Automatika

Journal for Control, Measurement, Electronics, Computing and

Communications

ISSN: 0005-1144 (Print) 1848-3380 (Online) Journal homepage: [Link]

Mathematical modelling and controller design

using electromagnetic techniques for sugar
industry process

[Link] & [Link] Kumar

To cite this article: [Link] & [Link] Kumar (2019): Mathematical modelling and
controller design using electromagnetic techniques for sugar industry process, Automatika, DOI:
10.1080/00051144.2019.1653667

To link to this article: [Link]

© 2019 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 16 Aug 2019.

Submit your article to this journal

View Crossmark data

Full Terms & Conditions of access and use can be found at

[Link]
AUTOMATIKA
[Link]

SPECIAL ISSUE: COMPUTATIONAL ELECTROMAGNETICS

Mathematical modelling and controller design using electromagnetic

techniques for sugar industry process

[Link] and [Link] Kumarb

a Department of Electronics and Instrumentation Engineering, [Link] College of Engineering, Karur, Tamil Nadu, India;
b Department of Electrical and Electronics Engineering, Bannari Amman Institute of Technology, Sathyamangalam, Tamil Nadu, India

ABSTRACT ARTICLE HISTORY

The design and tuning of controllers for the industrial process is an art, which plays a vital role in Received 8 January 2019
many of the process control industries. The industries such as sugar, chemical, paper and other Accepted 29 April 2019
process control-based industries hold the nature of non-linearity due to their uneven distribution KEYWORDS
of load influence with disturbances. This paper highlights the study and investigation of differ- Controller; level; stability;
ent controlling parameters associated with the sugar industry which is located in the southern PID; fuzzy; process analysis
region of Tamil Nadu, India. After the analysis of various processes, the impact of controlling strat-
egy needs to be implemented in an advanced way to minimize the resource utilization with high
productivity. The behaviour of various process parameters, such as flow, temperature, level, pres-
sure, is studied and one of these has been discussed in detail along with the controller design. The
controller, such as proportional, integral, derivative, has been modified with advanced control
phenomena and applied to process model with the presence of delay time provides the excellent
tracking of operating point to bring the stability of a certain process in a tremendous manner.

1. Introduction and Minyue fu have developed a mathematical model

There are nearly more than 400 sugar industries are along with a predictive controller for juice extraction
available in India. The states in India that mainly con- [2]. The white sugar is normally used in house as well
tribute in sugar production are Tamil Nadu, Andhra as food production. The size of the sugar particle is
Pradesh, Gujarat, etc. According to the survey of U.S. usually in mm and the final stage of the sugar pro-
department of Agriculture, India stands second place in duction process is called as crystallization. J. Michal
the production of sugar. Sugar industry process is the and M. Kminek derived an expert system with the
combination of huge control loops, cascade as well as process model and it was proved versatile [3]. Even
multiloops and variable inputs to the system. It requires though some advanced controlling technologies have
continues monitoring and control of non-linear pro- been employed in industries, the fault occurrence is one
cess parameters such as level, flow, pressure, tempera- of the unavoidable issue. Silvio simani has undergone
ture, etc., to ensure the process stability, labour safety the study of fault diagnosis at sugar cane crushing unit
with improving production rate. The process struc- and published an article using the fuzzy model [4]. The
ture or process model is analysing the variables such control of pH value in the chemical process becomes
as input, output and other constraints. The alternate very complex and a challenging task to maintain in a
method of process model identification will be based on steady-state condition. In the sugar industry, pH neu-
the nature of the system parameters. i.e. height, radius tralization is an important criterion to yield the product
and other system-dependent constraints. The mod- with good quality. O. Elfatni et T. Bounahmidi con-
elling techniques for various linear and linear processes ducted an experiment by determining the mathemat-
are dealt with in an extraordinary way by Bequette [1]. ical model for the acidification process on sugar extrac-
The flow chart of the sugar industry process is shown tion [5]. In the industrial process, the occurrence of
in Figure 1. All the units have its individual control delay in run-time process has never been eliminated at
loop architecture to have connectivity for its previous any controller action. Julio E. Normey-Rico and Alejan-
and next process unit. The sugar production starts from dro Merino proposed a compensation technique with
the extraction of cane juice from sugar cane in var- robust control in the evaporation process [6]. Astrom
ious stages at the sugar mill. The efficient extraction K. J. and T. Hagglund have developed a method of tun-
of cane juice ensures the rate of sugar production and ing PID controller with fuzzy rules for the interacting
minimizes the demand of sugar cane. Turker ozkocak process [7].

CONTACT [Link] vsaravananei@[Link]

© 2019 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ([Link] which permits unrestricted
use, distribution, and reproduction in any medium, provided the original work is properly cited.
2 [Link] AND [Link] KUMAR

Figure 1. Flowchart of sugar industry process.

Table 1. Few identified issue as major drawback of sugar industry if the controller are failure to control the certain process parameter.
Parameter Unit Identified problem Remedies
Level Raw juice tank level Consumption of heat energy is not in Need to maintain raw juice tank
balanced with input juice quantity level by regulating the cane feeder
at heater, evaporation through crushing process
Flow Raw juice, steam or any fluid flow Process stability affected Raw juice pump speed control using
VFD drive
pH Final pH to be control It will affect clarification syrup colour Lime dosing through by VFD pump
Temperature Final heater juice temperature It will increase the process time Inlet heating vapour to be controlled
by a control valve
Pressure Super heater water wash pressure control It will affect sugar colour Shower pump control using VFD drive

2. Problem identification from truck and to load in a cane feeder. It consists of

four sequence processes such as chopper (C), leveller
The important task in sugar industry process is the con-
(L), pusher (P) and fibriser (F).
trol of process parameter at the right time. According to
The four different operations are the primary pro-
this investigation at the sugar industry if any engineer
cesses in a sugar mill. In chopper, sugar canes are cut
fails to notice some important issues or an abnormal
into small pieces by using its knife-based arrangements.
condition in process that affects the entire process cycle
The cutted pieces are regulated through the leveller and
since all the processes are sequential.
by applying high pressure with the aid of a pusher. The
outer hard portion of sugar cane is removed by treat-
ing it with a fibriser. Finally, the sugar cane is fed to
3. Process study the mill through an elevator from the fibriser. Five sets
From Table 1, it is clear that we need to bring our of mills are consequently connected together and each
industrial process controlling technique with advance mill consists of three to six rollers made up of heavy
control tuning. The cane juice extraction process has iron with high pressure. The cane juice from mills 1
been taken for case study. The schematic diagram of and 2 is collected in tank-1 whereas the juice from
crushing process is shown in Figure 2. The purpose mills 3,4 and 5 is collected in their corresponding juice
of this process is to maximum the extraction of cane storage tank and it will be recycled, as shown in the
juice from sugar cane to produce sugar. This process is schematic diagram, using a pump. The juice from tank-
an example of a cascade-type control system; they are 1 is sent to a raw juice tank incorporated with a rotary
cane feeder, crushing mill and raw juice tank. The ini- screen and a sand catcher tank for purifying the cane
tial stage of the crushing process is to unload the cane juice. In the cascade process, the role of controller is to
 AUTOMATIKA 3

Figure 2. Schematic diagram for sugar cane crushing process.

Table 2. Level and speed calibration. to develop new controller schemes to enhance the pro-
Level cess performance and in this paper, we investigate on
transmitter Raw juice tank Cane carrier motor speed with the process characteristics with controller design. The
output in mA level in % respect to VFD output (RPM)
various observations have been taken based on the sim-
4 0 0
12 50 750
ulation results and summarized in the following sec-
15 62.5 937.5 tions. Non-linear systems are highly occupied in chem-
18 75 1125 ical industries, process control industries such as sugar
20 100 1500
plant, paper plant and cement plant. The important
criterion in controller design is tuning of gain parame-
maintain the raw juice tank in optimum level by varying ters. Tuning is an art of any controller implementation
the cane feeder or conveyor motor speed using variable process and a few popular tuning methods for PID
frequency drive. controllers are certain methods like Ziegler Nichols
tuning method, Cohen-Coon method and reaction
curve method. K. J. Astrom and T. Hagglund et al.
3.1. Need of raw juice tank-level control clearly explained the basic characteristics of PID con-
The raw juice tank is connected with a raw juice feed troller and its compatibility with different controlling
pump and it will be connected to further processes such issues [8]. T. Balasubramanian and N. Anandaraman
as clarification and evaporation. Figure 2 demonstrates exclusively designed the same controller for FOPDT
the data acquisition process through a level transmitter. and implemented successfully. The same controller has
The output of level transmitter is in the range 4–20 mA been modified using fuzzy logic i.e. a knowledge-based
and this will be calibrated as tank level from 0% to 100% controller also gives better control and both simula-
by the data acquisition system. The actual height of the tion and real-time results were compared [9]. One of
tank is 4 m and its radius is 3 m. The controller has a the critical tasks in petroleum separation process is to
reference set point and it has to be compared with the isolate the oil and gas due to its complex mixing and
transmitter output. Sensors are playing the major role structure of molecule [10]. The controller requires an
for data acquisition with many check points. advanced or intelligent system to yield the product with
The primary objective of controller is to drive the maximum quality. R. F. Liao et al propose an intelli-
cane carrier or conveyor motor at an optimum speed gent control system to constitute a knowledge-based
and this will be discussed briefly in the controller design system i.e. fuzzy logic controller to the petroleum sep-
session. Table 2 represents the theoretical calibration of aration process. It was implemented in real time and
level and speed, but in practice, the value may get differ performed well. The closed loop structure for level con-
slightly (Figure 3). trol is shown in Figure 4 and it was clearly explained
in the previous session. The mathematical model for
the above process was determined using system identi-
4. Controller design fication. The obtained model is second-order plus delay
Controllers act as the backbone of process control and time (SOPDT).
automation-based industries. PID controllers are the
only controllers where industries are often used because KeTd S
G(S) = . (1)
of their robustness and easy accessibility. It is required (1 + Tp1 S)(1 + Tp2 S)
4 [Link] AND [Link] KUMAR

Figure 3. Data acquisition system of raw juice tank.

Figure 5. Closed loop system with PID controller.

Figure 4. Closed loop system for tank-level control.

From Figure 4, it is required to choose the controller

adopt to this process and its design procedure will be
continued in the following session.

4.1. PID controller design

The PID controller is a combination of proportional,
integral and derivative controller, which is widely
employed in most of the industrial control applications
because of its simplicity. Most of the closed loop-based
industrial process is controlled only with the help of
PID controllers here (r) is the reference input signal and Figure 6. Closed loop system with fuzzy – PID controller.
(e) shows the tracking error corresponding to the feed-
back signal (y) where the controller action is based on The signal from the controller (u) fed to the plant com-
the tracking error signal. Kp is Proportional gain, Ki prises the gains of proportional, integral and derivative
is Integral gain, Kd is Derivative gain, U is controller mode. It can be expressed as
output and G(S) is plant model. Among the traditional  t
tuning methods for PID controllers in control theories, de
u = Kp e + Ki edt + Kd . (3)
Ziegler Nichols tuning method, one of the conventional 0 dt
controllers, is widely used in industrial PID tuning. The gain values are obtained based on one of the Ziegler
The controller equation of PID controller is gener- Nichols where without integral and derivative action
ally expressed as follows. slowly raise the proportional gain value till getting the
Ki sustained oscillations. The value at which the sustained
C(S) = Kp + + Kd . (2) value is denoted as K cr (K p = K cr ) and Pcr denotes the
s
 AUTOMATIKA 5

Figure 7. Two input and three output mamdani-type FIS editor.

Figure 8. a. Surface view of the output kp b. Surface view of the output ki . c. Surface view of the output kd .

time required for the period of oscillation. The meth- 4.2. Fuzzy-PID controller
ods of calculating PID controller gain parameters are
Even though our conventional controller shows a bet-
tabulated below.
ter response, the time domain characteristics are not
With respect to the sustained oscillations obtained
in a satisfactory level. The response exhibits more rise
the tuning rule Proportional gain (K p ) = 0.147, Inte-
time, settling time with peak overshoot. In order to
gral gain (K i ) = 5.265 and Derivative gain (K d ) =
improve the system performance criteria, the conven-
1.316. The transient response of PID controller is short-
tional controller should be driven by the knowledge-
listed from Figures 10 and 12 and its performance
based logical system. The various approaches like Neu-
analysis has been discussed using transient response
ral network, fuzzy logic controller and optimization
characteristics in result and discussion (Figure 5).
6 [Link] AND [Link] KUMAR

Figure 9. Rule viewer of the output kp , ki and kd .

techniques have to be invoked for designing intelli- controller reduces the overshoot with minimum set-
gent controllers. Among that fuzzy logic controller is tling time.
a knowledge-based rule base system which is coupled
with the PID controller to make a hybrid control system
called Fuzzy-PID controller, as shown in Figure 6. 5. Result and discussion
This kind of hybrid system is designed with error As of the analysis on process control variable and con-
feedback and it will ensure the system to reach an oper- troller design from sugar industry process, the entire
ating point in less time and reduce overshoot during controller design is based on process nature may be
any external disturbance. The four main steps of the linear or non-linear and its controller variable. PID
fuzzy logic controller are Fuzzification, Fuzzy Inference controllers are providing efficient control in most of
System with the knowledge base and Defuzzification. the industries. Many authors have investigated the
Fuzzification is the process of converting the physical characteristics of various industrial processes and pro-
variables into fuzzy variables with appropriate member- vided the solutions for better improvement in a certain
ship functions. Fuzzy rules such as IF . . . THEN rules process.
are designed in the inference system along with knowl- Several research studies are carried out in the area
edge base. Finally, the fuzzy variables are converted of process control and few of them have been discussed
into real-world physical variables using defuzzification. in controller design and even though the controller is
The design of fuzzy controller involves two inputs and the same, it was applied to various real-time applica-
three outputs of mamdani-type fuzzy inference system, tions. This paper describes how the controller reacts for
as shown in Figure 7. The two inputs are error e and the author’s application considered in the sugar indus-
change in error e whereas the three outputs are kp , kp try. Among the various non-linear process available
and kd . in the sugar industry, the raw juice tank-level control
The input (e and e) and output (kp , ki , kd ) are process is selected because of complex cascade sys-
normalized into the interval [−1,1] and [0.1], respec- tems associated with it. The PI, PID, Fuzzy-PID con-
tively. The fuzzy sets are constructed using linguis- troller are developed in the SOPDT system and its
tic labels “Negative Large” (NL), “Negative Medium” transient response characteristics have been given in
(NM), “Negative Small” (NS), “Zero” (ZO), “Positive Table 3. When compared with my result and other cited
Large” (PL), “Positive Medium” (PM), “Positive Small” research article the choice of controller depends on the
(PS). Figure 8(a)–(c). shows the surface viewer gen- application and not based on our choice.
erates based on 40 nine rules created for both error As shown in Table 3, PI controller provides better
and change in error. Figure 9 shows the rule viewer controlling compared to PID type. The settling time is
that provides information of rules execution during the not finite with the derivative controller and it shows
simulation. The response of Fuzzy-PID controller is continuous oscillations, as shown in Figure 10 and the
shown in Figures 13 and 14. It has been observed that response curve, as shown in Figure 11, proves that pro-
when compared to conventional controller proposed portionality with integral action exclusively gives good
 AUTOMATIKA 7

Table 3. Controller design based on sustained oscillations.

Controller Kp Ti Td
P 0.5 kcr ∞ 0
PI 0.45 kcr Pcr /1.2 0
PID 0.6 kcr Pcr /2 Pcr /8

Figure 12. Servo response of PI controller with set point

tracking.

Figure 10. Response of PID controller.

Figure 13. Response of Fuzzy-PID controller

Figure 11. Response of PI controller.

control stability with process delay. Since the process is

non-linear and an industrial application, the controller
should react with every set point change. Fortunately
for this process, my tuning parameter has given tremen-
dous response for every change in set point, as shown
in Figure 12. To improve the process performance, the
fuzzy controller has been designed and its responses Figure 14. Servo response of Fuzzy-PID controller with an
exhibit with less sustained oscillations and reduced set- operating point tracking system.
tling time, as shown in Figures 13 and 14, respectively
(Table 4).
6. Conclusion and future work
Research works prove that the real-time output
through SCADA and the proposed controller gain Research works designed and implemented PI, PID
parameters are almost close to the actual controller controller to enhance the performance of controller
plant. It shows that the proposed PI controller has given and tracking provision under various operating points
70% better result in real time and the following figure with delay time and its output has been shown in
has been taken by a trial test (Table 5). Figure 15. The research works identified that controller
The level and cane carrier motor speed are vary- performance need to be equipped well and to over-
ing with a control action, when the level decreases come those controller demerits a fuzzy-PID controller
immediately, carrier motor speed increases to raise the is designed to tune the gain parameters on knowl-
tank level and vice versa. Few experimental values for edge base with respect to error and change in error
different level and speed are tabulated. with delay and its response is shown in Figures 13 and
8 [Link] AND [Link] KUMAR

Table 4. Controller performance analysis based on transient response.

Rise time (tr ) Settling time Peak time (tp ) Overshoot (OS)
Controller type (min) (ts ) (min) (min) (%)
PID 1.8 > 60 2.7 19
PI 1.5 30 1.7 72
Fuzzy – PID 1.5 8 1.9 35

Table 5. Level and cane carrier speed. India for their kind support and provision for data access to
develop the plant model and controller design.
Level (percentage of Cane carrier motor
S. no process variable) speed in rpm
1 0 0 Disclosure statement
2 16 196
3 28 222 No potential conflict of interest was reported by the authors.
4 35 296
5 42 510
References
[1] Bequette BW. Process control: modeling, design, and
simulation. New York: Prentice Hall; 1998.
[2] Ozkocak Turker, Minyue Fu, Graham C. Goodwin.
Maceration control of a sugar cane crushing mill.
In: Proceedings of the American Control Conference;
Chicago, Illiois; 2000. p. 2255–2259.
[3] Michal J, Kminek M, Kminek P. Expert control
of vacuum pan crystallization. IEEE Control Syst.
1994;14(5):28–34.
[4] Simani S. Fuzzy model identification of a sugar cane
Figure 15. Real-time controller response with the proposed crushing process for fault diagnosis application. In: Pro-
PID controller gain values. ceedings of the 44th IEEE Conference on Decision and
Control, and the European Control Conference, Seville,
Spain; 2005. p. 2053–2057.
14, respectively. Added to that researcher deals with [5] Elfatni O, Bounahmidi T. Mathematical modeling and
the investigations of issues and fault occurrence in the simulation of acidification process for sugar extraction.
sugar industry which may affect the process perfor- Computational Engineering in Systems Applications;
IMACS Multiconference; 2006 Oct 4–6. p. 1202–1208.
mance in various aspects. The proposed controller may [6] Normey-Rico JE, Merino A, Cristea S, et al. Robust
not give better response all time and it is required to dead-time compensation of a evaporation process in
use sophisticated techniques for further excellent pro- sugar production. Prague: Elsevier, IFAC; 2005.
cess controlling strategy. The undergoing and future [7] Astrom KJ, Hagglund T. PID controllers, theory, design
assessment related to problem statement as discussed in and tuning. 2nd ed. Instrument Society of America;
1995.
the problem identification session, new approaches like
[8] Astrom KJ, Hagglund T. The future of PID control.
fault optimization, hybrid controller design, prototype Control Eng Pract. 2001;9:1163–1175.
model design are under progress. [9] Balasubramanian T, Anandaraman N. Design of intel-
ligent controllers for nonlinear processes. Asian J Appl
Sci. 2008;1:33–45.
Acknowledgement [10] Liao RF, Chan CW, Hromek J, et al. Fuzzy logic con-
The authors would like to acknowledge the unit head of sugar trol for a petroleum separation process. Eng Appl Artif
industry which is located in the southern part of Tamil Nadu, Intell. 2008;21:835–845.