CFD analysis of thermal dispersion in a structured pebble bed

Abstract

Heat transfer mechanisms in packed beds are not entirely understood, which gave rise to thermal dispersion correlations that contain considerable amounts of empiricism. The empiricism resulted from incorporating a limited number of heat transfer mechanisms, along with characterising porous structures with insufficient complexity. Therefore, these correlations have limited applicability and occasionally, uncertain validity. The High Pressure Test Unit (HPTU) was subsequently constructed to conduct a comprehensive set of separate effects tests for the purpose of validating existing thermohydraulic correlations. These tests included the Braiding Effects Test Section (BETS) experiments which investigated the increased thermal dispersion in packed beds that results from the effects of the porous structure on the flow. The BETS experimental temperature data required validation. Computational fluid dynamics (CFD) has been proven to accurately predict the thermal hydraulic phenomena associated with fluid flow in packed beds and can consequently also be used to analyse thermal dispersion in a packed bed (Cheng et al., 1999; Wen and Ding, 2006; Hassan, 2008; Rousseau and Van Staden, 2008; Van Antwerpen, 2009; Kgame, 2010; Preller, 2011; Van der Merwe, 2014). This study subsequently presents an explicit numerical analysis of thermal dispersion in a structured pebble bed using the CFD package STAR-CCM+®. This work is an extension of the preliminary work conducted by Preller (2011) wherein a single case of the BETS experiments was simulated, which corresponded to a Reynolds number of 3000. Furthermore, symmetry boundary conditions were imposed on the walls to reduce the cross-sectional area to 25% of its original size since this particular BETS packing configuration consisted of 3898 mono-sized spheres. The initial simulation was subsequently verified. Nevertheless, in addition to instabilities, it was observed that the symmetry planes affect the simulated temperature profiles to a certain degree. The full cross-sectional area was consequently simulated, which is also preferred in the analysis of radial thermal dispersion. The resulting meshes consisted of nearly 40 million volumetric cells, followed by several simulation challenges along with highly oscillatory flows and large temperature gradients. A suitable and universal simulation methodology was therefore developed based on a Reynolds number of 3000 by conducting comprehensive sets of analyses on, among others, mesh quality, mesh independency, turbulence models, temperature extraction methods, wall treatment, along with stability and convergence criteria. It was determined that the use of large eddy simulation (LES) along with mesh refinement in the packed region of the bed remained the most suitable approach. The resulting simulation methodology has been verified to accurately predict the radial temperature profile in a packed bed and displayed a considerable improvement over initial approaches in the agreement with experimental temperature data. The methodology was subsequently used to validate an extended range of BETS experiments with Reynolds numbers of up to 40000. In all cases, good agreement between the numerically simulated and experimentally measured temperatures has been achieved, which simultaneously verified the methodology for the entire set of Reynolds numbers and validated the temperature data from the selected BETS experiments. It can therefore be concluded that CFD can be used to provide an accurate analysis of thermal dispersion in a structured pebble bed.