Вы находитесь на странице: 1из 10

Energy Conversion and Management 50 (2009) 14261435

Contents lists available at ScienceDirect

Energy Conversion and Management


journal homepage: www.elsevier.com/locate/enconman

Experimental and modeling analysis of a batch gasication/pyrolysis reactor


P. Baggio a, M. Baratieri b,*, L. Fiori a, M. Grigiante a, D. Avi c, P. Tosi c
a b c

Department of Civil and Environmental Engineering, University of Trento, Via Mesiano 77, 38050 Trento, Italy Faculty of Science and Technology, Free University of Bolzano, Sernesi street 1, 39100 Bolzano, Italy Department of Physics, University of Trento, Via Sommarive 14, 38050 Trento, Italy

a r t i c l e

i n f o

a b s t r a c t
This paper presents some experimental results about biomass gasication obtained with a bench-scale gasier consisting of an indirectly heated batch reactor, inserted in a high temperature furnace. Spruce wood has been used as feedstock in different forms and sizes (as pellets and sawdust). The experimental activity includes the analysis of the thermal response of the system (using an inert bed material) and the characterization of the gasication products. The yield of the gaseous compounds found in the syngas has been measured using an on-line gas-chromatography technique. The experimental results have been compared against calculations obtained by applying a thermochemical equilibrium model, improved to predict both the gas and the solid phase product yields. The experimental (average) yield of reaction products has been found to be in a satisfying agreement with the thermodynamic model. 2009 Elsevier Ltd. All rights reserved.

Article history: Received 1 October 2007 Received in revised form 24 July 2008 Accepted 8 March 2009 Available online 5 April 2009 Keywords: Batch reactor Gasication Thermodynamics Chemical equilibrium

1. Introduction In the current energetic outlook, one of the strategies pursued is the energy generation from renewable sources, in order to reduce the fossil fuels dependence and to rely on environmental-compatible technologies. In this context, the biomass gasication process is considered as a viable fuel upgrading path, allowing the transformation of a solid fuel into a gaseous energy vector (i.e., syngas). This chance is particularly interesting for the current scenario, in which small and medium size conversion plant technologies could be integrated in a distributed energy generation model that is expected to increase its diffusion. In this perspective the biomass gasication represents an alternative to improve innovative energy processes, in particular for small rural communities. In fact, the large amount of waste biomass, widely spread on rural territories, looks promising for syngas production as fuel for electric power generation by direct utilization in conventional internal combustion engines or, after clean up and reforming stages, for innovative generation systems as fuel cells. This work presents the experimental results coming from a biomass gasication pilot plant and the assessment of its performance based on an equilibrium modeling approach. The proposed work refers specically to experimental activities carried out on a small allothermic batch reactor, in line with other studies reported in recent literature [1]. The apparatus, designed by the authors for the purpose of the present research, is presented in detail. The conguration of the reactor allows to investigate the
* Corresponding author. Tel.: +39 0471017127; fax: +39 0471017009. E-mail address: marco.baratieri@unibz.it (M. Baratieri). 0196-8904/$ - see front matter 2009 Elsevier Ltd. All rights reserved. doi:10.1016/j.enconman.2009.03.004

thermal response of the system itself and the energy transport phenomena occurring between the reactor shell and the bed. The role of the thermal contact resistance (reactor/bed) has been investigated. This is noteworthy noting that both modeling [2,3] and experimental works [4] concerning pyrolysis and gasication investigations focus on the heat transfer phenomena affecting single particles. This approach does not take into account the general view of the process that is controlled by heat transfer at a larger scale including contact resistance with the reactor wall. Other studies provide experimental thermal analysis applied to combustion furnaces [5] or auto-thermal pyrolysis-combustion systems [6,7], while these studies focus on allothermic conversion processes. An enhanced thermochemical equilibrium model has been implemented to describe the gasication/pyrolysis conversion yields. The adopted methodology based on the Gibbs energy minimization approach does not require the selection of the representative chemical reactions. Nevertheless, only the list of the chemical species inclusive of the ones expected in the product mixture must be assigned. This model, previously developed and widely detailed [8], represents an effective engineering tool to support the design of actual biomass conversion plants. Other approaches to estimate the product composition are available, such as kinetic/dynamic models [9,10] and neural network applications [11] allowing in some instances to obtain a better accuracy. Nevertheless, the proposed approach presents a high predictive character and does not require an extended set of data to train the model itself. More, the used solidgas formulation represents an improvement inside the traditional equilibrium calculations generally based on a single gas phase formulation [1218]. Similar

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

1427

applications have been presented in the literature only recently [1923]. The contribution of this paper encompass experimental results, theoretical analysis and model validation, and shows also that slow batch gasication processes can be treated in terms of the chemical reaction equilibrium. 2. Gasication equipment 2.1. Overview An experimental bench-scale apparatus for biomass gasication and pyrolysis has been designed: it consists of an indirectly heated batch reactor operating at variable pressure (17 barg), set in an external furnace capable of reaching temperatures up to 1000 C [24]. The reactor (Fig. 1) is inserted from above in the laboratory cylindrical furnace, open on the top side, where all the control and acquisition devices are connected. The vessel, half lled of feeding material (biomass), can be continuously purged with nitrogen to sweep the product gases from the reaction zone, either to avoid high concentration of the products in the efuent gas or to reduce the extent of secondary reactions. The temperature is monitored by a set of thermocouples placed in the reactor. In order to assess the yields of the syngas and its composition as a function of the parameters able to inuence the process principal reactions, the characterization of the efuent gas has been performed by means of gas-chromatographic analysis on gas samples. 2.2. Reactor The reactor consists of a cylindrical-shaped Incoloy vessel having internal diameter of 142 mm, height of 400 mm, thickness of 4 mm and internal net volume of 6.3 l. Two blind anges (210 mm diameter and 20 mm thickness), made of the same material, seal the tubular reactor on the top and bottom sides. The top ange is equipped with eight pipes (diameter 10 mm length 300 mm), used both for inserting the measure and control devices and for the gas inlets and outlets; on the same

ange there is also a central ue gas discharge duct, provided with a ball valve. The total reactor height, including anges and ancillary pipes, is equal to 746 mm. The reactor air-tightness is assured by 12 bolts and nuts, fastened on each ange; moreover, union sleeves with threads have been welded on the top of all the inlet/outlet tubes, in order to connect the necessary measure probes. The reactor has been manufactured using Incoloy 800 (30%Ni, 21%Cr, 46%Fe and other metals in traces), an alloy suitable for operating at high temperatures and pressures. This is a material widely used for construction of equipment requiring high corrosion resistance, heat resistance, strength, and stability for service up to 800 C. At high temperatures it offers resistance to oxidation, carburization and suldation along with rupture and creep strength. The reactor design parameters are a maximum temperature of 800 C at 3 barg pressure and 700 C at 7 barg. 2.3. Furnace The reactor is inserted in a pit furnace with a circular cross-section of 100 cm (external diameter) and 98 cm height; the net internal volume is about 51 l with 33 cm diameter and 60 cm height. The furnace is open at the top side, with a 21 cm circular hole. There is an insulation layer (33.5 cm thick) that keeps the external shell at low temperature (lower than 60 C); moreover, the internal chamber is lined by a coat of refractory material. The reactor can be handled by means of an electric cable hoist. The control panel is installed on the right side of the furnace and includes the master switch, the heating switch, the hoist switch, two temperature controllers, an amperometer, a voltmeter and a temperature safety alarm. The heating is provided by several electric resistances (maximum power 10.5 kW at 230 V, 50 Hz), up to a maximum work temperature of 1000 C. The power supplied to the resistances is regulated by a PID temperature controller (Fuji PX3 series) controlling both the maximum temperature (i.e., the desired temperature) and the heating rate (temperature ramp). After selecting the heating rate and the set point (it is possible to select up to 8 different set points and as many ramp for the same experimental run), the

Fig. 1. Gasication equipment: overview (top left); pit furnace (top right); reactor (bottom right) and reactor design data (bottom left).

1428

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

furnace begins the operation cycle using a constant power, proportional to the heating rate value, until the temperature is close to the set point, and then it reaches the set value (and keeps it stable) by means of a pulse width modulation (PWM) control strategy. Pulse-width modulation control works by switching the power supplied to the furnace electric resistors (i.e., the heating system) with a variable duty cycle. A second safety temperature controller is installed on the furnace panel, cutting the power to the resistors in case the temperature exceeds the safety temperature set point (usually selected 20 C above the rst controller set point). The temperature is measured using two K-type thermocouples insulated by a refractory sheath and inserted into the furnace through the insulation layer at 26 cm from the top side of the furnace.

2.4. Data acquisition unit and measurement devices The data acquisition unit consists of an Agilent34970A digital multimeter (DMM 6 accuracy) with a 20-channel plug-in module connected to a PC by means of a RS-232 port. For the temperature measurements inside the reactor, K-type thermocouples (ChromelAlumel) have been used, because they are well suited for corrosive and oxidizing environments and have a wide operating temperature range, between 200 C and 1370 C (accuracy range 1.1 C to 2.2 C); the DMM internal reference junction has been used since its accuracy (0.5 C) is of the same order of magnitude of the K-type thermocouples. To measure the temperature and its variation during the experimental runs, four thermocouples have been used placed as follows: 1. furnace: between the internal furnace wall and the external reactor wall; to measure the furnace heating rate; 2. wall: on the external reactor wall (usually at 50 mm from the reactor bottom); to measure the wall heating rate; 3. bed: in the biomass bed (usually at 30 mm from the reactor bottom); to measure the feedstock thermal behavior; 4. internal: in the upper side of the reactor (at 130 mm from the reactor top); to measure of the syngas temperature. Two different kinds of thermocouple junctions have been used: bare wire (for the furnace thermocouple) and sheathed (for the wall, bed and internal ones). The former typology consists of Chromel and Alumel wires (with glass ber insulation) twisted together at the tip so that a good mechanical contact is achieved. Sheathed thermocouples have a welded junction protected by an Inconel600 sheath (diameter: 3.2 mm and length: 500800 mm) with mineral oxide insulation; these thermocouples (usually named MIMS, i.e., mineral-insulated, metal-sheathed) have the advantage of the protection offered by a metal-sheath, while still retaining a reasonable amount of exibility. All the thermocouples have been calibrated before and after the experimental runs, by means of a precision bath (Hart Scientic series) and a reference Platinum Resistance thermometer (PT100). A 90 cm long tube, connected to the central reactor outlet (located on the top ange), acts as discharge duct. The discharge duct is open to the atmosphere and the tar compounds tend to condense on the chimney cold surfaces, thus a small container (for the condensate) has been located at the efuent discharge point. The gas sampling line consists of a 1/40 stainless steel tube connected to the discharge duct and equipped with a sampling valve and a diaphragm pump (N 820, KNF series, max ow rate 20 l/min) to let the desired syngas amount to ow towards the gas analysis unit.

Between the gas sampling valve and the gas analyzer a tar trap and a particulate lter have been installed; in particular a lter ask collects the liquid coming from a condensation coil immersed in a cooling water tank, while the remaining gases are piped to the analyzing unit. When nitrogen has been used as the inert gas carrier, in order to continuously purge the vessel, a owmeter (MKS Instruments series, operating range 220,000 Scm3) has been installed between the gas cylinder and the reactor inlet. The analyzing unit is an Agilent micro gas-chromatograph (3000 micro-GC series, repeatability less than 0.5% RSD at constant pressure and temperature) well suited for renery and reformer gas analysis. It consists of two complementary chromatographic channels, each consisting of an injection device, a separation column and a detector, that analyze a portion of sample in parallel. In the performed experiments Argon has been used as inert gas carrier, detecting nine different gaseous compounds: H2, O2, N2, CH4, CO, CO2, C2H4, C2H6 and C2H2. 3. Thermal analysis of the system 3.1. Experimental To characterize the thermal response of the system (furnace and reactor) six runs (named TH-type) have been carried out at different operating conditions as summarized in Table 1. In these experiments the reactor was lled with inert materials (gravel and sand) or empty (test 3TH). All the experiments have been performed at atmospheric pressure and at peak temperatures of 600 C. The same temperature measurement system has been used for all the runs. The characterization of the system thermal response is an important issue for the design of pyrolysis or gasication reactors, since the syngas production processes (gasication or pyrolysis of solid matter) need a signicant energy supply to support the conversion reactions (usually endothermic); the heat transfer phenomenon, between the vessel and the bed material subjected to the thermal treatment, is of primary importance for the conversion process efciency. Thus the experimental data have been used to test the application of a nite element model for the analysis of the temperature distribution inside the bed and of the contact thermal resistance between the bed material and the reactor wall. These experiments have been carried out using an inert feedstock, with granulometry similar to that of the biomass used in the actual experimental runs. In particular, two different types of stony material have been utilized: gravel (diameters between 6 and 10 mm) and sand (diameters between 1 and 3 mm), similar in dimension to wood pellets and sawdust, respectively. The ambient temperature (i.e., the temperature of the external environment, Tamb) was in the range 1722 C and the peak value of 600 C was reached by means of a heating rate equal to approximately 30 C min1. Fig. 2a and b shows the results of the thermal analysis carried out with the empty reactor (3TH) and with the reactor lled (par-

Table 1 TH-type experimental runs. Run 1TH 2TH 3TH 4TH 5TH 6TH Load Gravel Gravel Empty Gravel Sand Sand Wet Dry Dry Wet Dry Initial mass (g) 2450 2450 2450 2500 2500 Maximum temperature (C) 600 600 600 600 600 600

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

1429

700 600

300

250

Temperature [C]

Temperature [C]

500 400 300 200 100 0 0 2000 4000 6000 8000 10000 12000

200

150
T pellets (1BIO)

T bottom T internal T wall T furnace

T gravel wet (1TH) T gravel dry (4TH) T sand wet (5TH) T sand dry (6TH)

100
T sawdust (2BIO)

50

0 0 500 1000 1500 2000 2500 3000 3500

time [s]
700 600

time [s]
Fig. 3. Thermal eld behavior: comparison between inert and reactant materials.

Temperature [C]

500 400 300


T bed

Table 2 Experimental runs. Run 1BIO 2BIO 3BIO 4BIO 5BIO 6BIOa 7BIOa
a

Load Pellets Sawdust Sawdust Sawdust Sawdust Sawdust Sawdust Wet Wet Wet Wet Dry Wet Wet

Initial mass (g) 375 150 300 300 285 300 300

Maximum temperature (C) 600 600 600 800 800 800 800

200 100 0 0 2000 4000

T internal T wall T furnace

6000

8000

10000

12000

Use of N2 ux.

time [s]
Fig. 2. Thermal eld behavior: (a) empty reactor, 3TH run and (b) dried gravel, 4TH run.

tially) with dried gravel (4TH, sample mass of 2450 g, bed volume of 1.54 dm3, bed height of 100 mm). The rst gure shows the variation of the thermal eld for the empty vessel (3TH run). During the initial transient, at the beginning the temperature increase affects the external wall of the reactor (Twall), then the internal gases (air and Tinternal) and nally, the internal surface of the reactor bottom ange (Tbottom). After reaching stationary conditions (more than 9000 s runtime) all the temperatures tend to the set point value (600 C, asymptotic behavior) except for the Tinternal, that is affected by some heat losses from the reactor upper ange. Comparing Fig. 2a (empty) and b (gravel bed), it is worth mentioning the difference between Tbottom and Tbed, showing a signicant delay of the gravel thermal response (Tbed, measured at 30 mm from the vessel bottom and 30 mm from the internal wall); this is due to the low bulk thermal conductivity of the gravel bed, that has a high void fraction. Moreover, it is interesting to observe the deviation of Twall at 1800 s runtime Fig. 2b (measured on the external reactor wall at 50 mm from the bottom ange) with respect to Twall Fig. 2a since, under the same conditions (i.e., same available furnace thermal power), the global system thermal capacity is greater because of the presence of the gravel mass. In Fig. 3 the thermal behavior of different materials is reported. The curves of the inert wet materials (gravel 1TH and sand 5TH) are initially steeper than the ones of the inert dry materials (4TH and 6TH), due to the higher conductivity of the former compared to the latter. Increasing the temperature (70100 C) moisture evaporation lowers the slope of the heating curves (because of the apparent thermal capacity due to the latent heat of vaporization).

A similar behavior has been observed also for wet biomass feedstocks (1BIO and 2BIO refer to Table 2); in particular the wood sawdust heating curve is characterized by an inexion point with a positive rst derivative, while for the pellets curve the corresponding inexion point has a rst derivative value close to zero. This can be explained by the different status of the moisture in the bed material: for gravel, sand and sawdust, the moisture is mostly present as hygroscopic water in the interstices, while for the pellets there is a signicant water fraction bonded inside the single grain. In fact, the densication processes used for pellets manufacturing are carried out using starch, which causes a partial sintering of the pellet external surface. The higher temperature increase of pellets and sawdust (with respect to sand and gravel) is due to the different amount of material loaded into the vessel. 3.2. Thermal modeling As mentioned before, the experimental measurements have been compared with the results of a thermal simulation of the system carried out through a nite element thermal model. The whole system has been discretized using a 2D-axisymmetric geometry. Fig. 4 shows the geometry of the system, highlighting the points where experimental and simulated data have been compared. The Partial Differential Equations (PDEs) discretization and solution procedure has been performed by means of the Comsol Multiphysics [25] commercial software. As initial condition, an uniform temperature (ambient temperature Tamb) has been assigned to the whole computational domain, assuming that the initial state of an experimental run corresponds to the ambient temperature. For the reactor external wall a linear temperature change in time has been imposed, interpolating the experimental

1430

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

Fig. 4. Finite element model: geometric structure.

temperature data (Twall). The average value of the increasing temperature ramp, deduced from the experimental data, is approximately equal to a rate h = 32 C min1. On the bottom ange of the reactor an adiabatic condition has been assumed, while on the upper ange a natural convection ux boundary condition has been chosen (heat transfer coefcient h = 8 Wm2 K1), with the external air at Tamb. The symmetry axis (r = 0) is assumed adiabatic. The model has been calibrated in order to simulate the bed response, adjusting the (simulated and constant) bulk thermal conductivity (kbed) with a best tting procedure in order to obtain a good agreement with the experimental data (i.e., Tbed at a distance of 30 mm from the internal wall of the vessel); values equal to 0.32 and 0.35 Wm1 C1 for gravel and sand, respectively, have been obtained. Moreover, a further best tting procedure has been carried out considering a variable bed thermal conductivity (i.e., a linear and a quadratic function of the temperature). The resulting conductivity values, that vary between 0.2 and 0.4 Wm1 C1, give a better model calibration (Fig. 5) and are in agreement with some experimental data available in literature (e.g., Deru, 2003 found that

gravel thermal conductivity is an increasing function of the temperature [26]). In addition the contact resistance between the inert bed material and the reactor internal wall has been estimated. This parameter represents an additional resistance to the heat transfer and is caused by the contact between different materials (gravel or sand and Incoloy). For its calculation, the mixed bed composition (inert and air) in contact with the steel wall has been taken into account. This resistance has been simulated inserting an additional layer (2 mm thick) inside the reactor shell. The calibration of this model against experimental data gives values of the thermal resistance around 0.14 and 0.13 C W1 for the gravel and the sand bed, respectively [27]. These contact resistance values have been also compared against the value calculated using both a simplied model (lumped capacitance method) [27] and the correlation suggested by Holman [28], obtaining a good agreement (respectively 0.18 and 0.135 C W1 for gravel). In conclusion, for the considered scenarios the contact thermal resistance can be safely considered negligible. 4. Gasication activities 4.1. Experimental procedures The gasication experimental activity has been carried out by lling the reactor with spruce wood biomass (pellets or sawdust) and performing seven runs (named BIO-type) at the selected operating conditions summarized in Table 2. All the experiments have been performed at atmospheric pressure and at peak temperatures of 600 C or 800 C. The same temperature measurement system has been used for all the runs, while gas analysis has been performed only for experiments 47BIO carried out on spruce sawdust. The 6BIO and 7BIO runs have been conducted using nitrogen to continuously purge the vessel. Spruce sawdust has been used as feedstock for the 27BIO runs; only a test has been carried out on spruce pellets (1BIO) to characterize the thermal eld behavior and without performing any gas analysis. The sawdust has been characterized by means of its ultimate analysis and measuring its moisture and ash content. The ultimate analysis results (expressed in terms of d.a.f. mass fractions, i.e., on dry ash free basis) are in good agreement with other experimental data available in literature (Table 3). In particular Phyllis [29] and BIOBIB [30] databases have been chosen as they contain a large number of information on the composition of biomass (grass and wood) and waste. These databases are useful for analysis of individual available biomass or waste materials and offer the possibility to obtain an average composition of any combination of different materials. The moisture content has been measured using a precision thermobalance (Eurotherm series, 0.02% accuracy) performing several

700 600 500 400 300 200 100 0 0 3000 6000 9000 12000 15000

Temperature [C]

Table 3 Spruce sawdust characterization. (% mass, d.a.f.) C H Oa N S Moisture Ash Bulk density
a

Phyllis 50.10 6.00 43.90 0.00 0.00 (% mass, wet) (% mass, wet) (kg m3)

BIOBIB 49.53 5.77 44.50 0.19 0.01

Measured 51.58 5.90 42.44 0.08 0.00 8.25 0.90 192

time [s]
experimental data model estimate (linear) model estimate (quadratic) [ bed = a + bT ] [ bed = c + dT2]

Fig. 5. Finite element analysis: gravel thermal conductivity estimate (a = b = 2.55 104 Wm1 C2, c = 0.19 Wm1 C1 and d= 0.15 Wm1 C1, 3.00 107 Wm1 C3).

Measured by difference.

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

1431
Run: 5BIO H2
Signal out (V)
5000

runs at 105 C, drying the sample for long time (usually for a couple of hours, or even for 1 day), or using the balance constant mass function, which stops the run when the difference of the mass in a time step is less than a xed tolerance. The obtained moisture content values are in the range 7.68.7% with an average value (used for the model calculations) of 8.2%. The ash content has been measured analyzing the residual matter after an experimental run. The residual solid matter has an ash mass fraction of 4%, corresponding to a value of about 1% on the biomass sample as received. Values between 0.8% and 1.8% for spruce wood (as sawdust) have been found in literature [29,30]. The bulk density has been measured by means of a precision balance and a graduated cylinder obtaining 192 kg m3. 4.2. Solid product experimental characterization The experimental runs considered in the further analysis are the 34567BIO performed on spruce sawdust both at 600 and 800 C as maximum temperatures. These experiments have been carried out on the feedstock as received, except for the 5BIO run, performed using dried biomass. All the experiments have been conducted at atmospheric pressure and for the 67BIO, nitrogen has been uxed into the reactor. The ultimate analysis on the residual solid matter shows that it consists mainly of solid carbon (95.1% mass fraction), ash and other elements in traces (e.g., H, 0.46% and N, 0.07%). In Table 4 it is shown that for the 4567BIO runs there is a signicant production of solid residual during the process (19.5 22.5% of the wet samples). 4.3. Syngas experimental characterization Fig. 6 shows an example of the gas analysis performed during the 5BIO run by the adopted GC. The sampling rate is almost constant, with 30 measurements per hour. In the chromatograms there are nine different peaks corresponding to the many different gaseous compounds detected. The main components of the syngas are H2, CH4, CO, CO2, C2H4, C2H6, C2H2, and, possibly, some O2 and N2 due to the initial presence of air in the reactor volume not occupied by the biomass in the non-purged runs. Observing the experimental syngas molar composition versus the run time, for the non-purged experiments (e.g., 4BIO run in Fig. 7), it is possible to distinguish a pyrolysis zone (roughly extended between 1500 and 3500 s runtime), in which the conversion reactions take place producing a signicant percentage of the chemical species having a high caloric values. Before this zone (i.e., for lower syngas temperature) there are still signicant concentrations of O2 and N2 due to the presence of air in the reactor. This residual amount tends to ow out from the vessel because of the gas expansion due to the increasing temperature and because of the production of synthesis gas. On the contrary, after the pyrolysis zone the N2 molar concentration increases until it reaches an almost constant value (around atmospheric values). Since the system itself is airtight, this N2 concentration increase is probably due to air backow through the

Column 1

300000 250000

Signal out (V)

200000 150000 100000 50000 0 0 20

N2 O2
0

40

60

t (sec)

CH4

CO

40

60

t (sec)
Column 2
12000

CO2

Run 5BIO

500000 400000

H2, O2, N2
Signal out (V)
8000

C2H4

Signal out (V)

300000 200000 100000 0 0 10 20 30

4000

C2H6 C2H2
30 40 50

t (sec)

40

50

60

70

t (sec)
Fig. 6. Example of GC analysis layout.

Table 4 Residual solid (char and ash) mass fractions (% of the wet feedstock). Run 4BIO 5BIO 6BIO 7BIO Residual Char + ash Char + ash Char + ash Char + ash Mass fraction (%) 19.5 22.5 19.5 19.8

chimney occurring when the syngas production rate becomes negligible. In N2 purged experiments the conversion process has been characterized estimating the syngas yield versus the run time indirectly, i.e., calculating it from the nitrogen inlet mass ow rate and the (N2 diluted) produced gas composition (Fig. 8c). The N2 mass ow rate varies between 20 l min1 (6BIO) and 10 l min1 (7BIO). Thus, it is possible to compare purged and non-purged nitrogen free compositions for the pyrolysis zone (Fig. 8a and b shows the comparison of 47BIO run results). In Fig. 8 the oxygen concentration has not been reported, being negligible. In all the performed experiments, the biomass thermal conversion process starts with a signicant production of CO2, then, gradually, there is an increase of CO, CH4 and H2 concentrations. This early CO2 peak has been found also by other authors [31], reporting that for the thermal conversion of ne shredded wooden biomass (chips or sawdust) carbon dioxide precedes carbon monoxide formation. The effective reaction range occurs between 1500 and 3500 s run time and is characterized by the typical pyrolysis and gasication processes; specically, the gasication reactions (gas shift and reforming) use the steam coming from vaporization of the biomass water content. Table 5 shows the average syngas molar composition in the aforementioned pyrolysis zone (15003500 s run time); all the reported values are dry basis (since water is eliminated before injecting the syngas into the GC). The gaseous efuents mainly consist of the above mentioned gases with low concentrations of C2H4 and C2H6 and traces of C2H2. The small hydrocarbon molecules (alkanes, alkenes and alkynes compounds) found in the analyzed mixture are probably

1432

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

90 80
_ Molar fraction [%]

900 800 700 600 500 400 300 200 100 0 1000 2000 3000 4000 5000 6000 7000 8000 0 9000
Temperature [C]

RUN: 4BIO C2H2 C2H4 C2H6 CH4 CO CO2 H2 N2 O2 Tinternal

70 60 50 40 30 20 10 0

time [s]
Fig. 7. Syngas molar composition: experimental analysis for 4BIO run.

(a)
Molar fraction [%] _

100 90 80 70 60 50 40 30 20 10

900 800 700 600 500 400 300 200 100


Temperature [C]

RUN: 4BIO (N2 free) C2H2 C2H4 C2H6 CH4 CO CO2 H2 Tinternal

0 0 1100 1300 1500 1700 1900 2100 2300 2500 2700 2900 3100 3300 3500 3700 3900 4100 4300

time [s]

(b)
Molar fraction [%] _

100 90 80 70 60 50 40 30 20 10 0 1250 1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000 4250

900 800 700 600 500 400 300 200 100 0 4500
Temperature [C]

RUN: 7BIO (N2 free) C2H2 C2H4 C2H6 CH4 CO CO2 H2 Tinternal

time [s]

(c)
Syngas flow rate [Nl min ]
-1

3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 1250

RUN: 7BIO (N2 free) C2H2 C2H4 C2H6 CH4 CO CO2 H2 Sum 1500 1750 2000 2250 2500 2750 3000 3250 3500 3750 4000 4250 4500

time [s]
Fig. 8. Syngas characterization for 47BIO runs in the pyrolysis zone: (a), (b) molar composition and (c) ow rate.

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435 Table 5 Syngas average molar composition (N2 free) in the pyrolysis zone. 3BIO CH4 CO CO2 H2 C2H2 C2H4 C2H6 13.41 51.44 25.32 6.07 0.06 2.02 1.69 4BIO 18.07 36.70 19.31 21.53 0.17 3.30 0.91 5BIO 19.17 34.00 18.09 24.93 0.13 2.88 0.80 6BIO 14.11 30.44 13.74 40.50 0.74 0.46 7BIO 14.34 30.89 13.51 38.01 0.16 2.39 0.70

1433

the results of the cracking action of the pyrolysis process that tends to split large molecules (tar) into smaller ones. 4.4. Gasication model description For the considered conversion zone, the syngas temperature (i.e., the internal temperature) has been assumed as the local conversion temperature used in the equilibrium model (actually, the range of Tinternal temperature variation in the pyrolysis zone is roughly between 400 and 500 C and the peak temperature). The biomass conversion process has been modeled with a thermodynamic chemical equilibrium approach. To this purpose, a code written in Matlab environment [32] has been developed using the Cantera software library (a collection of object-oriented software tools for problems involving chemical kinetics, thermodynamics and transport phenomena [33]). The solver implemented in Cantera is a version of the VillarsCruiseSmith (VCS) algorithm (a well suited method to handle multiphase problems), that nds the composition minimizing the total Gibbs free energy of a mixture [34]. The NASA [35] and the GRI-MECH [36] databases have been used to evaluate the thermodynamic properties of the chemical species considered in the model. The applied procedure for solving the minimization problem is based on the stoichiometric formulation, in which the closed-system constraint is treated by means of the linearly independent stoichiometric Eq. (1) so as to result in an unconstrained minimization problem:

where ni represents the number of moles and li is the chemical potential. The VCS algorithm utilizes this procedure, and it is very useful when dealing with only a few independent stoichiometric equations. Among the set of states which satisfy the element mass balance, VCS algorithm nds the state with the lowest total Gibbs free energy. To estimate the yield of both the gaseous and solid phases, an improved two-phase formulation of the model has been used. As shown in Table 6, 61 chemical species, 60 in the gaseous phase and one in the solid state (i.e., the graphite allotropic form of carbon) are taken in account for the reaction products. The chosen compounds are composed of the C, H, O, N and S elements characterizing the biomass. The calculated yield of solid carbon can be used as an estimate of the actual charcoal residue of the thermal conversion process. For a more detailed description of the model the reader can refer to [8]. 4.5. Model performance Despite the experimental conversion process occurs in non-stationary conditions, the average values of the experimental data have been used to assess the reliability of the proposed model. The comparison of the measurements with the equilibrium model results has been carried out on the residual solid matter (char and ash) yield and on the syngas composition. The char yields are reported versus temperature (Fig. 9), and for the experimental data the representative temperature is the measured peak temperature (used as an estimate of the characteristic conversion temperature of the whole process see also Figs. 7 and 8a and b). Even if there are only few experimental data and the comparison is limited only to a narrow temperature range, the agreement between measured and computed values seems to be satisfying (maximum deviation less than 15% and +5% on the measured value, respectively, for wet and dry biomass data). Furthermore, it is interesting to observe that, as predicted by the equilibrium model, the water in the biomass acts as a gasifying agent, promoting a more efcient conversion process and so reducing the amount of residual solid carbon (compare 4 and 5BIO runs char yields). From the comparison between experimental and computed syngas average molar composition (Fig. 10), it is possible to afrm that in N2 purged runs there is a better agreement than in nonpurged runs. The effect of the carrier gas is to transport quickly the reaction products towards the measurement probe, lowering the system response time and minimizing the extent of secondary reactions. There is a good agreement in the prediction of the carbon monoxide and carbon dioxide production, while the calculated hydrogen and methane concentrations tend, respectively, to overestimate and underestimate the measured values. As previous studies have shown for coal gasication [37,38], the high measured

n n0

R X j1

mj nj

where n0 is the initial composition vector, mj represents the stoichiometric coefcients vector and nj are the so called reaction coordinates. The minimization procedure applied to the G function implies the computation of its partial derivatives with respect to the reaction coordinates (2) and gives the equilibrium condition (3):

@G @nj


T;P;nkj

N X  @G  i1

@ni

T;P;nkj

@ni @nj

 2
nkj

X
i

v ij li 0

Table 6 Chemical species considered by the equilibrium model. Phase Gas Group Inorganic carbon compounds Hydrogen compounds Nitrogen compounds Sulfur compounds Hydrocarbons Other organic compounds Carbon Compounds C(g) CO CO2 H H2 O O2 OH H2O HO2 H2O2 HCO N N2 NH NH2 NH3 NNH NO NO2 N2O HNO CN HCN H2CN HCNN HCNO HOCN HNCO NCO S SO2 SO3 H2S COS CS2 CH CH2 CH3 CH4 C2H C2H2 C2H3 C2H4 C2H5 C2H6 C3H7 C3H8 C6H6 C10H8 C12H10 CH2O CH2OH CH3O CH3OH HCCO CH2CO HCCOH CH2CHO CH3CHO C(s), graphite

Solid

1434

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435

model (wet biomass) measured (wet biomass)

model (dry biomass) measured (dry biomass)

Residual solid mass fraction (%) _

50 40 30 20 10 0 600

700

800

900

compositions is representative only of processes where the residence time is long enough to establish thermodynamic equilibrium conditions. This could be the typical case of some continuous real reactors (e.g., xed or uidized bed type), while for other kinds of process (e.g., fast pyrolysis) the equilibrium conditions will be not achieved. Moreover, the proposed model is still in the development stage since, even if the main reaction products (syngas and char) can already be computed, the assessment of the liquid phase yield is not yet available. The estimation of this third condensable phase (tar) is very important when designing a gasication system. Further research is then currently in progress to evaluate, on the modeling improvement, the feasibility of an advanced model to calculate also the liquid phase and, on the experimental activities, the evaluation of the tar yield and its chemical characterization. References
[1] Chen G, Andries J, Luo Z, Spliethoff H. Biomass pyrolysis/gasication for product gas production: the overall investigation of parametric effects. Energy Convers Manage 2003;44:187584. [2] Jalan RK, Srivastava VK. Studies on pyrolysis of a single biomass cylindrical pellet kinetic and heat transfer effects. Energy Convers Manage 1999;40: 46794. [3] Babu BV, Chaurasia AS. Pyrolysis of biomass: improved models for simultaneous kinetics and transport of heat, mass and momentum. Energy Convers Manage 2004;45:1297327. [4] Pyle DL, Zaror CA. Heat transfer and kinetics in the low temperature pyrolysis of solids. Chem Eng Sci 1984;39:14758. [5] Al-Omari S-AB. Experimental investigation on combustion and heat transfer characteristics in a furnace fueled with unconventional biomass fuels (date stones and palm stalks). Energy Convers Manage 2006;47:77890. [6] Sharma A, Rao RT. Analysis of an annular nned pyrolyser. Energy Convers Manage 1998;39:98597. [7] Batra D, Rao TR. Analysis of an annular nned pyrolyser-II. Energy Convers Manage 2000;41:57383. [8] Baratieri M, Baggio P, Fiori L, Grigiante M. Biomass as an energy source: thermodynamic constraints on the performance of the conversion process. Bioresour Technol 2008;99:706373. [9] Babu BV, Sheth PN. Modeling and simulation of reduction zone of downdraft biomass gasier: effect of char reactivity factor. Energy Convers Manage 2006;47:260211. [10] Gbel B, Henriksen U, Jensen TK, Qvale B, Houbak N. The development of a computer model for a xed gasier and its use for optimization and control. Bioresour Technol 2007;98:204352. [11] Guo B, Li D, Cheng C, L Z, Shen Y. Simulation of biomass gasication with a hybrid neural network. Bioresour Technol 2001;76:7783. [12] Altani CR, Wander PR, Barretto RM. Predictions of the working parameters of a wood waste gasier through an equilibrium model. Energy Convers Manage 2003;44:276377. [13] Fagbenle RL, Oguaka ABC, Olakoyejo OT. A thermodynamic analysis of a biogas-red integrated gasication steam injected gas turbine (BIG/STIG) plant. Appl Therm Eng 2007;27:22205. [14] Ginsburg J, de Lasa HI. Catalytic gasication of biomass in a CREC uidized riser simulator. Int J Chem React Eng 2005;3:A38. <http://www.bepress.com/ ijcre/vol3/A38> [accessed December 2007]. [15] Jarungthammachote S, Dutta A. Equilibrium modeling of gasication: Gibbs free energy minimization approach and its application to spouted bed and spout-uid bed gasiers. Energy Convers Manage 2008;49:134556. [16] Lutz AE, Bradshaw RW, Bromberg L, Rabinovich A. Thermodynamic analysis of hydrogen production by partial oxidation reforming. Int J Hydrogen Energy 2004;29:80916. [17] Schuster G, Lfer G, Weigl K, Hofbaure H. Biomass steam gasication an extensive parametric modeling study. Bioresour Technol 2001;77:719. [18] Zainal ZA, Ali R, Lean CH, Seetharamu KN. Prediction of performance of a downdraft gasier using equilibrium modelling for different biomass materials. Energy Convers Manage 2001;42:1499515. [19] Kuramochi H, Wu W, Kawamoto K. Prediction of the behaviors of H2S and HCl during gasication residual biomass fuels by equilibrium calculation. Fuel 2005;84:37787. [20] Lee DH, Yang H, Yan R, Liang DT. Prediction of gaseous products from biomass pyrolysis through combined kinetic and thermodynamic simulations. Fuel 2007;86:4107. [21] Li X, Grace JR, Watkinson AP, Lim CJ, Ergdenler A. Equilibrium modelling of gasication: a free energy minimization approach and its application to a circulating uidized bed coal gasier. Fuel 2001;80:195207. [22] Mahishi MR, Goswami DY. Thermodynamic optimization of biomass gasier for hydrogen production. Int J Hydrogen Energy 2007;32:383140. [23] Prins MJ, Ptasinski KJ, Janssen FJJG. From coal to biomass gasication: comparison of thermodynamic efciency. Energy 2007;32:124859.

Temperature [C]
Fig. 9. Residual solid matter (char and ash) yield versus temperature: comparison between measured and equilibrium computed data.

Average molar concentration (%)

4BIO

6BIO

7BIO

Model

60 50 40 30 20 10 0
CH 4 CO CO2 H2

Fig. 10. Syngas average molar composition (N2 free) in the 600800 C temperature range: comparison between measured and equilibrium computed data (6-7BIO are N2 purged runs).

concentrations of methane can result from incomplete conversion of pyrolysis products. This probably applies also to the biomass conversion process (i.e., measured methane concentration cannot be explained on a purely thermodynamic basis). The deviation can result from one or more non-equilibrium factors, such as incomplete cracking of the pyrolysis products, temperature gradients in the gasier, or catalysis of the methanation reaction taking place in the piping system towards the measurement probe or in the low temperature zones of the reactor bed in contact with the shell; in fact, many steel alloys (including Incoloy, used here) contains nickel, that can have a selective catalytic action on the methane formation kinetics. 5. Conclusions The overall agreement between model predictions and experimental results is reasonably satisfactory; in particular the residual solid yield (char) obtained experimentally at 800 C is very close (in the range 5% to +15%) to the value obtained by the simulation. Considering the gas phase, the agreement is quite good also for CO and CO2, it is fair for H2, while it is poor for CH4. The simulation code, based on a thermochemical equilibrium approach, can then be considered as an useful tool in the design of biomass thermal conversion processes lasting long enough to achieve (or to approach) equilibrium conditions. However, despite the satisfying results presented here, this study must be considered as a preliminary one for several reasons. First of all, as already mentioned, the computed reaction product

P. Baggio et al. / Energy Conversion and Management 50 (2009) 14261435 [24] Baggio P, Baratieri M, Fiori L, Grigiante M, Ragazzi M. Biomass gasication experimental investigation: literature review and design of a bench-scale reactor. In: Proceedings 14th European biomass conference; 2005. p. 6925. [25] Comsol multiphysics 3.3; 2008. <http://www.comsol.com>. [26] Deru M. A model for ground-coupled heat and moisture transfer from buildings. National Renewable Energy Laboratory. Technical report NREL/TP550-33954; 2003. [27] Baggio P, Baratieri M, Grigiante M. Indagine teorico-sperimentale sul ruolo della resistenza di contatto nei reattori per biomasse:risultati preliminary. In: Proceedings XII AIPT conference; 2006. p. 10514. [28] Holman JP. Heat transfer. 9th ed. Singapore: McGraw-Hill; 1992. [29] Phyllis. Database for biomass and waste. The Netherlands: ECN; 2003. <http:// www.ecn.nl/phyllis>. [30] Reisinger K, Haslinger C, Herger M, Hofbauer H. BIOBIB a database for biofuels. In: Thermie conference: renewable energy databases; 1996. [31] Simmons WW, Ragland KW. Burning rate of millimeter sized wood particles in a furnace. Combust Sci Technol 1986;46:115.

1435

[32] Matlab 7.0.4.365 (R14). The MathWorks, Inc.; 2005. <http://www.mathworks. com>. [33] Goodwin D. Cantera: object oriented software for reacting ows. California Institute for Technology (Caltech); 2005. <www.cantera.org>. [34] Smith WR, Missen RW. Chemical reaction equilibrium analysis: theory and algorithm. New York: Wiley Interscience; 1982. [35] McBride BJ, Gordon S, Reno MA. Coefcients for calculating thermodynamic and transport properties of individual species. NASA report TM-4513; 1993. [36] GRI-Mech thermodynamic values; 2000. <www.me.berkeley.edu/gri_mech>. [37] von Fredersdorff CG, Elliott MA. Coal gasication. In: Lowry HH, editor. Chemistry of coal utilization. New York: Wiley; 1963. [38] Coates RL, Chen CL, Pope BJ. Coal devolatilization in a low pressure low residence time entrained ow reactor. In: Massey LG, editor. Advances in chemistry series, vol. 131. Washington (DC): American Chemical Society; 1974. p. 92107.

Вам также может понравиться