The Large Eddy Simulation (LES) of two-phase flows with resolved scale interfaces is investigated through the a priori filtering of Direct Numerical Simulations (DNS) of one-fluid and multifield models. A phase inversion benchmark [1][2][3][4] is considered highlighting many coalescence and interface rupture events in a kind of atomization process. The order of magnitude of specific two-phase subgrid LES terms is first considered with the two modeling approaches. Then, different existing models such as Smagorinsky [5], Wall-Adapting Local Eddyviscosity (WALE) model [6], Bardina [7], Mixed [8] and Approximate Deconvolution Model (ADM) [9] are used to account for two-phase subgrid effects. These models are compared to filtered DNS results. The main conclusion concerning a priori LES filtering is that the inertia term is not predominant in two-phase flows with fragmentation and rupture of interface. This conclusion is different from that of the studies of [3,[10][11][12][13]. concerning LES models, functional modeling do 1 not correlate to filtered DNS results whereas structural approaches do. Bardina and ADM are clearly the good LES framework to consider for two-phase flows with resolved scale interfaces. ADM is clearly better than Bardina in our study. The authors greatly acknowledge the reviewer for its comments and remarks that have guided us in improving the technical and editorial aspects of the manuscript. From a general point of view, all the comments and remarks recommended by the reviewer have been taken into account. They are highlighted in the text in red. More specifically, we can detail them as follows:
COMPUTERS & FLUIDS
1) I would recommend to add the main conclusions to the abstract and to add a separate section conclusions to the paper.In agreement with reviewer remark and in order to improve the fast reading and understanding of the paper, we have included a conclusion part in the abstract. In addition, the last section have been extended to "Summary, conclusion and future work", with addition of conclusions in order to highlight the originality and main results of the work.
2) Abstract : DNS and ADM should be defined.Done. The same has been achieved for the Wall-Adapting Local Eddy-viscosity (WALE) model.
3) A streak of the interface points towards the upper left corner of the box. Is that physical?Yes, this a physical feature of the interface. As we are dealing with liquid-liquid twophase flows, when the light fluid (blue surface) goes to the top part of the cavity under gravity effects, heavy liquid can be trapped near vertical walls or vertical corners, inducing this king of interface topology.4) The sentence "Among the few existing …" seems to have a grammar mistake.
Detailed Response to ReviewersThis sentence has been rewritten as follows "Among the few existing references, it is worth mentioning the work on LES formulation for the one-fluid model and subsequent a priori LES filtering of DNS simulations for the estimate of new specific two-phase contributions in the mass, momentu...