Operations Recovery With Passenger Reaccomodation Considerations
Mitigating The Impact Of Schedule Disruptions On Passengers
Through data-driven decision-making, Sabre AirCentre Recovery Manager provides a major move toward integrated airline recovery by enhancing its optimization algorithms to consider passenger reaccommodation alternatives during schedule recovery. The algorithms deliver advanced decision-support capabilities, which are critical for airlines during irregular operations.
Regardless of how well airlines plan their flight schedules, aircraft rotations and crew scheduling, these operational plans often get derailed due to unexpected disruptions.
Extreme weather patterns during all seasons, unplanned aircraft maintenance that becomes necessary, air traffic congestion at airports, security issues, crew availability and numerous other factors frequently force airlines to change their flight schedules, causing flight delays, cancellations and routing diversions.
As with all adversity in life and business, the manner in which airlines respond to and manage their disruptions has a tremendous impact on the passenger experience, the airline’s operational efficiency in terms of how quickly the carrier can recover and resume normal operations, the airline’s brand and, ultimately, the airline’s bottom line.
Clearly, managing irregular operations is a complex decision problem.
Coordination and communication between the operations control center (OCC) and the airline’s operational units, including maintenance control, passenger services and crew controllers, are critical for effective disruption management.
The OCC must adhere to the operational constraints imposed by the disruption, decide among complex conflicting tradeoffs and utilize the airline’s resources efficiently to ensure that the new schedule remains operationally viable.
Often, in addition, these decisions must be made within an extremely short timeframe.
IN A HYPOTHETICAL SCENARIO, TWO FLIGHTS, K0100 AND K0200, ARE IMPACTED DUE TO A FLOW-RATE REDUCTION AT THE DFW AIRPORT, WHICH ALLOWS ONLY ONE FLIGHT TO DEPART BETWEEN 1530 AND 1800. USING CUMULATIVE DATA ABOUT EACH FLIGHT’S PASSENGERS, OPERATIONS CONTROLLERS, USING THE NEW RECOVERY MANAGER OPTIMIZER, CAN DETERMINE WHICH FLIGHT TO DELAY BASED ON WHICH SOLUTION WILL HAVE THE LEAST IMPACT ON THE FEWEST NUMBER OF PASSENGERS. THE RESULT CAN MEAN SIGNIFICANT RECOVERY SAVINGS FOR AN AIRLINE, AS WELL AS MITIGATE THE DISRUPTION IMPACT ON CUSTOMERS.
In a study sponsored by the U.S. Federal Aviation Administration, a study that still presents highly relevant findings even though it was conducted a decade ago, the National Center of Excellence for Aviation Operations Research (NEXTOR) estimated that in terms of actual value at that time, the total cost of U.S. air transportation delays during 2007 came to approximately US$31.2 billion. More than half of this cost (51 percent) was estimated to be at passengers’ expense due to additional expenses incurred, such as meals and lodging, as a result of delays.
More recently, a 2016 survey of global airline executives conducted by Sabre and Forbes Insights indicated that one of the primary factors in measuring the effectiveness of disruption management is the impact on passenger sentiment (this is according to the opinions of 72 percent of the executives polled).
During the past few years, passenger sentiment has become a much more intensely critical concern for airlines, not only from a guest-loyalty point of view but also from a greater branding perspective. This is due to the fact that passengers often very freely share their experiences on social media.
Therefore, for effective disruption management, an airline should consider passengers a top priority when making schedule-recovery decisions.
Complex Decision Relationships
Although on-time performance has become a standard metric for measuring airline operational performance, the relationship between flight delays and cancellations, in combination with the delays experienced by passengers as a result, can be extremely complex.
For example, delaying a specific flight can have a minimal impact on the final arrival time for point-to-point passengers. However, it can cause significant arrival delays for connecting passengers, especially those who are connecting to low-frequency long-haul flights.
Furthermore, the availability (or lack thereof) of passenger reaccommodation alternatives fundamentally dictates the arrival delays that passengers experience.
Although a flight is delayed, connecting passengers can be rebooked on an alternate flight, preserving their original connection. As a result, these passengers often do not experience an arrival delay.
Therefore, when there is a choice of flights to delay or cancel, making these decisions in isolation (without considering passenger reaccommodation alternatives) can lead to suboptimal decisions and higher costs of recovery.
For example, say that two flights, Flight KO100 going from Dallas/Fort Worth International Airport (DFW) to Denver International Airport (DEN) and Flight KO200 going from DFW to Los Angeles International Airport (LAX), are impacted by a flow-rate reduction at DFW, in which only one flight is allowed to depart between 1530 and 1800.
In this hypothetical example, 75 passengers booked on KO100 are point-to-point passengers and 25 are connecting on to Seattle-Tacoma International Airport (SEA) on KO400. Meanwhile, 75 passengers on KO200 are point-to-point and 10 are connecting on to Sydney Airport (SYD).
Based on this cumulative information, the operations controller can decide to delay either Flight KO100 or Flight KO200 by two hours. If Flight KO200 is delayed, 10 passengers on this flight who are connecting on to Sydney will have to miss their flight and can only be rebooked to the following day’s flight.
On the other hand, if Flight KO100 is delayed, 25 passengers will have to miss their connecting Flight KO400 to Seattle in Denver. But an alternate, Flight KO600 that departs only an hour and a half later, is available and can reaccommodate these passengers, who would then be delayed by only 90 minutes.
Both solutions impact the airline’s on-time performance in precisely the same way, because the flight delay is the same in both solutions. But the two solutions have drastically different passenger impact and outcomes.
Delaying Flight KO100 has significantly lesser total passenger delay (187.5 hours) compared to the excessive passenger delay in the second solution (75 passengers have an arrival delay of two hours and 10 Sydney-bound passengers have a delay of 24 hours, for a total of 390 hours).
Additionally, if the operations controller decides to deploy the second solution, the airline may incur additional costs of providing a hotel in Los Angeles for the Sydney-bound passengers.
BASED ON A CASE STUDY USING A LARGE ASIA-BASED AIRLINE, AIRPORT CLOSURE AT THE MAJOR HUB IMPACTS 89 FLIGHTS. THE TOTAL NUMBER OF PASSENGERS ON ALL SCHEDULED FLIGHTS IN THE RECOVERY WINDOW IS 41,148. WHEN TAKING PASSENGER RECOVERY INTO ACCOUNT, 93 FLIGHTS WOULD EXPERIENCE DELAYS, VERSUS 143 DELAYS WHEN USING THE TRADITIONAL RECOVERY APPROACH THAT DOES NOT CONSIDER PASSENGER RECOVERY.
Next Generation Of Algorithms
To consider passenger reaccommodation alternatives during schedule recovery, the operations controller must decide among myriad reaccommodation possibilities for several passenger itineraries. This is in addition to the difficulty posed by schedule and aircraft recovery, which are also challenging and time-consuming problems.
Sabre AirCentre Recovery Manager addresses the complex problem of decision support during irregular operations. It is integrated with Sabre AirCentre Movement Manager, a fully integrated operations-management system. In addition, the solution’s optimizer was developed using robust mathematical optimization models and algorithms.
These algorithms intelligently sift through the vast array of possible recovery options and deliver a holistic operations-recovery solution that minimizes controllable delays and cancellations. The algorithms take key operational-performance metrics into account and adhere to the airline’s business rules and restrictions.
By conscious design, the optimizer leverages the data integration provided by Movement Manager and uses the latest flight, aircraft, maintenance and crew schedules, as well as passenger-load information for near-real-time decision-making.
Airlines can use the optimizer to recover from small-scale day-to-day disruptions impacting only a few aircraft and flights, up to severe disruptions that impact a hub and hundreds of flights.
Operations controllers can also employ the tool to preplan for disruptions by conducting analyses on several what-if scenarios with different parameter settings before deciding to deploy a particular solution.
Traditionally, passenger consideration during schedule recovery has been limited to the passenger load and revenue from a flight. However, as the example in the preceding section clearly illustrates, considering passenger reaccommodation alternatives is extremely important when schedule-recovery decisions are being made.
Recent advancements to the algorithms in the Recovery Manager consider passenger reaccommodation as an important factor. The new feature has the ability to consider passenger itineraries and flight schedules of the airline network, as well as the flight schedules of codeshare and alliance partners.
These models evaluate the inconvenience to passengers and the cost of passenger reaccommodation and provide feedback to schedule-recovery models. The solution recommended by the optimizer minimizes the inconvenience to passengers caused by the new schedule. Key capabilities of the new algorithm include:
- A holistic view of passengers and reaccommodation alternatives in the airline network;
- The capability to consider passenger reaccommodation at an itinerary level and provide feedback to schedule recovery;
- The capability to prioritize high-value customers;
- Detailed modeling of passenger reaccommodation costs, including hotel accommodation and service-related costs, delays and interline costs;
- The capability to consider airline reaccommodation policies and preferences;
- The capacity to produce passenger-friendly schedule-recovery solutions;
- Minimization of the inconvenience and delays experienced by passengers.
These schedule-recovery models consider passenger itineraries, seat availability and reaccommodation costs at a macro level to recommend passenger-friendly schedule-recovery decisions that can potentially reduce rebooking-related costs incurred by the airline.
IT IS CRITICAL THAT AIRLINE OPERATIONS CONTROLLERS EFFECTIVELY CHOOSE THE “RIGHT” FLIGHTS TO DELAY OR CANCEL. IN THE EXAMPLE ABOVE, THERE ARE TWO POSSIBLE SOLUTIONS AN OPERATIONS CONTROLLER CAN DEPLOY. BOTH RESULT IN THE SAME ON-TIME-PERFORMANCE FOR THE AIRLINE; HOWEVER, THESE SOLUTIONS HAVE DRASTICALLY DIFFERENT OUTCOMES WHEN PASSENGER DELAY IS TAKEN INTO ACCOUNT
ALTHOUGH THE AVERAGE FLIGHT DELAY IS SLIGHTLY HIGHER, THE NEWER ALGORITHMS USED IN RECOVERY MANAGER OPTIMIZER, WHICH CONSIDERS PASSENGER REACCOMMODATION, PROVIDE SIGNIFICANTLY BETTER PERFORMANCE ON PASSENGER KEY PERFORMANCE INDICATORS.
Simulated Case Study
To demonstrate the power of the algorithms used in the optimizer, and the manner in which considering passenger recovery during irregular operations can bring benefits, following are results from a large disruption scenario. The various flight, aircraft, maintenance and crew data were taken from a large Asian carrier, and the passenger-booking information was simulated.
Total scope of the recovery was 27 hours, meaning flights that departed or arrived in this 27-hour window were candidates for delays or cancellations.
The carrier operates out of a major hub, and at the time of this scenario, it had a fleet of 151 aircraft. Within the recovery window, a three-hour airport closure was simulated at the hub, which impacted 89 of the 749 scheduled flights. In this scenario, the optimizer was run in two modes.
In the first mode, passenger reaccommodation was explicitly considered during optimization. In the second mode, the optimizer was run without consideration for passenger reaccommodation.
The first mode included the new algorithms for passenger recovery introduced in the optimizer, whereas the second mode simulated the traditional approach to schedule recovery, and simply evaluated the impact of the new schedule on passengers after the solution with the recovered schedule had been generated.
When passenger recovery was considered during schedule recovery, the optimizer produced a solution with 93 flight delays and 38 cancellations, whereas the traditional approach had more delays (143) but fewer cancellations (23). However, the number of passengers with no recovery options (stranded passengers) was significantly lower, with 6,979 stranded passengers compared with the traditional approach, which had 9,129 stranded.
Moreover, the model that considered passenger reaccommodation produced a solution in which the average passenger delay was 36 minutes less than the traditional recovery model.
The number of passengers who potentially needed overnight hotel accommodations was also significantly less in the solution that considered passenger recovery.
Toward Effective Airline Recovery
Clearly, the results of the case study demonstrate that joint decision-making involving both schedule and passenger recovery can significantly improve airline performance during irregular operations.
Previously, optimization models and algorithms that are scalable and deliver robust and holistic solutions for this complex decision problem had proven elusive.
With data and technological innovations, along with sophisticated mathematical modeling, Recovery Manager now offers airlines a robust and powerful decision-support tool to enhance efficiency and mitigate the impact to airline customers during irregular operations.