model development and validation_co-combustion of residual char, gases and volatile fuels in the...
TRANSCRIPT
1
2
3
4
5
6
78
9
1011121314151617
1819
20
2122232425262728293031323334
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
www.fuelfirst.com
Fuel xxx (2007) xxx–xxx
OO
F
Model development and validation: Co-combustion of residualchar, gases and volatile fuels in the fast fluidized combustion
chamber of a dual fluidized bed biomass gasifier
Priyanka Kaushal *, Tobias Proll, Hermann Hofbauer
Vienna University of Technology, Institute of Chemical Engineering, Getreidemarkt 9/166, A-1060, Vienna, Austria
Received 20 November 2006; received in revised form 14 March 2007; accepted 20 March 2007
R
CTED
PAbstract
A one-dimensional steady state model has been developed for the combustion reactor of a dual fluidized bed biomass steam gasifi-cation system. The combustion reactor is operated as fast fluidized bed (riser) with staged air introduction (bottom, primary and second-ary air). The main fuel i.e., residual biomass char (from the gasifier), is introduced together with the circulating bed material at thebottom of the riser. The riser is divided into two zones: bottom zone (modelled according to modified two phase theory) and upper zone(modelled with core-annulus approach). The model consists of sub-model for bed hydrodynamic, conversion and conservation. Biomasschar is assumed to be a homogeneous matrix of C, H and O and is modelled as partially volatile fuel. The exit gas composition and thetemperature profile predicted by the model are in good agreement with the measured value.� 2007 Published by Elsevier Ltd.
Keywords: Model; Fluidized bed; Char combustion
E
35363738394041424344454647484950
UN
CO
RR1. Introduction
Currently, biomass fuel provides 15% of world’s pri-mary energy and this fulfilment establishes biomass asworld’s fourth largest energy source after oil, coal andgas. Pyrolysis, gasification and combustion are the primetechnologies for thermal conversion of biomass. Gasifica-tion is the process where, through thermal decompositionin a low oxygen environment at temperatures of 800–900 �C, organic matter is converted to combustible gaseswith tars and chars as possible side-products. Air gasifica-tion results in the inferior quality gas (4–7 MJ/Nm3 LHV).Oxygen as a gasification agent produces superior qualitygas (10–16 MJ/Nm3 LHV) but requires additional costsfor oxygen production. A similar high quality gas can beproduced by using a dual fluidized bed (DFB) system with
51525354
0016-2361/$ - see front matter � 2007 Published by Elsevier Ltd.
doi:10.1016/j.fuel.2007.03.032
* Corresponding author. Tel.: +43 1 58801x15965; fax: +43 1 5880115999.
E-mail address: [email protected] (P. Kaushal).
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
steam as a gasification agent [1]. The fundamental idea ofthe DFB concept is to divide the gasifier into two fluidizedbed reactors, a gasification reactor and a combustion reac-tor. The energy needed for the endothermic gasificationreactions in the gasification reactor is provided by combus-tion of residual biomass called char in a combustion reac-tor. The energy released during combustion is transportedto the gasification reactor along with the circulating bedmaterial, thus making DFB gasifier a self sustainedprocess.
The DFB gasifier can be regarded as a circulating fluid-ized bed (CFB) system with the bubbling bed gasifier in thereturn loop of the solids. A number of different CFB mod-els have been developed and reviewed [2,3] but most ofthem are developed for coal combustion, i.e., for combus-tion of low volatile fuels. So far, biomass processing inCFBs has received little attention [4] and only a fewattempts have been made to model biomass char as partialvolatile fuel in CFB covering hydrodynamics and reactionkinetics. Further, confidentiality, numerous assumptions
ent and validation: Co-combustion of residual ..., Fuel (2007),
REC
TED
PR
OO
F
55565758596061
62
63646566676869707172
7374757677787980818283848586878889909192
Nomenclature
A area (cross-sectional area) (m2)a decay constant (m�1)Cp heat capacity (J/mol K)D diameter of column (m)D molecular diffusivity (m2/s)Da diameter of annulus (m)Dc diameter of core (m)dB bubble diameter (equivalent diameter of a
sphere) (m)dp particle diameter (m)dchar char diameter (m)g gravity (m/s2)_H enthalpy flow (J/s)
H(T) enthalpy at temperature T (J/mol)DH 0
f ;298 enthalpy of formation at 298.15 K (J/mol)hm mass-transfer coefficient (kg/m2 s bar)KB,E mass-transfer coefficient between bubble and
emulsion (s�1)Ki1 elutriation rate constant (kg/m2 s)kB,E mass-transfer coefficient between bubble and
emulsion (m/s)k reaction rate constant (dependent)Mc molecular weight of carbon (kg/mol)_m mass flow rate (kg/s)mbed mass hold-up (bed material) (kg)mchar mass hold-up (char) (kg)N no of cells (–)Nor number of orifices (–)_n molar flow rate (mol/s)n order of reaction (–)P partial pressure (bar)QB volumetric flow rate in bubble (m3/s)
QEX gas exchange (m3/s)q specific combustion rate of carbon on external
surface (kgcarbon/m2s)R universal gas constant (J/mol K)SB,E surface area of bubble in contact with emulsion
(m2)Schar surface area of char (m2)Sh Sherwood number (–)T temperature (K)U0 superficial velocity (m/s)UB bubble velocity (m/s)UB1 velocity of a single isolated bubble (m/s)UE velocity of gas in emulsion (m/s)Umf minimum fluidization velocity (m/s)Ut terminal velocity (m/s)Y correction factor for modified two phase theory
(–)y mole fraction (–)z height (m)z0 height of dense zone (m)
Greek symbols
dB bubble fraction (–)emf porosity at minimum fluidization condition (–)ez average porosity at height z (–)e0 average porosity in dense zone (–)e1 Porosity above TDH (–)/ mechanism factor for primary surface product
(1 for CO2, 2 for CO) (–)k air ratio (–)l viscosity (Pa s)qp density of particle (kg/m3)
2 P. Kaushal et al. / Fuel xxx (2007) xxx–xxx
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
UN
CO
Rvaried in a wide range makes the evaluation and compari-son of these models very difficult. Hence, a mathematicalmodel is developed for the riser to gain the better under-standing of the process. The model is based on knownapproaches and with respect to the data available (for val-idation) from the DFB gasification plant at Guessing/Austria.
2. Model development
The present work focuses on the mathematical model-ling of the combustion reactor indicated by the broken linein Fig. 1. The boundaries of the investigated system are theconnecting chute at the bottom of the riser where bed mate-rial enters together with the residual biomass char from thegasification reactor and the cyclone at the top of the riserwhere the hot solids are separated from the flue gas stream.The objective of this study is to develop a model to predictthe gas phase concentration and temperature profile alongthe height of the reactor.
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
The combustion reactor is divided into two zonesnamely dense and transport zone, having different hydro-dynamic characteristics. Transport zone is further subdi-vided into middle and upper zone as shown in Fig. 1.Preheated air is introduced into the riser at three differentpoints which are termed as bottom air, primary air and sec-ondary air. Preheated gaseous fuel (i.e., recycled producergas from the gasifier) is also introduced into the reactorto control the temperature of the entire system.
Each zone is further divided into cells and within eachcell the local hydrodynamic, kinetic and thermodynamicvariables are evaluated. Cells are solved sequentially withoutput of the Nth cell considered as input for the(N + 1)th cell. The energy conservation equation is solvedacross the entire zone, and it is assumed that within a zonethe temperature is uniform. This assumption is based on,efficient solids mixing and the dominance of the solid’s heatcapacity within a fluidized bed. The inputs for the modelare the riser geometry, particle properties, gas compositionand flow rate (Table 1)
ent and validation: Co-combustion of residual ..., Fuel (2007),
CT
93
949596979899
100101
102103104
105106107108
110110
111112113114115
117117
118119120121122123
125125
126127128129130131132133134
136136
137138139
141141
Flue gas
Solid
Solid
(Riser)
Secondaryair
Primaryair
Bottom air
Producergas
Producer gas
Biomass
Gasifier
Steam
Combustor
Modelboundary
Transportzone
Densezone
Fig. 1. Model structure and boundaries (1: dense zone, 2: middle zone, 3:upper zone).
P. Kaushal et al. / Fuel xxx (2007) xxx–xxx 3
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
OR
REModel assumptions:
The system is one-dimensional and in steady state.Gases are ideal and in plug flow.Solids are uniform in size and well mixed.Each of the three main zones has a uniformtemperature.The mass transfer of a single gas species between theemulsion phase and bubble phase is modelled by amass-transfer coefficient (kBE).
UN
C 142143
145145
146147148149150151152153154155
Table 1Input parameters to the model (typical operating condition from Guessingplant)
Diameter of column 0.61 (m)
Height of column 12 (m)Diameter of particle 500E�6 (m)Diameter of coke 0.001 (m)Density of bed material 2960 (kg/m3)Density of coke 600 (kg/m3)Bed circulation 37 (kg/s)Temperature of gasifier 883 (C)Bottom air feed rate 700 (Nm3/hr)Primary air feed rate 2222 (Nm3/hr)Secondary air feed rate 1043 (Nm3/hr)Char composition (C, H, O) (0.82, 0.04, 0.14) (wt/wt)
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
ED
PR
OO
F
The hydrodynamic parameters are evaluated based onthe properties of bed which is classified as Geldart groupB.
The flow regime map of Grace [5] between dimensionlessvelocity (U*) and dimensionless particle diameter (d�p) con-firmed that the fluidization state of the dense bed is in bub-bling mode. Gas balance based on superficial velocity is
U 0 ¼ UBdB þ ð1� dBÞemfU E: ð1ÞTwo phase theory (TPT) assumes that visible bubble flowcorresponds to the excess gas velocity i.e., (U0 � Umf).However, this theory tends to overestimate the bubbleflow. In order to limit the deviation from observations,the modified two phase theory [6] is used, where
QB ¼ Y ðU 0 � UmfÞ: ð2ÞY is always below unity and usually in the range from 0.7to 0.8. Y can also be 0.3 for coarse particles. The reportedvalues of Y in literature scatter in a wide range [7]. Y is afunction of the height above the gas distributor and hencethe correlation proposed by Johnsson et al. [8] is used toestimate the value of Y.
Y ¼ 0:26þ 0:7 expð�0:0033dpÞ0:15þ ðU 0 � U mfÞ½ �0:33
zþ 4
ffiffiffiffiffiffiffiA
Nor
r� �0:4
: ð3Þ
Bubbles are formed as soon as the superficial gas veloc-ity exceeds minimum fluidization velocity. The bubble sizeis one of the most critical parameters when modelling fluid-ized beds, it affects the bubble rising velocity, the bubblefraction and the mass transfer between bubble and emul-sion phase. As the bed height increases, bubbles coalesceand the bubble size increases. The bubble size is calculatedas a function of bed height and it is assumed that all bub-bles at any cross section are of uniform size [9].
dB ¼ 0:54ðU 0 � U mfÞ0:4 zþ 4
ffiffiffiffiffiffiffiA
Nor
r� �0:8
g�0:2: ð4Þ
Bubble’s interaction influences the bubble rising velocityand the common adaptation for bubble rising velocitygiven by Davidson and Harrison [10] is used in this work
UB ¼ UB;1 þ ðU 0 � UmfÞ: ð5Þ
For bubbles, the rising velocity of a single isolated bubbleis given below.
UB;1 ¼ 0:71ffiffiffiffiffiffiffiffigdB
pð6Þ
There is no theoretical basis to support this correlation,however, numerical simulations [11,12] show that thisapproximation is reasonable.
It is evident that a major part of the gas flows in the bub-ble phase. Thus, the interphase mass transfer between bub-ble and emulsion phase is essential for the gas–solidreactions. The interphase mass transfer is modelled by asemi-empirical approach using a mass-transfer coefficient,the exchange area and the concentration gradient accord-ing to Sit and Grace [13]
ent and validation: Co-combustion of residual ..., Fuel (2007),
T
157157
158159160161162163164165166167168169170171172173
175175
176177178179180181182
184184
185186187188
190190
191192
194194
195196197198200200
201202203204205206207208209210211212213214215216217218
220220
221222223224
226226
227228229230231232233234235
237237
4 P. Kaushal et al. / Fuel xxx (2007) xxx–xxx
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
CO
RR
EC
KBE ¼Umf
4þ
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4emfDU B
pdB
s: ð7Þ
Fig. 2 shows the gas flow streams in the dense zone and thegas exchange between the bubble and emulsion phase.After the injection of primary air into the combustion reac-tor, the superficial gas velocity is increased and the reactoroperates in fast fluidization regime. Fast fluidized beds ex-hibit strong radial gradients with respect to solid hold-upand solid movement. Near the column walls the particlemoves downwards and the concentration of particles ishigher than in the core of the reactor, where particles aretransported upwards. Different approaches are proposedto describe fast fluidized bed reactors [14–16]. In the pres-ent work, the transport zone is modelled by a simplifiedcore-annulus approach. The axial mean voidage is calcu-lated using an exponential decay function as proposed byZenz and Weil [17]
ez � e1e0 � e1
¼ exp½�aðz� z0Þ�: ð8Þ
Unknown parameters in Eq. (8) are the decay factor ‘a’ andthe porosity above transport disengaging height ‘e1’. Thevalue of ‘a’ varies widely in the literature and is approxi-mately in the range of 2–6.4 m�1 [18]. Cold flow studiesof the process showed that the decay constant dependson the fluidizing velocity to a higher power than one. Thus,the correlation proposed by Adanez et al. [19] was adopted.
aðU 0 � U tÞ2D0:6 ¼ K: ð9Þ
Based on the pressure drop measurement at standard con-ditions the constant K was determined to be K = 8.6. Thevalue of e1 is determined from the particle elutriation rateconstant Ki1 for a mono-sized bed material:
1� e1 ¼K i1
qpðU 0 � U tÞð10Þ
K i1 ¼ 0:011qp 1� U t
U 0
� �2
: ð11Þ
The voidage in the core at a given height is calculated fromthe mean porosity ez at that height [20]
eC ¼ 1� 0:6ð1� ezÞ: ð12Þ
UN
EMU
LSIO
N
BUBB
LE
out (N−1)EQ BQ
EXQ
out (N) out (N)EQ BQ
out (N−1)
Fig. 2. Gas balances in dense zone.
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
ED
PR
OO
F
It is assumed that the annulus is maintained at minimumfluidization conditions. The diameter of the core increaseswith the height of the riser and is estimated from Eq. (13).
eaD2a þ ecD2
c ¼ ezD2: ð13Þ
Unlike dense zone in the transport zone it is assumed that,there is a single gas stream across both the core and annu-lus. The local variables (e.g., reaction rate, hold-ups, etc.)are calculated separately for core and annulus but the massbalances are applied across the whole cell as shown by thebroken line in Fig. 3.
Char combustion is in general a complex process due tothe combination of different mechanisms as mass transfer,chemical reaction and heat transfer. In general, combustionof char begins after the evolution of volatiles. Sometimes,especially for large particles, there is a significant overlap-ping of these processes. The char composition is an impor-tant model parameter and is assumed to be a homogeneousmatrix of carbon, hydrogen and oxygen. Carbon present inchar is modelled as non-volatile fuel subject to heteroge-neous oxidation with O2 and if no O2 is present then withH2O. The primary products in the case of char combustionare CO and CO2 represented by reaction A.
ð1þ bÞCþ ð1þ b2ÞO2 ! bCOþ CO2; ðAÞ
where b is a stoichiometric coefficient for char combustiondescribing the ratio of CO to CO2 and is a function of sur-face temperature [21]. Char gasification reaction (B) is alsoconsidered in the model.
CþH2O !k COþH2: ðBÞ
The carbon combustion model is based on the combinedeffects of gas film diffusion and reaction kinetics on ashrinking particle and it is assumed that the diameter ofthe char particle decreases while the density of the char re-mains constant. Oxidation occurs in a thin layer close tothe particle surface and no ash layer is formed. The com-bustion rate in term of mass transfer through the gas filmcan be described as
q ¼ hmðP O2;g � P O2;sÞ ð14Þ
hm ¼M c/ShDdcharRT
: ð15Þ
N-1
N+1
EXQ
Fig. 3. Gas balances in transport zone.
ent and validation: Co-combustion of residual ..., Fuel (2007),
T
238239240242242
243244245246247248249250251252253
255255
256257
259259
260261262263264265266267268269270271
273273
274275276
278278
279280281283283
284
286286
287288289290291292293294295296297298299300
302302
303304305306307308
P. Kaushal et al. / Fuel xxx (2007) xxx–xxx 5
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
REC
The chemical reaction rate of carbon with oxygen per unittime and per unit external surface area of the particle is
q ¼ kP nO2;S
Schar: ð16Þ
Eqs. (14) and (16) are solved under steady state assump-tion,. The data obtained from experiments allow minimiza-tion of the root of mean square for the Arrheniusexpression [22].
The hydrogen and oxygen present in char are modelledas volatile fuel. Hydrogen and oxygen are released in a waythat the char composition remains constant as conversionproceeds. The hydrogen and oxygen released react to formwater (C). The ratio of CO, CO2, H2, O2 and H2O release iscalculated from the combustion kinetics of carbon and bal-ances of char.
H2 þ1
2O2 () H2O: ðCÞ
Carbon monoxide released from char is modelled to under-go combustion reaction and shift reaction
COþ 1
2O2 () CO2 ðDÞ
COþH2O() CO2 þH2: ðEÞ
The effect of the CO shift reaction becomes pronouncedwhen oxygen within a zone is consumed. As shown inFig. 1, a small amount of producer gas is also recycled intothe middle zone of the combustion reactor. The combus-tion kinetics and composition of the producer gas are givenin Tables 2 and 3, respectively.
The elementary mass balances inside every single cell iscovered by the stoichiometry of the homogeneous and het-erogeneous reaction formulations. The mass balance forsolid carbon and the energy balance are applied globallyacross each of the three zones as indicated in Fig. 1.
Char is evenly distributed in each cell of dense zone as
mcK ¼1
NðmcÞ; ð17Þ
UN
CO
R
Table 2Reaction rates
Reactions Reaction rate
A (1+b)C +(1+b/2)O2! CO + CO2 q = K[PO2]0.5 SB C + H2O! CO + H2 q = K[PH2O]0.5
C H2 +1/2 O2! H2O rH2 = K[H2]1.5[D CO + 1/2 O2! CO2 rCO = K[CO][OE CO +H2O! H2 + CO2 rCO = K[CO][HF CH4 + 3/2 O2! CO + 2H2O rCH4 = K[CH4]G C2H4 + 2O2! 2CO + 2H2O rC2H4 = K[C2HH C2H6 + 5/2 O2! 2CO + 3H2O rC2H6 = K[C2HI C3H8! C2H4+ CH4 rC3H8 = K[C3H
Table 3Producer gas composition
LHV (dry) Composition (vol%)
(MJ/Nm3) CO CO2 CH4 C2H4
13.1 21 20 10 2
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
ED
PR
OO
F
where as in transport zone char is distributed in cells insuch a way that the ratio of bed hold-up in adjacent cellis the same as the ratio of char hold-up.
mbedkþ1
mbedk
¼ mcharkþ1
mchark
: ð18Þ
Globally across the entire zone, the char balance is formu-lated as
_mChar;out þ _mChar;reacted ¼ _mChar;in: ð19Þ
The general carbon balance of each zone is written as
_mcin¼ _mcout þ
XN
k¼1
dmc
dt: ð20Þ
All feed rates are input parameters to the model. Becauseof the assumption of uniform char concentration in the sol-ids, the char exit rate and the char hold-up in the zone aredirectly linked to each other. The char conversion ratefrom solid to gas is dependent on several parameters withineach cell and depends directly on the available char surfacearea within the zone, the zone temperature and partialpressures of the reactants, i.e., O2 or H2O.
The energy balance is formulated globally across theentire zone using total enthalpies including the standardenthalpy of formation with the assumption that the totalamount of energy entering the system is actually leavingthe system.X
_Hout ¼X
_H in ð21Þ
HðT Þ ¼ DH 0f;298 þ
Z T
T¼298:15
CpðT ÞdT : ð22Þ
The enthalpy of ideal gas mixtures is calculated using theNASA polynomials with coefficients reported by Burcatand McBride [29]. The enthalpy of the inert bed materialis calculated by interpolation of data reported by Barin[30]. The sensible heat of char is calculated using the corre-lations reported by Merrick [31]. The enthalpy of forma-
K Ref.
char 8.56 · 10�2 exp(�2237/T) [22]7 2.62 · 108 exp(�237000/T) [23]O2] 1.63 · 109 T3/2 exp(�3420/T) [24]
2]0.5[H2O]0.5 3.25 · 107 exp(�15098/T) [25]
2O] 0.03 e exp(�7249/T) [26]0.7[O2]0.8 4.68 · 1018 T0.5 exp(�20087/T) [27]
4][O2] 1.0 · 1012 exp(�20844/T) [28]
6][O2] 2.34 · 1018 T0.5 exp(�20087/T) [27]
8] 1.0 · 1012 exp(�21145/T) [28]
C3H8 H2 H2O O2 N2
1 35 9 0 2
ent and validation: Co-combustion of residual ..., Fuel (2007),
OO
F
309310311312313314315316317318319320321322323324325
327327
328
329330331332333334335
336
337338339340341342343344345346347348349350351352353354355356
0
0.5
1
1.5
0.00 0.05 0.10 0.15 0.20
Hei
ght o
f ris
er (m
)
0 0.01 0.02Mole fraction (-)
BCO2
ECO2
BO2
EO2
EH2O
ECO
*EH2
BH2O
* Sec. axis
BCO
6 P. Kaushal et al. / Fuel xxx (2007) xxx–xxx
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
tion for char is calculated from the heating value, estimatedfrom ultimate analysis using the Boie formula [32]. _Hout is afunction of the exit temperature of the respective streams.Since a uniform temperature is assumed inside the zone,the problem is reduced to the determination of the twovariables char hold-up and temperature in order to fulfilEqs. (19) and (21).
A two-dimensional Newton–Raphson algorithm isapplied where the components of the Jacobian matrix arecalculated analytically. The iteration is performed for eachof the three zones where the order is again from bottomtowards top of the combustion reactor. After convergence,the model yields the gas phase concentration profiles, thetemperature in each zone, the solid char flow rate leavingthe reactor and the apparent global air ratio k which isdefined by neglecting the bypassing fraction of the char(i.e., only the actually converted char is considered).
PR
Mole fraction (-)
Fig. 4a. Gas composition profile in dense bed. B: bubble; E: emulsion.
k ¼½ _nyO2
�in½ _nyO2
�in � _nfyO2� 0:5yCO �
PCxHyðxþ 0:5yÞyCxHy
gh i
out
:
ð23Þ
TED
2
4
6
8
10
Hei
ght o
f ris
er (m
)
COCO2H2OO2
C3. Results and discussion
The solution of the mathematical model representssimultaneous convergence of the conservation equation(for which different subroutines have been defined). Table1 lists the input parameters for a typical operation case.The average residence time of the gas in the combustionreactor is in the range of 1.0–1.5 s while the air ratio inthe standard condition is approximately 1.02.
E357358359360361362363364365366367368369370371372373
00.00 0.10 0.20
Mole fraction (-)
Fig. 4b. Average gas composition profile along height of riser.
UN
CO
RR3.1. Predicted profile of gas composition
Fig. 4a shows the concentration profile inside the bot-tom bed (dense zone) separately for the bubble and emul-sion phases. As can be seen, the concentration of oxygenin emulsion phase goes down swiftly and when O2 is com-pletely consumed, the char gasification and shift reactiondominates; this can be observed by the increased concen-tration of H2 and CO2 in emulsion phase at the relevantheights. In the bubble phase, the concentrations of COand H2 are very low while that of oxygen is relatively high.This is one of the major limitations of one-dimensionalmodel that it fails to explain the three-dimensional mixingeffect. Nevertheless such behaviour of dense bed is oftenreported [4,33]. The overall gas composition profiles alongthe riser height are shown in Fig. 4b. The concentrationprofiles of CO, CO2, H2 and H2O locally go down at theheight of 2 and 4 m due to dilution effect caused by primaryand secondary air addition. Results show that the producergas added in the middle zone (rich in combustible gas)quickly combusts to CO2 and H2O. The predicted gas-pro-file inside the riser could not be verified at the demonstra-
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
tion plant at Guessing because of the lack of measurents.Therefore, to validate the model, the operating parametersof the present model were adjusted in order to match withthe operating parameters of the fluidized bed boiler atChalmers University (Sweden), reported by Adanez et al.[4]. Newton–Raphson method was used to iterate the woodfeed rate to fit the excess air ratio reported by Adanez et al.Fig. 5 shows a comparison between the measured and pre-dicted oxygen concentrations along the height of riser. Theoxygen profile predicted by the present model is in fairagreement with the measured values [4,34]. As seen inFig. 5, an increase in oxygen concentration with height isnot realistic and as indicated by Amand et al. [33], it canbe attributed to an insufficient penetration depth and mix-ing of the secondary air flow. Nevertheless Adanez et al.investigated various possibilities to deal the mixing prob-lem in one-dimensional model.
ent and validation: Co-combustion of residual ..., Fuel (2007),
UN
CO
RR
EC
T
374
375376377378379380
381
382383384385386387388389390391392393394395396397398399400401402403404405
0
2
4
6
8
10
0 0.1 0.2 0.3Mole fraction O2 (-)
Hei
ght o
f ris
er (m
)
Adanez et al. [4]Present model
Measured value from the fluidized bed boiler at ChalmersUniversity Sweden [4,34].
Center lineAverage
Fig. 5. Comparison of model prediction for O2 concentration.
0
2
4
6
8
10
880 900 920 940 960Temperature (°C)
Hei
ght o
f ris
er (m
)
Avg. temperature of zone( model prediction) Measured temperature
Den
sezo
ne
Mid
dle
zone
Upp
erzo
ne
Fig. 6. Measured and predicted temperature.
0
0.2
0.4
0.6
0.8
1
1.2
1.4
0 0.2 0.4 0.6 0.8Char feed
Ap
par
ent
air
rati
o (
-)
dc = 4 mm
dc = 8 mm
dc = 12 mm
dc = 16 mm
dc = 20 mm
dc = 24 mm
Fig. 7. Effect of char feed ra
P. Kaushal et al. / Fuel xxx (2007) xxx–xxx 7
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
ED
PR
OO
F
3.2. Predicted temperature profile
At the demonstration plant at Guessing, the tempera-ture inside the riser is measured at three different heights.The predicted temperature profile is in good agreementwith the measured value (Fig. 6). A little over-predictionin temperature is observed but the deviation in temperatureprediction is less than 5 K.
3.3. Effect of char diameter
Fig. 7 shows the effect of char feed rate on air ratio atdifferent char sizes. It should be noted that all the otheroperating parameter were fixed except char feed rate andsize. Each curve in Fig. 7 were generated by changing thechar feed rate at a given constant char size. Result showsthat with increase in char feed rate, the air ratio decreases.The slope of the curve changes at an air ratio of 1, suggest-ing that for lower char feed rates, char size have a veryprominent effect on char conversion. Simulation showsthat for constant char feed of 0.8 kg/s air ratio changesfrom 0.85 to 0.93 when char size is changed from 4 to8 mm. Therefore, it can be concluded that smaller sizefavours char conversion. The simulation results also showthat the residual char from the gasifier is only partly con-verted in the riser and the remaining un-combusted charis circulated back into the gasification reactor and poten-tially leads to an increase of char hold up in the dual fluid-ized bed system. Fig. 8 shows the exit CO concentration forthe various points shown in Fig. 7. It can be seen in the fig-ure, that for air ratio below 1, the CO concentrationincreases abruptly. It also shows that irrespective of thechar size the CO concentration follows the same trend linesuggesting that exit CO concentration is a strong functionof air ratio and is independent of char size.
1 1.2 1.4 1.6 1.8 rate (kg/s)
Increasing char diameter
te on apparent air ratio.
ent and validation: Co-combustion of residual ..., Fuel (2007),
T406
407408409410411412413414415416417418419420421422423424425426427428429430
431
432433434
435
436437438439440441442443444445446447448449450451452453454455456457458459460461462463464465466467468469470471472473474475476477478479480481482483484485486487488489490
0
20
40
60
80
100
120
0 0.4 0.8 1.2 1.6 2Apparent air ratio (-)
CO
(ppm
x 1
0-3 )
dc = 4 mm
dc = 8 mm
dc = 12 mm
dc = 16 mm
dc = 20 mm
dc = 24 mm
Fig. 8. CO concentration in flue gas.
8 P. Kaushal et al. / Fuel xxx (2007) xxx–xxx
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
NC
OR
REC
4. Conclusions
A one-dimensional steady state model of a riser has beendeveloped to analyze the combustion of biomass char incirculating fluidized bed. The model is based on mass andenergy balances. The model includes different sub-modelsthat are linked together to describe the overall combustionprocess. The hydrodynamic sub-model highlights the phys-ical characteristics of the bed material while the reactionsub-model deals with the chemical reactions in the differentzones of the combustion chamber (riser). The model is fuelflexible and offers the opportunity to evaluate different fueltypes over wide range of composition both for solid andgases. Global reaction rates are used for the descriptionof chemical kinetics. Model validations have been per-formed with measured results obtained at both 8 MWthdual fluidized bed gasification plant at Guessing, Austriaand with literature data on the 12 MWth circulating fluid-ized bed boiler at Chalmers University (Sweden). The pre-dicted trends are qualitatively compared with the publisheddata [4,33] and it turns out that they are in fair agreement.The model predictions show that the gas composition var-ies significantly in the upper section above the secondaryair injection. Simulation results also show that smaller sizeof wood or char favours the overall performance of thegasifier.
U 491492493494495496497498
Acknowledgement
The authors gratefully acknowledge the financial sup-port from RENET Austria (Knet/Kind public fund pro-gram, Austria).
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
ED
PR
OO
F
References
[1] Hofbauer H. CFB – steam gasification. 14th European biomassconference and exhibition biomass for energy, industry and climateprotection. 17–21 October 2005, Paris, France.
[2] Grace JR, Avidan AA, Knowlton TM. Circulating fluidized bed-s. New York: Chapman and Hall; 1997.
[3] Basu P. Combustion of coal in circulating fluidized-bed boilers: areview. Chem Eng Sci 1999;54(22):5547–57.
[4] Adanez J, Gayan P, de Diego LF, Garcia-Labiano F, Abad A.Combustion of Wood Chips in a CFBC. Modeling and Validation.Ind Eng Chem Res 2003;42(5):987–99.
[5] Grace JR. Fluidized beds as chemical reactors. In: Geldart D, editor.Gas Fluidization Technology. New York: John Wiley and Sons;1986.
[6] Grace JR, Clift R. The two phase theory of fluidization. Chem EngSci 1974;29(2):327–34.
[7] Baeyens J. Gas fluidization, short course. University of Brad-ford: Institute of Chemical Engineering; 1981.
[8] Johnsson F, Andersson S, Leckner B. Expansion of a freely bubblingfluidized bed. Powder Technol 1991;68(2):117–23.
[9] Darton RC, La Nauze RD, Davidson JF, Harrison D. Bubble growthdue to coalescence in fluidized beds. Trans Inst Chem Eng1977;55(4):274–80.
[10] Rowe PN. Experimental properties of bubbles. In: Davidson JF,Harrison D, editors. Fluidization. London: Academic Press; 1971. p.21–189.
[11] Davidson JF, Harrison D. Behaviour of a continuously bubblingfluidized bed. Chem Eng Sci 1966;21(9):731–8.
[12] Grace JR, Harrison D. Behaviour of freely bubbling bed. Chem EngSci 1969;24(3):497–508.
[13] Sit SP, Grace JR. Interphase mass transfer in an aggregative fluidizedbed. Chem Eng Sci 1978;33(8):1115–22.
[14] Fane AG, Wen CY. Fluidized-bed reactors. In: Hetsroni G, editor.Handbook of multiphase systems. Washington: Hemisphere; 1982.p. 8-105–51.
[15] Grace JR. Modeling and simulation of two-phase fluidized bedreactors. In: Lasa Hde. Editor. Chemical reactor design and technol-ogy, Martinus Nijhoff; 1986, p. 245–89.
[16] Van Swaaij WPM. The design of gas–solids fluid bed and relatedreactors. In: Luss D, Weekman VW, editors. ACS Symp Ser 72. NewYork: American Chemical Society; 1978. p. 193–222.
[17] Zenz FA, Weil NA. A theoretical-empirical approach to themechanism of particle entrainment from fluidized beds. AIChE J1958;4(4):472–9.
[18] Loffler G, Kaiser S, Bosch K, Hofbauer H. Hydrodynamics of a dualfluidized-bed gasifier – part I: simulation of a riser with gas injectionand diffuser. Chem Eng Sci 2003;58(18):4197–213.
[19] Adanez J, Gayan P, Garcia-Labiano F, de Diego LF. Axialvoidage profiles in fast fluidized beds. Powder Technol 1994;81(3):259–68.
[20] Namkung W, Kim SD. Gas back mixing in a circulating fluidizedbed. Powder Technol 1998;99(1):70–8.
[21] Arthur JA. Reaction between carbon and oxygen. Trans Faraday Soc1951;47:164–78.
[22] Winter F, Wartha C, Hofbauer H. Characterization and emission ofsingle fuel particles under FBC condition. Third internationalconference on combustion technologies for a clean environment, 3–6 July 1995, Lisbon, Portugal.
[23] Barrio M, Gobel B, Risnes H, Henriksen U, Hustad JE, SorensenLH. Steam gasification of wood and the effect of hydrogen inhibi-tation on the chemical kinetics. Progress in thermochemical biomassconversion, September 2000, Tirol, Austria.
[24] Vilienskii TV, Hezmalian DM. Dynamics of the combustion ofpulverized fuel. energiya 1978, Moscow.
ent and validation: Co-combustion of residual ..., Fuel (2007),
499500501502503504505506507508509510511512513
514515516517518519520521522523524525526527
P. Kaushal et al. / Fuel xxx (2007) xxx–xxx 9
JFUE 3518 No. of Pages 9, Model 5+
17 April 2007 Disk UsedARTICLE IN PRESS
[25] Dryer FL, Glassman I. High-temperature oxidation of CO and CH4.Fourteenth international symposium on combustion, The Combus-tion Institute: Pittsburgh, PA, 1973, p. 987–1003.
[26] Weimer A, Clough D. Modeling a low pressure steam oxygenfluidized bed coal gasifying reactor. Chem Eng Sci 1981;36(3):548–67.
[27] Zimont VL, Trushin YM. Total combustion kinetics of hydrocarbonfuels. Combust, Explosion Shockwaves 1969;5(4):567–73.
[28] Van der Vaart DR. PhD Thesis, University of Cambridge, 1985,England.
[29] Burcat A, McBride B. Ideal gas thermodynamic data for combustionand air pollution use. Technion isreal Institute of Technology,Aerospace Engineering Report, 1997, TAE 804 (Gar-field.chem.elte.hu/Burcat/burcat.html).
[30] Barin I. Thermochemical data of pure substance. 2nd ed. Ger-many: VCH; 1993.
UN
CO
RR
EC
T
Please cite this article in press as: Kaushal P et al., Model developmdoi:10.1016/j.fuel.2007.03.032
F
[31] Merrick D. Mathematical models of the thermal decomposition ofcoal part 1–6. Fuel 1983;62(5):534–70.
[32] Boie W. Vom Brennstoff zum Rauchgas – FeuerungstechnischesRechnen mit Brennstoffkenngroßen und seine Vereinfachung mitMitteln der Statistik, Teubner, Leipzig, 1957.
[33] Yan HM, Craig H, Zhang D. Mathematical modelling of a bubblingfluidized bed coal gasifier and the significance of net flow. Fuel1998;77(9/10):1067–79.
[34] Amand L, Lyngfelt A, Karlsson, M, Leckner B. Fuel loading of afluidized bed combustor burning bituminous coal, peat or wood chips.Report A97-221 1999; Deparment of Energy Conversion, ChalmersUniversity of Technology: Goeteborg, Sweden. (<http://www.entek.chalmers. se/~leam/Input_model/Fuel_loading1/Charload_rev1.pdf>).
ED
PR
OO
ent and validation: Co-combustion of residual ..., Fuel (2007),