X-ray...the practical way

TWI Bulletin, September - October 2003

George Georgiou is an independent NDT consultant and director of Jacobi Consulting Ltd. He is currently under contract to TWI as an NDT consultant and is carrying out research in a number of different areas. In particular he has been directly involved in developing techniques for Computed Radiography in National and European projects. He has had many years experience in mathematical modelling of NDT methods such as radiography, ultrasound and thermography. During the 1990s he was a BSI representative on a number of European committees involved in the development of CEN standards for all the major NDT methods carrying out bespoke pieces of research for the various committees. During much of the 1990s he was section manager of the NDT section at TWI.

Charles Schneider is a Principal Project Leader in the NDT Technology Section at TWI. He has worked in the fields of inspection reliability and theoretical modelling since 1985, holding posts within the Central Electricity Generating Board and its successor companies. Since joining TWI in 1997, his work has included inspection qualification, and the application of statistical techniques to both inspection reliability and fatigue testing.

Recent work on the practical aspects of radiography of planar defects in thin-section welds comes under the spotlight. This article looks at specimen manufacture, radiograph interpretation and analysis of the results.

During 1995-1999, TWI performed several detailed studies on the radiography of large planar defects in thick-section welds. The work considered a number of issues, such as the capability of 1950s and 1960s radiography, the use of statistical models to predict defect detectability and the effect of human factors on defect detectability. The thicknesses studied were 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 also confirmed that the Pollitt model was a valuable tool for predicting the detectability of planar defects in thick welds.

As George Georgiou and Charles Schneider explain this current work extends the original programme to look at the detectability of planar defects in thinner section welds (thickness range 10-51mm). Both x-ray and gamma radiography were used. 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. The Pollitt model will be considered in a future Bulletin feature.

Specimen and defect manufacture

Three defect specimens were manufactured at TWI (W1, W3 and W5) and were intended to contain 12 realistic planar defects with through wall extent (TWE) in the range 2-5mm and nominal lengths of 10mm. The intended defect types included lack of sidewall fusion, centreline solidification cracking and HAZ hydrogen cracking. The methods used to produce these defects are well established at TWI, although there were difficulties with such relatively small intended defect sizes. In some cases more than one attempt was necessary to produce the desired defect type and the sectioning programme revealed a total of 21 defects. Out of these 21 defects, 13 planar defects were selected for analysis ( Table 1).

Table 1 A summary of the final defect parameters for the 13 defects selected for analysis.

The sectioning programme revealed TWEs ranging between about 1-8mm. The lengths of the defects, as determined by NDT, were in the region of 7-15mm. However, the lengths were never verified as sectioning was generally only carried out at one position along the length of each defect. Overall, the manufacturing methods used for the intended defect specimens were successful in producing defect types with the intended TWE and lengths. However, there were no solidification cracks among the 13 planar defects selected, and the TWEs were generally rather smaller than intended. The three specimens were 10mm (ferritic), 17mm (austenitic) and 38mm (ferritic) thick respectively ( Table 1).

Summary of the radiography programme

Since only three specimens were manufactured, spacer plates were used to simulate different thicknesses between 10mm and 51mm. When spacer plates were used, they were usually placed on the source side, but for a substantial proportion (31%) of cases they were placed closest to the film. This was to simulate the effect of defects being at different depths within the welds.

The radiographic procedures used were largely based on Class B techniques ( ie improved), according to the current European standard, BS EN 1435, with class B films (Fuji 80 or Fuji 100). It was decided at the outset to assess the radiographic performance using the 'minimum requirements' of this quality class, that is, using the minimum allowed focus- or source-to-film distances (FFD/SFD) and the highest allowed kV. This of course does not necessarily represent day-to-day practice, but represents the minimum performance that should be achieved in practice. However, whilst the majority of radiographs were taken using 'minimum requirements', a number were also taken using more favourable radiographic settings to provide a comparison. Both normal incidence and angled radiographic exposures were taken.

A single-wall technique was used throughout with the Image Quality Indicators (IQI) always placed on the source side.

A total of 144 radiographic exposures were defined and taken using x-ray (70% of the total) and gamma (30% of the total) using both Iridium 192 and Selenium 75.

Results, analysis and conclusions

The radiographs were independently interpreted by two operators with suitable industrial qualifications ( ie PCN level 2) and who had no involvement in taking any of the radiographic exposures.

A great deal of effort went into the mounting and masking of the radiographs ( Fig.1) in order to minimise the risk of the interpreters learning the locations of defects on the different specimens, bearing in mind that there were only three specimens. The masking restricted the viewing window and meant that not all potential defect images were visible to the interpreters on every exposure. Further effort went into organising the order of the radiographs so that each interpreter received more challenging radiographs first in order to minimise the risk of the interpreter learning the defect locations from the easier exposures.

Comparison of interpreters: Achieving IQI sensitivity levels

An analysis of the interpreters' report forms shows that each interpreter reported at least Class A IQI sensitivity level in nearly all cases. Moreover, for a substantial proportion of the cases, each interpreter was reporting at least Class B IQI sensitivity level (40% for interpreter X and 67% for interpreter Y).

Comparison of interpreters: Overall detection of defects

On the 144 mounted radiographs viewed by each interpreter, there were a total of 284 cases where a defect image potentially lay within the viewing window. Each interpreter was asked to record each visible indication as either easily visible or barely visible, using a sample radiograph of some notches to make the distinction. The inspection results were subsequently analysed to determine the detection performance of each interpreter, using the known defect locations to determine whether or not each of the 284 potential indications had actually been detected. Interpreter Y detected more defects than interpreter X, by a margin of 4%. The higher detection level of Interpreter Y was consistent with reporting a higher average IQI sensitivity level (see above).

Comparison of defect detection at different FFD/SFD

Whilst the vast majority of radiographic exposures were taken at the minimum FFD/SFD settings ( ie 350-760mm), some were repeated at longer FFD/SFD ( ie 1000mm) to assess the difference in radiographic performance.

When all reported indications on the radiographs were considered there were about 10% fewer indications reported by each interpreter using minimum FFD/SFD compared with longer FFD/SFD settings. However, if the analysis was confined to the 13 selected planar defects there were only very slightly fewer of these reported at minimum FFD/SFD than at longer FFD/SFD.

Comparison of defect detection between x-ray and gamma

When taking all reported indications into account, each interpreter reported substantially more indications (between 30% and 40%) for x-ray radiography than for gamma radiography. This was consistent with more IQI wires being reported for x-ray than for gamma.

However, if the analysis was confined to the 13 selected planar defects, the interpreters reported only marginally more of these using x-ray radiography than when using gamma radiography, but there were also a few specific cases where the opposite was observed.

Comparison of defect detection for different defect positions

Spacer plates were used to simulate different defect positions through the specimen thickness. This change of defect position would also change the geometric unsharpness for each defect.

To simplify the comparison, only the corresponding normal incidence radiographs were selected. When all the reported indications were taken into account, slightly more indications were reported when the spacer plates were closest to the source (ie lower geometric unsharpness) as might be expected. However, there were also a few indications reported with the spacer plates closest to the film ( ie higher geometric unsharpness) that were not reported with the spacer plates closest to the source.

When the analysis was confined to the 13 selected planar defects, the defect detection results showed there was hardly any difference between the two geometric set-ups.

Defect detection as a function of other parameters

Other parameters such as penetrated thickness, the presence of the weld cap, the defect TWE, the mis-orientation angle and the typical gape were considered for their influence on defect 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 analysis showed, for example, that in general the interpreters' results supported the expected trends that increasing penetrated thickness, increasing mis-orientation angle and decreasing typical gape all contribute to making defect detection more difficult. This is consistent with similar trends observed in the earlier thick-section weld study.

Sectioning

The sectioning programme was guided by the results of fingerprint radiography and ultrasonic NDT. Each defect was sectioned at one position along its length and its TWE, tilt and gape were measured.

Overall, the sectioning programme revealed defects at each of the sectioning positions, as determined by fingerprint radiography and ultrasonic NDT, except for a couple of cases.

Two typical sectioning photos are given in Figures 2 and 3.

Realism of manufactured defects

The methods used by TWI for manufacturing and promoting realistic defects have been well established, but in some cases the methods used were difficult to control precisely. In this work the problems were made more difficult because relatively small defects were required.

The gape values for the 13 selected planar defects were compared with the comprehensive database at Oxford. However, in making comparisons it was important to note that in the current work gape values were provided at only one slice position. In addition, only three gape measurements were provided ( ie at the top, middle and bottom of the defect) as well as the maximum gape value. From these values a judgement was made as to the typical gape value ( ie average gape).

In order to assess the realism of defects in the specimens here, the graphs of typical gape values vs. TWE, provided by Oxford, were reproduced for lack of fusion and hydrogen cracking in the heat affected zone (HAZ). The typical gape value and maximum gape value for each of the 13 selected defects were added to the Oxford graphs.

The TWI defects show a reasonable overlap with the Oxford data, considering the limitations of the sectioning data for the TWI defects. There are however, some instances where the typical gape values from this study were much smaller ( eg typical gapes of the order 1-3µm) than the gapes for similar TWE values in the Oxford data.

Acknowledgements

The authors wish to thank R K Chapman and G S Woodcock of British Energy plc for their support. The paper is published by permission of the Industry Management Committee (IMC) of the UK nuclear power plant licensees, who also funded the work.