Fuel 86 (2007) 244255 www.fuelrst.

com

Some process fundamentals of biomass gasication in dual uidized bed

Takahiro Murakami, Guangwen Xu *, Toshiyuki Suda, Yoshiaki Matsuzawa, Hidehisa Tani, Toshiro Fujimori
Ishikawajima-Harima Heavy Industries Co., Ltd., Sin-nakahara-cho 1, Isogo-ku, Yokohama 235-8501, Japan Received 18 December 2005; received in revised form 28 April 2006; accepted 24 May 2006 Available online 23 June 2006

Abstract The dual uidised bed gasication technology is prospective because it produces high caloric product gas free of N2 dilution even when air is used to generate the gasication-required endothermic heat via in situ combustion. This study is devoted to providing the necessary process fundamentals for development of a bubbling uidized bed (BFB) biomass gasier coupled to a pneumatic transported riser (PTR) char combustor. In a steam-blown uidized bed of silica sand, gasication of 1.0 g biomass, a kind of dried coee grounds containing about 10 wt.% water, in batch format claried rst the characteristics of fuel pyrolysis (at 1073 K) under the conditions simulating that prevailing in the gasier intended to develop. The result shown that via pyrolysis more than 60% of fuel carbon and up to 75% of fuel mass could be converted into product gas, while the simultaneously formed char was about 22% of fuel mass. With all of these data as the known input, a process simulation using the software package ASPEN then revealed that the considered dual bed gasication plant, i.e. a BFB gasier + a PTR combustor, is able to sustain its independent heat and mass balances to allow cold gas eciencies higher than 75%, given that the fuel has suitable water contents and the heat carried with the product gas from the gasier and with the ue gas from the char combustor is eciently recovered inside the plant. In a dual uidized bed pilot gasication facility simulating the gasication plant for development, the article nally demonstrated experimentally that the necessary reaction time for fuel, i.e. the explicit residence time of fuel particles inside the BFB gasier computed according to a plug granular ow assumption, can be lower than 160 s. The results shown that varying the residence time from 160 to 1200 s only slightly increased the gasication eciency, but the reaction time available in the PTR, say, about 3 s in our case, was too short to assure the nish even of fuel pyrolysis. 2006 Elsevier Ltd. All rights reserved.
Keywords: Biomass; Coee grounds; Dual uidized bed gasication

1. Introduction In face of the global warming and climate change problems, the utilization of biomass as an alternative of fossil fuels has been on its boom for many years. To Japan, biomass is an indispensable supplementary guard of the nations energy safe supply because the nation nearly completely relies on imported oils, coals and gases. Statistics anticipated that the usable biomass per year in Japan might amount to 26 million kiloliter crude oils of heat equivalence [1]. The ecient use of this energy resource is surely impor*

Corresponding author. Tel.: +81 45 759 2867; fax: +81 45 759 2210. E-mail address: [email protected] (G. Xu).

tant not only to regulation of the nations CO2 emission but also to her energy diversication. Since 2003, The New Energy and Industrial Technology Development Organization (NEDO), Japan, had sponsored a three-year technical program to develop an advanced upgrading and pyrolytic gasication system for high-moisture biomass fuels, such as various wet wastes from beverage and food industries. The technical system consists of two parts, a reforming drier and a pyrolytic gasication plant. The drier works to reduce the water content of the wet biomass fuel to some reasonable values and to upgrade the fuel through reforming some of its fats to improve the fuels thermal properties, such as O/C ratio, heating value and so on [2]. For the research program

0016-2361/$ - see front matter 2006 Elsevier Ltd. All rights reserved. doi:10.1016/j.fuel.2006.05.025

T. Murakami et al. / Fuel 86 (2007) 244255

245

the considered model biomass is coee grounds containing water higher than 65 wt.%. This model fuel not only typies the high-moisture bio-wastes like tea grounds, soy sauce lees, vinegar lees and bagasse which are concentrated already, but has also a total production of up to 30 thousand tons per year in Japan. Instead of disposal or combustion for heat, many beverage works are thus interested with conversion of such fuels into middle-caloric gases to lower the works natural gas consumptions. The devised pyrolytic gasication plant is schematically illustrated in Fig. 1. It is a bubbling uidized bed (BFB) gasier coupled to a pneumatic transported riser (PTR) char combustor. As shown in the plot, the lower end of the PTR combustor is immersed into the particle bed (or particle bulk) of the BFB gasier, and the riser itself is seated around the central vertical line of the BFB gasier. Air nozzles extended to the downside entrance of the PTR supply air to the combustor. This airow not only provides the necessary oxidant for char combustion but catches also the particles from the BFB to let them move into the PTR. Steam, fed to the bottom of the BFB, is the usually employed gasication reagent. The gasier needs thus an external heat resource, which is the hightemperature heat carrier particles, commonly sand at about 1173 K, circulated from the PTR. The working procedure of the gasication plant is as follows. Biomass fuel is fed to the BFB gasier, wherein it is pyrolyzed and gasied through interactions with steam and the high-temperature heat carrier particles (i.e. sand) coming from the PTR combustor. The gaseous product of fuel pyrolysis/gasication (including tars) independently exhausts from the gasier, whereas the unreacted char remains in the gasier to move with the heat carrier particles. The amount of unreacted char is a function of gasication conditions, such as temperature, fuel particle residence time in the BFB, etc. With progress of fuel pyrolysis and gasication the temperature of the heat carrier particles has to decrease, but the particles, with char blended, could quickly reach the jetted air to allow them to be conveyed into the PTR. There, the expose of the

Combustion exhaust gas

Pneumatic transported bed char combustor

Fine powder (ash) Gasification gas

Biomass Bubbling fluidized bed gasifier Air

Fine powder (ash)

Steam

Fig. 1. Conception of the devised pyrolytic gasication plant.

unreacted char to air leads to char combustion and thereby to heat up the heat carrier particles again. The heated particles are in turn separated from the ue gas in the PTR cyclone and recirculated into the BFB gasier to serve continuously the fuel pyrolysis and gasication reactions by providing them the needed endothermic heat. Therefore, the distinctive advantage of the gasication plant is its separation of combustion exhaust from gasication gas product. This, similar to all the other dual bed gasication processes (such as FICFB [3]), allows the product gas to have high caloric values and low intake of inert gases (usually N2 and CO2). Furthermore, the quoted gasication plant has a more compact design than the other commonly encountered dual bed gasication assembles. In the other cases the char combustor and fuel gasier generally stand separately [3,4], which incurs surely more heat loss and requires a bigger space to construct the plant. Controlling both particle circulation and fuel mass and heat partitions between the BFB gasier and PTR char combustor is critical to the devised gasication plant and its implicated technology. While the circulation rate of heat carrier particles determines if it is possible to carry the required endothermic heat from the PTR combustor to BFB gasier, the partitions of fuels mass and heat species if suitable exothermic heat can be generated in the combustor and how high the available gasication eciency can be. Too less char from the BFB to PTR means insucient heat supply to the fuel pyrolysis/gasication reactions so that the temperatures of the whole system will become gradually lower and unstable with operation. On the contrary, if the char moved to the PTR combustor is too many, there must be an excessive heat production inside the combustor to cause its temperature out of control. Thus, the available gas production (or gasication) eciency via the plant is subject to the quoted char partition between the gasier and combustor, although the eciency varies as well with fuels water content. The present article is about the proposed gasication plant and intends to provide a few relative process fundamentals. In a batch uidized bed gasier it claried rst the available C and H conversions, product partition characteristics and gas composition at the end of fuel pyrolysis in a steam atmosphere. Using these data as the known input, a process simulation with ASPEN was in turn conducted to demonstrate the possibly available eciencies in the devised gasication plant for fuels, here coee grounds, containing water of up to 45 wt.%. This resulted in clarication of the suitable fuel water contents that assure the desired gasication eciency, such as over 75%. Finally, a few gasication tests in a 5 kg/h pilot dual uidised bed gasication facility simulating the gasication plant illustrated in Fig. 1 were carried out to identify the dependence of fuel conversion on reaction time of fuel, i.e. on the fuel particle residence time inside the gasier, so as to provide the rst basis for designs of the gasication reactor.

246

T. Murakami et al. / Fuel 86 (2007) 244255 Table 1 Properties of coee grounds No. 1 Proximate (wet-wt.%) Moisture VM FC Ash Ultimate (db-wt.%) C H N S O HHV (kcal/kg-db) Bulk density (kg/m3) Particle size 10.5 71.8 16.7 1.0 52.97 6.51 2.80 0.05 36.62 5260 No. 2 9.3 69.4 19.3 2.0 54.9 6.12 3.07 0.01 33.62 5682 Around 350 <2.0 mm

2. Methodology 2.1. Batch uidized bed gasier Fig. 2 shows a schematic diagram of the used lab-scale bubbling uidized bed gasier. As detailed in Xu et al. [5], the bed was made of quartz glass, and was 80 mm in i.d. and 1200 mm high. A golden furnace heated the bed, with temperatures inside the bed measured at 250, 400, 700 and 1100 mm above the bed distributor (a sintered porous plate). Via batch gasication test in this gasier we measured the available C and H conversions and the corresponding product distribution at the end of fuel pyrolysis in a steam atmosphere. As for the biomass gasication plant conceptualised in Fig. 1, the fuel pyrolysis may dominate the available gasication eciency. The facts that biomass fuel is abundant in volatile matters and the kinetic rate of fuel pyrolysis is much quicker than that of char steam gasication should be the causes. Meanwhile, the product distribution of biomass pyrolysis is closely dependent on measurement methods, such as heating device, because dierent methods may cause dierent heating rates to fuel particles. Choosing the bubbling uidized bed gasier illustrated in Fig. 2 expected to make the measured fuel pyrolysis characteristics representative of what actually occurs in the gasier conceptualised in Fig. 1. The test procedure was as follows. After the gasier was set to the desired conditions, one gram of fuel (fuel no. 2 in Table 2) was dropped into the uidized bed of silica sand particles to start the gasication reactions. The sand had a Sauter mean diameter of 190 lm, and the loaded amount in the bed was 3.2 kg. This led to a uidized particle bed of 500550 mm high under a steam feed (473 K) of 5.6 g/min and a N2 tracer gas ux of 4.0 LN/min. The gas from the reactor, which was rst dedusted, ltered and dried at the reactor exit, was sampled with gas bags at a time interval of 15 s. The sampled gas was then analysed in a micro gas chromatograph for its molar composition. Referring further to the N2 tracer ux, the measurement determined also the moles of various gaseous species (H2, CO, CO2 and

hydrocarbons of up to C3H8) and the total moles of released C and H. The tested biomass fuel was dry coee grounds obtained from drying/upgrading high-moisture coee grounds with water content over 65 wt.% according to the technology of slurry dewatering in kerosene [2]. This article was involved with two types of dry coee grounds (received from another company at dierent time). Table 1 summarizes their properties determined with proximate and ultimate analyses. We can see that both the fuels were similarly rich in oxygen and contained water of about 10 wt.% and volatiles of about 70 wt.%. Nonetheless, slight dierences existed in compositions, which might be due to some dierent drying/upgrading conditions applied to them. The fuel no. 2 was used for the batch gasication test specied above. 2.2. Process simulation The software package used was ASPEN plus. In order to simulate the chemical steps occurring in the gasication plant sketched in Fig. 1, a process model highlighted in Fig. 3 was constructed. The model consists of two basic modules, the fuel pyrolytic gasication module and char combustion module. The gasication module is further made of a fuel pyrolyzer, a char gasier and a gas/tar reformer. In actual gasication plant all these reactors should be integrated together, which have thus the same output temperature in Fig. 3. Unreacted char from the char gasier is completely sent to the char combustor of the combustion module, whereinto air preheated with hot product gas is supplied to provide oxygen for combustion. Between the two modules heat carrier particles (silica sand) are circulated. Independent exhausts are available to both the reformer and char combustor. Thus, the model simulated exactly the process features implicated in Fig. 1. This article considered only the temperatures of 1173 K and 1073 K for the combustion and gasication modules, respectively. On this basis Fig. 3 shows also how the heat

coffee grounds
T P T

filter cyclone

T : thermocouple P : pressure sensor

steam generator bag filter

P T

IDF

Gas analyzer

pump N 2

Fig. 2. A schematic diagram of the lab-scale bubbling uidized bed gasier.

T. Murakami et al. / Fuel 86 (2007) 244255

air

247

Gasification
Product gas

water

Gasification
Pyrolyzer H 0 Fuel Char + H2 + CO + + Tars, 298 > 0, ( H 0 Estimated from reactions enthalpy balance) : 298 Water Steam H 0 > 0 298 Gasifier
Char + H2O CO + H2 Char + CO2 2CO

Tm
573K Ta

1073 K

373 K

Reformer

water+steam
Ts1

1073 K

cyclone 1

heat
773K
fuel

Gasifier
1073 K
circulated sand

H 0 298 = 131 kJ/mol

H 0 298 = 172 kJ/mol H 0 298 = 41 kJ/mol H 0 298 = 206 kJ/mol H 0 298 > 0, Estimated.

1073 K

Reformer
CO + H2O CO2 + H2 CH4 + H2 O CO + 3H2

Pyrolyzer
air, Ta
Ts1

Combustion
flue gas 423K

Tar + H2O aCO + bH2

auxil. fuel
1173 K

Combustor
1173 K

Combustion
Combustor
Char + 1/2O2 CO CO + 1/2O2 CO2 Fuel + aO2 bH2O + cCO2

373K water+steam

cyclone 2

Fig. 3. A process model simulating the gasication plant sketched in Fig. 1.

H 0 298 = 111 kJ/mol H 0 298 = 284 kJ/mol H 0 298 , Based on HHV

Fig. 4. Considered reactions in process simulation.

carried with hot ue gas (1173 K) and hot product gas (1073 K) is recovered in the modelling. It was supposed that the virgin product gas from the reformer undergoes a heat exchange rst with water supplied for the gasier. This reduces the gas temperature to an intermediate value of Tm ($773 K) and also converts the water into a watersteam mixture of 373 K (under 1.0 atm). The remaining usable heat with the product gas, i.e. the enthalpy between Tm and its exhaust temperature of 573 K, is then consumed in heating the air fed to the char combustor by raising the airows temperature to Ta. In another heat exchanger, the heat from the ue gas converts further the aforementioned water-steam mixture into a steam stream of Ts1. The ue gas itself approaches in turn its nal exhaust temperature of 423 K. In case Ts1 is lower than 773 K, further heating the steam stream to 773 K, the desired temperature of gasication reagent, was conducted. In the depicted heat network the temperatures Ta and Tm are two dependent variables. Another dependent variable is the amount of circulated silica sand. Because the temperature dierence between the combustion and gasication modules is xed at 100 K, the circulated particle amount has to vary with the amount and composition of fuel feed. Herein, however, only the fuels water content is varied (based on coee grounds no. 2 in Table 1), while the fuel amount (wet-base) remains in a constant. Fig. 4 summarizes the considered chemical reactions, with as well their accompanying enthalpy variations, for each processor of the process model. In the char gasier, both steam and CO2 gasications were considered. The reformation of tars and CH4 (representative of CmHn) is suggested to occur only in the reformer. In the reformer water gas shift (WGS) reaction was supposed to be in its equilibrium state, noting that the gasication module has temperatures over 973 K. The char combustion was thought to

occur in two successive steps, rst into CO and then into CO2 via oxidizing the formed CO. At this stage, we calculated merely the heat and mass balance of the whole system so as to demonstrate the possibly available cold gas eciency and C and H conversions for fuels with various dierent water contents. The cold gas eciency refers to a ratio of HHVs between the produced gas and treated fuel. For the calculation it was assumed that the fuel fed into the pyrolyzer is completely pyrolyzed to form gas, tars and char according to the product percentage identied in the preceding batch uidised bed gasication test (the Section 2.1). Thus, increasing the fuel water content has to decrease the available amounts of dry fuel and the formed char. This would lead the char combustion to be unable to provide enough exothermal heat to maintain the systems temperature. Once this occurs, an auxiliary fuel feed (with the same water content as that of the fuel fed to the gasier) is directed to the char combustor so that the burning of this fuel can compensate for the insucient heat. Consequently, the computation was iteratively performed to determine the percents of gasied char (the left being combusted) or the necessarily required auxiliary fuel feed for char combustor. This in turn determines the circulated heat carrier particle amount and the temperatures Tm and Ta. It was suggested that 99% of the tars from pyrolysis is reformed, whereas reformation of CmHn was not considered. In addition, no kinetic limitation was applied to all reactions (such as char combustion and gasication). The needed enthalpy changes for reactions of fuel pyrolysis and tar reformation were estimated as the enthalpy dierence between their reaction products and reactants. In these estimations the standard enthalpy of tars was roughly determined according to the tars element composition

248

T. Murakami et al. / Fuel 86 (2007) 244255

obtained from element analysis (will be shown in the Section 3). 2.3. Pilot dual uidised bed gasication facility Fig. 5 shows a sketch of the pilot dual uidized bed gasication facility adopted to clarify the necessarily entailed reaction time for the fuel and to demonstrate the chemical possibility of the devised gasication process illustrated in Fig. 1. In the experimental facility, the pyrolytic gasier and char combustor were separately installed. This makes the whole facility be a modied circulating uidized bed (CFB) so that its operation can take advantage of the experiences from CFB combustion. On the other hand, the acquired result in this facility relative to chemical reactions is surely applicable to the compact conguration shown in Fig. 1 because for both the fuel pyrolysis/gasication proceeds in a BFB and char combustion in a PTR. However, the particle circulation control should be dierent from each other, raising thus another independent research of ours on it in a cold model uidised bed congured exactly according to Fig. 1 [6]. As indicated in Fig. 5, the gasier was a rectangular bed, with a cross section of 80 370 mm2 and a height of 1800 mm. The riser was a circular column of 52.7 mm in i.d. and 6400 mm in height. Both of these reactors were electrically heated (at least to $973 K), while additional feed of propane into them to facilitate temperature rise was also possible. The biomass fuel, as specied in Table 1, was supplied via a table feeder and the fed fuel was in turn carried into the gasier with a well-metered argon stream in a mass

ow controller from a port located slightly above the uidized-particle bed surface. With respective cyclones and gas exhaustion lines, the pressures inside the fuel gasier and char combustor could be independently controlled through adjusting the performance of their IDFs (induction draft fans). In order to prevent intermixing of gases between the riser and gasier, the gasier had a specially designed bed structure. This structure, called reactor siphon [7], allowed the reactions of fuel pyrolysis and char gasication to proceed inside the BFB, while it enabled as well the control of particle ow rate through the bed. The ue gas from the riser combustor was directly cooled down and vented into atmosphere after it passed through a baglter. Compared to this, the product gas from the BFB gasier was rst burned out in a combustion tube packed with ceramic balls and heated electrically to about 823 K. Then, the gas was subject to the same ventilation system as adopted for the ue gas from the PTR combustor. Herein, the presented test results were under given temperatures of about 1073 K for gasier and about 1103 K for char combustor. The tested fuel was fuel no. 1 specied in Table 1. The fuel feed rate and steam-to-fuel mass ratio were both xed at about 3.6 kg/h and 1.0 kg/kg (S/C = 1.22 mol/mol), respectively. Thus, the only varied parameter was the residence time of fuel particles inside the gasier, which was calculated from Particle amount held in the gasifier Particles circulated in unit time 1

under an assumption that the particles pass through the gasier in a ow of plug type. For this calculation, the

Exhaust gas

Gas cooler

Bag filter IDF

Gas analyzer

Gs measure

Combustor
50A 6400mmH Gasification gas filter Tar trap Biomass Gas analyzer IDF Combustion tube Gas cooler Bag filter

Gasifier
370 80 1800mmH 37 FDF Steam generator
Fig. 5. A schematic diagram of the employed pilot dual uidized bed gasication facility.

T. Murakami et al. / Fuel 86 (2007) 244255 Table 2 C and H conversions, product distribution and composition at the end of fuel pyrolysis in a steam-blown uidized bed of 1073 K Conversion to gas (%) Product distribution (wt.%)a Molar composition (vol.%) (free of tracer gas) Char composition (wt.%)b Tar composition (wt.%)c
a b c

249

15 Concentration [vol.%] 12 9 6 3 a 0 100 Element conversion [%] b 80 60 40 20 0 0 30 Fuel : 1g Temp. : 1073K C H 60 Time [sec] 90 120 Fuel : 1g Temp. : 1073K CO H2

C: 63.0, H: 88.7 Gas: 75, Char: 22.1, Tars: 2.9 19.3H2 + 38.2CO + 9.5CO2 + 17.5CH4 + 9.9C2H4 + 5.2C2H6 + 0.4C3H6 63.82C + 0.0H + 11.57N + 14.33O + 0.003S + 0.36Cl + 9.90 ash 64.40C + 6.23H + 11.10N + 15.10O + 3.01S + 0.11Cl

The fraction of tars was from pilot gasication test. Derived from element mass balance. From measuring the tar sample taken in pilot gasication test.

required particle circulation rate at the tested temperature of 1103 K was experimentally measured by using a particularly designed heat-resistant valve [8]. The product gas was sampled at the gas exit of the gasier cyclone to trap the tars with the gas (for tar element analysis) and to measure the gas molar composition. The sample gas ow, induced with a suction pump, was rst condensed in a water condenser to trap most of the steam present in the gas. Then, the gas passed through three water bubblers kept in an ice-water bath to further trap the tars with the gas. A wet gas meter was set behind the last water bubbler to measure the sample gas ow rate that was controlled at about 2.0 nL/min. By using a fabric lter to capture further the tars escaping from the water bubblers (BTX, phenols and poly-aromatics exclusively) it was demonstrated that the tar trapping eciency in our tar collection system was about 90%. The molar composition of product gas was measured in the same micro GC adopted for the batch gasication test, while the product gas volume was determined according to the rate of argon stream used to carry fuel. Extraction of the tars trapped in the condenser and water bubblers was conducted by following successively the steps of collecting the tarry water, washing the vessels (using acetone), ltrating the collected wateracetone liquid, vacuum vaporization of water and acetone (<333 K) and drying tars (the vaporization residue) in warm airow (<323 K). The resulting dry tars were then weighed to estimate the tar content in the product gas and analysed for its element composition (see Table 2). The tar content, however, will not be reported in this article because of its little relevance to the articles purpose. 3. Results and discussion 3.1. Pyrolytic characteristics in steam-blown uidized bed As indicated in Section 2.1, gasication tests in the batch uidized bed gasier (Fig. 2) were conducted to demonstrate the pyrolytic characteristics of the fuel coee grounds. Fig. 6 exemplies the time series of molar gas composition (Fig. 6a) measured at the gasier exit and

Fig. 6. Time series of (a) molar gas composition from a batch gasication test in the steam-blown bubbling uidised bed gasier and (b) the corresponding integrated C and H conversions estimated from the molar composition.

the corresponding accumulative C and H conversions (Fig. 6b) calculated according to the gas composition. Surely, in this batch test the produced gas, a mixture of H2, CO, CO2, CH4, C2H4, C2H6 and C3H6, should be rst gradually more and then gradually less with the progress of reactions. The concentrations of CO and H2 in the outlet gas (Fig. 6a) thus both exhibited a peak and nally approached values nearby zero. On account of the much slower char gasication in steam than fuel pyrolysis, we took the end of quick product gas release marked with the vertical broken line in Fig. 6 to represent the nish of fuel pyrolysis. According to this gure, the pyrolysis ended at about 43 s after dropping 1.0 g fuel into the reactor. This time is longer than our anticipated values necessary to fuel pyrolysis at the tested temperature of 1073 K. In a tubular reactor and N2 atmosphere, Bingyan et al. [9] reported a time of about 15 s for nishing the biomass pyrolysis at 1073 K. Here the longer time was considered to be due to the too big reactor used. Nonetheless, our purpose was other than measuring the time required for nishing fuel pyrolysis. It was for determination of the conversions of C and H into gas and the corresponding distribution of pyrolytic products, i.e. gas, char and tars, in steam-blown uidised bed. Thus, the reactor size would not much aect these intended data. Fig. 6b shows the integrative C and H conversions (into gas) corresponding to the gas molar composition exemplied in Fig. 6a. The integrative conversion for time t was calculated with (taking C as the example)

250

T. Murakami et al. / Fuel 86 (2007) 244255

Moles of released C until time t ; Moles of C in the 1:0 g fuel

the fuel. Table 2 veries this by showing that the C/H ratio is 64.4/6.23 for tars but 54.9/6.12 for the tested coee grounds (no. 2 in Table 1). 3.2. Theoretically possible gasication eciency The preceding batch test was for a dried fuel and also did not consider the heat balance between the fuel gasier and char combustor. Using the result of such a batch test (in Table 2) as the required known input, a process simulation with ASPEN according to the process and chemical models outlined in Figs. 3 and 4 was conducted to determine the possibly available gasication eciencies for the whole plant of Fig. 1. Table 3 summarizes the simulated conditions and the required known input. The fuel treated was the same coee grounds no. 2 specied in Table 1 but with water contents varying in 3.045.0 wt.%. The wet-base fuel feed rate was 250 kg/h. Steam at 773 K was supplied into the gasier according to a mass ratio of 1.0 kg/kg of steam to dry fuel. Water with the fuel was supposed to be completely vaporized inside the pyrolyzer (see chemical models in Fig. 4), causing it to mix with the supplied steam and in turn to raise its temperature to the gasication temperature of 1073 K. The employed heat carrier particles were silica sand of 190 lm in Sauter mean diameter. An air ratio of 1.5 against the char amount from the gasier was taken to determine the airow rate into the char combustor. Table 3 mentions also several other characteristic temperatures, which were all from the process model outlined in Fig. 3. The data about fuel pyrolysis were based on Table 2. Fig. 7 shows the simulation results acquired, with Fig. 7a for the available cold gas eciency (left Y, based on HHV) and C and H conversions (right Y) and Fig. 7b for the necessarily required conditions that assure the eciency and conversions. The conditions were presented with circulation rate of heat carrier particles (left Y) and auxiliary fuel amount fed to the combustor (right Y). All the parameters were expressed as a function of fuel water content. While the displayed circulation rate refers to a specic value against the treated dry fuel amount, the amount of auxiliary fuel (with the same water content as that fed to
Table 3 Conditions and known input for process simulation Fuel: coee grounds no. 2, 250 kg/h, with water content varied in 3.045.0 wt.% Gasier: temperature = 1073 k; Steam/dry-fuel = 1.0 (mass ratio); Steam temperature = 773 K Combustor: temperature = 1173 k; Air ratio = 1.5; Air temperature = Heated via hot product gas Heat carrier particles (sand): Sauter mean dp = 190 lm; Gs: by heat balance Exhaust temperatures: ue gas temperature = 423 K; Product gas temperature = 573 K Pyrolysis characteristics (product allocation, gas/tar composition): in Table 2

where, the moles of released C as well as H until time t were estimated from the time-series gas composition and dry gas ow rate determined according to the ow rate of N2 tracer. Nonetheless, the H conversion dened via Eq. (2) refers to an explicit value because, on the one hand, part of the fuels H would be converted into water in fuel pyrolysis and, on the other hand, the product gas contains also the H taken from steam via, for example, char steam gasication, tar/hydrocarbon reforming and water gas shift. After fuel feed, the integrative conversions of C and H increased with reaction time, in response to the gradually more product gas released from the reactor. At the end of fuel pyrolysis (broken-line annotated) they reached about 63% and 89%, respectively (see Table 2). These corresponded to a fuel mass conversion of about 75 wt.%. The remaining 25 wt.% should be char and tars. In order to determine their partition fractions we supposed that the tar yield is the same as that measured hereafter via pilot gasication test under the same 1073 K. There, it was found that the evolved tars amounted to 2.9 wt.% of dry fuel so that the char production became 22.1 wt.%. In this mass balance the produced H2O in fuel pyrolysis was suggested negligible and fuel ash was thought to be fully with char. The molar composition of product gas listed in Table 2 refers to the measured values but excluding N2 tracer, whereas the element composition of char was determined according to element mass balances. For this, the element content of tars, shown as well in Table 2, was treated as known data, which was obtained by measuring the tar sample taken in the pilot gasication test at 1073 K shown lately in Section 3.3. The similar batch gasication tests were conducted also at 973 and 1023 K (the other conditional parameters being the same as at 1073 K), revealing obvious decreases in the C and H conversions (into gas) and in the gas production percentage with reducing the reaction temperature. The acquired data claried also that for realizing C conversions higher than 60% the temperature has to be over 1073 K. Besides, the H conversion was found to be always higher than C conversion. While this reects a general phenomenon of biomass pyrolysis, in our case the presence of steam may make the H conversion (against fuel H only) rather large because steam reforming of tars and hydrocarbons would occur to convert some H of steam into product gas. The H conversion was lower than 100%. This is due to the fact that part of the fuel H had to become H2O during fuel pyrolysis. On the other hand, it shows also that the accompanying steam reforming reactions for the tested pyrolysis in steam did not occur to an extensive degree to make the H conversion over 100%. In the test we also observed serious tar deposition onto pipes behind the gasier, which made the pipes brown, even black. Because of the higher conversion of H than of C into the product gas, the tars must have a C-to-H ratio higher than that in

T. Murakami et al. / Fuel 86 (2007) 244255

251

100
Cold gas efficiency [%]

150
a
Element conversion [%]

90 80 70 60 50 60
Circulation rate / Dry fuel feed rate Auxiliary fuel
Cold gas efficiency C conversion H conversion

120

90

50
[-]

20
b

40 30

15 10

20 10 0 0 10 20 30 40 Fuel's water content [wt%] 5 0 50

Fig. 7. Simulation results of (a) gasication eciency shown with cold gas eciency and C and H conversions and (b) the correspondingly required particle circulation rate and auxiliary fuel feed to the combustor.

the gasier) indicates the value without heat loss from the system. Hence, the unavoidable heat loss in practical plants would make both the parameters surely higher, as will be claried in Fig. 8. As anticipated, the available cold gas eciency and C and H conversions (Fig. 7a) all decrease with raising the water content in the fuel. The higher the fuel water content, the more the heat required to vaporize the water and to heat the resulting steam to the gasier temperature (here 1073 K). On the other hand, this part of heat cannot be completely recovered from the product gas because its outlet temperature is 573 K. It was considered that until downstream scrubber the product gas is better to be over such a temperature in order to avoid substantial deposition of tars on pipeline. Hence, the higher water content means the more C needed to be combusted to maintain the temperatures of the combustor and fuel gasier, lowering conse100
Cold gas efficiency [%]

quently the C amount for gas production. The converted H also decreases because the decreased C amount for steam gasication reduces the converted H from H2O (i.e. steam). Corresponding to the simulation conditions that 99% tars are reformed and the combusted char is almost free of H, the resulting H conversion is always higher than 100%. When no heat loss is considered, the available cold gas eciency can be over 80% for fuels with water content lower than 40 wt.%. In order to reach a cold gas eciency of 85%, the fuels water content has to be lower than 10 wt.%. As the necessary conditions assuring the gasication eciencies mentioned above, Fig. 7b demonstrates that with raising the fuels water content the entailed relative circulation rate of heat carrier particles against the treated dry fuel amount becomes higher (left Y). Meanwhile, until a water content of 25 wt.% no auxiliary fuel is required (right Y), whereas further increase of the fuel water content has to lead to gradually higher consumption of auxiliary fuel. These clarications are obviously reasonable because they responded just to the change of fuel enthalpy with fuel water content. Nonetheless, the C and H conversions show in Fig. 7a are higher than the experimental data indicated in Table 2 (from batch gasication test). The steam gasication of char without kinetic limit, complete tar reforming (99% reformed) and zero heat loss considered in the simulation should be the causes. Although these assumptions are too ideal, the simulation itself shows its signicance by clarifying theoretically the available highest gasication eciencies for the examined fuels and technology. On the other hand, based on Fig. 7 we can simply investigate how heat loss aects the gasication eciency and its required conditions. Fig. 8 presents the variations of the available cold gas eciency (left Y) and entailed auxiliary fuel feed (right Y) with heat loss of up to 30% of the fuel energy inputted into the gasier. The considered case was for a water content of 10 wt.%. The revised cold gas eciency was computed from Efficiency wihout heat loss=1 heat loss percentage; 3 while the modied necessary auxiliary fuel feed was determined as 8 0 if equivalent energy of gasified char > > > > < at zero heat loss > Actual heat loss > Actual heat loss Equivalent energy of gasified char > > > : at zero heat loss; Otherwise: 4 As expected, raising the heat loss surely decreases the cold gas eciency and increases the required auxiliary fuel feed. If supposing that the heat loss varies generally from 5% to 10% of the treated fuel energy, we can see that the cold gas eciency without kinetic constrainments can be still 77% 81%. On this basis, we suggest that the water content of the

Auxiliary fuel [wet-kg/h]

Circulation rate Dry fuel feed rate

90 80 70 60 50 0

Fuel's water content : 10 wt. % Cold gas efficiency Auxiliary fuel

25
Auxiliary fuel [wet-kg/h]

20 15 10 5

10 15 20 Heat loss [%]

25

0 30

Fig. 8. Theoretically possible cold gas eciency with consideration of heat loss of up to 30% of fuel energy fed into the gasier under a fuel water content of 10 wt.%.

252

T. Murakami et al. / Fuel 86 (2007) 244255

coee grounds fuel for gasication is better to be about 10 wt.% in order to assure the gasication eciencies above 75%. Although no auxiliary fuel feed is needed in Fig. 7 for the fuel containing 10 wt.% water, the presence of heat loss makes it absolutely necessary when the loss is over 6%. The result judges further that the fuels water content should be around 10 wt.% in order to achieve energy conversion eciencies over 75%. This provides actually a design/operation standard for the upstream fuel drier. 3.3. Necessary reaction time How long should the reaction time be for fuel particles inside the gasier in order to nish fuel pyrolysis and to realize the cold gas eciency desired, such as over 75%. This is actually another unavoidable problem that has to be answered for developing the devised gasication plant. For this answer a few tests with the coee grounds no. 1 were conducted in the pilot dual uidised bed gasication facility sketched in Fig. 5 by varying the residence time of fuel particles inside the gasier. According to formulation (1) the dierent residence times were realized by adjusting the gas velocity in the riser combustor to vary the particle circulation rate. From these gasication tests we also gained the tar sample analysed for the tar element composition listed in Table 2. Fig. 9 exemplies the time series of molar composition (including tracer gas) and HHV of the produced gas (Fig. 9a) and their corresponding transient cold gas eciency and C and H conversions (Fig. 9b). The plotted test

1180 1160 Temperature [K]

Gasifier Combustor 0.30m 0.30m 0.65m 1.40m 1.40m 2.50m 1140 4.80m

1120 1100 1080 1060 0 10 20 30 Time [min] 40 50

Fig. 10. Typical time series of the temperatures inside the BFB gasier (d,j, m) and riser combustor (h, s, n, e) of the employed pilot dual bed facility during a test.

60
Gas concentration [%]

50 40 30 20 10 0

Gas heating value H2 CH4 CO C2H4+C2H6 CO2 C3H6

4500
a

4200 3900 3600 3300 3000

100

100
Cold gas efficiency [%]

b
90 80 70 60 50 40 0 10
Cold gas efficiency C conversion H conversion

90 80 70 60 50 40

20 30 40 Time [min]

50

Fig. 9. Time series of molar composition of rude product gas (with tracer) and the corresponding transient HHV, cold gas eciency and C and H conversions measured in the dual bed pilot gasication facility under an explicit fuel particle residence time of 160 s inside the gasier (the other conditional parameters being detailed in Section 2.3).

was for an explicit residence time of 160 s of fuel particles inside the gasier determined according to formulation (1). The correspondingly required airow into the riser was 70 nL/min. Fig. 10 shows the time series of a few typical temperatures measured for this test. The displayed temperatures were for dierent bed heights in the BFB gasier (solid marks) and riser combustor (open marks), and the time zero refers to the onset of fuel feed. As for fuel gasier we can see from Fig. 10 that its freeboard (m) remained in a relatively stable temperature, whereas its particle bed (d j) exhibited an evident local temperature drop with fuel feed. This is indicative of the occurrence of endothermic fuel pyrolysis/gasication reactions inside the particle bed. Being subject to electric heating, the dropped temperatures began to rebound to higher values since 5 min and in turn reached relatively steady values at about 30 min after the fuel feed. At last the upper particle bed (j) and freeboard (m) stagnated at nearly the same temperature, which was the situation expected in setting. The bed bottom (d, 0.3 m above the distributor) had slightly lower temperatures, say, for about 15 K, which shown just an eect of the inlet steam. Inside the riser the temperatures experienced a local drop as well with fuel feed by responding to the lowered particle temperatures inside the gasier. Nonetheless, the temperature drop itself was smaller than in the gasier, while its rebounding was much quicker. This allowed the risers temperatures at dierent elevations to reach their respective steady-states in about 15 min. Meanwhile, the riser bottom (s) possessed the lowest temperature, showing just the process feature that the particles from the gasier have the lowest temperature so that they need to be reheated in traveling through the riser combustor. In regard to the exemplied cases the most intensive char combustion occurred possibly in the middle section of the rise, causing the highest temperature inside the riser to emerge at the height of about 2.5 m (n). The gas compositions in Fig. 9 show evidently that the operation trended to become quasi-steady since 10 min after fuel feed (based on the displayed conversions and

Gas heating value [kcal/mN ]

Element conversion [%]

T. Murakami et al. / Fuel 86 (2007) 244255

253

HHVs). This, compared to the temperature proles in Fig. 10, reveals that the reaction temperature variation in about 10 K after 10 min of fuel feed did not greatly aect the gas generation characteristics. Under the tested conditions (see the Section 2.3) CO took the highest concentration of about 28 vol.% in the product gas (for the tested case the tracer concentration being 24 vol.%). Following this were H2, CH4, CO2, C2 and C3 hydrocarbons in succession (left Y, Fig. 9a). The molar ratio of H2 to CO was about 0.6, with the absolute H2 concentration being about 15 vol.%. Corresponding to the mentioned gas composition the gas HHV (right Y, Fig. 9a) was about 3900 kcal/nm3. Concerning the product gases from biomass gasication this HHV represents a high caloric value, verifying the distinctive merit of the dual uidised gasication technology mentioned in Section 1. The concentration diagrams in Fig. 9a clarify also that the product gas contained hydrocarbons (including CH4) more than 20 vol.%. In the view of H2 production it is important to convert these gas species into CO and H2 via reforming. Fig. 9b demonstrates that the achieved cold gas eciency was 69% or so, corresponding to C and H conversions of about 65% and 88%, respectively. These conversions are somehow very close to the values shown in Table 2 for fuel pyrolysis obtained in the batch gasication tests under a similar temperature. It indicates essentially that the reactions occurring inside the gasier of the pilot gasication facility (Fig. 5) were principally fuel pyrolysis, just aligning with our general anticipation about biomass gasication. That is, its gas production should mainly rely on fuel pyrolysis rather than on char gasication. Nonetheless, the reported test suered probably a fuel loss of about 5% of the original feed because the feed of fuel particles into the gasiers freeboard would lead to an easy elutriation of ne particles. In a strict sense we thus believe that the C and H conversions and cold gas eciency attainable in the pilot gasication test are slightly higher than those from fuel pyrolysis shown in Table 2. This tokens the occurrence of steam reforming of C inside the gasier. The longer reaction time in this case (160 s) than in Fig. 6 (concerned time: 45 s) should be the cause. A test similar to the illustration of Figs. 9 and 10 was conducted also at an explicit residence time of 1200 s of fuel particles inside the gasier. Meanwhile, another recent publication of ours [10] had reported a set of data obtained by gasifying the same fuel in the riser of the same facility. In this case, the riser becomes the gasier so that the fuel particle residence time is basically equal to the gas residence time because particles move with the gas. At the reported supercial gas velocity of 2.82 m/s (at bed temperature) the fuel particle residence time becomes 2.3 s or so (riser height being 6.4 m). By summarizing all of these mentioned data we have then Fig. 11 where the steady-state cold gas eciency and its corresponding C and H conversions are displayed as functions of the explicit fuel particle residence time. In the plot the data indicated by the keys s, h and n are from the gasication test inside the riser [10],

100
Cold gas efficiency [%]

100
Element conversion [%]

90 80 70 60 50 40 0 300 600
Cold gas efficiency C conversion H conversion

90 80 70 60

900

50 1200

Fuel residence time in gasifier [sec]

Fig. 11. Steady-state cold gas eciency and its corresponding C and H conversions under dierent explicit residence times of fuel particles inside the gasier of the pilot dual bed facility.

whose test conditions are similar to those employed here. That is, except for the similar reaction temperature of 1073 K, the fuel feed rate was also about 3.5 kg/h and steam-to-fuel mass ratio was around 1.0. The gure shows that with increasing the fuel particle residence time the conversions of C and H and their corresponding cold gas eciency increased gradually. However, when prolonging the time from 160 to 1200 s the increases in conversion and eciency were conned to a few percents (<5.0%). Corresponding to this, the increase in the residence time of fuel particles from 2.3 s to 160 s induced distinctive increases in both eciency and conversion. While the H conversion (n m) elevated to 88% from 62%, the cold gas eciency (s d) and C conversion (h j) exhibited increases of about 13%. Consequently, for the devised gasication plant (Fig. 1) the required explicit fuel particle residence time inside its gasier can be lower than 160 s. Rather longer residence time is benecial to fuel conversion, but the actually available benet is slight. On the other hand, the gasication inside the riser cannot provide the fuel suciently long time to nish even pyrolysis because inside the riser the available average residence time for the gas and particles is below 10 s [11]. Examining Fig. 11 we can see actually that the C and H conversions indicated with the keys s, h and n are truly much lower than the corresponding values mentioned in Table 2 for fuel pyrolysis. This, in essence, veries that in this case the fuel pyrolysis was not yet completed. This is also why the design of the gasication plant conceptualized in Fig. 1 deploys its fuel gasication into a BFB and its char combustion into a riser coupled to the BFB. Regarding this technical choice we had presented another publication [10] to compare the available eciencies and conversions when gasifying the same fuel inside the riser and BFB of a dual uidized bed system, respectively. Nonetheless, the element (C and H) conversions and cold gas eciency shown here are much lower than those indicated in Figs. 7a and 8 (left Y). For example, at the fuel water content of 10 wt.% the available C and H conversions and cold gas eciency can be, respectively, 75%, 140% and 85% in Fig. 7a but only 68%, 92% and 72% in

254

T. Murakami et al. / Fuel 86 (2007) 244255

Fig. 11 at the residence time of 1200 s. The production of tars and a low degree of C steam gasication involved in the gasication test should be the reasons for the mentioned lower experimental fuel conversion and cold gas eciency on comparison of their theoretical values predicted in the process simulation. As early mentioned in Section 3.2, the preconditions for the simulation results shown in Figs. 7 and 8 are that 99% of the tars generated in fuel pyrolysis is reformed and all unburned char (i.e. C) is gasied through reaction with steam. The fact in the experiment was that up to 3.0 wt.% of the fuel mass was present as tars (at 1073 K), and the slow char gasication reaction might be unable to aord to a desired amount of char to be gasied. Hence, the fuel conversion from experiment must be lower than the predicted value. Comparing Figs. 11 and 7a claries that the C conversion shared dierences only of a few percents (at a water content of 10 wt.%), but the dierences in H conversion are over 40%. This is indicative of the low degree of tar reforming and char steam gasication reactions occurring in the experiments. On the other hand, we should also note that in the presented tests a few percents of supplied fuel (up to 5.0%) were entrained with the gas ow from the gasier because the fuel was fed into the gasiers freeboard. This further lowered the experimental fuel conversion into gas. 4. Conclusions With both experimental tests and process simulation, the present study claried the following engineering fundamentals relative to chemical reactions and heat/mass balances that are necessary to the development of a newly devised dual bed gasication plant for biomass which adopts a bubbling uidized bed (BFB) as its gasier and a pneumatic transported riser (PTR) coupled to the BFB as its char combustor. (1) The gasication of, for example, 1.0 g biomass fuel in a steam-blown uidized bed in batch format appeared to be eective to measure the fuel pyrolysis characteristics in BFB reactors and steam atmosphere. As for our tested biomass fuel, a kind of dried coee grounds (water content being 10 wt.%), it was demonstrated that the fuel pyrolysis under a reaction temperature of 1073 K and a steam atmosphere was able to convert 63% of fuel C into product gas. The corresponding explicit H conversion with respect to fuel H reached about 90%. Both of these led about 75% of the fuel mass to be converted into product gas. (2) With the fuel pyrolysis characteristics claried above as the necessarily required known input, a process simulation using the software package ASPEN demonstrated that the devised dual uidised bed gasication plant can sustain its independent heat and mass loops with a cold gas eciency over 75% at a reaction temperature of 1073 K, provided (a) the heat carried

with the product gas from the BFB gasier and that with the ue gas from the char combustor are eciently recovered to produce steam reagent and to preheat combustion air, and (b) the tars generated in fuel pyrolysis are mostly reformed. The simulation claried also that with increasing the fuels water content the available cold gas eciency and C and H conversions decrease, whereas the required specic circulation rate of heat carrier particles with respect to the treated dry fuel amount increases. When considering the practical applications with more than 3.0% of fuels enthalpy as heat loss, we suggested that the fuel for treatment should contain water not much over 10 wt.% in order to maintain the available cold gas eciency higher than 75%. (3) Gasifying dried coee grounds in a pilot dual uidized bed gasication facility simulating the devised gasication plant claried that the explicit residence time of fuel particles inside the BFB gasier (using steam as the gasication reagent) calculated according to a plug granular ow assumption could be shorter than 160 s. This time can guarantee the nish of fuel pyrolysis so that further prolonging the fuels residence time did not much improve the realized fuel conversion and cold gas eciency. The result indicates essentially that as for biomass gasication the generated product gas comes basically from fuel pyrolysis. Through the tests we demonstrated also that the PTR reactor of the experimental facility was unable to provide the fuel suciently long reaction time to complete even fuel pyrolysis. This demonstration justies the superior technical choice of the devised dual bed gasication plant, which should arrange its BFB as the fuel gasier and its PTR as the char combustor. Acknowledgement The work was conducted during a research program nanced by The New Energy and Industrial Technology Development Organization (NEDO), Japan, on developing an advanced upgrading and pyrolytic gasication system for biomass with high water content. The authors are also grateful to Mr. Minoru Asai and Mr. Shigeru Kitano of the same company for their helps in experiment. References
[1] The Ministry of Agriculture, Forestry, Fisheries of Japan, Overall strategy of Japan on biomass, 2001 [in Japanese]. [2] Mito Y, Komatsu N, Hasegawa I, Mae K. Slurry dewatering process for biomass. In: Proc int conf on coal sci technol (ICCS&T), IEAClean Coal Center, 2005, Paper 2E01. [3] Pfeifer C, Rauch R, Hofbauer H. In-bed catalytic tar reduction in a dual uidized bed biomass steam gasier. Ing Eng Chem Res 2004;43:163440. [4] Paisley MA, Farries MC, Black JW, Irving JM, Overend RP. Preliminary operating results from the Battelle/FERCO gasication

T. Murakami et al. / Fuel 86 (2007) 244255 demonstration plant in Burlington, Vermont, USA. In: First word congress and exhibition on biomass for energy and industry, Sevilla, June, 2000. [5] Xu G, Murakami T, Suda T, Kusama S, Fujimori T. Distinctive of CaO additive on atmospheric gasication of biomass at dierent temperatures. Ind Eng Chem Res 2005;44:58648. [6] Murakami T. Biomass gasication characteristics in a pyrolytic uidised bed gasier, Presented in tutorial meeting of gasication section of the Japan institute of energy, January 2006, Tokyo. [7] Xu G, Murakami T, Suda T, Matsuzawa Y. Reactor siphon and its control of particle ow rate when integrated into a circulating uidized bed. Ind Eng Chem Res 2005;44:934754.

255

[8] Xu G, Murakami T, Suda T. An apparatus for measuring particle ow rate, Japanese patent 2005; Appl. no. 2005-295036 [in Japanese]. [9] Bingyan X, Chuangzhi W, Zhengfen L, Xiguang Z. Kinetic study on biomass gasication. Solar Energy 1992;49:199204. [10] Xu G, Murakami T, Suda T, Matsuzawa Y, Tani H. The superior technical choice for dual uidised bed gasication of biomass. Ind Eng Chem Res 2006;45:22816. [11] Petersen I. Sewage sludge gasication in the circulating uidized bed Experiments and modeling. Doctoral thesis; Hamburg: Technical University Hamburg-Harburg (TUHH); 2004.