季戊四醇
外推法
材料科学
粘度
热力学
润滑油
联轴节(管道)
润滑
分子动力学
高斯分布
生物系统
分子描述符
表征(材料科学)
分子
化学物理
分子模型
聚合物
实验设计
统计物理学
作者
Z X Sun,Pengpeng Bai,Hui Cao,Kai Ma,W P Chen,Zhonghuan Xiang,Yonggang Meng,Yu Tian
摘要
Synthetic pentaerythritol esters (PEEs) are widely used in advanced lubrication systems, but rational molecular design remains limited by structural complexity and the lack of quantitative structure–viscosity relationships under coupled temperature–pressure conditions. Here, a data‐driven molecular design framework is developed to correlate PEE molecular architecture with viscosity and its temperature‐ and pressure‐dependent evolution. By amplifying structure‐specific viscosity responses under elevated temperature and pressure, a structurally resolved dataset of 24 designed PEEs is constructed, covering short‐ and long‐chain components with controlled branching. Viscosity measurements span 25–100°C and pressures up to 1 GPa, enabling characterization under practical and extreme conditions. A Gaussian process regression (GPR) model integrating molecular descriptors and thermodynamic variables accurately predicts viscosity, pressure–viscosity coefficients, and temperature–viscosity coefficients in the Newtonian regime, while providing uncertainty estimates and supporting extrapolation toward extreme conditions. The learned structure–property mapping further enables forward molecular design, experimentally validated by a newly designed PEE lubricant, achieving R 2 > 0.98 and relative errors below 5%. This strategy transforms lubricant development from empirical trial‐and‐error into requirement‐oriented molecular engineering, accelerating materials discovery and formulation optimization for high‐performance lubrication systems.
科研通智能强力驱动
Strongly Powered by AbleSci AI