Professional Documents
Culture Documents
Abstract: This paper analyzes the performance of different controllers such as feedback, feedback plus
feed-forward and internal model controller to regulate the temperature of outlet fluid of a shell and tube heat
exchanger to a certain reference value. The transient performance and the error criteria of the controllers are
analyzed and the best controller is found out. From the simulation results, it is found out that the internal model
control outperforms feedback PID and feedback plus feed-forward controller.
KeyWords: Feedback controller, Feedback plus feed-forward controller, internal model controller
PC based Controller
Current to
DAQ Card
Compressor Pressure
(DAC)
Converter
4-20 mA
DAQ Card
(ADC)
3 - 15 psi
4-20 mA
Power Temperature
Source Transmitter
Kg/Sec Steam
Steam Input 2 Wire
RTD
Condensate
valve 1.6kg/ sec, time constant for control valve is 3 in Section 2.1 while developing the transfer function
sec, time constant for sensor is 10 sec. model of heat exchanger system.
From the experimental data linearized mathematical Transfer function model of heat exchanger system is
model of heat exchanger is developed.
50
Gp (s) = e1s (3)
30s + 1
2.2 Mathematical Model
To design a controller, a proper mathematical model Transfer function model of valve is
of the process has to be determined. Most of the
0.13
industrial system are non-linear in nature and can be Gv (s) = (4)
approximated as first order plus time delay (FOPTD) 3s + 1
or second order plus time delay (SOPTD) models.
Transfer function model of sensor is
The general form of FOPTD model can be expressed
as 0.16
Kp eD s H (s) = (5)
G (s) = (1) 10s + 1
s + 1
The general form of SOPTD model can be expressed Transfer function model of disturbance is
as
Kp eD s 1
G (s) = (2) Gd (s) = (6)
(1 s + 1) (2 s + 1) 10s + 1
Here Kp is the process gain, D is the time delay, The process transfer function is represented as
is the time constant of FOPTD system, 1 and
2 are the time constant of SOPTD system. The 5e1s
parameters are obtained from open loop step response G (s) = (7)
90s2 + 33s + 1
data or frequency response data. The time delays are
measured from the step response data. which is in the form of SOPTD represented in Eq. 2
This paper considers the experimental data mentioned
+
r(s) + Feedback
u(s)
Actuator Process
y(s)
Sensor
Controller
d(s)
Flow
Disturbance 1.4
u(s) +
r(s) + Internal Model
Actuator Process
y(s) 1.2
Controller
1
Process
Model
+ 0.8
Output
Sensor
0.6
Figure 6: Block diagram of internal model controller
0.4
1
Q (s) = Gp (s) (12)
(s + 1)n Figure 7: Set point and load disturbance response
using PID controller
4 Simulation Results
To control the temperature of a shell and tube heat controller shows 29.56% of overshoot and 115.2 sec
exchanger system different controllers are used and of settling time.
the simulated studies of the controller performance is Due to the high overshoot of classical PID
discussed in this section. Performance assessment of controller, feed-forward controller is added with
industrial controller is one of the widely researched feed back controller. The combination of feedback
area which determines the performance of the plus feed-forward controller reduces the overshoot
controller by various methods [16]. Oscillations in to 25.1%. The unit step response of feedback
process control loop is determined using different plus feed-forward controller for temperature control
parameters summarized below. The methods of of heat exchanger system is shown in Fig. 8.
oscillation detection was first introduced by [17]. Due to relatively higher overshoot of feedback plus
Some of the parameters used to evaluate the
performance of control loops are
1.4
0.6
IT AE = t |e (t)|dt (15)
0 0.4
IT SE = t2 e (t)dt (16) 0.2
0
0
0 100 200 300 400 500 600
Classical PID controller tuned using Time (sec)
Zigler-Nichols tuning method is used to control
the output temperature of heat exchanger. Set point
tracking and disturbance rejection of the feedback Figure 8: Set point and load disturbance response
controller is shown in Fig. 7. The feedback PID using feedback plus feed-forward controller
0.2
References:
0
0 50 100 150 200 250 300 350
Time in sec [1] L. Ljung, System identification. Springer, 1998.
[2] S. Skogestad, Tuning for smooth pid control
Figure 9: Set point response using internal model with acceptable disturbance rejection,
controller Industrial & engineering chemistry research,
vol. 45, no. 23, pp. 78177822, 2006.
The transient response (peak overshoot and
settling time) in unit step response of all the [3] V. R. Segovia, T. Hagglund, and K. Astrom,
controllers (feedback, feedback plus feed-forward Measurement noise filtering for pid
and internal model controller) is summarized in controllers, Journal of Process Control,
Table 1. The error response of all the controllers vol. 24, no. 4, pp. 299313, 2014.
(feedback, feedback plus feed-forward and internal
[4] O. Garpinger, T. Hagglund, and K. J. Astrom,
model controller) is summarized is tabulated in Table
Performance and robustness trade-offs in pid
2.
control, Journal of Process Control, 2014.
[5] K. H. Ang, G. Chong, and Y. Li, Pid control
Table 1: Results for transient response of controller system analysis, design, and technology, IEEE
Controller Overshoot Settling Time Transactions on Control Systems Technology,
Feedback PID 29.56% 115.2 sec vol. 13, no. 4, pp. 559576, 2005.
Feedback plus Feed-forward 25.1% 91.3 sec
[6] W. Tan, J. Liu, T. Chen, and H. J. Marquez,
Internal Model Controller 1.13% 77.79 sec
Comparison of some well-known pid tuning
formulas, Computers & chemical engineering,
vol. 30, no. 9, pp. 14161423, 2006.
Table 2: Results for error indices of controller [7] B. W. Bequette, Process control: modeling,
Controller IAE ISE ITAE ITSE design, and simulation. Prentice Hall
Feedback PID 5.55 0.3 610.8 11.75 Professional, 2003.
Feedback plus Feed-forward 4.14 0.25 340.1 5.107
[8] D. Seborg, T. F. Edgar, and D. Mellichamp,
Internal Model Controller 3.58 0.18 279.5 4.729
Process dynamics & control. John Wiley &
Sons, 2006.
From Table 2 it is observed that the error indices
(IAE, ISE, ITAE and ITSE) decreases as the overshoot [9] R. K. Mudi and N. R. Pal, A robust self-tuning
and settling time decreases. scheme for pi-and pd-type fuzzy controllers,