JABATAN KEJURUTERAAN KIMIA & PROSES FAKULTI KEJURUTERAAN& ALAM

BINA

KKKR3713 DYNAMICS AND PROCESS CONTROL

SEMESTER II 2023/2024

COURSE LECTURERS
PROF. MADYA DR. NORLIZA BINTI ABD RAHMAN
DR. JARINAH BINTI MOHD ALI

CHEMICAL ENGINEERING PROGRAMME

SIMULATION PROJECT
COMPUTER SIMULATION OF IDEAL BINARY DISTILLATION COLUMN

Name: YASMIN MARSYA BINTI SUHAIRI

Matrix No.: A188028
Submission Date:
Total Marks:
3. Design the control systems to control the composition of the top, 𝒙𝑫 and bottom, 𝒙𝑩 products,
using P, PI and PID. Test the controller based on servo and regulatory control. Discuss your
results. Which is the best controller for the system?

1. Manual Controller.

Manual Controller: Servo Control

Servo control is a method used to precisely control the position, velocity, or acceleration of a
mechanical system. It is commonly used in applications like robotics, CNC machinery, and
automated manufacturing. The core of servo control involves feedback loops that continuously
adjust the system to achieve the desired output.

P PI PID
Set point, xD 0.9488 0.9488 0.9488
Step time, xD 1 1 1
Set point, xB 0.4 0.4 0.4
Step time, xB 1 1 1
Kp,1 100 100 100
Kp,2 100 100 100
Ki,1 0 1 2
Ki,2 0 1 2
Kd,1 0 0 0.05
Kd,2 0 0 0.05
P Controller

PI Controller
PID Controller

Best controller: PI controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and PlD
controllers
- Zero offset

Manual Controller: Regulatory Control.

Regulatory control is a type of control system used to maintain a process variable at a desired
setpoint. It is commonly used in industrial processes to ensure that variables such as
temperature, pressure, flow rate, and chemical composition remain within specified limits.
Regulatory control systems use feedback loops to continuously monitor and adjust the process
to achieve the desired outcome.
P PI PID
Load, U1 50 50 50
Step time, U1 1 1 1
Load, U2 50 50 50
Step time, U2 1 1 1
Kp,1 100 100 100
Kp,2 100 100 100
Ki,1 0 1 2
Ki,2 0 1 2
Kd,1 0 0 0.05
Kd,2 0 0 0.05
P Controller

PI Controller
PID Controller

Best controller: P controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and PlD
controllers
- Zero offset

2. Toolbox Controller.
Toolbox Controller: Servo Control

P PI PID
Set point, xD 0.9488 0.9488 0.9488
Step time, xD 1 1 1
Set point, xB 0.4 0.4 0.4
Step time, xB 1 1 1
Kp,1 100 100 100
Kp,2 100 100 100
Ki,1 0 1 2
Ki,2 0 1 2
Kd,1 0 0 0.05
Kd,2 0 0 0.05

P Controller
PI Controller

PI Controller
Best controller: PID controller.
- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and PI
controllers
- Zero offset

Toolbox Controller: Regulatory Control

P PI PID
Load, U1 50 50 50
Step time, U1 1 1 1
Load, U2 50 50 50
Step time, U2 1 1 1
Kp,1 100 100 100
Kp,2 100 100 100
Ki,1 0 1 2
Ki,2 0 1 2
Kd,1 0 0 0.05
Kd,2 0 0 0.05

P Comtroller
PI Comtroller
PID Comtroller

Best controller: PID controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and Pl
controllers
- Zero offset

3. Multiple Set Point and Disturbance.

 Multiple Set Point: Servo Control

P PI PID
Set point, 𝑥𝐷⬚ 0.9488 0.9488 0.9488
Step time, 𝑥𝐷⬚ 1 1 1
Set point, 𝑥𝐷1 0.9488 0.9488 0.9488
Step time, 𝑥𝐷1 3 3 3
Set point, 𝑥𝐷2 0.9488 0.9488 0.9488
Step time, 𝑥𝐷2 5 5 5
Set point, 𝑥𝐵⬚ 0.4 0.4 0.4
Step time, 𝑥𝐵⬚ 1 1 1
Set point, 𝑥𝐵1 0.4 0.4 0.4
Step time, 𝑥𝐵1 3 3 3
Set point, 𝑥𝐵2 0.4 0.4 0.4
Step time, 𝑥𝐵2 5 5 5
Kp,1 100 100 100
Kp,2 100 100 100
Ki,1 0 1 2
Ki,2 0 1 2
Kd,1 0 0 0.05
Kd,2 0 0 0.05

P Controller
 PI Controller

PID Controller

Best controller: P controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than PI and PlD
controllers
- Zero offset
 4. Repeat Question 3 with multiple set points and disturbances.
Multiple Set Point: Regulatory Control

P Controller

PI Controller
PID Controller

Best controller: PID controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and Pl
controllers
- Zero offset

5. Tune your controller with TWO tuning methods of the following and compare the
results.
5.1. Ziegler-Nichols
Controller Tuning - Ziegler-Nichols (ZN)
1. Set the I (integral) and D (derivative) gains to zero
2. Turn on the P (proportional) mode and increase controller gain, Kc until the system
reaches the ultimate gain Ku, at which the output of the loop starts to oscillate
consistently.
3. Measure the period of oscillation, Pu.
4. Use the Ziegler-Nichols tuning table to calculate the P. I, and D values.
5.2. Cohen-Coon
 Formula of ZN for different control settings:
Type of Controller Kc 𝝉𝑰 𝝉𝑫
Porpotional (P) 50 50 50
Proportional-integral (PI) 1 1 1
Proportional-integral-derivative (PID) 0 0 0.05

Values of Kc, 𝝉𝑰 and 𝝉𝑫 for different types of controllers:

Type of controller Kc 𝝉𝑰 𝝉𝑫
P xD 6000 - -
xB 6000 - -
PI xD 5500 0.00073 -
xB 5500 0.00073 -
PID xD 7000 0.0004 0.000115
xB 7000 0.0004 0.000115
Controller Tuning - Ziegler-Nichols (ZN)
P Controller

PI Controller
 PID Controller

Best controller: PID controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and Pl
controllers
- Zero offset

5.2 Controller tuning - Cohen-Coon (CC)

1. Perform an open-loop test to determine the process reaction curve.
2. Draw a straight-line tangent to the curve at the point of inflection.
3. To determine,
Td = Intersection of tangent line with time axis
Bu = ultimate value of B at large t.
4. From the values obtained, T = Bu/S and Kp=Bu/M.
5. Use the CC tuning formulas to determine the PID parameters.
 Formula of CC for different control settings:

Type of Controller Kc 𝝉𝑰 𝝉𝑫
Porpotional (P) 1 𝑇 𝑇𝑑 - -
(1 + )
𝐾𝑝 𝑇𝑑 3𝑇
Proportional-integral (PI) 1 𝑇 9 𝑇𝑑 30 + 3𝑇𝑑 /𝑇 -
( + ) 𝑇𝑑 ( )
𝐾𝑝 𝑇𝑑 10 12𝑇 9 + 20𝑇𝑑 /𝑇
Proportional-integral-derivative (PID) 1 𝑇 4 𝑇𝑑 32 + 6𝑇𝑑 /𝑇 4
( + ) 𝑇𝑑 ( ) 𝑇𝑑 ( )
𝐾𝑝 𝑇𝑑 3 4𝑇 13 + 8𝑇𝑑 /𝑇 11 + 2𝑇𝑑 /𝑇

Values of Kc, 𝝉𝑰 and 𝝉𝑫 for different types of controllers:

Type of controller Kc 𝝉𝑰 𝝉𝑫
P 17.9226 - -
PI 14.6397 0.8885 -
PID 22.1366 0.8840 0.2219
Block diagram CC:

P Controller

PI Controller
PID Controller

Best controller: PID controller.

- Both responses for xD and xB show little oscillatory effect
- No overshoot for xD and lower overshoot for xB
- Shorter settling time and reach the steady state set point value faster than P and Pl
controllers
- Zero offset
Comparison between Ziegler-Nichols and Cohen-Coon.
Ziegler-Nichols Method
Pros:
• Simplicity: Easy to implement and requires minimal process information.
• Quick Tuning: Can quickly provide a set of tuning parameters.
• Widely Used: Well-known and widely accepted in industry.
Cons:
• Aggressive Tuning: Often results in aggressive control actions, which can lead to high
overshoot and oscillations.
• Less Accurate for Dead Time: Not very effective for processes with significant dead time.
Cohen-Coon Method
Pros:
• Better for Dead Time: More effective for processes with significant dead time.
• Balanced Tuning: Provides a more balanced tuning, often resulting in less aggressive
control actions compared to Ziegler-Nichols.
• Detailed Process Information: Utilizes more detailed process information (gain, dead
time, and time constant) for tuning.
Cons:
• Complexity: More complex to implement compared to Ziegler-Nichols.
• Requires Step Test: Needs a step test to determine process parameters, which can be
time-consuming.
Better tunning method: Cohen-Coon Method
- Have a quick and straightforward tuning method and your process does not have
significant dead time, Ziegler-Nichols might be the better choice.
6. Advanced Process Control Strategy (Optional)
For an advanced process control strategy, we will consider Model Predictive Control (MPC).
MPC uses a model of the process to predict future outputs and optimize control moves.
Here’s a basic outline:
1. Develop a dynamic model of your process.
2. Implement the MPC algorithm which involves:
o Predicting future outputs using the model.
o Solving an optimization problem to minimize a cost function (e.g., tracking
error, control effort).
o Applying the optimal control moves to the process.
3. Test the performance of the MPC and compare it with the best-tuned PID controller
from Question 5.