粘弾性流体バックステップ乱流におけるDNSと応力方程式モデル

抄録

It is well known that a small amount of polymer additive mixed into Newtonian fluid can reduce the frictional drag of wall turbulence. Such a fluid with the drag reduction effect is a viscoelastic fluid, and the prediction of viscoelastic turbulent flow by RANS (Reynolds-Averaged Navier-Stokes) simulation is required for engineering applications. In RANS to predict the viscoelastic turbulent flow, three unknown terms are involved into the ensembled-averaged constitutive equation. In the flat channel flow, Λ_ij is dominant compared to the other terms and they except for Λ_ij are neglected. However, the comparison of the three unknown terms with different analytical configurations was not performed. Therefore, we performed DNS of viscoelastic turbulent flows over a backward-facing step and investigated the influence of the three unknown terms. In the result, the turbulent diffusion term Ξ_ij is comparable to Λ_ij near the separated shear layer due to the separation. So, a model of Ξ_ij should be constructed considering the result that has a noticeable value in separated shear layer for improving the prediction accuracy of RANS.