Advanced Building Simulation

It was found that the differences are much larger in terms of airflow than in terms
of air temperatures. The temperature differences between the various methods
increases with the number of stacked zones.
The main conclusion from the case study is that the coupled solution method will
be able to generate accurate results, even with simulation time steps of 1h. Reducing
the time step will increase the computing resources used considerably, with a relatively
small improvement of the accuracy.
For equal length of time steps a coupled solution method will use more computer
resources than a decoupled solution.
For the decoupled method, it is necessary to reduce the time step to ensure the accuracy.
For the current case study, the decoupled solution method using a simulation
time step of 360s was less accurate than the coupled solution method with a time step
of 1h. However, the computer resources used were more than doubled.
Based on the current case study, it may be concluded that the coupled solution gives
the best overall results in terms of both accuracy and computer resources used.
Although the results presented here are for an imaginary (but realistic) building, the
observed trends may be expected to be more generally valid.
4.4 Quality assurance
Integrated building airflow simulation 107
Due to lack of available resources it usually has to be assumed in a practical design
study context that the models and the simulation environment, which is being used,
has been verified (i.e. the physics are represented accurately by the mathematical and
numerical models) and validated (i.e. the numerical models are implemented correctly).
Nevertheless, it is critically important to be aware of the limitations of each
modeling approach.
For example, when using the network approach it should be realized that most of
the pressure–flow relationships are based on experiments involving turbulent flow.
Von Grabe et al. (2001) demonstrate the sensitivity of temperature rise predictions in
a double-skin façade, and the difficulty of modeling the flow resistance of the various
components. There are many factors involved but assuming the same flow conditions
for natural ventilation as those used for mechanical ventilation causes the main problem,
that is using local loss factors � and friction factors from mechanical engineering
tables. These values have been developed in the past for velocities and velocity
profiles as they occur in pipes or ducts: symmetric and having the highest velocities
at the center. With natural ventilation however, buoyancy is the driving force. This
force is greater near the heat sources, thus near the surface and the shading device,
which will lead to nonsymmetric profiles. This is worsened because of the different
magnitudes of the heat sources on either side of the cavity.
One way forward would be to use CFD in separate studies to predict appropriate
local loss factors � and friction factors for use in network methods. Strigner and
Janak (2001) describe an example of such a CFD approach by predicting the aerodynamic
performance of a particular double-skin façade component, an inlet grill.
However, as indicated earlier, CFD is still very much being developed. At the same
time it seems to be very appealing to engineers and clients; the CFD � colors for
directors effect? Therefore it is essential that quality assurance procedures such as by
Chen and Srebric (2001) will be developed.