Bachelor's thesis Jan Trosdorff
Stochastic User Behavior Models for Non-Residential Buildings: Analysis, Implementation and ValidationCopyright: EBC
Thermal building simulations can be used during the planning stage of buildings, in order to achieve an ideal dimensioning considering the energetic and economical aspects.
Deterministic models with a specified presence are often used in such simulations to consider the occupant. However, these models are not able to simulate the random nature of occupant behavior regarding his actions. A deterministic profile can not describe the unique behavior of every single occupant. Therefore stochastic models are used to show more realistic occupant behavior.
In this thesis it is investigated based on a literature review which model is suitable to generate occupancy profiles as input to the thermal simulation of non-residential buildings. The stochastic model developed by Page is considered the most suitable model. It is implemented by using Python programming language and is later validated using monitoring data from the E.ON ERC main building.
This work’s main aspect is the generation of the occupancy profiles considering data from measurements. Hence, several methods are added to the model to provide necessary data to calibrate the simulation. Additionally a new method for simulating long absences is proposed. The validation of the long absence method is not conducted in this thesis due to a lack of monitoring data.
The validation of the model proposed by Page proves the ability of the model to create a time series of presence data with only little information of the framework. The created profiles show the same characteristics as the measured profiles for occupant presence. The occurrence of state changes in pairs, as a typical property of non-residential buildings, is confirmed by both, monitoring and simulation data.