Thumbnail Image

Distributed computing & solid oxide fuel cell modelling

Solid oxide fuel cells (SOFC) are a power generation technology that offers the potential to revolutionize the energy industry. Lower emissions and higher energy efficiencies than conventional thermal generation but, moreover, the ability to deliver these improvements in small generators, enables the distributed generation paradigm to flourish. The objective of this work was two-fold; to develop a model to aid in the optimization of a tubular SOFC design, and to develop a distributed computing platform for the solution of large scale models on a network of desktop computers. Distributed computing is the division of a computational problem amongst a number of computers. While a well established technology, having roots in the super-computer application space, distributed computing has never before been applied to SOFC modelling. A test application for distributed computing is detailed, showing the application of distributed computing techniques to a SOFC stack model. The test case is found to scale well with up to four clients, providing linear growth in processing overhead with each additional client. This illustrated the potential for a detailed SOFC model to be distributed across a network of PCs. Computer modelling of an anode supported tubular SOFC is rare in the literature. Validation of a finite-element model was undertaken against experimental data, showing good correlation. The model was then exercised to optimize aspects of the cell design revealing key limitations in the anode electronic conductivity and cathode activation overpotential. Modelling predicted that the addition of a metallic mesh inside the cell would increase the output voltage by 28 mV and an experiment with two samples and two controls showed a gain of 30 mV. Simulation of the cathode showed that improving the exchange current density to accepted levels (i.e. from 20 mA.cm⁻² to 200 mA.cm⁻²) would give a concomitant increase in cell voltage from 580 mV to 658 mV (all other variables constant). Reworking the distributed application to a generic interface for distributing Matlab problems allowed for a demonstration of the distributed solution of a short stack of detailed (finite element) fuel cells. Four computers were used in the test one coordinating server and three clients to solve the models. Even though the three client computers were extremely unbalanced (individual solution times of 5, 25, and 30 minutes), the net effect was a solution time close to the sum of the processing power of the three or a 17% reduction from the total time for the best performing client. Again, this demonstrated the potential advantage to be gained from applying distributed computing to fuel cell models and further investigation with additional clients is strongly recommended. This thesis describes advances in the design of a novel tubular SOFC gained through the use of a computer model. The demonstration of distributed computing technology for solving complex models is unique in the field of fuel cells and represents a significant contribution to fuel cell science and engineering.
Type of thesis
Foster, R. E. (2004). Distributed computing & solid oxide fuel cell modelling (Thesis, Doctor of Philosophy (PhD)). The University of Waikato, Hamilton, New Zealand. Retrieved from https://hdl.handle.net/10289/13256
The University of Waikato
All items in Research Commons are provided for private study and research purposes and are protected by copyright with all rights reserved unless otherwise indicated.