Probabilistic finite element methodology for the reliability assessment of stiffened plates

The present study reports a probabilistic finite element modeling and analysis procedure that is useful for the assessment of structural integrity and reliability. This procedure is efficient, accurate, and simultaneously accounts for the variabilities and uncertainties in the governing parameters as well as the nonlinear nature of the physical failure mechanisms. In the approach, nonlinear finite element methods are employed to model the physical response characteristics while an integrated stochastic finite element formulation that employs a FORM-based adaptive response strategy is used for probability calculations. Uncertainties in loads, material properties and structural properties are considered, and reliability levels for selected stiffened-panel configurations, with reference to some strain-based failure criteria, are determined. Probabilistic sensitivity studies are also conducted to study the importance of the influencing parameters as well as the uncertainty in these parameters. Results are compared to those obtained using approximate analytical models, and it is shown that the simplified models may predict reliability and sensitivity values that have significant errors. Finally, the paper suggests a strategy by which this methodology can be employed to improve the design of stiffened-plated structures.