Health monitoring, lifetimes estimation, and thermal management of IGBT-based power converters

• Main Author: Mohamed Halick Mohamed Sathik
• Other Authors: Josep Pou
• Format: Theses and Dissertations
• Language: English
• Published: 2017
• Subjects: DRNTU::Engineering::Electrical and electronic engineering
• Online Access: http://hdl.handle.net/10356/72745
• Institution: Nanyang Technological University

Power electronics plays a significant role in various applications such as renewable energy, aerospace and marine drive systems to achieve efficient electric energy conversion and to attain high system performance. As power electronics systems, have gained an important status in a wide range of industrial fields, reliability and availability of these units are of paramount importance in order to meet stringent requirements spelled in aviation and industrial standards. Reliability of a power converter is mainly verbalized by the failure rate of power semiconductor device and dc-link capacitor. For instance, in applications demanding higher reliability, electrolytic capacitors are swapped with film capacitors, since they exhibit lower equivalent series resistance (ESR) with increased reliability. However, due to the fact that the power devices are exposed to electrical, thermal and mechanical stresses, the failure rates of these components are comparatively high, and it was found to be the main reason for reducing reliability of the whole system. One promising solution to address this problem is to perform active health monitoring for early diagnosis of potential failures, implement thermal management and protection techniques to increase the reliability and conduct lifetime estimation of power devices through suitable prognostic methodologies. This is of special interest in mission critical applications such as avionics, marine, high-speed rail and wind turbines, where no failure is permissible until the next scheduled maintenance period. In general, the health monitoring approaches involve monitoring of appropriate parameters that are indicative of impending failures of power semiconductor modules. Most of the available health monitoring approaches are limited to only one failure mechanism and they require detailed device material properties, packaging factor, sensing elements with high bandwidth, and complex filtering circuits. Furthermore, a high-fidelity model is computationally expensive and often considered as impractical. To alleviate these problems, this thesis proposes a novel real-time health monitoring approach to enhance the power device reliability and efficiency of the power converter systems. The proposed health monitoring method is based on monitoring of potential electrical precursor parameters that are related to device ageing. This thesis proposes and discusses an online measurement technique for health monitoring of power modules in power converter systems. Lastly, a three-phase inverter system is developed for experiments to verify the proposed method in real time operating conditions. The inverter system consists of one IGBT power module aged through accelerated power cycling and two new IGBT power modules to verify the proposed online measurement method in real time operating conditions. It has to be mentioned that the proposed online condition monitoring approach is more suitable for custom made power converter systems where external measurement and filtering circuits can be integrated into the driver system, and cannot be used for off-the-shelf power converter systems. This thesis also proposes an online condition monitoring approach for off-the-shelf power converter systems. Ageing of IGBT modules and its effect on on-state resistance is investigated using accelerated power cycling in order to trigger solder die-attach degradation and bond wire lift-off damage. Experimental investigation verifies that the increase in on-state resistance is a reflection of device degradation. Therefore, an online computational model of on-state resistance is developed for real-time condition monitoring application to detect device incipient failure in off-the-shelf power converter systems. The accuracy of the proposed computation method is verified through a comparison with measured on-state resistance for new and aged power devices. In addition to the identification of device ageing and incipient failure, this thesis also studies lifetime estimation technique for three-phase power converter systems based on physics of failure approach. Normally, mean time to failure (MTTF) is used to check the reliability of power electronic components. However, most of the power electronic components failure distribution follows either Weibull or Lognormal, where the failure rate changes with respect to time. Hence, MTTF is not adequate for useful lifetime estimation in critical applications. Therefore, for mission critical applications, the failure rate function needs to be estimated according to Federal aviation administration (FAA) requirement. Thus, this thesis proposes a lifetime estimation model that includes both the MTTF and failure rate model to estimate corresponding useful lifetime of power converter systems. The proposed method is implemented using Monte Carlo simulation method, where the system lifetime is estimated according to the weakest component in the system. Based on the proposed method, a simulation case study is conducted and presented for a three-phase power converter system. A novel thermal controller of power semiconductor switches to minimize the thermal stress is proposed in this thesis. The developed control method developed in this work not only limits the over temperature, but also controls temperature swings (∆Tj) due to thermal cycling. The proposed thermal controller uses the temperature-related parameters to control the junction temperature online. The proposed junction temperature control technique is based on the availability of the junction temperature of an IGBT switching device. Therefore, this thesis proposes a junction-temperature estimation model based on power loss and transient thermal resistance characteristics for a three-phase power converter systems. The accuracy of the estimated junction temperature using the presented model is compared with the measured junction temperature (T_jm ) under real-time operating conditions for dynamic load profile conditions. A three-phase inverter prototype is developed to validate the proposed thermal control techniques. Experimental results demonstrate that demonstrate that the proposed thermal controller for power semiconductors devices effectively controls the junction temperature variation in real time operating condition. The proposed health monitoring and thermal control techniques prevent the power devices only from wear-out failure mechanisms. However, the power devices used in power converters also fail due to over-current or short-circuit current caused by external factors such as power supply transients, mechanical overload, load transients, etc. Some of these incidents can result in very high current (a few times higher than the system’s rated current) flowing through the electrical drive system. Usually, electrical machines have the capability to withstand this very high current (milliseconds to seconds depending on the size of the machine). However, power device withstanding time of short-circuit current is relatively small, in the order of microseconds (approximately 10µS) in IGBTs. Thus, a novel short-circuit detection methodology is proposed. It is based on sensing the collector-emitter voltage sensing using an external collector capacitor, and a protection circuit based on gate voltage clamping using current diode. This methodology protects a power devices from over current and short circuit current, and isolates the fault current from the rest of the system. The proposed short-circuit detection and protection circuit is compared with conventional gate-voltage clamping method, and results show that the proposed method aids in the soft shutdown of the IGBT, and considerably reduces the fault current and power dissipations compared to the conventional method.