ORCI: AI-based predictive spacing support for point merge operations — Case study in Lisbon (LPPT)

Background In Point Merge operations, the timing of the air traffic controller’s instruction clearing an aircraft from a sequencing leg towards the merge point is the primary tactical mechanism for controlling arrival spacing. Once this instruction is issued, the ability to further modify inter-aircraft separation is largely limited to speed-control interventions, making this decision critical for efficient arrival management. Methods This paper presents a data-driven decision support approach that predicts the spacing between consecutive aircraft at the merge point resulting from a given direct-to instruction. The study focuses on arrival operations to runway 02 at Lisbon Humberto Delgado Airport (LPPT), where arrivals are managed through a two-leg RNAV Point Merge configuration converging at the PESEX merge point. Explainable machine learning models were trained using historical radar surveillance data to estimate the expected spacing before the tactical instruction is issued. Results The proposed models accurately captured the spacing dynamics of Point Merge operations. Linear regression models achieved a mean absolute error of approximately 0.5 NM. In addition, a weighted training strategy reduced non-conservative spacing overestimations, shifting the error distribution towards safer operational outcomes while maintaining strong predictive performance. Conclusions The results demonstrate the feasibility of using explainable machine learning to support tactical spacing decisions in Point Merge operations. By providing real-time estimates of the resulting spacing, the proposed decision support tool has the potential to enhance controller situational awareness, improve spacing consistency, and support more efficient arrival management while preserving the controller’s authority over all operational decisions.