Dynamic probabilistic design for blade deformation with SVM-ERSM

Abstract

Purpose

This paper aims to reasonably quantify the radial deformation of turbine blade from a probabilistic design perspective. A probabilistic design for turbine blade radial deformation considering non-linear dynamic influences can quantify risk and thus control blade tip clearance to further develop the high performance and high reliability of aeroengine. Moreover, the need for a cost-effective design has resulted in the development of probabilistic design method with high computational efficiency and accuracy to quantify the effects of these uncertainties.

Design/methodology/approach

An extremum response surface method-based support vector machine (SVM-ERSM) was proposed based on SVM of regression to improve the computational efficiency and precision of blade radial deformation dynamic probabilistic design regarding non-linear material properties and dynamically thermal and mechanical loads.

Findings

Through the example calculation and comparison of methods, the results show that the blade radial deformation reaches at the maximum at t = 180 s; the probabilistic distribution and inverse probabilistic features of output parameters and the major factors (rotor speed and gas temperature) are gained; besides, the SVM-ERSM holds high computational efficiency and precision in the non-linear dynamic probabilistic design of aeroengine typical components.

Practical implications

The present efforts provide a method to design turbine besides other aeroengine components considering dynamic and non-linear factors base on probabilistic design for further research.

Social implications

Moreover, the present study provides a way to design dynamic (motion) structures from a probabilistic perspective.

Originality/value

It is proved that the dynamic probabilistic design-based SVM-ERSM could produce a more reasonable blade radial deformation while maintaining low failure probability, as well as offer a useful reference for blade-tip clearance control and a promising insight to the optimal design of aeroengine typical components.

Citation

Fei, C., Tang, W., Bai, G. and Ma, S. (2015), "Dynamic probabilistic design for blade deformation with SVM-ERSM", Aircraft Engineering and Aerospace Technology, Vol. 87 No. 4, pp. 312-321. https://doi.org/10.1108/AEAT-07-2013-0125