Sage Journals: Discover world-class research

Abstract

The main objective of this study is to characterize the folding strength of a thermoplastic polypropylene matrix composite reinforced with unidirectional (UD) flax fibers. Composite plates were produced by thermocompression technique, then cut into specimens and subjected to a thermo-folding shaping process. Mechanical tests instrumented by acoustic emission (AE) were carried out to identify and quantify the damage mechanisms leading to fracture. A multivariate analysis, based on the k-means algorithm, was used to distinguish damage modes. The AE recorded signals from bio-composites are categorized into signal clusters, allowing the attribution of damage mechanisms such as matrix cracking, interfacial decohesion or delamination, and fibre breakage. The chronological sequence of AE events revealed that matrix cracking occurs first, followed by fiber–matrix debonding, with fiber breakage appearing later. Matrix cracking persists until failure, while debonding and fiber breakage dominate in the post-failure phase. The combination of these three mechanisms ultimately leads to specimen rupture. These findings provide new insights into the progressive damage sequence in UD flax/polypropylene composites under bending, highlighting the potential of AE combined with clustering analysis to monitor and understand thermoformed bio-composite failure mechanisms.

Keywords

Bio-based composites polypropylene (PP)acoustic emission (AE)flax fibers multivariable analysis folding strength

Introduction

Due to the worldwide increase in the exploitation of raw materials to fulfill industrial needs, the use of bio-based composite materials is emerging as a highly promising alternative solution. Bio-composites, which consist of a polymer matrix reinforced with natural fibers, are used as an eco-friendly alternative to replace composites with traditional fibers. Natural fibers offer several advantages, including low density, high specific strength, improved energy recovery, CO₂ neutrality,¹ ease of processing and recycling, biodegradability, and low cost.^2–5

One of the primary challenges in applying bio-composites materials to structural components is the complexity of damage mechanisms. Unlike metallic materials, composites can experience multiple interacting failure modes, leading to a rapid and progressive decline in mechanical properties.^6,7 This complexity often makes it difficult to accurately predict their behavior and lifespan expectancy under load. Furthermore, certain defects may inevitably arise during the manufacturing process of bio composites, such as insufficient reinforcement, fiber misalignment, inclusions, and porosities, which can have detrimental effects when subjected to service loadings.

The literature review presents numerous Non Destructive Techniques (NDT) to better characterize the mechanical behavior and failure mechanism of composites materials including: SEM (Scanning Electron Microscopy)⁸; Ultrasound waves (US)⁹; Acoustic Emission (AE)¹⁰; Electrical resistance,¹¹ Computed Tomography (CT-scan)¹² and Digital Image Correlation (DIC).¹³

Among non-destructive testing (NDT) approach, acoustic emission (AE) is a passive, real-time monitoring technique for structural health monitoring (SHM). It has proved particularly effective in detecting and monitoring damage and failure in a wide variety of composite materials.^14–16 This technique is based on the analysis of AE signals generated by the propagation of cracks within the material under load, providing valuable insight into the presence, location and nature of damage. To provide context for this research, the following section presents a literature review on the use of AE techniques for characterizing damage and failure modes in bio-composite materials.

The literature review on the use of the AE technique for damage and failure characterization of bio-based composites is abundant.^17–20 Most of the research works in the literature utilizes two main approaches for analyzing data from acoustic instrumentation: single-parameter analysis²¹ and multivariable analysis.²² AE descriptors can include temporal, frequency, and time-frequency characteristics which can be extracted directly from the temporal signal or its frequency spectrum.²³

Haggui et al. (2019)²⁴ examined the mechanical behavior of thermoplastic composites reinforced with flax fiber under static and fatigue tensile loading. They attached to the specimens two piezoelectric sensors with a wideband to real-time monitor the mechanical behavior of the bio-composite. The AE signals analysis using k-mean clustering algorithm reveal four distinct classes of acoustic events during testing to failure, each corresponding to a specific damage mechanism: matrix micro-cracking, fiber-matrix debonding, fiber pull-out, and fiber breakage.^25,26 Czigany (2004)²⁷ used AE method to investigate crack propagation sensitivity in flax fiber-reinforced polypropylene composites with different fiber and moisture content. The author identified a correlation between the number of acoustic events occurring during SEN-T tests (Single-Edge Notched Tensile) and the fracture toughness. The four Micro30 piezoelectric sensors, with a frequency range of 100–600 kHz, were attached to the sample to localize the geometric positions of the acoustic signals. The planar projection of AE events reveals that, for wet specimens, the damage zone is larger compared to dry specimens, as the material’s strength is reduced due to the formation of microcracks. In a recent study, Habibi and Laperrière (2023)¹³ combined digital image correlation (DIC) with AE to locate and classify the types of damage occurring in flax fibre composites pre-impregnated with CORAL bio-based resins and tested in bending. The number of hits, their amplitude and the associated energy were used as descriptors to analyze the acoustic activity and identify the damage mechanisms. The damage was analyzed through a detailed interpretation of the acoustic emission signatures and k-means classification of the acoustic events. The evolution of each damage mechanism was correlated with the applied load and expressed as a function of the loading rate to emphasize the impact of the stacking sequence. In another study, Habibi et al. (2017)²⁸ performed tensile and flexural tests coupled with AE on composites reinforced with short flax fibers, which were produced using a papermaking process. The results indicate that, in comparison to tensile tests, AE events during flexural tests occur at much higher strains and with significantly lower accumulated energies, suggesting a lower incidence of AE events associated with matrix microcracking. They concluded that AE analysis provided a comprehensive overview of the failure behavior and enabled the identification of the dominant failure modes during tensile and flexural loadings by monitoring the evolution of the cumulative AE energy. Multivariable analysis of the AE data highlighted the clear dominance of fiber-matrix friction and fiber pull-out as the primary fracture modes, emphasizing the significance of the low adhesion between flax fibers and the epoxy matrix.

In recent research, Nasri et al. (2023)²⁹ examined the effects of UV-accelerated aging, under both dry and moist conditions, on the mechanical performance and damage mechanisms of a bio-composite material composed of a polypropylene matrix reinforced with flax and pinewood fibers. Damage mechanisms were monitored during tensile tests with the aid of two micro-80 sensors, which have a wideband frequency range of 100–1000 kHz. Analysis of AE energy and the number of AE-events over show that both materials exhibited linear elastic behavior with no acoustic activity being recorded until the behavior became nonlinear viscoelastic. Overall, AE analysis revealed that the main cause of the deterioration in mechanical properties was micro-cracking (contributing to 55.4% for PP-flax bio-composite deterioration), which resulted from the breakdown of polymer chains (i.e., photo-oxidation). The acoustic signals are significantly influenced by the propagation distance between the source and the sensor, owing to their inherently dispersive nature, which strongly affects their temporal characteristics.³⁰ This has led researchers in recent studies to analyze AE signal waveforms in both the frequency³¹ and time–frequency domains.³² In recent study, Barile et al. (2019)³³ demonstrated that the Continuous Wavelet Transform (CWT) effectively analyzes AE signals in carbon fiber reinforced polymer (CFRP) by distinguishing propagation modes and capturing time–frequency features. Using CWT, they successfully identified and separated different propagation modes of AE waves in CFRP laminates, providing valuable insights into dispersion behavior and wave attenuation.

Thermoplastic matrix composites are increasingly used in various industrial sectors, due to their recyclability and fire-resistance properties.³⁴ Further, their shaping by thermoforming is a major advantage for the production of parts with complex geometries.³⁵ In the field of material processing, thermo-folding is a promising technique for shaping bio-composite materials, into complex geometries.³⁶ This process is gaining traction due to its potential to create lightweight, durable structures with intricate shapes, offering advantages such as recyclability and reduced environmental impact. Several studies in the literature have investigated the shaping feasibility of bio-based composites using moulds or autoclave processing.^36–38 Building on these insights, our study focuses on the development of an experimental protocol for shaping bio-based composites via thermo-folding, without the need for molds or autoclave equipment, and providing a practical and accessible approach for bio-composite fabrication. The studies and tests have shown that folding is feasible, provided that temperature control of the material is properly managed. The observed damages are similar to those seen during the thermocompression, particularly the amalgamation of materials within the fold. The pressure applied to the specimen at the bending point helps to prevent the delamination of the layers.

However, this process can alter the structural performance of parts by generating undesirable effects within the material. It is therefore essential to develop an optimized manufacturing process that minimizes these effects while ensuring the required mechanical properties. In the first instance, this involves establishing a correlation between the process parameters and the mechanical performance obtained, so as to enable progressive optimization in line with the desired objectives.

The above-mentioned literature review highlights the limited number of quantitative investigations focusing on the characterization of polypropylene composites reinforced with flax fibers shaped by thermoforming process using the AE technique.

This study aims to explore the potential of a multivariate statistical approach, based on an unsupervised k-means clustering algorithm applied to extracted AE waveform parameters, for identifying damage initiation and underlying mechanisms in thermoplastic matrix reinforced with flax fiber. The folding response of the specimens was evaluated through bending tests, with a monitoring of failure progression using AE. The AE signals were analyzed with unsupervised k-means clustering algorithm, to identify and quantify the damage modes leading to failure. The novelty of this research lies in comprehensive characterisation of damage mechanisms and the folding strength of thermoformed bio-composite. This information’s can be used to optimize the mechanical properties of the composite components, matrix, fiber and interfaces, enabling the design of bio-composites into desired shapes for advanced engineering applications. This approach can improve the overall design of bio-composite materials to meet increasingly complex engineering requirements.

Materials and experimental methods

Materials and samples preparation

Manufacturing of experimental samples

The technical fabric is elaborated by DEPESTELE group which provides flax, including woven fabrics, made of comingled fibers and PP filaments ( $440 g / m^{2}$ ).

The thermocompression molding has been used to get a plate of 350 × 350 [mm²], dimensions defined by the molds. The stacking sequence adopted in this study was [0/−45/−45/0/45/90/90/45]. A thermocompression press was used (Fontijne Grotnes TPC 321) to cure the plate in 3-stage process: heating, curing and cooling,³⁹ according to the thermal and pressure cycles illustrated in Figure 1.

Figure 1.

Curing cycles for thermocompression process showing heating, consolidation and cooling stages.

According to the classical lamination theory, the sequence involved is suitable for folding and traction solicitations. Due to the asymmetric configuration of the plies, the bending stiffness matrix [B] was non-zero. This led to additional couplings between in-plane normal stresses and shear strains, as well as between bending moments and twisting curvatures. Then, the plate was cut to get test samples, by a laser cutting with the dimensions of 100 × 15 × 3.4 [mm³], which refers to rectangular samples (Figure 2) specified in the ISO 527 - 4 standard.⁴⁰ Sample homogeneity is influenced by both the manufacturing process and the fiber type.³⁹ To minimize the effect of processing variability on the analysis, all specimens were obtained from the same composite plate. The three specimens tested in this study were manufactured and thermoformed under the same conditions. They were selected on the basis of X-ray tomography results, showing no significant visible damage in the inner folding zone.

Figure 2.

(a) Cutting plate with laser (b) Composite samples.

Folding-based shaping of test specimens

A three-points bending test is a process that involves folding stresses on the material,⁴¹ the inner fold is compressed while the outer fold is stretched. A preliminary work has revealed the main issue of this technique: some fractures are located in the outer fold. This default originates from a lower ultimate elongation of the flax (2.7–3.2%) than the matrix (200–600%).⁴² Naturally reinforced by elementary fibers, the stiffness of the flax is due to both a low Microfibrillar Angle relative to the axis of the fiber (MFA) and a high cellulose content.^43,44 To reduce its rigidity, a heat input by the tooling is essential. A parametric study was carried out to determine the optimum thermo-folding process factor. The experimental bending equipment used for thermo-folding is illustrated in Figure 3. The specimens are heated not only by the die, but also by a flow of hot in to prevent heat loss.

Figure 3.

(a) Schematic representation of the experimental device (b) Shaping experimental set-up (c) Folded sample.

The main stages in the thermo-folding shaping process, and the parameters taken into account in this study, are outlined in three key phases below:

Preheating stage: the material is preheated for 30 minutes to ensure that the core reaches the target temperature of 160°C. The mold (punch and die) is thermally regulated through feedback system: thermal cartridges embedded in the mold are controlled based on real-time measurements of the tooling temperature using a PT100 probe.

The tooling surfaces were preheated to 145°C to prevent overheating of the specimen, and the punch was lowered at a constant speed. Once the composite was positioned between the punch and the die, tool heating was discontinued to allow cooling of the composite to 100°C. This sub-step lasted 6 minutes to ensure complete solidification of the polypropylene matrix (≈120°C). The feasibility of the fold was evaluated under low pressure (<0.5 bar), which was sufficient to guarantee material recohesion. Subsequently, the punch was returned to its initial position. The air projected onto the composite was maintained between 140°C and 155°C. The average thickness of the three tested specimens, measured both with a caliper and confirmed by imaging, are 4.1, 3.85 and 4.1 mm respectively for samples A, B and C. The three specimens examined in this study were fabricated under the identified optimal conditions and selected based on X-ray tomography results, showing no significant visible damage in the inner folding zone.

Mechanical loading and experimental set-up

The mechanical functionality is tested based on the mechanical strength of the fold. The procedures to define the strength of a 90° bend in a composite material are currently not usual.⁴⁵ Potential applications of this composite involve a maximum of stresses along the outer folds. To meet the needs of the current case study, a simple experimental system was designed to mechanically stress the external curvature in tension and the inner in compression (fold closure). This destructive test can be schematized as shown in Figure 4.

Figure 4.

(a) Experimental instrument with sample deflection, (b) Folding strength measurement protocol, (c) Specimen after mechanical testing. # The ruler is only shown to provide a visual reference of scale.

Acoustic emission instrumentation

During mechanical testing, AE signals were recorded using a Euro Physical Acoustics (EPA) system. Two piezoelectric transducers (PCA miniature Nano-30) were employed, having a resonant response at 300 kHz and a reliable frequency response in the range of 125–750 kHz. The sensors were labelled 1 and 2 according to their respective acquisition channels. They were attached to the specimens on the same side using a PVC mounting bracket, with a coupling agent (silicon grease) applied between the sensors and the specimens to improve coupling efficiency and minimize signal attenuation at the transducer-sample interface. The sensors are spaced 25 mm from each other. The amplification of EA signals was provided by an IL40S preamplifier with a gain factor set to 40 dB. The amplitude of the acquisition threshold (set at 41 dB) and the temporal acquisition parameters were calibrated using the Pencil Lead Breaking (PLB) protocol.⁴⁶ The acoustic emission signals are recorded simultaneously with the bending test.

Results and discussions

Mechanical properties

Figure 5 shows a representative force-displacement of flax-reinforced polypropylene under bending loading conditions until failure. It can be seen that the curves can be roughly categorized into three main stages. The curves exhibit an initial linear elastic region [0 to 3 mm], followed by a progressive increase in force as the material undergoes plastic deformation and damage [from 3 mm of displacement until reaching the maximum force value]. The peak force (F_max) values indicate the maximum load-bearing capacity of each specimen before failure. The differences in peak forces suggest variations in material mechanical response, likely caused by defects introduced during forming⁴⁷ and thickness variations, which lead to differences in local stress concentrations.⁴⁸ In future work, instrumenting the specimen during the thermoforming process could provide valuable insights into the material’s behavior and further enhance the analysis.

Figure 5.

Representative force-displacement curves of flax-reinforced polypropylene composite under bending loading.

Damage mechanisms analysis

Figure 6 depicts force-displacement curves for the three tested specimens, along with the corresponding AE activity. The AE activity curves follow a similar overall trend, demonstrating both good repeatability and reproducibility of the AE response. This reproducibility underlines the reliability of the experimental observations and confirms that the damage evolution detected by AE is consistent with the mechanical response. The AE activity is very low during the initial elastic stage. Notably, the peak AE activity, representing the maximum number of hits, consistently occurs after the force peak (F_max). This delay can be explained by the damage accumulation process: as the material reaches its maximum resistance, microcracks and fiber breakages start to propagate more rapidly. After the peak force, the structure begins to lose its load-bearing capacity, leading to intensified internal damage. This post-peak damage phase, characterized by matrix decohesion, fiber pull-out, and crack coalescence, generates a surge in AE events.

Figure 6.

AE activity evolution in terms of number of hits and Force-displacement curves for (a) specimen ‘A’ and (b) the three tested specimens.

Changes in mechanical properties are frequently associated with changes in damage modes. This means that knowing only the mechanical properties and overall degradation is not enough to provide an in-depth understanding of damage mechanisms. To identify the different failure modes of bio-composites, several methods have used single AE parameters to classify the AE signals.^49,50 In materials with very high damping properties, such as polypropylene reinforced with flax fibers, classification based solely on a single AE parameter often results in significant overlaps between failure mechanisms.⁵¹ Therefore, it is crucial to perform multivariable analysis of AE signals to assess the various damage mechanisms and monitor their evolution over time until reaching the final global failure. Given the consistency of the acoustic results and to simplify the analysis, the following figures refer exclusively to the specimen noted ‘A’.

The next sections aim to use the key characteristics of AE signals, such as: amplitude, duration, counts, counts to peak and the signal energy for clustering analysis. k-means clustering is a widely used algorithm in unsupervised machine learning due to its simplicity and efficiency in partitioning data into distinct clusters based on their similarities.⁵² The Elbow method and Davies-Bouldin index, a key step in k-means clustering algorithms, were applied to determine the optimal number of clusters (k values). Elbow method utilizes the Within-Cluster Sum of Squares (WCSS), which quantifies the total variation within a cluster. WCSS for each cluster was computed using the following equation:

W C S S = \sum_{p_{i}, c l u s 1} d {(p_{i}, C_{1})}^{2} + \sum_{p_{i}, c l u s 2} d {(p_{i}, C_{2})}^{2} + \sum_{p_{i}, c l u s 3} d {(p_{i}, C_{3})}^{2}

(1)

where

\sum_{p_{i}, c l u s 1} d

represents the sum of the squared distances between each data point and its centroid within a cluster, similarly, applied to the other two terms in the formula above. k-means clustering was conducted for various values of

k

(ranging from 1 to 10).

The Davies–Bouldin index quantifies the cost function in terms of two components.⁵³ The first reflects the minimization of intra-cluster variance, while the second accounts for the maximization of inter-cluster separation. The Davies–Bouldin (DB) index is defined as follows:

D B (k) = \frac{1}{k} \sum_{i = 1}^{k} \max_{i \neq j} {\frac{d_{i} + d_{j}}{D_{i j}}}

(2)

where: d_i and d_j denote the averages distances between the centroid clusters i and j and their respective elements. D_ij represents the distance between the centroids of clusters i and j.

The Elbow plots for the two sensors, presented in Figure 7, identify the optimal value of $k$ at the point where a distinct bend occurs. As it shown, the optimal number of clusters is clearly $k = 3$ .

Figure 7.

Elbow plot for (a) sensor 1 and (b) sensor 2.

Figure 8 illustrates the DB index computed for the two sensors. The DB index reaches its minimum value when the number of clusters is k = 3, indicating the most appropriate partitioning. Consequently, three clusters were retained in the k-means algorithm to characterize and assess the different damage modes. This choice ensures a balance between intra-cluster compactness and inter-cluster separation, leading to a more reliable interpretation of the sensor data.

Figure 8.

Evaluation of the DB index for (a) sensor 1 and (b) sensor 2.

Figure 9 illustrates the results of the acoustic data clustering in the plane defined by the amplitude and duration of the signals. The borders displayed no overlap and revealed well-defined domains between the different clusters. The literature review reveals that acoustic events characterized by low amplitude and short duration (cluster 1) are typically linked to matrix cracking. In contrast, events with higher amplitude and longer duration are associated with fiber breakage (cluster 2). In contrast, those with intermediate values (cluster 3) are commonly attributed to interfacial damage mechanisms (delamination and decohesion).⁵⁴ At low amplitude levels, the predominant damage mechanisms were identified as matrix cracking and interfacial degradation, including delamination and decohesion. These damage modes were associated with AE signals of short duration. These mechanisms are known to generate low-energy AE signals due to the relatively brittle failure of the polypropylene matrix. In contrast, higher amplitude and longer duration AE signals were indicative of damage within the flax fibers, which involve higher energy release and more prolonged signal activity. Similar trends have been reported in other studies from the literature, supporting these observations.⁵⁵

Figure 9.

Three clusters separated by the duration and peak amplitude for (a) sensor 1 and (b) sensor 2.

Figure 10 presents representative waveforms corresponding to the three damage modes detected in the samples. The acoustic signature of cluster 1, associated with matrix cracking, is characterized by low amplitude and short duration. Cluster 3, corresponding to fiber–matrix debonding, exhibits signals of intermediate amplitude and duration. In contrast, cluster 2, associated with fiber breakage, produces waveforms with high amplitude and long duration, reflecting more energetic phenomena and the more severe nature of this damage mode. Same tendences are presented in the literature review.^26,56

Figure 10.

Representative signal waveforms for (a) matrix cracking (b) interfacial damage and (c) fiber breakage.

Table 1 gives the percentage of each damage mode in the global acoustic activity recorded during the tests. The general trend indicates that specimen failure is mainly driven by matrix cracking and interfacial damage mechanisms, such as delamination and decohesion.

Table 1.

Damage relative contribution for the three modes.

		Specimen A	Specimen B	Specimen C	Std (%)	Mean (%)
Matrix cracking (%)	Sensor 1	45	48	27	28	39
Matrix cracking (%)	Sensor 2	41	46	26	28	39
Delamination/Decohesion (%)	Sensor 1	12	39	53	60	39
Delamination/Decohesion (%)	Sensor 2	35	45	52	19	39
Fiber breakage (%)	Sensor 1	43	13	20	62	22
Fiber breakage (%)	Sensor 2	24	9	23	46	22

The mean contribution from classes: matrix cracking, interfacial damage, and fiber breakage for the three tested specimen are, respectively 39%, 39% and 22%. This predominance of matrix cracking and interfacial damages is probably due to low interlaminar cohesion, as well as limited adhesion between the flax fibers and the polypropylene matrix. These phenomena can be attributed to the nature of the flax fiber, which are often less chemically compatible with non-polar polymers,⁵⁷ the local induced stress during thermo-folding, and probably on insufficient and various pressure rate between the punch and the die for these specimen.

Figure 11 presents the clustered AE amplitude superimposed on the corresponding force-displacement curves. The clustering analysis revealed three distinct groups of AE activity. Low-amplitude signals (cluster 1), appearing predominantly in the early stages of loading, are associated with minor damage mechanisms, such as matrix microcracking, which do not immediately compromise the material’s load-bearing capacity. At this stage, the material also exhibits an overall elastic behavior. Also, the matrix cracking cluster continues to appear until the sample breakage. The intermediate AE-signals amplitude (cluster 3) occurs throughout the loading process, reflecting progressive damage processes like stable crack propagation and interfacial damage (decohesion and delamination). High-amplitude AE events (cluster 2) become more pronounced after the maximum load (F_max) is reached, indicating the onset of critical failure mechanisms such as fiber breakage and unstable crack growth which continue to prevail in the post-failure phase. Furthermore, analysis of the amplitude trend presented in Figure 11 shows that fiber breakage occurs a displacement of approximately 4 mm, which corresponds to approximately 70% of F_max (folding resistance of the specimen). The alignment of these clusters with the force–displacement curve highlights the chronological sequence of damage evolution: the early initiation of matrix cracking (cluster 1), which occurs first due to the relatively low mechanical properties of the polypropylene matrix; the subsequent growth of delamination and interfacial decohesion (cluster 3); and finally, fiber rupture leading to failure (cluster 2). The same tendencies were systematically observed for all tested specimens, and they are in good agreement with previously reported results in the literature.⁵¹

Figure 11.

Load-displacement curve superimposed with clustered AE signals amplitude for (a) sensor 1 and (b) sensor 2.

Although the two sensors show differences in terms of percentage contribution of acoustic activity, the chronology of the damage modes - i.e., matrix cracking first, followed by interfacial damage and then fiber breakage - is systematically the same, whatever the sensor or specimen. This consistency confirms the reproducibility of the damage scenario and the reliability of the approach, despite the differences in sensitivity due to the position of the sensor.

Conclusions and outlooks

This study investigated the use of AE monitoring for the mechanical behavior of bio-based composites. The main conclusions that can be drawn are:

- Three distinct populations were identified corresponding to detected damage modes.

- The separation between clusters demonstrates the ability of the k-means algorithm to identify and distinguish these mechanisms effectively, providing insights into the composite material’s damage evolution.

- The analysis of acoustic emissions offers major potential for characterizing parts formed by bending, and is a promising tool for optimizing the manufacturing process.

Throughout their service life, composite structures are subjected to bending stresses, leading to the progressive appearance of various types of damage that compromise their durability. The increase in flexural load favours the accumulation of multiple degradations, such as matrix cracking, interfacial failure (fibre-matrix decohesion) and delamination between adjacent plies. In this context, bending tests coupled with acoustic emission monitoring can be carried out to identify the damage mechanisms and monitor their evolution.¹³

In future works, the experimental campaign will focus on combining Digital Image Correlation (DIC) and AE to characterize the flexural behavior of bio composites. A comparative study will be conducted on the mechanical behavior of specimens of similar geometry but produced by different processes - L-shaped thermoforming and U-shaped molding followed by cutting - will enable us to assess the influence of the thermo-folding approach on mechanical properties and durability, particularly in transition or curvature zones.

Although k-means clustering of AE parameters has demonstrated its ability to differentiate damage-related signal patterns, its empirical and unsupervised nature limits the physical interpretations. Hence, given the non-stationary behavior of AE signals in composite materials and their dispersive nature, integration of advanced time–frequency analyses—such as FFT, or wavelet transforms—offers a more robust framework for capturing the acoustic signature of damage mechanisms.^58,59 Future work will focus on combining these advanced signal processing techniques with supervised clustering approaches to improve the accuracy and reliability of AE-based damage identification.

Footnotes

Acknowledgements

The authors thank Dr Mokhfi TAKARLI from GC2D laboratory (GC2D: Génie Civil Diagnostic et Durabilité), Egletons, France, for providing AE acquisition system.

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The authors received no financial support for the research, authorship, and/or publication of this article.

ORCID iDs

Soufyane Benaboud

Valérie Dos Santos Martins

References

Correa

Montalvo-NavarreteHidalgo-Salazar

JMMA

. Carbon footprint considerations for biocomposite materials for sustainable products: a review. J Clean Prod 2019; 208: 785–794.

Pokharel

Falua

Babaei-GhazviniAcharya

. Biobased polymer composites: a review. J Compos Sci 2022; 6(9): 255.

Baley

BourmaudDavies

. Eighty years of composites reinforced by flax fibres: a historical review. Compos Appl Sci Manuf 2021; 144: 106333.

Karthik

Rajamanikkam

Venkatesan

, et al. State of the art: natural fibre-reinforced composites in advanced development and their physical/chemical/mechanical properties. Chin J Anal Chem 2024; 52(7): 100415.

Wong

Fabito

Debnath

, et al. A critical review: recent developments of natural fiber/rubber reinforced polymer composites. Cleaner Materials 2024; 13: 100261.

Chen

Wang

Luo

, et al. Bamboo as a naturally-optimized fiber-reinforced composite: interfacial mechanical properties and failure mechanisms. Compos B Eng 2024; 279: 111458.

LancioniAlessi

. Modeling micro-cracking and failure in short fiber-reinforced composites. J Mech Phys Solid 2020; 137: 103854.

Muthuraj

Misra

DefershaMohanty

FAK

. Influence of processing parameters on the impact strength of biocomposites: a statistical approach. Compos Appl Sci Manuf 2016; 83: 120–129.

Bouhamed

Souissi

Marechal

, et al. Ultrasound evaluation of the mechanical properties as an investigation tool for the wood-polymer composites including olive wood flour. Mech Mater 2020; 148: 103445.

10.

Habibi

AbbassiLaperrière

. Quasi-static indentation and acoustic emission to analyze failure and damage of bio-composites subjected to low-velocity impact. Compos Appl Sci Manuf 2022; 158: 106976.

11.

Morscher

Maxwell

. Monitoring tensile fatigue crack growth and fiber failure around a notch in laminate SIC/SIC composites utilizing acoustic emission, electrical resistance, and digital image correlation. J Eur Ceram Soc 2019; 39(2): 229–239.

12.

Frącz

Janowski

Pruchniak

, et al. The use of computed tomography in the study of microstructure of molded pieces made of Poly(3-hydroxybutyric-co-3-hydroxyvaleric acid) (PHBV) biocomposites with natural fiber. Polymers 2021; 13(17): 2942.

13.

Habibi

Laperrière

. Combining digital image correlation and acoustic emission to characterize the flexural behavior of flax biocomposites. Applied Mechanics 2023; 4(1): 371.

14.

Benaboud

Takarli

Pouteau

, et al. Fatigue process analysis of aged asphalt concrete from two-point bending test using acoustic emission and curve fitting techniques. Construction and Building Materials 2021; 301: 124109. doi:10.1016/j.conbuildmat.2021.124109.

15.

Thirumalaiselvi

Sasmal

. Acoustic emission monitoring and classification of signals in cement composites during early-age hydration. Constr Build Mater 2019; 196: 411–427.

16.

Liang

Liu

Wang

, et al. Flexural progressive failure mechanism of hybrid 3D woven composites: combination of X-ray tomography, acoustic emission and digital image correlation. Compos Struct 2022; 280: 114894.

17.

Seychal

Ramasso

Le Moal

, et al.

Towards in-situ acoustic emission-based health monitoring in bio-based composites structures: does embedment of sensors affect the mechanical behaviour of flax/epoxy laminates?

Compos B Eng 2022; 236: 109787.

18.

Allagui

El Mahi

Rebiere

J-L

, et al. In-situ health monitoring of thermoplastic bio-composites using acoustic emission. J Thermoplast Compos Mater 2023; 36(11): 4296–4316.

19.

Kek

Šturm

Bergant

. Tension and shear behaviour of basalt fiber bio-composites with digital image correlation and acoustic emission monitoring. Polymers 2024; 16(10): 1331.

20.

Kek

Potočnik

Misson

, et al. Characterization of biocomposites and glass fiber epoxy composites based on acoustic emission signals, deep feature extraction, and machine learning. Sensors 2022; 22: 6886.

21.

Bussiba

Kupiec

Ifergane

, et al. Damage evolution and fracture events sequence in various composites by acoustic emission technique. Compos Sci Technol 2008; 68(5): 1144–1155.

22.

Bravo

Toubal

Koffi

, et al. Damage characterization of bio and green polyethylene–birch composites under creep and cyclic testing with multivariable acoustic emissions. Materials 2015; 8(11): 7322–7341.

23.

. Wavelet-based AE characterization of composite materials. NDT E Int 2000; 33(3): 133–144.

24.

Haggui

El Mahi

Jendli

, et al. Static and fatigue characterization of flax fiber reinforced thermoplastic composites by acoustic emission. Appl Acoust 2019; 147: 100–110.

25.

Samborski

Rzeczkowski

Korzec-Strzałka

. Experimental study of delamination process in elastically coupled laminates with the acoustic emission technique. Eng Struct 2024; 300: 117196.

26.

Harizi

Chaki

Bourse

, et al. Damage mechanisms assessment of Glass Fiber-Reinforced Polymer (GFRP) composites using multivariable analysis methods applied to acoustic emission data. Compos Struct 2022; 289: 115470.

27.

Czigany

. An acoustic emission study of flax fiber-reinforced polypropylene composites. J Compos Mater 2004; 38(9): 769–778.

28.

Habibi

Lebrun

Laperrière

. Experimental characterization of short flax fiber mat composites: tensile and flexural properties and damage analysis using acoustic emission. J Mater Sci 2017; 52(11): 6567–6580.

29.

Nasri

Toubal

. Experimental and numerical investigation of damage and mechanical property retention by bio-composite plastic made with flax or pinewood fiber and aged by exposure to ultraviolet light. J Build Eng 2023; 79: 107899.

30.

Krishna

Raju

Desarkar

. Damage characterization of CFRP composites under compression using acoustic emission time–frequency analysis and unsupervised deep learning. Mech Syst Signal Process 2025; 238: 113224.

31.

Wang

Chegdani

Yalamarti

, et al. Acoustic emission characterization of natural fiber reinforced plastic composite machining using a random forest machine learning model. J Manuf Sci Eng 2020; 142: 031003.

32.

Huang

Zong

Liu

, et al. An in-situ experimental method enables one-to-one matching of acoustic emission signals and damage modes of 3D woven composites. Thin-Walled Struct 2026; 218: 114090.

33.

Barile

Casavola

Pappalettera

, et al. Experimental wavelet analysis of acoustic emission signal propagation in CFRP. Eng Fract Mech 2019; 210: 400–407.

34.

Zhang

Delichatsios

Fateh

, et al. Characterization of flammability and fire resistance of carbon fibre reinforced thermoset and thermoplastic composite materials. J Loss Prev Process Ind 2017; 50: 275–282.

35.

Zhai

Zhang

, et al. Experimental and numerical study of thermoforming-induced deformation in glass fiber/polyetherimide thermoplastic wing leading edge. Compos B Eng 2024; 275: 111319.

36.

Safwa

Yeddu

Leminen

. Modeling of the thermoforming process of paperboard composites for packaging. Bioresources 2024; 19(2): 2120–2134.

37.

Asghar

Nawab

Imad

, et al. Effect of gradual thermoforming pressure on the mechanical properties of jute/polypropylene commingled composites. J Compos Mater 2024; 58(4): 505–517.

38.

Limaye

Pradeep

Kothari

, et al. Thermoforming process effects on structural performance of carbon fiber reinforced thermoplastic composite parts through a manufacturing to response pathway. Compos B Eng 2022; 235: 109728.

39.

Mattlet

Sicot

Van Schoors

, et al. Influence of processing and matrix parameters on the manufacturing of unidirectional flax/polypropylene composites. SN Appl Sci 2023; 5(12): 314.

40.

ISO 527-1 . Part 1: general principles, plastics — determination of tensile properties. Switzerland, 2019. Disponible sur: https://221.224.170.180/2021-11-30/1638241309169.pdf

41.

ISO-178-2019 . Plastics — determination of flexural properties. Switzerland, 2019. Disponible sur: https://cdn.standards.iteh.ai/samples/70513/cd351551961842c590df25e23807b73c/ISO-178-2019.pdf

42.

Malkapuram

Kumar

Negi

. Recent development in natural fiber reinforced polypropylene composites. J Reinforc Plast Compos 2009; 28(10): 1169–1189.

43.

Melelli

Jamme

Legland

, et al. Microfibril angle of elementary flax fibres investigated with polarised second harmonic generation microscopy. Ind Crop Prod 2020; 156: 112847.

44.

Yan

Chouw

Jayaraman

. Flax fibre and its composites – a review. Compos B Eng 2014; 56: 296–317.

45.

Olsson

. A survey of test methods for multiaxial and out-of-plane strength of composite laminates. Compos Sci Technol 2011; 71(6): 773–783.

46.

Masmoudi

El Mahi

Turki

. Fatigue behaviour and structural health monitoring by acoustic emission of E-glass/epoxy laminates with piezoelectric implant. Appl Acoust 2016; 108: 50–58.

47.

Marissen

Van Der Drift

Sterk

. Folding of continuous fibre thermoplastic composites. Appl Compos Mater 1997; 4(5): 273–282.

48.

McCool

Murphy

Wilson

, et al. Thermoforming carbon fibre-reinforced thermoplastic composites. Proc Inst Mech Eng Part L 2012; 226(2): 91–102.

49.

Barré

Benzeggagh

. On the use of acoustic emission to investigate damage mechanisms in glass-fibre-reinforced polypropylene. Compos Sci Technol 1994; 52(3): 369–376.

50.

Bouzouita

Salvia

Ben Daly

, et al. Contribution of acoustic emission to evaluate the influence of hygrothermal aging on mechanical behavior of hemp reinforced polypropylene composites. In: Proceedings of the 6th European workshop - structural health monitoring 2012, EWSHM 2012, Dresden, Germany, July 2012, vol 1. Présenté à 6th European Workshop onStructural Health Monitoring -. In: Journal Abbreviation: Proceedings of the 6th European Workshop - Structural Health Monitoring 2012, EWSHM 2012. Resden, Germany, juill.

51.

Assarar

Scida

Zouari

, et al. Acoustic emission characterization of damage in short hemp‐fiber‐reinforced polypropylene composites. Polym Compos 2016; 37(4): 1101–1112.

52.

Lomov

Yan

, et al. Cluster analysis of acoustic emission signals for 2D and 3D woven glass/epoxy composites. Compos Struct 2014; 116: 286–299.

53.

Davies

Bouldin

. A cluster separation measure. IEEE Trans Pattern Anal Mach Intell 1979; PAMI-1(2): 224–227.

54.

Bravo

Toubal

Koffi

, et al. Characterization of tensile damage for a short birch fiber-reinforced polyethylene composite with acoustic emission. Int J Mater Sci 2013; 3: 79–89.

55.

Demirbas

Caliskan

Numan Zafar

. Damage detection on flexural loading of hybrid laminated composite by acoustic emission. Compos Struct 2025; 360: 119056.

56.

Marec

Thomas

J-H

El Guerjouma

. Damage characterization of polymer-based composite materials: multivariable analysis and wavelet transform for clustering acoustic emission data. Mech Syst Signal Process 2008; 22(6): 1441–1464.

57.

Stamboulis

Baillie

Schulz

. Interfacial characterisation of flax fibre-thermoplastic polymer composites by the pull-out test. Angew Makromol Chem 1999; 272(1): 117–120.

58.

Satour

Montrésor

Bentahar

, et al. Wavelet based clustering of acoustic emission hits to characterize damage mechanisms in composites. J Nondestruct Eval 2020; 39(2): 37.

59.

Barile

Casavola

Pappalettera

, et al. Acoustic emission waveforms for damage monitoring in composite materials: shifting in spectral density, entropy and wavelet packet transform. Struct Health Monit 2022; 21(4): 1768–1789.

Acoustic emission-based characterization technique for assessing folding strength of thermoformed flax-reinforced polypropylene composite

Abstract

Keywords

Introduction

Materials and experimental methods

Materials and samples preparation

Manufacturing of experimental samples

Folding-based shaping of test specimens

Mechanical loading and experimental set-up

Acoustic emission instrumentation

Results and discussions

Mechanical properties

Damage mechanisms analysis

Conclusions and outlooks

Footnotes

Acknowledgements

Declaration of conflicting interests

Funding

ORCID iDs

References