Geotextiles and Geomembranes 9 (1990) 261-267

Anchorage and Modulus in Geotextile-Reinforced

Unpaved Roads

Robert A. Douglas
Department of Forest Engineering, University of New Brunswick, Fredericton, New
Brunswick, Canada E3B 5A3

(Received 3 September 1989; accepted 16 October 1989)

ABSTRACT

In the light of recent comments, a programme of research into geotextile-

built model unpaved roads is summarized, and the question of reinforce-
ment by the provision of a geotextile in the road cross-section is discussed.
It is found that the improvement in the road section response to loading,
compared with the performance of the subgrade itself, is not dramatic and
reasons for this are given, referring to experiment and theory.

INTRODUCTION

In a recent Technical Note, B o u r d e a u et al. 1 c o m m e n t e d on a p r o g r a m m e

of research on geotextiles in unpaved resource road construction, reported
by Douglas et al. 2 and Douglas and Kelly. 3 As it was our intention to
perform well-controlled, large-scale model tests which would serve as a
foundation for testing our own theories and those of others, we are
grateful for the interest shown by Bourdeau et al. We now wish to point out
some misinterpretations we feel they have made.
A major criticism we had of the work reported by m a n y others was that
often researchers first assumed a reinforcement role to be played by the
geotextile, and then built an analysis on this assumption. Our experiments
were designed to test this original assumption about reinforcement,
starting from an unbiased viewpoint. Two hypotheses were proposed:
(1) If the geotextile does indeed play a reinforcement role, then an
increase in geotextile modulus will result in an increase in the road
stiffness.
261
Geotextiles and Geomembranes 0266-1144/90/$03.50 ¢~) 1990 Elsevier Science Publishers
Ltd, England. Printed in Great Britain
262 Robert A. Douglas

(2) If the geotextile plays a reinforcement role, then anchoring the

geotextile will increase the road stiffness over its value without
geotextile anchorage.
The road stiffness was defined as the average pressure p, applied by a long
narrow beam of width B, in these plane strain tests, divided by the average
vertical displacement, 3, of the beam. Thus road stiffness kr = p/& Loads
were applied at the centreline to model road cross-sections built on
reconstituted sphagnum peat subgrades, in a steel test bin 2.5 m square.
Beam width B = 250mm and gravel thickness H = 170 mm, giving
H / B = 0.67.
It was convenient in the experiments to anchor the fabric edges to the
test bin sides, at a distance x/B = 4.8. In this test programme, no attempt
was made to measure loads at the anchored fabric edge, nor was fabric
edge displacement measured during loading, as Bourdeau et al. ~ have
pointed out. In later test programmes, this improvement may be made.

T H E EFFECTS OF G E O T E X T I L E M O D U L U S

The first hypothesis was checked by measuring the road stiffness in load
tests carried out on model road sections built on a woven geotextile, a
nonwoven geotextile, and a weak polyethylene film. Fabric characteristics
are given in Table 1.
It can be seen from the table that the woven material was of medium
strength and stiffness, the nonwoven material was of low strength and
stiffness, and the polyethylene film was by comparison very weak and
extensible.

TABLEI
Average Cross-MachineCharacteristicsofFabrics~

Tensile Failure Strain at

Width modulus load failure
Fabric (mm) (kN m -z) (kN m -z) (%)

Woven 100 123 27 23

200 119 27 24
Nonwoven 100 26 14 56
200 27 14 53
Film 100 6.9 0-83 19
200 12 0.85 19

aSee Douglas et al. 2 and Douglas and Kelly3 for complete results.
 Geotextile-reinforced unpaved roads 263

Road stiffnesses were reported in terms of a nondimensional stiffness:

K* = kr/kp (1)
where kr is the stiffness of the road section (in kN m -3) and kp is the
stiffness of the peat subgrade alone (in kN m-3). The stiffnesses for plane
strain tests on the peat subgrade alone were found to be 170 kN m -3 for the
unanchored series of tests, and 145 kN m -3 for the anchored series.
Non-dimensional road stiffnesses are reported in Table 2.

TABLE 2
Non-dimensional Road Section Stiffnesses

Unanchored Anchored
non-dimensional non -dimensional
stiffness stiffness
Fabric i(* = kr/k r * = kr/k

Woven 1.21 1.45

Nonwoven 1.21 1-52
Film 1.09 1.31

The non-dimensional stiffnesses indicate how much of an improvement

the fabric-built road sections represent compared with the subgrade itself.
Although it is true that the improvement in the geotextile-built roads is
about double that of the road built on the weak film (21% compared with
9%), one must not lose sight of the fact that the real issue is whether the
improvement in the road performance with the addition of a geotextile
compared with the performance of the subgrade itself makes the provision
of the geotextile worth while. An improvement of only 21% in stiffness--
or put another way, a decrease in rut depth to 80% of what it would have
been for the same pressure on the subgrade itself (1/1.21 ~ 0.80)--would
probably not be worth the extra expense of building the road with a
geotextile. This argument holds if reinforcement is the object of placing a
geotextile in t h e road cross-section: other reasons (such as separation)
may still lead to the inclusion of the geotextile.
The reason the geotextile did not dramatically improve the model road
performance can be illustrated. Unpublished measurements 4 indicated
that at an applied pressure of 20 kN m -2 and a vertical displacement of
about 0-1 m (0-4B), at most the woven geotextile had distorted to an angle
of ca. 13 ° to horizontal at its steepest-sloped location, close to the loaded
zone (Fig. 1). Thus, the geotextile could contribute only about 45% (i.e.
264 Robert A. Douglas

2@ kN/m ~
.

Fig. 1. Schematic cross-section of model road section distorted under plane strain loading.

2 × sin 13°) of whatever tension it had developed, to the load-carrying

capacity of the road cross-section. Calculations showed 4 that at this toad
level, the woven geotextile had only strained a little over 3%, indicating
that the tension in the geotextile was small, well below the tensile strength.
With geotextiles placed in horizontal layers in road cross-sections, they
are not arranged to contribute much to the load-carrying capacity of the
road until such deep ruts have developed that the road is impassable.

T H E EFFECT OF A N C H O R A G E

Developing high tension in the geotextile to derive the greatest benefit

from its small vertical-component contribution to load resistance implies
the necessity of high anchorage forces. This impinges on the second
hypothesis of the test programme.
Table 2 indicates that when the geotextile edge was clamped a distance
of 4.8B from the loaded strip centreline, there was an improvement in the
response of all three fabric-built roads. Again, however, it is pointed out
that still there is no dramatic improvement over the performance of the
subgrade alone.
This anchorage arrangement represents a highly idealized situation, and
not one that is particularly practicable for real road construction. Usually,
anchorage is achieved through whatever friction and/or adhesion can be
developed on the two faces of the geotextile. Bourdeau et al. 1 agreed that
any contribution to anchorage made on the bottom (i.e. subgrade) face of
the geotextile can be neglected in the case of these model road sections and
therefore focused on the upper (i.e. the fill side) face. An equation (their
eqn (1)) is given, which predicts the anchorage force to be developed.
It appears that the equation is in error. It includes what are considered
to be the contributions to anchorage found in two separate locations:
 Geotextile-reinforced unpaved roads 265

under the loaded strip, and outside the influence of the loaded zone. We
submit that the error is that these two interface stresses act in opposite
directions. Figure 1 of Bourdeau et al.l shows the interface stress labelled
F~ acting in the wrong direction in the vicinity of the loaded strip.
Outside the loaded zone, the interface stress is directed toward the outer
edge of the geotextile. The downward displacement of the fabric under the
loaded zone drags the edges toward the centre. This could set up a relative
displacement between the gravel fill and the geotextile, although common
observation and experimental results 5 indicate that the relative displace-
ment is negligible.
Beneath the loaded zone, the direction of any potential interface stress
is the opposite. The gravel fill pulls the geotextile downward and the
interface stress must therefore be directed toward the centre. Outside the
loaded zone, the geotextile pulls the gravel fill, but inside the loaded zone,
the gravel pulls the geotextile.
By totalling the magnitudes of two components of interface stress acting
in opposite directions to arrive at the anchorage available, Bourdeau et
al.'s eqn (1) is in error.
The action of the interface stress on the geotextile results in a variation
in tension across the width of the geotextile. Therefore, an examination of
measured geotextile tensions should provide evidence of the variation in
interface stress. Jarrett 6 reported tensions measured by load cells mounted
on a geogrid strip. The measured loads given in his figs 3 and 4 are
replotted in Fig. 2.
Given that the curve of geotextile tension is smooth and continuous, and
is symmetrical about the centreline, its slope must be zero at the
centreline. The slope of the tension plot (Fig. 2), following it in towards
the centreline from the outside edge at 1.85 m, reaches a maximum at a
distance of around 0.4 m from the centreline, and then decreases to zero at
the centreline.
The plots, when followed from the outside edge toward the centre, are
in effect an integration of the interface stress applied to the geotextile. If
the slope is 'positive' (as the curves are drawn in Fig. 2, 'positive' slope is
upwards to the left), the interface stress is directed towards the outside
edge of the geotextile. However, if the slope is negative (downwards to the
left) the interface stress must act in the opposite direction, towards the
centre. Thus for Jarrett's particular test arrangements, 6 the interface
stress was greatest at about 0.4m (ca. 2B), but closer to the centreline it
decreased rapidly and was essentially zero within 0.1-0.2 m (0.5B-1.0B) of
the centreline, i.e. within the loaded zone.
The implication is that the anchorage available to the geotextile is that
which occurs outside the loaded zone. The potential interface stresses
266 Robert A. Douglas

IE.0

10.0

Z
.J£
8.e

O
.~[ 6.0
O
...J

r-~
4.0
t..9
O
b.I
tO
2.0

i I I I I I I I I !
O. 0.4 0,8 1.2 1.6 2.Q

DISTANCE FROM CENTRELINE [m]

Fig. 2. Geogrid loads for unpaved model road sections under plane strain loading. (After
Jarrett. 6)

occurring inside the loaded zone will not be anchorage stresses, but rather,
stresses which 'grip' the geotextile and pull it downward and inward. In the
case of Jarrett's experiments,6 these stresses were negligible, indicating no
relative displacement between gravel and geotextile in this zone. Because
of the thin fill depth typical of these roads, the anchorage available outside
the loaded zone is minimal, and in any event, as shown by experiment, 5 is
unlikely to arise at all because there is usually no relative displacement
between the gravel fill and the geotextile here either.

CONCLUSIONS

By beginning from a point of view where the existence of reinforcement is

not first assumed but is rather examined through the vehicle of the two
hypotheses in our experimental research programme, it is concluded that
the provision of a geotextile does not improve the load response of the
unpaved model road sections dramatically over that of the subgrade itself.
There may be other considerations, however, such as the need for
separation, that suggest the necessity of a geotextile in the pavement.
 Geotextile-reinforced unpaved roads 267

The research under discussion has focused on monotonic loading of

single strips. However, the real problem facing designers of unpaved
resource roads is one of cycled, closely spaced loads. Research using a
newly acquired programmable servo-hydraulic load system to test large-
scale model pavements is currently under way at the Department of Forest
Engineering, University of New Brunswick.
It is indeed the case that the 'mechanism of soil-fabric interaction as
applied to geotextile-reinforced unpaved roads is still not fully
understood'.l We are grateful that Bourdeau, Chapuis and Holtz have
provided an opportunity to discuss the mechanism further, and appreciate
their having commented.

ACKNOWLEDGEMENT

The author gratefully acknowledges the Natural Sciences and Engineering

Research Council of Canada (NSERC) for funding this research pro-
gramme.

REFERENCES

1. Bourdeau, P. L., Chapuis, J. & Holtz, R. D. Effect of anchorage and modulus

in geotextile-reinforced unpaved roads. Geotextiles and Geomembranes, 7(3)
(1988) 221-30.
2. Douglas, R. A., Bessy, B. B. & Small, R. P. The use of geotextiles in forest
road construction. Proceedings, 2nd Canadian Symposium on Geotextiles and
Geomembranes, Edmonton, Alberta, 23-24 September 1985. Canadian
Geotechnical Society, Edmonton, 1985, pp. 89-96.
3. Douglas, R. A. & Kelly, M. A. Geotextile 'reinforced' unpaved logging roads:
the effect of anchorage. Geotextiles and Geomembranes, 4 (1986) 93-106.
4. Addo, K. O. Geotextiles in unpaved road structures on peat subgrades.
MScEng thesis, Department of Civil Engineering, University of New Bruns-
wick, Fredericton, 1986, p. 38.
5. Jarrett, P. M. & Bathurst, R. J. 1985. Frictional development at a gravel-
geosynthetic-peat interface. Proceedings, 2nd Canadian Symposium on
Geotextiles and Geomembranes, Edmonton, Alberta, 23-24 September 1985.
Canadian Geotechnical Society, Edmonton, 1985, pp. 1-6.
6. Jarrett, P. M. Evaluation of geogrids for construction of roadways over
muskeg, Proceedings, Symposium on Polymer Grid Reinforcement in Civil
Engineering. Institution of Civil Engineers, London, 1984, Paper 4.5, p. 4.