DrillX: 1st prize in Total's Best Innovators 2020

format site sicara (Bannière moyenne (US) (Paysage)) (1)

Context

Since the 2010s, Total has launched its Digital Factory to support its ambition to become a leader in responsible energy.

On oil platforms, the main risks are kicks (gas influxes) and stuck pipes (pipes blocked by trapped material), which can halt production and even cause explosions.

In 2017, Total Exploration & Production launched the DrillX program to secure and optimize its drilling operations.

The challenge

Two challenges needed to be addressed:

Data latency and the speed of algorithm prediction calculations.
The reliability of kick and stuck pipe predictions. Unidentified risks can have human, environmental, and financial impacts. Raising too many false alarms increases non-productive time and the associated costs of drilling exploration.

The solution

A dashboard presenting predictions in near real-time (5-minute latency):

Predictions with a UX/UI displaying the values and mnemonics leading to such predictions.
A custom architecture using Azure Functions and Databricks, along with a data quality alerting system that can affect prediction quality.
Machine learning models specific to each prediction, retrained in near real-time.

Stack Technique

Results

Thanks to DrillX, Total became the first oil company to use predictive analysis in drilling operations and won the 1st prize in Total's "Best Innovators" 2020.

95%

Of kicks detected

100%

Of active wells monitored

Minutes lead time between sensor data retrieval and application display