178 T. Boulard et al. Agricultural and Forest Meteorology 100 2000 169–181
Fig. 15. Microscale λ m distribution of velocity component u in section II.
48 points in section II. An equivalent dependence was earlier signalled by Heber et al. 1996 in a
barn.
4.4.5. Energy spectra Comparison of the energy spectra of the external
wind Fig. 16, in the vent opening at position No. 3 Fig. 17 and that inside the tunnel at position No. 23
in section I Fig. 18, show that the air current in the windward opening was the most turbulent, followed
by the outside wind. The lowest values were found at position no. 23 between two openings. All locations
had similar spectral levels at higher frequencies in the dissipation region. However, the behaviour at lower
frequencies was quite divergent. In the vent opening, spectral densities at lower frequencies followed the
dominance of outside conditions. Most positions with low average air speed showed very similar curves
for u-, v- and w-components see position No. 23,
Fig. 16. Spectral energy distribution of the wind at a height of 1.65 m outside the greenhouse tunnel.
section I, with a spectral decay rate correspond- ing to the high frequency spectral energies equal to
about −53, as it is normally the case for isotropic turbulence.
5. Discussion
In experimental conditions marked by a strong external wind perpendicular to the tunnel axis and
moderate inside–outside temperature and humidity differences the wind driven ventilation flux prevails
substantially over the buoyancy forces de Jong, 1990; Boulard and Draoui, 1995. In these condi-
tions, the measurements demonstrated an intense air current crossing the tunnel between the windward
and leeward openings, while the air along the floor and in the vertical section between two consecutive
series of opening remained still. The air temperature
Fig. 17. Spectral energy distribution of the internal airflow at position 3 in section II.
T. Boulard et al. Agricultural and Forest Meteorology 100 2000 169–181 179
Fig. 18. Spectral energy distribution of the internal airflow at position 23 in section I.
and water vapour distributions were considerably influenced by these fluxes, with a high north–south
gradient due to the cold and dry air penetration through the windward vent opening. The turbulence
intensity notably increased from the centre of the tunnel toward the windward opening, where it was
about ten times larger than in the centre of the tun- nel. These experimental data can be compared with
the experimental air-speed profiles measured at crop level with the same wind conditions wind driven
ventilation in Mistral conditions in a multispan–span greenhouse equipped with roof openings parallel to
the wind direction Wang et al., 1999 ; Haxaire et al., 1999. The average air speeds in the major volume
of the tunnel were ranging between 20 and 80 of the outside wind velocity, against values between 10
and 20 only in the greenhouse equipped with roof openings. We can also observe that, contrary to the
tunnel, the high air speed and turbulence intensities in the multispan greenhouse were not concerning
the zones where the plants grow, but were confined near the roof openings in the upper volume of the
greenhouse.
6. Concluding remarks
With the current design of vent openings, the present results demonstrated that the mean and turbulent wind
conditions within a large part of the greenhouse tunnel volume were similar to those outside, particularly near
the soil surface where the crops would be growing. Plant growth would be enhanced by less heteroge-
neous and turbulent conditions, so there is a require- ment for improved designs of vent opening, which
would reduce the mean wind speed and the turbulence within the greenhouse tunnel. Computational fluid dy-
namics CFD simulations could be a valuable tool for analysing and designing better greenhouse venti-
lation. In this way, the size, position and shape of the vent openings can be designed so that the outdoor air
mixes more smoothly at crop level with the indoor air, without forming regions with a direct penetration of
outside air, or stagnant regions.
As suggested by previous CFD simulations for greenhouses Mistriotis et al., 1997; Boulard et al.,
1998, experimental validation is needed for quanti- tatively analysing the predictive accuracy of CFD in
detail, and particularly the fluctuating component of the flow. The results of the present study provide both,
a high-resolution database with which to validate on-going efforts with computer simulations of the
mean and turbulent characteristics of the greenhouse environment and a move towards a better understand-
ing of the plant environment behaviour under such conditions.
7. Notation