Reliability Analysis of TPMS Structures Considering Uncertainties in Operating Conditions

ABSTRACT Triply periodic minimal surface (TPMS) porous structures show considerable promise for lightweight load‐bearing implants. To address the fatigue reliability of radially graded gyroid‐type TPMS Ti‐6Al‐4V scaffolds under unavoidable uncertainties in loading and material properties, this study proposes an efficient reliability analysis framework by integrating finite element (FE) analysis with an active learning–based Gaussian process regression (GPR‐AL) surrogate model. Firstly, five bone‐mimetic cylindrical scaffolds with porosities ranging from 60% to 80% are designed and analyzed under displacement‐controlled compression. To improve mesh robustness, the 95th percentile of the von Mises stress field is employed as the high‐stress indicator. Subsequently, the GPR‐AL method is constructed by adopting the RMSE as an error indicator, with new training points iteratively selected from high‐uncertainty regions via the maximin distance criterion. Moreover, uncertainties are propagated through the GPR‐AL model coupled with Monte Carlo simulations, and fatigue life of TPMS is then evaluated using the Basquin model. The results demonstrate that the proposed method provides a computationally efficient tool for reliability‐oriented design and porosity selection for load‐bearing TPMS implants under operating uncertainties.