Sorbents for fluidized-bed combustion - ACS Publications

bed, so that the coal and sorbent will mix turbulently in the region above the bed, where the coal burns. Combustion occurs at the relatively low temp...
2 downloads 0 Views 10MB Size
Sorbents for fluidized-bed combustion Researchers are jinding many improved chemical sorbents of SO, that are more effective than natural limestone and dolomite Elmer J. Badin Tennessee Valley Authority Chananooga, Tenn. 37401 George C . Frazier University of Tennessee Knoxville. Tenn. 37996 Atmospheric sulfur dioxide (SO2) is a principal cause of acid deposition, frequently referred to as acid rain. A major source of SO2 is that liberated when coal containing sulfur impurities is burned. One emerging technology for

SO2 reduction is fluidized-bed combustion (FBC) of coal mixed with sorbent. In this article, we review representative literature on the chemical composition and SO2 sorption reactivity of limestone (CaCO,) and dolomite (CaCO,.MgCO,), chemical additives to increase sorption, and a variety of alternative sorbents. This review is based on our work in the chemistry of nonisothermal thermogravimetry (TG) combustion of coal-S02 sorbent mixtures (I), the TG combustion characteristics of highsarbon fluidized-bed cy-

!

clone ashes (2), and the preparation and testing of some new materials as SO2 sorbents for FBC (3). To carry out nonisothermal TG combustion, a stationary sample of coal-sorbent mixture, weighing 50 mg, is heated at a prcgrammed rate of 2 " C h i n , from ambient temperature to 550-600 "C, in a stream of filtered air flowing at 50 mL/ min ( I ) . Many varieties of limestone and some kinds of dolomite have low chemical reactivities, which limit their effectiveness in achieving high levels of SO2

sorption unless chemical and physical improvements to these sorbents are made. An example of improvement in the chemical reactivity of lime (CaO) comes from a contribution by the lime industry. It involves adding limestone to an aqueous solution of sodium bicarbonate (NaHCO,) or sodium chloride before calcination to form CaO. Calcination is simply a process by which limestone is thermally decomposed to form CaO and carbon dioxide (C02) (4).Another technique that provides significant improvements in FBC SO2 sorption by dolomite and in the percentage of complete combustion is recirculation of fly ash (a procedure derived from conventional fixed-bed coal combustion) to an experimental FBC combustor, which increases sorbent reactivity by its reaction with fly ash (5). Chemical sorbency Fundamental SO2 sorption reactions. The chemical sorbency of SO? by limestone occurs in two steps during FBC ofcoal-limestone mixtures (Equations l and 2). The calcination temperatures shown in Equation I for the limestone are those for limestones called aragonite and calcite.

-

CaCO, CaO + C 0 2 (825 "C, 899 "C) CaO

+ SO2 + Vz02

-

(I)

CaSO,

(2)

Calcium sulfate (CaS04) is the main product of SO2 sorbency in FBC. Calcium sulfite (CaSO,) and calcium sulfide (Cas) sometimes form in the reducing zones of the flame, but they are predominantly oxidized to CaSO,. A more reactive SO2 sorbency by dolomite involves half-calcination, forming a bandlike structure indicated in Equation 3 by [CaCO, + MgO] at the halfcalcination temperature for a dolomite. Most of the MgO is not directly converted to sulfate, and small quantities of Mg3Ca(S0,)4are formed (6).

-

+

CaCO,.MgCO, [CaCO, MgO] + CO? (730-760°C)

[CaCO, + MgO] + SO? + 1 / 2 0 ? [CaS04 + MgO] + CO?

(3)

-

(4)

A relatively low combustion temperature for FBC of coal with limestone (about 850 "C) follows from the endothermic reaction shown by Equation I and the exothermic reaction shown by Equation 2. The two reactions tend to be thermally balanced in the region of 850 "C (7). Analogous results apply to FBC of coal with dolomite. Natural carbonate stones. Natural carbonates, such as limestone and dolomite, always contain additional metal and semimetal compounds. For this reason, such stones differ in composi-

tion from CaCO, and CaC0,.MgC03. The results of limestone and dolomite analyses are usually reported as their principal oxide constiruenrs (actual oxides are seldom present) of CaO and MgO, as impurity oxide constituents (including S O 2 , Tior, AIIO,, Fe20r, MnO, Na20, K 2 0 , SrO, and P205), and sometimes as elements such as CI and Fe (8). Natural limestone is usually described as CaCO,. "Average limestone" is characterized as CaCO,. 0.26MgC0, (9),and typical limestone has a similar formula (10). Hence, the term dolomiric limesrone often is used. Natural dolomites may comprise different stones varying from CaCO,. 0.27MgC0, to CaCO,. 0.98MgC0, (8). and one dolomite used as an experimental FBC sorbent is C a C 0 3 . 0.75MgC0, (6). Large differences in SO> sorption reactivity occur among natural limestones and dolomites. A 15-fold range in the initial rate of SO2 sorption at 980 "C is shown by IO different calcined limestones and dolomites, each with a mean particle diameter of 0.0096 cm (11). An example for 26 samples of limestone and dolomite shows a 3.4-fold range in sulfation (sulfate-forming) reactivity. The degree of sulfation vs. corresponding percentages of Na20 impurity shows a trend of linear correlation. Other impurity constituents, such as Fe?O, do not give linear correlations, and pore diameters of 2-16 pm appear best for sulfation (8.12). A 6.5-fold range in maximum conversion to CaSO, is shown for IO kinds of precalcined limestone and dolomite (average particle size is loo0 pm) by TG reaction at 900 "C in a gas stream that contains 0.3% SO?; a correlation shows a trend toward increasing extent of sulfation with increasing MgCO, content (13, 14). There is a 13-fold range in sulfation reactivity via isothermal TG at 850 "C shown in 47 varieties of limestone and dolomite (15). The extent of sulfation, calcium-to-sulfurratio (Ca/S). and particle size are described for three classes of limestone (high, medium, and low reactivity) (16). Additives. Table 1 lists certain additives that increase the SO2 sorption ability of CaCO,, limestone, lime, and dw lomite. The list is based on results of laboratory TG reactions of sorbents at temperatures of about 850 "C. with a stream of synthetic SO?-containingflue gas. Only a few of the results are obtained from FBC: comparisons of a given listing use sorbents of about equal particle sizes. Distinct improvements come about when compounds of alkali metals, such a chlorides of Na and K, are added to sorbents. These improvements are sim-

How FBC w o r k s Fluidized-bed combustion normally involves several steps: A mixture of suitably sized coal and a sorbent is fed into a zone above the metal bed of a combustion chamber within which there are ash particles. Fluidizing air is introduced below the bed, so that the coal and sorbent will mix turbulently in the region above the bed, where the coal burns. Combustion occurs at the relatively low temperature of 850 OC (about 800 OC below the temperature of conventional coal combustion). Sorption of SO2 takes place during combustion, and some additional sorption may take place in the freeboard (the hot region directly above the FBC zone).

ilar to the results of earlier work in which NaHCO, or NaCl was added to limestone to produce more reactive lime (4).They are also similar to the trend of increasing N a 2 0 content, which causes increasing sulfation of natural limestone (8, 12). Alternative sorbents. Table 2 lists examples of alternative sorbents, some of which are analogous to the additives (Table I). Each entry in Table 2 is based on different experimental techniques by various workers, so results for a given entry cannot be compared directly with results for another entry. That is, a limestone used for comparison by one author or researcher may be completely different from a limestone used similarly by another. Comparisons of a given entry are usually made with sorbents of about equal particle size. An example is provided by the MSIMS4 sorbents in the last entry in Table 2, which lists results for comparisons of TG combustion of coal-sorbent mixtures with particle sizes of 2400-3300 pm (-6+8 US.mesh) and other comparisons for FBC using sorbents with particle sizes to 1700 pm (3,37). Figure I summarizes the results of examinations of alternative sorbents. Additional data for these sorbents include equilibrium partial-pressure calculations for oxides as sorbents that show a separation of efficient sorbents (CaO, SrO, Li20, and Na,O) from borderline sorbents (COO, CdO, La20!, MgO, and Mn203) (40). A systematic evaluation concludes that the three most promising are CaAI20,, BaCO,, and BaTiO, (41-42). Reactions for sorption of SO2 by alternative sorbents are generally analogues of Equation 2. These reactions are shown by equations for sorption by CaSi03, BaTi03, CaO na-A1203 in Environ. Sci. Technol.. Vol. 19. NO. 10, 1985 895

justed to total 100%). This composition suggests that the reaction shown by Equation 9 contributes to the formation of MS4. (5) Table 3 lists six groups of approximate relative SO2sorption effectiveness (6) values. Each value of unity is assigned separately to each group. The other values for a given group are then adjusted (7) to that group's value of unity; thus, val(8)

which n equals I or 2, and CuO on Al2O3support (26.30-32. 35).

-

CaSiO, + SO2 + V 2 0 2 CaSO, + Si02 BaTiOl + SO2 + I 1 2 0 2 BaSOo + Ti02 CaO a-A1203 + SO2 + V 2 0 2 CaS04 + a-Al2O3 CUO

-

+ so2 + 1/20>- c u s o 4

Acid-base reactions generally are involved in preparations of alternative sorbents. Acidic oxides, such as S O 2 , Al2O3,and FQO,, and basic oxides, including CaO and MgO, are defined either as Lewis acids and bases or as anhydrides of Br6nsted acids and bases by values of empirical ionic potential and by values of proton transfer acidity constants (43-45). Acid-base reactions are used in preparing portland cement. The cement has been shown to be sorbent component, and other data (47-49) indicate that it is an SO2 sorbent. Portland cement contains mixtures of primary constituents that have empirical formulas of Ca2Si04, 3 C a 0 . S i 0 2 , 3 C a 0 . A120,, and 4CaO'Fe203.A1203 (50); the formation of the first three is described by Equations 9, l l , and 12. The high efficiency of coal ash and fly ash in sorbents (Tables I and 2) is probably attributable to acid-base reactions among basic CaO and acidic moieties of ashes, such as SO2 and A1203. Recirculation can increase the extent of such reactions (5). Acid-base reactions are involved in the preparation of the four sorbents labeled MSI-MS4 (Table 2). Initial ingredients for MSI-MS3, in weight percent, are 65% for Fredonia limestone, 25% for fly ash (main oxide constituents are acidic, consisting of 46.8% Si02 and 21.3% A1203),and 10%for a binder of a clav (M&D ball clav) that contains acidic oxide constituefts of 57.3% SiO, and 27.3% AI,02. MS4 has the same initial ingredie&;except that the binder is 9% M&D ball clay by weight plus I % ammonium-stabilized colloidal Si02 (Ludox AS) by weight. The formation of granules (20 Ib each of MSI-MS4) was carried out in a laboratory scale pan granulator, with a diffuse spray of water for purposes of dampening and adhesion of the sorbent, by the Muscle Shoals Laboratory of the Tennessee Valley Authority. The last step consisted of varying the extent of precalcination after granulation (3). The MS4 sorbent is selected as an example to suggest that acid-base reactions play a role in its preparation. When fully precalcined, MS4 contains oxide constituents of 56.1% CaO. 30.1% S O 2 , and 13.8% A1203 (ad8913 Environ. Sci. Technol..Vol.19. NO.10. 1985

ues of relative effectiveness may be compared only within a given group. Inspection of the four sorbents in the first group reveals that a sorbent of equal weights of calcium aluminate cement and fly ash is 6.9 times more effective than limestone; this result reflects four cycles of regeneration (31). Spent sorbents can be regenerated with reducing agents, usually hydrogen

Acid-base reactions Examples of acid-base reactions are presented below. The products often are identifiable with sorbents and compounds shown by Figure 1 and Table 2. Calcium silicates

-

Ca2Si04 CaO + Si02 (base) (acid) (calcium orthosilicate) CaSiOs 2Ca0 + sio, (base) (acid) (calcium metasilicate) 3Ca0 + Si02 3Ca0. Si03 (base) (acid) (alite) Calcium aluminates; may have nonstoichiometric compositions (46)

+

3Ca0 (base)

-

A1203

(acid)

3Ca0.A1,03 (tricaicium aluminate)

+

A1203

CaO (base)

+

Al2O3 + 2Si02 (acid) (acid)

(acid)

+ SiOn

(acid)

-

-

(10)

(11)

(12)

CaO + AI,O, Ca0.AI2O3 (or CaA1204) (base) (acid) (calcium aluminate) Calcium aluminosilicates; might be nonstoichiometric (46) 2Ca0 (base)

(9)

(13)

2Ca0.AI2O3.SiO2 (gehlenite)

(14)

CaO. Al2O3.2SiOz (anorthite)

(15)

TABLE 1

Sorption-enhancing additive9 Addillve

ReSUltB

NaCl

Increases sorption; functions as combustion catalyst to decrease carbon content of FBC fly ash Increases sorption (by 2 % of compound in CaO); NaCl and

NaCI, KCI. Na,SO,. CaCI,, or

Na2Si0,

CaCI, or MgCh

Fe,03

Aqueous solution of FeSO,. FeSO,. Fe,(SOj),. and

Fe,(S04k

Fe>O, coating. from decomposition 01 Fe(NO,), on a dolomite

KCI are most effective. Na2Si03 is least effective More effective than NaCl (during

Doubles sorption by CaO via

4 % Fe,O, in CaO . _

RelelenEe

17, 18 19

20

22-25

Doubles Sorption rate by CaC03.

26

Increases sulfation capacity of dolomite by 65%

27

1 % Fe on CaCO,

"Additives lor CaCO,. limestone. lime. or dolomite

(H2). carbon monoxide (CO), or methane (CH,). Each cycle of regeneration will cause some loss of sorbent activity; the extent of loss must be measured for each sorbent. Examples of regeneration of a sorbent, such as CaO from CaS04 at 1040-1100 "C. are illustrated by Equations 16-18 (51, 52).

+ + SO:

CaSO, + Hz CaO + HzO + SO2

(16)

CaSO, CO CaO + COz

(17)

4CaS0, 2HzO

+ CH, - 4 C a 0 + + CO2 + 4SOz

(18) Another example involves regeneration of BaTiO, from BaSO, + TiO:, the latter two compounds shown in Equation 6 (31).

+

-

BaSO, Ti02 iCO BaTiO, C 0 2 + SOz

+

(19)

Oxidative wrbency Experimental evidence. Although it is clear that oxidative sorbencv occurs because the oxidation number of sulfur changes from i 4 in SO2 to +6 in sulfate, the chemical mechanisms for SO2 sorption and sulfation are not thoroughly understood. (Equations 2 and 4 show overall reactions.) Catalysis, or acceleration, by impurities in sorbents must be a primary cause of oxidative sorbency. The limited evidence s u p porting this concept is summarized below. Catalysis will involve superposition of two chemical effects. First, catalysis in the oxidation of SOz to SO, (or can be brought about by adding metal oxide constituents, such as oxides of V, Cr, or Fe, to sorbents. Second, StrUcNd rearrangements in sorbents during heating or calcination may cause desirable physical changes, assuming that undesirable sintering factors are absent. An example of desirable changes that may occur is the bandlike structure of half-calcined dolomite (Equations 3 and 4) (6). Catalysis of oxidative sorbency by Na+ is shown by impurities such as Na20, additives of Na compounds, or alternative sorbents that contain Na+ (Tables 1 and 2) (8, 12). An example of catalysis by metal ions is seen in the combustion of pure graphite in the range of 250-590 "C (53).This example gives values of relative catalytic activity-on the basis of unity for Be2+thatareAll' = 3 ; C a 2 + = 4 ; M g 2 + = 6; Na+ = 230; Cu ion = 500, and Mn ion = 86,000 (the oxidation numbers of Cu and Mn are not supplied, presumably because of their variable oxidation numbers). The Na' ion has about 58 times the catalytic activity of the Ca2+ ion. and about 38 times the

activity of the Mg'+ ion (53).A possible explanation is that the Na' ion is hydrolyzed by traces of water to form OH-. which can bring about low-temperature oxidation catalysis. A special case for catalysis by Na' is the additive NaCl (Table I). Because several publications deal with reasons for the effect of NaCl we will not review them here. However, we will suggest that physical changes caused by added NaCl may involve the opening of microvoids through the shrinkage of microcrystals (54. 5.5). Other proposed mechanisms for catalysis suggest that for increased oxidative sorbency in Equation 2 by addition of FezOl or its mixture with steam, the formation of an intermediate of [Fez03.S0,] might function as a catalyst (22, 24). The beneficial effect of fly ash on SO2 sorbency is also suggested as a chemical effect that is perhaps catalytic (31). After oxidative sorbency forms CaSO,, coal ash at 870 OC may chemically reduce CaSO, to form SO2, 02. and high-molecular-weight calcium silicates. X-ray diffraction analysis of these products shows that CaSiO, and

Ca2SiOa a r e absent and that Ca&10,2.H20 is present in a greater amount than Ca&0,7(0H)2 (56). Concerted chemical mechanism. As a step toward defining the chemical mechanism of SOz sorption or sulfation, we postulated a concerted chemical mechanism ( 3 ) . This mechanism explains the simultaneous decreases in SOz and carbon in FBC emissions. The latter is seen in the large carryover of fine particles of carbon in fly ash, shown either by the action of NaCl addition to limestone (17, 18). It is also shown bv a linear correlation we obtained (3j. Fieure 2 reoroduces this correlation as a-value o i the decrease of SOz in emissions, e , vs. corresponding values of the decrease of carbon in emissions, 01 (3). The regression line (correlation coefficient, r = 0.93) is determined by six data points for the four sorbents labeled MSI-MS4, the Fredonia limestone used to prepare MSI-MS4, and a mixture of 66% by weight of Fredonia limestone with 34% by weight of raw fly ash (Table 2) (3). The FBC data were obtained at the Department of Energy's Morgantown Energy Technology

FIGURE 1

Substances screened for their ability to sulfate SO2' Did not sulfate

Sulfated

Simple Oxides NazO CaO SrO BaO

M-30

coo NiO ZnO Ce203 ThOz Bi203

bo3

y203

Xtanates LizTi03 CaTi03 SrTi03 BaTiO,

PbTiO,

Aluminates LiZAl2O4( LizO.AIZO3) Ca3AlZO5(3Ca0.A1203) SrA120, (SrO.AIzO3) BaA120a(BaO'Alz03) Ferrites BaFe,20,9 SrFed& 'Conditions:

Sulfaled

Did not sulfate

CaO cornposities 3CaO. A1203 3Ca0.Ai,01 + 7% Na,O 3CaO.AI;Oj + 1.1% Na20 3Ca0. (SiOzA1203)1 (2 + 14.6% Na20 3Ca0. (SOz.A1203)1,2 i 0.5% NazO 3Ca0.Si02 + 3% NaPO 85% CaO + 10% SiOz + 5% Na20 CaO + portland cement CrYPe 1)

CaO + calcium aluminate cement (Atlas Refcon) Calcium aluminate cement (Atlas Refcon) Calcium aluminate cement (Atlas Lemnite) Calcium aluminate cement (Lonestar Fondu) Ca0, BaC03 CaSi03 BaSiO, BaZrO-

Other materials 3CaO. Si02

TG.870-900 OC, almmpheric pressure, 94.9% Nz + 5 % 0;+ 0.1% SO,

Sources: References 31.38.39

Enviran. Sci. Technol.. Vol. 19. No. 10. 1985 897

Center (METC, Morgantown, W.Va.), where a combustor was operated at a bed temperature of 1125 K (37). The direct proportionality between emissions Seems significant (Figure 2). It demonstrates for the first time that SO2 decrease and carbon decrease must occur in a concerted manner during combustion. The concept of concerted chemical mechanism comes from descriptions of catalysis by semiconductors and the electronic theory of catalysis by semiconductors (57, 58). It also is based on a proposal for equilibrium adsorption of O2 on a sorbent to form the anionradical (.02)-. followed by another equilibrium involving dissociation to the combustion-reactive anion-radical (.O)and the diradical .O. species (59). These equilibria are used in two examples of the concerted chemical mechanism that are outlined; the first specific for the MgCOl constituent in limestone or dolomite, the second dealing more generally with semiconducting impurity oxide COnStiNentS. Equation 20 describes the first example that is specific for the effect of MgCOl either in limestone or in dolomite, which is assumed to form the half-calcined bandlike StrucNre (Quations 3 and 4) in which MgO is a semiconductor (3, 6). ICaCOl + MgOl + O2 [CaCOl+ Mg6+(.02);] [CaCOl + MgO+(.O)- + .0.1

I

[CaSO,

(20)

[cao + M,ob] f 0 2 [cao f M.Ob+('Oz)-

+ MOOb]+ COz + CO

Sorbent

ReWtlS

CaO (precalcined limestone) b'-Ca,SiO, or CaO on SiO?

Doubles limestone sorbency Higher sorbency than CaO

support CaSi03 (Fibreboard powder)

CaSiO, and coal ash (equal weights)

BaTiO, or CaTiO, Cement of CaAI,O, containing oxides of Fe. Si, Ti Cement of CaAI,O, and fly ash (equal weights) Na20 in er-AI,OJ K 2 0 in n-A120,

Reference 29 30

Doubles the sorbency of reagent grade CaO Higher sorbency than CaSiO,

26, 30

Greater sorbency than limestone Greater sorbency than limestone

31 31

Three times the sorbency of

31

CaAI,O, cement alone High sorbency; Na,O offers best performance CaO in a-Alp03, SrO in a-A1203. High sorbency, but lower than Na,O. K 2 0 analogues Ea0 in aAIPOa High sorbency, >80% with pores CaO in AI,O,. via aqueous Ca(NO,), and calcination >0.2 pm High sorbency in FBC SiO2-AI2O3refractory CuO on A120J support High sorbency in FBC CaO with SiOI and Fe203 Excellent sulfation in circulating mullisolid fluidized-bed combustion Chemically improved Up to 2.8 times the sorbency by IimestoneS. limestone (via TG combustion with coal); up to 3 times the sorbency by limestone (via FBC)

26

32 32 33 34

35 36 3

.Label MS1-MS4 limestone (65%). fly ash (25%).and binder 01 clay or clay-colloidal SiO. (10%). precslcined IO varying extents

Alkaline oxldatlve sorbent Cycles of sorption (900 "C)

and regeneration

CaCO,, limestone CaA1204.calcium aluminate cement BaTiO,. barium titanate CaAI2O,. calcium aluminate cement and fly ash (equal weights) Sorption at 900 OC CaC03. reagent CaSiO,, reagent CaSiO,, reagent and coal ash (equal weights) CaC03. reagent and coal ash (equal weights) Sorption at 850 "C CaC03, reagent

-

CaSiO,. reaaent CaSi03. Fibreboard powder Sorption by precalclned limestone

I SOz and C

[CaSO,

Alternative sorbents

_.

+ M.Ob+(.O)- + .O.] I

TABLE 2

Relative SO2 sorption effectiveness values

The second example is for the generalized case. It is illustrated by Quation 2 I , which proposes that the initial functional SO2 sorbent is the [CaO + MOOb] complex in which MOO,, is a semiconducting acidic oxide such as Fe203, Ti02, ZnO, or another impurity oxide. Quation 20 also proposes that acid-base attractive forces form the [CaO + M,,Ob] complex (3).

[CaO

test unit. A first approach for models for FBC can be constructed on the basis of simple steady-state kinetics in which the

TABLE 3

SO2 and C

I + MgO] + C 0 2 + CO

proaches enable predictions of the way in which given sorbents will function with respect to Ca/S and feed rate to meet SO2 emission standards, without

(21)

Models relating SO, sorption data to FBC. For a series of sorbents, l a b ratory TG experiments for SO2 sorp tion are related to corresponding FBC data, the latter obtained by time-consuming operation of relatively expensive large-scale test units. The ap898 Envimn. Xi.Technol..Vol. 19, NO.IO. 1995

Relative effectiveness (1100 "C) 1

Reference

31

2.3 5.5 6.9

1 1.2 1.7

26

3.5 1

26

1.7

23

CaCO,, limestone 1 CaO. precalcined limestone 2 Sorption catalyzed by FelOI CaO. reaaent 1 2 CaO. reaient and 4% Fe203 Sorption by a chemically improved limestone, MS4 CaO, precalcined limestone 1 Precalcined composition of limestone (65%), 3 fly ash (25%), and binder (10%)

29

24

3

first order in SO2 and zero order (the rate that is unaffected by changes in O2 concentration) in 02.Equations in such models contain parameters from TG reactions of sorbent with SO2 in the absence of coal (31, 38, 39, 60-62). A second approach to a model uses a correlation parameter that we introduced as a h for sorbents (where a is the BNnauer-Emmett-Teller [BET] surface area and vis the pore volume) to obtain linear correlations for relative effectiveness of sorbents from FBC tests (3). However, neither approach has been tested sufficiently for use in FBC design. In the first approach, values of sorbent effectiveness that are correlated with corresponding results from FBC test units are briefly outlined for three current models. A Westinghouse model assumes that the gas in FBC is in “plug flow” for both the emulsion phase and bubble phase; isothermal TG data are used to compute a fractional conversion rate for the sorbent (16, 63, 64). The Lee and Georgakis (L&G) model is constructed for the emulsion and bubble phases (14). It assumes that the gas in emulsion phase is well mixed, and that the gas in bubble phase is in plug flow; the use of isothermal TG data gives two parameters that describe the maximum conversion of SO2 and establish a “pore plugging” constant. The Argonne National Laboratory (ANL) model makes the same assump tions for gas flow as does the L&G model (65,66). Isothermal TG data give two sorbent-specific parameters used in an FBC in-bed model either for a “slow bubble regime” or for a “fast bubble regime.” A newer ANL model shows that a significant amount of SO2 sorption occurs in the freeboard (67). Overall, the ANL models seem to be the most comprehensive because they cover in-bed sulfation, the slow- and fast-bubble regimes, and freeboard sulfation. The second approach shows that the previously defined sorbent parameter of a h supplies linear correlations with FBC test data for the relative effectiveness of the sorbent (3). We used METC measurements for the four MS sorbents and for FBC experiments with coalsorbent mixtures (37). The METC data include values of a in m2/g and v in cm’/g-estimating Y quantitatively by a special technique (@-and values of the relative effectiveness of the sorbents’ ability to reduce SO2 in FBC emissions (e) and to reduce carbon in FBC emission ((I) (Figure 2). We also obtained corresponding sorbent values for the relative effectiveness of SO2 sorption during nonisothermal TG combustion, R (3).

Figure 3 gives the linear correlation ( r = 0.97) for a / v vs. corresponding values of e. The regression line is determined by the four MS data points. The outlier is the data point for Fredonia limestone; its inclusion results in a somewhat reduced value of r = 0.91. The Fredonia limestone point is an outlier because its a h was measured for the uncalcined form of sorbent, whereas the MS sorbents had ah, measured for various degrees of calcination. Equation 22 shows a related linear correlation, a h vs. corresponding values of (I for the four MS sorbents; r falls to 0.83 when the outlier is included in the correlaton. (I

= 0.000366(a/v) + 1.07 ( r = 0.97)

(22) A linear correlation similar to that shown in Figure 3 cannot be obtained when values of E are replaced by corresponding values of R. The value of R is determined bv nonisothermal TG combustion of essentially stationary coal-sorbent mixtures at 500-560 “C, probably because the coal in the mixtures passes through a plastic state at about 350 “C. This state may cause particles of sorbent to be enveloped by the coal, thereby restricting the access of

’.

’‘--

.:.. ..

- ~I’



7hese sorbenrs will be analyzed for

ejiciciency of SO2 capture

FIGURE2

SMptiOn effecthrenessof limestone@

1

L

a.value of the relative effmiveness in decreasing camn emissions *Dar.aar in

and c a mmi&ms under FBC mndillw 01 Csls

YPiOlma,rn

-

2.0 and prrcll) size 01

omen1

Sourn: Re(anmca 3

Environ. Sci.Techno1.. MI. 19. No. I O . 1985 899

SO2 to reactive sites on the sorbent. It is believed that at temperaNreS below 350 “C, data for initial stages of SO> sorption and oxidation can be correlated with a h . Other linear correlations are obtained by defining an empirical quantity, A = R(a/v). where R is the relative effectiveness of TG SO2 sorption during the combustion of coal-sorbent mixtures (3). A correlation of values of A vs. corresponding values of E is nearly perfect ( r = 0.997) for the five data points for MSI-MS4 and Fredonia limestone. An analogous linear correlation, A vs. corresponding values of CY for the same sorbents, shows r = 0.92 (3). Uses of a h for sorbents to obtain linear correlations suggest ways of predicting values for the relative effectiveness of a sorbent in an FBC unit. This can eliminate the need for FBC tests once a correlation is established. It also is probable that the use of a h in areas other than combustion, such as catalysis, will be of value for correlations and predictions. Selecting sorbents A scientifically sound foundation has been established for the use of a large variety of chemically improved limestone and alternative SO2 sorbents in FBC of coal. Although such sorbents have never been selected for use in first-generation FBC pilot plants or small operational FBC facilities (these plants have always used natural limestone), improved and alternative sorbents are available to be considered for application in potential second-generation FBC plants. The analysis of all the factors that govern the selection of a particular sorbent for a given FBC application is well beyond the scope of this review. Nevertheless, we will briefly mention some of these considerations. Cost is of primary importance. The least expensive method of improving sorbents is the use of a small quantity of an additive such as Na2C03. The cost of generating electricity in an FBC plant operating at atmospheric pressure with a sorbent of regenerable portland cement is estimated (in 1980 dollars) to be 12% less than that in a corresponding FBC plant using once-through limestone. The cost is about 17% less than that of a conventional coal-fired plant using wet limestone scrubbers, and 12% less than that of a plant with a scrubber that uses regenerable MgO (48). Expenses are also lowered when any of the alternative sorbents or Na2C0, can be used to cause a decrease in carbon emissions. Other considerations for selecting the optimum sorbent include the overall cost for transportation, processing, and 9W Environ. Sci. Twhnol.. Vol. 19. NO. 10, 1985

FIGURE 3

Relationshipof decrease In SO, ernisslons to particle slze of sOrbenr

Surfacearea divided by pore volume. &

pulverization and for the use of fine particles of sorbents. SO2 sorption characteristics and Ca/S also must be considered. Other factors include the effect of sorbent on NO, emissions, the attrition of sorbent, regenerability, and the costs of regeneration. The erosion of combustor components during combustion, the degree of recirculation of fly ash to the combustor (5).the reuse or disposal of spent sorbent, and the environmental consequences of the use and disposal of spent sorbent also must be taken into account. Fortunately, there is an extensive body of literature that addresses many of these considerations. Acknowledgment This work was carried out under Contract TV49235A by the Tennessee Valley Authority. Division of Energy Demonstrations and Technology (Chattanooga, Tenn.), with the University of Tennessee (Knoxville). Before publication. this article was reviewed for suitability as an ES&T feature by Ralph T. Yang, Stale University of New York at Buffalo, Amherst, N.Y. 14260; Arthur M. Squires, Virginia Polytechnic Institute and State University. Blacksburg, Va. 24061; and Alexander Weir, lr., South-

ern California Fdison Company, Rosemead. Calif. 91770. References ( I ) Frarier, G. C. et al. Fuel 1982, 62, 1225. ( 2 ) Frarier, G. C.: Mason, C.; Badin E.J. Fuel 1984, 63. 499. ( 3 ) Frazier. G. C.: Badin. E. 1. Chem. Eng. Commun. 1983. 23, 259. (4) Boynton. R. S. “Chemistry and Technalogy of Lime and Limeslone.” 2nd ed.; Wiley: New York. N.Y.. 1980; pp. 179-80. ( 5 ) Cox, D. G.: Highley. J . ”Reduction of Atmospheric Pollution. Main Report.” DHB 060971: National Coal Board: London. U.K., 1971;p.A1.124. (6) Hubble. B. R. et al. In “Proceedings of the Fourth International Conference on Fluidired-Bed Combustion”; MITRE Corp.: McLean. Va.. 1976; p. 367. (7) Makansi. J.: Schweiger. B. “FluidizedBed Boilers. A Special Report”: Power. 1982. p. S-I. (8) Harvey, R. D. “Petrographic and Mineralogical Characteristics of Carbonale Rocks Related to Sorption of Sulfur Dioxide in Flue Gases. Interim Reporl”: Illinois State Geological Survey: Urbana. 111.. 1970. (9) Mason. B. “Principles of Geochemistry.” 3rd ed.; Wiley: New York. N.Y.. 1966: p. 153.

( I O ) Lea. F. M . “The Chemistry of Cement and Concrete.’’ 3rd ed.: Chemical Publishing: New York. N.Y.. 1970; p. 21. (11) Borgwardl. R.H.; Harvey. R. D. Envi,on. Sci. Techno/. 1 9 2 . 6. 350. (12) Harvey. R . D.: Steinmeti. J. C. “Petrographic Properties of Carbonate Rocks Re-

lated to Their Sorption of Sulfur Dioxide.” Environmental Geology Notes. Number 50: Illinois State Geological Survey: Urbana. Ill., 1971. (13) Vogel. G. 1.; Johnson, I.; Lee. S . H. “Supportive Studies in Fluidized-Bed Combustion. Quarterly Report, January-March 1977,” ANLIES-CEN-1019 and FE-1780-7; Argonne National Laboratory: Argonne. 111.. 1977, p. 31-34. (14) Lee. D.C.; Georgakis, C. AlChE J. 1981.27.472. (IS) Shearer, J. A,; Johnson, 1; Turner, C. 8. “Effect of CaC&Additive on the Reaction of S 0 2 0 2 Mixtures with Carbonate Rocks,” ANLICENIFE-79-7; Argonne National Laboratory: Argonne. Ill., 1979. (16) Newby, R. A,; Ulerich. N. H.: Keairns. D. L. In “Proceedings of the Sixth International Conference on Fluidized Bed Combustion”; Office of Coal Utilization. Department of Energy: Washington. D.C.. 1980; Vol. 3. p. 803. (17) Ehrlich. S. In “Institute of Fuel Sympcsium Series No. 1”; London, U.K., 1975; Vol I . p. C4-I. (18) Pope, M. In “Roceedings of the Fourth International Conference on Fluidized-Bed Combustion”; MITRECorp.: McLean, Va., 1976; p. 123. (19) Jonkc. A. A. et al. In “Roceedings of the Fluidized Bed Combustion Ttchnology Exchange Workshop.” CONF-770447-P. MITRE Corp.: McLean, Va.. 1977; Vol. 2, p. 343. (20) Johnson, 1. et al. “Support Studies in Fluidized Bed Combustion-I978 Annual Report, ANLICENIFE178-IO; Argonnc National Laboratory: Argonne. Ill., 1979. (21) Chopra. 0. K. et al. In “Proceedings of the Sixth International Conference on Fluidized Bed Combustion.” CONF-8W428; Courtesy Associates: Washington. D.C., 1980; Vol. 2. p. 496. (22) Yang. R. T. et al. “Regenerative Process for Desulfurization of High Temperature Combustion and Fuel Gases,” BNL 50706; Brookhaven National Laboratory: Upton, N.Y.. 1977; p. 8. (231 “Summarv Evaluation of Atmosvheric Pressure Fluidized Bed Combustion Applied to Elcrtric Utilit) Large Steam Generators.” EPRI FP-308; Electrtc Poucr Research lnstitutc Palo Alto. Calif.. 1976: Vol. I . pp.

1978,322, 17. (35) Demski, R.1. et al., presented at the I982 Summer National Meeting of the American Institute of Chemical Encineers. Cleveland, Ohio, September I Y 8 2 (36) Nack, H.; Liu. K. T.; Felton, 0. W. in “Proceedings of the Fifth International Conference on Fluidized-Bed Combustion”; Banellc Columbus Laboratory: Columbus, Ohio, 1978; Vol. 3 , p. 223. (37) Grimm, U. “Evaluation of the Sulphur Sorption Properties of Chcmrcally -Im. prwcd Lirncrtoncr During Fluidized-Bed Combustton.“ DOEIMETCIRI- 128. Department of Energy, Morgantown Energy Technology Center: Morgantown, W.Va., 1981. (38) Ruth. L. A. In “Proceedings of the Fluidized-Bed Cornbystion Technology Exchange Workshop. CONF-770447-P. MITRE Corp.: McLean. Va., 1977; Vol. 2. p. 301. (391 Lowell. !F S.; hr san. T B. “Identifica’ tion of Regeneible Mekl Oxide SOI Sorbents for Fluidized-Bed Coal Combustion.’‘ EPA-65012-75-065; Environmental h t e c tion Asencv: Washineton. D.C.. 1975. (40) L i t k i.L. “Suliur Removal in Fluidized-Bed Combustion.” MERCITPR-7711; Department of Energy, Morgantown Energy Research Center: Morgantown. W.Va., 1977. (41) Newby. R. A,; Kcairns. D. L. ”Alternatives to Calcium-Bad SO, Sorbents for Fluidized Bed Combustion: Conceptual Evaluation.” EPA-60017-78-005: Environmental Protection Agency: Washington. D.C.. 1978. (42) Newby, R. A,: Keairns. D. L. In “Proceedings of the Fifth International Canference on Fluid& Bed Combustion”; MITRE Cow.: McLean, Va., 1978; Vol. 2. p. 680. (43) Voms, K. S. In “Analytic Methods for Coal and Coal Products”; Karr, C., Jr., Ed.; Academic Press: New York, N.Y., 1979; Vol. 3, pp. 481-87. (44) Vorres. K. S. 1. Eng. Power 1979. 101, 497. (45) Badin. E. 1. “Coal Combustion Chcmisxu**;Elsevier: New

-~~~~ ~. ~~

TRE Corp.: McLean, Va.. 1977; Vol. 2, p. 285. (55) Canner, L.L.; Setesak, S.E. In “Procccdings of the Fifth International Conference an Fluidired Bed Combustion”; MITRE Corp.: McLean, Va.. 1978; Vol. 2. 1978; p. 163. (56) Yang, R. T.; Krishna, C. R.; Steinberg. M. I d . Eng. Chcm. Fwtdom. 1977, 16, 465. (57) Law. 1.T. In “Semiconducton”; Hannay, N.B., Ed.; Reinhold: New York, N.Y., 1959; pp. 696-703. (58) Wolkenstcin, Th. In “Advances in Catalysis and Related Subjects”; Eley, D. D.; Selwood, P. W.; Weisz, F! B.. Eds.; Academic Press: New York, N.Y. 1960. Vol. XII. pp.189-264. (59) Badin. E.J., MITRE Corp., personal communication, 1976. (60)Snyder, R.; Wilson, W.I.; Johnson, 1. 7hermochim. Acta 1978.26, 257. (61) Vogel, G.J.; Johnson, 1.; Lee. S.H. “Supponve Studies in Fluidid-Bed Combustion, ANLICENIFE-77-8; Argonne National Laboratory: Argannc. 111.. 1977. (62) O’Neil. E. F!; Ulerich, N. H. “Criteria for the Selection of SO2 Sorbents for Almospheric Pressure Fluidid-Bed Cambustors.” EPRI-FP-1307; Westinghouse Electric Corporation: Pittsburgh, Pa., 1979; Vols. I and 2. (63) Newby. R. A. et al. “Effect of SO2 Emission Requirements an Fluidized-Bed Combustion Systems: Preliminary TechnicaIlEconomic Assessment,” PB-286 971; Westinehouse Electric Cornration: Pittsburgh,Pa., 1978. (64)Ulerich. N. H.; Newby, R. A,; Kcairns. D. L. 7hhernwchim. Acta 1980.36, 1. (65) Fee. D.C. et 81. “Sorbent Utilization ’ Prediction Methodology: Sulfur Control in Fluidized-Bed Combustors.” ANLlCENl FE-80.1% Argonne National Laboratory: Argonne, 111.. 1980. (66) Fee, D. C. et al. Chem. fig. Sci. 1 9 Q 38. 1917. (67) Fee, D. C. et al. Chem. Eng. Sci. 1984, 39. 731. (68) Robem. B. F. 1. Colloid Inrcrfacc Sei. 1%7,23,266.

*-.),I “-A”.

(24) Yang, R. T.; Shen, M. S.; Steinberg, M. Environ. Sei. Echnol. 1978, 12. 915. 1251 Yam. R. T Fuel 1978.57. 709. for D&ulfurization of kigh Temperature Combustions Fuel Gases.” BNL 50809; Brookhaven National Laboratory: Upton.

,.

Y

I”,,

fo; FBC.” BNL 50992; Brookhaven National Laboratory: Upton. N.Y.. 1978; p. 12. (29) Ulerich. N. H.; ONeill, E. F!; Keairns, D. L. “Influence of Limestone Calcination on the Utilization of the Sulfur-Sorbent in Atmospheric Pressure Fluid-Bed Combust o n . Final Report,” EPRI FP-426: Electric Power Research Institute: Palo Alto. Calif., 1977. (30) Steinberg. M.; Yang, R. T. “Regcneralive Process for Desulfurization of High Temperature Combustion and Fuel Gases. Quarterly Report No. 9,” BNL 50891; Brookhaven National Laboratory: Upton, N V 197s .....,

(31) Ruth. L. A.; Varga. 0 .M., Jr. Environ.

Sei. Technol. 1979. 13.715. (32) Snyder, R. B. et al: 1. Air Pollur. Conrrol Asroc. 1!?77,27, 975. (33) Vogel. G. 1.; Jonke, A. A.; Snyder, R.B. U S . Patent4.091.076, 1978. (34) Gibson, J.; Highley, J. VDI-Berichre

21; Argonne National Laboratory: Argonne, 111.. 1977; p. 185. (47) Albanese. A. S.; Sethi. D. S. “Regenerative Process for Desulfurization of High Temperature Combustion and Fuel Gas&. Progress Report No. 14.’’ BNL 51223; Brookhaven National Laboratory: Uptan, x, 2..

..,

Y

,“Q,n

170”.

(48) Albanese. A. S . ; Sethi. D. S. “Regenerative Prccess for Desulfurization of High Temperature Combustion and Fuel Gases. Progress Report No. 15,’’ BNL 51235; Brookhaven National Laboratory: Uptan. N.Y., 1980. (49) Yoo, H. J.; McGauley. P: 1.; Steinberg, M. “Calcium Silicate Cements for Desulfurization of Combustion Gases.” BNL 51430: Brookhaven National Laboratory: Upton. N.Y.. 1981. (50) Bailey. R. A. et al. “Chemistry of the Environment”; Academic Press: New York, N.Y., 1978; pp. 4 6 - 4 7 , (51) Montagna. J. C. et al. In “Roccedings of the Fourth International Conference on Fluidized-Bed Combustion”; MITRE Corp.: McLean, Va., 1976; pp. 393423. (52) Ruth, L.A. In “Proceedings of the Fourth International Conference on Fluidized-Bed Combustion”; MITRE Corp.: McLean. Va.. 1976; pp. 425-38. (53) Amariglio, H.; Duval. X. Carbon 1966, A., V ?

(54) Gamer, L. L. In “Proceedings of the Fluidized Bcd Combustion Technology Ex-

change Workshop.’’ CONF-770447-P. MI-

d

I

EInrcr J. Badin (I.) is the author of “Coal Combuslion Chemistry-Correlation Aspecrs.”published by Ekevier Science Publishers in 1984. H e has an M.A. and a Ph.D. in organic chemistryfrom Princeton University. Badin has been a member of the Princeton University chemistry faculty and c h a i r m of the chemistry department (u Union College (Cmford. N.J.). He has recenrly retired after jive years with the Tennessee Valley Authoriry.

George C. Fnnier (c) is the Condra Distinguished Professor of Chem’cal Engineering (u the University of Tennessee. where he has taught since 1968. H e received his doctorate in chemical engineering from the Johns Hopkins Universiry. Frazier has written numerous articles and technical reports and does extensive consulting work in the U.S. and abroad. Envimn. Sci. TBCIInoI.. MI. 19, NO.10, 1%

801