You are on page 1of 9

Available online at www.sciencedirect.

com
Available online at www.sciencedirect.com
ScienceDirect
ScienceDirect
Available online at www.sciencedirect.com
Structural Integrity Procedia 00 (2021) 000–000
Structural Integrity Procedia 00 (2021) 000–000 www.elsevier.com/locate/procedia
ScienceDirect www.elsevier.com/locate/procedia

Procedia Structural Integrity 38 (2022) 631–639

FATIGUE DESIGN 2021, 9th Edition of the International Conference on Fatigue Design
FATIGUE DESIGN 2021, 9th Edition of the International Conference on Fatigue Design
Fatigue and Damage Assessment of CFRP Material Using Digital
Fatigue and Damage Assessment of CFRP Material Using Digital
Image Correlation
Image Correlation
Sara Eliassona,b,c,*, Lars Johan Wenner Berga,b,c, Per Wennhageb,c, Mathilda K. Hagnellb,c,
Sara Eliassona,b,c,*, Lars Johan Wenner BergBarsoum
Zuheir
a,b,c
, Per cWennhageb,c, Mathilda K. Hagnellb,c,
Zuheir Barsoumc
Scania CV AB, Södertälje, Sweden
a

The Centre for ECO


b a 2
Vehicle
Scania CV AB,Design, SE-100
Södertälje, 44 Stockholm, Sweden
Sweden
c
KTH Royal Institutebof Technology, The Department of Engineering
The Centre for ECO2 Vehicle Design, SE-100 44Mechanics,
Stockholm,SE-100
Sweden44 Stockholm, Sweden
c
KTH Royal Institute of Technology, The Department of Engineering Mechanics, SE-100 44 Stockholm, Sweden

Abstract
Abstract
Fatigue testing of a Carbon Fiber Reinforced Polymer (CFRP) in tension-tension loading has been conducted. In-situ surface
strain
Fatiguemeasurements were performed
testing of a Carbon to examine
Fiber Reinforced the gradual
Polymer (CFRP)elongation of the specimen
in tension-tension as this
loading has beenrelates to stiffness
conducted. In-situloss and
surface
fatigue damage. A methodology
strain measurements were performed capturing the specimen
to examine the gradualat peak load has
elongation beenspecimen
of the developed, including
as this relates an
to automated
stiffness loss trigger
and
mechanism
fatigue that activates
damage. the camera
A methodology at the desired
capturing cycle count.
the specimen The material
at peak load hastested
beenwas a Unidirectional
developed, including (UD)
an Non-Crimp Fabric
automated trigger
(NCF) with that
mechanism carbon fibers the
activates andcamera
an epoxy matrix.
at the Thecycle
desired fatigue test The
count. results revealed
material a wide
tested was scatter in the mid-range
a Unidirectional of the highFabric
(UD) Non-Crimp cycle
fatigue with
(NCF) region. By studying
carbon fibers and theanstrain
epoxy in matrix.
the earlyThe
fatigue
fatigueloading cyclesrevealed
test results and the stiffness loss over
a wide scatter time,
in the benchmark
mid-range of the
of the highfatigue
cycle
performance
fatigue region.between different
By studying material
the strain samples
in the could be
early fatigue carried
loading out, and
cycles explaining the scatter
the stiffness in the
loss over fatigue
time, testing.ofItthecould
benchmark be
fatigue
observed thatbetween
performance the fatigue limit of
different the UD
material CFRP could
samples material be in the fiber
carried out, direction
explainingis the
in the magnitude
scatter of 80 %
in the fatigue of theItmaterial’s
testing. could be
Ultimate
observed Tensile
that theStrength
fatigue (UTS).
limit of the UD CFRP material in the fiber direction is in the magnitude of 80 % of the material’s
© 2021 The
Ultimate Authors.
Tensile Published
Strength (UTS).by ELSEVIER B.V.
© 2021
This Theopen
is an Authors. Published
access article by Elsevier
under B.V.BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
the CC
© 2021
This is anThe
openAuthors. Published
access article under by
the ELSEVIER
CC BY-NC-ND B.V.
license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review
This
Peer-review under
is an open responsibility
access
under article under
responsibility of scientific
of the the scientific
CC BY-NC-NDcommittee
committee of the
of license Fatigue Design 2021 Organizers
(https://creativecommons.org/licenses/by-nc-nd/4.0)
the Fatigue Design 2021 Organizers
Keywords:
Peer-review CFRP;
underFatigue; Stiffness degradation;
responsibility DIC committee of the Fatigue Design 2021 Organizers
of the scientific
Keywords: CFRP; Fatigue; Stiffness degradation; DIC

* Corresponding author. Tel.: +46-8-553-54037.


* Corresponding saraeli@kth.se
E-mail address:author. Tel.: +46-8-553-54037.
E-mail address: saraeli@kth.se
2452-3216 © 2021 The Authors. Published by ELSEVIER B.V.
This is an open
2452-3216 access
© 2021 Thearticle under
Authors. the CC BY-NC-ND
Published by ELSEVIER license
B.V.(https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under
This is an open responsibility
access of the
article under the scientific
CC BY-NC-NDcommittee of the
license Fatigue Design 2021 Organizers
(https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the Fatigue Design 2021 Organizers

2452-3216 © 2021 The Authors. Published by Elsevier B.V.


This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0)
Peer-review under responsibility of the scientific committee of the Fatigue Design 2021 Organizers
10.1016/j.prostr.2022.04.065
632 Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639
2 Author name / Structural Integrity Procedia 00 (2021) 000–000

1. Introduction

Advanced lightweight materials, e.g. CFRP, are getting more attention in the lightweight design of structural
commercial vehicle components. Commercial vehicles have a wide range of service loads and the road induced
vibrations often sets the design criteria. To avoid redundancy and conservative structural design it is important to
understand the fatigue behavior of the materials when conducting fatigue life estimation calculations.
Damage during the fatigue life of a composite laminate can unfold differently depending on layup and material
combination. Reifsnider et al. (1983) declared the different damage modes throughout the fatigue life of a composite
laminate, illustrated in Fig. 1a. Gamstedt et al. (1999) studied the damage mechanisms of UD CFRP and observed
two types of behaviors: (1) propagation of small localized fiber-bridge cracks or (2) a more extensive distributed
damage with fiber breakage induced by longitudinal debonding or matrix cracking.
A way to monitor the damage development during fatigue loading is by monitoring the stiffness reduction
(Schulte, 1999). Senthilnathan et al. (2017) found that there is a steep reduction in stiffness in the beginning of a
fatigue test and then the microstructural damage increase monotonically with the number of cycles. A typical
stiffness degradation behavior can be described with three stages (Fig. 1b). The initial stage has a rapid stiffness
reduction, followed by a second intermediate stage with an approximately linear behavior. The second stage is
mainly driven by edge delamination and longitudinal cracks along the fibers. The final and third stage of the stiffness
reduction occurs in an abrupt manner ending in final failure (Van Paepegem, 2010).

Fig. 1. (a) The characteristic damage state during the fatigue life of a composite laminate (Reifsnider, et al., 1983); (b) typical stiffness
degradation behaviour during the fatigue life of a composite laminate (Van Paepegem, 2010).

The testing of the UD CFRP material is still a challenge due to the high specific stiffness and strength, and the
anisotropy of these materials. One of the challenges is for the failure to occur in the gauge length and knowing the
fatigue life is not underestimated due to unwanted artefacts such as stress concentrations and manufacturing induced
defects. Eliasson et al. (2019) developed a fatigue testing procedure to succeed with testing of UD CFRP. This
testing procedure is utilized in this study. Manufacturing induced defects highly affects the fatigue behavior and
strength of a composite laminate (Sisodia, et al., 2015) and reducing the prevalence and variability of manufacturing
induced defects is difficult and costly, but required if the spread in fatigue performance is to be reduced.
The study presents a tension-tension fatigue test series of a UD CFRP material tested in the fiber direction at
different load levels. In this study the surface displacements of the specimen was monitored with an image
processing method such that strains and stiffness could be calculated and monitored throughout the life cycle
(Sisodia, et al., 2011). An automated testing methodology to image the surface displacement captured at peak load
was developed. Analyzing the fatigue damage by studying the strain and comparing different test specimens, trends
were identified to explain the scatter in fatigue results and highlight challenges when conducting the test series.
Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639 633
Author name / Structural Integrity Procedia 00 (2021) 000–000 3

2. Material

The material used in the study was a stitch-bonded UD NCF from Zoltek with a fast-curing epoxy resin from
DOW. The material was received in 300x650 mm plates, with a thickness of approximately 2 mm. The plates were
manufactured with Resin Transfer Molding (RTM) using 8 layers of the fiber fabric, all stacked with the fibers in
the same direction. The UTS of the material in the fiber direction is 1295 MPa and was determined with static
tension testing according to ASTM D3039/D3039M-17 (ASTM International, 2017).

3. Fatigue Testing

Tension-tension fatigue testing was conducted according to ASTM D3479/D3479M-12 (ASTM International,
2012). The fatigue test rig used was a Schenk Hyrdopuls with a 100 kN load cell. The grips on the test rig were
hydraulic with the ability to adjust the clamping pressure. The clamping pressure was set to, 30-45 kN (Eliasson, et
al., 2019). The testing frequency was set to 5 Hz and a load ratio, 𝑅𝑅, of 0.1 was used for all tests. Two million cycles
were set as the runout limit for the specimens; however, some tests were run further. The fatigue testing was
performed at a load level of 70 – 96 % of the UTS. The specimens were manufactured according to ASTM
D3039/D3039M-17 (ASTM International, 2017) and the specimen geometry was chosen according to Eliasson et al.
(2019) (Fig. 2).

Fig. 2. The specimen geometry used in the fatigue testing (Lspec = 250 mm; wspec = 15 mm; Ltab = 50 mm; wtab = 20 mm).

3.1. Fatigue Test Setup

The fatigue testing methodology, measuring in-situ full-field displacements of the fatigue specimens was
automated such that no operator input was required after initialization. The measurement setup was disconnected
from the fatigue rig and could be connected to any testing system. Surface displacements, at peak load for a given
cycle of the ongoing fatigue test, were captured in-situ with a High-Speed Camera (HSC) and the images were
analyzed with Digital Image Correlation (DIC). Following sections outline the test setup, and for a more thorough
description see the thesis of Wenner Berg (2020).

3.2. Automated In-Situ Full-Field Displacement Measurements

DIC is an image processing technique that tracks surface displacement. It generates full-field measurements
without contact to the specimen. 2D DIC was utilized in this study to track the development of fatigue induced
displacement and strain. A speckle pattern in black and white was applied to the specimen to get high contrast
images (LePage, et al., 2017). The DIC measurements were compared with a second measurement from an
extensometer, to verify the correctness of the data.
The HSC was used to capture the peak loading of a cycle without pausing or slowing down the fatigue test. Each
captured peak was compared to follow the stiffness degradation, i.e. the fatigue damage growth. The HSC was
configured with the software MotionProX Studio and was triggered by closing a circuit with a transistor switch. The
sampling rate was 1000 frames per second and for every trigger event, 150 images were captured. Analyzing the
data, the peak loads were compared to an image with zero load taken before the test was started. The camera settings
are found in Table 1.
634 Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639
4 Author name / Structural Integrity Procedia 00 (2021) 000–000

The transistor switch triggering the HSC was operated by a microcontroller and Fig. 3 illustrates the process. The
trigger condition was controlled by a cycle count that was dependent on a threshold value (Fig. 4). The
microcontroller received an analog output signal from the load cell of the fatigue test rig and used this signal to
determine if the load was above the set threshold value. This allowed for counting the number of load cycles.

Fig. 3. Trigger mechanism flow chart.

Table 1. High Speed Camera settings.

Component Settings
Camera Dantec Dynamics Nano Sense MKIII
Lens Nikon AF-S 60/2-8 ED
Capture Distance 500-610mm
Trigger Settings Circular, 115-20/35-40 pre/post trigger
Ex. of Trigger 0 – 1000: every 200 cycles
Frequency 1000 – 10000: every 500 cycles
(cycle interval: 10000 – 50000: every 1000 cycles
trig cycle) 50000 – 250000: every 5000 cycles Fig. 4. Trigger description during cyclic loading.
250000 – End: every 10000 cycles

3.2.1. Data Handling


Sampling 150 images for each trigger event in a two million cycles fatigue test resulted in large sets of data that
could not be processed manually. There were two steps to extract the DIC data. First the peak images from the HSC
were extracted using a 2D DIC Python code (André, 2017). The next step of the analysis, calculating the strains for
the peak images, was conducted using the open source 2D DIC Matlab software, Ncorr (Blaber, et al., 2015). From
the Ncorr data the average strain over the surface of the specimen was calculated. In Fig. 5 the axis system for an
HSC image and a typical result image for the strain in y-direction is presented.
When analyzing the images with Ncorr the Region of Interest (ROI) must be chosen carefully. Effects of splinters
on the edges during the test can radically change the strain field if this is included in the analysis. Therefore, the ROI
was made slightly narrower than the full width of the specimen. The speckle pattern was also evaluated for damage
Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639 635
Author name / Structural Integrity Procedia 00 (2021) 000–000 5

before processing the images. However, for these uniform and unidirectional specimens the speckle pattern adhered
well over the course of the fatigue test.
Not all test data had sufficient quality, therefore, not all tested specimens have DIC data.

Fig. 5. An HSC image with coordinate system, and a result image of the strain in y-direction from the DIC analysis in Ncorr.

4. Results

4.1. S-N Data

Fatigue testing was successfully conducted for 18 test specimens (Table 2). There was a high spread in the
results, and the results were summarized in an S-N plot which is presented in Fig. 6. The results showed a high
spread for both load levels 90 % and 80 % of UTS. Specimens below 80 % of UTS were runouts.

Table 2. Tabular results for fatigue test specimens, with failure codes according to ASTM D3039/D3039M-17 (ASTM International, 2017).

Specimen Load [% of UTS] Cycles Failure Mode Specimen Load [% of UTS] Cycles Failure Mode
AT0_1 70 > 2 000 000 Runout AT0_10 80 78 370 LGM
AT0_2 78 > 2 000 000 Runout AT0_11 85 390 GAT
AT0_3 80 785 750 GIT AT0_12 85 370 LAV
AT0_4 80 500 370 SGM AT0_13 85 397 040 LAT
AT0_5 80 412 530 LIT AT0_14 85 1 620 LIT
AT0_6 80 > 5 000 000 Runout AT0_15 90 850 GAT
AT0_7 80 1 386 450 LIB AT0_16 90 2 113 880 LAV
AT0_8 80 > 3 330 000 Runout AT0_17 90 150 LIT
AT0_9 80 7 880 LIB AT0_18 96 170 LAT
636 Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639
6 Author name / Structural Integrity Procedia 00 (2021) 000–000

Fig. 6. S-N plot for the fatigue test series.

4.2. Failure Modes

The failure mode for each test was evaluated. A selection of different failure modes encountered from the fatigue
testing are presented in Fig. 7 (edge splitting, longitudinal failure, inside tab failure, and failure in the gauge length).
A common failure mode was number three in Fig. 7, edge failure inside tab. This failure mode will generally
underestimate the fatigue life, since this failure mode occurs inside the tab and the specimen could have been
affected by other factors, e.g. the clamping pressure, stress concentrations or adhesive failure.

Fig. 7. Different failure modes from fatigue testing at 80 % of UTS, (1) L/S I T/B; (2) SGM; (3) LI T/B; (4) LGM (ASTM International, 2017).

4.3. Load Level 80 % of UTS

For further understanding of the damage additional specimens were run at 80 % of UTS since this load level had
a high scatter, but at the same time indicated to be the fatigue limit. The fatigue specimens from the load level 80 %
of UTS failed earliest at 7 880 cycles and at the same time could endure more than five million cycles (Fig. 6).
The stiffness degradation of four specimens is presented in Fig. 8. The three specimens AT0_4, AT0_6, and
AT0_7 had a similar behaviour where the stiffness degradation curve flattens out after 2 – 6 % stiffness loss. Whilst
the fourth specimen, AT0_10, had a much higher stiffness degradation with up to 15 %. This specimen also had a
large jump in the stiffness degradation and strain at 20 000 cycles. Looking at the strain development over time (Fig.
10) and comparing this to the different failure modes, the specimen with the highest stiffness degradation was also
the specimen with the highest average strain and the shorter fatigue life. This was also the only specimen were the
failure occurred in the gauge length. The specimen AT0_6, was a runout and the average strain for this specimen
was the lowest, and it had a stable development of the stiffness degradation.
Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639 637
Author name / Structural Integrity Procedia 00 (2021) 000–000 7

Fig. 8. Stiffness degradation of specimens for fatigue load level of Fig. 9. A piecewise linear fit to the stiffness degradation of the
80 % of UTS. specimen AT0_6, finding the breaking point between region 1 and 2.

Fig. 10. DIC analysis results for the average strain in y-direction for the specimens (a) AT0_4; (b) AT0_6; (c) AT0_7; (d) AT0_10.
638 Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639
8 Author name / Structural Integrity Procedia 00 (2021) 000–000

The behaviour of the specimens could be sorted in two distinct regions, a clear increase in strain and a region of
almost no increase in strain until final failure. The two regions are separated by a breaking point, (𝑥𝑥3 , 𝑦𝑦3 ), illustrated
in Fig. 9. The breaking point was determined with a piecewise linear function, Eq. 1, and was extracted from the
stiffness degradation data (Fig. 8) and average strain data (Fig. 10). The breaking point occurred between 14 149
cycles all the way up to 419 548 cycles (Table 3).

(1)

Table 3. Breaking points (x3, y3) for the piecewise linear fit of stiffness degradation (*, Fig. 8) and average strain in y-direction (**, Fig. 10).

Specimen x3 y3 (*) y3 (**) Cycles


AT0_3 330 321 0.9533 0.01043 785 750
AT0_4 19 499 0.9895 0.00948 500 370
AT0_5 14 149 0.98546 0.01101 412 530
AT0_6 419 548 0.94891 0.00795 > 5 000 000
AT0_7 148 967 0.94677 0.01007 1 386 450
AT0_10 31 332 0.84529 0.01066 78 370

5. Discussion

The scatter of the tension-tension fatigue tests makes it challenging to fit a Wöhler curve to the S-N data. The
scatter for different load levels is significant and at the load level of 90 % of UTS there is a specimen failing after
two million cycles and a specimen failing before 1 000 cycles. The spread of the data at this high load fatigue
regime points towards the issues of transferring the load to these anisotropic materials with high specific strength
and stiffness. The fatigue limit seems not to be far off from the 80 % of UTS load level, where we see several runout
specimens.
Looking at the stiffness degradation, the first two stages from Fig. 1b are captured, but not the final stage with the
rapid stiffness reduction before final failure. The last stage of the stiffness degradation might occur so rapidly that
the frequency of the imaging is not high enough to capture the event. The limiting factors are the amount of data
produced if the HSC is to be triggered more frequently and when the failure occurs, since there is a large scatter.
Focusing on the stiffness degradation, the piecewise linear fit (Table 3) indicates that the first region (Fig. 9) of
higher stiffness degradation is ongoing until 14 149 – 419 548 cycles. The highest value of 419 548 cycles
corresponds to the runout specimen, AT0_6. All four specimens in Fig. 8 have different failure modes. The
specimens AT0_4 and AT0_7 have lateral failures inside the tab (Fig. 10a and c), and they have an incipient
leveling out of the stiffness degradation, meaning that the fatigue life of these specimens could have been
underestimated.
Looking at the strain measurements (Fig. 10) for the three specimens, AT0_4, AT0_7, and AT0_10 the first
region could be split in two or three substages. The first substage is where there is an initial strain increase and then
one jump (or two jumps) to a continued strain increase, before leveling out in region two. These jumps could be
indications of defects in the specimens inducing considerable damage before the system reached a more stable state.
The flaws could have arrested additional damage growth or led to an early failure. These substages are evident for
the specimen AT0_10, and the strain jumps are large and resulting in a very high stiffness degradation. This high
stiffness degradation could explain the early failure. The specimen, AT0_6, does not show these substages,
Sara Eliasson et al. / Procedia Structural Integrity 38 (2022) 631–639 639
Author name / Structural Integrity Procedia 00 (2021) 000–000 9

indicating a specimen with less flaws and a more stable fatigue damage development. The size of the strain for
AT0_6 is also lower.
Utilizing these data for design with a UD CFRP material it is important to consider the variation of the results.
Both the knowledge of the stiffness degradation behavior and the fatigue limit could be used as fatigue design
guidelines.

6. Conclusions

- The fatigue limit of the UD CFRP material indicates being at least 80 % of UTS.
- Differences in the strain development over time can be connected to the high spread in the fatigue results and
the varying failure modes of the specimens.
- A strain design criterion should be set after a fatigue specimen with a lower strain that has a stable
development of strain and stiffness degradation over time, since this is the most conservative.

Acknowledgements

The authors would like to thank the Centre for ECO2 Vehicle Design, funded by the Swedish Innovation Agency
Vinnova (Grant Number 2016-05195), and Scania CV AB for financial support.

References

André, D., 2017. Pydic - a python suite for local digital image correlation. Version 1.0. Limoges, France.
ASTM International, 2012. ASTM D3479/D3479M-12 Standard Test Method for Tension-Tension Fatigue of Polymer Matrix Composite
Materials. West Conshohocken, PA 19428-2959 United States.
ASTM International, 2017. ASTM D3039/D3039M-17 Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials.
West Conshohocken, PA 19428-2959 United States.
Blaber, J., Adair, B. and Antoniou, A., 2015. Ncorr: Open-Source 2D Digital Image Correlation Matlab Software. Experimental Mechanics, 7,
55(6), pp. 1105-1122.
Eliasson, S., Wanner, S., Barsoum, Z. and Wennhage, P., 2019. Development of fatigue testing procedure for unidirectional carbon fiber
composites. Procedia Structural Integrity, Volume 19, pp. 81-89.
Gamstedt, E. K. and Talreja, R., 1999. Fatigue damage mechanisms in unidirectional carbon-fibre-reinforced plastics. Journal of Materials
Science, Volume 34, pp. 2535-2546.
LePage, W. S., Shaw, J. A. and Daly, S. H., 2017. Optimum Paint Sequence for Speckle Patterns in Digital Image Correlation. Experimental
Techniques, 10, 41(5), pp. 557-563.
Reifsnider, K. L., Schulte, K. and Duke, J. C., 1983. Long-Term Fatigue Behavior of Composite Materials, ASTM International.
Schulte, K., 1999. Cyclic mechanical loading. In: Reinforced plastics durability. Cambridge: Woodhead Publishing Limited, pp. 151-185.
Senthilnathan, K. et al., 2017. Microstructural damage dependent stiffness prediction of unidirectional CFRP composite under cyclic loading.
Composites Part A: Applied Science and Manufacturing, 9, Volume 100, pp. 118-127.
Sisodia, S., Gamstedt, E. K., Edgren, F. and Varna, J., 2015. Effects of voids on quasi-static and tension fatigue behaviour of carbon-fibre
composite laminates. Journal of Composite Materials, 49(17), pp. 2137-2148.
Sisodia, S., Kazemahvazi, S., Zenkert, D. and Edgren, F., 2011. Fatigue testing of composites with in-situ full-field strain measurement. 18th
International Conference on Composite Materials. Jeju Island, Korea.
Van Paepegem, W., 2010. Fatigue damage modelling of composite materials with the phenomenological residual stiffness approach. In: Fatigue
life prediction of composites and composite structures. Woodhead Publishing Limited, pp. 102-138.
Wenner Berg, L. J., 2020. Fatigue Testing of a Unidirectional Carbon Fiber Reinforced Polymer - Investigation of damage development using
Digital Image Correlation, Degree Project in Mechanical Engineering. Stockholm: KTH Royal Institute of Technology. TRITA-SCI-GRU;
2020:354.

You might also like