Introduction

In recent years, sandwich sheets have been implemented in different applications to simultaneously reduce weight and increase rigidity [1]. Sandwich sheets are particularly effective in aircraft and satellites, because weight reduction directly affects energy efficiency and payload [2, 3]. Additionally, the material should be selected as light as possible [4]; and thus, continuous carbon fiber reinforced plastics (CFRP), which have excellent specific strength and stiffness, are considered suitable materials for sandwich sheets [5,6,7,8,9,10,11]. In fact, sandwich sheets comprising CFRP face sheets and an aluminum or Nomex honeycomb core are widely used in the aerospace industry [12]. Flying cars are currently being developed as next-generation mobility, and are expected to be used in delivery, transportation, entertainment, and other applications. They can also be an effective means of relieving traffic congestion in urban areas. To popularize flying cars, it is important to manufacture flying car structures that reduce the weight in a cost-effective manner. One promising solution to meet these demands is the use of CFRP sandwich sheets [13].

However, sandwich sheets require two face sheets and a core adhesively bonded separately, which increases production costs. Furthermore, the adhesive layer causes face sheet delamination, which often leads to failure. The strength of the adhesive layer depends on the properties of the adhesive, and face sheet delamination indicates that the mechanical properties of the sandwich sheet have not been fully demonstrated.

Xiong et al. developed a lightweight sandwich sheet by forming a CFRP truss core via hot press forming and bonding with two CFRP face sheets [14]. However, the results of bending tests revealed that the adhesive area between the core and face sheets was insufficient, resulting in delamination [15]. To solve this problem, Wu et al. installed an aluminum alloy (6061) connector between the truss core and face sheets, and improved the mechanical properties by adding an interlocking mechanism between the connector and truss core [16]. Compared with the experimental results obtained by Xiong et al., this structure suppressed the delamination of the adhesive layer. However, the formation of sandwich sheets using connectors is a costly production process involving several steps.

Zhang et al. developed a new sandwich sheet using steel for the face sheet and CFRP for the core, and proposed a truncated dome core considering the bending strength, rigidity, and mass producibility [17,18,19]. Sandwich sheets with truncated dome cores have excellent specific stiffness and formability, and can be mass-produced by hot stamping. Although the truncated dome core exhibited outstanding properties, it was fabricated empirically based on a conventional core structure without analytical optimization. Shibuya et al. optimized the core structure and verified the superior specific stiffness through bending tests and finite element analyses [20, 21]. In studies by both Zhang et al. and Shibuya et al., formability evaluations indicated that buckling of the face sheet or delamination of the adhesive layer were the main failure modes. However, neither failure mode fully exploits the properties of the sandwich sheets. From previous studies, it can be concluded that the delamination of the adhesive layer is a crucial problem for sandwich sheets.

The purpose of this study is to improve the mechanical properties of CFRP sandwich sheets by eliminating the adhesive layer via integral formation of sandwich sheets. The elimination of the adhesive interface is expected to prevent early failure due to the delamination of the adhesive layer. A 3D printer manufactured by Markforged, MarkTwo™, which is capable of printing continuous CFRP, was used because additive manufacturing is suitable for integral forming. MarkTwo™ is a novel 3D printer capable of printing continuous CFRP; it has a lower cost and shorter production time than that of conventional CFRP molding methods [22,23,24].

One of the advantages of forming sandwich sheets by additive manufacturing is its high design flexibility, where it is expected that the mechanical properties can be improved by optimizing the core shape. However, in this study, we focused on the differences between integral and adhesive forming. Therefore, only a simple corrugated shape is examined as the core structure. Integrally formed sandwich sheets and adhesively formed sandwich sheets were manufactured using additive manufacturing, and subjected to a three-point bending test to compare their mechanical properties. During the test, the strain distribution on the surface of the specimens was measured by digital image correlation (DIC), and the influence of the difference in the forming method was discussed. Moreover, as a step toward optimizing the core shape of the integrally formed sandwich sheets, specimens with different geometrical design parameters were manufactured and the optimal parameters were investigated.

Specimen configuration

The core of the sandwich sheet comprised a repeating arrangement of unit cells with identical geometries, and in this study, the unit cell of the corrugated core was trapezoidal. The geometrical design parameters were set as shown in Fig. 1, where the z axis represents the direction of lamination. The geometrical design parameters of the unit cell were as follows: \(l\) is the length of inclined strut, \(d\) is the length of horizontal strut, \({t}_{f}\) is the thickness of face sheets, \({t}_{c}\) is the thickness of core, \({h}_{c}\) is the height of core, \(\phi\) is the inclined angle of strut, and \(\theta\) is the angle of core in the width direction. Here, \(\theta\) is set so as to suppress the buckling of the face sheet and improve the mechanical properties by aligning the cores along both the width and longitudinal directions. The specimens were uniformly formed with \(d=8\text{m}\text{m}, {t}_{f}={t}_{c}=2.4\text{m}\text{m}\) and \(\phi =55^\circ\). The shear stiffness of the corrugated core is maximized at [25]. Then, \(l\) and \(\theta\) were set to \(l=5\text{m}\text{m}\)and \(3\text{m}\text{m},\) and \(\theta =0^\circ , 30^\circ ,\) and \(45^\circ\), respectively. Six different specimens were prepared for this study. From Eq. (1), the relative density of the cores was calculated to be \(\rho =0.37\) for the \(l=5\text{m}\text{m}\) specimen and \(\rho =0.49\) for the \(l=3\text{m}\text{m}\) specimen.

$$\begin{array}{c}\rho =\frac{{t}_{c}}{{h}_{c}}=\frac{{t}_{c}}{{t}_{c}+l\text{sin}\varphi}\end{array}$$
(1)

Figure 2 shows an integrally formed sandwich sheet. Onyx, comprising short carbon fibers and PA6, was used for the face sheets, whereas a continuous carbon fiber (CCF) filament and Onyx were used for the core. Table 1 lists the material properties of Onyx and CCF. The carbon fiber volume fraction of the CCF was 35% as per manufacturer specifications.

Fig. 1
figure 1

Geometric design parameters of the sandwich sheet

Full size image
Fig. 2
figure 2

Integrally formed sandwich sheet manufactured by additive manufacturing

Full size image
Table 1 Material properties of Onyx and CCF filament [26]
Full size table

The \(l=5\text{mm}, \theta =0^\circ\) specimens were manufactured not only by integral forming, but also by adhesive bonding. The other geometric design parameters were the same as those described above. The face sheet and core were printed separately and adhered using MOS-8 (Konishi Co., Ltd.), a two-component silicone polymer and epoxy adhesive with working temperature ranging from \(-40^\circ\mathrm C\) to \(120^\circ\mathrm C\). Adhesive was cured at room temperature with a curing time of 12 hours.

After specimen preparation, a random pattern of speckles was sprayed onto the surface of the specimens to track the strain distribution using DIC.

Three-point bending test

Figure 3 shows a schematic of the three-point bending test. The jig was made of an aluminum alloy with an indenter radius of 5 mm and distance of 50 mm between the lower indenters. The crosshead speed was set at 10 mm/min. The test was terminated when the stroke reached a depth of 20 mm. A multi stage compression testing machine (Thermecmaster, Fuji Electronic Industrial Co., Ltd.) was used. Six different integrally formed specimens with different geometric design parameters were tested 10 times each, and the adhesively formed \(l=5\text{m}\text{m}, \theta =0^\circ\) specimens were tested three times. The load and stroke were measured during the tests, and the maximum bending load and stroke at the maximum bending load point were compared for each type of specimen. To evaluate the effect of the geometric design of the core on the mechanical properties of the integrally formed specimens, t-tests at a \(5\%\) significance level were conducted with the null hypothesis that the core geometry does not affect the maximum bending load and stroke at the maximum bending load point.

Fig. 3
figure 3

Schematic of the three-point bending test

Full size image

The strain distribution was captured by DIC during a three-point bending test. The strain distributions of the integrally formed and adhesively formed specimens were compared.

Advantage of integral forming

First, the integrally formed and adhesively formed specimens were compared with \(\theta =0^\circ\). Table 2 presents the three-point bending test results of the integrally formed specimens with \(l=5\text{m}\text{m}\) and \(l=3\text{m}\text{m}\) and the adhesively formed specimens with \(l=5\text{m}\text{m}\). Figure 4 shows a typical load-stroke diagram. The test results showed that in the \(\theta =0^\circ\) geometry, integrally formed specimen have higher maximum bending load than adhesively formed specimen.

Table 2 Result of three-point bending tests (integrally formed specimen vs. adhesively formed specimen)
Full size table
Fig. 4
figure 4

Typical load-stroke diagram (integrally formed specimen vs. adhesively formed specimen)

Full size image

Figure 5 shows the strain distribution obtained using DIC. The integrally formed specimen exhibited a typical strain distribution during bending tests, with compressive deformation on the upper side of the specimen and tensile deformation on the lower side. The adhesively formed specimen exhibited a similar strain distribution at the beginning of the test. However, as the test progressed, a localized deformation appeared in the adhesive layer, as shown in Fig. 5. As the test progressed, the strain in the adhesive layer increased, and delamination occurred at approximately \(3\text{m}\text{m}\) of the stroke to induce specimen fracture. This is attributed to the low modulus of ductility of the adhesive layer compared to those of the face sheets and core. The stress singularity, where the adhesive layer becomes an interface, induces discontinuous of stress state. Therefore, the fracture occurred earlier, and a difference in the maximum bending load was observed.

Fig. 5
figure 5

Strain distribution during the three-point bending test

Full size image

Figure 6 shows the appearance of the specimens after the three-point bending test. The adhesively formed specimen clearly showed fracture owing to delamination of the adhesive layer. In the integrally formed specimens, the face sheets and cores fractured simultaneously. Further investigations, including finite element method (FEM) analyses, are required to determine the initiation point and fracture details.

Fig. 6
figure 6

Appearance of specimens after the three-point bending test

Full size image

Effect of core geometries manufactured by integral forming

The effects of the different core geometries on the mechanical properties of the integrally formed specimens were compared. Figure 7 shows a typical load-stroke diagram when the parameters \(l\) and \(\theta\) were changed. Table 3 shows the maximum bending load for different values of \(l\) and \(\theta\). Table 4 shows the stroke at the maximum bending load point, and Tables 5 and 6 show the p-values obtained using t-tests to verify the effects of \(l\) and \(\theta\), respectively. Table 5 shows that for all \(\theta\), the p-values are greater than 0.05 when comparing \(l=5\text{m}\text{m}\) vs. \(l=3\text{m}\text{m}\). Therefore, there was no significant difference in the maximum bending load or stroke at the maximum bending point in a range of \(3\text{m}\text{m}\) to \(5\text{m}\text{m}\). Further investigations on the effect of the parameter \(l\) will be continued through experiments and FEM analysis because it is possible to have some impact on mechanical properties seeing that \(l\)can determine the thickness of the specimen. Table 6 shows that for both \(l=5\text{m}\text{m}\) and \(3\text{m}\text{m}\), the p-value is above 0.05 for \(\theta =0^\circ\) vs. \(\theta =30^\circ\), and below 0.05 for \(\theta =30^\circ\) vs. \(\theta =45^\circ\) and \(\theta =45^\circ\) vs. \(\theta =0^\circ\). Therefore, it can be said statistically that the specimens with \(\theta =45^\circ\) have superior maximum bending load compared to the specimens with \(\theta =0^\circ\) and \(\theta =30^\circ\). As mentioned in the section on the specimen configuration, this can be attributed to the fact that the cores are aligned not only in the width direction, but also in the longitudinal direction, which improves the mechanical properties. Since the face sheet thickness was set to 2.4 mm, which is relatively high, buckling of the face sheet did not occur in any of the specimens.

Fig. 7
figure 7

Typical load-stroke diagram of integrally formed specimens where parameters \(l\) and \(\theta\) were changed

Full size image
Table 3 Maximum bending load\({F}_{max}\)[kN]
Full size table
Table 4 Stroke at maximum bending load point\({U}_{max}\)[mm]
Full size table
Table 5 The p-value calculated by the t-test (compared parameter: \(l\))
Full size table
Table 6 The p-value calculated by the t-test (compared parameter: \(\theta\))
Full size table

In the three- point bending test, the contact between the jig and the specimen affects the results. Furthermore, the specimens are thicker, the effect of shear deformation in the three-point bending test is greater. Therefore, the four-point bending test is considered effective for further investigation in order to rule out those error factors.

Summary

In this study, sandwich sheets were manufactured using additive manufacturing to eliminate delamination of the adhesive layer, which often causes early fracture of the sandwich sheets. The integrally formed sandwich sheet was subjected to three-point bending tests to investigate its mechanical properties. In addition, strain measurements using DIC, and a t-test were conducted. The results were compared with those of an adhesively formed specimen. The following conclusions were drawn.

  1. (1)

    The adhesively formed specimens fractured at the adhesive layer by delamination, whereas the integrally formed specimens fractured at the core and face sheets. This is consistent with the results of the strain distribution measurements using DIC.

  2. (2)

    Integrally formed specimens have superior mechanical properties compared to adhesively formed specimens, because integrally formed specimens avoid early fracture due to delamination of the adhesive layer.

  3. (3)

    In integrally formed specimens, the \(\theta =45^\circ\) specimen had the highest maximum bending load, which attributes to the core alignment.

In this study, only a simple corrugated structure was used for the core, and no optimization was performed. However, one of the major advantages of additive manufacturing is its high design flexibility; thus, appropriate optimization is effective. Therefore, optimization of the core structure to maximize the stiffness, strength, and formability is necessary in the future. In addition, the details of the fracture of the integrally formed specimens need to be further investigated, and the stress state and fracture should be analyzed using FEM.