Impact Damage Detection of a Glass Fabric Composite Using Carbon Fiber Sensors with Regard to Mechanical Loading

: Structural Health Monitoring (SHM) of composite structures leads to greater safety during operation and reduces the cost of regular inspections. Impact damage detection is an important SHM task. Since impact damage can signiﬁcantly reduce the lifetime of composite structures, sensors for impact damage are of great interest. Carbon Fiber Sensors (CFSs) can be used to detect composite damage. CFSs are lightweight and compact, and they can be integrated during the manufacturing process. In our study, CFSs were manufactured from three types of carbon ﬁber tows and were integrated into different layers of the lay-up in order to investigate the inﬂuence on impact damage detection. The effect of mechanical loading and temperature change on the measured electrical resistance was investigated during cyclic ﬂexural tests. It was revealed that, it is possible to distinguish between changes in measured signals due to impact and due mechanical loading. The change in the measured electrical signal caused by temperature can be eliminated. CFSs can be used for impact damage detection of a glass fabric composite. A combination of thermography and CFSs as an active heating element also provides good results in the ﬁeld of impact damage detection


Introduction
Applications of composite materials are nowadays spreading from the aircraft and marine industry to the manufacturing industry and sports equipment. In almost all applications, composites are exposed to cyclic mechanical loading. In addition, various types of impacts can occur during the lifetime of the product. Impacts on composite can be visible or barely visible (very difficult to visual inspection or not possible to detect by visual inspection) [1,2]. Although visible impacts cause greater damage to the material (visible surface dents, fiber breakage, matrix cracking, and multiple delaminations between plies) and bigger loss of compression and tension strength than barely visible impact damage (BVID), they can be revealed by visual inspection and by other methods easier than BVID. Barely visible impacts cause delamination of the composite material, which leads to buckling and a significant decrease of compressive strength of the composite [3]. The internal damage of the composite material in form of delamination also grows under cyclic loading and can lead to fatal damage [2,4]. Detection of BVID in composites is therefore of great interest.
to cyclic flexural loading before and after impact loading. An inspection of the specimens was carried out using thermographic inspection and computer tomography (CT).
The influence of the ultimate elongation of the material on impact damage detection is studied. The relationship between the mechanical loading and the measured electrical resistance of the CFSs is also investigated. The effects of the arrangement of the CFS in the composite lay-up are discussed.

Sample Preparation
The specimens were made of glass woven fabric prepreg sheets used in the aerospace industry. The autoclave curing technique was chosen. Specimens were prepared in LA Composite, s.r.o. in Prague, Czech Republic. Laminate lay-up [+45/0/-45/90] sym was used for preparing the specimens. Prepreg material made of woven fabric with epoxy resin HexPly 1454 GM/50%/1035 (Hexcel Composites SASU, Bouguenais Cedex, France) was chosen. The following curing process was applied according to the manufacturer's recommendation: 125 • C for 90 min, 6 bar. The specimens were cut into strips with dimensions of 140 mm × 23 mm and were 1.3 mm in thickness, see Figure 1.
made of 8 layers of woven glass prepreg sheets were manufactured with embedded CFSs. The CFSs were embedded in three different positions (close to the top of the specimen, close to the bottom of the specimen, and in the middle of the lay-up). The specimens were exposed to cyclic flexural loading before and after impact loading. An inspection of the specimens was carried out using thermographic inspection and computer tomography (CT).
The influence of the ultimate elongation of the material on impact damage detection is studied. The relationship between the mechanical loading and the measured electrical resistance of the CFSs is also investigated. The effects of the arrangement of the CFS in the composite lay-up are discussed.

Sample Preparation
The specimens were made of glass woven fabric prepreg sheets used in the aerospace industry. The autoclave curing technique was chosen. Specimens were prepared in LA Composite, s.r.o. in Prague, Czech Republic. Laminate lay-up [+45/0/-45/90]sym was used for preparing the specimens. Prepreg material made of woven fabric with epoxy resin HexPly 1454 GM/50%/1035 (Hexcel Composites SASU, Bouguenais Cedex, France) was chosen. The following curing process was applied according to the manufacturer's recommendation: 125 °C for 90 min, 6 bar. The specimens were cut into strips with dimensions of 140 mm × 23 mm and were 1.3 mm in thickness, see Figure 1.  Three different integration arrangements were applied. The carbon fiber sensors were placed between the first and second layer, between the fourth and fifth layer, and between the seventh and eighth layer. Table 1 presents an overview of the specimens for the mechanical tests and the different CFS materials that were used.

Carbon Fiber Sensors (CFSs)
The carbon fiber sensors were made of three different fiber tows. The manufacturing process is described in [25]. The material properties of each material are given in Table 2. The materials for the preparation of the CFSs were selected in such a way that the influence of the ultimate elongation on the impact damage detection could be assessed. The aim was to find whether an impact can damage the sensor fiber, and how the damage to the sensor influences the signal. The T300 1000-50A PAN tow that was chosen had already been used for another comparative study and also for various sensing applications [23][24][25][26][27].

Mechanical Testing and Impact Loading
All specimens were exposed to a three-point bending (3PB) test. This type of cyclic flexural test was chosen in order to prevent damage in the area of electrical contacts. The configuration of the loading shown in Figure 2 results in tension/compression loading of the sensor fiber, due to its non-symmetrical integration (see also Figure 8). The cyclic loading was performed by the hydraulic testing system MTS Mini Bionix (MTS, Eden Prairie, MN, USA).
The specimens were loaded with a loading force of 1.5-15 N (load ratio = 0.1) at a frequency of 0.1 Hz. The maximum loading force was established during the preliminary test. During this test, a strain-gage was installed in the middle part of the outer surface of the specimen. The maximum loading force of 15 N corresponds to a measured longitudinal strain of 3200 µm/m during the preliminary test. This strain level corresponds to a maximal loading level in many applications to avoid intralaminar matrix cracks. The specimens were exposed to 200 load cycles before impact loading and to 200 load cycles after impact loading. Measurements of the changes in electrical resistance were performed using a Keysight 344401A multimeter. The 4-wire resistance measurement method was used in order to eliminate test lead resistances and contact resistances. The specimens were loaded with a loading force of 1.5-15 N (load ratio = 0.1) at frequency of 0.1 Hz. The maximum loading force was established during the preliminary test. During this test, a strain-gage was installed in the middle part of the outer surface o the specimen. The maximum loading force of 15 N corresponds to a measured longitudinal strain of 3200 μm/m during the preliminary test. This strain leve corresponds to a maximal loading level in many applications to avoid intralamina matrix cracks. The specimens were exposed to 200 load cycles before impact loading and to 200 load cycles after impact loading. Measurements of the changes in electrica resistance were performed using a Keysight 344401A multimeter. The 4-wire resistanc measurement method was used in order to eliminate test lead resistances and contac resistances.
The possibility of detecting impact damage using CFSs was investigated by a simpl drop weight impact test using an impact device made in-house according to ASTM D 7136. The specimens were fixture to the supports according to Figure 3. The groove in th supports was made in order to prevent damage to the sensor during clamping of th specimen. The configuration of the impact test is shown in Figure 3. An impactor 16 mm in diameter and 410 g in weight (m) was used. The height (h) for the drop impact test wa 0.5 m, so the impact energy was 2 J. The impact energy was calculated according to formula (2). The possibility of detecting impact damage using CFSs was investigated by a simple drop weight impact test using an impact device made in-house according to ASTM D 7136. The specimens were fixture to the supports according to Figure 3. The groove in the supports was made in order to prevent damage to the sensor during clamping of the specimen. The configuration of the impact test is shown in Figure 3. An impactor 16 mm in diameter and 410 g in weight (m) was used. The height (h) for the drop impact test was 0.5 m, so the impact energy was 2 J. The impact energy was calculated according to Formula (1). E = m·g·h.
(1) The specimens were loaded with a loading force of 1.5-15 N (load ratio = 0.1) at a frequency of 0.1 Hz. The maximum loading force was established during the preliminary test. During this test, a strain-gage was installed in the middle part of the outer surface of the specimen. The maximum loading force of 15 N corresponds to a measured longitudinal strain of 3200 μm/m during the preliminary test. This strain level corresponds to a maximal loading level in many applications to avoid intralaminar matrix cracks. The specimens were exposed to 200 load cycles before impact loading and to 200 load cycles after impact loading. Measurements of the changes in electrical resistance were performed using a Keysight 344401A multimeter. The 4-wire resistance measurement method was used in order to eliminate test lead resistances and contact resistances.
The possibility of detecting impact damage using CFSs was investigated by a simple drop weight impact test using an impact device made in-house according to ASTM D 7136. The specimens were fixture to the supports according to Figure 3. The groove in the supports was made in order to prevent damage to the sensor during clamping of the specimen. The configuration of the impact test is shown in Figure 3. An impactor 16 mm in diameter and 410 g in weight (m) was used. The height (h) for the drop impact test was 0.5 m, so the impact energy was 2 J. The impact energy was calculated according to formula (2).
There was no visible damage to the impacted side of the specimen (the mold side of the specimen was impacted) after the 2 J impact. There were damages to the opposite side of the specimens which could be also detected by visual inspection for some specimens. There was no visible damage to the impacted side of the specimen (the mold side of the specimen was impacted) after the 2 J impact. There were damages to the opposite side of the specimens which could be also detected by visual inspection for some specimens.

Temperature Loading
Several specimens with integrated CFSs were manufactured for the experimental study on the influence of temperature on the electrical resistance of CFSs made of different materials (see Table 3). The CFSs were embedded between the same layers for all specimens (between the first and the second layer). The HBM QuantumX MX 1615 measuring amplifier was used at a 50 Hz sampling rate. The CFS-probes were connected by a 2-wire resistance measuring circuit. In addition, two PT-100 temperature sensors (Heraeus Nexensos M222, tolerance: F 0.1) were connected by a 4-wire resistance measurement (see Figure 1). One PT-100 temperature sensor was placed in such a way that the flowing air could pass freely. The other sensor was attached to the surface of the specimen to measure the temperature near the inertia of the specimen. The temperature of the oven (Zwick temperature chamber EC75A) was incremented with 20 • C temperature steps with a hold time of 30 min to a maximum temperature of 120 • C. After the maximum temperature was reached, the door of the oven was opened for a cool down period of 60 min.

Non-Destructive Testing
The FLIR A325sc infrared camera was employed for an active thermographic inspection of all specimens using the sensor fiber as heating element. All specimens were inspected after they were manufactured and after drop-weight impact testing. Several specimens were also inspected after 200 cycles of flexural loading and before the impact test.
The inspection was applied on the mold side of the specimen (=impacted side = outer surface during the cyclic flexural test). An electric current was applied to the CFSs integrated in the specimens during the thermographic inspection for 10 min via a direct current source (Agilent E3631A). The type T300 CFSs were subjected to an electric current of 0.09 A. An electric current of 0.3 A was used for CFS type 95 A, and an electric current of 0.24 A was used for CFSs type 70 A.
Zeiss Metrotom 1500 Computer Tomography (CT) was employed to evaluate the impact-induced damage to the composite laminates. A detector resolution of the CT was 2048 × 2048 px, 1950 frames were taken during scanning. Specimens that were investigated using CT are written in bold in Table 1. The specimens were investigated using CT after all mechanical tests had been completed. Figure 4 presents the raw measured values. In Figure 4a, the measured temperatures show the behavior of the temperature control of the oven. After a new temperature value has been set, the oven needs up to 15 min to reach a constant temperature level. The differences between the PT100 sensors just after a new temperature has been set due to the airflow inside the oven and the different location of the temperature sensor. At lower temperatures, the two PT100 sensors show a difference in the measured values of about 0.02 • C, which increases to about 1.97 • C or 1.64% at 120 • C. For further evaluation, the data of the sensor attached to the surface of the specimen was used. An increase in resistance with an increase in temperature can be observed for the 95 A CFS in Figure 4d. This effect is more pronounced with higher resistance values before the temperature measurement and can probably be attributed to pre-damage at the electrical contact point during manufacturing (see Section 3.3). No further examination was made of 95A-CFS.

Temperature Loading
temperatures, the two PT100 sensors show a difference in the measured values of about 0.02 °C, which increases to about 1.97 °C or 1.64% at 120 °C. For further evaluation, the data of the sensor attached to the surface of the specimen was used.  An increase in resistance with an increase in temperature can be observed for the 95 A CFS in Figure 4c. This effect is more pronounced with higher resistance values before the temperature measurement and can probably be attributed to pre-damage at the electrical contact point during manufacturing (see Section 3.3). No further examination was made of 95A-CFS.
The T300 samples show a resistance drop of 3% (0.8 Ω) while the 70 A samples show a drop of 8.5% (0.3 Ω) at a temperature level of 120 °C. This means that the effect of the temperature on the fiber resistance is greater for 70 A than for T300-CFS.
For practical applications, it is necessary to calculate the linear fit model from the measured data. For this purpose, the measured values were taken between 15 min after setting a new temperature and the next increase in temperature. For each temperature level, the mean values of the measured temperature were calculated and were used for the linear fit model (see Figure 5). 3 Ω) at a temperature level of 120 • C. This means that the effect of the temperature on the fiber resistance is greater for 70 A than for T300-CFS.
For practical applications, it is necessary to calculate the linear fit model from the measured data. For this purpose, the measured values were taken between 15 min after setting a new temperature and the next increase in temperature. For each temperature level, the mean values of the measured temperature were calculated and were used for the linear fit model (see Figure 5). A linear fit model was chosen, seen Equation (2), with the slope of the linear fit m, the resistance at 0 °C of the CFS R0°C, and the measured resistance of the CFS RT.   (2), with the slope of the linear fit m, the resistance at 0 • C of the CFS R 0 • C , and the measured resistance of the CFS R T .
The calculated coefficients of the linear model, and also the R 2 and root mean squared error (RMSE) values, are summarized in Table 4. The high R 2 values show, that the CFS signals show good linear behavior. The RMSE values show the magnitude of the error between the linear fit function and the measured mean values of the CFS samples. For an evaluation of the cyclic experiments and also for the possible practical utilization of the CFSs, it is necessary to evaluate a possible ±3 • C change in temperature. This corresponds to an increase of ±0.1% in the relative electrical resistance for a CFS made of T300 material and of ±0.3% for a CFS made of 70 A material (calculated for an averaged value of m and R 0 • C from the measured data).

Mechanical Testing
Most composite structures for which impact damage detection is an intending aspect are also exposed to cyclic loading. The influence of cyclic mechanical loading on the measured signal of impact sensors is therefore of great interest. The purpose of the mechanical testing was to describe the response of the measured signal of the investigated sensors with regard to: 1.
The positioning of the CFS in the composites lay-up; 3.
Influence of different materials on the CFS signal.
Specimens with integrated sensors were exposed to cyclic flexural loading before and after impact loading. During mechanical loading, the room temperature was measured in order to exclude this influence. The influence of a change in temperature on the measured signal of CFSs is described in Section 3.1.
In practical applications, it is not always possible to make temperature measurements simultaneously at the position where impact detection is needed. It is possible to use temperature compensation of the half-bridge circuit as used, for instance for strain-gauges [28]. Another option is to evaluate the peak-to-peak values of the cyclic loading [29] (see Figure 6). The schematic representation shows the evaluation procedure applied for the two signals. In the upper part of the figure is depicted an example of a signal from CFS during measurement under constant temperature. The first three cycles show signal before impact loading and the next three cycles represent signal after impact loading under cyclic loading and constant temperature. In the lower part of the figure depicts a graph of the measured signal from CFS under rising temperature before and after impact loading. We assume, that the rising of the temperature is much slower than the change of mechanical strain applied during cyclic loading. This is often the case in practical applications. The graph on the right side show measured peak to peak values of electrical resistance for constant and rising temperature. When using this procedure, temperature measurements are not needed. applied for the two signals. In the upper part of the figure is depicted an example of a signal from CFS during measurement under constant temperature. The first three cycles show signal before impact loading and the next three cycles represent signal after impact loading under cyclic loading and constant temperature. In the lower part of the figure depicts a graph of the measured signal from CFS under rising temperature before and after impact loading. We assume, that the rising of the temperature is much slower than the change of mechanical strain applied during cyclic loading. This is often the case in practical applications. The graph on the right side show measured peak to peak values of electrical resistance for constant and rising temperature. When using this procedure, temperature measurements are not needed. Figure 6. Evaluating the peak-to peak data. Figure 7 presents the peak-to-peak relative values of the measured electrical resistance for all specimens that underwent mechanical and impact loading.
(a) (b) Figure 6. Evaluating the peak-to peak data. Figure 7 presents the peak-to-peak relative values of the measured electrical resistance for all specimens that underwent mechanical and impact loading. applied for the two signals. In the upper part of the figure is depicted an example of a signal from CFS during measurement under constant temperature. The first three cycles show signal before impact loading and the next three cycles represent signal after impact loading under cyclic loading and constant temperature. In the lower part of the figure depicts a graph of the measured signal from CFS under rising temperature before and after impact loading. We assume, that the rising of the temperature is much slower than the change of mechanical strain applied during cyclic loading. This is often the case in practical applications. The graph on the right side show measured peak to peak values of electrical resistance for constant and rising temperature. When using this procedure, temperature measurements are not needed. Figure 6. Evaluating the peak-to peak data. Figure 7 presents the peak-to-peak relative values of the measured electrical resistance for all specimens that underwent mechanical and impact loading. The relative values were determined to allow an easier comparison of sensors made of different carbon fiber tows. The peak-to-peak relative values were determined for the first and last cycle before impact loading (cycle 1 and cycle 200) and for the first and last cycle after impact loading (cycle 201 and cycle 400). The peak-to-peak values were related to the measured value of the electrical resistance of the sensor after implementation. For The relative values were determined to allow an easier comparison of sensors made of different carbon fiber tows. The peak-to-peak relative values were determined for the first and last cycle before impact loading (cycle 1 and cycle 200) and for the first and last cycle after impact loading (cycle 201 and cycle 400). The peak-to-peak values were related to the measured value of the electrical resistance of the sensor after implementation. For the Xth cycle of the loading the equation is as follows: According to the measured data, it is evident, that all types of carbon fiber tows investigated here show piezoresistive behavior. A positive change in measured electrical resistance was observed for specimens, in which the sensor was integrated on the tension side of the specimen (between layers 1 and 2), see Figures 7 and 8. (c) (d) Figure 7. Peak-to-peak relative values of the measured signal from CFSs integrated into the specimens before and after impact loading: (a) results for specimens T300; (b) results for specimens T300-detail; (c) results for specimens 95 A; (d) results for specimens 70 A.
The relative values were determined to allow an easier comparison of sensors made of different carbon fiber tows. The peak-to-peak relative values were determined for the first and last cycle before impact loading (cycle 1 and cycle 200) and for the first and last cycle after impact loading (cycle 201 and cycle 400). The peak-to-peak values were related to the measured value of the electrical resistance of the sensor after implementation. For the Xth cycle of the loading the equation is as follows: According to the measured data, it is evident, that all types of carbon fiber tows investigated here show piezoresistive behavior. A positive change in measured electrical resistance was observed for specimens, in which the sensor was integrated on the tension side of the specimen (between layers 1 and 2), see Figures 7 and 8. (a) (b) (c) Figure 8. Configuration of the 3PB test and the measured signal from the integrated CFSs: (a) results for specimen with CFS between the 1st and 2nd layer during 3PB (measured data of T300-30), (b) results for specimen with CFS between the 4th and 5th layer during 3PB (measured data of T300-37), (c) results for specimen with CFS between the 7th and 8th layer during 3PB (measured data of T300-16).
The dependency between the number of cycles and the measured resistance for the sensors can be neglected in the case of material T300 before impact. Impacted sensors made of material T300 also show quite limited dependency between the number of cycles and the measured resistance. All specimens show a change lower than 0.7% in the peak-to-peak relative values between the first cycle and the 200th cycle, with the Figure 8. Configuration of the 3PB test and the measured signal from the integrated CFSs: (a) results for specimen with CFS between the 1st and 2nd layer during 3PB (measured data of T300-30), (b) results for specimen with CFS between the 4th and 5th layer during 3PB (measured data of T300-37), (c) results for specimen with CFS between the 7th and 8th layer during 3PB (measured data of T300-16).
The dependency between the number of cycles and the measured resistance for the sensors can be neglected in the case of material T300 before impact. Impacted sensors made of material T300 also show quite limited dependency between the number of cycles and the measured resistance. All specimens show a change lower than 0.7% in the peakto-peak relative values between the first cycle and the 200th cycle, with the exception of specimen T300-32, which shows a change of 7%. The influence of the number of cycles on the measured signal is much more pronounced for the specimens with integrated sensors made of pitch carbon fiber tow 95 A and 70 A. Compressive loading has a bigger impact on the change in the measured signal for both of these materials.
For material 70A, we also have data for cyclic loading after impact loading. Figure 7 shows that the bigger the change is in the measured electrical resistance after impact, the greater is the influence of the number of cycles after impact on the change in the measured electrical resistance. Peak-to-peak values for sensors made of the most brittle material 95 A are not given. The reason is that the specimens with sensors of this type were not subjected to a second campaign of cyclic loading, because damage in the contact area was detected for this type of CFS already after implementation, see Section 3.3.
After 200 cycles of flexural 3PB loading, the specimens were subjected to impact loading according to the configuration depicted in Figure 3. According to our hypothesis and in accordance with a previous experimental investigation published in [27], sensors made of a material with lower ultimate elongation should be more sensitive to impact loading. The relative change in the measured electrical resistance of the integrated sensor after impact should be greater for sensors made of a more brittle carbon fiber tow.
This hypothesis was not confirmed for all types of specimens. For the specimens with sensors on the impacted side, the biggest change in electrical resistance was measured for sensors made of material 95 A, which is the most brittle. For the specimens with sensors on the opposite side to the impact, the biggest change in electrical resistance was measured for material 70 A. This may have been caused by the fact that the sensors made of 95A had damaged electrical contacts already after implementation.
After 200 cycles of flexural 3PB loading, the specimens were subjected to impact loading according to the configuration depicted in Figure 3. According to our hypothesis and in accordance with a previous experimental investigation published in [29], sensors made of a material with lower ultimate elongation should be more sensitive to impact loading. The relative change in the measured electrical resistance of the integrated sensor after impact should be greater for sensors made of a more brittle carbon fiber tow.
This hypothesis was not confirmed for all types of specimens. For the specimens with sensors on the impacted side, the biggest change in electrical resistance was measured for sensors made of material 95 A, which is the most brittle. For the specimens with sensors on the opposite side to the impact, the biggest change in electrical resistance was measured for material 70 A. This may have been caused by the fact that the sensors made of 95A had damaged electrical contacts already after implementation.
Based on the measured data, positioning the CFS sensor on the opposite side to the impact seems to be more appropriate (specimens T300-16, T300-17, T300-18; 70A-7, 70A-8, 70A-9; 95A-9, 95A-11, 95A-12), because the change in the measured electrical resistance was the biggest (see Figures 9-11).   Peak-to-peak values after impact for sensors made of the most brittle material 95 A are not given, because the specimens with sensors of this type were not subjected to the second campaign of cyclic loading.  Peak-to-peak values after impact for sensors made of the most brittle material 95 A are not given, because the specimens with sensors of this type were not subjected to the second campaign of cyclic loading.

Active Thermographic Inspection and Computer Tomography
An active thermographic inspection of the specimens before loading revealed damage to the electrical contacts of the sensors made of pitch carbon fiber tow 95 A. The temperature profile along the sensor with the damaged contact area is shown in Figure  12. In Figure 13. The temperature profile of a sensor without any pre-damage is depicted. Six out of 9 specimens with implemented sensors made of material 95 A were definitely damaged during manufacturing or during implementation. Two specimens were indeterminate regarding damage to the sensor, and no data were provided for one specimen. This demonstrates the sensitivity for the structural integration of carbon fiber sensors made of brittle ex-pitch filaments.

Active Thermographic Inspection and Computer Tomography
An active thermographic inspection of the specimens before loading revealed damage to the electrical contacts of the sensors made of pitch carbon fiber tow 95 A. The temperature profile along the sensor with the damaged contact area is shown in Figure 12. In Figure 13. The temperature profile of a sensor without any pre-damage is depicted. Six out of 9 specimens with implemented sensors made of material 95 A were definitely damaged during manufacturing or during implementation. Two specimens were indeterminate regarding damage to the sensor, and no data were provided for one specimen. This demonstrates the sensitivity for the structural integration of carbon fiber sensors made of brittle ex-pitch filaments.   Sensors made of pitch 70 A carbon fiber tow and sensors made of pan T300 carbon fiber tow showed no signs of damage before loading.
Specimens T300-16, T300-17, 95 A-11, 95 A-12, 70 A-8, 70 A-7 were investigated after 200 cycles of flexural loading and before impact loading. No signs of damage caused by flexural cyclic loading were observed for specimens with a T300 sensor. Damage to the electrical contacts could be detected for specimens 95 A-11, 95 A-12. The temperature profile along the sensor changed slightly after cyclic loading in the case of sensors 70 A-7 and 70 A-8.
The specimens were thermographically inspected again after impact loading. The results of this observation are given in Table 5. Based on our measurements, T300 has the greatest potential for use as a heating element for active thermography. It was possible to detect the impact if the sensor was on the impacted site, and if the sensor was on the opposite side of the specimen to the impact. Based on observations with CFSs made of material 95A, the impact can also be detected by a sensor with damaged electrical Sensors made of pitch 70 A carbon fiber tow and sensors made of pan T300 carbon fiber tow showed no signs of damage before loading.
The specimens were thermographically inspected again after impact loading. The results of this observation are given in Table 5. Based on our measurements, T300 has the greatest potential for use as a heating element for active thermography. It was possible to detect the impact if the sensor was on the impacted site, and if the sensor was on the opposite side of the specimen to the impact. Based on observations with CFSs made of material 95A, the impact can also be detected by a sensor with damaged electrical contacts. In the case of specimens with an integrated CFS made of material 70 A, damage to the electrical contacts was observed for specimens 70 A-1, 70 A-3, and 70 A-9 after cyclic loading and impact loading. Unfortunately, only two specimens for each type of CFS material were inspected thermographically after 200 cycles, and before impact loading, for material type 70 A, specimens 70 A-7 and 70-8 were checked. Therefore, based on data from active thermography, it is not possible to determine whether the damage to the electrical contacts was caused by impact loading or by cyclic loading. Three specimens of each type of carbon fiber tow underwent CT, see Table 1. N signs of damages were detected using CT. Different sizes of the inserted carbon fiber tow are illustrated in pictures made by CT, see Figure 14.

Conclusions
Impact damage detection using sensors made of different types of carbon fiber tow has been demonstrated on rectangular coupons made of a prepreg glass fabric composite According to obtained results, it is possible to detect barely visible impact damage o glass fabric composite using integrated CFS.
Different ways of positioning the sensors in the lay-up were investigated. Accordin to our investigation, it is better to place the CFSs on the side opposite to an impact, rathe than on the impacted side. If a CFS is positioned on the side opposite to the impact, ther is a bigger probability that the impact can be detected. Integrating the sensor in th middle of the lay-up is not convenient for impact damage detection.
The Pan type of fiber tow used for integrated CFS showed stable cyclic loadin behavior. Sensors made of pitch carbon fiber tows were more sensitive to cyclic loading For industrial applications with a lot of cyclic loading, a more extensive experimenta campaign on the influence of cyclic loading on the measured signal need to be done.
It is not advisable to manufacture CFSs using extremely brittle pitch carbon fibe tows (Pitch 95 A), because the contact quality cannot be guaranteed, and there ar difficulties when manufacturing the sensor and during handling (during integration int the structure). The difficulties came from the brittleness of the Pitch 95 A fiber itself. It i assumed that failure occurred in the contact zone between the filaments and the solder The handling of the CFS made of 70 A and T300 material is problem-free for th manufacturing operator. Moreover, no damage after manufacturing was observed usin active thermography for specimens with sensors made of 70 A and T300 material.
The influence of the mechanical loading on the change in the electrical resistance o

Conclusions
Impact damage detection using sensors made of different types of carbon fiber tows has been demonstrated on rectangular coupons made of a prepreg glass fabric composite. According to obtained results, it is possible to detect barely visible impact damage of glass fabric composite using integrated CFS.
Different ways of positioning the sensors in the lay-up were investigated. According to our investigation, it is better to place the CFSs on the side opposite to an impact, rather than on the impacted side. If a CFS is positioned on the side opposite to the impact, there is a bigger probability that the impact can be detected. Integrating the sensor in the middle of the lay-up is not convenient for impact damage detection.
The Pan type of fiber tow used for integrated CFS showed stable cyclic loading behavior. Sensors made of pitch carbon fiber tows were more sensitive to cyclic loading. For industrial applications with a lot of cyclic loading, a more extensive experimental campaign on the influence of cyclic loading on the measured signal need to be done.
It is not advisable to manufacture CFSs using extremely brittle pitch carbon fiber tows (Pitch 95 A), because the contact quality cannot be guaranteed, and there are difficulties when manufacturing the sensor and during handling (during integration into the structure). The difficulties came from the brittleness of the Pitch 95 A fiber itself. It is assumed that failure occurred in the contact zone between the filaments and the solder. The handling of the CFS made of 70 A and T300 material is problem-free for the manufacturing operator. Moreover, no damage after manufacturing was observed using active thermography for specimens with sensors made of 70 A and T300 material.