Shravankumar, C2015-09-282023-10-262015-09-282023-10-262014ROLL NO. 09610304https://gyan.iitg.ac.in/handle/123456789/572Supervisor: Rajiv TiwariCondition-based maintenance is now-a-days recognized as the most efficient strategy for carrying out maintenance in industrial machineries. Fatigue cracks in the shaft are a very common but catastrophic fault, as accidents in turbine rotors have been reported since 1950s. Also, the shaft failure due to fatigue crack causes increase in downtime and maintenance costs. Condition monitoring employs methodologies broadly ranging from the signal-based and model-based methods to the model testing and non-traditional non-parametric methods. In this work, model based identification algorithms have been developed to identify a crack in a shaft. A switching crack rotor model from the literature is studied. The crack is identified using a single parameter, i.e. the additive stiffness it introduces in the shaft section. Unbalance eccentricity and viscous damping of the rotor are also identified simultaneously. Displacement responses in time domain are used as input to developed identification algorithms. The algorithms are then extended to identify crack forces, the estimates which can aid in understanding the actual breathing mechanism of a crack. The crack force and displacement time histories are transformed to frequency domain by means of full spectrum. This is done to directly use frequency responses as input to the identification algorithm, rather than using time responses. Compared to a normal spectrum, a full spectrum is helpful in obtaining the coefficients of reverse whirling frequency components, which constitute the crack force excitation and displacement responses.enMECHANICAL ENGINEERINGThesis