Applied Energy. Tar removal from biomass pyrolysis gas in two-step function of decomposition and adsorption

Transcription

1 Applied Energy 87 (2010) Contents lists available at ScienceDirect Applied Energy journal homepage: Tar removal from biomass pyrolysis gas in two-step function of decomposition and adsorption Thana Phuphuakrat a, *, Tomoaki Namioka a, Kunio Yoshikawa b a Department of Environmental Science and Technology, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama , Japan b Frontier Research Center, Tokyo Institute of Technology, 4259 Nagatsuta, Midori-ku, Yokohama , Japan article info abstract Article history: Received 12 August 2009 Received in revised form 29 November 2009 Accepted 1 December 2009 Available online 29 December 2009 Keywords: Biomass pyrolysis Tar removal Tar decomposition Thermal cracking Reforming Tar adsorption Tar content in syngas pyrolysis is a serious problem for fuel gas utilization in downstream applications. This paper investigated tar removal, by the two-step function of decomposition and adsorption, from the pyrolysis gas. The temperature of the tar decomposition process was fixed at 800 C both with and without steam, with air as the reforming agent. Both steam and air had a strong influence on the tar decomposition reaction. The reduction of the gravimetric tar mass was 78% in the case of the thermal cracking, whereas, it was in the range of 77 92% in the case of the steam and air forming. Under conditions of tar decomposition, the gravimetric tar mass reduced, while the yield of the combustible gaseous components in the syngas increased. Synchronously, the amount of light tars increased. This should be eliminated later by fixed-bed adsorption. Three adsorbents (activated carbon, wood chip, and synthetic porous cordierite) were selected to evaluate the adsorption performance of light tars, especially of condensable tar. Activated carbon showed the best adsorption performance among all light tars, in view of the adsorption capacity and breakthrough time. On the other hand, activated carbon decreased the efficiency of the system due to its high adsorption performance with non-condensable tar, which is a combustible substance in syngas. Synthetic porous cordierite showed very low adsorption performance with almost all light tars, whereas, wood chip showed a high adsorption performance with condensable tar and low adsorption performance with non-condensable tar. When compared with other adsorbents, wood chip showed a prominent adsorption selectivity that was suitable for practical use, by minimizing the condensable tar without decreasing the efficiency of the system. Ó 2009 Elsevier Ltd. All rights reserved. 1. Introduction Pyrolysis is one of the thermal conversion technologies for liquid fuel production and gas fuel production. Despite a low efficiency of gas fuel production, pyrolysis has the significant advantage of a higher heating value for the producer gas when compared to the gasification process. The pyrolysis process requires lesser carrier gas quantity, a small capacity of downstream gas cleaning. Moreover, controlling of the pyrolysis process is not complicated. For pyrolysis gas application, the liquid production as well as the tar should be converted to gas, or minimized, in order to prevent the damage of downstream applications and improve the gas production efficiency. Tar is an undesirable material typically described as a complex mixture of condensable hydrocarbons, which include single-ring to multiple-ring aromatic compounds along with other oxygen-containing hydrocarbons that easily condense in ambient conditions [1,2]. The condensable tar deposits on the gas pass-way may cause blockage and corrosion * Corresponding author. Tel.: ; fax: address: phuphuakrat.t.aa@m.titech.ac.jp (T. Phuphuakrat). of the downstream equipment. This problem necessitates frequent maintenance of the downstream equipment, resulting in lower plant reliability. Without the appropriate gas cleaning processes, higher investment is required for the duplicate installation of some downstream equipment, in order to operate the plant continuously [3]. There are five classes of tar: (1) undetectable, (2) heterocyclic, (3) light aromatic hydrocarbons, (4) light polyaromatic hydrocarbons, and (5) heavy polyaromatic hydrocarbons [3,4]. Excluding the light aromatic hydrocarbons, the others are considered as problematic, as they condense at an ambient temperature. Thermal decomposition is one conventional method used to convert tar to gas with high temperature and long residence time. Besides this effect, the tar is also converted into refractory tar [5]. In other words, thermal decomposition is able to minimize tar quantity and increase the yield of producer gas. In order to elucidate tar decomposition during the thermal tar decomposition process, many researchers have utilized one or numerous aromatics as model biomass tar compounds, to investigate the decomposition behavior [1,6 10]. Yet, those studies are based on aromatic combinations modeled as biomass tar compounds, and therefore, cannot be practically applied to the real process. Some papers have /$ - see front matter Ó 2009 Elsevier Ltd. All rights reserved. doi: /j.apenergy

2 2204 T. Phuphuakrat et al. / Applied Energy 87 (2010) reported on the effect of thermal cracking and reforming by using real biomass, but the qualitative and quantitative information of each refractory tar compound is rare. In this paper, we studied the real biomass and investigated the change not only in the gravimetric tar mass (total mass of tar) [11], but also in the concentration of major tar compounds, especially for class 3 and class 4 of tar. Although higher temperature and increasing reforming agents are able to meet the tar concentration requirement, the heating value and the cold gas efficiency will significantly decrease [12]. To optimize the system with maximum cold gas efficiency the remaining tar must be removed with the help of an additional approach. Thus, we have contributed fixed-bed adsorption as an additional equipment, in order to prevent downstream equipment damage from the remaining tar. Fixed-bed adsorption, using three kinds of adsorbents, that is, activated carbon, wood chip, and synthetic porous cordierite, has been conducted to study the adsorption characteristics of real biomass tar. Activated carbon is the most widely utilized adsorbent selected for adsorption of volatile organic compounds (VOCs), which are components of biomass tar, due to its high adsorption capacity. Wood chip is selected because they are low cost adsorbents, and synthetic porous cordierite is selected because it is a re-usable (regenerable) adsorbent. Their adsorption characteristics have been compared with those of activated carbon with real biomass tar. Although many researchers have demonstrated the adsorption performance of many kinds of activated carbon by using tar model compounds [13 16], the information on using real biomass tar is rare. Moreover, the literature carried out on the model tar compound, arranged in class 3 of tar (non-condensable tar) [14], and consequently, it cannot be practically applied to the real process due to a decrease in the efficiency of the system. This paper aims to improve pyrolysis production by converting pyrolysis tar to fuel gases and minimizing condensable tar by using the two-step function of decomposition and adsorption, for the purpose of fuel gas application. Quantitative information on the major tar components on account of the first step of thermal decomposition during the process is acquired. As the second step, fixed-bed adsorption is installed to remove the refractory tar in the system in order to capture the remaining tar. Finally, the performances of the three adsorbents, which are activated carbon, wood chip, and synthetic porous cordierite, are compared by their adsorption in a wide range of tar compounds from the real biomass tar. 2. Material and methods 2.1. Biomass pyrolysis A schematic representation of the experimental setup is shown in Fig. 1. The feedstock was prepared from Japanese cedar, crushed and sieved with a mesh size of mm. The feedstock was dried at a temperature of 105 C overnight in an oven, before it was packed in the feeder, to remove moisture. The proximate and ultimate analysis results of the feedstock are summarized in Table 1. The feedstock was introduced into the pyrolyzer by a controlled screw feeder with a continuous feed rate of 0.6 g/min. The pyrolyzer reactor was made of stainless steel with an inner diameter of 30 mm and a length of 280 mm, and was surrounded by an electric heater. In each experiment, the pyrolyzer was heated up to a temperature of 600 C and kept at this temperature for 30 min before starting the feeder, in order to ensure that pyrolysis tar production should be done under a steady-state operation. When the feedstock was introduced into the pyrolyzer, the feedstock released its volatiles in the form of syngas and tar aerosols, which were carried out of the pyrolyzer by nitrogen, used as a carrier gas, and entered the gas cleaning equipment with the flow rate Fig. 1. Schematic diagram of the experimental equipments. of 1.5 l/min. The product of the pyrolysis process was sampled at the exit of the reformer, which was kept at 400 C, in order to avoid tar decomposition or condensation Tar decomposition Pyrolysis tar decomposition experiments under different reforming agents were carried out in an externally heated reactor. The reactor, called as the reformer, was made of stainless steel with an inner diameter of 25 mm and a length of 1300 mm. The entire pyrolysis product in the form of gas was supplied directly into the reformer. In the tar decomposition experiment, the effect of each reforming condition was investigated on the basis of the change in gravimetric tar mass and concentration of tar compounds, and their changes were compared with the result of the pyrolysis condition. All experimental conditions of tar decomposition are summarized in Table 2. In the studies of steam and air reforming, water and fresh air were heated up to the temperature of reforming, before being supplied into the reformer for supporting the tar decomposition reaction. The parameters of steam and air reforming used to observe the effect on tar decomposition are given by the steam to biomass (S/B) ratio and the equivalent ratio (ER), respectively, where the S/B ratio is defined as the steam mass flow rate divided by the feedstock mass flow rate on the dry basis, while the ER is defined as the actual amount of air supplied to the reformer divided by the stoichiometric amount of air for complete combustion of the feedstock. Table 1 The proximate and ultimate analysis of wood chip. Proximate analysis (wt.% dry basis) Volatile matter 80.8 Fixed carbon 18.9 Ash 0.3 High heating value (MJ/kg) 19.3 Ultimate analysis (wt.% dry-ash free basis) C 49.7 H 6.4 N 0.1 O 43.8 S <0.1 Cl <0.1 Japanese cedar

3 T. Phuphuakrat et al. / Applied Energy 87 (2010) Table 2 Experimental conditions for tar decomposition Tar adsorption Pyrolysis Adsorption studies were carried out using a fixed-bed type adsorber, which was installed downstream of the reformer. The entire reformer product was supplied directly into the adsorption bed through a high temperature resistance tube connector. The reformed gas was cooled down to the temperature of C by natural convection, without any heat exchanger device. Before starting the tar adsorption experiments, the reformed gas was bypassed from the adsorption bed for 30 min in order to ensure that the upstream reactions were under a steady-state condition. Wood chip, as the same type of feedstock, and synthetic porous cordierite were selected to study the adsorption ability of tar compounds and to compare them with the adsorption ability of activated carbon, which has a good adsorption performance to adsorb hydrocarbons. In case of wood chip, it was dried in a controlled oven at a temperature of 105 C overnight before using as an adsorbent. The characteristics of each adsorbent are summarized in Table Tar sampling and analysis Thermal cracking Steam reforming Air reforming Pyrolyzer temperature Reformer temperature Reforming agent Steam Air Steam to biomass ratio Equivalence ratio Steam to biomass ratio = steam mass flow rate/feedstock mass flow rate in dry basis, equivalence ratio = actual amount of air supplied to the reformer/stoichiometric amount of air for complete combustion of the feedstock. The sampling of tar was done using either the wet type or dry type of sampling method at each sampling port. Two sampling ports were located at the exits of the reformer and the adsorption bed as shown in Fig. 1. The wet type (cold trap) is generally selected for the measurement of organic contaminants and particles in the producer gas from pyrolysis or the gasification process. However, the wet type needs long sampling period that cannot be used for a study on tar aerosol adsorption characteristics. The dry type sampling method was chosen for the purpose of tracing a change in tar concentration, to facilitate the study on tar adsorption. A simple method for dry type tar sampling and analysis has been introduced in our previous study, especially for light tar measurements, with a short sampling time, at room temperature [17]. Table 3 Adsorbents for tar adsorption study. Adsorbent Activated carbon Wood chip Synthetic porous cordierite Shape Granular Chip Pellet Mean particle diameter (mm) Volume of adsorbent (cm 3 ) Weight of adsorbent (g) Advantage Good adsorption performance Low cost Re-usable (regenerable) Manufacturer/ source material Ajinomoto Fine Techno Co., Inc. Japanese cedar Chemical Auto Co., Ltd. The previous study also reported higher measurement efficiency for light tar when compared to wet type tar sampling. In contrast, the collection efficiency of light PAH (polycyclic aromatic hydrocarbon) tars and heavy PAH tars (larger than the four-ring aromatic hydrocarbon) by dry type sampling was lower (approximately 25% for light PAH tars) than that with the wet type sampling method, as these tars might have condensed before sampling, due to the low sampling temperature [18]. Therefore, the wet type was selected to collect and measure products of the pyrolysis, which contained a lot of heavy tar and products of the reformer, so as to determine the tar removal efficiency of the reformer. On the other hand, the dry type was selected to investigate the change in concentrations of tar compounds by adsorption treatment. In this study, the tar removal performance was indicated by a decrease in either gravimetric tar mass or concentration of light tar compounds, that is, benzene, toluene, xylene, styrene, phenol, indene, naphthalene, phenanthrene, anthracene, and pyrene The wet type method This tar and particle sampling system consists of the tar collection module and the backup VOC collector as shown in Fig. 2. The wet type tar sampling used in this study was taken followed by the sampling and analysis tar of Biomass Technology Groups (BTGs). The accuracy of sampling was taken by BTG which reported in term of repeatability (5.2 29%) [19]. The tar collection module consists of two series of impinger bottles containing a solvent for tar absorption, placed separately in two cold baths. The first six impinger bottles were immersed in a mixture of salt/ice/water at a temperature below 5 C, whereas, the next four impinger bottles were immersed in isopropanol. The isopropanol bath was cooled by a mechanical cooling device to maintain a temperature under 20 C which tar and moisture will be completely collected. The tar aerosols were collected by both condensation and absorption in the solvent. The sampling gas with the controlled volume flow rate of approximately 0.8 l/min was passed through a total of 10 impinger bottles (250 ml). Each impinger bottle was filled with approximately 100 ml of isopropanol, which was considered to be the most suitable solvent for tar absorption [11]. A set of backup VOC adsorber was installed downstream of the series of impinger bottles to protect the column of the gas chromatography from the residual solvent or VOCs, which may have passed through the impinger train. The set of backup VOC adsorber consists of two cotton filters and an activated carbon filter connected in a series. Immediately after completing the sampling, the content of the impinger bottles were filtered through a filter paper. The filtered isopropanol solution was divided into two parts; the first was used to determine the gravimetric tar mass by means of solvent distillation and evaporation, and the second was used to determine the concentrations of light tar compounds using flame ionization detector type gas chromatography (GC-FID) The dry type method The dry type sampling method collects tar aerosols, using a charcoal tube and a silica gel tube connected in a series, using adsorption and condensation strategies, under room temperature. These sampling tubes are commercially available, supplied by Sibata Scientific Technology Ltd. The accuracy of sampling is approximate % tested by the manufacturer for some aromatic hydrocarbon standards. The charcoal tube contains a double layer of activated carbon of 100 mg and 50 mg, respectively, in a glass tube with an outer diameter of 6 mm, whereas, the silica gel tube contains a double layer of silica gel of 520 mg and 260 mg, respectively, in a glass tube with an outer diameter of 8 mm. The charcoal tube mainly adsorbs non-polar organic compounds without any cooling requirement, and the silica gel tube mainly adsorbs polar organic compounds. Therefore, the use of

4 2206 T. Phuphuakrat et al. / Applied Energy 87 (2010) Purge Suction pump P Flow adjustor GC-TCD Flow meter Activated carbon Cotton filter II Cotton filter I Sampling port COOLER Mixture of salt /water/ice bath Isopropanol bath (Temperature 3± 1 C) (Temperature -22±1 C) Fig. 2. Wet type tar sampling train. both the charcoal tube and the silica gel tube ensures that all organic compounds will be collected in the sampling tubes. After switching the gas flow into the adsorption bed, the sampling was done at the gas sampling ports located at the inlet and exit of the adsorption bed as shown in Fig. 1. The gas was sampled with the constant flow rate of 0.5 l/min for 3 min according to the recommendation of the manufacturer of the sampling tubes. The gas sampling controller consists of a flow meter, a flow control valve, and a suction pump, respectively, connected in a series, next to the sampling tubes, as shown in Fig. 3. Immediately after completing the sampling, these tubes were kept in a refrigerator, to prevent evaporation of the adsorbed tar. The charcoal tube was desorbed by carbon disulfide of 1 ml in each layer, whereas, the silica gel tube was desorbed by acetone of 2 ml in each layer. The desorbed solution was allowed to settle for 2 h before determining light tar compounds and their concentrations by GC-FID. More details about dry type tar sampling and analysis have been described in our pervious paper [17] Porous texture characterizations Characterization of the porous texture of the adsorbents was investigated by determining N 2 and Ar gas adsorption and desorption isotherms at 77 K and 87.5 K, respectively, using an Autosorb-I instrument (Quantachrome, USA). The Bruanuer Emmett Teller (BET) approach using adsorption data in the linear part, over the relative pressure range of , in accordance with the IUPAC recommendations, was utilized to determine the specific area of the samples. The Barrett Joyner Halenda (BJH) approach was used to calculate the pore size distribution using the adsorption data. The pore volume was determined from the maximum amount of adsorption at the relative pressure of Results and discussion 3.1. Effects of tar decomposition by thermal cracking and reforming The high temperature reaction had a significant effect on tar decomposition. The effects of tar decomposition were observed by changing either the quality or quantity of the reforming agents, at a constant temperature of 800 C, and upstream conditions. The experimental results for gravimetric tar mass reduction under different cracking and reforming conditions are summarized in Fig. 4. Without any addition of reforming agents, a reduction of 78% in the g/g-feedstock Gravimetric tar Fig. 4. Tar mass contribution before and after decomposition. Sampling port Charcoal tube Silica gel tube Purge P Suction pump Flow control valve Flow meter Fig. 3. Dry type tar sampling train.

5 T. Phuphuakrat et al. / Applied Energy 87 (2010) mg/g-feedstock Pyrolysis Thermal cracking S/B = 1 S/B = 2 S/B = 3 ER = 0.2 ER = 0.3 ER = Fig. 5. Light tar contribution before and after decomposition. gravimetric tar mass was observed under the thermal cracking condition. Similar to the investigation by Rath and Staudinger, approximately 22% of the tar mass remained [20]. Significant improvement in tar decomposition was observed by using reforming agents. Both steam and air enhanced the tar decomposition reaction. With an increase in the amount of a reforming agent, the gravimetric tar mass decreased. A similar tar decreasing trend was reported with respect to an increase in the amount of a reforming agent [21]. By increasing the amount of steam to the S/B ratio of three, however, only a slight decrease in gravimetric tar mass was observed. Despite a decrease in the gravimetric tar mass, both reforming and thermal cracking increased almost all the light tars (class 3 and 4 of tar). It is remarkable that the pyrolysis tar was composed mainly of heavy tars (class 5 of tar). The effects of decomposition or formation of light tars are shown in Fig. 5. With a high temperature reaction it is possible to break the chemical bonds of heavy tars, which results in the formation of lower aliphatic or aromatic hydrocarbons. It can be stated, therefore, that almost all the pyrolysis tar can be decomposed to light tars and gases, while only a small portion of the pyrolysis tar can hardly be decomposed. Thermal cracking had a stronger effect on the production of benzene and naphthalene compared to reforming. It can be attributed to the fact that benzene and naphthalene are stable aromatic compounds that are formed during the thermal cracking reaction. In air reforming, the yields of light tars were obviously increased, especially for light polyaromatic hydrocarbon tars (light PAH tars; class 4 of tar). One of the reasons for the enhancement of the formation of light PAH tars in air reforming conditions can be explained by the incomplete combustion of carbon-containing fuels, especially by the partial oxidation of pyrolysis tar. By increasing the ER from 0.2 to 0.4, the formation of all light PAH tars decreased. The increase in the ER also led to a decrease in the formation of light aromatic hydrocarbon tars (class 3 of tar). In contrast, the increases in both light aromatic hydrocarbon tars and light PAH tars were observed by increasing the S/B ratio, except for benzene and naphthalene. The highest yield of each light aromatic hydrocarbon tar was observed at the S/B ratio of three, excluding benzene. Benzene was present in the S/B ratio of 1, but not in the S/B ratio of two or three. The formation of benzene can proceed via the breakage of other light aromatic hydrocarbon tars. The increase in S/B ratio leads to a decrease in the gas residence time in the reformer. The formation of benzene may need a longer gas residence time in order to break the chemical bonds of the function groups in other light aromatic hydrocarbon tars [2]. Therefore, an increase in the S/B ratio resulted in the decrease in benzene content. Fig. 6 shows the detailed gas distribution before and after pyrolysis tar decomposition by thermal cracking or reforming. During thermal cracking, an increase in all gas products was observed. An increase in carbon dioxide and carbon monoxide after thermal cracking indicated that pyrolysis tar was composed of hydrogen, carbon, and oxygen. Similar gas production by thermal cracking mol/kg-feedstock Pyrolysis Thermal cracking S/B = 1 S/B = 2 S/B = 3 ER = 0.2 ER = 0.3 ER = H5 CH4 CO CO2 C2H 4 C2H 6 Fig. 6. Gas concentration before and after decomposition.

6 2208 T. Phuphuakrat et al. / Applied Energy 87 (2010) has been reported in the other study [5]. Significant hydrogen production was observed in the steam reforming conditions. Carbon monoxide reacted with steam during the water gas reaction to produce hydrogen and carbon dioxide. By increasing the S/B ratio, the formation of hydrogen and carbon dioxide increased, but the formation of carbon monoxide decreased. A decrease in the yield of ethylene and ethane was also clearly observed by increasing the S/B ratio, whereas, a decrease in the yield of methane was observed in a condition where the S/B ratio was three. It is remarkable that steam has a strong influence on their yield. Their reactions with steam promoted the production of hydrogen and carbon monoxide [22 24]. In case of the air reforming conditions, the results of gas distributions at the ER of 0.2 showed the highest formation of fuel gas products. An increase in the ER led to a decrease in the yield of hydrogen, methane, and carbon monoxide. The yield of methane at ERs of 0.3 and 0.4 was less than that in the pyrolysis condition. The decrease in the yield of each fuel gas was caused by their combustion reactions. The evidence of their combustion was observed by the increase in the concentration of carbon dioxide by increasing the ER Performance of tar adsorption of each adsorbent for fixed-bed adsorption Comparison of the porous texture characteristics of each adsorbent The porous texture characteristics of each adsorbent were measured (see Table 4). The activated carbon has pore size in the range of micropore, whereas, pores of wood chip and synthetic porous cordierite can be arranged in the range of mesopore. Moreover, although cordierite is generally a macroporous material, the one used in this study has been prepared by the manufacturer using its own technique, in order to improve the porosity. As a result, synthetic porous cordierite has a smaller pore size, but not as small as the microporous material. However, the surface area and pore volume cannot be much improved. Most practical adsorbents commonly selected for adsorption of hydrocarbons are porous materials in the range of micropore with a high surface area and high pore volume [13,14,25]. Therefore, we can expect activated carbon to show the best adsorption performance. Table 6 Adsorption capacity of each tar component. Amount adsorbed (g/g adsorbent) Activated carbon Wood chip Synthetic porous cordierite Benzene < Toluene < Xylene < Styrene < Phenol Indene Naphthalene Phenanthrene Anthracene Pyrene < < Sum Comparison of the adsorption capacity of each adsorbent The adsorption capacity of each adsorbent for the remaining tar coming from the reformer is presented in Table 5. The adsorption capacity was evaluated by the weight gain/weight of each adsorbent. The weight gain of each adsorbent was measured immediately after completing the experiment. This table indicates that wood chip had better adsorption capacity than activated carbon, which is contradictory to the earlier prediction based on the porous texture characteristics. In order to clarify the reason for this, the numerical integration technique was used to calculate the adsorption capacity of each tar component from their breakthrough curves as shown in Table 6. When we compared the summation of the adsorption capacities of all tar components shown in Table 6 with the overall adsorption capacity shown in Table 5, we can see that both the values are almost the same only in the case of activated carbon. A huge difference of these two values was observed for wood chip and synthetic porous cordierite. The substances adsorbed on the adsorbents were not only tar, but also water in the form of steam, produced during pyrolysis and thermal cracking processes. Although the gas stream was cooled down to the temperature of C before being introduced into the adsorption bed, the steam content in the gas was not completely removed. Therefore, it can be considered that the large amount of weight increase in the wood chip is the cause of the adsorption of steam. Because the hydrophilicity surface and mesoporous material favor water adsorption, the activated carbon adsorbed less amount of steam when compared to the wood chip and Table 4 Porous texture characterization of the adsorbents. Adsorbent Average pore size (nm) Specific surface area (S BET,m 2 /g) Total pore volume (ml/g) Activated carbon Wood chip Synthetic porous cordierite Table 5 Adsorption capacity of tar remaining after decomposition. Adsorbent Activated carbon Wood chip Synthetic porous cordierite Amount adsorbed (g/g adsorbent) Fig. 7. Breakthrough curves of activated carbon: (a) for light aromatic tar adsorption and (b) for light PAH tar adsorption.

7 T. Phuphuakrat et al. / Applied Energy 87 (2010) synthetic porous cordierite. In addition, according to Linders et al., who studied the adsorption equilibrium of hexafluoropropylene (HFP) on Norit R1 carbon, under humid conditions, HFP vapor was adsorbed in the mircopore volume left free by water due to the miscible solvent of HFP and water [26] Comparison of the adsorption performance of each adsorbent The breakthrough curves of each adsorbent for each tar component are shown in Figs. 7 9, where C 0 is the inlet concentration of a certain tar component and C is the exit concentration of that tar component. From these figures, the adsorption performance of each adsorbent can be found as follows: Activated carbon. The highest adsorption performance for both light aromatic hydrocarbon tars and light PAH tars was clearly observed from the breakthrough curves of the activated carbon adsorbent. Moreover, their breakthrough times were longer than the period of experiments. The highest adsorption for all light tars of activated carbon can be explained by its porous texture characterization. It is noted that this micropore material with a higher pore volume and larger surface area is the best adsorbent for all light tars. This observation is in agreement with that reported in the study of the adsorption performance of different porous texture characterization materials [13 15]. Some papers have also reported that the adsorbents had a negative influence of the presence of carbon dioxide and steam on hydrocarbon adsorption [13,27], but the present experimental data do not show any such prominent adverse effects on the adsorption of both light aromatic hydrocarbon tars and light PAH tars. After passing through the activated carbon adsorbent, the amount of residual tar in the syngas was very small and at the permissible level for downstream components. Fig. 9. Breakthrough curves of synthetic porous cordierite: (a) for light aromatic tar adsorption and (b) for light PAH tar adsorption. From the viewpoint of tar decomposition, most of heavy tars were cracked or reformed into light tars and gases in the reformer. If light tars can be delivered to a combustor or an engine without their condensation, the condensing problem in the downstream equipments will not occur, and thermal efficiency can be improved by the effective use of the calorific value of light tars. Some researchers have reported on light tars condensation [3,4,28,29]. They argued that tar condensation is related to the properties and compositions of tar. Not all light tars cause condensing problem. Therefore, only the condensable tar (light PAH tars) is recommended to be removed before supplying the syngas to downstream energy converters. In this regard, the high adsorption performance of activated carbon for all light tars may lead to a decrease in the thermal efficiency of the gasification system. The present experiment showed that activated carbon could reduce the concentration of residual condensable tar down to approximately 65 mg/m 3, while the calorific value loss reaches to approximately 20%, due to its high adsorption performance for noncondensable tar. Fig. 8. Breakthrough curves of wood chip: (a) for light aromatic tar adsorption and (b) for light PAH tar adsorption Wood chip. Wood chip could not adsorb most of the light aromatic hydrocarbon tars, except phenol, due to its water solubility. When water is adsorbed on the surface of the wood chip, phenol will be captured as it dissolves in the adsorbed water. Moreover, the surface chemistry of wood chip also influences the adsorption of phenol, due to its hydroxyl group structure. On the other hand, the adsorption performances of wood chip for light PAH tars were almost similar to that of activated carbon, except for naphthalene. The naphthalene concentration at the exit of the adsorption bed gradually increased, and the value of C/C 0 reached 33% after passing for 54 min from the start of the adsorption experiment.

8 2210 T. Phuphuakrat et al. / Applied Energy 87 (2010) The breakthrough curves showed that wood chip did not adsorb non-condensable tar. Such non-condensable tar is easily adsorbed by micropore adsorbent materials such as activated carbon [14], while the pores of wood chip are in the range of mesopore. Therefore, non-condensable tar can hardly be adsorbed by wood chip. It is expected that the thermal efficiency of the gasification system will not decrease due to the poor adsorption of non-condensable tar when wood chip is selected as adsorbents. However, the condensable tar adsorption performance of wood chip was lower than that of activated carbon, and the concentration of condensable tar at the exit of the adsorption bed of wood chip was quite high (1027 mg/m 3 ), exceeding the allowable level for internal combustion engines [21,30]. A smaller size of the wood chip and a lower gas flow rate may improve the condensable tar adsorption performance Synthetic porous cordierite. Considering both the adsorption performance and the capacity of synthetic porous cordierite for light tars, the breakthrough times for most of the light aromatic hydrocarbon tars and light PAH tars were shorter than the experimental period, except for phenanthrene, anthracene, pyrene, and phenol. That is, synthetic porous cordierite adsorbed light tars well at the beginning of the experiment, and the adsorption potential gradually decreased to almost zero. The adsorption performance for phenanthrene, anthracene, and pyrene monotonically decreased during the period of the experiment, but did not reach zero. In the case of phenol, the exit concentration was only less than 10% of the inlet concentration even at the end of the experiment. When synthetic porous cordierite fully adsorbed each light aromatic hydrocarbon tar, the outlet concentration of such a tar component was higher than the inlet concentration, which meant that C/C 0 exceeded 100%. This is caused by the displacement of one tar component by the other. This behavior agreed with the previous researches on the adsorption of hydrocarbon mixtures [21,31]. Here, it should be noted that using synthetic porous cordierite is inadequate to produce tar-free syngas. Although the decrease of calorific value is not so much as compared to using activated carbon, synthetic porous cordierite has a very low adsorption capacity for the adsorption of condensable tar. 4. Conclusions Tar elimination efficiencies of thermal tar decomposition combined with physical tar adsorption were investigated by using a reformer as the first step and a fixed-bed adsorber as the second step. Thermal tar decomposition was carried out at a temperature of 800 C. High temperature had a strong influence on tar decomposition. The pyrolysis tar was effectively decomposed, down to approximately 22% of the inlet level at the exit of the reformer. To improve the efficiency of tar reduction, either steam or air was introduced into the reformer, as a reforming agent. With an increase in the amount of each reforming agent, the gravimetric tar mass decreased. The maximum tar reduction rates were approximately 88% and 92% by steam and air reforming, respectively. Tar decomposition also had the benefit of increasing syngas production, while reforming could decrease either the calorific value of the syngas or the cold gas efficiency of the system. Despite a decrease in the gravimetric tar mass, heavy tar decomposition inevitably resulted in the increase of light tar production. Therefore, tar decomposition was not sufficient to eliminate tar from the syngas, to avoid damage to downstream equipment. A fixed-adsorption bed was installed downstream of the reformer. It was mainly designed to remove the light tars formed during the tar decomposition process in the reformer, but light aromatic hydrocarbon tars, which are a non-condensable tar, should not be removed. The non-condensable tar could be combusted in the downstream combustion equipment, without condensing, under operating temperature. Handling non-condensation tar led to an increase in the cost of gas treatment as well as a decrease in the thermal efficiency of the system. Three different adsorbents were conducted to study their performance of light tar removal. Activated carbon showed the best adsorption performance for both light aromatic hydrocarbon tars and light PAH tars. That is, the use of activated carbon provided very low tar content in the syngas, while it decreased the calorific value of the syngas. The adsorption result of activated carbon showed that approximately 65 mg/m 3 of condensable tar remained at the exit of the adsorption bed with about a 20% loss of the calorific value, based on the energy content of the feedstock. Although the adsorption performances of the other adsorbents for condensable tar were less than those of activated carbon, the loss of the calorific value of syngas was not so significant, as non-condensable tar adsorption was poor. When focusing on condensable tar removal, wood chip showed a better adsorption performance than synthetic porous cordierite. However, the quantity of condensable tar left after wood chip adsorption was still higher than the allowable level, for engines. Smaller size of wood chip and slower gas flow rate were expected to improve the adsorption performance of wood chip. References [1] Devi L, Ptasinski KJ, Janssen FJJG. A review of the primary measures for tar elimination in biomass gasification processes. Biomass Bioenergy 2003;24: [2] Devi L, Ptasinski KJ, Janssen FJJG. Decomposition of naphthalene as a biomass tar over pretreated olivine: effect of gas composition, kinetic approach, and reaction scheme. Ind Eng Chem Res 2005;44: [3] Bergman PCA, van Paasen SVB, Boerrigter H. The novel OLGA technology for complete tar removal from biomass producer gas. In: Pyrolysis and gasification of biomass and waste, expert meeting. France: Strasbourg; [4] Könemann HWJ, van Paasen SVB. OLGA tar removal technology; 4MW commercial demonstration. In: 15th European biomass conference and exhibition, Berlin, Germany; [5] Antal MJ. Effects of reactor severity on the gas-phase pyrolysis of cellulose- and kraft lignin-derived volatile matter. Ind Eng Chem Prod Res Dev 1983;22: [6] Coll R, Salvadó J, Farriol X, Montané D. Steam reforming model compounds of biomass gasification tars: conversion at different operating conditions and tendency towards coke formation. Fuel Process Technol 2001;74: [7] Garcia XA, Hüttinger KJ. Steam gasification of naphthalene as a model reaction of homogeneous gas/gas reactions during coal gasification. Fuel 1989;68: [8] Jess A. Mechanisms and kinetics of thermal reactions of aromatic hydrocarbons from pyrolysis of solid fuels. Fuel 1996;75: [9] Jess A. Catalytic upgrading of tarry fuel gases; a kinetic study with model components. Chem Eng Process 1996;35: [10] Simell PA, Hepola JO, Krause AOI. Effects of gasification gas components on tar and ammonia decomposition over hot gas cleanup catalysts. Fuel 1997;76: [11] Neeft JPA, Knoef HAM, Zielke U, Sjöström K, Hasler P, Simell PA, et al. Guideline for sampling and analysis of tar and particles in biomass producer gases, Version 3.3, Energy project ERK6-CT (Tar Protocol), Report No. ECN , Energy research Center of the Netherlands; [12] Wang Y, Yoshikawa K, Namioka T, Hashimoto Y. Performance optimization of two-staged gasification system for woody biomass. Fuel Process Technol 2007;88: [13] Hu X, Hanaoka T, Sakanishi K, Shinagawa T, Matsui S, Tada M, et al. Removal of tar model compounds produced from biomass gasification using activated carbons. J Jpn Inst Energy 2007;96: [14] Lillo-Ródenas MA, Fletcher AJ, Thomas KM, Cazorla-Amorós D, Linares-Solano A. Competitive adsorption of a benzene toluene mixture on activated carbons at low concentration. Carbon 2006;44: [15] Mastral AM, Garcia T, Callen MS, Navarro MV, Galban J. Assessment of phenanthrene removal from hot gas by porous carbons. Energy Fuels 2001;15:1 7. [16] Mastral AM, Garcia T, Murillo R, Callen MS, Lopez JM, Navarro MV. Measurements of polycyclic aromatic hydrocarbon adsorption on activated carbons at very low concentrations. Ind Eng Chem Res 2002;42: [17] Son Y, Sato M, Namioka T, Yoshikawa K. A study on measurement of light tar content in the fuel gas produced in small-scale gasification and power generation systems for solid wastes. J Environ Eng 2009;4:12 23.

9 T. Phuphuakrat et al. / Applied Energy 87 (2010) [18] Mukaida H, Son Y-i, Sato M, Namioka T, Kunio Y. Measurement of light tar contents in the fuel gas produced in waste gasification systems. In: 2nd Korea Japan joint lab workshop, Yokohama, Japan; [19] Good J, Ventress L, Knoef H, Zielke U, Hansen PL, Van de Kamp W, et al. Sampling and analysis of tar and particles in biomass producer gases, Technical Report, BTG biomass technology group, prepared under: CEN BT/ TF 143 Organic contaminants ( tar ) in biomass producer gases ; [20] Rath J, Staudinger G. Cracking reactions of tar from pyrolysis of spruce wood. Fuel 2001;80: [21] Morf P, Hasler P, Nussbaumer T. Mechanisms and kinetics of homogeneous secondary reactions of tar from continuous pyrolysis of wood chips. Fuel 2002;81: [22] Didenko L, Savchenko V, Arutyunov V, Sementsova L. Steam reforming of methane mixtures with ethylene over an industrial nickel catalyst. Petrol Chem 2008;48:22 7. [23] Laosiripojana N, Assabumrungrat S. Catalytic steam reforming of ethanol over high surface area CeO 2 : the role of CeO 2 as an internal pre-reforming catalyst. Appl Catal, B 2006;66: [24] Laosiripojana N, Assabumrungrat S. Catalytic steam reforming of methane, methanol, and ethanol over Ni/YSZ: the possible use of these fuels in internal reforming SOFC. J Power Sources 2007;163: [25] Mastral AM, GarcIa T, Callen MS, Navarro MV, Galban J. Removal of naphthalene, phenanthrene, and pyrene by sorbents from hot gas. Environ Sci Technol 2001;35: [26] Linders MJG, Broeke LJPvd, Kapteijn F, Moulijn JA, Bokhoven JJGMv. Binary adsorption equilibrium of organics and water on activated carbon. AIChE J 2001;47: [27] Mastral AM, Garcia T, Murillo R, Callen MS, Lopez JM, Navarro MV. Effects of CO 2 on the phenanthrene adsorption capacity of carbonaceous materials. Energy Fuels 2002;16: [28] Han J, Kim H. The reduction and control technology of tar during biomass gasification/pyrolysis: an overview. Renew Sust Energy Rev 2008;12: [29] Li C, Suzuki K. Tar property, analysis, reforming mechanism and model for biomass gasification an overview. Renew Sust Energy Rev 2009;13: [30] Hasler P, Nussbaumer T. Gas cleaning for IC engine applications from fixed bed biomass gasification. Biomass Bioenergy 1999;16: [31] Lee SW, Kam SK, Lee MG. Comparison of breakthrough characteristics for binary vapors composed of acetone and toluene based on adsorption intensity in activated carbon fixed-bed reactor. J Ind Eng Chem 2007;13:911 6.