DENSIFICATION OF GRAVELLY SAND F I L L USING

INTENSIVE SURFACE COMPACTION

By Nathan L. Whetten1 and James W. Weaver,2 Members, ASCE
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

INTRODUCTION

A case study is presented herein whereby a heavy vibratory roller was

used to densify a 6-ft- (1.8-m-) thick layer of gravelly sand fill located below
ground-water level. Geotechnical engineering evaluations had concluded that
the fill was susceptible to liquefaction during the design seismic event. Ex-
cavation and subsequent placement of the same fill in thin compacted lifts
would have required a difficult and expensive dewatering effort. The layer
was densified by repeated coverages of a heavy vibratory roller. Test borings
with continuous standard penetration tests were performed to verify that the
intended densification had been achieved.

LITERATURE REVIEW

Few cases have been reported relative to the effects of intensive surface
compaction where fill thickness has exceeded approximately 1-2 ft ( 0 . 3 -
0.6 m). A summary of relevant publications and a brief description of their
findings follows.

• DAppolonia et al. (1969) evaluated the densification of a uniformly graded

fine sand with less than 4% passing the No. 200 sieve located above ground
water. Densification consisted of 2, 5, 15, and 45 passes of a 12,500-lb
(5,670-kg) roller operating at a frequency of 27.5 cycles/s (Hz). Two cov-
erages of the roller resulted in densification to a maximum depth of ap-
proximately 4.5 ft (1.4 m); 45 coverages resulted in densification to a
maximum depth of approximately 6 ft (1.8 m).
• Moorhouse and Baker (1969) evaluated the densification of a 10-ft- (3.0-
m-) thick layer of clean sand backfill placed such that the surface of the
fill was approximately 0.5 to 2.5 ft (0.2 to 0.8 m) above ground water.
Densification consisted of 8 and 24 coverages of a 25,000-lb (11,340-kg)
vibratory roller operating at a frequency of 27.5 cycles/s (Hz) and a for-
ward speed of 2.2 ft/s (0.7 m/s). Densification was achieved to a max-
imum depth of approximately 8 ft (2.4 m) with eight coverages of the
roller; 24 coverages provided little additional densification.
• Schmertmann (1970) reports densifying sand above ground water to a
maximum depth of 4.8 ft (1.5 m), and below ground water to a maximum
'Sr. Geotech. Engr., Haley & Aldrich, Inc., 600 Roundwood Drive, Scarborough,
ME 04074.
2
Sr. Assoc, and Vice Pres., Haley & Aldrich, Inc., 600 Roundwood Drive, Scar-
borough, ME.
Note. Discussion open until December 1, 1991. To extend the closing date one
month, a written request must be filed with the ASCE Manager of Journals. The
manuscript for this paper was submitted for review and possible publication on Sep-
tember 15, 1990. This paper is part of the Journal of Geotechnical Engineering,
Vol. 117, No. 7, July, 1991. ©ASCE, ISSN 0733-9410/91/0007-1089/$1.00 +
$.15 per page. Paper No. 26023.

1089

J. Geotech. Engrg. 1991.117:1089-1094.

 depth of 7.2 ft (2.2 m), using 10,800-lb (4,900-kg) and 27,000-lb (12,250-
kg) rollers, respectively.
• Soydemir and LeCount (1984) evaluated the densification of a 5-6-ft- (1.5-
1.8-m-) thick layer of saturated medium to fine sand using vibratory rollers
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

providing combined static and dynamic forces of 38,000 lb (169 kN) and
63,000 lb (280 kN), operating at 25 cycles/s (Hz). Two test areas were
prepared and the densification resulting from 8, 14, 22, and 34 coverages,
both with and without "waiting periods" between coverage stages, were
compared.

These case studies reveal that the maximum depth of improvement increases
with roller energy. However, fines content, location of ground water, the
presence of a hard underlying layer that reflects vibrations, and other factors
can all have a significant impact on the maximum depth and effectiveness
of densification.

PROJECT DESCRIPTION

Dundee Dam was constructed in 1913 and is located on the Presumpscot

River between Gorham and Windham, Maine. Geotechnical engineering
evaluations revealed that the granular embankment fill, which forms the ma-
jority of the east embankment, was very loose and potentially subject to
liquefaction during the design seismic event. Embankment fill consisted of
gravelly coarse to fine sand. The range of embankment fill gradation is shown
in Fig. 1.
The majority of the liquefaction-susceptible embankment fill was densified
by draining the impoundment, excavating the loose embankment fill, and
subsequently placing the same fill in thin compacted lifts. However, an ap-
proximately 4-l/2-to-5 ft (1.4-1.5-m) thickness of liquefaction-susceptible
embankment fill extended below the drained impoundment level. Excavation
and replacement of the submerged embankment fill would have required a

GRAIN SIZE DISTRIBUTION

GRAIN SIZE IN MILLIMETERS

COBBLES GRAVEL SAND SILT OR CLAY

FIG. 1. Range of Embankment Fill Gradation

1090

J. Geotech. Engrg. 1991.117:1089-1094.

 difficult and expensive dewatering effort. Therefore, intensive surface com-
paction was recommended as an inexpensive expedient means of densifying
the submerged embankment fill.
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

DENSIFICATION

Approximately 12-18 in. (30-46 cm) of embankment fill was first placed
over the submerged fill to provide a stable working surface above ground
water. This resulted in an approximately 9,500 sq ft (880 m2) area where
thicknesses of embankment fill ranged from 0 to a maximum of approxi-
mately 6 ft (1.8 m). Densification consisted of 10-12 coverages of a 23,700-
lb (10,750-kg) vibratory roller operating at 30 cycles/s (Hz) and a forward
speed of 1 to 2 ft/sec (0.3 to 0.6 m/s). These energy parameters were se-
lected in advance of densification based on available case literature. Four
test borings with continuous standard penetration tests (SPT) had been drilled
in the area prior to densification; six test borings with continuous SPT were
drilled after densification to verify that the intended densification had been
achieved.
Schmertmann (1970) states that the cone penetrometer test (CPT) provides
higher quality data than SPT for purposes of evaluating changes in density
after compaction. The writers concur that CPT would likely have provided
higher-quality data than the SPT in this case. However, since the original
explorations were performed using the SPT, it was also used for postden-
sification explorations for direct comparison.

EVALUATION

SPT data obtained from pre- and postdensification test borings versus ef-
fective overburden pressure are plotted in Fig. 2. Predensification effective
overburden pressures differ from postdensification pressures because pre-
densification borings were drilled prior to draining the impoundment and
making the general excavation. SPT data plotted in Fig. 2 are from the same
elevation interval and general area as the postdensification SPT data. Rel-
ative density curves developed by Gibbs and Holtz (1957) are also plotted
on Fig. 2 and provide a means to compare the pre- and post-densification
SPT results. Relative densities estimated based on the Gibbs and Holtz (1957)
curves indicate predensification relative densities ranging from approxi-
mately 10 to 55%, and postdensification relative densities ranging from ap-
proximately 55 to in excess of 100%.
Postdensification SPT results are plotted versus depth in Fig. 3. The great-
est densification occurred within the 2-4-ft (0.6-1.2-m) depth interval. Rel-
ative densities within this interval all exceeded 85%. Therefore the 2-4-ft
(0.6-1.2-m) depth interval appears to be ideal for densification using the
23,700-lb (10,750-kg) roller on clean, saturated sand. Lesser densification
occurred within the 0-2-ft (0-0.6 m) and 4-6-ft (1.2-1.8-m) depth inter-
vals. Relative densities at these depth intervals ranged from approximately
55 to 85%.
The lesser densification within the 0-2-ft (0-0.6-m) depth interval is likely
due to insufficient overburden pressure for the selected vibration level and
compactive energy. The relatively high vibration level and compactive en-
ergy appears to have produced a loosening effect. This loosening effect was
1091

J. Geotech. Engrg. 1991.117:1089-1094.

 Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

STANDARD PENETRATION TEST (SPT)

N-VALUES (BLOWS PER FOOT)

40
—I

• PRE-DENSIFICATION SPT
N-VALUES
EI POST-DENSIFICATION SPT
N-VALUES

J. Geotech. Engrg. 1991.117:1089-1094.

100% APPROXIMATE RELATIVE
2000- DENSITY CURVE BASED ON
GIBBS AND HOLTZ (1957)

NOTE: SPT N-VALUES ARE UNCORRECTED AND

WERE OBTAINED USING A DONUT HAMMER.

FIG. 2. Pre- and Postdensification SPT versus Effective Overburden Pressure

 Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

STANDARD PENETRATION TEST (SPT)

N-VALUES (BLOWS PER FOOT)

35 40

•MEAN SPT
VALUE

O
CD
CO

o
I

J. Geotech. Engrg. 1991.117:1089-1094.

9-

10J

FIG. 3. Postdensification SPT versus Depth below Ground Surface

 also observed by Moorhouse and Baker (1969), D'Appolonia et. al. (1969),
and others. D'Appolonia et. al. (1969) state that this phenomenon, termed
"overcompaction" or "overvibration," occurs when particle accelerations are
large and effective stresses are small, and has been observed in both labo-
Downloaded from ascelibrary.org by Universidad Politecnica De Valencia on 06/09/15. Copyright ASCE. For personal use only; all rights reserved.

ratory and field investigations. Therefore, if a lower acceleration had been

applied, densification within the 0-2-ft (0-0.6-m) depth interval would likely
have been greater.
SPT data within the 4-6-ft (1.2-1.8-m) depth interval as compared to the
2-4-ft (0.6-1.2-m) interval suggest that vibration energy attenuates fairly
rapidly with depth, and that the degree of densification correspondingly drops
off. D'Appolonia et al. (1969), Moorhouse and Baker (1969), and Schmert-
mann (1970) all observed similar behavior.

CONCLUSIONS

Based on the data obtained from this evaluation, the following conclusions
are offered:

1. Ten to twelve coverages of the 23,700-lb (10,750-kg) vibratory roller op-

erating at 30 c/s (Hz) over a submerged, 6-ft- (1.8-m-) thick layer of clean,
gravelly sand resulted in relative densities ranging from approximately 55 to in
excess of 100% throughout the layer.
2. The greatest densification occurred within the 2-4-ft (0.6-1.2-m) depth
interval. This depth interval appears to be ideal for densification using the 23,700
lb (10,750 kg) roller operating at 30 cycles/s (Hz) over submerged, clean grav-
elly sand.
3. Lesser densification, termed "overcompaction" or "overvibration," oc-
curred within the 0-2-ft (0-0.6-m) depth interval due to insufficient overburden
and large particle acceleration.
4. Relative densities within the 4-6-ft (1.2-1.8-m) depth interval, as com-
pared to the 2-4-ft (0.6-1.2-m) depth interval, suggest that vibration energy
from intensive surface compaction attenuates fairly rapidly with depth and the
degree of densification correspondingly drops.

APPENDIX. REFERENCES

D'Appolonia, D. J., Whitman, R. V., and D'Appolonia, E. D. (1969). "Sand com-

paction with vibratory rollers." J. Soil Mech. and Found. Engrg. Div., ASCE,
95(1), 263-284.
Gibbs, H. S., and Holtz, W. G. (1957). "Research on determining the density of
sands by spoon penetration testing." Proc. Fourth Int. Conf. on Soil Mech. and
Found. Engrg., International Society of Soil Mechanics and Foundation Engi-
neering, 1, 35-39.
Moorhouse, D. C , and Baker, G. L. (1969). "Sand densification by heavy vibratory
compactor." J. Soil Mech. and Found. Engrg. Div., ASCE, 95(4), 985-994.
Schmertmann, J. H. (1970). Discussion of "Sand densification by heavy vibratory
compactor," by Douglas G. Moorhouse and Gerald L. Baker. J. Soil Mech. and
Found. Engrg. Div., ASCE, 96(1), 363-365.
Soydemir, C , and LeCount, P. L. (1984). "Foundation design for potential lique-
faction: A case study." Proc. Eighth World Conf. on Earthquake Engrg., Earth-
quake Engineering Research Institute, (3), 191-198.

1094

J. Geotech. Engrg. 1991.117:1089-1094.