The majority of projects in the area of vehicle optimization involve studies with large-scale variations in parameter and components on a limited palette of base vehicle models. These studies require high levels of CPU cycles on-demand. Not only SMEs, but even larger companies, struggle to provide sufficient computational resources necessary to accomplish optimization tasks in an acceptable time-frame. This case study addresses the use of on-demand, Cloud-based HPC resources to tackle the important requirement for the reduction of CO2 emissions in the design of vehicles.
The outcome of this case study has been to demonstrate the viability of on-demand computing resources in the design of powertrains with specific emphasis on the reduction of CO2 emissions. This solution involves the running of AVLs simulation codes on a Cloud-based HPC system where computer resources are made available on-demand.
The most clear cost benefit of using HPC-cloud resources is the possibility to lease a powerful computing cluster for single projects instead of acquiring and maintaining computational resources which would be underutilized for most of the time, and probably even not sufficient when really needed. Using a Cloud-based solution, taking into account all additional cloud overheads, short-term projects running millions of simulations on 400 cloud CPU cores for a period of a couple of weeks, several times a year, would run with costs reduced by up to 90% when compared to the total cost of ownership of a dedicated in-house system. This is the cost range where it becomes attractive for SMEs to participate in projects which require high CPU power for only a short time.
End-user and Code Owner: AVL
HPC Centre and HPC Expert: University of Stuttgart
The outcome of this case study has been to demonstrate the viability of on-demand computing resources in the design of powerchains with specific emphasis on the reduction of CO2 emissions.