Material Characteristics of 3-D FRP Sandwich Panels

Abstract This paper presents an innovative 3-D fiber reinforced polymer, (FRP), panels designed to overcome delamination problems typically encountered in traditional sandwich panels. The sandwich panels consist of GFRP laminates and foam core. The top and bottom consist of GFRP plates connected together with through-thickness fibers to achieve the composite action. The fundamental material characteristics of the panel in tension, compression, flexure and shear are critical for the use and structural design of these panels. This paper summarizes the findings of an extensive experimental program to determine the various parameters believed to affect the material characteristics of these sandwich panels. The influence of the panel thickness, through thickness fiber configuration and density, and other parameters on the tension, compression, flexure and shear behavior of the panels are discussed. Keywords: fiber reinforced polymers, sandwich panels, 3-D fibers, core shear, flexure, tension, compression INTRODUCTION International research efforts continuously looking for new, better and efficient construction materials. The main goal of these research works is to improve the structural efficiency, performance and durability of civil engineering and transportation applications. The introduction of new materials typically brings new challenges to designers to utilize the new properties of these materials. In the past decades various sandwich panels have been utilized in the construction of aerospace, marine, architectural and transportation industry. Light-weight, excellent corrosion characteristics and rapid installation capabilities created tremendous opportunities for these sandwich panels in the industry. Sandwich construction provides an efficient use of the materials and utilization of each component to its ultimate limit. The sandwich structure offers also very high stiffness-to-weight ratio. It enhances the flexural rigidity of a structure without adding a substantial weight therefore it provides significant advantageous in comparison to the use of the material alone for structural system. Sandwich constructions have superior fatigue strength and excellent acoustical and thermal insulation. Historically, the principle of using two cooperating faces separated by a distance in between was introduced in 1820 by Delau. The first extensive use of sandwich panel was during the World War II. In the “Mosquito” aircraft, sandwich structure was used, mainly because of the shortage of other materials in England during the war. The faces were made of veneer while the core consisted of balsa wood. One of the early uses of sandwich structures in an aerospace application was in 1937 where balsa wood core and cedar plywood face sheets was used in the construction of De Havilland albatross airplane. 1 During World War II the first theoretical analysis of sandwich theory was published. By the completion of World War II and in the late 1940’s, some of the first theoretical works on sandwich constructions were documented. Sandwich beams with parallel skins and a metallic honeycomb were considered by many researchers. Allen [1] and Plantema [2] had summarized the information available up to the end of the 1960s in two text books. Paydar and Libove [3] presented a small deflection theory to determine the stresses and the deflections of a sandwich plate with a variable height but symmetric about its midheight surface. Ko [4] analyzed the flexural behavior of a rotating sandwich beam. The core and the skins were modeled as either Timoshenko or Euler-Bernoulli beams, and the core was assumed to be incompressible in the vertical direction. Gordaninejad and Bert [5] analyzed a straight sandwich beam with thick skins considered as Timoshenko beams and the core was assumed to be of antiplane type. High-order theory for the analysis of beams and plates was used by researchers, Reddy [6-8] and Krishna [9]. They assume that the height of the beam remains unchanged and that the longitudinal displacement through the depth of the beam is expressed by a high-order polynomial with coefficients that are functions of the longitudinal coordinate and are determined by the boundary and the overall equilibrium conditions of the section. Frostig and Baruch [10] and Frostig et al. [11] studied the behavior of a uniform sandwich beam with identical and non-identical skins and a soft core using a superposition approach that determines the effects of the core flexibility on the stresses, on the deflections, and on the overall beam behavior. An enhanced highorder theory was developed for beams using a superposition method. It was improved 2 with a refined high order theory that uses a rigorous systematic approach which based on variational principles. A general systematic rigorous theory was developed by Frostig [12,13], a variational high-order theory that defines the vertical normal and the shear stresses at the skin-core interfaces as well as in the core. The present study is aimed to provide the characteristics of a new type glass fiber reinforced polymer (GFRP) sandwich panels. The sandwich panels presented in this study consist of GFRP laminates and foam core sandwich where top and bottom face sheets are connected together with through-thickness fibers as shown in Figure 1. The top and bottom GFRP face sheets are formed by the laminates layed-up in a 0/90 degree fiber orientation. E-glass fibers having density of 2.54 g/cm3 were used as the reinforcing material in the laminates. The number of the laminates in either face sheets may vary depending on the use of the sandwich panel. The sandwich foam used as a core material is polyurethane modified polyisocyanurate cellular plastic. “Through-thickness” unidirectional glass fibers are inserted through the top and bottom face sheets, and the foam core. The amount of the glass fibers forming the “through- thickness” fibers is 227 m/kg. The sandwich panel is fabricated using pultrusion process. After the glass fiber laminates and the foam core are sandwiched the “through-thickness” fibers are inserted in dry condition. Afterwards, the whole assembly goes into the resin tank and the heated die. 3 TENSILE BEHAVIOR The in-plane tensile properties of the face sheets of the various 3-D FRP sandwich panels were evaluated. A total of 33 tension specimens, having different number of plies and different configurations of through-thickness fibers, were tested according to ASTM D3039. The modulus of elasticity, stress-strain behavior and failure modes of the tension specimens were evaluated. Three repeated specimens for each type of the GFRP face sheet cut from different sandwich panels were tested. Typical tension specimen consists of flat strips with a total width of 38 mm and a total length of 430 mm. The specimen length was selected to minimize possible bending stresses which could be induced by minor grip eccentricities. Aluminum tabs were bonded to each end of the specimen to prevent premature failure at the ends of grips. The specimens were mounted in the grips of a ±980 kN capacity MTS machine and monotonically loaded in tension up to failure. A standard head displacement rate of 0.13 cm/min was used to load the specimen up to failure. The strain in the specimen was monitored using a strain gauge located at the mid-length of the specimen. A non-linear measured stress-strain relationship was observed for all tested tension specimens. Since the coupon specimens were cut from 3-D panels, the non-linear behavior could be due to one or combination of the presence of the fibers in the other direction, presence of the veil and the end insertions of the through thickness fibers. Typical stress-strain relationship of tension specimens having different through thickness fiber densities is shown in Figure 2. For design purposes, the nonlinear behavior of the 4 stress-strain relationship could be approximated by two linear behaviors with different stiffness. The initial portion can be used to determine the initial elastic modulus using regression analysis for the data up to 0.2 percent strain. Due to the significant nonlinear behavior observed beyond the strain level of 0.2 percent, the second slope, conservatively representing the reduced elastic modulus can be determined approximately based on the data measured between strains of 0.4 percent up to failure strain. These two calculated slopes are extended between 0.2% and 0.4% strain until they intersect each other in order to obtain the whole approximation of the tensile behavior of the face sheets of the panels as shown in Figure 3. Test results indicate that the initial modulus of elasticity of the face sheets was typically about 50 percent higher than the value within the range of 0.4 percent strain to failure. Test results indicated that that increasing the density of the through-thickness fibers creates zones of imperfection and waviness among the fibers therefore results in reduction of the elastic modulus as well as the tensile strength of the face sheets considerably as shown in Figure 4 and Figure 5 respectively. Failure of all specimens was due to the rupture of GFRP sheets within the gauge length of the specimen. SHEAR BEHAVIOR The main objective of the shear testing program was to evaluate the influence of the through-thickness fibers on the shear modulus of the proposed 3-D FRP sandwich panels. A total of 44 specimens with different skin configurations, core thicknesses, through thickness fiber insertion densities and patterns were tested using the configuration 5 provided by ASTM C273 for sandwich panels. It should be noted that due to the relatively large thickness of the sandwich used in this program, the length to the thickness ratio did not satisfy the ASTM C273 requirements. The recommended length-tothickness ratio could not be practically achieved for these sandwich panels. The low length to depth ratio used for these shear specimens could have an effect of stress distribution in comparison to typical thin sandwich panels. Therefore, the measured shear strengths reported in this paper may be less than the actual shear strengths of these sandwich panels. The width of the test specimens was equal to the width of the sandwich specimen. The total length was 290 mm for all tested specimens. Sandwich test specimens were bonded to 19 mm thick steel plates on each side using an epoxy. The test fixture was designed to have the line of the load action passes through the diagonally opposite corners of the specimen. The specimens were loaded in compression using a 9000 kN capacity machine and a rate of loading of 0.05 cm/min. The relative displacement between the two steel plates was measured, at the center of the steel plates at both sides, by using displacement transducers. A 1100 kN capacity load cell was used to measure applied load. The shear modulus in the plane normal to the facing sheets was evaluated for each specimen using Equation. (1); ))(( ))(( bL tS G = (1) where, G is the shear modulus; S is the slope of initial portion of the load versus the relative displacement between the steel plates; t is the thickness of the core; L is the 6 length of the specimen and b is the width of the specimen. Two different configurations of through-thickness fibers were investigated. The first pattern was a “regular array of through thickness fibers” pattern in which the through-thickness fibers were evenly spaced in each direction. The second pattern was a “continuous wall” pattern in which the through-thickness fibers were arranged in semi-solid rows, like in a closely spaced picket fence, in one direction forming a rigid web. The layout of both patterns is illustrated in Figure 6. Figure 7 shows the typical stress-strain relationship of the tested shear sandwich specimens, where shear stress is determined based on the applied load and the shear resisting area while shear strain is determined by using the measured relative displacements parallel to the steel plates divided by the thickness of the sandwich specimen. The results indicate a typical linear behavior up to the initiation of the first shear crack in the foam core followed by a non-linear behavior with significantly low shear modulus up to failure. It is observed that the significant reduction in the shear stiffness is mainly due to the cracking of the foam and possible formation of the plastic hinges at both ends of the through thickness fibers. Test results showed that the density and configuration of the 3-D fibers affect the core shear modulus considerably. Increasing the quantity of the 3-D fibers from 1.25 per cm2 to 2.5 per cm2 in the regular pattern increased the core shear modulus by 33 percent as shown in Figure 8. Furthermore, presence of a continuous web of 3-D fibers creates a mechanism similar to shear wall mechanism, therefore minimizing the stress concentrations at the connection between the 3-D fibers and the face sheets results in increase of the core shear modulus 7 considerably. Test results showed that increasing the quantity of the 3-D fibers from 2.5 per cm2 in the regular array pattern to 3.6 per cm2 in the continuous wall pattern, increased the core shear modulus by 765 percent as shown in Figure 8. Test results suggest that increasing the thickness of the sandwich panel does not have significant effect on the shear modulus of the sandwich panel. In fact, increasing the thickness from 60 mm to 100 mm decreased the shear strength of the sandwich panel 27 percent. The influence of the filling material on the core shear modulus was investigated by cracking one of the 60 mm thick panels prior to loading using a chisel. The shear stiffness of the specimen with the uncracked foam, based on the initial slope of the curve was about twice the stiffness of the panel that with cracked foam as shown in Figure 9. It was observed that uncracked foam plays an important role by confining the throughthickness fibers and therefore the shear modulus of the core shear increased significantly. All shear tests conducted in this program were terminated at a certain stage due to the large shear deformation and limitation of the stroke of the testing machine. Typical shear cracks formed in the shear test are shown in Figure 10