| Summary: | 博士 === 國立中興大學 === 機械工程學系所 === 99 === A reliable and accurate method of the buckling prediction of thin-walled cylindrical shell under an axial load is presented first in this work. The experimental arrangement and specimens are discussed in detail, including the measurement of the geometric imperfections of the specimen’s surface using a coordinate measuring machine. Different Finite Element (FE) models, in terms of complexity, are used to simulate the experiment arrangement in an attempt to get a good agreement with the experimental buckling loads. It has been demonstrated that FE models with simplified rigid support conditions overestimate the prediction of the experimental buckling load even though these models included the effects of the measured initial geometric imperfections and load eccentricity. By contrast, FE models with realistically modeled support conditions achieved the best result.
The experimentally verified FE model was then employed to study the buckling behavior and the effect of measured initial geometric imperfections, load eccentricity size, load eccentricity position along the shell’s circumferential direction, the potential effect of the shell’s height, and the method of increasing the load-carrying capacity of the thin-walled shell. The presented work demonstrated the strong influence of the eccentric load position along the imperfect shell’s circumferential direction on the buckling of the thin-walled shell. The strongest and the weakest eccentric loading location of the structure were found and their relations to the position of the maximal radial imperfection were indicated. The maximum load-carrying capacity of the shell was obtained by moving the applied load into the strongest eccentric load location of the structure.
Finally, based on the result in this work, the knockdown factor used in the design of thin-walled shell could be increased by adding the parameter αecc=0.207. The value of αecc was defined based on the value of the minimal scatter in the buckling load between the cases under the action of optimized and weakest load for the studied specimens. This scatter in the buckling load should be investigated for more specimens with different imperfection shapes caused by different manufacturing processes, and also for different radius/thickness ratios to further optimize the value of αecc parameter.
|
|---|
