Teaching Intelligent Fluid Dynamics through Engineering Cases: Design and Preliminary Practice of a Graduate Teaching Case Library

Effective integration of classroom instruction and research practice is essential to the quality of graduate engineering education. To address three persistent weaknesses in the teaching of fluid dynamics simulation courses, a graduate teaching case library, Intelligent Fluid Dynamics, was established. Specifically, theoretical instruction is delivered in isolation from research practice, systematic training in the application of acquired knowledge is absent, and assessment fails to reflect the stated course objectives. The reform takes the CFD simulation skills already acquired by students as its point of departure and targets research competence in intelligent fluid dynamics. Two complete teaching cases are under development: physics-informed neural network (PINN)-based flow field reconstruction and inversion, and deep reinforcement learning (DRL)-based active flow control of airfoils. Each case is supported by a knowledge-to-competence map and an instructor-oriented teaching guide. Learning is organized as a three-level progression, applied foundations, specialized depth and innovative practice, and is supported by purpose-built teaching procedures and an assessment scheme combining coursework with a final examination. Preliminary results indicate improved transfer of prior learning and enhanced specialized competence, and the experience is expected to be informative for comparable graduate courses.