Importance of emergency response actions to reactor accidents within a probabilistic consequence assessment model

An uncertainty and sensitivity analysis of early health consequences of severe accidents at nuclear power plants as a function of the emergency response parameters has been performed using a probabilistic consequence assessment code. The importance of various emergency response parameters in predicting the consequences for a range of accident source terms was determined through training a neural network algorithm which relates the sensitivity of the output to various choices of the input. Extensions of this approach should be helpful to planners in prioritizing the emergency responses at nuclear power plants.