Araştırmacılar(2)

• Demirel, Soner
• Özdemir, Mustafa

• ArticleYayıncıda açık erişim2026/r/ebyu/11281
  A MACHINE LEARNING APPROACH FOR EVALUATING AIR TRAFFIC CONTROLLER CANDIDATES: A COMPARISON WITH AHP AND ANP METHODS

  Özdemir, Mustafa; Özdemir, Müjgan Sağır

  The safety and efficiency of air traffic operations depends heavily on air traffic controllers, making the accurate identification of suitable candidates a critical challenge for air navigation service providers. The selection process is traditionally based on standardized aptitude testing combined with structured interviews and expert evaluation, which may be affected by subjectivity and require substantial time and resources. This study introduces a machine learning based decision support approach for air traffic controller (ATCO) candidate selection. The candidate evaluation task is formulated as a binary classification problem to predict interview outcomes using historical data from an actual selection process involving 194 candidates, including 30 who passed the interview phase. Three machine learning models, Logistic Regression (LR), Support Vector Machine (SVM), and Decision Tree (DT), are trained using candidate background variables such as prior examination scores, type of high school attended, and high school grade point average. The performance of the models is evaluated in comparison with existing approaches based on the Analytic Hierarchy Process and the Analytic Network Process. The results of this study provide preliminary evidence that, in the context of ATCO selection, machine learning models may approximate the outcomes of traditional interviews, with LR achieving an accuracy of 93%, and both DT and SVM achieving accuracies of 90%. Using the best-performing model, 46 eliminated candidates were correctly classified as eliminated (true negatives) and 9 selected candidates were correctly identified as selected (true positives), while 4 eliminated candidates were incorrectly classified as selected (false positives). These results suggest that the proposed approach may support air navigation service providers and related institutions by assisting early-stage screening through data driven prioritization of ATCO candidates.

  RDL Kalıcı Kaydıİncele
• Conference ObjectYayıncıda açık erişim2026/r/ebyu/10890
  Modeling Propagation of Tactical Altitude Interventions for Contrail Avoidance in a Socio-Technical Air Traffic System

  Sekine, Katsuhiro; Mori, Tomoki; Sato, Ren; Özdemir, Mustafa; Oshio, Kiyoto; Bouchaudon, Philippe; Itoh, Eri

  Air traffic control is a complex socio-technical system in which human operators, automation, and procedures interact within structured airspace. Local operational interventions can propagate non-linearly through traffic flows and coordination networks. Contrail mitigation requires altitude adjustments in dense airspace, yet limited attention has been paid to how vertical perturbations interact with traffic structures and cross-boundary coordination constraints. This study employs a human-in-the-loop simulation platform based on ESCAPE Light to examine system responses to contrail mitigation in a high-density Asian Flight Information Region. Altitude changes are treated as localized perturbations imposed on sequenced traffic with coordination logic. Three configurations were examined: baseline operations, tactical adjustments, and strategic mitigation. Rather than relying on subjective workload assessments, the analysis focuses on structural indicators including altitude-change events. Results show tactical interventions amplify coordination demand and trigger secondary flow adjustments beyond affected aircraft, whereas strategically planned adjustments produce fewer disruptions. These findings illustrate how human-in-the-loop simulation reveals perturbation propagation and adaptive dynamics in dense air traffic systems.

  RDL Kalıcı Kaydıİncele
• ArticleAçık Erişim2026/r/ebyu/10592
  Aircraft sequencing and passenger flow optimization: Istanbul Sabiha Gokcen Airport; [Uçuş sıralaması ve yolcu akışı optimizasyonu: İstanbul Sabiha Gökçen havalimanı]

  Demirel, Soner

  This study presents a model that combines aircraft sequencing and passenger transfer dynamics to analyze the transfer process in terms of delay and how air-to-air and air-to-train transfer passengers reach stations and gates. The model considers real-world factors such as distances inside the terminal, walking times depending on baggage and gender, and waiting times at bus/metro stations. The research includes identifying the flights with the highest number of transfer passengers and conducting the analysis by giving priority to the busiest five aircraft. Individual behaviors of passengers, such as baggage and walking times, are also included in the optimization framework. When the differences between the models are analyzed, passengers' early arrival times at transfer points vary 0-5 minutes. Although this is important, no statistical difference was found. The analysis shows that Model 3 (43%) is superior to the others. While Model 2 reached an average delay of 1,066 seconds, Model 3 produced a delay of 185 seconds. Model 1 and 3 have similar performance. The combination of aircraft sequencing and transfer dynamics is the original contribution of the model. In addition, by including passenger-specific factors and operational constraints in the model, it becomes possible to offer more realistic and practical solutions.

  RDL Kalıcı Kaydıİncele
• Conference ObjectYalnız künye2026/r/ebyu/10153
  Traffic Flow Optimization in TMAs: Modeling and User Interface Implementation

  Demirel, Soner

  This study aims to manage air traffic flow in the terminal maneuvering area (TMA) of Istanbul Sabiha Gokcen Airport by generating delay-based solutions. For this purpose, a mixed-integer mathematical modeling technique is used. The model is integrated into a graphical user interface that includes two modules: Generic Data and Real Data. In this way, planning air traffic controllers can gain foresight into the traffic flow within the TMA by using parameters derived from the traffic outside the TMA. Through the Generic Data module, preliminary runs with different traffic mixes allow researchers to observe the outcomes of various scenarios, while the Real Data module provides both planning support and delay-based recommendations for controllers. The interface includes a Gantt chart, a heatmap visualizing delays between waypoints, the frequency of radio usage by air traffic controllers for solutions, emission-related graphs and tables presenting statistics such as flight durations of aircraft within the TMA. By using the model and the interface, traffic flow can be managed optimally with the objective of minimizing delays. At the same time, the user-friendly interface offers visual references and can serve as an easy-To-use assistant for planning controllers.

  RDL Kalıcı Kaydıİncele
• Conference ObjectYalnız künye2026/r/ebyu/10021
  Hotspot Capacity-Driven Scheduling Strategies for Reducing Airborne Delays in TMAs

  Demirel, Soner

  This paper investigates the reduction of airborne delays in Terminal Maneuvering Areas (TMAs) through departure time adjustments based on hotspot levels at origin airports. While existing studies primarily focus on reducing airborne delays to support environmentally sustainable airport operations, this research proposes shifting such delays to the ground within the predefined capacity limits of critical areas (hotspots) at destination airports. Istanbul Sabiha Gokcen International Airport (LTFJ), its associated TMA routes and Point Merge System (PMS) were selected as the case study. Four virtual origin airports were defined for analysis. The proposed model represents hotspot locations (gates, taxiways, and runway holding positions) as virtual nodes with predefined congestion levels (low, medium, and high). The problem is formulated as a mixed integer linear programming model and solved using the CPLEX solver. The model achieves computational efficiency within reasonable solution times. Under the implementability bound in Eq. (31), the model reduces average airborne delay from 1552 s to 931.2 s and it corresponds to a 40 percent reduction within the prescribed acceptability band. Overall, the findings demonstrate that the proposed methods align with existing literature by enhancing delay management through realistic modeling at hotspot points.

  RDL Kalıcı Kaydıİncele
• ArticleAçık Erişim2025/r/ebyu/12235
  Optimizing Aircraft Operations in Case of Increasing Demand for Limited Airport Capacity

  Demirel, Soner; Çınar, Ertan

  The purpose of this paper is to minimize the total difference between the requested and assigned departure time of aircraft to enhance the efficiency of using limited airport capacity. The mathematical model was formed by employing mixed integer linear programming. The parameters, decision variables, and constraints were defined to cover the problem. The baseline and alternative scenarios were compared to present the improved results. Sensitivity analysis was performed to test the ability of the model with different parameters. Also, in order to test the mathematical model, distinct from both the baseline and alternative scenarios, CHQ airport was based by using its number of parking positions and taxi-in/out durations in the sensitivity analysis. The proposed model reduced the total difference by 20.78% for 30 aircraft for 5 parking positions in the alternative scenario. The results showed that it may well serve to improve the imbalance regarding operational conditions.

  RDL Kalıcı Kaydıİncele
• Conference ObjectYalnız künye2025/r/ebyu/11860
  Predicting Aircraft Climb and Descent Times in Conflict Resolution Maneuvers Using Machine Learning Models

  Ozdemir, Mustafa

  This study proposes a machine learning-based approach to predict aircraft climb and descent times during conflict resolution maneuvers using data collected from on-board flight data recorders. The proposed framework utilizes features such as the magnitude of altitude change, wind speed, true airspeed, and initial altitude to provide real-time predictions of climb and descent durations. Three machine learning models - Random Forest, XGBoost, and Deep Learning - are developed and evaluated. The magnitude of altitude change is identified as the most influential factor affecting climb and descent times, followed by initial altitude, true airspeed, and temperature, with variations depending on the selected model. The models are assessed using multiple performance metrics, including Mean Absolute Error (MAE), Root Mean Square Error (RMSE), and Mean Absolute Percentage Error (MAPE). The Deep Learning model outperforms the others, achieving a MAE of 31.37 and a MAPE of 17.38. This research has the potential to improve the predictability of altitude maneuvers in conflict resolution, enabling the models to assist air traffic controllers in making informed altitude change decisions, reducing conflict risks, and enhancing efficiency.

  RDL Kalıcı Kaydıİncele
• ArticleYayıncıda açık erişim2025/r/ebyu/11628
  Transit passenger-oriented optimisation of arrival aircraft sequencing

  Demirel, S.

  This study presents a model that aims to optimise the sequencing arrival aircraft around the terminal manoeuvring area (TMA). The model considers the transit passenger counts of these aircraft and employs the point merge at Sabiha Gokcen Airport. In this study, aircraft were categorised into two groups, namely ‘High or Low Transit Passenger (HTP/LTP)’. Subsequently, multi-objective models were employed to solve the test problems. Weighted sum scalarisation (WSS), conic scalarization (CS), and epsilon constraint (EC) models were utilised to increase robustness and their results were compared with a single-objective optimisation model. This approach aims to provide decision-makers with a variety of outcomes, thus expanding their options. Simultaneously, efforts are made within the model to allow aircraft with HTP counts to have minimal delays. Additionally, emission calculations were conducted to offer a critical perspective on the environmental implications, and the delay results of the multiobjective optimisation (MOO) models underwent statistical analysis.

  RDL Kalıcı Kaydıİncele
• ArticleYalnız künye2023/r/ebyu/13307
  Determining the Key Factors Affecting Pseudo-Pilot Workload Based on Real-Time Simulations

  Özdemir, Mustafa; Dönmez, Kadir; Demirel, Soner

  Pseudo-pilots work in real-time simulations in air traffic control training and carry out an important task to ensure appropriate training for the trainee controllers. They are in constant communication with controllers during the training and are responsible for fulfilling the instructions given by the controllers. Unlike real operations, pseudo-pilots are required to monitor many aircraft simultaneously on the simulator’s radar screen and perform various interventions. This situation can be challenging, and the impact of such situations on their workload is usually neglected. In the literature, various studies have revealed the factors affecting controller workload; however, the factors affecting the workload of pseudo-pilots have not been emphasized. In this study, statistical analyses were carried out using the data obtained from real-time simulations and NASA-TLX surveys. As a result, the number of aircraft, controller performance, and the interruption duration were determined as the key factors that affect the overall workload of pseudo-pilots.

  RDL Kalıcı Kaydıİncele
• ArticleYalnız künye2023/r/ebyu/13235
  Comparison of RECAT-EU and ICAO wake turbulence category on the Point Merge System

  Demirel, S.

  This study presents a mathematical model that schedules arrival aircraft regarding RECAT-EU that is new categorisation for applying separation minima and analyses its effect on the performance of the Point Merge System (PMS) at Sabiha Gökcen International Airport (LTFJ). There are two main scenarios: one of them uses RECAT-EU and the other employs the ICAO wake turbulence category. Both scenarios have ten different test problems to examine the mathematical model. The model applies RECAT-EU wake turbulence categories and compares the outcome with the ICAO wake turbulence categories. The model aims to minimise flight duration on the sequencing leg and ground delay in the departure queue using the RECAT-EU and ICAO wake turbulence categories individually. The results were analysed to reveal the PMS performance using the two different approaches to turbulence categories. Statistical analysis was also carried out to compare the means of the two groups in the model.

  RDL Kalıcı Kaydıİncele