RESPONSE SURFACE METHODOLOGY FOR SENSITIVITY AND UNCERTAINTY ANALYSIS - PERFORMANCES AND PERSPECTIVES
Recent developments of response surface methodology (RSM) in safety analysis have shown its effectiveness in studying accident processes and enlightened its performances, mainly with respect to: - Sensitivity analysis. The method exploits the engineering expertise and the inferential capability of RSM in order to select and to rank the variables which affect the consequences of a given accident. - Modelling. Polynomial and inverse polynomial response surfaces and appropriate sequential experimental designs enable the approach to complex accident situations, as those described by a strongly asymmetric code response. - Uncertainty analysis. Numerical convolution and Monte Carlo techniques allow the uncertainties on the input variables to be propagated through the response surface. As perspectives, specialized subdesigns can be added to the previous designs, if the modelling results are not sufficiently adequate. This procedure can be exploited in order to explore extreme regions of the variable space, where the expected code behaviour could present strong non linearities.
Bibliographic Reference: PROBABILISTIC SAFETY METHODS AND APPLICATIONS, SAN FRANCISCO (USA), FEB. 24-28, 1985 WRITE TO CEC LUXEMBOURG, DG XIII/A2, POB 1907 MENTIONING PAPER E 31734 ORA
Availability: Can be ordered online
Record Number: 1989123004100 / Last updated on: 1987-01-01
Available languages: en