Download as pdf or txt
Download as pdf or txt
You are on page 1of 22

1 Computational fluid dynamics simulation of tree effects on pedestrian

2 wind comfort and wind safety around coastline building resort

ed
3 Husna Aini Swarnoa, Nurul Huda Ahmada*, Ahmad Faiz Mohammadb, Nurnida Elmira Othmanc
4
5 a Razak Faculty of Technology and Informatics (RFTI), Universiti Teknologi Malaysia Kuala Lumpur, Jalan
6 Sultan Yahya Petra, 54100 Kuala Lumpur, Malaysia

iew
7 b Malaysia-Japan International Institute of Technology (MJIIT), Universiti Teknologi Malaysia Kuala Lumpur,

8 Jalan Sultan Yahya Petra, 54100 Kuala Lumpur, Malaysia


9 c School of Mechanical Engineering, College of Engineering, Universiti Teknologi Mara (UiTM), 40450 Shah

10 Alam, Selangor, Malaysia


11
12 *corresponding author: nurulhuda.kl@utm.my

13 Abstract

ev
14
15 This paper discusses the quantitative effect of vegetative windbreak on wind velocity at a resort in Langkawi,
16 Malaysia. The implementation of vegetative windbreaks was proposed to reduce the wind velocity and improve
17 pedestrian wind comfort, especially in the coastline areas in Cenang, Langkawi. Natural windbreaks such as
18 trees are efficient barriers against high-velocity winds. The windbreaks exert a drag on the wind, causing a net

r
19 loss of momentum. However, there are no in-depth studies of the effect of tree windbreaks on wind velocity in
20 coastal areas. Hence, suitable tree species to be used as windbreaks in coastline areas remain unanswered. In
21 this study, a computational fluid dynamics (CFD) simulation was conducted to investigate the flow
er
22 characteristics around tree windbreaks. In addition, the effects of tree porosity, canopy height, and distance
23 between the trees were also investigated. The main factors that could affect the efficiency of the windbreaks
24 include tree height, width, arrangement, and porosity. The simulation technique allows the trees to be modelled
25 as porous media, where the aerodynamic properties of the trees are performed in the model. The most common
pe
26 tree found in Cenang is the Rhu tree, which is scientifically known as Casuarina equisetifolia (CE), with a drag
27 coefficient of 1.56. The simulation provided a means to analyze the effect of single-row trees in reducing the
28 velocity of the wind flowing towards the building as well as the wind comfort at the pedestrian height. Finally,
29 the simulations revealed that in terms of percentage reduction of U/Uref between a single row of CE and no
30 windbreak at the pedestrian height are 1H = 8.69%, 2H = 17.72%, 3H = 18.04%, and 4H = 15.45%. Thus, it
31 can be deduced that CE trees could act as tree windbreaks and protect buildings against storms and strong
32 wind. This outcome is expected to guide future researchers in their studies of wind comfort and thus contribute
ot

33 to improving the quality of the environment in coastline areas.


34
35 Keywords: strong wind, coastal area, windbreak, Rhu, Casuarina equisetifolia, wind comfort.
tn

36 1 Introduction
37
38 The development of infrastructure, such as road networks and commercial properties, is taking place all
39 over Malaysia, especially at tourism sites. One such site is the Langkawi Island, which is undergoing rapid
40 urbanisation and modernisation to accommodate thousands of tourists from around the globe. Although such
41 development is necessary, it may lead to uncontrolled urbanization in the surrounding areas and this in turn
rin

42 would pose challenges to conserving the heritage value of the island [1],[2]. In addition, the high-speed and
43 grand-scale development has increased the deforestation rate and the community is now facing various
44 environmental impacts. Rapid unplanned development and land clearing have fragmented the natural
45 environment particularly the forest and threatened its biodiversity. Meanwhile, land reclamation projects along
46 the coastline for the development of beachside accommodation have caused water pollution and soil erosion
ep

47 [3].
48 Commercial developments in several tourist destinations in Langkawi, namely Pekan Kuah, Chenang
49 Beach, and Tanjung Rhu, have caused damage to the natural environment and coastal areas, such as coastal
50 degradation and soil erosion. Moreover, most land reclamations for the high-end development of the coastal
51 areas, which is known as the “Langkawi New City” project, have disrupted the ecological systems and affected
52 the stability of the island ecosystem. What is worse is that there is no area reserved for wildlife, according to a
Pr

53 statement from Langkawi District Office [4],[5]. Fortunately, in 2007, Langkawi has been declared a Global
54 Geopark by the United Nations Educational, Scientific and Cultural Organization (UNESCO). Therefore,

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 Langkawi Development Authority (LADA) aims to ensure the development in Langkawi is carried out
2 sustainably [6].
3 Due to the unsystematic development, the sustainability and aesthetic appeal of tourist destinations can be

ed
4 negatively affected. Examples of such development include the commercial development in Kuah, chalet
5 developments at Chenang Beach, and the development projects in Tanjung Rhu [7]. The rapid development of
6 high-end residential buildings, commercial properties, and infrastructure in coastal areas could change the
7 landscape of the island. For example, in Langkawi, mangrove forests and beach areas have been destroyed as
8 a result of development [8]. Mangroves can adapt to and withstand harsh environmental conditions such as

iew
9 storms, strong wind, high temperature, and high salinity. Most important of all, they can provide coastal
10 protection [9].
11 The high-frequency occurrence of severe storm and squall over the sea, which are accompanied by wind
12 and wave of varying strength, poses a potential risk to the coastlines and their inhabitants. On 18th June 2018,
13 a strong wind of up to 50–60 km/h was recorded, in addition to heavy rainfall and thunderstorm that caused
14 trees to be uprooted and electric poles to collapse [10]. The four-day event of the Tropical Storm Pabuk, which
15 occurred from 31st December 2018 to 4th January 2019, brought strong winds of up to 30–50 km/h, surging
16 waves, and heavy rains to Langkawi. The tropical storm had caused severe damage to several buildings and

ev
17 houses [11]. The presence of mangroves near the coastlines helps in altering the wind friction by increasing the
18 surface roughness of the coastlines. In addition, the evaporation and momentum transfer are also affected by
19 the mangroves, resulting in changes in the internal boundary layer.
20 On 9th August 2019, Typhoon Lekima hit Kampung Yooi, Langkawi and brought strong winds and
21 thunderstorms, which uprooted trees and damaged the infrastructure. On 7th November 2020, the storm and

r
22 strong winds once again hit Kampung Yooi with a wind force that was capable of ripping off the roofs of
23 several residential [12]. These extreme weather phenomena, which are partly due to rapid urbanization in
24 Langkawi, caused discomfort to the residents and tourists.
25
26
er
It has been observed that the severe storm and strong winds can affect the buildings structure especially in
coastal areas. Therefore, further research is needed to investigate and analyze the wind flow characteristics
27 around coastal buildings that are exposed to storms and strong wind. In addition, the windbreak strategies
28 involving vegetation, such as planting single or multiple rows of trees, should also be investigated. This can
pe
29 potentially reduce the impact of strong winds near the coastline. Windbreaks and shelterbelts are effective
30 barriers to reducing wind velocity and altering wind fields [13],[14].
31 The function of a natural or living windbreak, which can consist of single or multiple rows of trees or
32 shrubs, fences, or other materials. As the wind blows against it, the air pressure on the windward side increases,
33 which causes large quantities of air to move up and over the top or around the ends of the windbreak [15].
34 Apart from protecting areas such as agricultural farms and residential houses, the reduction in wind velocity
35 behind the windbreak could alter the environmental conditions or microclimate of the sheltered zone.
ot

36 According to United States Department of Agriculture [15], the wind velocity on the leeward side could be
37 reduced approximately six times, depending on factors such as height, density, width, shape, arrangement, and
38 porosity of the windbreak [16].
39 A computational fluid dynamics (CFD) simulation is an effective tool for quantitative visualization of
tn

40 wind flows around tree windbreaks [17]. However, to increase the accuracy of the simulation, significant data
41 such as the drag coefficient of the trees are required for the simulation process [18]. In this study, the simulation
42 was performed using models of vegetation that are suitable as windbreaks and are commonly found in the
43 coastal areas of Cenang, Langkawi as well as on most of the reclaimed lands in Malaysia.
44 A three-dimensional model of the common Rhu tree (Casuarina equisetifolia) was developed. The aim was
45 to simulate the tree windbreaks in their natural environment. The drag value of the experimental tree, which
rin

46 was determined in a previous study [19], was employed as input during the setting up of the boundary
47 conditions of the tree, which was modelled as a porous medium. By using the simulation technique, a
48 comprehensive numerical analysis of the effect of the windbreaks in reducing wind velocity, especially on the
49 leeward side, could be performed. The results obtained can be used to design an effective windbreak system
50 for the reclaimed lands and the coastal areas in Malaysia. In addition, they can also be utilized in future
ep

51 experimental and simulation studies of the effect of artificial or natural windbreaks and complex topographies
52 on wind flows.

53 1.1 Site Description


54
Pr

55 Langkawi was selected as the site location because of its increasing number of tourists and its large urban
56 development area of approximately 36,342 hectares. According to the Langkawi Municipal Council [4], the
57 island has a hilly topography and is surrounded by a forest reserve, which has an area of about 26,266 ha and

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 covers 54.6% of the total land area. To fulfill the ever-increasing demand of the tourism industry, the urban
2 development is mostly concentrated in the Cenang area. Several commercial buildings, hotels, resorts, and

ed
3 transportation networks are built to accommodate the large number of tourists [20].
4 Cenang was the focus of this study as it is the main tourist area on the island and many development
5 projects are still ongoing in this area. The Cenang beach, which is well-equipped and has several beach resorts,
6 is the most popular tourist attraction [22]. The Holiday Villa Beach Resort was chosen because of the
7 topography and location of its three-story building, which were ideal for this study. Figure 1 shows the layout
8 of the resort, which is situated at Jalan Pantai Cenang, Langkawi.

iew
9
(a) (b)

ev
10

r
11 Figure 1 The site location at Jalan Pantai Cenang, Langkawi: (a) top view, and (b) side view. The yellow
12 dashed box in (a) denotes the Holiday Villa Beach Resort [23],[24].
13 er
14 The resort is surrounded by coconut and Syzygium grande trees. Under strong wind conditions, the
15 coconuts might fall from the trees and this can be dangerous to the pedestrians. According to Foale [25], the
16 weight of a mature coconut vary from 600 g to 3 kg, depending on the type of palm. Therefore, the pedestrians
17 should be notified of the danger of falling coconuts near the resort to prevent any injuries or unwanted incidents.
pe
18 The Syzygium grande trees can be found throughout the tropical Asia region in a large variety of habitats [26].
19 These include coastal forests, which can be commonly found in Singapore. The common names of the tree
20 include Sea Apple and Jambu Laut or Jambu Air Laut [27]. It is capable of growing up to 30 m in canopy
21 height [28]. However, this tree does not tolerate strong wind and waterlogged conditions because of its fragile
22 wood, as reported by Whistler and Elevitch [29].
23
ot

24 2 Methodology

25 2.1 Numerical Model


26
tn

27 The computational fluid dynamics (CFD) method is gaining importance nowadays in investigations related
28 to wind engineering. The technique is widely used due to its advantages, which include low cost, effective
29 computational time, and flexibility in measuring various parameters. In this study, the simulation was based on
30 the Reynolds-averaged Navier–Stokes (RANS) equations and a non-rotating, non-hydrostatic, incompressible
31 airflow system was considered. The turbulence was parameterized using the renormalization group (RNG) k-ε
rin

32 turbulence closure scheme [30]. To account for the airflow pressure loss caused by trees, the tree drag terms
33 added to the equations of momentum, the turbulent kinetic energy, k, and turbulent kinetic energy dissipation
34 rates, ε, as in Ries and Eichhorn [31] and Balczo et al. [32]. The RANS for momentum and mass conservation
35 can be expressed as follows [33]:
36
ep

∂𝑈𝑖 ∂𝑈𝑖 1∂𝑃 ∗ ∂2𝑈𝑖 ∂ 1


37 ∂𝑡
+ 𝑈𝑗 ∂𝑥 =‒ 𝜌 ∂𝑥 + 𝑣∂𝑥 𝑥 ‒ ∂𝑥 (𝑢𝑖𝑢𝑗) + 𝜌𝐹𝑡𝑟𝑒𝑒,𝑖 (1)
𝑗 𝑖 𝑖 𝑗 𝑗

38
∂𝑈𝑗
39 =0 (2)
Pr

∂𝑥𝑗
40

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 where 𝑥𝑖 is the 𝑖th cartesian coordinate (𝑖 = 1, 2, 3), 𝑈𝑖 is the the 𝑖th mean velocity component, and 𝑃 ∗ is the
2 deviation of pressure from the reference value. 𝜌 and 𝑣 are the air density and kinematic viscosity respectively,

ed
3 whereas 𝑢𝑖 is the fluctuation from the 𝑖th mean velocity component. The Reynolds stresses [33] in Equation (1)
4 were parameterized as follows:
5
6 ‒ 𝑢𝑖𝑢𝑗 = 𝐾𝑚 ( ∂𝑈𝑖
∂𝑥𝑗
∂𝑈𝑗
)
+ ∂𝑥 ‒ 3𝛿𝑖𝑗𝑘
𝑖
2
(3)

iew
7
8 where 𝐾𝑚, k, and 𝛿𝑖𝑗 are the turbulent diffusivity of momentum, turbulent kinetic energy, and the Kronecker
9 delta respectively. The turbulent diffusivity of momentum was determined using k and ε [33] as follows:
10
𝑘2
11 𝐾𝑚 = 𝐶𝜇 𝜀 (4)
12

ev
13 where 𝐶𝜇 = 0.0845 is an empirical constant of the RNG k-ε turbulence closure scheme. The prognostic
14 equations of k and ε in the RNG k-ε turbulence closure scheme [33] were written as follows:
15
∂𝑈𝑖 ∂ 𝐾𝑚 ∂𝑘
16
∂𝑘 ∂𝑘
+ 𝑈𝑗∂𝑥 =‒ 𝑢𝑖𝑢𝑗 ∂𝑥 + ∂𝑥 𝜎 ∂𝑥 ( )‒𝜀+𝐹 (5)

r
∂𝑡 𝑗 𝑗 𝑗 𝑘 𝑗
𝑡𝑟𝑒𝑒,𝑘

17
∂𝑈𝑖 ∂ 𝐾𝑚 ∂𝑘 𝜀2
∂𝜀 ∂𝜀 𝜀
( )‒𝐶
18

19
∂𝑡 𝑗 𝑖
er
+ 𝑈𝑗∂𝑥 =‒ 𝐶𝜀1𝑘𝑢 𝑢𝑗 ∂𝑥 + ∂𝑥 𝜎 ∂𝑥
𝑗 𝑗 𝜀 𝑗
𝜀2 𝑘 ‒ 𝑅𝑠 + 𝐹𝑡𝑟𝑒𝑒,𝜀 (6)

20 where 𝑅𝑠 is a strain rate term given as:


21
pe
𝛾 2
𝐶𝜇𝛾3(1 ‒ 𝛾 )𝜀
0
22 𝑅𝑠 =
(1 + 𝛽0𝛾3)𝑘 (7)
23
∂𝑈𝑗 ∂𝑈𝑖 1/2
24 𝛾=𝜀
𝑘
[( ∂𝑈𝑖
∂𝑥𝑗
+ ∂𝑥
𝑖
) ]
∂𝑥𝑗 (8)
ot

25
26 where 𝐶𝜀1, 𝐶𝜀2, 𝜎𝑘, 𝜎𝜀 , 𝛾0 and 𝛽0 are empirical constants [33] specified as follows:
27
tn

28 ( 𝐶𝜀1, 𝐶𝜀2, 𝜎𝑘, 𝜎𝜀 , 𝛾0, 𝛽0) = (1.42, 1.68, 0.7179, 0.7179, 4.377, 0.012) (9)
29
30 The tree drag was represented by the sink and the source terms on the right-hand side of Equations (1), (5), and
31 (6) were expressed as follows [34],[35]:
32
rin

33 𝐹𝑡𝑟𝑒𝑒, 𝑖 =‒ 𝜌.𝑛𝑐3.𝐶𝑑.𝐿𝐴𝐷.𝑈𝑖.|𝑈| (10)


34 𝐹𝑡𝑟𝑒𝑒, 𝑘 = 𝑛𝑐 .𝐶𝑑.𝐿𝐴𝐷.|𝑢3| ‒ 4𝑛𝑐3.𝐶𝑑.𝑘.|𝑈|
3
(11)
3𝜀
35 𝐹𝑡𝑟𝑒𝑒, 𝜀 = 2𝑘.𝑛𝑐3.𝐶𝑑.𝐿𝐴𝐷.|𝑢3| ‒ 6𝑛𝑐3.𝐶𝑑.𝜀.|𝑈| (12)
36
ep

37 where 𝑛𝑐 is the fraction of the area covered by the vertical projection of leaves (set to 1) and 𝐶𝑑 is the leaf drag
38 coefficient. The leaf area density (LAD) is defined as the leaf surface area per unit volume and |𝑈| is the wind
39 velocity.
40 The governing equations were numerically solved using the semi-implicit method for a pressure-linked
41 equation (SIMPLE) algorithm and a finite volume method in a staggered grid system [33]. The velocity
Pr

42 components 𝑈𝑖 were defined at the center of the grid cell faces whereas the scalar variables were defined at the
43 center of the grid cell. This approach is more numerically accurate and stable than the grid system that stores
44 all variables at one grid point [36]. The combined advection-diffusion terms in the governing equations were
4

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 discretized using a power-law differencing scheme, which was proven to provide accurate approximations [36].
2 A fully implicit scheme with first-order accuracy was applied for the time integration [33].

ed
3

4 2.2 Boundary Condition


5
6 For the inlet plane, a power-law velocity profile was applied, as expressed by Equation (13). The inflow

iew
7 profiles of the wind velocity (k and ε) were provided by the Architectural Institute of Japan (AIJ) [37] as
8 follows:
9
𝑧 𝛼
10 𝑈(𝑧) = 𝑈𝑟𝑒𝑓 ( )
𝐻𝑟𝑒𝑓 (13)
11
𝑘(𝑧) = 𝑎(𝐼𝑢(𝑧)𝑈(𝑧))2

ev
12 (14)
13
1
𝑈𝑟𝑒𝑓 𝑧 (𝛼 ‒ 1)
14 𝜀(𝑧) = 𝐶𝜇2𝑘(𝑧)𝐻 𝛼
𝑟𝑒𝑓
( )
𝐻𝑟𝑒𝑓 (15)
15

r
16 where Uref = 8.46 m/s is the mean velocity at the reference height H, Href = 11 m, and α = 0.11. The power-law
17 exponent was set to 0.11, which was an approximation based on the open sea terrain roughness [38]. The
18 turbulence kinetic energy k(z) in Equation (14) and the turbulence dissipation rate 𝜀(𝑧) in Equation (15) were
19
er
calculated from the velocity profile U(z) with the model constant 𝐶𝜇 = 0.09 and 𝑎 = 1 [37]. The visualizations
20 of the boundary conditions are shown in Figure 2.
21
pe
ot
tn

22
23 Figure 2 Boundary conditions used in the CFD simulation.
24
rin

25 The details of the boundary conditions are explained as follows. The pressure was initially set to zero
26 inside the domain in OpenFOAM, which is open-source software. The pressure was also given a zero gradient
27 for the inlet, top, ground, and lateral sides, as well as an urban block with a uniform outlet value as the pressure
28 was assumed to be constant. For the other regions in the domain, the outlet flow was given a fully developed
29 flow condition, namely pressureInletOutletVelocity. The slip conditions were set for the top and lateral side
ep

30 regions, the no-slip condition for the ground and the urban block, and the near-wall conditions for ε and k (i.e.,
31 epsilonWallFunction and kqrWallFunction respectively). The initial values inside the domain for all parameters
32 were set to zero. Wind velocity data obtained during the wind tunnel experiment by Miyazaki and Tominaga
33 [39] served as input values for the y-component inlet velocities (uy) whereas the other velocity components (ux
34 and uz) were assumed to be zero. Table 1 lists down the grid parameters and boundary conditions of the
Pr

35 computational domain and Table 2 tabulates the settings for the calculated parameters. After the grid creation
36 was completed, the input data was imported for the calculation.
37
5

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1
2

ed
3
4
5 Table 1 Computational domain grid and boundary conditions
Inlet boundary Power-law velocity profile
Outlet boundary uniform (pressure outlet, 0 Pa)

iew
Top and sides surfaces of the domain slip
Urban block and ground no slip/wall
Mesh RNG k-epsilon/RANS
Number of elements Structural mesh (Hexahedron)
Number of iterations 100
6

ev
7 Table 2 Calculated parameter settings
Parameter Set value
Solver type Generalized geometric-algebraic multi-grid, GAMG
solver (P)
smoothSolver (U|k|omega|epsilon)

r
Discretization scheme Second-order upwind
Algorithm er Semi-implicit method for pressure-linked equations,
SIMPLE scheme
8
9 The three-dimensional (3D) geometries of the Syzygium grande (henceforth SG) and Casuarina
10 equisetifolia (henceforth CE) tree models were developed using the SketchUp software. The SG tree model
pe
11 was designed to be 30 m in height with a trunk length of 10 m and a maximum canopy width of approximately
12 15 m. The CE tree model was designed to be 30 m in height with a trunk length of 5 m and a maximum canopy
13 width of approximately 15 m. The porosity and leaf area index (LAI) of both trees were determined via a field
14 measurement at the site location using a smartphone camera and the VitiCanopy Version 1 application. The
15 porosity values of the SG and CE trees were 20.5 and 49.9% respectively. Meanwhile, the LAIs of the SG and
16 CE trees were 2.185 and 0.927 respectively. The drag coefficients Cd of the SG and CE trees were obtained
ot

17 from the previous study [19], which were 12.48 and 1.56 respectively. The shapes of the SG and CE tree models
18 were cylindrical and conical respectively. According to Jian et al. [19], the effect of the canopy shape on the
19 flow field was small. Thus, the cylindrical shape of the SG tree model would simplify the calculation and
20 improve efficiency. The design of the CE tree model was based on the actual conical shape of the tree at the
tn

21 site location. Table 3 summarizes the properties of the tree models.


22
23 Table 3 Properties of the tree models for the vegetation simulation
Tree Name Porosity Leaf Area Drag Coefficient Model Shape
Index
rin

SG 20.5% 2.185 12.48


ep
Pr

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
ed
CE 49.9% 0.927 1.56

iew
2
3

4 2.3 Computational Domain and Grid


5
6 The geometric idealization of the Holiday Villa Beach Resort building is shown in Figure 3. The

ev
7 computational domain was based on the standard geometry stated in the AIJ guidelines for performing a CFD
8 simulation of wind around buildings [37]. The height H and width W of the computational domain were 55 and
9 285 m respectively. The domain height was equivalent to 5h, where h = 11 m was the building height. The
10 entire length L of the domain was 418 m. This domain was reproduced using a grid resolution of 1 m containing
11 ground surface data, i.e., the building height while the topography was set to be flat. The number of cells and

r
12 points in the domain were 12,095,356 and 6,761,219 respectively.
13 er
pe
ot

14
15
tn

16 Figure 3 The isometric view of the built-up area at Holiday Villa Beach Resort.
17
18 The data of the prevailing wind from Northeast (NE) to East (E), was used in the simulation together with
19 the wind direction data obtained at a height of 10 m from the Langkawi Weather Station, which was forecasted
rin

20 by the Malaysian Meteorological Department at Langkawi International Airport. The data was collected from
21 2015 to 2020 [40]. From the data [40], frequency distribution of daily maximum wind velocity in Langkawi
22 from 2015 to 2020, noted that 47% of the measured wind velocity ranged from 6.0–9.0 m/s. From this data, it
23 was observed that the maximum daily wind velocity in Langkawi can reach up to 9 m/s. According to the Dutch
24 Wind Nuisance Standard (NEN 8100) wind comfort standard guideline, this value is beyond the threshold limit
ep

25 of pedestrian wind comfort, which should be less than 5 m/s [41]. Therefore, the wind velocity at the domain
26 inlet was set to 8.46 m/s at a reference height of 11 m.

27 3 Result and Discussion


Pr

28 3.1 Wind Simulation of the Resort Building


29

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 The side and top views of the resort building are shown in Figure 4. The SG trees are highlighted in the
2 red boxes in Figure 4(a) and the distances between the building, vegetation, and beachline are shown in Figure

ed
3 4(b). The ‘W’ shape of the resort building was taken into account in the study of the wind effect on the
4 buildings, to determine whether the shape of the building would result in natural or forced wind. This is
5 important so as to provide a comfortable environment for the pedestrians in the leisure areas near the building.
6 The SG trees, which are 74 m apart, are located near the swimming pool at the borderline between the grass
7 area of the resort at the beach, which is about 30 m from the beachline. The measured distances were included

iew
8 in the simulation. The large distance between the two SG trees created a big opening at the front of the resort,
9 facing the U-section and pool area.
10
11

(a)

r ev
er
12
pe
(b)
ot
tn

13
14
15 Figure 4 The (a) side view of the resort building and SG at the site location, and (b) the top view of the building
16 together with the dimensions.
rin

17
18 Figure 5 shows the wind streamlines across the pool and the U-section of the resort building. It can be
19 seen that there was a change in the wind flow towards the centre of the building, where the pool and the U-
20 section are located. This was caused by the venturi effect, which was due to the trees situated in front of the
21 building, generating a strong wind at the pedestrian level. In addition, the buildings acted as a barrier against
ep

22 the wind, thus creating a downward vortex zone, especially on the windward side of the building, and a
23 turbulence flow around the buildings [42].
24
Pr

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
(a) Downward vortex

ed
tion
Pool sec

iew
U-sectio

1
(b)

ev
on
lsecti

r
Poo
on
ecti
er U-s

2
3
4 Figure 5 The wind streamlines: (a) across the pool and the U-section (the red arrows indicate the zones, where
5 the downward vortex was generated near the building), and (b) the top view of the pool section within the
pe
6 building area.
7
8 The wind velocity characteristics in the vicinity of the resort building were analyzed at several locations
9 denoted by 1H, 2H, 3H, and 4H, where H = 11 m was the building height (see Figure 6). The reference wind
10 velocity Uref = 8.46 m/s was measured at the reference height h, which was equal to 11 m in the domain
11
ot
tn
rin

12
13 Figure 6 The lateral view of the locations, where the wind velocity characteristics were analyzed, upstream of
14 the resort building.
15
ep

16 3.2 Validating Wind Simulation via Wind Tunnel Experiment


17
18 The experiment was conducted in an open-loop wind tunnel located at Malaysia – Japan International
19 Institute of Technology (MJIIT), Universiti Teknologi Malaysia (UTM). The wind tunnel has a maximum
20 length of 22.7 m, which consists of two centrifugal fans, a wide-angle diffuser, a settling room front, a
Pr

21 contraction cone, and along test section divided into three parts. The air blow by the centrifugal fan will enter
22 the wide-angle diffuser. The diffuser allows for obtaining the best quality of air before entering the test section.
23 Meanwhile, the settling chamber which was installed with honeycombs and mesh screens helps to straighten
9

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 the airflow and reduce the turbulence effect. The contraction cone increases the speed of wind into the test
2 section by stabilizing the flow and reducing the pressure loss [43]. All data were measured at a reference

ed
3 streamwise velocity of 11 m/s at a vertical height of the building. The hot wire was calibrated every time before
4 a measurement was carried out using a pitot tube (Dwyer, series 160-36) with variable flow velocity in the
5 range of velocities U = (0-11) m/s. Both hot wire and pitot tube was brought into the wind tunnel test section
6 using a three-axis traverse system. The pitot tube calculated the wind velocity by measuring the differential
7 pressure inside the wind tunnel; the equation used for this calculation is given as equation (16) below:

iew
𝑃𝐷
𝑉 = 4.43 (16)
𝜌𝑎𝑖𝑟

8 where 𝑃𝐷 [mmH2O] represents the differential pressure measured from the pitot tube and 𝜌𝑎𝑖𝑟 [kg/m3] is
9 the air density inside the wind tunnel. The pitot tube was connected to the pneumatic cabinet for pressure

ev
10 transmission using a rubber hose. The pneumatic cabinet was directly connected to the control panel in the
11 control. Finally, the measurement was then recorded inside the wind tunnel, and the recorded data was taken
12 directly from the control panel.
13
14 Table 4 Specifications of the instruments used in the wind tunnel tests [43]

r
Item Hot Wire Hi-Speed USB Carrier Pitot Tube
Brand Dantec er National Instrument Dwyer
Model 55P11 NI USB-9162 Series 160-36
Type Constant Compact-DAQ chassis Pitot static tube
Temperature
Anemometer (CTA)
pe
Material Tungsten - Stainless steel
Diameter 5μm - 8 mm
Output Channel - 4 -
Current Consume - 500 mA -
Maximum Sampling 20,000 Hz - 1,000 Hz
Frequency
ot

15
16 The hot wire was set to collect data at the measurement frequency of 1 kHz for 10 seconds. A calibration
17 curve for the hot wire was obtained using polynomial curve fitting, shown as equation (17) below:
18
tn

𝑈 = 𝐴𝐸4 + 𝐵𝐸2 + 𝐶 (17)

19 where E is the corrected output voltage constant while A, B, and C are calibration constants. Once the calibration
20 was completed, the pitot tube was removed and the hot wire was moved to the target position. The wind speed
rin

21 was set constant at 11 m/s (the minimum speed required for the fan to be stable) inside the wind tunnel.
22 The measurement of the streamwise velocity without the building and tree was performed at the center of
23 the test section (y = 0) and started at the vertical position z = 7 mm from the wind tunnel surface. Then the
24 measurement height was vertically increased at the intervals of 1 mm (in the height range of 7 mm ≤ z ≤ 230
25 mm) with 223 measured points.
ep

26 To correlate the wind profile against building height in the wind tunnel and CFD simulation, Figure 7(a)
27 illustrate the wind profile of the wind tunnel without building at the streamwise positions. From the wind
28 profile, the calculation of the wind shear coefficient can be determined from the power law as below [44]:
29
ln (𝑣2) ‒ ln (𝑣1)
Pr

30 𝛼 = ln (𝑧 ) ‒ ln (𝑧 ) (18)
2 1

31

10

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 𝑣1 represents the wind speed at the reference position 𝑧1and 𝑣2 represents the wind speed at the elevated
2 position 𝑧2. Therefore, the wind shear coefficient can be calculated using equation (18) which is 0.11 which

ed
3 represents open surface terrain measured over smooth water surfaces according to Sisterson and Frenzen [45].
4 By comparing the results from CFD simulations and wind tunnel tests in Figure 7(b), it can be seen that
5 the normalized wind velocity obtained by the CFD simulation shows largely the same trend as the normalized
6 wind velocity in the wind tunnel test. Moreover, the CFD simulation results are closer to the wind tunnel test
7 results when the normalized wind velocity is lower than 1.1.

iew
8

(a) (b)

ev
9
10 Figure 7 Wind profile of the (a) wind tunnel test section without building at streamwise measurement positions,

r
11 and (b) normalized wind profile from wind tunnel and CFD simulation without building at streamwise
12 measurement positions. er
13
14 Further analysis was to validate the CFD simulation with building and SG tree via wind tunnel experiment.
15 The setup arrangement is illustrated in Figure 8. The building dimension in the wind tunnel is 600 mm (L) x
16 476 mm (W) x 44 mm (H) and SG dimensions in height and diameter are 120 mm and 60 mm. The distance
pe
17 between SG trees and the building is 300 mm and 300 mm, respectively as shows in Figure 8(b). The
18 measurement started at 2H, 3H, and 4H, which H referred to as building height in the wind tunnel at 44 mm.

(a) (b)
ot
tn

19

20 Figure 8 Setup diagram of the (a) wind tunnel test section with building and SG tree at streamwise
21 measurement positions, and (b) top view diagram of the wind tunnel test section with building and SG tree at
rin

22 streamwise measurement positions.


23
24 Figure 9 shows the comparison of normalized wind velocity between numerical simulations and wind
25 tunnel tests. The CFD results show very similar behavior as the experimental results and there is quite a good
ep

26 agreement for the entire range of measurement. The good results from these validation studies provide
27 additional support for the analyses of the CFD results.
Pr

11

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
ed
iew
1

r ev
2
3 Figure 9 Comparison of normalized wind velocity between numerical simulations and wind tunnel tests for
4 location 2H, 3H and 4H. er
5 3.3 Wind Simulation Results for the Pool Section of the Resort
6
7 The streamwise velocity measured at the distance of 1H, 2H, 3H, and 4H upstream of the building is
pe
8 presented in Figure 10. The simulation results for the pool section, which is an open area, show that the wind
9 velocity increased as the distance from the building increased. For example, at the distance of 1H, the vertical
10 wind velocity measured at the pedestrian height of 1.5 m was 3.04 m/s, while at the distance of 2H, 3H, and
11 4H, the vertical wind velocity was 5.41, 6.13, and 6.40 m/s respectively. It shows that the quality of the ambient
12 wind environment at the base of a building is influenced by the layout of the building. The street-level winds
13 are intensified as the layout allows the high-elevation strong winds to flow down the façade of the structure
ot

14 [46]. This process is commonly known as downwash, which increases the wind velocity at the pedestrian level
15 and causes turbulent wind conditions [47]. Nevertheless, a thorough understanding of the wind comfort criteria
16 is still required.
tn
rin
ep

17
18 Figure 10 Vertical wind velocity profiles at four locations near the pool section
Pr

19

12

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 Since there is no standard for determining wind comfort, the guideline from the Dutch Wind Nuisance
2 Standard NEN 8100 [48] was adopted in this experiment. In the NEN 8100 standard, the wind velocity

ed
3 discomfort threshold is set at 5 m/s for all levels of pedestrian activities and the local wind comfort is assessed
4 based on the possibility of the wind velocity exceeding this value [49]. Table 4 tabulates the criteria for wind
5 comfort according to the NEN 8100 standard. P(UTHR) is the probability values for wind velocity to exceed the
6 threshold wind velocity UTHR of 5 m/s at the pool section. The standard defined five quality grades of wind
7 comfort (A to E). The comfort criteria were divided into a number of activities, where each was associated with

iew
8 a quality grade. This helped to assess the comfort of the resort guests especially at the pedestrian level to
9 perform these activities in the pool area.
10
11 Table 4 Criteria for wind comfort according to the NEN 8100 standard [48].
P (UTHR > Grade Activity
5 m/s Traversing Strolling Sitting Comfort criteria
(%)

ev
< 2.5 A Good Good Good Sitting long
2.05 - 5.0 B Good Good Moderate Sitting short
5.0 - 10 C Good Moderate Poor Walking leisurely
10 - 20 D Moderate Poor Poor Walking fast
> 20 E Poor Poor Poor Uncomfortable

r
12
13 The simulation results listed in Table 5 show that the total probability of wind velocity at 1H, 2H, 3H, and
er
14 4H to exceed 5 m/s was more than 20%, which translated to grade E for wind comfort. The wind comfort
15 assessment revealed that the wind climate was unsuitable and uncomfortable for all ordinary activities, namely
16 traversing, strolling and sitting, in the vicinity of the pool. Thus, it is essential to improve the wind comfort of
17 the guests, especially in outdoor spaces such as the recreational pool area, as these are the spaces wherein the
pe
18 guests mainly lounge.
19
20 Table 5 Exceedance probability of wind velocity at the pedestrian level and wind comfort grades at the pool
21 section.
Location / Wind Probability Exceedence (%) Wind Comfort Activity Description
Velocity 5 m/s 6 m/s 7 m/s 8 m/s 9 m/s Total Grade Sitting Strolling Traversing
ot

1H 16.07 25.00 31.55 4.17 0 76.8 E poor poor poor


2H 7.74 21.43 36.90 33.93 0 100 E poor poor poor
3H 0 11.31 31.55 45.24 11.90 100 E poor poor poor
22 4H 0 7.14 25.60 42.86 24.40 100 E poor poor poor
tn

23
24 Figure 11(a) shows the vertical pressure profiles at four locations close to the pool section, which were
25 analyzed to ascertain the formation of the vortex in the windward direction in the vicinity of the building. As
26 the prevailing wind directly flowed towards the building, the highest vertical pressure recorded in the
27 streamwise direction was at the pedestrian height, especially at the distance of 1H, where the recorded pressure
rin

28 was 17.93 N/m2. According to Irwin [50], the high pressure recorded at the ground level at 1H was due to the
29 formation of a downward vortex, which was due to the strong air force of the wind flow. This result was
30 consistent with that from Irwin [50], which stated that the strong fluid rotation around a vortex core decreased
31 the turbulence. As a result, the high and rapidly varying velocities of a vortex could be maintained for an
32 extended period and the static pressure could be reduced over time. As the distance from the building increased,
ep

33 the static pressure at the pedestrian height gradually decreased to 9.00 N/m2 at 2H, 4.40 N/m2 at 3H, and 2.62
34 N/m2 at 4H.
35 The graph of the measured static pressure against the wind velocity at the pool section is presented in
36 Figure 11(b). The results show that the static pressure decreased as the velocity increased. A factor that can
37 contribute to this was the wind velocity that squeezed the air through the narrow spaces between the building
Pr

38 walls, which is known as the venturi effect [42]. The wind velocity at the pool section can be as high as 8.5
39 m/s, which is higher than the threshold value of pedestrian wind comfort.

13

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 Figure 11(c) shows the profile of vertical turbulent kinetic energy k against the building height. At the
2 pedestrian level, the k value was large in the streamwise direction on the downwind side of the building,

ed
3 especially at the distance of 1H, which was 0.31 m2/s2. The k value decreased as the distance from the building
4 increased but remained almost constant at 2H, 3H, and 4H, which was 0.18, 0.18, and 0.20 m2/s2 respectively.
5 The large k value at 1H was caused by the prevailing wind that flowed from the beach towards the building,
6 which generated a strong downwind flow towards the ground. Vortices were produced in a turbulent
7 environment and broken down by transferring kinetic energy to successive smaller eddies until viscous effects

iew
8 converted the energy into heat. Moreover, the turbulence was suppressed by the tight rotation of fluid around
9 the vortex core. As a result, the high and rapidly varying velocities of the vortex were maintained for a longer
10 time, thus increasing the value of k [50]. This large value of k could affect the pedestrian wind comfort in the
11 region near the pool section.
12

(a)

r ev
er
13
pe
(b)
ot
tn

14
rin

(c)
ep
Pr

15
16 Figure 11 The vertical (a) pressure profiles, (b) pressure profiles against the wind velocity, and (c) profiles of
17 the vertical turbulent kinetic energy k at four locations near the pool section.
14

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 In summary, the venturi effect caused by the prevailing wind flowing towards the building led to the wind
2 being strong and highly turbulent, especially in the pool section. The shape of the resort building resembles the

ed
3 shape of a courtyard, which is an unroofed area that is partially enclosed by walls or buildings in a large house
4 or a housing complex. Its purpose is to provide good ventilation for common areas such as a swimming pool
5 [52]. However, based on the results presented in this study, the courtyard shape may not be suitable for
6 providing good ventilation as the turbulence generated by the wind is high, causing wind discomfort to visitors
7 and tourists [53], especially in the pool area.

iew
8

9 3.4 Wind Simulation Results for the U-Section of the Resort


10
11 The simulation results for the small open section (U-section) are shown in Figure 12(a). It can be seen that
12 the vertical wind velocity increased as the distance from the building increased. At 1H, the wind velocity

ev
13 recorded at pedestrian height was 1.86 m/s and increased to 2.87, 3.96, and 4.28 at 2H, 3H, and 4H respectively,
14 which was lower than the threshold wind velocity of 5 m/s. This translates to better wind comfort at the
15 pedestrian height for the locals and tourists that walk or stroll in the U-section of the resort [41].
16 The pressure distribution in the U-section is shown in Figure 12(b). The figure shows that in the streamwise
17 direction and at the pedestrian height, the pressure at 1H and 2H was the lowest, which was 8.94 and 0.4 N/m2

r
18 respectively. In comparison to the pool section, the pressure measured at 1H and 2H was 17.93 and 9.00 N/m2
19 respectively.
20
er
(a)
pe
ot

Pedestrian height
tn

21

(b)
rin
ep

Pedestrian height

22
Pr

23 Figure 12 The (a) vertical wind velocity profile in the U-section, and (b) vertical pressure profiles in the U-
24 section.

15

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 The low pressure recorded in the U-section was due to the obstruction caused by the left and right wings
2 of the resort building, as shown in Figure 13(a). The wind flowed directly through the opening of the U-section,

ed
3 which caused changes in the vertical pressure near the inner faces of the building and the formation of a vortex
4 inside the U-section. The narrow opening of the U-section resulted in the lowest pressure at the ground level.
5 This result is in line with the finding by Sanyal [54], which revealed that the negative or lowest pressure is due
6 to the formation of a vortex and slight positive or higher pressure at the bottom is due to the reattachment wind
7 pressure being higher than the suction pressure. As the distance from the building and the opening size of the

iew
8 U-section increased, a rapid increase in pressure was observed from 0.04 N/m2 at 2H to 5.98 and 8.24 N/m2 at
9 3H and 4H respectively, as shown in Figure 12(b).
10 Figure 13(b) shows graph of building height against the vertical k of the U-section. At the pedestrian level,
11 the k values were greater in the streamwise direction on the downwind side of the building, especially at the
12 distance of 2H, which was 1.27 m2/s2. This was due to the wind flowing through the narrow channel of the U-
13 section, causing the wind velocity and k to increase. The increased value of k was the result of the net work
14 done to push the wind through the channel. At 2H, a pressure difference was observed as the channel narrowed,

ev
15 especially in the U-section. This pressure difference resulted in a net force acting on the wind flow, which was
16 in accordance with the work-energy theorem [55]. In summary, the net work done increased the value of k and
17 the high velocity of the wind caused the pressure to drop, which complied with Bernoulli’s theorem.
18

r
(a)
er
pe

19
ot

(b)
tn
rin

Pedestrian height

20
ep

21 Figure 13 The (a) top view of the U-section, and (b) profile of the vertical turbulent kinetic energy k in the U-
22 section.
23
24 Based on the analysis results presented in Section 3.3, it is concluded that strong wind is generated in the
25 pool section due to the section being partially enclosed by the building walls [54]. Action should be taken to
Pr

26 further investigate and improve the natural ventilation and the pedestrian wind comfort, especially in the leisure
27 and recreational areas near the resort building. One way is by planting windbreak trees at strategic locations
28 around the building.
16

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 3.5 Wind simulation results for the pool section in the presence of windbreak trees
2

ed
3 The CE trees have the potential to act as windbreak trees that could minimize the strong wind effects
4 resulting from the tropical storms that hit Langkawi Island every year. In this section, the numerical
5 investigation results of the windbreak effect of single-row CE trees are discussed.
6 Figure 14 illustrates the top view of the building resort and potential windbreak location. It can be seen in
7 Figure 14(a) that the CE trees are located 30 m from the beachline and 74 m from the building and swimming

iew
8 pool, which is similar distance inside the domain in CFD simulation (Figure 14b). The distance between the
9 canopy of the CE trees was 4 m, as shown in Figure 14(c). This distance was chosen based on the guideline for
10 the CE tree positioning within crop rows [56].
11
(a)

r ev
er
pe
12

(b)
ot

wind
tn

74 m
rin

13

(c)
ep

4m 4m 4m 4m
14
Pr

15 Figure 14 The view of the (a) building resort and potential windbreak location, (b) domain of the resort building
16 and CE tree models, and (c) canopy distance of the CE tree models in the simulation.
17
17

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 Figure 15(a) shows the streamwise velocity profiles obtained from the simulation in the pool section. It
2 can be seen that the vertical wind velocity increased as the distance from the building increased. At the

ed
3 pedestrian height, the wind velocity increased gradually from 1.56 m/s to 2.41, 3.07, and 3.78 m/s as the
4 distance increased stepwise from 1H to 4H.
5 Figure 15(b) presents the vertical static pressure profile in the pool section in the presence of a single row
6 of CE trees. The negative pressure was measured above the ground and over a large region throughout the
7 vertical domain. The lowest pressure was on the leeward side of the CE windbreak, which was due to the

iew
8 airflow being forced over the windbreak and down through the canopy [59],[60].
9 The effect of CE trees on the value of k is shown in Figure 15(c). The k value at 1H, 2H, 3H, and 4H was
10 0.49, 0.75, 0.92, and 1.04 m2/s2 respectively, which was larger than that presented in Figure 11(c). In the current
11 study, the CE tree model had a branchless trunk in the region between the bottom of the tree canopy and the
12 ground. This created a gap for the oncoming wind, which caused the wind velocity to increase as it flowed
13 through this region. Therefore, a strong turbulence zone was formed behind the tree trunks [61], which
14 increased the turbulence level and value of k.

ev
15 To investigate the windbreak effectiveness of single-row CE trees, the normalized vertical wind velocity
16 profiles with and without the CE trees in the pool section were compared (see Figure 15e). As the distance
17 from the building increased, the value of U/Uref at the pedestrian height without a windbreak was higher than
18 that with a windbreak. Without the CE trees, the U/Uref value was 0.36, 0.64, 0.72, and 0.76 at 1H, 2H, 3H, and

r
19 4H respectively. The presence of CE trees caused the U/Uref value to decrease to 0.18, 0.29, 0.36, and 0.45 at
20 1H, 2H, 3H, and 4H respectively. The percentage reduction of U/Uref value at the pedestrian height at each
21 location is as follows: 1H = 8.69%, 2H = 17.72%, 3H = 18.04%, and 4H = 15.45%.
22
23
er
It can be observed that CE trees have the potential to act as a tree windbreak and provide the resort building
with protection against storms or strong wind. These results are within the range of the previous research [57]
24 between the relationship of optical porosity of CE (in this study = 49.9%) to the minimum relative wind velocity
25 (wind velocity reduced less than 20%) at the leeward side of the windbreak with a significant width dimension
pe
26 of the 95% confidence limits for individual windbreak predicted values. Hence, planting a single row of CE
27 trees in front of the resort could greatly decrease the wind velocity within the building areas, especially in the
28 pool section, minimize the effect of wind damage from the tropical storm, and improve wind comfort [58] at
29 the pedestrian height.
30 To determine the effect of the windbreak on the pressure and wind velocity in the pool section, the graph
31 of static pressure against the wind velocity in the presence of a single row of CE trees was analyzed (see Figure
ot

32 15d). As the wind velocity increased, the static pressure decreased due to the venturi effect. This was in
33 accordance with the principle of continuity and mechanical energy conservation. The single row of CE trees
34 managed to reduce the wind velocity in the pool section to below the threshold value of pedestrian wind
tn

35 comfort, which is less than 5 m/s [41]. It shows that planting a single row of CE trees is an effective way to
36 reduce wind velocity. The findings in this study provide a scientific basis for future policymaking and the
37 proposed criteria can also help city planners to improve the wind comfort at the pedestrian level, especially in
38 the coastline areas.
39
rin

40
41
42
43
44
ep

45
46
47
48
49
Pr

50
51
52
18

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
(a) (b)

ed
iew
1
(c)
(d)

r ev
2 er
(e)
pe
ot

3
4 Figure 15 The vertical profile in the presence of a single row CE tree models in the pool section of the (a) wind
tn

5 velocity against building height, (b) static pressure against building height, (c) vertical k against building height,
6 (d) vertical static pressure against wind velocity, and (e) normalized vertical wind velocity (U/Uref) with and
7 without a single row of CE trees in the pool section.
8
rin

9 Conclusion
10 This paper presents the results of the wind flow simulation around the building of the Holiday Villa Beach
11 Resort in Cenang, Langkawi. The computational fluid dynamics (CFD) simulation was performed to
12 investigate the flow characteristics behind tree windbreaks. The significant outcomes of the current study are
ep

13 as follows:
14
15 • The distance between the two SG trees created a big opening, allowing the wind from the sea to flow
16 directly towards the resort building, especially towards the pool area. As the distance from the building
17 increased stepwise from 1H to 4H, the vertical wind velocity increased gradually. The increment of wind
Pr

18 velocity was due to the downwash process, which resulted in turbulent wind conditions and increased wind
19 velocity at the pedestrian level [44].

19

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 • Based on the criteria stated in the NEN 8100 standard, the pedestrian wind comfort in the pool section was
2 assessed [48]. The results show that the total exceedance probability of wind velocity at 1H, 2H, 3H and 4H

ed
3 was 76.8%, 100%, 100% and 100% respectively. Thus, the wind comfort in the pool section was given Grade
4 E, which signified that the wind climate in the pool section was unsuitable for all typical activities such as
5 traversing, strolling, and sitting.
6 • Due to the huge number of CE trees available at the resort, the potential of these trees to act as windbreak
7 trees and provide protection against strong wind was investigated. A simulation was performed, in which five

iew
8 CE trees were placed in a row at a distance of 74 m from the building. The gap between adjacent tree canopies
9 was set at 4 m. The results revealed that the wind velocity increased gradually as the distance from the building
10 increased.
11 • To investigate the effectiveness of the windbreak in the pool section, the value of U/Uref at the pedestrian
12 level was analyzed. It was observed that the presence of a single row of CE trees decreased the U/Uref value. It
13 can be deduced that CE trees have the potential to act as windbreaks and protect the resort building against
14 storms and strong wind. Moreover, the findings in this study provide a scientific basis for future policy-making

ev
15 and the proposed criteria can also help city planners to improve the wind comfort at the pedestrian level,
16 especially in the coastal areas.
17

r
18 Acknowledgement
19
20 The authors would like to thank Universiti Teknologi Malaysia for supporting this research through the
er
21 Transdisciplinary Research Grant (Vote 06G99). The authors would also like to express their gratitude to the
22 Malaysian Meteorological Department, Holiday Villa Beach Resort & Spa Langkawi, Wind Tunnel Laboratory
23 Malaysia - Japan International Institute of Technology (MJIIT) and Langkawi Development Authority (LADA)
24 for supporting this research.
25
pe
26 References
27
28 [1] R. A. Chan, G.K., Selvadurai, S. and Aziz, “The culture of heritage conservation in Malaysia: a study
29 of eco-tourism in Langkawi,” e-BANGI, vol. 17, no. 1, pp. 165–174, 2020.
30 [2] K. L. G. Chan, Jaringan Hubungan Sosial Pemangkin Pemuliharaan Seiring Pembangunan: Kajian
31 Kes Langkawi Geopark. 2014.
32 [3] S. I. Omar, A. G. Othman, and B. Mohamed, “The tourism life cycle: An overview of Langkawi
ot

33 Island, Malaysia,” Int. J. Cult. Tour. Hosp. Res., vol. 8, no. 3, pp. 272–289, 2014, doi:
34 10.1108/IJCTHR-09-2013-0069.
35 [4] Langkawi Municipal Council, “Rancangan Tempatan Daerah Langkawi 2001-2015 (Langkawi
36 District Local Plan 2001-2015),” Kedah: Langkawi Municipal Council, 2005.
tn

37 [5] M. Raman, “Review, scrap the Langkawi reclamation project,” New Straits Time, 2021.
38 [6] I. K. Sharina Abdul Halim, “The Geopark As A Potential Tool For Alleviating Community
39 Marginality: A case study of Langkawi Geopark, Malaysia,” Int. J. Res. into Isl. Cult., p. 5, 2011.
40 [7] Langkawi District Office, “Langkawi Structure Plan 1985-2015: Draft Report. Langkawi District
41 Council, Department of Town and Country Planning.,” 1990.
rin

42 [8] A. Marzuki, “Impacts of tourism development,” Anatolia, vol. 20, no. 2, pp. 450–455, 2009, doi:
43 10.1080/13032917.2009.10518921.
44 [9] B. Clough, Continuing the Journey Amongst Mangroves., vol. 1. 2013.
45 [10] Bernama, “Met Malaysia issues bad weather warning in several states,” 2018.
46 [11] MET, “Laporan Tahunan Annual Report MET Malaysia,” 2019. [Online]. Available:
47 https://www.met.gov.my/content/pdf/penerbitan/laporantahunan/laporantahunan2019.pdf.
ep

48 [12] Bernama, “Storm damages 17 houses, 15 cars in Langkawi,” 2020.


49 [13] X. H. Brandle, J.R., Hodges, L. and Zhou, Windbreaks in North American agricultural systems, In
50 New vis. Springer, Dordrecht, 2004.
51 [14] X. H. Zhou, J. R. Brandle, C. W. Mize, and E. S. Takle, “Three-dimensional aerodynamic structure
52 of a tree shelterbelt: Definition, characterization and working models,” Agrofor. Syst., vol. 63, no. 2,
Pr

53 pp. 133–147, 2005, doi: 10.1007/s10457-004-3147-5.


54 [15] USDA NRCS, “Technical notes,” Wind. Shelter. Landscaping Technol., p. 49, 2005, doi:
55 10.1080/00690805.1989.10438448.
56 [16] J. P. Bitog et al., “A wind tunnel study on aerodynamic porosity and windbreak drag,” Forest Sci.
20

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 Technol., vol. 7, no. 1, pp. 8–16, 2011, doi: 10.1080/21580103.2011.559939.
2 [17] D. R. Heisler, G.M. and Dewalle, “Effects of windbreak structure on wind flow,” Agric. Ecosyst.
3 Environ., vol. 22, pp. 41–69, 1988.

ed
4 [18] Z. Guo et al., “Optimal design for vegetative windbreaks using 3D numerical simulations,” Agric.
5 For. Meteorol., vol. 298–299, no. 19, 2021, doi: 10.1016/j.agrformet.2020.108290.
6 [19] Z. Jian, L. Bo, and W. Mingyue, “Study on windbreak performance of tree canopy by numerical
7 simulation method,” J. Comput. Multiph. Flows, vol. 10, no. 4, pp. 259–265, 2018, doi:
8 10.1177/1757482X18791901.

iew
9 [20] Wong PP., “Coastal tourism development in Southeast Asia: Relevance and lessons for coastal zone
10 management,” Ocean Coast. Manag., vol. 38, no. 2, pp. 89–109, 1998.
11 [21] ICIDM, “Map of Langkawi,” International Conference on Interactive Digital Media (ICIDM 2012),
12 2012. https://seminar.utmspace.edu.my/icidm2012/images/LangkawiMap.gif.
13 [22] A. Mehranian, H. and Marzuki, “Beach Users’ Perceptions Toward Beach Quality and Crowding: A
14 Case of Cenang Beach, Langkawi Island, Malaysia,” Sea Lev. Rise Coast. Infrastruct., p. 181, 2018.
15 [23] Tripadvisor, “Photo: Holiday Villa Langkawi,” TripAdvisor LLC, 2022.
16 https://www.tripadvisor.com/LocationPhotoDirectLink-g2053325-d1024498-i334069271-

ev
17 Holiday_Villa_Beach_Resort_Spa_Langkawi-Pantai_Tengah_Langkawi_Langkawi.html.
18 [24] Holiday Villa Beach Resort & Spa Langkawi, “Holiday Villa Langkawi,” 2020.
19 https://www.holidayvillahotels.com/holiday-villa-beach-resort-spa-langkawi/.
20 [25] M. Foale, “The coconut odyssey: the bounteous possibilities of the tree of life,” Aust. Cent. Int. Agric.
21 Res. Canberra, pp. 1–134, 2003.

r
22 [26] J. Parnell, P. Chantaranothai, “Myrtaceae. In: Santisuk T & Larsen K (eds.) Flora of Thailand,” For.
23 Herb. R. For. Dep. Bangkok, Thail., vol. 7, no. 4, pp. 778–914, 2002.
24 [27] E. Corner, “Wayside trees of Malaya,” Malayan Nat. Soc. Kuala Lumpur, vol. 2, p. 861, 1988.
25
26
[28]
er
P. Ashton, “Myrtaceae. In: Soepadmo E, Saw LG, Chung RCK & Kiew R (eds.) Tree Flora of Sabah
and Sarawak,” For. Res. Inst. Malaysia, vol. 7, pp. 87–330, 2011.
27 [29] W. A. Whistler and C. R. Elevitch, “Syzygium malaccense (Malay apple),” Species Profiles Pacific
28 Isl. Agrofor., vol. 2.1, no. April, pp. 1–13, 2006, [Online]. Available:
pe
29 https://agroforestry.org/images/pdfs/Syzygium-Malayapple.pdf.
30 [30] C. G. Yakhot, V., Orszag, S. A., Thangam, S., Gatski, T. B., & Speziale, “Development of turbulence
31 models for shear flows by a double expansion technique,” Phys. Fluids A., 1992.
32 [31] J. Ries, K. and Eichhorn, “Simulation of effects of vegetation on the dispersion of pollutants in street
33 canyons,” Meteorol. Zeitschrift, vol. 10, no. 4, pp. 229–234, 2001.
34 [32] B. Balczó, M., Gromke, C. and Ruck, “Numerical modeling of flow and pollutant dispersion in street
35 canyons with tree planting,” Meteorol. Zeitschrift, vol. 18, no. 2, p. 197, 2009.
ot

36 [33] G. Kang, J. J. Kim, D. J. Kim, W. Choi, and S. J. Park, “Development of a computational fluid
37 dynamics model with tree drag parameterizations: Application to pedestrian wind comfort in an urban
38 area,” Build. Environ., vol. 124, no. November, pp. 209–218, 2017, doi:
39 10.1016/j.buildenv.2017.08.008.
tn

40 [34] B. Balczó, M., Gromke, C., & Ruck, “Numerical modeling of flow and pollutant dispersion in street
41 canyons with tree planting,” Meteorol. Zeitschrift, 2009, doi: 10.1127/0941-2948/2009/0361.
42 [35] J. Ries, K., & Eichhorn, “Simulation of effects of vegetation on the dispersion of pollutants in street
43 canyons,” Meteorol. Zeitschrift, 2001, doi: https://doi.org/10.1127/ 0941-2948/2001/0010-0229.
44 [36] S. V. Patankar, Numerical heat transfer and fluid flow. CRC press, 2018.
45 [37] M. Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, “AIJ guidelines for
rin

46 practical applications of CFD to pedestrian wind environment around buildings,” J. Wind Eng. Ind.
47 Aerodyn., 2008.
48 [38] S. A. Hsu, “Determining the Power-Law Wind-Profile Exponent under Near-Neutral Stability
49 Conditions at Sea,” Am. Meteorol. Soc., p. 9, 1994.
50 [39] Y. Miyazaki, T., Tominaga, “Wind tunnel experiment on flow field around a building model with a
ep

51 scale ratio of 4:4:1 placed within the surface boundary layer,” Proc. Annu. Meet. Hokuriku Chapter,
52 Archit. Inst. Japan, vol. 201–204, 2003, [Online]. Available:
53 https://www.aij.or.jp/paper/detail.html?productId=313933.
54 [40] Malaysia Meteorology Department, “Weather Data Report Malaysia Meteorology Department.”
55 [41] J. P. A. J. Blocken, B., Stathopoulos, T. and Van Beeck, “Pedestrian-level wind conditions around
56 buildings: Review of wind-tunnel and CFD techniques and their accuracy for wind comfort
Pr

57 assessment,” Build. Environ., vol. 100, pp. 50–81, 2016.


58 [42] C. Fairclough, “Exploring the Venturi Effect,” COMSOL Blog, 2015.
59 https://www.comsol.com/blogs/exploring-the-venturi-effect/.

21

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052
1 [43] F. Vikneshvaran, V., Zaki, S.A., Rahmat, N.A., Ali, M.S.M. and Yakub, “Evaluation of Atmospheric
2 Boundary Layer in Open-Loop Boundary Layer Wind Tunnel Experiment,” J. Adv. Res. Fluid Mech.
3 Therm. Sci., vol. 72, no. 2, pp. 79–92, 2020.

ed
4 [44] R. Delgado, A., Gertig, C., Blesa, E., Loza, A., Hidalgo, C. and Ron, “Evaluation of the variability of
5 wind speed at different heights and its impact on the receiver efficiency of central receiver systems,”
6 AIP Conf. Proc., vol. 1734, no. 1, p. 30011, 2016.
7 [45] P. Sisterson, D.L. and Frenzen, “Nocturnal boundary-layer wind maxima and the problem of wind
8 power assessment,” Environ. Sci. Technol., vol. 12, no. 2, pp. 218–221, 1978.

iew
9 [46] L. Cochran, “Design features to change and/or ameliorate pedestrian wind conditions,” Struct. 2004
10 Build. Past, Secur. Futur., pp. 1–8, 2004.
11 [47] I. Serteser, N. and Karadag, “Design for improving pedestrian wind comfort: a case study on a
12 courtyard around a tall building,” Archit. Sci. Rev., vol. 61, no. 6, pp. 492–499, 2018.
13 [48] Netherlands Normalisation Institute, “Wind Comfort and Wind Danger in the Built Environment
14 (NEN 8100),” Dutch, 2006.
15 [49] Willemsen, E. and Wisse, J.A., “Design for wind comfort in The Netherlands: Procedures, criteria
16 and open research issues,” J. Wind Eng. Ind. Aerodyn., vol. 95, no. 9–11, pp. 1541–1550, 2007.

ev
17 [50] Irwin, Peter A., “Vortices and tall buildings: A recipe for resonance,” Phys. Today, vol. 63, no. 9, pp.
18 68–69, 2010.
19 [51] N. Mao, J. and Gao, “The airborne transmission of infection between flats in high-rise residential
20 buildings: a review.,” Build. Environ., vol. 94, pp. 516–531, 2015.
21 [52] P. Sanyal and S. Kumar, “Effects of courtyard and opening on a rectangular plan shaped tall building

r
22 under wind load,” Int. J. Adv. Struct. Eng., vol. 10, no. 2, pp. 169–188, 2018, doi: 10.1007/s40091-
23 018-0190-4.
24 [53] J. W. Emmerich, S.J., Dols, W.S. and Axley, “Natural ventilation review and plan for design and
25
26 [54]
er
analysis tools,” US Dep. Commer. Technol. Adm. Natl. Inst. Stand. Technol., pp. 1–56, 2001.
P. Sanyal and S. K. Dalui, “Effects of courtyard and opening on a rectangular plan shaped tall
27 building under wind load,” Int. J. Adv. Struct. Eng., vol. 10, no. 2, pp. 169–188, 2018, doi:
28 10.1007/s40091-018-0190-4.
pe
29 [55] OpenStax, “Bernoulli’s Equation,” University Physics Volume 1, 2020.
30 https://opentextbc.ca/universityphysicsv1openstax/chapter/14-6-bernoullis-equation/.
31 [56] Kementerian Air Tanah dan Sumber Asli, “Buku Panduan Penanaman Rhu,” 2022. [Online].
32 Available:
33 https://bakau.forestry.gov.my/index.php/component/flippingbook/book/18?page=1&Itemid=711.
34 [57] X. H. Zhou, J. R. Brandle, C. W. Mize, and E. S. Takle, “Three-dimensional aerodynamic structure
35 of a tree shelterbelt : Definition , characterization and working models,” pp. 133–147, 2004.
ot

36 [58] M. Danesh Miah, M., Abubokor Siddik, M. and Yong Shin, “Socio-economic and environmental
37 impacts of casuarina shelterbelt in the Chittagong coast of Bangladesh,” Forest Sci. Technol., vol. 9,
38 no. 3, pp. 156–163, 2013.
39 [59] E. G. Patton, “Large-eddy simulation of turbulent flow above and within a plant canopy,” University
tn

40 of California, 1997.
41 [60] E. S. Wang, H. and Takle, “A numerical simulation of boundary-layer flows near shelterbelts,”
42 Boundary-Layer Meterol, vol. 75, pp. 141–173, 1995.
43 [61] J. Lee, E. Lee, and S. Lee, “Shelter effect of a fir tree with different porosities,” J. Mech. Sci.
44 Technol., vol. 28, no. 2, pp. 565–572, 2014, doi: 10.1007/s12206-013-1123-6.
45
rin
ep
Pr

22

This preprint research paper has not been peer reviewed. Electronic copy available at: https://ssrn.com/abstract=4227052

You might also like