Journal of Building Performance Simulation

ISSN: 1940-1493 (Print) 1940-1507 (Online) Journal homepage: https://www.tandfonline.com/loi/tbps20

A cylindrical meshing methodology for annual

urban computational fluid dynamics simulations

Patrick Kastner & Timur Dogan

To cite this article: Patrick Kastner & Timur Dogan (2020) A cylindrical meshing methodology
for annual urban computational fluid dynamics simulations, Journal of Building Performance
Simulation, 13:1, 59-68, DOI: 10.1080/19401493.2019.1692906

To link to this article: https://doi.org/10.1080/19401493.2019.1692906

Published online: 18 Dec 2019.

Submit your article to this journal

View related articles

View Crossmark data

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=tbps20
JOURNAL OF BUILDING PERFORMANCE SIMULATION
2019, VOL. 13, NO. 1, 59–68
https://doi.org/10.1080/19401493.2019.1692906

A cylindrical meshing methodology for annual urban computational fluid dynamics

simulations
Patrick Kastner and Timur Dogan
Environmental Systems Lab, Cornell University, Ithaca, NY, USA

ABSTRACT ARTICLE HISTORY

For urban CFD simulations, it is considered a best practice to use a box-shaped simulation domain. Received 6 January 2019
Box-shaped domains, however, show drawbacks for airflow from several wind directions as remesh- Accepted 8 November 2019
ing and additional preprocessing steps become necessary. We introduce a routine to create a cylin- KEYWORDS
drical mesh that expedites the simulation of arbitrary wind directions using OpenFOAM. Results com- CFD; urban; meshing;
puted with the cylindrical domain are validated against wind tunnel data. We report that the cylin- box-shaped; cylindrical;
drical method yields comparable results in terms of accuracy and convergence behaviour. Further, multi-directional
run time comparisons in a real-world scenario are conducted to discuss its advantages and limita-
tions. Based on the findings, we recommend using the cylindrical approach if at least eight wind
directions are analyzed for which we report 18% run time savings. The cylindrical domain along
with automated best practice boundary conditions has been implemented in Eddy3D – a plugin for
Rhinoceros.

1. Introduction
geometries, and if surrounding buildings exist, further attention
Urbanization and population growth, along with a massive pre- is needed to avoid geometric oversimplification (Cheung and
dicted construction volume can be seen as a unique oppor- Liu 2011). This is especially important when it comes to annual
tunity to improve the built environment and its quality of analyses, that is, analyses for all hours of the year, for which
living through integrated and well-informed architectural urban multi-directional CFD simulations are needed. In these multi-
design processes. Such processes lead to high quality, climate- directional conditions, computational fluid dynamics (CFD) anal-
adaptive architecture that uses passive means to provide com- yses are used, either as a standalone tool or to inform the AFN
fortable environments with smaller carbon footprints. Not only method with more accurate pressure coefficient (cp ) input data.
in Mediterranean countries but also in countries with subtropi- CFD is a numerical methodology to calculate desired flow vari-
cal and tropical climates where the largest construction volumes ables on a lattice within a simulation domain by solving the
are expected, natural ventilation (NV) is one of the most effi- Navier-Stokes equations (NSE). The usual steps of a recurring
cient ways of cooling and promises a significant energy saving CFD analysis in a design process for the built environment con-
potential. In such regions, studies have shown the possibility sist of:
of saving up to 50% in energy compared to air conditioning
(Cardinale, Micucci, and Ruggiero 2003; Oropeza-Perez and (1) modelling the building geometry with CAD software;
Østergaard 2014). (2) meshing the building geometry and topography;
To evaluate such savings, building energy modelling (BEM) (3) simulating the problem with appropriately-assigned bound-
packages like EnergyPlus and TRNSYS come with capable air- ary conditions for multiple wind directions;
flow network (AFN) solutions for natural ventilation evaluation (4) postprocessing the variables of interest, likely followed by
in multi-zone building energy models (Huang, Winkelmann, and design alterations that lead back to (1) if goals or constraints
Buhl 1999; Weber et al. 2003). These solutions rely on pres- have not been met.
sure coefficient arrays for different wind directions and exte-
rior simulation nodes. For simple box-shaped buildings without The expertise needed to perform such analyses and the asso-
contextual obstructions, look-up tables and fast methods for ciated preprocessing overhead often impedes a wider use of
surface-averaged pressure coefficient generation exist (Swami natural ventilation studies in building and urban scale design
and Chandra 1988; Grosso 1992). Since then, many attempts workflows. As a result, NV studies are expensive and usually
have been made to deal with airflow sheltering effects for sim- are only undertaken late in the design process when significant
plified urban geometries, and there is evolving literature about changes to improve the NV potential are often no longer feasi-
wind pressure coefficients for sheltered buildings that are sum- ble. Hence, to incorporate natural ventilation analyses into early
marised by Costola, Blocken, and Hensen (2009). For unique design stages, the workflows for annual wind analysis need to be

CONTACT Patrick Kastner [email protected]

© 2019 International Building Performance Simulation Association (IBPSA)

60 P. KASTNER AND T. DOGAN

Figure 1. Collocation scheme of the investigated approach within the urban simulation process.

(1) streamlined, and (2) the time to produce actionable results domain dimensions, convergence criteria, and relaxation fac-
needs to be reduced. tors. Little focus, however, has been put on how to best
In this study, we propose a novel methodology to reduce the approach the meshing of a simulation domain for annual wind
preprocessing and simulation time of annual urban wind sim- simulations.
ulations by utilising the CFD library OpenFOAM. For this, we For urban CFD simulations that take into account a single
use a cylindrical simulation domain that allows for a seamless wind direction, it is considered best practice to construct a box-
assignment of boundary conditions for arbitrary wind directions. shaped wind tunnel (WT) with predefined dimensions in relation
Further, we discuss meshing issues that might occur when cylin- to the building geometry. A commonly-used practice proposed
drical meshes for urban CFD simulations are used. We introduce by Tominaga et al. suggests a simulation domain size of z = 6 ·
a robust workflow that expedites annual wind flow analyses Hmax , l = 20 · Hmax and w given a blocking ratio of ≤ 3%, where
that are relevant for applications shown in Figure 1. We investi- z, l, and w are the dimensions of the simulation domain and
gate the convergence behaviour, accuracy, and run times in two Hmax is the height of the tallest building in the building agglom-
studies: eration. The blocking ratio is defined as the ratio between the
projected facade area of the buildings perpendicular to the inlet
(1) In a validation study, we conduct a grid convergence study to the total area of the inlet in wind direction.
for four stages of mesh refinement. Then, we evaluate There are two approaches to account for different wind direc-
the accuracy and convergence behaviour of both the box- tions: one may either rotate the building geometry that is placed
shaped and cylindrical domains for a reference case by Jiang in the simulation domain (Figure 2(a)), or set up an entirely new
et al. (2003) from the previous grid convergence study.
(2) In a more detailed annual study, we use a square-shaped
building array to assess the differences to complete both
meshing and simulation referred to as run time.

2. Background
When it comes to spatial analysis and decision aiding for the
built environment with CFD, particular attention should be paid
to meshing methodologies. Several best practice guidelines for
urban flow problems have been published over the years (Franke
et al. 2004; Franke 2006; Franke and Baklanov 2007; Tominaga
et al. 2008; Franke et al. 2010; Ramponi and Blocken 2012;
Blocken 2015), all of which propose best practices concerning Figure 2. Top view of simulation domains to account for different wind directions.
 JOURNAL OF BUILDING PERFORMANCE SIMULATION 61

simulation domain with corresponding boundary conditions for more involved problems (many directions with surrounding
each additional wind direction (Figure 2(b)). In each subfigure, urban context) become increasingly complex to handle.
the dashed lines represent the wind tunnels and wind directions To avoid the creation of a new mesh for every wind direction,
after an arbitrary rotation. we propose a cylindrical computational mesh that allows for a
Both options come with disadvantages. In the first case, streamlined simulation of arbitrary wind directions, see Case C
the entire simulation domain needs to be remeshed for every and C’ in Figure 3. Further, the setup of boundary conditions is
additional wind direction. However, as the height of the WT automated such that a significant amount of time will be saved
depends on a constant Hmax , a change in the windward-facing for the setup of an annual wind analysis (up to 32 or more wind
projected facade area introduced by rotation, results in a dif- directions). More specifically, every lateral cylinder patch rep-
ferent width of the WT. In manual preprocessing setups, width resents an individually defined angular increment and can be
adjustments for adequate blocking ratios are often neglected. assigned to either an inlet or outlet boundary condition depend-
As a result, the same WT width is used for all wind direc- ing on the particular wind direction (Figure 4). Moreover, the
tions, which may lead to convergence problems (Figure 2(a)). angular increment determines the block size of the inner rectan-
For the second approach shown in Figure 2(b), the coordi- gle. This makes the same mesh reusable for the simulation of any
nates of the WT and the boundary conditions need to be arbitrary wind direction. Case A and A’ in Figure 3 illustrate the
adjusted to account for changes in wind directions, thus likely box-shaped wind tunnel, whereas Case B/B’ and C/C’ illustrate a
violating the best practice dimensions for the WT width if square-shaped and cylindrical simulation domain, each of them
omitted. Hence, box-shaped simulation domains show draw- with best practice dimensions imposed. Little is known about
backs when many wind directions need to be simulated because cylindrical meshes for this purpose, hence, a thorough study
of the given climate data. While (re)meshing of single, exposed of accuracy, convergence behaviours, and actual computational
building geometries for few wind directions is manageable, speed gains are required.

Figure 3. Top row: top views of the simulation domain for an arbitrary urban area; Case A shows a conventional simulation domain, Case B shows a square-shaped
simulation domain, and Case C shows a cylindrical simulation domain, each for a wind direction of 0◦ . Bottom row: Cases A’-C’ for wind directions of 45◦ . The illustrations
are not true to scale.
62 P. KASTNER AND T. DOGAN

Figure 4. Top view of the lateral cylinder patches. A coarse setting is shown left, a substantially finer mesh is shown on the right.

3. Methodology
3.1. Validation study
This section consists of two analyses: first, we conducted a
grid independence study for a box-shaped domain for a sin-
gle wind direction and a single building geometry. With the
sufficiently-refined grid, we compared its accuracy and con-
vergence behaviour to a mesh produced with the cylindrical
approach.
Jiang et al. (2003) conducted an extensive study in an atmo-
spheric boundary layer (ABL) WT in which a cuboid with two
openings had been investigated experimentally to estimate the
cross-ventilation behaviour (Figure 5).
The geometry in Figure 5(a) was modelled in Rhinoceros with
infinitely thin walls, neglecting the 6 mm wall thickness of the
scale model. The Grasshopper plugin Eddy3D was used to auto-
mate the preprocessing, including the assignment of bound-
ary conditions. The mesh was created by using the blockMesh
utility for the background mesh and snappyHexMesh to subse-
Figure 6. 3D visualisation of box-shaped wind tunnel (WT) and cylindrical WT with
quently snap the background mesh to the building geometry, dimensions.
producing a mixed polyhedral mesh. The dimensions of the box-
shaped simulation domain are (5.75 · 2.75 · 1.5) m. When using
a cylindrical mesh, while ensuring not to violate the best prac- lateral patches of the cylindrical domain consist of 5◦ straight line
tice dimensions for length, width, and height, the ground area of segments.
the mesh is larger. It is characterised by a higher cell count than The domain inlet was set to an atmospheric boundary
one would anticipate with the conventional approach shown in layer profile for U, k, and ω. At the outlet of the computational
Figure 3 for Case A. To ensure a fair comparison between both domain, constant pressure is assumed, while the other variables
approaches, we created the blockMesh with identical cell sizes are assumed to be zero-gradient. The ground and the build-
in the areas where the buildings were placed and used identical ing walls use the same boundary conditions: a no-slip condition
mesh refinement levels. Figure 6 illustrates both the box-shaped for velocity, a zero-gradient condition for the pressure, and wall
and the cylindrical domains considering best practice guide- functions for k and ω. For νt , the intelligent wall function called
lines as they were created in Grasshopper. In this example, the nutUSpaldingWallFunction was used. The front, back, and top

Figure 5. (a) Schematic of the validation wind tunnel geometry; (b) Vertical section through validation domain with geometry spanning from 0 to h.
 JOURNAL OF BUILDING PERFORMANCE SIMULATION 63

Table 1. Turbulence boundary conditions used for the running Windows 10. We used the Docker Version 17.12.0-ce-
validation study.
win47 (15139) to run OpenFOAM 4.1. At most, we ran a maxi-
Parameter Value mum of four OpenFOAM instances at a time on single CPUs on
k 0.03456 separate threads to avoid affecting the run times by other pro-
 0.0835 cesses. All other values including the discretization schemes not
ω 10.42
specifically mentioned here were selected according to current
best practice guidelines (Franke 2006).
To compare the OpenFOAM results of both meshing method-
faces are set to symmetric boundary conditions for all variables.
ologies against the WT data, measurements by Jiang et al. (2003)
The kinematic viscosity, ν, was set to 1.5 × 10−5 . The turbulence
were digitised and subsequently interpolated to yield 50 sam-
inlet parameters were calculated using the following equations:
pling points. The axes were normalised by h = 0.25 m and uref =
12 m s−1 . For later comparison, vertical, stream-wise velocity
(U∗ )2
k= √ (1) measurements were taken at h/2, as the largest deviation from
Cmu the measured data was found there, see Figure 5. First, the results
(U∗ )3 for the refinement study were sampled with the sample utility
ε= (2)
κ(z + z0 ) using the cellPoint interpolation scheme in OpenFOAM, which is
ε a linear-weighted interpolation using cell values. The sampled
ω= (3) results were then plotted against the experimental data and the
Cmu · k
coefficient of determination (R2 ) was calculated:
where U∗ is the friction velocity, and Cmu = 0.09 is a constant N
for the turbulence model. The values used are summarised in (yi − ŷi )2
2
R = 1 − i=1
N
, with 0 ≤ R2 ≤ 1 (6)
Table 1.
i=1 (yi − ȳi )2
The approach to model the ABL in OpenFOAM is based on the
following equations (Wallace and Hobbs 2006): To verify and report grid independence, refined meshes were
  created by increasing the number of divisions of the background
U∗ z + z0 mesh by factors of two in each Cartesian direction while keeping
U= ln (4)
κ z0 the levels of surface and feature refinements constant. To pro-
Uref vide a standard and consistent approach to report the results
U∗ = κ (5) of grid convergence studies (GCS) and error estimations, we
ln ((zref + z0 )/z0 )
adopted the concept of the grid convergence index (GCI) by
where U∗ is the friction velocity, κ is the von Karman constant, Roache (1994) which is based on a derived variable. The derived
Uref is the reference velocity at the reference height zref , and variable, in our case, is the volumetric flow rate through the
z0 is the aerodynamic roughness length. In the original experi- opening. The GCI measures the percentage that the computed
ment, the atmospheric boundary layer profile of the WT’s inlet value is deviating from the asymptotic numerical value which is
velocity was created by placing Lego Duplo blocks on the wind- to be interpreted as an error band. In other words, it measures
ward side of the scale model. Unfortunately, no visual docu- how much the solution would change by further refining the
mentation is provided to estimate the size of the resulting z0 , grid. Moreover, we use Richardson Extrapolation (RE) to predict
hence, a value of z0 = 0.005 m has been used in this study. All the flow rate for an ideal mesh (continuum), thereby estimat-
numerical simulations are based on OpenFOAM’s steady-state ing the magnitude of the numerical error. The mesh size for
simpleFoam solver in combination with a k−w−SST RANS tur- each refinement stage is summarised in Table 2, and the meshes
bulence model. The pressure-velocity coupling was established themselves are illustrated in Figure 7.
with the SIMPLE algorithm using three non-orthogonal correc-
tor loops. Buoyancy effects were neglected due to air velocities
3.2. Annual study
that are well above 1.8 m s−1 (Boulard et al. 1996; Tecle, Bitsuam-
lak, and Jiru 2013). Furthermore, we assumed that convergence A previous study found that a singled out, cubic building geom-
was obtained when reaching residuals of 1 × 10−4 for p and 1 × etry is not adequate to highlight the benefits of the proposed
10−5 for the remaining fields. For each case, we simulated until method in terms of overall run times (Kastner and Dogan 2018).
the convergence criteria were reached. The relaxation factors To establish a more real-world-like scenario, we created simu-
were chosen to be 0.7 for p and 0.3 for U, k and ω. All simulations lation domains for a square-shaped, equidistant building array
ran on an AMD Ryzen Threadripper 1950X 16-Core Processor consisting of cubes with a 20 m edge length. To ensure a fair

Table 2. Parameters of the grid convergence study for four different mesh sizes: very coarse, coarse, normal, and fine.
Case Cell count hg r v̇[m3 s−1 ] Error (experiment) Error (continuum) GCI [%]
very coarse 124456 8 1.3 0.0540 17% 15% 11.04
coarse 263919 4 1.6 0.0505 11% 10% 6.82
normal 1165733 2 1.6 0.0485 7% 6% 4.94
fine 4940280 1 – 0.0471 4% 3% –
RE – 0 – 0.0456 – – –
Experiment – – – 0.0450 – – –
64 P. KASTNER AND T. DOGAN

Figure 7. Excerpt of refined mesh sizes for the validation study. (a) very coarse, (b) coarse, (c) normal and (d) fine.

comparison between the meshes, the refined regions in the cen-

tre of the simulation domain are identical and the surrounding
coarser regions were set up with the same cell divisions. For this
study, uref was arbitrarily chosen to be 5 m s−1 and the turbu-
lence values were calculated according to Equations (1)–(3).
The Cases A and A’ exemplify a rotation of the simulation
domain (0◦ and 45◦ representing the two extremes in terms of
the projected facade area) according to the best practice rules
if the blocking ratio is fixed at 3%. This leads to an increase in
the size of the simulation domain for Case A’. By simulating these
two extreme cases, a full 360◦ study is reproduced with 45◦ step-
wise rotations by exploiting the rotational symmetry of A and
A’. Moreover, we assessed a square-shaped simulation domain
suggested by Franke and Baklanov (Case B), which is charac-
terised by automatically-generated inlet and outlet boundary
conditions identical to Case C. This approach, often used as a
workaround for expedited case setups, also benefits from only
having to be meshed once no matter which wind direction is
imposed. Finally, Case C represents the cylindrical approach that
we propose in this study. To estimate the final run times of a Figure 8. Coefficients of determination (R2 ) for vertical velocity probes uy at h/2
set of hypothetical wind directions, we summed up the meshing for different mesh sizes.
and simulation times and multiplied them by the correspond-
ing number of wind directions (or rotations). Hence, the meshing
time was taken into account multiple times for Case A and A’,
whereas only a single time for Case B and C.

4. Results
4.1. Validation study
Figure 7 shows a cropped stream-wise section around the build-
ing geometry for each mesh refinement. For each of the meshes,
the vertical velocity samples are depicted in Figure 8, in which
the normalised domain height is plotted over the normalised
reference velocity uy . By comparing the R2 of each refinement
stage, it is evident that the accuracy of the solution increases
for finer grids. The finest grid, however, under-predicts the nor- Figure 9. Richardson Extrapolation (RE) of the volumetric flow rate v̇ through the
malised velocity, especially in the opening region. windward opening based on the grid convergence study.
In the experiment conducted by Jiang et al. (2003), a volu-
metric flow rate of 0.045 m3 s−1 was measured for the open-
ing of the building geometry. Contrary to the vertical velocity asymptotic range of convergence (Kastner 2016). Given the rea-
probes, all stages of the grid refinement confirm the measure- sonable numerical error band of ∼ 10%, we decided to continue
ment of that derived variable in a converging manner toward with the ‘coarse’ grid for the subsequent accuracy comparison.
the finer grids (Figure 9 and Table 2). By evaluating the GCS, the To compare the accuracy of the box-shaped and the cylindri-
volumetric flow rate for an ideal mesh would be 0.047 m3 s−1 , cal method, the vertical, stream-wise probes uy are plotted at
which is in good agreement with the experimental results of h/2, for the ‘coarse’ mesh setting, see Figure 10. It is evident
0.045 m3 s−1 . Furthermore, the ratios of the CGIs confirm that that the box-shaped simulation domain achieves a marginally
the volumetric flow rates reported in the GCS are well within the higher1 accuracy (R2 = 96.7%) than the cylindrical simulation
 JOURNAL OF BUILDING PERFORMANCE SIMULATION 65

horizontal velocity plots at 2 m above the ground plane. This

confirms visually that all simulation domains are comparable in
capturing the flow field, as shown before.
Figure 13 provides an extrapolation of the individual mesh-
ing and simulation time for an annual analysis of an urban area
depending on the number of wind directions. For the individ-
ual meshing times, we report a significantly higher meshing time
for Case A’ compared to the three other cases. The extrapola-
tion in Figure 13 reveals that the cylindrical simulation domain
(Case C) yields 18% better run times compared to the conven-
tional approach (Case A/A’) if 8 or more wind directions are
simulated. Moreover, Case C shows 39% better overall run times
compared to the square-shaped approach (Case B) regardless of
the number of wind directions.

5. Discussion
The results show that it is possible and beneficial to employ
Figure 10. Vertical sample uy at h/2 for the box-shaped and the cylindrical simu- a cylindrical mesh for annual/multi-directional urban CFD sim-
lation domain. ulations with OpenFOAM. Where high-pressure gradients are
expected in the simulation domain, the mesh has to be appro-
priately refined to capture important flow features and to ade-
domain (R2 = 94.1%). However, both approaches either under- quately resolve the boundary layer. We estimated the numerical
and/or overestimate regions with high-pressure gradients which error resulting from our validation mesh setting to be ∼ 10%
is known as a deficiency of the steady-state RANS model. with the help of a grid convergence study (GCS). GCSs for large
Figure 11 depicts the residuals reported for both cases. Here, meshes are not a trivial task, as the simulation time grows expo-
the cylindrical domain shows better convergence behaviour for nentially with the number of background mesh divisions. How-
the particular convergence criteria (Figure 11(b)). This results in ever, this method is useful to both estimate the numerical error
a simulation that converges after ∼ 2900 iterations for the cylin- and to confirm an adequate convergence behaviour by utilising
drical case, whereas the box-shaped WT converges after ∼ 4100 a reasonable number of mesh refinements. To put this effort into
iterations. perspective, we would like to emphasise that this study focuses
on the comparison of two different meshing approaches, not to
achieve the best overall accuracy. Here, the GCS confirms that
4.2. Annual study
the refinement strategy follows commonly-accepted guidelines
The annual study is concerned with evaluating the run times and that the coarse mesh is sufficient for the goals of this study.
for a square-shaped, equidistant building array. The top row in Further, we show that for the coarse mesh, the box-shaped
Figure 12 shows the top view of the meshes after the blockMesh simulation domain achieves a marginally better accuracy (R2 =
routine in OpenFOAM. A detailed summary of the three simu- 96.7%) than the cylindrical simulation domain (R2 = 94.1%)
lation domains is given in Table 3. For Case A/A’, the width of while both domains either under- and/or overestimate regions
the WT varies with the angle of the approaching wind (which is with high-pressure gradients. The deficiency in accuracy in high-
largest in the particular Case A’) due to the constraint to keep pressure regions can be attributed to the steady-state RANS
the blocking ratio constant. The bottom row in Figure 12 shows model. While we are aware of the limited applicability of the

Figure 11. Residuals of box-shaped (a) and cylindrical simulation domain (b).
66 P. KASTNER AND T. DOGAN

Figure 12. Top row: meshes after the blockMesh routine. Case A and A’ correspond to a box-shaped tunnel for 0 and 45◦ , Case B’ corresponds to a square-shaped approach,
and Case C’ corresponds to a cylindrical approach. Bottom row: horizontal velocity plots at 2 m above the ground plane.

Table 3. Geometry and mesh information about the cases A, A’, B, and C.
Dimensions Mesh
Case In flow direction[m] ⊥ to flow direction[m] Ground area[m2 ] Projected facade area[m2 ] Cell count Cells per ground area[cells/m2 ]
A 1.200 399 4.79E + 05 4.000 8.1E + 05 1.69
A’ 1.200 1.408 1.69E + 06 10.182 1.4E + 06 0.83
B/B’ 1.600 1.600 2.56E + 06 10.182 1.3E + 06 0.51
C/C’ 1.600 1.600 2.01E + 06 10.182 9.7E + 05 0.48

RANS approach for urban flows, it is noteworthy that the appli- extrapolated from the individual meshing and simulation times.
cation of the method proposed in this study focuses on early Here, the run times reported for Case A and A’ were linearly
design exploration for the built environment. Thus, we empha- interpolated to estimate the theoretical run times in the case of
sise the interest in overall feasible run times rather than accuracy. 16 and 32 wind directions and the necessary rotations. For Case
However, the cylindrical meshing approach presented can be A and A’, it is sufficient to adhere to the best practice dimen-
used independently of the solver. Therefore, such deficiencies sions from the wind direction of that simulation instance (either
could be avoided by using more advanced LES or DNS solvers, 0◦ or 45◦ ), whereas a cylindrical domain needs to be created in a
should the need for very accurate results arise. way that wind from all directions can be simulated appropriately.
In Figure 13, we summarise the run time comparisons Hence, additional cells are introduced at the ground surface of
for an annual analysis with multiple wind directions that are the cylindrical domain, see Figure 6. Considering that trade-off,

Figure 13. Meshing and simulation times for cases A/A’, B/B’, and C/C’.
 JOURNAL OF BUILDING PERFORMANCE SIMULATION 67

we suggest that simulation domains according to Case A/A’ considering best practice dimensions for urban CFD studies.
be used for analyses with < 8 wind directions2 . For ≥ 8 wind Meshing and simulation time comparisons show that it is recom-
directions, the fact that the cylindrical domain only needs to be mended to use the box-shaped approach if fewer than 8 wind
meshed once makes up for the additional number of cells in the directions are intended to be studied. We show that the cylindri-
simulation domain. Thus, for ≥ 8 wind directions, we suggest cal simulation domain shows better overall run times in the case
using the cylindrical approach. of 8 or more wind directions. Concluding, we discuss how a cylin-
However, it should be emphasised that the advantages of drical simulation domain may likely have advantages over the
a cylindrical domain may outweigh the disadvantages even in box-shaped approach from a methodological perspective, even
cases with fewer than 8 wind directions as CFD simulations are if fewer than 8 wind directions are studied.
highly sensitive to the quality of the CAD input geometry. First, in
practical use with real-world problems, the scale of most urban Nomenclature
geometries suggest overnight or over-the-weekend simulations.
Given these time-spans, improvements in simulations time are ABL Atmospheric boundary layer
indeed desirable but only magnitudes in run time improvements AFN Air Flow Networks
would change the way of working with such large-scale simu- BC Boundary condition
lations. On the contrary, malfunctioning simulation setups that BEM Building energy modelling
stem from manually changing the boundary conditions might CAD Computer-aided design
result in weekends of wasted computing time. Furthermore, CFD Computational Fluid Dynamics
the creation of robust, high-quality meshes often requires time- DNS Direct numerical simulation
consuming and tedious preprocessing efforts, mostly for clean- GCI Grid convergence index
ing the CAD geometries. CFD analyses for the built environment GCS Grid convergence study
are usually characterised by iterative design changes that have NV Natural ventilation
been outlined above. Every change is likely to introduce new LES Large eddy simulation
mesh inconsistencies that, for Case A and A’, might elicit mesh- RANS Reynolds-averaged Navier-Stokes
ing or convergence issues for every additional wind direction. SIMPLE Semi-Implicit Method for Pressure Linked Equations
Considering these impediments, we suggest making use of the SST Shear stress transport
inherent advantages of a cylindrical simulation domain. Here, WT Wind tunnel
the simulation domain only needs to be validated once, which cp Pressure coefficients
subsequently guarantees valid simulation results from all direc- h Height, m
tions. Moreover, the simulations for all wind directions may even hg Normalized grid spacing
be started in parallel after the single mesh is created. This pos- Cmu Constant
sibility might help to identify problems in a buildings’ design  Rate of dissipation of turbulence energy, m2 s−3
early on before it might be discovered with a sequential sim- k Turbulence kinetic energy, m2 s−2
ulation approach. Consequently, one requirement we see for κ von Karman constant
the prospective automation of CFD workflows for the built envi- p Pressure, kg m−1 s2
ronment is the ability to produce robust, converging simulation r Grid refinement ratio
cases. As a first step in that direction, we suggest using a cylin- R2 Coefficient of determination
drical simulation domain if the meshing process of the particular U Velocity, m s−1
building geometry seems to be unstable and likely to fail in the uref Reference velocity, m s−1
case of further geometry rotations. U∗ Friction velocity, m s−1
Apart from mixed polyhedral meshes that were used in v̇ Volumetric flow rate, m3 s−1
this study, other commonly used meshes include hexahedral- ω Specific dissipation rate, s−1
only, tetrahedral-only meshes. As the two latter are known to ŷi Predicted values
achieve better performance in terms of meshing time, future ȳi Mean values
work should investigate whether such an approach might affect z Dimensions along z-axis, m
the conclusions drawn. From an implementation perspective, z0 Surface roughness length, m
the cylindrical meshing approach could take advantage of faster zref Reference velocity, m s−1
meshing times.
In future work, we plan to use the results from such annual Notes
wind studies as input for BEM software to inform the prediction
1. The marginally higher R2 (94.6% vs. 96.7%) in Figure 10 vs. Figure 8 stems
of the natural ventilation potential of buildings.
from using three additional mesh boundary layers for the ground and
the building surfaces.
2. In an earlier study, a less optimistic result had been reported which
6. Conclusion
neglected the effect that the rotation enforces with respect to the
In this study, we propose a cylindrical mesh for urban wind sim- width of the WT while keeping the blocking ratio constant (Kastner and
Dogan 2018).
ulations and show that such simulation domains are feasible
and beneficial with OpenFOAM. We examined a commonly-used
validation case for which we compared the box-shaped com- Disclosure statement
putational domain to the cylindrical simulation domain while No potential conflict of interest was reported by the authors.
68 P. KASTNER AND T. DOGAN

Funding : An Outcome of COST 732.” In The Fifth International Symposium on

Computational Wind Engineering (CWE2010), Chapel Hill, 1–10.
The authors would like to acknowledge the financial support by the Cor-
Franke, J., C. Hirsch, A. G. Jensen, H. Krus, M. Schatzmann, P. S. W. Miles, S. D.,
nell University David R. Atkinson Center for a Sustainable Future and the
J. A. Wisse, and N. G. Wright. 2004. Recommendations on the Use of CFD in
Cornell Center for Transportation, Environment, and Community Health -
Wind Engineering. Technical Report, Joint Publication.
CTECH which funded this research.
Grosso, M.. 1992. “Wind Pressure Distribution Around Buildings: A Paramet-
rical Model.” Energy and Buildings 18 (2): 101–131.
ORCID Huang, J., F. Winkelmann, and F. Buhl. 1999. Linking the COMIS Multizone
Airflow Model with the EnergyPlus.
Patrick Kastner http://orcid.org/0000-0003-4940-341X Jiang, Y., D. Alexander, H. Jenkins, R. Arthur, and Q. Chen. 2003. “Natural
Timur Dogan http://orcid.org/0000-0003-0749-8465 Ventilation in Buildings: Measurement in a Wind Tunnel and Numerical
Simulation with Large-Eddy Simulation.” Journal of Wind Engineering and
Industrial Aerodynamics 91 (3): 331–353.
Kastner, P.. 2016. “Customizing OpenFOAM to Assess Wind-Induced Natural
References Ventilation Potential of Classrooms: A Case Study for BRAC University.”
Blocken, B.. 2015. “Computational Fluid Dynamics for Urban Physics: Master’s thesis, Technische Universität München.
Importance, Scales, Possibilities, Limitations and Ten Tips and Tricks Kastner, P., and T. Dogan. 2018. “Streamlining Meshing Methodologies for
Towards Accurate and Reliable Simulations.” Building and Environment 91: Annual Urban CFD Simulations.” In eSim 2018.
219–245. Oropeza-Perez, I., and P. A. Østergaard. 2014. “Energy Saving Potential of
Boulard, T., J. Meneses, M. Mermier, and G. Papadakis. 1996. “The Mecha- Utilizing Natural Ventilation Under Warm Conditions – A Case Study of
nisms Involved in the Natural Ventilation of Greenhouses.” Agricultural Mexico.” Applied Energy 130: 20–32.
and Forest Meteorology 79 (1-2): 61–77. Ramponi, R., and B. Blocken. 2012. “CFD Simulation of Cross-Ventilation for a
Cardinale, N., M. Micucci, and F. Ruggiero. 2003. “Analysis of Energy Sav- Generic Isolated Building: Impact of Computational Parameters.” Building
ing Using Natural Ventilation in a Traditional Italian Building.” Energy and and Environment 53 (0): 34–48.
Buildings 35 (2): 153–159. Roache, P. J.. 1994. “Perspective: A Method for Uniform Reporting of Grid
Cheung, J. O. P., and C. H. Liu. 2011. “CFD Simulations of Natural Ventilation Refinement Studies.” Journal of Fluids Engineering 116 (3): 405.
Behaviour in High-Rise Buildings in Regular and Staggered Arrangements Swami, M., and S. Chandra. 1988. “Correlations for Pressure Distribution on
at Various Spacings.” Energy and Buildings 43 (5): 1149–1158. Buildings and Calculation of Natural-Ventilation Airflow.” ASHRAE Trans-
Costola, D., B. Blocken, and J. Hensen. 2009. “Overview of Pressure Coeffi- actions 94 (3112): 243–266.
cient Data in Building Energy Simulation and Airflow Network Programs.” Tecle, A., G. T. Bitsuamlak, and T. E. Jiru. 2013. “Wind-Driven Natural Ven-
Building and Environment 44 (10): 2027–2036. tilation in a Low-Rise Building: A Boundary Layer Wind Tunnel Study.”
Franke, J.. 2006. “Recommendations of the COST Action C14 on the Use of Building and Environment 59: 275–289.
CFD in Predicting Pedestrian Wind Environment.” In The Fourth Interna- Tominaga, Y., A. Mochida, R. Yoshie, H. Kataoka, T. Nozu, M. Yoshikawa, and
tional Symposium on Computational Wind Engineering, 529–532. T. Shirasawa. 2008. “AIJ Guidelines for Practical Applications of CFD to
Franke, J., and A. Baklanov. 2007. Best Practice Guideline for the CFD Simulation Pedestrian Wind Environment Around Buildings.” Journal of Wind Engi-
of Flows in the Urban Environment: COST Action 732 Quality Assurance and neering and Industrial Aerodynamics 96 (10-11): 1749–1761.
Improvement of Microscale Meteorological Models. Hamburg: University of Wallace, J. M., and P. V. Hobbs. 2006. Atmospheric Science: An Introductory
Hamburg. http://theairshed.com/pdf/COST%20732%20Best%20Practice Survey. 2nd ed. Burlington, VT: Academic Press.
%20Guideline%20May%202007.pdf. Weber, A., M. Koschenz, V. Dorer, M. Hiller, and S. Holst. 2003. TRNFLOW, a
Franke, J., A. Hellsten, H. Schlünzen, and B. Carissimo. 2010. “The Best Prac- New Tool for the Modelling of Heat, Air and Pollutant Transport in Buildings
tise Guideline for the CFD Simulation of Flows in the Urban Environment within TRNSYS.