Engine flow simulations, which show how forces and gases move within an engine, are critical to improving engine performance and efficiency. They are also notoriously complex, with the necessary flow calculations proving so computationally expensive that most engine simulations only represent one or two elements of the engine system at any given time. Now, researchers from Argonne National Laboratory have applied supercomputing power to push the limits of these processes, conducting the largest-ever simulation of flow inside of an internal combustion engine.
The simulation was run by Sibendu Som, manager of Argonne’s Computational Multi-Physics group (itself part of the Energy Systems division), and Muhsin Ameen, a research scientist at Argonne’s Center for Transportation Research (also part of the Energy Systems division).
Over the course of years, the duo retrofitted an in-house fluid-thermal simulation code called Nek5000, which won the prestigious Gordon Bell Prize in 1999, to run direct numerical simulation for engines. During that time, they ran a series of small simulations in preparation for their record-breaking attempt. Paul Fischer, the chief architect of the code, was even consulted early on in the simulation’s development.
Finally, this spring, they ran the full simulation on Argonne’s Theta supercomputer. Theta, a Cray XC40 system, comprises 4,392 nodes, each with a 64-core Intel Xeon Phi Knights Landing CPU, 16GB of multi-channel DRAM and 192GB of DDR4 memory. In total, Theta delivers roughly 6.9 Linpack petaflops, placing it 34th on the most recent Top500 list of the world’s most powerful computers. The duo ran the simulation on 51,328 of Theta’s cores to solve the calculation’s two billion degrees of freedom, which collectively determined elements like pressure and temperature.
“The current simulation effort is the first-ever direct numerical simulation of the flow and heat transfer inside an internal combustion engine for a real engine geometry and operating conditions,” Ameen said in an interview with Argonne’s Ginger Reilly. “This is one of the most detailed simulations ever of the flow in an internal combustion engine.”
“This is one of the key milestones, and there will be more such milestones from Argonne,” Som added.
Apart from breaking records, the data produced by the simulation is likely to prove useful to engine-driven industries. Automotive manufacturers, Argonne says, will benefit from the extremely detailed information on the distribution of forces within the engine, and the simulation dataset will also be able to serve as a benchmark for submodels and other lower-detail approximations.
To read the reporting from Argonne on this research, click here.