This paper updates a method for generating small, accurate kinetic models for applications in computational fluid dynamics programs. This particular method first uses a time-integrated flux-based algorithm to generate the smallest possible skeletal model based on the detailed kinetic model. Then, it uses a multi-stage optimization process in which multiple runs of a genetic algorithm are used to optimize the rate constant parameters of the retained reactions. This optimization technique provides the user with the flexibility needed to balance the fidelity of the model with their time constraints. The updated method was applied to the reduction of a methane-air model under conditions meant to approximate the end of a compression stroke of an internal combustion engine. When compared to previous techniques, the results showed that this method could produce a more accurate model in considerably less time. The best model obtained in this study resulted in relative errors ranging from 0.22 to 1.14% on all six optimization targets. This reduced model was also able to adequately predict optimization targets for certain operating conditions, which were not included in the optimization process.

本文更新了一种生成小型，精确动力学模型的方法，以用于计算流体动力学程序。该特定方法首先使用基于时间积分通量的算法，根据详细的动力学模型生成最小的骨骼模型。然后，它使用一个多阶段优化过程，其中使用遗传算法的多次运行来优化保留反应的速率常数参数。这种优化技术为用户提供了平衡模型的保真度和时间约束所需的灵活性。在意在近似于内燃机压缩冲程结束的条件下，将更新的方法应用于甲烷-空气模型的还原。与以前的技术相比，结果表明，该方法可以在相当短的时间内生成更准确的模型。在这项研究中获得的最佳模型导致所有六个优化目标的相对误差范围为0.22至1.14％。这种简化的模型还能够针对某些操作条件充分预测优化目标，而优化过程中并未包括这些目标。