Haughey Thesis 2018

Development of a Static Aeroelastic Database Using NASTRAN SOL 144 for
Aircraft Flight Loads Analysis from Conceptual Design Through Flight Test
by
Ryan M. Haughey
A thesis submitted to the College of Engineering at

Florida Institute of Technology
in partial fulfillment of the requirements
for the degree of
Master of Science
in
Aerospace Engineering
Melbourne, Florida
April 2018
We the undersigned committee hereby approve the attached thesis,
“Development of a Static Aeroelastic Database Using NASTRAN SOL 144 for
Aircraft Flight Loads Analysis from Conceptual Design Through Flight Test ” by
Ryan M. Haughey.
_________________________________________________
Dr. Razvan Rusovici
Associate Professor of Aerospace and Biomedical Engineering
College of Engineering and Computing
_________________________________________________
Dr. David Fleming
Associate Professor of Aerospace and Mechanical Engineering
_________________________________________________
Dr. Paul Cosentino
Professor of Civil Engineering and Construction Management
_________________________________________________
Dr. Hamid Hefazi
Professor of Mechanical Engineering
Abstract
Title: Development of a Static Aeroelastic Database Using NASTRAN SOL 144 for
Aircraft Flight Loads Analysis from Conceptual Design through Flight Test
Author: Ryan M. Haughey
Advisor: Razvan Rusovici, Ph. D.
This paper describes a method for predicting aircraft aerodynamic and inertial
loading on a structural finite element model, (FEM) based on static aeroelastic
coefficients. These coefficients are computed via interpolated spline methods within
NASTRAN Solution 144 (static aeroelastic solution) and the “TRIM” module to
connect the doublet-lattice model (DLM) and the structural finite element model for a
coupled solution. The database is created by selecting key breakpoints where linear
interpolation techniques can be utilized to develop and predict static aeroelastic
coefficients for the prediction of any aircraft state for a given transient solution. This
method is applicable from conceptual design through flight test.
The methodology described in this paper is essential for aircraft design and analysis
from initial stages of conceptual design through flight test and beyond. The procedure
for analysis varies slightly between the analysis types. However, the premise is, in
general constant. The variation stems most significantly from the source and
reliability of the data.
iii
Table of Contents
Table of Contents .................................................................................................... iv

List of Figures .......................................................................................................... vi
List of Table ............................................................................................................ vii
Acknowledgement ................................................................................................ viii
Dedication................................................................................................................ ix
Motivation .................................................................................................................1
Current Aircraft Design Practices .................................................................................... 1
Aircraft Design Phases ..................................................................................................... 3
Aircraft Flight Loads Requirements .........................................................................5
Military Specifications...................................................................................................... 5
Aircraft Integration Analysis ..................................................................................10
Integrated Schedule and Coordination ......................................................................... 10
NASTRAN Study Model ...........................................................................................16
NASTRAN Static Aeroelastics......................................................................................... 16
HA-144F NASTRAN Model ............................................................................................. 18
Database Generation Process .................................................................................25
Static Aeroelastic Trim Theory ...................................................................................... 25
Stability and Control Derivatives................................................................................... 34
Incremental Load Sets.................................................................................................... 42
Database Predicted Loads Results .........................................................................45
Control Surface Hinge Moments .................................................................................... 45
Utilizing the Database .............................................................................................48
Conceptual and Preliminary Design Usage ................................................................... 48
Detail Design and Flight Test ......................................................................................... 57
References ...............................................................................................................60
Appendix A Tabulated Rigid Stability Derivatives at Incremental Mach Values .61
Appendix B Tabulated Flexible Stability Derivatives at Incremental Mach Values
..................................................................................................................................63
Appendix C Flexible Stability Derivatives ..............................................................65
Appendix D Flexible vs. Rigid Stability Derivatives ..............................................69
Appendix E Unit Hinge Moment Database Per Mach and Variable ......................75
iv
Appendix F Proof of Concept Test Matrices ..........................................................78
Appendix G Baseline Matrix Generation ................................................................88
Appendix H Expanded Matrix Generation .............................................................89
Appendix I Trim Estimator .....................................................................................90
Appendix J Unit Loads Derivative Database Generator ........................................91
Appendix K Write Trim Include .............................................................................92
Appendix L Non-Dimensionalize Anglular Values ................................................93
Appendix M HA-144F Model Data..........................................................................94
v
List of Figures
Figure 1 – Vn Diagram for Symmetrical Flight .........................................................6

Figure 2 – Longitudinal Control and Load Factor Response for Dynamic Pitching
Maneuvers ..........................................................................................................9
Figure 3 SPLINE2 Bulk Data Entry ........................................................................17
Figure 4 NASTRAN Structural Model ....................................................................19
Figure 5 First Mode - Fuselage Yawing at 7.56Hz ..................................................20
Figure 6 Second Mode - First Fuselage and Wing Bending Mode at 9.79Hz .........21
Figure 7 Third Mode - Second Fuselage Bending and Wing Bending at 18.78Hz…
..........................................................................................................................21
Figure 8 NASTRAN Aero Model Mesh with Structural Model ..............................22
Figure 9 NASTRAN Aero Model with Control Surfaces and Structural Model .....22
Figure 10 NASTRAN Trim Input Bulk Data Entries ..............................................34
Figure 11 Baseline Trim Input .................................................................................36
Figure 12 .f06 Aerodynamic Derivative Ouptut ......................................................38
Figure 13 Derivative Database Representation ........................................................41
Figure 14 Example Monitor Point Loads .................................................................43
Figure 15 Elevator Hinge Moment vs. Pitch Rate, Mach 0.5, 0.6 ...........................44
Figure 16 Dynamic Pitch Initial Actuator Sizing Survey - Elevator Hinge Moment
..........................................................................................................................51
Sorted by Mach ................................................................................................52
vi
List of Table
Table 1 Stability Derivative Interpolation Check ....................................................39

Table 2 Incremental State Matrix .............................................................................42
Table 3 Unit Hinge Moments at Mach = 0.3 ...........................................................46
Table 4 Abrupt Pitch Initial Sizing Survey ..............................................................49
Table 5 Maximum Positive Elevator Hinge Moment Results .................................50
Table 6 Maximum Negative Elevator Hinge Moment Results ................................50
Table 7 Elevator Hinge Moment in Alpha and Surface Deflection Limit ...............54
vii
Acknowledgement
I would like to acknowledge the Northrop Grumman Corporation for assisting in

funding my education. Their continuous investment in my growth both financially
and in experience and education has helped to develop my skills.
I would also like to acknowledge my advisor, Dr. Razvan Rusovici for his continued
support over the past eight years working on multiple design projects and assisting
in undergraduate and graduate research.
Finally, I’d like to thank my co-workers, many of whom I had countless discussions
on this particular topic, and whose ideas helped to aid the foundation this paper
and who dealt with my early departures from the work day in the later portion of
the development of this thesis.
viii
Dedication
I would like to dedicate this paper to my parents, whose continued love and support
is by far the largest contributing factor to my success. They have supported my
dreams and have helped to guide me to where I am today.
ix
1
Motivation
Current Aircraft Design Practices

The field of flight loads analysis is over seemingly a small community of specialized
engineers, as compared to fields such as structural analysis or design, however the data and
analysis they produce is critical for the success of an aircraft development or modification
program. The common adage of aircraft design loads is that the loads are too high.
Consider the design of a brand-new airframe. Through the design phases, maturation of
data, process and other developments tend to cause iterations or revisions of analysis.
Through these cycles, structural analysts are working with structural design engineers to
optimize the airframe to lowest positive margins. If iteration i-1 has loads which are greater
than iteration i, the program can claim, “The loads from iteration i-1 were too high! Now
the structure has to be redesigned so that the aircraft can hit performance requirements.”
Whereas in the case where iteration i-1 loads are less than that of iteration i, the program
can claim, “The loads from iteration i are too high! Now the structure has to be redesigned
to meet airframe strength requirements.”
It is a continuous balance where the loads engineer must be sure that the changes between
cycles are gradual and predictable with aircraft (and data) maturation. For this to occur,
quick and computationally inexpensive methods must be utilized so that incremental
change impact can occur and the process for which the loads are developed is consistent
through each phase. Having consistency in methodology allows the analysts to focus on the
development of increasing fidelity, rather than developing new methods based on program
maturation. This paper aims to produce a methodology which can be used as a framework
for a common method from the aircraft conceptual design phase through flight test.
2
There is very little documentation on the fundamentals of aircraft loads analysis. One
textbook exists by Lomax, titled “Structural Loads Analysis for Commercial Transport
Aircraft: Theory and Practice” [1]. There are, however, no textbooks dedicated to the
structural loads analysis of military aircraft.
Even still, the Lomax textbook is significantly outdated and does not contain methods
which are required for design today’s aircraft, commercial or military. The major reasons
being that traditional loads development typically neglects aircraft aeroelastic effects and
the data produced has turned from external load tables and diagrams to distributed FEM
loads for aircraft detail analysis. This thesis aims to tackle both of those “new era”
methodologies of aircraft design through the usage of NASTRAN.
“In the design of aircraft, it is important to have an accurate simulation of both the
structural characteristics and the aerodynamic characteristics of the vehicle. For [quasi-
static] aerodynamic loads, MSC/NASTRAN uses unsteady aerodynamics at zero reduced
frequency.” [3] The methods in NASTRAN have become accepted practices for many
structural analysis methods for major defense contractors and civilian engineering
companies. It is assumed, in this paper, that the methods of NASTRAN are valid and that
the results of the NASTRAN codes are validated against industry standard.
A concern outlined in “Design Loads for Future Aircraft” [7], is stated as “With the
increased use of active control systems on aircraft, there is currently a strong need to revisit
some concepts used for conventional aircraft and to identify the correction to be brought
forward to existing procedures to compute the several loads affecting a military aircraft and
the effect of the active control system. Special attention has been given to cover these
items.” As well as “During the past few years there has been an increased interest of the
aircraft community on design loads for aircraft”. Consequently there was a workshop in
1996 SC73 on “Loads and Requirements for Military Aircraft” Elastic effects on design
loads were presented at a Workshop: “Static Aeroelastic Effects on High Performance
Aircraft.” These claims reflect the date of the release of the paper, which was 2002.
3
It is difficult to claim that the methods presented in this thesis are groundbreaking and
novel in that the methods used in industry practice vary greatly from company to company
and are typically held behind proprietary limits.
However, it can be said that the methods presented are unique in that there is not literature
that has been found to exist in such detail on the topic, and methods presented in textbooks
are only consistent with the foundation of the ideas presented by this thesis. The total
method presented (the development of the unit component loads database using
NASTRAN) is on the whole, a novel method.
Aircraft Design Phases

There are four main phases of aircraft development that this paper focuses on. These
phases are:
1. Conceptual Design – The initial design phase of an aircraft development

program where trade studies are conducted. Crude analysis is conducted,
typically with low fidelity models in order to run optimization simulations for
initial system level sizing requirements.
2. Preliminary Design – The second design phase of an aircraft development

program where one of the trade study designs from conceptual phase is
chosen to develop system level specification requirements.
3. Detail Design – The third design phase of an aircraft development program

where detail analysis and design is conducted for final aircraft design and
build. Typically at this point in the aircraft development program no
significant changes to the baseline design are made, as to not impede
development of surrounded and connected systems and structure.
4
4. Flight Test – The fourth phase of an aircraft development program where the
vehicle is tested through a flight test program. This paper is particularly
interested in the envelope expansion portion of the flight test program, which
is the responsibility of the loads team.
Conceptual design and flight test are perhaps two of the most challenging stages of
aircraft loads analysis because of the uncertainty of data. As the preliminary design
matures, and the detail design phase progresses, the aircraft design data should
mature to a point where the amount of conservative assumptions tends to go towards
zero. Unfortunately, during the conceptual design phase, the loads analyst must make
many assumptions so that the design optimization and trade studies can be efficiently
processed.
“Aircraft conceptual design traditionally utilizes simplified analysis methods and

empirical equations to establish the basic layout of new aircraft.” [4] The methods
provided in this paper reflect this theory.
During the flight test phase of a program, good correlation between the predicted
loading does not disturb the development of flight test efforts, such as envelope
expansion. [10] Typically, real time data prediction is needed as the aircraft is
undergoing the envelope expansion phase of the flight test program. Real time data is
necessary so that the engineers can safely assess and predict loading at the next test
point in the sky, based on correlated predicted data and data from the latest
maneuver.
5
Aircraft Flight Loads Requirements
This paper provides examples based on military specifications, however there are many
similarities between commercial and military requirements. The commercial and civilian
requirements are defined by the Federal Aviation Regulations Part 23 and Part 25.
Military Specifications
The primary specifications document for fixed wing military aircraft flight loads is known
as MIL-8861(b) Airplane Strength and Rigidity [2], which is a Department of Defense
document published in February of 1986. It is important to note the date of the latest
publication of the spec as it is over thirty years old.
The primary difference between Military specifications and commercial and civilian
requirements is the flexibility in requirements and deviations from the MIL-Spec is
common, while the commercial requirements are not as flexible. Typically, military aircraft
requirements are developed and agreed upon between the acquisition branch of the
department of defense and the contractor.
Symmetrical flight load factors are prescribed for each class of aircraft defined in the MIL-
Spec [2]. These classifications include: Fighter, Trainer, Attack, Transport, Reconnaissance
et. al. The symmetrical flight load envelope is described in the spec as well, shown in
Figure 1. The points on the diagram are outlined in the spec, and a special factor,
designated k, is used to apply uncertainty in regards to unsteady flow and buffet when the
aircraft is at stall.
The theory of quasi-steady flight loads typically does not (directly) account for unsteady
flow such as buffet at stall. These values can be confirmed during flight test, or if a
program can use wind tunnel testing to accurately predict this flow and corresponding
6
normal force coefficients, the value may be reduced or eliminated. Typically however,
critical flight loads tend to come at points not along the stall curve of the Vn diagram.
Figure 1 – Vn Diagram for Symmetrical Flight [2]
There are predefined maneuvers in the spec which outline typical flight maneuvers for
which the aircraft is designed. Some maneuvers are prescribed for only certain types of
aircraft. For example, an accelerated pitch maneuver and recovery is required for all
aircraft types, while the level flight roll (360 degree roll) is required only for aircraft of
type attack, fighter and trainer.
This paper demonstrates proof of concept for three types of quasi-steady maneuvers
prescribed by the MIL-8861(b) Specifications [2]. These maneuver types are accelerated
pitch and recovery maneuvers, accelerated roll and recovery maneuvers and abrupt rudder
kick maneuvers. All of these are muti-degree of freedom maneuvers. Of course, the same
methodology can be appended for three, four and five degree of freedom cases, which
include coupled degrees of freedom for maneuvers such as rolling pull-outs (RPO). Three
examples of this are also included that consider all 5 degrees of freedom, namely linear
accelerations in the Y and Z directions, and roll accelerations about all three principal axes
7
and roll rates about all three principle axes. A more detailed discussion on this is presented
in Database Predicted Loads Results.
The three maneuvers are described below in the MIL-8861(b) Specification [2]:
“3.2.2 Accelerated pitch maneuver and recovery. The airplane shall be in the basic
high-drag, and dive-recovery configurations. The airplane initially shall be in steady
unaccelerated flight at the airspeed specified for the maneuver and trimmed for zero
control forces at that airspeed. The airspeed shall be constant until the specified load factor
has been attained. The load factors to be attained shall be all values on and within the
envelope bounded by O, A, B, C, D, and E of Figure 1 (of this document). Except as noted
the load factor at each airspeed shall be attained as specified for all center of gravity
positions, and also for the maximum-aft center of gravity position, and by a cockpit
longitudinal control movement resulting in a triangular displacement-time curve as
illustrated by the solid straight lines of Figure 2 (of this document) provided that the
specified load factor can be attained by such a control movement; otherwise by the ramp-
style control movement illustrated by the dashed straight lines of Figure 2 (of this
document) . The time t, is specified in Table I (in reference 2). For the ramp-style control
movement, the time t2 shall be the minimum time that the control is held at the stops to
attain the specified load factor.” [2]
“3.3.1 Rolling maneuvers. The airplane shall be in the basic: high-drag and specified store
configurations. The airspeeds shall be all airspeeds up to limit speed (V,). During the
maneuver, the directional control shall be: a. Held fixed in its position for trim with zero
rudder control force in wings-level flight at the speed required, and b. Displaced as
necessary to maintain zero sideslip up to limits of the rudder authority. The cockpit lateral
control shall be displaced to all the displacements to the maximum available displacement
attainable by a pilot lateral control force of 60 pounds (two equal and opposite 48-pound
forces applied at the circumference of the control wheel) by application of the control force
in not more than 0.1 second for airplanes with stick controls and not more than 0.3 second
for airplanes with wheel controls; for automated flight control type systems application of
8
the maximum control surface(s) authority is required. The control force(s) or authority
shall be maintained until the required change in angle of bank is attained, except that, if a
roll rate greater than 270 degrees per second would result, the control position may be
lessened or authority modified, subsequent to attainment of the maximum rolling
acceleration, to that position resulting in a roll rate of 270 degrees per second. The
maneuver shall be checked by application of the maximum available displacement
attainable with a 60-pound lateral control force (two equal and opposite 48-pound forces
applied at the circumference of the control wheel) applied in not more than 0.1 second for
stick controls and in not more than 0.3 second for wheel controls. For automated flight
control type systems, maximum lateral control surface(s) authority shall be used.” [2]
“3.3.3.3 High speed rudder kick. The airplane shall be in the basic and high-drag
configuration at speeds up to V, for VA, VF, and VT airplanes, and up to VH for other
type aircraft. The cockpit directional control shall be displaced to the maximum
displacement attainable with a 180-pound directional-control force applied in not more
than 0.2 second. The control force shall be maintained until the maximum over-swing
angle of slide-slip is attained and the airplane attains a steady sideslip. Recovery shall be
made by reducing the directional-control displacement to zero in not more than 0.2
second.” [2]
The accelerated pitch maneuver and recovery references Figure 2 when stating Figure 3.
These maneuvers were chosen as example maneuvers for this paper due to their dominance
on aircraft design sizing. Particular components outlined in this paper are concerned with
control surface hinge moment, as there is a direct correlation for the reader to make a
connection to loading vs. maneuver. For other structural components, such as integrated
fuselage torsion/bending loads, or wing shear/bending loads are not always as directly
responsive to these maneuver types.
9
Figure 2 – Longitudinal Control and Load Factor Response for Dynamic Pitching Maneuvers
[2]
10
Aircraft Integration Analysis
Integrated Schedule and Coordination

Perhaps one of the biggest challenges of aircraft design is the integration of the system at
the aircraft level. It takes a deep understanding of the interaction of each particular system
(including the sciences of flight) to develop an integrated schedule so that the best data
available can be used for analysis throughout each of the development phases. While that
particular discussion in detail is outside the scope of this thesis, the assumption is that a
well planned integrated schedule allows for ideal data to be available to the flight loads
analyst through development. This section covers each of the primary disciplines which the
analyst needs to be concerned with, and the corresponding data to be expected throughout
each of the design phases.
FINITE ELEMENT MODEL (STRUCTURAL ANALYSIS AND DESIGN)
Typically, the customer of the flight loads team is the structural finite element modeling
(FEM) team, and this paper will assume that. The FEM team is responsible for, among
other things, developing a FEM of the vehicle so that detail analysis and design can be
completed, aircraft flexibility can be assessed and may even aid in the development of
parametric mass properties. Just as the fidelity of the analysis of the flight loads team tends
to increase as the program progresses, so does the finite element model. It is important that
during the various phases of development, the stiffness of the global vehicle is tracked. The
benefit of this tracking is predicting impacts on flight loads as a function of the effects of
static aeroelasticity, which is the effect on the aerodynamic loading due to the deformations
of the structure. Typically, as a structure becomes stiffer, the aerodynamic loads will
increase. This will be evident as the flexible to rigid aircraft derivatives are presented in
this paper.
11
The FEM is both an input and the end product of the flight loads analysis as the model will
have the “critical” loads applied to it so stress analysis can be completed.
FLIGHT CONTROLS
With the advent of fly-by-wire and even direct hydraulic systems, the job of the flight
loads analyst becomes increasingly difficult due to the quick and sometimes hard to
recognize changes in the flight control system of the vehicle which may have significant
effects on the flying characteristics of the vehicle. These changes can be strictly at the
software level, where feedback and control gains are modified to meet flying quality
standards or may change as aerodynamic data changes or mission requirements change.
Typically, flight control development lags behind structural design due to the nature of the
schedule of the aircraft development program. It is necessary for the analyst to understand
the impacts of potential change and carry relatively significant margin as to protect the
vehicle from under-design.
This paper focuses on various assumptions that the analyst will need to make through the
different portions of the aircraft development phases. Depending on the program,
maximum maneuvering capability may be defined which restrict both the structural design
and flight controls. While these values may be good values for initial sizing, if any margin
can be reduced by time accurate aircraft state data, i.e. transient time history data, it is
desired that these values be used, such that the risk of the outcome is not increased due to
potential flight controls updates. In other words, the vehicle should not be overdesigned to
maximum maneuvering capability at every point in the sky if reliable time histories are
available to the analyst. It is up to the analyst to carry margin in the analysis if this is in
fact the case, in addition to the factor of safety.
For the conceptual design phases of the program, it is assumed that no aircraft
maneuvering requirements are set, and that general kinematics should be studied. In the
later phases of the preliminary design and certainly during detail design, more accurate
aircraft flying qualities are set and aircraft maneuvering should be well defined. It is at
12
these stages where the analyst must consider particular data sets for the quasi-steady
analysis, whether it be time slices from time accurate data, tabulated maximum rates and
accelerations as defined by aircraft control laws, or some other form. The methods
selected, however, have no change to the method for developing the loads.
If the usage of the time accurate six degree of freedom simulation is utilized, however, it
would be of interest to integrate a method for calculating loads at every time point at
specific monitor points of interest. This is important due to the load lag during aircraft
maneuvering.
MASS PROPERTIES
The mass properties team is integral in tracking and tabulating aircraft mass properties,
both on the total aircraft and detail part level. Through the stages of aircraft development,
the aircraft sizing will vary due to the loads that are produced to design the vehicle, along
with the development of the stiffness of the FEM. Therefore, the mass properties team will
have to track that development. As the program progresses it is necessary to capture
updates as mass properties change from parametric and predicted weights to actual weights
once the drawing has been released.
Mass properties will also have data involving requirements for payload, such as passengers
and cargo for commercial application or weapons and cargo for military applications. The
distribution of these payload items can be assumed to have significant impact on the trim
of the vehicle and the resulting loads.
Both the total aircraft mass properties and the distribution of mass are important to
developing a reliable aeroelastic model. Since the aeroelastic effects are produced from the
mass and stiffness matrices, the data from the mass properties team is just as important as
the data from the FEM team, i.e. the mass and the stiffness data.
The modeling techniques for the mass distribution may vary throughout the program
development. There are many methods which the mass can be represented on the structural
13
finite element model. Within NASTRAN, there are methods of defining material densities,
which along with the volumetric properties of the elements will define the mass properties.
This method should be used with caution, as the data may not always (and probably won’t)
sum back to the total aircraft numbers defined by the mass properties team. Other methods
include the use of lumped masses. The refinement of the lumped masses is up to the
analyst to capture distribution and accurate mode shapes without inducing fictitious modes.
PROPULSION
The integration of the propulsion system into the aeroelastic model can be one of the more
challenging disciplines. The propulsion system can have both effects on the total vehicle,
and local effects due to inlet and exhaust pressure differentials.
The total vehicle effects of the propulsion system are any off axis thrust terms which cause
aircraft pitching moment or loads along the aircraft body axis.
The example used in this paper neglects effects due to propulsion, however some aircraft
may have considerable effects on trim due to the propulsion system. Some may be as
complex as aircraft such as the Lockheed F-35 which has thrust vectoring, or as simple as
the pitching moment due to the low hung wings of a Boeing 737. Ducted inlets, such as
those on fighters can cause accelerated flow on the surrounded surfaces which will vary the
nominal aerodynamics without propulsive effects.
AERODYNAMICS
There are many methods for developing aerodynamics of the vehicle. The term
aerodynamics includes both total aircraft aerodynamics, such as those obtained from force
balance data from a wind tunnel test, i.e. aircraft pitching and lift coefficients, as well as
control surface and door hinge moments and distributed pressure coefficients such as those
obtained from computational fluid dynamics (CFD) and pressure sensitive paint wind
tunnel tests.
14
Typically, wind tunnel testing tends to be very expensive and may only occur over certain
flight regimes. The point at which wind tunnel testing occurs for a program can vary,
however, it can be assumed that during the initial trade studies, no wind tunnel testing
exists.
CFD data is typically limited at initial stages of the design phase. The loads analyst should
work with the CFD analyst to obtain solutions that will be used as the design refinement
goes forward, so that the doublet-lattice methods can be replaced by a complete CFD
database of distributed pressure solutions. Note, these CFD results obtained from high
order methods such as Euler and Navier-Stokes will be rigid aerodynamics and do not
account for flexibility. It will be up to the loads analyst to take these results and correct
them for aeroelastic effects using NASTRAN. These methods will be described in more
detail.
INTEGRATION
It can easily be seen how interactive each of these disciplines are with each other. The
structural design analysts will use the loads developed to size parts. The designed/sized
parts will have weights, which the mass properties team will tabulate. These tabulated
values are fed into the flight controls analysis to tune gains for flying qualities. The
resulting flying quality values are based on aircraft coefficients developed from the
aerodynamics team. The fallout aircraft maneuvering capability is fed back into the loads
analysis along with mass properties inertial distributions, the FEM aircraft stiffness (based
on the aircraft part sizing) and the propulsion system effects.
It is absolutely critical that each of these disciplines are well integrated for the success of
the aircraft loads assessments as the process is circular, in that the cycle of analysis is
lagged by n-1 for each discipline. Take for example analysis cycle 2 for the aircraft quasi-
steady aeroelastic loads. The data the loads analyst will use is based on the stiffness of the
finite element model from the previous sizing effort, along with some aircraft maneuvering
capability based on the inertia properties from the previous sizing effort. This means that
15
the loads assessment is performed based on the result from the previous loads assessment
and the impacts are analyzed on the sizing from the previous assessment.
It is crucial that this inherent lag in design analysis is understood and captured in the loads
analysis. The methods described in this paper allow for easy transition of updated data to
be positioned into the same database that was previously used in the previous assessment
of loads.
Moreover, trade studies and impact analysis can be conducted based on the previously
developed database for trends of increased or decrease stiffness and inertial distribution.
For example, typically, an increase in wing stiffness tends to lead to increased wing
bending loads due to an inherent outboard shift in the center of pressure due to wing flex.
A scaled reduction in derivative stiffness in the database can give a quick back of the
envelope investigation into wing stiffness changes. Similar studies can be conducted based
on the inertial distribution, and maneuvering capability changes. Small mass property
changes can be assumed to have no impact on the mode shapes of the vehicle but may have
an impact on the inertial distribution and inherent inertia relief. This assessment can be
conducted by scaling the inertia relief to mimic what is expected. Similarly, for changes in
aircraft maneuvering rates and accelerations, a quick study on the previously defined
critical conditions with the existing database at new rates and accelerations can show
incremental impacts due to the increases or decrease in aircraft maneuvering capability.
16
NASTRAN Study Model
NASTRAN Static Aeroelastics

The method used to develop the flexible loads database is derived from solutions of the
NASTRAN Solution 144 module, which is the static aeroelastics module in NASTRAN.
The method uses “a new stiffness formulation [which] has been developed and added to
MSC/NASTRAN to solve the basic trim load problem and estimated aeroelastic stability
derivatives at subsonic speeds.” [5] There are two primary purposes of NASTRAN
Solution 144. These are the study of “Static Aeroelastic Response” and “Aeroelastic
Divergence.” [5] While the methods described in this paper can aid in the evaluation of
static aeroelastic divergence (such as aileron reversal), the intent of this paper focuses on
the former, “Static Aeroelastic Response”, namely the response of the structural vehicle to
the inertial and aerodynamic loading produced at various flight conditions.
This method produces a coupled solution of the structural stiffness response to the quasi-
steady aerodynamic loading. It is a method of splines which creates interpolated solution
sets on corresponding grid points between the aerodynamic model and the structural
model. This coupling allows for structural monitor points and corresponding derivatives
for the monitor points.
There are two methods of splines in NASTRAN. Both of which are inherently “two
dimensional”. They assume that the forces on the coupled models are normal to one or two
orthogonal planes and that the aerodynamic geometry consists of collections of points on a
plane or along an axis. The two methods are the Harder Desmarais infinite plate spline,
also known as SPLINE1 in NASTRAN, and the beam spline, also known as SPLINE2 in
17
NASTRAN. The spline methods used in the example in this paper are SPLINE2.
Figure 3 SPLINE2 Bulk Data Entry
SPLINE2 – Denotes the call in NASTRAN for the spline method in the following bulk
data entry
EID – Defines the element ID
CAERO – Interpolated aerodynamic panel ID
ID1/ID2 – First and last box or body element to be interpolated on
SETG – The set of inertia grids to interpolate the aerodynamic forces to
DZ – The linear attachment flexibility.
DTOR – The torsional flexibility ratio = EI/GJ.
CID – The coordinate system (rectangular) for which the y-axis defines the spline.
DTHX/DTHY – Rotational attachment flexibility.
USAGE – Input to define whether force, displacement or both methods are used.
The process of the aeroelastic solution can be formulated into distinct portions. These
portions are aerodynamic stability derivatives, aerodynamic pressure distributions
(including aeroelastic effects), static trim solutions, external discreet loading, and control
surface hinge-moments. These solutions are part of the family of aerodynamic solutions.
18
Additionally, internal freebody loads, deflections and element stresses can also be
calculated as parts of the structural solutions.
For the subsonic solution using CAERO1 and CAERO2, aerodynamic and body elements,
respectively, the doublet lattice method is used. Because no license for supersonic flow
was available (requires ZONA51 Panel License) all analysis subsonic element with the
aforementioned element types. For the purpose of this study, however, that is sufficient as
the database generation is not limited in any way by Mach number, and therefore,
supersonic flow is in fact unnecessary. However, supersonic analysis is a possible
extension of this paper.
NASTRAN solution sequence 144 has multiple methods of deriving aerodynamic stability
derivatives. These methods include rigid, unrestrained flexible and restrained flexible. This
paper is interested in the unrestrained flexible solutions, which will be what is presented.
These solutions allow for the inclusion of inertial effects to be used in the calculation of the
aerodynamic stability derivatives, so that trim solutions can be predicted, after an initial set
of stability derivatives is developed.
HA-144F NASTRAN Model

The model used for this research is one of the common NASTRAN Aeroelastic study
models. Aeroelastic models in NASTRAN consist of two independent finite element model
types which are then coupled through the use of splines. One of the models, which is
referred to as the structural model, contains classical NASTRAN structural elements, i.e.
CROD, CBAR, CQUAD4, etc., while the aerodynamic model contains non-structural
elements such as CAERO1 and CAERO2 panels.
The aerodynamic portion of the model is made up strictly of CAERO1 and CAERO2
panels. CAERO1 panels are aerodynamic panels that represent un-interfered panels.
CAERO2 panels are aerodynamic panels used to represent wing-body interaction.
19
The aerodynamic reference data for the model uses a reference area of 400 [ft2], a reference
span of 100 [ft] and a reference chord of 10 [ft].
The structural model, or the portion of the model which contains the stiffness and mass
properties, is shown in Figure 4.
Figure 4 NASTRAN Structural Model
The model consists of 19 bar elements (CBAR), 14 concentrated mass elements (CONM2),
and 10 RBE3 rigid elements. The 19 bar elements are colored yellow in the model shown
in Figure 4. Some CONM2 elements are shown in pink, light blue and dark blue. They are
held at the poles of the rigid RBE3 elements which are colored white. The rest of the
CONM2s are colored green along the fuselage. Notice, the mass elements are lumped and
are not located on the control surfaces themselves but along the hinge line. The
underhanging pylons also have no mass attached to them.
20
It would be expected that further along in aircraft development, this model will become
more refined with two dimensional elements such as CQUAD4s and CTRIA3s along with
beam, rod and bar elements. The inertial distribution should increase significantly as well.
The quick look at aircraft mode shapes can be a good indication on the aeroelastic effects
of the vehicle. As the vehicle FEM and inertial properties change, identifying key mode
shapes can be critical in predicted aeroelastic effects as the design matures. The first three
mode shapes (really these are the 7th, 8th and 9th mode shapes, given 6 unrestrained rigid
body modes), which are shown in Figure 5, Figure 6 and Figure 7. They occur at 7.56 Hz,
9.79Hz and 18.18Hz, respectively. The first mode is a fuselage yawing mode. The second
is a combined fuselage bending and symmetric wing bending mode. The third is second
combined fuselage bending and symmetric wing bending mode.
Figure 5 First Mode - Fuselage Yawing at 7.56Hz

21
Figure 6 Second Mode - First Fuselage and Wing Bending Mode at 9.79Hz
Figure 7 Third Mode - Second Fuselage Bending and Wing Bending at 18.78Hz
22
Figure 8 NASTRAN Aero Model Mesh with Structural Model
Figure 9 NASTRAN Aero Model with Control Surfaces and Structural Model
23
The NASTRAN Aero Model and the NASTRAN Structural Model on connected through a
spline, or interpolation method which “connects” specific grid points of both models to
transfer aerodynamic loading to the structural model and provide aircraft stiffness and mass
properties to the aero model for aeroelastic effects. These combined models can be seen in
Figure 8 and the modeled control surfaces can be seen in Figure 9.
The bulk data file input for the spline is shown in Appendix M
HA-144F Model Data. For reference, the bulk data entry format can be seen in Figure 3.
The splines used were the same that were existing on the NASTRAN HA144F model and
no modifications were made. The aerodynamic modeling and the structural modeling are
also consistent with what the initial example had defined, and no modifications were made.
The details of the modeling and the exact model itself are independent of the procedure
described.
In theory, the methods described in this paper are able to be used on any fixed wing aircraft
type, including forward swept wings, wings with positive and negative dihedral, V-tail
aircraft, variable sweep aircraft, traditional swept wings, high wing, low wing, canard,
blended wing body, and even flying wing body. Basically, if the aircraft can fly, it can be
modeled with these methods.
The HA 144F model was selected due to its particularly unique properties and most
notably the forward swept wings and canards. Forward swept wings tend to have issues
with static aeroelasticity, namely divergence. The particulars of that discussion are outside
the scope of this paper; however, the flexible effects of wing torsion can be seen in
Appendix D
24
Flexible vs. Rigid Stability Derivatives. The aircraft has a higher dCz/dα value when
flexibility is considered. This is due to the fact that as the aircraft increases in alpha, the
wing tends to torque to increases alpha, which in some cases can lead to divergence issues.
Portions of the processes described in this paper can be used to set requirements for wing
torsion stiffness due to static aeroelastic divergence. One of the most well-known examples
of this type of aircraft was the Grumman X-29 demonstrator.
It is interesting to note the first mode shape of the vehicle. The first mode is a fuselage
yawing mode which is typically uncharacteristic of flight vehicles. This indicates that
perhaps the fuselage design needs to be stiffer. Typically, the first modes are wing bending
or torsion modes before fuselage modes. Studies of the mode shapes can help identify
static aeroelastic effects due to stiffness properties of the vehicle.
25
Database Generation Process
Static Aeroelastic Trim Theory

The theory for the development of the database generation used in this thesis is reliant on
linear assumptions. The doublet-lattice model used in the aerodynamic generation (if
chosen) assumes a constant stability derivative as a function of Mach only. Therefore, the
linearity of the control derivatives is assumed across all variations of the control variables,
at a given Mach. This theory, however, though limited in direct application can be used to
capture non-linear effects. This can be accomplished via a piecewise-linear database
generation. This method is shown in the way the database is generated, piecewise linear, in
Mach, neglecting non-linearity in any other terms, such as stall effects at extreme alphas,
or betas. Additional non-linearities can occur due to downwash effects, propulsive effects,
etc. It is up to the analyst to understand these non-linearities and develop the breakpoints of
the database to capture them.
First, a database of aerodynamic stability derivatives must be developed. These stability

derivatives are a function of the control variables used in the trim analysis. These control
variables include alpha and beta, which are the relative flow angles, and control surface
deflections, i.e. aileron, elevator and rudder. Additionally, set parameters which need to be
considered are roll, pitch and yaw rates. Neglected aerodynamic effects are acceleration
terms, which describe the inertial loading.
A summary of the resulting rigid database can be seen in Appendix A

Tabulated Rigid Stability Derivatives at Incremental Mach Values and the
corresponding flexible database can be seen in Appendix B
Tabulated Flexible Stability Derivatives at Incremental Mach Values.
Note, the aerodynamic stability derivatives in X are always zero. This is because the
database assumes only 5 degrees of freedom for trim, Y, Z, MX, MY, and MZ. In order to
26
satisfy the determined system for static aeroelastic trim, the number of free variables must
equal the number of degrees of freedom for the equation of motion. In other words, there
must be n-number of unknowns to be solved for, given n-number of equations. The
solutions being sought are in the form Ax=B where B is the inertial loading, A is the
dimensionalized stability derivatives and x is the state vector of the free and prescribed
variables.
It is common to neglect the drag term for two particular reasons. The first is that the
doublet-lattice method does not consider viscous drag, which can only be predicted
through higher order CFD methods such as Navier-Stokes. More importantly, though, is
that the aircraft body axis is always assumed to be aligned to the aircraft inertial axis, and
that the aircraft can be in the same state whether thrust is at maximum power or it is at idle.
For example, in flight test a pilot may be steady level and pull back abruptly on the stick at
full throttle. Similarly, the pilot can be at the same airspeed and same altitude at idle power
by diving into the maneuver from a higher altitude, so that the aircraft has state data that is
(exactly) the same except for the thrust setting for that maneuver.
In the following derivations, roll, pitch and yaw rates are represented by the letters “p”, “q”
and “r”, respectively and their time derivatives, or accelerations are represented with dot
notation. The true airspeed is represented by “V’. The aircraft body forces are represented
by “X”, “Y” and “Z” and the aircraft body moments are represented by “L”, “M”, and “N”,
each representing the 3 forces and moments on the principle body axis, in the order of roll,
pitch and yaw. The relative flow angles of angle of attack and sideslip angles are
represented by “α” and “β”, respectively, and the control surface deflections are
represented by “δ”.
The general form of the aerodynamic loading can be represented as
Equation 1 CY Aero Loading
𝑝𝑏̅ 𝑞𝑐̅ 𝑟𝑏̅

𝐶𝑌 𝑎𝑒𝑟𝑜 = 𝐶𝑌𝛼 𝛼 + 𝐶𝑌𝛽 𝛽+𝐶𝑌𝛿𝑎 𝛿𝑎 + 𝐶𝑌𝛿𝑒 𝛿𝑒 +𝐶𝑌𝛿𝑟 𝛿𝑟 + 𝐶𝑌𝑝 + 𝐶𝑌𝑞 + 𝐶𝑌𝑟
2𝑉 2𝑉 2𝑉
27
and
Equation 2 CZ Aero Loading
𝑝𝑏̅ 𝑞𝑐̅ 𝑟𝑏̅

𝐶𝑍 𝑎𝑒𝑟𝑜 = 𝐶𝑍𝛼 𝛼 + 𝐶𝑍𝛽 𝛽+𝐶𝑍𝛿𝑎 𝛿𝑎 + 𝐶𝑍𝛿𝑒 𝛿𝑒 +𝐶𝑍𝛿𝑟 𝛿𝑟 + 𝐶𝑍𝑝 + 𝐶𝑍𝑞 + 𝐶𝑍𝑟
2𝑉 2𝑉 2𝑉
The terms can be combined and cast into matrix form yielding:
Equation 3 General Matrix Aerodynamic Loading
𝛼
𝛽
𝛿𝑎
𝛿𝑒
𝑌
𝐶𝑌𝛼 𝐶𝑌𝛽 𝐶𝑌𝛿𝑎 𝐶𝑌𝛿𝑒 𝐶𝑌𝛿𝑟 𝐶𝑌𝑝 𝐶𝑌𝑞 𝐶𝑌𝑟 𝛿𝑟
𝑞̅𝑆 ̅
= [ ] 𝑝𝑏
𝑍 𝐶𝑍𝛼 𝐶𝑍𝛽 𝐶𝑍𝛿𝑎 𝐶𝑍𝛿𝑒 𝐶𝑍𝛿𝑟 𝐶𝑍𝑝 𝐶𝑍𝑞 𝐶𝑍𝑟
{𝑞̅𝑆} 2𝑉
𝑞𝑐̅
2𝑉
𝑟𝑏̅
{ 2𝑉 }
Each of the stability derivatives are non-dimensionalized by the freestream dynamic

pressure, 𝑞̅, and the reference area, S. For trim, it is assumed that the free trim variables, or
variables used to trim the aircraft are 𝛼, 𝛽, and control surface deflection only. It is
convenient to rewrite Equation 3 in the form of a linear combination of free variables and
prescribed variables, where the prescribed variables are the aircraft rotational rates. These
rotational rates contribute to the aircraft roll, pitch and yaw damping terms. In some cases,
these terms may be negligible for inertial effects, but must be included for aerodynamic
28
effects. The roll, pitch and yaw rates induce local flow angles of alpha and beta on the
panels, resulting in local reductions or increases of aerodynamic pressures and may, such
as in the case of this example, be significant due to the high rates (>100 deg/s) of a fighter
type aircraft.
Equation 4 Modified Coefficient Aerodynamic Loading g
̅
𝑝𝑏
𝛼
𝑌 2𝑉
𝐶𝑌𝛼 𝐶𝑌𝛽 𝐶𝑌𝛿𝑎 𝐶𝑌𝛿𝑒 𝐶𝑌𝛿𝑟 𝛽
𝑞̅ 𝑆 𝐶𝑌𝑝 𝐶𝑌𝑞 𝐶𝑌𝑟 𝑞𝑐̅
= [ ] 𝛿𝑎 + [ ]
𝑍 𝐶𝑍𝛼 𝐶𝑍𝛽 𝐶𝑍𝛿𝑎 𝐶𝑍𝛿𝑒 𝐶𝑍𝛿𝑟 𝛿𝑒 𝐶𝑍𝑝 𝐶𝑍𝑞 𝐶𝑍𝑟 2𝑉
{𝑞̅ 𝑆} ̅
𝑟𝑏
{ 𝛿𝑟 }
{ 2𝑉 }
Since the aircraft rates are prescribed of roll rate, p, pitch rate, q and yaw rate, r, there are
three possible conditions that will exist; either the terms will be non-zero, zero or some
combination of both. The system can be solved for directly in either scenario. Therefore,
the second term on the RHS of the equation can be moved to the LHS of the equation.
Equation 5 Modified Coefficient Aerodyanamic Loading, Ax = B form
̅
𝑝𝑏 𝛼
𝑌𝐴 2𝑉
𝐶𝑌𝛼 𝐶𝑌𝛽 𝐶𝑌𝛿𝑎 𝐶𝑌𝛿𝑒 𝐶𝑌𝛿𝑟 𝛽
𝑞̅ 𝑆 𝐶𝑌𝑝 𝐶𝑌𝑞 𝐶𝑌𝑟 𝑞𝑐̅
−[ ] = [ ] 𝛿𝑎
𝑍𝐴 𝐶𝑍𝑝 𝐶𝑍𝑞 𝐶𝑍𝑟 2𝑉 𝐶𝑍𝛼 𝐶𝑍𝛽 𝐶𝑍𝛿𝑎 𝐶𝑍𝛿𝑒 𝐶𝑍𝛿𝑟 𝛿𝑒
{𝑞̅ 𝑆} ̅
𝑟𝑏 { 𝛿𝑟 }
{ 2𝑉 }
Similarly, the aircraft roll, pitch and yawing moments can be expressed:
29
Equation 6 Aeodynamic Moment Coefficients, Matrix Form
𝐿𝐴 ̅
𝑝𝑏 𝛼
𝑞̅ 𝑆𝑏̅ 𝐶𝐿 𝑝 𝐶 𝐿𝑞 𝐶𝐿𝑟 2𝑉 𝐶𝐿𝛼 𝐶𝐿𝛽 𝐶𝐿𝛿𝑎 𝐶𝐿𝛿𝑒 𝐶𝐿𝛿𝑟 𝛽
𝑀𝐴
− [𝐶𝑀𝑝 𝐶𝑀𝑞 𝐶𝑀𝑟 ] 𝑞𝑐̅ = [𝐶𝑀𝛼 𝐶𝑀𝛽 𝐶𝑀𝛿𝑎 𝐶𝑀𝛿𝑒 𝐶𝑀𝛿𝑟 ] 𝛿𝑎
𝑞̅ 𝑆𝑏̅ 2𝑉 𝛿𝑒
𝑁𝐴
𝐶𝑁𝑝 𝐶 𝑁𝑞 𝐶 𝑁𝑟 ̅ 𝐶𝑁𝛼 𝐶𝑁𝛽 𝐶𝑁𝛿𝑎 𝐶𝑁𝛿𝑒 𝐶𝑁𝛿𝑟
𝑟𝑏 { 𝛿𝑟 }
{ 𝑞̅ 𝑆𝑐̅ } { 2𝑉 }
Attention is turned towards the inertial portion of the problem to create a system of
equations to solve for the statically determinant trim analysis.
It is assumed that the aircraft inertial loading has small deformations such that
Euler rigid body motion can used, which is expressed as:
Equation 7 Inertial Linear Acceleration Loading
𝑋𝐼
[ 𝑌𝐼 ] = 𝑊[𝑁𝑥 𝑁𝑦 𝑁𝑧 ] = 𝑊[ 0 𝑁𝑦 𝑁𝑧 ]
𝑍𝐼
and
Equation 8 Euler Rigid Body Mechanics
𝐿𝐼 𝐼𝑥𝑥 𝐼𝑥𝑦 𝐼𝑥𝑧 𝑝̇ 𝑝 𝐼𝑥𝑥 𝐼𝑥𝑦 𝐼𝑥𝑧 𝑝

[𝑀𝐼 ] = [𝐼𝑦𝑥 𝐼𝑦𝑦 𝐼𝑦𝑧 ] {𝑞̇ } + {𝑞 } × [𝐼𝑦𝑥 𝐼𝑦𝑦 𝐼𝑦𝑧 ] {𝑞 }
𝑁𝐼 𝐼𝑧𝑥 𝐼𝑧𝑦 𝐼𝑧𝑧 𝑟̇ 𝑟 𝐼𝑧𝑥 𝐼𝑧𝑦 𝐼𝑧𝑧 𝑟
The rigid body moments can be expanded:

30
Equation 9 Euler Rigid Body Rolling Moments Expanded
𝐿𝐼 = 𝐼𝑥𝑥 𝑝̇ − 𝐼𝑦𝑧 (𝑞 2 − 𝑟 2 ) − 𝐼𝑧𝑥 (𝑟̇ + 𝑝𝑞 ) − 𝐼𝑥𝑦 (𝑞̇ − 𝑟𝑝) −

(𝐼𝑦𝑦 − 𝐼𝑧𝑧 )𝑞𝑟
Equation 10 Euler Rigid Body Pitching Moments Expanded
𝑀𝐼 = 𝐼𝑦𝑦 𝑞̇ − 𝐼𝑧𝑥 (𝑟 2 − 𝑝2 ) − 𝐼𝑥𝑦 (𝑝̇ + 𝑞𝑟) − 𝐼𝑦𝑧 (𝑟̇ − 𝑝𝑞 ) −

(𝐼𝑧𝑧 − 𝐼𝑥𝑥 )𝑟𝑝
Equation 11 Euler Rigid Body Yawing Moments Expanded
𝑁𝐼 = 𝐼𝑧𝑧 𝑟̇ − 𝐼𝑥𝑦 (𝑝2 − 𝑞 2 ) − 𝐼𝑦𝑧 (𝑞̇ + 𝑟𝑝) − 𝐼𝑧𝑥 (𝑝̇ − 𝑞𝑟) −

(𝐼𝑥𝑥 − 𝐼𝑦𝑦 )𝑝𝑞
If a symmetric aircraft (about the XZ-plane) is assumed, the inertia matrix reduces to:
Equation 12 Inertia Matrix
𝐼𝑥𝑥 0 𝐼𝑥𝑧
𝑰=[0 𝐼𝑦𝑦 0]
𝐼𝑧𝑥 0 𝐼𝑧𝑧
Equation 13 Symmetric Euler Rigid Body Pitching Moments
𝑳𝑰 = 𝑰𝒙𝒙 𝒑̇ − 𝑰𝒛𝒙 (𝒓̇ + 𝒑𝒒) − (𝑰𝒚𝒚 − 𝑰𝒛𝒛 )𝒒𝒓

31
Equation 14 Symmetric Euler Rigid Body Rolling Moments
𝑀𝐼 = 𝐼𝑦𝑦 𝑞̇ − 𝐼𝑧𝑥 (𝑟 2 − 𝑝2 ) − (𝐼𝑧𝑧 − 𝐼𝑥𝑥 )𝑟𝑝
Equation 15 Symmetric Euler Rigid Body Yawing Moments
𝑁𝐼 = 𝐼𝑧𝑧 𝑟̇ − 𝐼𝑧𝑥 (𝑝̇ − 𝑞𝑟) − (𝐼𝑥𝑥 − 𝐼𝑦𝑦 )𝑝𝑞
Equation 13 through Equation 15 is the case for the HA144F model and is typically a good
approximation for most practical aircraft analysis.
To satisfy static equilibrium, the inertial forces and moments plus the aerodynamic forces
and moments must sum to zero. Therefore, the relationship can be made such that
(assuming 𝑋𝐼 = 0 and 𝑋𝐴 = 0 always),
Equation 16 Aircraft Steady State Balanced Condition
𝑌𝐼 𝑌𝐴
𝑍𝐼 𝑍𝐴
𝐿𝐼 = − 𝐿𝐴
𝑀𝐼 𝑀𝐴
[ 𝑁𝐼 ] [ 𝑁𝐴 ]
Substituting in the inertia terms via the relationship of inertial and aerodynamic loading,
the combined equation becomes:
32
Equation 17 General Control Trim Solution for Balance Flight Maneuver
𝑌𝐼
𝑞
̅𝑆
𝑍𝐼 𝐶𝑌𝑝 𝐶𝑌𝑞 𝐶𝑌𝑟 ̅
𝑝𝑏 𝐶𝑌𝛼 𝐶𝑌𝛽 𝐶𝑌𝛿𝑎 𝐶𝑌𝛿𝑒 𝐶𝑌𝛿𝑟
𝛼
𝑞
̅𝑆 𝐶𝑍𝑝 𝐶𝑍𝑞 𝐶𝑍𝑟 2𝑉 𝐶𝑍𝛼 𝐶𝑍𝛽 𝐶𝑍𝛿𝑎 𝐶𝑍𝛿𝑒 𝐶𝑍𝛿𝑟 𝛽
𝐿𝐼 𝑞𝑐̅
− − 𝐶 𝐿𝑝 𝐶 𝐿𝑞 𝐶 𝐿𝑟 = 𝐶 𝐿𝛼 𝐶 𝐿𝛽 𝐶𝐿𝛿𝑎 𝐶𝐿𝛿𝑒 𝐶𝐿𝛿𝑟 𝛿𝑎
𝑞̅ 𝑆𝑏̅ 2𝑉 𝛿𝑒
𝑀𝐼 𝐶𝑀𝑝 𝐶𝑀𝑞 𝐶𝑀𝑟 𝑟𝑏 ̅ 𝐶𝑀𝛼 𝐶𝑀𝛽 𝐶𝑀𝛿𝑎 𝐶𝑀𝛿𝑒 𝐶𝑀𝛿𝑟
{ 𝛿𝑟 }
𝑞̅ 𝑆𝑏̅ [ 𝐶 𝑁𝑝 𝐶 𝑁𝑞 𝐶𝑁𝑟 ] { 2𝑉 } [ 𝐶 𝑁𝛼 𝐶 𝑁𝛽 𝐶𝑁𝛿𝑎 𝐶𝑁𝛿𝑒 𝐶𝑁𝛿𝑟 ]
𝑁𝐼
{ 𝑞̅ 𝑆𝑐̅ }
This equation is of the form B = Ax, where:
𝑌𝐼
𝑞𝑆
̅ 𝐶𝑌𝑝 𝐶𝑌𝑞 𝐶𝑌𝑟 ̅
𝑍𝐼 𝑝𝑏
𝑞𝑆
̅ 𝐶𝑍𝑝 𝐶𝑍𝑞 𝐶𝑍𝑟 2𝑉
𝐿𝐼 𝑞𝑐̅
𝑩= − 𝑞̅𝑆𝑏̅
− 𝐶 𝐿𝑝 𝐶 𝐿𝑞 𝐶 𝐿𝑟 2𝑉
,
𝑀𝐼 𝐶𝑀𝑝 𝐶𝑀𝑞 𝐶𝑀𝑟 𝑟𝑏̅
𝑞̅𝑆𝑏̅ { }
𝑁𝐼 [ 𝐶 𝑁𝑝 𝐶 𝑁𝑞 𝐶𝑁𝑟 ] 2𝑉
{𝑞̅𝑆𝑐̅}
𝐶𝑌𝛼 𝐶𝑌𝛽 𝐶𝑌𝛿𝑎 𝐶𝑌𝛿𝑒 𝐶𝑌𝛿𝑟

𝐶𝑍𝛼 𝐶𝑍𝛽 𝐶𝑍𝛿𝑎 𝐶𝑍𝛿𝑒 𝐶𝑍𝛿𝑟
A = 𝐶 𝐿𝛼 𝐶 𝐿𝛽 𝐶𝐿𝛿𝑎 𝐶𝐿𝛿𝑒 𝐶𝐿𝛿𝑟 ,
𝐶𝑀𝛼 𝐶𝑀𝛽 𝐶𝑀𝛿𝑎 𝐶𝑀𝛿𝑒 𝐶𝑀𝛿𝑟
[ 𝐶 𝑁𝛼 𝐶 𝑁𝛽 𝐶𝑁𝛿𝑎 𝐶𝑁𝛿𝑒 𝐶𝑁𝛿𝑟 ]
and the state vector,
𝛼
𝛽
𝑥= 𝛿 𝑎
𝛿𝑒
{ 𝛿𝑟 }
33
Where B and A are known, x can be found. The vector, x is the solution to the quasi-steady
trim solution.
There are only three control surfaces defined in this solution, however, more, and in fact
infinite number of surfaces or external control devices can be defined for the aircraft. For
example, thrust vectoring may be another consideration. In such a case, the A matrix would
change from a 5x5 matrix to a 5x6 matrix, where the additional terms would be derivatives
of forces and moments with respect to the thrust vectoring device and some variable. It is
up to the analyst to determine appropriate development of this matrix and include these in
the model.
It should be noted that while there are 5 linear equations, they are in fact not independent
equations but the respective position of the trim variable has no bearing on the
corresponding degree of freedom. For example, because 𝛼 and side force, Y are the first
terms of the state and output vector, the side-force is not necessarily dependent on alpha
alone. In fact, alpha has zero influence on that degree of freedom (in the presented
example).
The solution to the quasi-steady trim is necessary for “re-building” the aircraft loading. As
mentioned, there are different phases of the program where different data is available.
When a six degree of freedom simulation is available, there will be inherent differences
between the data for the simulation and the data within the aeroelastic model, unless there
are corrections to correlate the models. If the models are well correlated, the time accurate
aircraft state data should be able to be represented by the trim solution in NASTRAN,
however if the data is uncorrelated, there will be deviations in aircraft control position
solutions to the trim state.
It is always necessary to have the aircraft in a “balanced” state for the structural finite
element model analysis. The aircraft balance satisfies the assumption of quasi-steady flight
loads. The term quasi-steady is used because there is a distinction made between steady
maneuvering and non-steady maneuvering. Steady maneuvering is assumed to have zero
34
rotational acceleration terms, meaning the aircraft is steady in its maneuver and its angular
rates are not changing. It may however have some linear acceleration terms. For example, a
steady 7g (Nz) pull-up maneuver is typical for fighters, where the aircraft is in an
accelerated state from the Nz term, but has no rotational acceleration. Typically, there is an
assumed associated pitch rate as a function of the load factor, which the Mil-Spec 8861b
does outline.
Non-steady maneuvers are those where the pitch acceleration terms are non-zero. For
example, a dynamic pitch maneuver where the pilot would, in a quick fashion pull back on
the longitudinal stick to impose an immediate nose up reaction and then later return the
stuck for an abrupt nose down reaction. These maneuvers are considered to be non-steady
in their abruptness, however for the purpose of static loads analysis, a single time point will
be considered. This single time point is typically, but not always, assumed to be point at
the highest rotational acceleration. That is, the most positive and most negative
acceleration for the initiation and termination points along the maneuver. The analysis of a
single time slice creates a quasi-steady assumption since the aircraft inertial forces and
moments are assumed to be balanced. This is the assumption on which the static aeroelastic
database will be derived.
Stability and Control Derivatives

As described above the stability and control derivatives are necessary to predict aircraft
trim. These can be generated by running Sol 144 in NASTRAN with TRIM decks defined.
The NASTRAN User Guide describes the trim input shown in Figure 10.
Figure 10 NASTRAN Trim Input Bulk Data Entries
The input deck used to generate the “baseline” derivatives is shown in Figure 11.
35
The trim input deck was developed in a separate bulk data file altogether from the model,
and was connected to the model by use of an “INCLUDE” statement which indicates to
NASTRAN that external files exist that should be included in the solution sequence
identified in the bulk data file submitted to the NASTRAN executable.
The use of an “INCLUDE” statement reduces the file sizes for multiple runs of common
tasks, and allows the user to quickly “swap” out properties, components, etc. In this case, it
was useful to be able to “swap” out trim solutions, so that a file for the model did not have
to be developed each time, and reduced the manual efforts to a minimum. The ECHO
request was used, so that the combined model would be read back out into the .f06 and
.pch files.
36
Figure 11 Baseline Trim Input
The trim is varied in Mach at 0.1 increments from Mach = 0.1 to Mach = 0.9. The choice
in the increment of Mach is based on the assumption of piece-wise linearity between
breakpoints. In other words, there is negligibly small variation between increments of 0.1
Mach. This is a requirement because the database will interpolate across the next higher
and the next lower Mach values to predict what the aerodynamic derivative is. The
dynamic pressure is the only other variation in the TRIM decks, such that the altitude stays
37
constant. Each of the corresponding Mach and qbar, or dynamic pressure in pounds-per-
square foot, values corresponds to an altitude of 0. Again, it is assumed that the variation of
altitude has no effect on the derivatives. Typically, this is a valid assumption and should
only be affected by external energy of the system such as propulsive effects. For example,
propeller wash is not simply scalable by dynamic pressure, but must be a function of
dynamic pressure and altitude.
The selection of breakpoints at delta Mach = 0.1 was based on engineering judgement and
validated through the results presented. If the predicted database results had significant
error against the NASTRAN run results, this would indicate that poor breakpoints in Mach
were chosen and the assumption of piece-wise linearity failed. In other words, a smaller
Mach would have needed to be chosen. In such a case where the predicted HM was
matched to many decimals places, perhaps too small of an increment was chosen.
Convergence studies could be conducted to find the optimum delta Mach (or any variable)
for the database.
38
The resulting output file is known as an “.f06” file. An example of the output is shown in
Figure 12.
Figure 12 .f06 Aerodynamic Derivative Ouptut
In order to calculate the estimated trim for a given Mach number, single variable
interpolation across Mach is accomplished via a method of multivariable interpolation.
Multivariable interpolation is necessary due to the fact that the tables are 5x5 (trim
variables) and 5x3 (set variables). The method used is a MATLAB function, ‘interpn’
which can be used to interpolate across multi-dimensional arrays. Because the input
breakpoints are dummy arrays, 1 through 5 and 1 through 3, are the same as the query
arrays, the interpolation is one-dimensional across multiple two dimensional tables. The
39
interpolation method used is ‘linear’. The dummy arrays could in theory be any value so
long as the query arrays are the same as the breakpoint arrays.
A hand calculation was performed to verify that indeed the interpolation scheme and the
database produced what was expected.
Table 1 Stability Derivative Interpolation Check
Test Interpolate, Mach = 0.25

ANGLEA SIDES ELEV AILERON RUDDER ROLL PITCH YAW
CY 0 0.1194595 0 -0.2243 -0.02414 -0.37865 0 0.065861
CZ 3.8221785 0 0.117279 0 0 0 7.126874 0
MX -1.143025 0 0.2168985 0 0 0 -4.22647 0
MY 0 -0.668085 0 0.066483 0.207653 0.157521 0 -0.55717
MZ 0 -0.13301 0 -0.00175 0.088882 -0.00057 0 -0.1785
INTERPOLATED VALUES FROM MATLAB CODE
ANGLEA SIDES ELEV AILERON RUDDER ROLL PITCH YAW
CY 0 0.1194595 0 -0.2243 -0.02414 -0.37865 0 0.065861
CZ 3.8221785 0 0.117279 0 0 0 7.126874 0
MX -1.143025 0 0.2168985 0 0 0 -4.22647 0
MY 0 -0.668085 0 0.066483 0.207653 0.157521 0 -0.55717
MZ 0 -0.13301 0 -0.00175 0.088882 -0.00057 0 -0.1785
Now that the stability derivatives are created, a trim estimation can be generated for a
given aircraft maneuvering state. In order to verify that the trim estimation works properly,
a simple check is performed against the existing NASTRAN Aeroelastic Trim solutions
from the database generation. Using the methods prescribed, the same trim solution that
was generated from the actual NASTRAN run should be able to be generated from the
database.
A summary of the stability and control derivatives are plotted in Appendix C

Flexible Stability Derivatives to compare the derivatives in each of the six degrees of
freedom (only five are actually used in the TRIM analysis). These figures are useful in
40
determining the effectiveness or contribution from each variable on the aircraft trim. For
example, one can compare the effects on aircraft rolling moment between the baseline
mean flow state, angle of attack, sideslip angle, roll rate, pitch rate, yaw rate, Ny, Nz, Pdot,
Qdot, Rdot, elevator control surface deflection, aileron control surface deflection and
rudder control surface deflection. These plots should be used to validate modeling from a
qualitative perspective. Intuitively, it should be expected that variables such as aileron
deflection, roll rate, yaw rate, rudder deflection largest “effectiveness”. In fact, it can be
seen that roll rate has the most significant effectiveness, followed by aileron deflection,
sideslip angle and yaw rate.
Appendix E
Unit Hinge Moment Database Per Mach and Variable shows individual degrees of
freedom with respect to each of the stability and control derivatives. Additionally, both
flexible and rigid aircraft derivatives are shown, to compare the resulting variation of
variable effectiveness affected by the aircraft flexibility, or the static aeroelastic
increments.
These static aeroelastic increments can be considered “flex to rigid” ratios. These are
described as the ratio of the flexible values over the rigid values. These are assumed to
follow a linear relationship, by the assumption of small aircraft deformations. Of course, if
the structure deformed such that linear assumptions are no longer valid, more intricate
methods, outside the scope of this paper, are necessary. However, for most practical
aircraft applications, these assumptions of linearity are valid.
Notice, for this analysis it is assumed that the only non-linear variability is that each of the
derivatives are non-linear in Mach. In other words, the assumption is that the stability and
control derivatives are linearly independent functions of each other, and dependent only on
Mach. This paper uses single baseline variable non-linearity for simplicity purposes,
however, in many aircraft applications, these derivatives are dependent with respect to each
other. For example, the aileron control surface effectiveness at a beta value of 10 degrees,
will be in fact, different than that of of the aileron control surface effectiveness at a beta
41
value of 0 degrees. In order to capture these coupled effects, multi-variable database
interpolation methods are needed. These multi-variable interpolation methods are simply
an extension of the presented method, whereas instead of simply interpolating across a
single variable independently, each of the derivatives are interpolated across each of the
other variables as well. It should also be noted that if this multi-variable interpolation is
required for the aircraft analysis, the order of interpolation does matter. The ordering and
methodology for interpolation should be studied, so that the methods of interpolation do
not skew inteded results from what the “true” database values are.
Now that the aircraft stability and control derivatives are defined, the database can be
generated. As stated, a single variable, Mach, is interpolated across. Visually, this is
expressed in Figure 13.
Figure 13 Derivative Database Representation
Given that the aircraft trim state can be predicted for any load factor, Mach, Altitude and
rotational accelerations, the use of the aircraft hinge moment derivatives can be used to
42
develop predicted aircraft hinge moments. A hinge moment database is developed in a
similar fashion that the stability and control database is developed.
The aircraft derivatives will give a trim solution, but still do not give insight into aircraft
loading. To develop this database, another matrix of static aeroelastic runs is required.
These runs are used to develop aircraft loading increments as a function of the aircraft
degrees of freedom, i.e. Y, Z, L, M and N.
Incremental Load Sets

In order to predict loading at any aircraft state, incremental aircraft loading will need to be
computed at various states. For the example presented in this paper, the following matrix in
Table 2 was generated.
Table 2 Incremental State Matrix
𝐏̇ 𝐐̇ 𝑹̇ P Q R
Mach Ny Nz [deg/s2] [deg/s2] [deg/s2] [deg/s] [deg/s] [deg/s]
Start Value 0.1 -2 -4 -100 -100 -50 -100 -100 -50
Increment 0.1 1 1 50 50 25 50 50 25
End Value 0.9 2 9 100 100 50 100 100 50
The values chosen for this paper are somewhat arbitrary and the fact that multiple values
(greater than two) were chosen is unnecessary here but would be necessary in the event of
non-linearity. The only necessary set of multiple values in this example is in Mach, which
as was stated previously, is assumed to have an effect on the aerodynamic derivatives.
In this exercise, the aircraft control surface hinge-moments are virtually instrumented with
“monitor points”. Monitor points are groups of elements and nodes in the finite element
model that reference a coordinate system and whose output values in the NASTRAN
43
solution are tabulated.This is an easy way to get output load data from each run, without
having to use some form of external load calculation.
An example of the output of the aerodynamic monitor point integrated loads is shown in
Figure 14. These loads are parsed and read into a database (both flexible and rigid
solutions were run). The values for hinge moments output in the NASTRAN .f06 file are in
units of [lb-ft2]. The hinge moments per condition are divided by the dynamic pressure of
that condition. The purpose of this is such that a condition at any altitude at a given Mach
can be computed. Otherwise, there would need to be a matrix presented above developed
for each altitude desired – or at the minimum a maximum and minimum altitude which can
be interpolated on. Both of these methods would be unnecessary since scaling by dynamic
pressure to vary altitude at a given Mach is just as accurate.
44
Figure 14 Example Monitor Point Loads
An example plot of the hinge moment vs. aircraft pitch rate can be seen in Figure 15.
Because of the linear assumptions made by use of the doublet-lattice solution, and as
previously stated, only two values of pitch rate are necessary to predict the aircraft hinge
moment based on that pitch rate. However, any non-linearities due to other aircraft effects
would determine how many breakpoints in the database are necessary. Having excessive
break points does not affect the solution and should not affect runtime.
45
Elevator Hinge Moment vs. Pitch Rate

5000
4000
3000
Hinge Moment, [ft-lb2]
2000
1000
0 Mach = 0.5
-150 -100 -50 -1000 0 50 100 150 Mach = 0.6
-2000
-3000
-4000
-5000
Pitch Rate [deg/s]
Figure 15 Elevator Hinge Moment vs. Pitch Rate, Mach 0.5, 0.6
46
Database Predicted Loads Results
Control Surface Hinge Moments

In order to validate the control surface hinge moment loads, test cases were run with
varying degrees of freedom in the solution. These cases are a combination of “realistic”
restricted degrees of freedom, such as 2 degree of freedom dynamic pitching conditions,
and higher degree of freedom conditions, where Ny, Nz, Pdot, Qdot and Rdot are included
degrees of freedom with non-zero values. These conditions are used to validate that the
prediction methods are not limited to reduced degrees of freedom and can be used for all
degrees of freedom of interest.
To prove the theory presented, a set of nominal Mil-Spec maneuvers and those which are
“unordinary” maneuvers were run to show that the solution is valid where all degrees of
freedom and trim variables are used.
The first set of conditions run were for standard “dynamic pitching maneuvers”. See
section “Military Specifications” for more information regarding this and other type of
standard Mil-Spec maneuvers. There are two subsets of the dynamic pitching maneuvers
presented in Appendix F
Proof of Concept Test Matrices. The abrupt pitch initiation assumes that the pilot is at
steady flight, and in this case at elevated or negative-g load factors, and an abrupt
maneuver to the longitudinal stick is generated such that no pitch rate occurs by aircraft
response and only aircraft acceleration acts on the vehicle in the form of Nz and Qdot. The
abrupt pitch termination assumes that the pilot has already initiated a pitching maneuver
and the aircraft has responded with some pitch rate. The pilot then puts longitudinal stick
in such that the aircraft accelerates in the opposite direction from the pitch rate developed.
47
The unit control surface hinge moments are outlined in Appendix E
Unit Hinge Moment Database Per Mach and Variable. All hinge moments reference the
RHS aileron or RHS elevator hinge moments.
Suppose the condition Ny = 0.45 [g] , Nz = 4.3 [g] , Pdot = 13 [deg/s2], Qdot = 49 [deg/ss],
Rdot = 4 [deg/s2], P = -24 [deg/s], Q = 3 [deg/s] and R = 2 [deg/s] at Mach = 0.3. The
resulting unit hinge moment coefficients are presented in Table 3.
The coefficients are then multiplied through by the values of the condition, wehre URDD2
and URDD3 are in units of g’s (non-dimensional) and the rotational acceleration and
rotational rate terms are in units of deg/ss and deg/s, respectively.
The theory for this echoes that which was developed for the static aeroelastic trim, in Static
Aeroelastic Trim Theory.
Table 3 Unit Hinge Moments at Mach = 0.3
Aileron Hinge Elevator Hinge Rudder Hinge

RATE
Mach Moment/(qbar* Moment/(qbar* Moment/(qbar*
VAR
URDD2) URDD2) URDD2)
0.3 URDD2 12.839 -20.7369 31.7138
0.3 URDD3 0.961 57.4937 0
0.3 PDOT 0.5027 -0.0659 0.1212
0.3 QDOT -0.0161 4.5065 0
0.3 RDOT -0.5489 1.7987 -5.9859
0.3 P 0.064 -0.0192 -0.0046
0.3 Q 0.0065 0.1533 0
0.3 R 0.0045 -0.0165 -0.0296
48
For example, the total aileron hinge moment is calculated by Equation 18.
Equation 18 𝐓𝐨𝐭𝐚𝐥 𝐇𝐢𝐧𝐠𝐞 𝐌𝐨𝐦𝐞𝐧𝐭 𝐂𝐚𝐥𝐜𝐮𝐥𝐚𝐭𝐢𝐨𝐧
𝟏𝟐. 𝟖𝟑𝟗 −𝟐𝟎. 𝟕𝟑𝟔𝟗 𝟑𝟏. 𝟕𝟏𝟑𝟖 𝑻 𝟎. 𝟒𝟓

𝟎. 𝟗𝟔𝟏 𝟓𝟕. 𝟒𝟗𝟑𝟕 𝟎 𝟒. 𝟑
𝟎. 𝟓𝟎𝟐𝟕 −𝟎. 𝟎𝟔𝟓𝟗 𝟎. 𝟏𝟐𝟏𝟐 𝟏𝟑
𝑨𝒊𝒍𝒆𝒓𝒐𝒏𝑯𝑴
−𝟎. 𝟎𝟏𝟔𝟏 𝟒. 𝟓𝟎𝟔𝟓 𝟎 𝟒𝟗
{𝑬𝒍𝒆𝒗𝒂𝒕𝒐𝒓𝑯𝑴 } = 𝟏𝟑𝟑. 𝟑𝟐𝟏𝟔
−𝟎. 𝟓𝟒𝟖𝟗 𝟏. 𝟕𝟗𝟖𝟕 −𝟓. 𝟗𝟖𝟓𝟗 𝟒
𝑹𝒖𝒅𝒅𝒆𝒓𝑯𝑴
𝟎. 𝟎𝟔𝟒 −𝟎. 𝟎𝟏𝟗𝟐 −𝟎. 𝟎𝟎𝟒𝟔 −𝟐𝟒
𝟎. 𝟎𝟎𝟔𝟓 𝟎. 𝟏𝟓𝟑𝟑 𝟎 𝟑
([ 𝟎. 𝟎𝟎𝟒𝟓 −𝟎. 𝟎𝟏𝟔𝟓 −𝟎. 𝟎𝟐𝟗𝟔] { 𝟐 })
The result gives hinge moments for aileron, elevator and rudder as 1593 [ft-lb], 62,119 [ft-
lb] and -1072 [ft-lb], respectively. These results can be checked qualitatively against the
maneuver representation. It is an elevated Nz maneuver with a dominant pitch acceleration
term, therefore, the highest hinge moments are expected to be the elevator hinge moment
term.
49
Utilizing the Database
Now that the aircraft database has been generated, the analyst has the first set of tools
available to generate design aircraft load cases (at least for control surface component
cases). The goal is to develop full aircraft, flexible and balanced loads to deliver to the
stress team.
Conceptual and Preliminary Design Usage

Suppose the aircraft is in the conceptual design phase, and it is up to the flight loads
analysts to come up with trends of aircraft hinge moment for a particular concept. At this
stage in the design phase, a rudimentary aircraft shape, number of control surface and
control surface sizing and generic propulsion data is known, however there is no
aerodynamic data (high order CFD or wind tunnel). The method presented thus far
provides aerodynamic data. The analyst can generate a simplified stiffness and mass model
with doublet-lattice aerodynamics to produce aircraft loading. It is assumed that there are
some general aircraft requirements but even these may not be set in stone yet, or they are
being developed based on these studies.
Neill and Whiting state, “In the design of aircraft, it is important to have an accurate
simulation of both the structural characteristic and the aerodynamic characteristics of the
vehicle to produce accurate trimmed loads. While MSC/NASTRAN has long had a static
aeroelastic analysis capability, it utilizes the embedded unsteady aerodynamic methods
(e.g., Doublet-Lattice) at zero reduced frequency.” They go on to state that “…these
methods are very useful for aeroelastic analyses at the conceptual and preliminary design
stage”. [3]
Indeed, these methods are very powerful tools for conceptual and preliminary design. An
example of how this method can be implemented is developing initial criteria for
50
conceptual and preliminary design sizing. In order for the structural design and analysis
teams to come up with conceptual or parametric sizing, an initial set of loads need to be
balanced aircraft load set must be developed. The question arises as to what load cases
should be delivered.
The details of these decisions depend on the nature of the aircraft development, and usually
specify particular wing stations of interest, fuselage stations, control surface hinge
moments, elevator stations, and rudder stations. In this paper, only the control surfaces had
monitor point loads, however, it is just as easy for any component of the vehicle to have
the same methodology applied to it. The same process would be followed, where aircraft
loads are normalized to the angular rate, angular acceleration or load factor associated to
that incremental run and in turn a database of coefficients which can be multiplied by the
dynamic pressure and variable value can be used to quickly identify critical component
loads.
In this example, suppose one of the components to critique is the elevator hinge-moment to
set preliminary actuator sizing requirements. A survey could be run, much like the one
presented in Appendix F
Proof of Concept Test Matrices, however expanded for incremental pitch rates, associated
with some pitch accelerations at normal vertical load factors. Such survey increments are
presented in Table 4.
Table 4 Abrupt Pitch Initial Sizing Survey
Pdot Qdot Rdot P Q R

Ny Nz [deg/s ] [deg/s ] [deg/s2]
2 2
[deg/s] [deg/s] [deg/s]
Min Value 0 -2 0 -200 0 0 -100 0
Increment 0 1 0 25 0 0 25 0
Max Value 0 4 0 200 0 0 100 0
An expanded test matrix results in a combination of these values to 9639 combinations of

cases over increments of Mach from 0.1 to 0.9, following suit with the database break
51
points, though, any Mach value between the bounds of the database is valid. The resulting
array of elevator control surface hinge moments are sorted from most positive to most
negative (the sign of the hinge moment is important, especially for actuator sizing, since
depending on the control actuation device, it will be actuator extend upper/lower or
actuator extend and retract loads). All results of the survey are presented in Figure 16
Dynamic Pitch Initial Actuator Sizing Survey - Elevator Hinge Moment
The tabulated results of the critical conditions are provided in Table 5 for maximum
positive hinge moments and in Table 6 for maximum negative hinge moments.
Table 5 Maximum Positive Elevator Hinge Moment Results
Elevator
Case Pdot Qdot Rdot P Q R HM
ID Mach Ny Nz [deg/s2] [deg/s2] [deg/s2] [deg/s] [deg/s] [deg/s] [ft lb]
9639 0.9 0 4 0 200 0 0 100 0 164,402
9638 0.9 0 4 0 200 0 0 75 0 162,195
8568 0.8 0 4 0 200 0 0 100 0 161,029
9637 0.9 0 4 0 200 0 0 50 0 159,988
8567 0.8 0 4 0 200 0 0 75 0 159,312
Table 6 Maximum Negative Elevator Hinge Moment Results
Elevator
Case Pdot Qdot Rdot P Q R HM
ID Mach Ny Nz [deg/s2] [deg/s2] [deg/s2] [deg/s] [deg/s] [deg/s] [ft lb]
8569 0.9 0 -2 0 -200 0 0 -100 0 -147,868
8570 0.9 0 -2 0 -200 0 0 -75 0 -145,662
7498 0.8 0 -2 0 -200 0 0 -100 0 -144,896
8571 0.9 0 -2 0 -200 0 0 -50 0 -143,455
7499 0.8 0 -2 0 -200 0 0 -75 0 -143,179
52
Elevator Hinge Moment

200000
150000
100000
Hinge Moment [ft lb]
50000
-50000
-100000
-150000
-200000
The trend in Figure 16 Dynamic Pitch Initial Actuator Sizing Survey - Elevator Hinge
Moment that the results are not simply linear, however the tabulated results do show, that
the combination of the max positive pitch acceleration, with the max positive pitch rate at
the maximum positive load factor does cause the most positive hinge moment. A positive
hinge moment, as can be seen in the tabulated results in Appendix F
Proof of Concept Test Matrices, is from a trailing edge up deflection causing aircraft nose
up moment.
A figure is presented to show the trends of Mach on the elevator hinge moments. Because
the same conditions were run at each Mach, a direct case by case comparison can be done.
Presented in Figure 17 is a summary of the same data presented in Figure 16, except that
the data is sorted and colored by Mach. A trend of the peaks, per Mach, can be seen with
53
Figure 17 Dynamic Pitch Initial Actuator Sizing Survey - Elevator Hinge Moment Sorted by
Mach
54
the dashed black lines, where the dashed black line with circles trends the maximum peak
positive elevator hinge moments and the dashed black line with triangles trends the
maximum peak negative hinge moments. Of course, again, piecewise linearity between the
variations of Mach is the only non-linear assumed term in this survey.
There is, however, one important piece of information that has been neglected in the
assumptions of this survey, and this the fact that the aircraft control authority limits have
been neglected. The control derivatives can be extrapolated to solve any aircraft state that
is input, however, the question of whether the results are valid needs to be analyzed.
Therefore, the trim estimator must be used to predict control surface deflections.
Since the aircraft control derivatives were developed using doublet-lattice methods, there is
some inherent factor that should be maintained due to the primitive nature in the solution
methods. Therefore, when the down-select of critical hinge moments limited by aircraft
control authority is taken into account, some margin should be assumed for limits of the
control surface.
Suppose the aircraft was limited to 45 degrees trailing edge up and 45 degrees trailing edge
down. Additionally, since alpha is assumed to have no stall limits, stall in alpha is assumed
at +20 degrees and -15 degrees.
55
Table 7 Elevator Hinge Moment in Alpha and Surface Deflection Limit
Elevator Hinge Moments in Limit

200000
150000
100000
Hinge Moment [ft-lb]
50000
-50000
-100000
-150000
-200000
The resulting maximum positive and maximum negative hinge moments obtained from the
survey satisfy the “realistic” stall and surface travel limits. There were a reduction of about
20% of the cases, of which occurred at the lowest qbar and highest accelerations. This
validates logical reasoning because the aircraft capability is limited at lower airspeeds.
This type of survey can be useful in predicted aircraft limits during preliminary and
conceptual design phases. Perhaps the vehicle is maneuverable limited by airframe
structural design limits. As the aircraft requirements are being developed, there may be a
maximum allowable design load for some portion of the aircraft, which shall not be
exceeded. If this is the case, it would be up to the flight controls team to rectify this
requirement by setting aircraft control limits on the pilot, or reducing aircraft capability
from a controls standpoint.
56
Of course, this is dependent on the type of aircraft and its mission purpose. Perhaps for a
Cirrus-SR22 weight optimization may be key, while maneuverability can be sacrificed.
The highest weighted objective to some aircraft design is minimum weight and a lower
weighted objective is aircraft maneuverability.
In the case of fighter design, such as the example, aircraft maneuverability may be the
highest objective requirement to meet the customer’s expectations and therefore the
airframe structure is a fallout of that. Typically, though, as aircraft weight increases,
aircraft maneuverability tends to decrease and thus performance becomes a function of
itself.
In order to reduce the opportunity of getting caught in a circular requirement loop of

maneuverability and structural rigidity, clear requirements should be set early in the
developmental phases of the design, so that both teams can work together to come up with
trade space studies weighing out impacts of aircraft maneuvering capability and structural
strength and weight as a fallout, or vice-versa.
This type of survey, as mentioned, can be expanded to any component on the vehicle. It is
typical that the stress team will want to analyze the vehicle for the critical load conditions.
Suppose that one of the critical locations surveyed was the wing root, which stress wants to
analyze on the full vehicle for critical internal loads. That condition can be generated by
running a survey with monitor point loads integrated at that location to find a critical trim
associated to that load.
While it is of interest to the structural analysis team what the wing root bending and shear
are, for them to analyze that loading on the FEM, they will need distributed net loads.
57
These nets loads are the distribution of aerodynamic loads and inertial loads and select
points on their FEM. While the methods to convert aeroelastic model loads to the FEM
model (assuming they are different), the process to develop critical aeroelastic model loads
is discussed.
In running the database generation, the aerodynamic pressure associated with the flexible
solution is captured into a similar database. This database is an array of pressure
coefficients at that condition. In the same method for creating derivatives of total monitor
point loads at each of the flight conditions, incremented on aircraft state values, the same
can be done for the pressure coefficients. If there is an nx1 array where n is the number of
grid points on the aeroelastic model, that array can be divided by the increment, which can
then be later post-multiplied through to get the incremental aerodynamic pressure for that
condition.
The inertial forces at discreet points where lumped masses represent the aircraft structure
and subsystems, can be calculated using the same equations presented in Static
Aeroelastic Trim Theory, using Euler rigid body mechanics.
The combination of these can be used to develop “distributed point loads” on the
aeroelastic model. The method for transferring the distributed point loads on the aeroelastic
model to the finite element model for internal loads assessment is very complex and there
are many methods to accomplish this. One method is presented by Samareh. [9]
58
Detail Design and Flight Test
The process for generating a critical loads survey should vary only slightly between the
various design phases. In the previous section, detailed discussion the development of
critical elevator hinge moments were laid out. The assumptions were that no wind tunnel or
CFD data existed and that the loads analyst only had a basic structural beam model
representing aircraft rigidity and a basic doublet-lattice model for aircraft aerodynamics.
During the detail design phase, it can be assumed that high order CFD, such as Euler or
Navier-Stokes solutions are used along with wind tunnel test data. With this assumption,
the use of the doublet-lattice solution is unnecessary, however the methods of developing
derivatives can still be consistent with the previously described methodology.
There is a method of interfacing external, high order aerodynamics into NASTRAN for
aeroelastic analysis presented by Whiting and Neill. [3] The method describes replacing
the doublet-lattice solution with that of the high order CFD. With this, there are many
assumptions which can be eliminated and details which can be added.
With the use of high order CFD and wind tunnel data, accurate total aircraft coefficients
can be captured. The coefficients should be consistent across disciplines, so that the flight
controls team is working to the same data the loads team is analyzing to. In the conceptual
and preliminary design phases there may be discrepancies, as expected. There may,
however still be discrepancies even at the later detail design phases of design. The flight
controls team may have simulations with rigid aircraft assumptions, while the loads team
will have aeroelastic effects in their analysis. This should be noted if time history data is
used in calculating time accurate aircraft loads.
Modeling fidelity can increase with the use of CFD by capturing wing-body interaction and
complex flow such as effects due to inlet and exhaust effects. The doublet-lattice method is
restricted to neglect effects due to propulsion, unless a direct matrix input, or DMI is used
to supersede the doublet-lattice pressures. Of course, this is not possible with CFD
solutions defining those inputs.
59
With the usage of high order CFD and wind tunnel data, non-linear effects can be captured.
To utilize this data a multi-dimensional non-linear database would need to be generated.
The format of the database is consistent with that of the single-dimension non-linear
database which was generated in the example of this paper. In the sample presented, the
non-linear independent variable was Mach with dependent variables of alpha, beta, roll
rate, pitch rate, yaw rate, elevator deflection, aileron deflection and rudder deflection. The
variation is the extension to n-dimensions, where n is the number of independent variables.
The database becomes a multi-dimensional array of size m x n1 x n2 x n3 x … nk where k
is the number of dependent variables. The same method of interpolation within MATLAB
can be utilized.
Additionally, increased fidelity on time accurate aircraft simulation, or six degree of

freedom simulation becomes available as the aircraft design matures. Utilizing the flight
control’s simulation increases the reliability of the data, due to modeling of the intended
aircraft flight control system involving surface scheduling, accurate gains and other
parameters that a basic simulation would not capture.
In the flight test phase, it is critical to be able to produce and assess loads quickly and
accurately. Having a well correlated database to the instrumented flight test vehicle is
critical during the envelope expansion phase of flight test. There is limited literature on
instrumented loads flight testing. The extent of the discussion on flight test methods will be
limited in this paper as well. The methods for which a flight test operation is conducted can
vary greatly, from vehicle to vehicle, however, the intent is that the loads team has
predicted critical loading on the vehicle such that the loads experienced during flight test
do not exceed those which were used to design the vehicle.
In an effort to rapidly certify the aircraft through different flight speeds and altitudes,
having a derivative database which is well correlated to flight test data allows the analysts
to predict loads and “next” test points. If the analysts can accurately predict loads during
the envelope expansion phase, certain points may be able to be skipped, showing them as
60
confidently benign maneuver points, and show the aircraft “good” at the predicted critical
points. Good correlation also gives confidence through the rest of the envelope not tested,
since it is not reasonable to fly at every single point in the envelope, performing every
maneuver. [11]
Additionally, with the usage of the derivative database, the loads analysts can predict loads
real-time to assist in “knock-it-off” calls. These calls are to immediate stop the
maneuvering and have the pilot resume to level and safe flight attitudes and airspeeds.
There are many methods for correcting the database to correlate to flight test data points,
the details of which are up the analyst to prove, however, it is very efficient to have
methods of reducing manual efforts. With this, as the flight test is conducted, the database
can be automatically corrected. As the database is corrected for previous maneuvers, this
allows the analyst to predict loads for the next maneuver. With good correlation, the
program has confidence to be more aggressive in removing test points in the program and
thus saving money and time for the flight test program.
61
References
[1] Lomax, Ted. (1996) Structural Loads Analysis for Commercial Transport Aircraft:
Theory and Practice. American Institute of Aeronautics and Astronautics.
[2] MIL-A-8861B (1986), Military Specifications: Airplane Strength and Rigidity Flight
Loads
[3] Neill, Douglas J. & Whiting, Brent (1997), Interfacing External, High Order
Aerodynamics into MSC/NASTRAN for Aeroelastic Analysis
[4] Neufeld, D. (2010). Multidisciplinary aircraft conceptual design optimization

considering fidelity uncertainties (Order No. NR71403). Barber, B. C., & Author,
E. (2014).
[5] Rodden, W., Wilson, C., Herting, D., Bellinger E. and MacNeal, R. (1984) STATIC
AEROELASTIC ADDITION TO MSC/NASTRAN, The MacNeal-Schwendler
Corporation, Los Angeles, California 90041
[6] Rodden, William P., Taylor, Paul F., and McIntosh, Sameul C., Journal of Aircraft,
AIAA, Vol. 35, No. 5 (1998), “Further Refinement of the Subsonic Doublet-
Lattice Method”.
[7] Roskam, Jan. (1990) Airplane Design. Roskam Aviation and Engineering Corp.
[8] RTO Applied Vehicle Technology Panel, Prepared for the North Atlantic Treaty
Organization (NATO) (2002), Design Loads for Future Aircraft, RTO-TR-045
AC/323(AVT-024)TP/30
[9] Samareh, Jamshid A. (2007), NASA Langley Research Center, Hampton, VA,
“Discrete Data Transfer Technique for Fluid-Structure Interaction”.
[10] Sunil Kukreja (2007). "Aeroelastic Model Structure Computation for Envelope
Expansion", 48th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics,
and Materials Conference, Structures, Structural Dynamics, and Materials and Co-
located Conferences.
[11] Wolf, J, Sansone, A. (2002) "The U.S. Air Force Academy's flight test course -
Preparing tomorrow's flight testers", 40th AIAA Aerospace Sciences Meeting &
Exhibit, Aerospace Sciences Meetings
[12] Liu, Tianshu , Guille, M., Sullivan, M. (1999) "Accuracy of pressure sensitive paint",
30th Fluid Dynamics Conference, Fluid Dynamics and Co-located Conferences
62
Appendix A
Tabulated Rigid Stability Derivatives at Incremental
Mach Values
63
64
Appendix B
Tabulated Flexible Stability Derivatives at
Incremental Mach Values
65
66
Appendix C
Flexible Stability Derivatives
67
68
69
70
Appendix D
Flexible vs. Rigid Stability Derivatives
71
72
73
74
75
76
Appendix E
Unit Hinge Moment Database Per Mach and Variable
Elevator Hinge Rudder Hinge

RATE Aileron Hinge Moment/(qbar* Moment/(qbar*
Mach VAR Moment/(qbar* URDD2) URDD2) URDD2)
0.1 URDD2 113.8882 -187.5346 280.9917
0.2 URDD2 28.6236 -46.8023 70.6559
0.3 URDD2 12.8390 -20.7369 31.7138
0.4 URDD2 7.3203 -11.6080 18.0951
0.5 URDD2 4.7734 -7.3751 11.8064
0.6 URDD2 3.4001 -5.0662 8.4088
0.7 URDD2 2.5867 -3.6610 6.3841
0.8 URDD2 2.0818 -2.7296 5.1045
0.9 URDD2 1.7805 -2.0599 4.2906
Elevator Hinge Rudder Hinge
RATE Aileron Hinge Moment/(qbar* Moment/(qbar*
Mach VAR Moment/(qbar* URDD3) URDD3) URDD3)
0.1 URDD3 8.9319 514.5834 0.0000
0.2 URDD3 2.2069 128.9061 0.0000
0.3 URDD3 0.9610 57.4937 0.0000
0.4 URDD3 0.5241 32.5052 0.0000
0.5 URDD3 0.3209 20.9497 0.0000
0.6 URDD3 0.2091 14.6867 0.0000
0.7 URDD3 0.1398 10.9262 0.0000
0.8 URDD3 0.0915 8.5083 0.0000
0.9 URDD3 0.0514 6.8894 0.0000
RATE Aileron Hinge Elevator Hinge Rudder Hinge
Mach VAR Moment/(qbar* PDOT) Moment/(qbar* PDOT) Moment/(qbar* PDOT)
0.1 PDOT 4.4725 -0.5966 1.0783
0.2 PDOT 1.1229 -0.1489 0.2707
0.3 PDOT 0.5027 -0.0659 0.1212
0.4 PDOT 0.2859 -0.0369 0.0689
0.5 PDOT 0.1858 -0.0233 0.0447
0.6 PDOT 0.1317 -0.0159 0.0316
0.7 PDOT 0.0996 -0.0114 0.0237
0.8 PDOT 0.0795 -0.0082 0.0186
0.9 PDOT 0.0670 -0.0058 0.0151
77
Mach VAR Moment/(qbar* QDOT) Moment/(qbar* QDOT) Moment/(qbar* QDOT)
0.1 QDOT -0.1444 40.4923 0.0000
0.2 QDOT -0.0361 10.1293 0.0000
0.3 QDOT -0.0161 4.5065 0.0000
0.4 QDOT -0.0091 2.5388 0.0000
0.5 QDOT -0.0060 1.6282 0.0000
0.6 QDOT -0.0043 1.1338 0.0000
0.7 QDOT -0.0034 0.8359 0.0000
0.8 QDOT -0.0030 0.6429 0.0000
0.9 QDOT -0.0032 0.5105 0.0000
Mach VAR Moment/(qbar* RDOT) Moment/(qbar* RDOT) Moment(/qbar* RDOT)
0.1 RDOT -4.8396 16.3457 -53.2556
0.2 RDOT -1.2192 4.0721 -13.3706
0.3 RDOT -0.5489 1.7987 -5.9859
0.4 RDOT -0.3145 1.0024 -3.4029
0.5 RDOT -0.2061 0.6329 -2.2097
0.6 RDOT -0.1474 0.4311 -1.5645
0.7 RDOT -0.1124 0.3079 -1.1795
0.8 RDOT -0.0903 0.2259 -0.9354
0.9 RDOT -0.0764 0.1662 -0.7791
Mach VAR Moment/(qbar* P) Moment/(qbar* P) Moment/(qbar* P)
0.1 P 0.1861 -0.0574 -0.0127
0.2 P 0.0941 -0.0287 -0.0066
0.3 P 0.0640 -0.0192 -0.0046
0.4 P 0.0494 -0.0144 -0.0037
0.5 P 0.0411 -0.0116 -0.0032
0.6 P 0.0361 -0.0097 -0.0030
0.7 P 0.0331 -0.0083 -0.0030
0.8 P 0.0319 -0.0072 -0.0031
0.9 P 0.0325 -0.0063 -0.0036
78

Mach VAR Moment/(qbar* Q) Moment/(qbar* Q) Moment/(qbar* Q)
0.1 Q 0.0187 0.4498 0.0000
0.2 Q 0.0095 0.2267 0.0000
0.3 Q 0.0065 0.1533 0.0000
0.4 Q 0.0051 0.1174 0.0000
0.5 Q 0.0043 0.0968 0.0000
0.6 Q 0.0039 0.0840 0.0000
0.7 Q 0.0037 0.0762 0.0000
0.8 Q 0.0037 0.0725 0.0000
0.9 Q 0.0040 0.0736 0.0000
Mach VAR Moment/(qbar* R) Moment/(qbar* R) Moment/(qbar* R)
0.1 R 0.0130 -0.0485 -0.0871
0.2 R 0.0066 -0.0244 -0.0438
0.3 R 0.0045 -0.0165 -0.0296
0.4 R 0.0035 -0.0126 -0.0226
0.5 R 0.0030 -0.0104 -0.0185
0.6 R 0.0027 -0.0090 -0.0160
0.7 R 0.0025 -0.0080 -0.0143
0.8 R 0.0025 -0.0074 -0.0133
0.9 R 0.0027 -0.0070 -0.0129
79
Appendix F
Proof of Concept Test Matrices
80
81
82
83
84
85
86
87
88
89
Appendix G
Baseline Matrix Generation
90
Appendix H
Expanded Matrix Generation
91
Appendix I
Trim Estimator
92
Appendix J
Unit Loads Derivative Database Generator
93
Appendix K
Write Trim Include
94
Appendix L
Non-Dimensionalize Anglular Values
95
Appendix M
HA-144F Model Data
INIT MASTER(S)
NASTRAN SYSTEM(442)=-1,SYSTEM(319)=1
ID EDS,Femap
SOL AESTAT
TIME 10
CEND
TITLE = BASELINE MACH SOLUTION - THESIS MODEL
SUBTITLE = BASELINE RUN
LABEL = HA144F - NASTRAN MODEL
ECHO = NONE
DISPLACEMENT = ALL
FORCE = ALL
STRESS = ALL
AEROF = All
APRES = All
SPC = 1
MPC = 10
ECHO = BOTH
DISPLACEMENT = ALL
FORCE = ALL
STRESS = ALL
AEROF = All
APRES = All
SPC = 1
MPC = 10
OUTPUT(PLOT)
PLOTTER = NASTRAN
SET 1 = ALL
FIND SCALE, ORIGIN 1,SET 1
PLOT SET 1
PLOT STATIC DEFORMATION 0, ORIGIN 1, SET 1, OUTLINE
BEGIN BULK
$
PARAM,PRGPST,YES
PARAM,AUTOSPC,YES
PARAM,GRDPNT,0
CORD2C 1 0 0. 0. 0. 0. 0. 1.+FEMAPC1
+FEMAPC1 1. 0. 1.
CORD2S 2 0 0. 0. 0. 0. 0. 1.+FEMAPC2
+FEMAPC2 1. 0. 1.
$ Femap with NX NASTRAN Coordinate System 3 : Rectangular Coordinate System
CORD2R 3 0 24.2265 10. 0.24.311529.829954.9817617+
+ 25.09355 10.4981.0111873
CORD2R 4 0 12.5 0. 0. 12.5 0. 1.+
+ 13.5 0. 0.
CORD2R 5 0 30. 0. 0. 30. 0. 1.+
+ 30.86603.4999967 0.
CORD2R 20 0 30. 0. 0. 30. 0. 1.+
+ 30.86603-.499997 0.
CORD2R 30 0 24.2265 -10. 0.24.31152-9.82995 .981761+
+ 25.09355-10.4981.0111894
CORD2R 90 900 5. 0. 0. 5. 0. 1.+
+ 6. 0. 0.
CORD2R 100 0 15. 0. 0. 15. 0. -1.+
+ 14. 0. 0.
CORD2R 110 0 26.7265 10. 0. 26.7265 10. -1.+
+ 27.59253 10.5 0.
CORD2R 210 0 26.7265 -10. 0. 26.7265 -10. 1.+
+ 27.59253 -10.5 0.
CORD2R 300 0 30. 0. 0. 30. 1. 0.+
+ 29.13397 0..4999964
CORD2R 301 0 32.5 0. 0. 32.5 -1. 0.+
96
+ 31.63397 0..4999964
CORD2R 450 0 24.2265 10. 0. 24.2265 11. 0.+
+ 25.2265 10. 0.
CORD2R 460 0 24.2265 -10. 0. 24.2265 -9. 0.+
+ 25.2265 -10. 0.
CORD2R 900 0 10. 0. 1.5 10. 0. 2.5+
+ 11. 0. 1.5
CORD2R 100000 017.27607 0. .1552817.27607 0. 1.15528+
+ 18.27607 0. .15528
PARAM,AUNITS,.031081
AESTAT 1 ANGLEA
AESTAT 2 SIDES
AESTAT 3 ROLL
AESTAT 4 PITCH
AESTAT 5 YAW
AESTAT 6 URDD2
AESTAT 7 URDD3
AESTAT 8 URDD4
AESTAT 9 URDD5
AESTAT 10 URDD6
AEROS 100000 100000 10. 40. 400. 0 0
$ Femap with NX NASTRAN Aero Control Surface 505 : ELEV
AESURF 505 ELEV 90 1 90 2 LDW+
+
$ Femap with NX NASTRAN Aero Control Surface 517 : AILERON
AESURF 517 AILERON 110 3 210 4 LDW+
+
$ Femap with NX NASTRAN Aero Control Surface 518 : RUDDER
AESURF 518 RUDDER 301 5 LDW+
+
AELIST 1 1000 1001 1002 1003 1004 1005 1006+
+ 1007
AELIST 2 2000 2001 2002 2003 2004 2005 2006+
+ 2007
AELIST 3 1119 1123 1127 1131
AELIST 4 2103 2107 2111 2115
AELIST 5 3103 3107 3111
$ Femap with NX NASTRAN Constraint Set 1 : NASTRAN SPC 1
SPC1 1 1 90
$ Femap with NX NASTRAN Constraint Set 2 : NASTRAN SUPORT
SUPORT 90 23456
$ Femap with NX NASTRAN Constraint Set 3 : NASTRAN OMIT
OMIT 110 4
OMIT 120 4
OMIT 210 4
OMIT 220 4
OMIT 310 4
$ Femap with NX NASTRAN Constraint Set 10 : NASTRAN MPC 10
MPC 10 89 1 1. 890 1 -1.
MPC 10 89 2 1. 890 2 -1.
MPC 10 89 3 1. 890 3 -1.
MPC 10 89 4 1. 890 4 -1.
MPC 10 89 5 1. 890 5 -1.
MPC 10 89 6 1. 890 6 -1.
MPC 10 91 1 1. 910 1 -1.
MPC 10 91 2 1. 910 2 -1.
MPC 10 91 3 1. 910 3 -1.
MPC 10 91 4 1. 910 4 -1.
MPC 10 91 5 1. 910 5 -1.
MPC 10 91 6 1. 910 6 -1.
$ Femap with NX NASTRAN Property 100 : BAR Property
PBAR 100 1 4. .347222 .3 1. 0. +
+ 1. 1. 1. -1. -1. 1. -1. -1.+
+ 0.
$ Femap with NX NASTRAN Property 101 : BAR Property
PBAR 101 1 1.5 .173611 2. .462963 0. +
+ .5 3. .5 -3. -.5 3. -.5 -3.+
+ 0.
$ Femap with NX NASTRAN Material 1 : ISOTROPIC Material
MAT1 1 1.44+9 5.4+8 0. 0. 0.
GRID 88 900 5. -5. 0. 0
GRID 89 900 5. -2. 0. 0
GRID 90 0 15. 0. 0. 0
GRID 91 900 5. 2. 0. 0
GRID 92 900 5. 5. 0. 0
GRID 97 0 0. 0. 0. 0
GRID 98 0 10. 0. 0. 0
GRID 99 0 20. 0. 0. 0
GRID 100 0 30. 0. 0. 0
97
GRID 110 027.11325 5. 0. 0
GRID 111 024.61325 5. 0. 0
GRID 112 029.61325 5. 0. 0
GRID 115 0 24.2265 10. 0. 0
GRID 120 021.33975 15. 1. 0
GRID 121 018.83975 15. 1. 0
GRID 122 023.83975 15. 1. 0
GRID 150 0 19.2265 10. -1.5 0
GRID 151 0 24.2265 10. -1.5 0
GRID 210 027.11325 -5. 0. 0
GRID 211 024.61325 -5. 0. 0
GRID 212 029.61325 -5. 0. 0
GRID 215 0 24.2265 -10. 0. 0
GRID 220 021.33975 -15. 1. 0
GRID 221 018.83975 -15. 1. 0
GRID 222 023.83975 -15. 1. 0
GRID 250 0 19.2265 -10. -1.5 0
GRID 251 0 24.2265 -10. -1.5 0
GRID 310 032.88675 0. 5. 0
GRID 311 030.38675 0. 5. 0
GRID 312 035.38675 0. 5. 0
GRID 890 0 15. -2. 0. 0
GRID 910 0 15. 2. 0. 0
CBAR 89 101 88 89 0. 0. 1.
CBAR 90 101 890 90 0. 0. 1.
CBAR 91 101 90 910 0. 0. 1.
CBAR 92 101 91 92 0. 0. 1.
CONM2 97 97 0 93.243 0. 0. 0.
CONM2 98 98 0 93.243 0. 0. 0.
CONM2 99 99 0 93.243 0. 0. 0.
CBAR 100 100 90 99 0. 0. 1.
CBAR 101 100 97 98 0. 0. 1.
CBAR 102 100 98 90 0. 0. 1.
CBAR 103 100 99 100 0. 0. 1.
CBAR 110 101 100 110 0. 0. 1.
RBE2 111 110 123456 111
RBE2 112 110 123456 112
CBAR 115 101 110 115 0. 0. 1.
CBAR 120 101 115 120 0. 0. 1.
RBE2 121 120 123456 121
RBE2 122 120 123456 122
CBAR 150 101 150 151 0. 0. 1.
CBAR 151 101 151 115 1. 0. 0.
CBAR 210 101 100 210 0. 0. 1.
RBE2 211 210 123456 211
RBE2 212 210 123456 212
CBAR 215 101 210 215 0. 0. 1.
CBAR 220 101 215 220 0. 0. 1.
RBE2 221 220 123456 221
RBE2 222 220 123456 222
CBAR 250 101 250 251 0. 0. 1.
CBAR 251 101 251 215 1. 0. 0.
CBAR 310 101 100 310 0. 0. 1.
RBE2 311 310 123456 311
RBE2 312 310 123456 312
CONM2 313 100 0 93.243 0. 0. 0.
CONM2 314 111 0 18.6486 0. 0. 0.
CONM2 315 112 0 12.4324 0. 0. 0.
CONM2 316 121 0 18.6486 0. 0. 0.
CONM2 317 122 0 12.4324 0. 0. 0.
CONM2 318 211 0 18.6486 0. 0. 0.
CONM2 319 212 0 12.4324 0. 0. 0.
CONM2 320 221 0 18.6486 0. 0. 0.
CONM2 321 222 0 12.4324 0. 0. 0.
CONM2 322 311 0 1.86486 0. 0. 0.
CONM2 323 312 0 1.24324 0. 0. 0.
$ Femap with NX NASTRAN Aero Property 1000 : Aero Property
PAERO1 1000 4000 4510 4610
PAERO2 4020 ZY 2.5 1. 1 2 +
+ 1 8
AEFACT 1 0. 1.111 1.778 2. 2. 2. 2.5+
+ 2.5 2.5
AEFACT 2 45. 135. 225. 315.
PAERO2 4520 ZY .5 1. 3 4 +
+ 1 4
AEFACT 3 0. .5 .5 .5 0.
AEFACT 4 45. 135. 225. 315.
AEFACT 5 0. .16667 1.
$ Femap with NX NASTRAN Aero Panel 1000 : Aero Panel
CAERO1 1000 1000 900 4 5 1+
+ 0. 2. 0. 10. 0. 5. 0. 10.
98
CAERO1 1104 1000 0 3 4 1+
+ 23.55662 2.5 0. 10. 19.2265 10. 0. 10.
CAERO1 1116 1000 0 4 4 1+
+ 19.2265 10. 0. 10.13.45299 20. 2. 10.
AEFACT 6 0. .83333 1.
CAERO1 2000 1000 900 4 6 1+
+ 0. -5. 0. 10. 0. -2. 0. 10.
CAERO1 2100 1000 0 4 4 1+
+ 13.45299 -20. 2. 10. 19.2265 -10. 0. 10.
CAERO1 2116 1000 0 3 4 1+
+ 19.2265 -10. 0. 10.23.55662 -2.5 0. 10.
CAERO1 3100 1000 0 3 4 1+
+ 30.7735 0. 10. 10.26.44334 0. 2.5 10.
CAERO1 3500 1000 0 1 2 1+
+ 21.7265 10. 0. 5. 21.7265 10. -1. 5.
CAERO1 3600 1000 0 1 2 1+
+ 21.7265 -10. 0. 5. 21.7265 -10. -1. 5.
CAERO2 4000 4020 0 8 8 1+
+ -5. 0. 0. 40.
CAERO2 4510 4520 0 4 4 1+
+ 19.2265 10. -1.5 10.
CAERO2 4610 4520 0 4 4 1+
+ 19.2265 -10. -1.5 10.
SET1 1000 88 89 91 92
$ Femap with NX NASTRAN Aero Spline 1502 : Aero Spline
SPLINE2 1502 1000 1000 1007 1000 0. 1. 90+
+ 0. 0. BOTH
SET1 1101 100 111 112 115
SPLINE2 1602 1104 1104 1115 1101 0. 1. 5+
+ -1. -1. BOTH
SET1 1102 115 121 122
SPLINE2 1603 1116 1116 1131 1102 0. 1. 3+
+ -1. -1. BOTH
SPLINE2 2501 2000 2000 2007 1000 0. 1. 90+
+ 0. 0. BOTH
SET1 2102 215 221 222
SPLINE2 2601 2100 2100 2115 2102 0. 1. 30+
+ -1. -1. BOTH
SET1 2101 100 211 212 215
SPLINE2 2602 2116 2116 2127 2101 0. 1. 20+
+ -1. -1. BOTH
SET1 3100 99 100 311 312
SPLINE2 3100 3100 3100 3111 3100 0. 1. 300+
+ -1. -1. BOTH
SET1 4001 97 98 99 100
SPLINE2 4000 4000 4000 4007 4001 0. 1. +
+ 0. -1. BOTH
SET1 4521 115 150 151
SPLINE2 4520 3500 3500 3501 4521 0. 1. 450+
+ -1. -1. BOTH
SET1 4525 150 151
SPLINE2 4525 4510 4510 4513 4525 0. 1. +
+ 0. -1. BOTH
SET1 4621 215 250 251
SPLINE2 4620 3600 3600 3601 4621 0. 1. 460+
+ -1. -1. BOTH
SET1 4625 250 251
SPLINE2 4625 4610 4610 4613 4625 0. 1. +
+ 0. -1. BOTH
ENDDATA

Haughey Thesis 2018

Uploaded by

Copyright:

Available Formats

Haughey Thesis 2018

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Haughey Thesis 2018

Uploaded by

Copyright:

Available Formats

Development of a Static Aeroelastic Database Using NASTRAN SOL 144 for

A thesis submitted to the College of Engineering at

Author: Ryan M. Haughey

Advisor: Razvan Rusovici, Ph. D.

Table of Contents .................................................................................................... iv

Figure 1 – Vn Diagram for Symmetrical Flight .........................................................6

Table 1 Stability Derivative Interpolation Check ....................................................39

I would like to acknowledge the Northrop Grumman Corporation for assisting in

Current Aircraft Design Practices

Aircraft Design Phases

1. Conceptual Design – The initial design phase of an aircraft development

2. Preliminary Design – The second design phase of an aircraft development

3. Detail Design – The third design phase of an aircraft development program

“Aircraft conceptual design traditionally utilizes simplified analysis methods and

Figure 1 – Vn Diagram for Symmetrical Flight [2]

Integrated Schedule and Coordination

FINITE ELEMENT MODEL (STRUCTURAL ANALYSIS AND DESIGN)

NASTRAN Static Aeroelastics

Figure 3 SPLINE2 Bulk Data Entry

EID – Defines the element ID

CAERO – Interpolated aerodynamic panel ID

ID1/ID2 – First and last box or body element to be interpolated on

SETG – The set of inertia grids to interpolate the aerodynamic forces to

DZ – The linear attachment flexibility.

DTOR – The torsional flexibility ratio = EI/GJ.

DTHX/DTHY – Rotational attachment flexibility.

HA-144F NASTRAN Model

Figure 4 NASTRAN Structural Model

Figure 5 First Mode - Fuselage Yawing at 7.56Hz

Figure 8 NASTRAN Aero Model Mesh with Structural Model

Static Aeroelastic Trim Theory

First, a database of aerodynamic stability derivatives must be developed. These stability

A summary of the resulting rigid database can be seen in Appendix A

The general form of the aerodynamic loading can be represented as

Equation 1 CY Aero Loading

𝑝𝑏̅ 𝑞𝑐̅ 𝑟𝑏̅

Equation 2 CZ Aero Loading

𝑝𝑏̅ 𝑞𝑐̅ 𝑟𝑏̅

Equation 3 General Matrix Aerodynamic Loading

Each of the stability derivatives are non-dimensionalized by the freestream dynamic

Equation 4 Modified Coefficient Aerodynamic Loading g

Equation 5 Modified Coefficient Aerodyanamic Loading, Ax = B form

Equation 7 Inertial Linear Acceleration Loading

Equation 8 Euler Rigid Body Mechanics

𝐿𝐼 𝐼𝑥𝑥 𝐼𝑥𝑦 𝐼𝑥𝑧 𝑝̇ 𝑝 𝐼𝑥𝑥 𝐼𝑥𝑦 𝐼𝑥𝑧 𝑝

The rigid body moments can be expanded:

𝐿𝐼 = 𝐼𝑥𝑥 𝑝̇ − 𝐼𝑦𝑧 (𝑞 2 − 𝑟 2 ) − 𝐼𝑧𝑥 (𝑟̇ + 𝑝𝑞 ) − 𝐼𝑥𝑦 (𝑞̇ − 𝑟𝑝) −

Equation 10 Euler Rigid Body Pitching Moments Expanded

𝑀𝐼 = 𝐼𝑦𝑦 𝑞̇ − 𝐼𝑧𝑥 (𝑟 2 − 𝑝2 ) − 𝐼𝑥𝑦 (𝑝̇ + 𝑞𝑟) − 𝐼𝑦𝑧 (𝑟̇ − 𝑝𝑞 ) −

Equation 11 Euler Rigid Body Yawing Moments Expanded

𝑁𝐼 = 𝐼𝑧𝑧 𝑟̇ − 𝐼𝑥𝑦 (𝑝2 − 𝑞 2 ) − 𝐼𝑦𝑧 (𝑞̇ + 𝑟𝑝) − 𝐼𝑧𝑥 (𝑝̇ − 𝑞𝑟) −

Equation 12 Inertia Matrix

Equation 13 Symmetric Euler Rigid Body Pitching Moments

𝑳𝑰 = 𝑰𝒙𝒙 𝒑̇ − 𝑰𝒛𝒙 (𝒓̇ + 𝒑𝒒) − (𝑰𝒚𝒚 − 𝑰𝒛𝒛 )𝒒𝒓

Equation 14 Symmetric Euler Rigid Body Rolling Moments

𝑀𝐼 = 𝐼𝑦𝑦 𝑞̇ − 𝐼𝑧𝑥 (𝑟 2 − 𝑝2 ) − (𝐼𝑧𝑧 − 𝐼𝑥𝑥 )𝑟𝑝

Equation 15 Symmetric Euler Rigid Body Yawing Moments

𝑁𝐼 = 𝐼𝑧𝑧 𝑟̇ − 𝐼𝑧𝑥 (𝑝̇ − 𝑞𝑟) − (𝐼𝑥𝑥 − 𝐼𝑦𝑦 )𝑝𝑞

Equation 16 Aircraft Steady State Balanced Condition

This equation is of the form B = Ax, where: