?

Vibro Replacement
to Prevent
Earthquake Induced
Liquefaction

Dipl.-Ing. Heinz J. Priebe

Keller Grundbau GmbH

Presented by
Keller Grundbau GmbH
Kaiserleistr. 44
D-63067 Offenbach
Tel. 069 / 80 51 - 0
Fax 069 / 80 51 - 244

Proceedings of the Geotechnique-Colloquium

at Darmstadt, Germany, on March 19th, 1998
(also published in Ground Engineering, September 1998)

E-mail [email protected]
www.KellerGrundbau.com

Technical paper 12- 57 E

Vibro Replacement
to Prevent Earthquake Induced Liquefaction
Dipl.-Ing. Heinz-Joachim Priebe
Keller Grundbau GmbH, Offenbach

1 Introduction
Fortunately, Germany has only a few, localised regions that are at risk from earthquakes, and even
here the risk is small. The situation is, however, quite different in the south and south-east of
Europe. For example, Assisi in central Italy was hit by an earthquake last year.
Elsewhere, there are regions on earth, often densely populated, where the danger of both severe
and very severe earthquakes is quite real. This is of primary concern in south-east Asia and other
areas of the Pacific Rim, where approximately 85 per cent of all the worlds seismic movements
are recorded.
When a building becomes unusable, it does not necessarily mean that the structure has suffered
severe damage or a total collapse as a result of seismic shaking. Frequently the ground under and
around the building fails and the facility is lost, even though the structure itself might have survived with minor damage. In these cases, failure results from liquefaction.

2 The Phenomenon of Liquefaction

Earthquake induced soil liquefaction occurs relatively frequently in areas of more or less level
ground. The term phenomenon is used not because the events seldom occur, but because the
expression highlights the often astonishing result, such as the overturning of whole apartment
blocks or the floating of buried facilities. Figure 1a shows a capsized apartment block in Niigata,
Japan, resulting from the famous 1964 earthquake (Herzog, 1980).

Figure 1a: Capsized apartment

block in Niigata,
Japan, 1964

?
Heinz-Joachim Priebe

Figure 1b: Subsided mosque near Sungai Penuh,

Indonesia,1994

Figure 1c: Tilted tank in Kobe,

Japan, 1995

Other spectacular and more recent examples are shown in figure1b, where a mosque in Indonesia subsided during an earthquake in 1994, and in figure 1c, where a tank in Kobe, Japan, tilted at
the 1995 earthqake (Wenk /Schwarz, 1995).
Liquefaction of this type only occurs in loose to medium dense, saturated soils with fairly uniform grain size distributions, covering the silty sandy range. The most critical soil is fine sandy
grained with some silt content. Figure 2 shows the bandwidth susceptible to liquefaction. It extends from medium to coarse sand to medium to coarse silt. In loose to medium dense conditions, the dynamic forces of a seismic event lead to an adjustment of the grain structure to a
denser state. Initially, if the soil does not drain sufficiently, due to a low permeability or long

Clay

Sand

Silt

Gravel

St.

100

Percentage Passing [by Weight]

90
80
70

Vibro Replacement
60

Liquefiable
Soils

50
40

Vibro Compaction
30
20
10
0
0.002

0.006

0.02

0.06

0.2

0.6

Grain Size [mm]

Figure 2: Application ranges of the deep vibratory compaction techniques

20

60

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

drainage paths, effective stresses in the grain structure start to pass to the pore water. The pressure in the pore water, therefore, rises accordingly, and the previously existing shear resistance of
the soil diminishes.
In a limit state, the subsoil behaves like a liquid, and loses its bearing capacity. Even if this state
lasts only for a short period, extreme deformations may occur, which have little to do with normal ground failure.

3 Evaluation of the Liquefaction Potential

In the mid-sixties, extensive efforts were beginning to be made to deal with the engineering
problems of soil liquefaction. The reason might have been the 1964 Niigata earthquake, where
the phenomenon occurred extensively. However, there were also demands to reduce the risk of
liquefaction at the growing number of industrial plants, not only nuclear power stations but also
barrages, refineries or chemical plants, where construction or foundation failure could have catastrophic consequences.
Options for evaluating the liquefaction potential at a given earthquake magnitude can be roughly
subdivided into three substantial groups: theoretical approaches, laboratory tests and statistical
analyses.

X
IX

10
Richter-Scale
MM-Scale (Modified Mercalli)

MSK-Scale (Medwedew, Sponheuer, Karnik)

JMA-Scale (Japan Meteorological Agency)

VIII

Earthquake Zones acc. to DIN 4149

Earthquake Zones acc. to UBC (Uniform Building Code)

VII

VI

IV

III

II

0
0,01

0,02

0,03

0,04 0,05

0,1

0,2

0,3

0,4

Earthquake Acceleration in g

Figure 3: Correlation between common earthquake scales and earthquake acceleration

0,5

Magnitude M (Richter-Scale)

Earthquake Zones

Earthquake Intensity

With theoretical approaches, the problem that arises is how to consider the prevailing ground
conditions properly. In the second group, laboratory tests can at least deal with the material
concerned, but problems lie in the difficulty of realistically modelling natural conditions. Although
the soil density is the main parameter in evaluating the liquefaction potential, the influence of
ageing and naturally occurring anomalies must not be underestimated. This is equally applicable
when transmitting model behaviour to actual conditions.

?
Heinz-Joachim Priebe

So far, the simplest and probably most reliable method seems to be the evaluation of the soil
liquefaction potential on the basis of statistical analyses. For this, the forces expected during a
seismic event are compared with the forces that the subsoil under consideration can actually
resist. Generally, the maximum surface acceleration on level ground is used as the characteristic
value for the forces developed by an earthquake. In figure 3, common earthquake scales, based on
subjective perceptions and damage observations, are correlated to each other and to the more
objective Richter scale. Figure 3 also gives an idea of the maximum surface acceleration which is
attributable to the various scales and, as such, it can be used as support for design purposes. The
diagram shows values at the conservative side (Klein, 1990), at least for higher intensities. This is
intentional for safety reasons, as the statistical analyses do not contain any further safety factors.
First, in order to evaluate the liquefaction potential, the Seismic Stress Ratio SSR = h /v0 developed in the field at level ground conditions is determined. This is, depending on depth, the ratio
between the shear force created by the assumed earthquake magnitude and the effective overburden pressure (Seed/Idriss/Arango, 1983).

h
a

= 0.65 max v 0 rd
v 0
g v 0
where
h

seismic shear force [kN/m]

v0

total overburden pressure [kN/m]

v0 =

effective overburden pressure [kN/m]

a max =

maximum earthquake acceleration [m/sec2]

0.65 =

average value of accelaration in comparison to amax

acceleration due to gravity [m/sec2]

rd

reduction value depending on depth [ 1 0.012 z]

depth [m]

The stress ratio that the soil can resist, is found by statistical analyses of a multitude of soundings
at earthquake affected subsoils. The first of these analyses, related to a magnitude of M = 7.5, was
originally based on Standard Penetration Tests (Seed/Idriss/Arango, 1983). It was then continued,
attuned to the more reliable Static Cone Penetration Tests and refined with regard to the silt
content of soils, by other investigators (Robertson/Campanella, 1985; Stark /Olson, 1995).
Figure 4 shows the relationship between seismic stress ratio and modified cone resistance qc1 at
the boundary state.

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

0,5

0,4

Sandy Silt

Silty Sand

D50 [mm] < 0.10

F. C. [%] > 35

0.10 < D 50 [mm] < 0.25

5 < F. C. [%] < 35

Magnitude M = 7.5

0,3

0,2
Clean Sand
0.25 < D 50 [mm] < 2.0
F. C. [%] < 5

0,1

qc1 = 180 qc / (80 + v0)

0
0

10

12

14

16

Stark and Olson, 1995

Corrected CPT Tip Resistance qc1 [MPa]

Figure 4: Relationship between seismic stress ratio and CPT tip resistance for sandy soils

Cq

q c1 = Cq q c
where

1. 8
0. 8 + v 0 100

qc = cone resistance [MPa]

The evaluation proposed in the German manual Grundbau -Taschenbuch coincides principally with
the procedure outlined above (Klein, 1990).
The method was further modified with the introduction of the friction ratio between sleeve
friction and cone resistance. In this case, the determination of the grain size distribution can be
omitted (Suzuki/Koyamada/Tokimatsu, 1997). In the publication concerned a slightly modified formula, as given below, is used to determine the relevant seimic stress ratio (abbreviations as before).

a max v 0
h
= 01
. ( M 1)

(1 0.015 z )
g v 0
v 0
where

M = earthquake magnitude

In this formula, the average value of the acceleration is related to the magnitude M, and the reduction value rd is slightly modified. The diagram shown in figure 5 should be used with modified
parameters given below, where sleeve friction and cone resistance is considered via a soil behavior
type index Ic .

?
Heinz-Joachim Priebe

0,5
4
Q = (qc v0) / v0
F = 100 fs / (qc v0)

[%]

F(Ic)

Shear Stress Ratio d / v

0,4

Ic = [(3.47 log Q) + (log F + 1.22)2 ]0.5

2
1

0,3

0
1

1,2

1,4

1,6

1,8

2,2

2,4

2,6

Soil Behaviour Type Index, Ic

0,2

0,1

0
0

10

12

Adjusted Tip Resistance, qt1 x F(Ic) [MPa]

14

16

18

20

Suzuki et al., 1997

Figure 5: Relationship between seismic stress ratio and CPT tip resistance

Q = ( q c v 0 ) v0
F = 100 fs

(q

v 0 ) [%]

Ic = (3.47 log Q ) + ( log F + 1.22 )

2

fs = sleeve friction

2 0. 5

In conclusion, it should be noted that the evaluation of the liquefaction potential by comparing
the expected forces of an earthquake with those that the soil can actually resist, is not a safety
analysis in the customary way, as the magnitude of an earthquake is a random variable.

4 Performance of Vibro Replacement

Vibro Replacement is more suitable than most other foundation techniques in preventing liquefaction during a seismic event. Generally, the problem can be solved by three alternatives, for
which vibro replacement offers the best conditions: soil compaction, drainage and increase in
shear resistance.
Which of these prevails with regard to vibro replacement, depends on the grain size distribution
of the soil concerned. Figure 2 shows a dashed transition zone relating to deep vibratory
compaction techniques, which is located more or less in the middle of the area susceptible to
liquefaction, and plays a dominant role.

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

It is assumed here that the two main compaction techniques, vibro compaction and vibro replacement, are already well known. Therefore, only the main features are described here, in order
to understand their performance in soils susceptible to liquefaction. Vibro compaction
increases soil density, while vibro replacement, in its pure application, reinforces cohesive, noncompactible soils, using load carrying columns of imported coarse aggregate such as gravel or
crushed material. Often, vibro replacement has the combined effect of considerable densification
of the surrounding soil by vibrational effects during the installation of stone columns. As in static
cases, it is also useful in dynamic events to consider both components of improvement separately, namely the soil densification and the improvement by the stone columns.
Soils with a grain size distribution curve entirely outside the transition zone at the sandy, gravelly
side are suitable for treatment solely by vibro compaction. If the distribution curve is totally within
the dashed zone, or partly within it with the remainder on the coarse side, it is advisable to add
backfill material, that is, install a stone column. The reason for this is to improve transmission of
vibrations via the imported material, as the main target is still the densification of the surrounding soil. In all other cases, where the grain size distribution curve is partly or entirely to the left
of the transition zone, that is, in the silty clayey area, substantial improvement in the densification
due to vibration can not be expected, and stone columns have to be installed which provide
considerable improvement due to their stiffness and shear resistance.
Therefore, as shown in figure 2, the transition zone between vibro compaction and vibro replacement subdivides the range of soils susceptible to liquefaction into three parts. Generally, as described above, for soils that are within the sandy range, vibro compaction without imported backfill
material is suitable for preventing earthquake induced liquefaction. This can be substantiated by
cone penetration tests as outlined in section 3 above. In addition, in both the other cases, where
vibro replacement is either recommended or definitely required, the improvement by densification
can also be substantiated by the same method. However, if the favourable effects of the stone
columns with respect to drainage capability and increase in shear resistance are ignored, the result is a very conservative treatment design.
The type of drainage which contributes considerably to the reduction of pore water pressure
during the usually very short period of an earthquake, and which thereby reduces the risk of
liquefaction, can only be expected in densifiable soils where vibro compaction alone is recommended. In finer grained soils with relatively low permeability, substantiating a reduction in
porewater pressure, and hence proving that the risk of liquefaction has been reduced, becomes
difficult. However, even if a considerable reduction in the pore water pressure cannot be proved,
even at short distances from the column, the high permeability of the columns themselves is
decisive for their bearing performance. A high gradient at the periphery of the columns provides
the required lateral support, and hence the bearing capacity, even at moments where the soil
between the columns could tend to liquefy.
Columns which are laterally supported in the limiting state by the difference in hydraulic pressure provide considerable shear resistance, especially due to load concentration. However, compared to piles, they have the advantage of high flexibility which can absorb the amplitudes occurring during an earthquake, without losing their bearing capacity. This implies that seismic shocks
are not reduced significantly by vibro replacement. Therefore, the treatment does not necessarily
provide protection for the buildings, but is primarily concerned with the prevention of liquefaction. Any conclusion should not discount that in some cases, buildings which might have otherwise been destroyed by the shaking of non-liquefied soil have remained relatively undamaged
solely because of liquefaction.
7

?
Heinz-Joachim Priebe

Since it is difficult to quantify the favourable effects of stone columns in seismic events, in some
cases they were not considered in the design at all. In some other cases, where the densification
of the soil just missed the specifications, they were given a qualitative consideration only, in the
sense that they would compensate for the deficit in soil densification. Such an approach is not
satisfying and even rough approximations on a static basis are preferable.

5 Evaluation of Liquefaction Potential with Vibro Replacement

It is not possible to estimate, by statistical analyses, the extent to which the risk of liquefaction is
reduced by vibro replacement. Fortunately, no failure has occurred so far, even though the application of this method of improvement has been steadily growing. Nevertheless, such analyses for
vibro replacement would be difficult to be implemented due to additional variable parameters
attributable to the arrangement of the columns.
As the existing analyses collected so far have already proved reliable, a different approach can be
made in the design in order to include them. The question does not then become, to what extent have the shear resistance and load bearing capacity been changed by the stone columns, but
rather, which part of the forces exerted by an earthquake are borne by the columns without any
damages? Such a step, namely reducing the acting forces instead of increasing the resisting ones,
is permissible as long as the evaluation, as previously mentioned, is not a customary determination of safety.
It is difficult to determine what amount of the acting forces in a seismic event is taken by the
stone columns. Evaluations on the basis of computer simulations or theoretical approaches are
more suitable than laboratory tests, in which possibilities are limited. A relatively simple proce-

Improvement Factor n

C = 45.0

S = 0,5

C = 42.5

4
C = 40.0
C = 37.5

3
C = 35.0

1
1

Area Ratio A / AC

Figure 6: Design chart for vibro replacement

10
Priebe 1998

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

dure for the design of vibro replacement in static cases was outlined comprehensively in Ground
Engineering (Priebe, 1995). This method was introduced a while ago in Germany and has since
proved reliable. Static cases mostly comprise long-term processes, and therefore, as a general
approximation, a Poissons ratio of S = 1/3 was proposed.
On the other hand, for short-term seismic events, it seems more realistic to consider deformations of the soil with the volume remaining constant, that is, to calculate with S = 0.5 which also
simplifies the formulae. In the above mentioned procedure the improvement factor n0 , which is
the basic value of improvement by vibro replacement, is determined initially using some
simplifications and approximations. It is shown in figure 6.

n0 = 1+

AC
1

1
A K aC ( 1 A C A )

where

A
AC
C

=
=
=

K aC = tan 2 ( 45o C 2)

attributable area within the compaction grid

cross section of stone columns
friction angle of column material

The reciprocal value of this improvement factor is merely the ratio between the remaining stress
on the soil between the columns pS, and the total overburden pressure p taken as being uniformly distributed without soil improvement and, as such, can be used as a reduction factor. It is
simple to calculate this value or it can be taken from figure 7.

0,8

Stress Ratio pS / p

C = 35.0
C = 37.5

0,6
C = 40.0
C = 42.5

0,4
C = 45.0

S = 0,5

0,2

Area Ratio A / AC

Figure 7: Residual stress on soil between stone columns

10
Priebe 1998

?
Heinz-Joachim Priebe

= p S p = 1 n0 =

K aC (1 A C A)
A C A + K aC (1 A C A)

On the understanding that the loads taken by the columns from both the structure and the soil
do not contribute to liquefaction, it is proposed to use this factor to reduce the Seismic Stress
Ratio created by an earthquake, and hence evaluate the remaining liquefaction potential as outlined in section 3.
This procedure represents an approximation which, although not being completely satisfactory
from a geotechnical point of view, nevertheless does realistically consider the stabilising effect of
vibro compaction.

6 Case Histories
The first project were vibro replacement was used to reduce the risk of liquefaction in a seismic
event was at the waste water treatment plant of Santa Barbara, in California, USA. This project is
relatively well documented and especially interesting, as it has already withstood an earthquake
where the acceleration corresponded to the one considered in the design (Mitchell, 1986)
More recently, this improvement technique has been increasingly applied against the risk of earthquakes in a wide range of applications, from bridges and barrages to industrial and marine facilities (Dobson, 1987).
To date, only a few facilities constructed on ground improved by vibro replacement have actually
been subjected to earthquakes. However, in such cases, no serious results have been observed,
and therefore, at least qualitatively, vibro replacement has proved to be a success (Mitchell / Wentz,
1991).
On a more recent project for oil storage tanks on the Black Sea coast in Georgia, vibro replacement has been carried out to reduce both the settlements and the risk of liquefaction. A soft
silty clayey top layer, with thickness ranging from 1m to 4 m, is underlain by mostly sandy strata
with varying silt content up to a depth of approximately 22 m which, in turn, is underlain by
calcareous hard clay. The groundwater table lies at around 2 m below ground level.
The tanks, already under construction, have a diameter of 66 m and, in service condition, exert a
foundation pressure of up to 150 kN/m. With regard to earthquakes, the design was based on a
maximum acceleration of a max = 0.25 g which corresponds to a magnitude of M 6.5 according
to figure 3.
For the tank used in the design example below, the soft top layer was completely removed by
soil replacement up to a depth of 2.7 m. Vibro replacement was then carried out down to a
maximum depth of 11.5 m, with up to 1 m diameter columns placed on a square grid at 3 m
spacing.
Calculations for the design, utilising the relevant formulae and diagrams, were performed using a
computer program, and the results are shown in figure 8 below. The upper level of the layers
concerned is of main interest with regard to the required cone resistance. For the silty sands
beneath the soil replacement, fines contents of 15 per cent were assumed, and the bulk densities

10

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

estimated. The measured cone resistance decreased considerably at 8 m depth, most probably
due to a silt content higher than the 15 per cent considered, which is also demonstrated by
densification of the soil not being achieved by the treatment below this depth. Therefore, in this
instance, it is acceptable that the required cone resistance slightly exceeds the measured one.
The same also applies at the full depth of treatment. For the rest, the measured values are well
above those required and, therefore, there is no concern at all with regard to liquefaction of the
upper strata which are generally more crucial.

SUPSA Terminal, Tank D

Maximum Earthquake Acceleration amax = 0.25 g

Magnitude M = 6,5

Parameters of Vibro Replacement

Grid Size A
Column Diameter D
Area Ratio AC /A
Backfill Material
Coeff. of Act. Earth Pressure KaS
Reduction Factor 1)

General Parameters

Ground Water Table

Surface Load

9.00 m
1.00 m
0.09
42.50
0.19
0.71
1) Priebe, Kolloquium Darmstadt, 1998

Calculation

2.00 m
0.00 kN/m

2) Stark and Olson, Journal of

Geotechnical Engineering, 1995

2)

F. C.

Depth

[%]

[m]

5
5
15
15
15
15

0.00
2.00
2.70
4.00
8.00
11.50

[kN/m]

19.0
19.0
18.0
18.0
18.0
18.0

11.0
11.0
9.0
9.0
9.0
9.0

v0

v0
[kN/m]

0.0
38.0
52.7
77.4
153.4
219.9

0.0
38.0
45.7
57.4
93.4
124.9

rd

SSR

* SSR

qc1

Cq

qc, requ.

[MPa]

1.00
0.98
0.97
0.95
0.90
0.86

0.000
0.159
0.181
0.209
0.241
0.247

0.000
0.113
0.129
0.148
0.172
0.175

Figure 8: Evaluation of the liquefaction potential at SUPSA Terminal, Tank D

11

0.0
6.8
4.7
5.5
6.4
6.6

qc, meas.

[MPa]

2.250
1.525
1.432
1.310
1.038
0.878

0.0
4.5
3.3
4.2
6.2
7.5

0.0
10.0
10.0
10.0
6.0
7.0

?
Heinz-Joachim Priebe

References
Department of the Navy, Naval Facilities Engineering Command, Design Manual 7.3 (1983)
Soil Dynamics, Deep Stabilization, and Special Geotechnical Construction
Dobson, T. (1987)
Case Histories of the Vibro Systems to Minimize the Risk of Liquefaction.
Soil Improvement - A Ten Year Update, Geotechnical Special Publication No. 12
Engelhardt, K., Golding, H. C. (1975)
Field testing to evaluate stone column performance in a seismic area. Geotechnique 25
Herzog, M. (1980)
Zwei Versagensarten des Baugrundes unter Erdbebenwirkung. Bauingenieur 55
Klein, G. (1990)
Grundbau-Taschenbuch, 4. Auflage, Teil 1
Mitchell, J. K. (1986)
Material Behaviour and Ground Improvement.
International Conference on Deep Foundations, Beijing, Vol. 2
Mitchell, J. K., Wentz, F. J. 1991
Performance of Improved Ground during the Loma Prieta Earthquake.
Report No. UCB/EERC-91/12
Peck, R. B. (1979)
Liquefaction Potential: Science Versus Practice.
ASCE, Journal of the Geotechnical Engineering Division, Vol. 105
Prater, E. G. (1977)
Verflssigung von Bodenschichten infolge Erdbeben.
Mitteilungen der Schweizerischen Gesellschaft fr Boden- und Feldmechanik, Nr. 97
Priebe, H. J. (1989)
The prevention of liquefaction by vibro replacement.
Conference on Earthquake Resistant Construction and Design, Berlin
Priebe, H. J. (1995)
Die design of vibro replacement. Ground Engineering, Vol. 28, No. 10
Robertson, P. K., Campanella, R. G. (1985)
Liquefaction Potential of Sands Using the CPT.
ASCE, Journal of Geotechnical Engineering, Vol. 111
Seed, H. B., Booker, J. R. (1977)
Stabilization of Potentially Liquefiable Sand Deposits Using Gravel Drains.
ASCE, Journal of the Geotechnical Engineering Division, Vol. 103
Seed, H. B., Idriss, I. M., Arango, I. (1983)
Evaluation of Liquefaction Potential Using Field Performance Data.
ASCE, Journal of Geotechnical Engineering, Vol. 109

12

?
Vibro Replacement to Prevent Earthquake Induced Liquefaction

Singh, S. (1996)
Liquefaction characteristics of silts. Geotechnical and Geological Engineering
Stark, T. D., Olson, S. M. (1995)
Liquefaction Resistance Using CPT and Field Case Histories.
ASCE, Journal of Geotechnical Engineering, Vol. 121
Suzuki, Y., Koyamada, K., Tokimatsu, K. (1997)
Prediction of liquefaction resistance based on CPT tip resistance and sleeve friction.
International Conference on Soil Mechanics and Foundation Engineering, Hamburg
Wenk, T., Schwarz, J. (1995)
Das Groe Hanshin-Erdbeben vom 17. Januar 1995. Bautechnik 72

13