Prediction of Groundwater Pollution Using Multiprocessor Computers
Tech Area / Field
- ENV-MRA/Modelling and Risk Assessment/Environment
3 Approved without Funding
VNIIEF, Russia, N. Novgorod reg., Sarov
- IGE (Institute of Geoecology) / Institute for Geoecology (Saint-Petersburg Branch), Russia, St Petersburg
- Los-Alamos National Laboratory / Earth and Environmental Sciences Hydrology & Geochemistry, USA, NM, Los-Alamos\nUniversity of California / Department of Mechanical and Aerospace Engineering, USA, CA, La Jolla
Project summaryIn many regions of the Earth, groundwater constitutes the major and often the sole source of domestic and drinking water supply. Pollution of the Earth surface and open water bodies often poses environmental risks to the public residing not only in the vicinity of the source of pollution, but also in remote localities, because once they get into groundwater, pollutants actively migrate and are transported at far distances.
The common way of water quality monitoring in aquifers is to take samples and analyze these at regular intervals at inpidual stations of a monitored network of observation wells. This kind of monitoring has rather limited capabilities, as it does not predict conditions of groundwater basins, such as the levels of hazardous components, location of polluted zones in the subsurface space, spread of local pollution etc. This hampers efficient administrative decision making with respect to environmental protection, especially under emergency conditions related to the localization of a sudden discharge and recovery of polluted territories.
Numerical modeling makes it possible to assess the risk of environmental pollution and to semantically evaluate the environmental status and, consequently, reveal the most hazardous sources of pollution and rank them with respect to the hazard they pose to certain environmental objects.
However, a number of environmental geology problems cannot be modeled numerically on up-to-date single-processor computers, and they require multiprocessor computing, e.g.:
- Migration of pollutants that may consist of hundreds of components in groundwater. In such problems, separate differential equations of transport, dispersion, sorption and chemical kinetics need to be solved for each component. The domain of interest is rather large, and a difference mesh composed of dozens of millions of cells needs to be generated in order to provide the needed accuracy of prediction.
- Virtual environmental geology models for the regions with hundreds of operating industrial facilities that constitute the sources of groundwater pollution can be developed and operated only on multiprocessor computers.
- Uncertainty of initial data in underground hydromechanics (geometry and parameters of strata) can be taken into account only by self-learning expert systems that use geostatistics and neural networks. Such an approach however requires environmental geology problems to be run with varied initial data hundreds of times, which cannot be done in affordable time frames on single-processor computers.
In the recent years, under ISTC Projects 714-1998 and 1656-2001, a team of VNIIEF scientists jointly with specialists from NIIMM at Kazan University and the St.Peterburg branch of IGE RAS have developed a software product, NIMFA, which is intended for solving 3D environmental geology problems. Another team of VNIIEF specialists have developed a library of parallel and serial solvers for large sparse linear systems, PMLP/ParSol, under a lab-to-lab project (LANL – VNIIEF, 1997) and an NCI project (Parallel Mathematical Library Project, 2000).
The objective of this project is to develop efficient methods to parallelize hydroecology problem algorithms and to develop a highly parallel software suite to solve such problems based on the NIMFA code developed under ISTC Projects 714-1997 and 1565-2001 and the PMLP/ParSol library of parallel and serial solvers for large sparse linear systems developed under lab-to-lab (LANL – VNIIEF, 1997) and NCI (Parallel Mathematical Library Project, 2000) projects.
The following major tasks will be accomplished under the project:
- Efficient methods for parallel computing of filtration problems will be developed. This task provides for the development of highly parallel algorithms to compute coefficient arrays for the system of linear equations resulting from the discretization of filtration equation using an implicit difference scheme on quasi-structured 3D difference meshes. In developing the algorithms, the major requirement to be met will be for the algorithms to run efficiently on both structured and unstructured meshes being polar particular cases of quasi-structured meshes.
- Develop a NIMFA - PMLP/ParSol coupled code. This task provides for the development of a multi-platform version of NIMFA, interface modification, development of coupling interfaces and testing of the new code.
- Multi-level paralleling. Under this task, algorithms will be developed and migration computations will be paralleled in components of physical processes, geometry fragments of domain (each having its own mesh) and computing cells. The implementation within NIMFA of the parallel algorithms developed as applied to environmental hydrogeology problems will allow their testing and improvement by running real problems.
- Numerical performance and efficiency analysis of the parallel algorithms. This task provides for efficiency analysis of the parallel algorithms developed by numerical experimentation and their comparison with other techniques.
The project involves scientists and specialists from the Russian Federal Nuclear Center VNIIEF, and scientists from the St. Petersburg Branch of the Russian Academy of Sciences’ Hydroecology Institute (SPb IGE RAS). This cooperation will combine unique capabilities and expertise of nuclear weapons scientists in the area of parallel computations and software development and the wide experience of SPb IGE RAS specialists in using domestic and foreign software products for solving hydroecology problems.
Project methodology rests upon the use of the most advanced multi-component filtration and parallel computing solutions for modeling various physical processes on multi-processor computers.
The project will make use of the experience of its participants in developing a serial multi-component filtration code and parallel heat conductivity and hydromechanics algorithms for multi-processor computers.
The technical approach and methodology of the project will therefore consist in accomplishing the major integrated tasks as follows:
Existing parallel algorithms to compute coefficient arrays for the system of linear equations resulting from implicit discretization of diffusion equation will be analyzed, algorithms to account for specific features of filtration equation and the differential-projection technique incorporated in NIMFA will be developed and implemented.
A NIMFA-PMLP/ParSol coupled code will be developed, filtration computations will be made completely parallel.
Parallel computing algorithms for multi-component pollutant transport in groundwater flow will be developed. It is expected to provide multi-level paralleling:
- in alternative computations in the expert system;
- in physical process components;
- in geometry fragments;
- in mesh cells.
The third task also provides for the implementation and testing of the algorithms developed in the NIMFA code.
Numerical reliability and efficiency analysis of the parallel algorithms developed will be carried out.
The project results have prospects of their further development and use in the following areas:
- development of virtual environmental geology models for regions under stress caused by human activities;
- computations of optimal protection geometry and properties for toxic and radioactive waste storage facilities, and optimal location of protective screens on pollutant migration routes;
- numerical prediction of the consequences of large-scale pollution caused by industrial accidents or acts of terrorism.
This project is in full compliance with ISTC goals and objectives:
- It provides VNIIEF nuclear weapons scientist and engineers opportunities to redirect their talents to peaceful activities related to construction, hydrogeology studies and environmental protection;
- It rests upon the most advanced parallel computing techniques in modeling filtration and heat and mass transfer processes in permafrost rock types and thus promotes integration of nuclear weapons scientists into the international scientific community.
The project has certain commercial potential related to the demand for efficient environmental monitoring tools.
The International Science and Technology Center (ISTC) is an intergovernmental organization connecting scientists from Kazakhstan, Armenia, Tajikistan, Kyrgyzstan, and Georgia with their peers and research organizations in the EU, Japan, Republic of Korea, Norway and the United States.
ISTC facilitates international science projects and assists the global scientific and business community to source and engage with CIS and Georgian institutes that develop or possess an excellence of scientific know-how.