How do you address a problem before it’s a problem? Retrospective analysis requires learning from prior experience – and nobody wants a pilot or an automated system to have already experienced an accident. Much of NASA’s safety research, Nowinski explained, is focused on developing proactive tools through the mining of data from both the aircraft fleet and operations centers. Researchers can analyze this data and learn what to look for, in order to identify precursors to incidents.
“These algorithms analyze and learn from data collected from thousands of operations,” said Nowinski. “They are able to identify unusual, sometimes unsafe, events and the precursors to those events. In many cases we have identified incidents that were not previously known to the operator.” Much of the safety program’s early work was developed to function at the aircraft level: NASA supported the development of a prototype for an onboard algorithm, for example, that identified a precursor that flagged the need for a future engine shutdown.
One of the primary obstacles to evaluating a tool that aims to resolve a problem before it happens, of course, is access to real-time data that would enable it to function as designed. A key enabler of the SMART-NAS project is the SMART-NAS Test Bed development – with an open-architecture, networked system that will function in “shadow mode” as a simulator of the entire NAS, with live real-world data feeds from airlines, airports, FAA facilities, and other elements of the NAS. Four contracts were launched in December 2013 to provide NASA with architecture design alternatives along with cost and benefit assessments before moving to implementations.
“The algorithm identified some precursor that may not have been detected, just a little change in the operation of the engine, that then was associated with an engine shutdown some number of flights in the future,” said Nowinski. “If you can pick up the need for an engine shutdown several flights ahead, then obviously you’ve headed off a potential safety incident. What we’re hoping to do is use the same technologies to understand airspace data – looking forward and predicting where there will be problems, based on small perturbations in the system as it is operating.”
NASA researchers have helped to develop several decision-support tools that make use of data available from ongoing operations. The Dynamic Weather Routes (DWR) tool developed at the Ames Research Center, for example, continuously and automatically analyzes in-flight aircraft and changing weather conditions to find time- and fuel-saving corrections to weather avoidance routes. Version 2.0 of the DWR software was installed on an American Airlines trial system in Fort Worth, Texas, on July 1, 2013, after data analysis revealed that DWR had saved 46 American Airlines flights a total of 360 flying minutes in the previous month.
The SMART-NAS Project: Toward Proactive Safety Assurance
It’s one thing to verify and validate the effectiveness of a single decision-support tool – one concept at a time, with one airline and one FAA facility. It’s another matter altogether to demonstrate the feasibility and operational benefits of an integrated system across the U.S. National Airspace System (the NAS). To meet the need for more potent research of these systems operating together, the ARMD last year launched a new project: the Shadow Mode Assessment Using Realistic Technologies for the National Airspace System (SMART-NAS) for Safe Trajectory-Based Operations.
The data-driven detection of safety issues incorporates analysis of heterogenous data from multiple sources. NASA image
Dr. Shon Grabbe, who manages the SMART-NAS project, said it’s designed to evaluate NextGen technologies and concepts that remain to be implemented: “We’ll have some work on trajectory-based operations, focusing on aircraft conflict detection and resolution,” he said. The project’s initial TBO focus will be on procedures and methods that might be used to improve efficiencies in the New York City airspace. The other concepts to be evaluated include a system that determines how and where aircraft separation functions should be performed, and “networked ATM,” or the control functions that can benefit from networked and cloud-based architectures that allow real-time data sharing.
The evaluation of these concepts, Grabbe explained, requires discrete tools – including the real-time safety modeling and systems assurance technologies being developed by NASA researchers. The inherent challenge will be to develop an algorithm that’s responsive to the dynamic conditions of flight. “A lot of those previous studies really focused on kind of a post-operational assessment,” he said. “So maybe a month after a potentially unsafe event occurred, you could collect all the radar tracking data, or data from the flight deck of the aircraft itself, and then you could run that data through these advanced data-mining tools and capabilities, and they would identify these potentially unsafe situations – aircraft coming too close to one another, or maybe an aircraft not following a published procedure. There are many things that could make for an unsafe situation. But those analyses are typically post-operational. They happen after an event has already occurred.”
One of the primary obstacles to evaluating a tool that aims to resolve a problem before it happens, of course, is access to real-time data that would enable it to function as designed. A key enabler of the SMART-NAS project is the SMART-NAS Test Bed development – with an open-architecture, networked system that will function in “shadow mode” as a simulator of the entire NAS, with live real-world data feeds from airlines, airports, FAA facilities, and other elements of the NAS. Four contracts were launched in December 2013 to provide NASA with architecture design alternatives along with cost and benefit assessments before moving to implementations.
The Airspace Operations Lab at NASA Ames conducts simulation of Automatic Dependent Surveillance-Broadcast (ADS-B). NASA Ames Research Center’s advanced airspace modeling and simulation tools have been used extensively to model the flow of air traffic across the United States and to evaluate new concepts in airspace design, traffic flow management, and optimization. NASA Ames
The virtual world created by the SMART-NAS test bed will allow new air traffic capabilities and technologies to be demonstrated together, gate-to-gate, in real time, in order to confirm that they’ll perform as expected. “With the test bed,” said Grabbe, “we’re going to have the ability to maybe test out a couple of those concepts in parallel, so we can start exploring those interactions between one another before they are operationally deployed. We’re really focusing on the real-time aspects as we’re shadowing and monitoring current operations. Can we get the live data feeds – the live weather, for example – into the test bed and into these real-time safety algorithms?”