Structural Behavior of Durable Composite Sandwich Panels with High Performance Expanded Polystyrene Concrete

EPS concrete is a lightweight, low-strength material with good energy-absorbing characteristics. However, due to the light weight of EPS beads, and their hydrophobic surfaces, EPS concrete is prone to segregation during casting, resulting in poor workability and lower strength. Fine silica fume greatly improved the bond between the EPS beads and cement paste, and increased the compressive strength of EPS concrete (Cook 1972; Chen and Liu 2004). Fine silica powder is one of the main binders of UHPC, therefore the combination of UHPC and EPS aggregates is expected to perform better than normal concrete without silica fume.

Density is known to be an important parameter determining numerous physical properties of EPS concrete, and it is determined primarily by the volume fraction of polystyrene aggregates. The strength of EPS concrete increased with increasing concrete density and decreasing EPS volume fractions (Chen and Fang 2011). The thermal resistance of EPS concrete can also increase with an increase of EPS volume fractions, due to the decreased density (Schackow et al. 2014; Real et al. 2016). To evaluate optimized HPEPC as a core element in a sandwich panel, the mechanical properties of HPEPC, according to the replacement ratio of UHPC and EPS in a unit volume, were investigated. The tested range of the EPS volume fraction in this study ranged from 30 to 70%. The mix composition of UHPC and the material properties of EPS, used in this study, are presented in Tables 1 and 2, respectively.

The design compressive strength of the given mix composition of UHPC is 180 MPa (Richard and Cheyrezy 1995; Wille et al. 2011; KCI 2012). The Portland cement was Type I, as designated by ASTM C150. The silica fume used was made in Norway, and had a particle size distribution of 45–800 μm. Typically the silica fume/cement ratio used for UHPC is 0.25 considering the optimum filling performance, enhancement of the lubrication effect, and the complete consumption of total hydration of the cement (Richard and Cheyrezy 1995). Filler, sized between cement and silica fume sizes, improves the compressive strength of concrete by increasing the poured density. A mean particle size of the filler in Table 1 is 2.2 μm and the filler, as an Australian product, composed (> 99%) of SiO₂ components was used. Super plasticizer, that has a specific gravity of 1.01 g/cm³, was used. EPS beads were used as artificial lightweight aggregates for decreasing the weight of the concrete, and producing different grades of EPS concrete. The diameter of 85% of the EPS particles was approximately 3.5 mm, and their actual density was 50.58 kg/m³. Bulk density of EPS beads is less than half of true density, which means that EPS beads per unit volume have large pore space. The curing method involved removing the specimens 1 day after pouring, and then subjecting them to 90 °C high-temperature steam for 48 h. Figure 2 shows the cylindrical specimens with different EPS bulk ratios. Bulk ratios of HPEPC of 30 and 40% show that EPS plays a role in lightweight aggregates within the UHPC matrix, but HPEPC with bulk ratios greater than 50% shows that UHPC cannot completely fill the gaps between the EPS. Specimens with an EPS bulk ratio of 70% have massive pores in their structure. With increasing EPS replacement ratios, the HPEPC is expected to be lighter and more thermally resistant, but the compressive strength is expected to decrease.

The compressive strength tests were carried out in accordance with ASTM C109. On the seventh day, the strength of the cubic specimens (50 × 50 × 50 mm³) was measured using a universal testing machine. The specimens were subjected to displacement at a loading rate of 1 mm/min. The unit weight of the HPEPC was measured according to standard KS F 2462, that contains a method for determining the unit mass of structural lightweight concrete.

To determine the modulus of elasticity and Poisson’s ratio, cylindrical concrete specimens, 100 mm in diameter and 200 mm high, were tested according to ASTM C469 (2014). A compressometer was used to measure the modulus of elasticity. The mean strain was measured by two strain gauges, opposite each other halfway up the specimen, parallel to the vertical axis. The specimens were tested, and maintained within the defined curing state, within 1 h after removal from the mold. The transverse strain was determined by an unbonded extensometer, with an accuracy of 0.5 μm, that measured changes in diameter at the mid-height of the specimen. A combined compressometer and extensometer was used to obtain Poisson’s ratio.

The test results are summarized in Table 3. The compressive strength and density of UHPC measured on the 7th day were 196.83 MPa and 2311.07 kg/m³, respectively, for the same curing conditions as the other specimens. The density and compressive strength of HPEPC decreased with increasing EPS bulk ratios. The density of HPEPC decreased almost linearly with increase EPS bulk ratio till 60%, but the density of HPEPC with EPS bulk 70% decreased steeply. The massive pores in the specimens with EPS bulk 70% in Fig. 2 supports these test results. The compressive strength of HPEPC decreases at a rate greater than the rate of increase of the EPS bulk ratio. Figure 3a shows the normalized compressive strength and density, by material properties, of UHPC that contained no EPS. The modulus of elasticity shows tendencies similar to the compressive strength, but the Poisson’s ratio decreases negligibly compared to the other properties. Figure 3b shows all the test results, except for the average values, that indicate that the standard variation of Poisson’s ratio increases with an increase in the EPS bulk ratio, despite the relatively constant average value, 0.6.

From the test results, HPEPC with a 40% EPS bulk ratio was chosen as a core material for the composite sandwich panels, taking into consideration the decreased rate of compressive strength, modulus of elasticity, and density as a function of the EPS bulk ratio.

This can be regarded as a high strength lightweight concrete. According to fib Model Code 2010, it is recommended that lightweight aggregate concrete for structural applications have a density in the range 800–2000 kg/m³, and high strength concrete is typically accepted to have a characteristic compressive strength greater than 50 MPa (Fib 2012).

As can be seen in Fig. 1, composite sandwich panels comprise a prefabricated core and a prefabricated face sheet, with glue between them. To investigate the optimized composite sandwich panels using HPEPC cores, three types of face sheet materials, ultra-high-performance fiber-reinforced concrete (UHPFRC), UHPC with a glass fiber reinforced polymer (GFRP) mesh, and GFRP were tested with both adhesive mortar and epoxy bond as a glue. Specimens with EPS mortar cores were also tested, for comparison with the specimens with HPEPC cores. EPS mortar in this study is distinguished by EPS concrete by no silica fume, no coarse aggregates. EPS mortar was composed by ready mixed cement mortar and EPS beads. The mechanical properties of EPS mortar was targeted for structural applications. The test program is presented in Table 4. The first letter of the specimen ID is the core material, and the second letter the face sheet material.

The test set-ups and dimensions of the specimens are shown in Fig. 4. The thickness of the core and face plate are 55 and 5 mm, respectively, and the specimens are 65 mm thick. Results of flatwise compression tests can be used for the applications of the support zones under concentrated loads. To compare core compressive behavior only, specimens with two different cores, with GFRP face sheets, were tested. For applications of load bearing walls and slabs in high-rise buildings, specimens bonded by adhesive mortar were tested for edgewise compressive loading, as well as for flatwise flexural loading (Manalo et al. 2010). In addition, specimens with different face sheets, with their HPEPC cores bonded by epoxy, were tested to investigate delamination between cores and face sheets. Flexural behavior testing was planned for all composite sandwich panels, but the face sheet of the M-U2 specimen detached before the test was conducted, and the flexural test results of that specimen are not recorded.

The flatwise compression tests were conducted in accordance with ASTM C365 (2016). Compressive strength tests were performed on three sandwich specimens, 100 × 100 × 65 mm³, for the EPS mortar core and the HPEPC core. The load was applied with a universal test machine, with a 2000 kN capacity. The ultimate flatwise compressive strengths, based on the averaged test results, are 8.89 MPa for the M-G specimen, and 45.23 MPa for the U-G specimen (Table 7). The failure modes of both specimens are core compression failure, as shown in Fig. 7. When comparing the material properties of the core materials in Table 5, the ultimate compressive strength of the M-G specimen is greater than the compressive strength of EPS mortar, but the ultimate compressive strength of the U-G specimen is significantly lower than that of the HPEPC specimen. The material compressive strength of HPEPC is approximately eight times greater than that of EPS mortar, but the flatwise compressive strength of the U-G specimen is approximately five times greater than that of the M-G specimen. This phenomenon might be interpreted by size effect and high strength effect. The compressive strength of EPS concrete decreases significantly with an increase in EPS bead size for the same concrete porosity (Miled et al. 2004, 2007; Le Roy et al. 2005; Babu et al. 2006) It is observed that the EPS beads size effect is very pronounced for low porosity concretes. In addition, for 6.3 mm EPS beads concrete, unlike 2.5 mm beads EPS concrete, the compressive strength can be affected by specimen size (Miled et al. 2007). HPEPC has much less porosity than general EPS concrete and the EPS bead size applied in this study was 3–5 mm, which can lead to the specimen size effect.

Edgewise compression tests determine the compressive properties of structural sandwich construction in a direction parallel to the sandwich facing plane. The tests, to evaluate the in-plane compressive behavior of the sandwich panels made of both core materials, were carried out in accordance with ASTM C364 on specimens with nominal dimensions of 250 × 250 × 65 mm³. The loaded surfaces (250 × 65 mm²) were levelled to avoid any non-uniform or transverse loading. The load was applied with the same testing machine used in the tests described above. The test results in Table 7 show that the maximum strength of the specimens with EPS mortar core depends on the type of face sheets, but the maximum strengths of the specimens with HPEPC cores are greater, irrespective of the type of face sheets.

Failure mode is important in edgewise compression test results, and commonly observed failure modes of composite sandwich panels can be categorized as face sheet failure, comprising buckling or compression failure, or core failure, comprising compression or shear failure (ASTM C364 2016). In this study, it was observed that the specimens sheeted by GFRP plate experienced core failure, and the other specimens face sheet failure. In all specimens, initial cracking was observed in the core, and the M-G and U-G specimens suffered core failure, with severe delamination between the face sheets and cores. The other specimens behaved differently after core cracking, depending on the types of face sheets. The specimens sheeted with UHPFRC resisted additional compressive loading with less stiffness, but failed with face sheet buckling. The specimens sheeted by UHPC with GFRP mesh did not resist additional compressive loading, but ductile behavior was observed with constant compressive loading. The failure modes of the specimens with HPEPC cores, as shown in Fig. 8, depend on the face sheets. The U-U1 specimen failed by UHPFRC buckling in a brittle way, and the U-U2 specimen failed by compressive failure in the face sheet. Shear failure in the HPEPC core was observed in U-G specimen.

Flexural tests were conducted in a four-point bending configuration. The bond capacity between the face sheet and core plays an important role in the flexural capacity of a sandwich panel. The separation of the facing from the core in the adhesive bond zone is one of the most frequent failure modes for sandwich panels with concrete facings (Shams et al. 2014). Using externally bonded reinforcement as shown in Fig. 1a, the bond between the adhesive and the concrete often fails once the tensile strength of the concrete has been exceeded. Following the local debonding of externally bonded reinforcement and the concrete, the results mostly induces a total failure of the bond between the externally bonded reinforcement (Zilch et al. 2014). The representative stress distribution depending on the bond capacity between the face sheet and core can be explained by non-composite section and fully composite section in Fig. 9. For a non-composite sandwich panel, the three concrete wythes act independently, so that the distribution of loads is based on the relative flexural stiffness of each wythes. For a fully composite panel, the entire panel acts as a single unit in bending and the flexural design of fully composite panels can be considered as I-shaped section applying transformed section method (PCI Sandwich Wall Committee 1997). There are numerous techniques, for instance the shear connection between the outer face sheets, for ensuring linear composite behavior in the cross sections (Benayoune et al. 2008; Shams et al. 2014). To consider the importance of bond behavior between prefabricated concrete, epoxy bond as an adhesive material was also tested for the U-U1 and U-G specimens.

Table 8 and Fig. 10 shows the failure modes of the specimens. The flexural capacity of the specimens with EPS mortar cores, and the specimens with epoxy bonds, was significantly less than the flexural capacity of the specimens with HPEPC cores. The peak loads of the specimens with epoxy bonds was 0.3–0.6 times the peak loads of the specimens with the same components, but with different adhesive materials. Therefore, it can be concluded that the epoxy bond is not appropriate as an adhesive material in prefabricated concrete panels. Whether delamination was observed before the peak load is also indicated in Table 8, as delamination induces not only a decrease in the flexural capacity, but also occasional brittle structural behavior. Both specimens sheeted with GFRP plate, and the specimens with epoxy bonding, were clearly delaminated before the peak load, as shown in Fig. 10. For the specimens with a delamination, core cracking tends to govern the overall flexural capacity, which means that the maximum stresses in the EPS core determines the peak load without externally bonded reinforcement’s role as shown a non-composite section in a Fig. 9. On the other hand, the delamination in U-U2 specimen was observed after the peak load in a partial region, which induces ductile behavior till a total failure.

Figure 11 shows that load–displacement curves for the specimens except the specimens bonded by epoxy material. All panels exhibited an approximately linear behavior up to initial cracking in the core. Of the specimens with an EPS mortar core, despite the small peak load, the M-U1 specimen resisted in a stable manner, and failed by the UHPFRC plate yielding at the bottom, as shown in Fig. 11a. Figure 11b shows the load–displacement curves of the specimens with HPEPC cores. Depending on the type of face sheets, the elastic region before initial cracking in HPEPC cores showed some scatter. The initial cracking strength in the core of the U-G and U-U2 specimens differed, but the peak loads were similar. However, the load–displacement relationship of the U-G specimen shows behavior that is too brittle, so it is not recommended for practical applications. On the other hand, the U-U1 specimen also shows brittle behavior in the load–displacement relationship, but it did not collapse, as the UHPFRC face resisted after the peak load, without delamination. The post-peak behavior of the U-U2 specimen is outstanding, ensuring conservative flexural loading, and the specimen failed by the yielding of the GFRP mesh in the face sheet.

Measured strain values in Fig. 12 demonstrate the flexural behavior in detail. The specimens reinforced by UHPFRC face sheets shows partially composite flexural behavior. For for U-U1 specimen and M-U1 specimen, the initial crack was observed in a core and face at the same time and the final failure was related to the face yielding or bending. On the other hand, the specimens reinforced by GFPR plates shows non-composite flexural behavior. The face sheets in U-G specimen and M-G specimen hardly resist flexural loads, so that the initial crack was caused by tensile stress at the bottom of the core and the compressive stress in the top of the core induces the final failure. The provision in ACI 318-11 requires that reinforced flexural members need at least 1.2 times additional load beyond cracking to reach its flexural strength to prevent abrupt flexural failure developing immediately after cracking (ACI Committee 318 2011). U-U1 specimen and M-U1 specimen satisfy the statement, and U-G specimen also satisfies that in spite of almost single resistant behavior of HPEPC core.