SYSTEMS DEVELOPMENT FOR PLANARSOFC BASED POWER PLANT
ETSU F/01/00195/REP
DTI/Pub URN 02/868
ContractorALSTOM Research and Technology Centre
Prepared byS H Pyke
A J BurnettR T Leah
The work described in this report was carriedout under contract as part of the DTISustainable Energy Programmes. The viewsand judgements expressed in this report arethose of the contractor and do not necessarilyreflect those of the DTI.
Solid oxide fuel cells (SOFC) are now the subject of developmentprogrammes world-wide. The main focus of these programmes is in stationarypower applications, although considerable investment is also being made formobile auxiliary power units (APUs). The reasons for this interest are basedon several key advantages that SOFC systems potentially offer:− Very high system electrical efficiency (>50%)− Low emissions of harmful pollutants (NOx, SOx, unburned hydrocarbons,
etc)− Fuel flexibility (can be operated on a range of hydrocarbon fuels without
significant loss of efficiency or the need for complex processing plant)
In order to realise this potential SOFC systems must demonstrate a majorreduction in costs compared to systems demonstrated to date. This willrequire the design of a compact and highly integrated balance-of-plant systemand the development of low-cost SOFC stack technology. Reduced operatingtemperature (<800°C) SOFC stacks based on anode-supported cells have thepotential to achieve major cost reduction both in the stack and system.
As part of a programme to develop SOFC system technology for distributedpower applications, ALSTOM Research & Technology Centre has co-ordinated a European Commission Framework 5 Programme project (acronym“ProCon”). Project partners were Forschungszentrum Jülich (Germany) andPrototech (Norway), whose work is also reported in outline here.
The aims of the ProCon project were to investigate the design of a SOFCsystem based on intermediate temperature SOFC stack technology and todesign and test the performance of a SOFC stack based on anode-supportedcells. System design was focused on a 20 kW system since this size wouldhave the potential to be built as a demonstrator. However, the aim was todesign a system that could demonstrate the feasibility for commercially viableSOFC power plant, that could have a power rating in the range 100-200 kWe.The principal design targets for the system were that it should be compact(target maximum: 2 m x 2 m x 2 m) and have high electrical efficiency(>45%).
To meet these targets requires a reduction in the size of key components, suchas heat exchangers, afterburner and compressors, and a high level ofintegration and performance optimisation. It also implies a significant level ofinternal reforming of methane in the stack. Extensive process engineeringsimulations were performed to derive a flow sheet for a system that couldpotentially meet these targets.
The system designed included recycle loops for both the anode and cathodeexhaust gases. Recycling of the anode exhaust gas provides both steam andheat for the pre-reformer feed gas, which eliminates the need for a largeexternal steam generator and reduces the size of the fuel pre-heater. Recyclingof air from the stack lowers the necessary air-flow, which reduces the amountof heat ejected in the system exhaust and reduces the amount of heat that must
ii
be transferred in the pre-heater. The system performance for the unit withanode and cathode recycle is summarised in table 1i.
Result Value UnitsDC electrical power 20.06 kWGross AC electrical power 18.66 kWNet AC electrical power 16.83 kWUseful heat 12.80 kWHeat in fuel (LHV) 34.42 kWNet electrical efficiency 48.9 %Thermal efficiency 37.2 %Overall efficiency 86.1 %Power for air compressor -0.831 kWPower to run control system -1.0 kWPower loss in power conversion -1.40 kWTotal heat loss through insulation -2.23 kWCell voltage 0.715 VStack voltage 186 VCurrent density 299 mAcm-2
Air mass flow 21.0 gs-1
Fuel mass flow 0.772 gs-1
Specific fuel consumption 165 gkWhe-1
Specific CO2 emissions 418 gkWhe-1
Table 1i: Summary of performance of system with cathode recycle at fullpower
A dynamic model of the system was developed which enabled theinvestigation of system operability and transient performance, such as load-following capability. The model included simulation of the control systemrequired to operate the SOFC unit under varying load conditions, as well asstart-up and shut-down. The model indicated that the system was controllableover the range 50-100% of rated power and that the load following capabilitywas better than a priori expectations.
As well as the ‘gas balance-of-plant’, a design for the power conditioning wasalso developed that would meet the requirements for connection of the systemto the public electricity supply and that would interface with the particularcharacteristics of the SOFC stack. Simulations indicated that the resulting firstgeneration design of power conditioning has excellent performance and canregulate the load on the SOFC stack independently of the mains supply.
The SOFC design work by Forschungszentrum Jülich focused on the design ofa 20 kW SOFC module that would integrate into the system and the detailedengineering of a 5 kW stack based on anode-supported cells. The stackdevelopment, including materials and cell technology was performed outsidethe ProCon project. The stack design, which employs 20 cm square anode-supported cells, features an internal manifold and parallel flow arrangement.Stack tests with this design have achieved power densities over 0.5 Wcm-2 at800°C. To date, materials degradation problems have delayed the 2,000 hour
iii
test of a 5 kW stack of design. This test is now scheduled to be completed inlate 2002.
The design and performance modelling results clearly demonstrate thepotential to achieve a compact, low-cost and highly efficient SOFC systembased on a stack operating at temperatures below 800°C. The next phaseshould now be to prepare for a complete system demonstration. Work shouldaddress the testing of all components and sub-systems and of the controlsystem. Further evaluation of the stack performance is required to informsystem design, including, operation with internal reforming and transientperformance.
Although the development of the SOFC stack was outside the scope of thisproject, it is clear that in order to exploit the potential of planar, intermediatetemperature SOFC technology, development of the stack should focus ondevelopment of a cost-effective design and increasing the durability.
1
TABLE OF CONTENTS1. BACKGROUND ..................................................................................................22. OBJECTIVES ......................................................................................................43. FUNCTIONAL SPECIFICATION OF SOFC SYSTEM ................................54. DESIGN AND MODELLING OF A 20 KW SOFC SYSTEM........................6
4.1 Process flow engineering ................................................................................64.1.1 Anode gas recycling ..............................................................................64.1.2 Cathode gas recycling...........................................................................7
4.2 Conceptual design of the 20 kW integrated SOFC system .........................94.3 Dynamic modelling of the 20 kW integrated SOFC system......................10
5. SOFC SYSTEM POWER ELECTRONICS ...................................................125.1 SOFC Electrical Characteristics .................................................................125.2 The power conditioning system ...................................................................145.3 Optimisation of the power conditioning .....................................................15
6. DESIGN OF THE SOFC STACK....................................................................176.1 Modelling of the 20 kW module...................................................................176.2 Design engineering of the 5 kW SOFC stack .............................................18
Of the various fuel cell types currently under development, solid oxide fuel cells(SOFCs) appear best suited to the demands of the stationary electrical powergeneration market. SOFCs will offer the highest electrical efficiency when operatingon hydrocarbon fuels (in excess of 45%), so they can be expected to have the mostcompetitive running costs. They will also offer fuel flexibility and security of supplysince they can potentially be operated on a range of fuels, including pipeline naturalgas and bio-mass, without a significant loss of efficiency or increase in systemcomplexity and cost. With these attributes, and being essentially modular, SOFCsystems will be highly attractive for the distributed power market where units can beconfigured and sized to meet a particular local power generation demand. Highsystem efficiencies for SOFC should also make the biggest impact in reducingemissions of carbon dioxide which, combined with very low emissions of major localair pollutants (CO, NOx and unburned hydrocarbons), will make them an extremelyattractive generation technology as tighter emissions legislation is implemented.
In order for SOFC systems to realise their potential in the stationary power generationmarket, they will have to achieve substantial cost reductions compared to systemsdemonstrated to date. The major cost saving for SOFC will be achieved by exploitingthe potential for making systems simpler and more compact than is possible for otherfuel cell types. Low-temperature fuel cells (PEMFC, PAFC, etc) use noble-metalcatalyst electrodes which must be fed with a high purity hydrogen fuel. This requiresa complex and expensive balance-of-plant (BOP) in order to provide the necessarypre-processing for hydrocarbon fuels. The energy consumed in performing thisprocessing also limits the efficiency of these systems. By contrast, the high operatingtemperature of SOFCs gives them the possibility of steam reforming methane tohydrogen and carbon monoxide within the fuel cell stack by means of a CO-tolerantnickel catalyst. This means that SOFC can potentially operate on natural gas withminimal pre-processing of the fuel.
In addition to the simplified fuel processing, optimisation of SOFC system design canresult in much higher levels of component integration, a reduction in componentdimensions and minimisation of parasitic losses (heat losses, compressorwork/pressure drops, etc) in the system. This will not only reduce the capital cost ofSOFC systems but will also result in systems of high electrical efficiency.
The cost of the SOFC stack is clearly the other major factor in the overall system cost.Work on reducing SOFC costs is focussed on planar stack and cell designs whichpromise the lowest stack production costs since high-volume, low-cost manufacturingprocesses, such as tape-casting, can be employed. Efforts to reduce themanufacturing costs and improve durability of planar SOFC have resulted inevolutionary modifications to the designs. The use of a metallic, rather than aceramic, interconnect promises significantly lower manufacturing costs. However,the problems of high-temperature corrosion and the need to use readily available andeasily formed metals, such as stainless steels, require a reduction in the stackoperating temperature.
Currently, the most feasible way of enabling efficient, intermediate temperatureoperation is to base the stack on anode-supported cells. Conventionally, planar SOFC
3
cells use the electrolyte as the structural support for the electrodes, which limitsminimum electrolyte thickness to around 100 µm. By using a thick, structural anode,much thinner electrolytes can be used with a concomitant reduction in cell resistance.Work at ALSTOM, in conjunction with the development programmes atForschungszentrum Jülich and ECN, indicates that this will allow stacks based onanode-supported cells to operate at temperatures in the range 700-800°C compared tothe 800-900°C necessary for electrolyte-supported, planar cells.
The advantages of lower temperature stack operation should also extend to the costsof the surrounding balance-of-plant components, such as heat exchangers, piping andafterburners, since they too will not have to be made from high-cost, heat-resistantalloys.
In order to investigate the potential for building an efficient, low-cost system based onSOFC stacks operating in the temperature range 700-800°C, a study involving thedetailed design and performance modelling of an SOFC unit was performed. Themajor cost elements of the BOP system are heat exchangers, reformer, afterburner andcompressors, therefore the focus of the BOP design was to reduce the size of thesecomponents.
In parallel, the development of a planar stack with performance appropriate to thesystem operation was conducted. The design and analysis was based on a nominal20 kW SOFC system since this sizing would have the potential to be built as apilot-scale demonstration. The design, however, was also intended to be scaleable toa larger system of, say, 100-200 kW, which would have wide commercialapplicability. The work described in this report was performed and funded within aEuropean Commission Framework 5 Project (acronym: “ProCon”). Project partnerswere Forschungszentrum Jülich (Germany) and Prototech (Norway).
4
2. OBJECTIVES
The aim of this project was to investigate the design and model the performance of asystem based on a SOFC stack module using anode-supported cells and, therefore,with an operating temperature in the range 700-800°C. In parallel, the project aimedto demonstrate the feasibility and performance of a planar stack based onanode-supported cells under projected system operating conditions. Specific projectobjectives were:
• To assemble a 5 kW-scale planar SOFC stack• To design and build a suitably instrumented test rig for operation on reformed
methane• To test the performance of the 5 kW-scale stack during a 2000 hour trial• To design and model a conceptual 20 kW integrated system• To perform the detailed engineering of system components
5
3. FUNCTIONAL SPECIFICATION OF SOFC SYSTEM
At the outset of the project, a functional specification for a 20 kW SOFC system wasderived. This provided for an SOFC system that would have performance andattributes that would be attractive for a range of distributed power applications, ratherthan a detailed specification for a particular application. As well as thesespecifications the system had to be scaleable, based on a notional future systemsrequirement in the range 100-200 kW. Whilst no cost targets were set, the systemdesigned clearly also had to address the requirement for low production costs.
Key parameters for the functional specification of the 20 kW system were:
− electrical load driven, with heat a (useful) by-product− maximise electrical efficiency (target: 45% for 20 kW system)− overall efficiency 70%− turndown ratio: minimum operational load 50% (possibly lower for larger
systems)− start up time: 12 hours from cold, 3 hours from hot standby− shut down: safe and non-catastrophic in the event of loss of gas supply or grid− comply with requirements, for grid connection:
− 50 Hz− 400 V (or 415 V)− EN-50160 Power Quality Standard
− siting indoors− compact (target maximum: 2 x 2 x 2 m3)− operation on three specified gas compositions with varying sulphur and higher
hydrocarbon content
6
4. DESIGN AND MODELLING OF A 20 KW SOFC SYSTEM
4.1 Process flow engineeringThe aim of this task was to derive and model plant concepts that would meet thefunctional specifications. To meet these targets requires a high level of integrationand performance optimisation and implies a significant level of internal reforming ofmethane in the stack. Reduction in the size of key components such as heatexchangers is also a key requirement.
Process engineering for concept plant focused on the balance-of-plant layout but alsoconsidered the projected requirements for control and power conversion and stackmodule design and performance. In order that each of the partners could use theoutput data from process flow modelling, it was important at the outset that data couldbe generated and compared reliably and reproducibly. Since the project partners useddifferent software for process engineering, a validation case was defined, which wassimulated by all partners to check their compatibility.
The flow sheet for the validation case is shown in figure 1. In the system defined forthe validation case, inlet natural gas is mixed with a separate steam supply and passesthrough a partial reformer before being fed to the SOFC stack. On the air-side, inletair is preheated prior to feeding to the stack. Exhausts from the fuel and air-sides ofthe stack are then fed to an afterburner. The burner exhaust in this system is split toheat the reformer and air in parallel. Further downstream, residual exhaust heat isused to generate steam for the reformer and any remaining useful heat is used forspace heating.
The output data files for the three partners were compared and after clarification ofcertain definitions, such as fuel utilisation, air ratio, etc., only small differences (≤1%)remained. These were agreed to be tolerable for the ongoing simulation work. Asexpected, the net electrical efficiency for the validation case of 35 % (gross 47 %) wasrelatively low. In order to derive a system that was potentially compact and couldmeet the target electrical efficiency, operating parameters were varied and modifiedflow sheet designs were investigated. In particular, it was decided to:
− increase the fuel utilisation (max. 85 %)− increase the degree of internal reforming (with a minimum of 10% H2 at SOFC
inlet)
Flow sheet design modifications included:
− anode gas recycling with an ejector (Oxygen/Carbon ratio = 1.8 minimum)− cathode gas recycling with an ejector
4.1.1 Anode gas recycling
Where an anode gas recycle is adopted, much of the exhaust gas from the anode sideof the stack is recycled and mixed with fresh fuel before the mixture is fed to the pre-reformer. The recycling, generally, would be performed using an ejector (jet pump),
7
driven by the fresh fuel. It was assumed in this system that the natural gas wassupplied at 3 bar, and therefore no compressor work was required to drive the ejector,meaning the energy to drive it was effectively free.
The principal reason for anode gas recycling is to eliminate the need for a largeexternal steam generator, although a small steam supply is likely still to be requiredfor system start-up. This is beneficial for the system thermal efficiency, since the heatrequired to supply the steam would otherwise be extracted from the exhaust gases,which would require an expensive boiler and a supply of deionised water. In addition,the heat in the recycled exhaust gas provides much of the preheating required for thefresh fuel, reducing the size of, or even eliminating the need for, a fuel pre-heater.
A further potential benefit is that there may be sufficient heat in the recycled anodeexhaust to enable a low, but sufficient, degree of external reforming (includingcomplete conversion of higher hydrocarbons) to be achieved in a simple adiabatic pre-reformer. This would eliminate the need for an externally heated pre-reformer, whichis a more complex and expensive component. However, there were doubtsconcerning the effectiveness of an adiabatic pre-reformer under part-load operation,so in the final design a heated pre-reformer was adopted. However, to optimise thesystem efficiency the degree of pre-reforming was still relatively low, with most ofthe methane reformed internally within the stack.
The addition of anode gas recycling to the system was shown to increase significantlyboth the electrical and thermal efficiency of the system, with values of net electricalefficiency as high as 48% at full load. A generic flowsheet of a system with anodegas recycling is illustrated in figure 2.
4.1.2 Cathode gas recycling
The flow of air carries oxygen to the cathode and also removes excess heat from thestack. The flow required for thermal balance is usually larger than the minimumnecessary to supply oxygen in the stack. A large net air-flow should be avoided sinceit carries a greater proportion of heat out of the system at the final exhausttemperature of 80 °C. Additionally, with increasing air flow, a larger fraction of thepower generated in the stack is consumed by the air compressor and a larger heatexchange area is required to pre-heat the incoming air.
Recycling a proportion of the hot cathode exhaust reduces both the fresh air flow andthe amount of heat that must be transferred in the pre-heater. This allows asignificantly smaller air pre-heater to be specified with a corresponding cost saving.The penalty is that, whilst fresh air flow is reduced, the ejector is pressure-driven soadditional power is required to drive the compressor.
Simulation results showed that with the lower net air flow and reduced heat exchangein a system with air recycle, it is feasible to operate without an air pre-heater at thedesign load. However, investigation of system operation on part-load showed thatthis was not practicable since there is no margin of excess air that can be used toreduce the flow.
8
The main benefit of designing the system with cathode gas recycle was an increase inthermal efficiency and system compactness. The effect on electrical efficiency wasless significant than for anode gas recycle. Additionally, there was some concernover the controllability of a system with cathode recycle under part-load and start upconditions because of the non-linear variation in performance of the air-side ejectorwith changing pressure.
The predicted performances of the systems both with and without cathode gas recycle(both systems have an anode gas recycle loop) are summarised in tables 1 and 2.
Result Value UnitsDC electrical power 20.29 kWGross AC electrical power 18.87 kWNet AC electrical power 17.05 kWUseful heat 9.65 kWHeat in fuel (LHV) 34.42 kWNet electrical efficiency 49.5 %Thermal efficiency 28.0 %Overall efficiency 77.5 %Power for air compressor -0.818 kWPower to run control system -1.0 kWPower loss in power conversion -1.42 kWTotal heat loss through insulation -2.23 kWCell voltage 0.723 VStack voltage 188 VCurrent density 299 mAcm-2
Air mass flow 54.10 gs-1
Fuel mass flow 0.772 gs-1
Specific fuel consumption 163 gkWhe-1
Specific CO2 emissions 413 gkWhe-1
Table 1: Summary of performance of system without cathode recycle at fullpower
9
Result Value UnitsDC electrical power 20.06 kWGross AC electrical power 18.66 kWNet AC electrical power 16.83 kWUseful heat 12.80 kWHeat in fuel (LHV) 34.42 kWNet electrical efficiency 48.9 %Thermal efficiency 37.2 %Overall efficiency 86.1 %Power for air compressor -0.831 kWPower to run control system -1.0 kWPower loss in power conversion -1.40 kWTotal heat loss through insulation -2.23 kWCell voltage 0.715 VStack voltage 186 VCurrent density 299 mAcm-2
Air mass flow 21.0 gs-1
Fuel mass flow 0.772 gs-1
Specific fuel consumption 165 gkWhe-1
Specific CO2 emissions 418 gkWhe-1
Table 2: Summary of performance of system with cathode recycle at full power
4.2 Conceptual design of the 20 kW integrated SOFC system
The conceptual design of the balance-of-plant (BOP) system was performed byPrototech. The concept system was based on the process flow modelling discussed inthe preceding section and was for a system design that included recycling of both theanode and cathode off-gases. The BOP layout for this system is shown schematicallyin figure 3. The design of the system BOP focused on the major cost and sizeelements which are the heat exchangers, reformer, afterburner and compressor(s),since a reduction in size of these components will contribute not only to major costreductions but also to a more compact packaged system.
The complete system is illustrated schematically in figure 4, where it is separated intothree main sections: the stack module, the BOP module and the control and powerconditioning module. A photo-rendered impression of the packaged unit is shown infigure 5. The system is designed for siting indoors, with a key objective thereforebeing to make it compact. The unit designed has a total footprint of 1.6 m x 0.8 m =1.3 m2 which is within the initial specification, of 2 m x 2 m.
Heat exchangers and the pre-reformer were designed for durable and stable operation,and are therefore conservatively oversized in the concept shown with respect to thenumber of plates and plate spacing. The BOP module and the stack module areinsulated to minimise heat losses by means of a 100 mm layer of high performance,microporous insulation; total heat loss with this configuration was calculated to be835 W at the nominal full electrical power output of 20 kW.
10
The criterion for the sizing of the pipework was that the gas velocity was generallykept below 20 m/s to avoid unnecessary pressure drops. The heat exchangers weredimensioned using the CFD software package Star-CD. The analysis includedtemperature dependent density, viscosity and conductivity and was performed with a3D mesh representing the real 3D geometry. The heat exchangers are of plate typewith dimensions 500 mm x 200 mm and a thickness of 1 mm.
For the case with no cathode gas recycle, the heat exchange is 380 W per plate,requiring a total of 120 plates with a total area of 12 m2 (9.6 m2 net). For the systemwith cathode gas recycle, because of the much higher temperature difference betweenthe streams and the significantly lower flow, the heat exchange is 1.3 kW per plate.This requires a total of 3 plates with a total area of 0.3 m2. This large reduction insize, and therefore cost, of the air pre-heater is the key advantage of employing acathode gas recycle loop. A photo-rendered impression of the smaller heat exchangeris shown in figure 6 and the calculated temperature distribution is shown in figure 7.
4.3 Dynamic modelling of the 20 kW integrated SOFC system
A dynamic model of the 20 kW system was developed and tested by ALSTOM, usinga commercially available dynamic simulation tool, gPROMs. Two system concepts,based upon the steady-state simulations by Prototech, were modelled; one system thatincluded cathode gas recycle and one without. The model includes simulation of thecontrol scheme required to operate the system under varying load conditions, as wellas start-up and shut-down.
To this end, the model developed is more detailed than the steady-state models, andgives a greater insight into the operability of the system, particularly at part-load.This is an important consideration, since it is quite possible to design a system whichworks very efficiently on paper at maximum power, but which would be inoperable inpractice since it would be impossible to start.
Control loops, which relate the fuel supply to the applied stack current, regulate theanode gas recycle loop and control the stack temperature, were simulated. These aregenerally PI controllers, with a simulated supervisory PLC controller which modifiestheir set-points according to the required power output. The stack temperature isregulated by changes, either to the temperature or flow, of the air supply. The use ofthe air supply to control stack temperature may cause difficulties in the control of thesystem with cathode recycle at part-load, due to certain non-linearities in the recycleejector performance.
The following control objectives were specified:− maintain constant stack temperature from 50 to 100% rated power− fuel utilisation to remain constant at all power outputs− oxygen / carbon (in the fuel) ratio at the pre-reformer inlet always to exceed 1.8 in
order to avoid the potential for carbon deposition− water temperature to remain constant at 80°C− exhaust temperature not to fall below 50°C− oxygen (in air) / fuel ratio, lambda, at the stack inlet always to exceed 2.0
11
With both systems it was possible to meet these objectives, although the method forstack temperature control differs. Using the model, it has been demonstrated that itshould be possible to regulate the stack temperature to within a few degrees over therange 30-100% power by careful regulation of the air supply. This is largely as aresult of the high thermal mass of the stack, which means that any temperaturechanges occur slowly. This indicates that the dynamic performance of the SOFCgenerator system can be predicted to be relatively fast. In particular, the dynamicperformance of a planar SOFC system is predicted to be significantly better than atubular type system since, in this planar design, the stack is much more compact, andtherefore has a smaller inventory of reactants. Also the thermal conductivity of themetallic, planar stack components is much higher than the ceramic components of atubular stack, which should help to prevent large temperature gradients occurringwithin the stack.
Simulations have also been performed on a system cold start, and to this endadditional plant items (a steam generator, an external heat supply to the stack andvarious bypass valves) have been provided.
Further simulations have also been performed to investigate the control andcomponent requirements for putting the system into ‘hot standby’. This is important,for example, during a temporary loss of the mains supply, since it would avoid theneed for a complete shut-down of the system and would allow it to re-start exportingpower with only a short interruption.
It is intended to submit a patent to cover much of this work, so the details providedhere are limited. However, typical results are illustrated in figures 8 to 12. Figures 8and 9 show the predicted electrical and thermal power outputs of the systems with andwithout cathode gas recycle during a simulation where the load was reduced in stepsfrom 100% to 50% power. It can be seen that the electrical power output is predictedto closely follow changes in current. This is largely because the control enables thestack to be maintained at a constant temperature, as discussed earlier. The thermalpower output is shown to take longer to react to load changes. It can also be seen thatthe thermal power output of the system with cathode recycle is somewhat higher thanthat without recycle, since less heat is rejected in the exhaust.
Figure 10 shows the changes in electrical and thermal efficiencies with reduced poweroutput. Down to 50% of rated power output, electrical efficiency is seen to rise as aresult of the lower losses when operating the stack at lower current density and,therefore, higher voltage (figure 11). Figure 12 shows the results of a start-upsimulation.
12
5. SOFC SYSTEM POWER ELECTRONICS
The power conditioning for an SOFC power generation scheme sets significanttechnical challenges in order for it to interface with the unique characteristics of theSOFC stack and to achieve the required performance within a compact and low-costsolution. The power conditioning also represents the for the SOFC systemowner/operator’s interface to the grid, therefore an understanding of the electricitygenerating and distribution system can usefully serve to identify and guide thecomplete equipment requirements towards a design that meets the end applications.
As part of the detailed investigation of the 20 kW SOFC system, a paper design for apower conditioning system has been created and modelled to identify the major issuesand their proposed solutions. The modelling work includes detailed representations ofthe power electronics elements, the transformer and the public electricity supply(PES) to which the system is connected. The SOFC has been modelled as anapproximation of the predicted V-I characteristics as there is insufficient actualdynamic electrical performance data to model it in detail at the present time. Acomplete set of studies has been performed on the model to demonstrate theperformance of the system supplying into a PES with local load. This is consideredthe most demanding case in terms of the design of the power conditioning andperformance has been found to be in accordance with the required specifications.
The key findings of the work are presented here. The current understanding of theSOFC characteristics is discussed briefly since this imposed certain constraints on theapproach taken for designing the power conditioning. The power conditioning isdescribed and a discussion presented on the available system options and why thechosen configuration was adopted.
5.1 SOFC Electrical CharacteristicsThe ionic transport occurring within a SOFC has led to an analogy being drawn withthe operation of a battery and, whilst this is a helpful concept, the SOFC differssignificantly in its own specific characteristics. The dynamic performance of theSOFC is limited (not withstanding the discussion in section 4) and requires that thecurrent drawn must carefully regulated and must not rise or fall too rapidly.
The active cell area determines the current capacity of a stack with a typicaloperational current density being of the order of 0.5 Acm-2. Each cell layer of a stackwill produce an open circuit voltage of 1.05V that falls in a broadly linear mannerwith increasing current down to a set to a minimum that is typically 0.6V. Operationbelow this minimum voltage should be avoided (certainly for extended periods) sincethe mass transport limitations of the electrochemical reaction would tend to cause theanode of the cell to be electrochemically re-oxidised which will degrade cellperformance and ultimately shorten stack-life. The planar stack geometry and the cellarea under investigation in this project resulted in a 20 kW stack module thatproduced similar orders of voltage and current output at the maximum power point.This is considered as a relatively low voltage and high current source for conventionalpower electronics.
The SOFC balance-of-plant (BOP) has a control bandwidth (of the order of seconds)that is much lower than the fluctuations in power demanded from the connection to
13
the PES (100s µs) and it is important that the BOP control is maintained in such a wayas to protect the SOFC module within it. Therefore the electrical power drawn fromthe stack must be controlled so that it is kept within the transient capabilities of theBOP. Constraining increased PES demands in power is a matter of control but extrapower conditioning plant is required to absorb the excess SOFC output power forperiods when the PES is unreceptive so that the load change can be smoothed out tothe BOP capabilities.
Operation of a SOFC stack in the laboratory is usually performed with a load thatdraws a steady d.c. current. However, typical power conditioning equipment containsswitching elements that will draw the current in pulses. The variation imposed on thed.c. current by the pulsing, termed ripple current, is expressed as a percentage of thefull load d.c. current. Therefore for a SOFC stack connected to power conditioningequipment, the direct effect of the ripple current would be that it becomessuperimposed on the SOFC stack voltage and causes the cell voltage to transientlypeak and dip above and below the level normally required for the given output power.It is known that a large ripple current will reduce the maximum power outputavailable from the SOFC stack but apart from this little is known about the dynamicelectrical performance, particularly with regard to long-term effects. For the purposesof this study the power conditioning was designed to minimise the ripple current thatwas drawn from the SOFC stack. A further consideration was that back flow ofpower into the SOFC stack is also potentially damaging therefore the powerconditioning had to be designed to prevent this occurring.
The specification for a 20 kW SOFC stack that formed the basis of the designrequirements for the power conditioning is shown in table 3. It can be seen that theSOFC as a power source imposes specific requirements on the design of the powerconditioning. The ripple current requirement is particularly onerous and willcurrently preclude the use of most of the compact, high frequency power converters.
1 Fuel cell dimensions 20x20cm2 Number of cells in the stack 2603 Output power 20kW4 No load voltage 220V5 Operating voltage 185V6 Operating current 108A7 Maximum permissible ripple current 3.9A8 Maximum rate for a large change in current 1A/s9 Maximum step of current 5A10 Nominal system parasitic losses 1.28kW11 Additional losses for hot standby 2.4kW12 Minimum power for stack self heating 3.68kW
Table 1: SOFC specification for power conditioning
14
5.2 The power conditioning systemFigure 13 is a block diagram showing the basic SOFC power conditioning system thatwill connect the SOFC to the three-phase PES. The form is not dissimilar to that usedfor a photovoltaic power generating plant where the key component is the d.c. tothree-phase 50 Hz static invertor. There are several ways in which the powerconditioning system can be configured. The reasons for the system topology chosenare summarised as follows:− The voltage output from the 20 kW power level SOFC module requires a voltage
step up to reach the PES voltage of 415 V a.c. rms− Connecting the invertor to the SOFC stack directly will not comply with the
SOFC stack requirements of ripple current or power back feed− The ‘power system matching block’ was therefore added between the two and
selected to optimise the loading requirements for the SOFC stack− This restricts the operation of this converter such that it cannot currently provide
isolation or the full voltage step up− The best compromise is therefore to step up to an intermediate voltage suitable for
the optimum utilisation of the invertor semiconductors− A 50 Hz transformer was added downstream of the invertor to provide the balance
of the required step up to the PES
The system described above was designed as a first-generation system that is mostcertain of providing the correct performance. It is not the only configuration possibleand lacks some of the optimisation that could be achieved with other more integratedsolutions. In this design, the positioning of the transformer enables it to: provide therequired galvanic isolation for safety; help with filtering out the electrical harmonicnoise from the invertor and block any d.c. current passing from the invertor into thePES, which are all important legal requirements. However, the major disadvantage isthat 50Hz transformers are large and heavy. Indeed, if the power semiconductors arewater-cooled then the transformer becomes as big as the rest of the powerconditioning.
Since the SOFC output power is not capable of varying rapidly, then the maindisturbances to the system will come from the PES. The supply of power to the PEScan be made to operate at an appropriate bandwidth, so to account for small variationsis a matter of control. The problem occurs where a significant voltage dip occurs inthe PES or the supply becomes disconnected from the PES and it will, therefore,become unreceptive to power. This requires a load to absorb the excess power fromthe SOFC during the interruption or until the SOFC output power can be rampeddown. Energy storage can be used but batteries are not appropriate because theycannot absorb the power at a sufficient rate. New ultra capacitor technology canprovide sufficiently high capacitance at an appropriate volume but is currently tooexpensive. Therefore, the only viable alternative is a resistive load-bank producingwaste hot water. This is has been incorporated into the system model and found tooperate satisfactorily.
The control was configured to set the invertor to pass the power from the SOFC to themains. The power system matching block has to regulate the d.c. link and is thereforetransparent in terms of the power transfer process, i.e. the power out is the same aspower in (less losses) but at a fixed voltage. This transfer must be carried out in such
15
a way that the SOFC voltage does not dip excessively and without causing any rapidchanges of current. A control scheme was derived to use the current drawn by theinvertor in combination with the feedback from the load bank to regulate the SOFCterminal voltage against a desired set point.
Figure 14 shows an example case where regulation is taking place. In this case, thesystem was started up and then a 90% voltage dip of 10% of the mains cycle wascreated on the PES after 50 ms, in order to disturb the system. It can be seen that thiscreated slight repetitive “notches” in the SOFC current and voltage waveforms butwas otherwise insignificant because the control instantly diverts the SOFC power thatcannot be absorbed into the load bank. The start-up also shows how well the SOFC iscontrolled because the slopes of the waveforms are smooth gradients that are withinthe BOP limits. It should be noted that the initial uncontrolled transient is a functionof the modelled system and will not be present in practice.
Work on the dynamic model of the BOP, described in section 4, included results of asimulation to put the system into hot standby to avoid a complete system shutdown inthe event of temporary interruption of the load. The requirements for controlled rampdown of the power output of the SOFC to a hot standby in the absence of the mainspower supply have been investigated in the design and modelling of the powerconditioning. The results indicate that, regardless of the PES state, the powerconditioning can be successfully integrated with the BOP to supply the parasiticsystem loads (compressors, pumps, control electronics, etc) and maintain theappropriate operating conditions for the SOFC stack and BOP.
5.3 Optimisation of the power conditioningThe design basis, as set out earlier, for the power electronics was set by certainassumptions about the performance and protection requirements for the SOFC stack.With better understanding, which can only be obtained in the light of practicalverification, it may be possible to relax some of the requirements for the powerconditioning to produce a lower-cost system. The justification for, and design of, thepower system matching block hinges on the electrical impedance of the SOFC stackso a key priority is to confirm this by performing a practical evaluation of the ripplecurrent and transient performance of the stack.
The requirements for a more compact power conversion system have also beenconsidered and are briefly discussed. A more compact position for the transformerwould be within the power system matching block because it can operate at highfrequency and therefore would typically offer a 50% reduction in size and an 80%reduction in weight but would require a more ambitious converter design. Sincesmaller stacks are unlikely to reach the voltage for direct driving to the PES thendevelopment of this improved converter would be particularly attractive for a compactSOFC system.
Much larger SOFC stacks have the potential to achieve the voltage level for drivingthe PES directly but may still require the isolation provided by a transformer andtherefore a high frequency option is still attractive. It should also be noted that thehigh frequency transformer technology is only available up to 40 kW per device sointerleaving of multiple stages will certainly be required for the larger power ratings.
16
Recommendations for further work− Evaluate the SOFC stack source impedance and incorporate the findings into the
model to reassess the proposed design− Carry out the detailed circuit design of the power electronics elements− Bread board test each of the sections to confirm correct operation− Detailed design of the controls including process synchronisation and PES
interfacing protocol− Bread board test the complete system to confirm the operation
17
6. DESIGN OF THE SOFC STACK
Work within ProCon by Forschungszentrum Jülich addressed both the design andperformance modelling of the SOFC module for integration into the 20 kW system(this may comprise more than one stack and includes the necessary gas connections tothe stack or stacks) and also the engineering of the 5 kW stack. Development of thecell and stack technology was performed in parallel programmes.
6.1 Modelling of the 20 kW moduleStack modelling was performed first for one single stack based on the SOFC stackconcept using anode-supported thin electrolyte layer cells, in order to show thefeasibility of this technology in attaining stacks operating below 800 °C withreasonable power densities. Figure 15 shows the resulting current density andtemperature distribution in one layer of this stack and a summary of the main resultsof the calculations. The stack was assumed to be operated on 30% pre-reformedmethane fuel, i.e. 70% of the methane is internally reformed, with air and fuel inlettemperatures of 700°C. For a fuel utilisation of 70%, the stack attained an averagepower density of 0.21 W/cm², where the maximum temperature did not exceed 800°C.
The 20 kW SOFC module was designed to comprise several smaller stacks. Optionsfor combining these stacks in parallel and/or in series were investigated. One of theseoptions was a so-called cascade of stacks. In the cascade, the fuel flows through aseries of stacks: the exhaust fuel from two stacks in one stage is combined and fed toone single stack in the next stage. Fresh air is supplied in parallel to all stacks in allstages. An advantage of this arrangement is that the fuel utilisation in the stackremains low, whereas the overall fuel utilisation still can reach values above 60%.However, the overall power output of the cascade arrangement was not substantiallyincreased in comparison with a single stack operating under the same conditions.Similar results were obtained for another option where both air and fuel flowedthrough a series of stacks. This latter arrangement also showed the disadvantage ofhigh temperature air being fed into stacks in the next stage, raising the temperaturelevels to undesirably high values. Moreover, the series connection of gas flows leadsto higher pressure differences, which would require more work by the compressors,leading to higher electrical power losses in the system.
For these reasons it was decided to make the arrangement for the 20 kW module withfour stacks of 5 kW each, which are all fed with air and fuel in parallel.
Additional modelling of the 5 kW stack was performed, in particular at partial loadlevels. The 5 kW stack was design to contain 65 cells of 20 cm x 20 cm each(effective electrode area 19 x 19 = 361 cm²). For these calculations the fuel gascomposition exiting the adiabatic pre-reformer (derived from the process flowmodelling) was taken as input for the stack at a temperature level of 650°C. Theresults of the calculations are shown in figure 16. In the first series the airstoichiometry was kept constant at a value of 3.7, in the second series it was graduallylowered with decreasing load, but kept at or above the minimum value of 2. Theresults indicated that the minimum stack temperature was always in the range of 720to 740°C, more or less independent of the load and the air stoichiometry. Themaximum stack operating temperature was in most cases close to 800 °C, except for
18
the cases with 50% partial load and 25% partial load, in which the cooling effect ofthe air flow was evident.
6.2 Design engineering of the 5 kW SOFC stackThe cell and stack technology for the 5 kW SOFC stack was developed in parallel byForschungszentrum Jülich in the ongoing 'ZeuS'-project, funded by the GermanMinistry for Economic Affairs (BMWi), in which ALSTOM also co-operated in theDTI-supported project F/01/00195.
The established Forschungszentrum Jülich stack design was a block comprisingsubstrate cells and interconnect plates that was placed inside a metallic housingconstituting the external manifolds. This design allows the stacks to be operated in across-flow configuration only. To counter the problems originating from the severedemands on the sealing materials, the design was modified. In the new design with aninternal manifold, the sealing surfaces are all in planes perpendicular to the stackingdirection. A number of short-stacks were tested in this so-called D-design with cross-flow configuration.
Adopting an internal manifold design opens the possibility to change the gas flow to aparallel flow (either counter-flow or a co-flow) configuration, which was given thedesignation E-design. Modelling calculations show that temperature distributionsacross the cells are more symmetrical and temperature gradients are lower in thislayout. For the E-design with internal manifold and counter-flow configuration,extensive modelling of the gas flow was performed to optimise the geometry and thedimensions of the feed tubes, the manifolds and the gas channels in the interconnectplate.
A schematic view of the stack design is shown in figure 17. In the counter-flowarrangement, air and fuel gases enter via the base plate of the stack through the twoouter tube sections. In each stack layer, gases enter an enclosed plenum and are thendistributed across the entire surface of the anode or cathode of the single cell beforeflowing to the exhaust plenum from where the gases exit the stack via the central pipesection. The anode and cathode facing sides of the interconnect plate are identical.
The feasibility of this stack design was first demonstrated with short stack tests using10 cm x 10 cm cells. The stack design for the ProCon project was based on a20 cm x 20 cm cell size. Forschungszentrum Jülich has performed a series of testswith this stack design. For example, a test has been performed with a 10-cell stack.At an operation temperature of 800°C this stack gave a power output of 1.6 kW(220 A @ 7.30 V; i.e. 0.61 A/cm² and 0.44 W/cm²) with humidified hydrogen at afuel utilisation of 44%. At 880°C the power output was 2.4 kW (340 A @ 7.14 V;i.e. 0.94 A/cm² and 0.68 W/cm2) at a similar fuel utilisation of 47%. Performance ofthis stack design has also been demonstrated on methane. The results demonstratethat the stack design can meet performance requirements for the system with internalreforming.
To date, however, in these tests, it has been difficult to achieve stack lifetimes betterthan 700 hours with the existing ferritic stainless steel interconnect material (DIN1.4742). This is principally believed to be a result of oxidative degradation at theinterconnect/cathode interface. Plasma-sprayed anti-corrosion coatings on this
19
stainless steel produced by the German national aerospace research institute, DLR,have not resulted in the expected decrease in degradation. As a result, the proposed2,000 hour test of a stack on reformed methane has been delayed. Parallelinvestigation of ferritic stainless steels is ongoing and a candidate material has beendeveloped which should provide the necessary increase in stack lifetime. It isintended that this material will now be used to make the interconnect plates that willbe used for the 2,000 hour test.
20
7. CONCLUSIONS
The complete conceptual design of a 20 kW system based on a SOFC module with amaximum operating temperature of 800°C has been developed and its performancemodelled. The design and the performance meet the functional specification set forthe system. Key achievements of this design are:− Compact (footprint for the packaged unit = 1.6 m x 0.8 m)− Potentially low-cost− High electrical efficiency (49% at full load, up to 55% at half rated power output).
Given the data and assumptions used, it is reasonable to expect that when scaled-up to a power range 100-200 kWe a system of this design could attain an electricalefficiency exceeding 55%.
The targets were reached by designing a highly integrated system with a high level ofinternal reforming. Recycling of anode and cathode exhaust gases was integrated intothe balance-of-plant and enabled significant reductions in heat exchange area andeliminated the need for a separate steam generator for fuel reforming (other than forstart-up). Designing for low pressure-drops across the entire system minimised theparasitic losses and contributed to the high system efficiency.
A realistic dynamic model of the SOFC generator has been developed using suitablesoftware. The model has been used to simulate the performance of systems with andwithout cathode gas recycle. The results of the dynamic model show close correlationwith steady state simulations.
The dynamic model has enabled the testing of different control strategies, and allowedthe simulation of the dynamic response of the system to external changes. The resultsindicate that the proposed system is controllable over the specified range of 100-50%electrical power, and that the dynamic response to load changes is fast. It is alsopossible to control closely the stack temperature, minimising thermal stresses on thestack during dynamic operation.
It has also been demonstrated that at least one of the proposed systems can be easilyand relatively quickly started from cold, an important aspect for any practical system.A hot standby mode has also been demonstrated.
It has been shown that a system with cathode recycle gives a more efficient andcompact overall system but the control is more complex.
A design for the power conversion has been developed that allows for the particularattributes of the SOFC and meets the requirements for connecting the SOFC system tothe grid.
A converter system has been designed to provide power conversion into the mainssupply for a solid oxide fuel cell. This has required both the specifications for the fuelcell and the specifications for the mains supply to be identified so that a powerconversion system could be devised that can successfully marry the two. Thefollowing conclusions can be drawn from this work:
21
The design of a planar SOFC stack based on anode-supported cells has beencompleted. The planar stack design, which incorporates an internal manifoldstructure and a parallel flow arrangement, has been tested with 20 cm x 20 cm cells.Outputs over 0.5 Wcm-2 have been reached at 800°C on hydrogen / 3% steam.Internal reforming with this stack has also been demonstrated.
The degradation rate of the stacks is, currently, relatively high. The probableexplanation for this is corrosion of, and Cr-evaporation from, the ferritic stainlesssteel interconnect plate. Deterioration in the performance of the cathode currentcollection is also a problem. As a result of this degradation the testing of a 5 kWstack for 2,000 hour on reformed methane was delayed. It should be noted thatinvestigation and prevention of performance degradation lay outside the scope of thisproject.
22
8. RECOMMENDATIONS
The design and performance modelling results clearly demonstrate the potential toachieve a compact, low-cost and highly efficient SOFC system. The logical nextphase is the building of a complete system demonstration. In preparation for thesystem demonstration phase, the following areas should be addressed.
− Build and test balance-of-plant components and sub-systems to verify theirperformance
− Implementation of the balance-of-plant system control developed in the project− Detailed circuit design of power electronics elements− Detailed design of the controls including process synchronisation and public
electricity supply interfacing protocol
Evaluation of stack performance with respect to predicted system operatingconditions, in particular:− Operation with internal reforming− Transient performance− Evaluate stack source impedance to inform power conversion design
Finally, although the development of the SOFC stack was outside the scope of thisproject, it is clear that in order to exploit the potential of planar, intermediatetemperature SOFC technology, development of the stack should focus on:
The work of the SOFC team at ALSTOM Research & Technology Centre in supportof this work is acknowledged.
The work reported here includes the input of the project Partners who are thanked fortheir co-operation throughout the course of the project. In particular the work of thefollowing individuals is acknowledged: from Forschungszentrum Jülich; Bert deHaart, Ludger Blum, Ernst Riensche and Dieter Froening and from Prototech; ArildVik and Paal Bratland.
24
Figure 1: Process flowsheet for the 20 kW SOFC system validation case.
25
Figure 2: Flowsheet for a SOFC system with anode recycle loop.
26
Figure 3: Balance of plant layout for 20 kW SOFC system
air compressor
steamgenerator
air pre-heater
pre-reformer
afterburner
water heater
27
Figure 4: Schematic layout of 20 kW SOFC system
Figure 5: Photo-rendered impression of packaged 20 kW unit with footprint1.6 x 0.8 m2.
4 stacks
air heaterair compressor
power conversionand system control
steam generatorwater heater
afterburner
pre-reformer
28
Figure 6: Photo-rendered impression of air-side heat exchanger for system withcathode gas recycle.
Figure 7: Temperature profile in heat exchanger. Cold side on the left, hot sideon the right.
29
Figure 8: Predicted power output of the system with no cathode recycle duringload-change simulation.
Figure 9: Predicted power output of the system with cathode recycle duringload-change simulation.
Pow er output at 100-50% DC pow er
0
5
10
15
20
25
0 500 1000 1500 2000 2500 3000 3500
Tim e / s
Power / kW
Stack DC power System gross AC power System net AC power Useful heat
100%
80%
71%
62%
51%
Pow er output at 100-50% DC pow er
0
5
10
15
20
25
0 500 1000 1500 2000 2500 3000 3500
Tim e / s
Power / kW
Stack DC power System gross AC power System net AC power Useful heat
100%
80%
71%
63%
51%
30
Figure 10: Changes in electrical and thermal efficiencies for a system withoutcathode recycle, for a load-change simulation over the range 100-50% DC power
Cell voltage and current density at 100-50% DC pow er
0.700
0.720
0.740
0.760
0.780
0.800
0.820
0.840
0.860
0 500 1000 1500 2000 2500 3000 3500
Tim e / s
Cell voltage/
0
50
100
150
200
250
300
350
Current density/ mAcm
-2
Cell voltage M ean current density
100%
80%
71%
62%
51%
Figure 11: Changes in cell voltage and stack current density for a system withoutcathode recycle, for a load-change simulation over the range 100-50% DC power
Efficiencies at 100-50% DC pow er
0
10
20
30
40
50
60
0 500 1000 1500 2000 2500 3000 3500
Tim e / s
Efficiency (LHV basis
Net electrical efficiency Therm al efficiency
100%80%
71% 62% 51%
31
Pow er output/use during the startup procedure
-5
0
5
10
15
20
25
0 500 1000 1500 2000 2500 3000
Tim e /s
Power/ kW
Stack DC power System net AC power Power for steam generator Power to furnace
1
2
3 4
5 6
7
8
Figure 12: System electrical power output/demand during the start-upprocedure
Figure 13: Block diagram of a fuel cell power conditioning system
32
Figure 14: Example from power conditioning simulation of SOFC voltage andcurrent regulation
33
average current density : 300 mA/cm 2
30% pre-reformed methanefuel utilisation: 70%
heat-transferby radiationon 2 sides
λ = 6Conditions:
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cmair
fuel
current density distribution / mA/cm² temperature distribution / °C
700
800
Stack with 265 anode substrate cells (20 cm x 20 cm)
air and fuelinlet temperature:700 °C
720
740
760
780
0
500
100
200
300
400
counter-flow
Stack with 265 anode substrate cells (20 cm x 20 cm)
Temperatures °C Energy balance / per cell W Fuel gas compositionfuel inlet 700 electric power output 73.5 in out
air inlet 700 heat transfer by air 23.0 mol% mol%
fuel outlet 731 heat transfer by fuel 1.3 H2 28.63 18.58air outlet 782 heat transfer by radiation 13.9 CO 1.99 3.10maximum 800 CO 2 5.31 15.00
Electrical efficiency % CH 4 17.10 0.13Operation parameters related on converted fuel 65 H2O 46.97 63.19fuel utilisation % 70 related on input fuel 43 N2 0.00 0.00air stoichiometry mol/ mol 6current density mA/ cm² 300 Pressure loss mbar
cell voltage mV 679 in air channels 6power density mW/cm² 204 in fuel channels < 1
Figure 15: Modelled temperature distributions for a single stack layer andoverall for the 20 kW module
34
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
710
810
730
750
770
790
fuel: pre-reformed natural gas with anode gas recyclefuel utilisation: 69%
65 cells 20 cm x 20 cmcounter-flowadiabatic
λ=
3.7
Conditions: air and fuelinlet temperature:650 °C
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air
cell length x / cm10 201550
0
5
10
15
20
cell
wid
th y
/ cm
fuel
air s
toic
hiom
etry
cons
tant
electrical load: 100% 50% 25%
air s
toic
hiom
etry
varia
ble λ
=3.7
λ=
3.7
λ=
3.7
λ=
2.0
λ=
2.7
tem
pera
ture
/ °
C
fuel: pre-reformed natural gas with anode gas recyclefuel utilisation: 69%
