Customer Interview
Danielson studies Structural Defeat Charges
CAP Project uses new HPC hardware capabilities for complex problem
 |
Dr. Kent Danielson Senior Computational Engineer Network Computing Services, Inc. |
Q: Name a project that used many HPC hours.
A: “HPC Simulations of Army Special Structural Defeat Charges” is a current HPCMP Capability Applications Project (CAP). (CAP projects are project selected to allow capability computational workload on the newest, largest HPCMP HPC systems before they enter the formal allications process. These projects have essentially dedicated time on the whole machine) I’ve been working for several years with ERDC-Geotechnica land Structures Lbaoratory in explosive-structural modeling for the design of such charges. I’ve primarily used my own finite element code, ParaAble, and have had to implement several features such as the Johnson-Holmquist-Beissel ceramic material model. In the past, we’ve typically not been able to reasonably use the desired level of mesh refinement and/or run for sufficient duration to capture all of the desired transient responses. Dedicated CAP access to thousands of fast processors at the ARL MSRC has overcome these limitations.
Q: That project’s impact on DoD?
A: The results of these simulations are being used directly by the Army Units engaged in current theaters of operations to assist in the design of improved explosive charges for their specific applications. Prior to actual use in the field, all new designs must ultimately be tested on full-scale structures. Thus, the numerical simulations help improve designs, but also reduce the number of costly tests.
Q: What was the project’s objective?
A: Generally, these Army Units want to defeat a broad range of specific structures using high explosive charges. Ideally, each charge will properly defeat the structure with minimal explosive material and charge weight for the warfighter to carry while also minimizing collateral damage. Numerical simulations permit “virtual” testing of a wide range of design concepts such as various charge placement strategies and ignition timings.
Q: What HPC systems here did you use?
A: The ARL IBM Opteron cluster, stryker, in its pioneer mode and will perform the other aspects of the CAP on stryker when it becomes production.
Q: How many computer hours did you use?
A: I don’t know exactly because the system doesn’t have a queuing system yet (I call it the tk batch system, since I give my files to Tom Kendall to run!). I’d guess that I’ve used 100,000 to 150,000 cpu hours since it went online in mid December. Under this CAP, I’ll probably use twice that much after stryker becomes classified. My analyses typically use 6,000 to 12,000 cpu hours each and run in the range of 2 to 3 TeraFLOPS on nearly the full 2,300 processors.
Q: Did you use the Sci Vis Lab?
A: So far, all of the visualization has been done by transferring the files to the ERDC MSRC Scientific Visualization Center, but the classified visualization is planned to be done remotely on the ARL SGI Origin, fcs, via the fastlane SDREN connection in the ERDC MSRC vault.
Q: Which personnel assisted you?
A: Tom Kendall and Steve Thompson.
Q: The highlight of your career?
A: It has most definitely been working at the ERDC under auspices of the Army HPC Research Center (AHPCRC). The current personnel and administration has been competent, supportive, and a pleasure with which to work and the access to top HPC facilities has been tremendous. The AHPCRC arrangement has permitted me to “think outside of the box” in developing capabilities that might not be possible in other positions. The projects at ERDC are challenging, exciting, and include considerable test data that helps to provide confidence in computational predictions. It is very rewarding to see large-scale numerical modeling help solve complex real-world problems for the Army.
Q: Hobbies?
A: Sailing, Sailboat Racing, Scuba Diving, and Traveling.
Dr. Kent T. Danielson has been sponsored at the Engineer Research and Development Center Geotechnical and Structure Laboratory (ERDC-GSL) for almost nine years by the Army High Performance Computing Research Center (AHPCRC). Under this arrangement, he has been a faculty member with the University of Minnesota and with Northwestern University, and is presently a Senior Computational Engineer with Network Computing Services, Inc. He has also taught at the WES Graduate Institute as an Adjunct Professor of Aerospace Engineering at Mississippi State University