0
Research Papers

Prediction of Direct Methanol Fuel Cell Stack Performance Using Artificial Neural Network

[+] Author and Article Information
M. A. Rafe Biswas

Department of Mechanical Engineering,
University of Texas at Tyler,
Tyler, TX 75799
e-mail: mbiswas@uttyler.edu

Melvin D. Robinson

Department of Electrical Engineering,
University of Texas at Tyler,
Tyler, TX 75799
e-mail: mrobinson@uttyler.edu

1Corresponding author.

Manuscript received December 14, 2016; final manuscript received April 30, 2017; published online June 21, 2017. Assoc. Editor: Matthew Mench.

J. Electrochem. En. Conv. Stor. 14(3), 031008 (Jun 21, 2017) (7 pages) Paper No: JEECS-16-1162; doi: 10.1115/1.4036811 History: Received December 14, 2016; Revised April 30, 2017

A direct methanol fuel cell (DMFC) converts liquid fuel into electricity to power devices, while operating at relatively low temperatures and producing virtually no greenhouse gases. Since DMFC performance characteristics are inherently complex, it can be postulated that artificial neural networks (NN) represent a marked improvement in prediction capabilities. In this work, an artificial NN is employed to predict the performance of a DMFC under various operating conditions. Input variables for the analysis consist of methanol concentration, temperature, current density, number of cells, and anode flow rate. The addition of the two latter variables allows for a more distinctive model when compared to prior NN models. The key performance indicator of our NN model is cell voltage, which is an average voltage across the stack and ranges from 0 to 0.8 V. Experimental studies were conducted using DMFC stacks with membrane electrode assemblies consisting of an additional unique liquid barrier layer to minimize water loss to atmosphere. To determine the best fit to the experimental data, the model is trained using two second-order training algorithms: OWO-Newton and Levenberg–Marquardt (LM). The topology of OWO-Newton algorithm is slightly different from that of LM algorithm by employing bypass weights. The application of NN shows rapid construction of a predictive model of cell voltage for varying operating conditions with an accuracy on the order of 104, which can be comparable to literature. The coefficient of determination of the optimal model results using either algorithm were greater than 0.998.

FIGURES IN THIS ARTICLE
<>
Copyright © 2017 by ASME
Your Session has timed out. Please sign back in to continue.

References

Figures

Grahic Jump Location
Fig. 1

Schematic of the DMFC cell used in the experimental setup

Grahic Jump Location
Fig. 2

VI curve of a four-cell DMFC stack under a steady-state operating condition

Grahic Jump Location
Fig. 3

Illustration of a generic fully connected multilayer perceptron

Grahic Jump Location
Fig. 4

Comparison of model VI curves to experimental data of a four-cell DMFC stack where NN models are based on OWO-Newton and LM

Grahic Jump Location
Fig. 5

Comparison of model and experimental VI curves of an eight-cell DMFC stack where NN models are based on OWO-Newton and LM

Grahic Jump Location
Fig. 6

Experimental and NN-based OWO-Newton model curves of a four-cell DMFC for operating temperatures of 45 °C and 60 °C

Grahic Jump Location
Fig. 7

Experimental and NN-based LM Model curves of a four-cell DMFC for operating temperatures of 45 °C and 60 °C

Grahic Jump Location
Fig. 8

Experimental and NN-based OWO-Newton model curves of a four-cell DMFC for operating methanol concentration of 1 M and 1.5 M

Grahic Jump Location
Fig. 9

Experimental and NN-based LM Model curves of a four-cell DMFC for operating methanol concentration of 1 M and 1.5 M

Tables

Errata

Discussions

Some tools below are only available to our subscribers or users with an online account.

Related Content

Customize your page view by dragging and repositioning the boxes below.

Related Journal Articles
Related eBook Content
Topic Collections

Sorry! You do not have access to this content. For assistance or to subscribe, please contact us:

  • TELEPHONE: 1-800-843-2763 (Toll-free in the USA)
  • EMAIL: asmedigitalcollection@asme.org
Sign In