Zoran Pavlovic

Enhancement of the Performance of the Partial-Averaged Navier – Stokes Method by Using Scale-Adaptive Mesh Generation

The Partially-Averaged Navier-Stokes (PANS) approach is a recently proposed method which changes seamlessly from Reynolds-Averaged Navier-Stokes (RANS) to the direct numerical solution of the Navier-Stokes equations (DNS) as the unresolved-to-total ratios of kinetic energy and dissipation are varied. The parameter which determines the unresolved-to-total kinetic energy ratio f k is defined based on the grid spacing. The PANS asymptotic behavior goes smoothly from RANS to DNS with decreasing f k . In the work of Basara, Krajnovic and Girimaji (Proceedings of ERCOFTAC 7th International Symposium on Engineering Turbulence Modelling and Measurements ETMM7, 2/3, pp. 548–554, Lymassol, Cyprus, 2008), it was shown that a dynamic update of the PANS key parameter f k by changing at each point and at the end of every time step, is the promising approach to provide the optimum modeling on employed computational meshes. This work is extended here by introducing the adaptive local grid refinement which keeps in advance prescribed value of the parameter f k . The results show benefits of using such advanced numerical technique in conjunction with the PANS method.

Performance analysis of partially-averaged navier-stokes method for complex turbulent flows

The paper evaluates the performance of the Partially-Averaged Navier-Stokes (PANS) method on complex industrial Computational Fluid Dynamics (CFD) applications. The PANS method changes seamlessly from Reynolds-Averaged Navier-Stokes (RANS) to the Direct Numerical Solution (DNS) of the Navier-Stokes equations and supports any filter width or scale resolution and therefore, it has potential to be efficient tool to solve large and complex applications on available computational resources. The PANS variant, which is used in the present study, is based on the RANS k-e-ζ-f model (v2/k-f model). In the work presented here, the filter width is controlled by specifying one control parameter: unresolved-to-total ratios of kinetic energy fk. Furthermore, this parameter is defined based on the grid spacing. A dynamic update of fk at each computational point and at the end of every time step makes the computational procedure very simple and attractive for industrial CFD. The models performance is shown for complex CFD cases: an external car aerodynamics, a train aerodynamics and engine intake ports.

Statistical Analysis of Detailed 3-D CFD LES Simulations with Regard to CCV Modeling

Abstract The paper deals with statistical analysis of large amount of detailed 3-D CFD data in terms of cycle-to-cycle variations (CCVs). These data were obtained by means of LES calculations of many consecutive cycles. Due to non-linear nature of Navier-Stokes equation set, there is a relatively significant CCV. Hence, every cycle is slightly different – this leads to requirement to perform statistical analysis based on ensemble averaging procedure which enables better understanding of CCV in ICE including its quantification. The data obtained from the averaging procedure provides results on different space resolution levels. The procedure is applied locally, i.e., in every cell of the mesh. Hence there is detailed CCV information on local level – such information can be compared with RANS simulations. Next, volume/mass averaging provides information at specific locations – e.g., gap between electrodes of a spark plug. Finally, volume/mass averaging of the whole combustion chamber leads to global information which can be compared with experimental data or results of system simulation tools (which are based on 0-D/1-D approach).