A New and Practical Oil-Characterization Method for Thermal Projects: Application to Belridge Diatomite Steamflood

Abstract

Summary Most of the oil-characterization approaches for thermal recovery are designed for heavy oils at moderate temperatures, in which oil can be represented in very simplistic ways (such as “gas” and “oil”). However, when oil is exposed to very high steam temperatures (i.e., 550°F), and/or the oil is lighter than the classical range defined for heavy oils and is exposed to a wide spectrum of thermal effects, such as distillation of the lighter ends, the conventional methods of representing the interaction of steam and the in-situ fluids are not accurate. In many cases, we have to first evaluate the quality of the data, and then represent the average behavior with a single most likely fluid model per reservoir segment (plus other scenarios, as needed) to simulate the production performance. There is a need to develop a streamlined approach to bring such data into industrial simulators in a practical way. In this study, we have developed a fit-for-purpose approach to generate a consistent pressure/volume/temperature (PVT) model over the whole reservoir, reflecting both pressure and temperature changes through the entire oil accumulation. The model represents the oil viscosity for a wide spectrum of temperatures, from reservoir temperature to steam temperature (thermal-process range). A systematic lumping scheme enables conversion of the characterized PVT model for numerical simulators with the minimal number of pseudocomponents while still capturing the essence of thermal physics. To our knowledge, there is no systematic study of this nature available in the literature. We have tested this approach in the Belridge diatomite steamdrive project. The study confirmed that steamflood incremental oil production in a light-oil reservoir is sensitive to the component-lumping scheme because of distillation of the lighter ends. We also found that a five-component PVT model, representing the physics in “fit-for-purpose” dynamic simulation, best compromises between minimal number of components and physical description of the light-oil behavior.