Condition monitoring of capacitors in power electronic converters

• Main Author: Sundararajan, Prasanth
• Other Authors: -
• Format: Thesis-Doctor of Philosophy
• Language: English
• Published: Nanyang Technological University 2020
• Subjects: Engineering::Electrical and electronic engineering
• Online Access: https://hdl.handle.net/10356/137800
• Institution: Nanyang Technological University

Marine and aerospace industries are driven towards more electric aircraft and ships due to the demand to optimise performance using electrical systems which are more efficient than the traditional mechanical systems and to reduce greenhouse gas emissions caused by fossil fuels. With the indispensable nature of the power converters in applications such as hybrid propulsion, electric vehicles, and many other safety-critical systems, condition monitoring of power converters has become essential to ensure high safety and reliability. A power converter is made up of components like capacitors, inductors, power semiconductor devices, gate drivers, sensors and a control unit. The unexpected failure of one or more of these components may lead to system downtime. In power converters, aluminium electrolytic capacitor (AEC) plays significant roles such as filters, snubbers, and energy storage elements. However, they are also one of the most vulnerable components due to their high failure rates compared to other parts in a power converter. Therefore, with the increase in the use of AECs in critical applications, condition monitoring of AECs to prevent unexpected failures has become inevitable. The main factors affecting the lifetime of an AEC are environmental factors such as ambient temperature, vibration, humidity, etc., and electrical factors such as operating voltage, ripple current, and charge-discharge duty ratio. Several research findings have been published regarding the condition monitoring of AECs. However, none or a limited number of the methods are adopted by industries because the existing condition monitoring approaches are either unreliable or highly complicated and expensive to implement, which make them impracticable. Hence, this research focuses on developing and implementing a novel, reliable, and cost-effective prognostic and health monitoring technology for AECs to predict the failure during the early stages of its degradation and thereby to enable preventive maintenance. The first phase of research focuses on the study of construction, physical properties, equivalent circuit models, and failure mechanisms of an AEC to develop reliable methods to identify and track the dominant failure mechanisms. The most common failure mechanism in an AEC is the electrolyte evaporation, which causes an increase in equivalent series resistance (ESR) and a decrease in capacitance. Therefore, the ESR and capacitance are selected as the indicators to track AEC degradation. Reliable and accurate parameter estimation methods are developed to estimate the ESR and capacitance. The developed methods are tested experimentally in a three-phase inverter test rig and found that the error in estimation is less than 3% for ESR and less than 1% for capacitance. The degradation indicators are sensitive to both operating temperature and frequency. Therefore, they must be normalised before decisions can be taken. The second phase of research focuses on accelerated ageing tests to study the degradation behaviour and physical properties of degradation indicators. For this purpose, 30 samples of AECs are divided into five groups of six samples each and subjected to five different voltage levels under accelerated thermal stress of 10°C above the rated value. Based on the analysis of physical properties of AECs before and after ageing, normalisation methods are developed to account for frequency and temperature variations. The study of the impact of the ageing process on physical properties is one of the crucial contributions of this thesis as it increases the failure detection accuracy by appropriate normalisation techniques. The final phase of research aims to develop methods to estimate the remaining useful life (RUL) of AECs. The results from accelerated ageing tests are used to create the polynomial and exponential degradation models to estimate the remaining useful life based on the failure threshold for degradation indicators. RUL estimation methods are designed using the least-squares regression and Kalman filter techniques to determine the remaining useful life. A comparison of these methods in terms of accuracy and complexity is made to highlight the suitability of these techniques for different applications.