Beruflich Dokumente
Kultur Dokumente
Kannan Perumal
Postgraduate Research Student, Airline Pilot
p_kannan@singaporeair.com.sg
Abstract
This paper provides a brief history of the development of a PC-based flight dynamics engineering
simulator facility in Department of Aeronautical Engineering, Universiti Teknologi Malaysia (UTM-
FDES). The FDES is a simple, easy to use, flexible and user friendly aircraft simulation through the
implementation of the classical longitudinal and lateral aircraft equations of motion. FDES allows
students to rapidly assess aircraft stability and control parameters and able to give visual
representation of the effect of changing the stability derivatives. Currently it is utilized in the flight
dynamics and control class for demonstration, research and understanding the basic concepts of flight
dynamics, and flying and handling qualities. It has been an effective and efficient teaching and
learning tools of basic principles of aircraft dynamics and flight control systems evaluations in a
relatively short time.
Keywords : Flight Dynamics, Simulation, flying and handling qualities, flight control
1.0 Introduction
Flight simulator is a device that is used to simulate the behaviour of an aircraft in
flight on the ground. The essence of it is in creating an illusion of reality of an
aircraft in flight on ground by which expensive flying time and costs are saved. In
has been widely used in the aerospace and aviation industries for commercial and
military pilot training, test pilot and flight research activities.
Keyboard Input
Equation of Computer
Motion Graphics Output
Joystick Input
As its name implies, the objective of flight simulation is to reproduce on the ground
the behaviour of an aircraft in flight. The practical value of flight simulation is
obvious with the extensive use of the technique in aerospace research and
development and by the fact that many flight simulators are in use throughout the
world, for training and maintaining the skills of civilian and military aircrew.
For a vehicle flying in the air, the mathematical model is primarily the relationship
between the air reactions and the motion of the aircraft relative to the air. Thus, this
can be called aerodynamic model. Other external forces and moments arise from
engine thrust, the flight controls or external (atmospheric) input such as gust and
disturbances. Definitions of all these forces and moments components are the key
realistic description of an aircraft’s flight characteristics. Figure 2 shows the detail
elements of aircraft mathematical model.
Performance Dynamics Atmosphere Turbulence
Engine Environment
Ground effect
Mathematical Aircraft
Wings (rotor) Aerodynamics Pilot
Model response
Tail
Undercarriage Control system
The initial development of the first prototype for the flight simulator in the Faculty of
Mechanical Engineering, UTM started in mid 1995. The flight simulator was
developed and enhanced through five final year students as part of their final year
project. Many improvements and new features have been introduced in each project.
The project development was systematically monitored and supervised in order to
ensure the success of the project.
Phase I
The project was first started by simulating a simple Head-Up-Display (HUD)
displaying a ‘gull’ shape aircraft symbol incorporating the basic speed, altitude and
heading information. HUD is a projection that allows the pilot to take information
from the instruments without taking eye-off from the outside scene. This technology
is widely used in the fighter aircraft. The first part of it was to integrate the hardware
(a contacless joystick) with the software (Turbo Pascal 7 programming). Simple
longitudinal equations of motion were used to represent the dynamics of the aircraft.
Then, a simple mathematical model for actuator dynamics and control law was
included [2].
Phase II
The main concentration of this phase was to improve the aircraft equation of motion
by improving the aerodynamic modelling and adding different aircraft type. There
are six aircraft models in the program database to be simulated. This function will
allow student to experience the flying and handling qualities of different type of
aircraft, such as A4D Skyhawk, F104-A, NAVION, Convair 880, Boeing 747 and
Jetstar. There are six types of aircraft available in the program database. The main
purpose of this type of simulation is to allow the user to feel the flying and handling
of different kind of aircraft [3].
The simulation is done using the stability and aerodynamic variable of the aircraft
itself [7]. From there, student will be able to understand how the stability derivatives
of the aircraft affect the flying and handling qualities of an aircraft. Thus, it is an
inherent flying of the aircraft.
The computer codes were made to a modular-type for the ease of management and to
be more effective. Capability to use the keyboard as the inceptor to fly the aircraft
was also added. Graphics output of the program has been reviewed and improved
where the program will run and display HUD graphics with almost the same speed
no matter what is the CPU clock speed. Improvement on the HUD display itself was
made to a more standard form of HUD symbolic (Figure 3).
30 60 90
60
Phase III
The contactless joystick input was replaced to a normal game joystick. This made the
program more users friendly. Real time simulation was attempted by having a
reference to CPU clock speed. A Head-Down-Display (HDD) was designed, and
meshed with the previous Head-Up-Display (HUD) to create a realistic flight
simulator (Figure 4). Several other indicators such as gyroscope and roll and pitch
angle were added. Rudder input was added to give more realistic response to
lateral motions. Variable stability aircraft model was included to allow user to
modify the existing aircraft parameters. The flight dynamics control system was
designed to augment the aircraft response, this include the simple pitch rate feedback
and PI controller pitch rate error as well as roll rate damper.
The second function is the simulation using user-defined variable stability (Figure 5).
This is where the users can fly the aircraft created by their own. The meaning of
created here does not mean a real aircraft is created, it means the aircraft coefficients
and certain dimensions about the aircraft. There are 20 stability coefficients can be
inserted by the user in the interface. Another purpose for this type of simulation is
that, the user could know what would be the outcome towards the aircraft flying
qualities if one of the coefficients is modified. This function is only recommended
for advanced user who is expected to have understood the aircraft variable stability
well. The derivatives for longitudinal and lateral are editable and this will result the
change of aircraft motion. With this, student can see the effect of aircraft motion
when any of the derivatives is changed [4].
Figure 4. Main simulation screen
Phase IV
A new source code for the joystick input was developed in Pascal to give a better
control in setting the environment for the joystick hardware interface. The real time
simulation was further improved so the program will run at a constant speed
regardless of the CPU speed and also the intensity of the graphics [5].
As the flight simulator was aimed to be a teaching and learning tool, modification
was done to the program to provide better guidance to guide the student on the
handling of the program. At this stage the simulator was complete with four degrees
of freedom using linearized aircraft equation of motion. At this stage it allowed the
study on the short period and the phugoid motion with the integral controller,
proportional controller and rate control was to be conducted. The third function is the
simulation using flight control system. Student can choose from the menu for the
desired flight control system that to be applied on the aircraft. At the same time,
student may also alter the stability of the aircraft by inserting suitable values for the
control gain, damping ratio and natural frequency. There are six flight control
systems where student can choose to evaluate. They are:
At this stage a complete FDES main program flow chart has been developed shows
in Figure 6.
BEGIN
Display type
menu
Main Menu
Inceptor
Choice
1 = A4D
3 = NAVION
Main Menu
4 = Convair 880
6 = Jetstar
HDD Menu
ASI+Altimeter+Turn
No Indicator ASI + Altimeter indicator+Heading
All Indicator
Choice = 6 Choice = 2
Simulation
Center
Choice = 3 Choice = 1
Choice = 7
END
Some inaccuracies in the equations of motion, numerical solution, input and output
were also corrected. For example, in the previous programs, the simulation of the roll
motion is found to be stable but the steady state oscillating periodically with ±25
deg/sec was not giving the desired output. The equations of motion was reviewed
and improved to include more detailed derivatives to obtain a better solution. The
initial Euler solution was replaced to be more accurate using the Rungga-Kutta
numerical solutions. Beside that the graphics was further enhanced to look more
realistic with actual aircraft. The symbology, colouring and finer details of the
presentation were emphasized. To verify the accuracy of the simulation, a
comparison between simulation results and theory using MATLAB was carried out.
Finally a more user friendly graphic User Interface (GUI) was created for the aircraft
selection and input and output of various parameters. In the previous program,
Euler’s Method is used to solve the aircraft mathematical model. Testing was done to
compare the output values obtained from the program with the theory values. The
theory values were obtained from calculation using MATLAB (Figure 7 and 8).
q vs t
120
80
60
q (deg/s ec )
40
20
-20
0 2 4 6 8 10 12 14 16 18 20
Tim e (s ec )
Figure 7: Pitch rate response for A4-D Skyhawk using Euler’s Method, time step 0.04s.
Solid line – Simulator
Dashed line – MATLAB
Figure 8: Comparison of roll rate output using (a) 0.1s, (b) 0.04s and (c) 0.01s
time step for Euler’s Method
The equations of motion used in the FDES are a set of linearized equation derived
using the small perturbation theory. In applying the small perturbation theory, it is
assumed that the motions of the aircraft consist of small deviations about a steady
flight condition. Although this theory cannot be applied to problems in large
amplitude of motion, it yields sufficient accuracy for practical engineering purposes.
The simplicity the equations of motion can be separated into two groups -
Longitudinal motion and lateral motion with appropriate assumptions in the form of
state space equation without coupling between the both longitudinal and lateral
derivatives [7].
Longitudinal motion:
⎡Δu& ⎤ ⎡ X u Xw 0 − g ⎤ ⎡Δu ⎤ ⎡ X δe ⎤
⎢Δw& ⎥ ⎢
⎢ ⎥ = ⎢ Zu Zw u0 0 ⎥⎥ ⎢⎢Δw⎥⎥ ⎢⎢ Z δe ⎥⎥
⎢Δq& ⎥ ⎢ 0
+ [Δδ e ] (2.1)
Mw Mq 0 ⎥ ⎢Δq ⎥ ⎢ M δe ⎥
⎢ ⎥ ⎢ ⎥⎢ ⎥ ⎢ ⎥
⎢⎣Δθ& ⎥⎦ ⎣ 0 0 1 0 ⎦ ⎣Δθ ⎦ ⎣ 0 ⎦
Lateral motion:
⎡Y Yp ⎛ Y ⎞ g cos θ 0 ⎤ ⎡ Yδr ⎤
⎡Δβ& ⎤ ⎢ β − ⎜⎜1 − r ⎟⎟ ⎥ ⎡Δβ ⎤ 0
⎢ ⎥ ⎢ u0 u0 ⎝ u0 ⎠ u 0 ⎥ ⎢ ⎥ ⎢⎢ u0 ⎥
⎥ ⎡Δδ ⎤
⎢Δp& ⎥ = ⎢ L Lp Lr
Δp
0 ⎥ ⎢ ⎥ + ⎢ δa
L
Lδr ⎥ ⎢ a ⎥ (2.2)
⎢Δr& ⎥ ⎢ β ⎥ ⎢ Δ ⎥ Δδ
0 ⎥ ⎢ ⎥ ⎢ δa N δr ⎥ ⎣ r ⎦
r N
⎢ ⎥ ⎢N β Np Nr
⎢⎣Δφ& ⎥⎦ ⎢ Δφ ⎢ 0 ⎥
⎣ 0 1 0 0 ⎥⎦ ⎣ ⎦ ⎢⎣ 0 ⎥⎦
Numerical method was used to perform the task of integrating the first order
differential equation and produce the desired output. There are several numerical
analysis methods that can be used such as Euler’s, Runge-Kutta, Milne’s,
Hamming’s [6]. All these methods are varying in accuracy, complexity and error
produced, depends on the requirement of the user. Initially the Euler’s numerical
method was used. The general equation given by Euler’s Method is:
y1 = y 0 + hf 0 (2.3)
where h is the constant step size, or the integral of time, dt used in the simulation.
Longitudinal motion:
u = u 0 + ( X u u 0 + X w w0 − gθ 0 ) ⋅ dt (2.4)
w = w0 + (Z u u 0 + Z w w0 + u 0 q 0 + Z δeδe ) ⋅ dt (2.5)
(
q = q 0 + M w w0 + M q q0 + M δeδe ⋅ dt ) (2.6)
θ = θ 0 + q0 ⋅ dt (2.7)
Lateral motion:
β = β 0 + [(Yβ / u 0 )β 0 + (Y p / u 0 ) p0 − (1 − (Yr / u 0 ) )r0 + (g cos θ 0 / u 0 )φ0 + (Yδr δr / u 0 )]⋅ dt (2.8)
(
p = p0 + Lβ β 0 + L p p 0 + Lr r0 + Lδa δa + Lδr δr ⋅ dt ) (2.9)
(
r = r0 + N β β 0 + N p p 0 + N r r0 + N δa δa + N δr δr ⋅ dt ) (2.10)
φ = φ 0 + p 0 ⋅ dt (2.11)
ψ = ψ 0 + r0 ⋅ dt (2.12)
Euler’s method was later replaced by the use 5-stage 4th order Explicit Runge-Kutta-
Mason Method to obtain better accuracy with a times step of 0.1s. The 5-stage 4th
order Explicit Runge-Kutta-Mason Method employs the recurrence formula of the
form:
5
yn +1 = yn + h ∑ bi k i
i =1
where
k i = f ⎛⎜ t n + ci h, y n + h ∑ aij k j ⎞⎟
5
(2.14)
⎝ j =1 ⎠
1
a 21 =
3
1 1
a31 = , a 32 =
6 6
1 3
a 41 = , a 42 = 0 , a 43 =
8 8
1 3
a51 = , a 52 = 0 , a53 = − , a54 = 2
2 2
1 2 1
b1 = , b2 = 0 , b3 = 0 , b4 = , b5 =
6 3 6
1 1 1
c1 = 0 , c2 = , c3 = , c4 = , c5 = 1
3 3 2
h = time step
Since the mathematical model for each variable is too complex to be displayed here,
only the formulation for forward speed, u is shown.
k u ,1 = X u ⋅ u 0 + X w ⋅ wo − g ⋅ θ o
k u , 2 = X u ⋅ (u 0 + h ⋅ a 21 ⋅ k u ,1 ) + X w ⋅ (wo + h ⋅ a 21 ⋅ k w,1 ) − g ⋅ (θ o + h ⋅ a 21 ⋅ kθ ,1 )
k u ,3 = X u ⋅ (u 0 + h ⋅ a31 ⋅ k u ,1 + h ⋅ a32 ⋅ k u , 2 ) + X w ⋅ (wo + h ⋅ a31 ⋅ k u ,1 + h ⋅ a32 ⋅ k w, 2 )
− g ⋅ (θ o + h ⋅ a31 ⋅ kθ ,1 + h ⋅ a32 ⋅ kθ , 2 )
k u , 4 = X u ⋅ (u 0 + h ⋅ a 41 ⋅ k u ,1 + h ⋅ a 43 ⋅ k u ,3 ) + X w ⋅ (wo + h ⋅ a 41 ⋅ k u ,1 + h ⋅ a 43 ⋅ k w,3 )
− g ⋅ (θ o + h ⋅ a 41 ⋅ kθ ,1 + h ⋅ a 43 ⋅ kθ ,3 )
k u ,5 = X u ⋅ (u 0 + h ⋅ a51 ⋅ k u ,1 + h ⋅ a53 ⋅ k u ,3 + h ⋅ a54 ⋅ k u , 4 )
+ X w ⋅ (wo + h ⋅ a 51 ⋅ k u ,1 + h ⋅ a53 ⋅ k w,3 + h ⋅ a54 ⋅ k w, 4 )
− g ⋅ (θ o + h ⋅ a51 ⋅ kθ ,1 + h ⋅ a53 ⋅ kθ ,3 + h ⋅ a54 ⋅ kθ , 4 )
u = u o + hb1k u ,1 + hb2 k u , 2 + hb3 k u ,3 + hb4 k u , 4 + hb5 k u ,.5
The execution times and accuracies are the two main important factors in the flight
simulator. They directly affects the performance FDES. For higher accuracy (real
time), smaller time step should be used, however this reduces the execution time. As
such a good compromise between both is required. Time step of 0.1s using the
Rungga-Kutta method was found to give better accuracy the 0.04s using the Euler
method. Figure 10 and 11 show the effect of numerical solutions and time steps on
simulation results.
Figure 10: Comparison of roll rate between Runga-Kutta 4th and 5th order for time step 0.04s
and input 00.2618 rad
p vs t (RK 5, tim e step 0.1s) p vs t (RK 5, tim e step 0.2s)
4 4
3 3
2 2
1 1
0 0
0 2 4 6 0 2 4 6
3 3
2 2
1 1
0 0
0 2 4 6 0 2 4 6 8 10
Figure 11: Comparison of roll rate response using Runge-Kutta 5th order for various time step
and input 00.2618 rad.
As a conclusion for the discussion on the execution times and accuracies, if high
accuracy is desired in the simulation, smaller time step should be used. But, this will
require more execution time because the smaller the time step is used, the more time
required to perform the same load of calculation. To obtain the real time simulation,
higher time step is required in order to control the total program execution time. The
conflict between the time step used and the accuracy of the simulation require some
trade-off for optimum simulation result. Table 5.1 and 5.2 summarized the execution
time for different numerical method and time step.
\
Table 5.1: Comparison of execution time for different numerical method and time
step to obtain 100s simulation time.
Time Step, ts Number of loop Execution time for 100s simulation time, t100
(second in count for 100s (second in real time)
simulation simulation time.
time) (100/ts)
Euler RK5
0.04 2500 7.91 5.17
0.1 1000 - 3.40
0.2 500 - 1.65
0.3 333 - 1.05
Table 5.2: Comparison of one loop execution time for different
numerical method and time step
Time Step, ts Execution time for one loop [ t100 / number of loop ]
(second in (second in real time)
simulation time) Euler RK5
0.04 0.003164 0.002068
0.1 - 0.0034
0.2 - 0.0033
0.3 - 0.003153
START
Advanced Variable
Editing
(frmVarAdv) Splash Screen Type of FCS
(frmSplash) 1. Pitch Rate Feedback
(Simple)
View Calculated Variable
2. Pitch Rate Feedback
(frmVarView)
(P+I Controller)
3. Roll Angle Control
Calculate IAS & Mach Main Menu 4. Roll Rate Damper
END 5. Yaw Damper
(frmVarMach) (frmMain)
6. Spiral Mode Stabilization
Analysis Tool
(frmAnalysis)
Main Simulation
Screen
(frmHUD)
Option
1. Set input value Sim Option Keys Help
2. Preset Input (frmOption) (frmHelp1)
3. HUD Mode
Figure 13 to 16 show the current status and the capability of the FDES program
which show the important feature such as type of simulation, selection of aircraft and
head-up display.
Figure 13 FDES program shows option to select type of simulation and fly with
different types of aircraft, and option to save simulation data for analysis.
Figure 14 Calculation of natural frequency, damping ratio and time constant as a
function of aerodynamic stability and control derivatives.
8.0 Conclusion
It was concluded that PC-based flight simulation can be a valuable tool for
illustrating flight dynamics to aeronautical students and can be challenging,
interesting and fun. The synergistic learning potential has improved due to more of
hands-on experience with FDES and the ease of use has made the course easier to
teach, complementing the classical and modern classroom theories. This simple
FDES also serves as a motivation and a starting point for the UTM ambition of a
full-scale flight simulator to serve as an engineering, research and instructional tool.
References