Proceedings: Second International Conference on Case Histories in Geotechnical Engineering, June 1-5, 1988, St. Louis, Mo.

, Paper No. 3.56

Design and Construction of Geocell Mattress as Embankment Foundation

Alistair J.T. Gilchrist
Area Civil Engineer, Netlon Limited, United Kingdom

SYNOPSIS: This paper describes the use of a high strength Geocell mattress as a new solution to
the problem of the construction of embankments on soft foundations.
The paper deals in detail with the design of the Geocell mattress to support and distribute the
imposed loading onto soft foundation. It considers the plastic condition of the soft clay to
determine the factor of safety against foundation failure.
Two case studies will be examined to illustrate the construction techniques. Monitoring of
settlement during construction of the Geocell and during embankment filling is compared with
predicted values.
This ther,efore pushed the realignment to the
north into an area of marsh which once formed
the southern part of the Dougalston Loch which
had been drained many years before. This area
was also very heavily wooded and to achieve the
required vertical alignment and embankment, of
some 4.5m in height, was required to be
constructed across this marshland.

INTRODUCTION
The A807 Auchenhowie Road runs to the north of
the City of Glasgow, Strathclyde, Scotland and
follows a very torturous route with some very
narrow right angled bends.
The general poor alignment and the sub standard
condition of the road has contributed to a high
accident record, thus the local Highway
Authority embarked on a capital works scheme to
re-align the road along a new route (Fig 1).

SITE INVESTIGATION

To the south of the exisiting road runs four

large diameter water pipes which supply a large
proportion of the population of Central
Scotland with drinking water. Any disturbance
of these pipes would have incurred a very heavy
cost penalty (approximately 1.0 million).

In the area of the swamp (Fig 2) bore holes

103-105 penetrated between 31.1m and 3.gsm of
Very soft deposits generally comprising of 1.3m
to 1.5m of very soft brown organic silty clay
interbanded with black peat and occasional
lenses of fine sand and medium gravel. This
was underlain by between 0.9gm and 2.45m of
very soft brown laminated silty clay
interbanded with frequent partings of clayey
silt.
The bore holes were terminated at approximately
4.0m depth in very weak highly weathered
becoming weak friable grey mudstone.
Surprisingly the bore holes were generally dry
with water only becoming apparent at the clay
mudstone interface.
Hand bores number h.b. 101 to h.b. 104 again
penetrated the very soft deposits to a depth of
about 1.0m where they were terminated. Hand
bores h.b. 105 and h.b. 106 went deeper but
still remained within the soft laminated silty
clay.

scale 1-10.000

Figure 1

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

627

Figure 2
Bore hole number 101 was sunk into the bank of
the loch and penetrated 2.2m of firm to stiff
friable reddish brown very sandy clay underlain
by 1.35m of dense to very dense, very clayey
sand and gravel over decomposed mudstone.

In bore holes 103, 104 and 105, moisture

contents as high as 123% were recorded but
average figures were in the order of 40% with
average liquid limits of 50% and plasticity
index of 24% (Fig 3).

Three bore holes sunk in the location of bore

holes number 102 all terminated on obstructions.

To supplement the available information,

laboratory vane shear tests were carried out on
undisturbed samples from bore holes numbers
103, 104 and 105. These yielded undisturbed
shear strengths of between 11 kN/sq m to 30
kN/sq m with an average value of 17.0 kN/sq m.
The respective remoulded values varied from 3
kN/sq m to 8 kN/sq m with an average value of
5.0 kN/sq m.

Laboratory Results
Due t6 the poor nature of the underlying soils
many undisturbed samples proved unsuitable for
testing.
Where triaxial testing was possible, values of
apparent cohesion of between 12 and 19 kN/sq m
were recorded.

The soft organic clay silt had an average

organic content of 11.0%.

105
.,..,,.o.:D.

0.80-M0~5)
1.10 J1

0.70 :B1

1.0.5- 1. (2)

1-4,-1.,cl TJ1

"'
"'

1.80- 2.25 V1

....

3-20 ... ,_,, V't

4-15 :D1

Ii
~,,1

..

4..25-455 (!10)~1

I
Figure 3

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

628

(b) Construction of a high strength Geocell

mattress directly onto the soft layer to
form a foundation to the embankment.

SUMMARY OF GROUND CONDITIONS

From an inspection of the bore hole logs and
from the available laboratory results obtained
from site investigations the underlying strata
beneath the proposed embankment is summarised
as fallows:

This method of construction would require

no significant increase in site traffic.
It would also not be weather dependent and
require no specialised site operations.

0.4 m thick

(a) Peaty topsoil

(b) Loose brown organic
clayey silt and peat
(c) Very soft brown
laminated silty clay
(d) Weak weathered
friable grey mudstone

A straight cost comparison was made between

these two options:

1.50 m thick

Cost Comparisons

2.50 m thick

Excavation and replacement

Item

For design purposes the soft silty clay and

peat area was taken to an average depth of 4.0m
with an average undisturbed shear strength (Cu)
of 15.0 kN/sq m and = 0 degrees.

Strip top soil

Excavate U/s
( d = 3.5m)
Disposal off site
Import suitable
f i ll

EMBANKMENT FOUNDATIONS
Substituting these soil strengths into
Terzaghi's equation:

To ta 1
cost

Quantity
cu m

Rate

1840

1. 00

1840.00

21460
21460

0.50
3.00

10730.00
64380.00

21460

3.50

75110.00

TOTAL =152,060.00
Construction of Geocell mattress

qu

= CU X

Nc +

taking D = 0, Nq
following value

~ X

0, Nc

Nq +

(~ X

B/2)

X N~

(x + 2), gave the

Ultimate bearing capacity- surface material

qu = 77.1 kN/sq m

Quantity

Provide, assemble
and infill Geocell
Free draining
infill

6132 sq m 11.00

67452.00

6132 cum

49056.00

Rate

Total
cost

8.00

TOTAL = 116,508.00
Less cost of
10.m of
embankment

Assuming a density of the imported material

of 19.0 kN/Cu m and the height of the
proposed embankment of 4.5m then:

6132 cu m 2.00

Therefore effective
cost of Geocell

Imposed bearing pressure = 4.5 x 19.0

85.5 kN/sq m

12264.00
104,244.00

This represents an estimated saving of

approximately 31% or 48,000.00

Therefore factor of safety against

foundation failure= 0.9

This saving, by using the Geocell mattress,

does not take into account the reduction in the
environmental nuisance by cutting down off and
on site traffic. It was therefore decided to
proceed with the detailed design of the
mattress.

DESIGN CONSIDERATIONS
A number of design alternatives were looked at
for the embankment, although only two are
considered further by the author.
(a) Removal of the soft 1 ayer of material and
replacing with suitable rock fill.
This method would have involved a
considerable increase in traffic on and off
the site. No rock cut was locally
available on the site so all rock fill
would have to be imported.

DESIGN OF FOUNDATION MATTRESS

Netlon Limited were approached in October 1984
by Strathclyde Regional Council Roads
Department for advice on a suitable method of
constructing an embankment over this soft
material. At this time only approximate
parameters for the soft clay were available and
a preliminary design was prepared utilising a
high strength Geocell mattress which had been
successfully used on similar sites.

Owing to the high moisture content of the

soft silts excavation and maintaining a dry
site would have been difficult and weather
dependent as would the removal and disposal
of this highly fluid material off site.
Due to the local environment this
construction method would not be readily
acceptable.

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

Item

The main criteria for design was for an overall

factor of safety of 1.5 to be achieved against
foundation failure.

629

(1) Sufficient tensile strength to ensure that

the full Cu value is mobilised on the base.

DESIGN PHILOSOPHY
A Geocell mattress is a very stiff rigid
foundation mattress designed to support an
embankment over very soft ground. It is
generally installed at existing ground level
and with no soil strip (Fig 4).

(2) The mattress is rigid to ensure an even

distribution of load onto the foundation
material. This is provided by the high
tensile strength of the Tensar SR2
diaphragms forming a triangular cellular
construction.

(3) The base of the mattress is rough to

provide high frictional contact with the
underlying soils. The base material is
Tensar SS2 which allows partial penetration
of the infill material through the
apertures, creating a rough underside to
the mattress.
The preliminary design was carried out prior to
the site investigation being completed and was
based on a depth of soft silty clay layer of
2.5m with an average Cu = 30kN/sq m and ~ = 5
degrees.
From the preliminary design the factor of
safety against foundation failure was 3.51 and
against failure of the mattress of 1.71.
The very high factor of safety reflected the
under utilisation of the load carrying capacity
of the Geocell mattress but compared with the
alternative options available this still proved
to be economical and the design was accepted by
the Roads Department.

Figure 4
The mattress is constructed from high strength
Tensar geogrids to form a cellular construction.

However, when more accurate soil parameters

became available, a further design assessment
was carried out.

This rigid foundation mattress is assumed to

alter the direction of the normal slip circle
failure plane by forcing it to pass vertically
through the mattress.

The design parameters used in this final design

were:

This has the effect of forcing the slip plane

deeper into the underlying stiffer weathered
mudstone.

Depth of soil silty clay

1 ayer
Average strength of layer
Height of embankment
Width of embankment

Before this weathered mudstone can fail in

shear, a plastic failure would be initiated
within the softer silty clay beneath the
embankment. The plastic failure condition of
the soft silty clay is therefore considered in
the design.

Cu

4.0 m
15.0 kN/sq m
4.2 m
27.4 m

From this design the m1n1mumn factor of safety

against foundation was 1.58 and against
mattress failure was 1.93.

The foundation mattress is assumed to display

the following properties:

The mattress required to be edge stiffened over

the outer 6.0m. This was achieved by using 0.5m
cells along the outer edges and conventional
l.Om cells within the remainder (Fig 5).

7.3mwide
carriageway
~-------------------

--=---/--- ------------------Embankment
toe
Figure 5

l.Omcoll,

0.5mcells

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

630

It was also shown that where the m1n1mum value

of Cu = 8.0 kN/sq m, the minimum factor of
safety remained greater than unity.
The Geocell now showed to be a very economical
solution and was adopted for the contract.
The design procedure for the Geocell mattress
is outlined in detail in Appendix 1.

CONSTRUCTION
The Geocell foundation mattress was constructed
insitu during January and February 1986. The
dense vegetation of shrubs and trees was
removed in order to gain access to the site.
An island across the line of the mattress was
partially levelled. Other than this no site
preparation was required. The site was
waterlogged and under standing water during the
period of construction.

Figure 7
The infill material is not compacted and slight
overfilling compensates for any slight
settlement and also protects the top of the
diaphragms from damage.

4.0m wide lengths of Tensar 552 were rolled out

along the line of the road to form the base of
the mattress. These strips were lapped by
500mm. The main transverse diaphragms were
Tensar SR2 and were installed at l.Om centres
across the mattress. The Tensar SR2 was
stitched to the base material using HOPE braid
(Fig 6).

As the filled Geocell progresses the Contractor

utilises it as a construction platform for
infilling plant and the delivery of infill
material. At no time does the Contractor's
plant run on soft ground so delays from bogging
down are eliminated (Fig 8).

Figure 6
Figure 8

After tensioning of the main diaphragms, the

intermediate diaphragms, again Tensar SR2, were
installed and fixed to the main diaphragms
using steel bodkins. This forms l.Om
triangular cells. Over the areas where the
edge stiffening was required, 0.5m triangular
cells were constructed using short intermediate
transverse diaphragms (Fig 7).

This work was a new experience for the

Contractor, Raynesway Construction Services
Limited.
After the initial learning curve, outputs of
completed, filled mattresses in excess of 350
sq m per shift were achieved. Weather
c6nditions during the initial construction
period were very poor.

The cellular mattress was then infilled with

imported free draining granular material. The
filling procedure ensures that no cell is
filled to more than half the height of the
adjacent cell. This prevents excessive bulging
of the cell walls (Ref 6).

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

The total time for construction was six and a

half weeks and the actual installation cost was
some 111,000.00 being 5,500.00 less than the
original estimate.

~1

The embankment was then constructed directly

onto the Geocell mattress using conventional
earthmoving plant. The embankment was raised
to its full height rapidly and was completed in
about 7 days (Fig 9).

3. The Contractor had a dry working platform on

which to carry out Geocell filling
operations and any subsequent drainage works
if these were required. Wick drains can be
installed through the mattress with no
impairment. The mattress enables the
Contractor to work continuously through bad
weather with delivery vehicles running on
the infilled Geocell.
4. The infill material to the Geocell provides
a massive drainage blanket which aids
consolidation of the completed embankment
and reduces consolidation times.
5. The Geocell mattress showed considerable
cost savings over alternative methods of
constructuion.
6. No differential settlement had been evident

after completion of the embankment.

Settlement has been less than that protected
for an unreinforced embankment. The Geocell
mattress effectively spreads the load more
evenly along the embankment foundation.

Figure 9

SETTLEMENT
Using the information from the site
investigation, the overall settlement of the
completed embankment was expected to be in the
order of 175mm over 1 year. The expected
settlement during the 3 month construction
period was expected to be 140mm.
Tubular settlement pins, welded to .metal base
plates, were placed onto the top of the
completed Geocell base grid and supported by
the infill material of the embankment. They
were levelled every week as infilling and
embankment construction progressed and the
settlement recorded and plotted on a graph. An
average initial settlement of approximately
90mm was recorded after the embankment was
complete and before carriageway construction
was started.
The settlement was noted to be even across the
width of the embankment. No differential
settlement has been monitored along the length
of the completed road.

CONCLUSIONS
1. By using the Geocell mattress there was no
need to remove large quantities of
unsuitable material from the site with the
resulting problems of handling and disposal.
There was also no appreciable increase in
the off site traffic and environmental
nuisance.
2. The factor of safety against foundation
failure was significantly increased and
embankment constuction continued rapidly
without having to incorporate rest periods
into the construction programme to allow
excess pore water pressures to dissipate.

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

APPENDIX 1
In this design, reference is made to
"Engineering Plasticity" by W Johnson and P B
Mellor, (Chapter 12.5: pages 392 to 401).
This considers the plastic failure of material
compressed between two rough, rigid parallel
platens being brought together. The platen
widths exceed the soft material layer thickness
(Fig 10).
The Geocell mattress represents the upper
platen with the weathered mudstone representing
the lower.

Figure 10

DESIGN PARAMETERS
Weight of embankment
Embankment width
Depth of soft material
Strength of soft layer

4.2m
27.4m
d
4.0m
Cu = 15.0 kN/sq m
1

The pressure diagram for half the embankment is

evaluated. A value equal to twice the cohesion
is added on to take into acount the effect of
passive pressure beyond the toe and neglecting
the effect of upthrust within 0.45 x 2h of the
toe.

Load distribution on the base is obtained from

the equations given in the extract from
"Engineering Plasticity" and from the 15 degree
flow net (Fig 11) .

Cllt offload9271cN/m

..

~,:

t ._....ar--..

12

d Dlplllareolt-.

14
18
11

cu-81111........

PNamlll..-..anft1191._
..J-2.8m--tio---- 5.8m---+---S.Om--l

Figure 12

Figure 11
Calculation of the average base pressure over
the rigid zone:AX

1.966

1. 707

1.259

X = 4.932

yrad.

0.654

0.393

0.131

h = 0.4m

yxAX

1.286

0.671

0.165

%YXAX 2.122

p = 6. 0 Cu + 2.0 Cu = 8. 0 Cu .

from eqn . 12.11

!u 2 [ 2.%

Y 2X AX

+ h

J !;
+

= 2 [c2 x 2.1221 + 4.oJ + 8

2 X 4.932

4.57 CU

190.40

5.8 (3.43/2 + 4.57_) cu = 545.20

5.0

9.67 .Cu .

W((3.2x7.15)+(1.0x13.6)+(0.5x3.2x4.8))
X 20.0 = 883.2
Surcharge 3.65 x 12.0 = 43.8
Therefore Total Imposed Load= 927.0 kN/m
Therefore factor of safety against foundation
failure:F.o.S = 1.58
Consider load within the diaphragms of the
Geocell.
For the factor of safety reduced to 1.0. at the
centre of the embankment, the equivalent
apparent cohesion for the soft layer would
become:15.0 1 1.50 = 9.49 kN/sq m

9.67 cu

Therefore, adopting an average cohesion of 15.0

kN/sq m for the soft layer, the load to failure
for the half embankment is (Fig 12):2; 8

Imposed loading for half the embankment is:-

= 725.25

Therefore load to failure = 1460.85 kN/m

From a Mohr construction it can be shown that

the horizontal load to be resisted by the
Geocell mattress is:Tf = Cu

m-J
Based on the infill material properties 9 30
degrees. Therefore for a factor of safety of
1.0 the horizontal load is:T1 0 = 9.49= 18.98 kN/m run
~
Sin 30

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

This is the minimum condition and applies over

the worst centre portion of the embankment.
At the edge of the embankment assume the worst
condition where the factor of safety drops to
2. 0.

i.e. Cu beneath embankment= 18.98 kN/sq m

The embankment load/foundation diagram is shown
in Fig 12. The cut off line at support value
of 927 kN/m shows . the extent of the plastic
zone.
Over this zone the horizontal force to be
resisted by the mattress is:

Johnson, W. and Mellor, P.B., (VNR),

Engineering Plasticity

The diaphragms to the Geocell and Tensar SR2

geogrid.

= 21.48

kN/m

(Ref 5)

Therefore resistance of l.Om cells is:T = 21.48 +

rr

Thanks are . also due to Netlon Limited

andRaynesway Construction Services Limited
fortheir assistance.
BIBLIOGRAPHY

T2 0 = 18.98
= 37.96 kN/m run

Sin 30

Working load in the grid

ACKNOWLEDGEMENTS
The author wishes to thank Mr W S McAlonan, MSc
DipTp CEng, Director of Roads,
StrathclydeRegional Council and Mr D L Hynd,
CEng MICE,Divisional Engineer, Dumbarton, and
to othermembers of the Roads Department for
their help. The author would point out that the
views and opinions expressed are his own and
notnecessarily those of the Regional Council.

= 36.76 kN/m run

and resistance of O.Sm cells is:T = 2 x 36.67 = 73.34 kN/m run

Therefore factor of safety of Geocell against
failure= 1.93
Therefore use O. Sm cells within 6.0m of the
edge of the mattress and l.Om in the remainder.
Where the embankment height is less than 3.0m
high, l.Om cells were used throughout.

Second International Conference on Case Histories in Geotechnical Engineering

Missouri University of Science and Technology
http://ICCHGE1984-2013.mst.edu

Jones, C.J.F.P., Earth Reinforcement and Soil

Structures
Netlon Limited, Design suggestions for the use
of a High Strength Geocell mattress in the
Construction of the Auchenhowie Embankment
Netlon Limited, Tensar Geocell Mattress
Netlon Limited, Test Methods and Physical
Properties of Tensar Geogrids
Robertson, J. and Gilchrist, A.J.T., Design and
Construction of a Reinforced Embankment
Across ~oft Lake Bed Deposits
Smith, G.N., Elements of Soil Mechanics for
Civil and Mining Engineers