Analysis of the role of sidewall vents on buoyancy-driven natural ventilation in parral-type greenhouses with and without insect screens using computational fluid dynamics

biosystems engineering 104 (2009) 86–96

Available at www.sciencedirect.com

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

Research Paper: SEdStructures and Environment

Analysis of the role of sidewall vents on buoyancy-driven natural ventilation in parral-type greenhouses with and without insect screens using computational fluid dynamics Esteban J. Baezaa,*, Jero´nimo J. Pe´rez-Parraa, Juan I. Monterob, Bernard J. Baileya, Juan C. Lo´peza, Juan C. Ga´zqueza a

Fundacio´n Cajamar, Estacio´n Experimental, Paraje Las Palmerillas 25, 04710 El Ejido, Spain Institut de Recerca i tecnologı´a Agroalimentaries, 08348 Cabrils, Spain

b

article info Computational fluid dynamics (CFD) was used to study the effect of greenhouse sidewall Article history:

vents on buoyancy-driven natural ventilation, which is the most difficult situation for

Received 12 August 2008

greenhouse cooling. A CFD model was validated and then used to compare roof ventilation

Received in revised form

with combined roof and sidewall vents. The effects of the distance between opposing

19 March 2009

sidewall vents and the presence of anti-insect screens were investigated and quantified. In

Accepted 10 April 2009

a 20 span greenhouse with a distance of 152 m between the sidewalls, combining roof and

Published online 7 July 2009

sidewall vents gave a ventilation rate per unit ground area that was twice that with only roof vents. In a 3 span greenhouse with 22.8 m between sidewalls but with the same roof vent area per unit ground area, seven times more ventilation was obtained with combined ventilation compared to only roof ventilation. In the 7 and 20 span greenhouses with roof ventilation, 79% and 48.3%, respectively, of the cross sectional area had a temperature difference between inside and outside the greenhouse greater than 4  C, whereas with combined ventilation, these areas were 23.4% and 36.1%. In the latter case, these hot areas were located in the centre of the houses. These results show that, with buoyancy-driven ventilation, the contribution of the sidewall vents is important even for quite large greenhouses but is more critical for greenhouses with a small number of spans. Numerical simulations with an anti-insect screen having a porosity of 25% showed that the air exchange rate with combined ventilation was reduced by 77–87%, depending on greenhouse size. These reductions were much larger than those obtained for wind-driven ventilation. It was concluded that, to maximise ventilation when wind speeds are low, combined roof and sidewall ventilation should be used. Also, large greenhouses should be relatively narrow to make better use of the sidewall vents. ª 2009 IAgrE. Published by Elsevier Ltd. All rights reserved.

1.

Introduction

A major requirement of greenhouses used for horticultural production in Mediterranean climates is effective ventilation

since cooling requirements are high during at least part of most crop production cycles. Consequently, the greenhouses tend to be long and narrow, covering small areas with vents along the sidewalls. There are economic advantages in

* Corresponding author. E-mail address: [email protected] (E.J. Baeza). 1537-5110/$ – see front matter ª 2009 IAgrE. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.biosystemseng.2009.04.008

biosystems engineering 104 (2009) 86–96

Nomenclature F G SF U r T DT b k 3 m K

concentration of transported quantity diffusion coefficient (m2 s1) source term air velocity (m s1) density of air (kg m3) temperature (K) temperature difference (K) thermal expansion coefficient of air (K1) turbulent kinetic energy (m2 s2) turbulence dissipation rate (m2 s3) kinematic viscosity of air (m2 s1) permeability of the medium (m2)

building larger greenhouses. The capital cost per unit covered area is reduced, and in the winter the heat loss per unit is reduced, thereby reducing heating costs (Vollebregt & van de Braak, 1995). The average greenhouse area in The Netherlands has increased from less than 1 ha in 1998 to 1.5 in 2003 (Bunschoten & Pierik, 2003), while in Almerı´a it is around 0.7 ha (Ferna´ndez & Pe´rez-Parra, 2004). Large greenhouses generally do not have sidewall vents, and natural ventilation takes place exclusively through roof vents (De Jong, 1990; von Elsner et al., 2000). The tendency to build larger greenhouses without sidewall vents also exists in the Mediterranean region. However, these greenhouses often have an insufficient area of roof vents for the climatic conditions, and the vents are almost invariably covered with low porosity screens to exclude insects. Thus, growers have had problems controlling excess temperature and humidity and also their greenhouses tend to experience low CO2 concentrations resulting from suboptimal air exchange rates (Montero et al., 1985; Lagier, 1990; Castilla, 1994; Pe´rez-Parra et al., 2004). This situation is most extreme when wind speeds are low and natural ventilation is driven predominantly by buoyancy forces. For greenhouses in coastal areas, this situation occurs only in the hours before noon during warm months. Inland, however, this situation can be common during the middle of the day when temperatures are higher. There have been numerous studies of roof and sidewall ventilation under conditions where wind plays an important role in promoting air flow through greenhouse ventilators. Investigations by Kamaruddin (1999), Montero & Anton (2000), Montero et al. (2001), Pe´rez-Parra et al. (2004), Kacira et al. (2004) and Bartzanas et al. (2004) showed that, under these conditions, combined roof and sidewall vents provide significantly higher air exchange rates than those obtained with roof ventilation only. This result is especially important for narrow greenhouses, as shown by the theoretical calculations of Pe´rez-Parra (2002) who estimated that 50 m was the maximum distance between opposite sidewall vents beyond which air exchange was almost exclusively dependent on roof ventilation. However, there have been few experimental studies dealing with the role of sidewall vents in multispan greenhouses under calm conditions or when the wind speed is very low (<2 m s1) and ventilation is buoyancy-driven. Data for

Cf a Sr Cd F Re d

87

non-linear momentum loss coefficient porosity global radiation (W m-2) discharge coefficient loss coefficient Reynolds number diameter (m)

Subscripts o reference state s screen t screened ventilator v ventilator without screening

single span greenhouses was provided by Sase et al. (1984), who used scale models in a wind tunnel and also analysed thermally driven ventilation. Oca et al. (1999) presented a new laboratory technique to simulate natural ventilation by thermal effects using a tunnel greenhouse with two continuous roof and sidewall vents, but the paper dealt only with the validation of the experimental values with theoretical predictions. Numerical analysis using CFD was used by Mistriotis et al. (1997) in a systematic analysis of natural ventilation in greenhouses under no wind and low wind speed conditions. This analysis confirmed the importance of sidewall ventilators for efficient thermally driven ventilation. However, only greenhouses of up to four spans width were studied. Kacira et al. (1997) performed CFD simulations of ventilation in saw-tooth multispan greenhouses with two and four spans and observed reductions in ventilation rates of 80– 90% at low wind speeds (0.5–2 m s1) when the windward sidewall vent was closed. This paper uses CFD simulations to evaluate the effect of sidewall vents on buoyancy-driven natural ventilation in multispan parral type greenhouses. The effects of using combined roof and sidewall vents compared to roof vents alone on ventilation rates, temperature fields and air speed profiles were investigated. Also studied were the influences of greenhouse size, height of the vents in the sidewall, and the presence of insect screens.

2.

Material and methods

2.1.

Theoretical background

A CFD code (v.6.2.16, Fluent, Ansys-Fluent Inc., Sheffield, UK) was used to perform the simulations to study the role of sidewall vents on buoyancy-driven greenhouse natural ventilation. The program uses the finite volume method to numerically solve the Navier–Stokes equations, i.e., the mass, energy and momentum balances, permitting the calculation of air velocity and temperature fields (Patankar, 1980): X v  vF  vF X v ðUi FÞ ¼ GF þ SF þ vt vXi vXi vXi i¼1 i¼1

(1)

In order to account for gravity forces due to air density (temperature) changes, the Boussinesq hypothesis was used for

88

biosystems engineering 104 (2009) 86–96

the whole of the computational domain. This method treats density as a constant value in all the solved equations, except for the gravity term (thermal effect) of the momentum equation: ðr  r0 Þgz  r0 bðT  T0 Þg

(2)

where the subscript defines a reference state. The Boussinesq approach is valid if the density (temperature) gradients occurring in the computational domain are not too large; that is, if b(T  T0) << 1. In our case, with a naturally ventilated greenhouse, the temperature differences are never very large (<20  C). Therefore, the Boussinesq simplification can be applied. Using this approach, a better convergence in natural convection problems is achieved compared with treating density as a function of temperature. To account for the pressure–velocity coupling and turbulence, the SIMPLE algorithm and the RNG k–3 model (Launder & Spalding, 1974) were used. The RNG k–3 turbulence model has been used with success in CFD simulation studies of greenhouse natural buoyancy-driven ventilation, generally providing good agreement with the experimental results (Mistriotis et al., 1997). The body force-weighted model was used for the pressure discretisation. The crop was not included in order to simulate the most unfavourable conditions, which occur when the crop has just been transplanted and intercepts little solar radiation, and transpiration is almost negligible. Simulations were performed with and without an antiinsect screen on the vents. The screen was treated as a porous media, with the source term governed by the Darcy-Forcheimer equation:

the length of the greenhouse in the windward and leeward directions and 10 times the height of the ridge, ensuring that the domain boundaries had a negligible effect on the flow in the vicinity of the greenhouse (Richards & Hoxey, 1992). The simulation models were meshed with a squared ‘‘pave’’ mesh scheme with a cell size of 0.2 m inside the greenhouse and 0.4 m in the outside domain. The cell density was increased in the areas close to the greenhouse vents (0.1 m), where larger physical gradients were expected to occur (Fig. 1). As the floor is responsible for emitting most of the energy that heats the inside air, a boundary layer containing a much higher cell density was attached the greenhouse floor. In order to simplify the calculations, a homogeneous temperature condition (330 K) was imposed on the soil (with sand mulch). The greenhouse walls were considered adiabatic. The soil temperatures were measured and then averaged using infrared temperature sensors and correlated to an average summer day in an empty greenhouse at the Experimental Station of the Cajamar Foundation in Almerı´a, Spain. Table 1 shows the boundary conditions used in the simulations.

3.

Results and discussion

3.1. Validation of buoyancy-driven natural ventilation simulations

Values of K and Cf for the screen were obtained from the following equations (Miguel, 1998):

The availability of experimental data corresponding to temperature patterns inside greenhouses that are ventilated exclusively by buoyancy forces caused by thermal differences is very scarce. Oca (1996) and Oca et al. (1999) conducted experiments on laboratory-scale models and field measurements to study buoyancy-driven natural ventilation in tunnel greenhouses, and their results were used to validate the CFD model.

K ¼ 3:44  109 a1:6

3.1.1.

 SF ¼ 

 Cf ¼

2.2.

Cf m U þ rpffiffiffiffiU2 K K

4; 3  102 a2;13

 (3)

(4)

2 (5)

Greenhouse simulation models

A two-dimensional (2D) simulation model was created for an experimental parral-type greenhouse 23.2 m long with five 7.6-m wide spans. The ridge height was 4.4 m and the gutter height was 3.6 m. The greenhouse had one roof flap vent per span with dimensions 0.73  8.35 m. All roof vents were oriented in the same direction. There were rolling vents along the top of the two 23.2 m sidewalls, each with a maximum opening of 1.2 m. Taking the 5 span model as the reference model, additional models were created with 3, 7, 10, 15, and 20 spans, keeping the same dimensions for the spans and vents and also their number and distribution per span. The presence of an anti-insect screen with a porosity of 25% covering the vents was also included in each model.

2.3.

Computational domain and boundary conditions

The computational domain for the different greenhouse models was created with the following dimensions: five times

Validation with scale model laboratory results

The scale models were ventilated through a simple rolling vent located on the lower part of one of the greenhouse sidewalls and a rolling roof vent located near the ridge (Fig. 1). Both the combined ventilation (roof plus sidewall vents) and the sidewall-only open vent configurations were studied in the laboratory experiments. The laboratory technique consisted of immersing the inverted greenhouse scale model in a tank of water and injecting a solution of salt water and black dye through the greenhouse floor in order to produce density differences that simulated the buoyancy flux due to heating of the greenhouse air. The flow was recorded with a video camera. The images were processed to relate pixel intensity to density, and scaling laws were used to calculate the temperature increments associated with the relative density increments. Oca (1996) and Oca et al. (1999), carried out laboratory experiments in a scale model greenhouse for soil heat fluxes of 200, 350, and 500 W m2, an outside temperature of 300 K, and adiabatic greenhouse walls (no heat flux through the greenhouse walls). Two ventilator configurations were considered: ventilation through a single sidewall vent and the combination of a single sidewall vent and a roof vent. For comparison purposes, 2D CFD simulations were performed for the same greenhouse geometry with the same boundary

89

biosystems engineering 104 (2009) 86–96

Fig. 1 – 2D CFD model of the scale model greenhouse used for the validations.

conditions: identical soil fluxes, the same outside air temperature, adiabatic walls, and no wind conditions. Fig. 2(a,b) compares the average temperature difference between the inside and outside air for the cross section obtained from the laboratory experiments and CFD simulations. For the sidewall ventilation configuration (Fig. 2a) the agreement between the experimental data and simulations showed minimal differences for the three tested soil heat fluxes. For the combined ventilation configuration (Fig. 2b) experimental data and CFD simulations followed the same trend, but they differed by about 2 K. The discrepancy between the two methods increased as the soil heat flux increased (Fig. 2b).

Validation with field measurements

The field experiment was carried out in a 3-m wide tunnel greenhouse located inside of a 9-m wide greenhouse, in order to achieve zero wind conditions and guarantee that all the air exchange inside the 3-m wide greenhouse was due to the thermal effect (Oca, 1996). Three sets of measurements were carried out for the same greenhouse configuration. Measurements were made of the air temperature (14 solar-shielded but

a

35 30

Temperature difference °C

3.1.2.

non-aspirated Resistance Temperature Detectors (RTD) sensors) over the central cross section of the experimental tunnel greenhouse (Fig. 3) for the sidewall-only ventilation configuration used in the previous simulations (Oca, personal communication, 1996). The temperature sensors were not aspirated to avoid altering the natural temperature distribution created by thermal effects. To reduce experimental errors, the measured values were corrected to eliminate the effect of solar heating by using the adjustment proposed by AbdelGhany et al. (2006). At the same time, the temperature outside the tunnel greenhouse (aspirated psychrometer), the soil surface

25 20 15 10

Experiments

5

CFD

0 0

Table 1 – Boundary conditions used in the simulations (case studies) Value

Wind velocity

0 m s1

Outside air properties Temperature (K) Density (kg m3) Specific heat (J kg1 K1) Thermal conductivity (W m1 K1) Viscosity (kg m1 s1) Thermal expansion coefficient (K1)

303 1.16 1006.5 0.0263 1.864e5 0.003

Greenhouse soil properties Temperature (K) Density (kg m3) Specific heat (J kg1 K1) Thermal conductivity (W m1 K1)

200

300

400

b

600

10 8 6 4 Experiments

2

CFD

0 0

100

200

300

400

Heat flux from the soil 330 1600 295 0.35

500

Heat flux from the soil Wm-2

Temperature difference °C

Parameters

100

500

600

Wm-2

Fig. 2 – Temperature differences obtained for different heat fluxes from the soil for the only sidewall vent configuration (a) and for the combined sidewall-roof ventilation configuration (b).

90

biosystems engineering 104 (2009) 86–96

1.75 m

44.9 °C

1m

45.7 °C

0.75 m

45.5 °C

0.5 m

45.3 °C

45.5 °C 45.8 °C

46.4 °C 46.5 °C

45.3 °C

44.8 °C

0.25 m

44.3 °C SR=550 W m-2

0m

x = 0.75 m

CFD (°C) x= 0,75 m

1,5

45.9 °C

Height (m)

45.2 °C

1.25 m

Measured-corrected (°C) x=0,75 m

1,75

46.9 °C

1.5 m

2

a

Text= 38.3 °C

2m

SR=465 W m-2

SR=576 W m-2

x=1.5 m

x=2.25 m

1,25 1 0,75 0,5

TS= 58.8 °C

0,25

Fig. 3 – Schematic diagram of the central cross section of the scale model greenhouse with the three measuredcorrected vertical air temperature profiles, the soil temperature (TS), the outside temperature (Text) and the solar radiation values (SR).

0

0

1

2

3

4

5

6

7

8

9

10

7

8

9

10

7

8

9

10

Temperature (°C) 2

b

Measured-corrected (°C) x=1,5 m CFD (°C) x=1,5 m

1,75

Height (m)

1,5 1,25 1 0,75 0,5 0,25 0 0

1

2

3

4

5

6

Temperature (°C) 2

c

Measured-corrected (°C) x=2,25 CFD (°C) x=2,25

1,75 1,5

Height (m)

temperature (RTD) and the global radiation were also measured and averaged for a period of 1 h around noon. This measurement enabled the most important heat fluxes (convection between greenhouse soil and air, and between cover and outside air) to be calculated. Table 2 shows the measurement conditions and main heat fluxes for the three experimental tests conducted by Oca (1996). A 2D CFD model of this greenhouse was created using the experimental measurements and heat fluxes presented in Table 2 as boundary conditions for the simulations. Fig. 4(a–c) presents the corrected and CFD simulated internal–external temperature differences. The two methods show a good coincidence, with a difference of less than 2  C. The best agreement was achieved for the profile in the centre of the greenhouse. On the whole, the simulated temperatures were slightly higher than the experimental values. The reason for this minor discrepancy could be that, in the simulation, the heat losses through the cover were not included. Considering the average temperature of the full cross section, good agreement was observed between the measured data and the CFD simulations, with differences of less than 1  C. The results of these laboratory and field measurement validations indicate that the 2D CFD model used is suitable for

1,25 1 0,75 0,5 0,25 0 0

1

2

3

4

5

6

Temperature (°C) Table 2 – Measurement conditions and heat fluxes during the first scale model experiments used as boundary conditions in the first set of validations Experiment

2

Rn (W m ) Hc (W m2) H (W m2) U (m s1) Text ( C)

E1

E2

E3

313 26 185 0 42

288 26 169 0 40.6

298 27 164 0 35.8

Rn: net radiation inside the scale model greenhouse, Hc: heat losses per unit surface through the cladding, H: heat flux that heats the inside air in each experiment, U: wind velocity outside the scale model greenhouse, Text: average ambient temperature outside the scale model greenhouse.

Fig. 4 – Experimental (AAA) and calculated (____) vertical air temperature profiles at distances of (a) 0.75 m, (b) 1.5 m and (c) 2.25 m from the sidewall vent. Only sidewall vent configuration.

simulating pure buoyancy-driven natural ventilation in parral multispan greenhouses.

3.2. Simulations of ventilation in parral-type multispan greenhouses 3.2.1.

Sidewall ventilation

The results of the simulations using combined sidewall and roof vents and only roof vents are presented in Table 3. It is

biosystems engineering 104 (2009) 86–96

Table 3 – Air exchange values (m3 sL1), obtained numerically for greenhouse models with 3, 5, 7, 10, 15 and 20 spans for roof and sidewall ventilation and for only roof ventilation Number of spans 3 5 7 10 15 20

Roof and sidewall ventilation 15.2 19.2 25.1 29.2 31.4 35.7

Number of spans 3 5 7 10 15 20

Roof ventilation 2.2 2.9 4.5 9.0 11.5 16.5

clear that using sidewall vents significantly increases ventilation rates compared only roof vents. Combined ventilation gave almost three times more ventilation than roof ventilation for the 15 span case and seven times more ventilation for the smaller, 3 span case. If the ventilation rates are normalised by the ventilator open area, the combined ventilation was between 1.4 and 2.5 times more efficient than pure roof ventilation. This result clearly shows that the sidewall plus roof vent configuration results in a better performance than roof vents alone. The simulated flows of ventilation air for the 5 span case with only roof vents and for the case with combined roof and sidewall vents are shown in Fig. 5. The colours of the flow paths indicate their temperature. With combined ventilation, the colder and denser external air entered almost exclusively through the sidewall vents and the warmer, less dense air leaves through the roof vents. When only the roof vents were open, however, some of the vents operated as inlets and others operated as outlets, thereby establishing a lower ventilation flow. In this situation, the flow of outside cold air

37 36 36 36 35 35 35 34 34 33 33 33 32 32 31 31 31 30 30

91

through the roof vents took place against the temperature gradient, which caused the hot air to ascend. This explains the low ventilation rates shown in Table 3. The influence of sidewall vents on ventilation rate is reflected in the temperature fields produced inside the greenhouse. The results of simulations of the inside–outside temperature difference fields for the 5 span greenhouse model are shown in Fig. 6. With roof vents (Fig. 6a), large temperature differences were found (>4  C) over almost all the greenhouse cross section, except in the fourth span, through which the outside colder air was entering. In this case, the temperatures were more than 4  C above the outside air temperature for over 32% of the greenhouse cross section, with a maximum temperature difference of 9  C. Opening the sidewall vents (Fig. 6b) dramatically changed the temperature pattern. As noted previously, a flow of cold air entered the greenhouse through both sidewall vents, producing a cooling effect which, for this width of greenhouse (38 m), covered almost the whole house. In this case, only 16% of the greenhouse cross section (near the centre of the greenhouse) had an inside–outside temperature difference equal to or exceeding 4  C. Almost all the roof vents evacuated the warmer, less dense air from inside the greenhouse. The location of the sidewall vents has been shown to have an important influence on ventilation rates. The theoretical models proposed for greenhouses with combined roof and sidewall vents (Bruce, 1978; Timmons and Baughman, 1981; Kittas et al., 1997) indicate that the pressure difference between roof and sidewall vents is proportional to the vertical distance between the centres of the two sets of vents. Therefore, increasing this distance will increase the ventilation rate. This increase can be achieved by increasing the height of the greenhouse ridge and/or placing the sidewall vents in a lower part of the sidewall. A simulation was performed for the 5 span model with the vents located at the bottom of the sidewall (distance between the centres of the sidewall and roof vents was 3.91 m, vs. 2.71 m with sidewall vents located in the standard position). The simulation showed that this arrangement resulted in a ventilation flow

a

b

Fig. 5 – Air flow paths in the vents of the 5 span greenhouse with roof ventilation only (a) and with combined roof plus sidewall ventilation (b); paths coloured by temperature (8C).

92

biosystems engineering 104 (2009) 86–96

Fig. 6 – Temperature difference field (8C) of the 5 spans greenhouse with roof ventilation (a) and with combined roof plus sidewall ventilation (b).

3.2.2.

Greenhouse size

Another important issue is how buoyancy-driven natural ventilation is affected by greenhouse size. This effect was studied using the simulation models with 3, 5, 7, 10, 15, and 20 spans, both with combined sidewall and roof vents and with only roof vents. For these models, the ratio of roof ventilator area to ground area covered remained constant, whereas the corresponding ratio for the sidewall vents decreased with increasing span number. Fig. 7 shows the how the ventilation rate varied with the number of spans. It is clear that, for combined ventilation, the ventilation rate decreased as the greenhouse size (and therefore the distance between the sidewall vents) increased. The data was fitted to a function of the type y ¼ a þ b/x0.5 (Fig. 7), where y is

the ventilation rate and x the number of spans of the greenhouse. A regression equation was used to estimate the ventilation rates for greenhouses with up to 50 spans. When the number of spans became very large, the ventilation rate tended towards an asymptotic value. This value was very close to the ventilation rate with only roof vents, which, as can be seen from Fig. 7, was almost independent of the number of spans and therefore of greenhouse size. This pronounced decrease agrees with the results reported by Kacira et al. (2004) for a similar vent configuration (sidewall vents and leeward roof vents) in a gothic multi-tunnel greenhouse with winddriven natural ventilation. This result shows the importance of the role of sidewall vents both with and without wind.

0,035

Vent rate per unit greenhouse area (m3 m-2s-1)

per unit greenhouse covered area of 0.0024 m3 m2 s1, which was only slightly higher (10%) than the value of 0.0022 m3 m2 s1 obtained when the vent was in the standard position. In greenhouses with a higher ridge (or with a larger temperature difference), it is possible for this increase in ventilation rate to be larger, justifying this new location of the sidewall vent. A greenhouse 2D model was created with a higher ridge (6 m) and the sidewall vents at the bottom of the sidewalls (the distance between the centres of the sidewall and roof vents was 5.51 m). This model gave an air flow rate of 0.0031 m3 m2 s1 (an increase of 41%). However, locating vents in the lower part of the sidewalls may not be acceptable in practice because, under windy conditions, air entering the greenhouse through these vents would impinge directly onto the crop, which could be undesirable, especially when the outside air is very warm and dry. A solution might be to place an internal baffle device along the sidewall vent to direct the incoming air into the upper part of the greenhouse, thus avoiding the direct impact on the crop.

0,030 y = -0.0009+0.05/x0.5 R2 = 0.98

0,025 0,020 0,015 0,010 0,005 0,000

0

10

20

30

40

50

60

Number of spans r&sw vent CFD Roof CFD

Fig. 7 – Evolution of the ventilation flow per unit greenhouse covered area (m3 mL2 sL1) as the number of spans increases, for roof ventilation only (line shows the average of the simulated values), and combined roof D sidewall ventilation (with regression equation).

93

biosystems engineering 104 (2009) 86–96

average temperature difference was between 2 and 3  C for combined roof and sidewall vents and was only slightly affected by span number, whereas for roof ventilation only this difference increased to between 5 and 6  C (Table 4).

Ventilation increase factor Roof+Sidewall vents/Roof vents

10 9

from CFD

8 7 6 5

3.2.3.

4

The last aspect of ventilation investigated concerned the effects of an anti-insect screens placed over the vent openings. The screen, with a porosity of 25%, simulated the one commonly used by growers in Mediterranean countries to limit the entrance of the white fly (Bemisia tabaci) and some thrips (Frankliniella occidentalis). Table 5 shows the ventilation rates obtained from simulations with the greenhouse models having 3, 5, 7, and 10 spans with and without the screen over the vents. A large reduction in ventilation rate was observed, which varied between 77 and 87%, depending on the number of spans. This dramatic reduction in ventilation rate caused a general increase in temperatures in the whole greenhouse. The average temperatures increased by between 4  C and almost 6  C (Tables 4 and 6), and the majority of the greenhouse volume had a temperature more than 4  C above the external value (Table 6). This reduction in ventilation and general increase in temperature agrees with CFD results obtained on greenhouses under windy conditions with anti-insect screens of different porosities (Bartzanas et al., 2002; Fatnassi et al., 2003, 2006; Molina-Aiz et al., 2004; Campen, 2005). The equation proposed by Pe´rez-Parra et al. (2004) indicates that this screen should produce a reduction in ventilation rate of 56% for a Reynolds number (Re) of 100, which corresponds to a wind speed of 4–5 m s1. Re depends mainly on the air speed through the vents and on the geometric characteristics of the screen (thread diameter or screen thickness). According to Bailey et al. (2003), the resistance of a screen to air flow depends on Re and screen porosity in such a way that, at low flows, as Re decreases, the resistance increases. The practical implication of this model is that, when wind speed is low or when the airflow through the vents is driven by buoyancy forces and the air speed is low, the screen resistance will be higher than at larger Re. This hypothesis was tested using the simulation results from the 5 span model. The average air speed in the sidewall vents was obtained from the simulations (U ¼ 0.075 m s1) because all the air entering the greenhouse flowed through these vents. The thickness of the screen was considered equal to the diameter of its threads (d ¼ 1.9  104 m). These values

3 2 1

0

10

20

30

40

50

60

Number of spans Fig. 8 – Ventilation increase factor when using roof D sidewall vents compared to use of only roof vents.

The points shown in Fig. 8 represent the ratios between the ventilation rates obtained when using roof and sidewall vents to those obtained with roof vents alone. Both values were derived from the results of the CFD simulations. The line represents the same factor, where the ventilation rates were calculated from the regression equation shown in Fig. 7, divided by the average value of the ventilation rates for roof vents alone. It is clear that even for a very large number of spans (30 spans) the ventilation rate with combined ventilation was approximately twice that when only roof vents were used. However, the air entering the sidewall vents will only affect a limited number of spans close to the sidewalls. The reduced ventilation rates in large greenhouses produced changes in the interior temperature field. As the greenhouses became larger, the regions that showed temperature differences greater than 4  C above the outside temperature increased in area (Table 4). This result means that the average greenhouse temperature also increased. The colder air jets penetrating into the greenhouse from the sidewall vents remain constant when the span number increases, thus they become less effective in the middle of large greenhouses, and the temperatures become higher, as seen by comparing Fig. 9(a,b). The results from simulations with the 20 span model show that the percentage of the greenhouse cross section with a temperature difference above 4  C is around 36%. Thus, more than one third of the greenhouse has high temperatures. However, this is in sharp contrast with the value of 79% obtained with the same greenhouse model when only the roof vents were open. The

Anti-insect screens

Table 4 – Average temperatures, inside–outside temperature differences (average and maximum values) and percentage of greenhouse areas with high temperature differences (DT > 4 8C) obtained from CFD simulations for increasing number of spans (combined roof D sidewall ventilation) Roof þ sidewallvents 3 Spans 5 Spans 7 Spans 10 Spans 15 Spans 20 Spans

Average temperatures ( C)

Average temperature difference ( C)

Maximum temperature ( C)

Maximum temperature difference ( C)

Area with DT > 4  C %

31.8 32.4 32.6 32.9 33.2 33.4

1.8 2.4 2.6 2.9 3.2 3.4

36.1 37.6 38.5 39.4 40.4 40.6

6.1 7.6 8.5 9.4 10.4 10.6

3.4 15.7 21.8 31.2 34.8 36.1

94

biosystems engineering 104 (2009) 86–96

Fig. 9 – Temperature difference field (8C) with combined ventilation (roof plus sidewall vents) in a 5 spans parral greenhouse (a) and a 15 spans parral greenhouse (b).

gave a Re of 0.92. Using this value and the screen porosity, the loss coefficient Fs of the screen was calculated as (Bailey et al., 2003):   Fs ¼ ½18=Re þ 0:75=log ðRe þ 1:25Þ þ 0:055 log ðReÞ 1  32 32 (6) which gave a value of Fs ¼ 327. The total loss in the vent is the sum of the loss caused by the screen plus the loss caused by the vent without the screen, i.e., Ft ¼ Fv þ Fs (Bailey et al., 2003). The loss coefficient of the unscreened vent was obtained from the expression proposed by Bailey et al. (2003), which gives the discharge coefficient as a function of the basic dimensions of the rectangular opening. The loss coefficient is related to the discharge coefficient Cd through the following equation: F ¼ 1=C2d

(7)

In this case, the discharge coefficient was Cdv ¼ 0.65, and the corresponding value of Fv ¼ 2.37. Therefore, the total loss coefficient was Ft ¼ 329, which corresponded to a discharge coefficient Cdt ¼ 0.055 for the screened vent. Thus, the screen reduced the discharge coefficient by 92%. This value for the reduction in air flow through the vent is quite close to the values of 87–89% obtained from the CFD simulations and clearly differs from the value of 56% predicted for average wind speeds.

Table 5 – Ventilation rates (vol. hL1) obtained from simulations with and without anti-insect screen on the greenhouse vents (screen porosity 3 [ 25%) Ventilation rate

3 Spans 5 Spans 7 Spans 10 Spans

No screen

Screen

28.0 21.3 19.9 16.2

3.6 2.5 2.3 1.8

Even for a more porous screen (0.35 porosity and 0.3 mm thread thickness), the flow reduction was quite high, since Re is very low due to the low air velocities with buoyancy-driven ventilation. For instance, taking the air velocity as 0.1 m s1 and the kinematic viscosity as 15.49  106 m2 s1, then Re ¼ 1.94. From Eq. (6), Fs is 9.96  7.16 ¼ 71.4. Thus, Ft ¼ 73.7 and Cdt ¼ 0.12. For this ‘‘more popular’’ screen the flow reduction is around 82%. Thus, this result shows that screen characteristics are not so important when the Re is very low; in thermally driven ventilation, Re is approximately 2.

4.

Discussion

With only roof vents, the ventilation rate remained more or less constant as the number of spans increases. This result differs from that found by Baeza et al. (2006) for the same parral-type greenhouse with the same vent configuration but with wind-driven ventilation. In that study, the ventilation rate was found to decrease as the number of spans increased. With few spans, the dynamic pressure field over the greenhouse created by wind turbulence and the internal air circulation leads to the roof vents acting as air inlets, air outlets, or a combination of both, thus contributing to a large overall air

Table 6 – Temperatures obtained from the simulations for different number of spans with an anti-insect screen (porosity 3 [ 25%) on the greenhouse vents Thermal values with screen on the vents

Reduction in ventilation rate (%) 87.2 88.3 88.4 88.9

Average Maximum Area with temperature ( C) temperature ( C) DT > 4  C (%) 3 Spans 5 Spans 7 Spans 10 Spans

36.5 36.8 38.0 38.1

39.6 39.8 42.0 43.4

92.7 91.9 93.7 93.2

biosystems engineering 104 (2009) 86–96

exchange. However, when the span number increases, some roof vents do not contribute to the ventilation process, thus reducing the effective ventilation capacity. With no wind, the pressure distribution due to buoyancy is not affected by the number of spans, and as a result the air exchange is similar regardless of the greenhouse length. As greenhouses are increased in size and the distance between the sidewall vents increases, their contribution to the overall ventilation of the greenhouse becomes less important in relation to the roof vents. The results from the simulations presented here suggest that, in the case of buoyancy-driven ventilation, the effect of sidewall vents extends a distance of approximately 25 m into the greenhouse for the design of greenhouse used in this study. In a very long greenhouse with many spans, the overall ventilation rate with roof and sidewall vents is almost equal to the ventilation rate with only roof vents. However, as there will be higher ventilation rates near the sidewalls than in the middle of a large greenhouse, the environment is less likely to be uniform in larger greenhouses. For such greenhouses, a solution could be to place the wall vents in the longest sidewalls, irrespective of whether these are parallel to or at right angles to the spans. However, the best situation would be to build very large greenhouses that are approximately square, where vents could be placed in all four walls. The presence of a crop, which behaves as a semi-porous media, will act as a barrier to the flow of ventilation air. This effect has been studied by several authors using the CFD technique (Haxaire et al., 2000; Lee & Short, 2000; Boulard & Wang, 2002; Bartzanas et al., 2004; Molina-Aiz et al., 2004; Fatnassi et al., 2006), who have shown that crop height, density and especially crop row orientation relative to the sidewall vents (Sase 1989) have strong influences on ventilation airflow. Simulations with the 25% porosity anti-insect screen showed that it reduced ventilation airflows by 87–89%. A small variation occurred between greenhouses that differed in the distance between the sidewall vents. This reduction was much greater than the value of 56% predicted using an average wind speed (Pe´rez-Parra et al., 2004). Using the method developed by Bailey et al. (2003) for screens at low air speeds, where the flow resistance increased with decreasing Re, gave a flow reduction of 92%, which was closer to the simulated values. Thus, it appears that with buoyancy-driven ventilation, anti-insect screens produce a greater reduction in airflow than with wind-driven ventilation.

5.

Conclusions

The most unfavourable situation for greenhouse cooling occurs when the wind speed is low, and natural ventilation is predominantly driven by buoyancy forces. A CFD model designed to simulate buoyancy-driven natural ventilation in parral-type greenhouses was created and shown to provide acceptable estimates of greenhouse airflows and temperatures when compared with experimental data obtained from measurements on physical models of greenhouses. Simulations with the CFD model showed that under calm conditions, the air exchange rate per unit ground area of

95

a greenhouse with only roof ventilation was independent of greenhouse width and length. However, when ventilators on two opposing sidewalls were combined with the roof ventilators, there were higher air exchanges rates and lower temperatures inside the greenhouses. The air exchange rates per unit ground area were highest in greenhouses with a small distance between the sidewalls containing the ventilators and decreased in an asymptotic manner towards the value for roof-only ventilation as the distance between the sidewalls increased. In a three span greenhouse, when using combined roof and sidewall ventilation, the air exchange rate was increased by a factor of 7, while in a 20 span house the factor was 2. The increased air exchange had the beneficial effect of reducing greenhouse temperature. In the 3 span greenhouse with only roof ventilation, 79% of the greenhouse cross section had a temperature more than 4 K above the exterior, whereas with combined ventilation the value was reduced to 23%. The corresponding values for the 20 span greenhouse were 48% and 36%, respectively. These results clearly show that, with buoyancy-driven ventilation, the contribution of the sidewall vents is important even for quite large greenhouses, but it is critical for small greenhouses. Placing nets over ventilation openings to exclude insects dramatically reduces the air exchange rate when exchange is driven by buoyancy. With combined ventilation, an antiinsect screen with a porosity of 25% reduced air exchange by 77–87%, depending on greenhouse size. These reductions are much larger than those obtained with wind-driven ventilation and indicate that very high temperatures are a distinct possibility in greenhouses fitted with effective insect screens when wind speeds are low.

Acknowledgements This research work was partially financed by the Cajamar Foundation and Spanish Projects CICYT AGL2005-06492-C0301 and INIA RTA2005-00142-CO2.

references

Abdel-Ghany A M; Ishigami Y; Goto E; Kozai T (2006). A method for measuring greenhouse cover temperature using a thermocouple. Biosystems Engineering, 95, 99–109. Baeza E J; Pe´rez-Parra J J; Montero J I; Lo´pez J C (2006). CFD study of the natural ventilation performance of a parral type greenhouse with different numbers of spans and roof vent configurations. Acta Horticulturae, 719, 333–340. Bailey B J; Montero J I; Perez Parra J J; Robertson A P; Baeza E; Kamaruddin R (2003). Air flow resistance of greenhouse ventilators with and without insect screens. Biosystems Engineering, 86, 217–229. Bartzanas T; Boulard T; Kittas C (2002). Numerical simulation of the airflow and temperature distribution in a tunnel greenhouse equipped with insect-proof screen in the openings. Computers and Electronics in Agriculture, 34, 207–221. Bartzanas T; Boulard T; Kittas C (2004). Effect of vent arrangement on windward ventilation of a tunnel greenhouse. Biosystems Engineering, 88, 479–490.

96

biosystems engineering 104 (2009) 86–96

Boulard T; Wang S (2002). Experimental and numerical studies on the heterogeneity of crop transpiration in a plastic tunnel. Computers and Electronics in Agriculture, 34, 173–190. Bruce J M (1978). Natural convection through openings and its application to cattle building ventilation. Journal of Agricultural Engineering Research, 23, 151–167. Bunschoten B; Pierik C (2003). Kassenbouw neemt weer iets toe. CBS Webmagazine (Centraal Bureau voor de Statistiek) Available from: . Campen J B (2005). Greenhouse design applying CFD for Indonesian conditions. Acta Horticulturae, 691, 419–424. Castilla N (1994). Greenhouses in the Mediterranean area: technological level and strategic management. Acta Horticulturae, 361, 44–56. De Jong T (1990). Natural ventilation of large multispan greenhouses. PhD Dissertation. Agricultural University, Wageningen, The Netherlands, p. 116. von Elsner B; Briassoulis D; Waaijenberg D; Mistriotis A; von Zabeltitz Chr; Gratraud J; Russo G; Suay-Cortes R (2000). Review of structural and functional characteristics of greenhouses in European Union Countries, Part II: typical designs. Journal of Agricultural Engineering Research, 75, 111–126. Fatnassi H; Boulard T; Bouirden L (2003). Simulation of climatic conditions in full-scale greenhouse fitted with insect-proof screens. Agricultural and Forest Meteorology, 118, 97–111. Fatnassi H; Boulard T; Poncet C; Chave M (2006). Optimisation of greenhouse insect screening with computational fluid dynamics. Biosystems Engineering, 93, 301–312. Ferna´ndez C. y Pe´rez-Parra, J.J. (2004) ‘‘Caracterizacio´n de los invernaderos de la provincia de Almerı´a’’ Edited by Cajamar. Legal deposit: AL-457–2004. 20 pp. Haxaire R; Boulard T; Mermier M (2000). Greenhouse natural ventilation by wind forces. Acta Horticulturae, 534, 31–40. Kacira M; Short T H; Stowell R R (1997). A cfd evaluation of naturally ventilated, multi-span, sawtooth greenhouses. Transactions of the ASAE, 41, 833–836. Kacira M; Sase S; Okushima L (2004). Effects of side vents and span numbers on wind-induced natural ventilation of a gothic multi-span greenhouse. Journal of Agricultural Research Quarterly, 38, 227–233. Kamaruddin R (1999). A naturally ventilated crop protection structure for tropical conditions. PhD Dissertation, Cranfield University, Silsoe, UK. Kittas C; Boulard T; Papadakis G (1997). Natural ventilation of a greenhouse with ridge and side openings: sensitivity to temperature and wind effects. Transactions of the ASAE, 40, 415–425. Lagier J (1990). Proble`mes pose´s par la maıˆtrise du climat dess erres en re´gion me´diterranknne: efficience de l’ombrage et de la brumisation. Acta Horticulturae, 263, 325–334. Launder B E; Spalding D B (1974). The numerical computation of turbulent flows. Computer Methods in Applied Mechanics and Engineering, 3, 269–289.

Lee I B; Short T H (2000). Two dimensional numerical simulation of natural ventilation in a multi-span greenhouse. Transactions of the ASAE, 43, 745–753. Miguel A F (1998). Airflow through porous screens: from theory to practical considerations. Energy and Buildings, 31, 195–209. Mistriotis A; Arcidiacono C; Picuno P; Bot G P A; ScarasciaMugnozza G (1997). Computational analysis of ventilation in greenhouses at zero- and low-wind-speeds. Agricultural and Forest Meteorology, 88, 121–135. ´ lvarez A J (2004). Measurement and Molina-Aiz F D; Valera D L; A simulation of climate inside Almerı´a-type greenhouses using computational fluid dynamics. Agricultural and Forest Meteorology, 125, 33–51. Montero J; Anton A (2000). Buoyancy driven ventilation in tropical greenhouses. Acta Horticulturae, 534, 41–48. Montero J I; Castilla N; Gutierrez E; Bretones F (1985). Climate under plastic in the Almerı´a area. Acta Horticulturae, 170, 227–234. Montero J I; Hunt G R; Kamaruddin R; Anton A; Bailey B J (2001). Effect of ventilator configuration on wind-driven ventilation in a crop protection structure for the tropics. Journal of Agricultural Engineering Research, 80, 99–107. Oca J (1996). Estudi de la ventilacio´ natural per efecte te´rmic en hivernacles, mitjanc¸ant te´cniques de visualitzacio´ de fluids en models a escala. PhD Dissertation, University of Lleida, Lleida, Spain. Oca J; Montero J I; Anto´n A; Crespo D (1999). A method for studying natural ventilation by thermal effects in a tunnel greenhouse using laboratory-scale models. Journal of Agricultural Engineering Research, 72, 93–104. Patankar S V (1980). Numerical heat transfer and fluid flow. Hemisphere Publishing Corporation, New York. Pe´rez-Parra J J (2002). Ventilacio´n natural de invernaderos tipo parral. PhD Dissertation, University of Co´rdoba, Spain. Pe´rez-Parra J J; Baeza E; Montero J I; Bailey B J (2004). Natural ventilation of parral greenhouses. Biosystems Engineering, 87, 89–100. Richards P J; Hoxey R P (1992). Computational and wind tunnel modelling of mean wind loads on the Silsoe structures building. Journal of Wind Engineering and Industrial Aerodynamics, 43(1–3), 1641–1652. Sase S (1989). The effect of plant arrangement on airflow characteristics in a naturally ventilated glasshouse. Acta Horticulturae, 245, 429–435. Sase S; Takakura T; Nara M (1984). Wind tunnel testing on airflow and temperature distribution if a naturally ventilated greenhouse. Acta Horticulturae, 174, 329–336. Timmons M B; Baughman G R (1981). Similitude analysis of ventilation by the stack effect from an open ridge livestock structure. Transactions of the ASAE, 24, 1030–1033. Vollebregt H J M; van de Braak N J (1995). Analysis of radiative and convective heat exchange at greenhouse walls. Journal of Agricultural Engineering Research, 60, 99–106.