论文标题
原型燃料电池数值建模的进度报告
Progress Report on Numerical Modeling of a Prototype Fuel Cell
论文作者
论文摘要
报道了原型燃料电池的数值建模的进展。解决了先前发布的Alpha模型的一些已知局限性,并估计了由于改进模型Beta的离散化而导致的数值不确定性。在第1部分中,将Beta模型与Alpha进行了比较,尽管可以看到很小的差异。讨论了改进的模型的缺点,铺平了前进的方向,同时解决了先前结果的差异,进一步表明将达西·布林克曼(Darcy-Brinkman)用于自由和多孔介质流程而不是Stokes-Darcy公式。此外,进行了一项参数研究,限制了反应速率常数的合理值,从而确定了其他验证机会。在第2部分中,进行网格收敛研究以估计Beta模型的离散误差。降低的代理几何形状和两个外推方案用于估计精确解决方案,然后通过网格收敛索引框架来估计模型的不确定性。对于模拟流量范围的$ \ sim 10 \%$,误差估计值是模拟的,比可用的实验性估计值大。结果表明,即使在更简单的情况下,也很难在燃料电池样模型中获得网格收敛性。因此,在验证期间或从数值模型中设计预测时提出谨慎。最后,鉴于数值数据和可用的实验数据中的不确定性,结果缺乏验证能力,突出了对数值数据的其他实验数据的需求并提高了精度。
Progress on the numerical modeling of a prototype fuel cell is reported. Some known limitations of the previously published Alpha model are addressed, and the numerical uncertainty due to discretization of the improved model, Beta, was estimated. In Part 1, the Beta model is compared to Alpha, where significant albeit small differences are seen. Shortcomings of the improved model are discussed, paving the way forward, while a discrepancy with previous results is addressed, further suggesting the use of the Darcy-Brinkman over Stokes-Darcy formulation for free and porous media flow. Furthermore, a parametric study is carried out, constraining plausible values of the reaction rate constants identifying additional opportunities for validation. In Part 2, a mesh convergence study is carried out to estimate the discretization error of Beta model. A reduced, proxy geometry and two extrapolation schemes are used to estimate the exact solution, which is then used to estimate the model's uncertainty through the Grid Convergence Index framework. Error estimates are on average $\sim 10\%$ for the flow rate range simulated, larger than experimental ones available. Results suggest a difficulty in achieving mesh convergence in fuel cell-like models, even in simpler cases. Caution is thus suggested during validation or when devising predictions from numerical models. Finally, given the uncertainties in the numerical data and the available experimental data, the results lack validation power, highlighting the need for additional experimental data and improved precision for the numerical data.