Composites Part B 147 (2018) 128–134

Contents lists available at ScienceDirect

Composites Part B
journal homepage: www.elsevier.com/locate/compositesb

Depth determination of defects in CFRP-structures using lock-in T

thermography
Sarah Ekanayake∗, Sumit Gurram, Robert H. Schmitt
Laboratory for Machine Tools and Production Engineering (WZL), RWTH Aachen University, 52056 Aachen, Germany

A R T I C LE I N FO A B S T R A C T

Keywords: Due to the increasing application of CFRP in mass application, an economic repair process gains importance. As
D. Thermal analysis the typical repair process starts with the defect detection, the accuracy of the measurement system is a limiting
A. polymer-matrix composites (PMCs) factor of the repair process. Lock-in thermography is a promising approach for non-destructive defect detection.
B. Defect Since the phase images contain a superposition of depth information, the determination of the defects' depth is a
D. Non-destructive testing
challenge. Based on the complex wave field, this paper introduces a method to determine the thermal properties
Thermography
and the defects' depth in CFRP structures using lock-in thermography.

1. Introduction 2. Optical lock-in thermography

To guarantee an economically feasible application of carbon fiber
reinforced plastics (CFRP) long-term operational reliability and eco-
nomic repair processes have to be assured. As the application of CFRP
increases especially in mass markets (e.g. automotive industry), gen-
erously replacing damaged parts, which is an established in aircraft
industry, is not justifiable anymore. Alternatively, economic repair
processes are required. The typical repair process starts with the defect
detection and is limited by the accuracy of the defect detection system.
Due to small objects, e.g. stones or hail, impact damages can be caused
that are optically invisible, but propagate internally [1]. To assure the
safety of the automobile, the impact damage has to be detected and
repaired. If only delamination occurs and the fiber is undamaged, the
damage can be repaired by injection of polymer matrix or adhesive
between loose layers [2,3]. For the repair process, information of depth
position of the delamination is essential.
The non-destructive detection of delamination is still a challenge,
especially considering the repair workshop specific demands (e.g.
portability and cost efficiency). A promising approach is the principle of
lock-in thermography [4,5]. The measurement system is portable, in-
tuitive in use and economic as it covers large components in a single
image. Furthermore, thermography is non-tactile, reliable and capable
to detect impact damages with their respective failure characteristics
e.g. delamination, debonding and fibre fracture [6]. However, depth
information is not easily revealed by application of lock-in thermo-
graphy. This deficit is subject of this work.
Lock-in thermography is based on the measurement of the test
samples' infrared radiation. The principle is illustrated in Fig. 1. Ha-
logen spotlights induce a heat flow into the test object to disturb the
component's thermal equilibrium. The sinus shaped temperature mod-
ulation propagates from the surface through the component and in-
terferes with the reflected thermal wave from the samples' backside or
from inhomogeneities in the inside of the part. The infrared radiation
during the testing sequence is recorded by an infrared (IR-) camera and
Fourier transformed into the frequency domain. As a result, the mate-
rial inhomogeneities become visible on the resulting phase image. The
depth range for the optically excited lock-in thermography varies
through the excitation frequency. The lower the frequency, the deeper
the thermal waves propagates within the component.
The state of the art provides different approaches for depth de-
termination of defects using lock-in thermography. The DIN EN 15042-
2 standard introduces the layer thickness measurement with lock-in
thermography [7]. Nevertheless, either the thermal part properties or a
calibrated spot has to be given to assign the measured phase values to
the depth position. Moreover, only layer thickness and no depth posi-
tion of defects with contact resistance can be measured. Gleiter's ap-
proach is based on the model of the complex wave field [8]. Spiess-
berger apprehends Gleiter's method. In scatter diagrams, the phase
values of two different excitation frequencies are illustrated and ana-
lyzed considering the determination of layer thickness, delaminations
and the reflection coefficient. Spiessberger neglects lateral heat flows

∗
Corresponding author.
E-mail address: s.ekanayake@wzl.rwth-aachen.de (S. Ekanayake).

https://doi.org/10.1016/j.compositesb.2018.04.032
Received 15 March 2018; Accepted 9 April 2018
Available online 10 April 2018
1359-8368/ © 2018 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/BY-NC-ND/4.0/).
S. Ekanayake et al. Composites Part B 147 (2018) 128–134

1+i ω
σ= = (1 + i)
μ 2α (4)

where μ [m] represents the thermal penetration depth, σ [1/m] the

complex wave number, T (x , t ) [K] the locally and temporally oscil-
lating temperature, T0 [K] the amplitude of the temperature at the
surface, x [m] the depth beneath the surface, ω [Hz] the angular fre-
quency of excitation and α [m2/s] the thermal diffusivity.
If the thermal waves strike on an interface, the incident wave is
reflected. The incident and the reflected wave interfere with each other
and cause a phase shift. The reflection of the thermal wave repeats
infinitely on the front and backside of a specimen and is described as
geometric series. As the thermography inspection is reduced to the
examination of the parts' surface (x = 0) , the time dependent tem-
perature development is calculated in equation (5).
n =∞
T (0, t ) = T0⋅e−iωt (1 + R23 e−2dσ ) ∑ (R12 R23 e−2dσ )n
n=0 (5)

where d [m] represents layer thickness, R12 [-] the reflection coefficient
Fig. 1. Principle of lock-in thermography.
between air and specimen, R23 [-] the reflection coefficient between
specimen and substrate, n [-] the number of reflections and t [s] time.
For ideal excitation by the thermography system, the value of the
whereby his procedure looses accuracy for erratic differences in the reflection coefficient at the specimens' surface can be estimated as
depth profile [9]. Holtmann proves that Spiessberger's approach is not R12 = 1 [7]. The value of the reflection coefficient at the defect R23 is
suitable for impact damages. Holtmann compensates lateral heat flows limited by the value range between [-1; 1]. Hence, the geometric series
with a hardware solution and replaces the halogen spotlights with li- converges for n → ∞ to equation (6) [14].
quid crystal display (LCD) projector [10].
1 + R23 e−2dσ
The state of the art provides different approaches for depth de- T (0, t ) = T0⋅e−iωt
1 − R23 R12 e−2dσ (6)
termination of internal defects and highlights the importance of quan-
titative analysis of CFRP for its future success. Due to lateral heat flow, The reflection coefficient R23 is calculated by the effusivity e of the ad-
the accuracy of these methods is reduced significantly with decreasing joining materials at the reflecting interface. In case of reflection at the
defect size [11,12]. In addition, the appearance of contact resistance backside of a CFRP-plate e1 and e2 equals CFRP and air (equation (7))
increases the difficulty to detect delamination. Currently, no reliable
e1 − e2
method for defects' depth determination in CFRP-structures using lock- R23 =
e1 + e2 (7)
in thermography is available.
The subsequently proposed method for depth determination of in- The previous assumptions presuppose an ideal thermal contact be-
ternal defects considers not only the material dependent thermal tween the interfaces. Due to the surface roughness, the assumption of
properties but also the appearance of contact resistance and lateral heat ideal interfaces is not valid. In such cases, the heat transfer between the
flows. Resulting internal defects such as delaminations can not only be interfaces occurs via microscopic contacts. Because of the point like
detected, but the depth position determined precisely. Based on pre- area for heat transfer, a step in the temperature ΔT over the interface
vious studies, the introduced approach models the complex wave field occurs. A similar phenomenon can be observed examining delamina-
and extracts look-up tables for each used case. The measured phase tion. Due to delamination, the different layers of the CFRP separate
values are compared with the theoretical model. With a best-fit algo- from each other, but microscopic contacts remain. For the defects'
rithm, the material parameters and defect depth are calculated. In depth determination in CFRP-structures using lock-in thermography the
contrast to previous research activities, the focus lies on the lateral heat temperature step ΔT due to contact resistance Ω has to be considered
flow minimization and the consideration of contact resistance in order (cf. equation (8)) [15].
to realize a reliable examination of impact damages.
T (0, t ) + TR (0, t ) − TT (0, t ) = ΔT = ΩQ˙ G (8)
where Ω [JK/s] is the contact resistance and Q̇G [J/s] the heat flow
3. Measurement process for depth determination of defects in
through the interface.
CFRP-structures using lock-in thermography
The equations of the reflection coefficient and the temperature has
to be adjusted for the determination of internal defects considering
3.1. Heat transfer during the lock-in thermography measurement
contact resistance (c.f. (9) and (10)).
The heat transfer during the lock-in thermography measurement is 1+i
2
Ωe 2πf
described as thermal wave. Based on the wave equation and the heat R= 1+i
2 + 2 Ωe 2πf (9)
transportation in solid for the one-dimensional stationary case, the heat
propagation is calculated in equations (1) and (2) [13]. 1+i
Ωe 2πf
2 πf
1+ e−2d (1 + i)
T (x , t ) =
x
( x
T0⋅e− μ ⋅ei ωt − μ ) (1) 2+
1+i
2
Ωe 2πf α
T (0, t ) = T0⋅e−i2πft 1+i
Ωe 2πf
2 πf
1− e−2d (1 + i)
T (x , t ) = T0⋅e−σx ⋅eiωt (2) 2+
1+i
Ωe 2πf α
2 (10)

2α α Equation (9) illustrates the dependency of the reflection coefficient

μ= = not only on the effusivity but also on the contact resistance and ex-
ω πf (3)
citation frequency. The phase value is calculated in equation (11).

129
S. Ekanayake et al. Composites Part B 147 (2018) 128–134

φr = arg(T (0, t ))
The phase of the reflected wave is described as a function of fol-
lowing parameters:
φr = g (α, f , Ωe , d )
Since the frequency of the sinusoidal excitation φe is identical to the
frequency of the thermal wave at T (0, t ) , the phase shift φ is the
quantity measured by the thermography system and calculated as:
φ = φe − φr = φe − arg(T (0, t ))
3.2. Modeling of the complex wave field
Using optical lock-in thermography, the defects' depth position is
not determined directly. The depth position has to be assigned to the
measured phase value. Hence, the material dependent thermal diffu-
sivity, the influence of contact resistance on the measured phase value
and the distortion of the measured phase value due to lateral heat flow
has to be considered. The introduced method for depth determination is
based on the theoretical model of a five dimensional complex wave
field. The measurement problem depends on the following five un-
known parameters:
• Thermal diffusivity α
• Excitation frequency f
(11) measurement results. Fig. 2 shows the thermal diffusivity α for a given
excitation frequency f and depth d in dependence of the reflection
coefficient R23 and different phase values φ .
The measured values are fitted into the numerical model and the
(12) parameter combination of reflection coefficient and thermal diffusivity
with the smallest deviation is utilized for further investigations. As a
result, the specimen dependent thermal diffusivity α is determined.
3.4. Estimation of the optimal lock-in excitation frequency
(13)
To minimize the lateral heat flow, the optimized excitation fre-
quency is estimated for each defect. For reliable defect detection, the
measured phase value φ has to differ from −45°, which is the phase
value corresponding to infinite thermal thickness. Simultaneously, the
excitation frequency f has to be maximized to reduce lateral heat flows.
To fulfill these requirements, a thermal depth penetration of μ ≈ d is
endeavored. As the defect depth d is unknown initially, different ex-
citation frequencies across a value range are tested. A representative
measurement serious for a given thermal diffusivity α is shown in Fig. 3.
Starting with high excitation frequencies, defects far beneath the
thermal penetration depth d > > μ induce a phase value of −45° and
remain unrecognized. The excitation frequency is reduced stepwise
until the defect becomes visible. The optimal lock-in frequency is
reached the first time the phase value differs from −45°. As a result, the
optimized excitation frequency fopt and the corresponding phase value

• Phase values ρ φopt are estimated.

• Reflection properties: contact resistance and effusivity Ωe

• Depth position d 3.5. Determination of the contact resistance and effusivity Ωe

For the determination of the reflection properties and the defects'

The contact resistance and effusivity Ωe are summarized as re-
depth position the measurement of the optimized and the neighbouring
flection properties and considered as one single variable. The solution
lower excitation frequencies are employed. Similar to the thermal dif-
space is numerically calculated for every possible variable combination
fusivity determination, the data from the numerical model of the
using equation (11) and is deposited as a database. Table 1 exhibits the
complex wave field are recalled as look-up table for the selected mea-
value range and steps of the different input parameters. For the defects'
surement parameters. The numerical model of the complex wave field is
depth determination in CFRP-structures the five unknown parameters
limited to the measured thermal diffusivity α and the selected mea-
are estimated step by step considering the theoretical model until at last
surement frequencies f . The look-up table is illustrated in Table 2.
the defects' depth position is determined.
The look-up table contains all theoretical corresponding phase va-
lues, parameter combinations of reflection properties Ωe and depth
3.3. Measurement of thermal diffusivity α
position d. The experimentally measured values are fitted into each
parameter combination of reflection properties Ωe and depth position
The thermal diffusivity is material dependent and varies with the
d. The combination with the smallest deviation between measured and
composition of the CFRP part. Therefore, the thermal diffusivity is ex-
numerical value is chosen. As a result, the contact resistance and the
perimentally determined. Based on Spiessberger's approach, the
effusivity Ωe are determined.
thermal diffusivity is measured at a calibrated position with given
depth [9]. Experimental investigations are conducted at defined ex-
3.6. Determination of the defects' depth position d
citation frequencies and the corresponding phase values are measured.
From the initial five unknown parameters, only the reflection coeffi-
Once the parameters thermal diffusivity α , optimal excitation fre-
cient and the thermal diffusivity remain unknown. For the determina-
quency fopt , contact resistance and effusivity Ωe are determined, the
tion of the two remaining unknown parameters, at least two measure-
phase value can be assigned to the defects' depth position. Fig. 4 shows
ments are required. A look-up table corresponding to the measured
that the measured phase value φopt can be allocated to two different
parameters is extracted using the numerical database of the complex
depths.
wave field. Each parameter configuration of thermal diffusivity and
As the optimized excitation frequency fopt is estimated proceeding
reflection coefficient for the measured excitation frequencies and phase
lock-in thermography measurement from high to low excitation fre-
values of the theoretical model are compared to the actual
quencies, the defects' depth position cannot be right beneath the sur-
face. The second depth value represents the defects' depth position d.
Table 1
Dimensions for the calculation of the wave field.
4. Experimental investigation and results
Depth Contact resistance Excitation Thermal
position d and effusivity Ωe frequency diffusivity α
[mm] [s1/2] [m2/s]
4.1. Experimental set-up
f [Hz]

Range [0, 4] [0,50] [0.005, 0.5] [1·10−7, For the evaluation of the developed measurement process with re-
9·10−7] gard to the minimization of lateral heat flow and consideration of
Values 500 150 7 100
contact resistance, blind bore hole samples and water jet cut samples
Resolution 0.008 0.33 – 8·10−9
with artificial cavities are designed. The samples are calibrated with a
ZEISS MICURA coordinate measuring machine (CMM).

130
S. Ekanayake et al. Composites Part B 147 (2018) 128–134

Fig. 2. Wave field dependent on reflection coefficient R23 , thermal diffusivity α and different phase values φ .

Fig. 4. Phase value in dependence of depth d with given thermal diffusivity α,

optimal excitation frequency fopt and contact resistance and effusivity Ωe .
Fig. 3. Theoretical normalized phase sequence with marked phase values for
different excitation frequencies.

Table 2
Look-up table for the determination of the contact resistance and effusivity. Ωe

Row index i z Depth Factor Ωe Phase value for f1 … Phase value for fm

1 d1 Ωe1 φ (d1, Ωe1, f1 ) … φ (d1, Ωe1, fm )

… … Ωe1 … … …
j dj Ωe1 φ (dj , Ωe1, f1 ) … φ (dj , Ωe1, fm )
j+1 d1 Ωe2 φ (d1, Ωe2, f1 ) … φ (d1, Ωe2, fm )
… … Ωe2 … … …
2j dj Ωe2 φ (dj , Ωe2, f1 ) … φ (dj , Ωe2, fm )
… … … … … …
(k − 1) j d1 Ωek φ (d1, Ωek , f1 ) … φ (d1, Ωek , fm )
… … Ωek … … …
kj dj Ωek φ (dj , Ωek , f1 ) … φ (dj , Ωek , fm )

The CFRP borehole samples are designed using 8 layers of BIAX HPT
300 C45 with estimated single layer thickness of 0.5 mm. Blind bore
holes with varying diameter (4 mm, 8 mm, 12 mm, 15 mm, 20 mm) and
increasing remaining wall thicknesses (0.5 mm, 1 mm, 1.5 mm, 2 mm,
2.5 mm and 3 mm) are milled (Fig. 5). For the determination of the
thermal diffusivity, a wedge is included on the right hand side of the
sample. The remaining wall thicknesses and blind bore hole diameters
as well as the wedge are calibrated.
The five water jet cut CFRP-samples consist of woven fabric with the
dimensions 48 mm × 48 mm. The samples are water jet cut laterally
Fig. 5. Diagram of bore hole sample for depth determination with lateral heat
flow (all size indications in mm).
into two parts. The samples are cut with average top plate thicknesses
of d= [0.8, 1.0, 1.1, 2.8, 3.2] mm. Standard deviation of the top plate
thickness d over the 25 measured positions is [0.07, 0.06, 0.08, 0.03,
0.03] mm respectively. The resulting cut surfaces display a rough and
wavy texture. The thicknesses of the two parts are calibrated at 25
defined positions and reassembled using adhesive tape (Fig. 6).
The resulting cavity inside the sample has microscopic contact

131
S. Ekanayake et al.
Fig. 6. Diagram of water jet cut sample for depth determination with contact
resistance.
points between the surfaces with contact resistance that simulates de-
lamination at calibrated depth position. In contrast to impact damaged
CFRP-parts, lateral heat flow is minimized as the samples do not possess
erratic differences in the depth profile.
4.2. Data acquisition
For data acquisition the Edevis OTVis 5000 lock-in thermography
system consisting of two 1.25 kW halogen light sources and a FLIR
SC5650 infrared (IR) camera is used. For the experimental investiga-
tion, for each sample type (blind bore hole plate and water jet cut
sample) 30 replicate measurements at excitation frequencies of
f = [0.5, 0.2, 0.1, 0.05, 0.02, 0.01, 0.005] Hz are performed.
Additionally, for linearity investigation, series of 10 measurements each
were conducted on the four additional water jet cut samples.
4.3. Data analysis of measurements with lateral heat flow
Measurement data is analyzed for the case of lateral heat flow. For
each bore hole sample, the thermal diffusivity α is determined at the
wedge. The specimen is scanned from high to low excitation fre-
quencies using the lock-in thermography system and the developed
algorithm is applied for the remaining wall thickness determination of
each blind bore hole. The repeatability uEVR is investigated in 30 mea-
surements and is equal to the standard deviation. The results of the
thermography measurements are compared to the previously per-
formed CMM measurements. The systematic deviation of the thermo-
graphy and CMM measurement is calculated as bias uBi . The equation
(14) is illustrated below.
1 dcal ≤ 1.5 mm and bore hole diameter D≥ 12 mm. It rises to
uBi = dcal − d
3 (14)
Fig. 7. Repeatability uEVR (left) and Bias uBi (right) of bore hole sample repeat measurement.
132
Composites Part B 147 (2018) 128–134
where d [m] is the mean of measured wall thicknesses and dcal [m] is
the calibrated wall thickness.
The results of the experimental investigations for each blind hole on
the test samples are shown in Fig. 7.
In a second step, the influence of lateral heat flow by decreasing
blind bore hole diameter and increasing remaining wall thicknesses is
investigated. The linearity in dependency of both parameters is dis-
played in Fig. 8.
While the remaining wall thickness can be determined for large
blind bore diameter (20 mm) the accuracy for smaller diameters de-
creases significantly with increasing depth (Fig. 8 left). Fig. 8 right il-
lustrates the significant impact of the remaining wall thickness in
comparison to the blind bore hole diameter. For artefacts close to the
surface, depth position can be reliably determined even for small dia-
meters.
4.4. Data analysis of measurements with contact resistance
Measurement data is analyzed for the case of contact resistance. For
each water jet cut sample, the thermal diffusivity α is determined at the
top plate before reassembly. The reassembled samples' top plate
thickness is then measured at the defined and calibrated positions. The
repeatability uEVR and bias uBi are calculated and the results are shown
in Fig. 9.
The influence of the top plate thicknesses and reflection properties
Ωe are investigated. Fig. 10 shows that the top plate thickness and the
reflection properties Ωe do not have significant impact on the mean
deviation between measured and calibrated values. As the reflection
properties Ωe correlates with the gap size, for the proceeded mea-
surements no influence of the gap size can be perceived.
5. Discussion
Subject of this work is the development of a defect detection process
for CFRP repair. Delamination due to impact usually occurs close to the
surface (≤1 mm) [16]. Delaminated area generally displays linear
correlation with impact energy [17]. High energy impacts with sub-
sequent critical mitigation of structural strength thus generate large
delaminated areas (in range of cm2). Consequently, the developed al-
gorithm's accuracy has to be evaluated for greater bore hole diameters.
CFRP repair for delamination is commonly conducted by means of resin
injection between the delaminated layers [2,3]. With typical layer
thickness of around 0.5 mm, the developed algorithm needs to ensure
delamination depth measurement with accuracy within this range.
The repeat measurements of bore hole plate in Fig. 7 show a re-
peatability of uEVR < 0.06 mm in a range of wall thickness
uEVR < 0.6 mm for a range of dcal ≤ 2.5 mm, D≥ 8 mm and reaches
S. Ekanayake et al. Composites Part B 147 (2018) 128–134

Fig. 8. Mean deviation in dependency of varying calibrated wall thickness [with additional results of samples with contact resistance] (left) and bore hole diameter
(right).

Fig. 9. Box plots for repeatability (left) and bias (right) of water jet cut sample repeat measurement.

Fig. 10. Mean deviation in dependency of varying calibrated top plate thickness dcal and reflection properties Ωe with linear best fit.

maximal level of uEVR < 1 mm in range dcal = 3 mm, D= 4 mm. In
identical ranges of dcal and D , the bias uBi values attain higher un-
certainties of uBi < 0.15 mm, uBi < 1 mm and uBi < 1.33 mm.
Linearity diagrams in Fig. 8 show increasing mean deviation with
decreasing bore hole diameter D . As for the largest diameter D=
20 mm, mean deviation is constantly lower than 0.4 mm. For the
smallest diameter D= 4 mm, the magnitude of mean deviation increases
from 0.05 mm for lowest wall thickness dcal = 0.5 mm to −2.3 mm for
highest wall thickness dcal = 3 mm. For large defect diameter D , even
deep defects can be detected accurately. However, small defects can
only be measured for low depths. For deep defects, the measurements
are no longer linear over the bore hole diameter with a variance of
0.76 mm for dcal = 2.5 mm and 1.4 mm for dcal = 3 mm. Repeat
measurements of bore hole plate approve the assumption of influence of
bore hole diameter and wall thickness on the measured thicknesses'
repeatability, bias and linearity due to lateral heat flow. Nevertheless,
the developed algorithm for depth determination enables reliable
measurement for a range of d ≤ 1.5 mm in depth and D≥ 12 mm in
defect diameter. The identified limits of the algorithm's applicability
fulfil the requirements for CFRP repair as stated above.
For the water jet cut samples with contact resistance the lateral heat
flow through erratic differences in the depth profile is negligible. The
examination of the water jet cut samples focuses on the impact of the
contact resistance on the algorithm's performance. Fig. 9 shows lower

133
S. Ekanayake et al.
uncertainties of repeatability of uEVR < 0.04 mm of the water jet cut
samples in comparison to the bore hole sample. The uncertainty of
systematic deviations uBi are significantly higher than the uncertainties
through repeatability, but still with a maximum of uBi < 0.14 mm
comparable to best case scenario of the bore hole sample. The non-
infinitesimal gap size between the plates water jet cut samples could be
a possible reason for the higher mean and variance of the bias uBi .
Equation (8) assumes step in temperature ΔT due to contact resistance
over an infinitely thin interface. In reality the interface appears in form
of delamination between CFRP layers, which are not infinitely thin, but
have a width in sub-millimeter range [18].
Linearity diagrams in Fig. 10 show the minor influence of calibrated
top plate thickness dcal and reflection properties Ωe on the measured
mean deviation. The mean deviation increases by 0.05 mm between the
lowest and highest value of dcal and Ωe and is much lower in com-
parison to the variance of mean deviation within on measured series
with same value of dcal and Ωe . Therefore the impact of the two para-
meters dcal and Ωe is negligible concerting the measurement un-
certainty. Fig. 8 illustrates the accuracy of the thickness measurement
with contact resistance even in deeper depth positions. The maximum
deviation is less than 0.5 mm for the greatest calibrated thickness dcal
= 3.2 mm, which is only achieved for a similar depth for the bore hole
plate measurements of D= 20 mm. The experimental investigation
prove a lower level of error due to repeatability, bias and linearity of
the water jet cut samples in comparison to the bore hole samples. For
delamination due to impact damage, both lateral heat flow and contact
resistance occur as influencing factors on accuracy.
6. Conclusions
A method to reveal depth information of defects in CFRP-structures
using lock-in thermography is proposed. The lateral heat flow and the
contact resistance are identified as significant influencing factors on the
accuracy of depth measurement with thermography on delamination
due to impact damage. Therefore, the correlations of lateral heat flow
as a consequence of defect size and depth position as well as the in-
fluence of contact resistance are separately investigated. The experi-
mental results of the blind bore hole samples simulating lateral heat
flow show the dependency between defect properties and measurement
accuracy. The investigation illustrates that the developed method en-
ables an accurate depth determination over a wide range of thicknesses
and blind bore hole diameters with deviations of less than 0.5 mm.
Using water jet cut samples thermal contact resistance is simulated.
Experimental results show the capability of defect depth measurements
with contact resistance over a wide range of depth. The influence of
contact resistance on repeatability, bias and linearity is verified to be
lower than the impact of lateral heat flow.
The developed method essentially contributes to the qualification of
lock-in thermography in CFRP non-destructive testing. Since the precise
134
Composites Part B 147 (2018) 128–134
determination of the contour of impact damages is a great challenge, as
the defects' edges do not have a discrete but a continuous phase value
course, further investigations will focus on the quantitative determi-
nation of the defect area and position.
Acknowledgements
This work is supported by the Deutsche Forschungsgemeinschaft
DFG within the scope of the German-Brazilian Research Initiative
BRAGECRIM (reference SCHM1856/59-1).
Appendix A. Supplementary data
Supplementary data related to this article can be found at http://dx.
doi.org/10.1016/j.compositesb.2018.04.032.
References
[1] Kumar P, Rai B. Delaminations of barely visible impact damage in CFRP laminates.
Compos Struct 1993;23(4):313–8.
[2] Thunga M, Bauer A, Obusek K, Meilunas R, Akinc M, Kessler MR. Injection repair of
carbon fiber/bismaleimide composite panels with bisphenol E cyanate ester resin.
Compos Sci Technol 2014;100:174–81.
[3] Thunga M, Larson K, Lio W, Weerasekera T, Akinc M, Kessler MR. Low viscosity
cyanate ester resin for the injection repair of hole-edge delaminations in bismalei-
mide/carbon fiber composites. Compos Appl Sci Manuf 2013;52:31–7.
[4] Busse G. Subsurface imaging with photoacoustics. Appl Phys 1980;10:815–6.
[5] Busse G. Thermal wave imaging with phase sensitive modulated thermography. J
Appl Phys 1992;71:3962–5.
[6] Bai W, Wong SB. Evaluation of defects in composite plates under convective en-
vironments using lock-in thermography. Meas Sci Technol 2001;12:142–50.
[7] DIN EN 15042–2. Thickness measurement of coatings and characterization of sur-
faces with surface waves - Part 2: guide to the thickness measurement of coatings by
photothermic method. 2006-06.
[8] Gleiter A, Spiessberger C. Phase angle thermography for depth resolved defect
characterization. 35th annual review of progress in quantitative nondestructive
evaluation. 2008. p. 526–32. (Bd. 1096 AIP Conference Proceedings).
[9] Spiessberger C, Gleiter A, Busse G. Data fusion of lockin-thermography phase
images for innovative non-destructive evaluation. QIRT J 2009(6):149–61.
[10] Holtmann N, Artzt K, Gleiter A, Strunk HP, Busse G. Iterative improvement of
Lockin-thermography results by temporal and spatial adaption of optical excitation.
Quant InfraRed Thermogr J 2012(9):167–76.
[11] Aamodt LC. Effect of 3-D heat flow near edges in photothermal measurements. Appl
Optic 1982;21:111–5.
[12] Quimby RS. Three-dimensional heat-flow effects in photoacoustic spectroscopy of
solids. Appl Phys Lett 1979(35):43–5.
[13] Rosencwaig A, Gersho A. Theory of the photoacoustic effect with solids. J Appl Phys
1976;47(1):64–9.
[14] Bennett CA, Patty RR. Thermal wave interferometry: a potential application of the
photoacoustic effect. Appl Optic 1982(21):49–54.
[15] Carslaw HS. Conduction of heat in solids 23. Oxford Science Publications; 1946.
[16] Shen Q, Omar M, Dongri S. Ultrasonic NDE techniques for impact damage inspec-
tion on CFRP laminates. JMSR 2011;1(1).
[17] Hong S, Liu D. On the relationship between impact energy and delamination area.
Exp Mech 1989;29(2):115–20.
[18] Ehrlich I, Dinnebier H, Jost C. Comparison of impact delaminations in CFRP using
different test methods. J Achiev Mater Manuf Eng 2015;73(2):123–38.