Probabilistic deformation and strength prediction for a filament wound pressure vessel

• Published in: Composites. Part B, Engineering Vol. 34; no. 5; pp. 481 - 497
• Main Authors: Hwang, Tae-Kyung, Hong, Chang-Sun, Kim, Chun-Gon
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 01-01-2003; Elsevier
• Subjects: Applied sciences; Burst pressure; C. Statistical methods; E. Filament winding; Exact sciences and technology; Forms of application and semi-finished materials; Laminates; Polymer industry, paints, wood; Random variables; Technology of polymers; Random variables; E. Filament winding; C. Statistical methods; Burst pressure; Probabilistic approach; Laminate; Mechanical model; Fiber reinforced material; Mechanical properties; Epoxy resin; Theoretical study; Pressure vessel; Filament winding; Fracture criterion; Fabrication property relation; Modeling; Composite material; Shell theory; Statistical method; Bursting strength; Numerical simulation; Glass fiber; Comparative study
• ISSN: 1359-8368; 1879-1069
• DOI: 10.1016/S1359-8368(03)00021-0

In this study; probabilistic analysis and experimental tests are conducted to predict the probabilistic deformation and strength of composite pressure vessels subjected to inner pressure loading. A computer code was utilized for the progressive failure analysis and probabilistic strength analysis with the Edgeworth expansion method and the Monte-Carlo method. The effects of design random variables on the structural behavior of the composite pressure vessel were quantified by probabilistic analysis. Three design random variables with uncertainties were selected: (1) composite elastic constants; (2) lamina strength; and (3) the thickness of the helical and hoop layer. The accuracy of probabilistic analysis was verified by comparing the predicted results with the experimental hoop strain or with the burst pressure obtained by a hydro-test with ten composite pressure vessels. The improvement of the structural design based on the probabilistic analysis is also discussed. The sensitivities of each response to the design random variables were evaluated and conclusions are drawn concerning the relative importance of these variables in the design problem.