Robust Runway Scheduling using a time-indexed model

Abstract

The runway system is the main element that combines airside and groundside of the ATM System. Efficient models and planning algorithms are required. The best planning algorithm, however, is useless if the resulting plans cannot be implemented in the real word. This often happens because the input data of the planning algorithms is disturbed respectively it changes. For example, an estimated time of an aircraft is not stable for the next ten hours. These disturbances are not deterministic, but often their stochastic distributions with mean values and standard deviations are known. We present a robust model together with an optimization algorithm which explicitly incorporates the knowledge of uncertainty into each planning step. Our approach transforms the planning problem into an assignment problem with side constraints. We compute for every discretized point in time whether an aircraft is scheduled and if so, which one is. Experimentally, we show that running times are better than when using a different non-linear integer optimization model. Our Monte-Carlo simulation for a mixed-mode runway system shows that our approach results in fewer sequence changes and target time updates compared to the usual approach of just updating the plan if the actual plan is not feasible any more.