On the Structural Design of Tubular Joints

Abstract

ABSTRACT Stresses in thin-walled tubular joints can be calculated with finite-element computer programs that employ 'thin-shell' elements. Such a program, designated SATE, has been developed specifically with the object of determining stress distributions in complex tubular joints as generally encountered in the offshore industry. The program has been applied t()_ several joint geometries and good agreement has been observed between experimental and calculated results. The program thus provides insight into the structural design of tubular joints. Using the SATE program, the influence of changes in design of both unstuffy and stiffened tubular joints on the stress-concentration factor has been studied. In this context the current API recommendations for simple tubular joint design are discussed. Calculations have demonstrated that in the design of large tubular joints allowance should be made for membrane action rather than for bending of the chord and brace walls. Such an approach leads to joints of low stress concentration. 1. INTRODUCTION The stress analysis problem of tubular joints in offshore structures has been studied extensively, both experimentally and analytically over the past ten- yearsl. Owing to the complexity of this problem it is still far from being solved. The stress concentration of a joint is a measure of its strength in particular its fatigue strength. Here, the application of the finite-element (f. e.) method for calculating the stress concentration in such linear elastic, thin-walled structures is becoming more and more common practice. Under the assumptions of the thin-shell theory2, the f. e. method indeed suits the purpose of tackling, at least in principle, the wide range of problems related to the connections between flat plates, cylinders and other structural components. The SATE (§structural Analysis using Ihin-shel finite Elements) program has been developed specifically for determining stresses in complex tubular joints as generally encountered in the offshore industry. The program employs an element as described in ref. 3 following ideas of Herrmann4. In comparison with the Clough-element5, the incorporation of this element in a program is more complicated, but with the same number of unknowns more accurate stresses can be calculated. In the development of the SATE-program a considerable effort has been spent on making the program flexible and economical for both unsatisfied and stiffened tubular joints. The cost involved in solving a tubular joint problem are made up of the combination of input preparation and running of the computer. It would seem that high flexibility and minimum of input preparation are conflicting criteria, as long as the problem of writing a most efficient, general-purpose mesh generator has not been solved. Grete?s progratn6 is a first step in this direction. Some aspects of the SATE program are dealt with in section 2.