RANS predictions of trailing-edge slot flows using heat-flux closures developed with CFD-driven machine learning

Abstract

Accurate prediction of the wall temperature downstream of the trailing-edge slot is crucial to designing turbine blades that can withstand the harsh aerothermal environment in a modern gas turbine. Because of their computational efficiency, industry relies on low-fidelity tools like RANS for momentum and thermal field calculations, despite their known underprediction of wall temperature. In this paper, a novel framework using a branch of machine learning, gene-expression programming (GEP) [Zhao et al. 2020, J. Comp. Physics, 411:109413] is used to develop closures for the turbulent heat-flux to improve upon this underprediction. In the original use of GEP (“frozen” approach), the turbulent heat-flux from ahigh-fidelity database was used to evaluate the fitness of the candidate closures during the symbolic regression, however, the resulting closure had no information of the temperature field during the optimisation process. In this work, the regression process of the GEP instead incorpo-rates RANS calculations to evaluate the fitness of the candidate closures. This allows the inclusion of the temperature field from RANS to advance the iterative regression, leading to a more integrated heat-flux closure development, and consequently more accurate and robust models. The GEP-based CFD-driven framework is demonstrated on a trailing edge slot configuration with three blowing ratios. Full a posteriori predictions from the new closures are compared to high-fidelity reference data and both conventional RANS closures and closures obtained from the “frozen” approach.