CIRCUMFERENTIAL GUIDED WAVES FOR DEFECT DETECTION IN COATED PIPE

Transcription

1 CIRCUMFERENTIAL GUIDED WAVES FOR DEFECT DETECTION IN COATED PIPE W. Luo, J. L. Rose, J. K. Va Velsor, M. Avioli 2, Jack Spaer 3 Departmet of Egieerig Sciece & Mechaics, the Pesylvaia State Uiversity, Uiversity Park, PA 6802, USA 2 FBS, Ic., State College, PA 6803, USA 3 NDE Ceter, Sciece & Techology Developmet, EPRI, Charlotte, NC 2822 ABSTRACT. Circumferetial guided wave propagatio behavior i a viscoelastic multi-layered hollow cylider is studied to provide a baselie for defect detectio i tar coated pipelies. Theoretical work was carried out by developig the appropriate dispersio curves ad wave structures for circumferetial guided waves i a pipe coated with a viscoelastic material. Parameters that affect wave atteuatio were ivestigated with some iitial guidelies beig established for improved peetratio power. Low frequecies are suggested from both atteuatio ad detectio depth poits of view. Uder this guidace, experimets utilizig a liear trasducer array were coducted at a low frequecy for successfully detectig delamiatio ad volumetric defects i tar coated pipe. A study was carried out to fid the appropriate features for defect detectio i coated pipe ad a test protocol based o this study is recommeded ad summarized. KEYWORDS: circumferetial guided wave, multilayered structure, viscoelastic coatig, defect detectio, ultrasoics, pipe PACS: Mv, Mr, *43.40.Le, *43.35.Zc, 8.70.Cv INTRODUCTION Pipelies are used i almost every idustry that has a eed to provide a large scale distributio of their product, be it gas, oil, water, etc. The electrical utility idustry uses pipes as coduits for the distributio of electricity, with especially complicated cofiguratios, such as the trifurcatio joits, used i ier city maholes. These pipelies are both above groud ad below groud ad are thus exposed to a variety of chagig evirometal coditios that ca lead to corrosio ad/or coatig separatio at the outer diameter of the pipes. Therefore, viscoelastic coatig (coal tar, epoxy) is widely used i the pipelie idustry for corrosio protectio. Defective pipelies ca cause fatal failures, property damage, ad high litigatio ad replacemet costs. A effective NDE method is eeded to provide useful iformatio to the pipelie idustry as corrosio ad delamiatio detectio is critical for purposes of maiteace ad repair. Sice the middle of the 980 s, ultrasoic guided wave techiques have bee developed ad applied rapidly i the field of NDT because of their log-rage testig ability ad efficiecy compared with the traditioal poit-by-poit bulk wave methods []. Preseted 65

2 Receiver Trasmitter Viscoelastic Elastic Layer Displacemet r 3 r 2 r θ r Circumferetial FIGURE. The use of circumferetial guided waves to fid pipe defects uder tar tape coated pipes. z FIGURE 2. Cylidrical coordiates of a hollow viscoelastic coated cylider ad dimesios. i this paper is the developmet of a ultrasoic circumferetial guided wave techique, which exhibits the ability to detect defects [2-3]. Figure portrays the cocept of circumferetial guided waves for this applicatio. The difficulty of defect ispectio i a coated pipe usually comes from the atteuatio ad wave eergy distributio chage alog the pipe wall due to the existece of coatig, which may reduce the ispectio capability sigificatly. Therefore, theoretical work was carried out first by developig the dispersio curves ad wave structures i a pipe with viscoelastic coatig. The goal was to fid a appropriate frequecy with less atteuatio ad a appropriate eergy distributio. Based o the recommedatio from theoretical results, a piezoelectric liear trasducer array geeratig SH circumferetial guided waves at a low frequecy rage is desiged to provide the best results i terms of defect detectio ad sizig. Experimets o a coated pipe with a simulated defect were coducted ad a prelimiary featured-based decisio (defect, o defect) model was the costructed. DISPERSION CURVES For a pipe with a viscoelastic coatig, the phase velocity dispersio curve, atteuatio dispersio curve ad wave structure were calculated usig the global matrix techique. Figure 2 shows a viscoelastic coated pipe. Accordig to the correspodece priciple [4], uder a steady state, time harmoic assumptio, the guided wave propagatio solutio for a N-layered pipe with oe or more viscoelastic layers ca be acquired simply by substitutig the elastic modulus of the respective layers with the complex viscoelastic modulus ito the dispersio equatio for a totally elastic N-layered pipe. The dispersio equatio of a N-layered viscoelastic pipe ca be geerated similarly by just usig the complex viscoelastic costats istead of the elastic costats. The displacemet u z of circumferetial SH wave i the th layer ca the be expressed as i equatio (): ik ikr ( ) ( ) 2 ( ) + θ ωt uz == C J ) k r + C Y) k r e, r k k r r r + () + The stress ca be calculated by equatio (2): 2 ( ) ik ikr ( ) ( ( ) ( )) 2( ( ) ( )) + θ ωt σ rz = µ C J) k r J) k r + C Y) k r Y) k r e k k + k k + 2r (2) + where k is wave umber ad kˆ is a wave umber related parameter, J ad Y represet Bessel fuctios, C represets ukows, ad µ is a Lamé s costat. See [5] for more details about the wave theory ad omeclature. Boudary coditios are 2 tractio-free coditios at the ier ad outer surface ad 2(N-) displacemet ad stress cotiuity coditios at the (N-) iterfaces: 66

3 σ rz r r, r N+ = 0; = uz uz = 0, =,2,,N- σrz σ Layer= rz Layer= + r= r+ r= r+ (3) There is a total of 2N boudary coditios. Each layer has two ukows C ad C 2, ad thus there is a total of 2N ukows. The layer matrix is expressed i equatio (4): u z C ikr ( ) [ ] + θ ωt = A e (4) σ rz C Layer = 2 The 2 N 2N global matrix ca be costructed after applyig the 2N boudary coditios ad the dispersio equatio of the N-layered pipe ca be expressed as (5): A = 0 2N 2N (5) For a multi-layer pipe with viscoelastic layers, complex viscoelastic costats are the substituted ito equatio (5), i place of the elastic oes. Complex root searchig ca be carried out i a 3-dimesioal space (frequecy, real part of k*, ad imagiary part of k*) [6]. The eigevalues of the system are the used to costruct the dispersio curves ad the eigevectors are used to costruct wave structures. The wave umber root k* becomes a complex umber. The complex root is composed of the real part Re(k*), which results i the phase velocity dispersio curve, ad the imagiary part Im(k*) leads to the atteuatio costat as well as the atteuatio dispersio curve: ω c p ( ω ) = = ; Re( k*) Re(/ c*) ω α ( ω) = Im( k*) = Im(/ c*) (6) The calculated atteuatio dispersio curves for a 0 ich schedule 40 pipe coated with bitume tape are show i Figure 3, where the atteuatio icreased with the frequecy ad the layer thickess. It ca be see that the frequecy rage from 20 khz to 40 khz provides reasoably low atteuatio. A wave structure aalysis was carried out i order to tell the ispectio potetial at differet wall depths. The results show that the percet eergy carried by the elastic layer becomes less as the frequecy ad coatig thickess icrease. Therefore, it would be safer to use lower frequecy i order to realize a more reliable detectio Atteuatio (db/m) bare layer 2 layers 3 layers Frequecy (MHz) FIGURE 3. Calculated atteuatio dispersio curves of 0 circumferetial SH wave i a steel pipe (0 ich schedule 40) with differet coatig coditios (the thickess of each layer is mm), showig wave atteuatio icreases sigificatly as frequecy ad/or coatig thickess icrease. 67

4 EXPERIMENTAL STUDY For circumferetial guided wave ispectio as show i Figure, the ispected area was limited to the axial size of the trasducers used. To overcome this limitatio, a 26 log liear array with 6 chaels ad 24 trasducers was desiged as see i Figure 4. The beefit of usig a liear array is that beam spreadig ca be effectively reduced, especially for the middle two chaels, thus realizig cosistet ispectio ability. Experimets were carried out o a 0 schedule 40 pipe coated with tar tape as show i Figure 5(a). Frequecy sweepig was first carried out i the 20 to 40 khz rage suggested by previous theoretical studies, with the best frequecy foud to be 30 khz from sigal width ad SNR poits of view. A typical wave sigal at 30 KHz is show i Figure 5(b), where wrap aroud sigals atteuate gradually with circumferetial wave propagatio distace. Two key features were idetified: The atteuatio factor of the wave trai produced by the periodic traversal of the iput pulse aroud the circumferece of a pipe ad the appearace of a echo betwee the periodic pulse patter. The amplitude A ( x,ω) of a propagatig wave sigal is a fuctio of frequecy ad the wave propagatio distace: α ( ω) A x, ω e (7) ( ) x where α(ω ) is the atteuatio costat ad x is the wave propagatio distace. The atteuatio costat, α(ω ) ca be expressed i equatio (8): α ( ω ) = x x A l A2 2 x ( x, ω ) (, ω ) 2 (8) where A ad A 2 represet the amplitudes of two sigals propagatig with distace x ad x 2 respectively. Air valve Air bladder Chael Chael 2 Chael 3 Chael 4 Chael 5 Chael 6 FIGURE 4. Liear trasducer array: Couplig is ehaced by the use of a air bladder which applies pressure to the trasducers ad pushes them ito the coatig. (Pressure ~ 30 psi). Air supply coectio Each coectio cotrols 4 trasducers Break out box To TeleTest uit (a) (b) FIGURE 5. (a) Experimet set up ad (b) sample sigal of 30 KHz circumferetial SH wave, with a expoetial curve fittig to the wrap aroud sigal peaks which gives the atteuatio factor. 68

5 Atteuatio of the wrap aroud sigals could be a excellet feature for defect detectio i additio to the direct defect reflectio echoes ad wrap aroud wave amplitudes. The liear array trasducer used i this experimet was dry coupled by applyig sufficiet pressure o it, resultig i sigal amplitude chages due to the air bladder pressure variatios. However, the beefit of usig the atteuatio costat feature is that it is cosistet uder differet couplig pressures. This was proved by the experimetal study. Two experimets were performed i the pulse-echo mode. Oe was a hole drillig experimet ad the other was a delamiatio/gouge detectio experimet. Hole Detectio Experimet Experimets o a hole were carried out as show i Figure 6(a) where the array is about 90 away from the hole. The sigal i chael 3 is show i Figure 7. Apparet reflected echoes from the hole could be foud betwee two wrap aroud echoes for chael 3. Comparig Figure 7(a) ad (b), it ca be foud that oe-ply coatig reduces sigal amplitudes but ot ispectio capability. Because the hole is physically oe quarter of the circumferece away from the array, the reflected wave was right i the middle of the two wrap aroud waves, idicatig a wave propagatio distace of a half circumferece. The result i Figure 7(c) shows a testig blid area whe the hole is 80 away from array. The atteuatio costats were summarized for all 6 chaels i Figure 8 cosiderig with/without coatig ad with/without hole. Compared with the o-defect case, the atteuatio costats for the with-hole case do ot icrease much, which meas that small defects such as a hole caot cotribute a large effect o sigal atteuatio. hole gouge (a) (b) FIGURE 6. Detectio experimets of (a) a hole ad (b) gouge i a 0 schedule 40 pipe, usig 30 KHz circumferetial SH waves geerated by a liear array trasducer. Hole (a) o tar coatig, hole about 90 away from array Hole echoes (b) with oe-ply tar coatig, hole 90 away from array (c) with oe-ply tar coatig, hole 80 away from array FIGURE 7. 30KHz circumferetial SH wave sigal for Chael 3, showig the hole has strog itermediate reflectio echoes. 69

6 Atteuatio Costat o defect, o coatig hole, o coatig o defect, oe-ply coatig hole, oe-ply coatig Chael Number FIGURE 8. Atteuatio costats at chael -6 for 30 KHz SH wave i a 0 schedule 40 pipe with a hole 90 away from array. Note the existece of small hole does ot itroduce apparet atteuatio. Gouge Detectio Experimets As with the hole detectio, experimets o the gouge as show i Figure 6(b) were also doe. For coated pipe, a Teflo tape isert was used to simulate a delamiatio betwee the gouge ad the tar coatig. I this study, the gouge/isert was moved over differet distaces from the trasducer array positio i order to fid the appropriate ispectio techique. First, the array was placed 0 away (about 06 ) from the gouge s top edge. The results are show i Figure 9(a) ad (b) for o tar coatig ad tar coatig respectively. Gouge echoes ca be see for both cases, although the echoes are ot as apparet as those from the through hole show before. This is because the through hole has a sharp edge which easily results i wave reflectios. However, for the gouge, the edge is ot so sharp ad regular, thus leadig to lower reflectio amplitudes. The secod array positio was about 6 away from the gouge with results show i Figure 9(c). This time the echoes are closer to the wrap aroud sigals ad partially superimposed with them. This is caused by the wide pulse width for the 30 KHz frequecy, which geerates a certai ispectio blid area. To overcome this shortcomig, chagig array positios is a solutio. Atteuatio costats for the 6 chaels i Figure 9(a) ad (b) are summarized i Figure 0 ad compared with the two previous o-defect results, with/without coatigs. It ca be observed from the figure that the atteuatio costats mostly icrease i such a sequece:. o defect without coatig; 2. gouge/isert 0 away from trasducer without coatig; 3. o defect with oe-ply coatig; 4. gouge/isert 0 away from trasducer with oe-ply coatig. Gouge h Gouge Gouge (a) o tar coatig, gouge top edge about 0 (06 ) away from array (b) oe-ply tar coatig with Teflo isert, gouge top edge about 0 away from array (c) oe-ply tar coatig with Teflo isert, gouge top edge about 6 away from array blid ispectio area FIGURE khz circumferetial SH wave sigal of Chael 3 for the experimet o gouge. 70

7 The third array positio was 80 away from the gouge by puttig the array o top ad the gouge o the bottom. As predicted, o defect echoes were foud for both cases with/without coatig. Observatio of the atteuatio costats is similar with that i Figure 0. These observatios could give elightemet o how to use the atteuatio costat feature. If there are o reflected defect echoes but obvious higher atteuatio costats are foud, the trasducer positio could be chaged by a certai umber of degrees to ispect the blid area, which is wide for low frequecy. A test protocol based o this study is recommeded ad summarized i a flow chart as show i Figure. Two liear arrays are suggested, or the repositioig of oe, to acquire data. If itermediate echoes are foud, a delamiatio or other defect exists. If ot foud, atteuatio costats will be used as aother feature to tell whether there are o defects or the coatig is too thick to tell. Note that the protocol is based o a hypothesis that this method works if delamiatio is followed by corrosio. Geerally we will ot see a subtle delamiatio Atteuatio Costat o defect, o coatig gouge, o coatig o defect, oe-ply coatig gouge, oe-ply coatig Chael Number FIGURE 0. Atteuatio costats at chael -6 for 30 KHz SH wave i a 0 schedule 40 pipe for four cases: o defect without coatig; gouge/isert 0 away from trasducer without coatig; o defect with oe ply coatig; gouge/isert 0 away from trasducer with oe ply coatig. Look for presece of itermediate echo, usig 2 liear arrays or repositio to acquire data Yes No Defect Measure wrap aroud atteuatio costats < a > a No defect Either o-plaar defect or coatig too thick to tell Save as referece data for further examiatio, to moitor a chage OR measure coatig thickess, if < t, the o-plaar defect, if > t, the defect call is questioable FIGURE. A test protocol for defect detectio i a tar coated pipe usig circumferetial SH waves geerated by a liear trasducer array. This shows a dramatic accomplishmet for the tar coated pipelie ispectio. 7

8 CONCLUDING REMARKS Theoretical work has bee carried out by developig dispersio curves ad wave structures for circumferetial guided waves i a pipe coated with a viscoelastic material. Wave structure aalysis was coducted at selected frequecies, givig isight o defect ispectio by lookig at the eergy distributio alog the wall thickess directio. Both atteuatio ad wave structure studies suggest the use of low frequecies. A liear array trasducer has bee explored for defect detectio i coated pipe. Defect reflectio echo ad atteuatio of wrap aroud sigal amplitudes are foud to be two feasible features. A test protocol based o this study is recommeded ad summarized. Two liear arrays are suggested, or the repositioig of oe, to acquire data. This study has show a effective accomplishmet for the defect ispectio of a tar coated pipelie. REFERENCES. Rose J.L., J. Press. Vess. Tech. 24(8), (2003). 2. Liu, G. ad J. Qu, J. Appl. Mech. 65, (998a). 3. Luo, W., Rose, J. L. ad Kwu H., RNDE 5, -3 (2004). 4. Christese R.M., Theory of Viscoelasticity: A Itroductio, Academic Press, New York, Zhao X., Rose J.L., J. Acoust. Soc. Am. 5(5), (2004). 6. Xu, W., Jeot F., Ourak M., Modal Waves Solved i Complex Wave Number, i Review of Progress i Quatitative Nodestructive Evaluatio 24, edited by D. O. Thompso ad D. E. Chimeti, AIP Coferece Proceedigs vol. 760, America Istitute of Physics, Melville, NY, 2004, pp