Molecular dynamics simulation and machine learning prediction of tribological properties of graphene solid-liquid two-phase lubrication system

Abstract

The influence of solid-liquid interaction between graphene and base oil on lubrication performance and mechanism still lacks insight. This paper fully considered contact pressure, differences in base oil chain length, and surface roughness characteristics, the tribological properties and lubrication mechanism of graphene oil lubrication system under different lubrication regimes were revealed by molecular dynamics simulations, and frictional wear was predicted based on four machine learning algorithms. The results indicated that under fluid lubrication regimes, the increase in base oil chain length and contact pressure deteriorated lubrication performance, which was attributed to the long-chain base oils enhanced the interlayer sliding resistance at the solid-liquid interface and adsorbed on the friction pair surface to form a thicker solid-like film, while the high contact pressure induced an increase in lubricant viscosity and internal stress. Under boundary lubrication, increased chain length improved lubrication while contact pressure aggravated frictional wear. Attributed to long-chain base oils increased oil film thickness and strength, improved graphene bending wrinkles and lubricant fluidity, and facilitated lubrication state transitions. However, high contact pressure showed the opposite trend, resulting in worsening lubrication performance. Moreover, the ensemble learning algorithm exhibited optimal friction (R² =0.968) and wear (R² =0.951) prediction capabilities.