This paper presents a stability analysis of the Morton effect. The analysis is an extension of the Murphy and Lorenz method  and is based on better estimates of three influence coefficients linking the phenomena contributing to the Morton effect: the total response to the rotor unbalance, the temperature difference on the rotor surface induced by synchronous vibrations and the thermomechanical deformation of the rotor.
The models used in the present work are more complex and accurate because they are based on the non-linear unbalance response (large amplitude vibrations) of the rotor, on the non-isothermal analysis of the journal bearing flow and on a three-dimensional thermos-elastic analysis of the rotor. The results obtained with the original stability analysis of Murphy and Lorenz and with the modified one are compared with original experimental data obtained for a short (rigid) and long (flexible) rotor guided by a ball bearing and by a cylindrical bearing and presented in a previous work . Both methods confirm the experimental results obtained for a short (rigid) rotor. They show that this rotor is not subject to instabilities generated by the Morton effect. However, the results obtained for a long (flexible) rotor are different. The simplified method of Murphy and Lorenz shows a stable behavior while the modified method presented in this work confirms the findings of  and indicates that the rotor could be subject to a Morton effect at rotational speeds close to the experimental conditions. The improvements obtained by using the modified stability analysis are therefore clearly underlined, as well as its inherent limitations.