Cycle-to-Cycle Combustion Variation Modeling

Project Description

Dilute combustion with cooled EGR is an effective technology for improving fuel efficiency while meeting future emissions requirements. Dilution with EGR allows decreased pumping and heat transfer losses, while allowing the continued use of very effective exhaust aftertreatment with a 3-way catalyst. However, at the higher EGR rates corresponding to larger benefits, cycle-to-cycle variation in combustion becomes a limiting factor. A deeper understanding of the mechanisms leading to cycle-to-cycle variations would allow improved engine system designs that enable robust combustion at high EGR levels, thus achieving higher fuel efficiency and lower emissions. We believe that the world-leading edge super-computing facilities at OLCF have the capability to enable such large scale LES simulation of engine air-induction, fuel injection, spray atomization, fuel-air mixing, and combustion processes. We believe that it should be possible to adapt and reconfigure the in-cylinder combustion simulation software that Ford is currently using in a new way that exploits the large scaling potential of the OLCF computer facilities. This work will build on new algorithms, jointly developed by ORNL and Ford, for efficiently sampling and interactively guiding massively parallel, long-time-scale simulations of multi-scale

complex dynamic systems in ways that have not been demonstrated before. We expect the knowledge gained about the multi-scale physics and chemistry involved in this type of constrained turbulent combustion will also be applicable to other complex important complex reacting systems. The project will also demonstrate the viability of this large-scale computing intensive modeling approach, show scale-up of the software, and show value of High Performance Computing (HPC) at scale.