Session: 36-07 Multi-Disciplinary and Collaborative Optimization applications (1)

Paper Number: 122558

122558 - Multidisciplinary Analysis of the Impact of Blade Geometric Deviations on Compressor Performance Based on Fluid-Solid Grid Coordinated Deformation 

Advanced compressor blades have complex three-dimensional geometries, and it is common for geometric defects to occur during the manufacturing process, resulting in differences between the actual surface of the blade and the initial design. Existing research on constructing blade defect models is largely based on geometric parametrization (Wang et al 2022). While this allows for a relatively accurate analysis of the impact of geometric deviations, the need to frequently generate computational grids when dealing with large sample sets results in lower analysis efficiency.

Furthermore, the impact of manufacturing errors on compressor performance is diverse. While current research on the effects of blade manufacturing errors on compressor performance mainly focuses on aerodynamics (Wu et al 2022, Yan et al 2017), blade strength and vibrational characteristics are also influenced by these errors. Such errors may lead to a decline in the structural performance of the compressor. Therefore, a multidisciplinary analysis of the effects of blade geometric deviations on compressor performance is crucial.

To accurately assess the impact of this geometric uncertainty on the aerodynamic and structural performance of the blades, this study proposes a multidisciplinary analysis method for three-dimensional blade shape uncertainty using mesh deformation techniques. Considering the distinct characteristics of fluid and solid analysis grids, a technique is proposed that utilizes a set of blade geometric profiles to control the Coordinated deformation of both fluid and structural analysis grids. For the fluid grid, organized node sets are created based on the blade's circumferential, spanwise, and normal directions. This approach communicates the deviations of the control profiles in terms of node increments, effectively addressing the issue of staggered grids in boundary layer regions. For the solid grid, the Radial Basis Function (RBF) mesh deformation method is employed, which extracts nodes from the blade's surface and adjacent first-layer grid layer as the deformation domain, significantly reducing computational cost. Building upon this, The mathematical representation of geometric uncertainty is established by combining the spanwise averaging assumption and the non-stationary Gaussian process of Karhunen-Loève expansion. Random samples are generated using the Monte Carlo method, and ultimately, the BP neural network is employed to predict the impact of manufacturing errors on the aerodynamic and structural performance of a generic type of single-stage axial flow compressor blade.

The research results show that when considering the geometric deviation, the probability of the blade's maximum equivalent stress increasing, the pressure ratio, and the blade's maximum amplitude decreasing rises. The blade surface area and outlet mass flow rate are less influenced by the machining error, whereas the blade's maximum amplitude is notably more sensitive to these errors. As the thickness of the blade leading edge increases, the isentropic efficiency of the compressor shows a decreasing trend, and the corresponding geometric accuracy should be focused on. The research work in this paper provides a theoretical framework and practical reference for the optimal design of blades considering geometric uncertainty.

 

Keywords: grid deformation, geometric uncertainty, non-Gaussian stationary process, BP neural network, multidisciplinary analysis, axial flow compressor

Presenting Author: Zhaoxu Tong Dalian University of Technology

Presenting Author Biography: (a)Full Name:
Zhaoxu Tong
(b)Title/Position:
Ph.D. student at Dalian University of Technology
(c)Affiliation/Institution:
Dalian University of Technology
(d)Research Interests/Specializations:
Turbomachinery, Multi-disciplinary Analysis, Artificial Intelligence
(e)Paper:
TONG Zhaoxu, HAN Qixiang. Numerical simulation of combustion instability behind bluff body flame holder[J]. Aeroengine, 2023, 49 (1) : 81-88
(f)Contact Information:
Email: taa123@mail.dlut.edu.cn / tongtong19191@outlook.com

Authors:

Zhaoxu Tong Dalian University of Technology
Huiying Zhang Dalian University of Technology
Yan Zhou Dalian University of Technology
Rong Xie Dalian University of Technology
Shengli Xu Dalian University of Technology