Tuning PID Controllers using Internal Model Control

Jayanthi Vijay Sarathy, M.E, CEng, MIChemE, Chartered Chemical Engineer, IChemE, UK

Abstract: There is no engineering system makes IMC particularly appealing is that it

today where controllers are not used. presents a methodology for designing Q-
Controllers are everywhere. To successfully parameterized controllers that has both
keep a controller in doing what it is supposed fundamental and practical appeal. As a
to do, controllers need to be tuned properly consequence, IMC has been a popular design
and hence the name 'Controller Tuning'. procedure in the process industries,
There are enough text books out there to particularly as a means for tuning single loop,
explain the theory and complex equations PID-type controllers. The Author uses a case
behind controller tuning but from a practical study to demonstrate this.
stand point in the industry, it becomes Case Study
increasingly difficult to apply them in its raw
An inlet stream of 100 kgmol/hr at 5 barg
form with a paper, pencil and calculator while
containing 100 mol% methane passes
on the job.
through four valves. The pressure drop
The following article is based on using a well experienced at the four valves (VLV-
known software tool called Aspentech 100/VLV-101/VLV-102/VLV-103) is 0.1 bar,
HYSYS# that cuts down the mathematical 0.2 bar, 0.3 bar and 0.4 bar respectively to
headache to get your process facility to run exit at a final pressure of 4 barg at the outlet
the way you'd expect it to. A set of hand of VLV-103. VLV-100 has a hold-up volume of
calculations is also included to understand 2 m3 declared to that accounts for the delay in
how the controller is tuned. the output flow variation.
Design Procedure for Controller Tuning
A good controller is one that offers a suitable
trade-off between performance and
robustness and the standard type of
controllers used even to this day are the
proportional (P), proportional plus integral
(PI), and the proportional plus integral and
derivative (PID) controllers. To tune these Figure 1. Process Control Scheme

controllers, 'Gain' (Kc), Integral Time' (Ti) The minimum and maximum flow through
and 'Derivative Time' (Td) are the three basic valve VLV-100 is taken to be 0 kmol/h and
parameters needed and a procedure is 200 kmol/h respectively. The control valve
required to estimate it. In this article, the (VLV-100) opening is modelled to transport
procedure employed is called Internal Model 100 kmol/h at 50% opening. A closed loop,
Control (IMC). reverse acting flow controller (FIC-100) is
What is Internal Model Control added to VLV-100 and needs to be tuned as a
IMC refers to a systematic procedure for P-I controller.
control system design based on the Q- The purpose of FIC-100 is to ensure that the
parameterization concept that is the basis for flow through the control valve VLV-100 is
many modern control techniques. What always maintained at 100 kmol/h by

Page 1 of 4
adjusting the valve opening whenever the pass through the valve. In this case, for the
flow into it changes with time. Therefore the sake of the exercise, the maximum flow is
work to be done is to estimate what are the taken to be 200 kmol/h. Note: The
'Kc' and 'Ti' values of the P-I controller that maximum flow through a valve depends on
ensures a flow of 100 kmol/h is maintained the size of the valve and is represented by
by FIC-100 irrespective of the disturbances. the term 'Cv'. Higher the Cv, larger is the
valve and greater is the flow through the
Basics of a PID Controller
valve.
Before trying to model and tune a controller,
understanding a few basic terms helps. A 9. Process Time Constant - This describes
controller is defined by three parameters - how fast a measured process variable
Set Point (SP), Process Variable (PV) and responds when forced by a change in the
Output (OP). controller output. The Process Time
Constant is equal to the time it takes for
1. Set Point (SP) - The parameter that you
the process to change to 63.2% of the total
want to maintain. In this example, the flow
change in the measured process variable.
rate of 100 kmol/h becomes the set point.
In the current case study, FIC-100 in the
2. Process Variable (PV) - The actual flow
image is modelled as a P-I controller.
passing through the valve which can vary.
In this example, PV becomes stream '2' How is a Controller Tuned
through which the pressure driven flow A controller is tuned by causing a disturbance
can vary. to the process and seeing how it reacts.
Watching its reaction, the control parameters
3. Output (OP) - What the controller needs
(Kc, Ti) can be adjusted to ensure the
to change in order to ensure only 100
controller behaves as intended.
kmol/h of flow passes through the valve,
VLV-100. In this example, OP is the valve In the current undertaking, the disturbance is
opening that ranges between 0% to 100% introduced as - the valve (OP) which is at
opening. 50% opening for a flow rate of 100 kmol/h
(PV) is momentarily changed manually to
4. Proportional Control (P) - “How Far” the
60% opening (OP).
measured process variable (PV) has moved
away from the desired set point (SP).
5. Integral Control (I) - “How Long” the
measured process variable (PV) has been
away from the desired set point (SP). 
6. Derivative Control (D) - “How Fast” the
error value changes at an instant in time.
7. PV Min - Minimum Process Variable (PV
Min) is the lowest amount of flow that can
pass through the valve. In this case, the
minimum flow is 0 kmol/h.
8. PV Max - Maximum Process Variable (PV
Max) is the highest amount of flow that can

Page 2 of 4
 Figure 2. Controller Connections

Using the above values of Kp and T63.2%, the

overall gain (Kc) and Integral Time (Ti) as a
first order process with negligible dead time
can be calculated as,

Using the calculated value of Kc=12.5 and Ti of

0.45 sec, the controller is programmed
accordingly and executed to run.

Figure 3. Controller Configuration

Upon doing so, the flow through the control
valve, VLV-100(PV) is monitored as shown in
the below figure.

Figure 5. IMC Design Tab

Figure 4. Controller Output with OP variation from

50% to 60% (Manual Mode)
To tune the controller using IMC design
procedure, two steps are employed
1. Estimate the value of Process Gain (Kp)
and Process Time Constant [Tp] or T63.2%
2. Estimate the value of Overall Gain (K c) and
Integral Time (Ti) i.e., P-I action.
From the above graph, Process gain (Kp) and
T63.2% is calculated as,

Page 3 of 4
 Figure 6. IMC Tuned Controlled Values 4. PI and PID Parameters from IMC Designs,
The response of the tuned P-I controller (FIC- Chapter 6, http://www.bgu.ac.il
100) when activated to run in automatic 5. The IMC-Based PID procedure, Chapter 7,
mode, i.e., when the controller is set to run on http://rpi.edu
its own, gives us the result where the FIC-100
#
automatically adjusts the valve (VLV-100) % Aspentech HYSYS 2006.5 is a registered
opening to ensure 100 kmol/h of flow passes Trademark of Aspentech
through the system as follows. About the Author
Vijay Sarathy holds a Master’s
Degree in Chemical Engineering
from Birla Institute of Technology
& Science (BITS), Pilani, India and
is a Chartered Engineer from the
Institution of Chemical Engineers, UK. His
expertise over 10 years of professional
experience covers Front End Engineering,
Process Dynamic Simulation and
Subsea/Onshore pipeline flow assurance in
the Oil and Gas industry. Vijay has worked as
Figure 7. Controller Output with OP Variation with
Tuned Controller Values (Auto Mode)
an Upstream Process Engineer with major
conglomerates of General Electric, ENI
From the above graph, it is seen that the
Saipem and Shell.
controller FIC-100, automatically runs to
bring the valve to 50% opening with 100
kmol/h of flow passing through VLV-100 in
about roughly 180 sec (~3 min). Using the
tuned controller, with any changes hereafter
in the incoming flow would mean, the flow
controller (FIC-100) can automatically dictate
by how much the control valve (VLV-100) has
to open or close to ensure a flow of 100
kmol/h is maintained.

References
1. Aspentech HYSYS 2006.5 Documentation –
Dynamic Modelling
2. Control Systems, Robotics and Automation,
Vol. II - Internal Model Control - Daniel E.
Rivera and Melvin E. Flores
3. A Best-Practices Approach to
Understanding and Tuning PID Controllers,
First Edition by Robert C. Rice, PhD

Page 4 of 4