Experimental and Numerical Research on a Sand Cushion Geotechnical Seismic Isolation System in Strong Earthquakes and Cold Regions

Abstract

Masonry buildings in high-intensity seismic and cold regions of China face the dual challenges of frost heaving and seismic hazards. To explore the potential of a sand cushion instead of the frozen soil layer to deal with these problems, a cost-effective sand cushion-based Geotechnical Seismic Isolation System (GSI-SC) was developed in this study, where a sand cushion is introduced between the structural foundation and natural soil, while the space around the foundation is backfilled with sand. Shaking table tests on a one-story masonry structure equipped and non-equipped with the GSI-SC system were undertaken to investigate its effectiveness in seismic isolation, where the input wave adopted the north–south component of the EL Centro wave recorded in 1940, and the peak input acceleration (PIA) was set as 0.1 g, 0.2 g, and 0.4 g. It is found that the GSI-SC system significantly reduced the seismic response of the structure, effectively achieving seismic isolation. For a PIA of 0.4 g, the GSI-SC system reduced the acceleration of the roof panel and the inter-story displacement of the structure by 33% and 39%, respectively. Numerical simulations were performed to evaluate the seismic response of buildings equipped and non-equipped with the GSI-SC system. The simulation results matched well with the experimental results, verifying the effectiveness of the newly developed seismic isolation system. The GSI-SC system can provide the potential to reduce frost heave and earthquake disasters for buildings in high-intensity seismic and cold regions.

1. Introduction

Masonry structures are still very common in rural areas of some countries; due to the lack of appropriate seismic design and detailed design, their seismic performance is poor. Masonry buildings with poor seismic performance often suffer damage or even collapse under the action of earthquakes. The collapse and damage to houses will bring direct economic losses and a large number of refugees, causing a huge economic burden. However, in high-seismic and cold regions, the two threats of earthquakes and frost heave can cause damage to low-rise, rigid, and heavy masonry buildings. Therefore, for high-seismic and cold regions, efforts should be made to formulate technically feasible and cost-effective measures to simultaneously reduce the seismic and frost heave risks of brick–concrete buildings.

The application of structural energy dissipation and damping technology and base isolation technology makes it possible for buildings not to collapse during earthquakes and is one of the most effective means to mitigate earthquake hazards. Structural energy dissipation and damping technology is to install energy dissipation devices in certain parts of the structure, such as metal dampers and viscous dampers [1]. The base isolation technology places isolation materials with bearing capacity between the building foundation and the superstructure, such as laminated rubber isolation bearings and friction pendulum bearings [2,3]. Currently, structural energy dissipation, damping technology, and base isolation technology are widely used in improving the seismic performance of new or existing important buildings [4,5]. However, due to their high costs, these two methods are not feasible in masonry structure buildings in rural areas. Therefore, scholars have developed many economical and effective isolation systems [6,7,8,9,10,11], which are an attractive and feasible option for the isolation of rural buildings, but these isolation systems set flexible or sliding interfaces between the structure and its foundation. For high and cold regions, the frost heave risk is usually related to the freezing of moisture in the foundation soil, and these isolation technologies cannot be directly used to reduce the frost heave risk.

In recent years, geotechnical isolation technology using low-modulus materials around structure foundations has garnered significant research interest. This technology involves treating the foundation soil beneath structures, offering a potential solution to simultaneously mitigate the threats of earthquakes and frost heave in high-seismic and cold regions. The GSI technique that uses the rubber sand particles as the foundation pit backfill material (Geotechnical Seismic Isolation System based on rubber–soil mixtures, GSI-RSM) was first proposed by Tsang [12]. Subsequently, scholars conducted extensive research in both numerical and experimental aspects [13]. In terms of numerical research, Senetakis et al. [14] and Pitilakis et al. [15] investigated the response of a rubber sand layer through numerical simulation. Numerical studies by Pitilakis et al. [16] and Panjamani et al. [17] showed that the rubber sand layer was effective in decreasing base shear and maximum inter-layer displacement of mid- and high-rise buildings. Brunet et al. [18] found that a rubber–soil mixture of only 2–3 m thick was sufficient to achieve a good structural seismic response. Tsang and Pitilakis [19] analyzed the dynamic behavior of buildings equipped with a GSI system using a lumped parameter analysis model, and they also proposed the equivalent linear stiffness, viscous damping coefficient, and rocking radiation damping coefficient for ground and pre-built foundations. Forcellini [20,21] established a 3D numerical finite element model to evaluate the isolation potential of the GSI system. In terms of experimental research, Hazarika et al. [22] found that the use of tire debris and/or sand-mixed tire debris as a compressible cushion could not only reduce the dynamic load of the structure but also significantly reduce the dynamic-induced permanent displacement of the structure. Xiong and Li [23] and Bandyopadhyay et al. [24] carried out small-scale shaking table tests on the GSI-RSM system, where rubber–sand mixtures were positioned below the concrete or rigid block as the isolation material. Anastasios et al. [25] performed direct shear tests to sort out the failure mechanism of the rubber sand layer, further confirming its applicability in seismic isolation. Additionally, Banović et al. [26,27] proposed a GSI system based on a pebble layer sliding mechanism. The isolation performance of this system was validated through experiments, and the impact of foundation size on the system’s isolation efficiency was also examined. Tsiavos et al. [28] introduced a new isolation strategy, which was defined as the “PVC sandwich” isolation strategy and evaluated the effectiveness of the proposed design method through the experimental study of the isolation structure. Kuvat and Sadoglu [29] conducted cyclic triaxial tests on an asphalt and sand mixture that serves as damping material for the GSI system. The results showed that the asphalt and sand mixture had stronger rigidity and damping. Forcellini [30] discussed several geotechnical isolation (GSI) techniques for bridge structures to assess seismic vulnerability. Previous scholars have conducted extensive numerical and experimental research on the GSI-RSM system, but because of the relatively high material cost of rubber sand particles, the system has not yet been applied in engineering practice, and on-site data cannot be obtained to verify its reliability. Furthermore, it remains unclear whether replacing soil with rubber sand can effectively eliminate the threat of frost heave in foundations.

Sand is a widely used, low-cost building material. Replacing soil with sand can reduce the threat of frost heave, and sand also has isolation properties that help mitigate earthquake forces. However, there is still a lack of shaking table tests on entire masonry structures with sand cushion foundations. While a shaking table test was conducted on a brick masonry building with a sand cushion [31], the existing conclusions did not take the influence of foundation soil into account. This article proposed an economically efficient GSI-SC system, where a sand cushion is introduced between the structural foundation and natural soil to replace the original permanent frozen soil, and the sand that serves as a buffer layer is backfilled around the foundation to limit sliding displacement, as illustrated in Figure 1. Large-scale shaking table tests of the site foundation structure system were conducted on a single-layer masonry structure model to verify the effectiveness of the GSI-SC system, and numerical simulations were conducted on the shaking table tests using ABAQUS software version 6.14 to verify the seismic response of buildings equipped and non-equipped with the GSI-SC system.

2. Shaking Table Tests

2.1. Similarity Relationships

For shaking table tests with 1-g gravity acceleration, the gravity similarity of the soil is too difficult to achieve as a result of its uniqueness, and the similarity relationships between the structure and soil cannot be satisfied simultaneously. Thus, this study did not take into account the similarity relationship of soil. For the structural model similarity ratio, the length l, elastic modulus E, and density ρ were selected as three basic quantities. Considering the size limitation and bearing capacity of the shaking table, as well as the fact that the weight of the model should not affect the structural stiffness, the under-artificial mass model was adopted to approximately satisfy the similarity relationships.

Table 1. Structural model similarity.

Variable	Similarity Relation	Similarity Ratio	Remark
Length (l)	Sl	0.25	Basic quantity
Young’s modulus (E)	SE	1.00	Basic quantity
Density (ρ)	Sρ	2.00	Basic quantity
Stress (σ)	Sσ = SE	1.00	/
Time (t)	Sl(Sρ/SE)0.5	0.35	/
Displacement (u)	Sσ/SE	1.00	/
Velocity (v)	(SE/Sρ)0.5/Sl	2.83	/
Circular frequency (ω)	SE/(SlSρ)	2.00	/

summarizes the similarity relationships of the structural model.

2.2. Specimen Details

The prototype structure referenced in the tests is a brick masonry structure with ring beams and structural columns commonly used in rural areas. The plan size is 7200 mm × 5700 mm, and the story height is 3300 mm. The plan size of the scaled-down test structure model is 1800 mm × 1425 mm, and the layer height is 825 mm. The plan and elevation diagrams of the structural model and the overall picture after fabrication are shown in Figure 2.

To compare the seismic isolation effect of the GSI-SC system, two groups of tests were designed, i.e., the non-isolation test without the GSI-SC system and the isolation test with the GSI-SC system (hereinafter referred to as the GSI-SC test). The foundation soil models of the two groups of tests are displayed in Figure 3. The foundation soil model was completed in the laminated shear soil box by the divided compaction method, and the total thickness of the foundation soil was 1 m (Figure 4) [32]. The foundation soil model is constructed in five layers, with each layer having a thickness of 200 mm. Once the first layer is constructed, it is compacted and shaved, followed by the construction of the next layer.

2.3. Material Composition

To better simulate the actual project conditions, the natural soil and sand in the foundation materials were represented by silty clay and river sand, respectively.

Table 2. Physico-mechanical parameters of soil.

Type of Soil	Density (g·cm−3)	Moisture Content (%)	Internal Friction Angle (°)	Cohesion (kPa)
Clay	1.88	17.31	26.36	37.61
Sand	1.60	6.53	29.25	8.16

presents the physico-mechanical parameters of natural soil and sand. Figure 5 shows the cumulative gradation curve of sand. The coefficient of nonuniformity and curvature coefficient are 10.37 and 0.0247, respectively. Laboratory dynamic triaxial tests on natural soil and sand samples were carried out. The curves of the dynamic shear modulus ratio and the damping ratio versus shear strain are depicted in Figure 6. Compared with natural soil, the dynamic shear modulus of sand attenuated faster, and the damping ratio increased with increasing shear strain. These properties of sand under an earthquake will help the isolation system consume seismic energy. Masses of the test model and experimental device are shown in

Table 3. The masses of the test model and experimental setup.

Test Model	Masses (kg)
	Natural Soil	Sand	Laminated Shear Soil Box	Structural Model	Artificial Mass
NIT scenario	13,931	-	2300	1800	700
IT scenario	9570	3712	2300	1800	700

.

2.4. Test Setup and Measurement

In the shaking table test, the data acquisition system has an amplifier frequency response range of DC~100 kHz (−3 dB) and a maximum continuous sampling rate of 256 kHz/channel, and each channel is equipped with an independent 24-bit A/D converter. The sensor suite includes an acceleration sensor and a displacement sensor. The acceleration sensor is a piezoelectric three-way accelerometer (see Figure 7a) with a measurement range of ±5 g and a frequency range of 0.2~500 Hz. The displacement sensor (see Figure 7b) is a guyed displacement sensor with a guyed length of 5 m and a frequency range of 0–30 Hz. To compare the seismic response of the NIT and IT scenarios, the sensor arrangement for both test scenarios is identical. Figure 8 illustrates the sensor layout for the NIT and IT scenarios. Horizontal and vertical scales are carefully arranged within the shear soil box to ensure accurate positioning and distance between measuring points. The acceleration sensor within the soil is denoted as A, with its movement restricted by compacted surrounding soil. The acceleration sensor on the structure is represented as SA, forming a stable adhesion with the structure through a strong adhesive. The guyed wire displacement meter, indicated by D, is installed on a stable body outside of the table, with its fixing head secured onto the structure.

2.5. Ground Motion Input

According to the ‘seismic design standard for buildings’ (GB50011-2015) [33], the site type of the prototype structure is class II. The north–south component of the EL Centro seismic wave recorded in California, the United States, in 1940 was the first record of a strong earthquake with a peak acceleration greater than 0.3 g [34], which was used as the input wave to conduct horizontal excitation on the model. White noise refers to a type of noise with a power spectral density that is uniformly distributed across the frequency domain. In other words, it is random noise with an equal energy density at all frequencies. Frequency sweep, on the other hand, involves the repetitive scanning of narrow-band interference signals at a specific conversion speed and frequency interval within a certain frequency band. Both white noise and frequency sweep techniques are commonly utilized in shaking table tests for testing and evaluating the dynamic characteristics of various structures, which play play a crucial role in assessing structural performance under different scenarios. Prior to the test, white noise was input for frequency sweep, and the PIA of the seismic wave was adjusted to 0.1 g, 0.2 g, and 0.4 g, respectively. Figure 9 shows the ground motion characteristics of the input seismic waves.

3. Test Results and Discussion

3.1. Cracking Patterns and Damage Modes

NIT scenario: When PIA = 0.1 g, the structure model exhibited slight vibration. When PIA = 0.2 g, the vibration amplitude of the structure model increased, but no visible cracks were found on the brick wall yet. When PIA = 0.4 g, the structural model vibrated violently, and cracks appeared in weak parts, such as door and window openings on the brick wall. The cracks developed along the mortar joints between the bricks (Figure 10).

IT scenario: When PIA = 0.1 g, the response of the structure model was basically the same as that of the non-isolation test, and only slight vibration occurred. When PIA = 0.2 g, a slightly increased vibration amplitude was identified for the structure model. When PIA = 0.4 g, the vibration amplitude of the structure model further increased, but no cracks were observed on the brick wall.

By comparing the response features of the structural model, it could be concluded that there were no obvious cracks in the structure model in the IT scenario. However, cracks were identified in the doors and windows of the structure model after the non-isolation test with a PIA of 0.4 g. The model yielded better seismic performance in the IT scenario than that in the NIT scenario.

The vertical pressure on the backfill soil near the foundation is minimal, resulting in a small contact force between soil particles. Consequently, the frictional force between particles is also reduced, potentially leading to a decrease in the strength of the backfill soil. In cases where slippage occurs between the upper structure and the foundation soil, the backfill soil is susceptible to damage when compressed by the foundation beam. Figure 11 shows the damage at the backfill of the structural foundation when PIA = 0.4 g. It was observed that the backfill of the NIT scenario did not undergo apparent damage after seismic loading, indicating that the backfill effectively limited the movement of the structure foundation. Cracks were observed in the backfill in the IT scenario. A relative sliding movement was identified between the structural foundation and sand cushion, while the movement could reduce the transfer of seismic energy to the upper structure. The foundation had an extrusion effect on the backfill sand, resulting in the shear failure of the backfill sand and the formation of cracks.

3.2. Acceleration Response

Table 4. The peak acceleration of the roof panel of structural models.

PIA (g)	NIT Scenario			IT Scenario
	Table (g)	Roof Slab (g)	Amplification Coefficient	Table (g)	Roof Slab (g)	Amplification Coefficient
0.1	0.106	0.316	2.987	0.106	0.280	2.646
0.2	0.241	0.643	2.672	0.268	0.553	2.062
0.4	0.545	1.000	1.836	0.659	0.795	1.206

shows the peak acceleration of the roof panel of the structure models. The acceleration response of the roof panel was normalized to compare the acceleration response of the two groups of test roof panels more intuitively. The normalized time-dependent acceleration curves of the roof panel are presented in Figure 12. When PIA = 0.1 g, the acceleration responses of the roof panel in the IT and NIT scenarios were similar. When PIA = 0.2 g and PIA = 0.4 g, the peak accelerations of the roof panel in the IT scenario were only 77% and 66% of those in the NIT scenario (i.e., the peak accelerations were reduced by 23% and 34%, respectively, for the roof panel). It was identified that the GSI-SC system dramatically reduced the seismic response of the structure model. With the increase in PIA, the acceleration response of the roof panel equipped with the GSI-SC system decreased gradually.

3.3. Displacement Response

Figure 13 shows the relative displacement time-dependent curve between the structure foundation and natural soil. The peak relative displacement in the IT scenario was 3.4 mm when PIA = 0.4 g, and the residual relative displacement was also small. However, for the IT scenario with PIA = 0.1 g and 0.2 g under all loading conditions, the peak relative displacement was also less than 1 mm, with almost no relative displacement. The residual relative displacement in the IT scenario was caused by the effect of the isolation system. During the IT scenario, a portion of the seismic energy is converted into relative movement between the structural foundation and the foundation soil. This transformation results in a reduction in seismic energy transmitted to the superstructure, thereby decreasing the overall seismic response of the structure.

Table 5. Peak inter-story displacement.

PIA (g)	Peak Inter-Story Displacement (mm)
	NIT Scenario	IT Scenario
0.1	1.35	1.18
0.2	2.47	1.79
0.4	4.16	2.53

presents the peak inter-story displacement of the structure. When PIA = 0.1 g, the peak inter-story displacements of the structure in the IT and NIT scenarios were 1.35 mm and 1.18 mm, respectively, which indicated that the GSI-SC system reduced the peak inter-story displacement of the structure by 13%. When PIA = 0.2 g and PIA = 0.4 g, the peak inter-story displacements of the structure in the NIT scenario were 2.47 mm and 1.79 mm, respectively, while those in the IT scenario were 4.16 mm and 2.53 mm, respectively, which were reduced by 27% and 39%, respectively. The results revealed that the GSI-SC system was not able to provide significant damping in small-magnitude scenarios, whereas it could deliver better seismic isolation in larger-magnitude scenarios.

4. Numerical Simulation

Model Setup

Numerical simulations of the shaking table test were performed in ABAQUS software. Figure 14 presents the three-dimensional finite element model. The soil and structure were simulated by solid elements. In the shaking table test, the foundation soil model is placed within a laminated shear soil box. By setting the bottom boundary as a fixed boundary, it can restrict the displacement or deformation of the bottom of the foundation soil model. Additionally, by designating the side boundary as a bound boundary, it can effectively represent the shear deformation behavior of the foundation soil model in the shaking table test. The tie binding constraint and contact constraint were applied between the structure foundation and soil for the NIT and IT scenarios, respectively. Furthermore, normal and tangential behaviors were simulated by hard contact and friction models, respectively. The friction coefficient was set to be 0.48, according to the results of the friction coefficient test of concrete and sand cushions (

Table 6. Friction coefficient test.

Sequence	Normal Pressure (kPa)	Friction Coefficient	Average Friction Coefficient
1	3.6	0.475	0.475	0.480
2		0.475
3		0.475
4	7.2	0.465	0.476
5		0.484
6		0.478
7	10.8	0.529	0.490
8		0.475
9		0.466

).

The parameters of the structural model were obtained from the material properties test. The elastic modulus, Poisson’s ratio, and density of the concrete were 25.5 × 10⁹ Pa, 0.2, and 2.3 × 10³ kg/m³, respectively, while those of the brick wall were 1031 MPa, 0.15, and 1.8 × 10³ kg/m³, respectively. The foundation beams, constructional columns, ring beams, and roof panels of the structure were made of concrete. The CDP (Concrete Damaged Plasticity) model in ABAQUS was utilized to simulate the concrete. The CDP model is illustrated in Figure 15. The compressive behavior of the masonry was simulated by the compressive constitutive model proposed by Liu et al. [35]. The tensile constitutive of masonry is more complex with many influencing factors. The tensile failure mode of masonry is similar to that of concrete, so the tensile behavior of the masonry was simulated using the tensile constitutive model of the concrete. This model considers the anisotropic properties of materials caused by cracks. The behavior of material compression is assumed to be linear elasticity. The brittle fracture criterion can make the material fail when the tensile stress is too large, which is especially suitable for considering the brittle fracture of materials under tensile stress. The model parameters of concrete and masonry are listed in

Table 7. Model parameters for concrete and masonry.

Material	Elastic Modulus (MPa)	Poisson Ratio	Dilation Angle (deg)	Eccentricity	Coefficient of Viscosity
Masonry	1013	0.15	30	0.1	0.005
Concrete	25,500	0.2	30	0.1	0.005

, and the damaged plastic model for masonry is shown in Figure 16.

The Mohr–Coulomb model was adopted as the foundation soil model material, and the elastic modulus E was approximately determined by Equations (1)–(3) [36], where $f_1$ was obtained from test data. The maximum relative displacement D of the foundation soil model was determined by experimental data. The maximum shear strain of foundation soil was obtained by the ratio of $D_{max}$ to soil thickness H. According to the equivalent linearization, the damping ratio was determined by taking 0.65 times ${\gamma }_{max}$ of the damping curve of soil (Figure 9).

$$f_1={\displaystyle \frac{V_s}{4H}}$$

(1)

$$2G=\rho V_s^2$$

(2)

$$E=2G(1+\nu )$$

(3)

where $f_1$ is the first-order frequency of the foundation soil model, which is determined by the test data;  $V_s$ is the shear wave velocity of foundation soil; H is the thickness of foundation soil; G is the shear modulus of soil; $\rho$ is the density of soil; and $\nu$ is the Poisson’s ratio of soil.

5. Discussion

5.1. Simulation Method Reliability

The modal analysis, seismic damage, structural acceleration response, and isolation rate of the test and simulation results were investigated to verify the reliability of the numerical modeling method. Also, numerical simulations were carried out to further assess the effectiveness of the GSI-SC system in seismic isolation.

5.1.1. Modal Analysis

Figure 17 presents the modal analysis results of isolated and non-isolated models. It was indicated that the first-order frequencies of the isolated and non-isolated models obtained by simulations were 14.93 Hz and 15.17 Hz, respectively, while those obtained by shaking table tests were 15.25 Hz and 15.64 Hz, respectively, with errors of 2% and 3%, respectively. The simulation results matched well with the test results.

5.1.2. Seismic Damage

Figure 18 presents tensile damage programs in the structure model. There was no tensile damage in the masonry of the structure model in the IT scenario. When PIA = 0.4 g, the wall of the structure model in the NIT scenario experienced tensile damage, which was mainly distributed near the vulnerable parts. The numerical results were consistent with the seismic damage on the brick wall of the structure model in the test.

Table 8. The peak equivalent plastic strain (PEEQ) in backfill sand.

Amplitude	PIA = 0.1 g	PIA = 0.2 g	PIA = 0.4 g
Peak PEEQ	2.87 × 10−1	4.25 × 10−1	3.64

lists the peak equivalent plastic strain (PEEQ) in the backfill sand. Figure 19 displays the cumulative plastic deformation of soil when PIA = 0.4 g. The plastic strain appeared in the backfill sand, which was consistent with the cracking phenomenon in the IT scenario when PIA = 0.4 g. The backfill sand was damaged under the seismic action. The equivalent plastic strain of the backfill sand increased gradually with increasing the acceleration amplitude, indicating that the damage degree of the backfill sand increased gradually.

5.1.3. Structural Acceleration Response

Figure 20 shows the acceleration time-dependent curve of the roof panel under test and simulation conditions. The acceleration waveform changes to the simulation and the test results were basically the same, and the acceleration amplitudes were generally close to each other.

Table 9. Difference between the acceleration amplification coefficient of the structure from the simulation results and test records.

Simulated Conditions	Amplitude (g)	Position	Test Results	Simulation Results	Relative Error (%)
Simulation of IT scenario	0.1	Foundation	1.975	1.861	5.8
		Roof slab	2.646	2.616	1.1
	0.2	Foundation	1.318	1.435	8.9
		Roof slab	2.062	2.259	9.6
	0.4	Foundation	0.884	0.979	10.7
		Roof slab	1.206	1.341	11.2
Simulation of NIT scenario	0.1	Foundation	2.133	2.079	2.5
		Roof slab	2.987	3.224	7.9
	0.2	Foundation	1.447	1.574	8.8
		Roof slab	2.672	2.982	11.6
	0.4	Foundation	1.116	1.226	9.9
		Roof slab	1.836	2.051	11.7

lists the structural amplification coefficients obtained from the tests and simulations. There is a specific error in the acceleration amplification coefficient of the structural roof panel and the structural foundation obtained from both the shaking table test and the numerical analysis. This discrepancy may be attributed to variations in the physical and mechanical properties of the model materials used in the numerical analysis, as well as simplifications and assumptions made regarding sliding friction contact between the structural foundation and the foundation soil. In Table 9, the acceleration amplification coefficient relative errors are basically below 12%. The errors between the numerical simulation and the measurements can be acceptable. This indicates that the numerical analysis results are quite reasonable.

5.1.4. Isolation Rate

Table 10. The acceleration amplification factor and isolation rate of the structural roof panel.

Seismic Waves	Amplitude (g)	Acceleration Amplification Factor		Isolation Rate (%)
		Simulation of NIT Scenario	Simulation of IT Scenario
EL Centro wave	0.1	3.224	2.616	18.9
	0.2	2.982	2.259	24.3
	0.4	2.051	1.314	35.9

presents the acceleration amplification coefficient and isolation rate of the roof panel. With the increasing acceleration amplitude, the simulated acceleration amplification coefficient of the roof plate in the IT scenario was noticeably lower than that in the IT scenario. Additionally, this decreasing trend was more prominent as the acceleration amplitude continued to increase. The isolation rate increased with increasing acceleration amplitude. The isolation rates for PIA = 0.2 g and PIA = 0.4 g were 24.3% and 36.3%, respectively, reflecting the good isolation performance of the GSI-SC system. As the seismic action increased, sand had a higher damping ratio and could consume more seismic energy than natural soil. Furthermore, with the increase in seismic action, a relative slip would occur between the structure foundation and the sand cushion, limiting the transfer of seismic energy to the upper structure. The magnitudes of the isolation rate obtained by simulations and tests were similar.

In summary, the numerical simulations matched well with test results regarding the modal analysis, seismic damage, structural acceleration response, and isolation rate.

5.2. Isolation Mechanism

In this paper, the GSI-SC system was proposed for masonry buildings in strong earthquakes and cold regions. A large-scale shaking table test with and without a GSI-SC system was carried out on a single-story masonry structure model. The test results show that for PIA = 0.4 g, the GSI-SC system reduces the acceleration of the roof panel and the inter-story displacement of the structure by 33 % and 39 %, respectively. The GSI-SC system significantly reduces the seismic response of the structure and effectively achieves isolation.

In reality, the GSI-SC system primarily dissipates seismic energy through frictional slip between the foundation beam and the sand cushion, as well as collisions between the foundation beam and the backfill sand. This process can be referred to as friction and damage dissipation energy, which effectively reduces the shear force transmitted to the structure, consequently lowering the seismic demand for structural design. Furthermore, the backfill sand serves to limit excessive sliding displacement. These characteristics collectively constitute the primary isolation mechanism of the GSI-SC system.

5.3. Potential Practical Challenges

The isolation efficacy of the GSI-SC system has been validated through shaking table tests. However, experiments typically concentrate on a scientific issue and only take into account the key factors. Nevertheless, for promoting the GSI-SC system to practical engineering applications, it is essential to deliberate on its potential practical challenges, such as long-term performance, maintenance demands, and suitability for different soil types.

The long-term performance and maintenance requirements of a GSI-SC system are affected by multiple factors, including material selection, construction quality, environmental conditions, and soil type. The long-term performance of the GSI-SC system is mainly manifested in material durability, settlement and deformation, and water drainage. The sand bedding of the GSI-SC system, although stable in the majority of environments, might undergo particle breakage, abrasion, and decomposition under high-load circumstances. The sand cushion might settle due to particle rearrangement and compaction at the early stage; thus, it should be adequately compacted during construction to minimize subsequent settlement. Maintaining excellent drainage performance is crucial for the long-term stability of the sand mat, and if the surrounding soil or construction waste blocks the drainage path, it may lead to moisture accumulation and weaken isolation performance.

In terms of maintenance, regular inspection and monitoring of the settlement of the GSI-SC system and the structure above it, especially after major earthquakes or extreme load events, can help detect and repair settlement or deformation problems in a timely manner. Additionally, regular inspection and maintenance of the drainage system to ensure smooth drainage and prevent the accumulation of moisture that could lead to the deterioration of isolation performance are necessary. In areas with high groundwater levels, measures such as installing drainage wells or using waterproof materials to protect the sand cushion may be required to lower the groundwater level or prevent water from invading the sand cushion.

Overall, the sand cushion isolation technology has good long-term performance, but its effectiveness depends on material selection, construction quality, and post-construction maintenance. Regular monitoring and maintenance, especially the inspection of settlement and drainage performance, are important means to ensure the long-term effectiveness of the sand cushion isolation system. For specific environmental conditions, corresponding measures may need to be taken to optimize performance and extend its service life.

6. Conclusions

The GSI-SC system is a cost-effective and efficient solution, utilizing ordinary sand material that is easy to construct. The GSI-SC system serves the dual purpose of mitigating foundation frost heave and providing seismic isolation, making it particularly suitable for low-rise rural buildings in seismically active and cold regions. In practical engineering applications, the sand cushion must meet both the bearing capacity and isolation requirements of the foundation. A reasonable construction plan should then be designed based on the seismic demands.

The effectiveness of the isolation of the GSI-SC system has been verified through research. The majority of houses in rural areas are low-rise buildings consisting of one or two floors. Furthermore, the isolation system demonstrates a significant response to near-fault effects [37,38]. However, due to the limitations of the research conducted (such as using a simple building model with only one floor and applying only one ground motion excitation), further study is needed to strengthen the conclusions drawn. Therefore, it is necessary to conduct additional experimental research on this topic and utilize numerical models for investigation, including exploring the influence of structural floor height (height–width ratio of structure) and near-fault effects on the response of GSI-SC systems.