P L Moore 1,2*, E R Wallach 1 and D S Howse 2
1 University of Cambridge
2 TWI Ltd
Paper presented at 2nd Materials Research Conference organised by the Younger Members Committee of the IOM 3, London, 17-18 February 2003
Laser processes are being investigated for welding steel pipe sections, to replace the metal active gas (MAG) welding process currently used. Weld profiles and microstructures from both autogenous high power laser welds, and hybrid laser/arc welds have been compared. Reliable mechanical properties of welded structures can be obtained by understanding how to produce favourable microstructures in laser and hybrid welds. Weld microstructure is influenced by both the cooling profile and the weld metal composition. Hence, as part of the microstructural analysis, cooling curves were taken from the molten pool during both laser and hybrid welding using a thermocouple-harpooning temperature measurement method. By investigating laser weld microstructures in a number of steels of different compositions, it was shown that the optimal weld microstructure is acicular ferrite, since its small effective grain size results in high strength and toughness. This microstructure could be obtained in laser/arc hybrid welds using the MAG process with metal-cored wire. In this way, pipeline steels that previously gave unacceptable quality autogenous laser welds could be hybrid welded to meet current pipeline welding standards.
Currently, when welding steel pipe sections, about six arc welding passes are needed to complete a joint in 10-15 mm thick pipe. However, laser welding can replace several arc weld passes with one single pass as a consequence of its high power density which results in a deep and narrow profile. The formation of deep and narrow welds with low distortion makes laser welding increasingly attractive for industrial thick section applications. Currently, Nd:YAG lasers are commercially available up to powers of 6 kW, and steels up to 16 mm can be welded in one pass by combining beams from more than one Nd:YAG laser. The wavelength of the Nd:YAG laser (1.06 µm) is small enough to allow optic fibre delivery of the laser beam, essential for welding a variety of geometries including the circular geometry of pipe girth welds.
Unfortunately the narrow profile and low overall heat input in autogenous laser welds results in rapid cooling of the molten weld pool and the subsequent possible formation of hard martensitic/bainitic microstructures with corresponding low toughness. Centreline cracking is also common in welds with deep and narrow geometries, particularly laser welds.  Potential improvements to the laser welding process have come from the development of a laser and arc hybrid process, whereby the laser and arc are combined into the same molten pool. Hybrid welding has a greater tolerance to joint fit-up and allows faster welding speeds. The increased heat input in hybrid welding retards the cooling rate, and the filler wire, which is inherent to the arc process, can alter the composition of the weld metal and hence its resulting microstructure. In this way, acicular ferrite microstructures, which have good strength and toughness could be formed in hybrid welds in pipeline steels.
Mathematical models have the potential to predict weld microstructures and properties using as input only the welding parameters and weld metal composition. Comparing experimental cooling curves to those predicted from mathematical cooling equations, which are dependent on knowledge of the welding parameters, allows predictions of the coupling efficiency of the welding process to be made. Hence the models can predict the thermal profiles for solidifying weld metal, and link the resulting cooling rates, especially between 800°C and 500°C, and steel composition to provide predictions of the weld and heat affected zone (HAZ) microstructures and properties.
2. Welding trials
2.1. Autogenous Laser Welding
Fig.1. Diagram of laser keyhole welding
Seven different carbon-manganese structural steels (labelled A-G) were Nd:YAG laser welded under the same conditions using a power of 5.2 kW at a focus depth of -3 mm, and a travel speed of 0.3 m/min. The laser power was obtained by combining the beams of 3 kW and 4 kW Nd:YAG lasers. In this way, since the heat at the workpiece and hence cooling rate of the weld metal were identical for all seven steels, the differences in the microstructure and properties of the welds could be attributed to the compositional variations.  Hardness tests and Charpy impact tests were done on the laser welds from all seven steels. Radiography was used to determine the extent of cracking and porosity in the welds.
Most of the laser welds showed a microstructure of predominantly ferrite with an aligned second phase (FAS). This term covers a range of different microstructures that contain aligned sheaves of ferrite laths, separated by a second phase of carbides or retained austenite. These weld microstructures, which were mainly ferrite with aligned second phase, also contained Widmanstätten ferrite and bainite, but not martensite. This type of microstructure is undesirable in welds since the large colony size and oriented laths offer little resistance to crack propagation. However, two of the steels (B & G) formed acicular ferrite weld microstructures that also contained some grain boundary ferrite. Acicular ferrite microstructures consist of small particles of intragranularly nucleated ferrite, with no particular orientation relation. The resulting small effective grain size and lack of orientation, makes acicular ferrite a desirable weld microstructure with excellent mechanical properties.
The best overall weld performance was from steel G, a laboratory cast of a patented steel, designed for laser weldability.  Its acicular ferrite microstructure resulted in a low hardness overmatching between weld and parent of only 1.04. Charpy impact tests on this weld gave an average impact energy of 110 J at -10°C, easily passing the acceptance criterion of an average of 35 J minimum.  The laser weld in this steel also had low porosity (0.2%) and no cracking was observed.
The worst properties were in steel A, a commercially available X60 pipeline steel. This weld's microstructure was predominantly bainitic with some ferrite with aligned second phase, and there was a hardness overmatching of 1.33 between the weld and parent metal. Its average Charpy impact value was only 10 J at -10°C. This laser weld also showed considerable solidification cracking (47 mm in 100 mm of weld), which is unacceptable.
Fig.2. Weld in steel G showing an acicular ferrite microstructure, with some grain boundary ferrite and Widmanstätten ferrite
Fig.3. Weld in steel A showing a bainitic and FAS microstructure
2.1.2. Effect of Composition
Solidification cracking is caused by the formation of low melting point films, such as FeS, along the centreline of a weld where the solidification fronts meet.  Adding sufficient manganese will tie up the sulphur and phosphorous as inclusions, e.g. as inclusions instead of low melting point products. The least cracking was found in the steels with the most manganese (over 1.4 wt%), while the most solidification cracking was seen in those steels with the highest phosphorous and nickel contents (over 0.01 wt% and 0.5 wt% respectively). The presence of nickel in the weld suppresses ferrite formation and forms a relatively hard and brittle weld metal, making it more prone to cracking under restraint. Neither of the steels that had acicular ferrite microstructures (B & G) showed any cracking. The steels with higher contents of calcium, silicon and aluminium tended to have the most porosity. These elements alter the flow patterns of the molten pool, and restricting calcium to 20 ppm and silicon to 0.3 wt% seemed to help reduce weld porosity. 
Acicular ferrite formation was promoted by having low levels of aluminium and oxygen, but with titanium additions. Steel G contained 40 ppm aluminium, 11 ppm oxygen and 0.012 wt% titanium, and Steel B had the lowest aluminium content of the commercially available steels (0.027 wt%), and the highest titanium content (0.022 wt%) with 21 ppm oxygen. The Charpy and hardness tests done on this set of welds confirm the mechanical properties advantage of having acicular ferrite present in a weld microstructure.
2.2. Hybrid Laser/Arc Welding
Hybrid processes, whereby an arc and a laser process are combined, are becoming increasingly attractive in industrial fabrication for a variety of reasons. When combining the two processes, it is anticipated that the benefits of each individual process are obtained. For laser/arc hybrid welding, it is desired that the fast travel speed, deep penetration and low distortion of the laser weld is combined with the fit-up tolerance, filler addition control and resulting good mechanical properties of the arc process. The growth of interest in hybrid laser/arc processes has come from the increasing use of industrial lasers in fabrication shops; as lasers are more widely used, so they can be adapted to new applications and processes.
Fig.4. Diagram of the hybrid laser/arc welding set up
Steel A (which performed worst when Nd:YAG laser welded) was used to investigate the potential benefits of a hybrid laser/arc welding process, since its laser welds had the greatest scope for improvement. A 6 kW Nd:YAG laser beam (again, from beam combination of two individual lasers), and a MAG arc were combined in the same molten pool. The laser and MAG torch were kept at a fixed angle, with zero process separation and MAG leading. The hybrid welding travel speed was increased to 1 m/min, in order to match the joint completion rates that are achieved when laying pipelines in the field. Bead-on-plate welds were produced for comparisons of the process.
The effect of combining the two processes is illustrated in Fig.5, which shows the macro profiles of the separate arc weld and laser weld, compared to the resulting hybrid weld. The MAG weld does not penetrate into the plate, and has a high weld metal peak due to the filler wire, however this macro is not typical of a MAG weld since the arc parameters have been optimised for laser/arc hybrid, not MAG welding. The laser weld penetrates deeply, but has a narrow profile and no filler addition to control the weld microstructure. The hybrid weld shows a similar penetration to the laser weld, but also has the peaked surface profile from the filler wire addition.
Fig.5. Macro profiles of: a) MAG only b) laser only; and c) the resulting hybrid weld profile. Scale is in mm
2.2.1. Microstructure & Properties Improvement
Initially, hybrid welds were made using solid filler wire, which resulted in a microstructure of ferrite with an aligned second phase (FAS) and some intragranularly nucleated phases, such as acicular ferrite, in the weld. However, in order to improve the arc stability, a metal-cored MAG wire was used instead. Metal-cored wire also results in a higher oxygen content in the molten pool, and therefore increases the number of inclusions which promote acicular ferrite formation. The hybrid welds made using the metal-cored wire showed a more predominantly intragranular microstructure, with a greater amount of acicular ferrite.
The improvement in the microstructure that is possible by developing the welding process from autogenous laser welding to metal cored hybrid welding, for the same steel, is illustrated in Figs.6 & 7. The first microstructure is from a 5.2 kW autogenous Nd:YAG laser bead-on-plate weld, made at 0.6 m/min on 16 mm thick plate. The microstructure is predominantly ferrite with aligned second phase, and the weld hardness is 312 HV, corresponding to an overmatch ratio of 1.6 between the parent and weld metal. The second microstructure is from a hybrid weld, made using 6 kW Nd:YAG laser power and 200 amps MAG arc power at 1 m/min, in an 11 mm thick joint. This microstructure is predominantly acicular ferrite & grain boundary ferrite. Its weld hardness is now 275 HV, corresponding to an overmatch 1.4.
Acceptance criteria  give a maximum allowable weld hardness of 275 HV for pipeline welds, so the steel that originally failed to meet the hardness and toughness requirements when laser welded, shows an acceptable hardness when hybrid welded. In addition, the hybrid weld has been made at a faster speed than the laser weld (increased from 0.6 m/min to 1 m/min) and still has the better microstructure and hence properties.
Fig.6. Microstructure of autogenous Nd:YAG laser weld
Fig.7. Microstructure of laser/MAG hybrid weld using metal-cored filler wire
2.2.2. Molten Pool Mixing
When two very different welding processes such as arc welding and laser welding are combined, it is often uncertain how they will affect each other. One important aspect is the mixing of the filler wire within the hybrid molten pool. The laser penetrates to form a deep, narrow molten pool, but the filler wire added with the arc will not necessarily mix thoroughly all the way down the penetration finger before solidification occurs. This would result in different properties and microstructures occurring at different places in the weld. One way of improving the homogeneity in the weld is to use an edge preparation at the joint, instead of the closed butt joint simulated when welding bead-on-plate. When a narrow bevel joint design was used (shown in Fig.8a), the hybrid weld cross section became more parallel sided (Fig.8b) instead of the nail shaped profile shown in Fig.5c. The only problem with this joint design was the tendency to get small centreline cracks near the surface. These seemed to be caused by the constraint at the top of the molten pool, however, by using a compound bevel to open out the angle of the joint at the surface of the weld, this could be remedied  (Fig.8c). This adapted joint design also improved the profile of the hybrid weld.
Fig.8a) diagram of joint edge preparation, b) hybrid weld made using this joint, and c) hybrid weld on opened-out joint, now without cracking. Scale is in mm
Filler wire mixing in a bead-in-groove weld, such as that shown in Fig.8b, was investigated using a filler wire with a higher nickel content that the parent metal; the 1 mm diameter solid wire contained 2.4 wt% Ni compared to 0.02 wt% Ni in steel A. This difference in nickel content was sufficiently high to allow x-ray analysis of the weld metal using a scanning electron microscope with EDX detection. EDX analysis positions were chosen at the top above the small centreline crack, below the crack and down the broad portion in the middle of the weld, and the middle where the weld narrowed. The results are given in Table 1, and the error in the values is ±0.11 wt%. It can be seen that there is a marked reduction in the measured nickel content in the weld between the broad top part of the weld and the narrower region further down. Although, within the limits of experimental error, the apparently lower nickel content at the very top of the weld could be due to some loss to the atmosphere during welding, since it was localised to quite close to the surface. The lower nickel content at the bottom of the weld, only a third of that further up, is due to incomplete mixing of the filler wire in the molten pool before solidification occurs. However, it is still sixteen times the nickel content of the parent metal, so there is some filler wire mixing throughout the hybrid weld, even though it does not mix completely.
Table 1: EDX analysis results of nickel tracer hybrid weld
|Weld location||Nickel content|
|Middle-narrowest part of weld
3. Modelling of laser & hybrid welds
3.1. Temperature Measurement
A temperature measurement method was needed that could generate accurate cooling data, in order to obtain weld cooling profiles between 800°C and 500°C. This cooling time can be related to the resulting weld microstructure and properties in steels. If this critical cooling time is known, it is possible to compare existing predictive cooling models to the experimental data, to determine the coupling efficiency of the welding process. By comparing the coupling efficiency for laser and hybrid welds it is possible to see if the coupling is improved when using the hybrid welding process. In laser welds it is both the weld and the HAZ metal that determine the overall properties of the welded joint (in arc welds the HAZ is more significant since its composition cannot be altered, like the weld metal), so cooling curves are needed from the weld itself, rather than extrapolating from HAZ measurements. For this reason thermocouple harpooning was proposed, and it proved to be a successful temperature measurement method for these welds. In thermocouple harpooning the thermocouple tip is embedded into the molten pool behind the laser impingement point by means of a plunger device. Since the thermocouple is fused into the weld bead, it is possible to have confidence that the weld (and not the HAZ) cooling profile has been recorded. This method was used on both CO2 and Nd:YAG lasers, as well as Nd:YAG/MAG hybrid welds, which are all novel applications of this technique.
Cooling data were taken from the MAG-only, laser-only and hybrid welds in Fig.5, and the cooling curves are shown in Fig.12. The MAG-only weld had a heat input of 215 kJ/m, the laser-only weld's heat input was 281 kJ/m; these heat input values were calculated from the laser and arc powers, scaled by appropriate efficiency factors. The hybrid weld was made by directly adding these two welding sources and hence had a heat input of 496 kJ/m. It took 1.3 seconds for the MAG weld to cool from 800°C to 500°C, and this resulted in a fine microstructure of predominantly acicular ferrite, with some grain boundary ferrite. Although the cooling was rapid, the weld cross sectional area is small and the filler wire provides plenty of inclusions on which to nucleate acicular ferrite.
Fig.12. Cooling curves from hybrid, MAG and laser welds
The laser-only weld cooled from 800°C to 500°C in 1.3 s, the same as the MAG arc weld. However the laser weld had a microstructure of predominantly ferrite with an aligned second phase, and grains oriented perpendicular to the weld finger. This occurred due to the deep and narrow profile of the laser weld. Comparing the cross-sectional area of this weld to the arc weld, its fusion zone is about twice the area of that of the arc weld, so the heat has dissipated from a larger area in about the same time. This has caused the aligned ferrite type microstructure to form in this weld. When hybrid welding bead-on-plate, the cooling rate increased to 2.9 s, almost in direct proportion to the increase in heat input when the arc and laser are combined. The microstructure contained intragranularly nucleated phases, including acicular ferrite, due to the MAG filler wire, but had more ferrite with an aligned second phase further down the weld, similar to the laser-only weld. Fig.13 shows a schematic diagram of the heat flow direction in each of these welds; the strength of the arrow indicates the amount of heat that flows per second in each case. The differences in heat flow patterns means that existing mathematical models for thermal profiles in laser welds (usually two-dimensional) might not be appropriate to make the same predictions in hybrid laser/arc welds.
Fig.13. Diagram of heat flow amount and direction during cooling of a) MAG weld b) laser weld; and c) hybrid weld
3.2. Coupling Efficiency Prediction
Comparing the cooling curves obtained from the laser and hybrid welds to predictive equations allows values of coupling efficiency to be predicted. The coupling efficiency was determined from the empirical cooling equation for 3-dimensional heat flow in thick plate, shown below  and where Δt is the cooling time between 800°C and 500°C, T0 is the initial temperature in °C. Q is the input power (W), ν is the welding travel speed (cm/s) and A is the coupling efficiency for the process. By putting in all the welding parameters with an assumed coupling efficiency, A, of 1, and then changing the value of A until the predicted cooling time matched that measured from the cooling curve, a value of the coupling efficiency can be determined.
Using this method, cooling data taken from seven Nd:YAG laser welds gave a mean coupling efficiency of 0.72 with 0.19 standard deviation. Only two hybrid laser weld cooling curves were recorded, but the mean coupling efficiency was 0.78 with standard deviation of 0.12. Thus the coupling efficiency has increased slightly using the hybrid welding process instead of laser welding. Not only does this suggest that the hybrid process is slightly more efficient than laser welding, but provides a value of the coupling efficiency, A, that can be used to calculate the heat input, H, for a weld, using only the welding parameters Q, the laser (or combined laser and arc) power and ν , the welding speed ( H = AQ/ ν). When mathematically modelling heat flow in welding, accurate input data are needed in order to provide high quality results.
3.3. Mathematical Modelling
A relatively simple FORTRAN model 
was adapted to predict the cooling time, hardness and microstructure of laser welds in thick section (>10 mm) pipeline steel. This model was run using input welding parameters (composition, plate thickness, welding speed and laser power) matching those of welds made previously in this work. Values of coupling efficiency were used, determined for each laser process using experimental cooling data, as shown above. Twelve different sets of welding process and parameters were used as input data, including CO2
laser welds, Nd:YAG laser welds and hybrid welds, and the output compared to experimental results. In each case, the model's output gave a value of the time to cool between 800°C and 500°C, the HAZ hardness, and the relative amounts of microstructural phases.
The cooling time predictions were reasonably good, five of the model's results were within 5% of the experimental value, and only three were over 30% different from the experimental results. The HAZ hardness predictions from the model were even better, given the amount of scatter that is found in experimentally measured values of hardness. None of the predictions was more than 20% different to the experimental value, and five predictions were within 3% of the measured results. The use of this model for microstructural predictions is still in progress. The most desirable microstructural component in laser and hybrid welds is acicular ferrite, but mathematical predictions of its formation are non-trivial, and it is currently not included as one of the microstructural phases that can be predicted in this model. However, the model has proved to be a useful basis for predicting the properties of laser and hybrid welds.
The quality and properties of laser welds in pipeline steels can be improved significantly by using a hybrid laser/arc welding process, and selecting the appropriate consumables. The weld microstructure can be altered from a bainitic or aligned ferrite type microstructure to acicular ferrite by using metal-cored MAG wire in the laser/arc hybrid process.
Cooling curves taken from the molten pool of laser and laser/arc hybrid welds have been used to determine that the coupling efficiency of Nd:YAG laser welds is 72%, and for Nd:YAG laser/MAG hybrid welding has increased to 78%. Using these values in a mathematical model of laser welding allowed accurate predictions of HAZ hardness to be made, with the potential for microstructural predictions in future.
This work forms part of a larger collaborative project at TWI. Thanks to BP for providing the materials, and to EPSRC and the DTI for funding this work.
- J. D. Russell. 'Laser weldability of C-Mn steels', Assessment of Power Beam Welds, European Symposium, Geesthacht, Germany. February 1999.
- P. L. Moore, D. S. Howse and E. R. Wallach. 'Microstructure and properties of autogenous high power Nd:YAG laser welds in C-Mn steels', 6th International Conference on Trendsin Welding Research, Pine Mountain, Georgia, USA. April 2002.
- D. Senogles. 'A steel composition for laser welding'. UK Patent Application 2341613 A. 1998.
- TRANSCO P2. 'Welding of land pipelines and installations designed to operate at pressures greater than 7 bar (incorporating BS 4515)'. 1997.
- R. J. Scudamore, A. C. Woloszyn, D. S. Howse and G. S. Booth. 'Phase IV hybrid welding results for February - April 2002', TWI Internal Report 12940/34/02. April 2002.
- J. Degenkolbe, D. Uwer and H. G. Wegman. 'Characterisation of weld thermal cycles with regard to their effect on the mechanical properties of welded joints by the cooling time t 8/5 and its determination'. Report 91/79918 BSi. September 1991.
- S. Jackson & E. R. Wallach. 'Laser welding of steels for lightweight vehicles, LIVEMAN innovative manufacturing initiative', University of Cambridge. September 1999.