ReliPE

Conversion of electric energy by power electronic converters has an increasingly important role in the various parts of the power system. Examples are integration of renewables, industrial and household loads, HVDC-transmission and electrification of the offshore oil and gas sector. Consequently, the electric energy must normally pass through several power electronic conversion stages from generation to consumer load. The importance for the society of reliable energy conversion components is therefore fast growing. Knowledge-building research is needed for better understanding the fault mechanisms and then using the achieved knowledge for improving reliability of the components, for predicting and preventing faults, and for cost reduction by better component utilization and optimized maintenance routines.

A common factor for many applications is the high reliability requirements caused by maintenance challenges and/or high cost of downtime. The complexity of power electronic converter systems calls for high R&D-efforts in order to improve component designs and to increase the ability to identify and predict faults.

Reliability and ruggedness of power electronic systems and components can be improved by: 

• Better understand of failure mechanisms and modes, and how these are connected with stress levels 
• Improving component design, packaging and cooling 
• Improving condition monitoring, including both failure detection and failure prediction

The optimal design is always a trade-off between performance, reliability and cost, and a continual R&D-focus is required for providing high-performance and cost-efficient converter systems.

The power semiconductors, today mostly IGBTs (Insulated Gate Bipolar Transistor), are the key components, and at the same time the most vulnerable components in the power converter. Due to the rapid IGBT technology development, both the failure modes and the lifetime change as new generations of components are launched.

A well known source for lifetime reduction in power semiconductors is the so-called power cycling, where fluctuations in power level lead to temperature fluctuations and then cause mechanical wear and damage. Accelerated power cycling tests have been performed for decades, and are the most common way of characterizing and comparing the lifetime of IGBTs. The most important outcome of such tests is information of the dominant failure mechanism in the device.

A weakness of the well-established power cycling schemes is the difference in stress level between the test setup and the real application. For example, the stress level is significantly higher in an accelerated test in order to obtain failures within weeks to months, compared with the desired lifetime of 15-30 years. Furthermore, a power-cycling test is performed with an idealized temperature cycle profile, while typical applications can have complex temperature variations. Although some works have studied power-cycling lifetime in light of different applications, significant knowledge gaps must be closed in order to obtain trustworthy application dependent lifetime models. Then a fundamental question remains to answer: how is it possible to extrapolate results obtained from an accelerated lifetime test to estimate the expected lifetime in the real application? Such an estimate is extremely valuable during both design and operation of power electronic converters.

In a previous research project hosted by SINTEF on Reliable and Energy Efficient Power Renewable Energy Systems – OPE (2009-2014), the main focus was on power conversion systems for offshore wind turbines. In this project valuable methodologies, tools and special test equipment were provided for assessing operational dependent power-cycling lifetime of high power IGBT components. The present project, ReliPE, is exploiting these previous results for further investigating the above-mentioned reliability questions. Especial mentioning is a custom 2000A Power Cycling Tester provided by the previous OPE project, for running experiments for investigating impact from stress factors such as operating temperate swings and switching profiles.

While the OPE project had main focus on experiments with press-pack type IGBTs, ReliPE has established a program for power cycling of IGBT modules, representing the last generation of high power components aimed for high-voltage converter topologies such as the MMC (Modular Multilevel Converter). The achieved knowledge will be used for providing tools and models for estimating remaining lifetime or time to maintenance and repair, and by that to be able to take action prior to occurring problems.

ReliPE is also continuing the fruitful cooperation with Chemnitz University of Technology that was established in the previous OPE project by having TU Chemnitz as research partner in ReliPE. Of special importance is their valuable expertise on test methodologies, power-cycling lifetime modeling and IGBT short circuit ruggedness.