Advancing the organosolv biorefinery. Part 2: Modeling vapor stripping-vapor permeation system behavior for the recovery of 1-butanol from aqueous solutions

The successful emergence and evolution of the biorefining industry relies on development and implementation of innovative and energy efficient unit operations developed at the laboratory scale, including the ability to model and predict behavior across a wide range of operating conditions. One such emerging technology is vapor stripping-vapor permeation (VSVP), a separation technique that combines vapor stripping and vapor permeation into a single unit operation to recover organic liquids (in this case butanol and other volatile components) from aqueous solutions. Key VSVP operating conditions including feed temperature, feed vapor flow rate, feed vapor composition, and feed liquid composition were used to model VSVP behavior based on Arrhenius-type flux equations. Using the Arrhenius-type model, flux activation energies for butanol and water were determined to be 91.7 and 114.0 kJ/mol, respectively, consistent with the organophilic nature of the PDMS membrane used for the VSVP experiments. A simple Arrhenius-type model was insufficient for fitting experimental VSVP butanol removal data due to its sole dependence on temperature. An expanded Arrhenius-type model was successfully developed to incorporate the impact of feed vapor flow rate and feed vapor composition, which resulted in a much more accurate fit for experimental VSVP data. A concentration regime-based model also gave a good fit for experimental VSVP data using two distinct linear regimes with a crossover concentration of 3.7%wt butanol. Both the concentration regimes and expanded Arrhenius-type models were found to successfully fit experimental VSVP data as corroborated by R 2 values > 0.99 and residual sum of squares values < 0.5. • Successful modeling of butanol recovery from aqueous solutions via VSVP • Flux activation energies: butanol < water • Simple Arrhenius-type model is insufficient for fitting experimental VSVP data • Expanded Arrhenius-type model gave a good fit to experimental VSVP data • Model based on concentration regimes gave a good fit to experimental VSVP data