@inproceedings{hankins_bio-like_2019,
  title = {Bio-like {Composite} {Microstructure} {Designs} for {Enhanced} {Damage} {Tolerance} via {Machine} {Learning}},
  abstract = {Variations of unique and tailored composite microstructures have been observed in nature and have served as templates for the development of new synthetic materials. Microstructures are studied in fish scales for their penetration resistance, in spider webs for energy absorption, and in seashells and bone for their strength and toughness. However, it has proven difficult to reproduce the properties found in natural materials, due to the interaction between the intricate structures at different length scales. Rather than attempting to replicate these materials (biomimetics), the focus of this work is to use a bio-inspired pattern generation algorithm to search for new topologies that outperform traditional composite structures due to their naturelike design. The bio-inspired pattern generation algorithm employed in this research is known as the Gray-Scott model. This model was selected due to its unique ability to manufacture patterns that propagate with time, allowing the reinforcement volume fraction of the composite structure to be controlled. The model is capable of producing Turing patterns, propagating wave fronts, homogeneous oscillations, and chaos. Traditionally, Turing models have been primarily studied for their applications in morphogenesis and pattern development. However, this research extends the application of the Gray-Scott model by investigating the patterns as physical load bearing structures. A methodology was developed by which the patterns can be converted to structures, analyzed for a desired mechanical property, and optimized via Bayesian machine learning algorithms that yield an improvement of the average quality of structures produced by almost a factor of 10.},
  booktitle = {American {Society} for {Composites} 34th {Annual} {Technical} {Conference}},
  author = {Hankins, Sarah and Kotthoff, Lars and Fertig, Ray S.},
  month = sep,
  year = {2019},
  month_numeric = {9}
}