Occupant behaviour in buildings: Thermal performance implications of window use patterns

The longitudinal field studies of thirteen offices with different ventilation methods, including daytime only and day and night-time natural ventilation strategies and air-conditioning with operable windows, have been carried out in the summer of 2006 and 2007.   The physical monitoring variables include the occupant use of individual windows and indoor and outdoor thermal conditions with a high time resolution (i. e. observing the state of individual windows every half second).   The field studies have given new insights into the relationship between building design, occupant behaviour of window-control and the thermal performance of buildings. This dissertation has developed probabilistic window-control models for daytime only and day and night-time naturally ventilated offices from logistic regression analysis of the extensive monitoring data (i. e. more than 30,000 data points).   The models predict the transition probability of window states from open to closed and from closed to open as a function of (i) indoor temperature, (ii) the previous state of a window, (iii) the time of day and (iv) building design.   These models have been incorporated into a combined behaviour algorithm of probabilistic occupant behaviour and deterministic heat and mass balance models that is subsequently implemented into ESP-r, a European standard building simulation tool.   Markov chains and Monte Carlo methods were applied to the algorithm, so that it allows us to reflect more realistic window-control patterns in dynamic building simulation tools. Passive ventilation by occupant window-control behaviour could enable significant reductions in cooling energy requirements and CO2 emissions from the building sector by reducing the use of air-conditioning in buildings.