Development and numerical optimization of an Innovative Adhesion Model in the Railway field Benedetto Allotta a , Monica Malvezzi b , Enrico Meli a , Luca Pugi a , Alessandro Ridolfi a , Andrea Rindi a a University of Florence, Dept. of Industrial Engineering, MDM Lab, via Santa Marta 3, 50139, Florence, Italy b University of Siena, Dept. of Information Engineering, via Roma 56, 53100, Siena, Italy a.ridolfi@unifi.it The simulation of the railway braking under degraded adhesion conditions is very important since it affects both safety and interoperability issues of the railway network; unfortunately a realistic adhesion law is not easy to be defined and implemented because of the complex and non-linear behaviour of the adhesion coefficient. The presence of contaminants drastically reduces the available adhesion coefficient between the rolling surfaces of wheel and rail producing a limitation of the tangential forces that can be exchanged on the contact patch. Particularly under degraded adhesion conditions very high slidings occur and they usually produce a high dissipation of energy which has a cleaning effect on the rolling surfaces (i.e. the contaminants are removed), that allows to partially recover the adhesion coefficient . In this work this energetic criteria has been studied and the authors propose the implementation of an innovative adhesion law to the simulation of railway multibody models, with 3D multi-point contact algorithms. A benchmark case, the braking of a coach with WSP system is simulated and compared with experimental data available from testing; the new adhesion law permits to match the experimental reference results and to carry out simulated braking tests, with working WSP system, that comply with the current regulations. To achieve these results an accurate tuning of the characteristic parameters of the new adhesion model is mandatory: in particular the authors compare in the paper different numerical methods to fit in the best possible way the experimental data coming from the available tests. Inside the research activity many optimization techniques have been tested, based both on least square approaches and on derivative free heuristic methods: the proposed numerical procedures are quite suitable for this kind of problems (high degraded adhesion condition analysis), where there is a substantial lack of a stable theoretical background, representing an acceptable trade-off between good fitting of experimental data and applicability over different operating scenarios. 1. Introduction The crucial importance of an accurate adhesion modelling in tribology, vehicular and railway applications is mainly related to its deep impact on the dynamics, the wear processes and the safety of the considered systems, both theoretically and practically. As it is well known, the adhesion coefficient shows a highly complex and non-linear behaviour, especially when degraded adhesion and large sliding between the contact surfaces occur and when external unknown contaminants are present. Therefore the modelling and the complete understanding of degraded adhesion are today important open problems. Concerning the state of the art of the discipline and, particularly, the railway field where the multibody approaches are prevailing, the contact and adhesion models usually employed in the research activities do not consider the complex degraded adhesion behaviour and the presence of external contaminants. [1][2][3][4] However, in the last decades, many important studies and analyses have been performed to investigate the role of the so-called third body between the contact surfaces, e.g. wheels and rails. In particular the analyses have been carried out both on laboratory test rigs and through on-track railway tests by taking into account natural and artificial external contaminants and friction modifiers. [5][6] At the same time, also the main phenomena characterizing the