00 mi piace00 non mi piace

2 visualizzazioni51 paginemodeling process

Dec 05, 2019

© © All Rights Reserved

PPTX, PDF, TXT o leggi online da Scribd

modeling process

© All Rights Reserved

2 visualizzazioni

00 mi piace00 non mi piace

modeling process

© All Rights Reserved

Sei sulla pagina 1di 51

Modeling

By Hagos, 2019

Why a model is needed?

about system behaviour

To back up financial or other

decisions

To optimize a new or existing

process

To operate efficiently and safely an

existing process

For illustration / teaching

Basic Concepts

Modeling Fundamentals

Models are an integral part of any kind of human activity. However,

we are mostly unaware of this.

Most models are qualitative in nature and are not formulated

explicitly.

Such models are not reproducible and cannot easily be verified or

proven to be false.

Models guide our activities, and throughout our entire life. We are

constantly modifying those models that affect our everyday

behaviour.

The most scientific and technically useful types of models are

expressed in mathematical terms.

Modeling in Chemical Engineering

use of dynamic models, as opposed to the more traditional use of

steady-state models for chemical plant analysis, is much more

recent.

This is reflected in the development of new powerful commercial

software packages for dynamic simulation,

Dynamic simulation can thus be seen to be an essential part of any

hazard or operability study, both in assessing the consequences of

plant failure and in the mitigation of possible effects.

…Cont’d

process operations, as in other inherently dynamic operations such as batch,

semi-batch and cyclic manufacturing processes.

Dynamic simulation also aids in a very positive sense in enabling a better

understanding of process performance and is a powerful tool for plant

optimization, both at the operational and at the design stage.

Large scale commercial software packages for chemical engineering dynamic

simulation are now very powerful and contain highly sophisticated

mathematical procedures, which can solve both for the initial steady-state

condition as well as for the following dynamic changes.

…cont’d

They also contain extensive standard model libraries and the means

of synthesizing a complete process model by combining standard

library models.

Other important aspects are the provision for external data

interfaces and built-in model identification and optimization routines,

together with access to a physical property data package.

The complexity of the software, however, is such that the packages

are often non-user friendly and the simplicity of the basic modeling

approach can be lost in the detail of the solution procedures.

…Cont’d

The correct use of such design software requires a basic

understanding of the sub-model blocks and hence of the

methodology of modeling.

Our simplified approach to dynamic modeling and simulation

incorporates no large model library, no attached database and no

relevant physical property package. Nevertheless quite realistic

process phenomena can be demonstrated, using a very simple

approach.

Again this follows our general philosophy of starting simple and

building in complications as the work and as a full understanding of

the process model progresses. This allows the use of models to be an

explicit integral part of all our work.

…Cont’d

Kapur (1988) has listed thirty-six characteristics or principles of

mathematical modeling. They can be summarized as follows:

processes, which are often extremely complex and often only

partially understood. Thus models are themselves neither good nor

bad but should satisfy a previously well defined aim.

2. Modeling is a process of continuous development, in which it is

generally advisable to start off with the simplest conceptual

representation of the process and to build in more and more

complexities, as the model develops.

…Cont’d

3. Modeling is an art but also a very important learning process. One of

the most important factors in modeling is to understand the basic

cause and effect sequence of individual processes.

4. Models must be both realistic and robust.

General Aspects of the Modeling

Approach

An essential stage in the development of any model is the formulation

of the appropriate mass and energy balance equations.

To these there must be added appropriate kinetic equations for rates

of chemical reaction, rates of heat and mass transfer and equations

representing system property changes, phase equilibrium, and applied

control.

The combination of these relationships provides a basis for the

quantitative description of the process and comprises the basic

mathematical model.

The resulting model can range from a simple case of relatively few

equations to models of great complexity.

…Cont’d

then the difficulty in identifying the increased number of parameter

values.

One of the skills of modeling is thus to derive the simplest possible

model, capable of a realistic representation of the process.

…Cont’d

the following advantages:

Models help in experimental design.

Models may be used predictively for design and control.

Models may be used in training and education.

Models may be used for process optimization.

General Modeling Procedure

One of the more important features of modeling is the frequent need

to reassess both the basic theory (physical model), and the

mathematical equations, representing the physical model

(mathematical model), in order to achieve agreement, between the

model prediction and actual process behaviour (experimental data).

The following stages in the modeling procedure can be identified:

1. The first involves the proper definition of the problem and hence

the goals and objectives of the study.

2. All the available knowledge concerning the understanding of the

problem must be assessed

…Cont’d

terms and the mathematical model solved by computer simulation.

4. The validity of the computer prediction must be checked.

5. The model may now be used at the defined depth of development for

design, control and for other purposes.

…cont’d

Figure1: Steps

in model

building

Model Classification

Some model types are inappropriate in certain circumstances, such as

a steady-state model for batch reactor startup analysis.

Table1: Model Classification

Characteristic Nature of Process

Models

Mechanistic models are also referred to as phenomenological models

because of their basic derivation from system phenomena or

mechanisms such as mass, heat and momentum transfer.

Many commonplace models in process engineering applications are

derived from a knowledge of the underlying mechanisms.

However, most mechanistic models also contain empirical parts such

as rate expressions or heat transfer relations.

Mechanistic models often appear in design and optimization

applications.

They can be termed "white box" models since the mechanisms are

evident in the model description.

…Cont’d

Empirical models are the result of experiment and observation,

usually not relying on the knowledge of the basic principles and

mechanisms which are present in the system being studied. They

employ essentially equation fitting where the parameters have little

or no physical meaning.

Empirical models are widely used where the actual underlying

phenomena are not known or understood well.

These models are often termed "black box" models, reflecting the

fact that little is known about the real mechanisms of the process.

The most common form of model used in process engineering is a

combination of mechanistic and empirical parts and hence is termed

"grey box."

…Cont’d

Stochastic models arise when the description may contain elements

which have natural random variations typically described by

probability distributions.

This characteristic is often associated with phenomena which are not

describable in terms of cause and effect but rather by probabilities

or likelihoods.

Deterministic models are the final type of models characterized by

clear cause-effect relationships. Thus we can have a mechanistic

model with some stochastic parts to it. A very common occurrence is

a mechanistic model which includes empirical aspects such as

reaction rate expressions or heat transfer relationships.

Model Characteristics

affect our modeling and analysis .

Models can be developed in hierarchies, where we ca n have several

models for different tasks or models with varying complexity in

terms of their structure and application area .

Models exist with relative precision, which affect how and where we

can use them .

Models cause us to think about our system and force us to consider

the key issues.

…Cont’d

investigations.

Models are develop ed at a cost in terms of money and effort. These

need to b e considered in any application.

Models are always imperfect. It was once said by George E. Box, a

well-known statistician, "All models are wrong, some are useful"!

…Cont’d

the model such as kinetic rate constants, heat transfer and mass

transfer coefficients.

Models can often be transferred from one discipline to another.

Models should display the principle of parsimony, displaying the

simplest form to achieve the desired modeling goal.

Models should be identifiable in terms of their internal parameters.

Models may often need simplification, or model order reduction to

become useful tools.

…Cont’d

Models can become intractable in terms of their numerical solution.

We ca n keep some of these in mind when we come to develop models

of our own for a particular application.

A Systematic Approach to Model

Building

The notion of a process system

If we want to understand the notion of a process system , we can

start from the general notion of a system. This can be defined in an

abstract sense in system theory.

A system is a part of the real world with well-defined physical

boundaries.

A system is influenced by its surroundings or environment via its

inputs and generates influences on its surroundings by its outputs

which occur through its boundary. This is seen in Fig. 2

system

…cont’d

A system is by nature a dynamic object.

The system inputs u and the system outputs y can b e single valued,

giving a single input, single output (SISO) system.

Alternately, the system can be a multiple input, multiple output

(MIMO) system. Both inputs and outputs are assumed to be time

dependent possibly vector-valued functions which we call signals.

A process system is a system in which physical and chemical

processes are taking place, these being the main interest to the

modeler.

The system to be modeled could be seen as the whole process plant,

its environment, part of the plant, an operating unit or an item of

equipment.

…cont’d

To define our system we need to specify its boundaries, its inputs and

outputs and the physio-chemical processes taking place within the

system.

Process systems are conventionally specified in terms of a flowsheet

which defines the boundaries together with inputs and outputs.

Information is normally available about the internal structure of the

system in terms of the operating units and their connections.

Process systems are conventionally specified in terms of a flowsheet

which defines the boundaries together with inputs and outputs.

Information is normally available ab out the internal structure of the

system in terms of the operating units and their connections.

The Modeling Goal

The modeling goal has a major impact on the level of detail and on the

mathematical form of the model which will b e built.

Analogous to the idea of a process system shown in Fig. 2, a model acts

in some way to mimic the behaviour of the real system it purports to

represent.

Thus, Fig. 3 shows the model with certain inputs and outputs. The use

of the model can take various forms dep ending up on what is assumed

to be known and what is to be computed.

…cont’d

…cont’d

process engineering are the following:

Dynamic simulation: With the process model developed to represent

changes in time, it is possible to predict the outputs o given all inputs

i, the model structure M and parameters p.

Static or steady-state simulation: Here, the process system is

assumed to be at steady state, representing an operating point of the

system. Again the simulation problem computes the output values o

given specific inputs i, a model structure M and its parameters p. This

is sometimes known as a "rating" problem.

…cont’d

certain parameters p from the set of parameters p, given known

inputs i and desired outputs o and a fixed structure M. This type of

problem is normally solved using an optimization technique which finds

the parameter values which generates the desired outputs. It is also

called a "specification" problem.

A Systematic Modeling Procedure

constructed following a well defined sequence of steps.

These steps are arranged in a "Seven (7) Step Modeling Procedure"

which is introduced below and shown schematically in Fig. 4.

…cont’d

…cont’d

However, it should b e noted that model development is inherently

iterative in its nature. One must usually return to and rep eat an earlier

step in case of any problems, unusual or unwanted developments later in

the process.

No one gets it right first time!

In fact, we never get a perfect model, just one that is usable.

Before starting to setup a process model the problem definition should

be clearly stated.

This defines the process, the modeling goal and the validation criteria.

This is part of the formal description in the SEVEN STEP MODELING

PROCEDURE, which is given in the form of an algorithmic problem.

…cont’d

specify the following items:

the inputs to the problem in the given section,

the desired output of the procedure in the Find or Compute section,

the method description in the Procedure or Solution section.

following algorithmic problem statement can be constructed.

…cont’d

Given:

a process system

a modeling goal

validation criteria

Procedure:

seven steps.

Seven Step Modeling Procedure

Seven Step Modeling Procedure

The steps of the modeling procedure are as follows:.

Define the problem

This step refines the sections already present in the problem

definition: the description of the process system with the modeling

goal. Moreover, it fixes

the degree of detail relevant to the modeling goal and specifies:

inputs and outputs, hierarchy level relevant to the model or hierarchy

levels of the model in the case of hierarchical models

the type of spatial distribution (distributed or lumped model)

the necessary range and accuracy of the model and

the time characteristics (static versus dynamic) of the process

model.

Identify the controlling factors or

mechanisms

The next step is to investigate the physico-chemical processes and

phenomena taking place in the system relevant to the modeling goal.

These are termed controlling factors or mechanisms. The most important

and common controlling factors include:

chemical reaction

diffusion of mass

conduction of heat

…cont’d

free convection heat transfer

radiation heat transfer

Evaporation

turbulent mixing

heat or mass transfer through a boundary layer

fluid flow

…cont’d

As we consider the system under study, we recognize the following

issues:

There is a set of all process characteristics which are never fully

identified.

In the s et of all process characteristics for the system we often:

only identify and include in the model a subset of the essential

characteristics.

This means that some essential characteristics for the application

can be missing from our model description; include non-essential

characteristics of the system in our models, which lead to

unnecessary complexity and/or model order (or size).

…cont’d

actually not part of the system and include them in our model.

The previous issues are often difficult to resolve and very dependent

on our understanding of the system and our insight into what is

important.

Model validation and the principle of parsimony, which seeks the

simplest representation for the task are the tools that need to be

employed to address these issues.

…cont’d

set of all possible controlling mechanisms taking into account the

following key elements in the problem definition:

the type of spatial distribution,

the necessary range and accuracy,

the time characteristics.

Example: Modeling example: CSTR

Develop a process model of the continuously stirred tank for dynamic

prediction and control purposes following the SEVEN STEP MODELING

PROCEDURE with all of its ingredients.

Problem definition

The process system to be modeled is a continuously stirred tank with

continuous fluid flow in and out and with a single first order chemical

reaction A → B taking place in an inert solvent.

The tank is adiabatic with its wall perfectly insulated from the environment.

The flowsheet schematic of this tank is shown in the Fig. 4. (This

description of the process system is exactly the same as the Process

system section in the Problem definition of previous example)

…cont’d

and energy states of the tank contents if the inlet concentration is

changed over a stated range. The accuracy of the predictions should

be ± 10% of the real process.

Controlling factors or mechanisms

chemical reaction

perfect mixing

Data for the problem

…cont’d

parameter type data:

reaction kinetic data, heat of reaction

physico-chemical properties

equipment parameters from the literature or given by the process

documentation.

…cont’d

Process model

Assumptions

A1 : perfect mixing,

A2 : constant physico-chemical properties

A3 : equal inflow and outflow (implying constant liquid volume with V =

constant),

A4 : single first-order exothermic reaction, A→ P,

A5 : adiabatic operation.

…cont’d

Model equations and characterizing variables

Differential (balance) equations in molar units

dmA

dt = fAi − fA − rV ……………………………………………………………………..(1)

V ρcp dT

dt = fcpi ρi (Ti − T) + rV(−∆HR)………………………………....(2)

Constitutive equations.

r = k0e −E /(RT)CA …………………………………………………………………….(3)

mA = CAV ……………………………………………………………………………………..(4)

fAi = fCAi ……………………………………………………………………………………….(5)

fA = fCA …………………………………………………………………………………………(6)

…cont’d

Variables Ti : inlet temperature [K]

t: time[s] ∆HR : heat of reaction [J/mol]

CA : concentration in the tank [mol E : activation energy [J/mol]

/m3]; k0 : pre-exponential factor [s −1]

V : liquid volume [m3]; R : universal gas constant, 8.314

f : volumetric flowrate [m3/s ]; [J/(molK)]

CAi : inlet concentration [mol/m3]; ρ : density of mixture [mol/m3]

r : reaction rate [mol/(s.m3)]; ρi : density of feed i [mol/m3]

cp : specific heat of mixture mA: moles of A [mol]

[J/(molK)]; fAi : inlet flowrate of species A

cpi : specific heat of feed i [mol/s]

[J/(molK)] ;

T : temperature in the tank [K];

…cont’d

Initial conditions:

CA(0) = CAi , T(0) = Ti

Boundary conditions:

none

Parameters:

Values for the following parameters with 10% precision:

V , f , CAi , Ti , cp , cpi , ρ, ρi and for the reaction parameters with 30%

precision: k0, E, ∆HR

Solution procedure:

Solve using an ODE or differential-algebraic equation solver.

…cont’d

Model verification

Implement model equations using structured programming principles

for every balance volume in the system.

Provide measured data from pilot plant or real process.

Analyze plant data quality.

Carry out validation of predicted outputs from step test of system

using least squares estimation of error.

Apply hypothesis testing to validate model based on least squares

estimates.

Refine model as required by performance criteria.

## Molto più che documenti.

Scopri tutto ciò che Scribd ha da offrire, inclusi libri e audiolibri dei maggiori editori.

Annulla in qualsiasi momento.