A review of research on risk and safety modelling in civil aviation

Safety is considered as some of the most important operational characteristics of contemporary civil aviation. An extensive regulatory structure has been established to supplement the private airline, airport and air navigation systems, incentives to limit the risks of flying. This paper reviews the research on risk and safety modelling in civil aviation. In such a context, the basic concepts and definitions of risk, safety and their evaluation are described. The review focuses on four categories of models for safety assessment: causal for aircraft and air traffic control/management operations, collision risk, human factor error and third-party risk.     

A systematic approach to assess the effectiveness of airport noise mitigation strategies

Regional airports are often located very close to the urban area they serve and the increasing traffic rate that many of them have experienced in the last years has produced several impacts on the communities living close to the airport area, mainly aviation noise. If not properly managed, noise impacts produced by airport operations can cut down significantly the development of airport air traffic with direct effects on the economic and territorial systems. Aeronautical noise has greatly reduced in the last decade, due to aircraft design technological improvements and more severe regulations. However, the noise reduction during a single event does not make the issue of the airport location – and then the whole noise impact – less significant. This paper proposes an assessment process to evaluate the effects of actions adopted to reduce airport noise impacts on populated areas. Both airport-related factors – such as number of take-off; day-evening-night distributions of movements; aircraft type; flying paths – and land-use characteristics have been considered and combined in a density index that synthesizes the impacts of airport noise on the territory. The assessment process has been tested on a real case, the airport of Bologna in Northern Italy. The predicted results, compared with available real data for the test case, are significant and encourage the use of the proposed assessment process as decision support system for the airport management.     

Estimating entry counts and ATFM regulations during adverse weather conditions using machine learning

In recent years, convective weather has been the cause of significant delays in the European airspace. With climate experts anticipating the frequency and intensity of convective weather to increase in the future, it is necessary to find solutions that mitigate the impact of convective weather events on the airspace system. Analysis of historical air traffic and weather data will provide valuable insight on how to deal with disruptive convective events in the future. We propose a methodology for processing and integrating historic traffic and weather data to enable the use of machine learning algorithms to predict network performance during adverse weather. In this paper we develop regression and classification supervised learning algorithms to predict airspace performance characteristics such as entry count, number of flights impacted by weather regulations, and if a weather regulation is active. Examples using data from the Maastricht Upper Area Control Centre are presented with varying levels of predictive performance by the machine learning algorithms. Data sources include Demand Data Repository from EUROCONTROL and the Rapid Developing Thunderstorm product from EUMETSAT.     

The effect of management practices on aircraft incidents

Many models have been put forward in order to examine the human factors in aircraft accidents and incidents. Human Factors Analysis and Classification System (HFACS) which is the most widely used in literature is one of these models. HFACS is based on Reason's Swiss Cheese Model. The biggest disadvantage of the Reason's model is its post-accident applicability. Mostly HFACS aviation applications are usually based on accident data. This is a reagent (result-focused) approach. In this study, however, HFACS which is an improved version of Reason's model, was applied to aircraft incidents that did not result in an accident. This is a proactive approach. Thus, with this approach, the biggest disadvantage of Reason's model is turned into an advantage. In addition, a realistic application of this approach has been demonstrated in this study, focusing on aircraft incidents that took place between 2000 and 2018. The year 2000 forms a milestone in the manufacture of more technically advanced aircraft models which significantly reduced occurrence of technical errors in aircrafts, hence the choice of 2000 as base year. A total of 328 aircraft incident reports from the National Transportation Safety Board (NTSB) database were studied and among these reports cockpit crew related incidents were analyzed using HFACS. As a result of the analyzes, the root causes of incidents have been identified. In addition, unlike traditional HFACS analysis, the relationship between errors occurred at management levels of HFACS and the unsafe acts of the cockpit crew in aircraft incidents was statistically revealed.     

Predicting demand for air taxi urban aviation services using machine learning algorithms

This research focuses on predicting the demand for air taxi urban air mobility (UAM) services during different times of the day in various geographic regions of New York City using machine learning algorithms (MLAs). Several ride-related factors (such as month of the year, day of the week and time of the day) and weather-related variables (such as temperature, weather conditions and visibility) are used as predictors for four popular MLAs, namely, logistic regression, artificial neural networks, random forests, and gradient boosting. Experimental results suggest gradient boosting to consistently provide higher prediction performance. Specific locations, certain time periods and weekdays consistently emerged as critical predictors.     

Using causal machine learning for predicting the risk of flight delays in air transportation

Delays in air transportation are a major concern that has negative impacts on the airline industry and the economy. Given the complexity of the National Air Space system, predicting the risk of flight delays and identifying significant predictors is vital to risk mitigation. The purpose of this paper is to perform data mining using causal machine learning algorithms in the USELEI process (Understanding, Sampling, Exploring, Learning, Evaluating, and Inferring) to predict the probability of flight delays in air transportation using data collected from different sources. The findings indicated significant effects of predictors, including reported arrivals and departures, arrival and departure demands, capacity, efficiency, and traffic volume at the origin and destination airports on the risk of flight delays. More importantly, causal interrelationships among variables in a fully structural network are presented to how these predictors interact with one another and how these interactions lead to delay incidents. Finally, sensitivity analysis and causal inference can be performed to evaluate various what-if scenarios and form effective strategies to mitigate the risk of delays.     

Modelling of aircraft rotation in a multiple airport environment

The objective of this paper is to develop a simulation model to simulate aircraft rotation in a multiple airport environment. The developed aircraft rotation model (AR model) consists of two sub-models, namely the aircraft turnaround model, which describes the operation of aircraft turnaround activities at an airport and the Enroute model, which simulates the enroute flight time of an aircraft in the airspace between two airports. Delays due to operational disruptions from aircraft turnaround activities are modelled by stochastic variables in the aircraft turnaround model. Uncertainties from schedule punctuality are modelled by probability density functions in the Enroute model. The proposed aircraft rotation model is employed to carry out a case study by using real schedule and punctuality data from a European schedule airline. Simulation results when compared with observation data validate the effectiveness of the aircraft rotation model. The proposed model is also found suitable for airlines to serve as a schedule planning and analysis tool.     

The impact of variation in reporting practices on the validity of recommended birdstrike risk assessment processes for aerodromes

Birdstrikes are a major hazard to aviation; costing millions of pounds a year in damage and delays, as well as occasional hull losses and loss of life. The numbers and species of birds on and around airfields therefore need to be managed. To aid this process, airport staff often use risk assessments to identify which bird species cause the greatest risk and use the outcome to target their bird control effort. To this end, a number of national and international regulators, airports and other organisations recommend, or use, a derivation of a risk assessment process first published in 2006. This was developed using the UK Civil Aviation Authority's birdstrike database, employing data collected between 1976 and 1996. The risk assessment process relies on using the proportion of reported strikes that cause damage to the aircraft as a proxy for the likely severity of the outcome of strike incidents, so any change in the relative level of reporting of damaging and non-damaging strikes may significantly bias the results. The implementation of mandatory birdstrike reporting by the UK CAA in 2004 led to a significant increase in the number of strikes reported. If this involved a disproportionate increase in the number of non-damaging compared to damaging incidents reported, it may have impacted on the accuracy of the risk assessment process. This paper examines how differential reporting of damaging and non-damaging strikes can impact on the risk assessment process. It shows that changes in reporting practices since the original risk assessment was developed have impacted on the apparent birdstrike risk at UK airports, giving a false impression of increasing risk over the period. It makes recommendations for how the process can be better adapted to cope with such changes in the future, and how it should be modified for use in countries with different reporting regimes to that in the UK.     

The normalisation of aircraft overrun accident data

This paper is concerned with normalising runway overrun aircraft accident data so as to allow all accident data to be properly relevant to any overrun accident investigation. This task is part of a wider research task that addresses the need for models to assess the risk of aircraft operations at any particular airport based on risk management principles and to use all available data on previous accidents. The case of runway overruns is taken because new regulations require consideration of the provision of much longer Runway End Safety Areas than had been previously the norm. The reported research collects accident data and then describes its normalisation based on corrections made due to the effects of terrain, aircraft performance and required distances on the accident locations.     

Impact of landing fees on airlines’ choice of aircraft size and service frequency in duopoly markets

A one-shot simultaneous game-theoretic model is applied in a duopoly market to investigate how airport landing fees could influence airlines’ decisions on aircraft size and service frequency. It is found that higher landing fees will force airlines to use larger aircraft and less frequency, with higher load factor for the same number of passengers. It is also found that airlines will be better off if some of the extra landing fees are returned to airlines as a bonus for airlines using larger aircraft, which consequently reduces airport congestion.     

A global model for estimating the block time of commercial passenger aircraft

This paper uses a two-stage statistical model to estimate the block time of commercial passenger aircraft. The model considers many of the factors contributing to airport congestion and provides a basis for future development of multivariate statistical models of the flight operation process. The model is tested using 2 million US domestic flights by six airlines in 2004. Model analysis provided insight regarding the relative impact of weather conditions and airport utilization on block time. In particular, population, arrival time, airport utilization, ice, and the interaction of poor weather conditions and traffic were found to be significant predictors of block time.     

Statistical modelling of runway incursion occurences in the United States

Runway incursions are an important aviation safety concern; between 2002 and 2015 there were 16,785 runway incursions at United States airports ranging in size from small general aviation (GA) to large commercial airline hubs. When examining airports with the 50 highest incursion count over the past 5 years, the predominant categories were large hubs, which accounted for 21 airports and general aviation (GA) airports which accounted for 16 airports. In June 2015, the Federal Aviation Administration (FAA) announced the Runway Incursion Mitigation (RIM) program to identify airport risk factors that might contribute to a runway incursion and develop strategies to help airport stakeholders mitigate those risks. Different size airports serve different aircraft fleets, serve different operating volumes, and have different resources available (both funds and technologies) for incursion mitigation. Therefore, it is valuable to determine the correlating factors that affect incursions at different size airports. This paper uses econometrics based modelling techniques to identify statistically significant factors in data provided by the (FAA) public web sites on runway incursions. The model identified statistically significant variables that correlate with incursions, based on severity, for airports categories defined by the National Plan of Integrated Airport Systems (NPIAS).The model results indicate that operational incidents (OI) are more likely at large hub airports. In contrast, at GA/non-hub airports, pilot deviations (PD) were significant for less severe incursions (severity C and D). Only one variable, “number of years since 2002”, was found to be significant for all the three airport categories; this variable was correlated with severity A incursions and indicated a statistically significant reduction in severity A incursions, despite an overall 80% increase in incursions between 2002 and 2015.     

An explainable machine learning approach to improve take-off time predictions

Accurate aircraft trajectory predictions are necessary to compute exact traffic demand figures, which are crucial for an efficient and effective air traffic flow and capacity management. At present, the uncertainty of the take-off time is one of the major contributions to the loss of trajectory predictability. In the EUROCONTROL Maastricht Upper Area Control Centre, the predicted take-off time for each individual flight relies on the information received from the Enhanced Traffic Flow Management System. However, aircraft do not always take-off at the times reported by this system due to several factors, which effects and interactions are too complex to be expressed with hard-coded rules. Previous work proposed a machine learning model that, based on historical data, was able to predict the take-off time of individual flights from a set of input features that effectively captures some of these elements. The model demonstrated to reduce by 30% the take-off time prediction errors of the current system one hour before the time that flight is scheduled to depart from the parking position. This paper presents an extension of the model, which overcomes this look-ahead time constraint and allows to improve take-off time predictions as early as the initial flight plan is received. In addition, a subset of the original set of input features has been meticulously selected to facilitate the implementation of the solution in an operational air traffic flow and capacity management system, while minimising the loss of predictive power. Finally, the importance and interactions of the input features are thoroughly analysed with additive feature attribution methods.     

An empirical analysis of airport capacity expansion

Theoretical analyses of the impact of airport capacity expansion must model or make assumptions about the effect of capacity on demand, airline competition, aircraft types, fares and other characteristics of a given airport. In this paper, we use empirical data on historical schedules, fares, delays and demand for the busiest 150 airports in 2015 to examine the typical impact of historical capacity expansions. We find significant diversity in outcomes, with over half the expanded airports either using less than their pre-expansion capacity or remaining constrained even at post-expansion capacity by 2016. Many of the expected impacts, such as reductions in typical aircraft size, either do not materialise or are dominated by other effects (for example, recessions; airlines beginning or ending operations at an airport; changes in regulation). Behaviour on expansion is affected by slot control regulations and whether the airport is initially capacity-constrained. In particular, slot-controlled airports typically add new destinations and carriers on expansion rather than making significant changes to existing schedules.     

Measuring connectivity in the air freight industry

In this paper, we present an air transport connectivity model for air freight. For the purposes of this paper, connectivity is defined as all possible direct and indirect connections to or from an airport operated by wide-body aircraft, weighted for the quality of the connection in terms of transhipment and in-flight times. Using this model, we analyse the networks of seven European airports. Europe’s largest hub airports carry most air freight thanks to their extensive intercontinental passenger networks, while smaller airports with a strong focus on air freight carry large amounts of cargo on dedicated freighter aircraft. For air freight operations, the catchment area of an airport is much larger than it is for passenger services, as shipments are being trucked to their departure airport throughout all of mainland Europe. Since there are many airports sharing the same catchment area, potential competition for air freight is fierce. We found that well located regions between the four large European airports have access to large air freight networks, whilst regional air freight connectivity in northern and southern parts of Europe is substantially lower.     

Airport traffic complexity and environment efficiency metrics for evaluation of ATM measures

For a given (current or planned) traffic demand, different air traffic management measures could result in different airport traffic complexity and efficiency. This paper presents the research on the relationship between airport traffic complexity and time and environmental efficiency for different air traffic control (ATC) tactics applied to the given or planned airport layout. Emphasis is placed on the evaluation of airport traffic complexity, aircraft fuel consumption, gas emissions and time efficiency for different ATC tactics and/or airport airfield layouts. For busy airports during peak hours, arrival queuing delays, taxi-in, taxi-out times and departure queuing delays increase, which induces additional unnecessary fuel consumption, gas emission and time inefficiency. In order to find a tool which could indicate potential delay generators, a measure of airport traffic complexity – called Dynamic Complexity is proposed. Experiments were performed for airports with different airfield layouts, for different traffic demands and ATC applied tactics using SIMMOD simulation model. Traffic situations were analyzed and delays were measured. The values of airport traffic complexity, fuel consumption and gas emissions were also determined. A comparative analysis of the results show: first, the proposed airport traffic complexity metric quite satisfactorily reflects the influence of traffic characteristics upon the environmental state of the system, and second, different ATM strategic and tactical measures (airport airfield infrastructure development and applied ATC tactics) could significantly reduce traffic complexity and increase time and environmental efficiency at the airport.     

A data-driven operational model for traffic at the Dallas Fort Worth International Airport

Airports are on the front line of significant innovations, allowing the movement of more people and goods faster, cheaper, and with greater convenience. As air travel continues to grow, airports will face challenges in responding to increasing passenger vehicle traffic, which leads to lower operational efficiency, poor air quality, and security concerns. This paper evaluates methods for traffic demand forecasting combined with traffic microsimulation, which will allow airport operations staff to accurately predict traffic and congestion. Using two years of detailed data describing individual vehicle arrivals and departures, aircraft movements, and weather at Dallas-Fort Worth (DFW) International Airport, we evaluate multiple prediction methods including the Auto Regressive Integrated Moving Average (ARIMA) family of models, traditional machine learning models, and DeepAR, a modern recurrent neural network (RNN). We find that these algorithms are able to capture the diurnal trends in the surface traffic, and all do very well when predicting the next 30 minutes of demand. Longer forecast horizons are moderately effective, demonstrating the challenge of this problem and highlighting promising techniques as well as potential areas for improvement.Traffic demand is not the only factor that contributes to terminal congestion, because temporary changes to the road network, such as a lane closure, can make benign traffic demand highly congested. Combining a demand forecast with a traffic microsimulation framework provides a complete picture of traffic and its consequences. The result is an operational intelligence platform for exploring policy changes, as well as infrastructure expansion and disruption scenarios. To demonstrate the value of this approach, we present results from a case study at DFW Airport assessing the impact of a policy change for vehicle routing in high demand scenarios. This framework can assist airports like DFW as they tackle daily operational challenges, as well as explore the integration of emerging technology and expansion of their services into long term plans.     

Applying deep learning models to twitter data to detect airport service quality

Measuring airport service quality (ASQ) is an important process for identifying shortages and suggesting improvements that guide management decisions. This research, introduces a general framework for measuring ASQ using passengers’ tweets about airports. The proposed framework considers tweets in any language, not just in English, to support ASQ evaluation in non-speaking English countries where passengers communicate with other languages. Accordingly, this work uses a large dataset that includes tweets in two languages (English and Arabic) and from four airports. Additionally, to extract passenger evaluations from tweets, our framework applies two different deep learning models (CNN and LSTM) and compares their results. The two models are trained with both general data and data from the aviation domain in order to clarify the effect of data type on model performance. Results show that better performance is achieved with the LSTM model when trained with domain specific data. This study has clear implications for researchers and airport managers aiming to use alternative methods to measure ASQ.     

Regional public support to airlines and airports: An unsolved puzzle

This paper proposes a structural model to explain the motivation of regional public authorities to arrange marketing agreements for route and traffic development. Furthermore, using data from Spanish airports, we empirically test this model obtaining the demand function according to the preferences of public authorities. The results show that the public budget, airport’s attributes or intermodal competition affect to the demand for aircraft operations of regional public agencies. Finally, we propose an empirical method to determine the market power of airlines within these marketing agreements in a particular airport or route.