Abstract

In oil and gas, Production forecasting is the most vital process required for all the operational and investment decisions. The basis for any project-based resource estimation is a good production forecast. The objective of this paper is to demonstrate assisted production forecasting from the real-time data using AI (Artificial Intelligence) Techniques.

Decline curve analysis is one of the most used production forecasting methodology. Conventional Production Forecasting involves anomaly removal and applying best method for DCA (decline curve analysis). These processes are combined and automated Using Big data and AI techniques, in this method. The algorithm used clustering techniques like isolation forest and density-based clustering to automatically remove the anomalies and later uses neural network / conventional curve fitting (depending upon data frequency) for production forecasting. It also used LSTM (Long Short-term memory cells algorithm) neural network to predict the flowing bottom-hole pressure with the rates forecasted using automated production forecasting. This in-turn was used along with VFP (Vertical Flow Performance) curves of the well to assess the well ceasure date under existing condition. The algorithm also has a module to recommend the decline rates of workover wells based on the similar past events by using unsupervised clustering techniques.

The algorithm removed the anomalies data and perform production forecasting in a scalable manner. The actual and calculated EURs were matching very closely (very less root mean square error and a maximum deviation of 8%).

This approach is useful to debottleneck the processes involved in production forecasting like identifying the rate changes, production forecasting and FBHP prediction. This information can be coupled with an interactive dashboard to support quick diagnostics, easy interpretations and prompt business decisions.

You can access this article if you purchase or spend a download.