Failure Pressure Prediction of Semi Spherical GFRP Shells in Thermal Environment

Document Type: Research Paper

Authors

1 Materials and Manufacturing Technologies Complex, MUT, Lavizan, Tehran, Iran

2 Materials and Manufacturing Technologies Complex, MUT, Tehran, Iran

Abstract

The pressure tests and temperature effects on the failure of composite pressure vessels is an expensive and time-consuming process. In this research, a new process has been designed in the laboratory scale to manufacture a portion of the composite vessel as a test specimen. The com-posite specimen is manufactured in a close mold and is tested in an apparatus that is able to control gas pressure up to 100 bars and temperature up to 150 Celsius. A spherical cap of com-posite is designed as a test specimen to be pressure tested up to failure at constant temperature. A closed mold has been also designed and fabricated to manufacture the test specimens. Testing apparatus consisting of three-part holder, gas tank, pressure controller and temperature con-troller is designed and fabricated for the failure tests. To examine proper function of the fabri-cated mold and testing apparatus, twelve small spherical cap of the GFRP test specimens manu-factured by the hand layup process considering [0/90/45/-45/90/0] arangment in the fabricat-ed mold. Burn off tests has shown a closed tolerance of fiber volume fraction for all specimens. The test specimens subjected to internal pressure at constant temperature up to failure. A nu-merical analysis of the GFRP spherical cap has been also run using ABAQUS software consider-ing Hashin failure criterion to determine the failure pressure at constant temprature. Both the experimental results and finite element calculation of failure pressure have shown to be close enough to trust the mold and testing apparatus. These results also showed that the failure pres-sure decreases as the temperature increases. The numerical results for prediction of the failure pressure however have shown to be greater than experimental values and an average difference of 16.5% has shown to be all four tests.

Keywords