Benchmarking of capacitor power loss calculation methods for wear-out failure prediction in PV inverters

Abstract

Aluminium electrolytic capacitors (e-caps) are among the components most prone to failure in power electronic systems. Thermal stress is a critical factor which affects the lifetime of dc-link capacitors. Therefore, the estimation of power loss and temperature are important steps in lifetime prediction of these components. This paper aims to carry out a qualitative benchmarking of different capacitor loss estimation methods, focusing on the component level reliability. Four different equivalent series resistance (ESR) models are benchmarked for 2 different mission profiles based on complexity, estimated losses, thermal stresses, and B1 lifetime: 1) constant ESR, 2) temperature dependent ESR, 3) frequency dependent ESR, and 4) frequency/temperature dependent ESR. Capacitors in the dc-link of a PV inverter are evaluated using an offline look-up table (LUT) approach, considering two different mission profiles (MP), and following a normal distribution parametric variation. For the lifetime model and parameters employed, the results for this case study indicated a maximum difference of 7.1% in the B1 lifetime estimated by approaches 1 and 4.