Peramalan Produksi Hidrokarbon Berdasarkan Decline Curve Analysis (DCA) Dengan Metode Levenberg-Marquardt Algorithm (LMA) Pada Sumur HS-0105 Lapangan AG

Abstract

Hydrocarbon production forecasting is the process of projecting oil or gas  production over a period of time. One commonly used method is production  modeling with Decline Curve Analysis (DCA). Many previous studies have used the  Trial Error and Chisquare test method in DCA. However, this method has limitations  including: long calculation time, modeling is done manually and not automatically,  and the fit of the production curve with the data is often inaccurate. To overcome  these limitations, a method that is fast, automatic and produces accurate curves with  data is needed. One of them is the Levenberg-Marquardt Algorithm (LMA)  numerical optimization approach. The Levenberg-Marquardt Algorithm method is  an iterative method used to solve nonlinear optimization problems. The Levenberg Marquardt Algorithm method has advantages such as: reliability, fast convergence,  and has been widely applied in engineering optimization problems. This study aims  to forecast hydrocarbon production, calculate EUR (Estimated Ultimate Recovery)  and ERR (estimated remaining reserve) at well HS-0105 AG field. First, a DCA  program based on LMA was developed with the MATLAB programming language.  After that, the production data is inputted into the LMA program to do production  modeling until the optimum curve is obtained. After obtaining the optimum DCA  curve, a comparison was made with the curve resulting from trial and error. From  the results of DCA curve optimization with LMA, the RMSE (Root Mean Square  Error) value = 101,756, R2 (R-squared) value = 0,574, computation time 2-3  seconds in the Levenberg-Marquardt Algorithm (LMA) method while the Trial Error  and Chisquare test method RMSE (Root Mean Square Error) value 128.905 R2 (R squared) value = 0.569 and long computation time. From the comparison results,  the Levenberg-Marquardt Algorithm (LMA) method is much better than Trial Error  and is recommended for use. Based on the Decline Curve Analysis with the  Levenberg-Marquardt Algorithm (LMA) method, the EUR (Estimated Ultimate  Recovery) value is 14339,525 STB and the ERR (estimated remaining reserve) value  is 7693,091 STB at the HS-0105 well in the AG field. With the results of this  projection, it is expected that the production of the HS-0105 well in the AG field can  be stimulated to withstand the rate of decline in production and even increase.