You are on page 1of 6

Proceedings of the 2012 9th International Pipeline Conference IPC2012 September 24-28, 2012, Calgary, Alberta, Canada


Andrei Kotousov School of Mechanical Engineering, The University of Adelaide, SA 5005, Australia

Krzysztof Borkowski School of Mechanical Engineering, The University of Adelaide, SA 5005, Australia

Leigh Fletcher, Welding and Pipeline Integrity, PO Box 413, Bright, Vic. 3741, Australia

Reza Ghomashchi School of Mechanical Engineering, The University of Adelaide, SA 5005, Australia

ABSTRACT Due to significant cost and productivity advantages, low heat inputs, high welding speeds, severe loading conditions and the use of cellulosic electrodes in the construction of oil and gas pipelines are unavoidable in Australia. Another significant cost reduction directly related to the tonnage of steel pipe dictates the wider use of higher grade steels, such as X70, X80 or X100. These current tendencies raise a serious concern regarding potential problems associated with weld metal hydrogen assisted cold cracking, HACC. Although there are industry standards and guidelines for the avoidance of hydrogen cracking in the heat affected zone, this is not the case for the weld metal, which is now more likely source of crack initiation in modern pipeline steels. The current paper develops a simplified mathematical model to predict the risk of hydrogen cracking in weld metal. A sensitivity study is conducted to evaluate the effect of various welding parameters and geometry, such as heat input, preheat and ambient temperatures and wall thickness on the risk of hydrogen cracking. INTRODUCTION HACC has a long history and was frequently found in steel welded structures in the past (Petch, 1952). One of the main features of this physical phenomenon is its delayed nature; it occurs several minutes, a few hours or even days after completion of the weld. The hydrogen cracks can form longitudinal or transverse to the weld direction. The size of the cracks can also be very different. While the precise physicochemical mechanism(s) for HACC remain a subject of extensive debates, there is long-standing agreement that the fracture processes responsible for both heat affected zone (HAZ) and weld metal (WM) cracking are dependent on the combined effect of the following three conditions (Yurioka, 1999 among others): (1) sufficiently high concentration of

diffusible hydrogen; (2) sufficiently high stresses (they need to be tensile in nature to drive hydrogen cracks); and (3) a microstructure, which is susceptible to hydrogen cracking. In modern high strength pipeline steels HACC typically occurs in the weld metal rather than in the heat affected zone (HAZ). There are at least four reasons for such preferable location: (1) high strength steels accrue their mechanical properties under controlled thermo-mechanical processing, while weld metals gain high strength from additional alloying elements, and are therefore more susceptible to HACC; (2) the microstructure susceptible to hydrogen cracking, in particular, structural heterogeneity of weld metals, which can contribute significantly to initiation and propagation of hydrogen cracks; (3) welding of high strength steel normally generates higher level of residual stresses, which are localized in the vicinity of the weld increasing risk of HACC in weld metal; (4) the austenite () to ferrite () transformation in parent steel occurs at higher temperatures compared to that of the weld metal due to its lower carbon and alloy content. Since ferrite capability in dissolving hydrogen is lower than austenite, the rejected hydrogen due to transformation in the heat affected zone, HAZ, diffuses into the weldmetal, still austenitic, to further exacerbate weldmetal, which is already rich in H from the arc atmosphere, susceptibility to HACC. There has been a strong preference in Australia for cellulosic girth welding of oil and gas pipelines based on cost and productivity advantages. HACC is a potential serious problem because of the high hydrogen contents of cellulosic weld metal (presence of 30-40 ml/100g of weld metal is considered to be typical for cellulosic consumables). The current weld fabrication techniques utilized for thin walled small diameter line pipe have minimized the risk of hydrogen cold cracking, even in the absence of preheat. However,

Copyright 2012 by ASME

concerns about high strength grades such as X70 and X80 pipeline steel and strength matching cellulosic electrodes have driven intensive research on the subject of HACC (Dunne, 1999). The use of cellulosic electrodes, at least for the root pass is regarded as a key factor in cost effective pipeline construction in Australia (Alam, et al., 1996). The application of high strength grade pipeline steel for oil and gas transportation can provide significant savings in material costs through the use of thinner wall pipe capable of operating at high pressure. Despite the fact that there are industry standards and guidelines for the avoidance of HACC in the heat affected zone, this is not the case for the weld metal, which is a more likely source of crack initiation than HAZ in modern pipeline steels. In this case, the industry practice and academic research rely on a handful of qualitative weldability tests in order to determine the resistance of the weld metal to hydrogen cracking (Davidson, 1995). The purpose of this paper is to develop a simple theoretical model for prediction of HACC in WM. This simplified model utilizes the original McParlan and Graville approach (1976) developed specifically for weld metal hydrogen assisted cold cracking. A similar approach was also utilized by Glover and Graville (1999) to assess the risk of hydrogen cracking in multi-pass welds and its effect upon procedure design. The approach is based on the premise that cracking occurs when the local residual hydrogen concentration in a weld reaches a critical value. This value would be a function of the local stress and the susceptibility of the local microstructure, which can be assessed by a carbon index. In practice the determination of local stress concentration is very complex task and often impossible due to many factors affecting the stress distribution in weldments (Lindgren, 2006). However, the initiation of hydrogen cracks takes place in a weld metal region with a very high local stress concentration, where the stresses approach the yield stress of the weld metal. Therefore, a local increase in hydrogen concentration can be considered as the main factor contributing to the risk of HACC in WM with a fixed chemistry and microstructure of the weldments. The residual hydrogen concentration is a function of the temperature history of the weld. Therefore, this approach is coupled with an analytical equation for the prediction of the weld thermal history. The predictive WMHACC model requires a critical or threshold parameter, which can be determined from a limited number of weldability tests. NOMENCLATURE 2 2 = + + - moving coordinate: = , - time after welding, - travel speed, 2 ,

HYDROGEN DIFFUSION MODEL An extensive analysis of data on hydrogen diffusion in ferritic iron and steel by Coe (1973) and Yurioka and Suzuki (1990) is summarized in Fig. 1, which shows a steep drop in apparent diffusion coefficient for steel for temperatures below about 200oC. This decrease in atomic mobility is a result of hydrogen trapping mechanism or simply due to the kinetics of diffusion when temperature approaches ambient. Hydrogen trapping can be viewed as a manifestation of the interaction between hydrogen atoms and the various microstructural features in steel (Oriani, 1970).
D, m2/s 10-7 10-9 10-11 0 Lattice diffusion of hydrogen in -iron

- welding current, - welding voltage, - temperature at weld, - ambient temperature, - preheat temperature, - Carbon equivalent, - plate thickness, - thermal conductivity, - thermal diffusivity, - arc efficiency, - surface conductance at weld, - surface conductance at plate, - heat transfer coefficient, 2 - eigen values satisfying tan = 2 ( 2 2 ), 2 2 - Fourier coefficients, = ( + 2 2 + 2) 0 zeroth order 2nd kind Bessel function, - diffusivity of hydrogen, - density of trapping sites, 0 - average trap binding energy, - hydrogen concentration, - residual hydrogen concentration, - critical hydrogen concentration, - parameter, - parameter.

Trapping theory Ferritic steels 200 400 600 800 T, 0C

Figure 1: Diffusion coefficient of hydrogen in ferritic steel as a function of temperature. The scatter band reflects reduced diffusivities associated with hydrogen trapping (original data extracted from Yurioka & Suzuki, 1990)

Copyright 2012 by ASME

To describe the hydrogen diffusivity changes with the temperature Hirth (1978) has proposed the following temperature sensitivity of the diffusion coefficient: (1) () = 2 107 exp(828()) (m2/s) (t) is temperature in degrees Kelvin which is presented here as a function of time as the weld cools down. The apparent diffusivity , in the presence of traps, can be estimated by the Oriani equation: (2) = ()(1 + exp(0 ())) (m2/s)

Mn Cr + Mo Ni (8) + + 6 5 15 The value of in equation 5 was reported to be 49 cm-2 for the G-BOP test (Yurioka, 1999). = C +

where is the density of trapping sites, 0 is the average trap binding energy. The theoretical curve in Fig. 1 is based on = 4.4 103 traps/lattice site and 0 = 26 kJ/mole.

In general, the higher the strength or hardness of a weld the lower is its resistance to HACC. This general relationship underpins the use of carbon equivalent formulae, which reflects the potential hardness of the base steel (or the weld metal), to predict the susceptibility to HACC and to guide welding conditions (process, heat input, pre and post-heat) to ensure a hydrogen content below the critical level (Dunne, 1999). If the weld chemistry is defined and the initial hydrogen level in the weld metal does not change over a wide range, the criterion for the formation of hydrogen cracking can be rewritten in a simple form: (9) Where have different values for different weld metal compositions and welding consumables. In the case of fixed chemistry and welding consumables this critical value can be found from a limited number of weldability tests, such as WIC tests. HEAT TRANSFER MODEL Most of the published work on heat transfer during welding processes considers that the heat source is concentrated in a very small volume of the material. After such consideration, analytical solutions are obtained assuming a point, a line or a plane heat source, as those proposed by Rosenthal in 1946. These concentrated source models are satisfactory in regions outside the weld pool. In this study we will utilize an exact solution (equation 10) of the threedimensional heat conduction due to a point heat source moving on a preheated plate of finite thickness without neglecting heat transfer at the surfaces of the plate and the welded part as shown in Fig.2 (Yurioka etal. 1986).

The equation given by Coe (1973) for hydrogen diffusion in austenite is; (3) () = 2.5 106 exp(3133()) (m2/s) It is also recognized that there may be considerable uncertainty in the diffusivity, which could have significant effects on the results in the analysis. CRITERION OF WMHACC In McParlan and Graville (1976) approach, the HACC occurs when the mean residual hydrogen concentration in weld metal, , calculated at = 500 exceeds the critical level, , or, (4) . The residual hydrogen concentration, , can be found as; (5) = exp()
0 1(=1500) 2(=50)

Where is an empirical parameter, 0 is the initial weld metal hydrogen content and is called an integration diffusion parameter, which is given by integrating the apparent hydrogen diffusion coefficient () from 1500 oC to 50 oC as; = () (6)

The integration range (t1 and t2) are the times taken for the weld pool to reach 1500oC and 50 oC from the weld start point. The diffusion coefficient is presented as a time dependent parameters as Dapp changes with temperature and temperature varies with time as the weld cools down (see eq. 2). In this approach it was assumed that the critical level of hydrogen, , exponentially decreases with the increasing carbon equivalent, , (McParlan and Graville, 1977); This approach was derived from the results of G-BOP (gapped bead-on-plate) testing that examines WMHACC. The carbon equivalent, , in the equation 7 is given as follows: = exp() (7)

Figure 2: Heat transfer during welding

Copyright 2012 by ASME

This solution consists of several parts: the first part evaluates the change of the temperature with time due to the moving heat source of intensity = ; the second part of the solution provides an estimate of the temperature variation due to preheat. Description of all variables in Eq. (10) is given in the Nomenclature Section. The following thermal constants, which are assumed to be temperature independent for the purpose of analytical modeling, are utilised for calculating the thermal field between 1500 and 50 oC (Yurioka etal., 1986): (cal/cm0Cs) 0.146 (cm2/s) 0.146 (cal/cm0Cs) 0.0005 (cal/cm0Cs) 0.0005


2 2 = + exp cos + sin 0 + 2 2 +2 cos

=0 =0

2 sin cos 1 + sin exp 2 2


Copyright 2012 by ASME

It should be noted that the thermal history is also influenced by ambient temperature and heat transfer coefficient, the latter being influenced by wind velocity, air humidity, etc. In all calculations the ambient temperature was set at 20 oC. SELECTED RESULTS Figure 3 shows the variation of the weld cooling time between 1500 oC and 50 oC as a function of the plate thickness and various values of heat input. It can be seen that the cooling time changes more intensively for thinner plates and lower heat input. 15/50 (s)

the susceptibility to hydrogen cracking. A more pronounced thickness effect takes place for higher heat input, where a small variation in thickness has a large influence on , specifically for relatively thin plates. This kind of effect is also found on t8/5 at thickness where the heat sink conditions change from bithermal to tri-thermal. It is interesting that the &() dependency have some sort of saturation. For heat input of 0.4 kJ/mm a 16 mm is a critical thickness above which the thickness effect is negligible. For example, for a higher heat inputs (0.6 0.8 kJ/mm) this critical value moves to 24-30 mm. Test results of the X80 weld samples restrained in the WIC test rig for E9010 consumables have indicated that the critical value, , is around 1 106 m2/s oC (Alam et al., 1999). The combination of welding parameters, which are likely to cause hydrogen cracking, is shown as unsafe <cr area in Fig. 4. From this figure it is seen that the risk of hydrogen cracking remains for even moderate heat input of 0.8 kJ/mm and relatively thick plates. The risk of hydrogen cracking can be effectively controlled by preheat temperature. Fig. 5 shows the change of the hydrogen cracking controlling parameter as a function of time for three situations: no preheat, 40oC and 100oC preheats. Even relatively low preheat temperatures can fully mitigate the risk of hydrogen cracking for 16 mm thick plates. 106 (m2/s 0C)

600 400 200 0

Heat input increasing


Figure 3: Variation of cooling time at the bottom of the weld from 1500 oC to 50 oC as a function of plate thickness for different heat input = 0.4, 0.6 and 0.8 kJ/mm with no preheat. The variation of the critical parameter as a function of the plate thickness and for different heat inputs is shown in Fig.4. 4 3 2 1 0 8 12 16 20 24 Plate thickness, (mm) 28 106 (m2/s 0C)

16 20 24 Plate thickness, (mm)


Preheat temperature increasing

Heat input increasing Safe region >cr Un-Safe region <cr


Figure 5: Variation of the weld temperature and with time for 0.6 kJ/mm heat input, 16 mm plate for no preheat, 600C and 1000C preheat. The effusion of the hydrogen is determined by the temperature history () and the diffusion coefficient, , which is a function of the temperature as described above. The residual concentration of hydrogen is significantly affected by the time the weld spent under relatively low temperatures; say below 200 or 300 oC. This time is largely influenced by the preheat temperatures. The latter explains a high sensitivity of the risk of hydrogen cracking to the preheat temperatures, which can fully avoid the risk of HACC in both HAZ and WM.

200 300 Time, (s)



For relatively low heat input the plate thickness does not significantly affect the critical parameter, which characterizes

Figure 4: Variation of at the bottom of the weld as a function of plate thickness for different heat input = 0.4, 0.6 and 0.8 kJ/mm with no preheat.

Copyright 2012 by ASME

However, this method is generally expensive and largely slows down the speed of construction increasing the overall cost. An alternative approach to the preheat method is a strict control of welding parameters. However, this approach needs further research in order to identify the combination of welding parameters which avoid the risk of HACC. CONCLUSIONS The majority of published work on hydrogen cracking has been devoted to hydrogen cracking in the heat affected zone of weldments, nevertheless the susceptibility of the weld metal has been often acknowledged. A similar systematic study of the phenomenon has not yet been carried out and currently no guidelines exist for effective control of HACC in WM. This is particularly important for the pipeline industry where although the current precautionary measures for avoidance HAZ hydrogen cracking in high strength pipe grades provide some confidence, the cracking increasingly initiates in the weld metal (Barbaro, 1999). In this paper we developed a mathematical model, which is based on the residual hydrogen level in weld metal. This model disregards the effect of the stress concentration on the critical hydrogen concentration and assumes a fixed chemistry and weld metal microstructure. Variations in stress can have a major effect on the cracking risk. In principle, this could be accounted for in the model by increasing the critical value of with the increase of the stress level. However, experimental investigations are required to establish and validate this sort of relationship. From the modeling analysis conducted in this paper we can draw the following conclusions: Thickness of plate/pipe has a more pronounced effect at higher heat inputs; There is a critical value of the wall thickness above which the thickness effect on the risk of hydrogen cracking is negligible, this is important for qualification welding procedures; Results for X80 and heat input of ~ 0.5 kJ/mm indicate a high risk of hydrogen cracking even for relatively thin plates/pipes (around 8-10 mm); The hydrogen assisted cold cracking can be effectively controlled by preheat. It is also recognized that this study represents an initial step towards a more comprehensive analysis, which can be only possible if this will be validated with a support of experimental studies. ACKNOWLEDGMENTS This work is funded by the Energy Pipelines CRC, supported through the Australian Government Cooperative Research Centers Program. The cash and in-kind support from the APIA through its Research and Standards Committee is also gratefully acknowledged. We are grateful to Prof. V. Linton and

Dr. F. Barbaro for useful discussion and constructive comments.

REFERENCES Alam, N., Dunne, D., Squires, I., Barbaro, F. and Feng B. (1996) Weldment cold cracking - the effect of hydrogen and other factors, in: Proceedings of Joint Seminar Hydrogen Management in Steel Weldments, Melbourne, Australia, WTIA1996, 49-60. Barbaro, F. (1999) "Types of hydrogen cracking in pipeline girth welds", in: First International Conference on Weld Metal Hydrogen Cracking in Pipeline Girth Welds, Wollongong, Australia, 1999, WTIA Coe, F.R. (1973) Welding steels without hydrogen cracking, TWl, Abington. Davidson, J.L. (1995) Hydrogen induced cracking of low carbon-low alloy steel weldments. Met. Forum 19: 35-51. Dunne, D. (1999) A review of theoretical and experimental background of hydrogen assisted cold cracking of steel weldments, in: First International Conference on Weld Metal Hydrogen Cracking in Pipeline Girth Welds, Wollongong, Australia, 1999, WTIA. Glover, A. and Graville, B. (1999) The risk of hydrogen cracking in multi-pass welds and its effect upon procedure design. in: First International Conference on Weld Metal Hydrogen Cracking in Pipeline Girth Welds, Wollongong, Australia, 1999, WTIA. Hirth, J. P. (1978) Hydrogen adsorption at dislocations and cracks in Fe. Acta Metallurgica 26: 1795-1803 Lindgren, L.E. (2006) Computational modelling of welding. Computer Methods in Applied Mechanics and Engineering Vol.195(48-49): 6710-6736. McParlan, M. and Graville, B.A. (1976) Hydrogen cracking in weld metal. Welding Journal, Vol. 55(4), pp. 95s102s. Oriani, R. A. (1970) The diffusion and trapping of hydrogen in steel. Acta Metallurgica 18(1):147-157. Petch, N. J. (1952) Delayed fracture of metals under static load. Nature 169 (4307) Rosenthal, D. (1946) The theory of moving heat sources and its application to metal treatments. Trans. Am. Soc. Mech. Engrs. 68: 849-866. Williams, J.G. (2007). Advances in steels for high strength ERW spplication in Australia. Materials Forum 31: 1-10. Yurioka, N, Okumura, M., Kasuya, T. and Ohshita, S. (1986) Welding note. Second Edition. P2-47-86, Nippon Steel, Kanagawa, Japan. Yurioka, N. and Suzuki, H. (1990) Hydrogen assisted cracking in C-Mn and low alloy steel weldments. International Materials Reviews 35 (4): 217 248. Yurioka, N. (1999) Predictive methods for prevention and control of hydrogen assisted cold cracking", in: First International Conference on Weld Metal Hydrogen Cracking in Pipeline Girth Welds, Wollongong, Australia, 1999, WTI

Copyright 2012 by ASME