Agricultural and Forest Meteorology 96 (1999) 181188

Air speed proles in a naturally ventilated greenhouse with a tomato crop

S. Wang*, T. Boulard, R. Haxaire
de Bioclimatologie, I.N.R.A., Domaine St Paul, Site Agroparc, 84914 Avignon Cedex 9, France Unite Received 17 March 1999; received in revised form 14 June 1999; accepted 22 June 1999

Abstract Air speed distribution is a key factor inuencing heat and mass transfer in greenhouses. The air speed proles in the centre of a naturally ventilated greenhouse with a tomato crop were investigated by means of a customised multi-point twodimensional sonic anemometer system. The experimental results showed that air speed was linearly dependent both on external wind speed and greenhouse ventilation ux. Under leakage ventilation, however, the air speed remained nearly constant at a low value (<0.1 m s1), due to negligible wind and temperature effects on greenhouse air exchange rate. It was also shown that in the case of continuous openings parallel to the dominant wind, the air speed was smaller in the vent opening of the span situated at the greenhouse periphery than in the opening of the span situated in the middle of the greenhouse. Based on the measured air speed proles and ventilation ux, an estimation method for calculating the average internal air speed was established. This method can be applied to crop aerodynamic resistance calculations and irrigation management. # 1999 Elsevier Science B.V. All rights reserved.
Keywords: Airow; Natural ventilation; Greenhouse; Wind effect; Climate

1. Introduction The heat transport phenomenon is a `macro-scale' process characterised by an `aerodynamic resistance'. This resistance is not easy to quantify in greenhouses, since convective cells have been poorly documented. Data on actual greenhouse internal convection in the presence of a crop are scarce.

Corresponding author. Tel.: +33-4903-16077; fax: +33-490899810 E-mail address: wang@avignon.inra.fr (S. Wang)

Ventilation processes induce an air exchange between the interior air of a greenhouse and its external environment due to wind and temperature effects. Air movement provided by natural ventilation inuenced the convective heat exchange between the vegetation and the interior air, and thus the microclimate around the vegetation. Most recent studies on natural ventilation have used tracer gas techniques (Bot, 1983; De Jong, 1990; Fernandez and Bailey, 1992; Boulard and Draoui, 1995) and energy balance models (Fernandez and Bailey, 1992; Wang and Deltour, 1996). These two approaches, however, do not allow the airow patterns in greenhouses to be determined.

0168-1923/99/$ see front matter # 1999 Elsevier Science B.V. All rights reserved. PII: S 0 1 6 8 - 1 9 2 3 ( 9 9 ) 0 0 0 6 3 - 5

182

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188

Fernandez and Bailey (1994) investigated the daytime three dimensional distribution of air velocities, temperatures, humidity and carbon dioxide content, in a greenhouse planted with rows of tomatoes. They observed that this tall crop moderated air speeds measured in the upper greenhouse space. More recently, Boulard et al. (1996, 1997) and Wang (1998) measured air velocity directly by sonic anemometry in the opening and inside large empty greenhouses. However, no data were provided to support the hypothesis of a direct relation between greenhouse air exchange rate and inside average air speed. The objective of the present study is to investigate air speed proles in the centre of a greenhousewith a tomato crop. Air speeds at different heights are related to air ux within a greenhouse, external wind speed, buoyancy force and vent opening angle. The experimental results are compared with airow patterns already studied in the same greenhouse. Based on measurements of air speed proles, a method allowing an estimation of the average air speed in the greenhouse is proposed.

2. Experimental set-up 2.1. Site and greenhouse descriptions The experimental greenhouse was located in Avignon (448N latitude, 58E longitude and 24 m altitude), in the south of France. This region is characterised by a predominant northerly wind channelled ne valley. Measurements were performed by the Rho during 2 weeks (1022 August) in 1998 in a 416 m2 two-span greenhouse, equipped with two continuous roof vents (Fig. 1). The greenhouse contained a mature tomato crop, growing on a substrate of rock wool slabs. It was planted in two simple rows at two sides and ve double rows in the centre of the greenhouse. The distance for corridor between rows was 0.8 m. During the measurement period, its height and leaf area index were about 2.1 m and 2.3, respectively. Natural ventilation was provided when the temperature exceeded 248C. Neither heating nor misting systems were used.

Fig. 1. Schematic plan of the greenhouse with the measurement locations of sonic anemometers and climatic parameters (Ue, external wind speed; Ti, interior air temperature; , vent opening angle; h, vertical height of the opening; L0, vent length).

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188 Table 1 Statistics of climatic conditions during the measurements (1022 August) Parameters During the day Mean External wind speed, m s External wind directiona, 8 Outside air temperature, 8C Temperature difference, 8C Vent opening angle, 8
a 1

183

During the night Standard deviation 0.8 11 4.2 2.9 7 Mean 0.7 11 22.9 0.1 0 Standard deviation 0.6 14 2.4 1.1 0

1.8 9 29.4 3.0 45

08 is the direction of the north, the angle was countered positively in a easterly direction and negatively in a westerly direction.

2.2. Sonic anemometer and climatic measuring system Air speed proles in the centre of the greenhouse were measured by a synchronised two-dimensional (2D) sonic anemometer system described by Wang et al. (1997). The path length for both arms of each sonic anemometer was 60 cm with an accuracy of about 1.3% for air velocity in the range between 0 and 1.9 m s1. The system was composed of three 2D sonic anemometers aligned along a vertical support at 0.5, 1.25 and 2 m respectively, above the ground (Fig. 1). The two horizontal components of each sonic anemometer were mounted in the same orientation with an angle of 458 to the north to minimise the distortion introduced by the presence of the instrument itself. The data, were sampled with a frequency of 3 Hz and 2D resultant air speeds in the horizontal plane at six positions, were averaged over 15 min and recorded by a micro-computer. The external wind speed and direction were measured by a cup anemometer and wind vane situated 1 m above the greenhouse. Interior and exterior air temperatures were measured by ventilated platinum resistance thermometers situated in the centre and 1 m away from the greenhouse at a height of 1.5 m. The vents opening angle was monitored by a potentiometer. All climatic parameters were sampled every minute and averaged and recorded every 15 min using a data logger (Campbell, CR20). 2.3. Climatic conditions during the measurements Table 1 provides the external wind speed and direction, outside air temperature, interior-exterior air temperature difference and the vent opening angle

prevailing during the day and night over the whole sampling period. The day and night periods were dened based on the time of sunrise and sunset. External wind speed was higher during the day (1.84 m s1) than during the night (0.70 m s1). The constancy of wind direction from the north was observed due to the dominant regional wind the Mistral. The vents opening angle reached almost the maximum value during the day but dropped to 08 during the night. The vent opening angle was controlled by a climate microcomputer via a temperature sensor and a temperature set-point. The interior-exterior air temperature difference was nearly 38C when the vents were open but only 0.088C while the vents were closed. These signicantly different situations between day and night periods were well suited for evaluating the inside air speed induced by natural ventilation. 3. Results and analyses 3.1. Time variation of inside air speed Values for air speed averaged over 15 min periods, calculated from q the two components of sonic mea 2 V 2 ), were compared with the surements (V Vx y climatic measurements. Fig. 2 shows inside air speeds at positions 1, 3, 4 and 6 and the external wind speed (a) for 20 August, from 0 to 18 h, together with the vent opening angle (b). Before the vents were open, air speeds inside the greenhouse was smaller than 0.1 m s1 and independent of the external wind speed. When the vents were open during the day, there was a strong dependence of internal air speeds on the opening angle and the external wind speed.

184

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188 Table 2 The linear regression equations with R2 values for air speed (V) versus the external wind speed (Ue) at six positions from 208 observations under maximum opening angle Position 1 2 3 4 5 6 Linear regression equation V1 = 0.049 + 0.070Ue V2 = 0.008 + 0.098Ue V3 = 0.007 + 0.106Ue V4 = 0.056 + 0.082Ue V5 = 0.041 + 0.080Ue V6 = 0.085 + 0.190Ue Coefficient of determination (R2) 0.71 0.81 0.83 0.75 0.71 0.84

Fig. 2. Inside and external air speeds and vent opening angle for 20 August, (a) air speeds at positions 1, 3, 4 and 6 (V1, V3, V4 and V6) and external wind speed (Ue), (b) vent opening angle.

3.2. Effect of wind speed under maximum opening angle During the measurements in August 1998, the opening angle of roof vents always reached the maximum value (50.68) during the day, due to the strong solar radiation and high air temperature. In these conditions, the wind effect on internal air speeds was predominant and it was investigated rst. The relationship between V at different heights and the external wind speed was well tted by a straight line. Fig. 3af show the relationship between external and internal horizontal air speeds measured at six positions. Internal air speeds were proportional to the external wind speed, with coefcients of determination R2 ranging between 0.71 and 0.84 (Table 2). The slope of the relationship tended to increase from higher to lower levels and tended to be larger in the west span than that in the east span. 3.3. Effect of ventilation flux under maximum opening angle Air exchange between the greenhouse and its environment can be derived from the two main driving forces of natural ventilation wind and stack effects. For greenhouses and ventilation systems similar to

those in this study, both effects have been found to be of the same importance when the wind speed was lower than approximately 2 m s1 (Boulard and Draoui, 1995). Therefore, the relationship accounting for the combination of thermal and wind effects presented by Boulard and Baille (1995) was used to calculate the ventilation ux (0v in m3 s1): 4 5 3a2 L0 Cd Te gDT 2 2 3a2 h Cw Ue Cw Ue 0v Te 3gD T (1) where Cd and Cw are empirical discharge and wind effect coefficients, identified for this greenhouse as 0.644 and 0.09 by a tracer gas technique (Boulard and Baille, 1995), respectively; g is a gravity constant (m s2); h is the vertical height of the vent opening (m); L0 is the length of the continuous vents (m); Te and T are the exterior air temperature and the interior-exterior air temperature difference (K) and Ue is the external wind speed (m s1). Internal air speeds were linearly related to 0v (Fig. 4af) with signicant improvements in R2 at each position (Table 3) compared to the external wind speed relationship. It emphasises the importance of the combining wind and temperature effects on greenhouse air movements in summer. 3.4. Air speed under leakage ventilation Mean values and standard deviations of air speeds under leakage ventilation conditions (when the vents were closed) at six positions are listed in Table 4. It shows that the average greenhouse air speed during the night was much lower than during the day. During the

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188

185

Fig. 3. Horizontal air speeds at positions 16 (af) as a function of the external wind speed.

Fig. 4. 2D air speeds at positions 16 (af) as a function of the ventilation flux.

186

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188

Table 3 The linear regression equations with R2 values for air speed (V) versus the ventilation flux (0v) at six positions from 208 observations under maximum opening angle Position 1 2 3 4 5 6 Linear regression equation V1 = 0.034 + 0.0130v V2 = 0.012 + 0.0180v V3 = 0.016 + 0.0200v V4 = 0.039 + 0.0150v V5 = 0.026 + 0.0150v V6 = 0.127 + 0.0360v Coefficient of determination (R2) 0.73 0.82 0.86 0.75 0.71 0.87

Table 4 The mean results of air speed at six positions from 272 observations under leakage ventilation. Position 1 2 3 4 5 6 Mean value (m s1) 0.05 0.06 0.07 0.07 0.07 0.06 Standard deviation (m s1) 0.01 0.02 0.01 0.01 0.02 0.02 Fig. 5. Airflow pattern for a northerly wind (Ue) in the horizontal plane at 3.5 m height (vent opening level). Modules of vectors are proportional to the air speeds (after Haxaire et al., 1999).

night, the vents were closed and the air speed in the greenhouse was dependent on leakage and on natural convection induced by buoyancy forces due to the temperature difference between cold surface (greenhouse roof) and warmer surface (soil surface). The temperature difference between the soil surface and the roof surface was roughly proportional to the T between inside and outside. No signicant effect on inside air speeds was observed for either the temperature difference (between 0 and 2.18C) or the external wind speed (between 0.3 and 2.3 m s1). 3.5. Comparison with the airflow pattern determined in previous studies Fig. 5 shows the air ow pattern obtained using sonic anemometers in a horizontal plane situated at the level of the vent opening with the northerly predominant wind (Haxaire et al., 1999). It shows an inow at the downwind end of the ventilator and an outow at the windward end together with a general south to

north air circulation at plant level along the entire length of the greenhouse. It also shows that air circulation was much more important in the west vent than in the east opening. Since the air speed was higher at position 6 than at position 3, our continuous direct measurement of air speed at plant level (Table 2) conrms these previous measurements. 4. Discussion The heat and mass transfer between vegetation and interior air are largely dependent on the value of crop aerodynamic resistance. This resistance is mainly related to a mean inside air speed, assumed constant in most energy balance models (Bot, 1983; Kindelan, 1980; Wang and Deltour, 1997). However, this is only true when the greenhouse is closed or when natural ventilation is maintained at a small and constant value. The present results suggest that inside air speeds are well related to the ventilation ux Eq. (1). From analyses of the airow pattern in the greenhouse (Boulard et al., 1997; Mistriotis et al., 1997), vertical air speed proles perpendicular to the airow can

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188

187

5. Conclusions The average air speed value in greenhouses is a key factor for calculating heat transfers between greenhouse components and the interior air. Experimental air speed proles in the centre of a naturally ventilated greenhouse with a tomato crop were investigated at two locations by using six sonic anemometers. For the maximum roof opening angle, inside air speeds were highly correlated to the external wind speed (R2 = 0.710.84). When the air speed was related to the ventilation ux deduced from the previous models for the same greenhouse, the coefcients of determination (R2) were again improved. For the minimum roof opening angle, no signicant relationship between the air speed and the external wind speed or the interior-exterior air temperature difference was found because air speeds in the greenhouse were maintained at a low constant value. Based on the experimental airow pattern, an estimation method for calculating the average inside greenhouse air speed at crop level can be established from the global ventilation ux and the section area perpendicular to the airow. This relationship can be applied to crop aerodynamic resistance calculations, then to irrigation control. Acknowledgements The authors gratefully acknowledge Prof. J. Deltour for his assistance with the sonic anemometer system during the entire measurements, Prof. David B. Hannaway and Dr. K. Grifn for their helpful discussions in the preparation of the manuscript. References
Bot, G.P.A., 1983. Greenhouse climate: from physical process to a dynamic model. Ph.D. Thesis, Agric. Univ. Wageningen, The Netherlands. Boulard, T., Baille, A., 1995. Modeling of air exchange rate in a greenhouse equipped with continuous roof vents. J. Agric. Eng. Res. 61, 3748. Boulard, T., Draoui, B., 1995. Natural ventilation of a greenhouse with continuous roof vents: measurements and data analysis. J. Agric. Eng. Res. 61, 2736. Boulard, T., Meneses, J.F., Mermier, M., Papadakis, G., 1996. The mechanisms involved in the natural ventilation of greenhouses. Agric. For. Meteorol. 79, 6177.

Fig. 6. Comparison of average air speed between measurement and estimation from 208 observations. The straight line (Vcal = 0.036 + 0.837 Vmea) was obtained by linear regression (R2 = 0.87).

be used to estimate the mean greenhouse air speed (Vcal in m s1): Vcal 0v A (2)

where A is the greenhouse section area perpendicular to the ventilation flux. In this study, it was the vertical cross section area (13 4.08 m2) situated at the level of the speed profile measurements. This simple calculation was compared with the average air speed value deduced from the measurements at the six positions (Fig. 6). There was a good correlation between the measured and estimated air speeds (R2 = 0.87), with a small underestimation (slope = 0.837). This underestimation was probably caused by a larger airflow through the measured corridor section than through the tomato crop row. If the effect of different crops of varying density on the airflow could be taken into account, the improvement of the estimated average air speed in greenhouses can be made. It implies that the section area may be divided into crop rows part and void part. The crop row can be treated as a porous medium and its porosity can be introduced in the calculation of the discharge coefficient of the greenhouse. In this study, the simple Eq. (2) was acceptably used to estimate the average internal air speed due to low leaf area index (LAI = 2.3) and the associated high porosity of the tomato crop during the measurements.

188

S. Wang et al. / Agricultural and Forest Meteorology 96 (1999) 181188 Kindelan, M., 1980. Dynamic modelling of greenhouse environment. Trans. ASAE 23, 12321239. Mistriotis, A., Bot, G.P.A., Picuno, P., Scarascia-Mugnozza, G., 1997. Analysis of the efficiency of greenhouse ventilation using computational fluid dynamics. Agric. For. Meteorol. 85, 217 228. Wang, S., 1998. Measurement and modelling of natural ventilation in a large Venlo-type greenhouse. Ph.D. Thesis, Gembloux, Belgique. Wang, S., Deltour, J., 1996. An experimental ventilation function for large greenhouses based on a dynamic energy balance model. Agric. Eng. J. 5, 103112. Wang, S., Yernaux, M., Deltour, J., 1997. A multi-point ultrasonic anemometer system for measurement of airflow distribution. In: Proc. Third Int. Conf. on Fluid Dynamic Measurement and Its Applications, Int. Acad. Publ., 355359.

Boulard, T., Papadakis, G., Kittas, C., Mermier, M., 1997. Air flow and associated sensible heat exchanges in a naturally ventilated greenhouse. Agric. For. Meteorol. 88, 111119. De Jong, T., 1990. Natural ventilation of large multi-span greenhouses. Ph.D. Thesis, Agric. Univ. Wageningen, The Netherlands. Fernandez, J.E., Bailey, B.J., 1992. Measurement and prediction of greenhouse ventilation rates. Agric. For. Meteorol. 58, 229 245. Fernandez, J.E., Bailey, B.J., 1994. The influence of Fans on environmental conditions in greenhouses. J. Agric. Eng. Res. 58, 201210. Haxaire, R., Boulard, T., Mermier, M., 1999. Airflow, temperature and humidity patterns induced by wind effect driven ventilation in a two-span plastic house. Acta Horti., To appear.