Abstract
The growing demand for eco-friendly liquefied gas carriers has made it increasingly important to ensure the structural safety and economic viability of membrane-type CCS (Cargo Containment System). Mastic, applied between insulation panels and the inner hull, is a critical element that transmits sloshing impact loads and absorbs hull deformation; however, the same mastic pattern based on empirical criteria has conventionally been applied regardless of panel configuration or cargo conditions.
In this study, a systematic mastic pattern optimization methodology is proposed to minimize mastic usage while ensuring structural safety. In particular, to overcome the limitations of nonlinear structural analysis, which requires substantial computational cost, dynamic sloshing loads are converted into equivalent static pressures to determine the design loads efficiently. On this basis, multiple failure modes of the CCS — including crushing of R-PUF (Reinforced Polyurethane Foam) and failure of the bottom plywood — are formulated as structural constraints. An ensemble surrogate model, which aggregates multiple surrogate models to improve predictive robustness, is constructed from analysis data generated via LHS (Latin Hypercube Sampling), and a DE (Differential Evolution) algorithm is employed for the mastic pattern optimization, minimizing the mastic area under the multiple nonlinear constraints.
To demonstrate the proposed methodology, the flat insulation panels of the GTT Mark III Flex system in a 174K LNG carrier are selected as a representative application. The results confirmed that the proposed methodology can achieve up to a 25.0% improvement in structural response and a 5.2% reduction in mastic usage
compared to the conventional design, demonstrating its potential for extension to other panel geometries and cargo types.
In this study, a systematic mastic pattern optimization methodology is proposed to minimize mastic usage while ensuring structural safety. In particular, to overcome the limitations of nonlinear structural analysis, which requires substantial computational cost, dynamic sloshing loads are converted into equivalent static pressures to determine the design loads efficiently. On this basis, multiple failure modes of the CCS — including crushing of R-PUF (Reinforced Polyurethane Foam) and failure of the bottom plywood — are formulated as structural constraints. An ensemble surrogate model, which aggregates multiple surrogate models to improve predictive robustness, is constructed from analysis data generated via LHS (Latin Hypercube Sampling), and a DE (Differential Evolution) algorithm is employed for the mastic pattern optimization, minimizing the mastic area under the multiple nonlinear constraints.
To demonstrate the proposed methodology, the flat insulation panels of the GTT Mark III Flex system in a 174K LNG carrier are selected as a representative application. The results confirmed that the proposed methodology can achieve up to a 25.0% improvement in structural response and a 5.2% reduction in mastic usage
compared to the conventional design, demonstrating its potential for extension to other panel geometries and cargo types.
