See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/337040413

Evaluation of wind-driven natural ventilation of single-span greenhouses

built on reclaimed coastal land

Article  in  Biosystems Engineering · April 2018

DOI: 10.1016/j.biosystemseng.2018.04.015

CITATIONS READS

9 53

3 authors:

Sang-Yeon Lee I.-B. Lee

Seoul National University Seoul National University
16 PUBLICATIONS   25 CITATIONS    171 PUBLICATIONS   1,573 CITATIONS   

SEE PROFILE SEE PROFILE

Rackwoo Kim
Seoul National University
24 PUBLICATIONS   66 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

aviquitous sensor network View project

Agriculture facility environment View project

All content following this page was uploaded by Sang-Yeon Lee on 06 November 2019.

The user has requested enhancement of the downloaded file.

 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: www.elsevier.com/locate/issn/15375110

Research Paper

Evaluation of wind-driven natural ventilation of

single-span greenhouses built on reclaimed coastal
land

Sang-yeon Lee a, In-bok Lee a,b,*, Rack-woo Kim a

a
Department of Rural Systems Engineering, Research Institute of Agriculture and Life Sciences, College of
Agriculture and Life Sciences, Seoul National University, 1, Gwanakno, Gwanakgu, Seoul, 08826, Republic of Korea
b
Research Institute of Green Eco Engineering, Institute of Green Bio Science and Technology, Seoul National
University, 1477, Pyeongchang-daero, Daehwa-myeon, Pyeongchang-gun, Gangwon-do, 25354, Republic of Korea

article info
Recently, the government of South Korea announced the construction plan for a large-scale
Article history: greenhouse complex on reclaimed coastal lands. Wind characteristics of the coastal re-
Received 19 October 2017 gions are quite different from those of inland. Natural ventilation of greenhouses is
Received in revised form strongly dependent on the exterior wind characteristics; therefore, the effects of topo-
11 April 2018 graphical and meteorological characteristics of the coastal regions on ventilation rate
Accepted 18 April 2018 should be carefully evaluated to ensure stable and uniform production of the crops. In this
study, wind-driven natural ventilation rates of single-span greenhouses were evaluated
using a computational fluid dynamics (CFD) technique examining greenhouse type and
Keywords: typical wind conditions found in reclaimed lands. The accuracy of the designed CFD model
Computational fluid dynamics was validated by comparing the result of the CFD simulation with the results of particle
Natural ventilation rate image velocimetry (PIV) measurements that were conducted in a wind tunnel. Further-
Particle image velocimetry more, grid independence test and turbulence model test were conducted for validation of
Reclaimed land the designed CFD model. From these validated CFD model, the overall ventilation rates
Single-span greenhouse were estimated using two methods, mass flow rate (MFR) and tracer gas decay (TGD)
methods according to greenhouse type, wind speed, wind direction, and configuration of
the ventilator. Furthermore, local ventilation rates of the greenhouse were estimated using
the TGD method. The CFD computed ventilation rates of greenhouses were evaluated by
comparing them with the ventilation rates required for controlling the temperature in
greenhouses. Finally, charts, which can be used to predict the natural ventilation rate of
the single-span greenhouse built on reclaimed land, were developed.
© 2018 IAgrE. Published by Elsevier Ltd. All rights reserved.

* Corresponding author. 1, Gwanakno, Gwanakgu, Seoul, 08826, Republic of Korea. Fax: þ82 2 873 2087.
E-mail address: iblee@snu.ac.kr (I.-b. Lee).
https://doi.org/10.1016/j.biosystemseng.2018.04.015
1537-5110/© 2018 IAgrE. Published by Elsevier Ltd. All rights reserved.
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 121

Nomenclature 1. Introduction

AER Air exchange rate Since the 1970s, greenhouse cultivation in South Korea has
Ac Cover area of greenhouse (m2) been developing owing to such advantages as environmental
Af Floor area of greenhouse (m2) control, intensive cultivation, stable production and the pos-
Avent Area of the vent sibility of producing high-quality agricultural products. Total
CFD Computational fluid dynamics greenhouse area in South Korea has increased from 700 ha in
C0 Concentrations of tracer gas at t0 time (ppm) 1970 to 52,526 ha in 2015. The production of protected culti-
Ct Concentrations of tracer gas at t time (ppm) vation, 2,558,310 tonnes, accounted for approximately 30% of
Cv Volumetric specific heat of air (kcal m3  C1) the total agricultural production in 2015 (Ministry of
Cm Empirical constant Agriculture, Food and Rural Affairs of Korea, 2016).
d Index of agreement Environmental control of greenhouse is necessary to pre-
ESDU Engineering science data unit vent problems that can be derived from the excessively high
f Evapotranspiration coefficient air temperatures found in summer. Many methods to properly
G Mass flow at vent (kg s1) control heat environment have been used, such as evapora-
k Von Karman constant (approximately 0.42) tive cooling systems with high-pressure fog cooling and fan
k0 Overall heat transfer coefficient and pad system, natural ventilation system, forced ventilation
(kcal m2  C1 min1) system, etc. Because of low initial and maintenance costs,
MAE Mean absolute error natural ventilation has been recognised as an effective and
MFR Mass flow rate economical method to control indoor air quality, the air
MSE Mean square error temperature in summer, and excess humidity. Most single-
PE Potential error span greenhouses in South Korea have operated using natu-
PIV Particle image velocimetry ral ventilation because of the cost of installation and opera-
R2 Coefficient of determination tion of forced ventilation systems (Kwon et al., 2011).
RMSE Root mean square error The use of reclaimed land for agriculture has been steadily
RNG Re-normalisation group developing in South Korea since the 1910s. The government
RSR Root mean square error-observations standard recently announced a development plan for establishing
deviation ratio greenhouse complexes of 5185 ha on reclaimed lands to in-
S Solar radiation (W m2) crease the price competitiveness of local vegetables in the
SIMPLE Semi-implicit method for pressure-linked market (Ministry of Food, Agricultural, Forestry and Fisheries of
equations Korea, 2010). While natural ventilation is greatly influenced by
SST Shear stress transport external wind environments, the characteristics of the wind
STDEVobs Standard deviation of measured values environment, such as wind speed, wind direction, turbulence
t0 Initial time (s) intensity, and gust frequency in reclaimed lands are different
t Ventilation time (s) from those found inland owing to the low topography and
TGD Tracer gas decay oceanic climate (Korea Meteorological Administration, 2014).
DT Difference of air temperature ( C) Studies are required to analyse the ventilation of greenhouses
u *
Shear velocity (m s1) built on reclaimed land by considering the wind profiles and
UðzÞ Wind speed at z height (m s1) other characteristics of the reclaimed land.
UðzR Þ Wind speed at the reference height (m s1) The studies on greenhouses built on reclaimed land to date
UDF User-defined functions have mainly focused on the structural safety of greenhouses.
V Volume of greenhouse (m3) Yun, Choi, Lee, Lee, and Yoon (2013) analysed the design wind
vx Wind speed of x-axis at vent (m s1) speed and snow cover depth according to the return period
obs
Yi i-th measured value using the weather data of 72 weather stations located near
Y obs Average of measured values reclaimed land in South Korea. Yu, Ku, Cho, Ryu, and Moon
Yisi i-th simulation value (2014) analysed the problems that can occur when a green-
Y si Average of simulation values house is built on reclaimed land using wooden piles. Kim,
yþ Dimensionless distance from the wall Kim, Ryu, Kwon, and Lee (2014) evaluated the wind pressure
z0 Roughness length (m) coefficient of an Even-span type greenhouse considering the
a Coefficient of roughness wind environment of reclaimed land according to roof slope.
a0 Correction rate of the heated area Choi, Yun, Yu, Lee, and Yoon (2015) measured the settlement
ε Turbulent energy dissipation (m2 s3) of a foundation using wooden pile at a 1e2W greenhouse built
rair Air density (kg m3) on Gyehwa reclaimed land. Lee et al. (2015) tried to establish a
t Solar transmissivity of the greenhouse cover technique for designing the foundation of a greenhouse built
u Specific dissipation rate (s1) inside a greenhouse. Kwon, Kim, Kim, Ha, and Lee (2016)
conducted wind tunnel test to evaluate the wind pressure
122 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
coefficients of the four types of single-span greenhouses built
on reclaimed coastal land according to wind direction and
design factors of the facility. However, studies on the venti-
lation of greenhouses considering the environment of
reclaimed land are lacking. Therefore, natural ventilation
rates should be analysed quantitatively considering the wind
environment of reclaimed land.
Various studies on natural ventilation of greenhouse have
been actively conducted because natural ventilation is an
effective and economical method. Natural ventilation of
greenhouses has mainly been estimated by field experiments,
wind tunnel tests, and numerical analysis. Many researchers
have conducted field experiments to analyse airflow patterns
and to estimate the ventilation rate of greenhouses because of
their advantages in measuring real field data (Boulard,
Meneses, Mermier, & Papadakis, 1996; Fatnassi, Boulard,
Bouirden, & Sappe, 2002; Fatnassi, Leyronas, Boulard, Bardin,
& Nicot, 2009; Flores-Velazquez, Montero, Baeza, & Lopez,
2014; He, Chen, Sun, Liu, & Huang, 2015; Ishii et al., 2014;
Kittas, Boulard, Mermier, & Papadakis, 1996; Kittas, Draoui, &
Boulard, 1995; Pe  rez Parra, Baeza, Montero, & Bailey, 2004;
Rigakis, Katsoulas, Teitel, Bartzanas, & Kittas, 2015; Sase, 1989;
Teitel, Ziskind, Liran, Dubovsky, & Letan, 2008; Villarreal-
Guerrero et al., 2013). However, evaluation of the ventilation
efficiency is restricted owing to various experimental limita-
tions in the field such as difficult experimental control of the
variables due to unstable and unpredictable weather condi-
tions, point-measurement due to limited number of in-
struments, limitations to meet the theoretical assumptions at
the initial stage of the experiment (especially for tracer gas
method), and time and labour consumption problems. In this
context, wind tunnel experiment and computational fluid
dynamics (CFD) is an efficient technique for adequately eval-
uating the natural ventilation rate because these methods can
artificially control environmental conditions. The wind tunnel
experiments have been widely used for aerodynamics ana-
lyses of greenhouses (Lee, Choi, Yun, Jeun, & Kim, 2003; Lee,
Sase, Okushima, Ikeguchi, Choi, & Yun, 2003; Lee et al., 2005;
Okushima, Sase, Maekawa, & Ikeguchi, 1998; Okushima,
Sase, Maekawa, Ikeguchi, & Bailey, 2000; Sase, Takakura, &
Nara, 1984) because in a wind tunnel test, it is easy to con-
trol the wind environment and it is high reliability. However, a
wind tunnel test is laborious and costly. To overcome the
limitations of field experiments and wind tunnel tests, studies
in agriculture have extensively been using CFD with various
applications, because CFD addresses complex physical phe-
nomena with few limiting factors and can take massive data
under various environmental conditions (Lee et al., 2003). For
this reason, CFD simulations have been actively used to esti-
mate air flow patterns and the ventilation efficiency of
greenhouses (Aich et al., 2016; Baeza et al., 2009; Bartzanas,
Boulard, & Kittas, 2004; Benni, Tassinari, Bonora, Barbaresi,
& Torreggiani, 2016; Boulard, Papadakis, Kittas, & Mermier,
1997; Campen & Bot, 2003; Flores-Velazquez et al., 2014; He
et al., 2015; Hong et al., 2008; Kacira, Sase, & Okushima, 2004;
Lee & Short, 2000, 2001; Mistriotis, Bot, Picuno, & Scarascia-
Mugnozza, 1997; Molina-Aiz, Fatnassi, Boulard, Roy, &
Valera, 2010; Okushima, Sase, & Nara, 1989; Teitel & Wenger,
2014). When using numerical analyses, such as CFD simula-
tions, validation of the accuracy of the numerical model
should be preceded. To validate the accuracy of CFD
computed results, many researchers compared the results of
CFD simulations with the results of laboratory experiments
(Baeza et al., 2009; Boulard, Haxaire, Lamrani, Roy, & Jaffrin,
1999), field experiments (Baeza et al., 2009; Bartzanas et al.,
2004; Bournet, Khaoua, & Boulard, 2007; Bournet, Khaoua,
Boulard, Migeon, & Chasse riaux, 2007; Fatnassi, Boulard,
Poncet, & Chave, 2006; Flores-Velazquez et al., 2014; He
et al., 2015; Kim et al., 2008; Majdoubi, Boulard, Fatnassi, &
Bouirden, 2009; Francisco Domingo Molina-Aiz, Valera, &

Alvarez, 2004; Ould Khaoua, Bournet, Migeon, Boulard, &
Chasse  riaux, 2006; Roy & Boulard, 2004; Teitel et al., 2008),
wind tunnel tests (Lee, Okushima, Ikeguchi, Sase, & Short,
2000; Lee et al., 2005; Teitel et al., 2008), etc. Among the
various validation approaches of CFD simulations, particle
image velocimetry (PIV) tests in the wind tunnel have been
used to validate CFD models (Lee & Short, 2000; Lee et al.,
2005). A PIV test in a wind tunnel is very useful for analysing
accurately internal airflow patterns and the wind velocity
distribution because it can qualitatively estimate the airflow
pattern and quantitatively estimate the air velocity in the
greenhouse.
The objective of this study was to evaluate the overall and
local ventilation performances of single-span commercial
greenhouses according to the facility type and wind charac-
teristics of the reclaimed coastal lands using the CFD tech-
nique. The wind-driven natural ventilation rate was
computed and evaluated considering the strong wind condi-
tions at the coastal regions. The internal plants were ignored
to accurately analyse the effect of greenhouse structural
characteristics and the wind environment at the reclaimed
land on ventilation efficiency. Crops with low heights, such as
lettuce and strawberry are usually cultivated in single-span
greenhouses. Because crop heights are relatively low
compared with the greenhouse height, it was assumed that
there is no significant aerodynamic effect on the internal
airflow caused by the crop. A wind tunnel test with a PIV
experiment was conducted to validate the accuracy of the
designed CFD model. To realise the experimental boundary
conditions of the reclaimed coastal lands, the ESDU (Engi-
neering Science Data Unit, E01008) code was adopted for the
design of variations in the roughness characteristics of the
windward regions and vertical wind and turbulence profiles.
Using the validated CFD model, the ventilation rate was
calculated according to the experimental conditions based on
the mass flow rate (MFR) and tracer gas decay (TGD) methods.
The ventilation rate of the target greenhouses computed by
CFD were evaluated using the ventilation requirement (Kozai,
Sase, & Nara, 1980; Rural Research Institute of Korea, 1997) to
suggest new design standards for greenhouses built on
reclaimed coastal land.
2. Materials and methods
To realise the wind characteristics of the reclaimed coastal
regions, ESDU was adopted. The ESDU computed wind and
turbulence profiles were then used for the boundary condi-
tions of the wind tunnel and CFD model. To validate the ac-
curacies of the CFD model and to support the CFD design
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 123

criterion, a PIV test was conducted with a small-scale green-

house model in the wind tunnel considering the wind profile
of the reclaimed land. From the validated CFD model, natural
ventilation rates of the single-span greenhouse were calcu-
lated based on the MFR and TGD methods. The experimental
procedure is shown in Fig. 1.

2.1. Target reclaimed land & weather data

Representative reclaimed coastal lands in South Korea (Hwa-

ong, Si-hwa, Seok-mun, E-won, Go-heung, Yeongsan-river,
and Saemangeum) were selected as candidate areas. Weather
data collected by the Korean Meteorological Administration
(KMA) from January 1, 1983 to December 31, 2012 at nine
weather stations (In-cheon, Woo-jung, Sskeo-san, Goon-san,
Bu-an, Mok-po, Jin-do, Hae-nam, and Go-heung) located
near the seven representative areas were used as the input
data of the ESDU model and the CFD models. The locations of
the seven candidate areas and nine weather observation sta-
tions are shown in Fig. 2. To determine the experimental wind
speed conditions, a frequency analysis of wind speed was
conducted. Following the analysis, the experimental wind
speeds were determined by considering low-speed and high-
speed winds. Fig. 2 e Location of candidate reclaimed coastal land and
weather stations.
2.2. Single-span greenhouse models for evaluating
ventilation rate
each greenhouse model was selected based on the Korean
Four types of single-span greenhouses, that are typically used design standards of horticultural facilities (Ministry of
in South Korea, were studied to evaluate the overall and local Agriculture, Food and Rural Affairs of Korea, 2014) and the
ventilation rate using the CFD technique. Even-span and Korean design standards of glass-type greenhouses (Ministry
three-quarter type greenhouses are typical glass-covered of Land, Transport and Maritime Affairs of Korea, 1997)
greenhouses, and peach and wide-width type greenhouses (Fig. 3). Additionally, the area of the side vent and volume of
are typical plastic film greenhouses. The experimental size of the greenhouse according to greenhouse type are shown in

Fig. 1 e Flow chart of the experimental procedure of this study.

124 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Fig. 3 e Schematic diagrams of greenhouse models for the ventilation analysis.

Table 1. The even-span type greenhouse model was chosen so that it was close to the external air temperature, a higher
for the CFD validation study because it is most commonly ventilation rate is required. Ventilation requirements consider
used in South Korea. specific factors, such as the difference between the design air
temperature inside the greenhouse and the outside air tem-
2.3. Experimental wind tunnel perature, as well as the optional crop coefficient (evapo-
transpiration coefficient), solar radiation, and transmissivity
The wind tunnel test was conducted at TESolution, Co. of covering material, with the exception of wind conditions.
(Anseong city, Gyeonggi-do, Korea). The full length of the
0 
tunnel was 36.8 m and the size of the test section was 8.0 1 a tSð1  f ÞAf 1
Air Exchange RateVR ¼   k0 Ac  (1)
(W)  2.5 (H)  23.0 (L) m; the wind tunnel was an Eiffel type Cv DT V
wind tunnel. The diameter of the turntable was 3.0 m, and it When solar radiation (S) was 800, 600, 400 and 200 W m2,
could revolve ±270 (Fig. 4). Three suction-type blowers with ventilation requirements according to greenhouse types and
AC motor types of 132 kW capacity were installed. The the difference of air temperature (DT,  C) were calculated and
diameter of the blowers was 1.5 m with a maximum of illustrated in graphs. Ventilation requirements (AER min1)
1200 rpm, and the range of wind velocity control was were calculated using the value of factors suggested in the
approximately 0.3e11.5 m s1. greenhouse standard in Korea (Rural Research Institute of
Korea, 1997). The volumetric specific heat of air (Cv ) was
2.4. Estimation methods of ventilation rate 0.3 kcal m3  C1, the correction rate of the heated area (a0)
was 1.0, the solar transmissivity of the greenhouse cover (t)
In the summer, excessive increases in internal air tempera- was 0.7 and the overall heat transfer coefficient (k0) was
tures have negative effects on crop growth because of high 0.08 kcal m2  C1. Additionally, internal crops were ignored
temperature injury. Ventilation requirements of greenhouses as this study was the basic research for evaluating the venti-
for controlling the air temperature in the summer were lation rates of single span greenhouses. Hence, the evapo-
calculated by Eq. (1). To decrease the internal air temperature transpiration coefficient (f), which was determined by sensible
and latent heat using the growth conditions of crops, was
assumed to be 0. The floor area (Af , m2), cover area (Ac , m2),
Table 1 e Specification of the greenhouse according to
greenhouse type (Unit: m). and volume of the greenhouse (V, m3) were determined by the
specifications of the greenhouse.
Parameters Single-span greenhouse types
To evaluate the natural ventilation rates of the green-
Even- Peach Three- Wide- houses considering the environmental conditions of
span quarter width
reclaimed lands, the MFR and TGD methods were used. The
Width 9.00 8.20 15.00 16.00 MFR method, which has been used widely, was represented by
Ridge height 4.95 3.90 6.30 5.50 the number of air changes on the assumption that air flowing
Eave height 2.70 1.60 4.30 2.10
from outside was used completely to replace the internal air.
Length 45.00 45.00 45.00 45.00
Width of side 1.35 1.20 2.00 1.70
By applying the MFR method, the overall ventilation rate of a
vent greenhouse can be easily calculated using the volume of the
Width of roof 1.68 e 1.20 0.60 greenhouse, the vent area, wind velocity at the vent, and air
vent
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
Fig. 4 e Schematic information of the experimental Eiffel-type wind tunnel (Unit: mm).
density. However, it is difficult to evaluate the local ventilation
rate in the greenhouse.
1 1
Air Exchange RateMFR ¼ G   (2)
rair V
G ¼ vx  rair  Avent
where Air Exchange RateMFR is the air exchange rate of the
MFR (AER s1), G is the mass flow at vent (kg s1), V is the
volume of the greenhouse (m3), rair is the air density (kg m3),
vx is the wind speed of x-axis at vent (m s1), and Avent is the
area of the vent (m2).
To overcome the limitations of the MFR method and
analyse the local ventilation rate in the greenhouse, virtual
tracer gas was used in the CFD simulation. The ventilation
rates based on the TGD method were calculated by quantita-
tively analysing the concentration decay curve of the tracer
gas. The concentration of the virtual tracer gas was observed
and logged at the multiple measuring points in the green-
house model as the ventilation occurred over time. This
method can consider the internal air flow pattern of the
greenhouse and analyse the local ventilation rate inside the
greenhouse. Furthermore, because the CFD simulation was
used for analysing local ventilation rates, the limitations of
field experiments such as limited multipoint measurements,
irregular environmental conditions, and uncertainty of gas
control can be supplemented. The ventilation rate based on
the TGD method was described as follows:
 
ln CC0t
Air Exchange RateTGD ¼
ðt  t0 Þ
where Air Exchange RateTGD is air exchange rate computed
using the TGD method (AER s1), C0 and Ct are concentrations
of tracer gas (ppm) at t0 and t, respectively, where t0 is initial
time and t is ventilation time (s).
To compute the mass flow at vent openings, the mass flow
at all cells of the vent area was computed by the CFD simu-
lation. Inflow and outflow were separated with distinction of
directionality of mass flow at each cell at the surface of the
vent area, and the total mass flow was calculated through
125
surface weighted average. To compute the ventilation rate
based on TGD method, the averaged concentrations of tracer
gas in the target sections over time were computed through
volume weighted average. The ventilation rate based on TGD
method was then calculated by analysing the decay curve of
the tracer gas concentration. The initial concentration of vir-
(3)
tual tracer gas was set at 2000 ppm and the CFD simulation
was performed until the concentration of the tracer gas
decreased to 1% of the initial concentration (Ha, 2015).
2.5. Estimation methods for accuracy of computational
fluid dynamics simulation model
In this study, validation for the accuracy of the CFD simulation
model was conducted by comparing it results with the PIV
measured results in wind tunnel. In addition to qualitative
comparisons of the CFD computed results and the PIV
measured results, quantitative comparisons were conducted
by three statistical methods; the coefficient of determination
(R2), RSR, and d, mean absolute error (MAE).
The value of R2 is in the range of 0e1 and the two results
tend to be similar when it is close to 1. However, R2 has a
drawback that even if the error of the two datasets tends to be
large when the trends of the data are similar, the calculated
value is close to 1. RSR is a statistical indicator that normalised
the RMSE by representing the degree of error. As RSR is closer
to 0, the error is lower and both results are similar. d, which is
the ratio of the MSE and PE, is a statistical indicator designed by
Willmott et al. (1985) to complement the drawbacks of R2. d has
a value from 0 to 1, and the closer it is to 1, the more similar the
two results can be seen to be. MAE, which is calculated by the
(4) average absolute difference between two continuous variables,
is usually used as a measure of difference in statistics. Equa-
tions for calculating R2, RSR, d and, MAE are shown below.
0    12
Pn
B i¼1 Yi
obs
 Yobs Yi  Y si
si C
B C
R2 ¼ B
Bsffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 ffiC
C (5)
@ Pn P n A
i¼1 Y i
obs
 Y obs i¼1 Y i  Y
si si
126 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

2 rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi 3
Pn  obs 2 u*3
6  Ysii 7 εðzÞ ¼ (12)
6 i¼1 Y i 7
RMSE kðz þ z0 Þ
RSR ¼ ¼6
6sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
 2 7
7 (6)
STDEVobs 4 Pn 5
i¼1 Y i
obs
 Yobs ε
uðzÞ ¼ (13)
Cm k
Pn  obs 2
i¼1 Yi  Yisi where UðzÞ is the wind speed at z height (m s1), u* is the shear
d¼1  2 (7)
Pn si
obs þ Y obs  Y obs
velocity (m s1), k is the Karman constant (dimensionless), z0
i¼1 Yi  Y i is the roughness length (m), Cm is the empirical constant
(dimensionless), ε is the turbulent energy dissipation (m2 s3),
Pn obs
i¼1Y  Y si and u is the specific dissipation rate (s1).
MAE ¼ i i
(8)
n
where Yiobs is the i-th measured value, Y obs is the average of the
2.6.2. Experimental design of particle image velocimetry test
measured values, Yisi is the i-th simulation value, and Y si is the
in wind tunnel
A PIV test is an optical method used to map air flow patterns
average of the simulation values. STDEVobs is the standard
using smoke. The PIV technique is very useful for the analysis
deviation of the measured values.
of internal airflow patterns and wind velocity distribution.
2.6. Experimental procedures The displacement of moving particles on virtual two dimen-
sional planes, which were made by a pulsed laser in the
2.6.1. Design of wind profile greenhouse, was measured. The PIV test was conducted with
The ESDU VIEWpac E01008 code has been programmed based on an oil droplet generator (Model 9307, TSI, USA), which was
the ESDU 84011 and 84030 code, and it can design profiles by used to make visible particles, an ultrahigh-speed CCD cam-
applying the variations in surface roughness around the target era (Phantom Miro M140, KOMI, Korea), a DPSS laser equip-
area (ESDU, 1993a; ESDU, 1993b; ESDU, 2005). Using the ESDU ment (MGL-N-532-5W, 12080009, CNI Laser, China), a DPSSL
VIEWpac E01008 code, the vertical profiles of wind speed and driver (CNI Laser, China), and a synchroniser (LASERPULSE
turbulence intensity in target reclaimed lands were designed Synchronizer Model 610036, TSI, USA). INSIGHT-4G (TSI, USA)
following previous research (Kwon et al., 2016). To design the computer software was used to analyse the airflow and
wind speed and turbulence intensity profiles, Google Maps and extract the data.
GIS data were used for determining the surface roughness of The similarity of the small-scale model was designed
upwind areas out to a radius of 50 km from the target reclaimed following previous studies (Kim, 2015; Kwon et al., 2016). The
land, and geographical characteristic was characterised. Fetch, geometrical similarity was set to 1:30 to meet the proper
which is the horizontal distance from a representative point of blockage ratio which is defined as the ratio of the cross-
each windward region to a point on the reclaimed land, was sectional areas of the scaled greenhouse to that of the wind
calculated on the basis of Google Maps. Wind profiles and tur- tunnel. The time scale was determined as 1:5 considering
bulence intensity profiles of target reclaimed lands were measuring conditions in the wind tunnel and the resolution of
designed using the calculated surface roughness and fetch in- instruments. The wind velocity scale was derived as 1:6 based
formation of each reclaimed land as the input data for ESDU. The on the law of similarity. Experimental wind speeds of 6.0 and
coefficient of roughness (a) which represent the wind environ- 9.0 m s1 at the reference height (small-scaled model of wind
ment profiles of seven reclaimed lands was determined. Finally, tunnel: 1.0 and 1.5 m s1) were selected based on the level of
the coefficient of roughness (a) was applied for installation of wind speed control and stable airflow formation, among
roughness blocks, barriers and spires in the wind tunnel on the others, and the wind direction was selected as perpendicular
basis of previous studies (Kwon et al., 2016; Lee et al., 2004). to the side wall. To minimise the error of experimental fluc-
tuation, the PIV test was repeatedly conducted after airflow
 a
UðzÞ z was stabilised. The air flow pattern in the greenhouse was
¼ (9)
UðzR Þ zR qualitatively and quantitatively measured using the PIV test in
the wind tunnel, and then the PIV measured internal airflow
where UðzÞ is the wind speed at z height (m s1), UðzR Þ is the
distribution was used to validate the CFD simulation model.
wind speed at the reference height (m s1) and a is the coef-
The design characteristics of the PIV test in the wind tunnel
ficient of roughness.
are shown in Table 2 and an overview of the PIV test in the
The designed wind speed and turbulence profiles were
wind tunnel is shown in Fig. 5. Additionally, the CFD simula-
applied in the CFD simulation based on log law (Hoxey &
tion was conducted under the same conditions with PIV test in
Richards, 1992; Richards, 1989; Richardson & Blackmore,
the wind tunnel.
1995). The roughness length (z0 ) is 0.03 m, the empirical con-
stant (Cm ) is 0.09 and the wind speed at the reference height
2.6.3. Experimental condition of computational fluid
were selected through wind frequency analyses (Kim, 2015).
dynamics simulation
u* z þ z0 During the design of the CFD simulation model, associative
UðzÞ ¼ ln (10)
k z0
relationships between internal and external domains are very
u*2
kðzÞ ¼ pffiffiffiffiffiffi (11) important because they can affect the computed results.
Cm
Designing a proper computational domain that works in
accordance with the objectives of the study is essential. If the
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
Table 2 e Design characteristics of the PIV test in wind
tunnel.
Contents Value
Length scale 1:30
Wind velocity scale 1:6
Time scale 1:5
Wind velocity 6.0 and 9.0 m s1
Wind direction Perpendicular (90 ) to the side wall
of small-scaled greenhouse model
computational domain is not large enough, the accuracy of
the results can be reduced in terms of the internal airflow,
wind velocity, pressure, and so on. On the other hand, the
larger computational domains require longer time for
designing and computations. However, there are no exact
criteria regarding the design of computational domains, and
the design depends on empirical values. Therefore, the
computational domain in this study was designed with 15H
(H: Ridge height of the greenhouse) in the windward direction,
15H in the leeward direction and 10H height from the ground,
based on previous research (Bournet & Boulard, 2010; Bournet,
Khaoua, Boulard, et al., 2007; Franke et al., 2004). The designed
computational domain is shown in Fig. 6. The grids near and
Fig. 5 e Overall view of the PIV test in the wind tunnel.
127
inside the greenhouse were designed densely to more accu-
rately analyse the airflow pattern in and near the greenhouse.
To evaluate the local ventilation rate in the greenhouse,
the CFD greenhouse model was designed to be divided into
virtual spaces. Considering the height at which the crops were
generally located, the greenhouse model was divided into a
section from the ground to a height of 1 m and an upper
section of the greenhouse. Additionally, the section from the
ground to a height of 1 m was divided into nine sections in the
longitudinal direction, and three sections in the lengthwise
direction, for a total of 27 virtual sections. The homogeneity of
the ventilation rate at the section from ground to 1 m height
was analysed through the TGD results. The coefficient of
variation, which is defined as the ratio of the standard devi-
ation to the mean, was used for quantitatively evaluating the
homogeneity of the local ventilation rate. Because the average
ventilation rates were different according to analysis condi-
tions, using the coefficient of variation was more reasonable
than using the standard deviation for estimating the homo-
geneity of local ventilation rates.
To validate the CFD simulation model, the turbulence
model test and grid independence test were conducted. It is
important to determine the optimum turbulence model
because air flow patterns and turbulent characteristics vary
128 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Fig. 6 e Design of the CFD simulation model domain.

depending on the turbulence model in the CFD simulation.

The turbulence model test was conducted with five different
Table 3 e Constant input values for the CFD simulation.
turbulence models (Standard k-ε, RNG k-ε, Realisable k-ε,
Variables Setting of the model
Standard k-u, and SST k-u) when wind speeds were 6.0 and
9.0 m s1 (small-scaled model of wind tunnel: 1.0 and Solver Pressure-based solver
1.5 m s1); furthermore, the wind direction was 90 and only Numerical algorithm SIMPLE algorithm
Discretisation Second-order
the side vent was open. The grid size of the CFD simulation
Time condition Steady-stated, and transient state
model for the turbulence model test was 0.1 m, which made it
Dimension Tree-dimensional simulation
as fine as possible to ensure accuracy of the CFD simulation. Operating pressure 101,325 Pa
Although the accuracy of the results is generally increased at Gravitational acceleration 9.81 m s2
the densely designed grid, excessively finer grid sizes can in- Air density 1.225 kg m3
crease the computation time. For accurate results and time Air viscosity 1.7894  105 kg m1 s1
effective computation, six grid sizes (0.1, 0.2, 0.4, 0.6, 0.8 and
1.0 m) were selected. Grid independence tests, according to
grid size, were conducted when wind speeds were 1.0 m s1, to take account of low, averaged, and high wind speed based
the wind direction was 90 and only the side vent was open. on wind frequency analyses of the wind environment from
Through grid independence tests, the threshold grid size, target reclaimed lands. When the wind direction was
which did not affect the accuracy of the CFD simulation, was perpendicular to the side vent, the wind direction was defined
determined. The proper turbulence model and optimum grid as 90 . In the case of the three-quarter type, the wind di-
size were selected by comparing the PIV measured results and rections 45 and 90 were also analysed because of the
CFD computed results, such as the internal air flow pattern asymmetrical structure. The condition of vent opening when
and vertical profile of wind speed at the centre of the internal only the side vent was fully open was called the ‘side vent’
volume of the greenhouse. To evaluate the CFD simulation condition. Likewise, the conditions of vent opening when the
model, the similarity between the PIV results and the CFD side and roof vents were fully open, and only the roof vent was
results was compared by using statistical indices such as R2, fully open were called the ‘side-roof vent’, and ‘roof vent’
RSR, and d. conditions, respectively. The analysis conditions for evalua-
The pressure-based solver and the semi-implicit method tion of natural ventilation rates using the CFD simulation are
for the pressured-linked equations (SIMPLE) algorithm were shown in Table 4.
applied to solve the governing equation of the CFD simulation.
The second-order scheme was used as the discretisation
method. As convergence criteria, the residuals of all variables
were 1  106. To simulate the airflow near wall accurately, yþ
Table 4 e Analysis conditions for evaluation of natural
was considered by adjusting the first layer height near the
ventilation rates using the CFD simulation.
wall. Additionally, steady-stated and transient analysis were
performed to apply MFR and TGD methods, respectively. The Contents Conditions
constant input values for the CFD simulation are listed in Greenhouse type Even-span type, peach type, three-
Table 3. quarter type and wide-width type
greenhouse
Natural ventilation rates were calculated and evaluated for
Wind speed 1.0, 2.5 and 5.5 m s1 (at 10 m
four greenhouse types (even-span type, peach type, three-
height)
quarter type and wide-width type), three wind speeds (1.0, Wind direction 0, 45 and 90 (three-quarter type:
2.5 and 5.5 m s1 at 10 m height), three wind directions (0, 45 additionally, 45 and 90 )
and 90 to the side wall) and three vent openings (side vent, Vent openings Side vent, side-roof vent and roof
side-roof vent and roof vent). Wind speeds were determined vent
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 129

3. Results and discussions

3.1. Validation of the computational fluid dynamics

simulation model

3.1.1. Wind profile of reclaimed coastal land

Wind speed and turbulence intensity profiles were designed
using ESDU code according to the target reclaimed lands and
are shown in Fig. 7 together with the wind speed and turbu-
lence intensity profiles, which were calculated using the
power law according to a values. Through comparisons of
wind speed and turbulence intensity profiles that were
designed by using ESDU code and the power law, the a value,
which can represent the wind environment of the reclaimed
land, was selected. The error bar of ESDU computed data was
caused by the distribution of wind characteristics according to
target reclaimed lands. Figure 7 shows that wind speed and
turbulence intensity profile at the a value of 0.15 were very
close to the ESDU computed data. As shown in Fig. 7, the
profiles of wind speed and turbulence intensity were over-
estimated or underestimated at the a value of 0.13 and 0.17.
When the a value were 0.13, 0.15, and 0.17, the average dif-
ference between the profiles of wind speed by ESDU and the
power law were 6.7%, 1.3%, and 3.8%, respectively. Likewise,
when the a value were 0.13, 0.15, and 0.17, the average dif-
ference between the profiles of turbulence intensity by ESDU Fig. 8 e Comparison of profiles of wind speed designed in
and power law were 7.2%, 1.9%, and 3.9%, respectively. the wind tunnel (a ¼ 0.15) and in the CFD simulation.
Therefore, the a value, representing the profiles of the seven
reclaimed lands, was selected as 0.15. similar to the profile of wind speed designed in the CFD
The wind environments in the wind tunnel were designed simulation.
to apply the a value of 0.15 by using the power law. Figure 8
shows the profiles of wind speeds that were compared with 3.1.2. Turbulence model test
data designed by a power law in wind tunnel and by a log law Turbulence model validation was conducted by comparing
in CFD. The average difference between data designed in wind the CFD simulation results and the PIV results when only the
tunnel and in the CFD simulation was 0.1%. Therefore, the side vent was fully open, the wind direction was
profile of wind speed designed in the wind tunnel was very

Fig. 7 e Estimation of the a value by comparing with ESDU computed wind profiles for reclaimed coastal lands.
130 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
perpendicular to the side wall and wind speeds were 6.0 m s1,
and 9.0 m s1 (small-scaled model of wind tunnel: 1.0, and
1.5 m s1). Figure 9 shows the vertical distribution of flow
speed at the centre of the inner greenhouse, which was
calculated by the CFD simulation results and the PIV results.
To quantitatively compare the computed CFD results with the
measured PIV results, statistical indices (R2, RSR, d, and MAE)
for each of the conditions were calculated and are shown in
Table 5.
From the quantitative statistics results, R2 and RSR showed
a similar tendency. When the wind speed was 6.0 m s1, R2
values under condition of RNG k-ε, SST k-u, and Standard k-u
turbulence model were 0.99, showing very high correlation
between CFD computed results and PIV measured results.
However, the RSR values under condition of RNG k-ε turbu-
lence model was lowest, 0.08, while the RSR values under
condition of realizable k-ε, Standard k-ε, SST k-u, and Stan-
dard k-u turbulence model were 0.31, 0.40, 0.15, and 0.12,
respectively. Furthermore, the MAE between the CFD
computed result and PIV measured result according to tur-
bulence models was the lowest, 0.12 m s1, when the RNG k-ε
turbulence model was applied. When the wind speed was
9.0 m s1, the R2 values under condition of RNG k-ε, SST k-u
and Standard k-u turbulence model were 0.99. However, the
RSR value and MAE between CFD computed result and PIV
measured result were the lowest, 0.07 and 0.14 m s1,
respectively, when RNG k-ε turbulence model was applied.
Therefore, the RNG k-ε model was selected for the CFD
simulation turbulence model to analyse the greenhouse
ventilation in this study, because the CFD computed result
Fig. 9 e The PIV measured and CFD computed air speed distributions at the centre of the greenhouse internal volume when
the wind direction was selected as perpendicular to the side wall, grid size was 0.1 m, and the external wind speeds were
6.0 m s¡1 (left) and 9.0 m s¡1 (right).
with RNG k-ε turbulence model were much closer to the PIV
experiment results at both wind speeds conditions.
3.1.3. Grid independence test
The appropriate grid was selected from the comparison of the
CFD computed results and PIV measured results, and it was
applied to design models for the analysis of the greenhouse
ventilation rates. Distributions of vertical wind speed at the
centre of the greenhouse were computed according to grid
sizes and are shown in Fig. 10, when the turbulence model
was RNG k-ε, the wind speed was 6.0 m s1 and the wind di-
rection was perpendicular to the side vent (90 ). The green-
house internal grid sizes of the CFD model were 0.1, 0.2, 0.4,
0.6, 0.8, and 1.0 m, and the total number of grids were
6,220,000, 1,470,000, 410,000, 200,000, 150,000, and 110,000,
respectively. The results by comparing with the CFD
computed results and the PIV measured results according to
the greenhouse internal grid sizes are shown in Table 6 as
statistical indices (R2, RSR, d, and MAE). Based on the com-
parisons of the distribution of the vertical flow speed at the
centre of the greenhouse with the internal grid size of the
greenhouse and the PIV results, the CFD computed results at
the smaller grid size were similar to the PIV measured results.
When the grid sizes were 0.1 and 0.2 m, R2 values were 0.997
and 0.996, RSR were 0.080 and 0.115, d were 0.999 and 0.997,
the MAE between the CFD computed result and the PIV
measured result according to the turbulence models were 0.12
and 0.18 m s1, respectively. The statistical index shows a
good agreement at both the 0.1 m and 0.2 m grid sizes. In this
study, an appropriate grid size was selected as 0.2 m
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 131

Table 5 e Statistical index for comparing the vertical wind profiles of the PIV measured results and the CFD computed
results at the centre of the greenhouse when the grid size was 0.1 m.
Statistical index Realisable k-ε RNG k-ε Standard k-ε SST k-u Standard k-u
1 2
6.0 m s R 0.98 0.99 0.97 0.99 0.99
RSR 0.31 0.08 0.40 0.15 0.12
d 0.97 0.99 0.95 0.99 0.99
MAE (m s1) 0.51 0.12 0.64 0.18 0.21
9.0 m s1 R2 0.94 0.99 0.96 0.99 0.99
RSR 0.33 0.07 0.40 0.13 0.10
d 0.97 0.99 0.95 0.99 0.99
MAE (m s1) 0.71 0.14 0.91 0.27 0.20

considering the simulation time, and it was applied to the CFD

model designed for the greenhouse ventilation analysis.

3.1.4. Internal air flow pattern

Qualitative validation was performed through comparisons of
the vector field, which was derived from PIV measured results
and CFD computed results based on the RNG k-ε turbulence
model and the 0.2 m grid size model. Airflow patterns
computed by CFD simulation and measured with the PIV test
are shown in Figs. 11 and 12, respectively, when only the side
vents were fully open and the wind speeds were 6.0 m s1 and
9.0 m s1 (small-scaled model of wind tunnel: 1.0 and
1.5 m s1), and wind direction was perpendicular to the side
wall (90 ). The results show that, when the wind speed was
6.0 m s1 and 9.0 m s1, the flow patterns of the PIV measured
and the CFD computed results were quite similar qualita-
tively. Furthermore, the differences in wind speed between
CFD computed result and PIV measured result were very
similar. Incoming airflow from the windward side, made the
main air stream at the lower side of the greenhouse and, it
was exhausted at the leeward side. A counter-clockwise
reverse flow was observed at the upper side of the green-
house, but the air current was weaker than the main air
stream.

3.2. Estimation of ventilation requirements

Fig. 10 e Vertical distribution of wind speed according to In this study, the ventilation requirements were estimated
grid size when the turbulence model was RNG k-ε, the according to the level of solar radiation and greenhouse type
wind speed was 6.0 m s¡1, and the wind direction was and are shown in Fig. 13. The ventilation requirements at the
perpendicular to the side vent (90 ). same conditions were higher with increasing the solar radia-
tion. Furthermore, at the same environmental conditions, the
ventilation requirements were calculated as high in the
following order: peach type, wide-width type, even-span type,
and three-quarters type greenhouse. The ventilation re-
quirements of peach type, wide-width type, and even-span
type were about 80%, 50% and 38% higher than the ventila-
Table 6 e Statistical index according to grid size when the
tion requirement of thee-quarter type greenhouse. The dif-
turbulence model was RNG k-ε, the wind speed was
ferences in the computed results of the ventilation
6.0 m s¡1, and the wind direction was perpendicular to
the side vent (90 ). requirements were due to structural differences in the
greenhouse types such as the volume of the greenhouse, floor
Statistical 0.1 m 0.2 m 0.4 m 0.6 m 0.8 m 1.0 m
index area, etc. Figure 13 presented in this study can be used as a
standard for greenhouse ventilation design through the
R2 0.99 0.99 0.99 0.92 0.70 0.55
calculation of the ventilation requirements depending on
RSR 0.08 0.12 0.31 0.36 0.64 0.74
d 0.99 0.99 0.98 0.96 0.87 0.82 solar radiation conditions in the greenhouse installation area.
MAE (m s1) 0.12 0.18 0.45 0.52 1.01 1.16 For example, when the outside air temperature was 30  C and
solar radiation was 600 W m2 at the greenhouse installation
132 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Fig. 11 e Airflow distribution comparison of the measured data from the PIV test (left) and the computed results of the CFD
simulation (right) of the greenhouse when the external air speed was 6.0 m s¡1.

Fig. 12 e Airflow distribution comparison of the measured data from the PIV test (left) and the computed results of the CFD
simulation (right) of the greenhouse when the external air speed was 9.0 m s¡1.

area in the summer, the ventilation requirements to maintain
the air temperature inside the even-span greenhouse at 33  C
using natural ventilation was approximately 2.0 AER min1.
3.3. Analysis of natural ventilation rate
3.3.1. Overall ventilation rate
The MFR computed ventilation rates were usually larger than
the TGD computed ventilation rate in the same environmental
conditions because the MFR method does not consider the
internal air flow pattern of the greenhouse and only considers
the mass flow at the vent area and the internal volume of
greenhouse. Figure 14 shows an example of the tracer gas
concentration contour depending on vent type in the even-
span greenhouse. When the windward roof ventilator and
leeward roof ventilator were fully open at the same time in the
even-span greenhouse (Fig. 14 (c)), the fresh air that flows into
the windward roof did not affect the ventilation of the
greenhouse sufficiently and it flowed out directly through the
leeward roof ventilator. In this case, it was shown that the
natural ventilation rates calculated by the MFR method could
be overestimated compared to the actual ventilation. On the
other hand, the TGD method could consider the internal air
flow pattern of the greenhouse to calculate the ventilation rate
because the ventilation rates were calculated by analysing the
reduction in the concentration of tracer gas at various points
inside the greenhouse. Therefore, it was shown that the TGD
method achieved a result that was closer to the actual natural
ventilation performance than the MFR method when evalu-
ating the overall and local ventilation rate of the naturally
ventilated greenhouse.
The natural ventilation rates were calculated using the
MFR and TGD methods according to greenhouse type, wind
direction, wind speed and vent type (Tables 7 and 8). To
analyse the overall ventilation rates of the greenhouse ac-
cording to greenhouse type, ventilation rates under the con-
dition of the opening of side vents were compared because the
peach type greenhouse does not have a roof ventilator. The
greenhouse type in which the highest ventilation rate was
observed was derived differently according to the calculation
methods of natural ventilation rates. Based on the MFR
method, the ventilation rate of the peach type greenhouse
was the highest, 8.07 AER min1, under the condition of wind
direction of 90 and wind speed of 5.5 m s1, while the even-
span, three-quarter, and wide-width types had 79%, 46%,
and 69% lower rates, respectively. Based on the TGD method,
the ventilation rate of the wide-width type greenhouse was
the highest, 3.97 AER min1, under conditions of wind direc-
tion of 90 and wind speed of 5.5 m s1, while the even-span,
peach, and three-quarter types had 56%, 56%, and 28% lower
rates, respectively.
The tendencies of overall ventilation rates were different
for the different types of greenhouses. As a representative
case, Fig. 15 shows the local ventilation rates according to
sections divided by height when wind direction was 0 , wind
speed was 1.0 m s1 and vent opening was side vent. Because
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 133

Fig. 13 e Ventilation requirements according to greenhouse type and difference between the inside and outside air
temperature.

the MFR method considered the mass of inflow and outflow
and the volume of the greenhouse, the ventilation rate of the
peach type greenhouse, which had the largest ratio of side
vent area to greenhouse volume, 0.049 m1, was the highest
and the ventilation rate of the three-quarter type greenhouse,
which had the smallest ratio of side vent area to greenhouse
volume, 0.025 m1, was the lowest. Based on the TGD
method, the ventilation rate of the peach type greenhouse at
the section from the ground to a height of 1 m (b) was the
highest, while the ventilation rate of the wide-width
greenhouse in the upper section of the greenhouse (a) was
the highest. It was assumed that the ventilation rates of
greenhouses in the upper section of the greenhouse (a) had
great influence on the ventilation rates of the whole green-
house because the volume of the upper section (a) was higher
than the volume of the section from the ground to a height of
1 m (b). These finding indicate, when the TGD method was
used, the ventilation rate of the wide-width type greenhouse
was the highest among the types of greenhouses for the same
environmental conditions.

Fig. 14 e Ventilation pattern of even-span greenhouse according to vent openings when the wind speed was 1.0 m s¡1 at
10 m height and the wind direction was perpendicular to the side wall (90 ).
134 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Table 7 e CFD computed ventilation rates based on the MFR method according to greenhouse type, vent type, wind speed
and wind direction (unit: AER min¡1).
Type Wind direction Side vent Side-roof vent Roof vent
1
Wind speed (m s )
1.0 2.5 5.5 1.0 2.5 5.5 1.0 2.5 5.5

Even-span 90 1.18 2.92 6.40 2.75 6.92 15.1 1.33 3.21 7.20
45 0.89 2.48 5.39 2.2 5.61 12.1 1.17 2.78 6.42
0 0.26 0.65 1.40 0.30 0.89 1.98 0.13 0.37 0.84
Peach 90 1.48 3.68 8.07
45 1.01 2.62 5.88
0 0.32 0.80 1.79
Three-quarter 90 0.67 1.69 3.75 1.43 2.34 5.26 0.21 0.52 1.13
45 0.53 1.31 2.72 0.75 1.98 4.30 0.20 0.44 1.19
0 0.16 0.52 1.15 0.26 0.65 1.39 0.07 0.16 0.33
45 0.55 1.50 2.95 0.86 2.06 4.40 0.22 0.49 1.22
90 0.71 1.79 3.95 1.45 2.36 5.77 0.24 0.54 1.24
Wide-width 90 1.02 2.53 5.54 1.39 3.40 7.41 0.38 0.91 2.02
45 0.87 2.14 3.43 1.02 2.49 5.46 0.33 0.82 1.70
0 0.68 0.95 1.31 0.54 0.74 1.57 0.08 0.20 0.47

Table 8 e CFD computed ventilation rates based on the TGD method according to greenhouse type, vent type, wind speed
and wind direction (unit: AER min¡1).
Type Wind direction Side vent Side-roof vent Roof vent
Wind speed (m s1)
1.0 2.5 5.5 1.0 2.5 5.5 1.0 2.5 5.5
Even-span 90 0.43 1.04 2.23 1.06 2.69 5.92 0.32 0.87 1.81
45 0.27 0.71 1.51 1.01 2.55 5.32 0.25 0.78 1.44
0 0.17 0.47 1.02 0.22 0.57 1.27 0.14 0.45 1.02
Peach 90 0.46 1.06 2.22
45 0.31 0.99 1.67
0 0.22 0.53 1.18
Three-quarter 90 0.22 0.51 1.10 0.52 1.33 2.82 0.08 0.23 0.51
45 0.17 0.40 0.85 0.40 0.89 1.92 0.06 0.18 0.39
0 0.13 0.31 0.66 0.16 0.42 0.93 0.02 0.09 0.23
45 0.18 0.42 0.88 0.42 0.91 1.97 0.09 0.24 0.41
90 0.24 0.55 1.21 0.55 1.35 2.84 0.10 0.24 0.52
Wide-width 90 0.75 1.83 3.97 0.73 1.71 3.69 0.11 0.27 0.58
45 0.53 1.46 2.41 0.51 1.22 2.73 0.10 0.26 0.57
0 0.33 0.57 0.83 0.21 0.48 1.09 0.07 0.19 0.47

Fig. 15 e The TGD computed local and overall ventilation rates according to sections divided by height of the greenhouse
when wind direction was 0 , wind speed was 1.0 m s¡1 and only the side vent was fully open.
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 135

To evaluate the effect of the wind environment on natural
ventilation, the natural ventilation rates were calculated and
analysed according to wind speed and wind direction. The
results showed, as the wind speed increased to 2.5 m s1 and
5.5 m s1 from 1.0 m s1, the ventilation rates were increased
linearly in all cases for both the MFR and TGD methods. On
average, the MFR computed ventilation rates increased by
about 2.4 and 5.1 times as the wind speed increased from
1.0 m s1 to 2.5 m s1 and 5.5 m s1, respectively. On average,
the TGD computed ventilation rates increased by about 2.5
and 5.2 times as the wind speed increased from 1.0 m s1 to
2.5 m s1 and 5.5 m s1, respectively. As a result of analysing
ventilation rates according to wind directions, for all types of
the greenhouse, the highest ventilation rate was observed
with the wind perpendicular to the side ventilator, while the
lowest ventilation rates were obtained with the wind parallel
to the side ventilator. In particular, the difference in the
ventilation rates for the wind directions 90 and 0 was high.
This is because the amount of air inflow through the ventilator
at a wind direction of 0 was lower than that at a wind di-
rection of 90 . In the case of the three-quarter type green-
house, the ventilation rates at wind directions of 45 and
90 were slightly larger than those at 45 and 90 , respec-
tively. However, the non-symmetric structure of the three-
quarter greenhouse did not considerably affect the internal
air flow pattern and performance of ventilation.
To evaluate the effect of vent openings on natural venti-
lation, the natural ventilation rates were calculated and ana-
lysed according to vent openings (side vent, side-roof vent,
and roof vent). The results showed the MFR and TGD
computed ventilation rates increased in the following order:
side-roof vent, side vent, and roof vent condition. On average,
the natural ventilation rates under the side-roof vent and roof
vent conditions were 78% higher and 54% lower than those
under the side vent conditions.
As result of analysing the natural ventilation rates ac-
cording to evaluation methods (MFR and TGD), when the wind
speed were 1.0, 2.5, and 5.5 m s1, the average errors between
ventilation rates computed by the two method were 0.45, 1.02,
and 2.28 AER min1, respectively. Likewise, the average error
of the even-span, peach, three-quarter, and wide-width types
greenhouse were 2.12, 1.89, 0.86, and 0.8 AER min1, respec-
tively. Compared to other greenhouse types, the overall
ventilation rates of the wide-width type greenhouse were less
affected by differences in the calculation methods used for
estimating natural ventilation rates.
3.3.2. Local ventilation rate
To analyse the homogeneity of local ventilation rates, co-
efficients of variation in the local ventilation rate at the sec-
tion from the ground to a height of 1 m were calculated and
are shown in Table 9. The analysis according to greenhouse
type showed that the averaged coefficient of variation of the
even-span type greenhouse was the lowest at 0.14, while the
averaged coefficients of variation of the peach, three-quarter,
and wide-width type greenhouses were 2.9, 2.7, and 1.3 times
higher than that of the even-span type greenhouse, respec-
tively. Therefore, compared to the target greenhouse types,
the results from the even-span type greenhouses exhibited
relatively high homogeneity.
In addition, the same analysis according to wind speed
revealed that the standard deviations increased with an in-
crease in wind speed. However, coefficients of variation
scarcely changed because the average ventilation rates were
also increasing following the increase of the wind speed. As a
comparison result according to wind directions, the averaged
coefficients of variation under condition of wind direction
0 were the lowest, 0.08, while the averaged coefficient of
variation under condition of wind direction 45, 90 were 4.8
and 3.0 times higher than that under condition of wind di-
rection 0 , respectively. However, when the wind direction
was 0 , the average ventilation rates were very low and ven-
tilations were performed inadequately. Therefore, both the
average ventilation rate and homogeneity should be consid-
ered simultaneously for estimating proper ventilation
performance.
The homogeneities of local ventilation rates at the section
from the ground to a height of 1 m were analysed according to

Table 9 e Coefficient of variation at the section from the ground to a height of 1 m according to greenhouse type, vent type,
wind speed and wind direction.
Type Wind direction Side vent Side-roof vent Roof vent
Wind speed (m s1)
1.0 2.5 5.5 1.0 2.5 5.5 1.0 2.5 5.5
Even-span 90 0.28 0.30 0.32 0.20 0.20 0.19 0.04 0.08 0.06
45 0.29 0.32 0.30 0.12 0.15 0.10 0.11 0.13 0.10
0 0.06 0.04 0.05 0.05 0.04 0.03 0.14 0.04 0.04
Peach 90 0.39 0.39 0.40
45 0.68 0.98 0.66
0 0.05 0.04 0.04
Three-quarter 90 0.42 0.42 0.45 0.48 0.46 0.48 0.14 0.06 0.07
45 0.62 1.05 0.84 0.85 1.05 0.76 0.22 0.19 0.18
0 0.14 0.12 0.13 0.06 0.09 0.09 0.17 0.05 0.07
45 0.80 0.91 0.82 0.65 0.77 0.66 0.10 0.05 0.11
90 0.42 0.42 0.42 0.26 0.25 0.25 0.11 0.04 0.06
Wide-width 90 0.33 0.34 0.35 0.19 0.17 0.17 0.10 0.04 0.03
45 0.34 0.32 0.28 0.35 0.36 0.27 0.03 0.08 0.10
0 0.09 0.09 0.06 0.10 0.08 0.07 0.20 0.15 0.11
136 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

the conditions of vent openings. The results showed the

averaged coefficient of variation under condition of opening of
roof vent was 0.01, which was the lowest observed value,
while the averaged coefficients of variation under condition of
openings of side vent and side-roof vent were 3.8 and 3.1 times
higher than that under condition of wind direction 0 ,
respectively. Furthermore, coefficients of variation under
condition of opening of side-roof vent were about 0.07 lower
than those under condition of opening of side vent. When only
the roof vents were open, coefficients of variation were sub-
stantially lower than the other vent opening type. Therefore,
relatively high homogeneity of local ventilation rates at the
section from the ground to a height of 1 m was observed when
roof vents were only open.
To analyse the local ventilation rate specifically according
wind directions, the analysis results of the local ventilation
rate of Wide-width type greenhouse were presented as an
example cases because the differences in homogeneity ac-
cording to wind direction were relatively high among other
analysis conditions. Figure 16 shows the contour of the tracer
gas concentration of wide-width type greenhouse over time at
1 m height from ground when only the side ventilators were
fully open and the wind speed was 1.0 m s1. Figure 17 also
shows the distribution of TGD computed local ventilation
rates of the wide-width type greenhouse at the section from
the ground to a height of 1 m in the greenhouse according to
wind direction when the wind speed was 1.0 m s1 and only
the side ventilators were fully open. Favourable air exchange
of the overall greenhouse was observed under condition of
wind direction 90 because the fresh air come in through the
windward side ventilator flowed from the windward side to
the leeward side directly. Local ventilation rates at both end
walls of the windward side were relatively higher than those
at the centre on the leeward side. When the wind direction
was 45 , the end section of the windward side had a relatively
low ventilation rate. When the wind direction was 0 , the
overall ventilation rates were low. In particular, the local
ventilation rate at the end wall of the windward side was low.
The maximum ventilation rate among all of the sections from
ground to 1 m height were 3.0, 3.6, and 1.3 times higher than
minimum ventilation rate under the wind direction of 90, 45,
and 0 , respectively. Because the local ventilation rates at the
sections from ground to 1 m height were significantly
different, local ventilation in the greenhouse should be
considered when the natural ventilation system of the
greenhouse is designed or evaluated.

3.4. Evaluation of natural ventilation by comparing

with the ventilation requirement
Finally, the charts for expecting natural ventilation rates were
compiled according to greenhouse type. Figure 18 shows the
charts that were plotted using the natural ventilation rate
calculated from the MFR method and Fig. 19 shows the charts
that were plotted using the natural ventilation rate calculated
from the TGD method. To compare the natural ventilation rate
and ventilation requirements at some given conditions,
ventilation requirements for maintaining 2, 4, and 6  C of the
air temperature differences in 800 W m2 of solar radiation
were also expressed in the charts with dotted horizontal lines.
Fig. 16 e Distribution of tracer gas concentrations of wide-
width type greenhouse according to wind direction when
the wind speed was 1.0 m s¡1 and only the side vent was
fully open.
The charts suggested in this study will be proposed for
ventilation standards of greenhouses and be practically used.
The charts were used to evaluate whether the natural
ventilation rates meet the ventilation requirements at specific
conditions for each greenhouse type. Each greenhouse type
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
Fig. 17 e Distribution of local ventilation rates (AER min¡1) of wide-width type greenhouse at the section from the ground to
a height of 1 m according to wind direction when the wind speed was 1.0 m s¡1 and only the side vent was fully open.
needs different ventilation rates to maintain optimum envi-
ronments because of the difference in structural features. The
natural ventilation rates calculated from the TGD method
were used to explain the comparative analysis of the venti-
lation requirements and the natural ventilation rates because
it was judged that the TGD method evaluated the actual nat-
ural ventilation more closely than MFR method.
As shown in Fig. 19 (a), the straight lines indicating natural
ventilation rates of the even type greenhouse were located
below the horizontal lines indicating the ventilation require-
ment for maintaining the inside air temperature at the dif-
ference of 6  C from the outside air temperature when the
wind direction was 0 . This means that the natural ventilation
rates did not meet the ventilation requirement for maintain-
ing the inside air temperature at the difference of 6  C from
the outside air temperature when the wind direction was 0 .
However, the natural ventilation rates of the Even type
greenhouse satisfied the ventilation requirements for main-
taining inside air temperature at the difference of 6  C from
the outside air temperature at side-roof vent condition and
the wind direction condition of 90 or 45 when the wind speed
was over 1.0 m s1. Additionally, the natural ventilation rates
of the Even type greenhouse met the ventilation requirements
for maintaining inside air temperature at the difference of 2  C
from the outside air temperature at side-roof vent condition
and the wind direction condition of 90 or 45 when wind speed
was over 4.0 m s1.
For the peach type greenhouse, all of the straight lines
indicating natural ventilation rates were plotted below the
horizontal lines indicating the ventilation requirement for
maintaining the inside air temperature at the difference of 2
and 4  C from the outside air temperature when the wind di-
rection was 0 (Fig. 19 (b)). It means that the natural ventilation
137
could not satisfy the ventilation requirements for maintaining
the inside air temperature at differences of 2 and 4  C from the
outside in any condition. When the wind direction was 90
and the wind speed was over 4 m s1, natural ventilation rates
were enough for maintaining the inside air temperature at a
difference of 6  C from the outside air temperature. When the
wind direction was 45 or 0 , the ventilation rate could not
satisfy the ventilation requirements for maintaining the in-
side air temperature at a difference of 6  C from the outside air
temperature, even if the wind speed was over 5.5 m s1.
In the case of three-quarter type greenhouse, the natural
ventilation could not satisfy the ventilation requirements for
maintaining the inside air temperature at differences of 2  C
from the outside air temperature in any case of this study
(Fig. 19 (c)). However, under the side-roof vent condition, the
natural ventilation rates of three-quarter type greenhouse
satisfied the ventilation requirements for maintaining the
inside air temperature at a difference of 6  C from the outside
air temperature when the condition of wind speed was over
1.5 m s1 under conditions of the wind direction of 90 and
90 , and over 2.5 m s1 under conditions of the wind direction
of 45 and 45 .
In the case of the wide-width type greenhouse, natural
ventilation rates at the roof vent condition could not satisfy
the ventilation requirements for maintaining the inside air
temperature at differences of 2, 4, and 6  C from the outside air
temperature in any case of this study (Fig. 19 (d)). Further-
more, the natural ventilation rates at the side vent or side-roof
vent condition could not satisfy the ventilation requirements
for maintaining the inside air temperature at differences of 2,
4 and 6  C from the outside air temperature when the wind
direction was 0 . However, the natural ventilation rates at the
side vent or side-roof vent condition satisfied the ventilation
138 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2

Fig. 18 e Chart for estimating natural ventilation rates based on MFR method.
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2 139

Fig. 19 e Chart for estimating natural ventilation rates based on the TGD method.
140 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
requirements for maintaining the inside air temperature at
differences of 4 and 6  C from the outside air temperature
wind speed was over 3.0 m s1 under conditions of the wind
direction of 90 and 45 .
4. Conclusions
In this study, as basic research, wind-driven natural ventila-
tion rates of greenhouse built on reclaimed land were ana-
lysed according to wind speed, wind direction, vent openings,
and greenhouse types using a CFD simulation. The wind speed
and turbulence intensity profiles designed by using ESDU code
were used in a wind tunnel test and CFD simulation. The CFD
simulation models were verified by comparing CFD computed
results with PIV measured results qualitatively and quantita-
tively. The internal grid size of the greenhouse model was
selected as 0.2 m, and the turbulence model was selected as
the RNG k-ε model, considering the accuracy and computa-
tional speed.
By analysing the overall ventilation rates of the greenhouse,
the ratio of side vent area to greenhouse volume is an important
factor when estimating the natural ventilation rate by using the
MFR method. On the other hand, the internal airflow pattern of
a greenhouse is a main factor when estimating the natural
ventilation rate by using the TGD method. For this reason, the
overall natural ventilation rates computed by using the MFR
and TGD method showed different tendencies. Analysis of the
overall ventilation rates according to wind speed indicated that
the ventilation rates increased linearly as the wind speed
increased. In addition, analysis of the overall ventilation rates
according to wind direction reveal that ventilation rates
decreased as the wind direction changed from 90 to 0 . As a
result of analysing the homogeneity of the local ventilation
rates at the section from the ground to a height of 1 m, it was
found that the homogeneity was mainly influenced by wind
direction and the configuration of the ventilators, but not by
wind speed. Finally, charts for predicting natural ventilation
rates according to the design environment and greenhouse
type were suggested using the CFD computed natural ventila-
tion rates. The charts could be used for determining whether
natural ventilation rates of greenhouses meet the ventilation
requirements of the environmental condition of the installa-
tion area. Additionally, the results of the analysis of the local
ventilation rates could be used for uniformly controlling the
microclimate environments of single span greenhouses.
The significance of this study is that it considered wind
environments of reclaimed land to compute and evaluate the
wind-driven ventilation of single-span greenhouses (without
the presence of internal plants) after the CFD simulation
model was validated by comparing its results to those of a
wind tunnel test at wind-driven conditions. A study for eval-
uating the effects of crops and buoyancy on natural ventila-
tion of greenhouse built on reclaimed coastal lands in Korea is
still required. Additionally, although the wind velocity is very
high near reclaimed coastal land, buoyancy-driven ventilation
may be more important at large-sized multi-span green-
houses. Therefore, the next step will be the evaluation of
natural ventilation performance considering the buoyancy
effect and crop effect on natural ventilation of multi-span
greenhouses.
Acknowledgement
This work was carried out with the support of the "Develop-
ment of core technology about Korea type of ICT convergence
smart farm (Project No. PJ0120382018)" Rural Development
Administration, Republic of Korea.
references
Aich, W., Kolsi, L., Borjini, M. N., Aissia, H. B., Oztop, H. F., &
Abu-Hamdeh, N. (2016). Three dimensional CFD analysis of
buoyancy-driven natural ventilation and entropy
generation in a prismatic greenhouse. Thermal Science, (00),
52e52.
rez-Parra, J. J., Montero, J. I., Bailey, B. J., Lo
 pez, J. C.,
Baeza, E. J., Pe
& Ga zquez, J. C. (2009). Analysis of the role of sidewall vents on
buoyancy-driven natural ventilation in parral-type
greenhouses with and without insect screens using
computational fluid dynamics. Biosystems Engineering, 104(1),
86e96.
Bartzanas, T., Boulard, T., & Kittas, C. (2004). Effect of vent
arrangement on windward ventilation of a tunnel greenhouse.
Biosystems Engineering, 88(4), 479e490.
Benni, S., Tassinari, P., Bonora, F., Barbaresi, A., & Torreggiani, D.
(2016). Efficacy of greenhouse natural ventilation:
Environmental monitoring and CFD simulations of a study
case. Energy and Buildings, 125, 276e286.
Boulard, T., Haxaire, R., Lamrani, M. A., Roy, J. C., & Jaffrin, A.
(1999). Characterization and modelling of the air fluxes
induced by natural ventilation in a greenhouse. Journal of
Agricultural Engineering Research, 74(2), 135e144.
Boulard, T., Meneses, J. F., Mermier, M., & Papadakis, G. (1996).
The mechanisms involved in the natural ventilation of
greenhouses. Agricultural and Forest Meteorology, 79(1e2),
61e77.
Boulard, T., Papadakis, G., Kittas, C., & Mermier, M. (1997). Air flow
and associated sensible heat exchanges in a naturally
ventilated greenhouse. Agricultural and Forest Meteorology,
88(1e4), 111e119.
Bournet, P.-E., & Boulard, T. (2010). Effect of ventilator
configuration on the distributed climate of greenhouses: A
review of experimental and CFD studies. Computers and
Electronics in Agriculture, 74(2), 195e217.
Bournet, P., Khaoua, S. O., & Boulard, T. (2007). Numerical
prediction of the effect of vent arrangements on the
ventilation and energy transfer in a multi-span glasshouse
using a bi-band radiation model. Biosystems Engineering, 98(2),
224e234.
Bournet, P., Khaoua, S. O., Boulard, T., Migeon, C., &
Chasse riaux, G. (2007). Effect of roof and side opening
combinations on the ventilation of a greenhouse using
computer simulation. Transactions of the ASABE, 50(1), 201e212.
Campen, J., & Bot, G. (2003). Determination of greenhouse-specific
aspects of ventilation using three-dimensional computational
fluid dynamics. Biosystems Engineering, 84(1), 69e77.
Choi, M.-K., Yun, S.-W., Yu, I.-H., Lee, S.-Y., & Yoon, Y.-C. (2015).
Settlement instrumentation of greenhouse foundation in
reclaimed land. Protected Horticulture and Plant Factory, 24(2),
85e92.
 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
ESDU. (1993a). Wind speed profiles over terrain with roughness
changes. Engineering Sciences Data Unit International. Data
Item 84011.
ESDU. (1993b). Longitudinal turbulence intensities over terrain with
roughness changes. Engineering Science Data Unit
International. Data Item 84030.
ESDU. (2005). Computer program for wind speeds and turbulence
properties: Flat or hilly sites in terrain with roughness changes.
Engineering Science Data Unit International. Data Item 01008.
Fatnassi, H., Boulard, T., Bouirden, L., & Sappe, G. (2002).
Ventilation performance of a large canarian type greenhouse
equipped with insect-proof nets. Acta Horticulturae, 578, 79e88.
Fatnassi, H., Boulard, T., Poncet, C., & Chave, M. (2006).
Optimisation of greenhouse insect screening with
computational fluid dynamics. Biosystems Engineering, 93(3),
301e312.
Fatnassi, H., Leyronas, C., Boulard, T., Bardin, M., & Nicot, P.
(2009). Dependence of greenhouse tunnel ventilation on wind
direction and crop height. Biosystems Engineering, 103(3),
338e343.
Flores-Velazquez, J., Montero, J. I., Baeza, E. J., & Lopez, J. C. (2014).
Mechanical and natural ventilation systems in a greenhouse
designed using computational fluid dynamics. International
Journal of Agricultural and Biological Engineering, 7(1), 1.
Franke, J., Hirsch, C., Jensen, A., Krüs, H., Schatzmann, M.,
Westbury, P., et al. (2004). Recommendations on the use of
CFD in wind engineering. In Paper presented at the Cost Action C.
Ha, J. (2015). Evaluation of natural ventilation efficiency of protected
cultivation system in reclaimed land using aerodynamic simulation.
Master Thesis. Seoul, South Korea: Seoul National University.
He, K.-S., Chen, D.-Y., Sun, L.-J., Liu, Z.-L., & Huang, Z.-Y. (2015).
The effect of vent openings on the microclimate inside multi-
span greenhouses during summer and winter seasons.
Engineering Applications of Computational Fluid Mechanics, 9(1),
399e410.
Hong, S.-W., Lee, I.-B., Hwang, H.-S., Seo, I.-H., Bitog, J., Yoo, J.-I.,
et al. (2008). Numerical simulation of ventilation efficiencies of
naturally ventilated multi-span greenhouses in Korea.
Transactions of the ASABE, 51(4), 1417e1432.
Hoxey, R. P., & Richards, P. J. (1992). Structure of the atmospheric
boundary layer below 25 m and implications to wind loading
on low-rise buildings. Journal of Wind Engineering and Industrial
Aerodynamics, 41(1), 317e327.
Ishii, M., Okushima, L., Moriyama, H., Sase, S., Fukuchi, N., &
Both, A. (2014). Experimental study of natural ventilation in an
open-roof greenhouse during the summer. In Paper presented at
the XXIX International Horticultural Congress on Horticulture:
Sustaining Lives, Livelihoods and Landscapes (IHC2014) (p. 1107).
Kacira, M., Sase, S., & Okushima, L. (2004). Effects of side vents
and span numbers on wind-induced natural ventilation of a
Gothic multi-span greenhouse. Japan Agricultural Research
Quarterly: JARQ, 38(4), 227e233.
Kim, R.-W. (2015). Evaluation of wind pressure coefficients of
greenhouses using wind tunnel test and numerical model. Master
Thesis. Seoul, South Korea: Seoul National University.
Kim, R.-W., Kim, D.-W., Ryu, K.-C., Kwon, K.-S., & Lee, I.-B. (2014).
Estimation of wind pressure coefficients on even-span
greenhouse built in reclaimed land according to roof slop
using wind tunnel. Protected Horticulture and Plant Factory, 23(4),
269.
Kim, K., Yoon, J.-Y., Kwon, H.-J., Han, J.-H., Son, J. E., Nam, S.-W.,
et al. (2008). 3-D CFD analysis of relative humidity distribution
in greenhouse with a fog cooling system and refrigerative
dehumidifiers. Biosystems Engineering, 100(2), 245e255.
Kittas, C., Boulard, T., Mermier, M., & Papadakis, G. (1996). Wind
induced air exchange rates in a greenhouse tunnel with
continuous side openings. Journal of Agricultural Engineering
Research, 65(1), 37e49.
141
Kittas, C., Draoui, B., & Boulard, T. (1995). Quantification of the
ventilation of a greenhouse with a roof opening. Agricultural
and Forest Meteorology, 1(77), 95e111.
Korea Meteorological Administration. (2014). Available at http://
www.kma.go.kr. (Accessed May 2015).
Kozai, T., Sase, S., & Nara, M. (1980). A modelling approach to
greenhouse ventilation control. Acta Horticulturae, 106, 125e136.
Kwon, K.-S., Kim, D.-W., Kim, R.-W., Ha, T., & Lee, I.-B. (2016).
Evaluation of wind pressure coefficients of single-span
greenhouses built on reclaimed coastal land using a large-
sized wind tunnel. Biosystems Engineering, 141, 58e81.
Kwon, J.-K., Lee, S.-H., Seong, J.-H., Moon, J.-P., Lee, S.-J., Choi, B.-
M., et al. (2011). Analysis of natural ventilation characteristics
of venlo-type greenhouse with continuous roof vents. Journal
of Biosystems Engineering, 36(6), 444e452.
Lee, I.-B., Choi, K., Yun, J., Jeun, J., & Kim, G. (2003). Study of
aerodynamics in agricultureemodern technologies. In Paper
presented at the 2003 ASAE annual meeting.
Lee, S.-Y., Kang, D.-H., Kim, J.-G., Kim, Y.-J., Yoon, Y.-C., Son, C.-
S., et al. (2015). Development of the stabilization method of
reclaimed land to establish the foundation of greenhouse. South
Korea: Rural Development Administration.
Lee, I., Kang, C., Lee, S., Kim, G., Heo, J., & Sase, S. (2004).
Development of vertical wind and turbulence profiles of wind tunnel
boundary layers (Vol. 47). St. Joseph, MI: ETATS-UNIS: American
Society of Agricultural Engineers.
Lee, I., Lee, S., Kim, G., Sung, J., Sung, S., & Yoon, Y. (2005). PIV
verification of greenhouse ventilation air flows to evaluate
CFD accuracy. Transactions of the ASAE, 48(6), 2277e2288.
Lee, I., Okushima, L., Ikeguchi, A., Sase, S., & Short, T. (2000).
Prediction of natural ventilation of multi-span greenhouses
using CFD techniques and its verification with wind tunnel
test. In Paper presented at the 2000 ASAE annual international
meeting, Milwaukee, Wisconsin, USA, 9e12 July 2000.
Lee, I., Sase, S., Okushima, L., Ikeguchi, A., Choi, K., & Yun, J. (2003).
A wind tunnel study of natural ventilation for multiespan
greenhouse scale models using twoedimensional particle
image velocimetry (PIV). Transactions of the ASAE, 46(3), 763.
Lee, I., & Short, T. (2000). Two-dimensional numerical simulation
of natural ventilation in a multi-span greenhouse. Transactions
of the ASAE, 43(3), 745e753.
Lee, I.-B., & Short, T. (2001). Verification of computational fluid
dynamic temperature simulations in a full-scale naturally
ventilated greenhouse. Transactions of the ASAE, 44(1), 119e127.
Majdoubi, H., Boulard, T., Fatnassi, H., & Bouirden, L. (2009).
Airflow and microclimate patterns in a one-hectare canary
type greenhouse: An experimental and CFD assisted study.
Agricultural and Forest Meteorology, 149(6), 1050e1062.
Ministry of Agriculture, Food and Rural Affairs of Korea. (2014).
Korean design standards of horticultural facilities.
Ministry of Agriculture, Food and Rural Affairs of Korea. (2016).
Statistics of the agriculture, food and rural affairs.
Ministry of Food, Agricultural, Forestry and Fisheries. (2010).
Notification for using reclaimed land.
Ministry of Land, Transport and Maritime Affairs of Korea. (1997).
Korean design standards of glass-type greenhouse.
Mistriotis, A., Bot, G., Picuno, P., & Scarascia-Mugnozza, G. (1997).
Analysis of the efficiency of greenhouse ventilation using
computational fluid dynamics. Agricultural and Forest
Meteorology, 85(3), 217e228.
Molina-Aiz, F. D., Fatnassi, H., Boulard, T., Roy, J. C., & Valera, D. L.
(2010). Comparison of finite element and finite volume
methods for simulation of natural ventilation in greenhouses.
Computers and Electronics in Agriculture, 72(2), 69e86.

Molina-Aiz, F. D., Valera, D. L., & Alvarez, A. J. (2004).
Measurement and simulation of climate inside Almerıa-type
greenhouses using computational fluid dynamics. Agricultural
and Forest Meteorology, 125(1), 33e51.
142 b i o s y s t e m s e n g i n e e r i n g 1 7 1 ( 2 0 1 8 ) 1 2 0 e1 4 2
Okushima, L., Sase, S., Maekawa, T., & Ikeguchi, A. (1998). Airflow
patterns forced by wind effect in a Venlo type greenhouse.
Journal of the Society of Agricultural Structures, 29(1), 159e167.
Okushima, L., Sase, S., Maekawa, T., Ikeguchi, A., & Bailey, B. J.
(2000). Temperature distribution in a naturally ventilated
venlo greenhouse with six spans and high eaves: A wind
tunnel study. Nogyo Shisetsu (Journal of the Society of Agricultural
Structures, Japan), 30(4), 343e352.
Okushima, L., Sase, S., & Nara, M. (1989). A support system for
natural ventilation design of greenhouse based on
computational aerodynamics. Acta Horticulturae, 248, 129e136.
Ould Khaoua, S. A., Bournet, P. E., Migeon, C., Boulard, T., &
Chasse riaux, G. (2006). Analysis of greenhouse ventilation
efficiency based on computational fluid dynamics. Biosystems
Engineering, 95(1), 83e98.
rez Parra, J., Baeza, E., Montero, J. I., & Bailey, B. J. (2004). Natural
Pe
ventilation of parral greenhouses. Biosystems Engineering, 87(3),
355e366.
Richards, P. (1989). Computational modelling of wind flow around low-
rise buildings using PHOENICS. AFRC Institute of Engineering
Research Buildings and Livestock Division.
Richardson, G. M., & Blackmore, P. A. (1995). The silsoe structures
building: Comparison of 1 : 100 model-scale data with full-
scale data. Journal of Wind Engineering and Industrial
Aerodynamics, 57(2e3), 191e201.
Rigakis, N., Katsoulas, N., Teitel, M., Bartzanas, T., & Kittas, C.
(2015). A simple model for ventilation rate determination in
screenhouses. Energy and Buildings, 87, 293e301.
Roy, J., & Boulard, T. (2004). CFD prediction of the natural
ventilation in a tunnel-type greenhouse: Influence of wind
direction and sensibility to turbulence models. In Paper
presented at the International conference on sustainable greenhouse
systems-greensys 2004 691.
View publication stats
Rural Research Institute of Korea. (1997). Design standards for
greenhouse environment.
Sase, S. (1989). The effects of plant arrangement on airflow
characteristics in a naturally ventilated glasshouse. Acta
Horticulturae, 245, 429e435.
Sase, S., Takakura, T., & Nara, M. (1984). Wind tunnel testing on
airflow and temperature distribution of a naturally ventilated
greenhouse. Acta Horticulturae, 148, 329e336.
Teitel, M., & Wenger, E. (2014). Air exchange and ventilation
efficiencies of a monospan greenhouse with one inflow and
one outflow through longitudinal side openings. Biosystems
Engineering, 119, 98e107.
Teitel, M., Ziskind, G., Liran, O., Dubovsky, V., & Letan, R. (2008).
Effect of wind direction on greenhouse ventilation rate,
airflow patterns and temperature distributions. Biosystems
Engineering, 101(3), 351e369.
Villarreal-Guerrero, F., Kacira, M., Fitz-Rodrı́guez, E., Linker, R.,
Giacomelli, G., Arbel, A., et al. (2013). Implementation of a
greenhouse cooling strategy with natural ventilation and
variable fogging rates. Transactions of the ASABE, 56(1),
295e304.
Willmott, C. J., Ackleson, S. G., Davis, R. E., Feddema, J. J.,
Klink, K. M., Legates, D. R., et al. (1985). Statistics for the
evaluation and comparison of models.
Yu, I.-H., Ku, Y.-G., Cho, M.-W., Ryu, H.-R., & Moon, D.-G. (2014).
An analysis of problems and countermeasures in the
installation of plastic greenhouse on reclaimed lands. Korean
Journal of Agricultural Science, 41(4), 473e480.
Yun, S.-w., Choi, M.-k., Lee, J.-w., Lee, S.-y., & Yoon, Y.-c. (2013).
Estimation of design load for greenhouse applicable in coastal
reclaimed lands. Journal of Agriculture & Life Science, 47(3),
123e131.