The difficult challenge of a two-phase CFD modelling for all flow regimes

Abstract

System thermalhydraulic codes model all two-phase flow regimes but they are limited to a macroscopic description. Two-phase CFD tools predict two-phase flow with a much finer space resolution but the current modelling capabilities are limited to dispersed bubbly or droplet flow and separate-phase flow. Much less experience exists on more complex flow regimes which combine the existence of dispersed fields with the presence of large interfaces such as a free surface or a film surface. A list of possible reactor issues which might benefit from an “all flow regime CFD model” is given. The first difficulty is to identify the various types of local flow configuration. It is shown that a 4-field model has much better capabilities than a two-fluid approach to identify most complex regimes. Then the choice between time averaging, space averaging, or even ensemble averaging is discussed. It is shown that only the RANS-2-fluid and a space-filtered 4-field model may be reasonably envisaged. The latter has the capabilities to identify all types of interfaces and should be privileged if a good accuracy is expected or if time fluctuations in intermittent flow have to be predicted while the former may be used when a high accuracy is not necessary and if time fluctuations in intermittent flow are not of interest. Finally the closure issue is presented including wall transfers, interfacial transfers, mass transfers between dispersed and continuous fields, and turbulent transfers. An important effort is required to model all interactions between sub-filter phenomena and the transfers from the sub-filter domain to the simulated domain. The main difficulties are expected in modelling the transfer of sub-filter interfaces to predicted interfaces and the multiple effects of sub-filter deformations of large interfaces on inter-field transfers.