Summary

Mechanised and automated ultrasonic equipment is now gradually being taken into use as an alternative to radiography for examination of pipeline girth welds during pipelaying. When this is done, one meets with the requirement to formulate ultrasonic acceptance criteria to replace the radiographic workmanship level criteria used so far. To address this topic, and later also fitness-for-purpose acceptance criteria, on a general basis, a project has now been launched under the auspices of the International Pipe Line & Offshore Contractors Association. The objectives of the project are to develop ultrasonic acceptance criteria equivalent to the presently used radiographic ones, and to demonstrate their adequacy. The project therefore includes an experimental reliability data acquisition and verification programme, which is outlined. Results to be achieved, and evaluations to be done, are exemplified using available non-destructive examination reliability results from other investigations.

Introduction

Mechanised and automated ultrasonic pipeline girth weld examination systems (AUE), including the use of time-of-flight diffraction (TOFD), are now so far developed, that they seem to be able to serve as a reliable alternative to radiography during pipeline construction (1), (2), (3) and such equipment has also been taken into use for some land pipelines (4), (5). During offshore pipe laying, where time is very limited for doing non-destructive examination (NDE) between welding and lowering of the pipe onto the seabed, AUK have so far only found limited application. The available systems do, however, also for this type of application bear potentials, as they will shorten the time required for NDE, and thus contribute to shorten the fabrication time, making the pipe laying more cost-effective. In addition, they have the potential of revealing the more severe defect types like lack of fusion, which are not always readily detected by radiography, and, through the better quality control work, reduce repair rates.

The potential use of AUK during offshore pipelaying is now also been addressed in newer codes and standards, for instance the 1996 DNV Rules for Submarine Pipeline Systems (6).

One of the questions raised, when evaluating AUK to replace radiography, is the question of the reliability, and how the results should be interpreted, i.e. how should acceptance criteria be formulated. This topic has been addressed by users of AUK (4), (5), and will now be addressed on a general basis in an international project under the auspices of the International Pipe Line & Offshore Contractors Association (IPLOCA) with support from pipeline authorities, operators, contractors and NDE service companies. Further reasoning behind establishing the project, and why to develop acceptance criteria for pipeline girth weld defects, can be found in (1).

The Task

It is the objectives of the IPLOCA project to develop acceptance criteria for AUK for use during pipeline construction to inspect the fabrication girth welds, and to document the adequacy of these acceptance criteria for the task in question. This is primarily done with respect to replacing the fabrication quality control radiography, which in this case represents the accepted concept. Later also the topic of fitness-for-purpose acceptance criteria will be addressed.

With respect to the replacement of radiography by AUK the following items are crucial

• AUK must do 'as good a job' as radiography in revealing significant defects
• AUK must not produce to many false calls, or lead to increase in repair rates, rather a lowering, in order to provide for cost-effectiveness
• The documentation of the AUK performance must, in order to be useful, have sufficient statistical confidence

The Route

An assessment programme for the replacement of radiography by AUK can follow the outlines given in the Nordtest Guidelines for Replacing NDE Techniques with One Another (7), which also contains confidence criteria for comparison and replacement of the methods, designated Elementary Detection Criteria: The lower one-sided 95% confidence level of a Probability of Detection (POD) value of both the Replacement (in this case AUK) and Reference (radiography) Technique shall both be equal to or above the average value for the Reference Technique minus 0.1.

For the experimental reliability verification, it is important to have a sufficiently high number of defects of relevant types, sizes and locations. Approximately 50% of the defects, should for instance be lack of fusion ones, as this corresponds to available defect statistics. Further, it is important to cover relevant material qualities, weld preparations and thicknesses. Modern pipeline steels (API SL Grade X60, etc.) show a varying degree of anisotropy and have a tendency to contain segregations, which may impede the ultrasonic examination, and put special requirements on for instance calibration and gate settings. This, as well as examination at elevated temperature (around 100 °C), is particular to (offshore) pipeline girth welds and their examination, and must be reflected in a validation programme. Some of this validation might be limited, however, to showing that these factors are controllable by proper calibration and special precautions. In spite of the high degree of automation, human elements related to for instance calibration and decisiontaking are still important, and must be considered in the programme by having more inspection teams involved.

During planning and evaluation of results due regard should be taken to the results from other studies, like the Dutch Welding Society (NIL) thin plate project (8). Some of these results are, however, due to the reporting format used, not always easily accessible and thus useful for the task in question.

Examination should be done by both AUK and radiography in order to do a direct comparison of the two techniques. During the experimental verification not only recording of acceptance or rejection according to set acceptance criteria should be done, but more primary information like echo amplitudes, TOFD measured heights, lengths, locations, assessed type, etc. recorded. This will allow later tuning of the AUK acceptance criteria to fit radiographic criteria by adjusting threshold parameters.

When analysing results one should bear in mind, that quality control acceptance criteria are somewhat arbitrary in nature, and that quality control acceptance levels specified by physical, defect related parameters (type, size, location) represent what should be detected on an average by the quality control system, i.e. with a POD of approximately 0.5.

The Results

The IPLOCA project is at present in its planning stage, and no results are thus yet available.

The type of results achievable and evaluations to be done will thus be illustrated using results from the Nordtest NDE Programme on radiography and manual ultrasonic examination (9). Neither the radiographic, nor the ultrasonic data can be regarded representative for the pipeline girth welds: The bevel preparation encountered on pipeline girth welds is steeper than for the welds in the Nordtest programme, in the programme radiographic results were evaluated according to the IIW 1952 collection of radiographs (10), the modern AUK with probes dedicated to each part of a weld is far from the manual examination performed (even if the physics remain more or less the same), etc.

Nomenclature
The abbreviations used further below are:

RL: Ultrasonic reference level, corresponding to the echo from a 3 mm dia side drilled hole
E: Ultrasonic echo amplitude
L: Measured defect length
h: TOFD measured defect height
t: Material thickness


Acceptance Criteria
In the examples shown slight modifications of the acceptance criteria of the DNV pipeline rules (6) were used. For ultrasonics, these are very close to those of prEN 1712. In addition, acceptance criteria based on simulated TOFD measured heights¹ to replace the echo amplitude criteria of the DNV rules were applied. The applied rejection criteria, in a simplified version, are:
¹ Assuming no mean error in measurement and a standard deviation of 1.3 mm based on values found in (8).


Radiography	Ultrasonics2	Amplitude based	TOFD based
			E > RL- 12dB and
L > t, max.2 5mm	L < t/2:	E > RL+4dB	h > 4mm
L < t: IIW degree Red or Brown2	t/2 < L < t:	E > RL -2dB	h > 2mm
Evaluation as crack	L > t::	E > RL- 6 dB	h > l mm


² For surface defects. There are slightly other length criteria for embedded defects and thick plate (> 25 mrn thickness).
³ Reflecting also the rejection of spherical cavities or inclusions of 3 mm diameter or more.

Probability of Detection Curves
To describe the ability of an NDE technique to reveal defects, POD⁴ curves, or POD values for groups of defects, can be used and compared utilising the Elementary Detection Criteria of the Nordtest comparison and replacement guidelines (7) (see Clause 'The Route' above). The attached Figure 1 gives examples of such curves based on the above acceptance criteria and the data from the Nordtest NDE Prograrnme (9). Most of these diagrams contain maximum likelihood regression POD curves (thick lines) with lower onesided 95% confidence limit (thin lines), and for the reference radiography curve also the regression values minus 0.1 as used for the Nordtest criteria (thick chopped / greyed lines). All curves give POD as function of defect height, which is regarded the most predominant defect severity parameter.
⁴ POD is here also used a synonym for Probability of Rejection: The NDE system. including the acceptance criteria used. can be regarded a 'black box' either detecting a defect, or not.

Comments	Figure 1a - 1i: Examples of POD curves for different techniques and acceptance criteria (see comments). The data have been taken from the Nordtest NDE Programme (9).
Figs. la. and lb. show grouped POD observation data with fitted curves. Fig la. contains all the radiographic data contained in the Nordtest Programme, whereas Fig. lb. only contains approximately 150 randomly selected observations. In the latter case, the lower 95% confidence limit is just above the fitted mean curve minus 0.1, and the NordtestElementary Detection Criteria still fulfilled for the Reference Technique. Fig. lb. thus gives an indication of the number of observations required to fulfil the criteria.	Figure 1a Different defect types Radiography Minus 0.1 R Curve Number of Defects: 601 Number of Observations: 3414
	Figure 1b Different defect types Radiography Minus 0.1 R Curve Number of Defects: 153 Number of Observations: 3414
Figs. 1c. and 1d. give comparative curves for a mixture of all encountered defect types and, the pipeline girth weld important defect type, lack of fusion. The echo amplitude criteria can be regarded satisfactory for all defect heights, whereas the TOFD acceptance criteria are satisfied according to the Nordtest requirements above defect height 2 mm, corresponding to the radiography 0.5 POD. This must also be regarded satisfactory. In addition, it can be noted that the TOFD curves are steeper than the echo amplitude based curves. This is of course due to the better correlation between TOFD measured defect height and true defect height, than between echo amplitude and height, and implies a better quality examination work with TOFD, as less small and more large defects are revealed (please keep though in mind that the TOFD POD data are a simulation). The total (small and large defects) detection / rejection rate for TOFD is 45% compared to 46% for radiography.	Figure 1c Different defect types R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
	Figure 1d Lack of fusion defects R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
Fig. 1e. shows the inadequacy of the set acceptance criteria for ultrasonics for porosity. In order to reveal pores with ultrasound higher sensitivities than those used must be applied: At 30 mm distance a spherical cavity of 4 mm 0 gives an echo 11 dB below that of a 3 mm 0 side drilled hole, etc. One way to handle this problem is to do an evaluation of the severity of pores (and similar for slag inclusions), and possibly accept relaxed acceptance criteria compared to those for radiography, or use special ultrasonic pattern recognition techniques to map porosity.	Figure 1e Porosity R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
Figs. 1f. and1g. compare POD curves for lack of fusion in different thickness groups (average wall thicknesses 13 and 28 mm), and show that the ultrasonic echo amplitude criteria are not adequate for the thinner material.	Figure 1f Lack of fusion defects Plate thickness < 15mm R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
	Figure 1g Lack of fusion defects Plate thickness > 15mm R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
Fig. 1h. show the failure of the set ultrasonic echo amplitude acceptance criteria for lack of fusion of less length than wall thickness. Some caution is, however, required, when making this observation: The used acceptance criteria for radiography allow lack of fusion of length below wall thickness, and the detections made are due either to misinterpretation of defect type or length, or the evaluation as IIW degrees Red or Brown incorporated in the acceptance criteria. There is, however, as further analysis shows, a POD defect length dependency, and an evaluation of this length dependency is required, when lengths are not 'naturally' distributed, or defect significance is also length dependent, as when plastic collapse is the most relevant defect mechanism related to the girth weld defects. Further, a distinction may have to be made between surface and embedded defects.	Figure 1h Lack of fusion defects Defect lenght < Wall thickness R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve
	Figure 1i Lack of fusion defects Defect lenght > Wall thickness R: Radiography U: Ultrasonics TOFD: Time of flight diffraction Minus 0.1 R Curve

Fitness-for-Purpose Acceptance Criteria

The significance of defects can in general be assessed by fracture mechanics or large scale / wide plate tests, and is dependent on defect parameters like type, size and location, and on non-defect related parameters like stress, environmental factors and material properties. The severity of pipeline girth weld defects is determined from a number of factors and will be different for different application areas: Internal pressure and variations, laying forces, reeling forces, free spans and vibration, accidental loads, corrosion and erosion initiation, etc. For pipelines, absolute criteria are often based on critical defect sizes for plastic collapse, which are defect length, height, and wall thickness dependent (11). (12).

Formulating NDE acceptance criteria for a fitness-for-purpose assessment is a much more difficult task than for quality control. In principle every individual defect, or combination of defects, which might lead to an unacceptable condition of a construction, or a high probability for this, should be revealed. In order to formulate acceptance criteria it will be necessary to take into account defect severity, for instance in the form of probability of failure as function of defect severity parameters, anticipated defect distributions, or distribution forms, - and the anticipated number of defects. A formulation might be based on the reliability updating achieved through NDE results. In order to compensate for the unreliability inherent in many NDE methods, this might, for a number of applications, lead to acceptance criteria close to those for traditional quality control.

Ideally, a fitness-for-purpose approach should not be made by evaluating a single NDE indication. NDE should first be performed, the defect contents then assessed based on the NDE results, and acceptance or rejection of a construction as a whole (or defined part thereof) made from the combined severity of all defects present. Corrective actions as required should then be implemented. There is, however cases, for which this approach is not practical. One such important case is during offshore pipelaying, where time is short between welding and lowering of the pipe into the sea. A decision on acceptance or not of each weld has to be made within a short time window. An assessment of a pipeline as a whole is precluded. To meet this situation a special formulation of (discrete) defect or single weld based acceptance criteria has to be made reflecting the overall integrity of a pipeline.

References

1. F.H. Dijkstra, J. de Raad: Why develop acceptance criteria for pipeline girth weld defects? European-American Workshop Determination of Reliability and Validation Methods of NDE. Berlin. 18 - 20 June 1997.
2. P. Verhaeghe: Practical application of automated ultrasonic inspection on pipeline welds. 2nd International Pipeline Technology Conference. Ostend, Belgium. 11 - 14 September 1995.
3. J.A. de Raad, F.H. Dijkstra: Mechanized UT now can replace RT on girth welds. 2nd International PiPeline Technology Conference. Ostend, Belgium. 11 - 14 September 1995.
4. D. Hodgkinson, D.V. Dorling, A.G. Glover: Mechanized ultrasonic testing of pipeline girth welds produced by mechanized gas metal arc welding. 2nd International Pipeline Technology Conference. Ostend, Belgium. 11 - 14 September 1995.
5. H.A.M. van Merrienboer: Field application of mechanized US inspection of large-diameter pipelines: 2nd International Pipeline Technology Conference. Ostend, Belgium. 11 - 14 September 1995.
6. Rules for Submarine Pipeline Systems. Det Norske Veritas. OSIO, Norway. December 1996.
7. Guidelines for Replacing NDE Techniques with One Another. NT Techn Report 300. Nordtest. Espoo, Finland. October 1995.
8. Non destructive testing of thin plate. Doc.no. NDP 93-40. Nederlands Instituut voor Lastechniek. Voorschoten, The Netherlands. March 1995.
9. O. Forli: Reliability of radiography and ultrasonic testing. 5:e Nordiska NDT Symposiet. Espoo, Finland. 26 - 28 August 1990.
10. Collection of Reference Radiographs of Welds in Steel. International Institute of Welding. 1952.
11. R. Denys: A plastic collapse based procedure for girth weld defect acceptance. Proc. Int. Conf. on Pipeline Reliability, Vol. II, Paper VIII. Canmet, Pergamon, Calgary, Canada. Pergamon Press. 1992.
12. F.J. Horsley, A.G. Glover, R. Denys: An assessment technique for defects in under and over matched pipeline girth welds. Conf. Proceedings l 1th PRCI-EPRG Pipeline Research Conference. Paper 30. Arlington, Virginia. 7 - 11 April 1997.