Ground Delay Program Research Articles

Departure flight delay prediction due to ground delay program using Multilayer Perceptron with improved sparrow search algorithm

Abstract The ground delay program (GDP) is a commonly used tool in air traffic management. Developing a departure flight delay prediction model based on GDP can aid airlines and control authorities in better flight planning and adjusting air traffic control strategies. A model that combines the improved sparrow search algorithm (ISSA) and Multilayer Perceptron (MLP) has been proposed to minimise prediction errors. The ISSA uses tent chaotic mapping, dynamic adaptive weights, and Levy flight strategy to enhance the algorithm’s accuracy for the sparrow search algorithm (SSA). The MLP model’s hyperparameters are optimised using the ISSA to improve the model’s prediction accuracy and generalisation performance. Experiments were performed using actual GDP-generated departure flight delay data and compared with other machine learning techniques and optimisation algorithms. The results of the experiments show that the mean absolute error (MAE) and root mean square error (RMSE) of the ISSA-MLP model are 16.8 and 24.2, respectively. These values are 5.61%, 6.3% and 1.8% higher in MAE and 4.4%, 5.1% and 2.5% higher in RMSE compared to SSA, particle swarm optimisation (PSO) and grey wolf optimisation (GWO). The ISSA-MLP model has been verified to have good predictive and practical value.

Flight rescheduling of an airline under the ground delay program considering delay propagation in multiple airports

The purpose of this study is to reschedule flights for an airline's profit to correspond to the airport’s changed capacity. In the event of a ground delay program (GDP), the number of flights the airport can accommodate is reduced. We formulated a mixed-integer linear programming (MILP) model to reschedule flights. The MILP models were divided into two versions to handle the uncertainty of the future. In scenarios in which the GDP is changed again, an optimal model obtains solutions for each scenario. The stochastic model solution obtains a minimizing expectation cost of all scenarios. All flights are connected to both the origin and destination airports, and one aircraft may be used for more than one flight. Therefore, we considered delay propagation not only within the same airport but from other airports by extending the setup to include several airports at once. Because the objective of this study is to minimize the operation cost of airline, we also considered costs associated with airline resources such as aircrafts and crews. Related experiments were conducted including comparison between two suggested versions.

A Methodology for Predicting Ground Delay Program Incidence through Machine Learning

Effective ground delay programs (GDP) are needed to intervene when there are bad weather or airport capacity issues. This paper proposes a new methodology for predicting the incidence of effective ground delay programs by utilizing machine learning techniques, which can improve the safety and economic benefits of flights. We use the combination of local weather and flight operation data along with the ATM airport performance (ATMAP) algorithm to quantify the weather and to generate an ATMAP score. We then compared the accuracy of three machine learning models, Support Vector Machine, Random Forest, and XGBoost, to estimate the probability of GDPs. The results of the weather analysis, performed by the ATMAP algorithm, indicated that the ceiling was the most critical weather factor. Lastly, we used two linear regression models (ridge and LASSO) and a non-linear regression model (decision tree) to predict departure flight delays during GDP. The predictive accuracy of the regression models was enhanced by an increase in ATMAP scores, with the decision tree model outperforming the other models, resulting in an improvement of 8.8% in its correlation coefficient (R2).

Predictive and Prescriptive Analytics Toward Passenger-Centric Ground Delay Programs

Ground delay programs (GDPs) comprise the main interventions to optimize flight operations in congested air traffic networks. The core GDP objective is to minimize flight delays, but this may not result in optimal outcomes for passengers—especially with connecting itineraries. This paper proposes a novel passenger-centric optimization approach to GDPs by balancing flight and passenger delays in large-scale networks. For tractability, we decompose the problem using a rolling procedure, enabling the model’s implementation in manageable runtimes. Computational results based on real-world data suggest that our modeling and computational framework can reduce passenger delays significantly at small increases in flight delay costs through two main mechanisms: (i) delay allocation (delaying versus prioritizing flights) and (ii) delay introduction (holding flights to avoid passenger misconnections). In practice, however, passenger itineraries are unknown to air traffic managers; accordingly, we propose statistical learning models to predict passenger itineraries and optimize GDP operations accordingly. Results show that the proposed passenger-centric approach is highly robust to imperfect knowledge of passenger itineraries and can provide significant benefits even in the current decentralized environment based on collaborative decision making.

Scenario-based stochastic programming for an airline-driven flight rescheduling problem under ground delay programs

We address an airline-driven flight rescheduling problem within a single airport in which a series of ground delay programs (GDPs) are considered. The objective of the problem is to minimize an airline’s total relevant cost (TRC) consisting of delay costs, misconnection costs, and cancellation costs that would result from flight rescheduling. We introduce three solution approaches—the greedy approach, the stochastic approach, and the min-max approach—that revise the daily flight scheduling whenever the schedule is affected by a GDP or further GDP changes. The greedy approach simply searches for a solution using currently updated static GDP information, and the other two approaches provide a solution by considering possible scenarios for changes of the GDP. Using real-world data in existing literature and some generated scenarios, we present extensive computational results to assess the performance of the approaches. We also report the values of information on GDP the solution approaches refer to. Deliberating various cost parameter settings an airline might consider, we discuss the value of information in implementing the proposed solution approaches.

Modeling Arrival Flight Times within the Terminal Airspace

Air traffic management efficiency in the descent phase of flights is a key area of interest in aviation research for the United States, Europe, and recently other parts of the world. The efficiency of arrival travel times within the terminal airspace is one of nineteen key performance indicators defined by the Federal Aviation Administration (FAA) and the International Civil Aviation Organization, typically within 100 nmi of arrival airports. This study models the relationship between travel time within the terminal airspace and contributing factors using a multivariate log-linear model to quantify the impact that these factors have on the total travel time within the last 100 nmi. The results were compared with the baseline set of variables that are currently used for benchmarking at the FAA. The analyzed data included flight and weather data from January 1, 2018 to March 31, 2018 for five airports in the United States: Chicago O’Hare International Airport, Hartsfield-Jackson Atlanta International, San Francisco International Airport, John F. Kennedy International Airport, and LaGuardia Airport. The modeling results showed that there is a significant improvement in prediction accuracy of travel times compared with the baseline methodology when additional factors, such as wind, meteorological conditions, demand and capacity, ground delay programs, market distance, time of day, and day of week, are included. Root mean squared error values from out-of-sample testing were used to measure the accuracy of the estimated models.

Prediction of air traffic delays: An agent-based model introducing refined parameter estimation methods.

We propose an agent-based model for predicting individual flight delays in an entire air traffic network. In contrast to previous work, more detailed parameter estimation methods were incorporated into the agent-based model, acting on the state transitions of agents. Specifically, a conditional probability model was proposed for modifying the expected departure time, which was used to indicate whether a flight had experienced the necessary waiting due to Ground Delay Programs (GDPs) or carrier-related reasons. Additionally, two random forest regression models were presented for estimating the turnaround time and the elapsed time of flight agents in the agent-based delay prediction model. The parameter models were trained and fitted using the flight data for 2017 in the United States. The performance of the delay prediction model was tested for thirty days with three types of delay levels (low, medium, and high), which were randomly selected from 2018. The experimental results showed that the average absolute error in the test days was 6.8 min, and the classification accuracy with a 15 min threshold for a two-hour forecast horizon was 89.5%. The performance of our model outperformed that of existing research. Additionally, the positive effect of introducing parameter models and the negative impact of increasing the prediction horizon on the prediction performance were further studied.

Strategic Flight Cancellation Under Ground Delay Program Uncertainty

Under certain capacity constraints, flight operators will strategically cancel flights hours to days in advance of their scheduled departure times to improve their overall operating schedule. Although there are consequences for unnecessarily canceling flights (i.e., displaced passengers, lost revenue, etc.), any benefits associated with such cancellations are best realized if made early, often before any traffic flow rate limitation is imposed. This paper proposes a framework for optimizing an adaptive decision strategy based on the evolution of the forecast uncertainty. Using an ensemble forecast, a scenario tree is generated to highlight both key planning scenarios and the likelihood of these scenarios developing over the forecast horizon. By aligning decision points at the initial and intermediary nodes in the tree, the recommended flight cancellations and cancellation timing are optimized for minimum expected cost with respect to the forecast uncertainty. The proposed approach is applied to a case day and analyzes the recommended cancellations across the forecast horizon against different predicted scenarios. The results demonstrate that by directly planning for future strategy adaptation lower costs are incurred earlier, when a higher degree of uncertainty exists, as compared with decisions based on a single decision point.

Machine Learning Approach to the Analysis of Traffic Management Initiatives

Ground delay programs and ground stops are implemented to ensure that airport and flight operations remain safe in spite of constraint(s) at affected airports. Occasionally, ground stops are issued during ongoing ground delay programs, and vice versa, which hinders their efficient planning and implementation. Consequently, this research outlines a methodology for the analysis of their coincidence so as to identify underlying factors and better understand their occurrence. This is achieved by 1) fusing weather, ground delay program, and ground stop data; and 2) leveraging machine learning algorithms, partial dependence plots, and decision trees to assess and understand how different factors influence the coincidence of ground delay programs and ground stops. The methodology was implemented and tested using data from LaGuardia and San Francisco International Airports in order to test its robustness at airports with different weather conditions, locations, etcetera. The boosting ensemble algorithm was identified as the best-suited algorithm for both airports, and it was observed that different factors influence the occurrence of this phenomenon at both airports. This work also discusses how decision trees and partial dependence plots can be leveraged to analyze how different factors influence the coincidence of weather-related ground delay programs and ground stops.

Machine Learning Approach for Flight Departure Delay Prediction and Analysis

The expected growth in air travel demand and the positive correlation with the economic factors highlight the significant contribution of the aviation community to the U.S. economy. On‐time operations play a key role in airline performance and passenger satisfaction. Thus, an accurate investigation of the variables that cause delays is of major importance. The application of machine learning techniques in data mining has seen explosive growth in recent years and has garnered interest from a broadening variety of research domains including aviation. This study employed a support vector machine (SVM) model to explore the non-linear relationship between flight delay outcomes. Individual flight data were gathered from 20 days in 2018 to investigate causes and patterns of air traffic delay at three major New York City airports. Considering the black box characteristic of the SVM, a sensitivity analysis was performed to assess the relationship between dependent and explanatory variables. The impacts of various explanatory variables are examined in relation to delay, weather information, airport ground operation, demand-capacity, and flow management characteristics. The variable impact analysis reveals that factors such as pushback delay, taxi-out delay, ground delay program, and demand-capacity imbalance with the probabilities of 0.506, 0.478, 0.339, and 0.338, respectively, are significantly associated with flight departure delay. These findings provide insight for better understanding of the causes of departure delays and the impacts of various explanatory factors on flight delay patterns.

Predictive and Prescriptive Analytics toward Passenger-centric Ground Delay Programs

Ground Delay Programs (GDP) comprise the main interventions to optimize flight operations in congested air traffic networks. The core GDP objective is to minimize flight delays, but this may result in optimal outcomes for passengers — especially with connecting itineraries. This paper proposes an original passenger-centric approach to GDP by leveraging data on passenger itineraries in flight networks. First, we identify analytical drivers of passenger-centric operations in transportation systems. Second, we develop an integer program that balances flight delays and passenger delays in large-scale GDP operations. A rolling procedure decomposes the problem while ensuring global feasibility, enabling the model’s implementation in short computational times. Third, we propose statistical learning models to predict passenger itineraries and optimize GDP operations accordingly, enabling the model’s implementation when passenger itineraries are unknown by air traffic managers. Computational results based on real-world data suggest that our modeling and computational framework can reduce passenger delays significantly at small increases in flight delay costs, and that these benefits are robust to imperfect knowledge of passenger itineraries. Results highlight two major levers of passenger-centric operations: (i) delay allocation (which flights to delay vs. prioritize), and (ii) delay introduction (whether to deliberately hold flights to avoid passenger misconnections).

Using machine learning to analyze air traffic management actions: Ground delay program case study

We model the impact of weather and arrival demand on ground delay program (GDP) incidence. We use Support Vector Machine (SVM) to analyze how regional convective weather affects GDP incidence and find the impact depends on both distance and direction of convective activity from the airport. We then train and compare the performance of logistic regression (LR) and random forest (RF) in predicting GDP incidence using an SVM-generated regional weather variable, local weather and arrival demand. Generally, RF outperforms LR. Convective weather is the most important factor in predicting GDP incidence at Atlanta International Airport (ATL), while arrival demand has greater impact for the other airports studied. We also examined model transferability across different airports. Lastly, we build GDP duration prediction models to enable a user to assess how long a GDP is likely to continue, if it is in effect in a given hour.

Prediction of Weather Impacts on Airport Arrival Meter Fix Capacity

<div class="section abstract"><div class="htmlview paragraph">This paper introduces a data driven model for predicting airport arrival capacity with 2-8 hour look-ahead forecast data. The model is suitable for air traffic flow management by explicitly investigating the impact of convective weather on airport arrival meter fix throughput. Estimation of the arrival airport capacity under arrival meter fix flow constraints due to severe weather is an important part of Air Traffic Management (ATM). Airport arrival capacity can be reduced if one or more airport arrival meter fixes are partially or completely blocked by convective weather. When the predicted airport arrival demands exceed the predicted available airport’s arrival capacity for a sustained period, Ground Delay Program (GDP) operations will be triggered by ATM system. Severe imbalances between demand and capacity occur most frequently when the airport capacity is severely degraded due to either bad airport terminal surface weather or inclement convective weather around airport arrival fixes. A model that predicts the weather-impacted airport arrival meter fix throughput may help ATM personnel to plan GDP operations more efficiently. This paper identifies the characteristics of air traffic flow across arrival meter fixes at Newark Liberty International Airport (EWR). The proposed approach, based on machine-learning methods, is developed to predict the weather impacted EWR arrival Meter Fix (MF) throughput. Sector forecast coverage is used to envision the weather impact on airport arrival MF flow, and the validation is accomplished by using Convective Weather Avoidance Model (CWAM) 0.5 to 2-hour and Collaborative Convective Forecast Product (CCFP) 4 to 8-hour look-ahead forecast data for the period of April-September in 2014. Furthermore, the regression tree ensemble learning of random forests approach for translating a sector forecast coverage model to EWR arrival meter fix throughput is examined. The results suggest that ATM decision makers in charge of MF flow control and GDP planning may benefit from adopting the airport arrival meter capacity prediction models to estimate the inclement weather impacts.</div></div>

Predicting and planning airport acceptance rates in metroplex systems for improved traffic flow management decision support

Efficient planning of Airport Acceptance Rates (AARs) is key for the overall efficiency of Traffic Management Initiatives such as Ground Delay Programs (GDPs). Yet, precisely estimating future flow rates is a challenge for traffic managers during daily operations as capacity depends on a number of factors/decisions with very dynamic and uncertain profiles. This paper presents a data-driven framework for AAR prediction and planning towards improved traffic flow management decision support. A unique feature of this framework is to account for operational interdependency aspects that exist in metroplex systems and affect throughput performance. Gaussian Process regression is used to create an airport capacity prediction model capable of translating weather and metroplex configuration forecasts into probabilistic arrival capacity forecasts for strategic time horizons. To process the capacity forecasts and assist the design of traffic flow management strategies, an optimization model for capacity allocation is developed. The proposed models are found to outperform currently used methods in predicting throughput performance at the New York airports. Moreover, when used to prescribe optimal AARs in GDPs, an overall delay reduction of up to 9.7% is achieved. The results also reveal that incorporating robustness in the design of the traffic flow management plan can contribute to decrease delay costs while increasing predictability.

Exploration of the Evolution of Airport Ground Delay Programs

This study introduces a novel method of merging disparate but complementary datasets and applying machine learning techniques to ground delay program (GDP) data. More specifically, it aims to characterize GDPs with respect to changing weather forecasts, GDP plan parameters, and operational performance. The analysis aims to gain insights into GDP usage patterns (implementation and revisions), with respect to these key dimensions. It also aims to gain insights into how GDP cancelations and revisions correlate with operational efficiency and predictability. The results could be used to help traffic managers and air carriers understand complex patterns in the evolution of GDPs, so that they might, for example, better anticipate or even plan a response to a change in weather conditions. The focus is on GDPs at Newark Liberty International Airport (EWR), from 2010 through 2014. A master dataset was generated by merging several datasets on GDPs, weather forecasts, and individual flight information. Several scenarios of GDP evolution were then identified by reducing the dimensionality of the master GDP dataset, then applying cluster analysis on the lower dimensional data. It was found that GDPs at EWR can be categorized into 10 types based on weather forecasts, realized weather, GDP scope, arrival rates, and duration. The characteristics of these 10 GDP clusters were further explored by examining the relationships between GDP scenarios and their performance. It was found that GDPs under stable, low-severity weather and with large scope may score higher on the efficiency metric than expected. When GDPs called in the same weather conditions have high program rates, medium durations, and narrow scopes, capacity utilization was higher than expected—less affected flights lead to fewer cancelations and more arrivals (albeit delayed), and therefore, higher capacity utilization. Results also suggest that program rates are set more conservatively than needed for some poor weather conditions that end earlier than expected. GDPs with fewer revisions were associated with a higher predictability score but lower efficiency score. These findings can provide greater insights and knowledge about GDPs for future planning purposes. More specifically, the findings could, for example, be used to support discussion around, or even future guidance regarding, how to set and adjust GDP program rates. In future work additional data could be utilized to provide a more comprehensive operational picture of GDPs, and a wider range of performance metrics could be considered. It is also recommended that the patterns of how GDPs evolve over their lifetimes be further explored using other machine learning techniques that may provide new and useful insights.

Failures of critical systems at airports: Impact on aircraft operations and safety

Failures of critical systems at airports can significantly impact aircraft operations and cause substantial safety risks. This study develops a generalized method for measuring the impacts of unscheduled communication, navigation and surveillance systems outages on airport operational performance (i.e. throughput) and safety. The proposed method consists of two separate models – the aggregate and the disaggregate model. The aggregate model is a regression model that analyzes flight data, which are averaged per quarter hour and obtained from the Aviation System Performance Metrics database, and investigates the reductions in airport throughput due to an outage. The disaggregate model analyzes individual flights using the Performance Data Analysis and Reporting System database to investigate air traffic controllers’ terminal airspace traffic management procedures during an outage by examining the time-separation (i.e. headways) between consecutive arriving aircraft on each runway. More specifically, the model calculates the changes in the probability of a loss of separation and assesses the necessity of enacting a Traffic Management Initiative. The proposed method is tested on the case of two localizer outages at Los Angeles International Airport. The case-study results reveal that the throughput decreased after air traffic controllers enacted a Ground Delay Program during the first observed outage. The probability of a loss of separation doubled and the smaller aircraft were diverted to other airports. Further results indicate that arriving headways can be better described by a combination of two probability distributions, rather than a single distribution, in order to more accurately capture a complexity of air traffic controllers’ actions. The method can help aviation regulators: (1) make better decisions about the value of equipment and redundancies, (2) analyze operational and safety degradations caused by equipment failures, and (3) develop appropriate traffic management procedures and improve existing ones.

Stochastic models of Ground Delay Program implementation for prediction, simulation, and insight

We use historical data to build two types of stochastic model of hourly Ground Delay Program (GDP) implementation for the three main airports near New York City: Newark Liberty International, LaGuardia, and John F. Kennedy International. The models predict the probability that a GDP will be initial ized or canceled in a given hour based on hundreds of features describing the situation at the airport, including features describing forecasted weather conditions and scheduled traffic. One is a regularized logistic regression model that ignores system dynamics and the other model is based on inverse reinforcement learning, so it considers system dynamics and the impact that GDP implementation actions have over time on some metrics. We evaluate the models based on two objectives: their ability to predict and simulate GDP implementation decisions in historical test data sets. As is expected based on the motivation for and objective of each type of model, regularized logistic regression models make superior predictions while simulations controlled by inverse reinforcement learning models produce average metric values that more closely match those produced by historical GDP implementation decisions. Finally, we draw insights about GDP implementation from the trained model parameters. For example, parameters of both models suggest while weather conditions and values of performance metrics such as airborne delay play a role in GDP decision making, parameters related to the predictability or continuity of existing GDP plans are also important.

Predicting the initiation of a ground delay program

Predicting the initiation of a ground delay program

Airline-driven ground delay programs: A benefits assessment

Three decades of research studies in ground delay program (GDP) decision-making, and air traffic flow management in general, have produced several analytical models and decision support tools to design GDPs with minimum delay costs. Most of these models are centralized, i.e., the central authority almost completely decides the GDP design by optimizing certain centralized objectives. In this paper, we assess the benefits of an airline-driven decentralized approach for designing GDPs. The motivation for an airline-driven approach is the ability to incorporate the inherent differences between airlines when prioritizing, and responding to, different GDP designs. Such differences arise from the airlines’ diverse business objectives and operational characteristics. We develop an integrated platform for simulating flight operations during GDPs, an airline recovery module for mimicking the recovery actions of each individual airline under a GDP, and an algorithm for fast solution of the recovery problems to optimality. While some of the individual analytical components of our framework, model and algorithm share certain similarities with those used by previous researchers, to the best of our knowledge, this paper presents the first comprehensive platform for simulating and optimizing airline operations under a GDP and is the most important technological contribution of this paper. Using this framework, we conduct detailed computational experiments based on actual schedule data at three of the busiest airports in the United States. We choose the recently developed Majority Judgment voting and grading method as our airline-driven decentralized approach for GDP design because of the superior theoretical and practical benefits afforded by this approach as shown by multiple recent studies. The results of our evaluation suggest that adopting this airline-driven approach in designing the GDPs consistently and significantly reduces airport-wide delay costs compared to the state-of-the-research centralized approaches. Moreover, the cost reduction benefits of the resultant airline-driven GDP designs are equitably distributed across different airlines.

Stable two-sided matching of slot allocation in airport collaborative decision making by top trading cycles mechanism

The problem of capacity shortage in some airports needs to be dealt with sustainable solutions including a more efficient use of the existing runway slots at the airports. The Collaborative Decision Making (CDM) is an important approach applied to Air Traffic Management (ATM) to achieve this efficient use of the slots allocation. Using the Matching approach for two-sided markets of Game theory, the Top Trading Cycle CDM (TTC-CDM) algorithm developed in this research is an extension of the CDM approach aggregating the Ground Delay Program (GDP) of the air sector. The paper compared the developed TTC-CDM model to the existing models such as the conventional Compression algorithm in CDM, the Trade Cycle algorithm and the Deferred Acceptance CDM (DA-CDM) model to evaluate the performance of the proposed model. Through a case study, the results show the effective application of TTC-CDM model to slot allocation in ATM and also presents the advantage of considering the preferences of airport managers beside ATC controllers and airlines in the decision processing.