Subsea tidal current energy converters (TCECs) face distinctive reliability challenges due to limited accessibility, high operational and maintenance costs, and highly variable operating loads that can undermine conventional deterministic design margins. This paper presents a Probabilistic Physics-of-Failure (PPoF) methodology for reliability assessment of an infinitely variable transmission (IVT) in TCEC applications to support risk-informed design and operational planning under high-torque and low-speed conditions. The PPoF methodology combines failure mode and effect analysis-driven screening to identify dominant failure modes and mechanisms with an environment-to-stress mapping that propagates lab-based testing tidal current velocity through turbine dynamics and drivetrain kinematics to obtain gear-tooth contact forces and tooth contact stress distributions, including time- and angle-dependent effects relevant to kinematically modulated gear pairs. American Gear Manufacturers Association criteria are applied to evaluate tooth root bending strength and contact fatigue strength, enabling a probabilistic stress–strength assessment of failure probability across operating and environmental conditions.
The main contribution is the coupling of tidal-current-induced load modeling with standards-based mechanical strength analysis within a PPoF methodology , enabling uncertainty characterization and propagation with sparse test or field evidence. Bayesian updating is used to fuse model predictions with available experimental and condition-monitoring information, yielding posterior distributions for key uncertainties such as loading conditions. The sensitivity study for PPoF outputs can identify the dominant risk contributors and inform the improved design or maintenance planning. While PPoF methodology is used to demonstrate the IVTs in TCECs, it can be generalizable to other marine energy drivetrains where harsh environments, limited access, and operational variability challenge traditional reliability methods