Optimization of glass fiber reinforced MICP cured coastal sediments

Abstract

This study integrates glass fibers (GF) with Microbial Induced Carbonate Precipitation (MICP) to enhance the consolidation of coastal dredged sediments (CDS) and reduce treatment costs. Response Surface Methodology (RSM) was employed to optimize four parameters—moisture content, reaction time, fiber length, and fiber content—using unconfined compressive strength (UCS) and calcium carbonate content (CC) as response variables. The developed regression models exhibited high predictive accuracy (R2 0.98). Moisture content and reaction time were identified as the most influential factors for UCS and CC, respectively. Under the optimal conditions (moisture content 39.9%, reaction time 4 days, fiber length 5.6 mm, and fiber content 2.44%), the predicted UCS and CC reached 4.29 MPa and 3.21%, respectively, which closely matched experimental values. SEM and XRD analyses revealed that GF enhanced the formation of a CaCO₃–fiber network and increased calcite content from 10.5 to 11.5%. These findings demonstrate that integrating GF with MICP significantly improves mechanical strength and calcium carbonate deposition while reducing treatment costs, offering a sustainable solution for coastal sediment stabilization and reuse.

Supplementary Information

The online version contains supplementary material available at 10.1038/s41598-026-51790-7.

Introduction

Coastal dredging is an essential activity for maintaining navigable waterways and supporting global trade¹. However, this process generates enormous quantities of coastal dredged sediment (CDS), typically characterized by high moisture content, low strength, and poor permeability². Conventional disposal methods such as landfilling and chemical stabilization consume significant land and energy resources and often lead to secondary pollution³. Consequently, developing environmentally friendly and cost-effective approaches for the reuse and stabilization of CDS has become a pressing need.

Microbial-induced calcium carbonate precipitation (MICP or MICCP) has emerged as a sustainable technique for soil improvement due to its environmental compatibility, low energy demand, and controllability⁴. In this process, urease-producing bacteria (UPB) catalyze the hydrolysis of urea, leading to calcium carbonate precipitation that binds soil particles, thereby enhancing strength and reducing permeability^5,6. Despite its promise, large-scale applications of MICP remain limited by two major challenges: (1) high treatment costs associated with long curing times and high moisture content, and (2) brittle interparticle bonding caused by discrete CaCO₃ crystals, which restricts ductility and durability of the treated material.

The use of native UPB provides a potential solution to reduce cost and environmental risk. Compared with exogenous strains, indigenous UPB can adapt more effectively to local geochemical conditions, sustain higher urease activity, and eliminate the need for costly acclimation^7,8. In this study, the native Bacillus pacificus strain CM11 (OR770090), isolated from coastal sediments, was employed to ensure compatibility with the treatment environment and enhance carbonate precipitation efficiency.

To address the mechanical limitations of MICP-treated soils, fiber reinforcement has been proposed as a promising strategy. Fibers such as glass, basalt, or synthetic polymers improve unconfined compressive strength (UCS), toughness, and tensile resistance by bridging CaCO₃-bonded particles and restraining crack propagation^9,10. Glass fiber (GF), in particular, offers superior chemical resistance, dimensional stability, and cost-effectiveness, making it a suitable reinforcement material^11,12. However, previous studies have mostly focused on the overall strength improvement, while limited attention has been given to understanding the synergistic effects of GF incorporation with native UPB-mediated MICP in coastal dredged sediments. Moreover, systematic optimization of operational parameters governing both microbial activity and fiber reinforcement efficiency remains scarce.

Response Surface Methodology (RSM) provides a robust statistical framework to optimize multiple interacting variables simultaneously while reducing experimental workload^13,14. Although RSM has been used in MICP studies, few have applied it to evaluate the combined influence of moisture content, reaction time, fiber length, and fiber content on both mechanical performance and cost efficiency.

Therefore, this study employs RSM to optimize these key parameters for the glass fiber–reinforced MICP treatment of CDS. The objectives are to: (1) enhance sediment strength through the synergistic effect of GF and MICP; (2) reduce treatment cost by lowering moisture content and reaction time; and (3) elucidate the reinforcement mechanism through calcium carbonate deposition analysis using SEM and XRD. The Unconfined Compressive Strength (UCS) test is a widely utilized method for evaluating soil stability, as it provides a reliable measure of the inherent strength of the soil mass. This test holds significant value in the preliminary design and analysis of geotechnical engineering projects¹⁵.

This research contributes to a deeper understanding of the interaction between native UPB and glass fibers in MICP-treated sediments and provides an economical and sustainable strategy for large-scale coastal sediment reuse.

Materials and methods

Materials

CDS samples were collected from Huangye Harbor, Bohai Bay, China (38°17′50.77″N, 117°51′17.24″E). Basic physical and chemical properties are summarized in Table S1. The properties are suboptimal for UPB proliferation ,such as pH 7.41, SOC 1.71% and CC 1.29%. The sediments have a well-graded particle size distribution (Fig. S1).

Phylogenetic analysis (Fig. 1a) confirmed that the locally isolated CM11 was identified as Bacillus pacificus CM11 (OR770090). Growth curve analysis (Fig. 1b) showed that the strain reached a slight decline after 24 h. Bacterial cultures harvested at 24 h were used, corresponding to maximal metabolite accumulation under nutrient-limited conditions, ensuring controlled and efficient MICP reactions. Bio-CaCO₃ precipitates were characterized by XRD and SEM (Fig. S2), showing calcite and vaterite phases.

Fig. 1.

Neighbor-joining evolutionary tree (a) of B. pacificus CM11; Growth curve (b) of strain CM11.

Glass fibers (GF), known for their high strength and toughness, were incorporated as structural reinforcements. Their inclusion significantly enhances the mechanical properties of building materials while effectively preventing cracking and fracture under mechanical stress¹³. GF properties are summarized in Table 1.

Table 1.

Surface characteristics of glass fibers.

Specific gravity (kg/m3)	Elasticity modulus (MPa)	Tensile strength (MPa)	Length (mm)	Average diameter (mm)	Elongation at break (%)
2680	8720	2180	3\6\9	0.0174	2.52

Methods

Sample preparation

The specimen preparation process is depicted in Fig. 2. The strain CM11 cultures were incubated in nutrient broth (pH = 9.00) for 24 h to obtain a bacterial suspension (BS). The BS was then mixed with the cementing solution (CS), which contained urea and calcium chloride dihydrate each at 1 mol/L, at a 1:4 volume ratio (CS volume four times that of BS).

Fig. 2.

Sample preparation of “fiber-MICP” columns.

The resulting mixture was combined with GF-reinforced sediments, molded (Φ50 × 50 mm), cured (20 °C and ≥ 95% humidity), and demolded and dried (Fig. 2). The chosen parameters were based on initial experimental results and existing literature.

Test methods

UCS tests were performed according to ASTM D4219-08¹⁶. Axial loading rate was set at 0.1 mm/min, chosen to ensure accurate stress–strain measurements. Carbonate content was measured via neutralization titration. Mineralogy and morphology were analyzed by XRD (2θ = 10°–80°, step 0.02°) and SEM (gold-coated, 30 mA, 120 s).

Representative fracture patterns with and without GF are shown in Fig. S3. The sample without glass fibers (Fig. S3b) exhibited large, deep cracks, indicative of brittle failure, a typical behavior when internal stresses exceed tensile strength. Conversely, sample with GF (Fig. S3c) showed smaller, shallower cracks, reflecting a more ductile failure mode. This change is attributed to the reinforcing effects of the glass fibers, which enhance toughness, improve stress distribution, and prevent catastrophic failure.

RSM design

In this study, four factors, such as moisture content (A), reaction time (B), fiber length (C), and fiber content (D), were investigated using response surface methodology (RSM). UCS and carbonate content (CC) were selected as response variables. Based on the results of the single factor experiments in Fig. S4, a 4-factor, 3-level RSM experiment was designed using the Box-Behnken Design (BBD) method, resulting in 30 experimental runs. The input values for each factor, based on preliminary experiments, are presented in Table 2, with − 1 indicating the low level, 0 representing the center point, and + 1 representing the high level.

Table 2.

Initial input variable for the parameters.

	Name	Unit	−1	0	+ 1
A	Moisture content	%	32	38	44
B	Reaction time	Day	3	4	5
C	Fiber length	mm	3	6	9
D	Fiber content	%	2.0	2.5	3.0

Results and discussions

Table 3 lists the 30 samples with different compositions. All samples in the table are listed in random order to minimize potential experimental errors¹³. The CC and UCS data for each sample are also included in the table.

Table 3.

Input parameter and corresponding output response of the RSM.

Std	Run	A: Moisture content	B: Reaction time	C: Fiber length	D: Fiber content	UCS	CaCO3
		%	Day	mm	%	MPa	%
3	6	32	5	6	2.5	3.231	2.34
17	11	32	4	3	2.5	2.870	2.25
11	14	32	4	6	3	3.094	2.24
9	21	32	4	6	2	3.038	2.25
1	23	32	3	6	2.5	2.923	2.18
19	27	32	4	9	2.5	3.092	2.25
24	1	38	5	6	3	3.121	2.54
16	3	38	5	9	2.5	3.238	2.51
15	4	38	3	9	2.5	2.759	2.00
23	5	38	3	6	3	3.013	2.00
21	7	38	3	6	2	3.299	2.53
8	9	38	4	9	3	2.989	2.25
26	12	38	4	6	2.5	4.310	3.22
27	15	38	4	6	2.5	4.098	3.27
29	16	38	4	6	2.5	4.204	3.29
5	18	38	4	3	2	3.352	2.35
22	19	38	5	6	2	3.207	2.24
13	20	38	3	3	2.5	3.247	2.49
7	22	38	4	3	3	2.967	2.27
30	24	38	4	6	2.5	4.248	3.18
14	25	38	5	3	2.5	3.122	2.23
6	28	38	4	9	2	2.927	2.15
25	29	38	4	6	2.5	4.329	3.33
28	30	38	4	6	2.5	4.334	3.31
2	2	44	3	6	2.5	3.622	2.24
18	8	44	4	3	2.5	3.662	2.35
10	10	44	4	6	2	3.691	2.33
12	13	44	4	6	3	3.346	2.19
4	17	44	5	6	2.5	3.533	2.43
20	26	44	4	9	2.5	3.124	2.10

Optimization: RSM (ANOVE)

The length and content of fibers significantly influence both the CaCO₃ production process of the strain CM11 and the mechanical properties of MICP-treated samples. Shu et al. observed that CaCO₃ (%) in MICP-treated sand columns initially increased with fiber content before subsequently declining¹⁷. Excessive fiber distribution may obstruct bacterial migration, while fiber aggregation can occupy pore spaces, thereby reducing CaCO₃ precipitation. Long fibers are more prone to agglomeration than short fibers, which may further inhibit CaCO₃ deposition¹⁰.

A total of 30 samples with varying compositions were analyzed to evaluate the effects of moisture content (A), reaction time (B), fiber length (C), and fiber content (D) on unconfined compressive strength (UCS) and calcium carbonate content (CC) (Table 3). A quadratic response surface regression model was constructed, and ANOVA results (Tables 4 and 5) confirmed statistical significance for both UCS (F = 115.22, p < 0.0001) and CC (F = 176.19, p < 0.0001), with lack-of-fit terms insignificant (p 0.05), indicating model adequacy. Based on F-value ranking, the factor importance order for UCS was: moisture content (A) fiber length (C) fiber content (D) reaction time (B), while for CC, it was: reaction time (B) fiber length (C) fiber content (D) moisture content (A).

Table 4.

Analysis of the quadratic model for UCS.

Source	Sum of Squares	df	Mean Square	F-value	p-value
Model	6.90	14	0.4931	115.22	< 0.0001	Significant
A-Moisture content	0.6211	1	0.6211	145.13	< 0.0001
B-Reaction time	0.0289	1	0.0289	6.76	0.0201
C-Glass fiber length	0.0992	1	0.0992	23.18	0.0002
D-Glass fiber content	0.0807	1	0.0807	18.85	0.0006
AB	0.0394	1	0.0394	9.21	0.0084
AC	0.1444	1	0.1444	33.74	< 0.0001
AD	0.0402	1	0.0402	9.39	0.0079
BC	0.0912	1	0.0912	21.31	0.0003
BD	0.0100	1	0.0100	2.34	0.1472
CD	0.0500	1	0.0500	11.67	0.0038
A2	1.15	1	1.15	269.08	< 0.0001
B2	1.88	1	1.88	439.40	< 0.0001
C2	2.85	1	2.85	665.41	< 0.0001
D2	2.13	1	2.13	498.09	< 0.0001
Residual	0.0642	15	0.0043
Lack of fit	0.0222	10	0.0022	0.2636	0.9654	Not significant
Pure error	0.0420	5	0.0084
Cor total	6.97	29

Table 5.

Analysis of the quadratic model for CC.

Source	Sum of squares	df	Mean Square	F-value	p-value
Model	5.16	14	0.3687	176.19	< 0.0001	Significant
A-Moisture content	0.0014	1	0.0014	0.6729	0.4249
B-Reaction time	0.0602	1	0.0602	28.77	< 0.0001
C-Glass fiber length	0.0385	1	0.0385	18.41	0.0006
D-Glass fiber content	0.0108	1	0.0108	5.16	0.0383
AB	0.0002	1	0.0002	0.1075	0.7475
AC	0.0156	1	0.0156	7.47	0.0154
AD	0.0042	1	0.0042	2.02	0.1758
BC	0.1482	1	0.1482	70.83	< 0.0001
BD	0.1722	1	0.1722	82.29	< 0.0001
CD	0.0081	1	0.0081	3.87	0.0679
A2	1.85	1	1.85	883.15	< 0.0001
B2	1.37	1	1.37	653.71	< 0.0001
C2	1.80	1	1.80	862.01	< 0.0001
D2	1.68	1	1.68	804.19	< 0.0001
Residual	0.0314	15	0.0021
Lack of fit	0.0153	10	0.0015	0.4729	0.8532	Not significant
Pure error	0.0161	5	0.0032
Cor total	5.19	29

Predictive and adjusted R² values were high (UCS: 0.9730 and 0.9822; CC: 0.9786 and 0.9833), coefficients of variation were low (UCS: 1.92%; CC: 1.85%), and adequate precision values (UCS: 31.543; CC: 39.557) greatly exceeded the threshold of 4.0. These statistics collectively demonstrate the model’s strong predictive capability for elucidating relationships between process variables and material performance¹⁸.

The full quadratic equations and coefficients are provided in the Supplementary Material (Eq. S1–S2). In the UCS model (Table 4), the p-value for the BD interaction term exceeded 0.1000, indicating its minimal contribution to the model; hence, it can be omitted from Eq. S1. Similarly, in the CC model (Table 5), the AB and AD terms were statistically insignificant and contributed minimally, justifying their removal from Eq. S2. However, due to the significant influence of other A-related terms in the CC model, the A term must be retained in Eq. S2.

Residual analysis (Fig. S5a–b) indicated uniform distribution around zero, confirming normality and independence, while predicted versus observed plots (Fig. 3a, b) showed high alignment, further supporting predictive reliability. Pearson correlation analysis (Table 6) revealed a strong positive correlation between UCS and CC (r = 0.908, p < 0.001), highlighting the pivotal role of calcium carbonate deposition in enhancing mechanical strength. Kahani et al. reported that calcium carbonate production by Sporosarcina pasteurii significantly increased under optimized medium conditions, enhancing the overall UCS of treated samples to 520 kPa¹⁴.

Fig. 3.

Predicted against actual value plots (a) UCS (b) CaCO₃(%).

Table 6.

Tests of correlation.

	Pearson correlation	Sig. (2-tailed)	df
UCS	0.908	< 0.001	30
CaCO3(%)	0.908	< 0.001	30

2D/3D surface plots illustrating processing variables’ effect on UCS and CC

The 2D contour (Figs. 4 and 7) and 3D surface plots (Figs. 5 and 6) reveal mechanistic insights into how moisture content (A), reaction time (B), fiber length (C), and fiber content (D) jointly influence UCS and CC. Across all interactions, the trends reveal that UCS and CC are influenced by a combination of microstructural GF reinforcement and MICP mineral deposition, rather than by single parameters alone.

Fig. 4.

2D plots (a–b) A–B 2D plot for unconfined compressive strength, CaCO₃ content, (c–d) A–C 2D plot for unconfined compressive strength, CaCO₃ content, (e–f) A–D 2D plot for unconfined compressive strength, CaCO₃ content.

Fig. 7.

2D plots (a–b) B–C 2D plot for unconfined compressive strength, CaCO₃ content, (c–d) B–D 2D plot for unconfined compressive strength, CaCO₃ content, (e–f) C–D 2D plot for unconfined compressive strength, CaCO₃ content.

Fig. 5.

3D plots (a–b) A–B 3D plot for unconfined compressive strength, CaCO₃ content, (c–d) A–C 3D plot for unconfined compressive strength, CaCO₃ content, (e–f) A–D 3D plot for unconfined compressive strength, CaCO₃ content.

Fig. 6.

3D plots (a–b) B–C 3D plot for unconfined compressive strength, CaCO₃ content, (c–d) B–D 3D plot for unconfined compressive strength, CaCO₃ content, (e–f) C–D 3D plot for unconfined compressive strength, CaCO₃ content.

The interaction between moisture content and reaction time (A–B) is particularly critical. Moisture content strongly affects UCS, as indicated by the denser color gradients in the 2D contour plots (Figs. 4a and 5a), while reaction time primarily governs CC (Figs. 4b and 5b). As reaction time (B) increases, CaCO₃ deposition initially rises, enhancing particle bonding and structural stability, but prolonged durations lead to excessive crystal accumulation that can hinder fiber reinforcement and microbial activity, reducing mechanical performance. Tiwari et al. reported that clay samples subjected to a four-day reaction period exhibited superior UCS performance¹⁹. Moderate moisture (A ≈ 40%) levels optimize ionic mobility and microbial activity, promoting uniform CaCO₃ deposition on fibers and mitigating localized brittleness²⁰. Consequently, optimal UCS and CC require a balanced A–B combination, where reaction time and moisture content are sufficiently high to promote crystal growth but not so high as to induce heterogeneous deposition or stress concentration.

The effects of moisture content in combination with fiber length (A–C) show similar trends to the A–B interaction. Fiber length exhibits an optimal range of 5–6 mm, within which fibers provide structural reinforcement and promote uniform CaCO₃ deposition, enhancing both UCS and CC (Figs. 4c, d and 5a, b). Fibers that are too short fail to transfer stress effectively, whereas overly long fibers may entangle, causing localized heterogeneity and compromising matrix stability¹⁵. Moisture content modulates these effects by influencing fiber dispersion and microbial activity, such that the optimal A–C combination aligns closely with that of A–B, emphasizing the importance of maintaining moderate moisture to maximize mechanical performance.

Similarly, the interaction between moisture content and fiber content (A–D) follows a comparable pattern (Figs. 4e, f and 5e, f). Low fiber content results in insufficient structural support and uneven CaCO₃ deposition, while excessive fiber content can cause entanglement and disrupt uniform mineralization¹⁵. UCS peaks at an intermediate fiber content (~ 2.5%), where fibers effectively reinforce the matrix without impeding microbial mobility. CC trends are consistent, with optimal microbial-induced deposition occurring when fiber content and moisture are balanced, highlighting the necessity of simultaneously optimizing fiber and moisture parameters.

The combined effects of reaction time, fiber length, and fiber content (B–C, B–D, C–D) further illustrate the interplay of deposition kinetics and structural reinforcement. CC increases with reaction time, reaching a maximum around four days (Figs. 6a–d, 7a–d), whereas UCS shows a more complex dependence. Fiber length and content modulate the mechanical reinforcement: UCS peaks when fiber length is 5–6 mm and fiber content is ~ 2.5% (Figs. 6a, c, e and 7a, c, e). Beyond these ranges, fiber aggregation or excessive crystal clustering reduces stress transfer efficiency, leading to localized stress concentration and decreased UCS. The C–D interaction (Figs. 6e, f and 7e, f) alone has a relatively minor effect compared to B–C and B–D, suggesting that reaction time remains the dominant factor controlling CaCO₃ accumulation²¹, whereas fibers primarily influence structural reinforcement and stress distribution.

Collectively, these observations demonstrate that UCS and CC are nonlinearly coupled. Moisture content regulates microbial activity and ion availability, reaction time governs crystal growth kinetics, and fiber geometry controls reinforcement and nucleation sites. Optimal mechanical performance is achieved when these parameters synergistically promote uniform CaCO₃ deposition, efficient stress transfer, and crack bridging. Excessive moisture, prolonged reaction times, or unbalanced fiber parameters can decouple mineral content from mechanical strength, highlighting the need for mechanism-driven optimization to achieve both UCS and CC enhancement in MICP-treated materials.

Perturbation plot

The perturbation plots in Fig. 8 illustrate the sensitivity of UCS and CC to moisture content (A), reaction time (B), fiber length (C), and fiber content (D).

Fig. 8.

Perturbation plot (a) unconfined compressive strength (b) CaCO₃ content.

As shown in Fig. 8a, moisture content (A) exerts the most significant influence on UCS, consistent with the ANOVA results. Appropriate moisture ensures sufficient pore water for urea hydrolysis and nutrient transport, sustaining microbial urease activity and enabling uniform Ca2⁺–CO₃2⁻ diffusion and CaCO₃ nucleation. In contrast, insufficient moisture restricts microbial metabolism, while excessive moisture dilutes reactants and weakens interparticle bonding, ultimately reducing UCS. These results confirm that moisture content dominates the bio-cementation efficiency by controlling both the biochemical activity and the ionic environment of MICP.

Reaction time (B) markedly affects CC (Fig. 8b). The carbonate content increases rapidly and then stabilizes as microbial activity declines and crystal growth reaches equilibrium. Beyond the optimal duration, further precipitation coarsens the microstructure, leading to a slight increase in CC but no corresponding improvement in UCS. This indicates that the distribution and morphology of CaCO₃, rather than its total quantity, primarily govern mechanical performance.

Fiber parameters (C and D) show moderate but systematic effects on both UCS and CC. Appropriate fiber length and content enhance interfacial bonding, facilitate microbial attachment, and provide nucleation sites for uniform mineral deposition²². Conversely, overly short fiber or excessive fiber content leads to heterogeneity and weak zones within the matrix²³. These observations align with Figs. 4, 5, 6 and 7, confirming that glass fiber (GF) regulates both stress transfer and microbially mediated mineralization.

Overall, Fig. 8 reveals a clear hierarchy of influence: moisture content controls microbial and chemical kinetics, reaction time dictates the evolution of CaCO₃ deposition, and fiber characteristics shape the mechanical reinforcement network. Optimal combinations—moderate moisture (~ 40%), reaction time (~ 4 days), and balanced fiber geometry (~ 6 mm, ~ 2.5%)—promote synergistic interactions among microbial activity, CaCO₃ precipitation, and interfacial adhesion, thereby maximizing UCS and stabilizing CC.

Interaction plot

Figure 9 illustrates the interactions of moisture content (A), reaction time (B), fiber length (C), and fiber content (D) on UCS and calcium carbonate content (CC). The horizontal axis represents factor levels, while the vertical axis shows UCS, CC, or expected value. Curves of different colors indicate varying levels of the interacting factor, with other factors held constant at moderate levels.

Fig. 9.

Interaction plots (a–c) A–B interaction and (d–f) A–C interaction. Horizontal axes represent factor levels; vertical axes indicate UCS (MPa), CC (%), or desirability. Colored curves show the effects of interacting factors, while other factors are held at moderate levels. Key interactions for UCS and CC are highlighted; additional interaction plots are provided in the Supplementary Materials. Curve intersections and shaded regions indicate relative contributions of the factors to the response variables.

To quantify the relative influence of two interacting factors, a geometric curve-flattening method was employed. The area between intersecting curves is approximated as an obtuse or acute triangle: obtuse triangles indicate that the lower-axis factor exerts a stronger influence (+ a), while acute triangles indicate a weaker contribution (− a). Comparing the overhang distance (OA) to the triangle base (AB) provides an intuitive measure of factor dominance.

For example, Fig. 9a illustrates that overall confidence increases progressively with rising levels of factor A but declines at higher moisture contents. In Fig. 9b, the intersection of moisture content (A) and reaction time (B) divides the response into left and right regions. On the left, OA AB (obtuse triangle), showing factor A (lower axis) dominates UCS; on the right, OA < AB (acute triangle), indicating a stronger effect of factor B (upper axis). Additionally, the two dashed lines in Fig. 9b represent the 95% confidence intervals, within which the red and black solid lines fluctuate, demonstrating the high reliability of the model predictions.

A detailed analysis of Fig. 9b shows that at low moisture contents, the red curve (high reaction time, B) lies above the black curve (low B), indicating that longer reaction times increase UCS; above a threshold, the black curve surpasses the red, suggesting that shorter reaction times dominate at high moisture. The blue-shaded region, comprising obtuse triangles, confirms that factor A exerts a more pronounced effect on UCS than factor B, highlighting its dominant contribution. In contrast, Fig. 9c demonstrates that CC peaks at moderate moisture content, with the red curve consistently above the black curve across the moisture range, indicating that prolonged reaction time predominantly influences CC; the acute triangle at the leftmost intersection underscores the lesser role of factor A.

Interactions with fiber length (C) reveal that UCS follows a similar pattern to the A–B interaction (Fig. 9e), though curve intersections shift left due to fiber entanglement at longer lengths. Shorter fibers enhance UCS via uniform distribution and reduced entanglement, with obtuse triangles indicating factor A dominance at low fiber lengths and acute triangles highlighting increased C influence at higher lengths. CC trends (Fig. 9f) show that shorter fibers consistently promote precipitation. Especially at higher moisture, as excessively long fibers hinder material transport within pores²⁴. The acute triangle confirms factor C’s greater impact on CC relative to factor A. For the B–C interaction, UCS initially rises and then falls with reaction time (Fig. S7b), showing that short fibers favor strength, which is consistent with the findings of Lim et al.²⁵. CC increases under long-fiber conditions (Fig. S7c), reflecting the dominant role of factor B.

Interactions involving fiber content (D) indicate that UCS generally follows moisture and reaction time trends, with obtuse triangles confirming factor A’s stronger influence, while CC is more sensitive to fiber content, as acute triangles demonstrate factor D predominance (Fig. S7d–f). Notably, differences in the B–D (Fig. S8a–c) interaction between UCS and CC highlight that increased precipitation does not always translate into higher strength, as UCS is primarily governed by pore filling rather than total mineral content²⁶.

Finally, interactions between fiber length and fiber content (C–D) reveal that both UCS and CC initially increase and then decrease with fiber length, with obtuse triangles emphasizing the dominant role of C (Fig. S8d–f). The figure also reveals that short fibers provide a relative advantage in strength enhancement, a conclusion supported by previous studies^9,10. Despite minor deviations at extreme factor levels, the observed trends consistently align with prior analyses, confirming the robustness and reliability of the results.

Cube plot

The cube plot in Fig. 10 illustrates the combined effects of moisture content (A), reaction time (B), and fiber content (D) on UCS and CC, with fiber length (C) held constant. UCS decreases when fiber content exceeds an optimal range, as excessive fibers disrupt uniform CaCO₃ deposition, weaken interparticle bonding, and reduce compressive strength²⁷.

Fig. 10.

Cube plots (a) desirability (b) unconfined compressive strength (c) CaCO₃ content.

Confidence levels rise with increasing moisture, indicating that UCS is strongly influenced by the interaction of moisture, reaction time, and fiber content. Moisture facilitates microbial activity and Ca2⁺–CO₃2⁻ transport, while reaction time regulates the extent and morphology of mineral deposition. Optimal conditions occur at moderate factor levels rather than extremes. Figure 10a shows maximum desirability (0.979), while Fig. 10b, c shows peak UCS (4.301 MPa) and peak CC (3.22%) within the cube interior, highlighting the cooperative interaction between microbial activity, mineralization, and fiber reinforcement.

Ramp graphical representation of numerical optimization

The ramp plot in Fig. 11 visualizes the optimized parameter ranges for maximizing UCS and CC in GF-MICP-reinforced sediments. The theoretical maximum UCS is predicted at 4.301 MPa with a corresponding CC of 3.22%, surpassing reported 1.458 MP²⁸.

Fig. 11.

Ramp plots that encompass factors such as moisture content, reaction time, fiber length, fiber content.

To ensure practical feasibility and scalability, the optimal factors were slightly adjusted: moisture content (A) to 40.00%, reaction time (B) to 4 days, fiber length (C) to 6 mm, and fiber content (D) to 2.44%. Under these conditions, predicted UCS and CC were 4.290 MPa and 3.21%, respectively, closely matching the theoretical optimum.

Experimental validation was conducted with three repeated tests. The measured mean UCS and CC were 4.293 MPa and 3.20%, with standard deviations of 0.004 MPa and 0.008%, demonstrating excellent reproducibility. Comparison with model predictions (Tables S2–S3) shows that all experimental values fall within the 95% confidence intervals (UCS: 4.192–4.388 MPa; CC: 3.14–3.27%), confirming the reliability and accuracy of the numerical optimization.

Mechanisms of microstructure

Scanning electron microscopy (SEM) and X-ray diffraction (XRD) analyses were conducted to investigate the synergistic effects of fibers and CaCO₃ precipitation on sediment microstructure, and their influence on unconfined compressive strength (UCS) and calcium carbonate (CC) content. SEM revealed the structural interactions, while XRD provided mineralogical characterization, enabling a link between microstructure evolution and macroscopic mechanical behavior.

Microstructural analysis under SEM

Figure 12 presents representative SEM images of fractured surfaces. Figure 12a shows larger sediment particles coated with CaCO₃, with glass fibers embedded in the unfilled voids. Three principal mechanisms are observed: (1) confinement of sediment particles by fibers, (2) stress transfer along fibers and through fiber networks, and (3) cementation and encapsulation by CaCO₃ precipitates²⁹.

Fig. 12.

SEM images (a–d) under optimized conditions, (e) under 9mm fiber; (f–g) under 3.5% fiber.

Fiber–CaCO₃ interactions form composite fiber columns, as seen in Fig. 12b, which remain attached to sediment particles even after fracture, indicating strong interfacial bonding and effective stress transfer. This structure significantly enhances the interlocking effect at the interfaces, allowing the precipitate-coated fiber columns to effectively transfer external stresses while benefiting from both the strength of the precipitate and the high resistance of the fibers.

CaCO₃ precipitates fill interstitial spaces or wrap around fibers, forming a 3D network that stabilizes pores and distributes stresses efficiently (Fig. 12c, d). This process leads to the cementation of multiple fibers, forming a bridge-like structure that stabilizes sediment pore spaces. Within these pores, fibers provide a stress buffer, effectively preventing pore collapse.

Fiber length and content critically affect UCS: excessively long fibers cluster (Fig. 12e, f), disrupting stress transfer, whereas moderate-length fibers distribute evenly, maximizing reinforcement. Excessive fiber content can impede CaCO₃ deposition and reduce mechanical performance, indicating the necessity for optimal fiber parameters²⁴. Meanwhile, our SEM results confirm that fiber addition promotes uniform CaCO₃ nucleation, structural bridging, and effective stress transfer, leading to UCS improvement.

Mineral analysis under XRD

XRD spectra of the sediment feedstock, fiber-free, and fiber-added samples are shown in Fig. 13. Quartz and albite from sediments were observed, along with calcite produced by strain CM11. Fiber addition increased the number of calcite diffraction peaks (Fig. 13c) and raised calcite content from 10.5 to 11.5% (Fig. 14), indicating that fibers act as nucleation sites and enhance precipitation.

Fig. 13.

XRD patterns of sediment samples (a) raw material sediment, (b) without added fibers, (c) with added fibers.

Fig. 14.

XRD quantitative analysis of residual (a) with added fibers (b) without added fibers.

It is noteworthy that the XRD spectra in Fig. S2 show that the precipitates generated by the strain CM11 contain calcite and vaterite. Although vaterite peaks were not detected in Fig. 13, SEM (Fig. S6 and 12d) shows vaterite layers on fiber and sediment surfaces. This discrepancy is attributed to low vaterite crystallinity and its surface deposition, which may be masked by larger grains during XRD analysis. The combination of calcite and surface-bound vaterite forms a crystal-bridging network that stabilizes pores and improves interparticle cementation, contributing to UCS enhancement. Our results demonstrate that fiber-assisted MICP promotes structural reinforcement and uniform stress distribution, directly linking microstructure to UCS improvements.

Conclusions

This study systematically investigated the enhancement of MICP-treated sediments through the incorporation of glass fibers (GF) and optimization of treatment parameters. The main findings are summarized as follows:

1. Optimal treatment parameters: Response surface methodology (RSM) modeling and validation experiments identified 40% moisture content, 4 days reaction time, 6 mm GF length, and 2.44% fiber content as the optimal conditions. Under these conditions, the predicted UCS and CaCO₃ content were 4.29 MPa and 3.21%, respectively, with experimental results closely matching predictions, demonstrating the high reliability of the model.

2. Key influencing factors: Water content was found to be the dominant factor controlling UCS, while reaction time primarily influences calcium carbonate content. Fiber properties (length and content) interact with MICP parameters, where excessive fiber length or content can reduce UCS due to fiber clustering and impaired stress transfer.

3. Synergistic mechanism of GF-MICP: SEM and XRD analyses showed that GF provides nucleation sites for CaCO₃ precipitation, promotes fiber-mineral bridging, and stabilizes sediment pore structures. Calcite deposition increased from 10.5% to 11.5% with fiber addition, reinforcing the cementation effect. The combination of GF and CaCO₃ forms a composite network that effectively enhances stress transfer, adhesion, and mechanical strength of sediments.

4. Practical implications: These findings highlight the potential for both mechanical performance enhancement and cost reduction in MICP-treated sediments. By controlling moisture content, reaction time, and fiber parameters, GF-MICP treatment can be applied effectively in sediment stabilization and geotechnical engineering, offering a feasible approach for practical field implementation.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

Ximei Cai: Writing—review & editing, Writing—original draft, Methodology, Investigation. Shengli Wang: Supervision. Yifan He: Investigation. Tiantian Liang: Investigation. Fan He: Investigation. Chongbin Li: Investigation.

Funding

This study was financially supported by Gansu Provincial Department of Natural Resources 2024 Scientific and Technological Innovation and Application Research Project “Study on water acidification mechanism of typical massive sulfide deposits”(202410), Key Research and Development Program of Gansu Academy of Sciences(2025ZDYF-02), and Gansu Province Provincial Technology Innovation Guidance Plan Science and Technology Small and Medium Enterprises Innovation Fund Special Project “Research and development of key technologies for groundwater pollution prevention and control in closed pit mines in the Yellow River Basin of Gansu”(24CXGG003).

Data availability

The datasets used and/or analysed during the current study available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.