60
90
120 150 180 210 240
Distance along the screenhouse (m)
Distance along the screenhouse (m)
(a)
(b)
Fig. 10. Comparison of wind velocities between screenhouses with the same screen (S1 and
S2) and screen on the side (S1) and screen on the roof (S2). (a) With crop effect simulated. (b) Without crop effect simulated.
48
Applied Computational Fluid Dynamics
3.7 Solar radiation and temperature
Some studies have used solar radiation and transpiration models based on the heat and
water balances of the crop, to investigate the distributions of air temperature and humidity and also the interactions between the crop and the air, in addition to the airflow distribution (Sase, 2006). According to Tablada (2005), the factor of solar protection plays a crucial role in maintaining stable thermal conditions indoors, even if the outside air temperature is higher.
The slightly higher air speed on the top floor is insignificant in view of reducing the
negative effect of the solar radiation over the roof and facade. The temperature of the
greenhouse cover is an essential parameter needed for any analysis of energy transferred in
the greenhouse. A sub-model developed by Impron et al. (2007) calculated the transmission of radiation through the greenhouse, including the reduction of NIR transmission through
the roof. Tong et al. (2009) developed a numerical model to determine time-dependent temperature distributions based on hourly measured data for solar radiation, indoor air, soil and outside temperature, taking into account variable solar radiation and natural convection inside the greenhouse during the winter in northern China.
3.8 Temperature and air exchange
The effect of solar and thermal radiation is often taken into account by setting specific wall or heat fluxes at the physical boundaries of the greenhouse. Radiation transfer within the
crop itself is still the major concern since it determines the two main physiological crop
processes: transpiration and photosynthesis. This challenge is now launched and will
probably receive more attention within the next few years (Bournet and Boulard, 2010).
Pontikakos et al. (2006) analyzed data obtained from a CFD model, and showed that the external boundary temperature is a critical parameter in the pattern of internal greenhouse
temperatures and that for specific external temperatures and wind directions, airspeed
becomes the crucial parameter. According to Molina et al. (2006), opening vents affect the air flow, the ventilation rate and the air temperature distribution in a greenhouse; where the
mean air temperature at the middle varied from 28.2 to 32.9ºC with an outside air
temperature of 26ºC, there were regions inside the greenhouse that were 13ºC warmer than
the outside air. Nebbali et al. (2006) used a semi-analytical method to determine the ground temperature profile from weather parameters and other characteristics, to help in evaluating heat flux exchange between the surface and the air. Rico-García et al. (2008) showed that ventilation in greenhouses due to the temperature effect produces high air exchange rates;
however, those air patterns occur near the openings, causing almost no air exchange in the
central zone of the greenhouse due to a stagnant effect that reduces the wind effect
throughout the greenhouse. In agreement with the results of Majdoubi et al. (2009), convection and radiation are the dominant forms of heat transfer. The measurements show
that the difference between the air temperature inside and outside the greenhouse is
strongly linked to solar radiation and secondly to wind speed. However, Chow and Hold
(2010) obtained the following conclusions from studying buoyancy forces from thermal
gradients:
a. Thermal radiation without air involvement changes air temperature distribution by
radiating upper zone thermal energy in the wall towards the lower zone wall, which
then affects air temperature through conduction and convection;
b. The inclusion of air absorption increases the effect of radioactive thermal redistribution by allowing air to absorb and radiate heat, reducing temperature gradients further;
Advances in Computational Fluid Dynamics Applied to the Greenhouse Environment
49
c. Thermal boundary conditions and heat loads affect the predicted absolute temperature
bounds, but do not affect the temperature distribution.
Radiation conditions play an important role in redistributing heat. Atmospheric conditions,
especially relative humidity, are important for the calculation of radiation and heat transfer.
Flores-Velázquez (2010) recently found that without overhead natural ventilation, it is
possible to find linear relationships between temperature increase and the length of the
greenhouse respectively, and the fan power that determines the slope of the regression line.
3.9 Turbulence and buoyancy
As computing power has increased the complexity and sophistication of CFD models also
have increased. According to Norton and Sun (2006), the standard k-ε turbulence model
commonly used in CFD models for greenhouses, in some cases provides inadequate results,
and the choice of turbulence models must be based on the phenomena involved in the
simulation. Different turbulence models give rise to differences in speed, temperature and
humidity patterns, confirming the importance of choosing the model that most closely
matches the actual conditions of turbulence (Roy and Boulard, 2005). Teitel and Tanny
(2005) showed that the output of the turbulent heat flux is mainly due to cold air entering
the greenhouse, which produces hot and cold eddies coming in and out the greenhouse. Roy
and Boulard (2005) showed that the effects of wind direction on climate parameters inside
the greenhouse are usually simulated by using different turbulence models available, to
determine the energy balance between the flow of perspiration and the flow of radiation.
Under ventilation parameters based on Bernoulli's theorem, Majdoubi et al. (2007), showed that bad ventilation performance is not a result of the low value of the greenhouse
wind−related ventilation efficiency coefficient, but rather, that the low rate of discharge due to pressure drop in air flow is generated both by the use of anti-insect screens with small
openings as an obstruction due to the orientation of the rows of crops. Moreover, Rouboa
and Monteiro (2007) note that the RNG turbulence model is best suited to simulate
microclimates in arc-shaped greenhouses.
According to Baxevanou et al. (2007), the circulation of air buoyancy effect shows the importance of internal temperature gradients, forced convection resulting from natural
ventilation predominates. Rico-García et al. (2008) found that applying temperatures as the main driven forces for the buoyancy effect provides a simple way to study ventilation and
inner air patterns. Vera et al. (2010a) observed that differences in temperature and ventilation rates strongly influence the movement of air, pushing it through openings where
space is colder, while creating rising air currents when it is hot. Majdoubi et al. (2009) showed that the buoyancy forces induced by air temperature and increased humidity result
in loops of air between the crop and the roof windows, which in turn tend to accelerate the
pace of removal of heat and water vapor, enhancing indoor climate. Fidaros et al. (2010) studied turbulence in Greek greenhouses and found that external temperature variation is
very important because internal temperature is determined by convection induced by the
input current. The housing area had a higher circulation in the center of the greenhouse near the deck and in the corners of the ground, where the effect of the input current is weak.
Defraeye et al. (2010) used a RANS turbulence model in CFD simulations to evaluate heat transfer by forced convection at the surface of a cube immersed in a turbulent boundary
layer for applications in the atmospheric boundary layer (ABL), where wind speed is not
disturbed at a height of 10 m. In a study of airfoil wakes, three turbulence models were
simulated by Roberts and Cui (2010); the Reynolds Stress Model (RSM) is superior over the
50
Applied Computational Fluid Dynamics
k-ε model, and when a time-dependent solution is necessary, Large Eddy Simulation (LES)
is the desired option. However, LES does require the airfoil geometry to be included in the
domain because it performs poorly when given only inlet velocities, turbulence kinetic
energy and eddy dissipation at the trailing edge of the airfoil. According to Bournet and
Boulard (2010), although they have been used for a long time in both the agriculture and
environment studies, less empirical approaches to turbulence based on the use of LES have
never been applied to greenhouse climate modeling and might perhaps be used to look for a
solution to this complex situation.
Inside buildings, it is difficult to maintain a thermally stratified space with low ceilings, such as in offices and houses. Vera et al. (2010b) studied buoyancy in enclosed spaces, drawing the following conclusions:
a. Rising air currents and the exchange of humidity are closely related to the temperature
difference between the lower and upper space. Low temperature in the upper space
promotes the exchange of humidity and air flow through the opening; the hotter you
are, the greater the restriction of air and humidity transport.
b. The existence of upward air currents when the space is warmer than the bottom is
caused by local conditions such as non-uniform temperature distributions in the upper
space and convective warm currents of the base and the humidity source.
c. Compared with conditions without mechanical ventilation, ventilation severely restricts
the flow of air through the opening.
The main difficulty in the choice of the model is that greenhouse systems cover a range of
length and velocity scales that generally require different modeling approaches (Bournet
and Boulard (2010).
3.10 Incorporation crop effects and crop modeling
The effect of plants on greenhouse ventilation has also been studied in the past. For instance; Bournet et al. (2007), based on studies by Nederhoff (1985) and Lee and Short (1998), assumed that a crop of 90 cm high and low density decreases between 12 and 15%
greenhouse ventilation. Dayan et al. (2004) built a representative model of a greenhouse for three vertical segments, horizontally oriented to the directions of energy and vapor transfer between the segments containing plants, considering the external weather. They concluded
that Representative Plant Temperatures (RPTs) can be calculated instead of measured. Roy
and Boulard (2005) developed a 3D CFD model for the characterization of climatic
conditions in a greenhouse, incorporating five rows of ripe tomatoes as a porous medium
where the buoyancy, heat and moisture transfer between the crop and air flow inside were
considered. The heat and moisture transfer coefficients are deduced from the characteristics of the laminar boundary layer of the leaf, which are calculated with the velocity of flow in the crop. Khaoua et al. (2006) found that under external conditions of 1 ms-1 air velocity and 30° of temperature, wind speed at crops’ height varies according to the modalities of
ventilation from the windward 0.1 and 0.5 ms-1 for the leeward side, while temperature
differences ranged from 2.0 to 6.1 ° C. In a study with tomatoes, Majdoubi et al. (2007) found that crop rows oriented perpendicular to air movement reduce the rate of airflow through
the cultivation in a greenhouse by 50%. According to Baeza et al. (2008), a greenhouse with natural ventilation efficiency must combine an enough number of air changes to remove excess of heat, with good circulation of air through the crop. The effect of the crop was
evaluated by Impron et al. (2007) using a sub-model to determine its effects on ventilation, the properties of the cover, and crop transpiration. In agreement with Kruger and Pretorius
Advances in Computational Fluid Dynamics Applied to the Greenhouse Environment
51
(2007), the temperature and velocity at the plant level are influenced by the arrangement
and number of windows. A study carried out by Sapounas et al. (2007) simulated a tomato crop as a porous medium, taking into account the addition of buoyancy to develop a model
of the pressure drop of air flow due to the crop. The model depended on the area leaf stage
of growth and cultivation, under the RANS turbulence model together with the RNG k-ε
turbulence model. The results, validated with experimental measurements obtained at 1.2 m
inside the canopy; show that the evaporative cooling system is effective with numerical
parameters, providing a useful tool to improve system efficiency. A study performed by
Roy et al. (2008) on leaf level through an experimental setup based on Münger cells measured the temperature, relative humidity and different heat flows to the leaves of
soybeans, obtaining minimum stomatal resistance values ranging from 66 to 200 sm-1.
Teitel et al. (2008b) built a small-scale model and found that wind direction significantly affects the ventilation rate and temperature distribution in crops. A study by von Elsner et al. (2008) on the effect of near-infrared (NIR) reflecting pigments in microclimate and plant growth found that a temperature drop up to 4 ° C in a young crop is the result of a 18%
reduction in the transmission of global radiation in spring. At the same time, during the
rainy season, minimizing transpiration differences in temperature and shading reduces
water requirements in the plants, and they observed parthenocarpic fruit rot and yield-
reducing crop. In a tunnel-type greenhouse, a tomato crop was modeled by Bartzanas et al.
(2008) by designing a porous medium, where they emphasize the influence of the heating
system on greenhouse microclimate. The climatic behavior of the rows of the tomato crop is
taken into account using external user defined functions (Baxevanou et al. , 2007). According to Majdoubi et al. (2009), reorienting crop rows in simple ways improved climatic
conditions. Endalew et al. (2009) performed CFD modeling of a plant with leaves and branches of the canopy, using turbulent energy equations in porous sub-domains created
around the branches. Fidaros et al. (2010) simulated a greenhouse tomato crop as a porous medium so as to model radiation transport by discrete ordinates (DO). According to Teitel et al. (2010a), when applying the porous medium approach, the Forchheimer equation is often used, which gives rise to erroneous results with respect to the pressure drop through
screens. An alternative way to calculate it through several panels of porous media used to
simulate screens with realistic geometries. Moreover, the crop exerts a mechanical strain
(drag force) on the flow just above but also interacts through the transpiration process with the temperature and humidity distributions (Bournet and Boulard, 2010). A simple model of
transpiration of a crop was developed by Sun et al. (2010), who related it to the
characteristics of ventilation in a greenhouse in eastern China, obtaining a good
approximation. In general, there have been enormous efforts devoted to the analysis of ventilation in greenhouses (Norton, 2007); each new study provides new elements not only
in the movement of air in the greenhouse but also in the forms it takes due to interactions
occurring in the environment, such as position, shape and size of windows, and (one of the
most important), the presence of a crop (Flores-Velázquez, 2010).
3.11 Humidity
Roy and Boulard (2005) simulated wind directions of 0 °, 45 ° and 90 ° with respect to the
orientation of the greenhouse ridge to determine wind speed, temperature and humidity
distributions inside the greenhouse, getting a good approximation for the humidity. In
agreement with Demrati et al. (2007), models allow estimation, with better accuracy, of water requirements for a banana crop under cover and improved water saving in regions
52
Applied Computational Fluid Dynamics
where water is the main limiting factor for agriculture. Roy et al. (2008) studied moisture on the surface of leaves at low light levels; crop transpiration and air flow were integrated into a single parameter model of leaf stomatal response to air flow and radiation. Campen (2008)
showed that climate through a ventilation system is more homogeneous and the control is
more efficient than with the conventional method of steam extraction. Dehumidifiers and
cooling reduce the overall difference in humidity between the middle and lower areas of a
greenhouse, as demonstrated by Kim (2008) using a 3D model could identify the
heterogeneous distribution of relative humidity in a greenhouse. According to Majdobi et al.
(2009), an increase in air temperature precedes a more moderate increase in specific
humidity.
4. Main results of CFD models in greenhouses
Many CFD studies are focused on defining the conditions for a suitable environment. There
has been less work on automation and control variables. Investigations that seek for a
greater understanding of the interactions among climatic variables are increasing. Studies
such as those of Hooff, 2010; Teittel, 2010 and Fidaros, 2010, evaluating geometries, have
increased in the last year. Figure 11 shows the frequency of climatic variables studied during the period from the year 2005 to 2009 in the studies of CFD models in greenhouses.
Fig. 11. Frequency of climatic variables analyzed by CFD models applied to greenhoses.
Most studies show multi-variable relationships, of which temperature and air flow are
predominant. Humidity has been linked to temperature, while there are still few CO2
distribution models. Solar radiation is the subject of investigations that evaluate housing, and is also related to the temperature in simulations with a greater degree of realism.
Studies to determine the influence of windward and leeward wind direction indicate that
roof vents are of great importance for air renewal, where aperture settings that maximize the Advances in Computational Fluid Dynamics Applied to the Greenhouse Environment
53
number of air renewals are obtained only with windows open to the windward roof
(Bournet et al. , 2007; Kacira, 2008 and Majdoui, 2009). However, the combination of openings to the windward and leeward sides homogenizes the temperature inside the greenhouse better
(Bournet and Khaoua, 2007). There have been many studies to determine a kind of optimal
geometric design of greenhouses that encourage improvements in weather conditions and the
use of new technologies such as monitoring systems in real time, allowing improvements in
automation using Web technology (Pontikakos, 2005). Other studies have focused on the
evaluation of misting systems (Kim et al. , 2007; Gázquez et al. 2008), forced ventilation (Dayan et al. , 2004, Baeza et al. , 2008 and Hughes and Abdul 2010) looking for energy savings. As simulation technology and computing power have improved, accuracy and realism in research
based on CFD models, has increased as well, by defining more detailed models and by the use
of textures that define the materials of the facilities.
The use of insect-proof screens in commercial greenhouses is very important as a means of
crop protection; even though they reduce natural ventilation, by this, there have been many
research efforts to reduce its negative influence (Kittas et al. , 2005; Harmanto et al. , 2006; Majdoubi et al. , 2007; Teitel et al. , 2008a). These studies tested different designs in size of the box and tilt and determined the most affected areas within the greenhouse, where the use of
porous media allowed its CFD simulation.
Several studies have investigated the influence of solar radiation on temperature and
relative humidity (Tablada et al. , 2005; Impron et al. , 2007, Tong et al. , 2009), and the result in crop response (Baxevanou et al. , 2007). Other studies evaluated the use of pigments (Elsner et al. , 2008) taking into account the convection, and thermal gradients.
Most of the recent studies developed 3D CFD models, some of which reported the use of
models of turbulence and buoyancy, which appear more often during the past two years
(Fidaros, 2010; Defraeye, 2010; Norton, 2010; Majdoubi, 2009). By taking into account
turbulence, CDF models can make simulations more accurate, in turn increasing the
processing and memory requirements for computing resources. Norton and Sun (2006) and
Roy and Boulard (2005) discuss the importance of choosing the turbulence model that best
meets the conditions of the study. Moreover, the concept of buoyancy appears frequently in
order to incorporate the effects of growing space on the air flow and temperature gradients
into the models (Figure 12).
Many studies consider the growing space, some of which are designed to measure
phenomena based on their influence on the development and crop yield. Other studies are
focused on the influence of crops on the other elements, such as temperature, relative
humidity, CO2 concentration and air flow, where it is necessary to model the space
occupied by the crop by using porous media approach (Fidaros et al ., 2010). Other
investigations measure biological phenomena such as evapotranspiration and
Photosynthetically Active solar Radiation (PAR) by using indirect measures of climatic
variables (Baxevanou, 2007; Sun, 2008). However, some studies do not mention an
experimental phase aimed at validating the numerical model. In studies of air flow, the
experimental methods mostly used are scaled models and unidirectional anemometry; the
tracer gas technique is used less often, as well as three-dimensional anemometry, which is
considerably more expensive. Studies that have used new methods to assess ventilation
systems, such as those by Lu (2009), Molina (2010), Endalew (2009), van Henten (2008),
Mikulka (2010) and Defraeye (2010), have been increasing in the past three years. The main
question is the validation of these studies because they mainly concern to real scale
greenhouses, whereas the measurements and characterizations have merely been done on
scale models.
54
Applied Computational Fluid Dynamics
Fig. 12. Degree of realism and accuracy in the CFD models for greenhouse climate.
CFD modeling is an area of knowledge that in recent years has developed enormously
through the development of software and hardware, which has contributed to research on
natural ventilation a greater understanding of the interactions between the variables that
make up the climate inside greenhouses. In the past five years, CFD simulation has become
increasingly realistic and detailed, obtaining more accurate solutions. However, their use
requires depth and extensive knowledge of climatic variables, fluid dynamics and
turbulence. Simulating more accurately requires more processing power, so research tends
to use CFD models together with other tools. Further studies are required to incorporate
more realistic crops beyond a porous medium, taking into account the role of gas exchange,
which is necessary for an understanding of the physiology and phenology of crops. There is
still a need to develop high-precision systems in greenhouses, and CFD is a powerful tool
for defining parameters with high precision, in order to control better the greenhouse
environment.
5. Validation procedures for CFD models of greenhouse environment
5.1 Models and experimental validation
According to Sase (2006), recent progresses in CFD techniques have accelerated a more
detailed analysis of air movement in combination with verification tests. However, studies
in this area are required in order to address the detailed design of each element involved in the greenhouse climate, highlighting the difficulty involved in the analysis of air movement inside a greenhouse (Flores-Velázquez, 2010). The quality of the CFD models predictions is
often evaluated from the agreement with experimental data. Nevertheless, no standard
procedure exists yet in order to properly assess the accuracy of the simulations, and the type of comparison often differs from one study to the next (Bournet and Boulard, 2010). Figure
13 summarizes the main approaches used in the recent past to validate the CFD models of
the greenhouse environment.
