there is so much data that it is impossible for someone to look at all of it and as a result, intermittent anomalies may go undetected. The second problem is that you have to have the test technologies and computational resources necessary to detect the anomalies."
For an inlet test, a minimum of 40 measurements are required at the aerodynamic interface plane between the aircraft inlet and engine to characterize the flow entering the engine.
"If there are too many invalid measurements or, in the worst possible scenario, there are invalid measurements that you are unaware of, you can't characterize the distortion level entering the engine and what the effect of that distortion is on engine's ability to operate continuously without surging," Malloy said.
The flow field entering the engine is highly dynamic and susceptible to distortion from various sources including high angle of attack maneuvers, heavy cross winds and wake from aircraft and missiles.
"To ensure reliable engine and airframe integration, it is critical to fully assess this flow field for temporal and spatial variations," said Maj. Kurt Rouser, deputy chief of Arnold Engineering Development Center's Aeropropulsion Systems Test Division.
Malloy said the team was confident they could develop the propulsion capability into a technique that could be used in a wind tunnel.
"We have a lot of experience applying state-of-the-art test technologies to validate propulsion system data in ground test or flight test," he noted. "We've always known that these technologies could be extended to aero data - either internal aero data from a sub-scale inlet test or external aero data from a sub-scale aircraft model."
During the test, Malloy said the goal was to ensure that the data quality was consistent with pre-test estimates.
"We wanted to look at enough of the data to verify that the data was of the quality we promised the customer," he said. "We were hoping that there would be no anomalies and that if there were anomalies, that they could be quickly repaired or that there were enough good measurements that we wouldn't have to stop the test."
There was only one measurement anomaly and the analysis software alerted analysis engineers at the instant the measurement failed.