CHL4020: Process Control & AI Applications

Course Project Report

Project Title: Feedback control of a chemical Reactor
using PID controller

Group Members

Bhavneet Kaur (B20CH012)

Dwimalu Boro (B20CH017)
Divyanshu Pandey (B20CH0115)
Ayush Abad (B20CH011)

INDIAN INSTITUTE OF TECHNOLOGY JODHPUR

DEPARTMENT OF CHEMICAL ENGINEERING
 Implementation Of Temperature Regulation and Concentration Tracking of
CSTR with Fuzzy Controllers

Motivation
Industrial processes today are challenged by a dynamic market that is difficult to predict,
leading to consumer-oriented industries and personalised products. This has in turn resulted
in increasing energy costs, thereby encouraging the design of highly integrated thermal point
of view processes for which dynamic stability and control offer real challenges. Advanced
control is one of the most important directions in which the production situation can be
improved. The domain of CSTR (Continuous Stirred Tank Reactor) control problem in
particular has remained controversial in nonlinear dynamics. Conventional controllers are
limited to linear time invariant system applications. However, in real time applications of the
CSTR model, the functional parameters are defined in an environment of non-linear
characteristics like wear and tear. This requires the employment of intelligent and adaptive
controllers. The conventional approaches to deal with constraints suffer serious limitations
and hence are the motivation of researchers to seek more flexible and powerful algorithms
such as bio-inspired intelligent computing including The Genetic Algorithm, Neural
Network and Fuzzy logic. The global interest in fuzzy logic demonstrates industrial
performances of an approximate reasoning over crisp assumption models in real environment
applications.
With low computational cost, fuzzy logic is highly flexible to deal with complex nonlinear
problems.
Industrial applications:
1. Chemical industry:In chemical synthesis, polymerization etc. for controlling reaction
kinetics and maximising product yield and quality.
2. Pharmaceutical industry:For drug synthesis, purification, and isolation processes
for ensuring product quality, purity, and reproducibility.
3. Biotechnology industry: In the production of enzymes, proteins, and antibiotics
for maintaining the activity and stability of these products and ensuring the
reproducibility and scalability of the bioprocess.
4. Food and beverage industry: In fermentation processes in the production of
beverages, dairy products for controlling the fermentation process and ensuring
product consistency and quality.

Previous attempts
1. 1950s - First attempts at modeling and control of CSTRs using simple
mathematical models based on mass and energy balances were developed.
2. 1960s - Nonlinear models for control
3. 1970s-1980s - Control strategies including PID controllers and adaptive controllers
were developed.
4. 1990s-2000s - Advanced control techniques such as neural networks, fuzzy logic,
and model predictive control, were introduced for CSTR control.
5. 2010s-present - Intelligent control such as artificial neural networks, fuzzy logic,
and genetic algorithms, have been widely used in CSTR control.
Transfer function models developed

Figure 1: Schematic of CSTR

The development of the initial model is based on four hypotheses:

1. It is perfectly agitated.
2. The mass densities of the input component and that produced are equal and identical
3. The volume of liquid in the reactor is kept constant by an exhaust duct.
4. The reaction level ( r ) is of order one and is given by: r= kCA
The mathematical model of the process comes from two balances: a balance of matter and a
balance of energy. This model is nonlinear. Linearization requires, first, the search for a
point of operation in steady state, which requires an iterative resolution of the two nonlinear
algebraic equations using the NewtonRaphson method.

Transfer function:

Open Loop model:

Figure 2: Open loop block diagram of CSTR with step input

 Figure 3: Open Loop temperature response of CSTR

Control Strategy
The study proposes the use of two controllers: a PID controller and a fuzzy controller to
regulate the concentration and temperature of the CSTR. The PID controller is classical
controller that uses a linear model to calculate the control output and relies on a mathematical
model of the process being controlled to calculate the control signal, which is based on the
error (difference between setpoint and process variable), integral of the error, and derivative
of the error. The parameters of the PID controller, namely the proportional gain, integral
time, and derivative time, are tuned in order to achieve the desired performance. On the other
hand, fuzzy controllers are a type of intelligent control method that use fuzzy logic to
calculate the control output. Fuzzy logic is a form of multivalued logic that deals with
approximate reasoning and uncertainty. The fuzzy controller uses a set of fuzzy rules that are
defined based on expert knowledge and experience. The rules relate the input variables
(usually the error and the rate of change of error) to the output signal. The membership
functions of the input variables are typically triangular or trapezoidal, and the output is
calculated based on the degree of membership of each rule.

Controller Structure
PID controller: The PID controller uses just one input variable, that is the error in
temperature (ET). It typically consists of three components: proportional, integral, and
derivative. Each component is calculated based on the error between the setpoint and the
process variable. The proportional gain determines the sensitivity of the controller to the
error. The integral component is proportional to the integral of the error and helps to
eliminate steady-state errors in the system, it determines the rate at which the integral
component accumulates and corrects for the error. The derivative component is proportional
to the rate of change of the error and helps to dampen the response of the controller.

Fuzzy controller:
The fuzzy logic controller used is of mamdani type.The fuzzy controller uses two input
variables- error in temperature (ET) and error in concentration (EC). The output variable is
the control signal for the heater (Q). The fuzzy controller has two fuzzy sets for each input
variable: negative (N) and positive (P), which are represented by triangular membership
functions. The fuzzy rules are defined using the 25 IF-THEN format, where the antecedent
of each rule is a combination of the two input variables and the consequent is the
corresponding output value.

Controller Parameters
PID Controller:
Initially, default parameters provided in Simulink were used:

Figure 4: block diagram of closed loop system without tuning

Here, P=1; I=1; D=0

Closed Loop Response (PID controlled) before tuning:

Figure 5: Concentration tracking for CSTR using PID controller without tuning
 Figure 6: Comparison of the open and closed loop responses

Automated tuning of PID Controller

Case 1: Less response time, more overshoot

Figure 7: PID tuned parameters and response for case 2

 Figure 8: Comparison of case 1 with open and closed loop responses

Case 2: Slightly more response time, less overshoot

Figure 9: Comparison of case 2 with open and closed loop responses

 Figure 10: Case 2 responses with time

Figure 11: Comparison of all PID responses

Fuzzy Logic control:

Figure 12: Fuzzy control block diagram

Figure 13: Input and output membership functions for fuzzy controller
Figure 14: Rules for fuzzy controller

Figure 15: Fuzzy control surface

The initial output of fuzzy control:

Figure 16 (a): Fuzzy control response using ode(Adams) solver and (b) as mentioned in the
paper

Time delay response

A time delay was added to the current transfer function of the plant and the PID controller
was tuned to control the output temperature of the process

The block diagram of the open loop system and the temperature output without the controller
are as follows:

Figure 17: Open loop block diagram of time delay system

 Figure 18: Temperature response of open loop time delay system

Blue: System’s temperature output

Yellow: Step input

After tuning the PID controller parameters, the following results have been obtained:

Figure 19: Closed loop block diagram of time delay system

Figure 20: PID controller tuning parameters for time delay system
 Figure 21: Temperature response of close loop time delay system

Since the rise time of this system is very large, the PID controller was returned and the
following parameters were used

Figure 22: PID controller tuning parameters after further control

 Figure 23: Temperature response after further tuning

In this system, the rise time has decreased but at the cost of overshoot which can not be
afforded in the case of heat-sensitive materials. Hence the optimum solution would be to use
a fuzzy logic controller which avoids overshoots.

Comparisons
The following reponses where obtained considering a step input of magnitude 100, in an
open loop system, a system with a PID controller and a system with a fuzzy logic controller
(in a non-time delay system)

Figure 24: Comparative graph of temperature response for reference, open loop, closed loop
PID and closed loop Fuzzy approaches

Clearly, fuzzy logic provides lesser rise time as well as no overshoot whereas the
conventional PID controller has a greater rise time and also lead to overshoot.
Errors faced
1. While trying to obtain a concentration output from the CSTR plant transfer function,
a demux block was used, which led to certain errors (inconsistent input and output
ports) and could not be implemented. Hence, just the temperature of the CSTR is
observed, measured and controlled.
2. The fuzzy logic controller did not produce the desired response since the system is
non-linear and has to be linearized, besides the rigorous trial and error required in
tuning it. In the PID controller, the auto-tuning itself linearizes the system.
Different solver functions were implemented as a part of the trial and error process.

Summary
This study introduces an innovative approach to designing and implementing an intelligent
controller to optimise the performance of the CSTR system. Through simulation results, it
was demonstrated that the proposed method offers a more intelligent way to determine the
response parameters of the PID and Fuzzy controllers for the CSTR system, resulting in
better dynamic performance. When compared with the traditional PID and Fuzzy methods,
this process showed significant improvement in performance by selecting appropriate fuzzy
rules. In the short term, to improve the system's robustness against disturbances and
modelling uncertainty, it is recommended to include an identification phase in the
continuous mini-reactor control scheme.

Conclusion
In the case of heat-sensitive materials, an overshoot in the temperature needs to be avoided at
any cost. Although the PID controller gives a much faster response (lesser rise time) and
equal settling time as compared to the fuzzy logic controller, it leads to an overshoot. A
fuzzy logic controller avoids this situation and hence is preferred in this case.

References

1. Implementation Of Temperature Regulation and Concentration Tracking of CSTR

with Fuzzy Controllers, Anirudh Damodar Kakule and Palhavi Kerkar,
Proceedings of the Second International Conference on Intelligent Computing
and Control Systems (ICICCS 2018) IEEE
2. Design and Implementation of PI and PIFL Controllers for Continuous Stirred
Tank Reactor System, Abdul Albagul, Y Abujeela and Mustafa Saad, WSEAS
Press, May 2014
Contribution

Bhavneet Kaur (B20CH012): Theory (Motivation, Literature, Applications, Controller

strategy, structure and parameters), Results, Errors, Summary, Conclusion, Simulink model
and temperature response for PID and fuzzy controllers in plant without time delay.

Divyanshu pandey(B20CH015): Theory (Motivation, Literature), Implementation of simulink

model for open loop system without time delay, Fuzzy controls and its applications.

Dwimalu Boro (B20CH017): Simulink model and temperature response for PID controllers
in plant with time delay, tuning of PID controller with time delay.

Ayush Abad (B20CH011): Theory (Motivation, Literature), Implementation of simulink

model for open loop system with time delay, PID control and its applications.