Type 2 복합재료 압력용기의 피로손상시 발생하는 음향방출

Abstract

In this study, evaluation of damages on type 2 composite pressure vessel using acoustic emission, including the specimen destruction test, pressure vessel signal transmission test, burst and fatigue comparative test according to the destruction and evaluation of damages on the pressure vessel through the probabilistic neural network. The Amplitude of acoustic emission signal gets bigger as the cutting angle of knife increases. Accordingly, the number of hits in destruction of composite materials specimen have more in longitudinal direction (longitudinal direction to the fiber glass) than in hoop direction (horizontal direction to the fiber glass) while the amplitude of signals were bigger in hoop direction than longitudinal direction. Glass fiber It was found out that the amplitude where the glass fiber breakage is 40dB or more and that the amplitude of signal for matrix crack was 40dB or less. It was also found out that the source of acoustic emission signal during the destruction of specimen in the longitudinal direction is the glass fiber or main damage mechanism. The transmission speed of acoustic emission signal of type 2 composite pressure vessel according to the transmission angle showed little difference between when the water is filled or not. As the transmission angle increases from acoustic emission source relative to the direction when the glass fiber is wound (0°), the speed is reduced from 5700 m/sec to 4500 m/sec and the damping effects get increases, showing the anisotropy. The fracture mechanism of type 2 composite pressure vessel is follows; matrix cracking – delamination – glass fiber breakage – metal liner fracture. The Kaiser effect is observed when the pressure goes up to 60% of the bursting pressure. In more than 70%, the felicity effect is observed and the creep effect is found as the vessel is so much damaged. In addition, In 70% of bursting pressure, the matrix delamination and glass fiber breakage are found to be the key causes of damage. Just prior to the destruction of the vessel, the mass breakage of glass fiber bundle in the vessel and the rupture of metal liner occur. The analysis of acoustic emission signal occurring when type 2 composite pressure vessel is holding the load shows that the total count and the signal strength are the AE factors, which can represent the damage of vessel and that the signal variables, such as mean amplitude, rise time and duration and the AE signal, which is 60 dB or more, are the sound transmission signal variables, which can estimate the damage mechanism of the vessel. The pressure for destruction of type 2 composite pressure vessel which have the artificial defects, is not related to the defect of matrix but the direction of defect is related to the position of final leakage in the vessel. In the case of longitudinal direction, the thickness of wall of the wall around the vessel, where the defect is located, gets thin and thus become weak. So, it is likely that the leakage finally occurs where the crack appears on the metal liner though the acoustic emission source is located. The acoustic emission signal occurring in the fatigue test of type 2 composite pressure vessel does not show any increase or decrease according to the number of accumulated fatigues of the vessel. But it shows the increase or decreases after the specific signal variables such as amplitude, rise time and count decreases or increases when the new damage source is found. There are four patterns in the acoustic emission parameters according to the increase in the number of fatigues; pattern of gradual rise up to a certain level before going down, pattern of decreasing and then increasing, pattern of gradual rise up to the leakage after the specific fatigue cycle and the pattern of reduction to a certain fatigue cycle. As type 2 composite pressure vessel does not show any simple increase or decrease, it is effective to classify the pattern of acoustic emission variables coming from the fatigue damage using the probabilistic neural network. It shows that the learning input variables for the classification of signal pattern using the probabilistic neural network has better results when the features of AE parameters are more diverse, the more acoustic emission signals are included in the loading – holding – unloading zone and the single after 2nd cycles of collected signal is applied.