Over the years, oil and gas industries have been heavy users of HPC machinery and a driving force towards faster and more energy efficient computers. The consumption of HPC cycles in this industry is dominated by seismic processing, that selects promising locations of hydrocarbons reservoirs, promoted to candidates for exploratory drills.
Seismic processing input is a large survey of seismic data acquired over a target area. Seismic processing output is a set of images of the subsurface, containing locations and properties of geological layers that best fit survey data. Seismic data processing currently demands months of state-of-the art computing machinery.
Obviously, faster computers of the Exaflop era will accelerate this process under the same algorithms. But they will also allow the use of more computationally demanding algorithms, required to produce high quality images of complex geologies. They may also reduce the risk inherent to exploratory drills either by estimating the process uncertainty or by accelerating the evaluation of possible scenarios.
HPC4E works in three directions to prepare the oil and gas industry for the Exaflop era.
First direction is to produce more precise images by accelerating demanding kernels that solve various forms of the wave equation. Classical finite differences and promising finite elements kernels are optimized for speed on potential Exaflop architectures. New finite element methods are developed and optimized. The optimized kernels and the optimization procedures will be incorporated into the oil and gas industry production software for further testing.
Second direction is to develop a comprehensive benchmark suite of demanding geological models. These models will be the testbed of the optimization procedures of the first direction on the speed and quality of imaging software.
Third direction is the development of uncertainty quantification methods to enhance the exploration risk evaluation procedure.
So far, project milestones were achieved at due time. Target kernels were optimized and the optimization methods are being incorporated into the industry codes. They will be tested on the suite of geological models already produced. Promising uncertainty quantification procedures were developed and are currently being tested.
For further details, please refer to:
- Zanon Boito, F., Bez, J. L., Dupros, F., Dantas, M. A. R., Navaux, P. O. A., Aochi, H., "High Performance I/O for Seismic Wave Propagation Simulations". In 2017 25th Euromicro International Conference on Parallel, Distributed and Network-Based Processing (PDP). DOI: 10.1109/PDP.2017.92
- Camata, J.J., Rossa, A., Cortes, A., Coutinho, A.,: "Sedimentation and Deposition Simulation with LibMesh". In 2017 SIAM Conference on Computational Science and Engineering. Minisymposium 19 Applications and Computational Strategies for Finite Element Computations using LibMesh - Part I of II. Presentation (PDF)
- Silva, V., Camata, J.J., Mattoso, M., Valduriez, P., Coutinho, A.,: "In Situ Data Steering with Provenance Data". In 2017 SIAM Conference on Computational Science and Engineering. Minisymposium 19 Applications and Computational Strategies for Finite Element Computations using LibMesh - Part II of II Presentation (PDF)
- Kormann, J., Rodriguez, J E., Ferrer, M., Gutierrez, N., de la Puente, J., Hanzich, M., Cela, J.M., "Towards Automatic Full Waveform inversion: Synthetic Study Cases". In The Leading Edge (in press), Society of Exploration Geophysicists, 2016. Repository link: http://upcommons.upc.edu/handle/2117/103197 DOI: 10.1190/tle35121047.1
- Kormann, J., Rodriguez, J E., Ferrer, M., Gutierrez, N., de la Puente, J., Hanzich, M., Cela, J.M., "Elastic Full Waveform Inversion (FWI) of Reflection Data with a Phase Misfit Function". In Communications in Computer and Information Science, Springer, 2016. Repository link: http://upcommons.upc.edu/handle/2117/85902 DOI: 10.1007/978-3-319-32243-8_19
- Lavril, T., Mattoso, M., Silva, V., Costa, D., Rochinha, F., Miras, T., Coutinho, A., “Controlling Parallel Adaptive Sparse Grid Stochastic Collocation Simulations with Workflows”. In XXXVII Ibero-Latin American Congress on Computational Methods in Engineering (CILAMCE2016), Brasilia, DF, Brasil, 2016.
- Nasciutti, T.C., Panetta, J.: Impacto da Arquitetura de Memória de GPGPUs na Velocidade da Computaçâo de Estênceis. In XVII Simpósio em Sistemas Computacionais de Alto Desempenho (WSCAD 2016). Aracaju (Brazil). Repository link: http://www2.sbc.org.br/ceacpad/ijcae/
- Kormann, J., Rodriguez, J. E., Ferrer, M., Farrès, A., Gutierrez, N., de la Puente, J., Hanzich, M., and Cela, J. M., “Acceleration Strategies for Elastic Full Waveform Inversion Workflows in 2D and 3D”, accepted in Computational Geosciences. 2016. DOI: 10.1007/s10596-016-9593-0
- Böhm, C., Hanzich, M., de la Puente, J., and Fichtner, A., “Wavefield compression for adjoint methods in full-waveform inversion”, Geophysics, vol. 81. SEG, pp. R385-R397, 2016. Repository link: http://upcommons.upc.edu/handle/2117/103434 DOI: 10.1190/geo2015-0653.1
- Castillo-Reyes, O., De la Puente, J., Puzyrev, V., Cela, J.M., Edge-based parallel framework for the simulation of 3d CSEM surveys. In International Conference and Exhibition, Barcelona, Spain, 3-6 April 2016 SEG Global Meeting Abstracts, 2016. DOI: 10.1190/ice2016-6274394.1
Jairo Panetta - ITA (Instituto Tecnológico de Aeronáutica)
This article also appeared in the LinkedIn Group. Join us!
Other LinkedIn articles: