Interactive Buckling Tests on Cylindrical Shells Subjected to Axial Compression and External Pressure—A Comparison of Experiment, Theory and Various Codes

The buckling of welded steel cylindrical shells under the combined action of external pressure and axial compressive loads is of considerable interest to the offshore oil and nuclear industries. However, test results on this subject are scarce and some design rules which have been proposed recently have not been validated experimentally, especially in the plastic buckling region. In order to check these rules, and suggest others, interactive buckling tests were conducted at Liverpool University on cylindrical shells having R/t ≍ 100. One series of tests consisted of 19 machined and stress-relieved steel models having L/R ratios of 0.33, 0.74 and 1.45. The results obtained on these near-perfect machined models were compared with theoretical predictions of the behaviour of perfect cylindrical shells and the agreement between the two was good The other series consisted of 21 welded steel models and had geometric ratios which were similar to the machined ones. The linear interaction equation Sp + Sx = 1 was used to predict the failure loads of these welded steel models and the predictions were safe in all cases. However, for some combined loading cases the linear equation was rather conservative and, in consequence, some non-linear interaction equations were investigated. These seem promising for design purposes. Irrespective of whether a linear or a non-linear equation is chosen for design, more tests will be needed to establish the scatter bands of the interactive buckling curves for various values of R/t. Some tests were also carried out on (a) the effect of the loading path on the failure loads and (b) models with localized dents. Other topics discussed in the paper are: the effects of residual stresses and initial geometric imperfections, the general procedure adopted by Codes to predict buckling loads and some discrepancies between the predictions of various Codes.