Partially Averaged Navier Stokes Simulations of Turbulent Thermal Mixing in T-Junction

Abstract

Abstract High Cycle Thermal Fatigue is a major degradation mechanism associated with the nuclear power plant piping where hot and cold streams of fluids mix together with a phenomenon called turbulent thermal mixing. It induces significant thermal fluctuations on the piping wall called thermal stripping. T-junction has been known as a typical component susceptible to high cycle thermal fatigue. Accurate prediction of turbulent flow and thermal fields at the wall is necessary to predict the thermal fatigue degradation in such piping systems. Computational Fluid Dynamics (CFD) with advanced turbulent modelling is required to predict the flow and temperature fluctuations at the wall. Wall resolved LES is typically known to give good results for velocity and temperature fluctuations for such a flow phenomenon. However, LES is computationally expensive, and the cost of computation increases with the higher Reynolds number flows. Partially Averaged Navier Stokes (PANS) is a Hybrid turbulence Model. It is a bridging method from Reynolds Averaged Navier-Stokes (RANS) to Direct Numerical Simulation (DNS). PANS is usually controlled by two resolution control parameters, fk (unresolved to total kinetic energy) and fε (unresolved to total dissipation), fk = 0 being the DNS and resolving all scales of turbulence, and fk = 1 is RANS simulation. In this study PANS simulations for the Vattenfall T-junction Benchmark case are performed. fk is considered constant throughout the domain. Average and Root mean square temperature and Velocity profiles are compared with experimental data. It is concluded that PANS model can simulate turbulent thermal mixing phenomenon in a T-junction.