EE5101/ME5401 Linear Systems Mini-project 2021

Control System Design of a Continuous-Flow Stirred Tank Reactor (CSTR)

Important note: The due date is 19/11/2021. Late submission is absolutely not allowed as the
grades have to be submitted to the department very soon after the final exam. You may work together
with your classmates. But do write your report independently. And the results are supposed to be
different from each other as the parameters are based upon your matriculation numbers.

1 Background
This is a real control problem in chemical engineering, where we are trying to control the
reaction process that takes place in a tank shown in the following figure [1]. The chemical reaction is
described by 𝐴𝐴 → 𝐵𝐵. 𝐴𝐴 and 𝐵𝐵 stand for the reactant and product. The reaction is conducted in the
large container located at the central of the tank, meanwhile there is water flow inside the surrounded
container wall (we call it cooling jacket) to control the reaction temperature.

Figure 1. CSTR reactor [1]

The real plant is a non-linear, time delayed, time-varying system associated with variety of
inputs and outputs. However, for rough analysis, the system could be simplified so that it becomes a
3rd order LTI state-space model with 2 inputs and 3 outputs, as is presented in [1]. Now we are making
further assumption that the component concentration sensor is missing, which often happens in some
small factories since the component concentration sensor is very expensive and hard to install, the
plant becomes a 2-input-2-output LTI system.
This is an interesting application. You are supposed to control the component concentration of
the reactant. But you cannot measure it directly because the sensor is too expensive and you want to
keep the cost as low as possible. Can you solve the control problem with the cheap temperature
sensors? If you are confident that the model is accurate enough, then you can do it with the knowledge
you have learned from this module.

2 Modeling
For model-based control, the first step is to build an effective dynamic model for our target plant,
i.e., the continuous-flow stirred tank reactor (CSTR) in this project. Usually, there are two classes of
methods to build a dynamic model, either by first principles or by system identification. For many
complex process control applications, it is typically difficult, laborious, and time-consuming to derive
a dynamic model from first principle equations. Besides, the acquired non-linear system also requires
1 / 6
 EE5101/ME5401 Linear Systems Mini-project 2021

a significant time for calibration and verification, as well as more advanced control technologies.
In this project, the most important factors that we want to control is the component concentration
of the reactant during reaction and the reaction temperature. However, as aforementioned, only
temperature sensors are available. There are 2 sensors, one installed in the reaction container to detect
the reaction temperature, the other installed in the cooling jacket outlet pipe to detect the outflow
water temperature.
Notice that the input amount of the reactant and the temperature of the input water to the cooling
jacket are fixed. Our objective is to use the 2 valves VC1 and VC2, shown in Figure 1 to control the
outlet flow rate of the reaction and the flow rate of the water coming out from the cooling jacket in
order to maintain the whole reaction on the given operating point.
𝑇𝑇
Define the state vector 𝑥𝑥 = �𝐶𝐶𝑎𝑎 𝑇𝑇 𝑇𝑇𝑗𝑗 � , where 𝐶𝐶𝑎𝑎 is the component concentration of the
reactant, 𝑇𝑇 is the reaction temperature in the tank, and 𝑇𝑇𝑗𝑗 is the temperature of the outflow water
of the cooling jacket. Define the control signal 𝑢𝑢 = [𝐹𝐹 𝐹𝐹𝑗𝑗 ]𝑇𝑇 , where 𝐹𝐹 is the outlet flow rate of the
reaction; 𝐹𝐹𝑗𝑗 is the flow rate of the water in the cooling jacket. Define the measurement vector 𝑦𝑦 =
[𝑇𝑇 𝑇𝑇𝑗𝑗 ]𝑇𝑇 , because we only have two temperature sensors. The system is described by
x = Ax + Bu + Bw,
y = Cx
where
 −1.7 −0.25 0 
= 
A  23 −30 20 ,

 0 −200 − ab0 −220 − ba 0  (1)
 3+ a 0 

B = −30 − dc 0 

 0 −420 − cd 0 
0 1 0 
C= 
0 0 1 
where the vector w describes some possible load disturbances. The initial condition of the system
is assumed to be

x0 = [1 100 200] .
T
(2)

The parameter values of 𝑎𝑎, 𝑏𝑏, 𝑐𝑐, 𝑑𝑑 in equation (1) should be chosen to be the last four digits of
your matriculation number. For example, if your matriculation number is A0235004X, then 𝑎𝑎 = 5,
𝑏𝑏 = 0, 𝑐𝑐 = 0, 𝑑𝑑 = 4. In this case, we have
 −1.7 −0.25 0   3+5 0 

A = 23 −30  
20  , B = −30 − 40 0 
. (3)
 0 −200 − 500 −220 − 50   0 −420 − 40 

3 Control System Design

After all, we get a linear state space model (1) for the CSTR system. In the following, different
2 / 6
 EE5101/ME5401 Linear Systems Mini-project 2021

control strategies will be explored to achieve control of this system. We will target both the regulation
and set point tracking problems.
3.1 Design specifications
The transient response performance specifications for all the outputs y in state space model (1)
are as follows:
1) The overshoot is less than 10%.
2) The 2% settling time is less than 30 seconds.
Note: (a) This transient response is checked by giving a step reference signal for each input
channel, i.e., [1, 0] and [0, 1], with ZERO initial conditions; (b) For all the following task 1) to 5),
your control system should first satisfy this performance specification and then you are supposed to
finish the required investigation for each task as well.
3.2 Tasks
Your study should include, but not limited to
1) Assume that you can measure all the three state variables, design a state feedback controller
using the pole place method, simulate the designed system and show all the three state
responses to non-zero initial state 𝑥𝑥0 with zero external inputs. Discuss effects of the
positions of the poles on system performance and monitor control signal size. In this step,
both the disturbance and set point can be assumed to be zero. (10 points)
2) Assume that you can measure all the three state variables, design a state feedback controller
using the LQR method, simulate the designed system and show all the state responses to
non-zero initial state with zero external inputs. Discuss effects of weightings Q and R on
system performance and monitor control signal size. In this step, both the disturbance and
set point can be assumed to be zero. (10 points)
3) Assume you can only measure the two outputs in 𝑦𝑦. Design a state observer, simulate the
resultant observer-based LQR control system, monitor the state estimation error, investigate
effects of observer poles on state estimation error and closed-loop control performance. In
this step, both the disturbance and set point can be assumed to be zero. (10 points)
4) Assume that you can measure all the three state variables, design a decoupling controller
with closed-loop stability and simulate the step response of the resultant control system to
verify decoupling performance with stability. In this question, the disturbance can be
assumed to be zero. Is the decoupled system internally stable? Please provide both the step
(transient) response with zero initial states and the initial response with respect to 𝑥𝑥0 of the
decoupled system to support your conclusion. (10 points)
5) In an application, the operating set point for the two outputs is
ysp = [100, 150]T .

Assume that you only have two cheap sensors to measure the output. Design a controller such
that the plant can operate around the set point as close as possible at steady state even when step

3 / 6
 EE5101/ME5401 Linear Systems Mini-project 2021

disturbances are present at the plant input. Plot out both the control and output signals. In your
simulation, you may assume the step disturbance 𝑤𝑤 = [−2, 5]𝑇𝑇 takes effect from time 𝑡𝑡𝑑𝑑 =
10𝑠𝑠 afterwards. (10 points)
6) To make this task more interesting, suppose we intend to manipulate the three state variables
directly instead of the two outputs. Our target is to maintain the states 𝑥𝑥 around a given set
point 𝑥𝑥𝑠𝑠𝑠𝑠 = [5, 250, 300]𝑇𝑇 at steady state starting from the initial state 𝑥𝑥0 . Is it possible?
In this question, you may assume that all the state variables can be measured directly and
there are no disturbances. (10 points)
(a) If your answer is YES, please detail your control system design strategy to ensure x to be
𝑥𝑥𝑠𝑠𝑠𝑠 at steady state and demonstrate its effectiveness through simulation.
(b) If your answer is NO, explain why. In such a case, we may only want to keep the state
variables at steady state close enough to the set point 𝑥𝑥𝑠𝑠𝑠𝑠 . However, in practice, we
usually place different emphasis on the exactness of the three state variables. To address
our purpose quantitively, we aim to minimize the following objective function
1
J ( xs ) = ( xs − xsp )T W ( xs − xsp ), (4)
2
where 𝑊𝑊 = 𝑑𝑑𝑑𝑑𝑑𝑑𝑑𝑑(𝑎𝑎 + 𝑏𝑏 + 1, 𝑐𝑐 + 4, 𝑑𝑑 + 5) is a diagonal weight matrix and 𝑥𝑥𝑠𝑠 is the
state vector at steady state. Here, a, b, c, d are still the last four digits in your matriculation
number, as defined above. Please detail your control system design strategy to minimize
the objective 𝐽𝐽(𝑥𝑥𝑠𝑠 ) at steady state and demonstrate its effectiveness through simulation.
Can you prove that 𝐽𝐽(𝑥𝑥𝑠𝑠 ) at steady state is indeed minimized theoretically in your design?

Note that there are no unique answers to the above design questions. For the tasks in our project,
you can assume that the control input is unlimited. However, in practice all the physical actuators can
only provide a limited drive capacity. You need to make your own judgement assuming you are the
engineer responsible for the control system design in the real world. There are three major factors
you should consider when you design and justify your controller:
 Speed --- Transient response
 Accuracy --- Steady state error
 Cost ---- Size of the control signals
Please do follow the design procedures you have learned in linear systems to solve all the above
questions. List the necessary formulas and intermediate results in your report. If you only call the
MATLAB built-in functions for control system design with no details, for example, simply use place
to place poles or lqr to design the LQR regulator, you will get ZERO marks.

4 Reference
[1] P. Albertos, A.Sala, Multivariable Control Systems: An Engineering Approach, pp 19-27

4 / 6
 EE5101/ME5401 Linear Systems Mini-project 2021

5 Format of Reports
Your report should mainly contain the plant description, control and observer design method
description, your design details, simulation results, possible comparison, comments and discussion,
modification and refinements.
The report should include the following and be organized in the following sequence:
• A cover paper to indicate “Assignment for EE5101/ME5401 (or your specialization
code if else) Linear Systems”, a title of your report at your choice, your full name, your
Matriculation number, email address and date;
• An abstract of 50-100 words on a separate page;
• A contents table on a separate page;
• Section 1 Introduction
• The major materials of your report organized nicely in a few sections each with specific
focus. Label your equations, tables, and figures with number and caption for reference
in the text. Your figure size and figure quality should be high enough to facilitate the
verification of your results.
• The last section on conclusions.
• A list of reference books/papers if any;
• Appendices if any each on a separate page. Your MATLAB code should be in this
appendix. If you use Simulink, a screenshot of your Simulink model should be inserted
at proper position in the above major materials part as figures.
Pay attention to your presentation (English writing, organization, and layout et al). Make the
report formal, complete and readable. It is also advisable to write your report with a word-processing
software such as Word or LaTeX.
The final point to note about your report: it is the content that matters not the length. Keep in
mind that there are only TWENTY SEVEN pages in John Nash’s PhD thesis, which leaded to his
Nobel Prize. Therefore, you will be penalized if you put too much “copy and paste” material in your
report.

6 A Note on Access and Use of MATLAB

To complete the project, you are supposed to use SIMULINK and MATLAB. The easy way is
to learn how to build various block diagrams in SIMULINK first, and then try to solve the control
systems design for the mini-project. An excellent Control Tutorial for MATLAB and Simulink can be
found at http://ctms.engin.umich.edu/CTMS/index.php?aux=Home. Besides, a Matlab manual is
provided in IVLE for the first timers.
If you don’t have MATLAB on your PC currently, you can access MATLAB in either of the
following two ways:
1) Go to PC clusters located at the third floor of E2: http://www.eng.nus.edu.sg/eitu/pc.html.

5 / 6
 EE5101/ME5401 Linear Systems Mini-project 2021

2) Download MATLAB from NUS information technology center: every NUS student can
have a license. https://nusit.nus.edu.sg/services/software_and_os/software/software-
student/#install-matlab.

Hint on MATLAB/SIMULINK:
A. You can use functions such as step, initial and lsim to simulate the system’s
corresponding response. Also, all these simulations can be done with SIMULINK.
B. In some cases, it may be easier to use SIMULINK for the simulation, for example,
question 5).

6 / 6