Estimation of aircraft fuel consumption and emissions during taxiing processes

• Provides a comprehensive comparison of established and advanced taxiing emission models. • Evaluates real-world aircraft taxiing data from a major European hub airport (Frankfurt). • Demonstrates that machine learning models outperform traditional estimation methods. • Identifies factors significantly influencing emissions and offers practical recommendations. The aviation sector faces growing challenges in reducing emissions as global passenger numbers continue to rise. While most research focuses on flight operations, this research investigates aircraft emissions during taxiing. Therefore, existing taxiing fuel models are compared and evaluated on a theoretical and empirical basis. Using an extensive dataset of real-world taxiing data from a European hub-airport, we evaluate linear models widely used in practice based on ICAO fuel burn indices and more sophisticated models that account for operational factors and ambient conditions. Additionally, we estimate machine learning models to identify relationships not captured by current models. Results show that machine learning approaches provide more accurate fuel consumption predictions than traditional methods. Notably, empirically calibrated thrust settings are higher than literature defaults. However, unexplained variance remains, suggesting potential for further improvements. Our findings offer valuable insights for integrating emission models into airport operations, helping stakeholders implement effective emission reduction strategies.