Process-induced residual stresses in composite laminates by different constitutive laws and parametric investigation
Topic(s) : Manufacturing
Co-authors :
Hong MA (DENMARK), Robert PIERCE (DENMARK), Justine BEAUSONAbstract :
In composite manufacturing, the cure protocol significantly influences the quality of the final product, making it a critical aspect of the fabrication process. Numerical simulations offer a viable means to explore the development of residual stress during manufacturing. Therefore, selecting an appropriate constitutive law that accurately reflects material behaviour during cure is crucial for predicting the final residual stress of composites accurately.
This study introduces and compares various constitutive laws, including a linear law, cure-hardening instantaneously linear elastic material (CHILE) law, a viscoelastic law, and a path-dependent law, integrated with a cure kinetics model. These models were implemented into the finite element analysis software ABAQUS for the analysis of a carbon fibre/epoxy (AS4/3501-6) laminate. A flat cross-ply laminate with four symmetric plies [0o/90o/90o/0o] and dimensions of 101.6×101.6×25.4 mm3 was chosen. Due to the laminate's symmetry, one-eighth of the structure was simulated with symmetrical boundary conditions, as demonstrated in Fig. 1a. Initially, a simple viscoelastic model with a single Maxwell element was developed and validated against an analytical model to assess its accuracy (Figs. 1b and c). The simulations for composites using all the adopted models indicated that the development of residual stress mainly occurred during the cooling stage. The results obtained at the specific region (x=z=0, y=5.08 cm, refer to Fig. 1a) representing the highest interlaminar normal residual stress (σzz) within the composites from different models were subsequently compared (Fig. 1d). Models using linear and CHILE laws demonstrated interlaminar normal residual stresses reaching approximately 17.5 MPa before cooling (the degree of cure reached to >0.99), increasing to 44 MPa after cooling. Conversely, employing viscoelastic and path-dependent laws suggested only <5 MPa interlaminar normal residual stress before cooling. However, after cooling to room temperature, the path-dependent law predicted a interlaminar normal residual stress of 32 MPa, approximately 33% higher than that predicted by the viscoelastic law counterpart. This difference was attributed to the simplification of the viscoelastic behaviour used in the path-dependent law. A parametric study was also conducted to understand the influence of cure parameters on residual stress, which provided guidelines for optimising the cure protocol.
This study introduces and compares various constitutive laws, including a linear law, cure-hardening instantaneously linear elastic material (CHILE) law, a viscoelastic law, and a path-dependent law, integrated with a cure kinetics model. These models were implemented into the finite element analysis software ABAQUS for the analysis of a carbon fibre/epoxy (AS4/3501-6) laminate. A flat cross-ply laminate with four symmetric plies [0o/90o/90o/0o] and dimensions of 101.6×101.6×25.4 mm3 was chosen. Due to the laminate's symmetry, one-eighth of the structure was simulated with symmetrical boundary conditions, as demonstrated in Fig. 1a. Initially, a simple viscoelastic model with a single Maxwell element was developed and validated against an analytical model to assess its accuracy (Figs. 1b and c). The simulations for composites using all the adopted models indicated that the development of residual stress mainly occurred during the cooling stage. The results obtained at the specific region (x=z=0, y=5.08 cm, refer to Fig. 1a) representing the highest interlaminar normal residual stress (σzz) within the composites from different models were subsequently compared (Fig. 1d). Models using linear and CHILE laws demonstrated interlaminar normal residual stresses reaching approximately 17.5 MPa before cooling (the degree of cure reached to >0.99), increasing to 44 MPa after cooling. Conversely, employing viscoelastic and path-dependent laws suggested only <5 MPa interlaminar normal residual stress before cooling. However, after cooling to room temperature, the path-dependent law predicted a interlaminar normal residual stress of 32 MPa, approximately 33% higher than that predicted by the viscoelastic law counterpart. This difference was attributed to the simplification of the viscoelastic behaviour used in the path-dependent law. A parametric study was also conducted to understand the influence of cure parameters on residual stress, which provided guidelines for optimising the cure protocol.