mohammadali rasooli

Cylindrical shells under external pressure are increasingly used in underwater vessels and their collapse is mainly due to the buckling phenomenon. Improving the resistance of these structures to buckle can be achieved by modifying the geometry of the structures. For reinforcement of cylindrical shells increasing the shell thickness and using ring stingers have been amongst the most popular methods. However, increased pressure and weight constraints of structures demand for more advanced methods. In this study, by corrugating of surfaces between the rings and combining the geometric factors those affecting the buckling, the effecting parameters on the critical buckling pressure of the structure were investigated. For this purpose, first a cylinder with a specific diameter and length with reinforcements of different dimensions and distances was evaluated. Then by corrugating with different radius of curvatures in inter-stiffener conditions the effect of these modifications on the buckling pressure and strength to weight ratio were investigated. The results of the analysis indicated that in the case of inter-stiffener buckling the increase of critical buckling pressure of corrugated shells is very significant. In other words, in this method the effectiveness of geometric modification is highly dependent on the inter-stiffener buckling.

In recent years, the conical shells under external pressure are widely used in construction of the under-water pressure hulls, covers of aero-engines and storage tanks. Strength of thin-walled shells under external pressure are usually influenced by the buckling phenomenon. So, its study is in high degree of importance. The buckling analysis of thin conical shells based on theoretical and experimental methods is accompanied by shortcomings such as time consuming and complexity. In this paper, an efficient method based on Artificial Neural Network (ANN) is presented for prediction of buckling and post-buckling behavior of conical shells. Primarily, the linear and non-linear buckling loads of the truncated cones with various thickness and stiffener dimensions are obtained by using the Finite Element (FE) analysis. Then, these obtained results are submitted to the Neural Network for training. In order to verify the solution procedure, the predicted results of ANN are compared with those of extracted from FE analysis. It is shown, that the predictive model benefits from high convergence and accuracy. Finally, some predicted results of buckling and post-buckling analysis of conical shells is figured.