Paper presented at Materials Science & Technology 2010. Houston, Texas, USA, 17-21 October 2010.
Keywords: Dissimilar metal, cathodic protection, Hydrogen, sub-sea, fracture toughness.
Dissimilar metal joints are commonly used in the oil and gas industry for subsea applications. This arrangement allows for closure welds to be carried out in the field, without the need for PWHT of large structures. However, occasional failures have been observed, with the fracture path surprisingly following both the austenitic microstructures immediately adjacent to the fusion line and the fusion line itself. The fracture mechanism is attributed to a hydrogen assisted mechanism, with the hydrogen stemming from the cathodic polarization of the structure for corrosion protection. The challenge is to determine the conditions which lead to catastrophic failure of a very small number of components in the field. Important parameters related to this type of fracture mechanism will be presented, including, susceptibility of the microstructure to hydrogen assisted cracking mechanisms, effect of test conditions on determination of mechanical properties of the interface, and accelerated testing techniques.
Dissimilar metal joints are commonly used in the oil and gas industry for subsea applications. This arrangement allows for closure welds to be carried out in the field, without the need for PWHT of large structures. However, occasional failures have been observed, with the fracture path following both the austenitic microstructures immediately adjacent to the fusion line and the fusion line itself. The fracture mechanism is attributed to a reduction in toughness of the microstructure due to hydrogen stemming from the cathodic polarization of the structure for corrosion protection. The faster diffusivity of H in ferrite, and greater solubility in austenite, is believed to cause H to build-up on the austenitic side of the dissimilar metal interface. The challenge is to determine the conditions which lead to catastrophic failure of a very small number of components in the field. Although standard fracture toughness tests (CTOD) have been used extensively to ascertain the resistance to propagation of a pre-existing crack, the fracture in this case is mechanically driven by the constant displacement of the test frame. Alternatively, a new test setup is investigated to look at the nucleation and propagation of cracks from loss in toughness uniquely from the accumulation of hydrogen, as could happen after long service times.
To determine fracture toughness in terms of CTOD and J resistance curves, single edge notched bend (SENB) specimens (BxB) were manufactured to BS 7448:Parts 1, 2 and 4.[1-3] The AISI 8630M substrate welded with ERNiCrMo-3 (Alloy 625) filler metal dissimilar metal weld specimens were notched in the through-thickness direction at the dissimilar interface. Specimen thickness B was approximately 12mm.
Three specimens (W17-A2 to A4) were notched by electro-discharge machining (EDM) using a 25µm wire (actual notch width approximately 30µm). The nominal notch depth was 6mm (a/W=0.5) and the notch position was selected so that the notch tip would be on the dissimilar interface. Two more specimens (W17-A5 and A6), were notched by EDM up to approximately 4.7mm depth, then fatigue pre-cracked for 1.3mm, resulting in a total nominal notch depth of 6mm (a/W=0.5). These five specimens were tested in air at 4°C using the unloading compliance method at the standard testing rate. Two further specimens (W17-14 and 15) were fatigue pre-cracked (a/W=0.5) and tested at 4°C in air using the unloading compliance method at a low loading rate of 0.018mm/h.
Three specimens (W17-A8 to A10) were prepared, with fatigue pre-cracked notch tips (a/W=0.5) positioned at the dissimilar interface. These specimens were hydrogen pre-charged at -1100mVSCE in 3.5%NaCl for at least 48 hours before commencing the tests. Unloading compliance tests were carried out at 4°C in 3.5%NaCl aqueous solution under cathodic protection at -1100mVSCE, applied by potentiostat. A low loading rate of 0.018mm/h was employed to allow hydrogen diffusion to the crack tip.
Accelerated hydrogen cracking susceptibility testing without externally applied load
Accelerated hydrogen cracking test procedure
Although the fracture mechanics tests show the resistance to propagation of a pre-existing crack, the fracture in this case is mechanically driven by the displacement of the cross-head. Alternatively, there is interest in looking at the nucleation and propagation of cracks from loss in toughness uniquely from the accumulation of hydrogen, as could happen after long service times. In the present study, no externally applied load is used, so it is only the residual stress from the weld butter which is available to drive the cracks.
Three interfaces were selected for accelerated H testing: F6NM to alloy 625 with stacked and layered weld bead sequences and F65 to alloy 625 weld metal interfaces. The F6NM specimens were PWHT, while the F65 was not. Characterization of the mechanical properties along the interface is presented in the following section.
12x12mm cross section coupons of 100mm length, and containing the dissimilar weld interface at the middle length, were immersed in 0.1M H2SO4 + 1g/l Na2HAs2O4. The Na2HAs2O4 was added to the solution to poison the recombination of hydrogen to H2 and thus promote H adsorption into the steel.[4-6] The steel side of the specimens was fully immersed, with the solution covering the dissimilar metal interface plus 2mm of the nickel-rich side. A constant potential of -1100mV versus a standard calomel reference electrode was applied between the work-piece and a platinum counter electrode. An acoustic emission (AE) probe was attached to the top surface of the specimens, parallel to the dissimilar metal interface for detection of crack initiation time. After testing, all samples were cross sectioned perpendicular to the dissimilar metal interface, prepared for metallographic observation and surveyed with an light microscope with magnifications up to 500X.
Mechanical properties across dissimilar metal interface
Micro-hardness maps with 100g load and 62µm spacing between indents were carried out for an overview of the hardness distribution across the dissimilar metal interfaces.
For greater resolution when compared to the mico-hardness survey, nano-hardness measurements were carried out at a spatial resolution (indent distance) of 5µm. Hardness and elastic modulus of the welds were characterized at 25°C using Micromaterials Nanotester equipped with a Berkovich tip. Nanoindentation experiments were composed by arrays of 5x20 indentations across the weld to parent material interfaces. Measurements extended 25 and 75µm into the steel and Ni rich alloy, respectively. Indentation tests were performed at a maximum load of 2mN and hardness and elastic modulus were calculated by Oliver and Pharr analysis. 
The results from the fracture mechanics tests conducted in air are presented in Figures 1 and 2 for EDM and fatigue pre-cracked notched specimens, respectively, in terms of J (corrected for crack growth in accordance with BS 7448 Part 4) versus ductile crack extension or tearing (Δa). Only valid data points according to BS 7448-4 are shown in these figures.
A best fit curve was fitted using the following power law:
Where J is in N/mm and Δa in mm. Coefficients l and x, as well as values of J0.2BL for each fitting curve, are given in Table 1.
Table 1. J R-curve results (power law fit of valid data points to BS 7448 Part 4)
|J R-curve coefficients(1), N/mm
|Tests carried out in air
|EDM + fatigue pre-cracked
|Tests carried out in 3.5% NaCl
|EDM + fatigue pre-cracked
(1) Assuming the following power law: J=l (Δa)x, where J is in N/mm, Δa in mm.
As shown in Figures 1 and 2, the majority of the valid data points for all of the tests lie above the maximum value for J (Jmax) as defined by BS 7448 Part 4.
The results from the fracture mechanics tests conducted in 3.5% NaCl are presented in Figure 3. A best fit curve was fitted in each case. The coefficients l and x, along with values of J0.2BL are given in Table 1. This figure also includes J R-curves for EDM notched specimens tested previously by Beaugrand et al, for comparison.
Figure 4 shows all of the J R-curves obtained. It shows that the lower bound J R-curve and the minimum value of J0.2BL in air, ie 189N/mm, were obtained from specimen W17-A6 (fatigue pre-cracked). The same figure also shows that the J R-curves obtained from specimens tested in 3.5% NaCl with cathodic protection were considerably lower than those obtained in air. The lower bound J R-curve for the specimens tested in 3.5% NaCl was obtained from the fatigue pre-cracked specimen W17-A10. The lowest J0.2BL was measured for specimen W17-A9 (also fatigue pre-cracked). Of the specimens tested in 3.5% NaCl all three fatigue pre-cracked specimens gave lower curves than the EDM notched specimens.
Accelerated hydrogen cracking susceptibility test
The results of the accelerated H cracking susceptibility tests under no applied external load are summarized in Table 2. Although significant indications of up to 120dB intensity were recorded by AE, the cross sections revealed that no cracking occurred during the tests. An example of the AE output of energy versus time, for the F6NM to 625 sample welded using a layered weld bead sequence and tested for 20 hours, is shown in Figure 5. A minor crack was observed at the F6NM/625 stacked interface after 164hrs exposure, but the location of the crack, away from the interface suggests that it was most likely not a result of reduced toughness by H saturation. The AE signals are thought to be due to the bursting of H bubbles at the sample surface.
Table 2. Accelerated H cracking susceptibility under no applied external stress test results
|F65/625 (No PWHT)
Mechanical properties across dissimilar metal interface
The microhardness measurement results for the F65/625 interface, and F6NM/625 interfaces, are presented in Figures 6 and 7, respectively. Interestingly, higher hardness was not measured on the F65 side of the interface, even though no PWHT was carried out for stress relieving. The F6NM and alloy 625 had similar hardness, with a drop in hardness on the 625 side, adjacent to the interface.
Figures 8(a)-(c) show nanoindentation arrays performed in the F65 interface and resultant 2D maps of the hardness and elastic modulus obtained by probing regions shown in Figure 8(a). Comparing Figure 8(b) and (c) with Figure 8(a) it is possible to understand that nanoindentation was able to define the mechanical properties gradients contained within the microstructure scale. The average values of the hardness and of the elastic modulus across the y axis is presented in Figures 9(a) and (b), respectively. From the nanohardness measurements, an increase in hardness was measured along the interface, especially towards the F65 side.
Figures 10(a) and (b) show nanoindentation hardness and modulus maps of the F6Nm to alloy 625 interface with a stacked weld bead sequence and layered weld bead sequence, respectively. Higher hardness was measured on the alloy 625 side, with lower hardness on the F6NM, showing the efficiency of the PWHT. The nano-hardness measurements confirm the micro-hardness data, showing a slight decrease in hardness immediately adjacent to the interface, towards the 625 side. However, it must be pointed out that the indentation spacing of 5mm would not be capable of revealing features smaller than this.
The fracture mechanics tests show that the environment has an effect on fracture toughness: the specimens tested in the marine environment under cathodic polarization had lower J R-curves than those tested in air. Notch tip geometry also had an effect, with the tests on the fatigue pre-cracked specimens in marine environment showing lower J R-curves than EDM notched specimens. The effect of notch geometry was less pronounced for the specimens tested in air, however, the lower bound curve was obtained from a fatigue pre-cracked specimen.
The tests conducted on fatigue pre-cracked specimens in air at the slow strain rate had comparable J R-curves to those tested in air at the normal strain rate. This suggests that reducing the strain rate by itself has little effect on the measured fracture toughness, implying that the significantly lower results measured in 3.5% NaCl solution are due to the presence of hydrogen. However, it is important to note that strain-rate effect in the presence of H is significant, and lower J-R curves are usually associated with slower test load rates under these conditions.
The resultant nanoindentation maps were able to resolve the different micro-phases in terms of mechanical properties, and this type of information in real scale may help to improve understanding of failure modes at microstructure level. Although significant differences in local mechanical properties across the dissimilar metal interfaces were measured, all specimens appeared immune to H assisted cracking under the conditions tested. Until further testing is conducted, no correlation of susceptibility of the microstructure to H embrittlement to localized mechanical properties can be drawn.
From the large reduction in toughness observed in the fracture toughness tests carried out in simulated sea water under cathodic protection, cracking might also be expected under much higher H contents (albeit at much lower loading), such as in the accelerated H tests under no externally applied loads. It is not clear yet if the lack of cracking observed here are due to the high resistance to H-cracking of the dissimilar metal interfaces tested, or due to low residual stresses due to the small cross section of specimens. The authors will be extending the experimental work to include new dissimilar metal interfaces, such as the 8630/625 tested here (CTOD) and also to investigate the effect of specimen size on cracking susceptibility. Also, the use of acoustic emission will be explored further.
The authors would like to thank MicroMaterials for the nano-hardness testing and support in the course of this project.
- British Standards Institution, 1991: BS 7448-1 'Fracture mechanics toughness tests. Part 1: Method for determination of KIc, critical CTOD and critical J values of metallic materials', BSI.
- British Standards Institution, 1997a: BS 7448-2 'Fracture mechanics toughness tests. Part 2: Method for determination of KIc, critical CTOD and critical J values of welds in metallic materials', BSI.
- British Standards Institution, 1997b: BS 7448-4 'Fracture mechanics toughness tests. Part 4: Method for determination of fracture resistance curves and initiation values for stable crack extension inmetallic materials: Fracture mechanics toughness tests', BSI.
- S.M. Charca, Oswald N.C. Uwakweh, Basir Shafiq, and Vinod S. Agarwala, Characterization of Hydrogen Permeation in Armco-Fe during Cathodic Polarization in Aqueous Electrolytic Media, Journal ofMaterials Engineering and Performance Volume 17(1) February 2008 pp. 127 - 133.
- S. Homrossukon, Hydrogen Induced Cracking in low Strength Steels, Thammasat Int. J. Sc. Tech., Vol. 9, No. 2, 2004, pp 56 - 61.
- S.M. Charca, Master's thesis on Study of H permeation and Diffusion in Steels: Predictive Model for Determination of desorbed H Concentration.
- Beaugrand V, Smith L and Gittos M, 2009: OMAE 2009-80030 'Hydrogen embrittlement of 8630M/625 subsea dissimilar joints: factors that influence the performance', to be presented at OMAE 2009, May 31-June5, 2009, Honolulu, Hawaii, USA.
- A.C. Fisher - Cripps, Nanoindentation -Mechanical Engineering Series, Springer Science - Second Edition (2004).