EPA-650/2-74-001 January 1974 Environmental Protection Technology Series ------- EPA-650/2-74-001 A REGENERATIVE LIMESTONE PROCESS FOR FLUIDIZED-BED COAL COMBUSTION AND DESULFURIZATION by R. C. Hoke, M. S. Nutkis, L. A. Ruth, and H. Shaw Esso Research and Engineering Company P.O. Box 8 Linden, New Jersey 07036 Contract No. CPA 70-19 ROAP No. 21ADB-13 Program Element No. 1AB013 EPA Project Officer: S. L. Rakes Control Systems Laboratory National Environmental Research Center Research Triangle Park, North Carolina 27711 Prepared for OFFICE OF RESEARCH AND DEVELOPMENT U.S. ENVIRONMENTAL PROTECTION AGENCY WASHINGTON, D.C. 20460 January 1974 ------- This report has been reviewed by the Environmental Protection Agency and approved for publication. Approval does not signify that the contents necessarily reflect the views and policies of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use. 11 ------- AB STRACT An experimental study was begun of the pressurized combustion of coal in a fluidized bed of limestone and regeneration of sulfated limestone. The work is part of an overall program aimed at developing fluidized bed coal combustion as a means of desulfurizing flue gas in—Situ and generating clean power at low cost. The process includes regeneration of spent limestone by reduction to lime. This produces a gas stream containing a sufficient concentration of SO 2 to be fed to a by product sulfur recovery unit. The regeneration step was studied at pressures up to 10 atm and tempera- tures up to 2100°F. S02 concentrations measured.lin the product gas averaged about 2% at 10 atm pressure and 2100°F, about 40% of the concentrations calculated by assuming equilibrium between the solids and regenerating gas. High conversion of sulfated material to lime was achieved by injecting air into the bed, forming adjacent reducing and oxidizing zones, and minimizing formation of undesired CaS. Combustion studies also began. The combustion runs were limited by operating problems, especially plugging in the coal injection line. Initial SO 2 removal rates ‘were about 85%. However attrition rates were high with one SO 2 sorbent, Tyinochtee dolomite. This report was submitted in fulfillment of Contract CPA 70—19 by Esso Research and Engineering Company under the sponsorship of the Environ- mental Protection Agency. Work was completed as of July 31, 1973. iii ------- CONTENTS Page Abstract iii List of Figures v List of Tables vii Acknowledgments viii Sections I Conclusions 1 II Reconiinendations 2 III Introduction 3 IV Experimental Equipment, Materials, Procedures 5 V Development of Experimental Equipment and Procedures 30 VI Experimental Results 53 VII Discussion of Results 87 VIII Program for Operation of Niniplant 90 IX References 92 X List of Publications 93 XI Glossary 94 XII Appendix 96 iv ------- FIGURES No. Page 1 Cold model test unit 6 2 Cold model test unit solids reservoir 7 3 Cold model test unit simulated heat transfer coils 8 4 Fluidized bed coal combustion unit 10 5 Petrocarb coal injector 11 6 Fluidized bed coal combustor 12 7 Combustor fluidizing grid 14 8 Coal feeder test unit 16 9 Fluidized bed regeneration unit 18 10 Regenerator burner 19 11 Fluidized bed regenerator 20 12 Fluidized bed combustion and regeneration units 22 13 Fluidized bed unit control panel 23 14 Coal feeder orifice 32 15 Effect of injection air flow rate on coal feed rate — low pressure 40 16 Effect of feed tank pressure on coal feed rate 41 17 Effect of air flow rate on coal feed rate — high pressure 43 18 Effect of feed tank pressure on coal feed rate 44 19 Fused regenerator bed material and fluidizing grid 51 20 Effect of superficial bed velocity on expanded bed height — 2 ft. settled bed, simulated heat transfer coil 54 21 Effect of superficial bed velocity on expanded bed height — 3 ft. settled bed, simulated heat transfer coil 55 V ------- FIGURES (Cont d) No. 22 Effect of heat transfer coil on expanded bed height 56 23 Solids transfer rate 58 24 Adiabatic flame temperature versus % (CO + 112) at 8 atm total pressure 72 25 Regeneration studies — effect of pressure on approach to equilibrium SO 2 concentration 74 26 Regeneration studies — effect of pressure on SO 2 concentration 75 27 Regeneration studies — effect of temperature on SO 2 partial pressure at equilibrium 76 28 Regeneration studies — effect of temperature on SO 2 concentration 78 29 Regeneration studies — effect of temperature on approach to equilibrium SO 2 concentration 79 30 Regeneration studies — effect of temperature on CaO/CaS ratio in regenerated solids 86 vi ------- TABLES Page 1 Coal Particle Size Distribution 25 2 Composition of Coal Used in Esso Batch - Fluidized Bed Combustion Program 26 3 Properties of Limestone and Dolomites 26 4 Coal Feed Rates With a Modified Petrocarb Model 16-1 ABC Injector 33 5 Results of Regression Analysis of Coal Feed Test Data 45 6 Operating Parameters - Coal Combustion Runs 60 7 Composition of Effluent Streams - Coal Combustion RUnS 61 8 NO-CO Reactions 65 9 NO-CO Reactions - Effect of Pressure and Residence Time 65 10 NO-CO Reactions - Effect of 02 66 11 Run Summary - Fixed Bed Regeneration Studies 69 12 Summary of Runs With Batch Fluidized Bed Regenerator 70 13 Analysis of Solids from Regeneration Runs 82 A-i Fixed Bed Simulated Combustion Runs 97 A- 2 Air, Fuel and Solids Inputs for Regeneration Runs 99 A—3 Composition of Effluent Stream for Regeneration Runs 100 vii ------- ACKNOWLEDGEMENTS The authors wish to express their appreciation to Messrs. H. R. Silakowski and W. H. Reilly for performing the laboratory work described in this report. We also wish to ackowledge the efforts of the personnel of the Mechanical Division who were responsible for design and construction of the pilot plants and who assisted in the startup of the units, particulary Messrs. E. C. Vath and R. A. Van Sweringen. We wish to express our appreciation to Mr. S. L. Rakes, the EPA Project Officer, for his many and varied contributions to the conduct of this program. viii ------- SECTION 1 CONCLUSIONS The regeneration of CaSO4 to CaO and S02 appears to be limited by reaction rates. The measured SO 2 concentration in the product gas from the regenerator averaged about 40 to 50% of the maximum concentration calculated by assuming the solids and gases were in chemical equilibrium. At 10 atm* pressure, the measured SO 2 concentration averaged around 27.. Additional work will be necessary to determine if conditions can be found to allow a closer approach to equilibrium. T o routes could be followed — improving the quality of fluidization and increasing the gas/solids residence time. High conversion of CaSO4 to CaO was obtained by adding auxiliary air directly to the fluidized bed. This created adjacent reducing and oxidizing zones in the bed and reduced the tendency to form undesired CaS. Addition of a portion of the fuel normally used in the regenerator reducing gas generator directly to the fluidized bed produced higher and more uniform bed temperatures without causing bed agglomeration. Of the S02 released by the combustion of coal, 85 to 90% was retained by the fluidized bed when calcined limestone or half—calcined dolomite was used as the bed material at 5—8 atm pressure. However, the runs were very short because of plugging in the coal injection probe. Longer runs are necessary before definite conclusions can be drawn about the suitability of the limestone or the dolomite. The dolomite (Tymochtee) gave very high attrition rates and may not be suitable for once—through operation. Feeding of coal using a modified Petrocarb pneumatic injector at 10 atm pressure and low feed rates is possible. However, careful attention must be paid to operating conditions to prevent plugging of the injec—- tion lines. Additional modifications to the probe design are being made to reduce further coal plugging in the injection probe. A number ofcequipmeiit developments have been made incidding the design and construction of a very versatile burner used to generate heat and reducing gas for the regenerator. The burner can be operated over a wide range of pressures, flow rates and oxidizing or reducing condi- tions. A water cooled fluidizing grid was also developed which operates in the presence of very high gas temperatures in the regenerator. * EPA policy is to express all measurements in Agency documents in metric units. Because implementing this practice will result in undue cost, NERC/RTP is providing conversion factors for the particular non—metric units used in this document. For this report these factors are located on page 95. 1 ------- SECTION II RECc MENDATIONS Work in the area of fluidized bed coal combustion should proceed along two lines. First, work on a non—regenerative pressurized combustion system should proceed at this time to speed application of fluidized bed combustion and prevent any delays which could result from the development of a regenerative system. Second, work on the regeneration step should proceed as rapidly as permitted by the availability of funds, to develop a process which would minimize the disposal burden of spent limestone. It is recommended that the pressurized coal combustion program should first concentrate on the solution of the remaining operational problems. The effect of precalcination of a typical limestone and dolomite on S02 removal should then be studied. This would be done to study further the effects of precalcination conditions on stone capacity recently reported by Westinghouse Research Laboratory. A study would then be made of the effect of operating conditions on desulfurization effective- ness, NOx and trace emissions, attrition and utilization of the sorbent, combustion efficiency and heat transfer rates. The operating conditions studied would include pressure, temperature, bed depth, fluidizing velocity and excess oxygen. This study should be made with several combinations of coals and sorbents (stones). Suggested coals are Eastern, Western and possibly lignite. The regeneration program should be based on the use of sulfated lime- stone or dolomite rather than pure CaSO4. The study should determine the effects of improving fluidization quality and increasing gas residence time on the S02 product concentration and the effect of repeated combustion/regeneration cycles on the activity of the recycled limestone or dolomite. The fluidized bed combustion unit could be used further to screen other coals and sorbents and possibly even shale. It could also be run in conjunction with the fluidized bed Miniplant to guide the selection of operating conditions for the Miniplant and to help resolve problems arising from the operation of the Miniplant. 2 ------- SECTION III INTRODUCTION The fluidized bed combustion of coal is a new combustion technique which can reduce the emission of SO 2 from the burning of sulfur—con- taining coals. This is done by using limestone or dolomite as the bed material. This technique has other potential advantages over conven- tional coal combustion systems which could result in a more efficient, less costly method of power generation. High heat transfer coef- ficients between the fluidized bed and immersed steam generation sur- faces reduce the steam tubing requirements, and also permit opera- tion at lower and more uniform bed temperatures, In the vicinity of 1500 to 1700°F. The lower temperatures also reduce NO emissions and decrease steam tube corrosion. Lower grade coals can also be burned since the bed temperatures are lower than ash slagging temperatures. In the fluidized bed boiler, limestone or dolomite is calcined and reacts with SO 2 and oxygen in the flue gas to form CaSO 4 as shown in reaction (1). When used on a once—through basis, high limestone or CaO + SO 2 + 1/202 ‘ CaSO 4 (1) dolomite feed rates to the boiler are required if SO 2 removal of 90% or more is to be achieved. In order to reduce the solid waste disposal burden created by high limestone feed rates, a system is now under study in which the CaSO4 would be regenerated back to CaO in a separate fluidized bed reactor by reaction with a reducing gas at a temperature of about 2000°F as shown in reaction (2). The regenerated CaO would be CO CO 2 CaSO 4 + H 2 ) CaO + SO 2 + H 2 0 (2) returned to the boiler where it would again react with SO 2 and 02. Engineering and cost analyses carried out by Westinghouse Research Laboratory for the Environmental Protection Agency (EPA) Indicated a greater commercial potential for a pressurized combustion system when used in conjunction with a combined gas—steam turbine power generating plant. Based on theseanalyses, EPA requested Esso Research and Engineering Company to study the combustion and regeneration steps at pressures up to 10 atm. A new pilot plant capable of operating at these conditions was built and operated. The primary purpose of the pilot plant study was to explore the effect of operating parameters on combustion and regeneration at pressures up to 10 atm. In addition, the study was to include measurement of the activity maintenance of limestone or dolomite due to repeated combustion, regeneration cycles. Measurement of emission of trace elements from coal combustion was also to be made. Development and testing of equipment and operating procedures was also carried out as part of the program A test program 3 ------- for the operation of the large fluidized bed coal—combustion Miniplant was also developed as part of the program. In addition, a small amount of work was done in the cold model fluidization test unit to study fluidization characteristics and transfer of solids between the com- bustion and regenerator vessels. This was done in support of the Miniplant unit construction effort. A short laboratory program was also carried out in small fixed bed units studying the reduction of NO by CO under simulated combustion conditions. Some fixed or semi— fluidized bed regeneration studies were also carried out in small laboratory equipment. 4 ------- SECTION IV EXPERIMENTAL EQUIPMENT, MATERIALS, PROCEDURES COLD MODEL TEST UNIT For the purpose of verifying the design of the pulsed air solids transfer system used in the design of the fluidized bed coal combustion Miniplant, and to permit visual observation of its fluidization characteristics, a cold model test unit (CMTU) was built. The CMTU is a two vessel fluidized solids system constructed from Plexiglas, a transparent acrylic plastic. The CMTU was designed to operate at ambient temperature and pressures up to 60 psig. Figure 1 is a picture of the CMTU. The unit consists of two 5.5 inch I.D. vessels simulating the combustor and regenerator of the Miniplant. The simulated combustor (i.e., the vessel on the left in Figure 1) is 18 feet tall and was assembled from three flanged sections. The regener- ator is made of two flanged sections and stands 12 feet tall. Both vessels have bottom plenum chambers and grids for distributing the fluidizing air and single stage aluminum cyclones for removing entrained solids. These solids can be returned to the fluidized beds or collected as desired. Nylon reinforced transparent PVC hose, 1.5 inch I.D. is used for solids transfer between the two vessels. This permits visual observation of solids movement in these lines which connect the upper part of the vessel transferring the solids to the lower part of the vessel receiv- ing them. The small receiving—injection pots to which the bottom of these lines are attached are equipped with fluidizing and pulsed air. In some of the solids transfer studies, the use of an overflow collector was investigated as a means of creating an enlarged solids reservoir in the transmitting vessel. In the Miniplant, such a reservoir would serve to insure a continued solids seal in the transfer legs and minimize backflow of gases between reactors. The enlarged reservoir, shown on the regenerator vessel in Figure 2, was constructed of a 6 inch I.D. aluminum tee with a plate on the inside to serve as a solids baffle and a Plexiglas window on the outside for observation of the solids. In addition to studying solids transfer techniques, the CMTU was used to determine what effect the hair pin heat transfer ioops designed for the Miniplant would have on the fluidization characteristics of the combustor. Figure 3 shows a simulated heat transfer loop section that was used in the CMTU. It was fabricated of 3/8 inch aluminum as a prototype half—scale version with the same configuration and pitch as the Niniplant design. It is 68 inches in vertical length starting at a point 13.5 inches above the distributor grid. Inser- tion of the coil reduces the average cross section area by 8%. 5 ------- Figure 1. Cold model test unit 6 ------- ill I Figure 2. Cold model test unit solids reservoir 7 ------- I Figure 3. Cold model test unit simulated tieat transter coils 8 1 * ------- Gas flow rates to each of the two reactots of the CNTU could be varied. A control panel situated to the left of the Plexiglas vessels per- mitted measurement of their flow rates and of their individual bed pressure drops. It also controlled and measured the pressure differ- ential between beds, and provided a means of varying the on-off pulsing times and the air flow rates for the solids transfer pots. FLUIDIZED BED COAL CONBUSTION UNIT A schematic diagram of the Esso fluidized bed combustion unit is shown in Figure 4. The primary components of the unit are (1) the coal feeding system, (2) the fluidized bed combustor, (3) and the gas handl- ing and analytical equipment. Coal Feeding Equipment Figure 5 shows the Petrocarb Model 16-1 ABC injector. The main fea- tures are a conical-bottom tank that holds solids to be fed and an orifice and mixing valve assembly that mixes solids with carrier gas. Solids in the tank are aerated by a controlled stream of air at a selected pressure. Aerated solids flow through tl orifice at the bottom of the tank into the mixing valve assembly and are picked up by a controlled stream of carrier gas (air). Solids are pneumatically conveyed through a transport line into the receiving vessel. The feed rate of solids is controlled by pressure in the feed tank, carrier or injection air flow rate, and pressure differential between the feed tank and receiving vessel. The Petrocarb solids feeder shown in Figure 5 was modified to feed powdered coal (-30 mesh) at rates of 6—28 lbs/hr into the combustor against pressures of up to 10 attn. The feeder, as ut is supplied by Petrocarb, can handle only much higher feed rates. The diameter of the orifice was reduced to 3/16 inch and the Petrocarb injection hose was replaced with a 1/4 inch diameter (O.D.) x 15 feet long stainless steel tube. In order to make the feeder work satisfactorily with the batch combustor, the feeder to combustor presssure differential had to be held constant. This was accomplished with automatic controls which maintained the pressure in the feed tank above the pressure in the corn- bustor by the desired amount. A summary of the coal feeder test pro— gram and a detailed description of the modifications made on the Petrocarb injector are given in Section V. Fluidized Bed Combustor A schematic diagram of the fluidized bed coal combustor is given in Figure 6. The vessel was constructed from four sections of 10 inch standard wall carbon steel pipe and refractory lined with Gref cc Litecast No. 7528 to an inside diameter of 4 inches. The height of the vessel, above the fluidizing grid, was about 16 feet. Below the grid was a 2 foot long burner section lined with Bubbalite. The fluidizing grid, which was made of stainless steel, had seven air distributor caps 9 ------- VENT CYCLONES DRAIN REFRIGERATOR C COOLER COOLING COILS WATER WATER DEMINERALtZER FLUIDIZING GRID AIR FROM COAL FEEDER PROPANE WATER DRAIN INJECTION AIR Figure 4. Fluidized bed coal combustion unit ------- EXHAUST VALVE Figure 5. Petrocarb coal injector PRESSURE RELIEF VALVE LINE PRESSURE GAUGE DILUTER PRESSURE GAUGE DILUTER PRESSURE REGULATOR FEED VALVE FILLING VALVE TANK PRESSURE GAUGE REGULATOR VALVE GAS SHUTOFF VALVE MIXING -DILUTER HOSE INJECTION HOSE TO RECEIVING VESSEL BUSHING ‘QUICK DISCONNECT COUPLING 11 ------- GRE FCO LITECAST #7528 COOLING COILS (3 PAIR — EACH PAIR CENTERED ABOUT FLANGE) COOLING WATER INLET 2’ 0” 2’ 3’ COOLING WATER OUTLET PORT FLUIDIZING GRID (WATER COOLED) BUBBALITE BURNER Fluidized bed coal combustor THERMO- COUPLE 8’lO” 10” STD. WALL STEEL PIPE SOLIDS CHARGING PORT F LAN GE Figure 6. 12 ------- and was water-cooled. it is described in Figure 7. A propane burner, located at the bottom of the burner section, was used to preheat the unit to above the self—ignition temperature of coal. The burner is described in more detail in Section V. The maximum operating tempera- ture and pressure of the batch combustor were 1900°F and 10 atm., respectively. Two air compressors supplied sufficient air for opera- tion at a superficial velocity of 6 ft/sec at 10 atm and 1600°F (about 80 SCFM). Bed temperature was controlled by three pairs of serpentine wound cooling coils made of 1/4 inch diameter stainless steel tubing. The total surface area was about 6.3 sq. ft. The center of each pair of coils was located at a flange in the shell of the corn— bustor. Water flowed in parallel to each of the six coils; flow rate to the lower and upper pair of coils was manually controlled but flow to the center pair was automatically controlled to give the desired bed temperature. Thermocouples were located 6 inches apart in the lower sections of the combustor and 12 inches apart in the upper section. Solids were loaded into the combustor through a charging port located in the upper section. Solids could be removed through a port in the lower section or, alternatively, transferred directly to the regenera- tor by blowing them through a 2 inch pipe supplied for this purpose. Coal entered through a water-cooled coal injection probe to which was connected the end of the 15 feet x 1/4 inch diameter coal injection tube. The probe was a 14 inch length of the 1/4 inch diameter stainless steel tubing surrounded by a jacket through which cooling water flowed. Gas Handling and Analytical System Flow of air and fuel into the combustor, and system pressure, were under automatic control. Gases leaving the combustor first passed through two cyclones, which removed entrained solids and fly ash. An off-gas cooler, which followed the cyclones, reduced the temperature of the off-gas to the desired level. The off-gas then entered an expansion coil of 1 inch diameter stainless tubing which was electrically heated during startup to raise the temperature of gases above the dew point. A 1.5 inch Aerotec cyclone, following the heater, was used to remove particulates during startup of the combustor, when water vapor condens- ing in the first two cyclones caused them to operate at reduced effi- ciency. Fine particulates were removed with a Pall Model MD-600 filter, located upstream of the back-pressure control valve. This filter had a mean pore size of 165 microns and an area of 0.73 sq. ft. Before being vented, the off-gas entered a chiller and knockout to remove moisture so that the water content of the gas could be determined. Alternatively, the gas could be sent directly to a second knockout at room temperature and vented. A small portion of the off-gases were diverted to a refrig- erator which lowered the dew point of the gas to about 35°F before it was sent to the gas analysis equipment. 13 ------- WATER OUTLET 5 ”— 5 ’”SLOT 16 16 WATER-COOLING CHANNELS 5,I (Sj w\ 16 ‘16) #47 DRILL, 8 PLACES EQUALLY SPACED 32 AIR DISTRiBUTOR CAP 304 S.S., 7 REQ’D. THERMOWELL(” ”) GASKET SEALING GROOVES (BOTH SIDES) WATER COOLED GRID PLATE 304 S.S., 12 ” DIA., “ THICK Figure 7. Combustor fluid izing grid WATER tNLET 14 ------- Gas analysis equipment included Beckman Model NDIR 315 analyzers for SO 7 , NO, and CO, a Beckman Model 715 polarographic analyzer for 02, ana a LIRA Model 300 analyzer for CO 2 . The concentration ranges of the analyzers used with the combustor are given below. So 2 0-3000 ppm Instrument No. 1 Instrument No. 2 CO 0-1250 ppm 0-10 per cent 0-6250 ppm 0-30 per cent NO 0-500 ppm 0-1000 ppm 0-2500 ppm 02 0-5 per cent 0-25 per cent Co 2 0-25 per cent Coal Feeder Test Unit Figure 8 is a diagram of the apparatus used to test the Petrocarb coal feeder before use with the combustor. Air was used to pressurize the Petrocarb coal feeder and transport coal through the injection tube to the receiving vessel. The Petrocarb unit was equipped with pressure regulators for both the feed tank and injection air, a control valve for injection air, and a rotameter to indicate total air flow. The receiving vessel was mounted on a scale which indicated weight changes as small as 001 lb. Pressure in the receiving vessel was set at the desired value by means of a back pressure regulator. A differential pressure cell measured i P between the coal feed tank and receiving vessel. House air for the feeder first entered a bed of Drierite to reduce its moisture and thus keep the coal dry. Air leaving the receiv- ing vessel was passed through a 65 filter to remove coal fines. Experimental runs were first made at pressures of 70 psig and below in the feed tank. After some operating experience was obtained, modifica- tions wa made in the equipment to permit operation at pressures of 135 psig in the feed tank or about 10 atmospheres (absolute) in the receiv- ing vessel. For operation at higher pressures, the Petrocarb unit was connected to an air compressor. Air from the compressor passed through two silica gel driers to keep moisture reaching the coal to a minimum. The original receiving vessel, rated for 100 psig, was replaced with one good for 250 psig. Heavy rubber hose was substituted for Polyflow tubing for runs made at higher pressures. Pressure regulators supplied with the Petrocarb feeder had a 125 psig limit and were replaced with regulators designed for higher pressures. Air from the compressor was stepped down in pressure with an additional regulator. 15 ------- Figure 8. Coal feeder test unit (section within dashed lines provided by Petrocarb) ------- FLUID IZED BED REGENERATION UNIT Figure 9 is a schematic diagram of the Esso fluidized bed regeneration unit. The main components of this unit are (1) the burner, (2) the fluid bed regenerator, and (3) the gas handling and analytical equipment. Burner The burner, shown in Figure 10, was located at the bottom of the regen- erator, under the fluidizing grid. It produced heat and a reducing gas mixture containing CO and H 2 . The main parts were a tube to mix the fuel (propane + propylene) with air and a water-cooled grid to hold the flame. The tubular part of the burner was made of 304 stainless pipe, 2 inches in diameter and 5.5 inches long. At the bottom of the tube were two baffles which supported a bed of 1/8 inch alumina beads and also helped to prevent channeling of the air-fuel mixture up the sides of the tube. Air and fuel entered at a tee below the burner tube; fuel entered the tee through a sparging tube containing eight—3/64 inch diameter holes at its tip to promote even distribution. The alumina beads further mixed the air-fuel mixture. The burner grid was a 3 inch diameter x 1/4 inch thick brass plate which screwed onto the top of the burner tube. The grid was water- cooled and contained about 400—1/32 inch holes through which the air— fuel mixture flowed. On the underside of the grid was a disk of porous stainless steel which acted as a flame arrestor and prevented flashback into the burner tube. The burner was ignited by a pilot light which was positioned near the side of the burner tube below the burner grid. The pilot light, in turn, was ignited by an electrode. This arrangement kept the ignition equipment away from the flame of the burner, where it would be damaged by excessive heat. Batch Fluidized Bed Regenerator Vessel Figure 11 shows the batch fluidized bed regenerator. The vessel was constructed from three sections of 12 inch carbon steel pipe and refrac- tory lined with Grefco Litecast No. 7528. The interior of the regen- erator was lined with a 3.25 inch diameter alumina tube. The height of the unit above the fluidizing grid was about 16 feet. A burner sec- tion, about 2 feet long, and lined with Bubbalite, was located under the grid. The burner was mounted on a flange which attached to the bottom of the burner section. Air and fuel required for regeneration entered the regenerator at the bottom of the burner. However, for some runs, a portion of the fuel and secondary air were added at various points above the fluidizing grid. 17 ------- VENT REFRIGERATOR COMPRESSED AIR (FILTERED, DRIED) INFRARED ANALYZERS FRC OFF-GAS COOLER AIR CONTROL VALVE OFF-GAS CHILLER DRY GAS METER FLUIDIZING GRID BURNER PROPANE KNOCKOUT DRAIN Figure 9. Fluidized bed regeneration unit ------- 3” DIA. BURNER HEAD (BRASS) GASKETS COOLING COIL SCREEN 400 — HOLES WATER COOLING CHANNEL POROUS METAL DISK 2” DIA. S.S. TUBE, 2 LONG ENTIRE TUBE FILLED WITH ALUMINA BEADS 4 + 4 4 4 AIR&FUEL Figure 10. Regenerator burner FLANGE BAFFLES 19 ------- GREFCO LITECAST #7528 5’lO” ____ — LD. THERMOCOUPLE OLIDS CHARGING PORT _— 12 x 0.375 WALL STEEL PIPE — = = = — FLANGE = 5’ 0 - 5’ O SOLIDS REMOVAL PORT — = = ..— FLUIDIZ1NG GRID (WATER-COOLED) 2’ 3” — BUBBALITE SIG HTG LASS Figure 11. Fluidized bed regenerator 20 ------- The maximum operating temperature and pressure of the batch regenera- tor were 2100°F and 10 atm., respectively. Thermocouples, spaced 6 inches apart, were used to judge bed temperature. Besides air and fuel, N 2 and CO 2 could be fed into the reactor. For runs in which secondary air was not added above the fluidizing grid, °2 was added near the top of the regenerator to convert elemental sulfur, which might plug lines, to So 2 . All gas flows and system pressure were under automatic control. Air could be preheated before entering the burner by using an electri- cal air preheater with 7.5 kilowatts capacity. Kanthal wire was wound around the outside of the ceramic tube in each of the two lower sec- tions of the regenerator. Each heater had a capacity of 7.5 kilowatts and could be used to provide extra heat to the bed. The fluidizing grid for the regenerator was made of 304 stainless steel and contained 69-5/64 inch holes for the passage of fluidizing gases. A five inch square at the center of the grid was water-cooled. Solids were charged into the regenerator through a port located near the top of the unit. Solids could be removed through another port located just above the fluidizing grid. Gas Handling and Analytical System Gases leaving the regenerator were passed through a cyclone to remove entrained solids and then to an off-gas cooler which reduced tempera- ture to the desired level. Downstream of the off-gas cooler, the regenerator shared equipment with the batch combustor. Thusly, the gas stream entered a solids filter, back pressure control valve, chiller and knockout before being vented. The analytical instruments, too, were shared with the combustor. However, an additional Beckman Model NDIR 315 analyzer was used to determine the concentration of SO 2 in the 0-157. range. Air for the regenerator was supplied by a compressor. Other gases came from cylinders. Fuel cylinders (propane-propylene) had to be kept in a steam-heated hot-box at 120°F in order to maintain sufficient vapor pressure. In addition, fuel lines near the regenerator were heated to prevent fuel from liquefying. 1 igure 12 is a photograph of the fluidized cotnbustor and regenerator. Figure 13 is a photograph of the control panel used to operate the units. LABORATORY FIXED BED U TS Steel Reactor A 1 inch 0.D. stainless steel reactor (0.75 inch I.D.) was used for fixed bed studies examining reactions between NO and Co under simulated combustion conditions. The reactor was capable of operating at temper- atures up to 1600°F at pressures up to 10 atm. Feed gases were purch- ased with various concentrations of NO, CO, C0 2 , and 02 in N 2 or A 21 ------- Figure 12. Fluidized bed combustion and regeneration units COMBUSTOR r REGENERATOR I T 4 i — W I — ‘WV 22 ------- - —___p — I I ci - #lI It . 1 ‘ v -s’ S S S • L .J. I ‘. \ !• l \ I Figure 13. FLuldized bed unit control panel ------- diluent. Provisions were made to blend two gases before passing the gas mixture through the reactor. Rotameters were used to meter the gases. The reactor was packed with alumina beads which acted as a pre- heat section, followed by the bed material. The bed material was calcined limestone (1359) or dolomite (1337). The entire reactor was held in an electrically heated furnace. The off gases were anal- yzed for NO and CO by Beckman Model 315 NDIR analyzers. CO 2 was measured by an MSA Model 200 NDIR analyzer. The gases were cooled to 35°F before analysis to remove water. A Beckman Model 715 polarographic analyzer was used to measure 02 concentrations in the product. The blended feed gas could also be diverted to the analyzers for measurement. Ceramic Reactor A one inch I.D. ceramic reactor was used to study regeneration of CaSO 4 at temperatures up to 2000°F and pressures up to 9.5 atm. The reactor consisted of a one inch I.D. ceramic tube mounted axially in a 3 inch Schedule 5 304 stainless steel pipe. The reactor was heated electrically by resistance heater windings on the external surface of the ceramic tube. The overall length of the ceramic tube was 28 inches but only the mid 24 inch section was heated. For most of the runs, a 40 gin charge of CaSO4 (anhydrite) or partially sulfated lime was used and was placed about 17 inches above the bottom of the ceramic tube. The lower section of the ceramic tube was filled with ceramic beads and served as a preheat section for the feed gases flowing up through the beads. Temperature was measured by a therrno- couple inserted in a 1/4 inch O.D. ceramic well located axially in the center of the 1 inch reactor. A second thermocouple in this well oper- ated the temperature controller. The feed gas was produced by metering a CO/N 2 blend and a C0 2 /N blend to give the desired feed gas composition and flow rate. Rotame ers were used for gas flow measurement. The product gas was then cooled, passed through a filter, back pressure regulator, and chiller before analysis for SO 2 content in a Beckman Model 3l5A infrared analyzer. 24 ------- NATERIALS C oa 1 The coal used in the batch fluidized bed combustor, and in the test program for the coal feeds, was a high volatile (A) bituminous coal from Consolidation Coal Company’s Arkwright mine in West Virginia. It was ground to —30 mesh by Peun—Riliton Co. The specified particle size distribution is given in Table 1. A proximate and ultimate analysis was made on each of two samples of coal. Sample No. 1 was taken directly from the fifty pound bag in which it was supplied; sample No. 2 as taken from the receiving vessel used in the test program for the coal feeder, after it had been fed from the Petrocarb injector. The results of the analyses of both samples are given in Table 2. TABLE 1. COAL PARTICLE SIZE DISTRIBUTION Penn—Rillton Co. Grind B—2 Specification U. S. mesh size 10 20 30 40 100 200 pan wt. fraction on screen 0 4.5 15.5.14 35.5 12.5 18 Limestone and Dolomites A limestone and two dolomites were used in the experimental studies. The stones and their properties are given in Table 3. Various particle size ranges of these materials were used and are specified in the sections whera the experimental results are described. Ca S 04 A high purity form of anhydrous CaSO 4 was used in the regeneration studies. The material was purchased from W. A. Hammond Co. under the trade name “Drierite” in 10 x 20 mesh particle size range. EXPERIMENTAL PROCEDURES Cold Model Test Unit The measurements of bed expansion ratios, quality of fluidization and minimum fluidization velocity were made visually. The rate of solids transfer was evaluated by two methods. In one method solids transfer was permitted only from the prototype regenerator to the prototype combustor, and the increase in solids inventory in the combustor was measured over a timed interval. In the second method, transfer was accomplished in both directions, and a charge of painted 25 ------- TABLE 2. COMPOSITION OF COAL USED IN ESSO BATCH—FLUIDIZED BED COMBUSTION PROGRAM Source : Consolidation Coal Co. - Ark right Mine Proximate Analysis Wt. Percent Sample #1 Sample #2 Component ( bag) ( receiving tank ) Moisture 1.00 1.03 Ash 8.11 8.01 Volatile matter 36.86 36.69 Fixed carbon 54.03 54.28 Ultimate Analysis Wt. Percent Sample #1 Sample #2 Component ( bag) ( receiving tank ) Moisture 1.00 1.03 Ash 8.11 8.01 Total carbon 76.26 76.67 Hydrogen 5.30 5.30 Sulfur 2.66 2.49 Nitrogen 1.49 1.47 Chlorine 0.07 0.08 Oxygen (by difference) 5.11 4.95 Higher Heating Value, Btu/lb. Sample No. 1 Sample No. 2 14,045 14,070 TABLE 3. PROPERTIES OF LIMESTONE AND DOLOMITES Quarry Stone Chemical Analysis, Wt. Designation source — _ type 2 1 P Q2 _ 223 223 1359 Grove Lime Co. Limestone 97.0 1.2 1.1 0.3 0.2 (Stephen City, Va.) 1337 Chas. Pfizer Co. Dolomite 54.0 44.0 0.9 0.2 0.3 (Gibsonburg, 0.) Tymochtee C. F. Duff & Sons Dolomite 53.8 38.7 5.3 0.9 1.2 (Huntsville, 0.) 26 ------- limestone was added just after the combustor transfer catch pocket to serve as a tracer. The time required for the painted limestone to reach the regenerator was measured, and knowing the volume of the transfer line, the transfer rate was calculated. The stone used in the studies was a wide particle size distribution of calcined limestone (-6 to +20 U.S mesh) representative of solids from the limestone grinding equipment. Coal Feeder Test Unit Before a run, the coal feeder was checked to see that it contained 50-100 lbs. of coal. The feeder and receiving vessel (see Figure 8) were then pressurized and the back pressure regulator located downstream of the receiving vessel was adjusted so that the pressure in the vessel was less than the pressure in the feeder by the desired amount (routinely 3 psi) Flow of injection air was started by setting the injection air pressure regulator to about ten psi above the feed tank pressure, opening the injection air shutoff valve, and adjusting the metering valve until the desired air flow was indicated on the rotameter. The vibrator, which was attached to the feeder directly above the mixing tee, was turned on. If necessary, the back pressure regulator on the receiving vessel was readjusted to bring the feed tank to receiver t P to the desired value. Coal feed was started by opening the coal feed valve. Weight of the receiving vessel was recorded, typically every minute, so that the coal feed rate could be determined. To stop the flow of coal, the coal feed valve was shut. Injection air was left on for several minutes after the feed valve was closed so that the mixing tee and injection tube could empty. If injection air was stopped with coal remaining in the tee or injection tube, a plug usually formed when the feeder was restarted. Fluidized Bed Coal Combustion Unit Operation of the batch fluidized bed combustor can be divided into four phases: startup, ignition and pre-heating, coal feeding, and shutdown. Startup consisted of those activities preliminary to ignition of the propane burner. These activities included checking equipment to make sure it was ready for a run, checkout of the analyzer calibratLon, charging solids, turning on electrical circuits and the air compressor, turning on all cooling water systems (fluidizing grid, burner, coal probe, steam coils, condenser) and purge nitrogen systems (pressure taps, sight—glasses, d/p cells). To ignite the propane burner, air and fuel flows were set and the ignition electrode was activated. Safety devices shut down all flows if ignition was not obtained within ten seconds or if a flame—out occurred afterwards. A safety interlock prevented startup for 3 min- utes after an automatic shutdown to assure adequate purging of the combustor. Subsequent to ignition, air flow and coinbustor pressure 27 ------- were adjusted to the values desired for making the run. All gas flows and pressure were controlled automatically. Preparation of the coal feed system for a run consisted of setting the flow of injection air and activating and adjusting the coal feeder to combustor t P control system. Coal injection could be started only after the temperature in the combustor was high enough for self—ignition of the coal to occur. Propane flow to the burner was stopped automat- ically at the same time that feeding of coal was started. An automatic safety circuit would shut down coal injection if the combustor tempera- ture dropped too low to ensure combustion of the coal or if the feeder to combustor Al ’ dropped below a pre—set minimum (about one psi). Data on the weight of the coal feeder vs. time was taken so that the feed rate of coal could be determined. Another method of estimating the feed rate was to observe the oxygen concentration in the off-gas from the combustor. A rapid rise in oxygen concentration was usually the quickest way of determining that a problem was developing with the coal feeding System. Temperatures in the combustor could be adjusted by regulating the amount of water entering the three pairs of cooling coils. The feed rate of coal could be adjusted by changing the flow of injection air. To shut down the combustor routinely, the coal feed valve was shut, fluidizing air was stopped, and a nitrogen purge was started to preserve the solids. Flow of injection air was kept on for several minutes so that the coal feed line could be cleared of coal. All water flows were reduced. Solids could be discharged from the reactor (by blowing them out of a port located just above the fluidizing grid) after the unit bed cooled overnight. Fluidized Bed Regeneration Unit Operation of the regenerator was similar to the combustor except that no coal feeding was involved. Preliminary startup of the unit included checkout of the calibration of the analytical instruments, activation of electrical systems and air compressor, check of the propane system, and turning on all cooling water systems. The empty regenerator was first preheated to minimize condensation of water after solids were added. The burner was then shut off and solids were charged through a port located near the top of the unit. Reheat- ing of the bed was then begun and pressure and air flow were adjusted to meet the conditions selected for the run. If desired, air could be electrically preheated. When the bed was near the temperature of the run, reducing conditions were established by increasing the flow of fuel. Additional fuel could be added through the burner or, alterna- tively, directly into the bed. Auxiliary air could also be added, higher in the bed, to reduce formation of CaS. When auxiliary air was not used, it was necessary to provide a small stream of 02 near the top of the regenerator to oxidize any sulfur which formed and which might otherwise plug lines. 28 ------- Shut down was accomplished by stopping the flow of fuel and air and reducing the pressure in the system to atmospheric. A nitrogen purge was applied to prevent combustion of any carbon deposits in the reactor and to prevent moisture from reaching the solids until they had cooled sufficiently to be removed (overnight). Laboratory Fixed Bed Units Steel Reactor — The reactor was packed with 40 gm of calcined limestone or dolomite and inert alundum beads. The bed was heated to the operating temperature in the presence of N 2 . The proper blend of inlet gases was then made, analyzed in the continuous IR analyzers and then admitted downf low into the reactor. The outlet gases were continuously analyzed. Ceramic Reactor — The lower section of the reactor was first packed with Alundum beads to serve as the preheat section. For most runs, 40 gm of bed material was then placed above the preheat section. In some runs, the bed charge was increased to 60 gin. After heating to the proper temperature in the pre- sence of a stream of N 2 passing upf low through the bed, the feed gases were blended to give the proper composition and then admitted upf low into the reactor. The product gases were analyzed continuously. Since this unit was used to study regeneration of sulfated material, the pri- mary component in the product gas was S02 formed by the reduction of the sulfated bed material. The S02 concentration in the product gas decreased as the bed material was reduced, and the runs were terminated when the SO 2 concentration decreased to a very low level. The time of the runs ranged from 1 to 2 hours, during which time five to ten times the stoichiometric amount of reducing gas was passed over the bed. After the bed cooled, the charge was removed, weighed and analyzed for total sulfur content. 29 ------- SECTION V DEVELOPMENT OF EXPERIMENTAL EQUIPMENT AND PROCEDURES COAL FEEDER Introduction The coal feeding system for the pressurized batch—fluidized bed combus— tor must be capable of metering 6—28 lbs./hr. of powdered coal (—30 mesh) into the combustor against pressures of up to 10 atmospheres. A Petrocarb ABC injector, which is a small version of the solids feeder to be used with the combustor of the Miniplant, was chosen to feed coal into the batch combustor. An experimental program was undertaken to test the suitability of the Petrocarb feeder f or doing this. The purpose of the program was to determine what modifications of the Petrocarb system were needed to make It operate satisfactorily and to gain experience in feeding coal prior to startup of the combustor. Argonne National Laboratory had tested the suitability of the Petrocarb ABC injector for feeding 20—40 lbs./hr. of coal, limestone, and mixtures of coal and limestone, into a bench scale combustor— regenerator (1). Coal feed rates averaged 31.6—45.7 lbs./hr. for runs at atmospheric pressure to 97.8—179.3 lbs./hr. for runs with 75—82 psig pressure in the receiving vessel. Many runs were character- ized by feed rates that fluctuated widely. Argonne terminated work with the Petrocarb feeder when they concluded that it could not feed coal at steady rates in the flow ranges they required. Argonne made only minor modifications to the system supplied by Petrocarb. At Esso, it was hoped that the Petrocarb system could be made to perform satisfactorily if additional modifications were made. Theory of Operation Operation of the Petrocarb solids feeder depends on gravity flow of solids through an orifice and subsequent pneumatic conveying of the solids to the receiving vessel. These two operations have been studied by others with many different types of solids (2). For example, it is known that the reduction in “viscosity” of a bed of powder will result in an increased flowrate through an orifice. This reduction in viscosity can be accomplished by pressurizing the bin by means of a connection atop a closed vessel or by adding an aeration bleed at the base of an open bin. Either procedure results in a small stream of gas leaving with the solids through the orifice. This additional stream of gas “lubricates” the flow at the orifice and reduces the viscosity at this point. By means of this principle, solids can 30 ------- be made to flow through smaller openings than would be possible by simple gravity flow. Increasing the pressure, as can be done in the Petrocarb solids feeder, increases the lubricating effect at the orifice and permits solids to flow at a greater rate. Solids flow through the orifice at the bottom of the Petrocarb feeder and enter a mixing tee where a moving stream of air picks up and conveys the solids through a tube to the receiving vessel. The relationship between pressure drop per unit length of tube and superficial velocity is similar to that obtained for a tube with gas only flowing. However,pressure drop is higher due to the force required to maintain the solids in suspension and move them along with the gas stream. At constant superficial velocity, pressure drop increases with increasing solids loading. In order to maintain a given feed rate of solids through the tube, the superficial gas velocity must not fall below a certain critical value, called the saltation velocity. Below the saltation velocity, which is a function of solids flow rate for any particular gas—solid system, particles settle out in the tube. It is interesting to compare the saltation velocity for horizontal transport with the minimum velocity required to transport solids vertically, called the choking velocity. Experimental data indicates that saltation velocities and choking velocities are identical for uniform sized particles. However, for mixed size material saltation velocities are three to six times as great as choking velocities. This is in accord with numerous qualitative statements appearing in the literature on pneumatic transport to the effect that the velocity required to convey materials horizontally is several times as great as that required to convey vertically. Experimental Results Work with the Petrocarb solids feeder was divided into two sections. The purpose of the initial work was to modify the feeder so that it could deliver coal smoothly and dependably at the required rates. The modified equipment would be used to collect data on feed rate of coal vs. feed tank pressure, feed tank—receiving vessel pressure differential, and injection air flow rate. Table 4 gives results of 68 runs made with the modified Petrocarb Model 16—1 ABC injector. A 3/16 inch orifice was used for all runs except no. 22, in which a 7/32 inch orifice was used. After run no. 22, the taper of the original 3/16 inch orifice was made steeper (see Figure 14). Modification of Petrocarb Feeder The Petrocarb ABC injector was designed for much higher solids feed rates than 6—28 lbs./hr., which is required for use with the combustor. Therefore, modification of the unit to permit lower feed rates was 31 ------- 11716” DIA. 3/16” DIA. MATERIAL: 316 STAINLESS STEEL Figure 14. Coal feeder orifice 3/64” 32 ------- Table 4. COAL FEED RATES WITH A MODIFIED PETROCARB MODEL 16—i ABC INJECTOR Orifice size = 3/16 inch Coal = —30 mesh Receiving Feed tank vessel Air flow Average Run pressure, pressure, rate, coal feed No. Injection Hose psig psig SCFM rate, lb/hr. Comments 1 1/4 in. x 25 ft. 5 0 1.0 1.0 Injection hose wrapped around receiver. Polyflow Plugged after 15 mm. 2 20 0 2.0 7.8 Injection hose plugged after 35 mm. 3 1/2 in. x 25 ft. 5 0 1.0 10—44 Hose wrapped around receiver. Erratic Rubber (Petrocarb flow. Plugged after 13 mm. supplied) 4 5 0 1.0 20 5 5 0 1.6 6.5 Hose in 30 inch loops on floor. Ran well. 6 10 0 2.3 45 7 20 0 2.3 125 8 5 0 2.3 7.6 Conditions of Run No. 5 repeated. 9 1/4 in. x 25 ft. 20 0 2.3 8.0 Hose in 30 inch loop on floor. Flow Polyf low stopped after 19 mm. 10 20 0 2.3 21 Flow stopped after 9 mm. High static charge on injection hose. 11 20 0 2.6 17 Flow stopped after 10 mm. 12 1/4 in. x 15 ft. 10 0 2.3 12.3 Injection hose in one large loop on copper floor. 13 5 0 1,6 7.3 Coal flow increased when injection air 1,0 28 decreased. ------- 19 20 21 22 40 60 40 40 70 70 60 70 36 2.0 56 2.0 38 2.0 1.3 1.0 0.7 36 2.0 67 2.0/1.3 67 1.3 57 1.3 67 2.0 21.1 24.0 7.2 22.3 25.1 34 12.8 16.8/37.0 32.0 26.6 20.2 Table 4 (continued). COAL FEED RATES WITH A MODIFIED PETROCARB MODEL 16—1 ABC INJECTOR Receiving Feed tank vessel Air flow Averaged Run pressure, pressure, rate, coal feed No. Injection Hose psig psig SCFM - rate, lb/hr. Comments 14 1/4 in. x 15 ft. 20 16 1.6 18.0 copper 15 16 17 18 1/4 in. x 15 ft. copper 1/4 in. x 15 ft. Stainless steel 23 24 25 26 27 28 29 Vibrator used. Vibrator used. Coal flow increased as injection air decreased. Vibrator used. Ran only 10 minutes. 7/32 inch orifice used for this run. Taper of orifice made steeper. Higher coal rate ran only 3 mm. 70 70 60 40 40 70 70 67 67 57 37 37 67 67 1.3 2.0 2.0 1.3 2.0 2.6 1.6f1.3 28.7 18.6 16.5 19.8 8.2 9.8 24/31.6 ------- Table 4 (continued). COAL FEED RATES WITH A MODIFIED PETROCARB MODEL 16—1 ABC INJECTOR Run No. 30 Injection Hose 1/4 in. x 15 ft. stainless steel Receiving Feed tank vessel Air flow pressure, pressure, rate, psig psig SCFM 70 67 2.6 Average coal feed rate, lb/hr. 8.3 Comments 31 70 67 2.6 13.8 Bored out tee—injection tube connector. 32 40 37 2.0 5.8 33 60 . 57 2.0 15.4 34 60 57 1.3 30.6 35 40 37 1.3 21.9 36 60 57 2.3 8.0 L i i 37 38 70 20 67 17 2.0 1.3 24.3 10.1 39 20 17 0.65 29.9 40 60 57 1.6 28.8 41 40 37 1.0 35.4 Frequent plugging 42 40 37 1.6 16.3 43 70 67 2.3 15.8 New Syntron vibrator installed. 44 70 67 1.6 25.9 45 70 67 1.3 40.9 46 60 67 1.6 28.6 47 70 67 1.6 29.1 48 60 57 1.3 34.6 ------- Table 4 (continued). COAL FEED RATES WITH A MODIFIED PETROCARB MODEL 16—1 ABC INJECTOR Receiving Feed tank vessel Mr flow Average Run pressure, pressure, rate, coal feed No. Injection Hose psig psig SCFN rate, lb/hr. Comments 49 1/4 in. x 15 ft. 121 118 2.6 27.5 Pressure at rotameter = 155 psig stainless steel 50 135 132 3.5 15.7 51 135 132 4.0 6.1 52 135 132 3.1 25.6 53 106 103 2.6 19.4 54 106 103 2.6 26.3 Adjusted calibration of DP cell—was off slightly for runs 49—53. 55 135 132 3.5 12.6 56 121 118 3.1 17.7 57 106 103 3.1 17.1 58 91 88 2.2 24.9 Cleaned filter before run. 59 121 118 3.1 17.4 60 135 132 3.1 23.0 61 135 132 4.0 6.9 62 121 118 2.6 26.1 63 121 118 3.5 9.7 64 106 103 3.5 4.4 65 106 103 3.1 8.8—12.7, Problems with air flow. avg. 9.9 ------- Table 4 (continued). COAL FEED RATES WITH A MODIFIED PETROCARE MODEL 16—1 ABC INJECTOR Receiving Feed tank vessel Mr flow Average pressure, pressure, rate, coal feed _______________— psig psig SCFM rate, lb/hr. Comments 91 88 2.6 15.4 Minor problems with air flow. 135 132 3.7 11.3 Minor problems with air flow. 135 132 3.3 20.9 Minor problems with air flow. Run No. Injection Hose 66 1/4 in. x 15 ft. stainless steel 67 68 ------- necessary. Factors which affect feed rate are size of the orifice (5/16 inch supplied) and size of the tube connecting the feed tank with the receiving vessel (1/2 inch dia. x 25 feet hose supplied). An important part of the experimental program was to find that combination of orifice and injection hose size which would permit satisfactory operation over the range of required flowrates. The first task was to establish a satisfactory size for the orifice. Orifice sizes of 1/8 inch, 5/32 inch, 3/16 inch, and 7/32 inch were tried but, with the two smaller orifices, coal would either not flow at all or flow intermittently upon tapping the orifice assembly. The two larger orifices plugged occasionally, but the 7/32 inch size produced coal feed rates which were too high. Thus, 3/16 inch was the optimum size to be used with the coal on hand (—30 mesh). After run no. 22, the taper of the orifice was made steeper. This reduced plugging. Several injection tubes were tested. The first was a 3/16 inch x 25 feet copper tube (I.D. = 0.1275 inch) which was coiled around the receiving vessel so that the diameter of each coil was about 18 inches. The tube plugged frequently, and it was difficult to remove the plugs as this required disconnecting the tube, straightening it, and tapping an end to remove the plug. Polyflow tubing of 1/4 inch x 25 feet size (I.D. = 0.17 inches) was subsequently tried and had the advantage that it was translucent, so that plugs could be easily spotted, and that it could be readily put into any desired shape. When this tubing was wrapped around the receiving vessel, plugging usually occurred after only a few minutes. However, when placed in large (30 inch) loops on the floor, plugging occurred much less frequently. A 12 foot length of Polyf low tubing was also tested and gave considerably higher coal feed rates than did a 25 foot length. Use of Polyf low tubing was terminated because a static charge developed on the tubing during operation, which contributed to plugging. A copper tube, 1/4 inch x 15 feet (0.19 inch ID.), formed into one large loop was tried next. Decreasing the tube length made it easier to make a single loop and less plugging occurred than in longer tubes. After run No. 14, a vibrator was attached to the mixing assembly at the bottom of the feed tank. The vibrator caused both the mixing assembly and the entire injection tube to vibrate and reduced plugging at both the orifice and injection tube. Plugging had continued to be a problem with the 1/4 inch x 15 feet (I.D. 0.19 inch) copper tube, and it was replaced with a 1/4 inch x 15 feet (I.D. = 0.18 inch) tube of stainless steel. The stainless steel tube proved to be a significant improvement over the copper since it plugged less frequently. The improvement is probably due to the smoother surface of the stainless tube, which, reduces the tendency of the coal to stick. Plugging of the injection tube is a serious problem because to remove 38 ------- the plug it is usually necessary to disconnect the injection tube; hence, plugging of the tube during a run with the combustor would make shutdown of the combustor necessary. Plugging of the orifice in the mixing assembly is less serious as tapping in the area of the orifice usually removes the plug and coal flow resumes. S fl additional modification made in the equipment was to bore out the inside of the tube fitting which connects the 1/2 inch diameter mixing tee located at the bottom of the feeder to the 1/4 inch diameter injection tube. The abrupt reduction in diameter had caused coal to accumulate in the tee and eventually plug. By boring out the connector, so that a gradual transition was made from 1/2 inch to 114 inch diameter, the plugging problem was eliminated. This change caused also a moderate increase in coal flow. Calibration of Modified Feeder A program was undertaken to determine how feed tank pressure, feed tank—receiving vessel pressure differential, and air flow rate affect the feed rate of coal in the modified Petrocarb injector. Data col— lected would make it possible to pre—set operating conditions needed to produce the desired coal feed rate. For a given orifice, coal feed rate appears to be a function of pressure drop across the orifice of the coal feeder and the pressure in the feeder. The size of the injection tube and the injection air flow affect coal flow indirectly by causing more or less of the total pressure drop from feeder to receiving vessel to occur across the orifice rather than through the tube. Figure 15 shows the effect of air flow rate on coal feed rate for four pressures of up to 70 psig in the coal feed tank. AU data shown are for a 3/16 inch orifice, 0.18 inch I.D. x 15 feet long stainless steel injection tube, and a feed tank to receiving vessel pressure differential of 3 psig. Increasing the air flow rate through the injection tube causes a higher pressure drop across the injection tube and a correspondingly lower pressure drop across the orifice. Hence, coal flow decreases with increasing air flow. It should be noted that not all of the air which passes through the air rotometer enters the injection tube. A fraction of the total air flow enters the feed tank to compensate for the small flow of air leaving the feed tank with coal through the orifice. Thus, the actual air flow in the injection tube is slightly less than that indicated by the air rotometer setting. Figure 16 shows the effect of feed tank pressure on coal flow rate at constant air flow rate and feed tank to receiver P. An increase in feed tank pressure permits coal to flow more readily through the orifice. 39 ------- F I g u re Effect of injection air flow rate 15 on coal feed rate — tow pressure I I 3/16 INCH ORIFICE, 1/4 INCH x 15 FT STAIN- LESS STEEL INJECTION TUBE, P (FEEDER- RECEIVER) 3 PSI, —30 MESH COAL - 0 20 30 40 50 60 70 AIR ROTAMETER SETTING, 010 (10O% = 3.3 SCFM) 40 80 A 42 40 38 36 34 32 30 W26 24 w 22 -J < 20 0 C -) 18 16 14 12 10 8 6 4 U I I I 1 ------- Figure 16 Effect of feed tank pressure on coal feed rate 3/16 INCH ORIFICE, 1/4 INCH x 15 FT STAINLESS STEEL INJECTION TUBE, AIR ROTAMETER = 40% (1.3 SCFM), t P (FEEDER-RECEIVER) = 3 PSI, -30 MESH COAL 20 30 FEED TANK PRESSURE, psig 40 50 60 70 80 90 100 100 90 80 — 70 60 — 50 — 40 30 20 -o w I- w U i U- -J 0 C —) I I I I I_ 0/0 I I I I I 1_ 10 41 ------- Figure 17 shows the effect of air flow rate on coal feed rate at four levels of feed tank pressure from 91 to 135 psig. At any given pressure, small changes in air flow rate change the coal feed rate considerably. The effect is similar to that shown in Figure 15 for data at lower pressures. Figure 18 shows the effect of feed tank pressure on coal feed rates for three different air flow rates. Plugging at the orifice and in the injection tube occurred much less frequently in runs made at higher pressures (above 90 psig) in the coal feed tank than at lower pressures. For example, during runs 49—64 (See Table 4) the injection line plugged only twice. The orifice plugged during one run, but the vibrator was not turned on at the time. These runs represent a combined total of about eight hours of operation. The flow of coal was also steadier at higher pressures. A regression analysis was made on the data of runs 49—68 with receiver pressure and rotameter setting as the independent variables. The least squares correlation is F = 36.0 + O.37lP — 0.873R where F = coal feed rate, lb./hr. P = pressure in receiving vessel, psig R = air rotameter reading, % of scale This correlation is valid only over the range of Pand R in the experi- mental work (runs 49—68). These limits are 88—132 psig and 50—90%, respectively. The least squares lines are shown as the solid lines of Figures 17 and 18. Details of statistical tests of the model are given in Table 5. BURNER The burner for the batch fluidized bed regenerator had to handle a wide range of flowrates, up to a maximum of about 30 SCFM air and 1.5 SCFM fuel (60% propane + 40% propylene), in proportions ranging from strongly oxidizing to strongly reducing. Furthermore, it had to operate at pressures of 1—10 atm. and accept preheated air at temperatures up to about 800°F. As pressure, temperature, and fuel/air ratio increase, the tendency of a flame to flash back from the supporting grid into the burner tube also increases. These factors, combined with the need to produce a burner with a reasonably long service life, made the development of the burner a formidable task. All burners that were designed and tested consisted of two main 42 ------- Figure 17 Effect of air flow rate on coal feed rate — high pressure 3/16 INCH ORIFICE 1/4 INCH O.D. x 15 FT STAINLESS STEEL INJECTION TUBE, P (FEED TANK-RECEIVER) = 3 PSI 1 118 PSIG 40 50 60 70 80 90 AiR ROTAMETER SETTING, % (100% 4.4 SCFM) 100 RECEIVER PRESSURE = 132 PSIG U LU I— U i LU LL -I C C) 32 30 28 26 24 22 20 18 16 14 12 10 8 6 4 I I I I 43 ------- Figure 18 Effect of feed tank pressure on coal feed rate 3/16 INCH ORIFICE, 1/4 INCH x 15 FT STAINLESS STEEL INJECTION TUBE, L P (FEED TANK-RECEIVER) 3 PSI FEED TANK PRESSURE, psig 140 32 30 28 26 24 - w I — 0 U U -J Q C) 22 20 18 16 14 12 10 8 6 44 2 0 70 80 90 100 110 120 130 44 ------- Table 5. RESULTS OF RE(RESSION ANALYSIS OF COAL FEED TEST DATA Dependent variables = F (coal feed rate, lb../hr.) Independent variables = P (receiver pressure, psig) R (rotarneter reading, %) Correlation Matrix P R F P 1.000 R 0.687 1.000 F —0.126 —0.781 1.000 Standard error of estimate = 2.033 F ratio for regression 114.1 Fraction of explained variance = 0.931 Determination of correlation matrix = 0.528 - Residual degrees of freedom = 17 Std. error Co—variance Variable Coefficient of coeff. T—ratio F—ratio ratio Constant 36.0 P 0.371 0.042 8.86 78.46 0.47 R —0.873 0.058 14.98 224.3 0.47 45 ------- parts: the burner tube and the grid. (See Figure 10). The burner tube was essentially the same in all designs; it was a stainless steel tube about 6 inches long with a diameter of 2 inches. The burner tube was welded to a flange which attached to the bottom of the burner section of the regenerator. The burner tube contained baffles and was filled with 1/8 inch alumina beads to adequately mix fuel and air, which entered under the flange. Fuel entered through a sparging tube Which distributed the fuel radially into the surrounding air. The position of the sparging tube in the burner tube could be adjusted to achieve optimum mixing. The best mixing occurred when the openings of the sparging tube were positioned at the tee where the air and fuel lines joined. It was also found that the number of baffles in the burner tube could be reduced from nine to two without any noticeable change in the appearance of the flame. The first burner grid tested was a 1/8 inch thick plate of micro— metallic mesh, which was welded to the top of the burner tube. Directly under the grid was a water cooling channel in the shape of a circle with four spokes. ttThermonfl cement was placed between the mesh and the cooling tubes to promote good thermal contact. This burner assembly performed well in terms of its range of applicability for burning very lean to very rich. The reliability and durability of these grids was a problem, however. Three grids had been tried; the first and third lasted for only one to two runs while the second lasted considerably longer. Typically, cracks would develop in the mesh or around the weld. Variability in the service life of different grids was attributed to slight differences in construction. In order to learn more about the cause of the loss of burner grids, the fine micrometallic mesh was replaced with coarser “Rigimesh”. However, the Riginiesh grid, which caused a considerably lower pressure drop, burned out after only a few hours of operation. Since no clear indication was then available on how to design a burner with a reasonable life expectancy, a new burner assembly was built to allow easy replacement of burner grids. This new burner had a grid Which, rather than being welded, was attached to the top of the tube with a single bolt. This burner has been used successfully with the combustor for some time, perhaps because it has been used only for oxidizing flames. A new burner grid for the regenerator burner was designed and built. It was a water cooled stainless steel plate containing 48—1/16” holes. The grid worked well at 10 atm with an oxidizing flame, but a flashback occurred with a reducing flame. Another water cooled burner grid was built, this time of brass, to take advantage of the high thermal conductivity of this material. To help prevent flashback, the grid had small holes (400—l/32 inch), and a layer of micrometa].ljc mesh was placed under the grid to act 46 ------- as a flame arrester. This sandwich was held onto the burner tube by eight small studs. The burner performed very well for two runs (115R, 116R) but a flashback occurred during preparations for an additional run. The brass grid was not damaged but the baffles in the tubular section of the burner were destroyed. The probable cause of the flashback was a failure of the gasket sealing the metal- lic mesh flame arrester to the burner tube. The design was modified to provide a more positive seal of the mesh to the burner tube. The mesh was welded onto the top of the burner tube and the brass grid was placed over the mesh and attached with studs. A gasket was still needed under the brass grid but, since it was directly under the cooling channel of the grid, it was kept cool. This design eliminated the second gasket, which would be exposed to higher temperatures. A radiation shield for the tubular section of the burner, made out of copper tubing (water—cooled), was also added to preclude the possibility of exceeding the self— ignition temperature of the fuel—air mixture before it reached the grid. The modified burner worked very well except for the breakage of a stud on two occasions. Disassembly and inspection of the burner after a series of seven runs showed it to be in excellent shape. The last burner grid developed was of similar design but screwed onto the burner tube, so that the delicate studs could be eliminated. The mesh flame arrester was not welded to the burner tube but was seated on a shoulder cut into the wall of the burner tube at its top. Performance of this new burner has been excellent. It ignites easily, even at elevated pressures, the flame is stable over almost all operating conditions, and it should have a very long life expectancy. COOLING COILS The combustor contained three pair of serpentine wound cooling coils which extended from about 6 inches above the fluidizing grid upwards for about 6 feet. The coils were made from 1/4 inch diameter stainless steel tubing and had a total surface area of about 6.3 sq. feet. Each pair was centered at a flange so that one coil could be removed without disturbing the others. Water flow into each pair of coils was independently controlled. House water was denjineralized before its pressure was increased with a pump. The water flow then split into three branches which connected with the cooling coils. Each branch contained a rotanieter followed by a metering valve. The branch connecting with the middle pair of coils also contained an automatic control valve which could be used instead of the manual valve. 47 ------- During the first runs with the combustor, house water entered the cooling system at about 40 psig and no pump was used to increase pressure further. With this arrangement, control over the amount of water entering the coils was unsatisfactory. Because of expansion produced during the formation of steam, back pressures greater than the inlet pressure of the water developed in the coils. This resulted in water entering the coils one slug at a time, a limit to the maximum amount of water that could be fed into the coils, and very little control over the flow rate of water. It should be noted that it was not feasible to control heat removal from the conibustor by using enough water to maintain single phase flow. This is because the rate of heat removal is proportional to the log mean T, which would be nearly constant over the range of flowrates required to maintain single phase flow. Also, the use of air as a cooling medium had been investigated, but the flow of air required to keep the coil temperature at a safe level would be so high as to remove an excessive amount of heat from the bed. Control over water flowrate was improved by increasing the inlet pressure of the water by means of a pump. This created a much larger pressure drop across the control valve than existed previously. With increased pressure drop, downstream pressure fluctuations affected flowrate less. Initially, inlet pressure was increased to 150 psig; but this pressure was not high enough and a Delavan Model 20001 piston pump was then installed. This pump can deliver two gallons/mm. at pressures up to 500 psig; however, because of pressure limitations on the rotameters, operation was limited to about 300 psig. Fluctuations in water flow were greatly reduced and control over water flowrate was satisfactory with water inlet pressure at this value. COMBUSTOR CYCLONES AND FILTER Two cyclones normally removed flyash and entrained solids from the combustor. It was found; however, that during the first several minutes of some combustion runs, temperatures at the cyclones were not high enough to prevent condensation of steam from the off—gas. When this occurred, efficiencies of the cyclones were reduced sufficiently to allow flyash to reach the system filter. The filter then plugged rapidly, disturbing the pressure in the combustor, the pressure drop from the coal feeder to combustor, and the flow of coal. The problem to be solved was to prevent the filter from plugging with flyash. The easiest solution was to add an electrical heater and a third cyclone downstream of the first two cyclones. The heater was used during the early stages of runs to keep the temperature of 48 ------- the off—gas entering the third cyclone above its dew point, so that this cyclone could effectively remove flyash from the off—gas. This system worked well although it was needed on only several occasions when steam condensation was a problem. REGENERATOR BED HEATERS The lower section of the regenerator contained two 15 KW electrical heaters which were used during startup to bring the bed up to the desired temperature more rapidly than was possible with the propane burner alone. The heaters also smoothed the temperature distribution over the fluidized bed by increasing temperatures in the upper portion of the bed. The heaters were constructed from resistance wire which was wrapped around the alumina reaction tube located in the center of the shell of the regenerator. Two layers of ttFiberfraxU insulation surrounded the windings to provide a cushion for thermal expansion. The annular space between the alumina tube and the inside of the steel shell was filled with castable refractory insulation. A temperature runaway in the bed of the regenerator occurred during the first run and the heater windings were destroyed. New heaters were installed which lasted for eight runs. The heaters were examined and it was concluded that failure was caused by a combination of corrosion and overheating. Corrosive gases could have worked their way through cracks in the alumina reaction tube and contacted the heater windings. A new alumina tube plus a new set of heater windings were installed. The lower 10 inches of the tube was made removable so that it could be easily replaced if it cracked. It was decided to operate the new heaters at reduced power to minimize the possibility of overheating. Unfortunately, the new heaters failed after only two runs. Failure was attributed to chemical attack by reducing gases and SO 2 . It was decided not to re—install heaters as no simple way was known to prevent corrosive gases from reaching the heater windings. OFF—GAS COOLERS Heat exchangers were used to cool off—gases from the regenerator and combustor to temperatures just above their dew points. The original unit used with the regenerator was an American Standard shell—and—tube heat exchanger with ninety—l/ 4 inch stainless steel tubes through which gas flowed. The off—gas cooler had to be removed from the system when it was discovered that a number of tubes had cracked. Examination of the unit showed that some tubes had developed radial cracks and that many of the tubes were plugged with solids. The cause of the cracks was attributed to differences in thermal expansion between the tubes and a ceramic header, located near the gas inlet. 49 ------- Heat transfer calculations, based on experimental data, indicated that the original exchanger had too much area. As a result, the exchanger was rebuilt using three 1/2 inch stainless steel tubes inside the original shell. No problems have since developed in over eight months of operation. The heat exchanger used with the combustor contained nine 1/4 inch diameter (.049 inch wall) tubes of 304 stainless steel. Gas flowed through the tubes. No operating problems have developed in about five months of operation. MATERIALS OF CONSTRUCTION After the fluidized bed regenerator was started up, a number of materials problems developed caused by the combination of high temperatures and the chemical environment. Fluidizing Bed Grids The original fluidizing bed grid used in the regenerator was made of cast alumina. In the first few runs made in the regenerator, high temperatures occurred which resulted in melting of the bed charge, apparent reaction of the molten charge with the grid and cracking of the grid. The grid was destroyed in these runs. A photograph of the fused material and the grid is shown in Figure 19. Analysis of the molten material indicated the presence of CaSO 4 , CaO and Ca 3 A120 6 . Examination of the phase diagram of the CaO—A1 2 0 3 system indicated that a melt of calcium oxide and calcium aluininate could exist at temper- atures as low as 2550°F. Apparently, CaO was reacting with the alumina grid to form the oxide—alurninate mixture which melted at the lower temperatures. An additional run was made with an alumina grid under lower and better controlled bed temperatures conditions. Although the damage to the grid was much less, it still showed some signs of reaction with the bed charge,was bowed upward and cracked. A cast zirconia grid was then fabricated and tested. Zirconia not only has a higher melting point than alumina but also forms no low melting point mixtures with CaO. The zirconia grid showed no signs of reaction with the bed material, but was still slightly bowed and cracked. In order to impart maximum strength to the zirconia it must be cured at 3000°F. Available curing ovens could only cure at 400°F and the grids were therefore, not cured to maximum strength. Also, no assurance could have been given that the fully cured zirconia could stand up to the repeated oxidizing—reducing cycles in the regenerator. Therefore, it was decided to abandon the concept of cast ceramic grids and fabricate the water cooled stainless steel grid described in Section IV. The water cooled steel grid has worked very well after some early pro- blems of water leakage at a welded joint were solved. The grid was examined after being in operation for a number of months. The only ap- parent sign of damage was a small raised section which may have been caused by a blocked cooling channel. 50 ------- . U, Figure 19. Fused regenerator bed material and fluidizing grid ------- Ceramic Reactor Liner Melting and fusion of the bed charge in the regenerator which destroyed the alumina fluidizing grids also severly damaged the mullite ceramic bed liner. Mullite is an impure alumina, containing a signifi- cant amount of silica. It was-believed that lo :me tiiig ilicate systems could have formed by reaction of the liner with the molten bed charge. Since the silicates have even lower melting points than aluminates, it was decided to replace the mullite liner with an alumina liner. The alumina reactor showed no signs of reaction with the bed material during succeeding runs. However, due to its poorer resistance to thermal shock, the alumina tube cracked and had to be periodically patched with ceramic cement. Refractory Insulation The Litecast 7528 refractory insulation used in the regener- ator and combustor vessels has held up very well with use. The only damage sustained by the refractory occurred in the regenerator when the bed charge melted and attacked the ceramic liner and portions of the refractory also. Since the problem of bed charge melting was solved, no further damage to the refractory has occurred. The Bubbalite high temperature refractory used in the burner zones of the combustor and regenerator has also held up very well. The only damage sustained so far occurred when a failure in the regenerator burner caused direct flame impingement on the refractory. This resulted in the melting of a small portion of the refractory but was easily repaired by patching. 52 ------- SECTION VI EXPERIMENTAL RESULTS COLD MODEL TEST UNIT Effect of Simulated Steam Coils onFluidization Characteristics Tests were conducted to determine the internal baffling effect of the simulated coil upon the fluidizing and slugging character of the bed, and the anticipated particle entrainment. The tests were conducted in the combustor section of the CMTU over the following range of parameters: Superficial Bed Velocity (ft/sec) 4, 6, 8, 10, 12 Settled Bed Height (ft) 2, 2.5, 3, 3.5, 4 Vessel Pressure (psia) 20, 30, 40, 60 It was visually observed that the simulated coils inserted in the acrylic model produced a well fluidized bed, especially in the region where the expanded bed is below the upper coil level. It is therefore anticipated that this type of coil configuration will promote very satisfactory mixing and heat transfer in a high temperature combustor. Test results indicated that expanded bed height increases with increasing settled bed depth, pressure and superficial bed velocity (Fig. 20 and 21). Thus the tendency toward slugging and high entrainment rates is increased by operating with deeper beds, higher velocities and higher pressures. The insertion of the simulated coils, in addition to improving the bed fluidization behavior, also significantly reduced the slugging nature of the fluid bed (Fig. 22). For example, with a 3 feet settled bed at 40 psia and a superficial bed velocity of 6 feet/sec., the average expanded bed height was reduced from 132 inches to 78 inches when the simulated coil was inserted. This implies that the coils provide a modulating or baffling effect and that combustor outage height can be reduced without an increase in particle entrainment. Determination of the minimum fluidizing velocity was also obtained by increasing the fluidizing air velocity until the bed particles were just visually fluidized. The results indicated that the minimum fluidizing velocity is not affected by bed depth and decreases with increasing reactor pressure (see Fig. 20 and 21). 53 ------- Figure 20 Effect of superficial bed velocity on expanded bed height 2 ft. settled bed, simulated heat transfer coil 4- . ., — “I I I I SETTLED BED HEIGHT (24 IN) I I I I I 4 6 8 SUPERFICIAL BED VELOCITY, ft/sec 10 140 120 100 — 80 — 60 — 40— 20 U) U = w = LiJ = w uJ 0 x w 4u 0 2 12 54 ------- Figure 21 Effect of superficial bed velocity on expanded bed height 3 ft. settled bed, simulated heat transfer coil U) c i ) C) C w I Ui = LU LU x LU 140 120 100 80 60 40 20 0 0 2 4 6 8 10 SUPERFICIAL BED VELOCITY, ft/sec 12 55 ------- Figure 22 Effect of heat transfer coil on expanded bed height 40 psia Pressure I I I ct,, ‘c / (<, S 6 8 10 WITHOUT COIL WITH SIMULATED COIL (68 IN. LONG) SUPERFICIAL BED VELOCITY, ft/sec 180 . / a) 0 w F— w w w 0 x I J / / / 160 — 140— 120 — 100 — 80 60 40 / / / S F / / / / / / / I . I. / 2 4 I I I 12 14 56 ------- The minimum fluidizing velocity is considered to be most likely a function of fluidizing gas density rather than just pressure. Since the tests were run at the same temperature, the density varied directly with the pressure. At 20 psia, for the stone particle size distribution used, the minimum fluidizing velocity is 2.4 ft/sec; at 60 psia it is 1.6 ft/sec. It should be noted that the same density effect is also most likely true for the cases involving the expanded bed height. That is, it is probably the density of the fluidizing air, rather than just the reactor pressure, that influences the expanded bed height of the fluidized system. Continuous Transfer of Solids Studies were conducted in the Cold Model Test Unit to determine the effect of various parameters upon the solids transfer rate between prototype reactors and to measure this transfer rate. Tests were conducted at reactor vessel pressures to 40 psia, with settled bed heights of 3.0 feet. Since the solids were transferred using the pulse air technique, the main parameters varied were pulse air flow rate and pulse frequency, i.e., pulse on/off cycle. The results of the two- way transfer runs at 40 psia are presented in Figure 23. The tests verified the ability of the continuous system to transfer solids in both directions while operating under pressure and maintaining good bed level control in both vessels. Transfer rates of 6.1 lb/mm were attained while transferring solids continuously in both directions at 40 psia vessel pressure. The rate of solid transfer between vessels was controlled by varying the pulse cycle and also the pulse air flow rate. Observations of the continuous transfer operations in the CMTU indicated that the 1-inch I.D. catch tube inserted in the prototype combustor was a limiting or choking restriction on the maximum transfer rate attainable. However, the solids reservoir installed in the prototype regenerator (Figure 2) functioned satisfactorily in that it permitted freer solids flow and a superior transfer line seal than the catch tube. When transferring solids in one direction, from regenerator to combustor, solids transfer rates as high as 9.9 lb/mm were attained. This is attributed to the ample sizing and configuration of the regenerator transfer reservoir. Transfer line fittings and restrictions not only limit the solids flow rate, but induce hold-ups and plugs in the transfer line. As these line restrictions in the CMTU were systematically eliminated, improved solids transfer rates were achieved. It is anticipated that with adequate pipe line sizes and a minimum of restrictions, the Miniplant will be able to achieve satisfactory solids transfer with this pulse technique. The Miniplant transfer rate design is 680 lb/hr (11.3 lb/mm) under maximum conditions. 57 ------- Figure 23 Solids transfer rate 10 I REACTOR VESSEL PRESSURE, 40 psia SUPERFICIAL BED VELOCITY, 8 ft/sec E 8 PULSE AIR FLOW RATE, 2 SCFM LU I— _________ 6 LU . LL 2 4 / (I, -J 2— 0 I I 3 2 1 2 3 OFF/ON RATIO -p ON/OFF RATIO PULSE CYCLE ------- COMBUSTION STUDIES Fluidized Bed Combustion Unit Because of operating problems coal has not been fed continuously into the batch combustor for more than about twenty minutes at a time. These problems involved poor control over the flow of water entering the cooling coils, heavy accumulation of flyash in the off-gas filter, and unsatisfactory operation of the coal feeding system. The first two difficulties have essentially been resolved and progress has been made toward improving the operation of the coal feeding system. The last major obstacle appears to be preventing plugging in the coal probe and orienting the probe so that coal burns uniformly throughout the fluidized bed. Table 6 presents operating parameters for coal combustion runs. The compositions of effluent streams are given in Table 7. When evaluating this data, it is important to recognize that, because coal was fed only for brief periods, conditions in the coinbustor (e.g. coal feed rate, temperature, off-gas composition) were unsteady. Because there were problems with control of temperature, and also with the feeder, coal rates had to be kept low. As a result, excess air levels were far higher than they would normally be. Calcination of Limestone and Dolomite A 23 lb. charge of Grove #1359 limestone was calcined in the combustor in about one hour at an average temperature of 1560°F. It was desirable to calcine in the combustor because this eliminated the need to transfer stone to the combustor from the vessel in which it was calcined. A nearly stoichiometric air-propane mixture was used and the superficial velocity was 8-10 ft./sec. The total pressure was about 2.6 atm; the Co 2 concentration in the off-gas was approximately 2l7 ,. At the conclusion of calcination the bed temperature rose to 1700°F and the level of C02 in the off—gas fell to 15%, about the same level of CO 2 as observed before calcination. The combination of temperature rise (due to the end of endothermic calcination) and drop in CO 2 concentration gave a definite indication of the completion of calcination. Subsequent chemical analyses showed that more than 99.9% of the carbonate had been converted to oxide. Subsequently, a 16.6 lb. batch of -8 + 25 mesh Tymochtee dolomite was haif-calcined by heating in the combustor for about six hours at 1200-1400°F. Proper haif-calcination should have produced a material containing only CaCO 3 and MgO. Subsequent analysis of solids showed that some decomposition of the calcium portion of the stone had also occurred 83 wt. L (CaCO +MgO) 17 wt. 70 (CaO MgO) 59 ------- Table 6. OPERATING PARAMETERS - COAL COMBUSTION RUNS Coal: Arkwright Mine, —30 mesh Limestone: Grove #1359, —7 mesh, calcined. Dolomite: Tymochtee, —8+25 mesh, haif—calcined. Average coal rate varied over range of 10.0—18.0 lb./hr. Average coal rate varied over range of 15.0—17.1 lb./hr. C. Actual air — stoichiometric air /. Excess air = x 100 Stoichiometric air Charge, Run No. material, lbs. Settled height, bed ft. Press., psig Superficial velocity, ft./sec. Input Coal, Air, lb/hr SCFM 66—2690—ll2a limestone, 23 2.0 103 3•9 13 • 3 a 66—2690—112b limestone, 23 2.0 103 43 15 5 b 66—2690—127 dolomite, 16.6 2.0 63 5.4 8.6 40.0 66—2690—128a dolomite, 16.6 2.0 63 5.8 7.2 41.5 66—2690—128b dolomite, 16.6 2.0 104 5.6 8.8 62.0 66—2690—l28c dolomite, 16.6 2.0 104 6.1 9.2 63.5 Average bed temperature, °F 1480 + 100 1660 + 100 1480 + 100 1560 + 100 1510 ± 100 1640 + 100 % Excess airc 43 23 104 153 209 203 ------- Table 7. COMPOSITION OF EFFLUENT STREAMS - COAL COMBUSTION RUNS SO 2 , ppm NO, ppm C0 2 , % CO, % H 2 0, % N 2 , SO 2 Removald(%) SO 2 , ppm NO, ppm C0 2 , % CO, % 1120 , % N 2 , 7. SO 2 Removal (%) 66—269 0—112 a Measuredc Avg. Range 240 steady —— 310 230—370 12.0 10.6 6.5—19.0 —— 0.9 0.1—2.0 not collected 9.6 3.3—15.5 76.8 Avg. 145 — — 340 8.5 7 3.5 10.4 77.5 Range 100—140 160—550 3—11 .02—3.0 66—2690—112b Measured ______ Range steady 150—380 5. 2—23. 6 0.1—2.7 5.8 not collected 3.7 7.6 1.9—15.4 76.5 2.9 12.4 77.7 86.0 66—2690—128a Measured Range 125—550 250—550 4.5—9.5 .03 0.01—1.0 not collected 10 3—15 72.2 Run No. Caic . Caic. Avg . 1780 250 —— 290 13.9 13.5 —— 0.7 5.0 6. Run No. 84. 4 66—2690—127 Measured Caic . 1100 Caic. Avg . 900 250 —— 450 6.9 7.0 .03 not collected 12 2—17 86.8 61 ------- Table 7 (continued). C0?JIPOSITION OF EFFLUENT STREA1 IS — COAL COMBUSTION RUNS aA T O SO removed in bed. bA all C 2 —) CO 2 and H —v H 0, Corrected for water condensed pri r to analysis. Calculated S02 — observed SOl x 100. SO 2 removal = Calculated SO 2 Run No. 66—2690— 128b Measured 502, NO, ppm C0 2 , ppm CO, % H 2 0, % N 2 , % SO 2 Removal 66—269O—128c Measured ______ Range 100—600 170—490 5.0—11.5 .01—.4 Caic. Avg. - Range Calc. Avg. 700 320 225—550 700 320 —— 380 350—430 —— 340 5.7 8.0 7.5—9.5 5.7 8.0 —— .03 .01—0.4 —— .02 2.4 not collected 2.4 13.9 9 5—14 13.9 78.0 78.0 not collected 11 5—17 (%) 54.3 54.3 62 ------- Removal In two runs (11 2 a, ll2b) made with calcined Grove limestone, SO 7 removal was 84.4 and 86.O7 . It was estimated that conversion of CaO to CaSO in the bed was 2.O7 prior to run No. ll 2 a and 5.47,, after run No. ll2 . The combined coal feeding time for these runs was 47 minutes; 11.5 lbs. of coal were burned. Several brief runs were made with half-calcined Tymochtee dolomite; a total of about 15 lbs. of coal were burned, which corresponds to a conversion to sulfate of about l07 . It can be noted in Table 7 that the percentage of sulfur removal decreased with each subsequent run. This is probably due to loss of dolomite through attrition, which appeared to be very high, rather than to deactivation caused by sulfate loading. Even during the first combustion run with dolomite (127) sulfur removal was only 877g. This is not surprising in light of poor temperature control. Table 6 shows that the bed temperature for this run was 1480 ± 100°F. It was not possible to determine an average bed temperature more precisely because of large temperature changes during the run and steep temperature gradients in the bed. Poor temperature control was due to inadequate flow of water through the cooling coils, unsteady coal flow, and combustion of most of the coal in the immediate vicinity of the coal inlet point. Progress has been made to correct these problems. NO Levels —x Average concentrations of NO were typically 300-400 ppm although the instantaneous concentration fluctuated widely because the coal feed rate was unsteady. NO concentrations would undoubtedly have been considerably lower hadXit not been for the high levels of excess air, which averaged between 23 and 209 percent. Attrit ion Attrition during runs with calcined Grove limestone appeared to be fairly low; however, attrition was very high during runs made with half—calcined Tymochtee dolomite. With the latter material, about one-half of the bed wound up as fines in the cyclone diplegs. Reasons for the high attrition rate are being examined to determine if the stone, as it appears to be, or equipment (or both) was responsible. 63 ------- Fixed Bed Units Reaction of CO and NO under Simulated Combustion Conditions — The reaction of CO and NO under simulated combustion conditions was studied in the stainless steel fixed bed reactor. This study was car- ried out to extend earlier work which had identified this reaction as one which could be responsible for reduced NOx emissions observed in the fluidized bed combustion of coal. Calcined limestone (No. 1359) or dolomite (No. 1337) was used as bed material. In the study, pres- sures were held at 1 or 10 atm, temperature was 1600°F. Premixed gases were blended to maintain the CO and NO levels at either approx- imately 1000 or 2000 ppm each. The C02 level was held at 0 or about 15%, the 02 level was held at 0 or about 2.5%. Residence time at 1 atm pressure was held constant at 0.3 sec. At 10 atm pressure, residence time was varied from 0.4 to 3 sec by varying the feed gas flow rate. However, at residence times below about 0.5 sec, the flow rate was too high to maintain the temperature at 1600°F and the temperature fell to 1400°F at the highest flow rate studied. A summary of the runs is given in Appendix Table A—l. In the absence of C02, the reaction between CO and NO proceeds to 95—99% of the limiting reactant over calcined limestone or dolomite. This is shown in Table 8. The reaction proceeds in a 1/1 molar ratio of CO and NO, suggesting the reaction 2C0 + 2N0 2C0 2 + N 2 (3) As seen in Table 8, when C02 was added, conversion of the limiting reactant decreased to 19—26%. The effect of CO 2 was almost instanta- neous. In one case, the C02 blend was cut out of the feed and the flow of the CO/NO blend was increased to maintain a constant feed rate. The NO concentration fell from 625 ppm to 63 ppm in about 1 minute and then decreased more slowly to 22 ppm over the next 10 minutes. The lag time for the sampling and analytical system was measured at 50 sec, so the response of the system to the removal of C02 was virtually instantaneous. 64 ------- TABLE 8. NO—CO REACTIONS Bed Source CALCINED CALCINED LIMESTONE #1359 DOLOMITE #1337 Inlet Gas Comp . NO, ppm 1400 1800 860 1400 1990 840 CO, ppm 940 1870 990 900 2080 980 CO 2 ,% 0 0 17 0 0 16 Outlet Gas Comp . NO, ppm 400 20 640 350 240 680 CO, ppm 10 160 770 20 100 830 C0 2 ,% 0 0 17 0 0 17 Cony., , 99 99 26 98 95 19 Temperature : 1600°F Res. Time : 0.3 Sec. The effect of pressure was studied and, although increasing the pressure to 10 atm in the presence of CO 2 apparently increased the conversion, the increase was probably due to increased residence time. At equivalent residence times, increasing pressure appeared to decrease the conversion slightly. This is shown in Table 9. TABLE 9. NO-CO REACTIONS EFFECT OF PRESSURE AND RESIDENCE TINE Bed Source CALCINED CALCINED LIMESTONE #1359 DOLOMITE #1337 Press. (Atm. ) 1 10 10 1 10 Res. Time (Sec. ) 0.3 3 0.3 0.3 3 Inlet Gas Comp . NO, ppm 860 890 1150 840 840 CO, ppm 990 980 1240 980 980 CO 2 , % 17 18 13 16 16 Conv. 26 82 16 19 88 Temperature : 1600°F 65 ------- Changing the background gas from N 2 to argon appeared to increase the conversion very slightly, but the effect may not be significant. was added to the feed and appeared to increase conversion slightly. Thts is shown in Table 10. TABLE 10. NO—CO REACTIONS EFFECT OF 02 Bed Source CALCINED CALCINED LIMESTONE #1359 DOLOMITE #1337 Inlet Gas Comp . NO, 1150 970 840 980 cö, ppm 1240 1060 980 1080 C0 2 , % 13 16 16 15 02, % 0 2.4 0 2.3 Cony., % 84 95 88 91 Temperature : 1600°F Pressure : 10 Atm. Increasing the temperature from 1600 to 1700°F in the presence of CO 2 at 1 atm pressure had essentially no effect on conversion. The reason for the large CO 2 inhibiting effect is not understood. Carbonation of lime was considered but does not explain the effect since CaO does not carbonate at the conditions used in the 1 atm runs. At these conditions the inhibiting effect was pronounced. Carbonate is the stable form at 10 atm pressure and 1600°F but the oxide is the stable form at 1700°F. However, changing reaction temperature from 1600 to 1700°F at 10 atm had no effect on conversion. Inhibition by chemical reversibility was also considered, but does not explain the results. The most likely explanation for this effect is a kinetic limitation caused by the presence of the CO 2 . REGENERATION STUDIES Introduction A gas containing CO and I I2 can reduce calcium sulfate to the oxide at high temperatures. The reactions are CaSO 4 + CO 4 CaO + SO 2 + CO (4) CaSO 4 + ) CaO + SO 2 + H (5) 66 ------- Formation of CaS, via side reactions, is undesirable because large amounts of reductant are used and no S02 is produced: CaSO 4 + 4C0 — CaS + 4C0 2 (6) CaSO 4 + 4H 2 > CaS + 41120 (7) CaSO 4 can react with CaS to produce CaO but this reaction does not occur to a great extent: 3CaSO 4 + CaS > 4CaO + 4S0 2 (8) Reactions (4), (5), and (8) are favored by high temperatures whereas (6) and (7) are favored at lower temperatures. Gas composition has an effect through CO/C02 and 1 12/H 2 0 ratios. High ratios promote reactions (4) and (5) but promote reactions (6) and (7) to an even greater extent. There are at least nine process variables which can affect the perfor- mance of the regenerator. These are: 1) pressure 2) temperature 3) air/fuel ratio (concentration of CO + 112 in reducing gas) 4) space velocity (superficial velocity/bed depth) 5) manner of fuel introduction and use of auxiliary fuel 6) use of auxiliary air 7) particle size of bed material 8) nature of bed material (pure CaSO 4 , sulfated limestone or dolomite) 9) quality of fluidization To evaluate the overall performance of the regenerator several factors must be examined. The most important is concentration of SO 2 in the off—gas, which, of course, is to be maximized. Concentration can be considered both as an absolute molar percentage or as a percentage of the maximum concentration which is possible at equilibrium. Another index of performance is the conversion of CaSO 4 . Even a high concentration of is not alone sufficient to make performance of the regenerator accept- able if only a small fraction of the CaSO 4 charged has been regenerated. Moreover, CaSO4 can be reduced to CaS and CaO. It is desired to minimize the ratio of CaS/CaO in the regenerated product. Other indices of performance include the degree of attrition of the bed and the uniformity of temperatures in the bed (an indication of the quality of fluidization and gas—solids contacting). Fixed Bed Regeneration Studies Prior to the construction of the high pressure batch regeneration unit, a few regeneration runs were made in the ceramic tube fixed bed regenera- tion unit at pressures between 1 and 9.5 atm. This work was done primarily to measure SO 2 product concentrations at the higher pressures. Tempera- tures were held at 1900 or 2000°F. Anhydrous CaSO4 (Drierite) and 67 ------- sulfated limestone were used as bed materials. The reducing gas feed was blended from C0/N2 and C0 2 /N 2 mixtures to permit variation of the CO from 10 to 30% and the CO 2 from 10 to 20% in the feed. Gas linear velocities ranged from 0.16 to 1.45 ft/sec. corresponding to 20 to 150% of the calculated minimum velocity required to fluidize the beds. The results are given in Table 11. At lower pressures, 1 and 3 atm, S02 concentrations in the product peaked at about 7% using CaSO4 beds. These concentrations correspond to 35 and 50% respectively of the SO 2 concentration calculated at equilibrium. However, when the pressure was increased to 9.5 atm, the SO 2 concentration dropped to 1.6% or 34% of the equilibrium level. Two runs were then made with sulfated limestone obtained from earlier work at Esso and from the U.S. Bureau of Mines fluidized bed combustor. In these runs, the measured SO 2 concentrations were only about 0.5%, corresponding to about 10% of the equilibrium level. Moreover, the solids discharged from the reactor in the runs using sulfated limestone were partially agglomerated. This could have been due to softening of fly ash present in the sulfated limestone at regeneration temperatures. This could have then caused the bed to agglomerate in the absence of vigorous fluidization. The agglomeration may also have caused poor gas—solids contacting and low S02 product concentrations. Fixed bed studies were terminated at this point to permit operation of the batch, fluidized regenerator unit. Fluidized Bed Regeneration Studies Table 12 is a summary of runs made with the batch fluidized bed regener- ator. A more complete listing of the operating parameters and results for each run is given in Appendix Table A—2. Agglomeration of Bed The first runs made with the batch—fluidized regenerator resulted in fusion of the bed. Heat transfer calculations showed that temperatures on the top surface of the ceramic fluidizing grid could exceed the melting point of CaSO4 (about 2650°F). Additional calculations showed that radiation errors could result in thermocouples indicating tempera- tures several hundred degrees below the true temperature of the reducing gases. As a result, gas temperatures could have been high enough to melt CaSO 4 . It was decided to replace the ceramic fluidizing grid with a water— cooled stainless steel grid. This would preclude the possibility of localized melting of CaSO4 occurring when it contacted the hot surface of the grid. To reduce the temperature of the reducing gas, diluent nitrogen was added to the burner along with air and fuel. 68 ------- Table 11. RUN SUMMARY—FIXED BED REGENERATION STUDIES Bed Temp., Press., Feed Comp. Flow, Vel. Peak % of tnat’l. °F atm % CO % CO 2 SCFM ft./sec Vmf S0 2 % eguil . CaSO 4 1900 1 21 15 0.1 1.45 1.4 7.1 0.33 CaSO 4 2000 3 31 10 0.1 0.51 0.5 7.2 0.48 CaSO 4 2000 9.5 10 20 0.1 0.16 0.2 1.6 0.34 Suif. (a) 2000 9.5 10 20 0.1 0.16 0.2 0.6 0.13 Lime Sulf . 2000 9.5 10 20 0.19 0.31 0.3 0.4 0.08 Lime (a) Bureau of Mines source (b) Esso source (c) Velocity/minimum velocity to fluidize Charge wt. = 40 gm except 60 gm in run 36 Particle size = —18+20 mesh ------- Table 12. SUMMARY OF RUNS WITH BATCH FLUIDIZED BED REGENERATOR (—10+20 mesh CaSO 4 (Drierite) bed) Avg. Pressure, Feed Product Settled bed Superficial Observed Equil. % of Run No. temp., °F atm %(C0-I-H 2 ) CO/CO2 height, ft. velocity, ft/sec S02, % SO2, % eguil. -.1 0 66—2690-. 71 2040 3.1 2 0.013 2.8 2.7 3.4 12.9 26 72 1990 3.2 15 0.013 5.6 2.9 5.2 9.8 53 75 2100 6.2 9 0.016 5.4 5.4 7.5 11.3 66 78 1950 6.0 11 0.021 4.0 2.4 2.0 5.0 40 86 1870 6.0 15 0.026 4.0 2.4 1.8 3.8 47 89 1970 6.1 15 0.023 4.0 2.5 1.7 5.1 33 98 2030 9.0 15 0.021 2.0 4.0 3.1 6.7 46 100 2080 10 15 0.023 2.0 4.3 1.6 9.0 18 101 1920 10 14 0.017 2.0 2.0 1.1 2.7 41 105 b 1800 10 14 109 2030 ± 40 10 7 0.013 2.0 1.8 0.7 —— 2.0 3.5 3.0 1.9 6.0 37 50 115 2050 ± 20 10 11 0.048 2.0 3.6 1.9 6.7 28 116 1840 + 20 4.9 13 119 c 2000 ± 20 10 10 2020 ± 30 10 10 121 e 1900 + 30 5.0 17 122A 1840 + 20 5.0 13.5 122B 2000 + 20 10 14 l23 1970 + 20 5.0 13.5 0.040 2.0 4.1 1.1 0.064 2.0 3.5 — 2.9 1.5 0.059 2.0 3.8 1.6 —— 2.0 3.2 1 3 a —— 2.0 2.6 054 a —— 2.0 2.8 2 • 0 a —— 2.0 4.0 2 , 3 a 4.2 4.6 5.4 5.1 4.4 4.6 7.0 25 33 29 25 12 44 33 124 g 1940 + 40 10 23 0.061 4.0 2.6 20 a 2.8 72 l25 1950 + 30 10 0.16 2.0 3.0 062 a 3.0 21 126 gj 1880 + 80 10 0.062 4.0 3.0 12 a 2.4 50 Corrected for N2 added in au . air. Run terminated after 5 mm. fuel 6 inches above grid, aux. air grid overall 1.02 for 122A, .99 for 42 inches 122B, 1.06 above for Aux. air injected 12 inches above grid. Fuel injected 6 inches above grid. eAUX. fuel 6 inches above grid, aux. air 34 inches above grid overall = 10.8. 123. fuel 6 inches above grid, aux. air rid. .Sulfated limestone (Bureau of Nines) —2O+40 mesh CaSO4 (Drierite) 54 inches above ------- Computer calculations (3) were made to determine the flow rates of air, fuel, and nitrogen required to achieve the desired reducing gas com- position and temperature. Equilibrium gas composition and adiabatic flame temperaturewerefound as a function of equivalence ratio, pres- sure, and per cent dilution of air with nitrogen. Calculations were made over the range of 0.8—2.0 equivalence ratio, 1—10 atm, and 0—75% nitrogen dilution (a 75% dilution means that 75% of the air is replaced by N2). P n equivalence ratio of 0.8 is typical of that used during pre—heating of the regenerator (oxidizing conditions). Under reducing conditions, the equivalence ratio is greater than one. Figure 24 presents the major results of the computer study. Adiabatic flame temperatures are plotted as a function of the percentage of CO + H 2 in the reducing gas mixture for nitrogen dilutions of 0, 10, 15, 20, 25 and 50%. Lines of constant equivalence ratio are also shown. Curves indicating constant dilution are drawn as dashed lines between = 0.9 and = 1.2 because there is insufficient data in this region to draw the curves accurately. The actual maxima in temperature should occur at an equivalence ratio of slightly greater than 1. All the results of Figure 24 are for a total pressure of 8 atm. but no signi- ficant error would result if the data were used over the pressure range of 1 to 10 atm. From experience with the regenerator, it was found that the adiabatic flame temperature had to be below about 2900°F to be certain that no melting of CaSO 4 would occur. Hence, Figure 22 was used to determine the percentage of nitrogen dilution and the air/fuel ratio required to prevent melting at whatever concentration of CO + H2 was desired. The technique of diluting incoming air with nitrogen was found to be highly successful in preventing melting and agglomeration of the CaSO 4 bed. Pressure Runs were made over the pressure range of 3—10 atm. As pressure decreases, with superficial velocity constant, the amount of fuel added to the regenerator decreases. Hence, a minimum operating pres- sure exists which corresponds to a flow of fuel which is just sufficient to maintain the desired bed temperature. This pressure was about five atmospheres. The first runs were made at pressures of 3.1 and 3.2 atmospheres, respectively; however they were characterized by agglomera- tion and melting of solids and, as a result, there were very non—uniform temperature distributions in the bed. Thus, the average temperatures for these runs are not directly comparable to those reported for later runs, in which no agglomeration of solids occurred and in which temperatures were much more uniform. Also, somewhat higher air pre—heat temperatures were used in earlier runs (about 1000°F vs. 600—800°F for runs made later). 71 ------- Figure 24 Adiabatic flame temperature versus % (CO + H 2 ) at 8 atm total pressure O 2 4 6 8 10 12 14 16 18 20 22 24 26 28 % CO + H 2 c = 1.0 3700 3600 ct=O.9 3500 3400 3300 3200 3100 3000 2900 2800 2700 2600 2500 2400 2300 2200 2100 2000 1900 1800 1700 LL 0 uJ I- LU 0 LU I— 72 ------- Use of pre-heated air produces higher bed temperatures but increases the likelihood of melting solids in the regenerator or causing a flashback into the tubular section of the burner if the self-ignition temperature of the fuel-air mixture is exceeded. As the mole fraction of SO 2 in the regenerator off-gas increases, fuel costs for regeneration decrease. This provides an incentive to operate at low pressures because the mole fraction of SO 2 at equilibrium rises as the total pressure of the system is reduced. Hence, runs were made over a range of operating pressures. However, the percent approach to equilibrium was less at low pressures and, consequently, the mole fraction of SO 2 as not higher for runs made at low pressures than for runs made at higher pressures. Figure 25 is a plot of concentration of SO 2 , expressed as a percent of calculated equilibrium concentratior vs. pressure. If several runs are eliminated, for the reasons indicated in the figure, a trend can be observed toward increased approach to equilibrium at higher pressures. Figure 26 is a plot of concentration of SO 2 , expressed as a molar percentage, vs. pressure. The three runs with the highest SO 2 concentration were made early in the program and agglomeration of the bed occurred. Hence, temperatures in the bed were probably higher than indicated, which could account for high concentrations of SO 2 . The remaining data do not show any significant effect of pressure on concentration. Presumably, at higher pressures, the closer approach to equilibrium compensated for the lover equilibrium concentration. Temperature Runs were made over the temperature range of 1800-2100°F. Figure 27 shows that the equilibrium partial pressure of SO increases sharply with temperature in this range. Unfortunately, ttie risk of deactivating the solids by sintering, or deadburning, increases at higher temperatures, especially when the solids must undergo repeated cycles of sulfur absorption and regeneration. Hence, the optimum temperature for regeneration is the one which produces the maximum concentration of SO 2 over a number of cycles. Temperatures in the fluidized bed were judged by thermocouples spaced six to twelve inches apart. The height of bed must be known to determine which thermocouples are located in the bed. Since this height could only be approximated, the temperature of the bed had to be estimated. Another reason why temperature could not be determined precisely was that it often changed during a run. When this occurred, an average with respect to time was taken to calculate an average temperature. 73 ------- Figure 25 Regeneration studies — effect of pressure on approach to equilibrium SO 2 concentration o RUNS WITHOUT SECONDARY AIR AND/OR FUEL • RUNS WITH SECONDARY AIR AND/OR FUEL o LAB RUNS TEMPERATURES ARE GIVEN BELOW EACH POINT(°F ) 80 i i i i i t i i i 70 1940 LOW o SPACE 2100 HIGH VELOCITY TEMPERATURE 50 1990 BED FUSION, HOT o 2035 — SPOTS 1870 2O3Q s” 040 0 1950 0 1970—’• 18 0 30 ‘197O 2 OO — 2040— -o ‘ 1840 2020 (-) o” 1900 20 . 200O 0 2080 LARGE 0 • TEMPERATURE 10 — 2000 1845 CHANGES DURING RUN 0 I I I I I I I I I I 1 2 3 4 5 6 7 8 9 10 PRESSURE, atm 74 ------- Figure 26 Regeneration studies — effect of pressure on SO 2 concentration o RUNS WITHOUT AUXILIARY AIR AND/OR FUEL • RUNS WITH AUXILIARY AIR AND/OR FUEL TEMPERATURES ARE GIVEN BELOW EACH POINT (°F) 8 I I I I I I I I I 0 2100 HIGH TEMPERATURE 7 6— C -) 1- 0 BED FUSION, 5 1990 HOT SPOTS 0 E 0 4 I— U i 0 2040 o —205O 2— o ‘ 2030 1935 C-) 2000,2000 0 v - i • 1970 o —l95O S4—194O 2 •4—1880 4—1870 2O80- - —_195O 1970 “ —2O20 1 1840’ ’ 19O0 0 1920 0 • 1800 1845 I I I I I I I I I 0 0 1 2 3 4 5 6 7 8 9 10 PRESSURE, aim 75 ------- Figure 27 Regeneration studies — effect of temperature on SO 2 partial pressure at equilibrium E C t LsJ 1 -f ) —4 (1) LU -J C C a- (‘4 0 (1 ) 1.0 .9 .8 .7 .6 .5 ‘4 .3 .2 .1 0 1800 1900 2000 2100 TEMPERATURE, °F ------- Runs made at temperatures below 1900°F generally resulted in low levels of S02 in the regenerator off—gas. For example, of four runs made at temperatures below 1875°F, only one resulted in an S02 con- centration greater than 1.1% (volume) and two resulted in concentra- tions under 0.8%. On the other hand, for nine runs made in the tem- perature range 2000—2100°F, the lowest SO 2 concentration was 1.5% and five runs had average concentrations of over 2.0%. Figure 28 is a plot of mole percent of S02 vs. temperature. There is a definite trend toward higher SO 2 concentration at higher temperatures. In addition to the molar or volume percentage of SO 2 , the concentration of SO 2 , expressed as a percent approach to equilibrium, also increased with temperature. Figure 29 is a plot of the percentage of equilibrium concentration vs. temperature. The figure shows a trend toward closer approach to equilibrium as temperature increases. This trend is more clearly seen if several data points are omitted (noted in figure). Superficial Velocity and Settled Bed Height Varying superficial gas velocity and settled bed height can affect the concentration of S02 in the regenerator off—gas by changing both the contact time between gaseous and solid phases and the quality of fluid— ization. Decreasing the space velocity, e.g., by increasing bed depth, may not improve gas—solid contact if it causes a bed which is well fluidized to begin slugging. Space velocity can be defined as volumetric feed rate of gas based on average reactor conditions/volume of settled bed. It is equal to superficial velocity/settled bed height. Most runs, as Table 12 shows, were made with a settled bed height of two feet. The greatest bed height used was 5.6 feet. Superficial velocities ranged from 1.8 to 5.4 feet/second. Space velocities, based on condi- tions inside the regenerator, ranged from 1850 to 7700 hr . It is uncer- tain if changing settled bed height alone had a significant effect on the observed concentration of S02. Several earlier runs were made with different bed heights but these runs were characterized by poor fluid— ization, caused by melting and agglomeration of the bed. Thus, it is not possible to determine if different bed depths caused differences in results of these runs. It appears that a modest change in bed height affects regenerator prefortnance only to the extent that it affects quality of fluidization. The same can be said of superficial velocity. Toward the end of work with the regenerator, evidence was obtained which indicated that reducing space velocity would improve performance. In run No. 124 (see Table 12), the reactor space velocity was decreased to about one—third of that of many previous runs. At an average bed tem- perature of 1940°F, the concentration of SO2 was 2.0% or about 72% of equilibrium. This was a closer approach to equilibrium than was attained in any other run. From observations of the temperature 77 ------- Figure 28 Regeneration studies — effect of temperature on SO 2 concentration 8 7— • 3-6 ATM t o1OATM 1800 1900 2000 208C TEMPERATURE, °F 78 ------- Figure 29 Regeneration studies — effect of temperature on approach to equilibrium SO 2 concentration 80 o 1OATM • 5-6 ATM o LOW SPACE VELOCITY 70 — E 60— 50 — HIGH TEMP. EXCURSION — . DOWNSTREAM OF 02 0’ INJECTION POINT 0 .“ 105 .0 •o 0 LARGE TEMPERATURE 10 — CHANGES DURING RUN 0— I 1800 1900 2000 2080 TEMPERATURE, °F 79 ------- profile in the bed, it was believed that fluidization was not markedly different from earlier runs. Hence, the improvement in performance was attributed to an increase in gas—solid contact time. Concentration of CO + 2 in Reducing Gas The air/fuel ratio determines the concentration of CO and H 2 in the reducing gas. The fuel used was a mixture containing 60 weight per cent propane (C 3 H8) and 40 weight per cent propylene (C 3 H 6 ). A com- puter program (3) was used to estimate the air/fuel ratio necessary to give the desired concentration of CO + H 2 . The program calculated equilibrium concentrations as a function of equivalence ratio, , defined as = (air/fuel) stoichiometric/(air/fuel) operating An equivalence ratio of 1.4, for example, produces a gas product con- taining about 14% CO + H 2 . The percentage of CO + H 2 used in the reducing gas varied between 2 and 23%; however, the concentration most often used was about 15%. There appeared to be no significant effect of the concentration of reducing gas on the concentration of S02. Lower air/fuel ratios produce more CO + 112 and consequently higher CO/CO 2 and 112/1120 ratios. However, there was insufficient data to determine whether increased CO/CO 2 and 112/1120 ratios caused increased formation of CaS as would be expected from equilibrium considerations. Auxiliary Fuel Run 120 (2020°F, 10 atm) was the first run made with a portion of the fuel added above the fluidizing grid, directly into the bed of solids. The remainder of the fuel and all of the air entered through the burner and produced a strongly oxidizing flame. Additional fuel added above the fluidizing grid produced a reducing gas containing approximately 10 CO + B 2 . Injection of fuel above the fluidizing grid significantly affected the temperature distribution in the bed. Thermocouples located at points 6, 12, and 18 inches above the grid indicated temperatures less than 10°F apart. This is a much closer distribution than was ever achieved without help from the electrical bed heaters, which were inoperative. Not only is the uniformity of temperature significantly better when a portion of the fuel is burned in the bed, but combustion of fuel in this manner produces higher bed temperatures than introduction of the same amount of fuel into a burner located beneath the fluidizing grid. For example, run No. 123 (Table 12) was made at five atmospheres pressure with auxiliary fuel added 6 inches above the fluidizing grid (auxiliary air added 42 inches above grid). An object of this run was to determine 80 ------- the maximum bed temperature that could be attained at five atmospheres pressure using auxiliary fuel and preheated air entering the burner. The temperature achieved was 1970°F. This compares to a bed tempera- ture Of 1840°F, attained in run No. 116, in which all of the fuel and air entered the burner, with none entering the bed. Moreover, the total flow of fuel was about 10% less in run No. 123 than in No. 116. Another advantage of burning fuel directly in the bed is that the tendency of the bed to agglomerate is reduced. No melting or agglomeration of solids occurred in runs Nos. 120—124, made with auxiliary fuel, even though nitrogen was used to reduce the temperature of the burner flame in only two of these runs. Auxiliary Air A persistent problem with regeneration has been formation of large proportions of GaS, instead of CaO. Elemental sulfur and reduced sul- fur compounds have also formed and an oxygen stream near the top of the regenerator (above the fluidized bed) was provided early in this work to oxidize these substances to S02. For run No. 119, this oxygen stream was removed and, instead, air was introduced into the bed at a point 12 inches above the fluidizing grid. The expectation was that any CaS which formed near the bottom of the bed would be oxidized to CaSO4 or CaO upon contacting the air. Reducing gas was produced at the burner, in the usual manner. With good mixing in the fluidized bed, solids would be alternately exposed to strongly reducing and oxidizing (or midly reducing) environments. U-nfortunately, something less than this could be expected to occur because mixing of solids in the regenerator is poor. Poor mixing was indicated in previous runs where the temperature distribution in the beds of solids was poor and also by results with the cold model test unit which showed that slugging of solids occurred, rather than uniform fluidization. The results of run 119 (2000°F, 10 atm, air injected into bed) were different from most runs made previously at higher pressures. The average SO 2 concentration attained was 1.54% (corrected for dilution of the off—gas by N 2 which is present in the auxiliary air), or 33% of equilibrium, which is not unusual. However, the S02 concentration in the off—gas was nearly constant over the duration of the run, in previous work at 10 atm, the S02 concentration peaked sharply at the start of a run and continually decreased as the run progressed. An even more spectacular effect of auxiliary air was a sharp reduction in the amount of CaS formed during regeneration. Without auxiliary air, regenerated solids contained, typically, 40—70 mole percent GaS. With the addition of auxiliary air, the CaS content of solids was reduced to under one per cent in the majority of runs for which analyses were made. This is shown in Table 13. ------- Table 13. ANALYSIS OF SOLIDS FROM REGENERATION RUNS Run No. Temp., °F Pressure, atm. Sup. vel., ft./sec. % Stoich. air % Dii. N2 Set. bed ht., ft. Aux. air Aux. fuel 66—2690—98 2010—2060 9 4.0 66 16.3 2 NO NO 66—2690—100 2080 10 4.3 66 16.7 2 NO NO 66—2690—10 1 1910—1940 10 2.0 67 20 2 NO NO 66—2690—105 1720—1880 10 1.9 69 16.7 2 NO NO Run No. Temp., °F Pressure, atm. Sup. vel., ft./sec. % Stoich. air Z Dii. N2 Set. bed ht., ft. Aux. air Aux. fuel 66—2690—115 2050 10 3.6 77 0 2 NO NO 66—2690—116 1840 4.9 4.1 70 0 2 NO NO 66—2690—119 2000 10 3.7 78 5 2 12 in. above grid NO 6 6—2 690—120 2020 10 3.8 79 0 2 Nob 6 in. above grid Wt. % Mole % Wt. % Mole % wt. % 0.9 89.7 9.4 Solids Composition Note % Wt. % Mole % Wt. % Mole 7. Wt. 7. Mole 7. Wt.% CaS 47.1 50.6 62.9 66.8 73.4 73.9 73.5 69.1 cc CaO CaSO4 47.8 3.2 39.9 6.5 34.5 0.8 28.6 1.6 21.3 3.4 16.6 6.5 14.9 9.6 10.9 17.1 CaCO 3 + CaSO3 <1.8 <3.0 <1.8 <3.0 <1.9 <3.0 <2.0 <2.9 Solids composition Ca S CaO CaSO 4 CaCO3 + CaSO 3 Mole % Wt. 7. Mole 7 . 41.3 45.2 57.7 60.5 0.2 0.2 0.7 54.7 4.0 46.6 8.2 38.0 4.3 31.0 8.5 98.4 1.4 96.5 3.3 95.2 4.1 a ------- Table 13 (continued). ANALYSIS OF SOLIDS FROM REGENERATION RUNS Run No. 66—2690—122b 66—2690—123 Temp., °F 2000 1970 Pressure, atm. 10 5 Sup. vel., ft./sec. 2.8 4.0 % Stoich. air 72 72 %Dil. N 2 0 0 Set. bed ht., ft. 2 2 Aux. air 42 in. above grid 42 in. above grid Aux. fuel 6 in. above grid 6 in. above grid Solids composition Mole % Wt. % Mole % Wt. % CaS 8.3 10.0 0.2 0.3 CaO 88.5 82.8 ‘98.1 95.7 CaSO4 3.2 7.2 1.7 4.0 CaCO 3 + CaSO3 a Neg11g1b1e Solids exposed to oxidizing mixture (107% stoich. air) for several minutes after run. ------- Particle Size All but one run was made with —10 + 20 mesh material. To test the effect of particle size, run No. 126. was made with —20 + 40 mesh Drierite. Unfortunately, problems with equipment necessitated a shutdown several minutes after the start of regeneration. The unit was restarted several days later with the bed still in place. The temperature profile in the bed indicated poor fluidization. It is probable that moisture condensing in the reactor during pre—heating had caused some of the bed to agglomerate. Hot spots developed and the run had to be terminated. Average bed temperature was 1880 + 80°F. Examination of solids indicated that the bed had agglomerated but lit- tle fusion had taken place. The average concentration of S02 in the off—gas was 1.2% although a peak concentration of 1.9% (corrected for N2 addition in auxiliary air) was observed. Based on an average bed temperature of 1880°F, these levels represent approaches of 50 and 70 percent, respectively. How- ever, when the uncertainty in the average bed temperature is taken into account, the approach to equilibrium could be anywhere in the range of 35—100 percent. Hence, it cannot be clearly determined if SO 2 con- centration was increased by reducing particle size. Nevertheless, if poor fluidization is taken into account, an approach to equilibrium of only 35% still appears high when compared to other runs in which agglomeration of the bed had occurred. Use of Sulfated Limestone Drierite was the bed material for all runs except No. 125, in which —10 ÷ 20 mesh sulfated limestone from the Bureau of Mines was used. Early in the run, a hot spot developed near the point at which secondary air entered the bed. For a brief period, temperatures at this position approached 2370°F. The temperature profile in the bed was much poorer than usually observed with Drierite. In addition, a poor temperature profile was noted when the bed was preheated prior to regeneration. Examination of solids showed that much of the bed had agglomerated in the reactor. Fused pieces which were quite hard were also found. It is likely that the bed had begun to agglomerate at the outset of the run (possibly during preheating). Diluent nitrogen was introduced early in the run; however, by this time, high bed temperatures were already observed. The average SO 2 concentration was only 0.62% although a peak concentration of 1.57% was observed, which corresponds to 52% of equilibrium. Quality of Fluidization Quality of fluidization is probably the most poorly controlled and most difficult to evaluate of all factors which affect performance of the regenerator. The best means available to determine how well solids 84 ------- were fluidized was to check the closeness of the temperature distri- bution in the bed. In several runs, where solids agglomerated, tem- perature differences across the bed were as high as 500°F. A typical span in temperature for beds in which no agglomeration occurred was 150°F. Where auxiliary fuel was used, a spread in temperature of 50°F was common, but the improvement in temperature distribution was probably caused by the additonal fuel and not by better fluidization. Analysis of Solids Solids from some regeneration runs were analyzed for CaSO4, CaO, CaS, CaCO3, and CaSO 3 . The analytical techniques that were used are given in the Appendix. Table 13 gives results of the analyses. No signi- ficant amounts of CaCO3 or CaSO 3 were found. The amount of unconverted CaSO 4 was usually small. Run Nos. 119, 122b, and 123 were made with auxiliary air; solids from these runs had levels of.CaS which were sharply lower than solids from runs made without auxiliary air. Figure 30 is a plot of the molar ratio of CaO/CaS vs. temperature for runs made without auxiliary air. Increased temperatures appear to favor higher CaO/CaS ratios. All data except for run 116 are for ten atm. pressure. Run 116, made at five atm, has a higher CaO/CaS ratio than would be expected at this temperature from data at ten atmospheres. This might indicate that reduced pressure favors higher CaO/CaS ratios. Attrition Attrition in the batch fluidized bed regenerator was very low, typically 1—2% of the weight of Drierite charged. 85 ------- Figure 30 Regeneration studies - effect of temperature on CaO/CaS ratio in regenerated soI ds 2.0 1.8 — • 10 ATM RUNS o 5 ATM RUNS 1.6 — 1.4— .1... E 1.0— L) V.L-) 0 0.6 — 0.4 — 0.2 — 0 1800 1900 2000 2100 TEMPERATURE, °F ------- SECTION VII DISCUSSION OF RESULTS REGENERATION STUDIES The S02 concentration measured in the product gas at 10 atm pressure averaged about 2%. This corresponds to 40 to 50% of the concentration calculated if the gas and solids were in chemical equilibrium. This suggests that the SO 2 concentration is limited by reaction rates. This is supported further by data which indicate that higher temperatures, higher pressures, longer gas/solid residence times and possibly smaller solids particles give SO 2 concentrations which are closer to the calculated equilibrium levels. The reaction rates can be increased by any number of changes in operating conditions. However, only a few of these options are open because of certain practical limitations. The two most likely ways of improving the con- version of CaSO4 to SO2 and CaO are to increase the overall residence time and to improve the gas/solids residence time distribution by improving the quality of fluidzation. For a given regenerator vessel diameter, the best way to increase the residence time is to decrease the gas linear velocity. However, for a given SO 2 production rate, a larger diameter regenerator vessel with an increased inventory of solids would be required. Increasing the residence time by increasing bed depth is not as attractive since the bed pressure drop would be increased, solids entrainment from the regenerator would be increased and the quality of fluidization would probably be poorer, giving a less favorable residence time distribution. Improvements in the residence time distribution could be made by optimizing the distributor design and also using shallower beds. The use of baffles to promote better fluidization does not appear to be as likely since the very high temperatures occurring in the regenerator pose a materials problem. Also, if the bed were to agglomerate, baffles would complicate the job of cleaning out the bed. Other ways of increasing the S02 concentration in the regenerator off gas do not appear to be promising. For example, higher temperatures will increase the probability of bed agglomeration, especially when fly ash carried over from the combustor is present. Higher pressures may increase the rate, but decrease the equilibrium concentration of S02, so the net effect is small or negative. Using smaller particles does not appear practical since it would require a larger combustor diameter to prevent excessive solids entrainment. Dropping pressure to 1 to 2 atm will result in an increase in SO 2 concentration by making the equilibrium more favorable, even if rates are decreased. However, this would require operating the regenerator at a lower pressure than the combnstor, and devel3pment of high temperature lock hoppers and valves would be necessary. In addition, it may not be possible to provide enough thermal energy in the hot reducing gases fed to the 87 ------- regenerator to carry out the regeneration at low pressure without diluting the product gas and decreasing the 502 concentration. If this is the case, the SO 2 product concentration could be limited by the anlomit of feed gas required to provide the needed energy and not by either chemical equilibrium or reaction rates. Operating at lower pressure would also require higher fluidizing velocities and/or larger reactor diameters. Introduction of auxiliary air directly into the regenerator bed pro— motes high conversion of CaSO4 to CaO with very little CaS formation. It is believed that air injection creates an oxidizing zone in the bed above the reducing zone. The solids move back and forth between the zones so that CaS formed in the reducing zone by over—reduction of CaSO4 can be oxidized back to CaSO4 or directly to CaO and S02 in the oxidizing zone. This ultimately gives high conversion of CaSO4 and high selectivity to CaO. Injection of at least a portion of the fuel directly into the bed also improves operation by providing higher and more uniform bed tempera- tures. There was also less tendency to agglomerate the bed when direct injection was used. Higher bed temperatures are due to less heat losses compared to the case where all fuel is burned in an adjacent burner zone. Burning part of the fuel in the bed also causes the thermal energy to be released over a greater volume and in direct con- tact with the bed solids. This promotes more uniform bed temperatures and probably results in the transfer of energy between the gas and solids at a lower gas temperature. This, in turn, decreases the chance for localized overheating of solids which can cause solids to soften and agglomerate. The best method of operating the regenerator appears to be to add both auxiliary air and part of the fuel to the bed at separate points. The optimum split between air and fuel added to the burner and directly to the bed and the location of the injection points must be determined. COMBUSTION STUDIES Initial SO 2 removal of about 857. has been measured with both —7 mesh calcined limestone and haif—calcined dolomite at combustion pressures of 5—8 atm. However, the runs were too short to draw any general conclusions about the suitability of either sorbent at these conditions. Also, the Tymnochtee dolomite showed very high attrition rates. Unless this is related somehow to the equipment or the manner in which the dolomite was used, it would appear to rule out the use of Tymochtee dolomite in a once through operation. The NO levels measured in the flue gas are very high considering the high combustion pressures. However, because coal feeding was very erratic during the runs, the excess air level was much higher than planned. It is believed that high NO emissions were due to the high oxygen levels in the bed. 88 ------- After resolving the coal feeding problem, the combustion program will continue testing other sorbents and measuring the effects of operating conditions on SO 2 and NO emissions. EQUIPMENT DEVELOPMENT Development of equipment became a significant part of the experimental program. The use of high temperature ceramic materials in the regenerator did not prove successful due to poor resistance to thermal shock and reaction with molten bed material. Water cooled stainless steel components performed very well under the same conditions. A burner was developed in this study to heat and provide reducing gas feed for the regenerator and also to preheat the combustor. This burner proved to be very versatile, operating over a wide range of pressures, oxidizing and reducing conditions and flow rates. Coal feeding at pressures up to 10 atm is possible with a modified Petrocarb injector, but careful attention must be paid to the operation to prevent plugging. Close control of the pressure differential between the injector and the combustor, the flow of injection air, lack of sharp bends or corners in the transfer line and coal particle size and dryness must be maintained. Some problems still remain in the proper design and location of the coal probe which injects the coal into the hot combustor. The continuous transfer of solids between the coinbustor and regenerator by the use of gas pulses is feasible. The system works well at ambient temperature and based on other studies (4) should work even better at higher temperatures due to increased gas viscosity. 89 ------- SECTION VIII PROGRAM F0R OPERATION OF MINIPLANT The overall objective of the Miniplant program is to demonstrate the continuous fluidized bed combustion process at conditions anticipated for commercial plants. The specific objectives are to • Verify the reductions in SO 2 emissions • Measure the reduction in NO emissions x • Verify particulate loadings in the flue gas • Measure erosion of a stationary target simulating a turbine blade surface in the flue gas • Test various coals in combination with various sorbents (limestones, dolomites). Select sorbents, based on location, suitable for the 30MW demonstration unit • Verify overall operability of the continuous system including turndown ratio arid methods • Test and develop (if necessary) system components • Measure, verify and/or demonstrate other items needed for the design and operation of the 30MW demonstration unit At the present time, no plans are being made to operate the regenerator section of the Miniplant. Therefore the operation will be based on the once-through use of sorbent. The primary operating variables whose effects will be studied are: combustion pressure and temperature, sorbent/coal feed ratio, coal and sorbent type. Other variables which cannot be as readily changed or whose effects may be less will be studied to a lesser extent. These are: fluidized bed level, coal and sorbent particle size, excess oxygen in the flue gas and coal rate. Fluidizing velocity is fixed by the coal rate and excess oxygen. The primary measured quantities will be: • Flue gas composition • Used solids composition • Temperature distribution in the bed • Solids loss and particle size change 90 ------- • Corrosion and erosion effects including the effects on the stationary test piece in the flue gas • Burning front location by measurement of intra bed gas compos it ions • Overall system response to changes in operating conditions, including response to turndown and turnup measures • General equipment performance From these measurements, the following will be calculated: • Fraction SO 2 reduction • Combustion efficiency • Heat transfer coefficients • Fraction of bed attrited • Heat losses The initial operation of the combustor will consist of testing various coal and sorbent combinations. Because of the somewhat limited objectives of this initial study, only the effects of the primary operating variables will be studied, the other variables ‘will be fixed. The run plan will consist of a block of eight runs for each sorbent and coal combination. The runs will be as follows: Temp (°F) Ca/S Mole/Mole Press (atm ) 1500 2/1 10 1600 2/1 10 1700 2/1 10 1700 4/1 10 1500 4/1 10 1600 4/1 10 1600 3/1 10 1600 3/1 5 The sorbents and coals tested will be chosen in cooperation with the EPA. The choice will depend on the results of current experimental studies sponsored by the EPA and the location of the 30MW demonstration plant. Other test conditjOflS, such as the possible use of precalcined sorbent, will be set before the tests begin. After the initial program has been completed, the program will be directed toward solution of problems which become evident during the initial program. At this time, the effect of the secondary variables can also be studied if warranted. Another option would be to study regeneration of sulfated sorbent, if that appears to be warranted at that time. 91 ------- SECTION IX REFERENCES 1. Jonke, A. A., et al., Reduction of Atmospheric Pollution by the Application of Fluidized Bed Combustion. Argonne National Laboratory, Monthly Progress Reports 32—35, June — September 1971. 2. Zenz, F. A., and Othmer, D. F., Fluidization and Fluid Particle Systems, Reinhold Publishing Corporation, 1960, p. 136—174, 313—350. 3. Gordon, S., and McBride, B., Chemical Equilibrium Program. NASA Lewis Research Center, November 1970. 4. Craig, J. W. T., et al., Study of Chemically Active Fluid Bed Gasifier for Reduction of Sulphur Oxide flnissions. Final Report OAP Contract CPA 70—46, Esso Petroleum Co., June 1972. ADDITION REFERENCES 5. Archer, D. H., et al, Evaluation of the YIuidi.zed Bed Combustion Process, Vol. I, 11, III, Westinghouse Research Laboratories, November 1971. 6. Hammons, G. A., and Skopp, A., A Regenerative Limestone Process for Fluidized Bed Coal Combustion and Desulfurization, Esso Research and Engineering Co., February 28, 1971. 7. Skopp A., et al., Studies of the Fluidized Lime—Bed Coal Combustion Desulfurization System, Esso Research and Engineering Company, December 31, 1971. 92 ------- SECflON X LIST OF PUBLICATIONS 1. Hoke, R. C., Shaw, H., and Skopp, A., A Regenerative Limestone Process for Fluidized Bed Coal Combustion and Desulfurization. Presented at the Third International Conference on Fluidized Bed Combustion, College Corner, c io, cktober 29 — November 1, 1972. 2. Bertrand, R. R., Hoke, R. C., Shaw, H., and Skopp, A., Combustion of Coal in a Bed of Fluidized Lime. Presented at the American Chemical Society National Meeting, 1icago, Illinois, August 26—31, 1973, also submitted for publication in Hydrocarbon Processing . 3. Hoke, R. C., Fluidized Bed Combustion of Solid and Liquid Fuels. Presented at the Symposium on Modern Developments in Combustion Technology, Pennsylvania State University, August 1, 1973. PATENT MEMORANDA SUBMITTED 1. Ruth, L. A., Hoke, R. C., Reduction of CaSO4 by direct injection of fuel into a fluidized bed. 2. Hoke, .R. C., Vath, E. C., Removal of SO 2 from flue gas followed by sulfite precipitation. 3. Ruth, L. A., Shaw, H., Vath, E. C., High pressure gas generating burner. 93 ------- SECTION XI GLOSSARY Abbreviation Definition atm atmosphere — unit of pressure BTU British Thermal Unit CMTU Cold Model Test Unit dia. diameter d/p differential pressure pressure difference temperature difference degree Fahrenheit F coal feed rate, lb/hr FRC flow recorder controller ft. foot gm. gram hr. hour I.D. internal diameter IR infra—red Kw kilowatt lb. pound mm. minute NDIR non dispersive infra—red O.D. outside diameter P pressure in coal receiving vessel, psig ppm parts per million PRC pressure recorder controller 94 ------- psi pounds per square inch psia pounds per square inch absolute psig pounds per square inch gage R injection air rotameter setting, % of scale SCFM standard cubic feet per minute sec. second sq. ft. square feet wt. weight equivalence ratio p micron — unit of length Conversion Factors — English to Metric Units English system tric equivalent Length inch 2.54 centimeter foot 0.305 meter Area square foot 0.093 square meter Volume gallon 3.785 liter cubic feet 38.32 liters Mass pound 453.6 grams Pressure pound per square inch 51.70 millimeters Hg atmosphere 760 millimeters Hg Temperature ° Fahrenheit 1.8 (°Celsius) + 32 Energy British thermal unit 252 calories 95 ------- SECTION XII APPENDIX TECHNIQUES FOR ANALYSIS OF SOLIDS Solids from some regeneration runs were analyzed for CaSO 4 , CaO, CaS, CaCO3, and CaSO3. The analytical techniques that were used are described below. SO 4 2 — The sample was treated with acidic BaC1 2 solution. The BaSO 4 precipitate was weighed. CO 3 2 — Acidic H 2 02 was added to the sample. The solution was stripped with N 2 and the gas passed through H 2 0 2 /H 2 SO 4 , Drierite, CuSO 4 , and Ascarite. C0 3 2 was determined from the weight gain of the Ascarite. +2 +2 Ca — Ca was determined by atomic absorption. SO 3 2 — The sample was treated overnight with HgCl , then heated with acid and stripped with N2. The gas was passed through excess 12, S 2 0 3 2 , and H 2 0. The scrubbers were combined and the solution back titrated with S 2 0 3 2 . SO 3 was determined from 12 consumed. — Since no S0 3 2 was found, the sample was treated with acidic 12 and the excess 12 titrated with S 2 0 3 2 . S 2 was determined from 12 consumed. o 2 — CaO was determined by difference. The excess number of moles of calcium over the moles of anions was assumed to correspond to 2. 96 ------- Table A-i. FIXED BED SIMULATED COMBUSTION RUNS Bed Mat’l. Temp., °F 1600 Press., psig 0 Gas rate, SCFM 0.11 Feed gas comp. Prod. gas comp. Cony., % CO, ppm 945 NO, ppm 1400 C02, % —— 02, % —— CO, ppm 10 NO, 400 C0 2 , % —— 02, % —— CO 99 of NO 71 Limestone 1359 a 1600 0 0.10 988 862 17 —— 770 635 17 —— 22 26 1700 0 0.10 988 862 17 —— 800 625 17 —— 19 27 1700 0 0.10 1870 1800 —— —— 160 22 —— —— 92 99 1600 135 0.10 985 893 18 — — 300 165 13 —— 70 82 1600 135 0.10 1870 1800 —— —— 160 23 —— —— 91 99 1600 135 0.10 900 1400 —— —— 10 375 —— —— 99 73 Dolomite 1600 0 0.1 900 1400 20 350 —— —— 98 75 1600 135 0.1 900 1400 —— —— 10 280 —— —— 99 80 1600 0 0.1 983 840 16 —— 830 680 17 —— 16 19 1600 135 0.1 983 840 16 —— 270 105 13 —— 73 88 1600 135 0.10 1080 978 15 2.3 100 112 16 2.3 91 89 1600 135 0.10 2080 1987 —— —— 85 180 —— —— 96 91 1600 0 0.10 2080 1987 — — —— 95 243 —— — — 95 88 ------- Table A—i (continued). FIXED BED SINULATED COMBUSTION RUNS a_ 18 ÷ 20 mesh, calcined. b.. 16 + 25 mesh, calcined. Bed Mat’l. Limestone 1359 a Feed gas comp. Prod. gas comp. Cony., % Temp., °F Press., ps ig Gas rate, SC 4 — CO, pp i NO, pp n CO2, % 02, % CO, ppm’ NO, ppm C0 2 , Z 02, % of CO NO 1600 135 0.10 1245 1150 12.7 —— 330 355 15.2 —— 74 69 1400 135 0.82 1245 1150 12.7 —— 1055 965 15.7 —— 15 16 1600 135 0.40 1245 1150 12.7 990 888 15.7 —— 20 23 1600 135 0.55 1245 1150 12.7 1040 905 15.7 —— 16 21 1550 135 0.70 1245 1150 127 —— 1080 940 15.7 —— 13 18 1520 135 0.66 1245 1150 12.7 —— 1095 935 15.7 —— 12 19 1600 135 0.10 2080 1987 —— —— 155 177 —— —— 93 91 1600 135 0.10 1080 987 15.2 2.3 255 115 15.6 2.2 76 88 1600 135 0.10 1060 972 163 2.4 50 90 16.8 2.5 95 91 1600 135 0.10 2080 1987 —— —— 38 163 —— —— 98 92 1600 135 0.10 1245 1150 12.7 —— 205 300 15.5 —— 84 74 1600 1. 00 135 135 0.55 0.10 1245 965 1150 978 12.7 —— —— —— 890 0 840 100 15.5 —— —— —— 28 100 27 90 ------- Table A—2. AIR, FUEL AND SOLIDS INPUTS FOR REGENERATION RUNS Injected above bed to prevent sulfur deposition 0.17 SCFM C02 also added. C 20 40 mesh; —10+20 mesh used in all other runs. in exit lines. a 02 0.83 0.80 —— 2.34 —— 0.98 —— 1.00 —— 0.76 Run No. Charge material, weight, lbs. Fuel 0.21 0.32 Input in_SCF1 _ — Air 5.0 4.9 Aux. fuel Aux. air N 2 0.67 0.64 66—2690— 71 72 Drierite, 9.2 Drierite, 18.4 75 Drierite, 9.8 1.11 20.8 78 Drierite, 13.1 0.46 8.4 86 Drierite, 13.1 0.52 8.4 89 Drierite, 13.1 0.53 8.9 98 Drierite, 6.6 1.35 20.5 100 Drierite, 6.6 1.58 24.0 101 Drierlte, 6.6 0.75 11.6 105 Drierite, 6.6 0.73 11.6 109 Drierite, 6.6 1.23 24.0 115 Drierlte, 6.6 1.38 24.3 116 Drierite, 6.6 0.93 15.0 119 Drierite, 6.6 1.00 18.0 —— 3.46—7.52 0.82—2.32 120 Drierlte, 6.6 0.92 26.3 0.53—0.92 0—2.63 121 Drierite, 6.6 0.42 11.6 0.21—0.33 2.42—4.26 l22A Drierite, 6.6 0.36 9.6 0.22 3.464.34 122B Drierlte, 6.6 0.74 19.4 0.44 7.84--8.95 123 Drierite, 6.6 0.45 13.8 0.38 2.42—6.28 124 Drierite, 13.2 0.61 17.8 0.63—0.82 3.5—7.0 125 sulfated limestone, 9.35 1.2 20.0 0.12 7.4 2.7 126 Drieritec, 15.6 0.7 21.0 0.1 7.0 3.98 4.80 2.85 2.33 2.48 2.98 2.30 2.30 2.4 2.2 2.2 ------- Table A—i. COMPOSITION OF EF}1 UENT STREAM FOR RE(ENERATION RUNS Concentrations have been F un1d by difference. Second column gives concentrations dN2 which is contained in auxiliary air. Calculated. Water not measured. Run No. S02, % CO, ppm C02, % Component, molar % or ppma 66—2690— 71 3.04 3 .. 37 1120, % 02, % 9.83 N 2 , NO, ppm 1570 40 C 11.6 l 2 . 9 10.6 ll. 7 64.8 718 C 125 139 C 72 4.78 5.23 1750 1910 13.7 14.9 14.6 16.0 8.54 58.2 63.7 150 164 75 6.94 7.52 2360 2560 14.7 16.0 14.3 15.5 7.71 56.1 60.8 260 280 78 1.81 2.01 2590 2900 12.3 13.7 15.2 16.9 9.87 60.5 67.2 —— —— 86 1.58 1.79 3630 4130 14.0 15.9 14.7 16.7 11.9 57.4 65.2 —— —— 89 1.56 1.73 3000 3320 15.6 17.3 21.9 23.5 9.9 50.7 56.3 —— —— 98 3.0 3.1 1200 1230 12.5 12.8 14.6 15.0 2.5 66.9 68.6 —— —— 100 1.5 1.6 —— —— 14.9 15.7 15.0 15.8 4.8 63.8 67.0 —— —— 101 1.0 1.1 —— —— 12.7 14.4 13.5 15.3 12.0 60.8 69.0 —— —— 105 109 115 0.6 2.8 1.8 1.1 3.0 1.9 —— —— -- —— —— -- 14.8 13.0 13.3 16.9 14.0 14.1 11.9 13.0 15.7 13 • 6 d 14.0 16 • 7 d 12.5 7.0 5.8 60.2 68.4 64.2 69.0 63.4 67.3 —— —— 160 —— —— 170 ° 116 0.96 1.07 —— —— 13.7 15.2 13.0 14.4 10.0 62.1 68.5 210 230 119 1.30 1.54 630 750 14.1 16.7 12.0 14.2 <1 72.6 67.6 190 220 120 1.57 1.57 9000 —— 15.2 —— 13.0 —— <1 69.3 —— 180 —— 121 1.05 1.29 870 1070 13.1 16.1 12.7 15.6 <1 73.2 67.0 105 130 122A 0.44 0.54 17 21 11.8 14.6 12.6 15.6 <1 75.2 69.3 100 120 122B 1.60 2.02 61 77 12.7 16.0 12.4 15.6 <1 73.3 66.4 110 140 123 1.90 2.30 260 310 10.5 12.5 15.0 17.9 <1 72.6 67.3 100 120 124 125 126 1.67 0.42 1.0 2.05 0.62 1.2 880 2.0% 0.8% 1080 2.5% 1.0% 16.7 12.0 13.0 20.5 14.9 15.9 12.3 15.6 13.9 15.1 19 • 5 d l 7 .O <1 <1 —— 69.3 62.3 70.0 62.5 71.3 64.9 110 110 190 140 140 230 corrected for condensation of water prior to analysis. corrected also for injection of oxygen above bed OR addition of ------- BIBLIOGRAPHIC DATA 1. Report No. 2. SHEET E PA-650/2-74-001 3.’. ecipient’s Accession No. 4. Title and Subtitle 5. Report Date A Regenerative Limestone Process for Fluidized-Bed Combustion and Desulfurizat ion Coal January 1974 6. 7. Author(s) R. C. Hoke, M. S. Nutkis, L. ARuth, and H. Shaw 8. Performing Organization Rept. No. GRUS. 1 4GFGS. 74 9. Performing Organization Name and Address 10. Prpjcct/Task/Work Unit No. Esso Research and Engineering Co. P.O. Box 8, Linden, NJ 07036 ROAP 2IADB-13 U. Contract/Grant No. CPA 70-19 11 Sponsoring Organization Name and Address EPA, Office of Research and Development NERC-RTP, Control Syst ems Laboratory Research Triangle Park, NC 27711 13. Type of Report & Period Cot ered Final 14. 15. Supplementary Notes 16. Abstracts The report gives results of an experimental study of the pressurized comb- ustion of coal in a I luidized bed of limestone and regeneration of sulfated lime- stone. The study is part of a program to develop fluidized—bed coal combustion as a means of desuif urizirl.g flue gas in-situ and generating clean power at low cost. The process, including regeneration of spent limestone by reduction to lime, produces a gas stream containing a sufficient concentration of S02 to be fed to a by—product sulfur recovery unit. The combustion runs were limited by operating problems, es ecia1ly plugging in the coal injection line. Initial S02 removal rates were about 85%; however, attrition rates were high with one S02 sorbent, Tymochtee dolomite. The regeneration step was studied at pressures up to 10 atm and temperatures up to 2100°F. S02 concentrations measured in the product gas averaged about 2% at 10 atm and 2100°F, or about 40% of the concentrations calculated by assuming equilibrium be een the solids and regenerating gas. High conversion of sulfated material to lime was achieved by injecting air into the bed, by forming adjacent reducing and oxidizing zones, and by minimizing formation of undesired CaS. 17. Key words and Document Analysis. 170. Descriptors Air Pollution Des ul.furization Flue Gases Re generation (Engineering) Limestone Calcium Oxides Fluidized-Bed Processors Combustion 17b. identifiers Open-Ended Terms Air Pollution Control Stationary Sources Regenerative Limestone Process 17c. COSATI Field/Group 13B, 2lB Report 1 Unlimited UNCLASSIFIED j 101 18. Availability Statement r 9 - Security Class (This I2LNo. of Pages 20. Security Class (This 22. Price Page UNCLASSIFIED USCOMMDC r49 52- 0 72 FORM W V 5-35 REV. 3-72) THIS FORM MAY BE REPRODUCED 101 ------- |