Quantitative evaluation of the sheared edge of woven glass epoxy laminate after mechanical punching

Fiber-reinforced composites have a wide range of industrial applications owing to their light weight, high modulus, and good specific strength. Trimming using a mechanical punching tool is the fastest way to post process fiber-reinforced composites. However, quantitative analyses of the defects and process parameters are required to analyze the damage mechanisms and achieve high-quality cut surfaces. In this study, we quantitively investigated the defects on the sheared edge of woven glass epoxy laminates subjected to mechanical punching and explored the correlations between these defects and key process parameters, including punch-die clearance, weave alignment angle, and laminate thickness. The results demonstrate the necessity of considering glass fiber burr length and the dimensions of the cracked area simultaneously to evaluate the quality of the sheared surface. This study provides qualitative data on the interfacial damage mechanisms that occur during mechanical punching, which will help to tailor the process parameters of mechanical punching for fiber-reinforced polymers to obtain smooth cut surfaces.


Introduction
Fiber-reinforced polymer matrix composites [1] are widely used in the aerospace [2], automotive [3], construction [4], and biomedical [5] industries owing to their light weight, high modulus, and good specific strength. The application scope of fiber-reinforced polymers has expanded from large components such as aircraft wings and ship hulls to smaller, more delicate products such as unmanned aerial vehicles and mobile devices, which have more complex shapes and require greater dimensional accuracy. To satisfy this demand, post-processing methods for shaping fiber-reinforced polymers have attracted attention [6]. However, it is difficult to improve the dimensional accuracy of fiber-reinforced polymers by post-processing owing to the nonuniform nature of the fiber reinforcement. In addition, because of the abrasive nature of the reinforcing fibers, the usable lifetime of the tools is short. Accordingly, the productivity is relatively low, leading to high manufacturing costs. This, alongside the high price of raw materials, hinders the application of composite materials in products that require high dimensional accuracy. Therefore, it is necessary to improve the productivity of manufacturing and post-processing processes for fiber-reinforced polymers.
Several machining processes have been applied in the post-processing of fiber-reinforced polymers, including drilling, milling, turning, and computer numerical control (CNC) processes such as CNC milling, abrasive waterjet cutting, and laser ablation. Drilling is often used to produce holes for fastenings. However, the vertical movement of the drill can cause fiber interruption, stress concentration, delamination, resin erosion, and fiber pullout [7][8][9][10]. For CNC milling of polymer composites, the effects of cutting force and rotational speed on the surface finish have been widely studied [11][12][13]. The quality of the cut surface is determined by factors such as tool movement speed, nose radius, and wear [14][15][16]. These processes already have relatively long machining times because the desired shape is achieved by movement of the cutting tool; however, the constraints on tool speed limit how much the processing time can be shortened to improve productivity.
Mechanical trimming using a punch and die has been proposed as an alternative method of shaping fiber-reinforced polymers. Trimming using a mechanical punching tool is the fastest post-processing method, because the desired shape is achieved in a single piercing motion. This has benefits in terms of cost and productivity. Several researchers have explored ways to improve the quality of the sheared edge of fiber-reinforced composites trimmed by mechanical punching. Yokoi et al. [17] applied vibration to the punch to maximize the shear heat generated in the piercing step and achieved smooth sheared edges on various fiber-reinforced composite materials. Nojima [18] analyzed the effect of punch shape on the punching load acting on aramid-fiberreinforced polymers and found that the deformation load was concentrated on the slug when using a hollow punch. This effectively improved the quality of the sheared edge and produced a smooth cross section. Klocke et al. [19] experimentally investigated the relationship between punch diameter and dimensional and form accuracy during the circular piercing of carbon fiber-reinforced polymers. They reported that the punch diameter had a significant influence on the dimensional and shape precision and that the process parameters affecting the cutting resistance of carbon fiber-reinforced laminates were similar to those when shaping metals. Lambiase and Durante [20] performed tensile tests on thin glass fiber-reinforced laminates with a central hole to compare how punched and drilled holes affected the mechanical properties. They found that the mechanical properties were similar under a relatively small punch-die clearance of 0.1 mm, but as the clearance increased, the mechanical properties of the punched specimens decreased sharply. These studies indicate that the quality of the sheared edge can be improved by optimizing the punch shape and applying vibration. However, special equipment and tools are required to achieve stability and durability for mass production, which results in increased costs.
Researchers have identified several damage mechanisms that occur owing to mechanical loading during the mechanical punching of cross-ply laminated fiber-reinforced composites, including fiber-matrix interfacial debonding, matrix cracking, delamination between plies, and fiber breakage [21,22]. These phenomena have been analyzed based on the surface roughness and delamination factor. However, quantitative analyses of the defects and process parameters are required to analyze the damage mechanisms that occur during mechanical trimming and obtain smooth cross sections. In addition, there is a lack of quantitative data on interfacial damage mechanisms such as debonding, delamination, and matrix cracking.
This study quantitively investigates the defects on the sheared edge of woven glass epoxy laminates subjected to mechanical punching and explores the correlations between these defects and key process parameters (including punchdie clearance, the most critical variable) in relation to the interfacial damage mechanism. Mechanical punching was conducted on a machine press so that the defect mechanisms could be evaluated under the same conditions as those used in mass production. The results will help to tailor the process parameters of mechanical punching for fiber-reinforced polymers to obtain smooth cross sections, as well as providing qualitative data on the interfacial damage mechanisms that occur during mechanical punching.

Materials
Glass fabric prepregs with a plain weave (ply type 7628, epoxy resin 45-51 vol%, cured ply thickness 180-200 μm) were used to prepare 2-, 4-, and 8-ply laminates with nominal thicknesses of 0.5, 0.8, and 1.6 mm, respectively. Specimens were manufactured by prepreg compression molding process; the specimens were cut to a size of 40 × 120 mm and used in the experiment.
The woven glass epoxy laminate data are shown in Table 1. A hot-press laminating cycle was conducted as follows. The laminate was heated from 70 to 150 °C for 40 min under vacuum (40 torr) and a load of 7 kg/cm 2 , and the pressure was increased to 25 kg/cm 2 when the material temperature reached 100-110 °C. The temperature was then maintained at 200 °C for 100 min with the load lowered to 7 kg/cm 2 at 80 min. The process was completed by cooling to 60 °C for about 70 min. The heating rate between 80 and 130 °C was 1.3-1.9 °C/min, and the curing condition was holding above 165 °C for 90 min.

Mechanical trimming process
A Komatsu 200-ton servo-press was used as the machine press in the mechanical trimming experiments, and punching was conducted with a cutoff die structure. The apparatus and configuration of the machine press are shown in Fig. 1a. The cutoff dies have an open curve shape, which can interfere with trimming process because the punch is pushed by lateral pressure. Therefore, to prevent the punch being pushed by the lateral force and precisely control the clearance, a backup block was installed.
The backup block also acts as a stopper that determines the amount of material to be cut. The parts were fabricated as insert types and a fixed stripper type was adopted. The punch and die insert (made of KS D 3522, SKH51) both had an HRC (hardness on Rockwell scale C) of approximately 63, and the other parts (made of KS D 6753, STD11) had an HRC of 58.
To quantitatively analyze the effect of mechanical trimming on the glass weave epoxy laminate specimens, two major process parameters were varied, namely, punch-die clearance (C) and weave alignment angle. Schematics of the punch-die clearance and weave alignment angle are depicted in parts b and c of Fig. 1, respectively. Clearance refers to the gap between the punch and die and is generally expressed as a ratio of the thickness of the material. In this study, considering the characteristics of the multilayer structure, the clearance was set as a fixed dimension (0.03, 0.06, 0.09, 0.16, 0.32, or 0.5 mm) rather than a ratio of material thickness. The weave alignment angle refers to the angle between the warp of the glass-fiber fabric and the punching tool trim line. Alignment angles of 30°, 45°, and 90° were used, because the glass-fiber fabrics were woven from weft and warp fiber bundles. During mechanical trimming, the alignment angle affects the distribution and transfer of load to the weft and warp fibers. Consequently, the alignment angle affects the shear load applied to the fiber bundles, which is expected to correlate with the formation of burrs. If the trim line was parallel to the weft direction (alignment angle of 90°), the matrix between the weft fiber bundles would be separated by the load, causing the load to concentrate on the warp fibers. Therefore, the shape of the pulled-out fibers after shearing may differ from the original shape.
Manufacturers generally assess the quality of the trimming process based on the state of the glass fibers remaining on the cut surfaces of the laminate. In addition, the formation of the sheared edge is assumed to cause cracking. Therefore, glass-fiber burrs and cracks on the cut surface were selected as significant defects for quantitative analysis, as shown in Fig. 1c. The burr length (B lb ) was defined as the length of glass-fiber protrusions from the sheared edge, and the crack length and depth (L c and D c , respectively) were defined as the maximum dimensions of the cracked area in the horizontal direction (i.e., from the sheared edge to the farthest edge of the crack) and vertical direction (i.e., from the top of the die to the deepest point of the crack), respectively. Each experiment was repeated three times to increase Fig. 1 a Mechanical trimming apparatus and configuration. b Alignment angle of trimming punch and woven glass epoxy laminate. c Schematics of mechanical trimming of wovenglass epoxy laminates the reliability of results. An optical microscope was used for measurements. The specimens were mounted in epoxy resin and cut to the same size using an abrasive cutter before polishing with aluminum oxide polishing powder.

Glass fiber burrs after mechanical trimming of woven-glass epoxy laminate
To quantitatively analyze the effect of punch-die clearance on the formation of glass fiber burrs along the cut surface of the woven glass epoxy composite laminates, experiments were conducted with several clearance depths and laminate thicknesses. Figure 2a shows the lengths of the glass fiber burrs that formed with different clearances and laminate thicknesses. There was no clear trend in glass fiber burr length (B lb ) with clearance. Even for laminates of the same thickness, there was no clear increasing or decreasing trend of B lb with increasing clearance, and the error range was wide. It was not possible to infer the cause of this result. An exception to this is the clear decrease in B lb for the 8-ply laminate as the clearance increased from 0.03 to 0.16 mm; however, it was difficult to see a similar trend under other conditions. Comparing the average B l to laminate thickness indicated that B l increases with laminate thickness. The average B l values for the 2-, 4-, and 8-ply laminates were 127, 191, and 243 μm, respectively. Consequently, it was concluded that B lb is slightly greater in thicker materials. However, the correlation between clearance and B l was not confirmed. Although punch-die clearance is considered a significant parameter in mechanical trimming, it was difficult to prove the effect of clearance on the glass fiber burrs. Nevertheless, glass fiber burrs are an important part of quality assessments of trimmed woven glass epoxy composite laminates. Regardless of the clearance used, B l was irregular with a wide distribution range, making it difficult to determine a trend. Therefore, B ll was analyzed based on the weave alignment angle, shown in Fig. 2b.
As shown in Fig. 2b, there was no consistent trend in B l with alignment angle. When analyzed based on the mean values, B l seemed to decrease slightly at the alignment angle of 45°. However, it is difficult to generalize this because the error was greater than the interquartile range. B l values of a similar range were observed at alignment angles of 30° and 45°. At 90°, the range of error was noticeably larger than that at 30° and 45°, presumably because the punch load acted on the warp and weft fibers. Nevertheless, this result is insufficient to explain the results of B l . The cross sections of the trimmed specimens were analyzed by morphological analyses, which did not reveal any apparent differences in B l at different alignment angles. Accordingly, it was concluded that the formation of glass fiber burrs was not affected by the alignment angle between the shear line and warp direction.
As shown in Figs. 2a and b, in mechanical trimming, B l is not significantly affected by the clearance or alignment angle (i.e., the major process parameters). Additionally, there are serious deviations in B l depending on the position on the cut surface, even within a single sample. Thus, B l is not suitable to be used as an index for assessing the cut surface quality of mechanically trimmed glass-fiber laminates. Figure 3 shows the typical trimmed edge of laminates according to the controlled die clearance at three different Fig. 2 Influence of a punch-die clearance and b alignment angle on glass fiber burr length view angles. It is difficult to characterize and analyze sections of the sheared edge of woven glass epoxy laminates, because the sheared edge has an irregular shape. However, when measuring B l , fracture of the aggregated layers was observed after delamination of the reinforcing fibers. This was assumed to be related to crack formation. Therefore, the effect of clearance was analyzed by measuring the maximum length and depth of the cracked area of the laminate (L c and D c , respectively).

Analysis of crack penetration in sheared edge
As shown in Fig. 4a, there was no direct correlation between clearance and D c . Interestingly, for the 8-ply laminates, an increase in D c was observed as the clearance increased from 0.16 to 0.5 mm. This is consistent with the irregularity in B l under the same conditions. An increase in D c correlates to an increase in the number of separated layers. Consequently, it was inferred that the irregular B l value was caused by the detachment of the composite layers, rather than a reduction in the occurrence of glass fiber burrs. In For each material thickness, D c appeared within a certain range regardless of the clearance used. Therefore, there is a tendency for burr generation to increase in thicker materials. The average D c values for the 2-, 4-, and 8-ply laminates were 231.2, 353.4, and 634.7 µm, respectively. This indicates that the laminate thickness has a greater effect on D c than clearance. In contrast, L c clearly increased as the clearance increased. Even though the trend of the error range according to the clearance for each material thickness was irregular, the interquartile range was increased at clearances of 0.16 and 0.32 mm. As a result, it was confirmed that both D c and L c varied within a certain range. It was concluded that this range is dependent on the material thickness. Accordingly, the experiment results were converted into ratios of the laminate thickness to compare all conditions equally while excluding the effect of thickness. Figure 5 shows the trend line, 95% confidence band, and 95% prediction band of the D c and L c values as a ratio of the laminate thickness. D c had an adjusted R 2 value of 0.01, demonstrating that there was extremely low or no correlation between D c and clearance. Therefore, the distance that the crack penetrates perpendicular to the cutting plane has very low correlation with clearance. This agrees well with the results in Fig. 4a. In contrast, L c had a very high adjusted R 2 value of 0.8. This demonstrates that L c has a strong linear relationship with clearance. In other words, the horizontal growth of cracks increases as the clearance increases. It was hypothesized that this results from the effects of fiber pullout. Clearance is usually described as the gap between the punch and die. However, it can also be described as the local area between the punch and die that induces laminate deformation. Therefore, the increase in clearance induces delamination and increases L c in the same way as the expansion of the deformation region. In addition, when the clearance is greater than the laminate thickness, the deformation pattern is different because the cutting force imparts bending deformation rather than shear deformation. Accordingly, D c is affected by the material thickness, and the crack growth range is limited by the material thickness and punch-die clearance. It is possible to establish a predictive model according to the linear relationship between L c and clearance.

Morphological analysis
To analyze the shape and defects of the sheared edge, morphological analysis was conducted on the sheared edges of the samples. The top view and side view of the sheared edge of an 8-ply woven glass epoxy laminate after mechanical trimming with a clearance of 0.16 mm and alignment angle of 90° are shown in parts b and c of Fig. 6, respectively. The shape of the specimen was analyzed using a noncontact measurement method. The sheared edge is a continuous plane, as shown in Fig. 6a. Irregular protrusions are visible on the sheared edge. These protrusions are the glass fiber burrs, which are remnants of the weft fibers that were broken by the descent of the punch. In addition to the glass fiber burrs, thin and elongated whisker burrs were also observed along the sheared edge. The appearance of these whisker burrs varied from thin threads to thick bundles where dozens of fibers had been pulled out of the epoxy matrix (Fig. 6b). Whisker burrs appeared more frequently when the alignment angle was 90°, because the warp fiber bundles were not fixed to the matrix and were separated by the punch cutting in the direction parallel to the warp. This was the cause of the wide error range of burr length at an alignment angle of 90° (Fig. 2b). Figure 6c shows a depth color map of the sheared edge from the side view. This color map compares the relative heights of the protrusions and cracks. The maximum height was for a glass fiber burr that protruded 0.177 mm from the cut plane. On the other hand, the lowest point is − 0.303 mm. This value corresponds to the maximum crack length of the sample. The glass fiber burrs were generated at the boundary between the relatively smooth cross section and the separated slug layer. The color map in Fig. 6c confirms that burr generation occurs intensively in the center of the material. These results are the basis for the decreasing trend of glass fiber burr length for the 8-ply laminate as the clearance ranges from 0.16 and 0.5 mm, as discussed in Sects. 3.1 and 3.2. The upper part of the specimen exhibits a relatively smooth cross-sectional shape because the punch load is smoothly transferred. However, glass fiber burr generation occurs in the center of the specimen because of matrix-fiber debonding by crack penetration, and the reinforcement is forcibly broken. More plies fracture as the clearance increases because of the advanced debonding time of the matrix and fiber reinforcement, separating the slug layer before glass fiber burrs are formed. Therefore, the burr length tends to decrease, but the crack depth tends to increase. Consequently, it is necessary to consider glass fiber burr length and the dimensions of the cracked area simultaneously to evaluate the quality of the sheared surface.

Conclusions
In this study, the characteristics of the sheared edges of 2-, 4-, and 8-ply woven glass epoxy composite laminates after trimming using a mechanical punching tool were studied, and the effect of different processing parameters on the quality of the sheared edge were investigated. The equipment and experimental settings employed in this study are the same as those used in industrial mass production, and the variables are applicable to actual situations. Glass fiber burrs were formed on the sheared edge, which are the main defects of mechanically trimmed glass fiber laminates. The length of these burrs, which is the size of the normal-direction component of the sheared edge of the glass fiber, was measured with different punch-die clearances and alignment angles of the woven glass fabric.
The glass fiber burr length increased linearly with laminate thickness. However, it was difficult to confirm the effects of clearance and alignment angle on burr length, as large deviations occurred, even within a single specimen. This deviation was associated with the position on the sheared surface and the multilayer characteristics of the composite material. Consequently, glass fiber burr length is not a suitable parameter for assessing the quality of mechanical trimming.
Morphological analysis of the sheared edge using a noncontact measurement method identified oblique whisker burrs, particularly for the specimen sheared with an alignment angle of 90° (i.e., parallel to the weft), which resulted in increased error in the glass fiber burr length measurements. However, these whisker burrs otherwise did not significantly influence the experimental results. Because these whisker burrs can be removed in subsequent processes such as cold cleaning, the appearance of burrs in final laminate products may differ from those observed in this study.
Among the experimental results, a decrease in glass fiber burr length was observed for the 8-ply laminate when the clearance was 0.16-0.5 mm, which was confirmed to be a consequence of the increased crack depth in this clearance range. There was no substantial reduction in the occurrence of glass fiber burrs under these conditions. However, the measured length tended to decrease because more layers were slugged off under these conditions. These results demonstrate the complexity of evaluating defects on the sheared edges of mechanically trimmed woven glass epoxy composite laminates. A complex consideration of the glass fiber burr length, crack depth, and crack length is required.
In summary, minimizing the clearance does not reduce glass fiber burr length. However, it can be effective for reducing the crack penetration length. Selecting a suitable punch-die clearance with consideration to the tool's durability is critical to meet conflicting requirements.