Modeling the biaxial, rate-dependent response of filament-wound FRP tubes

This work studies the rate-dependent mechanical behavior of filament-wound fiber-reinforced polymer (FRP) composite pipes. Commercially available tubes with a filament winding angle ±55° were tested under cyclic axial compression for four loading rates. Stress relaxation constant strain was observed as well dependence stress on rate. A novel modeling methodology is presented to capture nonlinear response, including viscoelastic epoxy matrix and interaction hoop strains. accomplished by defining an element configuration separate elements glass fibers. The incorporated using generalized Maxwell model. machine learning (ML) calibration framework adapted this study used calibrate properties analytical model convolutional neural network (CNN). CNN trained identify understand interdependencies among parameters. calibrated parameters are simulate experimentally measured response FRP found be applicable across range proposed accurately predicted time histories during accuracy capturing stress-strain responses demonstrated synthetic dataset adequate training without requiring additional experimental data.