ISA Transactions 51 (2012) 271276

Contents lists available at SciVerse ScienceDirect

ISA Transactions
journal homepage: www.elsevier.com/locate/isatrans

Design of a simple setpoint filter for minimizing overshoot for low order
processes
V. Vijayan a , Rames C. Panda b,
a

Department of E&I, St. Josephs College of Engineering, IT Highway, Chennai 600119, India

Department of Chemical Engineering, CLRI (CSIR), Adyar, Chennai 600 020, India

article

info

Article history:
Received 25 January 2011
Received in revised form
19 September 2011
Accepted 6 October 2011
Available online 24 October 2011
Keywords:
Overshoot
PID controller
Set point filter
Closed-loop response
Tuning

abstract
Setpoint filters are widely used along with a PID controller. The aim of the present paper is to reduce the
peak overshoot to a desired/tolerable limit. To design a setpoint filter, numerous methods are available,
which need extensive calculations. Moreover, the existing methods need information regarding the
process parameters, values of controller settings and are laborious. But the proposed method is very
simple and requires only the information about the peak overshoot and peak time of the system response
regardless of type and order of the system with arbitrary PID parameters. Several examples are taken to
show efficacy of the process.
2011, ISA. Published by Elsevier Ltd. All rights reserved.

1. Introduction
Most of the industrial loops use PID controllers till today. These
types of controllers are popular because of their ease in operation,
robust behavior, and easy maintenance. Even after the invention
of advanced process control strategies, predictive controllers etc.,
use of a PID controller dominates process industries. Generally,
PID controllers have four different structures [1] out of which
three are implementable. The accuracy and performance of these
controllers are greatly dependent on the method of tuning
controller parameters, namely, KC , I , and D . Researchers have
proposed a number of tuning rules to improve loop performance.
There are many industrial processes, which need to be operated
at unstable operating points for economic and safety reasons.
Researchers [2] proposed setpoint weighted PID controllers to
control these systems. Presence of large dead time in unstable
processes makes the system more difficult to control. Different
structures (conventional feedback, modified smith predictor,
modified IMC, two-degrees of freedom etc.) have been proposed to
improve the closedloop performance. But, in all the above works,
either the closed-loop structure or the tuning designed for specific
systems has improved closed-loop performances (evaluated by
error criteria). The literature does not show much evidence to

Corresponding author. Fax: +91 44 24911589.

E-mail address: panda@clri.res.in (R.C. Panda).

reduce overshoot in closed-loop responses of these methods.

Shamsuzzoha and Lee [3] designed setpoint filters to improve loop
performances using an IMC Maclaurian PID controller. Recently,
Shamsuzzoha and Skogestad [4] derived PID tuning rules based
on closedloop tests that reduced overshoot of the system. Most
of the unstable systems and processes with numerator zero yield
overshoot in their closed loop responses mainly due to improper
tuning. Hence, in order to improve the time domain performances,
new design procedures are proposed here to achieve the desired
overshoot using simple calculations. Several examples from IPDT,
FOPDT, SOPDT, HOPDT and multivariable systems are chosen
to implement the present method and results are achieved
in this study. Thus the entire paper is organized as follows.
Section 2 discusses a design technique of the proposed setpoint
filter. Examples of processes with different model structure are
taken and setpoint filters are designed in Section 3. Real time
experimental results are presented in Section 4. Conclusion is
drawn in Section 5.
2. Set point filter design
Fig. 1 shows the actual closed-loop response of a typical process
(Gp) with a PID controller. The response oscillates around the
set point with first peak overshoot at Mp1 at a corresponding
time tp1. Let us think that the response can be approximated by
a FOPDT transfer function that will yield a desired closed loop
response with the desired overshoot. The design is based on the
idea that: if a first order system is assumed, the closed loop

0019-0578/$ see front matter 2011, ISA. Published by Elsevier Ltd. All rights reserved.
doi:10.1016/j.isatra.2011.10.006

272

V. Vijayan, R.C. Panda / ISA Transactions 51 (2012) 271276

(b) Assume that the approximated FOPDT process gain k =

Mp1/0.6321.
(c) Assume that the time constant of the approximated FOPDT
process = tp1.
(d) Find the desired overshoot (Mp2) and time corresponding peak
time (tp2 + D) from the response.
(e) Calculate filter time constant (f ) from Eq. (4).
3. Results and discussion
The examples considered for simulation are provided in Table 1.

Fig. 1. Peak over shoot response of existing process and approximation curve.

Fig. 2. The existing structure with PID settings. (The proposed filter is used along
with existing structure.)

response will pass through the peak overshoot (Mp1) only when
the peak time is equal to time constant ( ) of the first order system.
Let the desired overshoot of the closed loop response be Mp2 and
the corresponding peak time be tp2. Let us also design a FOPDT
system with gain k = Mp1/0.6321 so that the time constant
of the designed system can be written as = tp1. Thus the
y(s)
transfer function of the approximated system becomes u(s) =
B(s)eDs
A(s)

keDs
s+1

and actual process becomes GP =

where B and A
are polynomials of s. Now, after introducing a setpoint filter with
the existing structure (Fig. 2), the system transfer function can
y(s)

Ds

be written as u(s) = ke
f s1+1 . This setpoint filter will bring
s+1
down the peak over shoot from Mp1 to Mp2. The inverse Laplace
transform of the above TF for step input can be written as

y(t ) =

f et /f et / + k.

(1)

From Fig. 1, it can be noted that, the desired over shoot is Mp2
and its corresponding peak time is tp2. So, at t = tp2 + D, y(t ) is
equal to Mp2. So Eq. (1) becomes,

Mp2 =

f etp2/f etp2/ + k.

(2)

It is well known that the filter time constant tf is always less

than process time constant. So etp2/f is very less than etp2/ . So
Eq. (2) can be written as

Mp2 =

etp2/ + k.

(3)

From the above equation, f can be calculated

f =

k Mp2 k etp2/
k Mp2

(4)

The steps of the calculations under present procedure are as

follows.
(a) Note down dead time and actual peak over shoot with out
filter = Mp1 and the corresponding peak time = tp1 + D from
the response.

Example 1 (Stable First Order Plus Dead Time Process (FOPDT)). Consider the following stable FOPDT process [5].
Ex-1 and the controller settings used were kc = 1.8, I = 1.655
and set point weight = 0.63. A setpoint filter is designed for the
same process and controller settings to reduce the overshoot. The
1
proposed filter transfer function is GF = 0.5839s
. Results (Fig. 3)
+1
show almost similar values of overshoot (Table 1) with improved
performance (ITAE with present technique, using setpoint filter,
becomes 1.921 whereas with setpoint weight, Chidambaram [5]
obtained ITAE of 2.479).
Example 2 (Unstable FOPDT System). Let us take the following
0.5s

FOPDT process [6]: Gp = e s1 and the controller settings used

were kc = 1.5353, I = 7.5753 with setpoint filter GF =
1
. For the same process with same controller settings
7.5753s+1
the designed/proposed filter transfer function became GF =
1
. The closedloop simulation results (table) show lesser
7.9248s+1
peak overshoot and with improved time domain performances
proving the efficiency of present method. Moreover, the present
procedure of filter design is very simple and straight forward.
Present results are almost in close agreement to Shamsuzzoha
and Lee [3]. However, the method proposed by the latter is
laborious compared to the present one.
Example 3 (Stable Second Order Plus Dead Time Process (SOPDT)).
Consider a stable SOPDT process [3] as Ex-3 (Table 1) with the
PID controller parameters as kc = 9.8092, I = 5.4502, D =
1.6898. The peak overshoot (PO) reported was 1.009. Using the
present method, a setpoint filter is designed with f = 3.2612.
Closedloop simulation resulted in a PO of lesser value, 1.0002,
and better performance values (Table 1) are obtained compared to
Shamsuzzoha and Lee [3] who used a second order filter.
Example 4 (Second Order Plus Dead Time Plus Unstable Process
(SODUPOne Unstable Pole)). As an unstable SODUP process [3], let
us take Ex-4 as mentioned in Table 2. The PID controller settings
used are kc = 6.7051, I = 5.4738, D = 1.333 and a second
1.6421s+1
order filter with transfer function GF = 7.2966s
2 +5.4738s+1 was used
that yielded a PO of 1.03 (Table 1). Whereas by using the present
method, a first order filter was designed whose f = 3.9636.
After the closed-loop simulation with same PID settings, better
performance values are obtained with less PO value of 1.0055.
Example 5 (SOPDT with Inverse Response). Next example is chosen
as an SOPDT process with a zero in numerator that often shows
inverse response as also was considered by Shamsuzzoha and
Lee [3]. The PIOD parameter was set to be kc = 3.0819, I =
1.6399, D = 0.4295 for this Ex-5 and a second order filter
with transfer function GF = 0.7044s2 +11.6399s+1 was used. They
obtained a PO of 1.274 (printed value is 1.274 however, we
calculated it as 1.000) and ITAE of 2.751. By using the present
method, a setpoint filter with time constant f = 0.9024 is
obtained and after simulation, an ITAE value of 1.188 is obtained.
Thus the performance of the system is improved by the present
setpoint filter. The closed loop response is shown in Fig. 3 and the
performance values for this example are given in Table 1.

V. Vijayan, R.C. Panda / ISA Transactions 51 (2012) 271276

273

Table 1
Resulting performance for examples.
Process type

FOPDT (stable) Ex-1 Gp =

Method

PID parameters

Filter parameters

Performance

Proposed

kc = 1.8, I = 1.655

f = 0.5839

IAE = 1.561
ISE = 1.158
ITAE = 1.921
PO = 1.0012

Chidambaram [5]

kc = 1.8, I = 1.655

Setpoint wt = 0.63

IAE = 1.541
ISE = 0.9297
ITAE = 2.479
PO = 1.000

Proposed

kc = 1.5353, I = 7.5753

f = 7.9248

IAE = 2.587
ISE = 1.695
ITAE = 6.614
PO = 1.0266

Chan Sul et al. [6]

kc = 1.5353, I = 7.5753

1
GF = 7.5753s
+1

IAE = 2.892
ISE = 2.074
ITAE = 6.813
PO = 1.0366

Proposed

kc = 9.8092, I = 5.4502, D = 1.6898

f = 3.2612

IAE = 3.536
ISE = 2.562
ITAE = 8.737
PO = 1.0002

Shamsuzzoha and Lee [3]

kc = 9.8092, I = 5.4502, D = 1.6898

1.6351s+1
GF = 9.2099s
2 +5.4502s+1

Proposed

kc = 6.7051, I = 5.4738, D = 1.333

f = 3.9636

Shamsuzzoha and Lee [3]

kc = 6.7051, I = 5.4738, D = 1.333

1.6421s+1
GF = 7.2966s
2 +5.4738s+1

Proposed

kc = 3.0819, I = 1.6399, D = 0.4295

f = 0.7442

IAE = 1.331
ISE = 1.065
ITAE = 1.036
PO = 1.0192

Shamsuzzoha and Lee [3]

kc = 3.0819, I = 1.6399, D = 0.4295

GF = 0.7044s2 +11.6399s+1

IAE = 1.395
ISE = 1.106
ITAE = 1.14
PO = 1.0133

Proposed

kc = 0.3593, I = 12.13, D = 2.704

f = 9.5505

IAE = 9.551
ISE = 7.428
ITAE = 57.52
PO = 1.0001

Shamsuzzoha and Lee [3]

kc = 0.3593, I = 12.13, D = 2.704

GF =

e0.5s
s+1

FOUDT (unstable) Ex-2 Gp =

e0.5s
s 1

1s

SOPDT (stable) Ex-3 Gp = (10s+2e1)(5s+1)

0.9 3 9 s

SODUP Ex-4 Gp = (5se1)(2.07s+1)

SOPDTZ Ex-5 Gp =

(0.2s+1)e0.2s
(1s+1)(1s+1)

4s

FOIPDT Ex-6 Gp = s(1e4s+1)

1
32.8106s2 +12.1304s+1

IAE = 4.174
ISE = 3.058
ITAE = 10.89
PO = 1.009
IAE = 3.157
ISE = 2.329
ITAE = 6.651
PO = 1.0055
IAE = 3.244
ISE = 2.389
ITAE = 6.846
PO = 1.03

IAE = 12.33
ISE = 9.897
ITAE = 87.04
PO = 1.012

Example 6 (First Order Delay Integrating Process (FODIP)). Another

type of model structure (FODIP) is generally observed in process
industries. Due to presence of an integrator, the step response
becomes unstable for these systems. Shamsuzzoha and Lee [3]
used PID settings kc = 0.3593, I = 12.13, D = 2.704 and
designed a second order setpoint filter GF = 32.8106s2 +112.1304s+1
that yielded a PO of 1.012 and ITAE value of 87.31 whereas using
1
the present procedure, the filter is designed as GF = 9.5505s
.
+1
After simulating the performance is obtained as ITAE=57.52. Thus
improved performance values are obtained and are reported in
Table 1.

between inputs and outputs. The present setpoint filter can also
be implemented on MIMO systems. Wood and Berry [7] distillation
column is chosen for the implementation. The designed filters have
time constants f 1 = 2.0588 min and f 2 = 2.4528 s. When the
system is simulated with PID controllers, the POs are reduced and
improved performance is obtained.

Example 7 (Multivariable System). Chemical processes behave as

multi inputmulti output systems with interactions existing

The coupled tanks set-up (Fig. 5a) is a model of a liquid storage

system in process industries. Often tanks are coupled through

The present method is applied to Ex-5 in the presence of

measurement noise 2% and the resulting closedloop response can
be analyzed for performance (Fig. 4 shows a PO of 1.0).
4. Real time studies

274

V. Vijayan, R.C. Panda / ISA Transactions 51 (2012) 271276

Fig. 3. Response of the nominal system for all the examples (Ex-1, 3 and 5 in the 1st column and Ex-2, 4 and 6 in the 2nd column).

connecting pipes. These storage facilities contain fluids where the

reactant level and flow are to be controlled. Water is chosen as the
fluid. The experiment set-up of coupled tanks is designed so that

the system can be configured. The 33-041 coupled tank system

is interfaced with computer through MATLAB/SIMULINK and an
Advantech PCI 1711 data acquisition interface card. The set-up has

V. Vijayan, R.C. Panda / ISA Transactions 51 (2012) 271276

275

Table 2
The performance of closed-loop in the real time experiment.
Process type

Method

PID parameters

Filter parameters

Performance

Proposed

kc = 0.6121, I = 5.7783, D = 1.7056

1
Gf = 6.344s
+1

IAE = 961
ISE = 7233
ITAE = 8.91 104
PO = 3.5%

Shamsuzzoha and Lee [3]

kc = 0.6121, I = 5.7783, D = 1.7056

1.156s+1
Gf = 9.855s
2 +5.778s+1

0.00667e0.0667s
SOPDT (stable) Gp = 43
.61s2 +21.9s+1

IAE = 1031
ISE = 7771
ITAE = 9.96 104
PO = 6.38%

Fig. 5a. Experimental setup for two interacting tank level control system.

Fig. 4. Closedloop response of the process (Ex-5) under measurement noise.

four translucent tanks each with a pressure sensor to measure the

water levels. Two of these liquid level tanks of 3.7 l capacity each
are connected in series (Fig. 5b) where the exit flow from the 1st
tank enters the 2nd tank and level (h2 ) is measured and controlled
in the 2nd tank by manipulating flow rate of the liquid in the
1st tank. The input is the voltage given to the pump through the
pump control unit. The differential pressure type level transmitter
is used to measure the level in the tank. To measure the level in
both tanks, two DPT type level transmitters are implemented at the
bottom of the tanks. The hand valve HV1 and HV3 are introduced
in the pipe to adjust the inflow and outflow, respectively. The card
gives 05 V output corresponding to the level 025 cm in the
tank. The sampling time used here is 0.1 s. The controlled inputs
are voltage applied to pumps so as to adjust the feed flow rate
in the 1st tank. The maximum excitation for the 100% actuator
output is restricted to 3.5 V for the corresponding maximum flow
rate of 4907.8 cm3 /min. For open-loop studies, the level is set
at 11 cm in the 2nd tank. The nominal flow for this 11 cm is
maintained at 1648.3 cm3 /min (equivalent to an excitation of 2.2 V
to the pump). Step test is carried-out and the inputoutput data is
stored in the computer. Then the response graph is drawn using
MATLAB software. From the graph the time constant of the two
tanks is calculated. Thus the transfer function of the entire system
is derived. Using step test, the openloop transfer function, relating
height of the liquid in the 2nd tank to the feed flow rate of the 1st
0.0667s

0.00667e
.
tank of the system, is identified as Gp = 43
.61s2 +21.9s+1
For this process, designing set-point filter by Shamsuzzoha and
1.156s+1
Lee [3] method we get Gf = 9.855s
2 +5.778s+1 , and a = 0.033, b =
0.4342 for their PID structure

GC = KC

1+

I s

+ D s

1 + as
1 + bs

(5)

With an optimal value of = 1.5, the PID settings are obtained

as Kc = 0.6121, i = 5.7783, d = 1.7056. The value was

Fig. 5b. Schematic of experimental setup for two interacting tank level control
system.

Fig. 6. Real time closed-loop response of two tank level control system using
present controller (solid line) and [3] (dasheddot line).

selected after viewing and calculating closed-loop performances

with different values of on trial and error basis. These settings
are implemented on the present liquid-level control system and a
closed-loop performance is obtained as reported in Table 2.
When the set-point filter is designed by the present method, we
1
get the filter as Gf = 6.344s
, and controller parameters become
+1
kc = 0.6121, I = 5.7783, D = 1.7056. With these settings

276

V. Vijayan, R.C. Panda / ISA Transactions 51 (2012) 271276

on the PID structure (Eq. (5) and a = 0.033, b = 0.4342), the

present liquid level system is run for closed-loop. Table 2 shows
the performance values for the two tank level control system.
Fig. 6 shows the real time closed-loop responses for the above
system and with present controller (solid line) and with IMC-PID
by Shamsuzzoha and Lee [3]. This figure reveals that the present
controller produces better response (fast and less overshoot) than
that of Shamsuzzoha and Lee [3].
5. Conclusion
A first order filter for processes is proposed and designed to
reduce overshoot to a desired level. The design procedure to find
out the time constant of the filter is very simple. The proposed
method produces better result (compared to similar available
techniques) in obtaining closed loop performances, ISE, ITAE and
IAE. This method does not require prior knowledge of model

structures/equations and also does not require PID settings. The

existing PID parameters are enough to get improved performance.
References
[1] Panda RC. Synthesis of PID controller using the desired closed-loop response.
Ind Eng Chem Res 2008;47(22):868492.
[2] Padmasree R, Chidambaram M. Set point weighted PID controller for unstable
system. Chem Eng Commun 2005;192:113.
[3] Shamsuzzoha M, Lee Moonyong. Design of advanced PID controller for
enhanced disturbance rejection of second-order processes with time delay.
AIChE J 2008;54(6):152636.
[4] Shamsuzzoha M, Skogestad S. The set point overshoot method: a simple and
fast closed-loop approach for PID tuning. J Process Control 2010;20:122034.
[5] Chidambaram M. Set point weighted PI/PID controllers. Chem Eng Commun
2000;179:113.
[6] Jung Chan Sul, Song Hyung Keun, Hyun Jae Chun. A direct synthesis tuning
method of unstable first-order-plus-time-delay processes. J Process Control
1999;9:2659.
[7] Wood RK, Berry MW. Terminal composition control of a binary distillation
column. Chem Eng Sci 1973;28:170717.