By 2030, tens of thousands of Pratt & Whitney geared turbofans will be generating a staggering amount of data, as well as thrust. The combined fleet will stream 12 petabytes – equivalent to 12 million gigabytes – to servers on the ground every year, with each engine capturing 50 times the amount of data collected by P&W’s previous commercial powerplants.
As geared-fan architecture enables a step change in propulsive efficiency, P&W is counting on the power of all that incoming data to drive a substantial improvement in engine reliability. The ultimate result should be a reduction in maintenance costs as the manpower needed to support the fleet declines to one-tenth of staffing levels today.
“On average, it takes 10 people to manage a single engine,” says Matthew Bromberg, president of P&W’s aftermarket business. “Through analysing data on the engine and providing all the feedback to the customer that’s necessary they can go down to a single person.”
Capturing the data is the easy part. Each geared turbofan that powers an aircraft built by one of five major manufacturers – Airbus, Bombardier, Embraer, Irkut and Mitsubishi – comes equipped with sensors that capture 5,000 parameters, or 10 gigabytes of data every second. The types of parameters that will be measured include temperatures, air and liquid pressures, rotational speeds and vibrations.
The real challenge is developing a ground-based infrastructure capable of storing, processing and analysing 12 petabytes of information every year. A petabyte is so big a number it is difficult to conceptualise. As Bromberg grasps for intelligible examples, he sounds like an astronomy professor attempting to analogise the number of stars in a galaxy.
A petabyte is “six times the amount of data stored in all US research institutions today”, he says.
Like any aerospace company, P&W uses data-driven systems to inform design and engineering. Simulation models are used today to predict all aspects of engine performance, including aerodynamics, thermodynamics and structural integrity. The PW4000 also monitors about 100 parameters every second; this data is used to identify performance and reliability trends. The PW1000G family, however, requires a new level of support, and P&W decided it needed to look outside for help.
Last July, P&W announced forming a partnership with IBM to begin building the information-processing infrastructure that will be necessary to cope with the geared turbofan fleet’s data requirements. IBM has invested $24 billion recently in data and analytics technology, says Larry Volz, a P&W vice-president and the chief information officer.
“We believe [the IBM partnership] will be an accelerator,” Volz says. “They won’t the be only folks we use in this journey, and we’re already starting to partner with universities and colleges and other companies as well.”
A software algorithm applied to data gleaned from the installed PW4000 fleet has accurately predicted 90% of in-flight shutdowns, P&W says
P&W
Two geared turbofan engines – the PW1100G for the Airbus A320neo and the PW1500G – are scheduled to enter service with customers later this year. The PW1200G for the Mitsubishi Regional Jet and the PW1400G for the Irkut MC-21 are scheduled to complete first flights later this year. The PW1900G and the PW1700G for the Embraer E-190/195 E2 and the E-175 E2, respectively, have now started assembly. As the aircraft programmes ramp up production, each engine will be feeding a continuous stream of performance data that must be stored and analysed.
With IBM’s experts now on board, P&W has started building and testing the analytical tools to mine all of that collected information. There are now 14 projects in the demonstration phase, including a key new software algorithm used to predict the most serious engine events.
In P&W’s long-term vision, data will be used to detect and prevent uncommanded engine shutdowns. The rate of such events has declined to minuscule levels in modern engines, but P&W believes the rate can be further halved with the analytical tools already in development.
In laboratory testing, a software algorithm applied to data gleaned from the installed fleet of PW4000 engines accurately predicted 90% of the in-flight shutdowns, Bromberg says.
The software model is not yet perfect. The software predicts shutdowns, but it also generates false positives when no threat exists, he says.
“Clearly, there’s some false positives in there, meaning sometimes it’s going to send you an alert on an engine and in fact the parameter is not out of the design space,” Bromberg says. “But I can easily see a 50% reduction based on what we see today. A model that is that accurate means we can go in with operators, do on-wing inspection and prevent the event from happening.”
P&W plans to make the analytical tool operational on real engines later early next year, he adds.
The engine supplier is already in discussions with potential airline customers to define the timing of the alerts, says Lynn Fraga, a P&W analytics manager. “For some of the maintenance activities you only need a two-day heads-up,” she says. “Other things that require more logistics and scheduling you might want a longer timeframe.”
These are still early days in the Big Data era for P&W engines. It is arriving just as the overall business model for aftermarket services is changing. The majority of PW2000 and PW4000 engine maintenance services are provided using a transactional model, in which parts and services are provided as customers order them. That is being replaced by a power-by-the-hour service arrangement, in which airlines pay P&W a fee for maintaining a specified availability rate of engines across a customer’s fleet.
About 60% of aftermarket service for the V2500 engine fleet has converted to the power-by-the-hour scheme. The rate is even higher for the family of geared turbofan engines, with only one-fifth of engines covered under the older, transactional method, Bromberg says.
Underpinning the new data-driven aftermarket strategy is access to the data. Every power-by-the-hour service deal signed by P&W includes a clause guaranteeing access to the engine performance information, which is owned by the engine buyer. Although 80% of the geared turbofan fleet is covered by such deals, Bromberg says that number is likely to decline as the fleet ages.
“This is a new technology, new architecture and uncertain maintenance cost,” he says. “As our engine matures, the maintenance cost becomes known, competition develops on the scene – which is good for our customers and for us – and alternatives develop. As the industry matures there will be a variety of maintenance options and it will fragment.”
As more service providers enter the market, that does not mean that P&W will always lose access to the data streaming off the sensors inside each engine. Most customers using the old transactional service model for engines still freely share the information with P&W, even though there is no contractual requirement to do so.
“Today, the vast majority of airlines share the data with us because they want to perform the business intelligence analytics given current technology and provide the fleet data back on a confidential basis,” Bromberg says. “I anticipate going forward the vast majority of customers will be in the same boat.”
In reality, Bromberg acknowledges, reaching P&W’s goal of supporting each engine with only one staff member may be too ambitious. If an airline decides to keep line maintenance and outsource heavy maintenance, for example, that will increase the number of staff required to support the engines, no matter how much data-driven analytics improves the reliability.
“I honestly believe that over the next decade and a half we can drive airlines to a single point of contact within Pratt & Whitney,” he says. “I want to have such a comprehensive offering that they are really struggling with how to staff with more than one person because we’re going to provide everything to them.”
Source: Cirium Dashboard
