 |
Modern manufacturing is increasingly driven by the need for both high flexibility and speed in process and product development. HPCC technologies enable high fidelity computational prototyping in a shorter time, resulting in products that are lighter, stronger, safer, and cheaper to manufacture and use. HPCC infrastructure can also provide more efficient and flexible production systems through integration of planning, design, production, quality control, marketing, and user services. Such needs span many industries, including aerospace, chemical, electronics, and materials processing. |
|
|
A computer generated image of a metal flange [far left], part of a subway token collection machine, and a 3-D solid prototype [left] fabricated by a stereolithography device at the San Diego Supercomputing center's Telemanufacturing Facility. |
The continuing need for high end computing in manufacturing is illustrated by semiconductor and integrated circuit design where feature sizes have reached the submicron level. Since the cost for a new semiconductor process exceeds a billion dollars, accurate device and process simulation is key to maintaining industry leadership. However, such simulations necessitate modeling complex nonlinear effects not present in current simplified models, and require both teraflops of computational power and terabytes of memory. Agency sponsorship includes several projects in ultra small device modeling including a joint effort by several universities and private companies.
Aerodynamics research also has severe computational requirements but high payoff. For example, researchers used a dedicated 512-node computer for a 4.5 million grid point simulation analysis of the flow about a Delta II rocket, explaining an anomaly that had sent a launch vehicle into the wrong orbit. In another project, a heterogeneous parallel instrumentation, data collection, and visualization facility for wind tunnel testing uncovered previously unknown noise sources in the DC-10.
Optimizing complex multidisciplinary designs is also beginning to be possible thanks to distributed environments and toolkits such as DAKOTA and FIDO. DAKOTA has been used to improve chemical vapor decomposition reactors used in integrated circuit manufacturing.
Integrating advanced design technologies with production, quality control, marketing, and customer service systems is the focus of the Advanced Manufacturing Systems and Networking Testbed (AMSANT), an experimental facility for testing integration of distributed virtual manufacturing enterprises. Data exchange protocols are a necessary part of this infrastructure. Recently, the PlantSTEP Consortium developed the first such protocol for the process plant industries.
 |
The pressure field surrounding a Delta II rocket. |
 |
 |