AN ADAPTIVE MULTI-FIDELITY METHOD FOR SIMULATING PROGRESSIVE FAILURE IN COMPOSITE STRUCTURES

Abstract

High-fidelity simulation has been widely employed to model matrix cracking and delamination in composite laminates. However, high computational cost is always associated with the use of brick elements with fine mesh in capturing the composite failure accurately. It becomes impractical to analyze the progressive failure in composites beyond coupon-size laminates. In this thesis, we proposed a more computationally efficient framework known as Adaptive Multi-fidelity Model (AMF) that can change the model’s fidelity on the go as the damage propagates. Floating nodes are embedded in the user-defined element to allow it to change between four fidelity states, which are equivalent single layer (ESL) shell, brick, smeared brick and smeared shell. The proposed AMF model has shown that it is capable of modeling progressive failure in composites efficiently without compromising on accuracy. This work provides a pathway for virtual testing beyond coupon-size laminates without the need for intensive computational resources.