Health Monitoring of IGBTs in Automotive Power Converter Systems

UoM administered thesis: Phd

  • Authors:
  • Peter Andrew James


The use of IGBT power modules in the automotive industry is becoming increasingly common as manufacturers develop more hybrid and all electric vehicles. In an industry such as this, the reliability of a component is critical and vehicle manufacturers have conducted much research into diagnostic and prognostic systems for internal combustion engines that run in real time on the vehicle to determine when components will fail. Power electronic components do not have similar prognostics available. The traditional use of power electronic modules has been in applications where their life or duty cycle is well defined, and accelerated life tests are carried out to determine a mean time to failure. This type of prognostics is not appropriate for the automotive industry because the operating cycle of the vehicle varies greatly, both in driving style, duty cycle and environment. A new type of prognostics is therefore required which will calculate the life remaining in the power module in real time as the device is being used.Because of the high robustness of IGBT power modules, testing for time to failure can be a very lengthy process. A novel procedure and test rig based on Peltier effect thermoelectric coolers was developed, which can automatically temperature cycle IGBT power modules in a very short time and determine their life expectancy, all within their operating specifications. This was tested using several power modules. The failure modes of IGBT power modules are also investigated with a view to developing a failure prediction algorithm. The causes of failure are analysed and a prognostics algorithm is proposed. This prognostics algorithm uses thermal cycle history as a means to predict the life consumed for the power module. The data obtained by the accelerated life tests is used to calculate the coefficients for the prognostic algorithm. A simulation of a vehicle drive cycle is used to show how the prognostics algorithm can be used, and a value indicating the extent to which the IGBT power module has aged is calculated. It is also proved that by intelligently controlling the heat flowing from the heat sink on which the power module is mounted, the life of the IGBT power module can be increased by approximately three times.Hardware and software were developed to implement the health monitoring algorithm. Measurement and control circuits were designed, built and tested together with software that processes the input data, records the thermal cycle history of the IGBT power modules and calculates a value of age for the IGBT power modules in real time. This was tested on several modules to prove the validity of the algorithm.The new algorithms and methodology developed could enable vehicle manufacturers to predict the failure of power modules in hybrid and all electric vehicles. This technology could also benefit other industries such as the renewables (eg wind turbines) and aerospace, where the industry is moving towards all electric aircraft.


Original languageEnglish
Awarding Institution
Award date1 Aug 2013