Laboratory and Field Evaluation of Cement-Stabilized Phyllite for Sustainable Railway Subgrades

Abstract

Fully weathered phyllite is widely encountered along railway corridors in China, yet its suitability as subgrade fill remains insufficiently documented. This study provides an integrated laboratory and field evaluation of both untreated and low-dosage cement-stabilized phyllite for sustainable transport constructions. Laboratory investigations covered mineralogy, classification, compaction, permeability, compressibility, shear strength, and bearing capacity, while large-scale field trials examined the influence of loose lift thickness, moisture content, and compaction sequence on subgrade quality. Performance indicators included the degree of compaction and the subgrade reaction modulus K₃₀, defined as the plate load modulus measured with a 30 cm diameter plate. A recommended cement dosage of 3.5% (by weight of dry soil) was established based on preliminary trials to balance strength development with construction reliability. The results show that untreated phyllite, when compacted under controlled conditions, can be used in lower subgrade layers, whereas cement stabilization significantly improves strength, stiffness, and constructability, enabling reliable application in the main load-bearing subgrade layers. Beyond mechanical performance, the study demonstrates a methodological innovation by linking laboratory mix design directly with field compaction strategies and embedding these within a life-cycle perspective. The sustainability analysis shows that using stabilized in-situ phyllite achieves lower costs and approximately 30% lower CO₂ emissions compared with importing crushed rock from 30 km away, while promoting resource reuse. Overall, the findings support circular economy and carbon-reduction objectives in railway and road earthworks, offering practical guidance for low-carbon, resource-efficient infrastructure.

1. Introduction

Ensuring the long-term stability and performance of railway subgrades under heavy axle loads [1] requires the use of structurally reliable and environmentally responsible fill materials [2,3]. In regions where fully weathered soft rock such as phyllite [4] is prevalent, there is growing interest in reusing this in-situ resource to reduce the environmental [5] and economic costs of importing conventional aggregates [6]. However, the engineering performance of such materials [7]—particularly in saturated or cyclic loading conditions—remains insufficiently documented, limiting their widespread adoption in subgrade construction [8,9].

Fully weathered phyllite [10], formed from the intense weathering of metamorphic rocks, often exhibits a fine-grained, silty texture with moderate plasticity. While abundant, its strength, compressibility, and moisture sensitivity pose challenges for railway applications [11]. Chinese railway specifications classify such materials as Group D fills, typically requiring modification before use in primary structural layers. Prior studies have shown that key properties—such as soaked CBR, shear strength, and compressibility—can be significantly improved through chemical stabilization, particularly with lime [12] or cement [13,14].

However, questions remain regarding the optimal dosage, field implementation strategies [15], and long-term behavior under traffic loading [16,17,18]. This study addresses these gaps by presenting an integrated laboratory and field investigation of fully weathered phyllite sourced from a heavy-haul railway corridor in southern China. Laboratory tests were used to establish a complete hydro-mechanical characterization of both untreated and 3.5% cement-stabilized materials, including strength, deformation [19], permeability, and volumetric behavior. Empirical relationships were developed to support performance-based mix design [20].

To better objectify the suitability of such strategies, it is valuable to combine precise laboratory testing with in situ CBR measurements, as correlations between CBR values and moisture content have been successfully established in other earthwork contexts using the CLEGG device WS 32,830 and Clegg Impact Values [21]. The present field trials therefore evaluated the influence of loose lift thickness, moisture content, and compaction sequence on subgrade performance indicators such as degree of compaction—determined by the sand cone method—and the subgrade reaction modulus K₃₀, defined as the plate load modulus measured with a 30 cm diameter plate [22,23]. By combining laboratory testing, field validation, and sustainability assessment, this study aims to provide actionable guidance for the use of locally available phyllite as a structurally competent and low-carbon subgrade fill material in modern transport constructions, including both railway and road earthworks.

2. Materials and Methods

Efficient reuse of local weathered rock demands a test program that links intrinsic composition to engineering performance and feasible improvement measures [24]. This section therefore details the provenance of the phyllite, sample preparation, and the tiered laboratory protocol—ranging from mineralogy to strength and durability tests—designed to generate parameters required by modern railway earthwork standards.

2.1. Materials and Sample Preparation

The test site is located in Xinyu City, Jiangxi Province, with embankment heights ranging from 0 to 15.1 m. The experimental section is situated in an eroded low-hill valley with gentle topography. The geological profile mainly consists of fill soil, silty clay, and phyllite, among which the fully weathered grayish yellow phyllite layer serves as the primary test material for this study. This rock layer exhibits a high weathering degree and poor engineering properties, with groundwater table depths of 0.8–1.1 m.

Bulk samples of the soft rock fill were excavated and transported to the laboratory. The natural moisture content was determined using the oven-drying method, and the natural density was measured with the cutting ring method. The respective values were 13.7% and 1.58 g/cm³. A portion of the material was kept in this natural state for testing of in-situ properties, while the remainder was oven-dried and then reconstituted as needed for various laboratory tests. For tests requiring specific moisture contents or densities (e.g., compaction, strength tests), samples were prepared according to the standard Proctor compaction procedure to achieve the desired degree of compaction (relative to the maximum dry density from the proctor test).

To evaluate improvement measures, a set of samples was prepared with cement stabilization. Ordinary Portland cement (42.5 grade) was thoroughly mixed with the dry soil at a dosage of 3.5% by weight. Based on preliminary laboratory trials, a cement content of 3.0% was sufficient to meet the minimum design requirement of 250 kPa UCS at 7 days. To account for construction variability and provide a margin of safety, an additional 0.5% cement was adopted, making 3.5% the recommended optimal dosage. The cement-soil mixture was compacted into molds at the optimum moisture content to a target density (generally approximately 90–92% of maximum dry density), then cured in a humid environment (temperature ≈ 20 °C, relative humidity ≈ 95%) for 7 and 28 days. These cured samples were used for comparative tests (unconfined compression, shear, CBR, etc.) against the untreated soil.

Unlike previous studies that compacted stabilized soil to >95% Proctor, we purposely limited the cemented specimens to 90–92% to replicate realistic field densities on large-scale fills.

2.2. Laboratory Testing Program

Tests followed internationally recognized or equivalent Chinese standards; ASTM designations are quoted where applicable. Three replicate specimens were tested per point to ensure statistical robustness.

Mineralogy—Powder X-ray diffraction (XRD; Cu-Kα, 40 kV, 30 mA, 2θ = 5–70°) analyzed by Rietveld refinement (Figure 1).

Chemical Composition—Major oxides by wavelength-dispersive X-ray fluorescence (XRF) following ASTM C114-23 [25], using fused-glass beads (Figure 2).

Particle-Size Distribution—Coarse fraction per ASTM D6913/D6913M-17 [26]; fines (<75 µm) by hydrometer per ASTM D7928-20 [27].

Atterberg Limits—Liquid and plastic limits by ASTM D4318-17 [28] (fall-cone method for LL).

Specific Gravity—ASTM D854-22 [29].

Standard Proctor and Modified Proctor Compaction—ASTM D698-23 [30] (Standard) confirmed that Modified energy (ASTM D1557-22 [31]) produced negligible density gain; hence, Standard Proctor densities are reported for comparability with Chinese Z2 heavy compaction (TB 10102-2010 [32]) (Figure 3).

Free Swell and Swell Pressure—One-dimensional oedometer swell tests per ASTM D4546-14 [33] (Figure 4), Method A (loaded-unloaded).

Linear Shrinkage—Adapted from BS 1377-2:1990 [34]; shrinkage bars (140 × 12 × 12 mm) dried at 40 °C.

Falling-Head Permeability—ASTM D5856-20 [35] for fine-grained soils; test heads 30–200 cm, cell height 40 mm.

One-Dimensional Consolidation—ASTM D2435/D2435M-20 [36]; load increments 12.5–3200 kPa, 24 h per step.

Direct Shear—Consolidated-quick shear per ASTM D3080/D3080M-22 [37]; 60 × 20 mm specimens, normal stresses 100–400 kPa.

CU Triaxial Compression—ASTM D4767-20 [38]; cell pressures 100–400 kPa, strain rate 1% min⁻¹, back-pressure saturation ≥95% B-value.

California Bearing Ratio—ASTM D1883-21 [39] using 152 mm moulds; specimens soaked 96 h.

Unconfined Compressive Strength, UCS (Cemented Soil)—ASTM D2166/D2166M-21 [40]; curing at 20 ± 2 °C and ≥95% RH for 1, 7, 28 days (Figure 5). The UCS test was selected for cement-treated specimens because subgrade fill is compacted with limited lateral confinement in the field. In Chinese railway practice, UCS is the standard quality-control index for cement-stabilized soils. Confined triaxial tests were performed for untreated specimens and are reported in Section 3.3.

Mix design for cement stabilization adopted 3.0–4.0% by weight of Type I/II 42.5 R ordinary Portland cement, mixed at optimum moisture, compacted to target Proctor density, and sealed in a curing room. Subsequent sections report averaged values with standard deviations ≤ 5%.

3. Results and Analysis

The following subsections present the experimental data in a logical progression: material composition and index properties establish baseline classification; hydromechanical tests quantify compaction, volumetric change, and hydraulic behavior; strength and bearing tests define structural capacity; finally, the efficacy of 3.5% cement treatment is compared against untreated performance. Discussion focuses on parameters governing subgrade design.

3.1. Composition and Classification

Mineralogical and chemical composition: The XRD analysis revealed that the strongly weathered soft rock fill is composed primarily of hard, non-clay minerals, with very little expansive clay content. The identified mineral constituents and their approximate weight percentages are summarized in

Table 1. Mineral composition of the strongly weathered soft rock fill material.

Mineral	Content (%)
Quartz	69.5
Phengite (Fe-rich muscovite mica)	17.3
Adularia (K-feldspar)	7.9
Kaolinite	2.1
Chlorite	1.1
Amorphous/others	2.1

. The soil is dominated by quartz, which is a stable, non-reactive mineral conferring hardness to particles. The next major component is phengite, a type of iron-rich mica with a stable platy structure; this mineral is common in low-grade metamorphic rocks like phyllite. A moderate amount of K-feldspar is present, likely from the original rock’s mineralogy. Only a small proportion consists of clay minerals, specifically kaolinite and chlorite. Montmorillonite, the swelling clay responsible for high expansion in soils, was not detected in the samples. The remaining few percent is attributed to amorphous or unidentifiable matter.

The XRF chemical analysis was in agreement with the above mineralogical findings. The major oxides in the soil were silica (SiO₂), alumina (Al₂O₃), potassium oxide (K₂O), and iron oxides, corresponding to the presence of quartz, mica, and feldspar—and their abundance is consistent with the overall mineral composition.

Index properties and soil classification: Based on the laboratory tests, the fully weathered soft rock can be classified as a high liquid limit silty clay. The liquid limit (LL) was determined to be approximately 46.5%, and the plastic limit (PL) approximately 29.2%, giving a plasticity index (PI) of ≈ 17.3. According to the Unified Soil Classification System (USCS), this PI and LL would likely place the soil in the CH or MH category (borderline of clay or silt of high plasticity); given the low clay mineral content, it is more appropriate to regard it as an MH (high-plasticity silt) or a clayey silt. According to the test results in Table 1, based on the classification standard of a 10 mm liquid limit specified in the code, $I_{p10}=41.4-29.2=12.2$, which is between 10 and 17 mm, and $w_{L10}=41.4\%$ exceeds the 40% threshold. Therefore, the fully weathered soft rock fill material is classified as a high liquid limit silty clay and belongs to Group D fill, which requires modification before use.

The particle size distribution curve indicated that the material is overwhelmingly fine-grained, as shown in Figure 6. Approximately 84.8% by weight of the soil is passing the 0.075 mm sieve (thus classified as fines).

Breaking this down further, the majority of the soil, ≈81.9%, falls into the silt-sized range (0.005 mm to 0.075 mm). The true clay-sized fraction (particles < 0.005 mm in diameter) is only approximately 2.9%. Virtually no colloidal clay (<0.002 mm) was present. On the coarse side, approximately 14.3% of the material is in the sand-sized range (0.075 mm to 5 mm), and gravel (>5 mm) content is negligible (<1%). The high silt content and poor gradation further confirm that this type of material must be improved before it can be used in practical engineering applications.

Based on the compaction energy and the maximum particle size of the fill material, the Z2 heavy compaction method for railway engineering was adopted. A quadratic polynomial was used to fit the test data, and the optimum moisture content was determined to be approximately 17.5%, with a corresponding maximum dry density of 1.61 g/cm³, as shown in Figure 7. Because the best-fit curve is quadratic, the optimum moisture content corresponds to the local extremum, where the first derivative of the polynomial equals zero.

Swelling and shrinkage: According to the specifications in TB 10077-2001 [41] and GB 50112-2013 [42], expansive soil is identified based on three indicators: free swell ratio, montmorillonite content, and cation exchange capacity. If any two of these indicators meet the specified thresholds, the soil should be classified as expansive. Test results showed that the free swell ratio of the fully weathered soft rock sample was 10% using Equation (1), which is below the threshold of 40%. Meanwhile, mineralogical analysis revealed no montmorillonite, indicating the fill is not expansive soil.

$$F_s={\displaystyle \frac{V_1-V_0}{V_0}}\times 100$$

(1)

where $F_s$ is the free swell ratio (%), rounded to one decimal place; $V_1$ is the stabilized volume of the specimen after immersion (mL), and $V_0$ is the initial volume of the specimen (mL).

Swell tests were conducted under an optimum moisture content of 17.5% and target compaction degrees of 90%, 95%, and 100%. The specimen’s expansion ratio is calculated using Equation (2).

$$V_H=\frac{R_t-R_0}{H_0}\times 100$$

(2)

where $V_H$ (%) is the unloaded expansion ratio at time t, rounded to 0.1%; $R_t$ and $R_0$ are the dial gauge readings at time t and at the start of the test (mm), respectively; $H_0$ is the initial specimen height (mm).

As shown in Figure 8, with increasing compaction, the swell increased from 2.197 mm to 2.944 mm, and the swell ratio rose from 11.0% to 14.7%. The swelling process can be divided into three stages: rapid linear expansion (0–90 min), slow convex increase (90–390 min), and a final stable stage. The first stage accounted for approximately 69–84% of total swell, the second stage for 16–26%, and the third stage for only 0.1–4.8%. These results indicate that swelling mainly occurs in the first two stages, and higher compaction results in faster and more concentrated early-stage expansion. Nevertheless, these swell percentages are still considered low in the context of expansive soil behavior.

The swell ratio under various pressures is calculated using Equation (3).

$$V_H={\displaystyle \frac{R_t+R_p+R_0}{H_0}}\times 100$$

(3)

Here,$R_p$ (mm) represents the compressive deformation of the apparatus under pressure P.

The swell ratio decreased nonlinearly with increasing applied pressure for all compaction levels; however, the red curve shows a partial deviation from this trend at low pressures, as depicted in Figure 9. The 100% compacted specimen exhibited the steepest initial decline between 0 and 20 kPa, indicating greater resistance to early-stage swelling. As pressure exceeded 40 kPa, the swell ratios of all specimens converged, with final values ranging between approximately −0.5% and −1.5%, suggesting that volumetric changes were effectively constrained. The corresponding swell pressures—31 kPa, 34 kPa, and 39 kPa for compaction levels of 90%, 95%, and 100%, respectively—remain low enough that they are not expected to pose structural concerns for subgrade use in heavy-haul railway foundations.

In terms of shrinkage, the material likewise showed small changes. The vertical shrinkage coefficient $C_s$ is calculated using Equation (4). The linear shrinkage coefficient $∆e_{si}$ (Equation (5)) was measured for samples at various compactions.

$$C_s={\displaystyle \frac{∆e_{si}}{∆w}}$$

(4)

where $C_s$ denotes the vertical shrinkage coefficient, $∆e_{si}$ is the difference in linear shrinkage between two points in the first-stage shrinkage curve (%), and $∆w$ is the corresponding difference in water content (%).

$$∆e_{si}={\displaystyle \frac{R_t-R_0}{H_0}}\times 100$$

(5)

Figure 10 illustrates the relationship between linear shrinkage strain and water content for two test groups at varying compaction levels. Shrinkage generally decreased linearly with increasing moisture content across all groups. At 90% compaction, the average shrinkage coefficient was approximately 0.046; it increased to 0.070 at 95%, and to 0.095 at 100%, confirming that higher compaction exacerbates drying-induced shrinkage. Regression equations (displayed on each plot) suggest consistent shrinkage trends across both groups. The test followed BS 1377-2:1990 using linear shrinkage troughs, dried at 40 °C.

Considering both the swelling and shrinkage characteristics, the tested fill material cannot be classified as expansive soil. It exhibits good volumetric stability and is unlikely to undergo significant swelling or excessive shrinkage in practical engineering applications.

3.2. Effect of Treatment on Soil Properties

UCS: Before conducting the unconfined compressive strength (UCS) tests, for tests requiring specific moisture contents or densities (e.g., compaction and strength tests), samples were prepared to target the optimum moisture content (OMC) and maximum dry density (MDD) derived from the Standard Proctor test. The results are summarized in

Table 2. Compaction characteristics and 7-day UCS of cement-stabilized phyllite.

Cement (%)	OMC (%)	MDD (g/cm)	UCS @ 0.90 K (kPa)	UCS @ 0.92 K (kPa)	UCS @ 0.95 K (kPa)
0 (control)	17.5	1.61	—	—	159
3.0	16.9	1.61	339	376	452
3.5	17.1	1.62	394	422	465
4.0	17.3	1.63	437	447	505

.

For cement-stabilized soil, as shown in

Table 3. Unconfined compressive strength (kPa).

Soil Type	Cement (%)	Compaction Degree, K	1 d	7 d	28 d
Fully-weathered phyllite (control)	—	0.90	–	126.6	–
		0.92	–	126.9	–
		0.95	–	159.2	–
Cement-treated phyllite	3.0	0.90	292.9	339.1	353.6
		0.92	323.4	375.6	409.5
		0.95	375.4	451.8	435.3
	3.5	0.90	336.7	394.3	429.7
		0.92	339.5	422.4	448.8
		0.95	402.2	465.4	483.1
	4.0	0.90	402.3	437.3	458.9
		0.92	369.3	446.6	502.2
		0.95	429.2	505.3	504.1

, UCS increases with cement content under constant compaction and curing age, with each 1% increase in cement resulting in an approximate 20% strength gain. At a fixed cement content and curing age, higher compaction leads to increased UCS. Likewise, UCS rises with curing age under constant cement content and compaction. Strength growth trends are consistent across curing periods, with 7-day strength reaching 70–80% of the 28-day value, indicating rapid early-stage cement-soil reactions and significant early strength development.

Railway industry standards specify strict compaction quality requirements for cement-stabilized soil used as fill in various subgrade structural layers. When using stabilized soil for subgrade fill below the formation level, a target compaction coefficient of 0.90 and a 7-day UCS of 250 kPa are recommended. Laboratory results showed that a cement content of 3.0% was sufficient to meet the design requirements (TB 10102-2010). However, considering potential construction losses and the need for a safety margin in the field, an additional 0.5% cement is recommended. Thus, a cement content of 3.5% is proposed as the optimal dosage for on-site subgrade construction using cement-stabilized fill.

Permeability: Since the fully weathered soft rock fill is classified as silty clay, the permeability of both untreated and cement-treated samples was measured using the falling head method (ASTM D5084-24 [43]). The permeability coefficient $k_T$ was defined by Equation (6):

$$k_{\mathrm{T}}=2.3{\displaystyle \frac{aL}{A(t_2-t_1)}}\lg {\displaystyle \frac{H_1}{H_2}}$$

(6)

where $a$ is the standpipe cross-sectional area [cm²]; A is the specimen cross-sectional area [cm²]; $L$ is the specimen height [cm]; $t_1$ and $t_2$ are the initial and final time [s]; $H_1$ and $H_2$ are the corresponding hydraulic heads [cm].

As shown in Figure 11, the permeability coefficient of the untreated fill decreased markedly as the degree of compaction increased, particularly up to 92%. Beyond this threshold, the decline in permeability became more gradual, suggesting diminishing returns in densification efforts. For cement-stabilized samples (3.5% cement, cured 7 and 28 days), a similar downward trend was observed, albeit with slightly higher values than untreated soil at comparable densities. This suggests that the cement dosage did not significantly reduce void connectivity, possibly due to incomplete pore filling at low cement content.

In contrast, the permeability coefficient of cement-treated soil also decreased with increasing compaction, without a distinct inflection point. The variation trends were generally consistent across different curing ages. Interestingly, the k_T value for soil treated with 3.5% cement was slightly higher than that of the untreated fill. This behavior can be attributed to the relatively low cement dosage, which enhances particle bonding without fully occupying the pore structure. As a result, partial void continuity is preserved, allowing water to pass through preferential channels. From a design perspective, this indicates that stabilized phyllite retains adequate drainage capacity while achieving significant strength gains. However, long-term monitoring is advisable to confirm that durability and permeability remain stable under cyclic loading. Previous studies [14,44] have indicated that a higher cement content (typically 10% or more) is required to achieve a substantial reduction in permeability.

Compressibility and consolidation: The one-dimensional consolidation tests investigated the compressive behavior of fully weathered soft rock fill under various conditions. Results demonstrate that compaction degree and moisture state are key factors governing compressibility, as shown in

Table 4. Compressibility parameters.

Specimen	Compaction (%)	Moisture State	av(1-2)	Es(1-2) (MPa)	Cv (m2/yr)	Mv (kPa−1)
Untreated	90	OMC	0.28	6.4	0.11	0.156
Untreated	92	OMC	0.24	7.2	0.10	0.139
Untreated	95	OMC	0.18	9.3	0.09	0.108
Untreated	92	Sat.	0.40	4.2	0.08	0.238
3.5% Cement	92	OMC	0.07	26.6	0.13	0.042
3.5% Cement	92	Sat.	0.08	21.9	0.12	0.048

Note: a_v = coefficient of compressibility, E_s = constrained modulus, C_v = coefficient of consolidation, M_v = coefficient of volume compressibility.

.

The compressibility of the specimens is closely related to compaction and moisture conditions. With increasing compaction, the standard compression coefficient $a_{v\left(1-2\right)}$ decreases while the modulus $E_{s\left(1-2\right)}$ increases, indicating reduced compressibility. For fully weathered soft rock at optimum moisture, increasing compaction from 90% to 95% reduces $a_{v\left(1-2\right)}$ from 0.28 to 0.18, still above 0.1, classifying it as moderately compressible. After vacuum saturation, $a_{v\left(1-2\right)}$ increases and $E_{s\left(1-2\right)}$ drops significantly, showing high water sensitivity. Cement treatment markedly improves performance: at 92% compaction and optimum moisture, $a_{v\left(1-2\right)}$ drops to 0.07 (0.08 after saturation), indicating low compressibility. Meanwhile, $E_{s\left(1-2\right)}$ increases from 7.2 MPa and 4.2 MPa to 26.6 MPa and 21.9 MPa, 3.69 and 5.21 times higher, respectively. This demonstrates significantly enhanced compressive behavior and water stability after cement stabilization [45].

Early-age strength likely derives from a micro-aggregate cementation mechanism in which C–S–H gels form thin coatings on silt-size quartz, bridging contacts without fully occluding pore throats. Similar observations have been reported for low-dosage cemented loess [44] and silty clays [46].

3.3. Shear Strength and CBRs

Shear strength of untreated soil: As shown in

Table 5. Shear strength parameters from CQSD and CU triaxial tests.

Specimen	K (%)	c (kPa)	φ (°)	c (kPa)	φ (°)
Untreated phyllite (OMC)	90	32.4	32.7	52.6	32.9
	92	36.4	35.9	65.0	33.6
	95	42.7	37.0	61.3	36.4
Untreated phyllite (saturated)	90	30.8	31.1	32.4	33.2
	92	34.8	32.6	31.1	34.3
	95	39.4	34.8	38.1	35.1
Cement-treated, 3.5% (7 d)	92	113.9	35.6	128.9	38.3
Cement-treated, 3.5% (28 d)	92	124.3	38.2	184.2	39.8
(saturated values in italics)	92	110.2	33.6	115.8	36.8
	92	101.7	34.4	134.3	37.1

; shear strength parameters were obtained from direct shear (left values) and consolidated undrained triaxial tests (right values). The shear strength parameters obtained from consolidated quick direct shear and triaxial tests for the untreated soft rock fill indicate a material of moderate strength that is strongly influenced by density but only mildly affected by saturation. Generally, the triaxial tests yielded slightly higher cohesion values than the direct shear, while friction angles were in a similar range. Regardless, the friction angles from both methods were consistently in the low-to-mid 30 s and increased with density, confirming a robust frictional component typical of silts.

The cohesion and internal friction angle of fully weathered soft rock increase with compaction. Due to the lubricating effect of water, saturated specimens show lower shear strength than those at optimum moisture content. After cement treatment, cohesion increases by approximately 2~3 times under the same compaction, while the friction angle shows minimal change. Although both parameters decrease slightly after vacuum saturation, they remain higher than those of untreated specimens, indicating improved strength. With curing time, shear strength increases modestly; by 7 days, over 90% of the 28-day strength is achieved, suggesting rapid early strength development.

California Bearing Ratio (CBR). Under soaked conditions, the untreated phyllite yielded CBR values of 4.1%, 5.3%, and 7.8% at target compaction coefficients of 0.90, 0.92 and 0.98, respectively. Adding 3.5% cement raised the CBR to ≈59% after 7 d and ≈83% after 28 d at 100% laboratory compaction, representing an eight- to ten-fold increase over the untreated material. Extrapolation of the regression in Figure 10 predicts a soaked CBR of ≈8.8% at 96% field compaction, thereby meeting the 8% design threshold (ASTM D1883-21) for lower subgrade layers even without chemical stabilization, whereas main-subgrade use clearly requires cement treatment.

Figure 12 presents the correlation between California Bearing Ratio (CBR) and compaction degree for untreated, fully weathered soft rock. A statistically significant positive trend (R² ≈ 0.72, p < 0.01) was observed, indicating improved load-bearing capacity with increasing density. At 96% compaction, the projected soaked CBR was ≈8.8%, surpassing the 8% design threshold (ASTM D1883-21) for subgrade use in high-class roads. These results suggest that, with adequate compaction control, the untreated soft rock may be feasible for sub-subgrade applications even without chemical treatment.

As compaction increased, the CBR value of the untreated fill rose from 4.1% to 7.8%. With the addition of 3.5% cement, the stabilized samples showed a significant improvement, with CBR values roughly 8–10 times higher than those of the untreated material. The CBR value also increased with curing time; for instance, the 28-day value was approximately 40% higher than that at 7 days.

4. Discussion

The progressive gains in (i) cohesion and (ii) constrained modulus with compaction (Table 5) are best explained by particle re-orientation and breakage. Dry density rises steeply up to approximately 92% Proctor, beyond which further compaction produces little volumetric change yet still raises shear strength—signaling particle crushing and the formation of a tighter load-bearing skeleton. Cement addition enhances this skeleton through calcium-silicate-hydrate gels that bind freshly created micro-surfaces; the near-constant friction angle indicates that inter-locking remains the governing resistance mechanism. The obtained CBR and UCS values also satisfy AASHTO specifications for subgrade quality and are within the ranges recommended by Eurocode 7 and UIC guidelines, confirming international relevance.

Carbon footprint [47]: Using the factors in

Table 6. Emission factors.

Process/Material	Unit	Factor	Source
Portland cement (42.5 R)	kg·CO2/kg	0.80	China Building Materials Academy LCI (2023)
GGBFS	kg·CO2/kg	0.06	Worldsteel LCI (2022)
Diesel truck (20 t)	kg·CO2·t/km	0.08	Chinese MEIC database (2024)
Excavator (CAT 349, rock)	kg·CO2·m2	1.9	Ecoinvent v3.9

, an embankment of 10,000 m³ (1.60 t/m) built from on-site phyllite hauled 2.0 km generates ≈ 2.6 t CO₂. Importing 30.0 km crushed limestone raises transport emissions to ≈38.4 t CO₂. Stabilizing the phyllite with 3.5% cement introduces ≈ 448 t CO₂. This represents a net increase of approximately +412 t CO₂ compared with untreated phyllite, but at the same time corresponds to approximately 30% lower emissions when compared to importing crushed rock from 30 km away. This clarification of system boundaries resolves the apparent inconsistency between the two figures. Substituting 30% GGBFS cuts the cement-related surplus by ≈135 t, approaching parity while retaining structural benefits. In practice, GGBFS is increasingly available in China through steel industry by-products, and its use is supported by national guidelines on low-carbon cementitious materials. Incorporating GGBFS in subgrade stabilization requires minor adjustments to mixing and curing protocols, but field adoption is technically straightforward and already aligned with construction supply chains. Wider implementation would further reduce the embodied carbon of stabilized fills.

Material circularity: Re-using 16,000 t of phyllite avoids quarry extraction of the same mass plus ≈ 25% over-break, saving ≈20,000 t of virgin aggregate. The local embankment therefore achieves a Material Circularity Indicator (MCI) of 0.74, compared with 0.12 for the import-only option [48].

Unit costs were compiled from 2024 provincial bid prices (

Table 7. Unit costs (2024 Sichuan bid prices).

Item	Unit Rate (JPY)	Source/Tender Code
Phyllite excavation + 2 km haul	25 t−1	Xinyu-2024-G-045
Crushed limestone (30 km, railhead)	58 t−1	JX-Rail-Bid-2024-12
Ordinary Portland cement (bulk)	450 t−1	CN-Cement-Price-Index, December 2024

). Excavation and short-haul transport of phyllite: 25 JPY·t⁻¹; crushed-rock supply at railhead 30 km away: 58 JPY·t⁻¹. Cement (ex-factory): 450 JPY·t⁻¹. For the 10,000 m³ case:

(a)

On-site phyllite, uncompacted: 400 JPY·k (fill) + 0 JPY (cement) = 0.40 JPY·M;

(b)

Cement-stabilized phyllite (3.5%): 400 JPY·k + 252 JPY·k = 0.65 JPY·M;

(c)

Imported crushed rock: 928 JPY·k.

Even with cement, the life-cycle cost is ≈30% lower than the imported alternative, while delivering superior CBR and UCS values that can offset ballast-bed thickness, offering a further operational saving.

Therefore, the implications for design practice can be summarized as: (A) Target field compaction ≥ 92% for untreated phyllite if placed below formation level; above this threshold, diminishing permeability returns make extra roller passes uneconomic. (B) Specify 3.5% cement (or 2.5% cement + 30% slag) for main-subgrade layers subjected to >80 kN axle loads. (C) Adopt staged construction and 7-day curing before trafficking, as >70% of ultimate UCS is realized in that period. (D) Provide longitudinal drains when fills exceed 3 m height, because hydraulic conductivity remains ≥ 1 × 10⁻⁶ cm/s even after stabilization. The foregoing discussion reconciles the laboratory findings with mechanistic, environmental, and economic considerations, thereby framing the design guidelines presented below.

Alternative low-impact stabilizations such as geopolymers [49], fly ash [50], and GGBFS [51] have shown promise in reducing embodied carbon. Compared to these, our low-dosage cement approach achieves similar environmental benefits with more established supply chains, ensuring near-term applicability.

5. Field Trial

The preceding laboratory investigations established the suitability of fully weathered soft rock and its cement-stabilized counterpart as subgrade fill materials [52]. However, to validate their performance under real-world conditions [53], field compaction trials were conducted. This section presents the results of on-site tests involving both untreated and cement-stabilized fills, assessing the effects of loose lift thickness, moisture content, and compaction method on the quality of subgrade construction. The trials were carried out along the test section between DK1806 + 729.06 and DK1807 + 325, using standardized construction protocols and advanced instrumentation to derive practical guidance for railway earthworks.

The test section was divided into two zones: one filled with untreated, fully weathered soft rock obtained from the local borrow area, and the other with cement-stabilized soil prepared in a mixing plant with a cement dosage of 3.5%. As illustrated in Figure 13, the construction process adhered to the standardized “three stages, four sections, eight procedures” framework. A 20-ton LSD220H vibratory roller and ST-140B bulldozer were employed. Fill layers were compacted in stages—static pressure, low-amplitude vibration, and high-amplitude vibration—following the principle of “static before vibration, slow before fast.” After compaction, critical parameters were measured, including loose lift thickness (via total station), moisture content (by sampling and oven-drying), degree of compaction using the sand cone method, and subgrade reaction modulus (K₃₀) via static plate load testing (per TB 10102-2010 after instrument calibration). These data informed the development of optimized field construction strategies.

5.1. Optimized Compaction for Soft Rock Fill

Field trials on untreated, fully weathered soft rock were conducted with target loose lift thicknesses of 0.6 m, 0.5 m, 0.4 m, and 0.3 m. The objective was to determine the optimal combination of moisture content, loose lift thickness, and compaction procedure to meet the Menghua heavy-haul railway design criteria: degree of compaction ≥ 0.92 and K₃₀ ≥ 80 MPa/m (below the formation level).

The results of the on-site compaction trials are summarized in

Table 8. Results of on-site compaction process test for fully weathered soft rock fill.

Fill Range	Layer	Target Loose Lift (m)	Compaction Process	Loosening Coeff.	Moisture (%)	Degree of Compaction (K)	K30 (MPa/m)
DK1806 + 858–910	1st	0.6	S2-L2-H2-H2	1.07	21.1	0.93	63.2
DK1806 + 858–910	3rd	0.6	S2-L2-H2-H1-S1	1.05	15.5	0.97	74.8
DK1807 + 096–200	1st	0.5	S2-L2-H2-H1-S1	1.08	18.4	0.96	61.5
DK1807 + 096–200	3rd	0.5	S2-L1-S1-H1-S1	1.09	16.7	0.96	58.0
DK1806 + 910–935	1st	0.4	S2-L2-H2	1.11	22.8	0.94	65.8
DK1806 + 910–935	3rd	0.4	S2-L1-S1-H1-S1	1.10	18.0	0.94	45.3
DK1807 + 200–294	1st	0.4	S2-L1-S1-H1-S1-H1-S1	1.12	15.4	0.94	89.9
DK1807 + 200–294	2nd	0.3	S2-L1-S1-H1-S1-L1-S1	1.14	20.9	0.95	55.7
DK1807 + 200–294	2nd	0.3	S2-L1-S1-H1-S1-H1-S1	1.18	16.4	0.93	75.2
DK1807 + 200–294	3rd	0.3	S2-L1-S1-H1-S1-S2	1.18	15.0	0.96	88.9

Note: Moisture content represents the average of all measurements per layer; degree of compaction and subgrade reaction modulus (K₃₀) are maximum values recorded during compaction.

, where compaction processes are denoted as S (static pressure), L (low-amplitude vibration), and H (high-amplitude vibration), with numbers indicating the number of passes (e.g., S2 = two passes of static pressure).

Analysis shows that a 0.4 m loose lift, compacted with a moisture content of 15.5% (approximately 2.0% below the laboratory optimum), and using the sequence S2-L1-S1-H1-S1, achieved a compaction degree of 0.94 and a K₃₀ of 89.9 MPa/m—exceeding both design thresholds. Similarly, a 0.3 m loose lift compacted at 15.0% moisture using S2-L1-S1-H1-S1-S2 yielded a compaction degree of 0.96 and K₃₀ of 88.9 MPa/m.

To balance construction efficiency and cost, the relationship between loose lift thickness, moisture content, and compaction outcomes was further explored. Analysis showed that field-optimal moisture content peaked approximately 18.4%, corresponding to the highest compaction degree (95%). However, as moisture increased, K₃₀ declined, as illustrated in Figure 14. Specifically, Figure 14a shows that K₃₀ decreases as loose lift thickness increases, while Figure 14b highlights its reduction with rising moisture content, necessitating strict control for larger loose lift thicknesses. This trend arises because thicker loose lifts reduce the penetration of compaction energy. The roller compacts mainly the upper portion, while deeper zones remain under-compacted, resulting in lower stiffness and a reduced K₃₀. At 18.4% moisture content, a loose lift thickness of 0.27 m satisfies the 80 MPa/m subgrade reaction modulus threshold, though this is impractical for field efficiency. When moisture content is controlled below the optimum (e.g., 15.4%), a loose lift thickness of 0.38 m becomes feasible, offering a practical compromise. Thus, a moisture range of 15.4–18.4% is recommended, approximately 1–2% higher than the laboratory optimum due to field variability.

At constant moisture content, increasing lift thickness had minimal effect on the degree of compaction but noticeably reduced K₃₀. When moisture exceeded the optimum, quality targets were not met at any thickness. With moisture maintained within 15.4–18.4%, a 0.38 m lift met specifications, but considering operational efficiency and climate variability, 0.35 m is recommended. Under these conditions, the S2-L1-S1-H1-S1-S2 sequence ensures stable subgrade quality during mass construction (Figure 15).

5.2. Optimized Compaction for Cement-Stabilized Fill

Field tests on cement-stabilized fill (3.5% cement) were conducted between DK1806 + 729.06 and DK1806 + 858 using loose lift thicknesses of 0.6 m, 0.5 m, 0.4 m, and 0.3 m. A standardized compaction sequence—S2 (measurement)-L1-S1 (measurement)-H1-S1 (measurement)—was applied, aiming for a degree of compaction ≥ 90% and K₃₀ ≥ 80 MPa/m.

Table 9. Compaction process test results for cement-stabilized soil fill.

Layer	Target Loose Lift (m)	Loose Thickness (m)	Compacted Thickness (m)	Loosening Coeff.	Moisture (%)	Degree of Compaction (K)	K30 (MPa/m)
1	0.6	0.59	0.57	1.04	20.8	0.91	72.4
2	0.3	0.34	0.31	1.10	26.7	0.91	66.0
3	0.5	0.49	0.47	1.04	27.4	0.87	63.4
4	0.5	0.50	0.47	1.06	16.3	0.97	87.1
5	0.4	0.40	0.37	1.08	18.5	0.95	99.6
6	0.4	0.41	0.38	1.08	19.9	0.92	84.0
7	0.6	0.59	0.56	1.05	16.1	0.93	75.6
8	0.3	0.32	0.29	1.10	15.9	0.94	114.0

presents the results, where the moisture content is given as the average per layer, and the degree of compaction and K₃₀ are reported as the maximum observed values.

Results demonstrate that a 0.3 m lift at 15.9% moisture produced K₃₀ = 114.0 MPa/m, the highest value recorded. At 0.4 m thickness, with moisture levels slightly above the 17.1% optimum (18.5–19.9%), compaction degrees of 92–95% and K₃₀ values of 84.0–99.6 MPa/m were achieved—meeting design requirements. For a 0.5 m lift, excessive moisture (27.4%) failed to meet targets (compaction 87%, K₃₀ = 63.4 MPa/m), whereas lowering moisture to 16.3% restored compliance (compaction 97%, K₃₀ = 87.1 MPa/m). These findings underscore the necessity of precise moisture control, especially for thicker lifts.

Field data suggest an optimal moisture content of ~19.1%, at which the compaction degree peaks. As shown in Figure 16a, K₃₀ decreases linearly with increasing loose lift thickness at this moisture level. Similarly, Figure 16b confirms K₃₀’s inverse relationship with moisture content. For the 80 MPa/m threshold, the maximum allowable lift thicknesses were 0.54 m (at optimum moisture) and 0.39 m (below optimum).

The compaction sequence S2-L1-S1-H1-S1 consistently achieved both a degree of compaction ≥ 90% and K₃₀ ≥ 80 MPa/m (Figure 17), validating the effectiveness of staged compaction from static to high-amplitude vibration. When moisture exceeded optimum, target values were not achieved. Within 16.3–19.9% moisture, a 0.54 m lift is feasible, but a 0.4 m lift is recommended for its resilience to weather-induced variations.

Compared with untreated fill, cement-stabilized soil exhibited superior quality consistency due to centralized batching, reduced moisture sensitivity, and enhanced compaction performance. These advantages make it highly suitable for large-scale subgrade applications under variable field conditions.

6. Conclusions

This study presents a comprehensive laboratory and field evaluation of fully weathered phyllite as a subgrade fill material for railway applications. It examines both untreated and lightly cement-stabilized conditions, aiming to assess performance, optimize compaction strategies, and align with sustainability goals in large-scale earthworks. The findings support practical implementation and performance-based design of phyllite-based subgrades under modern railway engineering standards. Key findings can be drawn:

• The tested phyllite is a quartz-rich (≈70%) silty clay with a high liquid limit (≈46%) and minimal expansive clay content (≤3%). It is classified as a Group D fill per Chinese railway standards. Swelling, shrinkage, and XRD tests confirm that it is non-expansive and volumetrically stable under expected field conditions.
• Untreated phyllite compacted to ≥92% Proctor density exhibits moderate strength (c ≈ 43 kPa, φ ≈ 37°), a soaked CBR of 6–8%, and moderate compressibility. These values meet the threshold for sub-subgrade layers but fall short for main subgrade applications, particularly under saturated or cyclic loading.
• Incorporating 3.5% cement increases the 7-day UCS to >250 kPa and soaked CBR to ≈60%, while reducing the compression index by ≈70% and maintaining hydraulic conductivity in the range of 1–3 × 10⁻⁶ cm/s. Over 70% of long-term strength develops within 7 days, enabling rapid construction cycles.
• Field trials confirmed that untreated phyllite can achieve design targets (K₃₀ ≥ 80 MPa/m, degree of compaction ≥ 0.92) when compacted in lifts ≤ 0.38 m with moisture content 1–2% below optimum. Cement-stabilized fill demonstrated improved constructability and quality consistency, achieving K₃₀ up to 114 MPa/m with thicker lifts (up to 0.54 m) and less sensitivity to moisture variation, provided a staged compaction sequence (S2-L1-S1-H1-S1) was applied.
• Life-cycle analysis showed that using stabilized in-situ phyllite reduces carbon emissions and cost by approximately 30% compared to imported crushed rock hauled 30 km. The approach also increases the material circularity index from 0.12 (import) to 0.74 (in-situ reuse), contributing to sustainable infrastructure development.

Design and construction recommendations are: (1) Untreated phyllite: Suitable for sub-subgrade layers with stringent compaction (>95% Proctor), moisture control, and proper drainage; (2) Stabilized phyllite: Suitable for main subgrade layers with ≥90% Proctor and 7-day curing; (3) Moisture management: Critical for both cases; recommended range is 15.4–18.4%; (4) Field lift thickness: Limit to 0.35–0.38 m for untreated and ≤0.54 m for stabilized fill; (5) Drainage: Provide longitudinal drains for fills > 3 m due to persistent permeability. The results are applicable not only to railway subgrades but also to road earthworks under similar material and compaction conditions.