HPC Propels Fastest Production Car
Swedish car manufacturer Koenigsegg holds many speed records, including breaking the 0-300 km/h record in just 14.53 seconds in 2011, holding the Nürburgring speed record of 311 km/h in 2006 and also winning the title of fastest production car in 2005 for a 389 km/h top speed. To continue on this winning trajectory, Koenigsegg’s Technical Director, Jon Gunner, turned to another ultra-fast technology: computers.
The company had an aggressive goal of developing a market-leading hyper car in just six months. As Gunner explains, they were on a mission to deliver a vehicle that would “outperform all production cars on the track and, without modification, also be able to reach a top speed of 440 km/h.”
Smaller than many of its competitors, the 50-person company needed a competitive edge that would enable it to outdesign rivals. To facilitate the development of its latest hyper car, Koenigsegg partnered with High Performance Computing (HPC) Wales, ICON Technology & Process Consulting Ltd. (ICON) and Fujitsu.
With ICON’s assistance, Koenigsegg’s engineers are accessing HPC Wales’ advanced computing infrastructure to simulate the aerodynamics of the car. It’s the type of experiment that would previously require the use of expensive physical prototyping using wind tunnels, but Koenigsegg was committed to an all-virtual design process.
David Green, Commercial Director of ICON, discusses the partnership, “440 km/h is far and away faster than any wind tunnel can ever reproduce. Koenigsegg also don’t have the resources to build lots of prototypes. So, wherever possible they use virtual design for the structure, aerodynamics and fluid dynamics. This makes Koenigsegg unique; they are entirely committed to virtual design. We saw an opportunity for ICON to help Koenigsegg develop the car by utilising our relationship with Intel, Fujitsu and HPC Wales. We worked out a deal that gave them rapid access to the multiple cores on the HPC Wales system, required for highly computationally intensive CFD simulation.”
Because the team would not be verifying the results with physical testing, the simulations had to be accurate. The proof would come on racing day.
The project took a couple months to complete and during that time Koenigsegg’s engineers carried out more than thirty different simulations of the aerodynamics. Some runs used about 128 cores and ran for twenty-four hours to capture one second of air flow over the car. That was the kind of detail they were after.
Says Green: “We used a highly accurate method called Detached Eddy Simulation, where we make very few assumptions to simplify the simulation, but we can describe very accurately what happens to the air surrounding the car. There are cheaper ways of doing CFD simulations but, in a case like this, where Koenigsegg are not going to be able to do a lot of fine tuning with prototypes, we wanted the simulation to be as close to real life as possible. HPC Wales has enabled this prestigious project with Koenigsegg.”
The project is a perfect example of using HPC as a Service in order to augment existing computational resources. The Swedish designer car maker already used iconCFD software in-house, but its cluster tapped out at 32 cores. The system was sufficient for constructing models in-house and performing simple jobs, but it did not meet the requirements of this extravagant mission.
The payoff of all this computational work is a new supercar called One:1, which debuted at the 2014 Geneva Motor Show. In simulations, the “world’s first mega car” as its being called hit its aggressive 440 km-per-hour (273 miles-per-hour) target. That’s a few clicks higher than other reining speed champs the Hennessey Venom GT and the Bugatti Veyron Super Sport. The One:1 is also expected to go from 0-250 mph (400kph) in just 20 seconds. If One:1 can match the simulated speed on the track, the company will have achieved its goal of creating the fastest production car in the world.