Online Program

Design and Analysis of Computer-Simulated Experiments for Hydrocarbon Reserves Estimation (302722)

Keywords: Computer-simulated experiments, Latin hypercube designs, Multidimensional krigging, Optimal designs, Reservior modeling, probabilistic reserves curve, Statistical computing in R

In oil and gas industry, 3D reservoir modeling (3DRM) is performed to develop a spatial model of the fields for the purpose of estimating the ultimate recoverable (UR) reserves volumes. The estimates of UR are required for all economic decisions, from drilling wells to selling reserves. The 3DRM are developed based on the uncertain information on complex field geology on a fine grid. This leads to 3DRM that are so complex, a single run of the model takes weeks to generate a single UR value on the fastest computer in existence. The 3DRM are modern-day computer-simulated experiments (CSE) that have led to the revitalization of classical experimental design and analysis methods. As CSE are free from natural variations, we need to first develop an error structure for modeling and then select an adequate design and analysis methodology. In this talk, we will present the framework for estimation of probabilistic reserves volumes through 3DRM incorporating uncertainties in field parameters and challenges we faced in implementing this methodology for reserves estimation in Western Australia.