December 8, 2021 | 9:00 a.m. - 10:00 a.m. CDT (UTC-05:00)

Earth modeling involves generating 2D, or most often 3D digital representations of subsurface properties. Models of depositional facies, lithofacies, porosity, permeability and water saturation are typically built during hydrocarbon reservoir studies. These models can be used to estimate hydrocarbon volumes in place and can be sent to a flow simulator to forecast hydrocarbon production, and make important decisions about reserves booking, field development plans, or well drilling locations.  

Earth modeling is a data integration problem: the geomodeler needs to reconcile direct, but sparse, measures of the property to be modeled (well log data) along with abundant, but indirect, measures (seismic and production data), within a given context (depositional environment). This reconciliation calls for a combination of statistical data analysis and data interpretation technologies. Sophisticated technologies have been developed over the last forty years to capture more and more geological details; they have greatly improved the geological realism of models, providing pretty pictures for presentations, but have made earth modeling an increasingly complex and time-consuming exercise. Yet not all geological details have an impact on reservoir volume or flow performance forecasts. It is important for practitioners to identify the modeling parameters that impact decision-making, so that they can focus their time and effort on assessing, and, if possible, reducing the uncertainties associated with those parameters. Based on this sensitivity analysis, an ensemble of Earth models spanning the range of the major subsurface uncertainties should be built to estimate reservoir volume and flow performance uncertainty in order to communicate geological risks to decision-makers. 

In this talk, after providing a brief history of Earth modeling, and describing the main current geostatistical technologies, I will share best practices and make general recommendations for effective Earth modeling.