A crack may be defined as a local discontinuity produced by a fracture which can arise from the stresses generated on cooling or acting on the structure. It is the most serious type of imperfection found in a weld and should be removed. Cracks not only reduce the strength of the weld through the reduction in the cross section thickness but also can readily propagate through stress concentration at the tip, especially under impact loading or during service at low temperature.
Solidification cracks are normally readily distinguished from other types of cracks due to the following characteristic factors:
- they occur only in the weld metal
- they normally appear as straight lines along the centreline of the weld bead, as shown in Fig.1, but may occasionally appear as transverse cracking depending on the solidification structure
- solidification cracks in the final crater may have a branching appearance
- as the cracks are often 'open', they can be visible to the naked eye
On breaking open the weld, the crack surface in steel and nickel alloys may have a blue oxidised appearance, showing that they were formed while the weld metal was still hot.
The cracks form at the solidification boundaries and are characteristically interdendritic. The morphology reflects the weld solidification structure and there may be evidence of segregation associated with the solidification boundary.
The overriding cause of solidification cracking is that the weld bead in the final stage of solidification has insufficient strength to withstand the contraction stresses generated as the weld pool solidifies. Factors which increase the risk include:
- insufficient weld bead size or shape
- welding under high restraint
- material properties such as a high impurity content or a relatively large amount of shrinkage on solidification.
Joint design can have a significant influence on the level of residual stresses. Large gaps between component parts will increase the strain on the solidifying weld metal, especially if the depth of penetration is small. Therefore, weld beads with a small depth-to-width ratio, such as formed in bridging a large gap with a wide, thin bead, will be more susceptible to solidification cracking, as shown in Fig.2. In this case, the centre of the weld which is the last part to solidify, is a narrow zone with negligible cracking resistance.
Segregation of impurities to the centre of the weld also encourages cracking. Concentration of impurities ahead of the solidifying weld front forms a liquid film of low freezing point which, on solidification, produces a weak zone. As solidification proceeds, the zone is likely to crack as the stresses through normal thermal contraction build up. If liquid from the weld pool can feed into an incipent crack, it can be prevented. For this reason, an elliptically shaped weld pool is preferable to a tear drop shape, and fast welding speeds, which result in a large separation between the weld pool and cracking locations, increase the risk of cracking. Welding with contaminants such as cutting oils on the surface of the parent metal will also increase the build up of impurities in the weld pool and the risk of cracking.
As the compositions of the plate and the filler determine the weld metal composition they will, therefore, have a substantial influence on the susceptibility of the material to cracking.
Cracking is associated with impurities, particularly sulphur and phosphorus, and is promoted by carbon whereas manganese and silicon can help to reduce the risk. To minimise the risk of cracking, fillers with low carbon and impurity levels and a relatively high manganese content are preferred. As a general rule, for carbon-manganese steels, the total sulphur and phosphorus content should be no greater than 0.06%.
Weld metal composition is dominated by the consumable and as the filler is normally cleaner than the metal being welded, cracking is less likely with low dilution processes such as MMA and MIG. Plate composition assumes greater importance in high dilution situations such as when welding the root in butt welds, using an autogenous welding technique like TIG, or a high dilution process such as submerged arc welding.
In submerged arc welds, as described in EN 1011-2:2001 Annex E, the cracking risk may be assessed by calculating the Units of Crack Susceptibility (UCS) from the weld metal chemical composition (weight %):
UCS = 230C* + 190S + 75P + 45Nb - 12.3Si - 5.4Mn - 1
C* = carbon content or 0.08 whichever is higher
Although arbitrary units, a value of <10 indicates high cracking resistance whereas >30 indicates a low resistance. Within this range, the risk will be higher in a weld run with a high depth to width ratio, made at high welding speeds or where the fit-up is poor. For fillet welds, runs having a depth to width ratio of about one, UCS values of 20 and above will indicate a risk of cracking. For a butt weld, values of about 25 UCS are critical. If the depth to width ratio is decreased from 1 to 0.8, the allowable UCS is increased by about nine. However, very low depth to width ratios, such as obtained when penetration into the root is not achieved, also promote cracking.
The high thermal expansion (approximately twice that of steel) and substantial contraction on solidification (typically 5% more than in an equivalent steel weld) means that aluminium alloys are more prone to cracking. The risk can be reduced by using a crack resistant filler (usually from the 4xxx and 5xxx series alloys) but the disadvantage is that the resulting weld metal is likely to have non-matching properties such as a lower strength than the parent metal.
Austenitic Stainless Steel
A fully austenitic stainless steel weld is more prone to cracking than one containing between 5-10% of ferrite. The beneficial effect of ferrite has been attributed to its capacity to contain harmful impurities within the grains which would otherwise form low melting point segregates and consequently interdendritic cracks. Therefore the choice of filler material is important to suppress cracking so a type 308 filler is used to weld type 304 stainless steel.
Best practice in avoiding solidification cracking
Apart from the choice of material and filler, the principal techniques for minimising the risk of welding solidification cracking are:
- Control joint fit-up to reduce gaps.
- Before welding, clean off all contaminants from the material
- Ensure that the welding sequence will not lead to a build-up of thermally induced stresses.
- Select welding parameters and technique to produce a weld bead with an adequate depth to width ratio, or with sufficient throat thickness (fillet weld), to ensure the weld bead has sufficient resistance to the solidification stresses (recommend a depth to width ratio of at least 0.5:1).
- Avoid producing too large a depth to width ratio which will encourage segregation and excessive transverse strains in restrained joints. As a general rule, weld beads whose depth to width ratio exceeds 2:1 will be prone to solidification cracking.
- Avoid high welding speeds (at high current levels) which increase the amount of segregation and the stress level across the weld bead.
- At the run stop, ensure adequate filling of the crater to avoid an unfavourable concave shape.
As solidification cracks and crater cracks are linear imperfections with sharp edges, they are not permitted for welds meeting the quality levels B, C and D in accordance with the requirements of BS EN ISO 5817:2007. Crater pipes may be permitted for quality level D, depending on their size.
Detection and remedial action
Surface breaking solidification cracks can be readily detected using visual examination, liquid penetrant or magnetic particle testing techniques. Internal cracks require ultrasonic or radiographic examination techniques.
Most codes will specify that all cracks should be removed. A cracked component should be repaired by removing the cracks with a safety margin of approximately 5mm beyond the visible ends of the crack. The excavation is then re-welded using a filler which will not produce a crack sensitive deposit.
This Job Knowledge article was originally published in Connect, November/December 1999. It has been updated so the web page no longer reflects exactly the printed version.