Sensitivity Analysis of Fault Indicators in AC Drives

The combination of ac machines and power converters can be considered standard in modern industrial drive applications. Besides of all the well known benefits resulting from this kind of operation there are however also some side-effects like increased outage times resulting from defects in different components of the machine. This reduction in reliability can be one major concern with the increasing demands on system reliability and especially in security critical application of drives in by wire operation. When looking at the reasons of failures and breakdowns it is possible to identify three main components that are responsible for about 95% of all breakdowns. According to different studies on the breakdown of induction machines these components are the bearings (~50%) followed by the winding isolation (~35%) and the rotor cage (~10%). While a defect in the windings isolation needs immediate reaction to prevent further damage, bearing and cage defects usually develop slowly. In order to reduce the costs of preventive maintenance as well as to increase the reliability of the whole drive system, a reliable monitoring of the machine is thus of more and more importance. Many different methods have been proposed in the past to in the field of condition monitoring. However, most of the methods are limited to mains-fed, as well as steady state operation. There are only very few methods that enable a detection of fault conditions during inverter operation and that are not limited to specific, steady state points of operation. In a previous project financed by the FWF (P17595) different fault indicators have been developed and proposed that enable a detection of the mentioned fault conditions even during inverter operation. In order to further develop these fault indicators towards a monitoring scheme with possible industrial application it is, however imperative to answer the following main questions: - What are the drive operating limits for a reliable detection when using the different fault indicators? - Is it possible to identify threshold values to reliably distinguish between a healthy or faulty motor? - What is the influence of machine size and design on the performance of the fault indicators? These three main topics are going to be addressed in the proposed project. The first topic (drive operating limits of reliable detection) covers the influence of the machines point of operation, the current sensors used, inverter non-idealities, as well as the pulse width modulation method. These influences will be addressed by measurements on a special test stand based on the results and experience obtained from the previous project. For the identification of the threshold value (second topic) a special set of test measurements will be defined in combination with a training algorithm to identify the specific behaviour of the fault indicators with respect to the point of operation. The influence of the machine size and design will be addressed by finite element simulations that have to be verified by measurements on specific machines to allow a prediction of the detection capabilities. The proposed project is based on results obtained from a former project financed by the FWF.

The combination of ac machines and power converters can be considered standard in modern industrial drive applications. Besides of the well-known benefits resulting from this kind of operation there are however, also some side-effects like increased outage times resulting from increased stress in different components of the machine. This reduction in reliability can be one major concern with the increasing demands on system reliability and especially in security critical application of drives in by-wire operation.When looking at the reasons of failures and breakdowns it is possible to identify three main components that are responsible for about 95% of all breakdowns. According to different studies on the breakdown of induction machines these components are the bearings (~50%) followed by the winding isolation (~35%) and the rotor cage (~10%).While a defect in the windings isolation needs immediate reaction to prevent further damage, bearing and cage defects usually develop slowly. In order to reduce the costs of preventive maintenance as well as to increase the reliability of the whole drive system, a reliable monitoring of the machine is thus of more and more importance.Most methods proposed in the past in the field of condition monitoring are limited to mains-fed, as well as steady state operation. There are only very few methods that enable a detection of fault conditions during inverter operation and that are not limited to specific, steady state points of operation. At this research institution different fault indicators have been developed and proposed that enable a detection of the mentioned fault conditions even during inverter operation. In order to further develop these fault indicators towards a monitoring scheme with possible industrial application the following investigations were done in the current follow up project:Are there operating limits for a reliable detection when using the different fault indicators?Is it possible to identify threshold values to reliably distinguish between a healthy or faulty motor?What is the influence of machine design on the performance of the fault indicators?The first two topics (drive operating limits of reliable detection) cover the influence of the machines point of operation, the current sensors used, inverter non-idealities, as well as the pulse width modulation method. These influences were addressed by measurements on a special test stand based on the results and experience obtained from the previous project. The influence of the machine size and design has been addressed by finite element simulations that were verified by measurements on specific machines to allow a prediction of the detection capabilities.Based in the detailed investigations performed, a new indicator was developed that enables the measurement of the health state of the machines winding insulation. Thus a continuous monitoring of the insulation is possible even bevor an actual failure (short circuit) develops.