Wind Tunnel Simulation of the Atmospheric Boundary Layer for Studying the Wind Pattern at Centro de Lançamento de Alcântara

CLA Centro de Lançamento de Alcântara ABL Atmospheric Boundary Layer Boundary layer thickness H Wind tunnel height IBL Internal boundary layer Iu Turbulence intensity l Coastal cliff height MIT Mobile integration tower PIV Particle image velocimetry TA-2 Aeronautic Wind Tunnel of Institute of Aeronautics and Space u(z) L U(z),U(zr ) Mean velocities corresponding to heights z and zr Uinf Free stream velocity u* Friction velocity zr Reference height W Wind tunnel width INTRODUCTION


INTRODUCTION
The majority of the Brazilian rockets are launched from the Centro de Lançamento de Alcântara (CLA), which has a privileged geographical location, 2º 18'S that enables the operation of suborbital vehicles and satellites with safety launchings in several directions over the Atlantic Ocean (Pires et al., 2008;Avelar et al., 2010;Fisch et al., 2010, Pires et al. 2010).An effective use of the launch opportunities at CLA is possible due to the climate conditions with a demographical density allows the displacement of several sites for launching or logistic support.However, despite the many favorable aspects, mainly because of its proximity to the Equator, the launching center has a peculiar topography due to the existence of a coastal cliff with 40m height (Fig. 1), which can modify the atmospheric boundary layer (ABL) characteristics and consequently affect the safety of rocket launching operations, since the rockets launching pad and the place where the space vehicles are assembled, i.e., mobile integration tower (MIT), are located around 150 to 200m from the border, respectively.Another important physical feature occurrence at the CLA is the formation of an internal boundary layer (IBL) as a consequence of the surface roughness variation, from ocean surface to continental terrain.The wind blowing from the oceanic smooth surface interacts with the low woodland vegetation modifying itself with the formation of an IBL (Pires, 2009), which makes the study of region even more important.The simulation of an ABL in a wind tunnel with short-test section is quite complicated and there are several methods for this purpose discussed in the literature (Counihan, 1969).A simple way of generating thick boundary layers is by using passive methods (Barbosa et al., 2002;Loredo-Souza et al. a combination of spires, wedges or grids together with roughness elements distributed on the wall.Ten possible ways of simulating neutral, stable, and unstable atmospheric conditions in different wind tunnel types were described in Hunt and Fernholz (1975).A short review of the techniques used to thicken the boundary layer was presented by Barbosa et al. (2002).Besides, thickening devices with sophisticated geometry were described by Ligrani et al. (1979 and1983).designs, which have motivated researchers to choose satisfactory geometries by trial and error.
ABL physics is very complex, and the main reason for the surface, which occurs primarily through mechanical and thermal mechanisms.The mechanical interaction arises from the friction caused by the wind against the ground surface, and associated turbulence.In the absence of thermal process, the ABL is said to be neutral, and a logarithmic velocity u(z), characterized by the friction velocity u * and the terrain roughness height z o , is expected to be found (Loredo-Souza et al., 2004).According to Barbosa et al. (2000), for wind speeds higher than 10m/s, the turbulence produced buoyancy, therefore thermal effects become negligible.This is the case of CLA, where strong winds are observed during the dry season, from July to December.The ABL and atmospheric from observations, numerical simulations, and wind tunnel measurements (Pires et al., 2008;Avelar et al., 2010;Fisch et al., 2010;Marciotto et al., 2012) The present work is an extended version of a paper recently presented at the fourth AIAA Atmospheric and Space Environments Conference, in New Orleans, from 25 to 28 June 2012, Avelar et al. (2012), and it is also a continuation of a previous study (Avelar et al., 2010), in which the procedures for a boundary layer simulation in a short-test section wind tunnel (TA-2) were described and some preliminary results CLA region, were presented.
Herein, the ABL was simulated using a combination of spires, barrier, and bottom wall surface roughness.The results wind tunnel, TA-2, of the Instituto de Aeronáutica e Espaço, in Brazil, without using screens downstream of the spires, as in a previous work (Avelar et al., 2010).Three values of Reynolds number ( e l ) based on the coastal cliff height, l, ranging from 6.8×10 5 to 2.0×10 6 were considered.The sensitive to small Reynolds number variations.In addition, turbulence measurements from hot-wire techniques have been conducted.Some stereo PIV velocity measurements for the values of Reynolds number considered were also conducted, showing strong recirculation regions behind the TMI, and it small variations of this parameter.

METHODOLOGY
inside the ABL, for example, the logarithmic and power law equations (Arya, 2001).According to the logarithmic law, the vertical variation of the horizontal wind velocity, U, from the surface up to 100 to 150 m, which corresponds to the where, u * : is the friction velocity, : is the Von Kármán constant, z 0 : is the mean terrain roughness, and z r : is assumed to be 10m, which is the height suggested by the World Meteorology Organization to represent the horizontal wind surface.
The friction velocity, u *, is dependent on the wall shear stress, w , consequently being a measure of the logarithmic declivity close to the wall (Loredo-Souza et al., 2004).Such to the surface, however it is extensively employed also in the surface layer up to about 100m above sea level (Garratt, 1994).
where, U(z ref ): is the mean velocity correspondent to a reference height z ref .
The exponent is a characteristic of the type of terrain.It varies from 0.11 for smooth surface as lakes and the ocean to 0.34 for cities with high density of buildings.For the ocean surface, some studies consider between 0.11 (Hsu et al., 1994;Barbosa et al., 2002) and 0.15 (Blessmann, 1973).
Although commonly used, the power law equation has some drawbacks, which were pointed out by Loredo-Souza et al. (2004).Since this equation is valid for any value of z r , the top of the ABL is not recognized in this model.The second issue is that in spite of providing a good representation of the adjustment in Ekman's layer, but not into the surface layer.
In the present work, the power law equation was used obtaining z 0 and u * .In fact, according to Hsu (1988), in situ measurements of the aerodynamic roughness length are not always possible since it is related to both the wind speed and the wave characteristics of the ocean.The value of 0.11 for the exponent was assumed in the power law equation.

Wind tunnel atmospheric boundary layer modeling
The experiments were conducted in TA-2, which is a closed-circuit aeronautic subsonic wind tunnel.Its test section has a 2.10m height, H, and 3.00m width, W. A 1,600 HP motor produces a maximum speed of 120m/s through the test section.Spires, roughness elements, and a barrier positioned downstream of spires were used for simulation of a thick boundary layer.The entrance.The combination of these elements generates the depend on the desired boundary layer characteristics and on the wind tunnel size, and they were calculated following the methodology proposed by Blessmann (1973).
For the boundary layer formation, initially, a set of 180 small blocks with 80×80×20mm was displaced on the wind tunnel bottom wall separated by 150mm.A 200mm high barrier was positioned 350mm downstream of the spires.a=0.11, which was assumed to be the closest of what is found over the ocean (Hsu et al., 1994).
The positions where dynamic pressure measurements were carried out are represented in Fig. 3.
The circle in Fig. 3 is located in the middle of the test section.The distance between the spires and wind tunnel central line was of 7,860mm.

Turbulence measurements
Turbulence measurements were performed for the freestream velocity of, approximately, 40m/s.Mean velocity a constant temperature hot-wire anemometer, from Dantec Dynamics.These measurements were conducted only in the middle of the wind tunnel test section, in the location indicated as P1 in Fig. 3, after the simulation of the atmospheric boundary layer.It was used a straight golden-plated wire probe (55P01).For data collection, a sample rate of 10kHz was used.The measurements were conducted in several vertical positions.A manually controlled device (Fig. 4), which allowed the vertical displacement of the hot-wire probe during the experiment, was also used.Because of a physical limitation of this device, the highest vertical position where turbulence measurements were conducted was 765mm.

Particle Image Velocimetry measurements
topography was installed in the TA-2 test section, and PIV measurements were conducted at the edge of the coastal cliff and around the MIT.In the present study, the coastal cliff slope angle was assumed as 70º with the horizontal plane, and this value was then reproduced in the model.However, since this inclination angle is not constant along the coastal cliff length, as a continuation of the present analysis, other inclinations will be further considered.
Dynamics two-dimensional PIV system (Fig. 5).The system was a double-cavity pulsed laser, Nd:Yag, 15Hz, with an output power of 200mJ per pulse at the wavelength of 532 nm (New Wave Research, Inc.) and two HiSense 4M CCD camera, built by Hamamatsu Photonics, Inc. with acquisition rate of 11Hz, A Nikon f# 2.8 lenses with 105mm of focal length was used.The laser sheet was shot from the wind tunnel top wall, which was replaced by a glass window, and such sheet was produced using cylindrical lens placed at the end of an articulated optical arm, which transmits the laser from its source to the region of focus (ROF).This arm was used to allow the laser sheet displacement over the model.The red circles in Fig. 5 indicate locations where PIV measurements were conducted, at the edge of the cliff and around the TMI.  on an aluminum trail supported by a three-axis-positioning device.The number of image pairs captured per second was 5.6, and around 200 image pairs, from each camera, were averaged for one measurement condition.The instantaneous images were processed using the adaptive correlation option of the commercial software Dynamic Studio, developed by Dantec Dynamics.A 32×32-pixel interrogation window with 50% overlap and moving average validation was used.

RESULTS AND DISCUSSION
The configurations tested for the boundary layer spires, the barrier and the roughness elements were only tested to illustrate the role of these devices for an appropriated As can be noticed from

Turbulence measurement results
Table 1 shows the intensity turbulence, I u , measured for various vertical positions and associated h ratio in the central position of the TA-2 test section, where h is the distance from the the wind tunnel velocity of 40m/s.The turbulence measurements presented in Table 1 is presented in Fig. 24.The turbulence intensity values measured in the generated boundary layer, represented in Fig. 24, are in agreement with the values encountered by Wittwer et al. (2012), who experimentally studied CLA small scale models, 1:400 in the wind tunnel "Joaquim Blessmann" of the laboratory LAC / UF , in Porto Alegre, Brazil.In this study, mean hot-wire anemometer technique.
From Table 1 and Fig. 24, it can be observed that the frequency spectrums, for each vertical position where turbulence measurement were conducted, are shown in Fig. 25.From Fig. 25, it can be observed that in the inertial range the -5/3 Kolmogorov's law is followed by all curves.The PIV measurements were carried out with the in CLA region.From Figs. 27 and 28, it can be observed an already separated, the Reynolds number seems not to play an important role.In fact, according to Larose and D'Auteuil (2006), it is expected that bluff bodies with sharp edges, which is the case of MTI, the aerodynamics characteristics are almost insensitive to Reynolds Number as long as this parameter reaches 10,000.It can be pointed out also that the IBL seems to grow asymptotically.

CONCLUSIONS
Following a previous study on the simulation of the ABL in a short-test section wind tunnel, a combination of passive turbulence generators were tested in the present work.Good horizontal strips were added perpendicularly to the spires in the conventional setup (roughness, barrier, and spires).Whenever the power-law is well-followed, the dimensionless wind speed regime change for the range of speed studied (from 20 to 40m/s).around the step corner, representing the coastal cliff, and around the MIT.For the range of Reynolds number tested, pattern.In both cases, a very turbulent wake was downstream observed.A future analysis of this research will compare wind

Figure 1 .
Figure 1.A general view of Centro de Lançamento de Alcântara.

Figure 4 .
Figure 4. Hot-wire probe in the TA-2 test section.

Figures
Figures 18 to 21 were included to show some velocity

2
Figure 26.Schematic representation of the Centro de Lançamento de Âlcantara physical model.
Figure 28.Particle image velocimetry results around the mobile integration tower for different e l values.