Title

Session

Authors

Abstract

Running a numerical CFD simulation on a low-speed fan is found to be highly challenging as the flow is of low-Mach and generally consists of separations and complex flow structures. The robust, cost-efficient, and accurate pressure-based solver proposed by Cadence Design Systems, allows one to have a good understanding of low-speed fan performances and characteristics which is of critical importance in most working environments.
Typical density-based CFD solvers, solving the Navier-Stokes equations for continuity, momentum, and energy, are highly efficient to solve high-speed flows, but generally show a slow convergence speed for low-speed applications without dedicated preconditioning techniques. The simulation of low-speed flow fans can be accelerated through a pressure-based solver which directly solves for the pressure and momentum. The integration of the coupled pressure-based solver into Cadence's Fidelity environment, including best practices and interfacing, is presented. The solver is built on a fully coupled implicit all-Mach methodology.
The coupled pressure-based solver is used to compute the performances of a squirrel cage-type fan: commonly utilized for heating, ventilation, and air-conditioning applications. The fan consists of a centrifugal impeller and a volute: the impeller is typically a low-speed forward-curved centrifugal wheel, the volute tends to be a simple scroll with a rectangular cross-section. The coupled pressure-based flow solver is used to simulate and analyze the complex flow field, more specifically the strong impeller-volute interaction near the volute tongue and the flow separation due to rapid flow turning. The impact of different turbulence models is examined and discussed. The time-dependent nature of the flow is also verified through an unsteady simulation of the flow field. A variable time step is considered throughout the simulation to decrease computational time and save engineering resources. The results are compared to the ones of steady RANS computations.