Failure Prediction of Submodule Capacitors in Modular Multilevel Converter by Monitoring the Intrinsic Capacitor Voltage Fluctuations

Abstract

Modular multilevel converters (MMCs) are emerging as a promising topology for medium- and high-power applications. Aluminum electrolytic capacitors (AECs) are usually employed in MMCs as floating capacitors due to their high volumetric efficiency and low price. AECs gradually deteriorate over time due to electrolyte vaporization, and for this reason, they have been recognized as one of the most fragile components in the converter. As they continue to age, the AEC's capacitance decreases and its equivalent series resistance increases, which can result in an increase of submodule (SM) capacitor voltage ripple, power loss and could damage the operation of the MMC with prolonged use of aged capacitors. To prevent such damage, monitoring the health of capacitors is an important step to enhance the reliability of the MMC by predictive maintenance. This paper presents a failure prediction scheme for SM capacitors in the MMC by monitoring the SM capacitor voltage oscillations. Detailed simulation studies are carried out for a five-level MMC in the PLECS® platform and verified experimentally. The estimated capacitance through simulations and experiments is in close agreement to that value measured using the LCR meter.