Formation of conceptual design of axial turbines using machine learning methods

Read full text Summarise with VIRAT
Authors: G.М. Popov, S. A. Melnikov, E. S. Goryachkin, O.V. Baturin, Vasilii M. Zubanov, A. I. Shcherban
Cited by: 0
References: 11

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

The aim of the work was to find dependencies that make it possible to form sketches of the flow parts and disks of axial aircraft turbines at the initial design stages, when, according to the results of the design calculation, only the diameters at the inlet and outlet of the blade rings are known. For this purpose, a database was formed containing the results of measuring the geometric parameters of the flow parts and discs from drawings of meridian sections of the existing turbines. It was used to evaluate the accuracy of dependencies for calculating the sizes of flow parts and disks from literary sources, as well as for the formation of new dependencies using machine learning. The dependences found for calculating the dimensions of the flow part are in good agreement with the results of measuring the configuration of the stages of high- and medium-pressure turbines, but they have significant differences for low-pressure turbines. Using machine learning, new dependencies were formed to determine the size of the exact part of the turbines, which are in good agreement with the measurement results of both high- and medium-pressure and low-pressure turbine stages. This result was obtained by introducing a classifier parameter that separates different classes of turbine stages. In comparison with the known new dependencies, the coefficient of determination was increased from 0.58 to 0.76 for the working blades and from 0.58 to 0.71 for the nozzle devices. The search for dependencies for determining the shape of turbine disks showed that equations are known for determining only some of the disk sizes that have low accuracy. Using machine learning, a set of dependencies was obtained to determine all the necessary relative disk sizes. The use of these dependencies makes it possible to obtain a sketch of the turbine disk at the initial design stages, when only the shape of the flow part is known, to more accurately estimate the mass of the projected engine, and also to evaluate the strength of the turbine impeller disk.

Fields of Study & Subjects

Turbomachinery Performance and Optimization Aerospace Engineering and Applications Advanced Power Generation Technologies

Publication details

Journal / source
Вестник Самарского государственного аэрокосмического университета им. академика С.П. Королёва (национального исследовательского университета)
Volume
25
Issue
1
Pages
112–127
Publication date
26 March 2026
Language
EN