Geopolymer-based recycled fine aggregate shield synchronous grouting material and preparation method thereof

Abstract

The invention discloses a geopolymer-based recycled fine aggregate shield synchronous grouting material and a preparation method thereof. The grouting material is composed of a cementing material, an alkali activator, aggregate, a water reducing agent, a retarder and water, and the cementing material is a geopolymer precursor and is composed of fly ash, slag and metakaolin; the aggregate is composed of river sand and recycled fine aggregate. By optimizing the proportion and the preparation process, the material has the characteristics of excellent flowability, controllable setting time, high early strength, good impermeability, good durability and the like, and is suitable for synchronous grouting of shield tunnels in water-rich sand layers, high-water-pressure stratums or corrosion stratums. The invention also provides two optimization methods which are respectively based on a response surface method and a BP neural network combined genetic algorithm to realize multi-objective optimization of slurry performance. According to the material, dependence on traditional cement and gravel resources is remarkably reduced, energy consumption and carbon emission are reduced, high-added-value resource utilization of industrial solid waste and construction waste is achieved, and the material conforms to the development direction of green low-carbon building materials.

Description

Technical Field

[0001] This invention relates to the field of engineering materials technology, specifically to a geopolymer-based recycled fine aggregate shield tunneling synchronous grouting material and its preparation method. Background Technology

[0002] In shield tunnel construction, synchronous grouting plays a crucial role in controlling ground settlement and enhancing the overall integrity and long-term stability of the tunnel structure. Ideal synchronous grouting materials need to possess good flow properties, controllable setting time, high early and later strength, and excellent resistance to water erosion and durability to adapt to complex geological environments such as high water pressure and strong permeability. However, traditional cement-based synchronous grouting materials use cement as the main binder and river sand as aggregate. This not only results in high energy consumption and large carbon emissions during cement production, but also leads to the ecological damage caused by the large-scale mining of river sand, which does not meet the requirements of sustainable development. Furthermore, these materials have poor impermeability and corrosion resistance in water-rich strata, making them prone to grout loss and later leakage, thus failing to guarantee the long-term safety of the tunnel.

[0003] In recent years, the resource utilization of industrial solid waste has become a research hotspot in the field of building materials. If solid wastes such as fly ash and slag are not effectively utilized, they will occupy land and pollute the environment. Geopolymer-based cementitious materials, due to their advantages such as low carbon footprint, environmental friendliness, strong corrosion resistance, and rapid early strength development, are considered a potential substitute for some cement-based materials. Currently, geopolymer materials are mostly used in high-performance concrete and foundation reinforcement, while systematic research on synchronous grouting in tunnel boring machines is still relatively lacking. On the other hand, waste concrete, bricks, and tiles from construction waste can be processed into recycled aggregates to replace the increasingly scarce natural river sand, alleviating resource pressure and achieving waste resource utilization.

[0004] Currently, tunnel boring machine (TBM) synchronous grouting materials are mainly classified into three types: single-component inert grout, single-component rigid grout, and two-component grout. Single-component inert grout is mainly composed of inert components such as bentonite and fly ash, resulting in poor cementing performance. Traditional single-component rigid grout incorporates cement and other cementing materials, but still relies heavily on cement. Two-component grout typically consists of cement paste and water glass solution, which sets relatively quickly, but its workability control and long-term durability remain insufficient. Existing research largely focuses on optimizing these three types of grouts, failing to fundamentally reduce the dependence on cement and river sand.

[0005] Therefore, developing a new type of synchronous grouting material that uses geopolymers as cementing materials and recycled fine aggregates to replace natural river sand can not only significantly reduce cement consumption and carbon dioxide emissions, and realize the resource utilization of industrial solid waste and construction waste, but also improve the comprehensive performance of the grout, especially its impermeability and durability in strata rich in groundwater and highly karstified formations, providing a new technical approach for green, low-carbon, and high-performance shield tunnel synchronous grouting materials. Summary of the Invention

[0006] To improve the performance of traditional synchronous grouting materials, this invention provides a geopolymer-based recycled fine aggregate shield synchronous grouting material with high early strength, continued strength growth in the later stage, long setting time, large and adjustable initial fluidity, and good durability. This invention also provides two methods for precisely controlling the proportion of the grout according to the performance requirements of the grout, which have high precision, good reliability, and wide range of practical applications.

[0007] The above-mentioned objectives of the present invention are achieved through the following technical solutions:

[0008] A geopolymer-based recycled fine aggregate shield tunneling synchronous grouting material comprises: cementitious materials, aggregates, alkali activator, water-reducing agent, retarder, and water; wherein the mass ratio of alkali activator to water is 3:7, the water-reducing agent accounts for 1% of the mass of cementitious materials, the retarder accounts for 2% of the mass of cementitious materials, the water-cement ratio is 0.7, and the sand-cement ratio is 1.5.

[0009] The cementitious material is composed of a geopolymer precursor; the geopolymer precursor is composed of 56.43% fly ash, 26.63% slag and 16.94% metakaolin by mass percentage; or the geopolymer precursor is composed of 61.13% fly ash, 22.72% slag and 16.15% metakaolin by mass percentage.

[0010] When the geopolymer precursor is composed of 56.43% fly ash, 26.63% slag and 16.94% metakaolin by mass percentage, the corresponding aggregate is composed of 75% river sand and 25% recycled fine aggregate by mass percentage.

[0011] When the geopolymer precursor is composed of 61.13% fly ash, 22.72% slag and 16.15% metakaolin by mass percentage, the corresponding aggregate is composed of 61.02% river sand and 38.98% recycled fine aggregate by mass percentage.

[0012] Furthermore, the river sand is ordinary fine river sand with a fineness modulus of 1.9 and a mud content of 1.68%; the recycled fine aggregate is recycled concrete fine aggregate with a fineness modulus of 2.2 and a gradation belonging to the 2-gradation zone.

[0013] Furthermore, the fly ash is Class II, Type F fly ash, with a specific surface area ≥360m². 2 / kg; the slag is grade S95, with a specific surface area ≥429m². 2 / kg; the metakaolin content is industrial grade.

[0014] Furthermore, the alkali activator is anhydrous sodium silicate; the water-reducing agent is a naphthalene-based water-reducing agent; and the retarder is barium nitrate.

[0015] The present invention also provides a method for preparing the grouting material, comprising the following steps:

[0016] (1) Dry mixing: Weigh the fly ash, slag, metakaolin, river sand, recycled fine aggregate, anhydrous sodium silicate, water reducing agent and retarder according to the proportion, pour them into the mixing pot, and dry mix at low speed for 3 minutes.

[0017] (2) Wet mixing: Add the water weighed according to the water-cement ratio to the dry mixture at once, switch the mixing pot to the high speed and continue mixing for 4 minutes to obtain a uniform geopolymer-based recycled fine aggregate shield synchronous grouting material.

[0018] The present invention also provides an optimized method for preparing the grouting material, comprising: using metakaolin content A, recycled fine aggregate content B, and slag content C as independent variables; using fluidity Y1, bleeding rate Y2, stone rate Y3, and setting time Y4 as response variables; using the D-optimal module of Design-Expert software to construct a multi-objective expectation function, setting the upper limit, lower limit, and weight of each response to maximize the total expectation function D, and obtaining the optimized mix proportion: by mass percentage, metakaolin 16.94 wt%, recycled fine aggregate 25 wt%, and slag 26.63 wt%.

[0019] The present invention also provides another optimized method for preparing the grouting material, comprising: constructing a three-layer BP neural network with three input layer neurons X1, X2, and X3 for the proportions of metakaolin, recycled fine aggregate, and slag, and four output layer neurons Y1, Y2, Y3, and Y4 for the proportions of fluidity, bleeding rate, stone formation rate, and setting time; the number of input layer nodes is 3, the number of hidden layer nodes is 11, and the number of output layer nodes is 4; the experimental dataset is divided into training and testing sets at a ratio of 4:1, and the network is trained using RMSE, MAE, MAPE, and R² as evaluation indicators; the trained BP neural network is used as a predictor, and a genetic algorithm is employed to minimize the fitness function with the objectives of fluidity of 340 mm, bleeding rate of 0%, stone formation rate of 100%, and setting time of 180 min, to obtain the optimized mix proportion: by mass percentage, metakaolin 16.15 wt.%, recycled fine aggregate 38.98 wt.%, and slag 22.72 wt.%.

[0020] The grouting material of the present invention can be used in the synchronous grouting of shield tunnels in water-rich sand layers, high water pressure strata or karst strata, and is injected into the shield tail gap through the synchronous grouting system of the shield machine.

[0021] When used in water-rich sand layers or high water pressure formations, the geopolymer precursors used consist of 56.43% fly ash, 26.63% slag and 16.94% metakaolin by mass percentage, and the aggregates consist of 75% river sand and 25% recycled fine aggregates by mass percentage.

[0022] When used in karst formations, the geopolymer precursors consist of 61.13% fly ash, 22.72% slag, and 16.15% metakaolin by mass percentage, while the aggregates consist of 61.02% river sand and 38.98% recycled fine aggregates by mass percentage.

[0023] Technical principle of the invention:

[0024] Alkaline activators dissolve and activate aluminosilicate raw materials such as fly ash, slag, and metakaolin, promoting depolymerization-condensation reactions of active components to form a cementing system dominated by a three-dimensional network gel of aluminosilicates, thereby achieving efficient curing and performance enhancement. Fly ash is rich in active SiO2 and Al2O3, which can rapidly dissolve and release aluminosilicate species in an alkaline environment, participating in gel formation, improving the rheological properties of the slurry, and promoting later strength development. Slag has a high CaO content, and the dissolved Ca²⁺ can promote the formation of zeolite-like phases and CASH gel, enhancing the compactness and mechanical stability of the microstructure. Metakaolin has high reactivity and can effectively refine the pore structure, improving the material's impermeability and durability. Although the appropriate amount of recycled fine aggregate introduced into the material introduces porosity to some extent, by optimizing the cementing components and controlling the activator, the overall performance balance and resource utilization goals can still be achieved.

[0025] In the early stages of the slurry reaction, the alkaline activator rapidly breaks down the chemical bonds of Si-O, Al-O, and Ca-O in the raw materials, causing the silicon-aluminum phase to dissociate into silicon-oxygen tetrahedra and aluminum-oxygen tetrahedra monomers. As the ion concentration increases, preliminary polymerization occurs between monomers and with Na⁺, OH⁻, and other substances in the system, forming low-polymerization-degree aluminosilicate intermediates. At this stage, the slurry exhibits good fluidity and high stability, making it suitable for pumping and filling the shield tail gap.

[0026] During the mid-reaction phase, the oligomers undergo further condensation reactions, generating amorphous NASH and CASH gels. These gels gradually cover the surfaces of unreacted particles and recycled aggregates, forming a three-dimensional network structure through interlocking. Ca²⁺ in the slag can partially replace Na⁺ in NASH, forming a more stable CASH interwoven structure; aluminum oxide species can also enter the CSH chain, further enhancing gel stability and density. Multiple gels together constitute a composite (N,C)-(A)-SH system, significantly optimizing the slurry's setting behavior and early strength development.

[0027] In the later stages of the reaction, the slurry gradually hardens to form a stone-like mass. As curing progresses, the gel phase continues to dehydrate and compact, filling internal pores, improving the pore structure, and resulting in a continuous increase in compressive strength and significantly enhanced durability. The introduction of recycled fine aggregate needs to be combined with the synergistic effect of slag and metakaolin to optimize pore distribution, thereby achieving the comprehensive advantages of the material in terms of both mechanical properties and ecological benefits.

[0028] The beneficial effects of this invention are as follows:

[0029] 1. By completely replacing cement with industrial solid wastes such as fly ash, slag, and metakaolin as cementitious components, and replacing natural river sand with recycled fine aggregates from construction waste, the dependence on traditional cement and sand resources is significantly reduced, energy consumption and carbon emissions in cement production are greatly reduced, and the high-value-added resource utilization of industrial solid waste and construction waste is effectively promoted, which is in line with the development direction of green and low-carbon building materials.

[0030] 2. The application characteristics of geopolymer-based cementitious systems and recycled fine aggregates in synchronous grouting were systematically studied. By optimizing the mix design, the grout met the requirements of shield tunneling projects in terms of fluidity, setting time and water dispersibility, providing a technical basis for the large-scale application of geopolymer-based recycled aggregate grout in complex strata.

[0031] 3. Under the synergistic effect of geopolymer gel and recycled fine aggregate, the grout forms a dense and stable composite gel structure, which significantly improves the early strength, long-term impermeability and corrosion resistance of the grout body. It is especially suitable for harsh working conditions such as high water pressure and water-rich strata. While ensuring the grouting quality and the long-term stability of the tunnel, it has significant economic and environmental benefits. Attached Figure Description

[0032] Figure 1 This diagram illustrates the effects of slag content on slurry fluidity, bleeding rate, stone formation rate, setting time, compressive strength of the stone body, and the 28-day water-to-land strength ratio.

[0033] Figure 2 This diagram illustrates the effects of metakaolin content on slurry fluidity, bleeding rate, stone formation rate, setting time, compressive strength of the stone body, and the 28-day water-to-land strength ratio.

[0034] Figure 3 A schematic diagram showing the effect of recycled fine aggregate content on slurry fluidity, bleeding rate, stone formation rate, setting time, compressive strength of the stone body, and 28-day water-to-land strength ratio.

[0035] Figure 4 XRD patterns (28 days) for different slag contents.

[0036] Figure 5 XRD patterns (28 days) of different metakaolin content.

[0037] Figure 6 Design a three-factor experimental model.

[0038] Figure 7 Comparison of predicted values ​​from mathematical regression models and experimentally measured values ​​for various properties of the slurry: Figure 7 (a) shows the predicted and measured values ​​of fluidity; (b) shows the predicted and measured values ​​of bleeding rate; (c) shows the predicted and measured values ​​of stone formation rate; (d) shows the predicted and measured values ​​of setting time; (e) shows the predicted and measured values ​​of 3-day compressive strength; and (f) shows the predicted and measured values ​​of 28-day compressive strength.

[0039] Figure 8 For different numbers of hidden layer nodes, the output layer neurons R 2 And the impact of RMSE: Figure 8 In the middle (a), the RMSE and R of the output flowability are shown. 2 change; Figure 8 (b) shows the RMSE and R of the output permeability. 2 change; Figure 8 (c) represents the RMSE and R of the output stone rate. 2 change; Figure 8 (d) represents the RMSE and R of the output condensation time. 2 change.

[0040] Figure 9 The impact of the number of hidden layer nodes on the evaluation metrics of output layer neurons: Figure 9 In the middle (a), R has 4 outputs. 2 contrast; Figure 9 (b) shows a comparison of the RMSE of the four outputs; Figure 9 (c) shows a comparison of the MAE of the four outputs; Figure 9 The middle (d) shows a comparison of the four outputs of MAPE.

[0041] Figure 10 This is the BP model structure.

[0042] Figure 11 This is the training result of the BP model.

[0043] Figure 12 Compare the actual and predicted values ​​of the four outputs of the BP model training set.

[0044] Figure 13 Compare the actual and predicted values ​​of the four outputs of the BP model test set.

[0045] Figure 14 This is the output of the genetic algorithm. Detailed Implementation

[0046] The technical solution of the present invention will be further described in detail below with reference to the embodiments.

[0047] The raw materials and dosages used in Experimental Groups 1-13 and Examples 1 and 2 are as follows:

[0048] Fly ash: Class II, F type fly ash, specific surface area ≥360m² 2 / kg;

[0049] Slag: S95 grade slag, with a specific surface area of ​​429 m². 2 / kg, manufactured by Gongyi Longze Water Purification Materials Co., Ltd.

[0050] Metakaolin: Industrial grade;

[0051] River sand: Ordinary fine river sand, fineness modulus 1.9, mud content 1.68%;

[0052] Recycled fine aggregate: Recycled concrete fine aggregate with a fineness modulus of 2.2 and a gradation category of 2.

[0053] Alkali activator: anhydrous sodium silicate;

[0054] Water-reducing agent: Naphthalene-based water-reducing agent, manufactured by Shandong Yousuo Chemical Technology Co., Ltd.

[0055] Retarder: Barium nitrate.

[0056] The components of the geopolymer-based recycled fine aggregate shield tunneling synchronous grouting material include: cementitious materials, aggregates, alkali activator, water-reducing agent, retarder, and water. Specifically, the mass ratio of alkali activator to water is 3:7, the water-reducing agent accounts for 1% of the mass of the cementitious materials, the retarder accounts for 2% of the mass of the cementitious materials, the water-cement ratio is 0.7, and the sand-cement ratio is 1.5.

[0057] The chemical composition of the three industrial solid wastes, fly ash, slag, and metakaolin, is shown in Table 1. The basic parameters of river sand and recycled fine aggregate are shown in Tables 2 and 3. The composition and dosage of cementitious materials, aggregates, and alkali activators are shown in Table 4.

[0058] Table 1. Chemical composition content (%) of three types of industrial solid waste

[0059]

[0060] Table 2 Basic parameters of river sand

[0061]

[0062] Table 3 Basic parameters of recycled fine aggregate

[0063]

[0064] Table 4. Cementitious materials and aggregate ratios used in experimental groups 1-13

[0065]

[0066] Experimental groups 1-13 and Examples 1 and 2 all used the following method to prepare geopolymer-based recycled fine aggregate shield tunneling synchronous grouting materials:

[0067] (1) Dry mixing: Weigh the fly ash, slag, metakaolin, river sand, recycled fine aggregate, anhydrous sodium silicate, water reducing agent and retarder according to the proportion, pour them into the mixing pot, and dry mix at low speed for 3 minutes.

[0068] (2) Wet mixing: Add the water weighed according to the water-cement ratio to the dry mixture at once, switch the mixing pot to the high speed and continue mixing for 4 minutes to obtain a uniform geopolymer-based recycled fine aggregate shield synchronous grouting material.

[0069] The difference between Experimental Group 5 and Experimental Group 1 is that the fly ash content is 60% and the slag content is 40%, while the remaining steps and dosages are the same as those in Experimental Group 1.

[0070] The difference between Experimental Group 6 and Experimental Group 1 is that the fly ash content is 75%, the slag content is 20%, and the metakaolin content is 5%, while the other steps and dosages are the same as those in Experimental Group 1.

[0071] The difference between Experimental Group 7 and Experimental Group 4 is that the slag content is 20% and the metakaolin content is 10%, while the other steps and dosages are the same as those in Experimental Group 4.

[0072] The difference between Experimental Group 8 and Experimental Group 4 is that the fly ash content is 65%, the slag content is 20%, and the metakaolin content is 15%, while the remaining steps and dosages are the same as those in Experimental Group 4.

[0073] The difference between experimental group 9 and experimental group 8 is that the fly ash content is 60%, the slag content is 20%, and the metakaolin content is 20%, while the other steps and dosages are the same as those in experimental group 8.

[0074] The difference between Experimental Group 10 and Experimental Group 7 is that the amount of river sand is 75% and the amount of recycled fine aggregate is 25%, while the other steps and dosages are the same as those in Experimental Group 7.

[0075] The difference between Experimental Group 11 and Experimental Group 7 is that the amount of river sand and recycled fine aggregate is 50%, while the rest of the steps and dosages are the same as those in Experimental Group 7.

[0076] The difference between Experimental Group 12 and Experimental Group 7 is that the amount of river sand added is 25% and the amount of recycled fine aggregate added is 75%, while the remaining steps and amounts are the same as those in Experimental Group 7.

[0077] The difference between Experimental Group 13 and Experimental Group 7 is that the amount of river sand added is 0%, the amount of recycled fine aggregate added is 100%, and the rest of the steps and dosages are the same as those in Experimental Group 7.

[0078] Performance testing:

[0079] (a) Detection methods / test methods:

[0080] The material was prepared according to the preparation method of the geopolymer-based recycled fine aggregate shield synchronous grouting material prepared in Examples 1-13, and then tested according to the following testing methods. The test results are shown in Table 5.

[0081] Flowability: Refer to GB / T 50448-2015 "Technical Specification for Application of Cement-based Grouting Materials".

[0082] Bleeding rate: Refer to T / CECS 563-2018 "Technical Specification for Application of Synchronous Grouting Materials in Shield Tunnel".

[0083] Stone settling rate: Refer to T / CECS 563-2018 "Technical Specification for Application of Synchronous Grouting Materials in Shield Tunnel".

[0084] Setting time: Refer to GB / T 1346-2011 "Standard consistency water requirement, setting time and soundness test method for cement".

[0085] Compressive strength: Referring to GB / T 17671-2021 "Test Method for Strength of Cement Mortar (ISO)", the well-mixed grout was poured into a 40mm×40mm×160mm triple mold, then manually tamped for 2 minutes to remove air from the grout. The surface was then smoothed with a scraper and finally covered and sealed with plastic wrap. The sealed triple mold was placed in a curing chamber for 24 hours at an ambient temperature of 20±2℃ and a relative humidity of 95%. The demolded specimens were then cured until the appropriate test age. The compressive strength of the specimens was tested using a YTE-3000 flexural and compressive strength testing machine. The maximum load was 300KN, the loading rate was 2.4KN / s, and the failure constant was 2KN.

[0086] Water dispersibility: Water dispersibility tests were conducted in accordance with DL / T 5117-2000 "Test Procedure for Underwater Non-Dispersible Concrete", T / CECS 563-2018 "Technical Specification for Application of Synchronous Grouting Materials in Shield Tunneling", and the standards set by Hong Qiaoheng and Zhu Xianfa. First, a 40mm×40mm×160mm triple mold was placed in a water tank filled with water, with the water level 20mm above the mold surface. Then, the mixed grout was poured into the mold through a funnel. Pouring was stopped when the grout overflowed from the mold. The mold was then removed from the water tank and left to stand for 5-10 minutes to allow the water attached to the grout to flow out freely. The mold was then vibrated, the surface was leveled, and covered with plastic wrap. Finally, the mold was transferred to a standard environment with a temperature of 20±2℃ and a relative humidity of 95% for curing. After demolding for 24 hours, the slurry was cured under standard conditions for 28 days. The water-to-land strength ratio (water strength / land strength) was used as a quantitative indicator to evaluate the water dispersibility of the slurry.

[0087] The main performance requirements for synchronous grouting fluid in shield tunnels are shown in Table 6.

[0088] (II) Test Results:

[0089] Table 5. Detection results of experimental groups 1-13

[0090]

[0091] Table 6 Performance Requirements of Grout for Synchronous Grouting in Shield Tunnels

[0092]

[0093] Due to the slow reaction of the fly ash-based polymer slurry, the setting and hardening time of experimental group 1 was too long, the 3-day compressive strength was too low to be tested, and the 28-day compressive strength was only 3.2 MPa.

[0094] The experimental results show that increasing the slag content significantly reduces the fluidity of the grout. When the slag content increases from 0 wt.% to 40 wt.%, the fluidity decreases from 389 mm to 336 mm, a reduction of 13.6%, indicating that the addition of slag significantly deteriorates the rheological properties of the grout, which is detrimental to pumping construction. This phenomenon is closely related to the morphological characteristics of slag particles and their role in the reaction process. Simultaneously, with the increase of slag content, the bleeding rate decreases from 3.27% to 1.03%, while the stone formation rate increases from 96.43% to 98.70%, indicating that slag helps improve the water retention and volume stability of the grout, which is beneficial for the compact formation of the grout body. Regarding setting time, slag exhibits a significant accelerating effect; when the slag content is 0 wt.%, the grout setting time exceeds 24 hours, significantly exceeding the requirements of actual engineering projects. When the slag content increased from 10 wt.% to 40 wt.%, the setting time was shortened from 454 min to 76 min, a reduction of 83.3%, effectively ensuring the efficiency of the grouting process. Furthermore, slag plays a positive role in the development of compressive strength; the 3-day and 28-day compressive strengths significantly increased with increasing slag content, especially at 40 wt.%, where the early strength increase was more pronounced. The resistance to water dispersion also improved with increasing slag content, and the water-to-land strength ratio continuously increased, indicating that it helps improve the stability of the grout in an aquatic environment.

[0095] The incorporation of metakaolin also significantly affected the slurry properties. When its dosage increased from 0 wt.% to 20 wt.%, the fluidity decreased from 372 mm to 357 mm. Although the decrease was only 4.0%, it still indicated a certain negative impact on workability. The bleeding rate decreased from 2.48% to 1.66%, and the stone formation rate increased from 97.79% to 98.35%, indicating that metakaolin helps improve the water retention of the slurry and increase the solids formation rate. Regarding setting time, it slightly increased with a dosage of 0–5 wt.%, but shortened with further increases in dosage, indicating a significant dosage-dependent effect. As for compressive strength, the 3-day strength gradually increased with incorporation, while the 28-day strength reached its peak at 15 wt.% and then slightly decreased, indicating that an appropriate amount of metakaolin contributes to long-term strength development, but excessive amounts may cause adverse effects. Its effect on water dispersibility is quite complex. The water-to-land strength ratio reaches its lowest value at 10 wt.%, and then rises, indicating that the condensation time is a key factor affecting this property.

[0096] The incorporation of recycled fine aggregate significantly reduced the flowability of the slurry. When the replacement rate increased from 0 wt.% to 100 wt.%, the flowability decreased from 363 mm to 290 mm, a reduction of 20.1%, which had a significant adverse impact on pumping performance. Simultaneously, the bleeding rate increased from 1.92% to 3.53%, and the aggregate content decreased from 98.16% to 96.47%, indicating that recycled fine aggregate reduced the slurry's water retention capacity, decreased its volume stability, and made it more prone to segregation. Recycled fine aggregate also significantly shortened the slurry's setting time. The setting time decreased significantly within the 0–50 wt.% dosage range, but the change slowed down at higher dosages. Regarding compressive strength, the 3-day strength increased with increasing recycled fine aggregate dosage, while the 28-day strength rebounded after reaching a trough at 50 wt.%, reflecting its dual role of alkali-activated promotion and interfacial weakening. The water dispersion resistance continuously improves with the increase of recycled fine aggregate content, and the water-to-land strength ratio continuously increases, indicating that it helps to enhance the molding stability of the slurry in water.

[0097] Example 1:

[0098] Design a three-factor, three-level experiment based on response surface methodology, such as Figure 6 As shown. The water dispersibility basically meets the performance index requirements. Therefore, it is only necessary to study the fluidity, bleeding rate, stone rate, setting time, 3-day compressive strength, and 28-day compressive strength. These six properties are used as target responses Y1, Y2, Y3, Y4, Y5, and Y6, with the three factors of metakaolin content (A), recycled fine aggregate content (B), and slag content (C) as independent variables. The significance of the independent variables' influence on these six responses is the focus. The relationship between each factor and the various properties of the slurry is established through a second-order multivariate nonlinear regression equation. The slurry performance is then accurately predicted according to the following regression equation:

[0099] (1)

[0100] In the formula, Y represents the response of each characteristic, β0 is a constant term, and β1, β2, and β3 are linear regression coefficients. 12 β 13 and β 23 β represents the regression coefficients under the interaction. 11 β 22 and β 33The coefficients are second-order regression coefficients, with A, B, and C being first-order terms, AB, AC, and BC being interaction terms, and A2, B2, and C2 being second-order terms. Based on the results of experimental groups 1-13 above, the metakaolin content was set at 15 wt.%, 17.5 wt.%, and 20 wt.% by mass percentage; the recycled fine aggregate content was set at 25 wt.%, 37.5 wt.%, and 50 wt.%; and the slag content was set at 20 wt.%, 25 wt.%, and 30 wt.%. The three dosages for each factor correspond to the code values ​​-1, 0, and 1, representing low, medium, and high levels of dosage, respectively. The three factors, their code values, and the slurry raw material ratios are shown in Tables 7 and 8.

[0101] Table 7. Three factors and their coded values:

[0102]

[0103] Table 8. Slurry raw material ratio:

[0104]

[0105] The slurry described in Example 1 was prepared according to the preparation methods of Experimental Groups 1-13, and the fluidity, bleeding rate, stone rate, setting time, compressive strength and water dispersibility were tested according to the performance testing methods of Experimental Groups 1-13. The test results are shown in Table 9 below.

[0106] Table 9. Experimental Results:

[0107]

[0108] (Continued from Table 9)

[0109]

[0110] Based on the above experimental results, mathematical regression models for various properties of the slurry were obtained, as shown below:

[0111] (4)

[0112] (5)

[0113] (6)

[0114] (7)

[0115] (8)

[0116] (9)

[0117] Analysis of variance (ANOVA) can quantitatively describe the significance of each factor on each response in the regression model. The significance of the established model and response is typically determined using the F-test, with the p-value used to evaluate hypothesis testing. A p-value > 0.05 indicates the model is not significant, while a p-value ≤ 0.05 indicates it is significant, with smaller p-values ​​indicating higher significance. Furthermore, the coefficient of determination (R²) of the model... 2 The coefficient of determination (R²) can evaluate the degree of fit between the experimental results and the fitted model. The closer the value is to 1, the more reliable the fitted model is. 2 ) and predictive determination coefficient (Pre .R) 2 A higher SNR (signal-to-noise ratio) and a difference of less than 0.2 indicate good interpretability of the model. The SNR reflects the model's ability to resist interference; a higher SNR indicates higher prediction accuracy and greater reliability. Generally, a SNR greater than 4 is considered a usable model. Model analysis is shown in Tables 10 and 11.

[0118] Table 10. Analysis of variance of mathematical regression models for various properties of slurry:

[0119]

[0120] Table 11. Fitting accuracy analysis of mathematical regression models for various properties of slurry:

[0121]

[0122] As shown in Table 10, the analysis of variance reveals that the p-value for each theoretical regression model is on the order of 10. -4 The results indicate that the models for the six responses are highly significant. Furthermore, the p-values ​​of the lack-fit terms for all six response models are greater than 0.05, indicating that the theoretical regression models are not significantly misfitted for pure error, and that the interference from factors other than the content of kaolinite, recycled fine aggregate, and slag is minimal. Table 11 shows that the R-value for each response... 2 Both are close to 1, Adj.R 2 and Pre.R 2 The differences were all within 0.2, and the signal-to-noise ratios were all greater than 10, indicating that the six theoretical regression models can well describe the slurry properties and have high accuracy in predicting the slurry performance.

[0123] To further evaluate the accuracy of the mathematical regression model, such as Figure 7 As shown, a scatter plot is created comparing the measured and predicted values ​​of fluidity, bleeding rate, stone rate, setting time, and compressive strength at different ages. From... Figure 7As can be clearly seen in the figure, all points are distributed near the straight line y = x, indicating that the model's predicted values ​​match the measured values ​​well, the experimental error is small, and the model accuracy is high. In summary, the regression models for fitting various grout properties established based on response surface methodology all exhibit high accuracy and reliability. They can be used to analyze and predict the pumping performance, stability, filling performance, and mechanical properties of synchronous grouting grout, so as to achieve high levels of various grout properties and thus obtain the model proportions of grout materials.

[0124] The results of single-factor and multi-factor composite experiments show that the 3-day compressive strength, 28-day compressive strength, and water dispersibility of the geopolymer-based recycled fine aggregate synchronous grouting material all meet the requirements of actual shield tunneling construction and are at a relatively good level. However, the fluidity and setting time are difficult to meet the performance indicators required for construction, so optimization standards need to be set for them. Meanwhile, the bleeding rate and stone formation rate are important indicators characterizing the stability and filling performance of the grout, especially in complex strata where the quality of the bleeding rate and stone formation determines whether the grout can effectively fill the shield tail voids. Therefore, fluidity, bleeding rate, stone formation rate, and setting time were selected as four performance targets for optimization.

[0125] The characteristics of the grout prepared by the synchronous grouting material ratio obtained from the mathematical regression model are compared with those of the traditional grout, as shown in Table 12.

[0126] Table 12 Comparison of various properties of slurry between the model group and the traditional group:

[0127]

[0128] The performance of the D-optimal group, developed based on the optimization results, was compared with that of the control group. Compared to the control group, the D-optimal group showed a 42.8% increase in fluidity, a 62.4% decrease in bleeding rate, a 0.2% increase in stone rate, and a significant 65.8% reduction in setting time. All four properties—fluidity, bleeding rate, stone rate, and setting time—were at high levels, demonstrating significant improvements and a very clear optimization effect. Furthermore, compared to the control group's 3-day compressive strength (0.77 MPa) and 28-day compressive strength (3.39 MPa), the D-optimal group showed an 809.1% increase in 3-day compressive strength and a 385.0% increase in 28-day compressive strength, with significant improvements in the mechanical properties of the stone mass. In conclusion, the geopolymer-based recycled fine aggregate synchronous grouting material prepared in this invention not only possesses excellent pumping performance, stability, filling performance, and mechanical properties, but is also more economical and environmentally friendly, demonstrating significant engineering application value.

[0129] Using the D-optimal method in Example 1 above, the dosage of the three factors corresponding to the four responses being simultaneously optimized to a higher level was calculated: metakaolin = 16.94 wt.%, recycled fine aggregate = 25 wt.%, and slag = 26.63 wt.%.

[0130] Example 2: Based on the experimental results from the response surface methodology, a backpropagation (BP) model was trained and constructed, and further optimized using a genetic algorithm to ultimately achieve multi-objective optimization of slurry performance. Flowability, bleeding rate, stone formation rate, and setting time were used as output variables Y1, Y2, Y3, and Y4, respectively. The three factors—metakaolin content (A), recycled fine aggregate content (B), and slag content (C)—were used as input variables. The 17 sets of experimental data in Table 9 were used as the sample set. The specific steps are as follows:

[0131] (1) Construction of BP neural network model:

[0132] The network employs a three-layer feedforward structure (input layer - hidden layer - output layer), and the optimal number of hidden layer nodes is determined through trial and error. Model training utilizes the backpropagation algorithm, and error analysis during training and testing employs root mean square error (RMSE), mean absolute error (MAE), mean relative error (MAPE), and the coefficient of determination for goodness of fit (R²). 2 A comprehensive evaluation was conducted. Each set of hyperparameters underwent 10 parallel trials, and the average of the 10 trials was taken as the final result.

[0133] Root Mean Square Error (RMSE): (10)

[0134] Mean Absolute Error (MAE): (11)

[0135] Mean relative error (MAPE): (12)

[0136] goodness-of-fit coefficient (R²) 2 ): (13)

[0137] In the formula, This is a predicted value; These are measured values; The average of all measured values; Number of groups. RMSE, MAE, and R 2 RMSE, MAE, and MAPE are used to represent the degree of difference between predicted and measured values, and are the most important indicators for measuring the accuracy of the BP model. MAPE represents the mean absolute value of the relative error between predicted and measured values, and is a statistical indicator for evaluating the accuracy of the BP model. The smaller the values ​​of RMSE, MAE, and MAPE, the higher the accuracy of the BP model. 2The closer the value is to 1, the better the training effect and prediction performance of the BP model.

[0138] The number of hidden layer neurons can be determined by trial and error using empirical formulas:

[0139] (2)

[0140] In the formula, represents the number of neurons in the hidden layer of the BP model, represents the number of neurons in the input layer of the BP model, represents the number of neurons in the output layer of the BP model, and represents an adjustment constant ranging from 0 to 10. The number of hidden layers and the number of neurons are key parameters affecting the model's training effect and prediction performance. The general principle is: under the premise that the error reaches the expected target and the prediction performance is relatively good, the minimum number of hidden layers and neurons should be selected. According to Equation 2, the value range of the number of hidden layers and the number of neurons is shown in Table 13.

[0141] Table 13 Hyperparameter setting range for the BP model:

[0142]

[0143] Tables 14 to 17 show the fluctuations of RMSE, MAE, and MAPE for the four outputs Y1, Y2, Y3, and Y4 with different numbers of hidden layer nodes. The results show that the errors of outputs Y1 and Y4 are more significant compared to outputs Y2 and Y3. Therefore, when selecting the number of hidden layer nodes, priority should be given to minimizing the errors of outputs Y1 and Y4, while also selecting the number of hidden layer nodes that keeps the errors of Y2 and Y4 in single digits.

[0144] Table 14 Evaluation Indicators for Output Y1:

[0145]

[0146] Table 15 Output Y2 Evaluation Indicators:

[0147]

[0148] Table 16 Output Y3 Evaluation Indicators:

[0149]

[0150] Table 17 Output Y4 Evaluation Indicators:

[0151]

[0152] To further determine the number of hidden layer nodes, the RMSE and R of the training set were selected. 2 As a primary reference indicator. Figure 9 This demonstrates the effect of different numbers of hidden layer nodes on the MRSE and R of output layer neurons.2 The results showed that as the number of hidden layer neurons increased, the training set R of the four neurons in the output layer... 2 Both R and RMSE are changing irregularly. 2 The changes in Y1 and Y4 were relatively stable, while the changes in RMSE were significant, especially for Y1 and Y4, indicating that the BP model had significant errors in terms of fluidity and condensation time. Therefore, we prioritized selecting the number of hidden layer nodes with the smallest errors in outputs Y1 and Y4, and further compared and analyzed the RMSE and MAPE of the four output neurons with five groups of hidden layer node numbers: 3, 4, 5, 8, and 11.

[0153] Figure 9 This demonstrates 5 pairs of hidden layer node numbers for 4 output layer neurons R. 2 The effects of RMSE, MAE, and MAPE were observed. The results showed that considering R... 2 The optimal number of hidden layer nodes is 4 and 11. Considering RMSE, the optimal number of hidden layer nodes is 11; considering MAE, the optimal number of hidden layer nodes is 11; considering MAPE, the optimal number of hidden layer nodes is 11. In summary, selecting 1 hidden layer and 11 hidden layer nodes results in a 3-11-4 structure for the BP model. Figure 10 As shown.

[0154] The training results of the BP model are as follows Figure 11 As shown, it can be seen that at this time, the training set, test set, validation set, and all sample data points of the BP model are all distributed near the straight line, with almost no outliers, and R... 2 All values ​​reached above 0.99, indicating that the BP model trained well. The predicted values ​​of the four outputs from the training and test sets were compared with the actual values. Figure 12 and Figure 13 As shown, the predicted values ​​of the four outputs of the BP model in both the training and test sets are quite close to the measured values, with very few data points showing large errors. This indicates that the BP model can adequately explain and predict the data, and the consistency between the measured and predicted values ​​is good. The RMSE of the four outputs in both the training and test sets is controlled within 10, and the difference between the predicted and measured values ​​is small, indicating that the BP model has good training performance, high prediction accuracy, and can be optimized for multiple objectives.

[0155] (2) Genetic algorithm optimization design:

[0156] Using the trained BP neural network as the performance predictor, the genetic algorithm parameters are set as follows:

[0157] Encoding method: Real number encoding (three variables are directly used as genes).

[0158] Fitness function:

[0159] (3)

[0160] Target: Flowability approaching 340 mm, bleeding rate approaching 0%, stone rate approaching 100%, and setting time approaching 180 min.

[0161] Evolutionary parameters: population size 200, crossover probability 0.85, mutation probability 0.15.

[0162] Termination criteria: number of generations ≥ 200 or fitness change rate < 1%.

[0163] The optimization process is as follows Figure 14 As shown in Table 18, after 131 generations of evolution, the fitness converged to 6.41. The characteristics of the grout prepared by the synchronous grouting material were compared with those of the traditional grout.

[0164] Table 18 Comparison of various properties of slurry between the model group and the traditional group:

[0165]

[0166] Compared to the control group, the BP-GA group showed a 40.3% increase in fluidity, a 21.5% decrease in bleeding rate, a 0.42% increase in stone rate, and a significant 66.7% reduction in setting time. All three parameters—fluidity, bleeding rate, stone rate, and setting time—were significantly improved, demonstrating a very clear optimization effect. Furthermore, compared to the control group, the BP-GA group showed a 703.9% increase in 3-day compressive strength and a 313.3% increase in 28-day compressive strength, indicating a significant improvement in the mechanical properties of the stones.

[0167] Using the BP-GA optimization method in Example 2 above, the dosage of the three factors corresponding to the four responses being simultaneously optimized to a higher level was calculated: metakaolin = 16.15 wt.%, recycled fine aggregate = 38.98 wt.%, and slag = 22.72 wt.%.

[0168] Through Examples 1 and 2 above, the optimized mix proportions obtained by the D-optimal method are: metakaolin = 16.94 wt.%, recycled fine aggregate = 25 wt.%, and slag = 26.63 wt.%, while the optimized mix proportions obtained by the BP-GA method are: metakaolin = 16.15 wt.%, recycled fine aggregate = 38.98 wt.%, and slag = 22.72 wt.%. Compared with the traditional group, the fluidity of the D-optimal and BP-GA optimized groups increased by 42.8% and 40.2%, respectively; the bleeding rate decreased by 62.4% and 21.5%, respectively; the stone rate increased by 0.2% and 0.42%, respectively; the setting time was shortened by 65.8% and 66.7%, respectively; the 3-day compressive strength increased by 809.1% and 703.9%, respectively; and the 28-day compressive strength increased by 385.0% and 313.3%, respectively. These results demonstrate that the overall performance is significantly superior to the traditional group in all aspects, while also achieving a balance between performance and cost.

[0169] Comparing the D-optimal and BP-GA optimized mix proportions, it can be found that the D-optimal group has greater fluidity, lower bleeding rate, and higher compressive strength compared to the BP-GA group. The aggregate density and setting time of the two optimized groups are similar, and the setting time error, the largest error among the two groups, is lower in the D-optimal group, indicating that the slurry performance of the D-optimal group is superior. This demonstrates the superiority of using the response surface methodology (D-optimal) to optimize the geopolymer-based recycled fine aggregate synchronous grouting material. However, the mix proportions of the two optimized groups also show that the BP-GA group has a higher recycled fine aggregate content than the D-optimal group, which is more conducive to the recycling of recycled aggregate and leveraging its green and low-carbon advantages. In other words, the BP-GA method better balances performance and cost. Therefore, when considering the selection of the two optimized mix ratios, if the geological conditions at the shield tunneling site are complex and it is difficult to guarantee the performance of the grout, selecting the D-optimal group optimized mix ratio can maximize the effect of the synchronous grouting process; if the geological conditions are relatively stable and there is a lot of construction experience, selecting the BP-GA group optimized mix ratio can maximize the recycling of recycled fine aggregates while achieving the synchronous grouting target, and can also save grout costs.

[0170] It should be understood that the raw material ratio disclosed in this invention is a slurry ratio obtained through a mathematical regression model. In actual engineering, the raw material ratio can be appropriately adjusted within the given range according to different needs. For those skilled in the art, several optimizations can still be made without departing from the principle of this invention, and these optimizations should also be considered within the scope of protection of this invention.

Claims

1. A geopolymer-based recycled fine aggregate shield simultaneous grouting material, characterized by, The components include: cementitious material, aggregate, alkali activator, water reducing agent, retarder and water; wherein the mass ratio of alkali activator to water is 3:7, the water reducing agent accounts for 1% of the mass of the cementitious material, the retarder accounts for 2% of the mass of the cementitious material, the water-binder ratio is 0.7, and the sand-binder ratio is 1.5; The cementitious material is composed of a geopolymer precursor; the geopolymer precursor is composed of fly ash 56.43%, slag 26.63% and metakaolin 16.94% by mass percentage; or the geopolymer precursor is composed of fly ash 61.13%, slag 22.72% and metakaolin 16.15% by mass percentage; When the geopolymer precursor is composed of fly ash 56.43%, slag 26.63% and metakaolin 16.94% by mass percentage, the corresponding aggregate is composed of river sand 75% and recycled fine aggregate 25% by mass percentage; When the geopolymer precursor is composed of fly ash 61.13%, slag 22.72% and metakaolin 16.15% by mass percentage, the corresponding aggregate is composed of river sand 61.02% and recycled fine aggregate 38.98% by mass percentage.

2. The grouting material according to claim 1, characterized in that The river sand is ordinary fine river sand with a fineness modulus of 1.9 and a clay content of 1.68%; the recycled fine aggregate is recycled concrete fine aggregate with a fineness modulus of 2.2 and a gradation belonging to 2 gradation region.

3. The grouting material according to claim 1, characterized in that, The fly ash is grade II F-class fly ash, with specific surface area ≥360m 2 / kg; the slag is S95 grade, with specific surface area ≥429m 2 / kg; and the metakaolin is industrial grade.

4. The grouting material according to claim 3, characterized in that The alkali activator is anhydrous sodium silicate; the water reducing agent is a naphthalene-based water reducing agent; and the retarder is barium nitrate.

5. A method of preparing the grouting material according to any one of claims 1 to 4, characterized in that The method comprises the following steps: (1) Dry mixing: after fly ash, slag, metakaolin, river sand, recycled fine aggregate, anhydrous sodium silicate, water reducing agent and retarder are weighed according to the proportion, they are poured into a stirring pot and stirred at a low speed for 3 minutes; (2) Wet mixing: after water is added to the dry mixed material according to the water-binder ratio, the stirring pot is switched to a high speed position and continues to stir for 4 minutes to obtain a uniform geopolymer-based recycled fine aggregate shield synchronous grouting material.

6. An optimized method for the preparation of the grouting material according to any one of claims 1 to 4, characterized in that, The method comprises: Taking the metakaolin content A, the recycled fine aggregate content B and the slag content C as independent variables, taking the fluidity Y1, the bleeding rate Y2, the stone formation rate Y3 and the setting time Y4 as response variables, using the D-optimal module of Design-Expert software to construct a multi-objective expectation function, setting the upper limit, lower limit and weight of each response to maximize the total expectation function D, and obtaining the optimized ratio: metakaolin 16.94 wt%, recycled fine aggregate 25 wt% and slag 26.63 wt% by mass percentage.

7. An optimized method for the preparation of the grouting material according to any one of claims 1 to 4, characterized in that, The method comprises: The three-layer BP neural network was constructed with the metakaolin, recycled fine aggregate and slag content as the input layer of three neuron nodes X1, X2, X3, and the fluidity, bleeding rate, stone rate and setting time as the output layer of four neuron nodes Y1, Y2, Y3, Y4, the number of input layer nodes was 3, the number of hidden layer nodes was 11, and the number of output layer nodes was 4; the training set and the test set were divided according to the ratio of 4:1 based on the experimental data set, the network was trained with the evaluation indexes of RMSE, MAE, MAPE and R²; the trained BP neural network was used as the predictor, the genetic algorithm was used, the fluidity of 340 mm, the bleeding rate of 0%, the stone rate of 100% and the setting time of 180 min were used as the target, and the fitness function was minimized, and the optimized ratio was obtained: according to the mass percentage, the metakaolin was 16.15 wt.%, the recycled fine aggregate was 38.98 wt.%, and the slag was 22.72 wt.%.

8. The use of the grouting material according to any one of claims 1-4 in the synchronous grouting of a shield tunnel in a water-rich sand layer, a high water pressure stratum or a dissolution stratum, injected through a shield machine synchronous grouting system into a shield tail gap; When used in a water-rich sand layer or a high water pressure stratum, the geopolymer precursor used is composed of fly ash 56.43%, slag 26.63% and metakaolin 16.94% by mass percentage, and the aggregate is composed of river sand 75% and recycled fine aggregate 25% by mass percentage; When used in a dissolution stratum, the geopolymer precursor used is composed of fly ash 61.13%, slag 22.72% and metakaolin 16.15% by mass percentage, and the aggregate is composed of river sand 61.02% and recycled fine aggregate 38.98% by mass percentage.

Images