Theoretical and experimental modes of behaviour of cylindrical model prestressed concrete pressure vessels when pressurised to failure hydraulically and pneumatically

Abstract

If hydraulic pressure in a prestressed concrete vessel is increased until structural failure occurs, and the deflections are measured, it is found that there are three phases of behaviour. The first is a linear elastic phase, which embraces all the normal operating and test loadings: this phase can be thoroughly understood by theoretical analysis for the short term, and probably also for the long term. In the second phase cracks gradually spread through the concrete to produce a very non-linear, but still mainly elastic, response. The third phase involves large deflections with small increases of pressure, and the structure functions as a mechanism. Some knowledge of overload behaviour is essential to confirm that a proper margin of safety exists for normal loads, and this has generally involved an examination of the ultimate load factor, which is the termination of phase III. A method of calculating the pressure-deflection curve in phase III is explained, and experimental justification is given. In this work the liner is assumed to remain intact up to the ultimate pressure. If the vessel is pressurised pneumatically, when the liner fails the gas pressure will also act on the walls of the cracks in the concrete. The theoretical consequences of these extra loads are discussed both statically and dynamically. The possible effects of hot gas on the structure are mentioned. From these results it is suggested that the ultimate load factor is too far from reality to provide a really economic solution for the provision of adequate integrity, and it is suggested that a better criterion could be found in the phase II behaviour. The paper then discusses some of the basic problems in making a theoretical analysis of the structure in phase II, where major cracks are gradually spreading as the pressure increases. A relatively novel method of determining the extent of a stable crack is described, and the importance of shear transfer across the cracks is explained. These lead to a suggested application of simplified dynamic relaxation methods which should provide reasonably accurate results sults at a much reduced cost. Experimental checks on the theory are described.