Craters Formed in Granular Beds by Impinging Jets of Gas

Philip T. Metzger , Robert C. Latta, III † , Jason M. Schuler and Christopher D.

Immer
Granular Mechanics and Regolith Operations Laboratory, NASA Kennedy Space Center, Florida 32899, USA
†
Department of Aerospace Engineering, Embry-Riddle Aeronautical University, Daytona Beach, Florida 32114,
USA
ASRC Aerospace, Kennedy Space Center, Florida 32899, USA

Abstract. When a jet of gas impinges vertically on a granular bed and forms a crater, the grains may be moved by several
different mechanisms: viscous erosion, diffused gas eruption, bearing capacity failure, and/or diffusion-driven shearing. The
relative importance of these mechanisms depends upon the flow regime of the gas, the mechanical state of the granular
material, and other physical parameters. Here we report research in two specific regimes: viscous erosion forming scour holes
as a function of particle size and gravity; and bearing capacity failure forming deep transient craters as a function of soil
compaction.
Keywords: crater, erosion, scour hole, jet, bearing capacity, rocket exhaust, martian soil, gravity
PACS: 47.57.Gc, 47.56.+r, 47.55.Kf, 47.55.Lm

INTRODUCTION the soil can shear, then BCF is impossible and only DDF
can occur. However, if the sudden application of a jet
Jets of fluid impinging vertically on granular beds are onto the granular bed produces enough surface load to
scientifically interesting due to the variety of phenom- shear the grains faster than the fluid can diffuse into the
ena that have been observed. The present study was mo- pore spaces, then DDF is impossible and only BCF can
tivated by a desire to understand how landing rockets occur. In general the cratering is intermediate to these
will affect the soil of an extraterrestrial body. Small-scale two extremes. Experiments have shown that BCF causes
experiments help to identify basic relationships in the the grains to move perpendicularly away from the sur-
physics and provide a dataset to benchmark the high- face, whereas DDF causes them to move parallel to the
fidelity numerical flow codes that are being developed. surface, and in the intermediate cases the grains move
So far researchers have identified four basic ways that diagonally away from the surface [4].
an impinging jet of gas (or liquid) can cause a crater to It would be scientifically interesting and important to
form in a granular bed. First, viscous erosion (VE) is the a wide range of engineering applications to map the pa-
process of fluid lifting or rolling the top layer of grains rameter space of these four jet-induced cratering mecha-
along the surface [1], forming a scour-hole in the case of nisms to know where each becomes important and how
a localized jet. Second, diffused gas eruption (DGE) oc- they may interact. To-date this has not been attempted;
curs when the jet pressurizes the interstitial fluid of the research has only sampled the various behaviors at some
granular bed in a radially-expanding subsurface region, convenient points in parameter space. Here we report two
which may cause an eruption of granular material in an such samplings. We report on the scaling of VE forming
annular ring around the jet, or centrally if the jet is sud- scour holes beneath subsonic jets with variation in the
denly extinguished [2]. Third, bearing capacity failure granular particle size and gravity, and we report on the
(BCF) occurs when the dynamic pressure of the imping- insensitivity of BCF to soil properties beneath supersonic
ing jet applies mechanical loading to the top surface of jets.
the granular bed beyond its bearing capacity and thus
shoves it downward to form a depression [3]. Fourth,
diffusion-drivenflow(DDF) occurs when the fluid driven SCOUR HOLES BENEATH SUBSONIC
by a jet through the pore spaces of the soil creates a dis- JETS
tributed body force in the bulk of the soil—the drag force
of the fluid reacting against the grains—sufficient to un- The depth of the scour holes formed by VE under sub-
jam the material and shear it [4]. The shearing of DDF sonic jets was reported by Rajaratnam and Beltaos [5] to
is geometrically different than BCF, which is caused by grow logarithmically over several decades of time (what
forces acting on the free surface of the bed. If the fluid we shall call the “logarithmic period”) before slowing
diffusing through the soil reaches steady-state faster than

CPl 145, Powders and Grains 2009, Proceedings of the 6 International Conference on
Micromechanics of Granular Media edited by M. Nakagawa and S. Luding
© 2009 American Institute of Physics 978-0-7354-0682-7/09/S25.00

767
 Width (µ2.54 cm)
8

0 20 40 60 80 100
Time (sec)

FIGURE 2. Inner and outer craters widths versus time for

four sand sizes (in microns) 200-280 (lightest gray), 280-300,
300-450, and 500-600 (darkest gray). The top set of curves
FIGURE 1. Inner and outer craters beneath a subsonic jet. shows outer crater widths. The bottom set of curves (overlying
one another) shows inner crater widths. The noisiness of the
inner craters is caused by avalanching of the outer craters.
to reach an asymptotic depth. Prior studies have investi-
gated the asymptotic size and shape of these scour holes Depth (µ2.54 cm)

[5, 6, 7] and the transient growth of the holes [4, 8], but
3.0
much of the parameter space is yet to be explored. Here
we have used the experimental method reported in detail
2.5
^^^SC^Z^
in [4], directing a jet of nitrogen gas from a straight cir-
cular pipe into a small sand box so that it forms a crater. 2.0
The controlled parameters are jet velocity, height and di-
ameter of the pipe above the sand, and sand composition. 1.5
The experiment is effectively split in half by replacing
the front wall of the sand box with a viewing window 1.0
and centering the circular pipe over its outwardly beveled
0.5
edge. Thus we can see inside the crater as it forms, as
shown in Fig. 1. The crater’s inner surface is found to
become steeper than the angle of repose because it is 20 40 60 80 100
Time (sec)
supported by the traction of the gas exiting the crater.
Eventually the crater widens and traction is insufficient FIGURE 3. Craters depths versus time for the same set of
to maintain the steep slope, so the upper part of the crater sands described in Fig. 2. The top set of curves shows the
collapses to form an “outer crater” at the angle of repose overall crater depths. The bottom set of curves shows the inner
as shown in Fig. 1 We video-recorded the craters through crater depths, smoothed by a 101-point moving-average filter
the viewing window and measured their size and shape at to remove the noise.
each video frame using automated software. From these
measurements we extracted their scaling behavior.
rithmic period. This maximum width of the inner crater
is found to be independent of particle size, whereas the
outer crater widens faster for smaller particles. Figure 3
Scaling With Particle Size shows that unlike inner crater widths, the inner craters
depths continued to grow throughout the logarithmic pe-
To investigate how cratering scales with particle size riod and their rates were a function of particle size. This
we used quartz “construction sand” that had been sieved was unexpected and lacks an explanation.
to retain various size ranges. For a 0.95 cm diameter pipe, As described in [4], straight lines were fitted to the
7.62 cm height above the sand, and a jet velocity of 34 measurements of overall depth vs. logarithmic time. The
m/s, the crater widths vs. time are shown in Fig. 2. The slope of these straight lines is a, the dominant length-
inner crater width quickly reaches a maximum whereas scale in the logarithmic period. The intercept is alnb,
the outer crater continues to widen throughout the loga- where b is the inverse of the dominant time-scale. As

768
0.6 ~ a [cm] ,-- 51 Crater depth 1/6 g
0.5 ; • experiment [cm] ..." •"""""" 3/8 .g.
a ~ D1/2 • ^ ^ ^ * ™ — —
™ ™
4- .-•***
0.4 ; • •
1 3- :* / 1g
0.3
0.9
0.2 R2 ^ \ 2 f .'

0.8 2g
Power Law e x p o n e n t \
0 0.46 1 2 h
0 200 400 600 800 1000 ,r 5 10 15 20 25
Particle Diameter D [microns] Time (sec)
FIGURE 4. The cratering length-scale a vs. particle size D. FIGURE 6. Overall craters depths versus time for simulated
Inset: the Coefficient of Determination R2 for power laws of D martian soil JSC-Mars-1A at four different gravity levels with
between 0 and 2 (best fit is 0.46 ~ 1/2). all other parameters held constant.

200 r
Seven different soils were used at a variety of jet veloci­
ties and four different gravities: g « 1/6 gee (the Moon),
150 3/8 gee (Mars), 1 gee (Earth), and also 2 gee, which oc­
curred during the “pullouts” of the aircraft between suc­
cessive low-g parabolas. The low-g experiments could
100 b 1
[sec ] \ not go longer than a single parabola because the craters
caved-in during the periods of high-g. Several hundred
experiments were performed and the results are still be­
50
ing analyzed, but preliminarily we report one set of grav­
ities for the simulated martian soil known as JSC-Mars-
1A. It is a volcanic soil with a broad particle size dis­
0 200 400 600 800 tribution from silt to coarse sand. Figure 6 shows the
Particle Diameter D [microns] depths versus time with the expected decrease in crater­
ing rate at higher gravity levels. The 1 gee and 2 gee
FIGURE 5. The inverse time scale b vs. particle size D. experiments achieved asymptotic sizes within the dura­
Inset: the Coefficient of Determination R2 for power laws of tion of a parabola, with lesser gravity trending to deeper
D between 0 and 4 (best fit is 2.02).
asymptotic craters. Fig. 7 shows a and b for the logarith­
mic period at each gravity level. In the 2 gee experiment
the logarithmic period was very brief so a and b may
reported in [4], a varies with the height of the jet exit have larger errors than for the other gravities. Surpris­
plane above the soil, but surprisingly it is independent ingly neither parameter is strongly dependant on gravity,
of jet diameter, jet velocity v, and gas density ρ. How­ showing only a - g-0.25 while the scaling of b is incon­
ever, b varies with all these parameters and notably is clusive. Cohesion probably dominates at low gravity and
proportional to the jet’s dynamic pressure ρv2. This is this is probably responsible for b being non-monotonic
different than the scaling of the densimetric Froude num­ with gravity.
ber Fr ρ1/2v. Here we report in Fig. 4 and 5 the scal­
ing of a and b with respect to grain diameter D. We find
a D1/2 and b D-2. We note that the volumetric ero­
sion rate a3b ρv2/D1/2 has the same scaling as the BEARING CAPACITY FAILURE
Froude number for D but not for ρ or v. BENEATH SUPERSONIC JETS
We also used solid rocket motors to test supersonic jets
Scaling With Gravity impinging on a deep sand bed against a viewing win­
dow. When we placed layers of colored sand in the sand­
We have performed similar experiments on an aircraft box for some of thesefirings,the downward deflection of
flying parabolic trajectories to effectively reduce gravity. layers identified BCF as the dominant cratering mecha-

769
 b [sec 1 ]
70
1.5
•^Mars
60
Earth
50:
Cohesion
1.0
dominates 40:
at low g? -0.8
2 gee
30:

0.5 20
-1.2
10

-1.2 0 0
0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4
-2.0 -1.5 -1.0 -0.5 0.0 0.5
Time [sec]
Ln[Gravity, gee]
FIGURE 8. Depth of crater vs. time for two solid rocket
FIGURE 7. Cratering length-scale b vs. the log of gravity. motor firings into compacted and loosely deposited JSC-Mars-
The dashed line is a guide to the eye. Inset: log-log plot of 1A. Inset: Scorch marks on the viewing window of the sand
cratering inverse-time scale a vs. gravity. The dashed line is bed.
a g-0.25.

CONCLUSIONS
nism. We also noticed that in all our tests BCF continued
until the crater was just slightly deeper than the length Only small portions of the parameter space for jet-
of the jet, and beyond that the much slower VE became induced cratering have been sampled to-date. The VE
dominant. This led us to hypothesize that (within the pa- length a, time b-1, and volumetric erosion a3b scales in
rameter space of our experiments) the depth of BCF is the logarithmic period are inconsistent with the Froude
controlled almost entirely by jet length rather than soil number. Cohesion must be taken into account at low
conditions, gravity, or other parameters. gravity. BCF appears to be controlled primarily by the
To test this hypothesis, we performed two successive plume parameters (jet length). Much work remains to
firings into the sand bed filled with JSC-Mars-1A. In the understand the physics and scaling in the various flow
first test the soil was laid down in thin layers each vi- regimes.
brationally compacted. In the second test the soil was
poured in as loosely as possible. The resulting soil den-
sity in a core extracted from the top 25 cm of the bed REFERENCES
was measured to be four times higher in the vibrationally
compacted state (1.04 vs. 0.25 g/cm3), with three times 1. R. A. Bagnold, The Physics of Blown Sand and Desert
higher shear strength (2.20 vs. 0.70 kg/cm2 as measured Dunes, Dover, Mineola, New York, 2005.
by a Torvane shear tester), and seven times higher pene- 2. R. F. Scott and H.-Y. Ko, AIAA Journal 6, 258-264 (1968).
tration resistance (9.70 vs. 1.40 kg/cm2 as measured by 3. J. D. Alexander, W. M. Roberds and R. F. Scott, “Soil
handheld penetrometer). For both motor firings we took Erosion by Landing Rockets,” Contract NAS9-4825 Final
Report, Hayes International Corp., Birmingham, Alabama
high speed video and measured the depths of the craters (1966).
versus time. Also, the scorching of the viewing window 4. Metzger, P.T., C.D. Immer, C.M. Donahue, B.T. Vu, R.C.
provided witness marks for both the final crater depth Latta, III, M. Deyo-Svendsen., J. Aero. Eng. 22, 24-32
and the jet core length in each test. The results are shown (2009).
in Fig. 8. In both tests the BCF quickly achieved its max- 5. N. Rajaratnam and S. Beltaos , J. Hydraulics Div., 103,
imum depth in about 1.5 seconds or less followed by the 1191-1205 (1977).
6. N. Rajaratnam, J. Hydraulics Div., 108, 262-267 (1985).
slower VE. The final depth of the crater in loose soil was
7. S. A. Ansan, U. C. Kothyari, and K. G. Ranga Raju, J.
just slightly longer than in the compacted soil, and both Hydraulics Eng. 129, 1014-1019 (2003).
were just slightly longer than the jets. This confirmed 8. R. B. Haehnel, W. B. Dade, and B. Cushman-Roisin,
our hypothesis: in this region of the parameter space it “Crater evolution due to a jet impinging on a bed of loose
was primarily the jet length and not the state of the soil particles” in 11th Biennial ASCE Aerospace Division
that controlled the crater depth. However, the BCF did International Conference on Engineering, Construction
and Operations in Challenging Environments, Long
achieve this depth faster in the loose soil.
Beach, California, Mar. 3-5, 2008.

770