Assessment of CFD for unheated gas-liquid flows with high void fraction

• Published in: Nuclear engineering and design 2019-01, Vol.341, p.346-359
• Main Authors: Chen, Q., Podila, K., Rao, Y.F., Bowden, R., Yang, S.K.
• Format: Article
• Language: English
• Subjects: Aerodynamics; Bend tests; Bends; Breakage; Bubble breakage/coalescence; CFD; Closures; Coalescence; Coalescing; Computational fluid dynamics; Computer simulation; Correlation analysis; Finite element method; Fluid dynamics; Fluid flow; Fractions; Gases; Interphase; Nuclear engineering; Nuclear safety; PBE; Pipes; Reactor safety; Reactors; Turbulence; Turbulent flow; Two phase flow; Void fraction; Water flow
• ISSN: 0029-5493; 1872-759X
• DOI: 10.1016/j.nucengdes.2018.11.016

[Display omitted] •High void faction in horizontal pipe with bends were simulated using CFD.•ASME CFD best practice guidelines were used for performing the analyses.•CFD predictions were compared against a new in-house air-water experiment. Dispersed gas-liquid two-phase flows with high void fraction are commonly encountered in the nuclear and process industries. Detailed understanding of the fluid flow behavior under these conditions is imperative to reactor safety. Traditional modelling approaches (e.g. system thermalhydraulic codes) often use geometry specific correlations that limit their applicability. Canadian Nuclear Laboratories is systematically developing leading-edge CFD capabilities for applications to various reactor flows, including two-phase flows. In the present study, the existing turbulent momentum closures and bubble breakage/coalescence models were assessed for modeling the turbulent air-water flow in a horizontal pipe with two 90° bends. CFD predictions were compared against data from a new in-house air-water experiment that adopted an advanced wire-mesh sensor for the measurement of the void fraction distributions. An initial precursor CFD analysis for a simple pipe geometry was performed to determine the suitability and relative importance of interphase closures at the high void fraction range (15% ≤ αavg ≤ 30%); then a selected single set of interphase closures and breakage/coalescence kernels were applied to simulate three horizontal bend tests. The recommended procedure by ASME to determine discretization error was demonstrated through the Grid Convergence Index method. Fair agreement with measurements was obtained at the straight pipe section. Simulations qualitatively captured the kidney-bean shaped void distribution at the bend, but the location of the void peak value was offset by a minimum of 90° compared to the CNL measurements.