The Influence of Compressor Blade Row Interaction Modeling on Performance Estimates From Time-Accurate, Multistage, Navier–Stokes Simulations

Abstract

The time-accurate, multistage, Navier–Stokes, turbomachinery solver TURBO was used to calculate the aeroperformance of a 2 1∕2 stage, highly loaded, high-speed, axial compressor. The goals of the research project were to demonstrate completion times for multistage, time-accurate simulations that are consistent with inclusion in the design process and to assess the influence of differing approaches to modeling the effects of blade row interactions on aeroperformance estimates. Three different simulation setups were used to model blade row interactions: (1) single-passage per blade row with phase lag boundaries, (2) multiple passages per blade row with phase lag boundaries, and (3) a periodic sector (1∕2 annulus sector). The simulations used identical inlet and exit boundary conditions and identical meshes. To add more blade passages to the domain, the single-passage meshes were copied and rotated. This removed any issues of differing mesh topology or mesh density from the following results. The 1∕2 annulus simulation utilizing periodic boundary conditions required an order of magnitude fewer iterations to converge when all three simulations were converged to the same level as assessed by monitoring changes in overall adiabatic efficiency. When using phase lag boundary conditions, the necessity to converge the time history information requires more iterations to obtain the same convergence level. In addition to convergence differences, the three simulations gave different overall performance estimates where the 1∕2 annulus case was 1.0 point lower in adiabatic efficiency than the single-passage phase lag case. The interaction between blade rows in the same frame of reference sets up spatial variations of properties in the circumferential direction, which are stationary in that reference frame. The phase lag boundary condition formulation will not capture this effect because the blade rows are not moving relative to each other. Thus, for simulations of more than two blade rows and strong interactions, a periodic simulation is necessary to estimate the correct aeroperformance.