• Simon Pommerencke Melgaard
  • Ivan Titov Nikolaisson
The main focus of this project is how to model a double skin facade (DSF) with
computational fluid dynamics (CFD). Through an extensive literature review it was found
that there is no solid agreement on the topic of how to model a DSF with CFD. The
common trends regarding CFD modelling are investigated, such as; the use of turbulence
viscosity models (TVM), wall functions, radiation modelling, boundary conditions and
flow properties.
Due to this lack of agreement in the scientific community, a set of CFD models are simulated
and compared against experimental data. A DSF is simulated with 2D and 3D geometries
with various TVMs and radiation models. The results from the simulations are compared
against previous work, as well as experimental measurements.
The experimental measurements are performed using a particle image velocimetry (PIV)
setup, together with temperature measurements. The measurements are performed on
a naturally driven, but mechanically controlled wide DSF, under lab conditions. From
the results of the measurements various interesting phenomena are observed. The
possible occurrence of backflow in the outlet is documented. Additionally, inside the
DSF a stagnation point is found. This stagnation point appears to be in steady state.
Furthermore, the results indicate that the flow in certain areas of the DSF is transient.
This transient behaviour is indicated by the formation of vortexes inside the DSF.
From the CFD simulations it is concluded that the k-ε TVM family gives the best results.
The CFD simulations are able to predict the temperature field relatively well. The velocity
field is predicted well in the bottom region, but is overpredicted in the top of the DSF,
when compared with experimental data. In terms of radiation modelling, it is found that
simulating with radiation gives better results than with no radiation. It is furthermore
concluded that an enhanced wall function must be used in DSF CFD simulations, when
using the k-ε TVMs.
It is recommended that the backflow and the transient phenomena are further investigated,
since they might explain the overprediction from the CFD simulations. Additionally, the
stagnation point is an interesting phenomenon that has the potential of being a CFD
validation tool for DSFs
Publication date2019
Number of pages173
ID: 305256131