Session: 36-04 Neural-Network based approaches (3)

Paper Number: 128809

128809 - A Real Time AI-Based Strategy for the Design of a Low-Pressure Turbine Profile 

This paper focuses on the implementation of a data-driven, AI-powered design tool for Low Pressure Turbine profiles. This innovative tool is designed to exploit the power of a vast numerical database generated utilizing unsteady RANS (URANS) calculations.  The goal is to establish a rapid design tool for Low-Pressure Turbines (LPT) that can generate optimal profile geometry across a wide range of design parameters while ensuring accurate performance predictions.

The traditional workflow for LPT blade design starts with a preliminary design phase that utilizes simplified approaches (correlation-based) and limited computational power. This is followed by an optimization phase that is resource-intensive and iterative, utilizing CFD RANS calculations and metamodel-based techniques. The final stage involves refining the design to meet all constraints, a process that is repeated for each new design with little cumulative learning. The introduction of the new design tool aims to transform this approach, enabling the direct derivation of optimal designs from a comprehensive database, significantly reducing design time, and enhancing profile performance. This transition to a data-centric, single-shot CFD-driven approach is expected to significantly impact on the first phases of the industrial design procedure for turbines.

The paper presents the automatic workflow developed to create the large database of URANS calculations, to feed the new real-time framework for the profile design. The AI-based tool will be then presented and discussed, showing how in addition to suggesting optimal geometries, it offers additional outputs to inform the designer. Indeed, the tool can provide optimal geometries that match mechanical and geometrical constraints providing performance in both design and off-design conditions. Moreover, it can be used to create performance maps as a function of typical aero design parameters like the Zweifel, Diffusion rate, Reynolds number, etc. As a further result, the blade-loading and the entire 2D flow field are predicted in real-time for the optimal geometry at a given operating condition.

The integration of this tool into the design workflow represents a major step forward in achieving more efficient, reliable, and innovative turbine designs, ultimately contributing to the development of greener, quieter next-generation aeronautical engines.

Presenting Author: Juri Bellucci Morfo Design Srl

Presenting Author Biography: Juri Bellucci studied at the University of Florence, where he graduated in Mechanical and Energy
Engineering (master degree, 2010). He undertook his postgraduate studies in the group of
Prof. Andrea Arnone at the University of Florence, Department of Industrial Engineering, where he got
his PhD in December 2013. Starting from 2020 Juri has worked in Morfo Design.
His research is mainly concerned with the use of computational fluid dynamics and meta-models
for aerodynamic design and optimization of axial and centrifugal turbomachines, in cooperation
with industries, focusing his attention on the development of tools to improve and support
the optimization process.

Authors:

Juri Bellucci Morfo Design Srl
Angelo Alberto Granata Morfo Design Srl
Mattia Silei Università di Firenze
Matteo Giovannini Morfo Design Srl
Ennio Spano Avio Aero
Andrea Notaristefano Avio Aero
Davide Lengani University of Genova