Purpose This study aims to evaluate the technical efficiency (TE) of fuel flow during aircraft descents at Sabiha Gökçen Airport, the second busiest in Türkiye, and identify the operational factors most strongly influencing descent efficiency. Design/methodology/approach Real flight data (n = 42) from a commonly used narrow-body aircraft equipped with a CFM56-7B engine were analysed. Variable selection was performed using the least absolute shrinkage and selection operator (LASSO), followed by panel data analysis and stochastic frontier analysis (SFA), to estimate TE scores and determine key fuel flow drivers. The analysis reflects the behaviour of a single aircraft type operating in a specific airport environment. Findings Most flights achieved TE scores between 0.50 and 0.75, with the most efficient ranging from 0.70 to 0.74. Fuel efficiency was primarily affected by flight path angle (FPA) and true airspeed (TAS). Moderate FPAs (2.5°–3.0° at 3,000–18,000 ft and 3.5°–4.0° at 18,000–24,000 ft) and balanced speeds (191–270 knots) were associated with lower fuel consumption. These numerical results are context-dependent and should be interpreted as proof-of-concept rather than universally generalizable findings. Research limitations/implications Detailed flight profiles cannot be shown in the manuscript because individual descent trajectories derived from flight data record data are confidential. The data set is limited to one aircraft type and one airport; therefore, the empirical results should be interpreted within this operational context. Practical implications The results provide preliminary, context-specific strategies that may help airlines, pilots and air navigation service providers optimize descent profiles, reduce fuel burn and lower emissions at high-density airports. Originality/value To the best of the authors’ knowledge, this study is the first to integrate real flight data, LASSO-based variable selection and SFA to quantify descent fuel efficiency at the airport level, demonstrating a methodological framework that can be extended and validated using broader data sets in future research.
Yenilmez et al. (Sat,) studied this question.