Modern rotating machinery is prone to fatigue cracks due to the severe working and continuously varying loading conditions. To avoid failure of the rotating systems due to these cracks, the system needs to be continuously monitored. The reliability of the machinery can be enhanced with the fault identification at the early stages of their occurrence. Model based methods are used for crack identification in rotating systems successfully. Model based methods are of different types. One of the methods, the equivalent loads minimization method is studied here. It has a limitation that the error in identified crack depth increases with decrease in number of measured vibrations. In the present work, the theoretical fault model loads used in least squares minimization algorithm are transformed using modal expansion to reduce error in the identified fault parameters. The crack is identified accurately even in the case of less number of measured vibrations. The method is applied for crack identification in a centrifugal pump rotor using transverse vibrations. The results presented are obtained purely from FEM simulations. Copyright © 2010 by ASME.