Reliable models for determining the pressure-volume-temperature PVT properties using artificial intelligence technique

In petroleum industry, PVT properties are very important in predicting the performance of oil and gas reservoirs. Several laboratory measurements are used to determine these properties at reservoir condition. The PVT measurements are costly and time-consuming; therefore, numerous correlations were developed to predict the PVT properties based on primary inputs such as reservoir pressure, temperature and hydrocarbon gravities. However, significant deviations are reported between the actual values and the predicted results. The aim of this paper is to present reliable and rigorous models to determine the PVT properties using artificial intelligence technique. Artificial neural network was utilized to develop the pressure-volume-temperature models. The proposed models estimate the PVT properties based on the pressure, temperature, oil and gas densities, and the solution gas-oil ratio. The developed models determine the bubble point pressure, formation volume factor and solution gas-oil ratio. Total of 250 data sets were used to develop and evaluate the model reliability. Average absolute percentage error (AAPE) and coefficient of determination (R-value) were used to evaluate the model reliability. The new models are simple, accurate and easy to use when compared with the existing empirical equations. The obtained results showed that the developed models are able to determine the PVT properties with an average absolute error of around 7.7%. The intelligence models developed in this study outperform the popular PVT equations such as Standing and Al-Marhoun correlations. Average errors of 6.8% and 11.7% were obtained for the developed model and the popular PVT correlations, respectively. Therefore, the developed models can highly increase the quality of production management by providing an accurate estimation for the PVT properties. Also, the presented models can significantly reduce the time and cost required for conducting the PVT measurements.