STRUCTure-based URANS simulations of thermal mixing in T-junctions

Turbulent mixing of hot and cold streams in T-junction geometries is a critical safety issue for power plants, as temperature fluctuations have the potential to lead to high cycle thermal fatigue failures. Due to strong flow deformation in the mixing region, the time-scale separation assumption is not locally respected, and unsteady Reynolds-averaged Navier-Stokes (URANS) models fail to provide sufficient accuracy in predicting the temperature variations. Turbulence resolving methods such as large eddy simulation (LES) can in contrast provide reliable results, while being computationally overly expensive for industrial application. A STRUCTure-based second-generation URANS (2G-URANS) model was recently proposed at MIT, which aims at locally resolving unsteady flow structures; its applicability was demonstrated on a collection of classic flow tests. In the present work, the performance of the STRUCT model is assessed against low Reynolds number DNS data for a squared section T-junction, produced at the THTLAB (Fukushima et al., 2003). The mean and root mean square of the temperature and velocity predictions from the STRUCT solution are evaluated, confirming consistent predictions to the reference DNS results as well as the high-resolution LES simulations. The STRUCT turbulence model demonstrates LES-like accuracy on a URANS-like computational mesh. Spectral analysis of the velocity and temperature variation is also presented and provides consistent prediction of the frequencies of the large turbulent eddies.