Running a drill site can be an expensive venture, especially in the event that drilling operations cease. The sheer volume of the data collected on site can hold millions of different values and can take a substantial amount of time to analyze. Because of the distance from the ground level, operators on site must determine the status of the drill through feel and experience, rather than being able to rely on visual cues. All these factors result in setbacks and it can be difficult to understand exactly what is happening without significant study.
Trax Electronics specializes in machine learning, automation, and monitoring digital data in the oil field. The company wanted a way to predict whether a drill was rotating, sliding or properly drilling in order to maximize drill site efficiency. With their current system, Trax Electronics could determine what happened after a three day window. In order to minimize this waiting period, Ryan Schmitz, the president of Trax Electronics, reached out to VizworX for a speedier solution.
VizworX created a simple classification program that determines the drill state of a drill bit and its current function. It involves a decision tree that outputs a model within 30 minutes of receiving the data. The model has an accuracy rate of 97 percent. This is an increase from the 80 percent accuracy that could be determined by previous analysis of the same data. It is also possible to retrain the data and servers, allowing the program to constantly adapt for maximum efficiency.
The result minimizes the waiting period previously mentioned and allows operators to increase the amount of time spent drilling. By increasing the drill site efficiency, stakeholders in the oil and gas industry can maximize cost-effectiveness. Trax Electronics continues to serve their customers and reduce operating expenses.
To learn more about Trax Electronics and our work with them, visit their website at traxelectronics.ca.