Mechanical characteristics and solidification mechanism of sodium silicate activated fly ash cement composite loess based fluidized stabilized materials Authors: Hongjun Zhang, Dapeng Han, Bo Hua, Yun Cheng, Shubo Sui, Xiangpan Jiao, Zhiwei Xu, Binbin He, Lingyi Chen, Ruoqi Hu, Zhanping Song Published: 2026 ABSTRACT AbstractTo address the practical engineering challenges of compaction difficulty and low efficiency associated with traditional backfilling methods for narrow and confined urban foundation trenches, this study proposes an eco-friendly backfill material: sodium silicate-activated fly ash-cement composite fluidized stabilized soil, using in-situ loess from the Xiong’an New Area as raw material. Material components were optimized through curing agent ratio determination and single-factor tests. The development laws of mechanical properties, the micro-scale solidification mechanism, and the flow-filling characteristics within narrow trenches were systematically elucidated by integrating scanning electron microscopy and computational fluid dynamics (CFD) simulations. The results indicate that a fly ash-to-cement mass ratio of 1:1 serves as the critical benchmark for optimizing the composite curing agent proportion. This ratio effectively inhibits excessively rapid early water evaporation and provides a stable hydration environment for the later pozzolanic reaction of fly ash. The optimal mix proportion achieves a balance between high fluidity and high strength. A water content of 51.25% ensures sufficient cementitious reactions. A sodium silicate content of 2.0% strikes a balance between activating fly ash and avoiding excessive slurry viscosity. A curing agent content of 35% facilitates the formation of a continuous and dense cementitious network. Under the optimal proportion, the material exhibits a spread flow of 175 mm and a 14-day compressive strength of 4.53 MPa, meeting the requirements for pumping construction and strength. The compressive strength of fluidized stabilized soil results from the synergistic interaction of fluidity, molding compactness, and water loss behavior. This is manifested microscopically by the generation of cementitious products, pore filling, and particle cementation, and macroscopically reflected in the coupled effects of physical water migration and chemical water consumption. Adopting a single-pour length covering three utility tunnel Sect. (9 m) and a pouring speed of 5 m/s combined with a terminal speed reduction process can significantly enhance the compactness and construction quality of trench backfilling. This approach facilitates the formation of a stable flow field and reduces air bubble retention in corners. Field application demonstrates that this process can meet the dual requirements of construction efficiency and quality. The research findings provide a theoretical foundation and key technical guidance for the construction of underground comprehensive utility tunnels in the Xiong’an New Area.Keywords: Narrow backfill zone, Fluidized stabilized soil, In-situ loess, Microcosmic curing mechanism, Pouring processSubject terms: Engineering, Materials science Introduction As a critical infrastructure for integrating various municipal pipelines and ensuring the safe and stable operation of urban lifelines, the scale and quality requirements for urban underground comprehensive utility tunnels are continuously increasing. In the widely adopted open-cut method for utility tunnel construction, after the main structure of the tunnel is cast, a narrow and spatially constrained backfill zone, referred to as the narrow backfill zone, is formed between the outer side of the tunnel and the excavation support wall. Traditional backfill materials and techniques for utility tunnel backfill zones face significant challenges. Therefore, optimizing a new backfill material that combines good workability, appropriate mechanical strength, and low environmental impact has become a key issue to be addressed in the field of geotechnical engineering. The emergence of fluidized stabilized soil (FSS) technology offers a highly potential solution to the above engineering challenges. FSS typically uses excavated in-situ soil as the primary aggregate, mixed with an appropriate amount of curing agent and water to form a slurry material with good fluidity, capable of self-leveling and self-compaction. FSS offers advantages such as self-compaction, homogeneity, low compressibility, good impermeability, and high utilization of excavated soil, which translates to cost savings, making it particularly suitable for backfilling in confined spaces. As a result, FSS is especially well-suited for scenarios where traditional techniques are difficult to apply, such as urban confined spaces, irregular foundation pits, and areas around pipelines1. However, the performance of FSS is highly dependent on the selection and optimization of its core component-the curing agent system. Currently, mainstream curing agent systems face the following bottlenecks: although traditional cement-based systems are technologically mature and develop early strength rapidly, the extensive use of cement not only increases costs but also poses significant issues related to high energy consumption and carbon emissions during production. More critically, the hydration products of cement exhibit substantial drying shrinkage, which can easily lead to shrinkage cracks during the hardening process of FSS2, affecting its long-term integrity, impermeability, and interfacial stability with adjacent structures3. While purely solid waste-based cementitious systems can minimize costs and environmental footprints, aligning with the zero-waste city and dual carbon strategies, their chemical reactivity under ambient temperatures is generally low. These systems often rely on strong alkali activators at relatively high dosages to activate their latent reactivity. This results in slow early strength development, making it difficult to meet the demands of certain engineering applications requiring rapid support or continuous construction. Additionally, excessive alkali content may introduce new issues such as efflorescence and long-term volume stability4. To balance performance, cost, and environmental benefits, the low-cement, high-solid waste composite activation system has garnered widespread attention. This system aims to utilize a small amount of cement to provide early strength and alkalinity, with industrial solid waste serving as the primary cementitious component, supplemented by a chemical activator. Through the synergistic effects of multiple materials, it seeks to optimize overall performance. It is considered effective in promoting the dissolution of aluminosilicate materials like fly ash and the subsequent geopolymerization reaction, generating cementitious products with a three-dimensional network structure5. Gu et al.6 and Ozturk et al.7 investigated various solid waste composite systems for soil solidification, confirming that low early strength is a shortcoming limiting engineering application. Qiu et al.8 formulated a multi-source solid waste-based cementitious material. By optimizing the activation conditions, they achieved an unconfined compressive strength at 14 days reaching 80% to 90% of the 28 day strength, demonstrating the potential of material design and activation technology to improve early-age performance. These studies generally indicate that in composite systems, the ratio of solid waste to cement, the type and dosage of the activator (modulus, concentration), and the alkalinity of the liquid phase are the key controlling factors determining the process of hydration/geopolymerization, the composition of the reaction products, and the macroscopic properties. However, for region-specific soils, there remains a lack of in-depth and systematic reports on the optimal proportion range of fly ash to cement under activation by sodium silicate solution and on how this proportion, coupled with activator parameters and soil characteristics, influences the mechanical properties and microstructure of FSS. The fundamental reason for differences in the macroscopic properties of FSS lies in varying material mix proportions. An in-depth understanding of the solidification mechanism forms the theoretical basis for the designability and controllability of material performance. Shang et al.9 and Zhao et al.10, through scanning electron microscopy (SEM) and X-ray diffraction (XRD) tests, pointed out that ettringite (AFt) and calcium silicate hydrate (C-S-H) gel generated from cement hydration, along with amorphous or crystalline geopolymer gels formed by alkali activation, jointly contribute to cementing loose soil particles and filling pores. This is the fundamental reason for strength enhancement. Liu et al.11 further theorized the solidification process of alkali-activated fly ash into four consecutive yet overlapping stages. This framework deepens the understanding of the reaction kinetics and product formation in alkali activation. Liu et al.12 studied the mechanism of alkali activator on the setting time of Class F fly ash under medium-high temperature, summarizing the solidification process into three steps and revealing the action mechanism of the composite activator. Meng et al.13 found through research on cement-solidified marine-dredged clay that the strength of solidified soil is related to the saturation degree of free water and bound water. Li et al.14 used computational fluid dynamics software ANSYS fluent to simulate the flow and deposition characteristics of FSS and analyzed the influence of pumping parameters on the filling effect. However, for the narrow backfill zones of land-based utility tunnels, the boundary conditions and geometries vary significantly. How do factors such as the size of the backfill zone, the layout of pouring ports, and the pouring speed collectively affect the final filling degree, interfacial bonding quality, and the formation mechanisms of internal defects? This requires in-depth investigation through a combination of numerical simulation and physical model tests to extract the key process control parameters. This study takes the in-situ excavated loess from the Xiong’an New Area as the research object, systematically investigating the backfilling application of FSS in narrow backfill zones. First, after determining the fly ash-to-cement ratio, the mix proportions of the composite system comprising fly ash, cement, and sodium silicate solution were optimized through single-factor tests. This was done to obtain feasible proportions that meet the requirements of pumpability, rapid setting, and a certain strength level. Subsequently, representative samples were subjected to SEM and EDS analysis. The findings from these microstructural examinations, combined with macroscopic characteristics, were used to gain an in-depth understanding of the solidification mechanism. Finally, based on the rheological parameters determined from the optimized mix proportion, computational fluid dynamics simulations were employed to conduct a thorough investigation into the flow behavior of FSS within the narrow backfill zones. Furthermore, optimized pouring process parameters were proposed. This paper aims to provide an optimized mix proportion that meets both engineering performance requirements and green, low-carbon principles for this backfilling application. Additionally, based on the solidification mechanism and numerical simulations, it seeks to offer technical guidance for efficient on-site backfilling operations. Materials and methods Engineering background As a major national strategic development zone, the construction of underground comprehensive utility tunnels in the Xiong’an New Area is advancing comprehensively. The utility tunnel project in the Start-up Area, which serves as the foundation for this research, is located in Baoding City, Hebei Province, with a total length of 5.21 km. Certain sections run beneath municipal roads (Fig. 1). Considering that the backfill quality of these sections directly impacts the long-term stability of the road structure and they are subjected to the most critical stress conditions, they were selected as the focus of this study. The open-cut method was employed for construction. The excavation depth of the foundation pit was approximately 10 m, with the stable groundwater table depth ranging from 14.30 m to 15.20 m. Consequently, the influence of groundwater on the backfilling operation was negligible. During construction, to accommodate machinery operation and structural installation, a narrow lateral void, referred to as the narrow backfill zone was formed between both sides of the utility tunnel structure and the excavation support wall. This zone has a width of approximately 0.30 m and a height of about 4.50 m. The spatial distribution and dimensional characteristics of this zone are illustrated in Fig. 1. Due to the confined spatial constraints of the narrow backfill zone, traditional layered compaction backfilling techniques face significant challenges in terms of construction feasibility, efficiency, and quality uniformity. Spatial distribution and dimensional characteristics of the narrow backfill zone. Base map: Aliyun DataV.GeoAtlas (https://datav.aliyun.com/portal/school/atlas/area_selector), data from Amap. The project site (highlighted in yellow) and the location of Beijing were added by the authors for reference. To address the aforementioned technical challenges, this paper proposes an integrated excavation-pouring cyclic construction scheme. This scheme aims to utilize the excavator itself as a temporary formwork, enabling the immediate on-site preparation of FSS from the excavated in-situ soil within the trench and its subsequent pumping for backfilling. Once the material hardens, the machinery advances to the next cycle. This process imposes three core requirements on the backfill material, it must possess rapid-setting characteristics to align with the pace of the cyclic construction; the fresh slurry must exhibit excellent fluidity and self-compacting properties to ensure complete filling of the complex space within the narrow backfill zone without vibration; and the hardened material must meet the mechanical performance requirements for backfill structures beneath municipal roads. Therefore, to adhere to the principles of green construction and control project costs, the research proposes using sieved in-situ excavated loess as the primary aggregate, partially replacing cement with fly ash to form a composite solidification system, and introducing sodium silicate solution as an alkali activator. This approach aims to formulate a green, low-carbon, rapid-setting, and early-strength FSS material. Based on the technical challenges of the cyclic construction process and backfilling in narrow, elongated spaces, and in accordance with the standard DBJ51/T188-2022, this study aims to formulate an FSS that meets the following key performance indicators: a setting time not exceeding 4 h to accommodate rapid construction cycles, and a 14 day unconfined compressive strength not less than 1.00 MPa to ensure the structural stability of the backfill mass. Experimental materials To ensure that the experimental results provide direct guidance for on-site construction, this study strove to maximize the alignment of material selection with actual engineering conditions. All primary raw materials were sourced from the utility tunnel project site in the Start-up Area of the Xiong’an New Area or were equivalent common engineering materials. (1) In-situ excavated loess obtained from a depth of approximately 10 m at the project site. (2) Class F high-calcium (CaO) fly ash, actually used in the project and produced by Jinyu Jidong Cement Co., Ltd. (3) Ordinary portland cement (P.O 42.5) conforming to the national standard. (4) Industrial sodium silicate solution with a density of 1.45 g/cm³ used as the alkali activator, with the main component being Na₂O·nSiO₂. (5) Laboratory tap water as mixing water. The basic physical properties of the excavated loess were determined according to the Standard for Geotechnical Testing Method (GB/T 50123 − 2019), with results are presented in Table 1. The soil sample had a plasticity index of 11.30, a liquid limit of 29.80%, and a plastic limit of 18.50%. According to the classification standard, it is a typical silty clay, whose engineering properties have a direct impact on the formulation of the backfill material. Table 1Basic physical properties of undisturbed loess.Natural moisture content/%Liquid limit/%Plastic limit/%Liquidity indexPlasticity indexMoisture content after 24 h air drying/%13.4429.818.5−0.4511.33.09 Basic physical properties of undisturbed loess. To simulate the on-site soil handling process and ensure the uniformity and repeatability of the test results, standardized pretreatment was performed on the retrieved undisturbed soil. The soil sample was air-dried naturally for 24 h to eliminate significant fluctuations in initial moisture content15. Subsequently, it was passed through a 10 mm square-hole sieve to remove coarse particles and impurities such as plant roots. Finally, the sieved soil was subjected to moderate grinding using a laboratory ball mill to improve its particle size distribution and enhance subsequent mixing efficiency. Particle size analysis of the ground soil sample was conducted using the sieve analysis method. With reference to relevant standard methods16,17, the mass retained on each standard sieve was weighed, and the cumulative percentage retained was calculated to obtain the particle size distribution curve shown in Fig. 2. The analysis indicated that particles smaller than 1 mm accounted for up to 95% of the soil sample, classifying it as fine-grained soil. This has a decisive influence on the water demand, rheological properties, and solidification reaction interfaces of the fluidized stabilized soil. Particle size distribution curve of undisturbed loess. To gain an in-depth understanding of the soil’s chemical nature and its potential role in the solidification process, X-ray diffraction (XRD) phase analysis was conducted on a ground representative soil sample. The primary mineral constituents of the soil were identified as quartz (SiO₂) and calcite (CaCO₃), which aligns with the typical mineralogical composition of loess from northern China. Quartz, as an inert skeletal particle, primarily contributes to the physical filling effect of the material. The presence of calcite may provide an additional calcium source in the alkali-activated environment, potentially participating in the formation of cementitious products. The main chemical compositions of the fly ash and cement were analyzed by X-ray fluorescence (XRF) spectroscopy, and the results are presented in Table 2. Table 2Main chemical compositions of cement and fly ash.MaterialSiO2/%Al2O3/%Fe2O3/%CaO/%MgO/%Na2O/%Cement13.052.633.4941.462.39<0.60Fly ash52.021.03.218.02.43.4 Main chemical compositions of cement and fly ash. Experimental design To systematically optimize the mix proportion of the FSS, this study adopted a stepwise, progressive experimental design approach, aiming to balance experimental efficiency with the reliability of conclusions. First, the curing agent was considered as a composite whole, focusing on investigating the influence of its internal composition, specifically, the mass ratio of cement to fly ash—on material performance. Given the low early-age activity of fly ash, an excessively high proportion may lead to delayed strength development, potentially affecting the pace of cyclic construction18,19. Drawing on the common mix proportion ranges for cement-fly ash composite systems in similar studies20,21, two representative mass ratios were initially selected for comparison 2:1 (cement-dominant) and 1:1 (equal blend). By testing the water loss with these two ratios at the same curing age, differences in early-age hydration and solidification processes under different proportions can be indirectly reflected, thereby providing a basis for selecting the fundamental ratio within the curing agent. Based on the aforementioned principle and with reference to the suggested total curing agent content range in existing literature 19, three content levels were set as 25%, 30%, and 35%. This formed a benchmark experimental mix proportion scheme encompassing different cement-fly ash ratios combined with total curing agent contents, as detailed in Table 3. This baseline experiment systematically evaluated the early-age setting characteristics of the material under different proportions, establishing a benchmark for subsequent single factor optimization. Table 3Baseline experimental mix proportions.SpecimenCement/%Fly ash/%Curing agent content/%Water/%P123.3311.673550P217.5017.503550P320103050P415153050P516.678.332550 Baseline experimental mix proportions. After determining the suitable ratio between cement and fly ash within the curing agent, the influence of three key factors, including water content (W), sodium silicate solution content (SS), and total curing agent content (CA) on the mechanical properties of the material was further investigated separately using the single factor test method to determine their optimal content ranges. The water content, referenced against the baseline tests and workability requirements, was set at five gradient levels (48.00%, 51.25%, 54.50%, 57.25%, and 61.00%). The sodium silicate solution serving as the alkali activator was set at six gradient levels (0%, 1%, 2%, 3%, 4%, and 5%) to examine the relationship between activation effect and dosage. The total curing agent content was set at seven gradient levels (10%, 15%, 20%, 25%, 30%, 35%, and 40%) to comprehensively assess the trend of strength contribution from the cementitious material quantity. This study used the 7 day and 14 day unconfined compressive strength (UCS) as the primary evaluation metrics. By analyzing the strength response when each factor was varied individually, the optimal level for each factor was selected, which was then combined to form an optimized mix proportion proposal. During the single factor tests, to ensure comparability of results, only one factor was changed at a time, while the other two factors were fixed at their baseline levels (water content at 55%, sodium silicate solution content at 2.5%, and total curing agent content at 30%). For brevity in subsequent text, these three factors are denoted as W, SS, and CA respectively, with subscript numbers indicating the specific content (%). For example, W₅₅ represents a water content of 55%. Through this phased experimental design, the test scale can be effectively controlled while systematically revealing the influence mechanisms of each material component and the trends of their interactions. This provides a solid experimental basis for formulating an FSS that meets the requirements of rapid setting, early strength, high fluidity, and suitable strength. Specific experimental procedures (1) Sample preparation procedure. The dried soil sample, fly ash, and cement, along with the required sodium silicate solution and tap water, were precisely weighed according to the preset mix proportions. The soil sample and the curing agents (fly ash and cement) were placed in a mixing bowl and pre-mixed manually for 1 min to ensure homogeneity of the dry materials. Subsequently, the pre-prepared mixture of sodium silicate solution and tap water was added to the dry materials all at once. The mixture was then mechanically blended using an XREM-60 model electric mixer initially at a low speed (300 ± 10 rpm) for 1 min to achieve preliminary wetting and mixing, followed by a highspeed (600 ± 10 rpm) mixing for 2 min to ensure the slurry reached a fully uniform, lump-free state. After mixing, the slurry was immediately poured into cylindrical molds with an internal diameter of 50 mm and a height of 100 mm. The surface was leveled using a putty knife. The molded specimens were cured under ambient conditions (20 ± 2 °C, relative humidity > 90%) for 24 h before demolding. They were then transferred to a standard curing room (temperature 20 ± 2 °C, humidity ≥ 95%) and cured until the predetermined ages. The entire preparation process is illustrated in Fig. 3. Preparation of fluidized stabilized soil specimens. (2) Testing procedures. Macroscopic performance testing of soil samples. To minimize the influence of environmental fluctuations on test results, comparative specimens from the same batch were mixed, molded, and demolded within the same time period. To investigate the early-age water loss characteristics of the materials, an electronic balance with an accuracy of 0.01 g was used to periodically weigh the curing specimens. The mass loss due to water dispersion and the average water loss rate were obtained by calculating the mass difference between consecutive time points. To evaluate the mechanical properties of the FSS, unconfined compressive strength tests were conducted on specimens at 7 and 14 days, following the DBJ51/T188-2022 standard, which specifies that for early-strength materials, the testing age can be 7 or 14 days. Therefore, an MTS electro-hydraulic servo press was used in displacement control mode (loading rate of 1 mm/min)22,23. Load-displacement data were recorded synchronously during the test, and stress-strain curves were automatically generated for analyzing the material’s deformation characteristics and failure process24–27. Analysis of failure morphology characteristics. After mechanical testing, the failed specimens were systematically photographed. By comparatively observing the distribution patterns and propagation paths of surface cracks, as well as the characteristics of harmful pores, the failure modes and internal homogeneity of specimens with different mix proportions were qualitatively assessed. For quantitative comparison, image processing software was employed to annotate typical features. Main cracks were outlined with red dashed lines, where the line thickness visually reflected the crack width. Visible surface pores were encircled with red solid lines of uniform width, with the circle diameter corresponding to the pore size. Microstructural characterization and analysis of soil samples. Based on the macroscopic analysis, representative fragments from damaged specimens were selected, vacuum-dried, and sputter-coated with gold. Microstructural morphology observation was then performed using a Tescan Mira4 field emission SEM28,29, achieving a resolution of up to 2 nm, with an accelerating voltage of 15 kV and a working distance of 10 mm. Combined with the attached EDS and a collection time of 60 s, micro-area elemental composition analysis was conducted on typical zones of newly formed hydration products. This aimed to reveal, at the microscale, the morphology and distribution of cementitious products and the structural characteristics of the soil-cementitious matrix interface transition zone under different mix proportions, thereby establishing the intrinsic link between microstructural evolution and macroscopic mechanical properties. Construction simulation and analysis of optimized-mix FSS. Based on the rheological parameters measured from the optimized-mix specimen, a three-dimensional transient simulation of the FSS pouring process within a narrow backfill zone was conducted using the ANSYS Fluent fluid computation software. The multiphase volume of fluid (VOF) model was selected. By setting different working conditions, such as pouring inlet location, pouring speed, and backfill zone length, the simulation analyzed the slurry’s flow front morphology, filling uniformity, and potential defects. The objective was to determine the maximum achievable single-pour length and an economically reasonable pouring speed under the premise of ensuring pouring compactness and uniformity, providing a quantitative basis for shortening construction intervals and optimizing on-site techniques. Figure 4. Macro- and micro-mechanical test system. Experimental results and analysis Optimization of curing agent proportion To investigate the early hydration and water loss behavior of materials under different curing agent ratios, this study systematically measured the mass changes of specimens molded and demolded simultaneously, thereby obtaining the water loss rate at different time intervals. Within the relatively short observation period, the consumption of bound water by cement hydration and the pozzolanic reaction of fly ash could be neglected. Therefore, the measured water loss rate primarily reflects the macroscopic trend of free water migration from inside the specimen to the external environment. The test revealed that the surfaces of all specimens appeared relatively dense and free of significant voids after demolding, indicating that the freshly mixed slurry possessed good fluidity and self-compacting properties. Based on the observation period, the water loss process was divided into three stages for analysis: the first 8 h before demolding (WL1), 0–15.50 h after demolding (WL2), and 15.50–22.50 h after demolding (WL3). Additionally, the mass ratio of cement to fly ash was defined as α, and the total curing agent content was set according to Table 3. The water loss rates of specimens with different curing agent ratios are shown in Fig. 5. Water loss rate of specimens with different curing agent proportions. As shown in Fig. 5, during the WL1 stage before demolding, the water loss rate of all specimens exhibited an overall decreasing trend with the reduction of the total curing agent content, and the extent of the decrease grew progressively as the content decreased. At the same content level, the water loss rate of specimens with α = 2 was generally higher than that of specimens with α = 1. This suggests that a higher proportion of cement may promote the formation of capillary channels within the early-age slurry, thereby accelerating water migration. For specimens with α = 2 (P1, P3, P5), comparing their post-demolding water loss behavior revealed that with each 5% increase in curing agent content, the growth rate of the water loss rate in the WL2 stage increased from 5.50% to 8%, and in the WL3 stage from 0.50% to 5.10%. For individual specimens, the reduction rate of the water loss rate from the WL2 to WL3 stage was 5.30%, 9.80%, and 12.30%, respectively. This indicates that under a higher cement ratio, an increase in curing agent content not only accelerates early water evaporation but also leads to a more pronounced attenuation of the water loss rate over time. This is likely related to the rapid hydration of cement, which causes the surface layer of the slurry to form a denser structure earlier, thereby hindering the continued outward migration of internal moisture. However, for specimens with α = 1 (P2, P4), when the curing agent content was increased by 5%, the reduction rates of the water loss rate in the WL2 and WL3 stages were 4.20% and 4.10%, respectively. Furthermore, the reduction rate of the water loss rate across different time intervals remained stable at 9.70% for both specimens. This indicates that under the condition of an equal mass ratio of cement to fly ash, increasing the curing agent content did not promote water evaporation but rather suppressed the water loss process to some extent. Moreover, the temporal trend of water loss was relatively consistent. The reduction rate of the water loss rate for specimens with α = 1 was generally lower than that of their α = 2 counterparts, suggesting that an increased proportion of fly ash helps mitigate the dynamic changes in water loss. The above results demonstrate that the water loss behavior is influenced not only by the total curing agent content but is also closely related to the cement/fly ash ratio (α)30. From a mechanistic perspective, a higher cement content leads to rapid early hydration and the formation of more hydration products. This may, on one hand, increase the early porosity of the slurry, promoting water escape. On the other hand, it can also lead to the earlier formation of a solidified shell, causing a significant decrease in the water loss rate at later stages. Fly ash has lower reactivity in the early stages. Its particle morphology effect and micro-aggregate filling action help improve the flocculated structure of the slurry, slowing the development of water migration channels, thereby resulting in a more gradual water loss process. Based on this analysis, increasing the proportion of fly ash in the curing agent is not only beneficial for reducing material costs and enhancing solid waste utilization, aligning with the concept of green construction, but also effectively modulates the water loss characteristics of the slurry. This provides a more stable hydration environment for the cementitious materials, which is conducive to long-term strength development. Therefore, comprehensively considering water loss behavior, economic viability, and environmental requirements, this study determined a cement-to-fly ash mass ratio of 1:1 as the baseline proportion for the composite curing agent, to be used in subsequent single-factor experiments. Morphological characteristics of stress-strain curves To conduct an in-depth analysis of the mechanical behavior and deformation characteristics of FSS under different mix ratios, unconfined compression tests were performed on specimens at a 14 day curing age. The corresponding stress-strain curves are shown in Fig. 6. The characteristics of the stress-strain curves indicate that the peak stress for all specimens with different mix ratios occurred at a strain of approximately 2.50%, with peak strengths around 4.25 MPa. This demonstrates that, within the dosage range determined in this study, mix proportion optimization can enable the material to achieve comparable ultimate bearing capacities. By comparing the characteristics of the stress-strain curves31, it can be observed that the compaction stage occurred at the initial loading phase, with strains generally less than 0.50%. The stress-strain curve exhibited a concave shape, and the rate of stress increase with strain gradually accelerated. This stage primarily corresponds to the progressive closure of inherent micropores and defects within the specimen, leading to structural densification32,33. The elastic stage generally corresponded to strains of 0.50%–1.50%. The stress-strain curve approximated a linear relationship, with the specimen undergoing primarily recoverable elastic deformation, and the structure remained largely intact. The plastic stage occurred at strains of approximately 1.50%–2.50%. The slope of the stress-strain curve gradually decreased, and stress increase slowed until reaching the peak. During this stage, local damage is initiated at the bonds between the cementitious structure and soil particles, with internal microcracks gradually nucleating and propagating. Layered debris spalling and the appearance of fine cracks could be observed on the specimen sides. The failure stage was the post-peak softening segment, where stress decreased with increasing strain, reflecting that the specimen retained a certain residual strength after the loss of its overall bearing capacity, corresponding to the complete formation of the macroscopic failure pattern. Although the peak strengths of specimens with different mix ratios were similar, their stress-strain curves exhibited significant differences in the slopes of each stage, the positions of turning points, and the shapes of the softening segments. These differences reflect the substantial influence of varying material compositions on the internal structure and failure mechanism. Characteristics of stress-strain curves. (a) Water content, (b) Sodium silicate solution, (c) Curing agent. As seen in Figs. 6(a)-(b), with the increase in the content of water and sodium silicate solution, the specimen strength showed a trend of first rising rapidly and then declining slowly, accompanied by changes in the failure mode. At low dosages, the failure predominantly exhibited a plastic mode with good ductility. When the dosage increased to near the optimal level, the curve peak became sharp, showing certain brittle characteristics. Upon further increasing the dosage, the deformation characteristics of the specimens gradually reverted to a more ductile failure mode. Analysis suggests that an appropriate amount of water and sodium silicate is conducive to promoting thorough cementitious reactions34, forming a continuous and dense bonding network, which enhances strength but also increases brittleness. Conversely, an excessive liquid phase can lead to a looser initial slurry structure and increased porosity. While improved fluidity may contribute to some ductility, it results in a decrease in strength. As shown in Fig. 6(c), with the increase in curing agent content, the strength first increased slowly and then declined rapidly. The failure mode also gradually shifted from initial plastic failure towards brittle failure. A moderate amount of curing agent allows the cementing effect to be fully realized, resulting in a uniform and dense structure. However, excessively high content may, due to concentrated heat of hydration, increased autogenous shrinkage, or internal stress imbalance, introduce micro-defects, leading to strength reduction and increased brittleness. From this, it may be inferred that different material ratios not only ultimately determine the compressive strength of the specimens but also profoundly influence their deformation response and failure mode during loading. This reflects differences in the internal cementation state, pore structure, and characteristics of the interfacial transition zone within the material, providing an important basis for subsequent material design and optimization based on performance requirements. Effect of the curing agent on the mechanical properties Quantitative characteristics of peak strain To further elucidate the influence of different material ratios on the deformation performance of FSS, this section focuses on the quantitative characteristics of peak strain. Figure 7 summarizes the peak strain values for each test group. The peak strains in the water (W) and sodium silicate solution (SS) test groups exhibited significant fluctuations without a clear monotonic trend. In contrast, the curing agent (CA) test group displayed a V-shaped variation in peak strain, initially decreasing and then increasing, indicating a strong correlation between deformation behavior and the content of cementitious materials. Specifically, the rates of change for peak strain in the water test group were 11.54%, 1.15%, −10.61%, and 25.00%. It is noteworthy that specimens W51.25 and W54.50 had similar peak strains, yet the latter exhibited significantly higher strength. This is primarily attributed to the minor difference in their pore structures, with W54.50 having a slightly higher porosity than W51.25, which did not lead to a marked difference in plastic deformation prior to failure. This indicates that an appropriate increase in water content promotes cementitious reactions without substantially altering the material’s deformation mechanism. Specimens W57.25 and W61.00 showed comparable ultimate strengths, but their peak strains differed significantly. This suggests that under excessively high water content, the material’s ultimate bearing capacity has reached a critical threshold. Its deformation capacity is then primarily governed by the pore structure and compressible space. In other words, excessive water increases the initial porosity, providing greater space for deformation under compression, thereby manifesting as increased strain alongside a plateau in strength. Distribution characteristics of peak strain in fluidized stabilized soil. The rates of change for peak strain in the sodium silicate solution test group were − 31.09%, 29.30%, 8.63%, 23.51%, and 5.19%, successively. Specimen SS₂ exhibited the lowest peak strain, indicating that the 2% dosage represents the optimal balance between the activation effect and the defect effect. Analysis suggests that an appropriate amount of sodium silicate effectively activates the fly ash, promoting the formation of dense cementitious gels, thereby increasing material stiffness and resulting in reduced strain. At a 1% dosage, the activation effect was insufficient, leading to relatively higher strain. Notably, specimens SS₃ and SS₅ had similar strengths, but SS₅ showed a significantly larger peak strain. This reflects that although SS₅ possesses higher cementitious potential, the excessive alkali content may cause issues such as increased slurry viscosity and difficulty in bubble expulsion. This leads to increased initial pore defects and reduced compactness. Consequently, even though the final strengths are comparable, the deformation capacity of SS₅ is more significantly influenced by its pore structure. Figure 7 also shows that the rates of change for peak strain in the curing agent test group were 9.56%, −13.92%, −13.70%, −10.48%, −15.60%, and 47.06%, respectively. Specimen CA₃₅ had the highest strength and the smallest peak strain, exhibiting obvious brittle characteristics. This indicates that before reaching the optimal content (approximately 35%), the cementitious network continuously strengthens with increasing curing agent content, leading to improved material stiffness and a gradual decrease in peak strain. However, when the content increased to CA40, the strength was lower than that of CA30, while the peak strain increased sharply by 47.06%. This demonstrates that an excessive amount of curing agent has a detrimental effect on material performance. The analysis suggests that an excessive amount of cementitious material impairs the workability of the slurry, leading to defects such as entrapped air bubbles and uneven particle distribution during molding. These initial defects act as stress concentration points and deformation coordination zones under load, thereby reducing overall stiffness. This causes the material to undergo larger deformation at relatively lower stress levels, presenting a non-linear decline phenomenon. Solidification characteristics of elastic modulus Figure 8 shows the elastic modulus of fluidized stabilized soil. The elastic modulus exhibited significant variations under different influencing factors, revealing distinct mechanisms underlying the formation of the material’s internal structure. As can be seen from Fig. 8, the rates of change for the elastic modulus in the water content test group were 48.89%, −15.42%, 15.88%, and − 30.46%, respectively. The peak value occurring at a water content of 51.25% indicates that this dosage represents the optimal moisture state for the system35. Under this condition, the water is sufficient to ensure thorough cementitious reactions, promoting the formation and growth of hydration products and geopolymerization products. Concurrently, the amount of excess free water is minimal, resulting in low porosity after evaporation and the formation of a dense, continuous cementitious structure. Consequently, the stiffness of the specimens is generally high. When the water content is below the optimal value (W48.00), although initial porosity is lower, insufficient water restricts the rate and extent of cementitious reactions, limiting product formation and leading to inadequate filling and cementing effects, thus resulting in lower overall structural stiffness. Conversely, when the water content is excessively high (W54.50, W61.00), the sodium silicate solution becomes overly diluted, reducing the alkalinity and ion concentration of the system, which slows down the cementitious reactions. Moreover, the excess free water leaves behind numerous pores after evaporation, causing a decrease in the compactness of the cementitious matrix and a looser structure, leading to a significant reduction in the elastic modulus. Elastic modulus of fluidized stabilized soil. Furthermore, the sodium silicate solution test group revealed that the rates of change for the elastic modulus were 65.84%, −5.62%, −11.11%, −2.23%, and 16.44%, respectively. Specimen SS₀ without sodium silicate had the lowest elastic modulus, confirming that the activity of fly ash is not effectively activated in the absence of an activator. The system primarily relies on cement hydration, resulting in an insufficient total amount of cementitious products and a weak structural skeleton stiffness. Specimen SS₂ achieved the highest elastic modulus, indicating that a 2% dosage effectively activates the fly ash, promoting the generation of substantial cementitious products and pore filling, thereby significantly enhancing structural densification and overall stiffness. It is noteworthy that the elastic moduli of specimens SS₃ and SS₄ decreased instead of increasing. This is primarily attributed to air void defects formed during the pouring process due to increased slurry viscosity and air content, which disrupt structural continuity and weaken stress transfer efficiency. The recovery of the elastic modulus for specimen SS₅ suggests that although a high sodium silicate dosage may introduce some porosity, its strong activating effect can still generate sufficient cementitious products, partially compensating for the structural defects and allowing stiffness to recover. As can be seen from Fig. 8, the rates of change for the elastic modulus in the water content test group were 48.89%, −15.42%, 15.88%, and − 30.46%, respectively. The peak value occurring at a water content of 51.25% indicates that this dosage represents the optimal moisture state for the system. Under this condition, the water is sufficient to ensure thorough cementitious reactions, promoting the formation and growth of hydration products and geopolymerization products. Concurrently, the amount of excess free water is minimal, resulting in low porosity after evaporation and the formation of a dense, continuous cementitious structure. Consequently, the stiffness of the specimens is generally high. When the water content is below the optimal value (W48.00), although initial porosity is lower, insufficient water restricts the rate and extent of cementitious reactions, limiting product formation and leading to inadequate filling and cementing effects, thus resulting in lower overall structural stiffness. Conversely, when the water content is excessively high (W54.50, W61.00), the sodium silicate solution becomes overly diluted, reducing the alkalinity and ion concentration of the system, which slows down the cementitious reactions. Moreover, the excess free water leaves behind numerous pores after evaporation, causing a decrease in the compactness of the cementitious matrix and a looser structure, leading to a significant reduction in the elastic modulus. Solidification characteristics of bearing capacity Figure 9 presents the statistics on the UCS with different mix ratios at 7 day and 14 day. The strength in the three test groups increased with prolonged curing time and exhibited a trend of first increasing and then decreasing with variations in the dosage of each factor, demonstrating clear peak characteristics. Notably, within each test group, the specimen achieving peak strength at both 7 days and 14 days had the same mix proportion, indicating that this proportion possesses optimal cementation performance and structural stability in both the early and intermediate stages. Distribution characteristics of UCS of fluidized stabilized soil. (a) Water content, (b) Sodium silicate solution, (c) Curing agent. (1) Influence of curing age on unconfined compressive strength. As seen in Fig. 9, the UCS growth rates between 7 and 14 days for the water content test group were 96.63%, 46.55%, 42.25%, 21.52%, and 22.43%, respectively. This pattern clearly reflects the dual effects of water in the cementation process36. When the water content is relatively low (W48.00), the initial moisture is insufficient to fully support cement hydration and the pozzolanic reaction of fly ash, leading to delayed early-age strength development. However, as curing progresses, the remaining water continues to participate in the reactions, resulting in the continuous accumulation of cementitious products and showing higher potential for strength gain at later ages. Conversely, at higher water contents (W57.25, W61.00), the early-age moisture is ample and the reaction rate is faster, leading to relatively higher early strength. Yet, the evaporation of excess water leaves behind more pores, and the water available for sustained reactions in later stages is relatively limited. Consequently, the magnitude of strength gain significantly slows down, indicating insufficient potential for long-term performance development. This suggests that the beneficial effect of curing age on strength is most pronounced when the moisture condition is close to the optimal state. For the sodium silicate solution test group, the corresponding UCS growth rates were 17.31%, 32.09%, 45.99%, 20.07%, 13.53%, and 24.07%, respectively. The specimen with a 2% dosage (SS₂) exhibited the highest growth rate, indicating that it not only developed relatively high strength at early ages but also sustained an effective and ongoing cementitious reaction in later stages, leading to continuous structural densification and strength enhancement. When the dosage was below this level (SS₁), the activation effect was limited, resulting in an insufficient total amount of cementitious products, thus restricting the potential for long-term strength gain. When the dosage exceeded 2% (SS₃, SS₄), although the early-age activation effect was enhanced, the increased slurry viscosity37 and air content introduced more structural defects during the molding process. These defects not only reduced the early strength but also more severely hindered the subsequent formation and deposition of cementitious products, resulting in weak long-term strength development. It is noteworthy that specimen SS₅ maintained a certain growth rate even at a high dosage, suggesting that if initial molding defects could be overcome, the cementitious potential activated by the high-alkali environment could be partially realized. However, the defect effect often dominates, constraining its overall performance. Figure 9 also shows that the UCS growth rates in the curing agent test group were 40.82%, 35.94%, 38.27%, 20.07%, 38.74%, 24.07%, 25.63%, 35.42%, and 4.32%, respectively. When the content was below the optimal value (35%), the growth rates were generally high and showed a fluctuating upward trend. This indicates that within this range, the reaction system continuously generated cementitious products with the increase of cementitious materials, and the structure was progressively strengthened with age. Particularly within the CA25 to CA₃₅ range, the growth rates remained at a high level, reflecting sufficient cementitious reactions and stable structural development during this stage. However, when the content reached 40%, the growth rate plummeted to 4.32%, indicating nearly stagnant development of the material’s later-age performance. Based on data and mechanistic analysis, excessive curing agent content severely reduced the fluidity of the slurry, leading to decreased pouring compactness and increased initial porosity. The relative shortage of water within the system prevented some cementitious particles from fully hydrating, instead forming weakly cemented agglomerates. These factors collectively undermined the continuity and uniformity of the cementitious reaction, causing a significant decrease in the potential for long-term strength gain. (2) Influence of curing agent content on unconfined compressive strength. Figure 9 shows the variation trends of UCS at different ages, and the rates of strength change reveal the non-linear characteristics of the material’s performance response. As indicated in Fig. 9 for the water content influence, the rates of strength change at 7 days were 225.84%, −11.03%, −13.57%, and − 4.04%, respectively. At 14 days, they were 142.86%, −13.65%, −26.16%, and − 3.32%, respectively. It can be observed that before reaching the optimal dosage, only a small increase in water content could significantly enhance early strength, indicating the crucial role of appropriate water in promoting the early hydration of cementitious materials and ensuring uniform slurry distribution. However, when the water content continued to increase beyond the critical value, the strength showed a slow declining trend, and the potential for later-age strength gain was significantly constrained. This is mainly because excess water leads to a looser slurry structure and increased initial porosity, weakening the filling effect of the cementitious materials. Furthermore, the pores left behind after the evaporation of surplus water further compromise the compactness of the matrix, thereby limiting the continued development of strength. Regarding the influence of sodium silicate solution content, Fig. 9 shows that the rates of UCS change at 7 days were 3.08%, 7.09%, −2.79%, −4.66%, and 1.5%, respectively. At 14 days, they were 16.07%, 18.36%, −20.05%, −9.85%, and 10.93%, respectively. The test results indicate that a 2% dosage achieves the optimal balance between the activation effect and construction workability. An appropriate amount of sodium silicate solution can effectively activate the fly ash and promote the generation of cementitious products while avoiding casting defects such as air voids caused by excessive viscosity. Before reaching the optimal dosage, the UCS continuously increases with increasing dosage, and the magnitude of increase gradually expands, suggesting that the contribution per unit dosage is progressively enhanced. However, when the dosage is excessively high, the later-age strength development of the specimens is suppressed, indicating that density defects caused by thickening significantly affect the long-term stability of the structure38. Furthermore, for the same magnitude of dosage change, although the strength development of specimens with lower dosage is slower, their activation effect is still superior to that of specimens with excessively high dosage. This reflects the importance of rationally controlling the activator dosage for maintaining the material’s long-term performance. Figure 9 further reveals that the influence of curing agent content on strength development is the most significant. The rates of strength change at 7 days were 30.61%, 26.56%, 37.04%, 44.14%, 99.08%, and − 49.22%, respectively. At 14 days, they were 26.09%, 28.74%, 37.5%, 30.52%, 114.93%, and − 60.88%, respectively. When the content did not exceed the optimal value, the strength not only continued to increase with higher content, but the rate of increase showed an expanding trend. This indicates that the cementitious contribution per unit dosage gradually enhanced. The mechanism lies in the progressive formation of a continuous and dense skeletal structure by the cementitious materials, significantly improving the cementation and stabilization effect on soil particles. However, when the content exceeded the optimal value (40%), the strength experienced a cliff-like decline. Moreover, the magnitude of decline at 14 days was noticeably greater than at 7 days, suggesting that excessive curing agent not only reduces the absolute strength but also severely inhibits later-age strength development. This is primarily due to the imbalance in the water-to-solid ratio caused by the excessive curing agent, leading to insufficient and non-uniform cementitious reactions in some areas and the formation of weakly cemented agglomerates. This disrupts the structural continuity and compactness of the matrix, resulting in substantial strength deterioration. Through systematic analysis of the macroscopic mechanical properties of fluidized stabilized soil, the influence patterns and underlying mechanisms of each material component on strength development can be clarified. The optimal water content is 51.25%, which serves as a critical demarcation point for material strength development. Below this value, insufficient water limits cementitious reactions, preventing the cementation effect from being fully realized; above this value, the evaporation of excess water leaves pores, leading to a looser structure and decreased strength. The optimal sodium silicate solution content is 2.0%, demonstrating the best strength development across all ages. This dosage achieves a balance between fully activating the fly ash and avoiding excessive slurry viscosity and increased defects, effectively promoting the generation of amorphous aluminosilicate gels and enhancing cementation and filling effects. The optimal curing agent content is 35%. At this level, the total amount of cementitious material is sufficient to form a continuous and dense spatial network skeleton, adequately cementing soil particles and filling pores, endowing the material with both high strength and high stiffness, and achieving the optimal mechanical state. Therefore, under the condition of a 1:1 mass ratio of fly ash to cement, the recommended mix proportion is water content 51.25%, sodium silicate solution 2.0%, and curing agent 35%. This proportion unifies reaction sufficiency and structural compactness, providing a key basis for engineering applications. Macro- and micro-structure analysis Distribution patterns of macroscopic cracks To systematically compare the differences in failure morphology among specimens with different mix ratios, a unified qualitative characterization was performed on the surfaces of the failed specimens. Photographs summarizing the typical failure morphologies of the test groups are compiled in Fig. 10 reveals the macroscopic failure morphology and crack distribution characteristics of the water content test group. With the increase in water content, the macroscopic surface morphology of the specimens showed regular changes. The average pore size of harmful surface voids increased, and the specimen color gradually lightened, indicating a decrease in water retention capacity and an increase in drying. This primarily stems from the fact that excess free water more easily forms and entraps larger-sized air voids during mixing and setting. This not only increases the scale of initial defects but also provides rapid channels for water migration, thereby accelerating water loss. In terms of failure mode, all specimens formed distinct shear failure zones, but their specific morphologies differed. By comparison, specimen W₄₈.₀₀ had fewer cracks, with the main crack obliquely intersecting the loading direction and a length of 97.80 mm, exhibiting typical plastic shear failure characteristics. Specimen W51.25 had numerous cracks, with the main crack completely penetrating the specimen, a maximum width of 1.2 mm, and surface spalling, presenting a brittle tensile failure morphology. The crack characteristics of specimens W54.50 to W61.00 were between these two, showing a transitional plastic-brittle character. Among them, the main crack of W₅₄.₅₀ penetrated a surface void to form a through crack (maximum crack width 0.50 mm, maximum void length 19.80 mm). W57.25 had a main crack length of 67.50 mm and a large surface void measuring 30.50 mm in length; and W61.00 had a main crack length of 64.50 mm, a maximum width of 9.80 mm, and a maximum void size of 36.70 mm. Fracture modes of fluidized stabilized soil specimens. Combined with the strength analysis in Fig. 9(a), the strength peaked at a water content of 51.25%. Its underlying mechanism can be explained from both structural and reaction perspectives: When the water content is low (W48.00), although the initial porosity is lower and the darker color indicates better water retention, insufficient water restricts the full progression of cementitious reactions, leading to inadequate generation of cementitious products and a limited skeletal filling effect, hence the lower strength. When the water content is excessively high (W54.50, W61.00), although ample water facilitates the dissolution and migration of active components like SiO2, Al2O3, and CaO, promoting cementitious reactions that consume free water, the excess water leads to an increase in the number and size of air void defects in the mixture, significantly reducing the initial compactness of the material. Even with continued generation of cementitious products at later stages, their filling and cementing effects are still difficult to fully compensate for the structural degradation caused by the excessively high initial porosity, ultimately resulting in strength reduction. Therefore, the 51.25% water content essentially represents the optimal balance point between the water required for cementitious reactions and the demand for structural compactness. Under this condition, the material can ensure thorough reaction progression while minimizing harmful defects to the greatest extent, thereby achieving optimal mechanical performance. Figure 10 shows the macroscopic failure morphology and crack distribution characteristics of the sodium silicate solution test group. Specimens SS₁ and SS₂ exhibited a lower degree of drying, indicating slower internal moisture evaporation, ongoing cementitious reactions, and potential for later-age strength growth. Regarding failure mode, SS₀ and SS₁ had fewer cracks, with the main crack obliquely penetrating the specimen. Their maximum widths were 0.50 mm and 0.70 mm, respectively, showing plastic shear failure characteristics. The maximum surface void diameters for these two were 5.50 mm and 5.30 mm, respectively, reflecting that at zero or low sodium silicate content, the slurry viscosity is low, air bubbles are easily expelled but cementitious reactions are slower, product filling is insufficient, and the structure exhibits some plasticity but poor uniformity. As the sodium silicate content increased, the failure mode changed significantly. SS₂ and SS₃ had an increased number of surface cracks, with main crack widths reaching 3.30 mm and 1.50 mm, accompanied by surface spalling, showing brittle tensile failure. An appropriate amount of sodium silicate significantly promotes the pozzolanic reaction of fly ash, generating a large amount of cementitious products, increasing slurry viscosity and early strength, but brittleness also increases accordingly. The maximum void diameter of SS₂ decreased to 3.20 mm, indicating an improved filling effect, but rapid setting might hinder complete air expulsion, forming local defects. When the dosage was further increased to SS4 and SS5, the slurry viscosity increased sharply, severely impairing fluidity and compaction effectiveness. Their maximum surface void diameters reached 6.30 mm and 13.90 mm, respectively, and the main crack widths increased to 7.80 mm and 12.10 mm. Cracks propagated along the edges of voids and penetrated them, with the failure mode being between plastic shear and brittle tensile. Comparing SS2 and SS5 shows that although a high sodium silicate dosage further activates cementitious reactions, the drastic thickening effect leads to deteriorated fluidity, introducing more air bubbles and defects, and reducing structural uniformity. Notably, the strength of SS4 was lower than that of SS5, possibly because it was within a specific viscosity range where fluidity was poorest, molding quality fluctuated more, and more internal air bubbles remained. This not only reduced compactness but also hindered thorough cementitious reactions around the voids. This chain mechanism of fluidity—molding quality—defect distribution—strength development explains the unique softening behavior and ductile characteristics exhibited by SS4 and SS5 on their stress-strain curves. Figure 10 reveals the influence pattern of curing agent content on the macroscopic failure morphology of specimens. As the curing agent content increased, the water-to-solid ratio of the system correspondingly decreased, and the specimens exhibited a progressively increasing drying trend. This phenomenon reflects the enhanced binding and consumption of free water by the cementitious materials. Looking at the macroscopic characteristics of the failure morphology, the crack development patterns of specimens with different contents showed significant differences. Low-content specimens (CA10, CA15) mainly exhibited plastic shear failure. Their main cracks intersected obliquely with the axial direction, with maximum widths of 1.10 mm and 0.20 mm, respectively. Crack propagation was relatively slow, showing certain ductile characteristics. Specimen CA10 had a visible surface void with a maximum diameter of 7.70 mm, indicating that with insufficient cementitious material, the slurry’s filling ability and compactness were poor, and pore defects were evident. In contrast, high-content specimens (CA35, CA40) exhibited typical brittle tensile failure: the main cracks completely penetrated the specimens, with maximum widths of 2.1 mm and 7.2 mm, respectively. The failure process was accompanied by surface debris spalling, and CA40 even fractured entirely into two parts along the main crack. Although the maximum surface pore diameter of CA40 was only 2.9 mm, the number of harmful voids increased significantly, reflecting poorer internal structural uniformity. The failure characteristics of intermediate-content specimens (CA20, CA25, CA30) fell between the two types described above. Their main crack lengths were 94.5 mm, 91.5 mm, and fully penetrating (100 mm), with maximum widths of 0.70 mm, 0.50 mm, and 3.20 mm, respectively. The main crack of CA20 passed through a large void with a diameter of 17.80 mm, indicating that local defects guided the crack propagation path. From the macroscopic phenomena, it can be further inferred that specimens CA₁₀ and CA₂₀ exhibited a lower degree of surface drying, indicating their higher water-to-solid ratio and relatively abundant free water content. However, the degree of cementitious reaction was insufficient, and structural solidification was limited, hence they exhibited certain plastic deformation capacity. Although specimen CA40 had the highest curing agent content, its excessively low water-to-solid ratio resulted in poor fluidity of the fresh slurry, making it difficult to achieve full compaction during pouring and leading to the formation of more pores and defects. These initial defects became stress concentration points, significantly reducing material strength and limiting its potential for growth with age, ultimately resulting in pronounced brittleness and decreased load-bearing capacity. In contrast, specimen CA35 had the driest surface and good compactness, indicating that at this content, the ratio of cementitious material to water reached the optimum, hydration and activation reactions were sufficient, forming a uniform and dense microstructure. Consequently, it exhibited high strength, high stiffness and typical brittle failure characteristics macroscopically. The above patterns indicate that the curing agent content not only affects the final strength but, by altering the material composition and hydration environment, fundamentally determines its failure mode and structural performance. Characteristics of micro- and meso-scale crack distribution The SEM coupled with EDS can reveal the structural characteristics and solidification mechanism of FSS at the micro scale. EDS analysis of typical areas on the specimens provides the elemental composition of local micro-regions. The main elemental contents of representative specimens are summarized in Table 4.To clarify the influence of sodium silicate solution content on the material’s microstructure, representative specimens SS2 (2% content, appropriate) and SS5 (5% content) were selected for comparative analysis. Figure 11 shows that the interior of specimen SS₅ is dominated by relatively large, clearly outlined agglomerated crystals. While this indicates a high degree of polymerization reaction, evident gaps exist between the crystals, resulting in insufficient structural continuity. EDS analysis of the agglomerated crystals indicates that specimen SS₅ is primarily composed of elements such as O, Si, Al, and Ca, suggesting that the cementitious products are mainly calcium-aluminosilicate hydrate (C-A-S-H) gel. In contrast, the microstructure of specimen SS₂, besides containing some agglomerated crystals, is more notably characterized by a large number of fine crystals uniformly adhering to particle surfaces and effectively filling the inter-particle pores, forming a denser and more homogeneous overall structure. Table 4Elemental composition from EDS spectra of typical specimens.ElementSpecimen SS₅Specimen CA₁₀O50.1039.70Si17.9025.40Al8.107.70Ca8.0015.30Fe2.706.10Mg2.302.30K1.402.1 Elemental composition from EDS spectra of typical specimens. Influence of sodium silicate on micro-cracks. (a) SS2, (b) SS5. Furthermore, the compressive strength of specimen SS₂ is significantly better than that of specimen SS₅. This indicates that the compactness of hardened FSS does not have a simple negative correlation with the fluidity of the fresh slurry. The pursuing high-strength solidified soil does not necessarily require sacrificing the slurry’s workability. From the apparent state, the drier surface of specimen SS5 suggests faster internal moisture migration. The relatively moister surface of specimen SS2 indicates a lower water loss rate. These phenomena can be explained by the microscopic mechanism. An appropriate amount of sodium silicate solution effectively promotes cementitious reactions. During the reaction process, a considerable amount of free water is consumed and transformed into cementitious products that fill the pores, forming a dense microstructure. This dense pore system retards the physical process of water migrating outward, reducing the water evaporation rate. Therefore, the degree of surface dryness of the specimens is essentially the combined result of physical water migration and chemical water consumption. To analyze the influence pattern of curing agent content on the material’s physical properties, representative specimens CA10 (low content) and CA35 (appropriate content) were selected for microscopic comparison, as shown in Fig. 12. It can be seen that specimen CA10 has relatively large internal pores, loosely distributed flaky crystals, and few needle/rod-shaped crystals, resulting in a loose overall structure with poor compactness. In contrast, the interior of specimen CA35 is densely filled with flaky, agglomerated, and a large number of needle/rod-shaped crystals, with pores significantly reduced and compactness markedly improved. EDS analysis of the laminated crystals in CA₁₀ also detected elemental compositions dominated by O, Si, Al, and Ca, indicating they are primarily hydrated aluminosilicate gels. Based on hydration reaction mechanisms, the needle/rod-shaped crystals can be identified as ettringite (AFt). The improvement in compactness for CA35 stems from two aspects: firstly, the physical filling and compaction effect brought by the increased curing agent content; secondly, the participation of more curing agent in hydration reactions, generating a large amount of cementitious products that further fill the inter-particle pores. Influence pattern of curing agent content on micro-cracks. (a) CA10, (b) CA35. Combined with macroscopic analysis, specimen CA35 exhibits fewer surface bubbles and appears drier compared to specimen CA10. It can be inferred that CA35 possesses a superior particle gradation and curing system. This enables the mixture, under a fixed water content, to have better fluidity and stability, thereby reducing air entrainment during pouring and increasing molding compactness. Simultaneously, the favorable gradation and reaction environment promote the participation of more free water in cementitious reactions, generating more hydration products. This is the fundamental reason why the compressive strength of CA35 is significantly higher than that of CA1038. This further demonstrates that by optimizing the curing agent content, it is possible to enhance material strength while improving its microstructure and macroscopic performance. Figure 12 also indicates that although specimen CA35 has a denser structure and theoretically should have a lower water migration rate, its expected apparent drying degree should follow a pattern similar to that of specimen SS2. However, actual observation shows that CA35 exhibits a more pronounced apparent drying characteristic. This phenomenon reflects that during material hardening, the mechanism of water migration and consumption has transitioned from being dominated solely by physical diffusion to a complex stage where physical migration and chemical water consumption act together. Figure 13 illustrates the relationship between physico-chemical actions and free water, which can well reveal the solidification mechanism of the curing agent on fluidized soil. With continuous water loss, the system gradually transitions from a saturated state to an unsaturated state and finally reaches the shrinkage limit state, where volume shrinkage tends to stabilize. Solidification mechanism of curing agent on fluidized soil. For specimen CA35, the sufficient filling of pores by cementitious products forms a dense microstructure, which significantly narrows the physical migration paths for water, increases resistance, and substantially reduces the residual free water content. This leads to a sharp deceleration in the physical loss rate. However, the ongoing hydration and pozzolanic reactions continue to consume the remaining free water, making chemical water consumption gradually the dominant mechanism for water reduction at this stage. While the dense structure inhibits physical migration, it does not delay the continuation of cementitious reactions. Due to the sufficient total amount of cementitious materials and a stable reaction environment, the process of chemical water binding becomes more thorough/complete, thereby manifesting as a more apparent dry state externally. This evolution process of water migration and consumption mechanisms as illustrated in Fig. 13, revealing the synergistic influence of microstructural densification and sustained deepening of chemical reactions on the material’s apparent state and internal moisture condition under high curing agent content. Mechanical properties of fluidized soil under optimized mix proportion To verify the engineering applicability of the optimal mix proportion determined based on single-factor tests, FSS specimens were prepared in this section according to the established proportions: a fly ash/cement mass ratio of 1:1, water content of 51.25%, sodium silicate solution content of 2.0%, and total curing agent content of 35%. The measured density of the fresh slurry was 1950 kg/m³, and its dynamic viscosity was 0.24 kg/(m·s). Using the cylinder method, its spread flow was determined to be 175 mm, indicating that the fluidized soil material possesses good fluidity, meeting the requirements for pumping construction. Mechanical testing showed that the fluidized soil essentially reached initial set within 4 h after molding, and the unconfined compressive strength at 14 days of curing reached 4.53 MPa, demonstrating high early strength. To further investigate the formation mechanism of the mechanical properties, macro- and micro-characteristics of the specimen with the optimal mix proportion were analyzed as shown in Fig. 14. Macro- and micro-characteristics under optimized mix proportion. From the macroscopic failure morphology in Fig. 14, it can be observed that the specimen surface is dense and free of significant harmful voids, indicating that the FSS under the optimal mix proportion possesses good self-compacting ability. The specimen exhibited a high degree of surface dryness, reflecting the rapid progression of the solidification reactions. The failed specimen displayed typical brittle tensile failure characteristics: the crack essentially penetrated the entire specimen with a conical fracture surface, and significant fragment spalling occurred on the fractured surface. This indicates that the fluidized soil material rapidly loses its bearing capacity after reaching peak strength. From the microstructure shown in Fig. 14(b), it can be seen that the surfaces of fly ash particles are partially exposed and partially covered by cementitious products, indicating varying degrees of participation in the pozzolanic reaction. Cementitious products generated on the surfaces of the reacted fly ash gradually fill the surrounding pores and bond adjacent particles, effectively enhancing the skeletal compactness and integrity. Furthermore, a large number of needle/rod-shaped ettringite (AFt) crystals aggregate to form cluster-like structures39,40, playing a bridging role between particles and further strengthening the connectivity of the microscopic skeleton. The above macro- and micro-characteristics collectively reveal the solidification mechanism of the material under the optimal mix proportion. The curing agent particles effectively fill the spaces between soil particles, improving the particle gradation of the system through a micro-aggregate effect and enhancing inter-particle friction and interlocking. Activated by the sodium silicate solution, cement hydration and the fly ash pozzolanic reaction proceed synergistically, generating multi-form cementitious products such as C-(A)-S-H gel and AFt. On one hand, these products adhere to particle surfaces and fill pores, significantly increasing structural compactness and interfacial bond strength. On the other hand, their formation process also consumes water, influences slurry fluidity, and to some extent regulates the pathways and rate of water loss. Therefore, the final compressive strength of the material is essentially the result of the combined action of fluidity, molding compactness, and water loss behavior, which is the macroscopic manifestation of microstructural evolution. It is noteworthy that if air bubbles entrained in the slurry during pouring cannot be expelled promptly, structural defects will also form, leading to reduced strength41. Consequently, based on material proportion optimization, it is necessary to further integrate its rheological properties, conduct numerical simulations of the pouring process, and optimize construction parameters to achieve synergistic enhancement of material performance and process control. Discussion Air-liquid two-phase model This project employs precast utility tunnel segments, each 3 m in length. The construction plan aims to pour six segments per side per day, with only one pouring inlet on each side, and a preliminary pouring speed set at 4 m/s. Limited by the mixing tank capacity, the maximum length for a single continuous pour does not exceed three segments. Although FSS possesses self-leveling properties, the choice of pouring process directly impacts construction quality and efficiency42,43. Therefore, optimizing the pouring process hinges on reasonably determining the single-pour length and the pouring speed. This section will first determine the single-pour length and then further analyze the suitable range for the pouring speed. To avoid potential mixture leakage associated with simultaneous excavation and pouring, the construction adopts a sequence of first excavating to the predetermined length, followed by pump-backfilling. The narrow backfill zone is flanked by the excavation pit on one side and the installed utility tunnel on the other. The relative displacement between these boundaries and the FSS during pouring is negligible; therefore, deformation of the free surface of the backfill zone is ignored in the numerical simulation. Geometric modeling was performed using ICEM CFD. The backfill zone has a width of 0.30 m and a height of 4.50 m, with an angle of 77.5° between the inclined side and the bottom base. For the single-segment, double-segment, and triple-segment backfilling scenarios, the lengths were set to 3 m, 6 m, and 9 m, respectively. The pump pipe radius is 0.10 m, with the pouring inlet located at the center of the top surface. The mesh quality for all cases met the requirement of quality ≥ 0.49, with mesh counts of 36,290, 44,608, and 60,118 for the three scenarios, respectively. The schematic of the air-liquid two-phase model is shown in Fig. 15. Air-liquid two-phase model of fluidized soil. According to Ref. [14], the FSS during the pouring stage can be reasonably described as a Bingham plastic fluid. In addition to the measured density and dynamic viscosity, the yield stress is another key rheological parameter44. Based on Ref. [14] and the empirical relationship between spread flow and rheological parameters, the yield stress was set to 320 Pa. The process of backfilling the narrow zone with FSS essentially involves injecting a liquid phase into a space initially filled with air, constituting a typical air-liquid two-phase flow problem that requires a multiphase flow model for simulation45,46. This study employs the VOF (Volume of Fluid) model within ANSYS Fluent, setting air as the primary phase and FSS as the secondary phase. Considering that the maximum duration of a single continuous pour is about 96 s, within such a short time the viscosity decrease due to thixotropy is not significant. Therefore, to simplify the calculation and focus on the air-liquid interaction during pouring, the rheological properties of FSS are assumed to be time-invariant, neglecting its thixotropy. The impact of this simplification on the simulation results is analyzed as follows: thixotropy mainly manifests as viscosity recovery (thickening) after mixing when the slurry is at rest. However, during the pouring process, the slurry is continuously disturbed and remains in a shear state, so the thixotropic effect is actually weak. Neglecting thixotropy may lead to a slight overestimation of the early-stage bubble escape capacity (because the actual viscosity is slightly higher than the model value), but due to the extremely short pouring time, the deviation is within an acceptable range. To simplify the calculation and focus on the air-liquid interaction during the pouring process, the rheological properties of the FSS during the pouring stage are assumed to be time-invariant, neglecting its thixotropy. The surface tension model for the air phase is enabled to simulate the formation of air bubbles due to vortex-induced entrapment during liquid injection. To ensure stable and convergent calculations, the volume fraction discretization is set to the explicit scheme. The boundary conditions are set as follows: the inlet is defined as a velocity inlet, with velocities ranging from 2 to 6 m/s, the outlet as an outflow, and all other walls as stationary walls. For initialization, the air volume fraction within the backfill zone is set to 1. A time step of 0.01 s is used with a maximum of 20 iterations per time step, after which the solution calculation proceeds. Analysis of numerical results Air entrapment characteristics considering pouring length To investigate the influence of pouring length on air entrapment behavior, this study considered three typical working conditions: single-pour for one utility tunnel section (Condition 1), continuous pouring for two utility tunnel section (Condition 2), and continuous pouring for three utility tunnel section (Condition 3). A comparative analysis was conducted by extracting air volume fraction contours, streamline plots, pathline plots, and central pressure profile plots from the simulation results, as shown in Fig. 16. It can be seen that air entrapment regions under all three conditions are concentrated near the pouring inlet, at the bottom of the narrow backfill zone, and on the upper surface, but no through-hole voids are formed. Notably, significant air bubble accumulation exists in the bottom corners of the backfill zone for conditions 1 and 2, whereas this phenomenon is essentially absent in condition 3. Patterns of air entrapment caused by different pouring lengths. (a) One utility tunnel section, (b) Two utility tunnel section, (c) Three utility tunnel section. Regarding the characteristics of streamline distribution, in Condition 1, the streamlines on the right side rise closely along the wall and gradually converge towards the pouring inlet, resulting in insufficient filling and lower compactness in the top-right corner area. The streamlines on the left side exhibit higher dispersion during the ascent, flowing through the top-left corner, indicating better compactness in that region. However, as the streamlines fail to cover the bottom-left corner, the fluid there is nearly stagnant. Although air bubbles are subjected to buoyancy, they lack sufficient flow driving force to effectively escape, ultimately leading to entrapment. With increasing pouring length (Conditions 2 and 3), the flow resistance along the path increases. The ability of the streamlines to fill the edges of the backfill zone during their upward movement gradually weakens, causing the air content on the surface to show an increasing trend with the length of the backfill zone. Further analysis of the mechanisms of vortex and pressure distribution reveals that the area marked by the red circle in the figure is the impact vortex zone near the pouring inlet. This region features high flow velocity and forms a distinct recirculation vortex. The strong shear forces at the periphery of the vortex can entrain air bubbles and carry them into the vortex core. Due to the small pressure gradient within the vortex core, air bubbles entering it lack sufficient escape driving force, forming an air trap, resulting in the highest air volume fraction in this area. The pressure profile shows that the isobars within the red-circled area exhibit a non-uniform distribution, convex upward and concave downward. Above the air bubbles, the high velocity and low static pressure cause local upward convexity of the isobars. Below the bubbles, the low velocity and higher static pressure, combined with the internal pressure of the bubbles, create a localized high-pressure zone, leading to a corresponding downward concavity of the isobars. This distribution is closely related to capillary pressure effects47 and the Bernoulli principle48. That is, the pressure inside an air bubble is higher than the static pressure of the surrounding slurry, and the faster the fluid velocity, the lower its static pressure. In the bottom corner regions, due to the low air fraction and smaller flow velocity, the aforementioned pressure difference effects are not pronounced; therefore, the convex-concave characteristics of the isobars are only visible in magnified views. Pathline plots further indicate the existence of small-scale vortices in these regions, exacerbating air entrapment. It can be concluded that the pouring length significantly constrains the migration and escape behavior of air bubbles by influencing the flow field distribution, vortex structure, and pressure gradients, thereby affecting the final compactness and defect distribution of the molded body. The air volume fraction of the residual air within the trench at the final steady state was quantitatively calculated using the Volume Integral function in ANSYS Fluent. The variation pattern of residual air content under different pouring length conditions is obtained, as shown in Fig. 17. The calculation results show that the final residual air contents for condition 1, condition 2, and condition 3 are 1.69%, 1.04%, and 0.93%, respectively. It is evident that as the single-pour length increases, the residual air content in the molded body shows a decreasing trend. Condition 3 exhibits the lowest air content and the optimal compactness. This phenomenon can be explained by the flow field development and air transport mechanisms. A longer pouring segment facilitates the formation of a more continuous and stable flow front. The slurry fills the bottom and corners of the narrow backfill zone more fully, reducing areas not covered by streamlines, thereby lowering the probability of air entrapment in corners. Particularly in condition 3, there is essentially no air accumulation in the bottom corners. This is beneficial for avoiding soil damage during demolding caused by weak strength at the corners, ensuring the structural integrity of the backfill material. Furthermore, from the perspective of construction efficiency, condition 3 can reduce the number of pours while achieving similar compactness, thereby shortening the waiting/setting time between construction segments. This is conducive to accelerating the overall construction progress and improving project economy. Therefore, this study recommends adopting the single-pour length covering three utility tunnel sections as the optimized pouring scheme. This approach ensures molding quality while balancing construction efficiency and safety controllability. Statistical relationship between residual air content and pouring progress. Air entrapment characteristics considering pouring speed To investigate the effect of pouring speed on air entrapment behavior, the filling processes under two typical flow velocities, 6 m/s (Condition 4) and 2 m/s (Condition 5), were further simulated. Calculations show that the final residual air volume fractions under steady-state conditions are 1.96% and 0.75% for Conditions 4 and 5, respectively, indicating that increasing the flow velocity significantly increases residual air content. Compared to pouring length, flow velocity has a more pronounced impact on compactness. Figure 18 summarizes the air volume distribution, streamlines, and vector plots for conditions 3, 4, and 5, visually demonstrating the differences in flow field characteristics and air distribution under different speeds. Pattern of air entrapment under different pouring speeds. Analysis reveals that in the low-speed condition (2 m/s), the air entrapment area is significantly reduced, with entrapped air mainly concentrated in the impact vortex zone near the pouring inlet and asymmetrically distributed. In the high-speed condition (6 m/s), air accumulation is not only more significant in the impact zone but also evident in the bottom-right corner of the narrow backfill zone. It is concluded that reducing the flow velocity allows streamlines to fully diffuse, covering most of the trench space, thereby effectively expelling air and leaving only a small amount of air bubbles in localized low-pressure areas. This indicates that appropriately lowering the pouring speed facilitates smoother slurry filling, reduces turbulence-induced air entrainment, and consequently improves the compactness of the molded material. Based on macroscopic observations during sample preparation and the results of numerical simulations, this paper proposes an air bubble retention mechanism for fluidized stabilized soil during the mixing and pouring processes. This mechanism primarily encompasses two key processes: air bubble generation and air bubble entrapment. Figure 19 is a schematic diagram illustrating this air bubble retention mechanism. During sample preparation, high-speed stirring increases the Reynolds number, causing the slurry to enter a turbulent state. Vortices of various scales within the turbulence act to entrain and tear air, forming air bubbles. According to the Bernoulli equation, the high-speed region at the tip of the stirrer blade creates a localized low pressure, further promoting air entrainment. For an air bubble to rise, it must undergo some deformation into a non-spherical shape, overcoming the capillary pressure that confines it within the pores. Surface tension is the fundamental physical cause of this capillary pressure. That is, the viscous forces of the FSS hinder the coalescence and rise of air bubbles, allowing these entrained fine air bubbles to remain stably dispersed within the slurry in the absence of external forces. When FSS impacts a bottom corner at a relatively high speed, it splits into a wall jet flowing upwards along the wall and a radial jet. The energy driving fluid rotation in the corner region is insufficient, resulting in low shear forces and pressure gradients in this area, making it a flow dead zone. Air bubbles become trapped in this weak-flow region, unable to be effectively carried away, and ultimately remain entrapped in the corner. Schematic diagram of air bubble retention mechanism. To further optimize construction parameters, and referencing the speed control range in actual engineering, an additional condition 6 with a pouring speed of 5 m/s was simulated, focusing on examining air bubble entrapment in the bottom corners during the pouring process, as shown in Fig. 20. When the pouring speed is 5 m/s, the slurry rapidly spreads after impacting the trench bottom. Although some turbulence is generated, stable clockwise corner vortices are simultaneously produced in both bottom corner regions. These vortices have a continuous sweeping effect, effectively carrying air from these areas towards the main flow region, thereby preventing air bubble entrapment in the corners49. Therefore, a pouring speed of 5 m/s not only ensures construction efficiency but also leverages the corner vortex effect to eliminate bottom air bubbles, meeting the dual requirements for compactness and construction controllability in engineering applications. Pattern of air entrapment under 5 m/s pouring speed. Field verification and implementation of process optimization The preceding analysis indicates that the incorporation of sodium silicate solution significantly affects the rheological properties of fluidized soil. Its accelerating effect can trigger a rapid increase in system viscosity shortly after addition. To balance construction operability and material workability, a two-fluid pumping process was adopted on-site: first, the curing agent, soil, and mixing water are thoroughly blended into a semi-finished product; then, just before pumping, the sodium silicate solution is added. After brief agitation, the mixture is immediately pumped into the narrow backfill zone. This process can delay the setting progression, effectively maintaining fluidity during the pumping phase, and avoid a sharp viscosity increase due to excessively long mixing time, thereby ensuring pouring continuity and filling quality. Figure 21 shows a schematic diagram of the field verification. Previous analysis has shown that when the filling ratio approaches 95%, the discharge of air from the cavity is hindered, and the number of residual air bubbles rises sharply. This indicates that during the final stage of pouring, the efficiency of the liquid phase displacing the air phase decreases, making it prone to form air bubble accumulation in corner and near-surface regions. Therefore, in actual construction, a “terminal velocity reduction” strategy needs to be implemented. That is, when the filling ratio reaches approximately 95%, the pumping speed is proactively reduced to extend the air discharge time and improve the final compactness. Schematic diagram of field verification. For the field construction, based on the optimized mix proportion, the total water addition was dynamically adjusted according to the actual moisture content of the excavated soil. A bilateral synchronous pumping process was employed to achieve a single continuous pour covering three utility tunnel sections. The pouring speed was increased to 5 m/s and was gradually reduced when the filling ratio reached around 95%. Performance tests were conducted on the field mixture. The fluidized soil achieved a spread flow of 440 mm, meeting the fluidity requirements for pumping. The setting times were 50 min for the initial set and 150 min for the final set, fitting within the construction process time window. According to the relevant standard, cube compressive strength tests were conducted, and the 7 day strength of the fluidized soil exceeded 1 MPa, satisfying the design strength index. Post-demolding observation revealed that the surface of the backfilled material in the narrow zone was smooth and dense, with complete corner filling and no signs of corner spalling or voids. In summary, solely reducing the injection speed or increasing the single-pour length has a limited effect on improving overall compactness but significantly influences the distribution pattern of air bubbles. By employing the delayed two-fluid mixing process to maintain fluidity, coupled with terminal velocity reduction pouring control, air entrapment in corners can be effectively reduced, preventing soil damage during demolding caused by stress concentration. This scheme, verified on-site, successfully enhanced construction efficiency while effectively ensuring the internal compactness and surface quality of the backfill material, achieving the goal of process optimization. Conclusions This study focused on the in-situ excavated loess from the Xiong’an New Area, addressing the technical requirements for its backfilling in narrow and confined foundation trenches using FSS. Systematic research was conducted on material proportion optimization, solidification mechanism analysis, and pouring process simulation. The main conclusions are as follows: (1) A fly ash-to-cement mass ratio of 1:1 is identified as the key benchmark for optimizing the composite curing agent proportion. Within the cement-fly ash composite curing system, a mass ratio of 1:1 is determined as the optimal proportion, balancing material performance, cost-effectiveness, and environmental friendliness. By systematically comparing the water loss behavior and strength development patterns of specimens with different cement-fly ash ratios (α = 2 and α = 1), the study found that specimens with α = 1 (equal mass ratio) exhibited a more gradual water loss rate, which is conducive to sustained hydration of the cementitious materials and long-term strength development. This proportion not only effectively inhibits excessively rapid early water evaporation but also provides a stable hydration environment for the later pozzolanic reaction of fly ash. (2) The optimal material mix proportion is 35% curing agent, 2.0% sodium silicate, and 51.25% water content, which combines high fluidity with high strength. The influence of water content, sodium silicate content, and total curing agent content on material fluidity, compactness, and strength was systematically studied through single-factor tests. A water content of 51.25% ensures thorough cementitious reactions and a dense structure. A sodium silicate solution content of 2% achieves a balance between activating fly ash and avoiding excessive slurry viscosity. A curing agent content of 35% facilitates the formation of a continuous and dense cementitious network. Under this proportion, the material achieved a spread flow of 175 mm and a 14-day compressive strength of 4.53 MPa, meeting the requirements for pumping construction and strength. (3) The compressive strength of FSS is the synergistic result of fluidity, molding compactness, and water loss behavior, manifested microscopically by the generation of cementitious products, pore filling, and particle cementation. Macroscopically, the apparent drying degree of specimens reflects the coupling effect of physical water migration and chemical water consumption. Microscopic SEM-EDS analysis reveals that appropriate amounts of sodium silicate and curing agent promote the formation of C-(A)-S-H gel and ettringite, effectively filling pores and cementing soil particles. Excessively high or low component dosages lead to increased pore defects and reduced structural continuity, thereby affecting strength and failure mode. (4) Adopting a single-pour length covering three utility tunnel Sect. (9 m), a pouring speed of 5 m/s combined with a terminal speed reduction process can significantly enhance the compactness and construction quality of the backfill in narrow trenches. Increasing the single-pour length facilitates the formation of a stable flow field and reduces air bubble retention in corners. Appropriately reducing the pouring speed generates a corner vortex effect, promoting air expulsion. Combined with the two-fluid pumping process to delay setting and maintain fluidity and reducing speed at the final pouring stage to extend air expulsion time, field application demonstrates that this process can achieve a backfill with a smooth surface, intact corners, and no voids, meeting the dual requirements of construction efficiency and quality. Acknowledgements The authors thank the Science and Technology Innovation Team of Shaanxi Innovation Capability Support Plan (No. 2020TD005), the 2025 Jiangsu Province Young Science and Technology Talent Support Project (No. JSTJ-2025-319), the Basic Science Research Project of Higher Education Institutions in Jiangsu (No. 24KJB560024), and the 2023 High-level Talent Research Project from Yancheng Institute of Technology (No. xjr2023019). Author contributions Hongjun Zhang: Methodology, Experimental design. Dapeng Han: Experimental design, Validation, Writing–original draft, Writing–review & editing. Bo Hua: Investigation, Data curation. Yun Cheng: Methodology, Writing–review & editing.Shubo Sui: Validation, Writing–original draft. Xiangpan Jiao: Writing–review & editing. Zhiwei Xu: Methodology, Data curation, Writing–review & editing. Binbin He: Investigation, Data curation, Writing–review & editing. Lingyi Chen: Methodology, Experimental design, Validation, Writing–review & editing. Ruoqi Hu: Methodology, Data curation, Writing–review & editing. Zhanping Song: Methodology, Resources, Funding acquisition, Experimental design, Writing–review & editing. Funding This study was supported by the Science and Technology Innovation Team of Shaanxi Innovation Capability Support Plan (No. 2020TD005), the 2025 Jiangsu Province Young Science and Technology Talent Support Project (No. JSTJ-2025-319), the Basic Science Research Project of Higher Education Institutions in Jiangsu (No. 24KJB560024), the General Program of Jiangsu Province Natural Science Foundation (BK20251935), and the 2023 High-level Talent Research Project from Yancheng Institute of Technology (No. xjr2023019). Data availability The data used to support the findings of this study are available from the corresponding author upon reasonable request. Declarations Competing interests The authors declare no competing interests. Footnotes Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. Contributor Information Dapeng Han, Email: divebomb130929@163.com. Yun Cheng, Email: chengyun_ifu@163.com. Zhanping Song, Email: songzhpyt@xauat.edu.cn. References Associated Data Data Availability Statement The data used to support the findings of this study are available from the corresponding author upon reasonable request.