Die lateinamerikanischen Währungskrisen lenkten erst kürzlich wieder das Augenmerk auf die Gefahren wechselkursbasierter Stabilisierungen (WKBS). Dies sind Inflationsstabilisierungsprogramme, die den nominalen Wechselkurs als vorrangiges geldpolitisches Instrument einsetzen. Die vorliegende Dissertation dokumentiert die Wirkung der Stabilisierungen und präsentiert Erklärungsmodelle für deren stilisierte Fakten. Das erste Kapitel untersucht in Burns-Mitchell-Diagrammen typische reale und monetäre Effekte von 13 Stabilisierungsepisoden. Der anfängliche Anstieg des Konsums und des BIPs, die reale Aufwertung und die Verschlechterung der Leistungsbilanz sind dabei die auffälligsten stilisierten Fakten. Auf die Expansion folgt eine wirtschaftliche Abschwächung, d. h. niedrigeres oder Nullwachstum, falls die Stabilisierung noch andauert, und negative Wachstumsraten, falls das Programm bereits aufgegeben wurde. Die Kapitalimporte folgen einem ähnlichen Zyklus: Dem Anstieg zu Beginn der Stabilisierung folgt drei bis sechs Jahre später eine drastische Umkehr, die häufig mit dem Zusammenbruch des Programms einhergeht. Die Kurzlebigkeit von WKBS ist ein weiterer stilisierter Fakt: 70 % der betrachteten Stabilisierungen scheiterten innerhalb von 10 Jahren. Die anfängliche reale Aufwertung während WKBS wird meist als Anstieg des relativen Preises nicht-handelbarer Güter modelliert; empirische Ergebnisse hingegen unterstreichen die Bedeutung der internationalen Preisunterschiede handelbarer Güter. Kapitel 2 untersucht diese Ursachen durch die Anwendung von Engels Methode der Varianzzerlegung auf den realen Wechselkurs zwischen Brasilien und den USA. Die Ergebnisse bestätigen dabei sowohl die Modelle als auch den empirischen Befund: Bei Betrachtung der gesamten Stichprobe (von Januar 1981 bis Mai 2001) bestimmen Veränderungen der Preise handelbarer Güter und des nominalen Wechselkurses nahezu die gesamten Bewegungen des realen Wechselkurses. Während Perioden fester Wechselkurse hingegen sind die Preise nicht-handelbarer Güter von ähnlicher Bedeutung. Aufgrund dieses Ergebnisses wird in den in Kapitel 3 und 4 präsentierten Modellen der reale Wechselkurs in Abhängigkeit des relativen Preises nicht-handelbarer Güter dargestellt. Diese Modelle bilden kleine, offene Volkswirtschaften ab, die von nutzenmaximierenden repräsentativen Agenten mit perfekter Voraussicht bevölkert sind. Monetäre Größen sind aufgrund von cash-in-advance- Beschränkungen von Bedeutung. Weitere wichtige Modellelemente sind die Existenz von Marktimperfektionen (Preisstarrheiten und unvollständige Kapitalmobilität) sowie die mangelnde Glaubwürdigkeit der Stabilisierung. In Kapitel 3 ist diese durch die Antizipation einer Währungskrise à la Krugman (1979) begründet. Die reale Aufwertung kann dann mit vorausblickender Preissetzung der monopolistischen Produzenten nicht-handelbarer Güter erklärt werden: Aufgrund von Preisstarrheiten erhöhen diese ihre Preise in Erwartung der Währungsabwertung. Da die Preise handelbarer Güter durch das Gesetz des einheitlichen Preises bestimmt werden, folgt daraus ein Anstieg des relativen Preises nicht-handelbarer Güter und eine reale Aufwertung. Zudem wird aufgrund intertemporaler Konsumsubstitution der anfängliche Konsumboom reproduziert. Ökonometrische Evidenz bestätigt den Preissetzungsmechanismus: KQ-Schätzungen mit monatlichen mexikanischen Daten zeigen für Perioden fester Wechselkurse einen signifikanten positiven Zusammenhang zwischen dem relativen Preis nicht-handelbarer Güter und dem mexikanisch-US-amerikanischen Zinsdifferential als Approximation der Abwertungserwartung. Während das Ende der Stabilisierung in obigen Modell durch eine fundamentale Inkonsistenz von Wechselkursziel und Staatsausgaben bedingt ist, zeigt Kapitel 4, daß der Zusammenbruch auch aus sich-selbst-erfüllenden Erwartungen resultieren kann. Ein wesentliches Element ist dabei die Begrenzung der internationalen Kapitalmobilität. Diese erlaubt es, sowohl den anfänglichen Konsumboom als auch das Ende der Stabilisierung mit Erwartungen bezüglich der Dauer des Pegs und der nachfolgenden Geldpolitik zu erklären. Zusammenfassend zeigt die Dissertation die Gefahren wechselkursbasierter Stabilisierungen auf: Im Gegensatz zur herkömmlichen Meinung sind sogar relativ erfolgreiche und langlebige WKBS mittelfristig mit einer Kontraktion verbunden; nur wenige Programme erzielen eine nachhaltige Reduktion der Inflation. Die hier präsentierten Modelle zeigen, daß die mangelnde Glaubwürdigkeit der Programme - selbst wenn diese nicht durch Fundamentaldaten gerechtfertigt ist - zu Allokationsverzerrungen führt und den Erfolg der Stabilisierungsmaßnahmen gefährdet. / Exchange rate crises in Latin America recently put a spotlight on the perils of Exchange Rate-Based Stabilizations (ERBS), which use the nominal exchange rate as the main policy target for stabilizing inflation. This dissertation documents the effects of ERBS in high inflation economies and develops models to explain these stylized facts. The first chapter assesses the empirical regularities associated with ERBS. Based on a sample of 13 stabilization episodes, typical real and monetary dynamics are investigated in Burns-Mitchell diagrams. Stylized facts of ERBS are the initial increases in consumption and GDP, the real appreciation and the current account deterioration. Moreover, consumption and output are found to follow a boom-slowdown cycle, where slowdown means reduced or zero growth if the ERBS is still in effect, and negative growth rates for failed ERBS. Capital inflows follow a similar boom-bust cycle: Their increase at the stabilization''s inception is followed by a sharp reversal three to six years later, very often coinciding with the program''s collapse. This transitoriness of ERBS constitutes an additional stylized fact: 70 % of the programs under consideration failed within 10 years after their implementation. The origin of the initial real exchange rate appreciation during ERBS has been subject to controversy: Most models assume an increase in the relative price of non-traded goods. Empirical findings, in contrast, emphasize the contribution of traded goods'' cross-country prices. Chapter 2 sheds light on this issue by applying Engel''s (1995) method of variance decomposition on Brazilian-US real exchange rate fluctuations. The results confirm both the models and the empirical findings: When considering the full sample (from January 1981 to May 2001), changes in traded goods'' prices and the nominal exchange rate account for almost all of the observed real exchange rate movements. During periods of pegged exchange rates, however, non-traded goods'' prices are equally important for real exchange rate fluctuations. Thus, changes in relative non-tradables'' prices are incorporated as a determinant of real exchange rate fluctuations during ERBS in the theoretical frameworks presented in chapters 3 and 4. These explain the stylized facts in models of small, open economies populated by a utility-maximizing representative agent endowed with perfect foresight. Money matters due to cash-in-advance constraints. Further important features are the existence of market imperfections - price stickiness or imperfect capital mobility - and the stabilization''s deficient credibility. In chapter 3, the latter is due to the anticipation of a Krugman (1979)-style currency crisis. The initial real appreciation during ERBS can then be explained with forward-looking price setting by monopolistic non-tradable goods'' producers: These are subject to staggered price setting and incorporate the peg''s anticipated termination - i.e. higher future devaluation rates - by increasing their current prices. As tradables'' prices are determined by the law of one price, this implies higher relative non-tradables'' prices and thus a real exchange rate appreciation. Furthermore, due to intertemporal consumption substitution, the observed initial consumption boom is reproduced. Econometric evidence confirms the proposed price setting mechanism: Using the Mexican-US interest rate differential as an indicator for devaluation rate expectations, OLS regressions with monthly Mexican data find a significant positive relation between relative non-tradables'' prices and the interest rate spread during periods of pegged exchange rates. In the previous model, the stabilization effort collapses due to a fundamental inconsistency between the exchange rate target and government finance. Chapter 4 shows that the collapse can also result from self-fulfilling expectations. This is achieved by introducing partial international capital mobility. Given this constraint, both the initial consumption boom and the stabilization''s collapse can be shown to result from expectations about the duration of the peg and post-stabilization monetary policy. In conclusion, the dissertation points to the perils of ERBS in high inflation countries: Contrary to what is commonly believed, even relatively successful and long-lived exchange rate pegs are associated with a late slowdown; only very few ERBS are successful at stabilizing inflation rates in the medium and long run. The models show that stabilizations'' deficient credibility - regardless if justified by fundamentals or not - engenders real dynamics which distort economic activity and jeopardize the stabilization effort: The miracle of ERBS turns into a mirage.

Coal-fired power plants are the largest anthropogenic point sources of CO2 emissions worldwide. About 40% of the world's electricity comes from coal. Approximately 49% of the US electricity in 2008 and 23% of the total electricity generation of Canada in 2000 came from coal-fired power plant (World Coal Association, and Statistic Canada). It is likely that in the near future there might be some form of CO2 regulation. Therefore, it is highly probable that CO2 capture will need to be implemented at many US and Canadian coal fired power plants at some point. Several technologies are available for CO2 capture from coal-fired power plants. One option is to separate CO2 from the combustion products using conventional approach such as chemical absorption/stripping with amine solvents, which is commercially available. Another potential alternative, membrane gas separation, involves no moving parts, is compact and modular with a small footprint, is gaining more and more attention. Both technologies can be retrofitted to existing power plants, but they demands significant energy requirement to capture, purify and compress the CO2 for transporting to the sequestration sites. This thesis is a techno-economical evaluation of the two approaches mentioned above along with another approach known as hybrid. This evaluation is based on the recent advancement in membrane materials and properties, and the adoption of systemic design procedures and optimization approach with the help of a commercial process simulator. Comparison of the process performance is developed in AspenPlus process simulation environment with a detailed multicomponent gas separation membrane model, and several rigorous rate-based absorption/stripping models. Fifteen various single and multi-stage membrane process configurations with or without recycle streams are examined through simulation and design study for industrial scale post-combustion CO2 capture. It is found that only two process configurations are capable to satisfy the process specifications i.e., 85% CO2 recovery and 98% CO2 purity for EOR. The power and membrane area requirement can be saved by up to 13% and 8% respectively by the optimizing the base design. A post-optimality sensitivity analysis reveals that any changes in any of the factors such as feed flow rate, feed concentration (CO2), permeate vacuum and compression condition have great impact on plant performance especially on power consumption and product recovery. Two different absorption/stripping process configurations (conventional and Fluor concept) with monoethanolamine (30 wt% MEA) solvent were simulated and designed using same design basis as above with tray columns. Both the rate-based and the equilibrium-stage based modeling approaches were adopted. Two kinetic models for modeling reactive absorption/stripping reactions of CO2 with aqueous MEA solution were evaluated. Depending on the options to account for mass transfer, the chemical reactions in the liquid film/phase, film resistance and film non-ideality, eight different absorber/stripper models were categorized and investigated. From a parametric design study, the optimum CO2 lean solvent loading was determined with respect to minimum reboiler energy requirement by varying the lean solvent flow rate in a closed-loop simulation environment for each model. It was realized that the success of modeling CO2 capture with MEA depends upon how the film discretization is carried out. It revealed that most of the CO2 was reacted in the film not in the bulk liquid. This insight could not be recognized with the traditional equilibrium-stage modeling. It was found that the optimum/or minimum lean solvent loading ranges from 0.29 to 0.40 and the reboiler energy ranges from 3.3 to 5.1 (GJ/ton captured CO2) depending on the model considered. Between the two process alternatives, the Fluor concept process performs well in terms of plant operating (i.e., 8.5% less energy) and capital cost (i.e., 50% less number of strippers). The potentiality of hybrid processes which combines membrane permeation and conventional gas absorption/stripping using MEA were also examined for post-combustion CO2 capture in AspenPlus®. It was found that the hybrid process may not be a promising alternative for post-combustion CO2 capture in terms of energy requirement for capture and compression. On the other hand, a stand-alone membrane gas separation process showed the lowest energy demand for CO2 capture and compression, and could save up to 15 to 35% energy compare to the MEA capture process depending on the absorption/stripping model used.

Coal-fired power plants are the largest anthropogenic point sources of CO2 emissions worldwide. About 40% of the world's electricity comes from coal. Approximately 49% of the US electricity in 2008 and 23% of the total electricity generation of Canada in 2000 came from coal-fired power plant (World Coal Association, and Statistic Canada). It is likely that in the near future there might be some form of CO2 regulation. Therefore, it is highly probable that CO2 capture will need to be implemented at many US and Canadian coal fired power plants at some point. Several technologies are available for CO2 capture from coal-fired power plants. One option is to separate CO2 from the combustion products using conventional approach such as chemical absorption/stripping with amine solvents, which is commercially available. Another potential alternative, membrane gas separation, involves no moving parts, is compact and modular with a small footprint, is gaining more and more attention. Both technologies can be retrofitted to existing power plants, but they demands significant energy requirement to capture, purify and compress the CO2 for transporting to the sequestration sites. This thesis is a techno-economical evaluation of the two approaches mentioned above along with another approach known as hybrid. This evaluation is based on the recent advancement in membrane materials and properties, and the adoption of systemic design procedures and optimization approach with the help of a commercial process simulator. Comparison of the process performance is developed in AspenPlus process simulation environment with a detailed multicomponent gas separation membrane model, and several rigorous rate-based absorption/stripping models. Fifteen various single and multi-stage membrane process configurations with or without recycle streams are examined through simulation and design study for industrial scale post-combustion CO2 capture. It is found that only two process configurations are capable to satisfy the process specifications i.e., 85% CO2 recovery and 98% CO2 purity for EOR. The power and membrane area requirement can be saved by up to 13% and 8% respectively by the optimizing the base design. A post-optimality sensitivity analysis reveals that any changes in any of the factors such as feed flow rate, feed concentration (CO2), permeate vacuum and compression condition have great impact on plant performance especially on power consumption and product recovery. Two different absorption/stripping process configurations (conventional and Fluor concept) with monoethanolamine (30 wt% MEA) solvent were simulated and designed using same design basis as above with tray columns. Both the rate-based and the equilibrium-stage based modeling approaches were adopted. Two kinetic models for modeling reactive absorption/stripping reactions of CO2 with aqueous MEA solution were evaluated. Depending on the options to account for mass transfer, the chemical reactions in the liquid film/phase, film resistance and film non-ideality, eight different absorber/stripper models were categorized and investigated. From a parametric design study, the optimum CO2 lean solvent loading was determined with respect to minimum reboiler energy requirement by varying the lean solvent flow rate in a closed-loop simulation environment for each model. It was realized that the success of modeling CO2 capture with MEA depends upon how the film discretization is carried out. It revealed that most of the CO2 was reacted in the film not in the bulk liquid. This insight could not be recognized with the traditional equilibrium-stage modeling. It was found that the optimum/or minimum lean solvent loading ranges from 0.29 to 0.40 and the reboiler energy ranges from 3.3 to 5.1 (GJ/ton captured CO2) depending on the model considered. Between the two process alternatives, the Fluor concept process performs well in terms of plant operating (i.e., 8.5% less energy) and capital cost (i.e., 50% less number of strippers). The potentiality of hybrid processes which combines membrane permeation and conventional gas absorption/stripping using MEA were also examined for post-combustion CO2 capture in AspenPlus®. It was found that the hybrid process may not be a promising alternative for post-combustion CO2 capture in terms of energy requirement for capture and compression. On the other hand, a stand-alone membrane gas separation process showed the lowest energy demand for CO2 capture and compression, and could save up to 15 to 35% energy compare to the MEA capture process depending on the absorption/stripping model used.

本文研究目的有四：(1) 研究以本國通貨貶值率為政策工具的影響效果與相關的議題。(2) 探討由固定匯率到爬行釘住匯率再到浮動匯率崩潰的過程，並且分析有何關鍵因素，會影響匯率制度崩潰時間。(3) 討論匯率定錨政策失敗的原因，並提出以往文獻可能忽略的因素。(4) 分析通貨替代性是否會影響匯率政策的效果，甚至使匯率政策遭遇困難而失敗。為了達成研究目的，本文內容共分五章，除了第一章的緒論和第五章的總結外，其餘各章的內容為： 第二章延伸Calvo (1981) 的架構，加入Chen, Tsaur and Chou (1981)、曹添旺 (1987) 及張文雅、賴景昌和曹添旺 (1991) 的觀點，並仿照Liviatan (1891)、Engel (1989)、Calvo (1985) 的作法，設立一個具有通貨替代環境的小型開放經濟模型，探討本國通貨貶值率上升對經濟體系的影響。文中發現通貨之間若有較高 (較低) Edgeworth 替代關係，則本國通貨貶值率上升，經常帳會因此改善 (惡化)。本章亦針對通貨替代程度的不同，分析提升本國通貨貶值率政策對社會福利的影響效果，其中發現：不論是通貨替代程度大小，提升本國通貨貶值率政策對社會福利的影響均有兩種不同的情況，其中的一種是確定福利下降，另一種是不確定。 第三章仍然依據Chen, Tsaur and Chou (1981)、曹添旺 (1987) 及張文雅、賴景昌和曹添旺 (1991) 所強調通貨能提供流動性勞務，降低交易成本的功能，將本國通貨和外國通貨放入模型內，建立一個以交易功能為基礎的貨幣模型，探討由固定匯率崩潰到爬行釘住匯率，再由爬行釘住匯率崩潰到浮動匯率的過程，有何關鍵因素，會影響匯率制度崩潰時間。結果發現：(1) 不論是通貨替代、通貨互補或是通貨獨立的環境下，實際發生匯率制度崩潰的時間早於自然崩潰的時間。(2) 制度崩潰的時點由通貨替代或是通貨互補的程度決定，通貨替代性越是極端 (例如：完全互補或完全替代) 將使通貨危機越早發生，而通貨替代關係越低將使通貨危機發生時間延後。(3) 當固定匯率轉換到爬行釘住匯率時，所產生的外匯存底流失的額度將會是決定爬行釘住匯率制度是否能夠執行的關鍵。 第四章以代表性個人最適化模型的架構，加入Végh (1995) 主張外國通貨能提供流動性勞務，降低交易成本的觀點，建立一個具有通貨替代環境的交易成本貨幣成長模型，分析名目匯率定錨政策崩潰的過程與通貨替代程度的關連性，結果發現匯率定錨政策失敗的主因可能是名目匯率定錨政策違反經常帳跨期平衡所致。為了避免通貨危機提早發生，匯率政策須視通貨替代程度的高低作適當的調整。通貨替代程度較高時，貨幣當局應當在政策執行時，大幅的降低本國通貨貶值率，可使政策崩潰的時間延後；而通貨替代程度較低時，貨幣當局應當採用小幅度降低本國通貨貶值率，可拖延政策崩潰發生的時間。 / This paper has four purposes: (1) Research in the influence and relevant topics about adopting depreciation rate of the domestic currency as policy instrument. (2) Show the dynamics transition process from the fixed, to the crawling peg, and then to the flexible exchange rate regimes, and analyze the central factor about time of collapse of exchange rate regimes. (3) Research in the reason about the exchange-rate-based stabilization program fails, and it may be ignored by the existing literature. (4) Analyze the influence of the currency substitutability affects the effect of exchange rate policy, even to cause difficulty and failure. In order to achieve the research purposes, the content of this paper divides into five chapters, except that chapter 1 is the introduction and chapter 5 is conclusion; the rest is organized as follows: In chapter 2, this paper expands Calvo (1981) model, combining the spirit of Chen, Tsaur and Chou (1981), Tsaur (1987), Chang, Lai and Tsaur (1991) and the approach of Liviatan (1891), Engel (1989), Calvo (1985), we set up the framework for small open economy with currency substitution environments to explore the policy effects of a rise in the rate of devaluation domestic currency on economy. We find that if two currencies are higher (lower) degree of Edgeworth substitution, then a rise in the rate of devaluation domestic currency induced a current account surplus (deficit) on the transition path. This chapter also aims at the currency substitution degree the difference, the analysis about a rise in the rate of devaluation domestic currency to the social welfare influence effect. We show that no matter is the currency substitution degree size, a rise in the rate of devaluation domestic currency to have two kind of different situations to the social welfare influence, one is the definite welfare drops; another is an ambiguous. In chapters 3, we investigate the relationship between the collapse timing of exchange rate regime and degree of substitutability of foreign currency for domestic currency as a medium of exchange. According to the spirit of Chen, Tsaur and Chou (1981), Tsaur (1987), Chang, Lai and Tsaur (1991), we set up a transaction-based monetary model in which money provides liquidity services to reduce transaction costs, domestic and foreign currencies are introduced into the system. The results of study show (1) no matter what degree of substitutability of foreign currency for domestic currency, the collapse timing of the exchange rate regime is earlier than the natural collapse timing, and (2) the collapse timing depends on the degree of currency substitution or complement. We find that the extreme degree of substitutability of currencies led to earlier currency crisis; however, moderate degree of substitutability of currencies prolong currency crisis. We also explore the dynamic transition process from the fixed to the crawling peg, and then to the flexible exchange rate regimes. We find that the extent of the decrease in foreign exchange reserves is the key whether the fixed exchange rate is able to transform into the crawling peg exchange rate regime. In chapter 4, we investigate the relationship between the failures of exchange-rate-based stabilization program and degree of foreign currency substitutability. According to Végh (1995), we set up a transaction-based monetary model with currency substitution in which foreign currencies provides liquidity services to reduce transaction costs. The reason of failure on exchange-rate-based stabilization program is that it violates intertemporal current account balance constraint. In order to prolong the time of the currency crisis occurs, the exchange-rate-based stabilization program must regard the degree of currency substitution. When the degree of currency substitution is high, the monetary authority must reduce the rate of devaluation domestic currency largely, and when the degree of currency substitution is low, the monetary authority must narrow scale to reduce the rate of devaluation domestic currency, and prolong time of policy collapse occurrence.

While FPGAs provide flexibility for performing high performance DSP functions, they consume a significant amount of power. Often, a large portion of the dynamic power is wasted on unproductive signal glitches. Reducing glitching reduces dynamic energy consumption. In this study, retiming is used to reduce the unproductive energy wasted in signal glitches. Retiming can reduce energy by up to 92%. Evaluating energy consumption is an important part of energy reduction. In this work, an activity rate-based power estimation tool is introduced to provide FPGA architecture independent energy estimations at the gate level. This tool can accurately estimate power consumption to within 13% on average. This activation rate-based tool and retiming are combined in a single algorithm to reduce energy consumption of FPGA designs at the gate level. In this work, an energy evaluation metric called energy area delay is used to weigh the energy reduction and clock rate improvements gained from retiming against the area and latency costs. For a set of benchmark designs, the algorithm that combines retiming and the activation rate-based power estimator reduces power on average by 40% and improves clock rate by 54% for an average 1.1x area cost and a 1.5x latency increase.

Les gaz issus de la combustion des énergies fossiles dans les centrales électriques contiennent une grande variété de polluants tels que les gaz-acides et ne peuvent être rejetés directement dans l'atmosphère. Ces polluants gazeux doivent être traités par des méthodes de captage en post combustion dans des colonnes d'absorption utilisant des solvants chimiques. L'objectif est donc de concevoir une unité d'élimination sélective des gaz-acides tels que CO2, H2S et COS en utilisant des solvants de la famille des alcanolamines. Cette thèse développe dans un premier temps, un modèle de non-équilibre, adapté aux systèmes multiconstituants électrolytiques et réactifs, pour la représentation des colonnes d'absorption réactive. Le modèle comporte des modules pour représenter la thermodynamique en espèces vraies (espèces ioniques et moléculaires), le transfert simultané de masse et de chaleur, et les réactions chimiques. Les équations généralisées de Maxwell-Stefan sont utilisées pour quantifier les interactions multiconstituants lors de la diffusion. Le schéma réactionnel est intégralement pris en compte dans la phase liquide, et les réactions chimiques peuvent être cinétiquement contrôlées ou à l'équilibre chimique instantané. La séparation réactive en régime permanent est ainsi simulée avec une description rigoureuse des phénomènes de réaction-diffusion dans les films diffusionnels. Dans un deuxième temps, une installation pilote de captage du CO2 par une solution aqueuse de diéthanolamine est mise en fluvre pour la validation expérimentale du modèle. La diéthanolamine a une forte réactivité vis-à-vis du CO2; ceci engendre un profil important de concentration du soluté dans la colonne. Le pilote est dédié à la validation de modèle, il est donc équipé de plusieurs unités d'échantillonnage gaz et liquide à différentes hauteurs de garnissage. Les profils longitudinaux de la concentration du CO2 en gaz et en liquide, de l'humidité absolue et de la température liquide peuvent être établis expérimentalement et être comparés avec ceux provenant de la simulation. L'outil de simulation validé devient alors un outil de prédiction de l'efficacité des unités réelles de captage par l'absorption réactive. / The exhaust gases coming from the combustion of fossil fuels in power plants contain a wide variety of pollutants such as acid gases and can not be discharged directly into the atmosphere. These gaseous pollutants must be treated by postcombustion capture methods in absorption columns using chemical solvents. The objective is then to design a selective removal unit of acid-gases such as CO2, H2S and COS by using solvents of the alkanolamine family. This thesis develops as the first step, a non-equilibrium model, adapted to multicomponent electrolytic and reactive systems, for the representation of reactive absorption columns. The model includes modules to represent the thermodynamics on the basis of true species (ionic and molecular species), the simultaneous heat and mass transfer, and the chemical reactions. The generalized equations of Maxwell-Stefan are used to take into account the multicomponent interactions during diffusion. The reaction scheme is fully included in the liquid phase, and the chemical reactions can be kinetically controlled or at instantaneous equilibrium. The reactive separation at steady-state conditions is then simulated with a rigorous description of the reaction-diffusion phenomena in diffusional films. In the second step, a pilot plant of CO2 capture with an aqueous solution of diethanolamine is implemented for experimental validation of the model. Diethanolamine has high reactivity with respect to CO2; this generates an important concentration profile of solute in the column. The pilot plant is dedicated to model validation; it is equipped with several gas and liquid sampling units at different heights of packing. The longitudinal profiles of the gas and liquid CO2 concentration, the absolute humidity and the liquid temperature can be established experimentally and compared with those from the simulation. The validated simulation tool then becomes a tool to predict the effectiveness of real capture units by reactive absorption.