Any attempt at modelling natural phenomena includes a number of numerical assumptions on which we have little or no constraints. Basin and forward stratigraphic modelling are methodologies that aim at reproducing the history of sedimentary basins and their internal complexities. As are all deterministic models, these models are characterized by the none-uniqueness of their results. Meaning very different models can equally honor the calibration data. The traditional approach to such a problematic is a Monte-Carlo approach which requires running 100s or 1000s of simulations in order to capture the uncertainty and quantify the risk on an output of interest (e.g. Source rock maturity, charge, source rock presence, reservoir presence…). Such a large number of simulations requires days or weeks of computation and is thus not suitable for the operational needs of the industry.
In this presentation we will introduce an innovative approach for an efficient and effective sensitivity and risk analysis using response surface modelling. This method requires a small number of simulations, out of which a response surface can be constructed to mimic the behavior of the calculator. The predictivity of the response surface is checked with confirmation runs. The response surface can then be interrogated and producing thousands of results instantly for a thorough and quick sensitivity and risk analysis.
This method can be applied to basin and forward stratigraphic modelling and the analysis can be done on maps (whole model or per interval), along planned well paths (vertical or deviated), and in scalar mode.