Engineering & Technologyarticle2026-10-07

Reliability assessment of parallel-connected 3L-NPC machine-side converters in offshore wind turbines considering post-fault load redistribution

Open access0 citations

Abstract

Parallel-connected machine-side converters provide redundancy for offshore direct-drive wind turbines, but a local submodule failure increases the current stress and accelerates degradation of the surviving units. Conventional fixed-load k -out-of- n reliability models do not capture this state-dependent effect. This paper develops a mission-profile-based reliability framework for parallel-connected three-level neutral-point-clamped (3L-NPC) machine-side converters by coupling post-fault load redistribution, electrothermal degradation, and system-level reliability. Chronological annual wind-resource profiles are mapped to generator frequency and submodule current. Losses and junction temperatures of all five semiconductor positions are evaluated for each healthy-module state, and thermal-fatigue damage from mission-profile and generator fundamental-frequency cycles is accumulated. Device lifetimes are combined through a competing-risk submodule model and used in sequential Monte Carlo simulation. For a 10 MW direct-drive permanent-magnet synchronous generator system, the outer freewheeling diode is the lifetime-limiting device in rectifier operation, and fundamental-frequency cycling dominates its annual damage. The submodule lifetime varies approximately with the number of healthy submodules to the power of 6.4. Neglecting post-fault load redistribution overestimates the 20-year reliability by 53.7 percentage points for eight initial submodules and by 1.1 percentage points for nine. Under the baseline lifetime-model parameters and ideal current sharing, nine initial submodules satisfy the 20-year reliability target of 0.8 for all three investigated mission profiles, whereas the 10% current-imbalance case does not. The framework provides a quantitative basis for redundancy sizing while accounting for state-dependent survivor degradation.

// Source

View paper (DOI)Open access versionOpenAlexElectric Power Systems ResearchPublished 2026-10-07

Authors: Yilun Li, Yanbo Che

Institutions: Tianjin University