Reducing the burden of inspection29 December 2001
Reducing the time cost and radiation exposure incurred in inspecting steam generator tubes is a goal of all operators. Positive results have been obtained from field trials of R/D Tech's new X-Probe.
Eddy current inspection of steam generator tubes usually involves using rotating probes for high-resolution, high sensitivity inspection of tubes in four regions: tube sheet, top of tube sheet, tube support plates, U-bends and special interest areas.
Rotating probes are very slow, imposing inspection speeds that are, typically, under 0.5in per second. In comparison, bobbin coil probes run faster, up to 40in/sec, but are less sensitive. The high-sensitivity rotating probes used recently have surface-riding coils, which give the probe a very short life.
The slow speed and short life increase inspection time and consumable cost, and incur radiation dose to the personnel replacing the probes.
R/D Tech has developed a high-resolution ECT array probe in close collaboration with Atomic Energy of Canada (AECL), Nuclear Engineering Ltd (NEL), Electric Power Research Institute (EPRI), and Electricité de France (EDF). Known as the X-Probe, the new probe overcomes the limitations of rotating probes while providing inspection speeds comparable to that of bobbin probes.
Numerous field trials were conducted during the spring and autumn of 2000. The trials demonstrate that the X-Probe can provide considerable savings on inspection time and cost, consumables, and radiation exposure, without compromising data quality, flaw detectability, or introducing significant inspection transients. In addition, extended probe life will reduce the amount of contaminated waste.
The probe has recently been qualified at EPRI to appendix H, for the detection of a number of degradation mechanisms.
The X-Probe combines a bobbin coil and a high-resolution, high-sensitivity array into a single probe. The array is made of three rows of 16 coils each for the 3/4in and 7/8in versions. A 3x14 pattern is used for the smaller 5/8in version.
The coils operate in transmit-receive mode. This provides two important benefits. First, the signal phase angles are, for all practical purposes, not affected by lift-off. Expansions and dents are calibrated to produce horizontal signals and all other flaws produce vertical signal components.
As a result, there is minimal interference from dents, transition zones and expansions, greatly simplifying data analysis. Second, the probe is capable of clearly separating axial indications from circumferential indications, providing data that is easier to analyse.
The probe coils are multiplexed over a number of time slots. Not only are the active coil pairs different from one time-slot to the next, but a given coil may be used as a transmitter in one time slot and then as a receiver in another time slot. Altogether, there are 32 separate zones sensitive to axial indications and 32 separate zones sensitive to circumferential indications. Each of the 64 sensitivity zones is operated at four different frequencies.
It is clearly not practical to rely on conventional strip charts as the main method of detection when analysing data from a probe that has this many coils. To overcome this, new software was necessary. The new software, known as MultiView, takes the data from the 64 sensitivity zones and groups them into eight "composite" or "merged" channels. This allows convenient display of the information on 3D-views that are very similar to those used for the data analysis of rotating probes. There are separate views for circumferentially operated channels and for axially orientated channels as well as for each of the four operating channels of the array, yielding a total of eight 3D views for detection.
The array is operated at four frequencies. The 200kHz and 300kHz absolute channels are used for detection from 3D-views, and the other frequencies are used to confirm or size.
Detailed analysis is performed from the industry-standard Lissajou plots. Mixes are not required. From the analysis a few simple rules can be developed:
•Expansions and dents produce horizontal signals whereas all other indications produce a vertical signal.
•Cracks produce a vertical signal component on either axial or circumferential channels, depending on their orientation. The corresponding Lissajous exhibit the familiar phase shift with increasing frequency.
•Volumetric indications (wear, IGA, pitting, manufacturing burnished marks) produce vertical signal component on both axial and circumferential channels. Cracks at IGA clearly show up as crack-like indications.
•Copper and magnetic deposits produce negative vertical signal components on both axial and circumferential channels. This is an important feature of the probe as clear identification of deposits considerably reduces the number of false calls.
MultiView was designed to handle standard bobbins and rotating probes but also to include analysis tools optimised for the array probe. The software allows nearly fully automatic calibration of the X-Probe in under a minute, an operation that could take a whole day using conventional software. The software runs on ordinary PCs, equipped with a suitable amount of RAM, under Microsoft Windows NT or Windows 2000.
Field trials have now been conducted at Takahama 2, Palo Verde, Doel 2, Watts Barr, San Onofre, and Salem. There were also trials at Bruce and Gentilly 2, both CANDU plants, and one at Oconee (employing once-through steam generators). Most plants have elected to try the probe on tubes scheduled for plugging, allowing a direct comparison with data produced by the rotating probe of record.
The first field trial was at Japan's Takahama 2 in October 1999. The steam generators at Takahama 2 have already been replaced and the new ones have no known degradation mechanisms. The plant was selected solely on the basis of outage schedule.
NEL worked with five Japanese PWR utilities to carry out collaborative research in order to evaluate the insertion ability, handling ease, durability and quality of data concerning X-Probe.
This trial confirmed that the X-Probe has insertion ability and ease of handling equivalent to those of a bobbin probe. The inspections were performed with dual X-Probes, at 400mm/sec and at 500mm/sec. The tubes inspected included the Row-4 U-bend. After inspecting about 400 tubes each both probes were sound, and the quality of the data from the structures was good.
The most extensive trial so far was at Doel 2 in Belgium, where 5600 tubes were inspected over the first 43in, including an in-line calibration tube.
Conclusions from the field trials have been extremely positive:
•Most of the discrepancies initially found during the trials were resolved after careful review. No significant inspection transients are now expected from the use of the X-Probe. The trials confirmed once again that analyst training, as well as plant-specific training, is very important.
•Data is easy to analyse. The number of potentially false calls was considerably reduced, partly because the X-Probe can positively identify magnetic and non-magnetic deposits. In the early trials analysis speed was rather slow: the software has since been enhanced and analysis speed seems to be similar to that of rotating probes.
•The probe is suitable for full-length tube inspections: over 1000 tubes, some with U-bends with six inch bend radius, have been inspected over their full length during the various trials, with excellent data quality and probe life. Upcoming software, field tested in late 2000, will allow the high-resolution inspection to be limited to areas determined by the inspection scope (specified supports, tube sheet and U-bends) while retaining only bobbin data elsewhere. This will limit the amount of data to be analysed to the strict minimum.
•Inspections at 500mm/sec (20in/sec) are possible. At Doel the inspection speed averaged nearly 100 tubes per hour for the inspection of the first 43in of each tube, even while using higher than average axial resolution (75 points per inch instead of the usual 32). In comparison, inspection for half that length with surface riding probes averaged less than 20 tubes per hour. All quoted time figures include manipulator movements, calibration, regular data quality checks, and data acquisition.
•Probe life is at least one order of magnitude higher than that of surface-riding rotating probes. At Doel, each of the two probes inspected the first 43in of nearly 3000 tubes. In another trial, probe life was found to be more than 400 tubes per probe during a full-length inspection. At another trial, a single probe was used for the full-length inspection of 315 tubes with abrasive deposits on the inner surface, again without significant probe wear.
There have been no probe failures so far in any of the field trials and it is expected that the X-Probe will have a life similar to that of the conventional bobbin coil probes used for the same application.
The X-Probe has recently been qualified at EPRI to Appendix H.
Data was acquired from a set of nearly 200 tubes provided by EPRI. The tubes had a wide variety of flaws, including ODSCC, PWSCC, deformations and pits.
The analysis report was prepared by R/D Tech, AECL, and MoreTech, and submitted to EPRI along with the analysis procedure. A set of tubes was selected and rescanned with and without representative support structures. These tubes were sent for destructive examination and the qualification was completed after a peer review in summer 2000.
Following the field trials and qualification carried out in 2000, indications suggest that the X-Probe will help reduce outage time, radiation exposure, and the overall costs associated with inspecting steam generator tubes.