AN ADSORPTION SYSTEM FOR THE REMOVAL OF SULFUR OXIDES

by

GERALD JOHN FAUST, B.S.

A THESIS

IN

CHEMICAL ENGINEERING

Submitted to the Graduate School of Texas Tech University in

Partial Fulfillment of the Requirements for

the Degree of

MASTER OF SCIENCE

IN

CHEMICAL ENGINEERING

Approved

Accepted

August, 1977

^ei- IM/

C/^p >//Z^

ACKNOWLEDGMENTS

The author wishes to express his appreciation for the guidance

and counsel given by Dr. R. W. Tock. The author would also like to

express his special appreciation to the other committee members.

Dr. L. D. Clements and Dr. R. M. Bethea, for their individual assis­

tance.

The author is also indebted to the following companies:

Gulf Oil Corporation and Phillips Petroleum for their financial

assistance; Cosden Oil and the Linde division of Union Carbide for

supplying high sulfur fuel oil and molecular sieves, respectively,

for use in this research program; and to the U. S. Coast Guard for

the use of its incinerator.

n

TABLE OF CONTENTS

ACKNOWLEDGMENTS ii

ABSTRACT iv

LIST OF TABLES v

LIST OF FIGURES vi

Chapter

I. INTRODUCTION 1

II. THEORY OF ADSORPTION 6

III. THEORETICAL DEVELOPMENT OF EQUATIONS TO MODEL A

MOVING ADSORPTION BED 9

IV. SOLUTION OF ADSORPTION BED MODEL 12

V. EQUIPMENT AND PROCEDURES 22

VI. DISCUSSION OF RESULTS 26

VII. CONCLUSIONS 37

VIII. RECOMMENDATIONS 38

BIBLIOGRAPHY 39

NOMENCLATURE 42

APPENDIX 44

A. COMPUTER PROGRAM 45

B. ANALYTICAL PROCEDURES 52

C. SCALE-UP CALCULATIONS 55

m

ABSTRACT

The purpose of this study was to determine the SOp adsorption

efficiencies of a uniquely designed, panel bed adsorption system.

A vertical panel bed adsorber was attached to an incinerator fired

with high sulfur fuel oil. The efficiency of sulfur dioxide removal

from the combustion gases was then determined.

Two adsorbents were investigated, Barnebey Cheney type CH

activated carbon in 6 x 10 mesh size and Linde^^type AW-500 molecular

sieve in 1/8 inch cylindrical pellets with a pore diameter of 5A.

The data obtained from operation of the panel bed adsorber, using

these adsorbents, were then used to determine the characteristic

constants for our mathematical model of the adsorption system.

These data points were sufficient to yield an equation suitable

for scale-up purposes with regard to the two adsorbents tested.

IV

LIST OF TABLES

Table

1. Preliminary Data 27

2. Operating Data for Activated Carbon 28

3. Removal Efficiencies for Activated Carbon 29

4. Operating Data for Molecular Sieve 30

5. Removal Efficiencies for Molecular Sieve 31

6. Model Results for Activated Carbon 33

7. Calculated Error Results From Model 35

LIST OF FIGURES

Figure

1. Panel Bed Design 4

2. Differential Element of Bed 13

3. Gridwork for Numerical Computation IS

4. Computer Algorithm 20

5. Experimental System 23

VI

CHAPTER I

INTRODUCTION

During recent years energy supplies have become a major con­

cern to the United States. Coal is the largest and most readily

available supply of domestic energy. Therefore, it is reasonable to

assume that coal will provide an increasing share of the energy used

in the United States during the next two to three decades. Should

this be the case, then several major environmental problems associated

with coal usage must be overcome. One of these problems, and prob­

ably the most difficult to solve economically and effectively, is

sulfur dioxide emissions produced in the combustion of sulfur-bearing

coal.

If coal is to be used, some method of solving the sulfur

dioxide emission problem must be found. There are two main ways

of solving the problem. One, known as coal desulfurization,

attempts to remove the sulfur from the coal before it is burned.

The second procedure involves removal of the sulfur oxides from the

flue gases produced in the coal burning process. The purpose of

this study was to evaluate a dry adsorption system using activated

carbon and molecular sieves for flue gas cleaning. Coal desulfur­

ization will not be discussed.

Flue gas cleaning methods can be classified into two broad

catagories: (1) Throwaway processes exemplified by lime, limestone,

1

or double-alkali scrubbing operations in which the sulfur is dis­

carded as calcium salts. (2) Recovery processes, in which the

sulfur is recovered as sulfur dioxide, sulfuric acid, ammonium

sulfate, or elemental sulfur.^ ' The throwaway processes are

currently in greatest usage by industry. There can be serious

problems with these processes, however. Some of the main ones are:

(1) boiler fouling, traceable to limestone injection, (2) scrubber

scaling and corrosion from salt build up and acid solutions, (3)

erosion of the scrubber, (4) mist carryover, (5) excessive costs for

reheating the stack gases, and (6) disposal of solid wastes without

polluting natural water supplies. If one considers the amount of

waste solids which would be generated by these methods if coal were

to be used on a larger scale than at present, it is believed that

throwaway processes are not the best solution to the problem. This

is probably true from an economic as well as from an environmental

point of view.

Recovery processes eliminate the sludge disposal problem and

theoretically are capable of producing a marketable product. While

this technology is presently less well developed, these processes

possess the potential to solve the problem of sulfur dioxide

emissions with minimum impact on the environment.

One form of recovery process is dry adsorption. The advantages

of systems of this sort are: (1) no stack gas reheat, (2) no water

consumption, and (3) increased reliability (no scaling or plugging

problems). Dry adsorption systems also possess the advantage of

being able to choose the form in which the sulfur dioxide is recovered,

This is possible due to the fact that the product is determined by

the method of regeneration of the adsorbent. This enables one to

choose either sulfuric acid, sulfur dioxide, or elemental sulfur as

a recoverable product.

This Study

In this thesis a specific dry adsorption recovery system was

studied to determine its overall operating and adsorption character­

istics. This system consisted of a thin vertical, moving bed of

adsorbent. Such an arrangement has been called a panel bed filter,

(see Figure 1). Note the louvers on the down stream side. Their

configuration helps in retaining the solid in the bed. They achieve

this by making it necessary for the solid to be lifted out of the

louver instead of just being pushed off the edge as would happen

if the down stream louvers were the same configuration as the up

stream louvers.

The particular advantages of this design, in addition to those

already listed for recovery systems in general, are: (1) reduced

power requirements due to the low pressure drop resulting from the

thin bed design (2.54 cm), (2) the potential to also remove nitrogen

oxides created during combustion, and (3) filtration of particulate

or fly ash.

Using granular material in a louvered system to clean flue

gas is not a new concept. It has been in use since 1883 when Rew

used lime in trays to purify flue gases.^ ' Traditionally, research

T 1.3

0.75

2.2

1

Solid Flow

y"

/

0.8'

w 1.0

Gas Flow

y

/

2.54 *-1.8-^

0.9

Note all dimensions in cm, and figure is exact scale.

Figure 1. Panel Bed Design, modeled after a similar design proposed

by Squires (22) (23)

in this area of granular filters has been directed to the removal of

fly ash. Bench scale removal efficiencies of 99.9+% for coal fired

power plant fly ash has been reported by Squires using a panel bed

(22) filled with sand.^ ' The removal of gaseous pollutants by a panel

bed filter has also been demonstrated using activated carbon as the

adsorbent.^ ^ ^ ^ It has yet to be demonstrated that simultaneous

removal of particulates and gaseous pollutants can be effectively

achieved.

The panel bed filter used in this study is modeled after a

(22) similar design proposed by Squires.^ ' This study was done as a

first step in evaluating the effectiveness of the system for simul­

taneous filtration and adsorption. The objective of this study was

to evaluate the effectiveness of different adsorbents for SO^ removal

and to model the operating characteristics of the bed for SO2 removal.

In our experiments the panel bed was attached to a slip stream

from a fuel oil fired incinerator. This unit was operated with high

sulfur oil. The amount of sulfur dioxide produced during combustion

and adsorbed by the bed was measured for different operating condi­

tions. The data collected were then used in conjunction with a

computerized mathematical model to simulate the adsorption character­

istics of the bed.

CHAPTER II

THEORY OF ADSORPTION

When a gas or vapor is brought into contact with a solid

substance it has a tendency to collect on the surface of the solid.

This phenomenon is the result of various intermolecular forces of

attraction and is referred to as adsorption.

At ambient conditions, the amount of adsorption which occurs

on the surface of most solids is exceedingly small. Certain materials,

however, have an unusually high surface development in the form of a

microporous structure, thus possessing a wery large internal surface

area. This microporous structure allows a gaseous fluid to diffuse

into and through the solid. This enables the solid to adsorb greater

amounts of the gas on its much larger internal surface area. Ex­

amples of materials that possess this type of structure are acti­

vated carbon, silica gel, activated alumina, and molecular sieves.

The absorption that occurs on the internal surface area is

wery complex. Many mechanisms have been hypothesized to explain

this phenomena, which is generally referred to as physical adsorption

and/or chemical adsorption. In physical adsorption the forces of

attraction are of the van der Waal's type. In chemical adsorption,

the gases show a much stronger interaction with the surface, similar

to a chemical reaction.

All adsorption processes are exothermic. Chemical adsorption

is characterized by a larger liberation of heat than is physical

adsorption. For chemisorption, the heat liberated is similar in

magnitude to the heat of a chemical reaction. For gases the heat of

(5) physical adsorption is greater than the heat of condensation. '

With gaseous materials an increase in partial pressure pro­

duces a correspondingly greater amount of adsorption. As adsorption

progresses and the partial pressure is increased, adsorption goes

through three phases: (1) surface attraction, (2) wetting of the

smallest capillaries, and (3) wetting of the larger capillaries.

The relationship between pressure and the amount adsorbed is, there­

fore, a function of capillary size distribution, internal surface

(5) area, and the nature of adsorbent and gas.^ ' These relationships

are usually expressed at equilibrium conditions in the form of

adsorption isotherms. Such isotherms are found experimentally and

express the amount adsorbed as a function of vapor pressure at

constant temperature.

To recover sorptive capacity once a solid has been saturated

with a gas, regeneration or desorption is required. This can be

accomplished in a number of ways:

1. Raising the temperature of the solid. This is effective

since most gases exhibit decreased adsorption at elevated temperatures

This method could be used to produce a concentrated stream of sulfur

dioxide in a flue gas adsorption system.

8

2. Lowering the partial pressure of the adsorbate below its

equilibrium vapor pressure. This procedure would not be effective

for removal of sulfur dioxide adsorbed from a flue gas because it

would not serve to concentrate the sulfur dioxide.

3. By stripping with an inert gas. This method would also

not be applicable to a flue gas adsorption system because the

sulfur dioxide produced would not be concentrated.

4. By stripping with an easily condensed vapor such as steam.

By using this method it is possible to produce H^SO^ when stripping

SO2 from the adsorbent.

5. By displacement with a gas which is preferentially

adsorbed. This procedure would render the adsorbent useless for

future use in adsorbing the gas that is displaced.

CHAPTER III

THEORETICAL DEVELOPMENT OF EQUATIONS TO MODEL A

MOVING ADSORPTION BED

The main function of adsorption equipment is to bring the gas

and solid adsorbent into contact to facilitate adsorption. The

equipment to do this along with the adsorbent itself is typically

called an adsorption bed. Factors to be considered in modeling an

adsorption bed include the following: .(5)

(1

(2

(3

(4

(5

(6

(7

(8

(9

(10

(11

The particle size of the adsorbent.

The depth of the adsorbent through which the gas flows.

The gas velocity through the bed.

The solid adsorbent velocity through the bed. (For moving

beds only)

The temperature of both gas and adsorbent.

The pressure of the system.

The concentration of the adsorbate.

The concentration of other gas constituents which may be

adsorbed at the same time.

The concentration of gas constituents which may react

with the adsorbent.

The adsorptive capacity of the adsorbent for the adsorbate.

The efficiency of adsorbate removal required.

10

Four equations are required to model an adsorption system.

One describes the concentration of adsorbate in the bulk voids of

the system. The second describes the mass transfer of adsorbate

across a film surrounding the particles of adsorbent. The third

models the diffusion process within the pores of the particle, and

the fourth describes the adsorption phenomena on the internal surface

(21) f24) area of the particles. ^ ' '

The equation for the bulk concentration will be derived in the

next section. The equation describing the mass transfer across the

film surrounding the particles takes the form of a boundary condi-

t1on.('^)

Diffusion inside the par t i c le voids can be described by Fick's

laws. For spherical par t ic les the fol lowing mathematical formulation

. ^ (21) IS appropriate.

D {-^ + 2/r — ) = p ^ P 9^ ^ 8r

The surface adsorption rate can be modeled by the following

. (21) equation.^

(t)z = a ("sat - ") ^'^

In this equation w . refers to the adsorbate loading when the o CI w

carbon is saturated. This equation is based on the assumption that

the adsorption is first order and directly proportional to the degree

of saturation.

11

If the equilibrium relationship at a temperature is known,

equations (1) through (3) along with the mass balance on adsorbate

can be used to model the proposed adsorption system. The equations

must first be solved, however. If a solution is found it will be

limited to the boundary conditions chosen; i.e. to a specific

temperature and inlet concentration. This is because the equilibrium

relation will be valid only at constant temperature and constant

partial pressure of the adsorbate. For multi-component systems,

such a solution assumes that there is no competition for adsorption

from other constituents in the gas phase.

CHAPTER IV

SOLUTION OF ADSORPTION BED MODEL

The common method of solution for an adsorption system is to

assume that one of the adsorption steps outlined in the theoretical

section will control the overall rate of adsorption. This simplifies

the system so that it can be solved by either analytical or numerical

techniques. The method of solution employed in this model assumes

that the surface adsorption rate controls the overall adsorption,

i.e. diffusional processes are assumed to be very rapid.

In developing a model for this moving adsorption bed it is

necessary to first visualize a differential volume. Figure 2 shows

such a fixed elemental volume. The element is stationary, while

the adsorbent flows through the element in the x direction and the

gas flows through the element in the z direction.

The assumptions used in deriving this model are as follows:

(1) Isothermal operation.

(2) Isobaric operation.

(3) Rod-like flow of the adsorbent particles. This means that

a particle that enters the bed with a certain set of y

and z coordinates retains these coordinates as it flows

through the bed.

12

13

Figure 2. Differential Element of Bed.

14

(4) Diffusion in the z and x directions is negligible, i.e.

bulk flow controls.

(5) Adsorption does not vary in the y direction.

(6) Heat of adsorption has negligible effect.

One of the necessary equations in this model is given by the

overall mass balance. This can be developed by considering an in­

ventory of the adsorbate over the differential element shown in Fig­

ure 2. Adsorbate enters this element in three forms. One form is

as a gas which enters by bulk flow. The other two forms enter with

the solid. One of these is adsorbate that has been adsorbed on the

solid surface. This surface can be considered as the internal

surface of the particle, since the external surface area is small

compared to the internal surface area. The other form that the ad­

sorbate enters is as an unadsorbed gas in the internal voids of the

particle. One of the assumptions in deriving this model is that the

adsorption rate on the internal surface area controls the overall

rate. This assumption means that the concentration in the internal

particle voids should be essentially the same as the concentration

in the bulk gas phase. Therefore the adsorbate in the internal voids

will be included in the term for the concentration in the bulk gas

phase. An inventory of the adsorbate over the differential element

is as follows:

,adsorbate into element by bulk. ,adsorbate out of element by. ^ ^flow of gas ^ " ^bulk flow of gas ^

/adsorbate into element adsorbed. _ /adsorbate out of elementx ^on solid ' ' ^adsorbed on solid

15

/accumulation of adsorbate in. /accumulation of adsorbatex 'gas phase ) + (on solid )

In mathematical terms:

adsorbate on solid in = p v (Ay) (AZ) W

adsorbate on solid out = p v (Ay) (AZ) (W + AX ^ ) U o oX

(|^) (Ay) (AX) (AZ) P^ accumulation of ad­sorbate on solid = (^

adsorbate in by bulk flow = (Ay) (AX) C V

y

adsorbate out by bulk flow = (Ay) (AX) (C V + V^ I ^ A Z )

9 g o^

accumulation of ad­sorbate in gas phase = z^ (Ay) (AX) (AZ) |^ Combining and canceling terms gives:

n V -^ w e IC ^ •^b s 3x • ^g b dz

3w , ac /.x

Assuming quasi, steady-state further simplifies the equation:

^b ^s 9x g b 8z

As mentioned in the theoretical section, any number of solutions

are possible based on variations in temperature and concentration.

Thus, there are not sufficient initial or boundary conditions to per­

mit a direct solution of the general differential equations as they

were derived. Specifically, a new equilibrium relationship would

need to be determined for each temperature and concentration. Be­

cause of these limitations, it was decided to model our system

using an adaptation of an equation developed by the Westvaco

16 (2)

Corporation. Their equation was developed for the adsorption of

sulfur dioxide on activated carbon and includes terms to take into

account the effect of oxygen and water vapor content on the ad­

sorption behavior of sulfur dioxide. In this study, the oxygen and

water vapor content are nonvariant. Therefore, these terms were

included in the constant coefficient preceeding the equation. The

adaptation of the Westvaco approach is represented as follows:

aw at = A e /" (1 - W / W ^ ^ ^ ) ( C ) " (6)

A, B, and n are constants used to fit the data obtained from operation

of the adsorption bed.

On inspection of equation (6) it can be seen that it is

actually a modification of equation (3) which takes into account

temperature and concentration effects.

It is assumed that V and V are constant for any adsorption 5 g

run. This assumption combined with the rodlike flow assumption leads

to the following: ^ = 1/V ^ (7) ax ^ s at ' '

Combining equations (6) and (7) leads to the following dif­

ferential equation: aw _ A e^/^ (1 - w/w^ .) (c)" (8) ax V3

Combining equation (8) with equation (5) gives the overall

differential equation that can be used to model this adsorption

system.

17

-P, A e^/T (1 - w/w^^^) { c ) " - V g e , | | = 0 (9)

A numerical integration method using finite - difference

approximations was applied to obtain the solution of this model.

The technique is the same as used in Lapidus (page 138). The method

consists of writing the partial derivatives of interest in the form

of forward differences.

(I^^z =Tr(^(r + l,s) - ^(r,s)) ( ^

'"ai x ^ m ^(r,s + 1) ' ^(r,s)^ ' ^

r and s are used to locate the point of interest in the two

dimensional integration grid as shown in Figure 3. h and m are

the integration step sizes in the x and z directions respectively.

The errors introduced by using these equations to approximate a

continuous process are of the order of h and m respectively.

Substitution of equations (10) and (11) into equations (5)

and (8) yields:

hA B/T /, ' (r,s)x , xn . , (.^y. "(r+l.s) = v ; ^ (1 ' ^ ^ (" Cr.s)) ^"(r.s) ^'^^

' (r.s + I f - VT-FT^ (r + l,s) (r,s) (r.sj

To clarify the method of computation, the following computer

algorithm is presented:

1. The initial conditions specify the values at "n^sj and

"(r.l)-

18

J L ^ r. s + 1

r + 1, s

m r + 1, s + 1

Figure 3. Gridwork for Numerical Computation,

19

2. From the initial conditions of c, ,, and w,, , all (i^J) (1,1)

^(r,l) ^ ^ calculated using equation (12).

3. Using the w^^^^j calculated in 2, all c^^ 2) ^ ^ calculated

using equation (13).

4. Using the Cj^^2) calculated in 3 and equation (12), all

^(r,2) ^ ^ calculated.

5. Steps 3 and 4 are repeated, incrementing s until the end

of the bed is reached.

6. The c, . exiting the bed are averaged to get the outlet

concentration.

For clarification of this algorithm, see Figure 4.

The A, B and n used in equation (12) were determined by using

a simplex evolutionary operation optimization procedure.^ ' This

was done by using the data obtained from the operation of the ad­

sorption bed to calculate the outlet concentration from the bed,

using the algorithm in Figure 4. This outlet concentration was then

compared to the actual outlet concentration. The difference between

the two was used as a desirability coefficient for that data set.

The simplex procedure is basically just a technique for

determining the best set of values for A, B and n in the fastest

possible manner. The best set will be the set that most closely

represents the data. The first step in the simplex technique is to

choose a starting simplex of four sets of values for A, B and n and

to calculate the desirability coefficient for each set. The starting

simplex was chosen so that the values of A, B and n used by Westvaco

Read c

w (l,s)

(r,l)

s = 0

s = s + 1 <-

I r = 0

r = r + 1

20

w (r + 1, s) V h A ^B/T ( _ ^^^p:^ (

w sat (r,s)'"+"(r,s)

= r >^ no end.

yes

= '"-^ (V + l.s)-V,s))^ V,s) (r,s + 1) V

'out

1 ""( 'Sg )

end I END

Figure 4. Computer Algorithm,

21

were included within its range. Next, the set of A, B and n with

the lowest desirability coefficient is found. This set of A, B and

n is removed from the simplex and a new set is calculated to take

its place. The new values for A, B and n are found by taking the

average of the remaining points in the simplex, multiplying by

two and subtracting the values for A, B and n that were removed from

the simplex. These new values for A, B and n are then put into the

model to determine a desirability coefficient for them. This pro­

cedure is repeated until the set of A, B and n that best models the

adsorption system is determined. The computer program used for these

calculations is presented in Appendix A. Note that one set of values

for A, B and n was used to model all the adsorption runs for a

particular adsorbent, and that the overall desirability coefficient

for a set of A, B and n was determined by finding the lowest

desirability coefficient among the individual adsorption runs.

CHAPTER V

EQUIPMENT AND PROCEDURES

The experimental equipment for this study was set up as shown

in Figure 5. The incinerator was a Consumat^-^Model c-18 that was

borrowed from the U. S. Coast Guard. A light cycle oil containing

1.65% sulfur obtained from Cosden Oil Company was burned in this

incinerator to give a base sulfur dioxide loading of approximately

70 ppm.

The blower was used to draw a side stream from the incinerator

stack, and force it through the adsorption bed. The adsorbent bed

design is shown in Figure 1. The overall dimensions of this bed

were length = 32.4 cm, height = 29.6 cm, thickness =2.5 cm. The

bed was made of stainless steel.

Adsorbent was loaded into the solid hopper above the bed and

was collected in a hopper at the bottom of the bed. Adsorbent was

removed from this hopper by gravity. A flap covered the opening

at the bottom of the hopper. It was used to control the solid

flow from the hopper, and thereby control the flow from the bed.

This flap system did not allow good control of the adsorbent flow

rate. The system was too sensitive. Minor adjustments with the flap

made tremendous differences in the solid flow rates. Adsorbent that

fell from the flap was collected in sheets that were laid out below

22

23

o to

o

o

Q.

I 4-)

n3

c:

E 'I—

S -<u Q .

ID

<U

13 cn

O rtJ

CO

o (T3 s -(U c o c

24

the bed. This system of a hopper at both top and bottom was used to

seal the bed from gas leaks. There was enough adsorbent in the

hoppers at both top and bottom to prevent gas from leaking around the

bed.

Two different adsorbents were used in this study, Barnebey

Cheney type CH activated carbon in 6 x 10 particle size and Linde

molecular sieve type AW-500 in 1/8 inch cylinders with a 58 pore

diameter.

In Figure 5, the position of the sample ports should be noted.

These allowed sampling of the gas up stream and down stream of the

adsorbent bed. Sampling of the gas was done to determine the sulfur

dioxide loading. The analysis was done by the Modified West and

(14) Gaeke Method.^ ' For a complete description on this method see

Appendix B. Since the gas flows were in the turbulent region center

line samples were taken. The procedure for a run consisted of the

following:

1. Start the incinerator and let the system warm up for at

least ten minutes.

2. Start the blower to warm-up the side-stream equipment.

3. Turn off the blower and fill the bed with adsorbent.

4. Start the blower and set the sulfur dioxide rate, using

the SOp cylinder shown in Figure 5. Note: this was not done for

all runs.

5. Open the bottom of the bed by adjustment of the flap to

permit adsorbent flow.

25

6. Allow time for the system to come to steady-state, (see

below).

7. Turn on the flowmeter that has been placed in the stack

slip-stream and start the timer.

8. Change solid collection sheets.

9. Sample the gas up-stream and down-stream of the bed.

10. Record the pressure drop across the bed, as well as the

temperature up-stream and down-stream of the bed.

11. After shutdown record the weight of adsorbent used and the

gas phase sulfur dioxide concentrations as determined by the syringe

technique of Meador and Bethea.^^^^

In step 6, the time required to reach equilibrium was determined

from preliminary runs of the adsorbent system. Multiple measurements

of SO2 concentration were taken in these runs to determine when the

SO2 levels stabilized, this usually occurred after 10 minutes of

operation. Temperature also usually stabilized after 10 minutes.

Because of these observations 10 minutes were allowed for steady-

state conditions to be reached in all subsequent runs.

In measuring the SO2 concentrations two samples on the gas

entering the bed and two samples on the gas leaving were taken for

each run. Considerable variability occurred between the two measure­

ments for each. The maximum up-stream variability was approximately

10 ppm while the maximum down-stream difference was approximately

2 ppm.

The data obtained from the runs made in this study are presented

in the next section.

CHAPTER VI

DISCUSSION OF RESULTS

Table 1 shows the preliminary data obtained for this adsorption

system. The great difference in adsorptive capacity for SO^ between

activated carbon and molecular sieve should be noted.

The results of operating the adsorption system with activated

carbon for different temperatures, solid velocities, gas velocities

and SO2 loadings are presented in Tables 2 and 3. Tables 4 and 5

show the results for the molecular sieve runs. No control was

exercised over temperature or gas velocity. Solid velocities were

controlled by the flap system mentioned in the experimental section.

But a large variability was seen in any one setting of the flap., as

can be seen from runs 8 and 9 which occurred at the same flap setting.

Control of SOp levels was exercised only on runs 3 and 4 for activated

carbon and 1 and 2 for molecular sieve, and was done by adding SO2

from a cylinder at a constant rate. The variations in temperature,

gas velocity and SO , concentration between runs occurred by natural

variations within the system.

As can be seen by comparing the average efficiency in Table 3

with the average efficiency in Table 5, the activated carbon system

exhibits much better performance than the molecular sieve system.

This is to be expected because of the saturation levels shown in

Table 1. 26

TABLE 1

PRELIMINARY DATA

27

SO2 saturation loading

Bed density

Particle size

Bed void fraction

Activated Carbon

0.2703 g of SO2

g of carbon

0.57 g/cc

1.4 mm (dia.)

0.46 cc gas cc bed

SO2 saturation loading

Density Bed

Particle size

Bed void fraction

Molecular Sieve

0.0366 g of SO2

g of M.S.

0.58 g/cc

2.5 x 8 mm cyl

0.74 cc gas cc bed

o CO

E o

o CO

<

•it Q . O S-

Q

0) S-:3 to to O) %-a.

o cu to

E o

> •

• a

o CO

* * I — OD

o o 00

o o o

28

r r o o

a> ro CVJ o o

^ cr» o o o

00 "«d-1 —

o o

CO 'd-r—

o o

1 —

CO 1 —

o o

r—

o OJ o o

CM «;d-«d-O

O

r t-> CO t_>

o

CM

UJ _ 1 CO

< o

Q

tI3

CtL UJ D -O

O

t/)

(U

to

Ci3

o

ex o Q

O-E

o o

CL. E cu

>

c

cr> 0 0 CO 00

CM t^

00 r

,—

r cr» IT)

o LO

o CM

O CM

i n cr> i n C\J CM

00 CM

o CM 0 0 CM 0 0

cr» 0 0 0 0

0)

a. in o E

4-J

fO

0) to to <u to 03 -%

E 0) +-> to >> to

(U x: 4J

CU s-13 to to O) i -a. <u

03

+-> O

CM CO ID «i3 0 0 CT>

29

TABLE 3

REMOVAL EFFICIENCIES FOR ACTIVATED CARBON

Run SO2-IN, ppm SO2-OUT, ppm Removal Eff., %

0 100

0 100

9 90.9

15 91.4

9 81.6

8 86.9

5 93.9

4 95.2

2 96.7

1

2

3

4

5

6

7

8

9

70

82

99*

174*

49

61

82"

84^

61^

SO2 added from cylinder

regenerated carbon used

CO

UJ 1—(

oo

o UJ

o

Q

CD

O CM

E u

•K a. o s -

Q

(U s-:3 to to O) S-

o (U to

o

cu

XJ

o CO

o a; to

"E o

cu

to

cu

30

CT\

o O

0 0

o 0 0

o

o 0 0 CO o c>

cn ID CM o CD

VD CM r— o o

— ^ CO o o

C3^ C M CM

O CO

S -<u sz a. in o E +J 03

+J c cu to to <u to 03 5

OH UJ

o o

a. o

CM I— LD

CU

to >> to

O)

o. E (U

o o

Q . E

<u >

3

cn U3 1 ^

0 0 o

CM CO

O)

s -

to lO (U s -

Q.

O)

4-> OJ

CU +-> o

31

TABLE 5

REMOVAL EFFICIENCIES FOR MOLECULAR SIEVE

Run SO2-IN, ppm SO2-OUT, ppm Removal Eff., %

61 32.8

70 29.1

42 47.8

41 32.8

1

2

3

4

90*

99*

81

60

•ie

SOp added from cylinder

32

The average pressure drop for the activated carbon system was

0.9 cm of water. This is much better than other systems that clean

flue gases of SO2 (see scale-up results at the end of this section).

Table 6 shows the results of the computer simulation. The

results show good agreement between the model and the data. This

occurs even though the system was not truly isothermal or isobaric

as was assumed in the model. The input data to the model also had

some error, especially in the inlet concentration to the bed. This

has a large effect on the final results of the model as will be shown

presently.

The following calculations are not intended to show the actual

errors involved in the simulation model. They are presented to show

how the errors involved in the input data can multiply through the

course of calculation. The calculation here uses the method of

(8) predicting propagation of error presented in Jenson and Jeffreys.^ '

This method assumes that for large errors the percent error of a

product is equal to the sum of the percent errors for the individual

multipliers plus their product. It also assumes that for addition or

subtraction the errors are additive.

The calculations in this section are done for run 8. The inlet

concentration for this run is 84 j 7 ppm. The 7 ppm was determined

from multiple measurements of the inlet concentration. This cor­

responds to an error of 8.33% in measurement of the inlet concentra­

tion. To show how error propagates in this model it is assumed that

the error in concentration has the most effect, i.e. the other errors

in V , V , T, e^, P| , w^^^ are negligible.

33

Run

TABLE 6

MODEL RESULTS FOR ACTIVATED CARBON

Actual Removal Eff., % Calculated Removal Eff., % % Error

1 100

2 100

3 90.9

4 91.4

5 81.6

6 86.9

7 93.9

8 95.2

9 96.7

89.5

100

99.2

100

81.8

100

91.8

91.5

95.9

10.5

0.0

9.1

9.4

0.2

15.1

2.2

3.9

0.8

34

The following equations are used in the model:

W/ ^ , , = hA gB/T / _ '^(r,s) / .n (r + l,s) VT ' ^r-— ^Cfr.s)^ •"

sat (r.s)' ^«(r.s)

V = s

V = g

% =

^B =

^ a t

W / n .

0.0442

58.6

0.57

0.46

= 0.00418

.X = 0 . 0

'(r,s-M) ^ V ^ ^ ^ ^ ^ U s ) - ^ r , s ) ' ^ ^(r,s)

A = 0.989 X 10"^

B = 5350

n = 1.00

h = 5.920

m = 0.833

<=(r.O) = 0-278 X 10-8

The results obtained using the method outlined above are shown

in Table 7. This calculation is just for one iteration sequence.

Note that the error for w remains the same throughout the first

iteration, this is due to the fact that the solid concentration is

not near saturation.

This calculation shows that any error in the input parameters

can cause tremendous error in the final result because of the dif­

ferential nature of the computer model. Even so, the model did show

good results.

The constants in equation (8) found to best model the data from

the adsorption system were:

A = 0.989 X 10"^

B = 5350

n = 1.00

35

TABLE 7

CALCULATED ERROR RESULTS FROM MODEL

^ 1 , 1 )

^ 2 , 1 ]

^ 3 , 1 ]

^ 4 , i :

^ 5 , 1 ]

^ 6 , i :

"^(1,2

^(2,2

^(3,2

"^(4,2

""(5,2

Concentration

= 0.0

= 0.119

= 0.237

. = 0.356

. = 0.474

. = 0.591

J = 0.122

. = 0.122

. = 0.123

. = 0.123

. = 0.124

X 10"^

X 10"^

X 10'^

X 10"^

X 10"^

X 10'^

X 10"^

X 10"^

X 10"^

X 10'^

% Error

0.0

8.

8.

8.

8.

8.

29

50

71

92

112

3

3

3

3

3

.8

.9

.6

.7

.8

36

The Westvaco constants were:

A = 1.59 X 10"^

B ^ 5520

n = 0.4

The units on the Westvaco equation were different from the ones

used in this study and it also included terms for H O and O2 con­

centrations, so the constants should not be directly comparable.

Concentration in the Westvaco equation was in ppm while in the

adaptation used here it was in gram moles per cm .

The model was used to predict the size of an industrial process

to remove S02- The calculations involved are shown in Appendix C.

The inventory of carbon needed for a 1,000 megawatt coal burning

power plant would be approximately 111,000 lb with a flow rate of

68,000 Ib/hr of carbon. The pressure drop for such a system would

be approximately 6.8 inches of water. The pressure drop for scrubbers

to remove the SO^ would be greater than 10 inches of water. As can

be seen the adsorption system performs much better than scrubbers

in this area.

A complete design and economic analysis was not done for this

system because data on the cycle life of the carbon and regeneration

costs are lacking.

CHAPTER VII

CONCLUSIONS

1. At the conditions studied, the activated carbon used in

this study is a better adsorbent for sulfur dioxide than the molecular

sieves which were used.

2. The adsorbent bed, using activated carbon as the adsorbent,

exhibits satisfactory removal efficiencies for SO2 at the concen­

tration levels tested.

37

CHAPTER VIII

RECOMMENDATIONS

1. Change the method of removing solids from the bottom of

the bed, to improve the flow characteristics of the bed. This can

be done by using a motor driven star feeder.

2. Include analyses for NO2 (by the syringe technique),

H2O and O2 concentrations in the gas stream.

3. Insulate and install cooling coils so that the gas stream

to the bed can have a wider temperature range.

4. Increase the sulfur dioxide concentration entering the bed

to see if a marked decrease in efficiency occurs.

5. More runs with regenerated carbon should be made to find

the working life of the carbon.

38

BIBLIOGRAPHY

1. Balyhiser, R.E.: "Energy Options to the Year 2000", Chemical Engineering, January 3: 80-81 (1977).

2. Brown, G.N. and Torrence, S.L.: "SO2 Recovery via Activated Carbon", Chemical Eng. Prog., 68 (8):55 (1972).

3. C.E.K., "Outlook: Simultaneous SO2 and Fly Ash Removal", Environmental Science and Technology, pages 18-19, January 1971.

4. Ferrell, J.K. and Rousseau, R.W.: "The Development and Testing of a Mathematical Model for Complex Adsorption Beds", Ind. Eng. Chem., Process Des. Dev. 15 (1):114 (1976).

5. Fulker, R.D.: "Adsorption", p. 269 in Processes For Air Pol­lution Control, Nonhebel, G. (ed) CRC Press. Cleveland (1972).

6. Habib, Y.H. and Bischoff, W.F.: "FW-BF Dry Adsorption System for Flue Gas Cleanup a Status Report", p. 415, A.I.CH.E.

7. Howard, W.F.: "Adsorption of NO2 on Activated Carbon", pp. 50-59. M.S. Thesis, Library Texas Tech University, Lubbock, Texas (1969).

8. Jenson, V.G., Jeffreys, G.V.: Mathematical Methods in Chemical Engineering, pp. 356-360. Academic Press, New York, N.Y. (1963).

9. Juntgen, P.S. and Knoblouch, K.: "Removal and Reduction of Sulfur Dioxide from Polluted Gas Streams", p. 180 in Sulfur Removal and Recovery, Pfeiffer, J.B. (ed) American Chemical Society. Washington, D.C. (1975).

10. Kasten, P.R. and Amundson, N.R.: "Analytical Solution for Simple Systems in Moving Bed Adsorbers", Ind. and Eng. Chem. 44:1704 (1952).

11. Lapidus, L.: Digital Computation for Chemical Engineers, pp. 135-140. McGraw-Hill Book Co., New York, N. Y. (1962)

39

40

12. Lovett, W.D. and Cunniff, F.T.: "Air Pollution Control By Activated Carbon", Chem. Eng. Prog. 70; no. 5, 43-47 (1974). —

13. Lowe, C.W.: "Some Techniques of Evolutionary Operation", Trans. Inst. Chem. Engrs. 42 ; T339-T341 (1964).

14. Meador, M.C. and Bethea, R.M.: "Syringe Sampling Technique for Individual Colorimetric Analysis of Reactive Gases". Environmental Science and Technology. 4^:853-855 (1970).

15. Neretniek, I.: "Adsorption in Finite Bath and Countercurrent Flow with Systems Having a Nonlinear Isotherm", Chem. Eng. Sci. 31:107-114 (1976).

16. Noll, K. and Duncan J.: Industrial Air Pollution Control, pp. 86. Ann Arbor Science, Ann Arbor, Michigan (1973).

17. Perkins, H.C.: Air Pollution, pp. 274-275. McGraw-Hill Book Co., New York, N.Y. (1974).

18. Rew, H.C., Patent #290,928 (1883), Gas Purifier, Chicago, Illinois.

19. Ring, T.A. and Fox J.M.: "Fuel vs Stack Gas Desulfurization", pp. 1 in Pollution Control and Clean Energy 1976, A.I.CH. E

20. Schaum, J.: "Chrysler Initiates Odor Control at Huber Avenue Foundry", Modern Casting, Aug. p. 36 (1973).

21. Smith, J.M. and Masamune, S.: "Adsorption Rate Studies — Inter­action of Diffusion and Surface Processes", A.I.CH.E. Journal, ]±:3^ (1965).

22. Squires, Arthur M., and Pfeffer, R., "Panel Bed Filters for Simultaneous Removal of Fly Ash and Sulfur Dioxide: Introduction", A.P.C.A.J. 2^:534-538 (1970).

23. Squires, A.M., U. S. Patent 3,296,775, (January 10, 1967).

24. Weber, T.W. and Chakrovarti, R.K.: "Pore and Solid Diffusion Models for Fixed Bed Adsorbers", A.I.CH.E. 20:2,228 (1974).

25. Zahradnik, R.L., Anyigbo, J., Steinberg, R., and Joor, H.L., "Simultaneous Removal of Fly Ash and SO2 from Gas Streams by a Shaft — Filter-Sorber," Environmental Science & Technology. 4:663-667 (1970).

41

26. Zeny, F.A. and Krockta, H.: "The Evolution of Granular Beds for Gas Filtration and Adsorption", Brit. Chem. Eng. & Proc. Tech. 17_:224-227 (1972).

NOMENCLATURE

A = constant in adsorption rate equation

B = constant in adsorption rate equation

c = concentration of adsorbate in gas phase, g am moles cm

c^ = concentration in particle voids, S^am moles cm

2 D = intraparticle diffusivity,

h = integration step size in the x direction, cm

k = adsorption rate constant, cm"

k^ = mass transfer coefficient, cm/sec

m = integration step size in the z direction, cm

n = constant in adsorption rate equation

R = particle radius, cm

r = particle radial position, cm, also position indicator in the integration grid, in the x direction

s = position indicator in the integration grid, in the z direction

t = time, sec

T = temperature, K

V = superficial gas velocity, cm/sec

V = velocity of adsorbent, cm/sec ^ gram moles SO2

w = carbon loading of sulfur dioxide, —gram carbon

w gram moles SO2

sat = saturation loading of sulfur dioxide, gram carb on

42

43

x = direction of adsorbent flow

y = coordinate that is perpendicular to z and x

z = direction of gas flow

3 3 e. = adsorbent bed void fraction, cm /cm bed D

3 3 e = void fraction inside the particle, cm /cm bed

3 p, = density of adsorbent bed, grams/cm bed

3 p = apparent density of particle, grams/cm bed

APPENDIX

A. Computer Program

B. Analytical Procedures

C. Scale-up Calculations

44

45

APPENDIX A: COMPUTER PROGRAM

c

46 T H I S I S A Pf>0-,R/\M r n n P T I M i Z E THE CHNSTAMTS A , R , N l , AND M? USING THE SIMPLFX METHGU.

MAIN PR:»G'^A^' ' \'Cr>^ENCLATlJRE

C C C C C C C c c c c c c c C C C c

u-\

^'/ I

WSAT=S AKEA1= GAS FL RED-GM GMOLES T I i> E L)

USFO I A RUN-CM/SEC PR= Bb THRnUG THROUG OF EAC THAT M A , R , N I IN THE OF STE OF STE OF SE r THE SI

ATURATED AREA F3

OLES/CC, / : C , T = T E F ADSnRf> N AUS(.)R^ C : , V G = G »\/S= SOL D DENS IT H THE I E H THE B5 H COLUMN ENSURES •AND N 2 ,

7 D I R E :

PS ACROS PS DHWN S I N THE MPLEX,M=

Sn2 CAR R SOLID 2 ,CC= IN CF=CUTLE M^ERATUR TI ON RUM TION RUN AS VELOC ID VEIOC Y -G /CC,C D, GMOLES D - CMOL

IN THE THE DESI AI = THE TI ON,MN S THE BE THE BED SI MP LEX STEP SI

BON LOAD FLOW- CM LET CONC T CONCEN E-DEGREE -SEC. »S3 -GRAMS,G ITY THRO I TV THRO C= CONC. /CC,WW=C ES/G.P= SIMPLEX, RABILITY SIMPLEX

= NO. OF D IN THE IN THE X t N = NO 7E IN TH

ING-GMGL **2,AREA ENTRATIO TRATIDN S K,TI=L L=WEIGHT = VOLUME UGH THE UGH THE IN THE

ONC. ON A SUM AR DEZ = AM OF EACH MATRIX, DATA SE Z OIREC DIRECT I

. OF VAR E X DIRF

ES/ G 2= AREA FOR N TE THE FROM THE RED-ENGTH OF OF SOLID OF GAS FOP

BED-BED- CM/SEC, GAS PHASE THE SnLID RAY-SUM ARRAY SET HE

ML= STEP S I7F TS,K = NO. T I O N , LL= NC. ON,N = NO. lABLES IN C T I E N .

DOUB DOUB 1,0 UB DOUB DOUB N=3 M=A READ FORM AREA AREA EB = n PB=(^ WSAT MM = \

KKK = DO 1 READ CONT FHRM DO 2 VS ( 1 VG( I CONT R E A f

L : PRECISION IE PRECISION! LE PRECISION LE PRECISION LE PRECISION

7( 3 ) , P ( 3 ) , A 1 ( 4 , n , C 0 ( ' ^ ) WW{ i : , J 5 ) , C F ( 9 ) , P B , E P , V J S A T , G { 9)

H , ML, V S ( 9 ) , VG ( 0 ) , CC ( 1 0 , :• 5 ) DFZ (4 ) . A R E A I , A R E A 2 , T ( 9 ) , T I ( 9 ) SOL(9 ) . Z I T (9 )

40 0 , M M » H , ^ L , K , L L

A T ( I 2 , D 1 0 . 3,D 1 C . 3 , I ? , 12) 1 = 81 . 0 0 0 2=9C.C).0D0 . 4 6 D 0 . 5 7 D J = D , 4 18D-?

0 I = i , M N

40 1 ,CU(1 ) , C F ( I ) , T ( n , T I ( I ) ,SOL( I ) ,G( I )

INU^: AT ( ^ 0 1 ' ' . 3)

I = 1 , M N ) = s n L ( I ) / ( TI ( I ) *PB*AREA1 ) ) = G( I ) / ( A R E A 2 * T I ( I ) ) INUL" 40 2 , ( f AI ( I , J ) , J ^ l , r j ) , 1=1 ,M )

47

3 14

C C C C

IH 8

1

C C C

11

12 C C C c c

^5^

^)4

c c

DO 3 1 = 1 , M

CALL S U ^ U A M I , 1) , A M K 2 ) , l . j n j D ^ , A T ( I , 3 ) , H , ^^L , T , C - ' ,

^:^K, L L , V S , V G , C C , WW,CF,PB ,EB ,WSAT , n E Z ( I ) . M ' J , I , 71 r ) C o r j r i N U ^ -yM=MM+1 11 = 1 IF (KKK . r g , I I ) 11 = 2

F I N D I N G THE SMALLEST POINT IN THE SIMPLEX ^ X C L U D I N ' -THE NEWLY FOUND P O I N T .

DO H i= ;^ ,M I F ( K K K . ^ Q . I ) 3 0 TO 18 I F (DFZ ( ! ) . L T .DEZ ( I N ) 11 = 1 CONTINUE CONTINUE K K K = I I DO l U I = 1 , N Z( I ) = A I ( I I , 1 ) DO 9 J = l , N A I ( I I , J ) - O . D O

C A L C U L A T I O N OF MEV. P O I N T S .

on 11 J = I , N P( J ) = D . D O DO 11 I = 1 , M P( J ) = A I ( I , J ) + P ( J ) DO 12 J = 1 , N A I { I I , J) = ( (P ( J )*2 . D O ) / N ) - Z ( . J )

Cf^ECKlNG TO SEE I F NEW POINT I S ALREADY I N THE S i r ^PLEX I F SO THEN THF POINT I S ADDED TO THE S I M P L E X WITHOUT R E C A L C U L A T I N G .

DO 54 I = 1 , M I F ( I . E O . I I ) G? TO 54 MMM = 0 00 5 3 J = 1 , ! I E ( A I ( I I , J ) . E Q . A I ( I , J ) ) MMM=NMM+1

CONTINUE IF ( MMM.E O.N) DEZ ( I 1 ) = n F Z ( I ) IF ( MMM.E 0 . ' O 3 0 TO I ^

CONTINUE

C U L SU^'M A I ( I , 1) , A I ( I , 2 ) , l . C C 0 D n , A I { 1 , 3 ) , F , f ' L , T , C C ,

> : c K , L L , V S , v / G , C C , W W , C F , P B , h n , W S A T , DE 7 ( I ) , M ( i , I , 7 I T )

I F ( M M . L T . 5 ) GO TO 14

F I N D I N G LARGEST P O I N T I N TEE S P n > L E X .

48

15

40 2 5 M

L = l DO 15 1 = 2 , M IF (DEZ ( I ) .GT .DE7 ( L ) ) [.= ! CONTINUE PRINT2 2?. FORMAT ( ' 1 » . • OPTl MUM" PO I N T S ' ) I = L CALL SU'3( A M I , 1) , AI { I , 2 ) , 1 . COO DO, AI ( I , 3 ) ,F

^ : *K ,LL ,VS, VG,CC, W^J, CF , PB , EB , WS AT , DEZ ( I ) ,M'g, I FORMAT ( 3 0 1 ' ' ^ 5) CALL EX IT END

1 ( 1 , 3 ) , H , M L . T , C O , , Z I T )

49

C C C C

C C C c c c

1 c c c

1 V!.'

SUBROUTINE SUB(A,B,N 1,N2,H,M,T,CO,K,L,VS,VG, CC, WW. *CF,PR,EB,WSAT,7ED,MN,I,DEZ)

THIS PROGRAM DETERf^iINFS THE DESIRABILITY COEFFICIENT OF EACH SET OF A,B,N1,N2.

DOUBLE -PRECISION H , M , A , B, Nl, N2 , PB , E B, WS AT ,Z ED DOUBLE ^^RECISION DEZ ( MN ) , CO ( MN ) , VS ( MN ) , \/G ( MN I DOUBLE i^RECISlON CC ( L , K )., W V ( L, K ) , CF ( MN ) . T (MN )

USING THE A,B,N1,N2,T0 DETERMINE THE INDIVIDUAL DESIRABILITY COEFFICIENT FOR EACH RUN, DFSIRABILITV= CLOSENESS OF CALCULATED OUTLET CONCENTRATION TO REAL OUTLET CONC ENTRAT 1 0 ,N .

DO 1 K I < = 1 , M N CALL C A L C ( A , B , N 2 , H , M , T ( K I K ) , C J ( K I K ) , K , L , V S ( K I K ) ,

>:< VG ( KI K ) , CC , WW, CF ( K I K ) , PfW E B, WS AT , N1 , DE 7 ( K IK ) , MN ) CONTINUE

DETERMINING THE OVERALL D E S I R A B I L I T Y COEFFIC IENT.

IT = I DO 2 I 1 = ?,MN I F ( D E Z ( ! 1 ) . L T . D E Z ( IT ) ) I T = I I CONTINUE Z E 0 = D E 7 { I T ) P R I N T i n i , Z E D FORMAT ( • - « ,«0»/ER ALL D E S I R A B I L I T Y = • , 0 1 0 . 3 )

RETURN END

TEXAS TECH LIBRARY

50

C

C

2 C C C C

i

4 5 C C C C

8

i i a 10^ 10 3

1'.4

SUBROUTINE C AL C( A ,B , N, H . M , T , CO, K, L , VS , V<;, C , Ir. ,CF , P B. * E B , W S A T , F G , D E Z , M N )

THIS IS A PROGRAM TO SIMULATE A MOVING BED ADSORBER USING A F I N I T E DIFFERENCE TECHNIUUE.

DOUBLE PRECISION DEZ DOUBLE PRECISION CO , H , M , A , B , T , N , V S , V G . P B , W S A T , S U M DOUBLE PRECISION C ( L ,K ) , W ( L , K ) ,FB , C F , F G , D C F , A V E L 1 = L - I K 1 = K - 1 DO 1 1 = ] , L c( I , i ) = : o D02 1=1,K W( 1 , I ) = 0 . D 0

CALCULATING THE CONCENTRATION THROUGH THE BED, ON THE SOLID-W,AND I N THE GAS-C.

D05 J = 1 , K 1 DO 3 1 = 1 , L I W( I + 1 , J) = ( (HX^A ) / V S ) *DFXP(B/T) ' ' . - ( (DABS

* ( 1 - ( W ( I , J) /WSAT) ) )'?=*FG)'^-( (DABS (C( I , J ) ) )';='!'N )+W( I , J ) DO 4 I = 1 , L 1 C ( I , J « - l ) = - ( ( M * P B * V S ) / ( VG*H«EB) ) * ( W ( I +1 , J ) -W ( I , J ) ) +

* C ( I , J ) CONTINUE CONTINUE

TAKING AVERAGE 10 GET THE OVERALL OUTLET COMCENTRATIOrj.

SUM=0. [)G DO 6 1=1 ,L 1 SUM = SUM«C ( I , K ) P R I N T l ? f ) AVF=SUM/L I IF ( C F . E'O.O.D'..-) GO TO 7 DCF=AVE-CF i F ( D c r . E o . o . n o ) D C F = I . G O O D - l 3

P R I N T 1 0 3 , V S , V 3 , 3 0 PR I MTl J'-- , CF, AV E, DCF nFZ = DABS (1 . 0 O : . / D C F ) PRirJT 105 , A ,B ,FG , N , DG 7 r,(» T{1 9

FrjRMAT ( T^ • , •W= ' , D 1 C . 3) FORMAT ( ' O ' , •C= ' , 0 1'. . 3) FORMAT ( ' : )• , • SOLI D VEl . = ' , D l u

* , LU f . . 3 . 3 X , • INL ET CONC = * • DlC . . . ' . - l . r ^ . • r i i l T l r T

3 , 3 X , ' G A S VE 3 )

- I

= ,LUf. . - , 3 X , • INLET CONC. = ' , D U. . 3 I FT^.MAT ( ' 0* , 'OUTLE T CON C . = ' , D 10 . 3 , 3 X ,

51

V:J

7

9

^ •CALCULATED OUTLET CONC . = • , 0 10 . 3 , 3X , ^ • D I F F E R E N C E BETWEEN CALC. AND ACTUAL = ^ , D l ' - . 3 )

FORMAT C O * ,»A= ' , D 10 . 3, 3X , » H= • , D 10 . 3 , 3X , ' N = ' , D 10 . 3 , * 3 X , •N2 = ' , D 1 0 . 3 , 3 X , •DE7 = ' , D l 0 . i )

FORMAl ( ' - ' , • PR INT-OUT F ROM C ALCUL AT I ON t>Rt:GRAM') CONTINUE I F ( A V E . L T . O.DO ) AVE = 0. DO GO TO B

CONTINUE RETURN END

52

APPENDIX B: ANALYTICAL PROCEDURES

Determination of Sulfur Dioxide Concentration by the

Lyshkow Modified West and Gaeke Method

This method is intended for use when the sulfur dioxide con­

centration in the gas to be sampled is in the range of 0.17 to 50

ppm. The sulfur dioxide concentration is determined spectrophoto-

metrically by absorption in para-rosaniline solution.^^^^

The para-rosaniline solution is prepared as follows:

1. Concentrated Reagent

To 640 ml of 1:1 HCL/H2P add 0.800 g para-rosaniline

dihydrochloride. Allow to stand for 1 hour with periodic

mixing, then filter through Eaton-Dikeman filter paper

grade 512 or the equivalent.

2. Dilute Reagent

Dilute 32 ml of concentrated reagent to 900 ml using

distilled water. To this add 1.1 ml formaldehyde (37% ACS)

and dilute to 1 liter. Mix and allow to stand for 12

hours prior to use.

A standard sulfite solution, 0.02323 g/liter sodium metabisul-

fite, is required for the spectrophotometric determination. One ml

of this solution diluted to 25 ml produced a color equivalent to

5 ppm. This solution should be prepared fresh for each use by

dilution from a solution containing 2.323 g sodium metabisulfite

(assay 65.5% as SO2 in 1.0 liter of water). The stronger solution

53

should be kept in the refrigerator. The freshly prepared dilute

solution must be standardized by titration with standard 0.01 N

iodine using starch as the indicator, see Federal Register, vol. 36,

No. 228, Thursday, November 25, 1971, page 22386, also refer to

the reference by Meador and Bethea.^^^^

The apparatus used consisted of 50 ml disposable polypropylene

syringes and a spectrophotometer.

Analytical Procedure

1. Draw exactly 3 ml of the absorbing reagent into a 50 ml

syringe.

2. Draw a 47 ml gas sample into the syringe at the rate of

2 ml per second or less.

3. Shake the syringe gently for 3 minutes to develop the color

fully.

4. Transfer the sample to a 1 cm cuvette.

5. Read the transmittance at 560 my with respect to unexposed

reagent as 100% transmittance. Exposed reagent must be read within

an hour of exposure.

The syringes used must be cleaned with acetone before use to

remove the lubricating oil of the plunger. Also a high concentration

of sulfur dioxide should be left in the syringes overnight to pre­

condition them.

To determine the concentration in ppm from the transmittance

a plot of sulfur dioxide concentration vs In % transmittance must

be prepared. This can be done by using a permeation tube technique

54

or a chemical standardization technique. In this study chemical

standardization was used as follows:

Pipette 1 ml, 2 ml, 3 ml, etc. up to 10 ml of the standard

sodium metabisulfite solution into a series of 25 ml volumetric

flasks, and dilute to marks with the dilute para-rosaniline solution.

Mix and allow 3 minutes for color development before reading the

percent transmittance.

The data from the chemical standardization is then used to

determine a least squares fit to In % transmittance vs ppm. The

point at 100% transmittance and 0 ppm should be included.

Since the concentration of sulfur dioxide in the gas stream

was higher than the 0-50 ppm range that this method is used for,

a gas sample less than the 47 ml recommended was used. A ratio of

47 ml was multiplied with the ppm reading to get the actual ppm ml used

in the gas stream.

55

APPENDIX C: SCALE-UP CALCULATION

This design is based on a 1,000 megawatt plant having 35

percent thermal efficiency. The fuel for this plant is coal with

a sulfur content of 2.5 to 3.0 percent. The flue gas will be taken

as 2.3 x 10 acfm at 275° F with a SO2 concentration of 2000 ppm.^^^'

The computer model was used to find the size of bed necessary to

achieve an 85-90 percent reduction of SO2 concentration. The

program print out gives the following information:

T = 84° C SO2 removal = 88%

V^ = 0.0033 ft/sec V = 6.43 ft/sec s g

Height of bed - 19.7 ft Thickness of bed = 7.2 in.

Using this information the following calculations can be made.

Gas volume at 84° C = 2.3 x 10^ (UJ) = 2.0 x 10^ cfm

I n 4-h o-P K^^ - 2.0 x 10^ cfm = 265 ft Length of bed - (19.7 n ) (385.8 ft/min)

Amount of carbon in bed = (265) (0.6) (19.7) (35.58)*

= 1.11 X 10^ lb

Rate of carbon use = (265) (0.6) (0.0033) (35.58) (60)^

= 6.8 X 10^ Ib/hr

= 34 ton/hr

*(35.58 = density of bed in Ib/ft*^)

The pressure drop can be estimated by assuming linear scale-up.

The maximum pressure drop measured was 1.4 cm of H2O for 114 cm/sec

gas velocity. The system here uses a gas velocity of 196 cm/sec with

56

a bed thickness of 7.2 inches. The estimated pressure drop would

then be calculated as follows:

( •4) ({^) (^) (2:34) = 6.8 inches of H2O