Sie sind auf Seite 1von 13

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Contents lists available at ScienceDirect

Chemical Engineering Research and Design


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

An adaptable steady state Aspen Hysys model for the


methane fuelled solid oxide fuel cell
Timothy Anderson 1 , Periasamy Vijay 2 , Moses O. Tade
Centre for Process Systems Computations, Department of Chemical Engineering, Curtin University, Western Australia 6845, Australia

a b s t r a c t
An adaptable model for the methane fed internal reforming SOFC using the in built features of Aspen Hysys is presented in this paper. The model includes the electrochemistry, the diffusion phenomena and the reforming kinetics
in detail. Three potential methods for representing the SOFC are investigated out of which the recycled reforming
model is found to be capable of providing reasonable results over a wide range of operating conditions. The electrochemical model that gives good agreement with experimental data is also identied. From the simulations, it is
concluded that the developed model is reasonably accurate over a wide operating range and can be used for steady
state analysis. The computational challenges in the modelling are discussed. The model will be used for system level
optimisation studies of the SOFC system especially in conjuncture with gas turbines and steam turbines.
2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
Keywords: Aspen Hysys; Solid oxide fuel cell; Model adaptability; Steady state model; Model validation

1.

Introduction

Solid oxide fuel cell (SOFC) is the most promising fuel cell
technology especially for applications including stationary
standalone power generation, combined heat and power systems and hybrid systems where the fuel cell is coupled with
turbines to improve overall system efciency. Though the
SOFC can handle a variety of hydrocarbon fuels, hydrogen is
the preferred fuel. This hydrogen is usually produced externally in a steam reformer by reforming methane and utilising
the exhaust heat from the SOFC. Another alternative that is
considered in this work is the direct internal reforming of
methane in the anode of the SOFC, which is possible due to
the high temperatures present in the SOFC anode. This type
of design enables high energy conversion efciency for the
system. On the other hand it is also required that the anode
material must be a suitable catalyst for the steam reforming. One problem with direct internal reforming is the carbon
deposition in the anode which leads to catalyst deactivation and performance loss. A sufciently high steam/carbon
ratio can be maintained to minimise carbon deposition. The
addition of CeO2 to the Ni YSZ cermet material can also

be effective in preventing carbon deposition (Belyaev et al.,


1995).
A steady state model of the SOFC that is applicable for a
wide operating range as well as readily amenable for modications is necessary for the purpose of system design and
optimisation. Because many of the phenomena occurring in
a SOFC are not very well understood (for example the electrochemical kinetics), empirical models have been employed
in the literature to model them. In many cases, the empirical
models do not represent the wide operating ranges of the fuel
cell. Therefore, it becomes necessary to validate such empirical models against a wide range of experimental data. Also, for
performing system design and optimisation, it is convenient if
the model parameters are easily changeable (for the purpose
of variable sizing or to enrich the model with more details, for
example). Aspen Hysys (Hysys User Guide, 2004) is one of the
extremely versatile and popular softwares used in the process
industry for process modelling, conceptual design and optimisation. Owing to its features such as modular operation,
multi ow sheet architecture and the object oriented design,
this software offers the possibility of constructing steady state
models of the SOFC that will be easily adaptable.

Corresponding author. Tel.: +61 8 9266 7581; fax: +61 8 9266 2681.
E-mail addresses: V.Periasamy@curtin.edu.au (P. Vijay), M.O.Tade@curtin.edu.au (M.O. Tade).
Received 20 March 2013; Received in revised form 17 June 2013; Accepted 20 July 2013
1
Deceased.
2
Tel.: +61 8 9266 9890.
0263-8762/$ see front matter 2013 The Institution of Chemical Engineers. Published by Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.cherd.2013.07.025

296

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

There are a number of steady state models of the SOFC


in the literature. Some of them are not based on commercial softwares and others are based on software like Aspen
Hysys (Doherty et al., 2010; Palsson et al., 2000; Riensche
et al., 1998, 2000; Suther et al., 2011; Zhang et al., 2005). A
steady-state thermodynamic model of a hybrid solid oxide
fuel cellgas turbine cycle was developed using AspenPlus in
Suther et al. (2011). This hybrid cycle model incorporated a
zero-dimensional macro-level SOFC model, which was then
integrated with the Aspen plus model. In Palsson et al. (2000),
a fuel cell model including the effects of resistive cell losses,
reaction kinetics for the reforming reaction and heat conduction was presented. This model was integrated with Aspen to
analyse a system featuring external pre-reforming and recirculation of anode gases. In Riensche et al. (1998, 2000), an
SOFC model program as a Fortran routine was integrated into
the commercial ow sheet simulator PRO/II (Simsci) to analyse a clean combined cycle CO2 separating SOFC system. In
Zhang et al. (2005), a model for the tubular SOFC was presented using AspenPlus. The model was rather simple without
detailed consideration for cell voltage, diffusion phenomena
and the reforming kinetics. Also, it was not validated for a wide
range of data from across the literature. A tubular biomass
syngas fuelled solid oxide fuel cell stack was modelled using
Aspen plus in Doherty et al. (2010). Their model was built
using the inbuilt features of Aspen plus but was not validated
against experimental data pertaining to various temperature
and pressure conditions. The model was used to study the
effects of varying current and utilisation values on the SOFC
stack performance for the stack operating on wood and miscanthus syngas. This model was of equilibrium type and based
in Gibbs free energy minimisation. However, in our study we
found that this method is not satisfactory to represent the
SOFC with varying pressure and temperature operations.
The usual approach to include fuel cell into the Hysys models is to write separate codes for the cell in C++ or Fortran and
link it with the Hysys model of the overall system (Palsson
et al., 2000; Riensche et al., 1998, 2000; Suther et al., 2011). The
other approach is to model the fuel cell also using the inbuilt
components of Hysys. The second approach has the advantage
offering easy modications and exploitation of the features of
the commercial software. In this work, we develop a model
for the methane fed SOFC using inbuilt features of Aspen
Hysys. The model is detailed and includes the electrochemistry, the diffusion phenomena and the reforming kinetics in
detail. We investigate three possible methods for modelling
the SOFC viz. the equilibrium model, the 2-stage PFTR model
and the recycled reforming model, with respect to their ability to reproduce experimental data from the literature across a
wide range of operating temperatures and pressures. We also
investigate the validity of different methods that are available
in the literature for representing the SOFC electrochemistry.
In Section 2, the model construction is discussed in detail.
The model validation is presented in Section 3 and the paper
concludes in Section 4.

elements. The construction of the cell model is presented in


the following along with the rationale.

2.1.1.

Cathode

The Cathode was designed to be modelled as a variable oxygen separator; however this proved to be prone to calculation
difculties within Aspen Hysys and was instead replaced with
a 100% oxygen separator combined with a stream splitter that
adjusted the proportion of oxygen sent to the electrolyte, with
the remaining oxygen mixed back with the cathode exhaust
gas.

2.1.2.

Rate of oxygen transport/cell current

Within the electrolyte the oxygen is transported as two oxide


ions (Singhal and Kendal, 2003). However within our model it
is just treated as oxygen molecules. Nevertheless, the rate of
oxygen transport through the electrolyte determines the fuel
cell current due to the transport of these ions. As such, the cell
current I is given by:
I = 4FnO2 ,electrolyte ,

(1)

where, F is the Faradays constant and nO2 ,electrolyte is the oxygen mole ow rate. The cell current density j is determined by
dividing the cell current by the area (A):
j=

I
A

(2)

The maximum oxygen ow through the electrolyte is determined by full oxidation of the fuels supplied to the anode.
Within the model, only feeding of hydrogen and methane is
considered, so the two oxidation equations are as follows.
H2 + 1/2O2 + H2 O

(3)

CH4 + 2O2 2H2 O + CO2

(4)

The oxygen ow through the electrolyte can then be determined by the fuel utilisation factor, FU, which is the ratio of
actual oxygen ow rate to the maximum possible oxygen ow
rate and as such is effectively the extent of the oxidation reactions within the fuel cell. For the model, the fuel utilisation
is considered as a design variable, and needs to be specied
prior to simulation. The Oxygen ow through the electrolyte
can then be determined as:
n O2 ,electrolyte = FU(0.5n H2 ,feed + 2n CH4 ,feed )

(5)

The oxygen ow through the electrolyte is adjusted to that


specied by altering the split fraction in the cathode model.
Limits are included on the oxygen ow from trying to exceed
the oxygen content in the incoming air stream in the cases
where oxygen is the limiting reagent.

2.1.3.

Electrolyte

The electrolyte is simply modelled as a stream connecting the


cathode to the anode in which pure oxygen will travel through.

2.

Model construction

2.1.

Cell model

2.1.4.

Since the aim is to develop a model using the in-built features of Aspen Hysys, it becomes necessary to represent the
processes inside the cell using the standard Aspen Hysys

Anode

Due to the complex chemical reactions occurring within the


anode (reforming and electrochemical), several different models for the anode were considered. The models investigated
were an equilibrium model, kinetic CSTR for total anode, twostage PFTR and recycled reforming models. In this section we

297

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

provide a short description of these models. The detailed discussions and justications for their validity are deferred to
Section 3.
The equilibrium reactor is modelled as a Gibbs Reactor in
ASPEN HYSYS, which performs chemical and phase equilibrium calculations by Gibbs energy minimisation. The use of
the Gibbs reactor allows the simultaneous consideration of
oxidation reactions and the reforming of hydrocarbons within
one unit, and adjusts the outlet to equilibrium conditions. This
method showed good results for thick anodes with high fuel
utilisation, however for thinner anodes or low fuel utilisation
where kinetics are more signicant, the results from the equilibrium model showed considerable error as will be shown in
the validation section (Section 3). The primary disadvantage
of the model is that it does not account for the kinetics of the
reforming reactions.
The kinetic CSTR model replaced the Gibbs Reactor with a
CSTR reactor with kinetic models for the reforming and electrochemical reactions. These method experienced problems
with convergence of solutions and often resulted in no reactions occurring within the reactor. Therefore, this method is
not considered further in this work.
The two-stage PFTR model considers the anode to consisting of two sections, rstly the outer region where no
electrochemical reaction occurs, in this section there is only
steam reforming and water gas shift reactions occurring. The
second region is the active region at the electrolyte interface;
this is modelled as a thin region 10 m thick where both electrochemical and reforming reactions occur simultaneously.
The drawback of this model is that it does not consider the
participation of the water vapour formed during the electrochemical reaction in the reforming reaction of the rst PFTR.
The recycled reforming model was formulated to address this
drawback.
The recycled reforming model consisted of a PFTR which
performed the steam reforming of methane along with the
water gas shift reaction, followed by a Gibbs reactor with
methane inhibited to prevent further steam reforming, the
outlet from the Gibbs reactor is then recycled back into the
PFTR, with the recycle ratio as a tuning parameter for the
anode. The recycled reforming model is capable of providing
reasonable results over a wide range of conditions as will be
shown in Section 3. A schematic of the recycled reforming

model is shown in Fig. 1. The validity of these different models (viz., the equilibrium model, the 2 stage PFR model and
the recycled reforming model) are tested by comparing simulation results with literature data and are discussed in detail
in Section 3.

2.2.

Fuel and air preheaters

The fuel and air preheaters act to bring the inlet steams up to
the fuel cell operating temperature. This is achieved by modelling them as heaters with the outlet temperature set to that
of the specied fuel cell temperature. Note that it is assumed
there is no pressure drop across the preheater.

2.3.

Cathode exhaust heat exchanger

The cathode exhaust heat exchanger is included in the model


to act as an energy balance over the SOFC, thus heating or
cooling the outlet stream based on the overall energy balance.
The Cathode exhaust was chosen since the cathode ow rate is
typically larger than the anode and also the air ow rate modulation is more typically used in SOFC temperature control. The
combination of anode and cathode exhaust heat exchangers
may slightly improve the results; however the computational
difculties of the SOFC will be also increased, and at this stage
is not deemed necessary.

2.4.

Bulk parameters

To model internal conditions of the fuel cell, bulk parameters


are needed to be determined from the Aspen Hysys model, this
involved taking a combination of inlet and outlet conditions
from the anode and cathode. It was found that using purely
outlet conditions provided better results over a wide range of
operating conditions; this was particularly prevalent at high
utilisation of fuel or oxygen. Despite the use of outlet conditions from the anode and cathode, they will still be referred to
throughout the text as bulk parameters.

RECYCLE STREAM
ANODE
PLUG FLOW REACTOR

MIXER

GIBBS REACTOR

TEE

CATHODE

COMPRESSOR

SPLITTER

MIXER
AFTER BURNER

Fig. 1 The schematic of the recycled reforming model of the SOFC.

298

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

2.5.

The SOFC electrochemistry

2.5.1.

Cell voltage

The SOFC acts as a galvanic cell, and as such is governed by


electrochemical reactions on the anode and cathode. Within
a SOFC the electrolytic reactions are as follows
Cathode :

O2 + 4e 2O2

(6)

H2 + O2 H2 O + 2e

Anode :

(7)

CO + O2 CO2 + 2e

Where Q is the reaction quotient and E0 (T) is the standard


cell emf. If we assume that the species behave ideally, (their
fugacity and activity coefcients equal 1) then the reaction
quotient can be specied in terms of pressure (P) of the reacting species, such that
Q=

PH2 ,anode PO2 ,cathode

The actual cell voltage is obtained by subtracting the overpotentials from the open circuit voltage.

(8)

2.5.3.
The second anode reaction is much slower than the hydrogen oxidation and is typically not signicant (Singhal and
Kendal, 2003), so has not been considered within the model.
So for the purposes of modelling the SOFC, the overall electrochemical reaction within the fuel cell is given by the overall
cell reaction: H2 + 0.5O2 H2 O.
The standard cell voltage of a cell for a particular electrochemical reaction is the voltage when the cell is at
equilibrium. At equilibrium, the change of Gibbs free energy,
r G(T), of the reaction must be equivalent in change of electrical potential such that
r G(T)
E0 (T) =
zF

(9)

where z is the number of electrons participating the electrochemical reaction. This equation is used to determine the
equilibrium cell voltage. However the determination of the
change in Gibbs free energy is required. As
r G = r H Tr S

(10)

where H is enthalpy, T is absolute temperature and S is the


entropy. Then,
r G(T) = r G0 + (r H(T) r H0 ) (Tr S(T) T0 r S0 )

(11)

This can be expanded using the denitions of enthalpy and


entropy to

r G(T) = r G0 +

r Cp dT (T T0 )r S0 +
T0

T0

r Cp
dT (12)
T

If we assume that the change of heat capacity (Cp ) over the


temperature range is constant then the nal result is,
r G(T) = r G0 + r Cp (T T0 ) (T T0 )r S0 Tr Cp ln

T
T0
(13)

The values for r G(T), r S0 and r Cp are 474.4 kJ/mol,


88.84 J/K mol and 18.8 J/K mol as calculated from data
obtained from Atkins and de Paula (2010).

2.5.2.

RT
= E (T)
ln(Q)
zF
0

(14)

Activation over-potential: Butler Volmer equation

The ButlerVolmer equation is used to describe the kinetic


effects of electrochemical reactions on the current voltage output from a half cell. As such, it relates the current density to
the activation overpotential. The ButlerVolmer equation is
derived from assumptions of the rate limiting step within the
electrochemical reaction and is based on rst order reversible
reaction kinetics of the reacting species. The rate constants
within the equation are based on the activated complex theory, which obey the Arrhenius law (Atkins and de Paula, 2010).
The ButlerVolmer equation takes the form


j = jA jC = j0

CR
CR

C0


exp C


exp A

C0

F
act
RT

F
act
RT

(16)

where  is the overpotential, C represents concentrations


of reactants and products and R is the universal gas constant. The values of the parameters for the charge transfer
coefcients (A and C ), the expression to determine the
exchange current density (jo ) and the presence of the surface to bulk ratios vary signicantly between different SOFC
models in the literature. The values A and C are the
charge transfer coefcients, and according to Costamagna and
Honegger (1998) their values are 1.4 and 0.6 for the cathode and
2.0 and 1.0 for the anode, respectively.
The ButlerVolmer equation cannot be solved explicitly, so
an iterative method was implemented within Aspen Hysys. In
this method, the activation over-potential is assumed and the
corresponding cell current is calculated by the ButlerVolmer
equation and this is compared to the specied cell current. A
secant method is used to adjust the activation overpotential
so that the specied current and the calculated current are
within an error tolerance of less than 0.5%.
When there is zero current through a half cell, the rate of
the anodic and cathodic reactions are the same, this rate of
exchange on the electrode surface is termed the exchange current density. The exchange current density is dependent on a
large number of factors, including electrode material, species
concentrations and temperature. There are many different
approaches used within literature to simulate the exchange
current density. The formula for the exchange current density
is often given by the form:

Nernst equation

The Nernst equation is used to account for changes in species


concentrations on the equilibrium or open circuit voltage of
the cell. Such that the open circuit voltage EOCV is given by
OCV

(15)

PH2 O,anode

j0 = j

 P O  P R
R
O
Pref

Pref

exp

 E

act

RT


(17)

where j* is the pre exponential factor, Eact is the activation energy, PO and PR are the oxidant and reductant partial
pressures and  is the concentration exponent. The values

299

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Table 1 Exchange current density parameters.


Electrode

Anode

Cathode

2.9 10 A/m
120 J/mol
1.0
1.0
8

Pre exponential factor, j*


Activation energy, Eact
Reductant conc. exponent
Oxidant conc. exponent

7.0 108 A/m2


120 J/mol
0.25
0

determined, and can be estimated using pore size (rpore ) and


porosity () and active area thickness tactive
Volume of pore space = tactive A
2

Volume of pore = (rpore ) tactive


Number of pores, Npore =

for these factors are from Hosseini et al. (2011) and Singhal
and Kendal (2003) and are listed in Table 1. However for different SOFC materials different values of the parameters may be
required.
Other models have ignored the composition dependence
on the exchange current density and used either a constant
value or an activated complex approach with a form similar
to that given below:
j0 =

RT
exp
K
zF electrode

 E

act


(18)

RT

Some researchers have instead proposed a reaction mechanism at the electrode surface and derived a rate equation
for their proposed mechanism. Zhu et al. (2005) proposed a
hydrogen spill over mechanism at the anode and an oxygen
adsorption mechanism and derived formulas for the exchange
current density as follows

j0,anode = jH2

(PH2 /PH )

1/4

(PH2 O )

1 + (PH2 /PH )

1 + (PO2 /PO )

1/2

where P* is a parameter depending on adsorption equilibrium,


dened in detail in Zhu et al. (2005) and j* an empirical parameter use to t experimental data.
The exchange current density dened in Zhu et al. (2005)
yielded a better correlation to the experimental results for
SOFC pressure variations. However, the temperature uctuations were not well accounted for. Therefore, an activation
energy term accounting for the temperature was incorporated
into the models as given in the following.

j0,anode = jH2

(PH2 O )

1 + (PH2 /PH )

3/4

1/2

T exp

 E

act

(21)

and,

2.5.4.

(27)

The effect of the increased active surface area needs to be


accounted for in the Butler Volmer equation; this has been
achieved by nding the ratio of the active surface area to SOFC
surface area Ractive area and then adjusting the current density
j to a surface current density jsurface .
Ractive area =
jsurface =

Npore Spore
A

j
Ractive area

(28)

(29)

Ohmic overpotential

The ohmic overpotential losses are a result of electric current


running through the cell causing a resistive loss of electrical
potential. It is often more convenient to calculate the resistance per unit area and work with the current density, such
that:
ohm = jRcell

ti
Si

 E

act

(31)

The total cell resistance is then the sum of the different


component resistances so that:
Rcell = Ranode + Relectrolyte + Rcathode + Rinterconnect

1/4
(PO2 /PO )
2
jO2
T exp
1/2
1 + (PO2 /PO )
2

(30)

The resistance per unit area of a given component can be


dened by its thickness and conductivity, as:
Ri =

RT

j0,cathode =

(26)

2.5.5.

(PH2 /PH )

Active surface area = Npore Spore

1/4

1/4

Surface area of pore, Spore = 2rpore tactive + (rpore )

(25)

(20)

and,
(PO2 /PO )

Volume of pore space


Volume of pore

(19)

j0,cathode = jO2

(24)

The determination of the thickness of the active area


is somewhat problematic as there is no clear point where
electrochemical reactions will start. Clearly the maximum
thickness is that of the electrode. However this is often too
large as the conduction paths through the composite electrode would be very high. It has been found that by setting the
thickness of the active area to the thickness of the electrolyte
provides fairly good correlation with the literature.

3/4

1/2

(23)

(32)

RT

(22)

Active area for electrochemical reactions

Many electrodes in different types of SOFCs are made from


composite materials that contain the electrolyte material
(example, Ni/YSZ anodes). This composite structure has the
effect of extending the triple phase point into the electrode from the electrode electrolyte interface. To account for
this effect, the increase in active surface area needs to be

The conductivities of the different component have been


taken from different sources and should in theory be changed
for each different SOFC type. However since almost all the
types use YSZ electrolyte and in most designs this part dominates the overall cell resistance, the changes in cathode and
anode material will have minimal effect on the overall ohmic
overpotential.
On comparison with some literature results, it was sometimes observed that the cell resistance appeared to decrease
at higher current densities. This could be attributed to higher

300

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

temperatures within the electrolyte compared to the bulk


of the SOFC due to resistive losses within the electrolyte.
We obtain an approximation for the current dependence of
the electrolyte temperature to account for this phenomenon.
Since the power dissipated in the electrolyte Pelecrolyte is given
by resistive losses,

At steady state the heat dissipated from the electrolyte


must equal the power dissipated within the electrolyte.

where telectrode is the thickness of the electrode and Deff is


the effective diffusion coefcient. The plus or minus in the
above reaction depends on whether species i is being produced
or consumed at the electrode. Additionally the model above
assumes that the total concentration of species is constant;
otherwise there will be bulk ow as well as diffusive ow. However the above equation is the approximation that is generally
used in the literature (Mueller et al., 2006).
The maximum possible current occurs when the concentration of the reactants at the surface is zero, this maximum
current is known as the limiting current jlim and is given by

Q = UAT

jlim =

Pelectrolyte = j2 Relectrolyte

(33)

(34)

where, U is the heat transfer coefcient and A is cell area. The


temperature difference (T) is given by
T = Telectrolyte Tcell

(35)

Taking the energy balance gives,


Rcell 2
(36)T = UA
j
If we assume that Rcell and UA are roughly constant then
we can dene the electrolyte temperatures as
Telectrolyte = Tcell + kj2

(37)

where, k is a parameter that we call as the cell over temperature factor that depends on the ratio of electrolyte resistance
to heat transfer properties.
The electrochemistry model is required to dene the cell
voltage and current. These calculations are carried out separately in spreadsheets linked to the Hysys ow sheet model.
The Spreadsheet can be used to manipulate or perform custom calculations on ow sheet variables and the spreadsheet
cells are automatically updated when ow sheet variables
change. The temperatures, concentrations and ow rates from
the Hysys ow sheet are accessed by the spreadsheet program
and it calculates the output variables such as the cell current
and voltage.

2.6.

Diffusion model

The transport of material to and from the bulk of the fuel cell
to the active electrode surface has a signicant effect on the
performance of SOFCs. This typically results in some voltage
loss called as the concentration over-potential and imposes
a limit on the maximum current that can be drawn from the
cell. Thus, within the SOFC model, a model for predicting the
effects of transport phenomena is required. Since the overall
model is zero dimensional, a relatively simple one dimensional model through the electrode to the electrolyte surface
should prove sufciently accurate.
Assuming steady state conditions, the rate of mass transport to the surface is equal to the rate of reaction at the surface,
which depends on the current density.
Rate =

j
dC
= Deff i
zF
dx

(38)

Taking C i and C* i are the surface and bulk concentration of


species i then the above equation can be converted into

zFDeff Ci

(40)

telectrode

Substituting this into Eq. (39) and rearranging provides the


following,
Ci

Ci

=1+

j
jlim

(41)

This can then easily be substituted into the Butler Volmer


equation. However, signicant calculation problems occur if
the ratio is less than or equal to zero. To prevent this in the
program, if the ratio of current to limiting current for reactants
is greater than 0.9999, the cell voltage is set to zero and the
anode and cathode overpotential calculations are stopped.

2.6.1.

Diffusion coefcient

Determination of the effective diffusion coefcient is one of


the key parameters in the diffusion model. As the anode and
cathode are porous consideration for the pore structure needs
to be accounted for in the diffusion model. Since within pores
both Knudsen and binary diffusion can be signicant, both
need to be considered. Knudsen diffusion coefcient is given
by

Di,K =

2
rpore
3

8RTcell
Mi

(42)

The binary diffusion coefcient is calculated by a method


provided in Bird et al. (2006). This method is developed for
monotonic non-polar gases and shows good correlation for
polyatomic non-polar gases, so for application to polar gases
such as water there may be signicant error. Nevertheless this
method has been used in many other SOFC models (Severson
and Assadi, 2011; Todd and Young, 2002). The binary diffusion
coefcient is given by


DAB = 0.0018583

T3

 1
MA

1
MB

1
2
PAB
D,AB

(43)

where D,AB is the collision integral for diffusion and is


specied as a function of dimensionless temperature T/AB ,
and can be found in Bird et al. (2006). The values of  AB and
AB are determined from the LennardJones parameters of the
individual species, such that
AB =

1
(A + B )
2

(44)

and
jt
Ci = Ci electrode
zFDeff

(39)

AB =

A B

(45)

301

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

The properties LennardJones parameters for the species


were found in Bird et al. (2006) and Poling et al. (2000).
Within the anode there are multiple species present so
the use of a pure binary diffusion coefcient is no possible,
so a means of calculating a medium diffusion coefcient is
required. Note in the cathode this is not required as the major
components are only nitrogen and oxygen, other species are
in trace concentrations and should have minimal effect. In
cases of a small amount gas diffusing through medium of
roughly constant composition, the medium diffusion coefcient is given by Bird et al. (2006).

 xj

Di,m =

Di,j

(46)

decrease and become negative as the water concentration


increased. One such model was used by Nakagawa et al. (2001)
such that
rCH4 = KSR

 xj

Di,m = (1 xi )

Di,j

Ki = Ai exp

2.7.

Reforming model

2.7.1.

Modelling of steam reforming

Steam reforming within SOFC has had considerable investigation. However, there is still no consensus on the kinetic
expressions. There is a wide range of expressions investigated from power law expressions tting experimental data to
LanguirHinshelwood kinetic models assuming reaction pathways. This large variety of available models makes it difcult
to determine a suitable reforming model to predict cell performance.
The activation energy for the reforming reaction has a large
spread of reported values ranging from 58 kJ/mol to 229 kJ/mol,
however the value is typically slightly below 100 kJ/mol. The
reported order of reaction with respect to methane is 1 consistently. However there are considerable differences between
reaction order for water ranging between 1 and 1.25. This
large inconsistency with reaction order is likely due to different conditions (temperature, partial pressures) and reformer
materials used in the studies.
The use of a LangmuirHinshelwood kinetic based model
is appropriate, as this would allow for an order of approximately one when the water concentration was low and would

(48)

 E 
i

(49)

RT

Using this model they found the values for the parameters
as listed in Table 2. This model however is not equilibrium
limited and so was modied for use in Aspen Hysys so that

K1 K2 K3 PCH4 PH2 O
(1 + K2 PCH4 + K2 PH2 O )

PCO PH2 3
PCH4 PH2 O Keq,SR

mol
s m2

(50)

Additionally for modelling within Aspen Hysys, the values


appearing in the numerator need to be combined together,
these calculated values are displayed in Table 2. The equilibrium for the steam reforming reaction is given by the equation
Keq,SR =

PCO PH2 3
PCH4 PH2 O

2.7.2.

Modelling of shift reaction

(47)

The above calculation of the medium diffusion coefcient


is fairly inaccurate. However it should provide sufcient accuracy for a bulk parameter model. If further accuracy is required
then the dusty gas model could be used. However this greatly
increases computational difculty as it requires the solving of
an ordinary differential equation within the diffusion model.
The calculated diffusion coefcient is used in Eq. (39) to calculate the surface concentration, which in turn is used in
Eq. (41) to calculate the limiting current. The diffusion model
described in this section is also a subset of the electrochemical
model and is used to dene the concentration over-potential.
These calculations are performed separately in spreadsheets
that are linked to the Hysys ow sheet model.

(1 + K2 PCH4

mol
s
m2
+ K2 PH2 O )
2

where

rCH4 = KSR

However the assumption that the diffusing species is dilute


will not hold over the anode, so a correction factor is needed
to compensate for this. The use of the factor (1 xi ) will for a
two species system yield the binary diffusion coefcient, and
so has been used within the model. Thus,

K1 K2 K3 PCH4 PH2 O

(51)

Within modelling of anodes the shift reaction is usually


assumed to be in equilibrium. However as the kinetic reactors are used with Aspen Hysys, a kinetic model is required.
A model by Klein et al. (2007) presented a kinetic equation for
the shift reaction, as shown below:

 12509 

rCO = 1199T 2 exp

CCO CH2 O 6.77

 16909 

104 T 2 exp

CH2 CCO2

 k mol 
s m3

(52)

This kinetic model was then veried to have good correlation to the Gibbs equilibrium model within Aspen Hysys.
The rate of reaction was then increased so as to ensure that
equilibrium was achieved in the kinetic reactors.

2.7.3.

Modelling of electrochemical reaction

Modelling of the electrochemical reaction at the anode electrolyte interface is modied to a reaction with the transported
oxygen and hydrogen within the kinetic reactor. It was initially desired to use an equilibrium type reaction similar to
that used with the shift reactor. However, the use of three
reversible reactions combined within one reactor caused problems with convergence of solutions within Aspen Hysys. So it
was modied such that the reverse reaction was inhibited and
complete reaction of oxygen was consistently achieved. The
equation used within the Aspen Hysys model was:
rH2 = 1 106 (PH2 )(PO2 )

2.7.4.

0.5

 k mol 
s m3

(53)

Sizing of reactors

The sizing of the reactors for the anode reforming is somewhat


complicated due to the fact that the reaction rates are specied
by surface area. However within Aspen Hysys the rate needs to
be specied on a volumetric basis. This requires the specied
rate to be multiplied by the surface area to volume ratio of the
anode. Varying of the kinetic parameters within Aspen Hysys

302

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Table 2 Parameters of the LangmuirHinshelwood kinetic based model.


i

Ai

Average Ai

Ei (kJ/mol)

1
2
3

400612 mol/(s m )
3.01 104 4.96 104 kPa1
0.180.28 kPa1

506 mol/(s m )
3.895 104 kPa1
0.23 kPa1

Calculated values
K1 K2 K3
K1 K2 K3 /Keq,SR

0.02160.0850 mol/(s m2 kPa2 )


1.55 1019 6.08 1019 mol/(s m2 kPa4 )

0.0533 mol/(s m2 kPa2 )


3.82 1019 mol/(s m2 kPa4 )

is not easily achieved. An easier approach is to modify the


volume of the reactor instead. The conversion within a PFTR
for species A is specied by

XA

XA
rA

(54)

where V is the volume of the reactor, X is the conversion and


FAO is the inlet ow. Let ra denote the reaction rate with respect
to the surface area. Then we can also specify the conversion
as
Asurface
=
FAO

XA

XA
ra

(55)

Combining these two equations we get that

XA

XA
V
=
rA
Asurface

XA

XA
ra

(56)

Substituting this back into the original equation gives


V(Asurface /V)
=
FAO


0

XA

XA
ra

s = 6.92 104 exp(9681/T)


(57)

Thus showing that if we multiply the volume of the reactor


by the surface area (Asurface ) to volume ratio will produce the
same results is we directly use the volume specic reaction
rate.

3.

Model validation

In this section, the Aspen Hysys model developed is validated


by comparing it with a wide range of experimental and modelling data from the literature. Each of the following sections
provides a comparison of the results using the lumped parameter SOFC Aspen Hysys model compared to literature values
from the stated texts. These comparisons also show that the
model is exible and can be modied to represent different
geometrical and material properties. The development of the
solid oxide fuel cell model was strongly inuenced by comparison with the literature values, these decisions are discussed
as related to the literature investigated.

3.1.

11
215

effects of electrode kinetics on cell performance at varying


pressures are not well established. In the literature, various
methods for dening the exchange current density are proposed. In the following, we apply three different methods
for dening the exchange current density to the developed
model and study the inuence of pressure on the system
performance in terms of the polarisation curves. The results
are compared with the experimentally obtained curves from
Hashimoto et al. (2008). The rst method involves using a constant value for the exchange current density. The parameters
dening the current exchange density are given in Table 3 and
the comparison of our model results with Hashimoto et al.
(2008) are presented in Fig. 2.
The system in Hashimoto et al. (2008) has some differences
to the modelling system employed so slight modications
were required to account for these. The two most signicant
are the use of Sc doped Zirconia electrolyte rather than YSZ
requiring the conductivity (s) of the electrolyte to be modied
to

Effect of operating pressure

The effects of the electrochemical parameters on the cell


performance over a range of pressures are studied in the following. One of the key parameters affecting the performance
of the SOFC system is the operating pressure; as such it was
crucial to determine how pressure effects inuence the cell
performance. The pressure effects on transport properties and
electrochemical equilibrium are well understood, however the

(58)

Additionally the temperature of the furnace was found to


increase with current density and so also will fuel utilisation,
a rough correlation was determined such that
T = 650 C + 33FU

(59)

Also the fuel and air ow rates were not specied within the
text; however a fuel utilisation and current density were specied at maximum power allowing the fuel ow to be calculated.
The air ow was assumed to be in a 3 times excess to the fuel.
Note that due to the use of a pre-reformer in the anode, high
porosity and low tortuosity have been used in the model. From
Fig. 2, we can see that the error between the experimental and

0.1MPa Model
0.3MPa Model

0.8

Cell Voltage V

V
=
FAO

49
45
7

0.5MPa Model
0.7 MPa Model

0.6
0.1MPa
Experimental
0.3MPa Experimental

0.4

0.5MPa Experimental

0.2

0.7MPa Experimental

0
0

500

1000

1500

2000

2500

3000

Current Density mA/cm2

Fig. 2 Polarisation curves at various operating pressures


using constant exchange current density model.

303

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Table 3 Parameters corresponding to the three exchange current density models considered.
Model parameters
Cell over temp factor
Cathode Leakage rate
Recycle ratio

10
1%
1%

Electrode parameters
Porosity
Tortuosity
Pore radius
Anodic charge transfer coefcient, A
Cathodic charge transfer coefcient, C

Anode
0.55
1.1
1 m
1
1

Other modications
Cell temp
Electrolyte conductivity ScSZ rather than YSZ

650 + 33 FU
Conductivity 6.92E4 exp(9681/T)

Constant exchange current density


Exchange current pre exponential i* (A/cm2 )
Exchange current Activation energy Ea (J/mol)

Anode
7.00E + 00 A/m2

Cathode
9.00E+00 A/m2

Exchange current density model from Costamagna and Honegger (1998)


Exchange current pre-exponential, i* (A/cm2 )
Exchange current Activation energy, Ea (J/mol)

Anode
1.31E+03
120

Cathode
7161
120

The exchange current density model in Zhu et al. (2005)


Exchange current pre-exponential, i* (A/cm2 )
Exchange current Activation energy, Ea (J/mol)

Anode
7.87E+03
140

Cathode
1.06E+01
137

the numerical results using the constant exchange current


density model is huge. This is especially so at higher operating pressures. From these results, we conclude that constant
exchange current model is not suitable if we want our model
to be valid for a wide operating range.
The second method of dening exchange current density
is from Costamagna and Honegger (1998) and the parameters
used are given in Table 3. The comparison of model results
with Costamagna and Honegger (1998) is presented in Fig. 3.
Again we nd that this is a fair enough approximation at lower
operating pressures but not at high operating pressures. This
led to the investigation of a third method for dening current
exchange density based on Zhu et al. (2005).
The parameters and the results comparison for the third
method by Zhu et al. (2005) are given in Table 3 and Fig. 4,
respectively. From Fig. 4 we nd that this method gives much
better results than the other two methods. However, there
are still some discrepancies at higher operating pressures. For
want of more published results, we will adopt the method
by Zhu et al. (2005) in our model. Further, the over-potential
losses are compared to the experimentally measured values

Cathode
0.55
3
1 m
2
2

from Hashimoto et al. (2008) in Fig. 5. We can see that there


are some discrepancies between the results both at low and
high operating pressures, more so at high operating pressure.

3.2.

Effect of electrochemical parameters

Comparison of the model with results from Leng et al. (2004)


revealed a few different phenomena. Firstly, due to the high
current density achieved within the fuel cell, the ohmic overpotential should have caused signicantly low cell voltage; it
was deemed that the electrolyte temperature must be significantly higher than the furnace temperature. This led to the
development of the cell over temperature factor as discussed
in Section 2.5.5. Furthermore with the availability of cell data
over a range of temperatures, investigation into the temperature effects was possible. Initial activation energies used by
Aguiar et al. (2004) were employed as a starting point. In the
following, we validate the cell behaviour over a range of operating temperatures using data from the literature. Note that
the cathode is thin and very dense and therefore, low porosity, small pore radius and high tortuosity were required to get
1.2

0.1MPa Model

0.3MPa Model
0.5MPa Model
0.7 MPa Model

0.6

0.1MPa
Experimental
0.3MPa Experimental

0.4

0.5MPa Experimental
0.7MPa Experimental

0.2

Cell Voltage V

Cell Voltage V

0.8

0.1MPa Model

0.8

0.3MPa Model
0.5MPa Model

0.6

0.7 MPa Model


0.1MPa Exprimental
0.4

0.3MPa Experimental
0.5MPa Exprimental

0.2

0.7MPa Experimental

500

1000

1500

2000

2500

3000

Current Density mA/cm2

Fig. 3 Polarisation curves at various operating pressures


using exchange current density model from Costamagna
and Honegger (1998).

500

1000

1500

2000

2500

3000

Current Density mA/cm2

Fig. 4 Polarisation curves at various operating pressures


using exchange current density model from Zhu et al.
(2005).

304

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

0.6

1.2

0.5

Model (700 C)
Model (750 C)

0.4

0.8

0.3 A/cm^2 (Model)

0.3

2.8 A/cm^2 (Model)


0.2

0.3 A/cm^2 (Exp)

Cell Voltage V

Overpotenal (V)

Model (650 C)

Model (800 C)
Experimental (650 C)

0.6

Experimental (700 C)
0.4

Experimental (750 C)

2.8 A/cm^2 (Exp)

0.1

Experimental (800 C)

0.2

0.2

0.4

0.6

0.8
0

Cell Pressure (MPa)

oxygen depletion as observed in experimental results. The


model specications and the activation energy data as in
Aguiar et al. (2004) are given in Table 4.
As can be seen in Fig. 6, the model under predicts the cell
voltage as the temperature decreases. A comparison of the
model and the experimental values at 250 mA/cm2 is given in
Table 5.
From the table, we understand that the deviation in the
performance characteristics of the cell at low temperatures is
mainly due to the deviation in the activation over-potential.
Therefore an over-potential deviation factor is introduced. If
we assume that the overpotential characteristics of the cell are
symmetric and the current is small then using the sin h expansion (Qi et al., 2005; Vijay et al., 2009) as given in Eqs. (61) and
(62), we can assume that the current density is proportional
to activation overpotential. Thus at constant current density,
the variation in overpotential is proportional to the inverse
of the variation in exchange current density. If we assume the
deviation from expected overpotential and the modelled overpotential is the result of an activated process, then plotting
how the proportional difference in overpotential (overpotential deviation factor K) varies with temperature and tting to
it to an exponential curve the activation energy difference can
be found as depicted in Fig. 7.
act,anode =

act,cathode

2RT
sin h1
zF

2RT
=
sin h1
zF

0.5j


(61)

j0,anode

0.5j


(62)

j0,cathode

500

1000

1500

2000

2500

3000

3500

4000

Current Density mA/cm2

Fig. 5 Comparison of cell over-potential model results


with experimental result from the literature.

Table 4 Parameters corresponding to the exchange


current density model in Aguiar et al. (2004).

Fig. 6 Polarisation curves at various operating


temperatures using exchange current density model from
Aguiar et al. (2004).
As can be seen from the plot in Fig. 7, the activation
energy difference is roughly 22 kJ/mol. This indicates that
the exchange current activation needs to be changed by the
same factor. Thus the exchange current density calculations
were then modied by the factor shown below.
exp(Eact /RT)
exp(Eact /RTref )

(63)

where 800 C was selected as the activation energy. Using


this exchange energy modier, the model results show better
comparison with the experimental polarisation curves over
different temperatures as depicted in Fig. 8.

3.3.
Representation of the internally reformed SOFC in
Aspen Hysys
Having established the electrochemical properties of the
model so that it compares favourably with the experimental
data from the literature over a range of operating temperatures and pressures, we consider the methane reforming in
the SOFC in the following. We consider the performance of
the three anode models, viz., the equilibrium model, the two
stage PFTR and the recycled reforming models that were discussed in Section 2.1.4 with regards to methane reforming.
The performance of these models were compared with Koh
et al. (2002) in order to assess their performance with respect to
their polarisation characteristic curves. The parameters corresponding to the equilibrium, two stage PFTR and the recycled
reforming models are given in Table 6. The results are shown
in Figs. 911.

Model parameters (model tted at 800 C)


Cell over temp factor
7
0%
Cathode leakage rate
5%
Recycle ratio
Electrode parameters
Porosity
Tortuosity
Pore radius
Exchange current
pre-exponential, i* (A/cm2 )
Exchange current Activation
energy, Eact (kJ/mol)
Anodic charge transfer
coefcient, A
Cathodic charge transfer
coefcient, C

Anode
0.4
5
1 m
4.16E+03

Cathode
0.25
15
0.5 m
1.24E+02

140

137

1.5

1.4

0.5

0.6

Fig. 7 Determination of deviation in the activation energy.

305

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Table 5 Model and experimental cell over-potentials.


Temperature ( C)
Experimental voltage (Vexp )
Model voltage (Vmod )
Model anode overpotential
Model cathode overpotential
Total model activation overpotential (mod )
Model open circuit voltage (OCV)
Model ohmic overpotential (Ohm )
Expected overpotential exp = OCV Ohm Vexp
1/(RT)
Error in Overpotential = (exp -mod )
Shifted (error zeroed at 800 C) model overpotential shtf = mod + error (800 C)
Overpotential deviation factor K = shtf /exp

1.2

Model (650 C)

Model (700 C)

Model (800 C)
0.6
Experimental (650 C)
0.4

Experimental (700 C)

0.2

Experimental (750 C)
Experimental (800 C)

0
0

500

1000

1500

2000

2500

3000

3500

4000

Current Density mA/cm2

Fig. 8 Polarisation curves at different temperatures using


the modied activation energy.
From Figs. 911, we can see that the recycled reforming model gives the best agreement with the experimental
results. From Fig. 9, we can see that the equilibrium model
over-predicts the cell voltage at low current densities and
under-predicts it at mid and high current densities. This is
probably because at low current densities, the fuel utilisation
is also low. Therefore, we can conclude that the Gibbs energy
minimisation is not a valid assumption at low fuel utilisations
and the reforming kinetics need to be considered. From Fig. 10,
we can say that the 2 stage PFR model under-predicts the
cell voltage except for large current densities very near to the

700

750

0.55
0.379
0.349
0.207
0.556
1.167
0.237
0.38
0.0001304
0.176
0.572
1.5052631

0.73
0.675
0.201
0.148
0.349
1.158
0.1337
0.2943
0.000124
0.0547
0.365
1.240231

0.88
0.875
0.098
0.095
0.193
1.149
0.08
0.189
0.000118
0.004
0.209
1.10582

800
0.95
0.971
0.068
0.056
0.124
1.14
0.05
0.14
0.000112
0.016
0.14
1

limiting current density. This is probably due to the error in


species concentrations because of the segregated approach
to model reforming and electrochemistry. From Fig. 11, we
can see that the prediction of the cell voltage by the recycled
reforming model is better than the other models. From this
performance comparison between the three different models,
we can conclude that the recycled reforming model is superior
to the other two models.
To further establish the validity of the recycled reforming
model, we compare the model results with the experimental
results pertaining to considerably different operating conditions of a thick anode and low fuel utilisation (Chen et al.,
2007). The relevant model parameters are given in Table 7
and the results are plotted in Fig. 12. From the gure, we
can see that the recycled reforming model matches with the
experimental polarisation curves for a range of operating temperatures from 750 C to 900 C. This shows that the model
is valid across a wide range of operation and is suitable for
system design and optimisation purposes. The error in the
cell voltage as compared to the experimental values across
the current density range is shown in Fig. 13. As can be seen
from the results, the model provides good correlation (error
less than 3%) to the experimental data over the range of simulation. The only drawback of the recycled reforming model
is that the errors are higher at very low current densities
1.2

Table 6 Parameters for the equilibrium, two stage PFTR


and the recycled reforming models.
Model parameters (model tted at 800 C)
7
Cell over temp factor
0%
Cathode leakage rate
0%
Recycle ratio
10 kJ/mol
Activation energy modier
Reforming reaction water
1
exponent
Proportion reforming active area
N/A
Electrode parameters
Porosity
Tortuosity
Pore radius
Exchange current pre-exponential,
i* (A/cm2 )
Exchange current activation
energy, Eact (kJ/mol)
Anodic charge transfer coefcient,
A
Cathodic charge transfer
coefcient, C

Anode
0.5
2
0.5 m
4.16E+03
140

Cell Voltage V

1
0.8

Model

0.6

Experimental

0.4
0.2
0
0

100

200

300

400

500

600

700

Current Density mA/cm^2


0.16

Cathode
0.4
5
1 m
1.24E+02
137

0.14

Mole fracon

Cell Voltage V

Mode (750 C)
0.8

650

0.12
0.1

Mole fracon H2O

0.08

Mole fracon CH4

0.06

Mole fracon H2

0.04

Mole fracon CO

0.02

Mole fracon CO2

1.5

1.4

0.00

100.00 200.00 300.00 400.00 500.00 600.00 700.00 800.00

Current density mA/cm2

0.5

0.6

Fig. 9 Polarisation curves and the molar fraction proles


using the equilibrium model.

306

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

Table 7 Parameters for the recycled reforming model.


Model parameters
Cell over temp factor
Cathode leakage rate
Recycle ratio
Activation energy modier
Reforming reaction water
exponent
Proportion reforming active area
Electrode parameters
Porosity
Tortuosity
Pore radius
Exchange current pre-exponential,
i* (A/cm2 )
Exchange current Activation
energy, Ea (kJ/mol)
Anodic charge transfer coefcient,
A
Cathodic charge transfer
coefcient, C

1.2
1

Cell Voltage V

Model
0.8
0.6

0.03
Anode
0.4
2.3
0.3 m
615

Cathode
0.4
3.2
1.6 m
21.1

140

137

1.5

1.4

0.5

0.6

12

Error to Experimental (%)

Fig. 10 Polarisation curves and the molar fraction proles


using the 2 stage PFTR model.

2
0%
9.5%
10 kJ/mol
+1

10
8
6
4
2
0

Experimental

0
0.4

100

200

300

400

500

600

700

Current Density mA/cm2

0.2

Fig. 13 The error in recycled reforming model as


compared to experimental data from Chen et al. (2007).

0
0.00E+00

2.00E+02

4.00E+02

6.00E+02

8.00E+02

Current Density mA/cm2


0.12

Mole fracon

0.1
0.08

Mole fracon H2O

0.06

Mole fracon CH4

0.04

Mole fracon H2
Mole fracon CO

0.02

Mole fracon CO2

0
0.00

100.00 200.00 300.00 400.00 500.00 600.00 700.00 800.00

Curren density mA/cm2

Fig. 11 Polarisation curves and the molar fraction proles


using the recycled reforming model.

4.

0.8

Cell Voltage V

(less the 50 mA/cm2 ). Since this error can be found in the


other two models as well, it is probably due to inaccuracies
in the activation and ohmic over-potentials that are temperature dependent. At very low temperature, the overpotential is
largely activation controlled and the cell temperature is also
lower. Availability of more published data on the activation
energies and other electrochemical parameters could lead to
further improvement in results. However, this is not problematic for system design purposes because it is not likely to
design a system with such low current densities. The result
of all these simulations indicates that the recycle model is
satisfactory to represent the methane fed SOFC over a wide
operating range.

900 Experimental
850 Experimental

0.6

800 Experimental
750 Experimental
900 Model

0.4

850 Model
800 Model

0.2

750 Model
0
0

200

400

600

800

1000

1200

1400

Current Density mA/cm2

Fig. 12 Polarisation curves using the recycled reforming


model compared with data from Chen et al. (2007).

Conclusions

An Aspen Hysys model for the methane fed internal reforming SOFC is presented in this work. This model is constructed
using the inbuilt features of Aspen Hysys without any need
for linked code. The model is detailed and includes the electrochemistry, the diffusion phenomena and the reforming
kinetics in detail. We investigated three methods for modelling the SOFC viz. the equilibrium model, the 2 stage PFTR
model and the recycled reforming model. From the simulations it is concluded that recycled reforming model is capable
of providing reasonable results over a wide range of conditions. Methods for representing the exchange current density,
which is a key electrochemical parameter, are investigated. It
is found that the method by Zhu et al. (2005) provides much

chemical engineering research and design 9 2 ( 2 0 1 4 ) 295307

better correlation to pressure effects on a cell operating on


hydrogen. While investigating the temperature effects on the
polarisation curve, it is found that the model under predicts
the cell voltage as the temperature decreases. This effect is
attributed to the deviation in overpotential and an overpotential deviation factor is dened to correct it. By comparing
the model results with experimental data pertaining to the
wide temperature and pressure conditions it is concluded that
the developed model is reasonably accurate and can be used
for steady state analysis. In future studies, the model will be
used for system level optimisation studies of the SOFC system especially in conjuncture with gas turbines and steam
turbines.

Acknowledgements
The second and third authors gratefully acknowledge the rst
author, Mr. Tim Anderson, who did the bulk of the simulations for this manuscript as part of his Master of Philosophy
thesis at Curtin University. Unfortunately, Tim could not complete his thesis because he died on Friday, the 9th of March
2012 as a result of the injuries he sustained in a hit and
run accident while practicing for a charity youth ride. The
manuscript demonstrates the quality which his thesis would
have achieved. May his soul rest in peace!

References
Aguiar, P., Adjiman, C.S., Brandon, N.P., 2004. Anode-supported
intermediate temperature direct internal reforming solid
oxide fuel cell: I. Model-based steady-state performance.
Journal of Power Sources 138 (1/2), 120136.
Atkins, P.W., de Paula, J., 2010. Physical Chemistry, 9th ed. Oxford
University Press, Oxford.
Belyaev, V.D., Politova, T.I., Marina, O.A., Sobyain, V.A., 1995.
Internal steam reforming of methane over Ni-based electrode
in solid oxide fuel cells. Applied Catalysis A: General 133 (1),
4757.
Bird, R.B., Stewart, W.E., Lightfoot, E.N., 2006. Transport
Phenomena, 3rd ed. John Wiley & Sons.
Chen, X.J., Liu, Q.L., Chan, S.H., Brandon, N.P., Khor, K.A., 2007.
High performance cathode-supported SOFC with perovskite
anode operating in weakly humidied hydrogen and
methane. Electrochemistry Communications 9 (4), 767772.
Costamagna, P., Honegger, K., 1998. Modeling of solid oxide heat
exchanger integrated stacks and simulation at high fuel
utilization. Journal of the Electrochemical Society 145 (11),
39954007.
Doherty, W., Reynolds, A., Kennedy, D., 2010. Simulation of a
tubular solid oxide fuel cell stack operating on biomass
syngas using aspen plus. Journal of the Electrochemical
Society 157 (7), B975B981.
Hashimoto, S., Nishino, H., Liu, Y., Asano, K., Mori, M., Funahashi,
Y., Fujishiro, Y., 2008. Effects of pressurization on cell
performance of a microtubular SOFC with Sc-doped zirconia
electrolyte. Journal of the Electrochemical Society 155 (6),
B587B591.
Hosseini, S., Danilov, V.A., Vijay, P., Tade, M.O., 2011. Improved
tank in series model for the planar solid oxide fuel cell.
Industrial and Engineering Chemistry Research 50, 10561069.

307

2004. HYSYS 2004.2 User Guide. Aspen Technology Inc.,


Cambridge MA, USA.
Klein, J.M., Bultela, Y., Georgesa, S., Pons, M., 2007. Modeling of a
SOFC fuelled by methane: from direct internal reforming to
gradual internal reforming. Chemical Engineering Science 62,
16361649.
Koh, J.H., Yoo, Y.S., Park, J.W., Lim, H.C., 2002. Carbon deposition
and cell performance of NiYSZ anode support SOFC with
methane fuel. Solid State Ionics 149, 157166.
Leng, Y.J., Chan, S.H., Khor, K.A., Jiang, S.P., 2004. Performance
evaluation of anode-supported solid oxide fuel cells with thin
lm YSZ electrolyte. International Journal of Hydrogen Energy
29 (10), 10251033.
Mueller, F., Brouwer, J., Jabbari, F., Samuelsen, S., 2006. Dynamic
simulation of an integrated solid oxide fuel cell system
including current-based fuel ow control. Journal of Fuel Cell
Science and Technology 3, 144154.
Nakagawa, N., Sagara, H., Kato, K., 2001. Catalytic activity of
Ni YSZ CeO2 anode for the steam reforming of methane in
a direct internal-reforming solid oxide fuel cell. Journal of
Power Sources 92, 8894.
Palsson, J., Prausnitz, J., OConnell, J., 2000. Combined solid oxide
fuel cell and gas turbine systems for efcient power and heat
generation. Journal of Power Sources 86,
442448.
Poling, B., Prausnitz, J., O Connell, J., 2000. The Properties of
Gases and Liquids, 5th ed. McGraw Hill Professional.
Qi, Y., Huang, B., Chuang, K.T., 2005. Dynamic modeling of solid
oxide fuel cell: the effect of diffusion and inherent
impedance. Journal of Power Sources 150,
3247.
Riensche, E., Achenbach, E., Froning, D., Haines, M.R., Heidug,
W.K., Lokurlu, A., 2000. Clean combined-cycle SOFC power
plant cell modelling and process analysis. Journal of Power
Sources 86, 404410.
Riensche, E., Meusinger, J., Stimming, U., Unverzagt, G., 1998.
Optimization of a 200 kW SOFC cogeneration power plant:
part II. Variation of the owsheet. Journal of Power Sources 71,
306314.
Severson, H., Assadi, M., 2011. Modeling of overpotentials in an
anode-supported planar SOFC using a detailed simulation
model. Journal of Fuel Cell Science and Technology 8 (5),
051021051034.
Singhal, S.C., Kendal, K., 2003. High Temperature and Solid Oxide
Fuel Cells Fundamentals, Design and Applications. Elsevier.
Suther, T., Fung, A.S., Koksal, M., Zabihian, F., 2011. Effects of
operating and design parameters on the performance of a
solid oxide fuel cellgas turbine system. International Journal
of Energy Research 35, 616632.
Todd, B., Young, J.B., 2002. Thermodynamic and transport
properties of gases for use in solid oxide fuel cell modelling.
Journal of Power Sources 110, 186200.
Vijay, P., Samantaray, A.K., Mukherjee, A., 2009. A bond graph
model-based evaluation of a control scheme to improve the
dynamic performance of a solid oxide fuel cell. Mechatronics
19, 489502.
Zhang, W., Croiset, E., Douglas, P.L., Fowler, M.W., Entchev, E.,
2005. Simulation of a tubular solid oxide fuel cell stack using
AspenPlusTM unit operation models. Energy Conversion and
Management 46, 181196.
Zhu, H., Kee, R.J., Janardhanan, V.M., Deutschmann, O., Goodwin,
D.G., 2005. Modeling elementary heterogeneous chemistry
and electrochemistry in solid-oxide fuel cells. Journal of the
Electrochemical Society 152 (12), A2427A2440.

Das könnte Ihnen auch gefallen