As described in Job Knowledge articles 78, 79 and 80 (Fatigue testing, pts. 1, 2 and 3), welds have a low fatigue strength. S-N curves, published in standards such as BS 7608 or DNV-RP-C203, describe the fatigue strength of welded details for different combinations of weld geometry and applied loading. When a designer specifies a new welding procedure they must be confident that these new welds will have sufficient fatigue strength to survive the applied cyclic service loads without cracking. This is particularly important in structures with no redundancy in the design, such as girth welds in risers and flowlines.
Girth welds can have a wide range of fatigue strengths. Those made on a backing bar are BS 7608 Class F2 details whereas ‘defect-free’ girth welds with the weld caps ground flush are BS 7608 Class C details - this difference can result in as much as a factor of 10 on fatigue life for a given stress range. In addition to these weld-design considerations, there are many other factors which influence a weld’s fatigue strength which are less easy to control. These include weld profile, joint misalignment, the presence of defects and residual stresses. The way in which each of these factors influences the fatigue behaviour of a weld is difficult to predict. Therefore, the safest way to determine the fatigue strength of a girth weld that has been made using a new welding procedure is by testing representative specimens. This way, designers can gain confidence that the welds produced will withstand the expected service loads or that they will be at least as strong as the required design curve.
Options for determining fatigue strength
There are two main methods to determine the fatigue strength of girth welds through testing. One is by extracting strips from pipes and then testing these in hydraulic test machines. These tests are usually run at a frequency of up to 5Hz and can be run at high stress ranges and in environments other than air. However, when strips are cut from pipes containing girth welds, the residual stress profile is no longer representative of the complete joint and the specimens produced may not contain the most fatigue-critical section of the girth weld. This can result in an over-prediction of fatigue strength (particularly in the high cycle regime) and fatigue limit (Maddox and Zhang, 2008).
The alternative is to perform full-scale fatigue testing. The main benefit is that the whole girth weld, in its natural as-welded condition, is subjected to the fatigue load cycle. Conventional test methods require huge load capacities, but for testing in rotating bending the resonance method is a fast and energy-efficient approach. It is also a very efficient method for determining the fatigue strength of other tubular structures such as pipes with polymeric coatings and mechanical connectors. The test frequency is high, at around 30Hz. There is no standard that defines a resonance fatigue test; however, TWI has over 15 years’ experience of running resonance fatigue testing programmes.
The principle of resonance fatigue testing
Resonance fatigue testing involves exciting a test specimen close to its first mode of vibration by applying a rotating radial force to one end. A bending moment is generated in the specimen, which rotates about the pipe axis, resulting in all of the longitudinal fibres in the specimen experiencing the same bending moment during one revolution of the excitation force (Figure 1). The specimen vibrates in the first mode, and so there are two locations along the specimen length at which there is no deflection. Specimens are supported at these nodal points.
Along the specimen length, the bending moment is a maximum at the mid-length and decays to zero at the specimen ends, as shown in Figure 2.
The resonant frequency of a test specimen depends on the specimen’s mass and stiffness, and therefore the outer diameter and wall thickness of the pipe being tested. Specimen lengths are chosen so that they have a resonant frequency of around 30Hz. At this frequency, approximately 2.5million cycles are applied to the test specimen in each 24 hour period. TWI’s resonance test machines can accommodate pipes in the range 8in to 36in outer diameter. Typically, 8in outer diameter specimens are around 4.5m long and 36in outer diameter specimens are around 10.5m long (Figure 3).
Tests are run below the resonant frequency so that the applied stress range can be carefully controlled (Figure 4). By altering the speed of rotation of the excitation force (which in practice is achieved by altering the motor speed), the deflection and therefore strain range can be controlled.
The applied strain is monitored using strain gauges that are located in the area of interest (Figure 5).
When qualifying girth welds to determine whether their fatigue strength is at least as good as a particular design S-N curve, the nominal applied stress range is needed. In order to measure the nominal stress range, strain gauges are positioned such that they are remote from the weld to avoid any secondary bending stresses associated with misalignment at the joint, but close enough that they are not significantly affected by the decrease in applied bending moment remote along the specimen length (as shown in Figure 2). A distance of 60mm from the weld toe is the ideal strain gauge location for this purpose. The resonance method is capable of applying nominal stress ranges between around 50MPa and 250MPa.
The resonance test method applies a fully alternating stress cycle with a stress ratio, R, equal to -1. However, since the residual stress profile in girth welds is difficult to predict it is important that a high tensile mean stress is applied to them during fatigue testing so that the results are conservative. In almost all cases, a mean stress which is at least half of the highest stress range used in the tests is applied to specimens by internally pressurising them with water. This produces a positive R-ratio and so ensures that the full applied stress range is tensile.
The added benefit of using internal pressure to apply a mean stress is that it also acts as a means of crack detection, and so resonance tests are set up to stop automatically when the internal water pressure drops due to the presence of a through-wall crack. At TWI, alternatives to using internal pressurisation include carrying out tests with a mechanically applied tensile or compressive mean load, or with cooling water flowing through specimens.
When cracking occurs, stresses redistribute and this can also be detected by strain gauges located close to the crack position. The ability to detect cracking via the strain gauge readings is particularly useful in complex connector specimens in which, for example, a crack may initiate in a location which does not result in a drop of internal pressure, or in specimens tested with a mechanically applied mean stress rather than internal water pressure.
Typical test programmes
In a resonance testing programme to qualify girth welds, engineering judgement is used to select the number of specimens to test. The industrially accepted approach is to test nine specimens, three at each of three stress ranges.
In a typical test programme, high and medium stress range tests would be run until through-wall cracking occurred, while low stress range tests could be stopped as ‘runouts’ (above the target life but before cracking has occurred). The results from cracked welds would then be compared to a target curve which is based on a design S-N curve (from BS 7608 or DNV RP C203) and gives a specified level of statistical confidence that the results qualify to that fatigue class.
- BS 7608 (2014) ‘Guide to fatigue design and assessment of steel products’, British Standards Institution, London.
- DNV-RP-C203 ‘Fatigue design of offshore steel structures’ Det Norske Veritas, Norway.
- Job knowledge articles 78, 79 and 80 on fatigue testing.
- Zhang Y-H, 2011: ‘Comparison of the fatigue performance of full scale girth welded pipes and small scale strip and plate specimens: A literature review’ TWI Industrial Member Report 986/2011
For more information, please contact us.
Written by Carol Johnston.