1 Introduction

In recent years, there has been a rapid growth in the use of fibre reinforced composite materials in engineering applications, replacing metals in many components, and there is evident indication that this will be continuing in consequence of their excellent specific strength and stiffness [1]. In this context the most attractive features offered by thermoplastic composites are the potential of low-cost manufacturing, high fracture toughness, good damage tolerance and impact resistance, good resistance to microcracking, easy quality control and the possibility to recycle the raw materials. As consequence several thermoplastic resins have been investigated as matrices, including polypropylene.

One of the most important applications of glass reinforced polypropylene is in automotive body panels made by low cost thermoforming techniques. In these applications the fatigue loads are typical, which reduces the strength of material/component due to the damage that occurs during the fatigue process. In previous works, parameters like frequency, stress concentration, lay-up design and load conditions were investigated by the authors [2–4] in terms of fatigue performance.

The fatigue damage in composite materials can be modelled by theoretical and experimental approaches such as those based on the number of debonded fibres, fracture mechanics parameters, the strain energy density, the loss of stiffness [2–8] and residual strength [6, 9]. However some techniques are destructives and in many situations it is necessary take decisions with the components and/or structures in-service. In these cases is essential the recourse to the non-destructive techniques (NDT) in order to detect and evaluate the fatigue damages and consequent loss of fatigue strength.

This paper presents the results of a study on glass-fibre-reinforced polypropylene composite in which the fatigue damage was investigated in terms of the stiffness loss and temperature rise. The loss of stiffness to quantify the fatigue damage [2–8] was derived from the stress-strain curves obtained from all specimens tested. The damage mechanisms were interpreted with the assistance of the acoustic emission technique [10–13]. Failure mechanisms like matrix cracking, debonding fibre/matrix and fibre breaking were observed and, supported by the literature [4, 12], were responsible of this loss stiffness. The specific rise in temperature was measured for all of the tests using the thermocouple method with one instrumentation system based on three thermocouples placed in the failure region. In fact fatigue tests showed that the rise in specimen temperature is due to damage [3, 4, 14–16], which increases with the fatigue life especially close to the final failure. The acoustic emission results could explain the damages that are involved in this temperature increases. According with the literature this technique is able to detect, in real time, damage mechanisms like: matrix cracking, fibre–matrix interface debonding, fibre fracture and delamination [17–21]. Also the temperature can be monitored by thermography technique [10, 11] and good agreement was observed between results obtained by both techniques. Several works conclude that the measurement of surface temperature by thermography is an efficient process to evaluate the fatigue damage evolution [22–24].

2 Materials and Experimental Procedures

Composite sheets were manufactured using multiple layers of Vetrotex “Twintex T PP” which were processed in a mould under a pressure of 5 bar during 10 min, after heating at 190°C. This temperature is above the melting temperature of the polypropylene. For each sheet were used seven woven balanced bi-directional layers. The overall dimensions of the plates were 250 × 160 × 3 mm with a fibre volume fraction (Vf,) of 0.34, according with the supplier. The quality control of the plates was performed by sampling using the ultrasonic C-Scan, to evaluate the eventual presence of defects and voids resulting from manufacturing process, and by visual inspection of the colour.

Three layered plates were manufactured. For one series, all the layers have one of the two fibre directions oriented with the axis of the plate. The other two-layer distribution was obtained with the following laminate orientation with respect to the axis of the sheet: +45°/0°/−45°/0°/−45°/0°/+45° and +30°/−30°/+30°/0°/+30°/−30°/+30°. For convenience these three types are referred as 0°, +45°/0°/−45° and +30°/−30°/0° plates respectively.

The specimens used in the fatigue tests were prepared from these thin plates. The geometry and dimensions of the samples are shown in Fig. 1. The fatigue tests were carried out in a servohydraulic Instron machine, model 1341, in constant amplitude load and also in displacement amplitude control. All the tests were performed in tension with stress ratio (R = minimum stress/maximum stress of loading wave) of 0.025 and frequency 10 Hz at ambient temperature.

Fig. 1
figure 1

Geometry and dimensions of the specimens used in fatigue tests

Full size image

During the fatigue tests the temperature rise at three points on the surface of the specimens (center, top and bottom at curvature regions) was measured using thermocouples, type K, and recorded in a computer. The maximum error of the lecture’s equipment is ±0.4°. At the same time the temperature rise was obtained by San-ei thermography equipment, model Thermo Tracer TH 1100. From the room temperature, at the surface of the specimen, was combined the equipment’s emissivity. This value was 0.89. The stiffness modulus was derived from the stress-strain curves obtained. These curves where nearly linear in the low stress level. The stiffness modulus was calculated as the tangent to a nonlinear curve stress versus strain record at zero strain.

Complementary tests to aid the interpretation the fatigue damage mechanisms associated with the loss stiffness were performed using the acoustic emission technique. For these tests it was used a Marandy, model MR1004, acoustic emission analyser which provides the amplitude and the number of ringdown counts for each acoustic emission (AE) event. This equipment has a threshold voltage adjustable from 10 mV to 69.18 mV. The AE event amplitude is defined as the maximum AE signal level relative to 10 mV. The amplitude detector unit sorts the AE events into twenty five levels, each one with a bandwidth of 2.4 dB. Level zero corresponds with the amplitude above 10 mV and below 13.18 mV. For ringdown counting, the amplitude of the signal is compared with the threshold voltage, and the total ringdown count for each event is the number of times of the signal amplitude exceeds the threshold level. Despite being recommended prior to calibration of equipment, using for example the ASTM 1106 “Standard method for primary calibration of acoustic emission sensors”, this is not done under restricted this work.

The tensile properties for these materials [3, 4] were obtained using an electromechanical Instron Testing machine, model 4206. Table 1 summarizes the results obtained with a grip displacement rate of 2,000 mm/min. This test rate was chosen to be similar to the mean fatigue tests deformation rate. The tabled values are the average of four tests carried out for each test condition.

Table 1 Mechanical properties (strain load 0.333 s−1) [3, 4]
Full size table

3 Results and Discussion

Fatigue tests were carried out at room temperature in load amplitude control and displacement amplitude control. The results are plotted in Fig. 2, in terms of the stress range versus the number of cycles to failure. For the displacement control tests the stress range was measured for the first fatigue cycle. The analysis of the figure shows that the fatigue strength for 0° laminates is around 1.5–1.8 times higher than the other laminates. This can be explained by a change of failure mechanisms presented in [4]. For +45°/0°/−45° and +30°/−30°/0° laminates the predominant fatigue mechanism is the debonding between the fibres/matrix, caused by normal stresses, and matrix cracking. Relatively to the 0° laminates the normal stress is predominantly absorbed by longitudinal fibres and the failure is in transverse planes.

Fig. 2
figure 2

S-N curves for 0°, +45°/0°/−45° and +30°/−30°/0° laminates

Full size image

It is possible to observe a small increase in fatigue strength for the laminates tested in the displacement control mode, which results in longitudinal stress loss. Ferreira et al. [4] observed a significant drop of the stress (5–10%) at the early stage of the fatigue life (5%). After this stage the stress decreases slowly, and in the last 5% of fatigue life, has a sudden drop. These authors suggested that this behaviour is related to stress release, temperature rise and stiffness decrease.

During all fatigue tests were observed that the temperature in the specimens increases significantly, which was recorded by thermocouples system and thermography technique. Figure 3 shows the temperature rise versus N/Nf, in constant amplitude displacement tests, where N is the number of cycles at any given instant of the tests and Nf is the number of cycles to failure. The plotted temperature corresponds to the maximum of the temperatures measured at the three thermocouples and the difference was ever less than 3°C between thermocouples. The same behaviour was observed for the specimens tested at constant load.

Fig. 3
figure 3

Temperature increase for 0°, +45°/0°/−45° and +30°/−30°/0° laminates

Full size image

The figure shows a similar behaviour for all laminates, where the temperature increases with the number of cycles and the maximum temperature was obtained at failure. This behaviour agrees perfectly with literature results [25–27]. Three stages can be identified as the result of different damage mechanisms in the fatigue process. At the first stage temperature increase is about 10°C and result of energy produced by inelastic deformation. During the second stage, representing more than 80% of total fatigue life, a very small increase of temperature was observed which varies linearly with the number of cycles. This phenomenon results from equilibrium between the internal energy produced and material’s energy transference capability, mainly by convection. At this second stage, beyond deformation energy there is a friction energy produced at the interface matrix-fibres as a result of interface separation produced in the first stage. Finally a sudden increase of temperature occurs and then the specimen fails. In fact, the energy produced by the different phenomena indicated before is such high that increases the temperature, reaching the highest value at the failure.

Figure 4 shows temperature maps obtained by thermography technique in fatigue test carried out at constant amplitude load, Δσ = 160.5 MPa, and for 0° laminates. However, these maps are representative of all specimens tested (0°, +45°/0°/−45° and +30°/−30°/0°). Figure 4a) presents the specimen at room temperature before the test start. Although the specimen can be distinguished on the blue background, this results from different emissivity of the air and specimen’s material. Figure 4b) shows temperature map of the sample at 3,600 cycles, which corresponds to the end of the first stage. A zone with higher temperature can be seen perfectly and results of the mechanisms of damage occurred during this stage and in that zone. Figures 4c) and d) show the distribution of the temperature in the sample at 15,600 and 33,000 cycles, respectively. These temperature maps represent the evolution of the temperature in the second stage. However the zone where the temperature is higher is more pronounced and denotes the zone where the mechanisms of damage are more influent. Finally Fig. 4e) shows the temperature in the sample 170 cycles before rupture (final rupture at 48,170 cycles). This moment is the end of the third stage where the temperature increases significantly, in the same zone where the temperature was ever higher. In fact the zone where the temperature is higher is the site where the rupture will occur.

Fig. 4
figure 4figure 4

Temperature maps obtained by thermography technique at: a) Before the test start; b) 3,600 cycles; c) 15,600 cycles; d) 33,000 cycles; e) 48,000 cycles

Full size image

Then it is possible empathized that the termography technique presents temperature maps in agreement with the values obtained directly by thermocouples system, associating the advantage that can be used to detect the precision site where the rupture will occur.

According with the literature other fatigue damage parameter is the residual stiffness, which can be associated with the degradation that occurs during the fatigue life. In this context Fig. 5 plots E versus N/Nf, where E is the instantaneous stiffness modulus, N is the number of cycles at any given instant of the tests and Nf is the number of cycles to failure, obtained in different constant amplitude load tests. These curves are very similar for all laminates tested and the trend of the curves is the inverse of that for temperature (Fig. 3). It can be seen that E values at the beginning of the tests, essentially present two values (one for 0° laminates and another for +45°/0°/−45° and +30°/−30°/0°) and are very similar to static tests values presented in Table 1. After the initial value of E a significant drop in stiffness modulus occurs, around 5%, during the first 5% of the fatigue life. In the next regime, the second stage, the stiffness modulus decreases slowly until close to final failure. During the last 5% of fatigue life, the third stage, the stiffness modulus drops suddenly. Such behaviour can be explained by the stress release, temperature rise and internal damages [3, 4].

Fig. 5
figure 5

Loss stiffness modulus for 0°, +45°/0°/−45° and +30°/−30°/0° laminates

Full size image

In order to better understanding this phenomenon the acoustic emission technique was applied. Figure 6 presents the results obtained during fatigue test carried out at constant amplitude load, Δσ = 160.5 MPa for 0° laminates, and represents the evolution of the “Ringdown Counting” versus number of cycles. In present work the threshold for Ringdown Counting was fixed at 18.45 mV. It can be seen that after 20,000 cycles high frequencies sounds are dominant, resulting from fibre breakage [12]. Up to 20,000 cycles lower frequencies sounds are dominant, resulting from matrix cracking and debonding between fibre/matrix [12].

Fig. 6
figure 6

“Ringdown Count” against number of cycles

Full size image

Figure 7 shows the number of events versus number of cycles, for different sound intensities. In this plot level one represents lower intensity sounds and the others levels represent sounds with higher intensity. Each of the twenty five different intensity levels represent a band with 2.4 dB, and the lower level corresponds to 10–13.8 mV. For example the level of intensity number 7, that correspond to the phenomenon occurred between 16.8 and 19.2 dB, it is represented in Fig. 8. It can be seen that the first fibre breakage occurs only after 25,000 cycles and is much more frequent after 32,000 cycles. On the other hand Fig. 9 presents lower intensity sounds, between 7.2 and 9.6 dB, and it is possible to observe that many parts of these events occur between 20,000 and 28,000 cycles. The analysis of Figs. 8 and 9 indicates that about 40,000 cycles all curves represented, for different sound levels intensity, increase significantly and then the specimen fails at 48,170 cycles.

Fig. 7
figure 7

Number of events against number of cycles for different sound intensities

Full size image
Fig. 8
figure 8

Number of events against number of cycles for phenomena occurred between 16.8 and 19.2 dB

Full size image
Fig. 9
figure 9

Number of events against number of cycles for phenomena occurred between 7.2 and 9.6 dB

Full size image

If we compare the results obtained with acoustic emission, temperature rise and loss stiffness, Figs. 3, 5, 7, 8 and 9, it is possible to observe that after starting test the events with lower level sound intensity occur, reported in literature as matrix cracking and crack propagation [12]. This will be the consequence of the significant drop in stiffness modulus observed in the first stage (Fig. 5). At the same time this damage mechanisms are responsible for temperature increase during the first stage (Fig. 3). The propagation of the matrix cracks contribute for the debonding fibre/matrix resulting mainly from triaxial stress state developed inside the composite. The phenomenon, represented by medium levels sound intensity, can be associated with friction energy produced at the interface matrix-fibres causing a slow temperature increase in the second stage. However in this stage fibres breakage occurs at about 25,000 cycles, which corresponds to the second half of the stage two (N/Nf = 0.52), and is another type of damage responsible for temperature increase and stiffness loss. At 32,000 cycles (N/Nf = 0.67) the majority of fibres begins to break and at this moment the temperature increase and stiffness loss is more pronounced, but is only around 40,000 cycles (N/Nf = 0.83) that all damages have a sudden increase up to specimen failure (third stage).

4 Conclusions

Fatigue damage was monitored in terms of losing in stiffness modulus, surface temperature rise, thermography technique and acoustic emission. The fatigue strength is influenced by the layer design and also by the loading mode. The stiffness modulus decreases during the fatigue tests and present three different stages. Initially a sudden drop occurs followed by a second stage where the stiffness modulus decreases slowly until close to final failure. During the last stage the stiffness modulus drops suddenly and all damages have a sudden increase up to specimen failure. Relatively to the temperature increase, the curves are very similar for both laminates and the trend of the curves is the inverse of that for stiffness modulus. At the first stage temperature rise around 10–15°C and results essentially from deformation energy and from the energy produced by damage mechanisms. During the second stage a very small increase of temperature occurs, which varies linearly with the number of cycles. This phenomenon results from equilibrium between the internal energy produced and material’s energy transference capability.

Thermography technique denotes a good agreement with the results obtained by thermocouples. With thermography it is possible to obtain temperature maps and the precision site where the failure will occur.