Contact line phenomena entail multi-physics processes that operate at disparate scales, often posing formidable large-scale computing challenges.
Our current inability to leverage the predictive capabilities of computationally intensive and fully resolved models for the optimal design of wetting applications leads to approaches which are largely driven by intuition and empirical observations.
The proposed project aspires to address this challenge by providing efficient and accurate data-driven alternatives to fully resolved wetting hydrodynamics simulations. These data-driven surrogate models can be used for the design of surface features for controllable droplet transport, avoiding the need for computationally expensive parametric studies using large-scale simulations on HPC resources.
This project proposes the use of the efficient and highly scalable code Basilisk to generate wetting hydrodynamic datasets. More specifically, Basilisk will be used to simulate cases of initially spherical droplets that travel across chemically heterogeneous surfaces.
Each case will consider surfaces with different chemical heterogeneity profiles, such as smooth random features, striped features and checkerboard patterned features. The generated datasets will be used for undertaking an exploratory investigation of developing AI-assisted surrogate models, namely by augmenting low-accuracy models with a data-driven part.
The ultimate goal of this project is to develop a data-driven workflow able to learn the non-linear mappings between droplet trajectories and surface features in order to (i) enhance our insights about the morphology of surfaces and how droplets evolve on them, and (ii) help accelerate parametric studies aimed towards designing surfaces for controllable droplet transport.