Simulation of the fractionation of biomass by hydrothermal processes with sub and supercritical water

Resumen

Nowadays, biomass revalorisation is one of the most studied fields since it means the production of chemicals and energy from a renewable source. Thus, several and different processes have been developed. Among all of them, there are three especially remarkable since they just require a “green” fluid, such as water or carbon dioxide, and heat to do this biomass upgrading. These processes are: supercritical fluid extraction (SFE), pyrolysis-gasification and hydrothermal fractionation (HTF). For this reason, this thesis is focused on the modelling and simulation of these three processes. The general idea was to propose overall models, based on pseudo-compounds, to reproduce the experimental behaviour observed in the laboratory. The specific aims for each process were: (1) a new model for SFE, easily understandable, based on the Broken Intact Cells model, (2) an overall model for pyrolysis (based on the Waterloo’s mechanism) to estimate the initial sample composition and to analyse the structure role on thermal degradation and (3), a realistic model for the HTF of biomass in a packed bed reactor. Regarding SFE, its fundamentals have been deeply studied and a great deal of models can be found in literature. However, they generally require high knowledge about mass transfer. A knowledge that is not shared by the whole potential users of this technology. Thus, an innovative approach, solved by a free user-friendly Excel interface, was proposed in this thesis. This novel model involved non-stationary mass balances for the extracted compounds in both phases (solid and supercritical fluid) and it was based on the extraction characteristic times of the initial sample, which makes it far easier to understand the mass transfer. The validation was done by experimental data of samples where the mass transfer limitation was different: sesame seeds (controlled by external mass transfer and equilibrium) and coffee beans (controlled by internal diffusion). The equilibrium between the solid and liquid phases was simulated by a Henry’s linear relation and the average error of the model was lower than 11%. Similarly to SFE, biomass pyrolysis (or gasification if an oxidant is used) is a well known process and a huge number of studies about its modelling have been previously done. But, again, they need a deep knowledge about mass transfer and kinetics. Additionally, biomass pyrolysis is involved in one of the most used techniques to characterise a solid material, the thermogravimetric analysis (TGA). For these reasons, a kinetic model to simulate the slow pyrolysis of biomass during a TGA has been developed. This model was solved by another free user-friendly Excel interface and it could be used to easily analyse the effect of the heating rates, operational mode (isothermal or not), pre-treatments, atmosphere type (oxidant or not), raw material composition and biomass structure on the thermal behaviour of the sample. The reaction pathway was a modification of the Waterloo’s mechanism: any solid can produce gases and char, a char that can further degrade into more gases. Furthermore, this mechanism was completed by the vaporization of any liquid phase in the initial raw material. Auto-catalytic kinetics were used since the cleavage of the biomass produce oligomers that accelerate the further depolymerisation. The validation of the proposed model was done with data from completely different samples (pure hemicellulose, cellulose and lignin, woody biomass and winery residues) and the average error was always below 7%. Finally, HTF of biomass has been also previously studied. Nevertheless, a comprehensive physic-chemical model has not been utterly developed. Thus, a kinetic realistic model was proposed in this thesis. A model that also incorporates a novel reaction pathway for biomass fractionation, including the main processes involved. These processes are: biomass cleavage, biomass solubilisation, sugar production, pH changes and the mass transfer between both phases, the liquid and the water. With regards to the kinetics, an autocatalytic expression was again used to simulate the biomass depolymerisation. The model was validated by data about the hydrothermal fractionation of holm oak in a packed reactor at several temperatures around 180 ºC (to boost hemicellulose extraction). Additionally, different particle diameters and volumetric flows were also considered. However, the main variable was always the operational temperature. The fittings were focused on carbon content profiles, being the average error around 33 %. On the other hand, a populational model was also developed for the HTF of biomass. The aim was to complete the kinetic model by a more realistic simulation of the cleavage (up to 200 oligomers were simulated). Furthermore, sugar degradation and repolymerisation reactions were also included in this case since the data obtained at temperatures as high as 280 ºC. In summary, novel models for the simulation of the biomass pyrolysis, biomass extraction by supercritical fluids and biomass hydrothermal have been proposed and validated in this thesis. These models were able to accurately reproduce the experimental behaviour and they constitute a useful tool for a better understanding of the considered processes. Additionally, since they were solved by Excel-interface tools, they can be directly used by a wide range of people or being connected to commercial process simulators, like Aspen plus