We consume to much energy, which in turn causes excessive emissions of CO2. Reducing the consumption of energy is therefore one of the great challenges of our time. Heating and cooling of buildings account for approximately one fifth of the total energy consumption within the EU. Hence, the potential for financial and environmental savings is large.
By using mathematical models, which describes the thermal properties of a building, we can reduce the consumption of energy by optimizing the comfort temperature only when the building is in use. In addition, such models can be used to estimate and compare the thermal properties of a building in a more precise manor then the current energy classification systems.
The challenge of modeling buildings is their inherent complexity, being made from a large number of different materials and construction techniques. Additionally, heating of buildings depends on the weather conditions and the human occupants. This makes modeling of buildings a complex task. A possible solution is the use of simplified models, so called "thermal networks". These are simplified mathematical models that must be adapted (calibrated) to each specific building.
The thesis has studied the challenges linked with the use of these simplified models, and how measurement data from sensors can be used in the calibration of these models. The result of the work includes studying how the calibration of models can be quality ensured, such that they can be used to estimate the thermal properties of a given building. This in turn makes it possible to use the calibrated models as tools for more optimal control of temperature, thereby saving energy, money and reducing emissions of CO2. In addition, the thesis provides some answers as to what it takes for this kind of models to be usable for performing energy classification of an existing buildings.