The mechanical properties of EPS and steel were analyzed to understand their load-bearing behavior. The properties of EPS and galvanized steel wire are provided in Table 1. Additionally, oxide and mineral compositions of the Ordinary Portland Cement (OPC) Type I used in this study are detailed in Table 2. Locally available soil was used to prepare the stabilized mud mix for casting the panels wythes. The Atterberg limits, specific gravity, optimum moisture content and dry density of the soil were determined and is given in the Supplementary Tables S3, S4. According to the unified soil classification system the soil was classified as silty clay. The soil was mixed with 10% cement and 1% straw (10–15 mm length) by weight of dry soil.

Two full scale SMIP’s (Fp-1 and FP-2) were constructed for flexure testing. Each SMIP was 152.4 mm thick, with a central EPS core 51 mm and 51 mm thick mud mortar wythes on either side. The panels were setup vertically, with their width perpendicular to the ground and supported by wooden planks as shown in Figure 2A). Three SMIP’s were constructed for diagonal tension testing (DT-1, DT-2 and DT-3), three samples for compression test (CS1, CS2 and CS3) and 3 (WAS-1, WAS-2 and WAS-3) for water absorption test. The EPS core thickness for diagonal, compression and water absorption specimen are 102 mm.

The SMIP was erected by using wooden supports, placed at each corner on both sides and at mid for the flexure samples. After the panels were setup, water was sprinkled on them to remove dust and get wet surface for the application of mud plaster. Water cement grout was then applied to both sides of the panels. Following the grout, a cement stabilized mud mix was applied by hand from a distance to ensure food adhesion and the surface was finished with a steel trowel to create a uniform wythe thickness. This procedure was followed for all the SMIPs. The entire casting process is illustrated in Figures 2A–D.

During the plastering of the panels, three 50 mm × 50 mm × 50 mm cubes were made from the stabilized mud mix. These cubes underwent a compressive strength test after 28 days of curing, according to ASTM C-109, to determine their compressive strength and modulus of elasticity. The results showed a compressive strength of 0.75 N/mm² and a modulus of elasticity of 15.30 N/mm² (ASTM, 2020).

The SMIPs were cured at ambient temperature by sprinkling water on the wythes every 24 h for 28 days. While traditional mud construction usually does not require curing as no cementitious material is used, the cement-stabilized mud in this study needed curing to ensure proper cement hydration. The SMIPs were kept in an uncovered ambient environment until testing.

Flexural testing was conducted according to ASTM Standard C 78-02 to analyze the flexural behavior of SMIPs (Amran et al., 2016; ASTM, 2017b). The SMIPs were positioned on hinge and roller supports with a center-to-center distance of 2,438 mm, allowing for a 304.8 mm overhang on both sides, as depicted in Figure 3. Linear Variable Displacement Transducers (LVDTs) were strategically placed: two at the mid-span edges and one at each end of the SMIPs to measure deflection or rise of the SMIP. All LVDTs recorded positive values during contraction and negative values during elongation, with data automatically logged by a data logger.

For applying the load, a universal testing machine equipped with a hydraulic system capable of exerting up to 200 kN was used. The loading rate for both monotonic and cyclic loading was set at 10 mm/min. A W-shaped steel beam facilitated the transfer of load from the hydraulic jack to the steel plate and then to solid round steel bars (diameter 50 mm) positioned at specified locations on the SMIP. Figure 4 illustrates the experimental setup of SMIPs under flexural loading, focusing on their structural response as one-way slabs under load.

For the first flexural panel (FP-1), the load was applied while monitoring the deformations. The flexural panel (FP-2), was subjected to cyclic loading with three cycles (loading and unloading) for each load increment (1, 2, 3 and 4 kN). The test was conducted using load control, with an incremental load of 1 kN. After each increment, the load was reduced to zero. This process was repeated until the failure of FP-2. Detailed specifications of the specimens, including dimensions, composition, and testing parameters, can be found in the Supplementary Material.

Additionally, the stress at failure for a simply supported beam under loading, as per Equation 1 in ASTM C78 (ASTM, 2017b), was employed to analyze the results. R = PI / bd 2 (1)

Where, R = Bending stress in MPa.

P = Maximum applied load indicated by the testing machine in N.

l = Span length in mm.

b = Average width of the specimen in mm.

d = Average thickness of the specimen in mm.

Diagonal tension tests followed ASTM E-519 standards (ASTM, 2022). Samples were positioned diagonally in loading shoes to transfer uniaxial loads from the hydraulic actuator. Two displacement sensors, with a gauge length of 33.5″, were installed diagonally—one in tension and one in compression—1.4 feet from the top and 1 foot from each end as shown in the Figure 5A. These sensors along with the load cell were connected to a data logger, as shown in Figure 5B. The tests aimed to determine the shear strength of the SMIPs. Three samples were tested; one was strengthened at the corners due to local failure and then tested again.

The three compression SMIPs (CS1, CS2, and CS3) were tested following ASTM E-447 standards (ASTM, 2017a). The schematic diagram is as shown in the Figure 6A. Each specimen was positioned centrally on a layer of wet sand placed on the lower plate of the UTM. Additional wet sand was placed on top of each sample, and a 2 ft × 2 ft plate was placed over the sand to ensure uniform load distribution as shown in the Figure 6B. The sand served to provide a smooth surface for loading.

LVDT was installed to measure vertical deformation in the compression samples. Initially, a 20 mm dial gauge was selected for CS1, anticipating minimal vertical deformation. However, it was found that the gauge reached its maximum capacity of 20 mm while deformation continued. Consequently, for CS2 and CS3, a 50 mm LVDT was utilized as shown in the Figure 6B. Water absorption test was carried out according to ASTM D 559 – 03 (ASTM, 2012). Total of three samples were constructed. All the specimen of SMIP after curing for 28 days were then placed in the water tank for 24 h in order to find out the water absorption capacity of the specimen.

For the first flexural panel (FP-1), the load was applied slowly, and deformations were monitored. The second flexural panel (FP-2), was subjected to the cyclic loading with three cycles (loading and unloading) for each load increment (1, 2, 3 and 4 kN). The test used a load control method, with incremental loads of 1 kN applied and reduced to zero after each increment until FP-2 failed.

The load versus displacement curve for FP-1 and FP-2 is shown in Figure 7A. The ultimate load for FP-1 and FP-2 was 6 kN and 3 kN, respectively, with corresponding displacements at ultimate load of 119.83 mm and 36.04 mm. The maximum deformations for both panels were recorded as 123 mm and 66 mm, respectively. The cyclic loading graph of FP-2 is shown in Figure 7B.

The behavior of the flexural panels under different loading conditions can be observed in the Figure 7. From Figure 7A, it can be seen that as the load increases, the deformations at the middle also increase. At a load of 6000 N, deformation of 119.83 mm was achieved, indicating that the system is very ductile in bending. In contrast, for the cyclic loaded FP-2 load was applied in increments of 1 kN (0.1 ton). During the first two load increments, only small deformations were observed. However, when the load was increased from 2 kN to 3 kN (0.2–0.3 tons), very large permanent deformations occurred as shown in Figure 7B. When the load cycle of 4 kN (0.4 tons) began, the sample could no longer withstand more than 3 kN (0.3 tons) and suddenly collapsed, as shown in Figure 8. This sudden failure was considered due to a patch of the wooden support at the top mid-location as shown Figure 9A which was later filled with mud. Additionally, a crack at the center, as shown in Figure 9B, contributed to the failure. Bi-linear idealizations for both panels are presented in Figure 10. The initial stiffness values were 3 kN/mm for FP-1 and 1.2 kN/mm for FP-2.

Initially, minor cracks appeared in the lower wythes of the FP-1. The panel shows a linear elastic response. While further increasing load caused these minor cracks to widen and new cracks to form, especially in the lower wythe within the middle third region, where stresses and bending moments are highest (Fernando et al., 2017; Bishnoi et al., 2021; Mohamad et al., 2014). Most cracks occurred in the middle and extended across the width of the FP-1 without spalling of mud as shown in Figure 11.

Ductility is a crucial factor indicating a structure’s ability to undergo post-yield/plastic deformation and dissipate energy, thus reflecting its capacity to endure significant inelastic deformations before failure (Spadea et al., 1998). In this study, deflection ductility (µΔ) for the SMIPs is calculated, which is for 3.7 and 5.1 for FP-1 and FP-2 respectively. Deflection ductility is the ratio of ultimate deformation (Δu) to yield deformation (Δy) (Özkılıç et al., 2023). A higher deflection ductility (µΔ) means the structure can undergo substantial deformation before reaching the critical failure point, making it highly recommended for earthquake-resistant structures (Said and Razak, 2016). Additionally, the failure stress under flexural loading was calculated using from ASTM C78 for SMIP, EPS-based lightweight concrete sandwich walls panel (LCSWP), precast concrete sandwich panel (PCSP), reinforced mud sandwich panels RMSPs Equation 1 and precast lightweight foam concrete sandwich panels (PLFCSP) (Fernando et al., 2017; Daniel Ronald Joseph et al., 2018; Sarwar et al., 2024; Mohamad et al., 2014). The results revealed that the failure stress of SMIPs was higher than that of the LCSWP but lower than that of the PCSP and PLFCSP, and similar to that of RMSP, as illustrated in Figure 12.

From the flexure loading mechanism, it was observed that the ends of a panel along its length rise as the mid-span deflects, a behavior confirmed during the experiment as shown in the Figure 14. At an ultimate load of 3 kN, the uplift at the overhang portion was recorded at 36.9 mm while the theoretically calculated was 50.2 mm. At a load of 6 kN, the center span deformation was recorded at 119.83 mm, and the uplift deformation was 36.9 mm. For the same load, the uplift deformation at the overhang portion was 29% of the center span deformation. Similar to the mid-span deflection, the rise of the overhang portion of the panel was linear in the early loading stage, but beyond approximately 3 kN of loading, nonlinear behavior was observed.

The overhanging part of the SMIPs exhibits a slight deviation from its original position, resulting in a hogging shape, as shown in Figure 13. To analyze the sandwich effect, graphs were created to compare the experimental and predicted deformations at the cantilever ends of the panels Figure 14, using principle of similarity of triangles.

Initially, both panels closely follow the theoretical predictions, indicating linear elastic behavior. For FP-1, the experimental OED begins to deviate from the theoretical line as the load increases, showing greater deformations due to cracks and plastic deformation. FP-2 also initially aligns with the theoretical OED but diverges sooner, indicating earlier onset of damage due to cyclic loading and the preexisting crack at center. The experimental OEDs of both panels eventually fall below the theoretical predictions, demonstrating the sandwich effect, where layer interactions reduce overall deformation and enhance stability.

As the load increases, in the beginning, minor cracks appeared in the lower wythes of the SMIP. Initially the panels show a linear elastic response. While further increasing load caused these minor cracks to widen and new cracks to form, especially in the lower wythe within the middle third region, where stresses and bending moments are highest (Fernando et al., 2017; Bishnoi et al., 2021; Mohamad et al., 2014). Most cracks occurred in the middle and extended across the width of the SMIPs.

For the cyclic loaded FP-2, the load was incrementally applied in steps of 1 kN (0.1 ton). Initially, only minor deformations were observed during the first two load increments. However, a significant change occurred when the load increased from 2 kN to 3 kN (0.2 to 0.3 tons), resulting in substantial permanent deformations in FP-2. At the start of the 4 kN (0.4 tons) load cycle, the panel could no longer sustain more than 3 kN (0.3 tons), as illustrated in Figure 7B, leading to a sudden collapse depicted in Figure 8. The abrupt failure of FP-2 was primarily attributed to a weak patch due to the wooden support at the top mid-location, shown in Figure 9A, which was subsequently filled with mud. Additionally, a central crack, as highlighted in Figure 9B, further compromised the panel’s integrity. These defects collectively facilitated the panel’s inability to bear the increase.

The Figure 15 illustrates the compressive behavior of three specimens (CS1, CS2, and CS3) under loading conditions. Initially, all specimens exhibit elastic behavior, with the compressive load increasing linearly with deformation. CS1 reaches a peak load of approximately 40 kN at around 20 mm deformation. However, it is important to note that the displacement sensor used during the CS1 test was limited to measuring up to 20 mm. This limitation means that CS1’s full deformation was not captured, and it is expected that CS1 would have exhibited similar deformation behavior to CS2 and CS3 if a more capable sensor had been used. Despite this, CS1 demonstrates a high load-bearing capacity and significant residual strength, as indicated by the post-peak load maintenance and fluctuations.

CS2 reaches a peak load of around 39 kN at approximately 17 mm deformation, with a maximum deformation of 40.24 mm. Due to a pre-existing crack in the middle of CS2, its initial stiffness is lower compared to CS1 and CS3. CS3 reaches a peak load of about 35 kN at 22 mm deformation, with a maximum deformation of 41.56 mm. After the peak, CS3 demonstrates a gradual decrease in load capacity. For the CS2 and CS3 tests, a sensor with a measurement capacity of up to 50 mm was used, ensuring that their full deformation behavior was captured.

Bi-linear idealizations for CS2 and CS3 are presented in Figure 16. The initial stiffness values are 3 kN/mm for CS2 and 4.5 kN/mm for CS3, with an average stiffness of 4.1 kN/mm. Although the ductility of CS1 cannot be calculated due to the sensor’s limitation in capturing its ultimate deformation, its initial stiffness was found to be 5.7 kN/mm. The ductility ratios for CS2 and CS3 are 7.8 and 8.9, respectively, with an average ductility ratio of 8.3.

Figure 17 shows that CS1, CS2, and CS3 exhibit peak strengths of 0.861 MPa, 0.831 MPa, and 0.760 MPa, respectively. The corresponding strains for these values are 0.0313 mm/mm for CS1, 0.0278 mm/mm for CS2, and 0.0378 mm/mm for CS3. The Young’s moduli are 65 MPa for CS1, 37 MPa for CS2, and 64 MPa for CS3. The compressive strength of CSs was found to be nearly identical among the specimens with the average value of 0.82 MPa but lower than that of the cube specimens, which had a compressive strength of 1.06 MPa. Despite the lower compressive strength of the CSs compared to the cubes, the mud remained stable due to the presence of the sandwich panel, preventing it from buckling like a pure mud sample of the same size and thickness.

The behavior of the first and third compressive specimens (CS1 and CS3) was similar as shown in the Figure 17. As shown in Figure 18 in both CS1 and CS3, cracks appeared at the top and bottom of the sample at the ultimate load, specifically within a region extending up to 3 inches from both ends of the samples (Daniel Ronald Joseph et al., 2023; Mugahed Amran et al., 2016). In the CS2, chunks of mud spalling were observed at the ultimate load. Damage occurred on half of the top and a quarter of the bottom side of the sample. Both CS1 and CS3 experienced localized damage at the top region, with the depth of damage extending 3 inches from the top edge. In contrast, CS2 exhibited significant mud loss from the top half and partially from the bottom side. The primary mode of failure as shown in Figure 19 was the buckling of the wire mesh, detachment and spalling of mud from the specimens (Daniel Ronald Joseph et al., 2023).

The Figure 20 illustrates the ultimate load-carrying capacities of various structural panels, including PA4, PC1, FCS-F1, GA6, RCSP, and CSs (Benayoune et al., 2007; Daniel Ronald Joseph et al., 2023; Mohamad et al., 2017; Gara et al., 2012; Mugahed Amran et al., 2016; Khattak, 2024). PA4 exhibits the highest capacity of 1,075.00 kN/m, followed by PC1 of 917.60 kN/m, FCS-F1 of 830.67 kN/m, and GA6 of 616.80 kN/m. RCSP has a capacity of 188.68 kN/m, while CSs have the lowest of 83.03 kN/m (Khattak, 2024). Furthermore, when comparing the load-carrying capacities of CS and Reinforced Concrete Structural Panels (RCSPs), RCSP panels, with a capacity of 188.68 kN/m, can support loads for up to nine stores (Khattak, 2024). In contrast, mud panels (CSs) have a load-carrying capacity of 83.03 kN/m, allowing them to support approximately four stores under the same conditions. This indicates that while RCSP and other panels are suitable for high-rise structures, mud panels are a viable option for low-rise buildings due to their sufficient load-bearing capacity and sustainability advantages. In order to enhance the compressive capacity of SMIP CSs, it is recommended to increase the thickness of the mud plaster from 2 inches. This increase in thickness would likely provide additional stability and improve the overall compressive strength of the specimens.

The behavior of diagonal tension specimens was also studied in terms of cracking pattern, mode of failure, strain, and in-plane shear capacity. During the application of diagonal compression load, the vertical diagonal shortens, whereas the horizontal diagonal elongates. All the diagonal compression specimens were tested until collapse or until the load was reduced to below 80% of the peak load. The response was recorded in terms of total load versus horizontal and vertical displacements, using linear displacement sensors to measure these displacements. The total shear strain was calculated as the sum of strains along the two diagonals according to Equation 2 (ASTM, 2022). The applied load was used to evaluate the shear stress as per Equation 3 (ASTM, 2022). The shear stress versus strain plot is shown in Figure 21A. The nature of the shear stress versus strain curves is linear, as the cracking starts curves become non-linear. γ = ∆ V + ∆ H g (2) τ = 0.707   x   P 0.5   x   t   x   L + B (3) where, τ = average shear stress; γ = average shear strain; P = diagonal force; t = thickness of the specimen; L and B = length and width of the specimen; ΔV = decrease in the length of the loaded diagonal; ΔH = increase in the length of horizontal diagonal; and g = gauge length (equal along both the diagonals).

Figure 21B illustrates the relationship between shear stress (MPa) and shear strain (mm/mm) for four specimens: DT 01, DT 02, DT 03, and DT AR-03. DT 01, exhibits a linear increase in shear stress with strain, peaking at approximately 0.04 MPa at 0.0012 mm/mm before sharply dropping, indicating failure. For specimen DT 02, there is a sudden change in the slope of the shear stress-strain plot, which may be attributed to readjustment of the loading shoe of the specimen around a stress of 0.05 MPa. The peak strength was found out to be around 0.09 MPa at 0.0035 mm/mm and then declining sharply. The DT 03 also peaks at approximately 0.095 MPa at 0.005 mm/mm before a sharp drop. This sharp drop was because of the local crushing occurred at the loading shoes Figures 22A–C, indicating some out-of-plane bending due to accidental eccentricity caused by imperfections in construction or non-alignment of the centerline of the loading/supporting shoe, and the specimens (In-plane behavior of expanded polystyrene core reinforced concrete sandwich panels). This crushing failure occurred at very low applied stress at the loading shoes, which prevented the determination of the ultimate capacity and allowed only the measurement of the elastic capacity. In contrast, DT AR 03, continues to rise beyond the peaks of the other graph, reaching a maximum shear stress of about 0.17 MPa at 0.020 mm/mm before starting to decline, indicating it has higher resistance to shear stress and can withstand greater shear strain before failure compared to DT 01, DT 02, and DT 03. This was because one of the three samples was strengthened at the corners with cement-rich concrete (1:1:1.5) and cured for 28 days prior to testing, as shown in Figure 22D. The strengthening helped in capturing the ultimate shear behavior of the SMIP, and the shear strength was found to be 0.17 MPa with corresponding shear strain (mm/mm) 0.0208 while shear modulus was 12.5 MPa. Figure 21B shows that the SMIP has large shear deformation.

The ultimate shear load of different sandwich panels is shown in the Figure 23. It is clear that the shear strength or load carrying capacity of SMIP is lower than that of Concrete sandwich panels but it can be increased by increasing the strength of mud used (Gara et al., 2012; Ahmad and Singh, 2021b).

Water absorption tests for (SMIPs) were conducted following ASTM D 559-03. Three specimens, each measuring 2 ft high, 1 ft wide, and 8 in thick, were cured for 28 days before testing. The specimens were submerged in water for 24 h, then towel-dried and re-weighed to assess their water absorption capacity. The measured water absorption values were 14.1%, 12.3%, and 15.3%, with an average water absorption capacity of 13.8% that is within the limits (Jackson and Dhir, 1997).