Vous êtes sur la page 1sur 31

Modeling Aircraft Performance and Stability on X-Plane

Christopher W. S. Thong1
University of New South Wales at the Australian Defence Force Academy

Commercial flight simulator software has been long used by the public to hone their
piloting skills. X-Plane is the first commercially available flight simulator program
which directly outputs mathematical flight data without the need for programming
knowledge. This allows the public to readily simulate flight tests on a computer and also
acquire flight test data. This thesis models the F-15E in X-Plane and compares its
performance and handling characteristics to those predicted by theory and flight
manual data. These comparisons will be used to determine X-Planes usefulness as a
teaching and analytical tool in the demonstration of how forces and moments affect an
aircrafts flight path in flight.

Nomenclature
Terms
CAM
CFD
CFT
CG
CPU
DATCOM
FADEC
FAR
FDM
FSX
GB
Ghz
GUI
HUD
KIAS
KTAS
NACA
NASA
RAM
RDS
SFC
TR-824
USAF

Civil Aeronautics Manual


Computational Fluid Dynamics
Conformal Fuel Tanks
Centre of Gravity
Central Processing Unit
DATa COMpendium
Full Authority Digital Engine Control
Federal Aviation Regulation
Flight Dynamics Modules
Microsoft Flight Simulator X
Gigabyte
Gigahertz
Graphic User Interface
Heads-Up-Display
Knots Indicated Air Speed
Knots True Air Speed
National Advisory Committee for Aeronautics
National Aeronautics and Space Administration
Random Access Memory
Raymer Design Software
Specific Fuel Consumption
NACA Technical Report 824
United States Air Force

Body frontal area


Coefficient of Drag, consisting of skin friction, interference and pressure drag.
Average aircraft wing chord
Airfoil coefficient of lift
Coefficient of drag on the body
Maximum aircraft coefficient of lift
Drag on body
Wing Element Span
Frequency
Lift on wing element

Variables

Aeronautical Engineering Project, Thesis & Practical Experience ZEIT 4500


1
Final Thesis Report 2010, SEIT, UNSW@ADFA

Pitching moment
Dynamic pressure at flight condition
Reference wing area
Required Thrust
Wing element velocity
Stall speed
Aircraft weight
Angle of bank
Damping ratio
Air density
Natural frequency

I. Introduction
Computer flight simulation software has for decades been used extensively across the aerospace industry to
aid in the aircraft design. In this thesis, flight simulators will be taken to refer to computer software which
allows the user to pilot an aircraft with a visual representation of the cockpit and surroundings on the computer
monitor. It is also necessary to make the distinction between flight simulators and computer games which make
use of a similar cockpit GUI for the user to interact with. Thus, there is the additional constraint that flight
simulators must have a flight dynamics engine which sufficiently mimics real world physics. This is hard to
quantify so this thesis only considers software to be considered as a flight simulator if it focuses on the flight
dynamics and flying experience. While there are multiple benefits of flight simulation in relation to pilot
training (1 pp. 15-50), there are also numerous benefits for engineering applications. These include but are not
limited to:
time and cost savings as the aircraft can be simulated on the computer to determine its handling
characteristics and performance;
decreased personnel risk due to less actual test flights being flown ; and
the ability to test the effect of certain variables (such as the boundary layer effect or gravitational force)
on the aircraft which may be impossible to test in real life.
However, flight simulators built for engineering applications are usually expensive proprietary software,
built by major aircraft manufacturers to analyze aircraft before production. This thesis will use X-Plane, a
cheap, commercially available flight simulator, to model the F-15E aircraft.
Limitations of computer simulation are mainly based in the fundamental assumptions of the mathematical
model which drives internal calculations. Limitations may be exacerbated by the processing speed of the
computers used and accuracy required. Due to the usefulness of flight simulators, new software is constantly
being developed.
The thesis planning documents are contained in Appendices A, B and C whiles the essential and desirable
objectives of this thesis are listed below. Also, please note that due to the fact that numerous figures presented
in this thesis are in multiple colours, this report is best viewed in colour.
Essential Objectives
Collect dimensions, engine parameters and other required open source information on the F-15E
aircraft which will enable the creation of a model with maximum possible accuracy.
Build a model of the F-15E on X-Plane which can act as a teaching and building tool.
Obtain usable data relating to performance and handling qualities of the F-15E from public sources to
compare to X-Plane outputs.
Predict performance data and handling qualities of the F-15E by theoretical, non flight simulator means
to which the outputs obtained from X-Plane can be compared.
Compare the flight test outputs to predicted performance and stability data to assess the accuracy of the
simulator as well as the F-15E model.
Desirable Objectives
Develop as accurately as possible a working flight simulator model of the F-15E to assess the
possibility of making use of X-Plane to show the effect of changes on an aircraft.

2
Final Thesis Report 2010, SEIT, UNSW@ADFA

II. Methods of Predicting Aircraft Performance and Stability


There is a multitude of non-flight-simulator software which can provide estimates for aerodynamic
parameters, aerodynamic forces as well as stability and control characteristics of aircraft. These estimates will
be compared to certain X-Plane outputs so that the accuracy of using X-Plane can be established. The programs
elaborated on below are the more widely used software, each using different computational methods. It is also
possible to calculate certain parameters directly with the use of methods described by Raymer (2) and Roskam
(3) (4). Both these authors explain the method used to determine performance and stability characteristics and
provide necessary equations, graphs and historical data. They also explain the methodology as well as the
reasoning behind it and equation derivation in their books. However, their approaches are not covered in depth
in this section due to their relative simplicity.
A. USAF Stability and Control Digital DATCOM
DATCOM refers to a set of official USAF manuals which give guidance on how to calculate aerodynamic
coefficients, static and dynamic stability as well as control derivatives of fixed wing aircraft (5) while Digital
DATCOM refers to a USAF computer program written in FORTRAN which achieves the same task.
DATCOM requires the basic geometric properties, flight conditions and propulsion elements of the aircraft
under study in order to provide coefficients which approximate the performance and stability of the aircraft at
the stated flight conditions. It makes use of methods elaborated in the USAF Stability and Control DATCOM
volumes (6 p. 1) to compute 21000 different outputs from 125 inputs (7). The equations used by DATCOM are
stated in the official USAF Stability and Control DATCOM manual (8). DATCOM was developed on a
modular basis, with each module being a basic building block of an aircraft (6 p. 1). Examples of modules are:
body dimensions, synthesis parameters, wing plan form parameters and propeller/jet propulsion parameters (5).
Both Holy Cows, Inc. and the Open Aerospace Software Community have developed more user friendly
versions of DATCOM which are called DATCOM+ (9) and OpenDATCOM (10) respectively. While the only
difference between Digital DATCOM and DATCOM+ is some front end and back end adjustments to the
coding to increase user friendliness (7), Open DATCOM has a much easier to use GUI which was first released
as an alpha version in November 2009 (11). It is thus intended to use OpenDATCOM to predict the
performance and stability characteristics of the F-15E as this thesis author is unfamiliar with FORTRAN. .
Some limitations of DATCOM include not being able to calculate the effect of adding external stores or
inlets because DATCOM takes the fuselage to be a body of revolution as well as the fact that it is not possible to
attach two vertical tails to the aircraft. Possible remedies to these problems are increasing the cross section of
the fuselage such that the frontal cross section area of the enlarged fuselage is the same as that of the two inlets
in addition to the original fuselage and using a vertical tail with twice the area of the originals.
B. Raymer Design Software
RDS (12) primarily makes use of the aircraft design method outlined by Daniel P. Raymer Ph.D. in his book
Aircraft Design: A Conceptual Approach (13). RDS has multiple input parameters such as dimensions for
components including but not limited to the wings, fuselage, nacelles, seats and canopies. A component is
selected and the software then prompts the user for the required input data. RDS provides an extremely large
number of outputs and the ones most relevant to this thesis are performance outputs. These include takeoff and
landing speeds, rate of climb, turn rate and acceleration, which RDS can produce in graphical form.
While it was originally thought possible to use this software to predict the performance and handling
qualities of the F-15E, actual use of the software proved that it would be impractical to do so. One problem is
that the software requires aerodynamic inputs such as the aircrafts maximum, stall, climb and takeoff velocities,
as well as coefficients of lift and drag for the aircraft. This is because for aircraft design, the designer is
working towards designing an aircraft with certain characteristics in mind. Thus, this program cannot be used
for this thesis as most required RDS inputs are unknown.
C. Computational Fluid Dynamics
There are numerous different commercially developed computer programs which use CFD to solve fluid
flow problems including FLUENT, FLOW-3D and cfdesign. Due to the availability of FLUENT in the
university computer laboratories, it was initially planned make use of CFD by concentrating research on
FLUENT as well as learning how to use the software by enrolling in the CFD course. However, further work
on the subject proved overly time-consuming and CFD prediction was thus abandoned.
CFD is in essence fluid mechanic analysis of a situation making use of numerical methods and algorithms to
solve and analyze problems involving fluid flow. CFD makes use of three fundamental principles: conservation
of mass, conservation of energy and Newtons second law (14 p. 5). These three principles lead to the NavierStokes equations, as well as a variety of other equations necessary for modeling fluid flow. These mathematic
equations are usually partial differential equations in their most general form and CFD is the art of substituting
said equations for numbers and using the values along the boundaries to determine the fluid flow within the
3
Final Thesis Report 2010, SEIT, UNSW@ADFA

stated boundaries. CFD software does that by making use of


discretization methods such as the finite volume method, finite
element method, finite different method and the boundary
element method.
There are numerous ways of determining the fluid flow
within the boundaries. One could look at it as a two
dimensional situation or a three dimensional one. The user can
define a mesh which determines the number and position of grid
points at which the fluid flow must be determined.
Figure 1: A Hybrid Grid around an F-15
FLUENT is extremely flexible and capable, with the ability
(55)
to model both laminar and turbulent flows, eddy currents and
vortices with a variety of different user defined methods. It can calculate the force, moment and pressure
distributions on objects in a flow based on the test pieces geometry. FLUENT also presents data in a variety of
pictorial forms, one of which is shown in Fig. 1.
While it is fairly straightforward to develop simple models and meshes for use in FLUENT, it is a much
more complex task to model an aircraft in dynamic flight and was thus impractical to attempt considering the
scope of this thesis. This method can be very accurate in determining the forces on an aircraft but it would be
extremely time consuming to construct the model and complete the simulation. Even making use of a small
section of the aircraft (such as half the vertical tail) would require much calculation and modeling time.
Furthermore, the use of this extremely exact prediction method means that input data such as dimensions and
airfoils used would have to be much more accurate than those obtained for this thesis.
D. Comments
Due to the unavailability of the necessary specifications required to model the F-15E in RDS and the
resource intensity of CFD, these programs were not used to predict the performance of handling qualities of the
aircraft. However, OpenDATCOM appears to be a feasible method of predicting the required outputs. Of
utmost importance is the ability to validate our results. With the above stated software, the computer completes
all calculations and returns the requested outputs. This means that users are unable to obtain intermediate
calculations unless one writes new software to extract this data. Thus, certain calculations as described by
Roskam will be completed to validate a sample of OpenDATCOM outputs. The reason why Roskams
textbooks will be used instead of Raymers is due to the fact that Roskam is much more comprehensive and
detailed.

III. Existing Flight Simulators


This section reviews some of the best known flight simulators on the commercial market. All flight
simulators reviewed in this section appear to have adequate internal computations to determine aircraft flight
performance and handling characteristics. Unfortunately, they do not have a built in method to enable the user
to easily extract performance or handling data from the program without the need for programming knowledge.
While a large number of software tagged with the name flight simulator exist commercially, most are either
computer games or have inadequate FDMs which do not calculate aircraft motion accurately enough to be
considered a flight simulator comparable to X-Plane. Thus, only the following two simulators were chosen to be
discussed in detail so that X-Planes greater suitability for this thesis can be appreciated.
A. FlightGear
FlightGear (15) is a free flight simulator maintained by Curt Olson which can be readily hacked to allow
users to tailor the program to their specific need. There are multiple different usable FDMs which make use of
different methods to model aerodynamic forces and moments to predict aircraft flight paths (16).
The most commonly used FDM is JSBSim (16) an open source FDM in C++ which makes use of the
coefficient buildup method (17 pp. 5,16). JSBSim requires knowledge of the aerodynamic specifications such
as force and moment coefficients as well as stability and control derivatives (18 pp. 18 - 24). JSBSim also
requires data on graphs such as coefficient of lift versus angle of attack graphs. Due to its open source nature,
JSBSim is an extremely well developed FDM as an extensive range of users have contributed ideas for its
continual improvement (18 p. Preface).
Another notable FDM is YASim which, like X-Plane, takes primarily physical dimensions of the aircraft
into consideration for flight modeling (15 p. YASim Readme). The main difference between the two in terms of
data input is that YASim requires data such as cruise and approach information (15 p. YASim Readme) while
X-Plane takes airfoil data (19) instead. Thus, YASim, as with JSBSim, requires one to know to an extent the
abilities of the aircraft being modeled. There is a YASim helicopter FDM which makes use of blade element
theory (20) but throughout the process of literature review, no airplane FDMs were found which make use of
blade element theory.
4
Final Thesis Report 2010, SEIT, UNSW@ADFA

While calculations are performed to determine the aircrafts flight path in simulation, there are no prebuilt
options in the FlightGear simulation GUI or command line prompts for the user to automatically extract any
data to a spreadsheet or text file (15), (16). Thus, the user would have to write a program to interface with
FlightGear if he wanted to use this software to determine certain data values such as maximum thrust or lift to
drag ratio. One could even take the simulator out of the loop and feed data straight into JSBSim (17) or YASim
and extract data from there. However, these possibilities would require programming knowledge beyond that of
this thesis author.
B. Microsoft Flight Simulator X
FSX is one of the most popular flight simulation software products in the world (21). FSX makes use of an
FDM which consists of an *.air file and an aircraft.cfg file. The airfile contains hundreds of parameters
which define the aircrafts flight characteristics but these parameters usually cannot be freely accessed and
altered due to copyright restrictions outlined by the aircraft manufacturers and the aircraft model designers (22).
Fortunately, the online community has developed hacking programs which allow the user to edit the .air file,
two of which being available at Mudpond (23) and Flight Simulator Aircraft Dynamics (24). The cfg-file is
made up of parameters such as joystick force feedback, elevator effectiveness, roll stability, CG location, engine
performance, aircraft geometry, stall speed etc. Most of the parameters contained in the cfg-file are related to
altering the users flight simulator cockpit experience and do not affect the flight characteristics of the aircraft.
The FDM largely depends on stability derivatives to determine an aircrafts flight path. This flight simulator
does not provide any data output for users to graph or view and unfortunately for the flight simulator world,
Microsoft has laid off the entire Microsoft Flight Simulator team in order to align Microsofts resources with
(its) strategic priorities (25).
C. Comments
The discussed flight simulation programs have very different backgrounds but similar outcomes a usable
flight simulator which can be used for pilot training but no built in method of easily extracting flight data from
the simulator. This is largely due to the fact that most commercially available flight simulators make use of
stability derivatives to determine an aircrafts flight path. Eq. 1 (2 p. 470) shows the yaw stability derivative.
(1)
Not only is stability derivative information kept confidential by aircraft manufacturers and their clients, the
above equation indicates that stability derivatives change with aircraft weight, payload positioning and aircraft
damage. Thus, if one was to simulate an aircraft in one of the above flight simulators, he would usually have to
predict the stability derivative required to achieve certain handling qualities in flight. Thus, it is possible to
have a visual model of a Boeing-747 which actually flies like an F-15 due to the fact that the models flight
characteristics are determined by stability derivatives as opposed to the actual simulation model. As a result,
these flight simulators cannot predict aircraft motion like X-Plane does. However, it is possible for one to use
an FDM in FlightGear which does not make use of stability derivatives. However, information on this was
sparse and thus not elaborated on.

IV. The X Plane Software Package


X-Plane is a commercially available flight simulator which revolutionized the flight simulator industry by
using blade element theory in its FDM as opposed to the more traditional methods explained above (26). It is
not just a flight simulator used for pilot training and leisure; it can also be used to determine the forces on an
aircraft in flight and readily outputs flight data to the user without the need to do any programming. Numerous
companies and American government departments have purchased X-Plane in bulk to be used as an engineering
tool, including Cessna, Cirrus, Boeing, Lockheed Martin, NASA and the USAF (27). The software provided by
X-Plane consists of two relevant design programs with multiple inbuilt modules to enable its users to create
aircraft and airfoils as easily as possible.
The two programs are Plane Maker and Airfoil Maker. These two components, as their names suggest, are
used for designing aircraft and airfoils respectively. Due to the existence of an official X-Plane forum (28) and
the lack of specific details regarding X-Plane calculations in the user manual, this forum was engaged at
multiple times through the literature review stage so as to determine exactly how X-Plane completes
calculations and makes use of data for flight path prediction. Austin Meyer, the creator of X-Plane, was also
emailed a number of times for answers to more in-depth questions.
A. Mathematic Mechanics behind X-Plane

5
Final Thesis Report 2010, SEIT, UNSW@ADFA

Blade element theory is the method used by X-Plane to calculate an estimate of the forces acting on an
aircraft in flight. The aircrafts lifting surfaces are divided into multiple longitudinal strips and the calculated
forces on each strip are then added together. This gives an estimate for the total force due to lifting surface such
as the wings and empennage on the aircraft. As an example, the lift equation shown in Eq. 2 (29) with dr, the
element span, being depicted in Fig. 2. The velocity used is the trigonometric sum of the longitudinal and
transverse component of flow over the wing. This allows X-Plane to consider the effect of sideslip on the forces
on the aircraft. The same method is used to determine the lift, drag and moment on all lifting surfaces made up
of an airfoil.
(2)
Laminar Research claims that
using blade element theory to
compute the forces on the aircraft is
more accurate than the stability
derivative method (26).
Stability
derivatives indicate how quickly an
aircraft returns to straight and level
flight for each degree it is offset.
While this method is acceptable in
most civil aviation flight conditions, it
does not take into account effects
Figure 2: Depiction of how dr is determined on an F-15 wing (59)
such as engine failures, turbulence,
stalls, spins, sonic effects and spiraling slipstream to name a few (26). Furthermore, the stability derivative
method is not able to predict how a new aircraft prototype will fly in a given flight condition. The aircraft
model creator has to have full knowledge of the aircrafts performance and stability characteristics as the
software is unable to determine them.
On the other hand, using blade element theory, X-Plane determines an aircrafts flight characteristics with
the aircrafts geometry as well as engine and airfoil data (26). The cycle of calculations determining the flight
path of the aircraft is contained in Appendix D, as stated on X-Planes official website (26).
Inspection of the data and calculation output from X-Plane gives some insight into the intricate details
regarding calculations involved. For example, it can be determined from the cycle dump file that the three
dimensional drag forces and moments on each modeled component is taken into account in each cycle. These
components include the fuselage, external stores and landing gear (if extended).
B. Plane Maker
Plane Maker (30) has numerous functions which allow the user to create nearly any possible aircraft
geometry. While Plane Maker creates the visual model used for simulation, it must be noted that unlike most
other commercial flight simulators, the dimensions of this model are directly taken into consideration for flight
modeling.
The modules contained within Plane Maker are used for designing the fuselage, wings, empennage,
undercarriage as well as miscellaneous bodies and wings such as wing pylons, speed brakes and CFTs. Due to
the fact that not all wings have wings of steady taper and airfoil, it will be necessary to model the aircraft with
multiple wing sections. Since the F-15E makes use of a NACA64A006.6 airfoil at the wing root and a
NACA64A203 airfoil at the wing tip, it is also necessary to model this in X-Plane with the right airfoil data. XPlane can interpolate between wing root and tip to obtain a smooth transition between the two airfoils (31)
which meets the above stated requirement.
In most of the modules, the user simply enters values into boxes for each stated parameter. However, for the
fuselage and miscellaneous bodies, it is possible to load a bitmap image onto the background of the GUI. A
bitmap image of the left side view and top view can be loaded. Plane Maker allows one to edit the fuselage
cross section at a maximum of 20 locations along the fuselage length, with up to nine nodes at each cross
section. This allows for much more accurate modeling of the fuselage than simply stating the radius and length,
as is the case in most software which can only model simple bodies of revolution.
When the designer is required to input the body coefficient of drag, the designer must include interference,
skin friction and pressure drag in this body coefficient of drag value (32). The drag on each body (such as the
fuselage) is then calculated with use of Eq. 3.
(3)

6
Final Thesis Report 2010, SEIT, UNSW@ADFA

Other Plane Maker inputs include engine parameters and locations, forward, default and aft centers of
gravity and coefficient of friction between the wheels and the ground. Plane Maker can also blend parts of the
aircraft together, leading to the ability to smoothly join the fuselage and wings together like on the F-16 aircraft.
If necessary, Plane Maker has a GUI to program artificial stability and autopilot systems into the aircraft.
A notable characteristic of X-Plane is that it is compatible with numerous other well-known software
platforms such as Blender, SketchUp and FSX. Blender and Sketchup are drawing programs which allow one
to create objects. Aircraft models can be loaded from Blender into Plane Maker while scenery objects from all
three of the mentioned programs can be uploaded into X-Plane (33).
C. Airfoil Maker
Airfoil Maker is used to set the characteristics of the airfoils used by an aircraft. General airfoil information
required includes the Reynolds number at which the airfoil data is entered, the thickness to chord ratio and drag
divergence mach number at zero lift. Since a wings coefficients of lift, drag and moment vary with Reynolds
number, it is possible for one to insert data for the wing at multiple different Reynolds numbers and X-Plane
will interpolate between the data to determine the most realistic coefficients at the flight Reynolds number. If
data at only one Reynolds
number is input, X-Plane
will approximate for other
Reynolds numbers.
Next, the coefficient to
angle of attack graphs
must be set. The graph
shown in Fig. 3 depicts
the plot for the NACA
64A006.6 airfoil.
The
graph
shows
the
coefficients between an
angle of attack of -200 and
200. Notable points would
include the fact that
minimum drag appears to
Figure 3: Coefficient of Lift versus Angle of Attack (57) for NACA 64A006.6. Each
occur at zero angle of
graduation on the x and y-axes indicates 10 angle of attack and 0.1 coefficient value
attack, there exists a
respectively.
maximum and minimum
lift coefficient which indicates the stall condition and that the
wing has a predominately pitch-down moment, as indicated by
the coefficient which is negative for most angles of attack.
Numerous controls are available which enable the user to
produce these graphs.
Airfoil Maker also presents the coefficient versus angle of
attack graphs from -1800 to 1800 for the users information. The
user thus modifies the graphs for the -200 and 200 region and
determines the shape of the graphs over the remaining angles of
attack. The user can then check his airfoil data by calculating
the airfoil lift on drag ratio by inserting the aspect ratio and
Oswalds efficiency factor of the wing.
There are a number of airfoils already built in X-Plane.
These are mainly standard NACA airfoils. The F-15E uses the
NACA 64A006.6 and 64A203 airfoils (34) and the only NACA
64 airfoil available in Airfoil Maker is the NACA 64-208 airfoil.
Thus, appropriate airfoils need to be created for an accurate
F-15E model. Two sources will be consulted for this the
NACA Technical Report 824 (TR-824) and Javafoil. TR-824
contains airfoil data for numerous different airfoils and the ones
most similar to the F-15Es are the NACA 64-006 and NACA
64-206. Data includes mean line data, the coordinates for upper
Figure 4: Graphs of Coefficient of Drag
and lower surfaces, predicted critical Mach numbers and
against Angle of Attack at multiple
aerodynamic characteristics such as the coefficients of lift, drag
Reynolds Numbers for NACA 64A006
and moment. These coefficients are necessary for modeling the
(60 p. 177)
airfoil in Airfoil Maker but TR-824 only consists of data from 7
Final Thesis Report 2010, SEIT, UNSW@ADFA

120 to 160, as shown in Fig. 4. There were even less data points in some other graphs. Javafoil (35), a free
airfoil analysis program was used to approximate the coefficients over the rest of the required -200 to 200 range.
Javafoil outputs are contained in Appendix I.
Javafoil results are highly agreeable (36) with University of Illinois Urbana-Champaign (UIUC) wind tunnel
tests which is indicative of Javafoils accuracy. The UIUC experiments airfoil data were published in three
volumes (37) (38) (39) between 1995 and 1998 but unfortunately do not contain any data relevant to the airfoils
used in the F-15E (40). TR-824 was completed in 1945 and thus, the accuracy of the UIUC experiments can be
expected to be superior. Furthermore, the X-Plane stock airfoil data was compared to both TR-824 and Javafoil
data and was found to be more convergent to Javafoil data. Thus, when there was a deviation between Javafoil
and TR-824 coefficient values, the Javafoil value was favoured.
D. X-Plane Flight Simulator
The X-Plane Flight Simulator
program allows one to fly an aircraft
provided with the X-Plane software,
downloaded from the internet or
created by the user. There are many
features which increase X-Planes
realism. For example, if the payload
was changed, the new aircraft weight
and CG would be automatically
determined. As with other flight
simulators, system failures can be
programmed to occur as preset or
randomly.
Also, airports from
around the world can be selected in
X-Plane Version 9.
The weather of the simulation
Figure 5: A Screenshot from X-Plane of the Forces Acting on the
environment can be set which is
designed F-15E Model (27)
extremely useful when conducting
flight tests as the wind, cloud, precipitation and visibility can be set to optimum levels. Temperature and
pressure at sea level can be set to mimic the conditions at which the F-15E performance and handling data was
obtained. Turbulence and slipstream effects are also modeled in X-Plane (26).
The Flight Simulator has a Data Input & Output tab which is extremely useful for this thesis. The entire
flights data for up to 131 different parameter sets can be output into a text document for graphing or manual
interpolation. The table of all the possible data outputs is included in Appendix E. The 131 data sets provide
606 different parameters for the designed F-15E. The total number of data types would change dependent on
the number of components modeled in an aircraft. Up to 16 of these parameter sets can be displayed in real
time during flight. The user can set the disc rate, i.e. number of data points per second, to be between
0.1/second and 99.9/second. A disk rate of 15.0/second will be used since it is the default setting.
Structural limits can be set such that flying surfaces are removed in over-speed and over-G conditions, flaps
and gear doors are removed when they are extended beyond their stated never-exceed speed. This improves
realism for pilot training.
X-Plane also has the ability to pictorially present information about the aircrafts flight path so far, the forces
acting on each object and wing, the velocity vectors of the flow field around the aircraft as well as angle of
attack vectors. This information is displayed on the aircraft model and an example of forces on the aircraft is
shown above in Fig. 5. In order not to clutter the pictorial display, only forces on the wings, empennage and
engines are displayed. One point which deserves explanation is that the element where the wing joins the
fuselage contributes the most lift to the F-15E. This is because the leading edge extension increases the lift at
this point.
E. Data and Cycle Dump Text Files
These two files are output by X-Plane and can be used for further analysis. The data text file is made up of
up to 131 different parameter sets as indicated in the output parameter selection tab is shown in Appendix E.
The data test file output after the completion of each flight contained data which was used to graph and
determine the F-15E models flight characteristics.
The cycle dump file on the other hand, is an entire cycles worth of calculations output into a data file. A
cycle is as described in Appendix E. It shows the calculations for force buildup on each element, engine,
miscellaneous object and payload item. Downwash and finite wing loss calculations are also taken into account,

8
Final Thesis Report 2010, SEIT, UNSW@ADFA

as described in Appendix D. Cycle dump also gives the resultant radius of gyration in each of the three axes for
the simulated aircraft and the aircrafts static margin.
F. Limitations
Online forums indicate what users think of X-Plane and how it compares to other commercially available
flight simulator software. However, most forum members are interested in the simulation experience such as
memory usage and the smoothness of graphics. Other than the inability to model fuselage lift, technical
limitations discussed by forum members include the inaccuracy of X-Plane in modeling separated flow over
surfaces as well as the lack of vortex analysis and inaccuracy of transonic and supersonic modeling. Another
real world effect which is not simulated in X-Plane is the wing flex.
Of importance to this thesis is the issue of X-Planes transonic and supersonic modeling (26). While
compressible flow effects are considered using Prandtl-Glauert, transonic effects are only simulated by an
empirical Mach-divergent drag increase. This is the Mach number at which compressibility effects start to
become apparent. The only GUI in which the Mach-divergent drag can be input is Airfoil Maker.
In
supersonic flight, the aircraft airfoils are assumed to have a diamond shape with the same thickness ratio as the
input airfoils (26). Pressures behind the shock waves are then calculated on each of the plates on the diamond
airfoil and summed to give the total pressures on each airfoil element. This would enable the determination of
lift and drag on each airfoil element. Additional drag on each component due to supersonic effects is calculated
by determining the compression shock on each component. This compression shock is based on the angular
difference between neighbouring components and the local flow.
One interesting limitation of X-Plane is that if display options are set too high for the computer being used,
the cycle rate for completing calculations will decrease and the aircraft will suffer from abnormal oscillations.
Thus, in flight tests, graphics options were set to a minimum.
There are also some technical limitations discussed by online magazines. One such discussed limitation is
the inaccuracy of secondary aileron effect modeling. In reality, if a step aileron input acts on an aircraft, the
aircraft first rolls about its longitudinal axis and yaws in the opposite direction. After the turn is established, the
aircraft should yaw in the direction of the turn and if the turn is continued, the nose should drop and the aircraft
should enter a spiral descent. Sim Pilot Magazine writer looked at how a Cessna 172 in X-Plane reacted to an
aileron input (41). It eventually rolled in the opposite direction due to an increase in airspeed and a pitch up
moment developed after the initial roll. Thus, the spiral dive never developed, emphasizing the differences
between X-Planes mathematical modeling and reality. This scenario was recreated in X-Plane a number of
times to confirm the articles claim and it was validated. This indicates that the authors other claims are also
likely valid. Due to time constraints, no attempt was made to determine the exact deficiency in the X-Plane
modeling.

V. F-15E Overview
In order to accurately model the F-15E in X-Plane, accurate dimensions and engine parameters were
required. The F-15E makes use of the Pratt & Whitney F100-PW-220 or F100-PW-229 engines (42), (42).
Since the only performance charts obtainable were those of an F-15E equipped with two F100-PW-220 engines,
the F-15E was modeled as such. Information on the F-15E was limited to aircraft dimensions, some engine data
and CG locations.
A. Dimensions
Externally, the F-15A and F-15E base airframes are identical, with the largest visible differences being the
addition of CFTs and a tandem cockpit on the F-15E (43 p. 55). The most specific F-15 schematic found is
included in Appendix F and was used to base the F-15E model on. Actual photographs of the F-15E were used
to accurately model the canopy and CFTs.

Table 1: Data related to the F100-PW-220 (47)

B. Performance
Basic engine data can be seen in Table 1.
Performance charts and data included in the F-15E flight
manual includes stall speeds, one engine inoperable
engine rate of climb, all engines operable rate of climb,
fuel required to climb, combat ceiling, optimum long
range cruise, long range cruise, level flight envelope,
maximum speed at level flight, dive recovery, level flight acceleration and sustained level turns. Most charts
are available for a variety of different aircraft configurations.
Usage of X-Plane indicates that the most easily replicated results would be those for stall speeds, level flight
envelope, maximum speed at level flight as well as sustained level turns.
9
Final Thesis Report 2010, SEIT, UNSW@ADFA

An example of the data contained in the flight manual is shown in Appendix G (42 pp. A1-14) which
indicates the stall speed with gear and flaps up or down and provides a myriad of data which can be compared
with X-Plane outputs to test its ability to replicate real lift data.
C. Stability
Gross weight and CG locations for different payload combinations are contained in the F-15E flight manual
(42 pp. A1-8,9, A5-9). The aircraft has a 10 anhedral which has a slight destabilizing effect. This counters the
stabilizing effect of the high wing configuration, thus increasing the F-15Es maneuverability.

VI. Creation of the X-Plane Model in Plane Maker and Airfoil Maker
In order to create a working model of the F-15E for use in X-Plane, it was necessary to study freely available
flight simulation models of the F-15E to fine-tune the method used to create the model. Unfortunately, all F-15
X-Plane models were found to be based on the model initially developed by Anthony Booher which meant that
there was only one original F-15 model available. This model was deconstructed piece by piece and numerous
errors, inefficiencies and points for improvement were found. One such inaccuracy with Anthony Boohers F15E model is that it did not adhere to the official stall speeds.
Deconstruction and analysis of Anthony Boohers F-15 model and other X-Plane fighter aircraft models
produced the following framework for producing the F-15E X-Plane Model. Numerous different components
were used to model the complex F-15E shape accurately.
A. Fuselage
The F-15E fuselage was relatively simple to
model. The cross section of the aircraft was
defined at 20 stations along the fuselage length
with the maximum of nine nodes selected at each
station. A bitmap image was loaded into Plane
Maker and each of the nine points can be dragged
into position. Accurate 3-view diagrams of the F15A were used to build the fuselage. Fig 6 shows
the top and side view of the model, with the top
view bitmap loaded. The body coefficient of drag
must also be defined in this section. This was
defined to be equal to 0.03, as indicated by
Figure 6: The top and side view of the fuselage design GUI
Hoerners estimations of drag numerous different
(30), with a top view bitmap loaded to show how the
body shapes (44 pp. 3-9 to 3-22). Interference
definition points line up.
drag was neglected at this stage and would be
taken into consideration if total aircraft drag was lower than expected.
B. Airfoils
Six different airfoils were used on this aircraft, with two being
airfoils provided in the X-Plane package and four being created during
this thesis. The two stock airfoil files used were the flat_plate(very
thin).afl and NACA0009(symmetrical).afl which were used for the
horizontal and vertical tails as well as the connection of the
empennage to the fuselage.
The specially created
NACA64A006.6R9.afl, NACA64A203R9.afl and MidptR9.afl airfoil
files were used on the wings and leading edge extension while
eagle_fuse.afl was used within the F-15E fuselage to mimic the F15Es ability to create lift with its fuselage.
The horizontal and vertical tails were modeled as very thin flat
plates as indicated by observation of photographs. This airfoil did not
produce any moment between angle of attack of -200 and 200 and
produced no lift when the angle of attack was 00. In reality, however,
aircraft typically use an airfoil which generates a force opposing wing
lift (2 p. 471). The use of very thin plat plates was deduced to be a
reasonable assumption this is standard practice in the X-Plane
community. Many comparable aircraft provided in the X-Plane
package as well as those produced by the online community make use
of flat_plate(very thin).afl for the horizontal and vertical tails.

Figure 7: Side view of the F-15E


Empennage

10
Final Thesis Report 2010, SEIT, UNSW@ADFA

The empennage connection to the fuselage and wings was


modeled by making use of the NACA0009(symmetrical).afl
airfoil file. Being a symmetrical airfoil, it did not produce lift at
00 angle of attack. Also, as seen in Fig. 7 and 8, the airfoil is a
very similar shape to the actual empennage connection.
The NACA64A006.6R9.afl and NACA64A203R9.afl airfoil
files were designed by making use of two sources TR824 and
the Javafoil program. The relevant TR 824 and Javafoil charts
are included in Appendix H and I respectively which can be
compared to the X-Plane airfoil files shown in Appendix J.
MidptR9.afl was designed by setting each of its input parameters
to be the mean between those of NACA64A006.6R9.afl and
NACA64203R9.afl. This is because MidptR9.afl was used to
ensure that the airfoil at the midpoint between wing root and
wingtip was such that there would be a smooth transition in
airfoil shape from root to tip.
Figure 8: Bottom view of the F-15E
Eagle_fuse.afl was used to simulate the fuselage lift of the FEmpennage
15E. Due to the fact that X-Plane does not reduce the drag or
moment on overlapping components, the coefficient of drag and moment of this airfoil were set to be zero so
that the internal airfoil would not cause additional drag or moments.
C. Wings and Empennage
The wings and miscellaneous wings functions were used to
create five different symmetrical wing sections, as well as the
horizontal and vertical tail planes. Each wing section is defined
by several factors. The factors are the wing section span, root
chord, tip chord, wing sweep, dihedral and wing location.
In order to determine these dimensions for each wing
section, measurements of three-view diagrams were taken for
each wing section. A bitmap can also be loaded onto the main
background of the GUI and the aircraft model can be rotated
Figure 9: Four wing sections used to model
such that it lines up with the bitmap. This was used to ensure
the F-15 leading edge extension (1) and the
that the dimensions and location of each wing section was
main wing (2).
correct.
Each wing section can also have a different wing incidence at 20 points
along its length. The wing incidence was estimated by observation of
different photographs of F-15s such as those in Appendix K. Due to a lack
of data, the wing incidence was estimated.
The wing airfoil can be set either to be a single airfoil or linearly
changing from one airfoil to another. This function was used such that the
airfoil could be selected to be a NACA 64A006.6 airfoil at the root and a
NACA 64A203 airfoil at the tip with the midpoint airfoil in between.
One wing section is set to be within the aircraft fuselage and engine
inlets to simulate the lift generated by the F-15E body. This section made
Figure 10: The sections
use of eagle_fuse.afl.
initially used to design the
One wing section was designed as the leading edge extension up to the
horizontal tail
engine inlet. Again, precise measurements were not available so the shape
was approximated with the NACA64A006.6R9.afl airfoil file.
The remaining three wing sections were used to model the F-15E main wing. Three separate wing sections
were required for accurate modeling due to the differing trailing edge sweep angles, as can be seen in Fig. 9.
The wing root was modeled as a NACA 64A006.6 airfoil and the tip was a NACA 64A203 airfoil. The wing
airfoil appears to vary linearly from root to tip and this was modeled by manually interpolating and creating an
additional airfoil as stated in the above airfoil section.
The empennage was designed by using three wing sections one for each of the vertical tails and one for the
horizontal tail. Two wing sections initially used for the horizontal tail because of its unconventional stepped
shape, as shown in Fig. 10. Unfortunately, the more complex shape caused some issues as described in the
below discrepancies section. Thus, a simple unstepped horizontal stabilizer was used instead.
One unusual feature of the F-15 is that its fuselage creates enough lift to keep it airborne with only one wing.
This was demonstrated when an F-15I pilot was involved in a training accident which resulted in one entire
wing being ripped off from the wing root. (45 p. 420). Unfortunately, X-Plane is unable to determine if the
11
Final Thesis Report 2010, SEIT, UNSW@ADFA

fuselage will be able to generate lift. This was determined by reducing the coefficient of lift of all airfoils on the
aircraft to zero. The way around this was to insert a dragless wing within the models fuselage and X-Plane will
calculate the lift on that wing while still adding the drag of the fuselage to the model.

D. Engines
The engines modeled in X-Plane were F100-PW-220s. Their location and shape was determined by the
three view diagrams. The NASA Glenn EngineSim Program (46) was used to determine additional engine
parameters which were unavailable in the F100-PW-220 information sheet (47). The engines were set to have
specific fuel consumptions of 0.01/hr so as to have minimal effect on aircraft weight as the test flights
progressed. This meant that special test flights would not need to be designed to factor in changing aircraft
weight. Engines are only simulated to have drag if the engines are idling or flamed out.
E. Control Surfaces
The size and shape of ailerons, flaps and the rudders were determined from the three view diagrams. The
entire horizontal tail plane deflected to act as an elevator. The deflection angles were determined by observing
the deflection angles used to create similar X-Plane models. The ailerons were set to deflect 25 0 upwards and
150 downwards, the elevators were set to deflect 300 upwards and 180 downwards while the rudders were set to
deflect 300 in each direction. The flaps were plain flaps and set to deflect 300 downwards at full deflection.
F. Systems
Plane Maker has the ability for a designer to select properties in the aircrafts electrical, bus, pressurization,
hydraulics, starter, warning, radio altimeter, instrument arc limits, external light, internal light and spring
bouncer systems. Due to the fact that these systems did not alter the X-Plane models flight characteristics and
were not required for accurate flight modeling, this section of modeling was not considered.
G. Artificial Stability
The artificial stability GUI allows one to specify the artificial stability and autopilot characteristics of the
model. The artificial stability system aims to maintain the models angle of attack, sideslip and roll-rate.
Without the appropriate artificial stability system, the model suffers from unwanted oscillations. During test
flights to confirm the artificial stability settings, inappropriate settings occasionally caused the model to oscillate
at such a rate that the airframe exceeded its g-limit. Autopilot settings also needed to be entered and fine-tuned.
Again, inappropriate settings caused excessive oscillations which made accurate test flights impossible.
H. Landing Gear and Speed Brake
The only tests requiring the landing gear to be down were the stall tests with flaps and gear deployed. The
X-Plane software does internal calculations to determine drag due to landing gear. The landing gear and speed
brake were sized with the aid of three view diagrams and photographs of the F-15E. This was a useful piece of
equipment during test flights as it enabled rapid deceleration of the model to the required test velocity. Neither
the landing gear nor speed brake contribute any drag while stowed.
I. Engine Intakes and Conformal Fuel Tanks
The engine intakes and CFTs were modeled with six different miscellaneous bodies so as maximize the
accuracy of the models shape, with three bodies on each side of the fuselage. One body was required to model
the air intake, one for the shaft from the air intake to the engine and one for the CFT. Due to the fact that XPlane does not reduce the drag for overlapping components, no more bodies were used. As with the fuselage
design, the miscellaneous bodies GUI allows each body to have up to 20 stations, each with 9 definition points.
The coefficient of drag also needed to be selected for these components. Due to the fact that air flows straight
through the engine intakes, the coefficient of drag was set to be minimal. However, if the aircraft were meeting
the air at an angle to the engine intakes, the intakes would still contribute drag based on its frontal area to the
angle of attack. This is in line with real world physics.
J. Other Miscellaneous Bodies
Miscellaneous bodies were also used to design the underbelly of the model, the cockpit and the boom
located in between the engine nozzles. The underbelly was shaped by comparison to photographs of F-15E
aircraft. The cockpit was shaped by use of the 3 view diagram. The cockpits coefficient of drag was again
obtained by comparison to similar bodies in Hoerner (44 pp. 3-9 to 3-22). The boom was also included for
accurate portrayal of the F-15E. Accurate portrayal of all parts of the aircraft was necessary for X-Plane to
determine the models frontal area when calculating drag.

12
Final Thesis Report 2010, SEIT, UNSW@ADFA

K. Payload
The payloads used on the designed X-Plane
model were stock stores provided with the XPlane software. Table 2 provides information
on the stores used and their respective
asymmetric effects.
Payload of differing
weights and location were tested to determine
if there was any measurable effect on the
models flight characteristics. Actual flight
tests or CFD would be useful to quantify the
effect of the tested payloads.

Table

2:

Comparison

of

Payload

Asymmetric

Effects

L. Cockpit Layout
Due to the fact that the cockpit layout of model is insignificant to the flight dynamics of the aircraft, the
cockpit layout of an existing model (48) which already had the essential autopilot switches and HUD was used.
This was to facilitate easier conduct of flight tests.
M. Aircraft Paint Scheme
The aircraft paint schemes only purpose is to improve the models image. It is possible to alter the physical
look of the aircraft with the paint scheme. This is undesirable but is hard to prevent due to the fact that the paint
scheme is designed by making use of photographs of the actual aircraft which leads to obvious differences
between the actual aircraft and the model. While this does not affect flight simulation and flying characteristics,
it may confuse users who are not familiar with the software. Thus, the model was left unpainted as this would
enable one to see the models actual geometry during flight.
N. Discrepancies between Model and Reality
One of the discrepancies between the model and the actual
aircraft is the lack of a step in the horizontal stabilizer. If the step
is modeled, each horizontal stabilizer is made up of two wing
sections. Each wing section would have a different root and
average chord. Since the horizontal stabilizers automatically
rotate about the quarter chord point, each wing section rotates
about a different axis and the horizontal stabilizer resultantly
Figure 11: Separation of horizontal
separates during deflection. This effect is shown in Fig. 11. This
stabiliser due to use of two wing sections
leads to the moment about the CG of the model being slightly
incorrect as the forces on the outboard section of the horizontal
stabilizer act on a different plane. It will also cause concern to
users of the X-Plane model who are unaware of this issue. In
order to correct this, the F-15E was modeled with non-stepped
horizontal stabilizer. The surface area of the horizontal stabilizer
was kept the same so as to maximize accuracy of forces and
moments simulated.
One more physical difference between the model and real
aircraft is the fuselage to empennage connection region. The
difference is shown is Fig. 12. The difference exists due to
Figure 12: Slight shape discrepancy to
miscellaneous bodies in X-Planes requiring an axis of symmetry.
prevent overlapping of components
It would be possible to use another few miscellaneous bodies to
increase the visual accuracy of the model but this would not lead to significant improvements in simulation.
This is because X-Plane does not factor in the effect of fluid flow around miscellaneous body on another (49).
Thus, additional modeling would only lead to overly increased drag since X-Plane doesnt take into account the
overlapping of components.
Another difference is that the F-15E was modeled with an SFC of 0.01/hr. This meant that a negligible
amount of fuel was consumed in flight and that test flights need not be designed to factor in weight loss. Fuel
consumption is the only parameter affected by SFC so this does not alter the models flight characteristics at all.
The rationale behind being able to neglect weight loss in flight is explained in Appendix L. There is some
discrepancy in that the real F-15E has a rounded wing tip but the X-Plane model has a straight wing tip, this has
been decided to be acceptable as the wing area was kept the same.

13
Final Thesis Report 2010, SEIT, UNSW@ADFA

VII. Prediction of F-15E Performance and Stability by Methods other than X-Plane
As stated above, this thesis aims to compare X-Plane flight test results to theoretical predictions as well as
flight manual data. This section will analyse the predictions and determine if they can be used as a baseline to
which flight test results can be compared.
A. OpenDATCOM
Unfortunately, through use of the program, it was discovered that there are some errors present in the
OpenDATCOM GUI. Thus, no outputs were able to be extracted from OpenDATCOM. However, it would be
possible to use the determined inputs in DATCOM+ or the original Digital Datcom program. This would only
be possible if the user has experience in using FORTRAN, which this thesis author does not. Thus, while not
used in comparison of X-plane outputs, DATCOM inputs are attached in Appendix M while explanations to the
terms are contained in (50) so that future work can be readily initiated with regards to the use of DATCOM to
further refine the F-15E model.
Table 3: Percentage Uncertainty in Flight
B. Flight Manual Data
Manual Stall Speed Data
The F-15E flight manual contains data and charts which
have aided in the design of the F-15 X-Plane model. The
charts with which the experimental data are compared to are
contained in Appendix G and N. The charts in Appendix G
indicate the F-15s stall speed at 00, 150, 300, 450 and 600
angle of bank at a range of aircraft weights between 30000lbs
and 80000lbs while the charts in Appendix N shows the level
flight acceleration on the F-15 at set weights between
42800lbs and 66900lbs. The level flight acceleration graphs
taper off to the maximum speed for the thrust setting
(maximum or military) at 10000ft and 40000ft which is the
value we can compare the F-15 models maximum level flight
speed to. Maximum thrust is the aircrafts maximum thrust
with afterburners while military thrust is the aircrafts
maximum thrust without afterburners.
Table 4: Percentage Uncertainty in Flight Manual Maximum
There is also an uncertainty associated
Thrust Data
with the flight manual data. This is due to
the flight manuals method of data
presentation. The data is all presented on
graphs and it is necessary to read the data
from the graphs. Thus, the uncertainty in the
flight manual data at each point would be
equal to half a division of the graph. Tables
3 and 4 show the percentage uncertainty
associated with each data point taken from
the flight manual. This analysis led to
suspicion that the flight manual data may
contain some anomalous results. This is
because the maximum level flight speed at
61500lbs did not conform to the trend of the
other data points. The maximum level flight
speed for all conditions was expected to decrease with increasing aircraft weight but the data at 61200lbs
appears to be an outlier. The data in the flight manual definitely has an associated uncertainty and this would
most likely be the cause of this discrepancy. Unfortunately, the uncertainty in flight manual data was not
included in the flight manual.

C. Theoretical Calculations
Since DATCOM could not be used to provide the bulk of prediction data, Roskams aircraft design manuals
were used to calculate the expected stall speed, maximum level flight speed as well as the phugoid and short
period characteristics.
The stall speed is simply obtained by making use of Eq. 4. The wing reference area was simply obtained
from (34), with air density being dependent on altitude while weight is the chosen variable. The maximum
aircraft coefficient of lift was obtained from flight manual stall speed data by making use of Eq. 4. The chosen
maximum aircraft coefficient of lift was the one which led to the least percentage deviation between

14
Final Thesis Report 2010, SEIT, UNSW@ADFA

theoretically predicted stall speeds and flight manual stall speeds. This method was used as the exact required
data was unobtainable in open source sources.

(4)

The theoretical and official clean stall speed values are as stated in Table 5. Due to the lack of open source
information on the maximum aircraft coefficient of lift, this value was determined from the official stall speed.
The differences between the official and theoretical stall speeds for clean and flapped configurations for the full
range of aircraft weights were at their minimums with a clean maximum aircraft lift coefficient of 2.36 and
flapped maximum aircraft lift coefficient of 2.46. Raymer indicates that the flapped maximum aircraft lift
coefficient should be 0.9 more than the unflapped value due to the use of plain flaps (2 p. 326). This is not the
case in these calculations which may indicate that the F-15Es flaps are not as effective as those on other
aircraft.
The theoretical stall
Table 5: Comparison of Theoretical and Official Stall Speeds
speed in KTAS was
determined from Eq.4
and
subsequently
changed to KIAS. The
maximum
percentage
difference as seen is 3.64%. As can be seen,
the theoretical stall
speed is higher than
official values at lower
weights but is lower
than official values at
higher weights.
This
would indicate that the
assumption that the
aircraft has a single value for the maximum aircraft coefficient of lift is incorrect. The trend indicates that as the
F-15Es weight is increase, its maximum aircraft coefficient of lift actually decreases. This is contrary to what
Raymers equations suggest in (2 pp. 316 - 327).
The maximum level flight speed can be determined with Raymer (2 p. 518). However, the Raymer
calculations require knowledge of the aircrafts coefficient of drag. In order to determine the coefficient of drag
theoretically, it is necessary to factor in the skin friction drag, pressure drag, induced drag, interference drag as
well as wave drag. While not impossible to calculate, it will be extremely tedious to consider all required
factors.
Calculated engine power output would not be an appropriate baseline which X-Plane outputs could be
compared to. This is due to the fact that the F100-PW-220s on the F-15E make use of a FADEC and wastegate
to control the power output at different heights for maximum efficiency. Thus, maximum level flight speed
calculations will not be done.
The F-15E flight manual does not contain any numerical data or charts relating to phugoid or short period
motion. Thus, it was necessary to determine these characteristics through theoretical calculations. The
equations used in the calculations are contained in Appendix O. 31 input parameters are required, with 5 of
them either requiring excessive amounts of calculations (such as drag value) or unavailable data (such as
)
and 4 parameters being dependent on flight test conditions. Those 5 parameters were thus determined through
flight tests. The rest of the input parameters were
Table 6: Uncertainty in Dynamic Longitudinal
simply obtained from measurements of three-view
Stability Theoretical Calculations
diagrams or available F-15E data. The 31 input
parameters used in this set of calculations are listed and
valued in Appendix R. The phugoid and short period
and for the F-15E were thus calculated at altitudes of
10000ft and 40000ft and weights of 32000lbs, 55000lbs
and 80000lbs. The reason only six tests were conducted
for the phugoid and short period motion flight tests is
that much time is required for each test as well as its subsequent calculations.
Calculations were completed so as to determine the uncertainty associated with the short period and phugoid
and due to measurements taken. The uncertainty is only due to measurements taken and does not include
15
Final Thesis Report 2010, SEIT, UNSW@ADFA

the uncertainty associated with


and
. This is because those two parameters were extremely hard to
measure accurately due to the fact that the aircraft suffers from slight oscillations. The oscillations can be seen
in all the X-Plane output graphs contained within the Appendices. The uncertainty associated with

and

could not be measured due to this fact. Thus, a certain amount of trial and error was required to determine
and
values which were generally correct. Further testing would enable more accurate predictions of
. This is because the
is the only prediction which requires those two parameters as inputs.
Thus, since the uncertainty associated with
and
was not included in the measurements, the uncertainty
for and
is actually slightly more than as stated in Table 6.
The uncertainties in Table 6 also do not take into account the fact that Roskams equations are
approximations of real world effects. Assumptions are made so approximate an aircrafts motion to the
equations. This is an avenue for uncertainty but cannot be measured.
One fact which can be deduced from Table 6 is that there is much more uncertainty associated with short
period motion characteristics than phugoid ones.
D. Summary of Results
The data from the official F-15E flight manual was originally expected to be perfect. However, the above
analysis indicated that certain data (such as the maximum speeds at 61500lbs) could be anomalous.
Furthermore, there would definitely be an uncertainty associated with the data which was not stated in the flight
manual. There were no anomalous data points in the stall speed charts. The uncertainty associated with reading
the flight manual charts was between 0.42% and 1.18% for maximum level flight speeds and between 0.56%
and 0.91% for stall speeds. Thus, the F-15E model would be expected to achieve a maximum level flight within
1.2% of flight manual data and stall speeds less than 1.0% different to flight manual data.
Theoretical stall speeds show close compliance with an uncertainty of only 3.7% to flight manual data. This
would indicate that it is unlikely that the flight manual stall speed charts would contain significant errors.
However, analysis indicates that the aircraft coefficient of lift would be a changing value. Due to the lack of
any definitive data on the aircraft coefficient of lift, this parameter will be taken as a constant.
Phugoid and short period
and cannot be compared to any real world data. Thus, the theoretical data
cannot be validated against anything. However, they can be expected to be fairly accurate in giving a general
estimate of the actual aircraft
and . This is because the
and align with expected values. As expected,
is very close to unity, while
is closer to zero. The
are also significantly
shorter that
.
All considered, the predictions are expected to be fairly accurate baselines to which flight test data can be
compared.

VIII. Flight Test Methods


The models stall speed, level flight acceleration and maximum speed as well as dynamic longitudinal
stability are examined in this section. The reason these four performance and stability characteristics were
chosen is that they can be verified by methods mentioned in previous section as well as the fact that they test
different flight dynamics characteristics.
Stall speed (51 pp. 53, 54) is an indication of the aircrafts lift while level flight acceleration and maximum
speed is an indication of thrust and drag. Dynamic longitudinal stability is dependent on a multitude of aircraft
parameters, as indicated in Appendices O and P.
In order to accurately conduct the test flights and compare data extracted from the simulated flights to flight
manual data, the simulated test flights must conform to the required test standard. The different test procedures
are clearly stated below, with certain alterations made due to simplifications made possible by the use of a flight
simulator to conduct the test flights.
A. Stall Speeds
The following method was considered for stall speed testing. It is the method stated by both CAM3 and
FAR Part 23 and summarized in (52 pp. 47-52). The aircraft should first be trimmed at approximately 1.5 times
the stall speed. Following that, the throttle should be brought back to idle and the deceleration rate to stall be
kept at 1 kn/s. The first two conditions could be met easily but there was no easy way of measuring and
controlling the deceleration. Thus, this method was altered slightly such that the test pilot simply keeps the
aircraft flying straight and level till the aircraft stalls.

16
Final Thesis Report 2010, SEIT, UNSW@ADFA

The stall is defined to be the point at which the aircraft pitches nose down uncontrollably in level flight.
This point can be confirmed by using X-Planes data output function. The aircrafts elevator deflection, altitude
and calibrated airspeed can be displayed as a function of time.
The aircraft had no external payload, as stated in the official flight manual. The flight manual contains stall
graphs for the aircraft in the clean condition as well as stall graphs for the aircraft with flaps and gear down.
Thus, flight tests were conducted for the F-15E in both configurations. Appendix L shows the calculations
which indicate that the weight change in each flight test is negligible. While this is completely unrealistic in
mimicking real world physics, it is an advantage of making use of flight simulators and keeps the CG position
as well as the aircraft weight basically constant throughout the test flight.
In order to accurately measure the stall speed of an aircraft at a given altitude and weight, each test point was
tested three times. This is to ensure validity of data. Any anomalous data points would be replaced by data
from another test. The resultant stall speed is the average of the three data points. Due to the fact that official
stall speeds were determined to the closest whole number, flight test results shall also be presented to the closest
whole number.
Two main test groups were conducted one for the F-15E in the clean configuration and one in the flaps and
gears down configuration. The models stall speed was determined at 32000lbs, 45000lbs, 55000lbs, 65000lbs,
75000lbs and 80000lbs as well as at angles of bank of 0 0, 150, 300, 450 and 600 for both F-15 configurations. All
stall tests were conducted at 10000ft.
B. Level Flight Acceleration and Maximum Level Flight Speed Flight Test Procedure
The actual flight test technique for the speed power test of a jet aircraft is quite complex and it involves
many mathematical calculations, including the use of non-dimensionalisation to ensure accurate results. This is
because as fuel is burnt, the aircraft becomes lighter. In a real world flight test, the aircraft would be flown
while varying test altitude in order to keep a constant (52 pp. 99, 100). This would require a large amount of
preflight planning as well as a great amount of pilot skill as it would be imperative to fly exactly the correct
flight path. However, as explained in Appendix L, the aircrafts change in weight for these flight tests is
negligible. Thus, the method used would be to simply reach the required test altitude, stabilize at the test
altitude, engage the altitude hold autopilot, reach the initial test flight velocity then increase the throttle to its
appropriate maximum or military thrust setting and wait for the aircraft to reach its maximum speed.
The maximum level flight speed at maximum thrust was taken to be when the throttle was at maximum
thrust with afterburners engaged, altitude is constant and the aircraft was unable to accelerate anymore. The
maximum level flight speed at military thrust was the same but without afterburners.
The aircraft payload and weight were as stated for the stall speeds. The test flights were conducted with a
clean airframe. Due to the continuous nature of this flight test, only one test needs to be conducted at each
altitude and weight.
The tests were carried out at 10000ft and 40000ft, at 9 different aircraft weights between 42800lbs and
66900lbs. This is such that each available flight manual graph can be compared against. The tests at 10000ft
were conducted both at maximum thrust as well as at military thrust while those at 40000ft were conducted at
maximum thrust. This is due to limitations in flight manual data.
C. Dynamic Longitudinal Stability Flight Test Procedure
Phugoid Motion
(52 p. 248) indicates that the aircraft should first trimmed to
the test trim speed at the predetermined test altitude. Once these
conditions are met, the aircraft is then displaced by 10 to 15
knots from the trim by elevator deflection. The control column
is then released and the resultant aircraft oscillation is recorded.
The airspeed can be either increased or decreased and normal
procedure is to observe phugoid motion from both cases. Once
the control column is released to its trim position, the control
column can be left free or held steady. Again, both possibilities
are usually tested as differences between stick-fixed and stickfree cases normally exist. However, due to the lack of feedback
in the joystick used to interface with the flight simulator
program, the stick-fixed case cannot be simulated. This was the
exact method used to conduct these flight tests.
The method for calculating of the
from flight test
data is shown in Fig. 13 while the calculations required for
are shown in Eq. 5 and Eq. 6.
17
Final Thesis Report 2010, SEIT, UNSW@ADFA
Figure 13: Chart used to determine the
Phugoid Damping Ratio (52 p. 250)

(5)

Where

(6)

The phugoid motion is a large amplitude variation of aircraft airspeed, pitch angle and altitude. From a sideon perspective, the aircraft would appear to be travelling in a sinusoidal motion. The important values
determined by this flight test are
and the
. The
of an aircraft is its resonant frequency, the
frequency at which oscillations will increase amplitude with increased oscillations. is related to the ratio of an
oscillatory amplitude to the following amplitude. If the is positive, it is a measure of how quickly an
oscillation converges. If the is negative, it is a measure of how quickly an oscillation diverges.
The aircraft payload was as stated for the stall speeds. Due to the substantial time taken for each test and
their consequent required calculations, only three weights were considered for the phugoid motion tests. These
weights were 32000lbs, 50000lbs and 80000lbs. CG location and weight changes were not considered as
indicated by Appendix L.
Two flight tests were conducted at each flight condition so as to obtain results from both possible flight test
methods. The aircraft was allowed to undergo at least five full oscillations in each flight test so that an average
and
can be obtained.
The flight tests were conducted at altitudes of 10000ft and 40000ft, meaning a total of twelve different flight
tests.
D. Dynamic Longitudinal Stability Flight Test Procedure Short Period Motion
There are three possible methods to conduct the short period motion flight test the pulse input, doublet
input and 2-g pull up method. The method used in this thesis was the doublet input due to its ability to suppress
the phugoid motion as well as the lower required pilot ability. The doublet method requires the aircraft to first
be in a trimmed, unaccelerated level flight. Once the test begins, the pilot rapidly moves control column
forward (or backward), then backward (or forward), then back to trim. Again, due to the joystick used, it was
not possible to hold the control column at the trim position so the stick-free scenario was considered.
An aircrafts short period motion is usually a heavily damped oscillation with a maximum period of a few
seconds. The motion is a rapid pitching of
the aircraft about its CG. Due to the rapidity
of oscillations, it is important for an
aircrafts short period motion to be heavily
damped as the pilot would be unable to
correct the motion easily. Again, the short
period oscillations have an
and with
the same definitions as those for the phugoid
motion, with the method of determination
contained in Fig. 14.
The aircraft payload and weight will be
as stated for the phugoid motion tests.
As with the stall tests, each flight test
was conducted three times. Ideally, the
aircraft would have the same airspeed and
altitude at the start of each test. The doublet
method will be varied alternatively, i.e. if on
the first test the aircraft is displaced upwards
then downwards, in the following test it will
be varied downwards then upwards.
The flight tests were conducted at
Figure 14: Method to determine short period natural frequency
altitudes of 10000ft and 40000ft for a total
from flight test data (52 p. 252)
of six different flight tests.

IX. Quantitative Analysis and Verification of X-Plane Outputs


This section will focus on analyzing and verifying the results of the X-Plane flight tests. The analysis
section will consider the results of each set of flight tests and confirm if they conform to an expected trend. It
will indicate X-Planes ability to mimic real world physics. This is determined by whether each set of flight test
results conforms to the expected trend.
18
Final Thesis Report 2010, SEIT, UNSW@ADFA

The verification section will indicate the accuracy of X-Planes flight characteristic prediction. Verification
will be done in two ways - one will be to simply compare the obtained flight test results to the official and
theoretical values while the other will be to take into
account the change in tested flight characteristic with
the flight test variable.
The second form of
verification is called the delta method as it simply
considers the delta, or change, of a variable. This
would enable any inaccuracies in modeling the F-15E
to be neglected. The verification section will also
determine the source of inconsistencies between flight
test, theoretical and official values and look at
possible ways to improve the F-15E model if it is
determined that further improvements are required.
It should be noted that in this section, any
reference to flight testing would be specifically to
flight tests conducted in X-Plane with the F-15E
Figure 15: Depiction of Lift and Weight Forces on a
model designed by this thesis author.
Banked Aircraft (58)

A. Clean and Flapped Stall Speed Test Analysis


As an aircrafts weight increases, so does its stall
speed, as indicated in Eq. 4. The stall speed also
increases with aircraft angle of bank. This is because
roll decreases the lift force working against aircraft
weight, as depicted in Fig. 15. These trends are visibly
present in the Fig. 16 and Fig. 17 plot. As angle of
bank and aircraft weight increases, so does stall speed.
It can also be seen that at two points in each graph, the
plots for stall speeds at 00 and 150 angle of bank
overlap.
This can be attributed to pilot error.
Furthermore, the official clean stall speeds for
unbanked flight is114kts at 32000lbs and 174kts at
75000lbs while those for 150 banked flight is 116kts at
32000lbs and 177kts at 75000lbs. These values are
extremely close and a slight deviation in data may lead
to the plots overlapping. Similarly, the flapped stall
speed values are very close together for 00 and 150
angle of bank. Thus, apart from the infrequent
overlapping, the trend behaves as expected. This would
indicate that the X-Plane software simulates stall speeds
correctly. The following section will quantitatively
consider the ability of X-Planes simulation software to
accurately model stall.
B. Clean and Flapped Stall Speed Verification
The comparison plot and table of X-Plane
flight test to flight manual clean stall speeds
against aircraft weight at 00 angle of bank are
shown in Fig. 16 and Table 7 respectively. The
complete set of the clean stall speed comparison
plots are contained in Appendix Q. The difference
between X-Plane clean stall speed test results and
official values for the entire range of aircraft
weight and angle of bank is less than 4.7%, as
indicated by Table 7.
The comparison plot and table of flapped stall
speeds against aircraft weight at 00 angle of bank
are shown in Fig. 17 and Table. 8 respectively. The
rest of the flapped stall speed comparison plots are
contained in Appendix R. The difference between

Figure 16: Flight Test Clean Stall Speeds

Figure 17: Flight Test Flapped Stall Speeds


Table 7: Percentage Difference between Flight Test and
Official Clean Stall Speeds

Table 8: Percentage Difference between Flight Test and


Official Flapped Stall Speeds

19
Final Thesis Report 2010, SEIT, UNSW@ADFA

X-Plane flapped stall speed test results and official values for the entire range of aircraft weight and angle of
bank is less than 3.4%, as indicated by Table 8.
The clean and flapped stall speeds indicate both a high degree of accuracy of the X-Plane F-15E model in
terms of its stall characteristics as well as a high degree of accuracy of the X-Plane software in modeling the
stall condition.
C. Level Flight Acceleration and Maximum Speed Analysis
It was initially planned to only consider the maximum level flight speed. However, upon conducting initial
flight tests to refine the F-15E models accuracy, it was discovered that the F-15E model had substantial issues
with reaching its appropriate maximum level flight velocities. At 10000ft and maximum thrust, the aircraft
accelerates to a maximum speed of Mach 1 instead of the official value of Mach 1.35 while at 40000ft, 42800lbs
and maximum thrust, the aircraft only reaches Mach 1.4 as opposed to the official value of Mach 2.4. Thus, it
was determined to also consider the acceleration from 250KTAS such that it can also be compared to the charts
from the F-15E flight manual. 250 KTAS is approximately equivalent to Mach 0.39 at 10000ft and Mach 0.83
at 40000ft.
Figs 18, 19 and 20 are graphs of time taken to accelerate from 250 KTAS to its maximum speed with the F15Es simulated acceleration compared to the official values by putting them on the same graph. Flight manual
graphs were loaded onto the plot background with flight test results superimposed over them. The rest of the
comparison plots are shown in Appendix S.
D. Level Flight Acceleration and Maximum Speed Verification
The maximum test speed
at all altitudes and thrust
levels differ significantly
from the official values. At
10000ft, for both maximum
and military thrust, flight tests
indicated that the
F-15E
model accelerates to Mach 1
for all weights. At 40000ft,
the
aircraft
obtains
a
maximum
speed
of
approximately Mach 1.35.
Figure 18: Comparison Chart of X-Plane Outputs to Flight Manual Chart (42)
The
flight
manual
indicates that at 10000ft and
maximum thrust, the F-15E should accelerate to Mach
1.35 at 42800lbs and 1.06 at 66900lbs. At military
thrust, the aircraft be able to reach Mach 1 at the
minimum tested weight of 42800lbs but would be at
lower speeds at higher weights. At 40000ft, the
aircraft at minimum and maximum test weights should
accelerate to Mach 2.4 and Mach 1.6 respectively.
However, as can be seen from Fig 18, 19 and 20
the F-15E model accelerates much faster than stated in
the flight manual. This trend is consistent in all level
flight acceleration flight tests. These graphs may
prove hard to read in black and white.
Figure 19: Comparison Chart of X-Plane Outputs to
The lower maximum speed at maximum thrust
Flight Manual Chart (42)
indicates that X-Plane simulates too little thrust to
drag in supersonic flight, while the higher maximum
speed at military thrust and greater acceleration
indicates that X-Plane simulates too much thrust to
drag in subsonic flight. There are two possibilities for
the significant differences between simulated and real
world data either the X-Plane model could be
imperfectly built or the simulator could be incorrectly
modeling real world physics.
The X-Plane model may be inaccurate in two
aspects the thrust at 10000ft and 40000ft may have
been set too high or the total drag on the aircraft could
Figure 20: Comparison Chart of X-Plane Outputs to
Flight Manual Chart (42)

20
Final Thesis Report 2010, SEIT, UNSW@ADFA

be too low. This was deduced by the previous analysis.


The thrust at altitude was determined by the X-Plane software, with the only way to change this parameter
being to change the baseline thrust at 100% low compressor speed at sea level, the critical altitude at which the
engine can put out its maximum allowable thrust, and whether the FADEC or waste-gate system is engaged.
The FADEC controls the operating parameters of the engine such that it operates at its maximum efficiency for
all flight conditions while the waste-gate prevents the engine from exceeding its maximum allowable thrust.
The critical altitude and FADEC settings were chosen after consulting the official X-Plane forum and Anthony
Booher, the designer of the other available X-Plane F-15 models. This may be an avenue for error but the lack
of other open source information on these parameters led this thesis author to make use of values recommended
by the X-Plane forum. Another source of error may be that Janes rates the F100-PW-220 in terms of dry thrust
and maximum takeoff thrust instead of thrust at 100% low compressor speed (47). This refers to the thrust
when the stage 1 low pressure compressor is at its maximum speed. This dry thrust value was assumed to be
equal to the 100% low compressor speed, with the difference between maximum takeoff thrust and dry thrust to
be equal to the additional thrust provided by afterburners. This assumption may be found to be invalid with
more data but open source information does not state the 100% low compressor speed thrust.
The total drag on the aircraft could be overly low due to a number of factors. The drag coefficient on each
component could have been set too low or the software could be incorrectly determining skin friction,
interference and wave drag. The only possible way of altering the drag on the aircraft would be to change the
drag coefficient on each component.
At 10000ft and military thrust, the F-15E model flies faster than expected while at 40000ft and maximum
thrust, the F-15 model flies slower than expected. While increasing weight was appropriately simulated by a
resultant decreased maximum speed, the delta in decreased speed did not match up to that of official values.
The model appears to suffer from excessive drag once it reaches transonic and supersonic speeds. In order
to test the theory that X-Plane was modeling supersonic flight incorrectly, a simple level flight deceleration
flight test was conducted.
E. Troubleshooting Flight Tests
It was initially thought that a level
flight acceleration flight test would be
more appropriate but this led to
extreme oscillations at velocities
above Mach 1. In order to show the
trend of drag coefficient to Mach
number at 10000ft, the X-Plane model
was edited such that it had enough
thrust to accelerate well beyond Mach
1.4 and Mach 2.4 at 10000ft and
40000ft respectively. The aircraft
would then be allowed to decelerate
in level flight to near stall velocity.
The resultant trend of drag coefficient
against Mach number would then be
plotted. The aircraft would be in a
clean configuration and with a weight
Figure 21: Comparison of Flight Test Zero Lift Drag Coefficient to
Known Zero Lift Drag Coefficient Graphs (2 p. 346)
of 32000lbs.
As can be seen in Fig. 21, the drag
coefficient remains constant from Mach 0.5 to about Mach 0.8 before increasing slightly in a parabolic curve.
The drag instantly increases at Mach 1. The drag coefficient then increases to peak at about Mach 1.1 before
generally decreasing with increasing Mach number. A sketch of how the coefficient of drag versus Mach
number plot should look is superimposed over the plot in Fig. 21. Aircraft in Fig. 21 which are comparable to
the F-15E are the F104, F14, F4, F105 and the RA5C. Pictures of these aircraft and the F-15E are contained in
Appendix T.
As can be seen, all of them have a peak in drag coefficient at between Mach 1 and Mach 1.1. This is
accurately simulated in X-Plane. However, all of the drag coefficient peaks of about 0.04, except the F104 with
a peak at 0.05. This is very different from the F-15E X-Plane models peak drag coefficient of about 0.12.
However, the drag coefficients between Mach 0.5 to the critical Mach number are between 0.015 and 0.02 for
both the F-15E model at 10000ft and the comparable aircraft. It can be deduced that the X-Plane F-15E models
drag is thus modeled relatively accurately in the subsonic zone but is excessive in the transonic and supersonic
zone. This would seem to suggest an issue with the X-Plane softwares internal calculations as the aircraft

21
Final Thesis Report 2010, SEIT, UNSW@ADFA

designer only inputs one set of drag coefficient data without distinction between subsonic and supersonic
speeds.
Another difference is that the drag suddenly increases at Mach 1 instead of increasing from the critical Mach
number to a peak at around Mach 1. The drag coefficient obtained from flight tests then remains at nearly the
same level before further increasing and peaking at about Mach 1.1. This indicates that the mathematical
modeling involved in mimicking transonic and supersonic drag has significant room for improvement. The
imperfect transonic and supersonic drag modeling may be the reason for the F-15E aircraft to be unable to reach
its appropriate maximum velocities.
As indicated by the above analysis, the drag coefficient is accurately modeled at subsonic speeds. This
indicates that the drag coefficient of individual components is modeled correctly. At transonic and supersonic
speeds however, total drag coefficient is excessively high, more than twice the value of expected results. This
likely indicates that drag coefficient is overestimated by X-Plane. However, excessive acceleration in the
subsonic zone indicates that the X-Plane F-15E model has too much thrust at both tested altitudes.
Unfortunately, due to the lack of definitive thrust at 100% low compressor speed and the critical altitude of the
engines, accurate results would only be obtainable by trial and error. The correct thrust and drag settings will
improve and the method will be described in a later section.
F. Phugoid Motion Analysis
The F-15E aircraft underwent
phugoid motion as predicted, as
shown in Fig 22.
The first
downward trough is the second
pulse of the doublet input. The
oscillations lasted for many
minutes with gradually decreasing
amplitude. Using the test method
described above, the graph of
indicated airspeed against time
was plotted and from which the
and
were
determined.
Each test was
conducted until the aircrafts
Figure 22: Plot showing the F-15E Phugoid Motion obtained from X-Plane
altitude
and
airspeed
was
changing minimally from the initial test altitude and airspeed.
As stated in the above flight test method description, it was planned to do two flight tests for each test point.
One flight test was initiated by pitching upwards while the other was initiated by pitching downwards.
However, difficulties in obtaining the exact same altitude and starting test speed meant that it was not feasible to
accurately compare data from one flight test to another. Thus, while two flight tests were completed in each
condition only one would be used as the comparison to theoretical data. This also meant that statistical analysis
was not done for phugoid motion flight test data. The flight test which was compared to theoretical data was the
one with the most similar
and .
Theoretical predictions indicated that the
at both altitudes should remain fairly constant. Any
deviation from a constant value should be a slight decrease in
as weight increases. This trend was upheld by
flight test results, with the flight test
remaining fairly constant with an increase in aircraft weight. The
test results at 40000ft were extremely compliant to the expected horizontal trend line.
The
was also expected to have a slight decrease, with the theoretical
at 55000lbs and
80000lbs being approximately 0.60 and 0.40 times the value of the theoretical
at 32000lbs. This trend
was not represented in the data obtained from flight testing. The
obtained from flight tests at 10000ft
was basically constant while the values obtained at 40000ft indicated an increase with aircraft weight. Analysis
of the method used to determine
Table 9: Value of Theoretical Dynamic Longitudinal Stability Characteristics
led to a possibility as to
why the
values obtained
were trending differently from
theoretical values.
As can be seen in Fig 13, the
amplitude of each peak and
trough must be obtained and
then the can be obtained from
the ratio of a peak to its
22
Final Thesis Report 2010, SEIT, UNSW@ADFA

subsequent trough or vice versa. In order to increase accuracy, a minimum of five full oscillations were
conducted for each flight test. Multiple ratios of peak amplitude to trough amplitude were thus obtained and the
average value was determined for subsequent use. The peak to trough ratio was thus a fluctuating value
throughout the experiments. This led to a slight uncertainty in the obtained . Furthermore, the method used for
presenting the correlation between peak to trough ratio and was extremely coarse as shown in Fig 13. More
accurate graphs or data could be found so it would be expected that small changes in would not be detectable.
G. Phugoid Motion Verification
The calculated flight test
and
are
compared to theoretical predictions
in Tables 9, 10 and 11. As can be
seen in Table 11, the
obtained from flight tests are within
30% of theoretical predictions.
Considering
the
uncertainty
associated with the theoretical
calculation, the flight test
values show significant correlation
to theory. The
obtained
from flight tests were mostly within
84% deviation from the theoretical
predictions. Considering the small
absolute value of the difference
between those flight test results and
theoretical predictions, they can be
accepted as fairly accurate. Only
the
at 10000ft and
80000lbs had a higher difference of
567%. This may possibly be due to

Table 11: Value of Dynamic Longitudinal Stability Characteristics obtained


from Flight Tests

Table 10: Comparison of Flight Test Dynamic Longitudinal Stability Results


to Theoretical Values

incorrect
and
values as
well as the coarse method of
determining
from flight tests. Unfortunately, further testing did not result in an improved
.
This is because of a short coming of the X-Plane trim function. It was found to be extremely difficult to trim the
aircraft to exact straight and level flight. In most cases, the aircraft would ascend or descend at a slow rate.
This affected the repeatability of the flight tests, as well as the accuracy of flight test outputs. The phugoid
characteristics appear to correlate to theoretical predictions much better than the short period values. This is
likely due to the significantly lower uncertainty due to the fewer number of inputs and equations involved.
H. Short Period Motion Analysis
As with phugoid motion flight tests, it was not feasible to compare data between multiple similar flight tests.
This was due to the slightly different starting altitudes and aircraft velocities. Furthermore, only a few points in
the flight test were used for calculations. Thus, statistical analysis was not conducted on the short period motion
flight test data. Due to the quantity and uncertainty of different variables and equations used to determine the
short period motion characteristics, there was much more uncertainty associated with it. The actual uncertainty
cannot actually be calculated due to the fact that uncertainty also lies in the use of equations. This is because
empirical equations used make underlying assumptions which cannot be quantified. This is a possibility as to
why the calculated
seems to be overly low at 40000ft.
Even though there are no values to which the short period characteristics can be compared, we know that the
F-15E must satisfy the military safety regulations. The MIL-F-8785C for Class IV aircraft (fighter) category B
flight phase (cruise) specifies that the
for an aircraft of this class is to between 0.2 and 2.00 (53 p.
13). Also, FAR part 23.181 (a) states that the short period motion is to be heavily damped, meaning having not
more than two cycles.
Taking into consideration both these regulations, it can be determined that theoretical predictions at 10000ft
80000lbs, 40000ft 55000lbs and 40000ft 80000lbs do not adhere to these guidelines. The simulated aircraft at
these three conditions undergoes more than two cycles of oscillations before reaching a steady state. Thus, it
would indicate that either the uncertainties are excessive or the simplifying assumptions invalid for these flight
conditions. Thus, these theoretical predictions will not be used to as the baseline for flight test results. It must
23
Final Thesis Report 2010, SEIT, UNSW@ADFA

be noted that this analysis does not indicate accuracy of the calculated theoretical predictions, merely the
possibility that they are correct based on these references.
I. Short Period Motion Verification
Table 6 indicates the uncertainty associated with the dynamic longitudinal stability characteristic prediction.
There is a much larger uncertainty associated with the aircrafts short period motion than phugoid. Thus, it is
within expectations that the
obtained from flight tests has a larger difference than other results
when compared to theoretical results. However, the average difference is about 5000%, as indicated in Table
11. This is far too large for results to be considered comparable. Furthermore, the
are significantly
different from theoretical predictions. This would indicate that either the F-15E model is inaccurately modeled
with respect to short period motion or that X-Plane is inaccurately determining the aircraft motion in this region
of flight. Further testing would be required to pinpoint the exact reason for the deviation of short period motion
from expected results.

X. X-Plane Repeatability and Comparison Flight Tests with Payload


It is a desired outcome of this thesis to develop as accurately as possible a working flight simulator model of
the F-15E which can show the effect of changes to the aircraft. For ease of modeling, these changes were
chosen to be the addition of an asymmetrical payload. However, due to the lack of a method to which flight test
results can be compared, the effect of changes to the aircraft cannot be discussed quantitatively, only
qualitatively. It is also necessary to determine whether differences between each flight test are significant. By
choosing a flight test which can be easily replicated for multiple different payloads, systematic error between
flight tests can be made negligible. The optimum aircraft weight, altitude and speed for the flight tests should
also be determined for maximum accuracy in results and thus significance when comparing to results of other
flight tests.
A. Determination of Optimum Flight Test Method
Multiple experiments were conducted to compare X-Plane outputs with expected trends. Due to the lack of
quantitative information in the flight manual regarding the F-15E under different payload conditions, a simple
flight plan was determined for the following flight tests. The aircraft would simply be allowed to decelerate
under conditions of no thrust from one flight velocity to another. Thus, it would be necessary to determine at
which altitude, weight and velocity the tests would face the least uncertainty. This was due to the fact that the
inbuilt autopilot led to slight oscillations in autopilot-controlled flight, it was necessary to determine the best
flight condition for minimum oscillations.
1. Determination of the Optimum Aircraft Altitude and Weight for X-Plane Payload Flight Tests
Experiments were conducted to determine the normal spread of results between
Table 12: Percentage
identical flight tests. Nine flight tests were performed in each flight condition and the
Uncertainty in
standard deviation of the test results in each flight condition was determined as shown
Standard Deviation
in Appendix U.
For a normal distribution, approximately 68% of measurements will lie within one
standard deviation of their mean (54 p. 34) and about 95% of measurements within two
standard deviations. Thus, the smaller the standard deviation, the lower the spread of
test results and the more accurate a comparison between different flight conditions.
Table 12 indicates the number of test flights required for a certain percentage
uncertainty in standard deviation. It was decided to conduct nine test flights such that a
low percentage uncertainty would be obtained while minimizing the amount of time
spent conducting test flights.
Once the optimum flight condition has been determined, the standard deviation will
be determined. This standard deviation would be such that there is a 95% probability that the flight parameter
would be within two standard deviations of the population mean and the uncertainty associated with this
standard deviation would be 25%.
The F-15E model was tested in the following conditions:
1.
2.
3.
4.
5.

Clean, 10000ft, 32000lbs aircraft weight


Clean, 10000ft, 81900lbs aircraft weight
Clean, 40000ft, 32000lbs aircraft weight
2 x AIM-54 asymmetrical payload, 10000ft, 34028lbs aircraft weight
2 x AIM-54 asymmetrical payload, 40000ft, 34028lbs aircraft weight

24
Final Thesis Report 2010, SEIT, UNSW@ADFA

Nine flights were conducted for the F-15E model in each of the above five flight conditions, with conditions
[1] and [4] being used as controls. The flight tests were conducted to determine the aircrafts thrust, lift, drag,
side force, lift to drag ratio, total coefficient of lift, total coefficient of drag as well as the aileron, elevator and
rudder deflections. These parameters will be the only ones discussed as other parameters are either dependent
on the above parameters or are not meaningful to this analysis.
The standard deviation of the analyzed parameters for the above flight conditions were compared with the
control flights. A sample of the calculated standard deviations for the above set of flight tests is included in
Appendix U. Due to the large amount of data involved in each flight test, only a sample of standard deviations
for each flight condition is included in the Appendix. The data in Appendix U indicates that the standard
deviation of parameters was inversely proportionate to both altitude and aircraft weight. Thus, it would be
advantageous for the aircraft to have a lower altitude and weight. However, if the aircraft were flying too low,
X-Plane would take into consideration wing in ground effect which would increase the lift on the aircraft (19) ,
leading to errors in results. Ground effect affects aircraft when the aircraft is at a distance from the ground less
than half its wing span (13 p. 354). Thus, while excessive, the aircraft test altitude was chosen to be at 10000ft
such that it would be out of ground effect and would be at the same altitude as several other tests.
2. Determination of the Optimum Aircraft Speed Range for X-Plane Payload Flight Tests
The graphs of the above stated parameters against
Table 13: Comparison of the Oscillation Amplitudes
true airspeed in flight conditions [1] and [4] are
for Various Speed Ranges
contained in Appendix V and W respectively. These
graphs indicate that the F-15E model will suffer from
oscillations in lift, drag and side force which increase
with velocity. This trend is steady for the F-15E model
in both the clean and asymmetrical payload condition.
These oscillations are reflected in the graphs of aileron,
elevator and rudder deflections, also contained in
Appendices V and W. However, the control surface
deflection graphs for the clean aircraft indicate
abnormal aileron deflections between velocities of
260KTAS and 300KTAS as well as abnormal elevator
and rudder deflections between speeds of 340KTAS
and 370KTAS. The control surface deflection graphs
for the F-15E under payload shows similar trends.
Also, for the F-15E in a clean condition, the rudder
deflection graphs indicate that the rudder undergoes slight oscillations at velocities below 360KTAS.
As can be seen in Table 13, the speed range with the lowest oscillation amplitudes for the clean and
asymmetrical payload condition would be between 200KTAS and 250KTAS.
Thus, considering all the above factors, the optimum test speed range was determined to be 250KTAS to
200KTAS. This would minimize unwanted oscillations in aircraft forces and control surfaces, abnormal control
surface deflections as well as ensure the aircraft is well out of the transonic and supersonic zone.
As shown in Appendix V and W, the oscillations in aircraft lift follow a horizontal trend line. These
oscillation amplitudes can thus be determined easily. The lift was not exactly equal to weight because the
aircraft pitch varies with velocity and the resultant force which acts against aircraft weight would be a
trigonometric sum of aircraft thrust and total wing lift. The average total wing lift for all the clean flight tests at
10000ft was thus approximately 32030lbs, while that of the 2 x AIM-54 asymmetric payload tests was
34040lbs. Both these total wing lift values are slightly higher than the aircraft weight in those conditions. This
is because of two factors. Firstly, the actual force acting against gravity is a trigonometric sum of aircraft lift
and thrust, due to the fact that the aircraft is not flying at exactly 00 pitch. Secondly, the aircraft is undergoing
slight oscillations in pitch which causes the wing lift, total lift and resultant lift forces to vary.
B. Repeatability Flight Tests
1. Calculation of Standard Deviation
The optimum test flights were chosen to be conducted at 10000ft, 32000lbs aircraft weight (excluding
payload) and between 250KTAS and 200KTAS. Thus, the average standard deviations for the flight test output
parameters were calculated for the two extreme conditions of the flight tests no payload and maximum
asymmetrical payload. These results are tabulated in Table 14.
As can be seen, two standard deviations is less than 2% of the mean for most measured parameters. The
standard deviation for thrust is 2.18% of the mean thrust for the F-15E with 2 x AIM-54 missiles. It was
assumed that the clean and 2 x AIM-54 payload would be extreme cases in terms of oscillations in output
25
Final Thesis Report 2010, SEIT, UNSW@ADFA

Table 14: Calculations for the Percentage Uncertainties for the Stated Flight Parameters.

parameters. Thus, other payloads would have standard deviations lower than those determined for the
aforementioned two aircraft configurations. The reason why the 2 x AIM-54 payload is considered an extreme
is not just that its weight exceeds that of a other payloads but also because it provides the greatest lateral
asymmetry (in ft-lbs).
One output parameter which needs explanation is the thrust. This is because the thrust is a negative value.
This is because the minimum possible aircraft weight was chosen 32000lbs. This is only possible with nil
fuel. Thus, the engines are not operating so are producing drag, not thrust. This leads to the negative thrust
values. This is how X-Plane calculates drag on aircraft engines it is factored into the thrust calculations.
2. Conclusion
The combination of all the above calculations indicates that the above parameter results from a single test
would be within 2.2% or population mean 95% of the time, with an associated uncertainty of 25%.
This leads to numerous possible deductions about the mathematical modeling involved in X-Plane. One
possible deduction is that there are very few possibilities or random errors occurring which may distort
experimental results. The likely reason for the uncertainty in the repeatability flight tests is that the autopilot
does not act in exactly the same manner in each flight test, leading to slight differences. Another conclusion
which can be made is that flight tests a very repeatable as long as the flight path and initial conditions are
readily controllable.
C. Payload Comparison Flight Tests
1. Expected Trends
It is expected that as lateral asymmetry increases, so
would aileron and rudder deflection. This would be
because increased lateral asymmetry would cause the
aircrafts autopilot to increase aileron and rudder
deflection to as to maintain straight and level flight.
It is also expected that lift and drag would increase
with payload. This is because increased payload means
that total lift would have to increase to maintain straight
and level flight. However, total lift would be a
trigonometric addition of wing and horizontal stabilizer
lift and thrust. Since the aircraft is in straight and level
flight, drag would be acting purely horizontally.
Fortunately, this total lift value is the value provided in
the X-Plane data output file so the total lift can be easily
charted.

Figure 23: Indication of Lateral Asymmetry Effects

2. Analysis of Flight Test Results


The obtained trends for the different payload flight
tests were generally as expected. The plots of the
obtained results are contained in Appendix X. Aileron
and rudder deflection generally increases with lateral
asymmetry as shown in Fig 23 and 24 respectively.
Only the tests making use of the 2xAIM-54, 2xAIM-7,
Figure 24: Indication of Lateral Asymmetry Effects
1xAIM-54 and 1xAIM7 have a significantly different
trend to the others. As expected, the above four stated payloads are the payloads which result in the largest
lateral asymmetry, as indicated in Table 2. The same four stated payloads are also the only ones significantly
26
Final Thesis Report 2010, SEIT, UNSW@ADFA

different from the rest in terms of rudder deflection. However, the 2xAIM-7 payload is the one with the largest
rudder deflection instead of the 2xAIM-54 payload. This is unexpected because the 2xAIM-54 payload has a
larger lateral asymmetry than the 2xAIM-7 payload. The reason for the aircrafts deviation from the expected
trend is unknown and retests indicate similar results. This could be due to a postulated complex relationship
between lift, drag, pitch and control surface deflection and would be able to be pinpointed by use of further
testing.
The side force on the F-15E did not show any sort of discernible trend. This is contrary to what was
expected. However, since the rudder deflection is so minimal, it is possible that the effect of payload on side
force cannot be detected.
Lift and drag showed mixed trends. As expected,
aircraft lift increases with payload weight. This can
be seen in Fig. 25. Aircraft drag also increases with
payload weight and size.
The 2xAIM-54
configuration clearly has more drag than the 1xAIM54 configuration. This is to be expected as X-Plane
does drag calculations on the payloads.
As velocity increases, pitch decreases and thus
drag decreases due to the decrease in frontal area and
induced drag. However, when landing gear is
Figure 25: Indication of Lateral Asymmetry Effects
extended, aircraft drag increases with velocity. This
is because when landing gear is extended, the
increase of drag on the landing gear far outweighs
the decrease in drag due to decrease pitch and is
visually represented in Fig. 26.
The GUI for creating payloads has a body
coefficient of lift input. As with other bodies, the
designer must include interference, skin friction and
pressure drag in this value before the total drag force
is calculated by Eq. 3. This is the only way X-Plane
determines the interference effect of the payload on
the lift over the wing directly above it.
Figure 26: Indication of Lateral Asymmetry Effects
It can be seen that X-Plane does simulate
changes in the F-15E aircraft rather well in terms of
trending correctly. Unfortunately, quantitative analysis could not be completed for these flight tests in this
thesis due to the lack of data to which results could be compared.
D. Conclusion
The flight tests conducted for this section of the thesis have shed light on numerous aspects of X-Plane.
With the right flight conditions and number of test flights, the uncertainty in a result can be reduced to a
minimal value. The best flight condition to conduct flight tests in was determined to be a simple level flight
deceleration test between the velocities of 250KTAS and 200KTAS.
The desired aim of this thesis was to produce an F-15E X-Plane model which could show the effect of
changes to an aircraft. This change was modeled as increasing lateral asymmetry. The above analysis proves
that lateral asymmetry on the aircraft model does indeed produce expected trends. The effect of an alteration to
the F-15E airframe can also be tested. This would likely have to be conducted by adding a miscellaneous body
to the F-15E aircraft, taking note that an accurate coefficient of drag must be known beforehand. Only then can
accurate results be obtained.
One important point to note is that the X-Plane model suffers from slight oscillations at all flight conditions.
This is because of the time required to complete one cycle of calculations. Once one cycle is completed, the
flight parameters would have changed slightly, leading to the need for the autopilot to recorrect the aircrafts
flight path. Even if a supercomputer were used, oscillations would still exist. The only difference would be that
the magnitude of oscillations would be tiny in comparison. The oscillations can be seen in the flight output data
charts contained in this thesis paper as well as in the appendices. In order to improve the precision of output
data, this issue can be addressed with the use of a more powerful computer. The computer used for these flight
tests has an Intel Core i3 CPU processor, clocked at 3.07Ghz with 3.87 GB of usable RAM.

27
Final Thesis Report 2010, SEIT, UNSW@ADFA

XI. Final Comments


The above described flight tests gave valuable insight as to the accuracy of the F-15E model as well as the
ability of the X-Plane software to correctly mimic real world physics. Any inaccuracies with regards to the
tested flight characteristics have been determined. These are summarized below and the results of further
improvements have been explained too. However, there are still many possibilities to improve on the designed
F-15E model. Numerous other flight tests can also be conducted so as to analyse the models characteristics in
other modes of flight, such as the spiral spin or sustained level flight turn.
A. Issues with F-15E Model
There were numerous issues discovered with the designed F-15E model. These issues included excessive
thrust, excessive supersonic drag, and substantial differences between obtained short period characteristics to
expected values. Alterations were made to the F-15E model to attempt to lessen the issues and the results of
these improvements are contained in the following section.
B. Method to Improve the F-15E Model after Analysis of Test Results
Much consideration needed to be taken into account before attempting to improve the F-15E model. Firstly,
the obtained stall speeds were already extremely compliant with flight manual data, indicating that wing and
empennage dimensions, as well as airfoil lift curve slopes were fairly accurate. Thus, it would be important not
to change those parameters as that would cause the stall speeds to deviate from official values.
The following method was decided upon to improve the aircrafts maximum level speed and acceleration
characteristics. The drag settings would be corrected first, followed by the thrust settings. This was because the
aircrafts thrust settings would be altered by observing aircraft acceleration which is dependent on drag but drag
settings would be corrected by observing level flight deceleration.
It was hypothesized that the F-15E X-Plane model had two components which had excessive drag the
connection of the empennage to the fuselage and the wings. Thus, an altered X-Plane model would be created
which had the minimum drag setting on each of those components. If this led to a large decrease in drag
coefficient, it would most likely be the incorrectly modeled component. If a large decrease in drag coefficient
was not obtained, other components would be tested. If the nullification of each single components drag did
not lead to any components contributing excessive drag to the aircraft in the supersonic region of flight, then all
components coefficients of drag would be lowered and the aircraft would be tested again.
The X-Plane models thrust settings could then be corrected. For speedy testing, only the lightest and
heaviest aircraft weights would be tested. The aircraft (with corrected drag characteristics) would be accelerated
at 10000ft, military thrust and the thrust at 100% low compressor speed would be altered such that the correct
acceleration at 10000ft would be obtained. Once this is completed, the aircraft would be accelerated at 10000ft,
maximum thrust and its additional afterburner thrust would be edited so that its maximum thrust at 10000ft was
correct. Finally, the aircraft would be accelerated at 40000ft, maximum thrust and its critical altitude would be
altered so as to achieve the correct accelerations.
Once the above testing and corrections are completed, the aircrafts maximum level flight speeds would be
checked and the aircraft components drag values would be finely adjusted so that the maximum level flight
speeds would be correct.
C. Results of Attempt to Improve F-15E Model
Unfortunately, attempts to improve the drag characteristics of the aircraft were unsuccessful. The above
method was used and even with the minimum drag setting for every single aircraft component in the F-15E
model, the aircraft could neither exceed Mach 1 at 10000ft nor reach Mach 2.4 at 40000ft and 42800lbs aircraft
weight. The airfoils on the wings were even changed to have near zero drag for angles of attack between -200
and 200 but even this did not allow the aircraft to pass the speed of sound at 10000ft. The airfoils Mach
divergent drag number was also increased to no avail.
Thus, it can be seen that with even minimal drag settings chosen on every single component in the aircraft,
the X-Plane software still determines that the aircraft is unable to fly supersonic. This would indicate that the
X-Plane softwares transonic and supersonic drag calculations are rather inaccurate as the simulated aircraft
drag is far too high in these regions of flight. As stated in Section IV. F, X-Planes transonic and supersonic
modeling is makes a number of simplifying assumptions and approximations. This is likely the cause of the
inaccuracies of transonic and supersonic modeling.
Due to the fact that the issue of aircraft drag could not be reconciled, the aircraft thrust was not altered.
D. Future Work
Further work can be conducted on the F-15E model to further improve the aircrafts accuracy. This can be
facilitated by the conduct of additional flight tests which would indicate the necessary edits required on the F15E model. Additional flight tests to determine the F-15Es takeoff distances, rotation speeds, single engine
28
Final Thesis Report 2010, SEIT, UNSW@ADFA

rates of climb, climb performance, descent performance and sustained level flight turn performance would be
ideal. While the flight manual contains data on other flight characteristics, the above stated characteristics are
the easiest to compare flight test results to.
Flight characteristics not mentioned in the flight manual can also be tested with the use of theoretical
predictions. The aircrafts static longitudinal stability, dynamic lateral-directional stability and maneuvering
stability are a few characteristics which can be predicted by the use of (4). The directions for the correct
conduct of all the above mentioned flight tests are contained in (52).
Use of DATCOM will provide another avenue to which stability characteristics can be compared.
Knowledge of FORTRAN would enable one to use the DATCOM source code. Since slight alterations need to
be made to the aircraft before DATCOM can be used, such as the use of a single vertical tail as opposed to two,
an aircraft should be designed in X-Plane making use of the exact dimensions used in DATCOM. This would
ensure that the DATCOM outputs are valid for the F-15E.
With enough resources, CFD analysis of the F-15E could be conducted for more accurate determination of
forces and moments on the aircraft. This would provide a benchmark to which the payload comparison
deceleration test results can be compared such that those flight test results can be verified. The reason why it is
preferable to use X-Plane over CFD to determine the forces, moments and other characteristics of the F-15E in
flight is that X-Plane takes a much shorter time to do the essential calculations. CFD can be used to ensure that
the input parameters are correct and X-Plane can then be used to do the bulk of the calculations. However, if
CFD were to be used, it would be imperative to obtain much more accurate measurements of the F-15E. This
would be so as to achieve the best possible correlation between the X-Plane and CFD models and the real
aircraft.
Further study into possible flight conditions for precise measurements should be taken. This would enable
the F-15E model to be suffering from the least possible amount of oscillations. This would increase the
precision of data collected from X-Plane.
E. Conclusion
A large amount of research has been conducted thus far to ensure accuracy and validity of results. Future
work should focus on improvement of the F-15E model as well as verification of its abilities. The best possible
way to verify the F-15E model would be to make use of either CFD or actual F-15E flight test data. Even if the
X-Plane aircraft model is not extremely accurate, the delta method can be used to determine the effect of
changes to the aircraft. However, X-Plane has proven in this thesis to be a rather accurate tool in predicting the
flight characteristics of an aircraft if its input parameters are correct. This is seen in the very close correlation of
obtained flight test stall speeds to flight manual data, as well as the correlation of maximum level flight
acceleration and speeds to flight manual data once the aircraft was improved. This would lead to X-Plane being
a powerful prediction tool in determining the approximate result of a change to an aircraft, with more precise
analysis methods being used to focus on certain scenarios which may be been determined through use of XPlane.
Acknowledgements
This thesis would not have been possible without the help of my thesis supervisor Andrew Neely.
Throughout this thesis, he has been a great help on this thesis and a source of many fantastic ideas for
improvement. I would also like to thank Anthony Booher, who not only provided me with privately designed
aircraft models which are unavailable online but also aided me in understanding X-Plane and Javafoil. Without
him, my aims would have been much harder to meet. Lastly, I would like to thank Austin Meyer. One would
not expect the creator of such a well selling piece of software to have the time to reply an endless stream of
emails about the mathematics behind X-Plane but Austin has proved to be an exception to this rule. To all
others who have helped me along the way thank you.

29
Final Thesis Report 2010, SEIT, UNSW@ADFA

References
1. Williams, B. Microsoft Flight Simulator as a Training Aid. Newcastle, Western Australia : Aviation Supplies
& Academics, Inc., 2006.
2. Raymer, D. P. Aircraft Design: A Conceptual Approach (3rd Edition). Reston, Virginia : AIAA, 1999.
3. Roskam, J. Airplane Design Part VI: Preliminary Calculation of Aerodynamic, thrust and Power
Characteristics. Lawrence, Kansas : DAR Corporation, 2000.
4. . Airplane Design Part VII: Determination of Stability, Control and Performance Characteristics; FAR and
Military Requirements. Lawrence, Kansas : DAR Corporation, 2000.
5. Hoak, D.E. USAF Stability and Control DATCOM, Volumes 1 to 4. Irvine, CA : Global Engineering
Documents, 1978.
6. Williams, J. E., Vukelich, S. R. The USAF Stability and Control Digital DATCOM. St. Louis, Missouri :
USAF, 1979. AFFDL-TR-79-3032.
7. Holy Cows Inc. DATCOM+. [Online] Mar 17, 2010. [Cited: Mar 28, 2010.]
http://www.holycows.net/datcom/.
8. Hoak, D.E. USAF Stability and Control DATCOM. Dayton, Ohio : Flight Control Division, Air Force Flight
Dynamics Laboratory, 1978.
9. Holy Cows Inc. DATCOM+, Software Package, Ver 2.6c. Holy Cows, Inc. [Online] Mar 17, 2010. [Cited:
Mar 20, 2010.] http://www.holycows.net/datcom/.
10. Matyushev, A. OpenDATCOM, Software Package, Ver 1.1. openAE. [Online] Jan 5, 2010. [Cited: Feb 8,
2010.] http://www.openae.org/software.
11. OpenDatcom. OpenDatcom. Open Aerospace Software Commmunity. [Online] January 05, 2010. [Cited:
March 25, 2010.] http://openae.org/software.
12. Raymer, D. P. RDS-Student Software. Reston, Virgina : AIAA, 2006.
13. . RDS-Student Manual. Reston, Virginia : AIAA, 2006.
14. Wendt, J. F. Computational Fluid Dynamics: An Introduction (3rd Edition). Berlin, Germany : Springer,
2008.
15. Olson, C. L. FlightGear Ver 2.0, Software Package. FlightGear. [Online] 2010. [Cited: Mar 18, 2010.]
http://www.flightgear.org/.
16. . The FlightGear Manual. FlightGear. [Online] Feb 25, 2010. [Cited: Mar 15, 2010.]
http://www.flightgear.org/Docs/getstart/getstart.html.
17. Berndt, J. S. JSBSim Reference Manual. JSBSim. [Online] Mar 20, 2010. [Cited: Mar 28, 2010.]
http://jsbsim.sourceforge.net/JSBSimReferenceManual.pdf.
18. . JSBSim: An Open Source Flight Dynamics Model in C++. Providence, Rhode Island : AIAA Modeling
and Simulation Technologies Conference and Exhibit, Aug 16-19, 2004. Vols. AIAA 2004-4923, pp. 1-27.
19. Laminar Research. X-Plane 9 Operation Manual. s.l. : Graphsim Entertainment, 2008.
20. HHS. YASim helicopter model?? FlightGear Forum. [Online] Feb 26, 2008. [Cited: Apr 22, 2010.]
http://www.flightgear.org/forums/viewtopic.php?f=4&t=1049.
21. Away, Fly. Boost your Flight Simulator Experience with FSX Aircraft Add-ons. Fly Away. [Online] April
14, 2010. [Cited: May 2010, 13.] http://flyawaysimulation.com/article3983.html.
22. fsdownload. Flight Simulator X Downloads Directory for FSX FS2004 FS2002 FS2000 FS98 FS95.
www.fsdownload.com.
[Online]
Feb
23,
2008.
[Cited:
June
17,
2010.]
http://www.fsdownload.com/modules.php?name=News&file=article&sid=891.
23. Beckwith, J. Mudpond. Mudpond. [Online] 2010. [Cited: Apr 8, 2010.] http://www.mudpond.org/.
24. Sors, H. FS Airfile Editors and Design. Flight Dynamics and Navdata. [Online] 2010. [Cited: Apr 8, 2010.]
http://pagesperso-orange.fr/hsors/fsairfile.html.
25. Microsoft Flight Simulato. About the Aces Team. Microsoft Flight Simulator. [Online] Jan 26, 2009.
[Cited:
Apr
10,
2010.]
http://www.microsoft.com/Products/Games/FSInsider/news/Pages/AMessageFromAces.aspx.
26. Laminar Research. Appendix A: How X-Plane Works. X-Plane Desktop Manual. [Online] 2010. [Cited:
Mar 20, 2010.] http://wiki.x-plane.com/Appendix_A:_How_X-Plane_Works.
27. . X-Plane. [Online] 2010. [Cited: Apr 29, 2010.] http://www.x-plane.com/index_pro.html.
28. Taureau, N. Home of the X-Plane Flight Simulator Community. X-Plane.org. [Online] 2010. [Cited: Apr
10, 2010.] http://forums.x-plane.org/.
29. Ross, D. Structural Analysis of a Cessna 182 Wing During Manoeuvres using X-Plane Flight Simulator.
Canberra, Australia : UNSW@ADFA, 2008.
30. Laminar Research. Plane Maker, Software, v9.50rc.
31. . Supplement: Airfoil-Maker. X-Plane Desktop Manual. [Online] 2010. [Cited: Apr 1, 2010.]
http://wiki.x-plane.com/Supplement:_Airfoil-Maker.
32. Meyer, A., HansVanKampen. Plane-Maker Menus. X-Plane On-Line Instruction Manual. [Online] 5 2002.
[Cited: 5 20, 2010.] http://roust.gotdns.com/instructions/Manual_Files/Plane-Maker%20Menus.html.
30
Final Thesis Report 2010, SEIT, UNSW@ADFA

33. Harris, J. X-Plane Scenery Tools. X-Plane Scenery. [Online] 2009. [Cited: Mar 14, 2010.]
http://marginal.org.uk/x-planescenery/tools.html.
34. Jane's all the World's Aircraft. Boeing F-15E Eagle. Jane's. [Online] Feb 26, 2010. [Cited: Mar 20, 2010.]
http://www.janes.com, search for "F-15E".
35. Hepperle, M. Javafoil, Software Package, Ver 2.13. Javafoil. [Online] Mar 1, 2007. [Cited: Apr 12, 2010.]
http://www.mh-aerotools.de/airfoils/jf_applet.htm.
36. . Comparison between Theory and Experiment. Javafoil. [Online] Oct 15, 2007. [Cited: Aug 11, 2010.]
http://www.mh-aerotools.de/airfoils/methods.htm#ComparisonTheoryExperiment.
37. Selig, M.S., Guglielmo, J.J., Broeren, A.P., Giguere, P. Summary of Low-Speed Airfoil Data, Vol. 1.
Virginia Beach, VA : SoarTech Publications, 1995.
38. Selig, M.S., Lyon, C.A., Gigure, P., Ninham, C.N., Guglielmo, J.J. Summary of Low-Speed Airfoil Data,
Vol. 2. Virginia Beach, VA : SoarTech Publications, 1996.
39. Lyon, C.A., Broeren, A.P., Gigure, P., Gopalarathnam, A., Selig, M.S. Summary of Low-Speed Airfoil
Data, Vol. 3. Virginia Beach, VA : SoarTech Publications, 1998.
40. UIUC LSATs Airfoils Tested. UIUC Applied Aerodynamics Group. [Online] 2010. [Cited: August 11,
2010.] http://www.ae.illinois.edu/m-selig/uiuc_lsat_airfoilsTested.html.
41. Stock, C. Flight Dynamics: FSX and X-Plane Battle it out. Sim Pilot Magazine. [Online] 2010. [Cited: Apr
5, 2010.] http://www.simpilotnet.com/index.php?option=com_content&task=view&id=20&Itemid=9.
42. McDonnell Aircraft. USAF Series F-15E Aircraft Flight Manual. U.S.A. : McDonnell Aircraft, 1991. TO
1F-15E-1.
43. Jenkins, D. R. McDonnell Douglas F-15 Eagle. North Branch MN USA : Specialty Press, 1997.
44. Hoerner, S. F. Fluid-Dynamic Drag. Bakersfield, CA, USA : Hoerner Fluid Dynamics, 1992.
45. Cohen, E. Israel's Best Defense. New York : Orion Books, 1993.
46. National Aeronautics and Space Administration. EngineSim Version 1.7a. NASA. [Online] Oct 25, 2006.
[Cited: July 18, 2010.] http://www.grc.nasa.gov/WWW/K-12/airplane/ngnsim.html.
47. Jane's Aero-Engines. Pratt & Whitney F100. Jane's. [Online] Jul 10, 2009. [Cited: Mar 27, 2010.]
http://www.janes.com, search for "F100-PW-220".
48. Austin, D. , Booher,A. , Leger B. F-15E Strike Eagle, v9.1. X-Plane.org. [Online] Feb 3, 2009. [Cited: Feb
14, 2010.] http://forums.x-plane.org/index.php?autocom=downloads&showfile=7090.
49. Meyer, A. , Creator of X-Plane. Electronic Mail Correspondence. 2010.
50. Williams, J.E., Vukelich, S.R. The USAF Stability and Control Digital Datcom. Ohio : United States Air
Force, 1979. AFFDL-TR-79-3032, Vol I.
51. Smith, H. C. Understanding Performance flight Testing (2nd Edition). Hightstown, New Jersey : McGrawHill, 2002.
52. Kimberlin, R. D. Flight Testing of Fixed Wing Aircraft. Reston, Virginia : AIAA, 2003.
53. Specification, U.S. Military. MIL-F-8785C. s.l. : U.S. Military, 1980.
54. Mendenhall, W. , Sincich, T. Statistics for Engineering and the Sciences. Upper Saddle River, New
Jersey : Pearson Prentice Hall, 2007.
55. Pointwise .Inc. F-15E Spin Analysis. Pointwise .Inc. [Online] 2010. [Cited: Mar 28, 2010.]
http://www.pointwise.com/apps/f15e.shtml.
56. Jane's Aero-Engines. General Electric F110. Jane's. [Online] May 29, 2009. [Cited: Mar 27, 2010.]
http://www.janes.com, use the "main search" to search for "F110-GE-129".
57. Laminar Research. Airfoil Maker, Software Package, Ver 9.50rc3. 2010.
58.
Flickr.com.
Flickr.com.
[Online]
[Cited:
July
02,
2010.]
http://farm4.static.flickr.com/3428/3800915982_92a7db0bb0.jpg.
59. SRA D. M. CULLEN, USAF. DefenseImagery.Mil. [Online] Apr 30, 2001. [Cited: Feb 20, 2010.]
http://www.dodmedia.osd.mil/DVIC_View/Still_Details.cfm?SDAN=DFSD0405951&JPGPath=/Assets/Still/2
004/Air_Force/DF-SD-04-05951.JPG.
60. Abbot, I. H. , von Doenhoff,A. E. , Stivers Jr,L. S. Report No. 824 - Summary of Airfoil Data. Washington
D.C. : NACA, 1945.

31
Final Thesis Report 2010, SEIT, UNSW@ADFA

Vous aimerez peut-être aussi