A Validation of CFD Methods on Predicting Valve Performance Parameters Available to Purchase

The influence of mesh resolution, the abilities of various eddy viscosity models, and near wall flow treatments on predicting the flow coefficients of poppet valves, operating in water are investigated in this paper. The computational fluid dynamics (CFD) models are solved using STAR-CCM+ 12.04. Grid-convergence is studied first, followed by quantitative assessments of the ability of standard k-ε model, realizable k-ε model, EB k-ε model, Lag EB k-ε model, V2F model and k-ω-sst model, and different wall treatments, such as high y⁺ wall treatment, two-layer wall treatment and all y⁺ wall treatment, embedded in the solver. The flow discharge coefficient (C_q) of poppet valves predicted by CFD models are compared to physical measurements. It was demonstrated in the study that grid resolutions normal to the wall and mesh quality are key factors. Advanced near wall flow treatments produce similar or worse predictions when using the standard k-ε model, and the effects of the near wall flow treatments are marginal for the realizable k-ε model. The ability of turbulence models varies greatly in predicting flow in different valves and lift levels. The realizable k-ε model is the optimal option for the considered valve flows giving an acceptable error within ±5%.