Sloan Digital Sky Survey Telescope Enclosure: Flow
Visualization
Sloan Digital Sky Survey Telescope Technical Note
19940310
Charles H. Comfort, Jr., Mark Matheson, Siri
Limmongkol and Walter
Siegmund
Abstract
The telescope enclosure for the Sloan Digital Sky Survey 2.5 meter
telescope is the first roll-off design used on a telescope of this
size. The roll-off design eliminates dome-induced image degradation
because the telescope enclosure is rolled downwind of the telescope
during observations.
Flow visualization techniques were used to study a 1/72 scale
model of the enclosure in a water tunnel. The overall slope and the
vegetation present at the telescope site were represented in the
model. Results indicate that flow separates at the leading edge of
the telescope platform but that the turbulent boundary layer remains
below the level of the telescope and should not contribute to image
degradation. A close fitting telescope wind and light baffle is
intended to reduce wind-induced tracking error and scattered light
from off-axis sources. A model of a portion of the baffle was
examined, also. The baffle is very effective in reducing flow
velocity downstream of the baffle. Although the velocity of the flow
through openings in the baffle is high, these jets diffuse within 0.1
to 0.2 meters of the baffle. The resulting flow should have little
effect on telescope tracking while providing excellent flushing of
the telescope light path.
Introduction
The preservation of image quality is one of the most important
design criteria for a telescope enclosure. Image quality is adversely
affected by the presence of fluctuations of air temperature in the
light path. Pressure fluctuations, e.g., due to turbulent flow, do
not significantly degrade images in ground-based telescopes. Ideally,
the air in the telescope light path is continually replaced by
ambient air before it has the opportunity to be warmed or cooled by
the surfaces near the telescope.
A telescope enclosure represents a significant cost of a telescope
project. Unfortunately, a poor telescope enclosure design can degrade
the image quality of an otherwise excellent site. A better
understanding of the interaction of the fluid and the enclosure
should lead to improvements in the design of the telescope enclosures
and preservation of the intrinsic image quality of the telescope
site.
The SDSS 2.5-m telescope enclosure is designed to roll away from
the telescope leaving the telescope totally exposed to the prevailing
wind (see Sloan Digital Sky Survey telescope enclosure: design in
these Proceedings). Large doors at each end provide access for
servicing and allow the building to pass over the telescope. These
doors remain open during operation. This reduces the effect of the
building on upstream air flow. Mitigation of wind-induced tracking
error is provided by a wind baffle (Figure 1) that fits closely
around the telescope and that is also a component in the telescope
light baffling system. The wind baffle is driven in two axes
separately from the telescope and is supported by the telescope
building. The wind baffle drive systems are electronically slaved to
the telescope axes. The wind baffle is covered with panels (H. H.
Robertson model 5100) that are 25% porous to the wind. This allows
excellent flushing of the telescope optical path while reducing
wind-induced forces on the telescope. (These panels have the
desirable property of providing excellent light baffling.) The end of
the wind baffle is closed except for an annular opening that provides
clearance for light to enter the telescope. The inner edge of the
annular opening is defined by a circular plate that is supported from
the rest of the wind baffle by narrow vanes.
Figure 1. The SDSS 2.5-m telescope enclosure.
The telescope enclosure rolls away from the telescope on crane rails
to expose the telescope to the wind. A wind and light baffle, driven
and supported independent of the telescope, fits closely around the
telescope.
Description of the Experiments
This research was performed using the 13 m long water tunnel of
the Department of Aeronautics and Astronautics of the University of
Washington. The test section is 3 m long and 0.76 m square. The
maximum flow speed is 0.6 m/s. The flow speed used was 10 cm per
second. This scales to 7.2 m/s (16 mph). The Reynolds number is the
most important parameter affecting flow in this system. In
particular, typical wind speeds are high enough that buoyancy effects
are negligible in a reasonable telescope enclosure design. The
Reynolds number is
Re = VD/µ
where V is the flow velocity, D is a characteristic dimension, and
µ is the fluid kinematic viscosity (0.9E-6 m^2/s for water and
18E-6 m^2/s for air). Typically, the Reynolds number for a scale
model is too small because of the dependence of V and D on the scale.
The lower viscosity of water partially offsets the effect of the
scale change, making a water tunnel an ideal environment for
examining flow characteristics around sharp edged structures.
The Reynolds number for the model system is about 4400 based on
the width of the wind baffle. For the actual structure and typical
wind speeds, the Reynolds number will be 10^6. Fortunately, for a
shape which is not streamlined, flow patterns do not change much once
the Reynolds number is above several thousand. In several cases, we
increased the flow speed to about 50 cm/s to insure that our results
were not sensitive to the Reynolds number.
An acrylic model of the telescope enclosure, telescope and wind
baffle was used. The scale was 1/72. The model was mounted upside
down on a flat base plate 1.58 m long. Wind at site is observed to
flow parallel to the ground, i.e., up the slope. This relationship
was maintained in the water tunnel by mounting the base plate and
model with the proper orientation with respect to the flow direction.
Since flow in the water tunnel is horizontal, this meant that the
telescope enclosure model was installed at an angle. However, the
photographs of Figures 2 to 4 were taken with the camera tilted and
printed upside down so that the orientation of the building would
appear normal.
Realistic modeling of a telescope site must account for the
velocity profile at the site, which in this case includes the
presence of trees. The velocity profile was modeled using molded
plastic trees with approximately the same density and height
distribution found at the site. Plastic screening was added to the
forest to match the velocity profile. Horizontal velocity was
measured at several heights and adjusted to approximate that measured
at the site.
Fluid flow through and around the telescope was examined at
various angles of the flow with respect to the building orientation.
Only the range of angles from 0° to 45° were used (0°
corresponds to the wind coming directly up-slope, west-southwest at
the site). This is not a significant limitation since the wind
direction at the site is most frequently from the corresponding
quadrant and is associated with the best image quality. Also, the
telescope orientation with respect to the building is fixed. The
orientation chosen is believed to be the worst case for shielding of
the telescope by the wind baffle, i.e., with the enclosure oriented
at 0°, the telescope is pointed directly into the wind.
Three types of experiments were performed. In the first, dye was
injected into the flow upstream of the model. Several probe heights
were used for each model; these are chosen to examine the flow at
various levels in and around the telescope. Streamlines flowing
through and around the telescope do not efficiently examine stagnate
spaces since the flow rarely enters those spaces. In the second type
of experiment, dye was injected directly into the telescope wind
baffle and fills the volume in front of the primary mirror. Observing
the rate at which dye disappears from the wind baffle indicates the
flushing rate. Dye remains in stagnant spaces longer than in spaces
which are well flushed. In the third type of experiment, dye was
again injected into the interior of the wind baffle and video taped.
The tape was subsequently analyzed to determine flow velocities near
the primary and secondary mirrors. Features on the model were used to
determine the scale on a television monitor and the frame interval
(1/30 second) provided timing.
In a separate set of experiments, a model of a portion of a wind
baffle panel was examined. The model was approximately 0.2 m square
and its scale was 1/4. Because of the limited extent of the model,
and its low porosity, flow around the model perimeter tended to
affect the flow immediately downstream. However, by controlling the
width of the gap between the model and the base plate and by looking
at flow near the center of the model, we were largely able to
overcome these effects.
Results
In earlier studies, we have found that telescope buildings often
cause streamlines to rise as they approach the telescope.1,2
Since the height of telescope buildings is driven by the desire to
place the telescope high with respect to streamlines, a design that
prevents streamlines from rising may be more cost effective than
increasing the height of the telescope above the ground. Figure 2
shows dye tracing a streamline just above the leading edge of the
telescope platform. The telescope building is at 0°. (The
building orientation is described by the angle between the down-slope
direction and the direction from which flow occurs.) No elevation of
the streamline is apparent near the building. Most likely, this is
due to the presence of the forest that causes a strong gradient of
flow speed with height.
Figure 2. With the enclosure at 0°, dye
traces flow coming up-slope with mild turbulence induced by the
presence of trees. This and similar tests show that the volume
surrounding the wind baffle will be continually flushed by the wind
and that the stagnant volume just above the telescope platform will
be small. This and the other postage stamp images are linked to
detailed 40kB images.
The streamline shown encounters the telescope wind baffle near the
level of the primary mirror. This and other tests show that the
entire exterior of the wind baffle will be very well flushed by the
wind. At the leading edge of the telescope platform, flow separates
and a stagnation region develops over the telescope platform. Even
so, because this region is not enclosed, flushing is very rapid.
Still, we will want to minimize the production of warm or cold air in
this region. This will be more important with the telescope pointed
in the opposite direction since the thickness of the stagnation
region increases downstream.
Figure 3. These photographs, taken 10 seconds
apart, show the very rapid flushing of the wind baffle due to its
porosity and low volume. The full-scale flow speed is 7.2 m/s (16
mph).
Flushing of the telescope wind baffle is very rapid due to its 25%
porosity and low volume. Figure 3 shows that at 0°, flushing is
nearly complete in 10 seconds. Flushing is only slightly slower at
45° and Table 1 shows flushing times at other angles which were
examined.
Table 1. Model enclosure flushing times (as a
function of azimuth angle)
Azimuth Run 1 Run 2 Comments
Angle (sec) (sec)
0° 17 19
15° 24 25
30° 22 24
45° 23 24
60° 28 30 Stagnation regions formed
75° 24 25 on downstream bottom
90° 23 24 corners of enclosure
105° 19 16
120° 13 14 Stagnation region formed
135° 18 15 behind secnodary mirror
150° 17 18
165° 18 19
180° 17 18
Video tape recordings of dye injected into the interior of the
wind baffle were analyzed to determine the effectiveness of the wind
baffle in mitigating wind-induced tracking error. With the building
at 0°, fluid entered the wind baffle through the open end,
passed near the edge of the secondary and rotated in a clockwise
direction between the primary and secondary. The highest velocity
measured was near the edge of the secondary. The average was 0.39 of
the free stream flow velocity (see Table 2). This velocity reduction
corresponds to a dynamic pressure reduction of about 1/7 and is a
large decrease in the wind loading on the telescope. As mentioned
above, this appears to be the worst case orientation of the
telescope. At other orientations, the shielding will be even more
effective. Wind-induced moments on the secondary truss and frame are
a large portion of the total moment about the telescope altitude
axis. These components are very well shielded by the wind baffle so
that only the secondary and its baffles and support package are
likely to be much affected by the wind.
The motivation for examining the detailed model of the wind baffle
panel was to study how rapidly flow diffuses on the downstream side
of the panel. Since the dynamic pressure of the wind goes as the
square of the wind velocity, it is desirable that flow through the
baffle diffuse rapidly to avoid wind-induced tracking error. Many
flow diffusers produce persistent high velocity jets. This particular
design avoids this problem. Figure 4 shows that diffusion is
essentially complete, i.e., comparable to the spacing of panel
openings, about 0.2 m (full scale) downstream of the panel.
Figure 4. Dye approaching this model of a
portion of a wind baffle panel flows through the model and diffuses
very rapidly downstream. The lack of a persistent jet demonstrates
that this panel geometry is very effective at shielding downstream
structures from the wind while providing 25% porosity. The full-scale
thickness of the panel is 0.1 m.
Table 2. Velocities within the wind baffle (as
a fraction of the free stream velocity)
Location Velocity
At edge of secondary 0.39
In front of secondary 0.18
In front of primary 0.21
Conclusions
The roll-off telescope enclosure, in combination with a close
fitting independently driven wind baffle provides an excellent flow
environment for a telescope. The stagnation region above the
telescope platform is well flushed and is not likely to have any
significant effect on image quality even with the telescope pointed
downwind. The exterior of the wind baffle is continually flushed by
the wind. Air flowing through the wind baffle panels and the open end
of the baffle flush the interior of the baffle very rapidly. At the
same time, the wind baffle provides excellent shielding of the
telescope from wind-induced tracking error. The baffle panels provide
good porosity and rapid diffusion downstream.
Acknowledgments
It is a pleasure to thank Bob Breidenthal and Charlie Hull of the
University of Washington who variously contributed ideas,
encouragement and assistance.
References
- W.A. Siegmund, W. Wong, F.F. Forbes, C.H. Comfort, Jr., and S.
Limmongkol, Flow visualization of four 8-m telescope
enclosure designs, Proc. of S.P.I.E. 1236, p. 567,
1990.
- F. Forbes, W.-Y. Wong, J. Baldwin, W. Siegmund, S. Limmongkol,
and C. Comfort, Telescope Enclosure Flow
Visualization, Proc. of S.P.I.E. 1532, 1991.
A paper similar in content to this note is
published as "Sloan Digital Sky Survey telescope enclosure: flow
visualization", C. H. Comfort, Mark Matheson, S. Limmongkol, W. A.
Siegmund, Proc. of S.P.I.E., 2199, 1994, p.1074.
Date created: 3/10/94
Last modified: 5/26/99
Copyright © 1999, Walter A. Siegmund
Walter A. Siegmund
siegmund@astro.washington.edu