1.2 Background
The development of the commercial air transport and its continuous growth has leads to cause more and more massive delays along the years, which makes from delays one of the biggest challenge for the air transport industry. 'Punctuality' is one of the important performance indicators for operational air transport performance. It represents the proportion of flights that arrive "on-time" as opposition to delayed. This concept has many definitions but according to of the United States Department of Transportation's Bureau of Transportation Statistics (US DOT's BTS), a "flight is considered delayed if it arrived at (or departed) the gate 15 minutes or more after the schedule arrival (departure) time as reflected in the Computerized Reservation System" (2010). This definition is retained since data collected from the United States (US) will be used for this thesis.
Figure 1 depicts the schedule adherence on intra US flights between 2000 and 2009. Despite an improvement between 2000 and 2003, the percentage of delayed flight increased continuously from 2004 to 2008. 2009 and 2010 have seen their percentages of delayed flights decreasing materially but this has to be related to the economic crisis which materially affected the traffic.
Figure 1.: Proportion of intra US flights delayed [1]
Source: BTS, 2010
Because of the high level of competition in the airline industry and the high degree of public exposure, airlines always have the desire to ensure that their flights reach their destination within the accepted 15 minutes window. Their reputation is based on the services they offer, whose punctuality is part of. Also, punctuality raises many issues like time slot allocation, future flight cockpit and cabin crew schedules, curfews, aircraft maintenance plan, or resource utilisation.
Furthermore, On-Time Performance (OTP) is crucial for airlines since it influences shareholders' decision to invest or not in the company. Indeed, OTP directly impacts direct operational costs due to the disturbances in operations but also indirect operating costs essentially through the loss of market share (Wu, 2003). Therefore considerable effort is expended by airlines to improve their OTP.
Departure delays 'of a turnaround aircraft is influenced by the length of scheduled turnaround time, the arrival punctuality...as well as the operational efficiency of aircraft grounds services" (Wu, 2003). As for arrival delays, the BTS data shows that one third of the arrival delays was the result of a late arrivals of the previous flight (2010).
Thus a single initial delay can lead to the whole network disruption through its propagation throughout the airline's resources. Such delays caused by a previous one are classified as 'reactionary' or 'knock-on' delay. Those that propagate through the aircraft are known as rotational delay and those through the crew as non-rotational delay.
Reactionary delays are critical because of the costs they generate at two different levels. First they involve at the scheduling stage strategic costs due to the resources committed as contingency for delays. Secondly the day of operation the emergency measures taken incurred tactical costs (Cook, 2007). However, they are most of the time not well understood and consequently anticipated. Yet they are the cause of an even more serious impact than the root delay itself (AhmadBeygi, 2008). The CODAs annual DIGEST 2008 stresses that 40% of the total delayed minutes is caused by reactionary delays (2009).
As the consequences of delay can be very severe, airline take measures to try to minimise their propagation. The allocation of slack time or "buffer time" in the schedule is one of the existing management strategies to enhance the level of punctuality (Al-Haimi, 1991). Buffer time is extra time that can be used either in the turnaround time to prevent against the risk of delay propagation due to a late incoming aircraft or irregularities during the ground activities. Or it can be inserted in the flight time in case of disturbance. In that case the schedule flight time encompasses the actual flight time and the slack time (Cook, 2007). This practice requires a high level of predictability of operations since it will impact the extent to which slack time will be added in the flight duration. However, although adding slack time allow for a better management of punctuality it is not without knock-on effect on the costs. Indeed, the major consequence of padding schedule is a penalty in resources utilisation and hence in productivity and profits. Considering that an A320 buffer minute's cost is equal to €49, "cutting 5 minutes on average of 50% of all European schedules thanks to higher predictability would be worth some €1,000 million per annum, through savings or better use of airline and airport resources" (Rapajic, 2009). The cost of buffer time is equivalent to the strategic cost discussed earlier. However it is hard to calculate because buffer time also brings savings through the benefits realised by avoiding delays that are not tangible. However, Cook tried to calculate this cost by distinguishing three cases have to be distinguished. First a shorter buffer will involved the strategic cost of it and the cost of delay the day of operation it occurred. Secondly a buffer covering just the length of the delay will avoid any tactical cost and bring the satisfaction of passengers to be on-time. Finally a too long buffer means that the airline has paid an excessive strategic cost but again passengers are satisfied (Cook, 2004). Depending on assumptions about the aircraft model, the number of late arrival minutes, the usage or not of the minute and the moment of occurrence of the delay (in case of, taxi time, holding or en-route) he found a minute airborne buffer time cost ranging from 5 to 351€.
Therefore buffer time can be a very costly way to reduce delay if not used wisely. The theoretical buffer that should be added to airline's schedules should be 'up to the point at which the cost of doing this equals the expected cost of the tactical delays they are designed to absorb' (Guest, 2007).
Table 1.: Major inflation in the block time onUS routes between 1996 and 2010
Airline
Route
Block time, March 1996 (hours : minutes)
Block time, March 2010 (hours : minutes)
Added minutes
Percent change
Delta
Atlanta-Orlando, Fla.
1:14
1:43
29
39%
Delta
Chicago-New York JFK
1:54
2:36
42
37%
Southwest
Phoenix-Las Vegas
1:00
1:20
20
33%
American
Chicago-Newark
1:58
2:30
32
27%
US Airways
Charlotte, N.C.-Atlanta
1:00
1:16
16
27%
United
Los Angeles-San Francisco
1:17
1:36
19
25%
United
Chicago-Denver
2:17
2:46
29
21%
Continental
Atlanta-Newark, N.J.
2:07
2:32
25
20%
Southwest
Las Vegas-Austin, Texas
2:15
2:40
25
19%
Delta
New York JFK-Los Angeles
6:00
7:04
64
18%
AirTran
Philadelphia-Atlanta
2:00
2:22
22
18%
Continental
Ft. Lauderdale, Fla.-Newark, N.J.
2:52
3:15
23
13%
American
Miami-New York LGA
2:55
3:15
20
11%
Alaska
Portland, Ore.-Los Angeles
2:12
2:22
10
8%
Source: McCartney, 2010
Buffer times are widely used today in the airline industries. Table 1.1 gives a comparison of block times for a selection of US routes during the last 15 years. It shows that the schedule block time has increased and even up to more than 30% on some routes. At the same time, the number of flights arriving early is increasing, meaning that airline tend to pad to much their schedule which implies a loss of money. This leads to the conclusion that either improvement has to be done when calculating the length of buffer times or that airlines increase intentionally their schedule time to ensure a high level of punctuality.
In order to calculate the necessary buffer time, airlines base their calculations on historical data performance. However, the resulting schedules are most of the time not optimised in terms of cost, meaning that some contributing factors are probably not taken into account during the generation of schedules (Wu, 2002). Different algorithms have shown that airline schedule could be optimized either by reallocating the slack time differently in the schedule or by adding more or less buffer minutes (Wu, 2002). The number of research done about the contributing factors to the variation in buffer time is limited and in addition the few available studies often use the data for one particular airline. Mayer studied the effect of the number of competitors and the quality of airports as a major, medium, small or non-hub to the airline schedule (2003). However, one can imagine that contributing factors to the distribution of delays could influence the distribution of buffer time.
This research will try to understand the factors that influence the repartition of delay and hence buffer times and also evaluate the efficiency of buffer time in reducing delays.
1.3 Objectives
Research Question:
The aim of this research is to investigate the contributing factors to the variation in block-time buffer times.
Research Objectives:
Three research objectives have been established to achieve the research question.
To investigate the interplays between on-time performance, delays propagation and the implementation of buffer time in airlines' schedules.
To identify the contributing factors that influence the distribution of delay and buffer times in airline's network and investigate the predictability of delays
To give some recommendations to determine a sound number of buffer minutes to allocate in the schedule according to some factors in order to improve the chance to be on-time without arriving systematically early.
1.4 Thesis Structure
The study is organised in five main parts, each divided in subparts. The first chapter gives the background information related to subject to briefly familiarise the reader with the research. It provides the research question and objectives of the paper and finally includes an outline of the methodology and chapters of this thesis. This is followed by the second chapter whose aim is to provide a review of the relevant literature related to the topic. Both theoretical and empirical studies are covered to give a general idea about the subject. It is divided in three parts, examining first the importance of on-time performance for airlines, secondly the mechanism of propagation of delay and its impact and finally the implementation of buffer time in airline schedule. The methodology is presented in the third chapter. It includes an explanation of the data collection method and a description of the data used for the research. It also presents the hypothesis of the analysis and a justification of the statistical tests used to test them. Finally the limitations of the method used to answer the different objectives are presented. After those three first chapters, the results of the analysis are presented. In this case, it describes in a first part the systematic evolution of punctuality in relation with some factors and in a second part the contributing factors to the variation of buffer times. A discussion and interpretation of the results are provided in the same chapter. Finally, the conclusion of this thesis encompasses the summary of the key findings and some recommendations for the allocation of buffer time are drawn for the analyses. It ends up with a presentation of the possible area for further research.
