Neville Gregory's career as a metallurgist spanned 40 years up to his retirement in 1990. After working at The English Electric Company, Rugby, Murex Welding Processes and Enfield Rolling Mills he joined the London office of BWRA as a Welding Technologist. For six years he carried out liaison visits to members and dealt with technical enquiries.
This work continued when he joined the Research Station at Abington and was combined with contract research on various welding processes and materials including zinc coated steels, low alloy steels, reinforcing bars, and armour plate. Since 1975 Neville Gregory headed a Welding Advisory Service in the Arc Welding Department assisted by a team of Welding Engineers.
Rigorous observance of whether flaws are present in a weld is of no importance if they have no influence on structural performance. Neville Gregory outlines different types of flaw and discusses what is, and isn't acceptable.
National standards and codes specify acceptance levels for weld flaws in terms of their size, position and distribution. Some, such as cracks or lack of fusion, are prohibited.
These standards are invaluable for quality control purposes because it is necessary to produce welds to a predetermined quality and then use inspection to confirm that the requirements have been met.
These requirements have evolved from the weld quality that would be expected from a skilled welder, as well as practical experience over many years and data from extensive research work.
Because of difficulties in specifying the precise significance of weld flaws in the form of standard rules the British Standards Institution has issued a Published Document for guidance. The latest edition of this is PD 6493:1991 Guidance on methods for assessing the acceptability of flaws in fusion welded structures. Another document that applies a fitness-for-purpose concept based on linear elastic fracture mechanics is ASME Section XI In service inspection.
If the arbitrary acceptance levels are exceeded, in many cases it will be possible to use the guidance given in PD 6493 or ASME XI to determine whether the flaws present can be proved to be insignificant.
This approach, known as engineering critical assessment (ECA), has been able to save the cost of extensive repair operations on a number of occasions.
Application of the guidance documents should of course, be entrusted to appropriately qualified and experienced persons.
Classification
It is convenient to classify flaws into three categories to assess their significance, and these are as follows:
Planar flaws, i.e. flat with sharp edges:
- Cracks;
- Lack of fusion or penetration;
- Undercut or overlap.
Non-planar flaws, i.e. volumetric:
- Porosity;
- Slag inclusions.
Geometric weld shape:
- Misalignment;
- Incorrect profile.
Significance
National standards and codes specify acceptance levels for flaws that are based on the relative importance of the different types. The order of severity of different flaws depends on their relative effects on the integrity of a welded structure.
In decreasing order of significance, weld flaws can be listed as:
- Cracks;
- Lack of fusion or penetration;
- Undercut;
- Porosity or slag inclusions;
- Unsatisfactory weld shape.
Planar flaws
Cracks
A crack is considered to be the least acceptable type of flaw because it has sharp edges and therefore acts as a stress concentrator which reduces the resistance of a welded joint to either fatigue or brittle failure. Therefore all standards and codes prohibit cracks, which means that repairs have to be carried out unless an engineering critical assessment (ECA) based on fracture mechanics principles shows that the defect will not affect the integrity of the structure. With further advances in both fracture mechanics and non-destructive examination (NDE) techniques, incorporation of acceptance levels for cracks may well be feasible but is probably impracticable for the following reasons:
- The presence of cracks indicates some fault in the welding procedure and the fact that some cracks may be allowed could lead to some relaxation in the control of welding procedures.
- Cracks, unlike porosity or slag inclusions, generally occur in a completely unpredictable manner in terms of length, depth and sometimes orientation.
The present situation may be the best compromise, where cracking is not allowed by the code, but if repair is difficult then an engineering critical assessment can be carried out to determine whether any of the cracks are significant.
Engineering critical assessment can determine whether a crack will cause failure, provided that NDE can show its size and orientation and data are available on material properties and service and residual welding stresses.
In certain hardfacing applications cracks are quite acceptable, in fact they are a normal occurrence.
Weld deposits of high chromium irons in particular have a fine network of craze cracking which can extend to the weld junction. However, these cracks do not normally affect the service performance or the lives of components used to resist heavy abrasion in the mining, quarrying and mineral processing industries.
In contrast, some hardfacing applications require crack free deposits, for example:
- Sealing surfaces - valves and seats;
- Surfaces requiring both wear and corrosion resistance - chemical processing plant;
- Surfaces subject to mechanical and/or thermal fatigue - steel mill rolls, hot work dies.
These examples of the wide variety of types of service stresses and environments that can occur show some of the difficulties in specifying acceptance levels in a standard or code, because requirements are often unique to a particular application.
Lack of fusion or penetration These flaws are considered to be less serious than cracks because lack of fusion is generally restricted to the size of a single weld run and incomplete penetration will generally not be larger than the size of the root face in butt or T butt welds.
A crack, lack of fusion and incomplete penetration of similar size would probably be rated in this order of decreasing severity because this is the order of acuity of the tip of each defect. However, in practical terms it is not possible to use this difference to set a range of acceptance levels and therefore these flaws are all considered as planar and are disallowed by codes.
Lack of fusion and incomplete penetration can, when necessary, be analysed on a fitness-for-purpose basis and this can determine whether repairs are necessary.
Undercut Most codes specify maximum allowable depths of undercut and PD 6493 gives recommendations for acceptance levels for both butt and fillet welds for a range of material thicknesses. Undercut which exceeds the specified limits is assumed by PD 6493 to behave like a crack and can be assessed as a planar flaw on fracture mechanics principles.
Therefore, should the need arise, an engineering critical assessment can be carried out if undercut is deeper than, say, 1mm.
In a small scale fabrication it could be more cost effective to grind out the undercut rather than submit the case to a rigorous engineering critical assessment.
Porosity and slag inclusions
At maximum levels, porosity and slag inclusions generally have no effect on the static tensile or ductility properties of a welded joint.
Figure 1 shows a broken cruciform tensile test specimen from a steel plate coated with excess primer and welded by the submerged-arc process. The gross and extensive porosity shown on the fracture surface did not lower the tensile strength below that of a control test specimen made of uncoated steel and having welds free from porosity.
Fig. 1. Extensive wormhole porosity in broken cruciform tensile test specimen
Internal porosity in fillet welds has been shown to have no effect on fatigue strength when they are non-load bearing ( Fig.2). In such welds fatigue cracking is always initiated at the weld toe and propagates through the parent plate.
Fig. 2. Non-Load bearing fillet weld, fatigue failure through plate
In load bearing fillet welds, provided that the weld is large enough, fatigue failure also occurs at the weld toe ( Fig.3) and porosity is not significant. However, if the weld is small enough fatigue cracking is initiated from the root of the weld ( Fig.4) and the crack propagates through the weld throat.
Fig. 4. Load bearing small fillet weld fatigue failure through weld
In this case porosity that is both large and extensive enough to reduce the effective throat thickness will lower the fatigue strength.
Fillet welds should, in any case, be designed large enough to prevent fatigue failure through the weld throat, but obviously in borderline cases gross porosity could change the position of failure from the toe of the weld to the throat.
The maximum acceptable amount of porosity given in PD 6493 is that which occupies 5% of the projected area on a radiograph and the maximum pore diameter is quoted as a quarter of the thickness or 6mm, whichever is smaller.
Maximum sizes of slag inclusions recommended by PD 6493 are 3 x 3mm with no limit on length.
The above maxima in PD 6493 are stated to be suitable only for materials having certain minimum values of Charpy V impact strength or fracture toughness quoted in the PD. Materials having lower toughness can be assessed on the basis of their maximum dimensions and can be treated as having planar flaws.
Some codes allow pores having diameters that increase with plate thickness up to a maximum of 6mm for plate above 75mm. This philosophy may have originated in the days when larger welds meant larger flaws, but today, even if justified by engineering critical assessment, the presence of 6mm diameter pores would alert the welding engineer to the need to make some adjustment to the welding procedure.
Therefore there are instances when the requirements of quality control could be more onerous than those of fitness for purpose. The acceptance levels for flaws in pressure vessel and piping application standards are much lower than those quoted above.
In fact, some codes for nuclear work require a complete absence of porosity, for example in Co-Cr-W (Stellite type) weld coatings on valves and valve seats.
This requirement may well be too onerous and an engineering critical assessment could possibly reduce the need for the number of repairs that are carried out, with considerable economic benefit both to the client and to the fabricator.
Solid inclusions such as slag are considered to have the same effect on weld integrity as porosity. Slag inclusions amounting to, say, 3% of projected area on a radiograph might be acceptable on a fitness-for-purpose basis but would indicate a major fault in the welding procedure or technique and would fail any reasonable quality control criterion.
Occasionally slag inclusions may have one or more sharp edges and in this case they may have to be evaluated as planar defects. The presence of this type of slag, fortunately not common, may only be revealed when the weld is radiographed from different directions.
Misalignment and incorrect profile
These flaws include excess weld metal, excess penetration, root concavity, misalignment, convex or concave profile and unequal or undersize leg length and they have limits specified in most codes.
All these flaws are welder or operator related and do not arise because of material properties or associated metallurgy.
Visual inspection will find these flaws, and if they do not comply with the code, which can be decided by simple measurement, it is generally easier to make repairs than to perform an engineering critical assessment.
In borderline cases any variation in weld profile from that specified can only be accepted if ECA confirms that it gives a weld throat thickness or weld shape that will carry the maximum static or dynamic service stresses.
What of the future?
With increased knowledge of the significance of weld flaws and more effective NDE techniques, engineering critical assessment is a powerful tool for increasing efficiency in welded fabrication.
Together with these advances, there will no doubt be greater efficiency in controlling welding procedures which should be accompanied by a reduction in the incidence of weld flaws.
The greater the precision with which we can locate and define flaws, the more we understand their significance.
Unfortunately things are not well ordered in the real world. For instance, we first knew about hydrogen induced cracking about half a century ago but it has still not been completely eliminated.
