DATA UNCERTAINTY IN LONG-TERM PREDICTION. PRESENT TRENDS IN RISK ANALYSIS WITH MONTE CARLO TECHNIQUES
As the analysis of the risk linked to the existence of radioactive waste repositories into geological formations must extend over geological time periods, the parameters and models utilised are unavoidably affected by uncertainties. Uncertainties due to incomplete model descriptions or approximations can be overcome by systematically adopting the most conservative idealization of a possible evolution scenario. On the other hand uncertainties due to scarce knowledge of parameter values require a specific treatment if the risk has to be quantified. A methodology recently applied to Waste Disposal Performance Assessment involves the use of probability distributions for the uncertain parameters. Such distributions, which account for intrinsic parameter variability as well as for their degree of uncertainty, are then fed into a computer code able to produce a distribution of the model results. In the Monte Carlo approach an empirical distribution function for the output variable (e.g. a dose rate to man) is obtained by repeatedly sampling the input variables from their distributions and running the model for each set of input values. In this note some of the statistical techniques usually applied in the various steps of the Monte Carlo Analysis will be briefly illustrated.
Bibliographic Reference: WORKSHOP ON MATHEMATICAL MODELLING FOR RADIOACTIVE WASTE REPOSITORIES, MADRID (SPAIN), DEC. 11-12, 1986 WRITE TO CEC LUXEMBOURG, DG XIII/A2, POB 1907 MENTIONING PAPER E 33047 ORA
Availability: Can be ordered online
Record Number: 1989125076900 / Last updated on: 1987-12-01
Available languages: en