Detection and modelling of incipient failures in internal combustion engine driven generators using Electrical Signature Analysis

Condition-based maintenance of electric generators have been gaining increasing importance due to the electricity demand and the criticality that this equipment represents to electrical power systems. In this context, this paper proposes a methodology and a system for detection and modeling of incipient failures in the components of internal combustion engine-driven generators based on Electrical Signature Analysis (ESA). The proposed methodology enables the detection of incipient faults both in the prime mover and in the coupled synchronous generator, only relying on measurements of the generator stator voltages and currents. The proposed ESA failure patterns are based on defined frequencies and the structural features of the machine, so they can be reproduced in a wide range of engine-generators sets. The main advantages of the proposed system are its low intrusiveness, feasible installation and cost efficiency. A scale model laboratory has been designed to simulate faults in a small diesel generator and apply the ESA methodology to detect these faults and obtain the failure patterns. Experimental results are presented to prove the effectiveness of the proposed methodology. The main results include the findings that exciter generator unbalance induces electrical unbalance components, exciter diode short circuit induces even harmonics, intake valve failure and piston ring failure induce multiples of rotation frequency components, and mechanical misalignment of the engine generator set induces multiples of half order speed frequency components on ESA. Moreover, the proposed prototype is installed at two large in-service internal combustion engine-driven generators and examples of signal analysis are provided.