International Journal of

IS Systems , Algorithms &

J AA Applications

Design and Simulation of Fuzzy Controller for Heat Exchanger

1
Debi Prasad Dash, 2Mrs.Amruta S.Deshpande
1,2
Department of Instrumentation and Control
College Of Engineering Pune, Pune, India
e-mail: [email protected], [email protected]

Abstract - Shell and tube heat exchanger system is widely used sical PID controller is implemented in a feedback control loop
in chemical plants because it can sustain wide range of temper- so as to achieve the control objectives. PID controller exhibits
ature and pressure. The main purpose of a heat exchanger sys- high overshoots which is undesirable. Generally PID control-
tem is to transfer heat from a hot fluid to a cooler fluid, so tem- ler is only suitable for linear system and requires that the
perature control of outlet fluid is of prime importance. In this
mathematical model of controlled object is exact enough,
paper we design a fuzzy logic controller (FLC) for shell and
tube heat exchanger system, uses MATLAB as a visual model- while heat exchanger system in itself has traits of big delay,
ling tool and simulation research based on SIMULINK. The time varying, nonlinear and much destabilization which makes
purpose of this controller is to maintain a uniform temperature it difficult to build mathematical models. But fuzzy control
at the output of the heat exchanger system, there by varying the neednt build mathematical model of controlled objects, it has
input flow rate. The control action taken by the FLC is based benign robustness and nonlinear control characteristics and
on the deviation from set point i.e. error and rate of change of can take more effective control of time-varying, nonlinear and
error. The designed fuzzy logic controller for the heat exchang- complex controlled objects. Hence a fuzzy logic controller
er system is then compared with a conventional PID controller. (FLC) is proposed in this paper for control of output tempera-
The result shows that FLC has less overshoot and settling time
ture of heat exchanger, with parameters of control as error,
than conventional PID controller.
change in error. The implementation of FLC has reduced the
Keywords: FLC, PID, Shell and tube heat exchanger. overshoot further. Tuning of proposed FLC is very easy as a
gain at the output is used to tune the controller. Simulation is
I. INTRODUCTION done with the help of SIMULINK software. Simulation result
In practice, all chemical processes involve production or shows that fuzzy controller has good settling time and reduced
absorption of energy in the form of heat. Heat exchanger is overshoot as compared to the PID controller.
commonly used in chemical processes to transfer heat from a
hot fluid through a solid wall to a cooler fluid. There are II. SHELL AND TUBE HEAT EXCHANGER SYSTEM
different types of heat exchanger used in the industry but A typical interacting chemical process for heating consists of
most of the industry use shell and tube type heat exchanger a chemical reactor and a shell and tube heat exchanger sys-
system [1]. tem. The process fluid which is the output of the chemical
reactor is stored in the storage tank. The storage tank sup-
The principle of operation is simple. Two fluids of different plies the fluid to the heat exchanger system using a pump
temperature are brought into close contact but are prevented and a non returning valve. The heat exchanger heats up the
from mixing by physical barrier. The temperature of two fluid to a desired set point using super heated steam at 180C
fluids will tends to equalize and this is because of the heat supplied from the boiler. The super heated steam comes from
transfer taking place between them. Based on heat transfer the boiler and flows through the tubes, whereas, the process
efficiency, the cross flow shell and tube heat exchanger has fluid flows through the shells of the shell and tube heat ex-
proven efficient than its counter parts. They have large ratios changer system.
of heat transfer surface to volume than double-pipe heat ex-
changers, and they are easy to manufacture in a large variety Different assumptions have been considered in this research
of sizes and configurations. They can operate at high pres- paper. The first assumption is that the inflow and the outflow
sures, and their construction facilitates disassembly for peri- rate of fluid are same, so that the fluid level is maintained
odic maintenance and cleaning. Shell-and-tube heat ex- constant in the heat exchanger. The second assumption is the
changers find widespread use in refrigeration, power genera- heat storage capacity of the insulating wall is negligible
tion, heating and air conditioning, chemical processes, manu- [6].The objective is to maintain the outlet temperature of the
facturing, and medical applications. A shell and tube heat process fluid, C, at its desired value or set point, rC, in
exchanger is an extension of the double-pipe configuration. the presence of disturbances in the process fluid flow, qm(kg/
Instead of a single pipe within a larger pipe, it consists of a sec), flowing inside the tubes of the heat exchanger, and inlet
bundle of pipes or tubes enclosed within a cylindrical shell temperature iC. The steam flow, qms(kg/sec), is the varia-
[2]. ble that can be adjusted to control the outlet temperature. In
this feedback process control loop, the controller is reverse
This paper reports a work that considers a shell and tube heat acting , the valve used is of air to open (fail-close) type. A
exchanger and builds a single input-single output model of the thermocouple is used as the sensing element, which is imple-
system with the help of experimental data. The outlet tempera- mented in the feedback path of the control architecture. The
ture of the heat exchanger system has to be kept at a desired temperature of the outgoing fluid is measured by the thermo-
set point according to the process requirement. Firstly, a clas- couple and the output of the thermocouple (voltage) is sent
to the transmitter unit, which eventually converts the thermo-

Volume 3, Issue ICRAET13, March 2013, ISSN Online: 2277-2677 83

 Design and Simulation of Fuzzy Controller for Heat Exchanger International Journal of
IS Systems , Algorithms &
J AA Applications

couple output to a standardized signal in the range of 4-20 the transfer function block diagram of feedback control of
mA. This output of the transmitter unit is given to the control- shell and tube heat exchanger system.
ler unit. The controller implements the control algorithm,
compares the output with the set point and then gives neces- Table II. Transfer Function of Various Parts of Process
sary command to the final control element via the actuator
unit. The actuator unit is a current to pressure converter and 50
Transfer function of process
the final control unit is an air to open (fail-close) valve. The 30 s + 1
actuator unit takes the controller output in the range of 4-20
mA and converts it in to a standardized pressure signal, i.e. in 0 .1 3
Transfer function of valve
the range of 3-15 psi. The valve actuates according to the con-
3s + 1
troller decisions. Fig. 1 shows the basic feedback control
scheme implemented in a shell and tube heat exchanger sys- Gain of valve
tem. 0.13

Gain of current to pressure converter

0.75

Transfer function of disturbance variables 1 3

(flow and temperature disturbances re- ,
spectively) 30 s + 1 30 s + 1

0.16
Transfer function of thermocouple
10 s + 1

Fig.1 . Shell and tube heat exchanger system control scheme

III. MATHEMETICAL MODELING

In this section, the heat exchanger system, actuator, valve, sen-
sor are mathematically modelled using the available experi-
mental data [7]. The experimental process data is shown in Ta-
ble I .
Table I: Process Data
Exchanger response to the steam flow gain 50C/(kg/sec)
Fig. 2. Feedback control of shell and tube heat exchanger system
Time constants 30 sec

Exchanger response to variation of process fluid IV. CONTROLLER DESIGN

1 C/(kg/sec)
flow gain The design of FLC consists of several steps. First, the variables
Exchanger response to variation of process tem-
for the fuzzy control system are determined. The universe of
3C/C discourse for all the variables involved are then set. Here, the
perature gain
FLC is designed with two input variables and one output varia-
Control valve capacity 1.6 kg/sec for steam
ble. The input variables are error(e) and change in error(e) that
Time constant of control valve 3 sec is difference between the set point temperature and the actual
temperature and the rate of change of error and output is con-
The range of temperature sensor 50C to 150C troller output [5].The design is based upon using e-e space
consisting of the plot e(t) vs e(t) [3]. Fig. 3 shows the FLC to
Time constant of temperature sensor 10 sec
control the temperature in the outlet of the shell and tube heat
exchanger system with error and change in error as input.
From the experimental data, the transfer functions and gains
are shown in Table II. The input variables and the output variable is designed using
five membership functions. Triangular membership function is
To control the exit temperature of the heat exchanger system a used for both input and output variables. The rule table is
classical feedback controller is used. The PID control algo- formed in such a way that the table itself became fuzzy with a
rithm remains the most popular approach for industrial pro- zero main diagonal separating positive and negative fuzzy rules
cess control despite continual advances in control theory. This that are also partitioned into sub fuzzy rules among themselves.
is not only due to the simple structure which is conceptually This is shown in Table III. The rule base developed for this
easy to understand and, which makes manual tuning possible, system is based on MAX-MIN inference. The MAX-MIN
but also to the fact that the algorithm provides adequate per- method of composition effectively expresses the approximate
formance in the vast majority of applications. Fig. 2 represents and interpolative reasoning used by humans when they employ

Volume 3, Issue ICRAET13, March 2013, ISSN Online: 2277-2677 84

 International Journal of
Design and Simulation of Fuzzy Controller for Heat Exchanger
IS Systems , Algorithms &
J AA Applications

linguistic propositions for deductive reasoning [8]. That is MIN

is used for the 'and' conjunction and 'MAX' is used for the 'OR' Table IV. Comparison of PID and Fuzzy Controller
conjunction. For this system MIN conjunction is used [4]. Cen- Settling Time
troid method is used for defuzzyfication . S.NO Controller Overshoot (%)
(sec)
1 PID 180 90

2 Fuzzy 14 80

The PID parameters by Ziegler-Nichols tuning are P=14.64,

I=0.976, D=54.9. The gain which is used at the output of fuzzy
controller to match the signals with the predefined membership
parameters is 1.23.

Fig. 3. FLC to control the temperature in a heat exchanger system

Table III. Fuzzy Rule Table

e
Positive
Negative Big Negative Zero Positive
Small
(NB) Small (NS) (ZZ) Big (PB)
(PS)

Positive
Big(PB)
ZZ PS PS PB PB

Positive
NS ZZ PS PS PB
Fig. 4. Comparison of Fuzzy Logic Controller with PID Controller
Small (PS)

VI. CONCLUSION
e Zero(ZZ) NS NS ZZ PS PS
A Fuzzy Logic Controller is designed for a crossed flow shell
Negative and tube heat exchanger system and its performance is com-
Small (NS)
NB NS NS ZZ PS pared with conventional PID controller. The performance com-
Negative
parison results prove that the FLC has a lesser settling time and
Big(NB)
NB NB NS NS ZZ overshoot when compared with PID controller. Thereby FLC
outperforms conventional PID control.
The fuzzy sets are designed such that the response is faster, REFERENCES
smoother and rules are fired for all values within the range. The [1] Orlando Duran et.al, "Neural Networks for Cost Estimation of
rule is developed based on experts knowledge that is by thor- Shell and Tube Heat Exchangers, " in Proc. IntI Multi Corif-
oughly understanding the problem defined and the complica- Eng. Com put. Scient., vol II, pp. 1584-1589, Mar 2008.
tions involved; based on which rules are fired. Or by means of [2] Anton Sodja et.al, "Some Aspects of Modeling of Tube-and-
simple trial and basis also rules can be fired. A gain block is Shell Heat-Exchangers," in Proc of 7th Modelica Corif-, Italy,
used after the controller output to match the signals with the pp. 716-721, Sep 2009.
predefined membership parameters. In order to prevent zero [3] I.H. Altas , A.M. Sharaf, A Generalized Direct Approach for
Designing Fuzzy Logic Controllers in Matlab/Simulink GUI
division that causes simulation problems such as delayed simu- Environment, International Journal of Information Technology
lation time and simulation hanging, a signal route has been pro- and Intelligent Computing, Int. J. IT&IC no.4 vol.1, 2007.
vided. [4] Mridul Pandey, K. Ramkumar , V. Alagesan , Design of Fuzzy
Logic Controller for a Cross Flow Shell and Tube Heat-
V. RESULTS AND DISCUSSION Exchanger, IEEE-International Conference On Advances In
In order to validate control performance of the fuzzy controller Engineering, Science And Management (lCAESM -2012)
designed in this paper, it is compared with the conventional PID March 30, 31, 2012.
controller. To evaluate the performance of the different control- [5] Abdul Aziz Ishak, Dinah Fadhilah Nasir , Sharliza Mohamad,
lers this paper has considered two vital parameters of the step Tuning of an Industrial Fuzzy Logic Controller, IEEE, 978-1-
4244-7240-6/10 ,2010.
response of the system. The first parameter is the maximum
[6] Subhransu Padhee, Yuvraj Bhushan Khare, Yaduvir Singh,
overshoot and the second parameter is the settling time. A com- Internal Model Based PID Control of Shell and Tube Heat
parative study of their performance has been shown in the Table Exchanger System, IEEE ,978-1-4244-8943-5/11 ,2011.
IV below .The simulation result is shown in Fig. 4. [7] M. Gopal, Control Systems Principles and Design, Tata Mc
Graw Hill, 2007.
Table IV shows that overshoot of PID is 180%.To reduce this [8] Timothy J. Ross, Fuzzy Logic With Engineering Applica-
high overshoot FLC is implemented. With the implementation tions,Wiley India,2008.
of FLC overshoot reduced to 14%, an improvement of 92 %.
Settling time of FLC is also less than PID.

Volume 3, Issue ICRAET13, March 2013, ISSN Online: 2277-2677 85