Development of an automated process for turbine blade optimisation VII International Conference on Computational Methods for Coupled Problems in Science and Engineering COUPLED PROBLEMS 2017 M. Papadrakakis, E. O˜ nate and B. Schrefler (Eds) DEVELOPMENT OF AN AUTOMATED PROCESS FOR TURBINE BLADE OPTIMISATION TESSA UROI ´ C * , BORNA ˇ SOJAT * AND HRVOJE JASAK †,* * Faculty of Mechanical Engineering and Naval Architecture (FSB) University of Zagreb Ivana Luˇ ci´ ca 5, 10000 Zagreb, Croatia e-mail: tessa.uroic@fsb.hr, borna.sojat@stud.fsb.hr, hrvoje.jasak@fsb.hr web page: http://www.fsb.unizg.hr/cfd † Wikki Ltd. Unit 459, Southbank House, Black Prince Road, London SE1 7SJ, United Kingdom e-mail: h.jasak@wikki.co.uk, web page: http://wikki.gridcore.se/wikkiweb/company Key words: Turbomachinery, Optimisation, Genetic Algorithm, OpenFOAM Abstract. In this paper we present a fully automated procedure for turbine blade op- timisation. Optimisation process consists of geometry parametrisation using B-splines, mesh deformation using the dynamic mesh library in OpenFOAM, numerical simulation of transonic flow through the blade passage and finding the feasible solutions with the Multi-Objective Genetic Algorithm (MOGA). The process proved to be robust whether starting the optimisation from unfeasible geometry or a conventional blade profile. 1 INTRODUCTION The motivation for this work comes from industrial demands for faster and more ef- ficient design cycles. Turbomachinery components are regularly found in many energy conversion processes where the work load is predetermined and constant. Thus, it is pos- sible to optimise the components of a machine for a single operating point. The process of turbomachinery optimisation usually consists of several steps: geometry description in a mathematical sense (parametrisation), calculation of objective (fitness) functions, e.g. via computational fluid dynamics (CFD) simulation, and evaluation of the obtained solution(s). There are many optimisation approaches for different applications, from the method of trial-and-error to advanced adjoint optimisation algorithms [1]. The most common is the genetic or evolutionary algorithm which is based on Darwin’s theory of natural selection: only the fittest individuals survive and provide their genetic code to the following generation. There are many examples of turbomachinery optimisation using 859