(655a) Computational Study of Turbulent Single Phase and Multiphase Flows in 90°Bends

Flows in curved pipes are encountered in numerous industrial and biological systems and exhibit complex features due to the presence of centrifugal and Coriolis forces that induce secondary flow patterns called the Dean vortices. The deposition of particles on a pipe wall is particularly sensitive to these complex flow patterns and studied in this work with the help of computational fluid dynamics simulations. While CFD software offer an ideal tool for these studies, the impact of various options available on the accuracy of the computed results is sparsely discussed in the literature. An effort is thus made to discuss available modeling approaches on the accuracy of the results. In a first section, the Reynolds Stress Model (RSM) with astandard wall function for the near-wall treatment is used to simulate turbulent flows passing through 90 degree bends (L-bends). Primary and secondary flow patterns are studied qualitatively for a range of bend curvature ratios. Numerical calculations of pressure drop are compared to the experimental work of Sudo et al. (1998). An analysis of the computed Reynolds stress is also conducted to assess the quality of the results by verifying that the eigenvalues are non-negative. In the second part of this work, different wall function models are used in modeling the continuous air phase while a Lagrangian particle tracking algorithm is employed for the discrete droplet phase. The grade efficiencies of the L-bend are compared against the experimental works of Pui and Liu (1987). Results show that the RSM is able to accurately capture the pressure drop at the outer and inter parts of the bends but fails to capture the pressure drop on the sidewall. However, using the RSM in combination with enhanced wall functions can improve significantly accuracy of particle deposition predictions on the L-bend pipe wall.