Experimental Study of the Dynamic Characteristics and Microscopic Mechanism of Lightweight Soil Modified with Expanded Polystyrene and Sisal Fibre

Abstract

With the increasing demand and use of highways, railways and tunnels in China, the phenomena of foundation settlement, uneven deformation and ground cracking caused by the cyclic loading by traffic are becoming increasingly significant. There is now an emphasis on research to prevent or decrease these phenomena by mixing new materials into the soil body. In this study, cyclic loading tests were conducted on lightweight soils modified with expanded polystyrene (EPS) and sisal. A GDS true/dynamic triaxial apparatus was used to study the dynamic elastic modulus and damping ratio of clays with different dosages of EPS and sisal fibre. The modified soil samples were tested, and then, they underwent micro-scale analysis. The results showed that, with the continuous increase in EPS doping and dynamic stress, the trend of the growth of the dynamic strain of the specimens increased. At the same time, with the increase in the dynamic strain, the dynamic elastic modulus decreased, and the trend increased with increasing doping of the soil with EPS particles. A comparison of the improvement effect coefficient of the soil samples showed that the most suitable EPS doping volume was 5%. Different dosages of sisal fibre were added to the most suitable EPS-modified lightweight soil, and the dynamic elastic modulus first increased and then decreased with increasing sisal dosage. In addition, the damping ratio first increased and then decreased. The best dynamic performance of the soil was obtained when the dosage of sisal was 1.2%. Nuclear magnetic resonance and electron microscope scanning tests verified that, when the sisal doping was 1.2%, the soil particles had the largest compactness, the best interparticle bonding and the best improvement effect.

1. Introduction

Widely distributed throughout the world, clay is a fine-grained natural soil material that has good plasticity when wet but hardens and becomes brittle and is no longer plastic when dried or fired [1]. Clay is one of the materials commonly used in civil engineering projects. Due to its properties related to the mineralogical composition and water content, the clay layer often requires the use of various methods to strengthen/improve it or even completely replace it.

Expanded polystyrene foam granules (EPS granules) have the advantages of light weight, high compressibility, good stability, long-term durability, low moisture absorption and good thermal insulation, making them an ideal lightweight material for the improvement of other materials. Many scholars have conducted feasibility studies on EPS polypropylene as a road base filler. Fu et al. [2] found that the EPS particle admixture and particle size had a relatively obvious effect on the unconfined compressive strength of lightweight soil; they found that a higher EPS dosage or larger particle size reduces the unconfined compressive strength of the soil. Hou et al. [3] and Gao et al. [4] studied the dynamic shear modulus, dynamic elastic modulus and damping ratio characteristics of EPS-modified lightweight soil; Yang et al. [5] studied the degree of nonuniformity in the stress–strain distributions of EPS particles; and Hou et al. [6] investigated the static earth pressure characteristics of lightweight soil mixed with EPS particles behind a retaining wall. The results of all the above studies indicated that EPS-modified lightweight soils have good engineering properties. Sisal fibre has good mechanical properties and is widely used in military and civilian industries. As a high-quality natural fibre, it has many excellent properties, such as corrosion resistance, high yield, high fibre volume and high crack resistance. It consists mainly of cellulose, hemicellulose and lignin along with other small amounts of pectin and waxes, and these elements give it durability, strength and stiffness. Li et al. [7] found that sisal fibres could significantly improve the cohesion of the soil, but the increase in the angle of internal friction was small. Cao et al. [8] studied the early crack resistance of sisal fibre-modified concrete. Ma et al. [9] investigated the effect of organic polyurethane polymers and sisal fibres on the mechanical properties of sandy soils. Hao et al. [10] studied the unconfined compressive strength of soils improved with composites of fly ash and sisal fibre. Dai et al. [11] studied the strength, stability and curing of sisal fibres combined with an ionic curing agent under static loading.

In summary, scholars at home and abroad have studied the static loads of soils modified with EPS and sisal fibres and achieved a large number of scientific results, while less research has been conducted using dynamic loads. In addition, no one has yet studied the mechanics of EPS and sisal fibres incorporated into clay at the same time. The study of the dynamic properties of EPS-sisal-modified lightweight soils is particularly relevant because of the more complex situation of the forces affecting the soil under dynamic loading conditions. Based on this background, this study compared and analysed the effects of different sisal fibre admixtures on the dynamic properties of EPS-modified lightweight soils under the same surrounding pressure and frequency conditions and established a normalised model for the calculation of the decay of the dynamic elastic modulus. The microstructure and pore characteristics of the EPS-sisal-modified lightweight soils were analysed from a micro perspective through nuclear magnetic resonance and scanning electron microscopy. All the above research results will provide a reference basis for practical related projects.

2. Test Material and Programme

2.1. Test Apparatus

The test was conducted using a British GDS motorised true/dynamic triaxial apparatus, which could perform true and dynamic triaxial tests on soils; the maximum peripheral pressure, axial load and vibration frequency were 2 MPa, 20 KN and 5 Hz, respectively, as shown in Figure 1. The specimen was loaded using the dynamic loading module of the apparatus. The measurement system can accurately record the specimen axial pressure, circumferential pressure, counterpressure and other changes in the data; the test selection of the power loading module incorporates the use of 0.0001 mm high-precision sensors in real time to accurately determine the axial strain.

The test apparatus in Figure 1 consisted of an electronically operated display, the main test machine, a standard pressure volume controller, a single-channel perimeter pressure controller, a perimeter pressure, a counter pressure transducer and an actuator.

2.2. Test Materials

2.2.1. Clay

The clay was taken from a road project in Wuhan, which is slightly wet, hard plastic, rich in calcium nodules and mainly composed of montmorillonite, illite, kaolinite and other clay minerals. The proportion of soil particles with a diameter of ≤0.075 mm reaches 72.1%, and the free swelling rate is 20%, which is a low swelling rate and a non-expansive clay. And the soil sample was in the form of a large dark yellow lump with the physical parameters shown in

Table 1. Basic physical parameters of clay.

Natural Density ρ/(g/cm3)	Specific Gravity ds/(−)	Liquid Limit WL/(%)	Plastic Limit WP/(%)	Moisture Content/(%)
1.7	2.68	39.46	19.46	21.65

. According to the Standard for Geotechnical Test Methods (GBT50123-2019) [12], the soil sample should be compacted in 5 layers, and the moisture content should be controlled at 20% (optimum moisture content). Each compaction should maintain a uniform distribution of soil particles and composites and be scraped to prevent them from developing a weak surface.

2.2.2. EPS Particles

The test was conducted with a 0.5–1 mm particle size foam for blending. The bulk density was 0.030 g/cm³, the pure particle density was 0.044 g/cm³, and the compacted density was 0.060 g/cm³.

2.2.3. Sisal Fibre

The sisal and the physical and mechanical parameters of the sisal fibre were provided by Guangdong Oriental Sisal Co., Ltd. (Shunde, China), and the parameters are shown in

Table 2. Physical and mechanical parameters of the sisal fibre based on the data of Guangdong Oriental Sisal Co., Ltd.

Wire Density/(g/km)	Unevenness of Linear Density/(%)	Breaking Strength/(cN)	Strong Unevenness/(%)	Elongation at Break/(%)
32.7	13.47	684.87	36.17	1.87

. Wu et al. [13] found that, when the sisal fibre length was controlled at approximately 5 mm, the soil strength was greater than that of the other test groups, and the test results were the best. Thus, sisal fibre with a length of 4.5–5.5 cm was selected as the test material.

2.3. Test Programme

This is a small-scale, indoor, dynamic, triaxial test. In small-scale tests, manual mixing of soil with EPS pellets and sisal fibres is more capable of achieving homogeneity of their mixtures at the test level. Therefore, manual random homogeneous mixing and blending were used in this test. The remoulded soil sample was made into a standard cylinder that was 100 mm in height and 50 mm in diameter. It was first put into a saturator for pumping and saturation and then loaded into a GDS pressure chamber for backpressure saturation until the saturation coefficient B value reached above 0.98. The specified surrounding pressure was applied in a linearly increasing manner for drainage consolidation. Finally, a predetermined sinusoidally varying load was applied under drainage conditions [14]. The expression for each stage of the load is given by the following:

$${\sigma }_{\mathit{d}}={\sigma }_0+{\sigma }_{\mathit{m}}\mathit{sin}\left(2\pi \mathit{ft}\right)$$

(1)

where σ₀ is the initial static stress per stage, σ_m is the dynamic stress amplitude, f is the loading frequency, and t is the loading time.

The σ_m was divided into ten equal increments, keeping σ₀ 2.5 kPa greater than σ_m at each level to prevent the cap above the specimen from detaching from the soil sample during vibration. The test was terminated when the dynamic strain reached 5% or when the ten-stage load was reached. It was found that when controlling the EPS and sisal fibre as variables, the EPS volume fraction (EPS volume/whole soil sample volume × 100%) was within 20% [15] and the sisal fibre mass fraction (sisal fibre mass/whole soil sample mass × 100%) was less than 2% [7], it was more likely to find the optimum dose. Taking the above into account, the specific test protocol is shown in

Table 3. Testing program.

Specimen Number	Surrounding Pressure (kPa)	Frequency (Hz)	EPS Volume Ratio (%)	Dynamic Stress Amplitude (kPa)	Sisal Quality Ratio (%)
A-1	100	1	0	20–160	0
A-2	100	1	5	20–160	0
A-3	100	1	10	20–160	0
A-4	100	1	15	20–160	0
A-5	100	1	20	20–160	0
B-1	100	1	5	20–160	0.6
B-2	100	1	5	20–160	0.8
B-3	100	1	5	20–160	1.0
B-4	100	1	5	20–160	1.2
B-5	100	1	5	20–160	1.4

.

The test was divided into two parts, A and B, where the dynamic stress amplitude was controlled in the range of 20–160 [16] kPa. Part A was to find the optimum amount of polystyrene foam, using volume ratios of 0%, 5%, 10%, 15% and 20%, under the test conditions of 100 kPa perimeter pressure and 1 Hz [16] to find the maximum improvement effect coefficient. Part B was for the case where the test conditions remain unchanged; the optimum amount of EPS was added to the lightweight soil with mass ratios of 0.6%, 0.8%, 1.0%, 1.2% and 1.4% sisal fibres to find the optimum amount of sisal fibre to increase the dynamic modulus of elasticity and the damping ratio.

To further analyse the pore structure of the improved soil under the optimal dosage, the soil samples with different sisal dosages after the cyclic loading test were first put into a drying oven at 80 °C for 48 h to completely dry (This is because sisal fibres are organic fibres, which may cause damage to their internal structure under the action of high temperatures). Based on this, we chose to extend the drying time to make the soil in the conditions of 80 °C to achieve the effect of drying and then immersed the soil in water as a whole for 8 h to completely saturate. Finally, small, flat and smooth pieces of soil were taken from the middle of the soil samples and put into test tubes for the nuclear magnetic resonance (NMR) test.

SEM (scanning electron microscopy) tests were performed on soil samples with different sisal fibre densities at maximum dry density [17] to observe the structural system of particles and pores on the surface of the soil and to analyse the pore images. Attention should be given to the vacuum gold plating of the soil surface in advance to prevent charge build-up and interference with the image quality.

3. The Trend of Change in Dynamic Characteristics

3.1. Variation in the Dynamic Elastic Modulus of EPS-Modified Lightweight Soils

Figure 2 shows soil specimens with different EPS doping levels after the test (from left to right: EPS doping levels are 0, 5%, 10%, 15% and 20%).

The magnitude of the dynamic modulus of elasticity [18] expresses the magnitude of the stiffness and elastic kinetic energy of the soil at a given dynamic stress amplitude. It is expressed as the slope of the line connecting the two endpoints of each hysteresis curve, and it is given by the following:

$$E_{\mathit{d}}=\frac{{\sigma }_B-{\sigma }_A}{{\epsilon }_B-{\epsilon }_A}$$

(2)

where the numerators σ_B and σ_A are the maximum and minimum dynamic stresses in the hysteresis curve under each cyclic load, and the denominators ε_B and ε_A are the maximum and minimum dynamic strains in the hysteresis curve under each cyclic load.

Figure 3 and Figure 4 show the ε_d − σ_d and ε_d − E_d curves for light soils with different EPS admixtures, respectively. Figure 3 and Figure 4 show that the higher the EPS admixture is, the greater the tendency for the dynamic strain to increase with increasing dynamic stress, and the dynamic modulus of elasticity gradually decreases with increasing dynamic strain. This is because EPS has a lower strength relative to the soil and reduces the strength of the clay cementitious structure. When the dynamic load is increased, the soil with a higher dose of EPS undergoes greater deformation, the cement structure starts to breakdown slowly, pores and cracks gradually penetrate and the tendency of its dynamic strain growth becomes greater as the EPS dose increases. In the dynamic loading test, the deformation of the soil body primarily comes from the increase or decrease in the pore volume of the soil body and the compression of the EPS particles. The confined air in the EPS particles is compressed first, followed by the deformation of the EPS particles themselves, so that the external kinetic energy is lost and dissipated. The EPS particles are mixed with the clay to increase the spacing between the soil particles, and the bonding force between the EPS and the soil particles is very small, thus, reducing the cohesion of the clay.

When the weakest section is formed internally, the dynamic stress is primarily borne by the friction and cementation strength between the soil particles and EPS particles at which point the good elastic properties of EPS come into play and the compressive and tensile strains increase simultaneously until the cementation strength at the weakest part of the soil is completely lost and the dynamic load is borne entirely by the friction between the particles in the section, resulting in a sharp drop in the macroscopic dynamic modulus of elasticity, which is brittle damage [19]. At the same time, the excessive admixture of EPS can also lead to softening of the soil [15]; therefore, the admixture of EPS should not be too high.

To evaluate the improvement effect of EPS in lightweight soil, the soil sample improvement effect coefficient [20] is introduced, which is the ratio of the absolute mass reduction rate (i.e., the difference between the absolute mass at that EPS dosage and the absolute mass at zero EPS dosage divided by the mass at zero EPS dosage) to the absolute dynamic elastic modulus reduction rate (i.e., the difference between the value of the dynamic elastic modulus at that EPS dosage and the value of the dynamic elastic modulus at zero EPS dosage divided by the value of the dynamic elastic modulus at zero EPS dosage) at the same EPS dosage and reflects the efficiency of absolute mass reduction per unit of the dynamic elastic modulus reduction rate. The improvement effect coefficients for different EPS blending levels are shown in

Table 4. Improvement effect coefficient of light soil with different EPS contents.

EPS Volume Dose (%)	Absolute Quality Reduction Rate (%)	Absolute Dynamic Modulus of Elasticity Reduction (%)	Modified Effect Factor
5	4.82	12.36	0.39
10	9.64	28.24	0.34
15	14.46	45.12	0.32
20	19.28	62.56	0.31

. The improvement effect coefficients for the EPS lightweight soils were 0.39, 0.34, 0.32 and 0.31. As the improvement effect coefficient decreased with increasing EPS blending, the optimum volume blending level for the EPS lightweight soils was 5%.

3.2. Variation in the Dynamic Elastic Modulus of EPS-Sisal-Modified Lightweight Soils

Figure 5 shows the post-test EPS-sisal lightweight soils with different sisal admixtures at 5% EPS volume fraction (from left to right: the sisal fibre admixtures are 0.6%, 0.8%, 1.0%, 1.2% and 1.4%). Figure 6 and Figure 7 show the ε_d − σ_d and ε_d − E_d curves for the EPS lightweight soils modified with different sisal admixtures at 5% EPS volume fraction, respectively.

In the soil mixed with sisal fibres, the fibres in the soil body have an irregular staggered arrangement by means of a random distribution, forming a “network” structure. When the specimen is ruptured under a large load, the contact between the fibres and the soil particles and the skeleton structure formed at this time make the dislocation and deformation of the soil particles be subjected to a certain restriction and better inhibit the development of cracks in the soil, thus enhancing the integrity and toughness of the specimen. As shown in Figure 6, the stress–strain curves for the EPS-sisal-modified lightweight soils changed considerably at the beginning. With increasing amounts of sisal, the dynamic stresses first increased and then stabilised. This was because the increase in sisal fibres led to a rougher surface and an increase in inter-surface forces, and a greater dynamic stress was required to achieve the same strain [21]. This was because the increase in sisal incorporation under dynamic loading caused the difference between the maximum and minimum dynamic strains of the composite soil to decrease and then increase, changing the stiffness of the soil [14]. In Figure 7, the trend of each specimen was basically the same, between 0% and 0.5% of the strain, while the trend gradually diverged between 0.5% and 1%. When the strain was greater than 1%, the trend of each specimen was more divergent with different trends. This was because, when the soil deformation was 0 < ε_d < 0.5%, the soil sample was in the elastic deformation stage, the plastic deformation was small, and the soil was prone to rebound. When the deformation was ε_d > 0.5%, some of the specimens underwent plastic deformation, the resistance of the soil to deformation was reduced, and the difference in the dynamic modulus of elasticity of the soil with different sisal admixtures was significantly reflected at this time.

To further analyse the trend of the maximum value of the dynamic modulus of elasticity of the specimen [22] and the maximum value of E_dmax, the inverse of the dynamic modulus of elasticity and E_dmax curves were established, as shown in Figure 8 and Figure 9. The modulus of dynamic elasticity did not consistently increase or decrease with an increasing amount of sisal but increased when the amount of sisal was less than 1.2% and decreased when it was greater than 1.2%, i.e., the maximum was reached when the amount of sisal was 1.2% and the improvement effect was the best. According to the Kondner [23] model, the maximum dynamic modulus of elasticity is expressed as the following:

$$\frac{1}{E_{\mathrm{dmax}}}=\mathrm{a}+\mathrm{b}\times {\epsilon }_{\mathrm{d}}$$

(3)

where a and b are fitted parameters.

To study the relationship between the dynamic modulus ratio decay and dynamic strain for the EPS-sisal-modified lightweight soil, the E_d/E_dmax − ε_d relationship was plotted for different sisal fibre admixtures, as shown in Figure 10. Figure 10 shows that E_d/E_dmax − ε_d continued to increase with increasing sisal admixture and then tended to be stable. The Hardin–Drnevich model [24] cannot describe its curve variation trend well, according to the studies of Zhuang Xinshan [9] and Hou Tianshun [23] on the modified model. In this study, the improved Hardin–Drnevich curve model, i.e., the Darendeli model [25], was used; it is expressed as the following:

$$\frac{E_{\mathit{d}}}{E_{\mathit{dmax}}}=\frac{1}{1+{\left(\mathit{c}\times {\epsilon }_{\mathit{d}}\right)}^{\mathit{d}}}$$

(4)

$$\mathit{y}=\frac{1}{1+{\left(\mathit{c}\times \mathit{x}\right)}^{\mathit{d}}}$$

(5)

where c and d are fitted parameters.

The test points were closely distributed on both sides of the fitted curve, and the correlation coefficient of the modified model, R², was greater than 0.95, indicating a good fit. The model could accurately describe the variation in the dynamic modulus of the EPS-sisal-modified lightweight soil with dynamic strain, and the resulting decay model was as follows:

$$\frac{E_{\mathit{d}}}{E_{\mathit{dmax}}}=\frac{1}{1+{\left(4.6213{\epsilon }_{\mathit{d}}\right)}^{0.6985}}$$

(6)

3.3. Equivalent Damping Ratio Variation Pattern

The damping ratio is an important concept in the study of the dynamics of soil structures and expresses the significance of the damping magnitude of a soil structure, reflecting its ability to absorb energy and resist earthquakes [26]. The equivalent damping ratio λ of a soil is the proportion of energy dissipated by the soil under dynamic loads. The formula is as follows:

$$\lambda =\frac{1}{4\pi }\frac{\Delta W}{W}=\frac{1}{4\pi }\frac{S}{S_{\mathit{tri}}}$$

(7)

where W is the total energy obtained under each dynamic load, ΔW is the energy consumed under each dynamic load [27], S is the area of the hysteresis curve that indicates the energy consumption of the soil due to the damping ratio, and S_tri is the area of the triangle.

Figure 11 shows a typical hysteresis curve. The area enclosed by the hysteresis curve [13] can be approximated as ΔW under each dynamic load, and the area of triangle OAA′ is W under each dynamic load [8].

The formula for the total energy stored, W, is the following:

$$W=\frac{1}{2}{\sigma }_{\Delta }{\epsilon }_{\Delta }$$

(8)

where σ_Δ is the difference in value from the origin to the point of maximum dynamic stress in the hysteresis loop, and ε_Δ is the difference in value from the origin to the point of maximum dynamic strain.

Figure 12 shows the stress–strain hysteresis curve at one point in the test where 10 cyclic loads were performed at each stage of the test. The number of loads reaching the deformation protection value was approximately eight to twelve loads each time, and all the maximum deformation values at the end of the test were less than 15%.

Figure 13 shows the σ_d − ΔW and σ_d − W curves for the soil samples with different sisal admixtures at 5% EPS volume fraction. Figure 13 shows that, when the sisal admixture was less than 1.2%, the hysteresis curve area increased continuously with increasing admixture. This was because, at this time, the sisal fibre and soil particles could be well cemented together, and this enhanced the ability of the soil to resist deformation. The soil energy consumption became large, the sisal doping was greater than 1.2%, the hysteresis curve area was constantly shrinking, and energy consumption became smaller. This indicated that the sisal doping was too large. As a result, the soil particles did not have a good combination of sisal fibre, the soil body was relatively soft, and the consumption of energy decreased.

Figure 14 shows the variation in the damping ratio for normal clay and for the EPS-modified lightweight soil with different sisal dosages at 5% EPS volume fraction. Figure 14 shows that, at the beginning of the stress change, i.e., when the stress was less than 40 kPa, the damping ratio of all test groups, except the ordinary clay soil, showed a decreasing trend. This may have been because, under the action of the dynamic load, the internal pores of the soil and EPS particles were first and largely compressed. The overall shape of the modified soil remained almost unchanged at the beginning of the application of stress. The soil deformation under the action of dynamic loading was smaller, and less energy was consumed through the extrusion deformation This resulted in a smaller damping ratio. When the stress was greater than 40 kPa, the damping ratio increased with increasing stress in either test group for the following reasons. For the EPS-modified lightweight soils, EPS responded more to dynamic loads, consumed less energy through elasto-plastic deformation and had a denser internal structural connection. As the dynamic load increased, the soil sample gradually underwent greater elasto-plastic deformation, and the energy loss increased, which macroscopically expressed the increasing damping ratio of the soil. When the doping amount was greater than 1.2%, the degree of cementation between the sisal fibres and the soil body decreased. The tightness between the soil particles decreased so that, when it reached a certain degree of compactness, it showed incompressibility, reduced the efficiency of dynamic load transfer and consumed less energy, which was reflected in the decrease in the damping ratio.

4. Micro-Scale Analysis

4.1. Nuclear Magnetic Resonance (NMR)

The nuclear magnetic resonance (NMR) test was carried out using a MicroM12-025VR nuclear magnetic resonance analyser manufactured by Suzhou Newmax Analytical Instruments Co., Ltd. (Suzhou, China) with an instrumental strength of (0.28 ± 0.05) T, a resonance frequency of 12 MHz and a sampling coil size of 25.4 mm. Figure 15 shows the steps of the nuclear magnetic resonance (NMR) test. NMR is a technique often used in the field to analyse porosity, bound water saturation, pore size distribution, permeability and reservoir parameters or to observe the state of fluid distribution. When the sample is placed in a magnetic field, hydrogen plasmons absorb the energy of the signal and resonate when a certain pulse frequency is emitted. The relationship between the signal amplitude and the relaxation time T₂ of the pore water in the soil sample is obtained during the test. Since the sample is in a saturated state, T₂ can be regarded as the vector size of the pore, which is given by the following:

$$\frac{1}{T_2}\approx {\rho }_2\frac{S_{\mathit{pore}}}{V}=\frac{\lambda {\rho }_2}{\mathit{d}}$$

(9)

where T₂ is the lateral relaxation time (ms), ρ₂ is the relaxation coefficient, S_pore and V are the pore surface area and volume, respectively, d is the pore diameter, and λ is a factor related to the pore shape. According to the equation, the relaxation time T₂ is proportional to the pore volume V, i.e., the larger the relaxation time T₂ is, the larger the pore volume V.

Figure 16 shows the distribution curves of the transverse relaxation time T₂ for ordinary clay and for the EPS-modified lightweight soils with different sisal dosages at 5% EPS volume fraction. According to the above figure, the control and 1.0%, 1.2% and 1.4% sisal fibre admixtures all had bimodal distributions, which was due to the nonuniform diffusion coupling between the water molecules that remained in the saturated and unsaturated pores, where the trend of the signal intensity of each major peak was between 0.1 ms and 1 ms and that of the minor peaks was between 10 ms and 50 ms. However, the change in the peak value of the soil samples in each test group was relatively more obvious in the main peak, and the peak value of the secondary peak was too small and almost unchanged. Some scholars have shown that the relaxation time and soil pore size are directly proportional to the corresponding peak signal intensity, the number of pores is directly proportional to the pore size, and the peak area can indicate the soil pore size, that is, the pore volume and the water content of the specimen are positively correlated [17,27,28,29].

The comparison between controls 1 and 2 showed that the peak signal intensity and peak area of control 2 were larger, which indicated that the EPS-modified lightweight soil had more continuous small pores than the normal clay soil, the internal pore space was larger, the water content was higher, the EPS incorporation made the internal structure looser, and the number and size of the pores were larger than those of the normal clay soil. Comparisons between control 2 and the other test groups showed that the peak signal intensity of the sisal fibre significantly weakened after the incorporation of the sisal fibre. With increasing amounts of sisal fibre, the peak signal intensity decreased, and the peak area gradually decreased. This indicated that a series of physical reactions occurred after the sisal fibres were incorporated into the soil, and the soil grid structure that formed was conducive to improving its strength and dynamic properties. In addition, the sisal fibres made the soil particles more compact, which was equivalent to filling the soil pores and made the number and volume of the pores decrease continuously. However, when the sisal fibre admixture exceeded 1.2%, the peak signal strength rose, the peak area increased, the soil pore spacing and size increased in the reverse direction, the water content became larger, and the soil became loose. The reason for this was that the effect of the excessive admixture of sisal fibre on the internal structure of the soil produced damage. The degree of damage increased with the increase in the sisal fibre admixture. Excessive sisal fibre was not fully combined with the internal support structure of the soil. The excess sisal fibres were not completely combined with the internal support structure of the soil. The excess uncombined sisal fibres caused disruption to the internal support structure of the soil, resulting in a larger and more numerous pore spacing, Therefore, there was more water in the pores and an increase in the water content. In summary, at 1.2% sisal fibres, the internal structure of the soil was the most stable, the soil particles were the most compact, and the resulting mesh structure was the most conducive to improving the strength and dynamic properties of the soil [30]. The results of this test were consistent with the results of the study of the dynamic modulus and damping ratio.

4.2. Scanning Electron Microscopy (SEM)

The scanning electron microscope (SEM) test was carried out using a Japanese Zeiss SIGMA high-resolution field emission scanning electron microscope, which mainly consists of an electron optical system, a signal collection and processing system, an image display and recording system, a vacuum system, a power supply and a control system, etc. The resolution of the secondary electron image is approximately 5–10 nm, and the resolution of the backscattered electron image is approximately 50–200 nm. Figure 17 shows the steps of the scanning electron microscopy test. According to the dynamic triaxial and nuclear magnetic resonance test, which was performed for sisal fibre doping ≥1.2%, the structural strength of the soil body weakened with the increase in sisal fibre incorporation. So, the SEM test was conducted using two sisal-modified soils (1.2% and 1.4% sisal). Scanning electron microscopy was used to observe the soil particles, pore sizes and microstructure. To better explore the changes in the pores and structural characteristics, the electron microscope observation times were increased 1000 times, as shown in Figure 18. Figure 18a shows that, when the sisal fibre doping was 1.2%, the soil changed from the original large soil particles into smaller soil particles, and the bulk pores gradually changed to a wider distribution of small pores. When the sisal fibre doping reached 1.4%, as shown in Figure 18b, with the increase in sisal fibre doping, the internal structure of the soil body changed, part of the cross-section of the soil body contained more granular and spherical particles, and large or rough soil particles gradually cemented together to become larger particles of the soil block. Thus, the soil particles could not be better cemented between the filling [31], and the fine pores gradually changed into isolated larger pores that were spaced apart. This showed that too much sisal fibre led to pores that were more angular and more complex, and the soil was less homogeneous.

5. Conclusions

(1) Under the condition of constant surrounding pressure and frequency, with the increase in the expanded polystyrene (EPS) admixture, the growth trend of the soil dynamic strain became larger with increasing dynamic stress, while the dynamic elastic modulus continued to decrease. The optimal EPS volume admixture was 5%, which was obtained from the analysis of the improvement effect coefficient. This lightweight soil can be used for some pavement foundations where lightweight foundations are required and strength requirements are low.

(2) Lightweight soil was modified with 5% EPS by volume and doped with sisal fibres. When the sisal doping was less than 1.2% mass fraction, the elastic modulus and damping ratio increased with the increasing amount of sisal. When the amount of sisal doping was greater than 1.2% mass fraction, the dynamic elastic modulus and damping ratio decreased with increasing sisal doping.

(3) Compared to unmodified and EPS-modified lightweight soils, the EPS-modified lightweight soil with 1.2% sisal admixture had a significantly higher dynamic modulus of elasticity and an approximately two-fold increase in the damping ratio.

(4) E_d/E_dmax − ε_d shifted upward first and then stabilised with the increasing amount of sisal admixture, and the normalised model for the decay of the dynamic elastic modulus of the EPS-sisal-modified lightweight soil was established, based on the Darendeli model.

(5) Nuclear magnetic resonance (NMR) and search engine marketing (SEM) tests showed that the EPS-improved soil had the best structural form with smaller and tighter pores at 1.2% sisal admixture. The results of this microscopic test well verified the enhancement of the dynamic properties of the improved soil.

Based on this experiment, further investigations will include an attempt to answer the question of whether the improved soil body would still have the improvement of various mechanical properties and have the practical engineering applications under freeze–thaw and other special conditions.