Finally, in our consideration of shrinkage and distortion we must not ignore the impor­tance of heat input. As we have seen in Chapter 2 and 3, the heat from the weld pool during solidification flows into the plate adjacent to the fusion boundary. The width of metal heated to above room temperature is greater than that of the fused zone, and the picture used above of a hot weld-metal element between cold plates is an over-simplification. The heat flowing into the plates establishes a temperature gradient which falls from the melting point at the fusion boundary to ambient temperature at some point remote from the weld.

The heated-band width is directly proportional to the heat input in joules per mm length of weld and is therefore dependent on the process being used. It follows that the amount of distortion and shrinkage will also vary from one welding process to another. If the heat source moves slowly along the joint, the heat spreads into the plate and the width of hot metal which must contract is greater. The effect is less noticeable in thick plate but in sheet material, say 2 mm thick, the differences are most marked. The GMA system, with its fast speed of travel, gives a narrow heat band compared with the spread in oxy-acetylene welding, and it is possi­ble to arrange the manual processes in ascending level of distortion, i. e., GMA, SMAW, GTA and oxy-acetylene welding.

5.4.5 Residual Stresses

Solving the problem of distortion control during welding and determining shrinkage allow­ances for design purposes are of such importance in fabrication that it is easy to overlook the fact that they are the products of plastic deformation resulting from stresses induced by con­traction in the joint. As long as these stresses are above the yield point of the metal at the prevailing temperature, they continue to produce permanent deformation, but in so doing they are relieved and fall to yield-stress level. They then cease to cause further distortion. But, if at this point we could release the weld from the plate by cutting along the joint line, it would shrunk further because, even when distortion has stopped, the weld still contains an elastic strain equivalent to the yield stress. We can visualise the compeleted joint as an element of weld metal being stretched elastically between two plates.

The stresses left in the joint after welding are referred to as residual stresses. From our discussion of shrinkage and distortion, it can be seen that there will be both longitudinal and transverse tension. In the case of the longitudinal stresses, the weld itself and some of the plate which has been heated are at or near yield stress level (Fig. 5.22). Moving out into the plate from the heat-affected zone, the stresses first fall to zero. Beyond this there is a region of compressive stress.

It must be emphasised that all fusion welds which have not been subjected to post-weld treatments-in other words, the vast majority of welded joints contain residual stresses. Procedures developed to minimise distortion may well alter the distribution of the residual stresses but do not eliminate them or even reduce their peak level. Having said this, since we cannot avoid the formation of residual stresses, it is appropriate to ask if we are worried by their presence. As with so many engineering situations the answer is not a simple yes or no. There are numerous applications where the existence of residual stresses would have little or no influence on the service behaviour of the joint-storage tanks, building frames, low-pressure pipework, and domestic equipment all provide examples of situations where the joints can be used in the as welded condition without detriment.

If the service requirements do indicate that the residual stresses are undesirable, the designer must take them into account when selecting materials and deciding upon a safe working stress. This approach can be seen in the design of ships, where the combination of low temperatures and residual stress could lead to a type of failure known as brittle fracture. The designer selects a material which is not susceptible to this mode of failure even at the low temperatures which may be experienced during the working life of the ship; the presence of residual stresses is then important. Similarly, in many structures subjected to loads which fluctuate during service—for example, bridges, earth-moving equipment, and cranes—the designer recognises the existence of residual stresses by choosing a working-stress range which takes account of the role these stresses play in the formation and propagation of fatigue cracks.

There are, however, some specific applications where it is essential to reduce the level of residual stresses in the welded joint. With pressure vessels, because of the risk of a catastrophic failure by brittle fracture, stress-relieving is often a statutory or insurance requirement. Again, some metals in certain environments corrode rapidly in the presence of tensile stress, i. e., stress corosion will occur. In these cases, a joint in the as welded condition containing residual stresses suffers excessive attack; this is retarded if the joint is stress-relieved. Finally, when machining welded components, removing layers of metal near the joint may disturb the balance between the tensile and compressive residual stresses and further deformation or warping can occur. This can make it difficult to hold critical machining tolerances and it may be desirable in these circumstances to stress-relieve to achieve dimensional stability.