Hello, A new project on the LCRC cluster has been requested. Please forward the information on to the LCRC Allocation sub-committee. Applicant's name: Rui Hu Applicant's institution: ANL Applicant's division: NE Project Name: TMix Project title: Translating Petascale Simulation into Engineering Design Tools: Application to Mixing Associated funding: Argonne-LDRD Other Systems: Science: Mixing is a phenomenon that is central to a broad range of DOE science and engineering problems, such as convective heat transfer in energy systems, chemical deposition and materials processing, dispersion of contaminants in the ocean and in urban canyons. The extreme nonlinearity of the systems coupled with minimum dissipation leads to a broad range of scales that make solution of the governing equations of flow mixing intractable by analytic means and extremely demanding from a computational standpoint. Although the Department of Energy's leadership computing facilities provide simulation capabilities that allow scientists to explore a significant range of scales with unprecedented details, it is too expensive to be used for parameter-space exploration or design analysis. This project will develop a hierarchical methodology to model turbulent mixing in any given wall-bounded flow system. Using DOE’s leadership computing facilities, high fidelity simulations based on direct numerical simulation or large eddy simulation will be performed to probe the micro-details of the turbulent mixing phenomenon. The fine-scale simulation results will be used to develop a model to accurately estimate mixing through lower fidelity Reynolds-Averaged Navier-Stokes calculations on a small cluster such as Fusion. Such an approach would allow exploring a wider range of parametric design options for a given flow system. Through the wide exploration of the parameter space it will be possible to construct a general model for mixing to be used in reduced order methods for engineering design analysis. This approach provides a critical bridge among the micro-, meso-, and macro- scale modeling and simulation methods. Thus, the high-fidelity simulations on DOE exsa- or p eta- scale computing facilities can be integrated into quotidian design process. Project description: High fidelity calculations using direct numerical and large eddy simulations (DNS/LES) based on the ANL code Nek5000 will be performed in prototype geometries and a full rod-bundle in a reactor core and other cases. The probability density functions for each velocity component will be computed locally. From this data, one can evaluate the mixing parameter through Reynolds stresses in RANS based simulations with the commercial CFD code, STAR-CCM+. While the high fidelity calculation of the full rod-bundle simulation has to be executed in Blue Gene/P, the calculation of prototypical geometry can be run in Fusion. Once the pipeline from DNS/LES to RANS is established and validated, RANS will be used to examine more complex problems in their relevant parameter space: the geometric parameters, the Reynolds number, the Peclet Number, etc., which require intense use of computation time in a small clusters. Access to the Fusion cluster will allow us to explore a wide range of parameter space and to develop models/correlations to be used in engineering design tools. Nek5000 is a mature research DNS/LES computational fluid dynamics code using spectral element method. STAR-CCM+ is a commercial CFD tool using the finite volume formulation of the RANS method for thermal fluid dynamics simulation. Both Nek5000 and STAR-CCM+ has already been installed and used in Fusion. Nek5000 employs the MPI standard for parallelism. In 1999 the code was recognized with the Gordon Bell prize for algorithmic quality and sustained parallel performance. It has been scaled on Blue Gene/P to over 100,000 cores. The CFD codes STAR-CCM+ are routinely applied at Argonne to simulations that use 8-200 cores, and good parallel performance has been observed. As problem size increases, i.e. for finer meshes and for large pin bundles, larger numbers of cores would be utilized and based on data from the developers good parallel performance will continue to be expected. All jobs run on Fusion will be run in parallel. A typical case in parameter exploration may cost 8 cores/node * 25 nodes * 20 hours of run time = 4000 core-hours. Project URL: Requested allocation: 200000 Q1: 50000 Q2: 50000 Q3: 50000 Q4: 50000 Justification: The requester has used undetermined amount hours of their initial startup project. In addition to approving an initial amount, please specify a Category and Subcategory for this project. For a list of the current selection of approved categories, please see: https://wiki.lcrc.anl.gov/wiki/Processes/Categories Once the Allocation committee has approved the project, please go to the Project Management page to create it: https://accounts.lcrc.anl.gov/projects.php Thank You, The LCRC Accounts System