Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA

Brian Carnes, Ken S. Chen, Fangming Jiang, Gang Luo, Chao-yang Wang

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Current computational models for proton exchange membrane fuel cells (PEMFCs) include a large number of parameters such as boundary conditions, material properties, and numerous parameters used in sub-models for membrane transport, two-phase flow and electrochemistry. In order to successfully use a computational PEMFC model in design and optimization, it is important to identify critical parameters under a wide variety of operating conditions, such as relative humidity, current load, temperature, etc. Moreover, when experimental data is available in the form of polarization curves or local distribution of current and reactant/product species (e.g., O2, H2O concentrations), critical parameters can be estimated in order to enable the model to better fit the data. Sensitivity analysis and parameter estimation are typically performed using manual adjustment of parameters, which is also common in parameter studies. We present work to demonstrate a systematic approach based on using a widely available toolkit developed at Sandia called DAKOTA that supports many kinds of design studies, such as sensitivity analysis as well as optimization and uncertainty quantification. In the present work, we couple a multidimensional PEMFC model (which is being developed, tested and later validated in a joint effort by a team from Penn State Univ. and Sandia National Laboratories) with DAKOTA through the mapping of model parameters to system responses. Using this interface, we demonstrate the efficiency of performing simple parameter studies as well as identifying critical parameters using sensitivity analysis. Finally, we show examples of optimization and parameter estimation using the automated capability in DAKOTA.

Original languageEnglish (US)
Title of host publicationASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010
Pages455-461
Number of pages7
DOIs
StatePublished - Dec 1 2010
EventASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010 - Brooklyn, NY, United States
Duration: Jun 14 2010Jun 16 2010

Publication series

NameASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010
Volume1

Other

OtherASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010
CountryUnited States
CityBrooklyn, NY
Period6/14/106/16/10

Fingerprint

Proton exchange membrane fuel cells (PEMFC)
Parameter estimation
Sensitivity analysis
Electrochemistry
Two phase flow
Materials properties
Atmospheric humidity
Boundary conditions
Polarization
Membranes

All Science Journal Classification (ASJC) codes

  • Energy Engineering and Power Technology
  • Fuel Technology

Cite this

Carnes, B., Chen, K. S., Jiang, F., Luo, G., & Wang, C. (2010). Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA. In ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010 (pp. 455-461). (ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010; Vol. 1). https://doi.org/10.1115/FuelCell2010-33038
Carnes, Brian ; Chen, Ken S. ; Jiang, Fangming ; Luo, Gang ; Wang, Chao-yang. / Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA. ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010. 2010. pp. 455-461 (ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010).
@inproceedings{734d97d66b404dad960b1180b95c6b23,
title = "Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA",
abstract = "Current computational models for proton exchange membrane fuel cells (PEMFCs) include a large number of parameters such as boundary conditions, material properties, and numerous parameters used in sub-models for membrane transport, two-phase flow and electrochemistry. In order to successfully use a computational PEMFC model in design and optimization, it is important to identify critical parameters under a wide variety of operating conditions, such as relative humidity, current load, temperature, etc. Moreover, when experimental data is available in the form of polarization curves or local distribution of current and reactant/product species (e.g., O2, H2O concentrations), critical parameters can be estimated in order to enable the model to better fit the data. Sensitivity analysis and parameter estimation are typically performed using manual adjustment of parameters, which is also common in parameter studies. We present work to demonstrate a systematic approach based on using a widely available toolkit developed at Sandia called DAKOTA that supports many kinds of design studies, such as sensitivity analysis as well as optimization and uncertainty quantification. In the present work, we couple a multidimensional PEMFC model (which is being developed, tested and later validated in a joint effort by a team from Penn State Univ. and Sandia National Laboratories) with DAKOTA through the mapping of model parameters to system responses. Using this interface, we demonstrate the efficiency of performing simple parameter studies as well as identifying critical parameters using sensitivity analysis. Finally, we show examples of optimization and parameter estimation using the automated capability in DAKOTA.",
author = "Brian Carnes and Chen, {Ken S.} and Fangming Jiang and Gang Luo and Chao-yang Wang",
year = "2010",
month = "12",
day = "1",
doi = "10.1115/FuelCell2010-33038",
language = "English (US)",
isbn = "9780791844045",
series = "ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010",
pages = "455--461",
booktitle = "ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010",

}

Carnes, B, Chen, KS, Jiang, F, Luo, G & Wang, C 2010, Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA. in ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010. ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010, vol. 1, pp. 455-461, ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010, Brooklyn, NY, United States, 6/14/10. https://doi.org/10.1115/FuelCell2010-33038

Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA. / Carnes, Brian; Chen, Ken S.; Jiang, Fangming; Luo, Gang; Wang, Chao-yang.

ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010. 2010. p. 455-461 (ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010; Vol. 1).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

TY - GEN

T1 - Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA

AU - Carnes, Brian

AU - Chen, Ken S.

AU - Jiang, Fangming

AU - Luo, Gang

AU - Wang, Chao-yang

PY - 2010/12/1

Y1 - 2010/12/1

N2 - Current computational models for proton exchange membrane fuel cells (PEMFCs) include a large number of parameters such as boundary conditions, material properties, and numerous parameters used in sub-models for membrane transport, two-phase flow and electrochemistry. In order to successfully use a computational PEMFC model in design and optimization, it is important to identify critical parameters under a wide variety of operating conditions, such as relative humidity, current load, temperature, etc. Moreover, when experimental data is available in the form of polarization curves or local distribution of current and reactant/product species (e.g., O2, H2O concentrations), critical parameters can be estimated in order to enable the model to better fit the data. Sensitivity analysis and parameter estimation are typically performed using manual adjustment of parameters, which is also common in parameter studies. We present work to demonstrate a systematic approach based on using a widely available toolkit developed at Sandia called DAKOTA that supports many kinds of design studies, such as sensitivity analysis as well as optimization and uncertainty quantification. In the present work, we couple a multidimensional PEMFC model (which is being developed, tested and later validated in a joint effort by a team from Penn State Univ. and Sandia National Laboratories) with DAKOTA through the mapping of model parameters to system responses. Using this interface, we demonstrate the efficiency of performing simple parameter studies as well as identifying critical parameters using sensitivity analysis. Finally, we show examples of optimization and parameter estimation using the automated capability in DAKOTA.

AB - Current computational models for proton exchange membrane fuel cells (PEMFCs) include a large number of parameters such as boundary conditions, material properties, and numerous parameters used in sub-models for membrane transport, two-phase flow and electrochemistry. In order to successfully use a computational PEMFC model in design and optimization, it is important to identify critical parameters under a wide variety of operating conditions, such as relative humidity, current load, temperature, etc. Moreover, when experimental data is available in the form of polarization curves or local distribution of current and reactant/product species (e.g., O2, H2O concentrations), critical parameters can be estimated in order to enable the model to better fit the data. Sensitivity analysis and parameter estimation are typically performed using manual adjustment of parameters, which is also common in parameter studies. We present work to demonstrate a systematic approach based on using a widely available toolkit developed at Sandia called DAKOTA that supports many kinds of design studies, such as sensitivity analysis as well as optimization and uncertainty quantification. In the present work, we couple a multidimensional PEMFC model (which is being developed, tested and later validated in a joint effort by a team from Penn State Univ. and Sandia National Laboratories) with DAKOTA through the mapping of model parameters to system responses. Using this interface, we demonstrate the efficiency of performing simple parameter studies as well as identifying critical parameters using sensitivity analysis. Finally, we show examples of optimization and parameter estimation using the automated capability in DAKOTA.

UR - http://www.scopus.com/inward/record.url?scp=84860285133&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=84860285133&partnerID=8YFLogxK

U2 - 10.1115/FuelCell2010-33038

DO - 10.1115/FuelCell2010-33038

M3 - Conference contribution

AN - SCOPUS:84860285133

SN - 9780791844045

T3 - ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010

SP - 455

EP - 461

BT - ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010

ER -

Carnes B, Chen KS, Jiang F, Luo G, Wang C. Systematic parameter estimation and sensitivity analysis using a multidimensional PEMFC model coupled with DAKOTA. In ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010. 2010. p. 455-461. (ASME 2010 8th International Conference on Fuel Cell Science, Engineering and Technology, FUELCELL 2010). https://doi.org/10.1115/FuelCell2010-33038