About the project
You will be joining a collaborative group dedicated to addressing complex real-world engineering problems. The group is focused on conceptualizing cutting-edge data-driven topology and optimization methodologies.
These techniques are specifically developed with the aim of solving real-world engineering challenges such as
- fluid structures
- turbomachinery
- meta-materials, etc.
The University of Southampton boasts extensive High Performance Computing (HPC) and experimental facilities making this a unique opportunity to conduct high fidelity, multi-disciplinary research and collaborate with world-class researchers.
Mechanical design is non-intuitive. Even with years of experience, the non-intuitive behaviour of physical systems, due to our limited understanding of them, can mean the optimal geometry is surprising or even extreme (look up, for instance, the bulbous bow of a ship).
However, the expense of trying out novel designs is usually extremely prohibitive for real-world engineering problems restricting us to unadventurous design spaces. In addition, the geometric constraints and designer biases imposed by conventional methods can restrict the design process to a particular space and make non-intuitive designs impossible.
Data-driven methods have forced a major reconsideration of current research techniques methodologies. These methods, however, have yet to find a footing as a viable component of the conventional design processes due to lack of representative datasets and the inhibitive cost of generating real-world data sets.
The struggle of conventional machine learning algorithms to generate or predict performance of out-of-sample designs also limits their utility for real-world design processes.
