Author:

Keywords:

Lignocellulose;, Pervaporation;, Bio-ethanol;, Ethanol-water;, Techno-economics;, PDMS;, Production cost;, Fermentation;, By-products

Abstract:

This dissertation investigates the potential of the integration of pervaporation into the production process of lignocellulosic bio-ethanol. The aim is to assess the impact on an economic, as well as an ecological level, and to emphasize the sustainable character of lignocellulosic bio-ethanol production. The basis of this study is the overall production process from lignocellulosic feedstock to anhydrous ethanol. Based on a literature review, it was found that absolute production cost estimated in different studies varies between $234 and $1210 per m3 ethanol, although often the same processing methods and feedstock are assumed. The main key factors responsible for this discrepancy were pinpointed and a more accurate value of $651 per m3 was calculated as the current minimum ethanol selling price. Furthermore, it was concluded that lignocellulosic ethanol is already a cost-effective alternative to corn-ethanol and can become competitive with gasoline in the near future.Subsequently, the influence of the replacement of the first distillation column in the purification train by the pervaporation unit was studied. For predefined values of membrane parameters, it was found that this modification increased the ethanol production cost by ~2%. On the other hand, the ecological footprint decreased even more, due to a larger quantity of produced electricity. Membrane targets to become cost competitive with distillation were calculated and an ethanol production cost of $617 per m3 was obtained for the best available laboratory-made membrane. This low cost was mainly achieved due to a very high membrane flux, rather than a high selectivity factor.The pervaporation process itself was studied in detail and the influence of operation conditions on the membrane performance was investigated for two applications of hydrophobic pervaporation. It was found that an increased temperature had a positive effect on both the membrane flux and the selectivity. An increase in permeate pressure on the other hand, enhanced the selectivity, but decreased the membrane flux. Hence, this parameter must be optimized for the application of interest. Furthermore, it was observed that incorporation of inorganic fillers in the PDMS network increased both the selectivity and membrane flux.Ethanol-water pervaporation was compared to fermentation broth pervaporation, since several by-products can be present in a fermentation broth. No significant differences were found for unfilled PDMS membranes, whereas the water flux of zeolite-filled membranes was ~11% higher during fermentation broth pervaporation and the permeate ethanol concentration ~8% lower. Carboxylic acids were found to be responsible for this declined membrane performance, which rendered the membrane more hydrophilic due to the interaction with the zeolite surface. Membrane fouling could be avoided by increasing the pH to more neutral environments since dissociated ions hardly adsorb on zeolite particles.Finally, an extended study on the mass transportduring pervaporation wasmade. Different models for the prediction of an effective permeability were analyzed with finite element calculations for membranes and fillers with various geometries. Despite their promising theoretical approach, these models exhibited clear practical limitations, and actual performance predictions were highly inaccurate. The first sorption step during pervaporation is often described by the mass-based UNIQUAC model. A deeper study revealed that this model is erroneous and even leads to an incorrect description of simple vapor-liquid equilibrium of pure liquid mixtures. The reason for this is that a wrong conversion from molar to mass-based parameters is generally assumed in the literature, which should have been performed in a simple and straightforward way. Furthermore, an adaptation to the standard model was required for the description of sorption equilibrium in membranes. This model was successfully applied and even found to be very accurate for a highly non-ideal system.