VARIABLE AMPLITUDE CORROSION FATIGUE OF HIGH STRENGTH

WELDABLE STEEL

S. S. Ngiam and F. P. Brennan

NDE Centre, Department of Mechanical Engineering, University College London
Torrington Place, London WC1E 7JE, UK

ABSTRACT
In recent years there has been considerable interest in the offshore industry in using high strength steels in the
construction of offshore structures. This growing trend has seen high strength steel widely used in the fabrication of Jack-
up platforms for production purposes, largely due to the optimisation of light weight design. This is particularly important
due to the competitive Oil & Gas industry, as considerable reduction in weight can lead to a significant saving in both
manufacturing and operation costs.
High strength steels with nominal yield strength of circa 450 to 700MPa have been commonly used in the fabrication of
jack ups, especially in the leg structures. Early realistic loading fatigue studies were conducted on the offshore structures
in the North Sea, focusing only on fixed platforms. For example, with the aid of the advancement of high-speed computer
during the last two decades, several working groups such as COLOS (Common Load Sequence for the European Coal
and Steel Community Research Programme II), UKOSRP and Wave Action Standard Load History (WASH) [1, 2 and 3]
were established to perform realistic variable amplitude loading fatigue testing of offshore structures in the North Sea.
These committees provided the foundation of standard load histories for variable amplitude fatigue testing. However,
these studies looked only at fixed platforms. The load histories are not appropriate for Jack up structures as these are
subjected to different loading conditions due to the different dynamic response of these structures. In this study, Jack up
Offshore Standard Load History (JOSH) [4 and 5], which was developed under the framework of WASH, is used as the
standard load history generator.
Previous High Strength Steel projects have concentrated mainly on establishing the influence of cathodic protection, and
overprotection, on the fatigue life of welded joints as the HSE Guidance has proposed the need for CP limiters when using
High Strength Steels. A major finding from the early studies was that at least some of the newer steels were not very
susceptible to overprotection (-1050mV) in a corrosive environment at modest lives. The impetus for this research
programme stems from concerns over the lack of data for fatigue of high strength steels in the long life region. This study
has investigated the Long Life Corrosion Fatigue performance of SUPERELSO 702 (SE 702), a widely used weldable
High Strength Steel (made by Creusot-Loire Industrie as a NiCrMo steel of similar strength).
This paper reports the most recent work on longer life VACF and a modification of JOSH. The results presented include
the measured fatigue life, fatigue crack growth rate and crack shape evolution. Stress life (S-N) and linear elastic fracture
mechanics are the major approaches adopted for the analysis of the T-butt specimens tested. Experimental data is
compared with data previously obtained from the air test.
Experimental Programme
A total of five SE 702 T-butt welded specimens were supplied by CLi. The full-penetration welded T-butt specimens were
fabricated from 50mm thick plates, including the attachment. The welding consumable used was Oerlikon Fluxofil. Figure
1 shows the detailed dimensions of the test specimens.

1
 50

Width of plate =190mm

= 500

500 to 600

50

80

495

Figure 1: Detailed dimensions of T-butt specimens in mm

Prior to testing, the weld toe stress concentration factors were determined using published parametric equations [7]. A
total of four corrosion tests and one air test (to serve as control specimen) were conducted. Table 1 summarises the
general fatigue parameters for all the tests.

Cathodic Protection
Test Environment Equivalent Stress Range
Level
TS1 Air - 149MPa
TS2 Corrosive (Sea water) -800mV 149MPa
TS3 Corrosive (Sea water) -1050mV 149MPa
TS4 Corrosive (Sea water) -800mV 101MPa
TS5 Corrosive (Sea water) -1050mV 101MPa
Table 1: Summary of fatigue testing program

Jack-up Offshore Standard Load History (JOSH)

Jack-up offshore standard load history (JOSH) [5, 8] was developed under the framework of Wave Action Standard Load
History (WASH) [1]. The JOSH variable waveform generator has been updated and modified so that it can be adopted in
this study. The complete JOSH sequence was generated using 2000 sea state blocks of 10 minutes sub-block duration.
Sea state 1 has been set as the starting sea state for the sequence, as suggested by Tantbirojn [8] and one complete
sequence consists of 536,657 turning points, which is equivalent to 268,326cycles. The sequence was generated based
on the power spectral distribution for the corresponding 8 sea states recorded on a Jack up platform at North Sea and the
corresponding power spectra are shown in Figure 2. In order to ascertain that the sequence generated is appropriate for
fatigue testing purposes, relevant analysis was performed on the generated load history. The analysis performed included
the stationary point of the sequence, the adequate mixing of sea states, the sea state sub-block duration, the significance
of different starting sea states, and the convergence of the equivalent stress range and power spectral density function.

Power Spectrums
0.80

Sea State 8
0.70
Sea State 7

0.60 Sea State 6

Sea State 5
0.50
Normalised PSD

Sea State 4

0.40 Sea State 3

0.30 Sea State 2

Sea State 1
0.20

0.10

0.00
0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35
Frequency (Hz)

Figure 2: Power spectra for the 8 sea states used in JOSH

Figure 3 shows the real time sea state occurrences of the load history.

2
 Real Time Sea States Occurrences for JOSH10 Sequence
8

Sea States
4

0
0 5,000 10,000 15,000 20,000
Time (mins)

Figure 3: Real time sea states occurrences for JOSH10 sequence

Details of the JOSH10 sequence can be found in Table 2 while the stress range distribution is shown in Figure 4.

Parameter Value
Sea states used 1-8
Mean stress(MPa) 321.32
Maximum stress(MPa) 613.42
Minimum stress(MPa) 29.2
RMS(based on rain flow ranges) (MPa) 132.98
Clipping ratio 4.61
Equivalent stress range (MPa) 149.08
Total number of cycles (Rain flow) 268,326
Average frequency (Hz) 0.24

Table 2: Summary of parameters in JOSH10 sequence

Stress Range Distribution

25,000

20,000
Number of Occurances

15,000

10,000

5,000

0
0.00 100.00 200.00 300.00 400.00 500.00 600.00 700.00
Equivalent Stress Range (MPa)

Figure 4: Stress range distribution obtained using rainflow cycle counting

As mentioned earlier, relevant analysis were performed in order to verify that the load sequence is adequate and
representative of a variable amplitude conditions. For example, the convergence of the Equivalent stress range and
power spectral density function of the load sequence can be seen in Figure 5.

3
 Convergence of Power Spectrums
0.6

0.5

0.4 Number of data points

10k

Normalised PSD
100k
150k
0.3
300k
550k
1000k
0.2

0.1

0
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7
Frequency (Hz)

Figure 5: The convergence JOSH10 power spectra

Simulation of Environmental Conditions

In corrosion fatigue, the appropriate simulated conditions must be achieved in order to obtain representative results. As
mentioned before, four corrosion fatigue tests will be conducted in the current study, under simulated seawater
environment.
The welded intersection was immersed in artificial seawater made to ASTM D1141 standard [9]. The simulated seawater
was pumped from an insulated reservoir through a closed loop passing the environment chamber. The environment
chamber around the parent plate and welded section was constructed from 10mm Perspex sheet and silicon sealant,
which provides the required flexibility during the fatigue test. The fully aerated seawater was pumped from an insulated
reservoir through a closed loop circulation into the environment chamber. The seawater temperature limits was
o o
maintained between 8 C and 10 C using an auxiliary refrigeration system. The pH level was also monitored constantly in
the course of the test and maintained between 7.8 and 8.2. Two cathodic protection (CP) levels (-800mV and -1050mV)
with respect to Ag/AgCl reference electrode, were used in this study. All CP potentials were based on an impressed
current method, applied and maintained by using a Wenkin CP Potentiostat.
Prior to the fatigue test, the welded section was submerged in the artificial seawater with the adequate cathodic protection
level. A soak time of two weeks was adopted in order to achieve uniform polarization of the specimen and allow hydrogen
to diffuse through the steel. The soak time implemented was based on previous studies [4 and 6], which concluded that a
soak time of two weeks is sufficient for uniform hydrogen diffusion into the specimen.

S-N Results
The fatigue lives of the specimens are presented in the S-N format. Since the fatigue tests were conducted under variable
amplitude loading conditions, the stress range shown is the equivalent stress range for the variable amplitude sequence
used while the life represents the number of fatigue cycles to failure. The end of each test was defined as the attainment
of a half plate thickness crack. Figure 6 shows the S-N plot of the experimental results obtained, together with the Class P
(Class F) design curves. It should be noted that the correction factor to 16mm plate thickness was applied to all the S-N
data in the plots. The fatigue lives predicted by the P Design Curves for Air and Sea environments of the 16mm 50D joint
data are included in Table 6 for comparison.

4
 SN Curve for T-butt Welded Plates with Thickness correction to
16mm thick plate

1000

Equivalent Stress Range, S (MPa)

100

10

TS1- T Plate Air (149MPa)

TS2- T Plate CP -800 (149MPa)
TS3- T Plate CP -1050 (149MPa)
TS4- T Plate CP -800 (101MPa)
TS5- T Plate CP -1050 (101MPa)
P Curve (Class F) Air
P Curve (Class F) CP
1
10,000 100,000 1,000,000 10,000,000
Number of Cycles, N

Figure 6: S-N plots of experimental data and Class P Design Curves

Fatigue Crack Growth Data

The fatigue crack growth curves for all tests are presented in Figure 7 below.

Fatigue Crack Growth Curves for All Tests

30

25

20
Crack Depth, a (mm)

15

10
TS1 ESR=149MPa
TS2 ESR=149MPa

5 TS3 ESR=149MPa
TS4 ESR=101MPa
TS5 ESR=101MPa

0
0 500,000 1,000,000 1,500,000 2,000,000 2,500,000 3,000,000
No. of Cycles, N

Figure 7: Fatigue crack growth curves for all tests

For TS1, final inspection was performed after the specimens had undergone 321,778 of crack propagation cycles (total
fatigue life of 492,075) and the maximum crack depth recorded was 21.87mm. TS2 had a propagation life of 213,691
cycles and a maximum crack depth of 18.38mm was recorded from the final inspection. Although the crack depth had not
reached half the plate thickness, the test was terminated prematurely because there was severe leakage on the
environment chamber. It is believed that the decreasing rigidity of the specimen had caused the silicon sealant to break
down. Fatigue testing of TS3 was terminated at 231,287 cycles, after having undergone approximately 213,000 fatigue
crack propagation cycles. A maximum crack depth of 23.20mm was recorded from the final inspection. On the other
hand, TS4 had a fatigue crack propagation life of approximately 1,000,000 cycles. Finally, TS5 had undergone 507,197 of
crack propagation cycles (total fatigue life of 2,020,197) and the maximum crack depth recorded was 25.40mm.

5
 Discussion On Fatigue Test Results: Fatigue Life
The fatigue crack initiation data shows that TS1 and TS2 have similar initiation cycles. However, there is a 90% reduction
in the crack initiation cycles for TS3. On the other hand, TS4 and TS5, which were tested under an equivalent stress
range of 101MPa had fatigue initiation lives of approximately 1,850,000 and 1,500,000 cycles respectively. Although the
initiation cycles are significantly different between TS2 and TS3, the fatigue crack propagation cycles for both corrosion
tests are similar. For the long life corrosion tests, only a 17% reduction in the crack initiation cycles was observed in TS5.
The fatigue crack propagation cycles for TS2 and TS3 are similar while a 35% reduction was recorded in TS5 compared to
TS4. From the data obtained, it cannot be concluded that there is a susceptibility to hydrogen embrittlement in high
strength steel subjected to cathodic over-protection and it is believed that the significant reduction in initiation life of TS3 is
an isolated case. Furthermore, if the reduction in initiation life is solely caused by cathodic over-protection, a similar
reduction in fatigue crack propagation life should be observed in TS5.
The total fatigue life of TS1, TS2, TS4 and TS5 is about 490,000, 436,000, 2,600,000 and 2,000,000 cycles respectively.
The experimental fatigue lives for TS2 and TS4 are well above the P curve (Class F) CP line and also the Class F Air line.
Hence, it can be said that cathodic protection level of -800mV with respect to Ag/AgCl, as proposed by UK HSE Offshore
Guidance Notes [13], is adequate for the application to high strength steel. The most significant reduction in life (50%) is
observed in TS3, having a total fatigue life of just over 230,000 cycles. For TS5, there is a reduction of 22% in the total
fatigue life. In general, the shorter fatigue life of the corrosion tests is believed to be mainly caused by the hydrogen
embrittlement effect. In high strength steels, the effect of internal hydrogen super-saturation is significant, as a result, they
are more susceptible to hydrogen embrittlement. The mechanism of hydrogen embrittlement is known to increase the
crack propagation rate possibly through enhanced grain boundary attack. Although the results obtained have shown that
cathodic over-protection has a detrimental effect on the fatigue performance of high strength steel, it is not as detrimental
as suggested in some published literature [15]. The fatigue results have shown that the effect of cathodic over-protection
is less detrimental in the low stress regime. Overall, the results obtained are consistent with previous high strength steels
corrosion fatigue studies [6 and 14], where a reduction of 30 50% win life is observed for cases where the CP level was
increased from -800mV to -1050mV.
In the LLCF study conducted by Tantbirojn [6], the average fatigue life of the specimens was well under 300,000 cycles.
The short fatigue lives and the susceptibility of cathodic over-protection are significantly more severe than anticipated. As
can be seen from Figure 8, most of the results are close to and below the P (Class F) design curves.

SN Curve for T-butt Welded Plates with Thickness correction to

16mm thick plate

1000
Equivalent Stress Range, S (MPa)

100

TS1- T Plate Air (149MPa)

TS2- T Plate CP -800 (149MPa)

TS3- T Plate CP -1050 (149MPa)

10
TS4 - T Plate CP -800 (101MPa)

TS5- T Plate CP -1050 (101MPa)

P Curve (Class F) Air

P Curve (Class F) CP

Mean Line Air (Tantbirojn)

Mean Line CP -800 (Tantbirojn)

Mean Line CP -1050 (Tantbirojn)

1
10,000 100,000 1,000,000 10,000,000
Number of Cycles, N

Figure 8: Comparison of experimental fatigue lives with results obtained in the study conducted by Tantbirojn [6]

The variable amplitude load sequence used by Tantbirojn was found to have a decreasing amplitude against time. This is
mainly caused by the inadequate mixing of the sea states during the generation of the load history. In addition, the use of
inappropriate wave modelling parameters is also believed to have contributed to the undesired characteristic in the load
sequence. The relatively high stress ranges at the beginning of the sequence is found to have significantly accelerated the
fatigue crack initiation rate of the specimens.

6
 Fatigue Crack Growth
As can be observed from Figure 7, the fatigue crack growth under variable amplitude loading conditions is considered to
be steady although the variable amplitude load sequence (JOSH10) consists of multi-sea states with varying stress
intensity. However, this has been anticipated since the relevant analysis performed have shown that the JOSH10
sequence is stationary and the sea states are adequate mixed by using the Markov Chain-Monte Carlo technique [16].
This becomes more evident when the results are compared with the fatigue data obtained by Etube [4] and Tantbirojn [6].
Both the authors have used JOSH as the standard load history generator for fatigue tests but the sequence parameters
were different from those used in this study. As stated earlier, the Markovian transition matrix and numerical parameters
used in these studies have led to an inadequately mixed sequence. Hence, the statistical nature of the load sequence is
shown to have a detrimental effect on the corrosion fatigue crack growth. This is highly important as non-uniform fatigue
crack growth will cause problems during fatigue analysis and can lead to erroneous results. It must be noted that the
sequences used by the authors mentioned earlier are deemed to be inappropriate only for long life fatigue tests. This is
because in practice, a structure may be subjected to relatively higher loads (storm) immediately after installation.
However, the current load sequence is more adequate in this study since the emphasis is on the fatigue performance of
high strength steel in the long life region.
Conclusions
The S-N data obtained from the corrosion fatigue tests showed that there is an improvement in fatigue performance for SE
702 for both air and under cathodic protection than that shown for 50D material. Test results from TS2 to TS5 showed
that cathodic protection level of -800mV with respect to Ag/AgCl, as proposed by UK HSE Offshore Guidance Notes, is
adequate for the application to high strength steel. Although TS3 has a significant reduction in fatigue crack initiation life,
its fatigue life is well above the P curve (Class F) CP line. Results obtained from TS5 suggest that the significantly shorter
initiation life of TS3 may be an isolated case. Although increasing the CP level results in a shorter fatigue life, the effect of
cathodic over-protection in high strength steels is less detrimental than suggested. The LLCF test results suggest that the
effect of hydrogen embrittlement in high strength steels is less significant in the low stress regime. The load sequence
generated by using JOSH is found to be adequate for simulation of variable amplitude loading conditions. The relevant
analysis performed and fatigue crack growth data collected further confirmed the adequate mixing of sea states and
appropriate utilisation of wave modelling parameters. Since the number of fatigue tests conducted are limited, there would
appear that more tests are required in order to draw a clear conclusion on the fatigue performance of SE 702 at long life
regime.
References
1. Kam J C P (1989), Structural Integrity of Offshore Tubular Joints Subject to Fatigue, Ph.D. Thesis, University of
London, London.
2. IABG (1982), The Common Load Sequence for Fatigue Evaluation of Offshore Structures Background and
Generation, TF 1982 3342477401, United Kingdom.
3. Pook L P (1976), Proposed Standard Histories for Fatigue Testing Relevant to Offshore Structures, NEL Report 624,
National Engineering Laboratory, Glasgow.
4. L. S. Etube, Fatigue and Fracture Mechanics of Offshore Structures, Engineering Research Series- ERS-4,
Professional Engineering Publishing Ltd., ISBN 1-86058-312-1, 2001.
5. Etube L S, Brennan F P and Dover W D (1999), Modelling of Jack-up Response for Fatigue Under Simulated Service
Conditions, Marine structures, vol. 12, pp 327-348, United Kingdom.
6. Tantbirojn N, Etube L S, Dover W D, Kilgallon P J, Roberts T and Spurrier J (2001), Variable Amplitude Corrosion
Fatigue of Jack-up Steels, HSE VACF-T Report, London.
7. Brennan F P, Peleties P and Helier A K (2000), Predicting Weld Toe Stress Concentration Factors for T and Skewed
T-Joint Plate Connections, International Journal of Fatigue, Vol. 22, pp 573-584.
8. Tantbirojn N (2002), Fatigue Testing of Weldable High Strength Steels Under Simulated Service Condition, Ph.D.
Thesis, University of London, London.
9. American Society for Testing and Materials, Specification for Substitute Ocean Water, ASTM D 1141-75, 1980.
10. Dartec Limited (1997), Dartec 9600 Advanced Digital Controller Use Instructions, Issue no. 3, United Kingdom.
11. Collins R, Dover W D and Michael D H (1985), The use of AC Field Measurements for Non Destructive Testing,
Research Technique in NDT, Vol. Academic Press, United Kingdom.
12. Technical Software Consultant (1993), U10 Microgauge User Manual, TSC, United Kingdom.
13. UK Health and Safety Executive (2001), Offshore Technology Report- Steel, OTO 2001 015, United Kingdom.
14. Etube L S, Myers P, Brennan F P and Dover W D (1998), Constant and Variable Amplitude Corrosion Fatigue
th
Performance of a High Strength Jack-up Steel, Proceeding of the 8 International Offshore and Polar Engineering
Conference, Montreal, Canada.
15. Davey V S (1991), Hydrogen Assisted Cracking of High Strength Steels in the Leg of a Jack-up Rig, Health and
Safety Executive, Offshore Safety Division, United Kingdom.
16. Olagnon M, Lebas G and Thebault J (1988), Characterisation of Sea-States for Fatigue Testing Purposes,
th
Proceeding of 7 International Symposium of Offshore Mechanics and Artic Engineering, ASME, Houston, United
States of America.