-7 4.

-:7

NASA CP 2092 c.; 1

NASA

Conference

Publication

2092
LOAN COW:R i AWL TECH&l&,, KIRTLAND AFEz=

Aeropropulsion

1979

Proceedings of a conference held NASA Lewis Research Center Cleveland, Ohio May 15-16, 1979

at

I .

:.

TECH LIBRARY KAFB, NM

lllillllllll~illll~lllRll
ooss~7~

NASA

Conference

Publication

2092

Aeropropulsion

1979

Proceedings of a conference held NASA Lewis Research Center Cleveland, Ohio May 15-16, 1979

at

National Aeronautics and Space Administration Scientific and Technical Information Branch 1979

FOREWORD

The Lewis Research Center has a 37-year heritage of providing advances in aeronautical propulsion from the research activities of its staff and its university and industrial grantees and contractors. These advances have helped create the preeminence in aeronautics that has contributed to our national defense, has provided swift, reliable transportation for our people and their goods, and has so greatly aided our position in international trade. Although the results of the Center's projects and programs are reported as they are obtained, from time to time a conference such as this one affords the opportunity for a broad overview and an in-depth interpretation of a large body of results and for informal discussions about the subject. We hope you will find the material given herein informative and useful and that you will inquire further about whatever may be important to you. John F. Director McCarthy, Jr.

iii

CONTENTS

FOREWORD.........................
I.

Page iii

AIRCRAFT ENERGY EFFICIENCY (ACEE) STATUS REPORT Donald L. Nored, James F. Dugan, Jr., Neal T. Saunders, and Joseph A. Ziemianski . . . . . . . . . . . . . . . . EMISSION REDUCTION Donald A. Petrash, Larry A. Diehl, Robert E. Jones, and Edward J. Mularz . . . . . . . . . . . . . . . .

II.

. .

59

III.

NOISE REDUCTION Charles E. Feiler, John F. Groeneweg, Francis J. Montegani, John P. Raney, Edward J. Rice, and James R. Stone . . . . . . . . . . . . . . . . . . . ALTERNATIVE JET AIRCRAFT FUELS Jack Grobman . . . . . . . . . . . . . . . . . . . .

85

IV.

. . 129

V. MATERIALS AND STRUCTURES TECHNOLOGY Thomas K. Glasgow, Robert A. Signorelli, and Stanley R. Levine . . . . . . . Gary R. Halford,
VI.

. . 149

COMPUTATIONAL FLUID MECHANICS William D. David N. Bowditch, Bernhard H. Anderson, John J. Peter M. sockol. . . . . . .

OF INTERNAL FLOW McNally, Adamczyk, and . . . . . . . . . . . . . . 187

VII.

TURBOMACHINERY TECHNOLOGY Cavour H. Hauser, Jeffrey E. Haas, Lonnie Reid, and Francis S. Stepka. . . . . . . . . . . . . . . . . . . . 231 MECHANICAL COMPONENTS William J. Anderson, Robert C. Bill, and David P. Fleming . . . . . . . John J. Coy, . . . . . . . . . . . 273

VIII.

IX.

INSTRUMENTATION TECHNOLOGY David R. Englund, Jr., and William C. Nieberding, George E. Glawe. . . . . . . . . . . . . . . . . . . . . 309 CONTROL TECHNOLOGY John R. szuch. . . . . . . . . . . . . . . . . . . . . . 329

X.

XI.

SUPERSONIC PROPULSION TECHNOLOGY Albert G. Powers, Robert E. Coltrin, Leonard E. Stitt, Richard J. Weber, and John B. Whitlow, Jr. . . . . . . . 345

Page XII. XIII. HYPERSONIC PROPULSION H. Lee Beach, Jr.. . . . . . . . . . . . . . . . . . . . 387

VERTICAL TAKEOFF AND LANDING (VTOL) PROPULSION TECHNOLOGY John M. Abbott, Royce D. Moore, and Carl C. Ciepluch, James F. Sellers . . . . . . . . . . . . . . . . . . . HIGH-PERFORMANCE-VEHICLE TECHNOLOGY . . . . . . . . . Louis A. Povinelli . . . . . . . . .

. 409 . 445 . 463

XIV.

RESEARCH AND DEVELOPMENT CONTRACTORS AND GRANTEES . . .

vi

I. AIRCRAFT Donald

ENERGY

EFFICIENCY

(ACEE)

STATUS

REPORT

L. Nored,

James F. Dugan, Jr., Neal T. and Joseph A. Ziemianski

Saunders,

National

Aeronautics and Space Administration Lewis Research Center

Efficient air transportation is of national concern since commercial aircraft constitute a primary segment of public transHowever, the viability of the U.S. air transportaportation. tion industry and its ability to handle future traffic growth Figure I-l, are threatened by rapidly escalating fuel prices. taken from Civil Aeronautics Board (CAB) data (ref. l), illusFrom 1973 to 1975, fuel trates the recent trend in fuel prices. prices essentially tripled internationally and doubled domestiThey have since continued to increase about 12 percent cally. Such increases are directly reflected in the aireach year. craft direct operating cost (DOC). The major elements of the DOC are shown in figure I-2 (ref. 2). The elements are expressed in cents per available ton-mile of Before 1973, the mix of passengers and cargo for each year. these elements were about equal in their contribution to DOC. with the OPEC embargo and the subseStarting in 1973, however, quent large increases in fuel prices, these historical relationships changed. Fuel prices began to escalate faster than the rate of inflation and even faster than increased productivity As a result, not only has the total DOC incould reduce them. but fuel costs have become a much larger percentage of creased, as shown in figure I-3 for the Boeing 727 the DOC. For example, flying commercially in the domestic market, in 1973, fuel confuel pertributed about 26 percent of the DOC. By 1977, this centage had increased to about 41 percent. The percentage of the DOC related to fuel is expected to continue to increase in the future. Large increases in aircraft fuel needs are projected, as a result of an expected rapid growth in air travel. Aircraft fuel, however, is derived completely from petroleum, a dwindling national resource. Hence, scarcity of fuel and resulting higher prices are foreseen. This situation will be aggravated by various artificial (e.g., OPEC initiated) price increases. Commercial aircraft currently use over 10 billion gallons of fuel per year (ref. 1). This is conservatively projected to more than double by the year 2000. Obviously, an increase in fuel supply would alleviate the problems of cost and availability. For this reason, there is interest in producing

synthetic jet fuel from our large national resources of coal and However, such synthetic fuel will not be available oil shale. in large quantities for some years, and until it is, the aircraft industry must use its petroleum-based fuel more efficientIncreases in fuel efficiency will help counteract the eflY* fect of rising fuel prices on the DOC while also alleviating the critical problem of future fuel availability. To answer this need, NASA started the Aircraft Energy Efficiency This program is a focused reponse to (ACEE) program in 1976. the current importance of fuel effiency in aeronautics, for fuel conservation in general as well as for its effect on commercial aircraft operating economics. Included in the program are six major projects aimed at providing technology for more fuelconservative aircraft and propulsion systems for future commerThree of the projects - in the cial airline service (ref. 3). areas of aerodynamics and aircraft structures - are managed by the Langley Research Center. These aircraft-related projects are Energy Efficient Transport, Laminar Flow Control, and ComThe other three projects are propulsion reposite Structures. as indicated in figure I-4, lated, and are managed by the Lewis They are (1) Engine Component Improvement Research Center. at improving the engine components and the per(ECI) , directed formance retention of existing engines; (2) Energy Efficient directed at providing the technology base for the Engine (E3), next generation of turbofan engines; and (3) Advanced Turboprop, directed at advancing the technology of turboprop-powered aircraft to a point suitable for commercial airline service. It is these three projects that are discussed in this paper in some detail. ENGINE COMPONENT IMPROVEMENT The EC1 project is investigating the potentiai for reducing fuel usage in existing engines. It is these engines that will use the bulk of the commercial aircraft fuel between now and 1990. This project is expected to result in technology, by 1980 to 1982, that will permit as much as a 5 percent fuel savings over the operational life of the engines. Such fuel savings will be achieved by improving engine performance as well as by minimizing engine performance degradation in service. Thus, the project has two parts: (1) performance improvement and (2) engine diagnostics (fig. I-5). Performance Improvement part of the EC1 project is directed for improved, more fuel-efficient introduction into commercial seris being conducted by the manu-

The performance improvement at developing the technology engine components for early vice. The technical effort 2

facturers (General Electric and Pratt & Whitney) engines that power most of the current commercial These engines are the Pratt & Whitney JT8D I-6). Boeing 727 and 737 and the Douglas DC-9 - and the Whitney JT9D and the General Electric CF6 - used 747, Douglas DC-lo, and Airbus Industries A300. these latter two large, high-bypass-ratio engines power the newer airplanes such as the Boeing 757 Airbus Industries A310.

of the three fleet (fig. - used on the Pratt & on the Boeing Derivatives of will also and 767 and the

The components being investigated are shown in figure I-7, and include most major components of the engine. The specific component improvements will be derived generally from improved aerodynamics, reduced clearances, more effective cooling, and improved materials. An extensive feasibility and screening analysis was conducted on a variety of concepts by both General Electric and Pratt & Whitney before a specific concept was selected (refs. 4 and 5). This analysis was a team effort, with each engine manufacturer manufacturers (Boeing and being assisted by NASA, two aircraft and a number of airlines (United, American, TWA, Pan Douglas), Technical merits (e.g., performance, weight, Am, and Eastern). and maintenance) as well as economic merits (e.g., airline redirect operating cost, and payback period) turn on investment, were investigated. From considerations such as potential fuel economic benefits, and cost of development, NASA sesavings, lected 16 component improvement concepts for technology development by General Electric and Pratt & Whitney. For the JT8D development: (1) (2) (3) engine, four outer concepts air seal were for selected the (table I-l) for

An improved

high-pressure

turbine

A new high-pressure-turbine in the cooling air is

blade cooling concept, wheredischarged at the root of the blade stang materials on the the

An aerodynamically improved DC9/JT8D reverser fairing that incorporates advanced composite for lighter weight Longer blades high-pressure 1, 2, stang Co. and abradable compressor trenched rubstrips

(4) Concepts reverser Aircraft

and 4 are being developed fairing is to be developed

by Pratt by the

& Whitney; Douglas

Technology development has been completed on the improved JT8D Figure I-8 shows some features of the seal. outer air seal. Through the use of honeycomb seal material plus additional 3

knife-edges, an improved labyrinth arrangement was effectively added to the seal. Also, the blade cooling flow, which had been discharged completely from the blade tip, was rerouted to permit some discharge from the blade suction surface. Results have been obtained from back-to-back engine tests of both the current and improved configurations, as shown in figure I-9, over a range of thrust levels. At the normal (90 percent) cruise point, a specific fuel consumption (SFC) reduction of 0.6 perThis reduction (though small) is equivalent cent was achieved. to 6 percent of the total net income of the U.S. domestic airlines in 1977 (at an average fuel cost of 50d/gailon). It is also equivalent to an airline fuel savings of 200 million gallons by the end of the century, if the seal is incorporated both into new-production JT8D engines and into older engines through retrofit, where economically feasible. For I-l) the JT9D engine, for development (1) (2) (3) (4) Active four concepts were by Pratt & Whitney: control for only the also selected (table turbine along end with walls

clearance

high-pressure shroud vane

A new fan incorporating low-aspect-ratio blades Ceramic in the Ceramic thermal-barrier high-pressure outer air

a single on the

coatings turbine seals for the

high-pressure

turbine to

Development of the active clearance control concept has been As indicated in figure I-10, this is a technique completed. reduce tip clearances during cruise by cooling, and hence During the transient portions of flight shrinking, the case. (takeoff and landing), the cooling air is reduced or elimiThis increases the tip clearances between the turbine nated. blade and shroud and minimizes the potential for rubbing. The current The I-11. are (1) (2) (3) and improved JT9D significant features ring cooling for the impingement the cooling for configurations are shown in of the improved configuration that flow air supply distances air tubes that and a more tests of indicate the permits greater

figure

A modified support in tip clearance A doubling of the

reduction

A new configuration provides shorter effective use of are shown in and current

Results improved 4

figure I-12 configurations.

back-to-back These results

that cruise SFC can be reduced cruise thrust settings. For the (1) (2) CF6 engine, eight

0.65

percent (table

over I-2)

a wide were

range

of

concepts that

selected: in weight and

A short-core nozzle scrubbing drag A new front around the clearances A new fan Improved Improved

permits

reduction

mount that improves compressor case and with improved for control

the load distribution results in reduced design turbine

tip

(3) (4) (5) (6) (7) (8)

aerodynamic the for

aerodynamics roundness

high-pressure the high-pressure of

turbine air in turbine turbine all the

Reduced compressor bleed DC-10 cabin air-conditioning Active Active clearance clearance control control

by recirculation system for for the the high-pressure low-pressure

Concept 6 is under the other concepts

development by the are to be developed

Douglas Aircraft Co.; by General Electric.

Technology development has been completed on the first three The changes in the core nozzle resulted in a 0.9 perconcepts. cent SFC reduction at cruise, as demonstrated by flight tests on both the DC-10 and A300 aircraft. The new front mount can provide a 0.3 percent cruise SFC reduction. The improved fan can Figure I-13 lists provide a 2.0 percent cruise SFC reduction. the improved features for the new fan concept: basically, improved airfoils, rearward placement of the current single shroud to minimize blockage, and addition of a fan-case stiffening ring to keep the case round (hence, reducing tip clearances). Results of sea-level engine tests are shown in figure I-14. The indicated 3.2 percent SFC reduction at 40 000 pounds of sealevel thrust is equivalent to a 2 percent cruise SFC reduction. According to a projection based on various market predictions by the engine manufacturers, the performance improvement part of the Engine Component Improvement project should provide the airline industry with a cumulative fuel savings of at least 7 billion gallons by the year 2000. This savings would occur if all the concepts complete the technology development phase successfully, are carried through the certification phase by the manufacturers, and are successfully introduced into commercial service on future engine models or through retrofit. 5

Engine

Diagnostics

The engine disagostics part of the EC1 project is directed at identifying and quantifying the sources of the performance deterioration that occurs with time in the JT9D and CF6 engines. The effort will also provide design methodology and maintenance practices for minimizing such deterioration in current and fuAs with performance improvement, NASA has major ture engines. contracts with both Pratt & Whitney and General Electric for data analysis, and modeling. testing, The major contributors to performance degradation are shown in Examples of these degradation mechanisms are shown figure I-15. in figure I-16. In the cold section of the engine, foreign oband surface roughness are significant. In ject damage, erosion, thermal distortion is one of the the hot section of the engine, predominant degradation mechanisms - causing, for example, Clearance increases occur warpage or distortion of vanes. through the entire engine, as a result of blades rubbing with These clearance increases result in effithe outer shrouds. ciency losses. In investigating these contributors the general approach is as follows: (1) to performance degradation,

Gather existing (historical) data from airline in-flight recordings and from ground test cells at both airline (To date, and engine overhaul shops. this has been done for about one-third of the world's aircraft that use JT9D and CF6 engines.) Also, conduct inspections on a selected used parts that contribute to performance degradation. Augment this information with special engine tests and inspections in order to evaluate the effects of deteriorated components, and subsequent refurbishment, on both overall performance and engine module performance. (Overall performance deterioration was evaluated by securing an engine from an airline, testing it, conducting a complete teardown, and inspecting it. The engine was then refurbished and/or reassembled, retested, and returned to the airline. Module performance degradation was evaluated by tests before and after used modules were replaced.) Assess the causes of performance degradation during the first flight or flights of the aircraft as the engine structure first responds to the flight environment Assess the causes of long-term as a function of both cycles performance and time degradation

(2)

(3)

(4) 6

(5) (6)

Determine performance

the

effects

of

deteriorated

parts

on module and the

Establish statistical design criteria, with effect of maintenance

trends, analytical models, associated correlations of practices on SFC losses

The results of the module replacement and refurbishment test on the CF6 engine are shown in figure I-17. The numbers in parentheses represent the number of test modules. Simply cleaning the fan blades and recontouring the leading edges resulted in a specific-fuel-consumption reduction of 0.3 percent. Replacing the low-pressure turbine with new or refurbished modules resulted in a SFC reduction of 0.4 percent. The high-pressure-turbine since testing and analysis are resuits are not available yet, still under way. How the various performance degradation mechanisms - such as erosion and airfoil roughness, and thermal clearance increases, the cruise SFC deterioration of the JT9D distortion - affect engine modules is shown in figure I-18. These results, averaged are based on used-parts analysis for a large number of engines, Clearance and prerepair tests (i.e., the historical data). changes affect all the engine modules from the very first flight. As the engine matures, however, thermal distortion becomes of significance for the high-pressure turbine, and erosion and airfoil roughness become of significance for the highAlthough the data indicate that the highpressure compressor. and low-pressure-turbine modules do not deteriorate much beyond the 1000th flight, this is only because these modules are normally replaced every 1000 to 2000 flights. Thus, what is shown are deterioration values where the high-pressure turbine has been replaced between the 1000th and 3000th flights. Percent cruise SFC deterioration as a function of the number of flight cycles (where a flight cycle is a takeoff, flight, and landing) is shown in figure I-19. As evident, clearance increases cause more than 50 percent of the overall deterioration experienced in the JT9D engine. The major causes of these service-related changes in gas-path These are clearances are the loads on the engine in flight. experienced during airplane rotation, nacelle aerodynamic loads, as well as inertia loads, such as "g" and gyroscopic loads. schematically shown in figure I-20, cause the enThese loads, gine to bend and distort and the blade to rub against the outer They thus produce increased clearances and attendant shroud. performance losses. Pratt & Whitney and Boeing - in a cooperative NASTRAN finite-element structural model (fig. these flight-load effects on the JT9D engine. program - used a I-21) to analyze The NASTRAN model 7

was used to calculate the deflections or changes in clearances that occur throughout the engine, for both steady-state and dyThese clearance changes were then converted namic conditions. to performance losses. Shown in table I-3 are the NASTRAN theoretical results. In the steady-state case, the nacelle aerodynamic load affected all engine stages and contributed to about 87 percent of the total performance loss. The other two loading conditions were of lesser significance. In the dynamic cases, such as wind gust and hard landings, there were no significant changes from the steady-state results. To provide a data base for validating the NASTRAN theoretical steady-state load results, Pratt & Whitney will subject a JT9D engine to a simulated aerodynamic loads test. Aerodynamic loads will be simulated by placing a series of supporting bands (bellybands) around the inlet. These bands will then be pulled to exert loads on the inlet to simulate nacelle loading condiAs the engine is run through a performance tions (fig. I-22). X-ray and laser proximity probes will be used to determine test, changes in tip clearances. Additional instrumentation will be added to the engine to determine the engine and module performance changes that occur as the aerodynamic loads are simulated. A summary of engine diagnostics results to date for both the JT9D and CF6 engines is shown in figure I-23. Shown are the statistically averaged engine performance deterioration trends as a function of engine cycles. There is about a 3 percent increase in SFC after 3000 flight cycles, and about 1 percent is recovered during a normal overhaul (which is typically done on the engine hot section). Many factors influence the shape and values of the curves: Airline overhaul practice, route structure, derated takeoff, engine module mix, airplane model, and engine location are but a few. The output of the engine diagnostics effort is expected to provide industry with data necessary to cost effectively restore the performance of current engines. The magnitude and location of performance deterioration will also be pinpointed, and the modular performance analyses techniques required to diagnose the related performance losses associated with the engine will be improved. For derivative and future engines, this effort - by identifying unique degradation mechanisms and developing usagerelated deterioration models - will provide design tools for improving performance retention. ENERGY EFFICIENT ENGINE

The objective of the Energy Efficient Engine (E3) project is to provide an advanced technology base for a new generation of fuel-conservative turoofan engines for commercial transports. Specifically, this project involves aggressive development of 8

I-

followed by integration and advanced component technologies, testing as complete systems. A technology readiness date of At that 1983 has been set for completing these activities. time, the advanced component technologies will have been developed and demonstrated to a point where they are suitable for use In the late 1980's, in a future commercial engine development. E3 technology could appear in new advanced commercial turbofan sooner in advanced derivengines, or perhaps even a few years Benefits from E3 techative versions of current engines. nology would then start to accrue and would grow rapidly during the 1990's. Goals were established at the outset of the project to guide the selection of engine cycles and configurations and to serve as a These goals recognized that focus for the technology efforts. future engines must be not only fuel efficient, but also economically attractive to the airlines and environmentally acceptable Goals for fuel savings are to the public. (1) (2) At least factured a 12 percent engines reduction in SFC for in newly manureten-

At least a 50 percent improvement tion over the lives of the engines

performance

To provide economic incentives, NASA established a goal of at least a 5 percent reduction in the DOC. These fuel and economic goals are relative to current high-bypass-ratio turbofan engines (specifically, the JT9D-7A and the CF6-50C engine models). There is a high degree of uncertainty as to future environmental requirements. As a result, since any future advanced turbofan engine must meet the requirements prevailing at that time, NASA selected, as goals, the most stringent limits for noise and emissions that have been proposed to data by either the Federal Aviation Administration or the Environmental Protection Agency. These are the FAA's FAR-36 (1978) standards for noise and the EPA's proposed 1981 standards for emissions. The E3 project is being accomplished through parallel contracts with the two U.S. manufacturers of large, high-bypassratio engines: General Electric and Pratt & whtney. The effort The itself is structured and scheduled as shown in figure I-24. first element involves defining a baseline design for a future advanced propulsion system. This baseline design serves as a focus for defining technology needs, performance potential, component interfaces, and system trade-offs. To arrive at optimum cycle characteristics and engine configurations, both General Electric and Pratt & Whitney conducted trade-off evaluations, in which they concentrated on fuel usage and DOC for E3 technology engines operating on potential 1990-era aircraft. Each company was assisted by Boeing, Douglas, and Lockheed in conducting 9

these trade-offs. The initial design effort has been completed, but lower-level work will be conducted during the remainder of the project to support the experimental efforts and to use data from those efforts for refining the initial propulsion-system Thus, an on-going appraisal of the technology baseline design. Component technology status and potential will be obtained. and work has started on the needs have now been well defined, This element - which comprises second element of the project. the major portion of the project - involves design and extensive experimental development of the major engine components. The intent is to substantially improve the performance of each component. In the third element, the components will be integrated to assess their interactive effects and to work on those mechanical First, technologies that are unique to the integrated system. the high-pressure compressor, the combustor, and the highpressure turbine will be assembled in a core (or high spool) Then the core will be integrated with the low-spool package. booster or low-pressure compressor, lowcomponents (fan, pressure turbine, and mixer) and a metal "boilerplate" nacelle. package (ICLS) will then be This integrated core - low-spool tested to assess the overall performance, component interacand design integrity of the mechanical systems. tions, General Electric Configuration

The initial baseline propulsion-system design has been finFigure I-25 illustrates General Electric's E3 configished. which resulted from that effort. uration, Their configuration uses two spools plus a mixer and a long-duct nacelle. A mixer is included because, according to results from the initial performance analyses, exhaust-gas mixing offered the potential for about a 3 percent reduction in SFC from that of an unmixed engine. Table I-4 lists the cycle characteristics of General Electric's E3 configuration. As a comparison, values for General Electric's reference engine - the CF6-50C - are also shown. The bypass ratio of the E3 has been increased substantially, as has the compressor pressure ratio, over those of the CF6-50C. Turbine temperatures have been increased only moderately. Cycle trade-off studies indicated that, if the DOC and performanceretention goals are to be met, only moderate increases could be tolerated even with improved materials. A thrust of 36 500 pounds was used to size the hardware for this project. The technologies, however, are scalable over a wide range; hence, commercial engines based on E3 technology could have whatever appropriate thrust is required to meet airline needs. For this particular configuration, General Electric predicts a 14.2 percent cruise SFC reduction over the CF6-50C SFC. 10

The advanced technologies needed for the high-pressure-core components are listed in figure I-26. 'The aggressive compressor So that design has a compression ratio of 23 in only 10 stages. it can achieve a polytropic efficiency of over 90 percent, it will have advanced, low-aspect-ratio, highly loaded airfoils and active clearwill be made of advanced materials. In addition, ance control will be used on stages 6 to 10 in order to achieve much tighter clearances than those used in current engines. This The combustor is a two-zone, double-annular configuration. rather complex configuration is required to meet the stringent emission goals, which will be very difficult to achieve on an This combustor engine with an overall pressure ratio of 36. requires several technology advances, including improved diffuser designs and a segmented (or "shingled") liner, in order to achieve longer life. The high-pressure turbine has two stages and is designeu for an efficiency of over 92 percent at maximum cruise conditions. a number of technology advances This high efficiency requires including low leakages, improved cooling concepts, and advanced Also, several advances in materials active clearance controls. (directionally solidified and fabrication processes are required and near-net-shape rotors that use hot blades, ceramic shrouds, isostatic pressing of powder-metallurgy alloys). in technology advances is required for Similar aggressiveness the General Electric low-spool components, as shown in figure The I-27. Particularly important are the fan and the booster. fan uses low tip speeds and a low placement of the midspan damThe novel quarterper to achieve over 88 percent efficiency. It aids "island" booster design serves a dual function: stage and it centrifuges any matching of the fan and core streams, and thereby helps reforeign objects away from the core stream The five-stage, duce foreign-object-damage erosion in the core. low-pressure turbine has been specially configured to reduce noise. compressor and turbine, an And, as in the high-pressure active clearance control system will be used to permit closer The mixer will be designed to achieve 75 running clearances. percent mixing effectiveness and very low pressure losses in a short length. Pratt & Whitney Configuration

The Pratt & Whitney E3 design that resulted from the initial design and definition effort is illustrated in figure I-28. It is similar to the General Electric design in that it is a twolike General mixed-flow configuration. This design, spool, has also stressed greater durability, performance Electric's, Both designs feature a and ease of maintenance. retention, short, stiff, straddle-mounted core with five main bearings 11

located in easily accessible bearing compartments. Special attention also has been given to structural load paths in order to minimize the bending and twisting encountered in current enthus improving performance retention. gines, Both designs also feature careful attention to active and passive clearance control as an aid in retaining performance. The cycle characteristics of Pratt & Whitney's baseline design are listed in table I-5. Again, for comparison, values for their reference engine - in this case the JT9D-i'A - are also shown. The predicted cruise SFC reduction for the E3 design is 14.9 percent below that of the JT9D-7A. Advanced-technology features in the Pratt & Whitney E3 core configuration are shown in figure I-29. This core includes a lo-stage compressor with a compression ratio of 14. A number of advanced-technology features - including supercritical airfoils, trenched cases, and active clearance controls - help achieve a polytropic efficiency level of over 91 percent. The combustor also depends on a two-zone configuration for emissions control (in a similar manner to General Electric's combustor), but the Pratt & Whitney design has two burning zones arranged axially in series. Several technology advances are also planned for this combustor, including carburetor-type fuel nozzles and an advanced liner configuration that uses improved cooling concepts and segmented panels to extend life. Pratt & Whitney selected a single stage for the hign-pressure turbine in order to gain the cost and maintenance benefits associated with fewer hot-section parts. However, a cooled turbine efficiency of over 88 percent must be reached if the total engine performance required for fuel savings is to be achieved. Reaching this high efficiency in a single stage will require advances in several areas, including advanced airfoil designs for the transonic airflow in this turbine, improved cooling schemes, and several advanced materials. Pratt & Whitney's singlecrystal alloys will be used for the blades and vanes. An improved active clearance-control system is also planned for this turbine. in the iow-spool components of the Advanced-technolo y features are shown in figure I-30. Pratt & Whitney E4 design The fan blades are unshrouded. Eliminating the midspan dampers, used in the fan blades on all current engines, can improve efficiency and hence reduce SFC nearly 1 percent. However, the designs of unshrouded blades invariably result in very large, long-chord configurations that tend to increase engine weight. Thus, advanced design approaches are needed to generate lighter weight fan-blade configurations that are resistant to both aeroelastic Pratt & Wnitney has develflutter and foreign-object damage. oped a promising design approach that uses hollow titanium airfoils, and work is now under way to develop the manufacturing 12

technology approach titanium

for such unique blades. is to fabricate the hollow laminates.

the most promising To date, blades from diffusion-bonded

The other low-spool components include a four-stage, lowpressure compressor driven by a four-stage, counter-rotating, low-pressure turbine. The approach to exhaust-gas mixi,ng is also extremely aggressive, calling for a design mixing effecThese design tiveness of 85 percent and low pressure losses. goals are to be achieved by using a short, scalloped mixer to minimize the weight and drag penaities of the required long-duct nacelle. Component Technology

The component design phase is well under way, and both General Electric and Pratt & Whitney are conducting technology investiFigure I-31 illusgations in support of their design work. trates an example of such work at General Electric in support of Current engines have about their fan-frame acoustic design. Also twice as many vanes as fan blades for minimum fan noise. they have an axial spacing of about 1 chord width between the vanes and blades, along with separate struts to carry loads to General Electric will use a the fan case. In the E3 design, lower vane-blade ratio, as indicated, and a much wider axial spacing. This design permits the vanes to be integrated with the struts and hence reduces fan weight and cost. Acoustic tests of these two configurations were recently conducted, and the results are shown in figure I-32 for perceived noise level as a function of acoustic angle. The upper plot shows results for the approach condition; the lower plot shows results for the takeoff condition. The two configurations gave very similar acoustic results. Thus, GE's E3 fan is being designed with a vane-blade ratio of about 1.1 and very wide axial spacing, as illustrated in figure I-31. Another important series of supporting technology tests conducted by GE involves the combustor design. GE's prime combustor design is the double-annular configuration, shown at the left in figure I-33. Principles of this two-zone design were established under NASA's Experimental Clean Combustor program. The results of that program, when extrapolated to the higher pressure conditions for E3, indicate that this configuration offers potential for meeting the stringent E3 emissions goals. However, the two separate combustion zones require rather complex staging of the valves that control the fuel-flow to each zone. Thus, GE proposed an alternative design involving a simpler single-annular configuration. A series of early tests were run on this single-annular configuration in order to assess its potential for meeting the E3 emissions goals. 13

The prime resutls of those tests are shown in figure I-34. Plotted are the emissions levels for carbon monoxide (CO), hydrocarbons (HC), and oxides of nitrogen (NO,). The left bar in each pair shows single-annular combustor results from tests recently run as part of the E3 project. The right bar in each pair shows estimates for the double-annular combustor: These predictions are based on the earlier test results from the Experimental Clean Combustor program. These results indicate that the single-annular combustor does not meet the E3 goals for CO Also, these differences might be increased and NO, emissions. Thus, NASA has concluded with future broad-specification fuels. that the more-complex double-annular combustor is needed to meet All efforts are now directed the stringent E3 emissions goals. at the double-annular configuration so that the low emissions levels that might be required in future engines can be demonstrated. Another example of the component technology efforts completed to date involves Pratt & Whitney's single-stage turbine. Early emphasis was on aerodynamic tests in an uncooied turbine-rig so that the best airfoil designs could be selected. Two combinations of vanes and blades were tested; figure I-35 shows the Results for the uncooled-rig tests are shown in two blades. Plotted are the measured efficiency levels as figure I-36. functions of airfoil span for the two airfoil combinations. Both designs achieved good efficiencies, but the 43-percentThis reaction reaction airfoils produced higher efficiencies. value was thus selected for the E3 turbine design. Benefits The fuel savings that have been predicted for the E3 engines are shown in figure I-37. These are the projections made for both engine designs by three major aircraft manufacturers: BoeDouglas, and Lockheed. Using the predicted SFC reductions ing, the aircraft companies predicted fuel of 14 to 15 percent, usages for their own versions of advanced aircraft for various and then they compared these to the predicted flight missions, fuel usages for similar aircraft with current engines. Plotted are the differences in the fuel-usage estimates. (Note the orstarts at 10 percent.) dinate of the plot The projected fuel savings for E3 ranges from 13 to 22 percent. The fuel savings increase with longer flight distances since the prime fuel savings is achieved in the cruise portions of each flight mission. even the shorter-range flights with relatively short However, cruise durations show significant fuel savings. Projections were also made for the direct operating costs of the same aircraft-engine combinations, and a plot of these results is shown in figure I-38. As might be expected, the projected trends for DOC are similar to those for fuel savings since fuel 14

costs contribute about 40 percent to the direct operating Also important in these DOC reductions was the emphasis costs. both General Electric and Pratt & Whitney placed on minimizing maintenance costs. For example, the total numbers of airfoils in their engines were drastically reduced from the numbers in the reference engines. From the current status of the E3 project, these benefits apindicate that ail pear to be realistic. Present E3 designs with the possible exception of the NO, goals are achievable, emissions goal. The advanced technologies required to meet the goals are scheduled to be demonstrated by 1983. ADVANCED TURBOPROP The third ACEE propulsion effort, the Advanced Turboprop project, is directed at establishing the feasibility of radically improving propeller-driven propulsion systems to the point where they can be effectively applied to future commercial air transports. Advanced turboprop propulsion systems promise extremely large fuel savings. And, with an aggressive technology development program - demonstrating the advanced technology in the late 1980's - they would be available for the commercial air transports of the 1990's. Turboprop-powered aircraft are not are used today in various military, craft, and at one time were used in transports. Figure I-39 shows the which was introduced into Electra, vice in the late 1950's. It was a propeller-driven reciprocating-engine, mercial service. new, of course. Turboprops business, and commuter airmedium-range commercial air turboprop-powered Lockheed medium-range commercial sermajor improvement over the aircraft then in com-

such as the Boeing 707 shown in figure Jet-powered aircraft, I-40 (the first domestic commercial jet), were also introduced in the late 1950's. Such jet-powered aircraft offered better passenger comfort since they flew above the weather and had less They were noise and vibration in the cabin than the Electra. also faster. This, along with their larger size, provided greater productivity and convenience. And, they had newer engines with lower maintenance costs. The 707, and others like it, swept the market. Their major disadvantage was the higher fuel consumption of the turbojet engines. But, in an era of inexpensive fuel - around 5 to 10 cents a gallon at that time the price of fuel was not a critical factor in propulsion-system selection. Figure I-41 illustrates sumption for various per mile to account this last point. propulsion systems for speed differences. Shown is the fuel con- expressed in gallons These values are 15

plotted relative to the Electra engine fuel consumption. Also plotted are the dates of introduction into service (or possible introduction). Going from the Mach 0.6 turboprop-powered Electra to the Mach 0.8 turbojet-powered 707 resulted in a 50 percent increase in fuel consumption. Since the first commercial turbojet was put into service, a number of improvements have increased the efficiency of aircraft gas-turbine engines. Most significantly, the high-bypass-ratio turbofan engine has been a major factor in reducing fuel consumption. The Energy Efficient Engine promises another 15 percent reduction in fuel consumption, primarily througn better thermal efficiency. But the curve does appear to be flattening out, an indication of a maturing technology. Making the next significant step in fuel savings will necessitate a major increase in propulsive efficiency. Sucn an increase, about 15 percent at Mach 0.8, is offered by the advanced turboprop. Even larger savings would be possible at lower This propulsion-system concept will be considerably speeds. different, however, from the older, more conventional, Mach 0.6 Electra system. Figure I-42 illustrates this difference. The advanced propeller concept is compared with the Electra propeller, to the same scale. The shape, size, and number of blades are all radically different. The major features of the advanced propeller concept are shown in figure I-43. The advanced propeller would be driven by a lar'ge, modern turboshaft engine and a gearbox. The blades of the propeller would be very thin and highly swept to minimize both compressibility losses and propeller noise during highAn area-ruled spinner and an integrated nacelle speed cruise. shape would also be used to minimize compressibility iosses in the propeller-blade hub region. A high power loading, and hence a relatively small propeller diameter, would be achieved by using 8 or 10 blades. A final feature worthy of note is that modern propeller blade fabrication techniques would be used to make the thin, highly swept, twisted blades. The basic reason for the advanced turboprop being such an attractive concept is its potential for high propulsive efficiency in the Mach 0.7 to Mach 0.8 speed range, as shown in figure I-44. Older model turboprops had relatively thick, unswept propeller blades and experienced rapid increases in compressibility losses above Mach 0.6. Current high-bypass-ratio turbofans exhibit their highest propulsive efficiency (about 65 percent) at cruise speeds somewhat above Mach 0.8. The new advanced turboprops are estimated to be about 20 percent more efficient than high-bypass-ratio turbofans at Mach 0.8. At lower cruise the efficiency advantage of the advanced turboprop is speeds, This high propulsive efficiency of the advanced even larger. turboprop makes it an attractive powerplant for many applications. 16

A number of aircraft studies have been conducted to quantify the Figure I-45 shows benefits promised by the advanced turboprop. Two aircraft some of the passenger aircraft studied to date. a medium-range wide-body transport with four turboare shown: and a shorter-range narrow-body props mounted on the wing, transport with two turboprops mounted at the rear of the fuseCargo airplanes and military patrol aircraft could also lage. In every study, the benefit from using advanced turboprops. airplanes powered by advanced turboprops used much less fuel than competing turbofan-power airplanes. .These studies are summarized in figure I-46, where the trend in For shortfuel savings with aircraft design range is shown. where takeoff and descent dominate the fuel frachaul aircraft, the turboprop fuel savings can be as high as 30 percent. tion, fuel savings are 15 to 20 percent. For medium-range aircraft, For very long-range aircraft, where cruise dominates the fuel These turboprop fuel savings are 17 to 30 percent. fraction, fuel savings for the turboprop are relative to a turbofanpowered aircraft with the same level of component technology. will achieve a 15 percent fuel savings Thus, if an E3 turbofan over a conventional turbofan in a new medium-range transport, a new turboprop with an E 3 level of component technology would achieve a 30 to 35 percent fuei savings. It is this very large fuel savings potential that prompted NASA to include the Advanced Turboprop project in its ACEE program. The objective of this project is to develop the technology for efficient, reliable, and acceptable operation of advanced turboprop-powered aircraft at cruise speeds between Macn 0.7 and Mach 0.8. The turboprop goals of minimums of 15 percent fuel savings and 5 percent DOC improvement are relative to turbofans at equivalent levels of technology. The third turboprop goal is a cabin comfort level (noise and vibration) the same as that in modern turbofan-powered aircraft. To achieve these goals, Lewis began a phased Advanced Turboprop project in 1978. The major elements of phase I (an enabling technology and research phase) are shown in figure I-47. The first element - the propeller-nacelle - is concerned with propeller aerodynamics, acoustics, and structures. The second element addresses the cabin environment. Since the fuselage may be in the direct noise field of the propeller, the noise generated by the propeller must be attenuated by the cabin wall in order to provide a low-noise cabin environment. the turboprop Also, engines must be mounted in such a way that cabin vibration is low. The tnird major element of phase I, installation aerodynamics, is concerned with an accelerated, swirling propeller slipstream flowing over the wing. there is the challenge Here, of integrating propeller design with wing design to achieve the best combination of engine efficiency and aircraft lift-drag ratio. Also, airplanes powered by advanced turboprop engines 17

The must be configured to have adequate stability and control. fourth element involves the mechanical components of an advanced the engine drive, the gearbox, and turboprop propulsion system: the advanced propeller. These components must be designed and packaged in such a way that maintenance and reliability will be much improved over that experienced by the first generation of Since these four elements are so turboprop-powered aircraft. aircraft trade-off studies must be perstrongly interrelated, formed to obtain the match that will best achieve the goals of low fuel consumption, low operating cost, and passenger acceptance. Propeller-Nacelle The advanced propeller technology program in phase I is outlined The plan is to test a matrix of eight 2-footin table I-6. The range of the propeller design diameter wind tunnel models. Eight or 10 blades are parameters is indicated in the table. Sweep at used to keep the propeller diameter relatively small. be investigated to determine the blade tip as high as 60 will The tipits effect on propeller-generated noise at cruise. speed range results in transonic relative tip Mach numbers The high-cruise power loading also reduring Mach 0.8 cruise. Besides the exsults in relatively small propeller diameters. perimental work on propellers in phase I, an extensive analytical effort is under way to upgrade prediction and design capabilities. Three blades have been tested and are shown in figure I-48. Each blade is extremely thin over much of its span for minimum Also, the compressibility losses during high-speed operation. blade aspect ratio is relatively low as a result of designing for a relatively small propeller diameter (i.e., high power The blade on the left in the figure is straight, with loading). The middle blade has 300 of tip sweep. Oo of tip sweep. The blade on the right, with 450 tip sweep, was designed with both refined aerodynamic and acoustic design methodology. Indictated the rather unique blade planform shape, deed, acoustics which has a longer chord and more sweep than the other two This high-sweep propeller model is shown in figure I-49 blades. All three proinstalled in the Lewis 8- by 6-Foot Wind Tunnel. peller models were tested in this wind tunnel, and all were driven by the strut-mounted air-turbine propeller test rig shown in the background. - The measured efficiencies of all three propellers Efficiency. At Mach are shown in figure I-50 as a function of Mach number. 0.8 conditions, measured efficiency improved as the tip sweep of This confirms that blade sweep does rethe models increased. Also, efficiency improved as Mach duce compressibility losses. number was reduced. 18

Shown in figure I-51 is a more complete Mach 0.8 performance map Blade tip speed and blade pitch of the model with 450 sweep. angle were changed during testing in order to define lines of An efficiency of 79.5 percent has constant power loading. special significance since this efficiency was assumed in the early studies that indicated the potentiai of the advanced turMeasured propeller efficiencies exceeded 79.5 percent boprop. Since each point for a range of power loadings and tip speeds. on this map represents a unique set of values for propeller efficiency, diameter, weight, and tip speed, an airplane trade-off study must be performed to select the best propeller design for a particular application. - The relative noise generated by the three propeller Noise. as measured in the wind tunnel, is shown in figure models, the propeller with a tip I-52. At Mach 0.8 design conditions, Its peak sound pressure level sweep of 45O was the quietest. was about 6 decibels lower than that of the straight blade. Because of the many unknowns involved in noise measurements in wind tunnels, particularly in measurements of absolute noise levels, NASA is proceeding with a program that will allow propeller noise to be measured in flight. A NASA JetStar aircraft that is being modified to accept a small propeller and its air-turbine drive atop the fuselage is shown in figure I-53. The 2-foot-diameter propeller will be driven by an air turbine, whose working fluid is main-engine bleed air. Microphones will be mounted in the JetStar's fuselage to record propeller noise generated at flight conditions from takeoff to The Mach 0.8 cruise. Testing is scheduled to begin next year. data from these tests will be used to upgrade propeller noise The data on noise level, spectral content, prediction programs. and directionality will constitute the input needed to work on fuselage attenuation of propeller noise. structural techniques are Structural techniques. - Propeller Shown at the top of the figure is tne compared in figure I-54. structural design for a current-production fiberglass blade. It has a fiberglass shell, a solid aluminum load-carrying spar, and foam fill. Blades like this were introduced into service in the early 1960's. On the bottom is shown an example of how an advanced blade might be constructed. Materials and fabrication In place of techniques are similar to those for current blades. foam fill, however, honeycomb or some other material is used to This blade construction congive the thin blade more rigidity. along with others, will be analyzed in the coming year. cept, Cabin Environment the cabin depends mainly and on the attenuation on the of this noise noise generated by the 19

The noise inside by the propellers

fuselage wall are interrelated,

and

the cabin as shown

interior in figure

trim. I-55.

These

two

conditions

The maximum-cabin-noise goal is 75 decibels on the A scale For a typical propeller design, (dBA). sound pressure level in the cabin would be 90 decibels. With current conventional cabin exterior noise could be as high as 110 decibels. treatment, With improved fuselage design and cabin treatment, exterior noise of 140 decibels or more could be tolerated according to initial studies conducted by the Langley Research Center. Howthe price that must be paid for the higher noise attenuever, ation is an increase in fuselage weight. Influencing cabin noise, of course, is the propeller noise level. At the moment, there is uncertainty in predicting noise level and also uncertainty in interpreting wind tunnel noise measurements. The absolute noise level of the three models tested thus far is somewhere in the shaded area with a possible lower bound of approximately 140 decibels. With improved propeller design, it may be possible to reduce the noise level to Since the two bars overlap, a number of about 130 decibels. solutions are possible for achieving the desired maximum noise it will be necessary to conduct airlevel in the cabin. Thus, craft trade-off studies. Engine location is, of course, another The optimizations are quite differvariable in such studies. for example, when the turboprops are mounted at the rear of ent, the fuselage rather than on the wing. Installation Aerodynamics

The third major element of the Advanced Turboprop project is installation aerodynamics. So far only the question of drag has been addressed. Drag becomes of concern with the advanced turboprop because, at the Mach 0.7 to 0.8 cruise speeds of interlarge compressibility losses could be encountered on that est, area of the wing washed by the accelerated slipstream of the propeller. Preliminary results have been obtained with a propeller slipstream simulator (fig. I-56). This figure shows a wing-body aircraft model installed in an Ames wind tunnel. There were no model turboprops directly on the aircraft model; instead, the ejector device directed a simulated propeller slipstream over the wing. By using this device, slipstream Mach number and slipstream swirl angle were varied independently. It is these two parameters that are peculiar to a turboprop installation. The results are shown in figure I-57 for two cruise Mach numThe aircraft drag at a lift coefficient typical of cruise bers. is the zero reference point. When the simulator was turned on, generating a Oo- swirl slipstream, the increase in drag was 20

about 2.5 percent. This level was maintained to swirl angles of 70. At higher swirl angles, drag decreased as the wing recovered some of the swirl and actually became negative between 7O and 110. Since the amount of swirl of the slipstream increases as power loading is increased, propeller design seWing design lection must consider these slipstream effects. Such integramust also be optimized for the recovery of swirl. tion may yield an installation drag for the turboprop equal to or perhaps even lower than - that of the turbofan. Turboprop installation drag will be accurately assessed in the near future by testing an aircraft model with a 2-foot-diameter propeller, Data as shown in the artist's sketch of figure I-58. from this test will guide the integration of propeller design with wing design. Mechanical Components

The fourth major element in the Advanced Turboprop project inThis element was of particvolves the mechanical components. ular concern at the beginning of the project because of the high maintenance costs for the turboprop propulsion system on the Electra and the general feeling that turbofan maintenance costs For this reason, the Lewis would always be substantially lower. Research Center placed a contract with Detroit Diesel Allison (with Hamilton Standard as a subcontractor) to evaluate the Electra system. Results indicated that the 1970-era turboprop maintenance costs were indeed higher than those for the JT8D turbofan (a representative turbofan), but the difference was mainly in the core engine. The core engine of the turboprop was expensive to maintain largely because it was an older technology It was core originally designed for military application. therefore concluded that overall advanced-turboprop maintenance costs can be competitive with those of an advanced turbofan. based on equal levels of technology, Core maintenance costs, should be about equal. Preliminary conceptual designs were made to evaluate how turboprop maintenance costs couid be reduced. The advanced propeller and gearbox can be greatly improved by performing on-condition maintenance instead of scheduled overhauls, by using modular construction, and by emphasizing simplicity and reliability in design. Future Phases

Results of the subscale testing part of phase I has been very promising. Progress has been achieved in developing design and This phase is an enabling technology prediction methodologies. effort that will take about 3 years to accomplish, with the intent being to establish the feasibility of concepts and a data As a logical next step, NASA is planbase for future efforts. 21

ning large-scale component verification tests that will lead to To be evaluated are propelflight verification (fig. I-59). ler fabrication, propeller flutter, scale effects relative to and advanced fuselage-noisepropeller-generated noise, Operational effects for a full range of attenuation concepts. flight conditions would be an important feature in demonstrating Thus, at the conclusion of the Adtechnological readiness. technology would be available for use vanced Turboprop project, in designing very efficient turboprop-powered commercial aircraft and in establishing commercial acceptance of this advanced system. CONCLUDING REMARKS The technology readiness dates for the three Aircraft Energy EfBy the ficiency propulsion projects are shown in figure I-60. required for beginning commercial dates shown, the technology Such developments would be development will be demonstrated. expected to yield the fuel savings indicated in the figure. Figure I-61 indicates the potential benefits of the three projects in terms of both direct operating cost and fuel savings. DOC savings are, of course, important since future propulsion systems must be economically attractive to the airlines as we1.L as fuel efficient. There could be about a 5 percent fuel savings and a 3 percent DOC savings in derivative engines that use Engine Component ImWith a technology readiness data of provement (ECI) technology. 1980 to 1982, this project will yield results that are extremely applicable to the near-term needs of the airlines. Energy Efficient Engine (E3) benefits are about a 15 to 20 percent fuel savings and a consequent 5 to 10 percent DOC savings. These benefits could be realized by the late 1980's in new engines to be demonstrated by 1983. Advanced that use E3 technology turboprops are expected to provide about 30 to 35 percent fuel savings over current engines, if E3 technology is used for the turboshaft engine, with a possible 10 to 15 percent DOC savings. These major gains could begin to be achieved in the 1990's, if the technology is successfully demonstrated in the late 1980's. In a future where fuel prices (and possibly even fuel allocations) will influence aircraft design and selection, advanced turboprops are expected to play an important role. These three projects differ not only in their technology readiness dates but also in the level and type of technology that each is addressing. In going from the EC1 project to the E3 project to the Advanced Turboprop project, the emphasis shifts from the near term to the far term, with progressively larger benefits, but also with more unknown technology, with higher degrees of risk, with more difficulties in integration with the aircraft, and with ultimate applications being clouded as usual 22

by market represents culties And, if pected, mercial

The ACEE effort, however, and economic uncertainties. a comprehensive approach to resolving these diffiand unknowns and offers a number of technology options. the ACEE technology advances prove as attractive as exthis program should have a major impact on future comair transportation systems.

REFERENCES 1. 2. Fuel Cost Board. and Consumption, Monthly Reports. Civil Aeronautics

Trends in Airline Cost Elements. July 1977 plus nautics Board, Trends in Airline Unit Costs. nautics Board, Feb. 1979. Conservation of Aeronautics

Second ed., Civil AeroSupplement, Jan. 3, 1979; Fourth ed., Civil Aero-

and

3. Aircraft Fuel NASA-Office 1975. 4. Fasching, gram: Electric 1979.

Technology - Task Force and Space Technology,

Report. Sept. 10,

CF6 Jet Engine Performance Improvement ProW. A.: (Ri'9AEG295, General Task I - Feasibility Analysis. NASA CR-159458, NAS3-20629.) Co.: NASA Contract

JT8D and JT9D Jet Engine Per5. Gaffin, W. 0.; and Webb, D. E.: Task I - Feasibility Analformance Improvement Program: Pratt & Whitney Aircraft Group; NASA ysis. (PWA-5518-38, Contract NAS3-20630.) NASA CR-159449, 1979.

23

.PERFORMANCE IMPROVEMENT CONCEPTS PRATT & WHITNEY ENGINES CONCEPTS % SK REDUCTION COMPLETION DATE

P & WA JTSD ENGINE HPT OUTER AIR SEAL NEW HPT BLADE REVERSER STANG FAIRING TRENCHED ABRADABLE HPC P & WA JT9D ENGINE HPT ACTIVE CLEARANCE CONTROL NEWFANTECHNOLOGY HPT VANE THERMAL COATING HPT ABRADABLE OUTER SEAL

0. 6 .9 .5 .9 .7 1. 3 ::

I L

SEP JUL SEP DEC

1978 1980 1980 1980 1978 1979 1981 1981

NOV JUN FEB OCT

cs-79-1534

Table I-l

PERFORMANCE IMPROVEMENT CONCEPTS GENERAL ELECTRIC ENGINE CONCEPTS % SFC REDUCTION COMPLETION DATE

GE CF6 ENGINE SHORT CORE NOZZLE NEW FRONT MOUNT IMPROVED FAN HPT AERO REFINEMENT HPT ROUNDNESS CONTROL REDUCED HPC BLEED LPT ACTIVE CLEARANCE CONTROL HPTACTIVE CLEARANCE CONTROL 0. 9 .3 2. 0 1.3 .4 .7 .3 .6 FEB 1979 MAR 1979 MAY 1979 JUL 1979 MAY 1980 SEP 1980 OCT 1980 AUG 1981
cs-79-1535

Table I-2

24

PROPULSION

SYSTEM

LOADS

ANALYSIS

RESULTS

STEADY STATE AREA AFFECTED % OF TOTAL SFC LOSS 87 8 5

ALL STAGES HPT STAGES FAN STAGE DYNAMIC NO SIGNIFICANT CHANGE FROM STEADY STATE

cs-79-2089

Table I-3

CYCLE GENERAL

CHARACTERISTICS ELECTRICS

OF

ENGINES

AT MAX CRUISE CONDlTlONS EEE EXHAUST CONFIGURATION BYPASS RATIO FAN PRESS. RATlO COMPRESSOR PRESS. RATIO OVERALL PRESS. RATIO TURBINE TEMP, OF HOT-DAY TAKEOFF MAX CRUISE TAKEOFF THRUST, SLS-lb INSTALLED SFC, lbmlhrllbf
cs-79-2065

CF6-5DC SEPARATE 4.3 1. 72 12.9 32. 0 2400 2080 50 250 0.666

MIXED 6. 9 1. 61 22.6 36.1 2450 2170 36 500 0.572 L=-14.%A

Table I-4

25

CYCLE PRATT

CHARACTERISTICS 8, WHITNEY'S

OF

ENGINES

AT MAX CRUISE CONDITIONS EEE EXHAUST CONFIGURATION BYPASS RATIO FAN PRESS. RATIO COMPRESSOR PRESS. RATIO OVERALL PRESS. RATIO TURBINE TEMP, OF HOT-DAY TAKEOFF MAX CRUISE TAKEOFF THRUST, SLS-lb INSTALLED SFC, lbmlhrllbf
CS-79-2067

JT9D-7A SEPARATE 5.1 1.58 10. 0 25.4 2300 2000 46 300 0.677 = -14.9% 3

MIXED 6. 6 1. 71 13. 9 37. 4 2495 2200 41 115 0.576 IA Table I-5

ADVANCED

PROPELLER

TECHNOLOGY

PROGRAM

PHASE I

DESIGN MATRIX OF EIGHT 2-ft DIAM WIND TUNNEL MODELS NO. OF BLADES SWEEP AT BLADE TIP, deg TIP SPEED, ftlsec (RELATIVE TIP MACH NO. 1 POWER LOADING, hplft2 (RELATIVE DIAMJ 8 OR 10 0 TO 60 600 TO 800 (1.0 TO 1.15) 26 TO 37.5 (1. 2 TO 1.0)

USE DATA TO UPGRADE AERO, ACOUSTIC, & STRUCTURAL ANALYSIS & PREDICTION PROGRAMS CS-79-2155

Table I-6

26

U.S. AIRLINEJETFUELPRICE
50

r
-4'

MONTHLY AVERAGES

FUEL PRICE, algal

10

---

INTERNATIONAL TRUNKS DOMESTIC TRUNKS

YEAR

1977

I978

B-79-2295

Figure I-l

DIRECT OPERATING COST ELEMENTS

DOMESTIC OPERATIONS OFTHE DOMESTIC TRUNKS -----FUEL -MAINTENANCE {DIRECT AND INDIRECT] ---CREW .-.OTHER iDEPRECIATION. HENlALS. INSlJRANCEt ./. / , / / /

'76-5CENTS PER AVAILABLE TON-MILE 4-

/ .0'

/ /

7lI;69 1970 -1971

I Ii

1972 1973 1974 YEAR END Figure I-2

1975 1976

1977

:978

27

DIRECT OPERATING COSTS

B727-200; DOMESTIC OPERATION FUEL CREW MAINT OTHER

Figure I-3

ACEE

PROPULSION

PROJECTS

ENGINE COMPONENT IMPROVEMENT

_= 1

-=

\= X= =

ENERGY EFFICIENT ENGINE

ADVANCED TURBOPROP

Figure I-4

28

ENGINE COMPONENT

IMPROVEMENT

PROJECT

PERFORMANCE IMPROVEMENT GOAL OF % FUEL SAVINGS

ENGINE DlAGNOSTlCS
cs-79-2099

Figure I-5

PERFORMANCE

IMPROVEMENT

P & WA JT8D

P & WA JT9D

DEVELOP TECHNOLOGY FOR COMPONENTS TO REDUCE FUEL CONSUMPTlON IN NEW PRODUCTION OR RETROFIT OF CURRENT ENGINES BY 1980-1982

Figure I-6

29

COMPONENT
HIGH PRESS. COMPRESSOR (HPCh FAN-/ : :

AREAS
7 LOW PRESS. TURBINE (LPT) \ \\ EXHAUST NOZZLE

CS-79-2096

FigureI-

JT8D

HPT OUTER AIR SEAL

w
ADDED KNIFE-EDGE SEAL ADDED HONEYCOMB SEAL COOLING AIR DISCHARGE RELOCATED CRUISE SFC REDUCTION, 0.6% DEMONSTRATED r-l. I
c-79-2144

CURRENT

I IMPROVED

Figure I-8

30

JT8DHPTOUTER AIR SEALENGINE TEST

CRUISE SFC REDUCTION, -70 0 20

80 40 60 % MAX CRUISE THRUST

tS 7g ,;; - -

Figure I-9

ACTIVE CLEARANCE CONTROL

TAKEOFF CLEARANCE

CRUISE CLEARANCE

SHROUD & CASE

NO AIR

COOLING AIR

CS-79-208b

Figure I-10

31

-...,

mm

.,..---.--.--w

JTSD HPT ACTIVE

CLEARANCE

CONTROL

CURRENT MODIFIED SEAL SUPPORT RING INCREASED CASE COOLING AIR FLOW SHORTER IMPINGEMENT DISTANCE CRUISE SFC REDUCTION, 0.6R -. DEMONSTRATED

IMPROVED

Figure I-11

JT9DHPTACTIVE CLEARANCE CONTROL TESTDATA

1.0 CRUISE SFC REDUCTION, % (-)ooOo 5 c\ NORMAL CRUISE I 100

0 60
cs-79-2086

90 80 70 % MAX CRUISE THRUST

Figure I-12

32

CF6FAN

IMPROVED AIRFOILS REARWARD SHROUD FAN CASE STIFFFNFR CRUISE SFC REDUCTION, % DEMONSTRATED
cs-79-2097

CURRENT

IMPROVED

Figure I-13

CF6ENGINE TESTCOMPARISON
--o---I CURRENT FAN ENGINES IMPROVED FAN ENGINES

SFC

35 ooo

I I 40 000 45 OaJ SEA LEVEL THRUST. lb

I 50 cm0
cs-7g-2083

Figure I-14

33

CONTRIBUTORS TO ENGINE PERFORMANCE DETERIORATION

CLEARANCE INCREASES

Figure 1-15

EXAMPLES OFENGINE PERFORMANCE IIjETERlORATlON

F.O. D. ,fjHROUD

EROSION

--WARPAGE

CLEARANCE INCREASE

THERMAL DISTORTION cs-7g-zog4 Figure I-16

34

CF6MODULE REPLACEMENT/REFURBISHMENT TESTRESULTS

CRUISE CONDlTlONS

REPLACE VANES/SEALS mm.-% ASFC

NEW MODULE 0.4% ASFC

FAN (2) CLEAN BLADESlRECONTOUR LEADING EDGE 0.3% ASFC

Figure I-17

JTSDMODULAR PERFORMANCE DETERIORATION RESULTS

USED PARTS DATA & PRE-REPAIR TESTS 0 m lX$l CLEARANCE INCREASE EROSION/AIRFOIL ROUGHNESS THERMAL DISTORTION 1OOOTH FLIGHT CRUISE SFC DETERIORATION, %

cs-79-2091

FAN Lt% Hlk HPT CPT (ENGINE MODULE) Figure I-18

35

JT9D PERFORMANCE DETERIORATION BY DAMAGE MECHANISM

CRUISE CONDITIONS 4 EROSION (FAN, LPC, HPC)y /'

r
3CRUISE SFC DETERIORATION, %

1
cs-79-2088

> CLEARANCE INCREASE jj (ALL) 3000

1000

2DcQ FLIGHT CYCLES

FigureI-

JTSDEXTERNAL APPLIEDLOADS AND REACTIONS

THRUST

LOADS

NACELLE AERODYNAMIC 8" PRESS. DISTRIBUTION >' *NACELLE AERODYNAMIC LOADS *INERTIA LOADS

(s--"..!J:

FigureI-

36

JT9D/747 PROPULSION SYSTEM STRUCTURAL MODEL

11ODDSTATIC FREEDOMS 4 000 ELEMENTS

INLET REVERSER, COWLING, ETC INCLUDED IN ANALYSIS
135-79-2092

Figure I-21

JTSD ENGINE IN PRAll &WHITNEY X-RAY ENGINE TEST FACILITY

FigureI-

37

AVERAGE ENGINEPERFORMANCE DETERIORATION TRENDS

JT9D & CF6 4 3 CRUISE SFC 2 DETERIORATION, %

TYPICAL OVERHAUL RECOVERY CURRENT UNRESTORED P RFORMANCE

1lLcz!i
0

cs-79-2085

1000 2000

3000 USAGE, CYCLES

4MKl

Figure I-23

ENERGY

EFFICIENT

ENGINE

PROJECT

SUMMAl?cSCHEDULE

Icy
PROPULSION SYSTEM DEFINITION & DESIGN

COMPONENT TECHNOLOGIES

SYSTEMS INTEGRATION TECHNOLOGIES

GENERAL ELECTRIC
cs-79-2071

PRAlT & WHITNEY

Figure I-24

38

. --.....-..-_. -. .-.-.-... . ,....._._._. i T I

ENERGY

EFFICIENT

ENGINE

CONFIGURATION

GENERAL ELECTRIC

Figurer-25

39

ENERGY EFFICIENT ENGINE- GENERAL ELECTRIC CONFIGURATION

CORE COMPONENT TECHNOLOGIES

/ // /
TEN-STAGE H.P. COMPRESSOR 23~1 PRESS. RATIO HIGH AIRFOIL LOADINGS ACTIVE CLEARANCE CONTROL ADVANCED MATERIALS
(35-79-2187

i I I I
TWO-ZONE COMBUSTOR LOW EMISSIONS ADVANCED DIFFUSER SEGMENTED LINER DIGITAL-CONTROL STAGlNG TWO-STAGE H.P. TURBINE VERY HIGH EFFICIENCY LOW LEAKAGE IMPROVED COOLING ADVANCED MATERIALS ACTIVE CLEARANCE CONTROL

Figure I-26

40

ENERGY EFFICIENT ENGINE- GENERAL ELECTRIC CONFIGURATION

LOW-SPOOL COMPONENT TECHNOLOGIES

MIXER HIGH EFFECTIVENESS LOW PRESS. LOSS

:-.;

:-I _.

., :I i

.:.

__.. -.: . .

.. -

iASINGLE-STAGE FAN LOW TIP SPEED INTEGRAL STATORlFRAME SlEELlKEVlAR CONTAINMENT l/4-STAGE ISLAND BOOSTER AUTOMATlC CORE MATCHING REDUCED CORE FOD FIVE-STAGE L P. TURBINE LOW NOISE CONFIGURATION ACTIVE CLEARANCE CONTROL cs-79-ZIt36

Figure I-27

41

ENERGY

EFFICIENT

ENGINE

CONFIGURATION

PRATT & WHITNEY

Figure I-78

ENERGY EFFICIENT ENGINE - PRATT & WHITNEY CONFIGURATION

CORE COMPONENT TECHNOLOGIES

/ TEN-STAGE H. P. COMPRESSOR SUPERCRITKAL AIRFOILS TRENCHED CASES ACTlVE CLEARANCE CONTROL ADVANCED MATERIALS
cs-79-2189

I TWO-ZONE COMBUSTOR LOW EMISSIONS CARBURETOR FUEL NOZZLES ADVANCED LINER DIGITAL-CONTROL STAGING

\ SINGLE-STAGE H. P. TURBINE HIGH EFFKIENCY TRANSONIC FLOW IMPROVED COOLING ADVANCED MATERIALS ACTlVE CLEARANCE CONTROL

Figure I-29

42

ENERGY EFFICIENT ENGINE- PRATT & WHITNEY CONFIGURATION

LOW-SPOOL COMPONENT TECHNOLOGIES

/I SINGLE-STAGE FAN SHROUDLESS BLADES STRUCTURAL VANES AllKEVlAR CONTAINMENT

\ 4-STAGE L P. COMPRESSOR HIGH LOADING CANTED AIRFOILS

\ 4-STAGE L P. TURBINE COUNTER-ROTATING LOW LEAKAGE ACTIVE CLEARANCE CONTROL

MIXER HIGH EFFECTIVENESS LOW PRESS, LOSS

CS-79-2lA8

Figure 1-N

ENERGY EFFICIENT ENGINE- GENERAL ELECTRIC

FAN-FRAME ACOUSTIC CONFIGURATIONS CONVENTIONAL CONFlGURATlON E3 CONFIGURATION

0 STRUTS

VANkBLADE RATIO -2 Figure I-31

-1.1

CC.-79-07-!

43

ENERGY EFFICIENT ENGINE- GENERAL ELECTRIC

FAN-FRAME ACOUSTlC TEST RESULTS 0 E3 TYPE

NOI SE LEVEL,

130 r

TAKEOFF CONDlTlON

20

40 60 ACOUSTlC ANGLE, deg

80
cs-79-2077

100

Figure I-32

ENERGY EFFICIENT ENGINE-GENERAL ELECTRIC

COMBUSTOR CONFIGURATIONS DOUBLE-ANNULAR COMBUSTOR SINGLE-ANNULAR COMBUSTOR

. LOWER EMISSIONS MORE COMPLEX

l l

LOWER COST LESS WEIGHT cs-n-2070

Figure I-33

44

ENERGY EFFICIENT ENGINE- GENERAL ELECTRIC

COMBUSTOR EMISSION CHARACTERISTICS; 4% IDLE THRUST m m SINGLE-ANNULAR COMBUSTOR (TEST) DOUBLE-ANNULAR COMBUSTOR (PREDICTED)

5r 4 POUNDS PER 1000 Ibs 3 THRUST HOURS PER 2 CYCLE 1 n

--

.-

GOAL -----

co
Figure I-34

HC
cs-79-2073

ENERGY EFFICIENT ENGINE - PRATT & WHITNEY

H.P. TURBINE UNCOOLED RIG HARDWARE

3% REACTlON BLADE Figure I-35

43% REACTlON BLADE

45

ENERGY EFFICIENT ENGINE- PRATT & WHITNEY

H.P. TURBINE UNCOOLED-RIG TEST RESULTS

EFF, % 86 84 821 0 0 0 10 20 30

% EFF REACTION (UNCOOLED) 35 43 90. 1 90. 6 80 90 100

40 50 60 70 AIRFOIL SPAN, %

cs-79-2069

FigureI-

POTENTIAL FUELSAVINGS OF ENERGY EFFICIENT ENGINES

RELATIVE TO THE SAME AIRCRAFT WITH SCALED JT9D-7A OR CF6-50C 0 Cl A r BLOCK FUEL SAVINGS, % 20 BOEING DOUGLAS LOCKHEED

25

15 10 0

I
1000 4000 5000 2000 3000 FLIGHT DISTANCE, n mi 6000
cs-?g-?o7F

7000

FigureI-

46

POTENTIAL DOCIMPROVEMENTS OF ENERGY EFFICIENT ENGINES

RELATlVE TO THE SAME AIRCRAFT WITH SCALED JT9D-7A OR CF6-5DC 0 0 n BOEING DOUGLAS LOCKHEED

15

DOC IMPROVEMENT, %

cs-79-207.

I 1000

I I 3000 2000 5000 4000 FLIGHT DISTANCE, n mi

6000

7000

Figure I-38

LOCKHEED ELECTRA AIRCRAFT

Figure I-39

47

BOEING 707 AIRCRAFT

Figure I-40

PROPULSION SYSTEM RELATIVE FUELCONSUMPTION

1.6
1.4 vLOW REL4TlVE FUEL CONSUMPTlON, gal/mile l-2 l.. .HIGH \-.. .8
.fl

r I

vFlRST

COMMERCIAL TURBOJETS

BYPASS RATIO TURBOFANS

BYPASS RATIO TURBOFANS ENERGY EFF ;"::",, ENGlNE .' ,.:, _ ::,. ,".

mvmcm
L

TURBOPROP ENGIN (MACH 0.8) 2000

73 1960
?i-7y?-Z2U

I970 1980 1990 YEAR INTRODUCED INTO SERVICE Figure I-41

48

SIZE/CONFIGURATION

COMPARISON

ADVANCED PROPELLER
figure I-42

ELECTRA PROPELLER :
cs-79-2149

ADVANCED TURBOPROP PROPULSION SYSTEM

/) .c-THlN SWEPT BIADES

L LOADING (8 OR 10 BLADES) &ODERN BlADE FABRICATION

LARGE, MODERN TURBOSHAFT ENGINE & GEARBO:

CS-79-2165

Figure I-43

49

INSTALLED PROPULSIVE EFFICIENCY AT CRUISE

100
INSTALLED PROPULSIVE EFF, Lw / 90 80 70 60
cs-79-2153

HIGH BYPASS

511 -15 CRUISE MACH NO.

Figure I-44

ADVANCED TURBOPROP PASSENGER AIRCRAFT

Figure1-45

50

TREND OF POTENTIAL FUELSAVINGS FORADVANCED TURBOPROP-POWERED AIRCRAFT

RELATIVE TO TURBOFAN-POWERED AIRCRAFT WITH SAME LEVEL OF CORE TECHNOLOGY

10

t-

TAKEOFF & DESCENT -- -_... DOMINATFO - -....... ..--

CRUISE FUEL IIOMINATFD - . .. . - -

I
0 2000 4000 DESIGN RANGE, n mi 6000 _._~.,_ _.

Figure I-46

ADVANCED TURBOPROP PROJECT

n
b 0 0 0 MECHANICAL COMPONENTS ENGINE GEARBOX PROPELLER

PHASE I - MAJOR ELEMENTS

AIRCRAFT TRADEOFFS

9 INSTALLATION AERODYNAMICS 0 l DRAG STABILITY CONTROL

GOALS 0 LOW FUEL CONSUMPTION 0 LOW OPERATING COST 0 PASSENGER ACCEPTANCE CABIN ENVIRONMENT . . NOISE VIBRATION cs-79-2162 Figure I-47

51

MODEL PROPELLER

BLADES

SWEEP

O0

3o" Figure I-48

PROPELLER

MODELTEST IN LEWIS WIND TUNNEL

52

Figure I-49

PROPELLER

MODEL EFFICIENCY

LEWIS 8 x 6 WlND TUNNEL 85 POWER LOADING r 37.5 hp/ft* AT M = 0.8 MEASURED 80 PROPELLER EFF, % 75

70 .70
cs-79-2163

.75 .80 MACH NO.

.85

Figure I-50

PROPELLER PERFORMANCE AT MACH0.8

MODEL WITH 45 TIP SWEEP; LEWIS 8 x 6 WIND TUNNEL 81

r
37.5 POWER LOADING. hplft*

76i 650

700

740 a00 TIP SPEED, ftlsec

.-79-175

Figure I-51

NOISECOMPARISON OFADVANCED PROPELLER MODELS

MO = D.8 DESIGN CONDITIONS; LEWIS 8 x 6 WIND TUNNEL

-/-O OR--/
ON-, -.
%-. 0

REDUCTION IN 2 PEAK SPL OF BLADE PASSING FREQUENCY, dB 4

cs-7gd.156

_/--

/u

Figure I-52

Figure I-53

54

BLADE FABRICATION COMPARISON

FIBERGLASS SHELL-

CURRENT PRODUCTION FIBERGLASS PROPELLER BLADE

_-SOLID ALUMINUM SPAR HONEYCOMB FILL

-_.

DING EDGE ATH

ADVANCED TURBOPROP BLADE CONSTRUCTlON CONCEPT (EXAMPLE)

CS-79-2158

Figure I-54

CABINNOISE
ADVANCED TURBOPROP-POWERED AIRCRAFT AT MACH 0.80 CRUISE 160 ,$ABIN ATTENUATION /I .-PROP NOISE LEVEL

140 - 1 t SOUND ;;;;L* dB 100 120

-I I

-. -. -. -_

1 .. . -IMPROVED PROPELLE, DESIGN

-I --IMPROVED

ENGINE LOCATION

FUSELAGE DESIGN & CABIN TREATMENT

PROPELLER Al RCRAFT TRADES

80 -

CONVENTIONAL TREATMENT -B-B--DESIRED MAX CABIN NOISE LEVEL(75 dBA AT 160 Hz)

-F-:9-21,4

Figure I-55

55

Figure I-56

SIMULATED SLIPSTREAM EFFECT ON AIRCRAFT CRUISE DRAG

57 CRUISE MACH NO.

%
CHANGE IN AIRCRAFT DRAG (CL = 0.5) SLIPSTREAM SIMULATOR -5 =c$z$s$ EJECTOR DRIVEN NACELLE-"' I 4 7 DESIGN SWIRL ANGLE, deg
cs-79210

\o \
11

! -101 0

Figure I-57

56

POWERED-PROPELLER

SEMI-SPAN AIRCRAFT

MODEL

Figure I-58

ADVANCED TURBOPROP

FUTURE PHASES LARGE-SCALE COMPONENTS - DEVELOPMENT& VERIFICATlON . PROPELLER FABRICATION 0 PROPELLER AEROlACOUSTlClSTRUCTURAL SCALING VERlFlCATlON 0 FUSELAGE SEGMENTS/CABIN ENVIRONMENT . ADVANCED GEARBOX & PITCH CHANGE SYSTEM l AIRFRAME INTERACTIONS 0 INSTALLED PERFORMANCE l OPERATlONAL EFFECTS 0 ADVANCED ENGINE/GEARBOX/PITCH CHANGE SYSTEM PERFORMANCE Figure I-59 -,-7,-, 1. .,

57

ACEEPROPULSION PROJECTS
PROJECTED FUEL SAVINGS & TECHNOLOGY READINESS DATES 40 . MED RANGE MISSION . SAVINGS RELATIVE TO CURRENT ENGINES

30 F

ADVANCED TURBOPROP (ATPl

E ERGY EFFICIENT ENGINE YI (E loENGINE COMPONENT IMPROVEMENT (EC11 I I 1990 1995 1985 TECHNOLOGY READINESS DATE

0
(35-79-2293

I 1980

Figure I-60

POTENTIAL BENEFITS OFACEE PROPULSION PROJECTS

SAVINGS RELATIVE TO CURRENT ENGINES MACH 0.8 MEDIUM-RANGE MISSION 40 30 MISSION F
&$iy ~~~~~~~~~~~~ 1 : ..:G>,,gr .&.: :,. ENERGY EFFICIENT ENGINE

10 a 0 ENGINE COMPONENT IMPROVEMENT I 5 10 15 DOC SAVINGS, %

I
20
cs-79-2291

FigureI-

58

I I. Donald A.

EMISSION

REDUCTION E . Jones,

Petrash, Larry A. Diehl, Robert and Edward J. Mularz

National

Aeronautics and Space Administration Lewis Research Center

The quality of life in the United States and in particular the quality of the air we breathe came into focus as a national issue The Congress in 1970 amended the Clean Air Act to in the 1960's. include the control of undesirable gaseous pollutant emissions NASA responded by initiating an emissions from aircraft engines. This program, reduction research and technology program in 1971. and its future direction form the the results obtained to date, In 1973 the Environmental Protection Agency basis of this paper. issued aircraft engine gaseous pollutant emission standards, Since that time the EPA which were to be implemented by 1979. has continuously examined the effect of aircraft engine emissions on air quality and has closely followed the advancing technology Just last year, the EPA for the control of these pollutants. issued a Notice of Proposed Rule Making that would amend the standards. Final action by the EPA has not yet been taken with respect to the proposed amended standards. The Gaseous Pollutant emission standards in effect today for newThe three gasly manufactured engines are shown in table II-l. eous emissions that are controlled are carbon monoxide, unburned Six engine ciasses for and oxides of nitrogen. hydrocarbons, The Tl class gas-turbine-powered aircraft have been established. applies to turbofan engines of less than 8000 pounds thrust. Into this class fall most engines used in general-aviation aircraft. The T2 class applies to engines having thrust levels greater than 8000 pounds and for the most part applies to engines The T3 and T4 classes are specific used in commercial aviation. The T5 class to the Pratt & Whitney JT3D and the JT8D engines. applies to those engines used to power supersonic commercial airand the P2 class applies to turboprop engines. craft, 1971 NASA took several actions to address the issues of gasA program was constructed that had pollutant emissions. both near-term and far-term objectives. The near-term program concentrates on achieving a large and immediate reduction in the It addresses engines classes Tl, T2, T4, pollutant emissions. conducted largely under contract, is esand P2. This program, The far-term program has the overall goal of sentially complete. developing the technology necessary to define the minimum
eous 59

In

pollutant emission levels This program is achieve. search Center and through contracts with industry. NEAR-TERM PROGRAMS

that aircraft gas-turbine engines can being conducted both at the Lewis Reresearch grants to universities and

From 1970 to 1973 considerable effort was directed toward examining various methods of controlling engine emissions. Most of this work, however, was conducted in simplified test rigs. No serious efforts to integrate these emission-reduction methods into actual aircraft-engine combustors had been undertaken. Tne purpose of the near-term program was to study the various emission-reduction schemes, choose the best of these and evaluate their potential in realistic engine combustor geometries. figure II-1 illustrates schematically an By way of background, aircraft-engine combustor of the type used in large turbofan enBelow the sketch are bar graphs illustrating tne levels gines. The characteristic feaof pollutants typical of these engines. Tney have a tures of these combustors are noted in the sketch. the primary zone tends to operate fuel single burning stage; rich; and large amounts of air bypass the combustor primary zone and are admitted further downstream to cool and dilute the comThese combustors all use a step-louver, filmbustion products. and a large portion of the total combustor airflow cooled liner, The emission levels shown are exis used to cool the liner. pressed as values of EPAP (EPA parameter), and are obtained by integrating the engine emissions over the specified ianding and In general, emissions of carbon monoxide, total takeoff cycle. and oxides of nitrogen are all substantially hydrocarbons, Smoke levels are low, well below greater than the EPA standards. which reflects the work done to control the level of visibility, smoke in the late 1960's and early 1970's. Figure II-2 shows the combustion characteristics pollutant-emission levels and identifies the reduction techniques that are used. that cause high various emission-

At low-power conditions, typically engine idle, emissions of carbon monoxide and hydrocarbons are predominant, primarily because The combustion characteristics of poor combustion efficiency. responsible for low-power emissions are quenching of the burning gases near the film-cooled liner walls and quenching of the burnThe combustion stabiling mixture with cold dilution airflows. because of the low ity of the flames is also low, primarily inlet-air temperature and pressure that characterize the idle In addition, poor fuel atomization and distrioperating mode. bution contribute to poor combustion stability and low combustion contribute to greater emissions of efficiency, which, in turn, The emission-reduction techniques, unburned hydrocarbons. 60

1. -

itemized on the right of figure 11-2, all result in a marked reduction in carbon monoxide and hydrocarbon emissions. Increasing the burning zone residence time will allow time for the complete consumption of the hydrocarbon fuel and carbon monoxide. Reducing the flow velocity is one way to increase the residence time. Delaying the injection of dilution air will result in a longer primary zone and hence increased residence time. Delayed mixing also reduces quenching effects caused by rapid cooling of the burning mixture. Increasing the burningzone equivalence ratio, that is, adding more fuel to the burning zone, results in a higher local temperature, which accelerates Improvements in fuel atomization and the combustion reactions. distribution prevent large pockets of fuel-rich mixtures from occurring and make for smaller fuel droplets which can be more rapidly consumed. At high power conditions, typically takeoff and climb, the major pollutants are oxides of nitrogen and smoke. Carbon monoxide and hydrocarbons are present only in very small amounts, as comOxides of nitrogen bustion efficiency is virtually 100 percent. and smoke emissions are formed when residence times are long, when flame temperatures become very high due to the high air pressures and temperatures (which are typical of high-power operation), and where there is poor local fuel distribution, which These fuel-rich pockets or zones. causes very-high-temperature, pollutants can be reduced by decreasing the combustor residence time so that they do not have time to form in any significant amount. This can be done by increasing the velocity or by enhancing mixing, which effectively reduces the length of the burning zone. Decreasing the equivalence ratio, that is, operating fuel lean, reduces the maximum flame temperature and the rate at Improving fuel atomization which these pollutants are formed. and distribution results in a more uniform mixture of the fuel and air. An examination of the emission-reduction techniques reveals an interesting conflict between those that reduce idle emissions and Witn tne those that reduce high-power emissions (fig. 11-3): exception of improved fuel atomization and distribution, the reAs an example, to reduce duction techniques are in opposition. increased residence time is needed; however, low-power emissions, to reduce high-power emissions, decreased residence time is It is This poses a difficult combustor design dilemma. needed. one for low power and one for as if two different combustors, are required. high power, The problems, then, were to determine if these conflicting be integrated into a real engine combustor withapproaches could out compromising performance and, if so, to determine to what These were the level the resulting pollutants could be reduced. questions to be answered by our near-term emission-reduction proof this program were to Put another way, the objectives gram. 61

investigate new combustor concepts with the potential for signiflevels and to measure the emissions reicantly lower emission duction by engine test. Contracts were awarded to major aircraft engine manufacturers to devise and investigate new combustor conthese contracts were for a multiphase procepts. In general, The first phase consisted of the screening of a variety of gram. new combustor concepts to determine those with the greatest Those concepts so identified were emission-reduction potential. refined in the second phase, and finally the best or most "engine combustor design was tested in an engine to measure the ready" Table II-2 shows the engine manuemission reduction obtainable. The engines are arranged in facturers and the engines selected. order of increasing compressor pressure ratio. The EPA engine class designation is shown in the left column. Near-Term Program Results

The results of the programs conducted with the engines in EPA In general, these results are class T2 are discussed herein. quite similar to those obtained for combustors in the other enFigure II-4 shows the Vorbix combustor used in the gine classes. JT9D-7 engine. Vorbix is an acronym meaning vortex burning and A cross section of this combustor (fig. 11-4) shows that mixing. tnere has been a drastic departure from conventional combustors. The Vorbix combustor consists of two burning stages arranged in series: for low-power emission control, and a a pilot stage, for operation at all engine conditions beyond idle. main stage, the pilot stage is the igThe main stage is separately fueled; The bar graphs below the drawnition source for the main stage. ing compare the emissions of a production JT9D-7 combustor with those obtained with the Vorbix combustor tested in JT9D-7 engine. Carbon monoxide emissions were reduced by more than onewere reduced by over a factor total hydrocarbon emissions half; of nitrogen emissions were reduced by over oneof 10; and oxides A photograph of the Vorbix combustor is shown in figure half. 11-5. Figure II-6 shows the double-annular combustor tested in an exThis combustor also has two stages, perimental CF~-50 engine. the pilot and main stages are arranged in parallel, but, here, The pilot zone is used resulting in two annular burning zones. at all operating conditions and is designed to control low-power The main zone is functional at all engine conditions pollutants. above idle and is designed to reduce the high-power pollutants. The bar graphs compare the production CF6-50 combustor emissions Carbon monoxide with those of the double-annular combustor. emissions were reduced by about 40 percent; total hydrocarbons by of nittrogen by about 30 percent. A a factor of 10; and oxides photograph of the double-annular combustor is presented in figure 11-7. 62

With the completion of the engine tests, several important achievements had been made. All the gaseous pollutants were significantly reduced with these controlled-emission combustors. These combustors operated the engines in which they were tested to full takeoff power levels and essentially met the required engine acceleration times. In general, combustor performance was excellent, considering the extent of these programs. These combustors are, of course, not ready for manufacture, and many technical areas require further development. A major problem with these two-stage combustors is the coking of fuel in the mainstage fuel lines and injectors. This occurs when the engine is idling and no fuel is flowing through the main-stage injectors. Fuel in the lines or injectors is exposed to compressor-exit air temperatures for some time, resulting in the degradation of the This fuel degradation can become serious enough to affect fuel. the fuel flow and spray quality. Liner durability problems were particularly around the throat separating the pilot encountered, and main stages of the Vorbix combustor. The combustor exhausttemperature pattern factor can be improved and needs more work. Figure II-8 summarizes the advances made to control pollutant The combustor sketch illusemissions in these two programs. trates the typical features of controlled-emission combustors. a pilot stage for engine idle emission Note the multiple stages: control and a lean-burning main stage for all higher power operating conditions. Air-blast fuel injectors are often used in the main stage to achieve fine fuel drops intimately mixed with combustion air. Since most of the air is now used in controlling the combustion process, very little air is available for dilution the amount of air and temperature profile tailoring. Similarly, available for liner film cooling is reduced, and other, advanced Emission performance is summarized cooling schemes must be used. below. Carbon monoxide was reduced about 5u percent; total hydrocarbons by about a factor of 10; and oxides of nitrogen by Good reductions were achieved in carbon about 35 percent. monoxide and oxides of nitrogen emissions, but they may not be as Total hydrolow as may be required to meet emission standards. carbons on the other hand, virtually disappeared, which would suggest that no further work is needed to reduce levels of this Another aspect of our near-term program concentrated pollutant. on investigating other techniques to further reduce CarPOn monoxide emissions at low-power conditions. Idle Emissions Reduction Program

As mentioned earlier, carbon monoxide and unburned hydrocaroons are emitted from combustors mainly during low-power, or idle, Quenching of the burning fuel and air operating conditions. before combustion is complete is the characteristic most responsible for these emissions. Quenching takes place near the In conventional combustor liners and at the dilution air jets. 63

combustors, liners are cooled by injecting The interaction of combustion liner wall. air film sup- presses the completion of the Avoiding the use of this cool air film and surface to run hotter would greatly reduce reacting gases. Quenching can air jets with placed further zone can take combustion.

a film of air onto the gases with this cool combustion process. allowing the liner the quenching of the

also be reduced by delayed mixing of the dilutionthe combustion gases. If these dilution jets are downstream, the chemical reactions of the burning place for a longer time, resulting in more complete

With these facts in mind, the idle emissions reduction program had the objective of investigating new combustor concepts with potential for significantly lower engine-idle emissions. To achieve this objective a contract was awarded to an engine manufacturer to test and evaluate three new combustor concepts having unconventional design features. The testing and evaluation of the combustor were confined to typical idle conditions; no attempt was made to operate the combustors at the more severe high-power conditions. The application of this technology to a practical combustor system could be realized by using variablegeometry schemes or by using one of these designs as the pilot stage of a multistage combustor. The three combustor recuperative-cooling combustor. designs were the hot-wall combustor, combustor, and the catalytic-converter the

The main feature of the hot-wall combustor (fig. 11-9) is tne thermal barrier coating on the inside surface of the combustor liner to reduce wall quenching. These refractory coated along, with impingement-cooled liners and no film coolsurfaces, result in greatly reduced quenching losses at tne ing whatsoever, Also, walls. the secondary dilution-air jets are placed farther downstream to further reduce quenching. The main feature of the recuperative cooling combustor (fig. 11-10) is that all of the primary combustion air is first sent through the annular passages of the combustion liners as impingement cooling air before being admitted into the combustor through the air swirlers of the dome or the primary dilution Thus, the combustion air is first used to cool the liners holes. and in this way picks up heat before entering the combustor. This air temperature enhancement reduces pollutant emissions by increasing combustion reaction rates. The catalytic converter combustor (fig. initial burning zone followed ventional, The fuel is first burned catalyst bed. 64 11-11) consists of a conby a ceramic honeycomb in front of the catalyst

The catalyst then acts as a cleanup bed in a lean fuel mixture. reactor for the unburned hydrocarbon and carbon monoxide products in very low emissions of these in the combustion gas, resulting A photograph of the actual catalyst bed used in this pollutants. It consists of three sections design is shown in figure 11-12. of a ceramic honeycomb substrate cemented together and placed in a metal housing shaped to fit the 600 sector combustor test The ceramic honeycomb substrate is approximately 4 hardware. inches deep and is impregnated with the catalyst material. In addition designs, film-cooling, for good emissions. concepts dilution fuel and the main features of each of the combustor all featured impingement-cooled liners with no air-blast fuel injectors and they used improved, another factor in achieving low idle fuel atomization, the primary-zone equivalence ratio of these Finally, was designed to be very close to stoichiometric, and air was delayed until far downstream for maximum reaction of air. they to

The results of this experimental program are shown in figure II-13 in terms of carbon monoxide and unburned hydrocarbon 'These results are emissions at the design idle condition. compared with those of the controlled-emission combustor at a All three combustor designs dramatically similar idle condition. reduced carbon monoxide emissions and significantly reduced unThe large reduction in idle burned hydrocarbon emissions. emissions achieved by these concepts indicates the potential of cOmbUStOr this technology if it were to be applied into a full design for a gas turbine engine. Because the hot-wall combustor is the simplest, its design features are shown in figure II-14 as the pilot stage of a hypoThe refractory surfaces of the thetical multistage combustor. inner liner walls and the use of impingement cooling result in The pilot stage is designed minimized wall quenching effects. for optimum burning rates at the idle operation of the combustor. The projected emissions of such combustor operating in an engine over a standard landing-takeoff cycle are shown in the lower half Carbon monoxide emissions are seen to be draof the figure. matically reduced from the controlled-emission combustor. Unburned hydrocarbon emissions are further reduced from the already low levels of the controlled-emission combustor. Oxides of nitrogen are essentially unchanged from the controlled-emission because most of this pollutant is gencombustor, as expected, erated during high-power operation. To realize the benefits of work must be undertaken to apply this the reduced idle emissions, technology into a full combustor system.

65

FAR-TERM

PROGRAMS

At high-power operation high flame temperature is the most important factor in the formation of oxides of nitrogen. Experimental data as well as analytic predictions indicate that oxides of nitrogen emissions vary exponentially with flame temperature. Therefore, far-term efforts are concentrating on the technique of lean burning in which decreasing the combustion-zone equivalence ratio will lower the flame temperature with a resultant reduction in oxides of nitrogen. Reducing the combustion-zone equivalence ratio alone may not result in the lowest possible oxides of nitrogen emissions unless several other factors are also considered. Because the local flame temperature is a significant factor in controlling oxides of nitrogen production, fuel distributions with locally rich full pockets must be avoided. This requires the fuel and air to be uniformly mixed throughout the combustion zone. In addition, it may be necessary to prevaporize the fuel. Large fuel droplets in the combustion zone are consumed by a diffusion flame that surrounds the evaporating droplets. This process takes place at near-stoichiometric conditions, and the high temperatures produce Thus, combustors excessive oxides of nitrogen emissions. with provisions to prevaporize the fuel and premix the fuel and air may be necessary to realize the full potential of lean-burning techniques. Low levels of oxides of nitrogen emissions obtainable by lean, prevaporized combustion have been demonstrated in expremixed, The essential features of this perimental flame-tube combustors. Fuel is injected type of experiment are shown in figure 11-15. into a preheated airstream upstream of a simple flame-holding A vaporization and mixing zone provides sufficient time device. In sucn a wellfor the fuel to completely vaporize and mix. controlled experiment the fuel-air distribution at the flame holder is uniform. Order-of-magnitude reductions in oxides of nitrogen emissions have been obtained in experiments such as these. The concept of catalytic combustion offers the potential of even further reductions in pollutant emissions. By using a catalyst bed consisting of a ceramic honeycomb substrate impregnated with efficient combustion occurs at even catalytic material, stable, When operating in idealized leaner overall equivalence ratios. conditions, such as the flame-tube apparatus (fig. II-16), nearly pollutant-free combustion occurs. Even though lean, premixed, combustion have the potential levels. Considerably more nology could be applied to 66 prevaporized combustion and catalytic for achieving very low emission work is required before either techaircraft engine combustion systems.

This then program.

forms

the

basis

for

the

far-term

emission

reduction

The objective of our far-term program is to evoive the technology needed for the development of combustors with minimum pollutant levels. The achievement of this objective relies heavily on continuing basic and applied research. The degree of risk and overall level of complexity associated with the adaptation of advanced techniques is more severe than that of the near-term programs. Fundamental studies are viewed as a requirement to closing the gaps in our understanding of key areas and to bringing the new technology to a point where a new approach to combustor design is practical. As discussed previously two techniques appear particularly attractive: The lean, premixed, prevaporized and the catalytic combustion techniques. In late 1979 contracts will be awarded to evolve and evaluate both of these combustor concepts. Before lean, premixed, prevaporized combustors can be used in aircraft engines, additional research is required in several areas. Shown in figure II-17 is a conceptual drawing of a lean, prevaporized combustor, which is a staged type of premixed, design. The pilot stage has been configured to include features, such as a hot-wall liner to minimize idle pollutants. The main stage looks much like a flame-tube rig and contains a fuel injector, a premixing and prevaporizing section, and a flame holder. To maintain a wide operating range, while burning as controi of the airflow as well as the fuel flow lean as possible, between the two stages may be required. To achieve this required airflow control, a variable geometry device has been included in the diffuser section. Key areas requiring additional study are also shown in the figure 11-17. Combustor inlet airflow characteristics must be known to Engine transient cnarassure uniform fuel-air distributions. acteristics must be identified and studied to avoid autoignition and flashback in the fuel-air mixing passage. Practical scnemes for varying the combustor geometry and controlling the operation Techniques for predicting of the combustor must be identified. and achieving the required fuei distribution and vaporization in Autoignition the premixing section of the main stage as needed. More data on these and flashback may also be problems there. phenomena are needed over the full range of engine operating conOther areas of the combustor also including transients. ditions, require study. Lean stability and altitude relight capability need special attention with these advanced concepts. Because most of the combustor airflow must pass through the main stage to satisfy the lean burning requirement, the amount of air available to cool the combustor liner will be less tnan tnat of current It therefore appears likely that advanced technology combustors. liner cooling schemes will be required to avoid liner durability problems. 67

Digital engine controls will likely be required for the addiIt is expected that tional complexity of variable geometry. full-authority digital-control technology will be available in However, additional study is needed to examine the the future. control aspects of variable geometry combustors and to establish transient response requirements. Figure II-18 indicates the areas requiring research for catalytic In general, all of the problem areas associated with combustors. premixed combustion apply equally well to the catalytic conProblems unique to this concept include the activity of cepts. the catalytic materials over wide operating ranges, long-term and thermal durability degradation and poisoning of the catalyst, of the catalyst during continuous and cyclic operation of the While considerable progress has been made in the catalyst bed. last few years on research into catalyst and substrate materials, It is not the purpose of this considerably more is required. paper to fully document the results to date of these fundamental However, three studies in support of our far term program. studies that highlight some of the activities have been selected for discussion. Studies of the influence of flame-holder geometry on emissions Figure II-19 shows the and performance have been undertaken. flame zone structure for six flame-holder designs that have been The designs tested included wire grids, perforated evaluated. These open-duct burning photoplates, cones, and C-gutters. Actual testgraphs were taken only for visualization purposes. ing was done at high pressure in an enclosed flame-tube rig. Figure II-20 presents data from a high-pressure, lean, premixed, Emissions of oxides of niprevaporized flame-tube experiment. trogen are shown as a function of the adiabatic flame temperWhen this experiment was completed, several milestones ature. flame-tube experiments were for the First, had been passed. first time conducted at pressures well above 10 atmospheres. In these flame-tube experiments were successfully conducted at fact, pressures to 30 atmospheres and inlet-air temperatures comparable Second, although previous data had to those of modern engines. shown an inconsistentency in oxides of nitrogen emissions with this experiment demonstrated that, from 10 increasing pressure, to 30 atmospheres, pressure had no effect on oxides of nitrogen emissions in lean, premixed, prevaporized, combustors when corThird, this experirelated against adiabatic flame temperature. ment verified the emission levels projected from lower pressure tests. Tests are also underway to determine the characteristics of boundary-layer autoignition and flashback phenomena in premixed As discussed earlier, an understanding of fuel-air streams. autoignition and flashback phenomena are important in the design and operation of the premixer section. Figure II-21 is a photo68

graph taken through the viewing port of a high-pressure flame-tube apparatus with a quartz liner. A variety of test plates can be mounted in the premixed stream to map the autoignition and flashback limits of various boundary-layer profiles for high inlet temperature and pressure conditions. In this particular photograph the inlet temperature and pressure were such that autoignition of the fuel-air mixture occurred in the boundary layer. In another view taken from a high-speed movie sequence (fig. II-22), the flame was observed on the upper surface propagating upstream through the boundary layer. When the inlet-air temperature and pressure are further increased, autoignition occurs in the free stream. An example of this phenomenon is shown in figure 11-23. Details of the combustion on the test plate as well as residual droplet burning may also be Studies of this type apply to both premixed and catalytic seen. combustion. Looking into the future, we can pro-ject engine emission levels for lean, premixed, prevaporized and catalytic combustors. The drawing of the combustor in figure II-24 shows some essential features of these designs. The combustor is staged, has variable geometry, and has optimized pilot-stage technology. In the main stage lean combustion occurs downstream of the flame holder or, in the case of a catalyic combustor (see insert), in the catalyst bed. The bar graphs of carbon monoxide and hydrocarbons show achievable levels which are based on the successful integration of optimized pilot-stage features such as those discussed previously. The emission-control strategies used here were aimed at oxides of nitrogen reduction. The third set of bar graphs shows in terms of tne integrated EPA parameter, oxides of nithat, trogen levels may be further reduced from the previous controlled-emission techniques by a factor of 3 to 4. It is interesting to note that the pilot stage, which is necessary for engine startup and wide-range operation, may contribute more oxides of nitrogen during engine idle than the main stage contributes during high-power operation. Thus, the pilot stage is limiting the minimum achievable oxides of nitrogen emission levels for the specified landing-takeoff cycle used in computing the EPA parameter. The actual levels may be somewhat different when these emissioncontrol techniques are developed into operational engine hardware: Tradeoffs between emissions, performance, altitude-relight capability, durability, maintainability, and complexity will be evaluated in future experimental programs. Tne influence of the actual engine environment as opposed to carefully controlled rig experiments will be considered.

69

CONCLUDING REMARKS The combustion systems found in engines today may well be markedly different in the future if low-pollutant-emission combustion systems are required. Current combustors are characterized by burning in a single stage, film-cooled liners, single-plane fuel injection, and rich burning in the primary Current combustors are all fixed-geometry systems. zone. The prospective minimum pollutant combustors are far different. multiple stages will be needed to reduce both low-power First, and high-power emissions. Advanced liner-cooling concepts will in the pilot stages, to reduce quenching be used, particularly and to control low-power (idle) emissions. The main stage will burn lean to reduce oxides of nitrogen by reducing the flame temperature. To fuel the two stages multiple-point fuel inAnd last, jection will be needed. variable geometry will be incorporated to optimize overall combustion system performance. still needs to be done before these advanced Much work, however, systems can be considered for actual application. Quite obLow pollutant emission comviously there is a price to pay: bustion systems are more complex. These complexities arise from the very items that distinguish the advanced combustor from the current combustor, that is, the need for more injectors to fuel the multiple-burning stages, tne more complex liner cooling required, and the variable geometry. The There are price for this complexity will be higher initial cost. the most significant of which is the drastic benefits however, reduction in pollutant emissions. Because of the lower flame temperatures and uniform burning, the combustor-exit pattern factor should be improved over that in current engines, thereby resulting in longer turbine life. Additionally, hot streaks in the combustor should be minimized, thereby increasing combustor liner life. The ability to vary the airflow distribution in the combustor through the use of the variable geometry should improve the altitude relight characteristics of the system. At present we are a long way from realizing the potential of Research and technology these advanced combustion systems. programs indicated that dramatic reductions in the pollutant emissions of gas turbine engines can be achieved in the near Taking that technology and developing it into practical term. Continuing research flight systems remains to be accomplished. and technology programs must be pursued to validate that the minimum pollutant emission levels found in rig tests can in fact be realized in gas-turbine engine combustion systems.

70

GASEOUS

EMISSION

STANDARDS POLLUTANT" HC CO 9.4 4.3 4.3 4.3 30.1 NO, 3.7 3.0 3.0 3.0 9.0 12. 9

ENGINE CLASS

Tl

TURBOJET/TURBOFAN LESS THAN 8000 lb THRUST 1.6 .8 .8 .8 3.9

T2 TURBOJET/TURBOFAN GREATER THAN 8000 lb THRUST T3 T4 P&WJT3D P & W JT8D

T5 TURBOJET/TURBOFAN ENGINES FOR SUPERSONIC AIRCRAFT P2 TURBOPROP ENGINES "T STANDARDS AS lb/l000 lb THRUST-hr/CYCLE P STANDARDS AS lb/l000 hp-hr/CYCLE

4. 9 26.8

cs-/Y-175s

Table II-I

NEAR

TERM

EMISSION

REDUCTION

PROGRAM

SCOPE

ENGINE CLASS P2 Tl T4 T2 T2

MANUFACTURER

ENGINE

DETROIT-DIESEL-ALLISON GARREll AIRESEARCH PRAll & WHITNEY PRAll & WHITNEY GENERAL ELECTRIC

501-D22-A TFE-731-2 JT8D-17 JT9D-7 CF6-50

cs-79-1747

Table II-Z

71

CURRENTCOMBUSTORS

PRIMARY ZONE -/ 10 5 10

& FILM-COOLING AIR

2Or

0 LI'Itl CO, EPAP

0 Tl-lC, EPAP

0 NO,, EPAP

0
SMOKE NO. CS-icl-: j j,Y

Figure II-1

COMBUSTOR EMISSIONS
OPERAilNG CwDlTlONS LOW POWER/IDLE COMBUSTlON CHARACTERlSllCS QUENCHING POOR COMBUSTION STABILITY POOR FUEL ATOMIZATION & DISTRIBUTlON POLLUTANTS REDUCTlON TECHNIQUES INCREASE RESIDENCE TIME REDUCE FLOW VELOCITY & DELAY MIXING INCREASE EQUIV RATIO IMPROVE FUEL ATOMIZATION 8, DISTRIBUTION

CARBON MONOXIDE UNBURNED HYDROCARBONS

HIGH POWER/ TAKEOFF

EXCESS RESIDENCE TIME HIGH FLAME TEMP POOR LOCAL FUEL DISTRIBUTION

OXIDES OF NITROGEN SMOKE

DECREASE RESIDENCE nME INCREASE FLOW VELOCITY ENHANCE MIXING DECREASE EQUIV RATlO IMPROVE LOCAL FUEL DlSTRlBUllON
cs-79-2054

Figure II-2

72

EMISSION REDUCTION PROBLEM

INCREASE RESIDENCE TIME REDUCE FLOW VELOCITY DELAY MIXING INCREASE EQUIVALENCE RATlO IMPROVE FUEL ATOMIZATION & DISTRIBUTION

LOW POWER IDLE

HIGH POWER TAKEOFF

DECREASERESIDENCE TIME INCREASE FLOW VELOCITY ENHANCE MIXING DECREASE EQUIV RATIO IMPROVE LOCAL FUEL DISTRIBUTION
cs-79-2042

Figure II-3

VORBIX COMBUSTOR CONCEPT FORJTSD-7 ENGINE

PRODUCTlON

CO, EPAP

MC, EPAP Figure II-4

NO,, EPAP

73

PROTOTYPE

VORBIX COMBUSTOR

Fiaure II-5

DOUBLE/ANNULAR COMBUSTOR FOR(36-50 ENGINE

FUEL-, tl

PRODUCTION DOUBLE ANNULAR

CO, EPAP

THC, EPAP Figure II-6

NO,, EPAP

,- ,I. ! ,,I

74

PROTOTYPE DOUBLEANNULARCOMBUSTOR

FigureII-7

CONTROLLED-EMISSION COMBUSTOR
FUEL1 I\ , ,-AIR-BLAST INJECTORS ,,-MULTIPLE BURNING .ZONES

m fXl

CURRENT CONTROLLEDEMISSION

CO, EPAP

MC, EPAP Figure II-8

NO,, EPAP

75

-,

HOTWALLCOMBUSTOR
CONCEPT NO. 1 SHELLS WITH IMPINGEMENT COOLED LINERS 7

SWIRL CUP AIR mu

cs-79-1751

PRIMARY DILUTION Figure II-9

SECbNDARY DILUTION

RECUPERATIVE COOLING COMBUSTOR

CONCEPT NO. 2 SWIRLER AIR WITH ENHANCED TEMP I \ FUEL INTO SIMPLEX NOZZLES PRIMARY DILUTION AIR iii Ai HIGHER TEMP j Figure II-10
Ill

SHELLS WITH IMPINGEMENT m COOLED LINERS7

SECONDARY /I DILUTION u CS-,-:(/

76

II -

CATALYTIC CONVERTER COMBUSTOR

CONCEPT NO. 3 SHELLS WITH IMPINGEMENT ,-HONEYCOMB

LlNITIAL BURNING ZONE Figure II-11

CATALYTIC COMBUSTOR ASSEMBLY

Figu re II -12

77

EMISSIONS AT IDLE

co
EMISSION INDEX. IbllOOO lb

EC

HOT RECUP CAT WALL

.nJWALL

RECUP CAT
cs-79-2041

Figure II-13

PROJECTED ENGINE EMISSIONS INCORPORATING IDLE EMISSIONS CONTROL TECHNOLOGY

OPTIMIZED EQUIVALENCE RATlO AT IDLE IMPROVED AIR BLAST FUEL r---.,1 QUENCHING (IMPINGEMENT COOLED LINERS) ICONTROLLED EMISSION ICONTROLLED EMISSION & IDLE EMISSIONS CONTROL 10 5 0 CO, EPAP HC, EPAP hh NO,, EPAP
cs-79-2142

REFRACTORY

Figure II-14

78

PREMIXFLAMETUBE COMBUSTOR

41

/-FUEL INJECTOR r FLAME HOLDER

\\
i: <
<

COMBUSTOR

II

bAPOR,ZATlON & MIXING ZONE Figure II-15

CS-79-1752

CATALYTIC FLAME TUBECOMBUSTOR

,-FUEL INJECTOR

=
PREHE, AIR

i-VAPORlZATlON & MIXING ZONE Figure II-16

cs-79-1754

79

LEAN, PREMIXED, PREVAPORIZED COMBUSTOR

REQUIRED RESEARCH AREAS FUEL DISTRIBUTION FUEL VAPORlZATlON AUTOIGNITION FLASHBACK AIRFLOW UNIFORMITY TRANSIENT EFFECTS VARIABLE GEOMETRY

LEAN STABILITY ALTlllJDE RELIGHT LlNER DURABILITY CONTROLS

Figure II-17

CATALYTIC COMBUSTOR
REQUIRED RESEARCH AREAS FUEL DlSTRlBUTlON FUEL VAPORIZATION AUTOIGNITION FLASHBACK AIRFLOW UNIFORMITY TRANSIENT EFFECTS VARIABLE GEOMETRY

CATALYST ACTlVllY CATALYST DEGRADATlON THERMAL DURABILITY

Figure II-18

80

OF SIX FLAMEHOLDERS AT 70% BLOCKAGE OPENDUCT OPERATION

WIRE GRID

PERFORATED PLATE

MULTlPLECONE

CEE GUTTER

SINGLECONE

SWIRL CONE

Figure II-19

INLET TEST CONDITIONS PRESS., atm J-$;p.

NITROGEN OXIDES EMISSION INDEX, lb NO2 loo0 lb FUEL

lo

Figure II-20

81

STUDYOF BOUNDARY LAYERAUTOIGNITION AND FLASHBACK PHENOMENA

Figure II-21

FigureII-22

a2

Figure II-23

PROJECTED ENGINE EMISSION LEVELS FORFARTERMTECHNOLOGIES

PREMIX PREVAP PASSAGE 7 n \

15

(OPTIMIZED) -/ 5

I 10 5

CATALYTIC

5 10 0 CO, EPAP 0 ELHC, EPAP Figure II-24 0 NO,, EPAP

cs-79-2141

83

L--

III.

NOISE

REDUCTION Francis J . Montegani, and James R. Stone

Charles E. Feiler, John F. Groeneweg, Edward J . Rice, John P. Raney*, National

Aeronautics and Space Administration Lewis Research Center

Aircraft noise has been a major national environmental concern for a number of years. One indication of the public pressure to reduce noise is the number of major airports around the world that have noise restrictions. Some data are shown in figure III-1 for the years 1968, 1973, and 1978 (ref. 1). In 10 years The restricthe number of airports with restrictions doubled. tions take several forms: curfews on nighttime operations; flight routing and operating restrictions; and the use of prefthe Federal Aviation Adminerential runways. During this time, istration (FAA) has issued increasingly stringent noise certification standards that are critical design constraints on new NASA has been engaged in aircraft. To help with the problem, research-and-technology studies to advance the state of tne art The primary noise source is the in aircraft noise reduction. turbofan engine. Propulsion fan engine's reduced in the engine objective that will noise research is focused on understanding the turboso that noise can be noise-producing components efficient and economical ways that do not penaiize An additional performance or weight significantly. is to develop prediction procedures for each source allow aircraft noise to be estimated accurately. in figthe the The as propbe reand the engine

The noise sources from a turbofan engine are illustrated ure 111-2. The sources are botn internal and external to engine. The internal sources are the fan, the compressor, turbine, the combustor, and flow over the support struts. last three sources have usually been considered collectively engine core noise. Sound from the internal sources must agate through the engine ducts and nozzles, where it can acoustic treatment duced with acoustic treatment. Thus, sound propagation in ducts are very important elements in noise reduction.

*NASA Langley

Researcn

Center,

Hampton,

Va. 85

The external noise sources are the high-velocity jets mixing An important aspect with each other and with the ambient air. of the engine noise problem is the effects of flight on the varAs will be shown, the effects of flight, or ious noise sources. differ with the noise source. forward velocity, The discussion presented nerein is The topics discussed are fan noise, noise technology, combustor noise, FAN NOISE The fan is a dominant noise source in current high-bypass-ratio particularly during the landing approach. turbofan engines, advanced turbofan design studies, such as those Furthermore, associated with the Energy Efficient Engine program, indicate that the fan will continue to be a dominant noise source in fuThe ultimate objectives of fan noise research are ture engines. to propose noise-reducing fan design features that are compatible with good aerodynamic performance and to experimentally Our apconfirm the acoustic effectiveness of those designs. proach includes research to understand the generation mechanisms and to describe in detail the fan source characteristics. Describing the source is important because propagation, suppression, and radiation all strongly depend on the initial conditions at the source. An important constraint on experimental work in static facilities is that the test environment must lead to noise levels that correctly simulate flight. Two primary source mechanisms that are addressed in research to reduce fan noise are shown in the turbofan cross section in figure 111-3. Rotor-stator interactions in the form of rotor wakes and vortices impinge on the stators and can be particularly important noise generators at the subsonic tip speeds that occur during landing approach. The corresponding narrowband spectrum is shown in the upper portion of figure 111-4. The blade passing tone and its harmonics, which are due to periodic interactions of the rotor wakes with the stator blades, are superimposed on the broadband levels that result from interactions involving random flow disturbances. Rotor-alone noise production occurs because of nonuniformities in the rotor-locked shock wave patterns which form at the leading edges at supersonic tip These patterns radiate multiple pure tones during takespeeds. off and have a spectrum of the type shown in the lower portion of figure 111-4. Multiple pure tones can occur at all multiples of the shaft rotation frequency, and some of the individual tone levels often exceed the level of the blade passing frequency and its harmonics. based on recent research. acoustic suppression, jet and aircraft noise prediction.

86

One of the concepts that has been investigated to reduce shockmultiple-pure-tone noise is to sweep the rotor-blade generated, An experimental swept-rotor fan designed to exleading edges. plore the acoustic performance of swept blades is shown in figThe acoustic design of this fan was performed by ure 111-5. and the aerodynamic and mechanical Bolt Beranek and Newman Inc., (refs. designs were done by the Lycoming Division of AVCO Corp. The fan was acoustically tested at the NASA Lewis 2 and 3). The blade leading edges are swept Research Center (ref. 4). forward to midspan and then backward to the tip in order to limThese stressit the maximum blade stresses at the blade roots. es would be unacceptably high if the sweep were only in one direction. which is used in supersonic The aerodynamic concept involved, is shown in figure 111-6. The components swept-wing aircraft, of the blade leading-edge Mach numbers are shown in the planform and the variation of these components view of tne swept blade, Distance along the along the blade span is shown in the plot. span is plotted vertically to correspond with the blade sketch. It is the component of Mach number normal to the blade leading The sweep rate edge tnat controls leading-edge shock formation. at each spanwise location is designed to keep the normal component of blade-inlet relative Mach number subsonic over the Even though the total inlet relative Mach number entire span. three-fourths of the swept blade, is supersonic over tne outer tne normal component is limited to a maximum Mach number of Except for blade end effects and the sweep reversal about 0.9. this should eliminate a major portion of the strong point, leading-edge shock system and thereby reduce the multiple-puretone noise. The acoustic results obtained with this swept design are shown in figure 111-7. The multiple-pure-tone power levels for a unswept fan and the swept-rotor fan are compared as a function of fan-tip relative Mach number. Rotor sweep delayed the onset of multiple pure tones to higher Mach numbers (about 1.25 instead of 1.0) and reduced the levels over a large portion of the tipThese speed range, including speeds representative of takeoff. initial results are encouraging, and refinement of the aerodynamic design may lead to further multiple-pure-tone reductions. In addition to the rotor-alone and rotor-stator interaction another strong source of noise is present in all static sources, rotor interaction with inflow disturbances, as shown testing: in figure 111-8. The rotor blades cut externally produced turbulence, wakes, or vortices that are drawn into the inlet. At subsonic tip speeds, this source often obscures or completely masks the rotor-stator interaction source.

87

The reasons for the prominence of the inflow source statically and its greatly reduced importance in flight (ref. 5) are illusThe nature of the fan-inlet fiow field trated in figure 111-9. for static and flight cases is shown on the left side of figure 111-9, and the corresponding fan spectra are shown on the turbulence in the atmosphere and wakes and Statically, right. vortices from the proximity of the test stand and ground plane are dr,awn into the inlet through greatly contracting stream The contraction intensifies transverse turoulent fluctutubes. ations and stretches the disturbances axially so that the rotor blades cut each intensified disturbance many times. Tone bursts are generated that appear as a strong blade passing tone and In contrast, harmonics in the fan spectrum. with forwara velocthe stream tubes no longer contract to intensify ity in flight, and elongate atmospheric turbulence, and test-stand and groundThus, plane disturbances are not present. for fan stages that have been designed to limit noise caused by rotor-stator interthe tone levels - particularly those of the fundamental action, tone - are greatly reduced. To eliminate extraneous inflow disturbances during static tests, passive inflow control devices of the type shown in figure III-10 are being developed (refs. 6 to 8). A hollow, hemispherical structure surrounds the fan inlet to damp incoming turbulence and to smooth out local distortions. The structure consists of layers of honeycomb and screen. The honeycomb cells are essentially alined along streamlines, and the size of the structure (about 4 fan diameters) was chosen to keep the pressure drop low and to avoid affecting the acoustic transmission to the far field. The frontal view shows that a segmented construction was used to approximate a hemisphere. The segmented honeycomb surface of the inflow This device has been shown in figure 111-11. indoor anechoic chamber fan tests and outdoor Figure III-12 shows the device mounted around the Lewis anechoic chamber. control device is studied in both engine tests. the fan inlet in

Typical aeroacoustic results obtained in the anechoic chamber tests are shown in figure III-13 in terms of measured inlet turbulence intensities and scales and the narrowband fan noise With inflow control the transverse turbulence intenspectra. sities were reduced by more than a factor of 5. Axial length scales were also reduced and this indicates much less axial elongation of inflow disturbances. The narrowband spectra show large fan-tone reductions, particularly at the blade passing The fan tested was designed to minimize rotor-stator frequency. interaction and the results show that a large portion of the rotor-inflow disturbance noise that controlled the tone levels without inflow control was removed.

88

Development of inflow control methods (ref. 9) is also proceeding on the outdoor engine stand shown in figure 111-14. A JTlSD engine is shown fitted with the inflow control device. An exhaust muffler is used so that the inlet noise can be measured These tests are part of an intercenter program inseparately. volving the Lewis, Ames, and Langley Research Centers. Related NASA-sponsored work is also going on at engine and aircraft manufacturers. The purposes of the program are to measure the effects of forward velocity on fan noise and to develop static test methods that correctly simulate the measured flight behavior. A key feature of the program is the use of the same turbofan engine source in four test situations: the outdoor static engine tests and the indoor anechoic chamber fan tests, both to be done at Lewis; wind tunnel tests at Ames; and flyover noise measurements at Langley. The flight noise will be obtained by using an OV-1 aircraft as a flying testbed, as illustrated in figure 111-15, which shows an artist's conception of the planned JT15D engine installation under the wing. The flight results will serve as the standard by which the adequacy of the ground-test methods can be verified. As the inflow control methods are perfected, the rotor-stator interactions and any other sources controlling flight fan noise These are the sources will be clearly revealed in static tests. They also define the source to be minimized by fan design. characteristics that are the starting point for acoustic suppressor analyses and experiments. ACOUSTIC SUPPRESSION is aimed at making tnem more Basic research on noise suppressors efficient. It is believed that the weight gains and engine perThis research is formance losses they cause can be reduced. going on at the NASA Langley and Lewis Research Centers and at all the major engine and airframe companies. Summary of Sound Propagation Theory

The components of suppressor analysis are shown in figure 111-16. The The noise source is pictureu at the far right. characteristics of the source must be known to build a good suppressor. as the Some of the other phenomena are illustrated Tne sound is scattered at tne insound waves move to the left. terface between the hard wall and the acoustic liner. 'The sound More details of the is absorbed as it passes through tne liner. The usual duct liner as a boundary condition are shown later. geometry includes a variable area so tnat the sound is scattered and reflected, and possibly refracted because of the velocity gradients. The sound reaches the inlet lip, is furtner scatThe far-field tered, and finally radiates to the far field. 89

sound level directivity is most important since the far field is where the observer is located. All the elements mentioned have an influence on far-field directivity. One approach to solving the complete problem is to attack the entire system at once by using numerical techniques. This is a long-range solution that will not be considered here. Just a few of the elements on this figure are briefly discussed here. A simplified picture of a single sound wave traveling in a duct is shown in figure 111-17. Of course, many such waves at different angles would be present. Two angles associated with this wave are very important. The incidence angle of the wave on the liner is labeled. The other angle, between the wavefront normal and the duct axis, is called the axial propagation angle. The incidence angle correlates the optimum wall impedance (ref. 10); the axial propagation angle is related to the far-field radiation direction (ref. 11). The two angles, together with the determine the suppression that can be obtained. wall impedance, The correlation of optimum resistance with incidence angle is shown in figure III-18 (from ref. 10). All possible modes or angles of propagation were considered. Frequency and Mach number were held constant. A cylindrical duct was used with three boundary-layer thicknesses. For each boundary-layer thickness the calculated "best resistance for maximum damping" is very well correlated. Also, the boundary layer has an effect only for incidence angles above 650, which corresponds essentially to axial propagation. Similar correlations could be shown for the other component of wall impedance - the imaginary part, or Equations are available to describe the optimum imreactance. pedance behavior illustrated here (ref. 12). The axial propagation angle can be used to modify the existing inlet radiation theory, which is quite limited. Only zero flow or the same flow inside and outside the duct can be handled exactly. In a static engine test, there is a large flow Mach number in the inlet but essentially zero flow outside the duct. The radiation theory must therefore be modified (ref. 11). The results of such a modification are shown in figure III-19 (ref. level is plotted against the angle 13) I in which sound pressure from the inlet axis. Forty-one rods mounted in front of the rotor of a JT15D engine produced a single mode or angle of propagation to keep the experiment simple. The previous theory, which does not fit the data, is shown by the dashed line. The modified theory, which uses the axial propagation angle and fits the data quite well, is shown by the solid line. An additional correction for refraction appears necessary when the principal lobe of radiation falls near the inlet axis (ref. 13). The improvement in liner concepts is illustrated 90 efficiency in figure with III-20 the newer (from ref. suppressor 14). Sound

power reduction in the highest l/3-octave band is plotted The two liners were against the liner length-diameter ratio. tested on Quiet Engine C and the Lycoming YF-102 engine, which happened to be the available engines at the time of the tests. The old liner (plane-wave design) was quite long and was tested in three lengths. .The new liner (spinning-mode design) was very The increased suppression short and only one length was used. was quite dramatic, about a 3-to-l improvement. This improvement comes about by a better match of the liner to the estimated properties of the noise source. Extended-Reaction Liners

The final subject covered is sound propagation within the suppressor material itself and the resultant effect on the farAs illustrated in figfield sound pressure level suppression. ure III-21 the liner acts as a boundary condition that must be applied to the propagating sound waves within the duct. Two broad distinctions in the type of liners that can be built are shown in figure 111-22. On the left is shown a liner made of perforated plate bonded to honeycomb. This is a pointreaction liner, which means that sound can travel in the liner only in the direction normal to the liner. This type of suppressor provides an acoustic impedance type of boundary condition (as implied in fig. 111-21) and is the liner commonly Other variations could use wire used in aircraft applications. The pointmesh or other fibers bonded to the faceplate. reaction liner is often called a single-degree-of-freedom (SDOF) liner. A second type of liner, called an extended-reaction Sound is free to liner or bulk absorber, is shown on the right. travel in all directions in the bulk absorber, and impedance boundary conditions no longer apply. The wave equation must be solved both in the bulk material and in the duct, and these Tne bulk solutions must be matched at the common interface. absorber has been known for quite some time to provide a wider The probattenuation bandwidth than the point-reaction liner. lems with this liner have been associated with its wicking propor the ease with which it soaks up water, oil, and erties, This causes.the liner to lose its sound-absorbing propfuel. erties and can also be a safety problem. Recent work at NASA Langley (ref. 15) and in the aircraft industry has shown that some fibrous materials can be made nonwicking and thus flightTherefore, a detailed study of the physics of bulk worthy. absorbers is justified and is being done under a university grant at the University of Texas, Austin, and under a contract with Hersh Acoustical Engineering, as well as in house. A Lycoming lets were YF-102 engine on a test tested is shown in figure stand where bulk absorber 111-23. The large aft in91

suppressor assures that the inlet noise. Three inlet suppressor panels

aft fan and jet noise do not contaminate densities of Kevlar were tested in the over the full speed range of the engine.

The sound power attenuation as a function of frequency is shown in figure III-24 for the best bulk-absorber and perforated-plate honeycomb liners that were tested. More extensive results are Both liners were designed for a blade reported in reference 16. passing frequency of 5000 hertz, and their attenuations are about the same at that frequency. At higher and lower frequenThe higher cies the bulk absorber has a larger attenuation. frequency improvement can be explained by the difference in the acoustic reactance between the two materials. However, at the lower frequencies both materials have about tne same acoustic The extended-reaction properties of impedance characteristics. the Kevlar liner must be invoked to explain the low-frequency improvement. Just a few of the efforts to better understand the details of This suppressor performance have been discussed in this paper. improved understanding has resulted in better suppressors, and it can be expected that even more efficient suppressors will evolve. JET NOISE TECHNOLOGY For jet-powered aircraft the most important noise source at Considerable researcn takeoff is usually the jet exhaust noise. has been conducted over the last several years on the jet noise This section summarizes some reproblems of subsonic aircraft. cent advances in jet noise technology applicable to advanced supersonic cruise aircraft and progress toward understanding flight effects on jet engine exhaust noise. Inverted-Velocity-Profile Coannular Jets

coannular jets were A few years ago, inverted-velocity-profile identified as a breakthrough in jet noise suppression applicable As ilto advanced supersonic cruise aircraft (e.g., ref. 17). this approach is to exhaust the lustrated in figure 111-25, higher velocity stream through an annulus, with the lower veSuch velocity profiles can be oblocity stream in the middle. tained by crossducting the fan and core streams (e.g., ref. 18) or by burning in the fan duct and thereby increasing the fan stream velocity above that of the core stream (e.g., ref. 19). are further described in Engines incorporating tnese approaches paper XI, SUPERSONIC PROPULSION TECHNOLOGY. The noise benefits of the inverted velocity The normalized peak perceived figure III-26. 92 profile noise are shown level is in

plotted against the mass-averaged jet velocity (ideal specific thrust) for several of the many configurations tested (refs. 20 and 21). A reference curve is also shown for a hypotnetical, perfectly mixed conical nozzle. For all these coannular nozzle data, the outer-stream velocity is 1.5 to 2.0 times that of the inner stream. Noise reductions with the coannular nozzles, relative to the conical nozzles, generally improve as the ratio of the inner diameter to the outer diameter of the outer stream increases. The area ratio and the velocity ratio between the two streams also influence the noise levels at a given massaveraged velocity. Further results are discussed in references 20 to 32. Since the noise is a complicated function of flow-field and geometric parameters, it is necessary to go beyond simple plots such as figure III-26 to correlate the data (e.g., refs. 33 to 35). The complexity of the inverted-velocity-profile jetnoise-generating processes is illustrated in figure 111-27. As The many as four noise-generating regions must be considered. low-frequency noise is generated well downstream of the nozzle, where the two flows have mixed and can no longer be distinguished; The higher frequency jet this is the merged region. mixing noise is generated near the nozzle exit, where the individual jets can still be identified: this is tne premerged region. When either or both streams are supersonic, noise can be generated by turbulent eddies passing through shock waves: thus, we must in general consider both inner-stream and outerstream shock noise. We have developed empirical models relating these noise-generating processes to those of a simple conical nozzle (refs. 33 to coaxial-nozzle experimental spectra 35). Small-scale, plugless, (ref. 30) are compared with predictions based on the empirical In figure model of reference 33 in figures III-28 and 111-29. 111-28, sound pressure level is plotted against frequency for an For this case both in the rear quadrant. angle of 120, so all four noise sources must be constreams are supersonic, The sidered. But it is the jet mixing noises that dominate. shock noises, predicted by an empirical modification to the theory of Harper-Bourne and Fisher (ref. 36), contribute somebut not as much as the prewhat in the high-frequency range, Figure III-29 shows results for the same merged mixing noise. Shock noise conditions, but in the forward quadrant at 75O. The inner-stream shock noise becomes much more important here. dominates the midfrequency range and determines the peak sound The outer-stream shock noise controls the hignpressure level. Although the relative contributions of tne frequency range. various sources differ in the forward and rear quadrants, the spectra at both angles are predicted with good accuracy.

93

Typical results for an engine with a coannular plug nozzle are The perceived noise level is plotted as shown in figure 111-30. The symbols a function of angle at a typical sideline distance. denote model-scale experimental data obtained by General Electric under NASA Lewis sponsorship (contract NAS3-19777) scaled up,to engine size. The solid curve represents the total noise predicted from the empirical model of reference .33, the dashdotted curve represents the predicted mixing noise from both the merged and premerged regions, and the dasned curve represents the predicted shock noise from both streams. Although the jet mixing noises are most important in this case, the shock noises do contribute somewhat in the forward quadrant. At higher power settings and in flight, the shock noises become even more important and can contribute significantly to the effective perceived noise level. As discussed in paper XI, SUPERSONIC PROPULSION TECHNOLOGY, these results have now been verified statically at larger scale on the NASA/General Electric variable-cycle-engine testbed. It is also quite significant that the inverted-velocity-profile noise oenefits have been verified under simulated flignt conditions in model-scale tests conducted by Pratt & Whitney under NASA Lewis sponsorship (ref. 37). Flight Effects on Jet Noise

The subject of flight effects on jet noise has been a rather controversial one in recent years. Some of the terminology needed to describe flight effects is defined in figure 111-31. The cases considered herein are level flyovers at an airplane velocity of VU. The observer is located at an angle 8 from the engine inlet axis. In the present discussion, only the effect of flight on the overall sound pressure level is conSpectral effects, sidered. whicn usually are not very signifare not discussed. icant, Only conventional engines are considered, although analogous results have been obtained with inverted-velocity-profile jets. According to classical jet noise theory, in-flight jet noise should foilow a fairly simple relation, as the velocity arrows at the bottom of figure III-31 suggest. For a given absolute Vj (shown by the upper, longer arrow), injet velocity creasing the flight velocity VU (shown by the lower arrow) reduces the velocity of the jet relative to the air. This reduces the shear, and therefore the noise shouid be less in flight. The current interest in flight effects was greatly few years ago when Rolls-Royce (refs. 38 and 39) results like those shown in figure 111-32, where sound pressure level is plotted as a function of 94 stimulated reported the overall angle. The a

static case is shown by the solid curve, and the corresponding The noise in the flight case is shown by the dash-dotted curve. in some cases, However, rear quadrant was reduced, as expected. the noise in the forward quadrant such as the one shown here, Further confusing the issue is the fact increased in flight. that model-jet simulated flight tests indicate that in-flight noise should be reduced at all angles, as shown by the dasned Studies conducted or sponsored by NASA suggest that curve. these apparent anomalies can be resolved when the engine inThe internal noise ternal noise is considered (refs. 40 to 46). is amplified in the forward quadrant in flight by a sufficient 47) that the total in-flight noise exceeds the amount (ref. static level even though the jet noise is reduced. As can be inferred from the preceding discussion, the type of flight effects to be expected depends on the relative levels of To demonstrate this point, jet mixing noise and internal noise. figure III-33 shows the flight effects for engines with different levels of internai noise relative to jet mixing noise. On the left are overall sound pressure level (OASPL) data as a function of angle for an HS-125 airplane with a Viper 610 engine about 1100 feet per second (ref. at relatively low jet velocity, this represents relatively "high" internal noise. The pre38); For dictions are based on semiempirical NASA methods (ref. 43). this case a forward-quadrant noise increase is both predicted On the right is a similar plot for a DC-9 airand observed. plane with refanned JT8D engines at a jet velocity of about 1500 this represents relatively "low" infeet per second (ref. 41); For this case, ternal noise. noise reductions in flight are predicted and observed at all angles. Comparing improved prediction methods (ref. 48) with flight data shows agreement, with a standard deviation of about 1.5 decibels. The major developments in jet noise technology can be summarized Inverted-velocity-profile noise reductions have as follows: been demonstrated for a wide range of nozzle geometries at model scale and have now been verified statically at larger scale on the variable-cycle-engine testbed. Furthermore, invertedvelocity-profile noise reductions have been verified under simulated flight conditions in model-scale tests. Flight effects have proven to be predictable with reasonable accuracy wnen all the noise sources are properly accounted for. COMBUSTOR NOISE After the fan and the jet, the engine core is tne next most significant source of noise. Figure III-34 shows data from the YF-102 engine that illustrates this fact. When fan noise is reduced, the engine noise signature tends to be dominated by low-frequency noise, which is shown in figure III-34 as acoustic power data. This low-frequency noise comes from the jet and the 95

Moreover, the levels of the jet.and core noises are comcore. parable and intersect somewhere in the normal operating range of the engine, as shown. at low engine power setAs a result, tings, core noise tends to dominate. Low-frequency core noise is essentially all from the combustor. Consequently, this discussion focuses on combustor noise generation and propagation. Research for other engine noise sources is fairly mature and is focused on specific critical problems. Combustor noise research at Lewis is in a formative stage characterized by rather general objectives at this time. These objectives include the acquisition of a valid noise data base and information on the nature of combustor noise generation and propagation. From this information, empirical or analytical models of noise generation and propagation can be developed to give a predictive capability that in part will support the NASA noise prediction objectives The entire program, of course, points toward discussed here. the ultimate goal of developing noise reduction methods. Examination of engine operation reveals a number of problem areas associated with combustor noise research. Principal among these is the combustor itself. Since the combustor is an acoustic source, the noise generated is influenced by the environment in which it operates - in acoustical terms, by the impedances at its boundaries. Operation of a combustor outside an engine in general will be acoustically different from in situ operation. definitive information on combustor noise-generation Therefore, characteristics must be derived from actual engine operation. it must propagate through the turOnce the noise is generated, bine and then through the nozzle, both of which can subject it It then propagates through the flow and snear to attenuation. where refraction and scattering are possible, layers of the jet, Obviously, before it radiates to the far field. actual engine operation also is the only definitive source of combustor farfield noise. This creates an additional problem area since it requires some ingenuity to discriminate and measure only combustor noise in the presence of the other engine noises. As a practical matter, testing engines is costly and this proTherefore, component tests hibits parametric investigations. are mandatory in conducting research. (This is true of combustor source noise research as well as of turbine and nozzle propagation research.) Combustor component tests allow parametric investigations of noise-generating mechanisms at tolAlso, erable cost. they can be used to obtain acoustic power However, since it is likely that a combustor will measurements. generate a different noise in a rig than in an engine, an important problem associated with component testing is the need to understand any anomalies of combustor operation in test rigs. Some preliminary findings relating to this point are given in reference 49. 96

Direct Measurement of Combustor by Correlation Techniques

Far-Field

Noise

An important problem is the detection and measurement of only combustor far-field noise in the presence of noise from other engine sources. Figure III-35 schematicaliy illustrates the YF-102 engine and microphones used to obtain far-field noise. The usual method of obtaining engine comoustor: noise is to measure the total noise signature and then to use the properties shown in figure III-34 to infer from the changing slope of the data with engine speed when the noise is dominated by the core. This method has obvious drawbacks, one of which is the limited range of power settings for which it is useful. A more direct way of measuring far-field comuustor noise from an engine has been demonstrated. It has the advantage of being more definitive than the traditional method and is not limited to engine power settings where combustor noise is dominant. The method makes use of internal fluctuating pressure and the total far-field noise to obtain the far-field combustor noise. Some results from YF-102 engine tests are shown in figure III36. The top graph shows the coherence between the internal and far-field signals. The coherence is the square of the crossspectrum between the internal and far-field signals normalized It is the by the product of the internal and far-field spectra. The frequency-domain analog of tne cross-correlation function. coherence function can have a value only between 0 and 1 and can be interpreted in this case as the fraction of mean-square acoustic energy in the far-field signal due to the internal signal. In the bottom graph, the upper curve is the engine total far-field spectrum as measured directly by a single far-field microphone. By applying the coherence function to this spectrum, that portion which is identifiable with the combustor is extracted, as shown. This measurement technique is not novel, but this is thought to be its first use to obtain combustor noise from a complete engine. A significant contributor to the success of the technique was the development of the probes used to obtain the fluctuating internal pressure. This method provides a direct way of measuring combustor far-field noise from engines in order to contribute to a valid data base. This work is discussed more completely in references 50 to 52. Combustor Acoustic Power

A central problem in the combustor noise field is to determine the acoustic energy generated by the source. An analytical capability to do this is far beyond the state of the art, and current efforts are directed at developing empirical methods from test data. Figure III-37 shows much of the available data 97

L-

and how the acoustic power generated is related in some manner Data have been generated for a varito the heat release rate. and these data are largely from comety of combustor types, Equally important as the quantity and variety ponent rig tests. of data is the discovery of normalizing parameters for the acoustic power and the heat release rate that collapse the data The normalizing parameters are not exonto a reasonable curve. but they involve the various deplicitly noted, for simplicity, Although some sign and operational variables of the combustors. it is not unreasonable considering the data scatter remains, wide variety of combustors included and the present state of the This empirical correlation is a significant step toward art. The correlation and its developdeveloping a combustor model. ment are discussed fully in reference 53. In addition to combustor far-field noise and combustor source propagation of low-frequency core noise through acoustic power, and jet flow is receiving attention. Sigthe turbine, nozzles, nificant programs have been completed or are under way in these areas under contracts to the General Electric Co. and the Recent work completed is reported in refLockheed-Georgia Co. erences 54 and 55. AIRCRAFT NOISE PREDICTION: STATUS AND FUTURE APPLICATIONS

Designers of new aircraft have been forced in recent years to Tradeaccept propulsion noise as a critical design constraint. offs among many variables, including noise, must be considered before an aircraft that meets all mission requirements emerges An accurate, validated from the preliminary design process. system for aircraft noise prediction is clearly required. aircraft noise prediction is As indicated in figure 111-38, First, the aircraft trajectory based on three calculations. must be calculated from engine cycle and aerodynamic data. the noise characteristics of the various propulsion system Next, atmospheric components must be calculated and summed. Finally, propagation and ground effects must be calculated to represent the properties of the noise that actually reaches an observer (or microphone) on the ground. the Aircraft Noise Prediction A comprehensive computer program, implements the steps showrin tEe first Program (ANOPP), that The diagram shown figure has been developed by NASA (ref. 56). From enin figure III-39 illustrates the computational flow. gine cycle and aircraft aerodynamic data, the aircraft traFrom the engine jectory is computed as a function of time. thermodynamic state variables together with certain physical For the component noise characteristics are calculated. data,

98

the noise-critical parameters required by the example, combustor-noise prediction module include the mass flow rate, the combustor inlet temperature and pressure, and the turbine inlet and exhaust temperatures. For each time interval or airthe component noise characteristics are calcucraft position, lated and summed to give the total noise produced by the engine. Propagation and ground effects are then calculated to produce received-noise characteristics in the form of l/3-octave Levels for spectra at one or more arbitrary observer .positions. any of the subjective noise scales are then calculated and plotted. The ANOPP noise prediction modules and Flight Research Langley, Lewis, and updated by Langley and Lewis. ANOPP is available ANOPP is Cybernet currently to other available commercial have been developed by the Centers and are maintained

operational at Langley, where it is also NASA Centers through remote terminal access. to other users through the Control Data Corp. data processing network.

Some recent comparisons of ANOPP-predicted noise levels with measured data are shown in figures III-40 to 111-42. The first comparison, shown in figure 111-40, is for the turbojet-powered Concorde. There is good agreement between calculated and measured perceived noise levels plotted as a function of the directivity angle to the engine inlet 8. The spectra at a direcup to about 1 kilohertz. The distivity of 130 agree well crepancy at higher frequencies may be due to the omission of turbomachinery noise from the calculated levels. The next comparison, shown in figure 111-41, is for the lowbypass-ratio-powered DC-g. The calcuiatlons for tone-corrected perceived noise levels agree well with the measured data except for peak levels at a directivity of 120 where a 3- to 5-PNdB difference exists. ANOPP underpredicts the effective perceived noise level try 1.2 EPNdB. Tne precilcted and measured spectra at a directivity angle of 120 would be neariy coincident except for tne unexplained consistent difference in level. The final comparison, shown in figure 111-42, is for the highbypass-ratio-powered DC-10 at takeoff. The tone-corrected perceived noise levels and spectral comparisons show a slight average overprediction. ANOPP overpredicts the effective perceived noise level by 1.6 EPNdB. The noise levels shown in this figure are based on data from McDonnell-Douglas, who also made the ANOPP calculations. For all the previous aircraft, NASA had no role in the data acquisition and reduction process; and, for all cases, the values of the noise-critical parameters were not availaule as 99

Conrecorded data but were estimated by,others ex post facto. sidering the uncertainties inherent in. the measured noise levels and in the estimated values of the engine cycle parameters, the agreement between the measured flyover noise levels and the ANOPP-predicted levels is encouraging. However, because severe economic penalties result from each decibel of noise design marthe highest level of pre,diction accuracy is absolutely esgin, sential. ANOPP should continue to be developed until this goal has been met. As indicated, propulsion noise must be treated as a critical constraint during the preliminary aircraft design process. application of ANOPP to preliminary design systems studies parametric analyses is illustrated in figure 111-43. The or

A few of the key dimensionless variables are the thrust-weight ratio the propulsion system; the lift-drag T/W, which sizes represents the aircraft's aerodynamic ratio CL/CD I which characteristics; and the normalized specific thrust T/fiCa r which is an indicator of source noise. The interrelationships among these and other dimensionless variables must be carefully studied before the ultimate compromise between noise at the FAA certification points, performance, and economics can be reached. System trade-off studies involving aircraft noise often result in the identification of noise-qenerating phenomena or propagation effects that require individual research before final The role of ANOPP in identifying conclusions can be reached. critical noise research areas has been illustrated in a recent International Civil Aviation Organization (ICAO) study related to future SST noise standards. As indicated in figure 111-44, the effects of shock cell noise, advanced or automated flight procedures, and excess ground attenuation all emerged from the ICAO study as areas with unacceptably high levels of uncertainty. For example, the intensity of shock cell noise when the exhaust nozzle is operating at offdesign conditions is not fully understood. the noise Likewise, reductions actually obtainable from the use of flight-profile optimizing techniques have never been demonstrated. Finally, the actual amount of excess ground attenuation, expecially at shallow angles, is the subject of a great deal of controversy. Research activities that focus on each of these phenomena are in progress at the Langley Research Center. The value of ANOPP for design studies and, consequently, for quantifying the benefits of proposed noise reduction technology The Langley and Lewis Research Centers has been established. are committed to continued cooperative development and improvement of ANOPP for application to future parametric and pre-

100

liminary design studies of advanced aircraft system concepts. This would include executive system improvements, new prediction modules, and component and system vaiidation. SUMMARY Fan Noise Shock-related multiple-pure-tone noise from fans can be controlled to some extent by sweeping the rotor-blade leading edges. The concept is also applicable to the high-speed turboprop (see paper I, ACEE Status Report). Inflow control devices appear to be an adequate approach to simulating the in-flight characteristics of fan noise in static testing. Experiments relating to the development of inflow control devices have also provided a great deal of insight into the several sources of fan noise that had not been fully appreciated. Acoustic Substantial behavior show that, adequately suppressor standing fan source standing pression Jet Noise Suppression advances have been made in understanding suppressor and in simplifying suppressor concepts. The results when the various factors in a suppressor design are there can be a large improvement in accounted for, It now appears that improved underperformance. coupled with the new understanding of of suppression, noise, will soon permit us to achieve the longobjective of jointly optimizing the acoustic supand the fan design for low noise. Technology

The noise reductions associated with inverted-velocity-profile coannular jets have been demonstrated for a large number of nozzle geometries at model scale and have now been verified at 'These larger scale on the variable-cycle-engine testbed. inverted-velocity-profile noise reductions have also been verin model-scale tests. ified under simulated flight conditions Flight effects on jet noise and low-frequency noise are now fairly well understood, with experimental static-to-flight noise increments agreeing with prediction within a standard deviation of about 1.5 decibels.

101

Combustor

Noise

Combustor noise has been fairly well defined over the past few The sampling of results discussed is representative of a years. larger collection of new data and findings that has been develThese new findings relate to the combustor acoustic power oped. acoustic transmission through the turbine and nozgeneration, and the development of new measurement techniques and a zle, From these new results a fairly comprenensive model data base. This is a necessary step of combustor noise can be developed. toward reducing this noise. Aircraft Noise Prediction

A working ANOPP computer program is now avallable that appears to provide an acceptable level of accuracy for CTOL aircraft with low- or high-bypass-ratio engines and for SST aircraft. The usefulness of the program in defining areas needing research has also been demonstrated.

REFERENCES 1. The State of Aircraft Russell, R. E.: Presented at the EPA Noise Technology Dallas, Texas, Jan. 29, 1979. Bliss, D. B.; et al.: Source Noise Transonic 1976. Noise Technology. Research Symposium, for a Novel Low Paper 76-577, July

2.

Design Considerations AIAA Fan Stage.

3.

Analysis and Design of a High Speed, Hayden, R. E.; et al.: Low Noise Aircraft Fan Incorporating Swept Leading Edge (BBN-3332, Bolt Beranek and Rotor and Stator Blades. Newman Inc., NASA Contract NAS3-18512.) Lucas, J. Acoustic 78-1121, G.; Woodward, R. P.; and MacKinnon, Evaluation of a Novel Swept-Rotor July 1978. (Also NASA TM-78878.) M. J.: Fan. AIAA Paper

4.

5.

Feiler, C. E.; and Groeneweg, J. Velocity Effects on Fan Noise. (Also NASA TM-73722.) 1977. Shaw, L. M.; et al.: in an Anechoic Wind Flow Control Device. NASA TM-73723.)

F.: Summary of Forward AIAA Paper 77-1319, Oct.

6.

Inlet Turbulence and Fan Noise Measured Tunnel and Statically with an Inlet AIAA Paper 77-1345, Oct. 1977. (Also

102

Effectiveness 7. Woodward, R. P.; et al.: Turbulence Control Device to Simulate NASA TM-73855, an Anechoic Chamber.

of an Inlet Flow Flight Fan Noise 1977.

in

Reduction of Rotor8. Kantola, R. A.; and Warren, R. E.: Turbulence Interaction Noise in Static Fan Noise Testing. AIAA Paper 79-0656, Mar. 1979. 9. Jones, W. L.; McArdle, J. G.; and Homyak, L.: Evaluation of Two Inflow Control Devices for Flight Simulation of Fan AIAA Paper 79-0654, Mar. Noise Using a JT15D Engine. (Also NASA TM-79072.) 1979. Modal Propagation Angles in Ducts with Soft Rice, E.J.: Walls and Their Connection with Suppressor Performance. AIAA Paper 79-0624, Mar. 1979. (Also NASA TM-79081.) Rice, E. J.; Heidmann, M. F.; and Sofrin, T. G.: Modal Propagation Angles in a Cylindrical Duct with Flow and Their Relation to Sound Radiation. AIAA Paper 79-0183, Jan. 1979. (Also NASA TM-79030.) Rice, E. J.: Optimum Wall Impedance for A Correlation with Mode Cutoff Ratio. no. 5, May 1979, pp. 336-343. Spinning J. Aircr., Modes vol. 16,

10.

11.

12.

13.

Heidmann, M. F.; Saule, A. V.; and McArdle, J. G.: Analysis of Radiation Patterns of Interaction Tones Generated by Inlet Rods in the JT15D Engine. AIAA Paper 79-0581, Mar. (Also NASA TM-79074.) 1979. Jones, W. L.; and Groeneweg, J. F.: State-of-the-Art of Turbofan Engine Noise Control. NOISE-CON 77, G. C. Maling, Jr., ed., Noise Control Foundation, 1977, pp. 361-380. Smith, C. D.; and Parrott, T. L.: An Experimental the Effects of Water Repellant Treatment on the Properties of Kevlar. NASA TM-78654, 1978. Heidelberg, of Bulk 1979. L. J.; Absorber and Homyak, L.: Full-Scale Acoustic Inlet Treatment. Study of Acoustics

14.

15.

16.

Engine Tests NASA TM-79079, Supersonic Engines, Cruise AGARD

17.

Willis, Edward: Variable-Cycle Engines for Aircraft. Variable Geometry and Multicycle CP-205, 1976, pp. 7-l to 7-19.

18.

Krebs, J. N.: Advanced Supersonic Technology Study - Engine Program Summary: Supersonic Propulsion - 1971 to 1976. Proceedings of the SCAR Conference, NASA CP-001, Part 1, 1976, pp. 353-370. 103

13.

Variable Stream Control Engine Concept Howlett, Robert A.: for Advanced Supersonic Aircraft - Features and Benefits. Proceedings of the SCAR Conference, NASA CP-001, Part 1, 1976, pp. 341-352. Coannular Plug Nozzle Lee, Robert: Impact on Exhaust System Design. Conference, NASA CP-001, Part 2, Goodykoontz, Study of 79-0631, Noise Reduction and Proceedings of the SCAR 1976, pp. 505-524.

20.

21.

Jack H.; and Stone, James R.: Experimental Coaxial Nozzle Exhaust Noise. AIAA Paper Mar. 1979. (Also NASA TM-79090.)

22.

Stone, James R.; Goodykoontz, Jack H.; and Gutierrez, Orlando A.: Effects of Geometric and Flow-Field Variables on Inverted-Velocity-Profile Coaxial Jet Noise and Source Distributions. AIAA Paper 79-0635, Mar. 1979. (Also NASA TM-79095.) Sarohia, Inverted 1979. V.: Effect Velocity of Density on Noise Radiated from Profile Jets. AIAA Paper 79-0634, J.; and Lau, J. C.: of Inverted-Profile Co.; NASA Contract The Noise Coannular NASl-15018.) Mar.

23.

24.

Tanna, H. K.; Tester, B. and Flow Characteristics Jets. (Lockheed-Georgia NASA CR-158995, 1979.

25.

Flight Effects on Noise Generated Strout, Frank G.: by the JT8D Engine with Inverted Primary/Fan Flow as Measured in the NASA Ames 4O- by 80-Foot Wind Tunnel. NASA CR-2996, 1978. Cargill, istics Paper A. M.; and Duponchel, of Inverted Velocity 77-1263, Oct. 1977. J. P.: Profile The Noise CharacterCoannular Jets. AIAA Nozzle AIAA

26.

27.

Crouch, R. W.; Coughlin, C. L.; Exit Flow Profile Shaping for Paper 76-511, July 1976. Dosanjh, Coaxial 1976. D-S.; et al.: Cold/Heated Jet Supersonic Flows.

and Paynter, G. C.: Jet Noise Reduction. Jet-Noise AIAA Paper

28.

Suppression by 76-507, July

29. 30.

Knott, P. R.; et al.: Turbofan Jet Noise

Acoustic Simulation.

Tests of Duct-Burning NASA CR-2966, 1978. Allan B.: Turbofan Aerodynamic and Exhaust Nozzles.

Kozlowski, Hilary; and Packman, Acoustic Tests of Duct-Burning NASA CR-2628, 1976.

104

31.

Aeroacoustic Studies of Gutierrez, Orlando A.: Nozzles Suitable for Supersonic Cruise A.ircraft Proceedings of SCAR Conference, Applications. CP-001, Part 2, 1976, pp. 471-490.

Coannular NASA

32.

Coannular NOZZle Noise Characteristics Kozlowski, Hilary: and Application to Advanced Supersonic Transport Engines. Proceedings of the SCAR Conference, NASA CP-001, Part 2, 1976, pp. 491-504. An Empirical Stone, James R.: Profile Jet Noise Prediction. A Correlation Pao, S. Paul: Jets with Inverted Flow Model for Inverted-velocityNASA TM-73838, 1977.
COannUlar

33.

34. 35.

of Mixing Noise from NASA TP-1301, Profiles.

1979.

A Jet Exhaust Noise Prediction Larson, R. S.: Inverted Velocity Profile Coannular Nozzles. 79-0633, Mar. 1979.

Procedure for AIAA Paper

36.

The Noise from Shock Harper-Bourne, M.; and Fisher, M. J.: Noise Mechanisms, AGARD-CP-131, Waves in Supersonic Jets. to 11-13. 1974, pp. 11-l Flight Effects Kozlowski, Hilary; and Packman, Allan B.: on the Aerodynamic and Acoustic Characteristics of NASA CR-3018, 1978. Inverted Profile Coannular Nozzles. Brooks, Speed Paper Bushell, Flight. Stone, J. Noise. Merriman, Effects The Effects J. R.; and Woodrow, R. J.: on a Number of Turbojet Exhaust Silencers. 75-506, Mar. 1975. K. W.: AIAA Measurement Paper 75-461, and Prediction Mar. 1975. Flight of on Jet of Forward AIAA Noise in

37.

38.

39. 40. 41. 42.

Jet

On the Effects of R.: NASA TM X-71819, 1975. J. E.; et on Engine al.: Noise. Forward AIAA

Engine

Exhaust

Motion and Installation Paper 76-584, July 1976.

Blankenship, G. L.; et al.: (MDC-J7708, Engine Noise. NASA Contract NAS3-20031) Stone, for vol. Stone, for Prediction James R.: Turbojet and Turbofan 10, no. 1, Jan.-Feb. J. Jet

Effect of Forward Motion on Douglas Aircraft Co., Inc.; NASA CR-134954, 1977. of In-Flight Exhaust Noise Contr. Engines. 1978, pp. 40-46. Noise Eng.,

43.

44.

On the Use of Relative Velocity R.: NASA TM-78873, Engine Exhaust Noise.

Exponents 1978. 105

45.

Ahuja, K. K.; Tester, B. J.; and Tanna, Jet as a Simulator of Forward Velocity NASA CR-3056, 1978. Engine Noise.

H. K.: Effects

The Free on Jet

46.

Burcham, F. W., Jr.; Lasagna, P. L.; and Oas, S. C.: Measurements and Predictions of Flyover and Static Noise of a TF-30 Afterburning Turbofan Engine. NASA TP-1372, 1978. Morse, P. M.; McGraw-Hill and Ingard, K. Book Co., Inc., Theoretical U.: 1968. for Predicting NASA TM-79155, Acoustics. the Effects 1979.

47. 48. 49.

An Improved Stone, J. R.: of Flight on Jet Mixing Reshotko, Pressure Facility: Reshotko, Turbofan

Method Noise.

M.; and Karchmer, A. M.: Combustor Fluctuating Measurements In-Engine and in a Component Test NASA TM-73845, 1977. A Preliminary Comparison. M.; et al.: Core Noise Measurements of AIAA Paper 77-21, Jan. 1977. Engine, a YF-102

50. 51.

Karchmer, A. M.; and Reshotko, M.: Core Diagnostics on a Turbofan Engine Using Coherence Tecnniques. NASA TM X-73535,

Noise Source Correlation and 1976.

52.

Karchmer, A. M.; and Reshotko, M.; and Montegani, F. J.: Measurement of Far Field Combustion Noise from a Turbofan Engine Using Coherence Functions. AIAA Paper 77-1277, (Also NASA TM-73748.) Oct. 1977. von Glahn, U. H.: Correlation Levels Inferred from Internal Measurements. NASA TM-78986, of Comuustor Fluctuating 1978. Acoustic Pressure Power

53.

54.

Doyle, V. L.; and Matta, R. K.: Attenuation of UpstreamGenerated Low Frequency Noise by Gas Turbines. (R77AEG482, General Electric Co.; NASA Contract NAS3-19435.) NASA CR-135219, 1977. Matta, R. K.; and Mani, R.: Theory of Low Frequency Noise Transmission Through Turbines. (R77AEG570, General Electric Co.; NASA Contract NAS3-20027.) NASA CR-159457, 1979. Noise Prediction Raney, J. P.: NASA TM-78700, 1978. Technology for CTOL Aircraft,

55.

56.

106

NOISECONSTRAINTS AT MAJOR WORLD AIRPORTS

160 140 120 100 NO. OF AIRPORTS 80 60 40 20 m 1968

CURFEWS AIRPORTS FLIGHT PREFERENTIAL WITH NOISE ROUTING & RUNWAYS PROCEDURES r;-7s-7015 CONSTRAINTS Figure III-1

TURBOFAN ENGINE NOISESOURCES

rFAN JET NOISE :

I i I ; ,, I I I /ENGINE CORE INOISE (COMBUSTOR, ; ACOUSTIC TREATMENTl TURBINE. AND STRUTS)J COMPRESSOR/ NOISE Figure III-2

107

FANNOISESOURCES
T-ROTOR ALONE ,(LEADING EDGE SHOCK WAVES)

cs-79-1540

L ROTOR-STATOR INTERACTIONS (WAKES &VORTICES)

Figure III-3

FANNOISESPECTRA AT SUBSONIC AND SUPERSONIC TIP SPEEDS

SUBSONIC ,---BLADEPASSINGFREQUENCY 10 dB (I 1I

SOUND PRESS. LEVEL, dB I

SUPERSONIC . I /-

BLADE PASSINGFREQUENCY

FREQUENCY, kHz
Figure III-4

cs-79-1844

108

SWEPT-ROTOR FAN

Figure III-5

SWEPT ROTOR BLADE

/I NORMAL /-'RN' ' CCMPCNENT CF MR

.8

1.0

I_ .A 1.2 1.4 MACH NO.

1.6
cs-79-1839

Figure III-6

109

MULTIPLE PURETONEGENERATION OF UNSWEPT ANDSWEPT-ROTOR FANS

150 UNSWEPT FAN-\ MPT 140 -SWEPT-ROTOR FAN SOUND POWER 130 LEVEL, dB 120 110 1.0 1.1 1.2 1.3 1.4 1.5 TIP INLET RELATIVE MACH NO. Figure III-7 cs-79-1842

RANGE FOR ONSET OF MPT

FANNOISESOURCES
-ROTOR ALONE 'JLEADING EDGE SHOCK WAVES)

+-

--IL-,-

-LROTOR-STATOR INTERACTIONS (WAKES &VORTICES)

~~-79-1841

/' LROTOR - INFLOW DISTURBANCE INTERACTION (TURBULENCE, WAKES, VORTICES) Figure III-8

110

FORWARD VELOCITY EFFECTS ON INLETFLOW AND NOISE

FLIGHT -~ ,

FREOUENCY STATIC

CS-79-1845

Figure III-9

INFLOWCONTROL DEVICE
,-Cl.25 2 in. unhlcvrnnn R 7
/ /

in. GRID

BASE SCREEN & 0.05 in. GRID TURBULENCE REDUCING SCREEN

FILM PROBE

FRONTAL VIEW-SECTIONED CONSTRUCTION Figure III-10

CS-79-1846

111

INFLOW CONTROL DEVICE

Figure III-11

INFLOWCONTROL.DEVlCE IN ANECHOIC CHAMBER

Figure III-12

112

AEROACOUSTIC EFFECTS OF INFLOW CONTROL DEVICE

INLET TURBULENCE INFLOW TRANSVERSE AXIAL LENGTlCONTROL INTENSITY, SCALE, DEVICE ft % -WITHOUT ----WITH 1.1 .2 18.7 3.3 c

NARROWBAND FAN SPECTRA BPF 2x BPF I

FREOUENCY, kHz Figure III-13

CS-79-1843

INFLOW CONTROL DEVICE INSTALLED ON JT15D ENGINE

Figure III-14

113

OV-1 AIRCRAFTTO BE USEDFORJTl5D FLYOVER NOISETESTS

Figure III-15

COMPONENTS OF SUPPRESSOR ANALYSIS

y DUCT GEOMETRY \

SOUND

II I

L WALL BOUNDARY CONDITION (ACOUSTIC IMPEDANCE) C?-T?-l??l

L BOUNDARY CONDlTlON AT DUCT TERMINATION Figure III-16

114

COMPONENTS OF SUPPRESSOR ANALYSIS

INCIDENCE ANGLE y, SOURCE DESCRlPTlON (INPUT WAVE)

CS-79-1988

Figure III-U

EFFECT OFBOUNDARY LAYER THICKNESS UPON OPTIMUM RESISTANCE-INCIDENCE ANGLE CORRELATION

1. 0 .8 .6 .4 .2lo .l: .08: .06- -A0
.Mq

BOUNDARY LAYER THICKNESS/ DUCT RADIUS 0. 005

NORMALIZED OPTIMUM RESISTANCE

OR @Ah 8 @ % B 33
1

A .05
cs-79-1995

.02 20

:O

1~ -170 do5b 60 INCIDENCE ANGLE, deg

A,

80

90

Figure III-18

115

COMPARISON OF MEASURED AND PREDICTED TONEDIRECTIVITIES

0 lOO--DATA DUCT FLOW ALONE EQUAL FLOW EVERYWHERE > PREDCToNS JTlSD ENGINE 41 RODS BPF AT 8450 rpm

0 0 70 30 I I 90 60 ANGLE FROM INLET AXIS, deg Figure III-19 I 120

cs-79-1992

PERFORMANCE OFSUPPRESSORS BASED ON TWODESIGN METHODS

10 INLET QUADRANT SOUND POWER REDUCTION IN MPT PEAK l/S-OCTAVE BAND, dB 8 6

ENGINE)

.2 .4 .6 LENGTH TO DIAM RATlO OF TREATED SECTlON Figure III-M

cs-79-1994

116

COMPONENTS OF SUPPRESSOR ANALYSIS

T.S-79-1989

WALL BOUNDARY CONDlTlON (ACOUSTIC IMPEDANCE) Figure III-21

LINING MATERIALS ANDCONSTRUCTION

PERFORATED PLATEHONEYCOMB, SINGLE DEG OF FREEDOM(SDOF) Figure III-22

BULK ABSORBER, EXTENDED REACTlON

cs-7g-1990

117

AVCO-LYCOMING YF-102 ENGINE ONTEST STAND

Figure III-23

COMPARISON OF SOUND POWER ATTENUATION SPECTRA FORBULKAND SDOFLINERS

15

SOUND POWER AITENUATION, APWL, dB

-5 ( 50

100

200

500 1000 2000 FREOUENCY, Hz

5ooa 10 000 20 000

cs-79-1993

Figure III-24

118

INVERTED-VELOCITY-PROFILE COANNULAR JETS

CS-79-1850

Figure III-25

NORMALIZED PEAKPNL AS FUNCTION OF MASS-AVERAGED JETVELOCITY

STREAM

!\lORMALIZED PEAK PERCEIVED NOISE LEVEL, PNL

0

.90 .93

CS-79-1853

1000 1200 1500 2000 2500 3000 MASS-AVERAGED JET VELOCITY, ftlsec Figure III-26

119

INVERTED-VELOCITY-PROFILE JETNOISESOURCES

MERGED

PREMERGED

SHOCK SHOCK

SOUND PRESSURE LEVEL FREQUENCY Figure III-27

COMPARISON OF INVERTED-VELOCITY-PROFILE JETNOISE PREDICTION WITHSTATIC MODEL EXPERIMENTAL DATA

BOTH STREAMS SUPERSONIC REAR QUADRANT, B = 120 PREDICTED TOTAL PREMERGED -------------SHOCK (OUTER) -----SHOCK (INNER) MERGED --

EXPERIMENTAL

.2

.5

10 5 2 FREQUENCY, kHz Figure III-28

20

50

120

..- .

COMPARISON OF INVERTED-VELOCITY-PROFILE JETNOISE PREDICTION WITHSTATIC MODEL EXPERIMENTAL DATA

BOTH STREAMS SUPERSONIC FORWARD OUADRANT, 8 = 75 PREDICTED ------------------TOTAL SHOCK (OUTER) PREMERGED SHOCK (INNER) MERGED

120
SOUND PRESS. LEVEL, SPL, dB

EXPERIMENTAL

FREQUENCY, kHz Figure III-29

COMPARISON OF EXPERIMENTAL AND PREDICTEDPNL DIRECTIVITY

PREDICTED NOI SE 110 r - -e--w--TOTAL JET MIXING SHOCK 0 MODEL COANNULAR NOZZLE STATlC DATA SCALED TO FULL SIZE AT SIDELINE DISTANCE ^ ^

PERCEIVED NOISE LEVEL, PNL, PNdB

cs-79-1854

501 0

20

40 60 80 loo 120 140 ANGLE REFERREDTO INLET, deg Figure III-30

160

180

121

FLIGHT EFFECTS ON EXHAUST NOISE

GROUND PLANE----

/yl===a >vo
Figure III-31

cs-79-1851

TYPICAL FLIGHT EFFECTS ON EXHAUST NOISE

ROLLS-ROYCE DATA ------OVERALL SOUND PRESS. LEVEL, OASPL, dB ENGINE STATIC ENGINE FLIGHT SIMULATED FLIGHT, JET MIXING NOISE

I
10 30

I
130 150
cs-79-1852

50 70 90 110 ANGLE REFERREDTO INLET, 9, deg Figure III-32

122

STATIC AND FLIGHT DIRECTIVITIES FORDIFFERENT TYPEENGINES

"HIGH" INTERNALNOISE (VIPER 610 ENGINEIN HS-125 AIRPLANE, Vjlc,zl.O) r-f FLYOVER OASPL (JET PLUS INTERNAL), dB /
cs-78872

"LOW" INTERNALNOISE (REFANNED JT8D ENGINES ON DC-9 AIRPLANE, VjIC,"1.4) r-i

,I

F[LIG;T

(/,

0 30 60 90 120 150 180 0 30 60 90 120 150 180 ANGLE FROM ENGINEINLET AXIS, deg
Figure III-33

YF-102ENGINE FAR-FIELD ACOUSTIC POWER

160 -

ACOUSTIC POWER LEVEL, dB

110 30

I 40

I I I I 50 60 70 80 % ENGJNESPEED

I I 90 100

cs-7g-2060

30500 600 700 800900 EFFECTIVEJET EXHAUST VELOCITY, ftlsec Figure III-34

123

YF-102ENGINE TESTS

INTERNALPRESSURETRANSDUCERS

i -c--..-?o; 1

_-

Figure III-35

YF-102COHERENCE AND SOUND PRESSURE LEVEL SPECTRA

FOR 120' & 43% SPEED 1.00 .75 COHERENCE .50 .25 0 LL 70 60 SOUND PRESS. LEVEL, 56 dBlHz 40 30' 0 I 40 I 80 I. ENGINE TOTAL FAR-FIELD SPECTRUM F

COMBUSTOR COHERENCE
I I I I I IpI 120 160 200 240 280 320 360 400 FREQUENCY, Hz csm7g-2064

Figure III-36

124

CORRELATION OF COMBUSTOR ACOUSTIC POWER LEVELS INFERRED FROMINTERNAL FLUCTUATING PRESSURE MEASUREMENTS
170~ 165

t
160 NORMALIZED ACOUSTIC 155 COMBUSTOR 0 0 0 n

POWER LEVEL, 150 dB 145 -

SINGLE STAGE, COPLANAR 2-STAGE, PILOT ONLY SINGLE STAGE, NONCOPLANAR 140 2-STAGE I I 1-I 135 165 170 175 180 185 190 195 200 cs-7g-2063 NORMALIZED HEAT RELEASE RATE Figure III-37

PREDICTION METHODOLOGY

SOURCE INTENSITY

FLIGHT PATH

Y,
Figure III-38

/\

125

ANOPP CTOL NOISE PREDICTION

AIRCRAFT DATA ENGINE CYCLE FLIGHT DYNAMICS AIRCRAFT TRAJECTORY .

NOISE CRITICAL PARAMETERS-

SOURCE INTENSITYATMOSPHERIC AND GROUND EFFECTS/ NOISE PLOTS PROPAGATION TIME LOOP OASPL. PNL. EPNL AT OBSERVER LOCATIONS
L-5031-2

Figure III-39

ANOPP FLYOVER NOISE VALIDATION

CONCORDE / i:

10 P$

PERCEIVED NOISE

J-EVEL, JNdB$+++y+
8, deg

-J 180
Figure III-40

IK

2K 4K 8K

FREQUENCY, Hz

126

ANOPP FLYOVER

NOISE VALIDATION
DC-9

90 PERCEIVED O" NOISE '-EVEL. PNdB 80 70 C DATA SPL dB 50

AEPNL ANOPP-DATA= -1.2EPNdB 7OL ;-.L.I-I 20 60 100 140 180 8. deg Figure III-41

0 63

DATA

30 tl, 250 lK2KdK8K16K FREQUENCY, Hz

ANOPP FLYOVER

NOISE VALIDATION
DC-10 /7?

116 108 PERCEIVED NOISE 100 LEVEL, PNdB 92

r
t

ANOPP DATP

100 80 SPL dB 60

ANOP3

DATA =1.6 84 t 0
L-5031-5

EPNdB 140 180

40

I
20 60 100

i - i_l,L
63

-i250

.--

1K 2K dK SK

8. deg
Figure III-42

FREQUENCY. Hz

127

ANOPP PRELIMINARY

DESIGN SYSTEMS

STUDIES

1. PROPULSION SYSTEM

3.5 n. mi. 3. NOISE T lilCa i-J

2. AERODYNAMICS

(-9
cL cD Figure III-43

L-5031-6

ANOPP RESEARCH IDENTIFICATION

SHOCK CELL NOISE

ADVANCED FLIGHT PROCEDURES

GROUND EFFECTS

L-5031-7

ANGLE OF INCIDENCE Figure III-44

128

IV.

ALTERNATIVE Jack

JET AIRCRAFT Grobman

FUELS

National

Aeronautics and Space Administration Lewis Research Center

The first paper in this conference discusses NASA's research efforts to evolve energy-efficient aircraft technology and emphasizes tne fact that jet fuel costs have become the dominant factor affecting aircraft direct operating costs. The diminishing reserves of petroleum aggravate the problem of rising costs and, in addition, lead to concern about how we will be able to meet the growing need for jet fuel in the future. This paper discusses the changes in jet fuel properties that may occur in the future as a result of shifts in the supply and demand for fossil fuels, the effects that varying fuel properties would have on aircraft engines and fuel systems, and, finally, the technology that would be required to use these fuels. POTENTIAL CHANGES IN FUEL PROPERTIES

The options that we might have for producing jet fuel in the future are illustrated in figure IV-l. For the present, the aircraft industry is totally dependent on petroleum as a source of jet fuel. Jet fuel is currently produced by straight distillation followed by mild nydrotreating for removal of sulfur. Trends in the aromatic content of jet fuel produced in this manner are shown in figure IV-2. Aromatics, which have about half the hydrogen content of saturated hydrocarbons, have undesirable combustion characteristics as a jet fuel. An increase in aromatic content results in a corresponding reduction in the hydrogen content of a jet fuel. Tne average aromatic content has increased steadily over the past several years. This increase is due to the increased proportion of high-aromatic crude oils being used to produce jet fuel (ref. 1). The aromatic content is limited to a maximum of 20 percent by the current ASTM Jet A fuel specification; however, Jet A produced from some of the available sources of crude oil exceeds this limit. Jet A refined from a heavy Arabian crude oil has had aromatic contents as high as 22 percent, and projections indicate that Jet A refined from Alaskan crude oil may have aromatic contents as high as 25 percent. A waiver currently in effect permits the limited use of jet fuels with aromatic contents as high -as 25 percent. 129

There is increasing for aromatic content

pressure to as a result

raise the of these

specification trends.

limit

As the demand for the middle-distillate fuels such as jet fuel, diesel fuel, and heating oils increases, the supply of these natural distillation products will De exceeded by the demand Then it will become necessary to crack the heavier IV-l). (fig. ends of the barrel to increase the middle-distillate pool. Energy-intensive hydrocracking will be required to iimit the aromatic content of jet fuel produced in this manner (ref. 2). As the supply of natural petroleum diminishes, the next likely domestic source of fossil fuel would be shale oil. The production of jet fuel from shale oil requires relativeiy severe hydrotreating to remove undesirable impurities and to upgrade the hydrogen content. For shale oil, nitrogen removal is particularly important because of the relatively large quantity of organic nitrogen present in comparison with that in petroieum. Nitrogen is undesirable because it reduces the chemical stability of the fuel and because it may be converted to the pollutants oxides of nitrogen (NOx) during combustion. Ultimately the Nation's large reserves ot coal could be used as a source of jet fuel. The use of coal syncrudes, which have very high concentrations of aromatics, would require even more severe hydrogenation for the production of jet fuel. The cost of producing jet fuel to current specifications will certainly rise as we use less desirable feedstocks and more-energy-intensive refining Rising production costs will be an increasing incenprocesses. tive to broaden jet fuel specifications in order to minimize energy consumption and reduce total fuel costs. The degrees to which fuel properties may change within the next 20 years are shown in table IV-1 (ref. 3). The ranges of properties for a potential broad-specification fuel are compared with those for a typical current Jet A fuel. These are the more significant changes to be expected in terms of their effects on The increase in aromatic aircraft engines and fuel systems. content is approximately equivalent to the reduction in hydrogen Increasing the final boiling point as indicated genercontent. ally leads to the increase in freezing point shown. The thermal which is a measure of the degree to which the fuel stability, may be heated without incurring deposits within the fuel system, might become marginal in terms of current-day standards. The properties of the potential broad-specification fuel approach those of a number 2 diesel fuel. The feasibility of using cryogenic fuels such as liquid hydrogen is being studied (ref. 4); however, tne use of this fuel is a long way off. The general consensus is that liquid-hydrocarbon fuels will be used to fuel jet aircraft for many years to come, but that less stringent specifications may be necessary in the future for the reasons discussed previously. Therefore the al130

ternative jet uid-hydrocarbon the potential

aircraft fuels considered in this paper are liqfuels with the range of properties indicated for broad-specification fuel in table IV-l. ENGINES AND FUEL

EFFECTS OF FUEL PROPERTY CHANGES ON AIRCRAFT SYSTEMS

The effects that the property changes shown in table IV-1 would have on the performance and durability of aircraft engines and fuel systems are discussed in references 5 to 7. Reduced hydrogen content along with lower volatility and higher viscosity would adversely influence the combustion characteristics of jet A higher concentration of reactive constituents would fuel. lower the thermal stability of the fuel. A higher freezing point along with a higher viscosity would adversely influence the low-temperature flow and pumping characteristics. Combustion Characteristics

Decreases in hydrogen content have a pronounced effect on exCombustor tests have been haust smoke and liner temperatures. conducted with prepared fuel blends having varying hydrogen conAt simulated cruise and takeoff conditions typical for a tents. exhaust smoke increases significantly as the hydroJT8D engine, 'l'here is a corgen content of the fuel decreases (fig. IV-3). responding effect of hydrogen content on maximum liner temperaAs the hydrogen content is lowered, the flame ture (fig. IV-4). becomes more sooty and more luminous: tnus the neat radiated to The rate of increase in maximum the liner walls is increased. liner temperature as the hydrogen content is lowered is much The increases in liner wall greater at cruise tnan at takeoff. temperature at cruise would have a larger influence on liner durability because of the much larger exposure time during Higher liner temperatures reduce liner life as a result cruise. Other combustor of accelerated metal fatigue ana oxidation. performance and durability criteria that could be adversely affected by reduced hydrogen content, lower volatility, and higher viscosity include ignition and relight capability, gaseous exhaust emissions, and carbon deposition. Fuel Thermal Stability

The effects of reactive constituents in the fuel on tnermal stability were discussed at a recent workshop sponsored by the Lewis Research Center. The amount of reactive constituents in the fuel may increase as fuel specifications are broadened or as fuel is produced from alternative sources such as snale oil or Figure IV-5 shows tubes that were heated to varcoal (ref. 8). ious temperatures with fuel flowing over the outside of them. 131

They represent heated passages within a fuel injector or fuelThe figure illustrates what can occur with oil heat exchanger. jet fuels that are not thermally stable at the temperatures they encounter on their way through the fuel system to the combusThe deposits shown on the tubes result from the auto-oxitor. Deposit builddation of the reactive constituents in the fuel. up in a fuel injector may cause nonuniform fuel sprays, which can lead to local hotspots on the liner or nonuniformities in the exit temperature profile. The effects of both temperature and nitrogen content of the fuel on the rate of deposit formaAs the temperature intion are illustrated in figure IV-5. the rate of deposition increases, and at similar temcreases, the fuel with the higher nitrogen content has a perature levels, much higher rate of deposition. Similar results, shown in figure IV-6, were obtained in tne study described in reference 9. The effect of temperature on the deposit formation rate for a Jet A fuel containing negligible amounts of nitrogen is compared with that for a Jet A fuel doped with 0.1 percent nitrogen. Both fuels display a rapid increase in deposit formation rate as temperature is increased, 0.1 percent nitrogen shows a much but the fuel containing greater acceleration of deposit formation rate with increasing temperature. Fuel Pumpability at Low Temperature

Fuel stored in aircraft tanks can reacn very low temperatures Figure IV-7 shows a correlation of induring long flights. flight minimum fuel temperatures for Boeing 707 and 747 aircraft flying polar route missions. Data were obtaineci from about 1100 Each mission was greater than 4000 nautical miles. missions. The probability that the temperature of fuel in the tank will fall below a given minimum value is plotted against minimum fuel The minimum temperature that the fuel may reach temperature. of these missions. during flight is lower than -150 F for half For 10 percent of these flights, the temperature may be lower than -300 to -40O F. The differences in temperature between the two types of aircraft are mainly attributed to differences in For a probability of 1 flight Mach number or flight envelope. the minimum fuel temperature is as low as about day per year, -45O to -55O F. Airlines have always avoided potential fuel freezing problems by using fuels with low freezing points and, altering flight conditions. when necessary, Lockheed conducted an experimental study under a NASA contract to determine the flow behavior of fuels cooled below their freezing points in a simulated flight environment. The experimental fuel tank simulator used for this study is shown in figure IV-8. It represents a section of a Lockheed L-1011 wing tank. Cooling was provided at the top and bottom of an insula132

ted tank to simulate the in-flight cooling of the upper and Fuel sloshing could be lower surfaces of a section of the wing. simulated by rocking the tank with a mechanical agitation sysfuel was discharged from the tank by After being cooled, tem. Conditions in the tank were observed through the boost pump. Figure IV-9 shows an viewing ports on each side of the tank. interior view of the tank loaded with a test fuel at a temperaA surge box was loture above the freezing point of the fuel. cated at the inlet to the boost pump to maintain a constant pressure head. Ejector tubing connected to the pump inlet provided a means for removing fuel from the bottom of the tank. Temperature was measured at various locations by thermocouples. Stringers were located at the top and bottom to simulate the The effects of cooling a structural design of the wing tank. fuel below its freezing point are illustrated in figure IV-lo, which shows the frozen and liquid fuel remaining in the tank During the initial part of the pumpdown, the after pumpdown. two-phase mixture of wax and liquid fuel may be removed by the the pump inlet becomes plugged by boost pump, but eventually and the remaining fuel cannot be pumped this slushy mixture, The low-temperature flow behavior of fuels in from the tank. the fuel-tank simulator was assessed by measuring the percentage of a full load of fuel that could not be pumped from the tank as The percentage of fuel held in the temperature was lowered. tank was a function of boundary-layer temperature, as shown in figure IV-11. The boundary-layer temperature was measured about Data are shown for l/2 inch from the lower surface of tne tank. While the boundary-layer temperature for a specifthree fuels. for both lower-freezingic percentage of fuel holdup is lower the rate of increase in fuel holdup as the boundpoint fuels, The holaup for ary-layer temperature is reduced is different. the Jet A fuel refined from a paraffinic petroleum increases more rapidly as the boundary-layer temperature is reduced than that for either tne naphthenic Jet A or the broad-specification This may be attributed to the larger concentration of fuel. long-chain paraffin waxes contained in high-molecular-weight, This kind of research data enables us to have a betthis fuel. ter understanding of tne low-temperature behavior of fuels in Nevertheless, for the present, any aircraft fuel systems. amount of solid fuel formed in tne fuel tank is unacceptable to an airline and is avoided by routinely maintaining the fuel tank temperature about 5 o F above the freezing point of the fuel being used. COMBUSTOR AND FUEL SYSTEM TECHNOLOGY NEEDS The potential problem areas described previously are identified The in the sketch of a typical combustor shown in figure IV-12. higher liner temperatures associated with the use of fuels with higher aromatic content could be reduced or tolerated by evolvimproved cooling effectiveness, ading improved construction, 133

or ceramic coatings. Increasing the proportion vanced alloys, of air used to cool the liner would be unacceptable, however, because this approach would reduce the amount of air available for control of the exit temperature profile. Increasing coolant airflow might also increase carbon monoxide and unburnea-hydrocarbon emissions. Approaches which are successful in reducing exhaust emissions, such as lean burning or improved fuel atomization (paper II), can also be effective in reducing liner temperature by minimizing luminous flame radiation. These approaches could also prevent carbon deposition and reduce exhaust smoke. Multizone combustion or variable geometry, besides enabling the control of emissions at both high and low power levels, could also ensure reliable ignition and relight. Several multizone designs, including the Pratt & Whitney vorbix and the Generai Electric double annular combustors, are described in paper II. A cutaway drawing of the double annular combustor is shown in figure IV13. During idling and at low power, fuel is fed only to the outer annulus, and combustion occurs in a relatively fuel-rich At high power levels, fuel is fed to both the inner and zone. and combustion occurs at relatively fuel-lean conouter annuli, ditions. Both the double annular and vorbix combustors have been evaluated with several fuel blends varying in hydrogen content (refs. 10 and 11). The liner temperatures measured in tests of the double annular and vorbix multizone combustors are compared with data from several production combustors in figure IV-14. Since the various data were not all obtained at the same combustor-inlet conditions, the data are plotted as the difference between maximum liner temperature and combustor-inlet temThe curves for the production combustors exhibit a perature. strong dependence of maximum liner temperature on hydrogen con'The data for the multizone combustors show a tent of the fuel. relative insensitivity of maximum liner temperature to hydroMost of these data were obtained in combustor rig gen content. tests at pressures of 10 atmospheres or less; however, the two data points shown were obtained in the evaluation of the double annular combustor in the CF6-50 engine at takeoff conditions. While these findings represent limited laboratory tests and their practicality requires demonstration in a full development they do indicate that it may be possible to make imporprogram, tant advances in the ability of aircraft gas turbine engines to use low-hydrogen-content fuels. Additional combustor design concepts for broad-specification fuels being evaluated under a NASA contract with General Electric are shown in figure IV-15. Several variations of the double annular combustor are being investigated. Both of the modified configurations shown employ improvements in fuel nozzle including air atomization. design, A shorter main combustion stage is incorporated in one configuration to reduce liner surface exposure and to minimize smoke and NO, emissions. Pre134

mixing of fuel and air in the main stage will second configuration to reduce luminous flame haust emissions. Prevention of Fuel System Fouling

be studied radiation

in and

the ex-

Preventing fuel gumming and coking in the engine fuel system becomes more and more difficult as compressor discharge temperatures increase as a result of increasing engine pressure ratios or as fuel systems become more complex as a result of multistage combustion. Many approaches to preventing fuel system fouling are already in use. For example, designers attempt to limit tne maximum fuel temperature by insulating the parts of the fuel manifold exposed to high temperatures. 'The controi of reactive constituents in the fuel is attempted by the addition of antioxidants during the final stages of refining. However, furtner design improvements must be sought to limit fuel exposure temperature. Research is needed to obtain a oetter understanding of the effects of constituents in the fuel on fuel degradation. The dissolved oxygen in the fuel plays an important part in While the removal of this oxygen may be imthese reactions. practical, more efficient antioxidant additives could be developed and the removal of certain reactive or catalytic impurities might be effective. Fuel system purging could be investigated as a means of minimizing fuel deposits in multizone combustors. Fuel System Technology for Fuels With Higher Freezing Points

Modifications to aircraft fuel systems that mignt be necessary to permit the use of a higher-freezing-point fuel are being investigated (refs. 12 and 13). A number of fuel tank heating sources plus insulation have been evaluated as approacnes to keeping the fuel in the tanks above tne freezing point (fig. IV-16). Insulation of the fuel tanks is currently not practical because it entails a large weight penalty; however, future aircraft wing designs may permit the use of effective lightweight insulation. The use of a tailpipe heat exchanger is undesirable Fuel heating by means of compresbecause of design complexity. sor air bleed has the disadvantage of having a relatively high cycle penalty. The use of either boost pump recirculation or a heat exchanger in the cabin air conditioning system is limited Two systems by the amount of heat tnat either could provide. the lubrication oil are considered worthy of further study: heat exchanger and an electric heat excnanger powered by an engine-driven generator. The existing lubrication oil heat exchanger system would be modified by adding a second heat exchanger that would return heated The recirculation valve could be fuel to the tank (fig. IV-17). This shut off when fuel tank heating would not be required. 135

system has the advantages of using existing components, being and being a simple syshaving a low cycle penalty, low in cost, tem requiring only a relatively minor modification to the existThe calculated f uei ing lubrication oil heat rejection system. temperatures for a long-range flight using this system are compared with the temperatures for an unneated fuel tank in figure Heating the fuel permits the minimum in-flight fuel temIV-18. perature to be raised from about -4tj to -2S F. The latter The heat revalue points out the disadvantage of this system. jection limit of about 2800 Btu per minute would not be applicable to fuels with freezing points greater than -300 F, allowIn addition, margin. the heat avaiiing for a 5 o F operating able would be variable and woula depend on the engine operating point. The electrical fuel heating concept would overcome these disadvantages by providing much higher heating rates to permit the use of fuels with freezing points up to 00 F (fig. IV-19). Heater control would be independent of engine operating point, and this system would be adaptable to layover heating oy means In-flight power would be obtained by of auxiliary ground power. installing a generator on each engine. An inert fluid would be The heated fluid would heated by an electric resistance heater. in order to reject heat then be pumped to another heat exchanger A disadvantage of this system is that it into the fuel tank. volves major modifications and additions to the engines and airFurthermore tnere would be a significant weight penalty craft. and a cycle penalty when the system is used. The judgment to use such a fuel heating system will depend partly on the relative cost reduction of using a broad-specification fuel compared with its installation and operating costs. CONCLUDING REMARKS Projected shifts in the supply and quality of refinery feedstocks warrant the consideration of broadening jet aircraft fuel specifications. Advanced combustor and fuel system technologies that will be required to counteract adverse fuel property effects if fuel specifications are broadened are currently being evaluated. A considerable effort will be necessary by both government and industry to assemble the data base that must be developed in order to establish an acceptable tradeoff between future fuel specifications and aircraft engine tecnnology. REFERENCES 1. Dukek, W. G.; and Longwell, Fuels for Aviation. Exxon 92-96. PP. 136 J. P.: Alternative Air World, vol. 29, Hydrocarbon no. 4, 1977,

2.

Longwell, Fuels. 161. Longwell, nology.

J. P.; J. Eng.

and Grobman, Power, vol.

J.: 101,

Alternative no. 1, Jan.

Aircraft 1979, pp. Fuels Fuels 1978. Fuels 1978.

155Tech-

3.

John P., ed.: NASA CP-2033,

Jet Aircraft 1978.

Hydrocarbon

4. Witcofski, Aviation CP-2036, 5.

Robert D.: Progress on Coal-Derived Systems. CTOL Transport Technology, Pt. 2, 1978, pp. 927-950.

for NASA on FuNASA Future Air

Grobman, Jack S.: Impact of Broad-Specification ture Jet Aircraft. CTOL Transport Technology, CP-2036, Pt. 1, 1978, pp. 217-233.

6.

Blazowski, W. S.; and Jackson, T. A.: Evaluation of Jet Fuel: Combustion Characteristics. AFAPL-TR-77-93, Force Aero Propulsion Lab., 1978. (AD-AO60218.)

7. Rudey, R. A.: Effect of Broadened-Specification Fuels on NASA TM-79086, 197Y. Aircraft Engines and Fuel Systems. 8. Reynolds, Thaine W.: Thermal bine Fuels Derived from Oil x-3551, 1977. Taylor, posits. 1969. Stability of Some Aircraft Shale and Coal. NASA TM Tur-

9.

The Study of Hydrocarbon Fuel Vapor DeWilliam F.: GR-lo-VDP-69, Esso Research and Engineering Co., (AFAPL-TR-69-77, AD-863604.)

10.

Gleason, C. C.; and Bahr, D. W.: Experimental Clean CombusAlternate Fuels Addendum, Phase II. tor Program, (R76AEG268, General Electric Co., NASA Contract NAS318551.) NASA CR-134972, 1976. Roberts, R.; Peduzzi, A.; and Vitti, G. E.: Experimental (PWA-5370, Pratt & Clean Combustor Program, Phase II. Whitney Aircraft; NASA Contract NAS3-18544.) NASA CR-134970, 1976. Preliminary Analysis of Pasion, A. J.; and Thomas, I.: Aircraft Fuel Systems for Use with Broadened Specification (~6-44538, Boeing Commercial Airplane Co.; Jet Fuels. NASA CR-135198, 1977. Contract NAS3-19783.) Friedman, Aviation Robert: Turbine High Freezing Point Fuels Used NASA TM-79015, 1979. Engines. for

11.

12.

NASA

13.

137

POTENTIAL

CHANGES

IN FUEL

PROPERTIES

PROPERM

CURRENT JET A

POTENTIAL BROADSPEC FUEL

AROMATICS, voi % HYDROGEN, wt%

17-

25

30-

35

14 - 13.5

13.0 - 12.5 550-3o630 -20

FINAL BOILING POINT, OF 500 - 530 FREEZING POINT, OF THERMAL STABILITY -5o-40

ACCEPTABLE

MARGINAL
cs-79-1558

Table IV-1

JETFUELPRODUCTION ALTERNATIVES

SPECIFICATlnNq Figure IV-1

138

TRENDS IN AROMATIC CONTENT OFCOMMERCIAL JETA FUEL

JET A FROM ALASKAN CRUDE (PROJECTED)-\, ASTM JET A WAIVER LIMIT --------A-JET A FROM A HEAVY ARABIAN CRUDE7, cl ASTM JET A SPEC LIMIT -----A---------vol % (AVGlyrl * A--

cs-78-151

YEAR Figure IV-2

EFFECT OF HYDROGEN CONTENT OF FUELON SAESMOKENUMBER

S. A. E. SMOKE NO.

30 20

0111 10
cs-79-1568

11 12 13 14 15 HYDROGEN IN FUEL, % BY WT

16

Figure IV-3

139

EFFECT OF HYDROGEN CONTENT OF FUEL ON MAXIMUM LINER TEMPERATURES

1900

MAX LINER

TTY

1500 1400 ;:;;t , 10 , , , +RUl,SE

cs-77-569

11 12 13 14 i5 16 HYDROGEN IN FUEL, % BY WT Figure IV-4

THERMALSTABILITY OF FUELSCONTAINING NITROGEN

N -a175L N-aom

II 1 84 H III
525
CS-79-1225

,.i

500 TEMP,9

500

Figure IV-5

140

EFFECT OF FUELNITROGEN CONTENT ON THERMAL STABILITY

107

DEPOSIT FORMATlON RATE, pg/cm2-hr

I
300 350 TEMP, 9 400
CS-79-1567

I
450

Figure IV-6

MINIMUM FUELTEMPERATURES SUMMARY OF IN-FLIGHT

99.99 99. 9 99 95 PROBABILITY OF FUEL TEMPS BELOW MIN. SHOWN, % 9o POLAR ROUTE

/
50 10 5 --1 .Oll -80 -60 -40 -20 0 MIN. FUEL TEMP, OF 20
cs-79-1572

DAYlYR

Figure IV-7

141

LOWTEMPERATURE FUELTANK SIMULATOR

Figure IV-8

Figure IV-9

142

INTERIORVIEW OF FUEL TANK SIMULATOR

FUELCOOLEDDOWN BELOWFREEZING PT

Figure IV-10

LOWTEMPERATURE FUELTANKBEHAVIOR

0 0

JET A - PARAFFINIC - FREEZING PT -40 F JET A - NAPHTHENIC FREEZING PT -45 F

FUEL HOLDUP AFTER PUMPOUT, %

-60
cs-79-1571

-50

-40 -30 -20 BOUNDARY LAYER TEMP, % Figure IV-11

-10

143

COMBUSTOR TECHNOLOGY REQUIRED TO USE BROAD-SPEC FUELS

/-PREVENT FUEL GUMMING & COKING I ! ,-REDUCE LINER TEMPS

REDUCE SMOKE & GASEOUS EMISSIONS

z PREVENT CARBON DEPOSlTlON Figure IV-12

cs-79-1573

BLEANNULAR COMBUSTOR

Figure IV-13

144

EFFECT OF HYDROGEN CONTENT OF FUELON LINER TEMPERATURE FORDIFFERENT COMBUSTORS

lOOO/- CFd-50 FULL ANNULAR COMBUSTOR mlMAX LINER TEMP MINUS INLET AIR TEMP, AOF ,- JT8D COMBUSTOR CAN 600 -VoRBIX &

400 -

DATA POINTS, CF6-50 ENGINE DATA WITH DOUBLE ANNULAR COMBUSTOR I I I 12 13 14 15 HYDROGEN CONTENT, w-t % Figure IV-14

cs-79-1574

200. 11

16

COMBUSTOR DESIGN CONCEPTS FORBROAD-SPEC FUELS

DOUBLE/ANNULAR WITH AXIAL STAGED BURNING ZONES

DOUBLE/ANNULAR BASELINE
cs-79-1575

DOUBLE/ANNULAR WITH PREMIXED MAIN STAGE

Figure IV-15

145

FUEL TANK HEATING SOURCES

,/-

CABIN AIR COND HEAT EX rBOOST PUMP RECIRCULATION

- TAIL-PIPE HEAT EX ENGINE-DRIVE COMPRESSORAIR BLEED -

Figure IV-16

FUELHEATING SYSTEM STUDY - BOEING

FUEL/OIL HEAT EXCHANGER SYSTEM

FUEL

rckl I
i ,

+-HEAT EXCHANGERS .a I

ADVANTAGES l USES EXISTING HEAT REJECTIONHARDWARE l LOW COST - LOW PENALTY . SIMPLE SYSTEM, RELATIVELY MINOR MODIFICATION DISADVANTAGES l LIMITATION TO QUANTITY OF HEAT AVAILABLE l-30 F FREEZE PT MAX. 1 .VARfABLE HEAT AVAILABLE, DEPENDENTON ENGINE OPERATING POINT

cs-78-2880

Figure IV-17

146

PERFORMANCE OF LUBE-OIL HEATER FOR IN-FLIGHT FUELHEATING

2800 Btulmin 10 oAVG RATE

-40 -50 0

I 1

I 2

I 3

I I 5 4 TIME, hr -

I 6

I 7
cs-79-1565

I
8

Figure IV-18

ELECTRICAL FUELHEATING CONCEPT - BOEING

HEATED FUEL DISTRIBUTOR -/
rUtL/r

EXCHA

ADVANTAGES l HEAT AVAIL FOR FUEL WITH 0 F FREEZE PT . INDEPENDENT CONTROL l ADAPTABLE TO GROUND HEATING FOR LAYOVER DISADVANTAGES l MAJOR MODIFlCATIONADDITIONAL SYSTEM l WEIGHT PENALTY l CYCLE PENALTY WHEN USED \ L HEA;%$PORT FLUID HEATER

L GENERATOR (800 KVA AVAIL)

CS-79-1569

Figure IV-19

147

V. Robert A.

MATERIALS Signorelli,

AND

STRUCTURES

TECHNOLOGY Gary R. Halford,

Thomas K. Glasgow, and Stanley R. Levine

National

Aeronautics and Space Administration Lewis Research Center

There are several major thrusts of our materials and structures First, we research for aircraft turbine propulsion systems. must understand the materials and structures performance limitations, particularly for the hot section of the engine in which To do so, it is these limitations limit the life of components. necessary to understand the failure modes for components such as and combustors and how they are affected by the blades, vanes, environment for such components. we are attempting to improve the materials used for such Second, Turbine components with components using several approaches. through high strength at higher temperatures are being achieved Improved coatings the application of directional structures. are being applied to hot section parts to increase oxidation and corrosion resistance. The higher specific properties of composite materials are being applied to increase strength and stiffness with reduced weight. more cost effective proAlso, cessing such as near net shape powder methods are being applied to disks. Third, we are attempting to employ these advanced materials more effectively by developing more accurate life prediction techThese techniques will enable the designer to predict niques. component life more accurately in advance of service and also to incorporate advanced structural concepts in design. In addition to these structural concepts, structures work is an area of growing importance at Lewis. We are increasing our effort and resources in this area, and in the future we anticipate that there will be much to report in the field of engine structural dynamics. In this paper, hot section components are discussed first since they operate under a severe combination of stress, temperature, and environmental attack. After describing the various aspects of our program to improve materials and structures for the hot section, the work directed toward improving the intermediate temperature and cold section components of turbine engines will be discussed.
149

HOT SECTION COMPONENTS We consider first the major structural materials requirements Because of the comfor the engine hot section components. plexity of the loadings imposed and the environmental conditions to which these components are exposed, it is vital to understand the nature of these requirements in order for the engine designer to proceed effectively. Material Requirements for hot and

There are three major structural materials requirements thermal fatigue resistance, strength, section components: resistance to environmental attack.

- The traditional strength requirements are, of Strength. At high use temshort time yield and tensile strengtn. course, long time creep and rupture strength become an adperatures, Strengtn requirements genditional important consideration. Since the hiserally dictate the sizing of a component part. toric trend is to increase use temperatures, it is mandatory that alloys be identified that have higher strength to keep pace Examples of such alloys with these higher use temperatures. will be discussed in a later section. - The second and third requirements are not as Thermal fatigue. This is primarily because thermal fafamiliar as the first. tigue and environmental attack are more complex phenomena and are not as readily understood. Thermal fatigue cracking, is a common failure mode in hot section components nevertheless, of gas turbine engines. An example involving a film-cooled inThermal fatigue cracks let guide vane is shown in figure V-l. have initiated and propagated as a direct result of the cyclic thermal strains caused by the repeated heating and cooling on each engine startup and shutdown cycle. Thermal fatigue in engines usually involves a combination of fatigue and creep which can interact to seriously reduce cyclic lifetime. Figure V-2 dramatizes the cyclic life degradation brought about by the interaction of creep with fatigue. At lower temperatures and short times per cycle, the creep effect is not present and the cracking mode is transgranular and representative of classical fatigue cracking. However, as the temperature and the time per cycle increase, the creep influence increases and the cracking shifts to the more detrimental intergranular mode following along transverse grain boundaries (ref. 1). As may be seen from the figure, cyclic life is reduced substantially. By recognizing the mode of thermal fatigue cracking, we are in a much better position to cope with this crucial problem by providing materials with no transverse grain boundaries and by 150

developing life prediction methods for the creep-fatigue interaction.

which

can

accurately

account

- In addition to being mechanically deEnvironmental attack. the gas turbine environment is chemically aggressive manding, High temperatures, pressures, gas velocities, toward materials. and severe thermal cycles can cause rapid oxidation attack. Impurities in the gas stream can further accelerate the process by These reactions can also inducing hot corrosion and erosion. accelerate the creep-fatigue cracking process. Since the oxithe approach to controlling envidation process is unavoidable, ronmental attack is to select materials which react slowly to form an adherent, protective corrosion product. Three basic modes of oxidation attack are possible (ref. 2) as For parabolic oxide growth, an adherent oxshown in figure 3. In this case, a plot of oxide thickness ide is required. Hence, the term parabolic oxide squared against time is linear. blade alloys, grows growth. NiO, which forms on most turbine parabolically at unacceptably high rates. which is Cr203, can be useful as a protective scale at lower hot secvolatile, tion temperatures but not at the highest temperatures. Here Hownonvolatile oxides such as Al203 are most desirable. because of a thermal expansion mismatch with turbine mateever, Al203 is subject to spalling. rials, the hot corrosion and erosion failure modes (ref. In figure V-4, These potentially more aggressive modes of 2) are illustrated. environmental attack are induced by impurities in the air and The hot corrosion process is caused by the reaction of fuel. such as sea salt with sulfur, in the fuel. airborne impurities, At certain temperature, pressure, and concentration levels, sodium sulfate and other salts can condense as a liquid phase, react with and flux protective oxide scales, and cause rapid sulfidation attack of turbine alloys. Because the hot corrosion it can be far more damaging than process is induced by a liquid, oxidation even though the temperature is typically several hundred degrees lower. A second potentially catastrophic problem can be caused by imSuch deposits can purities which form heavy solid deposits. plug film cooling holes and lead to component overtemperature. the impact of Finally, bine components causes carbon metal or ingested loss via the dirt particles erosion process. on tur-

The seriousness of the hot corrosion problem for the first-stage vanes of a helicopter gas turbine is illustrated in figure V-5. hot corrosion attack along the leading edges has conHere, In addition, tributed to the formation of longitudinal cracks. heavy impurity deposits have formed near pressure surface film cooling holes. 151

Improved

Hot

Section

Component

Materials

turbine blades are subject to the Of the hot section components, most severe combination of temperature, stress, and oxidation a substantial portion of Consequently, and corrosion attack. the materials effort at NASA Lewis is devoted to turbine blade Figure V-6 illustrates the use temperature trend applications. Some specific data points are for turbine blade materials. shown for alloys subjected to typical blade operating conditions Convenof 30 000 psi stress and a rupture life of 5000 hours. tional cast alloys with random polycrystalline microstructure developed in the 1950's and 1960's had use temperatures ranging from 14000 to 16000 F under these conditions. - To go to higher temperatures for the Directional structures. materials having distinct directional same stress and life, These include single crystal structures are being developed. oxide dispersion strengthened superalloys (ODS), superalloys, Beyond the metals, ceand fiber reinforced superalloys (FRS). ramic materials offer considerable potential, though at this time we do not have the design skills required for their use as their use for stationary or nonload However, turbine blades. bearing components such as seals and coatings will occur much sooner than shown in figure V-6. Two of the Materials AiResearch ventional hidden in tection. are alined the single blade. cast directional structures under development in our for Advanced Turbine Engines (MATE) program at Garrett The first blade is a conare shown in figure V-7. it has randomly oriented grain boundaries casting; this view by a coating applied for oxidation proBy directional solidification, the grain boundaries in And, of course, in the major stress direction. crystal there are no grain boundaries to weaken the

to maintain a turbine inlet temperature of In past practice, 19200 F the conventional casting required cooling air and thus If directional sothe complexity of casting a hollow blade. solid blade capable of withlidification is used, a simpler, standing the same turbine inlet temperature can be employed with If a single crystal blade is no cooling as shown in the figure. it is anticipated that the turbine inlet temperature can used, be increased to 19650 F without requiring cooling (ref. 3). Oxide dispersion strengthened peratures a new development, offers considerable superalloy, superalloys. - For higher use teman oxide dispersion strengthened In oxide dispersion potential.

152

strengthened strengthening stable oxide effective in

superalloy alloys such as MA-6OOOE, the traditional mechanisms are augmented by a dispersion of fine, The particles are especially particles (ref. 4). improving high temperature properties.

The process steps by which an oxide dispersion strengthened Metal powders, elesuperalloy is made are shown in figure V-8. mental and prealloyed, are mixed together with oxides in a high energy stirred mill which kneads the fine oxides thoroughly into the metal. The very homogeneous powder that results is sealed hot rolling Optionally, in cans and consolidated by extrusion. may follow. Finally, the hot worked product is given its elongated microstructure by recrystallization in a thermal gradient. The macro and microstructures of the oxide dispersion strengthThe highly elonened alloy MA-6000E are shown in figure V-9. gated grain structure shown in the macrograph is oriented in the direction of highest applied stress. It can be noted that there are essentially no transverse grain boundaries available at The microstructure which thermal fatigue cracks can initiate. shows both cubic gamma prime precipitates, the traditional and the very fine oxide parstrengthening phase in superalloys, ticles that are within both the gamma prime and gamma phases. This oxide dispersion strengthened alloy MA-6000E has generated a great deal of interest among engine manufacturers, particularly for the smaller engines that are relatively difficult to cool. The stress rupture properties of this alloy account for much of this interest (fig. V-10). The grain structure of each alloy is illustrated schematically in the figure. Alloy B-1900 is typiDireccal of current conventionally cast gas turbine blades. tionally solidified cast alloys, polycrystalline and single crystal, provide some advantage. And the addition of an oxide dispersoid adds a considerable increment so that one can envision using the oxide dispersion strengthened alloys at temperatures of 2000 F or higher. It should be mentioned that extending the use temperature of superalloys to higher values aggravates the oxidation attack. Alloy MA-6000E will require a coating for extended life at high As will be discussed later, the process required temperature. to identify the best alloy/coating combination has already begun. In addition to the improved stress rupture capability, the oxide dispersion strengthened alloy MA-6000E has demonstrated very good thermal fatigue resistance (fig. V-11). Thermal fatigue testing is conducted by alternately immersing wedge-shaped specimens in hot and cool fluidized beds. Thermally induced strains crack typical current blade materials after relatively few cycles. For example, B-1900 and directionally solidified Mar 200

153

MA-6000E cracked after only 800 and 2300 cycles, respectively. is still being tested and has yet to show any cracks after 6500 It must already be considered among the very cycles (ref. 5). best of thermal fatigue resistance materials. Another hot section component, the combustor, may use a differWork on the advanced combustor is being perent ODS alloy. formed by Pratt & Whitney in our Materials for Advanced Turbine As shown in figure V-12, the outside Engines (MATE) program. cooler portion of the combustor is to be formed of the convenThe interior, hotter portion will be tional alloy Hastelloy X. lined with an ODS sheet alloy based on the FeCrAlY or the Another use for oxide dispersion NiCrAlY system (ref. 6). strengthened alloys is as vanes: in fact, one alloy is already specified as bill-of-material in an advanced military engine. - The next higher level of use Fiber reinforced superalloys. temperature (fig. V-6) capability, above that of the oxide disis occupied by the fiber reinforced persion strengthened alloys, These are composite materials in which we superalloys (FRS). can take advantage of the high strength at high temperature of reinforcing fibers, such as tungsten, and the ductility of superalloys as a matrix to achieve an outstanding combination of The lOOO-hour density-compensated rupture strength properties. of several fiber reinforced superalloy or FRS composites is shown in figure V-13 along with a typical superalloy for comparison. The relative strength advantage of each composite is The controlled by the wire used to reinforce the superalloy. composite using unalloyed tungsten lamp filament, such as 218 achieves a modest improvement, while the composite alloy wire, developed as part of the Lewis Research with WReHfC alloy wire, has more than a fourfold advantage in Center FRS program, - density at 20000 F over conventional superalloys strength 7 and 8). (refs. In addition to high temperature strength, FRS has the added advantages of high thermal conductivity and low thermal expansion The high thermal conductivity of tungas shown in figure V-14. sten provides a typical FRS composite with a thermal conductivThis ity about twice that of conventional superalloys (ref. 9). advantage can be used to increase the effectiveness of cooled turbine components by increasing life at a given cooling airflow or by reducing cooling airflow to increase engine efficiency. The lower thermal expansion of FRS compared with nickel and cobalt superalloys also aids in reducing the severity of the creep-fatigue interaction discussed previously. A question that has been addressed is how such a complex comFigure V-15 shows schematiposite blade would be fabricated. This cally the process employed for turbine blade fabrication. process is based on the technology evolved for boron-aluminum 154

A filament mat and matrix alloy foil are combined fan blades. which is cut into plies. These to form composite monotape, plies are stacked in a die and then diffusion bonded at temperature to form a blade. Figure V-16 shows how a cooled turbine blade has been made using A solid a modification of the basic process just described. After steel core was used along with monotapes and root blocks. the steel core was removed by acid leaching diffusion bonding, An impingement cooling insert to form a hollow internal cavity. can be used to provide more complex cooling passages to improve The photograph in figure V-17 shows a cooling effectiveness. JT-9D blade with an unmachined oversized base fabricated using about 4 inches long, was designed to The blade, this process. the same external contour as the bill-of-material blade and to within 10 percent of the weight of the superalloy blade (ref. The wall thickness was modified to take advantage of com10). This illustrates the feasibility of fab-. posite properties. ricating cooled hot section parts such as blades and vanes using the FRS process concept. Other hot section parts such as combustor liners are attractive candidates and are being explored for possible application of as the operating temperature goes up, the Unfortunately, FRS. severity of environmental effects such as oxidation and hot corrosion also increases. Improved Environmental Protection Systems re-

The development sistance to the has made surface Metallic selection lytical

of stronger oxidation protection

often at the expense of alloys, and hot corrosion attack processes, problems extremely challenging. coating of ana-

- Over the past 15 years, metallic coatings. criteria have been devised by a combination and experimental approaches.

Figure V-18 shows a simplified oxidation resistance map on the left for three key elements affecting the oxidation resistance Most of superalloys nickel, chromium, and aluminum (ref. 11). cast superalloys fall in the poor area and the lower part of the The two composition areas exhibiting good oxidation fair area. Simple resistance define two major coating classifications. which have compositions near the apex of the aluminide coatings, are typically formed by reaction between the alloy and diagram, Overlay or add-on coatings, a source of aluminum (ref. 12). which can be deposited by such methods as electron-beam evapotypically fall near the 13) or cladding (ref. 14), ration (ref. center of the diagram. A second factor to consider in coating selection is hot corro155

As discussed previously, hot corrosion is sion resistance. caused by impurities in the air reacting with sulfur' in the fuel The hot corrosion map (ref. to form a condensed fluxing liquid. 15) for the same three elements near the apex of the diagram can be undesirable for hot corrosion protection while coating compositions falling across the center of the diagram are good. The most favorable coating for a given application can be selected by properly balancing coating composition with additional fac16) and processing cost. The need tors such as ductility (ref. for high ductility at low temperature generally favors compositions near the center of the maps while cost considerations favor aluminide coatings having compositions near the apex. The same principles can also be applied in the selection of cobaltbase coating compositions (ref. 17). The significant degree to which metallic coatings can extend oxidation life compared to bare superalloys is illustrated by the life in a Mach 1 burner burner rig data of figure V-19. Here, rig is plotted as a function of temperature. Simple aluminide coatings which fall toward the lower end of the coated specimen range offer more than a fourfold extension of life (ref. 18) whereas overlay coatings offer as much as a fiftyfold extension The effectiveness of overlay coatings as protection in life. against hot corrosion is illustrated in figure V-20. Here specimen weight loss (ref. 19) in a Mach 0.3, Jet A fueled burner rig is shown. Hot corrosion is induced by injecting sea salt at a 5 ppm concentration based on air flow. In this accelerated most uncoated cast nickel-base alloys suffer immediate and test, A NiCrAlY overlay affords protecrapid hot corrosion attack. tion for 1100 hours. Coatings are also highly effective in suppressing the initiation of creep-fatigue induced cracks as illustrated in figure V-21 by the log scale plot of cycles to crack initiation. The data were obtained in a fluidized bed (ref. 20). The random polycrystalline cast MM-200 alloy exhibits low life. The addition of an aluminide coating more than doubled life. Directional solidification to obtain a more crack resistant strulcture dramatically improves thermal fatigue resistance. The addition of an overlay coating to this stronger alloy more than doubled life. Our current efforts in metallic coatings are aimed at developing better coatings for airfoil cooling passages, improving plasma spray deposited coatings so that they are equivalent in performance to the more costly electron beam evaporated coatings, and developing coatings for advanced alloys such as MA-6000E. Thermal barrier coatings. - Another concept undergoing velopment is the thermal barrier coating (refs. 21 to conventional metallic coated blade illustrated on the 156 rapid de24). The left in

figure V-22 can operate in gas turbines under conditions where the gas temperature is above the high temperature strength capability of the superalloy and even above its melting point beIf a 15-mil insulating oxide ceramic cause of air cooling. layer or thermal barrier is added (as shown on the right in fig. V-22), the difference between the gas and metal temperatures can be increased by 200 o to 600 F without additional cooling. The vital glue that locks the oxide onto the superalloy is an oxidation resistant metallic bond coating as can be seen from Both the bond coat and the oxide the optical photom,icrograph. layers shown in figure V-22 were deposited by a plasma spray process. The concepts discussed earlier for metallic coating selection are also being applied to bond coat selection. Besides proper selection of the bond coat composition, the selection of the oxide composition is key to the durability of the NASA thermal barrier coating (refs. 21 and 23). Because of the large temperature drop through the oxide layer, thermal barrier coatings are not just another new family of protective coatings. The thermal barrier coating is a revolutionary concept that can be applied to turbine airfoils in many ways. An example of the calculated benefits of a lo-mil oxide coating applied to the cooled components of a high bypass engine is shown in table V-l. Cooling air reductions to the first and second stages are shown in the center column as a percent of engine airflow. The 6.1 percent total cooling air saving yields a net thrust specific fuel consumption improvement of 1.3 percent. Alternatively, blade life can be improved by more than four times at the baseline coolant flow because of lower metal temperatures and less severe transient thermal stresses. Looking ahead, thermal barrier coatings may also provide gains in efficiency by permitting higher gas temperatures without increasing the coolant flow. Thus, thermal barrier coatings can greatly extend the capability of conventional superalloys as well as of advanced material concepts such as single crystals, ODS superalloys, and FRS alloys. The durability of the thermal barrier coating is illustrated by the cyclic Mach 1 burner rig data shown in figure V-23. The Zr02 12 weight percent Y2O3 NASA coating lasts about l-hour cycles at a surface temperature of 26400 F. This is the coating that survived on turbine blades for 500 cycles between full power and flameout in a J-75 engine (ref. 25). Reducing the yttria content to 8 weight percent to obtain a tougher, stronger oxide resulted in a dramatic life improvement as shown by the two lower bars in figure V-23. This improved coating lasted 2000 hours without failure at a 400 higher surface temperature. Finally, at a surface temperature of 28600 F, which is well above levels anticipated in any current engine, the coating survived over 700 l-hour cycles. 157

Recent JT-9D engine tests conducted on a cooperative basis with 26) indicate that the original Pratt & Whitney Aircraft (ref. 1976 NASA coating is not yet engine ready as an add-on for Currently, our efforts are directed first-stage turbine blades. toward improving coating durability by further composition and and toward developing a design methodology process refinement, The which treats the coating and airfoil as an integral system. aerodynamic and heat-transfer aspects of thermal barrier coatings are covered in the TURBOMACHINERY TECHNOLOGY paper of this conference proceedings. Abradable seal materials. - Besides applying our understanding of the environmental resistance of NiCrAl alloys to coatings, we have used it to develop an improved first-stage turbine abradable shroud material. This powder metallurgy material, known as Genaseal, was developed under NASA contract by the General Electric Company (ref. 27). An engine-tested shroud segment filled with Genaseal is compared to two types of shroud segments filled with Bradelloy in figure V-24. The Genaseal segment is in better condition than either Genaseal offers improved abradof the Bradelloy segments. ability and oxidation resistance at temperatures about 1500 F Consequently, turbine efficiency can be imabove Bradelloy. Also, turproved by reducing the cooling air to the shrouds. bine efficiency can be retained by preferential wear of the Genaseal is shrouds rather than wear of the blades during rubs. now the bill-of-material shroud in one aircraft gas turbine enThe degine and is under consideration for a number of others. velopment of an even higher temperature capability oxide ceramic turbine shroud is covered in the MECHANICAL COMPONENTS paper of this conference proceedings. Development of Life Prediction Methods

Of prime importance to the engine designer is the availability of techniques for predicting the life of engine hot section comThe creep-fatigue behavior is one major aspect that ponents. must be taken into account in order to achieve accurate component life prediction in advance of service. - A promising method has been under Strainrange Partitioning. development at the NASA Lewis Research Center during the past The several years called Strainrange Partitioning (ref. 28). it is a method for charmethod serves two primary functions: acterizing the creep-fatigue behavior of materials, and then, it can be used to predict the thermal faonce characterized, tigue lives of components made of those materials. The creep-fatigue cyclic inelastic 158 behavior deformation of any material which can be characterized undergoes by means of

the Strainrange Partitioning method. Figure V-25 represents the relationship between the cyclic strainrange imposed on a laboratory test specimen and the number of times the cycle can be repeated before failure occurs by cracking. Four curves are shown; each represents whay may happen when creep is imposed in When no creep is present, the uppermost curve, different ways. which represents the pure fatigue condition, results. Creep imposed in only the compressive half of the cycle reduces the cyclic life. Creep in both the tensile and compressive halves of the cycle gives rise to further life reductions. .Finally, when only tensile creep occurs, the greatest life loss is suffered, and the lower bound on cyclic life is established. Any conceivable reversed strain cycle can be broken down or partitioned into components of the four types of strainrange shown. Then, the lifetime of any cycle can be by employing a damage rule, predicted. Considerable experience has been gained in using Strainrange Partitioning to characterize the creep-fatigue behavior of high An international symposium temperature engineering alloys. 29) was held recently during which organizations from six (ref. NATO nations presented the results of a 2-year evaluation program of Strainrange Partitioning. By and large, the method proved to be a satisfactory approach for characterizing the creep-fatigue resistance of high temperature alloys. over 50 engineering alloys have been studied using the To date, Strainrange Partitioning method of characterization. In genthe cyclic lives can be correlated to within a factor of eral, 2. Considering that the scatter in cyclic life measurements is also on the order of a factor of 2, the correlation capabilities of the method are highly satisfactory. The second important aspect of Strainrange Partitioning is its ability to predict thermal fatigue lives. The method is currently being used to predict the cyclic lifetimes of hot section components which are life limited by thermal fatigue cracking. Under contract, the General Electric Company will be applying the method to the life prediction of a CF6-6 first-stage high pressure turbine blade squealer tip. Over the past few years Pratt & Whitney has been making life predictions of combustor liners in high bypass ratio engines using a rudimentary form of Strainrange Partitioning (ref. 30). A recently initiated contract program with Pratt & Whitney calls for a more sophisticated use of the method in making liner life calculations. Figure V-26 illustrates the severity of thermal fatigue cracking in combustor liners. Cracks have initiated and propagated from the lip edge of the louvered liner construction. The accuracy with which the thermal fatigue lives of combustor liners can be predicted is shown in figure V-27 (ref. 31). Observed lives are on the order of a few thousands of cycles to failure, with vari159

These missions result ations due to different flight missions. in different operating temperatures, strainranges, and times. Strainrange Partitioning takes into account these variations directly and accordingly predicts greater or lesser lives depending on the severity of the flight mission. Predicted lives agree with the observed lives to within factors of 2 as indiThis accuracy is considered remarkably cated in the figure. good since the correlation of laboratory specimen data is also within a factor of 2. Oxidation and corrosion prediction. - Considerable progress has also been made in developing techniques for predicting the deAt NASA gree of environmental attack at high temperatures. Lewis a technique known as COREST (corrosion estimation) has been developed for predicting oxidation attack of bare superCOREST is available as a computer program (ref. 32). alloys. With COREST the depth of oxidation attack can be estimated from weight change data. Estimated depths of oxidation attack for a number of high temperature alloys were calculated by using an early version of COREST, and these estimations are compared with actual depths of attack in figure V-28 (ref. 33). In many cases the measured values exceed the estimated values because the estimation does not take grain boundary oxidation into account. However, most actual values fall within a factor of 3 of the predictions. COREST can also be applied to predict long-time depth of attack from short-time oxidation data. In the area of hot corrosion, we are in the early stages of developing a method for predicting attack of bare superalloys. The basis for this method is a series of statistically designed, loo-hour, Mach 0.3 burner rig experiments. The variables are the corrosive elements and their concentration and the temperature. In figure V-29 good agreement between predicted 200hour attack and measured attack is shown for four superalloys exposed to a complex combination of corrosive elements. These limited results are most encouraging. The prediction of coating life is far more complex than for bare metals. The microstructure of an as-deposited NiCrAlY coating on an oxide dispersion strengthened superalloy is shown on the left in figure V-30. The protective oxide scale forming elements, aluminum and chromium, are concentrated in the darkly etched particles. After 200 hours of cyclic Mach 0.3 burner rig exposure at 20000 F, shown on the right of figure V-30, some protective elements are consumed by the oxidation reaction as can be seen from the disappearance of the darkly etched phase from the thin layer at the surface. However, the diffusion reaction between the coating and alloy has resulted in a greater loss of coating protective capability than has the reaction with the environment. We have developed a method for predicting the diffusion reaction kinetics in simple model alloy systems (ref. 160

Further research with this method and adaptation of the 34). environmental attack like prediction methods already discussed .are key approaches to obtaining a comprehensive coating life prediction methodology. INTERMEDIATE TEMPERATURE COMPONENTS - DISKS

Disks operate at lower temperatures than the hot section compoHowever, they are a critical components thus far discussed. nent in that failure poses a serious threat to safe aircraft disk materials were confined to Until recently, operation. wrought alloys and these provided definite limitations as to However, with the advances fabricability,. and cost. strength, very significant progress is in powder metallurgy techniques, being made in this area as well. The requirements for alloys to be used as disks are somewhat Environdifferent from those of the hot section components. is minimized by the lower temperamental attack, for example, Since disks are highly loaded parts, the first concern tures. uniformity of strength - so must be with strength - especially that we can design to the greatest advantage without some small Disks are large and weaker area developing a crack prematurely. heavy; therefore, cost, including material use efficiency, is a factor. And the loading on a disk is cyclic, so fatigue resistance must also be considered. The traditional practice of manufacturing disks has been to cast and then forge. This process wastes material. Also, nonuniformities inherent in the casting process are carried through The powder metallurgy proforging and weaken production disks. cess offers the potential of increased alloy additions without harmful segregation. Work on this process has also been part of NASA's MATE program. The process itself is illustrated in figure V-31. Prealloyed metal powders are loaded into a can shaped like the final product. The powder is consolidated by the combined effects of heat and pressure in a hot isostatic press. After removal of the can by etching, only minimal machining is required to bring the part into the shape necessary for quality assurance by sonic inspection. It should be noted that the scale marker below the two disks in figure V-31 is 12 inches long. So the diameter of the disks is 19 inches, more than double that which had previously been processed by powder metallurgy techniques. A finished lustrated commercial indicated, metallurgy disk made by the powder metallurgy process is ilin figure V-32. This disk would be used in a large engine to hold the first stage of blades. As already the primary advantage of making the disk by powder is uniformity; highly alloyed parts can be made with 161

confidence that all portions are similar in strength (ref. 35). The process is also conservative of materials, reducing machining by 34 percent, and is projected to result in cost savings on the order of 25 percent., As part of the continuing efforts at NASA Lewis to provide materials for hotter and more efficient engines, we are seeking to identify advanced disk materials. These are evaluated at temperatures well above current disk use temperatures. Consequentthe creep-fatigue mechanisms described previously become ly, very important. As shown in table V-2, the ultimate tensile strength attainable has been increased from about 180 000 for conventional wrought disk alloys to 221 000 pounds per square inch by means of powder metallurgy materials. Also, the number of cycles to first crack has increased dramatically from 8000 to 165 000 cycles (ref. 36). Unfortunately, the number of cycles to propagate such cracks to test specimen failure has decreased It would, of course, be desirable to improve all the alloy As is the case in all our material programs, deterproperties. mining the optimum balance of properties for a component, in continues to be a major area of investigation. this case a disk, COLD SECTION COMPONENTS Thus far the hot section and the intermediate temperature engine components have been discussed. It is appropriate to conclude by considering some of the advances being made with composite materials for the engine cold section components. Some of the components where composites can and to some extent are being used in this part of the engine include rotating fan blades and static structures such as the frame-containment ring and the exhaust nozzle flaps. A number of benefits can be gained by using composites, both polymer matrix and metal matrix, for these components. The lower density of composites combined with their greater stiffness and strength permits designers to tailor these materials to meet component requirements while improving performance and decreasing component weight. In addition, costs may be reduced by decreasing the number of parts needed for an engine. For examthe number of blades per stage or the number of stages may ple, be reduced, thereby lowering costs. Fan Blades The major impediment to using composite fan blades remains that of inadequate resistance to large object foreign object damage (refs. 37 and 38). Figure V-33 show; a number of design variations and materials combinations under study to apply composites 162

to fan blades. The designs include solid and hollow blades, a spar-shell configuration with a leading edge or central spar, and a composite patch design wherein the midspan damper is removed and a composite patch substituted for vibration dampening. Three types of composite materials are being used - polymer matrix, boron-aluminum, and superhybrids. The latter is a combination of polymer and metal matrix composites. Improvements are being made to overcome the foreign object damage limitation to composite fan blades. Engine Static Structures

There is a more immediate opportunity for application of composFigure V-34 illustrates a ites to static engine structures. graphite-epoxy fan frame for a quiet, clean, short-haul experimental engine (QCSEE) program (refs. 39 and 40). The QCSEE program conducted by NASA has been described in previous conferIt is expected that weight savings up ences held at NASA Lewis. to 30 percent can be achieved by this approach compared to curThe composite frame includes a composite containrent frames. ment ring. This component is designed to prevent a failed fan blade from penetrating the engine case. A heavy steel ring has been replaced by a lightweight fiber composite to satisfy the containment requirement. Figure V-35 shows a recent application of a graphite-polyimide composite for the nozzle flaps of the F-100 engine. The standand the substitution of composite flaps ard flaps are titanium, is intended both to increase service life and to decrease The figure shows a composite flap installed in an F15 weight. aircraft at Edwards Air Force Base to evaluate service perforgraphite-polyimide composites are mance. In addition to flaps, being applied to the thrust reverser stang of the JT-8D engine PMR polyimide, developed at the Lewis Reon the DC-9 aircraft. is used as the matrix for that component. search Center, Thus, composites appear to have a bright future for achieving significant increases in component durability, reducing weight, Further research in this area is and ultimately engine cost. continuing to fully realize their potential. SUMMARY The advances of materials are now summarized. We have made major strides of hot section component and environmental attack, and structures for propulsion systems

in developing a better understanding failure modes such as thermal fatigue and we have used this understanding

to 163

Direcprovide components with improved performance potential. such.as ODS superalloys and FRS, tionally structured materials, have demonstrated the potential for significant increases in hot section component use temperature (up to 21000 F). Coatings, both metallic and thermal barrier, can be used to increase the use temperature by retarding environmental attack and by therImproved materials for abradable mally insulating components. seals permit increases in service temperature (as much as 150 Accurate life prediction techF) with reduced leakage losses. fatigue, oxidation, niques for failure modes, such as thermal are being developed for use by designers to utiand corrosion, lize more of the service life potential of the improved materials for the engine hot section components. Powder metallurgy fabrication processing of disks, one of the key intermediate temperature engine components, permits lower fabrication costs and increased strength up to 220 ksi at 12000 F. Cold section components can be made lighter and stiffer by using composite materials. Composites also afford the potential for reduced cost through reduction of the number of required parts. Thus, it is apparent that vancing the state-of-the-art tural concepts which are gains desired in advanced REFERENCES 1. Manson, S. S.; Halford, G. R.; and Hirschberg, Fatigue Analysis by Strainrange Partitioning. Elevated Temperature Environment. American chanical Engineers, 1971, pp. 12-24. 2. Grisaffe, perature 1977. S. J.; Lowell, Environmental C. E.; Effects and Stearns, on Metals. M. H.: CreepDesign for Society for MeC. A.: High TemNASA TM-73878, major of so key turbine strides are being made in adthe engine materials and structo achieving the performance engines.

3.

Sink, L. W.; Hoppin, G. S., III, and Fujii, M.: Low-Cost Directionally-Solidified Turbine Blades, Vol. 1. (AIRESEARCH 21-2953-1, Airesearch Manufacturing Company of Arizona; NASA Contract NAS3-20073.) NASA CR-159464, 1979. Benjamin, J. S.: Dispersion chanical Alloying. Metall. 1970, pp. 2943-2951. Strengthened Trans., vol. Superalloys 1, no. 10, Alloy by MeOct. for

4.

5.

Glasgow, T. K.: An Oxide Dispersion Gas Turbine Blades. NASA TM-79088,

Strengthened 1979.

164

6.

Eng, R. D.; and Evans, D. J.: Manufacture of Low Carbon Astroloy Turbine Disk Shapes by Hot Isostatic Pressing, Vol. 1. (PWA 5574-12, Pratt and Whitney Aircraft Group; NASA CR-135409, 1978. NASA Contract NAS3-20072.) Petrasek, D. W.; and Signorelli, Strength and Microstructural Hafnium-Carbon-Wire-Reinforced NASA TN D-7773, 1974. R. A.: Stress Rupture Stability of TungstenSuperalloy Composites. of TungGas Turbine PredicRein1978.

7.

8.

Signorelli, R. A.: Review of Status and Potential sten Wire: Superalloy Composites for Advanced Engine Blades. NASA TM X-2599, 1972.

9. Winsa, E. A.; Westfall, L. J.; and Petrasek, D. W.: tion of Inlet Gas Temperatures for Tungsten-Fiber NASA TM-734842, forced Superalloy Turbine Blades. 10. Petrasek, D. W.; et A First Generation TM-79094, 1979. Barrett, Alloys Met.,

al.: Tungsten Fiber Reinforced FeCrAlY: NASA Composite Turbine Blade Material.

11.

C. A.: and Lowell, C. E.: Resistance of Ni-Cr-Al to Cyclic Oxidation at 11000 and 12000 C. Oxid. vol. 11, no. 4, 1977, pp. 199-223. and Kinetics J. Elec1051-1064.

12.

Levine, S. R.; and Caves, R. M.: Thermodynamics of Pack Aluminide Coating Formation on IN-100. trochem. Sot., vol. 121, no. 8, Aug. 1974, pp.

13.

Evaluation Talboom, F. P.; Elam, R. C.; and Wilson, L. W.: of Advanced Superalloy Protection Systems. (PWA-4055, NASA NASA Contract NAS3-12415.) Pratt & Whitney Aircraft; CR-72813, 1970. Gedwill, M. A.: Claddings for May 1972, pp. Santoro, tance Sot., and Grisaffe, Superalloys. 55-61. S. J.: Oxidation Met. Eng. Q., vol. Resistant 12, no. 2,

14.

15.

G. J.; and Barrett, C. A.: of Nickel-Chromium-Aluminum vol. 125, no. 2, Feb. 1978,

Hot Corrosion ResisJ. Electrochem. Alloys. pp. 271-278.

16.

Strangman, T. E.; Felten, E. J.; and Benden, R. S.: Refinement of Promising Coating Compositions for Directionally Pratt & Whitney Aircraft; NASA Cast Eutectics. (PWA-5441, Contract NAS3-18920.) NASA CR-135103, 1976.

165

17.

The Cyclic Oxidation ReBarrett, C. A.: and Lowell, C. E.: sistance of Cobalt-Chromium-Aluminum Alloys at 11000 and 12000 C and a Comparison with the Nickel-ChromiumAluminum Alloy System. Oxid. Met., vol. 12, no. 4, 1978, 293-311. PP= Grisaffe, space S. J.: Structural Protective Materials, Coating for Superalloys. NASA SP-227, 1969, pp. High Velocity D. L.: of Some ODS Alloys. Aero305-316.

18. 19.

Lowell, C. E.; and Deadmore, and Hot Corrosion Resistance ~-73656, 1977.

Oxidation NASA TM

20.

Comparative Thermal Fatigue Bizon, P. T.; and Spera, D. A.: Resistances of Twenty-Six Nickeland Cobalt-Base Alloys. NASA TN ~-8071, 1975. Two-Layer Thermal Barrier Coating Stecura, S.: Temperature Components. Am. Ceram. Sot. Bull., 1089. no. 12, Dec. 1977, pp. 1082-1085, Stepka, F. S.; Liebert, C. H.; and Stecura, NASA Research on Thermal Barrier Coatings. 770343, 1978. Stecura, mance 1979. Effects S.: of a Thermal of Compositional Barrier Coating Thermal Results. S.: for High vol. 56, of

21.

22.

Summary SAE Paper

23.

Changes on the PerforNASA TM-78976, System. Burner NASA ThermalBlades 1976. Rig

24.

Hodge, P. E.; et al.: Hot Corrosion Test TM-79005, 1978. Liebert, Barrier Cyclic

Barrier Coatings: DOE/NASA/2593-78/3,

25.

D. H.; et al.: Durability of Zirconia Ceramic Coatings on Air-Cooled Turbine NASA TM X-3410, Jet Engine Operation.

in

26.

An Analytical Study of Sevcik, W. R.; and Stoner, B. L.: Thermal Barrier Coated First Stage Blades in a JT9D En(PWA-5590, gine. Pratt & Whitney Aircraft Group; NASA Contract NAS3-21033.) NASA CR-135360, 1978. Bessen, I. I.; Rigney, D. V.; and Schwab, R. C.: Improved High Pressure Turbine Shroud. (R77AEG481, General Electric Co.; NASA Contract NAS3-18905.) NASA CR-135181, 1977. Hirschberg, M. H.; and Halford, G. R.: Partitioning to Predict High-Temperature NASA TN D-8072, 1976. Life. Use of Strainrange Low-Cycle Fatigue

27.

28.

166

_ _... . __.

..----_-._.-..--

29. 30.

Characterization the Strainrange

of Low Cycle Partitioning

High Temperature Fatigue AGARD CP-243, Method.

by 1978.

Vogel, W. H.; Soderquist, R. W.; and Schlein, B. C.: Awlication of Creep-LCF Cracking Model to Combustor Durability Fatigue Life Technology, T. A. Cruse, ed., Prediction. American Society of Mechanical Engineers, 1977, pp. 23-31. Reliability Prediction for Combustors Schlein, B. C.: FR-8187-Vol. 1, Pratt and Whitney Aircraft Turbines. (AFAPL-TR-77-8-Vol. 1, AD-AO45670.) Group, 1977. Barrett, C. A.; puter Program its Application ~-8132, 1976. and

31.

32.

COREST: A FORTRAN Comand Presler, A. F.: to Analyze Paralinear Oxidation Behavior and NASA TN to Chromic Oxide Forming Alloys.

33.

Comparison of Isothermal Barrett, C. A.; and Lowell, C. E.: and Cyclic Oxidation Behavior of Twenty-Five Commercial Sheet Alloys at 1150 C. NASA TN ~-7615, 1974. Reaction Diffusion Levine, S. R.: Metall. Trans., vol. Systems. 1237-1250. in the 9A, Sept. NiCrAl 1978, and CoCrAl pp. As-

34.

35.

Manufacture of Low Carbon Eng, R. D.; and Evans, D. J.: Pressing. troloy Turbine Disk Shapes by Hot Isostatic and Whitney Aircraft Group: (PwA-5574-12-Vol. 1, Pratt NASA CR-135409, 1978. NASA Contract NAS3-20072.)

36.

Evaluation of Cowles, B. A.; Sims, D. L.; and Warren, J. R.: the Cyclic Behavior of Aircraft Turbine Disk Alloys. Pratt and Whitney Aircraft Group; NASA Con(PWA-FR-10299, NASA CR-159409, 1978. tract NAS3-20367.) Improved ImpactMcDanels, D. L.; and Signorelli, R. A.: Resistance of Boron/Aluminum Composites for Use as Turbine NASA TM X-71875, 1976. Engine Fan Blades. Advanced MateSignorelli, R. A.; and Blankenship, C. P.: rials Research for Long-Haul Aircraft Turbine Engines. CTOL Transport Technology-1978, NASA CP-2036, Part 1, 1978, pp. 187-204. QCSEE - Under the Wing (UTW) Engine Composite Nacelle Test and Me(R78AEG573, Vol. 1. Summary, Aerodynamic Report. chanical Performance, General Electric Co.; NASA Contract NASA CR-159471, 1979. NAS3-18021, Proj. FEDD.) Quiet Clean Short-Haul Experimental Engine (General Electric Final Design Report. NASA CR-134847, 1977. NAS3-18021.) - Under the Wing Co.: NASA Contract 167

37.

38.

39.

40.

POTENTIAL

BENEFITS

OF 10 MIL

THERMAL ENGINE

BARRIER

COATINGS

ON HIGH

BYPASS

COMPONENT

COOLANT REDUCTION, % ENGINE AIRFLOW 4.2 1.2 .6 L 1 6. 1

TSFC, % -0.4 -. 4 -. 3 A-2 -1.3

1ST VANES 1ST BLADES 2ND VANES 2ND BLADES

ALTERNATIVELY, 4X LIFE IMPROVEMENT AT BASEUNE COOLANT FLOW.

cs-79-1612

Table V-l

ADVANCED

DISK

ALLOY

PROPERTIES

WROUGHT WROUGHT POWDER POWDER WASPALOY ASTROLOY METALLURGY METALLURGY IN 100 NASA II-B7 12CJll F UTS, ksi CYCLES TO INITIAL CRACK CYCLES TO PROPAGATE CRACK TO FAILURE 183 8000 725 197 24 000 265 196 64000 75 221 165 000 <lO

Table V-2

cs-79-1509

168

THERMALFATIGUE CRACKS IN COOLED INLET GUIDEVANE

Fig1 we V-

CYCLIC LIFE IS REDUCED BY CREEP AT TRANSVERSE GRAIN BOUNDARIES

TRANSGRANULAR (FATIGUE)

CYCLIC LIFE INTERGRANULAR (CREEP)

INCREASING TEMP OR INCREASING TIME/CYCLE Figure V-2

cs-79-1497

169

SCHEMATIC OF MODES OF OXIDATION ATTACK

PARABOLIC OXIDE GROWTH OXIDE VAPORIZATION OXIDE SP/iLLING

SCHEMATIC OF MODES OF HOTCORROSION/EROSION ATTACK

GASEOUS REACTION LIQUID DEPOSITION. OXIDE FLUXING HOT CORROSION LIQUID/SOLID DEPOSlTlON FOULING LIQUID OR SOLlD IMPACT EROSION

CULPRITS:
cs-79-1620

Na,Cl,

Ca, P, Mg Figure V-4

CARBON, ASH, DIRT

170

HOT CORRODED TURBINE VANES

Figure V-5

PROJECTED USETEMPERATURES FORTURBINE BLADEMATERIALS

2600 2400 2200 BER REINFORCED UPERALLOY (FRS) -OXIDE DISPERSION STRENGTHENED SUPERALLOY (ODS) SINGLE CRYSTAL SUPERALLOYS I. -MAR M-200 + Hf 1200 I I 1950 l%O 1970 1980 1990 2ooO Figure V-6
cs-79-1510

5:RucT;;;

171

CAST TURBINE
CONVENTIONAL CASTlNG DIRECTIONALLY SOLIDIFIED

BLADES
SINGLE CRYSTAL

1920 F COOLING REQUIRED

1920 F NO COOLING REQUIRED Figure V-7

1%5F NO COOLING REQUIRED cs-79-1579

MECHANICAL ALLOYING PROCESS

HIGH ENERGY MILL 0

UNIFORMLY ALLOYED POWDER

HOT EXTRUSION e

HEATING ELEMENT HOT ROLLING GRADIENT ANNEALING FigureV-8

cs-79-1298

172

MACRO AND MICROSTRUCTURE OF AN OXIDEDISPERSION STRENGTHENED SUPERALLOY,MA-6000E

EXTRUDED

HOT ROLLED

I 1w Figure V-9
cs-79-1507

USETEMPERATURE ADVANTAGE OF ODSSUPERALLOYS

21wr 2ow 1000 hr LIFE AT 15 DDOpsi

TEMP, OF

1900 I 1800 t 1700 1600 t-

i
B-1900 D. S. MAR M-200 + HI Figure V-10 SINGLE CRYSTAL MAR M-247 ODS MA-

cs-79-1511

173

SUPERIOR THERMAL FATIGUE RESISTANCE OF MA-6000E ODSSUPERALLOY

MAR-M200 DS SINGLE CRYSTAL MAR-M2OOt Hf + NiCrAlY OVERLAY

ODS MA-6000E ODS MA-6ODCE+ NiCrAlY OVERLAY

DS MAR-MZDO FLUIDIZED BED B-1900 3 min - 3 min I 1000 2000 3000 4000 5ooO 6000 7000 CS-79-1297 CYCLES TO FIRSTCRACK Figure V-11

EXPERIMENTAL SEGMENTED ODSCOMBUSTOR DESIGN

Figure V-12

174

STRENGTH ADVANTAGE OFTUNGSTEN FIBER REINFORCED SUPERALLOY(FRS)

TEMP: 200@ F

1000 hr STRESS/DENSITY

CS-79-1525

SUPERALLOYS

218

W-1-112 Th02

W-Re-Hf-C

Figure V-13

RELATIVE THERMAL CONDUCTIVITY AND THERMAL EXPANSION ADVANTAGE OF FRSOVER SUPERALLOYS

CONDITION: TYPICAL TURBINE BLADE TEMPS 2

THERMAL CONDUCTlVlTY 1

CS-79-1528

FRS SUPERALLOY COMPOSITE Figure V-14

FRS SUPERCOMPOSITE ALLOY

175

FIBERREINFORCED SUPERALLOY' FABRICATION PROCESS

FILAMENT MAT ALLOY FOIL MONOTAPE

;-u

CUTPf,ESFCK

AS-PRESSED BLADE

DIFFUSION BOND

CS-79-1516

Figure V-15

COMPOSITE BLADE ASSEMBLY SCHEMATIC

TRAILING k STEI

SUPERALLOY-4 OUTSERTS .;I Lfl & INSERTS L LEADING EDGE PLIES FORMED AROUND CORE Figure V-16

CS-79-1526

176

COOLED

FIBER REINFORCED SUPERALLOY TURBINE BLADE

Figure V-17

METALLICCOATING SELECTION METHOD

OXIDATION RESISTANCE AT 2OD@ F GOOD FAIR HOT CORROSION RESISTANCE

50 NICKEL CONTENT, at %

100 NICKEL CZNT, at. %

cs-79-1305

Figure V-18

177

METALLIC COATINGS EXTEND OXIDATION LIFE OF SUPERALLOYS

MACH 1 BURNER,1 hr AT TEMP= 3 min AT RT

TEMP, 20Do OF 1900 1800 10 I

UNCOATED I 100 LIFE, hr Figure V-19 1000 10 000

CS-79-1615

METALLIC COATINGS EXTEND HOTCORROSION LIFE OF SUPERALLOYS

M 0.3 BURNER RIG, 1 hr AT 900C= 3 min AT RT; 5 ppm SEA SALT

ALLOYS

/ \
0

I 4cm

I 800 hr Figure V-20

I 1200

1600
CS-79-1613

178

METALLIC COATINGS EXTEND THERMAL FATIGUE LIVESOF SUPERALLOYS

FLUIDIZED BED TEST; 2000 F (3 mid = 600 F (3 min) 1 DS + NiCrAlY COATlNG

DIRECTIONALLY SOLIDIFIED (DS) MAR M-200 I

I
10 100 1000 10 000 CYCLES TO CRACK INITIATION Figure V-21
CS-79-1617

SCHEMATICOFTHERMAL BARRIER COATINGCONCEPT

COATED CONVENTIONAL AIRFOIL THERMAL BARRIER COATED AIRFOIL

COOLING AIR

COOLING AIR I r -SUPERALLOY /- BOND COAT -.- OXIDE

I SUPERALLOY I i I I

METALLIC COATING

SUPERALLOY

CS-79-1622

Figure V-22

179

THERMAL BARRIER COATINGS OFFER LONGLIVESAT VERYHIGHTEMPERATURES

MACH 1 BURNER RIG; 1 hr HOT= 70 F (30 set) SURFACE TEMP 2640' F Zr02. 8Y 203 INiCrAlY Zr02. 8Y203/NiCrAIY 2680' F 2860' F

I
0 500 1000 1500 CYCLES TO FAILURE Figure V-23

I
2m
CS-79-1614

GENASEALIS IMPROVED SHROUD MATERIAL

CF6-5K ENGINE GROUND TEST RESULTS

BRADELLOY

BRA-

Figure V-24

180

CHARACTERIZATION OF CREEP-FATIGUE BEHAVIOR BY STRAINRANGE PARTITIONING

rN0 ,r CREEP & COMPRESSIVE CREEP COMPRESSIVE CREEP ONLY

,///,-TENSILE

CYCLIC STRAIN RANGE

CYCLES-TO-FAILURE Figure V-25

CS-79-1502

THERMAL FATIGUE CRACKS IN COMBUSTOR LINER

cs-79-1496

Figure V-26

181

ACCURACY OF STRAINRANGE PARTITIONING IN PREDICTING COMBUSTOR LINERLIFE IN HIGH BYPASS RATIOENGIN'ES

10

1 PREDICTED LIFE, CYCLES

cs-79-1503

10 10

1111

1111

Ill1

1cao 100 OBSERVED LIFE, CYCLES FigureV-27

10 ooo

#lCTlON OF OXIDATION ATTACK BY COREST PRED

1000

100 ESTIMATED DEPTH OF AllACK 10 ISOTHERMAL CYCLIC

h 1: 1

10 100 1000 MAX MEASURED DEPTH OF ATTACK cs-79-1619 FigureV-28

182

PREDICTION OF HOTCORROSION ATTACK

MACH 0.3 BURNER RIG, 1 hr 1740' F= 3 min RT, 200 CYCLES FUEL IMPURITIES IN ppm: 0.9 Na, 0.9 K, 0.45Ca, 0.45 Mg

PREDICTED DEPTH OFATrACK

CS-79-1616

MEASURED DEPTHOFAllACK Figure V-29

ENVIRONMENTAL AND SUBSTRATEREACTIONS DEGRADE COATINGS

AS-DEPOSITED NiCrAlY COATING

CS-79-1621

AFTER.200 hr AT 2000 F

Figure V-30

183

HOTISOSTATIC PRESSING OF LARGE DISK

FILLED CONTAINER
cs-79-1506

HOT ISOSTATIC PRESS Figure V-31

INSPECTION SHAPE

ADVANTAGES: UNIFORMITY 34% CHIP REDUCTlON 2% COST REDUCTlON

184

COMPOSITE FANBLADE
MATERIALS & DESIGNS DESIGN VARlATlONS SOLID HOLLOW LEADING EDGE SPAR CENTRAL SPAR COMPOSITE PATCH STIFFENED MATERIAL COMBINATIONS GRAPHITE GLASS KEVLAR BORON I EPOXY POLYIMIDE

BORON/ALUMINUM SUPERHYBRID

CS-79-1527

Figure V-33

COMPOSITE FANFRAME/CONTAINMENT RING - --l---Y

CS-79-1524

Figure V-34

185

COMPOSITE NOZZLE FLAP FOR F-100

ENGINE

Figure V-35

186

VI. David

COMPUTATIONAL

FLUID

MECHANICS

OF INTERNAL

FLOW

N. Bowditch, William D. McNally, Bernhard H. Anderson, John J . Adamczyk, and Peter M. Sockol National Aeronautics and Space Administration Lewis Research Center

Computational fluid mechanics, a method for determining the internal flows of a propulsion system, shares many solution techniques with external-flow computations. However, internal flows present unique problems that require different procedures in applying those solution techniques or, in some cases, completely different approaches. Some of these problems are illustrated by the variable-cycle propulsion system for supersonic cruise shown in figure VI-l. The internal flows of this propulsion system include supersonic, transonic, and subsonic flows in the inlet and nozzle and speed variations from subsonic to supersonic on a single fan blade row. The viscous layers are very important for internal flows. This is particularly true where the annulus heights are small and the adverse pressure gradients are large, such as in a compressor or an inlet, which can have pressure ratios of 20 or higher. The latter stages of the compressor encounter boundary layers over a major portion of the passage. These boundary layers interact with the flow-field curvature to Therefore, the procreate secondary flows in the blade rows. pulsion-system flow analysis poses many problems that are different from those encountered in computing the external flows around the aircraft. This paper discusses some major solution techniques for internal computational fluid mechanics and presents some examples from the Lewis Research Center program. It then discusses the major steps involved in developing a large computer code. Until recently, internal aerodynamics has relied on a combination of analysis and empirical correlations that were obtained from many hours of testing. Testing is becoming increasingly expensive as component performance becomes more demanding in order to satisfy the requirements of complex engines such as the variable-cycle engine in figure VI-l. While the cost of testing has been increasing, the relative computation cost has been dewhere the relative computacreasing, as shown in figure VI-2, tion cost for many of the computers introduced in the last 25 years is plotted at the year each new computer became available. A value of 1 corresponds to the IBM 360-67 cost. Currently, computation costs are one-hundredth of that cost and are still declining. internal computational fluid mechanics Therefore, becomes more attractive as computation costs continue to decline. 187

The goals for the Lewis program include rapid simulation of complex flows combined with detailed flow-field descriptions. Computations provide much more detailed information about the flow than most researchers can afford to measure. The goals also include simulation of flows that are too costly to duplicate experimentally and rapid configuration optimization. Meeting these goals will reduce the cost, time, and risk to arrive at improved designs. SOLUTION TECHNIQUES FLUID MECHANICS USED IN INTERNAL COMPUTATIONAL

The major solution techniques used in internal computational fluid mechanics are shown in figure VI-3. The most complex equations are the full viscous equations. Solving these equations for turbulent flow requires calculating all the eddies This photograph was shown in the upper left of figure VI-3. Even with taken with a camera that was moving with the flow. the special-purpose computers currently planned, solving these equations would require so long that it could only be done for research in fluid mechanics. Therefore, the first of many efforts to model the flow is done in a manner that simplifies the equations so that they can be The effect of the eddies on the flow is solved more quickly. simulated with a turbulence model so that the viscous profiles This gives shown on the right in figure VI-3 can be calculated. the Reynolds-averaged Navier-Stokes, or viscous, equations. These are still complex enough that current computers can only economically calculate relatively simple two- and three-dimensional flows. more modeling is needed to For less time-consuming solutions, Addiobtain the methods shown at the bottom of figure VI-3. tional modeling can provide faster viscous marching methods. By separating the viscous layer from the free-stream flow, we can model each separately and can still obtain certain interactions by using interacting free-stream - viscous methods. By neglecting all viscous effects and using numerical inviscid methods, we can concentrate the computation effort on defining the shocks and the large flow gradients in regions like the leading edge of For some time-dependent flows, such as a compressor blade. those associated with flutter and forced vibration, solutions Therefore, must be obtained in seconds. further modeling is required to obtain analytical methods. In the following discusthese methods are described, and examples from the Lewis sion, program are presented.

188

Analytical

Methods

for

Unsteady

Flow

Unsteady linearized aerodynamic theory has been used at Lewis to analyze the onset of supersonic flutter in axial-flow fan and These models assume that detailed steadycompressor stages. state flow physics is not required to model supersonic flutter. the unsteady flow field generated by the blades is In addition, viewed as a small perturbation superimposed on a steady base analytical solutions were develFrom these assumptions, flow. These oped for the supersonic flutter problem (refs. 1 to 3). solutions are computationally efficient and, hence, can be used in conjunction with large structural dynamic codes to perform parametric aeroelastic design studies. In figure VI-4 the flutter regions analyzed at Lewis are shown The abscissa on superimposed on a typical fan performance map. this figure is the weight flow entering the fan. The ordinate A typical operis the ratio of exit to inlet total pressure. ating line is shown. The stable operating range of the machine A region of supersonic unstalled extends up to the stall line. flutter extends from near the stall line to wide-open discharge. This unstalled flutter region forms a barrier that potentially At high limits the design of efficient, high-speed fan stages. back pressure the flutter mode in the unstalled flutter region At low back pressure the flutter is usually the bending mode. mode is primarily torsion. At moderate back pressure the flutter mode tends to be a combination of the two. The flutter region shown paralleling the stall line is supersonic stall flutter. The flutter mode in this region is primarily the bending mode. Supersonic unstalled flutter. - The supersonic flutter analyses developed at Lewis analyze the unsteady flow field surrounding a cascade of airfoils like those in the tip region of a high-speed fan or compressor. A physical model of the cascade flow field at low back pressure is shown in figure VI-5. The relative Mach The number approaching and exiting the cascade is supersonic. shock waves in the entrance, passage, and exit regions are weak and oblique. An analytical model of this flow field is shown on the right of the figure. The steady flow is a uniform, supersonic stream entering and exiting the cascade. The airfoils in the cascade are represented as flat plates oscillating in simple harmonic bending and torsional motion. This motion is the source of the unsteady flow perturbations. A physical model of the flow field at moderate back pressure is shown in figure VI-6. The entrance-region shock structure is essentially unchanged from its configuration at low back pressure. In the exit region of the passage a normal shock exists. Downstream of this shock the flow field is subsonic. In the analytical model the steady flow entering the cascade is a uniform supersonic stream. A normal shock located at the exit of 189

the passaqe The airfoils flat plates. in time are

takes the flow down to a uniform subsonic stream. are again assumed to be harmonically oscillating The shape and position of the shock at an instant determined as part of the solution.

A physical model of the flow field at high back pressure is At this level of back pressure, the norshown in figure VI-7. mal shock is located at the entrance, with subsonic flow downstream. In the analytical model the steady flow entering the supersonic stream shocked down to a cascade is again a uniform, uniform, subsonic stream by the normal shock at the passage entrance. as in the previous model, Is The motion of this shock, The airfoils in the cascade determined as part of the solution. are flat plates harmonically oscillating in time. The results of applying these three analytical models to a cascade of airfoils oscillating in torsional motion are shown in figure VI-8. The aerodynamic work performed by the airstream on an airfoil during a cycle of motion is shown as a function of The interblade phase angle at three levels of back pressure. interblade phase angle parameter is directly related to the relative displacement between adjacent airfoils in a cascade. If this parameter is Oo or 3600, all the airfoils in the cascade oscillate in unison; if this parameter is 1800, alternate airWhen the aerodynamic foils in the cascade oscillate in unison. the airstream is supplying energy to the airwork is positive, foils. Unless this energy is dissipated by damping within the any small motion imparted to the cascade will rotor structure, When the aerodynamic work is negative, the cause it to flutter. Hence, any airfoils are supplying energy to the airstream. small motion imparted to the airfoils will be quickly damped. These results show that in moving from a region of high back pressure to a region of low back pressure, the aerodynamic work becomes positive and this positive work causes torsional flutThis result is consistent with the experimental observater. tions shown in figure VI-4. The results of the analytical models applied to a cascade of At low and moderairfoils in bending are shown in figure VI-g. ate back pressures the aerodynamic work is always negative. at these operating conHence, bending flutter should not occur if the cascade were operated at high back ditions. However, pressure the analysis predicts that the aerodynamic work would phase angle. If become positive over a range of interblade damping in the rotor assembly is not sufficient to dissipate The this energy, the cascade will experience bending flutter. prediction of supersonic unstalled bending flutter at high back pressure is consistent with the experimental observations shown Experimental measurements are currently being in figure VI-4. obtained in the Detroit Diesel Allison supersonic cascade wind tunnel to validate the supersonic unstalled flutter analyses. 190

- As the operating point of Supersonic stalled bending flutter. the fan stage is moved toward its stall limit, supersonic The flow-field model for stalled bending flutter can occur. The flow apthis flutter regime is shown in figure VI-lo. The analytical model for proaching the cascade is supersonic. this flow field represents the cascade as a deforming actuator The disk that changes the direction and energy of the flow. motion of an airfoil in the cascade is represented by a combinaThis motion of simple harmonic bending and chordwise motion. tion is related to the deformation of the actuator disk in space and time. Strong detached shocks exist near the stall line at The strength of these waves is the entrance to the cascade. sufficient to cause the turbulent boundary layer on the suction The effects of these surface of the airfoils to separate. shocks and the separated flow regions on the unsteady flow field are modeled by using empirical rotor-blade-element loss and turning correlations. Results of a study using this supersonic stalled flutter model are shown in figure VI-11. The cascade geometry and the steadystate flow field were chosen to represent the tip region of the The General Electric Fan C from the NASA Quiet Engine Program. maximum aerodynamic work predicted over the entire range of interblade phase angle is shown as a function of weight flow and wheel speed. Also shown is the estimated energy dissipated in Flutter is assumed the rotor assembly during a cycle of motion. to occur if the aerodynamic work exceeds the energy dissipated in the structure. Transforming this stability boundary to the Fan C performance map yielded the results shown in figure VI-12. The agreement between theory and experiment is quite good. Currently, General Electric, under contract to NASA, is performing aeroelastic experiments on a l/3-scale model of Fan C. This program will supply additional data for validating the semi-empirical stalled flutter model. Current Lewis in-house research in unsteady aerodynamics involves developing analytical models to assess the effects of airfoil geometry on flutter and developing analyses to predict unsteady air-loads caused by flow irregularities. Numerical Inviscid Methods

Several methods of solving the full inviscid equations are shown in figure VI-13. The full inviscid equations, or Euler equations, cannot handle viscous dissipation such as boundary layers or mixing. However, they do retain the ability to handle totalpressure variations, entropy changes across shock waves, and compressibility. Solving these equations for subsonic flow requires successive calculations through the mesh to relax the solution or to march it in time to converge to a steady-state solution. Time-marching solutions, discussed in a later sec191

are applicable to either subsonic or supersonic flow. If tion, the flow is supersonic, it can also be simulated with the method of characteristics. By further modeling the flow to get the solutions ar,e limited to isentropic flow. potential equations, Finite difference methods can be used to simulate compressible Further modeling to an incompressible flows with weak shocks. flow allows paneling procedures to be used. Examples of these methods are described in the following sections. Three-dimensional paneling method. - The three-dimensional paneling method represented in figure VI-i4 was recently acquired by Lewis from Douglas Aircraft Corp. so that we could design three-dimensional subsonic inlets (ref. 4). Paneling methods are different from all the other steady-state techniques to be described in that any solution can be a linear combination of four fundamental solutions. The fundamental solutions require a large amount of computer However, for a given inlet geometry, the solution for any time. mass flow can be combination of angle of attack, yaw, and inlet obtained quickly by properly combining the four fundamental soThis computational technique is thus ideally suited lutions. for parametric studies. This new analysis gives Lewis the ability to predict the nonuniform pressure on three-dimensional inlets like the scoop inlet shown in the figure. The surface static pressure is for the inlet operating at zero angle of attack. The pressure distribution at the top of the inlet indicates an overexpansion and reThe pressure distribution at the side of the inlet compression. shows less overexpansion, and that at the lower lip shows almost no overexpansion. By combining these pressure distributions it is possible to estimate the with boundary-layer analysis, operating limits of the inlet. Two-dimensional potential-flow finite difference method. - Another potential-flow problem on which progress has been made is the solution to the full potential-flow equation for two-dimensional blade-to-blade flow through compressor blade rows. A calculation obtained with a finite difference relaxation solution, developed at Lewis, to the full potential equation is shown in figure VI-15. The solution is for flow through a cascade of thin blades at a 60 setting angle and a 30 angle of One of the keys to the success of this new method is attack. the use of a unique body-fitted solution grid that is drawn on the blades at the left side of the figure. The effect of the solution grid used on the accuracy of the computed results is discussed in more detail later. Calculated results are presented in figure VI-15 in terms of blade-surface relative Mach number as a function of distance along the blade chord. These results show the ability of the method to automatically capture 192

shocks in the transonic region when time-marching techniques this type of calculation tions,

of the flow field. In the past, were used with the Euler equatook much more computer time.

Three-dimensional method of characteristics. - Potential-flow solutionssa?~d for subsonic and transonic flows, but the method of characteristics is used primarily for supersonic flows. The three-dimensional method-of-characteristics analysis presented in figure VI-16 was developed at Purdue University 5 and 6) and is used for computing the flow field in (refs. supersonic, mixed-compression inlets operating at the angle of In this method, the shock waves are computed by a disattack. crete shock-wave-fitting scheme. At the bottom of figure VI-16, the centerbody surface static pressures on the windward and leeward sides are plotted as a function of distance from the cowl lip. These results are for a mixed-compression inlet operating at Mach 3.3 and 3O angle of The data for the leeward side are compared with the attack. solid calculated line. Good agreement is obtained to just ahead of the second shock pressure rise. At this point, viscous effects apparently feed the shock pressure rise forward over a wide region. Although the current inviscid analysis predicts significant differences between the windward and leeward sides, a viscous interaction must be added to obtain better agreement at the second shock near the inlet throat. This additional effort is currently planned for next year. Three-dimensional time-marching methods. - Time-marching solutions to the full inviscid, or Euler, equations are very useful for analyzing transonic flows with shock waves and have been used at Lewis for analyzing propeller and compressor flow fields. Transonic propeller: The three-dimensional solution developed at Ames for the eight-bladed turboprop configuration shown in figure VI-17 will be used to support design studies for the AdCalculating the flow through vanced Turboprop Project Office. the blade passage starts with an initial guess of the flow field in a coordinate system attached to the blade and progresses through time to a steady-state solution. For this transonic propeller operating at a free-stream Mach number of 0.8 and a relative tip Mach number of 1.15, staticpressure distributions are shown in figure VI-18 for the hub, mean, and tip sections. All the distributions reveal substanThe tial regions of supersonic flow along the blade section. sharp changes in the static-pressure coefficient at the hub, mean, and tip sections of the blade indicate that the shock extends over the entire span. Substantial spanwise variation in blade loading is indicated by the differences in the areas be-

193

tween the suctionplanned to provide

and pressure-surface test data to verify

curves. Experiments this computer code.

are

Another time-marching solution has 1800-Foot-per-second rotor: This code is been developed for NASA at MIT (refs. 7 and 8). for analyzing flow in a three-dimensional compressor blade row, This solution of the such as the one shown in figure VI-19. full inviscid (Euler) equations permits analysis of three-dimensional flow with strong shocks in a rotating blade row that conThe general three-dimensional solution tains midspan dampers. region includes the interblade passage and the nearby entrance and exit regions. A midpassage flow surface in the hub-to-shroud direction is also Figure VI-20 shows a contour plot of pressure on this shown. midpassage flow surface for an 1800-foot-per-second-tip-speed The darker regions indicate high pressure and low velocity fan. The low pressure and high velocity. and the lighter regions, closely spaced contours near the front of the passage show the which crosses the passage to the suction side of the bow shock, The contour lines around the damper indicate flow decelblade. eration on the upper side and flow acceleration on the bottom This flow variation is caused by a negative flow inciside. This information can dence at the damper of approximately 5O. be used to reset the damper for lower losses. The time-marching technique has 1600-Foot-per-second rotor: been applied to calculating three-dimensional flow in a 1600A blade-to-blade surface foot-per-second-tip-speed fan rotor. for flow near the tip of this blade row is shown in figure VI-21. Calculated and experimental results are compared for this blade in figure VI-22. The experimental results at the right in figure VI-22 were obtained at MIT with a unique laser-fluorescence technique in which seed particles in the test gas are excited with a flat The resultant fluorescent illumination is sheet of laser light. photographed and can be directly related to the density of the These experimental results show the presence of two flow. a strong normal shock at the exit region shocks in the passage: and a detached bow shock crossing the passage from the leading edge of the upper blade to the suction surface of the adjacent The presence of two shocks in the passage was unexpectblade. yet the calculated contours shown at the left in the figure ed, These are evident confirm that two shocks should be present. from the concentration of density contours at the exit region of The analytical denthe passage and also at the leading edge. sity ratio across the bow shock is in agreement with experiment the agreement is within i0 percent for the to within 2 percent; in-passage shock.

194

Interacting

Free-Stream

- Viscous

Methods

About 70 years ago, Prandtl divided flow near surfaces into a free-stream region and a viscous region, as shown in figure VI-23. This permitted modeling in each region and simplified In many internal flows, it is often each solution technique. desirable to determine the interaction between the viscous and There are many ways to determine this infree-stream regions. but only boundary-layer displacement of the inviscid teraction, flow region and matched fluid properties at the interface between the two regions are described here. Boundary-layer displacement - two-dimensional cascade. - The prediction of fluid turning and loss for two-dimensional blade sections is a key ingredient in most compressor design systems. Figure VI-24 shows schematically the inviscid and viscous regions in a two-dimensional cascade flow and indicates the possible presence of a separation near the trailing edge. In this vicinity the coupling between viscous and inviscid effects can be quite strong. The present method uses separate finite-difference solutions in each region and couples them by means of the displacement effect, which is modeled as a surface injection of fluid (ref. 9). Experimental and analytical values of fluid turning angle and loss coefficient are compared as a function of angle of attack in figure VI-25 for subsonic flow in a cascade. Also shown is the location of the calculated separation point on the blade suction surface. Although the agreement with the data, for both the results show that the turning and loss, is only qualitative, strong upturn in the experimental and predicted losses corresponds to the rapid shift of the separation point toward the leading edge. Work on this method is continuing at Iowa State University. Boundary-layer displacement - blade-passage end wall. - Another boundary-layer displacement approach is being developed here at NASA Lewis to analyze flows in the end-wall regions of axial compressors. Most losses in turbomachinery occur in these endwall regions. The flow along the end wall of a rotating blade A two-dimensional turbulent verow is shown in figure VI-26. locity profile is shown at the inlet. The turning of this flow by the blades generates a passage vortex that in turn modifies the local blade loading and produces a three-dimensional velocThe present method (ref. 10) uses a ity profile at the exit. pitch-averaged, integral, boundary-layer procedure in the viscous region. This procedure uses an approximate solution for the secondary flow in order to evaluate many of the three-dimensional effects. The inviscid solution is currently obtained from a quasi-one-dimensional model. The two solutions are iteratively coupled by means of the displacement effect. 195

Experimental and analytical values of equivalent blockage and end-wall loss are shown in figure VI-27 as a function of axial For this case the dramatic distance for flow through a cascade. increase in blockage inside the passage is well predicted by the but the agreement for end-wall loss is only qualitative. method, Supersonic-subsonic viscous interaction method. - A different approach to the interaction of the viscous layer with the free stream is used in the supersonic-subsonic viscous interaction The method was initially demethod depicted in figure VI-28. veloped in a joint program between Lewis and Case Western ReThe numerical solution for serve University (refs. 11 and 12). the flow in the supersonic viscous region is obtained by applying the method of characteristics to a system of equations that includes some viscous terms. The system of equations for the subsonic viscous region includes, in addition to the classical the effects of transverse curvature and boundary-layer terms, normal pressure gradients. Coupling the supersonic and subsonic viscous regions at the interface requires an iteration to match the streamwise and normal velocities as well as the static and total pressures and temperatures. Although this interaction analysis is still under development at The some preliminary results are shown in figure VI-29. Lewis, calculation was performed on an axisymmetric waisted body, which is shown in the lower part of the figure. The upper part of the figure shows Mach number profiles plotted as a function of distance from the surface at various stations along the waisted body. The calculation was able to pick up the embedded shock It is in this particular wave emanating from the waist region. region that it is very important to include the normal pressure gradients across the boundary layer in the viscous solution. Viscous Methods

The fully viscous methods under study at Lewis include both the viscous marching methods and the full Reynolds-averaged viscous equations (fig. VI-30). The solutions to the Reynolds-averaged viscous equations require successive calculations through the mesh in order to relax the solution or to march it in time to a steady solution. They typically require very long run times. Since run time increases with the number of mesh points in the only a moderate number can be used and this results in flow, limited flow resolution. Through flow modeling the problem can be partially separated into two steps: An inviscid solution is used to obtain upstream effects on the subsonic flow field, and then the viscous equations are marched down the flow field to define the viscous effects. Faster run times for these marching methods permit the use of more mesh points, and this results in

196

better flow resolution. Some recent results from these viscous marching methods and a solution to the full Reynolds-averaged viscous equations are described in the following section. Axisymmetric viscous marching method. - An axisymmetric viscous 13 and 14) was developed at United Techmarching method (refs. nologies Research Center to analyze annular flows in configuraThis method tions like the nozzle depicted in figure VI-31. treats the entire flow field in any annular duct with a single set of viscous equations. It constructs a mesh like that shown in the figure, with the stream function as the normal coordinate and the velocity potential as the streamwise coordinate. Since the inviscid-flow streamlines approximate the viscous-flow streamlines, the fully viscous equations can be reduced to a set of boundary-layer-like equations that can be solved by a forward marching method. The solution mesh for the curved nozzle shown in figure VI-31 is a typical mesh that is necessary to properly resolve the flow. Velocity profiles at different stations within the curved nozzle are shown in figure VI-32. Because of channel diverqence, the flow initially diffuses until it nearly separates on the inner wall. Then the flow rapidly accelerates and the boundary layer thins as the flow turns. Finally the flow rapidly diffuses until it again nearly separates at the nozzle exit. Also clearly evident in the figure is the significant flow shift from one side of the passage to the other. Three-dimensional viscous marching method. - An application of viscous marching methods to three-dimensional flows is described in figure VI-33. This method, developed by Scientific Research Associates, was specifically designed for configurations with large geometric curvature (refs. 15 and 16). Such flows are common in both diffusers and turbomachinery. Another variation of this numerical analysis, which uses generalized coordinates can compute flows in duct geometries with (refs. 17 and 18), cross-sectional transitioning and off-set bends. The ability of the present method to predict the formation and development of secondary flows is of primary interest here. A sample calculation was made on a circular-arc duct with a square VI-33) in order to evaluate the present cross section (fig. state of development of this method. For this calculation, a 20x20 mesh was used in the transverse plane and 45 streamwise This gave stations were used in the forward marching direction. a step size THETA of 5.00. The flow was assumed to be symmetric about the half-span plane. A computer motion picture made of the calculations shows the development of the secondary flows in the transverse planes. Sample frames from this motion picture are shown in figure VI-34 for angular positions (THETA) of the cross plane every 300. 197

there is no secondary flow entering the bend (THETA = The secondary flows along the top and bottom surfaces begin so that at THETA = 300 two to move toward the suction surface large counter-rotating vortices have been well established. These secondary flows are caused by the transverse pressure gradients associated with the turning of the primary flow. By THETA = 600, a second vortex pair has been formed close to the This vortex pair is convected topressure, or right, surface. ward the suction, or left, surface so that by THETA = 900 it A third vortex pair is has merged with the first vortex pair. forming near the plane of symmetry close to the pressure surface. This vortex pair builds up in strength (THETA = 1200) This allows a smaller and moves toward the main vortex pair. vortex pair to form in the corner near the pressure surface (THETA = 1500). This vortex pair grows in strength (THETA = 1800) and translates toward the suction surface to join the initial vortex pair (THETA = 2100). Initially,
0) l

These results demonstrate the complexity of geometry-driven secondary flows that stem from the curvature and the associated To verify that the calculations represent a pressure field. real flow, we have begun an experimental effort to obtain Doppler velocimeter measurements in a similar configuration. Forced mixer nozzle: In support of the Lewis Engine Component a three-dimensional compressible viscous Improvement program, marching code is under development at Scientific Research AssoIt will be used to analyze the flow in lobed mixer nozciates. of the type depicted in figures VI-35 and VI-36. In the zles the hot core flow passes up through the lobed mixer nozzle, lobed section and the cooler fan flow passes down through the this nozzle increases mixing by providing Very simply, valleys. a much greater interface between the hot core flow and cooler By mixing the core and fan flows in this manner, a fan flow. The small but significant performance gain can be realized. level of the gain depends on the trade-offs between the degree of mixing of the two streams and the viscous losses incurred in the mixing process. The present analysis concentrates on the actual mixing process Although the present analydownstream of the lobe exit plane. sis is preliminary in that only laminar flow is considered, it addresses the important underlying fundamentals of lobed mixer flows. A sample laminar calculation was performed with this preliminary deck in order to evaluate the overall method in its present state of development and to demonstrate the potential of this method for predicting the flow and the mixing process in lobed mixers. The solution was computed for the geometry shown in The Reynolds number based on the hot-stream vefigure VI-36. locity and the inlet casing diameter was 900, the Mach number 198

of the inlet plane was 0.35 in both streams, and the temperature Because of nozzle symmetry, ratio between the streams was 2. only a pie-shaped computational segment covering a half lobe had The computational mesh within this segment conto be analyzed. sisted of 20 points in the radial direction and 10 points in the circumferential direction. A computer motion picture was made showing the development of total temperature and streamwise vorSeveral simulated frames ticity at successive axial stations. showing profiles of temperature and streamwise vorticity are presented as figure VI-37. The locations are identified by the station number, which increases from 0 at the lobe exit to 32 at the mixer nozzle exit. At station 0, there is no streamwise vorticity, and the totalAs the sotemperature plots clearly define the lobe geometry. lution begins to move downstream, there is relatively little change in the total temperature. Vortices are generated on both sides of the radial interface between the core flow and the fan flow (station 6). As the solution progresses downstream, the vortices initially generated in the core flow dissipate and the fan vortex pairs dominate (station 12). As the vortex pairs intensify near the end of the plug (station 19), faster mixing appears to take place. The overall effect is to enhance mixing. The vortex intensity and resulting enhanced mixing are probably This due to the curvature of the flow at the end of the plug. suggests that the centerbody shape is important to the mixing Downprocess through the generation of streamwise vorticity. the streamwise vorticity dissistream of the plug (station 32), pates. Three-dimensional - -solution to the solution is for peller shown in full viscous method. - The final example full Reynolds-averaged viscous equations. flow in a blade passage of the centrifugal figure VI-38. is a The im-

There is currently considerable interest in centrifugal compresThe sors for small jet engines and for automotive applications. flows in these devices are quite complex and often contain large separated regions. The present technique was developed under an NASA contract with Therm0 Mechanical Systems (ref. 19). It solves the steady viscous equations, first on blade-to-blade surfaces that are swept from hub to shroud and then on crossBy altersectional surfaces that are swept from inlet to exit. nating between these two sweep directions, the full solution can eventually be obtained. The method is quite time consuming and takes several hours on the fastest computer that is now available. A velocity-profile plot of the solution on a blade-to-blade surThe face of a centrifugal impeller is shown in figure VI-39. calculation is for laminar flow and exhibits a large separation on the suction surface near the trailing edge. A turbulence 199

model has recently ment is now under

been way.

added

to

the

code,

and

further

develop-

DEVELOPMENT OF A COMPUTER CODE The steps in the development of a computer code and some of its Developing a code from key elements are shown in figure VI-40. a basic algorithm to a design-analysis code can take many years and includes geometry and graphics work that may have little to The process usually starts with the do with fluid mechanics. which is a numerical scheme to solve the approbasic algorithm, priate equations for a relatively simple geometry and flow. For the initial code may only treat laminar flow on a flat example, To extend that capability to more practical flows and to plate. obtain what we call a research code require considerable effort. Initial and boundary conditions, a turbulence model, and more complex geometry and mesh are all part of this effort. Geometry and Mesh

Geometry and mesh generation are often a major roadblock to the use of a research code. The geometry often needs to be quite and the mesh characteristics often affect detailed and smooth, Therefore, the computed results. computer assistance is frequently needed to overcome this barrier. Enrichment techniques. - Experience with three-dimensional potential-flow paneling techniques has indicated that the large number of panels depicted on the right in figure VI-41 are required to describe the inlet geometry in sufficient detail to obtain an accurate calculation. over a thousand Typically, However, panels can be necessary for complex inlet geometries. it is difficult and time consuming to input all the necessary information by hand. Therefore, techniques have been developed at Lewis to smooth and enrich the input geometry, that is, to input a few points and have the computer enrich and construct the necessary number of panels. - In addition to describing the geomeEffect of mesh geometry. try of a configuration to the computer, there is the subsequent problem of generating an appropriate grid, if a finite-difference solution is to be used. In numerical code development, the solution grid, or mesh, can greatly influence the accuracy of the calculated results. Shown in figure VI-42 are two finite-difference meshes applied to the same cascade of thin blades. At the left is a conventional sheared mesh of throughflow lines and blade-to-blade lines that has been used successfully with a number of numerical methods. When applied with the new full-potential relaxation scheme described previously, this mesh yielded 200

The terthe surface Mach number distribution in figure VI-43. minal shock expected at the end of the supersonic zone was completely smeared by the numerical algorithm applied on this The spike near the blade trailing edge is another sheared mesh. A proposed result of the poor performance of the sheared mesh. solution to this shock resolution problem was the application of in figa less conventional wraparound mesh, shown at the right ure VI-42. Although this mesh is more difficult to generate, it has the advantages of grid lines nearly normal to the blade surface in a direction alined with the shock and of clustering grid points near the leading and trailing edges of the blades, where The improved resolution of the terflow gradients are strong. minal shock to the supersonic bubble is very evident in figure VI-44, which now has the solution for the wraparound mesh superimposed on the original solution for the sheared mesh. Graphics and Data Handling

In addition to dealing with complex geometries and generating grids tailored to finite-difference schemes, there is the problem of extracting meaningful information from the tremendous volume of output generated by the large three-dimensional and Sophisticated computer graphics and data-hanviscous codes. dling techniques such as those shown in figure VI-45 are reThese techniques include quired to alleviate this problem. contour representation, geometry profile development, x-y plots, and three-dimensional streamline tracing, and vector plots, To make graphical techniques available function representation. to all Lewis flow analysis codes, we have modified them to build This file can be accessed by any of a standard bulk data file. the graphical techniques represented in this figure. It is also very important that the engineer have a variety of These might ingraphical hardware available to display data. clude a Calcomp plotter, cathode ray tubes, microfilms, and moColor capability must also be included because tion pictures. it can be used very effectively to enhance the meaning of graphical representations. Much work still has to be done particularly graphics techniques, and unsteady flow fields. Benchmark Experiments there is codes. are re201 in developing data-handling for three-dimensional and steady

In addition to the coding efforts previously outlined, the important problem of verifying the accuracy of the Carefully planned and executed (benchmark) experiments

quired resent

to prove that the models incorporated the physics of the real flows.

in

the

coding

rep-

Effective benchmark experiments have the following characteristics: First, they are conducted in simple geometries that isolate a particular flow phenomenon. The flow situation is simplified initially so that various aspects of the numerical algorithm can be verified. Geometry and boundary and initial conditions are clearly defined. Redundant measurements are often conducted with differing kinds of instrumentation in order to Finally, give credibility to the results. the results must be well documented and available to everyone. NASA Lewis recognizes the extreme importance of this type of activity and will be emphasizing benchmark experiments in the near future. Design-Analysis Codes

A large effort is required to obtain a verified production for the designer to use this code, additional However, code. assistance is needed. The code can require long computer run for efficient use, the number of full and therefore, times; solutions obtained must be minimized. This can be done by using the interpolation and optimization techniques described in the following paragraphs. Numerical optimization procedures. - So that long-running codes can be used for design, procedures must be developed to aid the user in selecting an optimum configuration. A numerical optimization procedure was developed at Lewis so that the computer instead of the wind tunnel, to search for the could be used, best inlet bleed configuration. This computer code is diagrammed in figure VI-46. Data are introduced to initialize the which are the bleed rates at the shock decision variables, boundary-layer interaction regions. The optimization routines make changes in these decision variables, and these changes are evaluated by the aerodynamic programs, subject to the objective and constraint functions. The objective function in this problem is maximum range; the constraint function is avoiding boundary-layer separation. This process is repeated until an optimum, or final, design has been reached. The variation in range decrement and total bleed rate as a function of the number of iterations is shown in figure VI-47. Forty iterations were required to obtain a minimum range decrement, or a maximum range. The bleed rate obtained in the wind tunnel for a Mach 2.5 mixed-compression inlet agreed with the results of the computer optimization procedure. This was accomplished in about 30 minutes of computing time on the Univac 1110; this compares to several weeks of varying the individual bleed zones on the inlet model in the lo- By iO-Foot Supersonic Wind Tunnel. 202

Fast-perturbation solutions for design. - There is also a need toassist the designer in using long-running codes throughout Information from existing solutions can be the design process. used to predict other solutions at slightly perturbed flow conThis approach is ditions or with slight changes in geometry. particularly helpful in blade profile design, where parametric studies are often conducted over a range of flow conditions and geometry perturbations. A method has been developed for NASA Lewis by Nielsen Engineering (ref. 20). In this method, perturbation solutions can be obtained easily and rapidly for changes in a number of flow and geometry parameters. 'These solutions use previously calculated In figure VI-48, base solutions from the longer running codes. the two dashed-line solutions represent the two base solutions obtained with traditional analysis codes. These were run on a parabolic-arc blade profile for angles of attack of 3.00 and 4.00. The perturbation code, which uses information from these base solutions, was then run to obtain a solution at an angle of attack of 3.750. This solution is represented by the circular symbols in the figure, and it was obtained in less than 1 percent of the computer time used by either of the base solutions. So that its accuracy can be evaluated, It is compared in the figure to a third full-base solution for an angle of attack of 3.750; and it is evident that the agreement is excellent. Results for perturbations of other variables show just as good agreement in the same short computer time. This code will soon be extended to perturb several design variables simultaneously. CONCLUDING REMARKS As discussed in the examples, many internal flows can now be analyzed by using computational fluid mechanics. A large effort is required to develop the computer codes needed to analyze these flows from.a basic algorithm to a code that can be used reliably for designing aircraft hardware. Most of the results discussed in the text are from codes that are only part way through that development cycle. Therefore a lot of work is still required to compile,the library of codes needed to analyze all the internal flows in a propulsion system. The Lewis Research Center is convinced that the effort is worthwhile and is currently constructing the Research Analysis Center shown in figure VI-49 to house new computer faciiities that will support this work. The results of the effort should allow the designer to produce a better product cheaper, faster, and with less risk.

203

REFERENCES 1. Goldstein,
FLOW

in Shocks.

M. E.; Braun, W.; and Adamczyk, J. J.: Unsteady a Supersonic Cascade with Strong In-Passage Fluid Mech., vol. 83, Dec. 5, 1977, pp. 569-604. Flow Locus. in a AIAA

Unsteady 2. Adamczyk, J. J.; and Goldstein, M. E.: Supersonic Cascade with Subsonic Leading-Edge J ., vol. 16, no. 12, Dec. 1978, pp. 1248-1254. 3. Analysis of Adamczyk, J. J.: ter in Axial-Flow Compressor NASA TP-1345, 1978. Supersonic by Actuator

Stall Bending Disk Theory.

Flut-

4.

Hess, J. L.; Mack, D. P.; and Stockman, N. 0.: An Efficient User-Oriented Method for Calculating Compressible Flow About Three-Dimensional Inlets. (Douglas Aircraft Co., NASA Contract NAS3-21135.) NASA CR-159578, 1979. Inc.; Vadyak, J.; and Hoffman, J. D.: Calculation of the Flow Field in Supersonic Mixed-Compression Inlets at Angle of Attack Using the Three-Dimensional Method of Characteristics with Discrete Shock Wave Fitting. (TSPC-78-1, Purdue NASA Grant NGR-15-005-162, NGR-15-005-191.) [Jniv.; NASA CR-135425, 1978. Vadyak, J.; Hoffman, J. D.; and Bishop, A. R.: Calculation of the Three-Dimensional Flow Field in Supersonic Inlets at Angle of Attack Using a Bicharacteristic Method with Discrete Shock Wave Fitting. AIAA Paper 79-379, Jan. 1979. Epstein, A. H.; Kerrebrock, J. L.; and Thompkins, Shock Structure in Transonic Compressor Rotors. vol. 17, no. 4, Apr. 1979, pp. 375-379. Thompkins, Computed Transonic W. T., Jr.: AIAA J.,

5.

6.

7.

8.

W. T.; and Epstein, A. H.: A Comparison of the and Experimental Three-Dimensional Flow in a Compressor Rotor. AIAA Paper 76-368, July 1976.

9.

Hansen, E. C.; Serovy, G. K.; and Sockol, P. M.: Axial-Flow Compressor Turning Angle and Loss by Inviscid-Viscous Interaction Blade-to-Blade Computation. ASME Paper 79-GT-5, Mar. 1979. Sockol, P. M.: Compressors. End-Wall Boundary Layer NASA TM-78928, 1978. Prediction for Axial

10. 11.

Anderson, R. H.; Tassa, Y.; and Reshotko, Procedure for Supersonic Flows Including Viscous Contribution to Flow Rotationality. 76-426, July 1976.

E.: Characteristic Consideration of AIAA Paper

204

12.

Tassa, Y.; Anderson, B. H.; and Reshotko, E.: Finite ence Procedure for Boundary Layers Including Effects Longitudinal and Transverse Curvatures. AIAA Paper 76-427, July 1976.

Differof

13.

Anderson, 0. L.: A Comparison of Theory and Experiments for Incompressible, Turbulent, Swirling Flows in Axisymmetric Ducts. AIAA Paper 72-42, Jan. 1972. Anderson, Swirling, Struts. 0. L.: Finite Compressible NASA CR-2365, Difference Solution Flow in Axisymmetric 1974. for Turbulent Ducts with of Inter-

14.

15. 16.

McDonald, H.: nal Flows.

Computational Fluid AIAA Paper 79-1445,

Dynamic Aspects July 1979.

Briley, W. R.; and McDonald, H.: Computation of ThreeDimensional Viscous Subsonic Flows with Large Streamwise Vorticity. AIAA Paper 79-1453, July 1979. Levy, R.; McDonald, H.; and Briley, W. R.: Calculation of Three-Dimensional Turbulent Subsonic Flows in Transition Ducts. 6th International Conference on Numerical Methods in Fluid Dynamics, Springer-Verlag, 1973, pp. 184-192. Eisman, P.: A Unification tions. tjth International ods in Fluid Dynamics, of Unidirectional Flow ApproximaConference on Numerical MethSpringer-Verlag, 1979, pp. 72-79. An Alternating Jr.: Three-Dimensional SAE Paper 7a1001,

17.

18.

19.

Walitt, L.; Liu, C. Y.; and Harp, J. L., Direction Explicit Method for Computing Viscous Fiow Fields in Turbomachines. Nov. 1978.

20.

PerStahara, S. S.; Chaussee, D. S.; and Spreiter, J. R.: turbation Solution for Transonic Flow on the Blade-toNASA CR-294i, Blade Surface of Compressor Blade Rows. 1978.

205

INTERNAL COMPUTATIONAL FLUIDMECHANICS FORPROPULSION SYSTEMS

SOLUTION TECHNIQUES

DEVELOPMENT OF A COMPUTER CODE Figure VI-l

cs-79-2346

TRENDIN COMPUTATION COST

100

IBM 650

RELATIVE COMPUTATlON COST

CS-7g-2262

1955

CRAY 2. I I I I 1960 1965 1970 1975 1980 1985 YEAR NEW COMPUTER AVAILABLE

B IBM 370-195 IL bbUU 0. -STAR 100 ^-a W.^^ LLJL IbUU l CRAY 1 - ... . .. . . . . . 8 ILLIAL-4

Figure VI-2

206

INTERNAL

COMPUTATIONAL FLUID SOLUTION TECHNIQUES

MECHANICS

TURBULENCE MODELING

UNSTEADY FLOW

Figure VI-3

cs-79-1819

REGIONSOF COMPRESSOR FLUTTER

SUPERSONIC

BACK PRESSURE HIGH (BENDING)

MODERATE (BENDING/ TORSION) LOW (TORSION)

WEIGHT FLOW Figure VI-4

(35-79-2286

207

SUPERSONIC UNSTALLED FLUTTER

LOW BACK PRESS. MODEL PHYSICAL ANALYTICAL

M>l

cs-79-2270

Figure VI-5

SUPERSONIC UNSTALLED FLUTTER

PHYSICAL MODERATE BACK PRESS. MODEL ANALYTICAL

M<l

Figure VI-6

208

SUPERSONIC UNSTALLED FLUTTER

HIGH BACK PRESS, MODEL PHYSICAL ANALYTICAL

M<I

CS-79-2271

Figure VI-7

SUPERSONIC UNSTALLED FLUTTER ANALYSES

PREDlCTlON FOR TORSIONAL MOTION

90 180 270 360 0 INTERBLADE PHASE ANGLE, deg Figure VI-8

CS-79-2285

209

SUPERSONIC UNSTALLED FLUTTER ANALYSES

PREDICTION FOR BENDING MOTION

I AERODYNAMIC 0

FLUTlER

BACK PRESSURE

NO FLUTTER I I I I 270 360 0 90 180 INTERBlADE PHASE ANGLE Figure VI-9

~~-79-2281

SUPERSONIC STALLED BENDING FLUllER SEMIEMPIRICAL ANALYSIS

,-EMPIRICAL ROTOR BLADE ELEMENT LOSS & ! TURNING CORRELATlONS ! !
I -I +---

M>l

(38-79-2269

Figure VI-10

210

SUPERSONIC STALLED BENDING FLUTTER

PREDICTION FOR NASA QUIET ENGINE FAN C

AERODYNAMIC WORK/CYCLE

50

60 70 80 % OF DESIGN WEIGHT FLOW Figure VI-11

90

100

SUPERSONIC STALLED BENDING FLUTTER

COMPARISON OF ANALYSIS & EXPERIMENT; NASA QUIET ENGINE FAN C 1.8 PREDICTED BENDING FLUllER

. % OF DESIGN 95 WHEEL SPEED

1.6 TOTAL PRESSURE RATIO

1.4

1.2
.

0 l.Ol 40 60 80 100 % OF DESIGN WEIGHT FLOW Figure VI-12

MEASURED FLUTTER

211

NUMERICAL INVISCIDMETHODS

CS-79-2267

Figure VI-13

3-D POTENTIAL FLOWPANELING METHOD

SCOOP INLET

SURFACE DISTANCE

Figure VI-14

212

2-D POTENTIAL FLOW FINITEDIFFERENCE METHOD

1.2SUCTlON

% CHORD
cs-79-2337

'VI-15

3-D METHOD OF CHARACTERISTICS

INLET AT 3' ANGLE OF ATTACK

CENTERBODY SURFACE STATlC PRESS.

C5-79-2192

DISTANCE FROM COWL LIP FigureVI-16

213

3-D TIME MARCHING METHOD

TRANSONIC PROPELLER; &= 0.8, MTlP = 1.15

,-MEAN

rTlP !'

CS-79-2178

Figure VI-17

3-D TIME MARCHING METHOD

TRANSONIC PROPELLER

r
STATIC PRESS. COEFF

HUB

MEAN Figure VI-18

np

(35-79-2179

214

3-D TIME MARCHING METHOD

1800 fps ROTOR FLOW SURFACE

r:-7: Figure VI-19

3-D TIME MARCHING METHOD

1800fps ROTOR PRESSURE CONTOURS

-i/

Figure VI-20

215

3-D TIME MARCHING METHOD

1600 fps ROTOR

BLADE-TO-BLADE FLOW SURFACE

CS-79-2273

FigureVI-21

3-D

TIME MARCHING METHOD

1600 fps ROTOR

CALCULATED DENSITY CONTOURS

EXPERIMENTAL SHOCK LOCATIONS

(35-79-2288

FigureVI-22

216

INTERACTING FREE-STREAM -VISCOUS METHODS

FREE-STREAM REGION

VISCOUS REGION

INTERACTION TECHNIQUES BOUNDARYLAYER DISPLACEMENT OF INVISCID REGION MATCHED FLUID PROPERTIES AT INTERFACE Figure VI-23
(35-79-2264

INTERACTING INVISCID/VISCOUS METHOD

BOUNDARY LAYER DISPLACEMENT - 2-D CASCADE

,VI scous REGION

Figure VI-24

217

INTERACTING INVISCID/VISCOUS METHOD

EXPERIMENTAL/ANALYTICAL COMPARISON - 2-D CASCADE

SEPARATION POINT

LOSS COEFF

FLUID TURNING ANGLE

TC-,,?P1

48hk7hk-Y
ANGLE OF AilACK Figure VI-25

18

INTERACTING INVISCID/VISCOUS METHOD

BOUNDARY LAYER DISPLACEMENT - BLADE-PASSAGE END WALL

PROFILE

MEAN 2-D VELOCITY

CY-79-2276

Figure VI-26

218

INTERACTING INVISCID/VISCOUS METHOD

MPERIMENTALIANALYTICAL COMPARISON BLADE-PASSAGE END WALL

LEADING EDGE

TRAILING EDGE Figure VI-27

es-79-2279

INTERACTING SUPERSONIC/SUBSONIC VISCOUS METHOD

MATCHED FLUID PROPERTIES AT INTERFACE

VISCOUS REGION

CS-79-2172

Figure VI-28

219

INTERACTING SUPERSONIC/SUBSONIC VISCOUS METHOD

WAISTED BODY MACH NO. PROFILES

DISTANCE FROM SURFACE MACH NO. PROFILES 3)

AXISYMMETRIC WAISTED BODY Figure VI-29

cs-79-2174

VISCOUS METHODS

AVERAGED VISCOUS LONG RUN TIME

Figure VI-30

220

AXISYMMETRIC VISCOUS MARCHlNG METHOD

CS-79-2176

Figure VI-31

AXISYMMETRIC VISCOUS MARCHING METHOD

CURVED NOZZLE VELOCITY PROFILES

Figure VI-32

221

_-

. ..__..._..

--

..e

.--

111-1

3-D VISCOUS MARCHING METHOD

CIRCULAR ARC DUCT

SURFACE

CS-79-21x1

FigureVI-33

3-D VISCOUS MARCHING METHOD

CIRCULAR-ARCDUCT

.... ..... ......... ......... ......... ......... ......... ......... ......... I:::::: : : :

......... ......... ......... ......... ......... ......... .........

\x.:::::5///... -///////ll... :-/--/-.-. j,,.,. _-// - - - r, * . . . . _--- - - - , ..,.

Figure VI-34

222

MIXER NOZZLE

LOBE GEOMETRY

Figure VI-35

3-D VISCOUS MARCHING METHOD

MIXER NOZZLE SOLUTION MESH COMPUTATIONAL ssT

cs-79-2170

Figure VI-36

223

3-D VISCOUS MARCHING METHOD

MIXER NOZZLE

STATION0

STATION6

STATION12

STATION19

STATION32

Figure VI-37

224

3-D FULLVISCOUS METHOD

BLADE PASSAGE IN CENTRIFUGAL IMPELLER

FLOW

\\

/ ,. -
\ \ \\ .__/C -l , I, ,

I I I 1
-c-y3-)27,-

Figure VI-38

3-D FULLVISCOUS METHOD

VELOCITY PROFILES ON BLADE-TO-BLADE SURFACE

L SUCTION SURFACE

(35-79-2274

Figure VI-39

225

DEVELOPMENT

OF A COMPUTER

CODE

PRODUCTION / CODE WORKING CODE RESEARCH / CODE BASIC ALGORITHM L I I I INlTlAL & BOUNDARY CONDITIONS I TURBULENCE MODEL GEOMETRY & MESH Figure VI-40 / I BENCHMARK EXPERIMENTS

ANALYSIS CODE I VALIDATION EXPERIMENTS I OPTIMIZATION PROCEDURES

I GRAPHICS & DATA HANDLING

cs-7%16?3

GEOMETRY ENRICHMENT

INPUT PANELS Figure VI-41

226

EFFECT OF MESHGEOMETRY
TRANSONIC CASCADEFLOW

SHEARED MESH

WRAPAROUNDMESH CC.-,c4-,7Tf Figure VI-42

EFFECT OF MESHGEOMETRY
SHEARED MESH RESULTS

SURFACE MACH NO. .6-

.4 t % CHORD FigureVI-43

227

EFFECT OF MESHGEOMETRY
COMPARISON OF SHEARED & WRAPAROUNDMESH RESULTS
(e-7 I

,-SHEARED MESH 2

1. 0

.a SURFACE MACH NO. .6

.4
cs-79-2381

% CHORD Figure VI-44

DATAHANDLING AND GRAPHICS

PROFILEDEVELOPMENT

3-DIMENSIONAL

STREAMLINES c5-79-2184

VECTORS

GEOMETRY

CONTOURS

Figure VI-45

228

NUMERICAL OPTIMIZATION FORDESIGN

LOW BLEED INLET

INITIALIZE DECISION VARIABLES e & INLET DATA

I) oPTIMIZATloN ROUTINES -

AERODYNAMICS EVALUATE OBJECTIVE & CONSTRAINT FUNCTIONS

FINAL DESIGN
5-79-2183

Figure VI-46

NUMERICAL OPTIMIZATION FORDESIGN

LOW BLEED INLET

DE:;%NT

BLEED /-WIND TUNNEL DATA

CS-79-7171

ITERATIONS Figure VI-47

229

FASTPERTURBATION SOLUTIONS FORDESIGN

PARABOLIC ARC BLADE ROW

-3.75 >

BASE SOLUTIONS PERTURBATION SOLUTION

BLADE SURFACE PRESS.

BLADE AT ANGLE OF AilACK

I
% OF CHORD

I
FigureVI-48

I
CS-7g-22R7

NASA

LEWIS

RESEARCH

ANALYSIS

CENTER

FigureVI-49

230

VII. Cavour

TURBOMACHINERY

TECHNOLOGY E. Haas*, S. Stepka Lonnie Reid,

H. Hauser, Jeffrey and Francis

National

Aeronautics and Space Administration Lewis Research Center

The goal of Lewis Researcn Center studies of turbomachinery is to provide the basic technology for the design of tne fan, compressor, and turbine components for future advanced aircraft engines. These engines will be quieter and more efficient and will require less maintenance largely because of improvements in their turbomachinery components. The example of an advanced engine shown in figure VII-l is one of the engines of the Aircraft Tnese engines represent the Energy Efficiency (ACEE) program. next forward step in engine component improvement. A wide range of research activities to support the ACES program and to develop further advanced design concepts for each of the turboAn understanding of the basic machinery components is underway. flow characteristics in compressors and turbines and the heattransfer phenomena in cooled turbines is necessary to our goal. This understanding is being attained through a combination of analytical studies and new experimental techniques. The core compressor and the high-pressure turbine are the key elements in the heart of the engine. They have a large effect on engine performance, on fuel consumption, and on maintenance The results of research projects for these two compocosts. nents are presented herein. The attainment of improved engine cycle efficiency requires increased pressure ratio for the core compressor and higher inlet temperature in the turbine. Advanced engines will achieve greater economy through overall pressure ratios as high as 40 to 1 and turbine inlet temperatures to 3000 F. For the core compressor the aerodynamic blading design techniques must be refined so that the required high-pressure ratios can be obtained with high overall efficiency. Further, the compressor must be designed to avoid engine flow instability and to retain its good performance even after thousands of hours of operation. Losses in efficiency due to erosion and wear of components should be reduced. Significant cost reductions may

*u.s. Army Research Cleveland, Ohio.

and Technology

Laboratories

(AVRADCOM),

231

be attained if tne total number of blades in the compressor can With higher tip speed a higher pressure ratio per be reduced. In the inlet stages the blading must be stage may be attained. tailored to accept high-Mach-number flow if high efficiency is Low-aspect-ratio blades, that is, blades that to be obtained. have relatively long chord as compared with their radial span, are effective in extending the useful flow range of high-MachFor middle and rear stages the highly loaded stages. number, through-flow velocities are not constant over the blade height, Stages in which the but fall off in the hub and tip region. blade profiles in the end-wall regions are designed to account A recessed casing for this velocity profile are being studied. contour over the rotor blading is shown to reduce tip clearance losses. The laser velocimeter has recently been applied to ment of local velocities within the rotating blades Some early results show mental compressor stages. velocities verify calculated values from analytical the measureof experithat measured codes.

For the hign-pressure turbine minimizing the cooling requirements for the desired increased temperatures is the dominant engineering challenge. Overhaul and replacement of the turbine hot parts is the largest cost in the long-term maintenance of an Thus increased turbine durability is a high priority engine. goal. For the turbine temperatures of interest, film cooling, where the coolant is bled out over the blade surfaces, is required. Cooled-blade configurations must be evolved so that the minimum coolant flow is required and so that a minimum penalty in aerodynamic efficiency, due to the coolant discharge over the Recent progress in heat-transfer blade surface, are incurred. studies on cooling, ceramic thermal barrier coatings, and the effect of the film-coolant effluent on the turoine aerodynamic as for the compressor, it performance, are discussed. Finally, has been found that turbine tip clearance losses can be reduced with a recessed casing. CORE-COMPRESSOR INLET STAGE STUDY

An experimental study was made to evaluate the performance of These stages have inlet stages of an advanced core compressor. high blade loadings and must operate in a transonic-flow environment. Rotor blading must be designed to accommodate high supersonic inlet relative Mach numbers and to diffuse the flow To achieve good efficiency and to subsonic exit velocities. flow range with these stages, it is necessary to carefully consider the flow physics within the blade row and to establish a rationale for the selection of the blade design parameters. A rotor-blade row is model of transonic flow through a compressor shown in figure VII-2. For supersonic inlet relative Mach numbers a bow shock exists at the blade leading edge and extends upstream of the adjacent blade. The bow shock is followed by a 232

series of expansion waves. The blade-passage throat occurs a short distance downstream of the passage shock. For these highly loaded blade rows the velocity just downstream of the passage shock is subsonic. These subsonic velocities must be diffused to a lower velocity at the exit of the blade passage. In the design of the blading it is necessary to minimize the amount of supersonic expansion ahead of the passage shock to avoid high shock losses and severe shock-boundary-layer interaction. This can be achieved by minimizing the amount of flow turning or blade camber ahead of the passage shock. The blade passage throat area must be large enough to pass the design mass The throat size is strongly influenced by the selection flow. of camber distribution. The requirement of low blade camber ahead of the passage shock necessitates relatively high camber downsream of the passage shock to achieve the overall flow turning through the blade row. The blade surface pressure distributions for this transonic flow model are shown in figure VII-3 where the ratio of local static pressure to the inlet relative total pressure is plotted as a function of blade chord. The pressure gradients on the blade pressure surface are relatively mild. However, the pressure gradients on the suction surface are quite severe and merit special considerations. Near the blade leading edge the flow is subsonic, due to the strong bow shock. The sharp drop in static pressure near the leading edge on the suction surface corresponds to local acceleration of the flow to supersonic conditions. This is followed by a gradual drop in static pressure, which represents the supersonic The expansion or flow acceleration on the suction surface. The flow large pressure rise is indicative of the strong shock. The static condition just downstream of the shock is subsonic. pressure rise from that point to the blade exit is indicative of the amount of subsonic diffusion that is required to reach the blade-exit conditions. Blade surface pressure distributions are shown in figure VII-4 for two blades with the same design blade shape and with the same inlet and exit flow conditions, but with different chord lengths. The strength of the shock or sharp static-pressure Howrise on the suction surface is the same for both blades. ever, the static-pressure rise from downstream of the shock to occurs over a longer disor subsonic diffusion, the blade exit, Therefore, the rate tance for the blade with the longer chord. of subsonic diffusion is lower for the blade with the longer This lower rate of diffusion reduces the tendency for chord. severe boundary-layer separation and high diffusion losses and implies that long-chord, or low-aspect-ratio, blading might perform well for these high-Mach-number, highly loaded stages. An experimental evaluation of the effects has been made using two high-Mach-number, of blade aspect ratio highly loaded stages 233

that are suitable as inlet stages of an advanced core compresThe rotors for the compressors are shown in figure VII-5. sor. One rotor has low-aspect-ratio blading; the other nas moderateaspect-ratio blading. Both compressors were designed for a pressure ratio of 2.05, a rotor tip speed of 1500 feet per second, and a rotor-inlet relative Mach number of 1.48. A photograph of corresponding individual rotor blades is shown in figure VII-6. The aspect ratio is 1.2 for the low-aspect-ratio blade and 1.6 for the moderate-aspect-ratio blade. These stages were tested in the Lewis single-stage compressor facility (fig. VII-7). A performance comparison of the two stages is shown in figure VII-S. Performance data are presented at speeds from 0.5 of design speed to design speed. The low-aspect-ratio stage has a substantially greater flow range than the moderate-aspect-ratio Both stages have about the same maximum flow for a given stage. speed but the low-aspect-ratio stage operates at a much lower flow and higher pressure ratio before it encounters the stall limit. The overall efficiency levels for the two stages are comparable. The results from this particular study show that for this high pressure ratio and hign-inlet-relative Mach number good efficiency and stall margin can be achieved with low aspect ratio blading. CORE-COMPRESSOR MIDDLE AND REAR STAGE STUDIES Reduced Losses in End-Wall Regions

Research studies on the middle and rear stages of core compressors are aimed at reducing the losses in the end-wall regions near the hub and tip of the blades and also reducing tne tipclearance losses. Some fundamental experimental studies of these areas are being carried out in a contract program with the General Electric Co. using the facility shown in figure VII-g. Because the flow in compressor rear stages is subsonic, the performance of these stages may be modeled in this low-speed, large-scale facility where detailed measurements of the flow can be obtained. The rig includes survey instrumentation that roallowing loss source to be identtates with the rotor blades, ified. Tests are conducted using four successive, identical stages (fig. VII-lo), so that the forward stages set up a true multistage environment for the latter stages and typical multiThe primary objective of stage loss mechanisms are simuiated. the program is to develop middleand rear-stage blade designs with improved efficiency through reduced losses in the end-wall boundary-layer regions. Blade design concepts that show promise of reducing these end-wall losses are evaluated. Part of the study bution of velocity 234 is aimed at an evaluation over the surfaces of the of the best distriTwo rotor blades.

rotor-blade designs (rotor A and rotor B) are compared. In figure VII-11 the blade-tip profiles are superimposed, and the design velocity distributions on these two profiles are compared. The velocity distributions were calculated using a blade-to-blade analysis code; the velocity field over a circumferential surface between adjacent blades was analyzed. The blade profile for rotor A is standard, having essentially circular arc curvatures over the suction and pressure surfaces. Rotor B was designed for the same velocities and flow angles both upstream and downstream of the blade row but the distribution of camber is changed to give higher blade loading in the rear portion and a more rapid diffusion of the velocity along the suction surface near the trailing edge. For the velocity distribution of rotor B, the maximum difference in velocity between the suction and pressure surfaces is reduced. This corresponds to a smaller pressure difference from the pressure to the suction surface across the tip of the blade over the forward portion of the blade chord. It is thought that this reduced pressure difference will reduce the strength of the tip clearance leakage vortex and thereby reduce the resulting losses. The objective of the second part of the study was to determine what distribution of axial velocity from the hub to the tip over the blade height would give the better performance. Two different stators were designed (stator A and stator B) to vary the velocity distribution (fig. VII-12). The axial velocity component is generally lower in the hub and tip regions, and the radial distribution for stator A is conventional. Stator B is designed to give the design axial velocity distribution shown in figure VII-12 with higher velocities over the midportion of the passage and lower velocities in the hub and tip regions than stator A. Stators A and B have essentially the same airfoil sections, but the sections in the hub and tip regions of stator B are twisted toward the tangential direction, which increases the flow angle 6 in these regions. These changes in twist at the extremities of stator B are evident in figure VII-13. The axial velocity distribution has less effect on rotor blade angles, and the two rotors were designed for either of the velocity distributions. To evaluate the relative performance for the two stators, the measurements were taken in the third stage of the four-stage group. These measurements confirmed that the desired velocity distributions had already been set up for each case by the inlet guide vanes and the first two stages. Thus the performance comparison of the stages with stators A and B is an evaluation of which of these two radial distributions of axial velocity is preferable. The relative performances of the two stages are presented in figure VII-14. For these low-speed experimental stages performance is expressed in terms of pressure coefficient and flow 235

Pressure coefficient is the rise in casing static coefficient. pressure normalized by one half the density times the square of The flow coefficient is the ratio of inlet axial the tip speed. The stage performance of rotor A-stator velocity to tip speed. The various combinations of A is considered as the baseline. but rotor B-stator B gave the best rotors and stators were run, the design characteristics for Apparently, overall performance. This stage had higher both rotor B and stator B were desirable. a greater flow range at high pressure coefficient at each flow, and a more desirable pressure coefficient characterefficiency, istic as stall conditions were approacned. Tne drop in pressure coefficient at low flow for rotor A-stator A probably results from a significant flow breakdown and separation, which has been observed to occur in the hub region. Reduced Tip Clearance T,oss

The effect of a casing with a recessed clearance configuration, as shown in figure VII-15, was evaluated for rotor B-stator B. Such a recessed clearance has proved effective in reducing the tip clearance losses in turbines. Tne actual running clearance is the same for tne standard casing and for the recessed clearance configuration. In tne recessed clearance configuration, the blade is longer so that the tip is line-on-line with the upstream and downstream casing. As indicated in figure VII-16, the recessed clearance? configuration apparently does reduce the losses in the tip cletirance region. A further small increase in efficiency at the ni.gher flows and a nigner pressure coefficient over the whole range of flow was obtained with tne recessed clearance. EXPERIMENTAL VERIFICATItiG OF ANALYTICAL CODES

With the laser Doppler velocimeter, detailed measurements throughout the blade passages for both the rotor and the stator may be obtained. This makes possible direct verification of the analytical codes and thereoy gives a far more comprenensive understanding of the complex physical phenomena occuring within the blade rows of turbomachinery. It is most promising. A photograph of the velocimeter installed in the compressor facility of figure VII-7 is shown in figure VII-17. The compressor test section is on the left. Two laser beams are projected through a window in tne compressor casing and cross one another at a point within the rotating blade row. The velocity of microscopic particles passing through this intense point of light is measured by the instrument. The data shown in figure VII-18 line output from the velocimeter. urement station is shown in the 236 direct reproductions The location of the two diagrcilgs of figure are of onmeasVII-1Y.

As indicated, the circumferential survey of figure VII-18 was taken at a station about one-third chord length downstream of The axial comthe blade leading edges and near the blade tip. against circumferential posiponent of velocity VZ is plotted The cross-hatched regions tion stations for each blade passage. represent the blade locations. Each of the points of this plot represents the measured axial velocity within the blade pasSimilar measurements of the tangential velocity are obsage. tained so that the distribution of the relative velocity across the blade passage can be calculated. This relative velocity distribution averaged over 20 blade passages is shown in the Measurements are taken at other stations through lower plot. the passage to obtain the relative velocity contour plots shown A blade-passage flow analysis code was used in figure VII-20. In this case to obtain the velocity contours shown on the left. the agreement with the measured contours on the right is quite Codes that can be used at higher velocities where there good. The are shock patterns in the passage are under development. ability to obtain detailed velocity measurements throughout the blade passages is an invaluable aid in verifying the results of analytical design codes. TURBINE COOLING

The gas temperature levels of current and advanced engines are above the melting point of turbine-blade materials; turbine This cooling is provided by cooling is, therefore, required. air that is bled from the engine compressor and ducted to the base of blades and then through cooling passages within tne blades. Bleeding air from the compressor, however, results in Tne primary research objective is to engine performance losses. provide the technology to minimize the cooling requirements and thus these losses. It is desired to obtain a fundamental understanding of the flow and heat-transfer phenomena for the more effective cooling methods and to improve the accuracy of predicting blade-wall temperatures so that the cooling air is effectively distributed. Cooled blades will typically use combinations of cooling metnFor example, ods. the blade in figure VII-21 is convectively cooled in the midchord region by the airflow througn radial passages. Part of this air passes through holes in an internal wall to impinge and cool the leading edge. It is then ejected through holes in the blade wall to form an insulating film layer on the blade surface. The other part of the air from the midchord is directed to passages in the rear to convectively cool the trailing edge. The various methods of cooling are schematically illustrated cooling with augmented surface or insulating turbine blade walls in figure VII-22. Convection area is tne simplest means of
237

Another method is impingement cooling, wnerein air cooling. holes from an insert within a from a row, or many rows of small For impacts the blade wall and reduces its temperature. blade, advanced engines tnese cooling methods alone are not adequate at some locations on the airfoil and need to be combined witn local wherein the addition of a thin layer of cool air film cooling, For temperatures insulates the blade from the hot gas stream. envisioned for future engines full-coverage film cooling consisting of additional film cooling air from a multitude of small holes will be required. A thermally more efficient extension of full-coverage film cooling is transpiration cooling, wherein a cool layer of air is deposited on the blade surface after passAnother method of ing through a porous or woven wall material. is to insulate the blade from reducing blade wall temperatures the hot gases by a thermal barrier, tnat is, a low conductivity ceramic coating deposited on the gas side of the blades. The need for the more effective cooling methods is indicated in Coolant flow requirements rise rapidly for configure VII-23. vection cooling when turbine-inlet gas temperatures and presThe conditions of current engines are indicated sures increase. (>bviousiy, more complex by tne vertical line on the abscissa. full-coverage film cooling or cooling methods, for example, will be required to reduce cooling-air transpiration cooling, The use of thermal-barrier flow requirements in future engines. coatings on cooled turbine parts can further reduce the cooling-air requirements. Although neat-transfer studies have been conducted for all of only research results on the methods depicted in figure VII-22, full-coverage film cooling, and the impingement cooling, thermal-barrier coating will be discussed. Impingement Cooling

In order to effectively utilize impingement cooling, the interrelationship of flow and geometry variables on heat transfer The schematic of an impingement-cooled wali conmust be known. figuration in figure VII-24 shows that, in addition to the ima crossflow of cooling air is often prespingement jet flow, This air is from previous cooling of the blade upstream of ent. the jet array or from the flow that has accumulated with disThe research on tance downstream of the initial impinging jets. impingement cooling discussed herein was conducted under a grant Experimental studies were conto Arizona State University. ducted at conditions that simulate those of advanced core turSome of the variables under investigation included Dine vanes. arrangements of the jet arrays, hole spacing, hole-diameter-toand impingement-to-crossflow velocity wall-distance ratios, ratios. 238

A photograph and a schematic of the test apparatus are shown in figures VII-25 and VII-26. Air enters the plenum, exits through hole arrays in an interchangeable jet plate, and then impinges upon the test wall. The test wall consists of individually heated and insulated copper strips. The local heat-transfer coefficients for each strip are obtained from measurements of the electrical heat input into the strip and the temperatures of the strip and impinging air. For the tests described herein, only the test strips beneath the jet plate were used. One of the purposes of the investigation was to evaluate the relative cooling effectiveness of in-line rows of jets and staggered rows of jets. Initial results are shown in figure VII-27 where the change in relative heat-transfer coefficient with distance is plotted. High coefficients are desirable so that heat will be taken from the blade wall and reduce wall metal temperatures. The results show that the coeficient is highest at the first row of holes and then decays with distance downstream for both the in-line and staggered jet arrays. Of greater significance, and contrary to initial expectations, the in-line jet array is more effective than the staggered array for the same At the farthest downstream coolant flow rate and hole density. row of jets the in-line array has a 60 percent higher heatThis superiority transfer coefficient than the staggered array. of the in-line array was greatest when the hole density was the were four and five dihighest, that is, when the hole spacings ameters and when the jet-to-wall distances were two and three hole diameters. Test are continuing to obtain more data on this cooling method. The current study is concentrated on developing the analysis that will correlate the many variables in the investigation so that optimum impingement-cooling configurations can be designed. Full-Coverage Film Cooling

Research on full-coverage film cooling has been conducted to obtain the experimental thermal performances of selected filmcooling hole arrays and to develop analytical methods to predict their performances. The study was conducted under a contract A photograph and a schematic of the with Stanford University. test appratus are shown in figures VII-28 and VII-29. In the test tunnel film-cooling air and main-stream gas-flow conditions simulate those over a turbine vane. Cooling air is injected into the main stream through 11 rows of holes in the test wall. The film-cooling holes extend over only the initial portion of the tunnel length. The tunnel test wall consists of individually insulated and heated copper strips. The heat-transfer coefficient for each strip is obtained (as in the impingement cooling study) from measurements of the electrical energy input to the individual strips and of the temperatures of the strip and main-stream air. This apparatus gives the flexibility to 239

investigate various hole geometries, spacings, and orientaThe two hole arrays discussed herein are shown in figure tions. The holes in both arrays are inclined 300 to the wall VII-30. The array on the righthas these holes alined in the surface. direction of the. main-stream flow (in-line injection). The holes in the other array are alined at an angle of 450 to the main stream (compound-angle injection). The compound-angle injection was included in the study because Lewis flowvisualization studies indicated that improved cooling might be obtained. The results of detailed measurements of the reiative heattransfer coefficients as a function of distance downstream of the test-section inlet for no air injection and for in-iine and compound-angle injection are given in figure VII-31. For these which represent conditions on the gas-side surface of a data, a low coefficient is desirable so that less heat turbine blade, will be transferred from the gas stream to the blade wall and thus give lower blade metal temperatures. Clearly, in-line injection significantly reduces the heat-transfer coefficient. Compound-angle injection is even more effective beyond 25 hole diameters downstream of initial injection. However, beyond the region of film injection, the heat-transfer coefficient for the compound-angle injection increases rapidly. As part of the study at Stanford, a two-dimensional boundarylayer computer program called STANCOOL was developed to predict heat-transfer coefficients with film cooling. Figure VII-32 shows a comparison of analytical predictions with experimental heat-transfer coefficients for no injection and for in-line hole The agreement is good. For no film injection the injection. prediction line runs through the experimental data. For film injection the cyclic pattern of the predicted local coefficients across each of the strips (between rows of film cooling holes) agrees well with the measured average coefficients for the individual strips. To evaluate the performance of the in-line and compound-angle hole arrays on an actual turbine vane a Lewis investigation was conducted using the turbine cascade shown in figure VII-33. One of the vanes in the center of the cascade had five rows of inline holes and another had five rows of compound-angle holes on the suction surface. The schematic of the airfoil in the figure shows the location of the rows of holes. Downstream of each row a copper strip with thermocouples was installed to obtain the average spanwise temperature for each row. The results of the test confirmed the superiority of the compound-angle injection over the in-line injection for all of the rows. Evidence of this superiority is shown in figure VII-34 in the plots of cooling effectiveness as a function of mass flux ratio for the second and last row of holes on the vanes. Cooling effectiveness is the ratio of the reduction of vane metal temperature 240

below the gas temperature to that ideally possible at a given The mass flux ratio is proportional to cooling-air temperature. The superiority of the compound-angle cooling-air flow rate. ratios above injection is particularly evident at mass flux about 0.5 for the last row of holes where the cooling effectiveness of the in-line injection begins to decline while that of the compound-angle injection increases. Thermal-Barrier Coating

The thermal-barrier coating is a means for reducing blade metal temperatures and/or coolant flow. The two-layered, thermalbarrier coating applied to a convection-cooled turbine blade is shown in figure VII-35. The first layer is a metallic bond coat that is typically about 0.004 inch thick, and the second layer that is typically 0.005 to 0.020 is yttria stabilized zirconia inch thick. Figure VII-36 shows turbine vane leading-edge metal wall temperatures obtained over a range of coolant-to-gas-flow ratios in a research engine for both an uncoated vane and one with a 0.011 inch thickness of the ceramic. The circles and squares are exThe perimental data and the lines are analytical predictions. gas temperature for these tests was 25000 F and the coolingThe results illustrate that sigair temperature was 115O F. nificant reductions of as much as 340 o F in vane wall metal temperatures were obtained with the coating. The tests also showed the ability to predict this reduction analyticaliy, thus indicating that the thermal conductivity data for the plasmaThe figure sprayed-on coating is adequate for design purposes. also shows that at a given metal temperature a threefold reduction in coolant flow was obtained. Although large metal temperature or coolant-flow reductions can there was concern about the effect be obtained with the coating, of the coating on aerodynamic loss. Figure VII-37 shows the kinetic energy loss coefficient (which is a measure of loss in aerodynamic efficiency) for coated and uncoated vanes as obtained in a cold-air cascade. The data show that the as-sprayed coating gave very high losses compared with an uncoated vane. However, when the coating was smoothed by abrasive poiishing, the loss was only about 0.7 of a point larger than that for the uncoated vane. This loss was attributed to tne 38 percent thicker trailing edge caused by the coating. Redesign of the blade for a thinner trailing edge would reduce this loss.

241

TURBINE

COOLING AERODYNAMICS

Ejecting air from film-cooling holes such as tnose shown on the vane in figure VII-38 disturbs the boundary layers along the blading and end walls and causes significant decreases in turThe design problem becomes one of devising bine efficiency. means for ejecting the coolant from the blaciillg and end walls so that it effectively insulates the blade yet causes a minimum disturbance to the boundary layers. Two-Dimensional Cascade Program

Several programs to investigate the effect of film-cooiant injection on stator-vane performance have been conducted using the This CaSCade concascade shown in figures VII-39 and VII-40. of 12 constant-section vanes, the middle three of which sists The main-stream air enters the cascade from the are cooled. and the air that is used to simulate coolant is introright, duced to the middle three vanes through the coolant manifold and A survey probe is mounted downstream of the associated piping. blading and traverses parallel to the plane of the vane trailing The flow is discharged from the cascade through exhaust edges. Variables that have been piping attached to the cascade base. investigated in the research programs include coolant hole size, the angle orientation of the coolant holes witn respect to both the vane surface and the main-stream flow direction, and the location of the coolant ejection holes on the vane surface. One of the experimental film-cooled vanes is shown in figure There were 12 rows of coolant holes oriented in-line VII-41. surface and 6 rows with the main-stream - 6 rows on the pressure There were about 80 holes per row. The on the suction surface. blading was hollow and of constant cross section. A summary of the effect of coolant ejection from individual locations on vane efficiency is shown in figure VII-42. In these tests the coolant was ejected individually from each of the 12 rows of coolant holes. These tests were conducted for a condition where the coolant temperature was equal to the main-stream temprature and the coolant pressure was equal to the main-stream The numbers in the figure refer to the percentage pressure. loss in efficiency for a 1 percent coolant flow from each row of For example, a value of 0.20 indicates that the coolant holes. vane efficiency would decrease by 0.20 percent for 1 percent For this coolant ejection from that row of coolant holes alone. vane configuration the largest losses occurred on the suction surface, particularly in the region of maximum diffusion. In this region the local surface static pressure is lower than the Ejecting coolant into static pressure downstream of the vane. the boundary layer in this region of diffusion where an adverse 242

pressure gradient ably and causes

exists thickens the boundary large loss in vane efficiency.

layer

consider-

It is of interest to know whether the total vane efficiency loss can be predicted by adding the individual single-row losses. Figure VII-43 shows that the single-row results do add together This figure compares the additive single-row losses quite well. with the multirow losses for coolant holes open on the pressure on the suction surface, surface, and around the entire vane. The numbers next to the blocks for the additive, single-row results indicate the location of the film cooling hole rows. The good agreement for all three cases indicates that coolant ejection from upstream rows of coolant holes has little effect on coolant ejected from downstream coolant rows. Analytical methods that can be used to predict the effect of coolant on the aerodynamic performance are being developed. A comparison of results from one of these methods (the previously mentioned S.TANCOOL computer analysis) with experimental results is shown in figure VII-44. The STANCOOL results, given in terms are compared with experimentally obtained of efficiency loss, efficiency losses from different multirow coolant ejection locaThe case where the coolant holes were sealed (labeled tions. "solid vane" in the fig.) is included as a reference. The next two bars show the predicted loss with the holes open only on the The last bar suction surface and then on the pressure surface. with the use of all of the coolant holes, the indicates that, The STANCOOL program predictions efficiency loss is 5 percent. agree very well with the experimental data. Another observation can be made from these results. If the solid vane loss of 2 percent is subtracted from each of the other three cases, the remaining loss is that due to the film cooling. It can be seen that the individual film cooling losses for the suction and pressure surfaces add together to equal the total cooling loss. Stator End-Wall Film Cooling

In another program the effect of stator end-wall film cooling on stator-vane performance was investigated in a cold-air, fullIn figure VII-45 the end-wall coolant-hole annular cascade. can be seen on the hub and tip hollow vane rings. patterns Figure VII-46 shows the arrangement of tne end-wall cooling noles with respect to the vane passage. it is desirable to Again, eject the coolant with the least disturbances of the main-stream flow. The radial variation in out end-wall coolant is and main-stream pressures vane total-pressure ratio witn and withThe coolant compared in figure VII-47. were equal, resulting in a l-percent 243

The low momentum coolant apparently migrates to coolant flow. the passage vortex regions that are formed from the movement of secondary flows in the vane passage and increases the losses as This increase in compared with those for the solid end walls. total-pressure loss corresponds to an approximate 0.8 percent Other configurations are being decrease in vane efficiency. studied in an effort to reduce these losses. Effect of Temperature Ratio on Vane and Turbine Efficiency

The results presented thus far have been from cold-air experimental studies where air at or near the same temperature as the main-stream flow was used to simulate cooling air. The tests were conducted in this manner because the test facility and test hardware costs were markedly lower and because the tests could be performed more rapidly and with fewer problems. However, a questions arises as to whether these tests accurately model the actual engine operating condition. A method was developed to predict the aerodynamic performance at actual engine operating conditions from results obtained where the coolant and mainstream temperatures were equal. The primary assumption used in this method is that the aerodynamic performance at the actual engine operating condition will be modeled if the ratio of the coolant momentum to that of the main-stream flow is held constant between the test condition and the engine operating condition. The results from this prediction method are first compared with experimental data for a full film-cooled vane (fig. VII-48). The vane efficiency is plotted against the ratio of the mainstream temperature to the coolant temperature. In a core turbine application this ratio will approach a value of three. shown in this figure is the variation in vane efficiency for three values of coolant-supply to main-stream total-pressure ratio. The agreement between the results of the prediction method and the experimental data is good. At a pressure ratio of 1, which is typical of a first-stage vane, the efficiency is constant with temperature ratio. At higher pressure ratios, which correspond to coolant ejected from downstream vanes in a multistage turbine, the efficiency increases with increasing temperature ratio. The results from this prediction method are compared with experimental data for a core turbine in figure VII-49. As with the vane results, there is good agreement. Of special interest is that the turbine efficiency remains almost constant as the temperature ratio increases. This indicates that the performance at the actual temperature ratio condition can be modeled quite well from reduced temperature ratio results.

244

TURBINE ROTOR TIP

CLEARANCE EFFECTS

The problem of maintaining good aerodynamic efficiency with film cooling of the stator and rotor blading is the most critical problem associated with the core turbine. However, because the low-aspect-ratio blading, core turbine has small-blade-height, other aerodynamic design considerations are also important. As the rotor tip losses are significant. in the core compressor, Several studies have been conducted at Lewis to investigate the rotor-tip clearance loss. Two rotors that were tested are shown in figure VII-50. Two tip-clearance configurations were investigated. The inset on the left in figure VII-51 shows a recessed casing configuration. In this configuration the rotor-blade tip was in line with the upstream and downstream casing, and the tip clearance was varied by changing the casing recess depth. In the conventional, reduced-blade-height configuration, which is shown in the inset on the right, the casing was not recessed, and the tip clearance was varied by cutting back the rotor-blade height. The tip-clearance losses for the two configurations are compared in figure VII-52. The recessed casing configuration was tested The reduced-blade-height with the 5-inch tip-diameter rotor. configuration was tested with both rotors. For a given rotor the loss for the recessed-casing configuration is tip clearance, less than the loss for the reduced-blade-height configuration. The difference in loss for the two rotors having a reducedblade-height configuration is attributed to a difference in the level of rotor reaction. Using a recessed-casing configuration instead of a reduced-blade-height configuration could result in a core turbine efficiency improvement of approximately 1.0 to 1.5 percent.

245

TURBOMACHINERY TECHNOLOGY GOALS

/CORE COMPRESSOR-/. . HIGH PRESS. RATIO . EFFICIENT . PERFORMANCE RETENTION . REDUCED NO. OF BLADES

i
iHlGH PRESS. TURBINE . HIGH TURBINE INLET TEMP . DURABLE l MIN. COOLANT FLOW l EFFICIENT / ;- ,I- s,,,

Figure VII-l

TRANSONIC COMPRESSOR FLOWMODEL

EXPANSION WAVE FLOW \

,79-2227

Figure VII-2

246

BLADESURFACE PRESSURES

STATIC TO TOTAL PRESS. RATlO '-SUCTION .31 0

~~-79-2228

SURFACE I 20

.4

.8 I.2 BLADE CHORD, in.

I.6

Figure VII-3

BLADE SURFACE PRESSURES

-/ / STATlC TO TOTAL PRESS. RATlO LSUCTION SURFACE .31 0 I .4 I I I I.6 I.2 .8 BLADE CHORD, in. I 2.0 I 2.4 I

(3%79-2242

Figure VII-4

247

LOW AND MODERATE ASPECT RATIO COMPRESSOR ROTORS

Figure VII-5

LOW AND MODERATE ASPECT RATIO COMPRESSOR ROTOR BLADES

Figure VII-6

248

LEWIS

SINGLE STAGE COMPRESSOR

FACILITY

Figure VII-7

STAGE PERFORMANCE COMPARISON

DESIGN SPEED 1.0 PRESS. RATIO 1.4 1 24 I 32 FLOW, lb/w I 48

(35-79-2233

ASPECT RATIO 0 . 16 40 LOW MODERATE

Figure VII-8

249

LOW SPEED COMPRESSOR

FACILITY

250

LOW SPEED COMPRESSOR BLADE ROWS

Figure VII-10

SURFACE VELOCITY COMPARISON

/-ROTOR A B

VELOCITY 1 RATIO 1.

Figure VII-11

251

COMPARISON OF RADIAL VARIATION OFAXIALVELOCITY

STATORS A & B d ,,-STATOR B rHUB

BLADE HEIGHT, %

STATOR B -50 60 70 80 INLET AXIAL VELOCITY, ftlsec 90 CS-79-7"5<

Figure VII-12

REAR STAGE BLADING

ROTOR A E-79-2208

STATOR A

ROTOR B -

STATOR B

FigureVII-13

252

STAGE OVERALL PERFORMANCE

A

o ROTOR A - STATOR A ROTOR B - STATOR B

EFF

.65r

.60

PRESS. COEFF .55

.50

.45

.3

.4 FLOW COEFF Figure VII-14

.5

CASING WITH RECESSED CLEARANCE

STANDARD CASING

RECESSED CLEARANCE
cs-n-2251

Figure VII-15

253

EFFECTOF RECESSED CASING .92, .90 EFF. .a .86

.65 0 PRESS. m60 COEFF. .55

STANDARD CASING RECESSED CASING

FLOW COEFF Figure VII-16

COMPRESSOR

FACILITY

WITH LASER

VELOCIMETER

Figure VII-17

254

LASER DOPPLER-VELOCITY MEASUREMENTS

AXIAL VELOCITY 1000 ,-BLADE

0 2000 1500 VREL, ftlsec Icoo 500 0

(35-79-2237

I I I I I I I I I I 25 50 75 100 125 150 175 200 225 250 AVERAGED RELAllVE VELOCITY Q..........................

10 15 20 25 30 35 40 45 50 CIRCUMFERENTIAL POSITION Figure VII-18

LDV MEASUREMENT LOCATIONS

MEASUREMENT STATION I MEASUREMENT STATION MEASUREMENT I

BLADE-TO-BLADE SECTION IN MEASUREMENT PLANE Figure VII-19

cs-79-2239

255

CALCULATED AND MEASURED VELOCITY CONTOURS

cs-794250

Figure VII-20

SCHEMATIC OF COOLED TURBINEBLADE

MINIMIZE COOLING AIR REQUIREMENTS l OBTAIN FLOW & HEAT TRANSFER l IMPROVE BLADE TEMP PREDICTIONS

Figure VII-21

256

TURBINEBLADE WALLCOOLING TECHNIQUES

COOLiNG AIR CONVECTION

IMPINGEMENT

LOCAL FILM

FULL COVERAGE FILM -CERAMIC COATING -BLADE WALL INTERNALLY AIR-COOLED

TRANSPIRATION

THERMAL BARRIER Figure VII -22

cs-794252

TURBINE COOLING REQUIREMENTS

-[I-+d CONVECTlON COOLING FULL COVERAGE FILM COOLING TRANSPIRATION COOLING

t INCREASING COOLING AIR REQUIRED

INCREASINGLY COMPLEX CONFIGURATTONS

IENGINE INCREASING TEMP & PRESS. -

(3%79-2241

Figure VII-23

257

IMPINGEMENT COOLING

/ COOLING
AIR FLOW

IMPINGEMENT INSERT

-HOT

GAS

Figure VII-24

IMPINGEMENT COOLING APPARATUS

Figure VII-25

258

SCHEMATIC OF IMPINGEMENT COOLING APPARATUS

AIR INLET y PLENUM

TEST WALL-\ v-:. . :: i _,N .. -. .^. :.. .: : . .

;.,,. y...../., .;:..,y.;-...: . ..:.y .,.. ,;.-;,, I

cs-79-2245

Figure VII-26

IMPINGEMENT COOLING HEAT TRANSFER

RELATIVE HEAT TRANSFER COEFF

. G
STREAMWISE DISTANCE

COPPER STRIPS

Figure VII-27

259

FILM COOLING APPARATUS

Figure VII-28

SCHEMATIC OF FILM COOLING APPARATUS

MAINSTREAM h

FILM COOLING AIR

AIR INJECTION TUBES ELECTRIC HEATERS i INSULATION

cs-79-2234

Figure VII-29

260

FILM COOLING HOLE ARRAYS

MAINSTREAM

MAINSTREAM COOLANT FLOW

COMPOUND ANGLE HOLES

IN-LINE HOLES

C-70-72L7

Figure VII-30

.ING 'ILM COOL EFFECT OF F

0
o%ooo 00 a RELATIVE HEAT TRANSFER COEFF 0 .6

NO INJECTlON IN-LINE COMPOUND ANGLE

oooooo~o

II
0 25

I
125 cs-7g-2256

100 50 75 DISTANCE, HOLE DIAMS Figure VII-31

261

_. .

.._ - .._. _._

EXPERIMENTAL AND PRED .ING ICTEDFILM COOL

1.2

r
NO INJECTION

RELATIVE HEAT TRANSFER COEFF

.6

.4

ILL
0
CS-79-22G

/ -I ,) ,

FILM INJECTION 100

50

DISTANCE, HOLE DIAMS Figure VII-32

CASCADE TESTSOF FILM COOLING

q--J$c \ L3
$2

.
-1
INLINE HOLES ,?

COMPOUND ANGLE HOLES.7

. cs-792215
Figure VII-33

262

FILM COOLING OF TURBINE VANEIN CASCADE

INJECTION 0 COMPOUND ANGLE r: IN LINE ROW 5 .6 .5 -4 .3 .* .l 0 r C-,?-?)l.
r

COOLING EFFECTIVENESS, Ts-T, Tg - Tc

/-!I .4

08 1.2 1.6 0 L,I .4 .8 1.2 COOLANT-TO-MAINSTREAM MASS FLUX RATIO

1.6

Figure VII-34

THERMAL BARRIER COATED TURBINE BLADE

Figure VII-35

263

EFFECT OF THERMAL BARRIER COATING ON VANEMETAL TEMPERATURES

0
Cl UNCOATED VANE COATED VANE PREDICTION

1600 m METAL

COOLANT-TO-GAS FLOW RATIO Figure VII-36

CS-79-2238

EFFECTOFTHERMAL BARRIER COATING ON VANELOSSES

.a3 COATEDA5 ;PR ED -

.06

LOSS COEFF .D4

COATEDSMOOTI HED

m-79-2226

Figure VII-37

264

FULL-FILM COOLED VANE

(35-79-2209

Figure VII-38

STATOR VANECASCADE

SURVEY PLANE

I
csqg-2222

. I

11

Figure VII-39

265

STATOR VANE CASCADE

SURVEY PROBE ACTUATOR-x_

Figure VII-40

266

FILM-COOLED VANE

_i 1 : i_ : :.

j :
: ; ; ; : : : : :. .. : :

: : ;,

CS-79-2220

PRESS. SURFACE

SUCTION SURFACE

FigureVII-41

EFFICIENCY LOSSFOR1% COOLING AIR EJECTION

.

TCOOLANT = TMAINSTREAM PCOOLANT = 'MAINSTREAM

.25

.26
C&79-2240

FigureVII-42

267

ADDITIVE FILM COOLING HOLELOSSES

r ADDlTlVE ! SINGLE ROW \ RESULTS I MULnROW _I I RF -TS I 4 EFF LOSS, 3 %

12 11

Lull
l-57 ;/ I PRESS. SURFACE COOLING HOLES OPEN SUCflON SURFACE 54EFF LOSS, % 32 lO-

ACE SUCnON ! IRF, COOLING HOLES

2 ALL COOLING HOLES OPEN

CS-TV-2248

Figure VII-43

EXPERIMENTAL AND PREDICTED FILM COOLING LOSSES

RESS. UR(ACE \ AILING EDGE GION -e- TOTAL LOSS (SUCTION SURFACE, SUCTiON PRESS. SURFACE & SURFACE TRAILING EDGE) -e-PRESS. --- SURFACE ---

PREDlCnON METHOD-, , -0. EXPERIMENTAL DATA 7 * SOLID --VANE

(35-79-2257

Figure VII-44

268

COOLING HOLES VANERINGSSHOWING ENDWAl

TIP ENDWALL
~~-79-2219

HIJB ENDWALL Figure VIIe45

L;,pEND<
c+79-2231

Figure VII-46

269

VANE TOTAL PRESSURE RATIO WITHENDWALL COOLING AIR

r SOU D ENDWALLS

TOTAL PRESS. RATIO

.%-I

COOLED ENDWALLS (1%COOLING AIR)

I I

.90 I
0
cs-79-2259

I
.2

,8 .6 .4 RADIAL POSITION

I 1.0

FigureVII-47

EXPERIMENTAL AND PREDICTED VANEEFFICIENCIES

COOLANT-TO .97 =: fKKii

VANE EFF

"+<::I. .89: 1. 0

METHOD 3.0 2. 5 1.5 2. 0 MAINSTREAM-TO-COOLANT TEMP RATlO

!z\-:9-2243

Figure VII-48

270

EXPERIMENTAL AND PREDICTED TURBINEEFFICIENCIES

-PREDICTION .86 METHOD

TURBINE . 82 a EFF

s7!!CI

1.15 h0 2!5 310 MAINSTREAM-TO-COOLANT TEMP RATIO

CS-TV-2223

FigureVII-49

TURBINE ROTORS USED IN TIP CLEARANCEINVESTIGATIONS

lo-in.

TIP DIAM ROTOR

5-in. TIP DIAM ROTOR

---:,-

'

Figure VII-50

271

TIP CLEARANCE CONFIGURATIONS

RECESSED CASING CONFIGURATION

REDUCED BLADE HEIGHT CONFIGURATION

Figure VII-51

EFFECT OF ROTOR TIP CLEARANCE ON TURBINEPERFORMANCE

,rRECESSED CASING l5-in. TURBINE)

.88 (5-in. TURBINE) I I .-.A 4 2 6 8 ROTOR TIP CLEARANCE, %

.84 0

Figure VII-52

272

VIII. William

MECHANICAL

COMPONENTS John J. Coy,

J. Anderson, Robert C. Bill, and David P. Fleming

National

Aeronautics and Space Administration Lewis Research Center

The term mechanical components incorporates machine elements such as bearings, gears, seals, clutches, and shafts or rotors. In this paper, some research on bearings, gears, seals, and rotor dynamics (spe fically high-speed balancing and dampers) pertains to problems in both airis reported. This L-;earch craft turbine engines and helicopter transmissions. TURBINE ENGINES bearings, of the comstages. VIII-l.

In turbine engines, the concerns are with main-shaft and gas-path seals at the tips dampers, shaft seals, pressor and turbine blades and between the compressor The locations of these components are shown in figure Main-Shaft Bearings

In turbine engines, main-shaft bearings are eitner ball or cylindrical-roller bearings. In present engines, they are required to operate at DN values up to about 2.4 million. DN is a speed severity parameter consisting of the product of bearing bore diameter, D, in millimeters, and the rotative speed, N, in Engine designers would like to have main-shaft bearings rpm. with a DN capability of 3 to 3.3 million for advanced engines. Extensive work has been done in conjunction with the General Electric Company to define and solve the probiems associated with operating ball bearings at 3 million DN. Approximately 70 000 operating hours have been accumulated at 3 million DN. The bearing design and lubrication problems associated with operating at 3 million DN have been solved, but catastrophic fracturing of the bearing inner ring occurs as a result of fatigue pitting. Figure VIII-2 shows one-half of the inner ring of a ball bearing run at 3 million DN. Fatigue pitting is the normal failure mode for rolling bearings. At lower speeds, it is a benign type of failure which can be detected by small chips in the oil. The bearing is then replaced. At very high speeds, the hoop stress in the rotating ring causes minute cracks in the 273

fatigue pits and complete

to propagate failure.

through

the

ring

and cause

fracture

Engine bearings are traditionally made of through hardened which have low fracture toughness. steels, One approach to this problem might be to use carburized materials with a hard surface layer and a soft core. Tapered roller bearings made of carburized materials are currently being tested at high speeds for This research may have application transmission applications. to engine bearing problems. The problem of bearing material fracture toughness is one that is confined to the extremely high speeds contemplated for fuA much more mundane problem that is advanced engines. ture, present every day in the operation of all machinery is dirt ingestion into the lubrication system. With the steady improvements that have been made in bearing material quality, more and more failures are traceable to contaminants in the lubrication For this reason, a research program was conducted to system. determine the effect of lubricant filtrat.ion on bearing life. Figure VIII-3 shows the bearing fatigue life results for these The lives are statistical lives, each being obtained tests. First a new, clean with a group of 65-millimeter ball bearings. Then a test series oil was used to determine a baseline life. was run with the oil purposely contaminated to simulate the normal operating conditions of a turbine engine in a real environA different filter size was used for each of four test ment. Using a 105-micrometer filter resulted in premature groups. With progressively finer filter failure due to severe wear. fatigue life continued to improve, although the mesh sizes, In fact, even with the finest (3-pm) filter, gains were modest. fatigue life was only about one-half as long as when the system Although the imwas perfectly clean and free of contaminants. provement in fatigue life with finer filtration was rather modit may be more significant in other systems. est in these tests, Another factor was the dramatic difference in appearance among Bearing surface distress the bearings in the different groups. and wear were much more prevalent in the tests run with coarser indicating that wear could be the mode of failure for filters, many of the longer running bearings with coarser filtration. filters are used in aircraft enNormally 30- to 40-micrometer Use of finer filters gine and transmission lubrication systems. could result in increased bearing life. Shaft Seals sliding speeds up to noncontact labyrinth proolems associ-

shaft seals operate at In present engines, Usually approximately 450 feet per second. seals are used because of wear and durability 274

ated with contact or rubbing seals. Engine designers would like to nave contact seals capable of operating at 700 to 800 feet per second. Sliding velocities of conventional contact seals are limited by the wear and heat generation associated With high-speed rubbing. The development of self-acting or filmriding seals has been pioneered at NASA. Early work on selfacting seals which incorporate Rayleigh pads to create a separating film was reported at the 1975 Aeronautical Propulsion Figure VIII-4 shows the spiral-groove Conference (ref. 1). which utilizes shallow spiral grooves in the self-acting seal, face of the stationary seal member to pump fluid and to estabbelish a very thin film which prevents direct rubbing contact tween the stationary and rotating seal elements. This design significantly reduces friction and temperature rise at high successsliding speeds. A spiral-groove seal has been operated fully for 100 hours at sliding speeds up to 800 feet per second. A seal of this type is presently being designed for retrofitting into a 1500-horsepower turboshaft engine in a helicopter. Gas-Path Seals

At the present time, an intensive program for the development of ceramic seals for the turbine blade shroud is being carried out. Present engines have metal shrouds which must be coolea with engine airflow. They also have low rub tolerance, which necessitates large tip clearances. These two factors - the need for cooling air and the need for large tip clearances - result in performance losses. Figure VIII-l, shows the locations of tne although simplified, gas-path seals. A large engine contains over 50 dynamic sealing locations. The purpose of gas-path sealing is to minimize leakage from the primary flow path through the compressor, burner, and turbine. This leakage occurs across the blade tips and at the labyrinth seals between stages; it has a very significant cumulative effect, not only on specific fuel consumption, but also on compressor stall margin and engine thrust. Furthermore gas-path sealing becomes more critical as engines advance to higher pressures and temperatures. One of the sealing locations most critical to specific fuel consumption is the highpressure-turbine blade tip, which is shown in detail in figure VIII-5. From a sealing standpoint, the operating environment is the most severe with gas temperatures near 26000 F and with pressures near maximum. In small engines, the turbine rub shroud is a single ring: it is constructed of segments in in large engines, order to reduce thermal strains. Typically each segment is impingement cooled and is held in place by a cooler outer case structure. 275

Figure sures clude

VIII-6 indicates and temperatures

the problems induced of advanced engines. requirements, compressor which shows and warping which up in can the

by the high presThese problems inbe 2 to form of 3 percent wear, eroof

(1) High cooling flow through the (2) Low durability, sion, cracking,

(3) Lack of abradability, roundness or eccentricity down by the turbine blades

which means the shroud out-ofcannot be corrected by being (illustrated in fig. VIII-7)

worn

minimum clearance between the blade and For engine efficiency, shroud is desired. However, the high-pressure-turbine shroud exaggerated in figure VIII-7, is common beout-of-roundness, cause of (1) (2) (3) (4) (5) Nonuniform Nonaxisymmetric Flange leakage distribution flow, in of particular, cooling hot air gas recirculation temperatures structures

Nonuniform Localized

An important point is that the out-of-roundness is often greater with convenTherefore, than the desired operating clearance. tional blade and shroud construction, either the initial assembly clearance must be large enough to preclude rubbing or the In either case, the out-of-roundness will cause blade wear. out-of-roundness results in an increased clearance which can be This clearance, as large as 1 to 2 percent of the blade height. can result in a significant penalty in terms of specifin turn, ic fuel consumption. Typically in a modern, highly loaded turbine, a clearance cnange of 1 percent of blade heignt changes the specific fuel consumption by 1.5 to 2.0 percent. Each 1 percent of specific fuel consumption is a very significant figure because fuel costs account for nearly 50 percent of the direct operating cost, which is most of the ticket price. A l-percent change in specific fuel consumption applied to the Nation's commercial fleet represents about 150 000 000 gallons of fuel per year. A significant portion of the fuel loss due to large out-of-roundness clearances could be saved if the local bumps were abraded away without blade wear as indicated in figure VIII-8.

276

Several new shroud concepts are being developed witn tne goal of mitigating the problems of high cooling requirements, low durability, and lack of abradability. Abradable metal shrouds have been developed for use in present engines. ceramic shrouds for even higher temperatures are under development. Ceramic shroud development is not yet complete, but some very interesting results have been obtained. The most challenging and immediate problem connected with the use of ceramic materials in the high-pressure-turbine seal application is retention of mechanical integrity under conditions of severe thermal gradients imposed by the engine operating Currently being evaluated are two approaches to ensuring cycle. mechanical integrity of ceramic seals, both based on plasmasprayed zirconium oxide (Zr02). incorporates metal-ceramic shown in figure VIII-g, One approach, composite intermediate layers between the alloy substrate and The function the fully ceramic layer adjacent to the gas path. is to mitigate thermal of the metal-ceramic intermediate layers stresses imposed on the brittle ceramic by gradually varying the thermal expansion and mechanical properties from those of the Variations on metal substrate to those of the ceramic layer. this approach include nickel or cobalt alloys with chromium, and yttrium (NiCrAlY or CoCrAlY) as the metal conaluminum, stituent and different arrangements and compositions of interPioneer analytical and experimental work on mediate layers. this approach was conducted under NASA contract by the Pratt & Whitney Division of United Technologies Corporation. shown in figure VIII-lo, employs a strain The Other approach, The isolator pad of porous metal with Zr02 sprayed on top. case a Brunswick fiber metal, looks strain isolator pad, in this but is structurally very very sparse on a microscopic scale, sound and protects the ceramic from the thermal expansion difThe grey ceramic is ference between the ceramic and the metal. Other strain isolator pad systems being evaluated the zr02. include alternate fiber and metal layers, low-density sintered and low-density plasma-sprayed metals. metals, The durability of the ceramic shroud materials is ~being investigated in thermal cycling tests in which the surface temperature As illustrated in figure of the ceramic is held at 24000 F. VIII-11, an oxyacetylene torch is used, and the ceramic is The cycled into the torch and then into a stream of cold air. ceramic is alternately heated to 24000 F and cooled to about 5000 F. In the heating portion of the cycle, the temperature of the ceramic surface reaches 2OOOo F in about 20 seconds and Cooling air is directed onto 24000 F in another 10 seconds. the metal backing so that the backing temperature is about lloo" F. The hot portion of each cycle lasts about 33 min277

utes. those

The thermal gradients imposed by the engine.

in

this

test

are

more

severe

than

Figure VIII-12 shows a graded layer specimen after 1000 thermal The shroud segment is in excellent condition; needless cycles. shroud that had been at 24000 F would be in to say, a metal very poor condition. Similar results were obtained with the strain isolator pad specimen, shown in figure VIII-13 after bO0 There is no delamination cracking of the bond between cycles. the ceramic and the porous metal substrate. The effect of high cooling flow requirements on engine core flow is illustrated in figure VIII-14. The cooling flow requirement of a conventional shroud, which must be held to about 18000 F, can be as high as 3 percent of the engine core flow; this is a In contrast, significant penalty. a ceramic shroud can be operThe ated at a higher temperature and requires less cooling. immediate goal is 24000 F, and for th's shroud temperature, i the cooling flow requirement is only -i-percent of the engine The differencelin cooling flow requirements in tnis typiflow. which translates into a specific cal illustration is 2Fpercent, For this reason fuel consumption decrease of about 1 percent. the ceramic shroud concepts are very attractive. alone, Another aspect of durability is erosion resistance to particles generally these particles have an imin the hot gas stream; pingement angle of about 150, as illustrated in figure VIIIOne inherent advantage of ceramics as compared with metals 15. is that the erosion resistance at low impingement angles, like 150, is generally At high impingement angles, ceramics good. resistant than metals. Fortunately are generally less erosion the impingement angles are small. The test setup used in work performed under NASA contract by Pratt & Whitney to measure erosion is shown in figure VIII-lb. The specimens are positioned at the end of a combustor nozzle at an angle that approximates the impingement angle in an engine. The surface temperature is brought up to 24UUo F, and then hard particles are introduced at a controlled rate. Some re'The zirconium oxide was neld sults are shown in figure VIII-17. at 24000 F; the metallic system was held at 2000 F. The erosion rate of the metal was significantly higher. The problem of lack of abradability is being investigated, and one approach is shown in figure VIII-18. Abradability is obtained by bonding abrasive grits to the turbine tips. The turbine then acts as a grinding wheel. At this time, only preliminary results have been obtained. A second approach is to make the ceramic layer more abradable by introducing porosity. Some methods of doing this are currently being evaluated under NASA contract by Detroit Diesel Allison Division of General Motors Figure VIII-19 shows a porous ceramic specimen Corporation. 278

that has been nificant wear

rather to the

cleanly blades.

cut

by turbine

blades

without

sig-

In summary, gas-path sealing becomes more critical as engine A particularly critical pressures and temperatures increase. sealing location is the high-pressure-turbine blade tip. Here the ceramic shroud offers a solution to the problems of high cooling airflow requirements, low durability, and lack of abradability in advanced engines, where the potential reduction in specific fuel consumption is a very significant 2 to 4 percent. Rotor Dynamics fuel economy is to make the seal clearance is limited by that is, clearance has to of the rotor or it will rub. rotor vibrations is called

One of the keys to achieving better The minimum seal clearance small. the vibration amplitude of the rotor; be greater than tne dynamic motion The art and science of controlling rotor dynamics.

The first is There are three principal goals in rotor dynamics. The second is to minimize to minimize vibration amplitudes. Obviousiy vibratory forces on the rotor support bearings. The best rotor supachieving this will increase bearing life. port system for low amplitude is not always the best one for low The third bearing forces, so the design has to be a compromise. goal is to reduce stresses due to vibration in the rotor and support structure. Rotor dynamics must be considered more and more in designing and operating rotating machinery because of two principal factors: first, engine and airframe structures are lighter and more fiexible, and second, macnines are being operated at high speeds, so that the power per pound increases. The effect of these two factors is that rotors are now running above one or more bending critical speeds. A bending critical speed is a resonance where a significant part of the motion that occurs is bending of the rotor itself, as opposed to deflection of the bearings or other static structures. Figure VIII-20 illustrates a bending critical speed. When the rotor speed is higher than a bending critical speed, the rotor is operating supercritically. Supercriticai operation is more difficult than subcritical. The problem occurs not in running supercritically, but in passing through the critical speed because the rotor is not balanced perfectly; that is, the mass axis, or center of gravity axis, does not coincide with the axis of rotation. Because the rotor is not balanced perfectly, there is a rotating force applied to the rotor, and at resonant speeds, the vibration amplitude can be quite large. 279

Certain things can be done to make supercritical operation feathe rotor can be balanced very accurately to resible. First, 'Second, the rotor can be accelerated duce the exciting force. through the critical speed rapidly so that the amplitude does Third, not have time to build up. the rotor supports can be modified to keep the amplitudes and bearing forces toleraole even with an imperfectly balanced rotor. This modification entails designing appropriate stiffness and damping in the supports between the bearing and the engine case (fig. VIII-l). Actually both accurate balancing and properly designed supports research is being carried out in both areas. are needed; The acceleration rate often cannot be chosen, so that option usually is not available. High-Speed, Multiplane Balancing

Under NASA contract, Mechanical Technology, Inc., has been working on applying the influence coefficient method of high-speed, multiplane balancing. Most recently the research has included Figure VIII-21 the use of a laser to do the actual balancing. shows a laser beam firing at a test rotor being balanced. A computer controls the laser and gives it the right number of pulses to remove the proper amount of materiai. With tne laser, the rotor can be balanced while spinning; it does not have to be stopped as when grinding or drilling is used. Typical results of the experimental laser balancing are shown in figure VIII-22. Successive corrections were made to reduce the More than one run is needed because rotor vibration amplitude. the calculated required number of laser shots was based on a When a large number of pulses were trial of only a few pulses. fired, the amount of metal removed per pulse decreased. It puts In this work, a fairly small laser was used. pulse every 2 seconds, and each pulse takes off about The next phase of this work grams of rotor material. volve larger scale rotors and a more powerful laser. out a 2 miliiwill in-

High-speed, multiplane balancing has also been applied to the shown in figure VIII-23. It power turbine of the T55 engine, looks substantial, but an analysis shows a good deal of bending at the critical speeds. As shown in figure VIII-24, there is not as much bending at the first critical, 12 000 rpm, as at the 21 000 rpm. second, A rotor response analysis snowed that at the normal operating speed, 16 000 rpm, the rotor moue shape is that of the second critical speed. So even though normal operation is at 5000 rpm below tne second critical speed, bending still occurs and causes trouble. In the iaboratory experiments, high-speed balancing did a good job of smoothing out this rotor: even a rotor that was previously low-speed balancea was a lot quieter. A hiyn-speed balancing rig is presently being in280

stalled at the Army's overhaul can accommodate power turbine the smaller T53 engine. Although the T55 rotor of the new engines are turboshaft engine rotor is about 3 feet long. ical speeds on the way manufacturer balances but high-speed balancing is coupled with a laser

depot rotors

at of

Corpus Christi. the T55 engine

The rig and also

is rather thick and rigid, rotors in some A current-technology much more flexible. This rotor is shown in figure VIII-25. It has to run through two bending critup to its normal operating speed. The this rotor by using traditional methods, may be advantageous, particularly if it in an automated procedure.

There are several areas in which computer-controlled laser balIt can,be most easily used for componancing might be applied. ent balancing, as an automated replacement of present methods. may be in high-speed balancing of The real potential, though, The best balancing would be done on complete rotor assemblies. an assembled engine because the balance would not be disturbed Most present engines are not by disassembly and reassembly. It will have to be planned for designed for this procedure. Access to the balancing areas will when the engine is designed. be needed as well as enough sacrifical material that can be reIf force or displacement sensors were moved by the laser. placed in the engine permanently, rotors could be rebalanced This even with the engines still installed on the airplane. balancing would provide smoother running engines; there would be fewer unscheduled removals for high vibration, and engines could be expected to run longer between overhauls. Elastomer Dampers

One means of allowing supercritical operation is to properly tailor the stiffness and damping properties of the rotor supGetting stiffness is quite easy with mechanical springs, port. Some of the so most of the research effort has been on dampers. functions of bearing support dampers are (1) to absorb vibraand in so doing reduce vibration amplitudes and tional energy, bearing forces; (2) to allow an engine to shut down safely if it and (3) to allow less accurate balancing, because loses a blade; dampers lower vibration amplitudes and bearing forces. There ery. really are three The first, suitable basic types of the dry-friction for high-speed dampers used in rotating or Coulomb damper, is machinery. machinnot

The second, the hydrodynamic or squeeze-film damper, has been Many applied very successfully in many different applications. Two principal turbine engines would hardly run without them. disadvantages are that an external oil supply is needed and that 281

the damping motions.

coefficient

increases

very

sharply

for

large

damper

The third, the elastomer damper, largely eliminates tnese disIt is self-contained and, for the range of ampliadvantages. tude usually designed for, does not exhibit the stiffening effect of the squeeze-film damper. Of course, it has some disadvantages of its own, but for some applications, it can perform For the past 7 years, very well. a technology effort has been carried out by Mechanical Technology, Inc., under NASA contract, to study the properties of damping materials from an engineer's point of view rather than a physicist's or a chemist's. At measurements were made from simple shear specimens on a first, Then the basic properties were used in designing a shaker. damper for a rotor support system. The damper design, shown in figure VIII-26, is extremely simple and versatile. In a typical the rotor turns in a ball bearing whose outer race application, is pressed into the inner damper housing. Between the inner and outer damper housing, three or more buttons of elastomer material are placed. This damper was evaluated on a rig simulating a power turbine The rotor was balanced very carefully and then run with rotor. varying known unbalance to evaluate damper performance. As a the elastomer buttons were removed from the damper final test, Tne change in rotor benavior and replaced with steel blocks. Figure VIII-27 shows a plot of rotor vibrawas quite dramatic. Witn the elastomer suption amplitude over a range of speed. there were no amplitude peaks in the operating speed port, range, though when discrete unbalances were used, a critical On rigid supports, the speed was noted at about 10 000 rpm. critical speed shifted up to about 19 UOU rpm, and the amplitude increased to almost 10 mils. So the elastomer support was extremely effective in reducing vibration amplitude. In anotner comparison, both a squeeze-film damper and an elastomer damper were designed for a long drive shaft tnat was intended to run supercritically. The two dampers were run successiveFigure VIII-26 snows a plot of ly with identical shaft balance. The squeeze-film vibration amplitude as a function of speed. damper was more effective in limiting rotor amplitude at the first critical speed, but the elastomer damper was more effective at the second critical. So, overall, the elastomer and squeeze-film dampers performed comparably. As discussed in this section, elastomer dampers have several they are self-contained; the design can oe very advantages: dimensional tolerances are not as critical as in other simple; and tne stiffness and damping do not increase rapidly dampers; Elastomer dampers have limitations as the amplitude increases. many rubbers have to be kept away from any oil; elastomer too: properties can change with age; and elastomers are sometimes 282

severely affected by temperature change (this effect can be crucial, since the vibration energy absorbed by the damper is conFor some verted to heat, whicn raises the damper temperature). an elastomer damper is the best kind. applications, though, HELICOPTER TRANSMISSIONS

A transmission is nothing more than an assembly of bearings, shafts, and seals which transfers power mechanically, gears, In a helicopter, it usually from a high speed to a low speed. transfers power from the engine to the main rotor and to the tail rotor drive. The goals of transmission research are to reduce maintenance intervals; increase life and reliability; and increase efficiency and reduce weight, noise, and vibration; In the area of gearing, new allowable operating temperature. and new methods of predicting gear gea,r steels are being tested, reliability are being developed which will result in more reliAlso being investigated are imable gears with longer lives. proved gear designs which will increase gear life and reliabilHigh-speed photography ity and also reduce noise and vibration. is being used to study ways of more effectively lubricating gears for more efficient operation and better cooling. to benefit that area and Improvements in one area are expected others as well. reductions in impact loading of For instance, gear teeth through improved design methods will increase gear life and at the same time reduce noise and vibration. Transmission Tests

Helicopter transmission tests are currently being conducted at The figure Lewis by using the test rig shown in figure VIII-29. through the obsershows a view from the control room, looking The helicopter transmission vation window into the test cell. is mounted in the test stand. The general method used is to first research bearing and gear components by using analysis and experimental bench testing. When a component looks promising after this testing is done, it is put into the transmission system for further testing and evaluation. Figure VIII-30 shows a cross section of the helicopter transmission currently being tested. This transmission consists of a high-speed input shaft with a spiral bevel gear driving the intermediate shaft by transmitting power through a planetary gear train. The planetary gear train consists of a sun gear, three planet gears, and a ring gear. The overall reduction ratio is approximately 17.5:l. This transmission contains 8 gears and 14 bearings. Note the stack of three ball bearings supporting the 283

input pinion shaft. This is one of the complexities designers would like to eliminate because proper load sharing in a threetandem-bearing array is difficult. Even if the bearings are made with perfect geometric accuracy, thermal and mechanical distortions can result in one bearing carrying more than its share of the load and in premature bearing failure. Figure VIII-31 shows the transmission retrofitted with advanced These components will offer improved reliability components. and higher power transmission capacity. Improved tapered roller bearings on the input and output shaft will be used, as will improved spiral bevel gears and high-contact-ratio gears with improved bearings for the planetary train. After these changes will be tested and compared with the are made, the transmission standard transmission. Gear Research used at Lewis in the 3.5 inches in pitch diin the spur gear test rigs are used to perform of materials, lubricastudied.

Figure VIII-32 shows the spur test gears These gears are gear fatigue test rig. They are run ameter and have 28 teeth. Four such rig shown in figure VIII-33. long-term endurance tests. Tne effects and design on gear life are being tion,

Several different types of gear failures occur, as shown in figure VIII-34. Scoring is caused by poor lubrication and highTooth fracture is caused by high tooth temperature operation. bending stresses and is aggravated by poor heat treatment. Fracture usually begins at the root of the tooth, but may also originate at a surface fatigue pit. Surface fatigue pitting is caused by repeated application of contact stress. The nigher the contact stress, the more rapidly failure occurs. Gear materials. - Scoring and fracture failures may be controlled by proper design, but fatigue pitting is an intrinsic problem in gear applications. Pitting is an unavoidable event that eventually ends the useful life of a gear. Figure VIII-35 shows data for test gear failures due to pitting fatigue. The failure points are plotted on Weibull coordinates, which give the cumulative percentage of failed specimens as a function of The notable feature of presenting the data this running time. way is that it emphasizes the relation between life and reliFor instance, the life at which 10 percent of the gear ability. specimens have failed is denoted the L-10 life. This is conceptually identical to the B-10 life for bearings. The L-10 life corresponds to YO percent reliability; the L-50 life corresponds to SO-percent reliability. Other life numbers may be defined as the degree of reliability required changes.

284

In figure VIII-35, the fatigue data for The L-10 life of CBS 1000 steel pared. 9310 steel, which was used as a baseline

two gear is twice material.

steels are comthat of AISI

Figure VIII-36 shows the results from the fatigue tests of six gear steels. AISI 9310, the baseline for comparison, is assigned a relative life of 100 percent. In the tests, only pitting The CBS 600 gears fatigue failures occurred in this material. had an L-10 life 7.5 times that of the AISI 9310 gears. The In some cases, the failure mechanism was primarily pitting. cracks near the pitting failure propagated through the tooth to The forged M-50 material behaved similarly, cause fracture. The M-50 with a life six times that of the baseline AISI 9310. and CBS 1000 gear steels have good high-temperature hardness retention, which results in longer life at elevated operating temperatures. Gear designs. since they have well as reducing shows test gears signs. Contact meshing contact. and low contact High-contact-ratio gears are being investigated, potential for improving life and reliability as Figure VIII-37 dynamic loading of gear teeth. of high-contact-ratio and standard-ratio deratio is simply the average number of teeth in High contact ratios are ratios greater than 2, ratios are those between 1 and 2.

High-contact-ratio and standard gears were tested at the same The results are presented power level to compare performance. in figure VIII-38. Compared with standard gears, high-contact The ratio gears show a 300-percent improvement in L-10 life. is not without a price. improvement in fatigue life, however, high-contact-ratio gears are more As found in additional tests, susceptible to scoring by a 22-percent margin in allowable load. The bending endurance fracture limit is 15 percent iower because the teeth of the high-contact-ratio gear are more slender. Also these gears run hotter because of increased sliding of the longer teeth. With proper lubrication and design, the high-contact-ratio gear can be used to achieve longthough, er life and more reliability. Gear lubrication. - Research into lubrication of high-speed The gears has been conducted by using high-speed photography. lubricant in a high-speed, highly loaded gear system has two functions: to provide a lubricant film between the meshing teeth and to provide cooling of the gears. The effectiveness in performing these two functions is directly related to the depth of penetration of the jet of lubricant between the gear teeth. Lubrication jets may be tangentially directed into the mesh exit side of the gear pair, as shown in figure VIII-3Y, or directed radially, as shown in figure VIII-40. High-speed photography showed that more effective penetration to the dedendum, or root of the tooth was achieved with the radially directed region, 285

From the results of the lubrication study, an analytical jet. model was developed. From the model, a lubricant supply 'system can be tailored to provide the desired penetration of the lubricant jet and thus assure good lubrication and cooling of the gear teeth. Tapered Roller Bearings

Tapered roller bearings can support high radial and thrust loads, so they can, in theory at least, be used to replace a stack of two or three ball bearings, such as used in conventional transmissions (fig. VIII-30). This replacement would simpliConventional tapered roiler bearings are fy mechanical design. not designed for high-speed operation, however, so it has been necessary to upgrade both bearing design and lubrication. A tapered roller bearing designed for high-speed operation is shown in figure VIII-41. Proper lubrication is especially important in high-speed bearing Figure VIII-42 shows the experimental results Obapplications. tained with two high-speed tapered rolier bearing designs and Bearing temperature is plotted as a three lubrication systems. function of speed for an optimized high-speed design and a more The test objective was'to operate at high conventional design. speed without exceeding temperature limits. First the baseline design bearing was jet lubricated in the conNext the ventional manner with a flow of 3 gailons per minute. flow was split so tnat 2 gallons per minute was jetted in and 1 gallon per minute was introduced into the cone-rib area through With a series of equally spaced radial holes in the inner ring. this change in the lubrication system, considerably higher be achieved without exceeding the temperature speeds could limits. Further gains were achieved with the optimized high-speed bearThis design consists of redesigned rolling elements and a ing. A new lubrication scheme was also new inner ring riding cage. the lubricant flow of 3 gallons per minute was equally used: divided among circumferentially spaced holes, from wnich it lubricates the cage running surface as well as the critical neat generation zones of the rollers. The new design makes it possible fast as with the baseline design temperature limits. A contractor bearings at is the to run without at shaft speeds twice as exceeding the operating tests of advanced tapered roller transmissions.

now conducting endurance high speeds required for

286

SUMMARY Some of nautics our are program results as follows: and their potential impact on aero-

(1) Finer lubricant filtration should especially where dirt rotating machinery, premature failures occur. The investment payoffs can be significant.

be considered for all ingestion, wear, and is modest, and the

(2) Self-acting seals offer tremendous promise for improved seal performance and life in high-speed rotating machinery. The 4-inch-diameter seal Lewis is retrofitting into tr;e i>U0-horsepower turboshaft engine will save 32 norsepower bec.:ii;se of reduced leakage. (3) Successful application blade tip seals would save tne of fuel per year oy the Nation's of abraiisal? use of lS0 commercial
cerdxiz
iLii:dltle

to

6uir million airline fleet.

gallons

(4) Control of rotor vibration is one of the essential elements entering into successful operation of ali high-speed High-speed, I;lultiplalle balancing is more effective machines. than conventional low-speed, two-plane balanclny. Supercrltical operation of shafting is feasible witn good balancing and proper use of dampers. Elastomer dampers can be very effective in attenuating rotor vibrational amplitudes in the regions of critical speeds.

(5) High operating cost is one of the major impediments to expanding the market for nelicopters. Extended gear life can be achieved with new gear materials. High-contact-ratio gear designs may also extend life and may reduce noise. Improved gear lubrication and the use of advanced-technology tapered roller bearings will increase transmission efficiency and reliability. REFERENCE 1. Aeronautical Propulsion. NASA SP-381, 1975.

287

MECHANICAL COMPONENTS AREAS OF RESEARCH

rLABYRINTH ',SEALS \ AIR \ / / /' ,rABRADABLE / 1' / /-DAMPER7 / \ ! TIP SEALb, \ \ '\ \ \ \

A-.-+

.-.-----&--.---

+------~BEAR,NGS

_/.j---L-----

\SHAFT
cs-79-1975

SEALS' Figure VIII-l

FRACTURED INNER RING FROM BEARING

RUN AT 3 MILLION DN

Figure VIII-2

288

EFFECT OF FILTRATION ON BEARING LIFE

25oor-

la
BASELINE DATA (NEW OIL) TEST SERIES (CONTAMINATED OIL) 0 -. ._

AVG FATIGUE LIFE, hr

q --

500 0

;Ti =I 49

SEVERE WEAR 105

? :I

& % .-_ 7 i0

49 FILTER SIZE, pm

Figure VIII-3

SPIRALGROOVE SELF-ACTING SEAL

,- 24 SPIRAL GROOVES

Figure VIII-4

289

TURBINEGASPATHSEALING
AIR

GAS FLOW '-SEGMENTED RING OF RUB MATERIAL Figure VIII-5

cs-73-1?53

TURBINE GASPATHSEALING PROBLEMS

r HIGH COCLING

LOW DURABILITY---

LACK OF ABRADABILITY-

I/

Figure VIII-6

290

TURBINEBLADE TIP CLEARANCE

IDEAL CONCENTRIC ,_-72-1

j7')

OUT-OF-ROUND NONABRADABLE Figure VIII-7

ABRADABLE TURBINE SHROUD CLEARANCE

OIJT-OF-ROUND ABRADABLE :'I-':-::'

Figure VIII-8

291

GRADED LAYER CERAMIC SHROUD

Figure VIII-9

TURBINE SHROUD
STRAIN ISOLATOR PAD CONCEPT BASE

CS-73-1934

figure VIII-10

292

THERMAL CYCLING TEST r -7, j .-.-;

COLD AIR NOZZLE

COOLING AIR TO BACK

ACETYLENE OXYGEN TORCH

CERAMIC SHROUD SPECIMEN

cs-79-1961

FigureVIII-11

GRADED LAYER CERAMIC SHROUD

Figure VIII-12

293

STRAIN ISOLATOR PAD SHROUD

Figure VIII-13

COOLING FLOWREQUIREMENTS
3

2 % ENGINE CORE FLOW 1AFLOW REQUIRED FOR CASE REDUhTION IN COOLING

ALLOWABLE SHROUD SURFACE TEMP, OF Figure VIII-14

y.--l,--

294

TURBINESHROUD EROSION

ANGLE

FiyureVIII-15

EROSION RIG

FigureVIII-16

295

EROSION RATE

EROSION RATE

CERAMIC Figure VIII-17

METAL

ABRADABLE TIP SEALING

3_:-5-1?5;

Figure VIII-18

296

ABRADABLE

CERAMIC

FiyureVIII-19

BENDING CRITICAL SPEED

FigureVIII-20

297

LASER REMOVINGMETAL FROM TEST ROTOR

Figure VIII-21

RESPONSE OFTESTROTOR TO LASER BALANCING

------5------4I ROTOR AMPLITUDE, MILS 32l/ I 6000 SPEED, rpm Figure VIII-22 I 8000
~~-79-1968

UNCORRECTED CORRECTION 1 CORRECTlON 2 CORRECTION 3

I\

II II .

----__

--I 4000

---

0 2000

298

T55 POWER TURBINE ASSEMBLY

,--BEARING

FigureVIII-23

T55 POWER TURBINE CRITICAL SPEED MODESHAPES

OPERATING SPEED, 16000 rpm ~~--$~;;u~wAL SPEED.

DRIVE END

I I BEAFilNG I

I I BEAhG I

TURBINE END

Figure VIII-24

299

POWER TURBINE OF CURRENTTECHNOLOGY TURBOSHAFT ENGINE

ROLLER BEARING7 ,-BALL BEARING

FigureVIII-25

ELASTOMERIC DAMPER

,-BALL

BEARING

ELASTOMER BUllON

FigureVIII-26

VIBRATION AMPLITUDE WITH ELASTOMER AND RIGID SUPPORTS

AMPLITUDE, MILS

ROTOR SPEED, rpm Figure VIII-27

CS-73-1983

VIBRATION AMPLITUDE OF SUPERCRITICAL DRIVESHAFT

25 I,r~~~~~~~ 20 : I FILM

I
VlBRATlON AMPLITUDE, MILS 15 10

,-SECOND

CRITICAL

5 uCS-79-1986

I 2ooO

I I 4ooO 6OCKI SPEED, rpm

I 8ooO

Figure VIII-28

301

HELICOPTER TRANSMISSION TEST RIG

Figure VIII-29

HELICOPTER TRANSMISSION WITHSTANDARD BEARINGS AND GEARS

OUTPUT SHAL-I, BEARINGS i$J GEARS g& SEALS

-=T

Lh-/Y-lb/

Figure VIII-30

302

HELI COPTER TRANSMISSION WITHADVANCED BEARINGS ANDGEARS

OUTPUT ,,-NEWTAPERED ROLLER PPORTBEARING

IMPROVED BEARINGS

llNPUT SHAFT

BEARINGS

ADVANCED DESIGN & ' MATERIAL BEVEL GEAR J

8 GEARS
@SEALS Figure VIII-31

' NEW DESIGNTAPERED ROLLERTHRUSTBEARING

n-791626

TEST

GEARS

Figure VIII-32

303

SPUR GEAR TEST RIGS

FigureVIII-33

TYPICAL GEAR TOOTH FAILURE MODES

SCORING

FRACTURE AT TOOM ROOT

NASA TESTGEAR

FRACTURE FROM SURFACEPIT

SURFACEPllllNG
cs-l&l?36

Figure VIII-34

304

WEIBULLPLOT- GEARPITTINGFATIGUE

STATISTICAL % OF GEAR SPECIMENS FAILED

cs-79-1913

1000 IO 20 40 60 100 200 400 LIFE, MILLIONS OF GEAR REVOLUTIONS Figure VIII-35

GEAR FATIGUE TEST RESULTS

MATERIAL -;.(. ,,, : :.;;..,: AISI 9310 SUPER NlTRALLOy a AISI M-50 FORGED M-50 CBS ID00 CBS 600 ;,.;;:.-. :;;1 ..,..~.,.~~~i;;-ii;i !:J; I 100 ,I.;.:-,,<l..~:

FAILURE MODE (PIlllNG) (PITTING/SCORING) (PITTING/FRACTURE) (PITTING - FRACTURE) (PllllNGIFRACTURE) (PI m NG e FRACTURE) ~;$,~~.c<~. -;>.2 ,;.,<y.; .:,:. .:;:,,:!::;~~.~..:;:;.;:.I., .T:. ~!...~<:t.:~~,,f 0 I 6 I 600 200 % RELATIVE LIFE Figure VIII-36 I 700 I 800
cs-79-1978

305

HIGH CONTACT RATIO 8 STANDARD GEARS

Figure VIII-37

PERFORMANCE COMPARISON
HIGH CONTACT RATIO GEARS VS STANDARD GEARS

PllllNG

FATIGUE LIFE

i I_ ,.,. *._. m_I. -22/ ._ -15% E

. . _ : .:,::,.i~-~;.:-j +3(-Jc% v

SCORING LOAD BENDING ENDURANCE LIMIT OPERATING TEMP

Figure VIII-38

306

OUT-OF-MESH JETLUBRICATION n

Figure VIII-39

RADIAL JETLUBRICATION

Figure VIII-40

307

TAPERED ROLLER BEARING

Figure VIII-41

TAPERED ROLLER BEARING TEST RESULTS

BASELINE DESIGN 0 0 JET LUBRICATION SPLIT FLOW (JET & UNDERRACEI HIGH SPEED CATION

BEARING TEMP, OF

10 000 15 000 SHAFT SPEED, rpm FigureVIII-42 308

2ooctJ
cs-79-2018

IX. William

INSTRUMENTATION C.

TECHNOLOGY Jr.,

Nieberding, David R. Englund, and George E. Glawe

National

Aeronautics and Space Administration Lewis Research Center

Current propulsion instrumentation needs are characterized by the requirement to make measurements under more severe conditemperatures, on the tions than in the past - such as at higher in a higher vibration environment, or in very rotating parts, There is also a need to get more high-frequency small passages. flow velocity and angle, temperdata sucn as dynamic pressure, This leads to vast stress, and blade position. ature, strain, To be of any practical use, these data must quantities of data. be processed and reduced to humanly understandable form very rapidly. This usually means some form of on-line data processing coupled with an easily interpreted display. new technological advances Fortunately, Those are the problems. Two of the can and are being exploited to solve these problems. most significant technological advances affecting new measurement system development are optics and microelectronics. The field of optics ,with its lasers, finer optics, and electrooptic Lievices makes possible noncontacting, nonintrusive, and The environmentaLiy tolerant sensors and transmlsaion schemes. field of microelectronics extends from the ultra-small semiconmeasurements to the data-handling auctor sensors for jynamic especially the small compower and econo,mics of the computer, puter. Ti:iS qd.t>er deScK i.Ie:; ;OinP of tne efforts made at Lewis in applying these new tools to today's measurement problems. The topics covered in this paper were selected as representative of the The first topic is a blade-tip areas Lewis is working in. clearance system. Tne second topic is a pulsed-thermocouple system used to measure gas temperature with a thermocouple at The temperatures aoove the melting point of the thermocouple. next topic is an optical technique for measuring blade flutter. This is followed by tne results of work aimed at measuring unThe last two topics deal steady pressure on rotating blades. The first is a probe for dynamic flow with flow measurement. and flow angle measurement, and the second is a laser anemometer system for rapidly mapping the flow profiles between tne blades of a rotating colmpressor.
309

BLADE-TIP

CLEARANCE MEASUREMENT

One of the conclusions of the Aircraft Energy Efficiency (ACEE) described earlier in this conference publication, is program, that increased blade-tip clearance is one of the major contributors to performance deterioration in modern commercial airTne importance of blade-tip clearance is also craft engines. emphasized by the work described in the preceding paper on It is abradable rub strips to reduce the effect of blade rubs. to be able to accurately measure the tip clearance of important and turbine blades in an operating engine. fan, compressor, An optical blade- tip clearance measuring system developed by Pratt & Whitney Aircraft under an NASA contract is described The objective of the development program and the perforhere. The range mance goals for the system are listed in table 1X-l. accuracy was of measurement was to be from 0.005 to 0.125 inch, and resolution was to be within 0.001 to be within 0.002 inch, The probe was to be cooled so that it could operate in a inch. turbine environment with a wall temperature as high as 19000 F. The probe used in this system works on the light triangulation With this technique a spot of principle shown in figure 1X-l. If the blade moves up laser light is focused onto a blade tip. which indicates a reduction in tip to the dashed-line position, the spot of laser light moves from point A to point clearance, An objective lens projects an image of the light spot onto B. With the blade in the original position, the detector array. the image falls at point A on the detector; with the blade in the image is at point B. The detector the dashed-line position, is a linear array of photodiodes that puts out a signal only Thus the system provides from the diodes that are illuminated. a discrete signal that indicates the blade-tip clearance. This technique has the advantage that the output signal is not determined by the magnitude of the light falling on the detector This means that a variation in the but rather by its position. reflectance of the blade tips does not cause an error; the only requirement is that sufficient light be reflected to provide a signal. The The actual system is shown schematically in figure 1X-2. engine-mounted sensor is a cylinder less than an inch in diamThe relatively small diameter is achieved by placing a eter. This makes the incoming laser prism in the tip of the probe. beam and the reflected light beam more nearly parallel to the For use at high temperatures the prism is made of probe axis. and nitrogen gas is passed through the probe for sapphire, The gas exits over the face of the prism to keep it cooling. Fiber optics are used to couple the probe to a remote clean. This arrangepackage that contains the laser and the detector. ment results in a smaller probe and a more acceptable environ310

ment for the contents of the remote package. The fiber optic bundle carrying the reflected light signal is a coherent bundle so as to preserve the relative position of the light spot. An image intensifier is inserted between the coherent fiber optic bundle and the detector array in order to provide sufficient light at the detector for measurements of individual blades. The detector has 256 photodiodes in a l/4-inch array to provide The output signal from the detector is the required resolution. fed into a data system incorporating a small computer that corMore detail on the design relates clearance with blade number. system is available in reference 1. and operation of this The system has been delivered to Lewis and is being evaluated in Test results show that the performance goals the laboratory. have been met, including accuracy within 0.002 inch and resoluTne probe mounted in a test fixture untion within 0.001 inch. der a compressor wheel is shown in figure 1X-3. Future plans in tip clearance measurement inciude the development of a multiple-probe system so that a number of optical Also, probes can be used with a single data-handling package. advanced capacitance systems are being considered with the hope of developing lower cost, flyable, tip clearance instrumentation. PULSED THERMOCOUPLE In selecting a thermocouple for measuring high-temperature gases, one important limiting consideration is the melting point Also to be considered are physiof the thermocouple material. cal and chemical changes that degrade the thermocouple and shorten its useful life when it operates continuously at high temperatures and pressures. A pulsed-thermocouple system has toeen designed to alleviate these problems (ref. 2). A representative sketch of a pulsed-thermocouple probe is shown It has a simple "V"- snaped thermocouple elein figure 1X-4. by a jet of inert gas coming out of a nozment, which is cooled The curve in figure zle on the leading edge of the probe body. The IX-4 shows the operating cycle on a temperature-time plot. top horizontal line represents gas temperature, and the second horizontal line represents the melting point of the tnermocouple The initial part of the curve during the cooling-on material. time period shows the temperature level to which the thermocouple is initially cooled by the gas jet. the cooling jet is turned off Wnen a measurement is to be taken, Before the thermocouple and tile thermocouple begins to heat up. cooling is automatically reapplied. reaches its melting point, This heating curve is recorded by a high-speed digital system. where they are curve fit The data are then sent to the computer, The extrapolated curve shads tne indicated and extrapolated. 311

equilibrium tained if

its

temperature melting

that the thermocouple point would have allowed

would have atit to do so.

A three-element pulsed-thermocouple probe used in a recent reThe probe was search combustor program is shown in figure 1X-5. where the extrapolated temperrun in a 38000 F gas stream, The thermocouple ature steps were each about 300 degrees F. and extends from a cone-shaped nozzle, element is "V"-shaped which brings the cooling jet within l/8 inch of the junction. Data taken from an actual computer printout are shown in figure 1X-6. Here the thermocouple was initially cooled to 22400 F. When the cooling was turned off, the thermocouple temperature line (the recorded data rose to 32000 F, as shown by the solid Cooling was reapplied at 3200 F. The recorded data curve). were sent to the computer and curve fit (circular symbols on the From the data curve-fit, data curve). the computer program generated the extrapolated curve (diamond symbols) extending to 3500O F. The extrapolation involves a complex computer program for curve fitting. To estimate the accuracy of the extrapolation, we operated the system at a lower temperature. A combustor run at a lower temperature level, where the thermocouple can go through the full heating curve without burnout, is shown in figure 1X-7. The data of the full heating curve (solid line) were recorded. Then as a check on the computer program, the computer is instructed to curve fit only the first 65 percent of the temperature rise (circular symbols) and then to extrapolate the rest (diamond symbols). Since the full curve has already been recorded, this is an in-situ check of the entire system that can be made just before a series of extrapolated runs. The agreement between the final computed temperature and the measured temperature is better than 1.5 percent, which is good agreement at this level. In conclusion, it looks as if the pulsed-thermocouple probe, which extends both the life and temperature range of thermocouples, will help make the necessary measurements in the new generation of combustors operating at high temperatures and high pressures. The idea of a pulsed-thermocouple system has been around for some time, but advances in data acquisition and microelectronics have finally made it practical. Work is continuing on this system to upgrade probe design, data acquisition, and computer software. OPTICAL MONITORING OF BLADE FLUTTER

A previous paper in this conference publication deals with the work being done on aerodynamic instability of blades, commonly called flutter. If flutter might occur, planned or not, in a fan or compressor stage during testing, it is crucial that it be 312

carefully

monitored.

The most common method of monitoring flutter has been to mount as many strain gages as possible on as many blades as posThe gages are usually located at spots preselected for sible. The problem with the strain-gage technique is maximum strain. that the available slip-ring capacity prevents the application of nearly enough gages to cover all the suspected modes on all In addition, gages did not last long in Lewis enthe blades. gine tests. Although the strain-gage technique has been used at Lewis, we felt that it should be supplemented with something better. An optical blade-tip detection scheme was further developed for Each blade It has some significant advantages. this purpose. This means that a researcher tip is sensed on every revolution. does not have to be lucky enough to have a strain gage on the The optic blade that is fluttering with maximum amplitude. so they provide the dual advantage of a probes are case mounted, noncontacting measurement without any sensors on the rotor. Shown is the System operation is illustrated in figure 1X-8. whose blades are to be monitored. compressor, or fan, From the shaft a once-per-revolution synchronization pulse is picked up. One or more optical probes (two in this illustration) are mounted in the engine casing. Each probe consists of a bifurHalf the fibers conduct light from a cated fiber optic bundle. the other half return light reflected source to the blade tips; from the tips to a photodetector. Both the detectors and the light source are outside the test tank. The inset in figure IX-S shows more detail of the probe installation. A lens focuses the incoming light onto a tiny spot on It also collects light reflected from this spot the blade tip. and directs it onto the end of the fiber bundle. Thus, a pulse of light is generated in each probe for each blade passage. In this illustration, one probe senses the trailing edge of each the other probe senses near midchord. blade tip; The photodetector signals and the once-per-revolution pulse are fed to an electronics box that uses television-like techniques to generate a display on an oscilloscope. Two oscilloscope displays - one during no flutter, tne other during flutter - are shown in figure 1X-9. The idea is to probe location display consists like a television whose horizontal that blade at spot stays still. back and forth measure the time of arrival of the blade at the relative to the once-per-revolution pulse. The of one horizontal line for eacn blade, much raster. On each line is a brightened spot position is proportional to the arrival time of the probe. If the blade is not fluttering, the If the blade is fluttering, the spot moves by a distance proportional to the tip motion 313

caused by the flutter. By using various probe signals and the once-per-revolution be set up to monitor a selected flutter as torsion or bending.

combinations of the signal, the system can mode of the blades, such

The displays depicted in figure IX-9 were taken during monitoring of an F-100 fan for the first torsion mode of flutter. It is obvious from this display that all the blades must be monitored because some blades flutter a lot more than others. At any one time during this particular test, the blade showing maximum flutter on this display exhibited an amplitude five times that of the blade showing maximum flutter based on straingage signals. This system is now used for all Lewis engine testing whenever flutter is a possibility. In some tests we have even been able to detect stalls and rubs from the display characteristics. UNSTEADY PRESSURE ON ROTATING BLADES

Previous papers in this conference publication have presented discussions on fan noise and blade flutter. Both these problems have generated great interest in measuring the unsteady pressure on the surfaces of the rotating blades. Fluctuating pressure (and therefore lift) is a source of both noise and flutter (vibration). A recently completed contract effort with Pratt & Whitney Aircraft to develop improved techniques for measuring unsteady pressure on rotating compressor blades is described here. The objectives of the program were to design configurations for mounting pressure transducers on the blades and then to identify and quantify the sources of measurement error. The mounting configurations included surface mounting of a subminiature, flat pressure transducer and more complex mountings that required a recess or hole in the blade but resulted in less disturbance of the aerodynamic shape of the blade. Figure IX-10 shows examples of two of these mountings on JT15D fan blades that will be tested here at Lewis as part of the forward velocity effects program that is described in paper III, NOISE REDUCTION. Surface-mounted transducers are shown on the second and fourth blades, counting from left to right. The diaphragm of the surface-mounted transducer appears as a white dot near the left end of the mounted assembly. Epoxy-coated lead wires are routed along the blade surface. The white material surrounding the transducer is silicone rubber, which is used to fair the installation to the blade surface. Soft rubber is used for this rather than epoxy in order to isolate the transducer from local strain in the blade. The total thickness of this mounting is about 0.015 inch and this, of course, represents a 314

disturbance of the aerodynamic surface at the point of measureA second configuration is shown on the first and third ment. In this case the transducer is mounted with its diablades. phragm facing the blade and centered over a 0.06-inch-diameter which acts as a pressure tap. This means that the point hole, of measurement is on the opposite surface of the blade. The installation is about 0.018 inch thick on the transducer side of but there is no disturbance on the side of the blade the blade, from which the measurement is made, except for the pressure tap. More detailed information on these mountings and the pressure transducers is available in references 3 and 4. The environmental conditions specified for this work included up to 90 000 g's centripetal acceleration. This is the acceleration expected on the blade of a 20-inch-diameter compressor operating at 18 000 rpm. The temperature was limited to 300 F, which is the maximum operating temperature of the diffused siliFinally, because the interest in this con pressure transducer. work originated with aeroelastic flutter, in which there is condynamic strain up to +lOOO microsiderable blade vibration, strain was included as part of the environment. -The transducers and mountings survived in the environment, and measurement errors for a typical application are estimated to be +6 percent of reading due to sensitivity changes and +0.02 psi due to vibraThe 0.02 psi Ts a signal that is intion and dynamic strain. duced in the. transducer and that cannot be differentiated from the pressure signal. This signal determines the minimum unThe 6-percent sensteady pressure signal that can be measured. sitivity error is a combination of a sensitivity change due to the rotating environment and uncertainty in calibrating the mounted transducers. Periodic recalibration after mounting is advisable; appreciable calibration shifts were encountered in mounting and testing these transducers. One further conclusion from this work is that, for most applicait is probably not feasible to measure steady-state prestions, sure with blade-mounted transducers of the configurations used in this work. Zero shifts of up to 2 psi were measured under conditions that were aimed at minimizing zero drift. It is expected that normal test conditions would lead to totally unacceptable zero drift errors. Further detail on this work is available from references 3 and 4.

DRAG FORCE ANEMOMETER Over the past few years, there have been more and more requests for instrumentation to make dynamic measurements in turboTwo parameters of particular interest are the dynamic machines. flow angle and the dynamic flow rate, which is related to dynam315

ic velocity developed.

nead. It is

A device to make such measurements called a drag force anemometer (ref.

has been 5).

The The operation of the device is depicted in figure 1X-11. measuring element is a cantilevered beam exposed to the flow. The drag force due to tne flow over the beam generates a bending moment that is measured with strain gages at the base of the The equation in figure IX-11 shows that tne drag force is beam. the product of the drag coefficient, the exposed area, and the velocity head. The drag coefficient varies with flow angle, as For flow angles from auout 61)o to 900, snown in this plot. This means that when the beam the drag coefficient is constant. to the flow (as shown in the is approximately perpendicular the anemometer signal is only a function of velocity sketch), wnen the flow angle is beiow 4Uo, the drag Conversely, head. Thus, wnen the beam coefficient is proportional to flow angle. is approximately parallel to the flow, the anemometer can be used to measure flow angle. A number of drag force anemometers have been buiit with both A probe that has a silicon beam pursilicon and steel beams. chased from a semiconductor transducer manufacturer is depicted The beam is mounted on a pedestal inside the in figure 1X-12. the strain gages. hollow tip of the probe so as to protect The beam is 10 mils thick and the exposed part of the beam is 60 The probe support is l/4 incn in diameter. mils on a side. The advantages of the silicon beam are its small size and its The natural frequency of the silicon high-frequency response. this gives a flat frequency response to beam is 40 kilohertz: The disadvantages of the silicon beam are a-bout 8 kilohertz. by particles entrained in the its susceptibility to breakage Steel beams with flow and its relatively poor zero stability. conventional metal-foil strain gages minimize these problems, is roughly half that of silicon. but the frequency response Different designs of the drag force anemometer have been used in Fluctuating flow was measured downa variety of applications. stream of the F-100 fan during flutter testing. By using the along with other measurements, it is posanemometer signals, sible to identify flutter frequencies and rotating nodal patA number of anemometer probes were also used in a comterns. One of these was a highpressor dynamics experiment. sensitivity beam to measure inlet flow at a velocity as low as 100 feet per second. Other measurements included dynamic flow angles downstream of a turbine and turbuience intensity. In summary, it appears that the drag force anemometer is a very versatile addition to the group of instruments used for dynamic The output of the anemometer is linear with flow measurements. The device can measvelocity head up to a Mach number of 0.6. ure transient flow even with flow reversals. It can measure dy316

namic flow angle over a range high-frequency response. LASER ANEMOMETER

of +4Oo and,

finally,

it

has

The laser anemometer (sometimes called the laser Doppler velocimeter) is used extensively in mapping flow velocity fields. In it used to be enough to obengine and engine component testing, tain the velocity map upstream and downstream of the stage. Mathematical models were then used to get the map within the This is no longer good enough. stage. Real data are needed beThe laser anemometer has tween the blades to verify the models. recently been applied to the acquisition of the velocity map Lewis within the rotating stage, that is, between the blades. has developed a system that obtains these data rapidly and efficiently. A previous paper in this conference publication gives some of The major optical comthe results obtained with this system. It works in ponents of the system are shown in figure 1X-13. the following way. The laser beam is split into two beams that, by means of a lens, are caused to intersect at a location within the compressor blade row. This location is called the probe volume. Within this tiny volume the two beams interfere to produce a series of parallel planes that are alternately bright and dark. These planes are perpendicular to the plane of the figure. Tiny seed particles, a micrometer or less in diameter, are intentionally injected into the flow upstream of the stage. Laser light is scattered from any of these particles that happen to pass through the probe volume. Some of this scattered light is collected by tne receiving optics and focused on the photomultiplier tube. As the particle traverses the series of light and dark planes, the scattered light flashes on and off. The frequency of this flashing is proportional to the particle velocity component perpendicular to the planes. Thus, for each particle that happens to pass through tne probe volume, a burst of signal comes from the photomultiplier with a frequency proportional to this velocity component. Since it is realiy airflow that is being measured, the particles must be small to follow the flow. The hardware in the Lewis instrument test facility is shown in figure 1X-14. The two beams cross to define the probe volume within the blade row. Recall that the data are ootained from this tiny probe volume, which is essentially a point in space. As the stage rotates, data can be obtained from all the points on a circumferential line as they pass through the probe volume. To get at other circumferential lines, stepper motors, as shown in the figure, are used to reposition the whole optics package and hence the probe volume. Thus, tne whole stage can be mapped. 317

A simplified biock diagram of tne system is shown in figure 1X-15. Shown is the rotor with its blades and a once-perrevolution pickup for synchronization. Each time a particle passes through the probe volume, a burst of signal comes from tne photomultiplier. Tne laser anemometer processor converts the signal burst into a digital number proportional to velocThis number is fed to the minicomputer. ity. The trick is to tell the computer where, on the circumferential this velocity number was obtained. segment, To solve this problem, we invented an electronic device that we call an electronic shaft-angle encoder. It always has at its output a digital number that is proportional to the shaft angular position relative In other words, it gives the to the once-per-revolution pulse. position on the circumference at any time. Each time a velocity event occurs, as signaled by the processor box, botn the velocity and the angle are fed into the computer. These velocityangle data pairs are stored, sorted, and processed by the computer. is a CRT terminal, A key component to the experimenter in practically tell whether tne data are any good which presents graphics data real time. From it he can and what he should do next.

The equipment in the control room is shown in figure 1X-16. Shown are the minicomputer, the laser anemometer signal processor, the electronic shaft-angle encoder, and the operator working at the graphics CRT terminal. The optics system installed in the compressor facility is shown in figure 1X-17. The laser beam travels through the optics and enters a specially contoured window in the compressor housing. Tne primary contribution to the state of the art witn tnis system is rapid data acquisition. Data are accepted over a whole circumferential segment at whatever time they arrive. Previously reported systems gated the electronics to accept the data from only one point at a time until enough velocity events were recorded for good statistics. This system, as currently accepts the data from 1000 different locations at operating, rates from 5000 to 150 000 points per minute. Thus, it is orThis, ders of magnitude faster. coupled witn the graphics to guide decisionmaking, cuts running time significantly. CONCLUDING REMARKS This paper has illustrated the application of optic and microelectronic technology to propulsion instrumentation. Other topics as well as otner technology fields, could have been coverThe point to be made is that the solution to measurement ed. problems depends heavily on drawing from a wide variety of technologies. 318

REFERENCES 1. Advanced Optical Blade Tip Clearance Ford, M. J.; et al.: Measuring System. (PWA FR-10200A, Pratt & Whitney Aircraft NASA CR-159402, 1978. Group; NASA Contract NAS3-20479.) Glawe, George E.; Will, A New Approach to the perature Measurements. Grant, H. P.; and Lanati, Measuring the Dynamic (PWA 5558-12, Blades. Contract NAS3-20296.) Herbert A.: and Krause, Lloyd N.: Pulsed Thermocouple for High Gas TemNASA TM X-71883, 1976. Instrumentation for G. A.: Pressure on Rotating Compressor Pratt & Whitney Aircraft Group; NASA CR-159466, 1978. and Lanati, Compressor G. A.: Blades. Miniature Measuring NASA Drag

2.

3.

NASA

4.

Englund, D. R.; Grant, H. P.; Unsteady Pressure on Rotating TM-79159, 1979. Krause, Force

5.

Lloyd N.; and Fralick, Gustave C.: NASA TM X-3507, 1977. Anemometer.

319

.- .__...-.-

.-- ..- - ---.

BLADE

TIP

CLEARANCE

OBJECTIVE DEVELOP LASER PROBE SYSTEM TO MEASURE INDIVIDUAL BLADE TIP CLEARANCE

PERFORMANCE RANGE OF MEASUREMENT ACCURACY RESOLUTION TEMP 5 TO 125 MILS 2 MILS 1 MIL UP TO 1900 F
cs-79-1553

Table IX-1

OPTICAL TIP CLEARANCE PROBE

I
320

/ -BLADE
cs-79-1552

Figure IX-I

OPTICALTIP CLEARANCE SYSTEM

-A

I -PROBE Figure IX-2

1
cs-79-1555

BLADE-TIP-CLEARANCE PROBEMOUNTED IN TESTFIXTURE

Figure IX-3

321

PULSED-THERMOCOUPLE SYSTEM
COOLING NOZZLECOOLING GAS\ RECORDED PORTlON OF HEATING CURVE1 I Ai49 -.PROBE BODY

**bHERMOCOPLE HOT # JUNCTION GAS GAS TEMP MELTING POINT OF THERMOCOUPLE -COMPUTER CURVE

t TEMP

REAPPLIED
cs-79-1775

TlMEFigure IX-4

THREE-ELEMENT PULSED THERMOCOUPLE PROBE

COOLING GAS PORT 7

WATER-COOLEDSHELLd

\ THERMOCOUPLE JUNCTION1

Figure IX-5

322

PULSED-THERMOCOUPLE HIGHTEMPERATURE RUN

0

o -

DATA CURVE FIT EXTRAPOLATED CURVE DATA CURVE r3500 F

REAPPLIED

-.2

.2

.4 .6 TIME, set

.8

1.0

1.2

cs-79-1777

Figure IX-6

PULSED-THERMOCOUPLE LOWTEMPERATURE RUN

0
0

DATA CURVE FIT EXTRAPOLATED CURVE DATA CURVE

,r 2958 F

-.2

.2

.4

.6 TIME. set

.8

1.0

1.2
cs-79-1776

1 1.4

Figure IX-7

323

OPTICAL DETECTION OF BLADE VIBRATION AMPLITUDE

BLADE-,

I
!E,,; ONCE1 R SENSOR LlBtK

SIGNAL BUNDLE SIGNAL BUNDLE PHOTODETECTORS

OPTIC BUNDLES; .I/ L BULKHEAD

URCE

t
I

-SIGNAL CONDlTlONlNG &DISPLAY

CS-79-1633

Figure IX-8

DISPLAY DURING QUIET RUNNING CONDITION Figure IX-9

DISPLAY DURING FLUllER

CS-79-1634

324

PRESSURETRANSDUCERS MOUNTED ON JT15DFANBLADES

Figure IX-10

DRAG FORCE ANEMOMETER

DRAG = CD x A x 112 pV2

60 30 FLOW ANGLE, a, ckg

90

figure IX-11

325

DRAG FORCE ANEMOMETER PROBE

\ \

I- 114.in. DIAM
CS-79-1778

Figure IX-12

PHOTOMUl (PMITUBE

', VELOCITY MEASURED

x Figure IX-13

COMPRESSOR ROTOR
cs-79-1632

326

Figure IX-14

CS-79-1610

ONCE/REV SIGNAL -\ OUTPUT SIGNAL t /

LA . PROCESSOR (VELOCITY) , I

-0

CS-79-1631

Figure IX-15

327

Figure IX-16 -

Figure IX-17

328

X.

CONTROL John

TECHNOLOGY R. Szuch

National

Aeronautics and Space Administration Lewis Research Center

Today propulsion control technology is advancing at a tremendous there are dramatic changes taking place pace, and, as a result, both in the things one can expect the control system to do and in the equipment that is becoming available to do the controls This paper focuses on gas turbine engine controls by provjob. iding an overview of engine control technology and a glimpse of The role of the government, where this technology is heading. in advancing this technology is also and NASA in particular, discussed. CONTROL REQUIREMENTS The basic functions that have to be performed by the engine control system include starting the engine, accelerating and decelerating the engine from one thrust level to another, regulating or governing the engine speed to maintain a desired thrust, providing some form of overspeed protection, metering and shutting off the fuel flow to the engine, and providing control for other engine variabl such as compressor vanes and bleed. These functions have been and still are, for the most part, being performed by hydromechanical control systems such as the one shown in figure X-l, which is a cutaway of a Pratt & Whitney F-100 engine fuel control. This hydromechanical control is similar to an automotive fuel control system in that it uses cams, springs, levers, and valves to deliver the right fuel flow for a given combination of throttle setting, engine speed, and ambient conditions. The hydromechanical units are rugged, dependable, and generally capable of performing the basic control functions. However, there are practical limits on the number and types of calculations one can reasonably expect from a hydromechanical computer before paying a significant penalty in control weight, cost, performance, and reliability (refs. 1 and 2). One factor that is dictating a search for new ways to control engines is the continuing trend (fig. X-2) toward more complex and demanding propulsion systems. If the number of controlled variables is used as a measure of system complexity, it can be seen that over the last two to three decades we have gone from simple turbojets like the J47 with only fuel flow to control to today's 329

afterburning turbofans like the F-100 which high-performance, This trend, which can be attribhas seven control variables. uted to requirements for increased thrust to weight ratios, redand lower life-cycle costs, can also be uced fuel consumption, And if one looks ahead seen in the commercial engines as well. to the next generation of variable-cycle engines and engines for VTOL propulsion (refs. 3 and 4), there probably will be 10 or more variables that have to be controlled. DIGITAL ENGINE CONTROLS

It is generally accepted that hydromechanical controls will be unable to satisfy the multivariable control requirements of fuAs a result, the Government and the ture propulsion systems. propulsion industry are aggressively pursuing the use of electronics for both military and commercial enqine control applications (refs. 5 and 6). In addition to multivariable control, electronic engine controls offer new opportunities for automatic thrust rating, temperature limiting, self-trim capability, engine condition monitoring, fault tolerance, and communication with the other aircraft systems. Figure X-3 illustrates how this move from hydromechanical controls toward electronic controls has been progressing. For military engines, we have already seen the introduction of control systems having some form of electronic supervisory trim capability - either analog or digital. Figure X-4 shows the F-100 engine controls. A hydromechanical control unit performs all the basic control functions, while an engine-mounted digital computer trims the fuel flow and exhaust nozzle area commands ,to increase thrust and to give better thrust regulation at high compower conditions throughout the flight envelope. The trim puter, or EEC as it is referred to, also has full authority control of the fan inlet guide vanes. After demonstrating the suitability of digital electronics for trimming hydromechanical controls, it would seem logical to turn more and more of the control tasks over to the digital computer. As shown in figure X-3, successive steps might include a full authority digital control with a simple hydromechanical backup to provide flyback capability in the event of an electronics failure and, finally, a full-authority digital control with backup provided by complementary or redundant electronics. This same evolutionary process is expected to occur with commercial engine controls but at a somewhat slower pace due to the simpler engines. The new Boeing 757 and 767 aircraft will have hydromechanical controls with supervisory digital electronics. There is a possibility that commercial engine manufacturers will then be able to go directly to all-electronic controls. A lot depends on the outcome of the Navy's FADEC program and the ongoing reliability flight tests of dual channel electronic controls
330

on the 727 aircraft (ref. 6). programs such as In any case, 7) and the Energy Efficient Engine program (refs. QCSEE (ref. and 9) indicate that digital electronic controls will play a prominent role in future commercial aircraft. DIGITAL CONTROL RELIABILITY

While the future for digital engine controls looks bright, the question of their reliability remains a major stumbling block to their widespread acceptance for man-rated systems. Of particular concern are future VTOL applications where the engine controls will become, in effect, the flight controls. Figure X-5 gives a quantitative look at the reliability of both the propulsion and flight controls on modern fighter-attack and commercial transport aircraft. Reliability is expressed as the It can be mean time between control system failures (MTBF). seen that the commercial transports exhibit higher control reliThis is due, in part, to ability than do the military aircraft. the hydromechanthe simpler commercial engines. As an example, ical control system on the Pratt & Whitney JT9-D engine, which powers many of today's wide-bodied transports, has an MTBF in the F-100 engine, with excess of 10 000 hours. By comparison, its supervisory digital control, exhibits an MTBF of about 1000 hours. Later this paper will describe a number of technology programs aimed at bridging this reliability gap and thus making possible the development of full-authority digital controls for more complex commercial engines. Figure X-5 indicates future, that the reliability of flight controls is orders of magnitude higher than propulsion control reliability (refs. 10 and 11). This poses a significant problem in the development of electronic controls for future VTOL propulsion systems since it is reasonable to expect that flight control reliability requirements will be imposed on the VTOL propulsion controls. The current approach to achieve this high reliability in flight controls is to use many levels of redundancy. this approach may However, result in severe penalties in control system cost and weight when considered for the engine control application. Therefore, it is expected that a significant number of improvements in control technology will be required if electronic controls are to become a reality for VTOL aircraft. CONTROL TECHNOLOGY PROGRAMS A number of technology areas have been identified as having a significant effect on the reliability of electronic controls. One is the computer itself. This includes such things as computer design, parts selection, fabrication, packaging, and test procedures. Another area is computer software. Given the digital computer and its capabilities, one can reevaluate how con331

trol laws are designed and how one can use digital logic to provide the desired level of reliability without resorting to many levels of redundancy. The third area involves computer interHere it is imface components such as sensors and actuators. portant to start thinking in terms of digital-compatible components that do not require a lot of signal conditioning electronics that can adversely affect the system reliability. And finally, there is, control integration. While one usually thinks in terms of the performance benefits of integrated controls, communication between the various aircraft computers can improve control system reliability since it opens up possibilities for computer health monitoring and the trading off of control functions if faults are detected. its NASA, in fulfilling and technology programs this research, has been obligations to fund high-risk research and to provide unique facilities for active in all of these technology areas.

Both separately funded and jointly funded NASA-DOD programs have been conducted to make maximum use of available resources and to Each program is contributing to avoid duplication of effort. the evolution of reliable electronic control systems. ADVANCED TECHNOLOGY DEMONSTRATOR ENGINE (ATDE) PROGRAM

The Army's Advanced Technology Demonstrator Engine (ATDE) program is a 4-year effort to provide a technology base for developing an 800-horsepower turboshaft helicopter engine in the mid-1980's. As part of the ATDE program, the engine developers (AVCO Lycoming and Detroit Diesel Allison) are using engineEven though the full-authority electronic controls. mounted, ATDE only has two control variables, fuel flow and inlet guide the Army recognizes the potential payoff from electronic vanes, controls in terms of improved performance and lower costs. The computer technology being used has evolved from a number of Army programs (refs. 12 and 13). Figure X-6 shows the internal construction of the hybrid (part-digital, part-analog) control unit developed by Chandler-Evans that was tested in 1976 on the Pratt The STAGG tech& Whitney Small Advanced Gas Generator (STAGG). Also nology subsequently formed the basis for the ATDE design. shown in figure X-6 are a burner pressure sensor and detector circuitry for use with an optical pyrometer to measure turbine A more advanced version of the control is blade temperatures. being developed for the ATDE and will be undergoing engine tests in 1979. QUIET, CLEAN, SHORT-HAUL EXPERIMENTAL ENGINE PROGRAM

Short-Haul Experimental Engine The NASA-sponsored Quiet, Clean, (QCSEE) program was covered in detail at the 1978 NASA Powered 332

It should be noted that General ElecLift Conference (ref. 7). tric did develop a full-authority digital control for the QCSEE by using the state-of-the-art OTW engine. General Electric, computer technology and sophisticated software techniques, was able to show in sea level tests at Lewis a number of performance and operational benefits of using digital controls for poweredThese included thrust-rating control, lift propulsion systems. rapid thrust response (essential for the powered-lift application), failure identification and corrective action software, and an aircraft computer interface, engine condition monitoring, which would allow integration of the engine and aircraft controls. FULL-AUTHORITY DIGITAL ELECTRONIC CONTROL PROGRAM

Probably the most ambitious of the computer hardware-oriented programs is the Navy's Full-Authority Digital Electronic Control (FADEC) program (ref. 5). which began in 1976, In this program, full-authority digital control system is being a dual-channel, developed to operate advanced variable-cycle engines (VCE) for future supersonic aircraft. In this case, the variable-cycle Both engines will have as many as 13 controlled variables. Pratt & Whitney and General Electric are designing, fabricating, and testing FADEC systems. The stated FADEC goals are to reduce control life cycle costs by 30 percent, control acquisition costs by 25 percent, control weight by 25 percent, and control maintenance time by 50 percent, and to achieve an MTBF of 2000 hours. This reliability goal represents a significant improvement relative to the F-100 EEC experience and the reliability that could be expected from a QCSEE-type computer operating the more complex VCE. This is illustrated in figure X-7 where the computer (or processor) MTBF is plotted against control complexity. Complexity is expressed as a function of both the number of control inputs and the number of control outputs. In this case, the VCE would be represented by a complexity of about 350. Data are plotted for both a hydromechanical computer and a QCSEE-type computer utilizing medium-scale integration of circuits and multilayer circuit board construction. From figure X-7 two things are obvious. First, while hydromechanical controls may be preferred for less complex control tasks, they are out of the question for the VCE task. Second, using QCSEE circuit technology would result in less than half the desired FADEC computer reliability. As a result, both Pratt & Whitney and General Electric have had to take advantage of the latest advancements in large scale integration of circuits and advanced circuit board construction techniques to try to reduce the number of potentially troublesome circuit interconnections. Indications are that the FADEC design goals will be met or exceeded. Altitude tests of the Pratt & Whitney FADEC are being conducted at the NASA Lewis Research Center on an F-401 engine; 333

engine :1979.

tests

of

the

General

Electric

FADEC are

to

begin

in

late

MULTIVARIABLE

CONTROL SYNTHESIS

PROGRAM

The Multivariable Control Synthesis (MVCS) program was sponsored by the NASA Lewis Research Center and the Air Force AeropropulThis joint program began in 1975 and was comsion Laboratory. In the MVCS program, it was decided to pleted early in 1979. abandon the idea of merely implementing conventional control laws in the digital computer and instead to exploit the capabilities of the computer to handle sophisticated control algorithms such as those coming out of the so-called modern or optimal conThis marked the first application of Linear trol theories. 14) to the design of conQuadratic Regulator (LQR) theory (ref. trols for a modern turbofan engine - in this case, the Pratt & This also involved a computer-aided design proWhitney F-100. cess whereby the control designer could specify a design criteria or measure of performance and have a high-speed digital computer then compute the set of optimum gains for the control. (Vt) was contracted to develop the control Systems Control, Inc. logic which was then programmed on a NASA minicomputer (ref. The control was then verified using a real-time hybrid 15). computer simulation of the F-100 engine (refs. 16 and and 17). Figure X-8 shows a photograph of the NASA hybrid computer system This approach to software that was used in the MVCS studies. verification has proved to be very convenient and cost effective in that it allows one to debug, modify, and fully evaluate the Following the hybrid control software prior to engine testing. the same control hardware and software were computer studies, used to remotely control the operation of an F-100 engine in the Figure X-9 shows a NASA Propulsion Systems Laboratory (PSL). controls engineer monitoring the MVCS engine tests from the hyFigure brid facility where the control computer was located. X-10 shows the F-100 engine installed and instrumented in the The engine tests were tremendously sucNASA altitude chamber. they represented a free world first for the LQR control cessful; concept. CONTROL INTERFACE COMPONENTS

Recognizing the importance of the control interface components the NASA Lewis Research Center to the overall control system, began a long-range research program in 1975 aimed at advancing digital-compatible sensors and the state of the art in reliable, Several contracts are underway to promote research actuators. Central to these activities is the extensive in these areas. use of fiber optics with its inherent advantages of simplicity, and compatibility with the digital compulow weight, passivity, ter. Devices being considered for the engine control applica334

tion include pulse-width-modulated fuel valves, optical tachometers, tip clearance sensors, and optical temperature sensors. Figure X-11 shows a photograph of the Fabry-Perot optical temperature sensor being developed for NASA by Rockwell. The sensor is quite small. Light from a source in the computer enters the sensor through a fiber optic bundle. The thermal expansion of a small brass ring at the tip of the sensor changes the size of an internal air gap and produces a filtering effect on the light which is reflected back to a receiver in the computer (also through a fiber optic bundle). The frequency content, or color of the reflected light, is a measure of the temperature. Figure X-12 shows results from laboratory tests of this sensor. Distinguishable color patterns have been produced over a 140 temperature range. The next steps in the development of this promising concept will be to extend the temperature range and come up with a suitable detecting and decoding scheme to convert the color patterns to digital words that can be used in a control. INTEGRATED PROPULSION CONTROL SYSTEM (IPCS) PROGRAM

In 1973, the Air Force and NASA initiated the Integrated Propulsion Control System (IPCS) program (ref. 18). The objective was to demonstrate in flight the benefit of integrating engine and air inlet controls on a high performance supersonic aircraft. Figure X-13 shows the F-111 aircraft that was used for the IPCS flight tests. An onboard digital computer provided fullauthority control of the inlet and the TF-30 gas generator, afterburner, and exhaust nozzle. The IPCS flight tests resulted in improved aircraft maneuverability, better steady-state and transient thrust performance, lower fuel consumption, and expanded stall-free operation. INTEGRATED PROPULSION-AIRFRAME CONTROL (IPAC) PROGRAM

While the IPCS program showed that integrating the inlet and engine controls could improve performance, a number of studies have indicated that even more benefits can be derived by integrating all the aircraft control systems, including the engine, inlet, and flight controls. Recognizing a need for research in this area, NASA, the Navy, and the Air Force have joined to sponsor the Integrated Propulsion-Airframe Control (IPAC) program. This program is aimed at developing a flexible flight research facility for conducting flight evaluations of concepts related to controls integration. In addition to these integrated control studies, the facility will be ideal for conducting research in energy management, full-authority digital electronic controls, engine diagnostics, failure accommodation, electronic and optical data bus technology, and other advanced control hardware and software. The IPAC system will be built on the 335

NASA F-15 aircraft (fig. X-14). The IPAC schedule calls for initial flight tests of the Navy FADEC control system in 1983 with integrated multivariable control research tests to follow. CONCLUDING REMARKS This paper has dealt with the increasing demands on propulsion control systems and a number of ongoing programs aimed at develResults to date indicate oping the required technology base. that significant advancements in computer technology are being made which should provide future control designers with an alThis will permit using most unlimited computational capability. sophisticated control algorithms, like LQR, and other software such as fault accommodation, engine diagnostics, and features, Also, there is every reason to integrated control functions. believe that the work going on in advanced sensors and actuators will result in a number of simple, yet accurate and reliable, These will then mesh with the improved computer harddevices. ware and software to form electronic control systems to provide better performance, higher reliability, and lower costs. REFERENCES 1. Kuhlberg, J. F.; and Newirth, Propulsion Control System 74-1068, Oct. 1974. 2. D. M.: - Problems The Digital and Payoffs. Electronic AIAA Paper

Sevich, G. J.; and Newirth, D. M.: Economic Benefits Digital Electronic Propulsion Controls for Advanced SAE Paper 760508, May 1976. cial Aircraft. Gertsma, L. W.; VTOL Transports. and Zigan, S.: ASME Paper Propulsion 73-GT-24, System for Apr. 1973.

of CommerResearch

3. 4.

Problem Areas for Lift Fan Propulsion for Civil Lieblein, S.: Presented at the Deutsche Gesellschaft Fuer VTOL Transports. Symposium on VTOL Propulsion (Munich, Luft-und Raumfahrt. West Germany), Oct. 22-23, 1970.

5. Barclay, B. A.; Lenox, T. G.; and BOSCO, C. J.: Full Authority of Next Generation Digital Electronic Control - Highlights ASME Paper 78-GT-165, Aor. Propulsion Control Technology. 1978. 6. Kamber, P. W.; Controls for 1974. Powered-Lift and Welliver, A. D.: Commercial Aircraft. Propulsion. Electronic AIAA Paper Propulsion 74-1065, 1978. Oct.

7. Quiet, 336

NASA CP-2077,

8.

Energy Efficient Engine - PrelimiJohnston, R. P.; et al.: (R78AEG510, General nary Design and Integration Studies. NASA CR-135444, Electric Co.: NASA Contract NAS3-20627.) 1978. Gray, D. E.: Energy Integration Studies. Group; NASA Contract Boudreau, J. A.: ccv. AIAA Paper Helfinstine, ability
Systems.

9.

Efficient Engine - Preliminarv (PWA-5500-18, Pratt b Whitney NASA CR-135396, NAS3-20628.)

System

Design and Aircraft 1978. Design for

10. 11.

Integrated Flight Control 76-941, Sept. 1976.

-902445. 12.

R. F.; Montague, and Redundancy Study Honeywell-21718-FR, )

L. L.; and Seller, G. L.: Relifor Electronic Flight Control Honeywell, Inc., 1972. (AD-

White, A. H.; and Wills, D. F.: Advanced Engine Control ProFabrication and Test of Electronic Engine Congram - Design, System for Small Turboshaft Engines. ~-492-31, Chandler trol (USAAMRDL-TR-72-59, AD-758173.) Evans, Inc., 1972. Hearn, R. J.; Cole, M. A.; and White, A. H.: Advanced Engine Electronic Equipment. Control Program - Using Computerized R-492-45, Colt Industries, Inc., 1973. (USAAMRDL-TR-73-81, AD-773660.) Bryson, Publ. A. E.; and Ho, Co., 1969. Y.: Applied Optimal Control, Blaisdell

13.

14. 15.

Dehoff, R. L.; et al.: FlOO Multivariable Control Synthesis Program. Vol. 1: Development of FlOO Control System. AFAPL-TR-77-35-VOL-1, 1977. (AD-AO52420.) Szuch, John R.; Seldner, Kurt; and Cwynar, ment and Verification of Real-Time, Hybrid tion of FlOO-PW-100 (3) Turbofan Engine. David S.: DevelopComputer SimulaNASA TP-1034, 1977.

16.

17.

Szuch, John R.; et al.: FlOO Multivariable Control Synthesis Program - Evaluation of a Multivariable Control Using a Real-Time Engine Simulation. NASA TP-1056, 1977. Lampard, grated 1975. G. W. N.; Propulsion and Batka, J. J.: Development of an InteControl System. AIAA Paper 75-1178, Sept.

18.

337

CUTAWAY OFF-100HYDROMECHANICALCONTROL

Figure X-l

TRENDS IN CONTROL COMPLEXITY

0 MILITARY VARIABLE CYCLE ENGINES CONTROLLED VARIABLES

cs-79-1421

TIME Figure X-2

338

EVOLUTION OF ENGINE CONTROLS

r FULL AUTHORITY / \ ELECTRONIC, / \ H. M. BACKUP

TIME -

HYDROMECHANICAL

TObAY FULL AUTHORITY ELECTRONIC, ELECTRONIC Figure X-3

cs-79-1419

HYDROMECHANICAL ENGINE CONTROL WITH ELECTRONIC TRIM -F-l00 HYDROMECHANICAL ELECTRONIC TRIM COMPUTER

Figure X-4

339

CONTROL SYSTEM RELIABILITY

t FIGHTER/ATTACK
COMMERCIAL TRANSPORT PROPULSION CONTROLS FIGHTER/ATTACK t Figure X-5
(35-79-2372

STAGG

ELECTRONIC

CONTROL

CONSTRUCTION

-.-..

.,..

C5-79-1546

Figure X-6

340

FADEC COMPUTER RELIABILITY

1OOooO 80 Ooo 60 ooo COMPUTER MTBF, hr 40 ooo LSIlHY BRID CONSTR-, ,-MSIlMULTILAYER BOARD CONSTR FADEC GOAL HYDROMECHANICAL

-. j
100

.a

QCSEE --A 1 400

cs-

200 300 CONTROL COMPLEXITY (INPUTS + INPUTS x OUTPUTS + OUTPUTS) Figure X-7

LeRC

HYBRID

COMPUTING

SYSTEM

Figure X-8

341

F-100

MULTIVARIABLE

CONTROL

TESTS

Figure X-9

F-100

TURBOFAN

IN LeRC ALTITUDE

CHAMBER

Figure X-10

342

COMPLETEFIBER-OPTIC FABRY-PEROT. TEMPERATURESENSOR

Figure X-11

FABRY-PEROT TEMPERATURE SENSORTEST RESULTS

730c

270 c

290 c

340c

310c

Figure X-12

343

INTEGRATED PROPULSION CONTROL SYSTEM

COMB I NED I NLET, ENGI NE EXHAUST NOZZLE CONTROL

l GREATERMANEUVEfZABILtTY *MAXIMUM THRUST PERFORMANCE l IMPROVED FUELCONSUMPTION . EXPANDED STALL FREEOPERATICft

a-79-1 545

Figure X-13

NASA

F-15 IPAC RESEARCH

AIRCRAFT

/
Figure X-14

cs-79-1539

344

Xl. Albert

SUPERSONIC

PROPULSION

TECHNOLOGY E. Stitt, Jr.

G. Powers, Robert E. Coltrin, Leonard Richard J . Weber, and John B. Whitlow, National Aeronautics and Space Administration Lewis Research Center

in the commercial supersonic At the close of the U.S. attempt both the government and transport market in the early 1970's, aircraft industry recognized that significant technical advancements would be required to make a second generation supersonic aircraft economically viable and environmentally acceptable. a limited effort to advance Consequently, in 1972, NASA initiated The intent of the Supersonic Cruise Resupersonic technology. search (SCR) program is to investigate areas requiring new and improved technology that will lead to substantial improvements in and cost effectiveness. performance, social acceptability, The Langley Research Center, the lead center for the overall SCR program, is working closely with three airframe contractors, BoeReference aircraft (fig. ing, McDonnell Douglas, and Lockheed. X1-1) configurations are being studied that show improvements in range, payload, gross weight, and, hence, cost effectiveness. As indicated in figure X1-2, these improvements are results of recent advancements in structures and materials (including manufacairframe aerodynamics, system integration and turing techniques), advanced control concepts, and propulsion, which is obviously the major environmental culprit. To adequately cover the progress of Langley Research Center is planning November 1979. However, since this conference, this report concentrates closely related disciplines. PROPULSION SYSTEM STUDIES In 1971 the national SST program was still in progress (fig. XIBoeing was designing a preprototype airplane that was pow3) ered by a General Electric afterburning turbojet. Full-scale engines had actually been built and were being tested. However, in retrospect, this engine was borderline in performance and was a poor neighbor in terms of noise and pollution. This was recognized even at that time, and alternative cycles for the production airplane were just starting to receive serious attention when Congress terminated the SST program. 345
l

the overall SCR program, a four-day conference in is primarily a propulsion on propulsion and a few

the

Determining the best propulsion system for a supersonic transport is not a simple matter because of the diverse requirements tnat must be satisfied. Figure XI-4 illustrates a typical flight path and identifies the critical operating requirements. Tne airplane must, of course, be capable of efficient, sustained supersonic cruising. there is the controversial need to limit NO, Also, emissions to avoid damage to the ozone layer. The best supersonic engine tends to be a low-bypass turbofan, possibly even a turbojet, and augmentation is often desirable. On the other hand, for low-noise takeoff and plus at least some amount of occasional climb, a high-bypass-ratio turbofan that is similar modern wide-body transports is preferred. efficient subsonic subsonic cruising, to those used in

These conflicts in engine requirements necessitate a search for some type of compromise design that can best suit the airplane on the average. Alternatively, one can in theory imagine an engine that is so versatile that it can modify its mode of operation during flight so as to best match the requirements of the airplane from moment to moment. This is the concept of the variable cycle engine (VCE). In practice, of course, providing this versatility causes penalties in weight, complexity, and component it is not entirely obvious that the VCE efficiencies; therefore, is really the final solution. As already mentioned, NASA started its SCR program in 1972. Since there were no longer the constraints of a rigid time schedit was possible to reexamine ule of a true development program, the engine situation with a fresh approach. As shown in figure x1-5, an extensive series of propulsion studies was initiated with General Electric and Pratt & Whitney. The goal was to screen all the engine concepts that seemed to have any merit. These included all the usual, conventional turbojets and turbofans plus a wide variety of VCE's. In successive years, the studies weeded out the least promising approaches and analyzed This process prothe survivors in progressively more detail. duced a small number of favored concepts that are presently being studied for such refinements as differences in complexity and maintainability and the technology requirements of the particular components. The relationship of these studies to other phases of the SCR program is indicated in figure X1-6. The propulsion results are fed and they in turn help identify into the Langley system studies, the desirable engine types and the particular versions needed to The engine studies match the various airplane configurations. were also supported by research in the key environmental areas of emissions and noise. By 1976 the engine concepts were understood well enough for large-scale experimentation; also, work has begun

346

on the inlets have evolved.

and nozzles

necessary

to

match

the

engines

that

Some of the surviving engine concepts are illustrated in figures XI-7 and X1-8. Figure XI-7 shows a fairly conventional turbofan; its bypass ratio is so low, however, that this turbofan is pracFigure XI-8 shows the presently favored VCE tically a turbojet. offerings of the two contractors. They are both versions of turbofans and capitalize on the coannular noise benefit that was described in paper III. Because of the unusual features of these two VU's, they were selected for investigation in the large-scale experimental proHowever, this is not intended to suggest tnat one of these gram. is necessarily the engine that would be chosen by industry if an actual development and production program were started in tne Considerations of development risk, cost, reliability, future. technology readiness, and so forth are more properly the responsibility of the manufacturer and his customer than they are of the government. Later sections will describe the operation of these VCE engine concepts and the experimental program that is now underway. The projected performance of these engine types, as applied to a is shown in figure X1-9. typical supersonic transport mission, Airplane takeoff weight for a fixed range and payload is used as The independent parameter is the unsupthe figure of merit. The circled pressed takeoff noise generated by the engines. point on the far right shows the capability of the 1971 turbojet, which is much noisier than the noise constraint that until recently was imposed on new subsonic transports (Federal Air Regulation 36, 1969). The present VCE concepts are competitive with each other and can reach the general level of FAR 36 (1969). It that these calculations are probably should be stressed, however, and the noise goal is not attained with enough rather optimistic, The more conventional engine margin to create much confidence. is inherently somewhat noisier, but if it were fitted with an it could possibly achieve similar advanced jet noise suppressor performance to the VCE's. Despite the significant improvement since 1971 (fig. XI-9), there The FAA has announced its intent are no grounds for complacency. to subject future supersonic transports to the same noise limits that will apply to new subsonic transports of the same as those The latest subsonic limit, shown at the left side time period. of figure X1-9, is about 8 decibels lower than the 1969 rule, which is barely achievable. Consequently, satisfying the new FAA goal poses quite a challenge. To elaborate on this important issue, consider figure X1-10, which shows the noise of a typical VCE at the three FAA measuring points: sideline, community, and approach. Tne open bars pre347

sume that jet noise is reduced by the coannular effect. Compared sideline noise is high, but community and apto the 1969 rule, proach noise are low. after trades are made, the traded Hence, limit is just about met. But this is not good enough for the 1978 goal. One way to improve the situation is to combine the coannular nozzle with a simple mechanical suppressor in the outer stream This is represented by the shaded bars. only. Since the situation is considerably improved, this is a useful direction to A later section will comment further on jet noise suppursue. pressors. A different type of noise is involved in the bar labeiea apAt this flight condition, the engine is throttled back, proach. and a major noise contributor is the fan, rather tnan just the With the present technology the multistage, highexhaust jet. However, pressure-ratio fan used in these engines is very noisy. in this figure it was presumed that a reduction of 15 to 20 decibels was achieved by such techniques as inlet choking and wall These techniques may also be needed at the community treatment. point. The section on inlet design discusses this requirement. It was mentioned when discussing figure X1-9, that the VCE's and the suppressed conventional engines were competitive. But this conclusion is influenced by the mission selected for the airplane for figure X1-9, is an all-supersonic-cruise case. Variwhich, ations in the mission are examined in figure X1-11. Direct operating cost is used as the figure of merit, and what is plotted is the value of that parameter for a conventional low-bypass engine (LBE) relative to the value for a VCE. These data were ail generated by a single contractor to minimize the influence of difFor a full-range, all supersonic ferent technology assumptions. mission (circled point), the two engines yield similar performance (fig. X1-9). However, when one looks at how airplanes are actually used once in the fleet, they rarely fly their maximumrange capability. In fact, the average range for these longdistance vehicles is more like 70 percent of maximum. This slightly enhances the importance of the always present subsonic phases involved with takeoff and descent; hence, the LBE suffers slightly with respect to the VCE. This effect really becomes noticeable if substantial amounts of subsonic cruise become necessary to avoid sonic boom exposure over land areas. After all, not all airports are located next to seacoasts. For the example shown, the LBE is 12 percent more expensive to operate than the VCE at the average-range point. The significance of this factor must again be assessed by the manufacturer and the customer airline when selecting a production engine. To summarize, it seems sion have been achieved both imaginative cycle 348 fair to say that major strides in propulsince the old SST program by means of concepts and improved engine components.

VARIABLE

CYCLE ENGINE COMPONENT PROGRAM

During the propulsion optimization studies, two unique and attractive VCE concepts were identified as possible solutions to the conflicting propulsion requirements for supersonic transports. These VCE concepts incorporate unique components and unusual flowpath arrangements. Certain technologies were identified as critical to the successful development of these engines. To assess the feasibility of the concepts and to develop the most critical of the engine components, the Lewis Research Center initiated a VCE component program in 1976 through a series of conThe next two tracts with Pratt & Whitney and General Electric. sections describe these engine cycles and address a few of the critical technology components. Variable Stream Control Engine

Figure XI-12 is a scnematic showing the basic arrangement of the major engine components of the Pratt & Whitney variable stream Also shown is an illustration of the incontrol engine (VSCE). The engine is a twinverted jet velocity profile at takeoff. spool configuration similar to a conventional turbofan but witn The VSCE derives the added feature of a burner in the fan duct. its name from its ability to independently.control the primary The fan and compressor both have variable and bypass streams. geometry components and are driven by advanced tecnnology turbines. The main burner and duct burner both use low-emissions, high-efficiency combustor concepts, based on NASA's Clean Combustor program. The coannular nozzle provides variable throat areas for the core and fan duct flow and also includes an system. ejector - thrust-reverser The flexibility of this concept to meet the diverse requirements of low jet noise at takeoff and good fuel consumption at cruise can best be illustrated by describing the operation at its tnree most critical operating conditions: subsonic cruise, takeoff, and supersonic cruise (fig. X1-13). At takeoff, the primary stream is throttled to an intermediate power setting while the duct burner is operated at its maximum design temperature. The independent control of the two streams provides the unique inverted velocity profile tnat is needed to take advantage of the coannular nozzle noise benefit. The bypass jet velocity is about 60 to 70 percent higher than tne primary This inverted velocity profile provides a signifijet velocity. cant reduction in takeoff jet noise. At subsonic cruise, the main burner is throttled to a low temperature and the duct burner is turned off. Variable geometry features are used to "high flow" the engine to match the inlet airflow and thus reduce both the inlet spillage drag and the nozzle 349

boattail drag. gine approaches turbofan engine

The velocity profile is nearly flat, and the enthe performance level of a moderate bypass ratio designed strictly for subsonic operation.

At supersonic cruise, the primary burner temperature is increased relative to takeoff, and the duct burner is operated at part power. The resulting velocity profile again is nearly flat for good propulsive efficiency, and this concept then provides a fuel consumption that approaches that of a turbojet cycle designed exclusively for supersonic cruise. VSCE/F-100 component testbed program. - A comparison of the VSCE with Pratt & Whitney's component testbed configuration is shown in figure X1-14. The NASA component testbed program at Lewis began in 1976 to provide a large-scale evaluation of some of the unique and critical components of the selected VCE concepts. To keep the costs to a minimum, the critical component hardware is applied to existing high-technology engines. In the Pratt & Whitney program, the testbed system is designed to provide a large-scale evaluation of two of the most critical technology components of the VSCE - the low-emissions duct burner and the low-noise coannular nozzle. These components are added at the back of an F-100 engine which is used as a gas generator. The F-100 engine has the best potential to simulate the desired exhaust conditions of the VSCE without any major modifications. Low-noise coannular nozzle. - Model nozzle tests are being conducted concurrently to support the component testbed program. An aero/acoustic design procedure for coannular nozzles was applied to several candidate exhaust systems to identify the most attractive nozzle for the VSCE. A schematic of the selected design is shown in figure X1-15. At supersonic cruise, the nozzle is a Two internal conventional convergent-divergent configuration. clamshells are positioned to provide the initial portion of the expansion surface of the ejector shroud. Variable throat areas are provided for both the primary and fan flows. At low-speed the nozzle converts to an auxiliary inlet ejector. conditions, Actuated inlet flaps are opened to admit external airflow into Panels located immediately downstream of the doublethe shroud. hinged doors are translated aft to provide additional area for and the internal clamshells are alined with the inthe ejector, let flow. Floating tail feathers are aerodynamically positioned The clamshells are also to provide the proper exit flow area. used for thrust reversal by rotating them back to the nozzle cenThe reversed flow is then expelled through the open terline. inlet doors. Scale models of this nozzle configuration have recently been Thrust tested in the Lewis 8 by 6 Foot Wind Tunnel (fig. X1-16). performance levels were established for this nozzle design at both subsonic and supersonic speeds. The results of this experimental investigation are shown in figure X1-17. The measured 350

nozzle thrust coefficients are compared to the SCR study goals at three critical flight conditions - takeoff, subsonic cruise, and The supersonic cruise point is simulated at supersonic cruise. the highest Mach number available from the wind tunMach 2.0, The performance at this flight condition is very good. nel. This is encouraging since this is the most critical operating The low-speed performance was dispoint for the exhaust system. especially at subsonic cruise when the measured perappointing, formance levels are from 5 to 6 counts lower than the study Diagnostic tests of the subsonic cruise configurations goals. showed that the lower performance levels were the result of an doors of the ejector. aerodynamic flow separation over the inlet It is obvious that additional work is required to improve tne off-design performance of this ejector nozzle concept. - Segment rig tests are also being Low-emissions duct burner. conducted concurrently to evaluate the performance and emissions An analytical characteristics of duct burner configurations. high-performance duct burner screening study of low-emissions, operating on tne concepts indicated that a three-stage burner, vortex burning and mixing concept (vorbix), offered the best configuration for the testbed engine within its risk and scneduie A schematic of the selected configuration is shown constraints. in figure X1-18. The requirement that the duct burner be capable of operating smoothly over a wide range of fuel-air ratios leads The first, or to the need for a multistage combustor system. stage is sized for stable operation at very low pilot prechamber, fuel-air ratios to provide a soft light and to minimize disturThe combined first two stages bances to the fan operation. are designed to operate at (pilot prechamber and pilot secondary) supersonic cruise, and the third, or high power, stage is designed to operate only during takeoff and transonic climb. Emissions and performance measurements have been obtained for this three-stage vorbix configuration in both a two-dimensional segment rig test and more recently in the F-100 testbed engine. Emissions measurements from these experimental tests at two simtakeoff and supersonic cruise, are ulated flight conditions, The design goals for carbon monoxide (CO) shown in figure X1-19. and unburned hydrocarbons (THC) are based on a combustion effiThese goals are intended only as a stanciency of 99 percent. dard for comparison and are not related to any proposed or estabSince the lished regulation for advanced supersonic aircraft. measured combustion efficiencies are very high (near a value of levels for CO and THC are well below the de1.0) I the emissions sign goals. The design goal for oxides of nitrogen (NOx) is the lowest value that can be obtained with this duct burner conThe measured NO, emissions cept and assumes complete mixing. levels are quite low and very near the goals at both takeoff and It should be pointed out that the duct burner supersonic cruise. only contributes a small part of the overall engine NO, emis351

sions x1-20.

when compared

to

the

main

burner,

as can

be seen

in

figure

In figure XI-20 the projected emissions characteristics of the VSCE have been updated to reflect the duct burner emissions measurements from both the segment rig and testbed engine tests. The data used for projecting tne emissions of the main burner were based on results from the Clean Combustor program, and the main burner was assumed to be a two-stage vorbix concept. Figure XI-20 shows the projected emissions levels for both the airport vicinity and at altitude cruise. The shaded area depicts emissions from the main burner, while the unshaded area shows the emissions from the duct burner. The results at the airport indicate that the engine is capable of meeting the 1984 emissions requirement for the class T5 advanced supersonic transport engines. Tne NO, emissions at high altitude cruise are higher than the proposed Climatic Impact Assessment Program (or CIAP) goal of 3.0. Although the requirements for altitude NO, are if they are constrained to this proposed not yet established, CIAP level, more advanced emissions-reduction technology must be This is particularly true for the employed to meet the goal. main burner, since it produces nearly 90 percent of the total emissions at altitude cruise. NO, One of the major concerns with a duct burner is its starting This problem was problem and how that influences fan stability. addressed in this concept by designing the pilot prechamber stage Soft lights have been to operate at a very low fuel-air ratio. demonstrated with no significant pressure disturbance at the design starting fuel-air ratio of 0.002. Another concern was No acoustic instabilities have been encountered to date screech. after 46 hours of duct burner operation. The cooling system was and the duct burner was able to operate on all three worked out, stages for periods of several hours. As mentioned previously, the performance and emissions characteristics have been very good. Combustion efficiencies are near a value of 1.0. The thrust efficiency at the cruise condition was better than 97 perwhich was well above the design goal of 94.5 percent. cent, Thrust efficiency is a measure of the uniformity of the temperature distribution at the fan duct nozzle exit plane. The temperature profile should be as uniform as possible to produce the However, maximum thrust. the uniformity of the exit temperature profile is limited by the necessity of cooling the nozzle. A thrust efficiency of 97 percent is considered excellent. A photograph of the testbed engine installed in the test stand at East Hartford is shown in figure X1-21. This is a rear view of the testbed with the ejector nozzle removed. A photograph of the exhaust flow with the duct burner lit is shown in figure X1-22. It is evident that the exhaust flow is extremely clean.

352

testbed program at Pratt & Whit- The component Program status. The ney is in the fourth year of a total five-year program. emissions and performance testing of the baseline duct burner There configuration on the testbed engine is nearly completed. are still items to be done in the present program. The model nozzle program is continuing with an effort to improve nozzle performance, especially at the off-design conditions where the Duct burner segment measured performance to date has been poor. rig tests will continue to refine the three-stage vorbix concept This and to initiate work on a more simplified configuration. effort will lead to a second emissions test on the testbed enThe large-scale aero/acoustic test of the coannular nozzle gine. on the testbed engine and the measurement of duct burner combustion noise levels remain to be completed.

the performance and emissions measurements obtained In summary, to date from the component testbed program at Pratt & Whitney Successful operation of these adhave been very encouraging. vanced and unique engine components will remove some of the technical barriers that now inhibit the development of an advanced supersonic cruise aircraft. Double Rypass Engine

identified as The General Electric Company's VCE, previously their most promising concept from the NASA-sponsored SCR propulis a double bypass turbofan - so called because the sion studies, A schematic drawing of this fan has been split into two blocks. Each fan block has its own sepengine is shown in figure X1-23. the terminology double bypass engine. arate bypass duct; hence, Splitting the fan apart gives the designer additional freedom in sizing this component and additional control over bypass ratio. Description of concept. - Some of the unique features of this which help control the flow, are illustrated in figure engine, X1-23. The split fan consists of an enlarged front block, which is designed to accommodate all the airflow required for takeoff lower capacity rear block, or with low jet noise, and a smaller, which has variable inlet guide vanes for addithird-stage fan, They are partially closed to divert flow tional flow control. through the outer duct in double bypass operation, which is the mode shown in the view above the engine centerline in figure XlThis third-stage fan is driven by the high-pressure turbine, 23. as opposed to the more normal low-pressure turbine drive system. This unique drive arrangement provides a better turbine work split between the highand low-pressure spools and allows the fan tip diameter to be reduced because of the higher core shaft The improved work split also reduces the cooling air respeed. quired for the low-pressure turbine because more energy is extracted from the high-pressure turbine. 353

Another unique component is the forward bypass control valve, which is located immediately downstream of the split fan at the In double bypass operation, juncture of the two bypass streams. it varies the discharge opening of the inner bypass duct to balance the static pressure in this duct with that of the outer bypass stream so that they can be combined at minimum loss into a This flow is then brought to the crossover single stream. these struts support the nozzle plug and channel the flow struts: through the hot stream when the inner plug is translated open. This setup provides the inverted flow with the hot gas on tne outside and the cold fan air on the inside for the maximum jet acoustic benefit. The more conventional single bypass operating mode of this engine The outer bypass is shown below the centerline in figure X1-23. duct is closed off by the bypass selector valve, and the inlet guide vanes on the third-stage fan are opened so that this stage In this can accommodate all the flow from the front block fan. the nozzle inner cone is translated aft to block any cold case, The rear bypass flow from discharging through the inner nozzle. located between the crossover struts, are opened control valves, to allow the bypass flow to be mixed with the main engine exhaust flow to provide a uniform jet velocity prof.ile. The simplified schematic drawings in figure XI-24 illustrate the three critical flight conditions referred to in the previous sectakeoff, subsonic cruise, tion about the Pratt & Whitney VCE: At takeoff, the General Electric engine and supersonic cruise. is operated in the double bypass mode with the fan at an overspeed condition to maximize airflow for reduced jet noise. Unengine does not have a duct like the Pratt & Whitney VCE, this burner to provide the inverted jet velocity profile desirable for it relies on the flow inversion of the engine low noise: instead, and fan exhaust streams for this effect. the engine is throttled back and is again At subsonic cruise, The higher byoperated in the double bypass mode at high flow. pass ratio improves the specific fuel consumption at this flight Because the airflow is kept high, the throttle depencondition. Most of the fan discharge dent installation losses remain low. air is mixed with the main exhaust flow ahead of the nozzle to low-velocity jet profile, which is desirable obtain a uniform, for good propulsive efficiency. the engine operates as a conventional sinAt supersonic cruise, The gle bypass mixed-flow turbofan for high specific thrust. bypass and core exhaust streams are mixed ahead of the exhaust nozzle to obtain a single uniform jet for high propulsive efficiency. Early shown 354 acoustic test schematically program. in figure - The early acoustic was assembled X1-25, testbed engine, to assess the

feasibility of some of the most critical technologies relevant to the double bypass VCE concept, especially with respect to the operation of the coannular exhaust system and the unique inverted velocity profile desirable for low jet noise. A series of aero/ acoustic tests with the testbed engine was completed in October 1978 at a General Electric test facility at Edwards, California. Prior tests were conducted with this engine using a conventional mixed-flow exhaust at a General Electric test cell at Lynn, Massachusetts. To save costs and to get an earlier test date, the NASA program started with a Navy-modified YJlOl split fan VCE configuration with double bypass flows. The forward bypass control valve system was then added for the Lynn tests and the coannular plug exhaust system was later added for the Edwards test. The reduced size YJlOl/testbed engine design is obviously very similar to the advanced study engine shown in figure X1-23. The is driven by the low-pressure turbine, third-stage fan, however, and the front block is not enlarged but is the same flow size as losses in the testbed engine might be exthe rear block. Also, pected to be somewhat higher because of the increased bypass stream turning in the vicinity of the forward bypass control valving and the increased fan block spacing necessitated by the In spite of these differences, use of existing YJlOl hardware. takeoff exhaust conditions of the advanced study engine can be simulated by the testbed engine. A photograph of the testbed engine on the Edwards test stand is shown in figure X1-26. A large protective screen is shown ahead In this photograph the engine is set up of a bellmouth inlet. for the initial calibration testing with a baseline single-stream Figure XI-27 is a photograph of the conic mixed-flow nozzle. engine on the stand with the coannular plug exhaust nozzle installation. A laser velocimeter is shown to the side of the exhaust nozzle. It was used to make a velocity survey of the jet Two plume at various positions downstream of the exit plane. laser beams can be seen emanating from the laser velocimeter, Through a combinwhich is on a track-mounted traversing system. ation of different plug crowns and outer nozzle shrouds, tests were conducted with three different outer nozzle radius ratios at a constant throat area. The effect of nozzle area ratio was also investigated by translating the inner cone of the plug noazle,.to Both conic and convervary the exit area of the inner stream. gent-divergent outer nozzle terminations were investigated. Figure XI-28 shows the thrust loss resulting from bypass system In this bargraph, thrust loss is compared at pressure losses. typical takeoff exhaust conditions for both single and double bypass operation. The single bypass case represents an exhaust condition with a mass-averaged jet velocity of 2300 feet per second and an inner to outer area ratio of 0.1. In other words, even though the fan is operated in the single bypass mode, a small amount of bypass air is allowed to flow through the crossover struts and out through the inner plug to obtain an area 355

ratio of interest from the VCE conceptual engine studies. The which was constrained by the cycle limitations of the testbed, necessitated operation in the single byuse of YJlOl hardware, pass mode to obtain this high a level of jet velocity; the conwould obtain this exhaust condiceptual study engines, however, The double bypass case, repretion in the double bypass mode. sented by the second pair of bars in figure X1-28, is for a massaveraged jet velocity of 2000 feet per second at an area ratio of The highest losses would be anticipated at these conditions 0.4. by the higher since the strut flow is at a maximum, as evidenced area ratio. Bypass system losses were of some concern prior to tne Edwards testing because of the rather tortuous flowpath from the fan discharge through the bypass ducting and the crossover struts to the inner plug nozzle entrance. There was concern that these losses might be higher than analytical prediction techniques indicated, thereby detracting from an otherwise attractive concept. These predictions, as well as the actual testbed engine results, are For single bypass operation, the predictshown in figure X1-28. ed losses are in close agreement with the actual test results and are fairly inconsequential. FOK double bypass operation at the the predicted thrust losses were strut design flow condition, but the actual measured loss was only slightly around 2 percent, test result. greater than for single bypass - a very encouraging Figure XI-29 shows engine specific fuel consumption as a function of fan airflow on the test stand at part throttle conditions with This type of operation is representative of constant thrust. cutback at the takeoff flyover noise measuring station OK subIt shows that a 29-percent increase in airflow is sonic cruise. possible for double bypass operation compared to the normal single bypass operation without exceeding the minimum specific fuel consumption in single bypass. At cutback, this could mean a 6 to 10 perceived noise decibel (PNdB) reduction in jet noise because of the lower jet velocity. At subsonic cruise, the greater mass flow implies reduced throttle-dependent installation drag because the inlet spillage and nozzle boattail are reduced. Trades are possible, however, between increased airflow in double bypass and lower uninstalled specific fuel consumption. FOK instance, an uninstalled specific fuel consumption about 5 percent lower could be obtained at an airflow about 13 percent greater than at the single bypass minimum specific fuel consumption point. Incidentally, another important objective in these tests was met by successful transitions between single and double bypass modes, and vice versa, without engine stall over a wide range of power settings. The acoustic test results the baseline single-stream Perceived noise level 30. 356 for both the coannular plug nozzle and conic nozzle are shown in figure XI(PNL) is plotted at the peak sideline

The data shown angle against the mass-averaged jet Velocity. here have been scaled to study engine size, and the distance has been scaled to the sideline measuring station in a flight situAll the coannular data shown are for an inner to outer ation. The outer nozzle has a convergentnozzle area ratio of 0.2. divergent termination designed for a 2.9 pressure ratio. Its The acoustic benefit of the coannular radius ratio was 0.85. plug is about 6 PNdB relative to the conic at the higher jet velocities representative of takeoff conditions. A velocity ratio of approximately 1.5 between the hot and cold exhaust streams was maintained above a mass-averaged velocity of about 1750 feet per Below this velocity, second. the velocity ratio and the coannular benefit began to deteriorate because of the fixed nozzle areas used in this test. The sideline directivity of the PNL observed in the Edwards testangles measured from the ing is snown in figure XI- 31 at various inlet axis. Test data are shown for both the reference mixed conic and the coannular plug nozzles. The peak sideline PNL occurs in the aft quadrant at an angle of 1.300 where the coannular nozzle is about 6 PNdB quieter than the baseline conic, the same result shown in the preceding figure for this mass-averaged jet velocity of 1350 feet per second. In the forward quadrant there also appears to be about 7 PNdB in shock noise reduction from the coannular nozzle. Small-scale model coannular data are also shown for the coannular plug as a broken line. Very close agreement was obtained between the model and testbed engine data when both are scaled to full size and a common distance. Core-driven fan stage VCE testbed. - The acoustic test of the initial testbed engine configuration shown in figure XI-25 was Figure XI-32 shows a testbed engine completed in October 1978. now being built which is more similar to the advanced double byThis engine, pass study engine. to be tested in 1980, will have a core-dKiven third-stage fan, closer spacing between the fan blocks and compressor, and certain exhaust system modifications to allow for testing a simple outer stream mechanical suppressor. Figure XI-33 shows a retractable shallow-chute suppressor as it might be applied to the outer stream of the advanced double bypass study engine. low-loss suppressor Such a lightweight, could probably be added to a flight-type engine with very little penalty to the aircraft mission or cost effectiveness. Further reductions to the sideline and community cutback noise would result relative to that obtained with only the coannular inverted velocity profile benefit alone. Studies are now underway to determine the mission/acoustic trades possible with such systems. Using suppressors such as these should bring the projected noise levels closer to those specified by FAR 36 (1978). The version to be used on the testbed is still in the initiai design phase, but it would be a similar, although nonretractable, boilerplate design. Yodel scale suppressor work, which is now starting, should provide some further design information for this suppres357

in addition sor, funded experimental

to

that available studies done

from the by General

previous Electric.

DOT/FAA-

Program status. - The accomplishments of the General Electric component testbed program have, thus far, been very encouraging. The coannular jet noise relief of 6 PNdB obtained at the sideline condition verified the acoustic predictions based on the scale The part-power airflow extension capabilities at model work. lower specific fuel consumptions in the double bypass mode have demonstrated another of the variable cycle benefits. Successful transitions between double and single bypass modes have also been accomplished. Furthermore, the bypass system losses from the fan discharge through the crossover struts to the inner nozzle were less than expected. a major goal in the General ElecNext year, tric program is to verify the unique core-drive fan system and evaluate any additional acoustic benefit derived from a simple outer stream suppressor. SUPERSONIC CRUISE INLETS

So far in this discussion only the engine and the exhaust nozzle Another important element of the propulsion have been discussed. package is the engine inlet. At the supersonic cruise speeds of the inlet develops a pressure ratio of up to 20 to 1, interest, thus generating over half of the propulsive thrust. Basic Inlet Properties

There are several design factors that an inlet designer must consider when designing an inlet. Each of these factors tries to These factors are shown around the shape the inlet its own way. while the basic inlet properties which outside of figure X1-34, the designer must select are shown inside the box. Each of these basic properties will be described, and the factors will be discussed as appropriate. - The first and most basic inlet property is geomGeometry type. The inlet would really like to be made of rubber so etry type. that it could change its shape to just match the airflow requireBut, to be practical, a basic ments at each aircraft speed. shape must be selected as well as a method of varying the interFigure XI-35 shows the trend of three inlet propnal flow area. erties for three typical geometry types. One curve shows the trend of inlet complexity and the other curve shows the trend of both inlet weight and inlet off-design airflow capacity. The inlet on the left is a twin duct two-dimensional expanding wedge inlet. A photograph of this inlet type with the ramps in This photograph vertical orientation is shown in figure X1-36. was taken from a position ahead of and below the inlet with the 358 a

The internal area of this inlet is flow going into the page. varied by a simple doorlike movement of the aft portion of the a hingeline on the ramp surface. The curves in figramp, about ure XI-35 show that this inlet tends to be the least complex. It also has the largest off-design airflow capacity due to its large throat area capability. However, it does tend to be the heaviest. The middle inlet in figure XI-3S is an axisymmetric expanding (axi expanding) centerbody inlet which is also known as a collapA photograph of this inlet type is shown sing centerbody inlet. in figure X1-37. The internal area of this inlet is varied by the movement of overlapping leaves about a hingeline on the cone This mechanization is similar to that of an iris nozsurface. Figure XI-35 shows that this inlet tends to be the most zle. But it is lighter than the expanding wedge inlet and it complex. has a moderate off-design airflow capacity. The inlet on the right in figure XI-35 is an axisymmetric transA photograph of this lating (axi-translating) centerbody inlet. inlet type is shown in figure X1-38. The internal area of tnis by translating the centerbody along a support inlet is varied tube. Figure XI-35 shows that the basic axi-translating inlet (lower edge of shaded bands) tends to be somewhat more complex than the expanding wedge inlet. It tends to have the lowest weight, but it also has the least off-design airflow capacity. inlet may not at speeds near Mach 1, the axi-translating In fact, The addition of be able to supply enough airflow to the engine. such as auxiliary inlets, can increase auxiliary (AUX) devices, its airflow capacity, but these devices will also make it heavier All three SCR airframers presently snow some and more complex. all the Also, type of axi-translating iniet on their aircraft. But It must be VCE engines have been matched to this inlet type. remembered that its potential 1acK of off-design airflow capacity may result in marginal or inadequate transonic thrust. Compression split. - The second inlet property is compression The upper sketcnes will split and is described in fiqure X1-39. be used to define some terms: The left sketch-shows that at supersonic cruise the inlet uses multiple shock waves to slow the air to subsonic speeds before It enters the engine. The oblique The shocks are produced by the inlet flow area contraction. final wave is a normal shock which sits downstream of the minimum External compression is defined as area called the inlet throat. the flow area contraction occurring ahead of the cowl lip, and internal compression is that occurring between the cowl lip and The inlet is said to ue started when the shocks are the throat. in this high-performance position. But if the normal snoc& is by something pushed into the converging area ahead of the throat, like increased angle-of-attack or a control error, it is unstable and will be abruptly expelled to a point in front of the cowling, This process is calLed an inlet unstart. as shown on the right. The unstart transient can be very violent and the unstarted re359

covery very low. symmetric forces

These on the

unstart aircraft.

characteristics

impart

large

un-

The curves in figure XI-39 show that the desire for higher recovery and lower cowl drag will push the inlet designer toward high internal compression. The penalties will be greater bleed lower angle-of-attack (~1) tolerance before unstart, greater drag I unstart severity, and lower unstarted recovery. The figure also indicates the amount of internal compression that would be employed by the translating and expanding inlet concepts. The axitranslating (TRANS) inlet must employ high internal compression. The high internal compression is required so the translating inlet can obtain the needed off-design area variation. The expanding (EXP) concepts allow the choice of more modest amounts of internal compression. The choice would be the least amount of internal compression consistent with relatively good recovery and thus reducing the bleed drag and unstart penalties. cowl drag, Variable geometry variable geometry figure shows the during a typical has already been at cruise speed, this condition. features. features, gyrations mission. described. the inlet - The third inlet property, required This is described in figure X1-40. of an axi-translating centerbody inlet The supersonic cruise shock structure Since the inlets are normally sized is always matched to the engine at

The supersonic throat is too small in the off-design speed range and must be increased by extending the centerbody. The throat now occurs near the cowl lip and the terminal shock sits in front of the cowling. Toward the upper end of this speed range, bypass may be needed to spill air the engine cannot use. However, as speed is further reduced to near Mach 1, the bypass will be closed. At this point, as mentioned before, the translating inlet has difficulty obtaining all the flow area the engine demands. The inlet must deliver a large quantity of high recovery airflow However, a to the engine at takeoff so as to reduce jet noise. sharp lip separation is present at this condition which forces the addition of auxiliary inlet airflow to supply enough good quality air to the engine. Fan noise suppression of up to 20 decibels may be required in the This suppression will most likely be accomplished terminal area. by a combination of acoustic treatment on the internal walls, similar to that described by the noise reduction panel, and operThe ation of the inlet at high throat Mach numbers near choke. auxiliary inlets will have to be open for some noise suppression conditions in order to supply enough air to the engine. Therefore, in addition to the main duct, the auxiliary inlets will also have to incorporate some form of noise suppression for these conditions. This figure shows what a complex machine the inlet can be. 360

Design

Factors

Now that the basic inlet properties have been described, the design factors of figure XI-34 will be discussed. Noise suppression, cruise speed, performance, weight, and alpha tolerance have already been covered. The need for good shock stability has also But it should also be mentioned that shock stabilbeen implied. ity systems are being studied that will tend to prevent unstarts, or at least reduce their frequency of occurrence. But these shock stability systems will add complexity. The engine cycle is alSO an important driver on.inlet selection. For a VCE to operate as advertised, it must be matched to an inlet which can efficiently provide the airflow that the engine demands over the full The mission is important as it defines such things speed range. as the need for subsonic cruise legs which in turn can dictate The nacelle location dictates the required inlet throat area. the Mach number and flow angles at the inlet station. Good compatibility and controlability are required for any supersonic cruise vehicle and must be considered early in the inlet selection process. Maintainability and reliability of the various inlet systems are nebulous quantities, particularly early in a development cycle, but can be all important to a final inlet selection. Axi-Expanding Inlet Evaluation

The previous review of supersonic cruise inlets has alluded to some potential airflow and low internal compression advantages of In 1978 it was decided to further exthe axi-expanding inlet. plore these potential advantages by having the three SCR airframers - Boeing, Douglas, and Lockheed - conduct preliminary axi-expanding inlet evaluation studies. Each airframer was to aerodynamically and mechanically design an axi-expanding inlet to match their aircraft and VCE. Then they were to estimate its performance, weight, complexity, and operating characteristics and define any unique features associated with this inlet concept. They were then to compare all of these characteristics with those of their present axi-translating inlets. Finally, they were asked to identify inlet technology areas that require Since each airframer was studying a different future effort. aircraft at a different Mach number, three different answers might be expected. General agreement. - There were several items on which all three contractors agreed. They agreed that the axi-expanding inlet requires less bleed, has a greater alpha tolerance and a milder unstart, and can deliver a larger off-design airflow. The axiexpanding inlet has a less complex centerbody bleed system, but it does have complex centerbody leaf seals with possible leakage Since very little is known about noise suppression in problems. supersonic cruise inlets, the relative suppression advantages of the two inlets could not be identified. All of the contractors 361

felt fan noise suppression testing is required. ings of each contractor are now reviewed in order design Mach number.

The other findof increasing

Douglas evaluation. - Figure XI-41 shows simplified cross sections of the two inlet types for the Douglas Mach 2.2 aircraft. The bypass and auxiliary inlet systems are shown in different positions on the top and bottom of the cowl to illustrate movement. The axi-translating inlet incorporates two unique feaOne is a telescoping centerbody for added transonic airtures. flow. The aft portion of the centerbody is translated forward, with the spike tip in the full aft position, in order to open an auxiliary flow path through the centerbody. The other unique feature is the forward translating cowl auxiliary inlet system which is used for added takeoff airflow. An aft bypass door system is also employed in which the doors rotate about an axial axis, somewhat like a furnace damper. The axi-expanding inlet, shown on the bottom, also incorporates the translating cowl auxiliary inlet system and the aft bypass rotating system. The bleed system on this inlet is much simpler than the bleed systems on the axi-translating inlet. Douglas found that the axi-expanding inlet resulted in 3.3 percent more range due to the 27 percent less weight, a better supersonic cruise specific fuel consumption, and more climb thrust. A negative factor It also required fewer operating mechanisms. on the axi-expanding inlet was less area for acoustic treatment due to the shorter length and overlapping leaves. Douglas found that the axi-expanding inlet is a valid VCE concept but that verAlso, since all the VCE's have been ification is required. matched to an axi-translating inlet, Douglas recommends that VCE cycles should be pursued that might take better advantage of the added flow capacity of the axi-expanding inlet. Boeing evaluation. - The Boeing inlet geometries for their Mach 2.4 aircraft are shown in figure X1-42. The Boeing axiThe aft door translating inlet is a typical translating design. system, which is used on both inlets, is a combination bypass and auxiliary inlet system. This aft door system rotates out for The axi-expanding bypass and in for auxiliary inlet operation. inlet incorporates the unique feature of an added hinge point in the subsonic diffuser in order to obtain better throat position the bleed system on the axi-expanding and area control. Again, inlet is much simpler than those on the axi-translating inlet. Boeing found that the axi-expanding inlet resulted in 1.7 percent less range; this was due mainly to the inlet's 42 percent greater weight even though it provided better SFC at cruise and climb. Although they felt that the axi-expanding inlet would have greater benefit for the lower cruise speeds and that it is a valid VCE concept, they also felt that verification is required, Boeing's main near term particularly on the subsonic diffuser. 362

is in combining existing three-dimensional interest, however, flow. codes into the three-dimensional analytical models needed to design supersonic cruise inlets and their associated control systems. They would prefer to use an existing axi-translating wind tunnel model, which has a large available data bank, to experimentally verify these analytical models. Lockheed evaluation. - The Lockheed inlet geometries for their Mach 2.55 aircraft that incorporates overwing and underwing inlets are shown in figure X1-43. Both the over and under inlets are designed for the local Mach number at the overwing position of 2.75. The axi-translating inlet is a typical translating design. The aft door system, which is used on both inlets, is a The aft door syscombination bypass and auxiliary inlet system. tem opens as one door for bypass but opens several small louvers for takeoff auxiliary airflow. The axi-expanding inlet is unique in that it incorporates translation to maintain better shock position and airflow control. Again, the bleed system on the axiexpanding inlet is much simpler than those on the axi-translating inlet. Lockheed found that the axi-expanding inlet had a 3 percent range advantage due to its better supersonic cruise SFC and better climb engine flow match. A better takeoff recovery is also obThe absence of cowi bleed tained with the axi-expanding inlet. on the axi-expanding inlet permits the installation of acoustic treatment between the engine face and the auxiliary inlets and However, also gives increased auxiliary inlet position options. Lockheed felt the axi-expanding inlet was 21 percent heavier. the axi-expanding inlet was that, within the scope of this study, their preferred inlet but that concept screening should continue. Evaluation summary. lowing conclusions: (1) The range comparable to the inlet. - The three SCR airframers reached the folwas

of an aircraft with range of an aircraft

an axi-expanding inlet with a axi-translating

(2) Factors other than performance were more important as These items indecision criteria in selecting an inlet type. clude engine flow matching, angle-of-attack tolerance, noise suppression characteristics, and complexity. which The the (3) The axi-expanding inlet is adds a VCE flow match option. three airframers inlet recommended verification should the a viable following: advantages of SCR inlet candidate

(1) Experimental axi-expanding

of the potential be obtained.

363

(2) New inlet concepts should be screened vised so as to generate new and perhaps better ations. (3) Early ance and fan cruise inlets. CONCLUSIONS testing is needed to define noise suppression characteristics the

or old concepts reinlet-VCE combinlow-speed performof supersonic

The objective of the Supersonic Cruise Research program is to provide this country with an advanced technology data base to support possible future civilian and military supersonic cruise aircraft. This report has shown some of the recent technical advancements in supersonic propulsion that indicate cost and environmental improvements are possible in a second generation supersonic The paper has also described, in some detail, the transport. work being done on variable cycle engines. These unique propultogether with advanced supersonic inlets and low sion concepts, noise coannular nozzles, promise to provide good operating performance at both supersonic and subsonic flight, are reasonably quiet during takeoff and landing, and have acceptable exnaust emissions.

364

SUPERSONIC

CRUISE

RESEARCH

AERODYNAMICS

SION

cs-79-2003

w ENVIRONMENTAL IMPACT

Figure XI-I

ADVANCED SUPERSONIC CRUISE AIRCRAFT CONFIGURATIONS

BOEING

MCDONNELL DOUGLAS

Figure XI-2

365

1971 PROPULSION STATUS

. MARGINAL PERFORMANCE . EXCESSIVE NOISE l HIGH D(HAUST EMISSIONS>

C-79-1639

GE-4 AFTERBURNING TURBOJET

Figure XI-3

ENGINE REQUIREMENTS
TYPICAL SUPERSONIC TRANSPORT ,-LOW-NOISE TAKEOFF + LOW EMISSIONS

!I / /
Il I /

,-EFFECTIVE SUBSONIC CRUISE

/ /
: /

,-EFFICIENT SUPERSONIC CRUISE + LOW NO, ,-EFFICIENT HOLD/

t
ALTlTUDE

//

j/#-c;,;W+d&;SE

RANGE -

cs-79-1636

Figure XI-4

366

EVOLUTION OF ENGINE STUDIES

SCREEN ALL

NO. OF CONCEPTS

1972 TIME-

PRESENT
CS-73-Lb35

FigureXI-5

PROPULSION PROGRAM
_--_--------.----------\ I- WLANE_S_YSJEM STU_DIES__ _ _ _ _ _ _ _ - 1
41 1 PROPULSION SYSTEM STUDIES II l-l 1972 \

c1 COANNULAR NOISE

cs-79-1638

1978 1INLET, NOZZLE, NACELL+

FigureXI-6

367

--

CS-79-1764

LOW BYPASS ENGINE (LBE)

Figure XI-7

DOUBLE BYPASS ENGINE (DBEl

VARIABLE STREAM CONTROL ENGINE (VSCE)

Fluure XI-8

368

COMPARISON OF ENGINES
M = 2.4; ALL SUPERSONIC CRUISE 292 PASSENGERS RANGE, 4500 n mi FAR 36 (1969) I I

850

:::y ;;II*:~~~~~~T~~;~N~,s
WEIGHT

FAR 36 ! SIDELINE

1971 0 TURBOJET

98

100

102

104

106 108 110 NOISE, EPNdB

112

114
CS-79-1637

116

Figure XI-9

AIRCRAFTNOISE LEVELS
G.E. DOUBLE BYPASS ENGINE EXHAUST 0 q 110 EFFECTIVE PERCEIVED NOISE LEVEL, EPNL, dB m ___-__---------s----11-m-FAR 36 11969) FAR 36 (1978) s ----I COANNULAR COANNULARISUPPRESSED

CS-79-1809

COMMUNITY
Figure

APPROAC H

XI-10

369

RELATIVE DIRECT OPERATING COST

PRATT &WHITNEY AIRCRAFT ENGINES AVG TRIP

[LUl3SONlC ;:; :1.: .

TRIP LENGTH, % OF DESIGN RANGE

cs-79-16

Figure XI-II

VARIABLE STREAM CONTROL ENGJNE

TAKEOFF

ADVANCED HIGH SPOOL

n-79-2379

LOW EMISSIONS MAIN BURNER

LOi EMISSIONS DUCT BURNER

REVERSER CDAMJLAR NOZZLE

Figure XI-12

370

PRATT AND WHITNEY AIRCRAFT VARIABLE STREAM CONTROL ENGINE

SUBSONIC CRUISE

f. .

Figure XI-13

COMPARISONOF VSCE-502BAND TESTBEDCONFIGURATION

Low iMiSSIoNS ~cOANNULAR VSCE-502B

DUCT BURNER NOZZLE DUCT BURNER

,;---7

FlOO/TESTBED
cs-79-1727

COANNULAR NOZZLE WITH TREATED EJECTOR Figure Xl-14

371

VSCE-502B EJECTOR NOZZLE

SUPERSONIC CRUISE

PRIMARY @

SUBSONIC CRUISE/TAKEOFF FigureXI-15

cs-79-1722

VSCE-5026 NOZZLEIN 8'X6'SUPERSONIC WIND TUNNEL

FigureXI-16

372

VSCE-502B NOZZLETHRUST PERFORMANCE

1.00 .98 .96 GROSS THRUST COEFF .94 .92 .90 .88 .86 I 0 I .5 SUBSONIC CRUISE 1.0 MACH NO. I 1.5
CS-79-172L

SCR STUDY GOALS

SUPERSONIC m CRUISE

0 TAKEOFF

2.0

Figure XI-17

THREE-STAGE VORBIX DUCT BURNER

PILOT PRECHAMBER

PILOT SECONDARY

HIGH POWER

cs-79-1723

Figure XI-18

373

MEASURED DUCT BURNER EMISSIONS

TAKEOFF COMBUSTION EFF 99.7% 30 EMISSIONS INDEX, lb EMISSIONS 1000 lb FUEL SUPERSONIC CRUISE COMBUSTlON EFF 99.9%

2.

lo

~~-79-1725

NO,

CO

THC

NO,

CO

THC

Figure XI-19

VSCE-502B EXHAUST EMISSIONS ESTIMATES

AIRPORT ALTlTUDE CRUISE

2o

0 ml

DUCT BURNER MAIN BURNER

15 EPAP, lb EMISSIONS 1000 lb FN-hr-CYCLE 10

cs-79-1726

Figure XI-20

374

VSCE/F-100 COMPONENT TESTBED

INSTALLED IN X-16STAND

Figure XI-21

VSCE/F-100 COMPONENT TESTBED WITHDUCT BURNER OPERATING

Figure XI-?2

375

DOUBLE BYPASS

ENGINE

cs-79-1996

Figure XI-23

DOUBLE BYPASS ENGINE

GENERAL ELECTRIC CO.

TAKEOFF

SUBSONIC CRUISE

SUPERSONIC CRUISE

Figure XI-24

376

YJIOVEARLY TESTBED

ACOUSTIC ENGINE

Figure XI-25

CALIBRATION

NOISE TEST

OF GE ENGINE

377

COANNULAR

NOISE TEST

OF GE ENGINE

Figure XI-27

BYPASS SYSTEM LOSS COMPARISON

0 m TESTBED PREDICTED TEST RESULT

i
THRUST LOSS, $&+ $

I
SINGLE " BYPASS DOUBLE 8YPASS
(35-79-2334

FigureXI-28

378

TESTED ENGINE PERFORMANCE DATA

1.10 CONSTANT 50% THROTTLE 1.05 SINGLE BYPASS

REEVE

1.00 -

.95 -

.90 .90

I .95

1.00 1.05 1.10 1.15 1.20 1.25 1.30 RELATIVE TOTAL AIRFLOW CS-79-1807

Figure XI-29

ACOUSTIC TESTDATA
SCALED TO FULL SIZE 8, 2400 ft SIDELINE

PEAK SIDELINE PERCEIVED NOISE LEVEL

LAR PLUG NOZZLE

I
cs-79-1810

1250 1500 1750 2000 2250 2500 MASS AVERAGED JET VELOCITY, ftkiec

Figure XI-30

379

NOISEDIRECTIVITY
SCALED TO FULL SIZE & 2400 ft SIDELINE; v, = 1950 ftlsec

I
PERCEIVED NOI SE LEVEL

CONIC NOZZLE7

10 dB

I IO

-----

TESTBED ENGINE MODEL NOZZLE

20

40 60 80 100 120 ANGLE FROM INLET AXIS, deg

140
CS-79-1808

160

Figure XI-31

YJ IOKORE-DRIVEN

TESTBED

ENGINE

Figure XI-37

380

COANNULAR NOZZLE WITH SIMPLESHALLOW CHUTE SUPPRESSOR

.,- DEPLOYED SUPPRESSOR

CS-79-1806

Figure XI-33

INLET DESIGN FACTORS

SUPPRESS NOISE \ / CRUISE SPEED

MAINTAINABILITY RELIABHJTY COMPATIBILIN 7 CONTROLLABILITY NACELLE / LOCATION

GEOMETRYTy PES COMPRESSION SPLIT VARIABLE GEOMETRY

/ 1

PERFORMANCE WEIGHT a TOLERANCE

SHOCK STABILIIY

MISSION /

ENGINE CYCLE

CS-79-2034

Figure XI-34

381

INLETGEOMETRY TYPES

j ,,-COMPLEXITY / 0 1 I f 1I I I/

TREND LO WEDGE EXPAND 1 AXI EXPAND Figure XI-35 I . AXI TRANSLATE cs-79-1833

EXPANDING WEDGE INLET

Fiyure Xl-36

382

EXPANDING

AXI INLET

Figure XI-37

AXI-TRANSLATING

INLET

Figure XI-38 383

EFFECT OF INLETINTERNAL COMPRESSION

COMPRESSION EXTERNAL INTERNAL

RECOVERY COWL DRAG

WORSE

BLEED DRAG a TOLERANCE UNSTART SEVERITY UNSTART RECOVERY HI INTERNAL COMPRESSION

CS-79-1838

LO

Figure XI-39

INLETVARIABLE GEOMETRY
SUPERSONIC CRUISE OFF DESIGN

TAKEOFF \ v-1 / Pi!! ._.w I

NOISE SUPPRESSION

!I
cs-79-1835

-7A

FigureXI-40

384

DOUGLAS INLETGEOMETRIES

EXPANDING

CS-79-1836

Figure XI-41

BOEING INLET GEOMETRIES

CRUISEMACH 2.4 TRANSLATING
111117\

--

JllJll

EXPANDING

Figure XI-42

385

LOCKHEEDINLET GEOMETRIES
CRUISEMACH = 2.55 TRANSLATING I rI In-r--_Ba EXPANDI NG IIIII~ I

Figure XI-43

386

XII.

HYPERSONIC

PROPULSION Jr.

H. Lee Beach, National

Aeronautics and Space Administration Lewis Research Center

Research at NASA on vehicles and propulsion systems for hypersonic flight in the atmosphere is aimed at providing viable technological options for future civil and military applicadeWithin the uncertain framework of social, economic, tions. a wide spectrum of possibiliand environmental issues, fense, It is imperative, therefore, that key technologies ties exists. be pursued that can provide the alternatives when they are reVarious airbreathing and rocket propulsion system alquired. Of tne airbreathing systernatives are shown in figure X11-1. have the highest performance (as measured by tems, turbojets Above specific impulse) up to Mach numbers of approximately 3. turbine inlet temperature constrains performance, this speed, About Macn 6, dissociation and ramjets become more attractive. losses caused by slowing the high-velocity stream to subsonic It is then more effispeeds greatly reduce ramjet performance. cient to allow the engine internal flow to remain supersonic and the engine to operate as a scramjet (supersonic combustion ramAll systems in figure XII-1 except the turbojet are shown jet). for both hydrocarbon and hydrogen fuels. Hypersonic propulsion research is concerned with Mach numbers of the scramjet is the primary At these speeds, 4 to 6 and beyond. of its obvious performance advansystem of interest because the research program now unoer way at tne Specifically, tages. NASA Langley Research Center deals with hydrogen-fueled scramHydrogen is being studied because it provides a very jets. large performance benefit that yields specific impulse levels at Mach 7 comparable to those of JP-fueled turbojets at about Mach Hydrogen also has the excellent and necessary capability to 2. cool both the engine and parts of the airframe in high Mach number flight. This paper discusses the current NASA program in hypersonic proThe program is first put into context by a brief dispulsion. The status cussion of past and other ongoing scramjet programs. and emphasis is given to comof the research is then discussed; computational metnods, and preliminary ponent developments, ground tests at Macn 4 and 7 of subscale scramjet engine modules.

387

II

1mmm.,- ,111, m m., m.,--l-l...

, .,.,--

..--.-------

_..-.--

BACKGROUND The United States' commitment to scramjet research began in the early 1960's and has continued at various levels of effort through support from NASA, the U.S. Navy, and the U.S. Air and considerable progress can Much work has been done, Force. An excellent rebe claimed in all pertinent technology areas. view of scramjet development efforts in the United States is Much of the discussion in the next secgiven in reference 1. tion is derived from that reference. Brief Review of Scramjet Research

The U.S. Air Force became involved with scramjets 20 years ago through interest in external-burning and single-stage Earth-toorbit vehicles. Engine development programs were sponsored with the General Electric the United Aircraft Research Laboratory, General Applied Science Laboratories, and the Marquardt Co. co., Four research engines were built and ground tested in the 1960's. These engines were hydrogen fueled and generally achieved the internal performance levels that had been predicted. Features included in one or more of these engines were staged fuel injection for operating the variable-geometry inlet; rearward-facing steps for comover a wide Mach number range; and thermal compression. At the conclubustor-inlet isolation; the Air Force began concentrating sion of these test programs, Subseon smaller missile systems that used hydrocarbon fuels. quent research has been related to piloting such missile systems. U.S. Navy-sponsored scramjet research has continued since the early 1960's under the direction of Johns Hopkins University's This work has been missile Applied Physics Laboratory (APL). related, with emphasis on unmanned systems that are constrained such sysby the launching and handling requirements of ships. storable fuels and passive tems are volume limited; therefore, The research has included testing cooling are most desirable. Inlet deof both engine components and complete engine models. velopments have resulted in inverted, internal, axisymmetric These compression flow fields split into three or four modules. inlets appear to perform well and use sweep to minimize heatThe combustor configuration that has evolved transfer problems. from many tests of autoignitible hydrocarbon fuel uses a reara constant-area isolation section, and a coniward-facing step, Free-jet tests cal section that expands to an area ratio of 2. have been conducted at Mach of engine models with these features Net thrust (thrust greater than ennumbers of 5, 5.8, and 7. gine drag) was measured for at least one configuration at all More recent research at APL has focused on three Mach numbers. dual-combustion ramjet for the volumean integral-rocket, This concept uses an embedded dump-combustor limited missions. 388

ramjet primary

system to scramjet

serve as a fuel-rich, system (ref. 2).

hot-gas

generator

for

NASA also began its scramjet research in the 1960's with a program to advance the technology for manned vehicles. The Hypersonic Research Engine (HRE) project provided the focus for this effort; the original goal was to test a regeneratively cooled, flight-weight engine on the X-15 research airplane. Although the cancellation of the X-15 program precluded the achievement of this goal, two engines were designed and built by Garrett AiResearch and ground tested. The HRE was axisymmetric and used a translating spike to position the shocks and to control the Combustor opinlet airflow over a Mach number range of 4 to 8. eration was controlled by staging fuel injection streamwise over the Mach number range. Structural, nonburning tests of a complete flight-weight, regeneratively cooled version of this engine were made in the Langley Research Center's 8-Foot HighTemperature Structures Tunnel. Thrust performance was determined in separate tests at the Lewis Research Center's Plum heavy-walled version of the enBrook Station of a water-cooled, The measured internal performance at Mach 5 to 7 was very gine. These two sets of tests verified the close to the design goal. structural and cooling design and manufacturing techniques and demonstrated the feasibility of good internal thrust over a range of flight speeds. Although good internal thrust performance remains a primary goal it must also achieve high installed for any propulsion system, This fact performance (internal thrust minus external drag). has driven the emphasis of scramjet research away from most of the engine types examined in the 1960's to concepts tnat are intimately integrated with the vehicles on which they are installed. Airframe-Integrated Scramjet Concept

The meaning of airframe - propulsion-system integration for In this concept, tne scramjets is illustrated in figure X11-2. entire undersurface of the vehicle is part of the propulsion Integration is needed at high Mach numbers because alsystem. most all the airflow between the vehicle and its bow shock is This suggests an inlet capture required for good performance. By splitting the annular area into area with an annular shape. the primary engine becomes a system smaller rectangular modules, of identical units of a size and shape appropriate for testing in ground facilities. Within this frameworK, the vehicle forebody performs a significant part of the inlet function, and the Engine module afterbody becomes the nozzle expansion surface. external surfaces can therefore be easily shaped to minimize installation losses, and the vehicle base region can provide the large expansion ratios that are necessary. 389

The rectangular module concept that takes advantaqe of this apSince the vehiproach is shown in the center of figure X11-2. cle compresses the flow in the vertical direction, the module inlet side walls have wedge shapes to compress the flow horizonThis tends to minimize flow distortion over a flight tally. in combination witn a Sweep of these wedges, Mach number range. allows spillage to occur efficiently with recess in the cowl, Inlet compression is completed by three wedgefixed geometry. shaped struts located at the minimum-area section of the modThese struts also provide multiple planes for fuel injecule. tion and therefore shorten the required combustor length. A cross-sectional view of the strut region, shown to the left in illustrates the technique used to control the disfigure X11-2, tribution of heat release in the combustor. At Mach numbers of 7 and higher, almost all the fuel is injected perpendicular to This is an inherently fast mixing the airflow over the struts. process and results in rapid heat release close to the struts At Mach numbers below 6, too much perfor maximum performance. pendicular injection produces too rapid mixing and reaction, and To prevent this, most of the fuel is the flow thermally chokes. injected from the base of the struts parallel to the local flow Mixing thus occurs much more slowly, and the heat direction. release is stretched out over the length of the combustor. The scramjet concept now under investigation is tnerefore a fixed-geometry module using three dimensionality and mixed-mode fuel injection to allow flight over a wide Mach number range. The sections that follow describe the nature and status of the research program to support this concept. THE AIRFRAME-INTEGRATED RESEARCH PROGRAM

The NASA Langley Research Center is currently engaged in a broad-based research program (refs. 3 and 4) to establish the aerothermodynamic lines and performance potential of this scramjet airframe-integrated module concept. The program includes research on engine components (inlets, combustors, and nozzles), computational fluid mechanics for internal flows, component integration (subscale engines), structures, and flow diagnostics. The performance of the module itself is the key technology area and is the focus of the remaining discussion. Airframe integration (including nozzles) and structures are also briefly covered for completeness. Inlet Research tests over a period of years to a relatively mature state of Figure XII-3 shows one of the test

Four generations of aerodynamic have brought the inlet concept development (refs. 5 and 6).
390

A foreplate with boundary-layer models. partially simulate the boundary layer on One side wall of the model has vehicle. the struts and instrumentation rake used files at the throat.

trips was installed to the undersurface of a been removed to show to measure flow pro-

Tests were conducted at Mach numbers ahead of the inlet from 2.3 Results to 6, corresponding to flight Mach numbers from 3 to 8. were presented in the proceedings of the 1975 Aeropropulsion 7) and indicated variations in throat Mach numConference (ref. and pressure recovery simratio, mass capture, ber, contraction It appears that the ilar to those of a variable-geometry inlet. vehicle boundary layer can be ingestea without separation or at least as long as the effects of the combustor other problems, flow do not feed forward. Although some inlet testing has taken place in the last few Its peryears, the basic inlet has not changed significantly. the key remaining quesformance is considered to be excellent; tion is that of inlet-combustor interaction, which is discussed in a later section. Combustor Research

The difficulties in achieving rapid mixing and combustion in the fraction-of-a-millisecond residence times available in a supersonic combustor have dictated that considerable research be conFuel ducted on the physics and thermodynamics of such flows. injector and combustor configurations are experimentally evaluin a combustion-heated facility ated by direct-connect tests The facility burns a mixture of hydrogen, oxy8 and 5). (refs. and air to form a high-enthalpy test gas with an oxygen gen, Stagnation conditions mole fraction identical to that of air. up to 500 psi and 4700 R can be produced; sucn values are represelltative of Mach 8 flight. A large number of configurations has been tested over the past 8 These have included fuel injection from walls (refs. 10 years. from single struts with perpendicular or parallel injecto 12), from multiple struts (ref. 14), and from swept tion (ref. 13), Through these tests, the mixing perstruts (refs. 15 and 16). Mixedformance of the baseline concept has been established. and parallel injection appears to provide a. mode, perpendicular viable mechanism to control mixing and heat-release, and tne design length of the module comoustor appears to be adequate for approximately 95 percent mixing. An interesting issue of potential importance emerged from several of tnese tests, as well as from the subscale engine tests discussed later. Inconsistencies in ignition and flameholding much-delayed ignition was behavior were noted. For example, 391

A simsometimes observed when rapid autoignition was expected. ple autoignition test program was begun (ref. 17) to investigate these inconsistencies and to solve the ignition and flameholding problems without altering the basic mixing performance that had evolved through years of research. The model used in these tests is shown schematically in figure X11-4. A Mach 2.7 nozzle was mounted at the exit of the combusThe injector model was a simtion heater described previously. ple flat plate on which fuel injectors (usually three) were Such steps are used on the drilled near a rearward-facing step. fuel injection struts in the engine concept to help isoiate the inlet from the combustor and to help stabilize the flame. The Variables step is shown swept to simulate sweep in the engine. in the test program were free-stream conditions, fuel temperaboundary-layer thickness, wall temperature, sweep angle, ture, and various injection parameters such as injector location relative to the step, injector size, and step size. Figure XII-5 shows three typical photographs taken during the All are for Mach 7-simulated conditions with a stagnatests. pressure of 1 atmotion temperature of 4000 o R and a static sphere. The flow is from left to right in these photographs, For inand the step location has been indicated with an arrow. jection 1.67 step heights upstream of tne step (x/h, = -1.67), ignition occurs at the point of injection, and the fuel continThe problem with this ues to burn as it expands over the step. configuration is its relatively high susceptibility to inletWhen the combustor interaction through boundary-layer feedback. x/h, = 1, injectors were moved just downstream of the step to the base region apparently became too fuel rich, and ignition have serious impliThis can, of course, was severely delayed. By moving the injectors an adcations for engine performance. problem x/h, = 3, this ditional 2 step heights downstream to was alleviated. Ignition began near the injectors, and the this condition is the one desired flame stabilized at the step; The 3-step-height criterion proved to be an opfor the engine. timum for locating the injectors, independent of sweep, as long as the distance was measured perpendicular to the sweep line. Several other important results were produced by the ignition tests. There was a very large effect of boundary-layer (energy) For examthickness on the temperature required for ignition. R higher temperatures were needed to ig-f+=, up to 1200 degrees nite the fuel with a thick (0.15 in.) boundary layer than with a In addition, up to 700boundary layer five times thinner. degree R differences in ignition temperature were found for wall temperature differences of 200 degrees R, where the boundaryHeating the hydrogen layer thicknesses were about the same. the hydrogen temperature However, fuel had only a small effect. was always much lower than the air temperature, and mixture tem392

peratures near the ignition substantially. Interestingly, confirmed in these tests to in scramjet flow fields.

point

were therefore not affected the pressure-scale product was be a scaling parameter for ignition

The results of this simple test program have explained inconsistencies observed in various tests and have provided valuable design and scaling information for use in future configurations. It is clear that sensitivities are large, and that extreme care must be taken in controlling boundary-layer thickness, exact injector location, and wall temperature for effective ground simulation. Computational Fluid Mechanics Research

To support the inlet and combustor component research, a significant effort in computational fluid mechanics for scramjets has been under way for several years. The objective is to provide reasonably detailed and reasonably accurate computational tools with emphasis on such perforfor engine design and analysis, mance parameters as combustion efficiency and heat-release control. The scramjet environment is an extremely challenging one for analysis since the flow field is highly three-dimensional and has turbulent mixing and chemical reaction, regions of sepaand shock waves throughout the inlet ration and recirculation, Relative to other propulsion systems, however, and combustor. Because the hydrothe scramjet presents a unique opportunity. the combustor flow is single phase. gen is injected as a gas, In addition, hydrogen-air chemistry is the most studied and best The geometry is also quite simple understood of any fuel. and the fact that tne flow remains (relative to turbojets), largely supersonic gives several computational advantages. Because of the relative lack of understanding of the combustor Threeflow field, recent emphasis has been on the combustor. dimensional codes have been developed (refs. 18 and 19) for computing scramjet flows downstream of the fuel-injector struts, where separation and recirculation no longer occur, and the These codes are equations describing the flow are parabolic. require two-dimensional storage, and give spatially marching, The good agreement with experimental data (refs. 20 and 21). most serious problem in applying them is the specification of initial conditions, which must come from experimental data or some other analysis describing the injector-strut near field. Research is now under way to address this near-field analysis problem. Figure focus. drogen XII-6 shows the sample problem being used for It depicts the sonic, underexpanded injection The features jet into a supersonic crossflow. near-field of a hyinclude 393

separation of the boundary jet shocks from the underexpansion, recirculation zones ahead of and behind layer ahead of the jet, and bow shocks off the separated zone and the jet. the jet, A semiSeveral methods are being used to solve this problem. 22) has been shown to give acceptable reempirical method (ref. a more exact metnod to solve the sults for certain cases; Navier-Stokes equations describing the flow is also being purthe differential equations are ellipIn the latter case, sued. and the computing requirements increase tremendously over tic, For example, complete computer cases with parabolic equations. Full threestorage is required for each spatial dimension. dimensional storage is therefore needed for three-dimensional Computer run times also tend to increase dramatically problems. because time-relaxation techniques requiring many steps are necessary. The problem shown in figure XII-6 has recently been analyzed two the fuel injector was a slot dimensionally (ref. 23); that is, Velocity vectors from such an rather than a discrete orifice. analysis with an air Mach number of 2.5 are shown in figure X11-7. The top and bottom of the plot correspond to the boundThe direction of the arrow indicates flow direction, ing walls. Some of the and the arrow length indicates relative velocity. flow-field structure shown in figure XII-6 has been superimposed Both the front and rear recirculation zones have for clarity. The vecbeen computed as shown by the vectors turning around. as the flow crosses the up-running tors also turn up, then back, The bow shock for bow shock and its down-running reflection. this 'particular calculation was strong enough to separate the Fuel mass contours are shown in top-surface boundary layer. The lines are for constant mass fraction and are figure X11-8. Similar plots are, useful to indicate where the fuel is going. generated for temperature, pressure, and other flow of course, variables. This two-dimensional analysis has progressed to the point that comparisons with experimental data are now being made, and plans to extend the technique to three dimensions are being formuThe ultimate objective is to couple this near-field aplated. proach (or a semi-empirical one) witn the three-dimensional downstream approach to provide a complete combustor analysis caIt should be emphasized that in all computations of pabilty. the results are only as reliable as the turbulence this type, Such modeling is perhaps and reaction models that go into them. Curthe weakest link of all in the analysis of scramjet flows. two-equation turbulence models are being used in the rently, parabolic codes, and the elliptic code uses an algebraic model. Reaction schemes ranging from complete reaction to full finiterate chemistry are being used, and modeling to account for comIt is bined turbulence and reaction effects is also being done.

394

clear that computational fluid mechanics for scramjets has come Complex two- and threea long way in the last several years. dimensional flows are being computed, and these calculations are For now beginning to affect both scramjet analysis and design. strut contours in the latest subscale engine model were example, and detailed mixing and performance designed with such analyses, The fudata from direct-connect tests are now being analyzed. ture importance of computational fluid mechanics for scramjets therefore appears to be large. Component Integration Research

Concepts, results, and experience in component testing have been assembled in the form of component integration, or subscale enInitial wind-tunnel tests at flight-simulated gine modules. Mach numbers of 4 and 7 have been conducted in two ground facilThese tests are the first known for an airities (ref. 24). frame-integrated concept and were intended for research rather Mach 4 testing was done at the than performance demonstration. General Applied Science Laboratories (GASL), Inc., Westbury, to date, there have been approximately 70 tests totaling N.Y.; 12 minutes test time. Mach 7 testing was conducted in the Langley Scramjet Test Facility; to date, there nave been apBoth proximately 100 tests totaling about 25 minutes test time. flight Mach numbers were simulated because they represent engine At Mach 7, almost all operation over a wide Mach number range. the captured flow passes through the engine, most of the fuel is injected perpendicular to the airflow coming by the struts, and Mach 4 spillage the fuel will probably autoignite. In contrast, can be as much as 40 percent, most of the fuel is injected parThe allel to the airflow, and the fuel will not autoignite. susceptibility to inlet combustor interactions should also be quite different at the two Mach numbers. The model being tested at Mach 7 is shown in figure X11-9. It is constructed of copper and is primarily a heat-sink design with water cooling only in high-heat-flux areas such as leading The engine is heavily instrumented to measure internal edges. wall static pressures, temperatures, and heating rates. It is mounted in the test facility on a one-component force balance to measure thrust and drag. Attention was given in the design to allow research on different configurations by changing or substituting components. The model is 6.4 inches wide, 8 inches high, and about 5 feet long. The primary differences in the Mach 4 engine are that it is constructed of nickel and is entirely a heat sink. The Langley Scramjet Test Facility is shown in figure XII-1U. It is an arc-heated facility, supplying 10 megawatts of power provide Mach 7 simulation of 40000 R stagnation temperature and 116 000 feet altitude. The front part of the model is to

395

visible in the test-section viewing port. In these tests, vehicle bow-shock precompression that would occur in flight is simulated by expanding the Mach 7 energy flow to only Mach 6. The facility at GASL uses a hydrogen combustion heater and expands to Mach 3.4 for a Mach 4 simulation of 16OOO R and 70 000 Ingestion of the vehicle undersurface boundary layer is feet. simulated at both Mach numbers by alining tne engine top surface with the nozzle exit so as to ingest the facility boundary layer. This is shown more clearly in the facility schematic at the top of figure X11-11. Preliminary scramjet engine performance is snown at the bottom The performance parameter AF, defined as the of figure X11-11. difference in the thrust balance reading with and without fuel, is plotted as a function of fuei-air ratio (percent stoichioThe symbols represent selected experimental data, and metric). the lines are theoretical predictions representing performance levels that would be required to cruise a hypersonic aircraft. For ratios less than 50 percent, the experimental data approach This is a very encouraging the predicted performance levels. result from preliminary tests. For flight application, however, it will be necessary to operate at equivalence ratios near 1. No experimental data are shown at higher ratios because at botn Mach numbers inlet-combustor interactions occurred that adversely affected performance. These interactions were not surpristo be caused by a combiing, particularly at Mach 4, and appear nation of boundary-layer feedback effects and local thermal choking leading to complete engine unstart. The challenge for future tests is obviously to raise the fuelair ratio without causing inlet-comoustor interactions. To help insure that this will occur, inlet and combustor component tests to examine various aspects of the interaction problem nave been started. The objective for the component integration research is to demonstrate useful levels of thrust within the next sevTwo more generations of models and tests will proberal-years. ably be required to accomplish this. Other Related Research directly involve include to the scramjet modsignificant commitairframe integration,

Several areas of research relating ule developments already discussed These ments in other disciplines. and flow diagnostics. structures,

intimate propulsion-system - airframe inteAs stated earlier, This ingration is necessary for efficient hypersonic cruise. volves careful design of the vehicle forebody so as to provide aerodynamic efficiency and high-quality flow to the module inlet. It also involves extremely careful design of the aft 396

vehicle surface (or engine nozzle), which at Mach 6 generates about half the net thrust of the propulsion system. The nozzle flow is supersonic and must account for interactions between module wakes, spillage from the inlet, and other effects that change with altitude and flight Mach number. These very complex problems are currently being studied through a combination of sophisticated computational methods (ref. 25) and experimental simulation (ref. 26). Structural concepts for regeneratively cooled scramjets are also being developed. The starting point for the current research was the HRE technology mentioned previously. The HRE used a Hastelloy skin backed by an integral offset-fin heat exchanger through which the hydrogen fuel was circulated for cooling. Thermal cycle life was estimated to be approximately 100 for that design. As discussed in reference 27, tnis is extended to about 500 cycles by changing to the airframe-integrated design An additional 500 cycles can be realized by improved approach. fabrication techniques in which fin coolant passages are photochemically etched into the aerodynamic skin and hot-skin brazing is eliminated. Future improvements in materials could raise tne expected engine cycle life at least another order of magnitude. These and other structural problems related to scramjets and fuel injection struts are now being investigated by AiResearch (ref. 28). Flow diagnostic limitations for both combustor and engine flow fields continue to affect both the kind and quality of experimental data acquired. In-stream probes for reacting flows tend to be quite large, since they must be water cooled, and they almost invariably perturb the state of tne stream in which they are attempting measurements. Using nonintrusive techniques is therefore very desirable. In response to the need for more and better measurements in reacting systems, several nonintrusive approaches are being examined to measure static temperature and/or species concentration. Infrared and sodium-line-reversal techniques have either been tried or are being implemented, although neither is expected to give the kind of detailed data that are being sought. A method that appears to have promise for making temperature measurements uses the low-resolution absorption spectra of the OH (hydroxyl) radical associated with electronic transitions in the ultraviolet (ref. 29). The nearterm application of this technique will involve spatially averaged measurements across a combustion duct at several streamwise stations. For spatially resolved measurements, a fiber-optics probe that reduces the optical path length to approximately 0.5 centimeter is being designed. Ultimately, tne solution to combustion measurement problems may be found in coherent antiStokes Raman spectroscopy (CARS). The signal strength and laser-like signal character of CARS give inherent advantages over conventional Raman spectroscopy. The feasibility of using this technique to make temperature and species measurements in 397

is currentiy being examined supersonic comr,ust ion flow fields (NASl-15491) witn United Technologies under an :;A% contract A conceptual CARS system Researcn Center and Dr. Aian Eckbreth. Scramjet Test Facility has been formudesign f or tne Langley lated.

co NCLUDING REXARKS
considerable progress nas been made Over the past several years, toward tne goai of deveioping a viable airframe-integrated The fixed-geometry inlet has been demonscramjet concept. strated to give good performance over a wide range of Mach numAlso, mixed-moae perpendicular and parallel fuel injecbers. tion has been shown to be an acceptable way to control heat reInitial research tests of two lease over a Mach number range. subscale engine modules have been conducted, and results look In addition, detailed analytical approaches to quite promising. supersonic combustion are now beginning to affect design, and the related technologies of airframe integration, structures, and flow diagnostics have continued to advance. Operation at that many challenges remain. It is clear, however, high fuel-air ratios must be achieved without inlet-combustor Tne aerothermodynamic design must be optimized, interactions. Ultialthough the detailed design might be mission dependent. the airframe-integrated scramjet tecnnology must be demmately, Flight tests will onstrated both on the ground and in flight. be required since the vehicle is actualiy an important part of the propulsion system. Such demonstrations are expected to verify the viability of an airframe-integrated scramjet propulsion system option for future systems. REFERENCES Supersonic 1. Waltrup, P. J.; Anderson, G. Y.; and Stull, F. D.: Combustion Ramjet (Scramjet) Engine Development in the Proceedings of Third International SympoUnited States. sium on Air Breathing Engines, DGLR-Fachbuch, Vol. 6, Deutsche Gesellschaft fuer Luftund Raumfanrt, Coiogne, 1976, pp. 835-861. 2. Billig, F. S.; Waltrup, P. J.; and Stockridge, R. D.: Tne Integral-Rocket, Dual-Combustion Ramjet: A New Propulsion Fourth International Symposium on Air Breathing Concept. Engines, American Institute of Aeronautics and Astronau1979, AIAA Paper 79-7044, pp. 433-444. tics, Inc., Jones, R. A.; Astronautics PPW 38-48. and Huber, P. W.: and Aeronautics, Toward Scramjet vol. 16, no. 2, Aircraft. Feb. 1578,

3.

398

4.

Airframe-Integrated PropulJones, R. A.; and Huber, P. W.: sion System for Hypersonic Cruise Vehicles. Recent Advances in Structures for Hypersonic Flight, NASA CP-20651, 1978, pp. 39-45. Pt.

Design and Performance 5. Trexler, C. A.; and Souders, S. W.: at a Local Mach Number of 6 of an Inlet for an Integrated NASA TN D-7944, 1975. Scramjet Concept. 6. Trexler, C. A.: Inlet Performance of the Integrated AIAA Paper 75-1212, Sept. 1975. Scramjet Module. 7. Anderson, pulsion, 8. G. Y.: Hypersonic NASA sP-381, lY75, Propulsion. pp. 453-474. Aeronautical Langley ProSimulate 1974. by a

Russin, W. R.: Air Entering

Performance of a Hydrogen Burner to NASA TN ~-7567, Scramjet Combustors.

9. Eggers, J. M.: Composition Surveys of Test Gas Produced NASA TM X-71Y64, lY74. Hydrogen-Oxygen-Air Burner. 10. Rogers, R. C.; and Eggers, J. M.: Supersonic Hydrogen Injected Perpendicuiar to a Ducted AIAA Paper 73-i322, Nov. 1973. stream.

Combustion of Vitiated Air-

11.

Anderson, G. Y.; et al.: Investigation of Step Fuel Injection for an Integrated Thir:tlodular Seramjet Engine. teenth JANNAF Combustion :\'eeting, Vol. 3, CPIA Publ-281, 1976, pp. 175-18Y. Rogers, Mixing 1977. Influence R. C.: and Comoustion of of Fuel Temperature on Supersonic ilyiroqen. AIAA Paper 77-17, Jan. P. E.: Expioratory Tests of Supersonic Combustion. NASA

12.

13.

Anderson, G. Y.; and Sootier-urn,, Two Strut Fuel Injec%ors for TN D-7581, 1Y74. McClinton, Test of tegrated bustion

14.

Charles R.; anu (;oo=Ierrui;i,, P. 3.: Direct-Connect a Hydrogen-Fueled, Tkree-Strut Injector for an InModular Scramjet Engine. Fourteentn JANNAF ComMeeting, Vol. 2, CPIA PiA.oi-292, 1977, pp. 489-505. Experimentai Concept 1977. for Investigation Scramjet of Applicaa

15.

Anderson, G. Y.; et al.: Swept-Strut Fuel-Injector tion. NASA TN ~-8454,

16.

Northam, G. B.; Trexler, C; A.; and Anderson, i;. Y.: Characterization of a Swept-Strut Hydrogen Fuel Injector for Scramjet Applications. Presenteci at the iJth JANNAF Lambustion Meeting, Newport, Rr,oae Isiani, Sept. l-i-14, iri6. 399

17.

McClinton, verse to 1979.

C. R.: Autoignition of a Supersonic Airstream.

Hydrogen Injected TransAIAA Paper 79-1239, June

18.

Markatos, N. C.; Spalding, D. B.; and Tatcneii, D. G.: Combustion of Hydrogen Injectea into a Supersonic Airstream (The SHIP Computer Program). NASA CR-2802, 1977. Pan, Y. S.: The Development of a Three-Dimensional Partially Elliptic Flow Corn?uter Program for Combustor Research. NASA CR-3057, 1978. Pan, Y. S.; Drummond, J. P.; and McClinton, C. Ii.: Comparison of Two Computer Programs by Predicting Turbulent Mixing of Helium in a Ducted Supersonic Airstream. NASA TP1166, 1978. Evaluation Pan, Y. S.: Flow Computer Program of the SHIP. Three-DimenslonaL NASA TM-74094, Parabolic
i978.

19.

20.

21.

22.

A Moue1 of Transverse Fuel Injection Rogers, R. C.: to the Computation of Supersonic Combustor Flow. Paper 79-0359, Jan. 1979.

Appiied AIAA Sonic

23.

Numerical Solution for Perpendicular Drummond, J. P.: Hydrogen Injection into a Ducted Supersonic Airstream. AIAA J., vol. 17, no. 5, May 1973, pp. 531-533.

24.

E. A.: Initial Wind Tunnel Tests Guy, R. W.; and Mackley, Mach 4 and 7 of a Hydrogen-Burning, Airframe-Integrated Fourth International Symposium on Air Breathing Scramjet. American Institute of Aeronautics and AstronauEngines. 1979, AIAA Paper 79-7045, pp. 347-358. tics, Inc., P. D.: Numerical Methods for Dash, S. M.; and Del Guidice, the Calculation of Three-Dimensional Nozzle Exhaust Flow Aerodynamic Analyses Requiring Advanced ComFields. 1575, pp. 655-701. puters, Part 1, NASA SP-347-Pt-1,
Cubbage,

at

25.

26.

J. M.; Talcott, Exhaust Simulation for and Aerodynamic Tests.

N.

A.; and Hunt, J. L.: Scramjet Hypersonic Aircraft Nozzle Design AIAA Paper 77-82, Jan. 1977.

27.

Recent Advances in Convectively Cooled Kelly, H. N.; et al.: Engine and Airframe Structures for Hypersonic Flight. Eleventh Congress of the International Council of the Aeronautical Sciences, J. Singer and R. Staufenbiel, eds., International Council of the Aeronautical Sciences SecreWest Germany, 1978, pp. 137-151. Cologne, tariat,

400

28.

Thermal-Structurai Design Study Killackey, J. J.; et al.: (AIRESEARCH-77-13967, an Airframe-Integrated Scramjet. NASl-13984.) AiResearch Mfg. Co.: NASA Contract NASA CR-145368, 1978. The Advanced Development of Burde, D. H.; and Neer, M. E.: High Frequency Response Spectroscopic Probe for Analysis (ARAP-335, Aeronauof Supersonic Combustor Flow Fields. tical Research Associates of Princeton, Inc.; NASA ConNASA CR-145363, lY78. tract NASl-14853.)

of

29.

401

PROPULSION ALTERNATIVES

5000

4000

-_ -L---[--.

_-- _ _----_ ---__--

3000 SPECIFIC IMPULSE,

set

2000

--I-- RAMJETS

1000 CARBON
J J

6 MACHNUMBER

IO

12

Figure XII-I

AIRFRAME-INTEGRATED SUPERSONIC COMBUSTION RAMJET

iid

,-COMBUSTOR

CROSS SECTION OF STRUT REGION

Figure XII-2

402

AIRFRAME-INTEGRATED INLETMODEL

Figure XII-3

DIRECT-CONNECT AUTOIGNITION MODEL

COMBUSTION HEATER

SIDE VIEW

INJECTOR DDAIL

,-S I DEWALL
wr
VILW

FigureXII-4
403

EFFECT OF INJECTOR LOCATION ON IGNITION

X/h

= -1.67

X/h,

= 1.0 FigureXII-5

X/h,

= 3.0

FUELINJECTOR FLOW-FIELD SCHEMATIC

FigureXII-6

404

VELOCITY VECTORS FROM ELLIPTIC ANALYSIS

FliEL

Figure XII-7

FUEL MASS CONTOURS FROM ELLIPTIC ANALYSIS

CONSTANT MASS FRACTION

FUEL

Figure XII-8
405

COMPONENT INTEGRATION MODEL

FigureXII-9

LANGLEY SCRAMJET TEST FACILITY

Figure XII-10

406

PRELIMINARY SCRAMJET ENGINE PERFORMANCE

h

I &.aAT-. DIRECTTHRUST MEASUREMENT AND SIMULATED AIRCRAFTFOREBODY INTEGRATION 800 600

AF

MACH4

160

MACH7

120 80 40
0 40 80 120

Ibs

400 200 0
L

ALT. = 70 000 ft
40 80 120

(PERCENT STOICHIOMETRIC)

Figure XII-11

407

XIII.

VERTICAL TAKEOFF AND LANDING PROPULSION TECHNOLOGY C. Ciepluch, John M. Abbott, Royce and James F. Sellers

(VTOL)

Carl

D. Moore,

National

Aeronautics and Space Administration Lewis Research Center

The propulsion systems of vertical-takeoff-and-landing (VTOL) Accordingly, an aircraft present some challenging problems. adequate propulsion technology base will be a key factor in developing these aircraft for either military or commercial use. This paper identifies the unique propulsion problems and adIt also disvanced-technology requirements of VTOL aircraft. cusses programs in this area currently under way at the Lewis Research Center. VTOL PROPULSION TECHNOLOGY REQUIREMENTS The propulsion-system technical challenge is illustrated by sevFigure XIII-1 shows one of several eral VTOL aircraft concepts. Verconcepts for a Navy subsonic, multimission VTOL aircraft. tical lift is obtained by rotating the engine nacelles to a vertical attitude and by incorporating a lift fan in the aircraft nose. This concept is usually referred to as the tiltingnacelle approach. In addition to providing the aircraft lift, the VTOL propulsion system must provide aircraft control during landing and takeoff. This is a new and relatively complex requirement for airRoll is controlled craft propulsion systems and their controls. by modulating the thrust of the engines on either side of the fuselage. Pitch is controlled by modulating the thrust of the Aircraft yaw is nose fan relative to that of the main engines. usually controlled by using aerodynamic vanes in the propulsion exhaust. Another unique requirement of the VTOL propulsion system is that its thrust response must be much faster than that available by the usual method of varying engine speed. Fast thrust response can be obtained by using either a variable-pitch fan or a variable-inlet-guide-vane fan. The Navy aircraft main engine core (fig. has to XIII-l) be shut is designed to down or fails. land safely The engine if a fans
409

are mechanically interconnected through shafts and gearing, and The the cores are sized so that either could land the aircraft. power for driving the nose fan is also provided through shafting This shafting and gearing must be capable of and gearing. transmitting 15 000 horsepower and also must be lightweight and The gear and bearing research reported in paper VIII, reliable. MECHANICAL COMPONENTS, of this conference publication is contributing to the VTOL propulsion-system technology base. During landing and takeoff the vertical attitude coupled with the aircraft forward velocity results severe inlet flow angle of attack. Lightweight, vanced-technology inlets a're therefore necessary pressure recovery and low flow distortion to the the nacelles in a very compact adfor acceptable fan. of

Another class of VTOL aircraft is characterized by fixed nacelles rather than the tilting nacelles shown in figure X111-1. A representative fixed-nacelle VTOL aircraft is shown in figure x111-2. Vertical lift is obtained with this type of propulsionsystem installation by deflecting the engine exhaust downward with deflector nozzles. These nozzles must be not only aerodybut also lightweight and compact. They namically efficient, therefore represent another advanced-technology area for VTOL propulsion systems. This particular fixed-nacelle aircraft has a tandem-fan propuleach engine has two interconnectIn this concept, sion system. The deflected flow from the forward-located fan proed fans. duces a lift force forward of the aircraft center of gravity, and the deflected flow from the aft-located fan and the engine Airexhaust produce a lift force aft of the center of gravity. craft pitch is controlled by modulating the forward and aft lift forces. Although the inlet flow environment is less severe for the fixtne tandem-fan propulsion concept illused-nacelle aircraft, trated in figure XIII-2 presents a special inlet problem because two closely coupled inlets are required for each nacelle. These are some of the special propulsion problems associated The following sections of this with subsonic VTOL aircraft. paper present results of the Lewis Research Center VTOL technolTo be discussed are ogy program that addresses these problems. rapid-thrust-modulation high-angle-of-attack-capability inlets, and propulsion-system - aircraft-control fans, integration. INLETS The aerodynamic performance craft is being investigated centrated on the tilting-nacelle 410 of at inlets Lewis. inlet for all types of VTOL airThis discussion is conbecause it encounters

more extreme tilting-nacelle configurations: the nacelles operation. as it would proach. The descending, 1200. Inlet

conditions than the fixed-nacelle inlet. A V/STOL aircraft is shown in figure XIII-3 in two with the nacelles level for cruise, and with tilted back 900 as they would be during vertical For this particular example, the aircraft is shown be configured during a portion of the landing apnacelles are tilted back 900, the aircraft is and the resultant inflow angle to the inlet is

flow

Requirements

A typical landing-approach flightpath for a tilting-nacelle VTOL aircraft is shown in figure X111-4. As the aircraft begins its landing approach, the nacelles begin to tilt back at a forward velocity VO of about 160 knots. The inlet reaches a maximum This is angle of attack of 1200 at 4U knots forward velocity. The inlet the example that is illustrated in figure X111-3. airflow also varies during the approach with the necessary variation in engine thrust. The primary objective of the inlet designer for a tiltingnacelle aircraft is to design an inlet that provides separationInlet free airflow to the engine at all operating conditions. internal-flow separation is to be avoided because the resulting sudden high pressure losses and high pressure distortions can lead to sudden changes in engine thrust and to fan or compressor blade stresses. The inlet is most likely to encounter internal-flow separation at a forward velocity of 120 knots, an inlet angle of attack of 600, and an inlet This flow airflow of 56 percent of maximum. condition is labeled "critical" in figure X111-4. An inlet that operates separation free at this critical flow condition will operate separation free at all other flow conditions encountered during approach and takeoff. At conditions prior to the inlet angle of attack is lower and the critical condition, inlet operation is therefore more stable. At conditions beyond the forward velocity is lower and the this critical condition, inlet airflow is higher. Therefore these iater conditions are less critical, even though the inlet angle of attack is actually higher. For the remainder of the discussion, this critical point in the landing approach is called the tilting-nacelleinlet design requirement. Inlet Performance with unseparated first approach.that lip thickness to internal comes effectively flow to 411

In designing an inlet to operate at this critical condition, the mind is to provide enough inlet

turn the flow into the inlet (refs. 1 to 6). how much lip thickness is needed for effective Lewis conducted an experimental test program 20-inch-diameter inlets (refs. 7 to 11).

To determine just flow turning, on both 12- and

The 20-inch-diameter inlets were tested on a 20-inch-diameter model fan installed in the Lewis 9- by 15-Foot Low-Speed Wind The model fan has a pressure ratio of Tunnel (fig. X111-5). about 1.2 and is powered by a core turbine driven by highpressure air supplied through the support post and pylon. A typical tilting-nacelle inlet is shown installed on the fan. The inlet was instrumented to determine its aerodynamic performThe model-fan blades were also instrumented with strain ance. and blade stress data were gathered as part of the regages, search data. The results of the test program to determine lip-thickness requirements for tilting-nacelle inlets are shown in figure XIIIThe figure illustrates how the inlet-lip contraction ratio 6. affects the angle of attack at which the internal flow sepais defined as the The lip contraction ratio rates. AHL/AT ratio of the inlet highlight or leading-edge area AHL to the inlet throat area AT and is hence a direct indication of The higher the contraction ratio, the inlet-lip thickness. The data are for the critical flow condition thicker the lip. a forward velocity of 120 knots and an inlet described earlier: The inlet angle of attack at airflow of 56 percent of maximum. which the internal flow separates from the inlet lower lip is plotted against the inlet-lip contraction ratio. aSEP The horizontal line drawn on the figure is the 6U0-angle-ofattack tilting-nacelle-inlet design requirement. The data in figure XIII-6 suggest that for separation-free inlet Thinner the lip contraction ratio must be about 1.65. flow, those with contraction ratios below this value, have seplips, arated internal flow at the critical 600 angle of attack. by making the inlet lip thick enough, a tilting-nacelle Hence, inlet can be designed to operate at this most critical condition It will then, of course, opin the aircraft landing approach. erate effectively over the entire approach and takeoff flightpaths. Effect of Inlet Flow Separation on Fan Blade Stress

How the fan blade stress is affected by separated inlet flow LS Blade stress levels from the 20-incnshown in figure X111-7. diameter model fan are plotted against inlet-air weight flow. The data are for the same flow conditions as tnose in figure Two curves are shown: one for a relatively thin inlet X111-6. and one for a relatively tnick with a contraction ratio of 1.46, 'The 1.46-contractioninlet with a contraction ratio of 1.76. 412

The resultratio inlet has separated flow at these conditions. ing fan blade stresses increase witn increasing inlet-air weight flow and exceed the limit at the required 56 percent airflow. The thicker, 1.76-contraction-ratio inlet initially has separated flow at low inlet-air weight flows and the fan blade stresses begin to increase. But the flow attaches before the critical inlet weight flow of 56 percent is reacned, and the The spike in the blade stress then decreases to low levels. blade stress curve after the initial attachment point corresponds to a first-bending-mode excitation of the fan blades. It results from the inlet flow not being completely attached along the entire length of the inlet internal surface. The point to be made is that, when the inlet flow is separated, the fan blade stresses can be unacceptable and, when the inlet flow is attached, fan blade stresses are not a problem. Methods to Thickness Prevent Flow Separation and Minimize Inlet-Lip

Lip thickness can be increased to the point where an inlet will meet the requirements of a tilting-nacelle VTOL aircraft. A thicker lip, however, does have some disadvantages in that . weight, surface area, and hence cruise drag will increase. These disadvantages can be minimized by making the inlet unsymwhere angle of attack is impormetrical - thick on the bottom, tant, and thinner on the sides and top. In the interst of reducing inlet weight and surface area even further, however, it would be desirable to somehow make the lower lip as thin as possible. That leads to considering methods by which the angle-of-attack capability of an inlet with a given lower lip thickness can be extended. Several methods are illustrated in figure X111-8: an extended centerbody, lower-lip blowing, a scarf inlet, lower-lip suction, a lower-lip slot, and lower-lip vortex generators. Each method is intended to improve the inlet lower-lip performance to the point where a given angle-of-attack capability can be attained with a thinner lip. Results for the first three methods are discussed here; the performance of the other methods will be evaluated in future tests. Extended centerbody. - The effect of extending the inlet centerbody on inlet flow separation is shown in figure XIII-g. The data are shown in the same format and at the same flow conditions as figures XIII-6 and X111-7, with the angle of attack at flow separation being plotted against the lip contraction ratio. The lip-thickness-effect data with a short centerbody are repeated here and labeled as "reference." As indicated, extending the centerbody forward increases the flow separation angle by about loo for a 1.46-contraction-ratio inlet - a significant improvement. The method works by changing the area distribution within the inlet duct to make it more favorable to 413

the inlet internal Additional angle.

flow, results

thereby extending are reported in

the separation reference 8.

of lower-lip blowing is shown in Lower-lip blowinq. - The effect tigure X111-10. The plot format is again the same, and the data indicate that lower-lip blowing can extend the separation angle for a 1.46-contraction-ratio inlet by an impressive 200 to a This value very close to the tilting-nacelle-inlet requirement. large improvement in angle-of-attack capability results from the flowing air energizing the lower-lip boundary layer and thereby This result was atmaking the flow less likely to separate. tained with a blowing-air pressure of 1.1 times atmospheric and a blowing-air mass flow of about 6 percent of the inlet flow. Additional results are reported in reference 12. Scarf inlet. - The third method of extending the inlet flow sepThe data indicate aration angle is shown in figure X111-11. that extending the inlet lower lip to form a scarf inlet can increase the flow separation angle for a 1.44-contraction-ratio inlet by about 200 - the same amount that was attained with This method works because a scarf inlet lower-lip blowing. draws in air more from above than from below. This effectively reduces the local angle of attack on the lower lip of the inlet and thereby permits a higher inlet angle of attack to be attainAdditional results are ed before the internal flow separates. reported in reference 13. - As stated previously, these different Benefit of thin lip. methods for extending the flow-separation angle of attack are being investigated so that a thinner inlet lip can be used to As an exammeet the tilting-nacelle-inlet design requirement. ple of the potential benefit of a thinner inlet lip, figure XIII-12 shows two nacelle layouts - the top one without lip blowing, the bottom one with lip blowing. Both inlets have been designed to provide exactly the same angle-of-attack capabilWithout lip blowing, the inlet lower lip is relatively ity. and it is therefore relatively heavy and has higher thick, cruise drag. With lip blowing, however, the angle-of-attack requirement is attained with a thinner lip and the inlet is In addition, lighter and has less cruise drag. the lip-blowing The inlet thickness also affects the inlet is somewhat shorter. nacelle thickness aft of the inlet. All in all, from a nacelleit is highly desirable to weight and cruise-drag standpoint, design an inlet that meets the tilting-nacelle-inlet flow design requirement and is as thin as possible. In summary, the tilting-nacelle VTOL aircraft concept presents the most challenging low-speed inlet design problems. Tests at Lewis have shown that, by increasing inlet lower-lip thickness the inlet can be made to provide separation-free interenough, nal flow over the full aircraft operating range. Other test results have shown the promising potential of various methods to 414

provide a given angle-of-attack capability lip. A strong motivation exists to use thinness results in reductions in overall length and in cruise drag.

with a thinner thin inlet lips nacelie weight

inlet since and

Lewis is also doing some work on non-tilting-nacelle inlets. A fixed-nacelle VTOL inlet that has been tested at Lewis is shown in figure X111-13. This is a tandem-fan propulsion system inlet model that is compatible with the fixed-nacelle, tandem-fan VTOL aircraft described previously. The model is shown installed in the Lewis lo- by lo-Foot Supersonic Wind Tunnel, where low-speed tests were just recently completed. The model is composed of two inlets - the front axisymmetric inlet for the front fan, and the top S-duct inlet for the aft fan. Only the top inlet has been tested, in isolation, with the front inlet faired over. The test results indicated only a 4 percent loss in total pressure and an 8 percent distortion in total pressure at the most extreme condition tested. Plans include testing the model as shown in the figure with two fans, one for each inlet, in order to determine the aerodynamic interaction of the two inlets. RAPID-THRUST-MODULATION FANS

As already stated in this paper, either a variable-pitch fan or a variable-inlet-guide-vane (IGV) fan is being considered to provide the rapid thrust modulation needed for aircraft stability and roll control. A sketch of a variable-pitch fan is shown in figure X111-14. The variable-pitch mechanism, located in the rotor hub region, is quite complicated. There has been some concern about the durability and reliability of variabie-pitch rotor blades, and therefore some approaches favor a fixed-pitch fan with variable IGV's. A sketch of a variable-inlet-guidevane fan stage is shown in figure X111-15. The rotor blades are at a fixed setting angle. The IGV angle-changing mechanism is much simpler;' a geared motor and a synchronous ring are shown. Since the mechanism is located on the outer casing, the engine thrust controls can easily be incorporated into the variable-IGV systems. All VTOL two-engine configurations must be capable of safe operation with one engine out. The fan stages in both nacelles must be capable of being driven by one engine and must each provide the required thrust for all modes of operation. The thrust level of either propulsion system can be controlled individually by adjusting its variable blade during constant-speed, oneengine-out operation. Both a variable-pitch-rotor-blade fan stage and a variable-IGV fan stage were tested at Lewis in order to determine the feasibility of such designs for rapid thrust modulation (refs. 14 and 15).

415

Variable-pitch fan staqes. - The variable-pitch fan stage was designed and fabricated by the Hamilton Standard Division of The stage was designed for a pressure United Technologies Corp. ratio of 1.38 at a flow of 65.3 pounds per second, which corresponds to a specific flow of 41 pounds per second per square foot of annulus area. The design tip speed was 950 feet per second. The rotor is shown in figure X111-16, and the stators are shown mounted in the casing in figure XIII-17. The 20-inch-diameter rotor has a hub-tip ratio of 0.52. The 19 variable-pitch rotor blades have an aspect ratio of 1.26. The rotor-blade setting angle was changed manually. The 38 stator blades were spaced about 2 rotor chords downstream of the rotor trailing edge. So that the variable-pitch rotor blades can turn to feather, tne blade tip contour is basically a circular arc (fig. X111-18) in the chordwise direction. In an attempt to reduce the effects of the casing above the rotor was recessed as tip clearance, The radius at the blade leading and trailing edges is shown. At the blade radial axis of equal to the nominal casing radius. the blade tip radius is greater than the nominal casrotation, In an actual engine application, this configuration ing radius. has the disadvantage that a split casing would have to oe used to allow for rotor removal. - Hamilton Standard designed a Variable-inlet-guide-vane fan. X111-19) to be added to the variable-pitch set of IGV's (fig. stage so that the effects of variable IGV's on performance could NACA-63-series airfoil shapes where used for be investigated. The front portion of the vanes was fixed and the IGV elements. For this series of tests, the only the rear portion rotated. Botn stage rotor blades were set at an angle of 60 closed. in the Lewis single-stage compressor configurations were tested TUKBOMACHINEKY TECHNOLOGY, test facility described in paper VII, of this conference publication. Performance with Variable-Pitch Fan

The performance with the variable-pitcn fan is shown in figure x111-20. Pressure ratio is plotted as a function of weight flow The data presented are for several delta blade setting angles. Also shown in the figure are the stall line for design speed. and an assumed operating line that corresponds to a constantThe definition of blade settiny angle throttle valve position. Positive setting angles close is presented in figure X111-21. tile blade passages from the design setting and reuuce the flow. tiegative setting angles open the blade passages and increase the At the design setting angle (Oo), the measured performflow. Howance agrees reasonably well with design (fig. XIII-2U). The stall margin the stall margin is only 5 percent. ever, 416

based on the operating-line performance is iU percent. As snown the fan can operate over a wide range in the performance plots, of blade setting angles. At high positive angles, the operating line crosses the low-pressure-ratio side of the performance curves. At -80, the operating line is very close to the stall line with only about a 2 percent stall margin. The effect of blade figure X111-22. Tne plotted as a function static thrust varies to about 1800 pounds setting angle on static thrust is snown in operating-line calculated static thrust is of the delta blade setting angle. The from about 100 pounds at an angle of 250 at an angle of -80. while this anthe

Variable-pitch rotor blades show promise for VTOL application because the thrust of each fan can be independently varied the engine is operating at a constant speed. However, for particular fan, the stall margin at. negative blade setting gles is low. If the expected inlet flow distortions occur, stall margin will be inadequate.

In an effort to improve the stall margin, several modifications First, were made to the rotor (ref. 14). the blade leading edge was recoined in the tip region. This resulted in only a slight Next (fig. X111-23), the recessed improvement in stall margin. contour above the rotor was replaced by a cylindrical Insert and the blade tips were trimmed. the skewed-slot casing Finally, treatment shown in figure XIII-24 was inserted above the rotor This insert replaced the cylindrical solid insert shown in tip. figure X111-23. 'The axial location of the fence separating the two rows of slots corresponds to the rotor-blade axis of rotation. Its slots are skewed 600 from radial in the direction The performance with both the solid casing and tne of rotation. skewed-slot casing treatment is presented in figure X111-25. Pressure ratio is plotted as a function of weight flow for three blade setting angles. The operating line is the same as tnat shown in figure X111-20. At the high flow blade angles, casing treatment substantially increased stall margin. At 150, casing treatment did not affect the stall point. The blade-element data show that the rotor-tip element was not controlling stall at tnat angle, and thus the casiny treatment provided no benefit. At the design angle, the casing treatment increased tne stall margin by 13 percent. However, there was a 4 percent loss in efficiency. Further studies are required to determine a casing treatment that will give the improved stall margin with better efficiency. Performance with Variable-Inlet-Guide-Vane Fan Stage

The effects of the variable-inlet-guide-van& fan stage on overall performance are shown in figure XIII-2b. Pressure ratio at design speed is plotted as a function of weight flow at several 417

The stall line and the operating line are also IGV angles. The operating line corresponds to a constant throttleshown. and it is different from that obtained with valve position, variable-pitch stage.

the

The definitions of vane setting angle are illustrated in figure Positive IGV angles turn the flow in the direction of X111-27. Negative IGV blade rotation and close tne rotor-blade passage. angles turn the flow opposite to the direction of rotation. The operating line of the variable-inlet-guide-vane stage is At the high flows, the stage limited at low flows by stall. experiences a maximum flow condition, as iliustrated by the -1-I. So and -25O IGV angles having the same performance (fig. This is a result of the very high losses associated X111-26). Figure XIII-28 shows the negative IGV angles. with the high, effect of vane setting angle on the calculated static thrust and The compares the tnrust with that for the variable-pitch stage. as a percentage of that obtained operating-line static thrust, is plotted as a function of delta setting at design angle, The change in thrust with angle is not as great with the angle. At the positive angles, IGV's as with the variable-pitch rotor. at the negative angles, thrust is thrust is limited by stall; limited by choke. both the variable-pitch and variable-inlet-guideIn summary, vane fan stages show promise for VTOL application because the thrust of each fan can be independently varied while the engine The type of fan stage selectis operated at a constant speed. The varied will probably depend on the specific application. However, able-pitch fan gives a much greater thrust modulation. if the required thrust modulation is about +2U percent, the variable-inlet-guide-vane fan might oe chosen because of its it is obvious that much From the data shown, simpler design. more work is needed before the optimum fan configuration for a paricular application can be identified. In addition to the model-fan program just described, Lewis also has in progress a large-scale evaluation of the variable-IGV A variable-IGV module has been inthrust modulating concept. as shown in figure X111-29. The stalled in the TF-34 engine, variable IGV's are designed to affect only the bypass flow and This is to prevent the core from not engine-core supercharging. being desupercharged when the guide vanes are adjusted to reduce engine thrust. The variable IGV's are shown in figure X111-30. This view of the engine front face shows the inner ring to which the end of each guide vane is attached. The trailing edges of the vanes articulate, but the leading edges are fixed. The thrust-modulating experiments are being conducted in one of altitude test facilities. The engine installation is Lewis' shown in figure XIII-31. This program, which has been under way 418

for only several months, thrust modulation range, range of inlet-guide-vane PROPULSION CONTROL

is investigating the variable-IGV-fan stall margin, and blade stress over deflection angles.

Some of the propulsion control problems of VTOL aircraft have already been discussed, for example, the use of variablegeometry fans to obtain rapid thrust response. Some of the methods used to integrate the controls of these complex propulsion systems are discussed here. During hovering flight, a VTOL propulsion system has to do the same job that the wings and control surfaces do during conventional flight. Therefore a much higher degree of aircraft - propulsion control integration is required in VTOL aircraft than in more conventionai aircraft. Piloted Simulators

The success of the control-system design for a VTOL airplane is measured by the pilot's ability to land the airplane safely on a small landing pad. In studying this problem, it is necessary to consult the pilot for his opinion on aircraft handling qualities. The safest and most economical way to do this is to use piloted simulators like the one shown in figure X111-32. This facility is located at the NASA Ames Research Center and is called the flight simulator for advanced aircraft (FSAA). The pilot sits inside the enclosed cab at the left of the figure. The cab is driven by actuators that simulate actual aircraft moments. The actuators are positioned by a digital computer that solves the aircraft equations of motion. Inside the cab, the pilot is provided with a complete set of controls and displays. His view of the landing area is provided by a television screen. This screen displays a picture taken by The a camera that traverses a scale model of the landing area. television camera is operated by the same computer that operates the cab. For proper simulation of VTOL operations, the FSAA's computer should be capable of solving the equations that model propulsion-system dynamics, ground effects, and even ship motion for Navy operations. the computer does not have enough At present, capacity to handle all these calculations in real time. Plans are to expand the computer's capacity; however, for the present, investigations have been confined to propulsion systems that are somewhat simpler than those required for VTOL aircraft. These investigations are nonetheless relevant to VTOL research since VTOL aircraft belong to the general class of powered-lift aircraft. This class of aircraft includes STOL aircraft and heli419

copters as well as VTOL aircraft, share many similar problems.

and

these

types

of

aircraft

For all these types of aircraft, power modulation is used to control aircraft lift. This implies that aircraft maneuvers cause changes in engine speeds, temperatures, and surge marMonitoring of these internal engine variables is imporgins. tant for an adequate integrated-control-system design. The pilot becomes an important element in the overall system, and consequently pilot opinion has a critical effect on propulsion conThe pilot has to be satisfied with the trol requirements. thrust response since he uses the engines to maneuver the airEngine failures are also a critical problem for all powplane. ered-lift aircraft since engine failures limit the lift available for controlling the flightpath. STOL Aircraft Simulation

These problems were investigated by performing a piloted simulation of the STOL propulsion system shown in figure XIII-33. This system is the Quiet Clean Short-Haul Experimental Engine has such VTOL technology features as a (QCSEE) . This engine variable-pitch fan and a digital control. The hydromechanical backup control is designed to provide engine protection in case Since QCSEE was designed for a of a digital control failure. it was a good candidate for a piloted powered-lift aircraft, In cooperation witn NASA Ames, the Lewis simulator evaluation. Research Center developed a simulation of QCSEE suitable for use on the FSAA. The engine simulation was combined with a simulation of the aircraft snown in figure X111-34. This is an externally blown-flap As for a VTOL airplane, one of its critical STOL airplane. This parproblems is engine failure during a landing approach. ticular airplane experiences rolling and yawiny moments after an engine failure. The piloted simulator is quite safe, in contrast to actual airso it is an ideal tool for investigating craft flight testing, the pilot's reaction to propulsion-system faiiures. A number of failure modes were investigated in this experiment, incluaing failure of the QCSEE digital control and engine failure. The pilot could easiiy compensate for failure of the (,&SEE digital The hydromechanical backup system proved adequate to control. prevent a serious loss of thrust. Total engine failure, which did cause a thrust loss, was considerably more difficuit for tne Some results pilot to handle. of simulated engine failures are discussed here because they illustrate the importance of the pilot's role in overall system performance and because tney denonstrate the need for adequate modeling of internal engine behavior. 420

A typical result of an engine failure is shown in figure XIII35. The upper trace shows distance from the center of the runway in feet as a function of time. The lower trace shows changes in engine turbine-inlet temperature. The vertical dotted line indicates when engine failure occurs. The pilot's first warning of failure comes from a light that illuminates when the engine pressure ratio drops below a specified value. He must compensate for the failure manually. The engine failure causes the airplane to veer off the center of the runway by There is a slight increase in turbine-inlet temabout 80 feet. perature about 1 second after the failure, when the pilot advances the throttle. For this flight, the temperature variations were small but flightpath control was marginal. An automatic thrust control system was devised to help the pilot correct for engine failure. The automatic system was designed to keep the sum of all four engine thrusts at a level requested by the pilot. The automatic system was engaged at all times; failed, it would automatically increase the sor when an engine thrusts of the working engines. The reThis system was evaluated with the piloted simulator. sults are shown in figure X111-36. When unexpected failures are simulated, it is impossible to draw any final conclusions from but some general observations the results of only two flights, First, can be made. it appears that the automatic system was helpful since the airplane stayed near the center of the runway. From the flightpath information alone, the automatic syssuccessful. this conclusion tem appears to be highly However, is questionable when the time history of turbine-inlet temperature is considered. The range of temperature variation is about 900 degrees F, much greater than for the manual system. The reason for this was the high gain of the automatic control and also the fact that the automatic system felt different to As a result, the pilot. his throttle activity was greater than with the manual system. Another surprising result was that the pilot used the throttle more before the failure than after. When the failure occurred, he turned his attention to the flight controls and let the automatic system take care of thrust. At present, it is not possible to determine exactly what these temperature variations mean. A likely problem is an excessive reduction of engine thermal fatigue life. Lewis has begun a study to determine whether the complex process of engine fatigue life prediction can be simplified enough to provide system design criteria. Future Controls Research sections are for a STOL The problems will be 421

The results discussed in the preceding aircraft landing on a 2000-foot runway.

Lewis even more severe for VTOL aircraft landing on small pads. is currently trying to simulate more complex VTOL propulsion systems, like the one shown in figure X111-37. This system includes variable-pitch fans connected by cross shafts and highangle-of-attack inlets. In cooperation with NASA Ames, Lewis will investigate problems similar to those discovered in the STOL aircraft simulation. These problems include thrust response, engine failure, and low-cycle fatigue. Lewis will also continue to use the piloted simulator as a tool for investigating VTOL propulsion control problems without the safety hazards and cost of actual aircraft flight testing. SUPERSONIC VTOL PROPULSION TECHNOLOGY The previous discussion dealt primarily with subsonic VTOL airBecause of the military's interest craft technology programs. Lewis is also conducting advancedin VTOL fighter aircraft, technology programs applicable to supersonic VTOL propulsion Work is beginning on inlets, nozzles, and propulsion systems. controls. The inlet must function efficiently at both supersonic and subsonic speeds, where it will encounter a relatively severe angle-of-attack environment during takeoff and landing. Ejectors could improve the performance of remote burner nozzles and aircraft control jets. The ejector could also reduce the problems associated with remote burner hot-gas impingement and reingestion during takeoff and landing. And finally, integrabe tion of the propulsion system and the aircraft controls will and programs to improve control-system reliability will studied, be undertaken. CONCLUDING REMARKS This paper has reviewed the unique propulsion problems and technology requirements associated with VTOL aircraft. The Lewis Research Center is currently active in technological development applicable to subsonic VTOL aircraft and is also beginning work applicable to supersonic VTOL aircraft. The objective of this overall program is to provide an adequate technology base so that the development of military and commercial VTOL aircraft can be undertaken with a reasonable degree of risk. REFERENCES 1. Albers, James A.; Stockman, Norbert 0.; and Hirn, John J.: Aerodynamic Analysis of Several High Throat Mach Number Inlets for the Quiet Clean Short-Haul Experimental Engine. NASA TM X-3183, 1975.

422

2.

Potential and Viscous Flow in VTOL, Stockman, Norbert 0.: STOL, or CTOL Propulsion System Inlets. AIAA Paper 75-1186, Sept. 1975. Jakubowski, A. K.; and Luidens, Separation at High Incidence Jan. 1975. Boles, M. A.; and Stockman, Limits in the Theoretical Paper 77-878, July 1977. R. W.: Angles. Internal CowlAIAA Paper 75-64,

3.

4.

N. 0.: Design

Use of Experimental of V/STOL Inlets.

AIAA

5.

Chou, D. C.; Luidens, R. W.; and Stockman, of Laminar and Turbulent Boundary Layer V/STOL Engine Inlets. NASA TM X-73575,

Prediction N. 0.: Flow Separation in 1977. Study NASA

6.

Theoretical and Stockman, Norbert 0.: Hawk, J. Dennis: of VTOL Tilt-Nacelle Axisymmetric Inlet Geometries. TP-1380, 1979.

7. Miller, Brent A.; Dastoli, Benjamin J.; and Wesoky, Howard L . : Effect of Entry-Lip Design on Aerodynamics and Acoustics of High Throat-Mach-Number Inlets for the Quiet, NASA TM X-3222, Clean, Short-Haul Experimental Engine. 1975.
8.

Effect of Lip and Centerbody Geometry Burley, R. R.: Aerodynamic Performance of Inlets for Tilting-Nacelle Aircraft. AIAA Paper 79-0381, Jan. 1979. Shaw, R. J.; Williams, R. C.; and Koncsek, Nacelle Aerodynamics and Its Relation to AIAA Paper 78-958, July 1978. Stresses.

on VTOL

9.

VSTOL Tilt J. L.: Fan Blade Design AIAA and

10.

Potonides, H. C.; Cea, R. A.: and Nelson, T. F.: Experimental Studies of a Type 'A' V/STOL Inlet. Paper 78-956, July 1978.

11.

Abbott, John M.; Diedrich, James H.; and Williams, Robert Low-Speed Aerodynamic Performance of 50.8-CentimeterDiameter Noise-Suppressing Inlets for the Quiet, Clean, NASA TP-1178, Short-Haul Experimental Engine (QCSEE). 1978. Proj. FEDD. The Effect of a Diffuser Blowing on the Johns, A. L.: AIAA Performance of a Short V/STOL Tilt Nacelle Inlet. Paper 79-1163, June 1979.

C.:

12.

423

13. 14.

Aerodynamic Abbott, J. M.: AIAA Paper 79-0380, Jan.

Performance 1979.

of

Scarf

Inlets.

Moore, Royce D.; and Osborn, Walter formance of a 1.38-Pressure-Ratio, NASA TP-1502, 1979. Stage. Moore, Royce C.; and Reid, Lonnie: of Axial-Flow Fan Stage Operated NASA TP-1510, 1979. Angles. at

M.: Aerodynamic PerVariable-Pitch Fan Aerodynamic Nine Inlet Performance Guide Vane

15.

424

REPRESENTATIVE TILTING-NACELLE VTOL AIRCRAFT

CS-79-1648

Figure XIII-l

REPRESENTATIVE FIXED-NACELLE VTOL AIRCRAFT

CS-79-1647

Figure XIII-2 425

TILTING-NACELLE WSTOL AIRCRAFT

/
0 FigureXIII-3

APPROACH
cs-79-1717

APPROACH FLIGHTPATH
V KN8iS >160/ WEIGHT FLOW, 2 100

120 4od-al

1202-fCkL.< \ /
L CRITICAL I / 82
cs-79-1718

FigureXIII-4

426

FigureXIII-5

EFFECT OF INLETLIP CONTRACTION RATIO ON FLOW SEPARATION

'SEP

KNOTS

ANGLE OF AllACK AT FLOW SEPARATION, 'SEPl wl

cs-79-1713

1.4 1.5 1.6 1.7 1.8 LIP CONTRACTION RATIO, AHLIAT FigureXIII-6

427

FANBLADE STRESS
LIP CONTRACllON RATlO loo r 80 BLADE STRESS, % OF LIMIT 60 t ,! $0 KNOTS / 1.46

SEP //In

INLET WEIGHT FLOW, ;, % Figure XIII -7

cs-79-1711

METHODS TO PREVENT FLOW SEPARATION

EXTENDED CENTERBODY

BLOWING

SCARF

SUCTION
CS-79-1716

SLOT

VORTEX GENERATOR

Figure XIII-8

428

EFFECT OF EXTENDED CENTERBODY

V, = 120 KNOTS G=56$ 80/ EXTENDED CENTERBODY ti#@ v\ RF~#e-6 _. *e-

60 7 ANGLE OF AllACK AT FLOW SEPARATION, deg 40 -Qfe-

REQ

-0

202
cs-79-1715

1.4

1.5 1.6 1.7 LIP CONTRACTION RATIO Figure XIII-9

1.8

EFFECT OF BLOWING
V, = 120 KNOTS 0=5&s gol6owREQ

ANGLE OF AllACK AT FLOW SEPARAIION, hg

40

2ou 1.4
cs-79-1714

1.5 1.6 1.7 LIP CONTRACTlON RATlO Figure XIII-10

1.8

429

EFFECT OFSCARF INLET

Jo = 120 KNOTS ti = 56% 80

REF ---~\c.0. /H .*-

,( 'REO

ANGLE OF AllACK AT FLOW SEPARATION, deg 40 -*c9,

20/ 1.4
cs-79-1712

1.5 1.6 1.7 LIP CONTRACTlON RATlO Figure XIII-11

1.8

BENEFIT OF BLOWING

-=01-1-f

LIP BLOWING

* FigureXIII-12

LIP BLOWING
cs-79-1710

430

TANDEM

FAN INLET

MODEL

FigureXIII-13

VARIABLE-PITCH FANSTAGE

cs-79-1918

Figure XIII-14
431

VARIABLE-INLET-GUIDE-VANE FANSTAGE

Figure XIII-15

VARIABLE-PITCH ROTOR

FigureXIII-16

432

-.

STATOR

Figure XIII-17

CASING-ROTOR TIP CONTOUR

c.a -- _.. __ ---

FigureXIII-18 433

VARIABLE-INLET-GUIDE-VANE FANSTAGE

FigureXIII-19

EFFECT OF SETTING ANGLE ON PERFORMANCE

DESIGN SPEED

PRESS. RATlO

0 DESIGN ---STALL ----0P LINE

WT FLOW, lblsec Figure XIII-20

cs-79-1917

434

BLADE SETTING ANGLE

POSITIVE DESIGN (0') NEGATlVE

Figure XIII-21

EFFECT OF SETTING ANGLE ON THRUST

20@$OPERATING LINE - DESIGN SPEED

STATIC THRUST, lb

I
40
CC-73-1312

I
-20

30 20 10 0 -10 DELTA BLADE SElTlNG ANGLE, deg FigureXIII-22

435

CYLINDRICAL INSERT

FigureXIII-23

CASING TREATMENTINSERT

-=-P- ROTATION

FigureXIII-24

436

EFFECT OF CASING TREATMENT ON PERFORMANCE

DESIGN SPEED

DELTA BLADE SEl-llNG ANGLE, deg

*-.OP

LINE 1 SOLID - SOLID OPEN -TREATMENTJ CASlNG I 70 I 80 ::--,d-l'?..

1.0' 30

1 40

I I 50 60 WT FLOW, lblsec FigureXIII-25

EFFECT OF VANEANGLE ON PERFORMANCE

DESIGN SPEED /

e-w STALL .w. OP LINE

.17.5

1.01 30

I 40

I I 50 60 WT FLOW, lblsec FigureXIII-26

I 70

I 80 Sj-,^^ ,-1!1-

437

VANESETTING ANGLE
POSITIVE DESIGN (0'1 NEGATIVE

cs-79-2017

ROTATlON

Figure XIII-27

THRUST VARIATIONS
120100 80 % OF DESIGN ANGLE STATlC THRUST 6040 20 OPERATING LINEDESIGN SPEED

I 30

VARIABLE 0 ROTOR BLADES Cl INLET GUIDE VANES I I I I I I 20 10 0 -10 -20 -30 DELTA SETTING ANGLE, deg I -40

0 40
cs-73-1315

FiqureXIII-28

438

VARIABLEINLETGUIDEVANE ONTF34 ENGINE

VARIABLE INLET GUIDE VANES-,
\

,-FAN I

BLADES

B-79-1646

FigureXIII-29

VARIABLEINLETGUIDEVANES MOUNTED ONTF-34ENGINE

Figure

XIII-30

439

VARIABLE-INLET-GlJIDE-VANE TF-34 ENGINE IN TEST FACILITY

FigureXIII-31

FLIGHT SIMULATOR FOR ADVANCED AIRCRAFT (FSAA)

FigureXIII-32

440

QCSEE EXPERIMENTAL PROPULSION SYSTEM

r HYDROMECHANICALCONTROL DIGITAL CONTROL-,

CS-79-1466

FigureXIII-33

Figure XIII-M

441

APPROACH WITHENGINE FAILURE

WITH VISUAL CUEING ONLY

DISTANCE FROM CENTEROF

UNFAILED ENGINE TLJRBINEINLET TEMPERATURE CHANGE, OF

(x-79-1464

1 ENGINE

10

20

30

40 50 TIME, set

60

70

I 80

Figure XIII-35

APPROACH WITHENGINE FAILURE

WITH THRUST COMMAND COMPENSATION lOOr DISTANCE FROM CENTEROF RUNWAY, ft I

UNFAILED ENGINE WRBINE INLET TEMPERATURE CHANGE, OF

CS-79-1465

T -Y 0 10 20 30

; ENGINE I FAILURE I I I

I 40 50 TIME, set

60

I 70

80

Figure XIII-36

442

CONTROLS ANALYSIS FORSUBSONIC VTOL PROPULSION SYSTEM

CONTROL PROBLEMS: FAST THRUST RESPONSE ENGINE FAILURE LOW CYCLE FATIGUE

cs-79-1463

FigureXIII-37

443

,111.

,111

..--

.--

XIV.

HIGH-PERFORMANCE-VEHICLE Louis A. Povinelli

TECHNOLOGY

National

Aeronautics and Space Administration Lewis Research Center

This paper discusses some of the special needs of one of the most the high-performance military fightimportant types of aircraft, er. Figure XIV-l shows a modern aircraft of this type used by Such aircraft use highly advanced techthe Air Force, the F-15. nology in all systems, especially in propulsion, to perform important functions in national defense. Aircraft of this general class may also constitute a major portion of international sales of U.S. manufacturers. The discussion in this paper is limited to the following propulsion-related topics: (1) inlet performance, (2) nozzle performance and cooling, and (3) afterburner performance. The discussion indicates some NASA activities to and industry to further performance aircraft. INLET PERFORMANCE trends in future concepts and describes supplement those of the military agencies advance the technology required for high-

Figures XIV-2 First consider the type of inlet that is required. to XIV-4 show three current high-performance aircraft and illustrate the fact that inlet geometry differs significantly with Figure XIV-2 shows the Air Force F-16. aircraft design. For the F-16, which is a single-engine aircraft, the inlet has a "smile" configuration at the entrance, is vertically offset, and is Figure XIV-3 shows the F-18, a mounted beneath the fuselage. Navy aircraft. The twin-engine F-18 has a D-shape inlet, the offset is primariy in the horizontal direction, and a wing root mounting is used. In both aircraft, the inlet concept is unique. A special research aircraft is shown new Air Force/NASA highly maneuverable (HiMAT) vehicle for a flight research The inlet has similarities piloted. fuselage mounted, vertically offset, figure XIV-4. It is the aircraft technology program. It is remotely to that for the F-16: it is and "smile" shaped. in

445

Figure XIV-5 gives examples of a concept that may be needed for it illustrates some interesting refuture Air Force missions; sults of recent study contracts of stealthy configurations. In each case, portions of the airframe are used to hide the inlet so that it can be less easily detected by ground radar. Inlets act reflecting radar signals in a very focused manner, as beacons, whereas the convex underside of a fuselage and wing scatter the Again the inlet geometry (fig. XIV-5) is difradar signals. ferent from any other. There are, quirements (1) (2) (3) (4) (5) High however, some common geometrical and operational which can be summarized as follows: for inlets, degree of integration with short and angle airframe length, of large attack offset, re-

Simple, lightweight design, and geometric transition Wide Stallrange and of Mach number

buzz-free cross section

operation

Low radar

It is important at this point to inquire as to the consequences of these requirements on the aerodynamics. The progression from rather conservative toward more advanced inlets may cause serious problems with the internal flow. Typical problems are shown in figure XIV-6. Of particular importance are a strong shock boundary-layer interaction at the entrance to the inlet, which could cause flow separation, and another tendency for flow separation farther downstream in the subsonic diffuser that results from the curvature of the duct. It is necessary to achieve reattachment of the flow at the entrance so that shock stability can be assured and also so that pressure recovery and distortion of the flow at the engine face are established at acceptable values. Boundary-layer control techniques are discussed in the previous 1) and are listed at the bottom of figure XIV-6. paper (ref. These techniques have been useful in the past and will undoubtedly be even more so in the future. The applications of these techniques have historically been developed through experimental research for each unique inlet. Analytical methods to guide experimental research in applying vortex generation, bleeding, and blowing are now becoming sufficiently advanced so that they will play a greater role in guiding the experimental work than they have in the past. As an example of the use of analysis, consider the use of one of the viscous computer codes discussed by the panel on computational fluid mechanics (ref. 2). It is called the Annular Diffuser Deck (ref. 3) and was developed under contract by United Technologies Research Corporation. One of the 446

configurations it was applied to is shown in figure XIV-7. It is the F-16 inlet model which was tested in the Lewis 8- by 6-foot supersonic wind tunnel. In order to perform the calculation, SeVeral assumptions were required regarding the F-16 geometry. It was asThe method of analysis was developed by this author. sumed that the passage was an annulus having a centerbody shaped like the ramp surface. A number of radial struts which extended through the length of the inlet divided the flow into compartments. The flow through each of these compartments resembled the F-16 flow. The proper entrance-exit area ratio was matched as The well as the correct contour on the ramp and cowi surfaces. offsets between the entrance and the exit on the cowl and ramp Figure XIV-8 surfaces were also matched to those of the F-16. shows results for this duct obtained from the computer analysis by using the Annular Diffuser Deck. Shown at the top are velocity profiles for the flow between the throat and the compressor face. An accurate indication of flow separation is the corresponding calculation of local skin friction on the ramp and cowl surfaces. As shown at the bottom of the figure, the coefficient was positive at all locations and, hence, would not be expected to separate. No separation was observed in this portion of the flow in the model tests. Compared with the F-16 inlet, the HiMAT inlet is considered to be a more aggressive design in that it combines a large amount of offset with a short length: it is required to perform over a wide attitude and speed range. Figure XIV-9 shows the test model of supersonic wind tunthe HiMAT mounted in the Lewis 8- by 6-foot nel. The same computer program used for the F-16 was used for The corresponding the HiMAT duct, with the same assumptions. Velocity profiles analytical results are shown in figure XIV-lo. A tendency are at the top, and skin friction is at the bottom. for flow separation was predicted on the ramp side and, in fact, was observed in the model test. The analysis, therefore, predicted the location of the flow separation. Unfortunately parabolic marching procedures do not calculate beyond the separation Navierpoint, and, hence, reattachment cannot be determined. Stokes analyses or some type of empirical modeling through the separated region would be required in order to calculate the entire flow. The current marching solutions do, however, predict where separation will occur and, therefore, where boundary-layer The controls should be employed in order to prevent separation. distortion at the HiMAT compressor face resulting from the separation was not so severe that engine stall would necessarily result. The tendency for separation was easily suppressed with vortex generators, which were designed before the test in anticipation of the problem as predicted by the analysis. Without question, this and other analysis methods will play a significant role in future inlet development efforts.

447

NOZZLE PERFORMANCE AND COOLING Consider now the other end of the engine - the exhaust As in the case of inlets, there also exist geometrical ational requirements for nozzles. Some of the special ments of current interest for the nozzle are (1) (2) (3) High Stealth Thrust degree of integration with airframe system. and operrequire-

considerations deflection and reversal

An important item is a high degree of integration with the airStudies at the Langley Research Center have underscored frame. the growing interest in nonaxisymmetric nozzles for this purIllustrated in figure XIV-11 are some study concepts that pose. demonstrate the dramatic effect that can be achieved in aircraft On the left is a twin-engine axisymmetric nozzle concept design. that quite likely will encounter external flow separation. On the right are wedge or two-dimensional nozzles that would clean up the external lines very effectively. In addition, requirements (2) and (3) may be more easily met. In particular, thrust deflection and reversal may be implemented with less mechanical Deflection and reversal capabilities are of growing complexity. interest in order to increase flight maneuverability and decrease ground roll after landing. Although the wedge nozzle provides an important simplification of the external flow field of the airit may also lead to complexities of the internal flow craft, To illustrate, field of the jet. figure XIV-12 shows some LangAt the top of figure XIV-12 is shown the ley results (ref. 4). jet flow on the lower side of the wedge nozzle. The sketch indicates the emergence of the jet at an off-design pressure ratio where overexpansion shock waves are present in the flow. Beneath that are plan views of the wedge surface and shock patterns for nozzle pressure ratios of 3 and 6. The position and nature of these shocks were determined experimentally by the use of oil patterns on the wedge surfaces. At the lower pressure ratio, the flow is basically two-dimensional. At the higher pressure ratio, three-dimensional flow is indicated by the unexpectedly complex The planar shocks become curved, recirculation shock pattern. and vortex slip lines emanate downstream of the patterns appear, shock interaction points. The pressures measured along the centerline of the nozzle on the wedge surface are plotted at the botton in figure XIV-12. For the results of a two-dimensional, inviscid, timecomparison, dependent computer program by Cline (ref. 5) are also shown. At the lower pressure ratio, the agreement between theory and experAt the higher pressure ratio, the same comiment is excellent. parison shows that the analysis does not predict the pressures 448

Additional correctly. is under way at Langley program for this flow.

work on the to develop

analysis is required, a three-dimensional

and work computer

Another problem of the jet internal flow that is being studied at Lewis is film cooling of the nozzle walls to protect them from In a the hot flow resulting from afterburner operation (ref. 6). a simple axisymmetric plug nozpreliminary phase of this effort, a more complex two-dimenzle is being used; for a later phase, Figure XIV-13 shows the axisymsional nozzle is being procured. metric plug nozzle mounted on a J-85 engine in an altitude test cell. Figure XIV-14 shows the plug nozzle during afterburner operajust downstream of the cowl. Film coolant is injected tion. FigOverexpansion shock waves are clearly evident in this flow. ure XIV-15 shows the variation in local film cooling effectiveness as a function of a nondimensionaiized parameter which inThe coolant mass flow ratio, and slot height. cludes length, data correlate well with an empirical design curve except for These points corspecific points well below the design curve. respond to conditions where shock waves in the jet disrupt the Additional effort is required for corflow of the cooling film. A future phase of this cooling rerelation of these effects. search program will progress to nonaxisymmetric nozzles of the type illustrated in figure XIV-16. convergent-divergent exhaust nozzle may be This two-dimensional, This nozzle concept used in a variety of operating positions. was developed by General Electric under contract to Lewis. On one of the side plates is removed. the model shown in the figure, The geometry at the top of the nozzle is set for low pressure ratios. The top and bottom plates are adjustable for changes in the plates power and pressure ratios. At the left in the figure are rotated and the nozzle is positioned for thrust d,flection. On the right in the figure, rotation of a clamshell-like arrangement closes off the normal exhaust path. The two halves of the clamshell turn the flow. Top and bottom ports are opened when the clamshell is rotated. These ports allow the flow to exit in the forward direction. This nozzle will be used on a J-85 engine, and both cooling requirements and thrust characteristics will be evaluated. An alternative concept for thrust reversal is to use external flaps in conjunction with wedge nozzles (ref. 7). In this case, the reversal is accomplished externally rather than internally as in the two-dimensional clamshell geometry. Figure XIV-17 shows a model of the F-18 aircraft with a twin-wedge nozzle installation. Deflection plates are instalied on the top and bottom surfaces of both wedges. The plates are stored in the retracted position when no thrust deflection is required. Deployment of the plates provides increased maneuverability or thrust reversal. This 449

study at the Langley Research Center evaluated the effect of flap for example, geometry, the effect of sideplates on the deflection flaps. A sideplate is used on both sides of each of the flaps so that the cross section is U-shaped. Flaps with sideplates were more effective than flaps without sideplates for increasing the amount of thrust reversal. AFTERBURNER PERFORMANCE The first goal of the afterburner program conducted at Lewis 8 and 9) was to increase altitude limits for stable com(refs. bustion without rumble. Rumble is defined as a traveling pressure disturbance of 20 to 300 hertz. Operation of the afterburner in high-amplitude rumble causes fatigue failure of the afterburner with subsequent nozzle damage in a matter of minutes. The second goal was to increase from idle to maximum augmentation F-100 engine was used. blowout limits throttle. for transients For this work, an

Figure XIV-18 shows the research afterburner. The flameholder and injection rings can be seen. The novel feature of this afterburner was the introduction of swirl in the core flow. The method used is shown in figure XIV-19. The standard production augmentor is shown at the top of the figure, and the new swirl augmentor at the bottom. The latter is called a partial swirl augmentor, since only the core flow is swirled, not the fan flow. The 20 of core swirl around the centerbody was achieved by recambering the turbine exit guide vanes. Additional pressure loss was created by the presence of this swirl. However, swirling allowed the removal of the flameholder fingers and thereby The net result was no overall eliminated some pressure loss. performance penalty. A digital electronic engine control system was also added in order to control fuel flow rates and nozzle opening more precisely. Performance results are shown in figure XIV-20. The outer lines show the altitude and Mach number operating envelope of the engine without afterburning. The engine The limits of the operates at any point inside this envelope. productuion afterburner are shown by the inner lines, which indicate a loss in rumble free operational capability for very As indicated by the crosshigh altitudes at low Mach numbers. the swirl afterburner pushed the operating limit hatched region, it as much as 15 000 feet. to much higher altitudes - increased the transient blowout altitude Although not shown in this figure, limit was simultaneously increased by a minimum of 8000 feet. This research afterburner may be flight tested in an F-15 at the Dryden Flight Research Center in the near future.

450

CONCLUDING REMARKS A few areas of NASA work which are related to high-performance vehicles have been discussed. Unique inlet concepts will continue to evolve, but their design will be aided by computational methods to a greater extent than in the past. Nonaxisymmetric nozzles will provide cleaner external lines and enhanced maneuverability, but the internal flows will be more complex. And finally, novel concepts for swirl afterburners show promise for enanced performance in the high-altitude, low Mach number region. REFERENCES 1. Ciepluch, C. C.; et al.: Propulsion Technology. Conference (Cleveland, Bowditch, D. N.; et al: ternal Flow. Presented Conference (Cleveland, Anderson, Swirling Struts. Carson, lytical Static Vertical Takeoff Presented at the Ohio), May 15-16, and Landing Aeropropulsion 1979. (VTOL) 1979 of In-

2.

Computational Fluid Mechanics at the Aeropropulsion 1979 Ohio), May 15-16, 1979. Solution Axisymmetric for Turbulent Ducts with

3.

0. L.: Finite-Difference Compressible Flow in NASA CR-2365, 1974. G. T., Jr.: Investigation Conditions.

4.

and Mason, M. L.: Experimental and Anaof a Nonaxisymmetric Wedge Nozzle at NASA TP-1188, 1978. A Computer Program for the Computation Time-Dependent, Inviscid Nozzle Flow. Scientific Laboratory, 1977. of a Full Expansion

5.

Cline, M. C.: NAP: of Two-Dimensional, LA-5984, Los Alamos

6. Straight, D. M.: Effect of Shocks on Film Cooling Scale Turbojet Exhaust Nozzle Having an External NASA TM-79157, 1979. Surface. 7. Capone, F. J.; Characteristics F-18 Airplane.

Gowadia, N. S.; and Wooten, W. H.: Performance of Nonaxisymmetric Nozzles Installed on the AIAA Paper 79-0101, Jan. 1979. H.: Design and VerificaAIAA Paper 78-1040, of a 1979.

8.

Egan, W. J., Jr.; and Schadowen, J. tion of a Turbofan Swirl Augmentor. July 1978. Hanloser, Partial K. J.; and Cullom, Swirl Afterburner.

9.

R. R.: Test Verification AIAA Paper 79-1199, June

451

L., ,,% i,. .I:, (,. ;

. ,,

-_ : .

_. .,, ,

Figure XIV-l

F-16

Figure XIV-Z

452

F-18

cs-

79-2359

Figure XIV-3

HiMAT

Figure XIV-4

453

STEALTH

CONFIGURATIONS

Figure XIV-5

OFADVANCED INLETS AERODYNAMIC FEATURES

STRONG SHOCK BOUNDARY INTERACTION ; REAllACHMENT SEPARATION 7 8 I

SHOCK STABILITY -I BOUNDARY LAYER CONTROL TECHNIQUES VORTEX GENERATORS BLEEDING BLOWING

GS-79-2202

Figure XIV-6

454

F-16 INLET MODEL TESTING IN LEWIS 8X6 TUNNEL

Figure XIV-7

COMPUTER ANALYSIS OF F-16 INLET

1
RADIUS THROAT

LIP---L
COMPRESSOR FACE

SKIN FRICTION COEFF LENGTH Figure XIV-8

cs-794197

455

HiMAT INLET TEST IN 8X6 TUNNEL

Figure XIV-9

456

COMPUTER ANALYSIS OF HiMATINLET

RADIUS

L
SKIN FRICTION COEFF 1

EXPER!MENTAL SEPARATION COMPRESSOR FACE -RAMP ---COWL

THROAT

I
~~---?I>
Figure XIV-IO

Gs-7v-2203

TWIN AXISYMMETRIC AND WEDGE NOZZLE INSTALLATION

Figure XIV-11

457

WEDGE NOZZLE AERODYNAMICS

WEDGE

OIL SHOCK PAlTERN7

SHOCK

PATTERN 7

P/PT

LENGTH Figure XIV-12

FILM COOLED PLUG APPARATUS

J-85 ENGINE

Figure XIV-13

458

AXISYMMETRIC FILM COOLED PLUG NOZZLE

PR =3.1 Figure XIV-14

EXPERIMENTAL FILM COOLING RESULTS

AXISYMMETRIC GEOMETRY ,,- JET BOUNDARY

LOCAL FILM COOLING EFFECTIVENESS ,-DESIGN CURVE

, u LENGTH/MASS RATlO x SLOT HGT I Figure XIV-15

,o

I
cs-794201

459

20 CD VERSmATILE EXHAUST

NOZZLE

MODEL

NORMALOPERARON

THRUSTDEFLECTION FigureXIV-16

THRUSTREVERSAL

460

F-18

MODEL

WITH

TAB

REVERSERS

Figure XIV-17

AFTERBURNER VIEW OF F-100 ENGINE

Figure XIV-18

461

SWIRLAND PRODUCTION AUGMENTOR COMPARISON

r 7 INJECTION RINGS FLOWS: FAN CORE PRODUCTlON AUGMENTOR -____PARTlAL SWIRL AUGMENTOR

FLOWS: CORE FAN

\

L FLAMEHOLDER LRECAMBERED EXIT GUIDE VANES Figure XIV-19

CS-79-2199

EFFECT OF SWIRLON AFTERBURNER

r NO AmRBURNER

ALTlTUDE

A = 15

PRODUCTION AB

FLIGHT MACH NO.

Figure XIV-20

462

RESEARCH AND DEVELOPMENT CONTRACTORSAND GRANTEES NASA relies heavily research organizations nautical propulsion. currently supplementing this program area. In vertent omissions are Industry Aeromexico Aeronautical Research Associates of Princeton, Inc. AiResearch Manufacturing Co. Air New Zealand, Ltd. American Airlines, Inc. Atlas Consortium Avco Corp., Avco Lycoming Division The Boeing Co. Bolt, Beranek, & Newman, Inc. Chandler Evans, Inc. Colt Industries, Inc. Continental Air Lines, Inc. Cooper Airmotive, Inc. Detroit Diesel Allison Div., GMC Douglas Aircraft Co. Eastern Air Lines, Inc. Exxon Research and Engineering Corp. Finnair OY Flow Research Flow Simulators, Inc. Fluidyne Engineering Corp. Frontier Airlines Garrett AiResearch Aviation Co. General Applied Science Laboratories, Inc. General Dynamics Corp. General Electric Co. Grumman Aerospace Corp. Hamilton Standard Hersh Acoustical Engineering Honeywell, Inc. Educational and Research Institutions Purdue University Stanford University State University of New York at Stoneybrook Trenton State University University of California at Los Angeles University of Colorado University of Michigan University-of Missouri University of Notre Dame University of Texas Wichita State University Industrial Tetronics, Inc. International Nickel Co. KSSU Consortium Arthur D. Little, Inc. Lockheed Aircraft Corp. Lockheed Missiles & Space Co. McDonnell Douglas Corp. Mechanical Technology, Inc. National Airlines Northwest Airlines, Inc. Nielsen Engineering Pacific Airmotive Corp. Pan American World Airways, Inc. Pratt & Whitney Aircraft Group Rockwell International Corp. Scientific Research Associates Division of International Solar, Harvester Co. Solar Turbines International Southwest Research Institute Systems Control, Inc. (VT) Tech Development, Inc. Therm0 Mechanical Systems, Inc. Trans World Airlines, Inc. TRW, Inc. United Air Lines, Inc. United Technologies Corp. United Technologies Research Center Vought Systems Div., LTV Aerospace Corp. Western Air Lines Western Gear Corp. on the support and expertise of industry, universities, and in conducting its research and technology programs in aeroThe following is a list of contractors and grantees who are the in-house efforts of the NASA Lewis Research Center in view of the large number of organizations involved, inadpossible, for which apology is duly expressed.

Arizona State University Battelle Memorial Institute, Columbia Laboratory Case Western Reserve University Cornell University George Washington University Georgia Institute of Technology Illinois Institute of Technology Iowa State University Massachusetts Institute of Technology New York University at Westbury Pennsylvania State University Federal Agencies

Federal Aviation Administration U.S. Air Force Aeropropulsion Laboratory U.S. Army Research & Development Command

U.S. Army Research & Technology Laboratories (AVRADCOM) U.S. Naval Air Propulsion Center U.S. Naval Sea System Command (NAVSEA)

463

1. Report

No.

2.

Government

Accession

No.

3.

Recipients

Catalog

No.

NASA CP-2092
4. Title and Subtitle

5. Report Date October 1979

AEROPROPULSION
7. Author(s)

1979

6.

Performing

Organiwtion

Code

8. Performing

Organization

Report

No.

E-079
10. 9. Performing Organization Name and Address 11, Contract or Grant No. Work Unit No.

National Aeronautics and Space Administration Lewis Research Center Cleveland, Ohio 44135
12. Sponsoring Agency Name and Address

13.

Type

of

Report

and

Period

Covered

Conference Publication
14. Sponsoring Agency Code

National Aeronautics and Space Administration Washington, D. C. 20546

15. Supplementary Notes

16. Abstract

A two-day conference was held at the NASA Lewis Research Center on May 15 and 16, 1979, to provide leaders from government, industry, and universities with the latest results of NASA Fourteen papers were presented on topics including emisprograms in aeronautical propulsion. sion and noise reduction, alternative fuels, computational fluid mechanics, and hypersonic propulsion. The proceedings contains all the papers presented.

7. Key

Words

(Suggested

by Author(s))

Emission; Noise reduction; Aircraft fuels; Ceramic coatings; Materials science; Turbomachinery; Bearings; Gears; Microeledtronics; Optical measurement; Engine control; Supersonic turbines;
Hypersonic inlets; Supersonic combustion engines; Vertical takeoff aircraft
9. Security Classif. (of this report) 20.

Propulsion

efficiency;

18.

Distribution

Statement

Unclassified - unlimited STAR Category 07

ramjet

Security

Classif.

(of this

page)

21.

No.

of

Pages

Unclassified

Unclassified

467

22. Price A20

NASA-Langley. 1979

For sale by the National Technical Information Service, Springfield. Virginia 22161