Rutting Behaviour in Geosynthetic - Reinforced Pavements

Dawson A R, Little P H and Brown S F

University of Nottingham

ABSTRACT: Design methods for geosynthetic-reinforced unsurfaced pavements are shown to rely on the assumptions
of static monotonic failure for design and on aggregate non-influence in predicting rut depths. The results of a full-scale
haul-road trial in Scotland are used to illustrate the lack of validity of these assumptions and the importance of aggregate
compression and shear deformation in rut development.

1 1 INTRODUCTION
An important use of geosynthetics in road construction is
as a reinforcing element in unsurfaced haul roads
constructed on soft subgrades. In this application, the
geosynthetic is typically placed on the subgrade surface
and aggregate is then placed and compacted above it.
Many design methods have been proposed for this
situation (of which only Giroud & Noiray, 1981; Milligan
et al, 1989a,b; Sellmeijer, 1990 are mentioned here).
The earliest methods use a membrane analysis as the
basis for design in which vehicle loading causes surface
deflection of the pavement. This generates a downwards
deflection of the geosynthetic which is treated as a
membrane in tension between the undeflecting shoulders
of the pavement. The downward displaced and
elongated geosynthetic membrane exerts an upwards
force which acts to oppose the loading and thus
reinforcement is provided.
Later methods postulated a modification to the
subgrade stress regime which increases the load-carrying
ability of the soil (Milligan et al, 1989a,b) or a more
sophisticated membrane approach in which material
stiffness and strain compatibility become important
(Sellmeijer, 1990).
2 BEHAVIOUR OF UNREINFORCED SYSTEMS
In the unreinforced, unsurfaced pavement, failure almost
invariably takes place by excessive vertical displacement
of the pavement surface beneath vehicle wheels. For
efficient use the transient deflections must be small,
otherwise the vehicles are, effectively, travelling uphill
even when on the level (Douglas & Valsangkar, 1992).

This will be an important consideration for high volume

haul roads during construction of a major project in
which they are used as part of an earth/rock haulage
exercise. More usually it will be the permanent
deformation due to repeated loading which will be of
concern as large ruts increase friction (and thus tyre wear
and fuel inefficiency) and decrease manoeuv-rability.
This is the aspect discussed in this paper.
It will thus be evident that the monotonic strength of the
unsurfaced pavement will rarely, if ever, be of prime
interest.
Instead, the pavement engineer will be
concerned with the permanent deformations within the
aggregate and subgrade layers which result from
repeated transient load applications by the traffic.
Table 1 summarises repeated load triaxial testing of 17
different aggregates by Thom & Brown (1989) and
shows that there is no simple ranking relationship
between resilient deformation characteristics, permanent
deformation characteristics and monotonic strength for
different aggregates.
In an unsaturated aggregate, permanent deformation
will result from :
- material compaction (volumetric compression) leading
to a lowering of the surface beneath the wheel path.
- shear strains leading to displacement of material from
beneath the wheel path to the unloaded parts of the
pavement either side which will thus move upwards.
- material dilation (volumetric expansion) causing local
heave either side of the wheel path. This is particularly
associated with shear.
Similar mechanisms will operate in the subgrade,
although volumetric strains, in particular, will probably be
considerably smaller.

3 APPLICABILITY OF DESIGN METHODS

Peak Angle of Shearing

Resistance ()

54

51

3.1 Static Failure Analysis

3.3 Aggregate Thickness
The design methods for geosynthetic-reinforced,
unsurfaced pavements mentioned in the Introduction all
assume some form of static equilibrium in which one or
more elements is at its failure condition. This approach
has the advantage of simplicity but cannot easily form the
basis of a satisfactory surrogate for an analysis of
deformation due to repeated loading. Not only is the
poor relationship of strength to permanent deformation
resistance (Table 1) ignored, but also no allowance is
made for soil viscosity because of which, fully or nearly
saturated soils can tolerate transient stresses greater than
their monotonic failure strength (Brown & Dawson,
1992).
Table 1 Aggregate Performance in Triaxial Tests
Resilient Modulus :
9 12 1 2 5 4 3 6 8 7 10 11 17 15 13 14 16
Shear Strength :
9 1 8 7 17 10 4 2 3 16 12 5 6 15 11 14 13
Permanent Deformation Resistance :
10 5 8 2 3 4 6 1 17 9 7 15 12 16 11 14 13
Note : Each number indicates a particular aggregate.
Performance is ranked - 'Good' to left, 'Poor' to right.

3.2 Load-Deformation Relationship

A static design method predicts the ultimate condition.
As the actual failure criterion is a rut depth of a certain,
finite amount (defined by the type of pavement and
traffic), it is not clear how this rut and static failure might
be related. In practice, design methods usually assume
that static failure represents the condition at which the
allowable rut is developed by a single application of load,
Ps . For a load less than Ps , the pavement should take
more applications to achieve the same rut and the amount
is computed from an empirical relationship such as a
power law or that proposed by Hammitt (1970). The
relationships (except that due to De Groot et al, 1986) all
use information derived from trials on unreinforced
pavements.
Table 2 Summary of Aggregate Triaxial Testing
Confining Stress (kPa)
Poisson's Ratio
Resilient Modulus (MPa)

Crushed Diorite
20
45
0.33
360
450

Sand & Gravel

11
50
0.4
325
600

A further limitation of all the analytical methods discussed

above is the assumption that the aggregate thickness
remains constant, the surface rut being mirrored at the
subgrade surface.
It would be surprising if the
mechanisms described at the end of Section 2, above,
were inapplicable to reinforced, unsurfaced pavements.
4 TRIAL OBSERVATIONS
A recent experimental unsurfaced road trial at the
Science and Engineering Research Council soft clay test
site at Bothkennar, outside Edinburgh in Scotland
(Dawson & Little, 1990) presented an opportunity to
observe the rutting behaviour of 10 reinforced and 4
unreinforced pavement sections.
4.1 Construction
The sub-grade at the site comprised a firm brown very
silty clay of intermediate to high plasticity lying just above
the Casagrande 'A' Line. Further details of the soil
conditions can be found in Hight et al (1992). Undrained
strength of the subgrade varied with the season but was
typically 75-80kPa at periods of trafficking.
A variety of geosynthetics were used - non-woven
geotextiles (heat bonded and needle punched), woven
geotextile and biaxially oriented geogrid. These were
placed onto the subgrade which was first carefully
excavated to remove the relatively unquantifiable grass
and topsoil layers from the site. One of two aggregate
types was then carefully placed onto the geosynthetic and
compacted in three of four layers, no layer exceeding
150mm in thickness. The aggregates were well-graded,
crushed diorite and sand and gravel. Compaction was
by a 1000 kg/m twin drum vibratory roller. Dry densities
of between 2060 and 2240 kg/m3 were obtained for
both aggregates. The unreinforced pavements were
similarly constructed.
The aggregate was subjected to repeated load triaxial
testing in specimens 280mm in diameter and 560mm high
to determine the resilient modulus. Each specimen was
then taken to failure under monotonic loading to
determine shear strength. The results are given in Table
2. The relatively high resilient modulus of the sand and

gravel is in line with the results of Thom & Brown

(1989).
The sub-grade was levelled on a 0.5m grid over the
length and width of each pavement (20m x 5m) and the
final aggregate surface was similarly measured. Thus the
actual aggregate thickness was determined with some
accuracy (see Column 2 of Table 3 for mean
thicknesses). In pavements A to E the aggregate was
generally thicker than elsewhere as these pavements
were intended to have a longer life.
Geosynthetic samples were taken before construction
and after trafficking for laboratory index testing (Little,
1993).

Distance along pavement (m)

5
10
15

1000 passes

10
20
30
40
Depression (mm)

Figure 3

2115 passes

Vertical Depression of Pavement L

4.3 Results

The results of the trial are summarised in Table 3 and

Figs. 1 to 4. Fig. 1 gives typical results showing the
development of rutting (measured by careful surveying) at
4.2 Loading
two locations (J1 and J2) from a pavement with a woven
The pavement was subjected to 1115 passes of a vehicle
geotextile. It will be noted that :
comprising a rear axle weight of about 80kN and a
a) There was a rapid initial rut development. This
variable front axle of 30-56kN, followed by 1000 further
parallels triaxial testing on aggregates in which large
passes of the same vehicle with 126.35kN and 50.03kN
permanent deformation is observed in the first few
axles. (This represents 1191 and 6376 standard 80kN
repetitions of loading.
axles assuming a fourth power load equivalency
b) Rut development was broadly linear until onset of
relationship).
failure behaviour and was not proportional to the
logarithm of number of load applications (as usually
Rut depth (mm)
assumed). Rut depth development in the non-failing
Loading:
pavements
varied
between
0.003mm
and
150
Low High
Section J1
0.025mm/pass, either increasing with the higher loading
100
or remaining unchanged. Aggregate type did not appear
to have much influence on the rate of rut development.
50
Section J2
c) When excessive deformation occured it was rapid
and accelerating (J1).
0
0
400
800
1200
1600
2000 d) Two notionally similar constructions behaved very
Number of Passes
differently.
Figure 1
Rut Development, Pavement J
Fig. 2 shows the profile along the wheel path in
pavement F (geogrid). The longitudinal roughness
Distance along pavement (m)
increased with trafficking and there was significant
0
5
10
15
20
0
vertical depression of the aggregate surface. Fig. 3 gives
the longitudinal profile for pavement L (needle-punched
10
geotextile) which had a greater roughness but slower
20
1000 passes
vertical depression at the higher applied loading.
30
Table 3 summarises the rut depth, the vertical
displacement of the bottom of the rut relative to a datum
40
and the mean pavement aggregate and soil vertical strains
50
2115 passes
beneath the wheel path at the end of trafficking. A large
Depression (mm)
value for the ratio in the sixth column indicates that the
wheel path depression is much less than the rut depth
Figure 2
Vertical Depression of Pavement F
(hence indicating significant heave adjacent to the wheel
path.) Thus, pavement E (woven) has undergone

20

significant volumetric compression whereas pavement N

(geogrid) shows significant aggregate heave. The fifth
column in the Table indicates how representative the
cross-section results are of the whole pavement.
The vertical strains (obtained using inductive coils at the
base of the aggregate and at the top of the subgrade) at
the end of trafficking indicated that the relative
contribution
of
Table 3 Final Vertical Permanent Deformation Results

the heave in the subgrade. In this case the aggregate has

thinned by around 10mm and heaved by 25-50mm.
Overall the subgrade depression is responsible for 67100% of the surface depression (mean 79%), but this
falls to 11-91% (mean 51%) when rutting is considered.
Pavements L and N have the largest ruts as a proportion
of soil depression.
Level above
datum (mm)
450

After 2115 passes

Pavement surface
Pavement &
Deformation (mm)
Vertical Strain 3 ()400
Original
thick. (mm) Rut1 Vert 1 Mean2 Vert Rut/Vert Aggregate Subgrade350
A 430
48
50
43.2
0.96
18800
7600
B 495
49
40
40.1
1.23
19700
11800 100
Subgrade surface Original
After 2115 passes
C 527
28
33
32.0
0.85
15800
2900
50
D 580
28
26
21.8
1.08
16400
-2300
-0.8
-0.4
0
0.4
E 536
18
28
29.0
0.64
8800
4400
Distance from wheel path (m)
Fs 436
35
39
40.5
0.90
30800
18900
Figure 4
Original & Final Cross Section,
Hs 428
34
19
21.2
1.79
Is 350 failed failed
48000
21900Pavement L
J 262
failed failed
97300
28350
K 394
24
22
19.5
1.09
14800
5100
L 354
41
28
28.0
1.46
24400
9000 5 CONCLUSIONS
M 372
24
21
17.6
1.14
5200
6800
N 332
32
17
21.2
1.88
8500
15600
It is concluded that present analytical design methods are
Notes 1 Measured at 6 instrumented cross sections.
not capable of making a valid prediction of the load to a
2 Measured along wheel path throughout pavement
length.
defined rut depth - the normal design case. Aggregate
3 Measured at 2 instrumented cross sections.
behaviour is an important influence on rut build-up and is
S Pavements in which sand and gravel aggregate used.
not readily modelled solely by an ultimate shear strength
Vert = Vertical depression of surface relative to datum.
parameter. In particular, at the trial site on a relatively
Rut = Maximum vertical height difference across surface.

the aggregate to that of the soil to permanent deformation

varied but that it is often large (even though the strain is
measured at some depth below the wheel loading).
Large strains are associated with pavements which
experienced higher surface deformations except that the
strains are reduced when the aggregate is thicker. This
reduction may result because the strains are measured at
a non-constant depth from the loaded surface, suggesting
that high strains may still occur in the middle of a thick
aggregate layer and large internal aggregate deformation
could then be expected.
4.4 Exhumation
Eight of the pavements were carefully exhumed and the
final cross-sectional shapes and dimensions measured.
Fig. 4 gives a typical result (pavement L with a needlepunched non-woven geotextile). It shows the depression
of the aggregate, the lesser depression of the subgrade
under the wheel path and the heave due to shear of the
aggregate either side of the wheel-path which exceeds

firm subgrade, rutting has been shown, in broad terms :

- to involve significant variability due to increasing surface
roughness and other (unknown) factors. This introduces
uncertainty into prediction of rut depths.
- to be due, principally, to aggregate thinning due to
aggregate compression and/or shear, to aggregate
shoulder heave due to shear and to subgrade depression.
- to be (approximately) equally due to subgrade and
aggregate deformation.
- to develop linearly with load application, prior to onset
of rapid failure, after an initial 'set'.
ACKNOWLEDGEMENTS
The Authors wish to thank the UK Science &
Engineering Research Council, DuPont, Polyfelt and
Netlon Ltd for sponsoring the project at Bothkennar and
the assistance of the University of Nottingham laboratory
staff.

REFERENCES
Brown, S.F. & Dawson, A.R. (1992) Two-stage
approach to asphalt pavement design, Proc. 7th Int.
Conf. Asphalt Pavements, Nottingham, June 1992,
1:16-34.
Dawson, A.R. & Little, P.H. (1990) Reinforced haulroads:- trials at Bothkennar, Scotland, Proc. 4th Int.
Conf. Geotext., Geomem. & Related Prods., The
Hague, 1:250.
De Groot, M., Janse, E., Maagdenberg, T.A.C. & Van
Den Berg, C. (1986) Design method and guidelines for
geotextile application in road construction, Proc. 3rd
Int. Conf. Geotext., Vienna, 2:741-746.
Douglas, R.A. & Valsangkar, A.J. (1992) Unpaved geosynthetic-built resource access: stiffness rather than rut
depth as the key design criterion, Geotext. &
Geomem., 11:45-60.
Giroud, J.P. & Noiray, L. (1981) Geotextile-reinforced
unpaved road design, J. Geotech. Engrg., ASCE,
107:123-54.
Hammitt, G.M. (1970) Thickness requirements for
unsurfaced roads and airfields bare base support,
US Army Engineer Waterways Experimental Station,
Vicksberg, Tech. Report S-70-5.
Hight, D.W., Bond, A.J & Legge, J.D. (1992)
Characterization
of
the
Bothkennar
clay,
Gotechnique, 42:289-302.
Little, P.H. (1993) The design of unsurfaced roads
using geosynthetics, PhD thesis, Dept. Civil Engrg.,
Univ. Nottingham.
Milligan, G.W.E., Jewel, R.A., Houlsby, G.T. & Burd,
H.J. (1989a,b) A new approach to the design of
unpaved roads, Part I, Ground Engrg., April, 25-29
& Part II Nov., 37-42.
Sellmeijer, J.B. (1990) Design of geotextile reinforced
paved roads and parking areas, Proc. 4th Int. Conf.
Geotext., Geomem. & Related Prods., The Hague,
1:177-182.
Thom, N.H. & Brown, S.F. (1989) The Mechanical
properties of unbound aggregates from various sources,
Unbound Aggregates in Roads, (ed. Jones, R.H. &
Dawson, A.R.), Butterworths, London, 130-142.