Permanent Deformation and Stiffness Degradation of Open Hole Glass/PA6 UD Thermoplastic Composite in Tension and Compression

UD glass/PA6 coupons with an open hole are subjected to tensile and compressive loading. Three layups: [0/90]5s, [+45/−45]5s and [+45/0/−45/90]3s with a shape based on ASTM D5766 were tested. Both monotonic loading as well as loading–unloading–reloading tests were executed. The strain field on the sample surface was measured with digital image correlation. This allowed identifying the distribution of the strain field during loading, permanent deformation and the evolution of the sample elastic modulus. This information is not frequently measured. Yet, it is vital for the development and validation of advanced failure models. The results indicate that the thermoplastic matrix allows large plastic deformation under tensile loading for the specimens with layup [+45/−45]5s. In addition, the specimen elastic modulus reduces by about 70%. The other layups show minor permanent deformation, while the elastic modulus reduces by up to 15%. Furthermore, the quasi-isotropic laminate shows a significant post-failure load-bearing capacity under compression loading. The results are complemented with post-mortem damage and fracture observations using optical microscopy and ultrasound inspection.


Introduction
Technical continuous fiber-reinforced polymers (FRP) where carbon or glass fibers are combined with a thermoplastic (TP) polymer matrix are strong candidates for use in the mass production of composite structures. The TP matrix can be melted, allowing traditional (metal-like) processing techniques, such as hot molding or forming, which have short cycle times. In addition to this, TP-FRP provides improved recyclability and cleaner processing technologies compared to thermoset (TS) FRP [1]. Before TP-FRP can be used in structural design, the mechanical behavior under several loading conditions must be known. This knowledge can then be used to develop and benchmark improved mechanical models.
For TS-FRP [2][3][4], detailed studies about the failure of several combinations of fibers and matrix in the presence of stress concentrations exist. In contrast, for TP-FRP, only a limited amount of studies are available. The effect of the hole drilling method was investigated by Mariatti et al. [5] for plain woven glass/ABS and Brown et al. [6] for unidirectional (UD) carbon/PEEK. In the studies above, the holes were drilled through the laminate after consolidation, cutting both fibers and matrix. With advanced fiber steering techniques, it is possible to place fibers around hole openings. With this, the fibers are not cut and run continuously through the material. This usually results in an increased notched and bearing strength, as shown in [7,8] for carbon/PA6. the plies. The QI layup is chosen to represent a common layup strategy for laminates from fiber-reinforced polymers.
Based on ASTM D5766 [15] and ASTM D6484 [16], rectangular specimens with dimensions 250 mm × 36 mm and 150 mm × 36 mm were cut using water jet cutting. A central hole with a diameter of 6 mm was carefully milled from the coupons. The specimens with 250 mm length were loaded in tension, while the 150 mm long specimens were loaded in compression. The clamping distance was 50 mm. This results in an effective gauge length of 150 mm and 50 mm for the tensile and compressive specimens, respectively. Figure 1 shows the specimen dimensions.
Based on ASTM D5766 [15] and ASTM D6484 [16], rectangular specimens with dimensions 250 mm × 36 mm and 150 mm × 36 mm were cut using water jet cutting. A central hole with a diameter of 6 mm was carefully milled from the coupons. The specimens with 250 mm length were loaded in tension, while the 150 mm long specimens were loaded in compression. The clamping distance was 50 mm. This results in an effective gauge length of 150 mm and 50 mm for the tensile and compressive specimens, respectively. Figure 1 shows the specimen dimensions. The high thickness is a deviation from the standard. This was deliberately chosen such that the short gauge length combined with the high thickness would prevent, or at least significantly delay, buckling. As such, an anti-buckling guide, which would obstruct the view on the specimen surface to allow identification of the strain field on the sample surface with DIC, is not required. Normally a high thickness is only required under compression. However, since the strength of a laminate can depend on thickness [17], the same laminate thickness was used for the tensile specimens such that both tension and compression loaded cases could be compared.
After production, the specimens were nondestructively inspected to ensure that neither water jet cutting nor hole milling had induced damage to the sample or hole edges. A summary of the specimens and load types is shown in Table 1. The specimens were tested with a 100 kN hydraulic Instron uniaxial test bench () in a climatized room at 23 °C with 50% room humidity (RH). The clamps of the machine were aligned before testing to ensure proper load introduction. Both monotonic loadings until failure and loading-unloading-reloading (LUR) tests were executed. In the LUR tests, consecutive reloadings were done at approximately 25%, 35%, 50%, 65% and 85% of the static strength. After this, the specimen was loaded until fracture. The high thickness is a deviation from the standard. This was deliberately chosen such that the short gauge length combined with the high thickness would prevent, or at least significantly delay, buckling. As such, an anti-buckling guide, which would obstruct the view on the specimen surface to allow identification of the strain field on the sample surface with DIC, is not required. Normally a high thickness is only required under compression. However, since the strength of a laminate can depend on thickness [17], the same laminate thickness was used for the tensile specimens such that both tension and compression loaded cases could be compared.
After production, the specimens were nondestructively inspected to ensure that neither water jet cutting nor hole milling had induced damage to the sample or hole edges. A summary of the specimens and load types is shown in Table 1. The specimens were tested with a 100 kN hydraulic Instron uniaxial test bench () in a climatized room at 23 • C with 50% room humidity (RH). The clamps of the machine were aligned before testing to ensure proper load introduction. Both monotonic loadings until failure and loading-unloading-reloading (LUR) tests were executed. In the LUR tests, consecutive reloadings were done at approximately 25%, 35%, 50%, 65% and 85% of the static strength. After this, the specimen was loaded until fracture. Force was measured using the load cell of the machine. The strain on the sample surface was measured using 3D DIC. For this, the specimen surface was first painted with white, water-based, acrylic airbrush paint (titanium white, Golden Artist Colors, Inc., New York, USA). This type of paint adheres well to the surface without forming a film. This ensures that the paint does not detach or peel from the specimen surface under large deformation, invalidating the measurement. A black speckle pattern was applied on top of the white surface using matte black-colored spray (Motip matt black acrylic). The 3D DIC system consisted of a custom setup with 2 × 5 megapixels (MP) Pointgrey cameras (Grasshopper 3 GS3-U3-51S5M-C, The images were analyzed using VIC-3D (version 9.1.6) [18], Correlated Solutions.
From the tests, the failure phenomena, the general stress-strain behavior, developing permanent strain and damage (indicated by sample stiffness reduction) are investigated. The failure type of the samples is derived from the images taken with the cameras and side microscopy. For the nominal stress-global strain behavior, the nominal stress, σ nom , is defined, as shown below: In Equation (1), F is the force from the load cell, w is the sample width, t is the sample thickness, and d is the hole diameter. The global sample strain is defined as the engineering strain between two points located close to the clamps, as indicated in Figure 2a. This definition indicates the global state of the specimen and allows a representative comparison between the samples. As a result, the forthcoming "stress-strain" curves of these experiments indicate specimen behavior and not material behavior.

Results and Discussion
First, the nominal stress-global strain is shown in Section 3.1. Next, in Section 3.2, the observed failure type for each configuration is identified. This is followed by an analysis of the global permanent strain and damage together with.the patterns of permanent strain, as observed by the DIC, in Section 3.3.

Nominal Stress-Global Strain
The nominal stress-global strain of the experiments is shown in Figure 3. A different color represents the result for each layup. For the results of samples with the same layup, different dashing is used. The specimen elastic modulus is defined as the slope between consecutive maximum and minimum loads in the unloading-reloading cycle. This is illustrated by the straight black lines in Figure 2b. In the figure, one can also see the development of permanent strain after each loading cycle. After mechanical testing, the specimen sides were polished to identify the fracture mechanisms occurring on the side. Additionally, several specimens were inspected through transmission ultrasound to identify the extent of internal delaminations across the gauge length. The ultrasonic C scans were recorded with an automated scanner system on a rectangular grid. The scan axis measured points with steps of 0.1 mm, while the index axis was moved in steps of 0.5 mm. A through-transmission setup was used with two shockwave transducers with 5 MHz nominal frequency (GE Measurement & Control H5, Rotterdam, The Netherlands) spaced 130 mm apart. For the waveform generation, a UTPR-CC-50 ultrasonic pulser/receiver (Tecscan, Quebec, Canada) with negative spike pulse excitation was used, while data acquisition was handled by a NI-PXIe-5172 card (National Instruments Corporation, Austin, TX, USA). Signals were recorded with a sampling frequency of 125 MS/s. Instrument control was done in LabVIEW (version 2020) [19]. The lateral resolution of the ultrasound is 3.33 mm.

Results and Discussion
First, the nominal stress-global strain is shown in Section 3.1. Next, in Section 3.2, the observed failure type for each configuration is identified. This is followed by an analysis of the global permanent strain and damage together with.the patterns of permanent strain, as observed by the DIC, in Section 3.3.

Nominal Stress-Global Strain
The nominal stress-global strain of the experiments is shown in Figure 3. A different color represents the result for each layup. For the results of samples with the same layup, different dashing is used.

Results and Discussion
First, the nominal stress-global strain is shown in Section 3.1. Next, in Section 3.2, the observed failure type for each configuration is identified. This is followed by an analysis of the global permanent strain and damage together with.the patterns of permanent strain, as observed by the DIC, in Section 3.3.

Nominal Stress-Global Strain
The nominal stress-global strain of the experiments is shown in Figure 3. A different color represents the result for each layup. For the results of samples with the same layup, different dashing is used. ]5s initially shows linear behavior. This, however, turns nonlinear at 1% strain. A failure strain of about 12-15% is observed with a failure load of about 150 MPa. These failure strains are significantly higher than for the other specimens because the +45/−45 plies are primarily loaded in shear. This loading type is matrix-dominated and illustrates the large deformation capability of the PA6 matrix. Remarkable is that the OHT strength of the QI laminate is higher than the OHT strength of the coupon with layup [0/90]5s. This is caused by the high OHT strength of the +45/−45 plies, which contribute significantly to the strength of the QI laminate. ] 5s initially shows linear behavior. This, however, turns nonlinear at 1% strain. A failure strain of about 12-15% is observed with a failure load of about 150 MPa. These failure strains are significantly higher than for the other specimens because the +45/−45 plies are primarily loaded in shear. This loading type is matrixdominated and illustrates the large deformation capability of the PA6 matrix. Remarkable is that the OHT strength of the QI laminate is higher than the OHT strength of the coupon with layup [0/90] 5s . This is caused by the high OHT strength of the +45/−45 plies, which contribute significantly to the strength of the QI laminate.
The curves for OHC are shown in Figure 3b.  5s does not indicate specimen failure. The specimens were removed after, respectively, kinking and buckling to protect the load cell of the machine and preserve the crosshead alignment. Yet, the specimens could still have been compressed more. This was illustrated with the last QI specimen. The result is shown in Figure 4a. After kinking failure, the load-carrying capacity of the material steadily reduces, while the global strain reaches more than 15% global compressive strain. Figure 4b shows the final state of the specimen in the test machine before stopping the test. Both specimen halves were still intact with significant plastic deformation and failure. black diamonds. Recall that the end of the curves for [+45/0/−45/90]3s and [+45/-45]5s does not indicate specimen failure. The specimens were removed after, respectively, kinking and buckling to protect the load cell of the machine and preserve the crosshead alignment. Yet, the specimens could still have been compressed more. This was illustrated with the last QI specimen. The result is shown in Figure 4a. After kinking failure, the load-carrying capacity of the material steadily reduces, while the global strain reaches more than 15% global compressive strain. Figure 4b shows the final state of the specimen in the test machine before stopping the test. Both specimen halves were still intact with significant plastic deformation and failure.   For the samples with layup [0/90]5s, both tensile and compressive loading showed a quasilinear stress-strain behavior with brittle material failure. Figures 5 and 6 show typical failure under, respectively, tensile and compressive load. The videos of the DIC images during loading were analyzed. It is observed that OHT failure of the samples with layup [0/90]5s shows no indication of growing cracks just before sudden failure. In an instant, symmetrical cracks appear from the edge of the holes to quickly span the entire width of the sample. Remarkable for this failure is that several fibers still bridged over the crack at the sides of the specimen. This connection was quite weak yet indicates that next to fiber breakage, fiber pull-out is a damaging phenomenon for this composite material. Microscopic observation of the side of the samples, Figure 5b, shows this. It can be seen that the "pulled-out" fibers (cyan arrows in the figure) continue as ply delaminations as far as 3 mm away from the crack face (red arrow in the figure). Matrix cracking is observed in the [90] degree plies. This is observed at a distance of up to 10 mm from the crack face. The green arrows on the figure indicate the ply cracks observed furthest away from the hole. High-resolution images can be consulted in the online version of the article. Ultrasound inspection, Figure 5c   For the samples with layup [0/90] 5s, both tensile and compressive loading showed a quasilinear stress-strain behavior with brittle material failure. Figures 5 and 6 show typical failure under, respectively, tensile and compressive load. The videos of the DIC images during loading were analyzed. It is observed that OHT failure of the samples with layup [0/90] 5s shows no indication of growing cracks just before sudden failure. In an instant, symmetrical cracks appear from the edge of the holes to quickly span the entire width of the sample. Remarkable for this failure is that several fibers still bridged over the crack at the sides of the specimen. This connection was quite weak yet indicates that next to fiber breakage, fiber pull-out is a damaging phenomenon for this composite material. Microscopic observation of the side of the samples, Figure 5b, shows this. It can be seen that the "pulled-out" fibers (cyan arrows in the figure) continue as ply delaminations as far as 3 mm away from the crack face (red arrow in the figure). Matrix cracking is observed in the [90] degree plies. This is observed at a distance of up to 10 mm from the crack face. The green arrows on the figure indicate the ply cracks observed furthest away from the hole. High-resolution images can be consulted in the online version of the article. Ultrasound inspection, Figure 5c, shows that a limited amount of delamination develops in the hole's vicinity. The red lines on the figure indicate the edges of the specimen. At the edges of the sample, delaminations are present up to 2 cm away from the crack face.

Failure Observations
to be the largest, one could expect this to cause a final fracture. Surprisingly, the crack causing load-bearing failure occurs from a different location. The main failure phenomenon is ply delamination. This is shown on the microscopic images. In Figure 7c the separation between the [−45] and [+45] plies, that each remained attached to one-half of the specimen. A limited amount of matrix cracks is observed. The extent of internal delaminations for this layup is shown in Figure 7d. It can be seen that the ply separations do not extend further than the center area. to be the largest, one could expect this to cause a final fracture. Surprisingly, the crack causing load-bearing failure occurs from a different location. The main failure phenomenon is ply delamination. This is shown on the microscopic images. In Figure 7c the separation between the [−45] and [+45] plies, that each remained attached to one-half of the specimen. A limited amount of matrix cracks is observed. The extent of internal delaminations for this layup is shown in Figure 7d. It can be seen that the ply separations do not extend further than the center area.
(a) (b) (c)  can be observed originating from the hole edge. This is indicated by arrows in Figure 6a. Next to that, microscopic observation shows that few matrix cracks developed. Yet, large delaminations occur where plies were pushed apart after kinking failure, Figure 6c. The extent of the internal delamination is illustrated in Figure 6d with ultrasound inspection.
The samples with layup [+45/−45] 5s showed large nonlinear behavior before failure for both tensile and compressive loading. Figure 7 shows typical observations under tensile loading, while Figure 8 shows typical observations under compressive loading. OHT failure for the samples with layup [+45/−45] 5s , Figure 7, shows a combination of crack growth and ductile failure. Before the failure, multiple cracks develop at the hole edge, as indicated in Figure 7a. However, these initial cracks do not necessarily grow to cause final failure. In Figure 7a,b, the same crack is indicated with a red circle. Since this crack appears to be the largest, one could expect this to cause a final fracture. Surprisingly, the crack causing load-bearing failure occurs from a different location. The main failure phenomenon is ply delamination. This is shown on the microscopic images. In Figure 7c the separation between the [−45] and [+45] plies, that each remained attached to one-half of the specimen. A limited amount of matrix cracks is observed. The extent of internal delaminations for this layup is shown in Figure 7d. It can be seen that the ply separations do not extend further than the center area. unloaded and removed after the occurrence of buckling. Remarkably, side microscopy shows, apart from a permanent bend of the specimen, no signs of internal damage, Figure  8b. This is confirmed by ultrasound, Figure 8c, which only shows a limited amount of delamination around the hole edge. The absence of matrix cracking or delamination after such severe out-of-plane deformation is attributed to the ductility of the PA6 matrix combined with a strong fiber-matrix bond.   Figure 8a shows the sample deformation, which could be observed. To protect the test machine, and especially the clamp alignment, it was not attempted to fully break the samples. Therefore, the samples were unloaded and removed after the occurrence of buckling. Remarkably, side microscopy shows, apart from a permanent bend of the specimen, no signs of internal damage, Figure 8b. This is confirmed by ultrasound, Figure 8c, which only shows a limited amount of delamination around the hole edge. The absence of matrix cracking or delamination after such severe out-of-plane deformation is attributed to the ductility of the PA6 matrix combined with a strong fiber-matrix bond. Figures 9 and 10 show the tensile and compressive failure for specimens with layup [+45/0/−45/90] 3s . For the tensile case, Figure 9a,b, cracks are seen to develop along the +45 fiber direction of the top ply. These grow from the hole edge towards the edges of the sample until final failure. The specimen fails in a brittle way due to the presence of the 0 degree plies. Side microscopy, Figure 9c Figure 10. Buckling was not observed during loading. However, it is noted that the specimens do not fracture in two pieces, as seen from the side view in Figure 10c. The kink band is also clearly visible in the ultrasound, Figure 10d. The delamination damage is contained within the kinking zone. Combining plies in all three primary directions can keep the plies from both sides of the specimen connected and attached, even after severe kink deformation. This allows the specimen to carry load after initial failure.

Evolution of Elastic Modulus and Permanent Strain
From the nominal stress-global strain of the LUR tests, both the evolution of permanent strain and elastic modulus are derived. Additionally, the strain field on the specimen surface after unloading indicates the locations and pattern of permanent deformation. Figure 11 shows the evolution of permanent strain and modulus of the specimen with layup [+45/−45]5s. On the vertical axis, figures show the longitudinal specimen modulus ( ) normalized concerning the undamaged elastic modulus ( 0) and longitudinal specimen permanent strain. One can clearly see that a large amount of permanent strain develops. In addition, the elastic modulus reduces with each load cycle up to about 30% of the original modulus for tension and 40% (before buckling) for compression. In figures, the same symbols are used for the same test. The data points in gray occurred after buckling. Between tensile and compression loading, the evolution of permanent strain is similar. The evolution of specimen elastic modulus is less steep for compression.  Figure 10. Buckling was not observed during loading. However, it is noted that the specimens do not fracture in two pieces, as seen from the side view in Figure 10c. The kink band is also clearly visible in the ultrasound, Figure 10d. The delamination damage is contained within the kinking zone. Combining plies in all three primary directions can keep the plies from both sides of the specimen connected and attached, even after severe kink deformation. This allows the specimen to carry load after initial failure.

Evolution of Elastic Modulus and Permanent Strain
From the nominal stress-global strain of the LUR tests, both the evolution of permanent strain and elastic modulus are derived. Additionally, the strain field on the specimen surface after unloading indicates the locations and pattern of permanent deformation. Figure 11 shows the evolution of permanent strain and modulus of the specimen with layup [+45/−45] 5s . On the vertical axis, figures show the longitudinal specimen modulus (E spec ) normalized concerning the undamaged elastic modulus (E0) and longitudinal specimen permanent strain. One can clearly see that a large amount of permanent strain develops. In addition, the elastic modulus reduces with each load cycle up to about 30% of the original modulus for tension and 40% (before buckling) for compression. In figures, the same symbols are used for the same test. The data points in gray occurred after buckling. Between tensile and compression loading, the evolution of permanent strain is similar. The evolution of specimen elastic modulus is less steep for compression.  Figure 12 shows the evolution of the permanent longitudinal strain and elastic modulus for [0/90]5s and [+45/0/−45/90]3s. Though the nominal stress-global strain curve is quasilinear, both layups show a small evolution in the permanent global strain with a modulus reduction of about 20%. This suggests that, at least locally, some damage and plasticity occur in the specimen volume. Similar evolutions are shown between tension, compression, and per layup. Note the initially higher elastic modulus after the first unloading cycle in the compression data of Figure 12b. This is caused by variability in strain measurement at small strain magnitude. Despite this, both elastic modulus evolution and permanent strain evolution are similar between tension and compression. Although the global specimen behavior reveals the evolution of permanent strain and elastic modulus, it does not show where the permanent strain occurs. This can be identified from the surface strain fields obtained by DIC. Figures 13 and 14 show the longitudinal strain on the specimen surface at unloading after either the last load cycle or just before buckling. Note that the scales associated with individual figures are different. This was done to provide the clearest view of the variation in strain for each figure. For the samples with layup [+45/−45]5s X-shaped shear bands with high permanent strain develop. These meet at the side of the hole. This is illustrated in Figure 13a for the tensile case where the longitudinal strain is positive. For compression, the longitudinal strain has the same  Figure 12 shows the evolution of the permanent longitudinal strain and elastic modulus for [0/90] 5s and [+45/0/−45/90] 3s . Though the nominal stress-global strain curve is quasilinear, both layups show a small evolution in the permanent global strain with a modulus reduction of about 20%. This suggests that, at least locally, some damage and plasticity occur in the specimen volume. Similar evolutions are shown between tension, compression, and per layup. Note the initially higher elastic modulus after the first unloading cycle in the compression data of Figure 12b. This is caused by variability in strain measurement at small strain magnitude. Despite this, both elastic modulus evolution and permanent strain evolution are similar between tension and compression.  Although the global specimen behavior reveals the evolution of permanent strain and elastic modulus, it does not show where the permanent strain occurs. This can be identified from the surface strain fields obtained by DIC. Figures 13 and 14 show the longitudinal strain on the specimen surface at unloading after either the last load cycle or just before buckling. Note that the scales associated with individual figures are different. This was done to provide the clearest view of the variation in strain for each figure. For the samples with layup [+45/−45]5s X-shaped shear bands with high permanent strain develop. These meet at the side of the hole. This is illustrated in Figure 13a for the tensile case where the longitudinal strain is positive. For compression, the longitudinal strain has the same shape yet with negative strain values. For the samples with layup [0/90]5s, Figure 13b, a Although the global specimen behavior reveals the evolution of permanent strain and elastic modulus, it does not show where the permanent strain occurs. This can be identified from the surface strain fields obtained by DIC. Figures 13 and 14 show the longitudinal strain on the specimen surface at unloading after either the last load cycle or just before buckling. Note that the scales associated with individual figures are different. This was done to provide the clearest view of the variation in strain for each figure. For the samples with layup [+45/−45] 5s X-shaped shear bands with high permanent strain develop. These meet at the side of the hole. This is illustrated in Figure 13a for the tensile case where the longitudinal strain is positive. For compression, the longitudinal strain has the same shape yet with negative strain values. For the samples with layup [0/90] 5s , Figure 13b, a region of nearly zero longitudinal strain occurs above and below the hole center. To the left and right of the hole, a local region with permanent strain occurs. For the tensile case, the longitudinal strains are positive, while for the compressive case, these are negative. The region with larger permanent strain is limited to the immediate vicinity of the hole, which is in sharp contrast with layup [+45/−45] 5s , where the region of permanent strain extends fully to the specimen edges.
the longitudinal strains are positive, while for the compressive case, these are neg The region with larger permanent strain is limited to the immediate vicinity of the which is in sharp contrast with layup [+45/−45]5s, where the region of permanent extends fully to the specimen edges.
The permanent strain pattern for the QI layup, Figure 14, shows a combination patterns from the [0/90]5s and [+45/−45]5s laminates. In the tensile case, Figure 14a, r with nearly zero permanent strain are present above and below the hole, while le right from the hole, a region with limited permanent strain develops. Further away the hole, this region splits into two arms, extending similarly to the X-shaped patter in the [+45/−45]5s laminate. In contrast to the other laminates, the QI laminate in com sion, Figure 14b, does not seem to show an inverse pattern as the tensile case. In fa seems more similar to the compressive case of the [0/90]5s laminate. The authors att this difference to the different failure phenomena occurring for this laminate unde sion and compression. In fact, in the compressive case, one can already see the growth of the kink band forming close to the hole, while it does not yet span the ful imen width.  The permanent strain pattern for the QI layup, Figure 14, shows a combination of the patterns from the [0/90] 5s and [+45/−45] 5s laminates. In the tensile case, Figure 14a, regions with nearly zero permanent strain are present above and below the hole, while left and right from the hole, a region with limited permanent strain develops. Further away from the hole, this region splits into two arms, extending similarly to the X-shaped pattern seen in the [+45/−45] 5s laminate. In contrast to the other laminates, the QI laminate in compression, Figure 14b, does not seem to show an inverse pattern as the tensile case. In fact, this seems more similar to the compressive case of the [0/90] 5s laminate. The authors attribute this difference to the different failure phenomena occurring for this laminate under tension and compression. In fact, in the compressive case, one can already see the initial growth of the kink band forming close to the hole, while it does not yet span the full specimen width.

Conclusions
In The identification of permanent deformation as well as the elastic modulus degradation of the samples before failure is particularly useful for engineers in mechanical design and to validate new material models for thermoplastic materials. Even more so, the availability of the entire strain field on the sample surface throughout the test can be used to verify and validate finite element simulations of open-hole tests for this material. A remarkable observation is that, while the evolution of permanent global strain and elastic modulus reduction between the tensile and compressive load cases is quite similar, the distribution of permanent strain development can be different. This indicates the necessity to include full-field strain measurements, as well as both tensile as compressive loading, in future experimental programs on composite mechanical behavior.
Author Contributions: Tensile testing and microscopic investigation, R.D.S. and W.V.P., ultrasound measurement and interpretation, E.V., X.Y. and M.K., writing-original draft preparation, R.D.S.,  The identification of permanent deformation as well as the elastic modulus degradation of the samples before failure is particularly useful for engineers in mechanical design and to validate new material models for thermoplastic materials. Even more so, the availability of the entire strain field on the sample surface throughout the test can be used to verify and validate finite element simulations of open-hole tests for this material. A remarkable observation is that, while the evolution of permanent global strain and elastic modulus reduction between the tensile and compressive load cases is quite similar, the distribution of permanent strain development can be different. This indicates the necessity to include full-field strain measurements, as well as both tensile as compressive loading, in future experimental programs on composite mechanical behavior.

Conclusions
Author Contributions: Tensile testing and microscopic investigation, R.D.S. and W.V.P., ultrasound measurement and interpretation, E.V., X.Y. and M.K., writing-original draft preparation, R.D.S., E.V. and X.Y., data curation, R.D.S.; writing-review and editing, R.D.S., M.K. and W.V.P.; supervision, W.V.P. and M.K. All authors have read and agreed to the published version of the manuscript.

Funding: This research was funded by Research Foundation Flanders FWO, grant number 12R3221N.
Institutional Review Board Statement: Not applicable.

Informed Consent Statement: Not applicable.
Data Availability Statement: The research data for this article is stored at Ghent University for a minimum of five years after publication. The data can be requested by the corresponding author.