In this article, the performance of polyethersulfone (PES) ultra‐ and nanofiltration membranes, prepared with the non‐toxic solvent dimethyl sulfoxide (DMSO), was investigated. The membranes were prepared by immersion precipitation via phase inversion. Experimental results proved that DMSO is a better alternative to N‐methyl‐2‐pyrrolidone (NMP) as solvent for PES ultrafiltration membranes as the membranes had a higher permeability and rejection of bovine serum albumin (BSA). An explanation was found based on experimental cloud point data and scanning electron microscopy images showing the morphology. The rejection of BSA and rose Bengal (RB) was proportional to the polymer concentration. On the contrary, the permeability decreased with increasing polymer concentration. For a casting thickness of 250 µm, an optimal balance between permeability and rejection of macromolecules for ultrafiltration was found at 24 wt % PES. The permeability was inversely proportional to the casting thickness, but a small decrease in rejection was observed when lowering the thickness. A good balance between permeability and rejection of RB was found, using a reference nanofiltration membrane of 28.5 wt % PES with 150 µm casting thickness. This membrane achieved a RB rejection of 95.3% and a pure water flux of 2.03 L m−2 h−1 bar−1. The membrane thickness and polymer concentration did not have a clear influence on the hydrophilicity of the membranes. It can be concluded that DMSO is a benign alternative as compared to traditional solvents such as NMP and also results in better PES membrane performances. DMSO is a perfectly suitable solvent for ultrafiltration applications and has potential to be used for nanofiltration applications. © 2018 Wiley Periodicals, Inc. J. Appl. Polym. Sci. 2018, 135, 46494.

A molecular-level kinetic model for the hydroprocessing of methyl laurate was constructed. The reaction network was deduced using experimental observations in the context of the delplot method for the discernment of product rank. The resulting 45 species and 83 reactions were used to construct the set of material balances in the kinetic model. Kinetic parameters of the model were determined by minimizing the difference between model outputs and experimental data for methyl laurate hydroprocessing. Differences in reactivity as a result of catalyst metal composition were modeled via the catalyst family concept. The model results show good agreement with the experimental results for a range of process conditions.

A molecular-level kinetic model was constructed for the hydroisomerization/hydrocracking of a hydrotreated deasphalted oil feed. A kinetic network was developed for the lube base oil production containing 1105 molecular species and 15 991 reactions grouped into nine reaction families. The molecular composition of the feedstock was reconstructed by minimizing the difference between experimental data and simulated mixture properties. More specifically, hydrocarbon types, carbon number distribution, and sulfur level were matched very accurately. To model the kinetics and reduce the computational load, the Langmuir–Hinshelwood–Hougen–Watson rate law parameters were constrained using the linear free-energy relationship principles. Using the experimental process and product data of hydroisomerization over a commercial catalyst, the reaction kinetics were optimized on the basis of 141 data points. Excellent agreement was found between experimental and simulated properties of the lube base product of hydroprocessing at three different temperatures.