Modeling, Simulation and Optimization

for Chemical Engineering

for

CH3133- CC01 and CC02

Ho Chi Minh City University of Technology
 Part III
Optimization of Chemical Processes
 Lecture Contents

• Multi-Objective Optimization
• Unconstrained Multivariable Optimization

Assumptions
 Multi-Objective Optimization

Course: Process Modeling and Simulation | Lecture: Process Modeling

 Motivation for Multiobjective Optimization

• With several goals in mind

• It is all about finding

compromises

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Conflicting Nature of Multiobjective Optimization

• Some needs to increase

• Some needs to

decrease

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Applications of MOO in Research

Paper Titles Authors

Halim,I., Srinvasen,R.
Integrated decision support system for waste
National University Singapore,
minimization analysis in chemical processes
2002
Mauricio M. D. & Karen, A.H.
Evaluation of waste minimization alternatives
Oklahama state university, USA
under uncertainty: a multi-objective approach
1999
Analyzing the interaction of design Motivation for
Luyben, M.L., Flounders, C.A.,
Multiobjective Optimization and control. A multi-
Computers and Chemikal
objective framework and application to binary
Engg., 18, 933 – 969, 1994
distillation synthesis
Cavin Laurent Swiss Federal
A systematic approach for multiobjective process
Institute of Technology Zurich.
design in multipurpose batch plants
2003

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Applications of MOO in Research

Paper Titles Authors

Assessment of chemical risk and safety by the use
Garribba, Ovi GASP, Italy, 1980
of multi-attribute decision techniques

Busacca, Marseguerra, Zio

Multi-objective optimization by genetic algorithms:
Dep. of Nuclear Eng.,
application to safety systems
CESNEF, Italy, 2001

Kim, Raymond, Smith U.S.

Systematic procedure for designing processes with
Environmental Protection
multiple environmental objectives
Agency, 2005

Multi-objective optimization for safety-related Kim, Yeo, Moon Yonsei

decision making in chemical processes University Korea 2004

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Multi-Objective Optimization (MOO) Problems

• Involve more than one objective function that are to be minimized or

maximized

• Answer is set of solutions that define the best tradeoff between

competing objectives

• There is no unique solution to Multi-objective Optimization problems

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Mathematical Definition of MOO

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Dominance

• In the single-objective optimization problem, the superiority of a solution

over other solutions is easily determined by comparing their objective

function values

• In multi-objective optimization problem, the goodness of a solution is

determined by the dominance

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Definition of Dominance

• x1 dominates x2, if

• Solution x1 is no worse than x2 in all objectives

• Solution x1 is strictly better than x2 in at least one objective

• x1 dominates x2 x2 is dominated by x1

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Example Dominance Test

• 1 Vs 2: 1 dominates 2

• 1 Vs 5: 5 dominates 1

• 1 Vs 4: Neither solution

dominates

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Pareto Optimal Solution

Non-dominated solution set

Given a set of solutions, the non-dominated solution set is a set of all the solutions that are not

dominated by any member of the solution set

• The non-dominated set of the entire feasible decision space is called the Pareto-optimal set

• The boundary defined by the set of all point mapped from the Pareto optimal set is called the

Pareto optimal front

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Graphical Depiction of Pareto Optimal Solution

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Goals in MOO

• Find set of solutions as close as

possible to Pareto-optimal front

• To find a set of solutions as

diverse as possible

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Different Pareto Optimal Trade-off Surfaces

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Methods for MOO

Two important methods are:

• Weighted sum method

• ε-Constraint Method

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Weighted sum method

• Scalarize a set of objectives into a

single objective by adding each

objective pre-multiplied by a user

supplied weight

• Weight of an objective is chosen in

proportion to the relative importance

of the objective

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Weighted sum method

Convex Non-Convex

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Weighted sum method

Advantage

• Simple

Disadvantage

• It is difficult to set the weight vectors to obtain a Pareto-optimal solution in a desired region

in the objective space

• It cannot find certain Pareto-optimal solutions in the case of a nonconvex objective space

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 ε-Constraint Method

• Keep just one of the

objective and restricting

the rest of the objectives

within user-specific values

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 ε-Constraint Method

• Keep f2 as an objective;

minimize f2(x)

• Treat f1 as a constraint;

f1(x) ≤ ε1

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 ε-Constraint Method

Advantage

• Applicable to either convex or non-convex problems

Disadvantage

• The εvector has to be chosen carefully so that it is within the minimum or maximum

values of the individual objective function

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Unconstrained Multivariable
Optimization

Course: Process Modeling and Simulation | Lecture: Process Modeling

 Problem & General Sequence

Problem

Sequence

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Methods for Search Direction

Direct Methods

• Use function values

Indirect Methods

• Can use derivative values

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Direct Methods for Search Direction

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Direct Methods for Search Direction

Random Search Method

• A random search method simply selects a starting vector xo, evaluates f(x) at xo , and

then randomly selects another vector x1 and evaluates f(x) at x1. In effect, both a

search direction and step length are chosen simultaneously. After one or more

stages, the value of f (xk) is compared with the best previous value of f(x) from among

the previous stages, and the decision is made to continue or terminate the procedure.

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Direct Methods for Search Direction
Grid Search Method

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Direct Methods for Search Direction
Univariate Search Method
• This method searches one dimension at a time thus optimizing only a single variable per iteration.

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Direct Methods for Search Direction

Simplex Search Method

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 In Direct Methods for Search Direction

• Steepest Decent Method using gradient

• conjugate gradient method

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 In Direct Methods for Search Direction
• Steepest Descent

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering