Pipelines/flowlines/risers

Fatigue Testing of Shrink-Fit Couplings for Joining High-Strength-Steel Riser Pipe

A program was devised to test the manufacturability of shrink-fit joints made from high-strength (greater than 100 ksi) mill pipe and forged couplings.

The shrink-fit connection is a method of joining pipes to couplings or flanges by thermally expanding the coupling or flange over the pipe; sealing and static capacity are achieved by the contact pressure and friction between the components once they have cooled. This method enables the manufacture of high-pressure deepwater riser systems when traditional pipe-join methods are not feasible. A program was devised to test the manufacturability of shrink-fit joints made from high-strength (greater than 100 ksi) mill pipe and forged couplings.

Introduction

The objective of the project was to qualify the shrink-fit technology to project-ready status—any further testing on the connector would be for project qualification rather than further development. This development project involved two stages: (1) proving the feasibility of machining and assembly from seamless pipe and (2) hydrostatic and resonant fatigue testing of shrink-fit connections.

Six specimens were manufactured to demonstrate that pressure containment is maintained for maximum working pressures and hydrotest pressures, while target fatigue life is achievable without loss of pressure integrity in the connection. Before testing, a finite-element model was used to address residual shrink-fit stresses, safe hydrotest pressure, predicted connection-fatigue response, and the choice of internal pressure for fatigue testing.

Current Technology

The use of pipe for offshore oil and gas risers in various types and sizes requires intermediate connections because continuously cast pipe is not used for large-diameter offshore risers, and the pipe tends to come in limited lengths. For all drilling risers, the riser is usually required to be deployed and retrieved on a regular basis, so a coupling is employed on the joint extremities, which are designed for repeated assembly/disassembly. For shallow-water drilling risers, joint lengths of 40 ft are commonplace, but, for deepwater, larger joint lengths of up to 90 ft are increasingly desirable to reduce riser running time (and thus reduce drilling costs). These larger joints usually require one or more midjoint pipe connections.

The shrink-fit connection has already been delivered for a full 19.25-in.-bore drilling-riser project for the North Sea. The benefits of this shrink fitting include a connection demonstrably stronger than the pipe. Also, very-high-strength material grades can be used. The problems with the technology include a lack of resonant-bending-fatigue data and a need for pipe that is produced to reasonable dimensional tolerances.

Shrink fitting would give operators the flexibility to join high-strength-steel pipe or dissimilar materials to weldable end caps. This would not eliminate the need for thick-walled welding, but would enable light-walled, flexible quad or hex joints, stress joints, and touchdown-zone joints to be created in preparation for welding in an S-lay or J-lay welding environment. In the case of titanium joints, flanges would not be necessary, only machined pipe, thus saving time and material costs.

Coupling Design

To provide an assembly that could be fatigue tested easily, we used a double-ended coupling design. This assembly method is also useful in proving the feasibility of assembling pipe into a single line, as might be used on a pipeline-lay barge. The couplings, once thermally expanded, swallow the pipe with the pipe butting against a shoulder, as shown in Fig. 1 (an isometric view of the fatigue-test specimen is seen in Fig. 2). Very-high-­pressure metal seals are created at the pipe ends. The interference pressure along the length of the coupling, along with the friction factor, determines the static capacity of the connection. The achievable interference pressure and sealing pressure are limited by the amount by which the coupling can be heated (hence expanded), the assembly clearance, and the residual stresses in the assembly.

jpt-2013-8-fatiguetestfig1.jpg
Fig. 1—Cross section of shrink-fit coupling.

 

jpt-2013-8-fatiguetestfig2.jpg
Fig. 2—Isometric view of fatigue-test specimen.

Locking balls are inserted through access ports in the coupling after assembly; these fill in grooves machined in both the coupling and the pipe to provide a backup against structural failure of the friction connection. The access ports also provide a convenient location to monitor the high-pressure seal.

Manufacture

The internal diameter (ID) of the pipe was left unmachined; only the outer diameter (OD) was turned down. Assuming worst-case supply tolerances, 0.2 in. should have been sufficient to achieve a machined round surface; the supplied pipe achieved a turned surface in less than 0.1 in.

The double-ended couplings were assembled vertically in two separate operations. The coupling was heated, and the first pipe inserted; the assembly was left to cool and then was flipped over; and then the coupling was heated once again for the second pipe installation. Vertical assembly in this manner proved reliable and efficient, using only an overhead gantry crane for alignment; such a technique could therefore be adapted for offshore construction.

Selective assembly was not implemented—that is, the couplings were not matched to pipes on the basis of tolerance, which underlines the manufacturability of the connection.

Testing

Hydrotest. All joints were statically hydrotested to 22,500 psig and then held for a minimum of 15 minutes using water as the pressure medium. All six specimens passed the hydrotest, providing good evidence of consistent sealing independent of machining tolerances (no selective assembly).

Resonant Bending-Fatigue Test. Six couplings (giving a total of 12 shrink-fit connections) were subjected to ­constant-amplitude resonant bending fatigue at three nominal stress ranges (two per connection).

The resonant-fatigue machine used consisted of two supports, a variable-speed electric motor, a drive housing, and a dead-weight housing. The variable-speed motor rotated an eccentric mass in the drive housing clamped to one end of the sample load. The rotational velocity of the motor was adjusted to load the sample near its natural frequency (24 Hz). The dead-weight housing was clamped to the other end of the sample to balance the assembly.

The stress was measured using eight axial strain gauges on the OD of the pipe (four on each side of the connector at 90° spacing). Maximum and minimum strain-gauge readings were averaged for 1-minute intervals and the averages recorded. Additional data recorded included the mean and standard deviation of the 1-minute averages, along with the cycle count.

Fatigue testing was carried out with water with an internal pressure of 3,000 psig in order to check for leakage and to provide an element of mean axial stress. The selection of 3,000 psig was reasonably representative of application for both drilling and production scenarios and tested the connection fatigue beyond a nominal near-ambient pressure.

Because the performance of the connection is unknown, the highest-stress range was tested first to establish a baseline of performance. As a result, the values of the medium- and low-stress ranges were adjusted to give confidence in the connection at a wide variety of stress ranges. Each sample was tested to failure. The fatigue-test results are shown in Table 1.

jpt-2013-8-fatiguetesttab1.jpg

Failure Analysis

Initial Observations. In all six specimens, the fatigue testing was halted following observation of a leak from between the coupling and the pipe. The access ports were checked and showed no signs of leaks, indicating that the failure was not of the high-­pressure seal but rather of the pipe outboard of the ports. Furthermore, the locking balls were freely retrieved from the access ports, indicating that there was no failure or movement of the friction connection after hydrotesting or fatigue testing.

Owing to the assembly method, one pipe/coupling connection was exposed to high temperatures twice (i.e., it was reheated). However, there was no proven link between reheated couplings and failure. In addition, there was no relationship between the sample orientation in the fatigue-test rig (drive-end or dead-end) and failure.

The bores and ODs of the pipes inside the couplings were examined for fatigue cracks using dye-penetrant inspection, and cracks were found at the entry point of the pipe to the coupling.

Five out of the six failures were identified as circumferential fatigue cracking of the pipe at approximately the same location. The remaining specimen did not indicate cracking; however, it is plausible that the dissection activities destroyed the crack.

Metallurgical Analysis. There was clear evidence on all examined specimens that crack propagation occurred from the OD of the pipe to the ID. Upon examination of the outer machined surface of the pipe, there was evidence that the machined surface had been degraded—where the pipe surfaces were in contact with the coupling, the machined surface had been flattened, and oxide-filled pits developed. The flattening of the machine marks would involve a fretting mechanism with fragment removal from the surface that could be oxidized and pushed into the external pipe surface. There was also evidence of oxide scale from the coupling within the same pits. The failed samples showed evidence of fretting, with cracks forming from the bottom of oxide-filled pits. There was evidence of oxide-scale debris compacted inside the crack, which could act as a wedge. The origin of the oxide-debris material was the relative movement between the pipe and the coupling in the presence of the high contact stress.

The oxide scale likely originated from degradation of the pipe surface caused by fretting and from oxidation of the coupling during the heating process.

The conclusion is that oxide scale plays a major part in crack initiation and propagation, and that limitation of this effect may give enhanced fatigue performance. Owing to the interference pressure required to generate sufficient static load and sealing capacity, it is unlikely that the 300°C limit could be achieved without cryogenic assistance. For example, the pipe ends could be soaked in liquid nitrogen before assembly; however, this would bring associated risks of metal embrittlement and would also pose some challenging safety hazards.

An alternative may be to use an inert shielding gas such as argon on nitrogen to prevent oxidation of the hot metal. A gas purge of the coupling during heating might be relatively easy to put in place; it would also be very easy to test the efficacy of such a technique in the small scale.

This article, written by JPT Technology Editor Chris Carpenter, contains highlights of paper OTC 24069, “Fatigue Testing of Shrink-Fit Couplings for Joining High-Strength-Steel Riser Pipe,” by J. Shield and J. Wightman, Subsea Riser Products; J. Pappas, RPSEA; and J. Bowman, Chevron, prepared for the 2013 Offshore Technology Conference, Houston, 6–9 May. The paper has not been peer reviewed. Copyright 2013 Offshore Technology Conference. Reproduced by permission.