190 likes | 502 Views
PERFORMANCE PREDICTION OF A PROTON EXCHANGE MEMBRANE FUEL CELL USING THE ANFIS MODEL. Yasemin Vural Centre for Computational Fluid Dynamics (CFD) University of Leeds, UK ICAT 08 Conference November 13-14, Istanbul, Turkey. OUTLINE Introduction Modeling & Results
E N D
PERFORMANCE PREDICTION OF A PROTON EXCHANGE MEMBRANE FUEL CELL USING THE ANFIS MODEL Yasemin Vural Centre for Computational Fluid Dynamics (CFD) University of Leeds, UK ICAT 08 Conference November 13-14, Istanbul, Turkey
OUTLINE • Introduction • Modeling & Results • Conclusion
Fuel Cells • Fuel cells are the electrochemical devices that converts • chemical energy into electrical energy • fuel water • oxidant heat • electricity • Clean, high efficiency, quite (no moving parts) energy production • Applications: automotive, stationary power industry, • portable applicatons (mobile phones, PCs) • Types: PEMFC, SOFC, DMFC, Alkaline Fuel cells etc. • Recent Research: material type, manufacturing, understand the processes (through modelling) Fuel Cell Introduction Modeling & Results Conclusion -1-
Proton Exchange Membrane Fuel Cell (PEMFC) Introduction Modeling & Results Conclusion -2- (Source: http://www.fueleconomy.gov)
Proton Exchange Membrane Fuel Cell (PEMFC) • Operating Temp : 60-80 C • Efficiency : 35-45 (%) • Applications : Automotive, small-scale stationary, portable • Challenges • Cost • Lifetime/ Degradation • Start up (subzero temperature, freezing) • Water Management Introduction Modeling & Results Conclusion -3-
Applicationsin Automotive Industry 2008 Honda FCX Clarity Toyota FCHC Ford Explorer Volkswagen Bora -4-
Typical Polarization curve of a PEFMC Voltage loss due to activation polarization Introduction Modeling & Results Conclusion Voltage loss due to ohmic polarization Voltage loss due to concentration polarization -5- (Source: Buasri P.and Salameh Z.H.)
Proton Exchange Membrane Fuel Cell (PEMFC) • Performance (I-V curve) prediction of a cell is important for design improvements. • Measurements in a fuel cell is usually difficult and expensive. • Modelling is an important tool for performance prediction. • Mathematical models: complicated, empirical parameters. • Soft computing models: easier, rapid. Introduction Modeling & Results Conclusion -6-
Proton Exchange Membrane Fuel Cell (PEMFC) Purpose of the study: To predict the performance of a PEM fuel cell using a soft computing technique, namely the ANFIS model and validate the model for different operational conditions. Introduction Modeling & Results Conclusion -7-
Artificial Neuro Fuzzy Inference System (ANFIS) Introduction Modeling & Results Conclusion • combines the advantages of the Artificial Neural Network(ANN) • and Fuzzy Logic (FL) • Advantages: No prior knowledge of the system is necessary. • Solution using MATLAB software, Fuzzy Logic Toolbox -8-
The ANFIS structure Experimental data of Wang et al J. of Hydrogen Energy, 2002. Current density 0 -1.68 A/cm2 Voltage (V) ANFIS Cell temperature 50-90 C Anode humidification temperature 25 -90 C Cathode humidification temperature 40 - 90 C Pressure 1.0-3.74 atm -9-
Results MAPE (%) =1.86 -10-
Results MAPE (%) =2.06 -11-
Effect of the Operational Conditions on the Cell Performance Anode and cathode humidification temperature: 70 C Effect of Cell Temperature: Voltage (V) Cell Temp (C) Current density (A/cm2) -12-
Effect of the Operational Conditions on the Cell Performance Cell temp and cathode humidification temperature: 70 C Effect of Anode Humidification Temperature: Voltage (V) Anode humid. temp (C) Current density (A/cm2) -13-
Effect of the Operational Conditions on the Cell Performance Cell temp and anode humidification temperature: 70 C Effect of Cathode Humidification Temperature: Voltage (V) Cathode humid. temp (C) Current density (A/cm2) -14-
Effect of the Operational Conditions on the Cell Performance Cell temp, anode and Cathode humidification temperature: 70 C Effect of Pressure: Voltage (V) Pressure (atm) Current density (A/cm2) -15-
Conclusion • Models are important tools for the prediction of a fuel cell performance. • The ANFIS model trained and tested with the set of experimental data. • The effects of the operational conditions on the cell performance • were discussed. • ANFIS can be used as a viable tool for the prediction of the cell performance. Introduction Modeling & Results Conclusion -16-