Professional Documents
Culture Documents
Authors
Ignacio Moreno Pubul
Pablo Manuel Mu
noz de la Flor
December 23, 2015
Assigment 1
1 of 20
Contents
1 Introduction
2 Longitudinal Controllers.
2.1 Pitch attitude controller for phugoid suppressions. . . . . . . . . . . . . . . . . . .
2.2 Speed controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2.3 Altitude-hold controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2
2
4
6
3 Lateral Controllers.
3.1 Root locus plot. Lateral control. . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.2 Yaw Damper. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3.3 Roll Controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
7
7
8
10
4 Controller Capabilities.
4.1 Thrust limit models. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.2 Non linear equations of motion and controls. . . . . . . . . . . . . . . . . . . . . .
4.3 Handling qualities assessment. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
12
12
13
14
5 Conclusion.
15
October 2015
Assigment 1
2 of 20
Introduction
This last laboratory report has as its main objective to design the controllers for the C5-A under
our flight conditions of sea level and M = 0.45, for both longitudinal and lateral motion. These
controllers are put to the test under a real world environment to evaluate their capabilities. The
data characteristics of this sessions are identical to the previous assignment, Assignment 1, Tables
1-2.
Longitudinal Controllers.
2.1
The purpose of this analysis is to design a simple controller for phugoid suppression. If we recall
from previous assignments, the phugoid mode was a lightly damp, and thus rather long, long
frequency oscillation in speed, pitch and altitude. Therefore, the phugoid mode is highly dependent
on the evolution of the pitch. If the controller can maintain a desired pitch, we will successfully
suppress the phugoid mode.
The input of this controller is a desired pitch angle c . The output is the actual pitch angle .
The evolution of the pitch angle depends on the deflection of the elevator, e . The controller, J(s),
corrects the measured error, (c ). The system is represented by a simple feedback loop.
Finding the closed loop transfer function, knowing Ge from the previous assignment, and J(s) is
relatively simple:
Ge J(s)
=
c
1 + Ge J(s)
(1)
The concern is then finding J(s). A reasonable general form for J(s) is:
ki
J(s) =
+ kp + kd s
s
s
N
1
+1
(2)
To asses the stability of the system under proportional control, using the Nyquist criterion, we
simply have to consider J(s) = kp , and analyze the Nyquist diagram of the open loop transer
function G(s) = kp Ge .
Bachelor in Aerospace Engineering
October 2015
Assigment 1
3 of 20
0.25
0.2
Imaginary Axis
0.15
0.1
0.05
0
-0.05
-0.1
-0.15
-0.2
-2
-1.5
-1
-0.5
Real Axis
Ge J(s)
1.39s2 1.098s 0.01559
=
=
4
c
1 + Ge J(s)
0.9339s + 2.171s3 + 2.414s2 + 0.02482s + 0.009217
These values of kp , ki , kd , and N , came from tuning the parameters to obtain a solution as close
to the statement requirements as possible. These values, for a step input of 2o render the following
results:
October 2015
Assigment 1
4 of 20
Close loop
0.04
0.035
0.03
P controller
PI controller
PID controller
0.025
0.02
0.015
0.01
0.005
10
15
Time (s)
20
25
30
Open loop
0.6
0.4
0.2
0.2
0.4
0.6
0.8
200
400
600
800
Time (s) (seconds)
1000
1200
1400
1600
Figure 3: Response of the Open Loop Pitch Angle to a unit Step Command.
2.2
Speed controller.
This controller will have two inputs, (p and e ) and two outputs( and u), although only u is
in the feedback loop. The controller must correct the measured error (uc u). The chart below
summarizes the workings of the system:
October 2015
Assigment 1
5 of 20
ki
s
+ kp + kd s
h
s
N
1
+1
The closed loop transfer functions are given by:CHECK THIS AT THE END AND SUBSTITUTE
FOR CORRECT ONE.
Gup =
Gup
1 + Gue J
Gp = Gp Ge JGup
(3)
(4)
Before analyzing the response, let us take a look at the open loop response of the system. As the
figure below demonstrates, a change in the throttle setting of the aircraft results in a slow, lightly
damped, oscillating transition to the new state, that takes quite long to complete. Logically in a
real-world setting this is unacceptable, and so the phugoid mode is effectively suppressed.
October 2015
Assigment 1
6 of 20
Phugoid approximation
Exact solution
Open loop
6
Open loop
4
4
3
(rad)
u (ft/s)
2
4
3
200
400
600
800
1000
Time (s) (seconds)
1200
1400
1600
1800
200
400
600
800
1000
Time (s) (seconds)
1200
1400
1600
1800
2.3
Altitude-hold controller.
It is the combination of the previous two controllers in a single feedback control. Therefore, we are
going to have two controllers, one for the elevator, and one for the throttle. We will not use thrust
limiters for this section.
October 2015
Assigment 1
7 of 20
Ge J(s)
Bla
=
=
c
1 + Ge J(s)
Ble
(5)
3
3.1
Lateral Controllers.
Root locus plot. Lateral control.
The simple feedback below will be analyzed, with a proportional control. Within the validity of
this model, we will analyze the root locus plot of the roll damper (p to a ), yaw damper (r to r ),
yaw stiffness modifier (v to r ), and synthetic bank stiffness ( to a ).
October 2015
Assigment 1
8 of 20
3.2
Yaw Damper.
The task is to design a yaw damper that allows us to respond to disturbances and also to maneuver
in a steady turn. To do so, we must include a high-pass or wash-out filter. We must find the
values of the gains and the parameter of the filter that produce the desired smooth transient from
the initial to the final condition.
October 2015
Assigment 1
9 of 20
(6)
JGrr
1 + W JG
(7)
W (s) =
Closed Loop Transfer Function.
The closed loop transfer function is one such
Grrc =
In this case we use a proportional controller with J = kp = 1.81, and the response to a yaw rate
impulse of 20o /s is:
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
20
40
60
80
100
120
140
October 2015
Assigment 1
10 of 20
0.1
0.1
0.2
0.3
0.4
0.5
0.6
0.7
50
100
150
Time (s)
3.3
Roll Controller.
Using the previous yaw controller, we must design a roll controller according to the block diagram
below (note the typo p, should be r ). Using proportional control, we will set the values of the
appropriate gains, and then analyze the response to a initial bank error of 10 degrees and to a
bank command of 15 degrees.
10
October 2015
Assigment 1
11 of 20
Bank command: 15
0.45
Azimuth angle, (rad)
Rudder deflection, (rad)
r
0.4
0.35
0.3
Variables
0.25
0.2
0.15
0.1
0.05
0.05
10
15
Time (s)
20
25
30
Figure 18: Response of the System: Azimuth angle, yaw rate and rudder deflection.
Figure 18 shows how the azimuth angle, , yaw rate, and rudder deflection respond to the system. The new steady state shows an oscillation with takes about 15-20 seconds to decay, and is
characterized by a linearly increasing azimuth angle, a constant yaw rate, and a very small rudder
deflection. It is a turn to the right. The plot below shows the new bank angle at which the system
stabilizes, c , and that other variables and control surface deflections are very small.
Bank command: 15
0.7
Bank angle, (rad)
Aileron deflection, a (rad)
Roll rate, p (rad/s)
0.6
0.5
Variables
0.4
0.3
0.2
0.1
0.1
10
15
Time (s)
20
25
30
Figure 19: Response of the System: Bank angle, roll rate and aileron deflection.
Now let us take a look at the response of the system for an initial bank error of 10o :
11
October 2015
Assigment 1
12 of 20
14
x 10
10
Variables
10
15
Time (s)
20
25
30
Figure 20: Response of the System: Azimuth angle, yaw rate and rudder deflection.
In Figures 20 and 21 we can see the response of the system to an initial bank error of 10o . Although
the shape of the oscillations are quite strange and unexpected, the time of the decay is consistent
with the expected reasonably well damped oscillation. The magnitudes are also quite coherent. We
see a maximum aileron deflection required of about 25o , and rudder deflection of -0.2o (although
the sign and low value hints at a slight incoherence).
All the variables except for the azimuth angle converge at 0, meaning that the new steady state
after level flight is reinstated is one with a heading different by approximately 0.6o .
Initial bank: 10
0.5
Bank angle, (rad)
Aileron deflection, a (rad)
Roll rate, p (rad/s)
0.4
0.3
Variables
0.2
0.1
0.1
0.2
0.3
10
15
Time (s)
20
25
30
Figure 21: Response of the System: Azimuth angle, yaw rate and rudder deflection.
4
4.1
Controller Capabilities.
Thrust limit models.
In order to emulate real environments it is necessary to include thrust limit models. We will assume
that the maximum thrust is 10% greater than cruise thrust (Tmax = 1.1 Tcr) and that idle thrust
is 10 % of the cruise thrust (Tidle = 0.1 Tcr).
The response of the system is as follows:
12
October 2015
Assigment 1
13 of 20
4.2
We will introduce in Simulink the complete set of non linear equations of motion, and add the
longitudinal and lateral directional controls. Then, we will select a longitudinal disturbance and
longitudinal maneuver and analyze the response of the aircraft. The process is repeated for the
lateral-directional dynamics.
13
October 2015
Assigment 1
4.3
14 of 20
The assigned aircraft, the Lockheed C5-A Galaxy, is an aircraft of class III, large, heavy, and with
low maneuverability. The flight condition corresponds to a class B, that is, non-terminal phase
of the flight where tracking is less precise and maneuvering is gradual but flight path control is
accurate.
Table 1: Longitudinal Motion.
Phugoid Mode Damping
Short Period Mode Damping
0.0361
0.6114
Level 2
Level 1
0.5127 s
41.8958 s
0.7951
0.2952
Level
Level
Level
Level
1
1
1
1
Most of the results obtained render flying qualities adequate for the mission flight phase, Level 1.
We do see however that the phugoid mode damping we achieve a Level 2 result, which is adequate
but may affect mission effectiveness.
14
October 2015
Assigment 1
15 of 20
Conclusion.
To conclude on a self-reflecting note, this deep analysis has served its purpose in teaching the authors the intricate considerations that lay the basic foundations of longitudinal and lateral response
to the controls, and has helped consolidate the theoretical background of the lectures.
We have been able to determine and analyze the most important parameters of the aircraft for each
control, and how the plane behaves when modifying the control inputs. It was possible to study
how the variations in elevator, throttle, rudder and aileron affected the motion of the aircraft.
The results obtained throughout the sessions show congruence and coherence for the given flight
conditions. For example, the poles obtained generated peaks which coincided perfectly with the
phugoid and short period modes that we could expect from exciting the system through the motion
of the control parameters. We saw how increasing throttle naturally sent the aircraft into a phugoid,
lightly damped oscillation, as would be expected, or how moving the elevator generated a new flight
condition with higher speed and angle of attack. Steady turn results were also coherent, as were
steady sideslip conditions, if accounting for the error in sign convention.
We also learned to interpret bode plots, and how poles and zeros affect the results, as well as
interpreting the time domain solutions of the step response, which gave great insight into the
stabilization conditions of the aircraft after a maintained change in the actuation controls.
Although the time investment has been high, this response analysis of the Lockheed C5-A has
proven to be a fantastic manner of understanding and gaining insight in the nature of response to
actuation of controls. We have been able to observe the behavior of the transfer functions, their
application, and approximation methods, and how the are deeply related to the control of the
system.
15
October 2015