The knowledge of PVT properties is of a high importance in the petroleum industry, precisely, in oil and gas reservoirs development, engineering, and production. In the practice of petroleum industry, these properties are obtained by laboratory analysis from downhole fluid samples which are highly recommended to be taken in the early time of the reservoir prior to production to get representative reservoir fluid. It becomes a challenge when production is started as the fluid composition in the reservoir changes with changing the reservoir pressure, in such case PVT properties could be estimated using measured data on the surface separator, well head, offset field, etc. However, in many cases this data is not available and the only source to estimate these properties is the empirically derived correlations. In fact, there are many correlations that have been developed in the past seven decades for different geological areas. Nevertheless, these correlations are merely reliable for a specific range of data. Therefore, significant errors might yield when these correlations are applied for different regions and crude oils.
This work is the first regional and formal analysis that studies the performance of the most popular PVT correlations and introduces a new set of correlations that estimate some of the PVT properties of different Sudanese crude oils which are: bubble point pressure (Pb), solution gas-oil ratio (Rs), and oil formation volume factor (Bo), at pressures below the bubble point pressure.
The data was obtained from different Sudanese oilfields, to represent a wide range of data. Then, the data was analyzed, prepared and fitted with correlations using non-linear regression techniques. Afterwards, statistical analyses were performed, and the new correlations were compared with the existing correlations to prove their applicability using the most common statistical parameters, i.e. average absolute percentage error, standard deviation and R-square.
The Statistical Error Analyses (SEA) show that all the popular existing correlations yield a variable range of accuracy when these correlations are applied to our database and the proposed correlations were found to be superior to all the popular existing correlations in terms of accuracy when applied for this specific region. Eventually, the performance of the newly developed correlations was tested against each oilfield individually to ensure their accuracy, and it was found that the error varies widely from field to another.
It is concluded from this study that PVT properties vary regionally and recommendation comes to perform such study before using such correlations in estimating PVT properties for a specific region.