Radiography of Thin Section Welds, Part 3: Additional Flaw Measurements
G A Georgiou (Jacobi Consulting Ltd, London)
and C R A Schneider (TWI, Cambridge)
Paper presented at BINDT Annual Conference 2004, Torquay, 14-16 Sept.
This paper, which focuses on the practical work, is an extension of earlier work first presented at BINDT 2002 (Southport) and subsequently published in INSIGHT. [2,3] This paper accompanies Part 4, the theoretical part of this new work, which is also in this issue of INSIGHT. In essence, further macrographs were taken of the same planar flaws from the earlier work, in order to obtain more physical information about each flaw. In addition, information on other suitable planar flaws was included from a previous TWI study from welds in the thickness range 10mm to 50mm. The effects of including all this additional information on the various practical analyses are presented here.
During 1995-1999, TWI performed several detailed studies on the radiography of large planar flaws in thick-section welds. The work considered a number of issues, such as the capability of 1950's and 1960's radiography, the use of statistical models to predict flaw detectability and the effect of human factors on flaw detectability. The thicknesses studied were generally in the range 50-114mm. The main application of the work was to quantify the capability of the construction radiography performed on the welds of the Magnox steel reactor pressure vessels, but the work also had generic value in providing a better understanding of radiographic capability. The studies confirmed that the Pollitt radiographic model was a valuable tool for predicting whether planar flaws would be detected in thick welds and a large number of contractual reports were produced with some in the public domain. 
The thick section work was then extended to look at the detectability of planar flaws in thinner section welds (thickness range 10-50mm) using both X-ray and gamma radiography. This work was presented at the BINDT annual conference in 2002 as Parts 1 and 2. [2,3]
The current work, and subject of this paper, is a continuation of the earlier thin section work and in essence includes some additional macrographs of the same flaws to provide further measurements of the flaw parameters (e.g.typical gape, through wall extent, position etc). In addition, other flaw specimens, in the same thickness range, have been identified from earlier studies and the relevant flaw measurements have been recorded. With this information from the additional macros and additional flaw data, a number of previous analyses were repeated to see the effect on the earlier results.
This paper concentrates on the practical aspects of the work and associated analyses. A separate paper (Part 4), in this issue of INSIGHT, focuses on theoretical modelling and statistical analysis and will assess the effect of the updated flaw measurements on the earlier predictions on radiographic capability for planar flaws in thin weld radiography.
This new work, while intended for specific applications in the nuclear power industry, should also be of generic value in improving the understanding of radiographic capability as well as evaluating Pollitt modelling for thin weld radiography.
2. Summary of earlier work
Details of the results and conclusions of the earlier work can of course be found in the published references [2,3] . However, it is felt worthwhile setting the scene and recalling the earlier practical work.
- Three butt-weld specimens were manufactured with 12 intended planar flaws
- Spacer plates were used to make up a range of thicknesses up to 50 mm
- A total 144 radiographs were taken (70% X-ray and 30% Gamma)
- All radiographs were based on the minimum requirements of BS EN 1435, class B (the improved technique)
- Whilst minimum requirements do not represent day to day practice, they do represent the minimum performance that should be achieved in practice with class B radiography
- The radiographs were interpreted independently by two PCN qualified interpreters
- For the IQI requirements, both Class A (basic) and Class B (improved) levels were considered
- The sectioning programme, based on fingerprint radiographic and ultrasonic NDT, revealed 21 flaws
- 13 planar flaws were selected for analysis
- The earlier practical results (2) included
- IQI levels achieved by each interpreter
- Detection performance at different FFD's/SFD's
- Detection performance of X-ray vs. Gamma
- Detection performance at different flaw positions (through the use of spacers)
- Detection performance vs. geometric parameters (using TWE, misorientation angle, typical and maximum gape, ligament)
3. Summary of current work
From the 13 planar flaws, 8 were selected for further sectioning and the new macrographs were produced from the opposite faces of the original sections. In effect, this meant that for each of the planar flaws in question, two macrographs were produced at approximately 3mm apart as this was the thickness of material lost in the sectioning process.
In order to obtain the final values of parameters such as the through wall extent (TWE), the tilt, the typical gape, the maximum gape and the ligament, both sets of flaw measurements were used from both sides of the flaw face (e.g.average values for typical gape, the maximum values for maximum gape).
The new range of values for the TWE's (determined by sectioning) and flaw lengths (determined by NDT) were 1 - 8mm and 7 - 15mm respectively. These were not too different from the earlier values, but the typical flaw gapes and maximum flaw gapes saw an increase, in some cases quite significant.
Additional flaw data were selected from another TWI project (4), with welds in the thickness range 10mm - 50mm. These 25 data had TWE's and flaw lengths in the range 1 - 6mm and about 2 - 30mm respectively. These data also contained solidification flaw types and will be referred to here as the 'Phase IV data'.
The values of the flaw parameters obtained as a result of the new macrographs were used in various analyses and these are considered in more detail below.
4. Results and analysis
4.1 The macrographs
Overall, the manufacturing methods used for the intended flaw specimens were successful in producing the type of flaws with the intended TWE and lengths. However, there were no solidification cracks among the 13 planar flaws selected, and the TWEs were generally rather smaller than planned. The failure to manufacture any new solidification cracks has, to some degree, been mitigated by the inclusion of three solidification cracks from the Phase IV data  .
A comparison is provided of macrographs for two particular planar flaws from opposite faces of the section ( Figures 1a and 2a correspond to the earlier macrographs and Figures 1b and 2b are the corresponding new macrographs respectively).
Fig.1a) Specimen W3: A photograph of flaw W3-2A revealed by sectioning (Neg. No. V3554)
Fig.1b) Specimen W3: A photograph of flaw W3-2A revealed by sectioning from the opposite face to Fig.1a, presented as a mirror image (Neg. No. V5796)
Fig.2a) Specimen W3: A photograph of flaw W3-4A revealed by sectioning (Neg. No. V3556)
Fig.2b) Specimen W3: A photograph of flaw W3-2A revealed by sectioning from the opposite face to Fig.2a, presented as a mirror image (Neg. No. V5794)
4.2 Flaw detection as a function of flaw parameters and NDT parameters
Other parameters such as penetrated thickness, the presence of the weld cap, the flaw TWE, the misorientation angle and the typical gape were considered for their influence on flaw detectability. The parameters were considered both singly and in pairs, as it is impossible to show all the relevant parameters in this way on just one graph. This illustrates the benefit of using the 'index of detectability' as discussed in the statistical analysis of Part 4, which includes many more relevant parameters.
In general, the results illustrated in Figures 3, 4 and 5 (data from the 13 planar flaws) support the expected trends that increasing penetrated thickness, increasing misorientation angle and decreasing typical gape all contribute to making flaw detection more difficult. Similar results were observed when carrying out the analysis for the Phase IV data. This is also consistent with similar trends observed in the earlier thick-section weld study. 
Fig. 3. Defect detection as a function of defect misorientation angle and penetrated thickness (ND: Not Detected, BV: Barely Visible, EV: Easily Visible)
Fig. 4. Defect detection as a function of penetrated thickness and defect typical gape (ND: Not Detected, BV: Barely Visible, EV: Easily Visible)
Fig. 5. Defect detection as a function of misorientation angle and defect typical gape (ND: Not Detected, BV: Barely Visible, EV: Easily Visible)
4.3 Realism of manufactured flaws
The methods used by TWI for manufacturing and promoting realistic flaws have been well established. Nevertheless there were never any certainties in the manufacturing process and in some cases the methods used were difficult to control precisely. In this project the problems were made more difficult because relatively small flaws were required.
The gape values for the 13 TWI planar flaws were compared with the database of Oxford  , obtained from published photographs of sectioned flaws in the literature together with other information available at TWI. However, in making comparisons it was important to note that in the current work the typical gape values were taken at only one section position, although for 8 out of the 13 flaws, measurements were taken from both sides of the section. Hence for these 8 flaws two sets of sectioning information were available at about 3mm apart(i.e. the approximate thickness of material loss during sectioning). For typical gape values, only three gape measurements were made (i.e. at the top, middle and bottom of the flaw). From these values a judgement was made as to the typical gape values and these were agreed with the Test House metallurgist and British Energy, although it has to be said there was scope for arriving at different values. In addition to typical gape values, maximum gape values were also measured for each flaw.
One obvious observed difference between the Oxford data and the data here is the ratio of the maximum gape value to the typical gape value. For example, in the Oxford  data the maximum gape value was generally about 1½ times the typical gape value, in particular for lack of sidewall fusion and hydrogen cracks in the HAZ. The ratio of maximum to typical gape in the current project was in many cases much higher and it was felt important to consider this in any comparison.
In order to assess the realism of flaws in the three manufactured weld specimens, the graphs of typical gape values vs. TWE, provided by Oxford  , were reproduced here for lack of fusion and hydrogen cracking in the heat affected zone (HAZ). The typical gape value for each TWI flaw was added to Figure 6a and Figure 6b for lack of fusion and hydrogen cracking in the HAZ respectively. The same approach was taken for the comparison between the Phase IV data and the Oxford data.
Fig.6a) Variation in typical gape with increasing defect TWE for lack of side wall fusion defects in the TWI specimens and similar defects studied by Oxford 
Fig.6b) Variation in typical gape with increasing defect TWE for hydrogen cracks (HAZ) in the TWI specimens and similar defects studied by Oxford 
The typical gape values for the 13 TWI flaws show a reasonable overlap with the Oxford data, especially considering the limitations of the sectioning data for the 13 TWI flaws. There are, however, some instances where the typical gape values from this study were relatively small (e.g. typical gapes of the order 1 mm to 3 mm) compared to gapes for similar TWE values in the Oxford data.
- For 8 out of the 13 planar flaws, new sectioning data has been collected by measuring the opposite face to the original section face (~3mm apart). Among these 8 cases, there are some significant differences in the new sectioning data compared to the earlier data.
- For the 13 planar flaws revealed by sectioning and selected for this analysis and the 25 planar flaws from the Phase IV data, the results for flaw detectability support the expected trends that increasing penetrated thickness, increasing misorientation angle and decreasing typical gape all contribute to making flaw detection more difficult.
- Given the limited sectioning data available, it is felt that the gape values for the TWI flaws in this study show reasonable agreement with the gapes of the natural and artificial flaws found in the Oxford (6) study. There are however, cases where the typical gape values for the TWI flaws were relatively small compared to the gapes for similar TWE values in the Oxford data.
The authors wish to thank British Energy plc and BNFL for funding the work. Special thanks also go to Dr R K Chapman and Mr G S Woodcock of British Energy plc for their support during the project and The Test House for their careful preparation and interpretation of the macrographs. The paper is published by kind permission British Energy plc, BNFL and TWI.
- Chapman R K, Woodcock G S, Wooldridge A B, Munns I J and Georgiou G A: 'The Performance of Radiography for Large Planar Defects in Thick Section Welds'. 7 th European Conference on NDT, Copenhagen, 26-29 May 1998, pp1187-1196
- Georgiou G A and Schneider C R A: 'Radiography of Thin-Section Welds, Part 1: Practical Approach'. INSIGHT, February 2003, Vol.45 No 2, pp116-118, 121
- Schneider C R A and Georgiou G A: 'Radiography of Thin-Section Welds, Part 2: Modelling'. INSIGHT, February 2003, Vol.45 No 2, pp119-121
- Munns I J and Georgiou G A: 'The reliability of radiography on welded specimens 20-75mm thick'. TWI report 621493/1/97.
- Munns I J and Georgiou G A: 'The feasibility of 1950s/1960s radiography in thick section welds'. TWI report no. 621491/1/97.
- Oxford C H: 'A review of the occurrence and morphology of potential planar defects in Magnox steel reactor pressure vessel submerged arc butt weldments'. TE/GEN/REP/0059/97, Issue 1.