Probing Rate-Dependent Liquid Shear Viscosity Using Combined Machine Learning and Nonequilibrium Molecular Dynamics.

Area of Science:

• Computational physics
• Materials science
• Rheology

Background:

• Measuring liquid dynamic viscosity at high shear rates is experimentally challenging.
• Controlling thermal effects and resolving high shear rates are key limitations.
• Understanding shear-thinning behavior is crucial for complex fluid dynamics.

Purpose of the Study:

• To develop a robust method for accurate viscosity prediction across shear rates.
• To integrate machine learning with nonequilibrium molecular dynamics (NEMD) simulations.
• To investigate the interplay of shear rate, pressure, and temperature on viscosity.

Main Methods:

• Developed a supervised artificial neural network (ANN) model for viscosity prediction.
• Utilized nonequilibrium molecular dynamics (NEMD) simulations with LAMMPS.
• Implemented 'fix npt/sllod' for precise constant-pressure control in simulations.

Main Results:

• The ANN model accurately predicts viscosity as a function of shear rate, pressure, and temperature.
• Observed distinct shear-thinning trends and nonmonotonic changes in molecular morphology.
• Demonstrated that temperature effects on viscosity diminish at high shear rates.

Conclusions:

• ML-enhanced NEMD provides an efficient and accurate framework for viscosity prediction.
• The study offers insights into molecular behavior under shear stress.
• This approach facilitates future research in complex fluid dynamics and material design.