LES Analysis of Temperature Fluctuations at T-Junctions for Prediction of Thermal Loading

Abstract

T-junctions are widely used for fluid mixing in nuclear power and chemical and refinery plants. Temperature fluctuations generated by the mixing of hot and cold fluids at a T-junction can cause high cycle thermal fatigue (HCTF) failure. Japan Society of Mechanical Engineers (JSME) published ‘Guideline for Evaluation of High Cycle Thermal Fatigue of a Pipe (JSME S017, 2003)’ which results in a very conservative evaluation. CFD/FEM coupling analysis is considered as a useful tool for the more rational evaluation of HCTF. The present paper aims at the validation of CFD simulations to establish a more rational method of evaluating thermal loading, prior to performing CFD/FEM coupling analysis. It is very important to choose the proper turbulence model for the analysis of unsteady phenomena such as the highly fluctuating flow and temperature fields at a T-junction. Here, large eddy simulation (LES) turbulence models suitable for the simulation of the unsteady phenomena were investigated. LES sub-grid scale (SGS) models used include standard Smagorinsky model (SM) and dynamic Smagorinsky model (DSM). The effects of numerical schemes for the calculation of the convective term in the energy equation on the simulation results were also investigated. LES analyses of the flow and temperature fields at a T-junction were carried out using the above SGS turbulence models. For the sake of comparison, the simulation conditions are the same as those of the WATLON experiments conducted at Japan Atomic Energy Agency (JAEA) in the literature. All of the simulation results show the flow pattern of the wall jet with the strong flow and temperature fluctuations, which is the same as that observed in the experiment. The simulation results indicate the numerical schemes have great effect on the temperature distribution and the temperature fluctuation intensity (TFI). The 1st-order upwind differencing (1UD) significantly underestimates the TFI for each LES model, although it exhibits a good numerical stability. On the other hand, the hybrid scheme, which is mainly the 2nd-order central differencing (2CD) blended with a small fraction of 1UD, can better predict the TFI for each LES model. Furthermore, the DSM model gives a prediction closer to the experimental results than the SM model while using the same numerical scheme. In this study, an important finding is that a combination of the DSM model and the hybrid scheme with a large blending factor can provide a prediction agreeing very well with the experimental results.