Oil pumps are critical to the efficiency of oil wells. When a pump fails, it’s very costly to remove and replace the part. It involves not only stopping oil production, but also mobilizing costly, specialized personnel and equipment to pull the pump out of the well and replace it with a new one. These pumps are designed to work in harsh environments, but choosing the right pump for a specific well’s geologic and environmental condition is very complex. The right pump can operate for a longer period of time without costly interventions.
This Fortune 20 oil company needed a way to extract data from across silos and generate actionable recommendations to help make optimal pump selections, increase billable hours and reduce overall costs through efficiency improvements.
They used the Maana Knowledge Platform to help maintenance employees choose the right pump for each site, anticipate failures likely to occur and implement an effective predictive maintenance strategy. Maintenance experts used the platform to collect data related to existing pump operations from a variety of sources (such as run-and-pull reports, pump failure reports, pump sensor data and high-frequency data flows) and analyze it to predict the likelihood of a pump failure.
Much of the data related to the inspection of failed pumps was entered by employees working in the field, who were responsible for describing what they observed when they retrieved failed pumps from wells. In addition to this language-based data, the company collected highly detailed sensor data during pump operations.
Using the Maana Knowledge Platform, the company was able to use both the machine data and human language data to understand the causes of pump failures. For example, subject-matter experts used it to crawl and index these disparate data sources. And the platform’s interactive exploration environment enabled them to validate various hypotheses related to failures and causes.