Probabilistic Risk Analysis Projects

Yucca Mountain Project

INTERA has been involved in all aspects of probabilistic performance assessments for the Viability Assessment, Site Recommendation and the ongoing License Application phases for the Yucca Mountain Project. 

Our efforts include:

Conducting training courses on statistical uncertainty characterization for project scientists and engineers; and providing technical guidance and review during the distribution development process.

Building and testing abstracted models of sub-system performance (based on detailed process model results) that are combined to form the integrated total system assessment model.

Developing and implementing statistical tests for checking stability of model response (sensitivity to sample size) and reliability of model results (error bars on model predictions).

Development and application of statistical sensitivity analysis methodologies (stepwise regression analysis, classification tree analysis, entropy-based significance analysis) to determine key drivers of risk in projected outcomes of sub-system and total system response.

Probabilistic Risk Assessment

INTERA developed a probabilistic risk assessment methodology for geologic radioactive waste disposal at a generic site.  We evaluated the pros and cons of several alternative methods to the commonly used Monte Carlo simulation for quantifying uncertainty in model outcomes.  These methods include: first-order second moment and point estimate methods for data-limited and/or computation-intensive models, first-order reliability method for models where only a limited number of outcomes are of interest, and logic tree analysis to characterize input uncertainty using a few discrete states. 

Parameter Distribution Assignment 

INTERA developed a systematic framework for assigning probability distributions to uncertain input parameters in integrated system assessment models.  Techniques considered include: data-driven distribution fitting, subjective probability assessment using maximum entropy methods, application of formal expert elicitation protocols, and Bayesian updating for combining different sources of information. 

Sensitivity Analysis for Safety Assessment 

INTERA is currently involved in the application of uncertainty and sensitivity analysis techniques for detailed process models that form an integrated safety assessment framework for Ontario Power Generation’s deep geologic repository project.  The application of uncertainty and sensitivity analyses techniques will facilitate the quantification of uncertainty bounds on predicted radiological and non-radiological consequences, as well as the identification of key processes and parameters governing disposal system response.