Strain ageing of weld metals tensile test behaviour

TWI Bulletin, July/August 1988

Strain ageing behaviour of steels in general has always been demonstrated most dramatically by the jerky flow or serrated yielding seen in tensile tests at moderately elevated temperatures. Before starting a current project on the significance of strain ageing behaviour of weld metals likely to be used for the as-welded repair of pressure vessels, some preliminary tests were undertaken to see whether tensile testing could be used as a simple sorting test to assess the relative susceptibility to strain ageing of different consumables.

When strain ageing behaviour is observed during elevated temperature testing, i.e. when straining and ageing are simultaneous, it is termed dynamic strain ageing. Susceptible steels can be embrittled either by such simultaneous straining and ageing or by ambient temperature straining followed by subsequent ageing. In either case toughness is impaired at ambient temperature and below. In terms of welds, such so-called static strain ageing used to be a serious problem when semi-killed steels were common, as it led to poor toughness in the sub-critical heat affected zone (HAZ). Indeed, the original form of the wide plate test was notched into this region because it was usually the least tough region of the HAZ. ^[4]

Modern steels usually contain sufficient aluminium so that in most heat treated conditions the nitrogen is combined as aluminium nitride (AlN, which contains about twice as much aluminium as nitrogen, by weight). Hence strain ageing of the steels after welding is nowadays not a serious problem. Unfortunately, because excessive amounts of aluminium can prevent achieving tough, as-welded microstructures in many systems, ^[5] most weld metals are low in aluminium and are thus potentially susceptible to strain ageing.

Indeed, strain ageing is one of the reasons why toughness of the root runs of multipass welds can be low, particularly if the welds are not strongly restrained to prevent the root regions deforming while they are still warm. ^[6] However, other factors need to be taken into account. One is that, in multipass welds, some nitrogen is likely to be precipitated as nitrides by the complex heat treatment cycle imposed by subsequent runs. Another factor is that weld metals contain a much higher dislocation density than plate steels, ^[7] so that more nitrogen might well be needed to produce serious embrittlement than the 10ppm or so needed in a plate steel. ^[1,2]

Table 1 Welding conditions for test welds

The purpose of the project was to examine the tensile behaviour of two typical manual metal arc welds at moderately elevated temperatures. This was done to establish whether there were significant changes in tensile properties because of dynamic strain ageing; whether the weld metals of different types (Mn-Ni-Mo and C-Mn) showed the maximum response to strain ageing at the same temperature; and finally how behaviour was influenced by the strain rate during the tensile tests. Reports ^[1,2] indicate that the lower the strain rate, the lower the temperature showing the maximum effect.

Most work was concentrated on a low strain rate, as this was more relevant to likely strain rates in pressure vessels when they are operating at moderately elevated temperatures and require good upper shelf toughness, and consequently resistance to dynamic strain ageing, to avoid failure under certain fault conditions. The project was, in part, aimed at providing preliminary data for a programme* on the strain ageing behaviour of as-welded repair weld metal for pressure vessels.

*TWI Group Sponsored Project 5555, 'Strain ageing embrittlement of repair weld metals for service up to 350°C'.

Experimental

As-welded repair welds, which had been deposited in deep grooves in thick plate using relatively low heat inputs, provided appropriate test material. Details of the welds are summarised in Table 1 and the weld metal compositions in Table 2. As the welds had been deposited with well dried basic electrodes and had been held overnight at the preheat temperature of 150°C after welding, no further steps were taken to reduce the weld hydrogen level. The ductility subsequently measured in tests at ambient temperature showed no evidence of low values and fisheyes were not observed on the fracture surfaces.

Table 2 Chemical analyses of test welds

Analyses on final, low dilution weld runs.

Longitudinal tensile specimens of approximately 30mm gauge length and 6mm gauge diameter were machined from the test welds and tested at a strain rate of either 0.5 or 5 mm/min (equivalent to 2.6 x 10 -4 or 2.6 x 10 -3 strain/sec) over a range of temperatures from 20 to 260°C. Yield or 0.2% proof stresses and tensile strengths were determined from the load/extension traces of the testing machine; elongation and reduction of area (RA) were measured directly. All fracture surfaces were examined at low magnification for defects. The position of the fracture was also noted, as it is known that fracture near the end of the gauge length can lead to lower ductility values than normal.

Results

Table 3 Results of tensile tests (specimen diameter 5.8mm, gauge length 31mm)

* Broke near end of gauge length.
x Starting beyond 0.2% PS.
+ With an intermediate smooth portion.

At 140 and 160°C ( Fig.1c), behaviour was quite different. After the first yield, the load dropped sharply and recovered many times until fracture ensued. In these tests, the tensile strength, but not the yield stress, was distinctly higher than at ambient, although RA values were somewhat lower. At 180°C ( Fig.1d), this so-called zig-zag, or irregular, yielding occurred only between yield and maximum load. At 200°C ( Fig.1e) multiple yielding only occurred up to maximum load from a strain roughly half way between maximum load and the 0.2% proof stress (no initial yielding being observed). The load drops were larger than at lower temperatures, the greatest being as much as 55 N/mm 2. At 220 ( Fig.1f) and 250°C, the weld metal did not yield, although tensile strength values were still higher and ductility values lower than at ambient.

With the higher straining rate, no new features were observed but the changes in behaviour occurred at a temperature about 40°C higher, i.e. irregular yielding up to 160 instead of 120°C and multiple yielding up to 240 instead of 200°C.

Few C-Mn weld metal specimens were available for tests, which were carried out at the low strain rate ( Table 3). Zig-zag yielding was found at 120°C, i.e. about 20°C lower than for the Mn-Ni-Mo weld metal, but this did not carry 0n beyond maximum load at any temperature and was only just evident at 200°C. Particularly low ductility values with this weld were in part caused by a tendency for fracture to occur near the end of the gauge length.

Because of scatter, trends of tensile properties with temperature ( Fig.2) should be interpreted cautiously. After initially weakening slowly with increasing temperature, the Mn-Ni-Mo specimens show increases in both tensile strength and yield or, more correctly, proof stress, the increase in proof stress being smaller and extending over a shorter temperature range. The higher strain rate gives an increase in both properties about 40°C higher - the same difference as was observed for the onset and cessation of multiple yielding.

Of the ductility parameters, a minimum in RA was not defined with any certainty up to 250 or 260°C with either series of Mn-Ni-Mo specimens. The curve of RA against temperature was, however, shifted to a higher temperature for the higher strain rate by a similar amount to the strength curves. The difference appeared to be less, however, with the elongation curves, both of which exhibited minima close to 200°C.

The C-Mn specimens, which were only tested at the lower strain rate and only up to 200°C, showed no rise in yield stress and a less marked one in tensile strength. Nevertheless, ductility values fell sharply to below 40% RA at 18o and 200°C and approximately 10% elongation at 140-200°C.

Discussion

In Fig.3, the curves of properties against temperature have been brought together; these show that the higher strain rate with the Mn-Ni-Mo weld metal shifts the curves up by about 40°C. The changes in the different properties do not, however, occur at the same temperature for either strain rate. With the Mn-Ni-Mo material, the multiple yielding behaviour occurs roughly up to the maximum tensile strength and includes the minimum in elongation, but stops before the minimum RA (which was not reached with certainty in any series).

Theminimum in elongation occurs at a lower temperature than in RA and this difference may be because much of the elongation (especially with low ductility levels) occurs fairly uniformly over the whole gauge length, whereas the RA values are associated with the final stages of deformation over a short part of the gauge length. At the constant crosshead speed used for the present tests, this final deformation must necessarily take place at a higher local strain rate than the earlier uniform deformation and this could explain why the minimum RA occurs at a higher temperature than the minimum elongation.

If this is the case, it poses something of a dilemma when trying to use tensile test behaviour to assess likely tearing behaviour in pressure vessel service when strain rates are likely to be low. Tearing behaviour leading to fracture is probably closer to the processes occurring during RA than during uniform elongation, but, because deformation during necking in a tensile test at constant crosshead speed is taking place at an increased rate, the strain rates in practice are likely to be closer to those involved with uniform elongation.

In Fig.3, it can also be seen that the C-Mn weld metal reached its minimum elongation at a lower temperatuure than the Mn-Ni-Mo weld. It is important, therefore, to carry out tests at sufficient temperatures to establish that the minimum RA value has been reached if it is required to compare the strain ageing behaviour of different weld metals in order, for example, to select one with a good resistance to dynamic strain ageing.

Summary

References