Intentional amendment of soil with charcoal (called biochar) is a promising new approach to sequester atmospheric carbon dioxide and increase soil fertility. However, the environmental properties of biochars can vary with production conditions, making it challenging to engineer biochars that are simultaneously optimized for carbon sequestration, nutrient storage, and water-holding capacity. For this reason, I have undertaken a systematic study to (a) determine the pyrolysis conditions that lead to biochars with desired chemical and physical properties, and (b) find how these properties affect the water-holding capacity and nutrient adsorption in biochar-soil mixtures. First, a library of biochars was produced in a custom-built pyrolysis reactor under precisely controlled conditions. The chemical and physical structures of the produced biochars were characterized with various analytical techniques including 13C NMR, XPS, EA and BET pore surface analysis. My results suggest that the chemical composition and pore structure of biochars are determined not just by the maximum heat treatment temperature, but also by several other factors that include the pyrolysis heating rate, treatment time at the maximum temperature and particle size. I also tested a new approach that combines thermogravimetric reactivity measurements, diffusion-reaction theory and structural models to achieve a better characterization of the complicated multi-scale pore structure of biochars. The structural models treat biochars as porous solids having micro- and macropores of different shapes and exhibiting widely ranging pore-size distributions. Simulations results are then compared to experimental data to identify the presence of ordered or random pore networks and test their size distributions and connectivity. I then developed a multi-solid one-dimensional model that can use experimentally determined biochar properties to predict their field performance in beds packed with soil/biochar mixtures. The model used a system of coupled partial differential equations to describe the dynamic adsorption/elution of ammonium nitrate, a model fertilizer, in columns packed with biochar/soil mixtures and perfused with aqueous solutions of the fertilizer. The PDE system was solved using orthogonal collocation on finite elements. My chromatographic model accounted for all the important processes occurring in this system, including external mass transfer between the fluid phase and the solid particles, as well as intraparticle diffusion and adsorption of the solute on the pore surface area of the sorbents. To our knowledge, this is the first chromatographic model that accounted explicitly for the presence of two solid phases with widely different pore structures and adsorption capacities. A systematic parametric study was carried out to determine the importance of each system parameter. The adsorption equilibrium parameters and the intraparticle effective diffusivity of ammonium had the most significant effect on environmental performance. To complete the theoretical analysis, I also developed a model to describe the saturation and drainage of water from the packed column. The model accounted for all the important processes occurring in this system: (a) water exchange between the interstitial pore region and two different smaller pore regions and (b) water flow inside the larger pore region and the two different smaller pore regions. The transient mass balances led to a system of partial differential equations that was solved using block centered finite difference.

La pyrolyse douce est un procédé prometteur et largement utilisé, mené à une température de 200 à 300 °C dans une atmosphère inerte afin de produire des matériaux durables (bois traité thermiquement) ou des combustibles solides (bois torréfié). Le but de cette étude est d’étudier les bois traités thermiquement dans un réacteur à l’échelle semi-industrielle pour une production durable de matériaux. Deux essences de bois européennes différentes, une essence de feuillus (peuplier, Populus nigra) et une essence de résineux (sapin, Abies pectinata), sont utilisées pour réaliser les expériences. La présente recherche est divisée en trois parties. Dans la première partie, le comportement thermique des planches de bois est étudié dans un réacteur à l’échelle semi-industrielle. Les expériences sont effectuées à 200-230 °C avec une vitesse de chauffe de 0.2 °C min-1 dans un environnement sous vide (200 hPa) pour intensifier la dégradation thermique. Quatre étapes différentes de dégradation thermique lors du traitement thermique du bois sont définies, en fonction de l'intensité de la perte de masse différentielle (DML). Les caractéristiques de dévolatilisation du bois traité sont évaluées à l'aide de l'indice de dévolatilisation (ID) basé sur les résultats de l'analyse immédiate. La corrélation de l'ID par rapport à la perte de masse des deux essences de bois est fortement caractérisée par une distribution linéaire, ce qui permet de fournir un outil simple et utile pour prédire la perte de masse du bois. Dans la seconde partie de l’étude, plusieurs analyses (spectroscopie infrarouge à transformée de Fourier, diffraction des rayons X, mesure du changement de couleur, teneur en humidité à l’équilibre et angle de contact) ont été réalisées. Les résultats obtenus démontrent clairement la dégradation thermique lors des réactions de déshydratation, de désacétylation, de dépolymérisation et de condensation au cours du traitement thermique. Les phénomènes de changement de couleur et de transformation hygroscopique observés sont illustrés et discutés en détail. La décarbonisation (DC), la déshydrogénation (DH) et la désoxygénation (DO) des bois traités sont également évaluées. Il s'avère que les trois indices peuvent être bien corrélés à la variation de couleur totale et à l'étendue de la réduction de l'hygroscopicité (HRE). Dans la dernière partie de l'étude, une modélisation cinétique du traitement thermique du bois est développée sur la base d’un schéma cinétique en deux étapes. La cinétique obtenue permet de prédire avec succès le rendement en solide de planches de bois lors du traitement dans un réacteur à l’échelle semi-industrielle. Dans le même temps, une prévision de la composition élémentaire est proposée. Celle-ci est basée sur les analyses élémentaires (ultimes) du bois non traité et du bois traité, ainsi que sur les rendements instantanés en solides. Les résultats indiquent que la prédiction des profils C, H et O est en bon accord avec les changements de composition attendus dans le matériau au cours du traitement. En résumé, les résultats obtenus et la cinétique établie sont propices à l’identification des mécanismes de dégradation thermique du bois et peuvent être utilisés pour le traitement thermique et la conception de réacteurs dans l'industrie afin de produire des matériaux bois adaptés à diverses applications. / Mild pyrolysis is a promising and widely applied process conducted at 200-300 °C in an inert condition to produce sustainable materials (i.e. heat treated wood) or solid fuel (i.e. torrefied wood). The aim of this study is to investigate the woods heat treated in a semi-industrial scale reactor for sustainable material production. Two different European wood species, a hardwood species (poplar, Populus nigra) and a softwood species (fir, Abies pectinata), are used to perform the experiments. The present research is divided into three parts. In the first part, the thermal behavior of wood boards is studied in a semi-industrial scale reactor. The experiments are carried out at 200-230 °C with a heating rate of 0.2 °C min-1 in a vacuum condition (200 hPa) to intensify the thermal degradation. Four different stages of thermal degradation during wood heat treatment are defined based on the intensity of differential mass loss (DML). The devolatilization characteristics of treated woods are evaluated by the devolatilization index (DI) based on the results of proximate analysis. The correlation of DI with respect to mass loss of the two wood species is strongly characterized by linear distribution, which is able to provide a simple tool to predict the mass loss of wood. In the second part of the study, a number of analyses, such as Fourier-transform infrared spectroscopy, X-ray diffraction, measurement of color change, equilibrium moisture content, and contact angle) are performed to evaluate the property changes of treated woods. The obtained results clearly demonstrate the thermal degradation through dehydration, deacetylation, depolymerization, and condensation reactions during the heat treatment. The observed phenomena of color change and hygroscopic transformation are illustrated and discussed in detail. The decarbonization, dehydrogenation, and deoxygenation of the treated woods are also evaluated. It is found that the three indexes can be well correlated to the total color difference and hygroscopicity reduction extent (HRE). In the last part of the study, the kinetic modeling of wood heat treatment is developed based on a two-step kinetic scheme. The obtained kinetics successfully predict dynamic solid yield of wood boards during the treatment in the semi-industrial reactor. Meanwhile, the prediction of elemental composition is also performed by a direct method based on the elemental analyses of untreated and treated woods at the end of the treatment, as well as the instantaneous solid yield. The results point out that the prediction of C, H, and O profiles are in good agreement with expected composition changes in the wood materials during treatment. In summary, the obtained results and established kinetics are conducive to recognizing the mechanisms of wood thermal degradation and can be used for heat treatment process and reactor design in industry to produce wood materials for various applications.

Prescribed (i.e., controlled) burning is a common practice used in many vegetation types in the world to accomplish a wide range of land management objectives including wildfire risk reduction, wildlife habitat improvement, forest regeneration, and land clearing. To properly apply controlled fire and reduce unwanted fire behavior, an improved understanding of fundamental processes related to combustion of live and dead vegetation is needed. Since the combustion process starts with pyrolysis, there is a need for more data and better models of pyrolysis of live and dead fuels. In this study, slow pyrolysis experiments were carried out in a pyrolyzer apparatus and a Thermogravimetric analyzer (TGA) under oxygen free environment in three groups of experiments. In the first group, the effects of temperature (400–800 °C), a slow heating rate (H.R.) (5–30 °C min−1), and carrier gas flow rate (50–350 ml min−1) on yields of tar and light gas obtained from pyrolysis of dead longleaf pine litter in the pyrolyzer apparatus were investigated to find the optimum condition which results in the maximum tar yield. In the second group of experiments, 14 plant species (live and dead) native to forests in the southern United States, were heated in the pyrolyzer apparatus at the optimum condition. A gas chromatograph equipped with a mass spectrometer (GC–MS) and a gas chromatograph equipped with a thermal conductivity detector (GC-TCD) were used to study the speciation of tar and light gases, respectively. In the third group of experiments, the slow pyrolysis experiments for all plant species (live and dead) were carried out in the TGA at 5 different heating rates ranged from 10 to 30 ℃ min-1 to study the kinetics of pyrolysis. The results showed that the highest tar yield was obtained at a temperature of 500 °C, heating rate of 30 °C min−1, and sweep gas flow rate of 100 ml min−1. In addition, the tar composition is dominated by oxygenated aromatic compounds consisting mainly of phenols. The light gas analysis showed that CO and CO2 were the dominant light gas species for all plant samples on a dry wt% basis, followed by CH4 and H2. The kinetics of pyrolysis was studied using one model-free method and three model-fitting methods. First, the model-free method of Kissinger-Akahira-Sunose (KAS) was used to calculate the rates of pyrolysis as a function of the extent of conversion. The results showed that different plant species had different rates at different conversions. Then, three model fitting methods were used to find the kinetic parameters to potentially provide a single rate for each plant species. The results showed that the simple one-step model did not fit the one-peak pyrolysis data as well as the distributed activation energy model (DAEM) model. The multiple-reaction DAEM model provided very good fits to the experimental data where multiple peaks were observed, even at different heating rates.