Date of this Version
"A 3-Dimensional Multiscale Finite Element Analysis Of Damage Accumulation in Carbon Fiber Unidirectional Composites."
S. Camaraa - MS, Engineering Mechanics
aUniversity of Nebraska, College of Engineering, 114 Othmer Hall, Lincoln, NE 68588, USA
The purpose of this thesis is to model the mechanical failure of carbon fiber-reinforced epoxy composites due to fiber failure and damage localization. A three-dimensional finite element model is presented in order to determine the lifetime of these structures. Carbon/epoxy composites are increasingly used as parts of industrial structures. These materials undergo damage that can be detrimental to the mechanical strength of these structures. Filament wound composite structures place the fibers on geodesic paths so that when the structure is pressurized the fibers are only subjected to tensile forces. This allows an analogy to be made between the behavior of a 3D filament wound composite structure, such as a pressure vessel, and the behavior of a unidirectional composite. This has been exploited in this study. It is clear that when a unidirectional composite is loaded parallel to the fibers, it is the fibers which determine the strength and which must break if the composite is to break. Among all kinds of damage in composites, such as matrix cracking, fiber/matrix debonding,etc, fiber failure becomes important when fiber breakage density increases, resulting in possible catastrophic destruction of the structure. In this work, this phenomenon is studied locally, on the scale of fibers, both in terms of its causes and consequences. Additionally, at the end of this investigation, an approach using the finite element technique is developed to aid engineers in designing an industrial carbon/epoxy unidirectional composite structure and predicting its lifetime based on fiber rupture. Thus, the originality of this approach is based on modeling the physical phenomena at the microscopic level that are associated with fiber breaks, allowing a prediction of failure in a reasonable amount of computational time. This approach takes into consideration heterogeneity at the microscopic level, which leads to a large number of degrees of freedom and therefore an extensive computational time. A simplified multiscale approach is considered for this study.