Fatigue life prediction of threaded pipe connection

Abstract

In the oil and gas industry, threaded pipe connection is frequently used to connect the casing string, drill pipe strings or production and transportation risers and pipelines. The connection is normally preloaded in order to maintain a sealed and secure connection while in service and avoid leakage. Tapered thread are a common connection and in order to introduce preload to the threaded connection when they are assembled a certain make-up torque is going to be applied. The make-up torque plus external loads result in a multiaxial stress distribution over the connection, where the threaded connections act as stress risers. Environment such as waves and currents cause dynamic loads acting on the pipe line and offshore structures. The weakest point in offshore structure is the pipe connection because of fatigue crack initiated in the connection’s threads. Researchers and engineers developed a variety of patented threaded pipe connection which all claiming to improve a connection’s fatigue life. The experimental data for patented designs, available in literature, is limited. Most published studies usually comprise experiments on a single connection type. For detailed fatigue analysis those published studies cannot be used since there is no uniformity in testing setup, loading conditions and damage detection technique exist. Moreover, current design curves in codes and standards lead to overly conservative or inaccurate results. The aim of this work is to provide a better understanding of the fatigue mechanisms of threaded pipe connections and to study the effect of different design features on a connection’s fatigue life. The final goal is to formulate guidelines for new fatigue resistant connection designs. API connection is used as a reference in this study. Several modifications and design features are applied to the connection type. To simulate the effect of these modifications, a parametric 2D axisymmetric finite element model, ABAQUS is used. 2D finite element result are compared with a 3D model to prove its validity for both make-up. In addition, the results of the 2D axisymmetric simulation are validated by static strain gauge measurements during a make-up test and an axial tension test. The validated model is then used to evaluated the influence of the connection properties and design features on the threaded connection’s behaviour. Test rigs were designed to perform axial fatigue experiment on two scales: the small-scale experiments on 1" (33.4 mm outer diameter) connections are performed in axial fatigue testing, the medium scale tests on 4.5" (114.3 mm) connections are carried out under axial tension for which a setup is developed. The majority of the performed fatigue tests are small scale experiments. Several modified configurations are tested. The S-N curve is constructed, so that the effect of certain configuration on the connection’s fatigue life can be quantified. The local modification of the threaded connection’s geometry as well as the connection’s contact condition’s contact conditions can have an important influence on the fatigue life of the connection. A beach marking technique is used to visualized the crack fronts at different moments during the tests so that exact crack shape can be seen during post-mortem analysis. The result shown that a crack initiates at the root of the last engaged thread of the male part of the connection, and propagates slowly over a large segment of the circumference, forming a long shallow crack. When the crack penetrates the pipe wall, it rapidly increases in size along two crack fronts. The shape of crack observed in beach mark analysis do not have a semi-elliptical shape as commonly used in fracture mechanics. A fatigue crack growth analysis that considers the crack as an annular flaw, is effective in describing the crack growth behaviour. The experimentally obtained S-N curves and the result from the finite element simulations are combined in multiaxial damage evolution law. The observed trend in fatigue lives of the configuration are explained by using the fatigue analysis. Using a connection’s thread load distribution as a measure for its fatigue life is proven to be inaccurate. The main reason for this is that the load distribution is related to axial stresses over the connection. The fatigue life of a threaded connection is determined by the local multiaxial stress distribution and strain range around the root of the last engaged thread. These local conditions are not only the result of the load distribution, but they are also affected by the hoop stress introduced during make-up, which can additionally be affected by a changed connection stiffness. The multiaxial damage evolution law is used to analyse the influence of several features on a connection’s fatigue life. It is not for all patented modifications that an increased fatigue life is predicted when applied to the API connection. The final conclusion reached is that, in order to optimize a fatigue resistant connection, several design features must be combined together. The thread shape can be optimized to obtained a low stress concentration factor and reduce the local strains at the thread root. The connection’s global geometry and make-up conditions can be optimized to improve the load distribution over the threads and reduce local stresses and strains at the threads.