Runway operations planning: a two-stage solution methodology

The airport runway is a scarce resource that must be shared by different runway operations (arrivals, departures and runway crossings). Given the possible sequences of runway events, careful runway operations planning (ROP) is required if runway utilization is to be maximized. Thus, runway operations planning (ROP) is a critical component of airport operations planning in general and surface operations planning in particular. From the perspective of departures, ROP solutions are aircraft departure schedules developed by optimally allocating runway time for departures given the time required for arrivals and crossings. In addition to the obvious objective of maximizing throughput, other objectives, such as guaranteeing fairness and minimizing environmental impact, may be incorporated into the ROP solution subject to constraints introduced by air traffic control (ATC) procedures. Generating optimal runway operations plans was approached in the paper by Anagnostakis et al. (2001) with a "one-stage" optimization routine that considered all the desired objectives and constraints, and the characteristics of each aircraft (weight class, destination, air traffic control (ATC) constraints) at the same time. Since, however, at any given point in time, there is less uncertainty in the predicted demand for departure resources in terms of weight class than in terms of specific aircraft, the ROP problem can be parsed into two stages. In the context of the departure planner (DP) research project, this paper introduces runway operations planning (ROP) as part of the wider surface operations optimization (SOO) and describes a proposed "two stage" heuristic algorithm for solving the runway operations planning (ROP) problem. Focus is specifically given on including runway crossings in the planning process of runway operations. In the first stage, sequences of departure class slots and runway crossings slots are generated and ranked based on departure runway throughput under stochastic conditions. In the second stage, the departure class slots are populated with specific flights from the pool of available aircraft, by solving an integer program. Preliminary results from the algorithm implementation on real-world traffic data are included in (Anagnostakis and Clarke, 2002).