Alignments and small scale statistics in the production region of grid turbulence

Direct Numerical Simulation (DNS) of turbulent flow generated by a single square grid is investigated using an unstructured finite volume method. The maximum value of the Taylor length-based Reynolds number throughout the computed flow field is about 40. The main focus of this study is on the production region which lies in the lee of the grid where turbulence builds up. Statistics of vorticity and of eigenvalues ($\lambda_{i}$, where i$=$1,2,3) and eigenvectors (e$_{i}$, where i$=$1,2,3) of the fluctuating strain rate tensor (S$_{ij})$ are analyzed. It is observed that the PDFs of all the eigenvalues in the production region are highly non-gaussian. The PDFs of the compressive ($\lambda_{3})$ and intermediate ($\lambda _{2})$ eigenvalues are strongly skewed to negative and positive values respectively. The energy spectrum of the streamwise fluctuating velocity has a well-defined power law with an exponent around -2 or -5/3 over more than one decade depending on the position in the production region. It is also observed that the most extensive eigenvector (e$_{1})$ and the intermediate eigenvector (e$_{2})$ align significantly with vorticity vector in the production region, which in turn increases average enstrophy production.