Why Formula1 CFD Simulations Should Go with a Unified Cloud HPC Platform

The Fédération Internationale de l’Automobile (FIA) released the latest version of sporting and technical regulations for 2017 Formula One Grand Prix on April 29, 2016.  This document is the rulebook that all Formula One racing teams must follow in the 2017 season.  All of the restrictions of CFD simulations are clearly defined in Appendix 8 (Aerodynamic Testing Restrictions) – section 2 of the sporting regulations.

Computational Fluid Dynamics (CFD), a widely accepted methodology in automobile aerodynamics R&D, has been proven to speed up the turnaround time effectively. The biggest upside is that it doesn’t involve any part manufacturing and all proof of concepts (POC) can be done on computers.  In the high-end auto industry, such as sports and racing vehicle makers, CFD has been used even more intensively. In this blog post, I’ll illustrate why Formula One racing teams should leverage the cloud to advance their CFD designs and why FIA, as the governing body of the sport, would also benefit from pushing it forward.

Why there are CFD restrictions

You might ask: if CFD is so effective, why is FIA restricting its usage?

There are two main reasons:

The first is to ensure the fairness of the sport. Not all F1 teams have equal scale. Teams backed by automakers like Mercedes and Ferrari have more financial resources than independent teams like Manor and Sauber.  Big teams have more capacity to invest in high-performance compute hardware and software to perform CFD simulations, which results in a faster car that small teams may not have a chance to compete with.

The second reason is cost control. F1 was once notorious for its lavishness. With the fluctuation of the global economy, we’ve seen some independent teams, even those backed by automakers, drop out in recent years. Making the sport more economical is a top priority for the FIA.

Do CFD restrictions serve a real purpose?

If you read carefully through Appendix 8 (“Aerodynamic Testing Restrictions”), Section 2 (“Restricted CFD Simulations”) of the F1 sporting regulation, you’ll find the restrictions apply to compute hardware only, for example, the number of processing unit cores in the cluster and the make and model of processing unit.  The objective of the rules is to put a cap on the theoretical maximum compute power, as measured in FLOPS (floating-point operations per second). However, there are no limitations on software innovation. Also, for hardware, wealthier teams can invest in improving the interconnections, with InfiniBand for example,  which will increase the computation speed.  In the net, the regulations still have some opportunities which can be explored by the teams.

Hurdles for teams and FIA for traditional on-premise CFD computation

For traditional on-premise CFD computation, the hurdles for the teams are the following:

  1. Cost to maintain on-premises hardware
  2. Cost to maintain the simulation software and associated licensing
  3. Poor utilization with wasted resources during the idle periods
  4. Insufficient compute power during the R&D periods
  5. Manage simulation data securely and efficiently

A large hurdle facing FIA is auditing.  After teams report their on-premise HPC information for CFD simulations and locations, the auditor has to be physically on-premise as well to verify the hardware specifications.  Additionally, there is also no guarantee that the hardware being audited will be the same hardware running those CFD simulations.

A unified cloud HPC platform – a solution for all

Believe it or not, a unified cloud HPC solution can solve all the problems above.  But it has to meet the following requirements:

  1. Data security and integrity.  Simulation data is one of the most valuable assets to a F1 racing team.  For a cloud HPC platform with multi teams running simulation jobs on it, being able to protect the data from unauthorized access and being tampered should always be the top priority.
  2. Provide unlimited compute hardware resources, all in an “on-demand” fashion. This resolves the issue of a lack of compute power during the R&D phase. Engineers can boot up a “big enough” cluster whenever needed and run simulation jobs as fast as possible.  Also, the user only pays for the time when jobs are running which is a lot more cost efficient than traditional on-premises HPC.
  3. Offer CFD software licensing options with hourly pricing.  For the software vendors that offer this, this provides a variety of selection in the CFD engineer’s tool box.  Software is charged hourly based on the size of the clusters and lengths of their CFD jobs.  From a cost perspective this is more efficient than purchasing those expensive long term licenses.  And engineers also don’t have to worry about not having enough licenses to run their job.
  4. Provide the same set of hardware and software selections for each team.  This restricts the set of hardware and software selections a team can use. More importantly, all the teams are given the same set of tools which guarantees fairness.
  5. Provide a friendly user interface for CFD engineers to run and monitor and retrieve results for their CFD jobs.  This helps CFD engineers set up their jobs and retrieve the results faster which also makes the process less error prone.
  6. Provide an administration portal for admin users (auditors from FIA) to monitor and budgeting both software and hardware usage.  This feature will facilitate managing the auditing process for FIA dramatically.  Staff of FIA can set the same budget (cap) for each team.  This budget applies to all of the hardware and software usage for each team.  No more jobs can be run once the budget cap is hit.  FIA auditors can easily monitor the jobs at a high level overview (total hardware usage in core-hours, total software usage in license-hours) or with a detailed breakdown for each job.  This can all be done in a web browser.
  7. Zero IT cost.  Imagine if the F1 teams and FIA wouldn’t have to maintain the hardware and software at all, but instead, the CFD simulations would always be able to run with the latest hardware and software.  With cloud HPC, this dream is realized, and even better, with no IT cost.

The world is embracing the cloud, and F1 will as well

CFD simulation is considered a critical engineering tool in Formula One racing car design and development works.  But even though most of the IT world are pivoting to the cloud environment, almost all of the F1 CFD simulations are still conducted on on-premise hardware and software.  With a unified cloud HPC platform, the innovations iterate faster while the hurdles introduced by traditional on-premise HPC disappear–definitely a good thing for leading edge F1 teams.

This article was written by Irwen Song.