3.1

Optical fiber distributed sensors

Optical fibers, besides the well-known applications in the telecom sector, can also be used as optical sensors to detect a wide range of parameters, like temperature, strain, or pressure. They present many advantages in comparison to conventional electronic sensors, such as lightweight, reduced space and cost, flexibility, and electromagnetic interference immunity[14].

Optical fiber sensors were initially realized as pointwise sensors, however, distributed sensors have gained most of the attention in the last years due to their key advantages. These specialty sensors benefit from an overall cost reduction by allowing measurements of tens of kilometers with a high spatial resolution and no dead zones, and they provide real-time monitoring of the health of the structures and their environmental conditions in a variety of applications [19]. There exist several distributed sensing techniques, but recently, the Optical Frequency Domain Reflectometry (OFDR) has gained significant attention in distributed sensing applications due to its relatively high spatial resolution combined with a large dynamic range. The ability of this technology to measure a high range of strain (even small changes in residual stresses), and the capacity the optical fibers provide to perform measurements in the presence of high temperature and pressure, make this type of sensor the perfect tool for monitoring the autoclave curing process of a CFRP structure. Therefore, a single fiber embedded in a composite material can provide a full understanding of the health of the whole structure at once. In this work, we employ the OFDR technique with a strain resolution of 1 με, a sampling resolution of 2 mm, and a repetition rate of 30 s.

3.2

Test samples

To analyze the strain evolution during the composite curing process six samples were manufactured. The samples consisted of a prepreg stack with optical fibers embedded. For the sake of comparison, we employed 2 different test layouts (3 samples per each). The first one with the optical fibers embedded in the middle of the composite stack, and the second one with the optical fibers at the top of the composite stack. In all the samples, the optical fiber crosses the composite sample longitudinally 3 times to provide 3D measurements of the structure strain. The sample layout is presented in Figure 1.

Figure 1.

Sample and optical fiber layout.

The material used for the samples is the space-qualified M18/M55J prepreg. This material was chosen to reproduce as close as possible a final application of the technology. The type of optical fibers employed is SMF (Single Mode Fiber) with a reduced diameter and a special coating. This fiber was chosen due to its high-temperature resistance and its minimum influence on the structure of the composite [20]. The sample dimensions are specially designed considering the size of the curing chamber. All samples were put inside a vacuum bag and placed on aluminum plates. An example of one of the samples prepared for curing is shown in Figure 2.

Figure 2.

Sample in the vacuum bag before curing.

3.3

Composite curing setup

The setup used for the curing of the tested samples is presented in Figure 3. This schematic represents the employed measuring setup which was prepared to ensure the optimal conditions for M18/M55J prepreg composite material curing and the possibility to route the optic fibers outside the pressure chamber.

Figure 3.

Composite curing setup.

The test setup was based on a curing pressure chamber (1) connected to a vacuum pump (2) and a compressor (3) to apply the proper values of pressure and vacuum in the vacuum bag. The chamber was connected to a thermal regulation device (4) working based on the information provided by a thermocouple in a feedback loop (5). To perform the measurements, the test sample (7) was connected to an interrogation unit (6) using an optical fiber pigtail coming out from the thermal chamber (8). The temperature and pressure change was applied following the standard curing procedure.

The pressure is set at 6 bars for the whole process. The vacuum also remains constant in the whole process at -0.25 bar. The first stage of the process begins with heating the sample from room temperature to 180 °C at a constant speed of 1.2 °C per minute. Once the maximum temperature is reached the curing stage begins, where the temperature remains constant at 180 °C for 2 hours. Finally, in the last stage, the sample is cooled down at a constant speed of 1.2 °C per minute. When the temperature reaches 50 °C the pressure and vacuum are released and the process is finished.

4.1

Predictions

As beforementioned, in a thermal curing process, three main stages can be distinguished: heating at a constant gradient of temperature, curing at a constant temperature, and cooling at a constant gradient of temperature. Based on the thermal coefficients of the materials involved in the test (carbon fiber, epoxy resin, aluminum, and glass optical fiber) the following predictions were made: In the first stage, the strains in the sample should grow (elongation) due to the positive thermal coefficients of the materials; In the second stage the strains should remain the constant or get smaller as the resin undergoes a cross-linking process; In the third stage, the strains should get smaller and even change from elongation to compression due to the different thermal coefficients of the materials [21][22].

4.2.

Experimental results

The optical fibers were spliced in series to measure all the parts embedded into composite simultaneously. This is possible due to the use of distributed sensing, which allows achieving measurements in each part of the fiber. The OFDR traces were continuously recorded along the whole curing process.

The raw data gathered from the device was in form of spectral shift. According to the current state of the art, it is assumed that the spectral shift is linearly proportional to the temperature or strain change of the sensing fiber. The proportionality coefficient, called respectively temperature sensitivity or strain sensitivity is a characteristic value for each particular fiber and it depends on the material and processes used in the manufacturing of the fiber. In the case of a typical single-mode fiber, those coefficients are equal to: and . The whole relation can be defined with the following equation:

Δv – spectral shift (wavelength frequency),

S_T, S_ε – temperature and strain sensitivity coefficients,

ΔT, Δε – temperature and strain change regarding equilibrium state.

As can be seen, the spectral shift depends on both temperature and strain. As during the experiment both parameters could be changed (the temperature due to the heating and the strain due to the residual stress development) a solution to distinguish both effects must be provided. Therefore, apart from the testing fibers we employed an additional fiber loop attached on the surface of the vacuum bag for reference and temperature compensation. The fiber was positioned in the way that it was convoluted in the area of the composite sample (it was not excessing its surface). This fiber was exposed only to temperature changes, not to other forces. The signal from the fiber inside the composite was compensated with the temperature data from the reference loop in postprocessing. Since both strain and temperature measurements were conducted optically, the compensation process could be easily provided by the simple subtraction of the raw data measurements. This way, the data from the fiber inside the composite could be translated into strain.

The compensated spectral shift for one of the samples is presented in Figure 4, the data of an OFDR trace before the curing process on the left, and after curing at room temperature on the right. The green areas marked on the plots correspond to the temperature reference fiber loop, whereas the red areas correspond to the optical fibers inside the composite. As can be seen, some residual stresses remain in the red areas after the curing process is finished.

Figure 4.

Spectral shift before and after curing.

After processing all the OFDR traces and obtaining the strain values from the spectral shifts, the evolution of the induced strain as a function of the curing time can be plotted. Figure 5 and Figure 6 show the results obtained for a sample with the optical fiber in the middle of the composite stack, and a sample with the optical fiber on the top of the stack respectively. For the sake of simplicity, only the results of these two samples are shown, as a repeatable pattern was found for all the samples of the same type. The plotted data correspond to a single fiber section that crossed the sample longitudinally. The strain data were averaged for the fiber length to get the mean value of the strain change for every timestamp.

Figure 5.

Evolution of the strain during curing - sample with fibers in the middle of the composite stack.

Figure 6.

Evolution of the strain during curing - sample with fibers on the surface of the composite stack.

Based on the plots above, three main strain scenarios can be identified. During the heating stage, positive strain starts appearing due to the elongation of the sample; during the curing process (stage exposed to a constant temperature) the strains remain constant; and during the cooling stage, the value of the tensile strains decreases with temperature. However, at a certain time, we can observe a change in the strain values from positive to negative (from elongation to shrinking). At this stage, the effect of the strain induced inside the sample during the curing process is observed. These results are in agreement with the predictions carried out before performing the test. It should be mentioned that only one major difference was found between the sample types. Through the curing process, the samples were heated to a certain temperature to activate the thermoset process and then cooled down. The samples with optical fibers in the top of the stack presented shrinkage strains earlier than the samples with the optical fibers embedded in the middle of the stack.

As mentioned before, the analysis presented above was performed by using the averaged value of strain along one of the fiber sections inside the sample to show the general dependence between temperature, time, and the induced strain. However, the strain measured along the fiber can also be visualized as functions of the curing time and the sample length providing a 3D map of the process. An example of this visualization is shown in Figure 7.

Figure 7.

3D Map of the strain distribution in the sample during the curing process.