Experimental Study on Single Fiber Tensile Properties of Sisal Fibers Using a Digital Image Correlation Method as a Strain Measurement

ABSTRACT

The development of natural fibers in engineering applications requires an accurate measurement of their dimensional characteristics and mechanical properties. Fiber cross-sectional area (CSA) obtained from lateral dimensional measurements should consider the specific cross-sectional shape (CSS) of the fibers and their wide lengthwise variations. In this study, the dimensional measurements of water-retted and unretted sisal fibers were conducted by laser scanning microscopy (LSM) and calculation-based methods. Single fiber tensile tests were performed using the digital image correlation (DIC) method. Results show that the diameters of water-retted and unretted fibers measured by LSM were 233 ± 45 µm and 236 ± 42 µm, respectively. The diameters of water-retted and unretted fibers obtained by calculation were 192 ± 17 µm and 178 ± 20 µm, respectively. The tensile strengths of water-retted and unretted fibers were 679 ± 118 MPa and 718 ± 106 MPa, respectively. The strain at failure and elastic modulus of water-retted and unretted fibers were 2.0 ± 0.6% and 34 ± 8 GPa, as well as 1.5 ± 0.2% and 48 ± 6 GPa, respectively. Water retting seems to predominantly influence the strain at failure and elastic modulus of sisal fibers, as confirmed by the ANOVA analysis.

摘要

天然纤维在工程应用中的发展需要精确测量其尺寸特性和机械性能. 从横向尺寸测量中获得的纤维横截面面积（CSA）应考虑纤维的特定横截面形状（CSS）及其广泛的纵向变化. 本研究采用激光扫描显微镜（LSM）和基于计算的方法对水脱胶和未脱胶剑麻纤维的尺寸进行了测量. 使用数字图像相关（DIC）方法进行单纤维拉伸试验. 结果表明，用LSM测量的水脱胶和未脱胶纤维的直径分别为233 ± 45 µm和236 ± 42 µm. 通过计算获得的水脱胶和未脱胶纤维的直径分别为192 ± 17 µm和178 ± 20 µm. 水脱胶纤维和未脱胶纤维的拉伸强度分别为679 ± 118MPa和718 ± 106MPa. 脱胶纤维和未脱胶纤维的破坏应变和弹性模量分别为2.0 ± 0.6%和34 ± 8GPa，以及1.5 ± 0.2%和48 ± 6GPa. ANOVA分析证实，水脱胶似乎主要影响剑麻纤维的失效应变和弹性模量.

KEYWORDS:

• Sisal fiber
• water-retted fiber
• unretted fiber
• dimensional analysis
• optical techniques
• mechanical properties

关键词:

• 剑麻纤维
• 水浸渍纤维
• 未经修饰的纤维
• 维度分析
• 光学技术
• 机械性能

Introduction

In recent years, there has been a significant focus on the development of biocomposites within the composite industry. Natural fibers are gaining importance as reinforcing fibers in the creation of biocomposite materials, providing solutions to environmental and economic challenges. However, unlike synthetic fibers, natural fibers lack a well-established scientific and technological background to consistently deliver controlled performance suitable for engineering applications (Haag et al. 2017). In this regard, the need for accurate and reliable properties of natural fibers increases.

Composite manufacturers need to rely on fiber properties to ensure the quality of products made from natural fiber-reinforced composites. To utilize natural fibers effectively in engineering applications, precise measurements of dimensional characteristics and mechanical properties are crucial. Despite the growing interest in using natural fibers as reinforcement in various composite materials and applications, previous studies have revealed significant variability in the measured properties of single fibers, even within the same species, such as chemical composition (Hussain et al. 2023; Naveen et al. 2018; Tran et al. 2015), mechanical and physical properties (Garat et al. 2018; Saaidia et al. 2023). This is attributed to the inherent variability of natural plants, irregular fiber geometry, aging, planting soil type, cultivation location and weather conditions (Fiore et al. 2016; Parikh 2023).

To ensure that variations in fiber properties are solely due to material variability and not testing setups, consistent testing methods need to be implemented. However, also the actual testing setup could introduce additional variations. Reliable and accurate testing methods are essential for minimizing variations caused by testing setups. Therefore, single fiber tensile testing is employed to determine the precise mechanical properties of natural fibers, including strength and strain, enabling the evaluation of their potential.

The computation of tensile strengths of natural fibers depends greatly on the methods used to define the cross-sectional areas (CSAs) of the fibers and the tensile testing setups. Gathering and processing dimensional data accurately are crucial for determining the CSAs of natural fibers. Additionally, precise measurement of fiber displacement is vital to accurately determine the actual fiber strain. In single fiber tensile testing, fiber displacement can be measured using two methods.

The first method, in accordance with ASTM D3822, involves direct measurement of displacement using a tensile test machine (Bermudo et al. 2019). In this method, the movement of the clamps determines the displacement measurement. However, this method may introduce errors due to fiber slippage within the grips. To account for fiber slippage, a large number of specimens from each sample are required for testing. Attaching an extensometer to correct for machine compliance is not feasible for single fiber tensile testing setups since it is impossible to attach the extensometer to a single fiber.

The second method, in accordance with ASTM C1557, involves displacement measurement using digital image correlation (DIC) (Sarasini, Tirillò, and Carolina Seghini 2018). DIC is an optical method that tracks changes in a series of deforming images captured by a DIC camera system. It is widely used in experimental solid mechanics for single fiber tensile tests due to its ability to capture full-field deformations and perform non-contact measurements. DIC relies on image-based measuring methods, digital image processing, and numerical computation to measure displacement and strain accurately while avoiding deformation errors caused by fiber slippage. It enables precise measurements of displacement and strain on the surface of the fibers, minimizing variations in mechanical properties caused by testing setups compared to the standard ASTM D3822 method (Depuydt et al. 2017; Hua et al. 2011; Kim et al. 2013; Mehdikhani et al. 2016; Pan, Lu, and Xie 2010; Tay et al. 2005).

The objective of this study is to investigate the mechanical properties of sisal fibers using single fiber tensile tests with the DIC method. By avoiding variations in fiber properties caused by testing methods such as fiber slippage and gripping effects at the ends of the fiber, this study aims to determine the actual strain and modulus of sisal fibers. Different methods of fiber cross-section determination were implemented to investigate their mechanical properties.

Materials and methods

Materials

The study utilized sisal fibers obtained from Abala Zone 2 in the Afar region of Ethiopia. Abala resides at an elevation of 1465 meters and experiences an average annual temperature of 27°C. The prevailing wind direction is from the northeast at a speed of 8 km/h, with a humidity level of 47%. The soil type in this location is classified as vertisol. Sisal fibers are estimated to have a density ranging from 1.30 to 1.55 g/cm³ (Lau et al. 2018; Naveen et al. 2018). The tensile strength of these fibers falls within the range of 385 to 725 MPa, while the modulus of elasticity ranges from 9 to 38 GPa (Mohapatra and Kar 2019; Thomason et al. 2011).

Fiber extraction

Sisal fibers were obtained by manually decorticating them from sisal leaves using hand tools such as blunt knives and cylindrical woods. The decortication process did not involve any water treatment before or during the extraction. The first batch of fibers was obtained directly from the decorticated fibers without undergoing any water treatment. Another batch of fibers was obtained after subjecting the decorticated fibers to retting, which involved immersing them in distilled water. The decorticated fibers were initially dried in the sun for a week, and then immersed in distilled water at room temperature for a duration of 30 days. After the immersion treatment, any impurities present on the fibers were manually removed. The fibers were then washed with distilled water and dried under sunlight. The study focused on two types of fibers: water-retted fibers and unretted fibers, representing the two different batches obtained.

Fiber density

To measure the densities of sisal fibers, a helium gas pycnometer was employed. Samples of both water-retted and unretted fibers were utilized, with fiber lengths of 5 mm and 10 mm. Additionally, samples of milled powder from unretted fibers, processed using a vibratory disc mill (Retsch RS200), were included in the density measurements. For each type of fiber, four samples were tested to ensure accuracy and reliability of the measurements.

Fiber cross sections

The measurement of fiber diameters was conducted using two methods: laser scanning microscopy (LSM) and calculation based on the weight of the fibers. The two methods are classified as equivalent diameter measurement methods. For the LSM method, three width measurements were taken at three different regions of each fiber in every sample (Figure 1). These measurements were obtained at approximately 1 mm intervals along a total fiber length of 4 mm. The average value of the three measurements for each fiber was considered as its diameter.

Figure 1. Diameter measurements of sisal fiber using LSM.

In the calculation method, the weight of the fiber and its density were used to calculate the cross-sectional area (CSA). The fiber weight was measured after drying the fibers in an oven at 60°C for 24 hours. The CSA was then determined by dividing the fiber weight by the product of the fiber density and fiber length. The fiber length was measured using a steel ruler with an accuracy of ±0.05 mm. Assuming a circular cross-section for the fiber, the diameter (d) was calculated using the equation: d = $\sqrt{4\mathit{a}/\mathit{\pi }}$, where “a” represents the CSA of the fiber.

To ensure accuracy and reliability, 50 samples of water-retted fibers and 50 samples of unretted fibers were used to determine the average diameters using both methods. The large sample sizes were chosen to obtain precise and dependable average diameter values. Furthermore, the average diameters were utilized to calculate the cross-sectional areas (CSAs) of the fibers, which were subsequently employed in determining the tensile strengths of the fibers during the tensile testing.

The study also focused on examining the cross-sectional shape (CSS) of the fibers. To prepare the sample specimens, the fibers were embedded in Technovit 4071 resin, which is a mixture of powder polymer and liquid hardener in a 1:1 ratio. The embedded fibers were then cured at room temperature for a period of 4 to 6 minutes. Subsequently, the embedded fibers underwent grinding and polishing processes using various grades of sandpaper and diamond paste. For the final polishing step, Struers DP-Paste M with a 6 µm diamond was utilized.

Cross-sectional images of the fibers were captured using laser scanning microscopy (LSM) and scanning electron microscopy (SEM) with magnification scales of 20× and 92×, respectively. Additionally, the actual CSAs of the fibers were determined using digital image analysis. ImageJ software package was employed for this purpose, utilizing image tracing techniques, as shown in Figure 2 (Quinaya and d’Almeida 2019). Approximately 25 fiber images were used to calculate and compare the CSAs of the fibers obtained through equivalent diameter measurements and ImageJ analysis. This methodology allowed for a comprehensive examination of the cross-sectional shape of the fibers and facilitated the comparison of CSAs of single fibers using both measurement techniques.

Figure 2. Determining CSA of the fiber using ImageJ software.

Single fiber tensile test

The single fiber tensile tests were conducted using the digital image correlation (DIC) method, which enables the optical measurement of fiber strain. In this testing approach, the fibers were gripped between upper and lower clamps of the test machine, similar to standard tensile test specimens. However, certain modifications were made to improve measurement accuracy due to the small thickness of the single fibers, which made gripping them with clamps challenging without causing slippage or damage.

To address this issue, a paper frame was prepared and attached to both sides of each individual fiber to prevent gripping effects and fiber slippage. Methyl cyanoacrylate glue, specifically SICOMET 99, was used for attaching the fibers to the paper frames, and the glue was allowed to cure for 24 hours prior to conducting the tests. Additionally, two speckle patterns or optical flags, approximately 3 mm in diameter, were created on the fiber surface using white correction fluid (Tipp-Ex Soft Grip 10 M correction pen). To enhance visibility, the white speckle patterns were sprayed with black paint, making the flags more distinct (Figure 3).

Figure 3. Specimen preparation for single fiber tensile testing.

The total length of the fiber specimens was 100 mm, with a gauge length of 66 mm. The fiber length between the optical flags was approximately 50 mm, chosen for ease of manipulation during specimen preparation and testing. About 50 water-retted and 50 unretted fiber samples were used for the tests. After gripping the specimens with the upper and lower clamps, the paper frame was cut at its center before conducting the tests.

For capturing of displacements and force during the testing, a digital camera (Limess messtechnik and GmbH software, Germany) with a resolution of 96 dpi and a Componon-S 2.8/50 lens (Schneider Kreuznach, Germany) was used. The camera was positioned approximately 1 m away from the specimen. The specimen was directly irradiated with a 350 W halogen lamp. Images and force data were recorded and saved at 1-second intervals during the test using acquisition software (LimShot, Limess messtechnik and GmbH software). The DIC method was employed by correlating the images with the applied load using the limess key and Vic-2D 2009 software.

The specimens for single fiber tensile testing were prepared according to ASTM C1557–03 (Semiautomatic and Analysis 2004) standards. The tests were conducted using an Instron 5943 machine with a 100 N load cell and a displacement rate of 1 mm/min (Figure 4). The tests were performed at room temperature and 50% relative humidity. The raw data of fiber extensions obtained from direct measurements and optically supported measurements were extracted and analyzed. A comparison was made to determine the significance of differences between the results of fiber strains obtained through the DIC method and those calculated from direct measurements of fiber extensions. For uniformly distributed data, the results were summarized using the average ± standard deviation notation.

Figure 4. Setup of single fiber tensile testing according to ASTM C1557–03 (a): universal testing machine, (b): detail of clamps and (c): specimen gripped with clamps.

During the tests, a series of deforming images were captured by the DIC camera system (Figure 5a). These images were imported into the software for data processing, allowing the determination of fiber strain between the two optical flags. Simulation was performed by correlating the imported images with the applied load using the limess key and Vic-2D 2009 software. The strain results were represented by colors on the deforming fiber between the two optical flags. The simulation color of the deformed fiber image matching the scale color (vertical line) indicated the maximum fiber strain. In the scale color, red represented the maximum fiber strain at failure (Figure 5b).

Figure 5. (a) Single fiber tensile testing, and (b) image data processing using limess key and vic-2D software.

Analysis of variance

To study the impact of two factors, measurement method and water retting, on various properties of the fibers, a two-factor ANOVA without replication was employed. This statistical analysis, using Excel software, was applied to the tensile tests for water-retted and unretted fibers. A 95% confidence level (alpha = 0.05) was chosen for the analysis.

Results and discussion

Fiber density

The average densities of water-retted and unretted fibers were compared at different fiber lengths to investigate the effects of retting and length on fiber density. The results revealed that the density of sisal powder increased by 9.9% and 5.8% compared to the densities of retted and unretted fibers at a length of 10 mm, respectively. Similarly, the density of sisal powder increased by 2.9% and 2.1% compared to the densities of retted and unretted fibers at a length of 5 mm, respectively. This can be attributed to the fact that all porosities are fully accessible in the milled powder, allowing helium gas in the pycnometer to reach all the porosities. In contrast, the density of retted and unretted fibers at a length of 10 mm decreased by 6.8% and 3.7% compared to the density of the fibers at a length of 5 mm, respectively. This is due to the fact that the number of closed porosities increases as fiber length increases, making it more difficult for the gas to reach all the porosities. Consequently, the measurement yields lower density for longer fibers. Therefore, the density measured in the powder state is considered the actual density of the fibers, and a density of 1.45 g/cm³ was used for calculating fiber diameters.

Table 1 presents the average densities of water-retted and unretted fibers, allowing for a comparison between the two. The results indicate that, at a length of 10 mm, the density of water-retted fibers decreased by 3.8% compared to unretted fibers of the same length. However, at a length of 5 mm, the density of water-retted fibers only decreased by 0.7% compared to unretted fibers, which is considered insignificant. This suggests that the impact of water retting becomes less pronounced as fiber length decreases. Generally, an increase in the duration of water retting leads to a reduction in the linear density of natural fibers (Wetaka et al. 2016), as water molecules penetrate the fiber lumens and porosities (Figure 6), causing cross-sectional swelling (Garat et al. 2020). As the depth increases, it becomes more difficult for the helium gas in the pycnometer to reach all the porosities, resulting in lower density. Additionally, water exposure can lead to the removal of denser elements, such as hemicellulose, further contributing to lower density (KılınÇ et al. 2018; Stawski et al. 2020).

Figure 6. Microstructures of sisal fiber using SEM at magnifications of (a): 500× and (b): 2000 × .

Table 1. Densities of the various sisal fiber types using helium gas pycnometer.

Sample fiber type, fiber length	Weight [g]	Mean Pressures [Pa]		Density [g/cm3]	Standard deviation
		P1	P2
Retted, 10 mm Unretted, 10 mm Retted, 5 mm Unretted, 5 mm Milled powder	1.23 1.40 2.24 2.16 3.84	124.24 × 10^3 124.18 × 10^3 124.45 × 10^3 125.07 × 10^3 124.18 × 10^3	36.75 × 10^3 36.82 × 10^3 37.44 × 10^3 37.58 × 10^3 38.47 × 10^3	1.32 1.37 1.41 1.42 1.45	0.010 0.004 0.005 0.007 0.002

Fiber cross sections

As described in section 2.4, the measurement of fiber diameter was carried out using laser scanning microscopy (LSM) and calculations based on fiber weight. Both methods assumed that the fibers have a consistent, circular cross-section with a constant diameter along their length (Kandemir et al. 2020; Lima et al. 2014). However, in reality, natural fibers have irregular and variable cross-sections along their length. Obtaining the actual cross-sectional areas (CSAs) of the fibers after testing is not feasible due to the distortion and variation in cross-sections caused by the testing process. Therefore, the diameters measured from fiber width using LSM and the fiber weight-based calculations were utilized to estimate the CSAs of the fibers.

Furthermore, a comparison was made between the diameters of water-retted fibers obtained through LSM and calculation methods. The diameters of the fibers were analyzed based on their frequency within specific ranges (Figure 7). A graph depicting the relative frequency versus diameter range was employed to display the ranges of average fiber diameters. The diameters of water-retted fibers determined by LSM and calculation were found to be 233 ± 45 µm and 192 ± 17 µm, respectively. Similarly, the diameters of unretted fibers determined by LSM and calculation were 236 ± 42 µm and 178 ± 20 µm, respectively. Each average diameter of the fibers fell within the range with the highest frequency.

Figure 7. Diameter histograms of (a) water retted fibers by calculation, (b) water retted fibers by measurement, (c) unretted fibers by calculation and (d) unretted fibers by measurement.

The findings reveal that the average diameters of the fibers obtained using laser scanning microscopy (LSM) are greater than those obtained through calculation. This disparity arises from the assumption made during LSM measurements that the fibers are solid, whereas in reality, they possess porosities. In the LSM method, these porosities are treated as part of the solid fiber structure, while in the calculation method, they are excluded from consideration when determining fiber weights. Consequently, the calculated cross-sectional areas (CSAs) are considered to be more reliable. Additionally, although the cross-sectional shape (CSS) of the fibers is assumed to be circular, it deviates significantly from a perfect circle (Figure 8). Therefore, the diameters obtained from both methods should be regarded as equivalent diameters.

Figure 8. Images of CSSs of sisal fibers captured by LSM.

The CSS of the fibers exhibits some variation even within the same species (Figure 8a). In order to obtain accurate fiber properties, the actual CSAs of individual fibers were determined using image processing via the ImageJ software. These CSAs were then compared with the CSAs calculated based on the assumption of circular shapes. The results indicate that the average CSA determined through shape tracing image processing was 15% to 19% lower than the average CSA determined through calculation under the assumption of circular shapes.

Single fiber tensile test

In the analysis, a subset of the tensile tests was considered for further examination due to experimental errors (fiber slippage and premature failure) encountered during the testing process. Specifically, out of the 50 tensile tests conducted on retted fibers, only 34 tests were deemed suitable for further analysis. Similarly, 45 out of the 50 tensile tests performed on unretted fibers were considered successful and included in the analysis.

Special attention was given to the issue of premature failure caused by the effects of glue during sample preparation. It was recognized that improper alignment of fibers on the paper frame, along with excessive glue, could compromise the load-carrying capacity of the fibers. To account for this, careful observation of the failure modes of the fibers was conducted to determine whether to include or exclude specific tests from further analysis. The failure modes observed in the fiber bundles included fiber breaking, pull-out of elementary fibers from the bundles, fiber slippage, and failure at the glued locations. Additionally, the location of fiber failures during the tensile tests provided valuable information in the decision-making process regarding test acceptance or rejection. These considerations were taken into account to ensure the reliability and accuracy of the results by accounting for factors such as fiber slippage and premature failure caused by the gluing effects.

Tensile strength

Figure 9 presents the results of the tensile strengths of water-retted and unretted fibers obtained using two different methods of diameter measurement. It is observed that the tensile strengths of both water-retted and unretted fibers, when diameters determined by laser scanning microscopy (LSM), exhibit a relatively high variation with a relative standard deviation of 40%. Several factors contribute to this variation, including the non-uniformity of the fibers, weak points resulting from fiber damage during extraction and handling, microstructural differences along the fiber length (Gudayu et al. 2020), and the wide range of porosities present in natural fibers (Rebolledo, Cloutier, and Claude Yemele 2018). Moreover, the wide range of fiber diameters obtained by LSM further contributes to the scattered nature of the tensile strengths for both water-retted and unretted fibers.

Figure 9. Tensile strengths of water retted and unretted sisal fibers.

Figure 9 also displays the average tensile strengths of water-retted and unretted fibers for diameters determined by both LSM and calculation. The results indicate that the average tensile strengths of water-retted and unretted fibers, when diameters determined by LSM, are reduced by 33% and 31% respectively, compared to the tensile strengths, when diameters determined by calculation. This reduction can be attributed to the fact that the cross-sectional areas (CSAs) of the fibers obtained through calculation are lower than those determined by LSM, as described in section 3.2.

The impact of the water retting process on the tensile strength of the fibers was also investigated. The results demonstrate that the average tensile strength of water-retted fibers decreases by 5.8% compared to the average tensile strength of unretted fibers. This suggests that water retting has a minimal effect on the tensile strength of sisal fiber. During the early stages of water retting, non-cellulosic constituents, primarily pectin and hemicellulose, are removed from the natural fibers (Garat et al. 2020), leading to an increase in the proportion of cellulosic constituents, which improves the mechanical properties of the fibers (Mazian et al. 2019). However, excessive retting can gradually diminish the mechanical properties of the fibers (Mazian et al. 2020), due to the degradation of the cellulosic constituents by anaerobic bacterial fermentation (Chabbert et al. 2020). Indeed, the cellulose component of natural fibers exhibits a slower degradation rate due to its lower water absorptivity, attributed to the presence of a small amount of -OH groups in the semi-crystalline polysaccharide structure of cellulose (Abdullah et al. 2020). This may explain the insignificant effect of water retting on the tensile strength of sisal fiber retted for 30 days, as cellulose is the primary contributor to the fiber’s stiffness, stability, and strength (Lee et al. 2020).

Table 2 also reveals statistically significant differences in average tensile strength based on fiber diameter measuring methods. This is confirmed by the ANOVA analysis, where F-value exceeded F critical value and P-value was very low (P-values <0.05). However, water retting process did not significantly affect average strength, as the ANOVA analysis in Table 2 revealed F-value close to the F critical value and P-value near the confidence level (alpha = 0.05).

Table 2. Summary of ANOVA analysis for tensile strength of water-retted and unretted fibers.

Summary of tensile strength [MPa]		Fiber diameter measurement
		Calculated		Measured
Water-retted	Average	678.5		511.7
	Variance	8806.2		26337.9
Unretted	Average	717.9		547.0
	Variance	11469.6		37322.2
ANOVA	Source of variation	F-value	F critical value		P-value
	Water retting	4.2	3.9		0.043
	Diameter measurement	16.4	4.0		0.001

Strain at failure

The strain at failure of each fiber, a measure of its maximum extension before breaking, was determined using two methods. The first method directly measured fiber extension using an Instron machine. The second method, digital image correlation (DIC), provided an indirect strain measurement through image analysis of deformed fibers. DIC avoids errors caused by fiber slippage, a limitation of direct measurements. For each fiber, DIC analysis yielded strain values which were then compared to those obtained from the direct measurement of Instron machine. In Figure 10, the light-red color of the deformed fiber within the optical flags reveals its peak strain, estimated at 1.4%. This visual approach was applied for all specimens to determine the maximum strain at failure. Comparison of the two methods revealed that DIC measurements consistently underestimated strain at failure compared to direct measurements (Figure 11). This discrepancy, a decrease of 53% for water-retted fibers and 42% for unretted fibers, suggests that the direct measurements overestimate strain due to fiber slippage. DIC, by avoiding this factor, provides a more accurate representation of true fiber strain.

Figure 10. Digital image correlation retrospective through image data processing.

Figure 11. Strains at failure of water retted and unretted sisal fibers.

The study also investigated the effect of water retting, a treatment process, on fiber strain. Water-retted fibers exhibited a significant increase in strain at failure compared to unretted fibers, regardless of the measurement method used. This 31% increase by DIC measurement and 41% increase by direct measurement indicates that water retting enhances the ductility of sisal fibers. The improved ductility of water-retted fibers is likely attributed to two factors. First, water retting removes non-cellulosic components and impurities from the fiber surface, potentially reducing brittleness. Second, water acts as a plasticizer (Balogun et al. 2015; Jiang et al. 2019), increasing the flexibility of the cellulose, the primary load-bearing component of natural fibers. This plasticization effect plays a crucial role in enhancing fiber ductility.

Table 3 also shows statistically significant differences in average tensile strain at failure of fibers based on both water retting and the chosen fiber diameter measurement method. This is confirmed by the ANOVA analysis, where F-values exceeded F critical values and P-values were very low.

Table 3. Summary of ANOVA analysis for tensile strain of water-retted and unretted fibers.

Summary strain at failure [%]		Measurement methods
		Direct		DIC
Unretted	Average	2.19		1.54
	Variance	0.10		0.05
Retted	Average	3.08		2.01
	Variance	0.56		0.22
ANOVA	Source of variation	F-value	F critical value		P-value
	Water retting	29.29	4.05		0.002
	Measurement method	65.61	4.05		0.000

Elastic modulus

The elastic moduli of water-retted and unretted sisal fibers were determined using strains obtained by both direct measurement and digital image correlation (DIC) techniques. Figure 12 shows the strain results of water-retted and unretted fibers. Interestingly, unretted fibers consistently exhibited higher elastic moduli, regardless of the measurement method. Compared to water-retted fibers measured by DIC, unretted fibers showed a 35% increase in average elastic modulus. Similarly, compared to water-retted fibers measured directly from fiber extension, unretted fibers displayed a 46% higher average elastic modulus. This difference can be attributed to the removal of hemicellulose and pectin during water retting. These components act as natural stiffeners by binding the fiber constituents together. Their removal in water-retted fibers weakens the fiber structure, and lowering the overall elastic modulus.

Figure 12. Elastic moduli of water retted and unretted sisal fibers.

There are two important results that need to be understood. As illustrated in section 3.3.2, water retting exhibits a plasticizing effect on the sisal fibers, which further reducing their rigidity. This is illustrated in Figure 12, where the elastic modulus of water-retted fibers is 42% (DIC) and 60% (direct measurement) lower than that of unretted fibers. Interestingly, DIC consistently yielded higher elastic modulus values than direct measurements for both fiber types. This difference, an increase of 22% for water-retted fibers and 8% for unretted fibers, warrants further investigation and discussion, particularly in the context of composite design. Unretted fibers’ inherently higher elastic modulus suggests that composites reinforced with them would likely exhibit greater stiffness. Accurate determination of this crucial property is essential for optimal composite design. Addressing potential measurement errors through larger sample sizes with varied fiber lengths can improve the accuracy of elastic modulus determination.

Table 4 also reveals statistically significant differences in average elastic modulus of fibers due to both water retting and the chosen fiber strain measurement method, as confirmed by the ANOVA analysis, where F-values > F critical values and P-values <0.05.

Table 4. Summary of ANOVA analysis for elastic modulus of water-retted and unretted fibers.

Summary modulus of elasticity [GPa]		Measurement methods
		Direct		DIC
Unretted	Average	44.62		48.33
	Variance	24.12		39.47
Retted	Average	27.92		34.03
	Variance	55.60		52.76
ANOVA	Source of variation	F-value	F critical value		P-value
	Water retting	83.93	4.05		0.000
	Measurement method	5.19	4.05		0.027

Table 5 compares the mechanical properties of unretted sisal fibers obtained in this study with those reported in previous studies. The comparison focuses on DIC-derived elastic modulus and strain at failure values. Notably, the tensile strength and elastic modulus obtained in this study significantly exceed compared to previously reported results. However, the DIC-determined strain at failure is lower than a previous study’s value. This discrepancy can be attributed to the more accurate measurement employed in this study, which avoids fiber slippage errors.

Table 5. Obtained mechanical properties of the sisal fiber with respect to the results reported by previous studies.

Density [g/cm^3]	Tensile strength [MPa]	Elastic modulus [GPa]	Strain at failure [%]	Reference
1.45 1.50 1.40 1.31 1.42 1.33 1.39	718 606 555 297 354 650 584	48 16 16 18 13 38 24	1.5 2.3 3.2 2.5 4.5	(*** Results of this study ***) (Gupta and Singh 2018) (Mohapatra and Kar 2019) (Mydin 2022) (Ndoumou et al. 2022) (Veerasimman et al. 2022) (Neto et al. 2019)

Conclusion

Single-fiber tensile testing using digital image correlation (DIC) provides valuable insights into the mechanical properties of natural fibers during initial material and product development stages. This study compares results from direct displacement measurement with those obtained through DIC. Water-retted fibers exhibited significantly higher strains at failure compared to unretted fibers, regardless of the measurement method. The average strains at failure were 41% (DIC) and 31% (direct measurement) higher in water-retted fibers. This suggests that water retting enhances the fiber’s ductility, allowing it to deform further before breaking. Surprisingly, water retting had negligible impact on tensile strength. While a slight decrease (5.8%) was observed through calculations, it was deemed statistically insignificant.

Table 5 highlights that the tensile strength and elastic modulus of sisal fibers in this study surpassed values reported in previous researches. However, the recorded strain at failure was lower than some previous findings. This discrepancy is likely due to the DIC technique’s ability to accurately capture local strains and avoid errors associated with fiber slippage, potentially leading to a more realistic assessment of failure initiation.

Image analysis using ImageJ revealed that the fibers’ actual cross-sectional area (CSA) was 15% to 19% lower than the calculated values. This discrepancy emphasizes the importance of accurate CSA determination, as it significantly impacts tensile strength calculations.

Highlights

• The sisal fibers were extracted from an agave sisalan plant by manually decorticating method.

• Retting was applied to some decorticated fibers for 30 consecutive days using distilled water.

• Fiber density was determined on unretted and retted fibers to investigate effects of retting.

• Cross sections of fibers were determined by microscopy, calculation and shape tracing methods.

• Single fiber tensile testing using DIC technique was conducted; results revealed that retting has no significant effect on strength but strain of retted fibers was higher than unretted fibers.

Acknowledgments

The authors are pleased to acknowledge the support of the colleagues from the Arenberg and Bruges campuses of KU Leuven. The authors also thank the secretary office of Arenberg campus - KU Leuven and the student services office of Bruges campus - KU Leuven for their support in providing consumable laboratory goods.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Data availability statement

The data used to support the findings of this study are included within the article.

Additional information

Funding

This work was financed by the IUPEPPE project under the framework of collaboration of the Ministry of Education (Ethiopia) and KU Leuven (Belgium).