Identifying and Quantifying Significant Uncertainties in Basin Modeling
Paul J. Hicks, Jr., Carmen M. Fraticelli, Martine J. Hardy, Jennifer D. Shosa, Michael B. Townsley, 2012. "Identifying and Quantifying Significant Uncertainties in Basin Modeling", Basin Modeling: New Horizons in Research and Applications, Kenneth E. Peters, David J. Curry, Marek Kacewicz
Download citation file:
Basin models are used to address a variety of questions concerning oil and gas generation, reservoir pressure and temperature, and oil quality. A large number of input parameters are required for a basin model, and many are functions of both space and time. Examples include isopach thicknesses and ages, amount of eroded/ missing section, rock properties (e.g., porosity, thermal conductivity), and heat flow and surface temperature boundary conditions. Most, if not all, of these model input parameters have associated uncertainties, and it can be difficult and time consuming to adequately quantify these uncertainties and propagate them through a basin model to assign error bars, probabilities, and risks to the output properties of interest.
In this chapter, we propose a workflow that allows a basin modeler to identify key input parameters and quantify and propagate uncertainties in these key input parameters through a model to evaluate the model results in light of a business question. We demonstrate this workflow using a hypothetical illustration in which uncertainties in key input parameters that control hydrocarbon generation, volumes, and timing are identified, quantified, and propagated through a basin model.
The workflow proposed in this chapter was designed to (1) identify the purpose(s) of the model; (2) develop a base-case scenario; (3) identify the input parameters whose uncertainty might affect the output property of interest; (4) perform screening simulations to identify the key input parameters; (5) evaluate the range of uncertainty in the key input parameters; (6) propagate the uncertainty in key input parameters through the model to the output properties of interest and estimate ranges of uncertainty for these input parameters; and (7) iterate as needed to fine tune the input parameters and dependencies between input parameters, fine tune error bars and weights for calibration data, and improve the base-case scenario.