-. _ _

0EIsl+u~’ --- Pub~coqopany,e -printedilltim

_‘. ..-~ -

L.lMEsfowREA- wrn3 so, As I)EXERMI&D BY &RMO- GRAVIMETRIC ANALYSIS AND -kS MEASURED INPILoT SCALE FLUIDIZED-BED COAL COMBUSTORS* :

ROBERT SNYD~W. IRA WILSON A.SD IRVING JOHNSON

CXcrnid EiTginrrrrirl Diviifbn, &gone N-d lhborarov, Argormr, UL 60439 (USA)

A thermogravimetri~ analyzer was used to zneasure the reactivity of several limestones with SO=_ The Laboratory resuhs can be directly used to predict coal combustion results for pilot-plant atmospheric pressure fluid&d beds. Also, the TGA information was fitted to two different straight-line least-squares fits. These equations allowed prediction (by two different methods) of calcium utilization in atmospheric-pressure piIot plant ffuidized-bed combustors (FBCs) from knowiedge of only the CaC03 content of the limestone. When the predictions based on laboratory work for seven limestones were compared with the pilot-plant experimental results of Pope, Evans and Robbins, Argonne National Laboratory, Consolidation Coal Co_; I.. and Morgantown Energy R esearch Center, agreement was good for ah three methods-

Fhridized-bed combustion of coal is receiktig considerable attention due to its potentiaify higher efficiency and lower capital costs’ than conventional methods. In this procws, coal is burned in a fiuidized bed of a partially sulfated solid SO,sorbent __ (such as limestone or doiomite) at 8%950°C and 1014010 kPa pressure. The heat of combustion is utihzed by generating steam in boiler tubes immersed in the bed __ with the rest of the heat recovered in a downstream steam generator. The bed materiai not only helps to transfer heat to the boiier tubes but also captures SO=. The CaC03

_ in the limestone (or doIon&) bed material reacts with SO2 and 0, at these operating conditions to form stable CaS04_

N-Y cxarring limestines have been intensively studied’-7 as bed material because of their highcakium content (for SO, removal) and their Iow cost.

- _ Harvey et aLs made an extensive study of the Petrographic B of limestones and attempted to correlate their findings with limestone desulfurimtion -

‘data. tITleit et aLg studied the reaction of specific !imestones~under a variety of ^ - - -.

258

conditions (1014010 kPa and 75&iO5O”C)_ The rate of cakination of the limestone was found to have a large effect on the SO1 capacity of the limestones_ Coutant et aLro studied 25 limestones; however, he used short residence times and smah particks (- 100 F)_ Potter * ’ studied 86 limestones and found that the surface area

of pores larger than 03 pm in hmestont particles was important in the capture of SO, but that smaller pores were ineffective_ Investigators * ‘* r 3 have attempt& to

model the limestone-SO2 reaction process. Most modds assume that two diffusional

processes occur; the diffusion of SO2 through pores and the diffusion of SO, through a solid reaction product which forms on the pore surfaces or grain surfaces_

Lacking from al1 of these studies is a method of predicting the limestone reaaivity with SO, at flukiizd-bed coal combustion conditions. When a high-sulfur

coaI (4% S) is used, approximately 35% of the SO, must be retained in the bed

material in order to meet EPA SO2 emission standards. Therefore, process R&D

engineers, in reporting results from experimental tests, usualiy give SO2 reduction or retention as a fimction of the molar ratio of calcium (in the limestone)jsulfur (;ti

the coal) in the feed. The calcium utiiization of the limestone can then be obtained as

fo~ows_

V=_R CajS

where U = the fraaion of the calcium converted to calcium sulfate, R = fraction of

sulfur dioxide reduction or retained in bed, and CajS = molar CajS feed ratio. The designer of a plant knows the value of R required to meet the EPA SO2

emission standards for a designated coal and needs to know the Ca/S ratio required, for desi_gnated limestone(s), to meet this EPA standard_ Therefore, if the calcium

utiIization, U, for a gihren Iimestone can be predicted, then the Ca/S ratio required to meet the

a function

a CajS ratio of 4 is used, the maximum cakium utilization is 0.25. Therefore, for a highly reactive limestone, the calcium utilization will be limited by theavailability of SO2 or the CajS ratio, whereas for a limestone which has a Iow reactivity with S02, the cakium utilization will be limited by the reactivity of the limestone_ When the vaIue of R is just large enough to meet the EPA SO, star&~& the minisnurn required

Ca/S ratio is knovvu The predictions preseuted below arc onIy for atmvheric fluidizxxi-bed combustion and cannot be directly used to predict pressun ‘zed flui sized- bed opera&s conditions_

A thermogravimetric analyzer (TGA), Fig 1, was used for gll kinetic studies- A sample (100-200 mg) was pIaced in a wire mesh platinum basket and was suspcndcd from one arm of a recording microbahurcc in a quartz reactor h&c. The reactor tube wzu heated by a pfatinurn-wound furnace which was con&led to *5”C at te&pcra-

259

Fig. l- t5ckmatic diagram of the TGA apparatus

tures up to 12OO’C Two Pt/Pt-lO”A Rh thermocouples were used for temperature monitoring and controllinp.

The _eas mixtures were prepared by bleeding streams of the individual con- stituents- Mzss spectrometric analyses were made on samples of the final z-as mixture

to verify its composition; the measured values of the SO, and O2 concentrations were within 5% of the values computed from the flow rates.

The gas mixtures, which reacted with the samples, passed upward through the bested reaction tube and exited through a condenser and a series of scrubbers_ A

nitrogen purge ~a that fiowed through the microbalance beil jar kept. it free of the corrosive reactant gases-

The TGA unit continuously recorti the weight change of the sample during its reaction with the gas mixture. Therefore, chemical analyses of the SampIe at the completion of a TGA run were utilized to help determine and quantify the chemical changes which had CMXIIIT~~_ The accuracy of the TGA was approximately &l rng or approximateiy *3% of the total weight change_

RESULTS

Seven limestones which h2d previously been tested in pilot-smle FBCs were

tested for reactivity with SO, and O2 in the TGA. AII reactions were performed at

900°C plith a 0.3 % SO,-5 % Oz in N2 syntheticcombnstion gas. All limestones, sized to -18 +20 mesh (loo0 lun), were precalM in either 20% COi or 100% CO2 prior to reaction with the synthetic combustion &as_

_ The conversion of the calcium in the limestone to c&&m sulfate is given in Kg_ 2 foi the seven limestone!5 calcinfxi in 100% C02. (For limestones precalcined in 20% COr, for Which daza are omirtedj initial rextion~rates were the same but total

in

I

I%R

, Oomrnny

i%R

Va

lley

Clc

mia

ny

' Vullay

WI%&

',

Q

uney

Grew

PER

char

PER 'I arccr

PER'

Orccr

MBRC

Qrccr

MERC

chr

ANL:

1359

ANL

13s9

ANL

1359

ANL :

1359

ANL

( 1359

ANC

,, if35

% ANL' "', ~ochlca

ANL,'

TytIjoclW

:z:

,I,

Tymochtca

,Tymochlco'

E 'Qmochtco

7)mochtce

ccc

Tymoclitco

Stan

dqrd

va

~iatlorr

419

2@6

1600

54

2

2t8

3,7

:::

209

244

5.76

2117

266

214

::i

2,5

;::

212

1.6

xs

8,O

410

109

,

65

2#6

2.6

:.6

:I

15

15

15

15

15

15

I5

1:

15

215

1:

::

25,o

21

,d

26,3

26

86

27#9

28

,O

1386

40

86

20

;:

36

32

s:

::

53

12*3

23

.3

47.4

6;

2 82

6#7

17*7

21

14

22,6

23

86

25,s

16

d 32

,6

10.7

IO,8

1017

II

,0

IO*7

10

.7

30.9

42

,7

55,s

k"'

ii5

5915

74

14.3

lBt7

20

,4

z::

27,6

31

,5

17s

43‘9

28

29

:t3

2Bt3

28

13

3815

42

‘4

56,4

59

05

12

E5

6013

81

#5

&4,2

’ 17

.2

,’ 16

05

30.1

"

* 24

,O "

‘;

22e8

22

,3'

24.7

24

17

27.2

26

17

29.0

28

11

35*1

31

tB

14e6

17

11

3Bd

39s

) 3O

aB

30,o

,’

3008

’

3008

30

8 3O

bO

3Oti

B 31

b2'

1 30

,B

‘, 30

,4

3018

3o

u4

36G6

37

,l

3813

,,

4312

52

06%

,! 55

s 56

81

5713

10

85

1263

-.

’

2111

24

15'

4413

49*7

( 561

1 82

d ’

51t2

'(

62

97,

&49

&5,5

I

g

- -

- -_

.

_ _

_.

. ..-

. _.

- .

262

average pore diameter of the cakined limestone, and thus increak% Iimestone reztiv@ and cakium utihzation. There is less effect on dolomitic Iimestones due to their

favorabIe pore structure created by the conversion of MgCOS to MgO during pre-

CalcinatiOn-

Three methods were investigated as possibIe means of predicting calcium utilization in atmospheric-pressure FBCs using the kinetic information obtained with the TGA. These predictions are shown in Table 1 and are mmpared with pilot phmt experimental reurhs. Pope, Evans and Robbins (PER) determined the desirhii- tion behavior of Greer, Germany VaII& andChaney iimestones; Argonne National

Laboratory (ANL) studkd Iimestones 1359, 1360, 1337, and Tymochtee dolomite. Consolidation Coal Co. (CCC) tested Tymochtcc dolomite, and Morgantown Energ] Research Center (MERC) tested Greer limestone.

it should bc noted that pilot-plant resu1t.s in Tabk 1 were for a wide range of operating conditions. The PER combustion runs were done at 815-832”C. The ANL

and MERC pilot-plant runs were aIi Performed at 871 “C. The Consolidation Coal Co. rrtns x&h Tymochtee dolomite were performed at 982°C. In contrast, ah TGA runs

were performed at 900°C. The superkiai _B vekcities ranged from 0.61 m/set in the MERC runs to

4.51 m/set used by PER. Pope, Evans and Robbins used -l/4 in. Iimestone; 50% by weight of the material was smaller than 30 mesh (660 m diam.). It wouid be expected that very IittIe of the material smaher than 30 mesh stayed in the bed for +ut appreciable time_ In the ANL experimental runs, an average Particle size of 54G630

m was wed while Consolidation Coal Co. used - 16 L2g mesh (119CUOOO p) mater%. AII TGA k&oratory kinetic data was taken using -18 -I-20 mesh (HKIO- 841 m) bmcstone.

The most involved cakulation method (method 1 in Table 1) requires the use of the kinetic information (Fig 2) along with a fluid-bed desulfurization equation ~

developed by &aims et ZSI.~, viz

where U = cakjum utiIizatio~_ fraction._ ca/S = cdcium to dur mole ratio,

V = supedciaI p docity (m/see), H = fluidizcd-bed height (m), E = bed voidage, assumed to be. 0.5, and k .= average particle reaction rate constant (set-‘). This fiuid-bed desuifurizttion equation gives the adcium utilization as a function of the “average” reaction rate constant of the particles in the bed (provided the superIj&l

gas v&city and bed height are known). Thw in or&r to d&ermine U, the “average” raiecodstant,k,mustbeknown.

The TGA kinetic information [the sIopes of curyes in eqn. (Z).j give& the instantaneous rate constant, F, as a function of cakium utiition- Therefore, ke .I average limestone calcium utiIizxtion, U, is that value at which k and K are equal. (see Kearin et aL6 for calculation derails). The predictions using method 1, shoti in Table I, are also plotted in Figs. 4 and;5 for linktoneS precaIcined in 20% CO2 -

263

PUECALCINED 2Q% CO2 DOLOMITE

and 100 oA CO,, respectiveiy. For limestones precalcbd in 20% .CO,, the agreement bet-men TGA and pilot pIant remIts is excellerit for the dolomiticlimestones; however, the predictcci calcium utilizations for cakitic-likstooes (1359, Germany Valley, Greer, and chanr=y) are low (o~crail standard deviation of 11). However, for the

slowly cakined timestone (Fig. 5), the agreemexit for all Jimestoaes ii excelknt; the standard deviation is 4.2.

The better agreement fir the stone precahzined in 100% CO, implies that the

limestones when used in a pilot-plant fiuidized bed undergo slow cakinations. It is difiicuit to imagine a bigb COz concenVation in the combustor or a slow r%e in particle

temperature when it is injected into the bed. Thus, the good agreement between slowcaicination TGA resu1t.s and pilot-plant resuhs has not been expIained_ These results ako imply that precakination may provide no enhancement of reactivity_

However, the effect of precakhation must still be determined in pilot-plant operation. The results in Figs. 4 and 5 were obtained from the reaction rate curve for each

designat& limestone on a TGA and by use of eqn. (2). However, the calcium utika- tion can be predicted by “method 2= by howing only the calcium carbonate content

of a limestone_ The maximum calcium utilization of ‘the limestone (only the slow

&&nation condition with iOO% CO, is considered in the rcxnaindcr df the paper)

is shown as a function of CaCO, content in Fig. 3. If it is assumed that partide residenoz time in the bed is long (on the order of hours), then there will be maximum utilization of the cakium in the particle, provided that the reztion is not Iimitexl by roe little suIfur (high Ca/S ratio). For exampIe, for Greer limestone, this method wouId predict a cakium utiWation of 57% (80.4% Caco,, see Fig. 2) for a ca/S ratio smaIler than l/O.57 and wouId predict a calcium utikation of I/(Ca/S) for a C&S ratio Luger than I/0.57. This method, for the seven limestones, predicts values of U which are higher than the pilot plant -Its by approximatdy I9 eA (data not included).

Since the TGA czkium utilization in Fig- 3 can be predicted from the calcium

too

90 u=u, ?F lhtuz

so

70

60

50

40

30 0 HlGHsuwuRFuEL

20 ouMT-FuQ

10

0 0 10zo30405060708390190 ’

PmJTPlANl- unu?!ATKk~. % : - .:

carbonate limestone content by the use of a straight line fit to the w the pilot-plant

calcium utiiization can be estimated by aaJust$g the results lower by a factor of - -’

0.842 to take into account the high predictions~ The equations for method 2 are given

belOW”

Ul = o-842 (l-7 - 1.4 [CaCOJ) (3) 472 = cL84~ [I/(cajS)-J WY--y

u = u, ifu, -c u* (w--

u = U2 if U2 < tit (6)

wherq U VI, U, = cakinm utiiization, [CaCOJ = C&OS conten& fraction.

Figure 6 (predictions obtained by method 2, also listed in Table 2) shows the good agreement between the pilot-plant experimental results and the calcium utilizations predicted by eqns- (3)-(6) us&g method 2. The st+dard deviation is 24.3.

The third method also requires only knowledge of the CaCO, content of the limestone. The equation for the third method was developed from data used in the

@it me&x& which reqnires plotting of the instanianeous reaction rate constant k’, as a function of cakium utiIi+ion, U, for each limestone. This ctnve is a straight line except at the beginning and end, Except for very high or low CajS ratios, the fiuid-bed desulftition curve (k vs. CQ and the instantaneous reaction rate curye (K vs. v) intercept on the straight-line portion of the K vs_ U curve_

It was also found that the straight lines for the various limestones were parall& A linear equation was obtained for the intercept as a function of calcium carbonate content. With this and the slope of the curves known, an equation was deveioped

which predicts ca?cium utiition as a function of (I) CaCO, content and (2) the instantaneous rate constark Using this equation in conjunction with the fluid-bed desuI&rizition equation, eqn, (2), allows the pilot-plant calcium utili7ation to be przdicted from only &e limestone calcium carbonate content Ca/S ratio, superfkkl gas velocity, and bed height:

U=l- k i 203 [C&&-j

296

u 1

=cai,l [ k_LE (I -- --e --)I U is then the predicted calcium utilization when K = k.

The comparison between predicted and experimentai results is shown in Fig, 7 (aIs.u in Table f under method 3). Except for one point for Tymochtec dolomite for which a low &j!S ratio of O-95 was us& the correIation is excellent (a standard

daiation of -&3_9 with one point excluded; for all points, a standard deviation of 5.5)_ Predicted results are in good agreement with pilot plant results using all three

methods, However, the predictions are more reliable using method 1 which requires kinetic TGA information and the kid-bed desuIfurization equatioa

A number of methods can be used to predict desulfurizatioo performance of a specific Iimestone in fi uidized-bed combustors for a specified ca/S ratio- One method requires onfy krlowledge of the CaCO, content of the limestone. Another requires knowledge of the &CO, limestone content, the superficial gas velocity, and the ffuidized-bed height Buth methods give good predictions of FBC limestone per- formance Method 1, which is even more reliable, requires detezmination of the cunversion of Ca to CaSO, as a function of time for the particular limestone, using a

thermo_mvimetric anaiyzer. The standard deviation between predicted and experi- mental results was +k2. fn ail of these methods, kinetic data was obtained on slowly cakined Limestones (L-1 ‘A h in 100% CO& Limestones cakined in 20% Co, (in I5 tin) did not correlate with FBC desulfurization results.

The designer of an atmospheric pressure FBC plant may wish to screen

potential Iimestones near a pbnt site Iocation by using methods 2 or 3. Potentially attractive fimestones could then be tested in a TGA for more reliable estimates of their desuifkrization potentials.

This work was performed uuder the auspices of the U.S. Energy Research and Development Admini&&on, We would Eke to thank-k A. Jonke, D,.Webster,

267

and L Bun-is for their support and guidance- We wodd also like to thank K- Jensen for 3is analyticai assistance and J- Simmons the Technical Editor.

REFERESCES

1

2

3

4

5

6

3

8

9 10

11 12 13

G_ J- Vogel, L Jobnsoz~, P- Cmmh@tam 3. Hubbk, S. he, f- Lcae, J- Monta8nq E Nrmes. S. Siegel, G. Smith, R Snyder, S. Saxaa,‘W- Swift, G- Teats, C- Tamxr, I- Wiiscm anti k Jo&e, A Devchpmeni Program on Prmurirc d~i=&BufCb&xsriarx,ERDARqortNaAN~ CEN-lOI& Argonne. Iliinois, JUry 1976. J. E Me&o, Mukiceli Hbxk%k&&d &i&r Daigq Gwsstrktiq aad Trd Progr~ ERDA Rqxnt No- FE-1237-Qi63, Marc& 1976s s- S- Strom, I- E- Dowdy_ W_ C Lapplc J- 8. Kitto, T. P. Stanocb, R I-L Boll and W. L. Sage, Ret- EvaI..iorx of Atmospkri~ Pressure l%&iik&W combrotion A&i&d fo Ekcfrik UtiIity Lczgc S:cfzm Gefwrators_ EPR.X NO- RP 412-l interim report, AlIiancc, Ohio, Febnsay 1976. R L Gali, 1 S. Md and J. S. Wilson, ffui&ed=&d Combustion Test of LuwQuuiizy F&v I_ &hracite RrfuK, Morgantown Energy F&each ccntcr, Morg2altcnsn west V&gini24 hiay x977_ R C- Hake, R R Batmnd, M- S- Nut&, D- D- K&&z, L k Ruth and h:. W- Gregory, Sfudicr of the Pressurized ?%kkc&Bcd Gxzi Gombzrsiion ProcLa, EPA-600/7-76011, Exxon F&scar& and Engineering Company, Linden, New Jersey, Septanber 1976- D.L.I(eairos.D-It~,J.RI-Zamm,S.A.J~~B-W-Lancaster.El.P.O?ridll,CH. Peterson. C C Sun, E F. Swrdrup, E. J. Vidt and W. C Yang, ff.&iize&I&i G.wnhszhz Rocrrr Evahafion, Phase II, Pressurizd E&ii&i-&d GmI Gwnbc~ion Devebpmem, EfA- 650/2-75M7-c wtzxingho= Raearch IabcKatory, Piluburgfi. Pemlsyh& Scptanbu 1975. N- H- Coata (Editor), Fbidixc&&d Gwnbrrsrrbn, M&an Vk&a, 1975, Mitn corporation, 1976s RI)-Harvey_RRFrostandJ.Thnmas, Jr-. Pe~rographii Chamcter&icsandPhysicuIPropehes of Maris_ Ckdks, She& and Iheir Gzhks l&&ted to DesuIfwization of Flrrp Gases, EPA- 65512-71044. Xliinois State Gah@cal Survey, Champaign, Hiinois, September 1973. E P. O’Ndl, Da L. Kcaims and W- F- Kit&, Xhmwchim. Act4 14 (1976) 209- R W- Coutant, J. S. McNuIby, R E Barrett, X J- Carson, R Fii and E H- Lam&a, Investigarbn of the Reacfivtiy of Limestone and Dohnife for Cbpmring SO2 from F7ue Gav, Batteile Memorial Institute Columbus, Ohio, August 1968. A- Potter. Ant &am- .Shz BUZZ-, 48 (1%9) 855. R L Pigfh-ci ad G. Sligcr. Znd m. Uurn. Z’rucexs Des Dt&, 12 (I973) 85_ M. HartIMa and R coughlm, A-tChz J, 22 (1976) 490_