Torefikacija yra perspektyvi biomasės apdirbimo technologija, leidžianti pagerinti biomasės fizikines savybes, kartu sumažinti išlaidas skirtas biomasės smulkinimui. Torefikuota biomasė gali būti panaudojama esamose anglimi kūrenamose elektrinėse, ją santykine dalimi maišant su įprastomis anglimis, taip sumažinant CO2 kiekį išsiskiriantį deginant iškastinį kurą (Kioto protokolas). Šiuo metu trūksta mokslinių tyrimų ir duomenų apie torefikacijos panaudojimą agrokultūrų atliekoms (šiaudams) apdirbti, pagaminant naują energetiškai patrauklų kurą pasižymintį unikaliomis savybėmis. Pagrindinis šio darbo tikslas buvo ištirti ir įvertinti iš skirtingų biomasės rūšių pagaminto torefikuoto kuro savybes ir torefikacijos proceso sąlygas. Tam būtina sąlyga: bandomojo torefikacijos reaktoriaus sukūrimas. Sauso kuro aukštutinė degimo šiluma nustatymo (HHV), anglies, vandenilio, azoto, sieros, chloro, lakiųjų organinių junginių kiekio, peleningumo ir pelenų lydumo, masės išeigos, energijos našumo, energetinio tankumo, reaktoriaus patikimumo lyginant su TGA, lakiųjų organinių junginių sudėties ir hidrofobiškumo testai – tai pagrindiniai parametrai kurie buvo nustatinėjami atliekant šį darbą. Torefikacijos procesas buvo atliekamas azotinėje aplinkoje, prie skirtingų temperatūrų (250 °C, 280 °C, 300 °C), bandinį reaktoriuje išlaikant 30 minučių, plius papildomas išlaikymas 10 minučių naudojant 300 °C. Gauti rezultatai leidžia teigti, kad po torefikacijos proceso medžiaga turi didesnę sausojo... [toliau žr. visą tekstą] / Torrefaction is a promising fuel pre-treatment technology for biomass, as it improves the physical characteristics, reduces the energy consumption for grinding, improves the co-firing process due to more stable characteristics of this fuel. However, there is still lack of data on torrefaction of agricultural waste which have more unequal composition. The aim of this work was to create fixed bed reactor for torrefaction process and investigate the dependence of properties of formed torrefied products on various biomass materials and process conditions. HHV, amounts of C, H, N, S, Cl, VOC, ash content and ash melting behavior, mass and energy yields, energy density, reactor reliability comparing with TGA results and composition of VOC analysed with TGA-GC/MS, hydrophobicity test were studied as the main factors for comparison. Torrefaction was carried out in the nitrogen environment at variuos temperatures (250°C, 280°C, 300°C) for 30 min and for 10 min at 300 °C too. HHV of woody and agricultural waste after torefaction was increased as energy density too (which allows cheaper logistics), also was found that torrefied material has better hydrophobic properties, biodegradation slows down, material is easy to grind. Determining energy yield for each type of biofuel, assess to find most suitable conditions for torrefied biofuel production. It was found that torrefaction solves one of the major straw as biofuel problems: it’s large amount of sulfur and chlorine levels. Torrefied... [to full text]

Torrefaction and densification of biomass can provide an important piece in the puzzle of phasing out fossil fuels in favor of renewable alternatives. This new energy carrier shares many of the advantages with fossil coal in terms of energy density, hydrophobicity and burner feeding but is carbon neutral and renewable. It also lacks the challenges of many other renewable alternatives, especially irregular availability. A model was developed in Excel as sales support for BioEndev, one of the leading actors in the process of taking torrefaction to a commercial market, assessing the black pellet supply chain from feedstock to end user and comparing it to white pellets. Data was obtained from literature, industry and BioEndev. The model can be used for different parameters for price of feedstock, capital and operating expenditures, transport and handling costs and analyze 28 different cases. It also includes simplified calculations for energy input and greenhouse gas emissions. A case study for two different supply chains was performed with the model. One assessed a production facility in northern Sweden with distribution to a consumer in Denmark. The other a torrefaction plant in southeastern USA with distribution to a consumer in the Netherlands. The cost for delivering black and white pellets from Sweden to Denmark was found to be 33.0 €/MWh and 35.3 €/MWh respectively. For the case of delivering from USA to the Netherlands, the total supply chain cost was 27.6 €/MWh for white pellets and 24.7 €/MWh for black pellets. Suggestions for further work are to 1) develop the model outside this study’s limitations, for example by adding integration options for the torrefaction facility or by different end user configurations, and 2) expand the scope to also comparing black pellets to coal to see how big the gap is and which political incentives that could shrink this gap.

Nowadays, there is an increasing awareness on the importance of biomass waste as a renewable source of energy, materials and chemicals. In this context, the European project MOBILE FLIP aims at developing and demonstrating mobile conversion processes suitable with variousunderexploited agro- and forest based biomass resources in order to produce energy carriers, materials and chemicals. One of these processes is torrefaction, which consists in a mild thermal treatment, occurring typically between 200 and 300°C during a few tens of minutes in a defaultoxygen atmosphere. The solid product obtained has thermal and processing properties closer to coal, and thus is suitable as fuel for combustion or gasification. During torrefaction, condensable coproducts are released, that may also be source of “green” chemicals. It is therefore crucial to characterize them to optimize the torrefaction process and design industrial units. Up to now, only few works have been focused on characterizing and modelling both solid and condensable species during torrefaction versus operating conditions and feedstock type. Furthermore, these studies typically include a reduced number of biomasses. Cellulose, hemicellulose and lignin,which constitute biomass macromolecular composition, are determining properties to predict biomass behaviour during torrefaction. However, torrefaction tests on these constituents are rare and always based on commercial compounds, which were proved as little representative of the native biomass. The objective of this study is to analyse the influence of biomass characteristics, mainly represented by the macromolecular composition in cellulose, hemicellulose and lignin, on the global behaviour of biomass in torrefaction, both in terms of solid mass loss and of productionprofiles of the volatile species released, in function of the operating conditions.14 biomasses from the main biomass families (deciduouswood, coniferous wood, agricultural byproductsand herbaceous crops) were selected for this study. An optimized extraction procedure was proposed to recover cellulose, hemicellulose and lignin fractions from 5 reference biomasses. Experiments were performed on a thermogravimetric analyzer coupled to a gas chromatography mass spectrometer device through a heated storage loop system (TGA-GC/MS). Solid degradation kinetics and volatile release profiles were followed during torrefaction experiments combining non-isothermal (200 to 300°C at 3°C/min) and isothermal (300°C, 30 min) conditions, ensuring the chemical regime thanks to the appropriate operating conditions. The results obtained with the raw materials demonstrated that biomass macromolecular composition is a main factor influencing biomass behavior in torrefaction. Consequently, the heterogeneity of the resource results in a diverse behavior in torrefaction, particularly in the case of agricultural biomasses. The results with the extracted components evidenced their very different behavior compared to thecommercial compounds, particularly in the case of cellulose. This suggests that a limitation could be induced by the common use in literature of commercial components for torrefaction modelling. The impact on the characterization of macromolecular components was also shown to be prevailing in their behavior in torrefaction, especially in the case of hemicellulose sugar composition and cellulose crystallinity. Furthermore, differences in release kinetics of volatile species during torrefaction were observed, even for volatiles belonging to the same chemical family (acids, furans, ketones). Derived from these results, a torrefaction model based on the additive contribution of extracted cellulose, hemicelluloses and lignin to the global behavior of biomass in torrefaction was proposed, and this for the 5 representative biomasses.

Le traitement thermique du bois est un procédé connu et étudié depuis plusieurs décennies. Il confère au matériau une meilleure stabilité dimensionnelle et une meilleure durabilité au détriment de ses propriétés mécaniques, notamment de sa résilience. En dépit de plusieurs travaux disponibles, il reste difficile d'optimiser le gain sur les propriétés recherchées et de mettre en vis-à-vis les pertes sur les qualités que l'on voudrait préserver. Ce constat nous a conduit à mener des études fondamentales pour comprendre les mécanismes mis en jeux lors du traitement thermique. Les possibilités offertes en matière de simulation numérique nous permettent de proposer une contribution innovante au problème d'homogénéité du traitement d'une pièce de bois. Elle s'appuiera d'une part sur l'adaptation d'un code de simulation développé pour le séchage du bois en y intégrant les cinétiques chimiques, d'autre part sur l'étude expérimentale du traitement thermique. Parallèlement, nous avons cherché à caractériser ces échantillons de bois massifs soumis à différents traitements thermiques par spectrométrie en réflexion diffuse dans le proche infrarouge (SPIR). <br />Les résultats montrent que le modèle de pyrolyse couplé au modèle de transport rend compte des différents événements caractéristiques se déroulant durant le traitement thermique, c'est-à-dire la présence de réactions exothermiques et les surpressions internes générées par les gaz produits. La méthode par analyse spectrale a révélé qu'il est possible d'une part de discriminer des échantillons de bois ayant subis différents traitements et d'autre part de retracer l'historique thermique d'une pièce de bois dans son épaisseur. La SPIR semble ainsi une technique prometteuse qui devrait permettre de valider les profils de dégradation simulés par le code de calcul. Elle devait offrir également des perspectives intéressantes en matière de contrôle qualité des bois traités à haute température pour le couplage des propriétés physiques et mécaniques du nouveau matériau à sa composition chimique.

Distributed mobile conversion facilities using either fast pyrolysis or torrefaction processes can be used to convert forest residues to more energy dense substances (bio-oil, bio-slurry or torrefied wood) that can be transported as feedstock for bio-fuel facilities. All feedstock are suited for gasification, which produces syngas that can be used to synthesise petrol or diesel via Fischer-Tropsch reactions, or produce hydrogen via water gas shift reactions. Alternatively, the bio-oil product of fast pyrolysis may be upgraded to produce petrol and diesel, or can undergo steam reformation to produce hydrogen. Implementing a network of mobile facilities reduces the energy content of forest residues delivered to a bio-fuel facility as mobile facilities use a fraction of the biomass energy content to meet thermal or electrical demands. The total energy delivered by bio-oil, bio-slurry and torrefied wood is 45%, 65% and 87% of the initial forest residue energy content, respectively. However, implementing mobile facilities is economically feasible when large transport distances are required. For an annual harvest of 1.717 million m3 (equivalent to 2000 ODTPD), transport costs are reduced to less than 40% of the total levelised delivered feedstock cost when mobile facilities are implemented; transport costs account for up to 80% of feedstock costs for conventional woodchip delivery. Torrefaction provides the lowest cost pathway of delivering a forest residue resource when using mobile facilities. Cost savings occur against woodchip delivery for annual forest residue harvests above 2.25 million m3 or when transport distances greater than 250 km are required. Important parameters that influence levelised delivered costs of feedstock are transport distances (forest residue spatial density), haul cost factors, thermal and electrical demands of mobile facilities, and initial moisture content of forest residues. Relocating mobile facilities can be optimised for lowest cost delivery as transport distances of raw biomass are reduced. The overall cost of bio-fuel production is determined by the feedstock delivery pathway and also the bio-fuel production process employed. Results show that the minimum cost of petrol and diesel production is 0.86 $ litre-1 when a bio-oil feedstock is upgraded. This corresponds to a 2750 TPD upgrading facility requiring an annual harvest of 4.30 million m3. The minimum cost of hydrogen production is 2.92 $ kg-1, via the gasification of a woodchip feedstock and subsequent water gas shift reactions. This corresponds to a 1100 ODTPD facility and requires an annual harvest of 947,000 m3. The levelised cost of bio-fuel strongly depends on the size of annual harvest required for bio-fuel facilities. There are optimal harvest volumes (bio-fuel facility sizes) for each bio-fuel production route, which yield minimum bio-fuel production costs. These occur as the benefits of economies of scale for larger bio-fuel facilities compete against increasing transport costs for larger harvests. Optimal harvest volumes are larger for bio-fuel production routes that use feedstock sourced from mobile facilities, as mobile facilities reduce total transport requirements. / Graduate / 0791 / drbrown@uvic.ca

La torréfaction est une étape nécessaire pour la production de gazoles à partir de biomasse lignocellulosique par voie thermochimique (chaîne Biomass To Liquid). Il s'agit d'un traitement thermique dans le domaine de température compris entre 200 et 300°C en milieu non oxydant ; le but de cette étape est de modifier la structure de la biomasse afin d'en faciliter le transport pneumatique après broyage. Cependant, des réactions exothermiques ont été observées et peuvent mener à un mauvais contrôle de la température au sein du réacteur et nuire à la qualité des produits, voire endommager l'installation. L'objectif de cette thèse est de quantifier la chaleur émise par les réactions exothermiques de torréfaction de la biomasse lignocellulosique, et d'en étudier les impactes lors du changement d'échelle, où les phénomènes de transfert de masse et de chaleur ne sont plus négligeables. Durant nos travaux, des mesures de perte de masse et de flux de chaleur ont été réalisées à l'échelle de la poudre (microparticule) sur trois types d'essence de bois (robinier, épicéa et eucalyptus) ainsi que sur les principaux constituants de la matière lignocellulosique (cellulose, xylane, glucomannane et lignine). Un modèle cinétique capable de reproduire la perte de masse ainsi que le flux de chaleur généré par les réactions exothermiques, a été développé. Il utilise le concept de distribution d'énergie d'activation. Tous les paramètres du modèle ont été identifiés par méthode inverse sur un ensemble de tests isothermes d'une durée de 10 heures. Cela permet de proposer des paramètres cinétiques robustes et des valeurs fiables d'énergie d'activation. Par la suite, des mesures de température pendant des essais de torréfaction sur des planches de bois (méso-échelle) et sur un lit fixe de particules (échelle macroscopique) ont permis de mesurer la propagation d'une onde thermique générée par les réactions exothermiques. Une modélisation macroscopique qui intègre le modèle cinétique développé permet de propager l'effet des réactions exothermiques à l'échelle de la macro particule. L'analyse de l'ensemble des résultats permet de mettre en exergue l'importance de l'échelle lit sur l'emballement thermique observé expérimentalement. L'ensemble du travail, mené à différentes échelles spatiales et complété par une analyse permettant de relier ces échelles entre-elles, constitue une avancée significative vers la prédiction de l'exothermicité de la torréfaction afin d'assurer la sécurité et la faisabilité à l'échelle industrielle. / Lignocellulosic biomass torrefaction is an important step for diesel production through the BTL (Biomass To Liquid) chain. Torrefaction is a non-oxidative thermal treatment in the temperature range from 200 to 300°C. The aim of this process is to modify biomass structure in order to facilitate pneumatic transportation after grinding. However, some exothermic reactions are triggered in this temperature range which can lead to a lack of temperature control inside the reactor with detrimental effects on the product quality or even destructive effects on the facility. The purpose of this study is to contribute to the development of a multi-scale model for simulating these thermal runaway phenomena. Starting with the smallest scale, anhydrous weight loss combined with heat flux measurement from powders have been performed in a TGA-DSC. A data base was developed from three types of woody biomass, namely locust, spruce and an eucalyptus and from the main lignocellulsic components which are cellulose, xylan, glucomannan and lignin. A thermo-kinetic model able to reproduce the measured mass loss and heat flux during torrefaction has been developed by using the Distributed Activation Energy Method. Kinetic parameters and reaction enthalpies have been identified by inverse method taking into account the comprehensive set of data over several isothermal conditions with residence times of up to ten hours. Proceeding to larger scales, temperature measurements under torrefaction conditions have been performed separately in individual macro-particles and in a large scale packed bed of wood chips in order to test models at these scales. Thermal excursions were observed both within the particles and the bed due to the exothermic reactions. In the fixed bed an actual amplifying thermal wave was observed to propagate along the axial direction have been performed. A macroscopic heat and mass transfer model coupled with the kinetic model developed in this work allowed to simulate the temperature field at the macro-particle scale. Further model developmental work is needed to simulate the bed scale, Experimental observations and modelling carried out in this work represent an important improvement for the prediction of the heat released by torrefaction reactions in order to make this thermal pre-treatment safer and economically valuable.

In the transition to a sustainable energy supply there is an increasing need to use biomass for replacement of fossil fuel. A key challenge is to utilize biomass conversion technologies in an environmentally sound manner. Important aspects are to minimize potential formation of persistent organic pollutants (POPs) such as dioxins and dioxin-like compounds. This thesis involves studies of formation characteristics of polychlorinated dibenzo-p-dioxins (PCDDs), dibenzofurans (PCDFs) and naphthalenes (PCNs) in microwave-assisted pyrolysis (MAP) and torrefaction using biomass as feedstock. The research focuses are on their levels, distributions, fingerprints (homologue profiles and isomer patterns) and the underlying formation pathways. The study also included efforts to optimize methods for extracting chlorinated aromatic compounds from thermally treated biomass. The overall objective was to contribute better understanding on the formation of dioxins and dioxin-like compounds in low temperature thermal processes. The main findings include the following: Pressurized liquid extraction (PLE) is applicable for simultaneous extraction of PCDDs, PCDFs, PCNs, polychlorinated phenols and benzenes from thermally treated wood. The choice of solvent for PLE is critical, and the extraction efficiency depends on the degrees of biomass carbonization. In MAP experiments PCDDs, PCDFs and PCNs were predominantly found in pyrolysis oils, while in torrefaction experiments they were mainly retained in solid chars with minor fractions in volatiles. In both cases, highly chlorinated congeners with low volatility tended to retain on particles whereas the less chlorinated congeners tended to volatize into the gas phase. Isomer patterns of PCDDs, PCDFs and PCNs generated in MAP were more selective than those reported in combustion processes. The presence of isomers with low thermodynamic stability suggests that the pathway of POPs formation in MAP may be governed not only by thermodynamic stabilities but also by kinetic factors. Formation of PCDDs, PCDFs and PCNs depends not only on the chlorine contents in biomass but also the presence of metal catalysts and organic/metal-based preservatives. Overall, the results provide information on the formation characteristics of PCDDs, PCDFs and PCNs in MAP and torrefaction. The obtained knowledge is useful regarding management and utilization of thermally treated biomass with minimum environmental impact.

Incineration with energy recovery is the main disposal strategy for waste that cannot be reused or recycled, and a well-established source of energy in Europe, especially in Sweden where 2.2 Mtonnes of waste including domestic and imported municipal solid waste (MSW) and waste wood (WW) were combusted during 2015. However, owing to its inherent heterogeneous composition, inclusion of such waste in Waste-to-energy (WtE) technologies is challenging. These heterogeneities may lead to operationally-related issues in the WtE facilities and contribute to toxic emissions, which can be reduced by waste pre-treatment technologies.    This thesis examines the variations in the composition of MSW and WW streams used as a fuel supply in WtE facilities after undergoing waste pre-treatment technologies, and the effect of fuel composition on its combustion behaviour and formation of persistent organic pollutants (POPs) such as polychlorinated dibenzo-p-dioxins (PCDDs) and polychlorinated dibenzofurans (PCDFs). The overall objective is to contribute to a more thorough understanding of the selection of waste pre-treatment technologies to mitigate harmful emissions into the atmosphere when waste fuels are combusted in WtE facilities.    This thesis describes the high variability of contaminants in domestic and imported WW and suggests adaptation of WW pre-treatment techniques to produce fuels with a low potential for generating pollutants. A comparison of mechanical solid waste pre-treatments revealed that screening and shredding is more efficient than extrusion for reducing emissions of pollutants such as PCDDs and PCDFs in combustion. The evaluation of the combustion behaviour of MSW-based fuels showed a three-stage oxidative decomposition, and an acceleration of the decomposition of the MSW compared to the lignocellulosic materials, which may be attributed to the presence of food waste and plastics in the MSW. Combustion tests of fuel blends containing WW and MSW-based fuels with different food waste content suggested that WW, not food waste content, is the key factor for the formation of PCDDs, PCDFs, and polychlorinated biphenyls (PCB), benzenes (PCBzs) and phenols (PCPhs). Torrefaction may be a suitable technology for improving the properties of waste as a fuel e.g. due to its low PCDD and PCDF emissions. / Förbränning med energiåtervinning är det huvudsakliga sättet att ta hand om avfall som inte kan återanvändas eller återvinnas. Det är en väletablerad energikälla i Europa och särskilt i Sverige där 2,2 miljoner ton avfall, däribland inhemskt och importerat hushållsavfall och returträ, förbrändes under 2015. På grund av den heterogena sammansättningen hos hushållsavfall och returträ är förbränning av dessa material i anläggningar med energiåtervinning (så kallade WtE-anläggningar) förknippade med en del driftsrelaterade utmaningar. Det kan även ge upphov till miljöfarliga utsläpp, som dock kan reduceras genom förbehandling av avfallet. I denna avhandling har variationer i sammansättningen hos hushållsavfall och returträ som förbränns i WtE-anläggningar undersökts. Effekten av bränslemixens sammansättning och ev förbehandling på bränslets förbränningsegenskaper samt bildning av långlivade organiska föroreningar (så kallade POPar) såsom polyklorerade dibenso-p-dioxiner och polyklorerade dibensofuraner vid förbränning har utvärderats. Det övergripande målet är att bidra till en djupare förståelse av hur valet av förbehandlingsteknik för avfall kan bidra till att minska skadliga utsläpp till luft när avfallsbränslen förbränns i WtE-anläggningar. Denna avhandling beskriver den stora variabiliteten av metall- och materialföroreningar i inhemskt och importerat returträ och föreslår förbehandlingstekniker för att producera bränslen med låg potential att generera föroreningar. En jämförelse av mekaniska förbehandlingstekniker visade att mekanisk sönderdelning och separering (krossning och siktning) är mer effektivt än s.k. högtrycks-pressning för att minska utsläppen av föroreningar som dioxiner och furaner vid förbränning. Utvärderingen av bränslemixar innehållande hushållsavfall uppvisade en oxidativ nedbrytning i tre steg vid förbränning, och en accelererad nedbrytning av avfallsmaterialet jämfört med vedmaterialet i bränslet, troligen som effekt av innehållet av matavfall och plast i hushållsavfallet. Förbränningsförsök med bränsleblandningar av returträ och hushållsavfall med olika innehåll av matavfall visade att mängden returträ, och inte mängden matavfall, är den viktigaste faktorn för bildning av dioxiner, furaner, klorbifenyler, klorbensener, och klorfenoler. Torrefiering kan vara en lämplig teknik för att förbättra avfallets bränsleegenskaper, t.ex. på grund av dess låga emissioner.

The global problems of climate change, by the emissions of CO2 have over the past decenniums, led to a development of new innovations of renewable energy technologies, with the goal to phase out fossil fuels such as coal and oil. Many forms of renewable energy have already solved part of the energy consumption problems, but there are still large energy intensive industries that rely heavily on fossil fuels. One possible renewable product that could phase out fossil fuels in these industries is the black pellet, which is a processed bioenergy product. If commercialized, the black pellet could change major parts of the industry, thus making it a radical innovation.   One alternative to produce the black pellet is by using the torrefaction technology. The torrefaction technology has a historical record of many failed introductions. The step from pilot production to full scale commercialization is problematic in many ways, especially for a smaller developer. This study is focusing on the external commercialization problems for a radical innovation, the product black pellet and the technology torrefaction. The thesis aim to understand which these external barriers are for a torrefaction developer in Sweden and to answer our research question:   “What is the industry specific external barrier for a new entry-firm to commercialize black pellet with torrefaction technology?”   The theoretical framework is structured in two parts. The first one has a broad focus of theories regarding external barriers for commercialization of radical innovations, with a focus on small- and medium size enterprises. The second part focuses on general effects of industry structure and these two parts are combined in a conceptual theoretical framework. The findings in the study are based on empirical data collected through a total of six interviews with a supplier of torrefaction and black pellet, potential customers and market experts in Sweden.   The study’s analysis combines the theoretical and empirical data together with the industrial chapter, to create an understanding of the external barriers to commercialize black pellet with torrefaction technology. From the analysis we have understood many barriers, which could be summarized in four main barriers; lack of credibility, political incitements, strategic leadership and the costs of commercialization.   The answer to our research question, regarding the industry specific external barrier within the case of torrefaction and black pellet, is that black pellet and especially the torrefaction technology suffers from a lack of credibility by the actors in the market. From this answer, we have contributed with extended theoretical insights, that failures by previous actors create an external barrier for the current and future actors in their commercialization of a new technology.

Actuellement, l’industrialisation de la torréfaction de biomasse se heurte notamment à un manque de connaissances de la nature et de la quantité des matières volatiles produites en fonction des conditions opératoires et de la matière première. L’objectif de ces travaux est donc de mieux comprendre comment s’opère la torréfaction de la biomasse, en se concentrant sur l’étude de la perte de masse du solide et des rendements en matières volatiles. La torréfaction est considérée à partir de bois sec, sous atmosphère inerte et suivant un palier à une température comprise entre 200°C et 300°C. Lors d’une étude expérimentale, du hêtre et ses constituants, à savoir cellulose, xylane et lignine, ont été torréfiés, en régime chimique, dans une thermobalance et dans un pilote de torréfaction à échelle laboratoire. Le bilan matière boucle entre 97% et 104%. Les principales matières volatiles émises par la torréfaction de ce bois sont l’eau, le formaldéhyde, l’acide acétique et le CO2. De l’acide formique, du CO, du méthanol et du furfural sont aussi mesurés en quantité moindre. Certaines de ces espèces ne sont pas produites par tous les constituants du hêtre. Il semble en particulier que l’acide acétique soit produit à partir de la dégradation des acétates contenus dans les hémicelluloses. Par ailleurs, il apparaît en première approximation que la transformation peut être correctement représentée par la loi d’additivité jusqu’à 250°C. Cela n’est plus le cas à 280°C et 300°C, du fait d’interactions entre la cellulose et les deux autres constituants du bois. Celles-ci ralentissent la vitesse de torréfaction de la cellulose. A partir de ces résultats expérimentaux, a été développé dans ces travaux un modèle de torréfaction du bois, basé sur la superposition de « sous-modèles » décrivant chacun la torréfaction d’un constituant du bois. Ce modèle, qui présente comme originalité de prévoir en fonction de la proportion du bois en cellulose/hémicelluloses/lignine à la fois le rendement en solide et en huit espèces volatiles, et de prendre en compte les interactions à l’aide d’un facteur empirique, a été validé sur les expériences de torréfaction du hêtre entre 220°C et 300°C. Son utilisation a mis en évidence l’influence significative des contenus en hémicelluloses et cellulose sur les rendements en produits de la torréfaction. / The industrialization of the biomass torrefaction process requires better knowledge of the volatile species release versus operating conditions and feedstock. In this context, the present work aimed at studying solid mass loss and volatile species yields during biomass torrefaction. This transformation was considered on dry wood, at a temperature plateau between 200°C and 300°C and under inert atmosphere. First, torrefaction experiments were conducted under chemical regime on beechwood and its constituents – cellulose, lignin and hemicelluloses – in a thermobalance and in a lab-scale device. The mass balance closure was achieved with values ranging from 97 and 104%. The main volatile species measured were water, formaldehyde, acetic acid and CO2. Smaller amounts of methanol, CO, formic acid and furfural were also quantified. All those gas species were not produced by the three biomass constituents. In particular acetic acid seems to be produced by the degradation of the acetate groups contained in hemicelluloses. The results showed that in a first approximation torrefaction can be described by the additive law up to 250°C. But this law is not valid at 280°C and 300°C because of interactions between cellulose and the two other wood constituents. These interactions lead to a decrease in the torrefaction rate of cellulose. Based on these experimental results, a model of wood torrefaction was developed. It consists in the superposition of “sub-models” describing the torrefaction of each wood constituent. The originality of this model lies in its ability to predict both solid yield and eight volatile species yields depending on cellulose/hemicellulose/lignin wood composition, and to take into account interactions by means of an empirical factor. It was validated on beechwood torrefaction experiments between 220°C and 300°C. Finally, this model highlighted the significant influence of the proportion of hemicellulose and cellulose on torrefaction product yields.