LES Studies of High Reynolds Number Wall Bounded and Far Field Stratified Wake Flows
Abstract
Numerical simulations will be conducted to study the turbulence around complex shapes and provide a tool for the US Navy to predict stratified wake flows. Multi-scale large-eddy simulation (LES) method is needed to simulate complex high Reynolds number flows since brute force direct simulation (DNS) cannot be used due to enormous resource requirements. Under past ONR funding we have developed a new multi-scale turbulence modeling strategy called two-level simulation (TLS) which has been coupled with conventional LES to form a hybrid approach for high Re flow simulations (called TLS-LES). Along with this development, a new incompressible fully conservative, energy conserving, high order code was developed and validated for both DNS and LES applications. This code, henceforth called Multi-scale INCompressible LES (MINCLES) is designed to work in complex flows and has been applied to many wall bounded and free shear flows as well as wake flows, and has been delivered to researchers in ARL (PSU) for their internal studies of wake flows of interest to US Navy. The proposed three-year effort will continue the development of the multi-scale TLS-LES approach for both wall bounded and far wake flows. In particular, we will focus on experimental studies of the very high Reynolds number flow over the SUBOFF spheroid configuration at Princeton University under ONR funding. For practical applications we will extend our modeling approach to use hybridized URANS-TLS-LES strategy thereby enabling automatic adjustment of the closure depending upon the local conditions (a similar strategy is already operational using the hybrid TLS-LES approach). The code will also be extended in a new task specifically designed to support the US Navy’s applied research by extending this code’s capability to simulate stratified wakes using LES and later on using the hybrid RANS-TLS-LES method. Transport for temperature, salinity and other passive scalars will also be implemented with attendant state relations for local density. Specifically, new closures for the stratified equations will be considered beginning initially with a localized dynamic eddy diffusivity type closure based on our past development effort. Validation against DNS and other data will be carried out as needed. This model will be revisited in the later stage of this research. Finally, since we are planning to simulate long time evolution of the stratified wake we will need to further optimize this code’s performance to achieve high efficiency in thousands of processors. So far we have shown that the code scales reasonably well for 2000+ processors and for 1 billion grid points but further performance optimization are needed to eliminate some of the existing bottlenecks in throughput and I/O. We plan to work with our internal research teams to develop a hybrid MPI/OpenMP paradigm to allow both strong and weak scaling to thousands of processors and to take advantage of next generation multi-core systems. Online data processing and extraction of key features of interest is needed when doing such large simulations and new paradigms in online data processing will be included as a part of this development effort. It is anticipated that this new code will be useful not only for fundamental studies of wall bounded and wake flows but also for large simulations of stratified wakes of interest to the US Navy.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 23, 2016
- Source ID
- N000141612577
Entities
People
- Suresh Menon
Organizations
- Georgia Tech Research Corporation
- Office of Naval Research
- United States Navy