Improved multiobjective differential evolution (MODE) approach for purified terephthalic acid (PTA) oxidation process

Research output: Contribution to journalArticle

33 Citations (Scopus)

Abstract

Evolutionary algorithms (EAs), which are basically the population-based search algorithms, are becoming increasingly popular for solving highly nonlinear and complex optimization problems (both the single-objective and multiobjective problems). Multiobjective Differential Evolution (MODE), which is an extension of Differential Evolution (DE) to multiobjective optimization, has been successfully applied to a few chemical process industrial problems. This article presents multiobjective optimization of p-xylene oxidation in the manufacturing process of industrial Purified Terephthalic Acid (PTA) oxidation using MODE, its improved strategy, i.e., Elitist-Multiobjective Differential Evolution (E-MODE), and NSGA-II. E-MODE algorithm includes the concept of elitism as present in NSGA-II. Four industrially important cases are studied from the multiobjective optimization point of view. The effect of six decision variables identified, namely, catalyst concentration, vent oxygen content, % water in solvent, feed xylene rate, temperature of reactor, and total feed rate, on the formulated objective function for the four cases is studied. The corresponding Pareto optimal sets (i.e., sets of equally good nondominated solutions) are obtained. All decision variables, corresponding to the Pareto optimal solutions, are plotted against each of the objectives. Comparison among the Pareto optimal fronts of four cases is also carried out. Pareto solutions obtained in this study (using MODE, E-MODE, and NSGA-II) lie on same (Global) Pareto front. However, MODE is able to cover a better range than E-MODE and NSGA-II. The results obtained are useful to a plant engineer for possible improvement in the process design decisions.

Original languageEnglish
Pages (from-to)303-319
Number of pages17
JournalMaterials and Manufacturing Processes
Volume24
Issue number3
DOIs
Publication statusPublished - Mar 2009

Fingerprint

Multiobjective optimization
Xylene
Oxidation
Acids
Xylenes
Vents
Evolutionary algorithms
Water content
Process design
Oxygen
Engineers
Catalysts
terephthalic acid
Temperature

Keywords

  • Evolutionary computation
  • Modeling and simulation
  • Multiobjective optimization
  • Optimization
  • Pareto front
  • Process design
  • Purified Terephthalic Acid (PTA)
  • Reaction engineering
  • Series reactions

ASJC Scopus subject areas

  • Mechanical Engineering
  • Mechanics of Materials
  • Industrial and Manufacturing Engineering
  • Materials Science(all)

Cite this

@article{a869ae5d5aef49379cf5c3962cc05d49,
title = "Improved multiobjective differential evolution (MODE) approach for purified terephthalic acid (PTA) oxidation process",
abstract = "Evolutionary algorithms (EAs), which are basically the population-based search algorithms, are becoming increasingly popular for solving highly nonlinear and complex optimization problems (both the single-objective and multiobjective problems). Multiobjective Differential Evolution (MODE), which is an extension of Differential Evolution (DE) to multiobjective optimization, has been successfully applied to a few chemical process industrial problems. This article presents multiobjective optimization of p-xylene oxidation in the manufacturing process of industrial Purified Terephthalic Acid (PTA) oxidation using MODE, its improved strategy, i.e., Elitist-Multiobjective Differential Evolution (E-MODE), and NSGA-II. E-MODE algorithm includes the concept of elitism as present in NSGA-II. Four industrially important cases are studied from the multiobjective optimization point of view. The effect of six decision variables identified, namely, catalyst concentration, vent oxygen content, {\%} water in solvent, feed xylene rate, temperature of reactor, and total feed rate, on the formulated objective function for the four cases is studied. The corresponding Pareto optimal sets (i.e., sets of equally good nondominated solutions) are obtained. All decision variables, corresponding to the Pareto optimal solutions, are plotted against each of the objectives. Comparison among the Pareto optimal fronts of four cases is also carried out. Pareto solutions obtained in this study (using MODE, E-MODE, and NSGA-II) lie on same (Global) Pareto front. However, MODE is able to cover a better range than E-MODE and NSGA-II. The results obtained are useful to a plant engineer for possible improvement in the process design decisions.",
keywords = "Evolutionary computation, Modeling and simulation, Multiobjective optimization, Optimization, Pareto front, Process design, Purified Terephthalic Acid (PTA), Reaction engineering, Series reactions",
author = "Gujarathi, {Ashish M.} and Babu, {B. V.}",
year = "2009",
month = "3",
doi = "10.1080/10426910802679337",
language = "English",
volume = "24",
pages = "303--319",
journal = "Materials and Manufacturing Processes",
issn = "1042-6914",
publisher = "Taylor and Francis Ltd.",
number = "3",

}

TY - JOUR

T1 - Improved multiobjective differential evolution (MODE) approach for purified terephthalic acid (PTA) oxidation process

AU - Gujarathi, Ashish M.

AU - Babu, B. V.

PY - 2009/3

Y1 - 2009/3

N2 - Evolutionary algorithms (EAs), which are basically the population-based search algorithms, are becoming increasingly popular for solving highly nonlinear and complex optimization problems (both the single-objective and multiobjective problems). Multiobjective Differential Evolution (MODE), which is an extension of Differential Evolution (DE) to multiobjective optimization, has been successfully applied to a few chemical process industrial problems. This article presents multiobjective optimization of p-xylene oxidation in the manufacturing process of industrial Purified Terephthalic Acid (PTA) oxidation using MODE, its improved strategy, i.e., Elitist-Multiobjective Differential Evolution (E-MODE), and NSGA-II. E-MODE algorithm includes the concept of elitism as present in NSGA-II. Four industrially important cases are studied from the multiobjective optimization point of view. The effect of six decision variables identified, namely, catalyst concentration, vent oxygen content, % water in solvent, feed xylene rate, temperature of reactor, and total feed rate, on the formulated objective function for the four cases is studied. The corresponding Pareto optimal sets (i.e., sets of equally good nondominated solutions) are obtained. All decision variables, corresponding to the Pareto optimal solutions, are plotted against each of the objectives. Comparison among the Pareto optimal fronts of four cases is also carried out. Pareto solutions obtained in this study (using MODE, E-MODE, and NSGA-II) lie on same (Global) Pareto front. However, MODE is able to cover a better range than E-MODE and NSGA-II. The results obtained are useful to a plant engineer for possible improvement in the process design decisions.

AB - Evolutionary algorithms (EAs), which are basically the population-based search algorithms, are becoming increasingly popular for solving highly nonlinear and complex optimization problems (both the single-objective and multiobjective problems). Multiobjective Differential Evolution (MODE), which is an extension of Differential Evolution (DE) to multiobjective optimization, has been successfully applied to a few chemical process industrial problems. This article presents multiobjective optimization of p-xylene oxidation in the manufacturing process of industrial Purified Terephthalic Acid (PTA) oxidation using MODE, its improved strategy, i.e., Elitist-Multiobjective Differential Evolution (E-MODE), and NSGA-II. E-MODE algorithm includes the concept of elitism as present in NSGA-II. Four industrially important cases are studied from the multiobjective optimization point of view. The effect of six decision variables identified, namely, catalyst concentration, vent oxygen content, % water in solvent, feed xylene rate, temperature of reactor, and total feed rate, on the formulated objective function for the four cases is studied. The corresponding Pareto optimal sets (i.e., sets of equally good nondominated solutions) are obtained. All decision variables, corresponding to the Pareto optimal solutions, are plotted against each of the objectives. Comparison among the Pareto optimal fronts of four cases is also carried out. Pareto solutions obtained in this study (using MODE, E-MODE, and NSGA-II) lie on same (Global) Pareto front. However, MODE is able to cover a better range than E-MODE and NSGA-II. The results obtained are useful to a plant engineer for possible improvement in the process design decisions.

KW - Evolutionary computation

KW - Modeling and simulation

KW - Multiobjective optimization

KW - Optimization

KW - Pareto front

KW - Process design

KW - Purified Terephthalic Acid (PTA)

KW - Reaction engineering

KW - Series reactions

UR - http://www.scopus.com/inward/record.url?scp=61449197826&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=61449197826&partnerID=8YFLogxK

U2 - 10.1080/10426910802679337

DO - 10.1080/10426910802679337

M3 - Article

VL - 24

SP - 303

EP - 319

JO - Materials and Manufacturing Processes

JF - Materials and Manufacturing Processes

SN - 1042-6914

IS - 3

ER -