Computational methods to assess CO2 storage concepts can be divided into two broad categories. Reservoir simulator models are generally used to evaluate the complex physical-chemical and fluid flow requirements within the storage formation (e.g., coal beds, oil reservoirs, etc.). These numerical models are, however, computationally too intensive to be practical to quantify the uncertainty associated with potential leakage of CO2 into and within the geosphere surrounding the proposed reservoir. To solve this type of process-driven problem, a methodology known as probabilistic risk assessment (PRA) was developed. In essence, the complexity and the uncertainty associated with the geosphere characteristics are subsumed into statistical functions. Model parameters can be defined as probability distribution or density functions (PDF) and Monte Carlo methods can then be used to randomly sample from the parameter PDFs. This is followed by the use of conventional statistical methods to provide a parameter sensitivity analysis and to quantify the uncertainty in model predictions.
Project completed March 2007
Marsha I. Sheppard