Viscosity of Multicomponent Organic Mixtures

This dataset contains 44,316 experimentally measured viscosity values for 100 aqueous and organic solutions, ranging in complexity from 2 to 17 components and spanning −20 °C to 35 °C. Each entry includes solution composition (mole and mass fractions), temperature, and measured viscosity (cP). The dataset was generated using a calibrated ViscoLab 3000 viscometer coupled with a Fluke circulating bath, with detailed experimental protocols provided in the associated manuscript. The data were used to benchmark classical viscosity models (Katti–Chaudhuri, Augmented Adam–Gibbs) and to train a physics-informed neural network (PINN) that enables accurate viscosity prediction across arbitrarily complex mixtures. This dataset supports the manuscript: Physics-Informed Neural Networks for Predicting Viscosity of Many-component Organic and Aqueous Solutions (submitted to Chemical Engineering Journal).