1
Infrared thermography for inspecting the adhesion integrity of plastic
welded joints
M. Omara, M. Hassana, K. Donohueb , K. Saitoa and R. Allooc
(a) Mechanical Engineering Department, University of Kentucky
(b) Electrical Engineering Department, University of Kentucky
Lexington KY 40506
(c)Toyota Motor Manufacturing North America Inc.
Erlanger KY 41018
Abstract
This work aim at developing a non-destructive tool for the evaluation of bonded plastic
joints. The paper examines infrared thermographic transmission and reflection mode
imaging and validates the feasibility of the thermal NDT approach for this application.
Results demonstrate good estimation performance for adhesion integrity, uniformity and
bond strength using a transmission mode application of infrared thermography. In
addition, results from a pulsed infrared thermographic application using a modified
dynamic infrared tomography scheme show good performance for estimating adhesion
layer thickness mapping and detecting delaminations.
Key words: Plastic kissing bonds, Adhesion uniformity, Bond strength, Pull force,
Dynamic tomography, Delaminations detection.
2
Introduction
Plastic bonded structures have found increased acceptance in new products and
applications due to its light weight and strength in preserving structural integrity. Infrared
thermography has been reported to be successfully applied in the evaluation of wide
range of materials including wood (wood poles inspection and lumber industry) [1,2] ,
brick and reinforced concrete (in buildings and smokestacks [3-5]) and finally a variety of
composites and polymers such as, Carbon Fiber Reinforced Plastic CFRP and graphite
epoxy composites [6,7] which motivated the use of the infrared thermography for this
application.
Motivation
The non-destructive investigation of the kissing bond region in plastic welded joints
constitutes an important challenge in the plastic molding industry because, of the wide
variety of defects that exist in these joints such as, delaminations, the adhesion layer
thickness and uniformity. Those defects may lead to the breakage of the bonds or,
misalignments in the final product structure. Since these joints are manufactured
through mass production lines, the application of a classical non-destructive testing
procedure may hinder the production cycle times of these products due to its contact
nature, or slow operation time. The infrared thermography doesn’t pose these limitations,
but has the potential to provide the flexibility of detecting wide variety of defective
behaviors by relating those defects to a specific thermal signature that could be detected
remotely and in real time. Thermal NDT techniques have been applied for the inspection
of some typical encountered defects in similar applications. These applications included
3
the composite crack detection [19,20], the evaluation of impact damage in graphite
epoxy [8] and the delaminations detection [9]. In more related applications Turler [10]
reported the use of steady state thermographic technique in predicting the geometry and
location of defects in adhesive and spot welded steel lap joints. Aglan et al [11] used the
infrared thermography for the detection of cumulative fatigue disbond of adhesive steel
joints. None of the currently applied techniques including classical and thermal NDT
provides a quantative value for the bond strength under inspection, which is a potential
that may result from this work. The use of infrared thermography for the inspection of
plastic bonded joints is considered to be a novel approach.
Problem Description
The plastic joints under study are made through an extrusion molding process. The
material involved is a composite of two layers of a polymer (High Density Polyethylene
HDPE) with an adhesive interface. This composition is shown in figure (1). Carbon
pigments have been added to one of HDPE layers for darkening to provide better
thermal absorption and emission properties for insulation purposes. The geometry of the
joints under inspection along with a cross section are shown in figure (2), this geometry
is represented in two cups welded together at the neck region. The bond interface is
located between the white HDPE layer and the rest of the geometry.
The main defectives encountered in this particular product could be categorized into
structural and adhesion. Structural defects include misalignment in the neck region and
the variations in the diameter of contact, which may affect the breakage mode of those
joints as shown in figure (3). Adhesion related defects include lack of adhesion,
adhesion thickness, uniformity and delaminations. All of these defectives ultimately
affect the strength of joint.
4
In the following the transmission mode of infrared thermography will be introduced and
applied for the validation of the technique feasibility through studying the effect of bond
integrity on the thermal wave conduction through the bond region. The pulsed reflected
mode also will be presented for the thickness mapping and delaminations detection.
The transmission mode
The transmission mode of infrared thermography is where the stimulation and the
infrared detector are located at opposite sides, to allow for monitoring of the transmitted
heat through the part. This mode is used as a first qualitative step due to its simplicity
[18] and the fact that it doesn't require complicated setup.
The experimental setup for this mode is shown in figure (4). Boiling water at 100oC is
(HDPE tolerates temperature of 120 Co ) used as a contact stimulant; the choice of
stimulant is based on the fact that water at phase change will preserve a constant
temperature. An infrared bolometer (commercial name ThermaCam SC 2000 of Flir, MA)
is used to monitor the temperature rise. The transient temperature evolution curve is
recorded for a number of samples. The evolution curves exhibited a linear behavior with
different slopes for some samples; this behavior is related to the strength of contact
since a well-contacted joint will conduct heat faster than a joint with air gaps presenting a
larger thermal resistance for the heat flow leading to heat entrapment over the defect.
To quantify this observation a criterion to describe the strength of the joint is needed.
This criterion is chosen to be the pull force needed to break the bond between the two
cups using a tensile machine to pull the two cups apart. Figure (5) shows the results of
this application, where the heat evolution curves are shown for joints corresponding to
different pull force values. In this application there are two sets of plastic joints that differ
in size, and therefore the heat evolution curves are recorded separately in figures (5 a,b).
5
It's clear that some of the samples with low slope values exhibit unique breakage modes
due to the weak bond such as samples number (1) and (2). The no-fusion sample refers
to the case of contact without bonding, even though this technique is not mainly aimed to
detect such cases. Transmitted mode infrared thermography provides a quantitative
measure of this plastic bond strength by applying a simple and inexpensive setup that
emphasizes its potential utility. The transmission tests results were consistent upon
repeatable measurements.
The transmission mode is found to be useful in predicting the uniformity of the joint weld.
Figure (6,a) shows a thermogram image captured simultaneously for two samples, and
the corresponding pulled samples are shown in figure (6,b). Seeking a qualitative
description for testing uniformity of weld; the time needed for each pixel within the bond
region for some samples to exhibit a certain temperature change is recorded and
rendered into 3 dimensional plots to produce figure (7) which shows the results for
a CT o0.3=∆ . From figure (7) it's clear that the regions of disbond require larger time
periods to conduct heat.
The reflection mode
For the reflection mode of infrared thermography, the stimulant and detector are situated
at the same side and monitor the reflected thermal wave effect on the surface
temperature. The main aim of using this technique is to investigate some of the
subsurface features such as delaminations using a non contact mode. The transmission
mode isn't suitable for non-contact rapid scanning applications due to the fact that the
thermal wave must conduct through the whole bond making the average observation
time period in the order of 100 seconds per scan for this mode.
6
An analytical model was devised in this study to help predict the behavior of this material
under the pulsed stimulation. This analytical model is shown in figure (8) and is intended
to simulate the bond region. This model utilizes an implicit finite difference approach to
represent the general homogenous heat conduction equation in cylindrical coordinates
equation (1).
tT
zTT
rrT
rrT
∂∂
=∂∂
+∂∂
+∂∂
+∂∂
αθ111
2
2
2
2
22
2
(1)
The symmetry simplifies the analysis to where only a pie slice is studied. The back and
side walls are assumed to be thermally adiabatic. The boundary conditions are applied
according to a radiation excitation source but could be adjusted to fit the transmission
mode with contact source.
Tossell [12] discussed the boundary conditions and the effects of the method of
stimulation delivery on the thermography numerical modeling.
The convection and radiation surface heat losses are included in this model as in
equations (2) & (3).
))(( 44ambsurfrad TtTq −⋅⋅= εσ (2)
))(( ambsurfconvfreeconv TtThq −⋅= − (3)
where 28 .1067.5 mW−×=σ , ε is the emissivity, assumed to be 1. surfT is the
temperature of a surface node. ambT is ambient temperature and convfreeh − is the free
convection heat transfer coefficient assumed to be 10.
7
The delaminations in the analytical study are modeled as an air layer [21] that will
constitute an interface to reflect the thermal waves due to the thermal mismatch factor
existed between the polymer and the air, the thermal mismatch factor is computed
through equation (4)
airHDPE
airHDPE
eeee
+−
=Γ (4)
where HDPEair ee , is the effusivity of air and HDPE respectively computed as:
cke ⋅⋅= ρ (5)
where k is the thermal conductivity KmW ./5.0= , ρ is the density 3/950 mkg= and
c kgKJ ./1900= is the specific heat.
The reflection of thermal waves from subsurface features affects the facial temperature
value at that point causing a deviation from its surroundings. This deviation is monitored
and recoded as the thermal contrast. Figure (9) shows the sample-cooling curve after
depositing a rectangular pulse of 2/100 mkWQ = over a sound (non-defective) spot
and another one that is located over an air gap. Figure (10) shows the thermal contrast
for different delaminations at different depths, such figure is computed from similar
figures as in (9). The contrast peak decays as the depth increases. which is expected
since the thermal front will weakens due to the diffusion effect. Figures (9) and (10)
provide useful information to guide the pulse infrared thermographic procedure in
deciding on the best observation time window and sensitivity limits.
8
The pulse reflected infrared thermographic setup is shown in figure (11) where a pulse
unit with radiation power of 6.4 kJ (through two banks of capacitors) and duration of 15
ms is used (commercial name BALCAR Source, France). The geometry of the joints
posed the challenge of scattering the radiation power off the sides of the cups, so the
use of a Bi-tube pulse head is warranted to provide a focused pulse. The pulsed
infrared thermography results could be interpreted and processed in different schemes
such as, Pulse Phase Thermography Maldague et al [13], Synthetic Infrared
Thermography Shepard et al [14] and Infrared Dynamic Tomography Vavilov et al [15].
In this study the Dynamic Infrared Tomography will be utilized to process the data
resulted from the pulse application with some modifications applied to this scheme.
Infrared Tomography
This technique is intended in this case study to map the thickness of the adhesion layer
of the kissing bond. The application of this technique is based on the procedures
proposed by Vavilov et al in 1986 (Vavilov et al 1990 [15]) the details of the computation
involved in this technique could be found in [16]. This technique is based on
establishing a maximum contrast matrix and a time-gram matrix. The maximum contrast
matrix presents the maximum deviation values exhibited through the transient response
to the pulse and the time of occurrence for these values is reported by the time-gram
matrix. The Full Width Half Maximum (FWHM) contrast could be used in other
application [22]. In this application a self-referencing procedure (Shepard et al [17]) will
be utilized in computing the thermal absolute contrast needed for the maximum contrast
matrix. The self-referencing procedure is based on computing the deviation between
each pixel in the thermogram image and a small local neighborhood (kernel of pixels)
surrounding it. This procedure eliminates the need for a known sound area within the
thermogram for thermal deviation calculations and guarantees the consistency since
9
each thermogram is referenced to itself. A computer code using MATLAB is prepared for
the modified infrared dynamic tomographic calculations. A further modification is applied
for the representation of the tomographic results to enable it for thickness mapping. The
traditional representation of this technique is based on slicing the material under
inspection into depth slices that correspond to the distribution of the thermal properties
at those depths. In this application the tomographic output will be represented in the
form of a thickness map in order to include all the information in a single image rather
than multiple slices. The thickness mapping scheme is applied using Balageas et al [7]
approach in relating the minimum effusivity curves to the depth through equation (6)
95.0min ][
oo e
etZ ⋅⋅= α (6)
Where Z is the depth of the defect, mint is the occurrence time of the minimum of the
normalized effusivity oee curve. Another form of this equation in terms of the thermal
contrast could be found in [18] as equation (7).
bjiCjitajiZ ),(),(),( maxmax ⋅⋅= (7)
where ),( jiZ is the depth at location ),( ji , 334.0,432.0 −== ba are constants
determined experimentally (dependant on the material under inspection), ),(max jit is the
time at position ),( ji taken from the time-gram matrix, ),(max jiC is the contrast value at
),( ji from the contrast matrix.
The result of using this approach is shown in figure (12) that shows a drop in thickness
map in the middle of the kissing bond indicating delaminations. To validate this results
an ultrasound C-scan was obtained with an Ultrascan 5 (US Ultratek, Martinez, CA). For
10
this scan the transducer frequency was 5 MHz, the sampling rate was 100 MHz, the X-Y
scan increment was .012 inches, and the sonic velocity was .0741 inches/microsecond.
The velocity of sound was estimated with a contact transducer applied over parts of
samples including regrind and virgin layers that could be measured directly with calipers.
The X-Y increment was chosen to be about ¼ the width of the beam filed of the
transducer (.04 inches). The C-scan result is shown in figure (13) which verifies the
existence of delaminations in this sample. The average error between the depths
reported by the thermal technique and those of ultrasound measurements is about 12 %.
Conclusion
The infrared thermographic applications in transmission and reflection modes have been
successfully used for the evaluation of adhesion integrity in welded polymer plastic joints.
This evaluation included the inspection of adhesion uniformity and thickness mapping
was achieved using a modified dynamic infrared tomographic scheme. The infrared
thermography provided a quantative non-destructive tool for evaluating the strength of
the plastic polymer welded joint by correlating it with the thermal wave travel within the
joint under inspection. The infrared thermography could be utilized for the inspection of
different defective behaviors in plastic kissing bond applications making it a flexible and
an effective inspection tool for this application.
Acknowledgement
This study was sponsored by Toyota Motor Manufacturing North America Inc. Kentucky.
11
References
[1] A. Wyckhuyse, X. Maldague, "Study of wood inspection by infrared thermography,
part I: Wood pole inspection by infrared thermography" Research in nondestructive
evaluation, Vol 13. Issue 1, PP:1-12, Mach 2001
[2] A. Wyckhuyse, X. Maldague, "Study of wood inspection by infrared thermography,
part II: Thermography for wood defects detection" Research in nondestructive evaluation,
[3] Takahide Sakagami, Shiro Kubo "Development of a new non-destructive testing
technique for quantative evaluations of delaminations defects in concrete structures
based on phase delay measurements using lock-in thermography" Infrared physics and
technology Vol. 43 2002, Pages: 311-316
[4] V. Vavilov, V. Demin, "Infrared thermographic inspection of operating smokestacks",
Infrared physics and technology Vol 43. PP:229-232, 2002.
Vol 13. Issue 1, PP:13-21, Mach 2001
[5] D.J. Titman "Applications of thermography in non-destructive testing of structures"
NDT and E International Vol [34] PP: 149-154, 2001.
[6] Ahmad T., Jin, H. J. , Wang, X., Favro, L.D. and Thomas, R.L. "Infrared thermal wave
studies of composites" rev of prog in quantative non-destructive evaluation. Vol 10, New
York 1991 P:2173.
[7] Balageas D. Deom A. Boscher D.,”Characterization and nondestructive testing of
carbon-epoxy composites by a pulsed photothermal method,” Materials Evaluation, Vol.
45,1987.
[8] Jones, T. and Berger, H. "Thermographic detection of impact damage in graphite
epoxy composites" Material Evaluation. (50) 1992 Page: 1446.
[9] J. Varis, J. Rantala and J. Hartikainen "An infrared line scanning technique for
detecting delaminations in carbon fiber tubes" NDT and E international Vol. 29 No. 6
PP:371-377, 1996
12
[10] Turler D. "Predicting the geometry and location of defects in adhesive and spot
welded lap joints using steady state thermographic techniques" Thermosense XXI, Vol :
3700 Orlando-Florida, PP: 54-62, 6-8 April 1999.
[11] H. Aglan, S. Shroff, Z. Abdo, T. Ahmed, L. Wang, L.D. Favro and R. L. Thomas.
"Cumulative fatigue disbond of adhesive joints and its detection using thermal wave
imaging" Review of progress in quantative non-destructive evaluation Vol: 14, PP: 431-
438, 1995
[12] D. A. Tossell "Numerical analysis of heat input effects in thermography" Journal of
nondestructive testing Vol 6 No.2 1987.
[13] X. Maldague, F. Galmiche, A. Ziadi "Advances in pulsed phased thermography".
Infrared Physics and Technology 43,PP: 175-181, 2002
[14] S. Shepard, J. Lahota, B. Rubadeux, T. Ahmed "Reconstruction and enhancements
of active thermographic images sequence", Optical Engineering 42 (5) PP: 1337-1342,
May 2003.
[15] Vavilov V. “Dynamic thermal tomography: perspective field of thermal NDT” in
Semanovich SA, ed. Thermosense XI, SPIE Proceedings, 1313. PP:178-182, 1990.
[16] X. Maldague, J. Cote, D. Poussart, V. Vavilov "Thermal Tomography for NDT of
industrial materials" Canadian Society of Non-destructive Testing Journal PP:22-32,
May-June 1992.
[17] S. Shepard, R. Ducar. ”Quantative Infrared Defect Detection in Composite
Aerospace Structures.” 45th international SAMPE Symposium 2000.
[18] X. Maldague “ Theory and Practice of Infrared Technology for Nondestructive
Testing”. Wiley Interscience Publication. 2001
[19] L.D. Favro, T. Ahmed, Xianyan Han, L. Wang and S. M. Shepard "Thermal wave
imaging of aircraft structures" Review of progress in quantitative non-destructive
evaluation Vol. 14 PP:461-466, 1995.
13
[20] Varis J. Lehtiniemi R. "A thermal non-destructive evaluation system for detecting
vertical cracks in unidirectional carbon fiber composites" Rev. Sci. Instrum. Vol 68 Issue
7, PP:2818-2821, 1997.
[21] D.L. Balageas, J.C. Karpez, P. Cielo "Pulsed Photothermal modeling of layered
materials" J. Appl. Phys. 59 (2), 15 January 1986.
[22] D.P. Almond and S.K. Lau "Defect sizing by transient thermography. I : an analytical
treatment" J Phys D: Appl Phys Vol [27] PP:1063-1069, 1994.
14
3
2
1
Legend: 1 Virgin High Density Polyethylene HDPE (black colored due to carbon addition) 2 Adhesive 3 Virgin High Density Polyethylene HDPE
Figure 1. HDPE material composition
18
Figure 5. Left: Heat evolution curves A,B Legend: shows breakage pull force values in Newtons Right: Sample 1 and 2 breakage mode
1
2
1
2
A
B
19
Figure 6. Adhesion uniformity effect on heat conduction A: Thermograms for the two samples, B: Corresponding breakage mode
A
B
Pull force = 3055 N Pull force = 7355 N
20
Figure 7. Time plots needed for temperature change of 3 oC. Z axis: time in seconds. X,Y axis : bond spatial coordinates
Time in seconds