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

Paper Number: 129265

129265 - Multidisciplinary Design Optimization With Multiple Degrees of Freedom for an Axial Compressor Based on Data-Driven 

Compressor design involves many disciplines such as aerodynamics, strength, vibration, and life span. The traditional sequential iterative design pattern is time-consuming and ignores the coupling relationship among different disciplines. Multidisciplinary design optimization (MDO) enables a comprehensive consideration of the coupling effects among these disciplines, and reasonably balances the conflicts among the requirements of various disciplines, so as to improve the comprehensive performance of products and shorten the development cycle. In this paper, a data-driven compressor MDO platform is established based on a surrogate model-assisted differential evolution algorithm(pre-SADE). A prescreening strategy is introduced in the construction of the surrogate model to reduce true evaluation samples. This algorithm is combined with the directly manipulated free-form deformation (DFFD), which is a flexible and intuitive three-dimensional space parameterization method, to achieve multi-degree-of-freedom parameterized control. The research focuses on optimizing the aerodynamic and structural design of a 1.5-stage axial flow compressor in a gas turbine, with 58 design variables selected for optimizing the compressor blades and flow path. Maximizing efficiency and surge margin are set as the optimization objectives, while the constraint conditions involve flow rate, pressure ratio, strength, vibration, and high cycle fatigue life of the compressor. To address the challenge of determining surge boundaries during optimization, an approximate surge margin calculation method is proposed to reduce computational efforts without compromising accuracy. The results demonstrate that the peak efficiency, design point efficiency and surge margin are significantly improvedby 0.99%, 1.78% and 4.74% respectively after optimization, while the strength, vibration and high cycle fatigue life meet the requirements. The proposed MDO platform can not only guarantee the optimization effect, but also greatly reduce the design variables and evaluation time, which is effectively applicable to the multidisciplinary design optimization problem of compressors.

Presenting Author: Chuwei LUO Beihang University

Presenting Author Biography: Chuwei LUO, a doctoral candidate, is mainly engaged in unsteady aerodynamics, aeroelasticity and multidisciplinary design optimization of turbomachinery.

Authors:

Chuwei LUO Beihang University
Jiang Chen Beihang University
Yi Liu Beihang University
Hang Xiang Beihang University