A method for improving expansive soil and a method for predicting its performance based on the clay mineral composition.

Abstract

This invention discloses a method for improving expansive soil based on its clay mineral composition and a method for predicting its performance. The method for improving expansive soil based on its clay mineral composition includes the following steps: obtaining the optimal admixtures for montmorillonite, illite, and kaolinite clay minerals in expansive soil under different moisture contents through molecular dynamics simulations; after determining the optimal admixtures for the three clay minerals, adding the corresponding admixtures to the corresponding minerals, preparing soil test blocks for testing, and determining the optimal dosage based on the test results; and configuring admixtures to improve the properties of the expansive soil based on its moisture content and mineral composition. This invention constructs a coarse-grained model of real soil particles and corresponding admixtures, tests various properties of the model, and combines machine learning to predict the improvement effect of admixtures on expansive soil under different working conditions. The method of this invention is applicable to the improvement and performance prediction of different expansive soils in different regions and has universality.

Description

Technical Field

[0001] This invention relates to the field of clay mineral technology, and specifically to a method for improving the clay mineral composition of expansive soil and a method for predicting its performance. Background Technology

[0002] As a special type of soil with high plasticity, expansive soil is composed of highly hydrophilic minerals (representative minerals include montmorillonite, kaolinite, and illite), making it extremely sensitive to water. It exhibits numerous characteristics such as swelling upon water absorption, shrinking upon water loss, and a sharp drop in strength upon contact with water. Furthermore, due to the high overlap between seasonally frozen soil areas and expansive soil regions in my country, the impact of freeze-thaw cycles on expansive soil is significant and cannot be ignored. Extensive engineering practice shows that the freeze-thaw environment in high-altitude, deep-season frozen soil regions severely affects the stability of strongly weathered and completely weathered expansive soil slopes, thereby posing a threat to the safety of various geotechnical engineering projects constructed on them and seriously jeopardizing the performance of these projects.

[0003] Traditional methods for improving expansive soil include physical methods, chemical methods, microbial methods, and composite methods. Chemical methods involve adding lime, cement, fly ash, or chemical curing agents to the expansive soil, which inhibits its expansion and contraction through a chemical reaction. Adding chemical curing agents to expansive soil can fundamentally solve its hazardous characteristics such as wet expansion and dry shrinkage.

[0004] The frost heave and deformation properties of expansive soil in its natural state are related to boundary conditions such as soil composition, moisture, and temperature. However, seasonally frozen soil areas cover approximately 53.5% of my country's land area, distributed across Heilongjiang, Jilin, Liaoning, Inner Mongolia, Gansu, Ningxia, Qinghai, and northern Xinjiang, covering a wide range. The composition, moisture content, and temperature of soils vary significantly across different regions. Existing technologies for modifying expansive soil properties using chemical admixtures primarily focus on soils in specific regions, adding admixtures and modifying these particular soil types. However, the temporal and spatial distribution differences of expansive soil result in significant variations in its inherent properties, making modification schemes tailored to a single soil type lack universal applicability. Summary of the Invention

[0005] To address the problems existing in the prior art, this invention proposes corresponding improvement schemes and predicts the improvement performance based on the clay mineral composition of expansive soil. The clay mineral composition of expansive soil is a crucial indicator determining soil properties, with kaolinite, montmorillonite, and illite being representative minerals. These three minerals have different molecular structures, resulting in significant differences in their frost heave and deformation characteristics. A single admixture is insufficient for simultaneously and efficiently modifying all three clay minerals. This invention uses molecular dynamics simulations to study the interfacial adsorption and adsorption water film characteristics between the three clay minerals (kaolinite, montmorillonite, and illite) and different admixtures under different moisture contents. This reveals the modification mechanism of admixture molecules on different minerals, identifies the admixtures with the best improvement effect on each of the three clay minerals (kaolinite, montmorillonite, and illite) under corresponding working conditions, and specifically configures multiple admixtures based on the content of these three clay minerals in the expansive soil to improve its performance. This achieves optimized design of admixture ratios for expansive soils with different clay mineral contents. We constructed a coarse-grained model of real soil particles and corresponding admixtures, calculated various properties of the model (mechanical properties, frost heave, compression, and water film thickness), and used experimental and coarse-grained simulation results to build a database of improved soil properties. We then used machine learning to predict the improvement effect of admixtures on expansive soil under different working conditions.

[0006] To achieve the above objectives, the technical solution adopted by the present invention is as follows:

[0007] The first aspect of this invention provides a method for improving expansive soil based on its clay mineral composition, comprising the following steps:

[0008] (1) The optimal admixtures for clay minerals montmorillonite, illite and kaolinite in expansive soil under different moisture contents were obtained by molecular dynamics simulation.

[0009] (2) After determining the best admixture for the three clay minerals, add the corresponding admixture to the corresponding mineral, prepare soil test blocks for testing, and determine the optimal dosage based on the test results;

[0010] (3) Based on the moisture content and mineral composition of the expansive soil to be improved, admixtures are prepared to improve the properties of the expansive soil.

[0011] Preferably, step (1) includes montmorillonite modeling, illite modeling, kaolinite modeling, admixture molecular modeling, and admixture / clay mineral molecular dynamics modeling;

[0012] The montmorillonite modeling described above uses sodium-based montmorillonite as the crystal unit cell, with the molecular formula Na. 0.75 (Si 7.75 Al 0.25 (Al) 3.5 Mg 0.5 )O 20(OH)4·nH2O; the molecular formula of the illite model is K 1.5 ·nH2O(Si7Al)(Al 3.5 Fe 0.5 )O 20 (OH)4; the molecular formula of the kaolinite model is 2SiO2.Al2O3.2H2O;

[0013] The admixtures used in the molecular dynamics modeling of the admixtures include lignin, acrylates, polyurethanes, phenolic resins, epoxy resins, polyvinyl alcohol, and methylpropionic acid. The monomer model of the admixture is drawn using Materials Studio, and the polymerization is set to 10, that is, the monomer model is copied 10 times to obtain the calculation model of the admixture.

[0014] The molecular dynamics modeling process for the admixture / clay mineral includes: based on the clay mineral and admixture models, a two-layer stacking model of mineral / admixture molecules is constructed using Materials Studio, with admixture molecules placed on top and mineral molecules on the bottom. Molecular dynamics simulations are performed using LAMMPS to analyze the adsorption performance at the adsorption interface of clay minerals and different types of admixtures; the adsorption energy at the adsorption interface is calculated, the radial distribution function is used to characterize the bonding between atoms at the adsorption interface, the mean square displacement and time correlation function are used to analyze the stability properties at the adsorption interface, and the radius of gyration and atomic concentration distribution are used to characterize the contact area at the adsorption interface; based on the above adsorption parameters, the optimal admixture for the three clay minerals is determined.

[0015] Preferably, the admixtures include p-hydroxyphenyl lignin, syringyl lignin, and guaiacyl lignin; acrylates include sodium acrylate and butyl acrylate; polyurethanes include polyurethane; phenolic resins include polyamide-modified phenolic resin, dicyandiamide-modified phenolic resin, epoxy-modified phenolic resin, and polyvinyl alcohol acetal-modified phenolic resin; epoxy resins include bisphenol A type epoxy resin, bisphenol F type epoxy resin, polyphenol type glycidyl ether epoxy resin, aliphatic glycidyl ether epoxy resin, and glycidyl ester type epoxy resin; polyvinyl alcohols include polyvinyl alcohol; and methylpropionic acid admixtures include methyl methacrylate and methacrylic acid.

[0016] Preferably, the montmorillonite modeling process is as follows:

[0017] 1) Obtain the lattice parameters of the cell structure, including lattice size, angle, symmetry, shape, and space group;

[0018] 2) Adjust the interlayer spacing of the cells according to different water contents;

[0019] 3) Expanding the cell to a 4a×5b×1c system: Expanding the montmorillonite cell model by 4 times along the x-direction and 5 times along the y-direction;

[0020] 4) Isomorphic substitution: Adjacent atoms cannot be substituted simultaneously; in an octahedral sheet, every 8 Al atoms... 3+ One of them is Mg 2+ Replacement, every 32 Si in the tetrahedral sheet 4+ One of them was Al 3+ replace;

[0021] 5) Compensating ions: using Na + Na-based montmorillonite was established as a compensating ion;

[0022] 6) Add water molecules: Add different amounts of water molecules between the montmorillonite layers according to different water contents;

[0023] 7) Further cell expansion: Based on the size requirements of the simulated system, the montmorillonite model after isomorphic substitution is further expanded, and then molecular dynamics simulation is performed.

[0024] Preferably, during the illite modeling process, different numbers of water molecules are randomly filled into the interlayer domains of illite according to different water contents, and illite models with different water contents are established by Materials Studio software.

[0025] More preferably, the lattice parameters during the illite modeling process are: α = γ = 90°, β = 101.4°, and the interlayer spacing c varies with water content; the supercell consists of 128 unit cells arranged in an 8a × 4b × 4c pattern, and the dimensions of the expanded model in the x and y directions are respectively... and The side length c along the z-axis varies with water content, and the space group is P1.

[0026] Preferably, the kaolinite modeling is based on the modeling software Materials Studio, and the main modeling process is as follows:

[0027] 1) Obtain the kaolinite cell model, with the following lattice parameters: α=91.926°, β=105.046°, γ=89.797°;

[0028] 2) Supplement the missing H atoms in the kaolinite cell model;

[0029] 3) Expand the cell to an 8a×4b×2c system, so that the dimensions of the kaolinite model in the x and y directions are respectively

[0030] 4) Obtain a fully relaxed kaolinite model.

[0031] More preferably, the specific process of obtaining a fully relaxed kaolinite model is as follows: the simulation box is set to three-dimensional periodic boundary conditions, the van der Waals force is calculated using the Lennard-Jones potential function, and the cutoff radius is set to... Long-range electrostatic interactions were calculated using the Ewald method, with the cutoff radius set to... The energy minimization convergence tolerance is set to 10. - 6 kcal / mol; maximum iteration steps were set to 1000 steps; the kaolinite model was equilibrated sequentially in NVE (100ps), NVT (100ps), and NPT (100ps) systems, with the temperature set to 300K, the pressure set to 1.0atm, and the time step set to 1.0fs.

[0032] The second aspect of the present invention provides a method for predicting the performance of improved soil, wherein the performance of expansive soil improved by the method is predicted by constructing a coarse-grained molecular dynamics model.

[0033] Preferably, the construction of the coarse-grained molecular dynamics model includes:

[0034] 1) Use Materials Studio to construct single models of montmorillonite, kaolinite, illite, and quartz; by calculating the face-to-face and edge-to-edge interactions of the minerals in the single all-atom models of montmorillonite, kaolinite, illite, and quartz, obtain the functional relationship between the change of system free energy and the distance between clay mineral cores; coarse-grain the clay sheets into ellipsoidal particles, use the Gay-Berne potential energy formula for coarse-grained clay to obtain the coarse-grained potential energy parameters, and then construct coarse-grained models of the four minerals; based on XRD and TG tests, obtain the composition ratio of montmorillonite, kaolinite, illite, and quartz in the soil, determine the size of the four single molecules, and establish the four molecules in the same unit cell, that is, establish the coarse-grained soil model;

[0035] 2) Based on the content of kaolinite, montmorillonite and illite in the soil, determine the dosage ratio of the additives corresponding to the three minerals, and use Materials Studio to construct a calculation model for the three composite additives;

[0036] 3) Construct a two-layer adsorption model of coarse-grained soil / composite admixture, and then conduct relevant performance tests;

[0037] 4) Based on the data of improved soil obtained from nuclear magnetic resonance, triaxial, expansion rate tests and coarse-grained simulation, a database of improved soil performance is constructed, and machine learning is used to predict the performance of improved soil with different mineral ratios.

[0038] In this invention, the initial parameters of the quartz model are: α = 90°, β = 90°, γ = 120°. Using Materials Studio, a supercell operation was performed on the initial model, replicating it 8 times in the X direction, 7 times in the Y direction, and 4 times in the Z direction, resulting in... Quartz model.

[0039] More preferably, in step 3) of constructing the coarse-grained molecular dynamics model, different confining pressures are applied to the coarse-grained mineral model to obtain model test parameters, and then the actual parameters of the corresponding minerals are tested to verify the accuracy of coarse-grained parameterization and coarse-grained model, and then mechanical response simulation is performed.

[0040] Compared with the prior art, the beneficial effects of the present invention are:

[0041] This invention uses molecular dynamics simulations to determine the optimal admixtures for each of the main clay minerals (montmorillonite, kaolinite, and illite) in expansive soil under specific moisture content conditions, allowing for targeted admixture formulation to improve the properties of expansive soil in different regions. The admixture ratios obtained using this method for formulating expansive soil admixtures are universally applicable.

[0042] Based on the local mineral composition, moisture content, and temperature conditions of expansive soil, a coarse-grained soil model was constructed. Following the clay mineral ratio set during the soil coarse-grained model construction, a corresponding coarse-grained model of admixture molecules was developed. Molecular dynamics simulations were used to mimic the real soil geological environment and to construct realistic soil models for simulation. A database of improved soil properties was established by combining experimental and simulation data. Machine learning was then used to predict the modification effects of composite admixtures on expansive soils with different mineral contents and operating conditions. Attached Figure Description

[0043] Figure 1 Modeling a montmorillonite model;

[0044] Figure 2 Modeling a kaolinite model;

[0045] Figure 3 This is a schematic diagram of the coarse-graining process in an embodiment;

[0046] Figure 4 A schematic diagram of fitting the Gay-Berne formula for coarsening kaolinite;

[0047] Figure 5 To simulate the change in mass density of the coarse-grained montmorillonite model under different confining pressures;

[0048] Figure 6 For machine learning steps. Detailed Implementation

[0049] The specific embodiments of the present invention will be further described below. It should be noted that these descriptions are for the purpose of aiding understanding the present invention, but do not constitute a limitation thereof. Furthermore, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0050] Unless otherwise specified, the experimental methods used in the following embodiments are conventional methods, and the experimental materials used in the following embodiments are all available through conventional commercial channels.

[0051] Example 1

[0052] Molecular dynamics modeling of clay minerals

[0053] 1. Montmorillonite modeling

[0054] The montmorillonite model uses typical Wyoming sodium-based montmorillonite as the crystal unit cell, with the molecular formula Na. 0.75 (Si 7.75 Al 0.25 (Al) 3.5 Mg 0.5 )O 20 (OH)4·nH2O. The main modeling process is as follows:

[0055] (1) First, obtain the lattice parameters of the cell structure, including lattice size, angles, symmetry, shape, and space group. The cell of the constructed montmorillonite model is shown below. Figure 1 (a).

[0056] (2) Adjusting the interlayer spacing of the cells: The interlayer spacing of montmorillonite is adjusted according to its water content. The interlayer spacing of montmorillonite is initially... Adjusted to 9.6, 11.0, 12.5 and These correspond to systems with water contents of 0%, 10%, 20%, and 30% by weight, respectively. Figure 1 (b) is an improved cellular model of a hydrated montmorillonite system with a water content of 10%.

[0057] (3) Expanding the cell to a 4a×5b×1c system: The cellular model of montmorillonite is first expanded 4 times along the x direction and 5 times along the y direction in order to prepare for its subsequent isomorphic substitution.

[0058] (4) Isomorphic substitution (lattice substitution): mainly follows the following principles: (a) adjacent atoms cannot be substituted simultaneously; (b) every 8 Al atoms in an octahedral sheet 3+ One of them is Mg 2+ Replacement, every 32 Si in the tetrahedral sheet 4+ One of them was Al3+ The replacement sites are randomly distributed within the clay lamellae, see... Figure 1 (c)

[0059] (5) Compensating ions: Montmorillonite supercells (4a×5b×1c) generate interlayer negative charges due to isomorphic substitution, requiring the adsorption of external cations to form compensating ions to achieve charge balance. For example, inland alluvial soils with Ca 2+ Mainly, the clay in coastal areas contains a large amount of Na due to the influence of seawater. + This invention uses Na + Na-based montmorillonite was established as a compensating ion.

[0060] (6) Add water molecules: Add different amounts of water molecules between montmorillonite layers according to different water contents, where the water molecules can be randomly distributed between layers.

[0061] (7) Further cell expansion: Based on the size requirements of the simulated system, the montmorillonite model after isomorphic substitution is further expanded, and then molecular dynamics simulation is performed.

[0062] 2. Illite modeling

[0063] Illite is a potassium-rich, 2:1 type layered aluminosilicate mica-like clay mineral. Due to lattice substitution, illite carries a negative charge and can be divided into two types: octahedral illite with Al₂O₃ crystals. 3+ Mg 2+ Fe 2+ and Fe 3+ Plasma substitution, Si in tetrahedrons 4+ By Al + Substitution. Illite lattice parameters α = γ = 90°, β = 101.4°, and the interlayer spacing c varies with water content. The supercell consists of 128 unit cells arranged in an 8a × 4b × 4c configuration. The dimensions of the expanded model in the x and y directions are respectively... and The side length c along the z-axis varies with water content, the space group is P1, and in the model, every 16 Si 4+ There are 2 that were AI 3+ Replacement, every 8 Al 3+ There is 1 Fe 2+ Substitution results in a layer charge of -1.5 e / cell from lattice substitution, which is balanced by interlayer K ions. The established molecular formula of illite is: K 1.5 ·nH2O(Si7Al)(Al 3.5 Fe 0.5 )O 20(OH)4. Based on different water contents, different numbers of water molecules were randomly filled into the interlayer domains of illite, and four illite models with different water contents of 0%, 10%, 20%, and 30% were established using Materials Studio software.

[0064] 3. Kaolinite Modeling

[0065] Kaolinite modeling was performed using the molecular modeling software Materials Studio 8.0, with the molecular formula 2SiO2.Al2O3.2H2O. The main modeling process is as follows:

[0066] (1) Obtain the kaolinite cell model. The selected kaolinite crystal structure parameters and atomic coordinates were obtained from Bish's neutron diffraction experiment data at 1.5K, such as... Figure 2 As shown in (a). Their lattice parameters are respectively... α=91.926°, β=105.046°, γ=89.797°.

[0067] (2) Supplementing the missing H atoms in the kaolinite cell model. The positions of the missing H atoms were determined using VESTA software, and then the missing H atoms were supplemented using Materials Studio modeling software, such as... Figure 2 As shown in (b).

[0068] (3) Expand the cell to an 8a×4b×2c system, so that the dimensions of the kaolinite model in the x and y directions are respectively

[0069] (4) Obtain a fully relaxed kaolinite model. The simulation box is set to three-dimensional periodic boundary conditions, and the van der Waals force is calculated using the Lennard-Jones potential function. The cutoff radius is set to... Long-range electrostatic interactions were calculated using the Ewald method, with the cutoff radius set to... The energy minimization convergence tolerance is set to 10. -6 The kcal / mol pressure was set, and the maximum number of iterations was set to 1000. The kaolinite model was equilibrated sequentially under the NVE (100 ps), NVT (100 ps), and NPT (100 ps) ensembles, with the temperature set at 300 K, the pressure at 1.0 atm, and the time step at 1.0 fs. The optimized kaolinite model is shown below. Figure 2 As shown in (c).

[0070] Example 2

[0071] Molecular dynamics modeling of polymer admixtures

[0072] Common polymer admixtures include lignin-based C6H... 10 O5, acrylates CH2=CHCOOH, polyurethanes CH2-NH-C(=NH)-OH, phenolic resins NH-C(NH2)=NH-CH2-O, epoxy resins CH2-CH2-O, polyvinyl alcohols CH(OH)-CH2, methylpropionic acid CH2=C(CH3)-COOH, etc. The admixtures selected in this invention include lignins: p-hydroxyphenyl lignin, syringyl lignin, and guaiac lignin; acrylates: sodium acrylate and butyl acrylate; polyurethanes: polyurethane; phenolic resins: polyamide-modified phenolic resin, dicyandiamide-modified phenolic resin, epoxy-modified phenolic resin, and polyvinyl alcohol acetal-modified phenolic resin; epoxy resins: bisphenol A type epoxy resin, bisphenol F type epoxy resin, polyphenol type glycidyl ether epoxy resin, aliphatic glycidyl ether epoxy resin, and glycidyl ester type epoxy resin; polyvinyl alcohol: polyvinyl alcohol; and methylpropionic acid: methyl methacrylate and methacrylic acid. As modifiers for expansive soil, different polymeric admixtures have different molecular structures and polarities, and the surface structures of the three clay minerals are also different. Therefore, different admixtures have different modification effects on different minerals, and it is necessary to select the optimal admixture type for the corresponding clay mineral. Using Materials Studio, we drew the monomer model of the above polymer admixture. Then, we set the polymerization to 10, which means we copied the monomer model 10 times to obtain the calculation model of the admixture.

[0073] Example 3

[0074] 1. Molecular dynamics simulation of admixtures / clay minerals

[0075] Based on the aforementioned mineral and admixture models, a two-layer packing model of mineral / admixture molecules was constructed using Materials Studio, placing the admixture molecules on top and the mineral molecules at the bottom. Molecular dynamics simulations were performed using LAMMPS. Specific simulation parameters were as follows: CLAYFF was used to describe the mineral molecules, and CVFF force fields were used to describe the admixture molecules. Before the molecular dynamics simulation, the computational model was geometrically optimized to obtain the computational system at its minimum potential energy state. The simulation was conducted in an NVT ensemble, with the Nosé-Hoover method used to maintain the system temperature at 298 K. The simulation time step was 1.0 fs, and the total simulation time was 5000 ps. The system ran for a total of 5,000,000 steps, with one frame acquired every 5000 steps. After the system reached equilibrium, the motion trajectory of each frame was extracted to analyze the properties of the adsorption interface.

[0076] 2. Selection of the optimal admixture type for the three clay minerals

[0077] Based on prior molecular dynamics simulations, the adsorption performance at the adsorption interface of clay minerals and different types of admixtures was analyzed. The adsorption energy at the adsorption interface was calculated, the radial distribution function was used to characterize the bonding between atoms at the adsorption interface, the mean square displacement and time correlation function were used to analyze the stability properties at the adsorption interface, and the radius of gyration and atomic concentration distribution were used to characterize the contact area at the adsorption interface. Based on these adsorption parameters, the optimal admixture type for the three clay minerals was determined.

[0078] 3. Experimental study on the optimal admixture dosage for three clay minerals

[0079] After determining the optimal admixtures for the three minerals, soil specimens were prepared by adding the corresponding admixtures at concentrations of 6%, 9%, 12%, 15%, and 18% to the minerals for testing. Nuclear magnetic resonance (NMR) was used to scan the test samples, obtaining T2 relaxation curves and parameters such as pore size distribution, pore throat, and porosity. A vibrating triaxial testing machine with cold bath cooling was used to test typical mechanical properties of the soil (failure mode, compressive strength, elastic modulus, and shear strength). Free swelling rate and no-load swelling rate tests were conducted on the soil to obtain the frost heave and thaw settlement in frozen soil environments. Based on the above test results, the optimal dosage of the admixtures for the three minerals was determined.

[0080] 4. Prepare composite admixtures specifically based on the clay mineral content in expansive soil.

[0081] XRD and TG techniques were used to quantitatively analyze the composition and proportion of expansive soil. The main components of expansive soil are montmorillonite, kaolinite, illite, and quartz (sand). Sand has a relatively small impact on the deformation and mechanical properties of expansive soil; therefore, the influence of the three clay minerals was mainly considered. The optimal admixture for single minerals was analyzed experimentally. Based on the percentage content of the three minerals in the expansive soil, the required admixture dosage for each mineral was determined, and a composite admixture was prepared. For example, 9% polyvinyl alcohol (PVA) admixture is needed for montmorillonite alone, 12% polyurethane for kaolinite alone, and 6% acrylamide for illite alone. Tests showed that expansive soil in a certain region contained 30% montmorillonite, 16% kaolinite, and 5% illite. In this case, a composite admixture was prepared by mixing 30% × 9% PVA, 16% × 12% polyurethane, and 5% × 6% acrylamide by weight of the soil. This method allows for the development of targeted improvement solutions for expansive soil in a specific region.

[0082] Example 4

[0083] Based on the clay mineral content in expansive soil, a coarse-grained molecular dynamics model is constructed.

[0084] Molecular dynamics models of montmorillonite, kaolinite, illite, and different curing agents were constructed using the methods described in Examples 1 and 2. Quartz models were obtained from a US mineral crystal library, with an initial quartz model size of [size missing]. α = 90°, β = 90°, γ = 120°. Using Materials Studio, a supercell operation was performed on the initial model, replicating it 8 times in the X direction, 7 times in the Y direction, and 4 times in the Z direction, resulting in... Quartz model.

[0085] Because expansive soil mainly contains montmorillonite, kaolinite, illite, and quartz, after constructing individual models of these four components, a coarse-graining process is performed to obtain a realistic soil model containing all four components. The specific coarse-graining steps are as follows: Figure 3 As shown:

[0086] Coarsening involves mixing multiple individual models and requires a potential energy applicable to all models to construct the hybrid model. The process of obtaining the appropriate potential energy is the parameterization process.

[0087] Parameterization steps: By calculating the face-to-face and edge-to-edge interactions of minerals in a single all-atom model, the functional relationship between the system's free energy change and the distance to the clay mineral core is obtained. The clay sheets are coarsened into ellipsoidal particles using the Gay-Berne potential energy formula commonly used for clay coarsening.

[0088]

[0089] Where ε is the potential well depth, σ is the minimum effective particle radius, and h 12 η is the shortest distance between two particles, r is the distance between the particle's center of mass, and η is the distance between the particle's centers of mass. 12 and χ 12 These are two potential energy functions related to particle orientation. Surface-to-surface and edge-to-edge interactions were calculated for a single all-atom clay model, and fitted using the Gay-Berne formula. Example results are shown below. Figure 4 (Kaolinite coarsening parameters), that is, the coarsening potential energy parameters ε and σ. For different mineral mixture systems, the potential energy parameters follow the geometric mixing criterion, that is, the coarsening particles i and j of different types can be obtained through the following mixing criterion.

[0090]

[0091]

[0092] Coarse-grained modeling of mixed soil

[0093] After obtaining the coarsening potential energy parameters, coarsened mineral models of montmorillonite, kaolinite, illite, and quartz were established. The individual diameter of the coarsened mineral particles can reach several hundred nanometers. By aggregating mineral particles of the same size, a micrometer-scale mineral model was constructed. The compositional proportions of montmorillonite, kaolinite, illite, and quartz in the expansive soil were obtained based on XRD and TG tests. The coarsened molecules of the four minerals were then incorporated into the same unit cell to obtain a coarsened soil model. Based on the content of kaolinite, montmorillonite, and illite in the soil, the dosage ratio of the corresponding admixtures for the three minerals was determined, and computational models for the three composite admixtures were constructed using Materials Studio.

[0094] By applying different confining pressures to the coarse-grained mineral model, parameters such as equilibrium mass density and porosity were obtained. Figure 5 (Simulation results for coarse-grained montmorillonite) were compared with experimental test data. Elastic parameters, such as elastic modulus, shear modulus, bulk modulus, and Poisson's ratio, were tested using a selected density model. These parameters were then compared with experimental and simulation data to verify the accuracy of the coarse-grained parameterization and the coarse-grained model.

[0095] After the coarse-grained model is validated, mechanical response simulations can be performed. Uniaxial compression, simple shear, and load cycle simulations are conducted on single and mixed coarse-grained soil models to explore the mechanical response of the coarse-grained model. The results are compared with molecular-scale and macroscopic experimental results (NMR, triaxial, and expansion rate tests) to reveal the influence of mineral content on the mechanical properties of minerals and the mechanical response characteristics at different scales, thus establishing a database for subsequent machine learning.

[0096] Machine Learning

[0097] Neural networks were selected to process the mechanical and deformation properties of the improved soil obtained from the experimental section and coarse-grained molecular dynamics simulation, and a database of improved soil mechanical and deformation properties was established.

[0098] Based on novel swarm intelligence optimization algorithms and techniques such as ant colony, gray wolf, whale, and hunter-prey models, and utilizing an established database of improved soil mechanics and deformation characteristics, the optimal solutions for the mechanical properties of improved soil with different clay mineral ratios are obtained. Furthermore, predictive models for the mechanical and deformation characteristics of improved soil with different mineral proportions are constructed. Figure 6 ), to predict the properties of improved soil.

[0099] The embodiments of the present invention have been described in detail above, but the present invention is not limited to the described embodiments. For those skilled in the art, various changes, modifications, substitutions, and variations can be made to these embodiments without departing from the principles and spirit of the present invention, and these variations still fall within the protection scope of the present invention.

Claims

1. A method for improving expansive soil based on its clay mineral composition, characterized in that, Includes the following steps: (1) Modeling of montmorillonite, illite, kaolinite, admixture molecules, and admixture / clay minerals. The optimal admixture for each of the clay minerals montmorillonite, illite, and kaolinite in expansive soil under different moisture contents is obtained through molecular dynamics simulation. (2) After determining the optimal admixture for the three clay minerals, the corresponding admixture is added to the corresponding mineral, soil test blocks are prepared for testing, and the optimal dosage is determined based on the test results; (3) Based on the moisture content and mineral composition of the expansive soil to be improved, admixtures are prepared to improve the properties of the expansive soil.

2. The method for improving expansive soil based on the clay mineral composition according to claim 1, characterized in that, Step (1) includes montmorillonite modeling, illite modeling, kaolinite modeling, admixture molecular modeling, and admixture / clay mineral molecular dynamics modeling; The montmorillonite modeling described above uses sodium-based montmorillonite as the crystal unit cell, with the molecular formula Na. 0.75 (Si 7.75 Al 0.25 (Al) 3.5 Mg 0.5 )O 20 (OH)4·nH2O; the molecular formula of the illite model is K 1.5 •nH2O(Si7Al)(Al 3.5 Fe 0.5 )O 20 (OH)4; the molecular formula of the kaolinite model is 2SiO2.Al2O3.2H2O; The admixtures used in the molecular dynamics modeling of the admixtures include lignin, acrylates, polyurethanes, phenolic resins, epoxy resins, polyvinyl alcohol, and methylpropionic acid. The monomer model of the admixture is drawn using Materials Studio, and the polymerization is set to 10, that is, the monomer model is copied 10 times to obtain the calculation model of the admixture. The molecular dynamics modeling process for the admixture / clay mineral includes: based on the clay mineral and admixture models, a two-layer stacking model of mineral / admixture molecules is constructed using Materials Studio, with admixture molecules placed on top and mineral molecules on the bottom. Molecular dynamics simulations are performed using LAMMPS to analyze the adsorption performance at the adsorption interface of clay minerals and different types of admixtures; the adsorption energy at the adsorption interface is calculated, the radial distribution function is used to characterize the bonding between atoms at the adsorption interface, the mean square displacement and time correlation function are used to analyze the stability properties at the adsorption interface, and the radius of gyration and atomic concentration distribution are used to characterize the contact area at the adsorption interface; based on the above adsorption parameters, the optimal admixture for the three clay minerals is determined.

3. The method for improving expansive soil based on the clay mineral composition according to claim 2, characterized in that, The lignins include p-hydroxyphenyl lignin, syringyl lignin, and guaiacyl lignin; the acrylates include sodium acrylate and butyl acrylate; the polyurethanes include polyurethane; the phenolic resins include polyamide-modified phenolic resin, dicyandiamide-modified phenolic resin, epoxy-modified phenolic resin, and polyvinyl alcohol acetal-modified phenolic resin; the epoxy resins include bisphenol A type epoxy resin, bisphenol F type epoxy resin, polyphenol type glycidyl ether epoxy resin, aliphatic glycidyl ether epoxy resin, and glycidyl ester type epoxy resin; the polyvinyl alcohols include polyvinyl alcohol; and the methylpropionic acid compounds include methyl methacrylate and methacrylic acid.

4. The method for improving expansive soil based on the clay mineral composition according to claim 2, characterized in that, The montmorillonite modeling process is as follows: 1) Obtain the lattice parameters of the cell structure, including lattice size, angle, symmetry, shape, and space group; 2) Adjust the interlayer spacing of the cells according to different water contents; 3) Expanding the cell to a 4a × 5b × 1c system: Expanding the montmorillonite cell model by 4 times along the x-direction and 5 times along the y-direction; 4) Isomorphic substitution: Adjacent atoms cannot be substituted simultaneously; in an octahedral sheet, every 8 Al atoms... 3+ One of them is Mg 2+ Replacement, every 32 Si atoms in the tetrahedral sheet 4+ One of them was Al 3+ replace; 5) Compensating ions: using Na+ + Na-based montmorillonite was established as a compensating ion; 6) Add water molecules: Add different amounts of water molecules between the montmorillonite layers according to different water contents; 7) Further cell expansion: Based on the size requirements of the simulated system, the montmorillonite model after isomorphic substitution is further expanded, and then molecular dynamics simulation is performed.

5. The method for improving expansive soil based on the clay mineral composition according to claim 2, characterized in that, During the illite modeling process, different numbers of water molecules are randomly filled into the interlayer domains of illite according to different water contents, and illite models with different water contents are created using Materials Studio software.

6. The method for improving expansive soil based on the clay mineral composition according to claim 5, characterized in that, The lattice parameters in the illite modeling process are a = 5.21 Å, b = 9.02 Å, α = γ = 90°, β = 101.4°, and the interlayer spacing c varies with the water content. The supercell consists of 128 unit cells arranged in an 8a × 4b × 4c pattern. The expanded cell model has dimensions of 41.68 Å and 36.08 Å in the x and y directions, respectively, and the side length c in the z-axis direction varies with the water content. The space group is P1.

7. The method for improving expansive soil based on the clay mineral composition according to claim 2, characterized in that, The kaolinite modeling was based on the modeling software Materials Studio, and the modeling process is as follows: 1) Obtain the kaolinite cell model with lattice parameters a = 5.154 Å, b = 8.942 Å, c = 7.391 Å, α = 91.926°, β = 105.046°, and γ = 89.797°. 2) Supplement the missing H atoms in the kaolinite cell model; 3) Expand the cell to an 8a × 4b × 2c system, so that the kaolinite model is in The dimensions in the y-direction are 41.23 Å ​​and 35.77 Å, respectively; 4) Obtain a fully relaxed kaolinite model.

8. The method for improving expansive soil according to claim 7, characterized in that, The specific process for obtaining a fully relaxed kaolinite model is as follows: the simulation box is set to three-dimensional periodic boundary conditions; van der Waals forces are calculated using the Lennard-Jones potential function with a cutoff radius of 12.5 Å; long-range electrostatic interactions are calculated using the Ewald method with a cutoff radius of 8.5 Å; and the energy minimization convergence tolerance is set to 10. -6 kcal / mol; maximum iteration steps set to 1000 steps; kaolinite model sequentially in NVE(100 ), NVT(100) ), NPT (100 The system was balanced under the following conditions: temperature set to 300 K, pressure set to 1.

0. The time step is set to 1.

0. .

9. A method for predicting the properties of improved soil, characterized in that, The performance of expansive soil modified using the modification method described in any one of claims 1-8 is predicted by constructing a coarse-grained molecular dynamics model.

10. The method for predicting the performance of improved soil according to claim 9, characterized in that, The construction of the coarse-grained molecular dynamics model includes: 1) Use Materials Studio to construct single models of montmorillonite, kaolinite, illite, and quartz; by calculating the face-to-face and edge-to-edge interactions of the minerals in the single all-atom models of montmorillonite, kaolinite, illite, and quartz, obtain the functional relationship between the change of system free energy and the distance between clay mineral cores; coarse-grain the clay sheets into ellipsoidal particles, use the Gay-Berne potential energy formula for coarse-grained clay to obtain the coarse-grained potential energy parameters, and then construct coarse-grained models of the four minerals; based on XRD and TG tests, obtain the composition ratio of montmorillonite, kaolinite, illite, and quartz in the soil, determine the size of the four single molecules, and establish the four molecules in the same unit cell, that is, establish the coarse-grained soil model; 2) Based on the content of kaolinite, montmorillonite and illite in the soil, determine the dosage ratio of the additives corresponding to the three minerals, and use Materials Studio to construct a calculation model for the three composite additives; 3) Construct a two-layer adsorption model of coarse-grained soil / composite admixture, and then conduct relevant performance tests; 4) Based on the data of improved soil obtained from nuclear magnetic resonance, triaxial, expansion rate tests and coarse-graining simulation, a database of improved soil performance is constructed, and machine learning is used to predict the performance of improved soil with different mineral ratios.

Images