Manual ultasonic inspection of thin metal welds Capucine Capentie and John Rudlin TWI Cambidge CB1 6AL, UK Telephone 01223 899000 Fax 01223 890689 E-mail capucine.capentie@twi.co.uk Abstact BS EN ISO 17640 contains standad ultasonic inspection techniques fo feitic steel welds. The techniques in geneal include scans with diffeent angles fom eithe side of the weld. The minimum thickness noted in this standad is 8mm. Howeve thee ae an inceasing numbe of applications involving smalle thicknesses (e.g., containments) and these cannot stictly be inspected by these methods. This pape descibes a seies of expeiments of conventional ultasonic and phased aay testing on thin butt welded sections with simulated flaws using the pocedues descibed in BS EN ISO 17640 and BS EN 13588 but allowing the use of high fequency and small pobes easily available in the maket. The esults show that the pocedues of BS EN ISO 17640 can be adapted in this way fo detection scanning and sizing fo thicknesses down to 4mm. 1. Intoduction Thin walled vessels and pipes ae inceasingly equied to be tested non-destuctively when they ae used to contain substances that can damage a local envionment. Examples include some designs of nuclea waste stoage vessels, nuclea pocessing plant, low pessue oil pipes and othe fuel systems. Diffeent welding methods fo stainless and duplex steel ae also unde investigation fo these aeas. Equally thee is a need to detemine the actual size of a emaining ligament as this has a significant effect on factue mechanics analyses. Conventional ultasonic testing (accoding to BS EN ISO 17640) (1) limits thicknesses that can be examined to less than 8mm. Similaly, the daft phased aay standad BS EN 13588 (2) goes down to 6mm. Smalle thicknesses ae fequently inspected by adiogaphy, this cannot nomally give the though thickness infomation equied fo factue mechanics analyses. Thee is a equiement fo non-destuctive test methods in the smalle thickness anges. Bid (3) epoted an expeiment compaing manual UT and phased aay with thicknesses down to 6mm with good esults fo good opeatos using both techniques. New o diffeent technologies, such as eddy cuent aays o high fequency ultasonic phased aays may be needed fo smalle thicknesses. A special
case whee ultasonic methods have been developed fo a thin wall has been given by Bid et al. (4) Some wok on eddy cuents is epoted in this confeence (5). Howeve, fom the manufactues point of view it is convenient to efe to a standad when equesting items to be made, and to know what the actual limitations ae of the cuent UT methods. The main pupose of this pape is theefoe to examine whethe the nomal manual ultasonics pocedues used in the standad can be applied with minimum modification to inspect ac welds in thin sections and whethe phased aay techniques can impove on this. The eason that the 8mm limit applies in most cases is that it is not possible with thinne sections to apply standad UT pobes because they cannot appoach the weld close enough no can the beam be easily defined. The modification to the standad pocedue adopted is to allow the use of high fequency 10MHz angled beam pobes. Phased aay technology fo weld inspection is becoming matue, with a daft standad fo weld inspection BS EN 13588 (6), and this standad and technique wee also adopted. 2. Test Samples The test samples specified consisted of plates 3, 4 and 5mm thick in feitic and stainless steel, each containing 4 flaws. Figue 1 shows the position of these in the 4mm plate, the othes wee simila. The weld has a 60 pepaation and thee ae thee embedded lack of fusion flaws at the weld cap and one oot lack of fusion flaw in each plate. The flaws have been made with intended dimensions and those supplied by the manufactue, and these ae used in the analysis. Howeve, it should be noted that manufactue and measuement of such flaws has some uncetainties so the esults obtained must be viewed with this in mind. A adiogaphic inspection of these welds was pefomed in ode to veify the length of these flaws (Table1). As expected, the adiogaphy was not able to eliably detect lack of side wall fusion. Table 1. Results of adiogaphic inspection of welds Cabon steel 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected? N Y N N N Y N Y N Y Y N Length \ 12 \ \ \ 12 3 \ 12 15 \ Stainless steel 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected? Y Y N N N Y N N N Y N Y Length 7 12 \ \ \ 12 \ \ \ 12 \ 15 2
Length: 10mm Length: 10mm Length: 5mm Length: 15mm Figue 1 Position of the simulated flaws in the welds 3. Manual UT 3.1. Standad Requiements The intenational standad fo the manual ultasonic testing of fusion-welded joints in metallic mateial BS EN 1714 has been withdawn in 2010 and eplaced by BS EN ISO 17640:2010. This standad is specified to be applied fo the testing of fusion welded joints in metallic mateial of thickness geate than o equal to 8mm and whee both the welded paent mateial ae feitic. The pobe fequency specified in this standad ae to be within the ange 2 and 5MHz. Howeve, the standad allows the use of highe fequency fo impoving ange esolution if necessay fo acceptance citeia based on chaacteisation of indications. It is also equied that at least one angle beam shall be nomal o nealy nomal to the weld fusion face. The testing volume is to include the weld body and at least 10mm on each side of the weld to cove the heat affected zone (HAZ). The standad is open to seveal techniques to set the sensitivity. One of the techniques specified in the standad allows the use of efeence fom a distance-amplitude cuve (DAC) fo side-dilled holes of diamete 3mm. Refeence can also be made on notches 3
of 1mm wide with a depth of 1mm only fo the thickness ange between 8 and 15mm and fo angle beam angles lage than 70. BS EN 17640:2010 efes to BS IS 11666:2010 fo the definition of the evaluation level above which an indication will be investigated. Fo an acceptance level 2, which coesponds to the most common quality level, the level of evaluation is set at -14dB below efeence level. 3.2. Equipment and Pocedue The equipment used to cay out the manual testing was a Sonatest Mastescan 380M with 10MHz tansduces 45, 60 and 70, 6.35mm cystal diamete. The tansduces used fo the testing ae shown in Figue 2. The footpints of these tansduces wee between 20 to 30mm and enabled the pobe index point to be close to the weld cap and hence educe the numbe of ultasonic skips on the back wall. As has been mentioned above the pocedues of BS17604 togethe with an allowance fo the use of high fequency pobes has been used. DAC was used elative to a 3mm SDHs. No attempt was made to have blind tials in this case. The opeato had some knowledge of the flaw location and was seeking to identify and use the indication. a) b) c) Figue 2 Pictues of ultasonic tansduce Paametics A544S 10 by 0.25 a) 45 wedge b) 60 wedge c) 70 wedge 4. Phased Aay UT 4.1. Standad Requiements The use of (semi-)automated phased aay technology fo the inspection of welds is coveed by the daft intenational standad BS EN ISO 13588 which is cuently open to public comments. This standad specifies the application of the phased aay technology fo fusion welded joints in metallic mateials equal to and above 6mm thickness. 4
This standad is open to the use of linea scan (E-scan) and secto scan (S-scan) at a fixed distance to the weld to geneate multi angle beam fom a single position of the tansduce. In addition, this standad specifies the use of scanning mechanism that allows data collection with a scanning incement of no moe than 1mm fo component thicknesses up to 10mm. It is also equied to pefom pio testing sensitivity setting fo each beam and focal point geneated by the phased aay pobe. The use of angle coected gain (ACG) and time coected gain (TCG) shall be used to display the signals fo all beam angles and all distances with the same amplitude. The standad povides ecommendation on the type, size and positions of efeence eflecto to be used fo the sensitivity settings. Fo a thickness between 6 and 25mm, a notch of 1mm deep o a 2.5mm diamete SDH can be used. In addition, thee is no specification in this standad egading a set level fo assessment of indication. This is based in accodance with specified acceptance level and acceptance level applied. 4.2. Techniques The samples wee inspected with ultasonic phased aay using the off the shelve tansduce 10L16 SA00 N60S manufactued by Olympus as shown in Figue 3. This tansduce geneates shea waves at 10MHz using 16 elements. This tansduce is taditionally used fo aeospace applications and pesents the advantages to have a elatively foot pint (21mm) which allow to come close to the weld cap. The weld oot and fusion face wee inspected using a secto scan geneating ultasonic beams fom 45 to 75 with an angula esolution of 0.3 and thee linea scans at 45, 60 and 70. The same delay laws wee used fo 5, 4 and 3mm thick plates. The ultasonic beams wee focussed at diffeent depths in ode to take into account the multiple skipping on the back wall fo diffeent beam angles and plate thickness. The welds wee examined fom both sides and at two index offsets fom the cente of the weld in ode to cove the weld fusion face at diffeent angle of incidence. The index offsets wee diffeent fo each thickness. Figue 4 shows the ay tacing of the beams geneated by the secto and linea scans fo a weld 5mm thick and at the two index position fom the weld cente. The aim was to choose the ight index offset in ode to ensue that the weld fusion face, cap and oot wee coveed by at least two diffeent angles. As equied in most codes and standads fo phased aay testing, the wedge attenuation and delay compensation was caied out using the 25mm adius of the V2 calibation block in both cabon and stainless steel. Fo each delay law, an ACG was set based on 3mm SDH at a depth of 10mm in both cabon and stainless steel. A TCG was applied on each delay law in ode to compensate the attenuation loss thoughout the ange of the sound path. The TCG cuves wee geneated using a seies of 3mm diamete SDH at a depth of 5, 10 and 20mm. The plates wee then scanned 5
using a sensitivity set at TCG + 6dB and TCG +12dB. The sensitivity level was also checked on notch with a depth equal to 10% of the thickness of the plate tested fo the welds in stainless steel. It was noted that the sensitivity on the notches was equivalent to the sensitivity set on the SDHs within 10% FSH. Data wee ecoded using the Omniscan MX with a sting encode at a scanning esolution of 0.5mm. Figue 3 Pictue of phased aay tansduce 10L16 SA00 N60S manufactued by Olympus and used in the study, a) b) Figue 4 Scan plan fo the inspection of the 5mm thick plate (geen lines = 70 beam, ed lines = 60, blue line = 45, pink line = secto scan) a) Index offset at 7mm b) Index offset at 11mm 6
5. Modelling In ode to evaluate the length sizing esolution o lateal esolution of the inspection and compae beam pofile between the conventional and phased aay testing, the ultasonic beams geneated wee modelled using the modelling platfom CIVA v10.1. The ultasonic beam size was extacted fom the coss section of the ultasonic beam sound pessue at a depth coesponding to the ange whee the weld is examined. Figue 5a shows the model configuation fo a beam geneated by the phased aay tansduce at 45 focussed at 7.5mm in depth and the ultasonic beam modelled in 3D. The ultasonic beam pofile was also displayed along thee coss section planes (X;Y), (Y;Z) and (X;Z) as shown in Figue 5b,5c and 5d. The maximum enegy is displayed in blue. The pojected 6dB beam width along the Y and X axis was exacted to evaluate the lateal esolution of the inspections and compae the pefomance of the techniques fo though wall sizing. Table 2 summaises the beam width along the Y axis of the ultasonic beams at 45, 60 and 70 geneated by the phased aay techniques using secto and linea scans and by the conventional method at the ange fo which the welds in the 3, 4 5mm plate ae examined. Fo the phased aay techniques the beam size pedicted was aveaged fo the stand-offs at 7 and 11mm. Along the 45 beam, and fo the thee thicknesses, the secto scan geneates a smalle beam of about 1mm smalle compaed with the linea scan in the conventional method. Along the 60 beam, the secto and linea scans ae equivalent but thee is a slight incease within 1mm of the beam width fo the conventional method. Along the 70 beam, the phased aay using the linea technique geneates a smalle beam in compaison with the sectoial technique and the conventional method. In oveall, the linea techniques and conventional method povide a moe egula beam width ove the thee angle beams. The lateal esolution with a beam geneated with the sectoial technique is bette fo the small angle beam but incease significantly at highe angle. Oveall, the length sizing esolution geneated with the methods used in this study is anticipated to be in aveage equal to 3mm. Table 2 Pediction of the 6dB ultasonic beam width along the Y axis geneated by phased aay (PA) and conventional techniques at 45, 60 and 70 angle beams: Plate thickness Techniques 45deg 60deg 70deg 5mm PA 2.3 2.8 4.3 PA Linea 3.3 2.8 3.3 Conv. 3.4 3.7 3.8 4mm PA 2.1 2.1 4 PA Linea 3.2 2.8 2.1 Conv. 3.4 3.7 3.8 3mm PA 1.8 1.8 4 PA Linea 3.1 2 2.1 Conv 3.4 3.7 3.8 7
Table 3 summaises the beam width along the X axis of ultasonic beams at 45, 60 and 70 geneated by the phased aay techniques using sectoial scan and linea scan and by the conventional method at the ange fo which the welds in the 3, 4 and 5mm plates ae examined. Fo the phased aay techniques, the beam size pedicted was aveaged fo the stand-off at 7 and 11mm. Though wall sizing esolution is pedicted by evaluating the beam dimension in the X/Z plane. The beam width given in Table 3 with modelling is the pojected dimension of the beam width along the X axis. The pojection of the beam width along X axis was compaed fo all inspection techniques in ode to evaluate the sizing capability. Fo the weld in the 5mm plate, the though wall esolution ae equivalent fo the thee techniques. As expected, the beam width inceases with the angle of popagation. Fo the welds in the 4mm and 3mm plates, beam width fo though wall esolution ae equivalent along the 45 beam fo the thee techniques. Along the 60 beam, thee is a significant diffeence in the beam width between the sectoial scan and the linea scan with the sectoial scan geneating a smalle beam. The conventional testing povides equivalent pefomance as the linea scans. Along the 70, thee is a significant incease in the beam width fo the thee techniques. The lagest incease and the lagest beam width ae geneated by the sectoial scan. In conclusion, the sectoial scan povides a bette beam chaacteistic at low angle compae with the linea and conventional methods. Howeve the pefomance with the secto scan deceased significantly and become less beneficial than the linea and conventional methods at high angle. Table 3 Pediction of the 6dB ultasonic beam width along the X axis geneated by phased aay (PA) and conventional (Conv.) techniques at 45, 60 and 70 angle beams: Plate thickness Technique 45deg 60deg 70deg 5mm PA 2.7 3.9 7.8 PA Linea 3.2 3.8 5.8 Conv. 3.2 4.4 7.1 4mm PA 2.6 2.6 9.3 PA Linea 3.3 4 6.1 Conv. 3.2 4.4 7.1 3mm PA 2.6 2.6 9.3 PA Linea 3.4 4.2 6.1 Conv 3.2 4.4 7.1 8
a) b) c) d) Figue 5 Pediction of the ultasonic beam pofile geneated at 45 with the phased aay tansduce a) 3D modelling b) Beam coss section along (Y;Z) plane c) Beam coss section along (X;Y) plane d) Beam coss section along (X;Z) plane 6. Results 6.1. Manual UT Tables 4 and 5 summaise the inspection esults of the conventional ultasonic testing on the weld sample fo cabon steel and stainless steel espectively. Some of the flaws wee not detected eithe because no signal has been disciminated fom the noise o because the signal amplitude was below the -14dB evaluation level but identified fom the noise. The latest case is identified in the tables when the amplitude is <14. Most of the flaws in the 5 and 4mm thick plates wee detected. Not all the flaws in the 3mm welds wee detected. It can be noted that the level of detection of flaws in cabon steel and stainless steel ae equivalent. Fo these small thicknesses, the anisotopic and attenuative popety of the stainless steel does not seem to have a paticula effect on the pefomance of the ultasonic testing. It was not possible to measue the though wall size of the flaws. The opeato was not able to identify the edge of the flaw eithe with the 6dB dop technique o max 9
amplitude. The spatial and tempoal esolution was not good enough fo though wall sizing. Table 4 Inspection esults fo the conventional ultasonic testing of welds in cabon steel: 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected? Y Y N N Y Y N Y N Y Y Y Beam angle 70 70 70 / 45 45 70 60 45 70 45 45 db fom DAC -12-14 <14 / -4-12 <14-8 <14-5 -8-7 Ligament / / / / / / / / / / / / Though wall / / / / / / / / / / / / Length 4 5 / / 6 18 / 3 / 16 6 16 Table 5 Inspection esult fo the conventional ultasonic testing of welds in stainless steel 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected? N Y N Y Y Y N Y Y Y Y Y Beam angle / 45 / 60 60 70 / 45 70 60 60 70 db fom DAC / -14 / -7-2 -11 / 1.5 0.5-5 4 4 Ligament / / / / / / / / / / / / Though wall / / / / / / / / / / / / Length / 3 / 6 11 5 / 5 12 8 7 7 6.2. Phased Aay Tables 6 and 7 summaise the inspection esults of the phased aay testing on the weld samples fo espectively cabon steel and stainless steel. Figues 6 to 8 show some example of the plane view (C-scan) of the data collected with the phased aay fom both side of the weld (skew 270 and 90 ). The blue axis povides the position and length of the flaws. On the geen axis the position of the flaws with egad to the weld centeline (at zeo) is shown. The phased aay inspection povided a good detection fo the welds 5 and 4 mm thick. Fom the plane view, in the 5mm thick weld in cabon steel and in the 4mm thick weld in stainless steel, the fou flaws wee clealy identified fom the data. The data on the stainless steel was faily clea and did not show a stong signal fom the oot as shown in Figue 7. In the 3mm thick weld in cabon steel and stainless steel (Figue 8), stong indications ae noted along the weld coming fom multi-eflection fom the cap and the oot which mask the eflection fom the flaws. In addition, signals fom the flaws in the welds 3mm thick ae not expected to be high amplitude. Gain was then inceased which made the 10
inspection on the stainless steel wost as the noise level fom the weld wee highe than the flaws signals. Figue 9 shows the data fom the flaw B in the 5mm thick weld in stainless steel. The flaw B is intended to be vetical lying at the weld oot. The data shows effectively an indication fom the cone and the tip of the flaws fom which the flaw though wall height was measued. Flaw C shown in Figue 10 fo the 5mm thick weld in stainless steel is lying along the weld fusion face, specula to the ultasonic beam. Assuming that the flaw was bigge than the beam width, the though wall sizing was possible using 6dB dop on eithe side of the flaw edge. Table 6 Inspection esult fo the phased aay testing of welds in cabon steel 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected Y N N N Y N Y Y Y Y Y Y? Techniqu e Beam angle Amplitud e / / / Sect o / Sect o Linea Linea Linea Linea Linea 48 / / / 48 / 60 60 45 70 45 45 30% / / / 80% / 84% (3dB ) 52% 76% 75% >100 % >100 % Ligament 0 / / / -0.6 / -2.5-2.2-0.3-3.3-1.8-1 Though wall 0.6 / / / 1.6 / 2.0 2.0 1.2 1.7 0.3 2.8 Length 9 / / / 11.5 / 5 6 11.5 13.5 6 15 Table 7 Inspection esult fo the phased aay testing of welds in stainless steel: 3mm plate 4mm plate 5mm plate Flaws A B C D A B C D A B C D Detected N N N Y Y Y Y Y Y Y Y Y? Techniqu e / / / Beam / / / 65 51 76 50 70 49 53 52 52 angle Amplitud e / / / >100 % 55% 70% 100% 55% (3dB) 80% (3dB) 41% >100 % >100 % Ligament / / / 1.4-1 -2.9-1.7-1.1-0.3-3.2-1.5-3.1 Though / / / 1.4 1.6 1.1 1.7 2.8 1.7 0.8 2.8 3.0 wall Length / / / 19 4 8.5 8.5 8 11 10 10 17 11
a) b) Figue 6 Plane view of the data collected with phased aay on the 5mm plate in cabon steel at sensitivity TCG +12dB fom a) secto scan at 11mm stand off at skew 270 b) secto scan at 7mm stand off at skew 90 a) 12
b) Figue 7 Plane view of the data collected with phased aay on the 4mm plate in stainless steel at sensitivity fom a) secto scan at 7mm stand off, TCG +3dB, at skew 270 b) secto scan at 7mm stand off, TCG +6dB, at skew 90 a) b) Figue 8 Plane view of the data collected with phased aay on the 3mm plate in cabon steel at sensitivity fom a) secto scan at 11mm stand off, TCG +12dB, at skew 270 b) secto scan at 11mm stand off, TCG +12dB, at skew 90 13
a) b) Figue 9 Flaw B phased aay data on sample 5mm in stainless steel a) secto scan, stand-off 7mm, 6dB, skew 90 b) detail of Flaw B, secto scan, stand-off 7mm, 6dB, skew 90 14
Figue 10 Flaw C phased aay data on sample 5mm in stainless steel, secto scan, stand-off 11mm, 3dB, skew 90 7. Discussion It can be noted that the level of detection between the conventional method and phased aay techniques is simila fo the plate 4 and 5mm. Fo the 3mm plate, the conventional method gives detection of moe flaws. This could be because in conventional testing the opeato can skew the pobe and hence disciminate the signal fom a flaw to othe geomety featues such the oot and geomety which ae vey close togethe in time fo weld in small thickness. Phased aay testing povides, by displaying the data in plane and coss section views, the advantage to ease the intepetation of the signals and hence allows in some cases though wall sizing using tip diffaction fo vetical flaw and specula eflection fo flaw such as LOF. In addition, length sizing seems to be close with phased aay to the intended flaw size and also the esult fom the adiogaphy. 8. Conclusions 1. BS EN standads fo the use of conventional ultasonic and phased aay testing fo welds is today limited fo the examination of thickness lage than 8mm fo the conventional method and 6mm fo the phased aay. Fom this study, it has been shown that this standad could be extended to weld lage than 4mm thick in the condition that a small foot pint and high fequency tansduce is used (highe than 10MHz) and the sensitivity settings is set on efeence taget epesentative to the size of flaws to be found in small thickness ie use of small diamete SDHs o notches. 15
2. The phased aay method has the advantage of a display which gives easie intepetation of signals and hence bette sizing and chaacteisation of flaws. Howeve, at the smallest thickness, it has been shown that allowing the pobe to be moved feely in all diection along the weld as used with the conventional testing helped detection of flaws in compaison with the igid scan fom a fixed distance fom the weld used with phased aay. Theefoe, fo this thickness below 4mm, a fee manual scan using phased aay techniques could be ecommended fo the detection of flaws. This could then be followed by encoded scans in the aea pesented indications to be investigated. Acknowledgements The wok was funded by TWI s Coe Reseach Pogamme. Full epots ae available to membe companies of TWI. Refeences 1. BS EN ISO 17640:2010, Non-destuctive testing of welds Ultasonic testing Techniques, testing levels, and assessment. 2. & 6. Daft BS EN ISO 13588, Non-destuctive testing of welds Ultasonic testing Use of (semi) automated phased aay technology. 3. Schneide C R A and Bid C R, Reliability of manually applied phased aay inspection, 4th Euopean-Ameican Wokshop on Reliability of NDE, Belin, June 2009, http://www.twi.co.uk/content/spcbjune09.html. 4. Bid C R and Pettigew I, Qualification of a phased aay inspection of thin weld. 18 th Wold Confeence on Non-destuctive Testing, Duban, Apil 2012. 5. Majidnia S and Rudlin J, Depth of Penetation Effects in Eddy Cuent Testing, 51st Annual Confeence of The Bitish Institute of Non-Destuctive Testing, Daventy, Septembe 2012. 16