HPC Made for Manufacturing 1
HPC Made for Manufacturing
DIGITAL ENGINEERING WITH DELL EMC & INTEL ON:
The manufacturing industry is in the midst of a revolution that is powered by HPC.
Produced by the editors of Digital EngineeringSponsored by Dell EMC & Intel
Cred
it: iS
tock
/ Liuf
uyu
HPC Made for Manufacturing 3HPC Made for Manufacturing 2
connected products will become a core competency for manufacturers hoping to domi-nate in the era of digital disruption. Data from connected devices represent a goldmine for manufacturers who are able to collect and analyze it to improve new product design iterations, predict when parts will fail and open up new lines of business based on that data. According to Gartner estimates, there will be 6.4 billion connected devices in use this year, up 30% over 2015, and the research firm is projecting nearly 21 billion Internet of Things devices by 2020.
CHALLENGES TO HPC IMPLEMENTATION
While larger manufacturers have been able to reap the benefits of HPC for advanced virtual prototyping workflows, the technology has remained out of reach for the ma-jority of small and medium-sized businesses (SMBs), which includes many companies that supply larger manufacturers. According to the National Center for Manufacturing Sciences (NCMS), out of the more than 300,000 manufacturers in the United States, 95% are categorized as small or medium sized (with less than 500 employees), and a dominant 94% have yet to fully adopt HPC.
Small and mid-size companies have encountered a variety of stumbling blocks to HPC adoption, most related to complexity and cost. A typical HPC environment can be quite complex and difficult to manage, requiring domain expertise in areas like Linux, cluster management software, job schedulers and network configurations. Smaller manufacturers, most with limited IT resources, simply don’t have these skill sets or the bandwidth to devote to HPC system design, which is highly specialized and different from traditional workstation deployments.
The high price tag of HPC clusters and CAE software has also put the technology out of reach for smaller manufacturers, not to mention, the significant fees for sup-port and consulting that are typically required for deployment. Even if they do have the budget for simulation and design software, and with subscription-based licensing becoming more common, smaller shops may not know how to fine tune their software to fully exploit HPC resources. Similarly, most small and mid-sized manufacturers lack the internal expertise to ensure HPC clusters are certified and optimized to run the software and workloads they require without assistance from outside consultants.
THE DELL EMC HPC SYSTEM FOR MANUFACTURING
The new Dell EMC HPC System for Manufacturing is designed specifically to address those pain points by providing a validated solution for small and mid-size manufac-turers looking to make the leap to HPC. The pre-configured solution, designed and tested by industry experts, is optimized specifically for the manufacturing domain and architected in a way that integrates HPC capabilities into existing workstation environ-ments without disruption and without breaking the budget.
The Dell EMC HPC System for Manufacturing was created specifically to deliver the capacity and performance required for computationally intensive workloads, includ-ing those involving widely used CAE tools like ANSYS Fluent, ANSYS Mechanical, SIMULIA Abaqus, MSC Nastran, NX Nastran, LSTC LS-DYNA, and CD-adapco’s StarCCM+, among many others. The solution integrates computing, storage, network-
Manufacturing is in the midst of profound change as companies large and small work to digitize product development and better connect to an integrated supply chain to drive Industry 4.0 and Factory of the Future initiatives. Thanks to this era of digital transformation, there’s never been
more opportunity for manufacturers to differentiate through product design innovation. Faced with increasing product complexity, shorter development cycles and a requirement for global collaboration, manufacturers are looking for new ways to turbocharge that innovation in the hopes of outpacing their competition.
Manufacturers across the automotive, aerospace, consumer electronics and other sec-tors are scrambling to produce more sophisticated products in shorter timeframes at lower costs. Many are turning to virtual prototyping tools and simulation-driven product design processes as a way to test drive concepts in the digital world—a more cost effective ap-proach compared to building physical prototypes. Enlisting simulation and 3D modeling solutions as part of a virtual prototyping workflow also encourages greater design freedom
and flexibility, enabling a more iterative design cycle.
In fact, optimizing and testing product designs in a virtual world has been proven to empower manufacturers to develop best-in-class products with far fewer resources and without the protracted cycles that characterize tra-ditional development efforts. Digital prototyping has far fewer physical and budgetary constraints compared to traditional prototyping practices, while encourag-ing almost unlimited design exploration. The technique also allows manufacturing processes to be planned and perfected in a virtual world, reducing tooling costs and supporting a more efficient and error-free engineering change order process throughout a product’s lifecycle.
As companies transition to these digital prototyping and virtual test workflows, they are bumping up against the limitations of existing multi-core workstations, especially for advanced simulation and digital design work. Many installed workstations are at maximum capacity, unable to scale to meet the de-mands of today’s peak computational workflows. As a result of these deficiencies, it’s common for manufacturers to encounter challenges such as slow or incomplete job runs or delays in critical modeling, simulations or data analysis work. Compare that to high performance computing (HPC) technologies, which are not bound by similar constraints.
Tight deadlines for increasingly complex product design have resulted in many large manufacturers relying more on their network of suppliers for simulation and analysis. HPC is a key enabler of effective virtual product development and testing, particularly for increasingly complex products that have a mix of mechanical, electrical and soft-ware components. HPC resources facilitate better collaboration by creating a centrally maintained and shared platform for digital design intellectual property (IP), including everything from robust 3D CAD models to system-level simulations. This approach is essential for keeping far-flung design partners on the same page, as well as for enabling digitalization of the design process across the entire product lifecycle.
HPC resources also deliver the scalability, storage and processing horsepower nec-essary for emerging applications in areas like data analytics. The ability to capture, process and store real-time data streams measured in zettabytes coming off of smart,
95% The portion of the 300,000+ U.S.
manufacturers categorized as small or medium sized, 94% of which have not yet
fully adopted high-performance computing.
Source: National Digital Engineering Manufacturing Consortium report, “Modeling, Simulation, and Analysis, and High Performance Computing: Fore Multiplier for American Innovation,” Published by
the U.S. Council on Competitiveness, March 2015.
HPC Made for Manufacturing 5HPC Made for Manufacturing 4
ing, unified management and monitoring and service capabilities in a single solu-tion, making it easy for SMBs to adopt an HPC platform while reducing costs and mitigating deployment risks.
The platform takes a building block approach, enabling customers to quickly and easily assemble and configure re-sources to meet their requirements while enabling choice and flexibility at every level. The Dell EMC HPC System for Manufacturing encompasses the follow-ing building blocks:
Explicit Solver: For crash, stamping, safety, impact analysis, fluid flow, combus-tion, aerodynamics, acoustics, and pump design simulations using CFD tools like Fluent, CFX, STAR-CD, STAR-CCM+, OpenFOAM, and PowerFLOW or ex-plicit structure tools like Abaqus, LS-DYNA, PAM-CRASH, and RADIOSS.
Implicit Solver: For Noise, Vibra-tion, Harshness (NVH) simulations along with structural integrity analysis, both linear and nonlinear supporting workloads for such tools as ANSYS-Me-chanical, SIMULIA Abaqus, MSC Nas-tran, NX Nastran, and OptiStruct. It is available with GPU technology to bol-ster performance of graphics-intensive workloads.
Virtual Desktop Infrastructure (VDI): For supporting remote visualiza-tion for multiple users on a single, virtu-alized server running a variety of CAE applications and workloads and support-ing such VDI software packages as Nice, Horizon, XenDesktop, NX, and VNC.
Management: To handle cluster management, cluster administration, and log-in tasks, and supporting all industry-leading IPMI-based cluster management tools, including Bright Cluster Manage-ment and PBS Pro.
The system comes cloud ready, and cus-tomers have access to a variety of support plans and remote cluster management ser-vices, with a single point of contact.
Learn More Online• “Democratizaton via Virtualization” white paper: digitaleng.news/de/dellwp1• Nissan Motor Company creates a reliable and highly available backup solution: goo.gl/980gji• Dell EMC virtual desktop solutions: goo.gl/BPmelG• Dell EMC data center virtualization: dell.com/en-us/work/learn/dc-virt • Dell EMC Blueprints for Virtualization: goo.gl/u1nhIL • Dell EMC HPC: dell.com/hpc• Dell EMC HPC Community: dellhpc.org• Dell EMC HPC Tech Center: HPCatDell.com• Dell Workstations: dell.com/precision• Dell EMC Services: dell.com/services
HPC WITHOUT THE GUESSWORKConfiguring HPC servers, storage and networking can be a guessing game that if not done right, can dramatically degrade performance of the environment and impede scalability.
The Dell EMC HPC System for Manufacturing takes the guesswork out of this equa-tion. Depending on your specific needs, Dell EMC might recommend a desk-side or rack-mounted chassis configured with the appropriate servers with Intel Xeon processors and memory, shared HPC storage and networking technologies.
For example, a Dell EMC PowerEdge C6320 server equipped with Intel Xeon E5-2697Av4 processors with 128GB of DDR4 memory may be best for highly parallel crash, fluid flow, aerodynamics or combustion analysis. Other simulations, such as noise, vibration & harshness or structural integrity analysis require more memory and hard drive scratch disk access like that found in a Dell EMC PowerEdge R730 server with Intel Xeon E5-2680v4 processors, an optional GPU to help specific workloads, 256GB of RAM and eight 300GB hard drives.
Some solutions can make use of shared storage. The Dell EMC HPC NFS Storage solu-tion with High Availability is designed for a cluster with up to 100 users and modest share storage performance requirements. The system delivers 30% better throughput compared to non-optimized NFS solutions and can scale from a minimum of 48TB of raw capacity to 480TB in a single name space.
Another HPC storage option is the Dell EMC HPC Lustre Storage aimed at clusters with more than 100 users, and designed for high-performance storage requirements. It can scale from 120TB to petabytes of data in a single file system namespace and is aimed at big data workloads using Intel Hadoop Adapter for Lustre (HAL).
The Dell EMC HPC System for Manufacturing also provides multiple networking op-tions based on each manufacturer’s needs, from simple gigabit Ethernet up to 100 GbE, to more sophisticated high performance networking choices and options.
The flexible converged systems approach can help manufacturers of all sizes embrace the benefits of high performance computing, bringing HPC out of research labs and further into industry. “HPC is no longer a tool only for the most sophisticated researchers,” said Jim Gan-thier, senior vice president, Validated Solutions Organization, CPSD, Dell EMC. “We’re taking what we’ve learned from working with some of the most advanced, sophisticated universities and research institutions and customizing that for delivery to mainstream enterprises.”
Nissan Integrates End-to-End Manufacturing InfrastructureThe need for scalable storage and data backup drove Nissan Motor Company to tap Dell EMC’s end-to-end infrastructure solutions. The automotive giant, striving to achieve efficient work procedures and maintain quality standards while controlling costs, transformed the infra-structure at its Tochigi plant and is now using it as a stan-dard for rolling out infrastructure at new plant locations. While its older servers and file systems struggled to keep up with increasing volumes of data and larger backups, the Dell EMC infrastructure is keeping Nissan on the cutting
edge, providing scalable storage capacity for handling up to 40TB of data, slashing data backup times by 93%, and providing access to services that ensure redundancy and high availability. The virtualized solution also helps keep costs in check by reducing the number of servers, thus lowering maintenance costs.
“This will become the model for opening new Nissan manu-facturing plants overseas,” said Takehiko Kashiwazaki, facilities maintenance section No. 2, Manufacturing and Engineering Department No. 2 at the Tochigi plant. “The deployment, with Dell EMC products, represents an impor-tant step in keeping Nissan Motor Company on the cutting edge of technology.”
The ability for large manufacturers like Nissan along with smaller and mid-size players to readily gain access to scal-able infrastructure like HPC resources is a critical element for success in the new age of digital transformation and product innovation.
