Modeling Circulating Fluidized Bed Biomass Gasifiers

Fuel Processing Technology 86 (2005) 1021 – 1053 www.elsevier.com/locate/fuproc

Modeling circulating fluidized bed biomass gasifiers. A pseudo-rigorous model for stationary state Jose Corella*, Alvaro Sanz Department of Chemical Engineering, University "Complutense" of Madrid, 28040-Madrid, Spain Received 11 June 2004; received in revised form 19 November 2004; accepted 19 November 2004

Abstract A 1-dimensional model for an atmospheric circulating fluidized bed biomass gasifier (CFBBG) under stationary state is presented in this paper. The model is based on the kinetic equations for the reaction network solved together with mass and heat balances and with several hydrodynamic considerations. Kinetics used include both our own kinetic data and published equations with some corrective factors. The reaction network used involves twelve different reactions. A sub-model for the tar generation-elimination in the CFBBG is included in the whole model. The model has an academic structure, but several assumptions were made because of lack of accurate data in some areas. The overall model has some empirical aspects and can therefore be considered as semirigorous. Hydrodynamics in the model were checked with a survey carried out worldwide among the existing pilot and commercial CFBBGs. The axial profiles of concentration of ten different species (H2, CO, CO2, tar, char, . . .) and temperature can be calculated with this model which was conceived to optimize both design and operation of CFBBGs. D 2004 Elsevier B.V. All rights reserved. Keywords: Reaction engineering; Energy; Fluidization; Mathematical modeling; Biomass gasification; Tar

1. Introduction Thermochemical gasification of biomass generates a useful gas (a mixture of H2, CO, CO2, CH4, small hydrocarbons,. . .) using a gasifying agent, usually air, also being the only * Corresponding author. Tel./fax: +34 91 394 4164. E-mail address: [email protected] (J. Corella). 0378-3820/$ - see front matter D 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.fuproc.2004.11.013

1022

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

gasifying agent considered in this paper. Commercial biomass gasifiers are nowadays facing serious problems, especially their economical feasibility: the energy from the produced or gasification gas has to be competitive with natural gas which is abundant and cheap. One biomass gasifier which can have some future is the atmospheric circulating fluidized bed gasifier (CFBBG) with a few commercial units operating in the Netherlands, Austria, Sweden, Finland and Germany [1]. The produced gas is bdirtyQ (contains some tar) and usually it is directly fired in an adjoining combustor. Nevertheless, the most promising future applications of the thermochemical gasification of biomass will require at least both (i) generating a bquite cleanQ gas (to be fired in gas engines or turbines, for example), and (ii) having no troubles in the gasifier (with a continuous non-stop operation). These two facts require that the gasifier has an optimized design and operation which does not occur very often. Although most gasifier manufacturers may claim that their respective gasifiers are already optimized, a good model might help to optimize both gasifier design and its operation. For this reason this work is devoted to develop a model bas good as possibleQ for CFBBGs. There is huge amount of papers on hydrodynamics of CFBs which are very useful for modeling CFBBGs, as the book edited by Grace et al. [2]. Concerning the modeling of fluidized bed gasifiers, the field of coal is well ahead of that of biomass. There are a lot of works (i.e., [3–16]) on modeling CFB coal combustors and gasifiers which contain very valuable information and help to model CFBBGs. They were used in different parts of the modeling of CFBBGs presented here. Nevertheless, it is well known and accepted that thermochemical processing of biomass has some important differences with respect to the processing of coal. Two of them are important for the CFBBG modeling: (1) biomass is much more reactive than coals, it pyrolyzes very quickly and its ash content is usually very low. For these and other reasons, another solid, sometimes called fluidizing, has to be used in the gasifier. It is usually silica sand. Besides, an additive (dolomite, limestone, olivine,. . .) is also used in the gasifier for tar cracking, alkali capture, etc., . . . The particle size of these two solids (silica sand and additive) is an important variable in the process. (2) Biomass gasification below 1000 8C always produces important amounts of tar whose content in the flue gas has to be estimated with a good model for it to be it useful. Literature on biomass is therefore more important for CFBBG modeling than coal, but it is less abundant and the existing approaches are not yet as rigorous and developed as those for coal. The literature on modeling biomass gasification in fluidized bed is much more related to this work than that of coal but it is so scarce that it can be fully cited. In modeling biomass gasification (with air) in bubbling fluidized beds (BFBBG), Belleville and Capart [17] developed an empirical but quite interesting model which was successfully applied to the biomass gasifier of Creusot Loire in Clamecy (France). Fan and Walawender [18] and Van den Aarsen [19] reported two of the pioneering models, which are well known today; Corella et al. [20] modeled some non-stationary states of BFBBGs; Bilodeau et al. [21] considered axial variations of temperature and concentration and applied their results to a 50 kg/h pilot gasifier; Jiang and Morey [22,23] introduced new concepts in this modeling, especially related to the freeboard and the fuel feed rate; Hamel and Krumm [24] provided interesting axial profiles of temperature, although their work was mainly focussed on gasification of coal and did not give many details of their model; Mansaray et al. [25,26] presented two models using the ASPEN PLUS process simulator

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1023

but referred to their specific bdual-distributor-typeQ BFBBG, limiting their usefulness. Concerning modeling of BFBBGs with steam, Corella et al. [27] presented a model based on the Kato and Wen model for fluidized beds. The main contribution of their model is that it identifies, amongst the complex gasification reaction network, the four main (for modeling purposes) chemical reactions. That model fits then the experimental gasification data with only four parameters (chemical kinetic constants). If there are only a few papers on modeling BFBBGs, literature on modeling CFBBGs is even more scarce. The absence of details in the literature on CFBBG modeling is not surprising because, besides the very few existing commercial CFBBGs, it was mainly written for marketing purposes. Models claimed to exist can neither be checked nor used by readers because most of the key information is missing. Most of the manufacturers and users of CFB biomass gasifiers at commercial and pilot scales claim to operate a good own model but not much more is said about it. Lurgi, ECN, TPS, IGT, etc.,. . . for instance, reported to have their own models but they did not provide details of such models. The UMSICHT Institute in Oberhausen, Germany, [28] and University of Siegen in Germany too [29,30] provided some data from their own models for CFBBGs, but what is considered to be the core of such models is again missing from their papers. Kersen et al. [31] recently provided a model for the pilot CFBBG at ECN but, as the same authors recognized, it is an interpretation model which cannot be used for design and scale-up purposes. Finally, the model for CFBBGs recently published by Liu and Gibbs [32], which were also our partners on the same project that financed the work presented here, is similar to the model presented in this paper, but theirs is manly addressed to NH3 and HCN emissions. A detailed and advanced model for CFBBGs does not exist in the open literature, up to now. Such a model would be useful to optimize both the design and operation of a CFBBG. The main aim of this paper is to develop a valuable model for CFBBGs and to give a description of its main parts or sub-models. Results from the model are being presented in the next paper [33].

2. Basis of the model 2.1. Topology or description of the considered CFBBG Atmospheric CFB biomass gasification nowadays is not yet a well established technology. There are very few commercial gasifiers worldwide and each gasifier manufacturer (TPS AB, Lurgi, Foster Wheeler Energy, Austrian Energy,. . .) has its own design. So far there is no consensus on the detailed design of a CFBBG. The selection of a standard scheme for a CFBBG was therefore not an easy task. After a detailed analysis of the existing technology, including the small CFBBG owned and handled by the authors, the selected scheme for a CFBBG is shown in Fig. 1. Some aspects of it are the following: i) Although some CFB coal gasifiers do not have a dense bed at the bottom, the CFBBG considered here has a bottom bed, which is quite important in biomass

1024

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

gasification. It increases the heating transfer rate to the particles of biomass, and decreases the tar yield [34]. Among the CFBBGs analysed by the authors, the ones with a stationary bottom bed and a biomass feeding into it generate less tar than if there is not such bottom bed or if the biomass is fed above it. An additive (D), such as calcined dolomite (OCad OMg), limestone or related materials, is also considered to be as stationary, permanent or fluidizing material, together with the silica sand (S). This additive decreases the tar yield at the gasifier exit and prevents bed agglomerations [35–37].

Gas

ε = 0.99

2nd Air

Mixing 2nd Air

Dilute Zone Hdz

H4 HT H3

ε = 0.90

H2nd

ε = 0.77

Biomass Feed

Pyrolysis

H2 H1

1st Air Fig. 1. Scheme for the selected CFBBG to be modeled.

Transition Zone Htz Bottom Zone Hbt

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1025

ii) Biomass feeding point is located at the bottom of the gasifier, the biomass is fed directly into the bottom bed. iii) The stationary bed and main circulating material will be a mixture of silica sand and 20–30 wt.% of calcined dolomite [35]. iv) Air flow will be handled as the equivalence ratio, ER [total air flow fed to the gasifier/stoichiometric (for total combustion)air flow]. The total air fed (to the gasifier) is split into the primary and secondary air flows shown in Fig. 1. The secondary air flow is located above the biomass feeding point; together with the biomass feeding there could also enter some additional sealing air which would be considered as secondary air. The height of such 2nd air flow inlet is called H2nd, Fig. 1. This secondary air flow is usually used to increase the temperature in the upper part of the gasifier to decrease, by thermal reactions such as cracking, the tar content in the gas flows. As a typical or reference case, the 2nd air flow is considered to be the 20% of the ER value. 2.2. Scales studied for the CFBBG Not only commercial CFBBGs are of interest. There are other CFBBG of smaller scale, pilot and demo sizes, which have also been considered and covered by the modeling. Three scales were therefore considered. They are indicated in Table 1, together with their representative values. The gasifiers corresponding to the three scales were considered to have the same area specific throughput (=1740 kg biomass fed, as received/h m2 of gasifier cross-sectional area in the dilute zone), weight hourly space velocity for the biomass (WHSV) [=1.9 (kg biomass fed, as received)/h]/kg solids inventory in the CFBBG] and total height for the riser (=14.8 m). These values have been selected from a careful analysis and survey of the existing CFBBGs worldwide. The inner diameters in the dilute zone of the CFBBGs for these three scales are 7.6 cm, 0.85 m, and 3.3 m, respectively. 2.3. Operation intervals considered After a careful analysis of biomass gasification itself and of existing bed biomass gasifiers, the intervals for the main operation variables were selected. These are shown in Table 2. Most of the existing fluidized biomass gasifiers operate, or should be operated, between the intervals or limits shown in Table 2.

Table 1 Scales studied for the CFBBG Scale

Pilot Demo Commercial

˙B m (kg a.r./h)

Basis/Reference Throughput (kg a.r./h m2)

WHSV (h
1)

W sand+dolomite (kg)

8.0 1000 15000

1740 1740 1740

1.9 1.9 1.9

4.2 520 7760

i.d. (m)

HT (m)

0.076 0.85 3.3

14.8 14.8 14.8

1026

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

Table 2 Intervals considered for the main operation variables Parameters

Intervals considered

ER: Biomass moisture (wt.%): Biomass flow rate (kg/h): WHSV (h
1): Throughput (kg biomass a.r./h m2): 2nd air flow (%): 2nd air inlet height (m): u o (m/s): T bottom bed (8C):

0.20–0.45 5–30 2–20,000 1.0–3.0 1000–7000 0–40 5–10 2–10 750–980

2.4. Inputs and outputs to/from the modeling The inputs for the model are indicated in Table 3. They include the biomass to be gasified, the air (gasifying agent), the gasifier design and the biomass feeding flow which can be expressed in several ways: as mass flow rate (kg/h), as weight hourly space velocity [WHSV in (kg biomass/h)/kg solids inventory in the CFBBG] or as throughput (kg biomass/h m2 cross-sectional area in the dilute zone). Two important input data concerning the biomass feedstock, particle size distribution and its alkali (K+Na) content (sintering or agglomeration problems), are not directly handled by the model in its present state of development. For now, these two parameters will have to be handled in parallel with the model shown here. Direct outputs from the model (the only ones presented in this paper) are the gas composition, the gas yield, the tar content in the produced gas, the carbon (in biomass) conversion to gas, and the temperatures both at the gasifier exit and inside the gasifier (axial profiles and bottom bed temperatures). Experts in biomass gasification should be able to draw some more and important (indirect) outputs from the above direct outputs. For instance, the model, if understood well, enables an optimized design of a CFBBG, possible revampings and improvements of existing CFBBGs, detection of limits of operation and zones in which the gasifier operates badly. All these outputs can appear quite ambitious but in fact they have already been applied by the authors to some gasifiers with positive results and, therefore, these outputs are realistic for the authors. 2.5. Strategy in this modeling CFBBGs are quite complex reactors and their modeling is difficult. Besides the need to know biomass gasification technology in depth, there is as yet not enough accurate data for some aspects. More needs to be known on the kinetics of some reactions under gasification conditions and on radial and axial profiles in the whole gasifier of the char and of the charred biomass, which are required to develop a good model [38]. During the modeling work the authors continuously experienced uncertainties and lack of accurate information. Besides, a model must be checked with existing CFBBGs but this checking was very difficult in our case. The manufactures of the few existing pilot and commercial CFBBGs

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1027

Table 3 Inputs and outputs from the model Inputs. variables considered Biomass

Air

Gasifier

Biomass feeding

Chemical analysis (C, H, O, N) Moisture Ash content (Particle size. Content in potassium: constraints in operation temperatures by possible agglomerations) Preheating temperature Volumetric or mass flow rate Equivalence ratio primary air secondary air flow Moisture Topology. Size, shape and main dimensions bBedQ composition (solids and particle sizes) inventory Heights of feeding: 1st air 2nd air biomass Mass flow rate (kg/h) WHSV (h
1), [mass flow rate/mass of solids (inventory) in the gasifier] Throughput

Outputs Direct Gas

Carbon Temperature

Indirect

Composition (H2, CO, CO2, CH4, C2Hn , H2O, O2 contents) Heating value (LHV, MJ/Nm3, dry basis) Yield (Nm3/kg biomass daf) Quality: tar content (g/Nm3) Exit temperature Conversion (%) Content in exit fly ash Longitudinal profiles Exit Bottom bbedQ Optimized design of a CFBBG Limits of operation, zones of bad-functioning Maximize throughput Possible rewampings of the CFBBGs Optimized MW/m2

did not publish many details about their gasifiers, and a private worldwide survey carried out by these authors did not provide enough data as was required. Due to the complexity of handling all aspects required in a model for a CFBBG, Corella and co-workers at the Universities of Saragossa and Madrid (Spain), after twelve years of work on this modeling, have developed two different models for CFBBGs. Each one uses a different approach. In order to develop a model, given that till now not enough accurate basic or scientific information exists, several assumptions were made. Depending on where such assumptions are introduced or localized, a different model can be generated. In this paper only one model is presented. It is 1-dimensional and for CFBBGs under steady state. The 2nd model mentioned above is based on an empiric partitioning of the O2 (air) fed to

1028

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

the gasifier, it does not consider axial profiles inside the CFBBG and is for non-stationary states. The 1-dimensional model presented here might seem academic because it is based on the Chemical Reaction Engineering (CRE) principles: chemical kinetics together with mass, heat and momentum balances. A rigorous academic approach would contain more than 20 parameters, some of them being absolutely unknown today [38]. Such unknown parameters would be some kinetic parameters concerning in-bed tar appearance and elimination and the ones concerning heterogeneities (plumes and radial gradients) existing in the biomass and 2nd air flow feeding zones (i.e., [39]). Therefore, several assumptions had to be made to solve and handle the initially very complex rigorous academic approach. Making these experience based assumptions, the model became semi-rigorous. The degree of empirism in it will decrease, hopefully, in the future when new and accurate information (for CFB biomass gasifiers) is available. The equations presented and solved here will basically be a set of kinetic equations for the complex reaction network together with the mass balances for all the species or reactants considered in such network. To calculate the values of all kinetic constants, axial profiles of temperature are required. A lot of hydrodynamic considerations and details are also required. These temperature profiles and hydrodynamic considerations will be handled as sub-models and/or sub-routines in parallel with the main program. A model without extensive good verification means nothing for these authors. Although some experimental verification is usually not enough to demonstrate the validity and usefulness of a model, it is absolutely required. For this reason, a lot of effort was made to verify the model presented here. Such validation was carried out in two different ways: (1) a worldwide survey with all the known commercial gasifiers manufacturers and owners [AMERGAS, LURGI, TPS, ECN, UMSICHT, LAHTI, FOSTER-WHEELER E. Oy, ¨ RNAMO (Sydkraft), CLAMECY Plant,. . .]. Some of them do not exist nowadays, for VA example, the gasification plant in Clamecy (France), but some useful information from such plants were obtained years ago and were used in this work. Some information obtained from some manufacturers and owners had a confidential character and cannot be explicitly shown here. Nevertheless, it was used in several parts and steps of the modeling presented here. Some data have also been obtained during the IEA Biomass Gasification task meetings. (2) Some tests were carried out on small pilot plant scale to check this model. For that objective, a BFB biomass gasifier at University Complutense of Madrid (UCM) was fully modified by increasing its height and the biomass and air flow rates and connecting a standpipe and an L-valve, thus generating a CFB. This CFB biomass gasifier was presented in [37] and is shown in Fig. 2. Its main dimensions are (a) bottom zone: 70 mm internal diameter (i.d.) and 1.2 m height; (b) a 2nd zone of 120 mm i.d. and 1.4 m height; (c) upper zone: 150 mm i.d. and 2.2 m height. Temperatures were measured at different heights with different thermocouples. Some tests were designed and carried out in this CFBBG to check the model. This verification continues. 2.6. Species considered in the gasification reaction network i) Concerning biomass, pine wood was selected as reference. The composition (dry, ash free) was 50.0 wt.% C, 5.8 wt.% H and 44 wt.% O. Its weight formula was C4.2H5.8O2.8.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1029

ii) Gases. Air (in primary and secondary flows) and gases formed in the pyrolysis and gasification reactions: H2, CO, CO2, CH4, C2H4 and H2O. H2O comes from biomass and air mixtures and as a product in some reactions.

1. Hopper. 2. Feeding system. 3. Primary air inlet. 4. Preheating zone. 5. Secondary air 6. Gasifier 7. Additives inlet

7

6

5

1

H2O 2

4

PRIMARY AIR

3

Fig. 2. New small CFBBG pilot plant at UCM used to check the model.

1030

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

iii) Tar. Although it is in gas phase and could be considered as one more gas, it will be studied separately due to its importance in biomass gasification. The first fast pyrolysis or devolatization generates a tar which will be called tar1. In the bottom bed tar1 very quickly undergoes a lot of thermal degradation reactions generating a tar called here tar2. It would be the one detected if the CFB reactor was operated as pyrolizer, with an inert gas. In a CFB gasifier this tar2 reacts, to some extent, with the primary air generating a tar called btargasifQ. If, besides, there is a catalyst like calcined dolomite (as it is the case considered here) in the bottom bed, this btargasifQ undergoes some steam and dry (CO2) reforming reactions which decreases its total amount and changes its composition. This new tar generated with in-bed dolomite will be called btardolQ. The 2nd air flow in the dilute zone additionally converts and transforms this tar (the btardolQ coming from the bottom bed) to another tar which will be called btar2ndairQ. Not only the amount of tar decreases but also its composition changes by the above reactions [40]. To show the difference between the tars (tar2, targasif, tardol and tar2ndair), the corresponding tar yields obtained in similar FB biomass gasifiers using the same feedstock, pine wood chips, are shown in Fig. 3. This Fig. 3 clearly shows how, at a given gasification temperature, the tar yield depends very much on the existence of in-bed dolomite and on a 2nd air flow. Concerning tar composition, the 6-lump model for tar evolution (generationelimination) developed by Corella et al. [40–42] is now being applied to study the axial profiles of the tar species. These axial profiles can be handled by a sub-routine with 9 kinetic constants. This sub-routine, not used in this paper, may be added or incorporated in the model presented here to predict or calculate the axial profiles (in the riser) of the species present in the tar. The different tars existing in a CFBBG will be represented in this model by two different species only: – tar2, obtained from the devolatization+thermal reactions at the bottom bed (see Eq. (1)). Its general formula is Cx Hy Oz , with x=1 and y, z to be determined from mass balances ( y/x for tar2 is less than 1 in the range of temperature considered, according to Van den Aarsen [19]). – tar4, generated by in-bed catalytic steam (and dry) reforming reactions (Eq. (9)). Its general formula is CxU HyU with xU=yU. The btotal or overall tar contentQ at the exit of the CFBBG will be the addition of, first, the tar2 generated in the pyrolysis step (Eq. (1)) minus the tar2 reacted by reactions (2) and (9), and, second, the tar4 generated by Eq. (9) minus the one reacted by reaction (2b). iv) Char. Two different compositions will be considered for the char: – char2, generated in the devolatization+in-bed thermal reactions (Eq. (1)). Its general formula is Cx V Hy V Oz V with xV=1 and yV, zV to be determined from mass balances [(H/C)char2b(H/C)tar2b1, in the range of temperatures considered].

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1031

160

`

Gonzalez-Saiz [43] Narvaez et al. [36] Gil et al. [35] Lammers et al. [46], Kurkela [73]

Tar yield (g/kg biomass, daf)

140 120 100 80

tar2

20

targasif

10

tardol tar2ndair

0 700

750

800

850

Bed temperature (°C) Fig. 3. Tar yields after pyrolysis (tar2) and after atmospheric gasification with air (ER=0.24) without (targasif) and with (tardol) in-bed dolomite, without and with (tar2ndair) secondary air flow.

– char3, generated in the gasification of char2 with CO2, reaction (11). Its general formula was determined from analysis of this char obtained during the experiments at UCM. It can be noticed that (H/C)char3N(H/C)char2. The bchar concentrationQ inside and at the exit of the CFBBG will be the addition of two contributions: first, the generated one by reaction (11) and the char3 unreacted in reaction (3b), and second, the char2 unreacted in reactions (3) and (10). The compositions of tar2 and char2, according to Van den Aarsen [35], depends on the gasifier temperature, Table 4, and on the type and composition of the biomass used. The chemical compositions of tar2, tar4, char2, char3 were calculated for a gasification temperature of 850 8C, for a given biomass (pine wood was used as reference), and for the Htar2/Hchar2 and Otar2/Ochar2 ratios (Ctar2/Cchar2 ratio is 1) indicated in Table 4. Table 4 Composition of tar2 and char2, according to Van den Aarsen [19] Pyrolysis temperature (8C)

tar2

char2

715 815 915 Averaged

CH0.81O0.20 CH0.75O0.13 CH1.00O0.17 CH0.85O0.17

CH0.25O0.14 CH0.21O0.12 CH0.14O0.14 CH0.20O0.13

1032

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

Table 5 Calculated compositions for tar2, tar4, char2 and char3 (T=850 8C) Specie

Formula

H/C

Biomass tar2 tar4 char2 char3

C4.2H5.8O2.8 CH0.85O0. . .17 C0.54H0.54 CH0.20O0.13 C0.30H0.15O0.46

1.40 0.85 1.0 0.20 0.50

Tar2, tar4, char2 and char3 compositions, calculated at 850 8C as reference, are shown in Table 5.

3. CFBBG modeling 3.1. Reaction network considered The most suitable reaction network for biomass gasification with air in a CFB was previously discussed by Corella and Toledo [38]. Such a network contains all the reactions which were considered relevant and it is shown below (reactions (1)–(12)). Axial profiles of concentration for all the species considered in this network will be further obtained from the model. 3.1.1. Reaction network considered for the zone between the biomass feeding point and the 2nd air inlet (zone which covers the bottom zone, the transition zone and part of the dilute zone, Fig. 1) Fast Pyrolysis:

(1) Oxidation with the primary air of the products formed in the pyrolysis step: tar2½CH0:85 O0:17  þ q2  1st O2 ! e2  CO þ g2  CO2 þ s2  H2 O

ð2Þ

char2½CH0:20 O0:13  þ q3  1st O2 ! e3  CO þ g3  CO2 þ s3  H2 O

ð3Þ

H2 þ O1st O2 ! H2 O

ð4Þ

CO þ O1st O2 ! CO2

ð5Þ

CH4 þ 1:51st O2 ! CO þ 2  H2 O

ð6Þ

C2 H4 þ 31st O2 ! 2CO2 þ 2H2 O

ð7Þ

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1033

Steam reforming of methane: CH4 þ H2 OYCO þ 3  H2

ð8Þ

Tar and char reforming reactions: (9) char2½CH0:20 O0:13  þ s10  H2 O ! e10 CO þ d10 H2

ð10Þ

char2½CH0:20 O0:13  þ g11  CO2 ! e11 CO þ u11 char3bCxx Hyy Ozz c

ð11Þ

3.1.2. Reaction network considered for the zone between the 2nd air inlet and the exit to cyclone (part of the dilute zone, Fig. 1) Besides the reactions given by Eqs. (2)–(11), Oxidation with the secondary air of products coming from first zone of reaction: tar4tCx00  Hy00 b þ q2b  2ndO2 ! e2b  CO þ g2b  CO2 þ s2b  H2 O

ð2bÞ

char3tCxx Hyy Ozz b þ q3b  2nd O2 ! e3b  CO þ g3b  CO2 þ s3b  H2 O

ð3bÞ

Shift reaction: (12) The non fixed stoichiometric coefficients in these reactions were calculated by mass balances for each of the components (C, H, O) in each chemical reaction. These coefficients, for a temperature of 850 8C, are shown in Table 6. 3.2. Kinetic equations used for the chemical reactions involved in the reaction network The set of kinetic equations for the reaction network given by Eqs. (1)–(12), used to calculate the product distribution, is shown in Table 7. These equations originate both from our own findings/research and from some published papers. Table 6 Stoichiometric coefficients for T=850 8C Eq.

Stoichiometric coefficients

(1) (2) (2b) (3) (3b) (9) (10) (11)

b 1=0.87; c 1=0.45; d 1=1.3; e 1=1.5; g 1=0.51, h 1=0.47; p 1=0.18 q 2=0.68; s 2=0.31; e 2=0.75; g 2=0.25 q 2b=0.60; s 2b=0.26; e 2b=0.093; g 2b=0.42 q 3=0.58; s 3=0.073; e 3=0.75; g 3=0.25 q 3b=0.032; s 3b=0 ; e 3b=0.037; g 3b=0.17 d 9=0.13; e 9=0.35; s 9=0.15 v 9=1.2 d 10=0.45; e 10=0.54; s 10=0.38 g 11=1; e 11=1.7; u 11=1

1034

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

Table 7 Reaction rates for the reaction network Reaction Reaction rate (mol/m3 s) (1)

y i=m o,i+m 1,i ST+m 2,iST 2

(2)

r2=k2SCtar2SCO2

(2b)

r2b=k2S(ftar4/ftar2)SCtar4SCO2

(3)

r3=k3ScvSUsSCchar2SCO2 0.53 or r3=k3VSpO S(1-Xchar)0.49 2

(3b)

r3b=k3ScvSUsSCchar3SCO2

Parameters

k2=ftar2S(1-cv)1.58S 1010S exp(-24200/T) The same of (2) with different value for the kinetic constant k3V=5.3S105Sexp(-15000/T)

The same of (3) with different value for the kinetic constant

(2), (3), e2 e3 e2b e3b (2b), g2 ¼ g3 ¼ g2b ¼ g3b (3b) ¼ 4:30expð
3390=T Þ

(4) (5) (6)

2 r 4=k4SCH SCO2/CCO 2 r 5=k5SCCO SCO2 -0.5 1.5 r 6=k6SyCH SyO2 4

k4=3.09S1011Sexp(-12000/T) k4=8.83S1011Sexp(-12000/T) k6=7.0S1011Sexp(-30200/T)ST

(7)

r 7=k7SCC0.72H4 SCO0.82

(8)

r8=k8SCCH4SCH2O

k7=fC2H4S(1-cv)1.58S1010S exp(-24200/T) k8=3S1005exp(-15000/T)

(9) (10)

0.25 1.75 r9=k9SCtar2 SCH2O r10=k10SCchar2SCH2O

k9=f9S70.0Sexp(-2000/T) k10=2.0S105Sexp(-6000/T)

(11)

r11=SSr11 W

r11 V =7.2SCCO0.83 exp(-20000/T) 2

(12)

  CCO2 CH2 a) T N 1123 8C (equilibrium) r12 ¼ k12 S CCO SCH2 O
S k12 KW KW ¼ k12V CH2 SCCO2 KW ¼ CCO SCH2 O ¼ 0:0027expð
3960=T Þ

Reference Jiang and Morey [22,23], Gonzalez-Saiz [43] Zanzi et al. [66], Dai et al. [67], Di Blasi et al. [68,69] Dryer and Glassman [44,45] Lammers et al. [46] Dryer and Glassman [44,45] Lammers et al. [46] Janse et al. [47] Fushimi et al. [50] Kulasekaran et al. [71] Janse et al. [47] Fushimi et al. [50] Kulasekaran et al. [71] Belleville and Capart [17]

Kulasekaran et al. [62] Hayhurst and Parmar [70] Cozzani et al. [72] Kim et al. [11] Kim et al. [11] Srinivasan et al. [7] De Souza-Santos [63] Dryer and Glassman [44,45] Philippek et al. [64] The´rien et al. [48] Liu and Gibbs [32], from Fletcher et al. [49] Gonzalez–Saiz [43] Gonzalez–Saiz [43] Fushimi et al. [50] Van den Aarsen [19] Van den Aarsen et al. [51] Tang et al. [65] Gonzalez–Saiz [43] Xu and Froment [52] Simell et al. [53]

b) Tb1123 8C (no-equilibrium) 12 KW ¼ kk12V ¼ 520expð
7230=T Þ k12 ¼ 106 expð
6370=T Þ

Some comments on these kinetic equations: 1st) For reaction number 1 (fast pyrolysis in fluidized bed) a product distribution, instead of a btypicalQ kinetic equation, was preferred because such product

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1035

distribution was obtained with the same type of biomass as the one considered in this paper (small pine wood chips) and in a fluidized bed working under experimental conditions similar to those used in a CFBBG. The values for the parameters m 0, m 1, and m 2 appearing in Table 7 are given in Table 8 for all species. The units for y i , in equation for reaction 1, when calculating tar yield and yield to char were [kg/kg daf] and when calculating H2, CO, CO2, CH4 and C2H4 contents were [mol i/total mol, dry basis]. The authors are confident of the validity of this equation and those m-values for this application. Nevertheless, it has to be pointed out how this product distribution may be quite different for some other types of biomass, because the word bbiomassQ covers a big spectrum of fuels which can be very different between themselves. 2nd) The selection of the rate expressions for reactions (2)–(12) was made after a deep analysis of the abundant bibliography on most of these reactions. Their respective reference or origin are given in Table 7. Since for some reactions there were different possible kinetic equations in the literature and since there was no discrimination method among those rival kinetic models, different references are given in Table 7 for some reactions. When there were several possible kinetic equations for a given reaction, the easiest or simplest one (1st order, for instance) was preferred although several other kinetic equations could also be used for that reaction. 3rd) Since some reaction rates originally had different units (the ones provided by the corresponding authors) an important effort was made to adopt the same units in all kinetic expressions. For this propose, a change of units which takes into account variables such as pressure and temperature in the riser, voidage, density or dolomite content in the considered zone, . . . had to be made in several kinetic equations. 4th) Since most of the kinetic equations shown in Table 7 were obtained under gas atmospheres and environment, such as the existence of calcined dolomite (OCad OMg) which may have some catalytic effects for several reactions, different to the ones existing in a CFB biomass gasifier, most of the rate equations shown in

Table 8 Values of parameters m 0, m 1, and m 2 for the fast pyrolysis of small pine wood chips in a bubbling fluidized bed (from Gonzalez-Saiz [43] and Jiang and Morey [22,23]) i (species)

m0

H2 CO CO2 CH4 C2H4 N2 Tar Char H2O

0.255
4.47 10
4 0.255 5.44 10
4 2.14
3.53 10
3
0.45 1.12 10
3
1.32 2.51 10
3 0.118
1.99 10
4 0.382
2.16 10
4 2.09
3.47 10
3 the corresponding to the initial biomass moisture

m1

m2 4.54 10
7
3.88 10
7 1.55 10
6
5.47 10
7
1.15 10
6 8.64 10
8 0 1.48 10
6

1036

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

Table 9 Corrective factors used in the kinetic equations Corrective factors

Intervals considered

f2 f 2b f3 f4 f5 f6 f7 f8 f9 f 10 f 11 f 12

0.6–1.7 1 1.6 0.5 0.1–0.9 0.1–1 1 2 1.6 0.4–0.7 0.2–2 0.3–0.7

Table 7 have to be understood as approximate, not exact ones. For this reason, and to fit some experimental data from existing commercial CFBBGs, some corrective factors ( f n ) were introduced in the kinetic equations above. These corrective factors are shown in Table 9. The value of 1 in Table 9 means that the corresponding kinetic equation remains without modification. 3.2.1. Overall volumetric rate equations Each i-th species (H2, CO, . . . , tar2, char2, . . . ) existing in the reaction network indicated by Eqs. (1)–(12) appears in several reactions, sometimes as reactant, sometimes as product. The overall or net rate of appearance (indicated by R i ) of the i-th species will be the badequateQ addition, by the well known rules of Chemical Reaction Engineering, of the rates for all the reactions in which such species appears. The two main zones considered in the CFBBG (till the 2nd air feeding point and from this point to the exit) were considered separately. Some examples of the calculation of these overall reactions rate, for the zone between the biomass feeding point to 2nd air inlet, are Rtar2 ¼ eq:1
r2
r9

ð13Þ

Rchar2 ¼ eq:1
r3
r9
r10
r11

ð14Þ

RH2 ¼ eq:1 þ d8  r8 þ d9  r9 þ d10  r10
r4 :

ð15Þ

RCO ¼ eq:1 þ e2  r2 þ e3  r3 þ e6  r6 þ e8  r8 þ e9  r9 þ e10  r10 þ e11  r11
r5

ð16Þ

and from the 2nd air inlet to the gasifier exit. Rtar2 ¼
r2
r9

ð17Þ

Rtar4 ¼ v9  r9
r2b

ð18Þ

RH2 ¼ d8  r8 þ d9  r9 þ d10  r10 þ r12
r4

ð19Þ

RCO ¼ e2  r2 þ e2b  r2b þ e3  r3 þ e3b  r3b þ e6  r6 þ e8  r8 þ e9  r9 þ e10  r10 þ e11  r11
r5
r12 ð20Þ

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1037

3.3. Mass balances for all species Two main assumptions were made: 1st) One-dimensional model. Piston or plug flow considered in the CFBBG for all the species. 2nd) Steady-state operation. Three different scales for the CFBBG were considered: pilot, demo and commercial scales (see Section 2.2). The temperatures used, to calculate the values for all the kinetic constants, come from the heat balances shown below. Pressure drop considered in the riser is 0.10–0.20 atm (10.1–20.2 kPa) and the percentage of the 2nd air flow between 0% and 40% (standard: 20%). Two different units for the space time (s) were used in the mass balances depending on the i-th species under consideration (gases, tars and chars). These units for s are indicated in the notation. 3.4. Heat balances Axial profiles of temperature (T) in the riser of a CFBBG are very important because all kinetic constants in the set depend on temperature. To calculate a given kinetic constant at a given height in the riser, the temperature at such height has to be known and to know such axial temperature profiles several heat balances have to be made. The contours for such heat balances are indicated in Fig. 4. For a given heat balance, the fuel gas composition has to be known at the inlet and exit of this contour which, in turn, depends on the temperature in such points. An iterative process therefore has to be used in each contour. For a CFBBG under steady state the contours under consideration are shown in Fig. 4: the bottom zone (contour 1, with the biomass feeding point in it), the 2nd air feeding zone (contour 2), the zones of the riser with no feeding points (contours 3a and 3b) and the whole riser (contour 4). Once fixed the process conditions the axial profiles of T are calculated but, since the heat released by each reaction in the network (Eqs. (1)–(12)) depends on its conversion and this conversion depends on T, an iterative calculus has to be made in each contour. The heat losses considered in each contour are the 2% of the overall heat released. For a CFBBG under a non steady state (starting up period, changing in biomass feeding flow, etc., . . .) the only contour considered was the whole gasifier (contour 4 in Fig. 4). In this case all the gasifier was considered isotherm. The evolution of the only one bgasifier temperatureQ then with time-on-stream, until a steady state was reached, was also calculated. 3.5. Hydrodynamic considerations and constrains The hydrodynamics in a CFBBG are extremely complex because at least there are the following four types of different solids in a CFBBG: – Silica sand (S), as fluidizing material. – Raw dolomite [(CO3)2CaMg] as additive for in-bed tar cracking and to prevent agglomerations. At gasifier temperature the raw dolomite quickly calcinates to

1038

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

Gas 4th

Heat losses

3rd b

2nd Air 2nd

3rd a

Biomass

1st Bottom ashes

1st Air

air conditions Tamb = 15 °C rel. moisture = 30%

Fig. 4. Zones or contours where heat balances have been made.

OCad OMg. Calcined dolomite (D) has a density half of that of the raw dolomite and a softness three times higher than the raw material dolomite. So, in fact, raw dolomite and calcined dolomite should be considered, both chemically and physically, as two different solids for hydrodynamics studies.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1039

– Biomass, which usually contains very different types of particles (from different origins) each having different shape factors. – Char, charred biomass and ash from biomass, with a very low density and high softness. They very easily erode at the bottom bed by the effect of the abrasive silica sand. When gasifying, the particle sizes of the biomass, charred biomass, char and ash change very much and very quickly [54]. The hydrodynamics of this mixture of solids in a CFBBG is very complex [58] and will not be well known until a lot of the remaining work is done in future. Some hydrodynamic aspects taken into account in the modeling of CFBBGs are the following: 1st) The zone between the inlet of the primary air and the biomass feeding point (height H1, see Fig. 1) is difficult to model: the axial profiles of biomass and chars (char1, char2 and char3) are not yet well known. Since there is a stationary fluidized bed in that zone [61], it might be supposed that there would be back-mixing for all species in solid phase, as biomass and chars. Nevertheless, chars, and also biomass, have such a low density compared with the one of the bfluidizing materialQ (silica sand and additive) that there is a near instantaneous segregation of the chars once generated in the pyrolysis zone, just above the biomass feeding point. Aznar et al. [55] studied this fast segregation and published a lot of axial profiles of the char and of some types of biomass in binary, ternary and quaternary mixtures. Until other even more accurate axial profiles for char and biomass are published, the axial profiles for char and biomass published by Aznar et al. [55] will be the ones used to calculate the extent of the reactions and the resulting profiles of the gases in that zone. 2nd) The comminution and carry over of the calcined dolomite in a CFBBG occurs by different simultaneous mechanisms, as studied by Aznar and co-workers [56]. They found the reason why 1 to 3 wt.% (respect to the biomass fed) of dolomite has to be continuously fed to the gasifier to replace the dolomite lost by elutriation. They also established why an enough good behaviour of a CFBBG occurs when there is a 20– 30 wt.% of dolomite in its bottom bed. The remaining 70–80 wt.% is silica sand. The char and ash content in the bottom bed is usually very low (1–4 wt.%) and can be omitted when calculating the weight hourly space velocity (WHSV) of the biomass [(kg biomass fed/h)/kg solids inventory (S+D)] in the CFBBG. 3rd) To get realistic data on hydrodynamics in commercial and in big (pilot, demo) biomass gasifiers, a worldwide survey was carried out among all the existing CFBBG manufacturers and owners. Most of them provided to the authors some important values, sometimes confidentially, about the following parameters in which their gasifiers operate: – WHSV [weight hourly space velocity, (kg/h)/kg inventory] from 1 to 3 h
1. – Throughputs (kg/h m2 cross-sectional area): between 1000 and 7000 kg/h m2. – Superficial gas velocities: 2–10 m/s. – Mean residence times for the gas in the bottom and in the dilute zones: 1 and 3 s, respectively, on average. – etc . . . (gas compositions, for instance).

1040

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

In the solution of the model shown here, all this information was taken into account. This information defined the limits of operation or constrains of the whole model. It was one of the main verifications (in the calculations carried out in the model) shown in Fig. 8.

4. Model development The simultaneous solution of the mass and heat balances with the kinetics and the hydrodynamic aspects was very complex. Although certain details are omitted some important results are given below: 4.1. Heat balances and the axial profiles of temperature Some results concerning the axial profiles of temperature are 1) Axial temperature profiles obtained from the model are of the same type as the few already existing in the literature for CFBBGs [28,57]. They were also confirmed by the measurements made in the small CFBBG shown in Fig. 2. 2) From the bottom of the gasifier until the 2nd air flow inlet point (H 1+H bt +H tz +H 3 in Fig. 1) the temperature decreases because the endothermal reactions in this zone consume more heat than that provided by the simultaneous exothermal ones. Nevertheless, one simplifying assumption can be made, at a first attempt, to facilitate further calculations in that zone: an average temperature can be considered in this zone. This average temperature is calculated supposing that all devolatilization occurs there and that all the 1st air reacts in this zone. To give one example, the temperatures in the bottom zone for different ER values are shown in Table 10 for a 2nd air flow of 20% of the total air flow, a biomass moisture of 15 wt.% and a 1st air flow preheated to 250 8C before feeding it to the CFBBG. These temperatures are the ones at the lowest part of Fig. 5. Temperatures at the bottom zone for several other process conditions will be shown in our next paper [33]. 3) One very important fact concerning the axial profiles of temperature (in the riser) is the increase of T (DT) because of the secondary air flow in the upper (dilute) zone. Such increase of temperature may also appear when the particle size of the biomass fed is very low. In that case it is carried out from the bottom bed by the flue gas and partially combusted, if there is still some O2 left there, in the upper (dilute) zone, as van der Drift and van Doorn [57] demonstrated at the ECN (The Netherlands) pilot

Table 10 Standard bottom zone temperature depending on the ER value (secondary flow=20% of ER; biomass moisture=15%; preheated air temperature=250 8C) ER T b. bed (8C)

0.20 824

0.25 837

0.30 850

0.35 863

0.40 877

0.45 890

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1041

16

14

12

Riser Height (m)

10

Total ER 0.20 0.25 0.30 0.35 0.40 0.45

8

6

4

2

0 800

850

900

950

1000

1050

Fig. 5. Longitudinal profiles of temperature for different total ER values with a 2nd air flow of 20% of total air.

plant. The reference (or average) DT by the 2nd air addition calculated for several percentages of this 2nd air is shown in Fig. 6. 4th) In the upper part of the riser, from the 2nd air inlet point till the gas exit point, there is a continuous decrease of the temperature. The heat consumed by the endothermal reactions there, together with the heat lost by the walls, is usually not compensated by the heat released by the exothermal reactions occurring with the O2 fed in the 2nd air flow. Again it can be assumed, as a first attempt and to facilitate some further calculations, that the upper zone of the riser is isothermal, with an average temperature in it. The simplified axial profiles of temperature in the riser of the CFBBG for the case of a 2nd air flow of 20% of the total air, and for a 2nd air inlet height of 6 m are shown in Fig. 5. These simplified profiles can be used in the whole model when more precise or accurate profiles

1042

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

∆T (= T exit-Tb.bed )

200

150

100

50

0 0

10

20

30

40

2nd air flow (%) Fig. 6. Difference of temperature (DT) used as reference between the averaged in the upper zone and in the bottom zone of the CFBBG, for different percentages of 2nd air flow.

cannot be calculated. In fact, in the present state of development of this model and with the existing data, the axial profiles of T cannot be calculated with accuracy for some experimental conditions (i.e., for some types of feedstocks or for some particle size distributions of the biomass). In such circumstances a simplified two-zone temperature profile, as the ones shown in Fig. 5, can be used. In such cases, the temperature is considered constant from the biomass feeding point until the 2nd air feeding point. Then, there is an increase of T (DT) by the 2nd air addition (which depends on the percentage of 2nd air used) and then, from the 2nd air feeding point to the exit of the riser, the temperature is considered constant. Only two calculations (not easy sometimes) are needed in this simplified approach: the temperature in the bottom zone and the DT value by the secondary air flow. 4.2. Hydrodynamics of the CFBBG Some noticeable comments about modeling the hydrodynamics in the CFBBG are: 1) Although it is well known how in CFBs there is a continuous, along their axis or height, variation of voidage (e) or solids fraction, the gasifier was considered as composed by the following three zones (see Fig. 1): – bottom zone, with a height of H bt . The reference porosity in this zone is 0.77. – transition or splash zone, with a height of H tz and e=0.90. – dilute zone, with H dz of height and e=0.99. Secondary air is fed into this zone.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1043

An exit zone may also be considered, if required. Other values for the porosities might also be used but these ones were considered as acceptable or representative; they are the averaged values from the published porosities for similar CFB reactors. 2) Clouds and strands of particles which may lead to a different velocity-behaviour of the gas and of the solids were not considered. Besides, the behavior of the solids is different in very slim reactors compared with wide commercial reactors. These facts, which have been very well studied in coal combustors for instance, have not yet been studied enough in biomass gasifiers. These phenomena are therefore recognized, but a piston flow (1-dimensional model) has always been used in the model. Some errors are recognized at this point in the model but this situation is acceptable until more useful data concerning CFBBGs are available. 3) Biomass devolatization once biomass enters into the bottom bed in the CFBBG, although it is very fast, it is not instantaneous. There is some evidence that a plume or something like a small jet of gas is formed (i.e., [39]). The flow in such feeding zone is not an academic piston or plug flow [39] and today there is not enough information published for a good sub-model for this zone. It will remain as one of the main unknowns or weak points of this model. Something similar occurs with the 2nd air flow and its mixing with the upwards flow. It is not instantaneous and generates a lot of effects (i.e., [74]). The authors think, from their own experience, that the bmixing zonesQ above both the biomass and 2nd air feeding points should be at least 1 m height each. Nevertheless, to facilitate handling the mathematics in the present model, it was considered that such feeding of biomass and 2nd air occurs at just bone point or heightQ, as Fig. 1 indicates. In real CFBBGs, however, the space above these points has a lot of heterogeneities. For instance, the pipe(s) for the biomass feeding has sometimes 0.50 m diameter, it is therefore not just a point. When examining longitudinal profiles for this model, it will have to be remembered that such two bpointsQ (biomass and 2nd air injection) are not bpointsQ but zones of around 1 m height. 4) The heights of the different zones are indicated in Fig. 1. From the works of Schlichthaerle and Werther [75], Malcus et al. [76], etc., . . . the standard values considered in this paper for these zones are H 1=1.5 m; H 2=1.0 m; H 4=0.5 m; H T =14.8 m. Although H bz and H tz (see Fig. 1) vary with the air flow rate or ER value, these heights were considered constant, 1.4 m each one, for all ER values. Nevertheless, all these heights are variables in the model and can be varied to adapt it to a existing gasifier. 5) The terminal velocity (u t) of the solids was calculated by well established and known methods, recently refined and applied by Hartman et al. [59] to the raw and calcined dolomites. u t for the four main solids (in the riser of the gasifier there could be another ones as big stones and irons too) existing in a CFBBG are shown in Fig. 7. The flue gas considered for such calculation was air at 850 8C and 1.1 atm of total pressure. The particle density and shape factor (very difficult to know for biomass char in a CFBBG under operation) of the solids considered are shown in Fig. 7. Since the biomass, the charred biomass and the char (and ash) usually have a broad spectrum of shape factors and of densities, a zone, instead of a line, was considered for the reacting biomass in this figure. If the superficial gas velocity at the exit of the riser of the CFBBG is 4 m/s, for instance, Fig. 7 indicates how a silica sand of 0.5–1.0 mm would remain at the bottom bed, not

1044

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

5

SILICA RAW SAND DOLOMITE

CALCINED DOLOMITE

Terminal velocity (m/s)

4 WOOD CHIPS CHAR FROM THE WOOD CHIPS

3

LEAFS

2

Silica sand, 2.6g/cm3, φ=0.9 Raw dolomite, 2.4g/cm3, φ=0.8 Calcined dolomite, 1.3g/cm3, φ=0.9 Wood chips, 0.85g/cm3, φ=0.4 Leafs, 0.5g/cm3, φ=0.1 Ash from wood, 0.2g/cm3

1

0 0

1

2

3

4

5

6

7

8

9

10

dp (mm) Fig. 7. Terminal velocity of the main solids existing in a CFB biomass gasifier (calculated for air, at 850 8C and at a total pressure of 1.1 atm).

transported (its u t is higher than 4 m/s). A raw dolomite of 0.6–1.0 mm would initially remain in the bed as well, but once and quickly calcined it would be elutriated out. Biomass, char and charred biomass particles below 1.0 cm approximately (depending on its initial shape, density and degree of charring) would be transported out from the bottom bed first and then out of the riser. Fig. 7 is interesting for the CFBBG modeling. For instance, the very important mean residence time of both gasification gas and biomass in the bottom bed can be varied by changing the bottom bed height which (for a given ER or superficial gas velocity) depends not only on the inventory of solids (kg of S+D) but mainly on their particle size distributions. Depending on their particle sizes they will generate a bed at the bottom of the CFBBG or carried out. Since the particle densities of the silica sand and of the raw dolomite are fixed, their particle sizes remain as the main variable which can be varied (between some intervals only) to increase or decrease the mean gas residence time in the bottom bed. This will modify, in turn, the composition, yield and quality (tar content) of the gas exiting the bottom zone. So, a good choice of the particle size distributions of both sand and dolomite is of paramount importance in the

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1045

CFBBG design and operation. Consequently, these particle sizes are very important in this modeling. 6) The main or first cyclone as well as a second and in-series high efficiency cyclone, usually required for some applications of the produced gasification gas, were taken into account but are not reported in this paper. The content of particles in the produced gas mainly refers to the top of the riser and not of the flue gas hence leaving the whole gasification unit. The calculus of the particle content in the fuel gas after the whole CFBBG still has a considerable error. In fact, the overall particle content is due to different solids (char, ash, calcined dolomite, etc. . .) and some of these contributions (to the whole content in particles) have not been deeply studied yet [56]. 7) Char recirculation: the total conversion of the char produced in a CFFBG is not produced in one pass of the solid through the riser [60]. The char is recycled in the CFBBG until the (highest) final conversion is obtained. The internal recycling of solids causes a spread in the residence time of the particles; some pass only once through the riser while others are recycled several times [60]. Some assumptions in the calculus of the char conversion include: – The inlet to the riser of the recycled char stream is made at the same height as the biomass feeding point. – The char composition in the recycled stream is the same as that at the outlet of the riser. – The conversion of the total char, the one generated in the pyrolysis step and the recycled stream of char, in each passage in the riser is the same in each passing. Several other hydrodynamic considerations were also used, but are not included here to keep this paper within limits. 4.3. Calculation procedure The overall model is protected and for reference purposes it will be called Al-CoreR. The flow diagram for the overall solution of the Al-CoreR program is shown in Fig. 8. For its solution, the following inputs must be given: – Biomass (B): flow rate, moisture and ash content, and its ultimate analysis daf (C,H,O wt.%). – Operational conditions: pressure, inventory or weight of bed material (assumed 30 wt.% dolomite and 70 wt.% silica sand), and temperature at inlet of bottom bed. – Inlet Air: ER, temperature of preheating, ambient temperature, relative moisture and 2nd/1st air flows. – Gasifier topology: main dimensions, wall thickness or weight of the gasifier (which is needed in a heat balance), etc . . . The flow diagram for the solution of the Al-CoreR program is shown in Fig. 9 with some more detail. Some procedural steps are given below. It is necessary to make a first assumption for the axial profile of voidage (e) and for the volumetric flow rate of the flue gas along the gasifier height ( Q 1).

1046

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

INPUTS B, E R ( ... )

Resolution of kinetics + mass balance eqns. (for each specie)

[1 ] [2 ] [3 ] .. .

Heat balances Temperature at inlet of bottom bed T1 (°C) - depending of ER

[ 12 ] Hydrodynamics: ε and char axial profiles

WHSV Total gas flow in the riser (Q). Sup. gas velocities, profiles Mean gas residence times T Xta r C ch a r .. . Xo 2 at exit of the gasifier

OUTPUTS Axial profiles of T (°C) Gas comp. (% vol.)

ni =

P · V1 Rg · T

LHV of gas (MJ /N m3)

· Xi

Q2 =

Rg ·T P

OK

YE S ·Σ n

i

Q2 = Q

Gas yield (Nm3/kg daf,dry basis)

NO T GOOD

NO

1

Tar content (g /Nm3)

Checking with exp. data

Char concentration (g/kg S+D )

Fig. 8. Simplified flow diagram of calculation for the overall solution of the 1st, 1-dimensional pseudo-rigorous model.

It is also assumed that the inlet air is heated at the bottom bed from the preheated air temperature, of around 250–350 8C, to the temperature at which the gasifier operates in the bottom bed. When the mass balance for all the species is solved, the composition (vol.%) of the gas species, char concentration and volumetric flow rate of gas ( Q 2) are obtained. Two zones of resolution have been considered (1st: bottom zone–transition zone–part of dilute zone; 2nd: rest of the dilute zone). In the 1st zone of resolution, two steps in series have been considered: the first one is the fast pyrolysis; with Eq. (1) the product distribution are obtained (see Section 3.2., kinetic equations used). The results from the fast pyrolysis are used in

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1047

START

INPUTS: B, ER, ... (As a function of the riser height) Supposed gas temperature Supposed riser voidage, total gas flow

INPUTS Supposed Q 1

MODULE 1 Bottom Zone-Pyrolysis SOLUTION: Subroutine 1

Xio, Cchar0 MODULE 2 1st O2 (air)

Transition, bottom zone and part of dilute zone SOLUTION: Subroutine 2 Subroutine 3

OUTPUTS: X1, Cchar, total gas flow LHV, C content in fly ash gas field Xi0: initial molar fraction of the components after the pyrolysis, for H2, CO, CO2, CH4, C2H4, H2O, N2, tar

XiA: molar fraction of the components before 2nd air inlet, for H2, CO, CO2, CH4, C2H4, H2O, N2, tar Xi: final molar fraction of the components, for H2, CO, CO2, CH4, C2H4, H2O, N2, tar

NO Q1 = Q1'

XiA, CcharA, Q1' Q1 = Q1'

YES XiA, CcharA, Q1' MODULE 3 2nd O2 (air)

CChar,0: char concentration after the pyrolysis in g char / (kg sand + dolomite) CChar,A: char concentration before 2nd inlet in g char / (kg sand + dolomite) CChar:

char concentration at riser exit in g char / (kg sand + dolomite)

Dilute zone SOLUTION: Subroutine 4 Subroutine 5 Subroutine 6

Subroutine 1: Pyrolysis reactions Subroutine 2: Oxidation of products from pyrolysis

Xi, Cchar,Q2

Subroutine 3: Tar - char reactions and steam reforming of methane

NO Q1'=Q2

Q2 = Q1'

YES OUTPUTS

Subroutine 4: Oxidation of products from the bottom bed Subroutine 5: Tar - char reactions and steam reforming of methane Subroutine 6: Shift reaction

STOP

Fig. 9. Flow diagram for the solution of the mass balances.

second step as initial conditions for the resolution of mass balances for oxidation, tar and char steam reforming and methane reactions. The values obtained at the exit of the first zone are used as initial conditions for the resolution of the mass balances in the 2nd zone. A fourth–fifth order Runge–Kutta method has been used to solve the mathematical problem.

1048

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

The volumetric flow rate of gas was considered constant along each zone. A volumetric flow rate of gas is assumed for the entry point of each zone and it is used to calculate the gas composition, tar content and char concentration at the end of that zone. The volumetric flow rate of gas is calculated then for the end of the zone. If the calculated volumetric flow rate of gas is not the same as the first value, Q 1pQ 1V, the mass balance is repeated until the two values agree. The volumetric flow rate of gas finally calculated for the first zone is used as the flow rate entering into the second one, then repeating the same calculation. The solution of the heat balance is carried out simultaneously with the kinetics and the mass balances. The char concentration in the flue gas was calculated in a two-step process: – Char concentration is calculated along the gasifier height without considering its recycling to the riser (first step). – Once fixed the efficiency (99% for instance) of the (first) cyclone, the char recycled to the bottom of the riser is known. The recycled (to the riser) char is taken into account in a second step for an improved calculation. When the kinetics and the mass and heat balances are solved, the axial profiles for tar content, char concentration, gas composition and LHV are obtained as model outputs. Besides, tar and char contents, C content in fly ash, gas composition, gas yield and LHV at the gasifier exit are obtained at very different operating conditions. Results from the model for very different operating conditions are shown in our next paper [33].

Notation b,d Reaction orders for the species i and w. B Biomass flow rate, kg daf/h. cv Volumetric fraction of solids, dimensionless. Ci Concentration of the i-th species in the gas phase, mol/m3. daf Dry, ash free. ER Equivalence ratio, air fed/stoichiometric air, dimensionless. fn Corrective factor in some kinetic equations. i (from (1)–(10)): H2,CO, CO2, CH4, C2H4, H2O, O2, tar, char. j Air, if the specie i considered is a gas; B daf, if the specie i is tar or char. kn Kinetic constant of the n-th reaction, units depending on the considered reaction. ˙ B Biomass flow rate (kg a.r./h). m ni Total moles of all the species in the flue gas, mol. P Pressure in the riser, atm. Q Volumetric gas flow rate, m3/h. R Ideal gas constant, J/mol K (and also atm l/mol 8C in R appearing in Fig. 8). Ri Overall reaction rate of the i-th species, units depend on the species considered. rj Rate of the j-th reaction, units (given by the authors who provided each rate) depend on the reaction and species considered.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

T tg uo ut W WHSV xi xw yi

1049

Temperature, K. Mean residence time of the gas, s. Superficial gas velocity at the bottom of the gasifier (gas inlet), m/s. Terminal velocity of the particle, m/s. Mass of sand and dolomite in the gasifier (kg). Weight hourly space velocity for the biomass [(kg biomass fed a.r.)/h]/kg solids inventory in the CFBBG. Molar fraction of the i-th species. For char, it is char concentration [kg char/kg (sand+dolomite)]. Molar fraction of w, botherQ gas species which affect to the kinetics of the i-th species. For char, it is char concentration [kg char/kg (sand+dolomite)]. Yield to the i-th species. For the gas species [H2,CO,CO2,CH4,C2H4,H2O, O2] its units are (mol i/mair in3). For tar and char its units are (mol i/kg B daf).

Greek symbols e Voidage in each zone of the riser, dimensionless. s Space-time, based on m3 of riser (mR3) when the i-th species considered is a gas component [units then: mR3/(mair,in3/h)], or based on mass of sand and dolomite (S+D) in the whole gasifier(W S+D) when the i-th species considered is tar or char [units then: kgS+D/(kg B daf/h)]. Acknowledgment This work was carried out under the EC-financed project no. JOR3-CT98-0306. A. Sanz thanks J.I. Civicos for his personal help in performing this modelling. Scientific discussions with Dr. Hao Liu from University of Leeds (Chem. Eng. Dept.) have been very stimulating for the development of this model.

References [1] K. Maniatis, Progress in biomass gasification: an overview, in: A.V. Bridgwater (Ed.), Progress in Thermochemical Biomass Conversion, vol. 1, Blackwell Science, London, UK, 2001, pp. 1 – 31. [2] J.R. Grace, A.A. Avidan, T.M. Knowlton (Circulating fluidized beds), Blackie Academic and Professional, Chapman and Hall, London, U.K. (1997). [3] R.R. Rhinehart, R.M. Felder, J.K. Ferrell, Dinamic modeling of a pilot-scale fluidized-bed coal gasification reactor, Industrial and Engineering Chemistry Research 26 (1987) 738. [4] E. Ciesielczyk, A. Gawdzik, Non-isothermal fluidized-bed reactor model for char gasification, taking into account bubble growth, Fuel 73 (1994) 105. [5] Y. Fang, J. Huang, Y. Wang, Mathematical modeling of a bench-scale CFB gasifier, 13th Annual International Pittsburg Coal Conference, 1996, pp. 1437 – 1442. [6] R. Sotudeh-Gharebaagh, R. Legros, J. Chaouki, J. Paris, Simulation of circulating fluidized bed reactors using ASPEN PLUS, Fuel 77 (1998) 327. [7] R.A. Srinivasan, S. Sriramulu, S. Kulasekaran, P.K. Agarwal, Mathematical modeling of fluidized bed combustion—2. Combustion of gases, Fuel 77 (1998) 1033. [8] H.M. Yan, C. Heidenreich, D.K. Zhang, Mathematical modelling of a bubbling fluidised-bed coal gasifier and the significance of bnet flowQ, Fuel 77 (1998) 1067.

1050

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

[9] Q. Wang, Z. Luo, X. Li, M. Fang, M. Ni, K. Cen, A mathematical model for a circulating fluidized be (CFB) boiler, Energy 24 (1999) 633 – 653. [10] L. Huilin, B. Rushan, L. Wenti, L. Binxi, Y. Lidan, Computations of a circulating fluidized-bed boiler with wide particle size distributions, Industrial and Engineering Chemistry Research 39 (2000) 3212. [11] Y.J. Kim, J.M. Lee, S.D. Kim, Modeling of coal gasification in an internally circulating fluidized bed reactor with draught tube, Fuel 79 (2000) 69. [12] C. Chen, M. Horio, T. Kojima, Numerical simulation of entrained flow coal gasifiers: Part I. Modeling of coal gasification in an entrained flow gasifier, Chemical Engineering Science 55 (2000) 3861. [13] C. Chen, M. Horio, T. Kojima, Numerical simulation of entrained flow coal gasifiers: Part II. Effects of operating conditions on gasifier performance, Chemical Engineering Science 55 (2000) 3875. [14] X. Li, J.R. Grace, A.P. Watkinson, C.J. Lim, A. Ergqdenler, Equilibrium modeling of gasification: a free energy minimization approach and its application to a circulating fluidized bed coal gasifier, Fuel 80 (2001) 195. [15] H.M. Yan, D.K. Zhang, Mathematical modelling of fluidised-bed gasifiers, Proceed 18th Ann. Int. Pittsburg Coal Conference, 2001. [16] F. Chejne, J.P. Hernandez, Modelling and simulation of coal gasification process in fluidized bed, Fuel 81 (2002) 1687 – 1702. [17] P. Belleville, R. Capart, A model for predicting outlet gas concentration from a wood gasifier, in: A.V. Bridgwater (Ed.), Thermochemical Processing of Biomass, Butterworks, London, UK, 1983, pp. 217 – 228. [18] C.C. Chang, L.T. Fan, W.P. Walawender, Dynamic modeling of biomass gasification in a fluidized bed, AICHE Symposium Series 80 (234) (1984) 80 – 90. [19] F.G. Van den Aarsen, Fluidized bed wood gasifier performance and modeling. PhD thesis, Univ. of Twente (NL), 1985. [20] J. Corella, F.J. Alday, J. Herguido, A model for the non-stationary states of a commercial fluidized bed air gasifier of biomass, in: G. Grassi, et al., (Eds.), Biomass for Energy and Industry, vol. 2, Elsevier, London, 1990, pp. 2804 – 2809. [21] J.-F. Bilodeau, N. The´rien, P. Proulx, S. Czernik, E. Chornet, A mathematical model of fluidized bed biomass gasification, The Canadian Journal of Chemical Engineering 71 (1993) 549. [22] H. Jiang, R.V. Morey, Pyrolysis of corncobs at fluidization, Biomass and Bioenergy 3 (1992) 81. [23] H. Jiang, R.V. Morey, A numerical model of fluidized bed biomass gasifier, Biomass and Bioenergy 3 (1992) 431. [24] S. Hamel, W. Krumm, Modelling and simulation of bubbling fluidized bed gasification reactors, Proceed. of 3rd European Conference on Fluidization, Toulouse, France, (2000), 29–31 May, 2000, pp. 519 – 527. [25] K.G. Mansaray, A.M. Al-Taweel, A.E. Ghaly, F. Hamdullahpur, V.I. Ugursal, Mathematical modeling of a fluidized bed rice husk gasifier: Part I—Model development, Energy Sources 22 (2000) 83. [26] K.G. Mansaray, A.M. Al-Taweel, A.E. Ghaly, F. Hamdullahpur, V.I. Ugursal, Mathematical modeling of a fluidized bed rice husk gasifier: Part II—Model sensitivity, Energy Sources 22 (2000) 167. [27] J. Corella, J. Herguido, J.M. Toledo, J.I. Gomez-Cı´vicos, Modeling fluidized bed biomass gasifiers: Part II. Gasification with steam in a bubbling fluidized bed, in: S. Kyritsis, et al., (Eds.), 1st World Conference on Biomass for Energy and Industry, vol. 2, James and James (Science Publishers), London, UK, 2001, pp. 1971 – 1975. [28] T. Jennen, R. Hiller, D. Kfneke, P.-M. Weinspach, Modeling of gasification of wood in a circulating fluidized bed, Chemical Engineering Technology 22 (1999) 822. [29] S. Hamel, W. Krumm, R. Abraham, A. Mittelst7dt, G. Karkowski, Modelling of pressurized fluidized bed gasification in comparison with experimental data from a commercial scale and pilot scale htw— gasification plant, Proceed. of 4th International Symposium on Coal Combustion, Beijing, P.R. China, 1999, pp. 411 – 420. [30] S. Hamel, W. Krumm, Mathematical modelling and simulation of bubbling fluidized bed gasiffiers, Powder Technology 120 (1999) 105 – 112. [31] S.R.A. Kersten, W. Prins, A. van der Drift, W.P.M. van Swaaij, Experimental fact-finding in CFB biomass gasification for ECN’s 500 kWth pilot plant, Industrial and Engineering Chemistry Research 42 (2003) 6755 – 6764.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1051

[32] H. Liu, B.M. Gibbs, Modeling NH3 and HCN emissions from biomass circulating fluidized biomass gasifiers, Fuel 82 (2003) 1591 – 1604. [33] A. Sanz, J. Corella, Modeling circulating fluidized bed biomass gasifiers: Part 4. Results from a pseudorigorous 1-dimensional model for stationary state, Fuel Processing Technology 86 (2005) 1021 – 1053 (submitted for publication). [34] J. Corella, J. Herguido, F.J. Alday, Gasification of biomass in fluidized bed. Influence of the location of the biomass feeding point on the product distribution, in: A.V. Bridgwater, J.L. Kuester (Eds.), Research in Thermochemical Biomass Conversion, Elsevier, London, 1988, pp. 384 – 397. [35] J. Gil, M.A. Caballero, J.A. Martı´n, M.P. Aznar, J. Corella, Biomass gasification with air in a fluidized bed: effect of the in-bed use of dolomite under different operation conditions, Industrial and Engineering Chemistry Research 38 (1999) 4226. [36] I. Narvaez, A. Orio, M.P. Aznar, J. Corella, Biomass gasification with air in atmospheric bubbling fluidized bed. Effect of six operational variables on the quality of the produced raw gas, Industrial and Engineering Chemistry Research 35 (1996) 2110. [37] J. Corella, J.M. Toledo, R. Padilla, Olivine or dolomite as in-bed additive in biomass gasification with air in a fluidized bed: which is better? Energy & Fuels 18 (2004) 713 – 720. [38] J. Corella, J.M. Toledo, Modeling a CFB biomass gasifier: Part I. Model formulation, in: A.V. Bridgwater (Ed.), Progress in Thermochemical Biomass Conversion, vol. 1, Blackwell Science, Oxford, UK, 2001, pp. 333 – 345. [39] E.-U. Hartge, K. Luecke, J. Werther, The role of mixing in the performance of CFB reactors, Chemical Engineering Science 54 (1999) 5393. [40] J. Corella, M.A. Caballero, M.P. Aznar, C. Brage, Two most advanced models for the kinetics of the tar composition variation in its catalytic elimination in biomass gasification, Industrial and Engineering Chemistry Research 42 (13) (2003) 3001 – 3011. [41] J. Corella, M.A. Caballero, M.P. Aznar, J. Gil, C. Brage, A 6-lump model for the kinetics of the catalytic tar removal in biomass gasification, in: S. Kyritsis, et al., (Eds.), 1st World Conference on Biomass for Energy and Industry, vol. 2, James & James (Science Publishers), London, UK, 2001, pp. 1471 – 1475. [42] J. Corella, J.M. Toledo, M.P. Aznar, Checking an advanced 6-lump model for the kinetics of the variation tar composition in its catalytic elimination in biomass gasification, in: Proceed. of 12th EC Conference on Biomass for Energy, Amsterdam (NL), 17–21 June 2002, vol. 2, 2002, pp. 927 – 930. [43] J. Gonza´lez-Saiz, Advances in biomass gasification in fluidized bed, PhD thesis, University of Saragossa (Dept. Chem. Eng.), 1988. [44] F.L. Dryer, I. Glassman, Fourteenth Symposium (Intl.) on Combustion, Pittsburg, PA, The Combustion Institute, Pittsburg, 1973, pp. 987 – 1003. [45] D.J. Hautman, F.L. Dryer, K.P. Schug, I. Glassman, Combustion Science and Technology 25 (1981) 219. [46] G. Lammers, A.A.C.M. Beenackers, J. Corella, Catalytic tar removal from biomass producer gas with secondary air, in: A.V. Bridgwater, D.G.B. Boocock (Eds.), Development in Thermochemical Biomass Conversion, vol. 2, Blackie Academic, London, UK, 1997, pp. 1179 – 1193. [47] A.M.C. Janse, H.G. de Jonge, W. Prins, W.P.M. van Swaaij, Combustio´n kinetics of char obtained by flash pyrolysis of pine wood, Industrial and Engineering Chemistry Research 37 (10) (1998) 3909 – 3918. [48] N. The´rien, P. Marchand, A. Chamberland, G. Gravel, Computer modeling and simulation of a biomass fluidized bed gasifier, Proceedings of the XVIII Congress: The Use of Computers in Chemical EngineeringCEF87, 26–30 April, Gianardi Naxos, Italy, 1987, pp. 187 – 192. [49] D.F. Fletcher, B.S. Haynes, F.C. Christo, S.D. Joseph, A CFD based combustion model of an entrained flow biomass gasifier, Applied Mathematical Modelling 24 (2000) 165 – 182. [50] C. Fushimi, K. Araki, Y. Yamaguchi, A. Tsutsumi, Effect of heating rate on steam gasification of biomass: 1. Reactivity of char, Industrial and Engineering Chemistry Research 42 (2003) 3922 – 3928. [51] F.G. van den Aarsen, A.A.C.M. Beenackers, W.P.M. van Swaaij, Wood pyrolysis and carbon dioxide char gasification kinetics in a fluidized bed, in: R. Overend (Ed.), Fundam. Thermochem. Biomass Conversion [Pap.-Int. Conf.], chap. 36, Elsevier, 1982, Pub. 1985.

1052

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

[52] J. Xu, G.F. Froment, Methane steam reforming, methanation and water–gas shift: I. Intrinsic kinetics, AIChE Journal 35 (1) (1989) 88 – 96. [53] P.A. Simell, E.K. Hirvensalo, S.T. Smolander, A.O. Krause, Steam reforming of gasification gas tar over dolomite with benzene as a model compound, Industrial and Engineering Chemistry Research 38 (1999) 1250. [54] F. Miccio, O. Moersch, H. Spliethoff, K.R.G. Hein, Generation and conversion of carbonaceous fine particles during bubbling fluidised bed gasification of a biomass fuel, Fuel 78 (1999) 1473. [55] M.P. Aznar, F.A. Garcı´a-Gorria, J. Corella, Axial profiles in the fluidization of sand-biomass and sand-char mixtures, Anales de Quimica (Madrid, Spain) 85 (1) (1989) 100 – 108. [56] M.P. Aznar, J. Gil, M. Caballero, J.A. Martin, Mechanisms of comminution and carry over of the calcined dolomite in fluidized bed biomass gasifiers, Personal Communication (2000) (to be published). [57] A. van der Drift, J. van Doorn, J.W. Vermeulen, Ten residual biomass fuels for circulating fluidized-bed gasification, Biomass and Bioenergy 20 (2001) 45 – 56. [58] M. Horio, Hydrodynamics. Chap. 2 of Circulating fluidized beds. in: Grace et al. (Ed.), Blackie Academic & Professional, Chapman and Hall, London, (1997) pp. 21–85. [59] M. Hartman, O. Trnka, K. Svoboda, Fluidization characteristics of dolomite and calcined dolomite particles, Chemical Engineering Science 55 (2000) 6269. [60] K. Smolders, D. Honsbein, J. Baeyens, Operating parameters for the circulating fluidized bed (CFB) combustion of biomass, in: A.V. Bridgwater (Ed.), Progress in Thermochemical Biomass Conversion, vol. 1, Blackwell Science, Oxford, UK, 2001, pp. 766 – 778. [61] T. Pugsley, G. Chaplin, S. Malcus, The hydrodynamics of the high-density bottom zone in a CFB riser analyzed by means of electrical capacitance tomography (ECT), Chemical Engineering Science 55 (2000) 4129 – 4138. [62] S. Kulasekaran, T.M. Linjewile, P.K. Agarwal, Mathematical modeling of fluidized bed combustion: 3. Simultaneous combustion of char and combustible gases, Fuel 78 (1999) 403. [63] M.L. de Souza-Santos, Modelling and simulation of fluidised-bed boilers and gasifiers for carbonaceous solid. PhD thesis, Department of Chemical Engineering and Fuel Technology, Univeristy of Heffield, UK, 1987. [Pub. In Fuel, 68 (1989) 1507 and Fuel, 73 (1994) 1459]. [64] C. Philippek, T. Knfbig, H. Schfnfelder, J. Werther, NOx formation and reduction during combustion of wet sewage sludge in the circulating fluidized bed—measurements and simulation, Proceed. int. conf. on fluidized bed combustion, vol. 2, ASME, New York, 1997, pp. 983 – 996. [65] Z. Tang, J. Zhang, Y. Wang, B. Zhang, Char gasification in fluidized bed with CO2, Journal of Combustion Science and Technology 2 (1) (1996) 31 – 37. [66] R. Zanzi, K. Sjfstrfm, E. Bjfrnbom, Rapid pirolysis of agricultural residues at high temperature, Presented at Expose´ over ffrbr7nningsforskning, Chalmers Tekniska Hfgskola, Gfteborg, October 21–22, 1998. [67] X. Dai, C. Wu, H. Li, Y. Chen, The fast pyrolisis of biomass in CFB reactor, Energy & Fuels 13 (1999) 426 – 432. [68] C. Di Blasi, G. Signorelli, C. Di Russo, G. Rea, Product distribution from pyrolysis of wood and agricultural residues, Industrial and Engineering Chemistry Research 38 (1999) 2216 – 2224. [69] C. Di Blasi, Modelling the fast pyrolysis of cellulosic particles in fluid-bed reactors, Chemical Engineering Science 55 (2000) 5999 – 6013. [70] A.N. Hayhurst, M.S. Parmar, Does solid carbon burn in oxygen to give the gaseous intermediate CO or produce CO2 directly? Some experiments in a hot bed of sand fluidized by air, Chemical Engineering Science 53 (3) (1998) 427 – 438. [71] S. Kulasekaran, T.M. Linjewile, P.K. Agarwal, M.J. Biggs, Combustion of a porous char particle in an incipiently fluidized bed, Fuel 77 (14) (1998) 1549 – 1560. [72] V. Cozzani, C. Nicolella, M. Rovatti, L. Tognotti, Modeling and experimental verification of physical and chemical proceses during Pyrolysis of a refuse-derived fuel, Industrial and Engineering Chemistry Research 35 (1) (1996) 90 – 98. [73] E. Kurkela, Formation and removal of biomass derived contaminants in fluidized-bed gasification processes, VTT Publications, vol. 287, VTT, Espoo, Finland, 1996. [74] Y. Kang, P.S. Song, J.S. Yun, Y.Y. Jeong, S.D. Kim, Effects of secondary air injection on gas–solid flow behavior in circulating fluidized beds, Chemical Engineering Communications 177 (2000) 31.

J. Corella, A. Sanz / Fuel Processing Technology 86 (2005) 1021–1053

1053

[75] P. Schlichthaerle, J. Werther, Axial pressure profiles and solids concentration distributions int the CFB bottom zone, Chemical Engineering Science 54 (1999) 5485. [76] S. Malcus, G. Chaplin, T. Pugsley, The hydrodynamics of the high-density bottom zone in a CFB riser analyzed by means of electrical capacitance tomography (ETC), Chemical Engineering Science 55 (2000) 4129 – 4138.