Académique Documents
Professionnel Documents
Culture Documents
) 2003 VGE
Fabian Kirsch
Dipl.-Ing., Institute for Soil Mechanics and Foundation Engineering, TU Braunschweig
Wolfgang Sondermann
Dr.-Ing., Keller Grundbau GmbH, Offenbach
ABSTRACT:
Since decades vibro stone columns are used to improve the bearing capacity and the settlement
behaviour of soft soils. The design is based on empirical or semi-analytical formulae, most of
which using the unit cell approach. Such design procedures, e.g. the Priebe method, proved their
reliability in many cases. Stone columns are also used to improve the ground below embankments
constructed for infrastructural measures. In such cases the columns act predominantly in order to
enhance the slope stability. To give a realistic picture of the actual situation the design of the stone
column pattern needs to take into account the stress distribution between columns and soil. All
approaches not considering the stress concentration in the columns gain results which are over-
conservative and thus lead to uneconomical solutions.
In the paper the results of numerical analyses considering the spatial nature of the problem are
presented. The results are compared with those of analytical approaches. In addition to the
numerical analysis the results of long term monitoring of the stresses at a case history are
presented. The knowledge of the stress distribution offers the possibility of a realistic approach for
the calculation of the slope stability of embankments on improved ground.
Stone columns can be used to improve soft layers under dams and embankments in order to
reduce the settlements, accelerate the consolidation process and increase the stability.
They are installed using either the vibro replacement or the vibro displacement process. Figure 1
depicts the different construction stages. More detailed descriptions of the equipment and the
procedure itself can be found in Moseley & Priebe (1993) or Kirsch & Sondermann (2003).
In order to assess the settlement reduction factor numerous analytical and semi-analytical
approaches exist. One of the most common design procedures is the method developed by
Priebe (1995). An example of his design is given below, where an embankment of 15 m height is
being constructed for the dam close to a bridge abutment at a highway crossing in Kuala Lumpur.
Figure 2 shows the geometry.
15,00
20 kN/m 2
2,00
5,00
10,00 10,00
2,00
15,00
5,00
1
3,00
25,00
10,00
5,00
± 0,0 1
2,00
Stopfsäulen: Abstand 1,7 m Stopfsäulen: Abstand 2,1m
Stone columns Ø1.1m @1.7m
Ø 1,10 m
Stone columns Ø1.1m @2.1m
Ø 1,10 m
Adopting the Priebe method to this type of structure calls for several idealisations in order to
calculate the stresses at the base of the embankment. In this case a total settlement of 192 cm was
calculated for the centre of the dam.
The calculation of the slope stability is usually performed using analytical procedures like the
Bishop method. In order to use these analytical procedures some sort of homogenisation of shear
parameters has to be adopted. Generally it is proposed to use a weighted average for the shear
parameters of the unimproved soil and the stone column material. It seems however to be over-
conservative to calculate an average on the basis of the area ratio of columns and soil AS/A since
the columns concentrate the vertical stresses and therefore create a higher resistance against slope
failure. Priebe suggests a weighting of the angle of inner friction and the cohesion by the stress
distribution, which is a result of his analysis. Weighting the cohesion by the stress distribution has
no physical justification but accounts for conservatism.
In figure 3 the results of stability analyses using different mean friction values are shown.
Column 3 shows the result of an approach in which the shear parameters are weighted as an
average of area ratio and Priebe's method. The subsoil was divided into 14 different soil
compounds to compute the indicated factors of safety. One should emphasize, that the stress
concentration factor calculated by the Priebe method cannot simply be taken at the dam base, but
needs to be calculated for each and every layer. This stands because the stress distribution between
column and soil is by no means constant along the column depth.
1 2 3 4 5 6
2 7a 8a 9a
Average type Area-Ratio Priebe Average of 1 and 2
14,00
3 7b 8b 9b
2,00
1 2 3 4 5 6
There is quite a scatter in the results and since none of the methods has thorough physical
justification further investigation was deemed necessary.
3 STRESS DISTRIBUTION
3.1 Measurement
In order to gain more information about the stress distribution between column material and
surrounding soil pressure cells can be installed. When applied at the dam base usually the measured
stress distribution values n=σCol/σSoil vary between 2 and 3. The stress concentration is depending
on various parameters such as the loading type (soft or rigid), the surcharge, the material
parameters of column and soil and the geometrical dimensions.
Figure 4 shows the result of a measurement at another embankment site in Kuala Lumpur,
which shows approximately the same conditions as the example above. The columns were installed
using a square pattern at 2,2 m. The result of approx. n=2,6 compares well with the results of an
in situ trial field reported by Gruber (1994). There values of n=2,8 were measured for the same
column pattern and a surcharge of 120 kN/m².
3,0
2,5
Stress distribution n [-]
2,0
1,5
PC2
2,2
1,0 PC1
2,2
0,5
0,0
0,00 100,00 200,00 300,00 400,00
Surcharge [kN/m2]
σ 1'
elastic
hardening
σ 3' softening
residual strength
σ 2'
peak strength
Figure 6 shows the results of a simplified FEM analysis in which the typical stress distribution
between columns and soil can be seen. The calculated stress concentration results in n=3,1.
y
x
A B C
-2 0 2 4 6 8 10 12
50 kN/m 2
0
7,00
stress concentration n=3,1
3,00
18,00 12,00
-50
Vertikalspannung [kN/m 2]
D D
-100
6,00
-150
15,00
12,00
-200
3,00
0,70
3,00 -250
3,00
column no. 0 1 2 3 4
30,00
-300
-350
0 1 2 3 4
A B C -400
cross section vertical stresses at the dam base
4 SLOPE STABILITY
100
95
1
90 η FEM = = 1,36
% of phi-c
0, 735
85
80
75
70
0,00
0,10
0,20
0,30
0,40
0,50
Focus should be given to the fact, that there is obviously a second failure mechanism below the
berm at the embankment toe. This was not included in the analytical analysis and is probably one
reason why the safety factor in the numerical simulation is smaller than in the analytical
procedures.
REFERENCES
Cai, F. & Ugai, K. 1999. 3D FE-analysis of the stability of slope reinforced with piles. In Pande,
Pietruszczak, Schweiger (eds.) NUMOG VII. Rotterdam: Balkema. 541-546.
Gruber, F.J. 1994. Verhalten einer Rüttelstopfverdichtung unter einem Straßendamm. Diss. TU Graz.
Indraratna, B., Balasubramaniam, A.S. & Sivaneswaran, N. 1997. Analysis of settlement and lateral defor-
mation of soft clay foundation beneath two embankments. I. J. Num. Anal. M. Geom. 21. 599-618
Kirsch, K. & Sondermann, W. 2003. Ground improvement. In U. Smoltczyk (ed.), Geotechnical Engineering
Handbook. Vol. 2: 1-56. Berlin: Ernst & Sohn.
Moseley, M.P. & Priebe, H.J. 1993. Vibro techniques. In M.P. Moseley (ed.), Ground Improvement: 1-19.
Glasgow: Blackie.
Naylor, D.J.. 1999. On the use of the F.E.M. for assessing the stability of cuts and fills. In Pande,
Pietruszczak, Schweiger (eds.) NUMOG VII. Rotterdam: Balkema. 553-560.
Priebe, H.J. 1995. Die Bemessung von Rüttelstopfverdichtungen. Bautechnik 72. Heft 3. 183-191.
Raju, V.R. & Hoffmann, G. 1996. Treatment of tin mine tailings in Kuala Lumpur using vibro replacement.
In Proc. 12th SEAGC.
Sondermann, W. 1996. Rüttelstopfverdichtung zur Baugrundverbesserung für die feste Fahrbahn im
Schnellbahnbau. 3. Darmstädter Geotechnik Kolloq. TU Darmstadt. Heft 35. 147-164.
Sondermann, W. & Jebe, W. 1996. Methoden zur Baugrundverbesserung für den Neu- und Ausbau von
Bahnstrecken auf Hochgeschwindigkeitslinien. Vorträge der Baugrundtagung 1996 in Berlin. 259-280.
Vittinghoff, T., Plaßmann, B. & Schmitt, J. 1997. Programmentwicklungen im ANSYS-Open-System für
Anwendungen in der Geotechnik. 15th CADFEM Users' Meeting. Fulda. Part 1-30. 1-17.