The BEMUSE project (Best-Estimate Methods – Uncertainty and Sensitivity Evaluation) is an international project coordinated by the OECD (Organisation for Economic Co-operation and Development) in which IRSN is involved.
Launched in 2003, BEMUSE focuses on the application of “best-estimate” methods in the event of a Large-Break Loss of Coolant Accident (LB-LOCA) in the primary system of a nuclear power plant. The project involves international experts of nine different nationalities from twelve different organisations (institutes and universities).
Within the framework of nuclear power plant safety demonstration, a large number of studies are underway regarding the types of accidents liable to occur during the operating life of a plant, focused on effective performance demonstration and the design basis of safety systems. Such studies make use of a wide variety of computational codes. The mathematical models implemented in these codes serve to approximate the physical phenomena entailed in the accidents studied. Consequently, the results are not precise but rather subject to uncertainty. In order to take account of such uncertainty in safety studies, two approaches or methods are used:
Using the conservative approach, the consequences in terms of safety are maximised by selecting conservative values for uncertain physical models and for initial and boundary conditions. These conservative values are selected according to sufficiently large margins to cover any uncertainty and to ensure as far as possible that the calculated consequences will be that much greater, from the point of view of safety, than the actual consequences in the case of a real accident.
Using the “best-estimate” approach, design basis accidents are studied in a more realistic manner using “best-estimate” codes which approximate real situations as closely as possible. The results obtained cannot be used in safety studies as they stand, because of the uncertainty entailed in the calculations. This is why it is essential to evaluate and quantify the degree of uncertainty.
For IRSN, the “best-estimate” approach provides a greater level of confidence when it comes to transposing physical models, determined on an experimental basis, to a reactor, thanks to more comprehensive physical knowledge of the result.
The main objectives of the BEMUSE programme are:
to assess the quality and reliability of “best-estimate” methods covering uncertainty evaluations in applications relative to nuclear reactor safety at nuclear power plants,
to develop a common understanding of existing methods shared by all the organisations involved in the project,
to promote the use of “best-estimate” methods by the safety authorities and in the nuclear industry worldwide.
The programme is divided into three main stages:
Evaluating uncertainty associated with calculations for the LOFT L2-5 test. LOFT (Loss of Fluid Test Device) was an American test loop built in the 1970s in Idaho Falls, which simulated loss of coolant in a PWR primary system. This stage compared the different methods used to carry out “best-estimate” studies. Generally speaking, the experimental values, such as system pressure or the temperature of the LOFT fuel rods, were restricted by calculations and their inherent uncertainty. The researchers involved in the project concluded that it was preferable to perform over 59 calculations (minimum number of calculations according to Wilks’ Formula) so as to evaluate uncertainty and carry out sensitivity analyses.
Sensitivity and uncertainty evaluation analyses in the case of “ZION” US PWR with similar characteristics to reactors built in France. ZION, which was shutdown in 1998 following 25 years in service, was mainly chosen for two reasons: the LOFT loop represents this type of reactor on a scale of 1/50 and the geometric data used to for the simulation are available to all the parties involved. The aim of this stage is to discover the degree of transposition between the results obtained for the test loop and the results for an actual reactor.
For IRSN, the aim of this stage is to answer the following two questions: Are the physical parameters that are deemed important for LOFT the same in the case of ZION? Is the uncertainty relative to the results greater for the ZION PWR than for the LOFT test? Etc.
Drawing up a report that will present the different method and approaches, together with conclusions and recommendations regarding their implementation in the cases under study. This report will be written jointly by all the different groups of international experts.
The first stage of the programme was completed in 2006, the second stage is due to end in 2008 and the last stage will be completed by the end of 2009.
IRSN’s involvement in Stage 1 of the programme
During the first stage, IRSN applied its own “best-estimate” method. This entailed quantifying the uncertainties associated with the input parameters that impact on assessment results. These uncertainties were then propagated n times for random sampling of input parameters using different statistical methods. IRSN used three statistical methods as part of the BEMUSE project, using the CATHARE thermal hydraulics system code (590 calculations) applied to the data from LOFT: Wilks’ Formula, rank statistics and Bootstrap.
The information provided using rank statistics and the Bootstrap method for the response calculated using CATHARE, is much richer and fuller than that using the Wilks method. Thanks to this study, IRSN has developed new methods that are more effective than the Wilks Formula, which is usually used in the nuclear sector for statistical processing of uncertainty.
In the future, these methods may be used to perform independent calculations to assess safety demonstrations presented by nuclear facility operators, both in France and worldwide.