To be able to handle tomorrows need for limited energy consumption we need to reduce our use of energy. The building sector stands for around 40 % of all energy consumption in the society. The government has put up a goal to reduce the energy consumption in our buildings with 20 % by year 2020 and 50 % by year 2050 compared with year 1995. To be able to do reach that goal we need a more energy efficient building stock. The main part of the energy used in our buildings is used for space heating. By installing ventilation systems with heat recovery on the exhaust air it is possible to use the heat-energy in the exhaust air to warm up the incoming air. This can contribute to a reduction in energy use. A ventilation system with heat recovery on the exhaust air is space demanding and there can be problems with finding enough space to do the installation indoors. Therefore it can be an advantage to place the aggregate and the ducts on the outside of the buildings climate shell. A placement exterior of the buildings climate shell or in an unheated space leads to thermal heat losses. The aim with this essay is to investigate how significant the heat losses are on exterior placed ventilation systems. The investigation has been done with help of theoretical calculations and measurements of the temperature difference in the ventilation ducts. Analysis has been made on life cycle costs on how to reduce the heat losses in an economic manner. To buildings, Höstvägen 14 and 22 in Växjö, which have been equipped with exterior placed ventilation systems have been studied. The two buildings have two different types of installation of the ducts. Our result shows that the heat losses through the ventilation systems on Höstvägen 14 and 22 are significant. The majority of the losses occur in the ducts. In the aggregate the thermal bridges in the framework accounts for the larger part.

L’objectif de cette thèse est de modéliser en temps réel la propagation d’un incendie dans des ensembles comportant un grand nombre de locaux. Un modèle semi-physique de réseau polydisperse amorphe prenant en compte les connexions à courte et longue distances entre sites, est proposé. Les phénomènes physiques liés au développement du feu dans un local et à sa transmission entre locaux par les parois sont simulés par des lois normales de probabilité. Les durées moyennes de transmission par les parois sont déterminées à l’aide d’un modèle à zones prenant en compte les spécificités du local en feu. Des expérimentations spécifiques dans un caisson en acier, représentatif d’un local de la Marine Nationale, ont permis de valider le modèle à zones. Un exemple détaillé du calcul par le modèle de réseau de la propagation d’un feu dans une maquette de navire à échelle un est ensuite décrit et analysé pour différents scénarios. Une analyse de sensibilité utilisant un plan factoriel complet à deux niveaux permet de hiérarchiser les paramètres du modèle et d’étudier la sensibilité de la solution aux variations de ces paramètres. Une étude statistique est conduite afin d’établir une cartographie du risque incendie à bord du navire. La transmission du feu par les gaines de ventilation est simulée par une loi normale de probabilité où la durée moyenne de transmission est déterminée à l’aide d’un code à champ unidimensionnel. Après avoir validé ce code sur des mesures obtenues par DGA dans une conduite cylindrique différentiellement chauffée, l’influence de ce mode de transmission sur la propagation du feu dans le navire est analysée. / This thesis work is devoted to the development of a semi-physical network model to predict real-time fire spread in polydisperse amorphous massively multi-compartmented spaces. This model takes into account short-range and long-range connections between adjacent and remote network sites. The physical phenomena of fire ignition and flashover, and of fire transmissions through the walls are simulated using time-dependent normal probability distributions. Mean durations of transmission though the walls are determined by a two-zone model which takes into account the fuel load, the room size and the thermal properties of walls. Specific experiments were conducted in a steel room, representative of a naval vessel compartment, in order to validate the zone model. Then a proof of concept is developed by applying the network model to different fire scenarios in a full-scale vessel mockup. A sensitivity analysis using a two-level full factorial design is performed to identify the most influential model parameters and to evaluate the sensitivity of the solution to variations of these parameters. A statistical study is conducted to produce fire risk maps. Finally, a special emphasis is put on the fire transmission by the ventilation ducts. This phenomenon is simulated using a time-dependent normal probability distribution where the mean duration is determined by means of a one-dimensional CFD model. This model is first validated using data obtained by DGA in a differentially heated duct and second, the influence of fire transmission through ventilation duct on its propagation throughout the vessel is investigated.