Unified Equation of State Tuning Strategy for Large PVT Sample Inventory

Abstract Development of a reservoir fluid description based on a single PVT study is usually a straightforward task. The availability of multiple samples complicates the equation of state (EOS) tuning process significantly and it is not intuitively obvious which studies to assign the most importance. We devise a general strategy for development of a unified EOS model, which applies for any number of PVT studies and test the approach in a reservoir from which more than 50 studies have been performed over time. The proposed methodology starts with a thorough data QC consisting of Hoffman plots, compositional trends versus MW, mass balance checks, liquid dropout tail adjustment, followed by cross-plots such as C1-Psat, Rs-Psat and others, but put into a regional context involving data from nearby reservoirs and fields. The combination of all these steps help to identify outliers that are then discarded from the subsequent EOS tuning process. A method involving tracking various mole fraction ratios help to determine if samples may have been contaminated by injection gas. While the EOS tuning aims to assign more weight to more recent and also more comprehensive studies, the included samples span the entire compositional range and are spread out across the reservoir. One of the key uncertainties for the reservoir investigated in this study is whether the observed compositional variation is vertical or lateral, or perhaps both. To augment the data set used for the fluid initialization process, fluid compositions can be derived from producing GOR and MDT pressure data. The production data analysis suggests the presence of more than one distinct lateral depth trend across the reservoir, although an explanation for this observation is not currently available.