b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Available online at www.sciencedirect.com
journal homepage: www.elsevier.com/locate/issn/15375110
Research Paper
Numerical simulation study of a tree windbreak Jessie P. Bitog a,b, In-Bok Lee a,*, Hyun-Seob Hwang a, Myeong-Ho Shin c, Se-Woon Hong a, Il-Hwan Seo a, Kyeong-Seok Kwon a, Ehab Mostafa a,d, Zhenzhen Pang a,e a
Department of Rural Systems Engineering and Research Institute for Agriculture and Life Sciences, College of Agriculture and Life Sciences, Seoul National University, 599, Gwanakno, Gwanakgu, 151-921 Seoul, Republic of Korea b Department of Agricultural Engineering, Nueva Vizcaya State University, 3700 Bayombong, Nueva Vizcaya, Philippines c Saemangeum Project Office, Korea Rural Community & Agriculture Corporation, 605-1 Shinpoong-Dong Gimje, Shi Jeon-buk, Republic of Korea d Department of Agricultural Engineering, Faculty of Agriculture, Cairo University, 12613 Giza, Egypt e College of Horticulture and Gardening, School of Tropical Agriculture and Life Sciences, Hainan University, Hainan province, China
article info
In this study, computational fluid dynamics (CFD) was utilised to investigate the flow
Article history:
characteristics around tree windbreaks. The efficiency of windbreaks depends on many
Received 30 March 2011
factors which can be investigated in field experiments, though this is limited due to several
Received in revised form
reasons such as unstable weather conditions, few measuring points, etc. Fortunately, the
30 August 2011
investigation is possible via computer simulations. The simulation technique allows the
Accepted 12 October 2011
trees to be modelled as a porous media where the aerodynamic properties of the trees are
Published online 16 November 2011
utilised in the model. The trees employed are Black pine trees (Pinus thunbergii) with a drag coefficient value of 0.55. The simulation provides analysis of the effect of gaps between trees, rows of trees, and tree arrangements in reducing wind velocity. The simulations revealed that 0.5 m gap between trees was more effective in reducing wind velocity than 0.75 and 1.0 m. The percentage reduction in velocity at the middle of the tree section for 0.5, 0.75 and 1.0 m gap distance was found to be 71, 65 and 56%, respectively. Two-rows of alternating trees were also found to be more effective than one-row and two-rows of trees. The reduction at the middle of the tree region for one-row and two-rows of trees and two-rows arranged alternately was 71, 88 and 91%, respectively. Results revealed that the percentage reduction in wind velocity measured at distance 15H, where H is the tree height, for one-row, two-rows of trees and two-rows arranged alternately was approximately 20, 30 and 50%, respectively. ª 2011 IAgrE. Published by Elsevier Ltd. All rights reserved.
1.
Introduction
South Korea is mostly surrounded by bodies of water with approximately 2413 km of coast line along three seas. To the west is the Yellow Sea, to the south is the East Sea, and to the east is Ulleung-do and Liancourt Rocks (Dokdo) in the East Sea. The country has geographical land mass of approximately
100,032 km2 with limited land resources. This prompted the country to implement several land reclamation projects especially near the coastlines. By 2006, 38% or 1550 km2 of coastal wetlands had been reclaimed, including 400 km2 in the Saemangeum area (Korea Statistical Information Service, 2006). The wind velocities in the coastal areas and in the reclaimed lands are usually higher because of the sea wind. In
* Corresponding author. Tel.: þ82 2 880 4586; fax: þ82 2 873 2087. E-mail address:
[email protected] (I.-B. Lee). 1537-5110/$ e see front matter ª 2011 IAgrE. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.biosystemseng.2011.10.006
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Nomenclature C13 C23 C33
Ca CD Gk I k l m p P
constant (1.42) constant (1.68) tanh(n1/n2), n1 and n2 are components of the flow velocities parallel and perpendicular, respectively to the gravitational vector empirical constant specified in the turbulence model (approximately 0.09) drag coefficient generation of turbulent kinetic energy due to the mean velocity gradients (kg m1 s2) turbulence intensity (dimensionless) turbulent kinetic energy (m2 s2) turbulence length scale (m) thickness or diameter of tree canopy (m) pressure (Pa) mean pressure (Pa)
these areas, the recorded average wind velocity at 5 m height can reach up to 7.0 m s1 especially during the dry months of the year from February to May (Hwang et al., 2006). This has caused generation and diffusion of dusts to nearby areas, such as agricultural and animal farms. In addition, the dusts from the reclaimed lands contain significant quantities of sodium chloride (NaCl) which is very detrimental to plants, animals and humans (Bitog et al., 2011). These problems can be reduced by minimising the wind velocity especially in the dust source areas, where it is the main catalyst of dust generation and diffusion. Constructing artificial barriers or planting natural windbreaks such as trees to control the wind velocity are the best options. However, for long term protection, tree windbreaks are strongly recommended (Zhou, Brandle, Mize, & Takle, 2004). Natural windbreaks, especially trees, direct winds over or around protected areas such as agricultural farms, and livestock farms. They are very effective in reducing wind speed in the protected area. Windbreaks operate by creating pressure at the windward side of the trees as wind blows against them, with the direction of large air flows shifted in direction over the top or around the ends of the windbreaks (Bitog et al., 2011). Especially in coastal areas and in reclaimed lands, windbreaks can control the wind velocity to a level usually lower than the threshold velocity required for the generation and diffusion of dust. The amount of wind speed reduction depends on several factors such as the tree height, density, width, shape, arrangement, porosity, etc. (Bitog et al., 2011). However, several studies have already shown that, among the factors, tree porosity has the most influence on windbreak efficiency (Bitog et al., 2009; Cornelis & Gabriels, 2005; Heisler & Dewalle, 1988). Determining the actual tree porosity is complex, considering the irregular size and shape of the trees as well as the varied distribution of the gaps. However, this can be well represented by the drag coefficient (CD). As mentioned by Jacobs (1985), the resistance to wind flow or the drag coefficient of the windbreak can provide information on its effectiveness and efficiency in reducing high velocity winds. Therefore, knowing the dimensionless drag coefficient value of the tree windbreak is very important index to
R Si t Vavg v vi v1i v1j xi
41
additional term source term. time (s) average wind velocity (m s1) velocity (m s1) respective mean velocity components (m s1) Reynolds stresses (m2 s2) component of length (m)
Greek symbols turbulent dissipation rate (m2 s3) inverse effective Prandtl number for k ak inverse effective Prandtl number for 3 a3 m the fluid viscosity (N s m2) a permeability constant r fluid density (kg m3) effective viscosity (N s m2) meff
3
evaluate the wind protection effect of the tree (Bitog et al., 2011). In principle, the flow characteristics around windbreaks are being disrupted because of a net loss of momentum as the tree exerts a drag force on the incoming winds (Raine & Stevenson, 1977). Performing computer simulation has been in the forefront for studying the effectiveness and efficiency of natural and artificial windbreaks including the relevant flow mechanisms around barriers. This is evidenced by the numerous published papers. A number of small-scale windbreaks studies were also conducted in wind tunnels (Dong et al., 2008; Dong, Qian, Luo, & Wang, 2006; Gromke & Ruck, 2008; Guan, Zhang, & Zhu, 2003) since the results could be applied to full scale model or actual field conditions with reliable scale-up procedures. The use of correct aerodynamic and engineering equations to resolve the scaling up technique, especially the dimensional difference of the models, must also be carefully managed. The continuing development of both wind tunnel and numerical simulation techniques has paved the way for more accurate and reliable laboratory investigations and computer simulation studies (Li, Wang, & Bell, 2007). The analysis of the flow characteristics of these simulation studies was based on the objectives for the windbreaks, with the main emphasis on the effect to the leeward. However, most of the simulations were still limited to 2-dimensional models because of insufficient computer memory to process the computations. The simulation study presented by Raine and Stevenson (1977) measured and analysed the wind velocity and energy spectra to leeward of a modelled fence. Similar studies by Castro and Garo (1998) and Judd, Raupach, and Finnigan (1996) were conducted on a porous barrier where the mean velocity and turbulence stress downstream of the barrier were investigated. A numerical model has been developed by Wilson (1985, 1987) and Wilson and Yee (2003) to accurately simulate the wind flow characteristics around a single fence and multi-array windbreaks. These simulation studies proved their reliability by showing similar results when compared to wind tunnel or field experimental results. Computer simulation, particularly computational fluid dynamics (CFD), has now become very popular for studying
42
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
wind flow characteristics around artificial and natural windbreaks. The most recent studies have been conducted by Bitog et al. (2009), Gromke and Ruck (2008), Rosenfeld, Maron, and Bitan (2010), and Santiago, Martin, Cuerva, Bezdenejnykh, and Sanz-Andres (2007). These studies have exploited the power of CFD in the investigation and analysis of wind flows over a terrain or an area protected by windbreaks. Because of rapid development of more powerful and high memory personal computers, simulations of 3-dimensional models can now be easily achieved. A study by Rosenfeld et al. (2010) established the significance and extent of 3-dimensional flow patterns across tree windbreak comprising of individual cypress trees. Their study was validated by comparing their simulation results with experimental data and showed good agreement. In this study, a 3-dimensional model was designed adopting the general shape and dimension of Black pine trees (Pinus thunbergii). The study was conducted with an attempt to simulate the tree windbreaks in the natural environment. The drag value of the experimental tree, which had been determined earlier (Bitog et al., 2011), was utilised as input in setting up the boundary conditions of the tree, which was designed as a porous media. With the simulation technique, numerical analysis on the effect of the windbreaks in reducing wind velocity, especially to leeward, can be thoroughly investigated. The effect in reducing wind velocity of gap distance between trees, number of rows of trees and their arrangement were determined, and the horizontal extent of the effect of the tree windbreak at varied wind velocities was also measured. The results obtained will be used in the design of an effective windbreak system for use in the reclaimed lands and in the coastal areas of Korea, and can also utilised in future experimental and simulation studies of wind flows as affected by artificial or natural windbreaks, considering complex topographies.
2.
Materials and methods
2.1.
Wind tunnel experiment
Wind tunnel experiments have already been conducted to determine the aerodynamic properties of the trees. Table 1 presents the test parameters in the experiment where five live trees were used and the pressure and velocity was measured in five locations. The velocity was also varied from 2 to 8 m s1. In wind tunnels, live trees instead of scaled models
Table 1 e Parameters tested in the wind tunnel. Parameters Tree Measuring location Velocity
Particulars
Total
T1, T2, T3, T4, T5 (5) 0, 5, 10, 15, 20 cm before and after the tree (5) 2, 4, 6, 8 m s1 (4)
5 5 4 ¼ 100
T1, T2, T3, T4, T5: the subscript 1e5 represents the tree number that was tested in the wind tunnel.
were used making the approach more reliable since the use of a scaling factor was eliminated. Under the controlled atmospheric conditions where the wind velocity can be easily managed, experiments were conducted based on the designed cases. The pressure and velocity at windward and leeward were recorded from several points after several minutes when the velocity and pressure was observed to be stable. The measured data were used to determine the aerodynamic properties of the experimental trees. The schematic diagram of the wind tunnel experiment detailing the measurement points located before and after the experimental trees is presented in Fig. 1. The real set-up of the pressure and velocity sensors without tree and with tree as well as the actual measurement inside the control room of the wind tunnel is are also presented in Fig. 2. Specific details on the experimental trees as well as the wind tunnel experimental procedures and the aerodynamic equations used in the computations can be found in our earlier paper (Bitog et al., 2011). The resistance or the pressure loss equation for the tree windbreaks, Δp (Pa) shown in Eq. (1), can be simplified from Lin, Barrington, Choinier, and Prasher (2007) where it is affected basically by the viscous loss term (Darcy’s law) and the inertial loss term (Fluent manual, 2006; Lin et al., 2007; Santiago et al., 2007). However, since the computed Reynolds number was very high even for the lowest wind velocity of 2 m s1, the flow inside the wind tunnel is expected to be turbulent, thus the viscous loss term can be neglected. The pressure loss and velocity measured from wind tunnel experiments were used to compute CD value in Eq. (1) (Fluent manual, 2006; Lee et al., 2006; Molina-Aiz, Valera, Alvarez, & Maduen˜o, 2006) and was utilised as input value in the simulation study. m Dp ¼ v þ CD rv2 m a
(1)
The measurement of the extent of velocity reduction cannot be obtained in wind tunnels because of the limited height and length of the test section. This can be determined through simulation analysis because the model can be designed to a desired length and quantitative data along the lateral distance can be determined. The contour profiles of velocity derived in the simulation study also allow more in-depth analysis and understanding of the effect of the tree windbreak. In addition, the simulation study also significantly saves time, labour and cost. In our previous wind tunnel work, the windbreak drag (CD) of the experimental tree was determined from the measured wind pressure and velocity data from wind tunnel experiment. Since the pressure is a function of the velocity, the measured pressure is expected to increase at increasing velocity. The relationship between the measured pressure difference before and after the tree and the term (rv2)/2 gives the drag coefficient (CD) of the experimental trees, which was found to be 0.55. The average drag coefficient value was obtained when the velocity is within the range from 2 to 8 m s1.
2.2.
Computational fluid dynamics (CFD)
CFD is one of the advanced and powerful simulation tools to investigate fluid flows. CFD basically employs the
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
43
Fig. 1 e Schematic diagram of the wind tunnel set-up showing the measurement points at windward and leeward (Not drawn to scale) (Reproduced with kind permission from Bitog et al., 2011. Copyright 2011 Forest Science and Technology).
mathematical formulations of the conservation laws that govern fluid flows. It starts by building a computational model which represents a system or device divided into cells or meshes using computer aided design (CAD) tools. The fluid flow physics and chemistry will be applied in the designed model which then allows the prediction of flow dynamics and determination of other related phenomena in the system. The CFD technique numerically solves the continuity, momentum and energy equations within each cell of the discretised computational domain. In this simulation study, the Reynolds-averaged process considers the instantaneous fluid velocity to be the sum of a mean and a fluctuating component of turbulence. The continuity equation and RANS equations (Fluent manual, 2006; Rosenfeld et al., 2010; Santiago et al., 2007) presented in Eqs. (2) and (3) respectively, were solved in each of the simulated cases. Air was assumed to be incompressible under a steady turbulent flow while heat and mass transfer processes were neglected.
vvi ¼0 vxi
mj
vmi 1 vP v2 m v 0 0 ¼ þ v 2i m m þ Si vxj r vxi vxj i j vxj
(2)
(3)
The turbulence closure model, RNG ke3 presented in Eqs. (4) and (5) (Bitog et al., 2009; Fluent manual, 2006; Santiago et al., 2007), was utilised in the simulation. As shown in Eq. (4), buoyancy was not taken into account in the study. The computed turbulent kinetic energy (k) and the turbulent dissipation rate (3 ) determined from our previous study (Bitog et al., 2009) were also utilised which were linked through the User Defined Functions (UDF) of the CFD program. RNG ke3 model has been shown to be the best for investigation of wind flows around barriers (Bourdin & Wilson, 2008; Packwood, 2000; Santiago et al., 2007). RNG k-3 model provides an analytically-derived differential formula for effective viscosity that accounts for low-Reynolds number
Fig. 2 e A view of the set-up of pressure and velocity sensors without tree and with tree and the actual measurement inside the control room of the wind tunnel (Reproduced with kind permission from Bitog et al., 2011. Copyright 2011 Forest Science and Technology).
44
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Table 2 e Data and variables used in the simulation. Pre-processing
Size (L W H )
Trees at 0.50 gap distance: 130 x 10.50 x 27.5 m Trees at 0.75 gap distance: 130 11.25 27.5 m Trees at 1.00 gap distance: 130 12.00 27.5 m Tree volume: tetrahedron Windward and Leeward volume: Hexahedron One row of trees: 371,118 Two-rows of trees: 422,436 Two-rows of trees alternating: 550,225
Mesh type Total mesh number
Main module
RNG ke3 turbulence model 0.2 Second order upwind Steady state
Turbulence Surface roughness Discretisation Condition
effects. It also provides an analytical formula for turbulent Prandtl numbers. The k and 3 values at different heights can be calculated using Eqs. (6) and (7) (Bitog et al., 2009; Charuchittipan & Wilson, 2009; Fluent manual, 2006). The turbulence length scale was determined following the theoretical equations discussed by Counihan (1975). Dk v vk ak meff þ G k r3 ¼ r Dt vxi vxi
(4)
2 D3 v v3 3 3 r ¼ a3 meff þ C13 ðGk þ C33 Þ C23 r R Dt vxi vxi k k
(5)
k¼
3
2 3 Vavg I 2 3=
¼ Ca 4
2.2.1.
(6)
3
k =2 l
(7)
CFD simulation
CFD simulation involves the solution of a set of partial differential equations describing the transport of momentum, energy and turbulence quantities. Partial differential equations are converted into a set of simultaneous algebraic equations. The domain under analysis is divided into a set of non-overlapping adjoining rectilinear cells (finite volume grid). Algebraic equations are set-up for each grid cell and the whole set of equations are solved using a numerical method. The basic calculation procedure of the Fluent software in this type of simulation study has been previously discussed (Bitog et al., 2009; Lee and Lim, 2001; Packwood, 2000; Qiu, Lee, Shimizu, Gao, & Ding, 2004; Rosenfeld, 2010; Santiago et al., 2007; Wilson & Yee, 2003). The vertical wind velocity profile which was earlier investigated through wind tunnel tests (Bitog et al., 2009) was utilised at the inlet boundary of the computational domain. The profile was linked to the CFD main module through the UDF. The profile was comparable to the logarithmic wind profile discussed by Calvert (2004). According to Calvert, in the intermediate case, the velocity profile is found to be logarithmic. A theoretical expression for the velocity is u ¼ (v*/k) ln(z/d). Here, v* is the friction velocity computed by O(s/r), where s is the shear stress at the v ¼ 0 plane, r is the density, k is the von Ka´rma´n constant, z is the vertical height and d is the
height when the velocity is zero. Friction velocity is a measure of the surface stress reflecting the effect of surface roughness and wind velocity. More details regarding wind velocity profiles near surfaces can be found in Calvert (2004). Several 3D geometries of the tree models were developed using the Gambit software. The tree was designed to be 5.5 m in height with 0.5 m trunk height and maximum canopy width of approximately 3 m. To realise the application of the inertial resistance in the simulation, the tree canopy volume was zoned as fluid where the porous zone was activated. The value of the windbreak drag which was determined from the wind tunnel experiment for a wind velocity of up to 8 m s1 was then used for the porous trees during the calculation. The effect of gap distance between trees, rows of trees and their arrangement were thoroughly analysed in the simulation. The percentage decrease of wind velocity was also investigated at varied wind velocities of 3.5, 5.0 and 6.5 m s1. The total length of the computational domain was 140 m with windward length designed at 5H and leeward length, 20H. The H referred to the tree height which was 5.5 m. The total height of the domain was also designed at 5H or 27.5 m. The 5H has been recommended as the minimum height of the computational domain in simulation studies by Blocken, Carmeliet, and Stathopoulos (2007); Blocken, Stathopoulos, and Carmeliet (2007); Franke (2006); Franke et al. (2004); Hall (1997) and Tominaga et al. (2008). The width of the computational domain was taken as 10.50, 11.25 and 12.00 m depending on the gap distance between trees of 0.50, 0.75 and 1.00 m, respectively. The total mesh sizes of the whole domain ranged from 370,000 to 550,000 cells. The minimum and maximum cell size was 0.2 and 1.0 m, respectively. The quality of the meshes was determined via the equiangle skewness value. In the windward and leeward region of the computational domain, the structured mesh was applied with a skewness value from 0.25 to 0.50 which is considered good, while in the tree region with unstructured mesh, the highest value was approximately 0.79 which is generally acceptable (Bakker, 2002; Fluent manual, 2006). For a timedependent CFD model, the initial airflow field around the model continuously changes according to several factors such as the topography and the airflow turbulence (Hong et al., 2008). Therefore, to maintain the vertical wind velocity profile used in the inlet boundary the airflow flow was assumed to be laminar up to 3H to windward, and then became turbulent as it approaches the trees. At a distance 3H
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
3.
Results and discussion
3.1.
CFD simulation
45
3.1.1. Validation of the simulation study based on the determined CD value
Fig. 3 e Comparison of CD values determined from simulations at varied mesh sizes with CD value obtained from wind tunnel (-: Experiment, ,: Sim., 0.1 mesh size; >: Sim., 0.2 mesh size; D: Sim., 0.3 mesh size; B: Sim., 0.4 mesh size).
before the trees, a change of the profile is expected because of the turbulent flow as well as the effect exerted by the trees. The effect of the vegetation cover was also considered through the roughness parameter in the wall function, to account for the effect of the plants at the ground surface which can be observed in most reclaimed lands in Korea. The surface roughness input of 0.2 was used, assuming a vegetation height of 0.5 m (Oke, 1987). Both sides of the computational domain were designed to be symmetrical while the bottom section was designed as surface wall. The top sections as well as the opposite side of the inlet boundary of the computational domain were designed as pressure outlet. A summary of the data and variables used in the simulation is presented in Table 2.
Fig. 4 e Simulated wind velocity measured at 1.0 m height at varied gap distances, number of rows and arrangement of trees (>: One row, 0.50 m gap distance; ,: One row, 0.75 m gap distance; D: One row, 1.00 m gap distance; B: Two-rows, 0.50 m gap distance; D: Two-rows, 0.5 m gap distance).
The validity of the simulation study was initially checked. Based on the pressure and velocity data obtained from the simulation study, the CD value was determined using Eq. (1). The mesh size in the tree region was varied until the CD value becomes comparable with the CD value determined from wind tunnel experiment. As shown in the Fig. 3, a mesh size in the tree region of 0.3 and 0.4 m resulted in higher CD values. A mesh size of 0.2 and 0.1 m was found to display similar CD values to wind tunnel experiment. However, using a mesh size of 0.1 m in the tree regions would greatly increase the number of cells and would require more time for the calculation. Therefore, a mesh size of 0.2 was chosen to be used in the subsequent simulations.
3.1.2. Effect of gap distances between trees, rows of trees and arrangement Field experimental studies conducted by Hwang et al. (2006) in Saemanguem reclaimed land in Korea found that dust concentration is very high up a height of 1.0 m. This shows the presence of a significant amount of dust near the ground surface. In this regard, with the goal of preventing dust generation and diffusion, the analysis and discussion of the effectiveness of the tree windbreaks were based on 1.0 m height where the dust concentration is very high. A gap distance between trees of 0.5 m was more effective in reducing wind velocity compared to 0.75 and 1.0 m gap distance. This can be observed in Fig. 4 where the average velocity measured at 1.0 m height is presented. Lower
Fig. 5 e Visualisation of the velocity contour (m sL1) of the tree windbreak at varied rows of trees and their arrangement (Top view at 1.0 m height; maximum velocity is 8 m sL1).
46
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Table 3 e Percentage decrease of wind velocity measured at 1.0 m height. Number of rows/arrangement
Gap distance (m)
Distance from the trees (H: tree height) e3H
1H
0
1H
3H
5H
7H
9H
11H
13H
15H
One-row
0.50 0.75 1.00
2.4 1.6 0.5
24.8 13.0 10.1
70.5 65.0 56.0
52.5 49.0 45.5
47.1 41.7 36.1
44.8 35.8 29.2
41.7 32.0 24.2
37.7 28.2 19.1
32.8 24.1 15.3
27.4 18.7 10.1
20.5 11.6 4.0
Two-rows
0.50
2.5
23.2
87.5
80.5
72.5
65.1
58.2
51.7
45.2
38.1
31.2
Two-rows in alternate
0.50
2.6
25.3
92.5
89.5
80.5
75.1
69.2
63.7
58.2
54.1
48.2
velocities had been observed to the windward and leeward when the gap distance between the trees was 0.5 m. The average velocity along the middle of the tree zone at varied gap distance of 0.50, 0.75 and 1.00 m decreased from 3.5 m s1 to approximately 1.0, 1.2 and 1.5 m s1, respectively. The percentage reduction of wind velocity is discussed in the subsequent section. Simulation results have shown that gaps in the windbreak became funnels where the wind flows were concentrated. Where there are gaps, it is expected that the effectiveness of the windbreak is diminished. This is clearly obvious in Fig. 5 where the top view of the velocity contour at 1.0 m height is presented. The velocity along the gap distance for one-row and two-rows of trees was greater. For the tworows of alternating trees, the gap distance between trees created from the first row was disturbed by the second row of trees, thus making the velocity at the leeward fairly uniform. Therefore, to implement the most effective tree windbreak, gap distance between trees should be kept to minimum. Tworows of alternating trees have also been shown to be more effective in reducing wind velocity compared to one-row and two-rows of trees especially to leeward.
3.1.3.
Quantitative analysis on the effect of tree windbreak
The fractional reduction factor (DU/Uo) ¼ kr/(1 þ 2kr)0.8 in mean wind velocity that was proposed by Wilson, Swaters, and Ustina (1990) can be calculated separately according to the windbreak drag and the velocity determined near the leeward side. In the equation, DU is the change in velocity (m s1), Uo is the unperturbed velocity (m s1), and kr is the pressure loss coefficient which is actually the windbreak drag. Using the windbreak drag of 0.55, the fractional reduction was approximately 0.20 while from the simulated results obtained for one-row of trees at 0.5 m gap distance with average velocity near leeward of 2.23 m s1, the fractional reduction was found to 0.36. The value of the fractional reduction factor in the simulation deviates from the equation proposed by Wilson et al. (1990). However, the proposed equation only applies to thin windbreaks which cannot be exactly applied to trees. Further, the effect of the ground surface was accounted for in the simulation model where vegetation in the ground was assumed to exist. The percentage decrease of wind velocity due to the tree windbreaks, as affected by gap distance between trees and rows of trees and arrangement, is presented in Table 3. The wind velocity is shown to decrease gradually on the windward side, especially for 0.5 gap distance between trees where the percentage decreases are very high. With wider gap distance, the measured velocities at the leeward are higher, which
resulted in a smaller percentage decrease. Therefore, in the windward zone with the same tree height, gap distance plays the vital role in minimising the wind velocity. In the tree region, the velocity decreased up to 60e90% depending on the condition. The simulation results clearly showed the effectiveness of tree windbreaks in reducing wind velocity especially near the leeward zones. The lateral extent of windbreak effects upwind and downwind is usually assumed to be proportional to windbreak height (Heisler & Dewalle, 1988). In this study, with the designed tree height of 5.5 m, the lateral extent of windbreak effect for one-row of trees could reach up to 15H, with a decrease in velocity of 21, 12 and 4% for 0.50, 0.75 and 1.00 m gap distance, respectively. This shows that windbreak height cannot be exclusively considered if the extent of wind reduction is one of the primary concerns. Gap distance between trees is also an important factor to consider. The results strongly supported our earlier discussion that gap distance between trees can decrease the effectiveness of the tree windbreak. For two-rows of alternating trees, the percentage decrease of wind velocity at distance 15H measured at 1.0 m height can reach 50% (Table 3). The results are within the range presented by Heisler and Dewalle (1988)
Fig. 6 e The velocity contour profile (m sL1) at varied wind velocity of 3.5, 5.0 and 6.5 m sL1 measured at 1.0 m height for two-rows of trees arranged alternately with 0.5 m gap distance between trees.
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Fig. 7 e Measured wind velocity at 1.0 m height for tworows of trees arranged alternately with 0.5 m gap distance between trees (Windward wind velocity, D: 3.5, >: 5.0, ,: 6.5 m sL1).
in which the reductions of wind speed can go as far as 50H to the leeward and can go down by 20% or more at a distance of 25H from the windbreaks.
3.1.4. Simulation of wind flow for different input wind velocities The velocity contour profile at wind velocities of 3.5, 5.0 and 6.5 m s1 measured at 1.0 m height and the measured velocity at a given distance are presented in Figs. 6 and 7 for two-rows of trees arranged alternately. No recirculation of wind flow was observed in the leeward region. This was expected since the trees are very porous and airflow penetrating the trees is dominant. This is in agreement with our earlier study on windbreak fences (Bitog et al., 2009) where solid fences caused re-circulating flows while porous fences (with 0.2 porosity or higher) caused no re-circulating flows. This shows that high porous windbreaks allow more wind to flow though. The results of the simulation also showed that the higher the windward wind velocity, the shorter was the protected horizontal distance until the velocity had almost recovered. For 5.0 and 6.5 m s1 wind velocity, the velocity could recover from its original velocity at 15H of lateral distance. Therefore, in the design a windbreak system, the maximum wind velocity in the area should be strongly considered.
4.
Conclusion
CFD has proved to be an effective tool to simulate flow around tree windbreaks. Exploiting the power of CFD could allow quantitative visualisation and velocity reduction investigation. However, for the accuracy of the simulation study, significant values required in the simulation process such as the drag coefficient of the tree must be precisely determined. In this study, computer simulation was performed for the Black pine tree (Pinus thunbergii) windbreak, one of the most typical tree windbreaks that are very appropriate in the coastal areas and in most of the reclaimed lands in South
47
Korea. The drag coefficient of the tree, which was earlier obtained from wind tunnel experiment, was employed in the simulation studies for the tree windbreaks to predict the effect of gap distance between trees, rows of trees and arrangement in decreasing wind velocity. The horizontal extent of its effect was also quantitatively determined. Results of the simulation studies have shown that the gap distance is a very important factor for the effectiveness of the tree windbreak especially when the effect in the lateral direction is one of the main goals of the windbreak system. The velocity reduction of the tree windbreak with a gap distance of 0.5 m could reach up to 20% at a streamwise distance of 15H. For 0.75 and 1.00 m gap distance, the reductions were approximately 11 and 4% respectively. Two-rows of trees arranged alternately were found to be the most effective in reducing the wind velocity compared to one-row and tworows of trees. The reduction of wind velocity determined along the middle of the tree zone for one-row is approximately 71%, two-rows, 88%, and two-rows alternately arranged, 92%. Considering the streamwise extent of the effect of the windbreak at 15H, the reduction of wind velocity for two-row of trees arranged alternately could reach up to 50%. For one and two-row of trees, the quantitative effect had lowered to approximately 20 and 30%, respectively. Findings of this study should be very useful in planning and setting up natural windbreak system for a long term control of the generation and diffusion of dust especially in the reclaimed lands and the coastal areas of Korea.
Acknowledgements The authors are grateful to the Saemangeum Project Office, Korea Rural Community & Agriculture Corporation for the financial assistance extended; and to the National Institute of Agricultural Engineering of Korea for allowing the authors to use the wind tunnel. The anonymous reviewers who made critical comments and recommendations are also highly appreciated.
references
Bakker, A. (2002). Applied computational fluid dynamics, Lecture 7: Meshing. Fluent Inc. Bitog, J. P., Lee, I. B., Shin, M. H., Hong, S. W., Hwang, H. S., SeoMostafa, E., & Pang, Z. (2011). A wind tunnel study on aerodynamic porosity and windbreak drag. Forest Science and Technology, 7(1), 8e16. Bitog, J. P., Lee, I. B., Shin, M. H., Hong, S. W., Hwang, H. S., Seo, I. H., Yoo, J. I., Kwon, K. S., Kim, Y. H., & Han, J. W. (2009). Numerical simulation of an array of fences in Saemangeum reclaimed land. Atmospheric Environment, 43, 4612e4621. Blocken, B., Carmeliet, J., & Stathopoulos, T. (2007). CFD evaluation of wind speed conditions in passages between parallel buildings - effect of wall-function roughness modifications for the atmospheric boundary layer flow. Journal of Wind Engineering and Industrial Aerodynamics, 95, 941e962. Blocken, B., Stathopoulos, T., & Carmeliet, J. (2007). CFD simulation of the atmospheric boundary layer: wall function problems. Atmospheric Environment, 41, 238e22.
48
b i o s y s t e m s e n g i n e e r i n g 1 1 1 ( 2 0 1 2 ) 4 0 e4 8
Bourdin, P., & Wilson, J. D. (2008). Windbreak aerodynamics: is computational fluid dynamics reliable? Boundary-Layer Meteorology, 126, 181e208. Calvert, J. B. (2004). Wind: an exploration of the wind. http:// mysite.du.edu/wetuttle/weather/wind.htm. Castro, I. P., & Garo, L. (1998). Flow through porous fences in thick boundary layers. Part II-Laboratory experiments. University of Surrey. ME-FD/98.91, Technical Report. Charuchittipan, C., & Wilson, J. D. (2009). Turbulent kinetic energy dissipation in the surface layer. Boundary-Layer Meteorology, 132, 193e204. Counihan, J. (1975). Adiabatic atmospheric boundary layers: a review and analysis of data from the period 1880e1972. Atmospheric Environment, 9, 871e905. Cornelis, W. M., & Gabriels, D. (2005). Optimal windbreak design for wind-erosion control. Journal of Arid Environments, 61, 315e332. Dong, Z., Mu, Q., Luo, W., Qinan, G., Lu, P., & Wang, H. (2008). An analysis of drag force and moment for upright porous wind fences. Journal of Geophysical Research, 113, D04103. doi:10.1029/ 2007JD009138. Dong, Z., Qian, G., Luo, W., & Wang, H. (2006). Threshold velocity for wind erosion: the effect of porous fences. Environmental Geology, 51, 471e475. Fluent manual. (2006). New Hampshire. USA: Fluent Co. Franke, J. (2006). Recommendations of the COST action C14 on the use of CFD in predicting pedestrian wind environment. In: Proceedings of the fourth international symposium on wind engineering, Yokohama, Japan. Franke, J., Hirsch, C., Jensen, A. G., Kru¨s, H. W., Schatzmann, M., Westbury, P. S., Miles, S. D., Wisse, J. A., & Wright, N. G. (2004). Recommendations on the use of CFD in wind engineering. In J. P. A. J. van Beeck (Ed.), COST Action C14, impact of wind and storm on city life built environment. Gromke, C., & Ruck, B. (2008). Aerodynamic modelling of trees for small-scale wind tunnel studies. Forestry Advances Forestry, . doi:10.1093/forestry/cpn027. Guan, D., Zhang, Y., & Zhu, T. (2003). A wind-tunnel study of windbreak drag. Agricultural Ecosystems and Environment, 118, 75e84. Hall, R. C. (1997). Evaluation of modeling uncertainty e CFD modeling of near field atmospheric dispersion. Project EMU final report to the European Commission. Surrey: WS Atkins Consultants Ltd. Heisler, G. M., & Dewalle, D. R. (1988). Effects of windbreak structure on wind flow. Agricultural Ecosystems and Environment, 22e23, 41e69. Hong, S. W., Lee, I. B., Hwang, H. S., Seo, I. H., Bitog, J. P., Yoo, J. I., Kim, K. S., Lee, S. H., Kim, K. W., & Yoon, N. K. (2008). Numerical simulation of ventilation efficiencies of naturally ventilated multi-span greenhouses in Korea. Transactions of the ASABE, 51(4), 1417e1432. Hwang, H. S., Lee, I. B., Chang, P. W., Hong, S. W., Seo, I. H., & Lee, S. Y. (2006). Monitoring of dust emission at a reclaimed land and its 3-D aerodynamic modeling. St. Joseph: Mich., ASABE. ASABE paper No. 062055. Jacobs, A. (1985). The normal force coefficient on a thin closed fence. Boundary-Layer Meteorology, 32, 329e335. Judd, M. J., Raupach, M. R., & Finnigan, J. J. (1996). A wind tunnel study of turbulent flow around single and multiple windbreaks, part I: velocity fields. Boundary-Layer Meteorology, 80, 127e165.
Korea Statistical Information Service. (2006). Lee, I. B., Yun, N. K., Boulard, T., Roy, J. C., Lee, S. H., Kim, G. W., Lee, S. K., & Kwon, S. H. (2006). Development of an aerodynamic simulation for studying microclimate of plant canopy in greenhouse e (1) study on aerodynamic resistance of tomato canopy through wind tunnel experiment. Journal of Bio-Environmental Control, 15(4), 289e295. Lee, S. J., & Lim, H. C. (2001). A numerical study on flow around a triangular prism located behind a porous fence. Fluid Dynamics Research, 28, 209e221. Li, W., Wang, F., & Bell, S. (2007). Simulating the sheltering effects of windbreaks in urban outdoor open space. Journal of Wind Engineering and Industrial Aerodynamics, 95, 533e549. Lin, X. J., Barrington, S., Choinier, D., & Prasher, S. (2007). Simulation of the effect of windbreaks on odour dispersion. Biosystems and Engineering, 98, 347e363. Molina-Aiz, F. D., Valera, D. L., Alvarez, A. J., & Maduen˜o, A. (2006). A wind tunnel study of airflow through horticultural crops: determination of drag coefficient. Biosystems and Engineering, 93(4), 447e457. Oke, T. R. (1987). Boundary layer climates. London, UK: Methuen & Company, Ltd.. pp. 435. Packwood, A. R. (2000). Flow through porous fences in thick boundary layers: comparisons between laboratory and numerical experiments. Journal of Wind Engineering and Industrial Aerodynamics, 88, 75e90. Qiu, Y. G., Lee, I. B., Shimizu, H., Gao, Y., & Ding, G. (2004). Principles of sand dune fixation with straw checkerboard technology and its effect on the environment. Journal of Arid Environments, 56, 449e464. Raine, J. K., & Stevenson, D. C. (1977). Wind protection by model fences in a simulated atmospheric boundary layer. Journal of Industrial Aerodynamics, 2, 159e180. Rosenfeld, M., Maron, G., & Bitan, A. (2010). Numerical simulation of the flow across trees in a windbreak. Boundary-Layer Meteorology, 135, 89e107. Santiago, J. L., Martin, F., Cuerva, A., Bezdenejnykh, N., & SanzAndres, A. (2007). Experimental and numerical study of wind flow behind windbreaks. Atmospheric Environment, 41, 6406e6420. Tominaga, Y., Mochida, A., Yoshie, R., Kataoka, H., Nozu, T., Yoshikawa, M., & Shirasawa, T. (2008). AIJ guidelines for practical applications of CFD to pedestrian wind environment around buildings. Journal of Wind Engineering and Industrial Aerodynamics, 96, 1749e1761. Wilson, J. D. (1985). Numerical studies of flow through a windbreak. Journal of Wind Engineering and Industrial Aerodynamics, 21, 119e154. Wilson, J. D. (1987). On the choice of a windbreak porosity profile. Boundary-Layer Meteorology, 38, 37e49. Wilson, J. D., Swaters, G. E., & Ustina, F. (1990). A perturbation analysis of turbulent flow through a porous barrier. Quarterly Journal of Royal Meteorolgical Society, 116, 989e1004. Wilson, J. D., & Yee, E. (2003). Calculation of winds disturbed by an array of fences. Agricultural and Forest Meteorology, 115, 31e50. Zhou, X. H., Brandle, J. R., Mize, C. W., & Takle, E. S. (2004). Threedimensional aerodynamic structure of a tree shelterbelt: definition, characterization and working models. Agroforestry Systems, 63, 133e147.
