Multiscale Modeling Of Fiber Reinforced Material Under Dynamics Loading Using The Embedded Finite Element Method
Open Access
- Author:
- Martin, Valerie
- Graduate Program:
- Mechanical Engineering (PHD)
- Degree:
- Doctor of Philosophy
- Document Type:
- Dissertation
- Date of Defense:
- July 25, 2023
- Committee Members:
- Robert Kunz, Professor in Charge/Director of Graduate Studies
Reuben Kraft, Chair & Dissertation Advisor
Parisa Shokouhi, Outside Field Member
Charles Bakis, Outside Unit Member
Francesco Costanzo, Major Field Member - Keywords:
- Fiber reinforced composite
Multi-scale modeling
Embedded elements
Finite Element Analysis
Impact modeling - Abstract:
- The purpose of this dissertation is to investigate an efficient finite element model for fiber reinforced composites to use in the design of armors with complex shapes, incorporating multi-scale modeling, high strain rate behavior, and damage by utilizing the embedded element method. Particularly, the focus is on modeling Dyneema®, a composite consisting of Ultra High Molecular Weight Polyethylene (UHWMPE) fibers set in a polymer matrix. The embedded element method is used to model bundles of fibers as truss elements embedded in the matrix material modeled by standard continuum elements. The truss elements can support both tensile and compressive loads, but not buckling. A large part of this effort involves modifying the embedded element method to make it more conducive to modeling such composites with large fiber volume fractions. To this end, an central difference finite element code is introduced that is programmed to correct the volume redundancy in the embedded element method on an algorithmic level. It is shown that this code is accurate when compared to a commercial code and correctly removes the volume redundancy. Additionally, the embedded element method is implemented in the commercial code Abaqus to model ballistic impact test. The method is shown to be able to capture plate back face deformation, delamination, indirect tension failure, and fiber snap back. The main research questions can be summarized as follows: 1. Can a truss element mesh be easily created for curved or complex geometry? 2. Can the embedded element method be modified to correct the volume redundancy and eliminate the need for material property smearing? 3. Does using embedded truss elements to model fiber bundles provide accurate predictions of material behavior on small scales (tension/compression tests) and larger scales (full plate impact)?