You are on page 1of 7

Use of sand drains for improving

characteristics of loessail soil: a case study


A. Fouda, T. Lewis & G. Khan
The Louis Berger Group, Inc., Florham Park, NJ, USA

Abstract
Loessail soil is a collapsing soil type, sometimes referred to as metastable soil,
which undergoes a structural collapse if saturated. The collapse is due to the loss
of strength in the clay binders between the silt size particles, leading to
construction and design problems for highways and buildings. Use of sand
drains is one of the methods used, to facilitate the pre-wetting of the soil in order
to induce collapse and thus improve its characteristics and strength parameters.
Challenges facing the contractor and the design engineer include predicting the
total settlement associated with the soil collapse, and estimating the water
volume and time needed to complete the soil collapse program. In addition, a
monitoring program employing settlement plates and moisture sensors is
required to monitor change is soil moisture content and subsequent collapse
behaviour, if any. With all the uncertainties of the soil parameters and the
significance of all these parameters on the construction cost, especially the time
for the contractor on site, there is a necessity to come up with an analytical
solution based on the spacing between sand drains to predict the time needed for
complete saturation based on sand drain spacing. Formulas should also be
verified with field work and measurements. This paper presents a case study of
using sand columns to pre-wet loessail soils. This paper also presents an
analytical approach for calculating the time needed for collapsing loessail soils
using sand columns and verifies the approach by analyzing collected field data
during a saturation/collapse process of the soil.
Keywords: loessail soil, soil collapse, sand columns, wetting time.

1 Introduction
This paper presents a case study of use of sand columns to pre-collapse loessail
soil in Asia prior to construction of a four story building. The paper includes

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
258 Geo-Environment

listing of soil laboratory testing done, an analytical approach for estimating the
duration required for saturation, monitoring results during the soil wetting
program and a discussion about the results.

2 Soil parameters and collapse potential


Laboratory testing was carried out to identify important soil characteristics such
as unit weight, void ratio and liquid limit. In order to check for the soil
susceptibility to collapse under a change in its moisture content, natural dry unit
weight and corresponding liquid limit for many tested soil samples were studied
to check if the natural dry unit weight was less than limiting values determined
by Das, 1983. Results are given in Figure 1. Results indicated that most of the
soil samples tested is not susceptible to collapse at a change in moisture content.
However, some samples were slightly collapsible indicating the presence of
collapsible soil pockets in the soil matrix that may lead to differential settlements
affecting the proposed structure, if the soils were exposed to a significant change
in water content.
Soil parameters were reported as follows based on average values from the
analyzed samples:

Void Ratio: 0.83


Porosity: 0.47
Dry Unit Weight: 1.43 t/m3 (89 pcf)
In-Situ Unit Weight: 1.73 t/m3 (108 pcf)
In-Situ Moisture Content: 20% by weight
Liquid Limit: 29.5%

2.50

2.00
Natural Dry Unit Weight (t/m3)

1.50

Loessial Soil Likely to Collapse


1.00

0.50

0.00
0 5 10 15 20 25 30 35 40 45 50
Liquid Limit (%)

Figure 1: Estimate soil susceptibility to collapse.

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
Geo-Environment 259

3 Pre-collapse approach
As given in Figure 1, the identified soil parameters indicated that most soil
samples is not collapsible. However, the existence of collapsible soil pockets
embedded in the soil matrix was confirmed with a low potential to collapse, as
most of the soil characteristics are close to the border of collapsible soils. Those
soil pockets tend to give the site susceptibility for differential settlement that
may affect the proposed structure. In order to get rid of those collapsible soil
pockets and minimize the potential of having differential settlement on site, the
following measures were taken:
1. Regardless of the soil collapse potential, soil had to be wetted to induce
collapse, if a considerable potential exists. Wet the soil under the
proposed building foot print (plus a 2 meter buffer all the way around
the building) using a wetting program.
2. The wetting program would consist of a number of sand columns drilled
through the site and distributed in a grid form to ensure adequate
coverage of the area. A set of moisture sensors and settlement plates
would also be installed to monitor change in moisture content along
with any soil movement.
3. Replace the top 2.5 m of soil after collapse with a well compacted sub-
grade soil under the proposed foundation.
4. Use mat foundations with a reinforced concrete slab of 2 foot thickness
to minimize the effect of differential settlement on the proposed
structure.
The measures described above represent a combination of several methods
that are normally used for dealing with soil that is susceptible to collapse due to a
change in their water content. The challenge inherent in the project nature was
to design the wetting program and estimate how long it would take to saturate
the soil.
Ground Surface

Direction of
Water Flow H

L
∆H
∆L

Sand
Column

Figure 2: Schematic presentation of the site model for wetting.

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
260 Geo-Environment

4 Estimation of wetting time


In order to estimate the time needed to wet the soil, the soil matrix had to be
divided into finite elements, each of which has dimensions of ∆H in the vertical
direction and ∆L in the horizontal direction. A schematic diagram of the site
around a sand column is given in Figure 2. Water is assumed to initially fill the
sand column which has a hydraulic conductivity much higher than that of the
loessail soil surrounding the sand column. Water, then, flows outwards under
the water head in the sand column into the soil surrounding the sand column.
For a certain element at a depth (H) from the ground surface and a distance
(L) from the sand column, the hydraulic gradient driving water flow from the
sand column and along the soil matrix is directly proportional to the depth from
the ground surface and inversely proportional the distance from the sand column.
The rate of water flow through the element is given by Darcy’s law:

Q = K*I*A (1)
V = K*I (2)
I = H/L (3)
V = K*H/L (4)
Tn = Tn-1 + ∆L/2V (5)
Tn = Tn-1 + ∆L*L/(2*K*H) (6)

Where (n) designates, the element through which water is flowing and (n-1) is
the previous element through which water was flowing before entering to that
element. Equation (6) given above helps estimate the time needed to wet the soil
particles at a distance (L) from the sand column and at a depth (H) from the
ground surface, assuming a homogenous soil profile with a constant Hydraulic
Conductivity (K) across the site.
Based on equation (6), a finite element grid was established using a
spreadsheet to determine the time it takes to wet each element in the soil matrix.
The shallowest element and furthest from the sand column would take the
longest time and consequently determine the duration of the wetting program.
This element with its characteristics including Hydraulic conductivity of the soil,
depth and distance from the sand column represents the design criterion for the
wetting program. Obviously and based on equation (6) an element at a depth
close to zero from the ground surface would take a very long time to wet, unless
the entire site is soaked with water from the top (i.e. soak the ground surface) in
addition to the water it is receiving from the sand columns. The latter approach
is expensive and not practical to soak the site surface with water, as it leads to
dispensing considerable volumes of water simply to wet the top 2 meters of the
site. To avoid such a scenario it was determined to use an element that is 2
meters below the ground surface to design the wetting program for the adequate

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
Geo-Environment 261

duration to saturate the soil, and that the top 2 meters will have to be replaced
anyway with a compacted granular soil to provide an adequate sub-grade under
the mat foundation.
Sand columns are normally drilled to form a matrix of sand columns with a
spacing of 5 meters, i.e. the maximum length of water to flow away from any
sand column is 2.5 meters. Assuming an incremental distance ∆L of 0.2 meters
for the analysis, one can now use equation (6) to determine the duration required
to wet the site. The only unknown would be K due to its variance from one part
of the site to another for the same silty soil. Therefore, a sensitivity analysis was
done on the soil model provided in equation (6) and the value of K used within
the typical range for silty soils. Results are provided in Figure 3. Results
indicate high sensitivity of the model to K value and an exponential drop in time
needed to wet the soil as K decreases.
Sensitivity of Time Needed to Hydraulic Conductivity

250.0
Time Needed For Soil to Collapse (days)

200.0

150.0

100.0

50.0

0.0
0.00E+00 2.00E-05 4.00E-05 6.00E-05 8.00E-05 1.00E-04 1.20E-04
Hydraulic Conductivity (cm/s)

Figure 3: Sensitivity of wetting time to soil hydraulic conductivity,

5 Field monitoring data and discussion


A pilot study was done in the field to measure the site response to the wetting
program. Moisture sensors reporting volumetric water content (i.e. Volume of
water in a given volume of a soil matrix (Vw/Vt)) were installed at depths of 10
and 15 meters. A settlement plate was also placed on the ground surface to
record soil movement due to wetting. Readings from the moisture sensors and
the settlement plate are included in Figure 4.
Based on the soil characteristics, average volumetric water content of 0.47
was expected, as the volumetric water content will be numerically equal to the
average soil porosity at saturation. The volumetric water content was

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
262 Geo-Environment

determined to facilitate comparison to data reported by the moisture sensors.


The estimated value was fairly close to that recorded by the moisture sensors as
given in Figure 4. Note how the moisture sensors at both depths recorded a
sudden increase in value of the volumetric water content when the water front
reached the element, where the sensor was installed. The time lapsed prior to
this sudden increase was observed to be between three and four weeks,
consistent with that anticipated by the soil model of equation (6) and indicating
that the average hydraulic conductivity of the loessail soil varied between 6E-05
and 9E-05 cm/s.
The sensor at a depth of 15 meters recorded the arrival of the water front after
that of the sensor at a depth of 10 meters, although the water head in the sand
column is higher at 15 meters depth than at 10 meters. This is attributed to the
change in soil characteristics from top to bottom along the soil profile leading to
a lower hydraulic conductivity at the bottom than that at the top. This conclusion
was confirmed by looking at Figure 4, which shows that the volumetric water
content reached a lower asymptote at a depth of 15 meters than that at depth of
10 meters, implying that the porosity value was approximately 0.36 and 0.50 at
these two depths, respectively.
Settlement was observed using the readings from the settlement plate and did
not exceed a value of 2 cm during the period of wetting the soil, as the soil had a
relatively low potential to collapse.
Based on practice, collapse – prone soil was soaked at its maximum moisture
content for at least four weeks. The contractor should also account for the time
needed to dry the soil after those four weeks. This was considered to be the
appropriate time needed for water to flow downwards across the site to deeper
soil layers.

0.6 825.40

Moist. Sens. 10m


825.40
Moist. Sens. 15m
0.5

Ground Elevation
825.40
Ground Surface Elevation (m)

0.4
Vol. Water Content

825.40

0.3 825.39

825.39
0.2

825.39

0.1
825.39

0 825.39
11/5 11/10 11/15 11/20 11/25 11/30 12/5 12/10 12/15 12/20 12/25 12/30 1/4 1/9 1/14
DATE

Figure 4: Water content and ground surface elevation change versus time.

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X
Geo-Environment 263

6 Conclusion
This paper presents a case study of wetting loessail soil to avoid problems
associated with its susceptibility to collapse due to a change in water content.
Laboratory test results identifying different soil characteristics were reported
along with a method to evaluate the soil susceptibility to collapse.
The geotechnical approach for dealing with the collapsible soil included a
hybrid of common and well known techniques usually used to deal with loessail
soils. The paper presented an analytical approach for estimating the time needed
to wet the soil, with a sensitivity analysis of the time dependence on the
hydraulic conductivity of the soil. The paper also included graphical
presentation and analysis of data measured in the field, and a comparison
between analytical results and field measured data. The two results were
consistent and thus provided some validation of the analytical approach
presented. The paper includes a guide for engineers dealing with loessail soils
and a method of estimating time required to wet the soil backed up with
measured field data.

References
[1] Joseph E. Bowles, 1996, Foundation Analysis and Design, McGraw-Hill
Companies, Inc, New York, NY.
[2] Braja M. Das, 1984, Principles of Foundation Engineering, PWS Publishers,
Boston, MA.

Geo-Environment, J. F. Martin-Duque, C. A. Brebbia, A. E. Godfrey & J. R. Diaz de Teran (Editors)


© 2004 WIT Press, www.witpress.com, ISBN 1-85312-723-X

You might also like