Nonlinear uncertainty quantification of the impact of geometric variability on compressor performance using an adjoint method

• DOI: 10.1016/j.cja.2021.06.007
• Publication Date: 2021-07-07T18:29:19Z
• Authors: Qian ZHANG; Shenren XU; Xianjun YU; Jiaxin LIU; Dingxi WANG; Xiuquan HUANG
• Keywords: 0103 physical sciences; 01 natural sciences

Manufactured blades are inevitably different from their design intent, which leads to a deviation of the performance intended value. To quantify associated uncertainty, many approaches have been developed. The traditional Monte Carlo method based on Computational Fluid Dynamics solver (MC-CFD) for three-dimensional compressor is prohibitively expensive. Existing alternatives MC-CFD, such as surrogate models and second-order derivatives adjoint method, can greatly reduce computational cost. Nevertheless, they will encounter 'the curse dimensionality' except linear model gradient (called MC-adj-linear). However, MC-adj-linear neglects nonlinearity function. In this work, an improved proposed circumvent low-accuracy problem without incurring high cost other alternative models. applied study aerodynamic annular transonic cascade, subject prescribed geometric variability with industrial relevance. It found that achieves significant accuracy improvement over low cost, showing great potential fast uncertainty quantification.