This thesis presents an experimental and theoretical study of the dynamic
response of convectively heated buildings and their respective space heating peak
demands for different room temperature set point profiles and thermal mass levels,
with a focus on the impact of thermal model resolution on the peak demand
calculation.
Experiments were conducted at two identical and highly instrumented houses.
One house is modified with different 
oor coverings, while the other is kept unchanged
and used for reference. Through experimentation and simulation, peak
power (due to space heating) reduction strategies are investigated.
Twelve equivalent RC thermal network models of varying model resolution
are developed for a north zone of the houses. Modelling approximations including
linearization of the heat transfer, spatial and/or temporal discretization and approximations
for reduction in model complexity are implemented into the models
and their effects are investigated. The focus is on simple and physically meaningful
building thermal models suited for model-based control.
The models are used to study the impact of set point ramping lengths and
\near-optimal" transition curves between two temperatures on peak demand reductions
for a very cold day. Alterations to walls and ceilings in the models
were done to hypothetically modify their properties in the zone and the effects
in combination with ramping profiles were analyzed. This work can inform the
development of new building materials.
A commercial building is also considered and two low order RC thermal network
are compared. The first model excludes the mass of the interior partitions,
while the second model incorporates them. An advantage of the model with interior
partitions is it can be used for retrofit studies.