Document Type


Publication Date


Publication Title

Journal of Engineering for Gas Turbines and Power


Atransverse shaft crack is a serious malfunction that can occurdue to cyclic loading, creep, stress corrosion, and other mechanismsto which rotating machines are subjected. Though studied for manyyears, the problems of early crack detection and warning arestill in the limelight of many researchers. This is dueto the fact that the crack has subtle influence onthe dynamic response of the machine and still there areno widely accepted, reliable methods of its early detection. Thispaper presents a new approach to these problems. The methodutilizes the coupling mechanism between the bending and torsional vibrationsof the cracked, nonrotating shaft. By applying an external lateralforce of constant amplitude, a small shaft deflection is induced.Simultaneously, a harmonic torque is applied to the shaft inducingits torsional vibrations. By changing the angular position of thelateral force application, the position of the deflection also changesopening or closing of the crack. This changes the waythe bending and torsional vibrations are being coupled. By studyingthe coupled lateral vibration response for each angular position ofthe lateral force one can assess the possible presence ofthe crack. The approach is demonstrated with a numerical modelof a rotor. The model is based on the rigidfinite element method (RFE), which has previously been successfully appliedfor the dynamic analysis of many complicated, mechanical structures. TheRFE method is extended and adopted for the modeling ofthe cracked shafts. An original concept of crack modeling utilizingthe RFE method is presented. The crack is modeled asa set of spring-damping elements (SDEs) of variable stiffness connectingtwo sections of the shaft. By calculating the axial deformationsof the SDEs, the opening/closing mechanism of the crack isintroduced. The results of numerical analysis demonstrate the potential ofthe suggested approach for effective shaft crack detection.