Existing viscosity prediction methods and relevant literature are reviewed. An exhaustive review of group contribution, corresponding states, and interpolative prediction methods finds that even the best of these models produces large prediction errors and often require significant experimental data. Molecular dynamics simulation techniques for viscosity prediction are evaluated and compared to one another to determine the best choice for this work. A thorough investigation finds that Equilibrium Molecular Dynamics (EMD) simulations are the best option for reproducible and reliable liquid viscosity predictions. The many tuning parameters available in molecular dynamics simulations are investigated for their effects on prediction uncertainty and accuracy. Challenges associated with molecular dynamics predictions are discussed and a rigorous simulation and data analysis methodology is developed which addresses these issues. The TLVMie force field is developed to describe linear alkanes, branched alkanes, alkylbenzenes, and cycloalkanes. The method is developed with a "training set" of compounds and the transferability is evaluated with a completely different "test set" of compounds. Predictive capability with mixture viscosities is evaluated next without any re-fitting of the parameters determined from pure-component data. The TLVMie potential is shown to be significantly more accurate for both pure-components and mixtures, more reliable for compounds that differ greatly from the training set, and predictions are made without the experimental data requirements of other methods.
| Identifer | oai:union.ndltd.org:BGMYU2/oai:scholarsarchive.byu.edu:etd-11265 |
| Date | 14 February 2023 |
| Creators | Carlson, Daniel J. |
| Publisher | BYU ScholarsArchive |
| Source Sets | Brigham Young University |
| Detected Language | English |
| Type | text |
| Format | application/pdf |
| Source | Theses and Dissertations |
| Rights | https://lib.byu.edu/about/copyright/ |
Page generated in 0.0023 seconds