IOP Conference Series: Materials Science and Engineering

PAPER • OPEN ACCESS Related content

- Design of Linear-Quadratic-Regulator for a
Self-tuning regulator for an interacting CSTR CSTR process
P R Meghna, V Saranya and B Jaganatha
process Pandian

- Biodiesel Production using Heterogeneous

Catalyst in CSTR: Sensitivity Analysis and
To cite this article: Niraj Rajendra Mungale et al 2017 IOP Conf. Ser.: Mater. Sci. Eng. 263 052012 Optimization
L S Keong, D S Patle, S R Shukor et al.

- Performance of continuous stirred tank

reactor (CSTR) on fermentative
biohydrogen production from melon waste
View the article online for updates and enhancements. K Cahyari, Sarto, S Syamsiah et al.

This content was downloaded from IP address 47.30.167.231 on 17/12/2019 at 16:51

 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

Self-tuning regulator for an interacting CSTR process

Niraj Rajendra Mungale, Akshay Upadhyay and B Jaganatha Pandian

School of Electrical Engineering, VIT University, Vellore - 632014, Tamil Nadu,

India.

E-mail: [email protected]

Abstract: In the paper we have laid emphasis on STR that is Self Tuning Regulator and its
application for an interacting process. CSTR has a great importance in Chemical Process when
we deal with controlling different parameters of a process using CSTR. Basically CSTR is
used to maintain a constant liquid temperature in the process. The proposed method called self-
tuning regulator, is a different scheme where process parameters are updated and the controller
parameters are obtained from the solution of a design problem. The paper deals with STR and
methods associated with it.

1. Introduction
Mostly the chemical process is non-linear by nature. A description of various Continuous Stirred Tank
Reactor Control schemes is shown below:
1.1. Gain Scheduling:
Here in this process the gain is measured then it schedules the controller for compensating process gain
changes. Following is the block diagram of the above scheme:

Figure 1. Gain Scheduling

Inner loop is made up of process & controller and the outer one is used to update the controller
parameters on the basis of the process working conditions [5].

Content from this work may be used under the terms of the Creative Commons Attribution 3.0 licence. Any further distribution
of this work must maintain attribution to the author(s) and the title of the work, journal citation and DOI.
Published under licence by IOP Publishing Ltd 1
 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

1.2. Model Reference Adaptive System:

In this scheme a closed loop controller is constructed with various parameters which can be modified to
get the desired response of the system [5]. The output is then compared with the reference model
desired output and an ERROR is generated.

Figure 2. MRAC

1.3. Self Tuning Regulator

Figure 3. STR

STR is an indirect method of controlling process. In many processes the dynamics of the process and
disturbances are unknown & so it becomes difficult to find the controller parameters. Therefore we need
to estimate the process parameters [4]. Such a controller can tune the parameters on its own to get the
requirements of a closed loop system.

2
 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

1.4. Adaptive Control:

Figure 4. Adaptive Control

We must know what happens if the process fluctuates from the point of operation. It may lead to
efficiency issues of the controllers. Thus we need to update and tune the parameters of the controller
again to overcome such a drawback [1]. For this we use an automatic tuning/updating scheme and hence
the controller is termed as Adaptive Controller.
The objective of our project is to validate a Self-Tuning-Regulator approach for controlling a high
dimensional control problem.

2. Methodology

2.1. System Description

System under consideration here is Interacting CONTINUOUSLY STIRRED TANK REACTOR

(CSTR). Schematic representation of CSTR system can be shown in the diagram below.

Figure 5. Interacting CSTR schematic

Since Self Tuning Regulator is to be considered, there is no need for linearization of model. Thus, direct
differential equation model for interacting CSTR system can be used. Ordinary Differential Equation
(ODE) model is used for CSTR system. It has following differential equations governing the system:

3
 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

( ) ( ) ( )

( ) ( )

( ( ))

Input in this case is Coolant flow rate. For model considered in this paper, it is 100 L/min. However,
since simulation is to be done on MATLAB, random white Gaussian noise is added in the input, so as to
get random input and random output data for estimation process. System considered here is higher
dimensional, hence linearzing it is a difficult task which cannot give sufficient performance satisfaction
[2]. Thus, direct estimating system parameters from given input output data, makes it faster and reliable.
Assuming numerator order 1 and denominator order 4 for given system, we run Recursive Least Square
algorithm for obtained I/O data to get system parameter. i.e. Transfer function of unknown system.
Parameter to be controlled here is temperature in Tank 2 for disturbances in input. Thus transfer
function obtained is with respect to temperature in tank two.
2.2. Controller Design

Controller used here is Self Tuning Regulator (STR). Main block diagram of system with STR can be
shown as,

Process
Parameters

Figure 6. Block diagram of system with STR

First part of any STR is an estimator. Estimator is a system which treats entire process as a black box
and tries to identify the system parameters with the help of input output data. As system parameters are
very much important part of STR, main emphasis should be on estimation of process parameters/
system parameters.

4
 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

Estimator used here is Recursive Least Square Estimator (RLS Estimator). Main feature of RLS
estimator is it minimizes sum of error squares. It takes input and output data as input, and also takes
desired system model as an input. It gives system parameters, in terms of numerator polynomial,
denominator polynomial and transportation delay [3]. Thus we can construct process transfer function
from RLS estimator output. As the process involved here is higher dimensional and not linearized, we
use numerator order as 1 and denominator order as 4. This gives almost exact performance for system
around current operating point.
Next stage of STR is controller design. We assume model structure which system needs to follow. Let
that model structure be Gm. From estimator model received is Gp. With the help of controller, we can
make system to follow Gm even though original transfer function is Gp. By using Minimum Degree Pole
Placement (MDPP) technique, controller can be designed as follow:
System Structure for MDPP is-

Figure 7. Block Diagram for MDPP

Using MDPP, R, S, T is designed which provide main function of controller. Then next stage is
developing control law for STR. As it can be seen from above block diagram, control law for STR
becomes,

Thus main part is designing R, S and T vectors, which will give controller a proper structure.
After designing R, S, T, we repeat steps from estimation up to building control law, for each sampling
instance.
3. Results

Input output data is used for estimation of process parameters. As, the process is considered to be 4th
order, expected transfer function is in the form of

And estimated results from RLS algorithm are

5
 14th ICSET-2017 IOP Publishing
IOP Conf. Series: Materials Science and Engineering 263 (2017) 052012 doi:10.1088/1757-899X/263/5/052012
1234567890

To verify whether estimated model gives satisfactory performance, same input was given to estimated
model but without noise added. Results for that simulation are as following-

Fig8. Simulation results for both, actual process and estimated transfer function.

As it can be observed, red graph is for original process, for coolant flow rate of 100 L/min. same input
was fed to estimated transfer function, and simulated in MATLAB. That result is in blue graph. It can be
observed that, both systems give almost identical performance, showing accuracy of RLS algorithm
used.

References:

[1] Aström K J, Borisson U, Ljung L and Wittenmark B 1977 Theory and applications of self-
tuning regulators Automatica 13(5) 457-476
[2] Prabhu K and V Murali Bhaskaran 2013 Optimization of a temperature control loop
using self-tuning regulator International Journal of Computer Applications 61 9
[3] Rahmat and Mohd Fua'ad 2011 Temperature Control Of A Continuous Stirred Tank
Reactor By Means Of Two Different Intelligent Strategies International journal on
smart sensing & intelligent systems 4 2
[4] Chopra, Vikram, Sunil K Singla and Lillie Dewan 2014 Comparative analysis of tuning
a PID controller using intelligent methods ACTA Polytechnica hungarica 11 (1)
235-48
[5] Åström K J and Wittenmark B 2013 Adaptive control Courier Corporation