·Table of Contents
·Methods and Instrumentation
Equipment Condition Monitoring using TOFD. - Experiences at DSMA.A.A. Scheerder
DSM Engineering Stamicarbon
P.O. Box 18, 6160 MD Geleen The Netherlands
|Feature||TOFD||X-ray||Manual UT PE||Mechanised UT PE|
|defect detection (POD %)||60-85||45-95||50||65-80|
|False Call rate (FCR %)||0-20||10-15||25||10-30|
|Defect sizing error (mm) Length||±2 - ±10||±1||±9||±2|
|Defect sizing error (mm) Height||±1||not possible||not possible||±1|
|Remaining Ligament (mean error) near surface||±3||not possible||±4||±4|
|Remaining Ligament (mean error) far surface||±4||not possible||±8||±4|
|Defect interaction (mean error) mm||±2||±10||±5||±2|
|correct defect characterisation % planar /non planar||not possible||70 - 80 / 80 - 90||70 / 65||25 - 60 /15 - 90|
|correct defect characterisation % classification||not possible||75||35||not possible|
|Table 1: performance of NDT methods in accordance with existing codes, standards and guidelines, source .|
3.1 Weld details to be tested.
Several types of welds have to be inspected. Besides the straightforward welds of the shell (butt welds, see figure 1), also different types of nozzle weld (symmetric as well as asymmetric) have to be inspected (some of them with reinforcement pads). For this purpose an inspection procedure was developed and validated, using modelling software as well as full scale test samples.
|Fig 1: TOFD inspection of longitudinal weld of the shell using a hand held scanning device.|
Since the welds have to be monitored for stress corrosion cracks initiated in the weld root, the inspection procedure was optimised for maximum sensitivity in this region. With the aid of computer modelling, the actual probe position and probe angle were calculated.
In order to verify the coverage of the weld root full-scale test blocks were used, see for example figure 2.
All nozzle welds were ground at both sides to get optimal results. The weld cap was ground in an appropriate radius for optimal coupling of one of the two transducers. Also the weld root was ground in order to have optimal sensitivity for stress corrosion cracks in this area.
Fig 2: Full-scale reference block.
A baseline fingerprint was obtained during manufacturing of the vessels in order to increase the probability of detection of service induced defects during subsequent in-service inspections. The inspections will be carried out at ambient temperature during a planned turnaround.
This alternative inspection procedure was accepted by the Dutch Regulatory Authority and will result in considerable costs reduction with respect to the mandatory periodic inspections in the future, since the vessels don't have to be opened.
|Fig 3: Propane sphere at DSM. Abseilers mounting AE sensors for AE test|
It was decided to check all the welds of the sphere for critical defects, meaning defects larger than 4.5 mm in depth. In order to do so it was proposed to monitor the hydrostatic proof testing with acoustic emission. A research programme was carried out, see [2, 3] in order to validate this approach. The project was aimed to establish the minimum stress level needed during hydrostatic proof testing to detect all critical defects (4.5 mm in depth). However this could not be obtained and the acoustic emission was not accepted at that time.
Despite the fact that large manufacturing welding defects are present to date no crack initiation and growth during the entire service life of the spheres has been observed. This means that the welding defects present, are no threads to the structural integrity under the prevailing service conditions. Furthermore the hydrostatic proof testing was not wanted any more as test method (heavy water load can damage the sphere). This resulted in a new inspection philosophy, focussing on finding only defects, which are active during operations. For this purpose again an AE test method is applied, but in this case only at pressures 10 % above the actual working pressure. Additional on-stream TOFD testing is performed to confirm the AE results.
In order to assess the weld defects found with the TOFD method specific acceptance criteria were established for this application.
The assessment is based on a fitness for purpose philosophy using the BS PD6493:1991. "Guidance on methods for assessing the acceptability of flaws in fusion welded structures", British Standards Institution, London, 1991.
The result is presented as the straight line in figure 4.
|Fig 4: Crack assessment graph. Defects underneath dotted line, no report. Defects in-between dotted and straight line will be reported. Defects above straight line will be assessed with FFP method.|
All defects found using the TOFD method with dimensions smaller than the dotted line will not be reported. All defects between the dotted and the straight line are reported. Defects above the straight line will be further assessed using fracture mechanical calculations in order to determine if repair is needed.
This approach resulted in a more realistic estimation of acceptable defect dimensions compared to the wide plate tests in the early eighties (4.5 mm). For instance as can be deduced from figure 4 a crack depth of 15 mm is still acceptable (nominal wall thickness of the sphere is 30 mm) if the crack length is below 25 mm. Also when the crack depth is below 6 mm, the length of the crack can be infinite.
The first two spheres tested in this way showed good results. With TOFD no defects were found beyond the acceptance criteria. This confirmed the assumption that the rather large manufacturing defects present are acceptable and thus are no thread to the structural integrity. This was also confirmed by the AE tests, since no severe AE sources were detected, meaning that the defects present are not active under the prevailing service conditions.
Repair was not possible because of the enamel coating at the inside.
The defects found probably are fabricating welding defects, but this could not be verified.
In figure 5 a schematic overview of the reactor and the nozzle is presented.
The assessment has been made according to BSI PD6493: 1991.
The reactors are operated in a batch process, and the nozzles are submitted to fatigue loading. As a result of the analysis the defects are acceptable at working pressure but there's not much room to critical flaw dimensions and there won't be "leakage before break". Fatigue analysis shows that the defects, if they have to be classified as "growing cracks", can grow to unacceptable dimensions in a relatively short time. It was however not plausible that this situation, if it really exists, didn't lead to a failure already and therefore the "growing crack" approach is probably too conservative.
It was therefore advised to monitor the largest defect found and to establish whenever this defect will growth. Successive inspections had to be performed within a few months. When no growth of the defect is observed then all the defects can be assumed not to be "growing defects" i.e. the fitness for purpose is guaranteed.
Based on this an inspection plan was set up using TOFD for monitoring the crack growth. In order to analyse the TOFD signals modelling software was used see for example figure 6.
|Fig 5: Schematic overview of the nozzle and the flaw detected. By FEM modelling the flaw was assessed.||Fig 6: Typical TOFD data obtained from the nozzle weld. The signals are evaluated using modelling software, as presented on the right hand side.||Fig 7: On-stream inspection of the nozzle weld of the PVC-reactor at DSM.|
After the first baseline measurement to date several inspections have been carried out on-stream, see figure 7. Successive inspections showed TOFD images, which matched very well with the base line measurement. This gave confidence that no crack growth starting from this critical defect occurred. In the first half year every month an inspection was carried out. Since no crack growth was observed the inspection interval was extended to once a half year.
Based on this monitoring programme the reactors at DSM are still in service.
|© AIPnD , created by NDT.net|||Home| |Top||