Maximum-Projection-Based Bayesian Optimization Utilizing Sensitivity Analysis for High-Efficiency Radial Turbine Design with Scarce Data

2026-03-18

AI Summary

This work proposes an efficient methodology for optimizing radial turbine efficiency under a stringent high-fidelity CFD simulation budget of only 330 function evaluations. By integrating maximum projection initial design, polynomial chaos expansion–based global variance-based sensitivity analysis, and Bayesian optimization, the approach first identifies a critical low-dimensional subspace of parameters that most significantly influence performance. Subsequently, a Gaussian process surrogate model combined with an upper confidence bound acquisition strategy focuses computational resources on this informative subspace. The method effectively achieves dimensionality reduction and precise search within a highly anisotropic parameter space, successfully increasing turbine efficiency from 85.77% to 91.77%. This demonstrates the feasibility and advantage of high-dimensional engineering optimization under extremely limited simulation budgets.

Technology Category

• Search and Optimization: Sampling/Simulation-based Search
• Machine Learning: Optimization
• Reasoning under Uncertainty: Stochastic Optimization

Application Category

• Search and Retrieval-Augmented AI: Efficiency and scalability of Web search engines
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

We propose a data-efficient workflow to optimize the efficiency of a radial turbine design under a strict budget of high-fidelity computational fluid dynamics simulations. Assuming anisotropic parameter impact, we use a maximum-projection initial experimental design to ensure space-filling and strong projection properties on low-dimensional subspaces. Bayesian optimization is performed using Gaussian process surrogates with an upper confidence bound acquisition function. In parallel, polynomial chaos expansions provide variance-based global sensitivity analysis metrics, which allow to identify a reduced subspace with the most influential parameters, wherein the optimization is continued. Turbine efficiency is increased from 85.77% initially to 91.77% at the end of the workflow, with a total budget of 330 simulations.

Problem

• radial turbine design
• scarce data
• efficiency optimization
• computational fluid dynamics
• Bayesian optimization

Innovation

• Maximum-Projection Design
• Bayesian Optimization
• Global Sensitivity Analysis
• Polynomial Chaos Expansion
• Data-Efficient Optimization