Low velocity impact characteristics of 3D integrated core sandwich composites

Abstract

In this paper, the low velocity impact characteristics and impact damage of sandwich composites, produced at four different core thicknesses from 3-dimensional (3D) integrated sandwich fabrics, with and without foam filling, have been examined. The 3D sandwich fabrics have been produced using the same yarn and weaving densities. Thus, the impact characteristics are only affected by the core thickness and whether foam filling is used or not. Low velocity impact tests have been conducted at 32 and 48 J energy levels. The impact behavior has been determined as a function of the peak load, the energy to peak load, the time to peak load and the absorbed energy. The impact damage and the change in the compressive strength after impact have been analyzed. The findings obtained indicate that core-skin delamination on 3D sandwich composites has been fully prevented. Impact tests carried out on integrated 3D sandwich structures have shown that impact damage is limited to the vicinity of the point of impact and does not affect the integrity of the structure. This indicates that such damage can be easily repaired and the service life of the product can be sustained.

Keywords

3D integrated sandwich low velocity impact damage tolerance

Introduction

Sandwich materials, which are produced by bonding stiff and strong skins to a low-density core material, are commonly used in engineering applications since they combine low weight with high flexural properties, in comparison with conventional materials.¹ Also thermal insulation, acoustic damping, flame retardance, ease of production, high specific strength and modulus, good fatigue performance and corrosion resistance can be pointed out as advantages of sandwich materials. On the other hand, the most significant disadvantage of sandwich materials is their susceptibility to damage development when subjected to concentrated out-of-plane loads, indentation loads and impact loads. Particularly, the drop of any tool or object during production or maintenance, impact effect of stones or other solid matter projected from the tires of other vehicles on a high road, and even hailstorms can be listed as specific examples. Impact originated damage can cause a severe reduction in the strength of these structures and damage characteristics are considerably different than those of monolithic composites.^2,3 Consequently, much research is carried out on the impact characteristics of sandwich composites.

Low velocity impact on sandwich structures causes damage to the composite skin, the core material and at the core-skin interface.⁴ This damage results in a significant decrease in the load bearing capacity of the structure. The type and severity of the damage depends on the properties of the core and skin materials, and the method of binding between them. Most studies on low velocity impact of sandwich materials are focused on honeycomb sandwich materials,^5,6 which are very sensitive to skin-core delaminations.

Foam cores are widely used in sandwich composite production. The most prominent advantage of foam cores is the large contact area for bonding the skin and core materials. Properties of foam core sandwich materials have been examined by many researchers.⁷^–⁹ The behavior of the foam core sandwich composites depends on the density and the mechanical properties of the foam.

Several studies asserting innovative ideas for improving the impact resistance of foam core sandwich materials have been carried out.^10,11 In these studies, the hollow sections of honeycomb or corrugated structures are generally filled with foam or the integration of the structures is improved using z-pins in foam core structures. Some improvement has been obtained in these studies, yet the skin-core delamination problem could not be solved except by using z-pin structures. Also the production processes of the structures stated in these studies have become more difficult and production costs have increased.

3D textile production methods have led to interesting developments in the composites industry.¹²^–¹⁴ The use of 3D integrated fabrics in the production of sandwich composites has presented a new concept on this topic.¹⁵^–¹⁷ 3D integrated sandwich fabrics are made by the velvet carpet weaving technique, where two parallel skins (top and bottom skins) are woven together using pile yarns,¹⁶ keeping a defined distance between the skins to form a core. This integrated connection provides a through-the-thickness reinforcement, in which the pile yarn architecture increases the shear rigidity, which is the most significant disadvantage of many core materials.^18,19 The warp and weft yarns constitute the skins while the pile yarns create the hollow core section, the pile yarn free length determines the core thickness. This sandwich fabric structure has the following advantages: (1) sandwich panels can be produced in a single step, production costs are reduced in line with the shortened production periods; (2) top and bottom skins are integrally woven together with the core section and this ensures a stronger binding between the layers, skin-core delamination is virtually impossible; (3) the hollow core section can be filled by several means, thus different functional capabilities can be given to the structure.

The impact resistance of 3D integrated sandwich composites has been examined by a number of researchers. Vaidya et al.¹⁷ have examined different types of integrated hollow composite sandwich structures. Because the skin layers were very thin in this study, the low velocity impact resistance is low. Hosur et al.²⁰ have examined low velocity impact characteristics on 3D integrated composite structures with different skin layers and with foam in the core section. Their study concludes that the foam core and the 3D structure support each other. While the foam constitutes a vertical support for the 3D core, the integrated 3D core forms a structure that prevents the foam core from breaking up and secures the structural integrity.²¹

There is an increasing interest shown towards 3D integrated sandwich composites and this requires that the impact properties of these materials are identified comprehensively. In this paper, the low velocity impact characteristics and impact damage of sandwich composites, produced with four different core thicknesses from 3D integrated sandwich fabrics have been examined, with and without foam filling. The same production parameters were used for producing the 3D fabrics, i.e. the same yarns and weaving densities and weaving architecture. In this way, the impact properties are affected only by the core thickness and the presence of foam. Low velocity impact tests have been conducted at energy levels of 32 J and 48 J. Impact behavior has been determined as a function of the peak load, the peak load energy, the duration to reach the peak load and the absorbed energy. Impact damage was examined with optical microscopy.

Materials and test methods

3D integrated fabrics

The 3D sandwich fabrics have been supplied by Parabeam BV (NL). The 3D integrated fabrics used in the research had a core thickness of 10, 15, 18 and 22 mm. All four fabrics are identical in terms of their top and bottom skin layers, the yarns used and the weaving architecture. The only difference is their free pile yarn length and thus the core thickness. Figure 1 illustrates the layout of the yarns in the structure, and a typical weaving architecture of a 3D integrated sandwich fabric. According to this, there are binding yarns between the top and bottom skins of the structure in thickness direction and by means of this, the core of the fabric becomes hollow. The parameters of the fabrics used in the study are given in Table 1. All yarns consist of type E-glass fibers. The linear density of the warp, weft and pile yarns is 300 tex.

Figure 1.

Cross-sectional view of 3D dry fabric (a) and resin impregnated plate (b); the schematic display of the 3D fabric structure (c).

Table 1.

Main properties of the composite plates

Core thickness (mm)	3D dry fabric (g/m²)	Total dry fabric (g/m²)	Fiber volume fraction (%)	3D composite core thickness (mm)	3D composite plate thickness with additional face sheets (mm)
22 mm	1680	4680	36%	22.1	24
18 mm	1720	4720	36%	18.0	19.5
15 mm	1600	4600	36%	14.9	16.3
10 mm	1430	4430	35%	10.1	12

The structure of 3D integrated sandwich fabrics is different from conventional woven fabrics.²² Therefore, the fabric geometry was established based on the yarn intersections within the fabrics, to understand the fabric structure and geometry. Then the fabric structure has been established on the basis of the parameters taken from fabric structure such as yarn width, yarn thickness, weave densities and the distance between yarns. The drawings have been carried out with the Katia CAD program. The schematic display of the 3D fabric structure is presented in Figure 1c.

With the purpose of improving the impact resistance of the sandwich composite panel, the top and bottom skins have been reinforced with an additional plain weave E-glass fabric. Three additional plain woven fabric layers of 500 g/m² were used in both the top and bottom skins during composite production.

Fabrication of 3D integrated core sandwich panels

3D integrated sandwich composites were produced from 3D integrated sandwich fabrics. An Atlac 580 AC 300 type vinylester urethane resin was used. A 6% cobalt naphthalate accelerator was added in a ratio of 0.25%. A 50% active methyl ethyl ketone peroxide catalyst was used with a mixing ratio of 2%. The vinylester resin has a Young’s modulus of 3 GPa and 3.4% strain to failure.

All test panels were manufactured using the hand lay-up technique. Production of the panels has been carried out on a glass plate, treated with a release agent. Special care was taken to control the fiber volume fraction during the production of each panel. After applying the resin to the skins from both sides, it has been ensured that the resin thoroughly penetrated the entire fabric and that all entrapped air was removed using aluminum impregnation rollers (Figure 2a and 2b).

Figure 2.

Fabrication of 3D integrated panels (a and b) and view of finished plates (c).

After the application of the resin, the fabrics have been kept for four hours at room temperature for curing. During this time, the impregnated fabric opens up due to the way the pile yarns are woven into the skins. The top and bottom skins of the panel have been reinforced with an additional plain weave glass fabric in order to provide sufficient impact resistance. The extra plies have been bonded to the top and bottom skins after the 3D panels have cured with the same resin by using hand lay-up. The additional skins cannot be co-laminated as their weight would affect the core thickness. Therefore, the additional fabric layers are laminated onto the sandwich panel after the curing of the 3D sandwich panels. Subsequently, the panels were kept for one day for complete curing and specimens were cut from the plates. The appearances of the plates are shown in Figure 2c and 2d.

The weights of the plates and the weight ratio of fiber and resin of each plate were determined. Since the core sections are hollow, the fiber weight ratios have been determined rather than fiber volume fraction. Table 2 shows the resin and fabric ratios separately for plates of different thicknesses.

Table 2.

Finished properties of sandwich samples and their sample codes

Core thickness, mm	Sample code	Foam filling	Density of foam filling, kg/m³	% Weight ratio of fabric/resin/ foam
10	A1	No	−	64/36/0
	A2	Yes	100	57/32/11
15	B1	No	−	65/35/0
	B2	Yes	100	56/29/15
18	C1	No	−	65/35
	C2	Yes	100	51/27/22
22	D1	No	−	64/36
	D2	Yes	100	51/24/25

Foam filling of 3D integrated core sandwich composites

Some of the panel core sections were filled with a rigid polyurethane (PU) foam, consisting of a mix of 1.08 g/cm³ polyol and 1.23 g/cm³ isocyanate, in a 107/100 isocyanate-polyol mixing ratio (by weight). The foaming process was performed in a mold. During the foam expansion, a pressure of approximately 0.5 bar was built up. A constant temperature of 50°C was maintained throughout the reaction to increase the penetration and the homogenity of the foam. The process of injecting foam into the composite plates in the mold is shown in Figure 3. The appearance of the obtained plates with and without foam filling are shown in Figure 3d. In the end, samples of four different thicknesses, with and without foam filling, have been obtained. The names of the produced 3D sandwich samples are given in Table 3.

Figure 3.

Fabrication steps of foam filling process (a and b) and view of foamed and unfoamed panels (c).

Table 3.

Impact test results of unfoamed samples at 32 J and 48 J energy level

Samples	10 mm-UF-32 J	15 mm-UF-32 J	18 mm-UF-32 J	22 mm-UF-32 J	10 mm-UF-48 J	15 mm-UF-48 J	18 mm-UF-48 J	22 mm-UF-48 J
Initial velocity (m/s)	4.27 ± .02	4.27 ± 0.00	4.28 ± 0.00	4.28 ± 0.01	4.41 ± 0.05	4.33 ± 0.01	4.32 ± 0.01	4.32 ± 0.02
Final velocity (m/s)	1.87 ± 0.32	1.94 ± 0.16	0.52 ± 0.24	2.27 ± 0.09	1.39 ± 0.48	1.47 ± 0.21	1.94 ± 0.15	1.94 ± 0.13
Max displacement (mm)	14.36 ± 2.8	14.67 ± 0.9	17.48 ± 0.7	23.2 ± 0.2	16.1 ± 1.42	18.25 ± 0.49	19.7 ± 1.12	26.6 ± 0.7
Initial kinetic energy (J)	29.57 ± 0.28	29.58 ± 0.02	29.64 ± 0.07	29.71 ± 0.14	48.00 ± 1.11	46.27 ± 0.18	46.13 ± 0.21	46.10 ± 0.36
Final kinetic energy (J)	5.79 ± 1.95	6.10 ± 0.99	7.51 ± 0.43	8.33 ± 0.66	5.19 ± 3.57	5.43 ± 1.48	9.34 ± 1.47	9.33 ± 1.22
Absorbed energy (J)	23.78 ± 1.89	23.47 ± 1.00	22.13 ± 0.44	21.39 ± 0.61	42.81 ± 2.88	40.84 ± 1.48	36.80 ± 1.27	36.78 ± 1.26
Absorbed energy (%)	79.75 ± 6.70	79.00 ± 3.56	74.66 ± 1.41	71.75 ± 2.06	88.75 ± 7.41	87.50 ± 3.32	79.25 ± 3.10	79.25 ± 2.50
Peak load −1 (N)	2750 ± 200	3800 ± 250	2350 ± 250	1200 ± 160	5000 ± 200	4900 ± 260	2100 ± 180	1900 ± 280
Time to peak load −1 (ms)	1.74 ± 0.19	1.53 ± 0.07	1.40 ± 0.14	0.87 ± 0.07	1.73 ± 0.13	1.63 ± 0.30	0.79 ± 0.12	1.13 ± 0.37
Peak load −2 (N)	3400 ± 175	4500 ± 225	−5300 ± − 200	5500 ± 250	6000 ± 300	5000 ± 300	6100 ± 200	6600 ± 350
Time to peak load −2 (ms)	5.20 ± 0.55	5.49 ± 0.08	−6.50 ± − 0.11	6.81 ± 0.17	5.50 ± 0.24	6.71 ± 0.51	6.60 ± 0.77	7.71 ± 0.24

Experimental methods

Experimental set-up

Impact tests were carried out by using an instrumented low velocity drop weight testing equipment (Figure 4). All tests were conducted by dropping a 12 mm diameter hemispherical impactor from a fixed height of 1 m. The impact energy was adjusted by changing the impactor weight. The total impactor weight was 3.24 kg for an energy level of 32 J, and 4.94 kg for the 48 J energy level. The potential energies were 32 J and 48 J. From the impactor mass and the initial velocity v₀, the real impact energy can be calculated; the difference between this value and the potential energy is the energy loss. If you have v₀ values, you can present here the actual impact energy.The samples were fixed on all edges using a clamping device with a circular opening of 75 mm. Samples of a fixed size of 100 mm × 100 mm were tested. Each sandwich type has undergone four tests.

Figure 4.

Low velocity impact test equipment.

The steel impactor includes a load cell with a capacity of 15.6 kN to measure the forces during impact. This load cell was placed between the impact head and the weight (Figure 4). A pneumatic rebound brake prevents multiple impacts. The impactor displacement was measured using a light detector positioned at the bottom of the test equipment, measuring the intensity of a light emitting diode mounted on the impactor. During the impact, both the force and the displacement of the impactor were recorded as a function of time.

Data reduction

From the force measurement, the impact energy and impactor displacement as a function of time can be calculated:

(1)

(2)

Equation (1) is a basic kinematic expression for velocity, written in energy terms and equation (2) for displacement, as seen in any basic course of mechanics. Where E(t) is the impactor energy during the impact, m is the impactor mass, F the force measured by the load cell, d₀ is the impactor position at the moment of first contact (usually the reference position, set to 0) and d_force is the displacement, calculated from the measured force. The initial impact velocity ν₀ is obtained from the slope of the measured displacement-time curves, just before the impact.

Although good correlation has been observed between the measured and calculated displacements for monolithic composites and metals, significant differences were obtained in the testing of sandwich materials: whereas the impactor rebounded after impact, the final velocity of the impactor calculated from the load curve was oriented downward, indicating full penetration of the test specimen. This proves that the force measured by the load cell underestimates the real forces at the impactor tip. This error can be due to incapability of uni-axial load cell placed between the impactor and the weight to register off-axis forces that are present during impact. Another probable cause for false measured force is related to the measurement location since the best location for measuring the impact force is situated at the tip of the impactor and not behind the impactor.^23,24 It is believed that the location of the load cell at a distance from the impact tip is the cause of error in the measurement. On the other hand, differentiating the displacement-time curves to obtain accelerations and velocities is numerically unstable and would require significant smoothening. Therefore, it was decided to correct the force-time curves using the displacement measurements by means of a correction factor K.

This correction factor K is assumed to be constant and is defined by modifying (2):

(3)

K is calculated from:

(4)

where ν_end is the end velocity at the moment of loss of contact between the test specimen and the impact head, calculated from the displacement-time curve.

Using the correction factor K, the force, acceleration, velocity and energy are obtained as a function of time. The energy dissipated during impact can also be calculated from ν₀ and ν_end, both calculated from the displacement measurement d_meas:

(5)

Post-impact properties

After impact testing, samples were sectioned and examined under an optical microscope to visualize the damage modes.

Edgewise compression testing was used to evaluate the post-impact mechanical properties. The compressive strength of sandwich panels is very sensitive to delaminations and skin damage. Therefore compression tests were conducted on the sectioned parts of the sandwich materials. The test specimen was clamped at two opposite edges and the clamps were used to end-load the sample in compression. The fixture contains side supports at the specimen edges to prevent the global buckling of the panel (Figure 5), while at the same time allowing local buckling of the skins in the damaged area. Special care was given to the alignment of the sample in the test rig. The tests were carried out with a standard instron test machine with 1 mm/min test speed on impacted and non-impacted samples. The maximum stress levels were determined.

Figure 5.

View of the edgewise compression after impact test set-up.

Results and discussion

Load histories and damage evaluation

A total of eight materials have been tested: four different core thicknesses have been used, both with and without foam filling, four tests have been conducted for each type of sample.

Tables 3 and 4 present the impact test results for samples without and with foam filling, respectively.

Table 4.

Impact test results of foamed samples at 32 J and 48 J energy level

Samples	10 mm-F-32 J	15 mm-F-32 J	18 mm-F-32 J	22 mm-F-32 J	10 mm-F-48 J	15 mm-F-48 J	18 mm-F-48 J	22 mm-F-48 J
Initial velocity (m/s)	4.28 ± 0.01	4.26 ± 0.02	4.28 ± 0.00	4.28 ± 0.01	4.37 ± 0.05	4.33 ± 0.01	4.33 ± 0.00	4.33 ± 0.01
Final velocity (m/s)	1.10 ± 0.12	0.80 ± 0.45	0.52 ± 0.24	0.82 ± 0.39	1.12 ± 0.19	1.08 ± 0.14	0.88 ± 0.15	0.95 ± 0.31
Max displacement (mm)	10.4 ± 1.2	9.7 ± 0.34	11.7 ± 1.3	10 ± 0.9	15.8 ± 0.8	17.7 ± 1.3	19.6 ± 1	24.5 ± 0.6
Initial kin energy (J)	29.63 ± 0.16	29.42 ± 0.22	29.64 ± 0.07	29.70 ± 0.09	48.46 ± 3.15	46.37 ± 0.15	46.31 ± 0.10	46.32 ± 0.17
Final kin energy (J)	1.97 ± 0.43	1.28 ± 1.27	0.51 ± 0.43	1.28 ± 0.87	3.23 ± 1.01	2.94 ± 0.75	1.97 ± 0.67	2.41 ± 1.29
Absorbed energy (J)	27.66 ± 0.27	28.15 ± 1.21	29.13 ± 0.44	28.42 ± 0.89	45.23 ± 3.19	43.43 ± 0.78	44.34 ± 0.71	43.91 ± 1.44
Absorbed energy (%)	92.67 ± 1.15	95.00 ± 4.55	98.00 ± 1.41	95.00 ± 2.94	92.75 ± 2.22	93.25 ± 1.50	95.25 ± 1.71	94.00 ± 2.83
Peak load −1 (N)	5200 ± 280	6300 ± 400	5300 ± 350	6600 ± 300	6800 ± 220	7300 ± 180	5800 ± 205	6500 ± 280
Time to peak load −1 (ms)	1.45 ± 0.06	1.71 ± 0.22	1.40 ± 0.14	1.54 ± 0.10	1.59 ± 0.09	1.58 ± 0.06	1.63 ± 0.37	1.40 ± 0.16
Peak load −2 (N)					5600 ± 220	5000 ± 250	2800 ± 285	2900 ± 220
Time to peak load −2 (ms)					5.76 ± 0.43	7.24 ± 0.34	8.82 ± 0.25	11.33 ± 0.60

Unfoamed panels

The load-time and energy-time curves of the unfoamed samples impacted at 32 J are presented in Figure 6. It can be seen that the load curves have two maxima. After the first narrow peak, the load rapidly decreases and then it rises up again to reach a second broader maximum. The first peak is related to the top skin and the core resistance. Because of the absence of foam in the core, the core compressive strength is dominated by the compressive resistance of the pile yarns. When an impact load is applied, the core under the impact point first deforms by pile bending until a critical buckling load is reached. Also breakages of pile yarns just below the point of impact takes place. The occurrence of these damage phenomena results in a sudden drop of the load. Earlier work has shown that the flatwise compression strength of the core is dominated by the critical buckling load of the piles, which strongly decreases with increasing pile length due to elastic Euler buckling.¹⁸ A geometric model was developed,¹⁸ taking into account the complex pile fiber architecture, to predict the flatwise compression stiffness and strength. While good correlation was found for stiffness prediction, the compression strength was strongly affected by local geometric features. In the current analysis where both a dynamic load and a hemispherical object are used, making a theoretical prediction of the failure loads is even more complex and beyond the scope of this research. The reduction of the core compression strength, as a function of increasing core thickness, results in a reduced support of the impacted top skin; and the collapse of the piles under the impact point results in a drop of the load during impact, creating a first peak in the impact curve. The first peak load is thus dependent on the core compression strength, and decreases with increasing pile length. In some curves, small load oscillations superimposed on the first peak point of the load-time curve are observed, especially in the specimens with core thicknesses of 10 mm and 15 mm. The oscillations indicate abrupt changes in load, caused by the formation of matrix cracks in the top skin.

Figure 6.

Typical impact force-time curves and impact energy-time curves of unfoamed sandwich panels for 32 J impact energy: (a) sample A1; (b) sample B1; (c) sample C1 and (d) sample D1.

As the core is compressed during the continued impact, the loads are transferred to the bottom skin, again increasing the load and accordingly a second peak is formed. Many oscillations on the second peak are observed in all samples, indicating more cracking, predominantly in the top skin. The spikes are especially observed in the specimens with core thicknesses of 18 mm and 22 mm, because the low first peak load did not reach a level that can cause significant skin damage.

Figure 7 presents the views of top and bottom skins and a cross-section at the point of impact for all unfoamed panels, tested at 32 J. The top skins show severe local damage, yet in none of samples, was the top skin perforated. The cross-section pictures show some damage in the core, especially directly under the point of impact. Since the core properties decrease with increasing core thickness, less impact energy is dissipated in the top skin and the core. This results in increased damage in the bottom skin with increasing core thickness. The damage characteristics are almost the same for all samples. The damage area is only slightly larger than the diameter of the impactor. The deformations have occurred in the forms of matrix deformation on the top skin point of impact, core yarn breakages below the point of impact, and minor delaminations between additional skin layers and also between additional skin and core top skin.

Figure 7.

Impact damage pictures of top (1), bottom surfaces (2) and cross-section (3) of unfoamed sandwich samples at impact location for 32 J impact energy: (a) 10 mm; (b) 15 mm; (c) 18 mm and (d) 22 mm core thicknesses.

Figure 8 presents the load-time and energy-time curves of the unfoamed samples impacted at 48 J, while Figure 9 shows the top and bottom skins and a cross-section at the point of impact. The tests conducted at an impact energy of 48 J produced a different result. In the sample with 10 mm core thickness, the top skin and the core section have been completely perforated. The impactor also hit the bottom skin and caused a significant damage without perforation. The first peak of this curve indicates the puncture resistance of the top skin, while the second one indicates the resistance of the bottom skin. The oscillations on the curve indicate that severe damage has occurred in both skins. The A1 sample with 10 mm core thickness has a higher shear stiffness and local compression strength, leading to localized loading and puncturing under the impact load.¹⁹ The damage in sample A1 can be summarized as matrix cracks, fiber failure and delamination between the skin layers of the top skin at the point of impact, buckling and fracture of pile yarns in the core, and matrix cracks, fiber failure and delamination between the skin layers of the bottom skin without penetration.

Figure 8.

Typical impact force-time curves and impact energy-time curves of unfoamed sandwich panels for 48 J impact energy: (a) sample A1; (b) sample B1; (c) sample C1 and (d) sample D1.

Figure 9.

Impact damage pictures of top (1), bottom surfaces (2) and cross-section (3) of unfoamed sandwich samples at impact location for 48 J impact energy: (a) 10 mm; (b) 15 mm; (c) 18 mm and (d) 22 mm core thicknesses.

Severe damage was also observed in the top skins of other samples, yet no complete puncturing was found. The damage in B1, C1 and D1 samples is nearly identical. Permanent deformations have occurred without any puncturing of the top skins. The excessive oscillations around the peak load and the observations made with optical microscopy prove that not only matrix cracking, but also fiber breakages and delaminations have occurred in the top skin (Figure 9). In these samples, the impact energy has been dissipated as core shear damage, since they have weaker cores. The deformations and the damage are spread over a wider area, as shown in Figure 9c, for pile yarn fracture.

Foamed panels

Filling the core with foam ensures support for both pile yarns and the top skin. Foam filling also has a structural damping quality. The behavior of the foam core depends on the density and the modulus of the foam. An increase in foam density leads to increased rigidity and brittleness. On the other hand softer foams will not be able to ensure sufficient impact resistance.⁷ Therefore, foam density has to be selected in conformity with the area of implication of the sandwich composite material. Medium rigidity and density foam was used in this study. The density of the foam varied between the values 95–105 kg/m³.

The impact results of the foam filled samples are quite different to the unfoamed samples. The load-time and energy-time curves obtained for foam filled samples impacted at 32 J are shown in Figure 10. Each curve has only one peak and for all sample types, the curves have the same characteristics. Already at 32 J, the impactor perforates the skin and partially penetrates the core, yet it could not proceed through the core material (Figure 11). The excessive spikes around the peak load and the sudden decrease of load indicate that the top skins have been fully penetrated. Compared to the unfoamed samples, the peak load achieved here is higher. This corresponds to the fact that usage of foam filling at the core section increases the core resistance, both by the contribution of the foam to the core properties and by their support of the pile yarns against buckling. The high shear and compression resistance has prevented shear and compression deformation of the core section and the impact energy has been spent by penetrating the top skin. After penetrating the top skin, the impactor proceeded through the core, crushing pile yarns and foam, but was stopped before reaching the bottom skin, where no damage was observed (Figure 11). Comparing the damage of the different samples, minimal damage was found in the 18 mm thick specimen, probably having the optimal core properties to dissipate the energy over a wider area on one hand, while still maintaining sufficient strength to resist impactor penetration. Scatter in the deformations probably stem from the matter of whether there are pile yarns just below the point of impact or not. The presence of pile yarns just below the point of impact may increase the impact resistance and the peak load values to a certain extent. What needs to be emphasized is the fact that the deformation has taken place only in the vicinity of the impact point and the sandwich specimen still maintains its structural integrity as there is no significant propagation of the impact deformations. Compression after impact tests should corroborate this.

Figure 10.

Typical impact force-time curves and impact energy-time curves of foamed sandwich panels for 32 J impact energy: (a) sample A1; (b) sample B1; (c) sample C1 and (d) sample D1.

Figure 11.

Impact damage pictures of top (1), bottom surfaces (2) and cross-section (3) of foamed sandwich samples at impact location for 32 J impact energy: (a) 10 mm; (b) 15 mm; (c) 18 mm and (d) 22 mm core thicknesses.

The load-time and energy-time curves obtained at 48 J are shown in Figure 12. The pictures of the samples tested at this energy level are presented in Figure 13. In the tests conducted at this energy level, the impactor has perforated the top skin and progressively penetrated into the core and hit the bottom skin. The peak load values achieved at the first peak of samples of different thicknesses are not very different from each other. After reaching the peak load, the top skin of the samples is perforated and the load decreases rapidly. After this point, the impactor starts to penetrate the core. During this phase, the load remains at a constant level, as can be observed in the the C2 and D2 samples.

Figure 12.

Typical impact force-time curves and impact energy-time curves of foamed sandwich panels for 48 J impact energy: (a) sample A1; (b) sample B1; (c) sample C1 and (d) sample D1.

Figure 13.

Impact damage pictures of top (1), bottom surfaces (2) and cross-section (3) of foamed sandwich samples at impact location for 48 J impact energy: (a) 10 mm; (b) 15 mm; (c) 18 mm and (d) 22 mm core thicknesses.

Despite the deep penetration of the impactor, it has been observed that the sandwich samples still maintain their structural integrity and no severe delamination takes place in the top skin layers. The integral woven sandwich core has greatly suppressed the onset of delaminations, thus sustaining the load carrying capacity of the skins. The worst damage in the bottom skin has been observed in the A2 sample with 10 mm core thickness, where matrix cracks, fiber fractures and delaminations are seen. The damage is significantly less in the other samples, reducing with increasing core thickness. During the penetration of the core, the impactor loses energy due to core damage and friction, and for thick cores, the impactor eventually comes to a stop without causing significant damage in the bottom skin. When the impactor hits the bottom skin and rebounds, a vacuum effect and friction hinders the rebound, resulting in negative forces, as seen in the final part of the load-time curves in Figure 12. No visible delamination was observed in these specimens.

Impact damage modes have been compared to the damage modes observed in other conventional sandwich materials. In monolithic laminated composites, it has been reported that the skin damage increases in a linear way with the increase of impact energy.²⁵^–²⁷ In honeycomb sandwich panels, among the widely used core material types for sandwich materials, breakage of cell walls has been observed during impact testing.²⁸ It has also been determined that the core damages increase in line with the increase of impact energy.²⁹ However the impact damage area in honeycomb core sandwich materials is much larger than in 3D integrated sandwich materials. This is mainly caused by the very small contact area of the core-skin interface in honeycomb sandwiches, resulting in delamination. The core-skin delamination in turn substantially reduces the residual strength of the sandwich material and adversely affects its impact load bearing capacity.

With foam cores, damage takes the form of an indentation for high energy impact and looks more like a crack for low energy impacts.³⁰ The core-skin interface is debonded in a region surrounding the point of impact usually in the foam just below the bond line, and the core experiences permanent deformations. Impact damage on foam cored conventional sandwich materials generally causes severe delaminations between the core and the skin. The magnitude of the delaminations increase with increasing impact energy. On the other hand, the tests conducted on integrated 3D structures have manifested that the impact damage area is restricted to the vicinity of the point of impact and does not affect the integrity of the structure. This indicates that the products can be easily repaired and its service life can be sustained.

Peak load

The peak load values of foamed and unfoamed sandwich samples with the variation of the core thicknesses are given in Tables 3 and 4 and in Figure 14. As explained earlier, several specimens showed two peak loads. The first peak value of the unfoamed samples, where no penetration has taken place, depends on the core resistance rather than the top skin resistance. Generally, as the core thickness increases core properties decrease. Due to this, core deformation and damage takes place before any damage occurs in the top skin. This is further supported by the decrease in time to peak load for the first peak, which decreases with increasing core thickness. Core deformation mostly takes place in the form of shearing. As the core resistance decreases, top skin damages take place after the onset of core compression, in other words at the second peak. The second peak load corresponds to the case where the bottom skin contributes to the load build-up. The second peak leads to further absorption of impact energy until all the energy has been transferred and springback starts, after which the load drops again. The time to second peak load logically increases with increasing core thickness. With increasing energy level, no significant change was observed in terms of load-time curve characteristics, except for the A1 sample with 10 mm core thickness. The increasing energy only resulted in an increase of the deformation of the top skin.

Figure 14.

Peak force of sandwich panels as a function of the core thickness for different impact energy levels: (a) unfoamed panels for 32 J impact energy; (b) unfoamed panels for 48 J impact energy; (c) foamed panels for 32 J impact energy; (d) foamed panels for 48 J impact energy.

In foam filled samples, the core pile yarns and foam filling support each other and this increases the strength of the core. This, in turn, significantly increases core resistance.²¹ Since the core resistance of foam filled samples is higher than the top skin penetration resistance, the impact results in the penetration of the top skin. The time to peak load seems independent of the core thickness. The load-time curves obtained for 32 J and 48 J energy levels are different. Only one peak load has been obtained in the tests conducted at 32 J. This indicates that only the top skins of the sandwich samples have been penetrated and the impactor was stopped before reaching the second skin. At 48 J impact energy, there are two peak loads: the impactor has penetrated the top skins of the samples, proceeded through the foam core and hit the bottom skins. Here, the second peak load indicates the bottom skin resistance. The time to reach the second peak load increases with increasing core thickness, and is significantly longer than for the unfoamed panels, showing that the foamed core slows down the impactor better than the unfoamed core.

Comparing the peak load values for foam filled samples, Figure 14 shows that the values decrease with increasing core thicknesses. The first peak load in foam filled sandwich samples is related to the perforation of the top skin. Although the top skins of all samples have been produced from the same materials, these differences in the peak loads can be explained by the reduction in core stiffness with increasing core thicknesses, causing the deformation and damage to spread over a wider area.

The obtained peak loads have been statistically evaluated with the ANOVA test at a 95% confidence interval. The findings indicate that the changes in core thickness and the use of foam are statistically significant in terms of the changes in the peak loads. The highest peak load among the unfoamed panels has been obtained from the 15 mm core thickness panels. Peak load values significantly decrease as the core thicknesses increase. However, the peak load value of the second peak increases with increasing core thickness: more compaction takes place in thicker samples. The highest peak load in the foamed panels has been obtained with 15 mm core thickness. Contrary to the unfoamed samples, in foamed samples no significant decreases in the peak loads have been observed with increasing core thickness. This is due to the fact that, in foamed samples, the impact energy is spent in perforating the top skin. There may be several reasons for the small varying peak loads of the samples, despite their top skin and core properties being the same. The decrease in the support provided by the core to the top skin, in line with the increase in core thicknesses, causes the top skin to be perforated at lower load values.

The findings indicate that the peak load values increase with increasing impact energy. In unfoamed samples, the time to reach the first peak load linearly decreases with increasing core thickness. Yet, no significant difference has been observed in the time to reach the first peak load for foam filled samples.

Energy absorption

The absorbed energy is presented in Figure 15. The foam filled samples absorb approximately 18% more energy than the unfoamed samples. In the unfoamed samples, the absorbed energy decreases with increasing thickness. This can be explained by the decreasing core properties: as the thickness of the samples increases, the amount of energy spread on the skin of the sample drops.

Figure 15.

% absorbed energy of unfoamed (UF) and foamed (F) sandwich panels as a function of core thickness.

Analysis with the ANOVA test have determined that the changes in thickness do not affect the absorbed energies in the foam filled sandwich samples: their top skin rigidities are the same, the core resistance is sufficient to resist the impact without severe core damage and the same damage modes have occurred. The absorbed energy is significantly higher in foam filled samples. This increase stems from the fact that the impact energy is absorbed by perforating the top skin and the core. It is believed that the foam core leads to localization of damage, causing the skin to be perforated. This would absorb less energy than a more widespread damage, but the crushing of foam absorbs more energy. This depends on the bending and shear stiffnesses and the core strength. If the core is very brittle, the localization effect will dominate, and the energy dissipation decreases. Penetration will occur at relatively low energy levels in that case.

The absorbed energy increases with increasing impact energy for all samples, but for the foamed samples, the ANOVA test shows that the relative absorbed energy (relative to the impact energy) does not significantly change with increasing impact energy.

Post-impact properties

The results of the compression after impact strengths (CAI) are presented in Figure 16. The strength was calculated by dividing the failure load by the cross-section of the skins. Only a minor decrease in CAI as a function of the core thickness is observed.

Figure 16.

Compression after impact test results of unfoamed (a) and foamed (b) sandwich samples.

For the unfoamed samples, subjected to a 32 J impact energy, the CAI strength decreases by 10 to 15%. The maximum degradation has taken place in the A1 sample with 10 mm core thickness, but analysis shows that this degradation is not significant. This is because the samples subjected to impact at 32 J energy are mostly damaged in the core. The damage in the skins is not severe and so the load carrying capability of the skins has not changed significantly. At 48 J impact energy, the CAI strength has decreased by 17–27%. The maximum degradation has taken place in the A1 sample with 10 mm core thickness. The high core strength of this material induced more skin damage, reducing the skin load bearing capacity. The major damage modes are local skin buckling and matrix cracking, originating from the impact damage zone. Figure 17 shows damage of unimpacted (Figure 17a and b) and impacted (Figure 17c to 17f) samples after compression.

Figure 17.

CAI images of sample surface: (a) unimpacted unfoamed sample; (b) unimpacted foamed sample; (c) 32 J impacted unfoamed sample; (d) 32 J impacted foamed sample; (e) 48 J impacted unfoamed sample; (f) 48 J impacted foamed sample.

In the foamed samples, the skin damage was more severe. Even at 32 J impact energy, the top skin was partially perforated decreasing its load bearing capacity. The CAI strength decreased by 18–25%, the highest decrease was observed in the samples, having respective core thicknesses of 18 mm and 22 mm. The samples tested at 48 J energy exhibit a decrease of 26–37%. The maximum decrease has taken place in the sample with 10 mm core thickness. In this material, the top skin has been completely perforated and also the bottom skin has suffered severe damage.

The decrease of the compressive strength after impact indicates that the load bearing capacity of the structure has deteriorated. Yet, the impact damage only occurs near the point of impact and the overall integrity of the structure is not significantly affected. This means that such deteriorations detected after impact do not apply to the whole of the sample. The best resistance among all samples has been determined to be provided by the samples with 15 mm core thickness.

Conclusions

The low speed impact behavior of 3D integrated woven sandwich composites has been evaluated and the following conclusions are drawn:

By means of foam filling in the core sections of 3D integrated sandwich composites, the pile yarns are supported by the foam in the core section. This results in a significant increase in core resistance and the structural integrity improves. Foam filling also constitutes additional damping properties.

Since the core resistance of the unfoamed samples is low, impact initially causes shearing in the core and pile buckling damage. As the core thickness increases, the core resistance decreases. This prevents the skin performance to be put into use. As the core thickness increases, the peak load values decrease with decreasing core resistance.

Since foam filling increases the core resistance, both the top skins and cores have been perforated in foam filled panels. The peak load is in line with the impact resistance of the top skin.

The peak load values obtained from foam filled samples where full perforation has taken place slightly differ from each other. This shows that the top skin resistance changes with the core thickness. Peak load values in foam filled samples are related to the skin rigidity.

The highest peak load values have been obtained from the sandwich composites of 15 mm core thickness.

The amount of absorbed energy decreases with increasing core thickness for unfilled samples, because the decreasing core resistance increases core shear deformations.

Foam filled samples absorb more energy than unfoamed samples.

In foam filled samples the amount of absorbed energy does not decrease with increasing core thickness. This is explained by the fact that the damage occurs as perforation of the top skin which is the same for all samples.

The CAI strength of the unfoamed samples drops 10–15% at 32 J, and 17–27% at 48 J impact energy. For the foam filled samples, a decrease of 18–25% at 32 J and 26–37% at 48 J impact energy was found.

The tests conducted on integrated 3D structures have manifested that impact damage is restricted to the point of impact and barely affects the integrity of the structure. This indicates that the products can be easily repaired and their service life can be sustained.

Footnotes

Acknowledgements

The help of Bart Pelgrims (Department MTM) is gratefully acknowledged. The authors thank Stepan V Lomov (Department MTM) for valuable discussion of this paper.

Funding

This work was supported by Uludag University Scientific Research Unit (UAP-TBMYO/2010-32) and Tubitak (109M350).

References

Karlsson

Astrom

. Manufacturing and application of structural sandwich components. Composites Part A 1997; 28A: 97–111.

Karahan

. Investigation of damage initiation and propagation in 2 × 2 twill woven carbon-epoxy multi-layer composites. Textile Res J 2011; 81: 412–428.

Karahan

. The effect of fibre volume fraction on damage initiation and propagation of woven carbon-epoxy multi-layer composites. Textile Res J 2011; 82(1): 45–62.

Abrate

. Localized impact on sandwich structures with laminated facings. ASME Appl Mech Rev 1997; 50: 69–82.

Kim

Jun

. Impact resistance of composite laminated sandwich plates. J Compos Mater 1992; 26: 2247–2261.

Herup

EWJ

Palazotto

. Low velocity impact damage initiation in graphite/epoxy/nomex honeycomb-sandwich plates. Compos Sci Technol 1997; 57: 1581–1598.

Hazizan

Cantwell

. The low velocity impact response of foam based sandwich structures. Composites Part B: Eng 2002; 33: 193–204.

Anderson

Madenci

. Experimental investigation of low velocity impact characteristics of sandwich composites. Compos Struct 2000; 50: 239–247.

Mines

RAW

Worrall

Gibson

. Low velocity perforation behaviour of polymer composite sandwich panels. Int J Impact Eng 1998; 21: 855–879.

10.

Torre

Kenny

. Impact testing and simulation of composite sandwich structures for civil transportation. Compos Struct 2000; 50: 257–267.

11.

Vaidya

Nelson

Sinn

. Processing and high strain rate impact response of multi-functional sandwich structures. Compos Struct 2001; 52: 429–440.

12.

Lomov

Bogdanovich

Ivanov

. A comparative study of tensile properties of non-crimp 3d orthogonal weave and multi-layer plain weave e-glass composites. Part 1: materials, methods and principal results. Composites Part A 2009; 40: 1134–1143.

13.

Ivanov

Lomov

Bogdanovich

. A comparative study of tensile properties of non-crimp 3d orthogonal weave and multi-layer plain weave e-glass composites. Part 2: Comprehensive experimental results. Composites Part A 2009; 40: 1144–1157.

14.

Karahan

Ulcay

Eren

. Investigation into the tensile properties of stitched and unstitched woven aramid/vinyl ester composites. Textile Res J 2010; 80: 880–891.

15.

Barrett

. A micromechanical model for the analysis of Z-fiber reinforcement. In: AIAA/ASCE/ASME/AHS SDM Conference , Salt Lake City, Utah, 1996, AIAA-96-1329-CP. pp.62–7.

16.

ZCL Composites Inc. Company Literature. ST. Edmonton, AB. Canada, 1998.

17.

Vaidya

Hosur

Earl

. Impact response of integrated hollow core sandwich composite panels. Composites Part A 2000; 31: 761–772.

18.

Van Vuure

Pflug

Ivens

. Modelling the core properties of composite panels based on woven sandwich-fabric preforms. Comp Sci Technol 2000; 60: 1263–1276.

19.

Judawisastra

Ivens

Verpoest

. Determination of core shear properties of 3D woven sandwich composites. Plastic, rubber and composites: Processing and application 1999; 28: 452–457.

20.

Hosur

Abdullah

Jeelani

. Manufacturing and low-velocity impact characterization of foam filled 3-D integrated core sandwich composites with hybrid face sheets. Compos Struct 2005; 69: 167–181.

21.

Bhuiyan

Hosur

Jeelani

. Low-velocity impact response of sandwich composites with nanophased foam core and biaxial ( ± 45°) braided face sheets. Composites Part B 2009; 40: 561–571.

22.

Karahan

Lomov

Bogdanovich

. Internal geometry evaluation of non-crimp 3D orthogonal woven carbon fabric composite. Composites Part A 2010; 41: 1301–1311.

23.

Judawisastra

Ivens

Verpoest

. Low velocity impact behaviour of 3D sandwich fabric panels by means of instrumented impact. In: Proc ECCM-9, 9th European Conference on Composite Materials, Brighton, UK, 4–7 June 2000.

24.

Philips

Verpoest

. 3D-knitted sandwich structures: production, properties and prospects. In: Crivelli Visconti

(ed.), Proc ECCM-8, 8th European Conference on Composite Materials – Science, Technologies and Applications: Vol. 1. Naples, Italy, 3–6 June 1998, pp.543–550.

25.

Abrate

. Impact on laminated composite materials. ASME Appl Mech Rev 1991; 44: 155–190.

26.

Abrate

. Impact on laminated composites: recent advances. ASME Appl Mech Rev 1994; 47: 517–544.

27.

Akay

Hanna

. A comparison of honeycomb-core and foam-core carbon fibre/epoxy sandwich panels. Composites 1990; 21: 325–331.

28.

Sun

. Low velocity impact in composite sandwich beams. Compos Struct 1996; 34: 21–27.

29.

Nettles

Hodge

. Impact testing of glass/phenolic honeycomb panels with graphite/epoxy face sheets. Proc 35th Int SAMPE Symp and Exhibition, Anaheim CA, April 2–5 1990, 35: 1430–1440.

30.

Bernard

Lagace

. Impact resistance of composite sandwich plates. J Reinf Plast Compos 1987; 8: 432–445.