Results in Control and Optimization 21 (2025) 100635

Contents lists available at ScienceDirect

Results in Control and Optimization

journal homepage: [Link]/locate/rico

Maximum sensitivity-based PID controller cascaded with 2/3 order

filter for integrating processes with time delay
K. Divakar a ,∗, Muni Reddy G. a , N.M.G. Kumar b ,1 , M. Ijaz Khan c
a
Department of EEE, Siddharth Institute of Engineering & Technology, Puttur, Andhra Pradesh, India
b
Department of EEE, Mohan Babu University, Formerly Sree Vidyanikethan Engineering College, A Rangam Pet, Tirupathi Adhara
Pradesh, 517102, India
c
Department of Mechanical Engineering, College of Engineering, Prince Mohammed bin Fahd University, AI-Khobar, Saudi Arabia

ARTICLE INFO ABSTRACT

Keywords: In the chemical industry, integrating processes are often encountered. Examples of integrating
PID controller processes include bottom level control of a distillation column, level control of a tank with a
2/3 order filter motor fixed at the outlet, current-controlled DC motor, fermentation reactors, spacecraft take-
Integrating processes with time delay
off dynamics, paper industry drying processes, continuous stirred tank reactor (CSTR) with
Maximum sensitivity
exothermic reactor, and so on. This article describes a new PID controller with a 2/3 order
filter for integrating processes with time delay. Time delay is approximated to ratio of two
polynomials using a second-order Pade’s approximation. A set point filter is used to minimize
overshoot and the settling time of servo responses. Simulation studies are conducted on some
of the benchmarking process models utilized in the literature. The comparative assessment is
based on various performance indices.

1. Introduction

In a chemical industry, processes that have one or more poles at the origin frequently present difficulties for effective control.
They lack self-regulation, therefore, any change in the environment or the inputs will cause them to deviate from their equilibrium
point, which will cause the output to change continuously and rapidly. The integrating processes are also effected by the time
delay (or) dead time. The controller design for non- self regulating processes along with dead time is tedious task for many of the
researchers. The various control structures are classified based on number control loops as follows:

1.1. Multi loop control structures

Several researchers have used multi loop control structures to obtain effective control. Researchers have used independent
controllers for servo and regulatory responses in the multi loop control designs. The benefit of being able to forecast the future output
has been utilized by the researchers in their multi loop control structures by using a Smith predictor [1–11]. Several researchers have
proposed multi loop control structures without using smith predictor. Majority of these multi loop schemes are Double-loop control
schemes i.e inner and outer loops. The internal loop is utilized for achieving stability, and the required performance is achieved by
the external loop [12–18,18–20]. The Several researchers have used PD/PI controllers in achieving inner/outer-loop stabilization
respectively. In order to improve the performance in regulatory conditions, researchers utilized PID and Internal Model Control(IMC)

∗ Corresponding author.
E-mail address: [Link]@[Link] (K. Divakar).
1 Professor.

[Link]
Received 25 May 2025; Received in revised form 12 November 2025; Accepted 2 December 2025
Available online 3 December 2025
2666-7207/© 2025 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
([Link]
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

in double loop control structures [21,22]. A PI-PD control scheme for a gasoline pressure tower has been designed by Li et al. [14].
Optimization based PI-PD control settings are proposed for integrating and unstable processes by some researchers [15–19].

1.2. Single loop control structures

Though multi loop structures are advantageous in terms of decoupling servo and regulatory processes, they are complex in terms
of structure and tuning. Several researches have put significant effort into developing effective single-loop controllers [23–31].
Most single-loop control structures are designed using Internal Model Control(IMC) and Direct Synthesis(DS) based approaches.
Researchers have been trying to derive better performance with single controller itself by means of proper tuning, associating filter
to PID etc. Compared to multi-loop control structures, conventional control structures are easier to understand and analyze. Various
researchers have developed single-loop control structures in order to achieve the desired responses [23–30,32]. From the literature,
several control systems with several loops and controllers that are based on Smith predictors have been presented. Disturbance
rejection and setpoint responses are decoupled and managed by different controllers in various multi-loop control structures, which
is complex and tedious to control. Even though many of the single-loop and double-loop controllers have been designed till now,
there is also a scope to enhance the closed-loop performance of the by properly designing controllers to eliminate steady-state error.
Two-degree-of-freedom (2-DOF) controller architectures facilitate independent tuning of set-point tracking and disturbance
rejection; however, they introduce a greater number of tuning parameters, thereby escalating design complexity, tuning effort,
and the potential for parameter interaction in practical applications. A recent study has concentrated on solutions that preserve
the performance advantages of 2-DOF controllers while alleviating their parametric complexity. One approach employs model
reduction and low-order controller synthesis to decrease the effective number of adjustable parameters while maintaining closed-
loop performance. Various authors employed model reduction techniques for many applications to design the controller [33,34].
Oversimplification of the higher-order system model may result in erroneous conclusions. Recently some researchers designed
reduced parametric controller for various integrating and unstable processes. I-PD controllers are structurally identical to PI-PD
controllers; however, they only need to tune one less parameter. Despite having one less tuning parameter, the I-PD controller
provides comparable performance to the PI-PD controller [18]. Recently, Peker and Kaya [20], Raja [35] have designed an I-PD
controller using integral performance indices for integrating processes. Single-loop controllers with fractional-order algorithms serve
as an alternative to double- and multi-loop control strategies [36]. Accordingly, cascade controllers also suggested for integrating
and unstable processes by various authors [37,37–39].

1.3. The contribution of the work:

• In this study, a single loop control structure with a PID controller cascaded to a 2/3 order filter is proposed. The controller
design criteria is based on the polynomial approach for various integration process with dead time.
• The polynomial approach is used to determine the filter coefficients and PID parameters. The benefits of the polynomial
technique include choosing the target characteristic polynomial to calculate the parameters and adding a specific number of
parameters to the controller.
• To achieve better performance for the given time-delay accuracy, the present control method used a second-order Pade’s time
delay approximation.
• A single tuning parameter is used to determine the PID controller’s settings, which are chosen based on the Maximum
Sensitivity (MS) requirement.

1.4. Content organized

A mathematical approach for design of the controller for process like Pure Integrating Process with Dead Time(PIPDT),
Integrating First Order Process with Dead Time(IFOPDT) and Integrating Unstable First Order Process with Dead Time(IUFOPDT)
explained in Section 3, selection of tuning parameter(𝜆) discussed in Section 3 and simulation results discussion and conclusion are
included in Section 4 and 5 respectively.

2. Mathematical analysis of the proposed controller

2.1. Design of controller 𝐺𝑐

The block diagram of the proposed structure of the controller is shown in Fig. 1. From Fig. 1, 𝑟 represents the set point, 𝐹
represents the set point filter, 𝐺𝑐 represents the controller, 𝐺𝑝 represents a process to be controlled, and 𝑦, 𝑑 are the output and
input disturbance signals, respectively.
The proposed controller structure for PIPDT, IFOPDT with negative zero is shown in Eq. (1).
( )
𝑞 𝑘 (𝑧2 𝑠2 + 𝑧1 𝑠 + 1)
𝐺𝑐 = = 𝑘𝑝 + 𝑖 + 𝑘𝑑 𝑠 (1a)
𝑝 𝑠 2
(𝑝2 𝑠 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)
Where,
𝑞 = (𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1) (1b)
2
𝑝 = 𝑠(𝑝2 𝑠 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1) (1c)

2
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 1. Proposed controller structure.

2.2. Design of 𝐺𝑐 for PIPDT

By the application of time delay approximation of Pade’s, the resultant process is expressed as a ratio of two polynomials shown
in Eq. (2).
𝑘 −𝑠𝜃 𝑎
𝐺𝑝 (𝑠) = 𝑒 = (2a)
𝑠 𝑏
Where,
𝑠𝜃
𝑎 = 𝑘(1 − ) (2b)
2
𝑠𝜃
𝑏 = 𝑠(1 + ) (2c)
2
The responses of servo and regulatory are represented in Eqs. 3 and 4 respectively.
𝑦 𝐹 𝐺𝐶 𝐺𝑝
= (3)
𝑟 (1 + 𝐺𝑐 𝐺𝑃 )
𝑦 𝐺𝑝
= (4)
𝑑 (1 + 𝐺𝑐 𝐺𝑝 )
By substituting Eqs. (1) and (2) in Eqs. (3) and (4), the resultant equations are represented in Eqs. (5) and (6).
𝑠𝜃
𝐹 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 − 2
)
𝑦∕𝑟 = (5)
𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(1 + 2 ) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 − 𝑠𝜃
𝑠𝜃
2
)
𝑠𝜃 2
𝑘𝑠(1 − 2 )(𝑝2 𝑠 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)
𝑦∕𝑑 = (6)
𝑠 (𝑝2 𝑠 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(1 + 𝑠𝜃
2 2
2
) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 − 𝑠𝜃
2
)
Eq. (7) represents the characteristic equation (CE), which is the servo and regulatory responses denominator polynomial.
𝑠𝜃 𝑠𝜃
𝐶𝐸 = 𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(1 + ) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 − ) (7)
2 2
𝑠𝜃
In Eq. (7), (1 − 2
) is approximated as a 𝑒−𝑠𝜃∕2 and the resultant CE is shown in Eq. (8).
𝑠𝜃
) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)𝑒−𝑠𝜃∕2 = 0
𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(1 + (8)
2
The CE must be computed against the desired CE using appropriate dead time approximation to retrieve the unknown parameters.
The first order Pade’s approximation is the widely accepted approximation for various controllers design [25,40]. To achieve even
more accuracy, the current work has used a Pade’s second-order approximation and it is represented in Eq. (9).
(1 − (𝜃𝑠∕4) + 𝑠2 𝜃 2 ∕48)
𝑒−𝑠𝜃∕2 = (9)
(1 + (𝜃𝑠∕4) + 𝑠2 𝜃 2 ∕48)
By considering 𝑧2 = 𝜃 2 ∕48 and 𝑧1 = 𝜃∕4 and substituting Eq. (9) in Eq. (8), the resultant CE is represented in Eq. (10).
𝑠𝜃
𝐶𝐸 = 𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(1 + ) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(1 − (𝑠𝜃∕4) + (𝑠2 𝜃 2 ∕48)) (10)
2
On simplifying,

𝑘𝑘𝑖 (𝑐6 𝑠6 + 𝑐5 𝑠5 + 𝑐4 𝑠4 + 𝑐3 𝑠3 + 𝑐2 𝑠2 + 𝑐1 𝑠 + 1) = 0 (11a)

𝑝 𝑝 𝜃
𝑐6 = 2 3 (11b)
2𝑘𝑘𝑖

3
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 2. Nominal response of Example 1.

𝑝1 𝑝3 𝜃
𝑝2 (𝑝3 + 𝜃2 ) + 2
𝑐5 = (11c)
𝑘𝑘𝑖
𝑝3 𝜃 𝑘𝑘𝑑 𝜃 2
𝑝2 + 2
+ 𝑝1 (𝑝3 + 𝜃2 ) + 48
𝑐4 = (11d)
𝑘𝑘𝑖
𝜃 𝑘𝑘𝑝 𝜃 2 𝑘𝑘𝑑
𝑝1 + 𝑝3 + 2
+ 48
− 4
𝑐3 = (11e)
𝑘𝑘𝑖
𝑘𝑘𝑖 𝜃 2 𝑘𝑘𝑝 𝜃
48
− 4
+ 𝑘𝑘𝑑 + 1
𝑐2 = (11f)
𝑘𝑘𝑖
𝑘𝑘𝑖 𝜃
𝑐1 = 𝑘𝑘𝑝 − (11g)
4
To obtain the closed-loop stable system response and the PID values, Eq. (11) is to be solved against a desired CE as shown in Eq.
(12).

(𝜆𝑠 + 1)3 (1 + 𝑠𝑧1 + 𝑠2 𝑧2 )(𝑝3 𝑠 + 1) = 0 (12)

Where 𝜆 is the tuning parameter. By canceling out a few zeros in the servo and regulatory responses, the selected target CE locate
the remaining poles to be at 𝑠 = −(1∕𝜆).

2.3. Controller(𝐺𝑐 ) design for IFOPDT with zero

The process is considered as

𝑘(1 + 𝑠𝑧) −𝑠𝜃 𝑓 −𝑠𝜃
𝐺𝑝 (𝑠) = 𝑒 = 𝑒 (13a)
𝑠(𝜏𝑠 + 1) 𝑔

4
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 3. Perturbed response of Example 1.

Where,

𝑓 = 𝑘(1 + 𝑠𝑧) (13b)

𝑔 = 𝑠(𝜏𝑠 + 1) (13c)

Substituting Eqs. (1) and (13) in Eqs. (3) and (4) results in Eqs. (14) and (15) respectively.
𝐹 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 + 𝑠𝑧)𝑒−𝑠𝜃
𝑦∕𝑟 = ( ) (14)
𝑠2 (𝑝 2
+ 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 + 1)
2𝑠
+ 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 + 𝑠𝑧)𝑒−𝑠𝜃
𝑘𝑠(1 + 𝑠𝑧)(𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)𝑒−𝑠𝜃
𝑦∕𝑑 = ( ) (15)
𝑠2 (𝑝2 𝑠2+ 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 + 1)+
𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 + 𝑠𝑧)𝑒−𝑠𝜃
The CE is represented in Eq. (16).

𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 + 1) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)(1 + 𝑠𝑧)𝑒−𝑠𝜃 = 0 (16)

second order time delay approximation of Pade’s is considered. By considering 𝑧2 = 𝜃 2 ∕12 and 𝑧1 = 𝜃∕2, the resultant CE is
represented in Eq. (17).

𝐶𝐸 = 𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 + 1) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(1 + 𝑠𝑧)(1 − (𝑠𝜃∕2) + (𝑠2 𝜃 2 ∕12)) (17)

Further simplification of Eq. (17) leads to,

𝑘𝑘𝑖 (𝑐6 𝑠6 + 𝑐5 𝑠5 + 𝑐4 𝑠4 + 𝑐3 𝑠3 + 𝑐2 𝑠2 + 𝑐1 𝑠 + 1) = 0 (18a)

𝑝 𝑝 𝜏
𝑐6 = 2 3 (18b)
𝑘𝑘𝑖
𝑘𝑘𝑑 𝑧𝜃 2
48
+ 𝑝2 (𝑝3 + 𝜏) + 𝑝1 𝑝3 𝜏
𝑐5 = (18c)
𝑘𝑘𝑖

5
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 4. Nominal response of Example 2.

𝜃 2 (𝑘𝑘𝑑 +𝑘𝑘𝑝 𝑧) 𝑘𝑘𝑑 𝜃𝑧

𝑝2 + 𝑝1 (𝑝3 + 𝜏) + 𝑝3 𝜏 + 48
− 4
𝑐4 = (18d)
𝑘𝑘𝑖
𝜃(𝑘𝑘𝑑 +𝑘𝑘𝑝 𝑧) 𝜃 2 (𝑘𝑘𝑝 +𝑘𝑘𝑖 𝑧)
𝑝1 + 𝑝3 + 𝜏 − 4
+ 48
+ 𝑘𝑘𝑑 𝑧
𝑐3 = (18e)
𝑘𝑘𝑖
𝜃(𝑘𝑘𝑝 +𝑘𝑘𝑖 𝑧) 𝑘𝑘𝑖 𝜃 2
𝑘𝑘𝑑 − 4
+ 48
+ 𝑘𝑘𝑝 𝑧 + 1
𝑐2 = (18f)
𝑘𝑘𝑖
𝑘𝑘𝑖 𝜃
𝑘𝑘𝑝 − 4
+ 𝑘𝑘𝑖 𝑧
𝑐1 = (18g)
𝑘𝑘𝑖
In order to find out the unknown parameters, Eq. (18) is solved against the desired CE which is considered as presented in Eq. (19)

(𝜆𝑠 + 1)3 (1 + 𝑠𝑧1 + 𝑠2 𝑧2 )(𝑠𝑧 + 1) = 0 (19)

2.4. Design of 𝐺𝑐 for IUFOPDT

𝐾
𝐺𝑝 (𝑠) = 𝑒−𝑠𝜃 (20)
𝑠(𝜏𝑠 − 1)
Substituting Eq. (20) and Eq. (1) in Eqs. (3) and (4), the resulted equations are shown in Eqs. (21) and (22) respectively.
𝐹 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)𝑒−𝑠𝜃
𝑦∕𝑟 = ( ) (21)
𝑠2 (𝑝 2
+ 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 − 1)
2𝑠
+ 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)𝑒−𝑠𝜃

6
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 5. Perturbed response of Example 2.

𝑘𝑠(𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)𝑒−𝑠𝜃

𝑦∕𝑑 = ( ) (22)
𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 − 1)+
𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)𝑒−𝑠𝜃
The equation of CE is presented in Eq. (23).

𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 − 1) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)𝑒−𝑠𝜃 = 0 (23)

The dead time is approximated by using pade’s second order approximations and considering 𝑧2 = 𝜃 2 ∕12 and 𝑧1 = 𝜃∕2, the resultant
CE is represented in Eq. (24).

𝐶𝐸 = 𝑠2 (𝑝2 𝑠2 + 𝑝1 𝑠 + 1)(𝑝3 𝑠 + 1)(𝜏𝑠 − 1) + 𝑘(𝑘𝑑 𝑠2 + 𝑘𝑝 𝑠 + 𝑘𝑖 )(1 − (𝑠𝜃∕2) + (𝑠2 𝜃 2 ∕12)) (24)

The Eq. (24) further simplified as

𝑘𝑘𝑖 (𝑐6 𝑠6 + 𝑐5 𝑠5 + 𝑐4 𝑠4 + 𝑐3 𝑠3 + 𝑐2 𝑠2 + 𝑐1 𝑠 + 1) = 0 (25a)

𝑝 𝑝 𝜏
𝑐6 = 2 3 (25b)
𝑘𝑘𝑖
𝑝 (𝑝 − 𝜏) − 𝑝1 𝑝3 𝜏
𝑐5 = − 2 3 (25c)
𝑘𝑘𝑖
𝑘𝑘𝑑 𝜃 2
𝑝2 − 𝑝3 𝜏 + 𝑝1 (𝑝3 − 𝜏) − 12
𝑐4 = − (25d)
𝑘𝑘𝑖
𝑘𝑘𝑝 𝜃 2 𝑘𝑘𝑑 𝜃
𝑝1 + 𝑝3 − 𝜏 − 12
+ 2
𝑐3 = − (25e)
𝑘𝑘𝑖
𝑘𝑘𝑖 𝜃 2 𝑘𝑘𝑝 𝜃
12
− 2
+ 𝑘𝑘𝑑 − 1
𝑐2 = (25f)
𝑘𝑘𝑖

7
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 6. Nominal response of Example 3.

𝑘𝑘𝑖 𝜃
𝑘𝑘𝑝 − 2
𝑐1 = (25g)
𝑘𝑘𝑖
To findout the unknown values Eq. (25) to be solved against the desired CE and it is represented in Eq. (26)
(𝜆𝑠 + 1)3 (1 + 𝑠𝑧1 + 𝑠2 𝑧2 )(𝑠𝑝3 + 1) = 0 (26)

2.5. Set point filtering

In servo response, the zeros inserted by the controller may cause overshoots. In the present work, the controller added zeros in
servo and regulatory transfer functions are compensated by clever selection of target CE. However, there are still some zeros left.
Pertaining to servo response, the effect of zeros introduced by the controller can be cancelled out by either set point weighting or
set point filter. In this design, analytically designed set point filter is applied. The suggested set point filter(𝐹 ) in servo response is
intended to counteract the controller inserted zeros. The mathematical form of 𝐹 is represented in Eq. (27) for PIPDT and IFOPDT
with zero and Eq. (28) represented the setpoint filter for IUFOPDT.
𝛾𝑠 + 1
𝐹 = 𝑘𝑝
(27)
𝑘
( 𝑘𝑑 𝑠2 + 𝑘𝑖
𝑠 + 1)(𝑧2 𝑠2 + 𝑧1 𝑠 + 1)
𝑖
𝛾𝑠 + 1
𝐹 = 𝑘𝑝
(28)
𝑘
( 𝑘𝑑 𝑠2 + 𝑘𝑖
𝑠 + 1)
𝑖

The denominator cancels the controller introduced zeros and the 𝛾 can be used to manipulate the speed of servo response. The
main advantage of the set point filter is that, while it able to alter the servo performance, it does not interfere with the regulatory
response.

8
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 7. Perturbed response of Example 3.

3. Selection of 𝝀

The choice of 𝜆 is significant because it is closely associated with the stability of the closed-loop system. Various authors [25,
40,41] have used MS-based tuning, which characterizes the robust stability of a closed-loop system. Maximum sensitivity (MS) is
the inverse of the shortest path between the critical point and the loop transfer function’s Nyquist plot [42]. In this chapter, for
the selection of 𝜆, MS based criteria is applied. The lower MS value will result in higher robust stability [41]. For open loop stable
systems, MS should be modified between 1.2 and 2 as a trade-off between robustness and response of speed [41]. However, for
unstable and integrated systems, it is not always possible to get faster responses at lower MS levels.

4. Simulation studies

This section conducts simulation analyses to evaluate the performance of the proposed strategy in comparison to other control
strategies found in the literature.

Example 1. In this example, a PIPDT (Eq. (29)) is considered for analysis. Level control of a tank with a motor installed at the
outflow and bottom-level control of a distillation column are the most notable examples of this type of process.
0.2𝑒−7.4𝑠
𝐺𝑝 (𝑠) = (29)
𝑠
The current technique is compared to the method of [25] . The controller settings for both methods are represented in Table 1. Both
methods are compared at an MS value of 2. For the proposed method, the adjustment parameter 𝜆 = 8.7 is considered for which
the MS is obtained as 2. To assess the tracking of the set point for both methods, a unit step change is applied at 𝑡 = 0 s and to
assess the rejection performance of the disturbance, a negative disturbance of 0.5 magnitude is applied at 𝑡 = 100 𝑠𝑒𝑐. Fig. 2 shows
the nominal response for both approaches, and Table 2 displays the performance assessment matrix. From Table 2, it is evident
that the proposed method has shown the same performance compared to the opposite method in all performance indices. But, in
perspective of Total Variance(TV) proposed method has delivered a marginally better response .
To know the perturbed response characteristics, +20% variation in gain(𝑘) and +20% change in time delay(𝜃) are imposed. The
performance curves are shown in Fig. 3 and the performance assessment matrix of both methods is represented in Table 3. From
Table 3, it is proved that both methods have shown equal performance with respect to all the performance indices. The proposed
method settled quickly compared to [25] method which is evident from better ITAE values of proposed method.

9
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 8. Nominal response of Example 4.

Example 2. Examine the pure integration process with time delay, where the transfer function is represented by Eq. (30).
1 −𝑠
𝐺𝑝 (𝑠) = 𝑒 (30)
𝑠
This above process transfer function studied by Kaya [18],Chakraborty et al. [43]. The proposed method compared to [18] in this
case. Table 1 lists the tuning settings for both designs. Both method are tuned such that their IAE values in the servo response in
nominal conditions should be same. In this condition, the value of 𝜆 is 1.38 and the MS value is 1.864. With a step input of unity
magnitude and a disturbance of −0.1 magnitude entering the system at time 𝑡 = 25 s under normal conditions, Fig. 4 displays the
closed-loop responses for both controllers. IAE,ISE,ITAE, settling time and TV of both approaches are provided in Table 2. From
Table 2, the [18] method provided better performance in ITAE and settling time and was lagged by the other indices in the servo
response and required a high control signal, which reduces the smoothness of the final control element. The proposed method gave
better value in settling time in case of regulatory condition and comparable performance in all other performance indices.

The performance of both methods in the case of uncertainty of the model is achieved by including the + 20% variation in 𝑘
and 𝜃. Both approaches’ response curves are shown in Fig. 5. The matrix of performance is presented in Table 3. The [18] method
gave more oscillatory response in case of servo and the current technique provided better values in all performance except ITAE.
In regulatory, Kaya [18] method gave better response in case IAE, ISE and ITAE but lagged in settling time and TV. The overall
consideration of perturbed performance, proposed method gave superior performance in case of servo and acceptable performance
in case of regulatory response.

Example 3. In this example, IFOPDT with zero is considered. This process is previously studied by Anil and Sree [23], Medarametla
[41]. The process is represented in Eq. (31).
(1 + 10𝑠) −𝑠
𝐺𝑝 (𝑠) = 𝑒 (31)
𝑠(2𝑠 + 1)

The current technique is compared with [23,41] methods. The suggested technique is tuned at 𝜆 = 1.2240, at which MS is equal
to 2.3458. The other methods are tuned at MS = 2.35. The control settings of the three methods are shown in Table 1. A unit step
change is considered at 𝑡 = 0 s as a set point and a unit step disturbance is at 𝑡 = 15 𝑠𝑒𝑐. Fig. 6 displays the performance curves

10
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Fig. 9. Perturbed response of Example 4.

under un perturbed condition, while Table 2 provides the comparison matrix. From Table 2, it is evident that the method of [41] has
given better response compared to the other methods in servo condition. In the case of regulatory condition, the proposed scheme
has shown a superior response in all aspects of the performance indices.
A +20% change in 𝑘 and a +20% change in 𝜃 are imposed in order to determine the performance in perturbed conditions. Fig.
7 shows the performance curves, while Table 3 displays the performance matrix. The suggested approach settles rapidly, and the
method of [41] has demonstrated good responsiveness under servo conditions, according to the performance analysis presented in
Table 3. The suggested approach also demonstrates improved settling time and more equivalent response characteristics in the event
of regulatory conditions.

Example 4. The transfer function of IUFOPDT is represented in Eq. (32).

1
𝐺𝑝 (𝑠) = 𝑒−0.2𝑠 (32)
𝑠(𝑠 − 1)
The suggested approach is contrasted with [44,45] for the mentioned process. Every method is compared at the same MS value for
a unique comparison. Table 1 lists the control setup settings for the three techniques. At 𝑡 = 0 and t=20 s, respectively, a unit step
magnitude and a unit disturbance are applied. In Fig. 8, the nominal performance is shown. Table 2 displays the performance-related
matrices. The suggested approach achieved shorter T.V. and almost similar values in the case of settling time in the case of servo
response, according to Fig. 8 and Table 2. When it came to regulatory response, the [44] technique performed better but lagged in
𝑇 .V.

The robustness of the proposed controller also tested by applying the model uncertainties. The model uncertainties are obtained
by inserting +10% change in k and 𝜃 along with −10% change in 𝜏. The performance indices of all the three method under perturbed
condition is listed in Table 3. In servo response, So [44] method had less values in case of IAE, ISE and ISTE but larger values in case
of T.V and 𝑡𝑠 . The [44] method produced 1.65 times greater value than proposed method in case of T.V and 1.18 times higher value
in case of settling time. In regulatory response also, So [44] method obtained better values in all indices but 1.34 times higher value

11
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Table 1
Controller settings for various techniques.
Process Method PID Parameters PID Filter Ms
𝑘𝑝 𝑘𝑖 𝑘𝑑
1.1408𝑠2 +1.85𝑠+1
0.2 −7.4𝑠 Proposeda 0.5068 0.0151 1.6682 2.2732𝑠3 +4.47𝑠2 +3.2509𝑠+1
2
𝑠
𝑒
1
[25]b 0.5169 0.0151 1.70499 1.4835𝑠+1
2
0.0208𝑠2 +0.25𝑠+1
1 −𝑠 Proposedc 0.650 0.1265 0.293 0.0069𝑠3 +0.6981𝑠2 +0.5116𝑠+1
–
𝑠
𝑒
[18] 𝐿𝑘𝑐 = 0.604𝑘𝑓 = 0.273 0.1483 0.3065 – –
0.0833𝑠2 +0.5𝑠+1
Proposedd 1.4688 0.3144 1.7246 0.2402𝑠3 +1.2153𝑠2 +10.1191𝑠+1
2.348
(1+10𝑠) −𝑠
𝑒 (0.25𝑠+1)2
𝑠(2𝑠+1)
[41]e 1.37 0.2866 1.6359 1.3672𝑠2 +10.1367𝑠+1
2.35
0.5214𝑠+1
[23] 1.1601 0.2251 1.4452 2.674𝑠2 +10.2674𝑠+1
2.35
f 0.0033𝑠2 +0.1𝑠+1
Proposed 5.1197 3.121 4.119 (0.0012𝑠2 +0.0291𝑠+1)(0.01𝑠+1)
2.3
1
𝑒−0.2𝑠 0.1102𝑠+1
𝑠(𝑠−1)
[44]g 5.0793 3.621 3.87 0.0065𝑠+1
2.3
0.100𝑠+1
[45]h 4.898 2.922 4.044 0.00348𝑠+1
2.3

𝑓𝑅 = 𝑆𝑒𝑡𝑝𝑜𝑖𝑛𝑡𝑓 𝑖𝑙𝑡𝑒𝑟.
a 17𝑠+1
𝑓𝑅 = 125.788𝑆 4 +242.19𝑆 3 +173.37𝑆 2 +35.35𝑆+1
.
b 17.25𝑠+1
𝑓𝑅 = 30.4𝑠+1
.
c 1.8𝑠+1
𝑓𝑅 = 0.048𝑠4 +0.687𝑠3 +3.625𝑠2 +5.39𝑠+1
.
d 2.4𝑠+1
𝑓𝑅 = 0.4571𝑠4 +3.132𝑠3 +7.905𝑠2 +5.172𝑠+1
.
e 1.310𝑠4 +4.90𝑠3 +6.869𝑠2 +4.28𝑠+1
𝑓𝑅 = 0.228𝑠6 +2.588𝑠5 +10.62𝑠4 +19.58𝑠3 +17.248𝑠2 +6.8802𝑠+1
.
f 0.5𝑠+1
𝑓𝑅 = 1.3195𝑠2 +1.64𝑠+1
.
g 1.0703𝑠+0.561𝑠+1
𝑓𝑅 = 1.0703𝑠+1.4026𝑠+1
.
h 1.3843𝑠2 +0.670𝑠+1
𝑓𝑅 = 1.384𝑠2 +1.676𝑠+1
.

Table 2
Performance measures under nominal condition.
Process Method Servo Regulatory
IAE ISE ITAE TV ts IAE ISE ITAE TV ts

0.2 −7.4𝑠 Proposed 20.14 16.1 257 0.707 63.9 33.06 23.58 4369 1.399 85.64
𝑒
𝑠 [25] 18.45 13.51 263 0.841 69.8 33.11 23.15 4395 1.397 89.85

1 −𝑠 Proposed 1.85 0.35 11.49 0.64 8.38 0.79 0.08 23.64 0.26 13.2
𝑒
𝑠 [18] 1.849 1.51 2.087 1.74 2.82 0.671 0.04 20.7 0.28 18.19
Proposed 3.154 2.48 6.263 0.244 9.388 12.96 31.09 379.1 3.421 12.00
(1+10𝑠) −𝑠
𝑠(2𝑠+1)
𝑒 [41] 2.6 2.077 4.027 0.2518 5.851 14.03 34.33 411.9 3.357 12.286
[23] 4.34 2.65 15.81 0.395 12.26 17.26 48.38 510.3 3.153 13.01
Proposed 1.14 0.78 0.872 15.84 2.93 0.32 0.039 6.95 4.13 4.55
1
𝑠(𝑠−1)
𝑒−0.2𝑠 [44] 0.93 0.67 0.602 28.67 2.912 0.28 0.036 6.031 7.5 3.53
[45] 1.01 0.72 0.65 42.60 2.28 0.34 0.043 7.45 7.99 4.58
𝑡𝑠 − 𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑇 𝑖𝑚𝑒 (𝑠𝑒𝑐).

in case of T.V. Overall proposed method produced comparable response w.r.t to latest reported method [44] and superior response
in all indices w.r.t to [45] method (see Fig. 9).

5. Conclusion

This paper describes the design and implementation of a PID controller cascaded with an order 2/3 filter for various integrating
processes. Using a polynomial approach, the filter coefficients and PID controller settings are calculated. An analytical approach
based on MS is used to determine the controller’s tuning parameter (𝜆). Dead time is approximated using the second-order Pade’s
approximation. Choosing the appropriate CE eliminates the overshoot impact of zeros. To illustrate the advantages of the suggested
control approach, various processes are examined through simulation. The recommended controller performs better than the
techniques found in the literature in many a cases.
The proposed methodology was validated using simulations of multiple benchmark integrating processes with dead time. The
suggested controller outperformed previous recent techniques, resulting in improvements in various performance indices ranging
from 20% to 65% across all models studied. The proposed controller outperformed the double loop controller, which resembles
the utilization in scenarios where the double controller is challenging to implement in real-time applications, as it is a single loop
controller which requires single tuning parameter.

12
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

Table 3
Performance measures under Perturbed condition.
Process Method Servo Regulatory
IAE ISE ITAE TV ts IAE ISE ITAE TV ts

0.24 −8.88𝑠 Proposed 22.83 16.99 381.4 1.151 87.50 33.03 29.45 4360 2.13 103.57
𝑒
𝑠 [25] 22.12 15.13 398.7 1.405 87.15 33.08 28.94 4387 2.14 105.711

1.2 −1.2𝑠 Proposed 1.817 0.40 9.73 0.85 10.3 0.79 0.1 23.6 0.35 14.29
𝑒
𝑠 [18] 2.68 1.766 6.58 3.60 10.55 0.67 0.05 20.64 0.47 17.32
Proposed 3.24 2.522 7.401 0.366 11.19 19.44 64.13 579.8 8.14 14.77
1.2(10𝑠+1) −1.2𝑠
𝑠(2𝑠+1)
𝑒 [41] 2.82 2.072 6.404 0.38 11.47 19.31 65.35 573 5.84 15.34
[23] 4.34 2.651 15.81 0.3951 12.26 27.33 97.98 842.4 6.18 19.72
Proposed 1.166 0.774 0.97 20.91 2.95 0.32 0.037 7.02 5.26 4.63
1.1
𝑠(0.9𝑠−1)
𝑒−0.22𝑠 [44] 0.922 0.637 0.616 34.55 3.49 0.28 0.033 6.06 7.06 3.72
[45] 1.05 0.706 0.804 49.2 2.55 0.344 0.04 7.522 12.6 4.75
𝑡𝑠 − 𝑆𝑒𝑡𝑡𝑙𝑖𝑛𝑔 𝑇 𝑖𝑚𝑒 (𝑠𝑒𝑐).

The Future endeavors will be focused on extending the real time application of this controller design to more fields, including
robots and process controllers where the time delay is more dominant.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing
interests: K DIVAKAR reports a relationship with NO that includes: employment. NO has patent pending to NO. No If there are other
authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to
influence the work reported in this paper.

Data availability

No data was used for the research described in the article.

References

[1] Rao A Seshagiri, Chidambaram M. Enhanced smith predictor for unstable processes with time delay. Ind Eng Chem Res 2005;44(22):8291–9.
[2] Zhang Weidong. Optimal design of the refined Ziegler- Nichols proportional-integral-derivative controller for stable and unstable processes with time
delays. Ind Eng Chem Res 2006;45(4):1408–19.
[3] Rao A Seshagiri, Rao VSR, Chidambaram M. Simple analytical design of modified smith predictor with improved performance for unstable first-order plus
time delay (FOPTD) processes. Ind Eng Chem Res 2007;46(13):4561–71.
[4] Shamsuzzoha M, Lee Moonyong. Analytical design of enhanced PID filter controller for integrating and first order unstable processes with time delay.
Chem Eng Sci 2008;63(10):2717–31.
[5] Tan Wen. Analysis and design of a double two-degree-of-freedom control scheme. ISA Trans 2010;49(3):311–7.
[6] del Muro-Cuéllar Basilio, Márquez-Rubio Juan Francisco, Velasco-Villa Martin, de Jesús Álvarez-Ramírez José. On the control of unstable first order
linear systems with large time lag: Observer based approach. Eur J Control 2012;18(5):439–51. [Link] URL https:
//[Link]/science/article/pii/S0947358012709627.
[7] Mehta Utkal, Rojas Rubén. Smith predictor based sliding mode control for a class of unstable processes. Trans Inst Meas Control 2017;39(5):706–14.
[8] Uma S, Rao A Seshagiri. Enhanced modified smith predictor for second-order non-minimum phase unstable processes. Int J Syst Sci 2016;47(4):966–81.
[9] Karan Somak, Dey Chanchal. Improved disturbance rejection with modified smith predictor for integrating FOPTD processes. SN Appl Sci 2019;1(10):1168.
[10] Karan Somak, Dey Chanchal, Mukherjee Surojit. Simple internal model control based modified smith predictor for integrating time delayed processes with
real-time verification. ISA Trans 2022;121:240–57.
[11] Aryan Pulakraj, Kumar Vivek, Raja G Lloyds, Chelliah Thanga Raj, Hote Yogesh V. Robust shifted-IMC based modified decoupled smith predictor for
delay-dominant integrating-type chemical processes. Int J Syst Sci 2025;1–19.
[12] Vijayan V, Panda Rames C. Design of PID controllers in double feedback loops for SISO systems with set-point filters. ISA Trans 2012;51(4):514–21.
[13] Ajmeri Moina, Ali Ahmad. Two degree of freedom control scheme for unstable processes with small time delay. ISA Trans 2015;56:308–26.
[14] Li Haisheng, et al. Tuning of PI–PD controller using extended non-minimal state space model predictive control for the stabilized gasoline vapor pressure
in a stabilized tower. Chemometr Intell Lab Syst 2015;142:1–8.
[15] Nema Swapnil, Kumar Padhy Prabin. Identification and cuckoo PI-PD controller design for stable and unstable processes. Trans Inst Meas Control
2015;37(6):708–20.
[16] Onat Cem. A new design method for PI–PD control of unstable processes with dead time. ISA Trans 2019;84:69–81.
[17] Irshad Mohammad, Ali Ahmad. Robust PI-PD controller design for integrating and unstable processes. IFAC-PapersOnLine 2020;53(1):135–40.
[18] Kaya Ibrahim. Optimal PI–PD controller design for pure integrating processes with time delay. J Control Autom Electr Syst 2021;32(3):563–72.
[19] Lloyds Raja G, Ali Ahmad. New PI-PD controller design strategy for industrial unstable and integrating processes with dead time and inverse response. J
Control Autom Electr Syst 2021;32(2):266–80.
[20] Peker Fuat, Kaya Ibrahim. Maximum sensitivity (Ms)-based I-PD controller design for the control of integrating processes with time delay. Int J Syst Sci
2023;54(2):313–32.
[21] Aryan Pulakraj, Raja G Lloyds. Equilibrium-optimized IMC-PD double-loop control strategy for industrial processes with dead time. In: Recent advances
in mechanical engineering: select proceedings of CAMSE 2021. Springer; 2022, p. 37–50.

13
K. Divakar et al. Results in Control and Optimization 21 (2025) 100635

[22] Aryan Pulakraj, Raja G Lloyds, Vilanova Ramon. Optimal iIMC-PD double-loop control strategy for integrating processes with dead-time. In: APCA
international conference on automatic control and soft computing. Springer; 2022, p. 521–31.
[23] Anil Ch, Sree R Padma. Tuning of PID controllers for integrating systems using direct synthesis method. ISA Trans 2015;57:211–9.
[24] Gopi Krishna Rao PV, Subramanyam MV, Satyaprasad K. Design of internal model control-proportional integral derivative controller with improved filter
for disturbance rejection. Syst Sci Control Eng 2014;2(1):583–92. [Link]
[25] Kumar DB Santosh, Sree R Padma. Tuning of IMC based PID controllers for integrating systems with time delay. ISA Trans 2016;63:242–55.
[26] Begum K Ghousiya, Rao A Seshagiri, Radhakrishnan TK. Maximum sensitivity based analytical tuning rules for PID controllers for unstable dead time
processes. Chem Eng Res Des 2016;109:593–606. [Link] URL [Link]
S0263876216300077.
[27] Medarametla Praveen Kumar, Komanapalli Venkata Lakshmi Narayana. Maximum sensitivity based new PID controller tuning for integrating systems using
polynomial method. Chem Prod Process Model 2017;12(3):20160070. [Link]
[28] Medarametla Praveen Kumar, Muthukumarasamy Manimozhi. A novel PID controller with second order lead/lag filter for stable and unstable first order
process with time delay:. Chem Prod Process Model 2018;13(1):20170010. [Link]
[29] Kumar Munna, Prasad Durga, Singh Ram Sharan. Performance enhancement of IMC-PID controller design for stable and unstable second-order time delay
processes. J Central South Univ 2020;27(1):88–100.
[30] Raza Ashraf, Anwar Md Nishat. A unified approach of PID controller design for unstable processes with time delay. J Central South Univ
2020;27(9):2643–61.
[31] Divakar Kuruna, Kumar M Praveen. A novel efficacious PID controller for processes with inverse response and time delay. IEEE Access 2022;10:63626–39.
[32] Divakar Kuruna, Kumar M Praveen. A novel PID controller cascaded with higher order filter for FOPDT with real time implementation. IEEE Access
2022;10:101168–85. [Link]
[33] Saxena Sahaj. Load frequency control strategy via fractional-order controller and reduced-order modeling. Int J Electr Power Energy Syst 2019;104:603–14.
[34] Gupta Ankur, Manocha Amit kumar, Singh Gurpreet. Comparative analysis of techniques of order reduction for analysis of vehicle model. Adv Math Sci
J 2020;9(7):4471–84.
[35] Raja G Lloyds. Robust I-PD controller design with case studies on boiler steam drum and bioreactor. In: 2023 15th international conference on computer
and automation engineering. IEEE; 2023, p. 486–91.
[36] Aryan Pulakraj, Raja G Lloyds, Mehta Utkal. FOI 𝜆 D 1-𝜆 controller: An alternative to double-loop control schemes for integrating processes. In: 2023
IEEE 3rd international conference on smart technologies for power, energy and control. IEEE; 2023, p. 1–6.
[37] Singha Prabir, Meena Rammurti, Chakraborty Sudipta, Raja G Lloyds. Robust pidf-pid cascade control scheme for delay-dominant stable and integrating
chemical processes. J Taiwan Inst Chem Eng 2025;173:106165.
[38] Ranganayakulu Rayalla, Seshagiri Rao Ambati, Uday Bhaskar Babu Gara. Analytical design of fractional IMC filter–PID control strategy for performance
enhancement of cascade control systems. Int J Syst Sci 2020;51(10):1699–713.
[39] Raja G Lloyds, Ali Ahmad. Design of cascade control structure for stable processes using method of moments. In: 2019 2nd international conference on
power and embedded drive control. IEEE; 2019, p. 445–9.
[40] Rao CV Nageswara, Sree R Padma. IMC based controller design for integrating systems with time delay. Indian Chem Eng 2010;52(3):194–218.
[41] Medarametla Praveen Kumar. Novel proportional–integral–derivative controller with second order filter for integrating processes. Asia-Pac. J Chem Eng
2018;13(3):e2195.
[42] Vanavil B, Anusha AVNL, Perumalsamy M, Rao A Seshagiri. Enhanced imc-pid controller design with lead-lag filter for unstable and integrating processes
with time delay. Chem Eng Commun 2014;201(11):1468–96. [Link]
[43] Chakraborty Sudipta, Ghosh Sandip, Naskar Asim Kumar. I–PD controller for integrating plus time-delay processes. IET Control Theory Appl
2017;11(17):3137–45.
[44] So GunBaek. DS based 2-DOF PID controller for various integrating processes with time delay. ISA Trans 2024.
[45] Rao A Seshagiri, Rao VSR, Chidambaram M. Direct synthesis-based controller design for integrating processes with time delay. J Franklin Inst
2009;346(1):38–56.

14