SCHEDULE RELIABILITY MODELING
ROBUST APPLICATION HELPS BETTER PREDICT RELIABILITY IN AIRLINE SCHEDULES
SCHEDULE DEVELOPMENT HAS TRADITIONALLY REPRESENTED AN INEXACT SCIENCE WHEN IT COMES TO RELIABILITY, BUT ACTIVE EXPERIENCE CAN BE EFFECTIVELY APPLIED TO HELP MAKE AN AIRLINE SCHEDULE MORE PREDICTABLE, AS WELL AS BUILD GREATER RELIABILITY INTO SCHEDULE PERFORMANCE.
In the airline industry, schedule development has always been challenging due to the sheer scale and complexity of any given carrier’s flight network, as well as numerous practical and regulatory constraints.
Consequently, in developing a schedule, a great preponderance of consideration has been placed on profitability and competitiveness, but a lot less attention has commonly been paid to schedule reliability.
Schedule reliability is most fundamentally and directly measured in terms of flight delays, cancellations, and missed crew and passenger connections. The consequences of an unreliable schedule can be both tangible and intangible, with a highly significant financial impact.
Tangible negative schedule outcomes result in lost revenue due to canceled flights and missed passenger connections, as well as increased costs due to disruptions to crew schedules, higher levels of jet-fuel consumption and inadequate ground staffing.
Examples of intangible impacts include less-than-desired customer satisfaction and loyalty, in addition to various potential environmental impacts.
Variability in flying times and taxi times, congested airports, air-traffic-control issues, airline ground resources and weather are major contributing factors to flight delays and disruptions.
BASED ON STATISTICS FROM THE U.S. BUREAU OF TRANSPORTATION, FROM 2006 TO 2013, IN THE MONTH OF JULY, FOUR U.S. MAJOR CARRIERS EXPERIENCED SIGNIFICANT ARRIVAL DELAYS. SABRE SCHEDULE RELIABILITY MODEL HELPS AIRLINES UNDERSTAND THE REASON FOR DELAYS, AS WELL AS HOW TO REDUCE THE POSSIBILITY OF DELAYS.
Naturally, most airline schedules are developed based on profitability and competitiveness, often using overly optimistic planning parameters. However, many schedules are deeply involved with and directly influenced by operations that are inherently interdependent.
As a result, a small delay at the beginning of the day may lead to a series of delays to “downstream” flights later in the day.
To help alleviate some of the potentially negative factors relating directly to schedule reliability, considerable efforts in planning and analysis have been invested by academia and the airline industry.
For example, there are mathematical and simulation models to predict flight delays. These methods can indicate which flights are more prone to delays; however, they do not reveal the reasons behind the delays and how to reduce the possibility of delays.
Procedures to reduce delays include adjusting planning parameters, such as increasing ground and block times, redistributing existing slack within the flight schedule, formulating mathematical-programming models for reduction in total propagated delays (as well as the number of disrupted passengers), and investigating the use of a passenger-delay calculator and passenger-delay metrics to analyze the impact of flight delays on passengers.
Significant efforts have also been made to measure the reliability of a schedule from a primarily financial perspective.
Airlines have long understood the importance of schedule reliability, but the lack of robust technology has prevented carriers from effectively measuring the schedule’s reliability, as well as better understanding why a particular flight has a high probability of delay and to being able to optimally evaluate schedule reliability as a whole.
As a result, the Sabre Schedule Reliability Model has been developed to help considerably improve an airline’s approach to schedule reliability.
HIGH ESTIMATION ACCURACY HAS BEEN ACHIEVED WITH SABRE SCHEDULE RELIABILITY MODEL BASED ON THE REAL DATA OF A MAJOR AIRLINE. IT SHOWS THE AIRLINE’S ACTUAL ARRIVAL DELAY COMPARED TO SIMULATED DELAY, AS WELL AS THE PERCENTAGES BEING FAIRLY CONSISTENT AND EQUIVALENT IN STATISTICS. FOR EXAMPLE, CLOSE TO 12,000 FLIGHTS HAS A SIMULATED ARRIVAL LATE PERCENTAGE IN THE RANGE OF 10 PERCENT TO 20 PERCENT, AND THE SAME BUCKET OF FLIGHTS HAS AN ACTUAL ARRIVAL LATE PERCENTAGE OF APPROXIMATELY 18 PERCENT.
Measurement And Improvement
Measuring and improving schedule reliability will undoubtedly always constitute a challenging and highly worthwhile effort. Key questions include: How much buffer should be added into block time and ground time to mitigate flight-delay propagation? How should the buffer be changed in the morning, as compared to the ways in which the buffer might be changed to most effectively enhance schedule reliability in the afternoon? Which rules should be used in developing a reliable schedule?
By design, the Schedule Reliability Model brings the reliability-model concept into the schedule-planning phase. In operations, once a schedule is completed, it generally represents a single observation that is not statistically reliable, nor is it viewable on a comparative basis.
Some flights, in actual operations, are delayed, and the causes of delay may be more schedule-driven or more accidental. Furthermore, there’s usually considerable lag time from schedule planning to execution of the schedule, with the inherent result that timely feedback on schedule-reliability performance has almost always been unavailable.
The Schedule Reliability Model is simulation-based, thus enabling statistically reliable results with many repetitions, and the reliability model can be used to evaluate a schedule almost instantaneously, without having to wait on the real-life feedback that can only come from actual schedule execution.
Once flights with a high probability of delays are identified, the model can help find the causes by using critical paths.
The model can also help extract planning rules and parameters for developing a schedule, perform what- if studies (such as the expected or potential impact of shortened ground time through the allocation of more ground resources), and produce economic measures by tabulating missed passenger connections and disrupted crew duties.
In addition, it is designed to model aircraft rotations, as well as crew and passenger flows, including full consideration of resource constraints such as airport gates and allocated airport slots.
SABRE SCHEDULE RELIABILITY MODEL SIMULATES THE FLOW OF AIRCRAFT, CREWS AND PASSENGERS WHILE CONSIDERING RESOURCE CONSTRAINTS SUCH AS SLOTS, GATES AND GROUND CREW.
The Schedule Reliability Model is calibrated using airlines’ schedule and historical operating data, such as block times, taxi times, ground times, seasonality and other potential sources and causes of delays, in company with advanced data-mining and regression techniques.
In practice, the model is validated until the simulation results become close to the actual results of operations.
Estimation accuracy in terms of actual results in comparison to simulated results (as conducted for a major airline) has clearly demonstrated the high quality of the simulation model.
SABRE SCHEDULE RELIABILITY MODEL IS DESIGNED TO BRING THE RELIABILITY MODEL INTO THE SCHEDULE-PLANNING PHASE. ACTUAL OPERATING DATA ARE USED TO CALIBRATE THE RELIABILITY MODEL. THE MODEL HELPS DEVELOP MORE RELIABLE SCHEDULES WITH EXTRACTED RULES AND CRITICAL PARAMETERS.
Greater Schedule Reliability
Once the model is well-calibrated with historical data, it can be used to predict the reliability performance of future schedules (with similar seasonality), and help airlines gain unique and essential schedule-reliability insights.
For example, what are the factors that contribute to high-frequency delays? Which critical rotations can be improved? Which other key performance indicators (KPIs) should be noted and measured?
Every simulation run represents one realization.
Various sensitivity analyses can be conducted under different scenarios to find the most reliable schedule for better operational-system performance, thus quite potentially bettering airlines’ service quality and contributing significantly to enhanced customer satisfaction and, by extension over time, to greater and stronger customer loyalty.