Under the multi-field coupling effects of extreme environments and complex loads, large equipment structures are highly susceptible to material property degradation at both macro and micro levels. This degradation can lead to a reduction in structural bearing capacity and pose significant safety risks [1]. Consequently, it is crucial to employ advanced sensors for synchronous health monitoring during the service life of these structures [2]. Optical fiber sensors have multiple advantages such as small size, light weight, electromagnetic interference resistance, and adaptability to harsh environments, which have been widely used in the field of structural health monitoring [3–7]. By embedding optical fiber sensors within a composite structure, we can create an intelligent composite structure capable of self-sensing strain and temperature. This capability significantly enhances the structure’s safety and reliability [8–12].

The common method of embedding optical fibers into composite materials involves placing the optical fibers between two layers of prepreg and curing them together [13]. However, since the diameter of a typical optical fiber (>100 μm) is more than an order of magnitude larger than that of the commonly used reinforcing fibers, this process may introduce defects within the material, potentially affecting the mechanical properties of the overall structure [14–17]. Therefore, it is crucial to study the impact of embedding optical fibers on the performance of composite structures, particularly at the microscopic scale.

Some scholars have conducted mechanical experiments on composite materials with embedded optical fibers. Etches and Fernando [18] found that embedded sensors have a certain impact on material properties under tension and compression fatigue loads. The study observed that sensor performance varied significantly with loading type—tensile loading caused sensor failure at 0.5% strain, while compressive loading allowed sensor operation until composite failure at 1.1% strain [19]. Hadzic et al. [20] embedded optical fibers with different densities and found that as the number and density of optical fibers increased, the tensile and compressive strengths of the material decreased more significantly. Abidin et al. [21] explored the impact of the material properties and thickness of the external coating of optical fibers on the mechanical properties of the structure. The results indicated that when using a polyimide coating, the structure’s Young’s modulus was greater compared to using an acrylate coating. Furthermore, as the coating thickness decreased, the difference in mechanical properties between structures with different coatings embedded became smaller. Shivakumar et al. [22,23] conducted tensile and compression experiments on laminates embedded with optical fibers using a tensile machine and observed the whole process of material fracture failure through a traveling microscope.

To understand the impact mechanisms of embedded optical fibers on the composite structures, some scholars have conducted numerical simulation on the mechanical behavior of composite materials embedded with optical fibers. Eaton et al. [24] used finite element techniques to characterize the mechanical behavior of composite laminates with integrated optical fibers, particularly focusing on stress concentration effects. Their findings revealed a complex interaction of the stress field between the embedded sensor and the laminated matrix material, resulting in significant stress concentrations within the optical fiber, the coating, and the laminate. Korepanov et al. [25] performed computational simulations using finite element method (FEM) to study optical fiber integration effects in braided composite materials’ mechanical performance. Through comparative analysis of stress distributions in fiber-embedded and non-embedded structures, results evidenced the presence of optical fibers would lead to significant local stress concentration, which is noteworthy in the strength analysis of the structure. Fedorov et al. [26] calculated the stress distribution when the optical fiber was embedded in the polymer composite material made of single-layer prepreg and found that the stress concentration factor could be obtained accurately enough by simulating the multi-layer model.

The aforementioned works provide valuable insights into stress distribution, potential failure modes, and the macro-scale impact of optical fiber embedment on composite material behavior. However, existing research has not sufficiently addressed the microstructure of composite materials and the interfacial behavior between optical fibers and structural reinforcement fibers. Consequently, the microscopic influence of embedded optical fibers on composite structures, as well as the failure mechanisms under load, remain unclear.

This paper conducts a numerical simulation of fiber reinforced composites with embedded optical fibers by establishing a microscopic model that includes randomly distributed reinforcing fibers, the optical fiber, matrix, and the interface phase. This study investigates structural damage initiation and progression under transverse loading conditions, analyzing how interface parameters affect computational modeling outcomes. The simulation-derived damage morphology shows strong correlation with experimental findings, offering enhanced insight into the structure’s micro-scale failure behavior.

To conduct reliable numerical modeling and damage behavior analysis, it is essential to accurately characterize the microstructure of the composite material embedded with optical fibers. A typical optical fiber is composed of three parts: the inner core, the cladding, and the coating layer, as illustrated in Fig. 1. Both the core and cladding are made of silicon dioxide, but they have slightly different refractive indices [27]. The coating material is typically a polymer, used to protect the optical fiber from mechanical damage. In this study, optical fibers with a polyimide (PI) coating were embedded into unidirectional glass fiber-reinforced epoxy composite laminates. Fig. 1b shows the microscopic structure of the composite laminate embedded with one optical fiber, as observed through a scanning electron microscope (SEM). Since the core and cladding are made of the same material, they cannot be distinctly identified in the image. The reinforcing glass fibers are randomly distributed around the optical fiber.

Based on the SEM image, plain strain finite element models of the composite laminate embedded with one optical fiber are constructed, as shown in Fig. 2, by using representative volume elements (RVEs) of the material with varying sizes. Given that the core and cladding of the optical fiber are made of the same material, they are treated as a single component in the model. The optical fiber is positioned at the center of the RVEs. The core and cladding together have a diameter of 125 μm, while the overall diameter of the optical fiber is 160 μm. The random distribution of the reinforcing fibers is generated using the random sequential expansion (RSE) algorithm [28] proposed by the author. Firstly, an RVE model of a specific size is generated using the RSE algorithm. Then, the optical fiber is positioned at the center of the model, and the fibers that overlap with it are removed, resulting in an RVE with an embedded optical fiber. Since the size of the optical fiber is fixed, the size of the RVE reflects the density of optical fibers embedded in the composite. In this study, an RVE with dimensions of 300 μm × 300 μm is used to simulate a highly dense embedding of optical fibers. For a relatively sparse embedding of optical fibers, RVEs with dimensions of 600 μm × 600 μm and 1000 μm × 1000 μm are used. Once the RVE size is determined, the number of fibers in the model can be determined according to the fiber volume fraction. Fig. 2 displays the three distinct RVEs with different sizes but same fiber volume fraction of 50%, which respectively contains 431, 2498, and 7377 reinforcing fibers with a diameter of 10 μm. To study the influence of the embedded optical fiber, another three RVEs without optical fiber are also constructed for comparison.

The finite element models of the RVEs are established by using software ABAQUS. The reinforcing fibers, matrix, and optical fiber are discretized by mainly 4-node bilinear plane strain reduction integration elements (CPE4R) and a few 3-node triangle elements (CPE3). At the same time, 4-node 2D cohesive elements (COH2D4) with zero thickness are inserted not only between reinforcing fibers and the matrix but also between the optical fiber and the matrix to simulate the interface behavior. The number of nodes at the RVE boundary is adjusted to be consistent so that the periodic boundary conditions can be applied. Transverse tensile and transverse compressive loads are applied to the RVEs respectively to analyze their mechanical responses. An analysis of the mesh convergence was carried out, with element sizes of 1, 0.9, 0.8, and 0.5 μm selected for separate calculations. The computational results indicate that there is little difference between the outcomes of varying mesh sizes; however, reducing the mesh size significantly increases the computation time. Therefore, a mesh size of 1 μm is used in this study.

The material composition of the model and corresponding elastic parameters are listed in Table 1. Since damage is unlikely to occur in the reinforcing fibers and the optical fiber under transverse loading, they are modeled as linear elastic materials. However, for the matrix and the interfaces, it is crucial to accurately describe the damage and failure behavior to ensure a realistic representation of the material’s mechanical response under load.

For the epoxy resin, whose behavior is sensitive to the hydrostatic stress, its yield behavior is described by the extended linear Drucker-Prager criterion [29]: (1)F=t−ptan⁡β−d=0,t=12q[1+1k−(1−1k)(rq)3]

The yield criterion incorporates the hydrostatic stress (p), Mises equivalent stress (q), and the third deviatoric stress invariant (r). The linear yield surface slope in the p–t stress plane is denoted by β, while d represents material cohesion. Additionally, k accounts for differing yield responses under tension and compression by defining the ratio of triaxial tensile yield stress to triaxial compressive yield stress.

Material failure initiation is determined using a ductile fracture criterion, with the critical plastic strain dependent on stress triaxiality conditions. Post-initiation, damage development follows a progressive failure model. The primary damage effects include yield stress reduction and elastic property deterioration, both regulated by an evolving damage parameter that increases with accumulated material degradation. For further details on the epoxy material model, please refer to our previous publication [29]. The specific damage parameters for the epoxy resin are provided in Table 2.

In addition to the matrix, composite material failure is strongly affected by interface mechanics. The cohesive model is applied here to simulate interfacial behavior, linking the traction vector to the displacement difference across the element’s surfaces. The bilinear constitutive model is employed to capture the stress-displacement relationship of the interface.

In the linear stage, the stress-strain relationship of the interface is given by [29]: (2)t={tnts}=[Kn00Ks]{εnεs}where, tn and ts represent the normal stress and the shear stress acting on the cohesive element, respectively, Kn and Ks represent the corresponding stiffness coefficients, respectively, εn and εs represent the corresponding strains, respectively. Damage is assumed to initiate when the maximum nominal stress ratio reaches a value of one [29]: (3)max{⟨tn⟩tn0,tsts0}=1where, tn0 and ts0 represent the normal strength and shear strength of the cohesive element, respectively. As damage initiates, the traction stress decreases in accordance with the interface damage parameter. The failure displacement is governed by the fracture energy, which represents the integral of the traction-separation curve [29]: (4)GC=∫0δft(δ)dδwhere, GC represents the critical strain energy release rate, δ represents the opening displacement of the cohesive element during the damage process, while δf denotes the opening displacement when the cohesive element ultimately fails. When δ = δf, it indicates that the cohesive element has completely failed. The interface parameters are selected from reference [24], with the interface stiffness of 10⁸ GPa/m and the fracture energy of 100 J/m². The glass fiber-matrix interface strength is 39 MPa, while the optical fiber-matrix interface strength is 27 MPa [30].

This paper adopts a micromechanical approach to investigate the influence of embedded optical fibers on the mechanical and failure behavior of composite materials. Representative Volume Elements (RVEs) of different sizes were constructed, and then transverse tensile and compressive loads were applied to these RVEs to study the micro-damage mechanisms and failure processes.

The results show that under transverse tensile loading, the interface between the optical fiber coating and the matrix first reaches its failure strength, leading to debonding at the interface and the formation of curved cracks. Then the surrounding matrix gradually fractures as well. Eventually, the microcracks caused by matrix fracture connect with the cracks formed by interface debonding, penetrating the RVE and causing material failure. Under transverse compressive loading, the damage also initiates in the optical fiber-matrix interface, but in a direction perpendicular to the load, and plastic cracks form in the matrix around the optical fiber. As the compressive load increases, the debonding area and matrix cracks expand, forming two parallel shear bands that ultimately lead to the failure of the RVE.

The investigation utilized representative volume elements (RVEs) of varying dimensions to simulate different optical fiber distributions. Computational results demonstrate that fiber density variations minimally impact the material’s final failure mode. Experimental characterization of post-loading microstructures corroborated these findings, confirming the model’s validity.

Furthermore, parametric studies of fiber-matrix interfacial properties showed substantial influence on mechanical response. Enhanced interface strength was found to improve tensile strength but reduce fracture strain. Beyond a critical interfacial strength threshold, stress concentrations near fibers became insufficient to initiate debonding, shifting the failure mechanism to matrix-dominated fracture.

In addition to the research content addressed in this paper, future studies could delve deeper into the following aspects. Investigate the mechanisms of material damage and failure after embedding optical fibers in composite materials along different directions, and develop a multiscale mechanical model for composite laminates with embedded optical fibers. Moreover, the laminates could employ various stacking angles to reveal the mechanisms of material damage and failure under different loading conditions. Additionally, explore the mechanical properties of composite materials with embedded optical fibers under fatigue, impact, and other types of loading, and test the survivability of the optical fiber sensors inside the materials to explore the application prospects of this material under more complex working conditions.