38 Modelling SO2 emissions from fluidized bed coal combustors

Chemrcal Engzneermg Scaence Vol 35, pp 302-306 PergamonPress Ltd ,1980, Pnntedln Great Bntiun

38

David

C

Department

MODELLING

SO2

EMISSIONS

FLUIDIZED

BED

COAL

Lee,

of

James

Hodges

L

Chemical

Englneerlng

Massachusetts

Institute

Cambridge,

FROM

COMBUSTORS

and

and of

Georgakls

Energy

Laboratory

Technology 02139

Massachusetts U.S

Chrlstos

A

ABSTRACT

An analytIca model 1s presented for the calculation of sulfur amlss~ons and sulfur retention by sorbent stones In fluldlzed bed coal combustors as a function of the molar calcium to sulfur feed ratlo The model 1s based on an approximate rate expresslon for each sorbent particle and the two phase theory of fluldlzatlon It 1s shown that the sulfur retention 1s not dependent on the amount of excess air used or the sulfur content of the coal, as long as the gas residence time and the calcium to sulfur ratlo are fixed. Furthermore, there exists a maxlmum value of sulfur retention that 1s attalned at high values of the calcium to sulfur ratlo and depends directly on the stone reacIZlVl-ty It 1s also shown that the ppm SO2 emlsslons are proportional to the sulfur Comparisons of content of the coal and Inversely dependent on the excess azr used. model predlctlons with experlmental data are quite satisfactory.

KEXWORDS

Desulfurrzatlon,

emlsslons,

SULFATION

REACTION

fluld

bed

combustors,

reactor

models.

gas-solId

ractlons

RATE

Several researchers (Chr1stowskl and Georgakls, 1978, Georgakls, Chang and Szekely, 1976) have modeled the sulfatlon reactlon of carbonate rocks, 1979. Hartam and Coughlln, which complicates thezr use In an These nonllnear models require numerlcal solution, overall fluldlzed bed sulfur balance, due to the existence of particles of different sizes and residence times Borgwardt's model (19701, although sunple, 1s valLd only Motivated for reactlon times much shorter than the resrdence times In a fluldlzed bed. by 1) the observat%on that almost all experImenta data show an exponential decrease of 1~) the observation that the the reaction rate with time (Vogel and co-workers, 1977). reactlon 1s first order with respect to sulfur dloxlde (Borgwardt, 1970, Yang and co-workers, 1975). and 111) analytlcal calculations from more detalled gas-solld the following reactlon rate reaction models (Georgakls and co-workers, 1978, 1979), expression 1s assumed R(t)=Roexp(-t/Tp) (1) Here

R(t)

and

R.

(Kmol/sec/partlcle)

are

the

reaction

rate

per

partzcle

and

the

lnltlal

rate, respectively, and Tp(sec) 1s the pore pluggzng time constant taken as equal to one third of the pore plugging time, up defined by Georgakls and co-workers (1978, The lnltlal rate 1s proportlonal to the emulsion phase SO2 concentration 1979)

302

Modellmg

G-38

SO2 ern~~~ona from fluldlzedbed coalcombustors

303

cpwl/m3) R0=4*r35/3T Here

r(m)

1s

the

radius

measure of the stone pore size dlstrlbutlon, of the pore pluggrng concentration so2

of

the

(2)

sf and

sorbent

particle T the sulfatlon time, 1s (set), sf depends on the stone size, Internal surface area, composltlon. Furthermore, analytical calculations that rp 1s Inversely proportIona to the amblent

ractlvlty, and and chemical time have shown

(3)

Tp=P*/c Consequently dence of the

the CaO

slmpllfled converslon,

rate expresslon a(t), to CaSO4-

1 can be

integrated

to

yield

the

time

depen-

!Zn(l-a(t)/a,)=-tcl/p, The final parameters

conversion, as follows

a

1s

often

less

pa(Kg/m3)

represents

the

being

the

weight

(4)

unity

and

1s

related

to

the

other

model

m a_=p,m

With

than

a

density

fractron

of of

CaC03 the

CaCO,

3

uncalclned In

(5)

"CsfPaWCaCO

the

3 carbonate

stone

while

W

CaC03

stone.

12

Fig.

1 Comparison experlmental

between sulfatlon

modeled and rate data

In Fig. 1 calcrum converszon date (Vogel and co-workers, 1977) are compared with the predlctlon obtained from this slmpllfled value of a J_S set equal to the fznal converslon lndzcated by by linear regressIon_ For Tymochtee Dolomite of P* 1s f?tted are 0.96 and 0.245 Kmol and 0.174 Kmol.s/m3. The ment of other experimental REACTOR

s/m3 and the corresponding ones for Dolomite 1337 are 0.85 accuracy of this simple model 1s quite good. Slmllar treatdata has resulted In equally satisfactory results (Lee, 1979:

MODEL

The Fluldlzed Bed Reactor bubble phase The emulsion the coal sulfur 1s assumed

1s

modelled as conslstlng of two phases, the phase which contains the solids 1s assumed uniformly released In the emulsion phase as

absorbed by the sorbent stones. The bubble model with an average bubble size and bubble 1s The

also

assumed

dlfferentlal

for two types of stones rate expressIon_ The the data while the value the values of a_ and p*

for

the

sulfure

mass

exchange

balance

In

phase rise

J_S characterized velocity ub(m/s)

coeffzclent, the

bubble

dc2/dc=M(cl-c2)

Kp(sec phase

-'I,

by

between

emulsion and well mlxed. All S02, which 1s

a plug flow An average value the

two

phases.

1s (6)

where c=x/h IS the dunentronless The value of the dlmentronless between 1s

the

two

On

phases.

distance In the bed and number M=Kph/u,, increases

the

other

Here F 1s the molar Inlet flow s N the bed cross sectlonal area, their residence time dlstrlbutron, average

residence

into

eq

time

7 we

By

hand,

the

sulfure

c1 +GAKph~/cl-c2)

Fs=(l-B)AUmf

1

G-38

Industrral Applzcatzons Waste Rocessmng - Combushon - Gaszficatzon

304

of

sulfur,

6

eq.

6

1s

for

the

Qo=uoA

1s

1s

a measure the

equal

In

the

emulsion

phase

bubble

(7)

volume

fraction,

c,(E)

and

A

1s

the

bed, and E(t) 1s where ts 1s the

substltutlng

the

result

and

eq

obtain

the

volumetric

flow

rate

of

of

the

acceptor

rmxlng

stones

between

equal

to

the

two

Uo

alre,

Cl-exp

The

1s

the

flu~d~zlng

velocity,

(--Ml ) 3 ho

phases,

4ar3N/3ts,

(8)

(rp+ts)

the

f= [ (i-6) umf+6u,

of

balance

expanded bed hexght. amount of mlxzng

In the particles to exp(-t/ts)/t,,

Fs=fclQo+clQatS~p/TSf Here

the the

d<+RyOE(t)R(t)dt

number of sorbent taken as equal

saving

h 1s with

and

Q

reactor

1s a exit

the

solumetrlc

SO2

concentration,

flow

rate c,

1s

to l(l-6)umfcl+b~c2(1~l/uo=fcl

If

the

stones

are

not

reactive

the sulfur use of eq.

Consequently, 8=1-c/&. By

(zsf*),

retalned 3 and eq.

the

exit

concentration

by the stones 8 the following

when they equation

1s

given

are active for 3 1s

by

S=F~/Q,.

1s equal obtalned

to

e2-(l+fg+y)e+y=o where

6trp/ts=p,/tse

pressed the

as

gas,

and

equal

to

the

other

on

The

Y=Q

P /Qorsf. ap 3Ah(l--6)(1-smf)/4~r3 hand,

are

The

equal

to

(9)

number

t

of

stone

particles

superflclal

'h/u0

and

true

N

can

be

residence

extune

of

and

go tg=(6+(1-6)E These

imply

that

It

Qa/Qo=(tgo-tq)/ts-

mf)h/u

can

also

Qa =o.z~~o .

cs

pa

0

be

shown

ws

that

(10)

-T-I

cc0

a

iq

3

(11) where

n

1s

the

molar

calcium

to

sulfur

fraction of sulfur In the coal and air-to-fuel weight ratio/stochlometrlc are proportional to n:

+

ratlo, LS

p

1s

0

the

air

density,

the fractional excess air air-to-fuel weight ratro.

W

S

LS

the

defined by Consequently

C-c -Tg) 90

8=a,tsfni

werght

l++ = S and

y

(12) (13)

It can now mum value

be

observed

that

for

large

values

of

n

the

e_=l/(l+fB/y)=l/(l~f~sf/(~

-r go

An Increase In a uore reactzve

6 _ stone

1s only possible by will for whach T sf

Nezther the sulfur content of the coal parameters 6 and y, and, consequently, as long as the calcium to sulfur ratlo, em~sslons

rn

ppm

at

the

stack

are

given

sulfur

rncreaslng be smaller.

nor the n, by

retention

attains

a maxi-

1) g

the

gas

reszdence

the excess air used affect sulfur retention ~~11 not On the remains constant.

time

or

selecting

the dlmenslonless be affected either other hand the SO

2

Modellmg SO2 emlssons from fluldned bed coal combustors

G-38

78.30x103Ws(1-8)/(l++)ppm and the

they are proportional to the sulfur content the average stone excess air used. Finally, COMPARISON

WITH

EXPERIMENTAL

305

SO2

(14)

of the coal conversion

and rnversely is equal to

dependent 8/n_

on

DATA

go-

RO-

70x ;: 2= 602

50--

Ltmastone

U K

cz at 40

-% s I” COOI

30-

20-

0

2 71

d

2 40

I

0

‘“ok+5-

I

IO co/s

Fig_ 2 Effect of retention in a

20 Mole

Ca/S 0.15m

25

1

I

25

30

I

I5

: 5

Rot10

molar rat10 diameter

on sulfur combustor

In Figure 2 the model predrctions for the dependence of the percent sulfur retention on the calcium to sulfur (Cd/S) mole ratio 1s compared with experimental data (National Coal Board, 1971) for coals with different sulfur contents. Taking into account the scatter in the experimental data the model predictions are quite satisfactory In this case the fluidized bed was operating at atmospheric pressures, its height was 0 6m and the temperature equal to 1070°K and the fluidizing velocity was O-9m/s By use of a computer program developed by a group parallel to ours at the Energy Laboratory of MIT, the fluldmechanical properties were calculated. 6=0

(Ub=2-32m/S,

The

values

of

T sf

357,

Kp=l.368,

-d

p*

were

In Fig. 3 (next page), the obtained from a pressurized 1s 1.98m/s, the bed height ub=3.25,

6=0.425,

accurate resulted

in in

Kp=O

presenting similarly

The desulfurlzatlon sulfur retention

425,

adJusted

SO

to

E mf'0.474) best

fit

the

experimental

data.

model predictions are compared against Here the bed reactor (Hoake, 1977). 5m and the calculated fluidmechanlcal u,f=O.173,

these data, satisfactory

model

and

~m~=0.032,

presented emissions

Eti=O_474_

Similar results

It

is

treatment of (Lee, 1979).

provides a as a function

seen other

simple analytical of the calcium

experimental fluidizing properties-

that sets

model of

is data

data velocity quite has

way to estimate to sulfur molar

the feed

2 shown that this dependence is influenced by three dimensionless groups f, 8 and y. The first one, f. is dependent upon the fluidmechanzcal characterlstxcs of the hed and is a measure of the mixing between the bubble and emulsion phases. The second variable, S, is proportional to the maximum conversion of the stone, cz,, the calcium to sulfur molar feed ratio n and also proportional to the ratio

ratlo. It was of parameters

306

Zndustrta2 Appbcahons

Waste Processrng -

co/S

Fig. 3 Effect of retention In a of a sulfatlon parameter, y, of the stone can be easily of stones.

Combushon

-

Gastficatton

G-38

(Mole/Mole)

Ca/S molar ratlo on sulfur 0.3/m drameter combustor

time constant over the reszdence time of the gas m the bed. The third 1s proportIona to the calcium to sulfur ratlo and the maximum conversion Comparison with experunental data 1s quz_te satisfactory. This model used to estimate the economic feaslblllty of the use of different types

REFERENCES Borgwardt, R. H. (1970). Environmental Sclentzflc Technlclan Chrostowskl, J. W. and C. Georgakls (1978)_ ACS Symposium .Se;lzL,(&: szk Georgakls, C , C. W chang, and J. Szekely (1979). Chem Eng. SC~., 34, 1072 Hartman, M. and R. W. Coughlln (1976). AIChE J-, 22, 490. Hoke, R. C. (1977) Exxon Research and Englneerlng Co. Report EPA-600/7-77-107. Lee, D C (1979). S.M Thesis, Massachusetts Institute of Technology. National Coal Board (1971) "Reduction of Atmospherzc Pollution" Vol. III, Appendix London, England. Vogel. G. J. (1977). Argonne National Laboratory Quarterly Report AWL/ES-CEM-1019. Yang, R. T., P. T. Gunnlnghsm, W. I. Wilson and S A Johnson (1975). Adv Chem. 149. Series, 139,

5,