The utilization of machine learning method to predict hydrocarbon flowrate for a better reservoir potential evaluation
The numbers of machine learning technologies used in subsurface characterization work is increasing with more company rely on data driven to assist in performing any evaluation. In this study, a supervised random forest machine learning approach was utilized in two stages; first stage was to predict...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Conference or Workshop Item |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/108685/ http://dx.doi.org/10.2523/IPTC-22025-MS |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | The numbers of machine learning technologies used in subsurface characterization work is increasing with more company rely on data driven to assist in performing any evaluation. In this study, a supervised random forest machine learning approach was utilized in two stages; first stage was to predict static reservoir using well logs and core as inputs. The output is then used as the basis in the second stage to predict initial oil rate (Qi) and subsequently to determine estimated ultimate recovery (EUR) at targeted interval as proposed in the first stage. Static reservoir machine learning prediction outputs were benchmark with available routine core analysis with the result showed R2 of 88% respectively. For initial oil rate (Qi) prediction, a total of 9000 observation points from 20 wells were extracted for training and blind testing process by using variables such as permeability, net thickness, well choke size, well flowing pressure, average pressure, water cut, irreducible water saturation (Swi), and historical production rate. The estimated ultimate recovery (EUR) is then predicted utilizing the thickness of that unit and the decline rate that is obtained from the neighboring wells that has produced from the said reservoir as the analogue. The Qi and EUR results from machine learning is compared with the estimated Qi and EUR using conventional methods for verification purpose. The results from machine learning dynamic properties prediction showed 97% R2 for training while the testing score mean is 87% against the historical data. High R2 from static and dynamic machine learning prediction indicated that the method was reliable and able to assist petroleum engineer in reservoir potential evaluation process. |
|---|