Abstract Enhancing logistics performance in the E&P industry is historically a major factor in cost efficiency. This paper presents an AI-driven operational tool supporting helicopter operations in the Danish North Sea, joining cost and carbon footprint benefits by enabling reductions of flight hours and rotations through continuous re-optimization of the seat and routes offerings. The primary challenge of such automated optimization solutions lies in integrating mathematically optimal outcomes with real-world operations that demand considerable flexibility. This paper details the implementation and insights gained from an operations research-based (OR) optimization digital solution, internally named ‘PAX’. The tool provides operational recommendations by utilizing state-of-the-art models while considering real-life constraints. Method wise, the strategy is to integrate Operations Research algorithms open-sourced by Google AI, encompassing the Capacitated Vehicle Routing Problem with Time Windows and Pickup and Delivery model (CVRPTWPD) with business specific algorithmic approaches. The business specific challenges that need to be addressed include the implementation of an optimal split-delivery search component, the management of various time-window constraints, and finally the design of seats assignments and departures. In the Danish part of the North Sea, where many near-neighbor platforms constitute a complex logistics network, the solution reaches its goal: scheduling and routings of helicopters are optimized, with a good compromise with operational flexibility. While operations research models are not frequently used in practice due to difficulties dealing with actual real-world processes, PAX is the result of a multi-disciplinary approach that succeeds in delivering prescriptive, optimal and operationally relevant recommendations. Since July 2024, the optimization of helicopters routing and scheduling and associated split delivery have successfully guided the operations teams to save multiple flights and run consistently the shortest combined routes. Importantly as well, the solution's design allows transparency and explainability. Theoptimization of operations is measurable, providing valuable insights: cost indicators and seat utilization are fully contextualized, enabling a complete understanding of logistics performance. The main two novel components of this innovative method are: the implementation of split delivery added to the classic VRP (Vehicle Routing Problem) models. Indeed, helicopters have the specificity to be generally too small vehicles for the statistical quantities to transport, which requires a powerful split delivery heuristic search method to ensure optimality of the VRP.the multi-time window scenario approach, that enables the integration of a Passenger-Transportation-System with advanced open-source Operations Research models. The shared lessons learned have potential to be applied in other contexts of our industry.
Dubourg et al. (2025) studied this question.