Evaluation of Crude Oil Thermophysical Properties Utilizing PVTp Software

الملخص

Reservoir fluid characteristics play a critical role in reservoir engineering calculations. While direct laboratory measurements provide the most reliable source for these parameters, such data are often unavailable or too expensive to obtain. Under these circumstances, researchers conventionally use empirical correlations to estimate the required fluid properties. Over the past decade, researchers have increasingly applied artificial neural networks to improve the precision of PVT correlations. In this study, the performance of the most prominent empirical PVT correlations available in the literature was evaluated using 50 laboratory PVT reports from local crude oils. The findings revealed that existing correlations failed to provide accurate predictions for the PVT properties of the XX reservoir. Consequently, this research utilizes two distinct approaches to predict these parameters. New empirical PVT correlations were specifically formulated for the XX reservoir oils based on readily available field production data. Using non-linear multiple regression analysis via the Statistical Analysis System (SAS), correlations were developed for bubble point pressure, solution gas-oil ratio at bubble point, oil formation volume factor at bubble point, dead oil viscosity, saturated oil viscosity, undersaturated oil viscosity, and undersaturated oil compressibility. A comprehensive dataset comprising more than 50 PVT reports collected from 10 distinct oil fields within the XX reservoir was utilized to build these models. The results demonstrate that the newly proposed correlations significantly outperform existing published models in estimating the PVT behavior of XX reservoir crudes.