Energy & Fuels 2006, 20, 2465-2470
2465
Formation and Prevention of Agglomerated Deposits During the Gasification of High-Sodium Lignite Robert S. Dahlin,*,† WanWang Peng,‡ Matt Nelson,‡ Pannalal Vimalchand,‡ and Guohai Liu‡ Southern Research Institute and Southern Company SerVices, Power Systems DeVelopment Facility, P.O. Box 1069, WilsonVille, Alabama 35186 ReceiVed May 19, 2006. ReVised Manuscript ReceiVed August 25, 2006
A high-sodium lignite from the Freedom mine in North Dakota was tested in the transport gasifier at the Power Systems Development Facility (PSDF). During the first use of the high-sodium lignite in October 2003, agglomerated deposits formed at various locations in the transport gasifier system. An extensive laboratory testing program was carried out to characterize the deposits, understand the mechanism of the deposit formation, and test various methods of preventing or minimizing the agglomeration. The results of the deposit analysis and initial lab studies suggested that sodium released from the lignite was deposited on the surface of the sand bed material, resulting in the formation of sticky sodium silicates. Additional laboratory tests indicated that the agglomeration could be avoided or minimized by replacing the sand with a nonreactive bed material (e.g., coarse coal ash), operating at slightly reduced temperatures and using certain types of additives. By using these procedures, we completely eliminated the deposition problems in a subsequent gasification run in August 2004.
1. Introduction High-sodium lignite from the Freedom mine in North Dakota was tested in a transport gasifier at the Power Systems Development Facility (PSDF). The transport gasifier is an advanced circulating fluidized bed unit for converting coal into syngas. Because the gasifier operates at considerably higher solids circulation rates, velocities, and riser densities than a conventional circulating fluidized bed, the unit exhibits higher throughput and better mixing and has higher mass and heat transfer rates than typical gasifiers. More than 9100 h of gasification have been achieved in the transport gasifier at the PSDF with different fuels, including PRB coal and various bituminous coals and lignites, in both air- and oxygen-blown operations. The transport gasifier system and operational experience with the system have been discussed in detail in previous papers by Leonard et al.,1 Smith et al.,2 and Vimalchand et al.3 Gasification tests with the Freedom high-sodium lignite were conducted at the PSDF in October 2003 and August 2004 for a total of 445 h of gasifier operation. The operating experience and deposition problems encountered have been discussed in * Corresponding author. E-mail:
[email protected] † Southern Research Institute, Power Systems Development Facility. ‡ Southern Company Services, Power Systems Development Facility. (1) Leonard, R.; Pinkston, T.; Rogers, L.; Rush, R.; Wheeldon, J. The PSDF-Commercial Readiness for Coal Power Revisited. Presented at the 19th Annual Pittsburgh Coal Conference, Pittsburgh, PA, Sept 23-27, 2002. (2) Smith, P. V.; Vimalchand, P; Pinkston, T. E.; Henningsen, G.; Longanbach, J. L. Transport Reactor Combustor and Gasifier Operations. In Proceedings of the 8th International Energy Forum (Energex 2000), Las Vegas, July 23-28, 2000. (3) Vimalchand, P.; Leonard, R. F.; Pinkston, T .E. Power Systems Development Facility: Operation of a Transport Reactor System with a Westinghouse Candle Filter. Presented at the Advanced Coal-Based Power and Environmental Systems ’98 Conference, Morgantown, WV, July 2123, 1998.
detail by Peng et al.4 This paper focuses on the laboratory testing that was conducted to characterize the deposits, understand the mechanism of the deposit formation, and test various methods of preventing or minimizing the agglomeration. 2. Characterization of Agglomerated Deposits The October 2003 gasification test was terminated because the buildup of agglomerated deposits prevented the recirculation of solids in the gasifier recycle loop. In particular, major buildups of deposits occurred in the recycle cyclone, eventually spalling off and falling into the cyclone loop seal, blocking the recirculation of solids. After the run, the loop seal deposit had to be rodded out, producing the large deposit chunks shown in Figure 1. The chunks were typically 12-15 cm (5-6 in.) in size. Although the chunks were black, chemical analysis showed that they actually contained only 0.1-0.4 wt % carbon. Microscopic examination revealed that some of the particles were black, but most of them appeared to be glassy sand, sorbent, and ash (see Figure 2). More detailed examination showed that the black particles were actually sand particles that were covered with a thin outer layer of char (Figure 3). The chunks of loop seal deposit were ground as fine as possible in a mortar and pestle, and the ground powder was used in a variety of studies designed to understand the mechanism of the agglomeration and evaluate various corrective measures. First, the ground deposit powder was dispersed in isopropyl alcohol and analyzed on a Leeds & Northrup Microtrac X-100 particle-size analyzer. Almost all of the particles were in the size range of 60-700 µm, with a mean (4) Peng, W. W.; Nelson, M.; Liu, G.; Vimalchand, P.; Dahlin, R. S. High-Sodium Lignite Gasification With The PSDF Transport Gasifier. Presented at the 22nd Annual Pittsburgh Coal Conference, Pittsburgh, PA, Sept 12-15, 2005.
10.1021/ef0602269 CCC: $33.50 © 2006 American Chemical Society Published on Web 10/18/2006
2466 Energy & Fuels, Vol. 20, No. 6, 2006
Dahlin et al.
Figure 4. Micrographs of loop seal deposit after heating in air at various temperatures (note reconsolidation and increasingly glassy appearance as temperature is increased).
Figure 1. Chunks of agglomerated deposit removed from the cyclone loop seal.
Figure 2. Micrograph of deposit showing glassy sand grains, sorbent, and ash.
Figure 3. Micrograph showing sand grain covered with fine char particles.
particle size of 265 µm. This result is consistent with the presence of a large amount of sand bed material, which is generally much coarser than the coal ash and sorbent. Bulk chemical analyses confirmed that the deposit contained about twice as much silica as the high-sodium lignite on an ignited basis (35-41 wt % SiO2 in deposit versus 17-21 wt % SiO2 in lignite ash). Therefore, it is clear that the deposit was enriched in sand from the carryover of bed material. 3. Laboratory Studies of Agglomeration and Possible Corrective Actions 3.1. Experiments with Heated Deposit Samples. Samples of the ground loop seal deposit were heated in a muffle furnace in air at various temperatures to see if they would reconsolidate. The photos in Figure 4 show micrographs of samples heated at 760, 870, and 1040 °C (1400, 1600, and 1900 °F). As the treatment temperature was increased, an increasing degree of sample consolidation was observed. Some degree of sintering and neck formation between particles was noted even at the lowest temperature and became more pronounced as the temperature was increased. The particles also became progressively glassier and appeared to be largely amorphous at the
Figure 5. Close up of neck formed between two sand grains.
highest temperature. A similar type of behavior was noted in deposit samples and standpipe samples that were subjected to a standard loss-on-ignition (LOI) test (750 °C in air overnight). The micrograph in Figure 5 shows a close up of a neck formed between two sand particles in a standpipe sample that was subjected to a standard LOI test. It is important to note that the LOI test was done at a lower temperature than the gasifier operating temperature, so the necks were presumably formed in the gasifier. Energy-dispersive X-ray analysis confirmed that the particles were almost pure silica with small amounts of impurities. The neck region between the particles also contained silica and was enriched in sodium. On the basis of this observation, we speculated that the sintering may involve the formation of lowmelting sodium silicates, as was suggested in previous work by Bhattacharya and Hartigg,5 Vuthaluru et al.,6 and Manzoori and Agrawal.7 3.2. Experiments with Sintering in Sand/Sodium Mixtures. To investigate the potential for sodium silicate formation at gasifier operating temperatures, we treated samples of the bed sand with various amounts of sodium (in the form of NaOH solution) and heated them at 900 °C (1650 °F). As shown in Figure 6, consolidation was observed with only 0.5 wt % added sodium. Addition of 5 wt % sodium produced a solid glass that could not be removed from the crucible. Because silica and sodium were essentially the only reactants in these experiments, the primary product is undoubtedly sodium silicate. It is interesting that the addition of 5 wt % sodium resulted in extensive melting, because the loop seal deposit contained 6-7 wt % sodium as the element (8-10 wt % as Na2O). Because the gasifier operating temperatures were actually somewhat higher than the temperature used in this experiment, it is very likely that sticky sodium silicate would have been formed in the gasifier under the October 2003 operating conditions. (5) Bhattacharya, S. P.; Hartigg, M. Energy Fuels 2003, 17 (4), 10141021. (6) Vuthaluru, H. B.; Linjewile, T. M.; Manzoori, A .R.; Zhang, D-K. Fuel 1999, 78, 419-425. (7) Manzoori, A. R.; Agrawal, P. K. Fuel 1993, 72, 1069-1076.
Agglomerated Deposits in the Gasification of High-Sodium Lignite
Figure 6. Effect of sodium addition on the consolidation of bed sand.
The glassy condition of the 5 wt % sodium product observed in the lab tests was not observed in the chunks of agglomerated deposits that were removed from the gasifier after the October 2003 test, even though the deposits contained 6-7 wt % sodium. Nevertheless, the degree of agglomeration observed in the gasifier deposits was sufficient to prevent adequate solid circulation in the gasifier loop and force a shutdown of the gasifier system. The difference between the gasifier deposit and the lab product is probably attributable to the presence of inherent calcium minerals in the lignite that retard the agglomeration. This effect is expected on the basis of the Na2OCaO-SiO2 phase diagram. 3.3. Experiments with Various Additives to Prevent Agglomeration. On the basis of the foregoing tests, it was theorized that the sodium in the lignite was being vaporized in the gasifier and then deposited on the sand particles, where the sodium and silica reacted to form sticky sodium silicates. With this mechanism in mind, we considered two possible approaches to preventing agglomeration. First, we could introduce an inert particulate material that would physically block contact between the sand particles coated with the sticky sodium silicates. Second, we could inject a chemically reactive particulate sorbent that would scavenge the sodium vapor and thereby minimize the formation of the sodium silicates on the sand particles. AdditiVes That Physically Block Agglomeration. The feasibility of using an inert powder to block the agglomeration was tested with three different materials: dolomite, calcite, and coarse coal ash. The tests were done by mixing the pulverized material with the ground loop seal deposit and heating the mixture in a crucible in a muffle furnace. Figure 7 shows the results obtained by heating mixtures of ground deposit with dolomite at 930 °C (1700 °F) for 5 h. Under the above conditions, the loop seal deposit was completely reconsolidated. The mixtures with 5 and 10 wt % dolomite were less consolidated but contained several agglomerated chunks that were on the order of 0.6-1.3 cm (1/4 to 1/2 in.) in size. There was essentially no consolidation in the mixture containing 20 wt % dolomite. The above tests were done with a dolomite that was pulverized to a mean size of 200-300 µm. Examination of the heated mixtures under the microscope revealed that the dolomite particles had fragmented to a mean size of 60-70 µm, with some of the dolomite particles being as small as 20 µm. The degree of fragmentation may be even greater in the gasifier, where the rate of heating is much higher. Under those conditions, the dolomite particles may become so small that they will not be captured in the cyclone and hence will not be available to block agglomeration in the loop seal. In view of the potential fragmentation problem, tests were also conducted on a coarser dolomite that was pulverized to a mean size of 500-700 µm. The coarser dolomite produced larger particles that would be more likely to be retained in the cyclone loop seal. Unfortunately, however, the larger particles
Energy & Fuels, Vol. 20, No. 6, 2006 2467
appeared to be less effective in preventing consolidation, requiring about 50 wt % dolomite in the mixture. Nevertheless, the dolomite tests suggested that consolidation could be avoided by using materials that block contact between sticky particles. Similar experiments were conducted with coarse calcite having a mean particle size of 700 µm. In these tests, the deposit, the raw calcite, and various mixtures of deposit with calcite were heated to 930 °C (1700 °F) for 5 h. Surprisingly, the calcite prevented reconsolidation using a concentration of only 10 wt %, as illustrated in Figure 8. Microtrac analysis of the treated calcite showed that it underwent very little fragmentation (reduction in mean size from 700 to 600 µm). Because of the large size of the calcite particles, the number ratio of calcite to deposit particles was very low. Therefore, it seems unlikely that the calcite particles could be physically blocking the agglomeration of the sticky deposit particles. We speculated that the calcite might be so effective because it somehow alters the composition of the surface layer on the deposit particles, thereby effectively reducing the concentration of the sticky sodium silicate. This is possible because the addition of CaO to the Na2O-SiO2 system tends to shift the system toward higher melting points, thereby reducing the tendency to form sticky phases. This effect is illustrated in the phase diagram shown in Figure 9, which was originally published by Morey and Bowen.8 This could explain why the calcite was more effective than the dolomite. Although the dolomite also contains calcium, it contains much less calcium than does the calcite. Ash from the combustion of PRB coal was also shown to be an effective additive for reducing the reconsolidation of the loop seal deposit. The effect of the ash addition appeared to be similar to the effect of adding the finer form of the dolomite. Unlike the dolomite, however, the PRB ash did not undergo any fragmentation. In the absence of fragmentation, the coarse fraction of the ash should be retained in the gasifier loop under conditions at which most of the dolomite would be blown out. This suggests one possible approach to dealing with the agglomeration problem: replace the sand bed material with coarse coal ash, reducing the amount of silica available to react with the sodium vapor and thereby reducing the formation of sticky sodium silicate. AdditiVes That ScaVenge Sodium Vapor. We evaluated three different sorbent materials that were known to adsorb or react with sodium vapor at high temperatures: sand flour, kaolinite, and a paper mill byproduct that contained kaolinite. The sand flour was a very fine grind of sand with a mean size of 14 µm. The kaolinite was a pure form of the mineral that is produced commercially for use in paint pigments. The mean particle size of the kaolinite was 5.5 µm. The paper mill byproduct was a decarbonized form of deinking sludge that contained about 50 wt % kaolinite. It had a mean particle size of about 15 µm. All of these materials were so fine that there would be very little retention of the additive in the gasifier loop. Since most of the additive particles would pass through the recycle cyclone without being collected, there would be only a few seconds of residence time available for capture of the sodium vapor. Research performed elsewhere has shown that the kaolinite and the paper mill byproduct are capable of such rapid capture of sodium vapor under combustion conditions.9,10 We are not aware (8) Morey, G. W.; Bowen, N. L. J. Soc. Glass Technol. 1925, 9, 232233. (9) Gale, T. K.; Wendt, J. O. L. Combust. Flame 2002, 131, 299-307. (10) Gale, T. K.; Wendt, J. O. L. Aerosol Sci. Technol. 2003, 37, 865876.
2468 Energy & Fuels, Vol. 20, No. 6, 2006
Dahlin et al.
Figure 7. Effect of dolomite addition on the reconsolidation of the loop seal deposit.
Figure 8. Effect of calcite addition on the reconsolidation of the loop seal deposit.
Figure 9. High-SiO2 portion of the phase diagram for the Na2O-CaO-SiO2 system.
of any comparable studies under gasification (reducing) conditions, but we do not believe that the reaction between the kaolinite and the sodium is dependent on the presence of oxygen. It should be pointed out that the lack of retention in the gasifier loop is actually a desirable feature with these materials, because the sand flour and the kaolinite tend to form sticky sodium silicates (and sodium aluminosilicates in the case of the kaolinite). Therefore, they would actually exacerbate the agglomeration problems if they were retained in the gasifier loop. If, on the other hand, they are carried out of the system, it might be possible to capture some of the sodium vapor and
remove it from the gasifier loop before it can be deposited on the sand bed material and form sticky sodium silicate in the gasifier. All three of the sodium-scavenging additives were tested by placing a relatively thin layer (∼0.6 cm or 1/4 in.) of the additive over a thicker (∼1.9 cm or 3/4 in.) underlying layer of the ground deposit and heating the layers in a muffle furnace at 930 °C (1700 °F) for 5 h. In the case of the sand flour and the kaolinite, the top layer became completely consolidated. The result obtained with the sand flour is shown in Figure 10. The pure sand flour and the pure kaolinite did not undergo any consolidation when heated in the absence of the deposit.
Agglomerated Deposits in the Gasification of High-Sodium Lignite
Figure 10. Layer of consolidated sand flour above underlying layer of the loop seal deposit.
Therefore, the consolidation of the upper layer appears to be caused by uptake of sodium vapor that was released from the underlying layer of deposit. This result is not too surprising, because the reported reaction products include sticky sodium silicates that would be expected to cause consolidation.9,10 Unlike the sand flour and kaolinite, the upper layer of paper mill byproduct did not consolidate. Moreover, deposit mixtures containing 5, 10, and 20 wt % paper mill byproduct did not show any signs of consolidation when heated at 930 °C (1700 °F). At the same concentrations and the same temperature, mixtures containing sand flour and kaolinite were strongly consolidated. To further test the theory that the sodium in the deposit was being released and then captured by the sorbent material, we completed tests with the deposit layer above a layer of sand flour and above a layer of kaolinite. In these tests, only a very thin layer of the sorbent material adjacent to the deposit was consolidated. There was no consolidation at all in 99% of the sorbent material. This result is consistent with the theory that the sorbent is being consolidated by reaction with sodium vapor. With the deposit layer on top, the sodium is released to the atmosphere without passing into the sorbent material. However, strong consolidation was observed when the sorbent layer was on top, where the sodium vapor could pass through it. The results obtained with the sand flour and the kaolinite suggest that these materials will capture sodium vapor at the temperatures of interest. However, they apparently form sticky reaction products that may actually make the deposition problem worse if the reaction products are not carried out of the gasifier loop. Because the paper mill byproduct contains about half kaolinite, it probably also captured some sodium, but it did not consolidate. The absence of consolidation may result from the presence of other components that alter the phase compositions and melting points, or the kaolinite in the byproduct may be less reactive or less accessible than the pure kaolinite. 4. Fluidized Bed Tests The ground loop seal deposit and various mixtures of the deposit with additives were also evaluated in a fluidized-bed minireactor. The minireactor is a 3.8 cm (1.5 in.) diameter vertical tube furnace with a sintered metal frit at the bottom through which syngas or bottled gases may be introduced. The heated portion of the tube above the sintered metal frit is about 0.9 m (3 ft) in length. The deposit and deposit/admixture powders were place on the frit, fluidized in an upward flow of nitrogen gas, and heated at a rate of 110-220 °C/min (200400 °F/min). After the desired test temperature was reached, steady flow and thermal conditions were maintained for 1-2 h. The reactor was then cooled, and the sample was taken out for analysis. When tested at 900-950 °C (1650-1750 °F), the ground loop seal deposit reconsolidated, even though the bed was
Energy & Fuels, Vol. 20, No. 6, 2006 2469
fluidized with nitrogen. Under similar test conditions, no consolidation was observed with high-sodium lignite ash, which was carefully generated at relatively low temperatures in order to lessen the potential of any sodium loss. Although the highsodium lignite ash itself did not consolidate, mixtures of the high-sodium lignite ash with sand did show some consolidation. This result suggests that the sand must be present to produce the consolidation, which is consistent with the interaction of the sand with sodium to form sticky sodium silicates, as suggested by the lab studies. As in the lab studies, the minireactor tests also confirmed that dolomite addition reduced the extent of consolidation. The lack of consolidation in the lignite ash by itself suggests that the calcium content of the ash is sufficient to inhibit agglomeration in the absence of an external source of silica. When sand is added, however, the silica content is apparently increased to a level that exceeds the capacity of the inherent calcium to prevent sintering and agglomeration. This type of behavior is expected from the Na2O-CaO-SiO2 phase diagram presented earlier. Minireactor tests with a mixture of high-sodium lignite ash and PRB coal ash also yielded no consolidation, suggesting that the PRB coal ash can be used as a start-up bed material instead of quartz sand. However, both the high-sodium lignite ash and the PRB ash produced consolidated deposits when mixed with small amounts of trona and other low-melting sodium compounds, suggesting that accumulation of sodium in the gasifier could eventually lead to deposits. 5. Conclusions The laboratory studies and minireactor tests described above suggest that the deposition problems encountered in the October 2003 gasification run were most probably related to the formation of sticky sodium silicates through the interaction of sodium from the high-sodium lignite and the silica in the sand bed material. The lab studies also provide some evidence to support the use of additives to either block the agglomeration of sand particles covered with sticky sodium silicates or to reduce the formation of the sodium silicates on the bed material by scavenging the sodium vapor before it can react with the sand. Dolomite, calcite, and PRB coal ash were found to be effective in reducing reconsolidation in the lab tests, but fragmentation may reduce the effectiveness of the dolomite and calcite in the gasifier loop. Although the calcite did not fragment in the lab tests, it may still do so in the gasifier loop because of the much higher rates of particle heating. The PRB coal ash offers the advantage of no fragmentation. If coarse PRB ash could be used as the gasifier bed material in lieu of sand, it would effectively reduce the silica content of the bed material to about 20% and thereby greatly reduce the potential to form sticky sodium silicates. The lab studies showed that both sand flour and kaolinite will react with sodium vapor at gasifier temperatures. However, both of these materials exhibited extensive consolidation during the sodium capture. This behavior may not cause a problem if the reacted sorbents escape collection in the recycle cyclone and are carried out of the gasifier loop. If some of this material is captured in the recycle cyclone and returned to the system, it could actually exacerbate the deposition problem, because the reaction products contain sticky sodium silicates. This could be a serious concern with the sand flour, because about 20 wt % sand flour is larger than 30 µm and would probably be collected in the cyclone. The concern may not be as great with the kaolinite, because less than 5 wt % kaolinite is larger than
2470 Energy & Fuels, Vol. 20, No. 6, 2006
that size. The kaolinite also offers the advantage that it may capture other trace metals such as lead and cadmium.9,10 Unlike the sand flour and the kaolinite, the paper mill byproduct did not consolidate under the test conditions. On the basis of the kaolinite content of the byproduct, it seems reasonable to assume that it captured some sodium vapor, but it was probably less effective than the pure kaolinite. Nevertheless, it may be an attractive sorbent material if it can remove some sodium without agglomerating. In a subsequent gasifier run in August 2004, several of the remedial procedures that were tested in the laboratory were applied during the gasification of the high-sodium Freedom lignite. Prior to the introduction of the high-sodium lignite, much of the sand bed material was replaced with coarse ash from PRB coal, and gasifier operating temperatures were slightly
Dahlin et al.
reduced (about 55 °C or 100 °F below the temperatures used in the October 2003 run). Dolomite was also fed into the gasifier to help minimize the contact between any sticky particles. By the use of these procedures, the deposition problems that were seen in the October 2003 run were completely eliminated in the August 2004 run. Acknowledgment. The authors gratefully acknowledge the financial and technical support provided by the U.S. Department of Energy and the other project participants who provided funding and technical assistance for the work at the PSDF. The funding partners include Southern Company Services; Kellogg, Brown, and Root; Siemens-Westinghouse, Peabody Energy, Burlington-Northern Santa Fe Railroad, and the Lignite Research Council. EF0602269
