Testing the waters - shipbuilding laser welds face toughness trials

TWI Bulletin, September/October 2000

After a degree from Cambridge University and a doctorate from Sheffield University, Isabel Hadley worked as a materials/structural integrity engineer in a range of industries, including nuclear power, offshore engineering and steel research. She joined TWI in 1992, working mainly on assessment of fitness-for-purpose of critical welded structures. Isabel is currently manager of the Defect Assessment and Reliability Engineering section of TWI's Structural Integrity Technology Group.

Laser welding has many attractive features, in particular the potential for high joining rate and low distortion. Consequently, as Isabel Hadley discusses here, there is considerable interest in this joining technique in the shipbuilding industry, which has hitherto relied on more conventional arc welding processes. One problem which has been identified in qualifying the weld, however, is that of measuring weld metal toughness. The conventional Charpy test, although widely accepted as a quality control test for arc welds, is unsuitable for laser welds, since the path taken by the propagating fracture tends to deviate away from the very hard, narrow weld zone into the softer parent material (see Fig.1). This phenomenon, known as fracture path deviation (FPD), is increasingly predominant as the test temperature is increased. The measurement of absorbed energy in a specimen failing by FPD is thus not truly representative of the weld metal toughness, since both weld metal and parent metal constituents contribute to the failure event, and the fracture path is not in the plane of the initial notch.

Fracture path deviation was identified over 20 years ago; it is known to be promoted by a high 'overmatch' (ratio of weld metal hardness to parent metal hardness), narrow weld zone width and the relatively blunt notch tip associated with the Charpy specimen. Goldak and Nguyen (1977) proposed that FPD was related to the formation of a plastic hinge in the softer parent material.

Several solutions have been proposed to the problem outlined above. One is to avoid Charpy testing altogether and to demonstrate the structural integrity of laser welds by large-scale testing. A second is to use fracture mechanics tests in place of Charpy tests, on the assumption that the sharper crack tip produced in a fracture mechanics specimen will not deviate into parent material. These are technically feasible solutions, but not realistic for routine production testing. A third possibility would be to reduce hardness overmatch and/or to increase weld zone width by altering the process parameters or adding filler metal. However, this may also have unwanted effects, such as sub-optimal production rates and/or an undesirable weld metal microstructure. A fourth alternative, which is the one explored in this work, is to modify the Charpy test sufficiently to avoid FPD, and to correlate the test results with those of fracture mechanics and/or structural tests, so that a realistic level of Charpy energy may be specified for laser welds.

TWI recently participated in a European collaborative project to develop the use of laser welding in shipbuilding (TWI, 1996). The project explored a number of these issues surrounding testing of laser welds and identified several possible modifications to the Charpy test. These are shown in Table 1, along with other ideas from the literature and from the more detailed research programme subsequently carried out by TWI and described in this article.

Table 1: Possible techniques for avoidance of FPD

The objective of TWI's programme was to recommend a test method (based on the Charpy test) for routine testing of 5-12mm thick laser welds in the shipbuilding industry. The steel investigated was a 12mm thick C-Mn steel plate with a composition particularly suitable for laser welding ( Table 2). Attention was restricted to the weld zone in the first instance.

Table 2: Composition of material tested

Method

Autogenous butt welds were made in 12mm thick plate using TWI's CO ₂ laser, with a power output of around 10kW and two different travel speeds - 1 and 0.5m/min. The different heat inputs (and thus cooling rates) produced two distinct weld metal microstructures and bead shapes, as shown in Fig.2. The weld produced using the higher travel speed shows a narrow-profile weld with a very high hardness overmatch. Clearly, a high travel speed is desirable for high production rates, but experience suggests that the weld formed is more likely to show FPD during Charpy testing. Welds were designated N (narrow) and W (wide). The microstructure of the narrow weld consisted mainly of a fine autotempered martensite, whilst that of the wider weld consisted mainly of bainite. 

Charpy specimens and modified specimens were then made from the welds as follows, and tested over a range of temperatures to produce full ductile-brittle transition curves and a range of behaviour from failure in the weld zone to failure by FPD. Where necessary, tests were carried out at elevated temperatures, up to 100°C. Specimen types were as follows:

• 'control' specimens; two sets of Charpy specimens were made in the usual way, with one set notched at the weld centreline and the other in the parent metal.
• 'side-welded' specimens; melt runs were made close to the edges of the two plates to be joined. The plates were then machined back, close to the melt run, and butt-welded from the opposite side ( Fig.3).
• 'side-grooved' specimens; full-thickness specimens were side-grooved to a depth of 1mm each side ( Fig.4).
• 'transverse-notched' specimens; the Charpy specimen was prepared with its long axis parallel to the weld and notched so that the notch tip sampled the edge of the fused zone ( Fig.5) - weld N only.
• 'narrow notch' specimens; these were notched to a depth of 2mm using a 0.15mm thick diamond saw in place of the usual 2mm deep 45° broach.
• 'fatigue cracked' specimens; these were notched with a fine saw up to a depth of 1mm, then fatigue-loaded until the crack length had reached a total of 2mm.
• 'hardened' specimens; the plate material to be joined was given an austenitising heat treatment before welding in the usual way - weld N only.

The above tests were all carried out on 10mm thick specimens (notched to a minimum thickness of 8mm in the case of the side-grooved specimens). Additional tests were also carried out on sub-size (5mm thick) specimens as follows:

• 'control' specimens; the 12mm plate was machined down to 6mm before welding, using the same welding parameters as for weld N. Consequently the weld width was a little higher than for weld N.
• 'hardened' specimens; an austenitising heat treatment was carried out on the 6mm plate before welding to produce a weld similar to the weld described above.

In addition, fracture mechanics tests were carried out over a range of temperatures on single-pass butt welds notched at the weld centreline, to determine critical values of CTOD and J at fracture. These were carried out in accordance with BS 7448: Part 2 (BSI, 1997), although modifications were made to the test method to account for the high strength mismatch between parent metal and weld zone. These modifications were in accordance with the draft weldment testing Annex to ASTM E1290 (ASTM, 1998).

Results

Basic weld properties

The basic properties of the welds in terms of weld width, the hardness of the weld zone and the hardness of the parent material, are shown in Table 3. Also shown is the ratio of weld zone to parent metal hardness, designated M (mismatch ratio). The following features can be seen:

• weld N has a high mismatch ratio (M = 2.4) in the as-welded condition, and a weld width at the mid-section of the plate of approx. 1mm.
  
  
• weld W has a lower mismatch ratio (M = 1.7) and wider weld zone (approx. 2mm) due to the slower cooling rate of the weld.
  
  
• hardening of the parent material is more effective in the 6mm thick plate compared with the original 12mm thick plate
  

Table 3: Basic properties of welds made under this project

Charpy energy

Table 4: Results of Charpy tests

Looking first at Weld N, which is more representative of the type of weld which would be considered desirable in the shipbuilding industry, it can be seen that most of the modifications had some effect, that is that they shifted the temperature at which FPD occurs from below -160 to -80°C in some cases. Nevertheless, Charpy testing for weld qualification is usually carried out at temperatures between -50 and 0°C, so none of the techniques were successful in shifting the FPD temperature sufficiently that FPD could be avoided during routine testing. The exception to this was the set of transverse-notched specimens which cannot show FPD by virtue of their design. Nevertheless, in view of the difficulties in placing the notch accurately at the edge of the weld, and in interpretation of the results, the transverse-notched specimen is unlikely to replace the other techniques.

Weld W, on the other hand, showed FPD at -30°C and above in the unmodified specimen. Modifications such as side-grooving, using a slit or fatigue crack in place of a notch, and use of a supporting side weld all succeeded in shifting T _FPD to a temperature above the normal Charpy test temperature for qualification tests.

Results of all Charpy tests on full-size specimens are shown in the form of transition curves in Figs.6-7 (Charpy energy is shown in terms of J/cm ² to allow for the decreased fracture area of side-grooved specimens). The transition curve for the parent material is also shown for comparison. Figure 6 (results from the narrow weld) shows that the values of Charpy energy recorded when a specimen fails by FPD are clustered around the transition curve for the parent material, reflecting its properties rather than those of the weld metal. In contrast, Fig.7 (results from the wider weld) shows a clear distinction between the transition curve for parent metal and those for weld metal, commensurate with the fact that the fracture occurred wholly in the weld metal in most cases. Whilst this undoubtedly makes the task of testing Weld W easier to interpret, the actual values of Charpy energy in Weld W at temperatures below the FPD temperature are below the requirements usually stipulated for welds in ships destined for low temperature service. For example, if the Charpy energy of the weld metal is required to meet the same requirements as those for a grade EH steel (47J or 59J/cm ² at -20°C), as implied by recent guidelines on laser welding (Lloyd's Register, 1997), then none of the weld W specimens would pass this criterion.

Fracture toughness

Continued in Part II
See also Charpy testing laser welds - the significance of fracture path deviation