Quantification of uncertainty in continuous and discrete parameters describing subterranean regions, employs modeling, sensitivity analysis and risk analysis

摘要

A factorial part of experimental mapping is formed by folding part of the factorial plan for quantitative factors and attributing at least one modality of a quantitative factor, to each of the blocks formed by folding. Results are then analyzed, combining a sensitivity analysis and a risk analysis, introducing marginal- and global models. The axial part of the qualitative factors is determined, following a D-optimization criterion. Sensitivity analysis using marginal models, is carried out to detect terms influencing each scenario. Global models are employed to detect terms globally-influencing the entire group of scenarios. Risk is analyzed, predicting response at points over an interval of prediction for a set of fixed parameter values, using the global model. Risk is analyzed, predicting responses, commencing with a large set sets of values of parameters selected randomly in the range of variation. The global model is employed when one or more discrete parameters cannot be monitored or controlled. Marginal models are used to analyze risk when each discrete parameter can be monitored or controlled and the global model has not detected particularly influential quantitative-quantitative interaction which is undetectable by the marginal models. The global model is used for risk analysis when each discrete parameter can be monitored and that the global model has detected a strongly-influential quantitative interaction, which is undetectable by the marginal models. Global and marginal models are used to determine respective aliased influences of quantitative-quantitative and quantitative-qualitative interactions.