Thermal analysis of greenhouses installed under semi

arid climate
Kamel Mesmoudi, Kheireddine Meguellati, Pierre-Emmanuel Bournet

To cite this version:

Kamel Mesmoudi, Kheireddine Meguellati, Pierre-Emmanuel Bournet. Thermal analysis of green-
houses installed under semi arid climate. International Journal of Heat and Technology, 2017, 35 (3),
pp.474-486. �10.18280/ijht.350304�. �hal-01817756�

HAL Id: hal-01817756

https://institut-agro-rennes-angers.hal.science/hal-01817756v1
Submitted on 18 Jun 2018

HAL is a multi-disciplinary open access L’archive ouverte pluridisciplinaire HAL, est

archive for the deposit and dissemination of sci- destinée au dépôt et à la diffusion de documents
entific research documents, whether they are pub- scientifiques de niveau recherche, publiés ou non,
lished or not. The documents may come from émanant des établissements d’enseignement et de
teaching and research institutions in France or recherche français ou étrangers, des laboratoires
abroad, or from public or private research centers. publics ou privés.
 INTERNATIONAL JOURNAL OF A publication of IIETA
HEAT AND TECHNOLOGY
ISSN: 0392-8764
Vol. 35, No. 3, September 2017, pp. 474-486
DOI: 10.18280/ijht.350304
Licensed under CC BY-NC 4.0 http://www.iieta.org/Journals/IJHT

Thermal analysis of greenhouses installed under semi arid climate

Kamel Mesmoudi1*, Kheireddine Meguellati2, Pierre-Emmanuel Bournet3
1
Laboratory of Mechanics Structural and Materials, University of Batna 2, Avenue Boukhlouf
Mohamed El Hadi 05000 Batna, Algeria
2
Department of Materials Science, University of Batna1, Avenue Boukhlouf
Mohamed El Hadi 05000 Batna, Algeria
3
Agro campus Ouest, EPHor, Environmental Physics and Horticulture Research Unit,
F-49045 Angers, France

Email: [email protected]

ABSTRACT

The greenhouse design as well as the cover material properties in particular may strongly impact the
greenhouse energy. To study the effect of these parameters, three typical unheated greenhouses equipped with
rows of canopy were considered. Experiments were launched to establish the boundary conditions and
validate the model. Two parametric studies were carried out: for the nocturnal period when the energy
performance of each type of greenhouse was investigated, and for the diurnal period, when the sun path was
simulated taking into account the type of the cover, its spectral optical and thermal properties. Results indicate
that for the nocturnal period, the ambient air temperature in the tunnel and vertical wall greenhouse was
relatively homogenous and warmer compared with the temperature distribution in the Venlo greenhouse. The
plastic greenhouse, especially the tunnel one had better performances concerning the homogenization of the
climate and the thermal energy storage. Concerning the diurnal period, and for both plastic greenhouses
equipped with fully opened side vents, the air located between the rows of canopy and ground surfaces
remained very slow, not exceeding 0.2 ms-1; for the Venlo glasshouse, the recirculation loop situated above
the crop improved the air mixing and induced a good homogenization. Results indicate that the cover material
with highest absorptivity, deteriorated the natural ventilation, increasing the air temperature by convection,
and reduced the available Photosynthetically Active Radiation.

Keywords: Greenhouse Design, Thermal Analysis, CFD Simulation, Radiation, Coupled Model.

1. INTRODUCTION variables with low economic cost (compared with

experiments). Nowadays the CFD technique is recognized as
A greenhouse is an enclosed structure, that protects the a powerful tool to model the climate generated inside
crops from the outside environment by creating favorable greenhouses and to test the performances of different
conditions, which traps the short wavelength solar radiation structural designs.
and stores the long wavelength thermal radiation to create a Since the pioneering work of Nara [1], CFD simulation has
favorable microclimate for higher productivity, together with been increasingly used to study assess distributed indoor
certain limitations that will depend on the bioclimatic stage climate for a wide range of greenhouse shapes, especially in
conditions of its location, the geometry of the structure, and the north Latitude. Several review papers present the state of
the spectral optical properties of covering materials in the art concerning CFD developments. Reicharth [2]
particular. presented the main conclusions derived from the published
Managing the greenhouse microclimate is essential to material together with their latest results on greenhouse
maintain an optimum inside environment during the different modeling. Norton [3] provided a state-of-the-art review of
stages of plant growth. Modeling is an interesting approach CFD and its applications in the design of ventilation systems
to assess the microclimate in greenhouses, and test different for agricultural production systems. They concluded that the
scenarios. Among the modeling tools, CFD (Computational greenhouse CFD modeling was a higher standard than that of
Fluid Dynamics) is an advanced technique for design in animal housing, owing to the incorporation of the crop
engineering. It has been increasingly used in different types biological responses as a function of the local environmental
of agricultural studies, such as livestock houses, greenhouses conditions. The main factors governing air movements inside
and broiler houses. CFD offers many advantages to the food the greenhouse were analyzed by [4], with a particular focus
industry as it provides a mean to test the influence of multiple on conclusions drawn from field experiments, laboratory

474
scale models and CFD simulations. The principles of CFD, is semi-arid and the use of greenhouses for crop production is
the modeling approach and its adaptation to greenhouse rapidly increasing. However, the characterization of the
climate simulation were described, paying attention to energy balance of the greenhouses for this bioclimatic zone
ventilation efficiency inside greenhouses with respect to the still remains to be done and achieving favorable environment
greenhouse geometry, opening arrangements, wind speed and becomes essential in order to warranty the greenhouse
direction, addition of insect-proof or shading screens, and feasibility [16,17]. Indeed, maintaining ventilation
interactions with the crop. More recently Bartzanas [5] performance during the diurnal period and controlling the
presented a review on various CFD applications to improve heat release during the nocturnal period are the major factors
crop farming systems such as, soil tillage, sprayers, influencing both climate control and yield quality over much
harvesting, machinery, and greenhouses, they discussed the of the year. These aspects are major challenges still facing
possibilities of incorporating the CFD models in decision designers and growers. Nevertheless, few investigations of
support tools for precision farming. the performance of greenhouses in southern Mediterranean
Specific processes involved in greenhouses were also climates have been undertaken so far and the involved
analyzed into details in the literature. These processes include physical mechanisms remain poorly understood. Some
ventilation, interaction with the crop and radiative effects. progress was made in recent years since the energy balance
The effect of vent arrangements on the ventilation and energy and the behavior of the indoor microclimate has long become
transfer in a multi-span glasshouse was studied by [6], using a matter of concern in the studies conducted by [18, 19, 20,
a bi-band radiation model. The analysis of the humidity 21].
issues in greenhouse climate using the CFD tools at different Performance criteria based on very different approaches
scales: the leaf, the canopy and the greenhouse itself was also are difficult to compare and a common approach based
conducted by Bournet [7]. The effect of the crop is clearly on the same bioclimatic stage is required so that
particularly important for greenhouses, as side openings may greenhouses performance can be simulated and examined
be partly obstructed by the crop rows. Hernandez [8] studied with respect to their engineering design (both greenhouse
the effects of crop row orientation (perpendicular or parallel geometry and covering material). Under arid climate
to the wall equipped with side openings) on the ventilation conditions, few CFD works that predict and analyze the
and microclimate of a plastic multi span greenhouse. More microclimate of greenhouses exist [22]. In the present study
recently, based on a computational fluid dynamics (CFD) we will present a numerical analysis of the thermal
model and an experimental approach, Majdoubi [9] analyzed environment of greenhouses in Batna (6˚11' East, 35˚33'
the effect of crop row orientation on internal climate in a North). The region is localized at altitudes of 900-1000 m
large type greenhouse, and found the ventilation rate to be above sea level and characterized by high winter insulation,
heavily dependent on the orientation of the crop rows with varying from 10.5 to 14 hours/day between October and
respect to the dominant wind direction. The relationship March, and by cold and dry winters, with average minimal
between ventilation and the characteristics of a tomato crop temperatures between -5 ˚C and 2 ˚C during the night periods
growing inside was systematically studied by [10] in a of January to March, with low levels of moisture. The aim of
naturally ventilated tunnel greenhouse using the tracer gas the present study is to examine the influence of greenhouse
method. It appears however that most early studies ignored or configuration on the inside microclimate and energy
failed to consider the presence of the crop, and did not consumption for three different unheated greenhouses (tunnel,
provide detailed information about the way solar and Venlo and plastic vertical wall greenhouse) during two
atmospheric radiations were taken into account. periods (diurnal and nocturnal) focusing in particular on the
In recent years, the use of Computational Fluid Dynamics ventilation mechanism, the thermal behavior and the heat
made it possible to analyze the factors that determine losses. In this prospect a CFD model was used, and
greenhouse microclimate with respect to its structural experiments were launched to establish the boundary
specifications and used equipment [11, 12]. However, in conditions and validate the model.
those studies, radiation was not simulated directly through a
radiative transfer equation but its effect was indirectly
incorporated either through specific boundary conditions or 2. MATERIALS AND METHODS
through the addition of an extra source term in the energy
transport equation. More recently, Kim [13] included the 2.1 Experimental greenhouses
short wave length radiation in their simulations (diurnal
conditions) while Bournet [6] implemented a bi-band In order to estimate the ability of the model to correctly
radiation model distinguishing short and long wave length simulate the thermal characteristics of the microclimate of
contributions to take account of the diverse optical properties the tested greenhouses, production greenhouses were
of the glass within these different bands. Moreover, few equipped with sensors to provide input data for the model,
studies have addressed the question of the dynamics of solar and for its validation. The measurements were carried out in
radiation and temperature distribution as [14] did it for a three experimental N-S oriented greenhouses (tunnel, Venlo
tunnel greenhouse at a daily time scale. More recently [15] and plastic vertical wall greenhouse) located at the
presented numerical simulations of the climatic parameter agricultural research farm of the department of agronomy of
distribution of a ventilated tunnel greenhouse on the basis of the University of Batna1 (35.330 N., 6.110 E.) in the north
a 3D CFD approach using a bi-band discrete ordinates (DO) area of Eastern Algeria. The geometrical characteristics of
model, and calculating the sensible and latent heat transfers the greenhouses were as follows: for the tunnel and the
between leaves and the surrounding air by including the long plastic vertical wall greenhouses eaves height of 2.4m, ridge
wave and shortwave radiation fluxes in each crop control height of 3.4m, total width of 4 m, and total length of 8m
volume taking account of the sun position at each time step. (Figure 1), for the Venlo glasshouse, the greenhouse was a
In the southern Mediterranean basin, the bioclimatic stage standard 4 m width, 3.60 m high under the ridge and 3.27 m

475
high under the gutter (Figure 1, Figure 2). The glasshouse on a mast, 10 m away to the East of the greenhouse (Figure
was covered with a 4 mm thick horticultural glass and 1). External wind speed and direction were monitored by two
equipped with two opposite roof openings; the tunnel and cup anemometers (Model 100075, accuracy ±0.1m.s-1,
plastic vertical wall greenhouses were covered with a Climatronic Corporation) and a wind vane (Model 100076,
polyethylene sheet and were equipped with two continuous accuracy•±2°, Climatronic Corporation). The outside global
side openings (roll-up type) located at 0.6m from the ground solar radiation was measured with a pyranometer (SP lite,
and with a maximum opening of 0.9m. The greenhouses were Kipp & Zonen, Netherlands). The outside air temperature and
grown with a tomato crop, which reached a height of 1m humidity were also measured using platinum probes in
during the experiments. statically ventilated shelters (Model MP601A, accuracy•±
0.2%, Rotronic instrument crop) located at the same height as
Wind Direction the outside pyranometer. All the above mentioned
East West measurements were recorded every 2s and then averaged
over 30min periods, using a data logger system (Campbell
Scientific Micro logger, CR3000, USA).
(b) Measurements of the temperature and the humidity
distribution in the middle section of the greenhouse were also
3.4
conducted. The measurement locations were distributed
0.9 along a cross-section at the center of each greenhouse in the
0.6 same vertical plane. The temperature and relative humidity of
8
the interior air were recorded by means of a data logger
(OAKTON Logger Plus) using a remote sensing system. The
: Cup anemometer : Pyranometer temperatures of the solid surfaces (ground, underground and
: Platinum probe : Wind vane wall surfaces of the cover) were measured every 2 second
(a) Plastic tunnel greenhouse with roll-up type openings with thermocouples, and then averaged over 30 min periods.
The incoming solar radiation was measured with a
Wind Direction
pyranometer (SP lite, Kipp & Zonen, Netherlands) placed
East West inside the greenhouse at the center and 1.5m above the
ground. The cover surface temperatures of the greenhouse
were measured at six positions distributed along the
greenhouse sides and roof using stick on thermocouples
secured to the cover with transparent adhesive tape. The
3.4
storage and the processing of data were carried out with the
0.9 Micro Lab plus Software. Figure 2 shows the sketch of the
0.6 Venlo type experimental greenhouse and the location of the
sensors. The beach of Measurements and the accuracy of all
8
the sensors used are specified in Table A in Appendix
(b) Plastic vertical wall greenhouse with roll-up type
openings
0.4 0.2
Wind Direction
East West 0.9

3.2

0.5
1.5 0.2
3.6
3.2
2.3

0.9 0.9
5
4
1.80 1.80
(c) Venlo glass greenhouse with pivoting roof door type
openings 4

Figure 1. Geometries of the greenhouses and configurations : Tomato crop : Pyranometer

considered for the ventilation efficiency study : Air temperature and humidity measurements by probe

2.2 Measurements Figure 2. Sketch of the experimental Venlo greenhouse

showing the location of the sensors (All the distances are in
Two different types of measurements were conducted: (a) linear meter)
outside the greenhouse to determine the characteristics of the
atmospheric boundary layer in order to provide the boundary 2.3 Numerical model
conditions to the model, (b) inside the greenhouse to validate
the simulations.
The commercially available CFD code Fluent v.6.1 was
(a) Measurements of the weather conditions surrounding
used for this study. A 2-D grid was built for each case, and
the greenhouse were conducted with sensors installed outside

476
the model was run in order to compare the numerical results computational domain. The standard k-ε model assuming
with the experimental data. Although 2D simulations do not isotropic turbulence was adopted to describe turbulent
represent precisely the reality inside the greenhouse, it could transport as it proved to be a good compromise for a realistic
be a computationally beneficial assumption for the description of turbulence and computational efficiency as
investigation of the transport phenomena especially at the reported by several studies of greenhouse microclimate. This
middle section of a long structure with open side vents along model is a semi-empirical model based on additional
the whole length. In addition, 2D modeling also makes it transport equations for the turbulent kinetic energy (k) and its
possible to save significant computing time for model dissipation rate (ε). The complete set of the equations of the
simulation development, meshing and convergence process. k-ε model can be found in [24] and their commonly used set
of parameters (empirically determined) are: Cμ= 0.09, Cε1 =
2.3.1 Grid definition and numerical procedure 1.44, Cε2 = 1.96, σκ=1, and σε = 1.3 (Fluent, 1998).
The calculation domain was restricted to the greenhouse
itself, in the middle plan of the greenhouses ensuring fast Radiative sub model (RTE Radiative Transfer Equation).
calculation. The grid was selected after several attempts in The discrete ordinates method has received significant
order to reduce the CPU time needed for convergence and to attention due to its good compromise between accuracy,
ensure the independency of the numerical results from the computational economy and flexibility [25, 26]. Up until now
grid. The grid was an unstructured, quadrilateral mesh with a however, most CFD studies did not include both the
higher density in critical portions of the flow subject to interchange of short and long wavelength radiation between
strong gradients. After several trials with different densities, the sky and the greenhouse cladding and only indirectly
the calculations were based on a 70 by 90 cell grid (Figure 3). introduce the effect of radiative transfers in the model. In
The area of calculation includes the canopy, the soil, and the order to simulate the effect of solar incident radiation on the
inner walls of the greenhouses. Different regions with greenhouse cover, the discrete ordinate DO model was used.
adapted meshes were considered and no mesh was applied to In this model it was assumed that radiation energy was
the outer space surrounding the greenhouse. The inner space ‘convected’ simultaneously in all directions through the
was meshed using an unstructured grid with sizes varying medium at its own speed. The DO model available in Fluent
from 0.2m in the center of the greenhouse to 0.06 m near the makes it possible to solve the Radiative Transfer Equation
greenhouse cover. The crop rows, considered as a porous (RTE) in semi-transparent media. It can be used to assess
medium, were meshed using a structured, cubic, 0.18 m grid non-gray radiation using a gray-band model. So it is adequate
and the soil mesh under the greenhouse consisted of three for participating media with a spectral absorption coefficient
layers (0.01 m, 0.15 m and 1.44 m) with 0.005 m, 0.03 m and αλ that varies in a stepwise fashion across the spectral bands.
0.2 m structured meshes. The discrete ordinates radiation model solves the RTE for a
finite number of discrete solid angles, each associated with a
r
Porosities (Tomato croppies) vector direction s in the global Cartesian system (x, y, z). It
transforms the RTE equation into a transport equation for the
luminance in the spatial coordinates (x, y, z). The DO model
r
solves as many transport equations as there are s directions.
r r
The RTE for spectral intensity I   r , s  is written as:

r r
dI   r , s  r r T 4
     s  I   r , s     
Figure 3. Geometry of the whole calculation domain and ds 
4
(2)
greenhouse mesh details s r r r r
 I   r , s  .  s , s '  d 
4 0
2.3.2 Governing equations
The CFD method allows the explicit calculation of the
average velocity vector field of a flow by numerically solving where in Eq. (2) Iλ is the radiation intensity for wavelength λ
r r
the corresponding transport equations. The two-dimensional (W.m-2.sr-1), r the position vector, s the radiation direction
conservation equations describing the transport phenomena vector, αλ the spectral absorption coefficient (m-1), λ the
for steady flows are of the general form: wavelength (m-1), σ the Stefan-Boltzmann constant
(σ=5.672x10-8 W.m-2.K-4), σs the scattering coefficient (m-1),
 U    V   Φ the phase function, and Ω the solid angle. We assumed the
    2  S (1) refraction index, the scattering coefficient and phase function
x y to be independent of the wavelength. The angular space 4π at
any spatial location was discretized into Nθ x Nφ solid angles
In equation (1), ϕ represents the concentration of the of extent ωi, called control angles. The angles θ and φ are the
transported quantity in a dimensionless form, namely the polar and azimuthal angles, and are measured with respect to
momentum (velocity components) and the mass and energy. the global Cartesian system (x, y, z). In our case a 3x3
U, and V and are the components of velocity vector;   is the pixilation was used. Although in this equation the refraction
diffusion coefficient; and Sϕ is the source term. The index is considered to be constant, in the calculation of black
governing equations were discretized following the body emission as well as in the calculation of boundary
procedure described by Patankar [23] using the finite volume conditions imposed by semi-transparent walls the band length
technique which consists in integrating the governing depended values of refractive index were used (provided in
equations over a control volume. The Boussinesq model was Table 1 and Table 2). The RTE equation was integrated over
r r
activated to take account of the buoyancy effects in the each wavelength. Then the total intensity I  r , s  in each

477
 r r
direction and position r , s was computed using equation 2.3.3 Boundary conditions
(3): Boundary conditions for each variable ϕ (transport
variable) must be specified for each boundary surface of the
r r r r
I  r , s    I x  r , s  x (3) domain. In particular, ϕ values for the upper boundary and
x the leeward lateral boundaries were determined with the
assumption of a null gradient of ϕ. For the other boundaries,
Where, the summation is undertaken over the wavelength ϕ was determined either directly from experimental data
bands. The RTE equation is coupled with the energy equation bases or deduced from specific models. The left opening was
through a volumetric source term given by the following supposed to face the East and the wind direction: the wind
equation (4) [27]: was normal to the ridge and a parabolic wind profile was
imposed at the opening of the greenhouse with a given
qr  4
 velocity profile and temperature 300 K, which is considered
0 r r r
Sr      4 I   r    I 0  r , s  d   (4) to be the temperature of the ambient air around the
xi  0  greenhouse. This profile was determined from the
measurements of the wind speed at each ventilation opening
With Sr is the radiation source term (J), qr the radiative of the greenhouse. Figure 4 illustrates the corresponding
flux (W), xi the component in i- direction (m), and I 0 is the imposed profile fitted by a parabolic law. At the inlet section,
black body intensity given by the Planck function (W.m-2). a fully developed turbulent profile was also considered. At
the outlet section (leeward right opening), a constant pressure
Table 1. Optical properties of the cover for the three (P=Patm) was imposed. Finally, the boundary conditions
greenhouse geometries prescribed a wall type boundary condition along the floor and
wall and the cover was considered as a finite thickness wall
Glass Polyethylene film
consisting of semi-transparent materials.
Cover type & Thickness
(hortical glass) (low density)
(mm) 1,6
4 0.1 Opening height of the ventilator y

UV 0.05 0.37
Absorptivity PAR 0.1 0.09 1,4
(α) NIR 0.2 0.05
UV 1.65 1.72 1,2
(m)

Refractive index PAR 1.65 1.79

(n) NIR 1.72 1.79 1 y = -10,14U2 + 21,36U - 8,996
R² = 0,978
Table 2. Mean values of the thermal and optical properties of 0,8
the greenhouse components
0,6
Material Polyethylene Glass Ground Canopy
film 0 0,5 1 1,5 2 2,5
Velocity (ms-1)
Density 923 840 1300 700
(kg.m-3)
Figure 4. Wind profile imposed at the opening of the plastic
Heat transfer 0.38 0.76 1.00 0.17
greenhouses
conductivity
(W.m-1.K-1)
The optical and thermal properties of the components of
Specific heat 2300 2700 800 2310
greenhouses are provided in Table 1 [30] and Table 2 [16]. A
capacity
heat flux boundary condition was applied at the external
(J.kg-1.K-1)
boundary of the cover region. It is a mixed heat flux
Absorptivity α table 1 table 0.95 0.46
boundary condition (combination of radiation and convection
1
with convective heat transfer coefficient). The corresponding
Refractive index table 1 table 1.92 2.77
convective coefficient depends on the wind speed, according
n 1
to the law established by [16] on the same type of greenhouse
and under similar climatic conditions h=2.56+2.3U0.69,
Crop sub model. The crop was simulated using the where U is the mean wind speed along the roof of the
equivalent porous medium approach through the addition of a greenhouse. Also, the same boundary condition was imposed
momentum source term, due to the drag effect of the crop, to along the internal wall surface where the solid and the fluid
the standard fluid flow equations [28]. The plants were zones are coupled, restoring a conjugated heat transfer
simulated as porous materials with a viscous resistance a-1 = treatment at the specific area. The convective coefficient
27380m-2 and inertial resistance C2=1.534m-1. These between the interior air and the interior wall depends on the
parameters used in the pressure drop expression for a tomato temperature gradient (interior air - interior wall) according to
crop were derived from [29] for a low velocity range. For the the following law [16] h=3.59ΔT0.33. Fixed air temperatures
purpose of the study, sensible and latent heat transfers were were imposed along the ground. The side walls were
omitted and attention was rather paid on the mechanical considered as adiabatic and opaque while the ground was
interaction of the crop with its environment. considered as a diffusively radiating opaque material.

478
2.3.4 Numerical procedure 3.1 Validation of the model
A second-order upwind discretization scheme was used for
momentum and turbulence transport equation. The In order to check on the validity of the performance of the
convergence criterion for all variables was 10-6. CFD model, the validation of the present work was
undertaken on the basis on experimental field surveys
2.4 Parametric studies conducted in the Venlo geometry greenhouse covered with
horticultural glass. Figure 5 shows the air temperature
Two parametric studies were carried out: diurnal and profiles for this glasshouse along the middle axis of the
nocturnal conditions, in order to investigate the effect of greenhouse at 2m from the inlet flow vent opening both for
greenhouse geometry, as well as the effect of two different the diurnal period Figure 5a, and for the nocturnal period
cover materials with different optical characteristics, on the Figure 5b.
thermal behavior, heat losses, and temperature patterns of the
tested greenhouses. For the first parametric study, i.e. the 4
diurnal period, a typical day of the spring season in the 3,5
region of Batna was chosen for simulation sand calculations

Greenouse High (m)

3 Numerical
were launched at a time corresponding to midday. The 2,5 Experiment
incident irradiance (the earth solar radiation) was distributed 2
in three wavelength bands: the ultra-violet (λ=0.01 - 0.4 μm),
1,5
the visible or PAR (λ=0.4 - 0.76μm) and the near infrared
1
(λ=0.76 - 1.1 μm). In Table 1, the normal irradiances per
0,5
wavelength band are presented. In all cases a fraction of 24% Temperature (K)

diffuse radiation was considered. For the second parametric 0

290 295 300 305 310 315 320
study, i.e. the nocturnal period, the same day was considered,
but at midnight, and two cover materials with different
optical properties were studied. The corresponding spectral
(a) Diurnal period
optical properties are provided in Table 1 and Table 2.
4

3,5

3
3. RESULTS AND DISCUSSIONS
Greenouse Hight (m)

2,5 Numerical
Experiment
The measurements were carried out on a, sunny springs 2

days (march 10th to April 1st 2015) at solar noon for the first 1,5

parametric study, and at midnight for the second parametric 1

study. The required parameters were measured, every 1min, 0,5

Temperature (K)
at the locations shown in Figure 1, averaged at every 15 min 0
and recorded in a data logger (CR3000 Micro logger, 286 287 288 289 290 291 292

Campbell Scientific, Inc.). These Measurements were

conducted over two periods, from March to April 2015inside
the tunnel, Venlo and plastic vertical wall greenhouses. (b) Nocturnal period
During the same period, climatic data was also recorded
outside the greenhouses. Sets of data were used not only to Figure 5. Computed and experimental vertical profiles of the
define the boundary conditions of the model, but also to temperature (K) at the middle of the Venlo Glasshouse,
validate the simulations. For the purpose of the study, two (a) Diurnal period, (b) Nocturnal period
contrasted cases were analyzed: (a) diurnal period, (b)
nocturnal period. The transport phenomena inside the For the diurnal case, a good agreement between the
experimental naturally ventilated greenhouses were measured and simulated profiles was reached, a correlation
investigated using the mean values of the outside climate coefficient of R2 = 0.9669 and a mean square error of χ =
conditions for specifying the boundary conditions Table 3. 2.2481K are denoted and the standard deviation of the
temperature (±1°C) may be ascribed to the experimental
Table 3. Mean values of the outside climate conditions errors and to the models used for the determination of the
during the measurements temperature. Figure 5a shows that temperature distribution
along the vertical axis disclosed two distinct areas, one at the
Min & Max Outside air Temperature (K) 279-297.3 bottom of the domain, where the temperature remained
Min & Max Outside air Temperature (K) 277-313 relatively high due to the energy exchange with plants, and a
second on eat the top half domain where the temperature was
Outside air Humidity (%) 52.08 clearly affected by the fresher temperature of the entering
Outside air velocity (m.s-1) 1.3 stream. A difference of about 5K was observed and simulated
Global solar radiation (W.m-2) 610.8 between these two areas. Concerning the nocturnal period,
UV (W.m-2) 33 Figure 5b shows the vertical distribution of the measured
temperatures and the numerical air temperatures again at 2m
PAR (W.m-2) 310
from the sides of the greenhouse. In this case also a good
NIR (W.m-2) 267.8 agreement was reached between the numerical and the
Hour angle degree (˚) 0 experimental values. While a correlation coefficient of R2 =
0.9440 and a mean square error of χ = 0.1036 are denoted.

479
When observing Figure 5b, we denote globally a good Modifying the optical properties of the covering material
correlation between predicted and measured air temperature. impacted the amount of solar energy entering the greenhouse,
Contrary to the diurnal case, the air temperature profile was causing variability in the flow pattern for each type of
relatively homogeneous except in areas close to the ground greenhouse and for each studied cover material described in
and roof where high temperature gradients were reported. Table 2. Two recirculation loops appeared above and below
These gradients were mainly induced by heat exchanges the inlet, trapping small amounts of fluid. The optical
along the ground and roof. properties of the cover determined not only the size but even
the existence of the upper corner recirculation for the plastic
3.1.1 Diurnal period greenhouses (tunnel and straight wall greenhouse) as can be
Flow field. CFD results concerning the first parametric seen in Fig. 8a and Fig. 8b. In these greenhouses, the main
study: i.e. the diurnal period for a clear day of the spring stream flew above the plants and the smallest one inside the
season are shown in Figures 6-10 for all considered crop with lower speed. The differences of the streamlines
geometries. The computed contours of the air velocities, contours between both plastic greenhouses are restricted to
stream function, air temperature, and PAR radiation profiles the upper domain close to the inlet. Thus the dependence of
at specific sections are provided. From the results, it comes the upper corner recirculation on the cover material and on
out that the main mechanism governing heat transfers is the greenhouse geometry is clear and its size was probably
convection associated with the entering air stream except in determined by the combination of both effects (the tunnel one
areas close to the cover and in the corners of the greenhouse was smaller than the vertical wall greenhouse one). This
where incident solar radiation and heat storage mainly impact recirculation plays a significant role in the total flow pattern
the temperature. As shown in Fig 6a, Fig.6b and Fig. 6c the and temperature distribution inside the greenhouse since it
computed contours of the air velocities for all cases showed divides the domain into two distinguished areas. The
that the flow was dominated by a strong convective airflow recirculation formed at the bottom corner near the entrance
through the windward opening. The internal flow had the seemed however to be independent of radiation. The air
same direction as the wind direction and was damped by situated between the rows of canopy was hardly affected by
plants (porosity). Due to the obstacle created by the crop, the the main entering stream, with velocities in this region not
flow separated into two unequal streams. Results thus exceeding 0.2 ms-1. The flow decelerated as a consequence of
indicate that the wind direction clearly influenced the air the viscous and inertial resistances. Above the height of the
velocity inside the greenhouse and hence its ventilation rates. ventilator (i.e. at 1.6 m), the air velocities progressively
reduced. The computed contours of the air velocities obtained
2.02 e+00
1.92 e+00
for these cases were characterized by a weak air current near
1.82 e+00
1.71 e+00 the ground, and a recirculation loop with slower speed near
1.65 e+00
1.55 e+00 the roof and flowing counter current with respect to the
1.45 e+00
1.35 e+00
1.25 e+00
outside wind. This recirculation loop improved the air mixing
1.15 e+00
1.05 e+00
but most of the air left the greenhouse volume without a good
9.06 e-01
8.07 e-01
homogenization.
7.06 e-01
6.06 e-01 Contrary to the plastic greenhouses, for the Venlo
5.05 e-01
4.04 e-01 greenhouse, no air recirculation close to the roof was
3.03 e-01
2.02 e-01
1.02 e-01
predicted but a large loop was simulated at the bottom,
9.03 e-04 trapping large amounts of fluid. The existence of this
(a) Tunnel plastic greenhouse recirculation was mainly governed by the geometry and the
2.02 e+00
1.92 e+00
1.82 e+00
vent location effects. In addition, the air flow near the roof
1.71 e+00
1.65 e+00
was mainly driven by the convective flow through the vent,
1.55 e+00
1.45 e+00 reaching maximum values within the range [1.4; 1.7 ms-1].
1.35 e+00
1.25 e+00 Fig.7a, Fig.7b and Fig.7c provide the horizontal u-velocity
1.15 e+00
1.05 e+00
9.06 e-01
profiles at different locations inside the greenhouses at a
8.07 e-01
7.06 e-01
distance of 2m, 4m and 6m respectively from the inlet for the
6.06 e-01
5.05 e-01
plastic greenhouses and at 0.25m, 2m, and 3m for the Venlo
4.04 e-01
3.03 e-01 glasshouse (the Venlo greenhouse is relatively small
2.02 e-01
1.02 e-01 compared with the plastic ones). Close to the ground the
9.03 e-04

(b) Straight wall plastic greenhouse velocity profiles were similar for all cases and characterized
1.68 e+00 by low velocities caused by the resistance of tomato plants
1.66 e+00
1.44 e+00
1.32 e+00
combined with shear along the floor. Large peak appeared
1.19 e+00
1.07 e+00
over the plants where the flow accelerated. The peak position
9.49 e-01
8.27 e-01
moved up according to the distance from the inlet, while its
7.05 e-01
5.80 e-01 magnitude decreased, following the spreading of the jet as it
4.60 e-01
3.57 e-01 mixed with the ambient air. For the plastic greenhouses (Fig.
2.15 e-01
9.25 e-02
-3.00 e-02
7a and Fig.7b), close to the roof, negative values of u-
-1.52 e-01
-2.75 e-01
velocity were predicted, corresponding to area backflow. A
-3.97 e-01
-5.20 e-01 difference in the profiles was observed for the straight wall
-6.42 e-01
-7.85 e-01 greenhouse in the upper part of the domain close to the vent
(c) Venlo glasshouse height: the mean negative air velocity in this region had
values within the range [-0.6; -0.1] ms-1 for the straight wall
Figure 6. Computed contours of the air velocity (ms-1) at the greenhouse, and within the range [-0.4; -0.01] ms-1 for the
middle of the greenhouses tunnel greenhouse), showing the dependence of the flow

480
(especially in terms of velocity magnitude) on the greenhouse 1.90 e+00
1.76 e+00
design. 1.66 e+00
1.56 e+00
1.46 e+00
1.37 e+00
1.27 e+00
1.17 e+00
1.07 e+00
9.76 e-01
8.78 e-01
7.81 e-01
6.83 e-01
5.85 e-01
4.88 e-01
3.90 e-01
2.93 e-01
1.95 e-01
9.76 e-02
0.00 e+00

(c) Venlo glasshouse

Figure 8. Computed contours of stream function (m2s-1) at

the middle of the greenhouses
(a) Tunnel plastic greenhouse Temperature distribution. Fig. 9a, Fig. 9b and Fig. 9c show
the air temperature distributions for the three types of
greenhouse design (for the two studied covering materials)
under diurnal conditions. Not surprisingly, the temperature
distribution followed the air velocity distribution. In the area
just above the crop, the air temperature was similar to that of
the outside air (295-298K) due to the strong air movement in
this region. The temperatures in the center of the greenhouse
were relatively homogeneous above the crop rows, while
they strongly vary in the vicinity of the walls. Lower
temperatures were predicted close to the main stream of air
(b) Straight wall plastic greenhouse (coming from the outside), while higher temperatures were
simulated near the ground and roof.
3.19 e+02
3.17 e+02
3.16 e+02
3.14 e+02
3.13 e+02
3.12 e+02
3.10 e+02
3.09 e+02
3.07 e+02
3.06 e+02
3.05 e+02
3.03 e+02
3.02 e+02
3.01 e+02
2.99 e+02
2.98 e+02
2.96 e+02
2.95 e+02
2.94 e+02
2.92 e+02

(a) Tunnel plastic greenhouse

3.19 e+02
3.17 e+02
3.16 e+02
(c) Venlo glasshouse 3.14 e+02
3.13 e+02
3.12 e+02
3.10 e+02

Figure 7. Computed profiles of air velocity (ms-1) at the 3.09 e+02

3.07 e+02
3.06 e+02
middle of the greenhouses for three positions x=2m, 4m, 6m 3.05 e+02
3.03 e+02
from the inlet for cases (a) and (b) and x=0.25m, 2m, 3.75m 3.02 e+02
3.01 e+02
for case (c) 2.99 e+02
2.98 e+02
2.95 e+02
2.94 e+02
1.82 e+00 2.92 e+02 :
1.52 e+00
1.36 e+00
(b) Straight wall plastic greenhouse Therm
1.25 e+00
1.13 e+00
ocoup
3.26 e+02
3.24 e+02
1.02 e+00 les
3.23 e+02
8.75 e-01 3.22 e+02
3.21 e+02
7.52 e-01 3.20 e+02
6.25 e-01 3.19 e+02
5.00 e-01 3.18 e+02
3.75 e-01 3.17 e+02
3.15 e+02
2.55 e-01 3.14 e+02
1.25 e-01 3.13 e+02
0.00 e+00 3.12 e+02
3.11 e+02
3.10 e+02
3.09 e+02
(a) Tunnel plastic greenhouse 3.08 e+02
3.06 e+02
1.82 e+00 3.05 e+02
1.52 e+00 3.04 e+02
1.36 e+00 3.03 e+02
3.02 e+02
1.25 e+00 3.01 e+02
1.13 e+00 3.00 e+02
1.02 e+00 2.99 e+02
2.97 e+02
8.75 e-01
2.96 e+02
7.52 e-01 2.95 e+02
6.25 e-01 2.94 e+02
5.00 e-01 2.93 e+02
3.75 e-01
2.55 e-01
(c) Venlo glasshouse
1.25 e-01
0.00 e+00
Figure 9. Computed contours of air temperature (K) at the
(b) Straight wall plastic greenhouse middle of the greenhouses

481
 For the plastic greenhouses, the main temperature the greenhouse where the temperature was mainly affected
gradients were predicted in the upper corner near the inlet, by the flow ventilation. In this area the temperature was close
since in this region the total heat transfer was stronger. The to the temperature of the entering stream. A core flow
rest of the domain presented similar patterns for the appeared at the center where the temperature of the air
examined case. For the vertical wall greenhouse covered with remains droughty homogenous. The recirculation region in
a material characterized by a high absorptivity (plastic the canopy zone was fully developed on contrary to the two
cover), the roof temperature reached a relatively high other designs, resulting in a temperature rise.
equilibrium temperature, causing heating by convection of
the nearby air. The temperature of the back flow air trapped
in this recirculation zone increased as this dead zone favored
the accumulation of heat provided by the incoming
transmitted radiation. The same behavior was also depicted in
Fig.10a and Fig.10b showing the temperature profiles at
predefined positions.

(a) Tunnel plastic greenhouse

(a) Tunnel plastic greenhouse

(b) Straight wall plastic greenhouse

(b) Straight wall plastic greenhouse

(c) Venlo glasshouse

Figure 11. Computed PAR (Wm-2) profiles along the

greenhouse width at a level of 1.5m from greenhouse ground

Radiation distribution. In Fig.11a and Fig.11b and Fig.11c

the PAR radiation profiles at 1.5 m from the ground are
presented for the three studied greenhouses, with the two
different covering materials (thin plastic film and
(c) Venlo glasshouse
horticultural glass). The impact of covering materials in
terms of PAR penetrating into the interior greenhouse was of
Figure 10. The computed profiles of air temperature (K) at
course directly linked to the material transmittance. From the
the middle of the greenhouses for three positions x=2m, 4m,
figures, two groups of greenhouses which present roughly
6m from the inlet for cases (a) and (b) and x=0.25m, 2m,
similar behaviors may be distinguished: the tunnel and
3.75m for case (c)
straight wall greenhouses with thin plastic film (Fig.11a and
Fig.11b) on the one hand, and the Venlo glass greenhouse
Fig.9c and Fig.10c present the air temperature contours
with Fig. 8c on the other hand. For the plastic greenhouses
and temperature profiles for the Venlo glasshouse. Two
the low value of the transmittance only allowed a small
distinct areas can be observed: one at the bottom of domain,
amount of PAR to enter the greenhouse and the PAR
where the temperature distribution was mainly governed by
distribution inside the greenhouse almost disclosed a uniform
the energy exchange with plants and soil. In this region, the
distribution over the plants. Both plastic greenhouses had
temperature was mostly affected by the reduced air velocity
analogous performance as they absorbed significant amount
and the recirculation. The second area is located at the top of

482
of incident solar PAR. In both cases, the maximum PAR forces and air movements. The highest velocities were also
intensity reached roughly the same value, proving the good predicted along the vertical medium axis (ascendant stream
functional performance of the tunnel design of the movement) and near the walls (descendant stream
greenhouse roof, compared with the straight wall greenhouse movement) for each of tested geometry.
design for which a lower percentage of the incoming PAR Temperature variation. Fig.13a, Fig.13b and Fig.13c show
radiation reached the crop. Conversely, the Venlo greenhouse the air temperature distributions for the three greenhouse
was covered with glass with a higher transmittance and designs. For all cases, the predicted air temperature inside the
therefore made it possible higher amounts of PAR to greenhouse reached the max values near the ground and in
penetrate inside the shelter (meaning a higher PAR reached the ascending streams at the center of greenhouses (290-288
the crop). K). The lowest values were obtained near the walls (287-285
K). Simulations also reveal that for the plastic greenhouses
3.1.2 Nocturnal period (and especially for the tunnel greenhouse), the ambient air
temperature distribution was relatively homogenous and
1.84 e-01
1.75 e-01 higher compared with the temperature distribution inside the
1.65 e-01
1.56 e-01 Venlo greenhouse. The average air temperature inside the
1.47 e-01
1.38 e-01
1.29 e-01
tunnel greenhouse was 290 K (standard deviation ±0.45), it
1.20 e-01
1.10 e-01
was 288 K (±0.5) in the vertical wall greenhouse and 287 K
1.01 e-01
9.21 e-02 (±1.32) in the Venlo glasshouse. The tunnel greenhouse
8.29 e-02
7.37 e-02 geometry disclosed the best air mixing during this period (at
6.45 e-02
5.54 e-02 night), probably caused and facilitated by the curvature of the
4.62 e-02
3.70 e-02
2.79 e-02
roof of the greenhouse. Air temperature was almost
1.87 e-02
9.51 e-03
uniformly distributed in the Venlo and vertical wall
3.33 e-04
greenhouses but it was 2 K less than the interior tunnel air
(a) Tunnel plastic greenhouse temperature.
1.84 e-01
1.75 e-01
1.65 e-01
1.56 e-01 2.91 e+02
1.47 e-01 2.91 e+02
1.38 e-01
1.29 e-01 2.90 e+02
1.20 e-01 2.90 e+02
1.10 e-01
1.01 e-01 2.90 e+02
9.21 e-02 2.90 e+02
8.29 e-02
7.37 e-02 2.89 e+02
6.45 e-02 2.89 e+02
5.54 e-02
4.62 e-02 2.89 e+02
3.70 e-02 2.88 e+02
2.79 e-02
1.87 e-02 2.88 e+02
9.51 e-03 2.88 e+02
3.33 e-04
2.88 e+02
(b) Straight wall plastic greenhouse 2.87 e+02
1.84 e-01 2.87 e+02
1.75 e-01
1.65 e-01
1.56 e-01
(a) Tunnel plastic greenhouse
1.47 e-01
1.38 e-01
1.29 e-01 2.91 e+02
1.20 e-01 2.91 e+02
1.10 e-01
1.01 e-01 2.90 e+02
9.21 e-02 2.90 e+02
8.29 e-02
7.37 e-02 2.90 e+02
6.45 e-02 2.90 e+02
5.54 e-02
4.62 e-02 2.89 e+02
3.70 e-02 2.89 e+02
2.79 e-02
1.87 e-02 2.89 e+02
9.51 e-03 2.88 e+02
3.33 e-04
2.88 e+02
(c) Venlo glasshouse 2.88 e+02
2.88 e+02
2.87 e+02
-1
Figure 12. Computed contours of air velocity (ms ) at the 2.87 e+02

middle of the greenhouses (b) Straight wall plastic greenhouse

2.91 e+02
Flow field. During the night, the greenhouses were closed, 2.91 e+02
unheated and deprived of any heating system. Under such 2.90 e+02
2.90 e+02
conditions, the movement of the interior air was 2.90 e+02
characterized by two counter-rotative convective loops 2.90 e+02

guided by the greenhouse walls and following a circular 2.89 e+02

2.89 e+02
trajectory along the internal surface of the walls and the roof. 2.89 e+02

The ascendance of the air in the center of the greenhouse in 2.88 e+02
2.88 e+02
that case was mainly driven by the convection induced by the 2.88 e+02

heat stored inside the ground during daytime (Fig.12a, 2.88 e+02
2.87 e+02
Fig.12b and Fig.12c) and released at night. Low values of air 2.87 e+02

velocities were predicted in the vicinity of the canopy rows (c) Venlo glasshouse
(0.03 - 0.01 ms-1). Velocity collapsed inside the crop as a
consequence of the viscous and inertial resistances. It reached Figure 13. Computed contours of air temperature (K) at the
maximum values (0.19 - 0.15 ms-1) near the soil under the middle of the greenhouses
canopy where the temperature gradients enhanced buoyancy

483
 1.64 e-01
1.53 e-01
greenhouses equipped with fully opened side vents, the air
1.43 e-01
1.33 e-01
located between the rows of canopy, and near the roof,
1.23 e-01
1.12 e-01
remained very slow, not exceeding 0.2 m s-1. For the Venlo
1.02 e-01
9.20 e-02
greenhouse, the recirculation loop situated above the crop
8.18 e-02
7.16 e-02
improved the air mixing and appeared to induce a good
6.13 e-02
5.11 e-02
homogenization compared with the plastic greenhouse
4.09 e-02 geometry. The flow recirculation, showed the importance of
3.07 e-02
2.04 e-02 internal temperature gradients, although forced convection
1.02 e-02
0.00 e+00 which resulted from natural ventilation was dominant.
(a) Tunnel plastic greenhouse Consequently the Venlo greenhouse had the best
1.64 e-01 performance in terms of ventilation, particularly in the area
1.53 e-01
1.43 e-01 covered by the crop (0.4-0.6 ms-1) compared with the plastic
1.33 e-01
1.23 e-01 greenhouse for which air velocities less than 0.3 ms-1 were
1.12 e-01
1.02 e-01 predicted. The Venlo greenhouse also maintained a relatively
9.20 e-02
8.18 e-02 low difference of temperature with the outside air (6-7 K)
7.16 e-02
6.13 e-02 compared with the plastic greenhouse (8-10 K). In the Venlo
5.11 e-02
4.09 e-02 glasshouse, the canopy located in the middle of the
3.07 e-02
2.04 e-02
greenhouse also received higher amounts of PAR compared
1.02 e-02
0.00 e+00
with plants located in the vicinity of the walls. Such
(b) Straight wall plastic greenhouse heterogeneity in the PAR distribution may lead to an
1.25 e-01 important variability in the crop activity, thus impacting the
1.20 e-01
1.16 e-01 crop growth and development.
1.12 e-01
1.08 e-01
1.04 e-01
Concerning the nocturnal case, the ambient air temperature
9.97 e-02
9.55 e-02
in the tunnel and vertical wall greenhouse was relatively
9.14 e-02
8.72 e-02 homogenous and higher compared with the temperature
8.31 e-02
7.89 e-02 distribution in the Venlo glasshouse. The air temperature at
7.48 e-02
7.06 e-02
6.64 e-02
the center of the tunnel greenhouse was 290 K, while it was
6.23 e-02
5.81 e-02
288 K and 287 K in the vertical wall and Venlo glasshouse
5.40 e-02
4.98 e-02
respectively. It can be concluded that for the nocturnal
4.57 e-02
4.15 e-02 period, the plastic greenhouse, especially the tunnel one had
3.74 e-02
3.32 e-02 better performances concerning the climate homogenization
2.49 e-02
1.66 e-02
1.25 e-02
and the thermal energy storage.
0.00 e+00 This study paves the way for future investigations on the
(c) Venlo glasshouse impact of greenhouse design and choice of the covering
material on greenhouse climate in semi-arid areas It also
Figure 13. Computed contours of stream function (m2s-1) at stresses the need to properly include thermal transfers as well
the middle of the greenhouses as radiative transfers in the modeling approach in order to
accurately predict canopy radiation absorption,
photosynthesis and transpiration in the next developments of
4. CONCLUSIONS the numerical tool.

Assessing the micro climate in greenhouses is of prime

interest as climate is one of the main factors impacting plant REFERENCES
growth and development. Indeed the climate governs two
important physiological mechanisms of the plants, namely [1] Nara M. (1979). Studies on air distribution in farm
transpiration and photosynthesis. It is highly dependent on buildings, two dimensional numerical analysis and
greenhouse geometry, thermo-physical and optical properties experiment, Journal of the Society of Agricultural
of the covering material, and on the outside weather Structures of Japan, Vol. 9, No. 2, pp. 18-25.
conditions. In the present study, the influence of greenhouse Available:
design and configuration on greenhouse microclimate and https://www.jstage.jst.go.jp/article/sasj1971/9/2/9_2_1
energy consumption for unheated greenhouses under semi- /pdf
arid climate was numerically investigated using a [2] Reichrath S., Davies T.W. (2002). Using CFD to
commercially available CFD code. A field survey was model the internal climate of greenhouses: past,
undertaken in order to establish the boundary conditions and present and future, Agronomie, Vol. 22, No. 1, pp. 3-
validate the model. Three different greenhouse designs with 19. DOI: 10.1051/agro:2001006
two different covering materials (plastic and glass) were [3] Norton T., Sun D.W., Grant J., Fallon R., Dodd V.
investigated during two periods (night and daytime) resulting (2007). Applications of computational fluid dynamics
in different airflow and temperature patterns. (CFD) in the modeling and design of ventilation
Results indicate that for the first parametric study, namely systems in the agricultural industry: a review,
during daytime, in greenhouses with the cover material with Bioresource Technology, Vol. 98, No. 12, pp. 2386-
the highest absorptivity (i.e. the plastic film). High 2414. DOI: 10.1016/j.biortech.2006.11.025
absorptivity reduced the available Photosynthetically Active [4] Bournet P.E., Boulard T. (2010). Effect of ventilator
Radiation (PAR) but it distributed it equally inside the configuration on the distributed climate of
greenhouse. It was also concluded that for the same greenhouses: a review of experimental and CFD

484
 studies, Comput Electron Agr, Vol. 74, No. 2, pp. 195- Journal of the Association of ArabUniversities for
217. DOI: 10.1016/j.compag.2010.08.007 Basic and Applied Sciences, Vol. 9, No. 1, pp. 27-37.
[5] Bartzanas T., Kacira M., Zhu H., Karmakar S., DOI: 10.1016/j.jaubas.2010.12.007
Tamimi E., Katsoulas N., Lee T.B., Kittas C. (2013). [17] Mesmoudi K., Soudani A., Bournet P.E. (2010). The
Computational fluid dynamics applications to improve determination of the inside air temperature of a
crop production systems, Computers and Electronics greenhouse with tomato crop, under hot and arid
in Agriculture, Vol. 93, pp. 151-167. DOI: climates, J. App. Sc.in Env. Sanit., Vol. 5, No. 2, pp.
10.1016/j.compag.2012.05.012 114-127. HAL Id: hal-00729706
[6] Bournet P.E., Khaoua S.A.O., Boulard T. (2007). [18] Mesmoudi K., Bougoul S., Bournet P.E. (2012).
Numerical prediction of theeffect of vent arrangements Thermal performance of an unheated greenhouse
on the ventilation and energy transfer in multi- under semi-arid conditions during the night, Acta
spanglasshouse using a bi-band radiation model, Hort, Vol. 952, pp. 417-424. DOI:
Biosystems Engineering, Vol. 98, No. 2, pp. 224-234. 10.17660/ActaHortic.2012.952.52
DOI: 10.1016/j.biosystemseng.2007.06.007 [19] Serir L., Bournet P.E., Benmoussa H., Mesmoudi K.
[7] Bournet P.E. (2014). Assessing greenhouse climate (2012). Thermal simulation of a greenhouse under a
using CFD: a focus on air humidity issues, Acta Hort, semi-arid climate, Acta Hort, Vol. 927, pp. 635-642.
Vol. 1037, pp. 971-985. DOI: DOI: 10.17660/ActaHortic.2012.927.78
10.17660/ActaHortic.2014.1037.129 [20] Mesmoudi K., Bournet P.E., Zitouni B., Outtas T.
[8] Hernandez J., Fernandez-Tapia J.M., Hita O., Soriano (2013). Numerical simulation of the airflow and
T., Morales M.I., Castilla N., Escobar I. (2006). Effect temperature distribution in a closed venlo greenhouse
of crop row orientation on the passive ventilation of a under hot and arid climate, Acta Hort, Vol. 1008, pp.
plastic greenhouse, Acta Hortic, Vol. 719, pp. 205- 235-240. DOI: 10.17660/ActaHortic.2013.1008.31
210. DOI: 10.17660/ActaHortic.2006.719.21 [21] Mesmoudi K., Soudani A., Bougoul S., Bournet P.E.
[9] Majdoubi H. (2007). Contribution à la modélisation (2012). On the determination of the convective heat
d’une serre: couplage des transferts aérauliques et transfer coefficient at the greenhouse cover under semi
radiatifs dans une serre de tomate de grande surface arid climatic conditions, Acta Hort, Vol. 927, pp. 619-
dotée de filet anti-insectes, thesis, Univ. IBN ZOHR, 626. DOI: 10.17660/ActaHortic.2012.927.76
Agadir, Morocco. [22] Mesmoudi K., Bournet P.E. (2014). Assessing the
[10] Fatnassi H., Leyronasb C., Boulard T., Bardinb M., daily evolution of the climate inside a greenhouse
Nicotb P. (2009). Dependence of greenhouse tunnel under semi-arid conditions using field surveys and
ventilation on wind direction and crop height, Biosys CFD modeling, Acta Hort, Vol. 1037, pp. 1049-1054.
Engineering, Vol. 103, No. 3, pp. 338-343. DOI: DOI: 10.17660/ActaHortic.2014.1037.138
10.1016/j.biosystemseng.2009.03.005 [23] Patankar S.V. (1980). Calculation of the flow field, in
[11] Fatnassi H., Boulard T., Poncet C., Chave M. (2006). Numerical Heat Transfer and Fluid Flow, 1st ed. New
Optimisation of greenhouse insect screening with York, McGaw Hill, USA, ch. 6, sec. 7, pp. 126-129.
computational fluid dynamics, Biosys Engineering, [24] Mohammadi B., Pironneau O. (1994). Analysis of the
Vol. 93, No. 3, pp. 301-312. DOI: k-epsilon turbulence model, Research in Applied
10.1016/j.biosystemseng.2005.11.014 Mathematics, 1st ed., Paris, Wiley-Masson, France.
[12] Khaoua S.A.O., Bournet P.E., Migeon C., Chassériaux [25] Coelho J.P., de Moura Oliveira P.B., Boaventura C.J.
G., Boulard T. (2006). Analysis of greenhouse (2005). Greenhouse air temperature predictive control
ventilation efficiency with CFD, Biosys Engineering, using the particle swarm optimization algorithm,
Vol. 95, No. 1, pp. 83-98. DOI: Computers and Electronics in Agriculture, Vol. 49,
10.1016/j.biosystemseng.2006.05.004 No. 3, pp. 330-344. DOI:
[13] Kim K., Yoon J.Y., Kwon H.J., Han J.H., Son J.E., 10.1016/j.compag.2005.08.003
Nam S.W., Giacomelli G.A., Lee I.B. (2008). 3-D [26] Raithby G.D., Chui E.H. (1990). A finite-volume
CFD analysis of relative humidity distribution in method for predicting a radiant heat transfer in
greenhouse with a fog cooling system and refrigerative enclosures with participating media, Transactions of
dehumidifiers, Biosys Engineering, Vol. 100, No. 2, ASME Journal of Heat Transfer, Vol. 112, No. 2, pp.
pp. 245-255. DOI: 415-423. DOI: 10.1115/1.2910394
10.1016/j.biosystemseng.2008.03.006 [27] Kim S.H., Huh K.Y. (2000). A new angular
[14] Fidaros D.K., Baxevanou C.A., Bartzanas T., Kittas C. discretization scheme of the finite volume method for
(2010). Numerical simulation ofthermal behavior of a 3-D radiative heat transfer in absorbing, emitting and
ventilated arc greenhouse during a solar day, anisotropically scattering media. International Journal
Renewable Energy, Vol. 35, No. 7, pp. 1380-1386. of Heat and Mass Transfer, Vol. 43, No. 07, pp. 1233-
DOI: 10.1016/j.renene.2009.11.013 1242. DOI: 10.1016/S0017-9310(99)00211-2
[15] Nebbali R., Roy J.C., Boulard T. (2012). Dynamic [28] Boulard T., Wang S. (2002). Experimental and
simulation of the distributed radiative and convective numerical studies on the heterogeneity of crop
climate within a cropped greenhouse, Renewable transpiration in a plastic tunnel, Computers and
Energy, Vol. 43, pp. 111-129. DOI: Electronics in Agriculture, Vol. 34, No. 1-3, pp. 173-
10.1016/j.renene.2011.12.003 190. DOI: 10.1016/S0168-1699(01)00186-7
[16] Mesmoudi K., Soudani A., Zitouni B., Bournet P.E., [29] Molina-Aiz F.D., Valera D.L., Pena A.A., Alvarez
Serir L. (2010). Experimental study of the energy A.J., Gil J.A. (2006). Analysis of the effect of rollup
balance of an unheated greenhouse under hot and arid vent arrangement and wind speed on Almeria-type
climates:Study for the night period of winter season, greenhouse ventilation performance with

485
 computational fluid dynamics, Acta Hort, Vol. 719, Φ Phase function [Dimensionless]
pp. 173-180. DOI: 10.17660/ActaHortic.2006.719.17 Ω Solid angle, sr
[30] Pieters J.G., Deltour J.M. (2006). Performances of Χ Mean square error
greenhouses with the presence of condensation on
cladding materials, Journal of Agricultural Subscripts
Engineering Research, Vol. 68, No. 2, pp. 125-137.
DOI: 10.1006/jaer.1997.0187 λ depends on wavelength
φ transported quantity like: U, V, T, C, k, ϵ

NOMENCLATURE
APPENDIX
Iλ Spectral radiation intensity, Wm-2.sr-1
𝐼𝜆0 Black body intensity given by the Planck function, Table A. Beach of measurements and the accuracy of all the
W.m-2 sensors used in experiments
P pressure, Pa
Pr Prandtl number [Dimensionless] Measuring Manufacturer's
Sensor Uncertainty
range Name
qr Radiative flux, W Thermocouple 0°C to
Re Reynolds number [Dimensionless] Cu-Cs 100°C
±0.1 °C TC Online
S Surface, m2 0.8% to OAKTON Logger
Air humidity ±2%
R2 Correlation coefficient [Dimensionless] 100% Plus
Sr Radiation source term, J Air -39°C to
±0.2°C
OAKTON Logger
Sϕ Dimensionless source term temperature 60°C Plus
PAR silicon Kipp & Zonen,
T Temperature, K cell
0 Wm-2 to
±20 Wm-2 Cambell
U Axial component of velocity vector, ms-1 200 Wm-2
Scientific Ltd
V Radial component of velocity vector, ms-1 Model 100075,
Cup 0 ms-1 to
xi Space component in i- direction, m ±0.1ms-1 Climatronic
anemometers 60 ms-1
H Convectif heat transfer coeficient, Wm-2k-1 Corporation
Model 100076,
wind vane 0˚ to 360˚ •±2° Climatronic
Greek symbols
Corporation
-30°C to
λ Spectral absorption coefficient, m-1 platinum 80°C ±0.1 °C
Model MP601A,
Rotronic
 Dimensionless diffusion coefficient probes 0.8% to ±0.2 %
instrument crop
Θ Polar angle, rd 100%
Λ Wavelength, m Anemometer 0.01 ms-1 Cambell Scientific
±0.03 ms-1
Ρ Density, kg.m-3 with hot wire to 1 ms-1 Ltd
Kipp & Zonen,
Σ Stefan-Boltzmann constant, σ =5.672 10-8 W.m-2K-4 10Wm-2 to
Pyranometer ±20 Wm-2 Cambell
σs Scattering coefficient, m-1 500Wm-2
Scientific Ltd
Φ Azimuthal angle, rd
ϕ Dimensionless concentration of the transported
quantity

486