ANDRA and Inria sign a partnership agreement on digital simulation
On 14 September 2009, Marie-Claude Dupuis, the CEO of the French National Radioactive Waste Management Agency (ANDRA) and Michel Cosnard, Chairman and CEO of the National Institute for Research in Computer Science and Control (Inria) signed a partnership agreement with particular emphasis on digital simulation.
The challenge behind digital simulation , and the core subject of ANDRA's research for many years, is to represent the changes in phenomena occurring in a radioactive waste storage facility over periods ranging from a century to a million years in order to design the safest possible storage site.
For Inria, such a partnership illustrates the degree to which the contribution from computer science proves decisive in responding to many scientific issues, especially in the field of sustainable development. Involving its own research teams alongside ANDRA experts shows the significance of digital simulation computing tools, these being the only way to analyse highly complex processes over broad swathes of both time and space.
The partnership agreement signed on 14 September 2009 by Marie-Claude Dupuis, the CEO of ANDRA, and Michel Cosnard, Chairman and CEO of Inria, will allow the two organisations to jointly develop research activities in three area s:
1. Simulation of physical processes
To simulate physical processes (or "phenomena"), physical, mathematical and computer models enabling them to be represented need to be identified. Such representations of the phenomena occurring in a storage facility (and the surrounding geological environment) must also include any interaction (or combination) there may be between different phenomena.
This means acquiring knowledge about the phenomena (physical models) and then translating them through conceptualisation into mathematical models. These mathematical models must then be digitised using existing or specially developed calculation codes. The phenomena studied by ANDRA as regards managing radioactive waste are mainly water flow in porous media, the release of radionuclides by waste packages and the migration of these radionuclides. However, also included are all the other thermal, mechanical, chemical and hydraulic phenomena that, at different times, affect storage and come to play a greater or lesser role in the release and transfer of radionuclides.
To do so, the calculation codes must allow computer models to accurately include increasing amounts of knowledge about these phenomena. But the more phenomena and interactions are added, the more difficult and complex the mathematical object is to represent. This will therefore be one avenue of progress for ANDRA and Inria to explore.
2. Digital solution methods
This second area, following logically from the first, involves the methods used to digitise mathematical models. This, in a way, means solving mathematical equations in a faster time and smaller space. These methods make use of high-performance computing, making it possible to increase representation capacity (mesh size, data variability, etc.) and improve result quality (robustness and accuracy). For some more complex mathematical models, special computing methods need to be developed.
The topics dealt with under the partnership include parallel processing (running calculations at the same time to gain speed), partitioning (concentrating machine calculations just on those storage components where a situation develops in a given space and time) or adaptive meshing (taking account of the fact that the geometry of the objects comprising the storage facility will change over time).
3. Analysis methods
Since knowledge based on digital simulation is the result of calculations rooted in complex models representing reality, we need to be able to measure, firstly, any uncertainty surrounding parameter values and secondly the impact of such uncertainty on the results obtained. Several types of analysis method can be used, to put in perspective and rank all the data obtained or to be processed, and to confirm them relative to the phenomena in question for a radioactive waste storage facility. These methods (sensitivity analysis, uncertainty analysis) shed different light on the matter and complement each other.
Increasing complexity is being added to digital simulation of physical processes. ANDRA must therefore continue to develop new, ever more effective analysis methods, use of which may become widespread, including in the most complex situations. This work will therefore also be a key area in the research conducted jointly with Inria.
ANDRA is a publicly-funded industrial and commercial institution established by the French law of 30 December 1991. Its remit was expanded by planning legislation passed on 28 June 2006 on sustainable management of radioactive matter and waste. Independent of radioactive waste producers, ANDRA reports to the Ministries for Energy, the Environment and Research. ANDRA is tasked with the sustainable management of all French radioactive waste. It provides the French government with its expertise and skill to design management solutions, and to operate and monitor radioactive waste storage facilities, protecting people and the environment against the effects of such waste over the short and long terms.
A public science and technology institution under the supervision of the ministries for Research and for Industry. Executives: Michel COSNARD, Chairman and CEO of Inria - Jean-Pierre VERJUS, Deputy Managing Director. Annual budget (2009): €200m, of which 21% is from its own resources. Regional research centres: Paris - Rocquencourt, Sophia Antipolis – Méditerranée, Grenoble - Rhône-Alpes, Nancy - Grand Est, Rennes - Bretagne Atlantique, Bordeaux – Sud Ouest, Lille - Nord Europe, Saclay - Île-de-France. 2,800 researchers, including more than 1,000 PhD students, working within more than 160 project-teams, the majority of which are shared with other bodies, Grandes Ecoles and universities. 790 ongoing research agreements. 79 associate teams around the world. 94 businesses started since 1984.