Two computer programs were developed and used to determine the optimum operating parameters of a fedbatch and a continuous two-stage process for fermentation of recombinant bacteria. The study was conducted in three phases: (a) developing two computer programs for simulation and optimization of the above processes, (b) conducting batch culture fermentations to verify the performance of the biokinetic model, and (c) conducting fedbatch and two-stage continuous fermentation experiments to closely examine the simulation and optimization results. The Miao and Kompala (1992) biokinetic model was used for simulation of the bacterial growth and cloned gene expression. The Pattern-Search method, developed by Hooke and Jeeves (1962), was incorporated in the programs to determine the optimum values of the parameters. Extensive studies of the optimization results showed 30-40% higher productivities for the two stage continuous process over the fedbatch process when using the same media in both processes. In addition, increasing the number of stages in the continuous two-stage process resulted in very limited improvement in the productivity of the process (10-12%). The information from the process optimization was then used to design batch, fedbatch nd two stage continuous experiments. Recombinant <i>E. coli </i>(strain BL21DE3) with an inducible gene (sensitive to IPTG, isopropyl-â-D-thiogalactopyranoside) was used throughout the experiments. The experimental results from the fedbatch and two stage continuous processes clearly showed good agreement with the simulation and optimization results $(\cong$15% deviation). The experiments also revealed that the maintenance of plasmid harboring cells over the long-term operation could be an important barrier in achieving the predicted high productivity in the two stage continuous process. Finally, in addition to computer programs for optimization of genetically modified microorganisms, a new computer program with a generic algorithm for optimization of multiple CFSTR fermentation with any kind of biokinetic model was developed. The program was used to optimize multiple CFSTRs with the cybernetic biokinetic model for the first time. Besides finding the optimum residence times for multiple CFSTRs operation, the effect of inaccuracies in different cybernetic model parameters on the overall productivity of the process was investigated. The simulation results illustrated that, a single CFSTR was more sensitive in its operation to inaccuracies in the biokinetic constants as compared to optimized CFSTRs in series (2-8 times more sensitive).

Le micromélange, ou mélange à l'échelle moléculaire, a une influence déterminante sur la qualité des produits, la sélectivité, et le rendement de réactions d'importance industrielle comme les réactions de polymérisation. A l'heure actuelle, les systèmes chimiques existants sont bien adaptés à l'étude du micromélange de produits faiblement visqueux, mais il n'existe pas de méthode simple et opérationnelle pour les fluides plus visqueux. Inscrite dans le cadre du projet ANR PROCIP, l'étude a consisté à mettre en place une nouvelle méthode de quantification de l'efficacité du micromélange en milieu visqueux dans des réacteurs agités. La méthode utilisée s'inspire d'un système chimique basé sur deux réactions concurrentes, le système de réaction test iodure-iodate, dont la sélectivité d'un des produits est une mesure de l'efficacité du micromélange. L'approche propose un nouveau protocole basé sur le choix d'un tampon phosphate et l'utilisation de l'HydroxyEthylCellulose (HEC, 720 000 g/mol) comme agent viscosifiant inerte, qui possède un fort pouvoir viscosifiant permettant d'augmenter significativement la viscosité sans en ajouter plus de 1 % en masse dans le milieu. Des essais de caractérisation du micromélange en réacteur agité ont mis en évidence le ralentissement du micromélange au fur et à mesure que la viscosité augmente pour des pourcentages en HEC inférieurs à 0,5 %. Par contre, au-delà de 0,5 %, une « amélioration » du micromélange a été observée dans une large gamme de conditions expérimentales (différentes concentrations de réactifs, positions de l'injection, vitesses d'agitation et mobiles d'agitation). Plusieurs causes potentielles ont été formulées en vue d'expliquer ce phénomène et les cinétiques des réactions mises en jeu ont également été étudiées en présence de l'agent viscosifiant HEC et s'avèrent être inchangées. Enfin, une étude des interactions possibles entre les espèces iodées et l'HEC a été réalisée pour étudier l'existence d'une réaction parasite qui pourrait expliquer les résultats obtenus / Micromixing, or mixing at the molecular scale, has a decisive influence on product quality, selectivity and yield of industrially important reactions such as polymerization reactions. Currently, the existing chemical systems are well suited to the study of micromixing of low viscosity products, but no reliable characterization method exists for highly viscous media. As a part of the ANR project PROCIP, the study consisted in developing a new method to quantify the efficiency of micromixing in viscous media in stirred reactors. The method used is inspired by a chemical system based on two competitive reactions, the iodide-iodate test reaction system whose selectivity of a product is a measure of the micromixing efficiency. The approach proposes a new protocol based on a phosphate buffer and the use of HydroxyEthylCellulose (HEC, 720,000 g / mol) as an inert viscosifying agent, enabling to significantly increase the viscosity without adding more than 1 % by mass in the medium. Characterization of micromixing in a stirred reactor showed the slowing down of micromixing as viscosity increases for percentages of HEC below 0.5 %. By contrast, above 0.5 %, an "improvement" of micromixing has been observed in a wide range of experimental conditions (different concentrations of reactants, injection positions, stirring speeds and impellers). Several possible causes have been made to explain this phenomenon unexpected and the kinetics of the reactions involved were also studied in the presence of the viscosifying agent HEC and appeared to be unchanged. Finally, a study of possible interactions between iodine species and HEC was performed to study the existence of a secondary reaction which could explain the results

Submitted by Neusa Fagundes (neusa.fagundes@unioeste.br) on 2017-09-05T19:08:26Z No. of bitstreams: 1 Thiago_Edwiges2017.pdf: 2041114 bytes, checksum: f435c702dd49431aa0534c4ce49e2533 (MD5) / Made available in DSpace on 2017-09-05T19:08:26Z (GMT). No. of bitstreams: 1 Thiago_Edwiges2017.pdf: 2041114 bytes, checksum: f435c702dd49431aa0534c4ce49e2533 (MD5) Previous issue date: 2017-08-18 / The economical development and urbanization process have contributed to the increase in solid waste generation and problems related to final disposal. Thus, the National Solid Waste Policy reinforces treatment in order to improve efficiency in the management of solid waste as one of its priorities and stimulates to recover and use energy. Anaerobic digestion has been suggested as a promising approach to the treatment of organic fraction of solid waste, mainly because it represents an opportunity to reduce environmental pollution and minimizes impacts concerning the energy crisis. So, this trial aimed at evaluating FVW biogas potential, collected in a wholesale market (Unit from Foz do Iguaçu/PR) through batch and semicontinuous tests. Fruits and vegetables trading from that unit generates approximately 70,500 t year-1, with 2,800 kg d-1 solid waste, composed of almost 85% organic matter. The influence of chemical composition of twelve FVW different samples with different compositions on biochemical methane potential (BMP) was analyzed. The BMP ranged from 288 LN CH4 kg VS-1 to 516 LN CH4 kg VS-1, with significant statistical differences among means. This was explained by in the waste chemical composition over time. BMP variation was most strongly correlated for lipid content and high calorific values (HCV). Linear regression analyses were carried out to develop statistical models to predict as fast as possible methane potential of this kind of waste. The models were analyzed according to chemical compounds and HCV based on simple and multiple regression. The model with the best statistical metrics included lipid, protein, cellulose, lignin, and HCV, with a 92.5% R² and lignin content negatively correlated to BMP. Since HCV and lipids were strongly correlated, and because HCV can be determined more rapidly than chemical composition, HCV may be useful for predicting BMP. In addition to batch tests, the performance of anaerobic digestion was also investigated in a semicontinuous mode using a FVW mixture as a single substrate. The performance of a continuous stirred tank reactor (CSTR) was monitored along with the gradual increase of organic loading rates (OLR) from 0.5 gVS L–1 d–1 to 5.0gVSL–1d–1. BMP of FVW used as a feedstock to CSTR was 370 LN CH4 kg VS-1 with 81% biodegradability. During the whole trial pH was stable and there was an adequate level of buffering capacity in the system. Volumetric biogas production (LN biogas L–1 d–1) increased linearly (R² = 94.4%) according to OLR increase. On the other hand, the specific methane production (LN CH4 kg VS-1) has registered the best performance at an OLR of 3.0 g VS L–1 d–1 and 30 days of hydraulic retention time, with 285 LN CH4 kg VS-1 added, reaching 74% of BMP measured by batch tests. A clear accumulation of volatile fatty acids (VFA) was monitored with a decreased specific methane yield was detected with an OLR above 3.0 g VS L–1 d–1. / O crescimento econômico e o processo de urbanização vêm contribuindo para o aumento da geração de resíduos sólidos e problemas relacionados à disposição final. A Política Nacional de Resíduos Sólidos reforça o tratamento com o intuito de melhorar a eficiência quanto ao gerenciamento dos resíduos sólidos no país, como uma de suas prioridades e enfatiza o incentivo à recuperação e ao aproveitamento energético. A biodigestão anaeróbia tem sido apontada como uma abordagem promissora ao tratamento da fração orgânica dos resíduos sólidos, sobretudo por representar oportunidade de reduzir a poluição ambiental e minimizar os impactos da crise energética. Assim, o objetivo deste trabalho foi avaliar o potencial de produção de biogás de resíduos de frutas e hortaliças (RFH) coletados na Central de Abastecimento (CEASA – Unidade de Foz do Iguaçu/PR) a partir de testes em batelada e em regime semicontínuo. A comercialização de frutas e hortaliças da unidade é de aproximadamente 70.500 t ano-1, com geração de resíduos sólidos de 2.800 kg d-1, compostos por aproximadamente 85% de matéria orgânica. A influência da composição química de doze amostras de RFH com diferentes composições sobre o potencial bioquímico de metano (PBM) foi analisada. O PBM dos resíduos variou entre 288 LN CH4 kg SV-1 e 516 LN CH4 kg SV-1, com diferenças estatísticas significativas entre as médias, o que pôde ser explicado pela variação na composição química dos resíduos ao longo do tempo. A variação do PBM foi mais fortemente correlacionada à variação do teor de lipídeos e do poder calorífico superior (PCS). Foram realizadas análises de regressão linear visando desenvolver modelos estatísticos para estimar, de forma mais rápida, o potencial metanogênico deste tipo de resíduo. Os modelos foram analisados considerando os componentes químicos e o PCS a partir de regressão linear simples e múltipla. O modelo com as melhores métricas estatísticas foi obtido considerando lipídio, proteína, celulose, lignina e PCS, com R² de 92,5% e com teor de lignina negativamente correlacionado ao PBM. Como o PCS e os lipídeos apresentaram forte correlação, e como o PCS pode ser determinado, em geral, de forma mais rápida que a composição química, o PCS pode ser considerado um parâmetro útil para a estimava do PBM. Além dos testes em batelada, investigou-se também o desempenho da biodigestão anaeróbia em regime semicontínuo utilizando-se uma mistura de RFH como único substrato. O desempenho de reator de mistura completa (do inglês CSTR) foi monitorado juntamente com o aumento gradativo da carga orgânica volumétrica (COV) de 0,5 g SV L-1 d-1 até 5,0g SV L-1 d-1. O PBM dos RFH utilizado para alimentar o reator foi 370 LN CH4 kg SV-1 com biodegradabilidade de 81%. Durante todo o ensaio o pH se manteve estável e o sistema apresentou boa capacidade tampão. A produção volumétrica de biogás (LN biogás L-1 d-1) aumentou de forma linear (R² = 94,4%) de acordo com aumento da COV. Já a produção específica de metano (LN CH4 kg SV-1) indicou o melhor desempenho com COV de 3,0 g SV L-1 d-1 e 30 dias de tempo de retenção hidráulica, com 285 LN CH4 kg SV-1 adicionados, e atingiu 74% de PBM, medido a partir dos testes em batelada. Observou-se um acúmulo evidente de ácidos graxos voláteis (AGV), com consequente redução da produção de metano a partir da COV acima de 3,0 g SV L-1 d-1.

In der vorliegenden Arbeit wird die Konzeptionierung und die Realisierung eines teilkontinuierlichen Thermolyseverfahrens zur rohstofflichen Verwertung von Polyethylen- und Polypropylenabfällen beschrieben. Die Ausführung in der zweistufigen Rührkesselkaskade gewährleistete dabei die Trennung der Prozessschritte Schmelzen sowie Cracken der Einsatzmaterialien, woraus ein Thermolyseöl hervorging, welches weitestgehend frei von den Verunreinigungen aus den Abfällen war. Das Thermolyseöl wies eine Zusammensetzung aus Benzin- und Dieselkomponenten auf und erfüllte einige entscheidende Kraftstoffkriterien. Zur Senkung der qualitätsmindernden Olefingehalte wurden in einem weiteren Teil dieser Arbeit Untersuchungen zur heterogen-katalysierten Hydrierung der Siedeschnitte im Schüttgutreaktor und im Rührautoklav durchgeführt. Dabei konnten die Olefingehalte durch den Einsatz eines Palladium-Platin-Trägerkatalysators sowie durch einen Nickel-Gerüstkatalysator maßgeblich gesenkt werden. Als unerwünschtes Nebenprodukt des Thermolyseprozesses resultierte ein geringer Anteil Koks, welcher in der Schmelze akkumulierte und sich teilweise an der Reaktorwand ablagerte. Die entstandenen Mengen wiesen eine Abhängigkeit von der Prozesstemperatur und -dauer auf und mussten aus dem Prozess geschleust werden. Als weitere Nebenprodukte traten ein leichtflüchtiges Spaltgas sowie ein Sumpfrückstand auf. Darüber hinaus wurde gezeigt, dass die aufzuwendende Energie für den Thermolyseprozess durch die Energiegehalte der Nebenprodukte gedeckt werden kann. Neben dem Einsatz als Kraftstoff empfahl sich das Thermolyseöl aufgrund seiner physikalischen und chemischen Eigenschaften gerade im Hinblick auf einen nachhaltigen Umgang mit Ressourcen als beachtenswerter Ersatz für fossiles Rohöl.

The loss of efficiency and performance of bioprocesses on scale-up is well known, but not fully understood. This work addresses this problem, by studying the effect of some fermentation gradients (pH, glucose and oxygen) at a larger scale in a bench-scale two compartment reactor (PFR + STR) using the cadaverine-producing recombinant bacterium, Corynebacterium glutamicum DM1945 Δact3 Ptuf-ldcC_OPT. The initial scale down strategy increased the magnitude of these gradients by only increasing the mean cell residence time in the plug flow reactor (τ_PFR). The cell growth and product related rate constants were compared as the τ_PFR was increased; differences were significant in some cases, but only up to 2 min residence time. For example, losses in cadaverine productivity when compared to the control fed-batch fermentation on average for the τ_PFR of 1 min, 2 min and 5 min were 25 %, 42 % and 46 % respectively. This indicated that the increasing the τ_PFR alone does not necessarily increase the magnitude of fermentation gradients. The new scale-down strategy developed here, increased the magnitude of fermentation gradients by not only increasing the τ_PFR, but also considering the mean frequency at which the bacterial cells entered the PFR section (f_m). The f_m was kept constant by reducing the broth volume in the STR. Hence, the bacterial cells also spent shorter times in the well mixed STR, as the τ_PFR was increased (hypothesised as giving the bacterial cells less time to recover the non-ideal PFR section of the SDR). On adoption of this strategy cadaverine productivity decreases for the τ_PFR of 1 min, 2 min and 5 min were 25 %, 32 % and 53 % respectively. Thus, highlighting that loss in performance is most likely to occur as the magnitude of heterogeneity within the fermentation environment increases. However, Corynebacterium glutamicum DM1945 Δact3 Ptuf-ldcC_OPT did show some resilience in its biomass productivity. It was only marginally affected in the harshest of conditions simulated here.

Submitted by Johnny Rodrigues (johnnyrodrigues@ufcg.edu.br) on 2018-03-20T13:04:38Z No. of bitstreams: 1 TARCISIO DAVID KONNA NUNES SANTOS - DISSERTAÇÃO PPGEQ 2016..pdf: 1645201 bytes, checksum: 6f9773a951b06aacdb26ff5287e21e18 (MD5) / Made available in DSpace on 2018-03-20T13:04:38Z (GMT). No. of bitstreams: 1 TARCISIO DAVID KONNA NUNES SANTOS - DISSERTAÇÃO PPGEQ 2016..pdf: 1645201 bytes, checksum: 6f9773a951b06aacdb26ff5287e21e18 (MD5) Previous issue date: 2016-03-23 / Este trabalho propôs um modelo matemático para um reator do tipo CSTR (continuous stirred-tank reactor) partindo de dados experimentais da cinética da reação de transesterificação entre o óleo de soja e o acetato de metila, tendo como catalisador uma enzima, a Candida Antartica. Um volume foi proposto para reator para que o tempo médio de residência da mistura reacional fosse suficiente tal que o rendimento esperado, com base nos dados obtidos pelo Matlab®, fosse obtido na saída do reator. O estudo reacional fluidodinâmico foi feito utilizando o software comercial ANSYS® versão 15.0. Duas configurações do reator foram propostas otimização, afim de tornar o processo o mais homogêneo possível, tentando atingir as configurações do reator de mistura perfeita. Os impelidores utilizados foram o tipo pás retas inclinadas e o ripo Ribbon (Âncora), afim de avaliar qual dos impelidores consegue o rendimento satisfatório com menor potência de operação. A potência requerida foi determinada através das equações para misturadores mecânicos, sendo avaliado apenas como variável a rotação das pás. / This work proposes a mathematical model to a reactor of CSTR-type (Continuous Stirred-tank reactor) starting from experimental data of the kinetics of transesterification of soybean oil and methyl acetate, with the catalyst enzyme, Candida Antarctica. A reactor volume was proposed for the average residence time of the reaction mixture was sufficient such that the expected yield, based on data obtained by Matlab, was obtained in the reactor outlet. The reaction fluid dynamic study was done using the commercial software ANSYS® version 15.0. The optimization of the reactor was proposed by comparing the reactor operation with two impellers separately in order to make the process more homogeneous as possible, trying to achieve the perfect blend of reactor configurations. The impellers used were the type inclined straight blades and Ribbon RIPO (Anchor), to determine which of the impellers can the satisfactory performance with lower power operation. The required power was determined by the equations for mechanical mixers, being rated only as a variable rotation of the blades

ALes gaz issus des procédés de pyrolyse-gazéification de la biomasse doivent être épurés. Ils contiennent des vapeurs condensables (goudrons), des aérosols, des particules solides fines, des composés soufrés et des métaux alcalins qu’il s’agit d’éliminer avant leur utilisation sur des turbines (production d’électricité) ou comme gaz de synthèse. Les expériences rapportées dans ce travail concernent les vapeurs condensables et leur conversion par craquage thermique. Les vapeurs sont produites par pyrolyse de la biomasse dans un premier réacteur (RP) à 540°C. Elles sont ensuite craquées dans un réacteur parfaitement auto-agité par jets gazeux (RPAA) associé en série avec le RP. Le RPAA fonctionne à plus haute température (550-1030°C) et le temps de séjour de la phase gazeuse dans le craqueur est compris entre 0,1 et 1 s. Tous les produits de réaction (charbon, vapeurs condensables et gaz permanents) sont récupérés et analysés. Le RPAA étant uniforme en température et en concentration, la détermination de constantes de vitesse à temps de séjour donné est assez aisée à partir de bilans de matière en vapeurs et gaz. Des schémas réactionnels globaux rendant compte du craquage des vapeurs en gaz mais aussi de leur possible maturation en composés plus réfractaires sont proposés et leurs constantes de vitesse optimisées à partir des résultats expérimentaux. Ces modèles permettent de simuler le craquage thermique d’une charge type issue d’un gazogène. On détermine les conditions optimales de fonctionnement (température, temps de séjour) du réacteur de craquage qui aboutissent à une concentration en vapeurs condensables la plus faible possible. On comparera ainsi l’efficacité du craquage thermique à celle des autres procédés d’épuration des goudrons. / Pyrolysis and gasification processes give rise to gases containing by-products such as condensable vapors (tars), aerosols, dust, sulfur compounds and inorganics which may considerably lower the efficiency of catalysts (if chemical synthesis is foreseen) or cause severe damages to motors and turbines (in case of electricity production). Hence, efficient gas treatments are needed. The experiments reported in the present work are related to thermal cracking of condensable vapors. These vapors are produced in a first reactor by biomass pyrolysis (PR) at 540°C. They undergo further cracking in a second vessel, a continuous serf stirred tank reactor (CSSTR), assembled in series with the PR. The CSSTR is operated at temperatures ranging from 550 to 1030°C and gas phase mean residence times ranging from 0,1 to 1 s. Reaction products (char, condensable vapors and permanent gases) are recovered and analyzed. Temperature as well as composition are uniform at any point of the CSSTR. Therefore, it is easy to derive values of kinetic constants from mass balances at a given residence time. Global vapor cracking schemes including gas formation as well as possible maturation into more refractory compounds are proposed. Their kinetic constants are optimized from the experimental results. These models are used to simulate the thermal cracking of a typical load flowing out from a gasifier. Optimal operating conditions of the cracking reactor (in terms of temperature and residence time) are determined to reach the lowest condensable vapors concentration. Thus, efficiency of thermal cracking can be compared to other gas treatment processes.

Le procédé de captage du dioxyde de carbone (CO2) par combustion fonctionnant en boucle chimique (Chemical Looping Combustion (CLC)) permet de produire de l’énergie à partir du méthane tout en captant le CO2 produit par la combustion. Ce procédé met en oeuvre un matériau porteur d’oxygène (NiO/NiAl2O4) qui est utilisé pour fournir de l’oxygène lors de la combustion du méthane et qui est ensuite régénéré sous air. Le matériau utilisé se dégrade au cours du temps ce qui accroît les coûts du procédé et diminue ses performances. L’étude présentée ici a pour objectif de déterminer quel est l’impact des phénomènes thermiques et chimiques sur la dégradation du matériau porteur d’oxygène. Les mécanismes réactionnels représentant la réduction et l’oxydation du porteur d’oxygène ont été déterminés et validés grâce à des études expérimentales et à la modélisation d’un réacteur parfaitement auto-agité (RPAA) et d’un réacteur à écoulement piston. L’importance du contrôle du dépôt de carbone dans le procédé a été démontrée. Ensuite, les paramètres cinétiques des réactions représentant la réduction de l’oxyde de nickel ont pu être déterminés grâce à un modèle original du RPAA, puis validés dans le réacteur piston. L’intérêt du RPAA pour la détermination de paramètres cinétiques dans le cas du procédé CLC a été présenté. Les paramètres obtenus permettent de prédire de manière correcte toutes les réactions même si un travail complémentaire est nécessaire pour obtenir une meilleure précision des résultats. Finalement, un mécanisme de dégradation du matériau porteur d’oxygène déduit des résultats expérimentaux a été proposé. Ce mécanisme décrit la production importante de fines particules se dissociant des grains et leur rôle dans les phénomènes d’agglomération observés. Le matériau support, supposé inerte, jouerait un rôle dans l’apport d’oxygène. La méthodologie développée dans ce travail pourrait être adaptée à l’analyse et la caractérisation d’autres matériaux porteurs d’oxygène / The Chemical Looping Combustion (CLC) process produces energy by combustion of methane while capturing the carbon dioxide (CO2). An oxygen carrier (NiO/NiAl2O4) is used to deliver oxygen during the combustion of methane. It is then regenerated by air. The oxygen carrier material degrades over time, which increases the costs of the process and reduces its performance. The present study aims at determining the impacts of thermal and chemical phenomena on the oxygen carrier degradations. The reaction mechanisms corresponding to the reduction and oxidation of the oxygen carrier are determined and validated through experimental studies and the modeling of a continuously auto-stirred tank reactor (CASTR) and a plug flow reactor. The importance of controlling the quantity of deposited carbon in the process is illustrated. Then, the kinetic parameters of the reactions representing the reduction of nickel oxide are determined with an original model of the CASTR and validated in the plug flow reactor. The interest of using the CASTR for the determination of kinetic constants of the reactions involved in CLC process is presented. The obtained parameters give a good description of all reactions even if additional work is required to obtain a better precision of the results. Finally, a degradation mechanism of the oxygen carrier has been proposed. This mechanism describes the large production of fine particles separated from the grains and their role in the observed agglomeration phenomena. The support material, supposed to be inert, provides some of its oxygen. The methodology developed in this work could be adapted for the analysis and the characterization of other oxygen-carriers

Användningen av kommunalt avfall för att producera förnybar energi, huvudsaklig biogas, uppfyller 8 av Sveriges 16 miljökvalitetsmål. Två typer av rötningsprocesser som används är kontinuerligt omrörda tankreaktorer (CSTR) och pluggflödesreaktorer (PFR). Rapportens syfte är att direkt jämföra de två reaktorkonfigurationerna, PFR och CSTR, med samma substratblandning. Båda reaktortyperna drevs med samma belastning (5,2 g VS/L dag), hydraulisk retentionstid (30 dagar) och driftstemperatur (520C). Två PFR-reaktorer drevs med återcirkulation av rötrester. Under projektets gång mättes olika parametrar, vilka var: pH, total gasvolym och gassammansättning, FOS/TAC (organiska syror/alkalinitet), flyktiga fettsyror (VFA), ammonium-kvävekoncentration och TS (torrsubstanshalt) och VS (organiska ämnen). Studieresultatet var uppdelat i två sektioner. Avsnitt 1 sammanfattar uppstartsperioden, där alla tre CSTR (D1-R, D2-NR och C1-NR) kördes i icke-återcirkulerande läge. Uppstartsperioden gjordes för att säkerställa reaktorernas stabilitet och det gällde att uppnå en liknande baslinje för detektion. Inom studien accepterades avvikelser för baslinjen inom intervallet 0-5%. För uppstartsperioden varierade den specifika metanproduktionen med 2 % mellan D1-NR och D2-R, D1-R och C1-NR med 6 %, och D2-NR och C1-NR med 8 %. För ammoniak-koncentrationen var skillnaden mellan D1-R och D2-NR 2 % och mellan D1-R och C1-NR 9 %, mellan D2-NR och C1-NR, 7%. Med dessa variationer togs beslutet att utesluta reaktor C1-NR från den experimentella perioden för projektet. Avsnitt 2 jämför PFR och CSTR återcirkulerat/icke återcirkulerat läge. En CSTR drevs med 30 % återcirkulation av rötrest och en CSTR drevs i icke-återcirkulationsläge. Att jämföra PFR och CSTR återcirkulerat/icke återcirkulerat läge visade att PFR:erna övervann båda CSTR:erna när det gäller den specifika metanproduktionen. Där PFR producerade 395±93 NL/KgVS och 372±42 (N)L/KgVS CSTR med återcirkulation 280±10 (N)L/KgVS och den icke-återcirkulerade CSTR producerade 250±7 (N)L/KgVS. NH4+-N-nivåerna mellan PFR och CSTR visade en skillnad på 1 g/L, där PFR var den med högre NH4+-N (2,53±0,39 g/L för PFR-A och 2,73±0,43 g/L för PFR-T). En av skillnaderna mellan de återcirkulerade och de icke-återcirkulerade CSTR:erna var VFA-koncentrationen. VFA-koncentrationen i de återcirkulerade var 1,05±0,90 g/L och 0,43±0,25 g/L i de icke-återcirkulerade. VFA-koncentrationen för PFR var i genomsnitt 0,28±0,12 g/L och 0,40 ±0,16 g/L. Graden av nedbrytning i CSTRs var 90-91%, och 63-65% i PFR. Den är möjligen felberäknad och visar en inkonsekvens mellan reaktorernas prestanda och specifik metanproduktion. / The use of municipal waste to produce renewable energy, in the form of biogas, fulfils eight of Sweden's sixteen environmental quality objectives. The most common technology to produce biogas is anaerobic digestion (AD). Two common types of AD processes are continuously stirred reactors (CSTR) and plug flow reactors (PFR). The thesis aims at directly comparing the two reactor configurations using the same substrate mix. For this, two PFRs (PFR-T and PFR-A) and three CSTRs (D1-R, D2-NR and C1-NR) were setup. All reactors were operated at the same organic loading rate (5.2 g VS/L-day), hydraulic retention time (30 days) and operational temperature (52°C). During the project, various parameters were measured to evaluate the reactor performance, such as pH, total gas volume and gas composition, FOS/TAC (organic acids/alkalinity), volatile fatty acids (VFA), ammonium-nitrogen concentration, and solid and volatile solid of digestate.   The study result was divided into two sections. Section 1 summarises the start-up period (42 days), where the CSTRs and the PFRs were operated under non-recirculation mode. The start-up period was done to ensure the stability of the reactors and was concerned with achieving a similar baseline for performance. For the start-up period, the specific methane production between D1-R and D2-NR varied by 2% while the variation between C1-NR and D2-NR was 8%. Likewise, for the ammonia concentration, the difference between D1-R and D2-NR was 2%, while C1-NR differed by 9% from D2-NR. Based on these findings, the decision was taken to exclude reactor C1-NR from the experimental period of the project.  Section 2 compares the PFR and CSTR. One CSTR was operated with 30% recirculation of digest, while the other CSTR was operated in non-recirculation mode. The two PFRs were operated with recirculation of digestate. The experiments showed that the PFRs outperformed both CSTRs regarding the specific methane production. The PFRs produced 395±93 (N)L/kg-VS and 372±42 (N)L/kg-VS. In contrast, the CSTR with recirculation reached 280±10 (N)L/kg-VS and the non-recirculated CSTR produced 250±7 (N)L/kg-VS. The NH4+-N levels for the PFRs and CSTRs showed a 1 g/L difference, with a higher value for the PFRs (2.53±0.39 g/L for PFR-A, and 2.73±0.43 g/L for PFR-T). One of the significant differences between the recirculated and the non-recirculated CSTRs was the VFA concentration. The VFA concentration in the recirculated and non-recirculated reactor was 1.05±0.90 g/L and 0.43±0.25 g/L, respectively. The VFA concentration for the two PFRs was on average 0.28±0.12 g/L and 0.40 ±0.16 g/L. The degree of degradation in the CSTRs was 90-91% compared to 63-65% in the PFRs. The degree of degradation is possibly miscalculated and shows an inconsistency between the reactor performance in terms of specific methane production.

Les systèmes réels sont généralement de nature non-linéaire, et leurs modélisations etsurveillance restent une tâche difficile à accomplir. Néanmoins, avec les progrès technologiqueson dispose maintenant d'un atout de taille sur ces systèmes qui est les données.Ce travail présente une technique de diagnostic de défaut et de modélisation basée en grandepartie sur la méthode d'apprentissage automatique « Les Machines à Vecteurs de Support,SVM » qui est basée sur les données. La méthodologie proposée est appliquée à différentessystèmes multivariables et non linéaires, à savoir : un procédé de traitement des eaux usées, unsystème éolien et un réacteur chimique parfaitement agité.L'objectif de cette thèse de doctorat est d'examiner la possibilité d'extraire le maximumd'information à partir de données afin de surveiller efficacement le comportement de systèmesréels et de détecter rapidement tout défaut qui peut compromettre leur bon fonctionnement. Lamême méthode est utilisée pour la modélisation des différents systèmes. Plusieurs défis ont étérelevés tels que la complexité du comportement des systèmes, le grand nombre de mesuresvariant à différentes échelles de temps, la présence de bruit et les perturbations. Une méthodegénérique de diagnostic de défauts est proposée par la génération des caractéristiques de chaquedéfaut suivie d’une étape d'évaluation de ces caractéristiques avec une amélioration du transfertde connaissances en modélisation.Dans cette thèse ont a démontré l'utilité de l'outil Machines à Vecteurs de Support, enclassification par la construction de modèles de décision SVM dédiés à l'évaluation descaractéristiques de défaut, et aussi en tant qu'estimateur non linéaire/ou pour la modélisation parl'utilisation des machines à vecteurs de support dédiés pour la régression (SVR).La combinaison de SVM et d’une méthode basée sur le modèle "observateur" a été aussi étudiéeet a été nécessaire dans certains cas pour garantir un bon diagnostic de défauts. / Real systems are usually nonlinear and their modeling and monitoring remains adifficult task. However, with advances in technology and the availability of big amounts of data,we have a facility to operate these systems.This work presents a methodology for fault diagnosis and modeling which is in large part basedon the method of Support Vector Machines (SVM) which data-based. The proposedmethodology is applied to various nonlinear multivariable systems including: wastewatertreatment processes, wind turbines and stirred tank reactors.The objective of this PhD is to examine the possibility of extracting the maximum of informationfrom data to effectively monitor the behavior of real systems and rapidly detect any faults whichmay impair their proper functioning. The same method is used for modeling the differentsystems. Several challenges were identified and surmounted such as the complexity of thesystem behavior, large amount of data varying at different time scales, the presence of noise anddisturbances. A generic method of fault diagnosis is proposed for the generation of the faultcharacteristics followed by an evaluation of these characteristics as well as an improved transferof knowledge in modeling.In this thesis the usefulness of the tool Support Vector Machines in Classification has beendemonstrated by the construction of decision models dedicated to evaluating the characteristicsof faults, and also its usefulness for modeling/ or as estimator for the nonlinear systems usingsupport vector machines dedicated for regression (SVR).The combination of SVM and a method based on models “observer” was also considered andwas found to be interesting in some cases to ensure proper fault diagnosis.