This article presents an approach based on numerical reduced-order modeling to analyze the thermal behavior of electric traction motors. In this article, a single conjugate heat transfer analysis provides the possibility to accurately predict thermal performances by incorporating both computational fluid dynamic and heat transfer modules. Then, the developed model is used as the basis for deriving a fast reduced-order model of the traction motor enabling prediction of motor thermal behavior in duty cycles with a high number of operating points. All the results achieved are verified using flow and temperature measurements carried out on a traction motor designed and built for a traction application. A good agreement between the measured and estimated values of flows and temperatures is achieved while keeping the computation time within a reasonable range for both the full-order and reduced-order conjugate heat transfer models. The optimized full-order model can be run in minutes and the reduced-order model computation time is less than one second per operating point. The transient simulation based on the reduced-order model is conducted and both the learning phase and validation results are well illustrated. It is shown than the deviation of the reduced-order model in estimating the motor thermal performance is less than one celsius degree from the full-order model.

中文翻译：

---

基于数值的牵引电机降阶热建模

本文提出了一种基于数值降阶建模的方法来分析牵引电机的热行为。在本文中，通过结合计算流体动力学和传热模块，单个共轭传热分析提供了准确预测热性能的可能性。然后，开发的模型用作推导牵引电机的快速降阶模型的基础，从而能够预测具有大量工作点的占空比中的电机热行为。使用在为牵引应用设计和制造的牵引电机上进行的流量和温度测量来验证所获得的所有结果。对于全阶和降阶共轭传热模型，在将计算时间保持在合理范围内的同时，实现了流量和温度的测量值和估计值之间的良好一致性。优化后的全阶模型可以在几分钟内运行，而降阶模型的计算时间每个操作点不到一秒。进行了基于降阶模型的瞬态仿真，并很好地说明了学习阶段和验证结果。结果表明，在估计电机热性能时，降阶模型与全阶模型的偏差小于 1 摄氏度。优化后的全阶模型可以在几分钟内运行，而降阶模型的计算时间每个操作点不到一秒。进行了基于降阶模型的瞬态仿真，并很好地说明了学习阶段和验证结果。结果表明，在估计电机热性能时，降阶模型与全阶模型的偏差小于 1 摄氏度。优化后的全阶模型可以在几分钟内运行，而降阶模型的计算时间每个操作点不到一秒。进行了基于降阶模型的瞬态仿真，并很好地说明了学习阶段和验证结果。结果表明，在估计电机热性能时，降阶模型与全阶模型的偏差小于 1 摄氏度。