Historical-Based Machine Learning for Accelerated Design of Film Cooling Holes

Film cooling holes play a critical role in protecting turbine components from extreme thermal loads, thereby enabling higher engine operating temperatures and improved thermal efficiency. Due to complexity of the design space and the performance sensitivity to geometric and flow parameters, there is a growing need for fast and reliable evaluation tools that can guide early-stage design decisions. Traditional design approaches rely heavily on high-fidelity computational fluid dynamics (CFD) simulations to assess the performance of various hole geometries. However, these simulations can be computationally expensive and time-consuming. To address this challenge, the current work introduces a historical-based machine learning framework for predicting performance of specific hole geometries by leveraging simulation data across a wide range of film cooling configurations. Rather than relying on a purely parametric approach to incorporate design parameters, the proposed method integrates a database of high-fidelity CFD simulations using the Lattice Boltzmann Method (LBM) spanning a wide variation of shapes and operating conditions allowing to generalize more effectively across novel configurations. Such an approach enables the ML model to provide a near-instantaneous performance prediction directly from the hole geometry and the blowing ratio parameter. Designers can rapidly assess the thermal performance of hole configurations without the need for additional time-intensive CFD simulations. This enables faster design iteration along with an improved exploration of the design space early in the process, which supports the development of a more effective turbine cooling strategy.