A new correlation for oil formation volume factor of oil and gas mixture using Group Method of Data Handling; an empirical approach

Numerous production engineering and reservoir analysis problems have been solved by estimating formation volume factor (Bo) for oil-gas mixture. It is considered as one of most important factor of physical properties of hydrocarbon that use in evaluation and enhancement of the reservoirs. In addition to, evaluation of formation volume factor for oil and gas mixture is considered as an important tool in any field project development as well as in reservoir performance evaluation, because both of reservoir engineering and production design operations usually evaluate the changes of the fluid properties, and one of these properties is the FVF.
This project aims to construct a new correlation that can estimate the formation volume factor (Bo) for oil and gas mixture with much more accuracy (less errors) than the current one utilized by the industry. In order to develop a new correlation for Bo for oil and gas mixture, Group Method of Data Handling (GMDH) algorithms has been used in this study to generate an optimum model. GMDH Approach will be utilized for the first time to predict this property.
Total of 268 data sets have been collected from different regions. In addition to, different statistical and graphical tools have been used to assess model accuracy after collecting the required data. The performance of GMDH model is compared against the best correlating adopted by the industry currently. Small range of absolute average relative errors (1.53%) has been obtained whereas the correlation coefficient has been calculated as 0.993. Moreover, the standard deviation for the new Bo model has been calculated as 0.0271% with 0.00229% of minimum error for this correlation. Trend analyses have confirmed that this new model for oil formation volume factor at bubble point pressure is physically correct.