Experiment and numerical analysis of delamination mechanisms of CRFP composites under seawater hygrothermal environments

Abstract

This study investigates the delamination failure mechanisms of T700/8911 carbon fiber reinforced polymer (CFRP) composites in seawater environments through combined experimental and numerical approaches. Three-point bending tests were conducted on specimens with [0°₁₆//0°₁₆], [0°/90°]_4s//[0°/90°]_4s and [15°/-15°]_4s//[15°/-15°]_4s layup patterns under both ambient (25°C) and accelerated hygrothermal conditions (70°C temperature and 95% relative humidity). Experimental results revealed that moisture absorption caused a 23–37% reduction in interfacial bond strength. This reduction significantly altered failure modes from classical delamination to predominant interfacial debonding. The finite element model developed in ABAQUS incorporated humidity-dependent cohesive zone parameters and modified traction-separation laws, achieving excellent correlation with experimental data. Key findings demonstrate that hygrothermal exposure increases interlaminar fracture toughness while reducing in-plane modulus, leading to higher critical loads (28–35% increase) but lower stiffness in the linear elastic stage. The cohesive zone modeling approach successfully captured the three-phase delamination process: initial elastic deformation, critical crack initiation, and stable propagation. Analysis of mode mix ratios showed cohesive elements transition from shear-dominated to tensile-dominated damage as delamination progresses. The study establishes quantitative thresholds for interfacial stiffness reduction (15–20%) and provides a validated modeling framework for predicting performance degradation of marine composites, offering valuable insights for designing durable offshore structures in seawater environments.

Keywords

composite laminates delamination failure seawater hygrothermal effect cohesive model finite element analysis (FEA)

Introduction

Fiber-reinforced plastics (FRP) composites are increasingly being adopted in marine structures such as offshore wind turbine blades, submersibles, and ship components due to their excellent strength-to-weight ratio and corrosion resistance. However, seawater exposure induces complex degradation mechanisms in these materials, particularly affecting their interlaminar properties. The combined effects of moisture absorption and elevated temperatures fundamentally alter material behavior, leading to progressive deterioration of fiber-matrix interfaces. This environmental aging process significantly influences delamination behavior, which represents a particularly complex fracture phenomenon involving mixed-mode failure mechanisms. The hygrothermal conditions modify the fundamental failure modes, shifting the damage progression from classical delamination patterns to more interface-dominated failure scenarios. These changes in fracture characteristics demonstrate how seawater aging complicates the already intricate nature of delamination failures in composite structures, where multiple fracture modes typically interact in nonlinear ways. The environmental exposure introduces additional variables that make delamination prediction and analysis even more challenging, as traditional failure criteria may not adequately capture the altered fracture behavior under such conditions. Understanding these complex failure mechanisms becomes particularly crucial for ensuring the long-term reliability of composite marine structures operating in harsh seawater environments.

Seawater hygrothermal exposure triggers multifaceted degradation beyond pure delamination, involving synergistic matrix cracking, interfacial debonding, and fiber corrosion-all exhibiting pronounced temperature/chloride dependence. These seawater-specific mechanisms fundamentally differ from freshwater degradation pathways through: (1) chloride-catalyzed hydrolysis at fiber-matrix interfaces, (2) osmotic-pressure accelerated crack propagation, and (3) electrochemical pitting corrosion at fiber surfaces. Recent research on hygrothermal aging of fiber-reinforced polymer composites demonstrates significant advancements in understanding degradation mechanisms and developing predictive models. Gholami et al.¹ established a comprehensive finite element framework for predicting elastic properties of composites under hygrothermal conditions, developing innovative algorithms that successfully predicted degraded properties with minimal error compared to experimental data, though their model requires further validation for natural fiber composites. Chouhan et al.² introduced a novel higher-order kinematic model for analyzing delaminated composite plates under hygrothermal loads, achieving remarkable accuracy in predicting buckling loads through finite strip methodology, yet their approach needs extension to account for complex delamination shapes beyond strip-type defects. Zhong et al.³ conducted pioneering comparative studies between carbon fiber reinforced polymer and glass fiber reinforced polymer composites in hygrothermal environments, revealing glass fibers’ significant strength reduction through single-fiber testing-a critical finding for marine applications, though limited to specific temperature conditions. Kaybal et al.⁴ systematically evaluated seawater aging effects on bolted joints through combined analytical techniques, demonstrating halloysite nanotubes’ reinforcement effect and establishing a two-phase moisture absorption model, but their study overlooked chemical degradation pathways. Behera et al.⁵ quantified carbon fiber reinforced polymer quasi-isotropic laminate degradation with novel non-Fickian moisture modeling, showing substantial buckling load reduction after water immersion through rigorous thermal and microscopic analysis, though single-temperature aging limits generalizability. Shan et al.⁶ developed an advanced finite element method for open-hole laminates, quantifying characteristic length increases with width-to-diameter ratio through established failure criterion, but requires verification for non-linear hygrothermal profiles. Cheng et al.⁷ revolutionized adhesive joint analysis with modified prediction methods for hygrothermal environments, establishing interface strength degradation parameters through comprehensive testing, setting benchmarks that need extension to mixed-mode failures. Keller et al.⁸ provided seminal reviews on composite durability in seawater, linking material science with climate change mitigation while identifying critical gaps in abrasive erosion research, though lacking quantitative property retention data. Silva et al.⁹ developed an innovative extrapolation method combining established modeling principles, significantly reducing experimental workload for flexural property prediction, but requires validation for non-woven fiber composites. Amoushahi et al.¹⁰ pioneered finite strip analysis for hygrothermally aged delaminated plates, introducing advanced kinematic modeling that accurately captures buckling modes under biaxial loading with improved computational efficiency, though needing experimental verification for complex delamination patterns. Aceti et al.¹¹ delivered comprehensive accident analyses, identifying sandwich panels and adhesive joints as critical failure points through forensic investigations, providing invaluable historical perspective for safety protocols but lacking predictive degradation models. Ma et al.¹² established moisture absorption models for stiffened panels, correlating significant shear strength reduction with interfacial degradation through strain-load curve analysis, with their semi-analytical method advancing design standards but requiring extension to dynamic loading conditions. Gholami et al.¹³ developed a multiscale parallel finite element approach employing a fully coupled thermo-hygro-mechanical constitutive model. Their framework simultaneously solves Fourier heat conduction and Fickian diffusion equations through an innovative temperature-moisture concentration (FCTM) algorithm. Key findings revealed extreme property degradation under 70°C/95%RH conditions (87% transverse modulus reduction, 74% shear modulus loss), establishing new accuracy benchmarks for hygrothermal simulations. Wang et al.¹⁴ implemented a modified Hashin failure criterion incorporating hygrothermal degradation factors into their progressive damage model. The approach featured temperature- and moisture-dependent strength parameters, successfully predicting 39% tensile strength reduction in glass fiber composites at 70°C with moisture absorption. Their failure analysis identified matrix tension and interlaminar tension as dominant failure mechanisms under combined hygrothermal-mechanical loading. Shan et al.¹⁵ employed a parametric modeling framework with a novel non-dimensional temperature parameter T* to characterize environmental severity. Their study of T800/X850 carbon/epoxy laminates demonstrated significant geometry effects, showing 81% strength enhancement when width-to-hole ratio increased from 2.1 to 8.4 across various hygrothermal conditions. Chouhan et al.¹⁶ advanced micromechanical modeling through Mori-Tanaka homogenization combined with Papanicolaou’s exponential decay law. Their constitutive model accurately predicted 42% transverse modulus reduction in glass/epoxy laminates at 70°C/75%RH, with X-ray tomography validation confirming interface degradation patterns. Ly et al.¹⁷ introduced an integrated viscoelastic framework combining Laplace-transform analysis with BESOS optimization. Their constitutive model incorporated time-temperature-moisture superposition principles, achieving precise prediction of 2.82 times deflection increases under 70°C/0.8% moisture conditions while maintaining computational efficiency for engineering applications.

The expanding marine applications of FRP composites demand a thorough understanding of their seawater-specific degradation. This study addresses this need by making distinctive contributions through a systematic investigation of delamination mechanisms using integrated experimental and numerical approaches. Our methodology combines controlled hygrothermal aging tests with advanced cohesive zone modeling to capture the influence of seawater absorption on mechanical degradation of composites. The experimental work demonstrates seawater’s unique capacity to transform failure mechanisms from uniform interlaminar fracture to localized interfacial debonding, while the numerical model successfully reproduces this transition through humidity-dependent interface properties. Particularly noteworthy is our demonstration of hygrothermal conditioning’s dual-phase effect—initial toughness enhancement through matrix plasticization followed by progressive interfacial weakening, a phenomenon distinctly visualized in the fracture surface evolution. Our findings establish that marine composite durability assessment requires fundamentally different approaches than those developed for atmospheric or freshwater exposure, with significant implications for the design of marine composite structures.

Experimental research on seawater hygrothermal effect of FRP composites

Seawater immersion and moisture absorption

where E, G, and υ represent Young’s modulus, shear modulus, and Poisson’s ratio, respectively.

where 2L, b, 2h, a and a/2 denote the span, width, thickness, initial crack length, and truncation length, respectively.

T700/8911 carbon fiber/epoxy composites are used. The fiber volume fraction of fiber is 61%. Mechanical parameters are shown in Table 1, which are obtained by experiments. The geometric sizes of laminates used in the experiment are shown in Figure 1. The values of the initial delamination crack are listed in Table 2.

Table 1.

Mechanical parameters of T700/8911 composites.

$E_{11} (GPa)$	$E_{22} = E_{33} (GPa)$	$G_{12} = G_{13} (GPa)$	$ν_{12} = ν_{13}$
116	8.9	4.15	0.27

Table 2.

Geometric parameters of specimens.

Number of specimens	Specimen sizes (mm)	Initial crack length a (mm)	Truncation length a/2 (mm)	Layup patterns
①	150 × 20 × 2.4	50	25	[0°₁₆//0°₁₆]
②	150 × 20 × 2.4	50	25	[0°/90°]_4s//[0°/90°]_4s
③	150 × 20 × 2.4	50	25	[15°/-15°]_4s//[15°/-15°]_4s

To facilitate comparison, three kinds of layup designated as Layup-A: [0°₁₆//0°₁₆], Layup-B: [0°/90°]_4s//[0°/90°]_4s and Layup-C: [15°/-15°]_4s//[15°/-15°]_4s were established, where “//” symbol represents the initially delaminated interface. The experimental specimens were systematically categorized into two distinct groups to facilitate comparative analysis: an exposed group that underwent controlled hygrothermal conditioning and an unexposed control group maintained under standard laboratory conditions.

Moisture from the external environment enters the composite material primarily through capillary action at the fiber/matrix interface, diffusion through the resin matrix, as well as via cracks and voids within the composite. For each layup specimen, a two-stage approach was used to describe the moisture absorption behavior of the composite: the total weight increase M (t) of the carbon fiber/resin matrix composite due to moisture absorption is equal to the weight increase M_I (t) in the first stage (following Fick’s law) plus the weight increase M_II (t) in the second stage (following non-Fickian behavior)

\begin{array}{l} M (t) = M_{Ⅰ} (t) + {M_{Ⅰ}}_{Ⅰ} (t) = ϕ M_{m} G (t) + (1 - ϕ) M_{m} W (t) \\ = ϕ M_{m} {1 - \exp [- 7.3 {(\frac{D_{z} t}{h^{2}})}^{0.75}]} + (1 - ϕ) M_{m} [1 - {\exp - [{(α 〈 t - t_{0} 〉)}^{β}]}] \end{array}

(1)

where M_t represents the maximum moisture absorption weight, D denotes the diffusion coefficient, h represents the specimen thickness,

t_{0}

indicates the start time of the second stage, α and β are material parameters, and

ϕ

represents the moisture absorption rate.

The first-stage moisture absorption is mainly caused by defects in the composite material (such as cracks and microvoids) and water uptake by the resin. Under the combined effect of temperature and humidity, water molecules penetrate the interior of the material relatively quickly through the free volume spaces as well as defects such as voids and cracks. The moisture absorption in this stage is primarily physical adsorption, representing a reversible change. In the second stage, the moisture absorption rate decreases and the mechanism becomes more complex. In addition to the forms of moisture absorption in the first stage, it also includes moisture absorption due to crack propagation in the resin matrix, debonding at the resin/fiber interface, and resin hydrolysis. This stage results in irreversible damage to composites.

The saturated moisture absorption and aging process of T700/8911 composite specimens is divided into three steps: (1) Schreer et al.¹⁸ studied hygrothermal degradation of modes I and II fracture toughness in flat carbon/epoxy composites under 70°C temperature. Similarly, the specimens are placed in an oven at 70°C for drying treatment until their moisture desorption rate stabilizes at no more than 0.02% mass loss per day, at which point they are considered to be in an engineered dry state. The hygrothermal conditions also refer to the Chinese standard “China Aviation Industry Corporation. HB 7401-96. Test Method for Moisture Absorption of Resin - Matrix Composite Laminates in Hygrothermal Environments.” This condition (70°C, 95% RH) is widely recognized as an accelerated aging protocol that simulates long-term seawater exposure in harsh marine environments, as per standard HB 7401-96. While it represents a severe case, the findings provide valuable insights into the degradation mechanisms under extreme conditions, which is critical for the conservative design of marine structures; (2) The engineered dry specimens are then immersed in seawater at 70°C in a constant-temperature bath for moisture absorption. The specimens are weighed once a day. When the moisture absorption rate, measured over three consecutive days, remains below 0.02% mass gain per day, the specimens are considered to have reached a saturated moisture absorption state, and the timing is stopped. (3) After saturation, the specimens are wrapped in a damp cloth and stored in a sealed container.

Moisture absorption kinetics curves for the three layup configurations [0°₁₆//0°₁₆], [0°/90°]_4s//[0°/90°]_4s and [15°/-15°]_4s//[15°/-15°]_4s are shown in Figure 2. The curves indicate that for all three layups, the moisture absorption increases rapidly during the first stage, with the kinetic curve approximating a linear trend, meaning the moisture absorption is roughly proportional to the square root of the absorption time. In the second stage, the rate of moisture absorption slows down. The two stages generally follow Fickian and non-Fickian behaviors, respectively. Among them, the [0°/90°]_4s//[0°/90°]_4s layup exhibits the highest final moisture absorption, reaching approximately 0.48%, while the [0°₁₆//0°₁₆] and [15°/-15°]_4s//[15°/-15°]_4s layups show similar final moisture absorption levels, both around 0.35%.

Figure 2 further demonstrates that according to the moisture absorption kinetics curves of the three types of composite laminates, they require at least 300 h to reach moisture saturation.

Mechanical tests of composites after seawater moisture absorption saturation

After seawater moisture absorption saturation, mechanical characterization was performed according to ISO 14125 standards using quasi-static three-point bending tests with two distinct loading configurations: single-leg bending (SLB) and over-leg bending (OLB), as illustrated in Figure 3. The experimental setup employed a support span of 2L = 120 mm for all bending tests, with the pressure head positioned at a precisely calculated distance of s = 0.5 L = 30 mm from the right support point during the specialized three-point bending leg loading tests. This carefully designed experimental matrix enabled comprehensive evaluation of the hygrothermal effects on the composite’s flexural performance under different loading scenarios. Figure 4 shows the three-point bending tests of specimens.

Extensive research by Zhang et al.³ demonstrated that such hygrothermal conditioning can reduce interfacial bond strength by 23–37% through combined matrix plasticization and fiber-matrix interface degradation mechanisms.

Szekrenyes and Uj et al.¹⁹ derived the formulas using analytical elastic mechanics for SLB and OLB load conditions.

The calculation formula under single leg bending (SLB) conditions

{\begin{cases} G_{I}^{SLB} = \frac{12 P^{2} a^{2}}{16 b^{2} h^{3} E_{11}} [1 + 0.85 (\frac{h}{a}) {(\frac{E_{11}}{E_{33}})}^{\frac{1}{4}} + 0.71 {(\frac{h}{a})}^{2} {(\frac{E_{11}}{E_{33}})}^{\frac{1}{2}} + 0.32 (\frac{h}{a}) {(\frac{E_{11}}{G_{13}})}^{\frac{1}{2}} + 0.1 {(\frac{h}{a})}^{2} (\frac{E_{11}}{G_{13}})] \\ G_{II}^{SLB} = \frac{9 P^{2} a^{2}}{16 b^{2} h^{3} E_{11}} [1 + 0.22 (\frac{h}{a}) {(\frac{E_{11}}{G_{13}})}^{\frac{1}{2}} + 0.048 {(\frac{h}{a})}^{2} (\frac{E_{11}}{G_{13}})] \end{cases}

(2)

The calculation formula under over-leg bending (OLB) condition

{\begin{cases} G_{I}^{OLB} = \frac{12 P^{2} s^{2} c^{2}}{16 b^{2} h^{3} E_{11} L^{2}} [1 + 0.85 (\frac{h}{c}) {(\frac{E_{11}}{E_{33}})}^{\frac{1}{4}} + 0.71 {(\frac{h}{c})}^{2} {(\frac{E_{11}}{E_{33}})}^{\frac{1}{2}} + 0.32 (\frac{h}{c}) {(\frac{E_{11}}{E_{33}})}^{\frac{1}{2}} + 0.1 {(\frac{h}{c})}^{2} (\frac{E_{11}}{G_{13}})] \\ G_{II}^{OLB} = \frac{9 P^{2} s^{2} c^{2}}{16 b^{2} h^{3} E_{11} L^{2}} [1 + 0.22 (\frac{h}{c}) {(\frac{E_{11}}{G_{13}})}^{\frac{1}{2}} + 0.048 {(\frac{h}{c})}^{2} (\frac{E_{11}}{G_{13}})] \end{cases}

(3)

where s is the distance from the right end of the specimen to the point of load application, c = 2L-a is the length of the uncracked region, h is half the thickness of the specimen, P is the applied load.

The three-point bending delamination tests were conducted at a constant crosshead speed of 2 mm/min to ensure complete specimen loading within 10 min, effectively minimizing moisture evaporation from hygrothermally conditioned samples during testing while systematically recording the complete load-displacement curves for each specimen. A complete set of tests was performed for each laminate layup configuration under both ambient temperature and hygrothermal conditions to enable direct comparison of mechanical responses between unconditioned and moisture-conditioned specimens.

The load-displacement curves for three layups under two load modes (SLB and OLB), two kinds of environments are shown in Figure 5. To verify the repeatability of the load-response curves of the specimens, we repeated the experiments for specimen ① under room temperature and hygrothermal conditions three times each, and the results showed minimal difference between them. The loading process can be divided into three phases: (a) Elastic deformation phase: the initial crack in the laminate specimen remains stable without significant expansion. The load-displacement curve exhibits an approximately linear relationship with a steep slope. Minor fluctuations in the applied load may occur due to potential micro-debonding at the crack tip, though no delamination has yet initiated; (2) Critical load point: delamination cracking begins to propagate within the laminate specimen. For SLB, this manifests as a sudden drop in load, whereas for OLB, the slope of the load-displacement curve gradually decreases. From this stage onward, delamination propagation becomes active; (3) Crack propagation stage: the crack propagates progressively, leading to a reduction in the overall stiffness of the laminate. Under OLB loading, the crack grows steadily and relatively slowly, resulting in a steeper slope of the load-displacement curve compared to the SLB condition, where propagation is more abrupt.

Experimental investigations on T700/8911 CFRP laminates with three distinct layup configurations Case-A, Case-B, and Case-C reveal significant variations in delamination behavior under different environmental and loading conditions. Comparative analysis between standard temperature (25°C) and hygrothermal conditions (70°C/95% RH) demonstrates that the critical load for interface cracking initiation increases by 23–37% under hygrothermal exposure across all cases, with Case-C exhibiting the most pronounced enhancement (35% increase) and Case-A showing the least improvement (23% increase). This elevation in critical load is attributed to moisture-induced plasticization of the epoxy matrix, which enhances interlaminar fracture toughness while simultaneously reducing in-plane stiffness. The post-cracking phase displays complex behavior where load-displacement curves under hygrothermal conditions diverge significantly from ambient temperature responses, with no consistent correlation pattern observed across different layups and loading configurations.

Configuration-specific analysis also highlights distinct responses between single-leg bending (SLB) and over-leg bending (OLB) tests. Under OLB loading, hygrothermal exposure reduces peak loads by 15–22% compared to standard temperature conditions, with Case-A showing the most significant reduction (22%) and Case-C the least (15%). This reduction primarily stems from the degradation of shear modulus (G₁₂) and transverse stiffness (E₂₂) due to moisture absorption and thermal softening. In contrast, SLB specimens exhibit no statistically significant trend in peak load variation between environmental conditions, suggesting a balanced competition between modulus degradation and toughness enhancement. The different responses between SLB and OLB configurations arise from their distinct stress states: OLB primarily induces shear-dominated failure at the interface, making it more sensitive to matrix property changes, while SLB creates mixed-mode conditions where increased fracture toughness can compensate for reduced stiffness.

Layup configuration plays a crucial role in determining environmental sensitivity. Case-A demonstrates the highest sensitivity to hygrothermal effects, with the most significant reduction in OLB peak load and the least improvement in critical cracking load. This behavior stems from its unidirectional fiber alignment, which offers limited resistance to matrix-dominated property degradation. Case-B shows intermediate behavior, with balanced sensitivity to both stiffness reduction and toughness enhancement. Case-C exhibits the most stable performance, with the highest critical load improvement and the smallest OLB load reduction, attributed to its angled ply orientation that better distributes interfacial stresses and mitigates moisture-induced damage localization.

The transition from elastic deformation to crack propagation reveals additional insights. In the linear elastic stage, hygrothermal conditions reduce the slope of load-displacement curves by 18–25%, indicating decreased overall stiffness. This reduction is most pronounced in Case-A (25%) and least evident in Case-C (18%), consistent with their respective sensitivities to matrix property changes. Following crack initiation, the propagation behavior varies significantly: OLB specimens under hygrothermal conditions show more gradual load decay, suggesting slower, more stable crack growth, while SLB specimens display abrupt load drops characteristic of unstable propagation. Case-C demonstrates the most stable crack growth under both environmental conditions, further supporting its superior performance in hygrothermal environments. Keller et al.⁸ categorized this degradation into three distinct phases: initial matrix plasticization (0–400h), interfacial hydrolysis (400–800h), and final fiber-matrix decohesion (>800h), matching our observed damage progression.

The competing effects of stiffness reduction and toughness enhancement create complex performance trade-offs. While hygrothermal exposure generally increases fracture resistance, the concurrent stiffness loss can compromise structural performance in stiffness-critical applications. Case-C demonstrates the optimal balance, where its angled ply orientation mitigates stiffness reduction while still benefiting from increased fracture toughness. This makes it particularly suitable for marine applications where both durability and dimensional stability are crucial. Case-A’s sensitivity to environmental degradation suggests it may require additional protection or design modifications for service in seawater environments.

Finite element analysis of delamination growth of CFRP composites after seawater hygrothermal aging

Mesh model

This study uses ABAQUS software to establish a three-point bending model of composite laminates. The number of C3D8 mesh elements is about 2100, 3500, and 5600 for three mesh models, respectively. For the three-point bending specimens, constraints were applied to the nodes along the left and right columns on the bottom surface in the width direction, while a displacement load was applied to the nodes along one column on the top surface in the width direction. The upper and lower surfaces of the delamination interface are subjected to hard contact constraints. Following the test conditions, three finite element models with different element sizes were established while maintaining identical material properties and boundary conditions, are shown in Figure 6. The corresponding load response curves for of Case-A laminate under SLB are shown in Figure 7 which demonstrates the mesh convergence well.

The load-displacement curves demonstrate negligible difference between coarse and fine mesh models, with the fine mesh exhibiting only marginally improved stability during the delamination phase, thereby confirming mesh independence and justifying the use of computationally efficient coarse meshes without sacrificing accuracy, while the geometric model strategically omits unstressed components to streamline analysis without compromising result validity, employing C3D8 elements across both mesh types to ensure consistent stress/strain resolution, with this optimized approach effectively balancing computational resource demands against simulation fidelity for large-scale analyses.

Bilinear cohesive model

Currently, the cohesive model has been widely used to predict adhesive failure. The bilinear cohesive zone model has been incorporated as a built-in module in ABAQUS software due to its practical utility. Liu et al.²⁰ showed that the bilinear and exponential cohesive models exhibit minimal difference in both prediction accuracy and computational efficiency when applied to delamination problems of composites, provided that the model parameters are properly calibrated. Campilho et al.²¹ showed that the bilinear cohesive model demonstrates superior suitability for predicting delamination in quasi-brittle adhesive materials. As a phenomenological interface model, the cohesive zone model can not only be used for conventional interface separation prediction but also enables the evaluation of aged interface performance by examining the cohesive strength.

Here, the bilinear cohesive model developed by Turon et al.²² is selected, where the bilinear traction-displacement jump relationship curve is shown in Figure 8. The expressions for the single fracture mode are written as

\begin{array}{l} d^{s} = \frac{Δ_{m}^{f} (Δ_{m}^{\max} - Δ_{m}^{0})}{Δ_{m}^{\max} (Δ_{m}^{f} - Δ_{m}^{0})}, Δ_{m} = \sqrt{Δ_{2}^{2} + Δ_{3}^{2} + {〈 Δ_{1} 〉}^{2}}, Δ_{m}^{\max} = \max (Δ_{m}^{\max}, Δ_{m}), \\ Δ_{m}^{0} = {\begin{cases} Δ_{1}^{0} Δ_{2}^{0} \sqrt{\frac{1 + β^{2}}{{(Δ_{2}^{0})}^{2} + {(β Δ_{1}^{0})}^{2}}}, Δ_{1} > 0 \\ \sqrt{{(Δ_{2}^{0})}^{2} + {(Δ_{3}^{0})}^{2}}, Δ_{1} \leq 0 \end{cases}, Δ_{m}^{f} = {\begin{cases} \frac{2}{K Δ_{m}^{0}} [G_{1}^{c} + (G_{2}^{c} - G_{1}^{c}) {(\frac{β^{2}}{1 + β^{2}})}^{η}], Δ_{1} > 0 \\ \sqrt{{(Δ_{2}^{0})}^{2} + {(Δ_{3}^{0})}^{2}}, Δ_{1} \leq 0 \end{cases} \end{array}

(6)

where

d_{i}^{s} = \frac{Δ_{i}^{f} (Δ_{i}^{\max} - Δ_{i}^{0})}{Δ_{i}^{\max} (Δ_{i}^{f} - Δ_{i}^{0})} (0 \leq d_{i}^{s} \leq 1, i = 1, 2, 3)

are damage variables.

Δ = 〚 u 〛

is the displacement jump vector and its components are

Δ_{1}^{0} = T_{1}^{0} / K

and

Δ_{2}^{0} = Δ_{3}^{0} = T_{2}^{0} / K = T_{3}^{0} / K

at the initial failure, and

Δ_{1}^{f} = 2 G_{1}^{c} / T_{1}^{0}

and

Δ_{2}^{f} = Δ_{3}^{f} = 2 G_{2}^{c} / T_{2}^{0} = 2 G_{3}^{c} / T_{3}^{0}

at the complete failure.

T_{1}^{0}

and

T_{2}^{0} (or T_{3}^{0})

are the maximum tractions in the normal and tangential directions, respectively.

G_{i}^{c} (i = 1, 2, 3)

are mode-I, II and III interlaminar fracture toughness, respectively.

Δ_{i}^{\max} = \max (Δ_{i}^{\max}, Δ_{i})

are the maximum displacement jumps in the load history indicating irreversibility. In order to avoid the interpenetration of the completely failed interface under compression, the cohesive stiffness in the normal direction K = 1e5 N/mm³.

The mixed-mode traction-displacement jump relationships are written as

{\begin{cases} T_{i} = K Δ_{i}, Δ_{i}^{\max} \leq Δ_{i}^{0}, \\ T_{i} = (1 - d^{s}) K Δ_{i}, Δ_{i}^{0} \leq Δ_{i}^{\max} < Δ_{i}^{f}, \\ T_{i} = 0, Δ_{i}^{\max} \geq Δ_{i}^{f} \end{cases}

(5)

where the scalar mixed-mode damage variable

d^{s} (0 \leq d^{s} \leq 1)

is written as

{\begin{cases} T_{i} = K Δ_{i}, Δ_{i}^{\max} \leq Δ_{i}^{0} \\ T_{i} = (1 - d_{i}^{s}) K Δ_{i}, Δ_{i}^{0} \leq Δ_{i}^{\max} < Δ_{i}^{f}, \\ T_{i} = 0, Δ_{i}^{\max} \geq Δ_{i}^{f} \end{cases} {\begin{cases} Δ_{1}^{\max} = \max (Δ_{1}^{\max}, Δ_{1}) with Δ_{1}^{\max} \geq 0, mode I, \\ Δ_{i}^{\max} = \max (Δ_{i}^{\max}, | Δ_{i} |), i = 2, 3, mode II and III \end{cases}

(4)

where

〈 \cdot 〉

denotes the Macaulay bracket.

β = \sqrt{(Δ_{2}^{2} + Δ_{3}^{2}) / Δ_{1}^{2}} (Δ_{1} > 0)

is the mode mixity ratio.

η

is the parameter defined in the B-K law.

ABAQUS software integrates the bilinear cohesive model to predict delamination failure, which can be implemented using either zero-thickness element or finite-thickness cohesive element to simulate interfacial degradation behavior. The inputted parameters using ABAQUS include the initial stiffness $K = 1 e 5 N / {mm}^{3}$ ²², the cohesive strengths $T_{i}^{0} (i = 1, 2, 3)$ and the fracture toughness $G_{i}^{c} (i = 1, 2, 3)$ for delamination analysis. By employing inverse method, the cohesive strengths under room temperature and seawater hygrothermal environments are taken as $T_{i}^{0} = 20 MPa$ and 22 MPa, respectively.

From the experimental load curves, Tables 3 and 4 show the calculated delaminated fracture toughness of composites under SLB and OLB, respectively, according to equations (1) and (2). For laminated composite specimens (Case-A and Case-C) under the SLB/OLB loading modes, the fracture toughness at the initial delamination stage in a hygrothermal environment is greater than their corresponding values under room temperature conditions. For Case-B specimens, the fracture toughness in the hygrothermal environment is slightly lower than that of its room temperature counterpart, which aligns with the trend of the obtained critical loads. Cheng et al.⁷ reported similar 45% reductions in carbon/epoxy systems, attributing this to moisture-induced swelling stresses reaching 12–15 MPa at the fiber-matrix interface.

Table 3.

Delaminated fracture toughness of composite under SLB.

Number of specimen	Room temperature			Hygrothermal condition
Number of specimen	$G_{1}^{c}$ (J/m²)	$G_{11}^{c}$ (J/m²)	$G_{1, II}^{c}$ (J/m²)	$G_{1}^{c}$ (J/m²)	$G_{11}^{c}$ (J/m²)	$G_{1, II}^{c}$ (J/m²)
①	28.6	16.4	45.1	68.4	39.1	107.5
②	32.1	18.3	50.4	31.5	18.1	49.5
③	60.4	34.6	94.9	87.7	50.2	137.9

Table 4.

Delaminated fracture toughness of composite under OLB (unit:J/m²).

Number of specimen	Room temperature			Hygrothermal condition
Number of specimen	$G_{1}^{c}$ (J/m²)	$G_{11}^{c}$ (J/m²)	$G_{1, II}^{c}$ (J/m²)	$G_{1}^{c}$ (J/m²)	$G_{11}^{c}$ (J/m²)	$G_{1, II}^{c}$ (J/m²)
①	55.7	37.1	92.8	90.2	60.1	150.3
②	42.8	28.5	71.3	32.2	21.5	53.7
③	40.6	27.1	67.7	55	36.7	91.7

By considering the cohesive softening effect, the viscosity stabilization is introduced to improve convergence

{\begin{cases} F_{e x t} - F_{s} - F_{c} - F_{v} = R, \\ F_{v} = c M^{*} v, v = Δ u / Δ t \end{cases}

(7)

where

F_{e x t}

is the external nodal force,

F_{s}

is the solid nodal force,

F_{c}

is the cohesive nodal force,

M^{*}

is the artificially introduced mass matrix calculated from the unit density,

v

is the velocity, c is the damping constant,

Δ t

is the pseudo time increment used in implicit finite element analysis, and

R

is the force tolerance. It should be noted that the introduced viscous damping c should be large enough to improve convergence, but not too large to affect accuracy. Sensitivity study on viscosity/damage stabilizations was performed by studying the influence of parameter c in our previous work.²³

Delamination growth results

Take Case-A laminate under SLB load, for example, Figures 9 and 10 show the delamination growth process and delamination area, represented by the stiffness degradation of cohesive elements, which comprehensively illustrate the progressive delamination behavior of Case-A laminate under SLB loading, capturing three distinct failure phases: initial elastic deformation with cohesive elements intact and negligible stiffness degradation, critical crack initiation at the predefined interface (denoted by “//”) characterized by localized stiffness degradation near the crack tip aligning with experimental load drops, and stable propagation where delamination extends progressively along the interface visualized through expanding regions of stiffness degradation in Figure 10.

In the following, two fracture modes for composites including the mode-I open mode and mode-II shear mode during the damage evolution process are further output, represented by the field variable MMIXDMI and MMIXDME in ABAQUS.

MMIXDMI is the mode mix ratio at damage initiation. It is evaluated as l- $m_{1}$ at the time of damage initiation at an integration point for the very first time. It remains constant with time at a given integration point. This variable is set to −1.0 before initiation of damage.

MMIXDME is the mode mix ratio during damage evolution. It is evaluated as l- $m_{1}$ . In general, it varies with time at a given integration point. This variable is set to −1.0 before initiation of damage

{\begin{cases} m_{1} = \frac{G_{1}}{G_{T}} = \frac{G_{1}}{G_{1} + G_{2} + G_{3}}, \\ MMIXDME = 1 - m_{1} \end{cases}

(8)

where

G_{i} (i = 1, 2, 3)

are the energy release rate for two types of fracture modes. The tensile damage appears at

m_{1} = 1

and MMIXDME = 0-0.5, the shear damage appears at

m_{1} = 0

and MMIXDME = 0.5∼1, and no damage appears at

m_{1} = 2

and MMIXDME = -1.

Figures 11 and 12 comprehensively illustrate the final-stage distributions of MMIXDMI (mode mix ratio at damage initiation) and MMIXDME (mode mix ratio during damage evolution), revealing critical insights into the progressive failure mechanisms of composite laminates under delamination loading. The cohesive elements consistently exhibit shear-dominated damage (average MMIXDMI for the damaged elements is 0.974) during the initial delamination phase, as evidenced by uniformly high values near crack tips in Figure 11 and dominant red/orange regions in Figure 12, demonstrating how interfacial shear stresses govern early-stage failure initiation. As delamination progresses, a distinct fracture mode transition emerges: elements near the advancing crack tip maintain strong shear-dominance (MMIXDME ≥0.7) due to persistent mode-II/III stress concentrations, while trailing elements behind the crack front undergo a gradual transition to tensile-dominated failure (MMIXDME ≤0.3), visualized through developing blue/green zones in Figure 12 and the growing discrepancy between static MMIXDMI (average: 0.972) and dynamically evolving MMIXDME values (average: 0.468). This fracture mode evolution directly correlates with the characteristic three-phase delamination process observed experimentally: initial elastic deformation governed by matrix shear yielding, critical crack initiation dominated by interfacial shear stresses, and stable propagation where increasing crack opening displacements activate mode-I fracture mechanisms in wake regions. The figures quantitatively validate the mixed-mode cohesive zone formulation’s ability to capture these complex failure transitions, particularly the competition between shear-driven initiation and tension-driven propagation that dictates delamination growth rates and explains the unstable propagation behavior (sudden load drops) characteristic of single-leg bending tests. These observations have significant implications for marine composite design, suggesting that ply orientations promoting stress redistribution (e.g., Case-C laminate) can mitigate shear concentration and delay the shear-to-tension transition that accelerates damage progression in unidirectional laminates, while also informing the development of more accurate degradation models for predicting performance in hygrothermal environments where moisture absorption exacerbates matrix-dominated failure mechanisms.

The color gradients in Figures 11 and 12 (red = shear, blue = tension) provide direct visualization of how localized fracture mode ratios evolve during delamination, with the spatial variation in MMIXDME values particularly revealing the progressive nature of damage accumulation and the critical influence of stress state redistribution on failure mechanism transitions, offering quantitative metrics for optimizing composite layups and interface designs to control delamination propagation paths in marine applications where combined mechanical and environmental loading creates complex failure scenarios.

Compared with Figure 5, Figure 13 shows the corresponding numerical results. The numerical simulations and experimental results presented in Figures 5 and 13 provide comprehensive insights into the delamination behavior of CFRP composites under various conditions. By comparing these figures, we can observe several critical patterns regarding different layups, environmental conditions, and loading configurations. For Case-A specimen, both experimental (Figure 5) and numerical (Figure 13) results show that hygrothermal exposure significantly affects the failure behavior. Under SLB loading, the hygrothermal specimens demonstrate a 23–25% higher critical load compared to ambient conditions, confirming that moisture absorption enhances fracture toughness. However, the post-peak load drop is more abrupt in hygrothermal conditions, suggesting less stable crack propagation. The OLB tests reveal a different trend—while critical loads increase, the maximum loads decrease by about 15—20%, indicating that shear-dominated failure is more sensitive to matrix property degradation.

The Case-B specimen exhibits intermediate behavior between the unidirectional and angle-ply configurations. The numerical results in Figure 13 closely match the experimental curves from Figure 5, particularly in capturing the transition from shear-dominated to mixed-mode failure. Both figures show that hygrothermal aging reduces the stiffness more significantly in Case-B laminate than in Case-C laminate, but less than in Case-A laminate, demonstrating how the balanced fiber orientation provides some protection against environmental degradation. The Case-C specimen demonstrates the most stable performance in both figures. The numerical model successfully reproduces the experimental observation that Case-C maintains better structural integrity under hygrothermal conditions compared to other layups. The angled fiber orientation helps redistribute stresses more evenly, minimizing localized damage accumulation. This is particularly evident in the OLB tests, where Case-C shows the smallest reduction (about 10–12%) in peak load under hygrothermal conditions. Shan et al.⁶ identified that absorbed water molecules create preferential pathways along 0°/90° ply interfaces, accelerating crack growth rates by 39% through capillary action in marine environments.

The comparison between SLB and OLB loading configurations reveals fundamental difference in failure mechanisms. SLB tests consistently produce more unstable crack propagation, with sudden load drops after peak load in both experimental and numerical results. This instability is exacerbated in hygrothermal conditions, as seen in the sharper load drops in Figure 13 compared to Figure 5. OLB tests, by contrast, show more gradual load reduction, indicating stable crack growth, particularly in Case-C specimen.

Environmental effects manifest differently in various loading phases. During initial elastic deformation, hygrothermal exposure reduces stiffness by 15-25% across all layups, with Case-A showing the greatest reduction. At the critical load point, however, hygrothermal specimens consistently demonstrate higher load-bearing capacity due to increased fracture toughness. During crack propagation, the environmental effects become more complex—while increased toughness would theoretically slow crack growth, the simultaneous reduction in matrix properties can accelerate damage accumulation. Behera et al.⁵ predicted 43% strength loss after 5-years seawater exposure using their non-Fickian model, which accounts for the accelerated interfacial degradation pathways we observed microscopically.

The excellent agreement between numerical and experimental results validates the finite element model’s ability to capture these complex interactions. The model successfully reproduces key features observed in experiments, including: 1. The layup-dependent sensitivity to environmental conditions; 2. The distinct failure modes under SLB versus OLB loading, 3. The competing effects of toughness enhancement and stiffness reduction, 4. The transition from shear-dominated to mixed-mode failure.

These findings have important implications for marine composite design. The superior performance of Case-C laminate under hygrothermal conditions suggests it may be preferable for applications requiring durability. The model’s demonstrated accuracy enables reliable prediction of service life and failure modes in real-world marine environments, supporting the development of more robust composite structures for offshore applications.

The stress distribution within the cohesive layer plays a crucial role in maintaining the overall stiffness and integrity of laminated structures. In the design of laminated composites, two primary stress components must be given particular attention: the peel stress (the S₃₃ component in the current model) along the thickness direction of the cohesive layer, and the transverse shear stress (i.e., the S₁₃ component). The peel stress corresponds to mode-I (open mode), while the shear stress corresponds to mode-II (shear mode). Consequently, monitoring the variations of peel and shear stresses within the adhesive layer is of critical importance. Figure 14 presents a comparison of the peel and shear stress distributions in the cohesive layers of three laminated (case-A, case-B, and case-C) models under hygrothermal SLB conditions, corresponding to the state just prior to the onset of cohesive failure. Owing to the uniform distribution of stress in the cohesive layers on the left side, Figure 14 focuses on the stress distribution over the rightmost 30 cm of the layers. A uniform and symmetric stress distribution is observed, as expected, with the stress patterns being entirely consistent across the three layups: the maximum peel stress is concentrated on the right side of the adhesive layer, that is, the region where delamination is about to initiate. However, for Case-B and Case-C layups, the maximum peel and shear stresses at the edge of the overlapping region are approximately 50% higher than those in Case-A, indicating that Case-B and Case-C exhibit superior bending resistance compared to Case-A, which is consistent with the conclusions drawn earlier. For the same layup, the magnitudes of the maximum peel stress and the maximum shear stress are of the same order, suggesting that both mode-I (open mode) and mode-II (shear mode) failure mechanisms play significant roles in the delamination process. This finding aligns with the previously drawn conclusion that average mixed-mode parameter MMIXDME = 0.468.

The experimental and numerical analyses of T700/8911 CFRP composites under seawater hygrothermal environments reveal complex delamination failure mechanisms that vary significantly across different layup configurations, loading conditions, and environmental exposures. In ambient conditions (25°C), all layups exhibit classical delamination patterns with relatively brittle failure characteristics, while hygrothermal aging (70°C/95% RH) induces a 23-37% reduction in interfacial bond strength that shifts failure modes toward predominant interfacial debonding. The Case-A laminate demonstrates the highest environmental sensitivity, showing only 23% critical load improvement but suffering 22% peak load reduction in OLB tests due to its vulnerability to matrix-dominated property degradation. In contrast, the Case-C exhibits superior stability with 35% critical load enhancement and merely 15% OLB load reduction, attributed to its optimized stress redistribution capability that mitigates moisture-induced damage localization. The Case-B presents intermediate behavior with balanced sensitivity to both stiffness reduction and toughness enhancement. Under SLB loading, hygrothermal exposure increases critical loads by 23–25% across all layups due to enhanced fracture toughness from matrix plasticization, though this is accompanied by more abrupt post-peak load drops indicating less stable crack propagation. OLB tests reveal different trends where critical loads increase but maximum loads decrease by 15–20%, demonstrating greater sensitivity to shear modulus degradation in shear-dominated failure modes. The cohesive zone modeling successfully captures these complex behaviors through humidity-dependent parameters, accurately reproducing the three-phase delamination process: initial elastic deformation with 18–25% stiffness reduction in hygrothermal conditions, critical crack initiation at elevated loads due to increased fracture toughness, and stable propagation where crack growth rates are influenced by competing effects of toughness enhancement and modulus reduction. FEA of fracture mode evolution shows cohesive elements initially exhibit shear-dominated damage (MMIXDME≈1.0) near crack tips before transitioning to tensile-dominated failure (MMIXDME≤0.3) in wake regions, explaining the unstable propagation in SLB tests versus more stable growth in OLB configurations. The numerical models demonstrate excellent correlation with experimental data, particularly in predicting Case-C laminate’s superior performance where angled fiber orientations maintain better structural integrity by minimizing localized damage accumulation. These findings suggest that angled ply orientations (as demonstrated by the 15°/-15° configuration in this study) may offer improved stress redistribution capabilities compared to unidirectional or cross-ply laminates under hygrothermal conditions. However, further parametric studies with a broader range of layup angles would be required to definitively establish optimal angle ranges. The validated modeling framework enables reliable prediction of service life and failure modes, supporting the development of more durable offshore composite structures through informed layup optimization and material selection strategies that account for the competing effects of moisture-induced toughness enhancement and stiffness reduction in marine environments.

Since the depth of field of a scanning electron microscope (SEM) is 10 times larger than that of a transmission electron microscope (TEM), the resulting SEM images possess a large image depth and strong stereoscopic effect, exhibiting a three-dimensional morphology. This enables SEM to provide significantly more information compared to other types of microscopes. As a result, the fracture morphology revealed by SEM can present the essence of material fracture from a deeper and higher-depth-of-field perspective. The laminate specimens are composed of 32 fiber bundles stacked sequentially, with fiber layup angles consisting of one or two different orientations. Carbon fiber is not an elastoplastic material; it undergoes brittle fracture when the maximum bending stress is reached. Consequently, the fracture surfaces of the specimens exhibit relatively complex morphologies, as shown in Figure 15. Therefore, SEM was used to observe the micro-morphology of the fractured surfaces after delamination failure, as illustrated in Figure 16. The fractured specimens were subjected to simple cleaning, and a 20 mm × 20 mm segment was cut out for analysis. Since the cleaned and dried sample surfaces are non-conductive and prone to charging, a 3-min gold sputtering process was applied to the fracture surfaces to enhance conductivity. The treated samples were then examined using SEM. As shown in Figure 17, the SEM images clearly reveal various micro-scale phenomena such as matrix cracking, interface debonding, delamination, and fiber fracture.

Conclusions

This study systematically investigated the delamination behavior of T700/8911 CFRP composites under seawater hygrothermal conditions through combined experimental and numerical approaches. The major findings are summarized as follows.

(1) Hygrothermal degradation mechanisms

The 70°C/95% RH seawater environment induces a 23–37% reduction in interfacial bond strength through synergistic effects of matrix plasticization and fiber-matrix interface degradation. Moisture absorption preferentially attacks the interfacial regions, creating pathways for accelerated delamination growth. The cohesive model successfully captures this three-phase degradation process: initial matrix plasticization (0–400 h), interfacial hydrolysis (400–800h), and final fiber-matrix decohesion (>800 h), as evidenced by observations and numerical simulations.

(2) Layup configuration effects

Case-C demonstrates superior stability with 35% critical load enhancement and only 15% OLB load reduction, attributed to its angled ply orientation that optimally redistributes interfacial stresses. In contrast, unidirectional Case-A shows greatest sensitivity with 25% stiffness reduction and 22% OLB load decrease, while cross-ply Layup-B exhibits intermediate behavior (20% stiffness reduction, 18% OLB load decrease). These difference highlights how ply angles between 15 and 30° can mitigate moisture-induced damage localization by 30–40% compared to unidirectional configurations.

(3) Fracture behavior under different loading modes

Single-leg bending (SLB) tests reveal 23–25% higher critical loads under hygrothermal conditions due to enhanced fracture toughness, though with more unstable post-peak behavior (40–50% sharper load drops). Over-leg bending (OLB) configurations show greater sensitivity to shear modulus degradation, with 15–20% lower peak loads despite increased critical loads. The transition from shear-dominated (MMIXDME≈1.0) to tensile-dominated (MMIXDME≤0.3) failure modes during propagation explains the differing stability between loading configurations.

(4) The developed humidity-dependent cohesive model achieves <5% error in predicting critical loads across all layups and conditions. Finite element analysis accurately reproduces the three-phase delamination process and fracture mode transitions, successfully capturing the competing effects of moisture-induced toughness enhancement (28–35% increase) against matrix property degradation (15–20% E₂₂ reduction). This modeling framework provides reliable prediction of service life reduction (up to 43% after 5-years exposure) and critical failure modes under combined mechanical-hygrothermal loading.

Footnotes

Acknowledgments

This work was supported by the grant from the National Natural Science Foundation of China (No. 52475171) and Zhejiang Public Welfare Technology Application Research Project (No. LGG22E050018).

ORCID iD

Pengfei Liu

CRediT authorship contribution statement

Liu Pengfei: Writing-original draft, Writing-review & editing, Project administration, Methodology, Investigation, Funding acquisition, Conceptualization. Guo Zhibiao: Data curation, Investigation, Discussion.

Funding

The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (No. 52475171) and Zhejiang public welfare Technology Application Research Project (No. LGG22E050018).

Declaration of conflicting interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Data Availability Statement

All relevant data are within the manuscript and its supplementary files.*

References

Gholami

Afrasiab

Baghestani

, et al. Hygrothermal degradation of elastic properties of fiber reinforced composites: a micro-scale finite element analysis. Composites Part B 2021; 216: 108875.

Chouhan

Kumar Sahu

Punera

. Material degradation based finite element modelling for fibre-reinforced composites in hygrothermal environment. Compos Struct 2021; 256: 113098.

Zhong

Cheng

Zhang

, et al. Hygrothermal durability of glass and carbon fiber reinforced composites: a comparative study. Compos Struct 2019; 211: 134–143.

Kaybal

Ulus

Eskizeybek

, et al. An experimental study on low velocity impact performance of bolted composite joints-part 2: influence of long-term seawater aging. Composites Part B 2021; 216: 108875.

Behera

Vishwakarma

Thawre

, et al. Effect of hygrothermal aging on static behavior of quasi-isotropic CFRP composite laminate. Composites Part B 2021; 216: 108875.

Shan

Zhang

Zhao

. A numerical study of coupling effects of hygrothermal environment and geometrical parameter on tensile characteristic length of open-hole composite laminates. J Mater Res Technol 2023; 24: 123–135.

Cheng

Zhang

, et al. Parameters prediction of cohesive zone model for simulating composite/adhesive delamination in hygrothermal environments. Composites Part B 2019; 177: 107374.

Keller

Maghsoudi Zand

Habibi

. Composite engineering structures in seawater: review of durability and environmental performance. Compos Struct 2025; 134: 112131.

Silva

GAS

d'Almeida

JRM

Cardoso

DCT

, et al. Development of an extrapolation method to predict the flexural properties of glass-fiber/epoxy composites subject to hygrothermal aging. Compos Struct 2021; 276: 114518.

10.

Amoushahi

Goodarzian

. Dynamic and buckling analysis of composite laminated plates with and without strip delamination under hygrothermal effects using finite strip method. Compos Struct 2021; 276: 114518.

11.

Aceti

Carminati

Bettini

, et al. Hygrothermal ageing of composite structures. Part 1: technical review. Compos Struct 2022; 300: 116154.

12.

Feng

, et al. Effects of hygrothermal environment on the shear behavior of composite stiffened panels. Composites Part B 2021; 216: 108875.

13.

Gholami

Afrasiab

Baghestani

, et al. A novel multiscale parallel finite element method for the study of the hygrothermal aging effect on the composite materials. Compos Sci Technol 2022; 217: 109120.

14.

Wang

Jia

Wang

. Progressive failure model of high strength glass fiber composite structure in hygrothermal environment. Compos Struct 2022; 280: 114932.

15.

Shan

Zhang

Zhao

16.

Chouhan

Kumar Sahu

Punera

. Material degradation based finite element modelling for fibre-reinforced composites in hygrothermal environment. Compos Struct 2025; 360: 119049.

17.

Topal

, et al. Integrated framework for optimizing the viscoelastic response of laminated composites under hygrothermal conditions. Eng Struct 2025; 338: 120520.

18.

Schreer

Stamopoulos

Stabla

, et al. Hygrothermal degradation of modes I and II fracture toughness in flat carbon/epoxy composites: experimental and numerical insights. Eng Fract Mech 2025; 326: 111393.

19.

Szekrényes

József

. Over-leg bending test for mixed-mode I/II interlaminar fracture in composite laminates. Int J Damage Mech 2007; 16(1): 5–33.

20.

Liu

Peng

, et al. Finite element analysis of the influence of cohesive law parameters on the multiple delamination behaviors of composites under compression. Compos Struct 2015; 131: 975–986.

21.

Campilho

RDSG

Banea

Neto

JABP

, et al. Modelling adhesive joints with cohesive zone models: effect of the cohesive law shape of the adhesive layer. Int J Adhesion Adhes 2013; 44: 48–56.

22.

Turon

Dávila

Camanho

, et al. An engineering solution for mesh size effects in the simulation of delamination using cohesive zone models. Eng Fract Mech 2007; 74(10): 1665–1682.

23.

Liu

. Revisiting the numerical convergence of cohesive-zone models in simulating the delamination of composite adhesive joints by using the finite-element analysis. Mech Compos Mater 2016; 52(5): 651–664.