A Multiscale Progressive Failure Modeling Methodology for Composites That Includes Fiber Strength Stochastics

• Published in: Computers, materials & continua materials & continua, 2014, Vol.40 (2), p.99-129
• Main Authors: Ricks, Trenton M, Lacy, Thomas E, Bednarcyk, Brett A, Steven M.Arnold, Hutchins, John W
• Format: Article
• Language: English
• Subjects: Computer simulation; Continuous fiber composites; Distribution functions; Failure; Failure analysis; Fiber composites; Fiber strength; Finite element method; Mathematical models; Metal matrix composites; Micromechanics; Multiscale analysis; Probability distribution; Silicon carbide; Statistical analysis; Unit cell
• ISSN: 1546-2218; 1546-2226
• DOI: 10.3970/cmc.2014.040.099

A multiscale modeling methodology was developed for continuous fiber composites that incorporates a statistical distribution of fiber strengths into coupled multiscale micromechanics/ finite element (FE) analyses. A modified twoparameter Weibull cumulative distribution function, which accounts for the effect of fiber length on the probability of failure, was used to characterize the statistical distribution of fiber strengths. A parametric study using the NASA Micromechanics Analysis Code with the Generalized Method of Cells (MAC/GMC) was performed to assess the effect of variable fiber strengths on local composite failure within a repeating unit cell (RUC) and subsequent global failure. The NASA code FEAMAC and the ABAQUS finite element solver were used to analyze the progressive failure of a unidirectional SCS-6/ TIMETAL 21S metal matrix composite tensile dogbone specimen at 650°C. Multiscale progressive failure analyses were performed to quantify the effect of spatially varying fiber strengths on the RUCaveraged and global stress-strain responses and failure. The ultimate composite strengths and distribution of failure locations (predominately within the gage section) reasonably matched the experimentally observed failure behavior. The predicted composite failure behavior suggests that use of macroscale models that exploit global geometric symmetries are inappropriate for cases where the actual distribution of local fiber strengths displays no such symmetries. This issue has not received much attention in the literature. Moreover, the model discretization at a specific length scale can have a profound effect on the computational costs associated with multiscale simulations.