TWI Knowledge Summary

Failure of welded structures

Failures of welded structures can and do occasionally occur, sometimes with serious human, environmental and economic consequences. Table 1 shows approximate failure rates for various types of welded structure. It shows amongst other things how the use of experience-based engineering codes and standards can reduce failure rate (the ASME Boiler Code Committee was established in 1911, when boiler explosions in the USA were occurring at the rate of virtually one per day). Although such occurrences are much less common a century on, the continued prevention of failure requires careful attention to design, materials, construction, inspection and maintenance.

A useful way of categorising failures in welded structures is to distinguish between instant failure modes and time-dependent failure processes. Examples of instant failure modes include brittle fracture, overload failure and buckling. In all cases, the failure occurs when the 'driving force' for failure (e.g. applied stress, applied stress intensity) exceeds the materials resistance (e.g. tensile strength, fracture toughness). Consequently, instant failure modes are quite likely to occur early in the life cycle of the structure, perhaps due to errors in design, construction, materials or inspection. Failure of pressure vessels during their final hydrotest^[1,2] (see also Catastrophic Failures of Steel Structures in Industry: Case Histories.' TWI report 632/1998, February 1998) and of bridges during construction^[3], come into this category. On the other hand, 'instant' failures may also occur after many years of operation because of statistically rare events, such as the loading induced by a severe storm or the reduction in fracture toughness caused by unusually low temperatures. It should also be borne in mind that fracture toughness can fall over time as a result of embrittlement (e.g. temper embrittlement, radiation embrittlement) so that a structure which was safe at the start of life may later become unsafe.

Time-dependent failure processes include fatigue and creep crack growth, stress corrosion, other forms of corrosion, and wear. These may eventually lead to final failure by one of the instant failure modes described above, but are more amenable to control by in-service inspection. For example, welded joints are particularly susceptible to fatigue, typically initiating from discontinuities at the weld toe. However, the relationship between applied stress range, joint geometry and fatigue life is well-documented, and failure can be prevented by basing the design on an appropriate design codes.^[4,5] . The FAQ 'Where a welded structure is subjected to cyclic rather than static loading, how is the allowable stress calculated?' provides further information on this topic.

Although design codes allow a notional design life to be assigned to a welded structure, life extension and/or change of use of a welded structure are often possible through the use of in-service inspection, materials assessment and application of fitness-for-purpose procedures^[6-10]. additional information is contained in the FAQ 'Are there any codified procedures to evaluate the significance of weld flaws in metallic structures and components?' .

Table 1. Examples of failure frequencies of welded structures

References

Further information

You can use the Weldasearch literature database to supplement what you find in JoinIT.