How Airbus solves digital thread problems with NLP, AI, ML and search engine technology.
Sinequa has sponsored this post.
In the engineering community, it’s trendy to say data is the new oil. This is a crude metaphor—pun intended—as the former represents an ever-growing and labyrinthine resource, and the latter is an ever-dwindling and limited energy source. A more apt analogy is that data is the desert. It’s a vast resource containing ancient structures and artifacts of useful knowledge lost to time under growing dunes of information.
No matter how much data is collected, it isn’t immediately valuable to a company. Consider the customer support team at Airbus Helicopters. According to a translated presentation during Inform Paris, Frederic Antoine, Technical Support Network manager & Business Solutions architect at Airbus Helicopters, their team fields over 30,000 questions from 3,000 clients every year. About 60 percent of those clients fly only one helicopter and have limited experience working with the equipment. Nonetheless, the support teams rely on the data of over 12,000 in-service helicopters—many of which are custom-made and thus require custom servicing.
If that data sits unprocessed, it becomes a wasteland that sucks up the support team’s time. Some more experienced members will navigate the data faster—with experience, they know where to dig for knowledge. But gaining that experience takes time and risks the introduction of human error as well as the loss of corporate knowledge once retirement comes around.
Raw data hides insights, and Airbus Helicopters needs those insights to ensure the safety of its clients. From a maintenance perspective, an engineering manual for hundreds of fifteen-year-old med-evac copters is just as important as the twenty custom manuals needed to repair each bespoke system that was sold five years ago. There is no guarantee that a maintenance engineer knows which manual is useful for a given problem or what page contains the information needed. But the knowledge does exist, somewhere, lost to time within various support applications.
So, how does Airbus Helicopters get the right information to the right people when they need it? The organization invested in digital thread and enterprise search technology. This tool wasn’t limited to one PLM system; it could dig into various data hidden in everything from email to CAD. Engineers looking for maintenance information no longer had to dig through a desert of files. Instead, they asked a question and got an answer.
“That’s pretty much what our project is all about,” said Antoine, “being able to place this search engine above a whole range of databases that we use daily for our searches to provide client support.”
How Airbus Helicopter Leverages a Desert of Organizational Data, Knowledge and Intelligence
The challenges for Airbus Helicopter become a lot more complex the more you think about it. If a copter is fifteen years old, chances are that many parts—perhaps even the entire engine—had been replaced or retrofitted a few times. Every time these parts are switched up, that data can affect how that specific copter needs to be serviced. Maintenance engineers need to know not only the model they are working on, but all its customizations, repairs and replacements to properly maintain the vehicle.
“So, to achieve our objectives,” said Antoine. “We need to be able to search through several sources of information in different formats. We have information, which is organized in SQL databases, we have PDF documents, we have XML formats, and a whole range of other formats. We want [to give not only] our clients a 360-degree view, but also our technical support teams, so [that] they can search more easily and that their searches are high performing.”
To make sense of all the data, Airbus created an internal search engine called Hyperion. The search engine, powered by Sinequa Natural Language Process technology (NLP), acts as a digital thread of sorts. It sits on top of the company’s current systems by connecting to the data spread across its many software tools. For help with the creation of Hyperion, the aerospace giant went to Sinequa, a company that specializes in AI-enabled search tools. The idea is that with NLP and neural search, maintenance engineers can change how they access information. Instead of searching through documents to find answers, the process is more akin to using an internet search engine.
“Sinequa is in insight engineering,” says Xavier Pornain, senior vice president of Corporate Development at Sinequa to engineering.com. “We can connect unstructured human generated data, from email to technical CAD documents outlining the history of a part to the industrial data from its manufacturing. Sinequa builds the bridge to help support engineers to be more efficient.” He added that third party engineering and PLM tools from PTC, Dassault Systèmes, Siemens Digital Industries Software, Microsoft and much more can be accessed by Sinequa technology.
As for why Sinequa was selected, Bastien Pesce, IM Senior Project Manager at Airbus Helicopters said in the presentation, “we chose it for [a few] reasons because it was simple to set up, efficient and corresponded fully to our operational constraints. We installed it, set it up, and then I moved on to other tasks.”
Pesce’s comment speaks to the ability of Sinequa to act as a non-disruptive tool that can be added on top of various systems and operations. Since the software works independently as it crawls and indexes data from various resources, there is no need for Airbus to change any ongoing engineering workflow. Instead, the search tool works in the background, and it continues to learn as the maintenance teams continue their day. With each query into the search tool, and as new documents are added to the various applications, the search tool learns and becomes better at assisting maintenance engineers.
How Sinequa Impacted Airbus and Others
During a proof-of-concept test of Hyperion, Airbus noted that the system could answer 20 percent of all the questions posed by its clients. This represented the simple questions that the search engine was able to provide straightforward answers to. By giving its clients access to the search tool, Airbus could eliminate most of these questions before they got them. As an added benefit, the interactive nature of the search tool enabled the clients to learn about their equipment in real-time, improving the user experience.
With another 60 percent of these client questions, Hyperion was able to provide an answer to the cause, but this led to more questions or engineering analysis by the maintenance team. To test how much time this could save its team, Airbus H175 experts were given access to the search engine and instructed to use it instead of their traditional methods to search for client solutions.
The team reported that often, with just a few words or by explaining the problem, they were able to get an answer to the issue, and the documents needed to solve it. Though this didn’t eliminate the need to do any engineering work, “We estimated that we would reduce the response time to our clients by approximately 5 to 10 percent. That means that if I responded in ten days before, I’ll now respond in nine days,” said Antoine.
An extra day might not sound like much on the surface, but in aggregate it can make a big difference. Over a year, each engineer using the tool could save 36 days and solve about four more service issues. The more engineers on the servicing team the more this benefit can scale.
Pornain explained that this tool has added benefits for Sinequa’s clients. For instance, he notes that any time one can reduce the time and effort to service or maintain something, especially something as complex as a fleet of helicopters, it will significantly change the cost of quality. The less time you have people working on a problem for the same resolution and with less waste, the more you improve the bottom line.
“There are safety applications,” says Pornain. “If they are missing a part that needs replacement due to a defect, then lives are on the line. They need to find the right information promptly to provide that information to the engineers and to the customers.”
He adds that sometimes it isn’t enough to find the right information; sometimes you need the right person. The tool can also do this by referencing which employees have worked on which projects. It can then link the person asking the question to the best individual to solve the problem—or bring context to the data and documentation. He described it as the difference between getting any engineer on the problem and getting the one with the most experience, not based on hype or a resume, but based on hard data.
Similarly, the search tool can be used to help get people new to a project, or Airbus itself, up to speed. They will still need training on incident resolution, but they will have easy access to the right information. “Everyone new to Airbus doesn’t need to be trained on various equipment,” said Pornain. “The equipment is so complex that people don’t know where to look. Now they are onboarded using NLP so the employee can find what they need.”
Pornain notes that this is the tip of the iceberg. To learn how Sinequa technology can improve the processes of other engineering organizations, read the white paper “Is Your Digital Thread Cut Short? Mend It with Intelligent Search.”