Abstract
This study investigates how the interplay between fiber alignment and matrix ductility governs the tensile performance and failure modes of uniaxial fique-fiber composites. Two approaches were evaluated: using a flexible resin with higher failure strain and applying mechanical treatment to the fique fabrics. The baseline composite-manufactured with untreated fibers and a rigid polyester resin-exhibited premature failure driven by multiple matrix cracks. Replacing the resin with a more flexible matrix increased the tensile strength by 28% and suppressed matrix fragmentation, resulting in a single-fracture failure mode, as confirmed by computed tomography (CT). In contrast, mechanical treatment improved fiber alignment, raising both strength and stiffness by 76%. Although the treated composite still exhibited matrix fragmentation, it achieved a tensile strength of 111 MPa and an elastic modulus of 6.1 GPa, among the highest reported for commercially available fique fabrics. Additionally, in situ CT analyses during tensile loading revealed how matrix cracking governs the nonlinear mechanical response. These results demonstrate that controlling the failure mode and using standardized fique fabrics enables the development of natural-fiber composites with mechanical performance comparable to high-end commercial systems.
Introduction
The use of natural fiber-reinforced polymers in engineering applications has gained increasing attention in recent decades due to the growing demand for environmentally friendly materials that reduce greenhouse gas emissions. Natural fiber composites are currently used in diverse sectors such as aerospace, construction, sports equipment, food packaging, marine components, and particularly in the automotive industry.1–6 Their advantages over synthetic fibers include lower production costs, CO2 neutrality, abundant availability, biodegradability, recyclability, reduced health hazards during processing, low density, and reduced wear on manufacturing equipment.3,7 Moreover, certain natural fibers exhibit specific strength and stiffness that can approach those of glass fiber-reinforced composites.8,9
Common natural fibers used for composite reinforcement include jute, sisal, flax, hemp, kenaf, bamboo, and cotton.4,10,11 In this context, the present work focuses on fique, a natural fiber native to Colombia obtained from the Agave furcraea plant and traditionally used in food packaging, ropes, and sacks.12–14 Despite its promising mechanical characteristics for reinforcing thermosetting matrices, the industrial use of fique in structural composites remains limited. This limitation is largely due to the scarcity of comprehensive studies evaluating its mechanical performance and the absence of commercially standardized fique fabrics specifically designed for composite applications.
Table 1 summarizes the tensile properties of fique- reinforced composites reported in the literature. These studies evaluate different resin systems and assess whether the composite exhibited improved mechanical performance compared to the neat resin, indicating reinforcement efficiency. However, a strong dispersion in tensile strength and stiffness is consistently observed. This variability is commonly attributed to natural heterogeneities in fiber microstructure, diameter, and length, which directly affect load transfer and the overall mechanical response of the composite.13,15,16 An additional source of variability arises from the use of non-standardized raw materials: most available studies rely on manually produced fique fabrics or repurposed food-packaging sacks, neither of which are intended for structural reinforcement. As a reference, Table 1 also includes a commercial flax-fiber composite from Bcomp, 17 illustrating how standardizing a natural-fiber fabric can enable the development of composites with mechanical properties comparable to those of synthetic-fiber systems. This comparison highlights the unrealized potential of fique, which currently shows high variability not because of its intrinsic limitations but due to the absence of controlled, industrially produced textile architectures.
Mechanical properties of fique fiber composites under tension.
Unidirectional continuous fibers (UD). Disordered fibers of different lengths (MAT).
Although the mechanical properties of fique composites have been studied, their failure mechanisms remain poorly understood. To the authors’ knowledge, no prior research has systematically examined how the mismatch in failure strain and stiffness between the fique fibers and the resin governs damage evolution. This knowledge gap is critical: the lack of standardized textiles combined with uncharacterized failure modes has resulted in composites with inconsistent and generally low mechanical performance.
Early composite micromechanics studies21,22 demonstrated that different failure sequences can occur depending on the properties of the composite constituents. One failure mode occurs when the failure strain of the fibers is higher than that of the matrix, as is the case in fique fiber composites, as shown in the results section. For instance, when axial strain is applied during a tension test, the matrix can be damaged before the fibers, resulting in multiple fragmentations of the matrix until the fibers begin to crack, causing premature failure of the composite. In contrast, when synthetic fibers such as glass or carbon are used, the fibers, which have a much higher elastic modulus and lower failure strain than the matrix, govern the failure behavior. This results in a single fracture once the fibers reach their tensile strength, which is the preferred failure mode to maximize the mechanical properties of the composite.
The first failure sequence, involving multiple matrix fractures, can be predicted when the following inequality is satisfied,
where the subscripts
Given that fique fibers typically exhibit high elongation and low stiffness compared to polymeric resins, matrix fragmentation is a likely failure mechanism in these composites, according to the previous inequality. One strategy to enhance the composite overall strength is to modify the fiber properties. In Gomez et al., 23 a combination of mechanical and chemical treatments was applied to increase the strength and stiffness of fique fibers. Two methods were tested: immersion in a 5 w/v% NaOH solution for 24 h (chemical treatment) and applying a 1 N tensile load during soaking (mechanical treatment). The results of tensile tests on treated fibers are presented in Table 2, demonstrating that the mechanical properties of individual fibers can be improved. However, no tests were conducted on fique fabrics or composites.
Tensile test of individual fique fiber submitted to different treatments. 23
This study aims to improve the mechanical performance of fique fiber composites made with a standardized fabric by promoting a single-fracture failure mode instead of the typical multiple-cracking behavior. To achieve this, two strategies are explored: (i) increasing the deformation capacity of the resin, and (ii) reducing the failure strain of fique fabrics through specific treatments. The resulting composites were fabricated and tested under tension, and their failure mechanisms were investigated using computed tomography (CT), an approach that has not been previously applied to fique fiber composites. The results show that it is possible to shift from matrix fragmentation to a single-fracture mode, significantly increasing the composite strength and stiffness. Furthermore, in situ CT scans during tensile loading reveal how the failure mode governs the nonlinear mechanical response. Overall, this work demonstrates that by using a standardized fique fabric and controlling the failure mode, it is possible to produce a natural-fiber composite with mechanical performance comparable to synthetic-fiber composites and high-end commercial natural-fiber systems.
Materials and methods
This section details the strategies employed to enhance the structural performance of composites reinforced with fique fibers. These strategies leverage the high tensile strength of the fique fibers avoiding premature failure mechanisms, such as multiple matrix cracking. The following approaches were implemented:
Matrix Modification: The matrix, typically more brittle than the natural fiber, was modified by mixing a rigid resin with a flexible resin. This modification aims to increase the failure strain of the matrix, allowing it to deform without damage until the fiber reaches its maximum strength.
Fiber Treatment: Chemical and mechanical treatments were employed to enhance the fibers strength and stiffness, effectively reducing their failure strain. This approach aims to achieve the desired single fracture failure mode even when using unmodified rigid matrices.
The following subsections provide a detailed account of the materials used, their properties, and the processes followed for the fabrication and testing of the fique fiber composite laminates.
Fique fibers fabric
A custom-made fique fiber fabric, produced in collaboration with Coohilados de Fonce LTDA, was used as the composite material reinforcement. The fabric had 4 weft and 60 warp bundles per each 10 cm, with an areal weight of 470
Fique fiber fabric made by Coohilados (CO). The composites made with this fabric were loaded in the direction of the warp bundles, the principal direction.
Polyester resin
Two unsaturated polyester resins were used in the experiments. The first resin, Cristalan 1970 (referred to as RTM), was selected for its low viscosity, which improves the impregnation of the fique fabric. RTM stands for Resin Transfer Moulding resin, a low-viscosity formulation specifically designed for infusion and transfer molding processes. Cristalan 1970 is pre-accelerated with cobalt octoate and was mixed with 1wt% MEK peroxide catalyst, resulting in a gel time of approximately 20 min. A 3-h post-cure at 80°C was applied to all composite laminates.
The second resin, Cristalan 872, is a non-pre-accelerated isophthalic resin used to modify the properties of the RTM resin, specifically to achieve a failure strain greater than that of the fique fibers. Both resins were mixed at various weight ratios using a bench electric mixer at 1000 rpm for 15 min, followed by ultrasound degassing. The objective of the mixture was to provide sufficient elongation to prevent multiple matrix cracking in the composite. The optimal ratio leading to this behavior was 60% Cristalan 1970 and 40% Cristalan 872, referred to here as the flexible resin. Both resins were supplied by Andercol SAS. Table 3 summarizes their mechanical properties.
Mechanical properties of the resins.
Experimentally measured properties.
Composite fabrication
Fique fiber composites were manufactured using the vacuum bag method with manual pre-impregnation. Three rectangles (25 × 20 cm) of fique fiber fabric were cut, weighed, and stacked to create a panel with a thickness of 5.7 mm, satisfying the minimum ASTM D638-14 requirement (3.2 mm). 24 A fixed resin-to-fiber mass ratio of 3:1 was used for all panels to ensure consistent matrix content and impregnation quality across samples. 9 The pre-impregnation involved distributing resin uniformly over the fabric within a sealed plastic bag, using a spatula to promote uniform resin distribution before vacuum consolidation.
Once the fabrics were impregnated with resin, the vacuum bagging manufacturing method was employed using a flat glass surface as a tool. Instead of using the typical peel-ply/release-film/breather/vacuum-bag configuration for synthetic fiber composites, a regular bag sheet was placed on top of the peel-ply, followed by a steel plate and a breather ply (for more details on this process, refer to Vargas 9 ). This modification was necessary due to the low chemical affinity between the resin and the fibre,9,13,14 which tends to cause dry zones when the conventional vacuum bag method is used. In this setup, the breather and the steel plate distribute the pressure uniformly over the composite, ensuring consistent thickness throughout the panel. Subsequently, the bag was sealed with vacuum tape, and the composite material was kept for 24 h under a negative pressure of 16 in-Hg. Afterward, the panel was de-molded and post-cured in an oven at 80°C for 3 h. The composites produced using this method exhibited fiber volume fractions ranging from 37% to 45%, depending on the fiber/matrix configuration, calculated from the constituent weights and densities. Due to the inherent low compatibility between the natural fibers and the resin, full resin impregnation through the fabric architecture was not achieved, as confirmed by the CT scan analysis presented later. The measured porosity of the manufactured composites ranged between 10% and 19%, depending on the configuration. Nevertheless, despite the presence of internal porosity, the proposed manufacturing method produced panels with consistent fiber volume fraction, thickness, and porosity among the different laminates. This consistency indicates that the modified vacuum-bagging procedure provides a reproducible route for manufacturing fique-fiber composites, even when complete impregnation is limited by the intrinsic fiber-matrix compatibility.
Tensile test
Tensile test samples were cut according to ASTM D638 specifications using a CO2 laser cutter (Senfeng SF1390). 24 The cutting parameters were optimized to minimize thermal damage, using a power of 40 W, 10% intensity, and a cutting speed of 20 mm/s. After cutting, each specimen was visually inspected under a stereomicroscope, and any specimen showing signs of thermal damage or cutting-induced defects was discarded.
The tensile tests were performed on a Shimadzu AGX 100 kN universal testing machine equipped with a biaxial contact strain-gaged extensometer. At least five samples were tested for each configuration. The tests were conducted under displacement control at a rate of 3 mm/min.
Computerized tomography (CT) scans
CT scans were used to analyze the failure mechanisms and internal porosity of the composites. The scans were performed using a Nikon/Xtek HMX system with a 225 kV source and a Perkin Elmer XRD 1621 CN14 HS detector, achieving a voxel resolution of 10
Image analysis and porosity measurements were performed using ImageJ. For each composite configuration, porosity was quantified by analyzing at least 100 cross-sectional CT slices taken along the scanned volume. The images were converted to binary format using a thresholding procedure to distinguish solid material, including fibers and matrix, from air-filled voids. The porosity was then calculated as the ratio between the void area and the total cross-sectional area of each slice, and the reported values correspond to the average porosity obtained from all analyzed slices.
For the in situ experiments, tensile tests were carried out using a Deben CT 5000 stage. A step-scanning procedure was used, in which the load was held constant for 5 min before each CT acquisition to minimize relaxation effects. Figure 2 shows the CT scan planes discussed in later sections.
Planes and view scheme of the CT scans.
Optical microscopy
Optical microscopy was performed to examine the fiber-matrix interface and internal defects in the manufactured composites. Cross-sectional observations of untested specimens were carried out using a Nikon Eclipse LV100D optical microscope at 100× magnification. The analysis focused on identifying interfacial gaps, voids within the fique fiber bundles, and regions of incomplete resin impregnation, as well as assessing whether the use of a flexible resin or mechanically treated fabrics produced noticeable changes in the fiber-matrix interface.
Fique fibers modification
Based on the procedure reported in Gomez et al., 23 CO fique fiber fabrics were first immersed in a 5 w/v% NaOH solution for 15 min and then mounted in the Monsanto tensile machine for pre-tensioning. However, during the initial loading stage, the fabric began to unravel and separate into individual bundles, failing to reach even 5% of the targeted pre-tensile load. This behavior is attributed to the structure of the fique bundles: they consist of fibers of heterogeneous diameters held together primarily by friction-based mechanical anchoring resulting from twisting. The NaOH solution reduced this friction and acted as a lubricant, allowing the fibers within each bundle to slip relative to each other. Consequently, although the treatment in Gomez et al. 23 is effective for improving the properties of single fibers, it was not feasible to apply it directly to woven fabrics.
Given this limitation, we adopted an alternative approach consisting solely of a mechanical treatment applied to the uniaxial fique fabric. Rectangular samples (10 cm in length) were subjected to a constant tensile load of 2 kN for 24 h. Composites manufactured using these mechanically treated (MT) fabrics, following the same procedure described previously, incorporated three layers of treated fabric. The resulting panels exhibited a final thickness of 3.9 mm, representing a 31% reduction compared to composites made with untreated CO fabrics. This decrease is attributed to improved fiber-bundle alignment and reduced fabric waviness produced by the mechanical treatment.
Results and discussion
This section first evaluates the quality and repeatability of the manufactured composite laminates in terms of thickness, fiber volume fraction, and CT-based porosity. These measurements provide the basis for assessing the consistency of the modified vacuum-bagging process before analyzing the mechanical response. The tensile properties of the different composite configurations and neat resins are then compared to evaluate the effect of the two proposed modification strategies: increasing matrix ductility and improving fiber-bundle alignment. Finally, CT scan analysis is used to examine the failure mechanisms of each configuration, followed by an in-situ CT-based discussion of the role of multiple matrix cracking in the nonlinear tensile response.
Laminate quality and internal porosity
Before analyzing the tensile response, the quality and repeatability of the manufactured laminates were evaluated in terms of thickness, fiber volume fraction, and internal porosity. These measurements are summarized in Table 4. The CO/RTM and CO/Flexible composites exhibited similar fiber volume fractions, 38.3% and 37.5%, respectively, indicating that changing the resin did not substantially affect the reinforcement content under the manufacturing conditions used in this study. In contrast, the MT/RTM composite exhibited a higher fiber volume fraction of 44.7%. This increase is attributed to the reduction in laminate thickness produced by the mechanical treatment, from 5.7 mm for the untreated CO fabrics to 3.9 mm for the MT fabrics, despite using the same number of fabric layers. As shown later by the CT scan analysis, this thickness reduction is associated with improved fiber-bundle alignment and reduced fabric waviness.
Laminate thickness, fiber volume fraction, and CT-based porosity of the manufactured composites.
The CT-based porosity measurements show that the CO/RTM and CO/Flexible composites had similar void contents, 9.46% and 10.80%, respectively. This suggests that, under the manufacturing conditions used in this study, changing the resin did not substantially reduce the porosity level. In contrast, the MT/RTM composite exhibited a higher porosity of 19.72%. This increase is likely associated with the change in fabric architecture induced by the mechanical treatment. The original waviness of the untreated woven fabric may help retain resin during vacuum bagging, whereas the improved alignment of the mechanically treated fiber bundles may facilitate resin flow away from the reinforcement. Given the reported low chemical affinity between fique fibers and polymeric matrices,13,14,23,26 these results suggest that both fabric architecture and fiber-matrix compatibility contribute to the residual porosity observed in the composites.
To further examine this interpretation, optical microscopy was used to assess the fiber-matrix interface and the location of voids in untested specimens. Figure 3 shows cross-sectional images of the CO/RTM, CO/Flexible, and MT/RTM composites. In all configurations, interfacial gaps were observed between the fique fibers and the resin, as indicated by the red arrows. This confirms that neither the flexible resin nor the mechanical treatment substantially improved fiber-matrix adhesion under the manufacturing conditions used in this study. Therefore, load transfer is likely governed not only by chemical adhesion, but also by mechanical contact and friction between the resin and the fiber bundles. The microscopy images also reveal air voids within or between fique fiber bundles, as indicated by the blue arrows in Figure 3. These voids are located in regions where the resin did not fully penetrate the internal bundle structure, consistent with incomplete impregnation and the CT-based porosity measurements. Consequently, improving the fiber-matrix interface and resin penetration into the bundles remains essential for further increasing the mechanical performance of these composites.
Optical microscopy images of cross sections from untested composite specimens: (a) CO/RTM, (b) CO/Flexible, and (c) MT/RTM. Red arrows indicate fiber–matrix interfacial gaps associated with poor adhesion and incomplete impregnation, while blue arrows indicate air voids.
Tensile properties of fique-reinforced composite
Figure 4 shows the tensile stress-strain curves for the original composite made with untreated fique fibers (CO) and RTM resin, alongside the proposed composites made with CO fibers and flexible resin, and the composite made with mechanically treated fibers and RTM resin. The tensile test results for both the unreinforced RTM and flexible resins are also included for comparison. Table 5 summarizes the average and standard deviation of the mechanical properties for all tested materials. With the laminate quality and repeatability established, the tensile response of each composite configuration can be compared in terms of the two modification strategies proposed in this work.
Tensile stress-strain curves of the composites CO/RTM, CO/Flexible, and MT/RTM; and the resins RTM and flexible.
Mechanical properties of composites and neat resin.
The first strategy aimed at modifying the composite’s failure mode by increasing the elongation to break of the matrix. The unreinforced flexible resin exhibits a failure strain of 5.1%, roughly three times that of the RTM resin, and close to the strain at failure of fique fibers (Table 2). While both resins show similar strength, the flexible resin has a lower modulus (1.65 GPa), which is reflected in the mechanical response of the corresponding composites. Incorporating this resin increased composite strength by 28% compared to the CO/RTM system. No major difference in initial stiffness was observed; however, the CO/Flexible composite retains stiffness beyond 1% strain, unlike CO/RTM, which shows significant softening. These results indicate that increasing the resin failure strain promotes a transition in the composite’s failure mechanism, as later confirmed by CT scans.
The second strategy focused on reducing the failure strain of the fiber architecture through mechanical treatment. The MT/RTM composite showed a substantial improvement: its tensile strength and stiffness increased by 76% compared to the CO/RTM composite. These gains highlight the strong influence of fiber-bundle alignment on composite performance. To distinguish whether the improvement stemmed from changes in fiber properties or from improved fabric configuration, tensile tests were performed on dry (unimpregnated) CO and MT fabrics.
For these tensile tests, the cross-sectional area of the fiber bundles was measured using CT scan images. Based on 30 measurements, the bundle area averaged 0.97 ± 0.08 mm2. The dry fabric tensile tests were performed on samples consisting of a single layer of CO fique fabric containing eight unidirectional fiber bundles. Prior studies6,9 have shown that the sample length significantly influences the mechanical properties of fique fibers during testing. Thus, specimens in this study were prepared with a consistent length of 115 mm, matching the length used for the composite fabrication.
Figure 5 presents the stress–strain curves of dry fiber bundles. The mechanical treatment does not enhance the intrinsic tensile properties of the fibers; rather, strength marginally decreases from 143 MPa to 133 MPa. Thus, the significant improvements observed in the MT composite can be attributed primarily to enhanced bundle alignment and reduced fabric waviness, rather than changes in fiber quality. Since the failure strain of the fibers remains unaffected, the occurrence of multiple matrix cracking is expected and further examined in the next section.
Tensile test results of the dry fiber bundles with and without mechanical treatment–MT.
Importantly, the strength and stiffness achieved by the MT/RTM composite are comparable to those reported for manually combed uniaxial fique reinforcements, 9 which represent the highest-performing systems currently available. Unlike manually combed fabrics—whose labor-intensive fabrication limits their commercial viability—the mechanical treatment proposed here can be applied directly to commercially produced woven fabrics, enabling high performance with low variability, as reflected by the small standard deviations in Table 5. This demonstrates that optimizing standardized fabrics can deliver natural fiber composites with mechanical performance approaching that of synthetic-fiber composites, without sacrificing scalability.
CT scans failure analysis
Composites with RTM rigid matrix and untreated fique fabrics (CO)
The CT scans in Figure 6 show multiple views of the failed composite sample manufactured with the rigid RTM matrix and untreated CO fique fabrics (see Figure 2 for the CT scan views scheme). These images establish a baseline for evaluating the effects of matrix modification or fiber treatment on the mechanical behavior of the material.
CT scans of the tested CO/RTM composite. The front view of the specimen is shown in (a) at 1 mm and (b) at 3 mm deep from the surface, the lateral view in (c), and the top views in (d) and (e).
Figures 6(a) and (b) are close-ups of the front view failure zone at different depths. The red circles highlight multiple matrix cracks in the resin-rich zones, which form due to the waving of the fiber bundles of the fabric. Additionally, Figure 6(b) shows that some fique fiber bundles remain intact within the failure zone, indicating partial fiber continuity post-failure.
In the lateral view shown in Figure 6(c), continuity in some fiber bundles within the fault zone is also visible. The green dotted lines in this figure indicate the slice planes of the top view CT scans of Figures 6(d) and (e). These cross-sectional views reveal that the cracks propagate around the fiber bundles, leading to fiber/matrix interface debonding and limiting effective load transfer. The CT scan sequence of the top view also reveals that certain fiber bundles, such as those marked in blue in Figure 6(e), retain their shape without discontinuity or fracture, even within the specimen’s failure zone.
The results indicate that a matrix with a lower failure strain than the fibers promotes multiple matrix cracks. These cracks then propagate through the fiber/matrix interface and create stress concentrations. As a result, some fiber bundles fail prematurely, causing the composite to fail, while other bundles remain undamaged.
Composites with flexible matrix and untreated fique fabrics (CO)
Figure 7 shows different views of the CT scans of the failed specimen made with flexible resin and untreated CO fique fabrics. In Figure 7(a), the front view away from the failure zone reveals the absence of matrix cracking in the composite. Subsequently, Figure 7(b), which displays the front view in the failure zone, demonstrates that using flexible resin results in a single fracture, as there is no continuity in the fiber bundles, unlike the composite made with RTM resin.
CT scans of the tested CO/Flexible composite. The front view of the specimen away from the failure zone is shown in (a) and in the failure zone in (b) at 3 mm deep from the surface, the lateral view in (c), and the top views in (d) and (e).
In the lateral view of the CT scans, Figure 7(c), red markings highlight the complete fracture of the fiber bundles. This suggests that using a higher failure strain resin allows for the full exploitation of the mechanical capacity of the fibers, effectively transferring loads from the matrix to the fibers until their ultimate failure strength is reached.
Furthermore, Figures 7(d) and (e) show that there is minimal loss of adhesion at the fiber/matrix interface, maintaining effective load transfer up to the point of single fracture. The fracture in the specimen is localized, with no crack propagation throughout the material, as evidenced by intact fiber bundles, such as the one marked with a blue circle. This behavior contrasts with that of the CO/RTM composite, as shown in Figure 6, where the cracks generated on the matrix propagate across the interface with the bundles, leading to fiber/matrix separation.
Composites with reinforced mechanically treated fabrics (MT) and RTM matrix
Figure 8 shows CT scans of the composite made with mechanically treated (MT) fibers and RTM resin. The red markings in Figures 8(a) and (b) indicate that this composite configuration still exhibits multiple matrix cracks. This behavior was anticipated, as the mechanical treatment of the fibers did not alter their inherent mechanical properties. Therefore, the failure strain of the fibers remains higher than that of the matrix, resulting in an undesired failure mode. As shown in Figure 8(d), the matrix cracks propagate along the fiber/matrix interface, inhibiting effective load transfer and potentially reducing the composite’s strength and stiffness.
CT scans of the tested MT/RTM composite. The front view of the specimen is shown in (a) at 1 mm and (b) at 3 mm deep from the surface, the lateral view in (c), and the top views in (d) and (e).
Despite not achieving a single-failure mode with MT fibers, this composite demonstrated a significant increase in strength and stiffness. This improvement can be attributed to the 24-h pre-loading of the fiber fabric, which aligned the fiber bundles along the loading direction (Figure 8(c)) and reduced resin-rich zones (Figure 8(a) and (b)). This is evidenced in the frontal CT views at different depths, with the MT fique fabrics, all fiber bundles appear continuously across the specimen, whereas with the untreated CO fabric, fiber bundles appear discontinuously due to waviness, as shown in Figure 7(a).
Marked in blue in Figure 8(a) and (e), it is evident that with the mechanical treatment, the weft bundles (fibers perpendicular to the loading direction) curve, enabling improved alignment of warp bundles (fibers parallel to the loading direction) along the loading axis, thereby reducing the composite thickness. In contrast, with the untreated CO fabric, the weft bundles remain straight while the warp bundles curve, creating resin-rich areas and increasing the composite’s thickness.
This analysis explains why composites made from MT fibers exhibit a significant increase in strength and elastic modulus, even with the occurrence of multiple matrix cracking, compared to composites made with untreated CO fabrics. The results of this study suggest that combining both proposed approaches—mechanical treatment of the fique fabrics and incorporation of a flexible resin—could further enhance the composite’s strength and tensile stiffness. This combined approach has the potential to produce a single-fracture failure mode in a composite where the fibers are aligned in the loading direction, maximizing load transfer and overall mechanical performance.
Although complete impregnation was not achieved for all composite configurations, as evidenced by the high porosity measured in the CT scans, the relatively low standard deviations in thickness, porosity, and mechanical properties indicate that the manufacturing process was repeatable (Tables 4 and 5). This is particularly relevant for natural-fiber composites, where larger variability is commonly expected due to the intrinsic heterogeneity of the fibers and the challenges associated with fiber-matrix impregnation. Therefore, while the modified vacuum-bagging procedure did not eliminate void formation, it produced composite panels with consistent internal quality across the different configurations.
Non-linear behavior
The stress-strain curves for the fique fiber composites presented in previous sections exhibit non-linear behavior under tensile loads, characterized by a reduction in stiffness as the applied force increases. The primary cause of this non-linear response lies in the inherent non-linear elasticity of the composite constituents. This was demonstrated in tensile tests on both the RTM and flexible resins (Figure 4), where both polymers exhibited non-linear responses. Additionally, previous tensile tests on fique fibers, as reported in Gomez et al., 23 Teles et al. 27 and shown in Figure 5, confirm a similar non-linear behavior. The effect of large deformations may also influence the material response. 28
A second factor contributing to the non-linear behavior occurs in fique fiber composites that experience multiple matrix cracking during tensile testing. According to the model by Aveston et al., 22 once matrix cracking initiates, the fibers begin to carry the full load of the composite, leading to a reduction in the composite’s overall stiffness. To investigate the correlation between stiffness changes and matrix cracking, as proposed by this model, an experiment was designed.
This experiment involved performing a tensile test with interruptions at specific load points, enabling in situ CT scans to be captured without releasing the load. The test was conducted on a CO/RTM specimen where multiple matrix cracking was anticipated. This experiment was used not only to identify the load level at which matrix cracking begins, but also to relate the damage evolution to pre-existing manufacturing defects, including voids and fiber-matrix interfacial gaps. Figure 9 illustrates the selected points of interest (load interruption points) on the stress-strain curve of the CO/RTM specimen. For comparison, this figure also includes the CO/flexible composite results, which do not exhibit a multiple matrix cracking behavior. The first interruption point (1) at 900 N marks the end of the material’s linear elastic behavior. At the second point (2), at 2500 N, the material’s stiffness stabilizes after an initial reduction, revealing a divergence between the CO/RTM and flexible resin composites. The third point (3), at 4000 N, was taken near the material’s ultimate strength.
Tensile stress-strain response of the CO/RTM composite showing the load interruption points selected for in situ CT scanning at 900 N, 2500 N, and 4000 N. The CO/Flexible response is included for comparison.
Figure 10 shows the front-view CT scans of the CO/RTM specimen at each load interruption during the tensile test. At the first interest point (1), the scan shows no visible crack formation, corresponding with the composite’s linear elastic behavior. However, the specimen already contains internal voids associated with incomplete impregnation during manufacturing, as quantified from the CT-based porosity measurements reported in Table 4. By the second interest point (2), the red-circled regions reveal the onset of matrix cracking, primarily around these voids, which act as local stress concentrators. At the final interest point (3), as tensile loading progresses, horizontal cracks in the matrix increase in number and extend through resin-rich regions and along the fiber-matrix interface, while the fibers continue to elongate.
Front view of the same CO/RTM specimen at different loads 900 N, 2500 N, and 4000 N.
The in-situ experiment results confirm that the reduction in stiffness observed in the CO/RTM composite, compared to the flexible resin composite, can be attributed to the initiation and propagation of matrix cracks at interest point (2). In the flexible resin composite, the non-linear response is primarily due to the tensile response of its constituents. However, in the CO/RTM composite, the presence of voids and imperfect fiber-matrix impregnation promotes premature matrix cracking, which significantly amplifies this non-linear behavior. The propagation of cracks along the fiber-matrix interface further suggests incomplete adhesion between the hydrophilic fique fibers and the resin, limiting effective load transfer once damage initiates.
These observations show that the nonlinear response and failure of the CO/RTM composite are not only controlled by the intrinsic nonlinearity of the constituents, but also by manufacturing-induced defects such as voids, resin-rich zones, and imperfect fiber-matrix bonding. Therefore, the in-situ CT scans underscore the importance of reducing porosity and improving fiber-matrix impregnation in fique-fiber composites. Such improvements would reduce local stress concentrations, delay matrix cracking, and promote more efficient fiber utilization before final failure.
Conclusions
This study demonstrates that the mechanical performance of fique fiber composites can be substantially improved by controlling their failure mode, shifting from the typical multiple matrix cracking to a more desirable single-fracture behavior. Two complementary strategies were evaluated: increasing the deformation capacity of the resin and enhancing fiber alignment through mechanical treatment.
Using a flexible resin with a failure strain comparable to that of the fique fibers successfully prevented the onset of matrix fragmentation. CT imaging revealed that this resin promoted a more uniform load distribution across the fiber bundles, suppressing premature local overloads and enabling the composite to reach higher stresses before failure. This approach increased the composite strength by 28% relative to the reference CO/RTM configuration, confirming that matrix ductility is a key parameter for achieving a single-fracture failure mode.
The mechanically treated (MT) configuration resulted in even more significant improvements. The MT composite reached a strength of 111 MPa and an elastic modulus of 6.1 GPa, the highest values reported for commercially available fique fabrics. Although this treatment did not eliminate matrix cracking, CT analysis revealed that the improved fiber bundle alignment reduced waviness and enhanced load transfer efficiency, highlighting alignment as a dominant factor in stiffness and strength optimization.
The chemical treatment described in Gomez et al., 23 while infeasible for whole fabrics due to fiber bundle separation, remains promising at the individual fiber scale. If adapted to preserve fabric integrity, it could further increase composite strength and potentially enable single-fracture behavior even when using rigid matrices.
Overall, this work demonstrates that high-performance natural-fiber composites can be fabricated using standardized, commercially available fique fabrics when failure mode is properly controlled. By combining a ductile matrix with improved fiber alignment, it is possible to achieve mechanical properties that rival those of synthetic-fiber composites and high-end natural-fiber systems. These results suggest that a composite combining mechanically treated fique fabrics with the flexible resin could provide further improvements by simultaneously enhancing fiber alignment and matrix deformation capacity. Therefore, the MT/Flexible configuration represents a direct continuation of this work and should be evaluated in future studies. These findings provide a scalable route to enhancing the structural performance and reliability of sustainable composites, and they open the door to future improvements through optimized chemical treatments and advanced textile architectures.
Footnotes
Acknowledgements
The authors acknowledge the µ-VIS X-ray Imaging Centre at the University of Southampton and the NXCT for their technical support and access to CT scanning facilities. The authors also thank the Universidad Nacional de Colombia for support through the LTDM Laboratory.
Handling Editor: Humberto Almeida Jr
Author contributions
Santiago Marin: Conceptualization, Investigation, Methodology, Writing – original draft.
Johnattan Vargas: Investigation, Methodology.
Guillermo Idarraga: Supervision, Writing – review and editing.
Meisam Jalalvand: Supervision, Funding acquisition.
Juan Meza: Supervision, Writing – review and editing, Funding acquisition.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the Royal Academy of Engineering through the Transforming Systems through Partnership programme (TPS1312).
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.