Evaluation of the DEEPSOIL Software on the DesignSafe Cyberinfrastructure
Publication: Journal of Geotechnical and Geoenvironmental Engineering
Volume 143, Issue 9
In seismic site response analysis, uncertainties in both soil properties and earthquake ground motions are typically assessed in a probabilistic manner by conducting Monte Carlo simulations for a large number of combinations of soil profiles and ground motions (e.g., Pehlivan et al. 2016) and can be computationally costly. The DesignSafe cyberinfrastructure (www.designsafe-ci.org, Rathje et al. 2017) component of the NSF-supported Natural Hazards Engineering Research Infrastructure (NHERI, Grant No. CMMI-1520817) aims to address the computational needs of researchers by providing a cloud-based computational platform for engineering simulations on the Stampede supercomputer at the Texas Advanced Computing Center (TACC) located at the University of Texas at Austin.
One of the first simulation tools to be evaluated for deployment as part of the DesignSafe cyberinfrastructure is DEEPSOIL. The DEEPSOIL software is an equivalent linear and nonlinear one-dimensional (1D) seismic site response analysis tool widely used in both academia and engineering practice that has been in active development at the University of Illinois at Urbana–Champaign since 1998.
The trial deployment of DEEPSOIL on Stampede was evaluated using 1.7 million analyses and a smaller subset of 73,000 analyses from the Next Generation Attenuation Relationships for Central and Eastern North America project (NGA-East) (Harmon 2017). This set of simulations was the largest study of its kind to date in the field of geotechnical earthquake engineering.
Tables 1 and 2 compare the runtimes for the NGA-East analyses on Stampede with the runtimes of the same set of analyses on a small cluster of servers with hardware similar to the computing nodes of Stampede. On average, each analysis completed 16.8% faster on Stampede as a result of slightly faster processors, a reduction in overhead coordinating thousands of concurrent analyses, and a reduction in the amount of data being read from and written to disk. By scaling the study and distributing the analyses to 4,096 CPUs (running 4,096 concurrent analyses), the total wall time required for the simulations was reduced from approximately 11 days to 4.5 h (a 98% reduction).
System metric | Number of CPUs | Runtime (core-hours) | Wall time (days) | Runtime speedup (%) | Wall time speedup (%) |
---|---|---|---|---|---|
Cluster | 8 | 1,686 | 8.5 | — | — |
Stampede | 64 | 1,540 | 1 | 8.7 | 88.2 |
System metric | Number of CPUs | Runtime (core-hours) | Wall time (days) | Runtime speedup (%) | Wall time speedup (%) |
---|---|---|---|---|---|
Cluster | 94 | 20,767 | 11 | — | — |
Stampede | 2,048 | 17,715 | 0.36 | 14.7 | 96.7 |
4,096 | 17,272 | 0.18 | 16.8 | 98.4 |
Implications
The availability of DEEPSOIL on the DesignSafe computing platform will allow researchers to conduct unprecedentedly large-scale simulations. The computational cost of site response simulations is dramatically reduced, allowing researchers and practitioners to focus on designing suites of simulations to capture the effects of variability and uncertainty of soil and ground motion properties and on understanding the role uncertainty has on site response and amplification for use with reliability-based design standards. Additional evaluation is currently underway to determine how DEEPSOIL results can be efficiently processed with the data analysis tools available on DesignSafe to enable rapid adoption of high-performance computing by researchers and practicing engineers in the near future.
References
DEEPSOIL version 6.1 [Computer software]. Board of Trustees of Univ. of Illinois at Urbana–Champaign, Urbana, IL.
Harmon, J. A. (2017). “Nonlinear site amplification functions for central and Eastern North America.” Ph.D. thesis, Univ. of Illinois at Urbana–Champaign, Champaign, IL.
Pehlivan, M., Rathje, E. M., and Gilbert, R. B. (2016). “Factors influencing soil surface seismic hazard curves.” Soil Dyn. Earthquake Eng., 83, 180–190.
Rathje, E. M., et al. (2017). “DesignSafe: A new cyberinfrastructure for natural hazards engineering.” Nat. Hazards Rev., 06017001.
Information & Authors
Information
Published In
Copyright
©2017 American Society of Civil Engineers.
History
Received: Jan 28, 2017
Accepted: Mar 21, 2017
Published online: Jun 10, 2017
Published in print: Sep 1, 2017
Discussion open until: Nov 10, 2017
Authors
Metrics & Citations
Metrics
Citations
Download citation
If you have the appropriate software installed, you can download article citation data to the citation manager of your choice. Simply select your manager software from the list below and click Download.