"Predicting Fracture Dip with Artificial Neural Networks and Image Log Data"

What is it about?

The work introduces a new technique for enhancing existing software used in oil and gas exploration, like Petrel and geoframe. It focuses on predicting fracture dips in three wells within the Gachsaran field using image logs and other geological data. By employing feed-forward artificial neural networks (ANN) with back-propagation learning, the technique forecasts fracture dip data in the third well based on information from the other two. The results demonstrate the ANN model's ability to simulate the relationship between fracture dips in these wells, with high accuracy in both training and test sets.

Featured Image

Photo by Jack Prommel on Unsplash

Photo by Jack Prommel on Unsplash

Why is it important?

Understanding fractures in oil and gas reservoirs is pivotal as these fissures serve as both reservoirs and conduits for oil and gas movement. Consequently, a thorough comprehension of fractures is essential. Companies continuously seek to enhance software capabilities within this field. This study introduces a new technique intended to augment established software like Petrel and geoframe. Using data from image logs and geological logs of three wells in Gachsaran field (GS-A, GS-B, GS-C), the study applies feed-forward artificial neural networks (ANN) with back-propagation learning to forecast fracture dip data in the third well by leveraging information from the other two. The results demonstrate the efficacy of the ANN model in replicating the relationship between fracture dips in these wells, indicating high accuracy in both training and test sets. This advancement holds significant importance in refining software functionalities for understanding and predicting fractures, critical for the oil and gas industry's exploration and extraction endeavors.

Perspectives