Azure HPC accelerates Nissan Motor’s CAE innovation by 30% and delivers cost savings

Nissan modernized from on-premises servers to a cloud infrastructure to expand processing capacity and accelerate engineering simulations, yet engineers still faced efficiency challenges and resource limitations.

The company migrated to Linux on Azure for superior performance, scalability, and flexibility. Azure CycleCloud helps ensure rapid provisioning and management of virtual machine environments for efficient design simulations.

Azure not only provides Nissan with a platform for future AI implementation, but also contributes to a 30% reduction in Computer-Aided Engineering (CAE), enhances engineering productivity, shortens time to market, and reduces total software costs.

Product development while managing costs requires sophisticated engineering and extensive computer simulations to assess the impacts of changes on crash safety, aerodynamics, and other vehicle characteristics.

The simulations, performed on a Computer-Aided Engineering (CAE) system, predict, analyze, and optimize vehicle and component performance. The CAE Analysis performs computational processing on voluminous data of material properties, three-dimensional models of engine parts, etc. As a result, the system requires significant computing resources. Although the use of a standard Linux cluster improved scalability and flexibility, capacity constraints still required Nissan engineers to wait for simulations to be completed until resources became available. The company needed a more scalable and high-performance computing (HPC) environment to work more efficiently, innovate more quickly, and maintain engineering productivity.

Improving performance by migrating to Azure

Over the past half-dozen years, Nissan migrated its CAE processing from on-premises servers to HPC environments in a multicloud infrastructure. The engineering team worked with multiple cloud providers over time. Although modernization increased processing capacity and scale, more improvement was needed.

Dai Matsubara, Senior Manager and CAE System Director (at the time of the interview) for Nissan Motor Company, said, “On multicloud, we can get more competitive cost of performance. But working with large computing volume meant simulations could still take about a day. That limited engineering efficiency and speed. The Azure HPC environment is more competitive in both cost and CAE application performance.”

After assessing various cloud options, Nissan transitioned to Microsoft Azure on Linux on Azure infrastructure. The decision was driven by superior performance of Azure Virtual Machines, including HBv4-series instances co-engineered with Advanced Micro Devices (AMD). Nissan also used Azure HBv4-series virtual machines, which are optimized for high-performance computing workloads such as computational fluid dynamics and finite element analysis, providing up to 176 AMD EPYC CPU cores with AMD’s 3D V-Cache technology. These virtual machines use AMD EPYC processors, known for their availability, scalability, and cost efficiency, for demanding simulations. In addition, Nissan used multiple Azure global regions and the latest generations of virtual machine instances, ensuring state-of-the-art performance tailored to its needs.

“The Azure HPC environment is more competitive in both cost and CAE application performance.”

Dai Matsubara, Senior Manager and CAE System Director, Nissan

Ensuring smooth and familiar workload orchestration

Over several months, Nissan migrated its engineering data and CAE workloads to an Azure HPC technology stack with managed workload orchestration services. Dentsu Soken helped align the solution with the company’s multicloud architecture, which includes Oracle Cloud Infrastructure.

With Linux on Azure in place, Nissan relies on Azure CycleCloud for virtual machine provisioning, management, and network interfaces. Using Azure Virtual Machine Scale Sets, engineers can create and manage groups of identical, load-balanced virtual machines that provide flexibility through automatic scaling and distribution across availability zones. This approach makes provisioning virtual machine environments for simulations fast and easy while minimizing costs based on actual use.

Matsubara explained, “If we submit job requirements, Azure CycleCloud can immediately provision instances to handle our request, and when the job ends, this instance can automatically go off. That’s a very useful on-demand function of Azure.”

The result is a single Azure platform for innovation that can unify data, computing, and use cases for distributed teams. Interactivity with popular HPC tools helps ensure Nissan can continue to use Altair PBS as its chosen solution for scheduling jobs and managing the workloads within its Azure HPC environments. Nissan also uses several NetApp storage systems, incorporating Azure NetApp Files to help ensure efficient data caching and access across those systems. Caching is particularly important, because Nissan uses multiple Azure global regions extending from Japan to the eastern United States. Given the distances involved, cache technology helps deliver successful and timely transfer of the abundant CAE data to facilitate processing without delays.

“If we submit job requirements, Azure CycleCloud can immediately provision instances to handle our request, and when the job ends, this instance can automatically go off. That’s a very useful on-demand function of Azure.”

Dai Matsubara, Senior Manager and CAE System Director, Nissan

Compute resources can be provisioned whenever needed and deprovisioned as simulations complete and CPU demand drops. The following architecture diagram illustrates how Nissan uses its Linux on Azure platform, which incorporates Azure CycleCloud and other tools, for processing of its CAE simulations.

Nissan architecture diagram

More about this diagram

Improving performance while reducing costs

The move to Azure improved productivity in CAE and engineering. “Azure gives us 30% faster performance,” said Matsubara. That means engineers can perform more simulations with less lead time—and often lower costs.

“We consider total cost,” explained Matsubara. “The performance improvements also contribute to overall cost reduction. If we get 30% faster, CAE software costs can be compressed by about 20%. That’s a big benefit for us because the software cost is twice to three times more expensive than processing costs.”

These improvements enable better design quality before the company invests in physical prototypes, thereby reducing product development cost. “Although the direct estimates are difficult, lower product development costs will ultimately contribute to lower vehicle prices, so Nissan customers will benefit, too,“ Matsubara said.

Migration provides foundation for future innovation

With the migration to Azure, Nissan now has a foundation for large-scale, high-speed CAE and can better use cloud CAE throughout the organization. The initial migration to a multicloud environment has provided the company with a better cloud environment.

Matsubara said, “The migration to Azure has provided the essential engineering simulation infrastructure for automotive OEMs (original equipment manufacturers). The platform supports our Linux CAE workloads and offers the agility needed to develop safe and reliable vehicles.”

“The migration to Azure has provided the essential engineering simulation infrastructure for automotive OEMs (original equipment manufacturers). The platform supports our Linux CAE workloads and offers the agility needed to develop safe and reliable vehicles.”

Dai Matsubara, Senior Manager and CAE System Director, Nissan

With Azure technologies, Nissan engineers can innovate faster and more efficiently, speeding time to market, creating safer and more reliable cars for drivers, and enabling the maintenance of innovative leadership.

Discover more about Nissan on LinkedIn, X/Twitter, and YouTube.

Dai Matsubara, Senior Manager, Engineering & Design System Department Business System Solution Division (at time of interview - Left); Takumi Yamada, Engineering & Design System Department Business System Solution Division (at time of interview - Right)