Title : Deep learning models for history matching and forecasting reservoir performance
Abstract:
Advances in new technologies particularly artificial intelligence (AI) and machine learning (ML) have now made it possible to use AI and ML based approaches for building reservoir models, which do not rely on conventional simulation tools for history matching and production forecasting. These new methods have the capabilities to eventually replace numerical simulator as a fast history matching and forecasting tool. Objective of this work is to use new AI and ML based deep learning algorithms for performing history matching and forecasting on a series of industrial dataset from middle east and North Sea. The results show that AL and ML based deep learning algorithm-based models are very good and gives close to 85% accuracy in history matching a few well patterns. In total model is evaluated on 3 different datasets with high accuracy. The models are also extended to generate long term forecast.
Audience Take Away:
- Use of machine learning in oil and gas application
- AI and ML based time series history matching and forecasting of oil and gas production for reservoirs
- Ai and ML based forecasting can reduce the time taken for history matching and forecasting from months to days. Computationally fast and accurate approach can reduce reliance on expensive history matching and forecasting tools.
