Greenhouse gas emissions from international aviation and allocation
In chapter 8 we have looked briefly at emission inventories and models. Another potential source of fuel consumption data is airline reporting on average overall yearly fuel intensity. Such reporting holds the advantage over models that the data reflect actual consumption data for actual flights with certain loads, i.e passengers and freight. Such data are unfortunately not generally publicly available as discussed further throughout this chapter. This chapter discusses the availability and usability of such data.
The fuel intensity per passenger kilometre of commercial civil air transport has been reduced by approximately 50% since the early 1970s. The use of more fuel-efficient aircraft engines and the introduction of bigger aircraft accommodating more seats per aircraft in combination with an increase in the average stage distances reduced the fuel use per available seat kilometre (ASK). The improvement in the specific fuel consumption has furthermore reduced the necessary amount of fuel that has to be carried on flights of comparable distances leading to additional fuel savings. Furthermore, the operation at higher passenger load factors has contributed to reduce the fuel use per revenue passenger kilometre (RPK)21.
The trend in the average specific fuel consumption per revenue passenger kilometre in commercial civil air transport is illustrated in Figure 18 that plots a number of different estimates that are given in the literature for all the US airlines [Davis 1999], for British Airways [British Airways 1999a], for all the UK airlines [Martin and Shock 1989] for the World's scheduled fleet [Greene 1990] [Balashov and Smith 1992] [Gardner et. al. 1998] and for the IATA fleet [Dobbie 2001]. We note that the estimates that are shown here include the total amount of fuel consumed by the airlines in question. The estimates are for various groups of airlines that operate at different routes at varying passenger load factors and freight load factors using fleets of various aircraft mixes. The major part of the fuel consumed by airlines is attributable to the carriage of passengers, but some is related to freight transport. Thus, the estimates for the average fuel consumption per passenger kilometre that are shown in Figure 18 can be said to be somewhat overrated. Section 9.3.2 discusses the relative importance of freight in passenger airline activities and analyses how the fuel consumption can be distributed between passenger and freight transport weights respectively.
This section analyses the specific fuel intensity of different types of aircraft, based on recent information from some European and Asian airlines' yearly environmental audits as well as some recent operating statistics for the American air carriers that are submitted to the US Department of Transport and some information from a number of academic studies that analyse the fuel intensity of aircraft in some earlier years.
Figure 19 plots the average specific fuel consumption per available seat kilometre (ASK) and per revenue passenger kilometre (RPK) for a range of different aircraft types for the average stage lengths at which they are used. Most of the data refer to the use in recent years, but a few older data are included as well (see the notes to Table 8 and Table 9 for a further description of the data that are included in Figure 19). The data includes subsonic jets and turboprops in operation in various years from the beginning of the 1970s and onwards. The data for the fuel consumption per ASK and per RPK for each aircraft type refers to the usage cycle for a specific airline, or a number of airlines, in a specific year, including the specific load factor and the average stage distance flown by type of aircraft in that year.
Some data for the older aircraft types are derived from academic studies analysing 1970s and 1980s fuel intensity of a number of American and British airlines by aircraft type and are summarised in Table 9. The data for the aircraft types that are currently in use are summarised in Table 8 and represents data for use in the period between 1998-2000 of airlines that are situated in the United States, in Europe and in the Asia/Pacific region. It should be noted that different airlines use different methodologies for calculating the specific fuel consumption per passenger kilometre. For example, Lufthansa subtracts the fuel attributable to lifting the bellyhold freight in the company's passenger aircraft. Therefore, the further analyses' in the following sections primarily concentrate on comparisons of data between airlines or groups of airlines for which the data are consistent, unless otherwise is mentioned.
The fuel use per revenue passenger kilometre (RPK) is higher than the fuel consumption per available seat kilometre (ASK), due to nonoptimal passenger load factors. Aircraft that are used for short-haul regional flights are typically operating at load factors below average and are typically quite fuel intensive, as compared to aircraft that are used at medium-haul and longhaul, using normally around 50-90g per RPK (Table 8 and Figure 19). The most fuel-intensive subsonic passenger aircraft that are currently in use (among the airlines studied here) are low-capacity regional turboprops and jets using up to 119g per RPK.
The aircraft used at medium-haul typically use around 30-50g/RPK, but the most fuel-efficient types consume less than 20g/RPK. However, the old DC9s operating in the medium-capacity market use up to around 111g per RPK on average when operated on short-haul routes at below average load factors.
Aircraft that are used for long-range flights normally consume around 40- 50g/RPK. The most fuel-efficient long-range aircraft consume below 30g/RPK whereas the least efficient types consume up to 60g/RPK. The supersonic Concorde, that has not been included in Figure 3.4, is in a class of its own among the long-range aircraft, using 175g/ASK and 313g/RPK. That is, the Concorde use about ten times as much fuel per revenue passenger kilometre as do the most efficient subsonic long-range jets. Furthermore, the Concorde cruise at much higher altitude (18 kilometres) than subsonic aircraft (typically around 10-12 kilometres), leading potentially to a more severe environmental impact per kilo of fuel burned than aircraft cruising at lower altitudes.
* US airline average, ** Turboprops, NR Not reported by any of the airlines, C In
charter all-economy class configuration, D In domestic all-economy class configuration, L
A look at the data presented in Table 8 and Table 9 reveals the impact of the technological improvements to some main aircraft models that were operated by the American air carriers in 1973 and 1998 respectively. For example, in 1973 the B737s consumed around 59g of fuel per ASK on average and had 94 seats on average. For comparison, the B737-500s that are operated by American major airlines in 1998 use 37g of fuel per ASK and have 110 seats on average. A second example for comparison is the long-range B747. In 1973 the B747s used 43g per ASK and had 332 seats on average. In 1998 the B747-400s use 32g per ASK and have 383 seats on average. A third example for comparison is that the 233-seat DC-10 tri-jet introduced in the early 1970s used around 44g per ASK while the 290-seat B777-200 twinjet introduced in the mid-1990s consume around 28g per ASK [Sarames 1984] [Aircraft Economics 1999f and 1999c]. Many airlines are today replacing their current aircraft by bigger types and this makes it possible to operate at lower specific fuel consumption.
Generally, the average yearly passenger load factors of commercial air carriers have been improved through the last decades from around 50 percent in the early 1970s to around 70 percent currently [ICAO 1999a]. There are considerable differences among airlines concerning load factors. Passenger load factors are reported from around 50% to above 75% by scheduled airlines, although most major airlines operate above 65% [ICAO 1998f] [AEA 1998]. European charter carriers generally operate at above average passenger load factors, some of them close to the optimum, one example being Premiair reporting a passenger load factor of 98% in 1999 [Premiair 2001].
The weight load factors, that is the weight of passengers and their baggage plus the weight of the freight transported as belly-hold over the available capacity (measured as available tonne kilometres, are generally lower than the passenger load factors. Freight's share in total scheduled traffic range from less than 10% to above 40% for the World's major airlines [Cranfield College of Aeronautics 2000b]. Freight's share of the total weight transported is generally higher on long-haul routes than on medium-haul while being almost insignificant on short-haul [AEA 1999] [DOT 2001], see Section 9.2.2 for a further discussion of this issue.
The fuel use per revenue passenger kilometre and per freight tonne kilometre is generally reduced at higher load factors. However, the total aircraft fuel use increases as the load factor increases, because of the weight that is added to the aircraft when carrying additional passengers and freight and this is also reinforced by the aircraft carrying more fuel. The connection between load factors and the fuel-burn per seat vary according to the aircraft type and the distance flown. A recent study proposes, that for modern medium- to largecapacity aircraft such as B747-400, B777-200, B757-200 and B737-700, the additional fuel burn at high load factors is rather small. For example, an increase in the passenger load factor from 70% to 100% is suggested to generally lead to an increase of less than 5% in the total fuel use on trips of average lengths for those aircraft [Daggett et. al. 1999]. For smaller shorthaul aircraft as well as for some older medium-capacity jets the fuel consumption increase considerably more than what is suggested for modern mediumhaul and long-haul jets [IPCC 1999, p. 280].
An example of the importance of the freight load factors for the fuel consumption per revenue freight tonne kilometre transported by allcargo carriers is illustrated in Figure 20. In 1998, the main types of aircraft used by the three major US all-cargo carriers (UPS, DHL and FedEx) operated at weight load factors of between 47% and 67%. The fuel consumption per revenue freight tonne kilometre is therefore around 1,5 to 2 times as high as the fuel consumed per available tonne kilometre, that is the available capacity. The aircraft shown to the left in Figure 20 are operating at short distances with average revenue loads of between 10-30 tonnes and those to the right are long-haul aircraft with revenue loads of up to 65 tonnes.
Airlines operate aircraft that are configured for different purposes. For example, some aircraft are configured in all-economy class high-density seatconfiguration while others are configured with two or three different classes, where the seats in the business class and first class sections are more spacious. Thereby three-class seat-configuration aircraft have lower seat density than all-economy class configured aircraft. Some scheduled airlines operate aircraft featuring high-density seat configuration at domestic routes while using lowdensity seat-configuration aircraft on international routes. For instance, All Nippon's B777-200s accommodates 376 passengers in all-economy seatconfiguration but only 250 in the three-class international version, while the B747-400 accommodates 569 passengers in the all-economy seatconfiguration and only 337 in three-class mode.
Similarly, European charter carriers and many low-cost scheduled carriers generally use high-density seat-configuration aircraft, thereby operating at lower fuel consumption per available seat kilometre than scheduled flag carriers. Many examples can be drawn from the data shown in Table 8, showing considerable differences in the seat-configurations, especially in the segment for long-haul aircraft. For example, the European charter carrier Premiair operates A330-200s and A330-300s accommodating 30% and 50% more seats respectively than similar aircraft types operated by Swissair and Cathay Pacific respectively. A similar comparison in the mediumrange segment shows that the European charter carrier Air 2000 operates A320s and A321s with 20-25 percent more seats than Lufthansa's aircraft. The operation at above average passenger load factors and the negligible amounts of freight loads combined with the use of new-generation aircraft in highdensity seat-configuration, explain why the fuel intensity of Premiair and Air 2000 is around half of the global average for the world fleet.
Some airlines publish environmental reports giving estimates for their fleets' average yearly fuel intensity. Such data are compared in Table 10 and Table 11 for a number of airlines.
The airlines represented in Table 10 report average fuel-burns of between 24- 46g/ASK, and between 26-81g/RPK. European charter airlines are generally the most fuel-efficient. Scheduled airlines that are operating relatively old aircraft mainly at short- and medium-haul routes, such as SAS in 1998, are more than twice as fuel intensive as the most efficient charter airlines. Commuter- and regional airlines, like Lufthansa City Line, which often operate at below average load factor at short-range routes, generally use around twice as much fuel per passenger kilometre than do the major scheduled airlines. It should be noted that the yearly averages reported by airlines constantly changes as a consequence of changes in the composition of their aircraft fleets as well as changes in load factors and other operating characteristics.
*The figures in brackets represent airline estimates where fuel used for lifting
bellyhold freight in passenger aircraft is subtracted. Note that the three airlines that
give such estimates for the fuel which is attributable to belly-hold freight all use
different methodologies in the calculation
Similarly, Air France and Lufthansa Cargo report their average fuel use per revenue freight tonne-kilometre (RFTK) performed. These are averages over the fuel consumed for freight transport in their all-cargo freighters and the fuel that is attributable to lifting the belly-hold freight in their passenger aircraft. According to these estimates from Air France and Lufthansa Cargo around five times as much fuel is consumed for transporting one tonne of freight one kilometre as is used per passenger kilometre on average. However, this is an average over a number of different aircraft models that operate at different stage lengths. The specific fuel consumption of airfreight on short haul in passenger aircraft can be more than twice as high as the average. Furthermore, the fuel consumption per RFTK in some of the allcargo aircraft that are operated by Air France, Lufthansa Cargo, KLM, UPS, FedEx and DHL are shown in Figure 20. These data show that the specific fuel consumption in all-cargo freighters ranges from around 165g per RFTK to 644g per RFTK. The lowest figures reported are for long-haul MD11s that operate at average loads of above 60 tonnes while the highest figures represents old B727s that operate at average stage distances of around 500- 1300 kilometres carrying average loads of around 10-20 tonnes, see Figure 20. The average specific fuel consumption of the operations performed by the three Major US all-cargo carriers in 1998 can be estimated from the data described earlier in Figure 20 at around 237g per revenue freight tonne kilometre (RFTK) transported and some 138g per available tonne kilometre (ATK). Note that these data do only cover the most used types of aircraft by the carriers in question [Aircraft Economics 1999d].
a) Average over a number of models, b) B747-F, c) B747-300F, d) various aircraft
models, e) B747-200F, f) MD11.
The fuel intensity estimates for the different airlines that are presented here are not directly comparable between airlines because of the differences in reporting methodologies. One example is that some airlines subtract a part of the fuel consumption which is attributable to freight transport in passenger aircraft, whereas others include this use in the estimate for the specific fuel use per revenue passenger kilometre. All airlines carry both passengers and freight. Some freight is carried in freight-only freighter aircraft, some in combi-aircraft where a freight section replaces a part of the passenger section, while some is carried as belly-hold freight in standard passenger aircraft.
For airlines that carry much freight in passenger aircraft the fuel used for lifting the freight can contribute to a rather high proportion of the total fuel consumption. For example, British Airways' average fuel consumption per RPK is 49g for the whole passenger fleet on average, but if taking freight into account the efficiency improves to 35g per RPK (see the figure in brackets in Table 10) [British Airways 1999b, p. 21]. Similarly, Lufthansa's and Air France's Scheduled services uses 42g and 49g per RPK respectively, but the numbers are reduced to around 37g and 42g when subtracting the fuel used for lifting belly-hold freight. The fuel consumption figures for scheduled airlines can be more realistically compared to charter carriers if using estimates for the fuel consumption where the fuel use attributable to lifting and carrying freight is subtracted, because charter carriers generally transport negligible amounts of freight.
The division of fuel use between passengers and freight is not straightforward. For example, British airways attributes 30% of their fuel use to freight because around 30% of its revenue load (measured in tonne-kilometres) is freight [British Airways 1999b, p. 21]. Other airlines argue, that transporting one tonne of freight requires less fuel than transporting one tonne of passengers and luggage. For example, Lufthansa attributes 1,7 times as much fuel to passenger weight than to freight weight [Lufthansa 2000b, p. 51] while Air France uses a factor of 1,4 for medium-haul aircraft and up to a factor of 2 for some long-haul jets [Air France 2000, p. 9]. These ratios are supposed to account for the weight and space within an aircraft that is acquired for inflight passenger services such as seats, galleys, flight crews, catering supplies etc. Only three of the airlines mentioned in Table 10 have reported specifically on both the passenger load factors and the freight loads in their passenger aircraft.
Another example of the differences in reporting methodologies between airlines is the use of different assumptions for the average weight of passengers and their baggage when calculating the ratio between the weight that is attributable to passengers and freight respectively.
Yet another example of the differences in airline reporting methodologies is that for some scheduled airlines the passenger load factor refers to passengers that have paid a certain percentage of the normal fare. Children oftentimes get discounts or travel for free, as do frequent flyers having earned bonus points. The actual load factor is therefore sometimes higher than seen from the statistics and the fuel use per passenger may be somewhat lower.
This section takes a closer look at the specific fuel consumption of the American air carriers. The data material shown here covers the overall traffic performance of all the US carriers in the years 1982 and 1999. These data are not biased by the fuel consumption of all-cargo carriers that was included in the overall data shown in Figure 18. However, the data still include the fuel consumption that is attributable to belly-hold freight in passenger aircraft.
Figure 21 shows the average yearly specific fuel consumption per ASK and per RPK of US air carriers on domestic routes in 1999. The specific consumption varies between 27g and 64g per ASK and between 36g and 102g per RPK, if excluding a single carrier that uses some 74g per ASK and 160g per RPK. Among the Major US airlines that performed around 90% of the domestic revenue passenger kilometres in the United States in 1999 the specific consumption ranges between 30-37g per ASK and 44-53g per RPK. The large regional carriers, such as American Eagle and Continental Express, typically use around 50% more fuel per ASK and per RPK than the Major air carriers do. The overall average specific fuel consumption on domestic routes is around 35g per ASK and 50g per RPK. In 1982 the average specific consumption per ASK was about 43g suggesting a reduction of approximately 8g per ASK in the period or about 19%.
We note that the data availability on fuel consumption by American air carriers is much better than for European and Asian carriers because the American air carriers are required by law to report their operating characteristics to the American Department of Transportation. Similar arrangements do not seem to exist in Europe or in Asia.
9.3.2 Methodologies for allocating airline fuel consumption between passengers and freight loads in passenger aircraft
This section quantifies how much freight that is transported as belly-hold in passenger aircraft by different aircraft and airlines and discusses how much of the fuel that is attributable to passenger and freight revenue weight respectively on different routes.
Currently, freight and passengers account for around 30% and 70% respectively of the total number of revenue tonne kilometres that is performed by the World's airlines. However, some of this freight is carried in allcargo aircraft.
In 1999, 29 billion revenue freight tonne kilometres (RFTKs) were transported by the American air carriers [DOT 2000, p. 326]. Eight allfreight carriers alone carried more than half of this total. That is, less than 13 billion RFTKs were carried by the passenger airlines, representing some 18% of the total amount of RTKs transported. The average weight share of freight is therefore less than 18% for the US passenger carriers.
Similarly, in Europe, around 44% of the freight that is carried by the scheduled airlines is transported in passenger aircraft and the residual in allcargo aircraft. The freight's weight share in the total scheduled passenger services is 23%. The share is 29% in international long-haul scheduled passenger services, 10% in international short/medium haul scheduled passenger services and around 4% in domestic scheduled passenger services [AEA 2001].
In Japan, All Nippon Airways report the weight shares of freight in their passenger aircraft at 13% on domestic routes and at 36% on international routes [All Nippon Airways 2000b].
These data suggest that, as a general rule of thumb, Asian carriers transport the highest shares of freight in their passenger aircraft while the US passenger airlines transport a lower share of freight than the European passenger airlines. This is probably due to the large share of domestic traffic performed by the US air carriers that accounts for around two-thirds of all the RTKs and about three fourths of all the RPKs [DOT 2000, p. 323].
Generally, the overall statistics mentioned above suggest that the freight share is higher in long-haul traffic than in medium-haul and short-haul. A look at some statistics on the freight weight shares in individual aircraft confirms this picture (see Figure 22).
As was touched upon briefly in section 9.3 the airlines use different methodologies for the allocation of their fuel consumption on passengers and freight. Most airlines attribute all the fuel consumed to their passenger services. Some airlines attribute the same amount of fuel to a tonne of freight as to one tonne of passenger weight (including their baggage). Other airlines multiply the passenger weight with a factor of between 1,4 and 2 to account for the weight that is attributable to a number of in-flight passenger services (see section 9.3 for a further description of this issue).
The average revenue of the World's airlines per tonne of freight is around 60% lower than the average revenue for a tonne of passengers [ICAO 1996c and 2000d]. One could argue that this factor should also be taken into account in a discussion of which methodology that could potentially be used for the allocation. Therefore, if the fuel is distributed between freight and passenger loads according to their revenue shares, the weight of the passengers should be multiplied by a factor of around 2,5.
The four different methodologies for distributing the fuel between freight and passengers are illustrated in Figure 22. Not surprisingly, the most extreme difference in the estimate for the specific fuel consumption per revenue passenger kilometre appears between the methodology where all the fuel is attributed to passenger transport and the methodology where the fuel is distributed evenly between passengers and freight on an equal weight basis. In the latter case the specific fuel consumption per RPK is reduced by around 24-35% for long-haul trips in a B747-400 and by around 5-13% on mediumhaul trips with B757s and A320s. The implication of this finding is that the figures for the specific fuel consumption of aircraft and airline operations that includes the fuel which is attributable to freight (for example those figures that are shown in Figure 19) would typically be reduced by 5-13% on medium range and by 24-35% on long-haul. We note that these are rough estimates and may differ between airlines and between different types of aircraft (see Figure 22)
The selected aircraft shown in Figure 22 are arranged with the most fuelefficient aircraft, measured in fuel consumption per RTK, on the left hand side of the figure. The B767-300/300ER is the most fuel efficient when considering the fuel consumption per RTK and therefore also per RPK when distributing the fuel consumption on an equal weight basis between passengers and freight (methodology 2). The relative difference between the specific fuel consumption figures of methodology 1 and 2 is greatest for the MD-11s, the B767s and the B777s. For these aircraft RPK2 is between 35- 38% smaller than RPK1. For the DC-10s, the 747s, the B767-200s and the A300-600s RPK2 is between 18-30% smaller than RPK1. That is, if comparing the specific fuel consumption figures of these long haul aircraft to the most fuel-efficient medium haul aircraft (B757-200s, A320s and B737- 800s) they are at level or even more fuel-efficient if using methodology 2.
Sources: Fuel consumption from [DOT 2001] and freight loads from [Air Transport Association 1999, 2000e and 2001].
In this section some of the problems related to the issues described in this chapter are presented and discussed briefly.
As we have shown throughout this chapter, the airlines that currently report their fuel intensity in environmental reports do not use a common standard. The fuel intensity estimates reported by different airlines are not directly comparable because of the differences in reporting methodologies. One example is that some airlines subtract a part of the fuel consumption which is attributable to freight transport in passenger aircraft, whereas others include this use in the estimate for the specific fuel use per revenue passenger kilometre. As we have illustrated in this chapter, the division of fuel use between passengers and freight is not straightforward.
In the United States all airlines of a certain size are required by law to report their operating statistics to the Department of Transportation (the socalled form 41 arrangement). Therefore, in the United States, a comprehensive database exists with data for the fuel consumption of airlines and their aircraft spanning back several decades. This type of data can be used to make comparisons between airlines and for indexing their fuel efficiency in the way it was shown in Figure 21 in section 9.3.1. To the knowledge of the author such data are not systematically reported to the same detail to governments, ICAO or elsewhere from airlines registered in other countries, although most airlines almost certainly gather such data for internal purposes.
One interesting question is whether it would be possible to establish some sort of global reporting requirements for all the World's airlines in line with the US Form 41 establishment. Since ICAO and CAEP are currently investigating possibilities for setting up voluntary agreements with airlines on reducing their specific emissions of CO2 that process might involve setting up a scheme for airline reporting of fuel consumption and emissions. Furthermore, ICAO and CAEP are currently investigating the possibility to set up a global system for emissions trading. Such a system may come to involve the setting of an emission cap and the allocation of certain emission quotas to airlines and may also involve new reporting requirements for airlines.
Even though airline fuel consumption could be estimated using bottomup modelling, for example by using the Corinair-model, actual fuel consumption data from airlines may be needed because airlines might not be likely to accept being accredited for modelled fuel consumption data. At least at present, the models that have been constructed to calculate emissions from air traffic on a global scale do not contain a comprehensive database on flights actually being performed and furthermore relies on calculating fuel consumption and emissions by using less detailed aircraft categories than those used in the detailed CORINAIR methodology. Furthermore, all the models constructed to date are disadvantaged by not containing detailed information on the actual passenger- and freight loads transported within the aircraft. These loads may become relevant for example in the case that airlines should become required to reduce their emissions per passenger kilometre and per freight tonne kilometre in a voluntary scheme. Detailed data on the passenger and freight loads may also become necessary for some of the more sophisticated models for allocating emissions from international aviation to Parties that are discussed in chapter 10. However, it should be mentioned that these sophisticated models of allocation currently do not seem to be the most likely to be chosen if Parties to the Climate Convention should agree upon implementing an allocation option.