Enhanced rig activity analysis with Machine Learning
-
WellAct has a very effective algorithm.
-
Like any mathematical function, it needs certain data to function. If they are missing, it will not run the analysis.
-
ML helps "plugging" those holes as we have trained it to recognize patterns even if data are incomplete or not good quality.
Generative AI integrates RT reporting
An AI-based Chatbot answers questions on the status of the drilling process.
Weight To Weight Metrics
-
Median Duration: 6.60 minutes
-
Maximum Duration: 13.03 minutes
-
Minimum Duration: 4.40 minutes
-
KPI Target: 5 minutes
Performance Evaluation
Inconsistent Weight-to-Weight Durations
-
Large variation between the longest and shortest intervals indicates performance inconsistencies.
-
Significant potential for time savings in the remaining drilling section.
KPI Target Exceeded
-
Median (+10%) is above the 5-minute KPI, signaling structural inefficiencies.
-
Optimizing operations could yield substantial time & cost savings.
Most Time-Consuming Activity (within Weight-to-Weight)
-
Circulating: The majority of connection time is used here.
-
Should be reviewed for efficiency improvements.
Results
Can we quantify improved performances?
Built a RT reporting system active 24hrs a day via chatbot
ML improved the accuracy of the automated rig activity identification by a margin between 5% and 10%. Bringing the overall rig activity identification between 85% and 90% of the total time.