Process simulation of a SOFC and double bubbling fluidized bed gasifier power plant

transcript

ww.sciencedirect.com

i n t e rn a t i o n a l j o u r n a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 3 2e5 4 2

Available online at w

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

Process simulation of a SOFC and double bubbling fluidizedbed gasifier power plant

Andrea Di Carlo a, Enrico Bocci b,*, Vincenzo Naso c

aDepartment of Chemistry, Chemical Engineering and Materials, University of Aquila, Via Campo di Pile, 67100 L’Aquila, ItalybEnergy and Mechanical Engineering Department, Marconi University of Rome, Via Virgilio 8, 00193 Rome, ItalycMechanical and Astronautical Engineering Department, Sapienza University of Rome, Via Eudossiana, 18, 00184 Rome, Italy

a r t i c l e i n f o

Article history:

Received 16 March 2012

Received in revised form

12 August 2012

Accepted 10 September 2012

Available online 12 October 2012

Keywords:

Fluidised bed biomass steam

gasification

Solid oxide fuel cells

Process model

* Corresponding author. Tel.: þ39 3288719698E-mail addresses: bocci@uniroma1.it, e.b

0360-3199/$ e see front matter Copyright ªhttp://dx.doi.org/10.1016/j.ijhydene.2012.09.0

a b s t r a c t

The development of reliable fuel cells power plant based on renewable fuels stands out as

one of the promising energy systems solutions for the future. Indeed fuel cells can increase

the efficiency and the cleaning of the electrical energy production from renewable fuels.

Process simulations of advanced power plants fed by low cost renewable fuels like biomass

waste are a key step to develop renewable resources based on high temperature fuel cells

applications. The aim of this work is to predict the component behaviour of a specific

power plant mainly composed of a small indirectly heated gasifier and a Solid Oxide Fuel

Cell (SOFC) and fed by chestnut coppice, waste available in great quantity in Central Italy,

as well as in several other European regions. The plant’s thermodynamic behaviour is

analysed by means of the process simulator CHEMCADª in which particular models for the

SOFC and the gasifier have been developed in FORTRAN by the authors and then interfaced

to commercial software. The results of the predictive model are presented and discussed,

showing the possibility of an extremely interesting “carbon neutral” small plant configu-

ration with high electrical and global efficiency exclusively based on the use of low cost

renewable resources.

Copyright ª 2012, Hydrogen Energy Publications, LLC. Published by Elsevier Ltd. All rights

reserved.

1. Introduction in the low-medium range following the low energy density of

The realization of many international and national strategic

plans, based upon renewable energy sources and hydrogen/

fuel cells plans, demonstrates the increasing interest in the

promotion and implementation of methods, technologies and

processes for the development of sustainable energy systems

[1e3].

In order to exploit biomass as a major source of energy in

the power generation and transport sectors, there is a need for

high efficient and clean energy conversion devices, especially

; fax: þ39 06233296906.occi@unimarconi.it (E. Bo2012, Hydrogen Energy P59

this fuel [4,5].

Large installations, based on boiler coupled to steam

turbine (or Integrated Gasification Combined Cycle power

plant, IGCC), are too complex at small scale, meanwhile small

biomass gasifiers coupled to Internal Combustion Engines,

ICE, have low electrical efficiency (15e30%) and generally not

negligible emissions [6].

High temperature fuel cells represent the most promising

technologies for achieving higher conversion efficiency and

reducing emissions especially at small scale. Due to its higher

i n t e r n a t i o n a l j o u rn a l o f h y d r o g e n en e r g y 3 8 ( 2 0 1 3 ) 5 3 2e5 4 2 533

power density and operating temperature, Solid Oxide Fuel

Cells (SOFCs) are considered as the major candidate for power

generation and cogeneration units [7].

In the last years many studies have assessed operating

conditions, system performance, potentials and limits of

a variety of SOFC power plant including the hybrid Gas

Turbine SOFC plant and the integration of SOFC systems with

biomass gasification where the SOFC electrical efficiency, the

total thermal efficiency and the total electrical efficiency can,

ideally, reach the values of 42%, 38% and 62% respectively,

based on the low heating value of the biomass [8e10].

This paper deals with a specific power plant configuration,

based on a particular small indirectly heated fluidised bed

gasifier, high temperature fuel cells (SOFC) and micro Gas

Turbine (mGT). In particular the gasifier is based on UNIQUE

concept. UNIQUE consists of a compact a gasifier integrating

the fluidized bed steam gasification of biomass and the hot gas

cleaning system into one reactor vessel, by means of a bundle

of ceramic filter candles that operates at high temperature

(800e850 �C) in the gasifier freeboard. Such configuration

produce a syngas free of tar and sulphur compounds and

allows for remarkable plant simplification and reduction of

costs [11e13].

The analysis is based on process models that have been

developed by the authors in earlier works [14e16] where they

carried out process simulations of MCFC system integrated

with biomass gasification and hot gas cleaning. Black-box and

empirical models were used for the gasification process.

Theplant’s thermodynamicbehaviour isanalysedbymeans

of the process simulator CHEMCADª. The plant operation is

optimized in terms of energy management, including cogene-

ration. The models for the SOFC and the gasifier have been

developed in FORTRAN by the authors and then interfaced to

thecommercial softwareCHEMCADª. Indeeddifferent typesof

models can be developed, from complex fluid dynamics

models, to simpler black-box or zero-dimensionmodels, but to

really predict thebehaviour of a complex system like a biomass

gasifier under different conditions the fluid dynamics models

are the best developedmodels up to date [17].

Finally a sensitivity study of the power produced was

carried out varying the moisture content in the biomass from

10 to 30%.

Regarding the error analysis, the mass and energy

conservation principles are inherently satisfied by the soft-

ware and the model, while the results are validated by

comparing simulated with experimental data.

2. Power plant description

The proposed power plant mainly consists of a particular

gasifier that produces a woodgas to feed an SOFC while the

high temperature residual heat is used in a Capstone C15mGT

[18], to produce further electrical power.

Fig. 1 shows the CHEMCADª plant flowsheet. The incoming

biomass is gasified only by steam in the small indirectly heated

fluidized bed gasifier (100 kWth). The steam is generated by the

residual heat of themGT exhausts. From the gasifier (Stream 1)

a woodgas is obtained at 800 �C. The char and bed material are

recirculated (Stream 2) in the burner of gasifier to produce the

process heat. The woodgas is cleaned up from tars and

particulate directly in the freeboard of the gasifier. This can be

obtained by placing a bundle of catalytic ceramic candles in the

gasifier freeboard. These candles convert tars by steam

reforming and remove particulate at a temperature as high as

the gasification temperature (800e850 �C). The cleaned wood-

gas is utilized in the SOFC module to produce the electrical

power. The anode exhausts, still rich of H2, CO and CH4, are

burned with residual char in the burner of gasifier, to produce

the necessary heat for the gasification process. The hot flue

gases from burner of gasifier at 950 �C (Stream 3) are exploited

to heat the compressed air for the turbine, to obtain further

power from the mGT.

In order to simulate the power plant SOFC and steam-

gasifier specific models have been developed and described

in the following paragraphs. The remaining components of

the plant were simulated using conventional CHEMCAD

blocks, in particular the catalytic reforming downstream

gasifier was simulated by using a Gibbs reactor, while for the

combustor of the gasifier a stoichiometric reactor was used.

3. Gasifier model

The gasifiermodel receives as input the pyrolysis products and

calculates the woodgas composition. Indeed in fluidized bed

gasifier the pyrolysis reactions can be considered as flash

pyrolysis; the time necessary to reach the final products of

pyrolysis canbeneglected. For this reason, itwasassumed that

the fuel feeding the gasifier is that obtainable frompyrolysis of

wood at the same operative conditions. The experimental

results of Jand and Foscolo [19] were used in this study.

The products of pyrolysis are composed of CO2, CO, H2O,

H2, and CH4, light and heavy hydrocarbons (tar) and char, as

showed in Table 1.

The gasification reactions considered in this study are:

C þ H2O / CO þ H2 (R1)

C þ CO2/2CO (R2)

C þ 2H2 / CH4 (R3)

CH4 þ H2O 4 CO þ 3H2 (R4)

CO þ H2O ;4 CO2 þH2 (R5)

C10H8 þ 10H2O ;4 10CO þ 14H2 (R6)

Napthalene have been used as tar representative.

Regarding the reactor fluid dynamics the Kunii and Lev-

enspiel bubbling bed model [20] has been used. An outline of

their model employed in this study is shown in Fig. 2.

Fig. 1 e Plant flowsheet.

In the model, the fluidized bed consists of two regions,

bubble and emulsion, interacting each other through one

interchange coefficient of gas, kbe, and several assumptions

are employed as follows. The wake and cloud region is

neglected. An intermediate particles model for Geldart B

particle (dp ¼ 0.1e1 mm) was selected for the reactor model-

ling. Moreover the following equations have been used.

In the emulsion phase, gas ascends at the minimum

fluidization velocity, umf:

umf ¼h�27:22 þ 0:0408$Ar

�0:5�27:2i$

dprgas

In the bubble phase, bubble gas ascends at the velocity of ub:

�z� ¼ 0:71

ffiffiffiffiffiffiffiffigdb

q� �QðzÞ=A� umf

�The bubble diameter is calculated by Davidson model at

each bed height:

Table 1 e Pyrolysis (devolatilization) products.

Char kg/kgbio,daf 0.187

Gas kg/kgbio,daf (N m3/kg bio,daf) 0.763 (0.9)

Tar kg/kgbio,daf 0.05

Gas mole fraction

H2 0.285

H2O 0.035

CO2 0.11

CO 0.38

CH4 0.19

Tar composition

C10H8 1

�z� ¼ 0:54

�QðzÞ=A� umf

�0:40:2

"zþ 4

ffiffiffiffiffiffiffiffiA

s #0:8

The volume fraction of bubble in the bed is d (while that of

emulsion is 1 � d):

d�z� ¼ QðzÞ=A� umf

uBðzÞGases ascend as plug flow in each phase and are exchanged

at the rate of gas interchange coefficient kbe:

�z� ¼ umf

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi4εmfDrub

�z�

pdBðzÞ

With these assumptions, the transport equations at steady

state are:

vz½QðzÞCtot� ¼

�Rbi þ Rs

ei þ Rgei

��Q�z�� umfAs

�Cbi

� ¼ kbeAsðCbi � CeiÞ 6dbdþ εmfAsd

nijRgbj

��umfAs

�Cei

� ¼ kbeAsðCei � CbiÞ 6db

ð1� dÞ þAsð1� dÞ24�1� εmf

�εc

nijRcej þ εmf

nijRgej

þ �1� εmf

�εolivni6R

olive6

35 _mout

¼ _minchar �Asð1� dÞεcrc

nijRsejεc ¼

c þ _Vin

¼ _minc

rcþ _min

where j are the reactions 1,., 6, i are the chemical species H2,

., C10H8 and nij are the stochiometric coefficients of species i

in reaction j, negative for reagents and positive for products.

Fig. 2 e Kunii and Levenspiel fluidized bed reactor model.

Table 3 e Results of simulation compared with

Finally ideal gas law was used to calculate the gas

concentration:

Ctot ¼ pRT

To complete the model, the kinetic expressions for the

reaction rates must be set. Extra and intra particle diffusion

resistances were neglected owing to the small particles

diameter. For reactions (R1) and (R2) (gasification of char with

steam and CO2) the expressions deducted by Barrio et al. [21]

and Barrio and Husted [22] have been used.

R1 ¼ kw1pH2O

1þ ðkw1=kw3ÞpH2O þ ðkw2=kw3ÞpH2

ð1=sÞ

R2 ¼ kc1pCO2

1þ ðkc1=kc3ÞpCO2þ ðkc2=kc3ÞpCO

ð1=sÞ

The values of k coefficients, shown inTable 2,were deducted

from the work of Konttinen et al. [23].

Reaction (R3) was neglected because the low operative

pressure and the slow rate of reaction. Gas-phase reactions

kinetic (R4, R5) were simulated using the expressions adopted

by Wang and Kinoshita [24]

R4 ¼ 2:79e�12;579=RT

CCOCH2O � CCO2

KWGSeq

! �mol=m3 s

R5 ¼ 1:2863e�36150=RT

CH2O � C3H2CCO

KREFeq

! �mol=m3 s

Finally for the heterogeneous reaction (R6), the expression

proposed by Espenas and Waldheim [25], was used:

R6 ¼ roliv1:46� 1012

3600e�321;000=RTpC10H8

$p�0:44H2

$p�0:56H2O

�mol=m3 s

Table 2 e Arrhenius and activation energy values.

k01(s�1 bar�1)

k02(s�1 bar�1)

k03(s�1)

EA1(J/mol)

R1 6.49e7 95.3 1.64e9 204,000 54,315 243,000

R2 1.64e7 4.59e2 8.83e7 188,000 88,265 225,000

3.1. Model validation

In order to validate the model, different simulations were

carried out and compared with literature results. Gasification

temperature has been taken equal to 850 �C, the steam to

biomass ratio 0.5, the superficial velocity has been set two

times theminimum fluidization velocity, the static bed height

60 cm.

Table 3 shows the comparisonof the results obtainedby the

model with the experimental data declared by Hofbauer [26].

The results obtained by simulation are in fair agreement

with that obtained experimentally on a similar gasifier except

for the concentration of tars that results higher in the model.

The main reason is probably due to the choice of tar repre-

sentative for the simulation; naphthalene is one of the most

stable tar compounds and is difficult to decompose. Anyway,

for the scope of the work the results were considered

acceptable.

Fig. 3 shows the composition of the gas (dry basis) at

different height of the bed as results of the model. Hydrogen

fraction increases from 0.285 (fraction obtained by pyrolysis)

to 0.42 reaching a plateau. This increase is a consequence of

the char gasification reaction (R1) as also of water gas shift

reaction (R5) and steam reforming of hydrocarbon ((R4)e(R6)).

CO reaches a negative peak and then it slightly increases

again. This increase is probably due to the slower reforming

reactions ((R4)e(R6)) that continuously produce CO and H2 but

require much more residence time (and thus bed height) to

reach the equilibrium conditions, if compared to thewater gas

shift reaction. The reduction of steam and the not favored

equilibrium conditions at 800 �C for the water gas shift reac-

tion are also further reasons (Fig. 4).

Finally the cold gas efficiency of the gasifier was calculated

hcold ¼ LHVH2_mH2

þ LHVCO _mCO þ LHVCH4_mCH4

LHVBIO _mBIO

where LHV is the low heating value (kJ/N m3 or kJ/kg) of each

compoundwhile _m is its flowrate (Nm3/s or kg/s). The cold gas

efficiency of the gasifier is around 90%. It is worth stressing

here that this result does not consider the required additional

fuel that must be burned in the combustor of the gasifier (see

Fig. 1); combustion of the residual char from gasification is not

sufficient to supply the required heat for the gasification

process and additional fuel is necessary, in this way the cold

gas efficiency would be lower. This problem is solved burning

the residual fuel outgoing the anode of the SOFC. The

experimental data.

Results Simulation Experimental

H2 42% 35e45%

CH4 13% 8e12%

CO 22% 20e30%

CO2 21% 15e25%

Dry gas/kg biomass daf 1.22 1e1.5

Tar 12 g/N m3 1.5e4.5 g/N m3

Fig. 3 e Composition of the produced gas (dry basis) during

gasification vs. the bed height.

necessary amount of residual fuel affects the fuel utilization

in the SOFC and thus also the efficiency of the entire system: if

the residual fuel from the SOFC is not enough for the gasifi-

cation process, the fuel utilization must be reduced,

increasing the fuel for the combustor but decreasing the SOFC

(and thus system) efficiency.

4. SOFC model

In this paragraph the model developed for a SOFC is illus-

trated. The fuel cell model is taken one-dimensional on the

horizontal cell layer, while temperature variations along the

vertical coordinate are neglected. The model geometry was

divided in three distinct zones: a planar solid zone (S)

comprehensive of the two electrodes, the bipolar plate and the

electrolytic matrix invested by the two counter-flow gaseous

stream (Anodic A and Cathodic C).

The model is based on the following hypotheses:

1. steady state conditions;

2. adiabatic conditions;

3. no radiation heat exchanges between solid components

and gas streams;

4. continuous description of the gas flow (distributed into

a number of discrete channels) in terms of a specific rate of

reactants per unit length of the fuel cell side;

5. fully developed velocity and temperature profiles in the gas

streams;

6. plug-flow balance equations for the gas streams where gas

species diffusion on gas phase is neglected;

7. the rate of the electrochemical reaction has been calculated

on the basis of Faraday’s law;

8. water gas shift reaction was considered at equilibrium;

9. reaction rate for steam reforming of CH4 was calculated

using the expressions proposed by Achenbach and Rien-

sche [27]:

_sref ¼ 4274pCH4exp

��82; 000RT

� �mol=m2 s

4.1. Electrochemical model

The difference between the thermodynamic potentials of the

electrode reactions determines the reversible cell voltage or

open-circuit potential, This open-circuit potential is a local

quantity, as it depends on the gas composition and tempera-

ture at the electrodes, and can be determined by the Nernst

equationwritten for theelectrochemical oxidationofhydrogen

E ¼ E0 � RTs

where E is the maximum theoretical potential that can be

achieved by a fuel cell. As current is drawn from a fuel cell, the

cell voltage falls due to internal resistances (Ohmic) and

overpotential losses. Electrode overpotential losses are asso-

ciated with the electrochemical reactions taking place at the

electrode/electrolyte interfaces and can be divided into

concentration and activation overpotentials.

Due to these internal losses the real voltage V obtainable

from a fuel cell can be calculated as:

V ¼ E0 ��RUJþ hconc;a þ hconc;c þ hact;a þ hact;c

4.2. Activation polarization

The activation polarization is the result of the kinetics

involved with the electrochemical reactions. Each reaction

has a certain activation energy barrier that must be overcome

in order to proceed and this barrier leads to the polarization.

This energy barrier is called the activation energy and results

in activation or charge-transfer polarization, which is due to

the transfer of charges between the electronic and the ionic

conductors. The activation polarization may be regarded as

the extra potential necessary to overcome the energy barrier

of the rate-determining step of the reaction to a value such

that the electrode reaction proceeds at a desired rate. Acti-

vation polarization is normally expressed by the Butlere

Volmer equation [28]:

J ¼ J0

�bneFhact

�� exp

� ð1� bÞneFhact

��

where b is the transfer coefficient and J0 the exchange current

density. The transfer coefficient is usually 0.5 for the fuel cell

applications. The exchange current density is the forward and

reverse electrode reaction rate at the equilibriumpotential; the

higher the exchange current density and the electrochemical

reaction rate, the better fuel cell performance can be expected.

If b is set to 0.5 the equation can be rewritten as follows:

J ¼ 2J0sinh

�neFhact

And thus

hact ¼2RTneF

sinh�1

�J2J0

In order to evaluate hact for the two electrodes J0 must be

defined. The following form was adopted:

J0;electrode ¼ RTneF

kelectrodeexp

��Eelectrode

Data used by Aguiar et al. [29] were adopted for k and E

whose values are reported in Table 4.

4.3. Concentration polarization

The rate of mass transport to the reaction sites in a porous

electrode of an SOFC can be described by the diffusion of gases

in the pores. Concentration polarization results from restric-

tions to the transport of the fuel gases to the reaction sites.

This usually occurs at high currents because the forming of

product water blocks the reaction sites. This polarization is

also affected by the physical restriction of the transfer of

a large atom, oxygen, to the reaction sites on the cathode side

of the fuel cell. Diffusion through the porous material is

typically described by either ordinary or Knudsen diffusion

and has been found to play an important role in catalytic

reaction. Ordinary diffusion occurs when the pore diameter of

the material is large in comparison to the mean free path of

the gas molecules. Molecular transport through pores which

are small in comparison to the mean free path of the gas

is regarded as a Knudsen type of diffusion. According to

hConc;c ¼ �RT4F

"�pc=dO2

�� pc=dO2

�� pfO2

�exp

�ðRT=4FÞ�dO2lc=Dcathode;effpc

�J�

Knudsen diffusion, molecules collide more frequently with

the pore walls than with other molecules. Upon collision, the

atoms are instantly adsorbed on to the surface and are then

desorbed in a diffusive manner. As a result of frequent colli-

sionswith thewall of the pore, the transport of themolecule is

prevented. The Knudsen diffusion coefficient can be predicted

using kinetic theory by relating the diameter of the pore and

the mean free path of the gas.

Molecular diffusion Dm( g) is calculated using the Wilke

equation [30]:

DA ¼ 1PAsB

The binary diffusion coefficient Dmn( g) has been calculated

using the ChapmaneEnskog equation:

DAB ¼ 0:0018583

ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiT3

�1MA

þ 1MB

ps2ABUAB

where the expressions to calculate the required parameters

can be found in Bird [31]. The effective diffusion coefficient in

the particle (DA,eff) is calculated by the following equation:

DA;eff ¼ DA0ε

Table 4 e Activation overpotential data.

kcathode 2.35e11 U�1 m�2 Ecathode 137 kJ mol�1

kanode 6.54e11 U�1 m�2 Eanode 140 kJ mol�1

DA0 was calculated using Bosanquet equation as reported in

Hayes [32]

DA;eff ¼ 11DA

þ 1DkA

DA was calculated with Wilke equation while DkA is the

Knudsen diffusion, calculated as:

DkA ¼ 23rav

ffiffiffiffiffiffiffiffiffiffiffi8RTpMA

To calculate hconc, a relation between the partial pressures

of H2, H2O, and O2 at the three-phase boundaries and the

current density is necessary. Different porous-media gas-

phase transport models have been developed to predict

concentration overpotentials [33e36]. In this work the model

developed by Chan et al. [33] was adopted.

hConc;a ¼ �RT2F

" ð1� RT=2FÞ la=Danode;effpf

ð1þ RT=2FÞ la=Danode;effpf

where pfH2; pf

H2O;pf

O2are the partial pressure of the chemical

species involved in the reaction in the flow phase, thus at the

boundary of the electrodes (anode, cathode) and la and lc the

thickness of anode and cathode respectively.

Danode,eff and Dcathode,eff are computed as

Danode;eff ¼ pfH2O

!DH2 ;eff þ

!DH2O;eff

Dcathode;eff ¼ DO2 ;eff

Finally dO2is calculated as:

dO2¼ DkO2 ;eff

DkO2 ;eff þ DO2N2 ;eff

4.4. Ohmic polarization

Ohmic losses occur because of resistance to the flow of ions in

the electrolyte and resistance to flow of electrons through the

electrode materials. The dominant ohmic losses, through the

electrolyte, can be reduced by decreasing the distance of

electrode separation and enhancing the ionic conductivity of

the electrolyte. The ionic flow in the electrolyte obeys Ohm’s

law, thus the ohmic losses can be expressed by the equation:

hU ¼ RUJ

RU ¼X

i¼a;e;c

Table 5 e Main parameters adopted for the validation ofthe SOFC model.

Cell length, L 0.4 m

Cell width, W 0.1 m

Anode thickness 500 mm

Cathode thickness 50 mm

Electrolyte thickness 20 mm

Fuel feed Completely reformed CH4 with S/C ¼ 2

Fig. 4 e Cell potential depending on current density, H2 flux

360 ml/min, air flux 2 l/min, 50 cm2 active area [37].

where li is thickness of anode, cathode and electrolyte while si

is the anode, cathode and electrolyte electronic conductivity.

4.5. Transport equations

4.5.1. Massespecies balanceAnode:

d�ugCi

vij _sj1

where Ci is the molar concentration of species i, ug is the gas

velocity, wa is the height of the fuel channel and j are the

reactions that occur in the anode:

1. _sel ¼ J=neF for electrochemical reaction.

2. _sWGS WGS reaction was considered at equilibrium.

3. _sref ¼ 4274pCH4expð�82; 000=ðRTÞÞ ðmol=m2 sÞ [27].

Cathode:

d�ugCi

¼ _sel1wc

4.5.2. Energy balanceAnode:

ugd�cpCiTa

¼ haðTs � TaÞ 1wa

Cathode:

ugd�cpCiTc

¼ hcðTs � TcÞ 1wc

Solid layer:

hcðTs � TcÞ þ haðTs � TcÞ ¼ l$s

�v2Ts

�þ Qreac

where Qreac ¼Pj

_sjDHj � VJ.where cp is the specific heat of the

species i, T is the temperature of anode gas (a), cathode gas (c),

solid zone (s), h is the convection coefficient between gas and

solid, l is thermal conductivity of the solid, s is the thickness

of the solid, Qreac is the heat of reactions, and VJ is the electric

power produced per unit area.

The proposed equations system was solved using a finite

difference method with relaxation writing a sub-routine in

Fortran 90.

4.6. Model validation

The results obtained by Autissier et al. [37] and those of Aguiar

et al. [29] were used as reference for the validation of the SOFC

model. In the work of Autissier et al. [37] a button cell anode

supported SOFC were tested to validate a CFD model. The

button cell used had an active area of 1 cm2. Cells are anode

supported, 200 mm thick with a 6e8 mm thick electrolyte layer.

The results available in [37] were obtained with a flow of

360 ml/min of pure Hydrogen at the anode and a flow of air at

the cathode of 2 l/min. The temperature of the cell was set to

750 �C (Fig. 4).

Even if the general trend of the curves is similar, the

experimental open-circuit voltage is still lower than the

calculated one of about�10MV. Similar results were observed

by [37] comparing the results of their model with the same

experimental data. Autissier et al. justified this difference

by omitting to calculate back-diffusion through outlets in the

model, so modifying the partial pressure near the outlet and

thus the OCV. This diffusion could lead to the observed losses.

Another difference between the results obtained by this work

and the experimental data of [37] is the different slope of the

two curves. The main reason of this difference can be attrib-

uted at the different polarization curves adopted in this work

compared with [37]. A much more accurate calibration of the

model with experimental data would be required to reduce

this small error but this is beyond the presented work’s scope.

Fuel composition plays a significant role in fuel cell

performance and further validation of the model is required

when the gas has no negligible concentration of CO, CO2 and

CH4. In order to verify the model with a syngas containing not

negligible concentration of CO in the gas the results of Aguiar

et al. [29] were also used as reference. In the work of Aguiar

et al. a numerical model for anode supported planar SOFC

were developed and tested with a syngas obtained by steam

reforming of CH4. The same conditions adopted by [29] were

utilized in order to compare the results of Aguiar and those of

this work. Table 5 shows the most important values adopted.

Fig. 5 shows the power density obtained from the model

compared with those from Aguiar et al. [29] at different

operative temperatures and current densities. The results

Fig. 5 e Cell power density depending on current density,

results of the model compared with that obtained by

Aguiar et al. [29] using syngas from steam reforming of

obtained by the model are in agreement with that available in

literature, small differences can be detected only at high

current densities. As expected, the performance of a cell is

significantly hindered when the operating temperature is

decreased: the maximum power density reduces from

840 W m�2 at 800 �C, to 303 W m�2 at 700 �C.As the fuel flows through a SOFC fuel channel, hydrogen

and carbon monoxide are consumed by the electrochemical

reactions and by the WGS reaction and the fuel stream

becomes diluted, being less rich close to the exit of the

channel. Therefore, in this region, concentration over-

potentials become more significant and result in the decrease

of SOFC output.

Fig. 6 illustrates the predicted cell voltage as a function of

current density for a fully reformed fuel mixture with 10%,

75%, and 90% fuel utilisation. For low fuel utilization (10%)

concentration overpotentials are not significant. As seen in

Fig. 6, for this case, the voltage versus current density curve

does not present a concave curvature for high current densi-

ties. However, for 75% and 90% fuel utilisation, the concavity

Fig. 6 e Cell voltage as a function of current density at

800 �C for a fully reformed fuel mixture with 10%, 75%, and

90% fuel utilisation.

becomes visible and voltage and power density drop more

rapidly to zero, due to concentration overpotentials in the

anode. It is clear from the Figure that the curve loses its

linearity around 2.25 A/cm2 for fuel utilization equal to 75%

and 2 A/cm2 for fuel utilization equal to 90%.

Finally the influence of gas composition on the cell voltage

was verified. Three different simulations were carried out, the

first with pure H2 humidified with 10% of steam, the second

with a syngas obtained by steam reforming of CH4 with SC ¼ 2

(H2 ¼ 64%, H2O ¼ 16%, CO2 ¼ 5%, CO ¼ 14% CH4 ¼ 1% vol/vol),

the third with a reformed woodgas that can be obtained by

biomass gasification (see Table 5). Total amount of H2 þ CO

was maintained constant (per each current density) for the

three cases. Fuel utilization was maintained constant at 70%

(Fig. 7).

As expected the decrease of H2 fraction and the increase of

steam and other diluents (like CO2) decrease dramatically the

output voltage at each current density. Concentration resis-

tances effect due to diffusion is evident for the woodgas: the

typical concavity become visible at 1.5 A/cm2, while it is

negligible for the pure H2 case. As expected the use of pure

hydrogen would improve the efficiency of the cell but the

necessary devices to purify hydrogen from other contaminant

would reduce the entire system efficiency. Therefore the use

of woodgas could be anyway a reasonable choice.

5. Catalytic filter candles

Ceramic filters are mainly used for removal of dust from hot

gases produced by industrial processes such as combustion

and gasification. Recently, advanced particle filtration

systems using ceramic candle filters for cleaning hot gases

have been developed by Pall Schumacher GmbH [38]. The

candle filters are hollow cylinders (closed on one end) typi-

cally with a diameter of about 6 cm and a length of 1e1.5 m;

the outer surface is made of a thin, micro porous layer that

forms the actual filtering surface. The dusty feed gas flows

outside such cylinders and percolates through their porous

structure driven by a differential pressure: fine particles

accumulate on the filtering surface building up a cake, while

the clean gas is collected in the hollow space. A filter vessel

Fig. 7 e Cell voltage vs current density for different gas

composition.

Table 7 e Comparison of the results obtained by themodel with those of [42,43].

Model Rapagna et al [42,43]

Gas yield (N m3/kg daf) 1.53 1.49e2.09

Composition mole frac (dry) mole frac (dry)

H2 0.53 0.50e0.56

CO2 0.18 0.19e0.23

CO 0.22 0.19e0.22

CH4 0.07 0.02e0.05

Tar (g/N m3) 0.0 0.15e0.91

for a commercial power plant requires a large number

(several hundreds) of candle filters arranged in several clus-

ters. Such clusters are periodically cleaned by an instanta-

neous reverse-gas flow back-pulse procedure to remove the

dust cake that builds up on the filtering surface. In addition,

a high catalytic capacity can be achieved by simply filling the

free hollow cylindrical volume of the filter element with

catalyst particles of a high active surface area. Several tar

reforming catalysts with different NiO loadings and different

catalyst support materials have been tested with synthetic

gases, resulting in complete conversion of naphthalene at

800 �C [39].

The catalytic filter candles of UNIQUE system were simu-

lated using two blocks of the CHEMCADª library:

1. a separator that removes all the fines particles;

2. a Gibbs reactor downstream the baghouse filter.

The experimental tests described in the work of Di Carlo

and Foscolo [40] have shown that the cake on the candle

surface increases continuously its thickness without particle

shedding and no fine particles could be detected upstream the

filter. Consequently it was assumed in the model that all

particles reaching the filter candles are “trapped” in the cake

and removed from the gas stream. Thanks to the back-pulse

cleaning system the pressure drops due to the cake forma-

tion are of the order of 50 mBar and thus they were neglected

in the simulation.

Experimental works developed by Rapagna et al. [41e43]

have shown that tars can be almost completely reformed

thanks to the catalytic filter candles, also steam reforming of

methane can reach equilibrium. For this reason a Gibbs

reactor was considered a good choice to simulate the catalytic

reaction occurring in the candles.

6. Results of the entire system simulation

Finally a sensitivity analysis of the power produced by the

power plant has been carried out, by varying the cathode gas

inlet temperature (and thus also the operative temperature of

the fuel cell) and varying the moisture content of the feeding

biomass. The composition of gas at anode inlet (and thus at

the catalytic filter candles outlet) is reported in Table 6.

Table 6e Composition of gas at anode inlet (catalytic filtercandle outlet).

T (�C) 620

p (bar) 1

Total flow (kg/h) 28.5

Composition mole frac (wet)

H2 0.44

H2O 0.17

CO2 0.15

CO 0.18

CH4 0.06

In order to partially validate the results obtained Table 7

shows the composition obtained during different tests

carried out by Rapagna et al. [42,43] compared with the dry

composition obtained by the model.

Most of the results are in line with that obtained experi-

mentally. As expected the model predicted the complete

removal of tar, while the experiments showed a small residual

concentration. This result is due to the inaccuracy of the Gibbs

reactor (equilibrium reactor), anyway for the purpose of this

work the model results were considered acceptable.

Inorder tomakemorerealisticsimulations thedimensionof

the cell utilized by themodel were taken from [44,45] (Table 8).

The current density was extrapolated from the data

available in [44,45]. A value of 0.4 A/cm2 was chosen for the

preliminary dimensioning of the system. A sensitivity study

would be probably necessary to optimize the value of the

current density but it is beyond the scope of this work.

Number of cells was chosen in order to have fuel utilization

around 0.75 with the current density chosen, flowrate and

composition of Table 6. Oxidant flowrate ratio was set equal to

10 in agreement with data also available in [44]. Once the

number of cells was set, current density was varied around

0.4 A/cm2 to obtain different fuel utilization (and thus

different amount of residual fuel from anode) as required by

the gasification process.

A sensitivity study of the power produced was carried out

varying the moisture content in the biomass from 10% to 30%.

The steam to carbon ratio wasmaintained constant and equal

to 0.5. This means that when the moisture is increased the

necessary external steam amount was reduced. Figs. 8 and 9

show the results obtained by the simulations.

One can observe that the electrical efficiency decreases

when the moisture content increases (from 48% to 34%). The

main reason is due to the greater amount of heat that must be

supplied from the burner to the gasifier. Indeed, in order to

vaporize the moisture of the biomass, a greater amount of

Table 8 e Dimensions of the cell used in the simulations[44,45].

Cell active area 200 cm2

Single cell power 60 W

Anode thickness 240 mm

Cathode thickness 40 mm

Electrolyte thickness 8 mm

Fig. 8 e SOFC power vs cathode inlet temperature and

moisture content.Fig. 10 e Temperature distribution inside the cell for

different cathode inlet temperatures moisture 10%.

woodgas must be circulated to the burner, reducing the

amount exploitable in the SOFC for electrical power produc-

tion (reduction of fuel utilization). The opposite is for the

cogeneration efficiency, because lower amount of heat is

required from the exhausts gases to generate steam.

Simulations showed that the best case occurs with

a temperature of the cathode gas of 800 �C and moisture of

10%, in this case the fuel utilization could be set equal to 0.79

and the electrical efficiency of the entire system is 48%. In the

worst case (temperature of the cathode gas of 650 �C and

moisture 30%) fuel utilization must be reduced to 0.73. In this

case the electrical efficiency of the entire system is 34%.

Fig. 10 shows the temperature distribution in the cell for

different anode and cathode inlet temperatures, for the case

of 10% of moisture.

As depicted the maximum temperature difference are

limited and lower than 100 �C. The worst case is that obtained

at 650 �C where the maximum temperature difference was

93 �C while the best was for the case at 800 �C with

a maximum difference of 74 �C. The main reason of these

results is explained by the higher electrical efficiency that

increases with the increase of operative temperature:

a smaller fraction of the incoming energy is dissipated in heat.

Fig. 9 e Power plant electrical efficiency (SOFC D mGT) vs

cathode inlet temperature and moisture content.

7. Conclusion

The paper deals with a new small cogeneration system con-

sisting of a fluidised bed gasifier, coupled to a SOFC and

a mGT. A detailed model for the gasifier and the SOFC has

been discussed and validated. The simulation showed higher

stack power production (29e42 kWe) and higher electrical

efficiency (34e48%). Thus the proposed coupling of a recircu-

lated fluidized bed gasifier and a SOFC/mGT system presents

conversion efficiency higher than those reached by the stan-

dard biomass power plants even at bigger size. Considering

the realistic model (e.g. overestimation of tar formation) and

the low performance of the biomass as fuel, the performance

obtained look fine even if are below the performance obtained

in bigger or traditional fossil fuel (natural gas) power plants.

Another related important innovation is the feeding of the

mGT via the high temperature flue combustor gases that

allowed a better mGT operation producing other electrical

power. As the most of the heat is recovered from exhausts at

quite high temperature (400 �C), it could be also used in

Organic Rankine Cycle to further increase the electrical power

produced and thus improving electrical efficiency of the plant.

Moreover, the very low environmental impacts make this

solution particularly suitable for distributed energy produc-

tion also in place with high environmental constraints.

r e f e r e n c e s

[1] D.G.f.E. European Commission. World energy technologyoutlook-2050 (WETO-H2). Brussels: EC Editor; 2006.

[2] Barbir F. Transition to renewable energy systems withhydrogen as an energy carrier. Energy 2009;34(3):308e12.

[3] Beurskens LWM, Hekkenberg M, Vethman P. Renewableenergy Projections [.] Covering all 27 EU Member States.ECN-E-10-069 28; 2011.

[4] Demirbas‚ A. Biomass resource facilities and biomassconversion processing for fuels and chemicals. EnergyConversion and Management 2001;42(11):1357e78.

[5] Rapagna S, Jand N, Kiennemann A, Foscolo PU. Steam-gasification of biomass in a fluidised-bed of olivine particles.Biomass and Bioenergy 2000;19(3):187e97.

[6] Dornburg V, Faaij APC. Efficiency and economy of wood-firedbiomass energy systems in relation to scale regarding heatand power generation using combustion and gasificationtechnologies. Biomass and Bioenergy 2001;21:91e108.

[7] National Energy Technology Laboratory. In: Fuel cellhandbook. 7th ed. U.S. Department of Energy; 2004.

[8] Zhaoa Y, Sadhukhanc J, Lanzinid A, Brandonb N, Shaha N.Optimal integration strategies for a syngas fuelled SOFC andgas turbinehybrid. Journal of PowerSources2011;196:9516e27.

[9] Athanasioua C, Vakouftsi E, Coutelieris FA, Marnellosb G,Zabaniotoua A. Efficiencies of olive kernel gasificationcombined cycle with solid oxide fuel cells. ChemicalEngineering Journal 2009;149:183e90.

[10] Toonssen R, Sollai S, Aravind PV, Woudstra N,Verkooijen AHM. Alternative system designs of biomassgasification SOFC/GT hybrid systems. IJHE 2011;36:10414e25.

[11] UNIQUE Cooperative Research Project, Contract N.2115177FP, Available from: www.uniqueproject.eu [accessedFebruary 2012].

[12] Foscolo PU, Gallucci K. Integration of particulate abatement,removal of trace elements and tar reforming in one biomasssteam gasification reactor yielding high purity syngas forefficient CHP and power plants. Proceedings of the 16thEuropean biomass conference and exhibition, Valencia; 2e6June 2008.

[13] Heidenreich S, Nacken M, Foscolo PU, Rapagna S.Gasification Apparatus and method for generating syngasfrom gasifiable feedstock material. U.S. patent application US2010/0223848 A1; 9 Sep. 2010.

[14] Di Carlo A, Bocci E, Dell’Era A. Comparison by the use ofnumerical simulation of a MCFC-IR and a MCFC-ER whenused with syngas obtained by atmospheric pressure biomassgasification. IJHE 2011;36(13):7976e84.

[15] Di Carlo A, Bocci E, Orecchini F. Process simulation ofa neutral emission plant using chestnut’s coppicegasification and molten carbonate fuel cells. Journal of FuelCell Science and Technology 2008;5(2). ISSN: 1550-624X.

[16] Di Carlo A, Bocci E. MCFC and microturbine power plantsimulation. Journal of Power Sources 2006 ISSN: 0378-7753.

[17] Gomeez-Barea A, Leckner B. Modeling of biomassgasification in fluidized bed. Progress in Energy andCombustion Science 2010;3:444e509.

[18] CAPSTONE mGT C15. Available from: www.capstoneturbine.com/_docs/datasheets/C15%20NatGas_331063A_lowres.pdf[accessed February 2012].

[19] Jand N, Foscolo PU. Decomposition of wood particles influidized beds. Industrial and Engineering ChemicalResearch 2005;44:5079e89.

[20] Kunii D, Levenspiel O. Fluidized reactor models. Industrialand Engineering Chemical Research 1990;29:1226.

[21] Barrio M, Gobel B, Risnes H, Henriksen U, Hustad JE,Sorensen LH. Steam gasification of wood char and the effectof hydrogen inhibition on the chemical kinetics. In:Bridgwater AV, editor. Progress in themochemical biomassconversion. Oxford: Blackwell Science Ltd.; 2001. p. 32e46.

[22] Barrio M, Hustad JE. CO2 gasification of birch char and theeffect of CO inhibition on the calculation of chemicalkinetics. In: Bridgwater AV, editor. Progress inthermochemical biomass conversion. Oxford: BlackwellScience Ltd.; 2001. p. 47e60.

[23] Konttinen J, Moilanen A, Vepsalainen J, Kallio S, Hupa M,Kurkela E. Modelling and experimental testing of gasificationof biomass char particles. Proceedings of the Europeancombustion meeting, Orleans, France; 25e28 Oct. 2003.

[24] Wang Y, Kinoshita CM. Kinetic model of biomassgasification. Solar Energy 1993;51.

[25] Espenas BG, Waldheim L. Advanced gasification of biomass:Upgrading of the crude gasification product gas for electricityand heat generation. Report TPS 96/17; Termiska ProcesserAB: Sweden; 1996.

[26] Hofbauer H, Rauch R, Bosch K, Koch R, Aichernig C. BiomassCHP plant Gussing e a success story. Proceedings of expertmeeting on pyrolysis and gasification of biomass and waste,Strasbourg, France;October 2002. Available from:http://www.cfb.at [accessed February 2012].

[27] Achenbach E, Riensche E. Methane/steam reforming kineticsfor solid oxide fuel cells. Journal of Power Sources 1994;52:283e8.

[28] Berger C. Handbook of fuel cell technology. Englewood Cliffs,NJ: Prentice-Hall; 1968.

[29] Aguiar P, Adjiman CS, Brandon NP. Anode-supportedintermediate temperature direct internal reforming solidoxide fuel cell. I: model-based steady-state performance.Journal of Power Sources 2004;138:120e36.

[30] Wilke C. Diffusional properties of multicomponent gases.Chemical Engineering Progress 1950;46:95e104.

[31] Bird RB, Stewart WE, Lightfoot EN. Transport phenomena.NY: Wiley; 2002.

[32] Hayes R, Kolaczkowski S. Introduction to catalyticcombustion. Amsterdam: Gordon and Breach Science; 1997.

[33] Chan SH, Khor KA, Xia ZT. A complete polarization model ofa solid oxide fuel cell and its sensitivity to the change of cellcomponentthickness. JournalofPowerSources2001;93:130e40.

[34] Virkar AV, Chen J, Tanner CW, Kim J. The role of electrodemicrostructure on activation and concentrationpolarizations in solid oxide fuel cells. Solid State Ionics 2000;131:189e98.

[35] Zhu H, Kee RJ. A general mathematical model for analyzingthe performance of fuel-cell membrane-electrodeassemblies. Journal of Power Sources 2003;117:61e74.

[36] Kim J, Virkar AV, Fung K, Mehta K, Singhal SC. Polarizationeffects in intermediate temperature, anode-supported solidoxide fuel cells. Journal of Electrochemical Society 1999;146:69e78.

[37] Autissier N, Larrain D, Van Herle J, Favrat D. CFD simulationtool for solid oxide fuel cells. Journal of Power Sources 2004;131:313e9.

[38] Advanced metal and ceramic filter systems for Critical gassolid separation processes. Pall Gas Solid SeparationSystems, <http://www.pall.com/pdf/GSS. pdf>; 2009.

[39] Nacken M, Engelen K, Heidenreich S, Baron GV. Developmentof a tar reforming catalyst for integration in a ceramic filterelement and use in hot gas cleaning. Industrial andEngineering Chemical Research 2007;46:1945e51.

[40] Di Carlo A, Foscolo PU. Hot syngas filtration in the freeboardof a fluidized bed gasifier: development of a CFD model.Powder Technology 2012;222:117e30.

[41] Rapagna S, Gallucci K, Di Marcello M, Foscolo PU, Nacken M,Heidenreich S. Situ catalytic ceramic candle filtration for tarreforming and particulate abatement in a fluidized-bedbiomass gasifier. Energy & Fuels 2009;23:3804e9.

[42] Rapagna S, Gallucci K, Di Marcello M, Foscolo PU, Nacken M,HeidenreichS.Gascleaning,gasconditioningandtarabatementbymeans of a catalytic filter candle in a biomass fluidized-bedgasifier. Bioresource Technology 2010;101:7134e41.

[43] Rapagna S, Gallucci K, Di Marcello M, Foscolo PU, NackenM,HeidenreichS.FirstAl2O3basedcatalyticfiltercandlesoperatingin the fluidized bedgasifier freeboard. Fuel 2012;97:718e24.

[44] http://www.htceramix.ch/upload/Stack%20F-design.pdf[accessed February 2012].

[45] http://www.sofcpower.com/uploaded/documenti/p_d-asc-700_042012.pdf [accessed February 2012].

Process simulation of a SOFC and double bubbling fluidized bed gasifier power plant

Documents