Experimental Analysis of the Mechanical and Thermal Behaviour of Epoxy and Polyester GFRP Composites for Wind Turbine Blade Skin

The transition to renewable energy has become a global priority to mitigate climate change. Indonesia targets Net Zero Emission (NZE) by 2060 and plans to expand wind power capacity to 597 MW by 2030 [1]. However, its current utilisation is still far below the target, reaching only 154.3 MW in 2024. One of the main challenges is the high cost of repairing wind turbine blades due to cracks in the transition zone on the blade, which significantly affects financial losses [2, 3]. According to Cavalcanti et al. [4], who conducted research in polymers, the results found that selecting matrix polymers has a fundamental role in determining wind turbine blades' structural, thermal, and economic performance.

The reliability of wind turbine blades plays a critical role in reducing maintenance expenses, as the transition zone of wind turbine blades are the most vulnerable component due to complex mechanical and environmental loads [5]. Validation is generally performed through full-scale testing under IEC 61400-23 standards to ensure structural safety, focusing on statistical loads and fatigue in flap-wise and edge-directions of blade wind turbine [6, 7]. While essential, these full-scale tests are costly and often insufficient to capture local failure mechanisms in blade subcomponents [8, 9]. Recent studies have emphasized the importance of subcomponent-level investigations, which can provide detailed insights into progressive damage phenomena such as delamination, debonding, and fiber fracture in composite turbine blades [10].

During operation, blades are subjected to a combination of tensile, torsional, and flexural loads, leading to critical shear stress between the skin and the core in sandwich composite structures [11]. Glass fiber reinforced polymer (GFRP) is commonly utilized as the primary skin material due to its favorable mechanical properties, including high specific strength, corrosion resistance, and relatively low cost [12, 13]. Recent works also highlight that GFRP offers stable performance under cyclic loading, making it a reliable choice for offshore and onshore wind turbine applications [14, 15]. Chen et al. [16] investigated the failure phenomena of the skin-core sandwich composite HAWT blade by combining static testing and FEA simulation on a large composite blade and using 3D failure criteria, showing an approach to understand skin and core cracks, including debonding and delamination. The Braga and Magalhaes [17] explained that glass fiber reinforced polymer matrix (GFRP) has good stability because the glass fiber inhibits polymer degradation as a matrix. Polymer resistance generally begins at the initial degradation temperature at 300℃.

Chichane et al. [18], have confirms that the prediction of composite mechanical properties is strongly influenced by fiber distribution, volume fraction, and matrix-reinforcement interactions. This study demonstrates that micromechanical models, such as the Halpin–Tsai models, effectively explain variations in modulus due to changes in composition and bond quality between components. These findings are relevant to the development of GFRP for wind turbine blade skins, where structural performance is determined mainly by the selection of the matrix and the stress transfer efficiency of the glass fibers. Halpin-Tsai prediction models integrated experimental and micromechanical approaches for epoxy and polyester composites to enhance the mechanical durability of wind turbine blades.

The manufacture of laminated composites with Vacuum Assisted Resin Transfer Moulding (VARTM) is a composite fabrication technique that utilises vacuum pressure to flow liquid resin into a stack of dry fibers hermetically sealed by bagging film. Once the vacuum is achieved, the resin is drawn through an inlet until all the fibers are evenly wetted. This process enhances impregnation, minimises voids, and yields a more controlled fiber volume fraction compared to conventional methods. VARTM is known for its efficiency in producing large structures and is capable of creating laminates with superior mechanical properties and consistency [19].

The novelty of this research is its focus on a comprehensive evaluation of the mechanical and thermal properties of wind turbine blade skin materials using twist-tensile and three-point bending test approaches. Unlike previous studies that focused on full-scale tests and numerical modelling, this study makes a new contribution by providing a local characterisation of HAWT blade skin materials through the integration of multi-angle twist-tensile tests (0°, 20°, 40°, 60°) and span-dependent flexural analysis. This study presents new design guidelines for the influence of fibers misalignment and span length on load transfer efficiency and resin selection in the blade transition zone, which is often damaged. No previous studies have analysed the mechanical behaviour of epoxy and polyester under combined twist–tension conditions that resemble blade transition loads.

3.1 The Halpin-Tsai model

The Halpin-Tsai approach is a micromechanical model widely used to predict the modulus of elasticity fiber-reinforced composites. Unlike the Rule of Mixtures, which tends to provide an upper-bound estimate, the Halpin–Tsai model incorporates the effects of fiber geometry, orientation, fibers-to-matrix ratio, and stiffness, resulting in more realistic predictions for various types of composites, including short fibers and random orientations such as in CSM. The Table 3 is data sheet the E-glass fiber and polymer resin. The calculate fiber volume fraction and modulus of elasticity composite shows on Eqs. (1)-(3).

Table 3. Properties data of polymer and E glass fiber

In general, the composite modulus is expressed in the equation:

$E_c=E_m \frac{1+\eta V_f}{1-\eta V_f}$     (1)

$\eta=E_m \frac{\left(E_f / E_m\right)-1}{\left(E_f / E_m\right)+\xi}$     (2)

$V_f=\frac{\text { mass per area }}{\rho_f \cdot t}$     (3)

Note:

$E_c=$ Modulus of composite (GPa)

$E_f=$ Modulus of fiber (GPa)

$E_m=$ Modulus of matrix (GPa)

$V_f=$ Fiber volume fraction (%)

$\rho_f=$ Fiber density (gr/cm³)

$\eta=$ reinforcement factor depend on geometry

$\xi=$ shape orientation parameter

$t=$ thickness (mm)

$L=$ Length of fiber (mm)

$d=$ diameter of fiber (mm)

Note:

$\xi=2$ for longitudinally oriented fibers

$\xi=\frac{2 . L}{d}$ for randomly oriented fibers

From the Table 3, the tensile test specimen has a density is 2.5 g/cm³. The four plies laminate (one ply WR or CSM is 300gsm), then to calculate the Fiber volume fraction (V_f):

$V_f=\frac{4(\text {plies}) \cdot 300 \mathrm{~g} / \mathrm{cm}^2}{2.5 \mathrm{~g} / \mathrm{cm}^3 \cdot 0.11 \mathrm{~cm}}=43 \%$

Figure 4 shows that at a fiber volume fraction of 43%, the aligned fibers with a longitudinally oriented fiber factor ξ = 2 predict a modulus of composite (E_c) of approximately 6.7 GPa. In comparison, randomly oriented fibers (CSM), using ξ = 2.L/d, produce a predicted modulus of composite (E_c) of approximately 3 GPa.

Figure 4. Prediction modulus of elasticity composite based on the Halpin-Tsai model

3.2 Twist-tension analysis

Twist-tensile testing was performed to provide a more realistic description of the bond quality between resin and glass fibers under complex loading conditions. In conventional tensile testing, the dominant load was axial; thereby, not fully representing real conditions encountered in wind turbine blades, where the material experiences a combination of tensile and torsional loads. The test approach approximating real conditions is depicted on the angle measurement unit connected to a clamp or jig (Figure 5).

Figure 5. Twist-tensile of GFRP composite set on upper and bottom clamping with angle measurement

Figure 6(a) portrays that GFRP-epoxy resin exhibits significantly higher tensile strength (334 MPa) than GFRP-polyester (250 MPa) at 0° twist, indicating superior load transfer capability along the fiber direction. As the twist angle rose, GFRP-epoxy strength declined rapidly, particularly between 20° and 40°. At the same time, polyester remained relatively stable with a slight increase, suggesting better load distribution through fiber-matrix interaction at moderate angles. The at 60°, both resins experienced a significant drop, GFRP-epoxy (198 MPa) slightly lower than GFRP-polyester (203 MPa), signifying reduced stress transfer efficiency at high fiber misalignment. 

The variations in the torsion angle directly affect the effectiveness of stress transfer, with the fiber-matrix interface being a determining factor. The decrease in strength and modulus at an angle of 60° can be explained by the reduced contribution of fiber orientation and the reduction of effective fiber length (effective fiber length) according to the Kelly–Tyson theory. Meanwhile, the relatively stable change in modulus at 0° to 40° of twist angle consistent with the Halpin–Tsai prediction, which suggests that stiffness is strongly influenced by the volume fraction and orientation of fibers.

The difference in tensile modulus between the two types of polymers is relatively small, so that the predicted Halpin-Tsai model value is obtained for short fiber (CSM) E_c = 3.0 GPa and longitudinal fiber (WR) E_c = 7.5 GPa. The actual fiber volume fraction (V_f) of the GFRP composite with a WR/CSM/WR/CSM configuration was 44%. These results show consistency between micromechanical theory and the structural characteristics of laminates, which confirms that the Halpin-Tsai model can be used as a reliable approach to predict the increase in stiffness due to fiber addition in glass fiber-reinforced composites in tensile testing.

The modulus of elasticity the GFRP composite in Figure 6(b) indicates that variations in the twist angle significantly influence its stiffness response. In both matrix systems, epoxy and polyester, there is a tendency for stiffness to increase at low angles, followed by a decrease at higher angles. This pattern indicates that fibers orientation remains effective in resisting flexural loads at minor angle (20^o) deviations; however, its effectiveness decreases as the twist angle increases due to fibers misalignment and increasingly dominant stress redistribution. In general, the epoxy-based composite exhibits higher stiffness than the polyester, reflecting a better interfacial bond between the matrix and the fibers. This finding aligns with the principles of composite micromechanics, where changes in fibers orientation directly affect the load transfer capacity and effective modulus of the laminate.

Figure 6. Twist-tensile test of GFRP composite (n_WR = 2 and n_CSM = 2): (a) tensile strength and (b) modulus of elasticity

3.3 Flexural test analysis

The flexural test results for the two span configurations (96 mm and 60 mm) are displayed in Figure 7, demonstrating the consistent superiority of GFRP-epoxy composite (n_WR = 5 and n_CSM = 5) in flexural strength resistance, reaching 503 MPa at the short span and 519 MPa at the long span. It was compared to GFRP-polyester at 334 and MPa 387 MPa, respectively. However, the modulus of elasticity depicted different behavior. The results of the durability test show that the modulus of elasticity on epoxy are 14.4 GPa (long span) and 19.2 GPa (short span), while polyester is 14.8 GPa (short span) and 16.8 GPa (long span). This indicates that the flexural loading condition is highly susceptible to the span length, where shorter spans reduce the influence of rotation and twisting shear, resulting in higher modulus values.

Figure 7. Three-point bending behavior of the GFRP composite (n_WR = 5 and n_CSM = 5): (a) Flexural strength and (b) modulus of elasticity

These results are consistent with the study by Beter et al. [25], indicating that fiber orientation causes the fiber contribution to the tensile load to be distributed on the surface of the laminate composite. When the specimen was twisted at an angle below 45°, the resulting tensile response of the laminate composite was relatively more homogeneous due to the contributing fibers, which is a critical factor in determining the performance of the composite.

This difference was related to the contribution of shear deformation and the bending moment of the laminated composite. The shear resistance of GFRP-epoxy was superior to that of polyester, making it more resistant to bending loads and capable of larger deflection before failure. In contrast, GFRP-polyester exhibited higher stiffness but lower resilience, corresponding to its higher modulus of elasticity. Therefore, GFRP-epoxy is more suitable for applications requiring high flexural strength and good deflection capacity, while GFRP-polyester is preferable where structural stiffness is the primary concern.

The Halpin-Tsai predictions, it is observed that the model theoretically yields a modulus of approximately 7.0 G.Pa for the dominant fiber orientation (ξ = 2) and approximately 3.0 GPa for the random CSM orientation (ξ = 2.L/d). The significantly higher experimental values, indicate that the flexural response is not only controlled by the fiber–matrix micromechanics, but also by the structural contribution of the laminate, especially the WR layer which is dominant at the surface of the cross-section and increases the flexural stiffness. Thus, the difference between the theoretical predictions and the experimental results can be explained by a more complex flexural mechanism than the tensile one.

The span length in the three-point test significantly affected the composite material’s flexural strength and modulus of elasticity. The results disclosed that at small span-to-thickness ratios (L/h), failure tended to be influenced by local deformation and compression, resulting in higher but less representative apparent strength values. However, at excessively large (L/h), local flexural failure predominated, deflection increased, and the measured strength and modulus of elasticity were likely to decline [26].

The difference in flexural modulus at various span lengths can be explained by the loading mechanism changing from fiber-dominated to matrix-dominated. At short spans, the bending response of GFRP-epoxy is primarily influenced by the outer fiber layer, exhibiting higher stiffness due to more efficient adhesion and load transfer. Conversely, at long spans, the components vibrate more, and the matrix plays a greater role in resisting the load. Polyester has a relatively larger shear modulus value at low–medium strains, so that GFRP-polyester composites can appear stiffer in long-span tests. It explains the phenomenon of span-dependent flexural behaviour seen in the experimental data.

3.4 Failure modes microstructure analysis

The SEM images in Figure 8(a) shows the fracture morphology of the GFRP-epoxy composite (n_WR = 5 and n_CSM = 5) after a tensile test. It was afforded to fill the inter-fiber spaces, resulting in a strong matrix-fiber bond. In addition, some fibers fractured brittlely, while others exhibited fiber pull-out, indicating a combination of fiber damage and debonding mechanisms. The presence of fibers still encased in epoxy confirms the epoxy resin’s good wetting ability than glass fibers.

Figure 8(b) demonstrates the failure modes of GFRP-polyester, showcasing significant damage of delamination, fiber pull-out, and split cracks in the matrix interface. These phenomena indicate a weak bond between the polyester resin and the glass fiber, generating suboptimal load transfer. Consequently, the material exhibited easier delamination fracture behavior. The polyester resin did not provide a good load distribution on the shear stress phenomena.

Figure 8. SEM image after tensile test; (a) GFRP-epoxy failure mode (b) GFRP-polyester failure mode

Figure 9. SEM image after bending test; (a) GFRP-epoxy failure mode (b) GFRP-polyester failure mode

Figure 9(a) depicts the SEM fracture image after the flexural test. In the GFRP-epoxy composite, the matrix exhibited fiber-matrix interfacial bonding, characterized by effective wetting. The load tension distributed evenly from the matrix to the fiber was due to the load tension. Debonding was minimal and localized, in contrast to the GFRP-polyester, which disclosed more frequent debonding and fiber pull-out phenomena. In addition, brittle fiber fracture and homogeneous matrix damage confirm the structural compactness of the Epoxy. Overall, GFRP-epoxy could provide stronger bonds, more effective load distribution, and higher buckling resistance than polyester.

Figure 9(b) illustrates that the failure mode on the SEM image is dominant fiber pull-out, delamination, and debonding, indicating suboptimal interfacial bonding between the glass fibers and the polyester matrix, due to non-homogeneous resin wetting. This prevented efficient load transfer to the fibers but released through a pull-out mechanism. Debonding the fibers and leaving the imprint on the matrix indicates poor bonding. Its failure is highly dependent on the wetting ability of the polymer resin and fiber arrangements, which have a significant role in the failure behaviour [27-29].

According the Chowdhury et al. [30] and Matykiewicz et al. [31], during the curing process, epoxy resin forms a dense three-dimensional network through cross-linking. This structure produces polar groups, such as -OH (hydroxyl) and ether bonds (C-O-C) that interact strongly, including hydrogen bonds with silanol groups (Si-O-H) on the glass fiber surface. This combination strengthens interfacial adhesion and enhances the resin's load transfer ability, as evidenced by the minimal fiber pull-out phenomenon in SEM images.

In contrast, polyester resins use free radical polymerization, resulting in a sparser cross-linked network and fewer polar groups. It makes the interaction with the glass fibers primarily dependent on weaker van der Waals forces, resulting in a greater tendency for debonding and fibers pull-out in polyester-based composites.

3.5 Thermal stability of GFRP composite

The results of Thermogravimetric Analysis (TGA) and Derivative Thermogravimetry (DTG) tests on GFRP-polyester and GFRP-epoxy (Figures 10 and 11). The TGA chart shows differences in thermal resistance. Based on the TGA curve, GFRP-epoxy experiences degradation T_onset at temperatures around 340℃ and T_max at range of 360–370℃, while GFRP-polyester begins to degrade at lower temperatures, around T_onset at 300℃ and T_max at range of 350–360℃. It indicates that epoxy resin has higher thermal stability than polyester. The residual value of both, indicating a slower decomposition rate and better char formation. The DTG chart shows that GFRP-epoxy reaches a maximum decomposition rate (-0.6%/min) higher than GFRP-polyester (-0.4%/min). Although epoxy decomposes faster at peak temperatures, both composites exhibit good thermal resistance within the test temperature range. The sharper DTG peak in epoxy indicates a single-stage degradation process, while the wider peak in polyester indicates a slow and gradual decomposition process.

Figure 10. TGA chart determine weight loss, T_onset and T_maximum

Figure 11. DTG chart determine maximum decomposition rate

Despite these characteristics, both materials have good thermal stability with the maximum temperature around 350℃, so they can be considered suitable for structural applications. These results are similar to the research of Yudhanto et al. [32] investigating thermal stability on glass fiber reinforced polyester, which obtained results of the initial temperature (T_onset) degradation of 300℃ and the maximum temperature (T_max) degradation of 350℃. Then another researcher, Boopalan et al. [33] reported that epoxy composites showed higher degradation temperatures and residual yields than polyester due to the denser cross-linked structure of epoxy.

The study of Braga and Magalhaes Jr [17] also supports the idea that adding glass fiber increases the thermal stability of both of resins. However, epoxy still showed better heat resistance to the matrix volatilisation process which initial degradation.

The authors would like to express their gratitude to the Ministry of Higher Education, Science, and Technology, Republic of Indonesia (KEMDIKTISAINTEK), for providing funding support through the Fundamental Research Grant with Contract Number: 0498.07/LL5-INT/AL.04/2025. This support is highly significant in the implementation and completion of this research.