Title: |
A New Machine Learning Algorithm for Production Well Analysis |
Authors: |
MUCHAMAD Harry, Jantiur SITUMORANG, PRABATA Welly |
Key Words: |
machine learning, wellbore simulation, ANN, decision tree regressor, JIWA Flow |
Conference: |
Stanford Geothermal Workshop |
Year: |
2021 |
Session: |
Reservoir Engineering |
Language: |
English |
Paper Number: |
Muchamad2 |
File Size: |
3239 KB |
View File: |
|
Artificial intelligence or machine learning has been one of the buzzwords recently. The rapid development in the technology sector has greatly affected our lives. They are of great help to humankind. In terms of geothermal, production well is an asset that must be well managed. One of the challenges faced by reservoir engineers in managing the geothermal field is detecting problems in production well early. AILIMA, as a company, try to solve this problem by utilizing machine learning. Using WHP and production mass flowrate data with or without reservoir pressure data, some machine learning algorithms have been trained and tested to estimate the production decline rate. JIWA Flow wellbore simulator has been used to generate synthetic production history datasets for the machine learning test. Fifteen of existing machine learning algorithms tested, no one can provide a good result compared to a wellbore simulator. Therefore, a new algorithm, namely AILIMA-ONE is proposed. This paper describes the application of AILIMA-ONE to estimate production decline. The comparison against ANN (the most complex algorithm) and Decision Tree Regressor (the highest R squared during training process) is also provided.
Press the Back button in your browser, or search again.
Copyright 2021, Stanford Geothermal Program: Readers who download papers from this site should honor the copyright of the original authors and may not copy or distribute the work further without the permission of the original publisher.
Attend the nwxt Stanford Geothermal Workshop,
click here for details.