M.Sreedhar and F.Stern
University of Iowa, USA
The authors should be congratulated for such an interesting paper. The dynamic two-parameter model (DTM) seems to have overcome many of the drawbacks of the original Smagorinsky (DSM) subgrid scale model. We have the following questions and comments.
Usually, in the quasi-two dimensional free surface turbulent field, the energy of the surface-parallel velocity components tends to increase while the surface normal velocity component decays. The one dimensional energy spectrum (Fig.7 in the paper) which shows the energy contents of the surface parallel velocity components obtained with the DTM model at different horizontal components planes does not seem to indicate this feature. Is it due to the decaying nature of the turbulent field? Any comments on this?
The figures and discussion on the time evolution of plane-averaged forward and back scatter are very interesting. The superiority of the DTM model is very clearly demonstrated. Inclusion of the results of forward and back scatter computed from the DNS data (after proper filtering) in a few of the figures would be very welcome.
This behavior is actually due to the decaying nature of the flow. Indeed, since the mechanisms generating turbulence are eliminated, the normal velocity component is generally lower than in the turbulent open channel and hence, the damping of this component at the free surface is less important. As a consequence, the increase in the surface normal components at the free surface becomes negligible; this can be seen, for example, from the r.m.s. of these components in Fig.2 that tend to become straight along the channel depth.
We agree with this excellent suggestion, and we have been developing more quantitative comparisons for this flow. In Salvetti and Banerjee (Phys. Fluids, vol. 7, n. 11, Nov 1995) quantitative comparisons were made between DNS, DMM, and DTM behaviors.
University of California, San Diego, USA
In this paper the decaying turbulence is considered in an open horizontally periodic flow with free-slip top and bottom conditions. Large eddy simulation (LES) is described for dynamic models with Smagorinsky term and with a combination of Smagorinsky and Bardina terms. A quantitative comparison is made between results, obtained with these models. Some qualitative comparison of LES results with direct numerical simulation (DNS) is also mentioned. The value of this work will be greatly increased if the comparison between LES models and DNS will be made quantitative.
As we noted in response to the second point by Sreedhar and Stern, we are continuing work on this flow and will add more quantitative comparisons in future publications. The quality of the DMM and DTM SGS models has, however, been directly compared to DNS results via a priori tests in Salvetti and Banerjee (Phys. Fluids, vol. 7, n. 11, Nov. 1995) as we note above.