Deep learning modelling of manufacturing and build variations on multi-stage axial compressors aerodynamics

2023-10-06

AI Summary

This study addresses the challenge of inaccurate aerodynamic performance prediction in multistage axial compressors—arising from manufacturing/assembly tolerances and off-design operating conditions—hindering low-carbon design and process optimization of gas turbines. We propose a physics-informed deep learning framework that jointly integrates physics-constrained dimensionality reduction with deep neural networks, transforming unstructured CFD flow-field regression into a structured modeling task. The approach achieves CFD-level accuracy while enhancing model interpretability and enabling quantitative attribution of key aerodynamic influencing factors. With millisecond-scale real-time inference capability, the model has been deployed within industrial manufacturing workflows to support performance deviation root-cause analysis and closed-loop process optimization. This work establishes a novel paradigm for intelligent, low-carbon development of aero-engines and gas turbines.

Abstract

Applications of deep learning to physical simulations such as Computational Fluid Dynamics have recently experienced a surge in interest, and their viability has been demonstrated in different domains. However, due to the highly complex, turbulent and three-dimensional flows, they have not yet been proven usable for turbomachinery applications. Multi-stage axial compressors for gas turbine applications represent a remarkably challenging case, due to the high-dimensionality of the regression of the flow-field from geometrical and operational variables. This paper demonstrates the development and application of a deep learning framework for predictions of the flow field and aerodynamic performance of multi-stage axial compressors. A physics-based dimensionality reduction unlocks the potential for flow-field predictions, as it re-formulates the regression problem from an un-structured to a structured one, as well as reducing the number of degrees of freedom. Compared to traditional"black-box"surrogate models, it provides explainability to the predictions of overall performance by identifying the corresponding aerodynamic drivers. This is applied to model the effect of manufacturing and build variations, as the associated performance scatter is known to have a significant impact on $CO_2$ emissions, therefore posing a challenge of great industrial and environmental relevance. The proposed architecture is proven to achieve an accuracy comparable to that of the CFD benchmark, in real-time, for an industrially relevant application. The deployed model, is readily integrated within the manufacturing and build process of gas turbines, thus providing the opportunity to analytically assess the impact on performance with actionable and explainable data.

Problem

• Predict flow field in multistage axial compressors
• Address manufacturing variations impact on performance
• Reduce CO2 emissions through accurate performance predictions

Innovation

• Deep learning predicts axial compressor aerodynamics.
• Physics-based dimensionality reduction enhances flow-field predictions.
• Model integrates real-time performance analysis in manufacturing.