NCMS Lightweight Automotive Materials Project Portal

Develop, maintain, and support a secure cloud portal for on-demand procurement and access to technical computing applications required to support the Lightweight Automotive Materials Project (LAMP) funded by the Department of Energy (DOE).

Phase 1 FEA

Nimbis Services, Inc. under contract from the National Center for Manufacturing Sciences (NCMS), developed, maintained, and supported a secure cloud portal for on-demand procurement and access to technical computing applications required to support the Lightweight Automotive Materials Project (LAMP) funded by the Department of Energy (DOE). 

The target high performance computing (HPC) platform for the portal was the Glenn cluster at the Ohio Supercomputer Center (OSC). The pilot LAMP users of the Nimbis LAMP portal at OSC were Wayne State University (WSU) students and researchers. The finite element analysis (FEA) modeling and simulation applications provided through the portal included Abaqus, ANSYS, LS-DYNA, and Hypermesh. The portal supports batch solvers running parallel jobs on a cluster of compute nodes and interactive pre and post processing of job inputs and outputs running on a graphics processing unit (GPU) node. Academic licenses for the application software were provided by OSC to the WSU users.

Nimbis developed and demonstrated two approaches for L&L Products to access Altair’s RADIOSS software through an on-demand cloud portal with ecommerce capability. The first approach resulted in an implementation using the open source PHP coded Drupal content management system (CMS) with Ubercart ecommerce framework hosted on Amazon Web Services Elastic Compute Cloud instance. This configuration had a backend interface for launching a remote desktop over a virtual private network connection to a single sign-on login node on an R Systems high performance computing cluster. The second approach resulted in an implementation using the open source Python coded Django web application framework with Django’s Mezzanine CMS application and Django’s Cartridge ecommerce application. This configuration was also hosted on an Amazon Web Services Elastic Compute Cloud instance. In this case the backend was an interface to Altair’s HyperWorks On-Demand cloud-based high performance computing platform running the Altair Compute Manager. Through this interface, L&L Products successfully ran their test cases on RADIOSS.

Phase 2 FEA

Under the LAMP pilot program, Nimbis developed and demonstrated an on-demand cloud portal with ecommerce capability. This approach resulted in an implementation using the open source PHP coded Drupal content management system (CMS) with Ubercart ecommerce framework hosted on Amazon Web Services Elastic Compute Cloud instance. This configuration had a backend interface for launching a remote desktop over a virtual private network connection to a single sign-on login node on the OSC Glenn high performance computing cluster. The Nimbis commercial deployment, funded and developed separately from the LAMP pilot, resulted in an implementation using the open source Python coded Django web application framework with Django’s Mezzanine CMS application and Django’s Cartridge ecommerce application. This configuration was also hosted on an Amazon Web Services Elastic Compute Cloud instance. The Django web application framework in the commercial deployment provided a superior software development environment compared to the Drupal implementation in the pilot program resulting in a superior portal product. The results of the LAMP pilot program provided motivation for the development of the commercial deployment which benefited from lessons learned on the pilot.

Phase 2 CFD

The initial goal of this project was to develop a tool to address some of the issues small to medium-sized manufacturers in specialty markets face in adopting HPC based modeling and simulation technologies. As it pertains to truck add-on parts suppliers, OSC and project partners, along with the DOE LAMP program and NCMS, achieved this goal through the development of an easy-to-use, web-based portal system that would allow these manufacturers to leverage HPC resources at reduced costs.

SPONSOR / PARTNER
National Center for Manufacturing Sciences

DESCRIPTION
Lightweight Automotive Materials Project (LAMP)

YEAR
2013

TOPIC
Portal Hosting

PRINCIPAL INVESTIGATOR
Brian Schott

PRINCIPAL-IN-CHARGE
Robert Graybill

The major deliverable resulting from this project is the actual Truck Add-On Predictor, which remains available online via the Nimbis Services Marketplace. Notable capabilities of this portal include the following:

• Inclusion of a baseline truck/trailer geometry contributed by SOLUS and rough wireframe of the model available for download
• User access to default simulation results from a suite of J1252 runs at 65 mph that can be altered for additional design variations
• Creation of the user option to include an arbitrary number of add-on STL files
• Calculation of a wind-averaged drag coefficient for a suite of J1252 runs during post-processing stages
• Production of a CSV report of important model parameters and CFD results
• Production of an HTML report with high level slice pictures and data
• Comparison of results for two or more CFD simulations
• Completion of a simulation stability analysis after each run

Major project accomplishments include the following:

• Development of a basic graphical user interface (GUI)
• Creation of OpenFOAM example jobs
• Creation of example output reports
• Completion of successful runs at OSC
• Completion of friendly user testing at OSC
• Production of a value proposition report in terms of run time vs. accuracy (see Figure 6)
• Creation of an “intelligent” feedback report based on results from multiple runs
• Launch of web-based Predictor portal accessible via multiple platforms
• Creation of flexible use options
• Implementation of a flexible tiered payment structure including subscription and pay-as-you-go pricing
• Implementation of Software-as-a-service model