Artificial Oil Lift Production Forecasting and Analysis Using Autoregressive and Deep Learning Models
6 Pages Posted: 13 Jun 2019
Date Written: May 30, 2019
Abstract
Oil and Gas industry is one of the renowned industries where prediction and forecasting of necessary oil statistics is important from both the productive and economic perspective. Artificial lift methods are being used to increase the production rate. Even if the natural pressure is appropriate at the bottom well head, we do depend on artificial lift mech-anisms like beam pumping, hydraulic pumping, ESPs (Electronic Submersible pumps) and gas lifts to artificially pump oil up the surface. ESPs is most commonly used method for the purpose. IIOT devices and sensors have been established for monitoring the bottom well pressure, amount of oil produced, the temperature of the wellhead etc. All these data are in the form of time series which are analyzed using time series algorithms for predicting the future production beforehand. This study presents an application of an Autoregressive model to estimate the future production performance of oil wells based on monthly-production time series data. Deep learning Model (LSTM) is applied to get more accuracy in prediction.
Keywords: Autoregressive models, Oil & Gas production, Dickey-Fuller Test, Root Mean Squared Error, Industrial Internet of Things (IIOT), Deep Learning, Prediction, Forecasting
JEL Classification: C53, C22, C32
Suggested Citation: Suggested Citation