This case study presents findings from the deployment of Assaia’s ApronAI, Turnaround Control solution at one of the US airports. The solution is already deployed across 87 gates, using existing camera infrastructure. Soon it will be rolled out to the remaining gates for full, airport-wide, coverage.
For a long time, data availability and data quality have been an issue in aviation. Efforts like Airport Collaborative Decision Making (A-CDM) have been undertaken to address this topic and have already had great positive impact. However, even at airports using A-CDM principles, still many data points are manually generated or sensitive to human tampering. In this case study we will specifically investigate the Actual Off-Block Time (AOBT).
For many airlines, the AOBT also marks the moment from which point onwards aircraft crews get paid. Different airlines use different ways to measure this milestone. Some options include: all doors closed, brakes released or pushback started. In practice, what we see is that often this milestone is artificially fabricated to ‘put the crew on the clock’ even though the flight is actually not yet leaving. This means that airlines are actually overpaying their crews. Furthermore, it also means that incorrect data is used for departure metering, flight planning, on-time performance reporting and air traffic control purposes.
Especially as airlines are emerging from almost two years of COVID crisis, the focus is very much on cost control and return to profitability. Practices like the one described above negatively impact these goals and furthermore might create a competitive disadvantage for airlines who do not control these practices.
At the airport, the ApronAI Turnaround Control solution uses existing cameras and computer vision technology to create timestamps for all key turnaround processes. The easiest event for the system to detect is actually the arrival and departure of an aircraft at a gate. Since the system operates entirely automatically, no human intervention is required or possible. This implies that the system is impartial, objective and ultimately just reports what is happening at the gate.
The airport system is deployed in the cloud. All timestamps generated are saved into a database as well as displayed to different user groups via the Turnaround Control user interface and integrated into the airport’s Aerobahn Surface Area Management application. The database holding all the turnaround timestamps is linked to Tableau, a data analysis application. This application is then used in order to analyse operational performance and find areas for improvement. For the purpose of this study, Tableau was used to analyse the difference between the AOBTs as used by the airport and airlines today versus the actual AOBT as generated by the Turnaround Control system based on video.
Using the Tableau data analysis software, we are analysing AOBT accuracy from May 19th 2021 (the moment the Turnaround Control solution was taken into production) to September 29th, 2021.
Firstly, we found that indeed there is a difference between the current AOBT (which is usually acquired from aircraft sensor data via ACARS) and the actual AOBT as recorded by the Turnaround Control system. The average AOBT inaccuracy for all flights is 180 seconds.
When we look closer (also see figure 1), we also found that there are actually large differences between the different airlines operating out of the airport. The difference between the best and worst performing airline is as much as 6 minutes on average!
Finally, we also found that the AOBT inaccuracy is much more prevalent on delayed flights relative to flights which depart on time. Figure 2 both shows the number of flights for which the difference between AOBT and the actual AOBT is more than 30 seconds as well as the average difference in seconds for both delayed and on-time departures.
Given the fact that crews in some areas of the world start to get paid from AOBT onwards, the fabrication of this milestone at the incorrect moment is a breach of procedure that has a direct negative cost effect for airlines. Table 1 shows the average cost for a crew across all aircraft types departing from the airport. Based on these costs, the benefit for the airlines from switching to the Assaia generated AOBT timestamp is $23 per flight or over $4.5 million per year.
If you are eager to learn more about this particular case study or about different ways in which Assaia’s ApronAI software can create value for your airport or airline contact us at firstname.lastname@example.org or via the website form.