Construction of Flat Glass Float Factory on Loess Deposits Using

Controlled Modulus Columns in Seismic Condition

C. Plomteux & P. Liausu

Menard Soltraitement, Nozay, France
[email protected]

Abstract: The Saint-Gobain Glass Calaraşi project consists of constructing a 21,500m² rom float in Calaraşi, Romania. This warehouse
has a metal structure supported on piles. Ground improvement is used under the slab-on-grades in order to reduce differential settle-
ment under heavy live loading (80 kN/m²) to a maximum of 1/500 differential settlement over the whole building. The ground condi-
tion comprises of 6 to 10m of loess deposits overlying 6m of alluvium clay. Controlled Modulus Columns are used to minimize settle-
ment under heavy live loadings. This consists of reinforcing the soil by means of semi-rigid cement grout inclusions. This paper
presents the procedures used for the design of the semi-rigid inclusions in soft soil under seismic condition and heavy loading.

1 INTRODUCTION granular material that distribute the loads uniformly throughout

the soil mass.
1.1 Ground Improvement with Semi-Rigid Inclusions The CMC system uses a displacement auger powered by an
equipment with very large torque capacity and very high down-
The concept of semi-rigid inclusions is fairly old. A network of ward thrust, which displaces the soil laterally with virtually no
wooden piles installed under ancient churches is indeed one of spoil or vibration. The auger is screwed into the soil, and when
the first examples of application. The concept is to improve the the required depth or a preset drilling criterion is reached, a
soil globally by the use of semi-rigid soil reinforcement columns. highly workable grout-cement mixture is pumped through the
These inclusions can be installed by various methods (percus- center of the hollow auger. The cement based grout then flows
sion, vibration, soil displacement, etc). The final objective is to under low pressure (typically less than 5 bars) out of the auger
obtain an improved ground with bearing capacity compatible base as it is retracting and results in a 100% cement grout column
with the structure to be built. This type of soil improvement solu- that can be used in close vicinity of sensitive structures and that
tion does not aim to bypass the compressible ground by installing generates virtually no above ground spoils. No soil mixing takes
piles that will directly support the entire load imposed by the place during the pressure grouting. Fig. 2 shows the installation
structure but rather, it is to improve the soil globally and to re- process.
duce its deformability. Fig. 1 illustrate the concept of ground im-
provement using semi-rigid inclusions compared with rigid piles.

Fig. 2 CMC installation process

The main objective is to develop an optimal distribution of

Fig. 1 Semi-rigid inclusion versus piles load between the soil and the columns while aiming to develop
the full potential of the founding strata. The dimensions, spacing,
1.2 Controlled Modulus Columns and material of the CMC are based upon the development of an
optimal combination of support from the columns and the soil
The principle of installing Controlled Modulus Columns (herein- mass to limit settlements within the allowable range and to obtain
after referred to as CMC) as semi-rigid inclusions is to form a the design value for the equivalent deformation modulus of the
composite material. These columns are usually associated with a improved soil.
load distribution platform made of good quality well compacted
2 PROJECT DESCRIPTION – SAINT-GOBAIN GLASS Table 1 General soil parameters
CALARAŞI Loess deposits Alluvium clay Sand & gravels
Thickness [m] 8.6 6.0 2.0
The Saint-Gobain Glass Calaraşi project consists of the construc- γ [kN/m3] 18 18 18
tion of a rom float in Calaraşi, Romania. Due to local conditions, E [MPa] 3. 5 13 45
Saint-Gobain has chosen the town of Calarasi (approximately ν [-] 0.3 0.3 0.3
100 kilometres east of Bucharest) being the project site where the c [kPa] 6 10 0
soil conditions which consisted of compressible subsiding loess ϕ [°] 25 25 36
and soft clay and having potential seismic risks inevitably re-
quired ground improvement for the construction. A test area was implemented in order to validate the settle-
ment potential of the loess deposits. A 30 m × 30 m trial em-
bankment of 6.5m high was instrumented for settlement. Fig. 4
2.1 Project Description shows the settlement monitoring results.
The warehouse consists of a metallic structure supported on piles. 450

The final slab level is about 1.5 m above natural ground level. 400

The slab-on-grades is 25 cm thick and construction joints are 350

made around each foundation.
300
Ground improvement works is mainly on the slab-on-grades

Settlement (mm)
reinforcement of the warehouse measuring an area of about 250

21,500 m². Ground improvement is carried out to limit the differ- 200

ential settlement under live load of 80 kN/m² to 1/500. 150

There is no limitation on the absolute settlement. However, 100

these absolute settlements need to be limited typically to 30 - 40
50
mm in order not to influence the reliability of the differential set-
tlement predictions. 0
18/03/05 25/03/05 01/04/05 08/04/05 15/04/05 22/04/05 29/04/05 06/05/05 13/05/05

Fig. 4 Test area – settlement versus time

2.2 Soil Conditions
Site investigation carried out in 2004 shows the following subsoil The results of the trial embankment were used to back-
profile: analysed the geotechnical design parameters and the soil profile
- Top soil of 0.3 m thick used. Compaction and vibration tests also confirm that the loess
- Loess deposits over a thickness of 6 to 10 m deposits were not collapsible.
The superficial ground water table was found to be about 5 m
- Alluvial clay down to 14 to 17 m deep
deep in the loess deposits while a deep pressurized water table
- Clay sand over a thickness varying from 0 to 3 m was about 14 to 17 m deep below the alluvium clay layer. The
- Sand and gravel down to 26 to 29 m deep phreatic surface of this water table head is pressurized up to the
- Clay-marl complex first water table level (5 m deep from NGL).
- sandstone The Calaraşi area has a significant seismic activity, and the
Design Basis Earthquake (DBE) characteristics are as follow:
The soil conditions are quite homogeneous over the project
- aN = 0.262 g,
site and typical average geotechnical conditions are presented in
Table 1. Fig. 3 represents a typical CPT profile on the project. - Design earthquake magnitude : MW = 7,
- Building class : C (appendix 2 in PS92),
qc (MPa) Rf (%)
0 2 4 6 8 10 12 14 0 2 4 6 8 10 12 14
- Site type : S2 (soil type b over thickness lower than 50 m).
0 0

2 2
2.3 Ground Improvement Solution
4 4 An alternative solution proposed by Menard was chosen in pref-
erence to the initial exhibited design calling for a pile supported
6 6 slab. It consists of constructing a conventional slab-on-grades
with a thickness of 25 cm resting on soil reinforced with a grid
8 8
network of CMC with a well compacted sand-gravel load distri-
z (m)

z (m)

bution platform of 60 cm thick below the slab.

10 10
The design of the ground improvement by CMC is based on
12 12
elastic and plastic FEM calculations. Design models are based on
axial-symmetrical calculations whose general principle is pre-
14 14 sented in Fig 5. Considering a quasi-infinite grid of CMC, a regu-
lar square grid with center-to-center spacing L and section L², is
16 16 quasi-equivalent to a cylindrical grid of radius L π (equiva-
lent volume) with an axial-revolution symmetry (Fig. 5).
18 18

Fig. 3 Typical CPT profile

 The base case corresponds to the design parameters presented
in Table 1. Based on the site investigation and laboratory testing,
these soil parameters correspond to the best possible correlation
to the existing soil conditions and they are considered as the base
conditions for the design of the CMC reinforcement.
Sensibility calculations have been made based on different pa-
rameters to evaluate their impacts on total and differential settle-
ment as presented in Table 2. Design variations have been made
on:
- Deformation modulus of the CMC from 5,000 MPa (base
case) to 15, 000 MPa
- Equivalent deformation modulus of the loess deposits from 1
MPa to 3.5MPa (base case)
- Length of the cut-off of the CMC from 0 to 0.2 m (base case)

Table 2 Sensibility calculation over settlement

Settlement Stress in CMC
(cm) (MPa)
Base case 3.34 3.77
CMC 10 000 MPa 3.05 3.84
Fig. 5 General principle of axial-symmetrical models modulus 15 000 MPa 2.95 3.86
Loess 1 MPa 4.24 3.97
Considering the heavy live load applied to the slab, the design modulus 2 MPa 3.72 3.88
is on the settlement evaluation and the determination of the bend-
Cut-off 0.0 m 3.57 3.75
ing moment in the slab-on-grades. Sensibility calculation have
Average value 3.48 cm 3.85 MPa
been made on the soil parameters and loading conditions.
In order to take into account different slab loadings, 2D plane
From Menard D.60.AN design documents, the maximum dif-
strain models where the treated soil is considered as an equiva-
ferential settlement is given by the standard deviation for points
lent improved ground are considered. From the previous axial-
at 10 m apart and it is expressed by the following equation:
symmetrical FEM calculations, the equivalent modulus of the
ground improved by CMC can be obtained as presented in Fig. 6.
∑ (w − w)² = 0.433 cm over 10 m, (2)
∆w = S = n

n
Thus, ∆w = 0.86/500 < 1/500 which is acceptable.

In order to estimate the imposed stresses in the slab, two types

of calculations are carried out as follow:
(1) Axial-symmetrical calculations to evaluate stresses induced
by the presence of the CMC columns. The main results are
presented in Table 3 below:

Table 3 Sensitivity calculation over bending moment in the slab

Bending moment in the slab
(kNm/m)
Fig. 6 General principle of axial-symmetrical models Base case 8.63
CMC modulus 10 000 MPa 8.99
The equivalent oedometric modulus of the improved ground is 15 000 MPa 9.10
calculated from the vertical displacement measured between the Loess modulus 1 MPa 11.49
top of the load distribution platform and below the tip of the 2 MPa 10.10
CMC column where settlement are homogeneous over the sec- Cut-off 0.0 m 11.27
tion using the following equation: Average value 9.93 kNm/m

∆σ × h (2) 2D plane-strain model using equivalent material for the

E oed = (1)
treated soil to evaluate the stress induced by the different live
∆h
loadings on the slab was carried out. The calculations are car-
ried out for an applied live load of 80 kN/m² of different
where ∆h is the differential settlement between the top of the
width ranging from 1 m to 6 m. The main results are pre-
load distribution platform and the tip of the CMC; h is the dis-
sented in Table 4 and Fig. 7:
tance between those two points; ∆σ is the applied load; Eoed is
the equivalent oedometric modulus of the reinforced ground.
Fig. 7 2D plan-strain model for evaluation of bending moment in
slab-on-grades
Fig. 9 CMC installation on site

Table 4 Bending moment versus width on loading

2.4 Seismic Effect on CMC – Soil Displacement
Differential Bending moment
Width of loading
settlement in the slab During an earthquake, the waves transmitted by the substratum
(80 kN/m²)
(mm) (kNm/m) are transformed into shear waves that are spreading upward to the
1m 0.13 4.23 surface layers.
2m 1.01 8.85
3m 2.62 10.08
2.4.1 Free field motion
4m 4.52 9.60
5m 6.44 8.77 During DBE (Design Basis Earthquake), the maximum associ-
6m 8.33 8.41 ated strain has the shape of a quarter sine curve. In which case, it
60 m - 7.87 has a maximum surface displacement dmax given by Eq. (3):
Average value 3.84 mm 8.26 kNm/m
2
ρ S ⎛ 2.H ⎞ = 3.04 cm
d max = λ.aN . ⎜ ⎟ (3)
12 GS ⎝ π ⎠
10 where
- λ = 0.9 (based on site type S2)
bending moment (kN.m/m)

8
- aN : ground peak acceleration (equal to 2.62 m/s²)
6
- H thickness of the soil to substratum (equal to 49 m)
4
- ρs = Σ ρi × Hi / Σ Hi is the unit volume mass of the soil
2 (equal to 1792 kg/m3)
- Gi = 0.5 Gimax is the dynamic shear modulus of the consid-
0
0 2 4 6 8 10 12 14 16 18 20 ered soil layer (cf. §9.4222 in PS92)
Load band w idth (m)
- Gs = Σ Gi × Hi / Σ Hi is the dynamic shear resistance modulus
of the soil profile (equal to 91 MPa; taking into account iter-
Fig. 8 Bending moment versus width of loading
ated shear modulus).
Bending moment reaches a peak value for a loading width of aN dmax
about 3 m. Fig. 8 shows the variation of bending moment against
loading width
These two effects are combined and the maximum total bend- ρ1, G1
ing moment in the slab-on-grades was estimated to be around M
= 18kN.m/m. Hence, the slab-on-grades needs to be lightly rein-
forced. ρ2, G 2 CMC
This solution was able to guarantee an absolute residual set-
tlement of about 3.5 cm and differential settlement lower than
H
1/500.
The final design of 2.25 m center-to-center square grid with ρ3, G 3
360 mm diameter CMC at an average depth of 17 m was adopted
and constructed in Calaraşi. The installation works commenced
in July 2005 working with 12 hours shifts. The works was suc-
cessfully completed in August 2005 in just two months’ time. Fig. 10 Free field displacement
Fig. 9 shows the CMC installation rig.
2.4.2 Bending moment and shear stress in the CMC Table 5 Bending moment and shear force in CMC
Applied solicitations Results
In this scheme, shear force and bending moment in the CMC are
Horizon- Maxi- Maximum
calculated as columns under lateral forces and displacement, us- Vertical Surface

case
tal load mum shear
ing a finite difference elasto-plastic calculation following the load in displace-
at CMC bending force in
equation below: the CMC ment
head moment CMC
δσ.Β = Ks × B × δy (4) 1 0 kN 0 kN 0.0304 m
0.4 kNm 0.3 kN
2 384 kN 0 kN 0.0304 m
where:
- ks.B : reaction modulus of the soil applied on the width of
the CMC (B), 2.4.3 Compression, tensile and shear stress in the CMC
- δσ : differential pressure of the soil between each side of the Under both axial force and bending moment (Fig. 12), stresses in
CMC with δσ limited to the creep pressure pf, the CMC are given by Eq. (7):
- δy : differential displacement between soil and inclusion
Ri Mi
The reaction modulus of the soil against the CMC is calcu- σ = ± (7)
lated using the short-term pressuremeter Eq. (5) below: π .D 4 π .D 3 32
2

12Em where Ri is the axial compression force and Mi the bending mo-
KsB = ment
4
2.65α + α
3 (5)

pl
and p f .B = .B (for B = 0.36 m < 0.6 m) (6)
1.8 σ σ ≤ σ
The main results of those calculations are presented in Fig. 11 σadm = 4 MPa
σM = σN
and Table 5. Those results show that bending moment and shear
force induced by soil horizontal displacement during earthquake σ σ σ σ
are limited to low values.

Shear force (kN) Bending moments (kN.m) + =

0,00
0,05
0,10
0,15
0,20
0,25
0,30
0,35
0,40
0,45
-0,40
-0,30
-0,20
-0,10

σ σ
0,00
0,10
0,20
0,30

0
0 σ

2 σ σ σ
2
σadm = 4 MPa
4 σM = σ σN
4

6
6 σ σ σ σ
depth Z (m)

depth Z (m)

8 8

10 10 Fig. 12 CMC under axial force and bending moment

12 12 Table 6 Bending moment and shear force in CMC

Applied solicitations Results
Maximum
case

14 14
Vertical load bending Maximum
compression
in the CMC moment tensile stress
16 16 stress
1 0 kN 0.087 MPa -0.087 MPa
18
0.4 kNm
18 2 384 kN 3.859 MPa 3.685 MPa

Fig. 11 Shear force and bending moment under free field The compression strength at 28 days of the CMC material is
displacement around 12 MPa, giving a factor of safety of more than 3 on com-
pression strength during earthquake.
Concrete with compression strength fck28 = 12 MPa have a ap-
proximate tensile strength fctm = 1.3 MPa. Tensile stresses occur
in unloaded CMC during earthquake but they remain within ac-
ceptable ranges (less than 1.3 MPa). Steel reinforcement is thus - Calculation of the critical buckling force Nc
not necessary.
In this case, the critical buckling force calculated with
Mandel’s method is Nc = 277 tons. Usually, a 2.5 factor of safety
2.4.4 Buckling of the CMC is adopted for serviceability limit state (SLS). The maximum
load applied to the CMC is N= 384 kN = 38.4 tons which is
Check for non-buckling of CMC (Φ=360 mm and D= 16 m) is much lesser than the safe critical buckling load of 111 tons
carried out using Mandel’s method (Fig. 13). This method is used (computed as Nc/2.5 = 111 tons).
to evaluate the critical buckling load Nc of a CMC installed in a Hence, there is no risk of column buckling for the CMC inclu-
soil characterized by its surface horizontal reaction coefficient kh. sions in this case.

REFERENCES

Combarieu, O. 1988. Amélioration des sols par inclusions rigides

verticales – application à l’édification de remblais sur sols
médiocres, Revue Française de géotechnique n°44: 57-59
Combarieu, O. 1988. Calcul d’une fondation mixte, Note
d’information mixte LCPC
D.60.AN - Interpretation and Application of Pressuremeter test
results to foundation design, Sols Soil N°26, 1975
Fascicule N°62 – Titre V
Seed R. & Booker J. 1976. Stabilization of potentially liquefiable
sand deposits using gravel drain systems, ASCE GT Journal
407 – p201-255
Fig. 13 Graph from Mandel for buckling assessment Youd T. L. & Al. 2001. Liquefaction Resistance of Soils: Sum-
mary Report from the 1996 NCEER and 1998 NCEER/NSF
Workshops on Evaluation on Liquefaction Resistance of
The calculation is conducted with the following steps:
Soils, Journal of Geotechnical and environmental engineer-
- Calculation of the reduced semi-length λ (no unit) ing / October 2001 / 817
- Calculation of reduced force φ from Mandel’s abacus accord-
ing to the corresponding limit conditions.