Proceedings of ASME Turbo Expo 2015: Turbine Technical Conference and Exposition

GT2015
June 15 – 19, 2015, Montréal, Canada

GT2015-43251

AN INTEGRATED METHOD FOR PROPULSION SYSTEM CONCEPTUAL DESIGN

R.-G. Becker, S. Reitenbach, C. Klein, T. Otten, M. Nauroz

Institute of Propulsion Technology
German Aerospace Center (DLR)
Cologne, Germany

M. Siggel
Simulation and Software Technology
German Aerospace Center (DLR)
Cologne, Germany

ABSTRACT
The conceptual design of future, potentially highly NOMENCLATURE
integrated aircraft engines pose a variety of new design options A Area
to the propulsion system engineers. In order to find the best AR Aspect Ratio
conceptual design, rapid evaluation of many design choices is c Absolute Velocity, Chord Length
essential. However, traditional, fast evaluation methods CAD Computer Aided Design
employing historical and empirical data can only be applied to D Diameter
novel engine concepts to a very limited degree. Thus, swift DF Diffusion Factor
conceptual design methods based on physical approaches DH DeHaller Number
providing a sophisticated level of detail are needed. The current E Modulus of Elasticity
paper presents a methodology focused on conceptual engine h Enthalpy, Blade Height
design. The methodology is based on the gas turbine simulation HTR Hub to Tip Ratio
framework GTlab, which integrates software tools for engine IGV Inlet Guide Vane
performance, component aerodynamics and structural design. Ma Mach Number
For conceptual design a dedicated set of design tools exists – N Mechanical Spool Speed, Blade Number
the so called GTlab-Sketchpad. Sketchpad tools have full OGV Outlet Guide Vane
access to the thermodynamic design data of the engine p Blade Pitch
performance module. Based on cycle analysis, the tool set q Dynamic Head
generates parametric representations of the propulsion system r Radius, Diffuser Loss Coefficient
components and stores the results back to the frameworks data RQL Rich Burn / Quick Quench / Lean Burn
model. Computational time is limited to a few seconds, to S Stress
ensure interactivity during the design process. The graphical T Temperature
user interface provides means to interactively modify the design t Thickness
parameters and to immediately evaluate their impact on the u Displacement, Circumferential Velocity
overall design. Since the internal data model facilitates three z Number of Stages, Zweifel Coefficient
dimensional parameterizations of the engine components, 3D W Mass Flow
representations of the engine designs can be generated by w Relative Velocity
interfacing an open source CAD-kernel. For the present paper,
the conceptual design process of a commercial jet engine SYMBOLS
utilizing GTlab-Sketchpad is shown. The underlying α Angle, Heat Expansion Coefficient
computational methods are described and the resulting 3D- β Flow Angle
geometry is presented. γ Ratio of Specific Heats
η Efficiency

1 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Total Pressure Ratio analysis, noise and cost estimations to the multi-disciplinary
υ Poisson Ratio process chain. In 2009 Brophy [6] described the PMDO-lite
σ Solidity, Stress methodology which was generated at Pratt&Whittney Canada.
ω Angular Velocity PMDO-lite similarly relies on cycle, meanline and simplified
structural computations and emphasizes the importance of the
SUBSCRIPTS ability to quickly model a wide range of engine cycles and
avg Average architectures. However, since all three tool developments were
ax Axial conducted as propriety software for the engine companies,
D Dome details of the computation processes have not been disclosed.
diff Diffuser Regarding the research community, Tong [7] added an
in Inlet object oriented update to the WATE [3] methodology in 2008.
out Outlet WATE++ is closely connected to NASA’s performance code
pol Polytropic NPSS and is capable of drawing conceptual flow paths and
m Meridional overall engine layouts as well as calculating the propulsion
r Radial system masses from analytical approaches. Another ongoing
ref Reference research work is TERA [8] which combines TURBOMATCH
uts Ultimate Tensile Strength [9] for engine performance computations and WATE to perform
s Static the size and weight estimations.
Tangential The German Aerospace Center (DLR) has started an
t Total initiative to meet the challenge. The current paper presents a
methodology that is developed during the DLR internal engine
INTRODUCTION preliminary design project PEGASUS and is focused on
Increasing environmental concerns and soaring fuel prices conceptual engine design. The methodology is based on the gas
drive designers to reduce the fuel consumption and emissions turbine simulation framework GTlab, which integrates software
of future aircraft. To fully exploit the optimization potential of tools for engine performance, component aerodynamics and
the aircraft system, computational methods are needed that structural design. By means of the framework the tools
support the engineers to cope with the large amount of design exchange information over a common central data model and
choices and to better understand their effects on the overall are interchangeable in order to provide different levels of
aircraft system. At the beginning of the conceptual design phase fidelity during the engine preliminary design process. For
every new engine concept is represented by an engine cycle conceptual design a dedicated set of design tools exists – the so
model that predicts the corresponding thermodynamic called GTlab-Sketchpad. Sketchpad tools have full access to the
performance. To evaluate the effects on engine-aircraft thermodynamic design data of the engine performance module.
integration, further information on the engines dimensions, Based on cycle analysis, the tool set generates parametric
mass, center of gravity and moments of inertia is needed. This representations of the propulsion system components and stores
is met by subsequently adding tools for the prediction of the results back to the frameworks data model. Computational
component aerodynamics and structural mechanics. time of the Sketchpad tools on current personal computer
According to [1] the primary task of a conceptual design hardware is limited to a few seconds to ensure interactivity
code is to determine which design candidates of a large field of during the design process. The graphical user interface provides
competitors, should be further pursued. Thus, in order to find means to interactively modify the design parameters and to
the best conceptual design, rapid evaluation of many design immediately evaluate their impact on the overall design. Since
choices is essential. However, traditional, fast evaluation the common data model facilitates three dimensional
methods employing historical data as described in e.g. [2], [3] parameterizations of the engine components, 3D
can only be applied to novel engine concepts to a very limited representations of the sketched engine designs can be generated
degree. Hence, swift conceptual design methods based on by interfacing an open source CAD-kernel, which allows for
physical approaches providing a sophisticated level of detail are the precise evaluation of installation space, weight and the
needed. moments of inertia as well as a standardized geometry export to
The necessity of such computer aided conceptual design third party customers.
software has been identified before. In 1991 Stricker [4] In the following sections the details of the framework as
presented a methodology developed at GE Aero Engines to well as the parameterizations and modelling techniques are
predict the engine dimensions and masses on basis of shown. Finally the method is applied to the exemplary design
thermodynamic cycle computations, mean line predictions and of a mixed flow turbofan engine and the resulting 3D-geometry
simplified structural approaches. A further development was is presented.
published in 2002 by Jeschke et al [5] that describes the MTU
internal modular preliminary design code MOPEDS which is INTEGRATED DESIGN METHOD
also based on a modular performance code and adds The development of complex software systems requires a
sophisticated analysis tools for aerodynamics, structural well-planned data structure to allow quick and easy access,

2 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 maintenance and enhancement of components. Data modeling inertia tensor and thus the moments of inertia along arbitrary
is introduced to explicitly determine the data structures of those axes using the Huygens–Steiner theorem. The CAD abstraction
complex systems. During the DLR internal engine preliminary layer completely encapsulates the OpenCASCADE kernel such
design project PEGASUS a detailed data model was developed that no OpenCASCADE specific knowledge is required. Also,
to characterize the complex components of a gas turbine and this encapsulation theoretically allows a transparent
their relationships. The gas turbine simulation framework replacement of the underlying CAD engine.
GTlab uses this data model consistently across the whole In GTlab-Sketchpad, the most important parts of the engine are
system to achieve compatibility of data. All disciplines use the modeled as solid objects. The data model provides contours of
same data structures to store and access data and share disks, nacelles, shafts, and hub and tip walls as two dimensional
information seamlessly. Generated results are stored directly in polygon courses which are then transformed into 3D shapes
the data model and are available for other internal processes using the volume of revolution modeling algorithm. The blade
without any restrictions. This offers the possibility to easily geometries are computed by DLRs in house program
share data across different fidelity level within the design BladeGenerator [12]. Subsequently they are automatically
process. imported into the Sketchpad and finally transformed into solids.
The use of solid shape construction of the important parts
ENGINE PERFORMANCE enables the computation of the physical properties of single
Engine performance simulations are performed by GTlab- components as well as of the overall engine. The integration of
Performance [10] which is an object-oriented, component based the rendering engine allows a photorealistic visualization of the
performance synthesis program. Through its modular design complete aircraft engine which is a significant improvement
GTlab-Performance is capable of simulating arbitrary gas over pure 2D sketches.
turbine and turbo engine concepts. Both design and off-design
simulations as well as stationary and basic transient simulation BLADE PARAMETRIZATION
can be performed. GTlab-Performance allows for the All blade geometries within GTlab are represented by the
generation of customer decks which fit into the design and BladeGenerator parameterization, which was developed at DLR
simulation process of the airframe research group of the to support automated 3D-CFD blade optimization processes for
German Aerospace Center. fan, compressor and turbine geometries [12]. It features high
degrees of freedom in the selective design of curvature
CAD GEOMETRY REPRESENTATION distributions to influence shock-boundary-layer interactions as
The integration of a computer aided design (CAD) kernel well as global and precise local blade geometry modifications.
into GTlab-Sketchpad opens up many opportunities. Typical Parameter sets are chosen to maintain the use of typical blade
CAD systems offer numerous modeling algorithms that allow design parameters such as stagger angles, thickness
creating almost arbitrary shapes. These shapes are represented distributions, trailing and leading edge angles etc.
by mathematical models like NURBS curves and surfaces. In principle, the blade generation process can be divided
Some of the benefits of the complete three dimensional into two steps: Profile generation and profile stacking. Profiles
mathematical description of the engine parts are: three- are described in the two dimensional m’- coordinate system
dimensional visualization and accurate estimation of physical introduced by Drela [13] and consist of four B-splines
properties like mass, center of gravity, and moments of inertia. representing leading edge, trailing edge, suction side and
Also, by using standardized file exchange formats, the pressure side which are finally merged into a single B-spline.
modelled parts can be exported for direct use in other CAD All profiles are then radially assembled by means of a stacking
systems or as inputs for mesh generation algorithms, which are algorithm, such as e.g. centroid, leading edge or trailing edge
the basis for computational fluid dynamics simulations or FEM stacking.
structural analysis codes. On the other hand, the import of By utilizing a single central blade parameterization
geometry files into GTlab-Sketchpad allows the integration of methodology, GTlab is capable of seamlessly blending between
external geometrical components – that might be designed in the geometry outputs of tools of different fidelity and thus
another CAD program – into the tool chain. allowing for zooming applications such as given in [14].
The CAD kernel integration is realized by an abstraction layer
around the open source framework OpenCASCADE [11]. This FAN
powerful C++ based framework comes with many sub- Within GTlab-Sketchpad the fan section comprising the
packages, such as geometric modeling, topological abstraction, spinner, the fan rotor, the core section IGV and the bypass
an OpenGL based 3D rendering engine and shape analysis section OGV is pooled in one module (Fig. 1). Since the flow
algorithms. Some of the modeling algorithms used in GTlab- path is separated in core and bypass stream also the splitter
Sketchpad are volumes of revolution, extrusions/prisms, positioning is considered.
lofting, boolean operations, and B-spline interpolation of point The annulus geometry of the fan section is determined by
cloud data. The shape analysis layer offers functions to means of a meanline computation method as it is presented by
calculate physical properties of solid objects or groups of Bräunling [15]. For the calculation process, the thermodynamic
objects. These are the object mass, its center of gravity, the inlet conditions such as pressure, temperature and mass flow

3 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 are required input parameters. Besides thermodynamic The blade aspect ratio as well as the relative distance between
boundary conditions, additional aerodynamic and geometric rotor and Splitter/IGV/OGV in rotor chord lengths is given as
values must be assumed to translate fan performance data into an input. An exemplary CAD Output of the fan section is
geometric design. shown in figure 2.

Figure 1: FAN SECTION GEOMETRY DESCRIPTION

Assuming the axial Mach number and hub-to-tip ratio in

the rotor inlet section, size and radial positioning of the pre Figure 2: 3D-CAD REPRESENTATION OF THE SKECTHPAD
rotor area are distinctly defined. After the leading edge, the FAN MODULE
flow path is split up into bypass and core section. This enables
the direct use of the thermodynamic data for each section. COMPRESSOR
The meanline speed triangles behind the rotor are derived Similar to the Sketchpad fan module, the compressor
from Euler’s equation (Eq. 1) applying the rotor DeHaller module is mainly based on meanline calculation routines as
Number as an additional input parameter (Eq. 2). they are exemplarily presented by Bräunling [15] or Mattingly
et al. [16]. Using input values derived from the engine
∆ , (1) performance simulation, the meanline computation provides a
first estimation of annulus geometry as well as thermo- and
(2)
aerodynamic parameters in all compressor stages at slight
(3) computational effort. Regarding the compressor annulus, the
user may choose between constant tip, constant hub or constant
mean radius calculation. By means of aerodynamic loss
The Mach number as well as static thermodynamic correlations presented by Johnsen and Bullock [17] combined
parameters behind the rotor are determined by means of the with a free vortex approach, airfoil angles at hub, mean and tip
stator De Haller Numbers for IGV and OGV (Eq. 3) combined levels are also estimated.
with the assumption of an axial flow behind the stators. The meanline computation requires the following set of
The splitter radius depends on the user defined bypass- input parameters provided by the engine performance
ratio. By setting the wall contour angle Byp,Hub and CoreTip and simulation:
choosing constant tip, -mean or -hub radii for all blade rows, all
flow areas and their related radial positions are defined. Compressor total pressure ratio
In the following, speed triangles for other radii, such as Inlet massflow
hub and tip are derived, solving the radial equilibrium by means Mechanical spool speed
of the free vortex approach (Eq. 4, 5). , Total inlet temperature
, Total inlet pressure
(4) Polytropic Efficiency

. (5) Since the amount of information offered by these parameters is

not sufficient to proceed the meanline algorithm, the following
parameters have to be chosen additionally by the user:
The airfoil camber angle at each considered radius is
simplistically determined as the mean angle between relative Number of stages
inlet and outlet flow. Inlet Mach Number
Currently, no detailed profile geometry is calculated, so Absolut inlet flow angle
standard values are used for the 3D description of the blade.

4 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 Hub to tip ratio at compressor inlet By means of the free vortex approach (Eq. 4 and 5) and the
De Haller Number velocities on the mean radius, velocity triangles at hub and tip
Diffusion Factor radius can be determined for all flow stations within the
Aspect Ratio considered compressor stage. At this point the desired
aerodynamic data is complete and will be used in the following
De Haller number, Diffusion factor and aspect ratio may be for an evaluation regarding airfoil size and shape.
chosen separately for rotor and stator as well as for inlet and The calculated velocity triangles for rotor and stator are
outlet. Based on the input, a linear distribution of these used as input for correlations regarding reference incidence and
parameters is applied to all stages. The total enthalpy rise per deviation presented by Johnsen and Bullock [17]. This leads to
stage is assumed as constant for all stages and is given by: estimations of ideal airfoil camber and stagger angles for the
three considered radius levels. The chord length for each
1 (6) blade row is calculated by means of the user defined aspect
∆ , , 1 ratio and the blade height which is known from the annulus
1
geometry (Eq. 10).
Since total inlet conditions, inlet Mach number and mass (10)
flow are known, the inlet area is distinctly defined and it´s
radial positioning follows directly from . Static ∆
thermodynamic parameters at the inlet are calculated by means 1 ,
of simple gas-dynamical relations as they are presented by 2
(11)
Bräunling [15]. ∆
1 ,
Since calculations are performed on the mean radius, the 2
chosen annulus mode has influence on the further calculation (12)
procedure. In case of a constant mean radius, the
circumferential velocity at the post rotor station is already 2
known. By means of Euler´s equation (Eq. 1) and the rotor De (13)
Haller Number (Eq. 2), the velocity triangle behind the rotor is
defined. The total temperature behind the rotor follows from the Applying the definition of the Diffusion factor given in
stage enthalpy rise and since the absolute velocity at this station equation (Eq. 11) combined with the calculated mean radii for
is also known, the post rotor Mach number can be iterated from rotor and stator, the appropriate blade numbers for all blade
equations (Eq. 7 and 8). rows can be determined (Eq. 13).
The calculated airfoil parameters are used in the following
(7) to generate 3D airfoil geometries. Figure 3 shows an exemplary
3D compressor geometry generated with the presented
proceedings including airfoils and appropriate blade disks.
1 (8)
1 2

By means of Mach number and total temperature all static

thermodynamic parameters at the post rotor station as well as
the required flow area including hub and tip radius can be
determined. The calculation of the stator outlet area and related
thermodynamic and aerodynamic parameters is performed
analogously using equations (Eq. 9) and (Eq. 3). The
thermodynamic conditions at the stage outlet are set as input
parameters for the following stage and the presented calculation
process is repeated according to the total number of stages.

(9)

For a constant hub or constant tip annulus, the calculation

procedure described above ends up in an iterative process for
each stage since the mean radius behind the rotor is initially FIGURE 3: 3D-CAD REPRESENTATION OF THE
unknown. SKETCHPAD COMPRESSOR WITH INITIAL DISK
SETUP.

5 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 COMBUSTOR where designates the fuel flow and is the combustor
Within GTlab-Sketchpad combustor design is based on mass flow. For a typical combustor, dome cooling flow is
empirical correlations from design data of existing combustor estimated to be 10% - 15% of the overall combustor mass flow.
systems. Currently, a combustor is parameterized by a common Using the nominal dome reference velocity of 12 ⁄ the
geometry definition, which resembles the layout of a standard dome area follows from mass continuity.
annular RQL-combustor. A cross-section of the standard
annular layout with the main characteristics and components is The pressure loss method presented by Lefebvre [19]
shown in figure 4. specifies typical values for the overall pressure loss ∆ relative
to the inlet pressure and a reference dynamic head :

Δ
0.06 (18)
Δ
20 (19)
Thus, the reference velocity can be estimated as a function
of combustor inlet temperature with the empirical equation:

0.006 ∙ ∙ (20)

Figure 4: COMBUSTOR CROSS SECTION DESCRIPTION The dome area is finally obtained by

∙ (21)
In the present approach the geometry of the inlet and outlet with
gas path of the combustor is either matched to the geometries of
the previously sketched compressor and turbine components or
can be determined by defined inlet and outlet conditions from 1 (22)
engine cycle analysis. For the latter case, typical axial Mach Δ
∙
numbers and hub to tip ratios must be additionally defined by
the user.
Presuming the diffusor inlet area, the diffusor exit where the diffuser loss coefficient and are defined as
geometry is defined by a given inlet to exit area ratio and a follows.
diffusion angle. For the following determination of the
combustor dome area , there are currently three different Δ
(23)
empirical approaches implemented: Δ

The velocity method introduced by Mellor [18] estimates A

the dome area from mass continuity by means of empirically (24)
A
scaled reference velocities and mass flow distributions. The
dome airflow is divided into: The third methodology is based on an approach presented
by Lefebvre and Halls [20]. It estimates the combustor
(14) reference area based on a correlation between the combustor
efficiency and the correlation parameter (Eq. 25).
The airflows are:
.
 : Dome mass flow ∙ ∙ ∙ (25)
 : Fuel atomizing airflow
 : Swirler airflow
 : Dome cooling flow According to experimental data, empirical parameters and
are set to 0.75 and 300. The equation allows drawing
With the empirically defined airflows conclusions from combustor design and performance data to
the combustor efficiency. New combustor design is generally
15 ∙ (15) based on previous experience. Figure 5 shows efficiency data
7.5 ∙ (16) of a large number of multi-can, can-annular and annular
∙ (17) combustion chambers against the θ parameter that can be used
for new combustor design to meet stipulated performance

6 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 requirements. For example during wind milling condition, each blade row station and the mean line blading angles. The
where a minimum combustor efficiency of about 80% is being three-dimensional blade shapes are calculated by means of
achieved. Therefore the mean value for can be obtained from predefined thickness distributions for rotor and stator profiles
graph which defines an appropriate geometry. and blade angle distributions given by the free vortex approach
together with the assumption of constant meridional velocity
as given by equations (Eq. 4 and 5). The number of blades is
derived from the Zweifel coefficient, which has to be defined as
an input variable for each blade row. Using the equation given
in [22], the blade number can be calculated by:

4 ∙ ∙ cos (26)
∙

The required cooling air flows are estimated by means of

the methodology proposed by Horlock et. al [23] which
provides an semi-analytical heat balance approach for both
external film cooling and internal convective blade cooling.
Figure 5: CORRELATION OF COMBUSTOR EFFICIENCY The thermodynamic state and gas properties of the blade row
DATA FOR CONVENTIONAL COMBUSTORS [20] main flow and the cooling air flow are provided by the engine
cycle model. The Stanton number, cooling efficiency as well as
Finally the combustor length and the geometry of the the ratio of the exposed area for heat transfer to the cross-
pressure casing are determined by input parameters dependent sectional area of the main hot gas flow must be given as input
on the previously calculated dome area, and the passage area, to the calculation process. By assuming an allowable blade
which can be determined by using a passage reference velocity material temperature the cooling air fractions can be estimated.
of 50 ⁄ and a passage mass flow of . Currently no parameterization of the internal cooling
Figure 6 shows an exemplary 3D-output of the combustion passages of the turbine blades is implemented. To account for
chamber sketching module including liners, pressure chamber the blade mass reduction through cooling cavities a cavity
and fuel injectors. correction factor on the blade volume is applied during the
mass calculation algorithm.
Figure 7 shows a three-dimensional output of a Sketchpad-
modelled two stage high pressure turbine.

Figure 6: 3D-CAD REPRESENTATION OF THE SKETCHPAD

COMBUSTOR

TURBINE
Within GTlab-Sketchpad the turbine annulus geometry is
calculated by a modification of the simplified mean-line
approach described by Glassman [21]. Input design parameters Figure 7: 3D-CAD MODELL OF A SKETCHED TWO STAGE
HIGH PRESSURE TURBINE WITH APPROPRIATE DISKS.
are the required shaft power, the mass flow rate and the inlet
temperatures and pressures. Additionally, the rotational speed
as well as the inlet and exit Mach numbers, the stator angle and
the number of stages have to be defined by the user. The
BLADE ATTACHMENTS
method is restricted to three types of velocity diagrams:
According to the methods described in [24] GTlab-
symmetrical, impulse or zero exit swirl.
Sketchpad implements three types of blade attachment
Applying the input data, the GTlab-Sketchpad module
geometrical representations: axial and circumferential single-
computes the annulus dimensions, the thermodynamic state at
tooth dovetails primarily used as compressor blade attachments

7 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 and three- or four-teeth fir tree attachments for turbine blade
rows. Figures 8 and 9 show the parameterization provided by where
Sawyer [24], which was adopted by the sketching module.
1 Δ (28)
1

1 Δ . (29)
1
.

Substituting the above equations leads to an ordinary

differential equation for the displacement field which can be
solved by the finite difference approach. Re-substitution of the
displacements leads to the resulting radial and tangential disk
stresses. The material properties such as the modulus of
elasticity E, the Poisson ratio  and the thermal expansion ratio
Figure 8: FIR TREE PARAMETERIZATION AS DEFINED BY α are provided by the internal material database described later.
[24]
Temperature distributions can be defined as constant, linear, by
Combining the blades, the blade platform and the disk tangs, Fourier’s law or by an arbitrary array of temperature values.
the attachment components sum up as centrifugal loads that are The sketching module provides three disk types to approximate
equally distributed on the disk rim. gas turbine disks, as shown in figure 10.

Hyperbolic
Radius

Web

Ring
Figure 9: DOVE TAIL PARAMETERIZATION GIVEN IN [24]

dr
For bladed disks (BLISK), no extra attachment component is t

added to the blade and disk assembly and the disk is directly
attached to the blade platform. Thus, the disk rim stress is
calculated by the centrifugal forces of the blades only.
Currently no automated optimization procedure for the
adaption of the blade attachment size exists within GTlab-
Sketchpad. Engine Axis
Figure 10: AVAILABLE DISK TYPES OF THE GTLAB DISK
DISKS SKETCHER MODULE.
The GTlab disk sketching methodology is based on the
approaches of Armand [25] and Tong [26]. Therein the most
important design loads for rotating disks are considered to be
Each disk type has its own geometrical parameterization and
body forces, blade loads and thermal loads due to radial
facilitates interface functions to modify the blade shapes in the
temperature gradients. Other loads such as shaft torque or
according parameter space. In order to find the optimal blade
bearing loads are neglected. The methodology utilizes a set of
shapes for a given rim stress and temperature distribution, the
simplifications and assumptions: First of, only symmetric disks
shape parameters of the disks are varied by an optimization
are considered. The thickness of such disks needs to be small
algorithm. The objective function of the optimization process is
compared to their radius in order to assume a plane stress
the blade mass. Feasible solutions must comply with given
condition and only homogenous isotropic materials are applied.
geometrical restrictions such as minimum bore radii as well as
For stress calculations the disks are discretized by a set of thin
the design criteria proposed by Tong [26]:
rings of linearly changing thicknesses. Following the notation
of [26] the equation of equilibrium for the disk is then given by
1.0 0 (30)
0 (27)

8 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 .
-1.0 > 0 (31) NACELLE
, For the sketching process, the nacelle´s contour is divided
into three sections (figure 12). The first section is represented
Currently the GTlab-Sketchpad disk module utilizes the by an elliptic geometry and mainly responsible for the shape of
COBYLA [27] algorithm for fast local optimization and the the nacelle´s leading edge. It is defined by the throat point of
ISRES [28] algorithm for a more time consuming global the engine inlet and the foremost point of the nacelle whose
search. Both algorithms are provided by the NLopt optimizer positioning is dependent on the chosen input parameters. The
library [29]. section between the foremost and the highest point of the outer
casing is approximated by a NACA airfoil geometry. The final
SHAFTS segment between the highest point and the endpoint is given by
The shafts of an aero engine transfer the power from a B-spline.
turbines to compressors. In GTlab-Sketchpad, a simple
preliminary design method is applied. The required shaft
thickness is calculated based on torsion stress, axial stress and
centrifugal stress. Other potentially limiting design criteria like
critical speeds or longitudinal stiffness are not taken into
account.
The axial stress, resulting from axial forces induced by the
compressor and turbine is calculated from equation (Eq. 32).
(32) FIGURE 12: SEGMENTATION OF THE OUTER CASING FOR
∗
THE SKETCHING PROCESS
The tangential stress caused by centrifugal force can be
For the geometrical definition of the ellipsis, the following
expressed as follows [cf 30]:
three parameters are provided:
, 0.825 ∙ 1 0.212 ∙ (33)
horizontal radius
vertical radius
The torque of the shaft creates a shear stress. Its maximum rotation angle
value can be expressed as [31]:
The throat point of the engine inlet defined by and is
2∗ ∗
(34) given as a boundary condition and may not be changed. The
∗ lowest point of the ellipsis is automatically moved to the throat
point. A change of the rotation angle will hence not lead to a
All stresses are summarized using the von Mises criterion rotation of the ellipsis around its center but to a tangential
and need to match the materials allowable stress and a safety movement through the throat point.
factor. The NACA geometry is designed by means of a
symmetrical airfoil from the NACA-4-digit-series. For this
, , 3 (35)
series the maximum airfoil thickness is reached at 30% of the
chord length. A distinct definition of the NACA geometry
The shaft geometric layout is currently limited to a simple requires the positioning of the nacelle´s highest point as an
setup, A conical section of the shaft is attached to a cylindrical additional input defined by and . By means of these
section, followed by another conical section as shown in figure parameters, airfoil characteristics such as thickness and
11. required chord length can be calculated (Eq. 36 and 37). A
detailed description regarding the aerodynamic and geometric
properties of the considered airfoil family is given by Abbott et
al. [32].

FIGURE 11: GEOMETRIC PARAMETERS OF THE SHAFT 2 ∆ (36)

SKETCHING MODULE.
1
∆ (37)
Different materials can be selected for the calculation. 0.3

The endpoint of the outer casing is given as an additional

boundary and hence the final segment can be easily
approximated by a common B-spline to secure a smooth

9 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 junction from the NACA geometry to the nacelle´s trailing Since no real component off-design characteristics are available
edge. in this design phase, the off-design behavior of the engine
components is represented by standard performance maps
MATERIAL DATABASE which are scaled to the desired design values.
The GTlab-Suite ships with an extensible database for
commonly applied aero-space materials. Table 1 shows the list
of physical material properties that can be accessed via the
database interface.

Table 1: PHYSICAL MATERIAL PROPERTIES ACCESSIBLE

BY THE GTLAB MATERIAL DATABASE INTERFACE.

Property Unit Figure 13: PERFORMANCE SYNTHESIS SCHEMATIC OF

Density [kg/m^3] THE INVESTIGATED MIXED FLOW TURBOFAN ENGINE
Elasticity module [Pa] WITH DETAILS OF THE MODELLED SECONDARY AIR
Shear module [Pa] SYSTEM.
Temperature

Tensile strength [Pa]

dependent

By means of the synthesis model, all operating points

Yield strength [Pa]
relevant as input for the sketching modules are computed and
Poisson ratio [-] stored in the central data model. For the present investigation
Thermal expansion coefficient [K^-1] the points end of field (EOF), top of climb (TOC) and mid
Thermal conductivity [W/(Km^2)] cruise (MCR) have been simulated to provide the data for the
Heat capacity [J/K] fan, compressor, turbine and structural elements sketching
computations. In addition to that, the operating points idle
Material properties may be exported and imported through (IDL) and maximum take-off (MTO) have been evaluated for
the MatML markup language [33]. Table 2 provides an excerpt the combustor computations.
list of publicly available material data mostly collected from After finishing the synthesis model setup and completing
[34]. the engine cycle computations, the GTlab graphical user
interface allows for a quick changeover from the performance
Table 2: SELECTED PUBLICLY AVAILABLE MATERIALS to the preliminary design perspective. Sketchpad components
PROVIDED BY THE GTLAB MATERIAL DATABASE.
can be added to the perspective by appending the corresponding
Sketchpad calculator modules to the process management
INCONEL 600, 617, 625 Waspaloy
system. Thus, the overall sketching task is represented by a
INCONEL 706, 718, X-750 UDIMET 700, 720
process chain of calculator modules. The resulting modular
NIMONIC 75, 80A, 86, 90 UDIMAR 250, 300 setup allows the sketching of arbitrary engine concepts,
NIMONIC 105, 263, 901 Ti-6Al4V, Ti-17 providing that all engine components find a Sketchpad
INCOLOY 903, 909 TiAl representation. Once a calculator module has been added, it has
read-access to the full data model and write-access to the
DESIGN PROCESS engine component it was assigned to. Thereby, the calculator
To demonstrate the capabilities of the presented integrated model can fetch its thermodynamic design parameters from the
conceptual design method, an exemplary design process of a previously computed operating points. The geometric output
mixed flow turbofan has been conducted. As mentioned earlier, parameters are instantaneously displayed in the 2-D editor
every aero engine concept is initially modelled as a window of the preliminary design perspective, once the process
thermodynamic engine cycle that fulfils all requirements of the is finished. Changing the calculator input variables - either by
transport mission defined by the airframe manufacturer. In this editing the numerical input parameters or by using mouse
example the transport task is chosen to be a 150 passenger short dragging capabilities of the 2-D editor - and restarting the
to medium range commercial aircraft similar to the A320 or process gives immediate visual feedback to the engineer. In
737 airliners. To be able to compare the results of the addition to that, the design engineer may switch over to the 3D-
Sketchpad output with data of a real engine, a cycle model was CAD view to gain further detailed analysis capabilities. Figure
set up, that resembles an existing engine in service today. By 14 shows a screenshot of the two and three dimensional output
means of GTlab-Performance, thermodynamic component of the compressor Sketchpad module during an initial phase of
models for the intake, fan, compressors, combustor, turbines, the high pressure compressor sketching process.
ducts and exhaust systems have been assembled graphically as
shown in figure 13. The thermodynamic design parameters of
the component modules have been adapted to publicly available
engine certification data such as [35] as well as internal engine
measurement data taken from DLRs research aircraft ATRA.

10 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 effects of the variation of certain design parameters on the
overall engine system.
The employment of a common central data model which
incorporates high fidelity parameterizations of the involved gas
turbine components introduces seamless shifting between
conceptual, preliminary and detailed design tools. This feature
predestines GTlab to be used in future component zooming
applications.
In addition to the sketching modules described above, a
variety of additional sketchers are currently under development,
such as modules for planetary differential gear boxes, single
and counter rotating propellers, casings, lean combustor
injectors and reversed flow annular combustors. The existing
Figure 14: GTLAB-PREDESIGN PERSPECTIVE WITH
INITIAL GEOMETRY GUESS OF SKECTPAD COMPRESSOR Sketchpad modules for compressor and turbine design will be
MODULE. updated to improve the intuitive, interactive modification of the
annulus design through B-spline parameterizations of the hub,
After adequate input parameters for each sketched component mean and tip lines. Furthermore, the blade attachment design
have been found, the sketching process is finished. Subsequent routines will be automated.
changes to the overall cycle parameters will show up in GTlab-Sketchpad and parts of the GTlab-Performance
modified geometries after the Sketchpad process chain is re- library are planned to be distributed as a free to use educational
run. The CAD geometries of the new engine design may be software package.
exported to third party programs via standard export interfaces.
GTlab-PreDesign is able to compute all component masses as ACKNOWLEDGMENTS
well as their center of gravity and moments of inertia from the The authors like to thank all colleagues involved with the
generated solid geometry representation. Figure 15 shows a PEGASUS preliminary design project for their technical
final assembly of the investigated mixed flow turbofan. The support and fruitful discussions.
annuli of the low and high pressure compressors and turbines
have been manually refined to generate smoother flow paths. REFERENCES
[1] Halliwell, I., 1998, “Preliminary Engine Design - A
Practical Overview”, AIAA 98-3891, Cleveland, Ohio,
USA
[2] Sagerser, D. A., Lieblein, S., and Krebs, R. P., 1971,
“Empirical Expressions for Estimating Length and Weight
of Axial-flow Components of VTOL Powerplants”, NASA-
TM-X-2406, E-6191
[3] Onat, E., and Klees, G. W., 1979, “A Method to Estimate
Weight and Dimensions of Large and Small Gas Turbine
Engines”, NASA-CR-159481
[4] Stricker, J. M., and Norden, C. M., 1991, “Computerized
Preliminary Design of Turbomachinery”, 91-GT-391, Gas
Turbine and Aero Engine Congress and Exposition,
Orlando, Florida
Figure 15: MANUALLY REFINED 3D-CAD MODEL OF THE [5] Jeschke, P., Kurzke, J., Schaber, R., and Riegler, C., 2002,
SKETCHED MIXED FLOW TURBOFAN ENGINE. BEARINGS “Preliminary Gas Turbine Design Using the
AND CASINGS NOT INCLUDED. Multidisciplinary Design Tool MOPEDS”, ASME Paper
No. GT-2002-30496
CONCLUSION AND OUTLOOK [6] Brophy, F., Mah, S., and Turcotte, J., 2009, “Preliminary
An integrated approach to computer aided conceptual Multi-Diciplinary Optimization (PMDO) an Example at
design of propulsion systems has been presented. The software Engine Level”, NATO RTO-EN-AVT-167
offers a technical combination of engine performance, [7] Tong, M. T., and Naylor, B. A., 2008, “An Object-
component sketching and preliminary design and opens up new Oriented Computer Code for Aircraft Engine Weight
options to rapid engine design. The presented sketching Estimation”, ASME GT-2008-50062
modules provide means to swiftly conduct sophisticated [8] Ogaji, S., Pilidis, P., and Hales, R., 2007, “TERA- A Tool
conceptual design studies on basis of physical approaches and for Aero-engine Modelling and Management”, 2nd World
give the designer direct quantitative and visual feedback on the Congress on Engineering Asset Management and 4th

11 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use

 International Conference on Condition Monitoring, [26] Tong, M., T., Halliwell, I., and Ghosn, L., J., 2002, “A
Harrogate, UK computer code for gas turbine engine weight an disk life
[9] Pilidis, P., 2004, “Turbomatch Scheme for Aero/ Industrial estimation”, GT-2002-30500, ASME Turbo Expo: Land,
Gas Turbine Engine Design Point/Off Design Performance Sea & Air, Amsterdam, The Netherlands
Calculation”, Cranfield University [27] Powell, M. J. D., 1998, "Direct search algorithms for
[10] Becker, R.-G., Wolters, F., Nauroz, M., and Otten, T., 2001, optimization calculations", Acta Numerica 7, 287-336
“Development of a Gas Turbine Performance Code and its [28] Runarsson T. R., and Yao X., 2005, “Search biases in
Application to Preliminary Engine Design”, DGLRK 2011, constrained evolutionary optimization”, IEEE Trans. on
27.-29. Sept, Bremen, Germany Systems, Man, and Cybernetics Part C: Applications and
[11] Open Cascade Technologies, Open Cascade S.A.S. 2014, Reviews, vol. 35 (no. 2), pp. 233-243
http://www.opencascade.org, accessed October 15th 2014 [29] Johnson, S. G., 2014, “The NLopt nonlinear-optimization
[12] Voß, C., and Nicke E., 2008, “Automatische Optimierung package”, http://ab-initio.mit.edu/nlopt, accessed October
von Verdichterstufen“, Fachlicher Abschlussbericht 19th 2014
Forschungsvorhaben (FKZ: 0327713 B), AG Turbo [30] Beiz, W., and Grote, K.H., 2001, “Dubbel – Taschenbuch
COOREFF-T für den Maschinenbau“, 20th edition, ISBN 3-540-67777-1
[13] Drela, M., and Youngren H., 1998, “A User’s guide to [31] Schlecht, B., 2010, „Maschinenelemente. 1. Festigkeit,
Mises”, MIT Computational Aerospace Sciences Wellen, Verbindungen, Federn, Kupplungen“, Pearson
Laboratory Studium
[14] Reitenbach, S., Schnös, M., Becker, R.-G., and Otten, T., [32] Abbott, I.H. et al., 1945, “Summary of Airfoil Data”,
2015, „Optimization of Compressor Variable Geometry National Advisory Committee for Aeronautics, Report No.
Settings Using Multi-Fidelity Simulation”, ASME 824
GT2015-42832 (in print) [33] ASM International, http://www.matml.org/, accessed
[15] Bräunling, W., 2009, „Flugzeugtriebwerke: Grundlagen, October 20th 2014
Aero-Thermodynamik, ideale und reale Kreisprozesse, [34] Special Metals Corporation, Huntington, West Virginia,
Thermische Turbomaschinen, Komponenten, Emissionen USA, www.specialmetals.com, accessed October 20th 2014
und Systeme“, Springer Verlag, Berlin, Germany [35] ICAO, Aircraft Engine Emissions Databank, 2014,
[16] Mattingly J. D., Heiser W. H., and Pratt, D. T., 2002, http://easa.europa.eu/document-library/icao-aircraft-
“Aircraft Engine Design”, Second Edition, American engine-emissions-databank, accessed September 15th 2014
Institute of Aeronautics and Astronautics
[17] Johnsen, I. A., and Bullock, R. O., 1965, “Aerodynamic
Design of Axial-Flow Compressors”, NASA SP36,
National Aeronautics and Space Administration
[18] Mellor, A.M., 1990, “Design of Modern Turbine
Combustors”, Academic Press
[19] Lefebvre, A. H., 2010, “Gas Turbine Combustion –
Alternative Fuels and Emissions”, 3rd edition, CRC Press
[20] Lefebvre, A. H., and Herbert, M.V., 1960, “Heat-Transfer
Processes in Gas Turbine Combustion Chambers”,
Proceedings of the Institution of Mechanical Engineers,
Vol. 174, No. 12, pp. 463-73, 7960
[21] Glassman, A. J., 1972, “Computer program for preliminary
design analysis of axial-flow turbines”, NASA TN D-6702
[22] Bretschneider, S., 2011, “Knowledge-Based Preliminary
Design of Aero-Engine Gas-Generators”, PhD-Thesis,
Institut für Luftfahrtantriebe der Universität Stuttgart
[23] Horlock, J. H., Watson, D. T., and Jones , T. V., 2001,
“Limitations on Gas Turbine Performance Imposed by
Large Turbine Cooling Flows”, Journal of Engineering for
Gas Turbines and Power, Vol. 123
[24] Saywer, J.W., 1985 , “Sawyer’s gas turbine engineering
handbook, Volume I: Theory and Design”, Turbomachinery
International Publications
[25] Armand, S. C., 1995, “Structural Optimization
Methodology for Rotating Disks of Aircraft Engines”,
NASA Technical Memorandum 4693, Lewis Research
Center, Cleveland, Ohio

12 Copyright © 2015 by ASME

Downloaded From: http://proceedings.asmedigitalcollection.asme.org/ on 01/22/2016 Terms of Use: http://www.asme.org/about-asme/terms-of-use