The power of Predix is enabling the GE Aviation Fleet Support team to capture a higher volume of engine data and to analyze it more quickly and accurately, resulting in significant improvement in diagnostic capabilities offered to customers.
Operating from the 24/7 Fleet Support Center just north of the Evendale headquarters and from Shanghai, China, the team is tasked with monitoring data from some 35,000 GE and CFM engines operating worldwide, processing in excess of 100 million flight records per year, a figure that is rapidly growing with the introduction of new engines like GEnx and CFM LEAP that have significantly more sensors than legacy engines.
“Our job is to quickly and accurately sift through all that data looking for unknown unknowns, to catch any technical issues before they cause operational disruption for our customers,” says Fleet Support Executive Director Vijay Singh.
While it’s a tall order, the power of the Predix platform is making the job easier, enabling a stronger marriage of physics and analytics that is driving earlier detection of issues with greater accuracy than ever before.
The metrics are impressive: since the migration to Predix was completed, fidelity has improved by more than 30%, detection rates are up by 25%, and accuracy has climbed by 14 percentage points.
“Our product support engineers are looking at a more focused set of possible issues, and they are giving customers more lead time to address issues,” Singh says.
Predix is an industrial cloud-based service designed to meet the unique security, scale and speed requirements of industrial-strength machines. More than 7,500 developers from within and outside GE and its customers, including Boeing, ConocoPhillips and RasGas are building on the Predix platform. Predix enables GE to quickly and efficiently bring software innovations like its Asset Performance Management and Brilliant Manufacturing solutions to customers and to update them seamlessly with new improvements.
For Fleet Support, the Predix platform replaces a legacy database system that dates back to the mid-1980s. Using that system to monitor the fleet required lots of manual data manipulation and offered only a limited toolset. With Predix, data integration is far more sophisticated and the team can develop advanced, multi-variate analytics that combine several data sources for more accurate prediction. Predix also feeds customer views on myGEAviation.com customer portal, giving them additional real-time access to operating data.
One of the most impressive metrics from the migration is the reduction in raw alerts the team analyzes. In 2014, Fleet Support assessed some 350,000 potential alerts and only about two percent of which resulted in a customer notification. In 2015, that number dropped to 230,000, and the goal for this year is to cut the number by more than 50%.
“Predix enables us to have a learning system that gets smarter every day,” Singh says. “By driving down the number of basic alerts by two-thirds, our human expertise and experience is focused on the highest value data. That’s the power of analytics + physics.”