Killer consequences of defective welds - a plan for prevention

TWI Bulletin, January/February 2002

Bob Apps graduated from Birmingham University in the fifties having completed his doctorate in heat flow in arc welding. After periods with ICI and Atomic Power Constructions where he headed the welding group, he joined the College of Aeronautics, later Cranfield Institute of Technology and Cranfield University. Following retirement in 1989, he became Professor Associate at Brunel University.

Bernard Crossland started working life with Rolls-Royce in the late thirties. With an external London degree to his name, he later joined Bristol University as an assistant lecturer in mechanical engineering. By the sixties he had been appointed Professor of Mechanical Engineering at Queen's University Belfast. Since retirement in 1984 he has specialised in disaster investigation, most recently in relation to the Southall and Ladbroke Grove railway accidents.

Colin Evans is a director of Marcham Technical Services with more than twenty years experience in management of aerospace, oil and gas, nuclear, steel construction and power generation projects.

Against a backcloth of welding related catastrophes, a quartet of failure specialists, Bob Apps, Bernard Crossland, Colin Evans and Robert Fenn, consider five examples of accidents and disasters arising from such failures and then highlight the lessons learned from these incidents. It is concluded that these incidents arise from inadequate or non-existent education and training of engineers, technicians and welders involved in design and fabrication.

A brief review of the education and training courses currently available is given. It is suggested that there is a dearth of courses and confusion about those available, which makes it difficult for the most appropriate course to be identified.

It is concluded that there must be a wider acceptance of the need for high quality education and training courses at all levels. There is also a need for rationalisation of courses.

The authors draw on examples from their experience and knowledge of weld failures, which have led to serious financial losses and even more regrettably to death and injury. They discuss the lessons to be learned and the actions to be undertaken and emphasise that successful welding requires an understanding of the processes involved and also well considered training. Modern fabrication depends for its success on much more than the out-of-date skills of the blacksmith, welding being the last truly 'high tech' process that is still undertaken manually.

All the authors are concerned with the many defective welds that they have encountered in the course of their careers. These failures have arisen from abysmal, or total, lack of knowledge of the metallurgical consequences of welding. On other occasions the problems have arisen from an inadequate analysis or understanding of the stresses arising from static and dynamic loading. In other cases the design did not allow for adequate access to allow the welder to make sound joints, or even the NDT inspector to inspect the weld made. Frequently when a professional engineer was involved they had little or no formal education and training in welding, and this frequently applied to the welder and NDT inspector.

European Health and Safety Directives have been incorporated in British Laws and clearly place the responsibility for safety on the management. Failure to accept this responsibility for safety can result in a substantial fine and, in the case of death, the increasing possibility of imprisonment under the law relating to corporate manslaughter, which is currently being stiffened by the present Government.

Examples of weld failures

• Failure of a large liquid containment tank In large thin-walled tanks for the containment of liquids at low temperatures it is necessary to fit 'wind stiffeners' ( Fig.1) in the form of angle iron circular hoops welded to the inner wall of the tank. If this is not done then under transverse wind loading the tank is excessively flabby.

The wind stiffener is made out of angle iron arcs seam welded to the tank wall, and butt welded to the adjacent arcs. With large tanks several tens of metres in diameter, it is difficult to get good fit-up between the arcs. In the case under consideration there was a significant gap between the two adjoining arcs and the welder had 'smeared' or 'buttered' the gap. Additionally, the access to the underside of the wind stiffener was difficult. The buttered weld was full of porosity, inclusions and lack of fusion, which under the service stresses and low operating temperature constituted a defect larger than the critical crack size. Consequently the butt weld failed and the fracture ran some metres up and down the tank wall before arresting. Considerable expense and a major scare resulted from the ensuing leak of inflammable fluid.

The weld design did not recognise the practical problems of making a sound weld in the absence of good fit-up or of inspecting the weld. Too frequently attachments to main structures, in this case the wind stiffener, are neither recognised nor appreciated to be an integral part of the primary structure and thus demanding the same welding and inspection quality. The butt weld in question experienced the same stresses as did the tank wall, thus both the weld and inspection standards need to be as rigorously maintained as those in the main structure. According to the standard ^[1] to which the tank had been built, all the welds should have been visually inspected. From the topside of the wind stiffener the weld appeared sound, but from beneath it would have appeared defective; however, it was not readily possible to view the underside of the weld. In this example it would have been easy to replace the butt welds with welded on steel straps across the gap between the arcs ( Fig.2), with welds which would be easily made and inspected.

• Failure of stainless steel selector chains in a rotary kiln These 10m long chains hang down across the entrance of, and run into, the rotary kiln to retard the progress of the large pieces of limestone. These chains wear out over time and are replaced by metallic arc welding new lengths of chain onto existing undamaged links external to the kiln.

  After a second unscheduled 'down' within a few weeks, caused when the chains broke and long lengths entered the kiln, it was planned to take legal action against the chain supplier, on the assumption that the chain was defective. Prior to proceeding an investigation into the material was initiated. Examination of recovered links, which had connected the new to the existing lengths of chain, showed that they had opened up when the weld failed. The material of the weld was easily broken and crushed and it reminded the investigator of slag.

  When the maintenance engineer in charge of refurbishment was interviewed it was discovered that he had carried out the welding. He was certain that he had welded and dressed the link perfectly, but on examining the welding electrodes used it was found that they were nickel/iron used for welding cast iron. On further questioning his response was 'they are electrodes that welders like and, anyway they are all alike'! When the correct stainless steel electrodes were used the company suffered no further unscheduled downs and no further financial losses.

  This simple example illustrates, as in many other cases, that welding is perceived as a low cost operation with no technical requirements.

  
  

• Failure of a reinforcing, 're-bar', structure Part of a harbour installation required the construction of a 'mole' to which ships tied up. The mole looked like a conical beehive constructed of reinforced concrete. The rebar specification was for high tensile steel 38mm diameter rods, cold bent into shape ( Fig.3), wired and bolted together, which is standard practice, and then lowered into position. Lifting the frame caused it to move out of correct shape and after several attempts it was suggested that the joints be welded in position. Initially this idea was rejected, as no welding was allowed unless the weld procedure was controlled by a suitably qualified welding engineer. Ultimately a person was found by the contractors who were satisfied with his credentials. This man gave a verbal specification to the welders, including defining the electrodes to be used. After welding, as the framework was being hoisted in position, it came apart crushing two men beneath it.
  

The case went to court where a qualified welding engineer gave evidence against the construction company. His evidence was that the choice of electrode was suitable, but that welding had been performed with short, tack-like welds, which had produced martensite within the heat-affected-zone (HAZ). This martensite was so brittle that it effectively embrittled the whole re-bar section, additionally there were signs of hydrogen cracking in the martensite. He backed his evidence by mechanical tests, which demonstrated that the welded re-bar section was almost glass brittle in the welded regions.

On questioning, the welding engineer who had produced the welding specification, answered that the welding electrode was selected from the catalogue (it was an excellent choice). He had told the welders 'to use welds as small as possible'. Expert evidence noted that small welds cooled extremely rapidly and for the re-bar material, which was a high carbon steel (0.8%), such rapid cooling caused the formation of martensite in the HAZ. This expert went further to suggest that the welds should have been as large as possible to give a slow rate of cooling and that the welds should have been given a post-weld heat treatment to temper any existing martensite.

The so-called 'welding engineer' who had specified the welding procedure was jailed for two years. He was, in fact, an electrode company's salesman and had had only two days training. His company was heavily fined for allowing one of its employees to give information beyond his capabilities, and the construction company was also heavily fined for using an unqualified consultant.

• Self-steering axles A major motor vehicle manufacturer required a self-steering axle for large tandem trailers. The job was put out to tender, the successful company getting the sole rights to manufacture, sell, install and maintain these axles in the vehicle manufacturer's name. After an unduly short service life, a number of serious failures were reported in these axles and the vehicle manufacturer sent two of them for independent examination.

  These axles had been made from 100mm square section tubing of 10mm wall thickness ( Fig.4) and were fabricated by fusion welding cut sections together to form the 'stirrup' shaped axle. Both axles examined had failed through the welds. On examination of the macrostructure it was found that the tubing had no weld preparation, consequently the weld had only penetrated 3mm. Set-up of the three component tubes forming the stirrup was very variable. Some of the tubes having been cut at the wrong angle, though laid out to establish the correct overall shape and dimensions. This caused the 'root gap' to vary from 0-10mm on some limbs. MAG (CO ₂) welding had been used and inside each joint were lengths of wire, indicating that the welder had attempted 're-starts' and the wire had simply passed through the root gap and made contact with the inner wall. There was a large amount of weld splatter inside the tubes. Radiography of surviving joints confirmed these faults and exposed other cracks, which had grown and could have led to failure.

  A legal case initiated against the axle manufacturer found them liable for all the problems associated with these failures, including consequential damage. As a result the company was forced into liquidation. Fortunately nobody had been killed and only a few injured, although considerable damage had been caused.
  
  

• Collapse of a maritime passenger walkway On 14 September 1996 a passenger walkway, connecting the passenger ramp building on shore via a loosely moored pontoon to a roll-on roll-off ferry, collapsed. The seaward end of the walkway fell 10m on to the deck of the pontoon.

  The passenger walkway was of a lattice construction ( Fig.5) clad with 6mm steel plate, which resulted in a torsionally stiff structure. To allow for the movement of the pontoon relative to the shore, caused by tide, waves and impact of docking ferries the walkway was supported on feet at its four corners. The shoreward pair of feet rested on slideways in the passenger ramp building. The seaward pair of feet rested on a support platform welded to one of the portal frames supporting the upper roadway across the pontoon. The right foot looking seaward included a vertical 80mm diameter pin or 'pintle', which was located in a hole through the platform ( Fig.6). Except for the pintle all four feet were identical.
  

On examination it was found that the stub axle to collar weld ( Fig.6) on the right-hand foot of the seaward end of the walkway had failed by fatigue. This freed the walkway so that it could work its way off the supporting platform, allowing it to fall to the deck of the pontoon. The left-hand foot at the seaward end became detached when it impacted the deck of the pontoon. On examination it was found that the stub axle to collar weld had nearly failed by fatigue, before it impacted the deck. The two feet at the landward end were still intact, but on inspection it was found that both had fatigue cracks in the stub axle to collar welds. Metallurgical examination showed lack of penetration, lack of fusion, entrapped slag, surface breaking porosity and cracks in the first weld runs, which had only been partially removed by grinding.

Checking the design calculations, it was discovered that what was a simple bending calculation to determine the static stress in the stub axle, had been carried out completely incorrectly by the designer. A separate calculation by a classification society, appointed to give plan approval, was also completely incorrect for entirely different reasons. Both sets of calculations had grossly underestimated the bending stress in the stub axle by a factor of twenty or more. The designer had not attempted to calculate the stress in the weld connection and, though the classification society had, they grossly underestimated the bending moment and as a consequence the resulting stress. Neither party had attempted to calculate the fatigue life, but possibly they thought the stresses they had calculated were so low that it was unnecessary.

The stub axle was a medium carbon steel (0.486%) whereas the collar was a low carbon steel (0.145%). Qualified welding engineers would recognise that precautions are required in making such a weld to prevent HAZ cracking. According to the rules of the classification society, welding should have been carried out to an approved welding procedure document, but no such document was provided by the fabricator. Such a document would have defined the need for pre-heat, the electrode to be used, the need for a qualified welder and post-weld heat treatment. It was clear that no post-weld heat treatment had been carried out and the quality of the welding suggested that the welder was unqualified.

There were many other problems associated with the design and erection, which are outside the scope of the present paper. ^[2-4]

The walkway collapsed shortly after midnight when a few stragglers were dashing to catch the midnight ferry. Six passengers were killed and seven others were seriously injured. If the collapse had occurred a little earlier many more people would have been involved.

The companies responsible for the design, construction and erection of the walkway and the operating company were found guilty and heavily fined. They were also faced with substantial costs. The classification society pleaded guilty and was fined and had to pay costs. No doubt there were considerable claims against the operating company and their insurers from the relatives of the deceased and the injured, which had to be met.

Lessons to be learned

Failures have arisen from incompetent management who have neglected to ensure that their professional engineers, welding and NDT operatives were adequately educated and trained for their functions, and that their qualifications were regularly updated. It appears that too often welding is regarded as a blacksmith's job, with no appreciation of the stringent quality requirements now demanded of welded fabrications. Most importantly, we need to learn from the disasters resulting from bad welding practices.

The construction industry has been notorious for management failings leading to delays in the completion of contracts, non-compliance with design requirements and minor and major accidents leading to considerable financial losses and much more worrying, to injuries and deaths. This has led to the Construction, Design and Management (CDM) Regulations 1994, and these appear to be particularly relevant to the fabrication industry; perhaps we can learn from this experience. However small or large a project may be it merits the appointment of somebody to take overall control and responsibility for the successful completion on time and to specification.

The duties of a project manager may include budgetary control, ensuring the dissemination of information between the parties involved, quality management, progress monitoring and reporting. It is his/her duty to ensure that these aspects are in place and are fully implemented as a team effort. In particular quality management is a crucial element in insuring that all aspects of the design, fabrication and erection are fully assessed and the checking procedures are in place. Importantly this includes checking that the individuals involved are adequately trained and educated for their function. It can be seen in the examples quoted earlier that good project management would probably have avoided the problems which arose.

Education and training for welding

Perhaps the best way of achieving compliance with EN 719, short of a legal requirement, is to encourage independent qualification/certification of the personnel involved, though cost to the companies and individuals may be seen as a barrier. A serious problem is the maze of qualifications/certifications available ( Table), which is confusing for the individuals and their employers. Simplification and clarification of this maze is urgently required, but in the meanwhile the Professional Division of TWI is available for advice.

Table: Welding related courses, qualifications and certificates currently available in the UK

The Engineering Council is the overarching body for the whole engineering profession in the UK and is concerned with recognising the levels of Chartered and Incorporated Engineers and Engineering Technicians. Admission to one or other of these grades recognises both educational and experience components required for the holder to operate at such a level. Professional bodies in the various engineering disciplines, such as The Welding Institute (TWI), are recognised as Nominated Bodies by the Engineering Council. They approve candidates in the various grades and recommend them to the Engineering Council for the award of the appropriate professional title.

The problem is that the knowledge and competence required for professional recognition is well defined and not job specific, additionally there is no requirement for updating. This makes it essential to extend the core knowledge and encourage and make provision for continuing professional development (CPD). Ultimately CPD may become a mandatory requirement for continuing professional recognition, as in some other professions.

The take-up of welding courses at all levels in the UK has always been low in comparison with other EC countries and it has deteriorated further in recent times. Due to lack of industrial support, post-graduate courses, which produced graduates to professional standards, have withered away. There has been a decline in the take-up of the TWI certificate course in welding offered as a final year module in engineering degree courses. In the technical colleges there has been a great decline in the availability of welding in their course curriculum, even in the regions where there is an apparent industrial need. Internationally harmonised training and qualification schemes, which have emerged and are referred to in EN 719, have not been widely taken-up in the UK. This decline in education and training has continued despite industry having been warned of the trend and its consequences. ^[5-10]

Another European Standard EN 729: 1994 (Quality requirements for welding - fusion welding of metallic materials) gives guidance to companies involved with welding on how to ensure adequate control. If this standard was adopted as a management tool, as ISO 9001/2 has been, the chances of errors occurring would be significantly reduced. For example one of the requirements of EN 729 is to ensure that people in the company with welding related responsibilities must demonstrate competence by compliance with EN 719. Unfortunately there is little evidence of this being readily accepted in the UK.

Conclusions

This must surely reflect a completely unacceptable state of affairs, which brings the engineering profession into public disrespect. It also reflects badly on the competence and social responsibility of the companies responsible. Besides involving considerable financial loss and possible legal proceedings, it must affect their competitiveness in home and overseas markets.

Recently some fabrication companies have realised that investing in their welders' understanding of the technology and its applications is a cost efficient way of achieving the required quality level. It builds quality into the job rather than attempting to achieve quality by inspection. Dramatic increases in quality and productivity have accompanied this initiative, with associated cost reductions. All companies should be encouraged to adopt this policy.

There are many other actions, which must be taken to rectify the problems that have been identified. More universities should create centres of welding excellence and provide taught courses, supported by the fabrication industry and other industries involved with welding. The Engineering Council should make it mandatory that Chartered and Incorporated Engineers and Engineering Technicians in appropriate engineering disciplines should have as a bare minimum a short awareness course in welding. The take-up of the TWI certificate course as part of engineering degree courses should be strongly encouraged. Agreement should be reached on courses approved by the Engineering Council to be made widely available in technical colleges. All welders, besides having a broadly based training, should have an elementary education in welding, including examples of failures arising from bad practice. Companies involved in welding should accept their responsibility to ensure their staff associated with welding are appropriately educated and trained, failing this legislation should be introduced to enforce it.

References