An overview of current design practices for submarine pressure hulls is presented, along with the results of a survey of the literature that was conducted to determine standard nonlinear numerical modelling practices for those structures. The accuracies of the conventional submarine design formulae (SDF) and nonlinear numerical analyses for predicting pressure hull collapse are estimated by comparing predicted and experimental collapse loads from the literature. The conventional SDF are found to be accurate within approximately 20%, with 95% confidence, for intact pressure hulls. The accuracy of a wide range of nonlinear numerical methods, including axisymmetric finite difference and general shell finite element (FE) models, is found to be within approximately 16% with 95% confidence. The accuracy is found to be within 9% when only higher fidelity general shell FE models are considered. It is shown how the observations taken from the survey could serve as a starting point for establishing modelling guidelines, quantifying the accuracy of nonlinear FE analysis in pressure hull collapse calculations, and introducing this method into a design procedure by way of a partial safety factor.@en

This paper presents experimental investigations into the mechanics of collapse for externally loaded pressure vessels with simulated corrosion damage.@en

A submarine may have to operate for a period of time with local corrosion damage in the pressure hull if a suitable repair method is unavailable or too expensive for implementation. This paper describes collapse tests on twenty ring-stiffened aluminium cylinders, which were conducted to study the effect of corrosion damage on hull strength and stability. Artificial hull thinning was found to reduce the collapse strength of experimental models through high local stresses in the corroded region. Leading to early onset of yielding and inelastic buckling. Bending associated with the eccentricity due to one-sided thinning was found to further increase the local stresses in the hull. Overall collapse pressures were more severely affected by corrosion damage than interframe collapse pressures. The percentage reduction in overall collapse pressure, compared with intact experimental models, was found to be closely related to the percentage depth of thinning. The accuracy of conventional collapse pressure predictions for the experimental models was significantly better for intact than for corroded cylinders.@en

The effect of corrosion damage on overall collapse strength of submarine pressure hulls was studied experimentally. Ring-stiffened cylinders were machined from aluminium tubing and loaded to collapse under external pressure. In selected specimens, some of the outer shell material was machined away in large single patches, representing general corrosion. Other specimens had many smaller patches, representing corrosion pitting from the outside of the hull, followed by grinding. Large-amplitude out-of-circularity (OOC) was introduced by mechanical deforming selected cylinders. Clusters of artificial corrosion pits were found to have approximately the same effect on collapse pressure as equal-depth general corrosion covering the same region of plating. General corrosion was found to be most severe when it was "in-phase" with OOC, since, during pressure loading, high compressive stresses resulting from corrosion were compounded by compressive bending stresses associated with OOC, and furthermore, the corrosion tended to increase the geometric imperfection itself. On the other hand, out-of-phase corrosion reduced the effect of OOC, while at the same time, the thinning associated compressive stresses were counteracted by local tensile bending stresses associated with OOC, so that strength reductions were correspondingly smaller. Overall collapse pressures for corroded specimens were reduced by, on average, 0.85% for each 1% of shell thinning. That result is based on a linear approximation of the nonlinear relationship between thinning and collapse pressure. The linear trend-line, which was used to account for the experimental scatter, is based on specimens with 13 to 27% shell thinning, and with a variety of corrosion areas and OOC amplitudes.@en