Control Engineering Homework 2024-25

Wheel Suspension System

System 3 Case 2

Claudia Silva Lozano, David Raggi, Qiu Jiayue

Table of Contents

1. Introduction​ 3
2. Assumptions and modeling.​ 4
2.1. Assumptions​ 4
2.2. Model setup​ 4
2.3. Transfer Function​ 6
3. Uncontrolled system analysis.​ 7
3.1. Step response.​ 7
3.3. Stability analysis of the uncontrolled system.​ 8
3.3.1. Poles of the transfer function.​ 8
3.3.2. Impulse response.​ 9
4. Controlled system.​ 10
4.1. Block diagram.​ 10
4.2. Control Law.​ 11
4.3. Performance analysis.​ 14
4.4. Stability analysis of the controlled system.​ 16
5. Conclusions​ 17
Appendix​ 18
 1.​Introduction
The suspension system in a vehicle is a critical component that ensures stability, safety and
comfort of the ride. In this report, a quarter car suspension system is studied to analyze its
response to a 10cm step on the ground, while moving with an initial horizontal velocity of 1m/s.

The goal is to minimize the resulting vibrations, particularly overshoot and oscillations,
experienced by the main body of the vehicle. This study involves modeling the system
dynamics, analyzing the uncontrolled behavior, and designing a control strategy that can be
implemented via the electromagnetic actuator to achieve the desired outcome. Using
mathematical tools and simulation, the stability, controllability, and overall effectiveness of the
proposed control solution is evaluated.
 2.​Assumptions and modeling.
In order to proceed with analysis, it is necessary to develop the mathematical model for the
system under consideration. Suitable assumptions are also made to facilitate the development of
the equations. The resulting transfer functions are derived from the fundamental equations of
motion, and serve as the foundation for analyzing and designing control strategies for the
system.

2.1. Assumptions
To simplify the modeling of the quarter-car suspension system and facilitate the development of
the model, the following assumptions have been made:

1.​ The initial conditions are zero.

2.​ The weight of the system is considered negligible.
3.​ The initial displacement of the springs is assumed to be zero.
4.​ The wheel's radius is treated as constant and not factored into dynamic calculations.
5.​ The initial velocity is not factored into the transfer functions.
6.​ Nonlinearities such as tire deformation are assumed to be linear springs, and
suspensions are modeled as linear springs.
7.​ External forces, including air resistance, are neglected.

2.2. Model setup

The wheel suspension system consists of two main components: solid 1, which represents one
quarter of the vehicle’s body, and solid 2, which corresponds to the wheel. This model simplifies
the dynamics of the vehicle by isolating a single wheel and its interactions with the suspension
system.

When the vehicle is in motion and encounters a disturbance in the road, the wheel experiences
an upward force which causes the wheel to displace vertically. The upward motion is then
transferred to the suspension system which then tries to oppose it. The spring resists
compression by exerting a restoring force, which returns the wheel and vehicle body to their
equilibrium positions while the damper reduces oscillation and prevents solid 1 from
experiencing excessive vibrations.

The following table characterizes the system by providing key parameters:

Mass 1 𝑚1 = 100𝑘𝑔

Mass of wheel 𝑚2 = 10𝑘𝑔

Suspension stiffness 𝑘 = 10000𝑁/𝑚

Damping coefficient ξ = 0. 01

Tire stiffness 𝑘𝑇 = 5000𝑁/𝑚

Table 1: System data.

The variables represent the positions of different components are:

𝑥1: position of the wheel

𝑥2: position of the car
𝑟: position of the road

The various forces acting on the vehicle system, which are a result of the interactions between
the wheel, car body, and road surface are:

Force from spring of wheel and tire: 𝐹𝑇 = 𝐾𝑇(𝑥2 − 𝑟)

Force from suspension system: 𝐹𝑘 = 𝑘(𝑥1 − 𝑥2)

Force from damping system: 𝐹𝑑 = 𝑏( 𝑥˙1 − 𝑥˙2)

The equations of motion for the wheel and the car body in the vehicle system are the following:

𝑚1𝑥¨1 =− 𝐹𝑘 − 𝐹𝑑

𝑚2𝑥¨2 = 𝐹𝑘 + 𝐹𝑑 − 𝐹𝑇
Applying the Laplace transform to the dynamic equations allows us to convert the second-order
differential equations into algebraic equations in the s-domain. This simplification eliminates
the time derivatives, transforming the system into a set of equations that are easier to solve:

2
𝑚1𝑠 𝑋1(𝑠) = 𝑘[𝑋2(𝑠) − 𝑋1(𝑠)] + 𝑏𝑠[𝑋2(𝑠) − 𝑋1(𝑠)]
2
𝑚2𝑠 𝑋2(𝑠) = 𝑘[𝑋1(𝑠) − 𝑋2(𝑠)] + 𝑏𝑠[𝑋1(𝑠) − 𝑋2(𝑠)] + 𝑘𝑇[𝑅(𝑠) − 𝑋2(𝑠)]

Solving the two equations give the relationship between solid 1 and solid 2, which can be
represented by the following equation:

2
𝑋2(𝑠) 𝑚1𝑠 +𝑏𝑠+𝑘
𝑋1(𝑠)
= 𝑏𝑠+𝑘

2.3. Transfer Function

A transfer function is a convenient way to represent a linear, time-invariant system in terms of
its input-output relationship. From the equations in the previous section, the transfer function
between solid 1 relative to the input ground step is as follows:

𝑋1(𝑠) 𝑘𝑇(𝑏𝑠+𝑘)
𝐻(𝑠) = 𝑅(𝑠)
= 2 2 2
(𝑚2𝑠 +𝑘+𝑏𝑠+𝑘𝑇)(𝑚1𝑠 +𝑘+𝑏𝑠+𝑘)−(𝑏𝑠+𝑘)
 3.​Uncontrolled system analysis.
Based on the given values and the transfer functions obtained, the uncontrolled system is
modelled and an analysis is conducted.

3.1. Step response.

The system experiences a 10 cm step while traveling at a horizontal velocity of 1𝑚/𝑠. Using
MATLAB, the 𝑠𝑡𝑒𝑝(𝐻) command is chosen to model the response of the system.

​
Figure 1: Uncontrolled system response to a 10cm step.

Using the stepinfo command in MATLAB, we can analyze the time-domain response of the
system to a step input. This function provides important performance metrics, such as rise time,
settling time, overshoot, and steady-state value, which are critical for evaluating the system's
dynamic behavior and stability. The calculated values for these metrics are summarized in the
figure below.
 ​
Figure 2: Step response data for uncontrolled system.

The system exhibits a high overshoot of 100%, indicating that the output exceeds the desired
value by a significant margin before settling. Additionally, the settling time is quite long,
measured at 405 seconds, which suggests that the system takes an extended period to stabilize
around the steady-state value.

3.3. Stability analysis of the uncontrolled system.

3.3.1. Poles of the transfer function.

A linear system of transfer function H(s) is stable if and only if all the poles of H(s) have strictly
negative real part. Mathematically this can be expressed as:

Using the pole(H) command in MATLAB, the poles of the transfer function can be easily
obtained:

Since the real parts of all the poles are negative, the system satisfies the stability criterion.
Therefore, the system is stable because the real parts of all poles are strictly negative.
3.3.2. Impulse response.
The impulse response stability criterion requires that the linear system's impulse response h(t)
tend to zero as time approaches infinity. This ensures that the system's response to an impulse
diminishes over time, preventing the system from becoming unstable. Mathematically, this can
be expressed as:

​
Figure 3: Impulse response of the uncontrolled system.

As seen from the figure, the system's impulse response effectively tends to zero as time
approaches infinity. This behavior confirms that the system is stable, as the impulse response
decays over time and does not grow unbounded.
 4.​Controlled system.
4.1. Block diagram.

​
Figure 4: General Feedback Control System Architecture.

In this system, the block diagram illustrates the interaction between the components, including
the sensors, actuators, and the control strategy.

​
Figure 5: Block diagram representation.

In the above figure:

●​ Gc represents the controller. It defines the control law used to generate the
control signal based on the system's error or feedback. The controller adjusts
the input to the plant (the system being controlled) in order to achieve the
desired output performance.
●​ Gp is the linear plant model. The plant represents the physical system being
controlled.
 ●​ H is a linear model of the sensor, which can be modeled as a constant such as 1
if the sensor dynamics are negligible (a perfect sensor is assumed).

The equivalent system, derived using Mason's Gain Rule, simplifies the feedback control system
into a single transfer function for easier analysis and design.

​
Figure 6: Equivalent System Representation.

4.2. Control Law.

The tuning of the control system is performed using MATLAB’s SISOTOOL, a graphical tool
designed to design and analyze Single-Input Single-Output (SISO) control systems. SISOTOOL
offers an interactive and intuitive interface, making it easier to adjust controller parameters and
visualize system performance in real-time.

The initial tuning of the system was performed using a proportional controller. In this phase,
the proportional gain was adjusted while ensuring that the system remained stable.

However, after several attempts to adjust the gain, it was observed that the response showed
high oscillations in the step response and failed to reach the desired steady-state final value. To
address these issues, a PID controller was selected as the next step. The addition of the
derivative and integral terms in the PID controller helps reduce oscillations and eliminate
steady-state error, allowing the system to reach its final value effectively.

The general form of a PID controller includes a gain K, two zeros, and one integrator. The form
of the PID controller is as follows:

A general PID controller was initially applied, but achieving an acceptable response proved
difficult. Despite adjusting the PID parameters, the system still exhibited undesirable behaviors,
such as high overshoot or slow settling. Due to these challenges, a PID controller with a
derivative filter was used to improve performance.

The tuning process involved adjusting the gain, as well as the locations of the two zeros and the
pole. These parameters were iteratively modified to minimize overshoot and oscillations,
ensuring the system's response met the desired performance criteria.

The PID controller with the derivative filter was applied in the following form:

​
Figure 7: Comparison of Step Responses: With and Without Derivative Filter.

As seen from the figure, the addition of a pole helped address the issues associated with
high-frequency noise and oscillations in the derivative term, ultimately leading to a more stable
and well-behaved system.

The final tuning process involved adjusting the gain in order to minimize overshoot and
oscillations. Several candidate controllers were saved during this iterative design process for
comparison.
 ​
Figure 8: Step response comparison for the final considered controllers.

The final PID compensator selected is the one that minimizes overshoot and oscillations. The
transfer function of the selected controller is given by:

​
Figure 9: Sisotool PID compensator.
4.3. Performance analysis.
This section focuses on the performance analysis of the controlled system in comparison to its
uncontrolled counterpart. To assess the system's behavior, the step responses of both the
controlled and uncontrolled systems are compared. Key aspects such as overshoot, oscillations,
and the final value reached are analyzed to determine the improvements achieved by the
control system.

​
Figure 10: Controlled system 10cm step response.

​
Figure 11: Step response data for controlled system.
 ​
Figure 12: Step response comparison.

By observing the above figure and comparing the step response data for each case, we can
deduce the following:

The uncontrolled system has a fast rise time of 0.1755 seconds but experiences significant
instability, with a long settling time of 404 seconds, a large overshoot of 100%, and substantial
oscillations. The system also reaches its peak quickly, with a peak time of 0.5563 seconds.

In contrast, the controlled system has a slower rise time of 13.3448 seconds but offers much
better stability. It has a settling time of 21.8747 seconds, a minimal overshoot of 0.2561%, and a
controlled peak of 1.0026. The controlled system reaches its final value more steadily,
demonstrating better performance and faster stabilization.
4.4. Stability analysis of the controlled system.

​
Figure 13: Bode diagram and stability margins of the controlled system.

The system's gain margin is 24.9 dB and the phase margin is 73.5°, indicating that the controlled
system is stable. Additionally, the required gain margin of 6 dB and phase margin of 45°
(commonly required in mechanical engineering applications) are met, further confirming the
system's stability against model simplifications, unmodeled phenomenon or unknown
disturbances.
 5.​Conclusions
The control law was successfully applied to the system, resulting in significant improvements in
stability, overshoot reduction, and oscillation minimization. By carefully tuning the controller
parameters, the system was able to achieve a much smoother and more controlled response
compared to the uncontrolled system.

In terms of performance, the controlled system performs as intended, especially in applications

that demand stability and minimal oscillations. While the uncontrolled system may have a
smaller rise time, its poor performance in stability and settling behavior makes it unsuitable for
tasks requiring accuracy and steady control.

While there was a slight compromise in the system's responsiveness, the overall performance
was enhanced, ensuring precise control and faster stabilization. The controlled system now
meets the objectives set, making it well-suited for applications requiring reliable and accurate
performance.
Appendix
%% UNCONTROLLED SYSTEM
clear;
close all;
Clc;

% Data
m1 = 100; %car mass
m2= 10; %wheel mass
k= 10000; %suspension spring stiffness
kt= 5000; %tire spring stiffness
damping_ratio= 0.01;
b= 2*damping_ratio*sqrt(m1*k);

% Transfer function
s= tf('s');
num= kt*(b*s+k);
den= (m2*s^2 + k + b*s + kt)*(m1*s^2 + b*s + k)-(b*s + k)^2;
H= num/den; %system transfer function

% Custom input (step of 10cm = 0.1m)

opts = stepDataOptions('StepAmplitude',0.1);

figure(1)
step(H,opts)
title('Uncontrolled System Response to 10cm Step')
xlabel('Time (s)')
ylabel('Displacement (m)')
grid on

% Stability analysis for uncontrolled system: Impulse response converges to

% zero and all the poles have a negative real part
figure(2)
impulse(H)
grid on
title('Uncontrolled System Impulse Response')
disp('Step response for uncontrolled system info:')
stepinfo(H)
disp('Poles of uncontrolled system:')
pole(H)

%% CONTROLLED SYSTEM
C= (-1.3045e-07*(s-3571)*(s+100))/(s*(s+0.3882));
H_controlled= feedback(C*H,1);
figure(3)
step(H_controlled,opts)
grid on
title('Controlled System Response to 10cm step')
ylabel('Amplitude (meters)')
disp('Step response for controlled system info:')
stepinfo(H_controlled)

% Stability analysis for controlled system: Positive gain and phase margins
figure(4)
margin(C*H)

%% CONTROLLED VS UNCONTROLLED STEP RESPONSES

figure(5)
step(H_controlled,opts); % 'b' is blue color for sys1
hold on;
step(H,opts);
grid on
title('Superimposed Step Responses of Controlled vs. Uncontrolled System')
ylabel('Amplitude (meters)')
legend('Controlled system', 'Uncontrolled system');
hold off;