Heliyon 7 (2021) e06723

Contents lists available at ScienceDirect

Heliyon
journal homepage: www.cell.com/heliyon

Research article

Optimising the computational domain size in CFD simulations of

tall buildings
Yousef Abu-Zidan *, Priyan Mendis, Tharaka Gunawardena
Department of Infrastructure Engineering, The University of Melbourne, VIC 3010, Australia

A R T I C L E I N F O A B S T R A C T

Keywords: Recently, there has been a growing interest in utilizing computational fluid dynamics (CFD) for wind resistant
Computational domain design of tall buildings. A key factor that influences the accuracy and computational expense of CFD simulations is
Tall buildings the size of the computational domain. In this paper, the effect of the computational domain on CFD predictions of
CFD simulation
wind loads on tall buildings is investigated with a series of sensitivity studies. Four distinct sources of domain
Wind loading
Venturi effect
error are identified which include wind-blocking effects caused by short upstream length, flow recirculation due
to insufficient downstream length, global venturi effects due to large blockage ratios, and local venturi effects
caused by insufficient clearance between the building and top and lateral domain boundaries. Domains based on
computational wind engineering guidelines are found to be overly conservative when applied to tall buildings,
resulting in uneconomic grids with a large cell count. A framework for optimizing the computational domain is
proposed which is based on monitoring sensitivity of key output metrics to variations in domain dimensions. The
findings of this paper help inform modellers of potential issues when optimizing the computational domain size
for tall building simulations.

1. Introduction (CWE) best practice guidelines (Blocken, 2015; Franke et al., 2007;
Tominaga et al., 2008). These guidelines attribute domain errors to
The use of computational fluid dynamics (CFD) for wind resistant comparable issues in wind tunnel testing such as blockage effects in
design of tall buildings has gained interest in recent years as advance- domains of small cross-sectional area, and artificial acceleration of local
ment in computing power has made solving complex flow problems flow in domains with inadequate clearance between the building model
increasingly affordable. A key factor that influences the accuracy and cost and domain boundaries. Hence, sizing requirements are specified in
of CFD simulations is the size of the computational domain. terms of maximum blockage ratios, minimum distances between the
The computational domain (Figure 1) refers to an external volumetric domain boundaries and the building model, or a combination of both
region that surrounds the building model, where the basic flow equations (Blocken, 2015).
are discretised and solved. A total of six boundaries define the extents of a Franke et al. (2007) recommend a maximum blockage ratio of 3%
typical domain of cuboid shape. Aside from the bottom of the domain, based on an early study of domain effects in wind load predictions of a
these boundaries are mostly non-physical, so their influences on the flow low-rise building (Baetke et al., 1990). A similar limit on blockage ratio is
region constitute a source of error in the simulation (hereby termed recommended by Tominaga et al. (2008) for pedestrian wind comfort
domain errors). Non-physical boundaries should be placed far enough studies around tall buildings, although this was justified based on
from the building to avoid significant influences on the results. However, experience from wind tunnel testing. In addition to the blockage ratio,
placing the boundaries too far could increase the computational cost of Franke et al. (2007) impose minimum upstream, downstream, and lateral
the model. Consideration for both computational cost and solution ac- distances of 5H, 15H and
5H, respectively, where H is the height of the
curacy dictates the need for optimizing the size of the computational tallest building in the model. These distances were derived from earlier
domain. studies that were a part of a multi-partner project that evaluated un-
The importance of an adequately sized computational domain for certainties in CFD modelling of pollutant dispersion around buildings
solution accuracy is recognized by computational wind engineering (Castro et al., 1999; Cowan et al., 1997; Hall, 1997). Tominaga et al.

* Corresponding author.
E-mail address: [email protected] (Y. Abu-Zidan).

https://doi.org/10.1016/j.heliyon.2021.e06723
Received 11 June 2020; Received in revised form 20 December 2020; Accepted 1 April 2021
2405-8440/© 2021 The Author(s). Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-
nc-nd/4.0/).
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

(2008) recommend similar upstream and lateral distances but specify a will ensure that domain errors are investigated in isolation. The findings of
shorter downstream length of 10H. this paper will help inform modellers of potential issues when optimizing
While these guidelines are sufficient for preventing domain effects in the computational domain size for tall building simulations.
most CWE applications (Blocken, 2015), they were found to be overly
conservative when applied to tall buildings (Revuz et al., 2012). Many 2. Analysis cases and numerical setup
studies of tall buildings use smaller domains than those recommended
above, but there is little consistency between the domain sizes used in In this paper, domain investigations are performed for a generic
such studies [e.g. (Braun and Awruch, 2009; Huang et al., 2007, 2011; rectangular tall building with full-scale dimensions of 40 m in both width
Ricci et al., 2018; Zhang et al., 2015)]. (B) and depth (D), and 200 m in height (H). The size of the computational
For CFD simulations of wind loading on tall buildings, domain opti- domain is defined by four variables; lu, ld, b, and h (see Figure 1); where lu
mization is necessary because it can greatly impact the computational is the upstream domain length from the inlet to the windward building
cost. Tall buildings are governed by dynamic wind effects such as face, ld is the downstream length from the leeward building face to the
crosswind vibrations due to vortex shedding, and simulating these effects outlet plane, b is the side clearance on either side of the building, and h is
requires the use of scale-resolving simulations which require high grid the total height of the domain.
refinement throughout the domain. High refinement is particularly This paper consists of three distinct studies, each comprising several
needed in the upstream region to accurately transport inflow turbulence simulation cases. In the first study, the domain extents lu, ld, b, and h are
from the inlet boundary to the location of the building. Because of the each varied independently to assess the impact of each of these param-
high grid refinement, the size of the computational domain will greatly eters in isolation. In the second study, the impact of the blockage ratio
influence the cell count in the model. Moreover, even if optimizing the (BR) is investigated by varying b and h simultaneously. Finally, in the
domain only achieves a modest reduction in cell count, savings in third study, the impact of wind speed and geometric scale on domain
computational cost would accrue in a transient simulation over thou- errors is examined. Details of the simulation cases for each of the three
sands of timesteps. studies are summarized in Table 1.
Optimal domain sizing is difficult to codify because it is highly The domain dimensions shown in Table 1 with a single asterisk (*) are
problem-specific. However, a basic investigation into the nature of based on the commonly adopted guidelines by Franke et al. (2007) and
domain errors and their impact on the results would greatly help in are henceforth referred to as the base case (BC). Simulation cases are
guiding the optimization process. To our knowledge, no such study has constructed by altering the parameters of interest from the BC, while the
been performed for tall building applications. Furthermore, domain other parameters remain at the BC values. Most of the domain sizes in
studies in the literature do not demonstrate horizontal homogeneity of Table 1 are selected to be smaller than the BC since the BC is conserva-
the atmospheric boundary layer (ABL) which is needed to provide con- tively large for the tall building in this study. Nonetheless, for each of
fidence that domain errors are investigated independently of ABL in- studies 1 and 2, at least one case is selected to be significantly larger than
homogeneity errors since both domain errors and ABL inhomogeneity the BC. This large case is assumed to represent an infinitely large domain
errors are sensitive to upstream domain length (Abu-Zidan et al., 2020). that is free from domain errors.
Many studies do not address the issue at all (Buccolieri and Di Sabatino,
2007; Ramponi and Blocken, 2012; Xiang and Wang, 2007), while others 2.1. Demonstrating ABL homogeneity
acknowledge this issue but do not quantify its impact on their findings
(Ai and Mak, 2015; Revuz et al., 2012). To investigate the nature of domain errors, it is important to ensure
To address these limitations, this paper will investigate the effect of the that these errors act in isolation. A potential secondary source of error in
computational domain size in CFD simulations of wind loading on tall the current study is due to ABL inhomogeneity. ABL inhomogeneity re-
buildings with a series of sensitivity studies that include examining the fers to an unintended adaptation in the ABL profile often caused by an
effect of (1) domain dimensions and (2) blockage ratio, as well as the in- incompatibility of the inlet profile equations with the turbulence model
fluence of (3) additional parameters of wind speed and geometric scale. The and near-wall functions. Controlling ABL inhomogeneity in this study is
first two studies involve a direct investigation of domain effects on wind important because both ABL inhomogeneity errors and domain errors are
load predictions. Domain parameters are varied in isolation to help identify influenced by upstream domain length lu. A fuller discussion of the
various sources of domain error in the simulation. The third sensitivity relationship between ABL inhomogeneity errors and upstream domain
study is performed to generalise the findings from studies 1 and 2 to cases of length has been presented by Abu-Zidan et al. (2020).
various wind speeds and geometric scales. Moreover, the study will ABL homogeneity is achieved by ensuring that flow continuity
demonstrate the achievement of a horizontally homogenous ABL, which equations are in balance with the turbulence model that is adopted in the

symmetry

Figure 1. Computational domain dimensions and boundary conditions.

2
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

Table 1. Details of simulation cases for sensitivity studies in this paper.

Sensitivity study 1. Impact of domain extents 2. Impact of blockage ratio 3. Impact of wind speed and geometric scale

Parameter lu ld b h BR Wind speed Geometric scale

Simulation cases 1H 3H** 1.5H 2.2H 2.16% Uref ¼ 10 m/s (OD**) 1:400 (OD)
(b ¼ 2H, h ¼ 2.2H)
1.5H 5H 2H 3H 1.59%
(b ¼ 2H, h ¼ 3H)
2H 8H 3H** 4H** 1.28%
(b ¼ 2.5H, h ¼ 3H)
3H** 10H 4H 6H* 0.81%** Uref ¼ 10 m/s (BC) 1:400 (BC)
(b ¼ 3H, h ¼ 4H)
4H 15H* 5H* 8H 0.33%*
(b ¼ 5H, h ¼ 6H)
5H* 20H 7H 10H 0.10%
7H 25H 10H (b ¼ 10H, h ¼ 10H)
10H
*
Dimensions of base case (BC) based on recommendations by Franke et al. (2007).
**
Optimised domain (OD) selected based on findings from analyses 1 and 2.

simulation, the ABL profiles that are specified at the inlet, and the wall area where tall buildings are typically located (Blocken, 2015). The inlet
function that replicates roughness conditions at the ground. At the inlet k and ϵ profiles are based on the theoretical Eqs. (2) and (3). The inlet
boundary, theoretical ABL profiles are specified based on expressions profiles and the modified wall function are implemented in FLUENT with
derived by Richards and Norris (2011) [Eqs. (1), (2), and (3)] that are in a user-defined function (Abu-Zidan, 2019). The top shear requirement is
balance with the RNG k-ϵ turbulence model adopted in this study. A satisfied by specifying a momentum source term at the topmost cell layer
driving shear stress is specified at the top boundary [Eq.(4)], and a in the domain.
modified wall function proposed by Parente et al. (2011) is applied to the ABL homogeneity is verified by simulating flow in an empty compu-
ground. The modified wall function is based on aerodynamic roughness tational domain, and comparing the incident profiles of U, k, and ϵ to the
height z0 and can accurately produce ground roughness conditions that inlet profiles. The results are plotted in Figure 2. The incident profiles are
are necessary for maintaining homogeneity of the ABL. Full details for plotted at a horizontal distance of x ¼ 2000 m from the inlet, which cor-
this modelling approach can be found in a previous study by the authors responds to the largest upstream fetch used in this study (lu ¼ 10H).
(Abu-Zidan et al., 2020). The plots reveal clear conformity between the inlet and incident
profiles at all heights in the domain (Figure 2a) and particularly near the
u* z þ z0
U¼ ln ð Þ; Cμ ¼ 0:085; κ ¼ 0:4 (1) ground (Figure 2b). This demonstrates that the inlet profiles are pre-
κ z0 served as they travel through the computational domain. In other words,
a fully horizontally homogenous atmospheric boundary layer (HHABL) is
u2* achieved, and inhomogeneity errors are contained.
k ¼ pffiffiffiffiffi
ffi (2)
Cμ

u3* 2.2. Mesh configuration

ε¼ (3)
κðz þ z0 Þ
Another potential source of error in this study is due to spatial dis-
cretisation. The impact of this error is mitigated by maintaining a
τ¼ρ u2* (4)
consistent meshing configuration for all simulation cases in this study.
A velocity profile [Eq. (1)] is selected with u* ¼ 6.12 m/s and z0 ¼ 2.0 This prevents variation in the results that are caused by variation in the
m. This corresponds to a wind speed of Uref ¼ 60 m/s at 100 m height. mesh (Revuz et al., 2012). To maintain a constant mesh near the building,
The roughness length z0 value corresponds to a well-developed urban the larger domain grids are generated by extruding the boundary cells

(a) (b)
1000 25

800 20

600 15
(m)
(m)

400 10

200 5

0 0
0 50 100 150 0 50 100 150
(m/s), (m2/s2), (m2/s3) (m/s), (m2/s2), (m2/s3)

Figure 2. Inlet and incident ABL profiles for velocity U, turbulence kinetic energy k, and turbulence dissipation rate ϵ; from (a) z ¼ 0–1000 m; (b) z ¼ 0–25 m near the
ground surface. x is the streamwise distance from the inlet.

3
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

from smaller domains. Furthermore, a grid of high refinement is achieved sensitivity analysis of the blockage ratio, and (3) analysis of the impact of
near the building with structured hexahedron elements. The structured wind speed and geometric scale. The results of the three sensitivity
mesh resulted in high-quality elements with skewness <0.5. studies are presented in the following three subsections.
To assess the adequacy of the grid, a grid convergence study is per-
formed based on Richardson's extrapolation method (Celik et al., 2008).
Three grids of increasing refinement were generated for the base case 3.1. Sensitivity study on the impact domain extents
(BC) in Table (1). The number of elements for the coarse, middle, and fine
grid was 1.6, 2.3 and 3.1 million elements respectively. For each of the In the first sensitivity study, the impact of each of the four
three grids, results of the building's base reactions and surface pressure domain extents, lu, ld, b, and h, on domain errors is investigated.
along the centreline were obtained. Based on these results, spatial dis- This is done by varying each dimension in isolation while all other
cretisation errors are quantified in terms of the Grid Convergence Index domain dimensions are held at BC values. For instance, when
(GCI) (Celik et al., 2008). Spatial discretisation errors for the middle grid investigating the impact of upstream length, lu is varied incremen-
were minimal, with GCI values of 0.91% and 0.31% for the base moments tally from 1.5H to 10H (according to Table 1), while ld, b, and h,
and shear respectively, and an average GCI of 0.86% for pressure values are fixed at 15H, 5H, and 6H respectively.
extracted along the centreline of the building. For each simulation case, pressure coefficients Cp are plotted at the
Based on the results of the mesh sensitivity analysis, the middle grid centreline of the windward, leeward, side, and top faces of the building
with 2.3 million elements was selected for the BC, with a minimum (see Figure 4e). Since a uniform mesh and HHABL conditions are ach-
element size of 0.06 m at the edges of the building, a maximum element ieved, any variations in the pressure plots can be attributed to variations
size of 50 m at the boundaries of the domain, and an average mesh in the domain size. The pressure coefficients are normalised according to
growth ratio of 1.10. The maximum height of the wall-adjacent elements Eq. (5) with UH ¼ 70.6 m/s and Pref ¼ 0 Pa for all cases.
is 0.07 m at the building surface, and 1.84 m at the ground. The resulting
P  Pref
mesh is shown in Figure 3. CP ¼ (5)
0:5ρUH2
Figure 4 and Figure 5 show the impact of upstream length lu and
2.3. Solver settings
domain width b on surface pressure, respectively. The impact of domain
height h and downstream length ld on surface pressure was minimal even
Steady-state RANS simulations are performed in CFD code FLUENT
for the smallest dimensions of h and ld in this study. The maximum
with the RNG k-ϵ model. The standard wall function is applied to all
percentage error for both dimensions was below 2%. The pressure plots
surfaces of the building, while the modified wall function is used at the
of h and ld are not presented here as no trends can be visually discerned
ground boundary. Symmetry boundary conditions are applied to the top
from the plots.
and side boundaries of the domain and a pressure outlet condition is
To further investigate the influence of domain extents on wind load
specified at the outlet boundary. A pressure-based solver is used with a
predictions, domain error is calculated as per Eq. (6):
full pressure-velocity coupling algorithm that solves momentum and
 
continuity equations simultaneously. Spatial discretisation of moment, CP  CPlargest domain 
turbulent kinetic energy, and turbulent dissipation rate is performed Error ¼    100
 (6)
C Plargest domain
using a second-order upwind scheme. A second-order scheme is also used
for pressure interpolation, and gradient discretisation is performed using where the actual value is assumed to correspond to the largest sized
the Least Squares Cell Based method (ANSYS Inc., 2013). The solution is domain corresponding to each of the four domain parameters lu, ld, b and
iterated until all scaled residuals dropped below 105. Convergence was h as listed in Table 1. For instance, the error in Figure 6a is calculated as
verified by monitoring key metrics in the model including base reactions the deviation of results for cases lu ¼ 1H, 1.5H, 2H …etc. from results
and surface pressure. corresponding to the largest case of lu ¼ 10H. Similarly, errors in
Figure 6b for cases ld ¼ 3H, 5H, 8H …etc. are calculated as deviations
3. Simulation results and analysis from the largest case of ld ¼ 25H, and so on. The size of the largest case is
exaggerated to ensure that domain error is virtually negligible, which
As previously mentioned, three sensitivity studies are performed in would allow the use of the largest case as a reference for calculating error
this paper: (1) sensitivity analysis of all four domain extents, (2) in the smaller domains.

Figure 3. (a) Top view and (b) perspective view of structured hexahedral mesh with high refinement near the building.

4
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

(a) (
(b) (c)
1.0

1H
0.8
1.5H
2H
0.6 3H
5H (BC)
/

0.4 10H

0.2

0.0
0.2 0.4 0.6 0.8 1.0 1.2 -0.4 -0.3 -0.2 -0.8 -0.6 -0.4 -0.2
Windward Leeward Side

(d) (e) d
-0.3
-0.4
-0.5
-0.6 b c
Top

-0.7 a
-0.8
-0.9
-1.0
-1.1 wind
0 0.25 0.5 0.75 1
/
Figure 4. Impact of upstream length lu on surface pressure. Cp plotted along centreline of (a) windward, (b) leeward, (c) side, and (d) top surfaces of building. Plot (e)
illustrates the sampling lines in plots (a)–(d), labelled correspondingly.

The following observations are made from the results presented in (Figure 6c). This occurred at the top, leeward, and side faces of the
this section: building, while the error on the windward face was less prominent.
 The impact of domain height h was minimal even for the smallest case
 For all four domain dimensions, the error decayed exponentially as of h ¼ 2.2H (Figure 6d). The maximum error for all domain height
the domain is extended in each direction. This behaviour is seen in the cases was <2%.
error plots in Figure 6. The same behaviour can also be seen, although  The BC domain size recommended by Franke et al. (2007) was suf-
less apparent, in the pressure plots in Figure 4 and Figure 5 where the ficient to ensure negligible impact of domain extents. The maximum
pressure profiles converge onto the profiles from the largest domain. domain extent error for the BC was 0.4%, which occurred at the
This trend justifies the assumption that the large domains in this study leeward surface of the building.
are free from domain errors.
 Of the four domain dimensions in this study, the upstream length lu 3.2. Sensitivity study on the impact of blockage ratio
was found to have the most significant impact on the accuracy of the
results. The largest magnitude of error occurred for domains with a The second study examines the impact of the blockage ratio on sur-
short upstream length (Figure 6a). This error occurred on all surfaces face pressure. The blockage ratio (BR) refers to the ratio of the projected
of the building but was largest on the windward surface. frontal area of the building to the frontal area of the domain:
 Short upstream lengths resulted in a noticeable increase in pressure
on the windward face, and a decrease in pressure on the side, top, and frontal area of building BH
BR ¼ ¼ (7)
leeward faces of the building (Figure 4). The increase in pressure on frontal area of domain hð2b þ BÞ
the windward face is uniform at all heights of the building. This Since the building dimensions are fixed in this study, the blockage
suggests the absence of ABL inhomogeneity errors, which are largest ratio is a function of the domain width b and domain height h.
near the ground and decrease with building height (Abu-Zidan et al., For the BR study, a domain length of lu ¼ 3H and ld ¼ 3H is selected to
2020). reduce computational cost. These values have been shown in Figure 6 to
 The downstream length ld had minimal influence on surface pressure. produce minimal domain error. BR is varied incrementally from 0.1% to
The maximum error was <1.9% for the smallest case of ld ¼ 3H. This 2.16% by varying b and h simultaneously as per Table 1. The blockage
error occurred at the leeward face of the building (Figure 6b). ratio corresponding to the BC is 0.33%.
 The domain width b had the second biggest impact on surface pres- Once again, CP values are extracted at the centreline of windward,
sure; however, these were much smaller than errors due to a short lu. leeward, side, and top faces of the building (Figure 4e). Error is calcu-
A maximum error of 4.5% occurred for the smallest case of b ¼ 1.5H lated as variation from the largest domain in the BR study with b ¼ h ¼

5
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

(a) (b) (c)

1.0

1.5H
0.8
2H
3H
0.6
5H (BC)
/

10H
0.4

0.2

0.0
0.2 0.4 0.6 0.8 1.0 1.2 -0.4 -0.3 -0.2 -0.8 -0.6 -0.4 -0.2
Windward Leeward Side
(d)
-0.3
-0.4
-0.5
-0.6
Top

-0.7
-0.8
-0.9
-1.0
-1.1
0 0.25 0.5 0.75 1
/

Figure 5. Impact of domain width b on surface pressure coefficient CP along centreline of (a) windward, (b) leeward, (c) side, and (d) top surfaces of building
in Figure 4e.

10H (BR ¼ 0.1%). The maximum error due to BR is presented in Figure 7 investigated with cases at full-scale and 1:400 scale in both the BC and
for each of the building surfaces. the OD. The resulting windward CP profiles for the two parameters are
Figure 7 shows that blockage error manifests predominantly on the shown in Figure 8.
windward surface of the building, where a higher BR resulted in larger Figure 8 shows that the impact of wind speed and geometric scale on
pressures on that surface. This behaviour is comparable to blockage ef- domain errors are almost negligible. The plots for Uref ¼ 10 m/s and Uref
fects observed in wind tunnels (Holmes, 2015). Variations in surface ¼ 60 m/s are identical when performed in similar domain sizes, and the
pressure also occur on the leeward, side, and top faces, although these are same is true for the full-scale and 1:400 cases. The minor (and almost
secondary in magnitude. For the largest blockage ratio case (BR ¼ negligible) variations in the CP plots occurred mainly due to domain size
2.16%), the maximum error at the windward face was 7.3%, while the (i.e. between BC and OD), regardless of the wind speed and geometric
maximum error on the other three faces was smaller (1.8%–3.5%). scale. This suggests that domain optimisation studies using steady RANS
may be performed for any reasonable values of wind speeds and geo-
3.3. Sensitivity study on the impact of wind speed and geometric scale metric scales.

Two additional parameters are investigated for their impact on 4. Discussion

domain errors; namely the mean wind speed and the geometric scale of
the model. These two parameters are analysed for two domain sizes: the Based on the results of the sensitivity studies, four main sources of
BC domain size, and an optimised domain (OD) size that is selected based domain errors have been identified: (1) wind-blocking errors due to short
on the findings from the studies of domain extents and blockage ratio. upstream length lu, (2) flow recirculation errors due to insufficient
The dimensions of the OD are selected so that all maximum errors are downstream length ld, (3) global venturi effects (GVE) due to large
limited to <4% on all faces. The resulting OD dimensions are lu ¼ 3H, ld ¼ blockage ratios, and (4) local venturi effects (LVE) due to insufficient
3H, b ¼ 3H, and h ¼ 4H, with a corresponding blockage ratio of 0.81%. clearance between the building and top and lateral domain boundaries.
The OD represents an 87.5% reduction in volume from the BC, and a 40% Each of these is discussed in the following subsections.
reduction in mesh count (1.4 106 in the OD). The maximum CP error in
the OD occurred at the windward surface, with a value of 3.8%. 4.1. Wind-blocking effects
The impact of wind speed on domain errors is investigated by simu-
lating cases for Uref ¼ 10 m/s and Uref ¼ 60 m/s in both the BC and the The largest source of domain error in this paper is attributed to the so-
OD. Similarly, the impact of geometric scale on domain errors is called wind-blocking effect that is caused by short upstream distances.

6
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

(a) (b)
25 Windward face 2 Windward face
3
1 Leeward face Leeward face
20 Side face 1.6 5 Side face

Maximum error (%)

Maximum error (%)
Top face Top face

15 1.2 8
1.5 10
10 0.8
15
2 (BC)
5 0.4 20
3 5
4 (BC) 7
0 0
1 2 3 4 5 6 7 0 5 10 15 20
(multiples of ) (multiples of )

(c) (d)
5 Windward face 2 Windward face
1.5 Leeward face 2.2 Leeward face
4 Side face 1.6 Side face
Maximum error (%)

Maximum error (%)

Top face Top face

3 1.2
2 3
2 0.8

3 4
1 5 0.4 6
4 (BC)
(BC) 7 8
0 0
1 2 3 4 5 6 7 2 4 6 8
(multiples of ) (multiples of )

Figure 6. Impact of domain extents lu, ld, b, and h on domain error of surface pressure. Maximum error (%) computed from centreline CP at windward, leeward, side,
and top building surfaces.

The term wind-blocking effect is used in this paper to refer to “the Wind-blocking errors arise when the upstream domain length lu is
disturbance of the wind-flow pattern by the presence of the building, and too short to fully contain the wind-blocking region that forms in front of
the associated decrease of the upstream stream-wise wind-velocity the building. This results in an interaction between the inlet boundary
component near the building” (Blocken and Carmeliet, 2006). This and the wind-blocking region, as shown in Figure 9a. When the up-
should not be confused with blockage effects that are caused by a large stream length is too short, the inlet velocity profiles will be imposed at a
blockage ratio BR. The latter will henceforth be termed global venturi location where the flow, realistically, ought to slow down and deflect
effect (GVE) to avoid confusion. around the building. This causes overprediction of positive pressures on
the windward surface (Figure 4a), and overprediction of negative
pressures on the leeward, side, and top surfaces (Figure 4b,c,d). Figure 9
10 Windward face shows how increasing lu allows the wind-blocking region to fully
Leeward face develop.
=
Side face It is important to note that quantifying wind-blocking error in isola-
2.16%
8 tion is not possible without achieving horizontal homogeneity of the
Top face
Maximum error (%)

= 1.59% atmospheric boundary layer (HHABL). Wind-blocking errors are sensi-

= tive to upstream domain length lu, and so too are ABL inhomogeneity
6 1.28% errors (Abu-Zidan et al., 2020). If inhomogeneity errors are present in the
simulation, then lu will influence both wind-blocking errors and in-
= 0.33% = 0.81%
(BC) homogeneity errors simultaneously, such that increasing lu will minimise
4
wind-blocking error while maximising inhomogeneity errors (and vice
versa). Demonstrating horizontal homogeneity will confirm that in-
2 homogeneity errors are absent in the simulation which will allow for
quantification of domain errors in isolation. To the authors’ best
knowledge, quantification of wind-blocking errors in isolation has not
0 previously been performed in CFD studies of tall buildings.
0.0 0.5 1.0 1.5 2.0 2.5
Blockage ratio (%)
4.2. Flow recirculation
Figure 7. Impact of blockage ratio on domain error of surface pressure.
Maximum error (%) computed from centreline CP at windward, leeward, side, Flow recirculation error is caused by an insufficient downstream
and top building surfaces. domain length ld, where the outlet boundary interacts with the wake

7
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

(a) (b)
1.0 1.0
Wind speed Geometric scale
Uref=60 (BC) Full scale (BC)
0.8 Uref=60 (OD) 0.8 Full scale (OD)
Uref=10 (BC) 1:400 (BC)
0.6 Uref=10 (OD) 0.6 1:400 (OD)
/

/
0.4 0.4

0.2 0.2

0.0 0.0
0.2 0.4 0.6 0.8 1.0 1.2 0.2 0.4 0.6 0.8 1.0 1.2

Windward Windward
Figure 8. Impact of (a) wind speed and (b) geometric scale on windward surface CP.

Figure 9. Wind-blocking effect visualised with positive pressure contours (log scale) for different upstream lengths lu. Dashed red lines represent location of inlet
plane. Flow direction from left to right.

region of the building. This error is often detected during the solution The recommended ld ¼ 15H by Franke et al. (2007) seems overly
process as it tends to cause numerical instability that prevents conver- conservative considering that ld ¼ 3H was seen to be sufficient in
gence. Many CFD codes such as FLUENT alert the user to the presence of the current study. Nonetheless, larger ld may still be required in
recirculation error during the solution process. unsteady simulations in order to accurately capture the shedding
Flow recirculation error was insignificant for the domain sizes in this vortices in the wake (Revuz et al., 2012). Practical experience
study. Even the smallest ld ¼ 3H case allowed for full development of the suggests that shedding vortices in unsteady simulations tend to
wake region (Figure 10) and resulted in errors of magnitude <2% travel farther downstream compared to the static wake region in
(Figure 6b). Visual assessment of the wake region (as done in Figure 10) steady-state simulations. In such a case, special boundary conditions
is a crude method for identifying domains that are too short. However, may be employed to artificially dissipate vortices as they leave the
quantification of error may still be required when optimising ld. computational domain.

8
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

Figure 10. Flow recirculation region visualised with velocity contour plots and streamlines. Flow direction from left to right.

4.3. Global venturi effects (GVE) Kwon, 1998), while CWE guidelines specify a more stringent ratio of
<3% (Blocken, 2015; Tominaga et al., 2008). The findings of the current
As defined previously, the term global venturi effect (GVE) refers to study support those by Revuz et al. (2012) that suggest even a smaller
the commonly cited blockage phenomenon in wind tunnel tests, where blockage ratio for CWE may be required. As shown in Figure 7, the BR ¼
global acceleration of flow occurs due to a large blockage ratio (Choi and 2.16% case still resulted in a maximum error of 7.3% on the windward
Kwon, 1998; Holmes, 2015; Takeda and Kato, 1992). This behaviour has face of the building.
also been observed in CFD studies and is considered a primary source of
error in small domains (Blocken, 2015; Franke et al., 2007; Revuz et al., 4.4. Local venturi effects (LVE)
2012).
While GVE errors occur in both CFD and wind tunnel tests, it is Local venturi effects (LVE) are caused by insufficient clearance be-
important to note that the two may not be equivalent even for the same tween the building model and the lateral or top boundaries of the
blockage ratio. This is because the top and side boundaries in CFD do not domain. The symmetry condition specified at the top and lateral
typically represent the solid walls of the wind tunnel. In some CFD boundaries is only appropriate if these boundaries are located far from
studies, it may be necessary to replicate GVE by sizing the domain ac- the influence of the building. Otherwise, the symmetry condition will
cording to the cross-sectional dimension of the wind tunnel and treating enforce parallel-flow and zero-flux conditions that are non-physical,
the top and side boundaries as solid walls (Franke et al., 2007). This may resulting in flow field errors. These errors are concentrated at the
be required in validation studies based on small wind tunnels where boundaries and propagate inwards towards the building, eventually
blockage effects are inevitable. resulting in surface pressure errors.
GVE errors are controlled by limiting the blockage ratio. In wind Because LVE and GVE are both influenced by the same domain pa-
tunnel tests, a blockage ratio of < 5–10% is typically allowed (Choi and rameters (b and h), they are interrelated and often intertwined in an

Figure 11. Effect of domain width b on flow field errors. Flow direction into page. LVE ¼ local venturi effect.

9
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

indecipherable way. Nonetheless, they are two distinct sources of domain which resulted in a maximum courant flow number of less than 5 for the
error and they impact the results differently. Errors due to LVE dominate selected grid. The model is solved for 10,000 timesteps (100s) to initi-
in cases of small local clearance and small blockage ratio, while GVE alise the solution and then solved for a further 35,000 timesteps (350s)
errors dominate in cases with large blockage ratio. The impact of LVE over which instantaneous values of pressure and velocity were averaged.
errors on building surface pressure is largest on the side, top, and leeward Aside from varying the domain size, all numerical parameters are held
surfaces of the building (Figure 6c), while GVE errors are largest on the constant between the cases to ensure that the influence of domain size is
windward surface (Figure 7). To illustrate the differences between LVE assessed in isolation. The resulting time-averaged CP profiles for the base
and GVE, contour plots of wind speed error are presented in Figure 11, case and optimised domain are plotted in Figure 14.
Figure 12, and Figure 13 for the b, h, and BR sensitivity studies, respec- Figure 14a demonstrates that the optimised domain was able to
tively. The contour plots illustrate wind speed error in the cross-sectional accurately replicate the windward pressure profiles of the base case,
plane through the centre of the building, where the direction of flow is where the average deviation between the two cases was 0.8%. On the
into the page. leeward and side faces, however, Figures 14b,c show that the optimised
Figure 11 presents two cases from the domain width study (b ¼ 1.5H domain resulted in a noticeable overprediction of negative pressure
and b ¼ 3H) where LVE errors are dominant. These errors are concen- compared to the base case, where the average deviation on the leeward
trated at the side boundaries of the domain. As b is increased from 1.5H in and side faces was 10.7% and 7.7%, respectively.
Figure 11a to 3H in Figure 11b, the magnitude of LVE error decreases To investigate the cause of this discrepancy, a third case (case 3) is
drastically due to increased side clearance. Nonetheless, LVE remains performed with an upstream length lu equivalent to that of the base case,
dominant over GVE in both cases. This is due to the large domain height and all other domain dimensions equivalent to the optimised domain.
of 6H, which ensures a small blockage ratio and prevents GVE errors from The results, also plotted in Figure 14, show close agreement to those of
dominating. the base case, where the average deviation on the windward, leeward,
In Figure 12, two cases from the domain height study are presented (h and side faces was 0.1%, 3.7%, and 1.1%, respectively. This suggests that
¼ 2.2H and h ¼ 3H), where the GVE and LVE errors are equally dominant. the primary cause of discrepancies in the results from the OD and BC is
As the height of the domain is increased from 2.2H in Figure 12a to 3H in due to differences in upstream domain length lu, but it is unclear whether
Figure 12b, the magnitude of both LVE and GVE errors decrease. LVE these variations are caused by the wind-blocking effect observed in
errors drop due to added clearance between the roof of the building and steady RANS results. It is also unclear whether optimisation of lu using
the top boundary, while GVE errors drop due to a reduction in overall BR. steady RANS could potentially translate to unsteady LES. Further
The equally dominant LVE and GVE errors seen in Figure 12a align with research is needed to study the influence of upstream domain length on
the findings from Figure 6d, where the maximum error on the windward LES results.
surface due to GVE is roughly the same magnitude as the maximum error Nonetheless, the results of this study suggest that steady RANS can be
on the side, leeward, and top surfaces due to LVE. effective for optimising the height h, width b, and downstream length ld
Finally, Figure 13 presents three cases from the blockage ratio study of the domain for use in LES. Optimising these parameters resulted in a
that demonstrate the interrelated nature of LVE and GVE. For the case of 50% reduction of computational time compared to the base case.
high BR ¼ 2.16% in Figure 13a, GVE errors are significant and occur
throughout the entire domain cross-section. LVE errors are also present in 6. Framework for domain optimisation study
that case due to the small side and top clearances. As BR is reduced to
0.81% in Figure 13b, both GVE and LVE errors significantly drop in The findings of this study indicate that domain sizing guidelines by
magnitude compared to Figure 13a. The colour scale had to be adjusted Franke et al. (2007), although sufficient for eliminating domain errors,
to visualise the error pattern. As the BR is further reduced to 0.33% in are exceedingly conservative when applied to tall building simulations.
Figure 13c, LVE errors significantly drop in comparison to GVE errors, Optimising the domain size is strongly recommended as it can lead to
and are no longer visible in the plot. This indicates that for larger domain significant reduction in mesh size and computational cost. These savings
sizes with small blockage ratios, GVE errors will likely dominate LVE. become critical when performing unsteady simulations that involve
Hence, domain sizing guidelines based on maximum BR may be adequate thousands of timesteps.
for controlling both GVE and LVE errors. Domain size optimisation is highly problem-specific and depends on
multiple factors that include the building geometry and orientation, as
5. Large-eddy simulation of optimised domain well as the acceptable level of error for the application of interest. This
level of complexity is difficult to codify. Hence, instead of providing
To extend the findings of this study to unsteady large-eddy simula- specific sizing guidelines, this section proposes a general framework for
tions (LES), additional cases are performed where LES results from the optimising the computational domain size:
base case (BC) and optimised domain (OD) are compared. Details of these
cases are presented in Table 2. 6.1. Aim and requirements of domain optimisation study
For each of these cases, unsteady LES is performed with the Dynamic
Smagorinsky subgrid model (Lilly, 1992). The segregated algorithm PISO The aim of a domain size optimisation study is to identify the smallest
is used for pressure-velocity decoupling (ANSYS Inc, 2013). Momentum possible domain size for which domain error remains within the
terms are discretised using a bounded central differencing scheme, acceptable tolerance. Domain errors are estimated by comparing vari-
pressure terms are discretised using a second-order scheme, and temporal ables of interest of the optimised domain with corresponding values in an
discretisation is achieved with a bounded second-order implicit scheme. excessively large domain. When setting up a domain optimisation study,
A non-iterative time advancement (NITA) solver is used to reduce the following requirements should be taken into consideration:
computational cost by removing the need for outer iterations.
A mean wind speed of Uref ¼ 30 m/s at zref ¼ 100 m is selected, which  Domain errors should be assessed in isolation from other sources of
corresponds to serviceability design wind speed of 25-year return period error. Thus, ABL homogeneity is a critical requirement. HHABL
in Melbourne, Australia (Standards Australia, 2011). Inflow turbulence conditions can be achieved in FLUENT for RNG k-ϵ model by adopting
was ignored to reduce computational cost and to prevent potential in- the approach described by Abu-Zidan et al. (2020). Moreover, mesh
homogeneity errors that may arise due to the decay of inflow turbulence variation between the different domain sizes should be limited,
in the upstream fetch of the domain. A timestep of Δt ¼ 0.01s is selected particularly near the building model. For irregular geometries and

10
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

Figure 12. Effect of domain height on flow field errors. Flow direction into page. LVE ¼ local venturi effect; GVE ¼ global venturi effect.

unstructured grids, this can be achieved with a nested meshing Solution convergence should be verified by monitoring variables of
approach. interest during the solution process.
 Proper convergence of the solution is critical in domain optimisation  Wind velocity and geometric scale were found to have minimal
studies as error due to unconverged results can be significant. Nega- impact on domain errors for steady RANS. Hence, domain optimisa-
tive pressures on the side and leeward building surfaces are particu- tion studies can be performed for any reasonable wind speed profile
larly volatile, requiring many iterations before reaching convergence. and geometric scale.

Figure 13. Effect of blockage ratio on flow field errors. Flow direction into page.

11
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

Table 2. Details of LES cases.

Case 1 (BC) Case 2 (OD) Case 3

lu 5H 3H 5H
ld 15H 3H 3H
b 5H 3H 3H
h 6H 4H 4H
Cell count (106) 2.3 1.4 1.5

(a) (b) (c)

1.0
Case 1 (BC)
0.8 Case 2 (OD)
Case 3
0.6
/

0.4

0.2

0.0
0.2 0.4 0.6 0.8 1.0 -0.8 -0.7 -0.6 -0.5 -0.8 -0.7 -0.6
Windward Leeward Side
Figure 14. Comparison of LES results for base case (BC) and optimised domain (OD).

6.2. Optimising the domain dimensions blockage ratio of <1% is recommended. This stringent value
would ensure that both GVE and LVE errors are contained within
The domain size is optimised based on sensitivity studies that involve acceptable limits. Nonetheless, it may be worthwhile verifying the
parametrising each of the four domain dimensions (lu, ld, b, and h). absence of these errors with sensitivity studies.
Appropriate domain dimensions are then selected based on the accept-
able level of domain error in the variables of interest.
7. Conclusion
 Selecting upstream length lu
An adequate upstream domain length should be selected to limit This paper investigates the effect of computational domain size
the magnitude of wind-blocking errors. lu can vary with building on CFD predictions of wind loads on tall buildings. Four distinct
geometry and orientation, as these parameters affect wind- sources of domain error are identified which include wind-blocking
blockage (Blocken and Carmeliet, 2006). The greater the errors caused by short upstream length, flow recirculation errors
wind-blocking effect and the further upstream it propagates, the due to insufficient downstream length, global venturi effects (GVE)
larger the requirement of lu. Upstream length can have a sub- due to large blockage ratios, and local venturi effects (LVE) caused
stantial impact on computational cost in unsteady simulations by insufficient clearance between the building and top and lateral
where high mesh refinement in the upstream region is required to domain boundaries. Domain errors were found to be significant for
ensure an accurate transfer of inflow turbulence from the inlet domains with a short upstream fetch and large blockage ratio, and
boundary to the building. It is unclear from the findings of this these errors can significantly compromise the reliability of the so-
study whether optimisation of lu using steady RANS will suffi- lution. A framework for optimizing domain size in building studies
ciently translate to unsteady LES. This is an area for future is proposed which involves parametrizing the domain dimensions
research. and performing sensitivity study on variables of interest. The find-
 Selecting downstream length ld ings and recommendations of this paper are intended to assist the
The downstream length has a significant impact on computational modelling process for practitioners and result in more computa-
cost since a high mesh refinement is required in the wake region to tionally efficient and reliable CFD simulations for tall buildings.
capture complex flow features. An adequate ld should be provided
to ensure that the outflow boundary is sufficiently distant from the Declarations
wake region, thus minimizing flow recirculation errors. In this
study, an optimised value of ld ¼ 3H, determined using steady Author contribution statement
RANS, was found to be sufficient for unsteady LES.
 Selecting domain width b and height h Yousef Abu-Zidan: Conceived and designed the experiments; Per-
Conservative values for b and h may be selected since these do not formed the experiments; Analyzed and interpreted the data; Contributed
significantly influence computational cost. This is because mesh reagents, materials, analysis tools or data; Wrote the paper.
refinement is not required at the lateral and top extents of the Priyan Mendis & Tharaka Gunawardena: Conceived and designed the
domain. For the rectangular building in this study, a maximum experiments; Wrote the paper.

12
Y. Abu-Zidan et al. Heliyon 7 (2021) e06723

Funding statement Buccolieri, R., Di Sabatino, S., 2007. Flow and pollutant dispersion in urban arrays for the
standardization of CFD modelling practise. In: 11th International Conference on
Harmonisation within Atmospheric Dispersion Modelling for Regulatory Purposes,
This research was supported by the Cooperative Research Centres Cambridge.
Projects Round 8 Grant CRCPEIGHT000084: Upcycling solutions for Castro, I.P., Cowan, I.R., Robins, A.G., 1999. Simulations of flow and dispersion around
hazardous claddings and co-mingled waste, the Australian Research buildings. J. Aero. Eng. 12, 145–160.
Celik, I.B., Ghia, U., Roache, P.J., Freitas, C.J., Coleman, H., Raad, P.E., 2008. Procedure
Council Industrial Transformation Research Programme IC150100023: for estimation and reporting of uncertainty due to discretization in CFD applications.
ARC Training Centre for Advanced Manufacturing of Prefabricated J. Fluid Eng. 130, 078001–078001-078004.
Housing, and an Australian Government Research Training Program Choi, C.K.K., Kwon, D.K., 1998. Wind tunnel blockage effects on aerodynamic behaviour
of bluff body. Wind Struct. 1, 351–364.
(RTP) Scholarship. Cowan, I.R., Castro, I.P., Robins, A.G., 1997. Numerical considerations for simulations
of flow and dispersion around buildings. J. Wind Eng. Ind. Aerod. 67–68, 535–545.
Franke, J., Hellsten, A., Schlünzen, H., Carissimo, B., 2007. Best Practice Guideline for the
Data availability statement CFD Simulation of Flows in the Urban Environment. COST Office, Brussels.
Hall, R.C., 1997. Application of Computational Fluid Dynamics to Near-Field Atmospheric
Data will be made available on request. Dispersion, Atmospheric Dispersion Modelling Liaison Committee Annual Report
1995-1996 (Annex B), Chilton.
Holmes, J.D., 2015. Wind Loading of Structures. CRC Press.
Declaration of interests statement Huang, M.F., Lau, I.W.H., Chan, C.M., Kwok, K.C.S., Li, G., 2011. A hybrid RANS and
kinematic simulation of wind load effects on full-scale tall buildings. J. Wind Eng.
Ind. Aerod. 99, 1126–1138.
The authors declare no conflict of interest. Huang, S., Li, Q.S., Xu, S., 2007. Numerical evaluation of wind effects on a tall steel
building by CFD. J. Constr. Steel Res. 63, 612–627.
Lilly, D.K., 1992. A proposed modification of the Germano subgrid-scale closure method.
Additional information Physics of Fluids A: Fluid Dynam. 4, 633–635.
Parente, A., Gorle, C., van Beeck, J., Benocci, C., 2011. Improved k–ε model and wall
No additional information is available for this paper. function formulation for the RANS simulation of ABL flows. J. Wind Eng. Ind. Aerod.
99, 267–278.
Ramponi, R., Blocken, B., 2012. CFD simulation of cross-ventilation for a generic isolated
References building: impact of computational parameters. Build. Environ. 53, 34–48.
Revuz, J., Hargreaves, D.M., Owen, J.S., 2012. On the domain size for the steady-state
Abu-Zidan, Y., 2019. Verification and Validation Framework for Computational Fluid CFD modelling of a tall building. Wind Struct. 15, 313–329.
Dynamics Simulation of Wind Loads on Tall Buildings. Department of Infrastructure Ricci, M., Patruno, L., Kalkman, I., de Miranda, S., Blocken, B., 2018. Towards LES as a
Engineering. The University of Melbourne. design tool: wind loads assessment on a high-rise building. J. Wind Eng. Ind. Aerod.
Abu-Zidan, Y., Mendis, P., Gunawardena, T., 2020. Impact of atmospheric boundary layer 180, 1–18.
inhomogeneity in CFD simulations of tall buildings. Heliyon 6, e04274. Richards, P.J., Norris, S.E., 2011. Appropriate boundary conditions for computational
Ai, Z.T., Mak, C.M., 2015. Large-eddy Simulation of flow and dispersion around an wind engineering models revisited. J. Wind Eng. Ind. Aerod. 99, 257–266.
isolated building: analysis of influencing factors. Comput. Fluid 118, 89–100. Standards Australia, 2011. Structural Design Actions. Part 2: Wind Actions, AS/NZS
ANSYS Inc, 2013. ANSYS Fluent Theory Guide, USA. 1170.2:2011. Standards Australia. New South Wales, Sydney.
Baetke, F., Werner, H., Wengle, H., 1990. Numerical simulation of turbulent flow over Takeda, K., Kato, M., 1992. Wind tunnel blockage effects on drag coefficient and wind-
surface-mounted obstacles with sharp edges and corners. J. Wind Eng. Ind. Aerod. 35, induced vibration. J. Wind Eng. Ind. Aerod. 42, 897–908.
129–147. Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M.,
Blocken, B., 2015. Computational Fluid Dynamics for urban physics: importance, scales, Shirasawa, T., 2008. AIJ guidelines for practical applications of CFD to pedestrian
possibilities, limitations and ten tips and tricks towards accurate and reliable wind environment around buildings. J. Wind Eng. Ind. Aerod. 96, 1749–1761.
simulations. Build. Environ. 91, 219–245. Xiang, W.Z., Wang, H.S., 2007. Discussion on Grid Size and Computational Domain in
Blocken, B., Carmeliet, J., 2006. The influence of the wind-blocking effect by a building CFD Simulation of Pedestrian Wind Environment Around Buildings, Building
on its wind-driven rain exposure. J. Wind Eng. Ind. Aerod. 94, 101–127. Simulation, Beijing, China, p. 1150.
Braun, A.L., Awruch, A.M., 2009. Aerodynamic and aeroelastic analyses on the CAARC Zhang, Y., Habashi, W.G., Khurram, R.A., 2015. Predicting wind-induced vibrations of
standard tall building model using numerical simulation. Comput. Struct. 87, high-rise buildings using unsteady CFD and modal analysis. J. Wind Eng. Ind. Aerod.
564–581. 136, 165–179.

13