The inspection principles have been previously described by the authors (1) and are summarized as follows;
The weld is divided into several zones (see Figure 1). Each zone covers approximately 2-3 mm of thickness of the weld. Ultrasonic transducers are designed and positioned to investigate each zone from both sides of the weld centreline (See Figure 2). The array of probes is moved around the girth weld by a motorized carrier which moves along the same track the welding apparatus uses. Ultrasonic signals received by the instruments are monitored by electronic gates. Both amplitude of signal and time of flight of the signal in the gated region are monitored. The region gated is from just before the weld preparation line to just after the weld centreline. Gated output, either time or amplitude or both, is digitized and displayed in a chart format. An operator evaluates the chart results and makes a decision as to weld acceptability based on the length of signals exceeding a threshold as set out in specifications and regulating codes.
Inspection systems using these principles have proved to be very effective and statistics have shown good correlation with most radiographic calls made. The advantages of the system are speed and accuracy; e.g. a 48" diameter weld can be scanned AND ACCURATELY EVALUATED in about 2 minutes. Inspection is carried out as soon as the metal temperature has cooled sufficiently to permit the probes to be placed on the metal (typically four to five welds behind the last welding crew). With inspection results available so readily the UT system provide a process control for the welding.
Not only does the system locate the defective areas along the length of the weld but by means of discrete zone discrimination it can also establish the vertical position and vertical extent of the defect. This latter advantage over radiography permits more precise engineering evaluation as to the seriousness of the defects located.(2) This is the principle of the applied Engineering Critical Assessment (ECA).
Maintaining zone discrimination is a critical aspect of the technique. Problems encountered in the early stages of development led to the discovery that velocity variations were causing significant zone changes of the beams. These variations in velocity were from two sources; steel type and temperature. Acoustic velocity variations from 3100m/s to 3400m/s were found to exist in the steels from various pipe mills. As well, the effect of temperature on the acoustic velocity of the perspex refracting wedges became significant. For example; in winter projects the ambient air temperature can drop to -50(C while the steel temperature after welding is around 40(C. Winter projects require use of heated calibration blocks and 60% methyl alcohol in water is used as a recyclable couplant.
Development of ultrasonic girth weld inspection systems has continued in an effort to improve reliability of evaluation, as well as detection of flaws and sizing of defects. This development has taken the form of increased data acquisition. Increasing the amount of information provides the operator with a better opportunity to make accurate evaluations. However, the speed of inspection cannot be compromised by slower acquisition speeds nor can the inspection time be increased by the operator spending more time sorting through more data. Only with the developments in computer technology have these two problems been overcome.
Aspects of recent developments can be grouped into three areas;
Unlike manual scanning, in the mechanised UT systems only lateral motion is possible as the probes are moved around the weld. The echo envelope that is provided as the hard copy of the output for the gated peak amplitude and transit time is used by the operator to evaluate the weld. If the operator had the ability to follow the subtle changes seen on the A-scan presentation used by manual ultrasonic inspectors it would afford some of the advantage of echo dynamics seen by the manual scanner. The use of echo dynamics incorporates how a signal of interest moves with respect to not only probe movement but also to any other signals that may be present on the screen. The operator of a mechanised system would be given a further advantage if the whole A-scan could be collected instead of just the maximum amplitude and time in a gate. This can be done and the result of stacking collected A-scans is usually termed a B-scan. Some (5) differentiate between a B-scan which is generated when the probe motion is at right angles to the weld axis and a D-scan where probe motion is parallel to the weld axis, but generally B-scans are considered stacked A-scans irrespective of the direction of probe movement.
Only with the recent availability of high speed computer hardware and increased storage capacity has the possibility of B-scan presentations on multiple channels been practical. This B-scan mapping, when used in conjunction with the amplitude and transit time information, allows the operator to see trends associated with geometries and to differentiate geometric reflectors from defect reflectors.
Figure 3 illustrates root bead misalignment and the improved ability to discern that from a root defect.
Note: Mapping and amplitude/transit-time channels are arranged symmetrically on either side of the root bead map channels in the middle of the image. Amplitude of signals in the gate is represented by dark solid lines in the Root, LCP, Hot pass and Fill channels. Transit time of signals in these gates is represented by bands of grey-shading.
Symmetry of the porosity indication in Figure 4 would be the same sort of evidence used to evaluate porosity when seen in the porosity maps for the Fill regions of the weld.
An operator inspecting a heavy wall linepipe girth weld (about 17mm thick) may have 18 amplitude and time channels to monitor as well as four B-scan channels plus coupling monitor channels. Fitting these all onto the hardcopy and evaluating the results would suffer two main problems. First the delay of waiting for the hardcopy increases time spent on the weld. Secondly, fitting over twenty channels of information on an 8 inch (20cm) wide piece of paper provides the operator with very small traces to examine.
To overcome these problems operators now use large screen monitors to display the information. Twenty inch (50cm) monitors or larger allow nearly double the size of image to look at compared to paper hardcopies. Also, the images on the screen are in colour. This feature can be used to bring the operator's attention to potential problem locations. As the scan progresses information is buffered and the data scrolls off the screen. When the scan is completed the operator can scroll the data up and down and use tools such as magnification windows and individual A-scan viewing. With the colour enhancements and larger image, an operator now makes all evaluations based on the video image. The paper hardcopy is merely printed as a permanent record and stored as per the regulatory requirements of the governing body.
Calculating memory requirements for B-scans requires a knowledge of several items.
Digitizing an A-scan is the first step in constructing a B-scan. But each sample must be saved to computer memory, therefore larger scan lengths and larger time of the gated period, require more memory than small scans and short gated times.
For a simple B-scan using a 4 MHz contact 70° angle beam probe on a 16mm thick plate we would like to gate the entire half skip distance for display. We would use the recommended minimum ADC rate of 16 MHz. We must also consider that nearly 100mm time equivalent is traversed by the transverse wave (two way trip in pulse-echo); hence (50 x 2) ( 5.9 = 31.3 microsecond (µs).
At 16 MHz ADC 16 samples are made each µs, so for the gated time of 16.9µs, 500 samples will be recorded for each A-scan. At each point 8 bits of amplitude information are collected (8 bits = 1 byte). If our B-scan is to be collected along a 42" diameter weld, a 3400mm travel is needed. If an A-scan is collected at 1mm intervals, the data generated would be:
A single 42" weld with both the traditional amplitude & time strip-chart style information plus four B-scans could result in a data file of over 7MBytes.
RTD files require setup parameter information as well as data file information but typically on the order of 1Megabyte per weld is stored in a 42" diameter weld. This significant reduction of file size is possible by a clever technique that stores only a bit mapped image of the amplitude along the gated region collected by the flash digitizer for each B-scan map plus the traditional amplitude and transit time stripchart information. Although the entire waveform is not stored there is sufficient information for the operator to see trends associated with geometries such as weld root bead and weld cap and it would be rare for the operator to need to review the waveform. This B-scan map would be similar to the B-scan described by McMaster back in 1959.
Shaw Pipeline Services saves the entire A-scan but also achieves much smaller file sizes by employing data compression for saving the files.
| 55 Kbyte|
Figure 5: Positioning scanning head on 48" diameter pipe on the Canadian prairies. Daytime summer temperatures often reach 35°to 40°C. (Shaw Pipeline Services)
| 105 Kbyte|
Figure 6: UT inspection truck ready to test within five welds of last weld completed. Winter work in rock cuts in the Canadian Shield near the Manitoba-Ontario border (Shaw Pipeline Services).
| 65 Kbyte|
Figure 7: Loading the scanning head on a 42" diameter pipe in the winter northern Ontario. Equipment must also be able to perform under conditions of daytime highs of -40°C. (RTD Quality Services Ltd.)
In a recent study by one of the authors (Ginzel), a 20" diameter girth weld was tested using a modified immersion technique. All significant defects identified by radiography were detected and sized using the TOFD technique (See Figure 8). When used independently the TOFD technique does not identify the side of the weld where a defect is located. This drawback can be addressed by using two or three pulse-echo probes on either side of the weld and collect the amplitude and time response as in the present systems. The requirement for pulse-echo probes is also practical for areas where high-low conditions could mask flaws from the TOFD setup.
TCPL's TWE-1 specification is a very comprehensive specification and only minor modifications would have to be made to address inclusion of TOFD technique.
The single greatest advantage of TOFD is its ability to precisely provide accurate vertical extent of a defect. Accuracies of 0.1 to 0.2 mm are possible using TOFD. This could greatly extend allowable defect lengths in Engineering Critical Assessment calculations when applying the code CSA-Z184 (6 or the newer ISO/CSA-Z662 which combines both gas and oil pipeline construction.
Application of mechanized UT systems will, in future, need to adapt to other welding technologies. Until 1995 only one company provided a mechanized welding process used in Canada. This American company has provided mechanized welding on projects all over the world but ironically it is not used on pipeline construction in the USA. More recently two other systems have been developed, one from Italy and another from France. These have slightly different configurations but inital studies on pilot projects indicate mechanized UT will be as effective for these new systems as it has been for the older system.
Mapping enhancements, video evaluation techniques and mass storage of data have been features added to the systems. These have all been based on computer technology developments which will probably be the foundation for future development. Future improvements may require that even more information be acquired. If this presents the operator with a longer required evaluation time it may be necessary to increase data collection speeds. This would mean the probe arrays would be moved faster. Not only will there be mechanical concerns to consider such as maintaining coupling and inspection apparatus integrity, there may also be a need to use faster digitizing boards, faster CPU's and higher pulse-repetition frequencies in the ultrasonic pulser-receiver.
Although these potential developments may change the appearance of the apparatus seen on the pipeline right-of-way, the existing specification will be able to address nearly all future development without significant modification.
Merv Hoff is the senior NDT coordinator for TransCanada PipeLines Limited.
He can be contracted at
111 5th Avenue SE
tel. (403) 267-6401 fax (403) 267-6242
|Ed Ginzel is an independent consultant with the Materials Research Institute.
He can be contacted at