Table of Contents ECNDT '98
Quantifying the damage state of quasi-isotropic CFRP with embedded optical fibres during fatigue testing using acoustic emission and microfocus radiographyM. Surgeon*, M. Wevers
Department of Metallurgy and Materials Engineering, De Croylaan 2, B-3001 Heverlee (Belgium)
|TABLE OF CONTENTS|
Optical fibre technology is currently under investigation as a possible candidate for an in situ and continuous damage detection system for composite structures. Optical fibres offer some advantages as compared to e.g. traditional acoustic emission (AE) sensors: small dimensions, incorporation in the structure, no disturbance by electromagnetic interference,... These advantages are discussed more in detail in another paper in these proceedings (1).
One of the most important items that has to be considered during the development of an optical fibre sensor is the disturbance of the microstructure caused by the optical fibres. This disturbance can cause stress concentrations and thus a degradation of the mechanical properties of the resulting structure. A first step in this research has been an investigation into the mechanical properties of a quasi-isotropic laminate with and without embedded optical fibres (1). The results of this investigation have shown that any influence the optical fibres might have on the mechanical properties becomes most apparent during fatigue testing at an intermediate stress level. This type of test was therefore chosen in a second step to investigate the influence of the embedded optical fibres on the damage evolution sequence. Interrupted fatigue tests were carried out, which will be discussed below.
The specimens used for fatigue testing were 150 mm long, 12 mm wide and 1 mm thick. Previous results (1) had shown a considerable degradation in fatigue properties in the type B and C material. The aim of this study is to determine the reason for this degradation and to determine if any differences in damage evolution are the cause for it.
Three specimens of each material type were tested in an MTS 810 loading frame at a maximum stress level of 450 MPa. An R-factor of 0.1 and a frequency of 5 Hz were used during each test. Each test was interrupted at the following number of cycles (if the sample didn't fail prematurely) : 100, 1000, 10000, 25000, 50000 and 100000 cycles. During each interruption penetrant enhanced microfocus radiography was used to quantify the damage state of each specimen. A 40 mm long region around the middle of the specimen was used for evaluation. During each test the AE technique was used to monitor damage evolution in an attempt to make a correlation between the measured AE activity or AE signal parameters and the damage state observed on the radiographs. AE testing was done with a Vallen AMS3 using two broadband B1025 sensors (Digital Wave Corp) placed at a distance of 70 mm from each other.
The number of 90°-cracks
The 90°-cracks could easily be identified on the radiographs. Figure 1 shows the evolution of the number of cracks as a function of the number of cycles for the five different material types. As can be seen the behaviour is generally the same for all types : cracks initiate between 1000 and 10000 cycles after which a steep increase in the number of cracks is observed. The number of cracks appears to saturate at 105 cycles.
Fig 1: Number of matrix cracks as a function of the number of cycles for each material type
The curves for material types A, B and E are nearly identical throughout the test. The number of cracks for type C is somewhat lower up to 10000 cycles, but then it grows to about the same number as the above mentioned types, indicating that no distinct differences are noticeable for types A, B, C and E.
In type D matrix cracking appeared to initiate earlier. The number of cracks continued to be higher throughout the whole test, giving a first indication of a negative influence of embedded optical fibres in the -45/90-interface, corresponding well with the earlier observation of a diminished number of cycles to fracture (1).
The 90°-crack length
As most of the 90°-cracks didn't grow immediately over the entire width of the specimen, the total crack length was measured on each radiograph to get a complete picture of the damage state in each specimen. Figure 2 shows the crack length as a function of the number of cycles for each material type. The curves show the same behaviour as the ones in figure 1.
Fig 2: Total matrix crack length as a function of the number of cycles for each material type
It is clear that the crack evolution is similar for material types A, B, C and E. Again it is interesting to note that material type D has an earlier crack initiation and continues to have a significantly higher crack length than the other material types. The difference is clearer than in figure 1, once again indicating a negative influence of embedding optical fibres in the -45/90-interface.
The delamination size
The second damage mechanism that could easily be identified on the radiographs is a delamination growing between the -45 and 90-plies. The size of this delamination was measured during each test interruption. Figure 3 shows the evolution of delamination size as a function of the number of cycles for each material type.
The types A, B, C and E again show a very similar behaviour, once again indicating that no influence of the optical fibres on the damage evolution is present. The delamination appears to initiate between 10000 and 25000 cycles after which a steep increase in delamination size follows. At 100000 cycles an evolution to a saturation value is already obvious.
Once again the interesting material type is type D. In this case the delamination initiates earlier, between 1000 and 10000 cycles, which is significantly earlier than in the other types. Moreover the delamination size is continuously significantly higher.
Focusing on the number of cracks, the crack length and the delamination size, it has been noticed that the samples of the type D material exhibited higher values than the other types, indicating a negative influence of incorporating optical fibres in the -45/90-interface. This result confirms the results of the earlier continuous fatigue tests (1) during which all of the type D specimens failed prematurely. It seems that the explanation for the premature failure is the earlier initiation and faster growth of both the matrix cracks and the delamination size.
The continuous fatigue tests also revealed a negative influence of embedded optical fibres in the 0/45-interface. This influence obviously isn't caused by the damage mechanisms that were studied here. The degradation must be caused by phenomena that couldn't be observed by radiography, probably early fibre failure in the 0°-layer due to the bending of the plies around the embedded optical fibre.
Fig 3: Delamination size as a function of the number of cycles for each material type
High amplitude events
As was stated before, three fatigue tests were carried out for each material type. During each test interruption the accumulated number of high amplitude AE events (amplitude > 80 dB) was counted, whereas the number of 90°-cracks could be counted on the radiograph. A distinction was made here between the total number of 90°-cracks and the number of 90°-cracks that transversed the whole width of the specimen, since some literature results had indicated that matrix cracks growing in an unstable manner over the entire width of the specimen are likely to cause high amplitude events.
The obtained results were averaged for each material type. Due to space limitations nly the results for the type B material, which were typical for all material types, will be shown here. Figure 4 shows the number of high amplitude AE events and the number of 90°-cracks as a function of the number of cycles. If the curves are plotted with different scales it can be seen that both the number of high amplitude AE events and the number of 90°-cracks follow the same trend throughout the whole test. However, there is no one to one correlation, so it can not be concluded that each new 90°-crack leads to a high amplitude AE event.
Fig 4: Number of cracks and number of high amplitude AE events as a function of number of cycles for type B
Fig 5: Number of full width cracks and number of high amplitude AE events as a function of number of cycles for the type B material
Figure 5 shows the number of high amplitude AE events and the number of 90°-cracks transversing the whole width of the specimen as a function of the number of cycles. The scale for both curves is the same here. It seems that the correlation is approximately one to one during the first 25000 cycles of the test. However, no correlation is obvious during the later stages of the test, more full width cracks are observed than high amplitude AE events. This can probably be explained by the fact that unstable full width cracks are created more easily during the early stages of the test, when fewer cracks are present, whereas during the later stages of the test, when more cracks are present, the stress state hinders the creation of new unstable full width cracks and all new full width cracks are cracks that have grown progressively through the width in a stable manner.
It can be concluded that the correlation between the number of high amplitude AE events and the number of 90°-cracks isn't perfect, but that a relation between both appears to exist, certainly during the early stages of testing.
Middle amplitude events
Figure 6 shows the accumulated number of middle amplitude AE events (60 dB < amplitude < 80 dB) and the delamination size as a function of the number of cycles for the type B material which is once again taken as a typical example. When both curves are plotted on adequate scales it can be seen that they follow the same trend throughout the whole test. There appears to be a good correlation between the number of middle amplitude AE events and the delamination size, so it can be concluded that delamination growth causes mainly middle amplitude AE events.
In this case, however, it is not possible to talk about a one to one correlation. A delamination is, in contrast to e.g. a full width matrix crack, a more global form of damage, growing progressively throughout the whole specimen, which makes a more global correlation approach necessary.
Fig 6: Number of middle amplitude events and delamination size as a function of number of cycles for type B material
The AE results obtained during testing where used to check some statements put forward in literature. The one to one correlation between the number of high amplitude AE events and the number of full width matrix cracks is not followed perfectly in this material, but some relation still appears to exist, especially during the early stages of testing. A good correlation exists between the number of middle amplitude events and the delamination size.
As could be seen the correlations are not perfect, which can be explained by the fact that the separation between middle amplitude and high amplitude events is chosen arbitrarily at 80 dB. A perfect correlation would also require a good reproducibility of the test set-up (e.g. sensor coupling) from test to test.
Nevertheless there are clear indications that some of the damage phenomena in this laminate can be separated based on AE results. Care has to be taken, however, in extrapolating these results to other test circumstances. A change in laminate lay-up, sample size, type of sensors,... could lead to different results.
As one of the steps in the development of a damage detection sensor for composite materials based on optical fibre technology, a study was made of the damage phenomena appearing in a quasi-isotropic laminate with optical fibres embedded in the different interfaces during interrupted fatigue testing. Microfocus radiography was used to identify and quantify the different damage types. The main conclusion that could be drawn from the results was that an effect of the optical fibres was only observed when they were embedded in the -45/90-interface. In this case both the matrix crack length and the delamination size increased faster than in the other material types, which confirmed results obtained during continuous fatigue testing.
Additionally acoustic emission (AE) was used to monitor the damage evolution throughout the tests. The main conclusions that could be drawn from these tests is that it appears to be possible to separate some of the active damage phenomena based on AE results. Care has to be taken however in extrapolating these results to other test circumstances.
MS wishes to thank the Flemish IWT for granting him a research fellowship.