Aviation Integrated Model

AIM uses a modular, integrated approach to simulate the global aviation system and its response to new policies and technologies. AIM consists of seven interconnected modules. The Demand and Fare Module projects true origin-ultimate destination demand between a set of cities representing approximately 95% of global scheduled RPK, using a gravity-type demand model (Dray et al. 2014) where demand is a function of origin and destination population and income, and city-pair trip characteristics such as fare and journey time. Within each city-city passenger flow, itinerary, airport and routing choice (including hub airport for multi-segment journeys) are handled using a multinomial logit model (Dray & Doyme 2019) which assesses choices as a function of itinerary fare, time, frequency and other factors, and itinerary  fares are simulated using a fare model (see Dray et al,. 2019 and references therein) based on segment airline costs and other factors. These models are complemented by simpler models for freight flows and non-scheduled passengers.

The Airline and Airport Activity Module, given segment-level demand, assesses which aircraft will be used to fly these routes and at what frequency, using a multinomial logit model estimated from historical scheduling data (Sabre, 2017) and dividing the fleet into nine size categories. Given these aircraft movements per airport, a queuing model then estimates what the resulting airport-level delays would be (see Dray et al,. 2019 and references therein). Given the lack of long-term airport capacity forecasts, in most cases this delay model is used to estimate the amount of (city-level) capacity that would be required to keep delays at current levels.

The aircraft movement module assesses the corresponding airborne routes and the consequent location of emissions. In particular, routing inefficiencies which increase ground track distance flown beyond great circle distance, and fuel use above optimal for the given flight distance, are modelled using distance-based regional inefficiency factors based on an analysis of radar track data.

Given typical aircraft utilization, the aircraft technology and cost module assesses the size, composition, age and technology use of the aircraft fleet, and the resulting costs for airlines and emissions implications. First, aircraft movements by size class including routing inefficiency from the Aircraft Movement Module are input to a performance model (estimated from outputs of the PIANO-X (Lissys, 2017) model with reference aircraft types and missions for CO2 and NOx, the FOX methodology (Stettler et al. 2013) for PM2.5, and Wood et al. (2008) for NO2). Second, the costs of operating this fleet for the given schedule are estimated based on historical cost data by category and aircraft type (see Dray et al,. 2019 and references therein). Third, emissions and costs are adjusted to account for the current age distribution and technology utilization of the fleet, including typical retirement and freighter conversion behavior (e.g. Dray, 2013). Finally, any shortfall in aircraft required to perform the given schedule is assumed made up by new purchases, and the uptake of technology and emissions mitigation measures by both new aircraft and existing ones is assessed on a net present value basis, as described in Dray et al. (2018), and the impact of this on costs and emissions is assessed.

These four modules are run iteratively until a stable solution is reached. Data is then output which can be used in the global climate, air quality and noise, and regional economics modules. The global climate module is a rapid, reduced-form climate model which calculates the resulting climate metrics (e.g. CO2e in terms of global temperature potential (GTP) and global warming potential (GWP) at different time horizons; see Krammer et al.,  2013). The air quality and noise module are similarly rapid, reduced-form models which provide metrics by airport for the noise and local/regional air quality impacts of the projected aviation system. In the case of air quality, dispersion modelling for primary pollutants uses a version of the RDC code (e.g. Yim et al., 2015). The type of noise modelling carried out depends on whether data on standard flight routes per airport is available, but for all airports noise modelling based on total noise energy is carried out (Torija et al. 2017). The regional economics module looks in more detail at the economic impacts, including benefits such as increased employment as well as costing of noise and air quality impacts.

The output data from the first four AIM modules can also be used more generally as input to external impacts models: for example, the model includes the option to produce detailed emissions inventories which can be input into climate models. Further information on the individual sub-models, on model validation, and on typical model inputs and outputs can be found in the papers cited above and in the model documentation (Dray, 2020. AIM2015: Documentation. http://www.atslab.org/wp-content/uploads/2020/01/AIM-2015-Documentation-v9-270120.pdf).

• Population, GDP and urbanisation projections at regional or country level, by year, across the model time horizon
• Energy price projections (oil, electricity, gas), by year and country/region where relevant, across the model time horizon
• Carbon price projections by year and applicable scheme (e.g. EU ETS, CORSIA) across the model time horizon
• Carbon intensity of electricity generation by year and country/region across the model time horizon
• Technology characteristics (e.g. entry into service date of next aircraft generation, biofuel feedstock costs, etc; dependent on what is being evaluated, default assumptions are included in the model)
• Other policy characteristics as required (dependent on what is being evaluated; e.g. contrail or NOx pricing)