The first significant systematic investigation into the effect of environment on the fatigue behaviour of welded joints was the United Kingdom Offshore Steels Research Project - Phase I (UKOSRPI) [1.45]. The results suggested the following environmental effects which were subsequently used for the basis of the design guidance.
1. Free corrosion can reduce endurance by a factor of two
2. Cathodic protection can restore the in air life
3. Under free corrosion conditions, fatigue damage continues to occur even below the air fatigue limit.
It has know been recognised that the above conclusions can lead to non conservative life estimates in some cases mainly due to the fact that recent tests have shown that cathodic protection does not always restore the in air life.
Austin [1.76] has considered a significant amount of data generated under free corrosion conditions outside UKOSRPI and he concludes that a factor of 2.5 would seem appropriate to account for the effect of environment. The data considered included both plate and tubular joint tests. Additionally, higher reduction factors were noted for post weld heat treated joints (PWHT) leading to the conclusion that free corrosion conditions can negate the beneficial effect of PWHT. A similar review [1.102] suggests the ratio between air and fatigue lives is closer to 2.2. Additionally a distinct laboratory bias was noted in the results with the UKOSRP falling towards the higher end of the scatter band.
The application of ‘adequate’ levels of cathodic protection make the trends less clear as the reduction factor changes with apphed stress range. Several research programmes testing welded plate joints suggest that fatigue hves are always at least as long as those in air (reduction factor <1) especially so at lower stress ranges e.g.
[1.103, 1.104]. However a significant amount of data has also been generated suggesting that welded plates with adequate levels of CP result in hves between 2 and 3 times shorter than in air. In general it appears that the SN curve is rotated counter clockwise resulting in lives longer than air only at low stress ranges.
Vosikovsky and Tyson [1.103] noted that the scatter in the CP data was increased compared to the free corrosion data and that the noted laboratory bias was even stronger than for free corrosion conditions. The best fit of all the adequate CP data reviewed (from Canadian, Japanese, ECSC Dutch, ECSC German and UKOSRP) showed a neghgible effect of CP on hves compared to those found in air. However exclusion of the UKOSRP generated data from the analysis results in an average reduction factor on life of 1.5.
The most plausible explanation of the bias is identified as the weld procedures used to deposit the weld toe bead. Corrosion fatigue is thought to be more sensitive to weld toe quahty than fatigue in air. Vosikovsky et al [1.105] monitored the number of initiation sites and found that in general the number of sites increased under free corrosion and decreased under CP than found in air tests. Examination of the relationship between the resulting fatigue hves and the number of initiation sites shows that the tests with few initiation sites produced significantly longer hves. Thus the transition to an edge crack (in plates) is delayed and the defects grow as semi-eUiptical fatigue cracks throughout most of the test. AdditionaUy, semi elhptical defects are more susceptible to the beneficial effects of reduced stress intensity factor due to crack wedging effects of calcareous deposits that may form on the crack surfaces under CP conditions. The random variation of the number of
Chapter 1 - Introduction and Background to Fatigue of Tubular Joints
initial defects severe enough to initiate a fatigue crack under CP conditions also explains the increased scatter observed on the CP results.
Tubular joints performed as part of UKOSRPII [1.46] did not exhibit the same beneficial effect of adequate CP levels found in UKOSRPI [1.45] plate tests. Consideration of experimental data from various world wide sources [1.105] suggests a factor of two on the corresponding air life accounts for the application of adequate CP levels. The beneficial effect at lower stress ranges found for welded plate tests was not found for welded tubular joints.
Cathodic over protection is limited mainly to welded plate tests and tends to show a similar trend to the lower CP levels but with a greater reduction in life longer lives are needed before the results approach the in air results. Smith [1.106] performed ten fatigue tests on tubular joints fabricated from the high strength steel API 5L X85. Five of the tests were performed with CP levels of -KXXlmV Ag/AgCl. This steel has a yield strength of approximately 600MPa and is commonly used in jack up construction. Smith concludes that the results of the tests fell within the scatter of the tests used in the derivation of the UK fatigue guidance and above the design curve. No comparison between the relative endurance found from the air and CP tests is given. The welding procedure used during the fabrication of some of the specimens was un-representative of those used in previous tubular joint fatigue tests (weave rather than stringer beads) and the through thickness crack paths highly unusual making interpretation of the results difficult.