The present article investigates the dynamic behavior of a fully assembled turbogenerator system influenced by misalignment. In the past, most of the researchers have neglected the foundation flexibility in the turbogenerator systems in their study, to overcome this modelling error a more realistic model of a turbogenerator system has been attempted by considering flexible shafts, flexible coupling, flexible bearings and flexible foundation. Equations of motion for fully assembled turbogenerator system including flexible foundations have been derived by using finite element method. The methodology developed based on least squares technique requires forced response information to quantify the bearing–coupling–foundation dynamic parameters of the system associated with different faults along with residual unbalances. The proposed methodology is tested for the various level of measurement noise and modelling error in the system parameters, i.e., 5% deviation in E (modulus of elasticity) and ρ (density), respectively, for robustness of the algorithm. In a practical sense, the condition analyzed in the present article relates to the identification of misalignment and other dynamic parameters viz. bearing and residual unbalance in a rotor integrated with flexible foundation.

In rotating machineries, misalignment is considered as the second most major cause of failure after unbalance. In this article, model-based multiple fault identification technique is presented to estimate speed-dependent coupling misalignment and bearing dynamic parameters in addition with speed independent residual unbalances. For brevity in analysis, a simple coupled rotor bearing system is considered and analytical approach is used to develop the identification algorithm. Equations of motion ingeneralized co-ordinates are derived with the help of Lagrange’s equation and least squares fitting approach is used to estimate the speed-dependent fault parameters. Present identification algorithm requires independent sets of forced response data which are generated with the help of different sets of trial unbalances. To avoid/suppress the ill-conditioning of regression equation, independent sets of forced response data are obtained by rotating the rotor in clock-wise and counter clock-wise directions, alternatively. Robustness of algorithm is checked for different levels of measurement noise.