Air Traffic Flow Optimisation With Trajectory Uncertainty
Aviation has changed society dramatically over the past 100 years. The economic and social benefits throughout the world have been immense in “shrinking the world” with the efficient and fast transportation of people and goods. The growth of air traffic over the past 50 years has been spectacular, and will continue in the future, particularly in the growing markets of the Far East. However, since 2000, society’s perception of Air Transport has changed due growing environmental awareness, the rise of oil prices in 2008, and the recent financial crisis. In the future, aviation is likely to face even more radical challenges – with some arising from its own success. Globally civil aviation emitted 666m tonnes of CO2 into the atmosphere in 2008. In response to the likely volume of activity in the future, aviation must bring about step changes in technology and operational procedures to improve its environmental performance by keeping total climate effects at sustainable levels. With the aim of a global solution, International Civil Aviation Organisation (ICAO) is promoting efforts in four key areas: improved technology, efficient operations, effective infrastructure and positive economic measures. The International Air Transport Association (IATA) has declared a target to stabilise net CO2 emissions (carbon neutral growth) by 2020 with a long-term aspirational goal to reduce aviation net carbon emissions by 50% in 2050 compared to 2005 level. ICAO is seeking the mandate to implement the necessary actions for aviation. The current state of air traffic is restricted and congested due to the increasing number of aircraft and an out-dated ATC infrastructure. The Federal Aviation Administration (FAA) has predicted air travel demand to more than double within the next 20 years. Congested skies and flight delays lead to higher fuel consumption and the technology of current air traffic management (ATM) systems is dated in comparison to the demands of emerging fleets. The development and implementation of a new ATM concept is proposed by the United States (NexGen) and the Single European Sky ATM Research Programme (SESAR) in Europe. These systems are based on increased data sharing across all stakeholders, ie. airports, ATC, weather monitoring, airlines, etc. Improved operational practices and aircraft deployment across a network can reduce fuel-burn by about 5%, through measures such as better flight planning, speed management, aircraft selection, equipment weight reduction and taxiing with one engine shut down after landing. Improved air traffic control resulting in more direct routes and reduced delays could reduce overall fuel burn by 6 - 12%. In 2006, in Europe alone, additional distance flown due to non-optimal routing was 441 million km, equivalent to about 4% CO2 emissions. This paper proposes the development of a realistic Air Traffic Flow Optimisation framework that defines optimal trajectories based on intended destination, aircraft performance, conflict detection and resolution and weather conditions. It is intended to explore the fuel saving benefits of utilising prevailing tailwinds. A key element in this Air Traffic Flow Optimisation framework is that it is information rich, i.e. all necessary information is available and accessible, and flight data and intentions, are known in advance. Wind Model En-route wind data are commonly provided by the local weather or meteorological services agencies. Flight crew use this information for flight planning to determine if favourable tailwinds are available and if adverse weather conditions exist along the route. Wind predictions, even to very high altitude, are also used for balloon flights as well launch and recovery of spacecraft. In the USA, the National Weather Service of the National Oceanic and Atmospheric Administration (NOAA) provides this service, while in Australia it is the responsibility of the Bureau of Meteorology (BoM). The BoM also provides wind data for the Indian Ocean, Pacific Ocean and Tasman Sea regions. Wind data is usually provided as wind direction and wind speed over a regular grid for a number of discrete flight levels and at various times of the day, usually 6 hrs intervals. The understanding of the psychical properties and behaviour of the atmosphere as increased significantly over the past decades that show that correlation between measured data and forecasted data is within 1 m/s. As wind properties can be represented as a vector, 4D interpolation will be performed to estimate the wind properties at every point. The cubic interpolation function utilises a third order polynomial curve fit, whereas the spline interpolation function utilised a higher order function. The small difference between the computational times of these functions in addition to this factor was the primary decision driver towards the selection of the cubic interpolation function. The cubic interpolation function as utilised in the proceeding research is governed by the convolution sum. The format of the initially retrieved wind vector data was provided as magnitude and true bearing values over the desired longitude, latitude and time domain. The first interpolation step was designed to increase the number of available wind data points over the initial time window. A second interpolation step takes place after the selection of the optimal aircraft flight level, in order to increase the density of data points in the local longitude and latitude directions. Two separate four dimensional interpolation functions were required for each of these steps to generate interpolated data for both the local wind bearing and magnitude data sets. Fig. 3 illustrates the difference between the wind vector data initial input fidelity and that of after the second interpolation step. The red arrows indicate the local wind bearing and magnitude values of the initial wind vector data for the given longitude and latitude coordinates at a sample flight level and time instance. For the purpose of this research, level of accuracy in the initial wind vector model is grossly insufficient to ensure the generation of an optimal flight path. The blue arrows demonstrate the fidelity of the wind vector data after computing the second interpolation function. The introduction of basic aircraft flight parameters was necessary to further evaluate the suitability of each flight level for an optimal flight path. These metrics include the approximate flight bearing of the aircraft from origin to destination, and the time window over which the flight is occurring. To determine the most effective flight level to operate at, the total dynamic pressure of each flight level was used as a comparison metric.