United States Environmental Protection Agency Air and Energy Engineering Research Laboratory Research Triangle Park NC 27711 Research and Development EPA/600/S2-88/069 Mar. 1989 x°/EPA Project Summary Fundamental Studies of Dry Injection of Calcium-Based Sorbents for SCte Control in Utility Boilers G. H. Newton, D. K.Moyeda, G. Kindt, J. M. McCarthy, S. L Chen, J. A. Cole, and J. C. Kramlich A research program was con- ducted to determine the mecha- nisms which limit the extent of reaction between sulfur dioxide (SOa) and calcium-based sorbents [CaCOa and Ca(OH)2l by measuring the in situ physical structure and reactivity of sorbent injected into a combustion environment for resi- dence times as short as 35 ms. Four models of the sulfatlon reaction were used to guide the research and interpret the data. The extent of sorbent utilization was found to be limited by porosity losses during the sorbent activation process. In-sltu porosities a fraction of that theo- retically possible were measured In the absence of SO2- At temperatures below 1000°C, this porosity loss was determined to be caused by COa* actlvated sintering. The presence of SO2 during calcination reduced the extent of porosity loss and at optimal temperature sulfation conditions no loss in porosity was observed. At temperatures above 1200°C, porosity losses may result from an increased rate of thermal sintering or a decrease in the rate of the sulfatlon reaction. Calcines from CaCOa suf- fered greater losses in porosity than those from Ca(OH>2 which, along with the larger CaCOa particle size, accounts for the substantial differ- ences in SO2 capture between these two sorbents. This Project Summary was devel- oped by EPA's Air and Energy Engineering Research Laboratory, Research Triangle Park, NC, to an- nounce key findings of the research project that is fully documented In a separate report of the same title (see Project Report ordering Information at back). Introduction Upper furnace injection of calcium based sorbents to adsorb SOa has been studied extensively in recent years. A complete understanding of the funda- mental physical and chemical processes which occur at furnace temperatures has not, however, been achieved. Previous studies of the sulfation process and well established theories have allowed the rates of external mass transfer, pore diffusion, and product layer diffusion to be calculated. These processes coupled with the chemical reaction, the unknown physical structure of a calcined sorbent, and the extremely high rate of the overall process (the reaction is primarily over within a few hundred milliseconds) make it difficult to determine which of these mechanisms controls the extent of sorbent utilization. Other studies have determined the physical structure of calcined sorbent but only at times much longer than those available in full scale utility boilers. This program was designed to determine the physical structure of calcined sorbents at these early times so ------- that the mechanisms that limit sorbent utilization can be understood and higher levels of S02 capture achieved. Experimental Facilities Fundamental experiments were con- ducted in three research facilities: the Isothermal Reactor (ITR); the Short Time Reactor (SIR); and the Controlled Temperature Tower (CTT). Firing rates in these three facilities range from 9.4 kW for the ITR to 26.4 kW for the CTT. The ITR is an electrically heated drop-tube furnace down-fired by a premixed flat-flame burner. Sampling residence times normally range from 100 to 500 ms with isothermal operating temperatures of 800-1500°C. The STR is a back-fired furnace which operates isothermally at temperatures of 930- 1370°C with residence times of 35-400 ms. The STR can operate either isothermally or under quenched condi- tions and can fire coal as well as gaseous fuels. The CTT is a back-fired reactor designed to simulate the time- temperature histories of a wide variety of coal-fired boilers. The CTT can also fire either gas or coal. The ITR was used to obtain data on SC>2 capture as a function of time, temperature, and sorbent type. This information was supplemented with data obtained from the CTT on the influence of quench rate on S02 capture. To determine the in-situ physical structure of calcined sorbent, solids were sampled from the STR as a function of time, temperature, SC>2 concentration and sorbent type. A CO flame was used during solid sampling to allow sorbent to be obtained without the presence of H20 in the sampling system while still providing a sulfation environment similar to that in a coal or natural gas fired furnace. Surface area, porosity, and pore size distribution were determined for sorbent collected from the STR. Modeling The complexity of the sulfation reaction requires that models of the sulfation process be used to define which process controls the extent of sorbent utilization. Four sulfation modeling approaches were considered in this program: a grain model, a pore tree model, and two distributed pore models. Each model assumes that sorbent particles are spherical and fully calcined prior to the onset of sulfation. In addition, an activation model of the calcination reaction based on the grain model was utilized. The activation model was developed to simulate simultaneous calcination and surface area loss. It considers CaCOa decomposition at the CaO/CaCOs inter- face, diffusion of CC>2 through the CaO to the particle surface, diffusion of CC>2 from the particle surface to the bulk gas, and continuous, finite rate surface area loss for the calcined material (sintering). The calcination process is represented by a spherical, shrinking core model with the intrinsic calcination rate dependent only on the chemical rate. The sintering rate is assumed to depend on CC>2 concentration and surface area. All of the sulfation models include the following sequential mechanisms: exter- nal mass transfer; pore diffusion; solid state diffusion; heterogeneous chemical reaction; and product layer buildup on the internal surfaces. Each model, how- ever, view the physical structure of the calcine differently. The grain model treats the CaO particle as an agglomerate of CaO grains whose distribution of sizes is set to match the measured BET surface area. The pore tree model describes the sorbent pore structure as a set of trees of various sizes whose trunks are located at the particle surface. The size distribution of the pores is proportional to 1/rp3, where rp is the pore radius. Two versions of the distributed pore model were considered: one viewed the pore structure as being made up of an interconnected network of pores, and the other viewed the pore structure as having non-intersecting pores. Both distributed pore models use experimentally deter- mined pore size distributions. Results Influence of Temperature Capture of SOa for sorbents sulfated in the ITR and for the Linwood hydroxide sulfated in the STR fired with a CO flame shows that a maximum in SO2 capture occurs between 1100 and 12008C. The activation and grain models indicate that the surface area of a sorbent will increase to a maximum immediately upon calcination and then rapidly de- crease to an equilibrium value within a few tenths of a second. The surface area of the Linwood hydroxide sampled from the STR, both with and without SO2 present, does not show this. The surface area of these samples did not significantly vary with either time or temperature. The measured porosity of these solids, plotted in Figure 1, also did not vary with time or temperature and are much less than the porosity theoretical! possible from Ca(OH)2 (0.49). The porosity of the calcined sorbei sampled from the STR with S02 preset has, however, been reduced from il original value by the buildup of th CaS04 product layer within the por structure. These original porosities, whic can be calculated when the extents < sorbent utilization are known, are plotte in Figure 2. A maximum in calculate original porosity is observed betwee 1100 and 1200°C which is nearly equ; to the theoretical porosity of Ca(OH)2. / lower and higher temperatures th calculated original porosities were les than the theoretical value. A compariso of these calculated original porosities t the measured porosities without SO present (Figure 1) reveals that when SO is present the extent of porosity loss fror the theoretical porosity is either reduce or prevented. This dependency of the calculate original porosity on temperature and th presence of S02 allows a new hypothesi to be proposed: The process which causes this porosit loss (sintering) occurs at approximatel the same temperature as the sulfatio reaction. Porosity loss and sulfatio may therefore be viewed as processe which compete for available sulfatio sites. When the mechanism whic causes porosity loss is fast compare to sulfation, the loss in porosity result in low sorbent utilization. Whe sulfation is fast, porosity loss does nc occur and greater levels of sorber utilization result. The dependence of porosity loss o temperature could be caused by either c two scenarios. At low temperature (~970°C) the sulfation reaction is sky due to kinetic and diffusional limitation; and a large loss in porosity occurs. / intermediate temperatures (~1180°C the sulfation reaction is fast enough t prevent porosity losses. At highe temperatures (>1200°C) the increasin rate of CaS04 decomposition reactio results in a net decrease in the rate of th overall sulfation reaction, and porositie again drop. The most likely mechanism respor sible for this porosity loss is eithe thermal or C02 activated sintering. EPA' B. K. Gullett conducted a similar set c experiments in an electrically heate nitrogen (CO2 free) flow furnace c 1000°C. The measured porosities (with- out SO2) and the calculated origin* porosities (with S02) showed n ------- 1* I! oi o !«• x 5 0.25 0.2 0.1 0 0.2 0.1 0 0.2 0.1 •o- 970°C J270°C Linwood CO Flame Q w/o SO2 _ I 0.1 0.2 0.3 Time (sec) 0.4 0.5 Figure 1. Measured porosities of sorbent sampled from the STR. significant drop in porosity from the theoretical porosity of a calcined hy- droxide within the first 500 ms. Porosity losses therefore occur only in the presence of COg, indicating that CCV activated sintering is responsible for porosities less than the theoretical value (at the low temperature range). The second scenario describing the dependence of porosity loss on temper- ature is that at the lower temperatures (970 °C) the high rate of COa-activated sintering is responsible for porosity losses. At the intermediate temperatures (~11808C) the rate of COa-activated sintering slows, and porosity losses no '•>nger occur. At higher temperature inge, porosity loss is caused by an increasing rate of thermal sintering rather than a decreasing rate of sulfation. Data are not currently available to determine which hypotheses is correct. A modeling approach was developed based on the data obtained in this study. Pore size distributions from the earliest time sampled and an average calculated porosity at each temperature were used as inputs for the interconnected distributed pore model. The results, plotted as solid lines in Figure 3, match the data extremely well. To predict SOa capture at the highest temperature (1270°C), the equilibrium concentration of 802 above CaSC^ had to be included in the model, as indicated by the dashed line. Influence of Mixing Rate Mixing of sorbent is known to influence the extent of 862 capture. A venturi throat in the ITR provides a completely mixed stream within 7-10 ms, while a straight throat doesn't provide complete mixing until 24-40 ms. Use of the faster mixing throat results in significantly increased levels of SOa capture. It was concluded that slower mixing in an isothermal furnace affects 862 capture by: (1) Delaying the onset of sulfation. (2) Exposing the sorbent to a distribution of temperatures. When mixing is slow, eddies within the sorbent/air jets may obtain temperatures between the jets' initial temperature and the furnace temperature. While at these intermediate temperatures the sorbent will calcine and obtain an original porosity (see Figure 2) different than sorbent which calcines at the furnace temperature. The distribution of original porosities will result in a lower average porosity and lower average sorbent reactivity. (3) Allowing incomplete mixing to occur. Mixing which is not complete at the end of a furnace may result in eddies with high levels of sorbent (and depleted of 862) and eddies containing low levels of sorbent (and high levels of The ITR fitted with the venturi throat has extremely rapid mixing and none of these phenomena are likely to influence SOa capture. When fitted with the straight throat, factors (1) and (2) above affect capture. Influence of Quench Rate SOa capture at higher quench rates result from two mechanisms: (1) The time available at temperatures where sulfation occurs is reduced. (2) A finite mixing time exposes the sorbent to a distribution of temperatures. Calcination occurs at a range of temperatures, and a distribution of original porosities results. The lower average sorbent porosity leads to lower sorbent reactivity and lower 802 capture. This effect also results in a shift in the temperature where the maximum 802 capture occurs. When this second mechanism was included in the interconnected distributed pore model, it is able to account for the ------- 0.5 0.4 0.3 I 0.2 0.1 ™ Theoretical Porosity Calculated Original Porosity w/SOz Measured Porosity w/o SO2 -318-369 ms I 900 1000 1100 1200 Temperature (°C) Figure 2. Measured porosities and calculated original porosities. 1300 1400 differences in SOg capture due to quench rate. Influence of Sorbent Type Ca(OH>2 from different commercial sources are known to vary in their ability to capture 802- Capture by a Mississippi hydroxide in the STR is significantly less than capture by the Linwood hydroxide. The calculated original porosities for the Mississippi hydroxide, plotted in Figure 4, are less for those than for the Linwood hydroxide. Interconnected distributed pore model predictions indicate that the differences in capture by these two hydroxides result from the slightly larger particle size of the Mississippi Ca(OH)2 and from the greater porosity loss it experiences. The difference in S02 capture between carbonates and hydroxides was investi- gated in the STR with a CO flame. It has been hypothesized that lower levels of capture by carbonates result from their large particle size. Calculation of their original porosities, plotted in Figure 5, reveals that carbonates are subject to greater losses in porosity than hydroxides. Predictions by the inter- connected distributed pore model indicate that lower SC-2 capture by carbonates results (approximately) equally from their larger particle size and their greater porosity loss. Analysis of the porosity data also indicates that those portions of the carbonates which calcine most rapidly have low original porosities while those fractions calcining more slowly have the theoretical porosity of a calcined calcium carbonate. Influence of Physical Structure Average pore diameters for the Linwood hydroxide sampled from the STR without S02 present, varied as a function of temperature but not as a function of time. Pore size distributio corrected to a constant porosity a plotted in Figure 6 ofor the Linwoi hydroxide. Above 80 A, the distributio are about the same while at below 80 the distributions are scattered. determine the extent that the: differences make in SOa capture, tl pore size distributions in Figure 6 we entered in the interconnected distribut pore model. The results of the; calculations, presented in Figure indicate that the observed differences pore size distributions have litt influence on S02 capture. To furth determine the influence of pore si, distribution on sulfation pore si. distributions from the Linwood sorbe sampled at 970°C and 1270°C, tl Vicron 45-3 carbonate, the Mississip hydroxide, and the Fredonia hydroxi* sampled from a nitrogen reactor 1000°C were corrected to a consta porosity and used as inputs for tl model. The predictions exhibit no mo than 6% variation in SC>2 capture. Tl variations in sorbent pore size distribute resulting from different sulfatic conditions therefore have no significa effect on SC>2 capture. Conclusions The goal of this program was determine the mechanism(s) which tin calcium utilization. This was accor plished by investigating the relationsh between sorbent physical structure ai various sulfation parameters. In-si sorbent physical structure was dete mined by sampling from the Short Tin Reactor (STR), an isothermal react fired with a CO flame. The use of a C flame allowed sorbent to be sampled the absence of H2O which was shown degrade calcined sorbent durir sampling. Investigation of the influence injection temperature on SO2 capture ai sorbent physical structure revealed th calcination of sorbent in a combustic environment resulted in porositi< dramatically less than theoretical possible - 0.49 for Ca(OH)2. Tt presence of S02 was found to reduce tt extent of or completely prevent porosi loss, depending on the injectic temperature. The cause of this porosi loss was determined to be CO activated sintering. A hypothesis wj proposed based on this information: Porosity loss and sulfation may I viewed as processes which compe for available sulfation sites. When tl mechanism which causes porosity lo ------- 60 50 3 40 I Cj 30 20 10 Linwood CO Flame Ca/S = 2 t t -1400 ppmS02 1180°C Equilibrium Concentration of SOi Above CaSOt 0.0 0.05 0.10 0.15 0.20 0.25 0.30 Sorbent Residence Time (sec) 0.35 0.40 cining at longer times obtained the theoretical porosity of a carbonate (0.54). Modeling indicated that, although average pore diameters varied as a function of temperature (but not as a function of time), the observed variations in pore size distribution had little effect on SO2 capture when corrected to a constant porosity. The use of pore size distributions from carbonates and from a sorbent sampled from a nitrogen environment did not change the predicted levels of SOa capture. Figure 3. Interconnected distributed pore model predictions based on calculated original porosities. is fast compared to sulfation (at low temperatures, ~970°C), the loss in porosity results in low sorbent utilization. When sulfation is fast (~1180°C), porosity loss does not occur and greater levels of sorbent utilization result. At higher tempera- tures (>1200°C) the increasing rate of CaS04 decomposition reaction results in a net decrease in the rate of the overall sulfation reaction and porosities again drop. It was also noted that, at the higher temperature range porosity losses may result from an increased rate of thermal sintering rather than a decrease in the rate of the sulfation reaction. A distributed pore model, which used the measured pore size distributions and these porosities, was able to adequately predict the observed SO2 capture without the use of adjustable parameters. Mixing was found to influence BO2 capture by (1) delaying the sulfation reaction, and (2) altering the extent of porosity loss by changing the sorbent thermal history. In an isothermal environment with slow mixing, transient temperatures (between a sorbent's initial temperature and the ambient temper- ature) may occur resulting in a porosity loss not directly related to the injection temperature. In a quenched environment, mixing which occurs over a period of time will expose the sorbent to a distribution of temperatures and a resulting distribution of porosity losses. This results in (1) an average porosity different from the isothermal case (for a given injection temperature) and a resulting different capture and (2) a shift in the temperature where maximum SOg capture occurs. The extent of porosity loss and sorbent particle size were found to be the primary factors which determine sorbent reactivity. Different hydroxides experi- enced significantly different levels of porosity loss, and carbonates suffered dramatically greater losses in porosity than did hydroxides. The degree of porosity loss in carbonates varied with time. Those portions of a carbonate which calcined under 35 ms had a porosity equal to that of a carbonate calcined without SC>2 present (the lowest porosity possible), while portions cal- ------- 0.5 0.4 0.3 0.2 0.1 0: Theoretical Porosity Linwood ft 1BO°C) •o—. <1160°C> 1400ppmS02 CO Flame 0.1 0.2 Time (sec) 0.3 0.4 Figure 4. Calculated original porosities of two hydroxides. ------- 0.6 0.5 0.4 g 0.3 §. 0.2 0.1 0.0 0 III Theoretical Porosity Projected fj Calculated -f^ Original (w/SOz A£ N r |£ £ Measured (w/o SOii Vicron 45-3 (1 1 urn) III 0 0.1 0.2 0.3 Time (sec) • • • • • • i Theoretical Porosity ["] Projected _ JJ* • ^VO Calculated Cj Original (w/SOz ' ^ t^N Measured (w/o SOi) 7fjm Vicron ill 0.4 0.1 0.2 0.3 Time (sec) 0.4 Figure 5. Measured porosities of solids sampled without SO2 and calculated original porosities of solids sampled with SOi corrected for extent of calcination. The projected original porosities were calculated by assuming that the uncalcined portion of the collected sorbent would have a porosity of 0.54 upon complete calcination. 15 W 10 t= 35ms t= 69ms t=110ms t=369ms I I I I Linwood 1180°C I I I I I I I II i I I I I I I 100 Pore Diameter (A) 1000 Figure 6. Pore size distributions corrected to a constant porosity. ------- Linwood, STR CO Flame. 1180°C 1400 ppmSOz Ca/S = 2 0.2 Time (sec) 0.3 0.4 Figure 7. Predictions of the interconnected distributed pore model based on pore size distributions from Figure 6. ------- G. H. Newton, D. K. Moyeda, G. Kindt, J. M. McCarthy, S. L. Chen, J. A. Cote, „„« J. C. Kramlich are with Energy and Environmental Research Corp., Irvine, QA 92718-2798. Brian K. Gullett is the EPA Project Officer (see below). The complete report, entitled "Fundamental Studies of Dry Injection of Calcium- Based Sorbents for SO2 Control in Utility Boilers," (Order No. PB 89-134 1421 AS; Cost: $36.95 will be available only from: National Technical Information Service 5285 Port Royal Road Springfield, VA 22161 Telephone: 703-487-4650 The EPA Project Officer can be contacted at: Air and Energy Engineering Research Laboratory U.S. Environmental Protection Agency Research Triangle Park, NC 27711 United States Center for Environmental Research Environmental Protection Information Agency Cincinnati OH 45268 Official Business Penalty for Private Use $300 EPA/600/S2-88/069 QOOQ3Z9 PS u S u S l ------- |