Damage behavior of potting materials in sandwich composites with pinned joints

1 Introduction

Structural sandwich composites have been widely used in many application areas, such as aircraft structures, ship hulls, wind turbine blades, and bridge decks, due to their very high stiffness-to-weight ratio and also high bending strength-to-weight ratio. They typically consist of two thin, stiff, and strong faces that are separated by a thick, light, and weaker core. Thanks to the core material, the flexural rigidity of sandwich structures/panels is enhanced without any additional weight. Unfortunately, the low strength of the core material causes a problem at the junction with the main structure in mechanically fastened joints.

The recognized methods for joining composite structures to other composites or to metallic parts are adhesive bonding and mechanical fastening. Adhesive joints provide good load distribution and lower cost; however, in many cases, the mechanically fastened joints must be used because of requirements for disassembling the joint to replace damaged structure or to achieve access to underlying structure [1]. Therefore, it is necessary to locally reinforce the weak junction points, in mechanical fastening, with materials stronger than the core alone, called the insert. They transfer normally three types of loads: transverse, shear, and torque (Figure 1).

Figure 1

Three types of loads at junction points in sandwich composites: (A) shear, (B) transverse (tension), and (C) torque.

In the literature, a considerable amount of paper on the mechanics of adhesively bonded joints in composite materials can be found [2–5]. Of these, Banea and da Silva [2] have presented a review of the investigations on adhesively bonded joints of fiber-reinforced plastic (FRP) composite structures considering single-skin and sandwich composites. They noted that the effects of surface preparation, joint configuration, adhesive properties, and environmental factors on the joint behavior have been considered for adhesively bonded FRP composite structures. The analytical and numerical methods of stress analysis required before failure prediction have also been discussed. Although there are also a lot of studies on the mechanical joints/fastening of conventional composites structures [6–8], there are only few published works [9–24] on failure modes and failure mechanisms of mechanically fastened joints in sandwich composites. Thoppul et al. [7] have presented a comprehensive review of the recent literature in the broader area of mechanics of mechanically fastened joints in polymer-matrix composite structures. Their article begins with a review of the relevant mechanical test methods and standards followed by a discussion of the mechanics aspects of design, including joint design methodologies, considerations of the influence of geometric effects, and fastener preload selection. The failure modes, failure prediction for joints, time-dependent joint preload relaxation, effects of temperature and moisture on joint strength and failure, and nondestructive evaluation techniques for monitoring the joints are also summarized. They also offer some useful comments and recommendations on the remaining problems in this area.

Among those of sandwich composites, Bunyawanichakul et al. [10, 11] have done experimental and numerical analyses of inserts in sandwich structures manufactured from carbon fiber/epoxy face sheets and Nomex™ sandwich core. Heimbs and Pein [13] have examined the failure behavior of the different types of potted inserts and corner joints in Nomex honeycomb sandwich structures, experimentally and numerically. The damage process of the samples has been explained by using force-displacement curves and photos of the damaged specimens. Lim and Lee [15] have developed a new insert for assembling satellite sandwich structures by reinforcing the web of insert with high-strength carbon composite to increase the loading capability. The load-carrying capability and the failure modes of the composite sandwich structures assembled with the inserts were measured by the pullout and shear tests and the results were compared with the calculated results obtained from a finite element analysis. Nguyen et al. [17] have conducted the failure analysis of foam core sandwich joints under pullout loading by testing and numerical analysis. The experiments of two different types of joints, with and without inserts, showed that the failure modes of sandwich joints were a combination of different failure modes, such as core shear failure, the debonding of face and the potting material, composite face failure, and local failure. Song et al. [21] have addressed an experimental study on the failure behavior of sandwich composites subjected to the pullout and shear failure load of insert joints. The effect of design variables such as the core height and density, the face thickness, and the insert clearance on the failure loads of a sandwich insert joint was examined. They noted that, during pullout loading, the core density most affected the failure loads. Zabihpoor et al. [23] have presented a finite element model to predict the bearing strength, failure modes, and failure load of bolted joints in foam-core sandwich composites.

In this study, the damage behavior of the potting material itself (i.e., without a metallic insert) under shear (in-plane) loading is examined experimentally. The sandwich composite samples were manufactured by the vacuum-assisted resin infusion technique.

2 Materials and methods

The sandwich composite panels were manufactured by the vacuum-assisted resin infusion technique, as shown in Figure 2. In this technique, the dry fabrics and core materials are placed in an open mould according to the stacking sequence chosen and a vacuum bag is placed on the top of the mould, which is connected with a resin source and a vacuum pump firstly. Then, the liquid resin infuses into the reinforcing fibers, thanks to the vacuum drawn through the mould, followed by the curing and demoulding. In this work, the top and bottom face sheets were of two woven E-glass/epoxy layers. The areal density of the woven fabric was 500 g/m², while the density of the PVC foam core (AIREX^® C70.55) was 60 kg/m³. The composite panels were first cured at 50°C for 0.5 h and then at 90°C for 2 h. The nominal thickness of the panels manufactured was 11.5 mm.

Figure 2

Manufacturing of samples by vacuum-assisted resin infusion process: (A) schematic view and (B) a photo from manufacturing.

The test specimens having constant dimensions of L=135 mm and W=36 mm, as shown in Figure 3, were cut from the sandwich composite panels. Afterwards, the pinholes with and without the potting material were drilled. Two different holes having the diameters of d and D were drilled at the ends of the specimens. For the specimens without the potting material, pinholes with diameter d=6 mm were drilled directly. However, for those with the potting material, the holes of different diameters (D=10, 12, 15, and 18 mm) were drilled followed by the injection of the spotting material that consists of an epoxy resin and hardener system mixed with short E-glass fibers (4 wt%), as shown in Figure 4.

Figure 3

Dimensions of the test samples.

Figure 4

Preparation of the specimens with the potting material.

After curing, at the center of the potting material, pinholes of d=6 mm were finally drilled on all the sandwich composite specimens for the in-plane pin loading. In order to avoid possible damages to the specimens by grip jaws during tensile testing, as seen in the figures, pinholes were drilled at both sides of the samples. The schematic illustrations of a sample and test apparatus are shown in Figure 5. In addition to the diameter of the potting material D, the edge distance E with values of 15, 20, and 25 mm has also been considered as a research parameter. Manufacturing of the sandwich composites, sample preparation and tensile tests were done at Composites Research, Manufacturing and Testing Laboratory, Dokuz Eylül Üniversity, Turkey.

Figure 5

Schematic illustrations of a test sample and tensile test apparatus.

3 Results and discussion

The damage mechanisms of the sandwich composite specimens under in-plane loading due to a pinned connection have been investigated. The damage modes/mechanisms of the samples are explained in the light of the cross-examination of graphs and images of the damaged samples. In Figure 6, the typical damage modes of pinned joints [24] for sandwich composites are illustrated.

Figure 6

Typical damage modes of foam-based sandwich composites with and without the potting materials (PM).

Based on the geometry of the samples, for example, the ratios of width-to-diameter (W/D) and edge distance-to-diameter (E/D), the damage process may occur as the combination of the described modes. In Figure 6A–C, the typical damage modes (net-tension, shear-out, and bearing, respectively) for the samples without the potting material are given. For small values of D, the diameter of the potting material, the common damage mode is net-tension of the potting material itself, as shown in Figure 6D. As the diameter D increases, based on the geometry as indicated earlier, the potting material either sinks into the PVC foam (Figure 6E) or causes the net-tension and shear-out of the sandwich structure (Figure 6F and G), respectively.

Since the damage mechanisms of a pinned joint have been presented adequately in the literature, the damage modes of the samples with the potting material have been the focus of this paper particularly. Therefore, only the “bearing” mode given in Figure 6C, in which D=d=6 mm, is considered herein for comparison and discussion, together with the damage modes of the samples with the potting material. In Figure 7, the load-displacement variation and image of a damaged sample for E=25 mm and D=6 mm are given. Since the dimensions “a” and “b” are high enough, the bearing mode is observed as expected, rather than the net-tension and shear-out.

Figure 7

Load-displacement variation and image of the damaged sample for E=25 mm and D=6 mm, no potting material.

For each configuration of the samples with the potting material, the tensile tests were repeated three times. In Figure 8, the load-displacement variation and images of samples with E=15 mm and D=10 mm are given. The ascending sections of the curves are nearly of the same slope, indicating that the samples are of the same stiffness at the beginning of the tests. The horizontal offset of the ascending section of sample 1510-1 stems from the clearance existent between pin and hole in the tensile test; the load does not increase until a contact takes place. Here, sample 1510-1 represents the coding of sample 1 with E=15 mm and D=10 mm. As seen from the images, for samples 1 and 2, the potting material undergoes net-tension firstly, followed by the bearing and shear-out of the sandwich structure. However, for sample 3, the damage is initiated by the debonding of the interface between the potting material and the core material. For this, sample testing is stopped after the first failure to avoid following damage modes, for a better inspection. Similar damage mechanisms are observed for the configuration of E=15 mm and D=12 mm as well. In Figures 9 and 10, for comparison, the load-displacement curves and images of the damaged samples with E=15 mm and D=15 and 18 mm are given. As seen in Figure 9, the interface debonding between the potting material and the sandwich sample is followed by the shear-out and net-tension of the sample, for sample 1515-1. In samples 1515-2 and 1515-3, the crack of the potting material and shear-out of the sandwich part take place. Similar damage mechanisms can be observed in Figure 10.

Figure 8

Load-displacement variations and images of the damaged samples for E=15 mm and D=10 mm.

Figure 9

Load-displacement variations and images of the damaged samples for E=15 mm and D=15 mm.

Figure 10

Load-displacement variations and images of the damaged samples for E=15 mm and D=18 mm.

Considering the average values of the three tests for each configuration, the maximum loads carried by samples are given in Figure 11. As seen in the figure, for the samples without the potting material that is D=6 mm, the maximum failure load is the smallest. With the increase of D, it seems that the maximum load values are increased dramatically. One of the reasons is probably that, with the increase of D, the through-thickness contact surface between the potting material and face sheets transferring load increases. For sufficiently high values of “a” and “b”, as shown in Figure 3, the samples carry higher loads. For example, for E=25 mm and D=18 mm, a threefold increase in maximum load is observed compared to the samples without the potting material, D=6 mm.

Figure 11

Average maximum load values carried by the samples.

Since the load-displacement characteristics are similar to those given above, the images of the damaged samples are only given in the following for the other configurations. Since the diameter of the potting material is small, for all samples of the configurations 2010 (E=20 mm and D=10 mm), 2012, 2510, and 2512, the first failure generally took place as net-tension of the potting material followed by the bearing of sandwich composite.

The images of some damaged samples for E=20 and 25 mm and D=15 mm are given in Figure 12. It is seen that the damage mechanisms are significantly based on the first damage mode occurred. For sample 2015-1, it is observed that the bearing of the face sheet is followed by the crack of the potting material. However, for sample 2015-2, the potting material first sinks into core in the loading direction along with the core crack at the potting material/core interface at opposite side followed by the net-tension and shear damage of the sandwich part. The main reason of the sinking of the potting material into core is probably the smaller thickness of face sheets in this study. If the thickness of the face sheets increases sufficiently leading to a larger through-thickness contact surface between the potting material and face sheets, as shown in Figure 13, this damage mode may not be observed, resulting in probably a higher load-carrying capacity of the sandwich structure. For E=25 mm, since “a” is high enough, shear-out and net-tension are not observed. The images of some damaged samples for E=20 and 25 mm and D=18 mm are also given in Figure 14 for comparison.

Figure 12

Images of the damaged samples of configurations E=20 and 25 mm and D=15 mm.

Figure 13

Schematic of the through-thickness contact surface between the potting material and face sheets: (A) for small face-sheet thicknesses and (B) for larger face-sheet thicknesses.

Figure 14

Images of the damaged samples of configurations E=20 and 25 mm and D=18 mm.

4 Conclusion

In this paper, the damage behavior of the potting material around a pinhole, which may be used in a mechanical joint of sandwich composites, under in-plane loading is investigated experimentally. From the results, it is noted that, by using the potting material, the failure load of such mechanical joints may be increased significantly. Provided that the sample dimensions W and E are large enough, the increase of the diameter D results in a higher load-carrying capacity. Larger through-thickness bonding/contact surface between the potting material and face sheets may help to provide the potting material from sinking into core as the first damage mode. For the same purpose, bridging layers consisting of woven glass fibers and epoxy would also be used at the top and bottom surfaces of the potting material when the samples with the potting material are prepared. That is, this topic needs further investigation considering the different parameters such as the stacking sequence and the thickness of face sheets and the length of short fibers used in the potting material by using metallic inserts together with the potting materials for justification, before any use in engineering applications.