This thesis discusses the results and insights gained from developing a CFD model of a pilot-scale pressurized dry coal-feed system using the Barracuda CFD software and modeling various design concepts and operating conditions. The feed system was required to transport approximately 0.00378 kg/s (30 lb/hr) of pulverized coal from a vertical hopper to a 2.07 MPa (20.4 atm or 300 psi) reactor with a CO2-to-coal mass flow ratio of 1-2. Two feed system concepts were developed and tested for coal mass flow, CO2-to-coal mass ratio, steadiness, and uniformity. Piping system components also were evaluated for pressure drop and coal roping.With the first system concept, Barracuda software model parameters were explored to observe their effect on gas and particle flow. A mesh sensitivity study revealed there exists too fine of a mesh for dual-phase flow with Barracuda due to the particle initialization process. A relatively coarse mesh was found to be acceptable since the results did not change with increasing mesh refinement. Barracuda sub-model parameters that control particle interaction were investigated. Other than the close pack volume fraction, coal flow results were insensitive to changes in these parameters. Default Barracuda parameters were used for design simulations.The gravity-fed system (first concept) relied on gravity to transfer coal from a hopper into the CO2 carrier gas. This design was unable to deliver the required coal mass flow rate due to the cohesion and packing of the particles being greater than the gravity forces acting on the particles. The fluidized bed (second concept) relied on CO2 flow injected at the bottom of the hopper to fluidize the particles and transport them through a horizontal exit pipe. Additional CO2 was added post-hopper to dilute the flow and increase the velocity to minimize particle layout. This concept was shown to decouple the fluidized particle flow and dilution CO2 flow, providing significant design and operating flexibility. A non-uniform mesh was implemented to maintain a high mesh refinement in the 0.635-cm (¼-in) diameter transport pipe with less refinement in the hopper/bed region. The two main hopper diameters evaluated measured 5.08-cm (2-in) and 15.24-cm (6-in). Successful designs were achieved for each with appropriate coal mass flow rates and CO2-to-coal ratios. The particle flow was sufficiently steady for use with a coal burner.A piping system study was performed to test pneumatic transport and the effects of pipe length and bend radius. For a 1-to-1 gas-to-particle mass flow, particle layout occurred after 30 cm of travel. Particle roping occurred to various extents depending on the pipe bend radius. Bend radii of 0.318, 60.96, and 182.88 centimeters were simulated. Roping increased with bend radius and high pressure. Greater gas flow rates increased particle flow steadiness and uniformity. A simple methodology was identified to estimate the pressure drop for different piping system configurations based on the piping components simulated.

The world energy supply and demand, together with limited available resources have resulted in the need to develop alternative energy sources to ensure sustainable and expanding economies. Hydrogen is being considered a viable option with particular application to fuel cells. The Hybrid Sulphur cycle has been identified as a process to produce clean hydrogen (carbon free process) and can have economic benefits when coupled to nuclear reactors (High Temperature Gas Reactor) or solar heaters for the supply of the required process energy. The sulphur trioxide decomposition reactor producing sulphur dioxide for the electrolytic cells in a closed loop system has been examined, but it is clear that development with respect to a more durable active catalyst in a reactor operating under severe conditions needs to be investigated. A suitable sulphur trioxide reactor needs to operate at a high temperature with efficient heating in view of the endothermic reaction, and has to consist of special materials of construction to handle the very corrosive reactants and products. This investigation was undertaken to address (1) the synthesis, characterisation, reactivity and stability of a suitable catalyst (2), determination the reaction rate of the chosen catalyst with a suitable micro reactor (3) construction and evaluation of a packed bed reactor for the required reaction, and (4) the development and validation of a reactor model using computational fluid dynamics with associated chemical reactions. A supported catalyst consisting of 0.5 wt% platinum and 0.5 wt% palladium on rutile (TiO2, titania) was prepared by the sintering of an anatase/rutile supported catalyst with the same noble metal composition, synthesized according to an incipient impregnation procedure using cylindrical porous pellets (±1.7 mm diameter and ±5 mm long). Characterization involving: surface area, porosity, metal composition, - dispersion, - particle size, support phase and sulphur content was carried out and it was found from reactivity determinations that the sintered catalyst, which was very different from the synthesized catalyst, had an acceptable activity and stability which was suitable for further evaluation. A micro pellet reactor was constructed and operated and consisted of a small number of pellets (five) placed apart from each other in a two-stage quartz reactor with sulphur trioxide generated from sulphuric acid in the first stage and the conversion of sulphur trioxide in the second stage, respectively. Attention was only confined to the second stage involving the conversion of sulphur trioxide with the supported catalyst. The overall reaction kinetics of the pellets involving momentum, heat and mass transfer and chemical reaction was evaluated and validated with constants obtained from literature and with an unknown reaction rate equation for which constants were obtained by regression. As result of the complexity of the flow, mass and heat transfer fields in the micro pellet reactor it was necessary to use a CFD model with chemical reactions which was accomplished with a commercial code COMSOL MultiPhysics® 4.3b. A reversible reaction rate equation was used and a least squares regression procedure was used to evaluate the activation energy and pre-exponential factor. The activation energy obtained for the first order forward reaction was higher than values obtained from literature for a first order reaction rate (irreversible reaction) for the platinum group metals on titania catalysts. Detailed analyses of the velocity, temperature and concentration profile revealed the importance of using a complex model for determination of the reaction parameters. A fixed bed reactor system consisting of a sulphuric acid vaporizer, a single reactor tube (1 m length, 25 mm OD) heated with a surrounding electrical furnace followed, by a series of condensers for the analysis of the products was constructed and operated. Three process variables were investigated, which included the inlet temperature, the weight hourly velocity and the residence time in order to assess the performance of the reactor and generate results for developing a model. The results obtained included the wall and reactor centreline temperature profiles together with average conversion. As a result of the complexity of the chemistry and the phases present containing the products from the reactor a detailed calculation was done using vapour/liquid equilibrium with the accompanying mass balance (Aspen-Plus®) to determine the distribution of sulphur trioxide, sulphur dioxide, oxygen and steam. A mass balance was successfully completed with analyses including SO2 with a GC, O2 with a paramagnetic cell analyser, acid/base titrations with sodium hydroxide, SO2 titrations with iodine and measurement of condensables (mass and volume). The results obtained showed that a steady state (constant conversion) was obtained after approximately six hours and that it was possible to obtain sulphur trioxide conversion approaching equilibrium conditions for bed lengths of 100 mm with very low weight hourly space velocities. A heterogeneous 2D model consisting of the relevant continuity, momentum, heat transfer and mass transfer and the reaction rate equation determined in this investigation was developed and solved with the use of the commercial code COMSOL MultiPhysics® 4.3b with an appropriate mesh structure. The geometry of the packed bed (geometry) was accomplished by generating a randomly packed bed with a commercial package DigiPac™. The model predicted results that agreed with experimental results with conversions up to 56%, obtained over the following ranges: weight hourly space velocity equal to 15 h-1, temperatures between 903 K and 1053 K and residence times between 0.1 and 0.07 seconds. The post-processing results were most useful for assessing the effect of the controlling mechanisms and associated parameters. / PhD (Chemical Engineering), North-West University, Potchefstroom Campus, 2014

The world energy supply and demand, together with limited available resources have resulted in the need to develop alternative energy sources to ensure sustainable and expanding economies. Hydrogen is being considered a viable option with particular application to fuel cells. The Hybrid Sulphur cycle has been identified as a process to produce clean hydrogen (carbon free process) and can have economic benefits when coupled to nuclear reactors (High Temperature Gas Reactor) or solar heaters for the supply of the required process energy. The sulphur trioxide decomposition reactor producing sulphur dioxide for the electrolytic cells in a closed loop system has been examined, but it is clear that development with respect to a more durable active catalyst in a reactor operating under severe conditions needs to be investigated. A suitable sulphur trioxide reactor needs to operate at a high temperature with efficient heating in view of the endothermic reaction, and has to consist of special materials of construction to handle the very corrosive reactants and products. This investigation was undertaken to address (1) the synthesis, characterisation, reactivity and stability of a suitable catalyst (2), determination the reaction rate of the chosen catalyst with a suitable micro reactor (3) construction and evaluation of a packed bed reactor for the required reaction, and (4) the development and validation of a reactor model using computational fluid dynamics with associated chemical reactions. A supported catalyst consisting of 0.5 wt% platinum and 0.5 wt% palladium on rutile (TiO2, titania) was prepared by the sintering of an anatase/rutile supported catalyst with the same noble metal composition, synthesized according to an incipient impregnation procedure using cylindrical porous pellets (±1.7 mm diameter and ±5 mm long). Characterization involving: surface area, porosity, metal composition, - dispersion, - particle size, support phase and sulphur content was carried out and it was found from reactivity determinations that the sintered catalyst, which was very different from the synthesized catalyst, had an acceptable activity and stability which was suitable for further evaluation. A micro pellet reactor was constructed and operated and consisted of a small number of pellets (five) placed apart from each other in a two-stage quartz reactor with sulphur trioxide generated from sulphuric acid in the first stage and the conversion of sulphur trioxide in the second stage, respectively. Attention was only confined to the second stage involving the conversion of sulphur trioxide with the supported catalyst. The overall reaction kinetics of the pellets involving momentum, heat and mass transfer and chemical reaction was evaluated and validated with constants obtained from literature and with an unknown reaction rate equation for which constants were obtained by regression. As result of the complexity of the flow, mass and heat transfer fields in the micro pellet reactor it was necessary to use a CFD model with chemical reactions which was accomplished with a commercial code COMSOL MultiPhysics® 4.3b. A reversible reaction rate equation was used and a least squares regression procedure was used to evaluate the activation energy and pre-exponential factor. The activation energy obtained for the first order forward reaction was higher than values obtained from literature for a first order reaction rate (irreversible reaction) for the platinum group metals on titania catalysts. Detailed analyses of the velocity, temperature and concentration profile revealed the importance of using a complex model for determination of the reaction parameters. A fixed bed reactor system consisting of a sulphuric acid vaporizer, a single reactor tube (1 m length, 25 mm OD) heated with a surrounding electrical furnace followed, by a series of condensers for the analysis of the products was constructed and operated. Three process variables were investigated, which included the inlet temperature, the weight hourly velocity and the residence time in order to assess the performance of the reactor and generate results for developing a model. The results obtained included the wall and reactor centreline temperature profiles together with average conversion. As a result of the complexity of the chemistry and the phases present containing the products from the reactor a detailed calculation was done using vapour/liquid equilibrium with the accompanying mass balance (Aspen-Plus®) to determine the distribution of sulphur trioxide, sulphur dioxide, oxygen and steam. A mass balance was successfully completed with analyses including SO2 with a GC, O2 with a paramagnetic cell analyser, acid/base titrations with sodium hydroxide, SO2 titrations with iodine and measurement of condensables (mass and volume). The results obtained showed that a steady state (constant conversion) was obtained after approximately six hours and that it was possible to obtain sulphur trioxide conversion approaching equilibrium conditions for bed lengths of 100 mm with very low weight hourly space velocities. A heterogeneous 2D model consisting of the relevant continuity, momentum, heat transfer and mass transfer and the reaction rate equation determined in this investigation was developed and solved with the use of the commercial code COMSOL MultiPhysics® 4.3b with an appropriate mesh structure. The geometry of the packed bed (geometry) was accomplished by generating a randomly packed bed with a commercial package DigiPac™. The model predicted results that agreed with experimental results with conversions up to 56%, obtained over the following ranges: weight hourly space velocity equal to 15 h-1, temperatures between 903 K and 1053 K and residence times between 0.1 and 0.07 seconds. The post-processing results were most useful for assessing the effect of the controlling mechanisms and associated parameters. / PhD (Chemical Engineering), North-West University, Potchefstroom Campus, 2014

Significant maldistribution negatively influences performance of equipment containing dense tube bundles and, moreover, it can cause a wide range of operating issues. This thesis therefore focuses on analysis of fluid flow in complete distribution systems via computational fluid dynamics (CFD). Data obtained from simulations carried out using the software ANSYS Fluent were also statistically analysed. Influence of system arrangement, tube bundle parameters and operating parameters on flow distribution non-uniformity and pressure drop was investigated. According to the results, system arrangement is the crucial differentiating parameter in terms of flow distribution as well as pressure drop. Additionally, data obtained via the classical CFD approach were compared with those yielded by a simplified CFD model for three selected distribution systems. Simplified CFD approach can - given its low computational demand - be utilised in optimization algorithms as well as in the course of the initial stage of equipment design process. Furthermore, this thesis discusses a simulation tool which is based on the simplified CFD approach. Although this tool is still being developed, the results it yields are very promising.

Tato diplomová práce se zaměřuje na modelování proudění tekutiny v paralelních distribučních systémech, kde hraje důležitou roli rovnoměrnost distribuce tekutin. Pro tento účel je vytvořen vlastní CFD kód. Kód je napsán v programovacím jazyce Java a používá ke zlepšení výkonu knihovny třetích stran, které se vyznačují přímým přístupem ke hardwarovým a systémovým prostředkům. Kód se také vyznačuje určitými zjednodušeními, u nichž se očekává, že sníží výpočetní časy. Vliv použitých zjednodušujících opatření je vyhodnocen porovnáním výsledků simulací proudění na několika geometriích s údaji získanými z podrobných modelů CFD. Geometrie použitých svazků trubek se odlišují různými uspořádáními toku a trubek a také různým počtem trubek.

In considering methods of reducing the emission of carbon dioxide; there is a growing interest for use of wind power at domestic building in U.K. But the technology of wind turbines development in building environment is more complicated than in open areas. Small wind turbines in suburban areas have been reported as having unsatisfactory energy output, but it is not clear whether this is due to insufficient wind resource or low turbine efficiency. The aim of this research is to discover whether the wind resource in suburban areas is large enough for small wind turbines to produce a useful energy output.Historical wind data and manufacturers' turbine characteristics were used to estimate the hourly wind speed and energy output for different U.K. cities, terrain zones and turbines. It was found that for turbines at 10 m height in suburban areas and depending on city, the annual wind energy conversion efficiency ranged from about 20 to 40%, while the number of turbines required to produce the annual average electricity consumption of a UK dwelling ranged from about 6 for the smallest turbine (5.3 m² rotor area) to about 1 for the largest (35.26 m² rotor area).This analysis was based on average conditions, but the wind speed near buildings can vary considerably from one point to another. In order to predict the performance of wind turbines more accurately, the atmospheric boundary layer (ABL) of suburban areas was simulated in both CFD and wind tunnel models, and models of groups of semi-detached and terraced houses were set in this ABL. It was found that at 10 m height in the area of the houses, the turbulence intensity was too high for satisfactory operation of wind turbines (19 to 35%) while the mean velocity at different points ranged from 86 to 108% of the 10m reference velocity. At 30m height the turbulence intensity was satisfactory (less than 19 %), while the mean velocity ranged from 92 to 103 % of the 30 m reference velocity. It is concluded that for wind turbines in suburban areas, at 10 m height the wind speed is too low and the turbulence is too high for satisfactory performance, while at 30 m height the wind speed is much higher and the turbulence is low enough.

The importance of hydraulic structures has become an essential mitigating mean for floodsthat occur more often due to climate change. Thus, the importance and safety of flooddischarge tunnels has promoted further studies and experiments on the topic to mitigatedamages, such as cavitation that arise because of high speed flows.After an experimental study on a physical model was carried out on the flood discharge tunnelin Zipingpu Water Conservancy project, a CFD model was designed and simulated in thecommercial software ANSYS Fluent. The simulations aimed to evaluate and examine the riskfor cavitation in the tunnel, examine the design problems of the structure and analyse theinstalled aerators for the mitigation of cavitation. Moreover, using CFD models as acomplementary form to physical models was analyzed.A three dimensional geometry of the discharge tunnel was built in ANSYS Spaceclaim and themesh conducted with ANSYS mesh generator. The known boundary condition such as thedesign flow conditions, velocity inlet, pressure inlets and pressure outlet were set. For themodel a multiphase VOF scheme with RANS approach, k-ϵ turbulence model and a standardwall function was set.The results from the initial simulations showed that the discharge tunnel was under cavitationrisk, since the recorded cavitation index in the tunnel was below 1.8. After having revised thelayout of the aerators in order to mitigate cavitation risk, the results from the simulations withadded aerators were sufficient to mitigate the risk as the cavitation index was still below 1.8.The results for the cavitation index remained unchanged even in the simulated models with adifferent solver setup that were used in the comparison with the experimental data in order tovalidate them.As a conclusion, it was recommended that the tunnel design has to be revised and improvedby adding more aerators and air vents to mitigate the cavitation risk. Furthermore, more studieson the discharge tunnel or similar tunnels with similar conditions should be carried out in orderto validate the results of this study and determine if numerical models are preferable to physicalmodels

In order for wind-farm operators to deal with challenges regarding their fleet management, it is useful for them to estimate their units’ performance for different conditions. To perform such estimations, Computational Fluid Dynamics (CFD) may be used. This project focuses on the development of a CFD model for the aerodynamic analysis of wind turbine rotors, depending on their surface roughness. The work has been carried out in collaboration with the KTH Royal Institute of Technology and the Vattenfall AB R&D department. The open-source software OpenFOAM has been used to develop the desired model. A rigid body incompressible steady state, Reynolds-Averaged Navier-Stokes equations, k – ω SST CFD case has been set up. The NREL 5-MW rotor geometry has been used and the effect of four different surface roughness height values {1mm, 0.5mm, 100 μm, 30 μm} on its aerodynamic performance has been investigated for an incoming wind velocity of 10m/s. The referred roughness height values have been applied on the whole rotor surface. A 120° wedge type computational domain of unstructured mesh has been developed for the present simulations. The results indicate that a roughness-height increase leads to earlier flow separation over the blade suction side and increases the turbulent area of the boundary layer. That leads to a decrease for the extracted Torque and the Thrust force on the wind turbine rotor. Moreover, it is concluded that the rotor aerodynamic performance is more sensitive to low roughness heights rather than to high ones.

Le taux de pyrolyse est un paramètre clé du comportement du feu qui, à son tour, contrôle le transfert de chaleur à la surface du combustible. Dans cette étude, un modèle de pyrolyse volumique d’un combustible solide semi–transparent prenant en compte le rayonnement au sein du solide et la régression de l’interface a été couplé au code ISIS. Un algorithme génétique a été utilisé afin de déterminer un jeu optimal de paramètres cinétiques pour la pyrolyse du PMMA à partir d’une expérience de pyrolyse pure sous cône calorimètre en atmosphère normalement oxygénée. Des expériences de pyrolyse avec flamme de plaques carrées ont été réalisées afin de valider les résultats du modèle. L’analyse des résultats obtenus au centre de l’échantillon montre que la vitesse de régression de la surface devient constante en fonction du temps et que la contribution radiative au flux de chaleur total reste pratiquement constante. Les résultats du modèle couplé sont en bon accord avec la littérature et les mesures de cette étude. Néanmoins, le flux de chaleur incident aux bords de l’échantillon est sous-estimé. Une bonne concordance est obtenue entre les hauteurs de flammes prédites et celles déduites de la corrélation de Heskestad. Enfin, pour simuler la pyrolyse du PMMA en atmosphère sous–oxygénée, une chimie à deux étapes avec prise en compte du phénomène d’extinction de la flamme a été utilisée. Les résultats des simulations ont été comparés aux mesures réalisées dans le dispositif CADUCEE pour des fractions volumiques d’oxygène de 18,5% et 19,5%. La baisse du taux de pyrolyse et des températures de flamme avec la fraction volumique d’oxygène est bien reproduite par le modèle. / The pyrolysis rate is a key parameter controlling fire behavior, which in turn drives the heat feedback from the flame to the fuel surface. In the present study an in–depth pyrolysis model of a semi–transparent solid fuel with spectrally–resolved radiation and a moving gas/solid interface was coupled with the CFD code ISIS. A combined genetic algorithm/pyrolysis model was used with Cone Calorimeter data from a pure pyrolysis experiment to estimate a unique set of kinetic parameters for PMMA pyrolysis. In order to validate the coupled model, ambient air flaming experiments were conducted on square slabs of PMMA. From measurements at the center of the slab, it was found that the experimental regression rate becomes almost constant with time, and that the radiative contribution to the total heat flux remains almost constant. Coupled model results show a fairly good agreement with the literature and with current measurements. Nevertheless, the flame heat flux feedback at the edges of the slab is underestimated. Predicted flame heights based on a threshold temperature criterion were found to be close to those deduced from the correlation of Heskestad. Finally, in order to predict the pyrolysis of PMMA under reduced ambient oxygen concentration, a two–step chemical reaction and a flammability diagram for flame extinction was used. Model results are compared with data obtained in the experimental facility CADUCEE for ambient oxygen concentrations of 18.2 and 19.5%. Data show that the total mass loss rate and flame temperature decrease with the oxygen concentration, which is well reproduced by the model.

Le travail réalisé dans le cadre de cette thèse porte sur l'effet que les bâtiments exercent sur l'atmosphère urbaine et notamment sur les échanges énergétiques qui s'opèrent entre les deux systèmes. Afin de modéliser plus finement les effets thermiques du bâtiment sur les écoulements atmosphériques lors de simulations réalisées par le logiciel de CFD Code_Saturne, nous procédons au couplage de cet outil avec le modèle de bâtiment BuildSysPro. Cette bibliothèque fonctionne sous Dymola et peut calculer des matrices descriptives du bâtiment utilisables ensuite en dehors du logiciel. Ce sont donc ces matrices qui sont utilisées pour le couplage par l'intermédiaire d'un code assurant l'échange de données entre les calculs de thermique du bâtiment et ceux de CFD. Après une revue des phénomènes physiques en lien avec l'atmosphère urbaine et des modèles existants, nous nous intéressons aux interactions entre l'atmosphère et le milieu urbain, notamment les bâtiments. Ceux-ci peuvent avoir un impact sur les écoulements aussi bien dynamique, en tant qu'obstacles, que thermique, via leurs températures de parois. Parallèlement à la mise en place du couplage entre les deux logiciels, nous étudions les données de la campagne de mesures EM2PAU que nous utilisons pour notre validation. EM2PAU, réalisée en 2011 à Nantes, représente une rue canyon idéalisée par deux rangées de conteneurs. La campagne a pour spécificité de prendre simultanément les mesures de températures d'air et de parois ainsi que les vitesses du vent de référence et des écoulements dans le canyon par un anémomètre sonique placé à 10 m d'altitude et six autres positionnés en six emplacements dans le canyon. Nous cherchons donc à mettre en évidence les effets dynamiques et thermiques des bâtiments sur les écoulements à partir des résultats de cette campagne, pour ensuite les simuler. Puis la modélisation numérique des écoulements sur le domaine de EM2PAU est réalisée. L'objectif de ce travail est de mettre en évidence l'influence des effets thermiques des parois sur les flux atmosphériques. Nous comparons des simulations avec différentes méthodes pour donner les valeurs des températures de surface des conteneurs. La première méthode consiste à imposer ces températures d'après les mesures ; ainsi la température de chaque paroi sera fixée à la température de surface mesurée lors de l'instrumentation de EM2PAU. Quant à la deuxième méthode, on impose la température de l'air extérieur mesurée à l'instant simulé à toutes les parois des conteneurs, afin de créer un cas où l'on n'observe que peu ou pas d'échanges de chaleur. Enfin la troisième méthode est la simulation couplée de Code_Saturne et BuildSysPro. Les résultats des différentes simulations sont alors comparés afin de distinguer les effets thermiques des parois des bâtiments sur les écoulements d'air. Nous observons que les effets dynamiques sont primordiaux et peuvent engendrer des vitesses verticales de l'écoulement dans le canyon de l'ordre plusieurs mètres par seconde, tandis que des écarts de températures de surface de l'ordre de 15°C peuvent modifier les vitesses verticales du vent de moins de 0, 5 mètres par seconde. Si ces effets thermiques sont difficiles à isoler sur des mesures en raison des autres phénomènes susceptibles d'influencer les écoulements atmosphériques, les études numériques peuvent toutefois mieux quantifier ces différences / This thesis work is about the effect of buildings on the urban atmosphere and more precisely the energetic exchanges that take place between these two systems. In order to model more finely the thermal effects of buildings on the atmospheric flows in simulations run under the CFD software Code_Saturne, we proceed to couple this tool with the building model BuildSysPro. This library is run under Dymola and can generate matrices describing the building thermal properties that can be used outside this software. In order to carry out the coupling, we use these matrices in a code that allows the building thermal calculations and the CFD to exchange their results. After a review about the physical phenomena and the existing models, we explain the interactions between the atmosphere and the urban elements, especially buildings. The latter can impact the air flows dynamically, as they act as obstacles, and thermally, through their surface temperatures. At first, we analyse the data obtained from the measurement campaign EM2PAU that we use in order to validate the coupled model. EM2PAU was carried out in Nantes in 2011 and represents a canyon street with two rows of four containers. Its distinctive feature lies in the simultaneous measurements of the air and wall temperatures as well as the wind speeds with anemometers located on a 10 m-high mast for the reference wind and on six locations in the canyon. This aims for studying the thermal influence of buildings on the air flows. Then the numerical simulations of the air flows in EM2PAU is carried out with different methods that allow us to calculate or impose the surface temperature we use, for each of the container walls. The first method consists in imposing their temperatures from the measurements. For each wall, we set the temperature to the surface temperature that was measured during the EM2PAU campaign. The second method involves imposing the outdoor air temperature that was measured at a given time to all the surfaces, reducing every heat exchange to almost zero. The third method at last is the coupled simulation of Code_Saturne and BuildSysPro where BuildSysPro calculates the wall temperature from the Code_Saturne data. . The results of these different ways of modelling the wall temperatures are then compared in order to show the thermal effects of building wall heating on the air flows. We notice that the dynamic effects are dominant and can generate vertical wind speed that can pass several meters per second. On the other hand, differences of surface temperatures higher than 15°C can influence the vertical wind speed for less than 0.5 meters per second. These thermal effects are not easily highlighted with measured data because of the other phenomena that can impact the air flows. However they can be quantified with numerical studies