Factrs affecting the detectability f vids by infrared thermgraphy P. Wyssl, Th. Luthil, R. Primas 1 and O. Zgmal 2 ISwiss Federal Labratries fr Materials Testing and Research (MPA), Uberlandstrasse 129, CH-86 Oubendrf, Switzerland; 2SC SA, Hinterbergstrasse 32, P.O. Bx 55, CH-633 Cham, Switzerland Abstract During the manufacturing f laminar plastics, debndings between layers can ccur due t bad curing. Tw inspectin techniques by infrared thermgraphy have t be cnsidered: Impulse r transient thermgraphy and lck-in thermgrapy. Fr bth techniques, the detectability f a defect and the factrs affecting it are majr issues. The first investigated material is istrpic plyvinyl chlride, the secnd is anistrpic unidirectinal carbn fiberreinfrced plastic. This article will discuss the abve mentined factrs and will derive sme practical rules abut the detectability f a defect. Fr the tw base materials analysed, the cmplex cntrast is studied as a functin f the defect diameter t depth rati and the mdulatin frequency f the heating surce. Practical rules will be derived and a cmparisn between the tw techniques will be made. 1. Intrductin Laminar plastics such as wven fiber-reinfrced plastics are becming mre and mre widely used due t their gd mechanical prperties and their ability t be tailred t the intended applicatin. During their manufacturing, debndings between layers can ccur due t bad curing. Fiber-reinfrced plastics in frm f sheets are als used t reinfrce civil engineering structures like bridges [1]. The presence f air inclusins r delaminatins in the bnding layer between the sheet and the structure can drastically reduce the mechanical strength f the reinfrcement. Therefre, a fast testing methd is needed in rder t detect such flaws. Until nw, nly the time cnsuming ultrasnic inspectin methd came int cnsideratin. Infrared thermgraphy is a particularly well suited inspectin methd, because it is fast and gives an image f the lcatin f the defects. Tw inspectin techniques by infrared thermgraphy have t be cnsidered: the impulse r transient thermgraphy [2] and the lck-in thermgraphy [3]. 2. xperimental setup As defects act as thermal barriers, the thermal cntrast between the temperature measured abve a defect and the ne measured n the sund material is dependent n the diameter D f the defect, its depth x and the thermal prperties f the materials. The experiments were made with istrpic plyvinyl chlride (PVC) and unidirectinal carbn fiber reinfrced epxy resin (CFRP); their thermal prperties are summarized in table 1. It is t pint ut, hwever, that the thermal prperties f CFRP are very strngly dependent n the fiber cntent and the fiber type [4]. material density p heat capacity c thermal cnductivity A. diffusivity IX [kg m- 3 ] [J kg-1 K-1] [Wm-1 K-1] [m2 S-1] PVC 1426 94.17.131-6 CFRPT 135 19.61.45 1-6 CFRP" 135 19 4.97 3.651-6 Table 1 : Thermal prperties f PVC and CFRP (The thermal diffusivity alng the fibers fr CFRP was taken frm the literature) QIRT 96 - urtherm Series 5 - dizini TS, Pisa 1997
A frnt disc f the relevant material was put n a PVC back disc where the defect was simulated by a hle drilled in its center with a diameter x between 3 and 1 mm. A thermal paste prvided a gd thermal cntact between the discs. By evacuating the air behind the secnd disk, the tw are pressed unifrmly and the hle simulates a perfect defect where the heat cannt flw. The frnt face f the cver disc was painted black in rder t get an unifrm and knwn value f the emissivity n the surface. Figure 1 shws this experimental setup. The heating surces were: fr the impulse methd tw flash lamps with a ttal utput energy f 6 kj r 12 kj, with a distance f.3 m t the sample and an incidence angle f 45, fr the lck-in methd a halgen lamp with an electric pwer f 1 kw, equipped with a filter in rder t suppress the infrared radiatin in the sensitivity range f the camera (8-12 IJm). The distance t the sample was.35 m and the angle f incidence 9. This lck-in heating equipment was delivered by Agema. The infrared camera used is an Agema 9 LW. Due t the anistrpic behaviur f CFRP, the thermal image f a rund shaped flaw is nt a circle but mre an val [5]. Figure 2 illustrates this fact clearly. Since the samples were analysed with bth the impulse and the lck-in methds it is useful t review the relatins between them. Seidel [6] shwed that the heat pulse can be apprximately cnsidered as a mnchrmic thermal wave with a time dependent diffusin-length ljeff (t). Under this assumptin, the thermal cntrast K(t) in the time dmain is given by the abslute value f the cmplex cntrast K(f) btained frm the frequency dmain where 1J(f) is replaced by ljell(t). Table 2 summarizes the relatins. i! Frequency dmain ~(f) =~ 1tf K(f) = KA(f) + i ~(f) Time dmain ~eff (t) = 2..rat [7] K(t) = I K(f = _1 ) I 41tt Table 2 : Relatins between the frequency dmain and the time dmain 3. Transient r impulse thermgraphy,,i The detectin limit f impulse thermgraphy depends n several parameters. By means f finite element cmputatin and measurement the dependencies belw were analysed: 1. the aspect rati that means the rati Dx, where D is the diameter and x the defect's depth 2. the anistrpy f the sample's material 3. the intensity f the heating pulse Figure 3 displays the thermal cntrast versus the aspect rati fr PVC as an example f istrpic plastic material. The thermal cntrast diminishes faster, when the depth is varied while the diameter is being kept cnstant. The detectin limit is Dx = 3. Figure 4 shws the thermal cntrast versus the aspect rati fr unidirectinal CFRP. The cmputed curves fr the anistrpic material and the istrpic thught material with a thermal cnductivity parallel t the fibers are clse tgether and in gd agreement with the measurement fr the same heat pulse intensity. The istrpic material with a thermal cnductivity perpendicular t the fibers (the resin nly) wuld allw a quite better detectability than the cmputatin suggests. In fact twice the intensity is needed t reach this detectability fr the real CFRP sample. It is well knwn that the limited slit respnse f the camera can als affect the detectability. Thrughut these experiments, hwever, care was taken t keep this influence ut. 4. Lck-in thermgraphy Due t the scanning frequency f the camera, nly discrete heating frequencies are pssible. Fr the fllwing interpretatins f the results, the abslute cntrast values are nt f interest. Figure 5 and 6 shw the results fr PVC (thickness 1 mm) and CFRP (thickness 1.7 mm), respectively. 228.i
Bth materials have a similar behavir, the results f the istrpic PVC, hwever, are less nisy. The cntrast develpment behaves like a cunter-clckwise slpe in the negative phase cntrast - psitive amplitude cntrast quarter f the respective crdinate system, beginning at r near the rigin. The different curves are crrespnding t the Dx-ratis, the different measurementpints are crrespnding t heating perids between 8.5 t 25 s fr PVC and 4.3 t 51.2 s fr CFRP, respectively (nte that sme lng measurements are missing). Fr PVC the third pint f each slpe crrespnds t a perid f 25.6 s, equal t a thermal diffusin length f 1.3 mm. With the exceptin f Dx = 3, where all measurements are belw a reliable value f phase, these measurements are clearly away frm the rigin. Fr CFRP the secnd pint f each slpe crrespnds t a perid f 8.53 s, equal t a thermal diffusin length f 1.11 mm, taking the thermal prperties perpendicular t the fiber directin and 3.15 mm taking the ther value. T take the first value seems mre crrect, as the relevant thermal diffusin length is fr the directin perpendicular t the fiber. Fr bth materials the amplitude cntrast is lw fr small perids and the maximum cntrast appears at a much larger perid than the ne crrespnding t the thermal diffusin length f the defect, arund 2.5 mm fr PVC and arund 2. mm fr CFRP. Within the measurement uncertainty these statements are nt influenced by the aspect rati. Fr the measurements n PVC, hwever, it seems that the aspect rati is slightly influencing the slpe, turning it cunterclckwise fr increasing values. 5. Cnclusins Fr the transient technique it was shwn clearly that the heat input is influencing the detectability. Fr the lck-in technique similar experiments were nt carried ut; it is assumed, hwever, that the same wuld be true. The maximum thermal cntrast crrespnds fr bth techniques t a similar thermal diffusin length, arund 2 t 3 times larger than the effective defects depth. Fr the examined materials, fr small heating perids with the lck-in technique the thermal cntrast is mainly due t the phase cntrast and detectable if the thermal diffusin length reaches the defects depth. Therefre wrking with the transient technique, the peratr will cncentrate n the maximum thermal cntrast with the disadvantage f having n precise infrmatin abut the depth f the defect riginating it. It can mre exactly be fund using the lck-in technique and cncentrating n a grwth f the phase cntrast. 6. Acknwledgments Mr. Zgmal is particularly indebted t Dr. Walther and Dr. Seidel, frm the Friedrich-Schiller University Jena (D), fr very fruitful discussins. RFRNCS [1] MIR (U.), DURING (M.), MIR (H.) and SCHWGLR (G.). - Strengthening f structures with advanced cmpsites. Clark (J.L.), Alternative Materials fr the Reinfrcement and Prestressing f cncrete, Champman & Hall, Lndn, 1993, p. 153-171 [2] LAU (S.K.), ALMOND (D.P.), PATL (P.M.), CORBTT (J.), QUIGLY (M.B.C.). Analysis f transient thermal inspectin. Infrared Technlgy and Applicatins, SPI 132, 199 [3] BUSS (G.), WU (D.) and KARPN (W.). Thermal wave Imaging with phase sensitive mdulated thermgraphy. J. Appl. Phys. 71, 1992, p. 3962-3965 [4] HANCOX (N.L.) and MAYR (R.M.). Design Data fr Reinfrced Plastics, A Guide fr ngineers and Designers. Chapman & Hall, Lndn, 1994 [5] LOTH I (Th.), MIR (H.), PRIMAS (R.), ZOGMAL (.). Infrared inspectin f external bnded GFRP-sheets. Schickert (G.) and Wiggenhauser (H.) ed., Nn-destructive testing in civil engineering (NDT-GJ, DGZfP, Berlin, 1995, p. 689-696 [6] SIDL (U.). Quantitative Auswertung phtthermischer Messungen zur rkennung und Gharakterisierung vn Inhmgenitaten in Festkrpem. Ph.D. thesis, Friedrich-Schiller University, Jena, 1996 [in German] 229
[7] LAU (S.K.), ALMOND (D.P.) and PATL (P.M.). Transient thermal wave techniques fr the evaluatin f surface catings. J. Phys. : Appl. Phys. 24, 1991, p. 428-436 Cver-sheet PVC r CFRP Diameter D f the hle x Fig. 1. xperimental setup 3(j~9 3QA 2~~S l~a 28.. 9 28.. 5 Fig. 2. Typical image f a circular vid in a CFRP-sheet i I i u; f! " (.) iti CI> S ::I 'x s :;;.1 deltat max. [C] PVCPVC istrpic D=12 mm; 1= 4.6 kjm 2 cmputed --- deltat max. [C] PVCPVC istrpic x=1. mm; I = 4.4 kjm 2 D varying deltat max. [C] PVCPVC istrpic D=12 mm; I = 4.4 kjm 2 x varying.4 1. Rati defect diameter t the defect's depth (Dx) 1. Fig. 3. Thermal cntrast versus Dx fr PVC 23
1~-----------------------------------------------------. t. measured J =4 kjm 2 measured J = 8.5 kjm 2 cmputed J = 4 kjm 2... cmputed istrp II J = 4 kjm 2 - - - - - cmputed istrp T J = 4 kjm 2,.,... z;.1 +--------Y-----'----'---'---'---'--'--'-+--------'-------'---'---''--'--I...L--'-I 1. 1. aspect rati Dx 1. Fig. 4. Thermal cntrast versus Dx fr CFRP 5 I I ---.--- Dx = 3-2 4 6 ~Dx=4-5...: ---.-- Dx = 5 CI II> :!:!. ---- Dx = 6 "lii -1 l!! (.) II> III ca.c a. -15 ---.. -- Dx = 8 -----6- Dx = 1 amplitude cntrast [%] Fig. 5. Cmplex cntrast develpment fr PVC 231
-1 2 3 4-2..: I tn + Q) --.-- Dx = 3.7 ~ -4 "lij "'. I -G-Dx=4.7 ~ c -+ -6.. --+-- Dx = 5.6.~ Q) I) c Dx = 7.5.s: C -8 --..-- Dx = 9.3-1 ~~ -12 amplitude cntrast [%] Fig. 6. Cmplex cntrast develpment fr CFRP 232