A research team from Skoltech has introduced a model to speed up the planning stage of oil well development. This helps to obtain important data about a well—for example, the model can compare a prospective well with those already operating nearby to predict its oil-producing properties and improve well drilling. The study came out IEEE Geoscience and Remote Sensing Letters.
Oil and gas well development can be divided into three stages—discovery of the oil field, evaluation, and so on. The evaluation covers, for example, the number of oil reserves and their distribution. This stage includes drilling exploratory wells and checking them to record some indicators: layer radioactivity, groundwater mobility, etc. Later, this information will help make decisions related to well development.
“Currently, the evaluation of the oil field provides a set of fragmentary data and no one knows how to use it. Our study aims to build a model based on these data that will make in a mathematical representation-a vector that perfectly describes the well,” commented Alexander Marusov, the first author and Skoltech research engineer.
The vector returned by the model contains useful information about the well in a compressed format. Besides predicting its properties, the model helps to solve the issue of drilling in the wrong direction. While moving deeper into the layer, it is important to confine the drilling to one and the same type of rock. Otherwise, you have to drill again, in a different direction, and this is very expensive.
“Our model helps to determine the rock type and adjust drilling. Our model’s rock type prediction accuracy is 82%, while previously the best result was 59%. Our model makes it easier to make a decision on the development of the oil well, “Marusov added.
The model is trained by self-supervised learning. This is different from traditional machine learning methods, which require labeled data. Self-supervised learning does not use this. For example, a probe may register radiation or other geophysical signals in an exploration well. Self-supervised learning uses this raw data, without any labels.
“Self-directed learning methods can be divided into contrastive and noncontrastive. We use the latter method, which deals with pairs of similar objects—for example, the intervals of signals from one and the same well,” explained the researcher.
Alexander E. Marusov et al, Noncontrastive Representation Learning for Intervals From Well Logs, IEEE Geoscience and Remote Sensing Letters (2023). DOI: 10.1109/LGRS.2023.3277214
Provided by the Skolkovo Institute of Science and Technology
Citation: Deep learning model could improve oil extraction (2023, July 13) retrieved 13 July 2023 from https://techxplore.com/news/2023-07-deep-oil.html
This document is subject to copyright. Except for any fair dealing for the purpose of private study or research, no part may be reproduced without written permission. Content is provided for informational purposes only.