Aerodynamic performance of supersonic partial admission impulse turbine by quantifying the uncertainties of turbulence model

Supersonic partial admission impulse (SPAI) turbines are critical for advanced aerospace power systems due to high thrust-to-weight ratio and compactness. However, their asymmetric supersonic admission poses serious challenges to accurate aerodynamic predictions. In numerical simulations, the Boussinesq eddy viscosity assumption in the Reynolds averaged Navier Stokes (RANS) model introduces uncertainties, which may lead to deviations in flow prediction and affect design reliability. To evaluate the aerodynamic prediction bias, this paper employs a non-uniform Reynolds stress tensor perturbation method to quantify the turbulence model uncertainty in a SPAI turbine using six specific perturbation cases. Firstly, deterministic simulations revealed key flow characteristics, including supersonic nozzle outflow, admission/non-admission region disparities, and loss mechanisms dominated by shock-induced separations and wake mixing. Then, uncertainty quantification revealed maximum deviations of 7.15 %, 4.26 %, and 2.34 % for expansion ratio, power output, and isentropic efficiency, respectively, across all perturbation cases, with the largest efficiency increase of 3.31 % in the 1C_min perturbation case. Flow sensitivity was highest in the rotor transition zones, admission regions, and nozzle wake regions, where perturbations altered shock-induced separation positions by up to 13 % of the chord length. Symmetric nozzle arrangements reduced lateral forces and overturning moments by over 99 % in each perturbation case but increased losses due to expanded wakes, demonstrating nozzle spacing dominates efficiency trends over turbulence uncertainty. This paper quantifies the uncertainty of the SST model and constructs an uncertainty prediction interval for the numerical simulations of the SPAI turbine, providing crucial support for high-fidelity turbine design.