Massive Parallel Computation of Non-Linear Multi-Field Problems in Terrestrial Systems
Local Project ID:
HPC Platform used:
JUQUEEN of JSC
In past decades, high performance computing has become a valuable tool in many fields of environmental science and technology to utilize computational power for better characterization of the complexity of environmental systems as well as predicting their evolution in time. The challenge is to develop efficient numerical schemes and software implementations which can take full advantage of today’s supercomputing hardware (i.e. PetaFlop platforms). The JUQUEEN project „Massive parallel computation of non-linear multi-field problems in terrestrial systems“ strives to meet this challenge for the solution of thermo-hydro-mechanical-chemical problems in fractured porous media with applications in hydrology, geotechnical engineering and renewable energy resources (Kolditz et al. 2012a).
As an example a parallel computing application for the numerical simulation of two-phase flow processes in porous media is shown. The PETSc package is utilized for parallelization of the computational task in both the global assembly of the system of linear equations and the linear solver. In order to parallelize the global assembly of the linear equation system, the overlapping domain decomposition method is used. The present parallel FEM approach is realized within the framework of OpenGeoSys (www.opengeosys.org), an open source finite element code for numerical simulation of thermal, hydraulic, mechanical, and chemical processes in fractured porous media (Kolditz 2012b).
The computational efficiency of the approach has been tested with three examples of increasing complexity, the five spot benchmark, dense non-aquaeous phase liquid infiltration into an inhomogeneous porous medium, and a real-world application to the CO2 storage research site: Ketzin, in Germany, which is operated by the Helmholtz Centre for Geosciences in Potsdam.
Supercomputing was definitely required to solve the application problem for CO2 storage. Up to 15,000 CPUs have been used for computing on the JUQUEEN system. The interdisciplinary team from the Helmholtz Centre for Environmental Research UFZ in Leipzig comprises scientists from different fields, geologists, and numerical modellers as well as IT experts. The project goal was not just the parallel computation but developing a complete work flow for modelling including data integration and visualization. Scientific visualization is particularly important for supercomputing application for multiple purposes, visual validation of large data sets and their visualization together with simulation results (Bilke et al. 2014).
The next steps of the JUQUEEN project HLZ23: Massive parallel computation of non-linear multi-field problems in terrestrial systems are dealing with uncertainty analysis, reactive transport processes and hyperresolution simulation for various environmental applications.
Bilke L, et al. (2014): VISLAB – Laboratory for Scientific Visualization. Environ. Earth Sci., Thematic Issue on “Environmental Visualization”, submitted.
Wang W, Fischer T, Zehner B, Boettcher N, Goerke U-J, Kolditz O (2014): Parallel finite element solver for modeling two-phase flow processes in porous media: OpenGeoSys with PETSc. Environ. Earth Sci., DOI: 10.1007/s12665-014-3576-z.
Kolditz O, Shao H, Görke U-J, Wang W (eds) (2012a): Thermo-hydro-mechanical-chemical processes in fractured porous media. Lecture Notes in Computational Science and Engineering, Vol. 86, Springer, Heidelberg, ISBN 978-3-642-27176-2, DOI: 10.1007/978-3-642-27177-9.
Kolditz O., Bauer S., Bilke L., Böttcher N., Delfs J.O., Fischer T., Görke U.J., Kalbacher T., Kosakowski G., McDermott C.I., Park C.H., Radu F., Rink K., Shao H., Shao HB., Sun F., Sun Y.Y., Singh A.K., Taron J., Walther M., Wang W., Watanabe N., Wu N., Xie M., Xu W., Zehner B. (2012b): OpenGeoSys: an open-source initiative for numerical simulation of thermo-hydro-mechanical/chemical (THM/C) processes in porous media. Environ. Earth Sci., 67(2): 589-599, DOI: 10.1007/s12665-012-1546-x.
Prof. Dr.-Ing. habil. Olaf Kolditz (TU Dresden)
Department of Environmental Informatics (ENVINF)
Head of the Helmholtz Graduate School (HIGRADE)
Environmental Earth Sciences | Office Leipzig (EES)
Helmholtz Centre for Environmental Research (UFZ)
Permoserstraße 15, D-04318 Leipzig/Germany