Accurate OOOI Data: Implications for Efficient Resource Utilization

This paper compares the accuracy of sources of measurement of four aircraft event times: pushback from the departure gate, take-off from the departing runway, landing at the arrival runway, and arrival at the arrival gate. These events are commonly referred to as (gate) Out, (wheels) Off, (wheels) On and (gate) In, or OOOI. OOOI times are significant because these times are used as the basis of predictions and in calculation of delays. For these and other reasons it is important that OOOI data have sufficient completeness and accuracy. The OOOI data analyzed herein cover Northwest Airlines (NWA) flights to and from Detroit Metropolitan Wayne County Airport (DTW) during May 1-30,2006. Surface surveillance radar track data were used as the reference "truth" for OOOI times. Aircraft Situation Data to Industry (ASDI) and OOOI data provided by NWA (Flight Event Data Store: FEDS) were compared to the OOOI event times extracted from the surface surveillance data. For a given flight the pertinent and available OOOI events were estimated or parsed from the different data sources and compared in a pairwise fashion. This kind of experimental design controls other sources of variability except that of interest, which in this instance are putative differences between data sources. The ASDI ON times contained in the AZ message header were found to be more accurate and much more frequent than the ON time contained in the "A" field of the AZ message. The FEDS data were expected to always be more accurate than ASDI data, but that was found to not always be true. The surface surveillance radar data provide independent and potentially the most accurate data on OOOI event times that may be more generally available compared to proprietary data (e.g., FEDS). Accurate OOOI data make possible the development of models for taxi-time of aircraft on the airport surface. The paper concludes with a brief exposition on the development of a departure taxi-time forecast model. Preliminary results - - suggest that a standard error of 2 minutes on the predicted departure taxi-time duration can be achieved for departures up to 10 minutes before the aircraft leave the ramp area