Modelagem matemática aplicada ao estudo de células a combustível a etanol direto

Abstract

The current fuel cell market is dominated by Hydrogen cells. Nevertheless, due to Hydrogen’s high reactivity, the difficulty in storage and the lack of infrastructure to distribute it, Hydrogen cells might not be the most attractive ones for everyday applications. Direct alcohol cells stand as an alternative because volatile alcohols (methanol and ethanol) can be oxidized at low temperatures (90 °C) and are easily stored and transported since they are liquids. Direct methanol fuel cells are already more commercially established than direct ethanol fuel cells (DEFC), but ethanol has advantages over methanol, namely: lower toxicity, higher theoretical energy density, and greater availability. For DEFC to become economically and technically viable, many challenges still need to be overcome, in specific, the electro-oxidation kinetics at the cell anode, which proceeds slowly, forms less-oxidized intermediate products and reduces cell efficiency. The objective of this master’s project is to model and simulate the ethanol oxidation kinetics in a DEFC. Three models were considered and adjusted to previously collected experimental data: 1) a first-principles ideal model based on Tafel kinetics and complete ethanol oxidation for Pt-Sn catalysts; 2) a realistic first-principles model for the incomplete electro-oxidation of ethanol also for several Pt-Sn catalysts; and 3) a fuzzy model to relate the structural and electronic properties of the Pt3Sn catalyst to cell performance (i.e., predicting current density). All models were implemented in MATLAB. The realistic model had a very good fit for all catalysts studied, as seen by small RMSE (Root Mean Squared Error) values, ranging from 0,22 to 4,21 A/cm2, while also predicting the surface coverage distributions in agreement with previous literature works (experimental basis). The fuzzy model structure also exhibited an excellent fit to experimental data. The addition of the integrated intensity as an input variable did not affect the fitted parameters for the other input variables (crystal size, surface area, presence of PtSn phase and cell potential) and, therefore, it was possible to create a broader and more relevant model that includes an electronic property of the catalyst. Response surface analyses, corroborated by particle swarm optimization, indicated that in order to maximize power density the greatest effect comes from decreasing crystal size. Medium potentials and medium integrated intensity are also favorable. In addition, presence of the PtSn phase in moderate amounts is not unfavorable. With these values it was possible to predict the optimization of the power density value to 24,3 mW/cm2, compared to a maximum experimental value of 19,6 mW/cm2.