Вы находитесь на странице: 1из 75

1 Optimization of the Flash Carbonization Process

A Thesis Presented to the Faculty of the Engineering College of Ohio University

In Partial Fulf illsnent of the Requirement for the Degree

Master of Science

BY

Yeong-Siarg Chang, August 1984

-

-

1

ACKNOWLEDGEMENTS

I would especially like to thank Dr. Wen-Jia Chen, my thesis advisor, who suggested this topic and gave generously of his tirne and guidance during the course of this study.

I am grateful to Dr. Robert L. Savage for his constructive criticisms and useful suggestions to this thesis. Ny thanks are also due Dr. J. R. Collier and Dr. N.

Dinos for their help during my stay in Ohio University.

My wife, Huoy-Jen, has offered me her understanding and

I am more than

support over

the duration

of this task.

grateful for that. Finally,

parents for their encouragement and

I

wish to express

a deep gratitude

to my

support which make this

studjr possible.

This work is dedicated to them.

Pee

TABLE OF CONTENTS

v

LIST OF FIGURES

vi

LIST OF TABLES

vii

1.0

INTRODUCTION

 

01

1.1 Coal Gasif'ication and the Flash Carbonization

 

Process

01

 

1.2 Econorni cs of

the Flash Carbonization Process

05

1.3 Purpose of the Thesis

09

2.0

LIrnTJRE &'VIEW

 

10

2.1 Gasification Reaction

10

2.2 Equilibrium Computation

12

2.3 Overview of Optirmun Seeking Methods

15

2.4 Overview of Experimental Designs

18

3.0

OBJECTIVE FUNCTION

 

26

3.1

Assumptions

26

3.2

Data and Parameters

28

3.3

Flnal Form of Objective Function

29

4.0

OUTLINE OF OPTIMIZATION STKAmY

32

4.1 A Sample Calculation of Response Value

32

4.2

The Considerations of Optimization Strategy

43

v

Ascent

44

 

4.4 Second Order Design

46

5.0

REsULTS AND DISCUSSIONS

48

6.0

CONCLUSIONS

64

APPENDIX I

65

REFFWSNCES

66

LIST OF FIGURES

Figure 3-1 The Block Diagram of a Single-stage Gasifier

30

vii

LIST OF TABLES

Table 2-1 Heat of Reactions(HR : Kcal) and Equilibrium

 

Constants(KP) of Selected Reactions

23

Table 2-2

Yatels Algorithm

25

Table 3-1-A Unit Cost of Feedstock and Utilities

31

Table 3-1-B Unit Varket Price of Products

31

Table 4-1 Composition of Coal(Ohio Clarion 4A)

37

Table 4-2 Calculation of Heat of Formation of Coal Table 4-3 The Equilibrium Composition and Moles of Fach

38

 

Component

39

Table 4-4 Calculation of Heat of Reaction at 1200K

40

Table 4-5 Total Cost in the Sample Calculation

41

Table 4-6 Total Credit in the Sample Calculation

42

Table 5-1 Search on Stage-1

53

Table 5-2 Search on Stage-2

55

Table 5-3 Search on Stage-3

57

Table 5-4 Search on Stage-4

59

Table 5-5 Search on Stage-5

Table 5-6 Search on Stage-6

61

63

Zxtensive

processes for

1.1 Coal Gasification

programs have

been

undertaken to

develop

synthetic fuels

the comnercial production of

from

coal,

oil

shale,

tar

sands,

or biomass.(l)

The

objective is

to replace

exhausted or

costly supplies

of

natural gas and petroleum-based

a flexible

Products can be varied to include low-, medium-, high-Btu

gas,

methanol, and petrochemlcal products. It is also less

and more operable on lower

quality coals than coal liquefaction.

Many gasification processes have been developed with

differences in mdes of operation and characteris tics of

the products produced. Each gasifier has been described for

fuels.

Coal gasification is

of synthetic

method for

the production

fuels.

and raw materials for liquid fuels,

such as gasoline,

costly for chemical mufacture,

a specific application

and the

type of

coal avaiable

as

feedstock.

The three

min

characteristics of

gasifiers

sur,mrized by

Probstein and

Hicks(2),

are

the method

of

supplying the heat,

reactor type.

characteristics such as the state

the gasifying medium and amount, and the

Once they are specified, the other dependent

of the solid residue (dry

or

slagging),

the

properties of

the product

gas,

and the

gasification temperature

will be

fixed.

There

are three

types

The reactor type heavily

dominates the temperature distribution, therefore, those

and

the entrained flow rzactor.

of

reactors:

the

moving bed,

the

fluidized bed,

dependent

characteristics

are

all

influenced.

There

are

also

two

nethods

of

supplying

the

heat:

direct

and

indirect.

 

The

direct

method

is

to

supply

oxygen

or air,

where

heat

is

generated

from

the

combustion

reactions

of

coal.

The

indirect

method

applies

an

external

heating

source,

steam

or

electricity,

 

to

supply

heat

for

gasification

reactions.

The possible

gasifying

media

for

gasification

are

oxygen,

air,

hydrogen,

steam.

Oxygen

and

air

are

the

sources

of

oxygen

in

the

reactions

for

the

production

of

carbon monoxide.

Hydrogen

gas

and

steam are

the

source

of

hydrogen

for

the

production

of

methane

and

hydrogen respectively.

It

is .also

obvious

that

the

anount

and

the

kind

of

heating

medium

or

gasifying medium

influence

the

three

dependent

characteristics.

In

this

section,

a

brief

description

of

the

characteristics

of

the

major

single-reactor, direct-heat gasifiers is given.

The dry-ash and slagging-ash Lurgi processes(3,4) are

moving bed processes and primarily to be considered for the

production of Synthetic Natural Gas(SNG). Moving bed

gasifiers

operate

with

countercurrent

flow.

The

coal

is

introduced

downward

through

the

upward

flowing

gases.

It

is

dried

first,

then

devolatilized,

and

then

gasified

at

the

lower

section.

The

bottom

section

is

the

combustion

zone

where

the

renaining

coal

is

burnt

to

supply

heat

for the

3

gasification zone. The sla~ing-ashprocess is better than the dry-ash process both in the tt- roughp put and in thennal

efficiency because of higher temperature operation. By

reducing the steam injection

investment is reduced, and the process them1 efficiency is

methane

cantent yield and leads to a significant overall process

economy because, in most cases, the gas will be processed

further at elevated pressure. The disadvantages is that it

requires a sized noncaking coal which increases the cost of

coal. The reactor with a slowly

basically a low-throughput device that requires a large

number of gasifiers occupying a

the

gasification process for

downward moving bed is

requirements, the capital

increased. The pressurized gas FTier favors high

large area.

However,

Lurgi dry ash

process is the only

which the

technology has

been commercialized.

'The Sasol

projects have applied this process

to produce

Synthetic

Nqtural Gas.

The Texaco and Shell-Koppers ( 5,6,7) processes

mnufacture either low- or medium-Btu synthesis gas in a

manner that is closely similiar to the production of

synthesis gas from petroleum fuel residues. The 'Texaco

water slurry

injection system produces

hydrogen to

and solves the problem of

a high

carbon monoxide ratio product gas

feeding coal into a pressurized gasifier, but introducing a

therml inefficiency in operation. The potential for the

considerable . Tennessee

application of this process is

%stman Co. has applied it to the production of acetic

4

anhydride from co-al(8).

developed from Kopper-Totzek process, uses the ,Ininimum

mount of stem for high thermal efficiency and produces a

low hydrogen to carbon monoxide ratio. Both gasifiers are

entrained bed processes in high temperature operation,

The

Shell-Kogpers

process,

therefore, both process units require a high-temperature

heat exchanger which represents a costly technical problem.

A waste

energy in the fonn of stean(Mangold, 1982, p.122), which can

be utilized in a refinery or chemical plant, but my not be

used in other applications.

All the gasification processes discussed so far aim to

gasify completely the coal fed into the gasif'ier. Though it

a

difficult or

consideration. In

practice, it is impossible to convert the last five to ten

is possible to

coal

thermodynamic point of

expensive to do it from a kinetics

heat boiler

my give about

half of

the available

convert

totally

to

gas

frorn

view,

it is extremely

percent of coal to gas even under the rnost.favorable kinetic

condition.

coal gasification process, if both the gas and the solid

products can be utilized.

Savage and Chen(9) are actively developing the Flash

Carbonization process at Ohio University for the concurrent

production of synthesis

char which

along the Ohio River.

plant use

and a low sulf ur, low volatile

Therefore,

it

is preferable to use

a partial

gas

is suitable for

industrial and povJer

Using an entrained bed reactor, oxygen and steam, along

5

with finely-&rounded coal(-40+100, -100+200, ?ksh No.), are introduced from the top of the reactor. A char with forty

to eighty percent of the original carbon, together with a

produced at the botton of the reactor.

Coal, oxygen, and steam are fed at rates of 2.27Kg per hour, 5000c.c per ninute, and 2c. c. per minute respectively.

Atmospheric pressure and temperatures ranging from 1180~to

1450K were chosen for the operating conditions in a series

of experiments. Also, the residence times were varied.

In general, being an entrained bed system, the process

gasification

processes (lo), minly ;

has several advantages over the other

systhesis gas is

(a)

the ability

to handle

caking coal

and low

grade

coal, and

(b)

the product gas is free of

tars and Phenols.

 

More importantly, the moderate temperature (1100K-1500K)

employed

avoids the

-

aisadvantages

( >1500K) process such as (lo) :

of

a high

temperature

(a)

the large mounts of energy(oxygen) required to

maintain a high temperature condition,

(b)

the high cost in refractories and construction

material necessary in the combustion zone,

(c) the large munt

of

energy loss in

product gas or

high cost in heat recovery system.

1.2 Econmics of

the Flash Carbonization Process

The rigorous economic analysis of a chemical process

and total

includes detailed

narket research,

captial cost

production cost estimation,

and other economic factors.(ll)

At

the present

stage

economic

evaluation for

the

Flash

Carbonization

process

is

difficult

because

the

new

technology

involves

rnany

uncertainties

in

process

performance, operability, an reliability . A rigorous estimtion is, theref ore, not necessary. Alternatively, a

quick calculation always meets the need in priliminary design stages. Savage(9) has shown the predesign capital

and operating cost of the Flash Carbonization process, based

He also compared

the production cost at the experimental operatiw point with

other coal gasification processes, such as, Texaco process

and Koppers-Totzek process. The next step, followed by his work, in economic

analysis is to find the best operating point in the process

itself. In general, the best or optim operating point

on the assumgtion of equilibrium

yields.

implies that it will yield a maxlmum profit. Net profit, by definition, equals total income minus all expenses. Total

income is the sum of each product amount multiplied by its selling price. All expenses are the total production costs.

A typical total production cost analysis(l2) contains several items, direct production cost, fixed charges, plant

overhead costs, administrative expenses, and distribution

7

and marketing expenses. The first three items are sometimes referred as manufacturing cost. The last two items are so called general expenses. The first item is the dominating

factor in evaluating the optirm operating point. list of direct production cost contains(Peters,

A typical

1968

p.192)

raw materials

utilities cost(steam, electricity, fuel,

ref rigeration,

water,

etc. )

operating labor and supervision

maintenance and repairs

operating supplies

laboratory charges

catalysts and solvents

For a given process or a plant, raw materials and utilities

cost are the rmst influential factors in direct production

cost. Therefore, the profit model can be simplified and

related to the following factors:

(a)

the amount and the price of each product.

(b)

the munt and the cost of each raw material.

(c)

utilities cost.

These three factors can be estimated if the material and

energy balance are known. In general, a kinetic model

should be used to predict the material and energy balance at

different operating conditions. However, for high

temperature processes,

4t the

present stage, no kinetic model for the Flash Carbonization

thermodynamic equilibrium model

produces as good a prediction as the kinetic model.

the

8

is available. We will use a thermodynamic equilibrium mdel

for the profit calculations. Any profit model for a chemical process is naturslly constrained by chemical and

physical principles and economics rules. Chemical and

physical contraints

therrmdynamics laws. Gcononic constraints are the rules of

existing mrket syste~m. The search for the rmxirnum prof it

will give us an explanation of how rnuch the constraints

influence t'ne prof it and operating point.

The characteristics of partial gasification in Flash

Carbonization has raised a series of questions. Since the

process is not energy self-sufficient, which way is more

economical in supplying reaction heat-by electricity

(indirect method) or by oxygen (direct method) - is unknown.

operation favors

decreases energy

as heat

The

and

consist of mss conservation and

low temperature

char production

as well

required in

the reactor

waste in the product stream;

it also decreases the

production of syngas because of themdynamic equilibria in

the system. From the economical point of view, what is the

Flash Carbonization for

naxFinum prof it

optimal

in the

operating

point

is not clear.

1.3 Purpose of the Thesis

the

optimum operating point of the Flash Carbonization process.

Constrained by both thermodynamic and

considemtions, an objective function has been defined for the profit of each operating poigt. 4n experimental design

economic

This

study

is

to

find

a

auick estimation

for

is

implementation of

discussion on the search path is given with the cornbination

of chedcal stoichiometry, equilibrium constant, and reaction heats involved in the system. The result, at the

the

experimental points using the same profit model.

optirm operating point, is

used

as

an

optinim seeking optimization

method

in

the

since

real

it

is

an

A

world.

cmpared

with

those of

2 .o LITERATURE m1w

2.1 Gasif ication Reaction

The principal chemical and physical changes of coal in

a gasifier can be described by four categories: drying, pyrolysis, combustion, and gasification. Each physical or

chemical transformation can be silnply represented by the

following

equations : (bhngold,

1982, p. 132)

Drying

Coal (high moisture) ---- > Coal (dry)

Pyrolysis

Coal (dry) -----> Char + Volatiles

(CO,

H2

CH4

, Tar,

H

2

Combustion Combustible volatiles

(CO,

S,

Ha

0,

etc.)

H2

, CH4

, Tar)

-----> co2

+ 5 0

, C02

(2.2)

11

Gasif'ication

Char

+ 5 0 ---- > co + 5

+ H

2

s + N

2

c~sh(2.5)

Char + C02

----->2CO+H20+% +H2S+

 
 

N2

+

AS^

(2.6)

Char + 2H2

-----> CB4 + H2

3 + H2 S + N2

+

 

Ash

(2.7)

apply the present data, the

main reactions in a gasifier are simplified and summarized

in Table 2-1. The overall reaction can be represented by the

following equation:

In order

to discuss and

aCoal + b02

+ cH

2

0 =====

dCO + eH2

+ fH2

0

hCH4

+ iH2

S + jN2

+ kChar + etc.

+ gC02

+

(2.9)

Reactions involved in a reactor are generally defined and constrained by stoichiometry, themdynamic equilibrium, kinetics, and transport rates (~nass,energy, and mentun).

Since this is

a general analysis of the operating point in

any gasif'ier

with

syngas

production,

the discussion

A

concentrates

on

stoichiometry

and

equilibrium.

stoichiornetric analysis of coal gasification has been made

by

Wei(l3).

Through

detailed

vectorial

and

gemetric

12

explanations and comparisons of

pilot plant

data,

Wei

found a

the result to comercial and

operating

narrow feasible

regian which is only limited

stoichiometric constraints. However, the basis of his studies is on the complete gasification and thermal balance

by them1 balance and

which are different fron the assumption of this study.

2.2 Equilibrium Computation

model for

coal

factors (such as the mechanism of pyrolysis, gasification reaction, hydrodynamics, and the difference in coals) involved in the reactions are not cmpletely understood. As

an alternative, themdynamic analysis is always a guide in

the preliminary design work. The information about chemical equilibnun composition of a reaction allows us to estimate the theoretical mass and energy balance for the system. The

calculation can be applied to the design and analysis of the process. l%ny examples were presented in Shewood's

text (14). In a complex

equilibrium computation may provide information about the upper boundary of yield for the first step in assessing a

synfuel technology. Batchelder and Sternberg(l5) had a discussion of equilibrium composition for suspension

gasification of pulverized coal. Recently, Wiser and

At the

present time,

there

is no unifying

effects

gasification

kinetics

because the

of

~nany

reaction such as coal gasification,

13

Kithany ( 16) investigated a new catalytic hydrogenation of

coal slurry- prepared in s hydrogen-donor solvent, with

steam and hydrogen gas. The potential application and

operating point of the process was found by estimating the

equilibrium composition and heating value of product gas.

An entrainment gasifier is always operated under high

tanperature conditions. Therefore, the overall performance in these gasifiers can be determined approximately by

equilibrium considerations. Furtherimre, for an idealized

or a large reactor, the residence tirne of coal particles is

supposed to be long enough to reach equilibrium condition.

In a chemical system, equilibrium constraint includes

stoichiometric constraints and mass conservation constraints

for each principle reactions, the calculated equilibrium

cmposition fulfills both constraints. Two categories for the computation(l7) are the

equilibrium constant method and the free energy minimization

method. The former aethod uses equilibrium constants to

express certain species in terms of a set of arbitrarily

calculation

chosen species . Kandiner and Brinkley ( 18) had a

of the combustion of propane in air by this method. The

equilibrium system contained ten gaseous constituents, with

or without the formation of solid carbon. This method was

some

designed for a specific problem

advantage of special characteristics of the particular

problem. So,

which satisfy the mass balance, the total presssure

and

often

took

it is necessary to find those compositions

spcifications, and all the simultaneous equilibria

involved. It seems tedious for a rnoderateljr complex system.

An alternative method, free energy minimization method, based on the fact that the total Gibbs free energy of the

system reaches its minimum value at guilibriurn, subject to

the constraints of the material balance. The necessary

data, chanical potential of each species, can be calculated

from spectroscopic constants by

evaluating the

canonical

partition function of

statistical thermodynamics.

Oliver,

Stephanou

equilibrium

moles of

oxygen with 1 rnole of methane at 873K by this method. They also cited some additional application, such as, calculation of the adiabatic flame temperature at a sgecified pressure

distribution of

and

Baier ( 19)

computed

the

species

resulting frorn reacti-ng 5

and rocket motor performance calculations.

developed shce

the 1950's for multipurpose application. Rased on the latter

A NASA

cmputer progrm(20)

has been

method,

for

themdynamic parameters.

show the properties of the system, the (H,P) problem gives

adiabatic constant pressure combustion properties, the (U,V) problem gives adiabatic constant volume combustion properties, the rocket problen uses (T,P), or (H,P), or

where

S=E;ntropy,

T=Temperature, V=Volu-;le, P=Pressure, Y=Enthalpy,

U=Internal Energy.

described by two

the program calculates

thermodynamic

the equilibrium canposition

is

any

state which

The two thernodynanic paraneters

(S,P),

the detonation problem uses

(H,P)

or (T,P),

15

Savage

and

Chen

have

applied

it

to

calculate

equilibrium yields for the comparison with the experimental data. They found the experimental yields and the

canpositions of synthesis (H2 and CO) gas approached the

results calculated from the NASA program.

2.3 Overview of 3ptir;wn Seeking iflethods

Optimization

 

is the

way

of

finding the

~naxirnum or

i-illnim

values

of

an objective

function.

In chemical

engineering , optimization plays an important role for process evaluation, either economically or technically. For

example, such economic improvement as minim costs or

maxim profit is required for a process design, and the

possible technical aims might contain the maximum amounts of

yield from a reactor or a minimum size of a cooling tower.

In spite of various kinds of mathematical rnodels of

objectives, the

Which

be

applied to a specific objective function with high

for an

efficiency and accuracy is still an

basic optirm

what particular

seeking methods

group of

are fixed.

method or

methods can

active field

engineer.

Many texts(21,22,23) presented the information of

optimization techniques for most engineering problem. In

general, optimization techniques can be classified into two

broad categories:

analytical methods and numerical methods.

16

These are applied to two different types of objective mdels

respectively. (Beveridge 1977,

p. 26)

The first type

is

a

mathematical model which is a set of analytical expressions.

The second type is the so-called black-box model in which

the response to a particular input is detemined by

numerical computation, m experiment, or a computer program.

Sometimes a inathematical mdel is too coinplex to find an

optimum with analytic methods, then, it might be solved ~ith

nunerical methods.(Bveridge

1977, p.53)

optimization

techniques .

optimum seekirg method in the unconstrained multivariable

problem. The review here will be a basis for the selection

of an optimization technique in this study. There are three

methods summrized from most texts.

only concerned with the possible

We

are

not

We

going to

are

review

all

the

(a) univariate search

(b)

steepest ascent and

(c)

simplex method

(a) univariate search

is

sometLzes,

p.355-363) By

the k variables

fixed at some level, a mximm or minirum value can be found

starting an initial point and keeping k-1 of

or sectioning method1.

The

Univariate search

a

kind of

direct

method;

it is also called 'one

factor at a time methodf

(Beveridge 1977,

along this

variable search is necessary for

dimension.

Therefore,

an

efficient

single-

The

this type of search.

17

the

objective function is again optimized with another variable.

The process continues :mtil the successive change of

variables and object value is less than a tolerance.(24) 'The

optiml point Is

substituted

into

the function

and

disadvantage of univariant search is that it is difficult to

be used in a system containning a ridge or steep contours.

The step size must not be kept too large because the process

my stop at a n~noptimlpoint.

(b) steepest ascent

-

It is a kind of gradient method. (kveridge, 1977 p.407)

The gradient vector is nor,ml to the contour line or surface

and indicates the direction

The steps for this method(Stocker, 1980 p.180) are as

follows :

of steepest ascent(or descent).

(a)

Select a trial point.

 

(b)

Evaluate the gradient at the current point and the

relationship of the changes of the x variable.

(c)

Decide step size and then move that distance.

(d)

Determine the

maximum or

minimurn point along the

direction.

(e)

Check whether

the optim has been achieved. If

not,

return to step(b) .

There are

depend on the chamcterization of the system.

many variations of

step(c) and step(d), which

18

(c) simplex ;nethod

variables,

(k+l) points are necessary to form a simplex. For example,

-

If

we have

an objective

function

with k

a

dimensions

simplex in

is

two dimensions

a

tetrahedron.

is

a triangle

The

and in

three

of

general direction

search

my be

taken in a direction away from the worst

poht.

A new

point is then selected

along this direction

and passes through the center of gravity of the remaining

points. (Beveridge, 1977 p. 367) [The search will stop on a

The step size

search from there until the

region,

where no knprovenent

can be achieved.

shall be decreased to start t'ne

desired accuracy is reached for optirnurn.

2.4

berview of Experimental Designs

The purpose of any experimental work is to understand

mona bout the system being investigated. Experinental

designs have been introduced to provide the least number of

experimental trials. In section(2.2) , only the response

values without experimental error have been considered. kJe

have to use a method derived from experimental designs that

is applicable when experimental error is significant. In

this study, we are considering an effective experimental

method of fitting response surface and of locating an optimum operating point. The basic experimental designs are

indicated in many texts(25,26,27) and reviewed in this

1Y

section as a basis

of solution algorithm for this study.

Factorial Designs at two levels A general factorial design is

number of levels for each of a number of

possible

two level factorial desigs are

combinations.

varlables(factors)

the selection of a fixed

--

and

the experiments

with all

In general,

more important by the following reasons: (Pox, 1978 p.306)

(a)

They require relatively few runs per' factor

studied.

(b)

\*en needed, they are easily augmnted to form a

composite desig.

(c)

Through two level f ractional factorial design, the

number of runs can be decreased further.

(d)

The use of building blocks reduces the degree of

complexity of the problem.

(e)

It is easy to interpret the observations.

In general, the two level factorial design gives a

first order equation to represent local surface. There are

many texts(Davis, 1967 p.271; kx, 1978 p.510; ax 1954)

comparing the two level factorial design with the one factor

at a time nethod. In general, the differences between them

are:

(a) When interaction effects are significant,

a

factorial design avoid leading wrong conclusions. For exanple, one factor at a time will be valueless when the response surf ace contains a ridge, but

20

factorial

desim

my

identify it

and

give

the

 

direction of the axis of

the

ridge

so

that

improvement is possible.

(b)

The discovery of factor dependence of a particular

type provides

the information

in connection

with

the experimenter's

theoretical knowledge.

 

It is

helpful for futher experimentation.

 

(c)

For multivariables experimental

design,

factorial

design

tells exactly

which

factor

and how

.my

factors should be varied.

 

Fractional factorial design

 

When a model

contains more than three

variables,

the

full factorial designs are tedious and unmanageable.

Fractional factorial design is needed for fewer design

points and enough informtion about the nature of the

response function we are exploring. There are many examples

presented in the well hewn textbooks (Davis, 1976 p. 454;

Orthogonal design

If

we arrange

the levels of each factor in

such a way

that

the

diagonal

be

terms in

the

would be

performing least square)

said

to

orthogonal.

The

normal

vanished,

equations

matrix

(when

the design is

and

the

design

Composite desi~

Box and Wilson(24) originally introduced the concept in

1951.

Factorial design can be

augmented to form composite

design.

Therefore,

fitting a response surface is possible

with a second order equation. Many examples are presented in the paper. (28,29)

Response

has been selected with

success on locating the optim point through experinent

design. Two survey articles, Hill and Hunter(30),

and Nead and

PFke(31), listed references to those studies.

studied the rapid entrainment

carbonization of powdered coal under pressure in a partial

Surf ace Method

The response surface method(24)

Belt

and

Roder(32)

hydrogen atmosphere for

They established the relationship

the production of

low

sulfur char.

betrreen process variables

md char

yield as

well as

quality by

the application

of

response surface

analysis.

It

is

clear that in

the real

world the exact form of a response function would be

unknown. Also, the complete theoretical mchanisms of nost industrial processes are not available. In fact, the exact

function is not necessary because the mediate concerns for process design are questions such as(25)

(a) What values of a given set of inputs will yield a

22

maximn or reach a maximurn profit?

(b) What is the shape of

the response surface close to

this maximum,

or over

some specified

regions of

interest?

The response surface nethod has been applied to answer these

questions. One strikiw application of

is 'Evolutionary 0peration1(EVOP).(33,34) The basic idea is

the response surface method

la process should be

run so

as

to

generate

product

plus

inf'ormation

on

how

to

improve

the

product'. (34)

The

experiments for true

the full-scale

plant.

optimum yields nust be

carried out on

TABLE 2-1

HEAT OF REACTIONS ( kB : KCAL ) AND EQUILIBRIUJ'JI CONSTANTS ( KP )

OF SELFXTED @ACTIONS

3eactions

298K

700K

lOOOK

1500K

Combustion

 

I-tR

-26.4

-26.4

-26.74

-27 75

Continued

24

Gasification

Gas Eieactions

---> H3+C02-

KP

HR

CO + 3H2

--->CH +H 0 KP

4

2

0.07

0.15

0.20

0.28

-49.26

-52.68

-53 87

-54 59

14.83

1.15

-0.g6

-1-75

Source from Reference (1)

Unit

Temperature

(K)

- - - - - - --- ---- -

Oxygen

(wt%>

Stearn

(wt%)

 

T

0

H

Center Condition

 

1180

0.20127

0.09526

Step Size

20

0.01

0.002

 

+

1200

0.21127

0.09726

-

1160

0.19127

0.09326

TOH

Y

(1)

(2)

(3)

divisor

effect

---

96.00

90.36

186.09 372.46

8

46.56 ave

+

-

-

44.36

95.73

186.37

-6.55

4

-1.64

T

-

+

-

48.68

90.50

-3.27

10.74

4

2.69

O

min effects: T, 0, H two factor interaction: TO, THY HO three factor interaction: T9H

One.

3.0

A rigorous profit

OBJECTIVE FUNCTION

3.1

Assumptions

mdel has Seen discussed

in chapter

A simplified one contains the following factors :

(a)

the aiiunt and price of each product.

(b)

the amunt and cost of

raw mterials.

(c)

utilities cost.

Once the amunt and cost of raw materials and the price of

each

balances for each unit process and unit operation give us

rigorous mss and energy

product

are

specified,

the mount of each product and the utilities cost. Here, we will also apply the simplified profit model for the Flash Carbonization process. The necessary mass and enerw

balances

for

the

objective

function

of

the

Flash

Carbonization

process

have

been based

on

the

following

assumptions :

process

(a) Only a single-stage gasifier is in the system.

(b )

The operating cost of transmission machines (screw feeder, pump, compressor) is negligible.

27

(d)

No steam recovery system.

(e)

Electricity is the only indirect heating source for

coal gasification in endothemic condition.

gasifier

(f ) Operated in

isothermal condition at

1 atmospheric

pressure.

The

block diagram

of

it

is show

in

Figure 3-1.

(g) The products are under thermodynamic equilibrim

the

data bank of the

condition.

'The

NASA

program

In the

provided

equilibrium calculation.

NASA program,

Therefore,

Graphite carbon(C).

there is no molecular fom for char.

is necessary

to replace

char with

The element form of coal,

i.e.

it

Cv Hw

0,

Ny

S,

2.11.

aC

H

0

N

S,

vwxy

fH2

0 + gC02

,

also replaces coal in Reaction

+ b02

+ cH2 0 ======

S

+ jN2

dCO + eH2

+ etc.

+ hCH4

+ iH2

+ kc

+

(h) applying Hessls law for the

(i)applying the ideal

calculation of heat of

reaction at the reactiorl temperature.

calculation of

sensible heat content for each conponent of product

gas law for the

gas.

3.2 Data and Parameters

The necessary data and parameters are discussed in this section. Since cost and price are dependent on time, we

arbitrarily referred them from the following reference.

--- cost and price (j) The unit cost of

electricity

shown in Table 3-1-A.

and raw mterials are

(k) The unit price of

in Table 3-1-B.

char and medium-Btu

gas are shown

(1) No

separation cost for char and product gas.

(m) The price

of

product

should be

varied with

the

composition of

it.

However,

the credit of product

gas

here is

only

calculated

based on

the

heat

content of

it.

physical and chemical data

-

-

 

(n) Heat

capacity

equations

and

heat

capacity

coefficients are f ran the data bank prpgrun.

of the

NPSA

(0) Heats of Formation at 298K are shown in Table 2-1.

(p) Composition of Reference (9).

coal(0hio Clarion 4A)

is

shown in

29

3.3 Final Form of Objective Furction

After

making

function will be

the above

assumptions,

the

objective

Ul

y

=

,U2

z

y:

=

f

(xl

9

+Z2U2

X2

-

, X3)

(cl

u

11

+c2x2 + c3x3

prof it, response value

+ E)

f:

u1,u2

NASA program

: equilibrium mount for char and

product gas

c

,c

,c

123

:unit cost of

coal,

oxygen,

: amount of oxygen and steam

'2jX3

x 1 : temperature

stem

z19z2 : unit price of char and medium-Btu gas

E:

cost of

electricity

HEAT

4

I

I

\t/

YIGURE: 3-1 THE BLOCK DIAGRAM OF A SINGLE-STAGE GASIFIEB

TABLE 3-1-A

UNIT COST OF FEEDSMCK AND UTILITIFS

Item

Feedstock

Coal(0hio Clarion 4A) (Dry and Ash Free) Oxygen

Unit cost

$15.00 per ton

$26.90 per ton

Stem

$3.50

per Lvl

lb

Utility

Electric Power

Source from Reference(9).

$0.035 per IQih

UNIT !'hWKST PRICE OF PRODUCTS

Item

Unit price

 

Char

$14.00

per ton

%

Medim-BTU

gas

$5.30

per MM Btu

*

%Sourcefrom Reference(9). *Source from Reference(37).

4.0 OUTLINE OF THE OPTIMIZATION STHAmY

4.1 R SAMPLE CALCULATION OF RESPONSE VALUE

based on the objective

ble

arbitrarily select 1 ton of

to calculate the necessary operating cost, the possible

credit, and the profit-response value.

Heat of formation of coal

function in Chapter Three is

dry and ash free coal as a basis

A calculation of response value

shown

in this

section.

--

7-

(a) Molecular formula of

coal

de derive the molecular formula from the element

composition table of coal.(Table 4-1) The

molecular form is:

(b) Heating value of

coal

Frotn Dulongls

formula(35),

the

heating value

of

coal can be expressed by:

 

Q

(Btu/lb)

=

14544

* C

+

62028 *

(H - 0/8)

+ 4050

* S

 

where C,

H, 0,

and S are weight fraction of

each

element .

Combining the

formula and the

elenent composition

33

table (Table 4-1) , the heating value of coal (Ohio

Clarion 4A) is 12880 Btu per lb,

which is eqlal to

124.8 Kcal per gmle dry and ash free coal.

(c) Heat of formation of coal at 298K

In order to calculate the heat of formation of

coal,

we have to apply

the combustion reaction of

coal.

Standard heat of

Higher heating

Combustion per

=

value of

1 ple

ple of coal

of coal

The sum of heats The sum of heats

= of formation of products at 298K

-

of

formation of

reactants at 298K

the last equation is the

Standard heat of formation of coal. Heating value

has been estimted in Page 32. Standard heats of

The only unkno\m value in

formtion of other component can be found in many

texts. Table 4-2 gives a

Therefore, the standard heat of formation of coal

is -2.08 (-126.88-(-124.8) ) Kcal/gmle at 298K.

detail

calculation.

-- The NASA program

(a)

Input data

 

We

arbitrarily

assume

the flow

rates

of

coal,

oxygen, steam so that the weight ratio of oxygen to

coal

is

0.21127 and the

weight

ratio of

stem to

coal is 0.09726. Therefore, the input data for the

NASA program in this sample calculation are:

 
 

Molecular formula of coal, oxygen, and stem Pressure = 1 atm

Temperature = 1200K

 

The amount of

coal

=

1

The amount of

oxygen = 0.21127

 

The amount of steam = 0.09726

(b)

Output result

 

The

output

result

for

calculated

equilibrium

component distribution is listed in Table 4-3. In order to convert mle fraction into mole basis of

each

moles of

component,

we

have to

calculate the

Here,

total

we use the

the equilibrium mixture.

element balance of carbon to get the total moles of

carbon atorn in the system and the total mole fractions of carbon compounds in the mixture. Therefore,

The total moles of

carbon atom

= 1000000 (kg coal) /

= 65972 ple

The

15.15

total

mole fractions

(g / gmle of coal)

of

carbon

compounds

Heat of

--

(Table 4-3).

= 0.429

(for C) + 0.003

(for CH4 )

+ 0.219

(for CO) + 0.00036

(for COS )

+ 0.002

= 0.653

(for C02 ) + 0.000014

(for CS2 )

The tot a1 moles of

equilibrium mixtures

-

-

The total moles of

The total mole fractions of

carbon atom

carbon compounds

Once the total

calculated, the moles of

estirrated and the result is indicated in Table 4-3.

moles of

equilibrium mixtures

each

component can

is

be

reaction

at 1200K

--

Heat of

reaction =

at 1200K

Each item in

is calculated in the Table 4-4.

Heat of

reaction -

at 298"

Sensible

heat of

react ants

the right hand side

The heat of

reaction at 1200K

 

Sensible

+

heat of

products

of

this equation

Therefore,

It

= 350000 Ycal/ton

coal

is

an

endot-hermic

calculation.

reaction

in

this

sample

Cost,

-

fm

Note sensible heat is :

that

the

formula

to

calculate

+A5 * (~5- 2985 )/5 where Cp = heat capacity

R = gas constant

!I1, A;,

A3,

A4,

A5 = parameters of

each

canponent are from data bank of

the NASA

program.

Credit,

and Profit

-

Total credit

is indicated in Table 4-6. The profit, response value is equal to total credit .minus total cost

for one ton dry

which is $88.03 -

Total cost is indicated in Table 4-5.

$40.19

(= $47.12)

and ash free coal in this sample calculation.

TABLE 4-1

COMPOSITION OF COL (OHIO CLARION 4A)

Component

Weight%

Molecular

Mole%

Mole%

 

Weight

set C=l

C

70.3

12

5-86

1

H

5.3

1