PSI - Issue 39

Nabam Teyi et al. / Procedia Structural Integrity 39 (2022) 333–346 Author name / Structural Integrity Procedia 00 (2019) 000–000

337

5

1.2. Analysis of cracks Crack detection or identification provides a qualitative indicator of the presence of a fault in the machine, whereas crack classification determines if the fault is a rotor crack or another type of fault. It is possible to determine the location of the crack in the rotor using crack localization; however, it is more difficult to determine the shape and size or severity of a crack using crack assessment. Crack prediction is a method of estimating the remaining useful life of a rotating component. 2. Relevant literatures from 2010 to 2012 Karthikeyan and Tiwari (2010) developed flaw algorithm for detection, localization and sizing of flaw in circular beam on forced response measurement. They experimentally obtained the displacement responses of the flawed beam using transducers and proximity sensors. FE model was developed as well. Flaw model was in good agreement with the experimental results. However, they overestimated the flaw size due to inherent low sensitivity owing to the beam being very thin and relatively long. Based on previous research, Srinivas et al. (2010) projected that spinning machinery would exhibit higher vibration frequencies if it had unbalance and a crack in the shaft. A three-layered networked ANN trained with the Levenberg-Marquardt algorithm and the wavelet transform processed the data. The WT and ANN were used to accomplish three distinct tasks: data collection, feature extraction, and fault detection. Barella et al. (2011) studied a 60 MW thermal power plant rotor turbine failure. After approximately 10 years of operation, the CrMoNiV steel rotor failed. The cause of the crack was discovered not to be corrosion or fretting fatigue but rather a heat gradient during startup and the resulting high stress concentration which was amplified by the indentions. Based on fractography analysis, Momcilovic et al. (2011) investigated a major failure of a 28 MW horizontal hydro turbine shaft in terms of load carrying capacity of critical radius, and concluded that the seal box design resulted in constant flow of river water in the critical radius zone, resulting in corrosion fatigue cracks and major turbine shaft failure. In order to assess the rotor's health, Chana et al. (2011) used an eddy current sensor and Reasoner software to isolate the crack. LCF spin-pit testing was performed on a pre-flawed disc. The QinetiQ eddy current sensor based tip timing device collected time of arrival data to evaluate blade movement induced by disc crack propagation. A probabilistic model of the component's remaining useful life was developed to account for measurement uncertainty. Cheng et al. (2011) found that broad equations of motion for a Jeffcott rotor/de Laval roror with a transverse crack acknowledged the constraints of weight supremacy. The shape of the fracture in the Jeffcott rotor was modelled with fixed and rotating coordinates, and breathing cracks with and without weight dominance. The numerical analysis of a fractured rotor was done at critical speed. This speed was determined by the unbalance orientation angle of a damaged rotor. On the other hand, Dong and Chen (2011) introduced crack identification methods for one unknown crack case and two unknown cracks case for stationary rotors. The FE model used transfer matrix analysis, local flexibility theorem, and contour diagram analysis to locate cracks. Changes in Eigen frequencies of broken rotors were dependent on crack closeness to mode shape nodes, and these changes monotonically increased with crack depth for rotors with low slenderness ratio. Kulesza and Sawicki (2012) suggested a rigid Finite Element (RFE) approach to represent rotating machinery. The RFE model was verified using the Campbell diagram. The breathing mechanism was intuitively explained by modest scalar stochastic differential equations (SDEs) stiffness changes. The eccentricity value and its angular location influenced the damaged rotor’s breathing behaviour. Ricci and Pennacchi (2012) explored system anisotropy and crack position using a model of a genuine hyperstatic rotor with several degrees of freedom. The generator was modelled via FE. For the generator model, they employed linearized stiffness and damping coefficient analysis, and steering function for crack breathing. The Floquet analysis revealed that there was no instability. And the behaviour matched previous reports of significant cracks in generators. A new method for analysing the nonlinear dynamics of Jeffcott rotors/de Laval rotors was proposed by Rubio and Fernández-Sáez (2012). The linearized equations of motion were solved implicitly by Newmark. The nonlinear rotor orbits and resonance curves were studied. The model was efficient and saved time (upto 100 times). However, when the vertical whirl amplitude exceeded the shaft weight static deflection, the simplified quasi-static stiffness matrix approaches failed. Singh and Tiwari (2012) investigated the influence of steps in the shaft on crack detection near and distant from the step. They imitated the rotor system mathematically. Rotor parameter normalised quadratic coefficients were studied. Their programme spotted cracks at