Development, Performance, and Mechanism of Fluidized Solidified Soil Treated with Multi-Source Industrial Solid Waste Cementitious Materials

Abstract

Insufficient utilization of industrial solid waste and the high carbon emissions caused by the use of cement in engineering construction are two challenges faced by China. This study aimed to develop a multi-source industrial solid waste cementitious material (MSWC) for fluidized solidified soil (FSS) in soil backfill projects. First, the response surface models for the unconfined compressive strength (UCS) of MSWC-FSS were established, and the optimal mixing ratio of MSWC was determined. Subsequently, laboratory tests were conducted to compare the differences in flow expansion, UCS, and dry shrinkage between MSWC and ordinary Portland cement (OPC) in FSS, and the feasibility of MSWC-FSS was verified through on-site tests. Finally, the curing mechanism of MSWC-FSS was analyzed by XRD and SEM. The results showed that MSWC had an optimal mix ratio: steel slag (SS): ground granulated blast-furnace slag (GGBS): circulating fluidized bed fly ash (CFBFA): flue gas desulfurization gypsum (FGDG): OPC = 20:40:15:5:20. MSWC-FSS had good flow expansion, and its UCS and drying shrinkage resistance were more than 10% better than OPC-FSS. The on-site test also proved the practicability and progressiveness of MSWC-FSS. According to the chemical composition and microstructure, MSWC-FSS generated more ettringite than OPC-FSS, making MSWC-FSS denser.

1. Introduction

There are a large number of soil backfilling projects in the foundation construction of civil engineering. In certain backfill areas, such as bridge abutment back, retaining wall back, foundation pits, and trenches, there are challenges of limited and complex workspace, as well as large backfill depths. In these projects, traditional compaction techniques fail to ensure engineering quality and subsequent usability, resulting in issues like insufficient compaction and severe structural settlement [1].

Fluidized solidified soil (FSS) offers an effective solution to these engineering problems. FSS is a novel backfill material with characteristics such as self-compaction and high fluidity [2]. FSS can be pumped or chute-poured without the need for vibration, reducing the compaction steps (rolling, tamping) required for traditional soil and lime backfills, which is especially advantageous for backfilling in narrow spaces and provides irreplaceable technical benefits [3,4].

Portland cement has been the most widely used cementing material in FSS [5]. However, the production of cement consumes a large amount of energy and non-renewable resources and emits high levels of carbon dioxide, accounting for 15% of industrial CO₂ emissions [6,7]. In addition, cement often has a poor solidification effect on certain soils, especially in high moisture content [8,9,10]. Moreover, the drying shrinkage performance of FSS treated with cement is poor, and it is prone to drying shrinkage cracks, which seriously affects the durability of FSS [11,12]. Therefore, there is an urgent need to develop new environmentally friendly and effective cementitious materials suitable for FSS.

At the same time, China is facing severe pressure in the treatment of industrial solid waste. Currently, the cumulative stockpile of major solid waste in China is 60 billion tons, with an annual increase of about 3 billion tons [13,14]. Particularly, China leads the world in steel slag production, yet the utilization rate is low, with large stockpiles and a resource utilization rate of only about 20% [15]. The massive accumulation and landfilling of industrial solid waste occupy substantial land, severely polluting soil, groundwater, and air. The industrial solid wastes contain active components of CaO, SiO₂, and Al₂O₃, which can undergo pozzolanic reactions and have potential hydration activity, making them suitable as cementitious materials to replace OPC in road engineering [16,17,18]. Therefore, it is urgent to accelerate the application of industrial solid wastes as binders to enhance their resource utilization.

To address the aforementioned issues, the development of solid waste cementitious materials suitable for FSS has become a current research focus. Zhou et al. [1] used a composite binder consisting of rice husk ash, fly ash, and steel slag powder in a ratio of 1:6:3, with an additional 8% activator, to stabilize shale soil, comparing it with cement. Feng [19] utilized a composite solidification material composed of cement, fly ash, and phosphogypsum for the fluidized solidification of sludge, addressing the poor performance of cement-stabilized sludge. Zhao et al. [20] prepared a composite cementitious material for FSS using slag, municipal solid waste incineration bottom ash, gypsum, and cement, with a certain amount of NaOH and water glass additives, discovering certain advantages over traditional materials. Gu et al. [21] investigated the feasibility of preparing soil stabilizers using slag, carbide slag, cement, and so on for FSS; then, they also explored the adaptability of this type of FSS to various additives.

However, most studies only involve the simple application of one or a few types of solid waste, and the synergistic laws and mechanisms between multiple solid wastes are not clear, which is not conducive to fully utilizing the beneficial components of various solid wastes and the collaborative disposal of multiple solid wastes. In addition, there are currently only a small number of research reports on multi-source solid waste cementitious materials (MSWC) suitable for FSS, and very few have been validated on-site, resulting in unclear practical engineering application effects. In order to promote the large-scale utilization of industrial solid waste, a more comprehensive exploration of the impact of MSWC on FSS is needed.

To address these issues, this study aimed to develop an economical and environmentally friendly MSWC suitable for FSS, using four industrial solid wastes including steel slag powder (SS), ground granulated blast-furnace slag (GGBS), circulating fluidized bed fly ash (CFBFA), flue gas desulfurization gypsum (FGDG), and a small amount of ordinary Portland cement (OPC) as raw materials. First, the optimal mix proportions of MSWC were determined using response surface methodology (RSM). Second, the flow expansion, compressive strength, and drying shrinkage of FSS treated with MSWC (MSWC-FSS) were analyzed through laboratory and on-site tests. Finally, the curing mechanism of MSWC-FSS was analyzed by XRD and SEM. The experimental results indicate that MSWC can effectively replace OPC for FSS.

2. Materials

2.1. Soil

The soil used in this study was collected from the construction site in Heze City, China.

Table 1. Basic physical performance indicators of soil.

Items	Values
Initial water content (%)	5.9
Specific gravity	2.69
Liquid limit (%)	23.5
Plastic limit (%)	18.0
Plasticity index (%)	5.5
Clay fraction (%)	7.58
Silt fraction (%)	82.59
Sand fraction (%)	9.83
Maximum dry density (g/cm3)	1.71
Optimum moisture content (%)	13.1

presents the basic physical properties of the soil. The soil was identified as low-liquid-limit silt. Particle size analysis of the soil (Figure 1) was conducted using a laser particle size analyzer, revealing that 90.17% of the soil particles have a diameter less than 75 μm. The mineral composition (Figure 2) was determined using an X-ray diffractometer (XRD), identifying quartz, illite, and kaolinite as the primary minerals. Micrographs of the soil (Figure 3) were obtained via scanning electron microscopy (SEM), showing that the particle sizes were relatively uniform, with high sphericity and no bonding between particles.

2.2. Raw Materials of MSWC

The raw materials of MSWC include SS, GGBS, CFBFA, FGDG, and OPC. SS is a byproduct of steel making from Laiwu ironworks in Laiwu City, China. SS is treated by the thermal self-decomposition method. GGBS is a byproduct of blast furnace ironmaking from Laiwu ironworks, and is of grade I quality. CFBFA and FGDG are products of flue gas desulfurization, obtained from a power plant in Ningbo City, China. The basic parameters of OPC (P.O 42.5) produced by a cement company in Jinan City are shown in

Table 2. Basic parameters of OPC.

Setting Time (min)		Compressive Strength (MPa)
Initial	Final	3 d	7 d	28 d
115	184	33.8	43.5	51.6

, China. The specific surface areas of raw materials are shown in

Table 3. Specific surface areas of raw materials.

Materials	Specific Surface Area (m2/kg)
SS	497
GGBS	463
CFBFA	536
FGDG	358
OPC	381

. These materials are mixed in specific proportions with a certain amount of OPC to produce the MSWC.

The mineral and chemical compositions of each raw material were analyzed using XRD and X-ray fluorescence (XRF) analysis, as presented in Figure 4 and

Table 4. Chemical composition of raw materials.

Materials	Chemical Composition (wt%)
	CaO	SiO2	Al2O3	MgO	Fe2O3	SO3	TiO2	K2O
SS	30.62	16.28	8.72	6.29	28.57	—	—	—
GGBS	44.71	29.29	14.85	7.33	0.39	1.28	0.68	0.41
CFBFA	3.44	49.93	36.17	0.79	5.8	1.12	1.01	1.17
FGDG	45.35	1.56	0.8	0.35	0.12	50.63	0.02	0.41
OPC	51.42	24.99	8.26	3.71	4.03	2.51	—	—

. The main chemical compositions of SS, GGBS, CFBFA, and OPC are CaO, SiO₂, and Al₂O₃. SS contained a certain amount of tricalcium silicate (C₃S) and dicalcium silicate (C₂S), which were the main components of cement clinker and were also included in OPC [22]. GGBS had no obvious diffraction peaks and was an amorphous glass phase substance [23]. CFBFA was mainly composed of SiO₂ and Al₂O₃, and its crystal minerals included quartz, Fe₂O₃, NaAlSi₃O₈, etc. The main chemical composition of FGDG was CaSO₄.

Figure 5 shows the SEM images of each raw material of MSWC. The SS particles had rough surfaces with numerous attachments. GGBS particles had smooth surfaces with clear edges. CFBFA particles were mostly spherical with a loose structure. FGDG particles were roughly prismatic and relatively large. OPC particles were polyhedral minerals with varying shapes.

3. Experimental Methods

3.1. Test Method

3.1.1. Unconfined Compressive Strength (UCS) Test

The UCS test was carried out according to Chinese specification JTG E40-2007 [24]. The cured cube samples of FSS with a side length of 70.7 mm were vertically loaded at a loading rate of 1 mm/min using a universal testing machine until the samples were destroyed. The peak load was recorded, and the UCS was calculated.

3.1.2. Flow Expansion Test

The flow expansion test was carried out according to the specifications ASTM D6103-2017 and DBJ51/T 188-2022 [25,26]. First, a cylindrical mold with an inner diameter of 75 mm, an outer diameter of 85 mm, and a height of 150 mm was placed on a smooth glass plate. Second, the FSS was filled into the mold, and the surface of the FSS was leveled. Third, the mold was lifted vertically in one motion, allowing the FSS to collapse and flow outward to form a disk shape under gravity. Finally, the maximum diameter of the base and its perpendicular diameter were measured, and the average of these two measurements was taken as the flow expansion of FSS.

3.1.3. Drying Shrinkage Test

The drying shrinkage test was carried out according to Chinese specification JTG 3441-2024 [27]. First, after curing for 7 days, the FSS beam specimen was placed on a smooth glass rod of a shrinkage tester. Organic glass sheets were attached to both ends of the beam specimen, and two displacement sensors were installed on the outside of these glass sheets to test the displacement at both ends of the beam specimen. Second, the FSS beam specimen installed on the shrinkage tester was placed in an environment with a temperature of 20 ± 0.5 °C and a humidity of 60 ± 5%, and the readings of displacement sensors were recorded daily for the first seven days and every other day from the 7th to the 30th day. Water loss measurements were also taken during this period. Finally, after 30 days, the specimens were dried to a constant mass. The indexes of the drying shrinkage test were calculated as follows:

$$w_i=\left(m_i-m_{i+1}\right)/m_p,$$

(1)

$${\epsilon }_i=\left({\displaystyle \sum _{j=1}^2{X}_{i,j}-}{\displaystyle \sum _{j=1}^2{X}_{i+1,j}}\right)/l,$$

(2)

$${\alpha }_i={\epsilon }_i/w_i,$$

(3)

$$\alpha ={\displaystyle \frac{{\displaystyle \sum {\epsilon }_i}}{{\displaystyle \sum w_i}}},$$

(4)

where w_i is the water loss rate (%); i is the days the specimen was placed in a dry environment (i = 1, 2, 3, …, 30 d); m_i is the mass of the standard specimen (g); m_p is the mass of the standard specimen after drying (g); ε_i is the drying shrinkage strain (%); X_i,j is the reading of the displacement sensor (mm); j represents different displacement sensors (j = 1, 2);$l$ is the length of the specimen at the beginning of the test (mm); α_i is the drying shrinkage coefficient on the i-th day (%); α is the cumulative drying shrinkage coefficient (%); Σε_i is the cumulative drying shrinkage strain (%); and Σw_i is the cumulative water loss rate (%).

3.1.4. On-Site Test

The on-site test of FSS was conducted at the China-Europe Industrial Park project near Yaoqiang Airport in Jinan City, China. To investigate the engineering quality of MSWC-FSS, OPC-FSS was used as a control group. In the on-site test, FSS mixed mechanically for 3 min was used for backfilling foundation pits, with each pit measuring 4 m in length, 1.5 m in width, and approximately 0.5 m in backfill depth. The curing agent and soil were poured into a pre-dug mixing pit, and mixing water, which C_w = 0.45, was added. The mixture was then blended on-site using custom-made mixing machinery. Finally, the prepared FSS was poured into the foundation pit for backfilling. The construction process is shown in Figure 6. The experimental project included flow expansion, core drilling, and UCS of core samples.

3.1.5. X-Ray Diffraction (XRD) Test

The XRD tests of FSS powder with a particle size less than 0.075 mm were conducted using a D8 ADVANCE powder diffractometer (Bruker Corporation, Billerica, Massachusetts) with a Cu Kα source, operating at a step size of 0.02° and a scan rate of 2°/min.

3.1.6. Scanning Electron Microscope (SEM) Test

The samples were sublimated under vacuum for 48 h. The samples were coated with a layer of Pt to induce conductivity before imaging. Micrographs were obtained using a TESCAN MIRA LMS scanning electron microscope.

3.1.7. Specimen Preparation

The preparation parameters of FSS used in the experiment include the cementitious material dosage (C_b) and water-to-solid ratios (C_w), as shown in

Table 5. Preparation parameters of FSS samples.

Research Objective	Type of Test	Cb (%)	Cw (%)
Development of MSWC by RSM	UCS test	15	45
Performance and mechanism analysis of MSWC-FSS	Flow expansion test, UCS test	10, 15, 20	45, 50
	Drying shrinkage test, SEM test, XRD test, on-site test	15	45

. The FSS prepared according to the parameters was first stirred slowly for 2 min and then rapidly for 2 min using a slurry mixer. The uniformly mixed FSS could be directly subjected to flow expansion testing. For the UCS test, the steps for specimen preparation are as follows. First, the uniformly mixed FSS was poured into a mold, scraped flat on the surface, wrapped in cling film, and placed in a curing box at 20 °C and 96% humidity. Secondly, after curing for 2 days, the FSS was taken out of the curing box and demolded. The FSS specimen was wrapped in cling film and placed in a curing box at 20 °C and 96% humidity. Finally, the specimens of FSS that had reached the curing time were taken out of the curing box and tested. The samples for the SEM test and the XRD test were taken from fragments of UCS test samples.

3.2. Experimental Program and Flowchart

First, the optimal mixing ratio of MSWC for FSS was determined through RSM, with the UCS of the MSWC-FSS as the response value and different dosages of raw materials as the influencing factors. Subsequently, to evaluate the engineering performance of MSWC-FSS, the flow expansion, UCS, and drying shrinkage of MSWC-FSS were tested through laboratory tests, and the engineering application effect of MSWC-FSS was verified through on-site tests and compared with FSS treated with OPC (OPC-FSS). Finally, the curing mechanism of MSWC-FSS was analyzed by XRD and SEM tests. The overall experimental flow chart is shown in Figure 7.

3.3. Experimental Design and Statistical Analysis

RSM is a mathematical and statistical technique often used for experimental optimization design [28,29,30]. In this study, a Box–Behnken Design (BBD) of RSM with four factors at three levels was used to determine the optimal mixing ratio of MSWC. The mass proportions of SS, GGBS, CFBFA, and FGDG in MSWC were selected as four influencing factors A, B, C, and D, respectively. Based on the results of the previous single-factor experiments, engineering experience, and other literature, the mass proportions of raw materials of MSWC are as follows: SS (15–25%), GGBS (35–45%), CFBFA (10–18%), FGDG (4–8%), and the remaining ones are OPC. The UCS with a curing time of 7 days and 28 days for MSWC-FSS was determined as the response value. The RSM data analysis was conducted using Design Expert 11.0 software. The coded levels of the response surface are shown in

Table 6. The real values and coded values of variable levels.

Number	Factors	Variable Level
		−1	0	1
A	SS (%)	15	20	25
B	GGBS (%)	35	40	45
C	CFBFA (%)	10	14	18
D	FGDG (%)	4	6	8

. The specific steps of statistical analysis are as follows. First, the BBD of RSM was used to design an experimental scheme; next, the polynomial models between the factors and the responses were established based on experimental data, with model reliability verified through analysis of variance (ANOVA); then, statistical tests were conducted to evaluate the significance levels of both the overall model and individual variables; finally, the optimal mixing ratio of the MSWC was determined according to the application requirements of UCS [31,32].

4. Results and Discussion

4.1. Development of MSWC by RSM

4.1.1. BBD Experimental Results

Table 7. BBD experiment scheme and the obtained experimental results.

Run	Variables				Responses
	A	B	C	D	Y1	Y2
	SS (%)	GGBS (%)	CFBFA (%)	FGDG (%)	7 d UCS (MPa)	28 d UCS (MPa)
1	20	35	14	8	0.63	1.07
2	20	35	18	6	0.65	1.10
3	15	40	18	6	0.49	0.92
4	15	35	14	6	0.48	0.94
5	20	40	18	4	0.67	1.11
6	15	40	14	8	0.52	0.95
7	15	40	10	6	0.52	0.93
8	20	45	14	4	0.70	1.16
9	20	40	14	6	0.74	1.23
10	20	40	14	6	0.73	1.27
11	20	45	10	6	0.66	1.12
12	25	40	14	8	0.61	1.03
13	20	40	14	6	0.72	1.23
14	25	45	14	6	0.66	1.07
15	20	45	14	8	0.69	1.18
16	15	45	14	6	0.49	0.96
17	25	40	14	4	0.59	0.99
18	20	40	14	6	0.73	1.25
19	20	40	18	8	0.68	1.20
20	25	40	10	6	0.57	0.96
21	20	40	14	6	0.74	1.21
22	20	35	14	4	0.63	1.09
23	15	40	14	4	0.52	0.94
24	20	45	18	6	0.74	1.20
25	25	35	14	6	0.55	0.93
26	20	40	10	8	0.66	1.15
27	20	40	10	4	0.65	1.13
28	25	40	18	6	0.65	1.05
29	20	35	10	6	0.61	1.07

shows the 7 d UCS and 28 d UCS of 29 groups of MSWC-FSS with different mixing ratios.

4.1.2. Response Surface Fitting Model

By analyzing the experimental results in Table 7, the second-order polynomial regression equation for the effects of various industrial solid waste dosages on the 7 d UCS (Y₁) and 28 d UCS (Y₂) of MSWC-FSS was established as follows:

Y₁ = 0.732 + 0.0508A + 0.0325B + 0.0175C + 0.0025D + 0.025AB + 0.0275AC + 0.005AD + 0.01BC − 0.0025BD − 0.1439A² − 0.0389B² − 0.0314C² − 0.0314D².

(5)

Y₂ = 1.24 + 0.0308A + 0.0392B + 0.0183C + 0.0133D + 0.035AB + 0.025AC + 0.0075AD + 0.0125BC + 0.01BD + 0.0175CD − 0.2107A² − 0.0582B² − 0.0544C² − 0.04692D².

(6)

4.1.3. Analysis of Variance (ANOVA)

The validity of the response surface fitting model and the significance of factors were evaluated through ANOVA. The results of ANOVA are shown in

Table 8. ANOVA for 7 d UCS and 28 d UCS models.

Source	Degree of Freedom	7 d UCS (Y1)				28 d UCS (Y2)
		Sum of Squares	Mean Square	F-Value	p-Value	Sum of Squares	Mean Square	F-Value	p-Value
Model	14	0.1888	0.0135	76.14	<0.0001 *	0.3370	0.0241	48.45	<0.0001 *
A (SS)	1	0.0310	0.0310	175.05	<0.0001 *	0.0114	0.0114	22.96	0.0003 *
B (GGBS)	1	0.0127	0.0127	71.55	<0.0001 *	0.0184	0.0184	37.05	<0.0001 *
C (CFBFA)	1	0.0037	0.0037	20.75	0.0004 *	0.0040	0.0040	8.12	0.0129 *
D (FGDG)	1	0.0001	0.0001	0.4234	0.5258	0.0021	0.0021	4.29	0.0572
AB	1	0.0025	0.0025	14.11	0.0021 *	0.0049	0.0049	9.86	0.0072 *
AC	1	0.0030	0.0030	17.08	0.0010 *	0.0025	0.0025	5.03	0.0416 *
AD	1	0.0001	0.0001	0.5645	0.4649	0.0002	0.0002	0.4529	0.5119
BC	1	0.0004	0.0004	2.26	0.1551	0.0006	0.0006	1.26	0.2809
BD	1	0.0000	0.0000	0.1411	0.7128	0.0004	0.0004	0.8052	0.3847
CD	1	0.0000	0.0000	0.0000	1.0000	0.0012	0.0012	2.47	0.1387
A2	1	0.1343	0.1343	758.42	<0.0001 *	0.2879	0.2879	579.47	<0.0001 *
B2	1	0.0098	0.0098	55.46	<0.0001 *	0.0219	0.0219	44.18	<0.0001 *
C2	1	0.0064	0.0064	36.14	<0.0001 *	0.0192	0.0192	38.66	<0.0001 *
D2	1	0.0064	0.0064	36.14	<0.0001 *	0.0143	0.0143	28.74	0.0001 *
Residual Error	14	0.0025	0.0002	-	-	0.0070	0.0005	-	-
Lack of Fit	10	0.0022	0.0002	3.14	0.1405	0.0049	0.0005	0.9375	0.5778
Pure Error	4	0.0003	0.0001	-	-	0.0021	0.0005	-	-
Total	28	0.1913	-	-	-	0.3439	-	-	-

Note: * indicates that the item is significant (p-value < 0.05).

. The significance level represented the extent to which each factor affects the UCS. In statistics, the significance level α was typically set at 0.05, and a p-value less than 0.05 was considered significant [33]. The p-values of the 7 d and 28 d UCS models were less than 0.0001, indicating that the models were highly significant. This confirmed the high accuracy of the RSM analysis. The order of significance for the factors at curing 7 d was A (SS) > B (GGBS) > C (CFBFA) > D (FGDG), whereas at curing 28 d, the order changed to B (GGBS) > A (SS) > C (CFBFA) > D (FGDG). The p-values for factors A, B, and C were less than 0.05 at 7 and 28 days of curing, indicating that SS, GGBS, and CFBFA had an extremely significant impact on UCS, and SS and GGBS played a dominant role in the UCS. From the interaction effects of the factors, the p-values for AB and AC were both less than 0.05, whether cured for 7 or 28 days, indicating that the combined interaction effects of SS and GGBS, as well as SS and CFBFA, had a significant influence on UCS. This was because the hydration process of C₂S and C₃S in SS would generate Ca(OH)₂, which would promote the dissolution of glass in GGBS and silicoaluminal raw materials in CFBFA and react to form gel materials such as C-(A)-S-H, C-A-H, etc. [34,35].

The determination coefficient (R²), adjusted determination coefficient (adj-R²), and predicted determination coefficient (pred-R²) of the 7 d UCS model were 0.9870, 0.9741, and 0.9315, respectively. R², adj-R², and pred-R² of the 28 d UCS model were 0.9789, 0.9596, and 0.9089, respectively. R², adj-R², and pred-R² were all very close to 1, and the difference between adj-R² and pred-R² of all fitted models was far less than 0.2, indicating that the UCS models cured for 7 and 28 days had high fitting accuracy and predictive ability.

The normal probability of the studentized residuals for the two UCS fitting models with 7 and 28 days of curing is shown in Figure 8. The residuals of the two fitting models were mostly distributed along a reasonable straight line, indicating that the errors of the two fitting models were normally distributed and met the normality assumption of ANOVA.

The experimental and predicted values of two UCS fitting models were compared under the same mix ratio, and the results are shown in Figure 9. The ratio of all predicted values to experimental values of the two fitting models was around 1, indicating that all fitting models had high accuracy.

4.1.4. Effect of Solid Waste Dosage on UCS of MSWC-FSS

The 3D response surface plots of UCS fitting models with 7 and 28 days of curing are shown in Figure 10 and Figure 11. All 3D response surfaces of UCS were convex surfaces, and the maximum UCS values of the convex surfaces were within the range of the solid waste dosage in the experiment. With varying proportions of the four materials, the UCS values of MSWC-FSS showed a trend of first increasing and then decreasing.

4.1.5. Optimization and Validation

According to the optimization results of the 28 d UCS model, the optimal mixing ratio for the MSWC was determined to be 20.95% SS, 42.39% GGBS, 15.42% CFBFA, 6.34% FGDG, and 14.9% OPC. To simplify subsequent experimental operations and engineering applications, the optimal mixing ratios were adjusted to 20% SS, 40% GGBS, 15% CFBFA, 5% FGDG, and 20% OPC. Subsequent experiments in this study were conducted at the adjusted optimal mixing ratio.

To verify the reliability of the model, MSWC-FSS specimens were remade and retested under the adjusted optimal mixing ratio.

Table 9. Comparison of measured and predicted values at the optimal mixing ratio.

Index	7 d UCS (MPa)	28 d UCS (MPa)
Measured values	0.77	1.21
Predicted values	0.75	1.25
Relative error	2.67%	3.20%

shows that the relative error between the experimental values and the predicted values was less than 5%, indicating that the fitted model’s predictions were accurate. The 28 d UCS measurement results of MSWC-FSS met the strength requirements of the Chinese standard DBJ51/T 188-2022 [25] for backfilling bridge abutments and culverts (≥0.8 MPa), as well as filling foundations and roadbeds (≥1.0 MPa).

4.2. Performance and Mechanism Analysis of MSWC-FSS

4.2.1. Flow Expansion

Figure 12 shows the flow expansion of MSWC-FSS and OPC-FSS under different C_b and C_w. At the same C_w and C_b, the flow expansion of MSWC-FSS was smaller than that of OPC-FSS, but the difference was usually only 5% or less. As C_w increased, the flow expansion of MSWC-FSS and OPC-FSS rapidly increased. As C_b increased, the flow expansion of MSWC-FSS and OPC-FSS decreased because the hydration reaction of these two cementitious materials consumed a certain amount of free water. According to Chinese Standard DBJ51/T 188-2022 [25], the flow expansion of FSS used for backfilling bridge abutments and culverts and paving general highway roadbeds should be between 160 and 220 mm. The flow expansion of MSWC-FSS and OPC-FSS with 45% C_w could meet this requirement.

4.2.2. Unconfined Compressive Strength (UCS)

Figure 13 shows the UCS of MSWC-FSS and OPC-FSS cured for 7 and 28 days under different C_b and C_w. Regardless of curing for 7 days or 28 days, under the same C_b and C_w, the UCS of MSWC-FSS was generally higher than that of OPC-FSS by 10% or more, indicating that MSWC had a better solidification effect on FSS than OPC. The UCS of MSWC-FSS and OPC-FSS increased with curing time, and the increase gradually slowed down. With the increase of C_w, the UCS of MSWC-FSS and OPC-FSS both increased. However, with the increase of C_w, the UCS of MSWC-FSS and OPC-FSS both decreased. According to Chinese standard DBJ51/T 188-2022 [25], under all C_b and C_w in the experiment, the 28 d UCS of MSWC-FSS and OPC-FSS met the strength requirements for backfilling the bridge abutment and culvert back (≥0.8 MPa). Except with 10% C_b and 55% C_w, both types of FSS met the strength requirements for filling foundations and roadbeds (≥1.0 MPa).

4.2.3. Drying Shrinkage

To ensure fluidity, FSS contains a large amount of water, which leads to significant drying shrinkage during the curing process, which may cause cracking of the FSS and reduce its durability. Therefore, the drying shrinkage performance of MSWC-FSS and OPC-FSS was analyzed, as shown in Figure 14, Figure 15 and Figure 16. Figure 14 and Figure 15 shows the accumulative water loss rate and the accumulative drying shrinkage strain of MSWC-FSS and OPC-FSS over time, respectively. As the drying shrinkage time increased, the cumulative water loss rate and the accumulative drying shrinkage strain of all FSS gradually increased, with rapid growth in the first 8 days and slow growth after 8 days [36,37]. At the same time, the cumulative water loss rate and cumulative shrinkage strain of MSWC-FSS were lower than those of OPC-FSS.

Figure 16 shows the drying shrinkage coefficients and the accumulative drying shrinkage coefficient of MSWC-FSS and OPC-FSS over time. As shown in Figure 16a, as the drying shrinkage time increased, the drying shrinkage coefficient of all FSS generally exhibited a fluctuating trend rather than a monotonic increase or decrease. During the first 8 days, the drying shrinkage coefficient changed rapidly, reaching a peak around the 8th day. This was due to the rapid evaporation of moisture in the early stage, leading to significant volume shrinkage. As the drying shrinkage time further increased (8–30 days), the changes in the drying shrinkage coefficient of all FSS gradually leveled off. At this stage, the evaporation rate of moisture in the FSS slowed down, and some of the moisture underwent chemical reactions with the binder to form bound water, no longer contributing to volume shrinkage. As shown in Figure 16b, as the drying shrinkage time increased, the accumulative drying shrinkage coefficient of all FSS increased gradually, with rapid growth in the first 8 days and slow growth after 8 days. At the same time, the drying shrinkage coefficients and the accumulative drying shrinkage coefficient of MSWC-FSS were lower than those of OPC-FSS.

Considering these four drying shrinkage coefficients, MSWC-FSS had better resistance to drying shrinkage than OPC-FSS. This phenomenon could be explained from the following perspectives. Due to the high content of active Al₂O₃ in SS and CFBFA, as well as the high content of CaSO₄ in FGDG, the C-A-H generated by the hydration of MSWC was converted into a large amount of ettringite (AFt), which was more abundant than that generated by the hydration of OPC. The expansive AFt filled the internal pores of MSWC-FSS, enhancing its compactness and compensating for some drying shrinkage [38]. The reaction equation is as follows:

$$3\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\cdot 6{\mathrm{H}}_2\mathrm{O}+3\mathrm{C}\mathrm{a}\mathrm{S}{\mathrm{O}}_4\cdot 2{\mathrm{H}}_2\mathrm{O}+24{\mathrm{H}}_2\mathrm{O}\to 3\mathrm{C}\mathrm{a}\mathrm{O}\cdot \mathrm{A}{\mathrm{l}}_2{\mathrm{O}}_3\cdot 3\mathrm{C}\mathrm{a}\mathrm{S}{\mathrm{O}}_4\cdot 32{\mathrm{H}}_2\mathrm{O}$$

(7)

4.2.4. On-Site Test Analysis

At 0, 0.5, and 1 h after the preparation of the FSS, multiple samples were taken to conduct flow expansion tests to verify whether flow expansion met construction requirements. The test results are shown in

Table 10. On-site flow expansion test data of FSS (mm).

Binder Type	Time (h)
	0.0	0.5	1.0
MSWC	183	167	121
OPC	191	169	114

. The flow expansion of the three types of FSS prepared on-site exceeded 160 mm within 0.5 h, which generally met the construction requirements and demonstrated good uniformity. However, due to outdoor factors such as wind, temperature, and human operations, the on-site flow expansion was slightly lower than the indoor test results. Additionally, it was observed that the flow expansion of the on-site FSS decreased rapidly. Therefore, it is recommended to cover the mixing pit after preparation to minimize environmental impact.

Figure 17 shows the condition of the FSS backfill 12 h after pouring. The surface of OPC-FSS showed a cracking phenomenon, as indicated by the yellow dashed circles, while the surface of MSWC-FSS had no cracks, which was consistent with the conclusion obtained from the drying shrinkage test and further verifies the unique advantage of MSWC-FSS in resisting drying shrinkage.

Figure 18 shows the core samples with a diameter of 8 cm obtained from drilling into the FSS cured for 28 d. It could be seen that MSWC-FSS was denser and had fewer internal pores than OPC-FSS. The core samples obtained from drilling were processed into cylindrical specimens with a height and diameter of 8 cm using a cutting machine, and UCS was tested using a universal testing machine. The UCS results are shown in

Table 11. UCS of FSS core samplings (MPa).

Binder Type	Curing Time
	7 d	28 d
MSWC	0.78	1.12
OPC	0.66	1.06

. Compared to OPC-FSS, MSWC-FSS had higher strength.

4.2.5. XRD Analysis

Figure 19 shows the XRD patterns of MSWC-FSS and OPC-FSS cured for 7 and 28 days. Due to the high proportion of soil particles in FSS, the content of quartz, illite, kaolinite, and calcite was high, resulting in prominent XRD diffraction peaks. The diffraction patterns of the samples indicated that the solidification products of the two types of FSS were similar. However, the AFt diffraction peaks of MSWC-FSS were higher than OPC-FSS, indicating that there was more AFt in the hydration products of MSWC-FSS, which was consistent with the behavior of AFt in the microstructure of MSWC-FSS and OPC-FSS. This was because the Ca(OH)₂ generated by the hydration of cement clinker in OPC and SS reacted with the active SiO₂ and active Al₂O₃ contained in SS, GGBS, and CFBFA to form calcium aluminate hydrate (C-A-H). C-A-H reacted with CaSO₄ in FGDG to form AFt. Compared with OPC, MSWC contained more Al₂O₃ and CaSO₄, generating more expansive AFt, making MSWC-FSS denser than OPC-FSS. This explained why MSWC-FSS had better strength and resistance to dry shrinkage than OPC-FSS [39,40]. After 28 days of curing, the intensity of the diffraction peaks increased, indicating that hydration and pozzolanic reactions continued during the curing period from 7 to 28 days. With the increase in curing time, the generated hydration and pozzolanic products further bonded the soil particles.

4.2.6. SEM Analysis

Figure 20 shows the microstructure of MSWC-FSS and OPC-FSS cured for 7 and 28 days. It could be seen that as the curing time increased, the hydration products in all FSS became more abundant, filling the pores between soil particles and making FSS denser. Both MSWC-FSS and OPC-FSS hydration products contained flocculent and networked calcium silicate hydrate (C-S-H) and needle-shaped AFt, with the following differences: Firstly, MSWC-FSS contained more AFt than OPC-FSS at the same curing time because MSWC contained more active Al₂O₃ and CaSO₄ than OPC, resulting in the formation of more AFt during the hydration, which was consistent with the conclusion of XRD analysis. Secondly, Ca(OH)₂ existed in OPC-FSS but not in MSWC-FSS because Ca(OH)₂ produced by hydration of OPC and SS underwent a volcanic ash reaction with active SiO₂ and Al₂O₃ in SS, GGBS, and CFBFA.

5. Conclusions

This study utilized multi-source industrial solid wastes to prepare MSWC for application in FSS, validating the feasibility of MSWC as a replacement for OPC and revealing the performance of MSWC-FSS. The main conclusions were as follows:

(1)

Based on the BBD of RSM, the high-reliability UCS models of MSWC-FSS were established. The optimal mix ratio of MSWC determined by the models was SS:GGBS:CFBFA:FGDG:OPC = 20:40:15:5:20. SS, GGBS, and CFBFA had an extremely significant impact on UCS of MSWC-FSS, and SS and GGBS played a dominant role in the UCS. The combined interaction effects of SS and GGBS, as well as SS and CFBFA, had a significant influence on UCS of MSWC-FSS.

(2)

At the same C_b and C_w, the flow expansion of MSWC-FSS was less than 5% lower than that of OPC-FSS, while the UCS of MSWC-FSS was more than 10% higher than that of OPC-FSS. MSWC-FSS with a C_w of 45% and a C_b of 15% could simultaneously meet the requirements for slump and strength in the Chinese standard and had better resistance to dry shrinkage than OPC-FSS. The above conclusion has been confirmed by both laboratory and on-site tests, indicating that MSWC was more suitable for FSS than OPC.

(3)

The chemical composition and microstructure analysis showed that the curing mechanism of MSWC-FSS and OPC-FSS was similar. However, compared with OPC-FSS, MSWC-FSS generated more AFt with an expansion effect during the solidification process, making MSWC-FSS denser. This was the reason why MSWC-FSS had stronger strength and resistance to drying shrinkage.