The car manufacturer Porsche has for many years used computer simulation resources at the High-Performance Computing Center Stuttgart (HLRS) to develop better vehicles. In the early 2000s, however, it became increasingly clear that existing commercial software packages were not keeping pace with the constantly doubling performance of new supercomputing hardware. In an effort to accelerate the development of more efficient and more effective simulation software, Porsche joined forces in 2008 with HLRS, Daimler, Opel, and other companies—with support of the Baden-Württemberg Ministry for Science, Research, and the Arts—to found the Automotive Simulation Center Stuttgart, also called asc(s.
Aug 16, 2018
HPC in Industry
Automotive
Interview
Solution Centers
See all news
Since then the asc(s has operated as the center of a network of large and small companies from various corners of the automobile industry. Because computer simulation is the focus of the asc(s, HLRS continues to be an important node in this network. As a testing site for new software solutions and a source for the latest knowledge about high-performance computing (HPC), HLRS works closely with other members of the asc(s.
At a gala event on June 6, members of the asc(s took the opportunity to look back on the last 10 years. With lectures by Michael Resch (HLRS director and chairman of the asc(s board of directors), Ulrich Steinbach (director and head of the Baden-Württemberg Ministry for Science, Research, and the Arts), and Christoph Gümbel (futurist and former director of virtual vehicles at Porsche), the attendees reflected on the organization's accomplishments and how simulation could open up new opportunities for the auto industry in the future.
Shortly before the gala we spoke with asc(s Director Alexander F. Walser about the center's activities, what role HLRS has played in it, and why simulation continues to grow in importance for the field.
The asc(s is a nonprofit association that brings car manufacturers, software developers, hardware manufacturers, engineering service providers, start-ups, and scientific organizations together around a table to guide and steer the future of virtual automobile development. Just as an OEM (Original Equipment Manufacturer) can buy a screwdriver to build a car, we try to support virtual automobile development, so that manufacturers have access to the simulation methods that they need.
To make this possible, the asc(s brings together the critical mass of interested parties necessary to get projects off the ground, whether it be in terms of expertise or financing.
It is also an important part of our mandate to gather and integrate small and medium-sized enterprises (SMEs), as well as scientific institutions. SMEs can sometimes develop a very innovative software package, but if the interfaces aren't built in the right way or if the workflow doesn't fit, industry might not be able to use it. Getting OEMs and SMEs to work together reduces the risk of developing something in the wrong direction. At the same time, building these relationships incorporates industry's needs into software development very early and makes it possible to design more general solutions that meet the needs of many different OEMs at the same time.
Ten years ago the foundational idea of the asc(s was very innovative. Today, it fits contemporary needs perfectly, and in the future it will surely be indispensable for driving innovation.
High-performance computing is a key focus at the asc(s, a fact that distinguishes our organization from other interest groups in the automobile industry. It is only through our cooperation with HLRS that we can maintain this strong focus on HPC.
Our cooperation with HLRS makes it possible for us to always have access to the newest hardware architecture. With its frequent updates, the HLRS computing architecture is always a step ahead of what's available in industry, where the update cycles are longer and the cluster sizes are smaller. Using HLRS resources, we can show relatively early that new methods will work on future cluster systems.
By including HLRS in project planning, we also learn about potential problems. The projects are very diverse in terms of the number of computing cores used or of the parallelization necessary to achieve research goals. By working together with HLRS we learn how we can better address specific kinds of problems.
Another very important question for OEMs and small businesses is how to design a workflow that connects the workstation under your desk with the HLRS cluster system. In the past we have built many kinds of workflows in which processes can be automatically connected between classical workstations and HLRS. The data and results are automatically transferred back and forth using only the necessary software resources. In practice this means that engineers don't just push everything onto Hazel Hen, but rather they can integrate this powerful computer into their workflows.
Ten years ago no one was talking about self-driving vehicles. Electric cars were a theme that was already in the air, but not something that was as prominent as it is today. Now everything must develop much, much faster.
Virtual crash tests are also becoming increasingly important. Countless test scenarios need to be conducted, raising the question of how HPC and driving simulation can be better combined. In the medium-term, such tests will have to be conducted on a supercomputer. This also shows how important it is to move this large number of tests—which are necessary for safe and efficient development—into the virtual world.
Technological trends related to artificial intelligence are also gaining in importance. This will result not only in the use of the newest AI-technologies in driving systems, but also in asking how we can use AI to support and lessen the burden on engineers during the virtual development process. How can we apply AI to design simulation models more rapidly? How can we extract more knowledge out of the enormous number of simulation results that now exist?
Because of these new trends, the horizontal networking of automobile manufacturers that the asc)s promotes is very important, particularly because new key players are entering the market that have never before been integrated into the supply chain.
Simulation continues to become ever more important for the auto industry. This presents an additional important question: what qualifications will tomorrow's simulation engineer or even an engineer in another field need to have?
At the asc(s we have been asking ourselves how we could support our member organizations by helping to ensure that university students gain the appropriate basic knowledge during their studies. This includes engaging in dialogue with universities as well as finding ways to motivate students to study virtual automobile development, showing them the potential applications and making it easier to select the best program of study.
It's not just that methods that get developed need to be implemented quickly, but also that junior team members—who come from the University of Stuttgart, for example—must quickly become integrated into the development process. In order to do so they need to have a basic understanding of a variety of concepts related to simulation and IT, how workflows function, how hardware works, and how cloud technologies work.
The need for experts in simulation is constantly growing. By promoting young investigators, we can not only offer advantages for industry, but can also strengthen Baden-Württemberg as a center for research by directly integrating young talent into our joint projects. We expect that this kind of cooperation will grow stronger in the future.
—Interview by Christopher Williams
