This study develops an advanced framework for analyzing aerothermal performance uncertainties in high-pressure turbine guide vanes under combustor-exit non-uniform conditions. To address computational challenges in high-dimensional flow field analysis, a reduced-order methodology integrating proper orthogonal decomposition with sparse polynomial chaos expansion is established, enabling efficient uncertainty quantification and sensitivity analysis from global performance to local flow structures. By enhancing parametric characterization methods for combustor-outlet fields—particularly through improved temperature field generation (combining radial temperature distribution function and hot spots) and pressure–velocity coupling algorithms—the proposed approach better captures real-engine thermal–flow features. Applied to GE Energy Efficient Engine vane with endwall cooling, the results identify swirl strength as the most critical parameter. Key findings include the following: (1) swirl-induced flow deflection causes hot gas accumulation on the pressure side, reducing cooling effectiveness at high swirl strength. The pressure-side trailing edge shows peak sensitivity (σ=0.035); (2) total pressure losses exhibit positive skewness (70% samples exceeding baseline) due to nonlinear viscous dissipation from vortex core growth; and (3) compared with traditional methods, this analysis framework significantly reduces the computational cost of flow-field uncertainty quantification and sensitivity analysis, thereby providing a practical tool for robust turbine design under realistic inflow uncertainties.
Zhang et al. (Sun,) studied this question.