Manufacturing of welded joints with realistic defects

TWI Ltd

Paper presented at NDT 2011 - 50th annual conference of the British Institute of Non-Destructive Testing, 13-15 Sept. 2011. Telford, UK

Abstract

Intentional weld defect or flaw specimens can be required for training purposes, developing new non- destructive testing techniques, qualifying non-destructive testing procedures, obtaining mechanical property data and in support of safety cases. The single most important criterion in producing defects or imperfections is that they must accurately simulate flaws which can occur in welded components and structures. For this reason, in certain applications, saw cuts or machined slots which are more easily detected may not be considered acceptable as planar imperfections/defects for the purpose of NDT training or validation. Therefore, TWI has developed techniques for producing realistic imperfections/defects and, in the case of cracks, the desired morphology, including roughness, angles of tilt and skew to the surface.

This paper describes the techniques used to obtain the abovementioned defects and, for the most commonly required defect types, the qualification procedure used by TWI. This consists of inspecting by testing by surface crack detection, ultrasonic or radiographic inspection and/or sectioning to demonstrate that the dimensional tolerance of the simulated imperfections (ie actual size of the imperfection vs required size) can be generally guaranteed within ±0.5mm in through-wall extent and ±1mm in length.

1. Background

Intentional weld defect or flaw specimens can be required for non-destructive testing (NDT) operator training and validation, to develop validated NDT procedures or new NDT techniques, to obtain mechanical property data and in support of safety cases. The single most important criterion in producing defects or imperfections is that they must accurately simulate flaws which can occur in welded components and structures. In particular, when summarising the work carried out within the PISC project series (Project for the Inspection of Steel Components), Crutzen et al^[1] concluded that the NDT procedure has to be validated and tested on structures containing defects that, not being necessarily real ones, still do induce the real physical phenomena that the inspection techniques must be able to handle. Crutzen et al also stated that the use of very artificial discontinuities (side-drilled holes, SDH or flat-bottomed-holes, FBH) to demonstrate the capabilities of NDT techniques often results in optimistic statements and hazardous use of the technique on structures containing real defects. When comparing the use of welded joints with real or artificial realistic flaws (see definitions in Section 2), Crutzen listed the following advantages for the latter:

• Less expensive and time-consuming fabrication
• The possibility of more certain characterisation
• The provision of non-contaminated assemblies that can be more easily used for effectiveness assessments
• The consideration of more relevant selections of structural geometry and material

In this same review, facts appearing in favour of the use of artificial crack-like defects for NDT performance assessment were reported.

For these reasons, in certain applications, saw cuts or machined holes and slots, as well as structures containing 'real' defects, may not be considered acceptable as planar for the purpose of NDT training or validation. Therefore, TWI has developed techniques for producing realistic imperfections/defects and, in the case of cracks, the desired morphology, including roughness, and angles of tilt and skew to the surface.

TWI can reliably produce weld specimens with defects such as: lack of root fusion, lack of penetration, lack of sidewall or inter-run fusion, joint misalignment, porosity, solidification cracking, cluster cracking, heat affected zone (hydrogen) cracking, undercut, brittle fracture or fatigue cracks, under or overfill of weld metal, inclusions (slag or metallic). Some of these are essentially produced by using bad welding practice (lack of root fusion, porosity, solidification cracks), by welding with techniques such as tungsten inert gas (TIG) bridging to obtain lack of side wall fusion (LOSWF) or by welding under crack promoting conditions.

This paper presents a review of the techniques used to obtain the most commonly required defect types and the welding qualification procedure used by TWI. This consists of characterising the defects by surface NDT methods, ultrasonic or radiographic inspection and/or sectioning to demonstrate that the dimensional tolerance of the realistic defects (ie actual size of the defect vs required size) can be generally guaranteed within ±0.5mm in through-wall extent and ±1mm in length.

In addition, a case study is presented in Section 5. This describes a project recently completed project in which TWI manufactured bespoke non-standard defect specimens for NDT validation and operator training, developed the NDT technique, prepared the relevant procedures and finally demonstrated the NDT procedures at the customer's site.

2. Definitions

The following definitions were provided by Neundorf et al^[2] and are quoted from a glossary by ENIQ^[3] (European Network for Inspection and Qualification):

• Reference reflector: a reference reflector is a reflector in a test block whose response to the NDT technique provides a reference against which other responses can be compared. eg a side-drilled hole or a saw or electric discharge machined (EDM) notch whose exact dimensions are known.
• Real flaw: a flaw which has developed in a component during its manufacture or service without any steps having been taken to deliberately encourage its development.
• Artificial realistic flaw (also realistic flaw): a flaw deliberately inserted into a test assembly which is intended to produce a response to the NDT method under assessment which resembles that of a real flaw.

An ENIQ working document was quoted by Virkkunen et al^[4], which identifies four main techniques to obtain weld defects. These are listed below.

1. Implanted defects: where a pre-existing defect is attached to the test piece. The attachment usually takes the form of a weld in a machined recess.
2. Weld doping or weld modification: where for instance crack prone material is added to a weld to promote localised weld cracking. Other examples include introduction of porosity or slag.
3. Machined defects: where a defect can consist of a cut or machined void. EDM is perhaps the most relied upon technology in this area where a shaped electrode is used to erode the test piece. The process is most suitable for production of surface defects, although it is possible to use it in combination with welding to produce buried defects.
4. Grown defects: where cracking is initiated and propagated into test pieces in much the same way as would occur in plant, simply accelerated to make fabrication times practical. The main processes used for this class of defect are thermal fatigue and stress corrosion cracking.

3. Deposition of realistic defects at TWI

3.1 General

The vast majority of realistic defects produced by TWI are obtained by weld modification, machining or by growing defects (methods 2 to 4 as defined in Section 2). A full list of defects that can be deliberately inserted into welded joints is given below:

• Lack of side-wall fusion
• Lack of root fusion
• Slag inclusion
• Solidification cracking
• Cluster cracks
• Weld metal transverse cracking
• Porosity
• Heat affected Zone (HAZ) cracking
• Brittle fracture and fatigue cracks

This section shows a few recent examples of the techniques used to obtain the above defects and their application, for the different defect types. Details of the techniques not described in this paper are provided in a previous publication by TWI, which summarises the work carried out in support of the safety case for the Sizewell 'B' PWR power station in the UK (Lucas^[5]).

3.2 Lack of side-wall fusion (LOSWF)

Lack of side wall fusion defects are obtained with two techniques:

• TIG bridging
• Use of a metallic or non metallic insert

The 'TIG bridging' technique consists of outlining the edges of the defect on the weld edge with TIG runs, then bridging the area between them with further TIG runs, deposited so that no fusion with the base metal is obtained. The morphology of defects obtained with this technique is shown in Figure 1.

LOSWF obtained by using a metallic or non-metallic insert are deposited by tack welding an insert on the weld edge in the required position, welding it in position with TIG runs and then completing the weld according to the applicable welding procedure specification (WPS), see Figure 2. The metallic insert is normally made of a different material from that of the plates to be welded (eg a medium/high-carbon steel).

In both the above cases, due to contraction of the weld metal deposited to complete the weld, the TIG bridging runs and the metallic insert are 'pushed' towards the weld edge producing a very tight defect, which simulates the morphology of a real LOSWF. Both these techniques allow a very accurate control of the defect size.

Surface breaking LOSWF defects are always produced by TIG bridging. As shown in Figure 1c, the crack mouth tends to open due to solidification shrinkage of the weld; hence, it is not possible to obtain very tight (crack-like) defects by this method.

3.3 Lack of root fusion

Lack of fusion defects (similar to lack of penetration) at the weld root can be obtained by EDM notching or by TIG. Although EDM notching is precisely controlled, the resulting defect is characterised by a relatively large gape (Figure 3c) and cannot replicate a real lack of fusion defect (Figure 3a and b), which is better simulated by manual TIG welding. In order to obtain realistic lack of root fusion defects by manual welding, TIG is applied to obtain a weld metal build up at the weld root. This is then ground parallel to the opposite root face, according to the required defect size. Small TIG runs are deposited on top of the build-up, making sure that the contact surface between the build-up and the opposite root face is not melted. This leaves an unfused land which simulates the lack of root fusion defect.

The manual procedure allows defects within the required tolerances, even when very small sizes are required (1 to 3mm in through wall height).

3.4 Slag inclusion

Lucas described the procedure to obtain slag inclusions at TWI^[5]. Slag is formed from the residue of the electrode coating, which is principally deoxidation products from the reaction with the air and surface oxide. The slag becomes trapped in the weld when two adjacent weld beads are deposited with inadequate overlap and a gap is formed. When the next layer is deposited, the entrapped slag is not melted out. Thus slag may become trapped in cavities in multipass welds through excessive undercut or the uneven surface profile of preceding weld runs. The normal occurrence of slag is in the form of elongated lines which may be either continuous or discontinuous along the length of the weld.

As reported by Lucas, slag inclusions can be inserted in any position in the weld by stopping the welding operation for the length of the desired defect. Adjacent passes are then carried out to produce a groove in which powdered slag can be inserted, as shown in Figure 4. The top of the groove is sealed by small TIG runs. The slag is fused by the heat of the sealing runs and subsequent passes.

3.5 Solidification cracking

Solidification cracks normally occur through a poor weld bead size or shape. Cracks occur longitudinally and within the weld metal. A solidification crack can be induced by weld design and use of crack-prone filler metals (Figure 5a) or by using a specific welding technique (Figure 5b). The 'welding technique' route is the preferred one when the defect size, location and orientation are to be controlled.

3.6 EDM notching

As discussed in Section 2, machining or spark eroding are the most controlled ways to produce defects. Due to the nature of the machining operation itself, such defects would be classified more as 'reference reflectors' than 'realistic defects', as per Section 2.

However, there are cases where according to the requirements of the relevant code or standard and based on engineering considerations by NDT experts, the full control of the size, location and orientation of the deliberate defects is more critical than their resemblance to a real defect.

A specific example is that of a nozzle-to-shell weld mock-up prepared by TWI for NDT validation, with a weld thickness of approximately 140mm, manufactured by submerged-arc welding (SAW).

The location, orientation and sizes of the defects to be inserted in the nozzle to shell welds and on the nozzle inner radius, were selected to match the acceptance criteria in ASME section XI article IWB-3512. Following qualification of the defect production techniques as per the procedure described in Section 4 below, it was determined that if manual techniques were applied, it would not have been possible to guarantee acceptable tolerances on the required tilt and skew angles. In addition, for the purpose of the validation, it was not considered critical to obtain realistic defects.

Therefore, all defects were produced by EDM notching, with the results shown in Figure 6 below. In the case of defects located at mid-thickness, to prevent the subsequent SAW runs from melting of the defects, small TIG runs were deposited after notching, before resuming SAW welding. The parameters used to deposit these TIG runs were recorded during the weld procedure qualification, so that the same results could be obtained on the actual validation test piece.

4. Production of defective specimens at TWI

4.1 General

The production of defective welds at TWI usually involves three steps, which are detailed in Sections 4.2 to 4.4 below.

As and a quality plan is not necessarily required for the fabrication of defective test pieces, the authors suggest that NDT experts and standardisation committees could issue a standard qualification procedure for the production of realistic flaws, which would maximise uniformity throughout the industry.

This suggestion is in line with the guidelines recently published by ENIQ^[6] with regard to the design of test pieces and the conduction of test piece trials, which stipulates that the test-piece fabrication specification should place tolerances on test-piece defect parameters, and that the extents of fabricated defects should be checked as far as possible.

4.2 Fabrication specification

The first step when manufacturing a defective weld is to specify the type, quantities, location (embedded, sub-surface, surface breaking in HAZ, base metal or weld metal), orientation (tilt and skew) and size of the defects, as well as the joint design and the welding procedure.

The joint design and the welding procedure are usually identical to that of the welds to be inspected in production. On the other hand, one or more of the following factors will influence the selection of defect types, sizes, locations and orientations:

• Applicable inspection standard(s) or code(s): for instance, Code Case 2235-9 in Section V of the ASME B&PV Code^[7] may be used to select the defect types and sizes, based on tabulated height/length ratios.
• Fracture mechanics aspects: a fracture mechanics based fitness-for-service assessment would provide critical flaw sizes to be reproduced in the weld, to demonstrate that the selected NDT technique is capable of detecting them.
• In-service experience: the defective weld may be designed to simulate actual defects found on components in service, in order to develop inspection techniques to be applied to other components operating in similar conditions (see case study in Section 5).
• Other NDT considerations: for example, the validation of a radiographic inspection will be more conservative a lack of side wall fusion defect were located on the source side and if the fusion face angle were shallow.

Typically, a series of diagrams and tables are produced, which show a cross section of the defective weld and provide the above information.

4.3 Trials and qualification tests

The main drawback of weld modification and of some defect-growing techniques is that the exact size of the deposited defect cannot be controlled during manufacturing and can only be monitored by NDT, hence with an inherent measurement error.

In order to overcome these limitations and ensure that the defect obtained is as close as possible to the required size, prior to commencing the manufacturing of the defective welds, trial and qualification samples for all required flaw types are manufactured. These are typically butt-welds in plate or pipe (according to the geometry of the actual defective weld) in which the required flaw types are implanted. A TWI internal qualification sequence is then applied, which is similar to that provided by welding qualification standards (eg ASME IX, BS EN ISO 14614 series) to qualify welding procedures and welders:

1. During manufacturing of the trial plate(s)/pipe(s), the welding parameters and defect deposition techniques are monitored and registered, so that they can be repeated when manufacturing the actual defective weld.
2. After welding, the qualification samples are assessed by a combination of metallographic evaluations (macro and micrographs) and radiographic inspection, aimed at measuring the exact size of the implanted flaws and at assessing their morphology and any induced metallurgical variations.
3. The obtained sizes are compared with the required ones and with acceptance criteria (dimensional tolerances) selected by TWI.
4. A defect deposition procedure is considered satisfactory if these two conditions are satisfied:

• The morphology of the realistic defect is similar to that of the corresponding real defect
• The difference between the actual size of the defect and the required size is within ±0.5mm in through-wall extent and ±1mm in length.

1. If the qualification fails, the procedure is repeated from (1) above.

A number of defect sizes were measured within various confidential projects, for which the required defect through-wall extent and length ranged from 1 to 8mm and 6.8 to 30mm, respectively. The maximum absolute errors measured in the through-wall extent and in the length of the deposited defects were +0.88 (target ±0.5mm) and 1.1mm (target ±1mm), respectively. It should be noted that such unacceptable values were only observed in one instance each and that the average errors measured were +0.16 and -0.2mm for through-wall extent and length, respectively, which are well within the abovementioned targeted limits.

It should also be noted that, as the length of embedded defects was measured by radiographic inspection, it was not possible to detect and size embedded LOSWF defects, due to their tilt angle with respect to the plate/pipe surface. However, the techniques used for such defects allow a very tight control of the defect length (Section 3.2); hence this limitation is not considered significant towards the evaluation of the defect deposition techniques.

4.4 Manufacture of defective weld and final inspection

Once the qualification procedure is complete and it is has been demonstrated that all required defect types can be deposited within the target tolerances on size, the defective welds are manufactured. Following completion of the welds, UT inspection is normally carried out to verify that all required defects have been inserted and are detectable. Any additional indication which does not correspond to any of the required defects is also recorded.

5. Case study

Following the discovery of leaks from two tube-to-header welds in a Waste Heat Boiler, a company requested that TWI identify and evaluate suitable non-destructive testing (NDT) techniques to establish the integrity of the remaining welds. The aims of the inspection were to detect and sentence surface cracks, sub-surface cracks and original welding flaws. TWI performed the following tasks:

• Manufacture of a test block containing six artificial lack of fusion flaws with different sizes
• Development of NDT techniques allowing the detection and accurate sizing of all six flaws (with through-wall sizes of 2mm upwards)
• Establishment of approved procedures for site deployment.

Prior to manufacturing of the test block, extensive trials were carried out due to the small size of the weld and the difficulties associated with obtaining realistic lack of fusion defects by manual welding. The results of such trials are shown in Figure 7 (a, b).

The final procedures were based on:

• A swept beam Phased Array Ultrasonic Testing (PAUT) technique for detection and sizing of flaws embedded/root flaws
• An encircling coil MPI technique for detection of surface flaws.

The PAUT technique was deployed in a tube scanner, which allows full access even where adjacent tubes are closely spaced. Finally, the inspection procedures were approved for site use and were successfully demonstrated to the company at its site.

The equipment is, in general, able to inspect a wide range of pipes, with diameters ranging from 0.84" (21mm) up to 4.5" (114mm), and it can be applied to pipe-to-pipe welds as well as pipe-to-component welds. The PAUT scanner has a height clearance of just 12mm (0.5") allowing inspections in limited access areas, eg where there are nearby obstructions such as adjacent pipes or other structures.

6. Conclusions

1. It has been demonstrated that TWI can reliably produce welded joints with realistic defects, representative of the morphology of the most common defect types.
2. Using the different defect production techniques available at TWI, the dimensional tolerance of the realistic defects (ie actual size of the defect vs required size) can be generally guaranteed within ±0.5mm in through-wall extent and ±1mm in length.
3. A welding qualification procedure has been designed by TWI, whereby welding trials followed by metallographic assessment and NDT inspections are carried out and the results assessed against the acceptance criteria in (2) above. Welding parameters and manufacturing techniques are recorded.
4. To ensure consistency in defect size and morphology, manufacturing of the defective welds is carried out with parameters and operation sequences similar to those recorded during welding procedure qualification.

7. Acknowledgements

The authors wish to acknowledge the colleagues: Dave Howse, Nigel Allison, Mark Tiplady, Rita Banks, Ivan Pinson, Nathan Decourcelle, Bill Lucas and all TWI customers who gave permission to use images and data from confidential projects.

8. References

1. Crutzen S; Lemaitre P; Iacono I Realistic defects suitable for ISI (in service inspection) capability evaluation and qualification. In: NDE in the Nuclear and Pressure Vessel Industries. Proceedings, 14th International Conference, Stockholm, Sweden, 24-26 Sept.1996.
2. Neundorf B; Csapo G; Erhard A, Optimising the NDT (nondestructive testing) of boiling water reactors by using realistic flaws in the cladding. In: NDT at Work. Proceedings, 7th European Conference on Non-destructive Testing, Copenhagen, Denmark, 26-29 May.1998.
3. Development of ENIQ terminology taking into account new standards: Glossary of terms used in qualification, European Commision, DG Joint Research Centre, Institut for Advanced materials, Petten Establishment, Structural Components Integrity Unit.
4. Virkkunen I; Kempainen M; Ostermeyer H; Paussu R: Grown cracks for NDT development and qualification, www.trueflaw.com/pub/BiNDT2008/GrownCracks_gm_red.pdf
5. Lucas W, Making defective welds for Sizewell 'B', Welding & Metal fabrication, March 1992.
6. Söderstrand H, 'ENIQ Recommended Practice 5: Guidelines for the Design of Test Pieces and Conduct of Test Piece Trials (ISSUE 2)' (ENIQ Report No 42), JRC Scientific and technical reports, European Commission.
7. 2010 ASME Boiler and Pressure Vessel Code, Section V: Nondestructive Examination, American Society of Mechanical Engineers / 01-Jul-2010 / 700 pages.