Automatic Detection of Anomalous Density Measurements due to
We present a workflow for automatic detection and flagging of faulty formation density log measurements associated with wellbore cave-in (“bad holes”). We use an unsupervised time-series clustering algorithm to simultaneously cluster caliper and density logs, resulting in a labeled data set. Subsequently, we train a number of supervised learning algorithms on the labeled data set to detect bad holes when caliper measurements are unavailable. The workflow is shown to offer superior performance to conventional bad-hole detection methods, such as rugosity calculation, while requiring minimal user intervention. The workflow has been applied to a set of 3,762 Permian wells in order to tag and delete density values recorded at wellbore cave-ins. A density prediction model trained on the deleted data set is used to repredict the densities at the cave-in sections. This is shown to reduce erratic oscillations in density brought about by wellbore cave-in.
Standard price: 10.00
Discounted price: 1.00
10.00
You could save: 90.0%
Author(s):
Deepthi Sen, Cen Ong, Sribharath Kainkaryam, and Arvind Sharma
Company(s):
Texas A&M University; TGS Houston
Year:
2020