IEEE TRANSACTIONS ON ENERGY CONVERSION, VOL. 26, NO. 2, JUNE 2011 671 Influence of Rotor Position in FRA Response for Detection of Insulation Failures in Salient-Pole Synchronous Machines Carlos A. Platero, Member, IEEE, F. Bl´ azquez, Member, IEEE, Pablo Fr´ ıas, and Dionisio Ram´ ırez, Member, IEEE Abstract—Frequency response analysis (FRA) is a very common test for the diagnosis of power transformers. This paper presents some relevant results on the application of FRA to the diagnosis of rotating machines. First, a reference rotor position for obtaining the reference FRA response of a rotating machine is proposed. Then, FRA with the proposed rotor position is used to identify faults in the stator of the machine. This paper studies turn-to-turn and ground faults in the stator for different fault resistance values. Several laboratory tests demonstrate the applicability and value of the use of FRA in the diagnosis of rotating machines. Index Terms—Fault diagnosis, fault location, frequency response analysis, synchronous machines. I. INTRODUCTION A COUNTRY’s industrial development is directly related to its energy consumption, mainly electricity. To maintain and increase the development level, it is necessary to have a reliable power system, both in the generation and transportation of electric energy. Therefore, the development of diagnostic and protective techniques for generators and transformers is a primary concern. Frequency response analysis (FRA) is a well-known tech- nique for the diagnosis of power transformer windings. Since the first proposals for FRA [1], many improvements and new appli- cations have been developed that have resulted in new standards for the application of the FRA method, in both International Council on Large Electric Systems (CIGRE) and IEEE [2]. Since its appearance, the industrial interest of FRA has been in the possibility of identifying small strains that could appear in the coils of power transformers as a result of forces that occur during a short circuit or any possible shocks during transporta- tion [3], [4]. Basically, this technique is based on analyzing the impedance of the windings in the frequency domain. Since a winding can be modeled as an equivalent circuit with a complex network Manuscript received August 1, 2010; revised January 3, 2011; accepted January 4, 2011. Date of publication February 24, 2011; date of current version May 18, 2011. Paper no. TEC-00312-2010. C. A. Platero, F. Bl´ azquez, and D. Ram´ ırez are with the Electrical Engineering Department, Escuela T´ ecnica Superior de Ingenieros Indus- triales, Polytechnic University of Madrid, Madrid 28006, Spain (e-mail: carlosantonio.platero@upm.es; fblazquez@etsii.upm.es; dionisio.ramirez@ upm.es). P. Fr´ ıas is with the Department of Electrical and Power Systems, Universidad Pontificia Comillas Madrid, Spain (e-mail: pablo.frias@iit.upcomillas.es). Color versions of one or more of the figures in this paper are available online at http://ieeexplore.ieee.org. Digital Object Identifier 10.1109/TEC.2011.2106214 of capacitances, inductances, and resistances, its frequency re- sponse is unique. Thus, any alteration affecting the winding results in a variation in the equivalent circuit, and, therefore, in its frequency response. Applying the FRA technique correctly requires accurately characterizing the winding with an equivalent circuit [5]–[7]. However, the interpretation of the FRA results is sometimes complex because these results may be affected by exogenous situations such as the magnetization level of the core [8] or the connections of the measuring equipment [9]. Thus, reliable FRA requires many detailed laboratory tests [10]. The detection of winding deformation is not the only use of this technique. FRA can also be utilized to detect faults in the windings [11] and some authors suggest the possible detection of insulation failure in its early stage [12]. This latter application could be used in a predictive maintenance system. As concerns the protection of generating units, the literature presents different systems for detecting ground faults of both the rotor [13] and the stator winding [14], [15]. Some systems are also described that detect turn-to-turn faults and even in- terlaminar currents in the stator core [16]. Occasionally, other additional advanced diagnostic methods based on the analysis of high-frequency signals have been proposed [17]–[19]. However, despite its apparent advantages in power transform- ers, the FRA technique has gone practically unused in the diag- nosis of rotating machines. This is because the high-frequency equivalent circuit of the windings in rotating machines is more complex than that of transformers: the stator winding is divided into slots, and there is a rotor winding. Some authors propose the use of FRA for induction mo- tors [20] or for synchronous generators. This application would require removing the rotor in order to eliminate its influence in the analysis [21]. The need to remove the rotor makes this method less appealing from an industrial point of view. In order to avoid removing the rotor, this paper presents the results of a new implementation of FRA for diagnosing wind- ings on rotating electrical machines. During the laboratory tests, we encountered problems obtaining reliable measurements, es- pecially in salient-pole synchronous machines. This paper aims to lay the background for the future implementation of FRA in the diagnosis and maintenance of rotating machines in industry. This paper is structured as follows: Section II presents the operating principle of FRA diagnosis. The next section analyzes the results for applying FRA response to rotating machines and proposes a positioning of the rotor to obtain the reference FRA response. Section IV applies the previous rotor positioning to 0885-8969/$26.00 © 2011 IEEE