Three case studies illustrate the value of HPC for engineering.
Digital transformation is a buzzword of the decade in manufacturing. Factories are investing in digital technology solutions such as:
- Smarter, network-connected machine controllers and PLCs
- Automation
- Data analysis software
- IoT sensors
This shift is part of the larger trend toward Industry 4.0, or a manufacturing industry enhanced with automation, analytics and AI. “Data is the new oil” is the catchy sound-bite. However, as many engineers, IT departments and production leaders have learned, data is useless unless it produces actionable insights. In addition, other high-tech manufacturing technologies, such as advanced CAD and simulation, require advanced processing and powerful computer analysis to produce value. When you get to the bottom of it, much of industry 4.0 rests on raw computing power.
In many factories today, you’ll find equipment and processes running on old, unsupported hardware and software. Many manufacturing decision-makers have an “if it ain’t broke…” mentality about upgrading IT. However, today’s processors can perform the work of several older-generation processors, and new hardware in general is more stable, reliable and secure than older technology. In today’s manufacturing landscape, digital transformation running on high-performance computers is increasingly a must-have for organizations looking to stay competitive. It is no longer a nice-to-have.
What Is High Performance Computing?
Compared to a Model T, my 2008 Hyundai is a high-performance car. Similarly, it’s intuitive that the term “high performance computing,” or HPC, is a relative term. After all, today’s AMD Ryzen 9 processors run 5,000 times faster than 1969’s Apollo Guidance Computer.
For science and engineering, HPC typically refers to computing at a much higher level of complexity and speed—generally achieved by aggregating the processing power of a cluster of smaller processors. A high-performance computer cluster coordinates multiple processing units to perform one singular task, such as a simulation.
In this article, we’ll take a look at three case studies demonstrating the clear value of high performance computing in manufacturing.
Rimac Technology: Simulation for Success
Do you buy American cars or Japanese? Or German? How about Croatian? A recent case study from Microsoft Azure describes how Rimac Technology set out to design and build the most powerful electric sports car in a garage, when common sense seemed to suggest it wasn’t possible. But by leveraging advanced simulation to iterate and test designs for the electric motors, inverters, gearboxes and battery packs, the small company has been able to improve and validate its designs faster and at a lower cost compared to building prototypes.
Ivan Krajinović, head of Simulations at Rimac Technology, states, “The power of simulation is that you can build really complex models, test them virtually and then iterate and modify the design to do the simulation again—all without needing to produce anything.” In addition to reducing costs by crashing virtual cars instead of metal ones, simulations provide comprehensive test data that can be analyzed and processed from any perspective.
Krajinović continues, “When a car crashes against a wall, you can’t see, for example, that a bracket somewhere in the middle of the car is deformed and the source of a larger problem. That’s only identifiable through the power of digital virtual prototyping.”
While some manufacturing facilities still maintain servers on premises, much of this processing demand has been shifted to tech giants such as Microsoft Azure, Amazon Web Services (AWS) and Siemens through software-as-a-service (SaaS) models. In this case, customers access software, as well as massive processing and storage, via the cloud. This was the case for Rimac, which initially ran its own HPC cluster before migrating to Microsoft Azure to expand capability without increasing its internal IT costs.
“We needed more computational power to work with more complex simulation models to match increasingly challenging customer requests,” says Krajinović. “Our internal cluster was really strong, but the availability of the nodes and the calculations we could run with Azure HPC were just amazing.”
By using cloud-based HPC to enable better computational speeds, Rimac can run simulations faster and with greater scale and computing power. In fact, Rimac has reduced its processing time from weeks to days. In addition, the cloud-based architecture enables greater agility for tackling bigger and bigger problems. “However complex the model we need to create, we know that we can manage it with Azure HPC,” says Krajinović. “We now produce more highly complex models that simply wouldn’t have been possible on our old infrastructure—they were just too big. And we’re doing iterations for our customers that we couldn’t have done previously.”
JOST: Improving Operational Efficiency with Data-Driven Insights
This HPC case study from Hewlett Packard Enterprise details how JOST, a manufacturer of automotive and heavy equipment such as axles, hydraulic systems and fifth-wheel trailer hitches, leverages HPC to run analytics that drive greater efficiency in manufacturing operations at its 32 facilities across the globe.
JOST is focused on improving efficiency through automation and quality improvements. For example, data is collected during high-volume assembly for traceability purposes. “Sensor data such as the number of wrench turns and torque in assembly operations is automatically collected via Wi-Fi,” said Martin Frischkorn, IT team lead at JOST in Germany. At other JOST facilities, highly complex CAD models of custom products are needed.
According to Frischkorn, cloud computing is essential for meeting the needs of different facilities. “Manufacturing sites need server redundancy to help prevent downtime. Sites where CAD data is generated need more disk space, allowing engineers to work efficiently on increasingly complex models,” says Frischkorn. Larger JOST sites will use up to four new servers, while smaller sites will typically receive one or two. “These new servers are much more powerful than the ones they replaced. A single processor now does what used to take two processors. That helps us cut maintenance costs plus software licensing fees,” Frischkorn explains.
For JOST, HPC is used to support a vast array of needs across a global business, securely and reliably, with built-in scalability.
Mercedes-AMG Petronas F1: Racing to Perform
Formula 1 (F1) is an elite motorsport that attracts the world’s top engineering talent. In an attempt to keep the playing field level for the drivers, the Fédération Internationale de l’Automobile (FIA), the F1’s governing body, imposes strict regulations on nearly every aspect of the design, build and operation of the racecars.
In this case study from AMD, Mercedes F1 engineers aimed to fully optimize the aerodynamics of their cars. But FIA regulations regulate the amount of computational fluid dynamics (CFD) analysis and wind tunnel time each team can use to prevent teams with more resources from gaining an advantage. The limits include how many model geometries can be tested, as well as the budget for CFD. In addition, similar to draft order in other team sports such as football, a balancing rule is in place that allows lower-place finishing teams more resources than top finishers. As F1 fans know, Mercedes was the top finisher of the 2021 season, meaning the company has even more reason to make the most of the limited resources it is allowed to use.
“There’s also a regulation of how many geometries we can run in a certain period, which usually spans eight weeks. We’re trying to maximize everything we can do in that period to get the most out of our CFD,” says Simon Williams, head of Aero Development Software at Mercedes-AMG Petronas F1. “It’s about trying to maximize the work that the CFD solve can do per clock cycle.”
To deliver the best CFD performance, the Mercedes team leveraged HPC. Specifically, the Mercedes F1 team chose 2nd Generation AMD EPYC processors to replace the three-and-a-half-year-old system it had from another vendor. “We gained a 20 percent efficiency improvement on the old system. This is a big step because we’re usually looking at one or 2 percent gains. The new system is allowing us to focus our effort on aerodynamic performance,” says Williams.
With the speed and power of HPC, the Mercedes F1 team can iterate faster and more reliably. Of course, performance automotive is a niche manufacturing area—building one car instead of one million—but the value of faster process iteration is apparent to all areas of manufacturing.
In the coming F1 season, Williams is working to anticipate further planned limits to aerodynamics analysis and resources. “There’s a huge upheaval in the technical regulations coming for 2022,” says Williams. “Because it’s such a big change, we need to get as early a start as we can on our car. The AMD EPYC servers have enabled us to run a lot more work in parallel. Aerodynamicists, if they get a result back during the day, they can do another design and do a second run overnight. When they return in the morning, they will have completed two iterations, rather than one. You’ve got to have the right hardware to do that.”