Shape optimization of bellmouth to improve flow uniformity under non-aligned inflow condition in gas turbine casing

Abstract

A shape optimization process for a bellmouth in a gas turbine casing under high Reynolds number was conducted using computational fluid dynamics. The bellmouth connected to the nozzle guide vane was simplified assuming a cylindrically-shaped flow passage. The height of the casing entrance and the curvature of the bellmouth were selected as the geometric parameters. The curvature of the bellmouth was defined as an ellipse consisting of a major axis and a minor axis. Two hundred forty three cases were calculated with a combination of 5 parameters and 3 levels of change, and correlations for flow uniformity and pressure drop were derived. The optimized shape of the bellmouth was obtained by using the two correlations and the genetic algorithm. In addition, the optimized bellmouth was found using conditional U-Net, which predicted the flow uniformity image according to the given shape conditions. The shape of the baseline and optimized bellmouth were compared, and performances were compared in terms of flow uniformity and pressure drop. In addition, the results of the particle distribution, for particles generated after the combustion and entered the casing, were compared.

Keywords

• Bellmouth
• optimization
• uniformity
• pressure drop
• gas turbine casing