Abstract
For decades, scientists and engineers have been striving to develop materials that are both light-weight and strong, stable, and resilient, with applications ranging from aerospace to the biomedical industry. While conventional materials have been able to deliver the necessary mechanical properties for many applications, they are often associated with a trade-off between weight, stiffness, and strength. In order to address this issue, cellular solids have emerged as a promising solution, which allocate material efficiently along specific directions to reduce the weight-stiffness/strength trade-off.
With recent advances in additive manufacturing, it is now possible to create geometrically complex architected materials with unprecedented mechanical properties and functions. This has rapidly expanded the design space by combining different base materials and architectures. However, exploring and exploiting these large and complex design spaces quickly becomes unfeasible with analytical modeling or brute-force algorithms, and computationally intractable with traditional physics-based numerical simulations.
In response to these challenges, this thesis aims to combine machine learning (ML) with numerical simulations to develop computational frameworks that can understand and discover novel high-performing architected materials. In particular, the goal is to apply Deep Learning algorithms to predict the stiffness properties of an open-cell lattice structure.
This approach is highly innovative, as it enables the development of highly efficient and accurate prediction models that can guide the design of materials with desired properties. By leveraging the power of ML algorithms, this approach can potentially save significant amounts of time and resources that would otherwise be required to conduct traditional numerical simulations. Moreover, the approach has the potential to lead to the discovery of novel architected materials with unprecedented mechanical properties, which can have transformative impacts on a wide range of industries.