Power plants are industrial facilities used to generate power. Because the process of converting mechanical power into electrical current generates a great deal of heat, plants need to be monitored for safety, reliability, and efficiency. Ideally, monitoring helps workers identify problems before a failure occurs, thus avoiding costly downtime or accidents.
Many power plants have critical components outfitted with basic mechanical instrumentation for field service engineers to monitor operations. However, there are many other parts that are uninstrumented and therefore cannot be monitored for heat loss or other anomalies. This lack of visibility prevents field service engineers from being able to identify issues with these components and proactively repair or replace them before costly unplanned downtime occurs. These anomalies can also lead to reduced efficiency in turbine performance through heat loss.
Enter TITAN, a Predix app based on Cloud Foundry. An acronym for Thermal Imaging Tool for Anomaly Notification, TITAN uses thermal images (shot on an iPhone with a $250 camera attachment) to track anomalies. Workers load images into TITAN, which can run a machine learning algorithm on future images and automatically classify them as either normal or anomalous, according to GE.
Initially conceived as part of a hackathon, TITAN has been deployed at six different power plants so far. Via Titan, workers have identified several issues that, when addressed, will help their plants run more efficiently, according to the GE case study. In one case, managers estimated the application could save the company up to $50,000 a year.
Read An App for Asset Analytics: TITAN Thermal Imaging for the details.
Resources
- An App for Asset Analytics: TITAN Thermal Imaging
- PAN1: Thermal Imaging Analysis ( Predix Transform 2016)
