Lab Experiment 2: Regulation of a Heat

Exchanger (Warm Water Control)

CEE30007 Process Control and Measurements

Name Student ID
Charmaine Garnett Agan 100071394
Chong Siaw Yen 100071954
Emilly Nie anak Lint 100080798
Gracylla Rose anak Guncheng 4334914

Semester 1, 2019
1.0 Summary
Proportional-Integral-Derivative (PID) controllers has become part in the process control
system and predominant types of feedback control. Furthermore, PID controller has been
used in industry due to its simplicity and easy-tuning. This lab experiment is conducted to
control temperature by feedback and ensure that warm water can be controlled in the system.
Another objectives are to determine the optimum PID controller parameters of system to the
desired set point by feedback control and to compare the performance between Ziegler-
Nichols (closed loop) and Cohen-Coon (open loop) that will be used in the selection and
tuning of controller. In this report also covers the issues related to the process in terms of
safety, environmental and economic. Besides, the performance of Ziegler-Nichols method in
Figure 8 did not perform better compared to Cohen-Coon method in Figure 12 where the
response of output process is stable. Thus, conclude that Cohen-Coon method is more
suitable in tuning the PID controller compared with Ziegler-Nichols method. Additionally,
this can reduce the energy consumption and reduce the output cost.

2.0 Control Objectives

One of the main control objectives of this lab experiment is to control temperature by
feedback and ensure that warm water can be controlled in the system. Another objective is to
determine the optimum PID controller parameters of system to the desired set point by
feedback control. In this lab experiment, Ziegler-Nichols and Cohen-Coon method are used
in tuning the feedback controller. Apart from that, comparing the performance analysis
between Ziegler-Nichols (closed loop) and Cohen-Coon (open loop) will be used in the
selection and tuning of controller.
3.0 Sketch Control and Instrumentation Diagram

Figure 1: Shell and tube heat exchanger system (Padhee et al. 2011).

4.0 Issues Related to Process

Safety

It is safe to conduct this experiment because it did not involve any hazardous chemicals to
operate the equipment. Furthermore, control system is needed to ensure safety (essential
requirements) when conducting the experiment. This is because control systems can protect
the equipment from any damages due to excessive load. Apart from that, even if there is any
leakage happens, the process in the system is still safe and does not cause an explosion since
water is used in the process. Beware of hot water valve and keep hands off from the hot area
on heat exchanger. This is to ensure no accident happens in the lab. Moreover, the allowed
operating conditions of equipment is the “Output” value should be at 40% and the output
signal of PID controller should be at between 4 mA to 20 mA.

Environmental

There is no environmental issue caused by the process. This is because the process
contributes to only zero emissions and keep water as feedstock.
Economic

No extra materials nor do extra equipment needed in the process. All the equipment needed
are all in the lab. Besides, the process should be operated at the allowed operating conditions
in order to reduce wastage of source. In addition, obeying proper procedures and precaution
to handle the process will prevent from any breakdown of equipment and this will help to
reduce the maintenance cost of equipment.

5.0 Develop Transfer Function

Energy balance of a heat exchanger

T1 = Temperature hot in, T2 = Temperature cold in, T3 = Temperature hot out, T4 =

Temperature cold out

d
[ρVcp (T4 + T3)] = (F1cpT1 + F2cpT2) – (F3cpT3 + F4cpT4)
dt

d 1
(T4 + T3) = [(F1T1 + F2T2) – (F3T3 + F4T4)], where F1 = F3 and F2 = F4
dt ρv

d
ρv (T4 + T3) = (F1T1 + F2T2) – (F1T3 + F2T4)
dt

d
ρv (T4 + T3) = F1 (T1 – T3) + F2 (T2 – T4)
dt

d F F
ρv (T4 + T3) = 1 (T1 – T3) + 2 (T2 – T4)
dt ρv ρv

F1 F
(T4’ (S) + T3’(S)) S = (T1’(S) – T3’(S)) + 2 (T2’(S) – T4’(S))
ρv ρv

F1 F
(T4'(S) + T3’(S)) = (T1’(S) – T3’(S)) + 2 (T2’(S) – T4’(S))
ρvS ρvS

F1
Gd (S) = ; D(s) = T1 – T3
Sρv

F2
Gp (S) = ; M (S) = T2 – T4
Sρv

Proportional Controller
Closed loop transfer Function CLTF

GpsGfsGcs Gds
Y(S) = Y sp (S) + D
1+ GpsGfsGcsGms 1+ GpsGfsGcsGms (S)

Assume Gc (S) = Kc, Gm (S) = Gf (S) = 1,

F 2 Kc
GpsKc Gds F1
Y(S) = Y + D = F2 Y + (T -T )
1+ GpsKc sp ( S ) 1+ GpsKc (S ) Sρv 1+ (
Sρv )
Kc sp (S) Sρv 1 3

F 2 Kc/ F 2 Kc F 1/ F 1 Kc
Y(S) = S ρV F 2 Kc Y sp (S) + SρV F 1 Kc (T1 – T3)
+ +( )
F 2 KC F 2 Kc F 1 Kc F 1 Kc

1 Kp
Y(S) = Y sp + (T – T3)
τ P1+1 τ P2+1 1

F1 S ρV SρV
where Kp = , τ P1 = , τ P2 =
F 1 Kc F 2 Kc F 1 Kc

Proportional Integral (PI)

Figure 2: PI controller result from Ziegler-Nichols method.

Based on Figure 2, the process variables had reach set setpoint of 50°C at the end of the
system when Kcrit value was 1.0 and period of sustained oscillation, τp was 57s. The students
assumed that when the gain increased, the output will show declination in the system. Thus,
higher value of gain will let the controller output to start to oscillate.

Transfer function for PI

Laplace transforms for this system

F1 F
(T4'(S) + T3’(S)) = (T1’(S) – T3’(S)) + 2 (T2’(S) – T4’(S))
ρvS ρvS

Thus, the transfer functions for the process;

Kp F2
Gp (S) = =
τ p S+1 Sρv

Kc
PI control of Gc (S) = ,
τ1 S

Gp( s) Gf ( s)Gc(s)
General form of CLTF: Y(s) = Y sp (S)
1+ Gp(s)Gf (s)Gc( s)Gm (s)

Assume Gm (S) = Gf (S) = 1

Substitute in the values of Gm (S), Gf (S), Gc (S) and Gp (S) into CLTF equation.

F2 Kc
(
)(1)( )
Sρv τ1 S
Y(s) = Y sp (S)
F2 Kc
1+( )(1)( )( 1)
Sρv τ1 S
1 1
(S )( )
τp τ1 S ρv
Y(s) = Y sp (S), where τp =
1 1 F 2 Kc
1+( S)( )
τp τ1 S
1
Y(s) = Y sp (S), where τ = (τI.)(τp)
(τp)( τ ¿¿ 1) S 2+ 1¿
1
Thus, Y(s) = Y sp (S)
τ S 2+ 1

⸫ Increased integral control influenced the order of dynamics in this process.

6.0 Analyse response of Tuned PID Controller

Figure 3: P controller when gain = 1.4.

Figure 3 above showed that the process variable was unstable when gain value was set to 1.4
and the output reached its high oscillation.

Figure 4: P controller when gain = 1.2.

Based on Figure 4 above, the response of process variable was almost stable when gain value
is 1.2. Besides, small oscillation of output variable was also obtained.

Figure 5: P controller when gain = 1.0.

Figure 5 showed when gain value was set to 1.0, the process variable had reached its steady
state. It showed that the observed response of variable was mostly stable. Therefore, the most
suitable gain value in this process would be 1.0.

Proportional Action Controller (P)

Figure 6: P controller when K critical = 1.0.

Based on Figure 6 showed that using the Ziegler-Nichols method, the K c value was set to 0.5
from the K critical value of 1.0. Besides, there was small offset occurred between the process
variable and the set point. Thus, it showed that the offset presented in the system when using
P controller.

Figure 7 showed that when Kc value was 0.5 and τI was 48s, the response of the system was
stable. In addition, zero offset obtained between the process variable and the setpoint at the
end of the system.

Proportional Integral Action and Derivative Controller (PID)

Figure 8: PID controller when Kc = 0.6, τI = 29s and τD = 7.1s .

The main aim of using PID is to reduce the oscillations. Based on Figure 8 above showed
that the oscillations occurred were tolerable thus the use of derivative control action for
oscillation reduction was unnecessary. However, there were no offset occurred and the
process variable was equal to the set point in the system.
7.0 Demonstrate Cohen and Coon Method

Figure 9: Process reaction curve for the shell and tube heat exchanger (62.3% method).

The output variable response when reached 62.3% of its ultimate value,

0.623 ×0.4=0.2492≈ 0.25

Tp = 0.25
In order to achieve the steady state at initial of the system, set point was set to 45 ˚C.
Moreover, the value of K, τ and τd were calculated.

τ = 12s τd = 0.1s

∆ output 0.4
K= = =0.01
∆ input 30

Proportional Only Controller (P)

 Figure 10: P controller used when Kc = 233.3.

Figure 10 above showed when using P controller in the system, K c value was 233.3. This
showed that high oscillation obtained in the system and that the process variable remained
unstable.

Proportional and Integral Action Controller (PI)

Figure 11: PI controller used when Kc = 188.3 and τI = 1.0.

Figure 11 showed that the process variable was almost stable when Kc = 188.3 and τI = 1.0.
However, small offset occurred, and high oscillation was observed in the system.

Proportional Integral Action and Derivative Controlle (PID)

 Figure 12: PID controller used when Kc = 291.7, τI = 1.0 and τd = 1.0.

Based on Figure 12 obtained, the oscillation decreased when Kc = 291.7, τ I and τd remained
constant at 1.0. Thus, this achieved the main purpose of using PID which to reduce the
oscillation in the system. However, there was still offset remained and the process variable
considered to be stable.

8.0 Comparison between Ziegler-Nichols Method with Cohen

Coon Method
Ziegler-Nichols method was set at automatic mode while Cohen-Coon method was set at
manual mode. Based on experimental data analysis, Cohen-Coon method was better option to
Ziegler-Nichols methods in tuning PID controller. This can be seen where the response of the
system using Ziegler-Nichols method (Figure 8) did not perform better compared to Cohen-
Coon method (Figure 12). Besides, in terms of time taken in the system, Cohen-Coon
method (12s) was faster than Ziegler-Nichols method (57s). This is due to the time taken in
Ziegler-Nichols method was calculated from one peak to another peak of oscillation in the
system. Whereas, Cohen-Coon method was calculated starting from the process variable start
to rise until became a steady-state. Thus, conclude that Cohen-Coon method is more suitable
in tuning the PID controller compared with Ziegler-Nichols method. Additionally, this can
reduce the energy consumption and reduce the output cost.

Furthermore, the analysis of comparing the Ziegler-Nichols method with Cohen-Coon

method can be better by determining or calculating the percentage overshoot in the process
system. By doing this, it can conclude more that Cohen-Coon method is preferable as it has a
faster time taken and high overshoot in the settling time compared with Ziegler-Nichols
method where it reached high overshoot at only minimum settling time to get ideal response
of the system (Joseph et. al 2018).

Reference
Joseph, EA and Olaiya, OO 2018, ‘Cohen-Coon PID Tuning Method: A Better Option to
Ziegler Nichols-Pid Tuning Method’, Computer Engineering and Intelligent System, vol. 9,
no. 5, pp. 33–37.
Padhee, S, Khare, YB & Singh, Y 2011, ‘Internal model based PID control of shell and tube
heat exchanger system’, TechSym 2011 - Proceedings of the 2011 IEEE Students’
Technology Symposium, pp. 297–302.