Composite Damage Prediction

Overview

There has been significant progress in models that serve as predictive tools for evaluation of damage resistance and damage tolerance in composite laminates. High-fidelity finite element models have been demonstrated with close correlation to experimental data in both damage prediction (ie. crack locations/paths/sizes, delamination locations/areas/shapes), and with errors in compressive residual strength within 10% of experimental results.

The main drawback of these models is high computational cost. If an efficient (enough), validated model can be developed, then iterative methods useful to design, such as optimization and uncertainty quantification, become tractable. The low-hanging fruit of improving the computational efficiency of these finite element models is by reducing the complexity of the model, ideally by developing a model that yields accurate results with a coarser discretization, such as moving from solid to shell FE formulations. To achieve accurate predictions for general laminates, delamination must be captured discreetly - either through modeling each ply or adaptive fidelity techniques, both of which I have considered over the course of this work.

Final results and publications in progress.

Contributions

• Validated material model and modeling technique in Abaqus for predicting damage propagation and residual strength for aerospace composites

• Advanced state of user element and damage algorithm formulated for capturing delamination damage discreetly with adaptive fidelity - exhibiting 1-2 orders of magnitude improvement in computational cost

Building on previous work in our lab, I have developed a progressive damage modeling technique based on (1) intraply Hashin damage criteria and energy-based damage propagation and (2) delamination damage based on cohesive zone modeling in Abaqus/Explicit. An example of results from this model is shown below for simulations of ASTM standard LVI/CSAI tests, in this case predicting residual strength within ~1%.

Low Velocity Impact (LVI)

Residual Compression Strength After Impact (CSAI)

Structural Response and CSAI Prediction

Modeling LVI/CSAI with AFShell

To mitigate computational costs associated with modeling composite damage, a novel shell element enriched with the Floating Node Method and a damage algorithm based on the Virtual Crack Closure Technique has been demonstrated that captures complex damage mechanisms such as delamination propagation and transverse matrix cracks while exhibiting a runtimes 92 times faster than existing high-fidelity models. These efficiency gains are realized by the ability to model an entire laminate with just one layer of elements, while still discretely modeling delamination damage. AFShell is currently implemented as an Abaqus/Standard user element (Fortran). Example of delamination propagation predicted during an impact simulation is shown below.

Results using this model show onset of delamination predicted properly, and a pattern of damage qualitatively matching that seen in experiments.