Hello, A yearly allocation for the LCRC cluster has been requested with the following updated information: Submitter/PI: Sibendu Som Project Name: CAT-engine-modeling Division: ES, MCS Project title: Simulation of Internal Combustion Engines with High-Performance Computing Tools Associated funding: DOE Office of vehicle technologies, Caterpillar, Inc., Convergent Science Inc. Other Systems: CATERPILLAR Cluster, CONVERGENT Science Cluster, Electro-motive Diesel Cluster Science: This project will use an Eulerian-Lagrangian Computational Fluid Dynamics (CFD) software called CONVERGE to simulate internal combustion (IC) engines using High Performance Computing (HPC). The objective is to develop “best-practices” for performing IC engine CFD simulations using HPC, in the context of multi-cycle and multi-cylinder simulations. Project description: This project is an extension of our work done in FY15 using LCRC resources, which resulted in the development of best-practices and recommendations for performing multi-cycle simulations of a single cylinder of a six-cylinder heavy-duty engine. Several grid resolution and scaling studies were performed to determine the optimum grid resolution to use in-cylinder and in the ports, and number of cores to run these simulations on. The goal for FY16 is to use the LCRC allocation to extend these single-cylinder simulations to multi-cylinder simulations, and simulate all six cylinders concurrently, along with intake and exhaust runners. This would require scaling up the code to run on several hundreds of cores, and we plan to leverage our ongoing efforts in the area of improving the scaling of CONVERGE on thousands of cores on Mira. Multi-cylinder and multi-cycle engine simulations are critical to capture effects such as cylinder-to-cylinder and cycle-to-cycle variations in the combustion process in order to develop strategies and engine designs to simultaneously improve efficiency while reducing emissions. The goal is to develop guidelines for Caterpillar to establish a set of best-practices and a streamlined workflow in terms of running such high-fidelity simulations on their own HPC resources towards supporting their production engine design and development efforts. Industry partnership: The project is a CRADA with Argonne, Convergent Science, and Caterpillar Inc. It is expected that the results of the industry project will not only advance the state-of-the-art in engine simulations but also significantly benefit Caterpillar in terms of modifying their best practices for engine simulations. Argonne will be performing the simulations, post-processing results, and writing technical reports. Convergent Science will be implementing the suggestions from Argonne and CAT for further development of their code for HPC applications. Caterpillar Inc. will provide the geometry to be modeled, give guidance on simulation results, and their internal needs. Project URL: http://verifi.anl.gov/ Current FY Hours Used: undetermined amount New FY Requested allocation: 500000 Q1: 125000 Q2: 125000 Q3: 125000 Q4: 125000 Justification: During the past four years, we have been working in close collaborations with the developers of CONVERGE code to ensure that it scales well both on clusters and super-computers. We have implemented METIS algorithm which enables us to load balance significantly better than the original load balancing in CONVERGE. This was the first step towards enabling CONVERGE code for HPC use. This project was done through the core hours provided by LCRC in collaboration with CAT and CSI researchers. In collaboration with researchers at ALCF (during the past couple of years), we have further performed scaling studies with the code on Mira. Based on implementation of MPI-I/O, advancements in chemistry calculation with a stiffness based load balancing criteria and CFD load balancing, we have been able to show good scaling of the CONVERGE code on Mira on up to 4096 processors for a closed cycle engine simulation with about 10 million cells. These advancements were showcased during the VERIFI workshop in Nov. 2014 and we have published a detailed paper highlighting these changes. The above changes are not only expected to benefit calculations on Mira but also help on computing clusters like Fusion and Blues. More details and paper can be made available on request. Storage requirements: 1 TB Thank You, The LCRC Accounts System