Predicting Impact Outcomes and Maximum Spreading of Drop Impact on Heated Nanostructures Using Machine Learning

Area of Science:

• Fluid dynamics
• Materials science
• Machine learning

Background:

• Predicting droplet behavior on heated surfaces is crucial for industrial applications.
• Nanostructured surfaces offer unique thermal and fluidic properties.
• Existing models often lack comprehensive parameter integration.

Purpose of the Study:

• To develop data-driven machine learning models for predicting droplet impact outcomes and spreading on nanostructured surfaces.
• To identify critical parameters governing droplet-surface interactions.
• To construct a predictive phase diagram of droplet impact behaviors.

Main Methods:

• Utilized artificial neural network classification (ANNC) for phase diagram construction.
• Employed support vector regression (SVR) to model the maximum spreading factor.
• Incorporated key parameters: impact velocity, surface temperature, nanopillar packing fraction, and surface roughness.

Main Results:

• ANNC successfully generated a phase diagram encompassing all experimental impact behaviors.
• SVR accurately modeled the maximum spreading factor as a function of Weber number and surface temperature.
• Nanostructures were shown to influence deposition, rebound, and splashing regimes, introducing behaviors like central jetting.

Conclusions:

• Nanostructure parameters (packing fraction and roughness) critically affect droplet impact outcomes.
• Increased packing fraction promotes deposition and spreading.
• Increased roughness enhances heat transfer, promoting the Leidenfrost effect and splashing.