Microclimate evaluation of a new design of insect-proof screens in a Mediterranean greenhouse

This work studies natural ventilation in a Mediterranean greenhouse, comparing a new experimental screen of 13 × 30 threads cm (porosity 39.0%) with a commercial control screen of 10 × 20 threads cm (porosity 33.5%). In addition, both screens were tested in a wind tunnel to determine the discharge coefficients Cd of the greenhouse side and roof vents, which proved to be 0.16 for the commercial control screen and 0.18 for the experimental screen at both vents. These values represent a theoretical increase of 11% (Cd,φ-10 × 20 /Cd,φ-13 × 30 = 0.89) in the natural ventilation capacity of the greenhouse when the experimental screen is used. The greenhouse was divided into two separate sections allowing us to analyze natural ventilation in both sectors simultaneously. Air velocity was measured in the lateral and roof vents with two 3D and six 2D sonic anemometers. Using the commercial control screen there was an average reduction of 16% in ventilation rate, and an average increase of 0.5°C in the average indoor air temperature, compared to the experimental screen. In addition, the ventilation efficiency ηT was higher with the experimental screen (mean value of 0.9) than with the control (mean value 0.6). We have designed an experimental insect-proof screen (13 × 30 threads cm) with smaller thread diameter, higher thread density, smaller pore size and higher porosity than are used in most commercial meshes. All of these factors promote natural ventilation and improve the greenhouse microclimate. Additional key words: natural ventilation; sonic anemometry; discharge coefficient; ventilation efficiency. * Corresponding author: dvalera@ual.es Received: 05-09-13 Accepted: 07-05-14. Abbreviations used. Nomenclature: Cd (total discharge coefficient of the opening); Cd,LH (discharge coefficient due to the shape of the opening); Cd,φ (discharge coefficient due to the presence of insect-proof screens); Df [thread density according to the manufacturer (threads cm)]; Dh [diameter of the threads (μm)]; Di [diameter of the inside circumference of the pore (μm)]; Dr [thread density measurement (threads cm)]; E [thickness of the screen (μm)]; EG [error in the calculation of volumetric flow rate (%)]; Fφ (pressure drop coefficient due to the presence of an insect-proof screen); G [mean volumetric flow rate (m s)]; HR [relative air humidity (%)]; Kp [screen permeability (m)]; Lpx [length of the pore in the weft direction (μm)]; Lpy [length of the pore in the warp direction (μm)]; LV [length of the vent (m)]; P [pressure (Pa)]; q [specific humidity (g g)]; R [ventilation rate (h)]; Rep (Reynolds number based on the screen’s permeability); Ro [incoming shortwave radiation (W m)]; SA [greenhouse surface area (m)]; Sj [unit surface area corresponding to each measurement point in the vent (m)]; Sp [area of the pore (mm)]; SV [vent surface area (m)]; T [temperature (°C)]; u [air velocity (m s)]; Y (inertial factor). Greek letters: Δ (difference); β [angle of opening (°)]; ηT (ventilation efficiency for the temperature); θ [wind direction (°)]; μ [dynamic viscosity of air (kg s m)]; ρa [air density (kg m)]; φ [insect-proof screen porosity (%)]. Subscripts: C (corrected); i (inside); j (measurement point); LS (leeward side vent); M (average value); O (outside); S (sonic); WR (windward roof vent); WS (windward side vent); x (longitudinal component). Superscripts: * (corrected with wind speed); c (corrected with inside-outside air temperature). Instituto Nacional de Investigación y Tecnología Agraria y Alimentaria (INIA) Spanish Journal of Agricultural Research 2014 12(2): 338-352 http://dx.doi.org/10.5424/sjar/2014122-4956 ISSN: 1695-971-X eISSN: 2171-9292 RESEARCH ARTICLE OPEN ACCESS 2008; Baeza et al., 2009; Molina-Aiz et al., 2009), which has negative repercussions on the greenhouse microclimate, increasing the interior temperature and humidity (Fatnassi et al., 2006; Harmanto et al., 2006). Growers must install insect-proof screens with an average pore size smaller than the size of the tiniest insect pest (Teitel, 2007), but at the same time the porosity of the chosen screens must not be detrimental to the natural ventilation of the greenhouse. Given this scenario, the growers’ choice of mesh is limited, as lower porosity meshes reduce the ventilation rate and increase both temperature and humidity inside the greenhouse (Fatnassi et al., 2003). For instance, a mesh of 25% porosity can cause a drop in the greenhouse ventilation rate of 77-87% (Baeza et al., 2009); one of 53% porosity was found to reduce the ventilation rate by 50% and increase the greenhouse temperature by 4°C compared to a greenhouse without meshes (Bartzanas et al., 2002). Other studies have obtained the following results: 27% reduction in ventilation rate with a mesh of 50% porosity (Katsoulas et al., 2006), 50% reduction with a mesh of 45% porosity (Muñoz et al., 1999), 35% reduction with a mesh of 39% porosity (Pérez-Parra et al., 2004), 35% reduction with a mesh of 34% porosity (Campen & Bot, 2003), 38% reduction with a mesh of 34% porosity, and 55% reduction with a mesh of 26% porosity (Campen, 2005). The above-mentioned negative aspects of the screens may be exacerbated by the accumulation of dust and dirt in the pores of the mesh (Linker et al., 2002; López et al., 2013). Over time the structure of the mesh deteriorates, and López et al. (2013) observed that the thread diameter increases by an average of 3.1% and the porosity of the screen suffers a mean reduction of 6.5%. The accumulation of dust and dirt in the insect-proof screens gives rise to a mean reduction in the porosity of the insect-proof screens of 20.3%. The combined total of these two negative effects (mesh degradation and accumulation of dust) amounted to a 26.8% mean reduction in porosity (López et al., 2013). To avoid the entrance of insects into greenhouse, insect-proof screens are indispensable, but solutions must be sought to minimize their effect on the natural ventilation and the greenhouse microclimate. At this point several possibilities may be put forward: — (i) Increasing the total ventilation surface in order to limit as much as possible the reduction in natural ventilation capacity brought about by the use of insect-proof screens (Muñoz et al., 1999; Teitel, 2007). One option is to increase the maximum angle to which the roof vent can be opened in multi-span greenhouses (Bailey, 2003), though this is hardly viable in areas with frequent strong winds. Another way is to fit the frames of the openings with pre-formed concertina-shape screens that unfold as the vents open and fold up again when they close (Bailey, 2003). — (ii) Combining insect-proof screens with other means of pest control which allow the use of mesh with lower thread density and greater pore size. For instance, biological control methods making use of other insects, mites, fungi, bacteria, viruses, nematodes and weeds (auxiliary), which act as parasites, predators, pathogens, antagonists or competitors of the organisms that cause damage to the crop. The green revolution that Almería (Spain) has undergone over the last four years is particularly noteworthy. In this area many of the crops are produced using 100% biological control methods, which also constitute a competitive advantage over rival areas of production. The following examples provide an idea of the solutions available to growers: for the biological control of whitefly (Trialeurodes vaporariorum Westwood), the parasitic wasp Encarsia formosa (Castañé et al., 2004) can be used, and also autochthonous Mediterranean predatory bugs such as Macrolophus caliginosus or Dicyphus tamaninii Wagner (Lucas & Alomar, 2002; Castañé et al., 2004); to combat the Bemisia argentifolii whitefly parasitic wasps such as Eretmocerus eremicus and Encarsia Formosa (Hoddle et al., 1998) can be used; while to combat thrips (Frankliniella occidentalis), the parasitic nematode Thripinema nicklewoodi Siddiqi (Arthurs et al., 2003), the fungus Metarhizium anisopliae (Ansari et al., 2007), the predatory mite Amblyseius cucumeris or the wasp Orius insidiosus (Shipp & Wang, 2003) can all be used. Other methods for combating whitefly have been studied: planting a layer of covering vegetation that attracts these insects in order to reduce their incidence on crops (Hilje & Stansly, 2008); placing chromatographic adhesive traps shaped like a chrysanthemum flower (Mainali & Lim, 2008); and fitting electrostatic meshes (Tanaka et al., 2008). — (iii) Designing new, more porous insect-proof screens with larger pores than the current marketed models. Along these lines, the present work presents an experimental mesh of 13 × 30 threads cm. The threads of this mesh are of very small diameter, allowing us to obtain reduced pore size while at the same Microclimate evaluation of a new design of insect-proof screens 339


Introduction
With a view to reducing the entrance of certain insects in the greenhouse (aphids, leaf miners, whitefly, thrips or mealybugs), insect-proof screens are installed on the vents, thus reducing the need for phyto-sanitary treatments (Baker & Jones, 1989;Berlinger et al., 1992), while at the same time avoiding the exit of beneficial insects (Taylor et al., 2001;Teitel, 2007).On the other hand, the use of these materials reduces the greenhouses' capacity for natural ventilation (Muñoz et al., 1999;Fatnassi et al., 2003;Kittas et al.,

Microclimate evaluation of a new design of insect-proof screens in a Mediterranean greenhouse
2008; Baeza et al., 2009;Molina-Aiz et al., 2009), which has negative repercussions on the greenhouse microclimate, increasing the interior temperature and humidity (Fatnassi et al., 2006;Harmanto et al., 2006).Growers must install insect-proof screens with an average pore size smaller than the size of the tiniest insect pest (Teitel, 2007), but at the same time the porosity of the chosen screens must not be detrimental to the natural ventilation of the greenhouse.Given this scenario, the growers' choice of mesh is limited, as lower porosity meshes reduce the ventilation rate and increase both temperature and humidity inside the greenhouse (Fatnassi et al., 2003).For instance, a mesh of 25% porosity can cause a drop in the greenhouse ventilation rate of 77-87% (Baeza et al., 2009); one of 53% porosity was found to reduce the ventilation rate by 50% and increase the greenhouse temperature by 4°C compared to a greenhouse without meshes (Bartzanas et al., 2002).Other studies have obtained the following results: 27% reduction in ventilation rate with a mesh of 50% porosity (Katsoulas et al., 2006), 50% reduction with a mesh of 45% porosity (Muñoz et al., 1999), 35% reduction with a mesh of 39% porosity (Pérez-Parra et al., 2004), 35% reduction with a mesh of 34% porosity (Campen & Bot, 2003), 38% reduction with a mesh of 34% porosity, and 55% reduction with a mesh of 26% porosity (Campen, 2005).
The above-mentioned negative aspects of the screens may be exacerbated by the accumulation of dust and dirt in the pores of the mesh (Linker et al., 2002;López et al., 2013).Over time the structure of the mesh deteriorates, and López et al. (2013) observed that the thread diameter increases by an average of 3.1% and the porosity of the screen suffers a mean reduction of 6.5%.The accumulation of dust and dirt in the insect-proof screens gives rise to a mean reduction in the porosity of the insect-proof screens of 20.3%.The combined total of these two negative effects (mesh degradation and accumulation of dust) amounted to a 26.8% mean reduction in porosity (López et al., 2013).
To avoid the entrance of insects into greenhouse, insect-proof screens are indispensable, but solutions must be sought to minimize their effect on the natural ventilation and the greenhouse microclimate.At this point several possibilities may be put forward: -(i) Increasing the total ventilation surface in order to limit as much as possible the reduction in natural ventilation capacity brought about by the use of in-sect-proof screens (Muñoz et al., 1999;Teitel, 2007).One option is to increase the maximum angle to which the roof vent can be opened in multi-span greenhouses (Bailey, 2003), though this is hardly viable in areas with frequent strong winds.Another way is to fit the frames of the openings with pre-formed concertina-shape screens that unfold as the vents open and fold up again when they close (Bailey, 2003).
-(ii) Combining insect-proof screens with other means of pest control which allow the use of mesh with lower thread density and greater pore size.For instance, biological control methods making use of other insects, mites, fungi, bacteria, viruses, nematodes and weeds (auxiliary), which act as parasites, predators, pathogens, antagonists or competitors of the organisms that cause damage to the crop.The green revolution that Almería (Spain) has undergone over the last four years is particularly noteworthy.In this area many of the crops are produced using 100% biological control methods, which also constitute a competitive advantage over rival areas of production.The following examples provide an idea of the solutions available to growers: for the biological control of whitefly (Trialeurodes vaporariorum Westwood), the parasitic wasp Encarsia formosa (Castañé et al., 2004) can be used, and also autochthonous Mediterranean predatory bugs such as Macrolophus caliginosus or Dicyphus tamaninii Wagner (Lucas & Alomar, 2002;Castañé et al., 2004); to combat the Bemisia argentifolii whitefly parasitic wasps such as Eretmocerus eremicus and Encarsia Formosa (Hoddle et al., 1998) can be used; while to combat thrips (Frankliniella occidentalis), the parasitic nematode Thripinema nicklewoodi Siddiqi (Arthurs et al., 2003), the fungus Metarhizium anisopliae (Ansari et al., 2007), the predatory mite Amblyseius cucumeris or the wasp Orius insidiosus (Shipp & Wang, 2003) can all be used.Other methods for combating whitefly have been studied: planting a layer of covering vegetation that attracts these insects in order to reduce their incidence on crops (Hilje & Stansly, 2008); placing chromatographic adhesive traps shaped like a chrysanthemum flower (Mainali & Lim, 2008); and fitting electrostatic meshes (Tanaka et al., 2008).
-(iii) Designing new, more porous insect-proof screens with larger pores than the current marketed models.Along these lines, the present work presents an experimental mesh of 13 × 30 threads cm -2 .The threads of this mesh are of very small diameter, allowing us to obtain reduced pore size while at the same time maintaining an acceptable level of porosity.This experimental insect-proof screen was designed by the "Rural Engineering" Research Group (AGR-198) of Almería University in collaboration with Spanish companies that manufacture and market such screens.The design allows the cost, and therefore the sale price, to be similar to that of screens currently on the market.The efficiency of the screen in preventing the entrance of whitefly into the greenhouse was tested using yellow sticky traps to calculate the whitefly population inside the experimental greenhouse described below over four crop cycles of Solanum lycopersicum L. In the first crop cycle (autumn-winter 2007/2008) a 66% reduction in whitefly population was observed in the western sector of the greenhouse fitted with the experimental mesh (13 × 30 threads cm -2 ), as compared to the control mesh (10 × 20 threads cm -2 ); in the second cycle (spring-summer 2008) this reduction was 12%; in the third one (autumn-winter 2008/2009) it was 7%; and in the fourth (spring-summer 2009) the reduction was 22% (Escamirosa, 2009).
Sonic anemometry has been used to study the effect of two insect-proof screens on the natural ventilation of a Mediterranean greenhouse; a commercial mesh (10 × 20 threads cm -2 ) and the experimental one (13 × 30 threads cm -2 ).

Material and methods
Two types of experiment have been carried out in order to evaluate the new design of insect-proof screens (13 × 30 threads cm -2 ): A), wind tunnel tests to evaluate the aerodynamic characteristics of the experimental mesh (13 × 30 threads cm -2 ) and the control mesh (10 × 20 threads cm -2 ) in the laboratory (under controlled conditions); B), experimental measurements in the greenhouse, using two 3D and six 2D sonic anemometers, as well as other sensors, to analyze the microclimate and natural ventilation of the experimental greenhouse fitted with both screens.

Wind tunnel tests
In order to obtain the aerodynamic characteristics of the insect-proof screens, tests were carried out in a 4.74 m-long low velocity wind tunnel (Fig. 1) with a circular cross-section of 38.8 cm diameter.This wind tunnel was designed in the Engineering Department of the University of Almería.This device is able to produce airflows of up to 10 m s -1 , and is divided into the following principal parts (Fig. 1): flow conditioner and contraction, test section, diffuser and fan (Valera et al., 2006;Molina-Aiz et al., 2006).The experimental device forces a current of air through the porous meshes, allowing measurement of both the air velocity and the pressure drop produced.For the technical characteristics of the wind tunnel, the sensors used to measure air velocity and pressure drop and the different electronic components, see Valera et al. (2005 and2006) or Molina-Aiz et al. (2006).
For geometric characterization of the insect-proof screens (Table 1) a specific software tool was used (Valera et al., 2003(Valera et al., , 2006;;Álvarez et al., 2012).The average diameter of the threads in the experimental screen (165.5 ± 7.0 µm) was much lower than that of the control screen (254.3 ± 11.7 µm) analysed by Álvarez (2010), and so its porosity was greater despite the higher density of threads.
Once we have determined the relationship between the pressure drop produced when the airflow passes through the mesh, the air velocity through the porous medium and the geometric characteristics of the meshes (porosity and thickness), we can calculate the screen permeability K p , the inertial factor Y, the pressure drop coefficient F ϕ and the discharge coefficient due to the presence of insect-proof screens C d,ϕ .Darcy's equation depends linearly on velocity for Reynolds numbers below the unit.However, for Reynolds number Re > 10 a break is observed in linearity, since the resistance to solid objects (in this case the mesh) is comparable to the resistance to friction with solid surfaces (Nield & Bejan, 1998).The airflow through the porous medium (the insect-proof screen) can be described by modifying Darcy's equation (Forchheimer, 1901): [1] where P is the pressure (Pa), x the direction of airflow, u the air velocity (m s -1 ), Y the inertia factor, a dimensionless drag constant that depends on the characteristics of the porous material, K p (m 2 ) is an independent coefficient of the nature of the fluid that depends on the geometry of the porous medium and is known as the specific permeability of the porous medium (Nield & Bejan, 1998), µ is the dynamic viscosity of air (kg s -1 m -1 ) and ρ a is the air density (kg m -3 ).In line with Forchheimer's equation, a second degree polynomial can be used (Miguel et al., 1997;Dierickx, 1998;Muñoz et al., 1999) to adjust the experimental values of pressure drop through the mesh (Molina-Aiz et al., 2006;Valera et al., 2006): The independent term cannot be considered if it is compared with the other terms (Miguel et al., 1997).Equating the coefficients of the first and second order terms of Forchheimer's Eq. [1] and of the second degree polynomial that results from the fit of the experimental data Eq.[2], we can obtain the specific permeability K p (m 2 ) and the inertial factor Y (dimensionless) from the following expressions (Molina-Aiz et al., 2009): where e is the thickness of the screen (m).To obtain an accurate measurement of the thickness of the mesh, dimensional contact-free metrology has been used.The optical measurements were taken with a TESA-VISIO 300 (TESA SA, Switzerland) fitted with video and with a resolution of 0.05 µm and an uncertainty in measurement of (3 + 10 • e/1,000) (µm), where e is the dimension measured, in this case the thickness of the mesh.For the magnitude of the measurements taken the uncertainty was less than 10 µm.
Bernoulli's equation can also be used to describe the relationship between pressure drop and air velocity through the mesh (Kosmos et al., 1993;Montero et al., 1997;Teitel & Shklyar, 1998): [5] where F ϕ is the pressure drop coefficient due to the presence of an insect-proof screen.This coefficient can be calculated from the wind tunnel tests according to the expression obtained by equaling Eq.[1] and Eq.
[5] for ∂P/∂x = ΔP/e (Molina-Aiz et al., 2009): [6] This coefficient can be used to predict the pressure drop through the mesh for Reynolds numbers Re p < 10 5 (Teitel, 2001).Re p is the Reynolds number based on the screen's permeability and it can be calculated by considering the square root of specific permeability K p (Nield & Bejan, 1998): [7] In the present study, to determine Re p we have used a value of 0.25 m s -1 for u (maximum mean value of the longitudinal component u x , perpendicular to the vents, in the experimental greenhouse).The discharge coeff icient due to the presence of insect-proof screens C d,ϕ can be calculated from the pressure drop coefficient F ϕ as follows (Molina-Aiz et al., 2009):

Microclimate evaluation of a new design of insect-proof screens 341
Table 1.Geometric characteristics of the screens (average values ± standard deviation).D f and D r are the thread densities according to the manufacturer and measurement, respectively (threads cm -2 ).Average value and standard deviation of: ϕ, porosity (m 2 m -2 ); L px and L py , the lengths of the pore (µm) in the direction of the weft and warp, respectively; D h , diameter of the threads (µm); D i , diameter of the inside circumference of the pore (µm); S p , area of the pore (mm 2 ) Eastern sector 10 × 20 9.9 × 19.7 0.335 ± 0.011 233.7 ± 23.9 734.0 ± 29.2 274.5 ± 11.0 236.6 ± 24.0 0.171 ± 0.019 Western sector 13 × 30 13.1 × 30.5 0.390 ± 0.006 164.6 ± 9.3 593.3 ± 19.0 165.5 ± 7.0 167.4 ± 9.6 0.098 ± 0.006 [8] Bernouilli's law establishes that when a vent is opened the pressure drop between its two sides (potential energy) becomes kinetic energy.The main parameters that determine the aerodynamic behavior of the vents are their geometry (height and width), the angle to which they are opened and the presence of insect proof screens (porosity and airflow).The coefficient C d for each vent can be calculated as follows (Arbel et al., 2000;Kittas et al., 2002;Teitel, 2007): [9] where C d,LH is the discharge coefficient due to the shape of the opening which can be obtained as follows (Bailey et al., 2003): [10] where L V is the length of the vent [m], h the height (m) and ‚ the angle of opening, which is 90°for side vents without spoiler and 14°for the roof vent.

Experimental setup for greenhouse measurements
The experimental measurements were carried out in a multi-tunnel greenhouse (24 × 45 m 2 ), divided into two sectors (eastern sector 24 × 25 m 2 and western sector 24 × 20 m 2 ), located at the "Catedrático Eduardo Fernández" farm of the UAL-ANECOOP Foundation (36°51' N, 2°16' W) in the province of Almería (Fig. 2).The greenhouse was divided in two parts by a three-layer plastic film of the same material as the greenhouse cover.The plastic was held in place using metal profiles which are the same as those used on the greenhouse side walls, and in this way no air could pass from one sector to the other.Experimental measurements were carried out during the spring-summer crop cycle of Solanum lycopersicum L. cv.Salomee (average height 1.20 and 1.88 m, and average leaf area index (LAI) 0.78 and 2.75 m 2 m -2 , for the first and the last experimental measurements, respectively).
One roof vent and both side vents were fully open during the experimental measurements (Fig. 2, Table 2).The commercial insect-proof screen control (10 × 20 threads cm -2 ; Fig. 3a) was placed on the vents in the eastern sector of the greenhouse, while the experimental one (13 × 30 threads cm -2 ; Fig. 3b) with a higher thread density was installed in the western sector.The natural ventilation of the greenhouse was studied in both sectors simultaneously.The air velocity was measured at 12 points in each side vent (Fig. 4a) and at 3 points in each roof vent (Fig. 4b).The vent area corresponding to each measurement point was 2.0 m 2 (side vents) and 7.5 m 2 (roof vent) in the eastern sector, and 1.5 m 2 (side vents) and 5.8 m 2 (roof vent) in the western sector.These values are similar to those used by Boulard et al. (1998)  Two 3D and six 2D sonic anemometers were used: half of the total in each sector.The 3D anemometers measured the air velocity over 3 minutes (Molina-Aiz et al., 2009;López et al., 2011López et al., , 2012) ) at each point in the side vents (Fig. 4c).The 2D anemometers were Microclimate evaluation of a new design of insect-proof screens 343   used to measure continuously the air velocity at each point in the roof vent (Fig. 4d).The methodology employed has been explained in greater detail in López et al. (2011López et al. ( , 2012)).
The side vent of the northern sector of the greenhouse was partially blocked by a small auxiliary warehouse (Fig. 2).This implies a negative influence on the experimental design, as the two sectors of the greenhouse are not identical.In order to complement the results obtained from the experimental measurements (B), wind tunnel tests (A) were also carried out.In this way the aerodynamic behaviour of the mesh was analyzed without the influence of external factors.
Outside climatic conditions were recorded by a meteorological station at a height of 10 m (Fig. 2).The meteorological station included a BUTRON II (Hortimax S.L., Almería, Spain) measurement box with a Pt1000 temperature sensor and a capacitive humidity sensor, with a temperature measurement range of -25°C to 75°C and accuracy of ± 0.01°C, and a humidity range of 0% to 100% and accuracy of ± 3%.Outside wind speed was measured with a Meteostation II (Hortimax S.L.), incorporating a cup anemometer with a measurement range of 0 to 40 m s -1 , accuracy of ± 5%, and resolution of 0.01 m s -1 .Wind direction was measured with a vane (accuracy ± 5°and resolution 1°).Solar radiation was measured using a Kipp Solari (Hortimax S.L.) sensor, with a measurement range of 0 to 2,000 W m -2 , accuracy of ± 20 W m -2 , and resolution of 1 W m -2 .
Temperature and humidity inside the greenhouse were measured using 12 autonomous dataloggers (HOBO Pro Temp-HR U23-001, Onset Comput.Corp., Bourne, MA, USA) placed at heights of 1 and 2 m.These fixed devices allowed temperature measurement in a range of -40°C to 70°C with an accuracy of ± 0.18°C.They were all programmed to register data at 0.5 Hz and were protected against direct solar radiation with passive solar radiation open boxes, allowing natural air movement around the sensors (Molina-Aiz et al., 2004).From the data of inside humidity we can obtain the specific humidity q and correct the sonic anemometer temperature T s using the following expression (Tanny et al., 2008):

Statistical analysis
Temperature data inside the different sectors of the experimental greenhouse were subjected to Analysis of Variance (ANOVA) using Statgraphics Plus 4.1 Software (Manugistics, Inc., Rockville, MD, USA).Oneway ANOVA and possible significant differences between the temperature inside the eastern and western sectors were evaluated by Least Significant Differences (LSD) multiple comparison tests with a confidence level of 95%.On the other hand, regression analyses were carried out to study the relationship between the inside temperature (T i ), as dependent variable, and the outside temperature (T o ), the wind speed (u o ) and the porosity of the insect-proof screens (ϕ), as independent variables.

Calculation of the airflow passing through the vents
With only one sampling position possible (for lateral vents) at any one time, a difficulty arises from how to deal with changing external conditions throughout the time needed to record data at all the measurement positions in the lateral vents of each sector.This problem can be overcome by correcting the air velocities measured by the 3D sonic anemometer at each position j at the lateral vents through a process of scaling with the wind speed (Boulard et al., 2000;Molina-Aiz et al., 2009;López et al., 2011): [12] where u o is the mean wind speed during the experimental measurements [m s -1 ], u o,j and u x,j the mean va- T sc = T s 1 + 0.51q lue of the wind speed and of the longitudinal component of the air velocity u x corresponding to measurement point j.The volumetric flow rate G is calculated by multiplying u * x,j by the vent surface area corresponding to each measurement point (Boulard et al., 1998): [13]

Wind tunnel tests
The tests in the wind tunnel (Table 3) allowed the discharge coefficient C d,ϕ to be determined (for u = 0.25 m s -1 ) as 0.17 for the control screen (10 × 20 threads cm -2 ) and as 0.19 for the experimental one (13 × 30 threads cm -2 ).The discharge coefficient C d for the side and roof vents was then calculated as 0.16 for the control and 0.18 for the experimental screen at both vents.These values are similar to those obtained in other studies on insect-proof screens: C d = 0.253 in a mono-span greenhouse with side vents and a mesh of 35% porosity (Teitel et al., 2008); C d = 0.194 for an Almería-type greenhouse with a mesh of 34% porosity (Molina-Aiz et al., 2009).In the latter, the authors report the values obtained in a number of other works, ranging between 0.42 and 0.89 for vents without screens and between 0.16 and 0.51 for vents with insect-proof screens of 25% and 45% porosity, respectively.Fatnassi et al. (2002) establish that the relationship between the ventilation rates of the same greenhouse with and without insect-proof screens can be conside-red as proportional to the ratio of the discharge coefficients of the vents with and without screens.Bearing this in mind, the results obtained would lead us to expect a theoretical increase of 11% (C d-10 × 20 /C d-13 × 30 = 0.89) in the natural ventilation capacity in the greenhouse's western sector fitted with the experimental screen (13 × 30 threads cm -2 ) with respect to that of the eastern sector with the commercial screen (10 × 20 threads cm -2 ).The vents' discharge coefficients were also calculated for values of u from 0.05 to 0.50 m s -1 (Fig. 5), and in all cases the C d-10 × 20 /C d-13 × 30 ratios obtained were equal to 0.89.

Experimental measurements in greenhouse
The measurement tests were carried out under Poniente wind from the southwest.The outside climatic conditions remained relatively stable over the six measurement tests (Table 4).

Airflow
Given the characteristics of the experimental greenhouse (with a windward facing roof vent, a windward facing side vent that is blocked by another greenhouse and a leeward facing side vent which is partially blocked in the eastern sector) the eolic and thermal effects are contrary.Due to the eolic effect the incoming air enters through the roof vent (wind pressure) and exits through the leeward side vent (suction of the wind), but also through the windward side vent, as it is in the shelter of the adjacent greenhouse.Due to the thermal effect, the hot air rises to exit through the roof vent, which favours the entrance of air from outside through the side vents.The combination of these two effects at the three vents means that there are opposing incoming and outgoing airflows (Fig. 6).The same ventilation pattern was previously observed for the western sector of the greenhouse under prevailing Poniente winds (López et al., 2011 and2012).
Despite the difference in porosity of the tested screens (5.5 points), no major differences were found in the ventilation patterns of the two greenhouse sectors (Fig. 6).
The greenhouse ventilation is affected by the buoyancy effect generated by the average characteristic temperature difference between the inside and the out-side air ΔT io .Papadakis et al. (1996) showed that the effect of buoyancy on greenhouse ventilation could not be neglected at wind speeds lower than about 1.8 m s -1 .In greenhouses with both roof and side openings, Kittas et al. (1997) considered that the stack effect is important if the ratio u o /ΔT io 0.5 < 1.In the present case, air was found to both enter and leave the greenhouse through the roof vent in all experimental measurements, which indicates that the thermal effect played a major role in the natural ventilation, even though the ratio u o /ΔT io 0.5 was at times higher than 1 (Table 4).Eastern sector with the control mesh 10 × 20 threads cm -2 (green mesh, red vectors) and western sector with the experimental mesh 13 × 30 threads cm -2 (orange mesh, blue vectors).

Air exchange rate
The precision of the average air exchange measurements can be checked by summing inflows and outflows (Boulard et al., 1996) through all the openings' surfaces in order to verify the degree to which mass conservation in the greenhouse is satisfied (Boulard et al., 1997).The mean ventilation flux G M has been calculated from Eq. [13] using velocities scaled with the wind speed Eq. [12] and reducing the average difference between outflow and inflow to 11.8% and 67.0% for the western and eastern sectors, respectively (Table 5).
Surprisingly, although the same methodology was followed and the tests were carried out simultaneously, the average difference between outflow and inflow in the western sector was much less than that in the eastern one.This major difference may well be attributed to the partial blockage at the northern section of the side vent in the eastern sector (Fig. 2), which surely affects the airflow at that vent.More measurement points would therefore be required at that vent.In the western sector, on the other hand, where there are no obstacles, the airflow is more uniform at all vent surfaces, and therefore the number of measurement points is considered to be sufficient.
The air exchange rates observed in the present work (Table 5) are very low in both sectors when compared to recommended values, which for a temperature rise of 5°C between exit and inlet air temperatures vary from 0.02 m 3 s -1 m -2 to 0.09 m 3 s -1 m -2 (15 to 65 h -1 for a greenhouse with an average height of 5 m) depending on the solar radiation and the crop transpiration (ASABE Standards, 2008), with optimum values of between 45 and 60 h -1 (Hellickson & Walker, 1983).An air exchange rate of at least 45 h -1 would limit the inside-outside thermal gradient to approximately 5°C (Molina-Aiz, 2010).There are two main reasons for these low rates: (i) the combination of vents in the greenhouse gives rise to opposing wind and thermal effects; (ii) the insect-proof screens drastically reduce the discharge coefficients C d at the vents as compared to the obtained values of the coeff icient C d,LH , which is an approximation of the values at vents without screens (Table 3).Below we indicate the values of air exchange rates observed by other authors using similar methodology, i.e. sonic anemometers in greenhouses with insect-proof screens: in an Almeríatype greenhouse (S A = 1,750 m 2 ; screen porosity = = 34%), with two side vents and two roof vents, Molina-Aiz et al. (2009) observed values for R M from 5.0 to 16.8 h -1 (for u o from 3.34 to 8.40 m s -1 , respectively); in a mono-span greenhouse (S A = 74.4m 2 ; screen porosity = 35%), with two side vents, Teitel et al. (2008) observed values for R M of 20.7 h -1 (u o = 4.5) and 15.7 h -1 (u o = 4.6 m s -1 ).The R M values observed in the present work are similar to those found by Molina-Aiz et al. (2009) at similar wind speeds.The higher air exchange rate recorded by Teitel et al. (2008)

Microclimate evaluation of a new design of insect-proof screens 347
Table 5. Ventilation volumetric flow rates [m 3 s -1 ] through each vent opening calculated from Equation [3] with values of longitudinal component of air velocitiy u * x corrected with wind speed: leeward side G LS , windward side G WS , and windward roof G WR .Error in the calculation of ventilation flow rates E G [%] and air exchange rate due to the small dimensions of the greenhouse used, with a distance between vents of only 9.3 m and a greenhouse volume of approximately 300 m 3 .The lower porosity of the control screen (33.5% as opposed to 39% for the experimental one) may well be the main cause of the reduction in air exchange rate observed in the eastern sector (Table 5, Fig. 7a).However, the partial obstacle at the northern end of the eastern sector might also contribute to the reduction.Bearing in mind the exchange rate R M calculated from the average flow G M , a 16% mean reduction in R M is found in the eastern sector (with a ratio R M-10 × 20 /R M-13 × 30 = = 0.84).The maximum reduction in the air exchange rate, R M = 39%, was recorded in test 3 (Table 5).
The results of the wind tunnel tests indicated that the different aerodynamic characteristics of the insectproof screens lead to a theoretical 11% reduction in the ventilation capacity of the eastern sector compared to the western one (C d-10 × 20 /C d-13 × 30 = 0.89).The difference between C d-10 × 20 /C d-13 × 30 = 0.89 and R M-10 × 20 /R M-13 × 30 = 0.84 is thought to be due in the main to the influence of the obstacle that partially blocks the eastern sector of the greenhouse.In addition, it is important to remember that the ventilation surface in the eastern sector (S V /S A = 11.6) is slightly larger than that in the western one (S V /S A = 11.3).
The difference in the ventilation rate observed for the two types of mesh tested (5% difference in porosity) differs considerably from the results of Harmanto et al. (2006), who observed that two different meshes (30% and 38% porosity) reduced the ventilation rate by 50% and 35%, respectively, when compared to a third mesh of 41% porosity.Fatnassi et al. (2003) found that a mesh of 19% porosity reduced the air velocity inside the greenhouse by 50% as compared to a mesh of 69% porosity.The differences between the re-sults recorded by Harmanto et al. (2006) and Fatnassi et al. (2003) are due to the different experimental methods employed.Harmanto et al. (2006) worked with water vapor mass balance and energy balance methods; while Fatnassi et al. (2003) worked with simulations based on Computer Fluid Dynamics (CFD).The method followed and the placement of the sensors in each case can have a significant bearing on the results obtained.Van Buggenhout et al. (2009) studied the influence of the position of sensors on results when the tracer gas method is employed, and they found that the errors involved reached up to 86%.This fact should therefore be taken into account, especially when the tracer gas method is used to validate CFD simulations, as is the case of Fatnassi et al. (2003).

Interior microclimate
The greenhouse microclimate was monitored while the experimental measurements were being carried out (approx. 2 h).The 3D anemometers were used to determine the temperature at each of the measurement points in the side vents consecutively for 3 min each.The effect of changes in outside temperature throughout the duration of the tests should be considered.To do so, we have used the average inside and outside temperatures as the parameter to scale the inside temperature measured with the anemometers (López et al., 2012): [14] where T sc,j is the corrected sonic temperature (Eq.[11]) for position j in the lateral vents, T o and T i are the mean outside and inside air temperatures during the test and T o,j and T i,j are the mean outside and inside air ▫, western sector with the experimental mesh 13 × 30 threads cm -2 ; ⅙, eastern sector with the control mesh 10 × 20 threads cm -2 .
temperatures recorded by the fixed sensors over the 3 min used for measurement at position j.
In order to judge objectively the effect of ventilation on the inside temperature, we have determined the ventilation efficiency for the temperature η T (Qingyan et al., 1988): [15] where T i-o is the mean temperature of the air that leaves the greenhouse through the side vents and ΔT io is the difference in temperature between inside and outside the greenhouse during the course of the experimental measurements.The term η T represents the effectiveness to eliminate heat of the greenhouse area occupied by the crop.A value of η T equal to 1 indicates that the outside air enters into greenhouse and it mixes perfectly with the inside air.
A mean drop in the inside temperature of 0.5°C was observed in the sector with the experimental mesh, compared to the sector that was f itted with the commercial model (Table 6).The maximum temperature difference between both sectors (0.9°C) was recorded during experimental measurement number 3. The lower air exchange rate recorded in the sector with the commercial mesh means that there is less capacity of natural ventilation to force out the heat from the greenhouse.The ventilation efficiency was always greater in the sector which was fitted with the experimental mesh (mean value of η T = 0.9) than in the one with the commercial mesh (mean value of η T = 0.6) (Fig. 7b).
In the western sector with the experimental mesh, where there was no obstacle by the leeward facing side vent, the air that entered the greenhouse through the roof vent mixed better with the air in the crop zone and left the greenhouse through the side vents, at higher temperature than the air in the eastern sector with the control mesh.
The results obtained in the present work confirm the mean increase in temperature observed by Harmanto et al. (2006), i.e. 0.1°C for each percentage point difference in the porosity of the screens.The increase in the temperature difference ΔT io due to the lower porosity of the control mesh, expressed as the ratio ΔT io-10 × 20 /ΔT io-13 × 30 = 1.16, is similar to the value of 1.25 recorded by Kittas et al. (2002), who compared meshes of de 50 and 60%.Fatnassi et al. (2003Fatnassi et al. ( , 2006) ) obtained different values for the same ratio, 1.50, 1.43 and 2.67, recorded between meshes of 29 and 40%; of 20 and 41%, and of 56 and 69% porosity, respectively.
The statistical analysis based on the multiple regression between the porosity of the mesh ϕ (per 1%), the inside and outside temperatures T i and T o (°C) and the air velocity u o (m s -1 ) provide us with the following expression: [16] The inside temperature increases with the outside temperature and decreases with the air velocity and porosity.A 10% increase in the mesh porosity gives rise to a 0.9°C drop in inside temperature, while a 1 m s -1 increase in air velocity produces a 2.5°C temperature drop.When the difference in porosity between meshes is not so marked, as in this study, the wind velocity is more influential.However, when the ventilation of the greenhouse is due to the thermal effect, on days with little wind, higher porosity of the insectproof screen proves essential.The results obtained indicate that reducing the porosity of the insect-proof screen also reduces the air exchange rate.Consequently, the porosity can also exert a considerable influence on the greenhouse inside temperature (increasing it by 0.1°C per percentage difference in mesh porosity).This negative effect on the greenhouse microclimate is particularly notable around midday when the vents are fully open.Nevertheless, during the rest of the day the negative effect can be offset by opening the windows wider.In previous works analyzing four crop cycles (2007-09) in the same greenhouse, the experimental mesh was not found to cause significant variations in mean temperature or yield (Molina-Aiz et al., 2012), but it did reduce the incidence of whitefly (Escamirosa, 2009).
As final conclusions, by using finer threads, insectproof screens can be designed to combine smaller pore size and greater porosity; this results in improvements in the greenhouse microclimate due to the improved natural ventilation, which is a key feature of Mediterranean greenhouses.
-(ii) does not have a negative effect on the greenhouse airflow pattern.
-(iii) produces a mean increase of 16% in the air exchange rate.
-(v) improves the ventilation efficiency for the temperature η T (mean value of 0.9 as compared to 0.6 with the commercial mesh).
Insect-proof screens with smaller thread diameter enable a greater mesh porosity, which improves greenhouse ventilation when greater cooling is required (around midday).At the same time the pore size is reduced, which prevents the entrance of pests and the exit of beneficial insects used in Integrated Pest Management, a widespread practice in Almería's greenhouses.

Figure 2 .
Figure 2. Location of the experimental greenhouse at the farm.

352 Figure 6 .
Figure 6.Airflow at the vents during experimental measurement number 1 (height of measurements at the side vents 2.13 m).Eastern sector with the control mesh 10 × 20 threads cm -2 (green mesh, red vectors) and western sector with the experimental mesh 13 × 30 threads cm -2 (orange mesh, blue vectors).

352 Figure 7 .
Figure7.Air exchange rate R M (a) and ventilation efficiency for the temperature η T (b).▫, western sector with the experimental mesh 13 × 30 threads cm -2 ; ⅙, eastern sector with the control mesh 10 × 20 threads cm -2 .

Table 3 .
Aerodynamic characteristics of the insect-proof screens: e, thickness (µm); a, b and c are the coefficients of the polynomial fit from the wind tunnel tests (Eq.[5]);

Table 6 .
Microclimate conditions inside greenhouses for the different tests (average values ± standard deviation): T i , inside temperature [°C]; T WS and T LS , air temperature [°C] near windward and leeward vents, respectively (measured with 3D sonic anemometers and corrected with Equation [1]).ΔT io , inside to outside temperature difference [°C]; η T ventilation efficiency for the temperature.N, experimental measurement number * Indicate statistically significant differences between eastern and western sectors (95% confidence level).