-
Biosystems Engineering (2004) 88 (4),
479490doi:10.1016/j.biosystemseng.2003.10.006
Available online at www.sciencedirect.com
SE}Structures and Environment
Effect of Vent Arrangement on Windward Ventilation of a Tunnel
Greenhouse
T. Bartzanas1; T. Boulard2; C. Kittas1
1Department of Agriculture Crop Production and Agriculture
Environment, University of Thessaly, School of Agriculture
Science,Fytokou St., 38446, N.Ionia Magnisias, Greece; e-mail of
corresponding author: [email protected]
2 INRA, Unite Plantes et Systemes de Culture Horticoles, Domaine
St Paul, Site Agroparc, 84914 Avignon Cedex 09, France;e-mail:
[email protected]
(Received 22 May 2003; accepted in revised form 8 October 2003;
published online 1 July 2004)
The effect of ventilation conguration of a tunnel greenhouse
with crop on airow and temperature patternswas numerically
investigated using a commercial computational uid dynamics (CFD)
code. The numericalmodel was rstly validated against experimental
data collected in a tunnel greenhouse identical with the oneused in
simulations. The airow patterns were measured and collected using a
three-dimensional sonicanemometer and the greenhouse ventilation
rate was deduced using a tracer gas technique. A good
qualitativeand quantitative agreement was found between the
numerical results and the experimental measurements.After its
validation, the CFD model was used to study the consequences of
four different ventilatorcongurations on the natural ventilation
system. The ventilation conguration affects the ventilation rate
ofthe greenhouse and the airow and air temperature distributions as
well. For the different congurations,computed ventilation rates
varied from 10 to 58 air changes per hour for an outside wind speed
of 3m s1 andfor a wind direction perpendicular to the openings.
Likewise, the simulations highlight that while the mean
airtemperature at the middle of the tunnels varied from 282 to
2988C, for an outside air temperature of 288C,there are regions
inside tunnels 68C warmer than outside air. Average air velocity in
the crop cover variedaccording to the arrangement of the vents from
02 to 07m s1. The consequences of the marked climateheterogeneity
on plant activity through the variation of crop aerodynamic
resistance as well as the inuence ofthe vent congurations on the
efciencies of ventilation on ow rate and air temperature
differences betweeninside and outside, are also discussed.# 2003
Silsoe Research Institute. All rights reserved
Published by Elsevier Ltd
1. Introduction
Greenhouse tunnels are widely used in the wholeworld due to
their low cost, simple structure and easymanagement. However
spatial heterogeneity of airow,air temperature and humidity
strongly vary in thesestructures. The distribution of microclimate
variablesinside the greenhouse cause non-uniform productionand
quality but generates also problems with pests anddiseases (Bot,
2001). Therefore, quantitative under-standing of this heterogeneity
can help to optimisegreenhouse production in terms of cost
efciency, cropquality and quantity.Natural ventilation directly
affects the transport of
heat and mass between the environment and the interiorof the
greenhouse so that it strongly inuences the inside
greenhouse climate. Thus ventilation performance is amajor
factor in production, inuencing both climatecontrol and yield
quality over much of the year. Itsdriving force is the combination
of buoyancy and windeffects, and their relative importance depends
on thewind speed and the insideoutside temperature
differ-ence.Several studies on natural ventilation were based
on
estimations of a global air exchange rate from tracer
gasmeasurements (Bot, 1983; Kittas et al., 1995) and energybalance
(Kindellan, 1980; Wang & Deltour, 1996;Demrati et al., 2001).
However, these methods do notallow clear mapping of airow patterns
and temperatureproles. Moreover, little information is available on
theproper design of ventilation systems, because it is verydifcult
to establish fairly identical and stable conditions
ARTICLE IN PRESS
1537-5110/$30.00 479 # 2003 Silsoe Research Institute. All
rights reservedPublished by Elsevier Ltd
-
in a eld experiment. More recently, sonic anemometrywas used to
measure airow distribution in roof vents ofa double-span greenhouse
(Boulard et al., 1997) and tomap airow patterns in an empty tunnel
greenhousewith discontinuous vent openings (Boulard et al.,
2000)and in a cultivated tunnel with continuous roof vents(Wang et
al., 1999; Haxaire, 1999).On the other hand recent progress in ow
modelling
by means of computational uid dynamics programs(CFD) allows one
to investigate and analyse airowdistribution and to predict
ventilation rates in green-houses. Actual weather conditions and
structuralspecications could be simulated and changed in theCFD
model while maintaining stable and intenticalboundary conditions.
Computational uid dynamicssimulations can be a valuable tool for
analysingthe internal airow and understanding the effectsof the
greenhouse structural characteristics with respectto ventilation.
The signicant advantage of CFD isnot only the prediction of the
ventilation rate and thusthe greenhouse performance, but the
detailed investiga-tions of airow and temperature distribution in
the
greenhouse interior and especially in the region near
thecrop.Short (1996) introduced the use of commercial CFD
models for solving naturally ventilated greenhouseventilation
problems. Mistriotis et al. (1997a, b)analysed the ventilation
process in greenhouses withouta crop. Their numerical results were
agreed with theexperimental results of Sase et al. (1984) and of
Boulardet al. (1997). Boulard et al. (1999) investigated thenatural
ventilation (thermal and wind driven) in a small-scale greenhouse
by comparing their results withexperimental data from the same
greenhouse. Alwaysin greenhouses without a crop, Reichrath and
Davies(2002) simulated a two-dimensional full size
commercialmulti-span glasshouse comprising 60 spans and
theycompared their numerical results with the experimentaldata of
Hoxey and Moran (1991). Woodruff (1997),Kacira et al. (1998), and
Lee and Short (2000, 2001)studied various naturally and
mechanically ventilatedgreenhouses types by using a CFD numerical
model.They mainly investigated the effects of weather condi-tions,
greenhouse structural specications, internal and
ARTICLE IN PRESS
Notation
A vent opening area, m2
C tracer gas concentration, ppmCD drag coefcientCm model
constant of the ke modelCp specic heat of air, J kg
1 8C1
Cm constant tting parameterC1e model constant of the ke modelC2e
model constant of the ke modelDi saturation vapour pressure decit,
Pad characteristic leaf length, mF tracer gas uxG ventilation rate,
m3 s1
h reference height, mILA leaf area indexK von Karman constantk
turbulent kinetic energy, m2 s2
L leaf area density, m2m3
N air changes per hourQsen sensible heat exchange, Wm
2
Qlat latent heat exchange, Wm2
Rgi internal global solar radiation, Wm2
ra aerodynamic resistance, sm1
rs stomatal resistance, sm1
S surface area, m2
SF source termT air temperature, 8CTi inside air temperature,
8C
Tc crop temperature, 8CTo outside air temperature, 8Ct time, sU,
V, W components of velocity vectorUinl inlet velocity, m s
1
u air velocity, m s1
uh reference velocity, m s1
u* friction velocity, m s1
Vg greenhouse volume, m3
wa air absolute humidity, kg kg1
wc crop absolute humidity, kg kg1
Y non-linear momentum loss coefcientz height, mzo friction
length, ma crop permeablility, m2
G diffusion coefcientDT temperature difference, 8CDT mean
temperature difference, 8Ce dissipation rate of turbulent kinetic
energy,
m2 s3
l latent heat of water vaporisation, J kg1
m dynamic viscosity kgm1 s1
r air density, kgm3
s standard deviationsk turbulent Prandtl number for
turbulent
kinetic energyF concentration of transported quantity
T. BARTZANAS ET AL.480
-
external shading screens, number of spans and presenceof plants
and benches on the air exchange rate.Bartzanas et al. (2002)
investigated the three-dimen-sional air ux and temperature effects
in a tunnelgreenhouse equipped with an insect screen with
specialattention devoted to the inuence of the wind
direction.However, all these simulations basically suffer from
a
lack of realism due to the lack of modelling of the heatand
water exchanges between the crop cover and itsenvironment. More
recently, Boulard and Wang (2002)used a commercial CFD code
(CFD2000) and incorpo-rated solar radiation and transpiration
models into theoriginal code together with the modelling of the
dynamiceffect of the crop by the equivalent macro-porousmedium.
They validated their numerical results withexperimental data and
found a good agreement betweenmeasured and computed for both
climatic and croptranspiration elds. With the same approach
andboundary conditions measured in the experimental study,Fatnassi
et al. (2001) investigate the three-dimensional airux, temperature
and humidity effects in a very large-scale greenhouse (1/2 h)
equipped with an insect screen.Based on the same realistic approach
of CFD
simulation which considers both the dynamic, thermaland
transpiration exchange between the greenhousecrop and its
environment, the aim of present paper is,after validating the code
and the method againstexperimental data, to determine the effects
of differentvent arrangements on windward ventilation of a
tunnelgreenhouse with mature tomato crop.
2. Materials and methods
2.1. Experimental greenhouse
The measurements were carried out in an experimentalNS oriented
tunnel greenhouse located at the Universityof Thessaly near Volos,
(Latitude 398440, Longitude228790) on the coastal area of Eastern
Greece. Thegeometrical characteristics of the greenhouse were
asfollows: eaves height of 24m; ridge height of 41m; totalwidth of
8m; and total length of 20m. The greenhousewas covered with a
polyethylene sheet and was equippedwith two continuous side
openings (roll-up type) located06m from the ground with a maximum
opening of09m. The greenhouse was cultivated with a tomato
crop,which reached a height 15m during the experiments.
2.2. Measurements
Two different types of measurements were conductedin order to
validate the simulations: (a) measurements of
the three components of air velocity; and (b) measure-ments of
the ventilation rate of the greenhouse.
2.2.1. Air velocity measurementsRapid uctuations in air velocity
were measured by
means of one three-dimensional (3-D) sonic anem-ometer
(omnidirectional, R3, research ultrasonic anem-ometer, Gill
R&D). The three components of windvelocity (U, V, W) were
measured at six positions in themiddle of the greenhouse and along
its eastwest widthat 1, 2, 3, 5, 6 and 7m from the east side. The
height ofthe measurements was at 11m above ground, whichcoincided
with the middle height of the full opening. Themanufacturers
calibration was accepted for U, V, Wmeasurements. Sampling
frequency was 5Hz. The timeduration of each measurement record was
about 5min.Dry and wet bulb temperatures were also recorded atthe
same points using an aspirated psychrometer. Thepsychrometer was
placed within 02m of the samplingvolume of the sonic anemometer in
order to minimisethe ow distortion.A weather station tower was
installed outside the
greenhouse to measure the local climate such as dry andwet bulb
air temperatures, wind speed (AN1-UM-3,Delta-T devices Cambridge,
UK), wind direction (AN1-UM-3, Delta-T devices Cambridge, UK) and
solarradiation (CM-6, Kipp and Zonen, Delft, Netherlands).Outside
dry and wet air temperatures were measured at11m above ground, air
speed and direction as well assolar radiation at 1m above the top
of the greenhouse.The above variables were also measured each
secondand averaged over the length of each record.As the CFD
simulations describe only steady-state
conditions, so data for the validation were collectedwhen
weather conditions such as wind speed, winddirection and solar
radiation were stable, mainlybetween 1 h before and after solar
noon. Table 1presents the mean value (in a time interval of 5min)
ofclimate conditions during the measurements with thesonic
anemometer.
2.2.2. Tracer gas measurementsThe decay rate method, using N2O
as tracer gas, was
used to deduce the ventilation rate of the greenhouse. Adetailed
description of the procedure for these measure-ments as well as the
corresponding equations can befound in Boulard and Draoui (1995).
The tracer gas wasinjected into the greenhouse, its concentration
washomogenised using the fans of the heating system;
thenventilation openings were opened at a known height
andsimultaneously wind velocity, wind direction and
gasconcentration were recorded during the tests. Airsamples were
continuously taken at six points in thegreenhouse, by means of six
equally distributed plastic
ARTICLE IN PRESSVENTILATION OF A TUNNEL GREENHOUSE 481
-
pipes of the same length, located at a height ofapproximately
18m from the ground. The air fromthe six positions was then mixed
and pumped to aninfrared gas analyser (model 7000, ADC gas
analyser,analysis up to 200 vpm, accuracy at 5 vpm). Theduration of
each experiment varied depending onenvironmental conditions and on
ventilation opening,ranging between 5 and 20min. During the
experi-ments wind speeds varied between 15 and 5m s1 andventilation
opening from 0 to 090m.
2.3. Numerical model
The CFD method allows the explicit calculation ofthe average
velocity vector eld of a ow by numericallysolving the corresponding
transport equations. Thethree-dimensional conservation equations
describingthe transport phenomena for steady ows in freeconvection
are of the general form:
@UF@x
@VF@y
@WF
@z Gr2F SF 1
In Eqn (1), F represents the concentration of thetransport
quantity in a dimensionless form, namely thethree momentum
conservation equations (the NavierStokes equations) and the scalars
mass and energyconservation equations; U, V and W are the
componentsof velocity vector; G is the diffusion coefcient; and SF
isthe source term. The governing equations are discretisedfollowing
the procedure described by Patankar (1980).This consists of
integrating the governing equations overa control volume.The
commercially available CFD code Fluent v.5.3.18
(Fluent, 1998) was used for this study. As the prevailingwind
direction was parallel to the greenhouse during theexperiments, a
3-D model was rst built in order tocompare the numerical results
with the experimentaldata. All the other simulations, used for case
studies,were two-dimensional since the selected wind direction
for the simulations was perpendicular to the axis of
thegreenhouse.To achieve an accurate result, second-order
upwind
discretisation schemes were used for momentum andturbulence
equations. A semi-implicit method forpressure linked equations
algorithm was used for thecoupling between pressure and velocity.
The conver-gence criterion for all variables was 1 106.
2.3.1. Mesh and boundary conditionsFor the geometry, a control
volume was selected
representing a large domain including the greenhouse.The grid
structure was an unstructured, quadrilateralmesh with a higher
density in critical portions of theow subject to strong gradients.
Body-tted coordinateswere also applied to exactly conform the grid
to thecontours of the boundary conditions. After several trieswith
different densities, the calculations were based on a48 by 20 by 80
grid. This results from an empiricalcompromise between a dense
grid, associated with along computational time, and a less dense
one,associated with a marked deterioration of the
simulatedresults.The boundary conditions prescribed a null
pressure
gradient in the air, at the limits of the computationaldomain,
and wall-type boundary conditions along theoor and the roof whereas
the side walls were treated asadiabatic (Table 2). The Boussinesq
model (Launder &Spalding, 1974; Fluent, 1998) was activated for
thebuoyancy effect in the computational domain.As shown by the
measurements of turbulent airows
and microclimate patterns in a greenhouse tunnel(Boulard et al.,
2000), the airows were highly turbulent.Consequently, turbulent
models must be introduced inthe Reynolds equations written to
separate the meanow from its uctuating components. The standard
kemodel (Launder & Spalding, 1974) assuming isotropicturbulence
was adopted to describe turbulent transport.This choice is a good
compromise for a realisticdescription of turbulence and
computational efciency(Jones & Whittle, 1992). The complete set
of the
ARTICLE IN PRESS
Table 1Mean values (in a 5min interval) of climate conditions
during the measurements with the sonic anemometer
Measurement Positions Temperature, 8C Solar radiation, W m2 Wind
Speed, m s1 Wind direction* deg
Inside air Outside air
1 2900 2820 789 370 302 3020 2980 843 206 403 3090 2960 867 330
454 3120 2910 874 330 505 3130 2820 843 260 356 3150 2850 835 27
30
*0 denotes wind direction parallel to the greenhouses axis.
T. BARTZANAS ET AL.482
-
equations of the kemodel can be found in Mohammadiand Pironneau
(1994) and their commonly used set ofparameters (empirically
determined) are (Cm 009,C1e 144, C2e 1,91, sk 1) (Fluent, 1998).The
wind direction was normal to the ridge for the 3-
D model and perpendicular to the ridge for the 2-Dsimulations. A
reference velocity was chosen to be3m s1 at a reference height
(10m). At the inlet of thecomputational domain a wind prole was
imposed. Inletvelocity was dened as:
Uinl u
Kln
z zozo
2
with
u Kuh
lnh zo=zo3
with Uinl the inlet velocity in m s1, u* the friction
velocity in m s1, K the von Karman constant(K 042), z the height
in m, zo the friction length inm, Uh the reference velocity in m
s
1 and h the referenceheight in m. The friction length zo was
chosen as 001mcorresponding to a ploughed up eld. The
distributionof turbulent kinetic energy, k in m2 s2 and of
theturbulent dissipation rate, e in m2 s3 in the incomingwind prole
are described by the relationships:
k u2Cm
p 4
e u3
Kz z05
where Cm is a constant tting parameter.
2.3.2. The equivalent porous medium approachThe crop was
simulated using the equivalent porous
medium approach by the addition of a momentumsource term, due to
the drag effect of the crop, to thestandard uid ow equations. The
drag force perunit volume of the crop can be expressed as
(Wilson,1985):
SF rLCDu2 6
where: u is the air velocity in m s1; L the leaf areadensity in
m2m3; r the air density in kgm3, and CDthe drag coefcient. The
source term SF is composed oftwo parts, a viscous loss term
(Darcy), and an inertialloses term. In the case of a simple
homogenous porousmedia the source term was described as:
SF ma
u Y1
2rjuju 7
where: a is the permeability of the porous medium (crop)in m2; Y
the non-linear momentum loss coefcient;and m the dynamic viscosity
in kgm1 s1. In the caseof crop, for reasons of simplicity, it was
assumedthat pressure forces contributed the major portionof total
canopy drag (Thom, 1971). Using wind tunnelfacilities, for a mature
greenhouse tomato crop with aleaf area index ILA of 4, Haxaire
(1999) has evaluatedthe total drag of the canopy (CD 032). Usingthe
relationships in Eqns (6) and (7) (Boulard &Wang, 2002) the
appropriate values for permeabilityand non-linear momentum loss
coefcient can bededuced.The exchange of heat and water vapour
between crop
and air was considered through the heat and massbalance of crop
with the air. The sensible heat Qsen inWm2, from the crop was
calculated using the followingequation:
Qsen 2ILArCpTc Ti
ra8
where: ILA is the leaf area index; Cp is the specic heat ofair
at constant pressure in J kg1K1; Tc and Ti are thecrop and air
temperatures in 8C; and ra is theaerodynamic resistance of the crop
in sm1. Notethat performing leaf heat and water vapour balancesmake
it necessary to introduce a new phenomenologicalvariable
Tc.Following Campbell (1977), if the interior air speed
u501m s1, then:
ra 840d
jTc Tij
0259
else:
ra 220d02
u08
10
ARTICLE IN PRESS
Table 2Boundary values used in the simulations
Parameters Numerical value
Wind direction3-D model Parallel to the ridge2-D model
Perpendicular to the
ridgeTemperatureOf the cover, 8C 3200Of inside ground, 8C 4500Of
outside ground, 8C 3000Of outside air, 8C 2800
Inlet airVelocity, m s1 300Relative humidity, % 4000Density,
kgm3 122Gravitational acceleration, m s2 981Specic heat, J kg1 8C
100400Thermal conductivity, Wm2 8C1 00263
Plant canopyPressure drop coefcient 0395Inertial loss factor
020
VENTILATION OF A TUNNEL GREENHOUSE 483
-
where: d is the characteristic length of the leaf in m; andu is
the local air speed in the same mesh at the givenlocation in m
s1.The latent heat exchange between crop and air Qlat in
Wm2 was calculated according to the followingequation:
Qlat ILArlwc wara rs
11
where: wc and wa are the absolute humidity of crop andair in kg
kg1 also in the same mesh; l is the latent heatof water
vaporisation in J kg1; and rs is the cropstomatal resistance in
sm1, which was calculatedaccording to the relationship found by
Boulard et al.(1991) for greenhouse tomato leaves:
rs 200 11
exp0005Rgi 50
1 011 exp 034Di
100 10
12
where: Rgi is the internal global solar radiation inWm2; and Di
is the saturation vapour pressure decitin Pa. In this relationship,
only the radiation and airhumidity dependence of the stomatal
conductance areconsidered. The rst factor on the right-hand side
ofEqn (12) shows that stomatal resistance decreaseswith the solar
radiation Rgi. When Rgi is greater than100Wm2, rs is maintained at
a constant value of200 sm1. The second factor on the right-hand
sideof Eqn (12) shows that leaf stomatal resistanceincreases with
the air drying over 10 Pa of saturationdecit.The numerical model
was customised in C++ in
order to perform the balance described by Eqns (8)(12),based on
the local, computed air speed and climaticconditions within each
mesh of the porous medium(crop cover).
2.3.3. Simulation of the ventilations ratesThe ventilation rate
of the greenhouse was numerical
calculated by means of the continuity equation for everyinternal
cell for a virtual tracer gas (N2O):
dC
dt FF dS 13
where: C is the concentration of the tracer gas in a cell inppm;
t is the time in s; and F the tracer gas ux. On theright-hand side
of Eqn (13) is given the surface integralof the tracer gas ux F
through the surface area of thecells in m2. The average values of
the air velocity can beused for calculating F.In fact, the results
of the virtual tracer gas measure-
ments were obtained in two major steps. First aconverged
solution under steady-state conditions isobtained. Then, the ow is
considered unsteady and
the species model is used to inject the virtual tracer
gas.Initially all the cells in the greenhouse have a xed tracergas
concentration equal to unity and all the externalcells equal to
zero. In this way, if a time step, dt, isselected, the continuity
equation can be solved as adifference equation with respect to
time. The tracer gasconcentration decreases in the greenhouse at a
ratedepending on the local value of the air velocity. Then,the
average tracer gas concentration %CC of that volume iscalculated as
a function of time. This function usuallyexhibits an exponential
decay. For this reason, it is ttedby an exponential of the form
%CCt %CC0eNt 14
The identied exponent N value is the decay rate ofthis function;
therefore it describes the ventilation rateof the studied volume
(air changes per hour).
2.4. Vent configurations used for simulation
The following commonly found vent congurationshave been used for
the simulation of the inuence of thevent design on windward
ventilation of a tunnel typegreenhouse (Fig. 1). In order to
characterise the openingof the vent, the chord of the opening was
dened as thedistance between the free end of the vent to its
restplace on the greenhouse when the vent is closed and thewindow
aperture area as the product of the length ofthe vent by the chord
of the opening. In all cases thedistance between the articulation
of the vent and its restplace on the greenhouse structure when the
vent isclosed measures 09m.
2.4.1. Configuration (a): side openings only(roll-up type)
The greenhouse is equipped with two continuous roll-up type
openings located at 06m above ground with anopening height of 09m.
This conguration leads to atotal opening area of 36m2.
2.4.2. Configuration (b): side only openings(pivoting door
type)
The greenhouse is equipped with two continuouspivoting door type
side openings. The base of thewindow is at 06m above ground and the
height of thewindow is 09m. The aperture angle is 458, which
leadsto an opening area of 275m2.
2.4.3. Configuration (c): roof opening onlyThe greenhouse is
equipped with a pivoting type roof
opening. The free end of the opening is at the ridge ofthe
greenhouse and the articulation at 09m leewardfrom the ridge. When
opened, the opening faces the
ARTICLE IN PRESST. BARTZANAS ET AL.484
-
wind and its chord is 09m. This conguration leads toan opening
area of 18m2.
2.4.4. Configuration (d): combined roof and sideopenings
(roll-up type)
This conguration combines the roll-up side openingsof
conguration (a) with the roof opening of congura-tion (c). This
conguration leads to a total opening areaof 54m2.
3. Results
3.1. Numerical model validation
Figure 2 shows both the experimentally and thenumerically
obtained average transverse horizontalcomponent of the normalised
air velocity along thegreenhouse width at a height of 11m at the
middle ofthe greenhouse. The normalised air velocity wasobtained by
the ratio of the interior air velocity to themean external wind
speed. For roll-up openings (type(a) of Fig. 1) and a wind
direction parallel to thegreenhouse axis, both computed and
simulated valuesshow that air speed has relative high values
nearthe openings and reduced values near the centre of
thegreenhouse. However, when the wind is parallel to thegreenhouse
axis both openings acted simultaneously asinlets and outlets. Air
then entered the greenhousethrough the leeward section of the
openings and exitedthrough the windward part. A similar airow
patternwas measured in a greenhouse with a continuous roof(Boulard
et al., 1997) and this phenomenon is alsocomparable to the side
wall effect deduced from tracer
gas measurements (Fernandez & Bailey, 1992). Thewindward
gable end induced a positive static pressureeld whose relative
contribution to the whole ventilationrate is inverse to the size of
the greenhouse in thedirection of the wind.
Figure 3 presents the simulated and experimentalventilation
rates (air changes per hour) versus theproduct of ventilation
opening area and outside windspeed. Good agreement was also found
in this case andthe differences between simulated and measured
ventila-tion rate varied only by 1215%. In all cases the
valuededuced from CFD simulations was larger than thevalue deduced
from the experimental measurements.The explanations for this are
twofold: (a) The experi-mental wind velocity represents an average
value overthe measurement period and therefore neglects
theturbulent part, while the estimated value by the CFDmodel
includes the turbulent part even if the used kemodel is a rough
approximation of the reality; and (b)The homogenisation of the
tracer gas, in the experiment,is not perfect although fans were
used for this purpose,whereas in the CFD model a perfect
homogenisationis assumed. According to Boulard et al. (1996)
thisnon-perfect homogenisation leads to the fact thattracer gas
techniques allow the determination of theeffective airow, which can
be lower than the realairow.
Figure 4 presents the computed and measured airtemperature
difference Ti2To in a horizontal plane atthe middle of the
greenhouse. For an external windspeed parallel to the axis of the
greenhouse, themeasured and CFD-computed results indicated thatthe
inside air temperature gradually increases from theside wall to the
middle of the greenhouse where its valuestarts to decrease.
ARTICLE IN PRESS
00.020.040.060.08
0.10.120.140.160.18
0 3 7Greenhouse width, m
Nor
mal
ised
velo
city
21 654 8
Fig. 2. Experimentally (& & &) and numerically
obtained(& & &) average transverse horizontal component
of the airvelocity along the greenhouse width at a height of 11
from the
greenhouse ground normalised by the outside wind speed
Fig. 1. Geometries of the four different configurations for
theventilation efficiency study:(a) roll-up type openings;
(b)pivoting door type openings; (c) roof only openings; and (d)
side and roof openings.
VENTILATION OF A TUNNEL GREENHOUSE 485
-
3.2. Airflow patterns and temperature distributionfor different
ventilator configurations
Considering that the model was globally validated asillustrated
by the good t between experimental andnumerical obtained values, we
have used the CFD forinvestigating the inuence of the arrangement
of variousvent openings on airow and temperature distributionsin
the greenhouse tunnel.
3.2.1. Configuration (a): side openings only(roll-up type)
This rst conguration is similar to the greenhouseused for the
model validation when the vents are fullyopened. Only the wind
direction (now perpendicular tothe ridge) was changed. Figure 5
presents the computedcontours of the air velocities obtained for
this case. Itwas characterised by a strong air current near
theground and a recirculation loop with slower speedsituated near
the roof and owing counter current with
respect to the wind outside. This recirculation loopimproves the
air mixing but most of the air leaves thegreenhouse volume without
a good homogenisation.Above the height of the ventilator (i.e.,
1.5m) the airvelocities were strongly reduced. The internal ow
isdifferent from that which was observed during thevalidation
study. These results indicate that the winddirection clearly
inuences the air velocity inside thegreenhouse and thus its
ventilation rates. Figure 6presents the distribution of temperature
for this case.It is clear that temperature distribution follows the
airprole and in regions with small air velocities (especiallynear
the corners of the greenhouse) the air temperaturewas 318C compared
with a outside temperature of 288C.In the region covered by the
crop the air temperature issimilar to the outside (285298C) due to
the strong airmovement in this region.
3.2.2. Configuration (b): side only openings(pivoting door
type)
To prevent the incoming jet (through the windwardopening) from
impinging directly on the crop, the typeof the openings were
changed from roll-up (congura-tion a) to side-open. Airow patterns
show that theincoming air through the windward side ventilator
tendsto move up immediately by the inuence of inclinedventilator ap
and mainly follows the inner surface ofthe roof. In the space to be
occupied with a crop, thereverse ow due to secondary circulation
results inthe signicant decrease of the air velocity (Fig. 7).The
distribution of temperature for this case is presentedin Fig. 8.
Due to low air velocities near the greenhouseoor there are high air
temperatures in this region. Thetemperature elevation in the
corners of the greenhousewas 48C higher than outside air
temperature, and in themiddle of the greenhouse the air temperature
was 18Chigher than the temperature of outside air.
ARTICLE IN PRESS
00.20.40.60.8
11.21.41.61.8
2
0 2 5 7Greenhouse width, m
Tem
pera
ture
diff
eren
ce (T
i - T o
), C
431 6 8
Fig. 4. Experimentally (& & &) and numerically
obtained(}}}) air temperature difference (Ti To) along
thegreenhouse width at a height of 11 m from the greenhouse
ground
Fig. 5. Computed contours of the air velocities of a
tunnelgreenhouse with side (roll up type) openings only
(configura-
tion a)
0
5
10
15
20
25
0 20 40 60 80 100 120Opening surface x wind velocity (Su), m3
s-1
Air
chan
ges p
er h
our
Fig. 3. Experimentally (& & &) and numerically
obtained(& & &) air changes per hour; S, surface area
(m2), u air
velocity (m s1)
T. BARTZANAS ET AL.486
-
3.2.3. Configuration (c): roof opening onlyThe efciency of only
roof openings was examined
with this conguration. The incoming air from the roofopening
guided by the greenhouse walls follows a semi-spiral trajectory and
leaves the greenhouse by followingthe internal surface of the walls
and the roof. Still airconditions prevails at the centre of the
greenhouse(Fig. 9). As a result of the low values of air
velocitieswith this conguration the air temperature inside
thegreenhouse reached very high values. Air temperature atthe
leeward wall of the greenhouse was 68C higher thanoutside. Due to a
better air mixing with this congura-tion, caused by the air
circulation cell developed at thegreenhouse interior, air
temperature was uniformlydistributed in most of the greenhouse but
it was 28Chigher than outside air temperature (Fig. 10).
3.2.4. Configuration (d): combined roof and sideopenings
(roll-up type)
With this conguration the inuence the combinationof sides and
roof openings was tested. Qualitatively theairow was similar to
conguration (a) because littleexchange was observed through the
roof opening asthe external ow passed directly through the
sideopenings. Air velocities were slightly higher in the inletand
outlet of the greenhouse and almost the same in the
rest of greenhouse interior compared with the airvelocities of
conguration (a). Temperature distributionfollowed the airow pattern
with warm sections neargreenhouse oor (38C warmer than outside air)
andsections with similar to outside air temperatures in themiddle
of the greenhouse (Fig. 11).
ARTICLE IN PRESS
Fig. 6. Temperature distribution of a tunnel greenhouse withside
(roll-up type) openings only (configuration a)
Fig. 7. Computed contours of the air velocities of a
tunnelgreenhouse with side (sliding door type) openings only
(config-
uration b)
Fig. 8. Temperature distribution of a tunnel greenhouse withside
(sliding door type) openings only (configuration b)
Fig. 9. Computed contours of the air velocities of a
tunnelgreenhouse with roof openings only (configuration c)
Fig. 10. Temperature distribution of a tunnel greenhouse
withroof openings only (configuration c)
VENTILATION OF A TUNNEL GREENHOUSE 487
-
4. Discussion
The ventilation of a greenhouse is the exchange of airbetween
the inside and outside in order to: (1) dissipatethe surplus heat;
(2) enhance the exchange of carbondioxide and oxygen; and (3) to
maintain acceptablehumidity levels. For the four tested
congurations,conguration (d) (combined roof and sides
openings)achieves the highest ventilation rate, and conguration(c)
(roof opening only) the lowest. As it was stated fromthe
introduction, the largest ventilation rates are not, apriori, the
best indicator for the ventilation perfor-mances of a greenhouse.
The air velocity near the cropand the temperature difference that a
given type canachieve must also be taken into account since there
areimportant factors inuencing the uniform growth ofcrop. Spatial
heterogeneity of air velocity and climateinside greenhouse
interfere with plant activity andinuence largely crop behaviour
through their effectson crop gas exchanges, particularly
transpiration andphotosynthesis. For instance increasing air
velocityinside the greenhouse increases convective heat transferand
hence reduces the leafair temperature differences.Furthermore, air
velocity might be expected to increasephotosynthesis because of the
reduced boundary layerresistance to the transport of carbon
dioxide, unless it isnot limiting. Waggoner et al. (1963) have
observed withsugar cane plants that an airow rate of 05m s1 at aCO2
concentration of 200 vpm, gave growth equivalentto 300 vpm. If the
increased air speed raises transpira-tion to such extent that water
stress and hence stomatalclosure occurs, then photosynthesis will
be reduced as aconsequence. High air velocities (>1m s1) in the
cropcover should also be avoided since they lead toreduction in
leaf area and dry matter accumulation(Kalma & Kuiper, 1966).
Figure 12 presents the averagetransverse horizontal component of
the air velocity forthe four different congurations along the
greenhouse
width at the middle of the greenhouse normalised by theoutside
wind speed and Fig. 13 the corresponding cropaerodynamic resistance
which was calculated afterCampbell (1977). For the congurations (a)
and (d)the normalised air velocity at the middle of thegreenhouse
has relative high values (0609m s1)resulted in low values of crop
aerodynamic resistance(7090 sm1). Conversely with conguration (b)
and (c),the normalised air velocity is relatively low (00502m s1)
resulting in high values of crop aerodynamicresistance (600900
sm1). For the same outside climateconditions a reduction of 90% of
the crop aerodynamicresistance was observed which will lead to an
increase ofthe convective heat transfer to the same extend. For
themass transfer the effects are not so obvious becausestomatal
resistance, which depends on other microcli-mate factors, plays an
important role too.
ARTICLE IN PRESS
Fig. 11. Temperature distribution of a tunnel greenhouse
withcombined roof and side (roll- up type) opening (configuration
d)
00.10.20.30.40.50.60.70.80.9
1
0 2 3 6Greenhouse width, m
Nor
mal
ised
velo
city
1 54 87
Fig. 12. Average transverse horizontal component of the
airvelocity for the four different tested configurations along
thegreenhouse width at the middle of the greenhouse normalised
bythe outside wind speed: configuration (a) (}}}), configura-tion
(b) (& & &), configuration (c) (m m m) and
configuration (d)4 (- - - -)
0
200
400
600
800
1000
1200
0 2 4 5 6 7 8Greenhouse width, m
Crop
aer
odyn
amic
re
siste
nce,
sm-1
1 3
Fig. 13. Variation of crop aerodynamic resistance for the
fourdifferent tested configurations along the greenhouse width at
themiddle of the crop cover: configuration (a) (}}}),
config-uration (b) (& & &), configuration (c) (m m m)
and
configuration (d) (- - - -)
T. BARTZANAS ET AL.488
-
To better exploit our CFD analysis, the results wereexpressed
with respect to the effects of the differentcongurations on (i) air
exchange and (ii) temperaturedifference between inside and outside.
Likewise for eachventilation type, the efciency of the ventilation
wasconsidered by reducing the global ow rate Q or thetemperature
difference DT between inside and outside bythe vent opening area A.
The homogeneity of thetemperature distribution has been evaluated
by reducingthe standard deviation of DT (s(DT)) by its means
valueDT. A summary of the main results for the fourcongurations is
presented in Table 3.It is rst clear that conguration (c), with
only roof
opening gives the worst results and presents the
lowestventilation efciency by elementary surface of opening.On
contrast, conguration (d) with both roof and sideventilation
presents the best ventilation efciency. Sideopenings only have very
similar results.Referring to the efciency of ventilation on the
temperature difference between inside and outside, itcan be
stated that an even larger difference betweentypes (d) and (a), the
combination of side and roofventilations again giving the best
results, with anefciency of about 55 times more than that for
theconguration with only a roof opening and 15 timesmore than that
for conguration (a). For the sideopenings congurations, the type
(a), with roll-upopenings is about two times more efcient than the
type(b) with pivoting vent openings. For conguration (d),the high
ventilation efciency is not only due to thehighest total vent
opening surface, but mainly to thehigh efciency of the combination
of both openingtypes. This is a very important point because a
large partof the greenhouse cost is due to the cost of the
openingsand the determination of a good opening efciency
isprimordial.Considering the homogeneity of the temperature
eld,
conguration (c) gives the best results and conguration(d) the
worst. The highest is the efciency on the cooling,the lowest is the
homogeneity of the temperature eldand conversely.
5. Conclusions
The inuence of vent arrangement on windwardventilation of a
tunnel greenhouse was numericallyinvestigated using commercial uid
dynamics code. Thenumerical model was rst validated against
experimen-tal data. Four different congurations of ventilatorswere
investigated resulting in different ventilation ratesand different
airow and temperatures patterns. Theseresults indicate that the
highest ventilation rates are notalways the best criterion for
evaluating the performanceof different ventilation systems in
greenhouses. Suchcriteria are: the air velocities in the region
covered by thecrop, the corresponding air aerodynamic resistances
aswell as the efciencies of ventilation on ow rate and theair
temperature differences between inside and outside.For the
congurations studied in this work, the abovecriteria show that the
best solution is the combination ofroof and side openings. Whenever
there are only sidewindows available for ventilation, roll-up
openings aremore efcient than pivoting window openings.The
numerical model simulates reasonably well the
ventilation performance of greenhouses. For a givengreenhouse
type the CFD model can be used as adesign tool to propose the
ventilation openings design(typessizeposition) in order to achieve
a well-venti-lated greenhouse and uniform climate conditions in
thecrop cover. However, one must keep in mind that theresults
presented in this paper concern only the specicexamined cases. With
a different wind direction or adifferent greenhouse type, results
could be different.
References
Bartzanas T; Boulard T; Kittas C (2002). Numerical simulationof
airow and temperature distribution in a tunnel green-house equipped
with a insect-proof screen on the openings.Computers and
Electronics in Agriculture, 34, 207221
Bot G P A (1983). Greenhouse climate: from physicalprocesses to
a dynamic model. PhD Dissertation, Agricul-tural University of
Wageningen, Netherlands, 240pp
ARTICLE IN PRESS
Table 3Recapitulation of the main results of the four studied
cases
Conguration Air exchange density (G/A), m s1 Temperature
difference
Specific rise (DT/A) 8C/m2 Homogeneity (s(DT)DT)
1 015 0022 0712 013 0040 0563 007 0090 0334 016 0016 073
G, ventilation rate in m3 s1; A, opening surface area in m2; DT
temperature difference between inside greenhouse air and outsideair
in 8C; DT , mean value of temperature difference between inside
greenhouse air and outside air in 8C, s(DT), standarddeviation of
temperature difference between inside greenhouse air and outside
air in 8C.
VENTILATION OF A TUNNEL GREENHOUSE 489
-
Bot G P A (2001). Developments in indoor sustainable
plantproduction with emphasis on energy saving. Computers
andElectronics in Agriculture, 30, 151165
Boulard T; Baille A; Mermier M; Villette F (1991). Mesures
etmodelisation de la resistance stomatique foliaire et de
latranspiration dun couvert de tomates de serre [Measure-ment and
modelling of stomatal resistance and transpirationin a greenhouse
tomato canopy]. Agronomie, 11(4), 259274
Boulard T; Draoui B (1995). Natural ventilation of a green-house
with continuous roof vents: measurements and dataanalysis. Journal
of Agricultural Engineering Research, 61,2736
Boulard T; Haxaire R; Lamrani M A; Roy J C; Jaffrin A(1999).
Characterization and modeling of the air uxesinduced by natural
ventilation in a greenhouse. Journal ofAgricultural Engineering
Research, 74, 135144
Boulard T; Meneses J F; Mermier M; Papadakis G (1996).
Themechanisms involved in the natural ventilation of green-houses.
Agricultural and Forest Meteorology, 79, 6177
Boulard T; Papadakis G; Kittas C; Mermier M (1997). Air owand
associated sensible heat exchanges in a naturallyventilated
greenhouse. Agriculture and Forest Meteorology,88, 111119
Boulard T; Wang S; Haxaire R (2000). Mean and turbulentair ows
and microclimate patterns in an empty green-house tunnel.
Agricultural and Forest Meteorology, 100,169181
Boulard T; Wang S (2002). Experimental and numerical studieson
the heterogeneity of crop transpiration in a plastictunnel.
Computers and Electronics in Agriculture, 34,173190
Campbell G S (1977). An Introduction to EnvironmentalBiophysics,
p 159. Springer, New York.
Demrati H; Boulard T; Bekkaoui A; Bouirden L (2001).
Naturalventilation and climatic performance of a large-scale
bananagreenhouse. Journal of Agriculture Engineering
Research,80(3), 261271
Fatnassi H; Boulard T; Bouirden L (2001). Simulation of airux
and temperature patterns in a large scale greenhouseequipped with
insect-proof nets. Acta Horticulturae, 578,329337
Fernandez J E, Bailey BJ (1992). Measurements and predictionof
greenhouse ventilation rates. Agricultural and ForestMeteorology,
58, 229245
Fluent, 1998. FLUENT, v.5. Fluent Europe Ltd, Shefeld,UK
Haxaire R (1999). Characterization et modelisation
desecoulements dair dans une serre. [Characterizationand modeling
of the air uxes in a greenhouse] PhD Thesis,149p
Hoxey R P; Moran P (1991). Full scale wind pressure and
loadexperiments}Multispan 167 111m glasshouse (Venlo).Divisional
Note 1594, AFRC Institute of EngineeringResearch, Wrest Park,
Sisloe, Bedford, January
Jones P J; Whittle G E (1992). Computational uid Dynamicsfor
building airow prediction. Buildings and Environment,27, 321338
Kacira M; Short T H; Stowell R (1998). A CFD evaluation
ofnaturally ventilated multi-span, sawtooth
greenhouses.Transactions of the ASAE, 41(3), 833836
Kalma J D; Kuiper F (1966). Transpiration and growth
ofPhaseoulus vulgaris L. as affected by wind speed. Mededelin-gen
Landbouwhogeschool, Wageningen, 66(8), 19
Kindellan M (1980). Dynamic modeling of greenhouseenvironment.
Transactions of the ASAE, 23, 12321239
Kittas C; Draoui B; Boulard T (1995). Quantication of
theventilation of a greenhouse with a roof opening. Agricultur-al
and Forest Meteorology, 77, 95111
Launder B E; Spalding D B (1974). The numerical computationof
turbulent ows. Computer Methods in Applied Me-chanics and
Engineering, 3, 269289
Lee I B; Short T H (2000). Two-dimensional numericalsimulation
of natural ventilation in a multi-span greenhouse.Transactions of
the ASAE, 43(3), 745753.
Lee I B; Short T H (2001). Verication of computational
uiddynamic temperature simulations in a full-scale
naturallyventilated greenhouse. Transactions of the ASAE,
44(1),119127
Mistriotis A; Arcidianoco C; Picuno P; Bot G P A;
ScarasciaMugnozza G (1997b). Computational analysis of the
naturalventilation in greenhouses at low wind speed.
Agriculturaland Forest Meteorology, 88, 121135
Mistriotis A; Bot G P A; Picuno P; Scarascia Mugnozza G(1997a).
Analysis of the efciency of greenhouse ventilationwith
computational uid dynamics. Agricultural and ForestMeteorology, 85,
317328
Mohammadi B; Pironneau O (1994). Analysis of the
kepsilonTurbulence Model. Research in Applied Mathematics.Wiley,
New York, Masson, Paris
Patankar S V (1980). Numerical Heat Transfer.
Hemisphere,Washington
Reichrath S; Davies T W (2002). Computational uiddynamics
simulations and validation of the pressuredistribution on the roof
of a commercial mutli - spanVenlo-type glasshouse. Journal of Wind
Engineering andIndustrial Aerodynamics, 90(3), 139149
Sase S; Takakura T; Nara M (1984). Wind tunnel testing onairow
and temperature distribution of a naturally venti-lated greenhouse.
Acta Horticulturae, 148, 329336
Short T H (1996). Selecting the greenhouse structure your
cropneeds. Grower Talks, Summer Issue, July, 8-9
Thom A S (1971). Momentum absorption by vegetation.Quarterly
Journal of Royal Meteorology Society, 97,414428
Waggoner P E; Moss D N; Hesketh J D (1963). Radiation inthe
plant environment and photosynthesis. AgronomieJournal, 55,
3639.
Wang S; Boulard T; Haxaire R (1999). Air speed proles in
anaturally}ventilated greenhouse with a tomato crop.Agricultural
and Forest Meteorology, 96(4), 181188
Wang S; Deltour J (1996). An experimental ventilationfunction
for large greenhouse based on a dynamicenergy balance model.
Agricultural Engineering Journal, 5,103112
Wilson JD (1985). Numerical studies of ow through awindbreak.
Journal of Wind Engineering and IndustrialAerodynamics, 21,
119154
Woodruff V I (1997). Analysis of vent designs for
naturallyventilated gutter connected greenhouses. MS Thesis,
OhioState University, Colombus, OH
ARTICLE IN PRESST. BARTZANAS ET AL.490
Effect of Vent Arrangement on Windward Ventilation of a Tunnel
GreenhouseIntroductionMaterials and methodsExperimental
greenhouseMeasurementsAir velocity measurementsTracer gas
measurements
Numerical modelMesh and boundary conditionsThe equivalent porous
medium approachSimulation of the ventilations rates
Vent configurations used for simulationConfiguration (a): side
openings only (roll-up type)Configuration (b): side only openings
(pivoting door type)Configuration (c): roof opening
onlyConfiguration (d): combined roof and side openings (roll-up
type)
ResultsNumerical model validationAirflow patterns and
temperature distribution for different ventilator
configurationsConfiguration (a): side openings only (roll-up
type)Configuration (b): side only openings (pivoting door
type)Configuration (c): roof opening onlyConfiguration (d):
combined roof and side openings (roll-up type)
DiscussionConclusionsReferences