Oil Country Tubular Goods Fatigue Testing: Do We Test Them Enough?

Abstract

Fatigue is the most common known problem of drill pipes, since the combination of make-ups performed to connect the pipes and all the external loads, together with the threaded geometry of the connections, will stimulate the appearance of high stress points, cracks and finally promoting considerable economic losses. When threaded connections are used to connect the casing string, the fatigue resistance of the connection will affect the whole integrity of the string, and thus, in most cases, it is lower as the casing body. Generally, fatigue is classified as low-cycle fatigue and multi- or high-cycle fatigue. For Oil Country Tubular Goods (OCTG), a typical high cycle fatigue is represented by drill pipe fatigue in deviated wells. Unlike drill pipe, the casing may be exposed both to low-cycle as well as to high-cycle fatigue. Low-cycle fatigue is a common type of failure when the applied loads induce high stresses in the metallic material. The number of cycles may vary from as low as 10 up to 100. High-cycle fatigue requires a large number of cycles to failure. In order to avoid catastrophic failures, high-cycle fatigue resistance is usually considered to be sufficient if the number of cycles is above 106. The oil business has focused excessively on testing drilling risers and drill pipes under fatigue loads, but when it comes to casing and tubing the experimental approach may require different solutions. Drilling with casing opened the intensive testing of casing connections against fatigue resistance. Moreover, recent papers have shown intensive work on redesigning connections to withstand fatigue. New applications like rotating while running require a rethinking of testing strategy of Casing and Tubing. The following paper focuses on answering the question whether we test enough. The first part compares existing testing facilities, followed by an intensive discussion about the true loads of a casing or tubing connection. Using public testing data, the second part of the paper tries to identify how far the results provided by various types of testing machines can be compared with each other. For example, we found that low cycle fatigue results may not fully reflect the predictions based on extrapolations of high cycle fatigue results.