evolutionary based optimisation of multivariable fuzzy...

6
Evolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary Distillation Column Yousif Al-Dunainawi Electronic and Computer Engineering Department College of Engineering, Design and Physical Sciences Brunel University London Uxbridge, London, UK [email protected] Maysam F. Abbod Electronic and Computer Engineering Department College of Engineering, Design and Physical Sciences Brunel University London Uxbridge, London, UK [email protected] Abstract—Genetic Algorithms (GA), Simulated Annealing (SA) and Particle Swarm Optimisation (PSO) are population- based stochastic search algorithms that categorised into the taxonomy of evolutionary optimisation. These methods have been employed independently to tune a fuzzy controller for maintaining the product compositions of a binary distillation column. An analytical investigation has been conducted to distinguish the optimal tuning approach of the controller among these techniques. Based on simulation results, particle swarm optimisation approach combined with the fuzzy logic controller is identified as a comparatively better configuration regarding to its performance index as well as computational efficiency. KeywordsFuzzy Logic Control; MIMO; Distillation; Evolu- tionary Optimisation; I. I NTRODUCTION The pioneering studies of Zadeh on the theory of fuzzy set, logic and approaches [1]–[3], have motivated many researchers to establish a new discipline of the modern control system. Thus, Mamdani and his co-workers had proposed an innovative fuzzy control system for various applications to model and control dynamic, nonlinear, ill- defined and complex processes, based on fuzzy logic theory [4], [5] Later, the conception of fuzzy logic control FLC has been investigated and applied widely to the most of the engineer- ing and applied science realms [6]–[9]. Additionally, hy- bridisation of the various of evolutionary-based algorithms, providing intelligent techniques that gives FLC system a step ahead to produce powerful and more efficient solutions for different applications [10]–[12]. Employing these algorithms have been applied recently to both conventional and modern control system to find the optimal tuning parameters of the different configuration controllers [13], [14]. Instead of the traditional emphasis on accuracy and certainty, soft computing techniques can deal greatly with imprecision and uncertainty to allow reasoning and computation usually required for practical applications [11], as Zadeh stated “Fuzzy logic is not fuzzy. Basically, fuzzy logic is a precise logic of imprecision and approximate reasoning” [15]. This paper proposes an evolutionary-based multi-input multi-output fuzzy control system MIMO-FLC to maintain the product compositions of a binary distillation column as near as to desired requirements. The distillation process itself is considered a high nonlinear and characterised by uncertainty and the ill-defined relationship between the inputs and the outputs [16], [17]. The most motivated aim to introduce a new control configuration of the column is trying to find more robustness and effectiveness against the process perturbations, therefore, energy efficient columns. II. DISTILLATION COLUMNS It is well known that distillation columns are the most unit that used in oil refineries, chemical and petrochemical plants . These columns are mainly used to separate mixtures into their individuals components depending, basically, on the difference of boiling points. Distillation is reported as a highly demanding energy process. A report from the US Department of Energy has indicated that distillation column unit is the largest consumer of energy in the chemical indus- try; typically, it accounts for 40% of the energy consumed by petrochemical plant [18]. Regardless of its “hunger” for energy, distillation continues to be a widely used process for separation and purification. Therefore, efficiently oper- ating these columns necessitates a high degree of automatic control [19]. A. Process Description The column has a number of trays (plates) which are employed to enrich the components separation process. The mixture is usually fed in the middle (or around) in the column. Vapour is produced by a reboiler, which is supplied by enough heat. The steam travels up through trays inside the column to reach the top and, then, comes out to be liquefied in a condenser. Liquid from the condenser, at that point, enters into the reflux drum. As a final point, the distillate product is removed from this drum as a pure product. In addition, some liquid is fed back (reflux) close to the top, 2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation 978-1-5090-0888-9/16 $31.00 © 2016 IEEE DOI 10.1109/UKSim.2016.9 121 2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation 978-1-5090-0888-9/16 $31.00 © 2016 IEEE DOI 10.1109/UKSim.2016.9 127 2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation 978-1-5090-0888-9/16 $31.00 © 2016 IEEE DOI 10.1109/UKSim.2016.9 127

Upload: others

Post on 07-Jan-2020

7 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

Evolutionary Based Optimisation of Multivariable FuzzyControl System of a Binary Distillation Column

Yousif Al-Dunainawi

Electronic and Computer Engineering Department

College of Engineering, Design and Physical Sciences

Brunel University London

Uxbridge, London, UK

[email protected]

Maysam F. Abbod

Electronic and Computer Engineering Department

College of Engineering, Design and Physical Sciences

Brunel University London

Uxbridge, London, UK

[email protected]

Abstract—Genetic Algorithms (GA), Simulated Annealing(SA) and Particle Swarm Optimisation (PSO) are population-based stochastic search algorithms that categorised into thetaxonomy of evolutionary optimisation. These methods havebeen employed independently to tune a fuzzy controller formaintaining the product compositions of a binary distillationcolumn. An analytical investigation has been conducted todistinguish the optimal tuning approach of the controlleramong these techniques. Based on simulation results, particleswarm optimisation approach combined with the fuzzy logiccontroller is identified as a comparatively better configurationregarding to its performance index as well as computationalefficiency.

Keywords–Fuzzy Logic Control; MIMO; Distillation; Evolu-tionary Optimisation;

I. INTRODUCTION

The pioneering studies of Zadeh on the theory of fuzzy

set, logic and approaches [1]–[3], have motivated many

researchers to establish a new discipline of the modern

control system. Thus, Mamdani and his co-workers had

proposed an innovative fuzzy control system for various

applications to model and control dynamic, nonlinear, ill-

defined and complex processes, based on fuzzy logic theory

[4], [5]Later, the conception of fuzzy logic control FLC has been

investigated and applied widely to the most of the engineer-

ing and applied science realms [6]–[9]. Additionally, hy-

bridisation of the various of evolutionary-based algorithms,

providing intelligent techniques that gives FLC system a step

ahead to produce powerful and more efficient solutions for

different applications [10]–[12]. Employing these algorithms

have been applied recently to both conventional and modern

control system to find the optimal tuning parameters of

the different configuration controllers [13], [14]. Instead

of the traditional emphasis on accuracy and certainty, soft

computing techniques can deal greatly with imprecision

and uncertainty to allow reasoning and computation usually

required for practical applications [11], as Zadeh stated

“Fuzzy logic is not fuzzy. Basically, fuzzy logic is a preciselogic of imprecision and approximate reasoning” [15].

This paper proposes an evolutionary-based multi-input

multi-output fuzzy control system MIMO-FLC to maintain

the product compositions of a binary distillation column

as near as to desired requirements. The distillation process

itself is considered a high nonlinear and characterised by

uncertainty and the ill-defined relationship between the

inputs and the outputs [16], [17]. The most motivated aim

to introduce a new control configuration of the column is

trying to find more robustness and effectiveness against the

process perturbations, therefore, energy efficient columns.

II. DISTILLATION COLUMNS

It is well known that distillation columns are the most

unit that used in oil refineries, chemical and petrochemical

plants . These columns are mainly used to separate mixtures

into their individuals components depending, basically, on

the difference of boiling points. Distillation is reported as

a highly demanding energy process. A report from the US

Department of Energy has indicated that distillation column

unit is the largest consumer of energy in the chemical indus-

try; typically, it accounts for 40% of the energy consumed

by petrochemical plant [18]. Regardless of its “hunger” for

energy, distillation continues to be a widely used process

for separation and purification. Therefore, efficiently oper-

ating these columns necessitates a high degree of automatic

control [19].

A. Process Description

The column has a number of trays (plates) which are

employed to enrich the components separation process. The

mixture is usually fed in the middle (or around) in the

column. Vapour is produced by a reboiler, which is supplied

by enough heat. The steam travels up through trays inside the

column to reach the top and, then, comes out to be liquefied

in a condenser. Liquid from the condenser, at that point,

enters into the reflux drum. As a final point, the distillate

product is removed from this drum as a pure product. In

addition, some liquid is fed back (reflux) close to the top,

2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation

978-1-5090-0888-9/16 $31.00 © 2016 IEEE

DOI 10.1109/UKSim.2016.9

121

2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation

978-1-5090-0888-9/16 $31.00 © 2016 IEEE

DOI 10.1109/UKSim.2016.9

127

2016 UKSim-AMSS 18th International Conference on Computer Modelling and Simulation

978-1-5090-0888-9/16 $31.00 © 2016 IEEE

DOI 10.1109/UKSim.2016.9

127

Page 2: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

while the impure product is produced at the bottom outlet.

The distillation column diagram is depicted in Fig. 1.

Figure 1. Schematic representation of a binary distillation column

B. Model Representation

The model of a binary distillation column simulated in

this research is considered under the following assumptions:

1) No chemical reactions occur inside the column

2) Constant pressure

3) Binary mixture

4) Constant relative volatility

5) No vapour hold-up occurs at all stages

6) Constant hold-up liquid at all trays

7) Perfect mixing and equilibrium for vapour-liquid on

all stages

Hence, the mathematical expression of the model can be

represented by the following equations:

• On each tray (excluding reboiler, feed and condenser

stages):

Midxi

t= Li+1xi+1 + Vi−1yi−1 − Lixi − Viyi (1)

• Above the feed stage i = NF + 1:

Midxi

t= Li+1xi+1+Vi−1yi−1−Lixi−Viyi+FV yF

(2)

• Below the feed stage i = NF :

Midxi

t= Li+1xi+1+Vi−1yi−1−Lixi−Viyi+FLXF

(3)

• In the reboiler and column base i = 1, xi = xB:

MBdxi

t= Li+1xi+1 − Viyi +BxB (4)

• In the condenser, i = N + 1, xD = xN + 1:

MDdxi

t= Vi−1yi−1 − LixD −DxD (5)

• Vapour-liquid equilibrium relationship for each tray

[20]:

yi =αxi

1 + (α− 1)xi(6)

• The flow rate at constant molar flow:

Li = L, V i = V + FV (7)

since

FL = qF × F (8)

Fv = F + FL (9)

• The flowrate of both condenser and reboiler as: Re-

boiler:

B = L+ FL − V (10)

Condenser:

D = V + FV − L (11)

• The feed compositions xF and yF are found from the

flash equation as:

FzF = FL × xF − FV × yF (12)

The nominal and operation conditions of the column are

shown in the appendix in the end of this paper, and the

schematic diagram of a theoretical stage of the column is

shown in the Fig. 2.

Figure 2. Schematic diagram of ith stage of a binary distillation column

122128128

Page 3: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

III. CONTROL SYSTEM CONFIGURATION

In the present work, MIMO configuration is investigated

to control the product compositions of a binary distillation

column. One of the most common control loops of the binary

column is so-called (L− V ) configuration [19]. Where, the

reflux flow (L) is selected to control the mole fraction of

the top product (xD), while the reboiler steam flow (V ) is

chosen to control the composition of the bottom product, as

expressed in Eq. 13.[xDxB

]= GLV

[LV

](13)

where GLV is the column’s transfer function.

As it is known, the performance of any controller

depends as much on its design as on its tuning. Tuning

must be applied by operators to fit the controller to the

process. Therefore, there are many different approaches for

tuning, based on the particular performance criteria selected.

Therefore, evolutionary algorithms are employed as an

optimal tuner of the control parameters; genetic algorithm,

simulated annealing and particle swarm optimisation in this

paper for this purpose.

The control system aims to maintain the product compo-

sitions as close to the desired ones as possible despite the

expected disturbances.

A. Fuzzy logic control

A control system, based on fuzzy theory, simply trans-

forms a linguistic control strategy into automatic capabilities

for managing the nonlinearities and uncertainties of the

process. The proposed FLC structure has Mamdani infer-

ence system and the centroid defuzzification mechanisms.

The universe of discourse of the inputs and output were

normalised in the interval [1, 1]. Thus, actual system values

were converted through scaling parameters (gains). Gaus-

sian,the most common membership function was used to

defined the inputs and output of the fuzzy controller. the

relationship of the inputs as error and change in error, and

the output as control signal is shown in the surface viewer

of FLC in Fig. 3 as well as the if-then rules are detailed in

Table I .

TABLE IRULES BASED OF THE FLC.

ErrorNB NM NS SS PS PM PB

Ch

ang

ein

Err

or NB SS NS NM NM NB NB NB

NM PS SS NS NM NM NB NBNS PM PS SS NS NM NM NBSS PM PM PS SS NS NM NMPS PB PM PM PS SS NS NMPM PB PB PM PM PS SS NSPB PB PB PB PM PM PS SS

Linguistic expressions demonstrated to these fuzzy sets

are: PB: positive big, PM: positive medium, PS: positive

small, ZE: zero, NS: negatives small, NM: negative medium

and NB: negative big.

The Gaussian membership membership function is ex-

pressed as:

y(x) = e−(x−c)2

α (14)

where c and α are the mean and deviation of a Gaussian

membership function, respectively. The resulting fuzzy set

must be converted to a signal that can be sent to the process

as a control input. Centroid of the area has been used here

for the defuzzification process.

B. Genetic algorithms

Natural evolution inspired procedure as known genetic

algorithms GAs, can search for optimal or close-optimal

solutions for an optimization problem over the search space.

It generates an initially random population of candidate

solutions toward the optimal fitness (objective function) by

performing specific techniques, which are mimic the natural

selection processes, such as reproduction, crossover, and

mutation. The procedures are repeated until the prescribed

fitness is accepted, or the predetermined number of itera-

tions (generations) is implemented. The research topic of

tuning various control systems via GAs has already been

investigated in many researches [9], [21], [22].

C. Simulated Annealing

Simulated annealing is a heuristic optimisation method,

which is mimicking the process of metals annealing to

find the optimum solution via controlling the temperature.

SA initialises a candidate temperature, in optimization and

searches for optimal global fitness by slowly reducing the

temperature alike to the physical annealing process. Its

Figure 3. The relationship between inputs and output of the fuzzycontroller.

123129129

Page 4: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

advantages are reported as the considerable ability to find

the minimal fitness function in specific conditions as well

as it can deal with any objective function [23].

D. Particle Swarm Optimisation

Particle swarm optimisation (PSO) has been proposed by

Kennedy and Eberhart in 1995 [20] and 2001 [24], PSO

algorithm turned to be vastly successful. The several of

researchers have presented the merit of the implementation

of PSO as an optimiser for various applications [25], [26].

In PSO procedure, all particles are located randomly and

theoretical to move arbitrarily in a defined direction in

the search domain. Each particle direction is then changed

steadily to travel along the direction of its best previous

positions to discover a new better position according to

predefined objective function (fitness).

IV. SIMULATION AND RESULTS

In this paper, MIMO FLC has been designed of a binary

distillation column. EAs such as GA, SA and PSO were

employed to find the optimal scaling factors of the controller.

The integral of the time-weighted absolute error (ITAE) was

selected as the quantitative criterion for measuring control

performance. Minimisation of this index expressed in Eq.

15 is considered as fitness or objective function of EAs that

is used in this research.

ITAE =

∫ T

t=0

T × |E| dt (15)

where, E is the error between the desired value and output

of the product compositions of the column and T is a

simulation time, the schematic diagram of the designed

control system is depicted in Fig. 4.

To compare the performance of the various EAs for the

FLC controller design, three simulation experiments were

performed as follows:

1) Design of MIMO FLC controller without a compen-

sator

2) Design of MIMO FLC controller with a compensator

Figure 4. EA-based FLC design

3) Design of MIMO PID controller tuned by conventional

Ziegler-Nichols method [27]

Extensive simulations were carried out to find the optimal

initial parameters of EAs like the population size, the initial

condition, weight, etc.

Due to the randomness of EAs at initialisation stage,

20 times of runs had been done independently of each

algorithm. MATLAB R© and Simulink R© R2014b platform

were used for simulation via processor 3.6 GHz, with 8 GB

of RAM.

The FLC configuration with and without compensator

are shown in Fig. 5. The performance index of the different

controllers with various tuning method is given in Table II.

Clearly, all of the controllers designed using different

approaches pass the transient response requirements. Never-

theless, the performance of the PSO-base FLC with compen-

sator indicated better achievement regarding the performance

index and transit response. In addition, PSO outperformed

the other EAs techniques with minimum computation time.

(a)

(b)

Figure 5. MIMO FLC (a) without compensator, (b) with compensator

124130130

Page 5: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

TABLE IIPERFORMANCE INDEX OF CONTROLLERS TUNED BY VARIOUS

APPROACHES

Controller Tuning method ITAE Time (hour)PID Z-N 62.633 -

FLCGA 48.335 6.845SA 53.07 6.236

PSO 42.99032 4.5734FLC-compensator PSO 40.079 5.231

The compensator slightly improved the performance by

eliminating the interacting of loops, with more time cost due

to the number of parameters involved. For the convenience,

the comparisons of the time response behaviour of the

PID and PSO-based FLC with and without compensator is

presented in Fig. 6.

Time(min)0 50 100 150

xD

0

0.2

0.4

0.6

0.8

1

1.2

1.4PIDPSOFLC PSOFLC with Compensator

(a)

Time (min)0 50 100 150

xB

0

0.2

0.4

0.6

0.8

1

1.2

1.4PIDPSOFLC with CompensatorPSOFLC

(b)

Figure 6. Time Response of step setpoints of different control configura-tions (a) distillate and (b) bottoms product.

A. Robustness of the optimal controller

To check the robustness of the PSO-based FLC

with compensator against external disturbances, the

desired compositions of the column were set to change

asynchronously as follows; The desired compositions of

distillate and bottoms products are changing every 100

minutes for a thousand minutes. It can be observed that the

control actions of the controller were successfully adapted

to eliminate the effect of the external disturbances, where

the convergence of the desired responses was achieved after

the adaptation of the control output. The process output

responses to setpoints changes in the distillate and bottoms

compositions. Better performance is achieved with fuzzy

logic controller that tuned by PSO with a compensator to

eliminate the effect of inputs variance as shown in Fig. 7.

This result gives an indication that the proposed controller

can cope efficiently with disturbances.

(a)

(b)

Figure 7. Time response of changed-step setpoints of the product compo-sitions of the column (a) distillate and (b) bottoms product.

125131131

Page 6: Evolutionary Based Optimisation of Multivariable Fuzzy ...uksim.info/uksim2016/CD/data/0888a127.pdfEvolutionary Based Optimisation of Multivariable Fuzzy Control System of a Binary

V. CONCLUSIONS

In this research, three of the common techniques of EA

were performed independently as a tuner to FLC. GA,

SA and PSO had been combined with FLC to control a

binary distillation column. The results showed that PSO

outperformed GA and SA by achieving improvement to

the performance of the controller as well as computational

efficiency. For comparison purposes, the conventional PID

controller was also simulated and applied to the same

column. PSO-based FLC with compensator proved its feasi-

bility and superiority by handling disturbances with minimal

ITAE performance index. Different control configurations

could be applied to distillation columns with various tuning

method like Gravitational Search Algorithm, which is to be

the subject of future work.

ACKNOWLEDGMENT

The corresponding author is grateful to the Iraqi Ministry

of Higher Education and Scientific Research for supporting

the current research.

REFERENCES

[1] L. A. Zadeh, “Fuzzy sets,” Information and control, vol. 8, no. 3, pp.338–353, 1965.

[2] L. Zedeh, “Fuzzy algorithms,” Information and Control, vol. 12, pp.94–102, 1968.

[3] L. A. Zadeh, “Fuzzy logic,” Computer, no. 4, pp. 83–93, 1988.

[4] P. J. King and E. H. Mamdani, “The application of fuzzy controlsystems to industrial processes,” Automatica, vol. 13, no. 3, pp. 235–242, 1977.

[5] E. H. Mamdani, “Twenty years of fuzzy control: experiences gainedand lessons learnt,” in Fuzzy Systems, 1993., Second IEEE Interna-tional Conference on. IEEE, 1993, pp. 339–344.

[6] T. Takagi and M. Sugeno, “Fuzzy identification of systems and itsapplications to modeling and control,” Systems, Man and Cybernetics,IEEE Transactions on, no. 1, pp. 116–132, 1985.

[7] B. N. Alajmi, K. H. Ahmed, S. J. Finney, and B. W. Williams,“Fuzzy-logic-control approach of a modified hill-climbing method formaximum power point in microgrid standalone photovoltaic system,”Power Electronics, IEEE Transactions on, vol. 26, no. 4, pp. 1022–1030, 2011.

[8] M. M. Algazar, H. A. EL-halim, M. E. E. K. Salem et al., “Maximumpower point tracking using fuzzy logic control,” International Journalof Electrical Power & Energy Systems, vol. 39, no. 1, pp. 21–28, 2012.

[9] A. Abbadi, L. Nezli, and D. Boukhetala, “A nonlinear voltagecontroller based on interval type 2 fuzzy logic control system formultimachine power systems,” International Journal of ElectricalPower & Energy Systems, vol. 45, no. 1, pp. 456–467, 2013.

[10] W. Pedrycz, Fuzzy modelling: paradigms and practice. SpringerScience & Business Media, 2012, vol. 7.

[11] L. R. Medsker, Hybrid intelligent systems. Springer Science &Business Media, 2012.

[12] A. Abraham, “Hybrid intelligent systems: evolving intelligence inhierarchical layers,” in Do Smart Adaptive Systems Exist? Springer,2005, pp. 159–179.

[13] M. I. Menhas, L. Wang, M. Fei, and H. Pan, “Comparative perfor-mance analysis of various binary coded pso algorithms in multivari-able pid controller design,” Expert systems with applications, vol. 39,no. 4, pp. 4390–4401, 2012.

[14] M. Unal, A. Ak, V. Topuz, and H. Erdal, Optimization of PIDcontrollers using ant colony and genetic algorithms. Springer, 2012,vol. 449.

[15] L. A. Zadeh, “Is there a need for fuzzy logic?” Information sciences,vol. 178, no. 13, pp. 2751–2779, 2008.

[16] A. A. Kiss, Advanced distillation technologies: design, control andapplications. John Wiley & Sons, 2013.

[17] R. W. Baker, Membrane separation systems: recent developments andfuture directions. Noyes Publications, 1991.

[18] C. L. Smith, Distillation control: An engineering perspective. JohnWiley & Sons, 2012.

[19] S. Skogestad, “Dynamics and control of distillation columns: Atutorial introduction,” Chemical Engineering Research and Design,vol. 75, no. 6, pp. 539–562, 1997.

[20] M. Tsuzuki and T. Martins, Simulated Annealing: Strategies, PotentialUses and Advantages. Nova Science Publishers, Inc.

[21] H.-X. Li and H. Gatland, “Conventional fuzzy control and its enhance-ment,” Systems, Man, and Cybernetics, Part B: Cybernetics, IEEETransactions on, vol. 26, no. 5, pp. 791–797, 1996.

[22] D. Pelusi, L. Vazquez, D. Diaz, and R. Mascella, “Fuzzy algorithmcontrol effectiveness on drum boiler simulated dynamics,” in Telecom-munications and Signal Processing (TSP), 2013 36th InternationalConference on, July 2013, pp. 272–276.

[23] F. Herrera, M. Lozano, and J. L. Verdegay, “Tuning fuzzy logic con-trollers by genetic algorithms,” International Journal of ApproximateReasoning, vol. 12, no. 3, pp. 299–315, 1995.

[24] R. C. Eberhart, J. Kennedy et al., “A new optimizer using particleswarm theory,” in Proceedings of the sixth international symposiumon micro machine and human science, vol. 1. New York, NY, 1995,pp. 39–43.

[25] J. Kennedy, J. F. Kennedy, R. C. Eberhart, and Y. Shi, Swarmintelligence. Morgan Kaufmann, 2001.

[26] Y. Al-Dunainawi and M. F. Abbod, “Pso-pd fuzzy control of distilla-tion column,” in SAI Intelligent Systems Conference (IntelliSys), 2015,Nov 2015, pp. 554–558.

[27] J. G. Ziegler and N. B. Nichols, “Optimum settings for automaticcontrollers,” trans. ASME, vol. 64, no. 11, 1942.

APPENDIX

Abbreviations, the operating conditions and technical as-

pects of the distillation column are detailed in following

table.

Symbol Description Value UnitN Number of trays 20 -NF Feed stage location 11 -F Typical inlet flow rate to the col-

umn1 kmol/min

D Typical distillate flow rate 0.5 kmol/minB Typical bottoms flow rate 0.5 kmol/minzF Light component in the feed (mole

fraction)0.5 -

qF Mole fraction of the liquid in thefeed

1 -

L Typical reflux flow rate 1.28 kmol/minV Typical boil-up flow rate 1.78 kmol/minα Relative volatility 2 -xD Distillate composition (mole frac-

tion)0.98 -

xB Bottoms composition (mole frac-tion)

0.02 -

i Stage number during distillation - -x Mole fraction of light component

in liquid- -

y Mole fraction of light componentin vapour

- -

M Tray hold-up liquid 0.5 kmolMD Condenser hold-up liquid 0.5 kmolMB Reboiler hold-up liquid 0.5 kmol

126132132