Title: Runway excursions: Risk assessment with Bayesian network models
Authors: Fernando Calle-Alonso - University of Extremadura (Spain) [presenting]
Carlos Javier Perez Sanchez - University of Extremadura (Spain)
Eduardo Sanchez Ayra - Iberia Airlines (Spain)
David Rios Insua - CMAT-CSIC (Spain)
Abstract: Aviation has grown from the first commercial flight in 1914 up to the present, achieving for the first time more than 100,000 landings a day in 2014. Forecasts are very optimistic, expecting to duplicate the number of passengers in 2030. Although accidents are extremely rare in this sector, their consequences can be dramatic. One of the most important problems are runway excursions, taking place when the plane overexceeds the end (over-run) or the lateral (ver-off) of the landing strip. In order to analyze the most influential variables for runway excursions and to provide information on risk scenarios in landing procedures, a Bayesian network-based approach has been considered. A model selection has been performed among different Bayesian networks. The proposed methodology has been applied to a database containing a number of variables related to landing operations on three landing strips. A runway excursion risk ranking is provided for the three landing strips, and the main risk variables (and states) are identified.