Desenvolvimento de modelos de previsão de coeficiente de atrito em pista de pouso e decolagem dos aeroportos internacionais de Fortaleza e Brasília com Redes Neurais Artificiais

Abstract

Landing and take-off operations represent the most critical phases of a flight, in which several factors influence, such as the pilot's skill, weather conditions and tire-pavement adherence. In this context, the texture parameters of the coatings on airport pavements, represented by the coefficient of friction and the macrotexture, are objective criteria for operational safety in terms of tire-pavement adherence. These parameters are influenced by different aspects, such as the age of the pavement covering, the relative humidity of the air, the temperature, the accumulation of rubber on the runways and takeoff from aircraft tires, and the presence of transverse grooves, among others. In view of this, the general objective of this research is to develop prediction models for the coefficient of friction measured at 3 and 6 meters from the axis of landing and takeoff runways by GripTester, Skiddometer and Mu-meter equipment in different locations and coating surface conditions in order to assist the aerodrome operator in ensuring operational safety. In addition, this research also aimed to verify the influence of grooving on the performance of the coefficient of friction. That said, the technique of Artificial Neural Networks was used to develop predictive models of the coefficient of friction obtained through reports of measurements in the field, which were carried out at the international airports of Fortaleza and Brasília, between the years of 2012 and 2018. In addition to these data, information about weather conditions and the number of landing and takeoff operations obtained by the companies operating the aerodromes and by the National Civil Aviation Agency (ANAC) was also applied. Thus, this study produced efficient models to predict the coefficient of friction, in addition to providing a contribution to existing models. The developed models showed satisfactory results given the complexity of the problem, demonstrating that, despite the need for future improvements, they can contribute to the safety of landing and takeoff operations and to the management of airport pavements by operators of Brazilian aerodromes, as well as helping inspection services by ANAC.