DIC-Based Crack Mode Identification and Constitutive Modeling of Magnesium-Based Wood-like Materials Under Uniaxial Compression

Abstract

This study investigates the uniaxial compression failure of magnesium-based wood-like material (MWM) prisms (100 × 100 × 300 mm³) using digital image correlation (DIC). The results revealed an average compressive strength of 8.76 MPa and a dominant failure mode with Y-shaped or inclined penetrating cracks. A novel piecewise constitutive model was established, combining a quartic polynomial and a rational fraction, demonstrating high fitting accuracy. Critically, the proportional limit was identified to be very low (20–35% of peak stress), attributed to early-stage damage from fiber–matrix interfacial defects. DIC analysis quantitatively distinguished dual crack initiation modes, pure mode I (occurring at ≈100% peak load) and mixed mode I/II (initiating earlier at 90.02% peak load), demonstrating that tensile shear coupling accelerates failure. These findings provide critical mechanistic insights and a reliable model for optimizing MWM in sustainable construction. Future work will explore the material’s behavior under multiaxial loading.

1. Introduction

Amid global carbon neutrality initiatives, the substantial carbon emissions and resource intensity of conventional construction materials have reached critical levels [1,2,3], compounded by mounting pressures from agricultural waste disposal [4,5,6]. Developing waste-valorized low-carbon alternatives is therefore imperative for sustainable construction. Magnesium-based wood-like material (MWM), an innovative biocomposite integrating magnesium oxychloride cement (MOC) with plant fibers, presents a promising solution. MOC offers rapid hardening, low density, and fire resistance, with 40–50% lower CO₂ emissions during production compared to ordinary Portland cement, alongside carbon sequestration capabilities [7,8,9,10,11]. The incorporation of plant fibers (e.g., straw, sawdust) imparts wood-mimetic structures and lightweight characteristics (density: 1135 kg/m³) to MWM, while fiber bridging enhances its crack resistance (unstable fracture toughness: 0.81 MPa·m^1/2; fracture energy: 709.25 N·m⁻¹), enabling agricultural waste upcycling [12,13]. These attributes position MWM as an ideal material for conserving timber-structured cultural heritage.

Significant research efforts have been devoted to optimizing the composition and enhancing the performance of MWM. For instance, Li et al. [14] systematically investigated the influence of mix proportions and modifiers on the mechanical properties and water resistance of MWM, establishing a foundation for rational mix design. Yang et al. [15] further elucidated the modification mechanisms of specific additives, revealing their role in improving microstructure and durability. Regarding environmental performance, Wu et al. [16] evaluated the effectiveness of various modifiers under dry–wet cycling conditions, providing insights into long-term durability. From a processing perspective, He et al. [17] demonstrated that optimizing the pressing pressure significantly enhances the mechanical properties and water resistance of straw/sawdust–MWM composites, highlighting the importance of manufacturing parameters. Furthermore, structural applications have been explored, as Feng et al. [18] investigated the compressive behavior of CFRP-strengthened MWM columns under eccentric loading, contributing to the understanding of MWM’s performance in reinforced configurations.

Regarding characterization of the fracture mechanics of MWM, research has primarily focused on splitting tensile tests and three-point bending tests. Under splitting tension, Li et al. [13] conducted preliminary investigations on the fracture behavior of MWM, identifying secondary crack patterns in cubic specimens and quantifying damage evolution throughout the failure process [19]. In a mode I fracture assessment via three-point bending, Li et al. [12] determined key fracture parameters such as initiation toughness and unstable fracture toughness and established K_R-resistance curves to characterize the material’s crack growth behavior. Additional research has analyzed the evolution and diffusion mechanisms of the fracture process zone in MWM under bending conditions [20]. However, despite these advances, a systematic investigation into the crack evolution mechanisms, full-process failure behavior, and fracture mode classification of standard MWM prisms under uniaxial compression remains lacking, representing a significant gap in the current understanding of MWM’s mechanical behavior.

To address these research gaps, this study employs a full-field optical monitoring approach to investigate the complete failure process of standard MWM prisms under uniaxial compression. Given the rapid fracture propagation and limitations of direct visual observation in capturing subtle crack initiation, advanced measurement techniques are essential. Digital image correlation (DIC) has been identified as an ideal solution owing to its non-contact operation, full-field measurement capability, high environmental robustness, and cost-effectiveness [21,22,23]. This technique has demonstrated exceptional efficacy in deformation and fracture analyses across various engineering materials [24,25,26,27,28], particularly in characterizing complex crack propagation patterns and quantifying strain localization phenomena. The aforementioned studies on the fracture properties of MWM also utilized DIC for auxiliary observation and data collection.

Building upon this methodological foundation, the present study provides the first comprehensive analysis of crack evolution and failure mechanisms in MWM under uniaxial compression using DIC technology. The technical process is shown in Figure 1. Our work systematically examines the load–displacement response, establishes a novel piecewise constitutive model for uniaxial compression and, for the first time, identifies and distinguishes crack initiation modes at critical locations. The findings from this investigation provide critical insights for optimizing MWM design and promoting its engineering applications in sustainable construction, particularly for conservation of timber-structured heritage, where understanding failure mechanisms is essential for structural integrity assessment.

2. Materials and Experiments

2.1. Material Preparation and Specimen Fabrication

The MWM synthesis process is illustrated in Figure 2. The base formulation adopted a molar ratio of MgO:MgCl₂:H₂O = 11:1:19 [12,13,15,17]. Light-burned MgO powder (Liaoning source, 91.09% purity, 59.98% active content) was progressively added to industrial magnesium chloride solution (Qinghai source, 46% purity, 23°Bé). Mechanical blending using a mixer achieved homogeneous MOC paste.

Plant fillers (supplied by Guiyang Huaxi Yuanhai Wood Factory, Guiyang, Guizhou Province, China) included the following: straw: 10.5% moisture content, 0.24 g/cm³ density; sawdust: 13.2% moisture content, 0.30 g/cm³ density.

Pretreatment sequence: Impurity removal (<1% mud content, no fungal contamination) → rotary pulverization → 8 mm sieving.

The formulation incorporated 20 wt% plant fibers, 0.25 wt% Composite Additive A (1:1 sulfate–phosphate system, 5% concentration), and 0.25 wt% analytical-grade tartaric acid as hydration regulator. This ternary additive system enhanced hydrothermal stability while moderating exothermic reactions.

Molding procedure:

1. Steel mold dimensions: 1500 × 1000 × 1000 mm³;

2. 3.3 MPa compaction pressure maintained for 24 h via hydraulic press;

3. 28-day ambient curing post-demolding (20 ± 3 °C, 70 ± 5% RH).

The resultant MWM exhibited a bulk density of 1135.52 kg/m³. Figure 3 shows a uniform wood-like macrostructure without defects. Uniaxial compression tests employed 100 × 100 × 300 mm³ prisms (labeled AC-1 to AC-9), with supplementary specimens (e.g., 5 specimens for modulus of elasticity, E1~E5) being prepared as per GB/T 50081-2019 [29]. The specimens used in the experiment were cut by a high-precision diamond sawing machine, and the cut surfaces were lightly ground and polished using a concrete grinding machine. Their fundamental properties are summarized in

Table 1. Characterization of MWM.

Light-Burned MgO (wt%)		Mechanical Properties
Parameter	Value	Parameter	Value
MgO	91.09%	Tensile strength fts (MPa)	1.79
SiO2	5.77%	Elastic modulus E (GPa)	2.21
CaO	2.00%	Poisson’s ratio ν	0.21
Al2O3	0.58%	Crack initiation toughness KIcini (MPa·m1/2)	0.29
Fe2O3	0.29%	Unstable fracture toughness KIcun (MPa·m1/2)	0.81
Others	<0.27%	Fracture energy Gf (N·m−1)	709.25

.

Microstructural characterization was performed via scanning electron microscopy (SEM), as shown in Figure 4. Low-magnification imaging reveals discontinuous interfacial contacts between the MOC matrix and plant fibers within the MWM, along with irregular slit-like pores. These pores likely originated from the combined action of phase-transformation stresses and moisture transport during curing. Higher-magnification observations show the interfacial transition zone features: fiber surfaces exhibit micrometer-scale topological undulations, with MOC hydration products (Mg(OH)₂·MgCl₂·8H₂O) that are anchored within surface asperities, evidencing matrix infiltration during the rheological stage [19].

2.2. Uniaxial Compression Test

2.2.1. DIC System

DIC was employed as the primary full-field deformation monitoring technique. This method computes displacement fields by correlating intensity distributions of sub-regions, as the reference, and deformed specimen images [30]. As depicted in Figure 5, a subset, $ABCD$, that is centered at $P\left(x_0,y_0\right)$ deforms to $A\prime B\prime C\prime D\prime$ with a new centroid $P\prime \left(x_0\prime ,y_0\prime \right)$. The displacement of any point $Q\left(x_i,y_i\right)$ to $Q\prime \left(x_i\prime ,y_i\prime \right)$ is defined by a first-order shape function:

$$\left\{\begin{array}{c}{x}_i^{\prime }=x_i+u+{\displaystyle \frac{\partial u}{\partial x}}\mathsf{\Delta }x+{\displaystyle \frac{\partial u}{\partial y}}\mathsf{\Delta }y\\ y_i^{\prime }=y_i+v+{\displaystyle \frac{\partial v}{\partial x}}\mathsf{\Delta }x+{\displaystyle \frac{\partial v}{\partial y}}\mathsf{\Delta }y\end{array}\right.$$

(1)

where $\mathsf{∆}x=x_0-x_i$, $\mathsf{∆}y=y_0-y_i$. $u$ and $v$ are horizontal/vertical displacements of $P$. And ${\displaystyle \frac{\partial u}{\partial x}}$, ${\displaystyle \frac{\partial u}{\partial y}}$, ${\displaystyle \frac{\partial v}{\partial x}}$, and ${\displaystyle \frac{\partial v}{\partial y}}$ are the gradients of displacement in the corresponding directions.

The target area was abrasively polished and degreased with ethanol to create an optically neutral substrate. High-contrast stochastic speckle patterns were then generated via pneumatic spray deposition of a white base coat, followed by black speckles [31,32].

The DIC setup (Figure 6) incorporated twin Prosilica GE4900 cameras (4872 × 3248 pixels) with LED illumination. A 300 × 100 mm² region of interest (ROI) was defined at the specimen surface (Figure 7). Due to size constraints, the resolution of the ROI is approximately 3240 pixels in the vertical direction and about 1080 pixels in the horizontal direction. Images were acquired at 3 Hz and processed in VIC-3D-9 software using 29 × 29-pixel subsets with a 7-pixel step for displacement gradient calculation. The global axial compressive strains were measured via a virtual 150 mm extensometer centered on the specimen (Figure 7).

2.2.2. Loading Configuration

The testing system comprised an MTS electromechanical universal testing machine integrated with the aforementioned DIC system. Specimen surfaces were prepared as described in Section 2.2.1, and loading platens were cleaned to minimize frictional effects prior to testing. The experimental layout is shown in Figure 8.

Quasi-static compression was applied under force control at a constant rate of 3000 N/s. The axial compressive strength was calculated as follows:

$$f=0.95{\displaystyle \frac{F}{A}}$$

(2)

where $F$ is the failure load; $A$ is the bearing area; and 0.95 is the size effect correction factor specified in GB/T 50081-2019 [29] for 100 mm × 100 mm × 300 mm prisms.

The two data streams (load from MTS and strain from DIC) were synchronized using a common timeline. This hybrid approach leverages the high accuracy of the calibrated MTS load cell for force measurement and the superior capability of DIC for full-field, non-contact strain measurement.

3. Results and Discussion

3.1. Experimental Phenomena and Results

During the initial loading stage, no significant changes were observed on the specimen surface. As the load continued to increase and approached the peak load, multiple discontinuous cracks, visible to the naked eye, appeared on the specimen surface. With further load increases beyond the peak load, the cracks rapidly propagated and interconnected, forming distinct inclined cracks. Subsequently, these inclined cracks continued to propagate or connect with others, ultimately penetrating through the entire specimen. An audible tearing sound accompanied this penetration failure process. Following this, axial displacement continued to increase, while the load exhibited a declining trend, culminating in the complete failure of the specimen, which fragmented into multiple pieces. Figure 9 illustrates the typical failure morphology of the MWM prism after the axial compression test. It was observed that almost no spalling occurred on the specimen surface. The main fracture pattern consisted of “Y”-shaped cracks or inclined cracks penetrating through the specimen. Failure resulted in the specimen being split into several columns, which could be completely separated.

The compressive strength test results are presented in

Table 2. Compressive strength.

Specimen	Contact Area/[mm2]	Peak Load /[kN]	Compressive Strength/[MPa]	Average Strength/[MPa]	Standard Deviation/[MPa]
AC-1	101.28 × 103.50	102.48	9.29	8.76	0.33
AC-2	101.75 × 101.04	98.08	9.06
AC-3	101.46 × 100.40	91.04	8.49
AC-4	101.46 × 103.06	96.75	8.79
AC-5	101.01 × 101.67	92.98	8.60
AC-6	100.85 × 102.03	88.58	8.18
AC-7	101.54 × 101.44	97.62	9.00
AC-8	104.15 × 99.47	97.76	8.96
AC-9	105.06 × 98.55	92.56	8.49

. The measured average axial compressive strength of the specimens was 8.76 MPa, with a standard deviation of 0.33 MPa. The results of the modulus of elasticity test are shown in

Table 3. Modulus of elasticity.

Specimen	E/[GPa]	Average (E)/[GPa]	Standard Deviation (E)/[GPa]
E1	1.846	2.210	0.279
E2	2.605
E3	2.317
E4	2.342
E5	1.943

, and the average value is 2.21 GPa.

3.2. Stress–Strain Response

To accurately characterize the uniaxial compression behavior of the MWM, this section systematically investigates its stress–strain response through experimental tests. The constitutive model established herein will serve as the theoretical foundation for subsequent numerical simulations and engineering applications.

The stress–strain curves were derived from load–time data that were automatically recorded by the MTS testing machine’s acquisition system and strain–time data obtained using the digital image correlation (DIC) technique. The stress data was calculated in real time from the load cell of the MTS testing machine. The strain data was obtained from DIC analysis. As described in Section 2.2.1 and illustrated in Figure 7, a virtual extensometer with a gauge length of 150 mm was defined along the central axis of the specimen’s surface. The DIC software (VIC-3D) calculated the global axial engineering strain.

Valid data from nine specimens were collected during axial compression tests, with their complete stress–strain curves being presented in Figure 10.

As observed in Figure 10, all specimens exhibit similar geometric characteristics in their stress–strain curves. Using the peak stress σ₀ as a reference, the curve evolution can be divided into four distinct phases:

1. Initial Phase (σ ≤ 0.3σ₀): The curve displays high initial stiffness that rapidly decreases under loading, accompanied by a brief elastic deformation segment.

2. Stable Development Phase (0.3σ₀ < σ ≤ 0.8σ₀): The slope transitions from steep to gradual, with a reduced rate of change.

3. Softening Transition Phase (0.8σ₀ < σ ≤ σ₀): The slope decreases, forming a quasi-plateau until reaching zero slope at σ₀.

4. Failure Phase (after σ₀): The stress precipitously drops to 0.8σ₀ ~ 0.9σ₀, followed by abrupt curve descent.

To quantify the curve morphology’s evolution, the Euler forward difference method was employed to compute the slope variation rate [33]:

$${\displaystyle \frac{y\left(x_{i+1}\right)-y\left(x_i\right)}{h}}\approx y^{\prime }\left(x_i\right)$$

(3)

The resulting first derivative dσ/dε is shown in Figure 11. The results indicate a globally monotonic decreasing slope with accelerated decay at both ends, aligning with stress–strain curve characteristics. Using Equation (3) again, the second derivative d²σ/dε² can be obtained. The result shows that in the initial loading stage, the second derivative fluctuates around 0, which reveals a narrow elastic deformation range—the proportional limit is merely 20~35% of the peak stress, significantly lower than concrete’s 40~50% [34].

Although both MWM and concrete materials are biphasic composites (matrix-dispersed phase), their microstructures fundamentally differ: Concrete forms a dense structure through frictional forces and mechanical interlocking between the mortar matrix and aggregate dispersion [35]. MWM exhibits a flocculent network (Figure 4), where plant fiber bridging generates micro-pores that are incompletely filled by chemical crystals, increasing initial defects [36]. Poor compaction of fiber dispersion weakens structural continuity, exacerbating microcrack propagation during forming and ultimately reducing its proportional limit.

Building upon the stress–strain response and microstructural analysis, a constitutive model that accurately describes MWM’s mechanical behavior is required. Based on experimental stress–strain response data from uniaxial compression tests, this study establishes a constitutive model for MWM. Existing experimentally derived constitutive models primarily fall into two categories [37]: full-range continuous functions (e.g., Carreira D.J. model [38], Wee T.H. model [39]) and piecewise functions (e.g., Guo Z.H. model [40]). Referring to these frameworks and considering geometric curve characteristics, the MWM constitutive model is formulated as a piecewise function with a polynomial ascending branch and a rational fraction descending branch.

Using measured stress–strain curves, dimensionless parameters are defined:

$$y=\sigma /{\sigma }_0$$

(4)

$$x=\epsilon /{\epsilon }_0$$

(5)

where $\sigma$ denotes stress, ${\sigma }_0$ is peak stress, $\epsilon$ is strain, and ${\epsilon }_0$ is peak strain.

Through trial fitting based on curve geometry, the complete stress–strain curve is divided into the following:

Ascending branch: Fitted with a quartic polynomial;

Descending branch: Fitted with a rational fraction.

The constitutive equation is given by Equation (6):

$$y=\left\{\begin{array}{cc}0.12+2.99x-4.58x^2+3.80x^3-1.33x^4& 0⩽x<1\\ {\displaystyle \frac{x}{0.96(x-1{)}^2+x}}& 1⩽x\end{array}\right.$$

(6)

Figure 12 compares the fitted curves with experimental data. The ascending branch exhibits excellent agreement (R² = 0.94776) without significant deviation. For the descending branch, limited data points and high dispersion (nonlinear analysis being sensitive to data density) result in lower fitting accuracy. Using pre-0.9σ₀ data yields R² = 0.91857, sufficiently capturing the degradation trend.

Geometric features of the constitutive relationship (Figure 13) include the following:

(1) Initial point: x = 0, y = 0.12.

(2) Peak point: x = 1, y = 1.

(3) Zero slope at peak: x = 1, dy /dx = 0.

(4) Convexity: 0 ≤ x < 1, d²y/dx² < 0, monotonically decreasing slope.

(5) Boundedness: x ≥ 0, 0 ≤ y ≤ 1.

3.3. Crack Mode Identification and Evolution

Understanding crack patterns under uniaxial compression is crucial for elucidating the failure mechanisms of MWM and optimizing their engineering applications. Different crack modes significantly impact macroscopic mechanical behaviors, such as crack propagation velocity and fracture process zone evolution [41]. Although the global loading condition is compressive, the failure of quasi-brittle materials like MWM initiates locally through tensile or shear cracking due to stress concentrations caused by material heterogeneity. Consequently, deciphering crack formation mechanisms is essential for understanding the failure processes of multidimensional complex materials. Based on loading conditions, cracks typically manifest in three fundamental modes: mode I (tensile), mode II (shear), and mode III (tearing) [42]. All cracks can be formed by the combination of these three basic types of cracks. DIC technology primarily focuses on mode I and mode II cracks (Figure 14) in practice because their in-plane displacements (opening and sliding) are directly measurable from surface strain fields.

Traditional identification of crack modes in brittle materials relies on subjective fracture morphology assessment: clean or feather-patterned surfaces indicate mode I tensile cracks, while rough or powder-covered surfaces suggest mode II shear cracks [43,44,45]. Modern approaches quantify relative displacement across crack faces [23,46] or employ strain-field-based criteria [13]. The Strain Judgment Method (SJM) proposed by Li et al. [13] utilizes principal tensile strain ($e_1$) and maximum shear strain ($e_{\tau }$) at crack initiation zones, with the following criteria:

When the maximum tensile strain exceeds the critical threshold, it is considered that mode I crack propagation occurs. At this time, $e_1$ is greater than 0, and $e_{\tau }$ is approximately equal to 0 (satisfying Equation (7)). Alternatively, when the maximum shear strain exceeds the critical threshold, it is considered that mode II crack propagation occurs. At this time, $e_{\tau }$ is greater than 0, and $e_1$ is approximately equal to 0 (satisfying Equation (8)).

$$e_1>0\cup e_{\tau }\approx 0$$

(7)

$$e_1\approx 0\cup e_{\tau }>0$$

(8)

where $e_1$ is the principal tensile strain; $e_{\tau }$ is the maximum shear strain.

This study employed DIC technology to monitor surface strain throughout testing and applied SJM to analyze MWM crack modes. The core of our methodology is to objectively identify these modes from the local strain fields at the initiation of cracks, thereby bridging the theoretical framework of fracture mechanics with the experimental observation under compressive loading.

During uniaxial compression, multiple cracks initiated on prismatic specimens and coalesced into Y-shaped or inclined cracks. We focused on representative initial crack locations, identifying two modes: mode I cracks and mixed mode I/II cracks (exemplified by AC-1 and AC-2):

The Strain Judgment Method (SJM) was employed to identify initial crack modes during uniaxial compression. Regarding AC-1, as shown in Figure 15, both the principal tensile strain $e_1$ and maximum shear strain $e_{\tau }$ exhibited negligible changes during initial axial loading. At approximately 15 s, $e_1$ commenced sustained growth, while $e_{\tau }$ remained stable. By 37 s (Point A), significant strain localization emerged at the initial cracking location (Figure 16b), indicating imminent surface cracking. Measured values reached $e_1$ = 0.0292 and $e_{\tau }$ = 0.0022. The markedly steeper slope of the $e_1$-time curve compared to $e_{\tau }$ confirmed crack dominance by opening deformation, thus identifying a mode I crack.

Figure 16a shows the tensile strain concentration at AC-1’s initial crack location during early loading, and the localized area was limited ($e_1$ ≈ 0.0031). Notably, a light-green enveloped secondary concentration zone ($e_1$ = 0.0023~0.0025) surrounded the core region. With increasing load, the strain-concentrated area expanded along the secondary zone. Immediately prior to cracking, the maximum $e_1$ peaked at 0.0292. Post-fracture energy release caused continued strain increases but dissipated localization.

The strain–time relationship for AC-2 (Figure 17) differed fundamentally: both $e_1$ and $e_{\tau }$ increased synchronously soon after loading commenced. At 30 s (Point B), when surface cracking occurred, $e_1$ = 0.0106 and $e_{\tau }$ = 0.0079. Comparable magnitudes and slopes throughout loading indicated simultaneous normal opening and tangential sliding deformations, confirming a mixed mode I/II crack.

Figure 18a demonstrates the strains concentration at AC-2’s crack origin. The concentrated areas of the maximum shear strain and the principal tensile strain are basically overlapping, which indicates that the combined action of the two factors led to the failure of the specimen. The pre-crack maximum $e_1$ (0.0106) constituted merely 36.3% of AC-1’s value (Figure 18b), and the $e_{\tau }$ of AC-2 (0.0079) was 3.6 times that of AC-1 (0.0022), verifying shear stress involvement alongside tensile stress in crack initiation.

The crack mode identification results for all the specimens are presented in

Table 4. Results of crack identification.

Specimen	Crack Mode	e1 Before Crack Initiation	eτ Before Crack Initiation	Macro-Crack Initiation Load
AC-1	mode I	0.0292	-	≈ Pmax
AC-2	mode I/II	0.0106	0.0079	89.25% Pmax
AC-3	mode I	0.0310	-	≈Pmax
AC-4	mode I	0.0286	-	≈Pmax
AC-5	mode I	0.0298	-	≈Pmax
AC-6	mode I/II	0.0112	0.0074	92.13% Pmax
AC-7	mode I/II	0.0095	0.0085	87.58% Pmax
AC-8	mode I/II	0.0109	0.0072	91.14% Pmax
AC-9	mode I	0.0306	-	≈Pmax

. Analysis of the table reveals that failure in five of the nine tested specimens was governed by mode I cracking and four by mixed mode I/II cracking. A significant difference in the pre-cracking strain magnitude was observed: the average principal tensile strain ($e_1$) for mode I cracks was 0.02984, which is substantially higher than the value of 0.01055 recorded for mixed-mode cracks, with the latter constituting only 35.36% of the former. Critically, the macroscopic crack initiation load nearly coincided with the peak load (100%) for mode I specimens but occurred at a significantly lower load level (90.02% of the peak) for mixed-mode specimens. This behavior is attributed to the distinct failure mechanisms: pure mode I cracking is driven by localized tensile stress concentration, whereas the superposition of tensile and shear stresses in mixed-mode cracking facilitates earlier damage accumulation and thus accelerates failure.

4. Conclusions

Using the DIC technique, the failure of MWM under uniaxial compression was analyzed, and the following conclusions were drawn:

(1) The failure of MWM prisms under uniaxial compression is characterized by Y-shaped or inclined penetrating cracks, resulting in an average axial compressive strength of 8.76 MPa and an elastic modulus of 2.21 GPa, with no surface spalling being observed.

(2) A piecewise constitutive model was established with high fidelity, where the ascending branch is described by a quartic polynomial (R² = 0.94776) and the descending branch by a rational fraction (R² = 0.91857), providing a reliable theoretical basis for numerical simulations of MWM structures.

(3) The proportional limit of MWM was quantitatively identified to be very low, at only 20–35% of the peak stress, which is substantially lower than that of conventional concrete (40–50%). This is attributed to the weakened structural continuity induced by fiber–matrix interface defects.

(4) Two distinct crack initiation modes of MWM under uniaxial compression were quantitatively identified for the first time: pure mode I and mixed mode I/II. Statistical results from nine specimens showed that five failed via mode I and four via mixed mode.

(5) A critical difference in failure initiation was observed: pure mode I cracks initiated at a load very close to the peak load (≈100% P_max), whereas mixed-mode cracks initiated significantly earlier, at an average of 90.02% P_max. This demonstrates that the coupling of tensile and shear mechanisms accelerates the damage process and leads to premature failure.