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

• Models for Optimization

• Formulation of the Objective
Function
Assumptions

• Basic Concepts of Optimization

Models for Optimization
 Models Classification

Principle based: Behavior based: Change based:

• Based on physical • Linear • Steady state

principles
• Non-linear • Unsteady state

• Empirical based

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

Parameter based: Variable based:

• Lumped • Continuous

• Distributed • Discrete

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

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Model Elements
• A Reactor
Exercise to relate the properties of a B

process system influenced by the process.

• Mathematical representation of properties A+ B →C

Assumptions

Elements of
Process Relations
Modeling

Variables &
Parameters

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Model Elements
Assumptions A Reactor

B
• Conditions which need to be satisfied without which model is
invalid

• If any assumption is not satisfied the process model needs to

A+ B →C

be redefined with new conditions

• It is used to simplify the process and remove the

complications

• It limits the scope and define the boundary conditions

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

Variable

property that could vary with time during the execution of a process

Parameter

property that is fixed or specified

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Regression and Types
Regression analysis is a set of statistical
processes that estimates the relationship Linear Regressiom

between a dependent variable & one or

more independent variable. Regression

Analysis
Regression analysis produces a regression
equation where the coefficients represent the Non-Linear
Regression

relationship between each independent

variable and the dependent variable.

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Regression and Types
Linear Regression
Linear regression requires a linear model. In statistics, a regression
equation (or function) is linear when it is linear in the parameters
Y = b o + b1X1 + b2X2 + ... + bkXk
2
Y = b o + b1X1 + b2X1

Non-linear Regression
If the equation doesn’t meet the criteria above for a linear equation,
it’s nonlinear.

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

It is the process used to determine if a material balance problem has sufficient specifications to
be solved.

Four Steps

a) draw and completely label the flowchart

b) count the unknown variables on the chart

c) count the independent equations relating these variables

d) calculate degrees of freedom by subtracting step (b) from step (c)

nDOF = nvariables - nindep_eqns

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

nDOF = nvariables - nindep_eqns

• If nDOF = 0, problem can be solved (in principle).

• If nDOF > 0, problem is underspecified and at least nDOF additional variables
must be specified before the remaining variable values can be determined.

• If nDOF < 0, the problem is over-specified with redundant and possibly

inconsistent relations.

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

Sources of equations relating unknown process stream variables

• Material balances

• Energy balance

• Process specifications

• Physical properties and laws

• Physical constraints

• Stoichiometric relations

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

• Equality Constraint

• Inequality Constraint

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Examples of Inequality Constraint

• Production limitation due to equipment size

• Raw material limitations

• Safety restrictions

• Physical property of chemicals

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
Formulation of the
Objective Function
 Objective Function

• Can be considered in terms of Cost

• Can be single or multiple

• Cost considers profit, processing, raw material

• Processing involves labor, maintenance, utilities etc.

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

Suppose you want to maintain the quality of a product in terms of two of its properties.

One property is the deviation of the variable yi (i designates the sample number) from
the setpoint for the variable, y,. The other property is the variability of yi from its mean y

(which during a transition may not be equal to y,). If you want to simultaneously use

both criteria, youcan minimize f:

where the wi are weighting factors to be selected by engineering judgment

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Time Value of Money

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Measures of Profitability
Payback period (PBP)

Return on investment (ROI)

Net present value (NPV) Size + profitabaility

Internal rate of return (IRR) profitability

• If the NPV is positive, the investment increases

the company's assets: The investment is • If the IRR is greater than the designated rate of
financially attractive. return, the investment is financially attractive.
• If the NPV is zero, the investment does not • If the IRR is equal to the designated rate of
change the value of the company’s assets: The return, the investment is marginal.
investment is neutral. • If the IRR is less than the designated rate of
• If the NPV is negative, the investment return, the investment is financially unattractive.
decreases the company's assets: The
investment is not financially attractive.

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

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
Basic Concepts of
Optimization
 Continuity

• Continuous functions are preferred than discontinuous functions.

• Function of a single variable x is continuous at a point xo if

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

Functions with discontinuities in the function or derivatives

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

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

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

Problem 4.1 is nonlinear if one or more of the functions f, g,, . . . , gm, are

nonlinear. It is unconstrained if there are no constraint functions gi, and no

bounds on the xi, and it is bound-constrained if only the xi are bounded. In

linearly constrained problems all constraint functions gi are linear, and the

objective f is nonlinear.

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

• vector x is feasible if it satisfies

all the constraints. The set of all

feasible points is called the

feasible region F

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Example 1

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Example 2

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

Convex Set
A set of points (or a region) is
defined as a convex set in n-
dimensional space if, for all pairs
of points x, and x, in the set, the
straight-line segment joining them
is also entirely in the set.

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

Convex Function

concave function is the negative of a convex function

Course:
Course: Modeling, Process Modeling
Simulation and Simulation
and Optimization | Lecture:
for Chemical Process Modeling
Engineering
 Why Convexity Desired
• Although convexity is desirable, many real-world problems turn out
to be nonconvex. In addition, there is no simple way to
demonstrate that a nonlinear problem is a convex problem for
all feasible points.

Why Convexity is desired:

many significant mathematical results have been derived in the field

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