four particles were larger than 50 m. A random
108 particles before plugging. Design guidance
sampling of these particles using an SEM/EDX
suggested using a pressure drop of 34 kPa (5 lb
in.2) across the filter to determine pumping re-
found a copper sulfate grain, an alumino silicate
with some CuS on its surface, a fibrous particle
quirements. We expected that this loss would
(probably plastic) with some mineral fragments
dominate all pumping losses.
adhered, a very platey alumino silicate that con-
We estimated the required pumping rate from
tained Mg, Ca, and Fe, possibly a clay, several plas-
the velocity needed to entrain 1-mm-diameter
tic pieces, and four particles containing Fe, Ni, Cr,
sand particles in water (ASCE 1975). The average
Pb, Al, Si, and a trace of S. Most of these particles
velocity in the intake slot must exceed the particle
probably come from the heat exchanger, pump,
fall velocity of about 10 cm/s to carry these par-
and hoses used to operate the well. Assuming that
ticles into the filter. In addition, the friction veloc-
the number of particles found in the 8-L sample is
ity between the filter arm and the ice surface must
representative of what would be found in 2000 m3
exceed the threshold value to initiate particle
movement of about 3 cm s1. This requires that
(the amount of water melted during one year of
the average flow exceed about 15 cm s1, if we as-
operation), that all these particles were from the
well equipment, and that they settle to the bottom
sume that the gap between the filter arm and the
and are collected by our device, 108 terrestrial par-
ice surface is about 1 mm. Both conditions are met
if the average velocity exceeds 30 cm s1 in a 1-
ticles would be added to our collection, or about
mm-wide intake slot (or 15 cm s1 in a 2-mm-wide
1000 terrestrial particles for each extraterrestrial
particle larger than 50 m. Annual collection
intake slot). We selected a compact, submersible
pump that could deliver 1.6 L s1 at 34 kPa or about
would thus yield a total of about 100 g of material.
50 cm/s through a 1-mm-wide 3-m-long intake
The 8-L sample also revealed the presence of
slot. Near zero pressure, the pump delivers about
submicron particles of iron oxide in the well wa-
3 L s1 (100 cm/s through a 1-mm-wide 3-m-
ter. The white, cellulose nitrate membrane filter
long intake slot).
turned a bright orangish color after filtration, and
We expected that the high-pressure drop
energy-dispersive X-ray analyses of the filter
through the filter and the narrow intake slot would
showed only the presence of iron. The rust is
establish relatively uniform intake velocities along
thought to be from the heat exchangers or pump
the slot. Nevertheless, we conducted flow tests on
used for the well. Until January 1994, the water in
a 30-cm-long 13-cm-wide model of the filter arm
the SPWW was acidic (pH 4.8), very soft,* and cor-
(front-slot version, Fig. 3) to verify particle pickup
rosive to metals. NSF has neutralized the water
and assess the velocity distribution. This model
(pH 6.9) by running it through a limestone bed.
This measure may diminish the formation of iron
contained pressure taps that allowed us to mea-
oxide. In addition, a fire retardant (Ansul dust,
sure the pressure drop across the intake, the filter,
and the plenum areas at three cross-sections along
similar to baking soda) was used during the 1994
the collector. We also measured the total flow rate.
well fire. Particles of this fire retardant, soot, and
For these tests, we prepared mixtures of simulated
melted insulation could also have reached the well
pool, although most should have deposited along
extraterrestrial particles consisting of stainless steel
the well neck.
and glass spheres and silica sand, covering a size
range of 50420 m.
We found no appreciable plugging of the filter
(negligible increase in pressure drop across filter,
PUMPING REQUIREMENTS AND
negligible decrease in flow rate) even after collect-
PRELIMINARY TESTS
ing 68 g of material. The maximum pressure drop
across the 53-m filter was 4.2 kPa, and the corre-
Our collector design allowed for almost 1 m2
sponding minimum flow rate was 0.31 L s1 (rep-
of 30-cm-wide filter fabric inside a 3-m-long filter
resenting 100 cm s1 through the 1-mm-wide 30-
arm using a single fold of the filter to create a
pouch. We obtained a polyester filter consisting
cm-long intake slot). The collector model easily
of 53-m mesh openings and 108 openings/m2.
suctioned up particles lying several millimeters in
Thus, 1 m2 of this filter should trap the expected
front of the intake. However, because the flow
velocity drops quickly inside the collector, the col-
lected particles were deposited on the bottom of
the filter pouch downstream of the slot rather than
*H. Mahar, 1996, National Science Foundation, personal
evenly plugging the fabric.
communication.
7
Previous Page
