EXPERIMENTAL VALIDATION OF A ROTOR-BEARINGS SYSTEM MODEL FOR TESLA TURBINES RESONANCES PREDICTION

The main task of the present paper is the development and the implementation of a suitable analytical model to correctly capture rolling bearing radial stiffness, particularly nearby the critical speeds of the investigated rotor-bearings system. In this paper, such bearing non-linear stiffness model is applied to air bladeless turbines (or Tesla turbines) high speed rotors, in order to assess their global rotordynamic behavior when they are mounted on hybrid ball bearings. In order to properly investigate all the issues related to critical speeds identification, an adequate number of tests was performed by exploiting an experimental air Tesla turbine prototype located at TPG experimental facility of the University of Genoa. The correlation between experimentally detected flexural critical speeds and their numerical predictions is markedly conditioned by the correct identification of ball bearings dynamic characteristics; in particular, bearings stiffness effect may play a significant role in terms of rotor-bearings system natural frequencies and therefore it must be accurately assessed. Indeed, Tesla turbine rotor FE model previously employed for numerical modal analysis relies on rigid bearings assumption and therefore it does not account for bearings stiffness overall contribution, which may become crucial in case of "hard mounting" of rotor-bearings systems. Subsequently, high-speed air Tesla rotor is investigated by means of an enhanced FE model for numerical modal analysis within Ansys((R)) environment, where ball bearings are modelled as springs whose stiffness is expressed according to non-linear analytic model implemented in Matlab((R)). The obtained results in terms of rotor-bearings system modal analysis exhibit an improvement in experimental-numerical results correlation by relying on such ball bearing stiffness model; moreover, beam-based FE model critical speeds predictions are coherent with experimental evidence and with respect to the previously employed FE model based on solid elements it is characterized by lower computational time and it is more easily interpretable. Thus, such experimentally validated numerical model represents a reliable and easily adaptable tool for high-speed rotating machinery critical speeds prediction in practical industrial application cases.