32
SOFC Development at PNNL: Overview J. Stevenson, B. Koeppel, Y. Chou, J. Hardy, C. Coyle, J. Choi, Z. Xu, N. Karri, K. Recknagle, C. Wang, J. Bao, N. Canfield, and J. Bonnett April 30, 2019
SOFC Development at PNNL: Overview
J. Stevenson, B. Koeppel, Y. Chou, J. Hardy,C. Coyle, J. Choi, Z. Xu, N. Karri, K. Recknagle, C. Wang, J. Bao, N. Canfield, and J. Bonnett
April 30, 2019
2
Scope of Work• Core Technology Program
Materials Development Cathode materials and interactions
• Effects of volatile species (Cr, Sr) on cell performance• Mitigation of Cr poisoning: Evaluation of Cr capture materials• Cathode contact materials: Enhancing reliability of cathode/contact materials interfaces
Interconnects/BOP• Co-free protective coatings for metallic interconnects
• Core Technology Program Modeling/Simulation
SOFC Stack and System Modeling Tool Development Modeling of Stack Degradation and Reliability
• Small-Scale SOFC Test Platform Evaluation of performance and reliability of new stack technologies (3-10 kW)
3
Cr Poisoning
• Challenges Quantitative understanding of threshold concentrations and mechanisms for Cr
poisoning of SOFC cathodes Mitigation of effects of volatile Cr species on cathode performance
• Approaches Determination of relationship between Cr concentration in cathode air stream and rate
of degradation in cathode performance LSM and LSCF-based cathodes Poster: Effects of Cr Concentrations in Air on LSM/YSZ and LSCF Cathode Degradation
(John Hardy)
Evaluation/optimization of Cr “getter” materials intended to capture volatile Cr species May be located upstream of stack and/or within stack (“on-cell” capture) Possibly use upstream getter as primary, and “on-cell” getter as secondary (“polishing”)
4
Cr Gettering Materials
• In previous work, LSCF perovskites with high Sr content were shown to be effective as upstream getters due to high reactivity with Cr vapor species (forming SrCrO4 as reaction product).
• For on-cell applications, Cr-gettering material needs to have matched CTE, high electrical conductivity, chemical compatibility, and thermal stability.
• Approach: Evaluate LSCF / LSM mixtures as dual purpose cathode contact / Cr getter materials.
02468
1012141618
0.0 0.1 0.2 0.3 0.4 0.5
CTE
(x10
-6/o
C)
volume fraction of LSCF4628
CTE of LSM20/LSCF4628 series as-sintered
prediction experimental
Poster: Cr Mitigation by LSM-LSCF Composites for Solid Oxide Fuel Cells (Matt Chou)
Chart1
00
0.10.1
0.20.2
0.30.3
0.40.4
prediction
experimental
volume fraction of LSCF4628
CTE (x10-6/oC)
CTE of LSM20/LSCF4628 series as-sintered
12.26
12.26
13.1965748622
12.49
14.1158289259
13.17
15.0182382791
13.41
15.9042617202
13.87
composite
att6h 1550C4h sintered
estimate CTE by turner equation2.779full density
bulk modulus =E/(3*(1-2u)), u=0.21 of alumina2075000033350000measuredmeas'd5000000060000000CTEbulk modulusdensityE Gpapoisson
alpha(comp)=sum(aixKixVi)/sum(KixVi)LSM20LSCF4628LSCYSZCeO2mullitemullitemullitecordieriteSiO2SiCSi3N4Al2O3Al2O3Al2O3LSM2012.24926.6751100.3
E (Gpa)135.9159.2142.4143.1230.0158.1390.5344.8413.8
G (Gpa)50.160.453.563155.5
K (Gpa)133.2145.8140.311510091146107.48142266.3232273
a12.2620.917.910.58.925.45.45.41.70.568.88.88.8
poisson0.330.3180.3310.2380.2380.2380.2380.2550.2380.2380.2380.2380.2560.2560.256
measuredlowerupper0.21-0.27
VfCTE (ZrO2/LSCo)CTE (CeO2/LSCo)CTE (mullite/LSCo)CTE (mullite/LSCo)CTE (mullite/LSCo)CTE (cordierite/LSCo)CTE (quartz/LSCo)CTE aluminaCTE aluminaCTE alumina
0.0017.917.917.917.917.917.917.917.917.917.9
0.0517.617.216.917.116.616.917.317.117.217.1
0.1017.316.916.416.616.016.417.016.316.516.3
0.1517.016.515.916.215.315.916.715.615.815.6
0.2016.616.215.415.714.715.316.315.015.214.9
0.2516.315.814.815.214.014.715.914.414.714.3
0.3016.015.013.714.112.813.415.013.814.113.8
0.4015.314.212.513.011.512.013.912.813.112.8
0.5014.611.912.211.9
K/Ko10.90.8
Vfmullite/LSComullite/LSComullite/LSCo
0.0017.917.917.9
0.1016.616.516.3
0.2015.315.114.8
0.2514.714.414.1
0.3014.013.813.4
0.4012.812.512.1
0.5011.511.210.8
as sinteredaged
LSM20/LSCF4628LSM20/LSCF4628LSM20/LSCF4628
Vfpredictionexperimentalexperimental
0.0012.2612.2612.11
0.1013.2012.4912.36
0.214.1213.1712.88
0.3015.0213.4113.42
0.4015.9013.8713.68
1.0020.9020.9021.08
VfCTE (ZrO2/LSCo)exp'tCTE aluminaexp'tlCTE aluminaCTE alumina
017.917.90.0%17.917.90.0%17.917.9
0.0517.617.7-0.6%17.117.2-0.7%17.217.1
0.117.317.01.6%16.316.20.7%16.516.3
0.1517.016.15.1%15.615.22.7%15.815.6
0.216.615.015.214.9
0.2516.314.414.714.3
0.316.013.814.113.8
0.415.312.813.112.8
0.514.611.912.211.9
composite
CTE (ZrO2/LSCo)
CTE (mullite/LSCo)
CTE (cordierite/LSCo)
CTE (quartz/LSCo)
volume fraction of low CTE phase
CTE (ppm/oC)
CTE of composite LSCo
R=1
R=0.9
R=0.8
volume fraction of low CTE phase
CTE (ppm/oC)
CTE of composite LSCo/mullite
CTE (ZrO2/LSCo)
CTE (mullite/LSCo)
CTE alumina
CTE alumina
volume fraction of low CTE phase
CTE (ppm/oC)
CTE of composite LSCo
CTE (ZrO2/LSCo)
CTE (mullite/LSCo)
CTE alumina
CTE alumina
volume fraction of low CTE phase
CTE (ppm/oC)
CTE of composite LSCo
mullite 1550C4h measured
E (lower bound)
E (upper bound)
volume fraction of low CTE phase
CTE (ppm/oC)
CTE of composite LSCo
prediction
experimental
volume fraction of LSCF4628
CTE (x10-6/oC)
CTE of LSM20/LSCF4628 series as-sintered
aged
as-sintered
volume fraction of LSCF4628
CTE (x10-6/oC)
LSM20/LSCF4628 series before/after ageing
prediction
experimental
volume fraction of YSZ
CTE (x10-6/oC)
CTE of LSCo/YSZ
prediction
experimental
volume fraction of Al2O3
CTE (x10-6/oC)
CTE of LSCo/Al2O3
5
Ceria Barrier Layers: Sr Volatility
• Challenge Increased cell resistance through formation of insulating Sr zirconate at
ceria/electrolyte interface during sintering of LSCF-based cathodes with doped ceria barrier layers
After cathode sintering, Sr observed in cathode and at YSZ interface, but not in ceria layer
• Approach Investigate likelihood of vapor phase transport of Sr from cathode to ceria/electrolyte
interface
Poster: Investigating Sr Vapor Phase Evolution from LSM/YSZ and LSCF Cathodes During and After Sintering (John Hardy)
6
Ceria Barrier Layers: Sr Volatility
• Thermodynamic calculations show that SrO heated to 1100°C can produce Sr vapor pressures of the same order of magnitude as Cr vapor from chromia at 750°C.
Variables SettingsCathode State Sintered; Unsintered
Cathode Composition LSM/YSZ; LSCFSubstrate (Sr Sink)
CompositionGDC; YSZ
Spacer Thickness 1 mm; 10 mmTest Temperature 1000°C; 1100°C; 1200°C
Time at Temperature 0.5 h; 2 h
SEM-EDS is performed on cathode-facing surface of top substrate
As compared to unexposed substrates, statistically significant increase in Sr content was measured for tests with:• No cathode presintering (vs 1100°C for 2 h)• LSCF cathodes (vs LSM/YSZ)• 1 mm distance to substrate (vs 10 mm)• 1100 or 1200°C temperature (vs 1000°C)
7
Cathode / Interconnect Contact Materials
• Challenge Electrical contact materials at cathode / interconnect interfaces in planar stacks tend to
be mechanical “weak link,” especially during thermal cycling, due to brittle nature of ceramic materials and/or thermal expansion mismatch with adjacent components Low processing temperatures and constrained sintering conditions during stack fabrication lead to
low intrinsic strength and low bonding strength of ceramic contact materials, especially at contact-to-cathode interface
Use of metallic contact materials limited by cost, volatility, and/or electromigration
• Approach Use composite approach to develop ceramic-based contact materials having improved
mechanical reliability by reducing thermal expansion mismatch and increasing contact strength/toughness
Poster: Composite Cathode Contact Material Development - Validation in Stack Fixture Test and Effect of Strong Fibers (Matt Chou)
8
LSCo / mullite / fiber composite contact materials
• LSCo perovskite offers very high electrical conductivity but also has high CTE (~18x10-6/oC) as cathode contact one needs to overcome the large residual stresses by:
• Reduce thermal stresses by adding low CTE phase - mullite (~5.4x10-6/oC)
• Enhance the strength/toughness by reinforcement with strong short fibers with high elastic modulus (YSZ or Al2O3)
9
LSCo / mullite / fiber composite contact materials
• Presence of short fibers enhanced the contact bonding strength, initially and after 10 thermal cycles
850oC3h sinteredMaterials strength (MPa) stdev (MPa) strength (MPa) stdev (MPa)
LSCo 0 na 0 naLSCo/10%mull 0 na 0 na
LSCo/10/5%Al2O3 2.08 0.81 2.11 0.51LSCo/10%mull/10%Al2O3 1.74 0.44 1.89 0.56
950oC3h sinteredMaterials strength (MPa) stdev (MPa) strength (MPa) stdev (MPa)
LSCo 0 na 0 naLSCo/10%mull 0 na 0 na
LSCo/10/5%Al2O3 3.25 1.09 3.44 1.19LSCo/10%mull/10%Al2O3 2.72 0.8 2.68 0.46
as-sintered after 10 TC
as-sintered after 10 TC
10
Interconnect / BOP Coatings
• Challenges Metallic interconnects susceptible to oxidation (leading to high electrical resistance), Cr
volatilization (leading to Cr poisoning), and reactions with seals (leading to mechanical failure)
Other metallic components susceptible to Cr volatilization
• Approaches Electrically conductive Mn-Co spinel coatings exhibit good performance; due to
possible issues with Co cost and availability, developing Co-free alternatives Cu-Mn-O; Ni-Mn-O; Cu-Fe-O
Reactive air aluminization for applications that don’t require electrical conductivity Simple slurry-based process Fabrication in air at temperatures as low as 900°C
11
Co-free Electrically Conductive Protective Coatings: DoE Optimization of Spray Coating Parameters
Factors Level 1 Level 2 Level 3 Level 4Viscosity 37cP 17cP 9cP 5cP
Coating speed 40mm/sec 60mm/sec 80mm/sec 100mm/sec
Head height 15mm 25mm 35mm 45mm
Ink feeding rate 0.5ml/sec 1ml/sec 1.5ml/sec 2ml/sec
Air flow rate 30ml/sec 40ml/sec 50ml/sec 60ml/sec
Composition Viscosity Coating speed
Head height
Ink feeding rate
Air flow rate
(Cu1.3Mn1.7O4) 3 4 1 1 2(Cu1.5Mn1.5O4) 3 3 4 1 2
(NiMn2O4) 4 4 1 3 2(Ni1.5Mn1.5O4) 4 4 2 3 2(Cu1.5Fe1.5O4) 4 4 2 2 2
(CuFe2O4) 4 4 4 3 1
Table of Factors and Levels for DoE Optimization
Optimized Conditions for each Candidate Composition
Preliminary coating characterization has been completed
Isothermal (800 and 900°C) and thermal cyclic testing is in progress
Poster: (M, Mn or Fe)3O4 spinel for Advanced Electrical Conductive Layer for SOFC Stacks(Jung-Pyung Choi)
12
Small-Scale SOFC Test Platform• Purpose:
Evaluate performance and reliability of emerging stack technologies (3-10 kW)under realistic operating conditions
Estimated completion: May, 2019
• Test conditions: Steady-state isothermal
Variables: temperature, current, voltage, fuel Thermal cycling E-stop cycles (redox tolerance) Variable anode recycle rates
Poster: Small-Scale SOFC Test Platform (Brent Kirby)
13
Small-Scale SOFC Test PlatformKey features:
• Operation on methane via steam reforming
• Anode recirculation loop
• High efficiency microchannel heat exchangers for heat recuperation and anode/cathode stream temperature equalization
• Automated control system
14
Focus of Current PNNL Modeling Efforts
Focus mainly on SOFC systems model support
Stack Reduced Order Model
(ROM)
Pressurized Systems
Mechanical Reliability
NETL ASPEN+ System Models
Cell Channel Model
System Operating Conditions
Current-Voltage
Performance
Atmospheric NGFC
Atmospheric IGFC
3D Stack FEA Model
State of Art (SOA)
Future Performance
End of Life (EOL)
NETL Coarsening
and Poisoning
Models
15
Modeling Tools and Analysis Overview
Challenges: Develop modeling tools to evaluate SOFC behavior Integrate modeling results at different scales to improve design Understand performance degradation mechanisms and control strategies
FY19 Approach:1. Develop ROMs to support NETL system evaluations
Provides more accurate stack representation for system design Poster: Use of Reduced Order Models (ROMs) to Predict SOFC Stacks Performance (Jie Bao)
2. Evaluation of Cr poisoning Incorporate NETL model to understand long-term performance impacts at the stack level
3. Evaluation of creep Use FEA to understand time-dependent deformation on mechanical reliability of the stack Poster: Influence of Anode Creep on the Structural Reliability of SOFCs (Brian Koeppel)
4. Evaluation of metal-supported cells
16
1A. Generic Material Flowchart for ROM
May 7, 2019
VGR: vent gas recirculation concept
NGFC: Natural gas fuel w/ external reformer
IGFC: Syngas fuel w/o external reformer
CCS: w/ O2 separator
No CCS: w/o O2 separator
17
1A. ROMs Generated for Various SOFC Systems
• Provided 23 ROMs to NETL for supporting Pathway Studies Type: Natural gas fuel cell (NGFC), integrated gasification fuel cell (IGFC-
conventional, enhanced, and catalytic); Pressure: atmospheric or pressurized; Performance: state-of-art (SOA), future performance with reduced cell losses; Configuration: with or without carbon capture (CCS or w/o CCS), inclusion of vent
gas recirculation (VGR).
May 7, 2019
Average Current Density 2000-6000 A/m2Internal Reforming 0-100%Oxidant Recirculation 0-80%Oxygen-to-Carbon Ratio Target @ Stack Inlet 1.5-3.0Fuel Utilization (including recirculation loop) 40-95%Oxidant Utilization (including recirculation loop) 12.5-83.3%Oxidant Stack Inlet Temperature 550-800oCFuel Loop Inlet Temperature 15-600oCSystem Pressure 1-5 atmVent Gas Recirculation 30-97%
18
1B. Machine Learning (ML) Classification
• Issue: ROM will provide a result for non-physical parameter combinations
• Goal: Identify true operating domain• Approach:
Implement ML methods: support vector machine (SVM), random forest, decision tree, and neural network (NN)
Apply cross-validation to determine prediction accuracy
• Results:
May 7, 2019
ML Method Prediction AccuracyNN 93.0%
SVM 91.4%Random Forest 89.0%Decision Tree 82.6%
Classified Correctly
ClassifiedIncorrectly
Evaluation of 100 Non-Physical Cases
19
1C. Use of ML for ROM Generation
• Goal: Evaluate deep learning regression-based ROM as alternative to the Kriging regression-based ROM to predict SOFC stack performance
• Approach: Built a deep neural network (DNN)-based ROM• Results: DNN ROM can provide better prediction accuracy and reduce the
prediction error by a factor of 2-3 compared with existing Kriging ROM
May 7, 2019
Algorithm of Individual NeuronGeneral DNN Framework
20
1C. Deep Learning (DL) vs. Kriging ROM Results
May 7, 2019
Parameters from PDF DNN Kriging Improvement Ratio UB for 95% CI 0.0057 0.0135 2.36LB for 95% CI -0.0060 -0.0136 2.27
Max Error 0.0130 0.0354 2.72Min Error -0.0140 -0.0362 2.59
Voltage Prediction Results
21
2. Modeling of Stack Performance DegradationObjective: Collaborate with NETL and university partner modelers to bridge the scales of degradation from microstructure to stack.Approach: • Incorporate mechanisms affecting cell microstructure and electrochemical
performance into PNNL stack modeling tools (SOFC-MP)Accomplishments:• Thermal coarsening of the electrodes was added in FY17 • Chromium poisoning of the cathode was added in FY19• Demonstrated simulation of multi-mode stack performance degradation
22
2. Chromium Poisoning Mechanism• Chromium poisoning begins with interconnects and components upstream of the stack
containing chromium• When the components are exposed to inflowing cathode air (including humidity), chromium
oxide (Cr2O3(s)) scale is formed on the surface which results in chromium vapor species CrO2(OH)2(g) in the air stream
• The gas reacts with the cathode at the triple-phase-boundary (TPB) reaction sites depositing a solid chromium oxide (Cr2O3) with a deposition reaction current (𝑖𝑖𝐷𝐷,Cr2O3)
• The oxide irreversibly covers the TPB area (𝜃𝜃𝑇𝑇𝑇𝑇𝑇𝑇,Cr ) such that 𝐋𝐋𝐓𝐓𝐓𝐓𝐓𝐓 is decreased and the oxygen reduction reaction is diminished over time causing the electrochemical performance to degrade
𝑃𝑃CrO2 OH 2,𝑒𝑒𝑒𝑒 = 4.15 × 10−3𝑎𝑎Cr2O3
0.5 𝑃𝑃O20.75𝑃𝑃H2O exp −
5.35 × 104
𝑅𝑅𝑅𝑅
2CrO2 OH 2 𝑔𝑔 + 6e− → Cr2O3 𝑠𝑠 + 2H2𝑂𝑂 𝑔𝑔 + 3O2−
𝑖𝑖𝐷𝐷,Cr2O3 = 𝑖𝑖0𝐷𝐷,Cr2O3𝑙𝑙𝑇𝑇𝑇𝑇𝑇𝑇𝑤𝑤𝑇𝑇𝑇𝑇𝑇𝑇𝑃𝑃CrO2 OH 2𝑃𝑃H2O exp𝐹𝐹
2𝑅𝑅𝑅𝑅𝜂𝜂 − exp −
𝐹𝐹2𝑅𝑅𝑅𝑅
𝜂𝜂
𝜃𝜃𝑇𝑇𝑇𝑇𝑇𝑇,Cr 𝑡𝑡 + Δ𝑡𝑡 = 𝜃𝜃𝑇𝑇𝑇𝑇𝑇𝑇,Cr 𝑡𝑡 − Δ𝑡𝑡1
2𝐹𝐹𝑀𝑀𝐶𝐶𝐶𝐶2𝑂𝑂3
𝜌𝜌𝐶𝐶𝐶𝐶2𝑂𝑂3ℎ𝑇𝑇𝑇𝑇𝑇𝑇∗𝑖𝑖𝐷𝐷,Cr2O3𝑙𝑙𝑇𝑇𝑇𝑇𝑇𝑇𝑤𝑤𝑇𝑇𝑇𝑇𝑇𝑇
𝑙𝑙𝑇𝑇𝑇𝑇𝑇𝑇 = 𝑙𝑙𝑇𝑇𝑇𝑇𝑇𝑇,0 1 − 𝜃𝜃𝑇𝑇𝑇𝑇𝑇𝑇,Cr
23
2. Effect of Fuel on Long-Term Performance
• Example case operating at 800°C and 0.5 A/cm2, 1% H2O in air. On H2 fuel degradation due to coarsening
is greater than that of chromium poisoning Coarsening is a large factor at 800°C
On partially reformed natural gas degradation due to chromium poisoning is increased Due to increased activation polarization (η)
and chromium oxide deposition rate (𝑖𝑖𝐷𝐷,Cr2O3) For both fueling scenarios the degradation
is larger with both modes together than the sum of each mode occurring separately LTPB is decreased by both mechanisms
accelerating the degradation
per 1000 hrCoarse 0.061%Chrome 0.038%both 0.117%
per 1000 hrCoarse 0.054%Chrome 0.059%both 0.135%
24
2. Effect of Cathode H2O on Long-Term Performance
• Example case operating at 800°C and 0.5 A/cm2
• Baseline: 1% H2O in air resulted in 0.038% per 1khr
• When the cathode H2O is increased to 1.5% the degradation is more than tripled to 0.129% per 1khr. Equilibrium partial pressure of chromium
vapor species and deposition reaction current is first order with steam
• If the H2O is decreased to 0.5% the degradation decreases to less than a quarter of the baseline at 0.008% per 1khr
per 1000 hrCoarse 0.054%Chrome 0.059%both 0.135%
Effect of Steam on Cr Poisoning
25
2. Effect of Geometry on Long Term Performance
• Operating on Natural Gas with 60% IR at 750°C and 0.4 A/cm2 at 70% Fuel Utilization
• “Coarsening only” degradation is greater for co-flow For co-flow much of power is generated on 2nd half of cell where
temperature and current density values are highest Coarsening and decrease of LTPB is greatest where temperature
was highest (maximum 19°C higher for co-flow)
• “Chromium only” degradation is greater for counter-flow For counter-flow much of power is generated on 1st half of cell where
temperature and current density values are highest Locally high temperature increases deposition current (𝑖𝑖𝐷𝐷,Cr2O3) and
LTPB coverage (𝜃𝜃𝑇𝑇𝑇𝑇𝑇𝑇,Cr ) degrading LTPB most where power generation is highest
• “Chromium & Coarsening” mechanisms together are synergisticand degradation is greater for counter-flow Activation polarization (η) is decreased toward fuel exit thus deposition
rate is decreased such that synergetic coarsening and coverage of LTPB is not present for this co-flow case (at this temperature)
Flow Configuration
Degradation Mechanism
Chromium Only Degradation (per
1khr)
Coarsening Only Degradation (per
1khr)
Chromium & Coarsening
Degradation (per 1khr)
Co-flow 0.0402% 0.0253% 0.0664%Counter-flow 0.0428% 0.0203% 0.0705%
Fuel Air
T
J
LTPB
26
3. Modeling of Stack Reliability under CreepObjective: Investigate effect of creep on long-term operational reliability of SOFC stacks.Approach: • Implement creep models and study the
influence on reliability using FEA .Accomplishments:• Material creep model parameters were
identified for the SOFC operational range (700 – 800°C)
• Simulations were carried out for realistic operating temperatures for generic multi-cell stack designs
32
1
CC Tcr C eε σ
−
=
1.E-20
1.E-18
1.E-16
1.E-14
1.E-12
1.E-10
1.E-08
1.E-06
1.E-04
0 200 400 600 800 1000
Cre
ep R
ate
(s-1
)
Temperature (°C)
Steel
8YSZ
Ni:8YSZ
LSM
G18
C1 – Creep constantC2 – Stress ExponentC3 – the ratio Q/R
Sample creep strain rates for SOFC materials under 10 MPa stress
27
3. Temperatures from SOFC-MP 3D Tool
1-CellTmax=795°C
15-CellTmax=810°C
45-CellTmax=828°C
°K
Operating Temperature Contours in 1, 15, and 45-cell Stack Models (Tavg≈750°C, FU=86%, AU=16.4% (V=0.7908×NCELL, Idens=0.4 A/cm2)
28
3. 45-Cell Stack Results and Conclusions
• Creep increases failure probabilities of cathode and electrolyte
• Creep typically relaxes peak stresses in the PEN assembly however, the redistributed stresses produce higher net tension regions in electrolyte and cathode leading to higher failure probabilities.
• Effect is more pronounced in the cells near stack end (load frame).
• Pre-load significantly alters creep influence.
0%
20%
40%
60%
0 10 20 30 40 50
Failu
re P
roba
bilit
y [P
f]
Cell Number [Bottom=1 → Top=45]
Cathode Reliability
t=0 hours
t=40,000 hrs
0%
20%
40%
60%
0 10 20 30 40 50
Failu
re P
roba
bilit
y [P
f]
Cell Number [Bottom=1 → Top=45]
Electrolyte Reliability
t=0 hours
t=40,000 hrs
50
60
70
80
0 10 20 30 40 50
Max
Prin
cipa
l Str
ess,
S11
[MPa
]
Cell Number [Bottom=1 → Top=45]
Electrolyte Peak Principal Stress
t=0 hours
t=40,000 hrs
10
15
20
25
0 10 20 30 40 50
Peak
Max
Prin
cipa
l Str
ess [
Pa]
Cell Number [Bottom=1 → Top=45]
Cathode Peak Principal Stress
t=0 hours
t=40,000 hrs
(a) (b)
(c) (d)
0%
20%
40%
60%
0 10 20 30 40 50
Failu
re P
roba
bilit
y [P
f]
Cell Number [Bottom=1 → Top=45]
Cathode Reliability
t=0 hours
t=40,000 hrs
0%
20%
40%
60%
0 10 20 30 40 50
Failu
re P
roba
bilit
y [P
f]
Cell Number [Bottom=1 → Top=45]
Electrolyte Reliability
t=0 hours
t=40,000 hrs
50
60
70
80
0 10 20 30 40 50
Max
Prin
cipa
l Str
ess,
S11
[MPa
]
Cell Number [Bottom=1 → Top=45]
Electrolyte Peak Principal Stress
t=0 hours
t=40,000 hrs
10
15
20
25
0 10 20 30 40 50Pe
ak M
ax P
rinci
pal S
tres
s [Pa
]Cell Number [Bottom=1 → Top=45]
Cathode Peak Principal Stress
t=0 hours
t=40,000 hrs
(a) (b)
(c) (d)
Stack Failure Probabilities / Peak Stresses
29
4. Evaluation of Metal-Supported CellsObjective: Advance modeling capabilityof the PNNL SOFC-MP codes to includea generic metal-supported cell (MSC)Approach: • Identify SoA MSC performance and
implement in PNNL EC model • Simulate performance with SOFC-MP
and evaluate structural reliability w/ FEA.Accomplishments:• MSC performance was validated• Compressive preload and metal support
porosity showed significant effect on mechanical reliability
SourceInflow Temp,
°C
Average Stack
Temp, °C
Current Density, A/cm2
Cell Voltage
(V)
Cell ∆T, C°
TestData N.A. 700 0.4 0.921 N.A.
Model 674 701 0.4 0.923 24.9
SOFC-MP 2D solution for 400 cm2, 1-cell 2-D stack model compared to Nielsen* MSC data (Fuel: 80% H2, 20% H2O) at 13% FU, 3% AU.
# Jimmi Nielsen, Asa H. Persson, Thuy Thanh Muhl, and Karen Brodersen. Towards High Power Density Metal Supported Solid Oxide Fuel Cell for Mobile Applications, Journal of the Electrochemical Society 2018 165: F90-F96
30
Summary• PNNL is using experimental and computational capabilities to extend the knowledge
base in order to accelerate the commercialization of SOFC power systems.
• Posters Effects of Cr Concentrations in Air on LSM/YSZ and LSCF Cathode Degradation (John Hardy) Investigating Sr Vapor Phase Evolution from LSM/YSZ and LSCF Cathodes During and After
Sintering (John Hardy) Cr Mitigation by LSM-LSCF Composites for Solid Oxide Fuel Cells (Matt Chou) Composite Cathode Contact Material Development: Validation in Stack Fixture Test and Effect of
Strong Fiber (Matt Chou) (M, Mn or Fe)3O4 spinel for Advanced Electrical Conductive Layer for SOFC Stacks (Jung-Pyung
Choi) Use of Reduced Order Models (ROMs) to Predict SOFC Stacks Performance (Jie Bao) Influence of Anode Creep on the Structural Reliability of SOFCs (Brian Koeppel) Small-Scale SOFC Test Platform (Brent Kirby)
31
Acknowledgements
The work summarized in this presentation was funded by the U.S. Department of Energy’s Office of Fossil Energy Solid Oxide Fuel Cell Program.
NETL: Shailesh Vora, Joseph Stoffa, Patcharin Burke, and Greg Hackett NETL Site Support: Arun Iyengar, Harry Abernathy, J. Hunter Mason
Thank you
32
SOFC Development �at PNNL: OverviewScope of WorkCr PoisoningCr Gettering MaterialsCeria Barrier Layers: Sr VolatilityCeria Barrier Layers: Sr VolatilityCathode / Interconnect Contact MaterialsLSCo / mullite / fiber composite contact materialsLSCo / mullite / fiber composite contact materialsInterconnect / BOP CoatingsCo-free Electrically Conductive Protective Coatings: DoE Optimization of Spray Coating ParametersSmall-Scale SOFC Test PlatformSmall-Scale SOFC Test PlatformFocus of Current PNNL Modeling EffortsModeling Tools and Analysis Overview1A. Generic Material Flowchart for ROM1A. ROMs Generated for Various SOFC Systems1B. Machine Learning (ML) Classification1C. Use of ML for ROM Generation1C. Deep Learning (DL) vs. Kriging ROM Results2. Modeling of Stack Performance Degradation2. Chromium Poisoning Mechanism2. Effect of Fuel on Long-Term Performance2. Effect of Cathode H2O on Long-Term Performance2. Effect of Geometry on Long Term Performance 3. Modeling of Stack Reliability under Creep3. Temperatures from SOFC-MP 3D Tool3. 45-Cell Stack Results and Conclusions4. Evaluation of Metal-Supported CellsSummaryAcknowledgementsSlide Number 32
