TWI Frequently asked questions
by John Wintle
In general, the frequency of flaws in a weld depends on the skill of the welder and the specific parameters of the welding process and procedure. There can be considerable variations in skill between two welders with the same welderapproval. Relevant parameters include the process variables, type of materials and consumables, heating rate, pre-heat, access, fit-up, position, restraint, location (shop or field), section thickness and run length. Different weldswill therefore have different expected flaw frequencies.
Early estimates suggest that for the joining of CMn steel by the manual metal arc (MMA) process, with other conditions assessed as average, the total number of flaws might be expected to be around 25 per km of weld bead laid. Withinthis total will be flaws of different types (slag, pores, hydrogen cracks, lack of fusion, etc). The susceptibility to each flaw type is determined by the underlying mechanisms by which they can be formed and the associatedmetallurgical and process factors.
A number of studies have been made to determine flaw frequencies from examinations of welded equipment. Much attention has focused on thick-section welds used for reactor pressure vessels in the nuclear industry. This suggests thatthe frequency of large flaws (>0.5in) in vessels containing around 20 weldments manufactured in the early 1970s to Class 1 standards is on average about 0.4 flaws per vessel [e.g. Marshall (1982), Smith and Warwick (1981)]. Fromexamining inspection reports, Thurlbeck et al (1996) and Baker and Kountouris (1989) have derived estimates for the frequency of fabrication flaws in submerged-arc welds used for offshore jacket structures manufactured since 1970.
Structures fabricated using modern automated welding would be expected to have a lower frequency of flaws than those fabricated using manual processes. Wagner and Watchett have published data for girth weld defects in mechanised GMAfield welded pipelines. They found the repair rate due to flaws was around 1 repair per 10m of weld laid.
Welds are routinely inspected after fabrication and this can lead to the detection of significant flaws and their removal and repair. The frequency of flaws in welds entering service may therefore be different from that afterfabrication. Further changes to the flaw population can occur in service as new flaws become initiated by mechanisms such as fatigue or stress corrosion cracking.
Understanding the factors that control the frequency of flaws in welds and differentiating according to flaw type and size is of continuing interest at TWI. Core research report 749/2002 (only available to TWI Industrial Members) gives an extensive review. An expert system based on judgement and calibration is being developed to estimate flaw frequency of specified welds[Chapman (1992)].
References
Baker M J and Kountouris I S, Defect assessment analysis of the dimensions of defects detected by ultrasonic inspection of an offshore structure. CESLIC Report OR8:1989.
Chapman V, Simulation of defects in weld construction, ASME PVP Conf. 1992
Marshall W, Report of the Study Group on LWR Pressure Vessel Integrity. HMSO Publications 1976, 1982.
Smith T A, Warwick R G, A survey of defects in pressure vessels in the UK for the period 1962 to 1978 and its relevance to nuclear pressure circuits, UKAEA Report SRD R203, 1981
Thurlbeck S D, Stacey A, Sharp J V, Nichols N W. Welding fabrication defects in two offshore jacket structures. OMAE Conf., Florence, 1996
Wagner M J and Watchett B M, Girth weld defects in mechanised GMA field welded pipelines. Welding Journal, 75 - 82, June 1991.
