Forecasting Oil Production Rates for Oil Wells Using Decline Curve Analysis and Polynomial Regression Model

In the course of employing decline curves to predict future oil production rates, there is a need to first determine the initial flowrate which often times presents an extent of difficulty especially for harmonic and hyperbolic decline trends. Conventional methods of initial flowrate determination require extrapolation of charts on study or spreadsheets which is time consuming. It is therefore essential to provide an algorithm that would help speed up initial flowrate calculations. This study employs the use of decline curve analysis (DCA) and a polynomial regression model in predicting future production rates. Model identification was performed by plotting relative decline rates on flow rates, thus allowing for the fitting of appropriate decline models (exponential, harmonic and hyperbolic decline rate) into given production history data. More so, a fourth order polynomial regression model was fitted into a range of production history data via a proposed computer algorithm – to specify a value for the initial production rate (qi). While incorporating the regression model into a computer model, Gauss-Jordan algorithm was employed in providing a solution to the unknown constants. Results obtained from a number of production history data (Well AB, Well AC, Well XY, Well XZ) through the use of the developed model (FORECAST) revealed that, regardless of the nature of production history data for an oil well (exponential, harmonic and hyperbolic), production rate forecasts can be easily performed. The accuracy of fitted regression model into each production data was validated by using R² values. However, the estimated R² values revealed 100% level of accuracy for each sample production data.