| Popis: |
The standard method for propagation of uncertainty in a risk analysis requires rerunning the risk calculation numerous times with model parameters chosen from their uncertainty distributions. This was not practical for the Cassini nuclear safety analysis, due to the computationally intense nature of the risk calculation. A less computationally intense procedure was developed which requires only two calculations for each accident case. The first of these is the standard “best-estimate” calculation. In the second calculation, variables and parameters change simultaneously. The mathematical technique of deconvolution is then used to separate out an uncertainty multiplier distribution, which can be used to calculate distribution functions at various levels of confidence. |
