Statistical Analysis of Permian Basin Equipment Maintenance at Chevron

• One of Chevron's significant areas of opportunity is streamlining the maintenance activities for their upstream, midstream, and downstream equipment. The proper and optimal performance of their instruments is essential for all other processes within the company, and any failure or disruption in their operations results in a loss of company resources. Given the nature of the processes for which the equipment is used, constant preventive maintenance operations are expected and scheduled in advance. However, there is also a need for corrective maintenance operations when equipment fails, stops, or underperforms.

• Instrument failures are often unexpected and sudden, making them challenging to predict, even for experienced workers or teams. Additionally, the corrective maintenance operations required to address and resolve these issues tend to be time-consuming and result in significant resource losses. In the past, before the growth and expansion of information technologies, companies had limited opportunities to proactively prepare for such challenges. However, with the advent of modern methodologies such as statistical analysis and prediction models, there are now numerous possibilities to forecast equipment failures.

• In this report, we will delve into the methodology and results derived from the development of a statistical analysis on equipment maintenance activities. The primary purpose of this analysis is to present a proof of concept for streamlining corrective maintenance operations using machine learning techniques. The main objective of the statistical analysis is to accurately forecast the frequency, duration, and monetary impact of maintenance activities related to Permian Basin equipment.