Geotechnical and Geological Engineering (2006) 24: 757–774
Springer 2006

DOI 10.1007/s10706-005-1707-8

Technical Note

Experience of vibrocompaction in calcareous

sand of UAE

AZM S. AL-HOMOUD1,w and W. WEHR2

1
American University of Sharjah, P.O. Box 26666, Sharjah, UAE
2
Keller Grundbau Company GmbH (Offenbach), Germany

(Received 7 June 2004; accepted 2 February 2005)

Abstract. The study is concerned with the influence of various factors on vibro compaction in
calcareous soils, This include effect of carbonate content, and influence of pore water pressures
in sand, settlements during vibrocompaction, vibroaccelerations during sand column instal-
lation compared to earthquake accelerations and finally the liquefaction design based on
average CPT parameters. Previous experiences with calcareous sands and literature review
confirm the necessity to apply a correlation factor for calcareous material to the raw cone
penetration resistance CPT values obtained in the field. Because of the very fragile grains of
the Dubai calcareous sands the penetration resistance will be influenced by crushing and
grinding. To accommodate for effect of calcareous sands on measured CPT values, a best fit
correlations formula called here ‘‘correlation factor’’ is developed. A parametric dynamic
triaxial liquefaction testing programme has been executed on representative samples with
loose and dense zones. Representative testing parameters for field conditions and earthquake
design accelerations have been applied indicating that ground improvement is necessary. It
was observed during the dynamic triaxial tests that the whole soil mass consisting of dense
columns and looser zones in between behaves jointly, and therefore compaction control using
an average-CPT approach is possible.
Key words. calcareous sand, cone penetration test, design criteria, dynamic triaxial testing,
earthquakes, liquefaction potential, relative density, shells correction factor, UAE, vibro-
compaction.

1. Introduction and Problem Description

As part of the construction works required for the realization of the land reclamation
projects in Dubai, ground improvement via deep vibratory compaction is taking
place for a reclaimed calcareous sandy soil.
It is demanded that a minimum relative density of 60% after compaction and a
penetration resistance qc required to achieve the requested relative density is as given
in Table 1.
This criteria of specified penetration resistance was based on correlation charts
available in the literature for silica sand (e.g. Robertson and Campanella, 1985).

w
Corresponding author: American University of Sharjah, P.O. Box 26666, Sharjah, UAE
e-mail: ahomoud@aus.ac.ae
758 AZM S. AL-HOMOUD AND W. WEHR

Table 1. Specified penetration resistance after compaction

Depth [m] qc [MPa]

<4 m 6
4–8 m 8
>8 m 10

The shear resistance of granular soils depends on the granulometric properties of

the granular skeleton (grain minerals with hardness, grain shape and grain size
distribution) and on the soil state, which is defined primarily by the density and the
effective pressure. Therefore, soils with different granulometric properties exhibit
quantitatively different relationships between the cone penetration resistance and the
initial state variables. This is clearly shown by the results of many experimental
investigations of the cone penetration resistance of sands in large calibration
chambers (Baldi et al., 1986; Been et al., 1987; Iwasaki et al., 1988; Schnaid, 1990;
Fioravante et al., 1991; Rodrigues, 1993).
The penetration resistance of calcareous sands was investigated by Nutt (1993)
and Almeida et al. (1991). They found that for the same initial state, qc for a
calcareous sand (qcc) can be as low as 20% of that for a quartz sand (qcq) with a
similar grain size distribution. The higher crushability of calcareous sand grains is
responsible for the relatively low penetration resistance. Since grain crushing
increases with increasing density, and increasing pressure therefore, the largest values
of qcc/qcq are obtained for very dense states and larger mean pressures.
Due to the difficulties in achieving the requested penetration resistance in some
zones of the compacted fill, there is a need to verify whether the silica sand based qc-
values from Table 1 relate to relative density values significantly higher than 60%.

2. Geotechnical Properties of Investigated Sand

2.1. TESTED SOIL
The calcareous sand of this study is from one of the construction sites in Dubai. The
granulometric properties of the Dubai sand are listed in Table 2.
For the preparation of the soil sample with the pluviation method only the frac-
tion less than 4 mm was used. The coarser fraction (>4 mm), which is almost 8% by
weight, was separated before the testing programme was started.

2.2. EFFECT OF CARBONATE CONTENT

The behaviour of carbonate sands may be analysed by various field and laboratory
tests. This way, the influence of the carbonate content on the densification (com-
paction and grain breakage) and on compaction control can be estimated.
EXPERIENCE OF VIBROCOMPACTION 759

Table 2. Granulometric properties of the Dubai sand

Specific gravity 2.805

d50/d10/d60 0.4/0.13/0.53 mm
Angle of repose [] at lowest state 35
Maximum void ratio 1.223
Minimum void ratio 0.762
Grain shape Very angular (shell fragments)
Grain hardness Very crushable
Calc content 90%
Fines <0.5%

2.3. CONE PENETRATION TESTS

Correlations between the CPT cone resistance and the relative density are well
established for silica sands (e.g. Robertson and Campanella, 1985). The available
references concerning calcareous sands are reviewed in this section.
Vesic (1965) added 10% of shells to quartz sand which resulted in a decrease of the
CPT cone resistance by a factor of 2.3, but without a noticeable decrease in angle of
friction. Bellotti and Jamiolkowski (1991) compared CPT cone resistances and
determined a relationship of qc ðsilicaÞ=qc ðshellsÞ
1 þ 0:015 (Dr)20) yielding ratios
between 1.3 and 2.2 increasing with relative density, where Dr is the relative density
of sand. Moreover, Jamiolkowski and Pasqualini (1992) stated: ‘‘When correlations
obtained for silica sands are used to predict the behaviour of calcareous sands Dr was
underestimated ...’’
Almeida et al. (1992) compared normalized CPT cone resistances from calcareous
Quiou sand and silica Ticino sand which yields ratios between 1.8 and 2.2 increasing
with relative density. And finally Cudmani (2001) looked at normalized cone resis-
tances of seven sands yielding ratios between 1.4 and 3.5 depending on initial soil
pressure and relative density.

2.4. PRESSUREMETER TESTS

Foray et al. (1999) compared pressuremeter limit pressures of silica sands and car-
bonate sands at Grenoble University in France. It is stated ‘‘The limit pressures for
carbonate sands are around one third to one half of the values corresponding to
silica sands.’’ The ratios were between 2 and 3, increasing with increasing initial
vertical stress.

2.5. SUMMARY OF LABORATORY TESTS WITH CPT IN CALIBRATION CHAMBER

A study on Dubai calcareous sand has been undertaken using a calibration
chamber at Karlsruhe University by Gudehus and Cudmani (unpublished) in order
760 AZM S. AL-HOMOUD AND W. WEHR

to investigate the relationship between Cone Penetration Test (CPT) resistance qc ,

relative density Dr and the mean pressure for Dubai sand. CPT tests were con-
ducted at three different densities and two different initial pressures. The diameter
and the height of the calibration chamber are 0.95 and 1.50 m, respectively. The
sand sample is placed in the chamber by pluviation (sand raining). The CPT is
carried out by pushing a rod with a 36 mm diameter instrumented cone through a
hole in the cap of the calibration chamber into the soil sample. A hydraulic jack
pushed in the cone with a velocity of about 2.5 mm/s up to a maximum depth of
1.2 m. A detailed description of the complete testing apparatus is presented by
Cudmani (2001).
The penetration resistances for Dubai sand was compared with that for a medium
quartz sand (Karlsruhe sand) for a medium dense state with the same initial vertical
and horizontal pressure.
The main findings of Gudehus and Cudmani (unpublished) study are:
• The penetration resistance for the Dubai sand is about 37% lower than that for
medium quartz sand (Karlsruhe sand) for a medium dense state.
• The measured qc-values for the Dubai calcareous/carbonate sand have to be
multiplied by the following shell correlation/correction factors (Table 3) to
achieve the quartz sand equivalent specified CPT cone resistances.
• Grains of the Dubai calcareous sand are very fragile, therefore the penetration
resistance will be influenced by crushing and grinding. According to a mecha-
nism of ‘‘soil plug’’ formation and collapse, sudden not easily quantifiable
variations of the penetration resistance must be expected in the field.
• The above correlation factors should be seen as lower limit and conservative for
calcareous sand if the material in the field is much coarser than the soil fractions
used in the experiments since larger shells crush easier than very small ones.
This is true in this study because in the conducted tests all shells larger than
4 mm grain size (approximately 8%) were removed to produce reliable and
reproducable test results.
• A pronounced variation of the penetration resistance has to be expected at high
sand densities due to a soil plug formation including particle crushing below the
CPT-cone leading to significant variations of cone resistances at short depth
intervals which requires a higher value than 5% tolerance.

Table 3. Recommended shell correlation factors for calcareous Dubai sand

Depth[m] qc measured from Shell correlation Specified cone

calibration chamber factor resistance

<4 3.5 1.7 6

4–8 5.1 1.6 8
>8 6.7 1.5 10
EXPERIENCE OF VIBROCOMPACTION 761

3. Development of Shell Correlation Factor for Cone Penetration Tests

In this study, all experimental data from the literature and the tests by Gudehus and
Cudmani (unpublished) on Dubai calcareous sand have been compiled and plotted
in Figure 1. On the y-axis, the cone resistance qc has been normalized with respect to
pressures applied i.e. norm qc ¼ qc  pref =pm with the mean pressure being
pm ¼ ðrv þ 2rH Þ=3 and an approximate average reference pressures of all tests of
pref = 50 kPa.
Additionally exponential best fit curves have been calculated for all quartz
sands (all qua.) and for all calcareous sands (all cal.) yielding the equations given
in Figure 2.
It has been found that the influence of density is much stronger than the influence
of pressure taking into account a rang between 0 and 100 kPa in all tests. Therefore
the influence of pressure has been neglected in arriving to the to shell correlation
factor shown in Figure 2.
Dividing the equation for quartz sands and the equation for calcareous
sands of Figure 1 yields a linear equation for the shell correlation factor
(Figure 2).
fSHELL ¼ 0:0046  Dr ½% þ 1:3629
It is recommended to apply this factor to all CPT evaluations for calcareous sands.
In all tests mean pressures (0–100 kPa) and grain sizes (0.06–4 mm) were used. These
values should be regarded as application limits. If the mean pressure or the grain size
are larger the shell correlation factor tends to be higher.

Figure 1. Normalized cone resistance depending on relative density.

762 AZM S. AL-HOMOUD AND W. WEHR

Figure 2. Shell correlation factor depending on relative density.

4. Field Experience with Calcareous Sands

Jurong Island in Singapore (Wehr and Raju, 2002). At the pipeline area the
carbonate content was negligible and compaction was carried out to 70% relative
density using the Schmertmann curves. The average compaction depth was 20 m
below the ground. A triangular grid with 3.5 m spacing was chosen to achieve the
70% relative density requirement.
At the sand bund area, the carbonate content was 11% (without taking into
account large gravel size shells) and 1:3 inclined slopes to be densified to the Jurong
Town Council (JTC) specification which corresponds to a cone resistance of 4, 6, 8
and 10 MPa in depths of 0–2 m, 2–8 m, 8–24 m and 24–27 m below ground level.
The JTC specifications thus require a lot less compaction than the 70% Schmermann
criterion which would result in a much larger grid spacing assuming pure silica sand.
Because of the carbonate content, only a slightly larger triangular grid with 4.0 m
spacing was chosen in the sea side slope area to achieve the JTC specification.

4.1. EFFECT OF WATER PRESSURE DISSIPATION

Supported by various literature on vibrocompaction, one should let water pressures
dissipate for at least 2–4 weeks before taking CPT tests for checking compliance with
specifications.

5. Design Criteria of Vibrocompaction

If one follows Robertson and Campanella (1985) charts considering the soil to be
silica sand, a 6–8–10 MPa CPT vs. depth is required to prevent liquefaction
occurrence for a design Peak Ground Acceleration of 0.20 g. However, based on this
EXPERIENCE OF VIBROCOMPACTION 763

study, such prescribed CPT profile correspond to the relative density profile shown
in Figure 3, which far above the required relative density value of 60% needed to
avoid liquefaction.
Therefore 6, 8 and 10 MPa criterion yields much higher relative densities (70–
100%) than required for the liquefaction design, see Figure 3 and thus appears to be
too conservative.

5.1. SETTLEMENT DUE TO VIBROCOMPACTION FOR THE STUDY PROJECT IN DUBAI

The easiest measure of the success of the vibrocompaction procedure is possible by
measuring the site levels before and after compaction. At an investigated site in
Dubai, encountering calcareous sand, the settlements caused by one contractor’s
vibrators are summarized in Table 4.

Figure 3. Relative densities corresponding to the 6/8/10 MPa criterion.

764 AZM S. AL-HOMOUD AND W. WEHR

Table 4. Settlement due to vibrocompaction at study site in Dubai

Chainage Triangular grid Average settlement

0–100 m 3.75 m 1.00 m

100–200 m 3.50 m 1.10 m
200–300 m 3.00 m 1.10 m

These values of settlement upon vibrocompaction for a 12 m high fill with an initial
relative density (in its unimproved state) of less than 30% clearly results in significant
increase in the relative density by far exceeding the 60% threshold value required by
specifications of the compaction work. Considering the initial relative density of the
soil before treatment, it is our view that such settlement indicates clearly that ground
improvement contractor practically is achieving an overall relative densities exceeding
80–90%.
Therefore, it is recommended to adopt the amount of settlement observed in the field
after vibrocompaction as a supportive evidence of achieving required compaction to
the specified level.
Photographs of the site performance and field observations during vibrocom-
paction in this study are given in Figures 4 and 5.

5.2. VIBROACCELERATIONS DURING COLUMN INSTALLATION

Baez and Martin (1992) were measuring ground accelerations and pore water
pressures during stone column installation with a Keller S-vibrator. The same type of

Figure 4. Vibrator above ground with tip water to ease penetration at start of penetration.
EXPERIENCE OF VIBROCOMPACTION 765

Figure 5. Cracks within 4 m distance from the vibrator reflecting large influence radius.

S-vibrator is used in this study in Dubai. Baez and Martin (1992) reported the
following accelerations at different distances from the centre of the column:
1.7 g at 0.91 m distance,
0.9 g at 1.52 m distance,
0.6 g at 2.13 m distance.
Considering a design acceleration due to earthquakes in Dubai to be 0.2 g, the
acceleration during column installation lies well above the design value. Baez and
Martin (1992) stated: ‘‘The above observations clearly demonstrate that one of the
mechanisms of densification was a controlled vibrator induced liquefaction (ratio of
excess pore water pressure to initial effective stress equal to one) in the surrounding
soils. These observations also indicated that in order for a state of liquefaction to be
achieved a second time, the levels of acceleration must be significantly higher.’’

6. Evaluation of Liquefaction Potential of the Compacted Soil Via

Experimental Study Using a 50 kN Dynamic Triaxial Testing Facility
Earthquakes induce into the soil a cyclic shear stress. This shear stress in turn causes
strain between the grains. After a certain number of cycles and build up of such
766 AZM S. AL-HOMOUD AND W. WEHR

strains, pore water pressure starts to increase and with increasing number of shear
stress cycles, pore water pressure continues to increase, resulting in reduction of
effective stress, soil stiffness and strength. This will result in development of large
strains and degradation in strength.
At a pore pressure equals total stress, effective stress reaches zero and the friction
between particles becomes too small if not zero. Once that happens, the particles
cannot support any loading stresses any more. The soil behaves then for all practical
purposes like a liquid. Therefore this phenomenon received the name ‘‘Soil lique-
faction’’.
It is common practice to evaluate liquefaction potential from a correlation
between liquefaction resistance as determined from field performance of soil deposits
during past earthquake events and in-situ penetration test results. Historically, Seed
and co-workers (Seed, 1979; Seed et al., 1983; Seed et al., 1984) started the corre-
lation with SPT N values. This approach, is described by Youd (2001) and the
NCEER (1998) report. Given also in the NCEER (1998) report is the latest corre-
lation of liquefaction resistance with static cone tip resistance that was started in the
early 80s of the last century.
In Seed’s approach, an extensive database of field performance was established
for soil deposits which did or did not liquefy during past earthquake events. At
each site, the layer with the lowest penetration resistance was considered to be the
critical layer and the average standard penetration test blow count N value was
selected as representative for that layer. The N value for that layer was corrected
for in-situ testing procedure to N60 according to guide lines provided by Seed
et al. (1984). In the NCEER (1998), updated guide lines were given based on
Skempton (1986) and modified by Robertson and Wride (1998). The N60 value
was further corrected for the effect of the effective overburden pressure, r¢vo, by
converting it to a blow count at effective overburden pressure of 100 kPa, (N1)60.
Terzaghi et al. (1996) and NCEER (1998) suggested the use of the Liao and
Whitman (1985) equation:
 1=2
Pa
ðN1 Þ60 ¼ N60 where CN ¼
r0vo
where r¢vo, is in kPa, Pa is about 100 kPa (1 atm). The factor CN should not exceed a
value of 2. In the case of using cone penetration resistance, the qc of the critical layer
is corrected for the influence of effective overburden pressure to qc1N using the
following equation:
 n
Pa
qc1N ¼ ðqc =Pa ÞC where CQ ¼
r0vo
where CQ should not exceed a value of 1.7 and n is an exponent that varies from
0.5 to 1.0 depending on the grain properties of the soil (Olsen, 1997). For clean
sand, n is equal to 0.5. At the other extreme, n is equal to 1.0 in the case of
clayey soils.
EXPERIENCE OF VIBROCOMPACTION 767

The earthquake-induced equivalent uniform seismic shear–stress ratio or the cyclic

stress ratio, CSR=s(seismic)r¢vo, is calculated for the critical layer using the Seed
and Idriss (1971 ) equation:

sðseismicÞ amax rvo

CSR ¼ 0
¼ 0:65 rd
rvo g r0vo

where amax is the maximum ground surface acceleration in gals, g is the acceleration
due to gravity (980 gals), rvo is the total overburden pressure at the depth of the
considered layer, and rd is a reduction factor varying from a value of 1 at the ground
surface to a value of 0.9 at a depth of 10 m. Youd (2001) suggested that the
following equation may be used to estimate rd:

ð1:000  0:4113z0:5 þ 0:04052z þ 0:001753z1:5 Þ

rd ¼
ð1:000  0:4177z0:5 þ 0:05729z  0:006205z1:5 þ 0:001210z2 Þ

where z is the depth of the considered layer in meters.

The field performance data, for deposits with fines content less than 5% and for
earthquake magnitude M = 7.5, have been summarized. A curve was drawn as the
best boundary between the cases that did liquefy and those that did not liquefy. The
drawn curve represents the relationship between the cyclic resistance ratio, CRR,
which reflects mobilized undrained yield strength normalized by the effective over-
burden pressure, su/r¢vo, and either (N1)60 or qclN. Such a relationship is expected
because both CRR and (N1)60 or qc1N should be affected by factors such as density,
aging, preconsolidation, and preshearing of sand (Terzaghi et al., 1996). CRR
equations are given in details by Youd (2001) using either (N1)60 or qclN.
A practical parameter to evaluate the liquefaction potential and the consequences
of liquefaction is the factor of safety against liquefaction that is calculated using the
following equation:

CRR
FSL ¼
CSR

However, liquefaction behaviour can be studied in more details through laboratory

experimentation. In this study, a dynamic triaxial testing programme is also con-
ducted to investigate liquefaction behaviour of zoned soil columns resulting from
vibrocompaction at the site characterized by calcareous soil. Tests were conducted to
find out the optimum spacing between probes for the design Peak Ground Accel-
eration (i.e. the maximum spacing between probes that will result in zoned soil
columns that will not liquefy under design Peak Ground Acceleration).
Samples are prepared to correspond to field condition upon treatment using
triangular grid of vibrocompaction (i.e. cylindrical dense zones surrounding loose
zones). Samples with percentages of loose to dense area of 35 and 50% are prepared
and tested. These correspond to triangular vibrocompaction grid spacings 3.75 and
4.0 m, respectively. Samples for these tests were obtained from representative areas
768 AZM S. AL-HOMOUD AND W. WEHR

at the investigated site, where vibrocompaction equipment is performing the ground

improvement works.
The two zoned lab samples are as follows,
(a) A cylindrical soil sample with 15 cm outside diameter and 30 cm height. Thirty-
five percent by area contains loose sand with relative density equal to or less than
15% surrounded by 65% by area of dense sand with relative density 60%. The
inside diameter in this case is 8.874 cm.
(b) A cylindrical soil sample with 15 cm outside diameter and 30 cm height. Fifty-
percent by area contains loose sand with relative density equal to or less than
15% surrounded by 50% by area of dense sand with relative density 60%. The
inside diameter in this case is 10.607 cm.
Moreover, a 15 cm diameter and 30 cm height cylindrical sample with uniform
density (100% area) at loose state (i.e. with relative density equal or less than
15%) is prepared and tested for comparison purposes with the above. This
sample represent the worst scenario of relative density of the uncompacted soil
before treatment.
Samples are compacted to the required density in each zone of the two zones
within the cylinder by tamping using a rod. A weight of soil calculated to correspond
to the required density in each zone is compacted to fill in the required volume within
the cylinder. The soil passing Sieve No. 4 is used in the study. Based on lab tests, the
soil has a minimum void ratio of 0.765, and a maximum void ratio of 1.225 and
specific gravity of 2.81.
At the triaxial chamber, horizontal and vertical stresses and pore pressure are
applied at the sample to correspond to the following condition:
– Depth of sample from ground surface = 8m, water table is at 2 m below ground
surface, at rest lateral earth pressure coefficient = 0.5, unit weight of soil above
ground water table = 14 kN/m3, unit weight of soil below ground water table
level = 18.8 kN/m3.
– Initial total overburden pressure equal to 140.8 kN/m2, initial pore pressure =
58.8 kN/m2.
Cyclic shear is applied via dynamic deviatoric stress corresponding to design
earthquake with Peak Ground Acceleration of 0.20 g and frequency of 3 Hz. Each
sample was subjected to a number of load cycles causing failure or 500 cycles
whichever occurs first.
Cyclic shear stress is computed using the well known Seed equation.
Following are the computed amplitudes of cyclic shear stresses for three different
values of peak ground acceleration:
– For PGA = 0.2 g, shear stress amplitude is computed to be 17.39 kN/m2.
– For PGA = 0.3 g, shear stress amplitude is computed to be 26.10 kN/m2.
– For PGA = 0.4 g, shear stress amplitude is computed to be 34.78 kN/m2.
EXPERIENCE OF VIBROCOMPACTION 769

With the drainage valves open, initial total vertical and horizontal stresses cor-
responding to site condition were applied, then a back pressure equivalent to 6 m
water height was applied and sample was saturated with the drainage valves still
open. At the end of the saturation stage the valves were closed, and dynamic testing
was conducted at undrained condition. Cylic shear is applied via a cyclic vertical
loading, with the following parameters recorded with time: total confining pressure,
pore pressure, vertical load, and vertical settlement.
In the course of dynamic load application, in case the sample didn’t experience
significant deformations for the 500 cycles of applied shear stress with amplitude that
correspond to a Peak Ground Acceleration of 0.20 g at 3 Hz, the sample is then
subjected to cycles of shear stress with amplitude that correspond to 0.30 g. Once
more, if the samples didnot experience large deformations at such stress up to 500
cycles, then the sample is further subjected to cyclic shear stresses with amplitude
corresponding to a Peak Ground Acceleration of 0.4 g and frequency of 3 Hz, and
so on. This is done in order to understand the dynamic behaviour of the prepared
soil sample at different levels of ground shaking.
Results were analysed and plots prepared for the following parameters as shown in
Figures 6 through 9: Settlement vs. time, pore pressure vs. time, and effective stress
vs. time, respectively.
Results from the time tests (uncompated, treated via vibrocompaction resulting in
a state 35% area of loose sand, and treated via vibrocompaction resulting in a state
50% area of loose sand) showed clearly that soil sample corresponding to a trian-
gular vibrocompaction grid of 3.75 m didnot experience any deformation even at
dynamic shear stresses corresponding to PGA of 0.4 g. However, the soil sample that

Figure 6. Summary of pore water pressure development U (kPa) with time (s) from dynamic triaxial tests.
770 AZM S. AL-HOMOUD AND W. WEHR

Figure 7. Summary of effective confining pressure (kPa) vs. time (s) from dynamic triaxial tests.

Figure 8. Summary of vertical displacements (mm) vs. time (s) from dynamic triaxial tests.

correspond to a vibrocompaction grid of 4.0 m started to experienced large defor-

mations after 100 cycles for dynamic shear stresses corresponding to PGA of 0.2 g.
The untreated soil sample experienced large deformations upon application of the
first few cycles for dynamic shear stresses corresponding to PGA of 0.20 g indicating
that the untreated soil experienced liquefaction failure upon cycle loading corre-
sponding to the design earthquake (Figure 9).
Test results indicate that ground improvement is necessary and the 3.75 triangular
grid represent an optimum spacing to avoid liquefaction under design earthquake.
EXPERIENCE OF VIBROCOMPACTION 771

Figure 9. Liquefied specimen in the triaxial cell.

6.1. AVERAGE CPT VALUES

It was observed during the dynamic triaxial tests that the whole soil mass consisting
of densified columns and looser zones in between behave jointly, and therefore
compaction control using a real average CPT approach with two testing points is
possible. One point in the compacted column ‘‘best point’’ and one point at the
centroid of the grid ‘‘worst point’’ should be averaged. The ‘‘best points’’ are just
executed on site for the first areas.

7. Summary and Conclusions

Large calibration chamber tests made in Germany on representative three tons soil
sample from the site of ground improvement works executed at a land reclamation
site in Dubai reveal that to accomodate effect of calcareous nature of the sand, the
measured ‘‘raw’’ CPT cone resistance values have to be multiplied by a correlation
factor in the range between 1.5 and 1.7 depending on the depth to achieve the
specified CPT cone resistances of 6, 8 and 10 MPa. Alternatively, the specifications
in calcareous sands should be relaxed to accommodate these.
Previous experiences with calcareous sands and literature review confirm the
necessity to apply a correlation factor for calcareous material to the raw cone
penetration resistance values obtained in the field. Based on the correlation devel-
oped in this study, the correlation factor is found to be a sound technical fact. It
increases with increasing density of the sand and with increasing depth.
772 AZM S. AL-HOMOUD AND W. WEHR

Summarizing all above test results and correlation values proposed in the litera-
ture for similar cases to accommodate effect of calcareous sands on measured CPT
values, a best fit correlations formula called here ‘‘correlation factor’’ (fSHELL) is
developed. The study showed that fSHELL ¼ 0:0046  Dr ½% þ 1:3629:
Adopting this correlation formula results in a correlation factor of 1.64 for cal-
careous sands with 60% relative density. This also corresponds very well to the
calibration chamber test results on Dubai calcareous sand.
It is recommended to adopt the amount of settlement observed in the field after
vibrocompaction as a supportive evidence of achieving required compaction to the
specified level.
A parametric dynamic triaxial liquefaction testing programme has been executed
on representative samples with loose and dense zones from the work areas. Repre-
sentative testing parameters for field conditions and earthquake design accelerations
have been applied indicating that ground improvement is necessary and the 3.75
triangular grid represents an optimum spacing to avoid liquefaction occurrence
under the adopted design earthquake peak ground acceleration.
It was observed during the dynamic triaxial tests that the whole soil mass
consisting of dense columns and looser zones in between behaves jointly, and
therefore compaction control using an average-CPT approach is possible. For
quality control and checking compliance with specifications upon compaction, it
is recommended that two testing points for each 1000 m2 compacted area shall be
executed, one in the compacted column and one at the centroid of the grid. Both
values should be averaged and the correlation relation for calcareous material
developed in this study should be applied to evaluate the ‘‘raw’’ CPT values
obtained from the field.
Based on the correlations developed here for corrected CPT field ‘‘raw’’ test results
at the already performed ground improvement works and the results of liquefaction
study, it is highly recommended to adopt a 3.75 m triangular grid.

References
Almeida, M. S. S., Jamiolkowski, M. and Peterson, R. W. (1992) Preliminary result of CPT
tests in calcareous Quiou sand, In: Proceedings of the International Symposium on Cali-
bration Chamber Testing, Potsdam, New York, 1991, Elsevier, pp. 41–53 (cited by Lunne,
Robertson, Powell, Cone Penetration Testing in Geotechnical Practice, Blackie academical
and professional, p. 106).
Baez, J. I. and Martin, G. R. (1992) Liquefaction observations during installation of stone
columns using the vibro-replacement technique, Ground Modification, Geotechnical news,
September, pp. 41–44.
Baldi, G., Bellotti, R., Ghionna, N., Jamiolkowski, M. and Pasqualini, E. (1986) Interpretation
of CPT’s and CPTU’s, 2nd Part: drained penetration of sands, In: 4th International Geo-
technical Seminar, Field Instrumentation and In-Situ Measurements, Singapore, pp. 143–155.
Been, K., Jefferies, M., Crooks, J. and Rothenburg, L. (1987) The cone penetration test in
sands: part II, general inference of state, Ge´otechnique, 37(3), 285–299.
EXPERIENCE OF VIBROCOMPACTION 773

Bellotti, R. and Jamiolkowski, M. (1991) Evaluation of CPT and DMT in Crushable and Silty
Sands, Third interim report ENEL, C.R.I.S., Milano.
Cudmani, R. O. (2001) Static Alternating and Dynamic Penetration of Cohesionless Soils,
Dissertation, Veröffentlichungen des Institutes für Bodenmechanik und Felsmechanik der
Universität Fridericiana in Karlsruhe, Heft 152.
Fioravante, V., Jamiolkowski, M., Tanizawa, F. and Tatsuoka, F. (1991) Results of CPT’s in
Toyoura quartz sand, In: Huang (ed.). Calibration Chamber Testing, Elsevier, pp. 13–39.
Foray, P. Y., Nauroy, J.-F. and Colliat, J. L. (1999) Mechanisms governing the behaviour of
carbonate sands and influence on the design of deep foundations, In: Al-Shafei (ed.).
Engineering for Calcareous Sediments, Balkema, Rotterdam, pp. 55–67.
Herle, I. (1997) Hypoplastizität und Granulometrie einfacher Korngerüste, Dissertation, Kar-
lsruhe University.
Iwasaki, K., Tanizawa, F., Zhou, S. and Tatsuoka, F. (1988) Cone resistance and liquefaction
strength of sand, In: Proceedings 1st International Symposium on Penetration Test ISOPT-1,
Vol. 2, pp. 785–791.
Jamiolkowski, M. and Pasqualini, E. (1992) Compaction of granular soils – remarks on
quality control, Proc. Conf. On Grouting Soil Improvement an Geosynthetics, New Orleans.
Liao, S. C. and Whitman, R. V. (1985) Overburden correction factors for SPT in sand, Journal
of Geotechnical Engineering, ASCE, 112(3), 373–377.
Nutt, N. R. F. (1993) Development of the Cone Pressuremeter, PhD Thesis, St. Catherine
College, University of Oxford, Oxford.
Olsen, R. S. (1997) Cyclic liquefaction based on the cone penetration test, Proceedings of
NCEER Workshop on Evaluation of Liquefaction Resistance of soils, National Center for
Earthquake Engineering Research, pp. 225–276.
Robertson, P. K. and Campanella, R. G. (1985) Liquefaction potential of sands using the
CPT, Journal of Geotechnical Engineering, ASCE, 111(3), 384–403.
Robertson, P. K. and Wride, C. E. (1998) Evaluating cyclic liquefaction potential using the
cone penetration test, Canadian Geotechnical Journal, 35(3), 442–459.
Rodrigues, R. S. (1993) Analysis of Penetration Resistance in Sand, PhD Thesis, Graduate
Division of the University of California at Berkeley, Berkeley.
Schnaid, F. (1990) A Study of the Cone-Pressuremeter Test in Sand, PhD Thesis, Magdalen
College, University of Oxford, Oxford.
Seed, H. B. (1979) Soil liquefaction and cyclic mobility evaluation for level ground during
earthquakes, Journal of Geotechnical Engineering, ASCE, 105(2), 201–255.
Seed, H. B., Idriss, I. M. and Arango, I. (1983) Evaluation of liquefaction potential using field
performance data, Journal of Geotechnical Engineering, ASCE, 109(3), 458–482.
Seed, H. B., Tokimatsu, K. and Harder, L. H. (1984) The Influence of SPT Procedures in Soil
Liquefaction Resistance Evaluations, Earthquake Engineering Research Center, University
of California, Berkeley, Report No. EERC 84–15.
Skempton, A. W. (1986) Standard penetration test procedures and the effects in sands of
overburden pressure, relative density, particle size, aging and overconsolidation, Geotech-
nique, 36(3), 425–447.
Terzaghi, K., Peck, R. B. and Mesri, G. (1996) Soil Mechanics in Engineering Practice, 3rd ed.,
J. Wiley and Sons, New York, pp. 549.
Vesic, A. S. (1965) Ultimate loads and settlements of deep foundations in sand, Proceedings of
the Symposium on Bearing Capacity and Settlement of Foundations in Sand, Duke Uni-
versity, Durham, pp. 53–68 (cited by L. Decourt, General report session 2: SPT, CPT,
Pressuremeter Testing and Recent Developments in Insitu Testing, ICSMFE, Rio de Janeiro,
1989, pp. 2405–2412).
774 AZM S. AL-HOMOUD AND W. WEHR

Wehr, W. and Raju, V. R. (2002) On- and offshore vibro compaction for a crude oil pipeline
on Jurong Island, Singapore, 4th International Conference on Ground Improvement Tech-
niques, Vol. 2, Kuala Lumpur, Malaysia, pp. 731–736.
Youd, T. L. et al. (2001) Liquefaction resistance of soils: summary report from the 1996
NCEER and 1998 NCEER/NSF workshops on evaluation of liquefaction resistance of
soils, Journal of Geotechnical and Geoenvironmental Engineering, ASCE, 127(10), 817–833.