Net ZEB Design Processes and Tools - Modeling, Design, and Optimization of Net-Zero Energy Buildings - page 196

Fig. 4.12 Full factorial and fractional factorial design
A further advantage to decoupling models, where it is technically sound, is
a reduction in simulation time. If the simulation is based on simultaneously
solving equations by inverting a matrix, then the computational time to
invert the matrix is proportional to X 3 , where X is the number of rows in the
matrix.
In certain situations, it may be appropriate to simulate different building
phenomena in two or more different tools sequentially. This is permissible
if multiple phenomena do not interact dynamically yet one may affect
other(s). This allows the most suitable tool to be chosen for modeling each
phenomenon. For example, it is usually assumed for indoor air quality
(IAQ) modeling that contaminants are trace elements, meaning that their
mass is very small compared to the total mass of the air in a building.
Therefore, the airflow rates and pressures may be resolved first and then
this information, after the entire simulation has run, can be provided to the
contaminant model. The caveats to this are the following: (a) the tools must
be able to read from and write to a file that is recognized by both (e.g., a
simple text file), and (b) the assumption of unidirectional data flow must be
acceptable. In the IAQ and airflow example, sequential simulation does not
allow demand controlled ventilation (DCV) to be modeled because for such
buildings the ventilation rate depends on the carbon dioxide concentrations
and vice versa.