Available online at www.sciencedirect.

com

Journal of Power Sources 181 (2008) 313–326

Thermoeconomic modeling and parametric study of hybrid SOFC–gas

turbine–steam turbine power plants ranging from 1.5 to 10 MWe
Alexandros Arsalis ∗
Department of Mechanical Engineering, Virginia Polytechnic Institute and State University, Blacksburg, VA 24061, USA
Received 24 September 2007; received in revised form 26 November 2007; accepted 30 November 2007
Available online 22 January 2008

Abstract
Detailed thermodynamic, kinetic, geometric, and cost models are developed, implemented, and validated for the synthesis/design and operational
analysis of hybrid SOFC–gas turbine–steam turbine systems ranging in size from 1.5 to 10 MWe. The fuel cell model used in this research work
is based on a tubular Siemens-Westinghouse-type SOFC, which is integrated with a gas turbine and a heat recovery steam generator (HRSG)
integrated in turn with a steam turbine cycle. The current work considers the possible benefits of using the exhaust gases in a HRSG in order to
produce steam which drives a steam turbine for additional power output. Four different steam turbine cycles are considered in this research work:
a single-pressure, a dual-pressure, a triple pressure, and a triple pressure with reheat. The models have been developed to function both at design
(full load) and off-design (partial load) conditions. In addition, different solid oxide fuel cell sizes are examined to assure a proper selection of
SOFC size based on efficiency or cost. The thermoeconomic analysis includes cost functions developed specifically for the different system and
component sizes (capacities) analyzed. A parametric study is used to determine the most viable system/component syntheses/designs based on
maximizing total system efficiency or minimizing total system life cycle cost.
© 2007 Elsevier B.V. All rights reserved.

Keywords: Hybrid systems; SOFC; Gas turbine; Steam turbine; Cogeneration

1. Introduction provide power to a large number of residential/commercial

buildings. For example, a 10 MWe hybrid system can fulfill the
The combination of SOFCs with gas turbines is one of the needs of 2000 family residences based on an average four person
most promising power generating applications [1,2]. The SOFC family residence in the US which requires on average 5 kWe.
stack forms the combustor unit in a gas turbine system. Com- To model and then analyze the hybrid system configurations
pressed air is fed into the SOFC stack where fuel is injected as realistically as possible, detailed system/component thermo-
and electrical power drawn off. Operating close to a 46% con- dynamic, kinetic, geometric, and cost models are developed,
version of fuel to electrical power [3], this SOFC then provides implemented, validated and then used to conduct a parametric
pressurized hot gases to a turbine operating at 35% electrical effi- analysis of the key system/component parameters to investigate
ciency. The theoretical overall electrical conversion efficiency of both thermodynamically (efficiency maximization) and eco-
this system can approach 65+%, which can be further improved nomically (total life cycle cost minimization) the advantages
by adding a steam turbine cycle to drive the overall electrical that such hybrid systems might have over conventional GT–ST
efficiency into the mid seventies [1,2]. combined cycle systems, standalone SOFC systems, and hybrid
The objective of this research work is to make a rigorous SOFC–GT systems.
investigation of the design and performance characteristics of
hybrid system configurations consisting of a SOFC, gas tur- 2. System layout
bine, and steam turbine for stationary power applications which
As a starting point for this research work, the modeling, com-
puter code, and optimization results of a previously developed
∗ Tel.: +45 27282984; fax: +357 22 389423. 1.5 MWe hybrid SOFC–GT plant Calise et al. [4–6] are used.
E-mail address: arsalis@vt.edu. In this work, the gas turbine exhaust mixture is re-circulated

0378-7753/$ – see front matter © 2007 Elsevier B.V. All rights reserved.
doi:10.1016/j.jpowsour.2007.11.104
314 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

partment of the fuel cell. The mixture at state point 25 consists

Nomenclature of methane and steam. Thus, in the pre-reformer (PR), the first
step in the fuel reforming process occurs. The energy required
A heat transfer area (m2 )
to support the pre-reforming reaction is derived from the hot
D diameter (m)
stream at state point 26. The non-reacted fuel at state point 2
L length (m)
is involved in the internal reforming reaction within the anode
ṁ mass flow rate (kg s−1 )
compartment of the SOFC stack. Here, it is converted into the
ncell number of cells
hydrogen that participates in the electrochemical reaction.
Q̇ heat transfer rate (kW)
• On the cathode side, air is first preheated by a counter-flow
T temperature (K)
heat exchanger air injection pipe (HEC) and then brought
V volume (m3 )
into the annulus (air pipe) of the SOFC where, at the three-
Ẇ power output (kW)
phase boundaries, the cathode electrochemical reaction occur
Subscripts [1–2,7,8].
cell SOFC stack • The electrochemical reactions, occurring in the fuel cell, pro-
comp compressor duce DC electrical current and release thermal energy. The
cond condenser first of these is converted into AC current by the inverter; the
cool cool water stream latter is used by the internal reforming reaction and to heat up
gas gas side stream the fuel cell stack.
GT gas turbine • The high energy flow rate at state point 8 is first used to preheat
HEC counter-flow heat exchanger air in the counter-flow heat exchanger and then to supply
i inlet stream energy to the pre-reforming reaction. This stream at state point
o outlet stream 21 enters the gas turbine.
PR pre-reformer • The expansion in the GT supplies mechanical power which
pump pump in turn is converted into electric power.
ST steam turbine • The operation of the steam turbine bottoming cycle, e.g., the
steam steam side stream triple pressure with reheat variation (see Fig. 1, bottom part),
can be summarized as follows:
• The GT exhaust stream (state point 33) flows to the heat recov-
ery steam generator (HRSG). The gas mixture side of the
and used to preheat the input air and fuel streams by means HRSG passes through the ten heat exchanger sections – high-
of heat exchangers, while the remaining energy is recovered pressure (HP) superheater (SU), reheater (RH), HP evaporator
to heat water for residential usage, while in the current work, (EV), HP economizer (EC), intermediate-pressure (IP) SU, IP
this system has been modified and expanded to include a sec- EV, IP EC, low-pressure (LP) SU, LP EV, and LP EC – and
ond bottoming cycle, utilizing various types (based on pressure is exhausted at state point 34.
level) of heat recovery steam generators and a steam turbine. In • The superheated steam produced by the HP SU (state point
fact, four configurations are modeled and analyzed in detail here 35) is supplied to the HP stage of the steam turbine. After
with the variations occurring with regard to the steam turbine expansion the cold reheat (state point 64) at an intermedi-
(ST) bottoming cycle, i.e. a single-pressure level, a dual-pressure ate pressure returns to the HRSG and there by means of a
level, and a triple-pressure level with and without reheat. reheater is superheated (state point 66) and returned to the
The purpose of using multiple-pressure levels is to achieve a IP/LP steam turbine stage. Also the IP SU (state point 56)
higher power output from the steam turbine at the expense, of and the LP SU (state point 48) supply superheated steam
course, of extra equipment. The operation of the SOFC–GT to the double-admission IP/LP ST which during expansion
topping cycle (see Fig. 1, top part) can be summarized as produces mechanical power which in turn is converted into
follows: electric power in a generator. A small fraction of super-
heated steam at low pressure is extracted (state point 37)
• Air is compressed by the air compressor (AC) up to the fuel to the deaerator (DE) to be used later on for feedwater
cell operating pressure. The air is then brought to the cathode preheating.
inlet of the SOFC stack (state point 18). Similarly, fuel is • The wet steam (state point 38) is then condensed in the con-
compressed by the fuel compressor (FC) and then brought to denser (CON). The condensate (state point 39) enters the
the anode compartment of the stack (state point 1). condensate pump (CP) and is then pumped to the DE at state
• Both fuel and air can by-pass the fuel cell, i.e. a certain amount point 40.
of fuel can flow directly to the combustor (C) by-passing • In the DE, any air oddments and impurities contained by the
the electrochemical reaction occurring within the stack (state water are removed while the water is preheated at 60 ◦ C. The
point 23), while excess air can flow to the GT (state point 20). preheated water (state points 57, 49, 41) enters the HP FP
• At the stack, fuel (state point 24) is mixed with the anode re- (feedwater pump), IP FP, and LP FP, and is then pumped to
circulation stream (state point 5) in order to support the steam the HP EC, IP EC and LP EC at state points 58, 50, and 42,
reforming reaction in the pre-reformer and in the anode com- respectively.
 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 315

Fig. 1. SOFC–GT integrated with a triple pressure with reheat ST cycle.

• In the economizers, water is heated up to the saturated liquid where V is the actual fuel cell potential, E the open circuit
point. Then it is evaporated at constant temperature/pressure reversible voltage calculated on the basis of the Nernst equa-
in the evaporators. tion, Vact the activation overvoltage calculated on the basis of
• Water and saturated steam are separated in the drums, and the the Butler-Volmer equation and experimental correlations for the
steam is supplied to the superheaters where it is superheated anode and cathode exchange current densities, Vohm the ohmic
to the desired live steam temperatures and fed to the ST to overvoltage, and Vconc the concentration overvoltage. Eq. (1)
repeat the cycle. suggests that in the case of SOFCs, it is possible to neglect
crossover, fuel, and internal current losses. The assumptions
3. Plant model and calculation details for all aforementioned overvoltages are
discussed more extensively in [6].
The thermodynamic, geometric, kinetic and cost models were
developed in MATLAB® . They are used to simulate the behavior
of the hybrid fuel cell system configuration. The model is based 3.2. Pre-reformer model
on the following assumptions: one-dimensional flow; steady
state; no gas leakage; negligible heat losses to the environment; One of the main advantages of using high temperature fuel
negligible kinetic and gravitational terms in the energy balances. cells is the possibility of feeding the SOFC with natural gas
directly, since the reforming process can be supported inside
3.1. Internal reforming SOFC model the stack [1,7–10]. In practice, however, a pre-reforming pro-
cess is usually necessary. The pre-reformer unit consists of a
The overall voltage of the single cell can be calculated as a number of tubes located inside a shell and filled with a particu-
function of current density, temperatures, pressures, chemical lar catalyst [11,12]. The reformate gas flows inside these tubes.
composition, and geometric/material characteristics by calcu- Hot gases, coming from the combustor, flow inside the shell
lating the difference between the reversible potential and all the external to the tubes, supplying the thermal energy needed to
overvoltages [6], i.e. support the process, since the energy provided by the exother-
mic water–gas shift reaction is not sufficient for the endothermic
V = E − Vact − Vohm − Vconc (1) demethanization of the reforming process [11,12]. The pre-
316 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

reformer heat-exchange model is simulated taking into account point temperature is the difference between the evaporator’s
both equilibrium and the kinetics as described in [6]. outlet temperature on the water/steam side and the inlet tem-
perature on the exhaust gas side. The lower the pinch point, the
3.3. Gas turbine cycle model more heating surface is required and the more steam is generated
[18].
The SOFC–GT subsystem utilizes an air compressor, fuel The desired live steam temperatures and pressures are fixed.
compressor, and gas turbine. The air compressor and gas tur- The evaporator drum pressure can be determined based on a
bine are connected together with a single shaft. The shaft is 7–10% loss from the live steam pressure. The pinch points are
also connected to an electric generator converting the mechani- also selected and fixed. The energy balances on the gas and steam
cal power to electrical power. Mass flow rates and rotor speeds sides are:
are corrected on the basis of their inlet conditions according to gas
[4–6,13,14]. Q̇SU+EV = ṁGTexh cp (TSUin − TEVout ) (2)

3.4. Counter-flow heat exchanger air injection pipe, Q̇steam

SU+EV = ṁSTin (hSUout − hEVin ) (3)
catalytic combustor, mixer, inverter, electric generator
The heat transfer rate is determined from Eq. (2), and since the
models
two heat transfer rates on the left hand side of each equation are
equal to each other, Eq. (3) is solved for the live steam mass flow
A counter-flow tube-in-tube heat exchanger is required in
rate. Using simple energy balances, identical to the preceding
order to simulate the heat transfer in the air injection pipe
ones, all temperatures and heat transfer rates can be calculated
between the air flowing through the fuel cell air tube and the
for all the heat exchangers.
stream coming from inside the stack [1–2,5–8,10,14–16]. The
For the geometric models of the heat exchangers, both the
heat exchange is simulated on the basis of existing models in [5]
LMTD and effectiveness-NTU methods are used depending on
and improved to include the effects of pressure drops and to take
the exchanger. The geometric models are needed for determining
into account the dependence of the thermophysical and transport
off-design behavior. All the heat exchangers are shell-and-tube
properties on temperature. The details of the counter-flow heat
since they are the appropriate type for compact heat recovery
exchanger heat transfer model are given in [4].
steam generators [19]. The necessary equations for shell-and-
The combustor burns any non-reacted fuel coming out of
tube heat exchangers are obtained from [19]. A detailed analysis
the fuel cell and, therefore, produces thermal energy for use
of all the heat exchangers’ geometric models can be found in
elsewhere in the system. The hybrid plant makes use of three
[20].
mixers. These are necessary for the operation and the regulation
The axial-flow steam turbine can be single, dual, or triple
of the plant. The exerted electric signal needs to be conditioned
admission depending on the HRSG’s pressure level. Further-
before usage, converted to AC current, and filtered from possi-
more, in the triple-pressure reheat cycle configuration, it is
ble oscillations. This is done by a DC–AC inverter. Similarly,
divided into two sections: a high pressure (HP) section and
the mechanical energy produced by the gas turbine must be con-
an intermediate/low-pressure (IP/LP) section. In this particular
verted to electric power. This conversion is accomplished by an
configuration, the HP section is supplied with live steam from the
electric generator.
superheater. After expansion, the steam returns to the reheater
in the HRSG where it is superheated and supplied to the IP/LP
3.5. Steam turbine cycle model
section for further expansion. Following expansion, the exhaust
is fed to the condenser. All the configurations include extraction
The main components of the steam turbine cycle include a
outlets for deaerating/preheating. The mass and energy balances
steam turbine with an electric generator; a heat recovery steam
for the triple-pressure reheat cycle are
generator which includes economizer(s), evaporator(s), super-
heater(s), and a reheater (triple-pressure reheat cycle only); a ṁST
HPin + ṁRHin + ṁIPin + ṁLPin = ṁRHout + ṁext + ṁout
ST ST ST ST ST ST
(4)
condenser which is dimensioned according to the turbine exit
pressure and mass flow rate as well as ambient conditions; a
deaerator heated by steam extracted from the steam turbine; a triple RH
ẆST = ṁST
HPin hHPin + ṁRHin hRHin +ṁIPin hIPin +ṁLPin hLPin
ST ST ST ST ST ST ST
condensate pump; and one to three feedwater pumps.
The HRSG model calculates the live steam mass flow rates − ṁST
RHout hRHout − ṁext hext − ṁout hout
ST ST ST ST ST
(5)
and also the exhaust gas conditions at the HRSG exit. In addi-
tion, it sizes the different types of heat exchangers included in where ṁSTHPin , ṁRHin , ṁIPin , and ṁLPin are the mass flow
ST ST ST

the HRSG. Depending on the HRSG’s number of pressure lev- rates of the HP, RH, IP, and LP superheated steam entering the
els, the corresponding live steam mass flow rates are calculated. steam turbine, respectively, and ṁSTRHout the mass flow rate of the
The water/steam conditions at the inlet and exit of every heat steam after expansion in the HP section of the steam turbine.
exchanger are defined either directly by the desired live steam For off-design purposes steam turbine maps are used in order to
conditions or indirectly through conditions on the saturation capture the effects of geometry on turbine performance. To gen-
curve [17]. An important parameter defining the heating surface erate these maps for different size turbines, data is taken from
and performance of the HRSG is the pinch point. The pinch- [21].
 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 317

The steam turbine cycle includes a condensate pump and one steam turbine extraction, ṁdea
out the mass flow rate of the deaer-
to three feedwater pumps depending on the number of HRSG ated/preheated water exiting the deaerator. An energy balance
pressure levels. Since the thermodynamic states in the inlet are on this component yields
known and the outlet thermodynamic states can be fixed as
desired, what is left is a calculation of the pump power con- out hout − ṁext hext − ṁin hin = 0
ṁdea dea dea dea dea dea
(12)
sumed. The corresponding mass and energy balances are given
by out , hext , and hin are the enthalpies for the correspond-
where hdea dea dea

ing mass flow rates.

pump pump
ṁin = ṁout (6)
4. Cost model
Ẇpump = [ṁin (hout − hin )]pump (7)

where ṁin is the non-pressurized mass flow rate entering the For the thermoeconomic analysis of the plant, appropriate
pump, ṁout the pressurized mass flow rate exiting the pump, cost functions must be formulated to include the purchase cost
pump pump
Ẇpump the pump work rate consumption, and hin and hout for every component, the capital cost per annum, the operating
are the corresponding enthalpies for the mass flow. Again, for cost per annum, and the total cost per annum. The expressions
off-design purposes, pump maps are rescaled and modified from for all the component purchase costs are summarized in detail
actual pump maps found in the literature. For the condensate in Table 1, while those for the capital, operating, and total costs
pump, a map from [22] for a centrifugal type pump is used per annum are summarized in Table 2.
while for the feedwater pumps, a map for a displacement type For the gas turbine, the cost function proposed by [24] is used.
pump from [23] is employed. For the centrifugal compressors (air and fuel compressors), the
The condenser, which is a shell-and-tube heat exchanger, corresponding costs are calculated by interpolating data from the
receives wet steam from the steam turbine’s exhaust and con- manufacturers as a function of the maximum power required and
denses it to a saturated liquid. In the condensing process, the using information provided by [25]. For the counter-flow heat
temperature and pressure are kept constant. For the purposes of exchanger, the capital cost is determined on the basis of a cost
this study, they have been fixed at 31 ◦ C and 0.045 bar as in [18]. function from [26] while the cost of the SOFC stack is esti-
The working side mass balance is: mated with reference to market studies in which the expected
cost for the case of a significant increase in production volume
ṁcond
in = ṁcond
out (8) is assumed. A detailed work performed by [9] relates the SOFC
purchase cost to the active area and the operating temperature.
where ṁcond
in is the mass flow rate of the wet steam entering the Furthermore, the electric energy produced by the SOFC must be
condenser and ṁcond
out the mass flow rate of the saturated liquid filtered by an inverter, whose cost is not negligible and should,
exiting. The heat rejected to the cooling water is found from an therefore, be taken into account [9]. The SOFC system also con-
energy balance on the condensing steam, i.e. sists of a pre-reformer, whose cost is calculated on the basis of its
Q̇cond = [ṁcond cond
− hcond catalysts volume and the finned exchange area [11–12,26] which
in (hin out )] (9)
in turn is related to the number, diameter, and length of tubes.
where Q̇cond is the rejected heat transfer rate and hcond
in , hout are
cond Thus, based on these references and updating the functions with
the enthalpies for the corresponding mass flow rates. The cooling literature data, the pre-reformer component cost function is for-
water mass flow rate can be calculated by an energy balance on mulated by [4]. The total cost for SOFC auxiliary devices such
the cooling water entering and exiting the condenser: as the combustor, mixers, and by-pass valves are calculated as a
fixed percentage (10%) of the stack cost.
Q̇cond For the steam turbine cycle, all cost equations, except that for
ṁcw = (10)
(Tcw,out − Tcw,in )Cpcw the steam turbine, are based on [17] and have been appropriately
adjusted for inflation by using [27]. For the steam turbine, the
where ṁcw is the mass flow rate of the cooling water, Tcw,in
cost function, which is based on the steam turbine power output,
and Tcw,out are the inlet and outlet cooling water temperatures,
is developed based on [28]. For the HRSG (which includes the
respectively, and Cp,cw is the average cooling water specific heat.
drum and piping costs), the total cost is composed of the cost for
The LMTD method is applied to the geometric analysis of the
the various heat exchangers, the piping, the gas conduit, and the
condenser analyzed in detail in [20].
pump. It is based on a function used by [29]. The total cost of the
The deaerator removes dissolved gases and impurities from
heat exchangers is formed by the sum of the cost for the various
the condensate by keeping it in a reservoir at the state of a sat-
heat exchange units (e.g., HP superheater, HP evaporator, HP
urated liquid absorbing heat extracted from the steam turbine
economizer, reheater, etc.) indicated by the index i. Also the
at a pressure slightly higher than the deaerator pressure. The
LMTD correction factor, Ki is based on the logarithmic mean
corresponding mass and energy balances are
temperature difference, Tlm,i , while fpi , fTi,steam , and fTi,gas
ṁdea are cost correction factors. The cost functions for the piping
in + ṁext = ṁout
dea dea
(11)
and the gas conduit include the factors fpi , fTi,steam , and fTi,gas .
where ṁdea
in is the mass flow rate of the saturated liquid com- The factors introduce a sensitivity of cost to pressure as well
ing from the condensate pump, ṁdea
ext the mass flow rate of the as to steam and exhaust gas temperature. The pressure factor is
318 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

Table 1
Component cost models
Variable description Model equation

CGT Gas turbine component cost ($) CGT = (−98.328 ln(ẆGT ) + 1318.5)ẆGT
0.67
Ccomp Compressor component cost ($) Ccomp = 91562(Ẇcomp /445)
CHEC Counter-flow heat exchanger component cost ($) CHEC = 130(AHEC /0.093)0.78
CSOFC SOFC stack component cost ($) CSOFC = (ncells πDcell Lcell )(2.96Tcell − 1907)
0.70
Cinv Inverter component cost ($) Cinv = 105 (Ẇcell /500)
CPR Pre-reformer component cost ($) CPR = 130(APR,fin /0.093)0.78 + 3240(VPR )0.4 + 21,280.5VPR
Caux,SOFC SOFC auxiliary components cost ($) Caux,SOFC = 0.10CSOFC
0.7 0.95
CST Steam turbine component cost ($) CST = 3644.3(ẆST ) − 61.3(ẆST )
fpi Heat exchanger pressure factor fpi = 0.0971(pi /30) + 0.9029
fT,steam Steam-side temperature factor fT,steam = 1 + exp(Tout,steam − 830/500)
fT,gas Gas-side temperature factor fT,gas = 1 + exp(Tout,gas − 990/500)
Ki LMTD correction factor (kW K−1 ) Ki = (Q̇i /Tlm,i )

CHE(HRSG) HRSG’s heat exchangers component cost ($) CHE(HRSG) = 3650 (fpi fTi,steam fTi,gas K0.8 )i
i
fpj Piping pressure factor fpj = 0.0971(pj /30) + 0.9029

Cpiping HRSG’s piping component cost ($) Cpiping = 11, 820 (fpj ṁj,steam )
j
Cgas HRSG’s gas conduit cost ($) Cgas = 658ṁ1.2
gas
CHRSG HRSG component cost ($) CHRSG = CHE(HRSG) + Cpiping + Cgas
Ccond Condenser component cost ($) Ccond = 248Acond + 659ṁcool
fη Efficiency correction factor Fη = 1 + (1 − 0.8/1 − ηpump )
0.71
Cpump Pump component cost ($) Cpump = 442(Ẇpump ) 1.41fη

calculated as a function of live steam pressure Pi and comes from This loss of value needs to be distributed over the lifetime of the
curve fit data for heat exchangers found in [26]. The temperature component. This results in a realistic estimation of the cost of the
factors are developed using the [29] form of the temperature equipment and indicates how much money has to be spent every
correction factors and the fact that the investment of superheaters year in order to save money for future replacement or to pay
is about twice as high as the investment cost for evaporators [17]. back loans if the equipment was purchased with outside capital.
The temperature values indicating technical limits are taken from In the context of thermoeconomic modeling, the common linear
[17]. depreciation method is used for this cost estimation. Therefore,
For the condenser, the cost function is based on [29]. It is the annual depreciation cost is determined by dividing the total
calculated as a function of the condenser surface area, Acond , purchase cost, Cpur , by the depreciation time, ndep , measured in
and the cooling water mass flow rate, ṁcool . For the deaerator, years. For this research work, ndep has been assumed to be 10
the cost function is formulated using a cost function found in years [4].
[26]. The cost function for the pumps is taken from [29] and The purchase or capital cost must also be financed from
calculates the cost as a function of the electric power consumed, outside sources such as bank loans. The associated interest is
Ẇpump , and an efficiency correction factor fη . considered a cost [30]. For the current research work, some
The capital cost must be placed on an annual basis in order simplifying assumptions are made: (i) a single interest rate is
to account for the cost of the investment required. This annual assumed for the cost of borrowed capital as well as for the
cost is composed of the depreciation cost, Ċdep , interest on the opportunity cost of having invested ones own capital and (ii)
investment, Ċint , maintenance cost, Ċmai , insurance cost, Ċins , the capital cost is distributed over the lifetime or depreciation
and tax cost, Ċtax . The depreciation cost is based on the fact that time of the plant. The interest rate, i, is assumed to be 0.0926
the equipment deteriorates with time [30] and, thus, looses value. [30].

Table 2
Capital, operating, and total cost models
Variable description Model equation

Ċdep Depreciation cost ($ year−1 ) Ċdep = Cpur /ndep , Ċint = (Cpur /ndep )i,
Ċint Interest on outside capital cost ($ year−1 ) Ċmai = (Cpur /ndep )fmai
Ċmai Maintenance cost ($ year−1 )
Ċins Insurance cost ($ year−1 ) Ċins = (Cpur /ndep )fins ,
Ċtax Taxation cost ($ year−1 ) Ċtax = (Cpur /ndep )ftax
Ċcap Capital cost ($ year−1 ) Ċcap = Ċdep + Ċint + Ċmai + Ċins + Ċtax , Ċope = cf V̇f Nh
Ċope Operating cost ($ year−1 ) Ċtotal = Ċcap + Ċope
Ċtotal Total cost ($ year−1 )
 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 319

The maintenance cost may vary over the lifetime of an instal- 6. Parametric study: results and discussion
lation as the equipment degrades and depends largely on the
number of operating hours, the frequency of shutdowns and star- The purpose of the parametric study is to either maximize
tups, and the operating environment. A total maintenance cost the average efficiency or minimize the total cost of the hybrid
for the above suggests annual maintenance expenses on the order SOFC–GT–ST power plant. To achieve this, a systematic vari-
of 6% of the annual depreciation cost [11]. Thus, fmai which is ation in the values of a number of key decision variables as
the maintenance cost factor is 0.06. Similarly, the insurance and well as the relative sizes of the SOFC, GT, and ST for a given
taxation cost factors are chosen as 0.2 and 0.54%, respectively hybrid plant size must be made. The key decision variable ranges
[17]. and initial values are chosen on the basis of typical hybrid fuel
The operating cost per annum is based on [4]. In this cost cell systems found in [4–6]. The fuel utilization factor is varied
function, cf is the cost of fuel in $ Nm−3 , V̇f is the volumetric from 0.75 to 0.90 in 0.05 increments. Values below 0.75 are not
flow rate of the fuel in Nm3 h−1 , and Nh are the annual hours of applicable because such values cause an increase in temperature
operation. The latter is assumed to be 8760 h year−1 [4]. Once beyond the maximum possible turbine inlet temperature (TIT)
both the annual operating cost and the capital cost per annum are and, therefore, a coupling point of the air compressor and the
known, the total cost per annum becomes the sum of the annual gas turbine cannot be reached. In addition, lower fuel utilization
capital and operating costs. factors result in lower efficiencies since fuel is converted mainly
in the combustor.
5. Model validation The steam-to-carbon ratio is varied from 2 to 3.5 in 0.5
increments. Values below 2 are not included in order to avoid
The SOFC–GT–ST hybrid system needs to be validated in problems of carbon deposition on the anode of the SOFC stack
order to have confidence in the model predictions. The vali- as reported in [1–2,8].
dation procedure helps determine the degree of accuracy of The SOFC operating temperature is varied from 950 to
the model and any possible mismatches and discrepancies. 1100 ◦ C in 50 ◦ C increments. A value beyond 1100 is not used
Each subsystem of the hybrid system is validated separately because it would exceed the operating limit of the SOFC. Also,
using manufacturer’s data for the SOFC model [4–6] and a value lower than 950 ◦ C at full load (i.e. at the design point) is
measured data from the literature for the steam turbine cycle infeasible since the minimum part load (25%) SOFC operating
models. temperature is not high enough to heat the SOFC stack.
As mentioned previously, the steam turbine cycle subsys- The SOFC operating pressure is varied from 7 to 10 bar in
tem models were modeled based on [18]. Fortunately, the same 1 bar increments. A value beyond 10 bar is not used because the
source includes measured data for the performance of all four differential pressure between the anode and the cathode com-
steam turbine cycle configurations. This data includes temper- partments of the SOFC would exceed the operating limits set by
atures, pressures, mass flow rates, and steam turbine power the simulation process as described in [6].
output. Detailed comparisons of these data can be found in Finally, the unit cost of fuel is varied as follows: 0.1, 0.3, 0.6,
[20]. The comparison indicates good agreement between the 0.9, and 1.2. The initial value of 0.3 is the one published in [31]
measured values from the literature and those calculated from for the year 2005.
the models. The few configurational differences which do exist Three different hybrid plant sizes are considered: 1.5, 5, and
are justified by the differences which exist between the cycles 10 Mwe, where two different SOFC sizes are considered based
found in the literature and the model configurations used here. on current density. The operating current density for the selected
These differences include the following: simplification of the fuel cell operates from 100 to 650 mA cm−2 [32]. Since the cur-
HRSG’s heat exchangers from dual flow to single flow, a differ- rent density decreases at off-design, the small SOFC is selected
ent chemical composition for the GT exhaust gases, a pressure based on a maximum possible current density of 550 mA cm−2
loss (2%) for every heat exchanger in the HRSG for all model (full load or design condition) while the large SOFC is selected
calculations different from that in [18], and a model pump power based on a minimum possible current density of 100 mA cm−2
expenditure calculation somewhat different than that found in (25% full load condition). The larger SOFC yields higher effi-
[18]. The slightly lower live steam production, and in effect the ciencies as compared with the smaller SOFC but the latter has a
slightly lower ST power output for all model configurations as lower capital cost which is significant since the SOFC purchase
compared with [18], is mainly caused by the absence of a GT cost is the most dominant of all the equipment purchase costs.
cooler. In the parametric study conducted here, the smaller SOFC mini-
A second validation was also made to check in more detail mizes the total cost while the larger one maximizes the efficiency.
the steam production using results more suitable to fuel cell Also, the gas turbine and steam turbine sizing is determined
hybrid systems. These results are from a simple, single-pressure based on the extra power needed to reach the desired total power
HRSG. These were found in [2]. The model and literature results output.
comparison can be found in [20]. The minor differences between The performance of each individual system is analyzed at
the literature and the model results are caused by the assumptions full and part load conditions to determine the average and total
of a constant specific heat throughout the HRSG in the literature efficiencies and total operating cost. The load profile is based
results as well as of no heat exchanger pressure losses in the on a 2-day (one average winter day and one average summer
literature results. day), electrical power demand profile for an average four person
320 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

family household [33] scaled appropriately to coincide with the Table 3

three different sized hybrid plants analyzed here and extended Time intervals for Eq. (13)
over an entire year. The average efficiency for the plant becomes Variable description Value (days) Variable description Value (days)
a time-averaged value based on the time intervals. It is defined t1 25% load 137 t3 75% load 106
as follows: t2 50% load 46 t4 100% load 76

t1 η1 + t2 η2 + t3 η3 + t4 η4

ηave =  (13)
ti In addition, the total efficiency is defined as

Ẇi ti
where the t1 , t2 , t3 , and t4 are the time intervals cor- ηtot =  (14)
Q̇H ti
responding to 25, 50, 75, and 100% loads (values shown in
Table 3), and the η1 , η2 , η3 , and η4 are the hybrid plant energetic where Ẇi is the power output corresponding to 25, 50, 75,
electrical efficiencies corresponding to the above time intervals. and 100% load, and Q̇H the heat input to the system based

Table 4
Parametric results for the 10 MWe triple-pressure w/RH ST hybrid system
Decision variables Objective functions Power output (MW)

Uf TSOFC S/C pSOFC cfuel ηave Ctotal ηtot ẆSOFC ẆGT ẆST

Hybrid system employed with a small SOFC

0.75 1000 2 8 0.3 0.5864 3,846,790 0.6240 7.863 2.308 1.337
0.8 1000 2 8 0.3 0.5852 3,852,258 0.6213 7.902 2.356 1.398
0.85 1000 2 8 0.3 0.5889 3,637,267 0.6242 7.958 2.400 1.454
0.9 1000 2 8 0.3 0.5685 3,741,024 0.6039 8.001 2.447 1.510
0.85 950 2 8 0.3 0.5665 3,690,959 0.6030 7.999 2.439 1.407
0.85 1000 2 8 0.3 0.5889 3,637,267 0.6242 7.958 2.400 1.454
0.85 1050 2 8 0.3 0.5960 3,660,018 0.6300 7.902 2.358 1.506
0.85 1100 2 8 0.3 0.6032 3,692,544 0.6368 7.856 2.310 1.548
0.85 1000 2 8 0.3 0.5889 3,637,267 0.6242 7.958 2.400 1.454
0.85 1000 2.5 8 0.3 0.5822 3,653,267 0.6178 8.006 2.447 1.405
0.85 1000 3 8 0.3 0.5810 3,644,914 0.6166 8.059 2.489 1.360
0.85 1000 3.5 8 0.3 0.5716 3,689,910 0.6079 8.102 2.537 1.307
0.85 1000 2 7 0.3 0.5848 3,645,623 0.6192 7.906 2.354 1.403
0.85 1000 2 8 0.3 0.5889 3,637,267 0.6242 7.958 2.400 1.454
0.85 1000 2 9 0.3 0.5869 3,641,883 0.6214 8.001 2.441 1.498
0.85 1000 2 10 0.3 0.5852 3,648,621 0.6197 8.054 2.496 1.543
0.85 1000 2 8 0.1 0.5889 1,744,829 0.6242 7.958 2.400 1.454
0.85 1000 2 8 0.3 0.5889 3,637,267 0.6242 7.958 2.400 1.454
0.85 1000 2 8 0.6 0.5889 6,415,426 0.6242 7.958 2.400 1.454
0.85 1000 2 8 0.9 0.5889 9,217,784 0.6242 7.958 2.400 1.454
0.85 1000 2 8 1.2 0.5889 12,032,142 0.6242 7.958 2.400 1.454
Hybrid system employed with a large SOFC
0.75 1000 2 8 0.3 0.6298 3,798,306 0.6616 8.303 2.154 1.187
0.8 1000 2 8 0.3 0.6367 3,756,709 0.6659 8.347 2.203 1.126
0.85 1000 2 8 0.3 0.6529 3,697,652 0.6836 8.391 2.254 1.079
0.9 1000 2 8 0.3 0.6304 3,781,486 0.6605 8.442 2.304 1.005
0.85 950 2 8 0.3 0.6447 3,625,057 0.6741 8.440 2.997 1.007
0.85 1000 2 8 0.3 0.6529 3,697,652 0.6836 8.391 2.254 1.079
0.85 1050 2 8 0.3 0.6412 3,834,151 0.6707 8.344 2.201 1.130
0.85 1100 2 8 0.3 0.6386 3,926,399 0.6677 8.301 2.149 1.189
0.85 1000 2 8 0.3 0.6529 3,697,652 0.6836 8.391 2.254 1.079
0.85 1000 2.5 8 0.3 0.6433 3,737,498 0.6722 8.442 2.305 1.011
0.85 1000 3 8 0.3 0.6357 3,748,615 0.6657 8.496 2.347 0.952
0.85 1000 3.5 8 0.3 0.6337 3,774,777 0.6641 8.534 2.396 0.894
0.85 1000 2 7 0.3 0.6368 3,823,306 0.6658 8.346 2.203 1.008
0.85 1000 2 8 0.3 0.6529 3,697,652 0.6836 8.391 2.254 1.079
0.85 1000 2 9 0.3 0.6348 3,840,558 0.6644 8.435 2.301 1.136
0.85 1000 2 10 0.3 0.6316 3,871,101 0.6610 8.493 2.348 1.189
0.85 1000 2 8 0.1 0.6529 1,985,021 0.6836 8.391 2.254 1.079
0.85 1000 2 8 0.3 0.6529 3,697,652 0.6836 8.391 2.254 1.079
0.85 1000 2 8 0.6 0.6529 6,281,601 0.6836 8.391 2.254 1.079
0.85 1000 2 8 0.9 0.6529 8,859,549 0.6836 8.391 2.254 1.079
0.85 1000 2 8 1.2 0.6529 11,449,497 0.6836 8.391 2.254 1.079
 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 321

on the LHV and fuel flow rates for the aforementioned time From an observation of the trend lines shown in these figures,
intervals. some general remarks can be made that apply to all the hybrid
Results for all the hybrid systems modeled and simulated plants analyzed here. The optimum fuel utilization factor (Uf ) is
here are analyzed thoroughly in [20]. A representative set of 0.85. For the small SOFC where the cost is the optimizing func-
results for the parametric study performed for the 10 MWe tion of interest, the minimum total cost is achieved at this value
SOFC–GT–ST hybrid system with a triple pressure with reheat because although the capital cost decreases slightly at lower val-
ST cycle is shown in Table 4 and Figs. 2 and 3. The top half of ues of Uf , the more dominating operating cost is decreased at
Table 4 shows the results for the small SOFC, while the bottom higher values of Uf . Therefore, the higher efficiency is the main
half shows those for the large SOFC. The objective function is reason that the higher fuel utilization is more economical even
the average efficiency when using a large SOFC and the total though only slightly so.
cycle cost for the small SOFC. Figs. 2 and 3 show the deci- For the large SOFC, where the average efficiency is the opti-
sion variable values plotted against the corresponding objective mizing function of interest, as expected, the higher the fuel
function values. utilization factor is the higher the efficiency. The optimum value

Fig. 2. Optimizing variables vs. objective functions for the 10 MWe triple-pressure w/RH ST small SOFC.
322 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

Fig. 3. Optimizing variables vs. objective functions for the 10 MWe triple-pressure w/RH ST large SOFC.

is 0.85 and not 0.90 because at the latter value although the duced by the internal reforming reactions is consumed within
SOFC efficiency is slightly higher, the total plant efficiency the fuel cell by the anode electrochemical reaction and more
drops because not much heat is left for recovery by the gas efficient stacks release less heat. Therefore, the electrical power
turbine and the steam turbine since almost all the hydrogen pro- produced by the SOFC increases, causing a raise in its electro-
 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 323

chemical rate of reaction. On the other hand, this effect decreases Table 5
significantly the turbomachinery efficiencies. Optimal component costs for the 10 MWe triple-pressure w/RH ST hybrid
system
The optimum SOFC operating temperature for both SOFC
sizes and all models is 1000 ◦ C. For the small SOFC, where Component Cost
the total cost is the optimizing function, the lower the SOFC Large SOFC Small SOFC
operating temperature the lower the SOFC capital cost. This
SOFC 5,031,872 2,334,223
trend reaches a minimum at 1000 ◦ C and not 950 ◦ C because GT 1,595,840 1,662,272
the operating cost manages to slightly overcome the capital Combustor/mixer 1,236,623 941,679
cost. For the large SOFC, where the average efficiency is the HRSG 584,085 660,761
optimizing function, the best SOFC operating temperature is ST 438,749 536,340
still 1000 ◦ C because although the SOFC stacks operate slightly Inlet air tubes 345,616 189,093
AC 258,163 264,960
more efficiently at increased temperatures, the air and fuel Deaerator 57,278 67,761
compressors require significantly higher power consumption, Pre-reformer 45,124 36,994
thereby decreasing slightly the overall efficiency. FC 41,321 42,856
The steam-to-carbon ratio reaches its optimum efficiency and Condenser 8,262 8,262
lower cost at the lowest possible value of 2. The capital cost Pumps 7,958 9,734
Total investment 9,650,891 6,754,935
remains constant throughout the S/C variation; and, therefore,
the only significantly varying cost is the operating cost which
will be determined by the overall efficiency. The average effi-
ciency slightly decreases as the S/C increases since higher steam sure is increased beyond a certain level. Therefore, the optimum
partial pressures cause higher Nernst overvoltages [6]. Thus, efficiency for both the total cost and efficiency objectives is at
higher efficiencies and lower operating/total costs are achieved an intermediate point (8 bar) where the capital cost is not as
at lower S/C values. high as compared to a capital cost for an even higher pressure
Theoretically increasing the SOFC operating pressure will where the capital cost of the turbomachinery would increase to
result in an increase of the cell voltage because of the higher reac- accommodate the higher pressure needed. On the other hand,
tant partial pressure available, therefore, improving the SOFC a lower operating pressure causes lower efficiencies and, there-
efficiency. On the other hand, the expansion of the gas in the fore, although the capital cost is lower the operating cost is higher
turbomachinery is in a temperature region where the turbo- resulting in an increased total cost.
machinery produces less power as the pressure increases as Finally, the unit cost of fuel variation helps determine how
indicated in [34]. Also, the additional thermal input to the sys- efficiently the overall hybrid plant should be designed. At lower
tem for delivering air and fuel at the desired conditions leads costs of fuel, e.g., 0.1 and 0.3 $ Nm−3 , the capital cost competes
to a decreasing trend for the overall efficiency when the pres- more evenly with the operating cost to determine the optimum

Table 6
Cost and efficiency breakdown for all optimal models
Configuration Description ẆSOFC (MW) Total cost Operating cost Capital cost ηave ηtot ηmax

1 1.5 MWe single-pressure ST small SOFC 1.275 604,743 438,971 165,772 0.5696 0.6022 0.6638
2 1.5 MWe single-pressure ST large SOFC 1.330 613,655 405,328 208,327 0.6235 0.6525 0.6971
3 1.5 MWe dual-pressure ST small SOFC 1.275 602,033 435,146 166,887 0.5705 0.6031 0.6647
4 1.5 MWe dual-pressure ST large SOFC 1.330 611,026 401,575 209,451 0.6243 0.6533 0.6978
5 1.5 MWe triple-pressure ST small SOFC 1.238 599,321 431,428 167,893 0.5712 0.6035 0.6654
6 1.5 MWe triple-pressure ST large SOFC 1.314 608,046 397,831 210,215 0.6251 0.6536 0.6987
7 1.5 MWe triple RH-pressure ST small SOFC 1.234 598,691 430,598 168,093 0.5716 0.6038 0.6657
8 1.5 MWe triple RH-pressure ST large SOFC 1.300 607,446 397,031 210,415 0.6255 0.6539 0.6992
9 5 MWe single-pressure ST small SOFC 4.116 1,887,264 1,437,339 449,925 0.5791 0.6131 0.6795
10 5 MWe single-pressure ST large SOFC 4.333 1,935,053 1,340,421 594,632 0.6264 0.6573 0.7115
11 5 MWe dual-pressure ST small SOFC 4.069 1,872,891 1,411,966 460,925 0.5808 0.6149 0.6815
12 5 MWe dual-pressure ST large SOFC 4.325 1,920,649 1,316,017 604,632 0.6279 0.6589 0.7132
13 5 MWe triple-pressure ST small SOFC 4.054 1,871,891 1,405,966 465,925 0.5814 0.6155 0.6825
14 5 MWe triple-pressure ST large SOFC 4.317 1,924,649 1,310,017 614,632 0.6286 0.6596 0.7142
15 5 MWe triple RH-pressure ST small SOFC 4.037 1,869,891 1,402,966 466,925 0.5817 0.6159 0.6837
16 5 MWe triple RH-pressure ST large SOFC 4.281 1,923,649 1,307,017 616,632 0.6289 0.6601 0.7157
17 10 MWe single-pressure ST small SOFC 7.999 3,630,071 2,831,362 798,710 0.5876 0.6222 0.6922
18 10 MWe single-pressure ST large SOFC 8.495 3,717,621 2,603,916 1,113,705 0.6473 0.6766 0.7272
19 10 MWe dual-pressure ST small SOFC 7.929 3,630,067 2,819,358 810,710 0.5880 0.6228 0.6943
20 10 MWe dual-pressure ST large SOFC 8.485 3,717,624 2,591,919 1,125,705 0.6478 0.6772 0.7292
21 10 MWe triple-pressure ST small SOFC 7.905 3,640,067 2,805,358 834,710 0.5884 0.6234 0.6961
22 10 MWe triple-pressure ST large SOFC 8.474 3,715,624 2,577,919 1,137,705 0.6480 0.6774 0.7292
23 10 MWe triple RH-pressure ST small SOFC 7.958 3,606,144 2,822,572 783,573 0.5891 0.6242 0.6973
24 10 MWe triple RH-pressure ST large SOFC 8.391 3,697,652 2,577,948 1,119,703 0.6529 0.6836 0.7370
324 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

Fig. 4. Cost breakdown for all optimal configurations.

system. On the other hand, at higher values of unit cost of fuel, All twenty-four system configurations are compared in terms
e.g., 0.9 and 1.2 $ Nm−3 , the operating cost increases far and of cost (total, operating, and capital) and efficiency (maximum,
away above the capital cost and, therefore, a minimization of total, and average) in Table 6. From Fig. 4 it can be concluded that
fuel consumption is required. In such a case, a more efficient as expected the higher the degree of complexity in the heat recov-
system is required, and, therefore, the large SOFC, although ery steam generator, the higher the capital cost is. On the other
having a higher capital cost than the smaller SOFC would be hand, the operating cost decreases as this complexity increases
selected. because of the increasing efficiency. This decrease in operating
The component cost breakdown for the 10 MWe triple- cost is significant enough to balance and even slightly decrease
pressure reheat ST hybrid system is shown in Table 5. As the total cost for a more complex system. In terms of SOFC
expected, the most dominant component cost is the SOFC size, the configurations equipped with a smaller SOFC also as
purchase (depreciation) cost (especially when using the large expected have a total cost lower than those with the larger SOFC.
SOFC), while the turbomachinery, combustor, and HRSG costs Finally, from Fig. 5 it is evident that the efficiency is higher
are also significant. for a system equipped with a larger SOFC than those with a

Fig. 5. Efficiency breakdown for all optimal configurations.

 A. Arsalis / Journal of Power Sources 181 (2008) 313–326 325

smaller SOFC for any given configuration pair. In addition, a achieve the maximum benefits offered by the hybrid system.
global comparison of all twenty-four systems shows that the The air compressor has an unexpectedly high consumption of
most efficient system is a system at the larger power capacity power (about 75% of the gas turbine power output) meaning
level, i.e. the 10 MWe configuration is more efficient than the that a study must be made in order to minimize this trend if
5 MWe configuration or the 5 MWe configuration is more effi- possible. Based on current technology, the SOFC operates at
cient than the 1.5 MWe configuration. This is mainly due to the pressures of 7–9 bars meaning that a high degree of compres-
higher efficiencies achieved by the turbomachinery (gas turbine, sion is required to fulfill this need. New developments in lower
steam turbine, air compressor, fuel compressor, pumps) at the temperature and pressure SOFCs may benefit the overall system
larger capacities. since less compression will be required.
Since the SOFC–GT–ST system involves a large amount of
7. Conclusions equipment with a much larger number of decision variables
than actually considered in this parametric study, a more com-
The high efficiencies developed by some of the hybrid sys- plete optimization of the systems should be done in order to
tems are of great interest since they show the potential for determine more detailed syntheses/designs than those presented
exceeding those of the best commercial heat engine cycles here.
currently available or projected. For instance, the 10 MWe Finally, since SOFCs are not fully commercialized, a more
SOFC–GT–ST hybrid triple pressure with reheat system exhibits accurate economic analysis than that made here cannot be made
efficiencies (maximum efficiency of 73.8%, an average effi- at this time. The high capital cost suggested in this research
ciency of 65.3%, and a total efficiency of 68.4%) that cannot work (even adjusted for production volume) could decrease in
be matched by other conventional and non-conventional cycles the near future leading to minimized cost syntheses/designs
(e.g., a standalone SOFC, SOFC–GT hybrid cycles, etc.). Inter- which exhibit even higher efficiencies than those determined
estingly, the SOFC–GT–ST system develops high efficiencies here.
at off-design conditions as well. However, the off-design strat-
egy followed in this research work was a simplistic one which References
involved the lowering of the fuel flow rate while keeping con-
stant the air flow rate. This strategy was necessary because a [1] J. Larminie, A. Dicks, Fuel Cell Systems Explained, John Wiley, Chich-
constant air-to-fuel ratio strategy (which maintains high effi- ester, West Sussex, 2003.
ciencies) has a very restricted field of operation (above 80% [2] Fuel Cell Handbook, seventh ed., Office of Fossil Energy, National Energy
of full load) not applicable for the load profile (down to 25% Technology Laboratory, Morgantown, West Virginia, November, 2004.
[3] Siemens Power Generation, http://www.powergeneration.siemens.com,
of full load) considered in this research study. Therefore, this 2007.
simplistic strategy creates conditions at which, for example, the [4] F. Calise, Modellazione, Analisi Exergetica ed Ottimizzazione Termoeco-
SOFC operating temperature drops significantly, leading to the nomica di Cicli Ibridi SOFC–GT, Università Degli Studi di Napoli Federico
difficulty of maintaining higher efficiencies than those actually II, Napoli, Italy, 2005.
exhibited. Thus, a more in-depth analysis of the off-design strat- [5] F. Calise, M. Dentice d’Accadia, A. Palombo, L. Vanoli, R. Vanoli, Third
International Symposium on Energy and Environment, Sorrento, Italy,
egy should be done to see if a better strategy can indeed be found 2004.
(e.g., the removal of part of the SOFC stack at lower loads could [6] F. Calise, M. Dentice d’Accadia, L. Vanoli, M.R. von Spakovsky, J. Power
conceivably maintain higher efficiencies). Sources 159 (2) (2006) 1169–1185.
The parametric study identified a number of unforeseen com- [7] S.C. Singhal, K. Kendall, High Temperature Solid Oxide Fuel Cells: Fun-
plexities which only became evident after the integration and damentals, Design, and Applications, Elsevier, Oxford, New York, 2003.
[8] T.G. Benjamin, E.H. Camera, L.G. Marianowski, Handbook of Fuel Cell
development of the total system configurations. These difficul- Performance, Institute of Gas Technology, 1995.
ties included the proper selection of the SOFC stack size and [9] S.H. Chan, C.F. Low, O.L. Ding, J. Power Sources 103 (2) (2002) 188–200.
the difficulty of finding the proper steam turbines to match the [10] S. Campanari, Power Plants Based on Solid Oxide Fuel Cells Combined
system. For a realistic system, a 1.5 MWe SOFC–GT–ST is not With Gas Turbine Cycles, PhD Dissertation, Politecnico di Milano, Milano,
as attractive and efficient as a 5 or a 10 MWe system because the Italy, 1998.
[11] N. Georgopoulos, Application of a Decomposition Strategy to the Optimal
gas turbine and especially the steam turbine are very inefficient Synthesis/Design and Operation of a Fuel Cell Based Total Energy System,
at small sizes resulting in lower overall system efficiencies. M.S. Thesis, advisor: M.R. von Spakovsky, Virginia Polytechnic Institute
A careful selection of component designs in the steam turbine & State University, Blacksburg, VA, 2002.
cycle was made to achieve efficient conversion of the thermal [12] B. Oyarzabal, Application of a Decomposition Strategy to the Optimal
energy to power output based on the thermodynamic, geometric, Synthesis/Design of a Fuel Cell Sub-system, M.S. Thesis, Department of
Mechanical Engineering, advisor: M.R. von Spakovsky, Virginia Polytech-
and cost models. The uniqueness of the system required in many nic Institute & State University, Blacksburg, VA, 2001.
instances the rescaling and remodeling of existing components [13] W.R. Dunbar, N. Lior, R. Gaggioli, Energy: Int. J. 16 (10) (1991)
to fulfill the needs of the system. Thus, if not careful, one risks 1259–1274.
making unrealistic selections and cost analyses of equipment. [14] S. Campanari, J. Eng. Gas Turbines Power 122 (2) (2000) 239–246.
Therefore, the design of a hybrid SOFC–GT–ST power plant [15] P. Costamagna, L. Magistri, A. Massardo, J. Power Sources 96 (2) (2001)
352–368.
must focus on all the components and not only on the SOFC. [16] S. Campanari, J. Power Sources 112 (1) (2002) 273–289.
Special attention must be given in the coupling of the turbo- [17] S. Pelster, Environomic Modeling & Optimization of Advanced Combined
machinery with the SOFC and the heat exchangers in order to Cycle Cogeneration Power Plants Including CO2 Separation Options, Doc-
326 A. Arsalis / Journal of Power Sources 181 (2008) 313–326

toral Thesis, advisors: M.R. von Spakovsky, D. Favrat, Ecole Polytechnique [26] R.F. Boehm, Design Analysis of Thermal Systems, John Wiley and Sons,
Federale de Lausanne, Lausanne, Switzerland, 1998. Inc., New York, 1987.
[18] R. Kehlhofer, Combined-Cycle Gas & Steam Turbine Power Plants, Pen- [27] Chemical Engineering Plant Cost Index, Chemical Engineering, McGraw-
nWell Pub. Co., Tulsa, Oklahoma, 1999. Hill, New York, 2006.
[19] S. Kakaç, H. Liu, Heat Exchangers: Selection, Rating, and Thermal Design, [28] A. Traverso, Personal Communication on the Purchase Cost Function of
second ed., CRC Press, Boca Raton, Florida, 2002. Condensing, Axial Type Steam Turbines, Virginia Tech, Blacksburg, Vir-
[20] A. Arsalis, Thermoeconomic Modeling and Parametric Study of Hybrid ginia, 2006.
Solid Oxide Fuel Cell-Gas Turbine-Steam Turbine Power Plants Ranging [29] C.A. Frangopoulos, Int. J. Energy Environ. Econ. 1 (4) (1991) 275–
from 1.5 MWe to 10 MWe, M.S. Thesis, advisor: M.R. von Spakovsky, 287.
Virginia Polytechnic Institute and State University, Blacksburg, VA, 2007. [30] M.S. Peters, K.D. Timmerhaus, R.E. West, Plant Design and Economics
[21] J.K. Salisbury, Steam Turbines and Their Cycles, R.E. Krieger Publication for Chemical Engineers, McGraw-Hill, New York, 2003.
Company, Huntington, New York, 1974. [31] Energy Information Administration, U.S. Natural Gas Prices, http://www.
[22] B.G.A. Skrotzki, W.A. Vopat, Power Station Engineering and Economy, eia.doe.gov, 2006.
McGraw-Hill, New York, 1960. [32] S.C. Singhal, Fifth International Symposium on SOFC, Aachen, Germany,
[23] P.J. Potter, Power Plant Theory and Design, Ronald Press Co., New York, July 2–5, 1997.
1959. [33] D. Rancruel, Optimization of Solid-Oxide Fuel Cell Balance-of-Plant Sys-
[24] A. Traverso, A. Massardo, W. Cazzola, G. Lagorio, ASME Paper 2004- tem, PhD Dissertation, advisor: M.R. von Spakovsky, Virginia Polytechnic
GT-54115, 14–17 June, 2004, Proceedings of ASME-IGTI TURBO EXPO Institute & State University, Blacksburg, VA, 2005.
2004, Vienna, Austria, 2004. [34] T.P. Chen, J.D. Wright, K. Krist, NETL Proceedings of the Fuel Cells ’97
[25] P. Chiesa, S. Consonni, Second Annual Conference on Carbon Sequestra- Review Meeting, GRI study, 1997.
tion, Washington, DC, USA, 2003.