mechanically reinforced granular shoulders on soft subgrade laboratory and full

8/9/2019 Mechanically Reinforced Granular Shoulders on Soft Subgrade Laboratory and Full

1/12

Mechanically reinforced granular shoulders on soft subgrade: Laboratory and full

scale studies

Mohamed M. Mekkawy a,*, David J. White b,1, Muhannad T. Suleiman c,2, Charles T. Jahren d,3

a Fugro Atlantic Inc., 350 World Trade Center, Norfolk, VA 23510, United Statesb Department of Civil, Construction and Environmental Engineering, Iowa State University, 476 Town Engineering, 50011-3232, United Statesc Department of Civil and Environmental Engineering, Lehigh University, 326 STEPs Building, 1 W Packer Avenue, Bethlehem, PA 18015, United Statesd Department of Civil, Construction and Environmental Engineering, Iowa State University, 458 Town Engineering, 50011 3232, United States

a r t i c l e i n f o

Article history:

Received 13 March 2010

Received in revised form

24 August 2010

Accepted 16 October 2010

Available online xxx

Keywords:

Geogrids

Geosynthetics

Granular shoulders

Rutting

Soil stabilization

Mechanical reinforcement

a b s t r a c t

A recently completed eld study in Iowa showed that many granular shoulders overlie clayey subgrade

layer with California Bearing Ratio (CBR) value of 10 or less. When subjected to repeated trafc loads,

some of these sections develop considerable rutting. Due to costly recurring maintenance and safety

concerns, the authors evaluated the use of biaxial geogrids in stabilizing a severely rutted 310 m tests

section supported on soft subgrade soils. Monitoring the test section for about one year, demonstrated

the application of geogrid as a relatively simple method for improving the shoulder performance. The

eld test was supplemented with a laboratory testing program, where cyclic loading was used to study

the performance of nine granular shoulder models. Each laboratory model simulated a granular shoulder

supported on soft subgrade with geogrid reinforcement at the interface between both layers. Based on

the research ndings, a design chart correlating rut depth and number of load cycles to subgrade CBR

was developed. The chart was veried by eld and laboratory measurements and used to optimize the

granular shoulder design parameters and better predict the performance of granular shoulders.

2010 Elsevier Ltd. All rights reserved.

1. Introduction

Shoulder rutting performance problems are common in areas

where the granular shoulder material is supported by weak

subgrade. With repeated trafc loads,bearing capacity failure in the

subgrade occurs leading to progressive rutting. In addition to being

hazardous to drivers, severely rutted shoulders are expensive to

maintain. Ruts are commonly maintained by shoulder blading and

as necessary adding granular material. These maintenance prac-

tices, however, are considered temporary solutions as they neither

address the problem nor prevent it from reoccurring.

The performance of 25 granular shoulder sections in Iowa was

recently evaluated using various in situ tests with the objective

improving shoulder performance while keeping ownership costslow (Mekkawy et al., 2010). At about 40% of the inspected sections,

the subgrade layer had a California Bearing Ratio(CBR) of 10 or less.

As a result, bearing capacity failure of the subgrade as well as lateral

displacement of the granular and subgrade materials with repeated

trafc loads were frequently observed. Based on the ndings of the

eld study and on a pilot study basis, a granular shoulder test

section overlying a soft subgrade was constructed and monitored.

The test section involved three sections with different geogrids at

the interface between the subgrade and the granular layer. Moni-

toring the test section for a period of about one year demonstrated

the success of geogrid stabilization in eliminating rutting. To

supplement theeld study, a laboratory box model was designed to

evaluate several stabilized models, which were subjected to cyclic

loading with three incremental loading stages. The soil properties

and displacement before and after each test were recorded and

compared. The laboratory box model comprised of a loading frame,

reaction beam, hydraulic actuator, and a steel box to contain the

soil. The overall scope of the eld and laboratory experimentations,which is the focus of this paper, was:

Evaluate the use of geogrid reinforcement to eliminate rutting.

Compare and contrast, through laboratory testing, selected

geogrid stabilizers.

Develop simple designs tools, which will result in more stable

shoulder sections and better prediction of performance in

terms of rut depth and number of loading cycles.

To help design stable granular shoulders, a design chart was

developed from the semi-empirical method proposed by Giroud

* Corresponding author. Tel.: 1 510 267 4436; fax: 1 510 268 0545.

E-mail addresses: [email protected] (M.M. Mekkawy), djwhite@iastate.

edu (D.J. White), [email protected] (M.T. Suleiman), [email protected] (C.T.

Jahren).1 Tel.: 1 515 294 1463; fax: 1 515 294 8216.2 Tel.: 1 610 330 5413.3 Tel.: 1 515 294 3829; fax: 1 515 294 3845.

Contents lists available atScienceDirect

Geotextiles and Geomembranes

j o u r n a l h o m e p a g e : w w w . e l s e v i e r .c o m / l o c a t e / g e o t e xm e m

0266-1144/$e see front matter 2010 Elsevier Ltd. All rights reserved.

doi:10.1016/j.geotexmem.2010.10.006

Geotextiles and Geomembranes 29 (2010) 149e160

Please cite this article in press as: Mekkawy, M.M., et al., Mechanically reinforced granular shoulders on soft subgrade: Laboratory and full scalestudies, Geotextiles and Geomembranes (2010), doi:10.1016/j.geotexmem.2010.10.006
mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]://www.sciencedirect.com/science/journal/02661144http://www.elsevier.com/locate/geotexmemhttp://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://dx.doi.org/10.1016/j.geotexmem.2010.10.006http://www.elsevier.com/locate/geotexmemhttp://www.sciencedirect.com/science/journal/02661144mailto:[email protected]:[email protected]:[email protected]:[email protected]:[email protected]


2/12

and Han (2004a). The chart correlates rut depth and subgrade CBR

(CBRSG). The reliability of the chart was conrmed by comparing it

to eld and laboratory rut depth measurements. Field measure-

ments that were visibly due to subgrade deformation only were

used in this comparison since rutting can sometimes be due to the

cumulative deformation of both the granular and subgrade layers.The chart was developed for a specic axle load, granular layer

thickness, tire pressure, and granular layer CBR (CBRGL). Similar

charts can be generated for a selected set of CBR proles and

loading parameters.

2. Background

According to Fannin and Sigurdsson (1996), where trafc is

channelized, rut is dened as the distance between the initial

elevation of the surface before trafcking and the lower point in the

rut beneath the wheel. Where trafc is not channelized, an erratic

pattern of ruts develop, which can be dened as the distance

between adjacent high and low spots of the base course thickness(Giroud and Han, 2004a). According to Giroud and Han (2004a),

surface rutting occurs by one or more of the following mechanisms:

Compaction of the base course aggregate and/or subgrade soil

under repeated trafc loading.

Bearing capacity failure of the base course or subgrade due to

normal and shear stresses induced by initial trafc.

Bearing capacity failure of the base course or subgrade after

repeated trafc loads which can result in progressive deterio-

ration of the base course, reduction in effective base course

thickness due to contamination by the subgrade soil, a reduc-

tion in the ability of the base course to distribute trafc loads to

the subgrade, or a decrease in the subgrade strength due to

pore pressure build up or disturbance. Lateral displacement of base course and subgrade material due

to accumulation of incremental plastic strains induced by each

load cycle.

During recent years the use of geosynthetics to reinforce

unpaved structures has shown a marked increase. Geosynthetics,

which are typically placed at the interface between the base course

and the subgrade, can carry higher trafc volumes, and can prevent

lateral movement of the base aggregate stiffening the layer and

distributing the wheel loads over a greater area of the subgrade

(U.S. Army Corps of Engineers, 2003). The following are benets of

using geosynthetics as mechanical reinforcement (Berg et al., 2000;

Giroud and Han, 2004a; Subaida et al., 2009; Hufenus et al., 2006,

andPowell et al., 1999):

Reduce the stress on the subgrade;

Enhance subgrade connement and reducing heave

Reduce pumping of the subgrade nes into the base layer;

Prevent contamination of the base materials allowing for more

open graded, free-draining aggregates;

Reduce the excavation depth required to remove unsuitablesubgrade materials;

Reduce the aggregate layer thickness required to stabilize the

subgrade;

Minimize subgrade disturbance during construction;

Reduce maintenance and extend the life of the pavement; and

Prevent development and growth of local shear zones and

allow the subgrade to support stresses close to the plastic limit

while acting elastic.

Two types of geosynthetics are typically used: geogrids and

geotextiles. Geogrids and woven geotextiles have been used as

a reinforcement to increase the resistance to trafc loadings (Giroud

and Noiray,1981). Non-woven geotextiles havebeen mainly usedfor

separation of the base course aggregate and the subgrade.The fundamental reinforcement mechanisms involving the use of

geogrids are: lateral restraint; improved bearing capacity; and

tensioned membrane effect (Giroud and Noiray, 1981and Hufenus

et al., 2006). Lateral restraint refers to the interlocking and conne-

ment of aggregate during loading restricting the lateral ow of the

material (Hufenus et al., 2006). This increases the modulus of the

base course material, which subsequently, increases the area on

which the vertical stress is applied on the subgrade. The bearing

capacity of the shoulder system is improved since more shear stress

is transferred to the reinforcement, which would otherwise be

applied to the soft subgrade layer (Hufenus et al., 2006). The

tensioned membrane effect refers to the deformation of a geogrid

under tensile stress, which in turn improves the vertical stress

distribution. In early research stages, the tensioned membraneeffectwas believed to govern the reinforcement mechanism. However,

later research demonstrated that reinforcement benets are

obtained without signicant deformation, and that lateral restraint

can be the primary reinforcement mechanism followed by the

improved bearing capacity concept (U.S. Army Corps of Engineers,

2003).

The design method for geogrid-reinforced unpaved roads pre-

sented byGiroud and Han (2004a,b) was adopted for comparing

measured and predicted soil displacement as well as fordeveloping

a shoulder design chart. The chart, presented later in this paper,

shows the relationship between rut depth, CBRSG, and number of

load cycles. In their design method, Giroud and Han (2004a,b)

account for the inuence of geogrids by the bearing capacity

factor (Nc), which implies interlock between the geogrid and the

Fig. 1. Shoulder section on new Highway 34 bypass (a) visually suitable shoulder (b) 76 mm rut developed with a few truck passes.

M.M. Mekkawy et al. / Geotextiles and Geomembranes 29 (2010) 149e160150



3/12

base course materials, and aperture stability modulus (J), which isdened as the resistance of the geogrid to in-plane rotational

movementor in-plane rigidity, and is linked to the increase in stress

distribution angle. The design method also accounts for the quality

of the base course materials, the variation in the stress distribution

angle with number of load cycles, and the inuence of rut depth.

Only failure of the subgrade, which is assumed to be saturated and

have low permeability (behaves in an undrained manner), is

considered in this method (Giroud and Han, 2004a). Relationships

between rut depth, CBRSG, and the number of load cycles are pre-

sented later in the text for the date collected in this study.

3. Field observations

Encountering weak subgrade conditions was a reoccurring

observation during a eld study aimed to evaluate the performance

of granular shoulders in Iowa (Mekkawy et al., 2010). By conducting

Dynamic Cone Penetration (DCP) tests (ASTM D6951, 2003), it was

revealed that about 40% of the inspected shoulder sections had

a CBRSGof 10 or less. Of those sections, rutting was observed with

varying degrees depending on the level of off-tracking. Initially,

a section with a soft subgrade layer may appear stable immediately

after construction (Fig. 1a). However, when loaded by trafc,

pumping and rutting develops along the wheel path (Fig.1b). In the

example shown, 76 mm of rut developed only after six truck pas-

ses.DCP tests conducted with varying distance from the pavement

edge yielded CBR values of 6e12 in the upper 200 mm of the

crushed limestone layer, 4e10 in the underlying earth shoulder ll

underlying layer, and 2e29 in the subgrade underlying the earth

shoulderll. It should be noted that some of the rut developed in

this section can be caused by lateral deformation of the granular

layer; however, the authors attribute the majority of the developed

rut to subgrade deformation since contamination of the granular

layer by the underlying subgrade material was observed (Fig.1b) as

well as punching and shear failures. Currently, the Iowa Depart-

ment of Transportation (DOT) has no design requirement for theCBRSG, which consequently does not restrict shoulders from being

constructed over soft foundation soils.

4. Geogrid stabilizatione highway 218 Nashua, IA

4.1. Site description

The inside granular shoulder was experiencing severe rutting

(up to 200 mm at some locations) due to soft subgrade conditions.

The problematic section extended a distance of about 9.6 km. Soft

regions that needed repair were identied and isolated by driving

a fully loaded dump truck weighing 21,337 kg over the shoulder

section and measuring the rut depth at pre-identi

ed locationsalong the wheel path, conducting Clegg Impact tests using a 20 kg

hammer (ASTM D5874, 2007), and DCP tests. The prole of rut

depth and Clegg Impact Value (CIV) with distance starting from

milepost 224 is shown in Fig. 2. The region with the highest rut

depth and lowest CIV, indicating soft conditions, extended about

2000 m (from milepost 220.85e219.60). DCP tests conducted

within this region showed a weighted average CBR of 6 in the upper

200 mm and 5 at a depth between 200 and 500 mm.

Bulk soil samples were obtained from the subgrade and granular

material and classied as SC (A-4) and GW (A-1-a) (Table 1). The

subgrade soil optimum moisture content and maximum dry unit

weight determined using standard Proctor (ASTM D698, 2000) test

were15% and 17.9 kN/m3, respectively. The in situ moisturecontents

and unit weights were measured at 13 locations along the shoulder

section using driven cores and compared to the standard Proctor

curve (Fig. 3). Generally, all in situ dry unit weights were lower than

the standard Proctor curve even in good performing sections. Unit

Distance (m)

0 2000 4000 6000 8000 10000

CIV

0

2

4

6

8

10

12

14

16

18

Ru

tdepth(mm)

0

20

40

60

80

100

120

140

160

180

200

Milepost

218219220221222223224

CIV

Rut depth

Test section

Fig. 2. Rut depth prole measured inside the wheel path and CIV measured at 0.6 m

from the pavement edge.

Table 1

Engineering properties of test section and laboratory box study materials.

Properties of shoulder test section material

Material D10 D30 D60 Cu Cc %P200 %P#4 LL PI USCS AASHTO

Granular 0.75 2.9 6.1 8.1 1.8 4 52 e e GW A-4

Subgrade 0.001 0.008 0.19 190 0.3 49 95 32 23 SC A-1-a

Properties of laboratory box study material

Limestone 0.09 2.0 5.9 66 7.5 10 51 e e S P-SM A-1-a

RAP 0.9 3.1 6.8 7.2 1.5 0.5 42 e e GP A-1-a

Subgrade e 0.006 0.04 e e 78 100 50 32 CH A-7-6

Moisture content (%)

6 8 10 12 14 16 18 20 22 24 26

Dryun

itweight(kN/m3)

13

14

15

16

17

18

19M.P. 223.20

M.P. 224.0

M.P. 223.80

M.P. 223.0

M.P. 223.60

M.P.223.40

M.P. 221.40

M.P. 220.35

M.P. 222.20

M.P. 220.60M.P. 219.20

M.P. 218.40

M.P. 219.80

dmax= 17.9 kN/m

3

wopt= 15%

zav

Fig. 3. Lower eld densities relative to the standard Proctor curve.

M.M. Mekkawy et al. / Geotextiles and Geomembranes 29 (2010) 149e160 151



4/12

weights measured between milepost 220.60 and 219.80 in partic-

ular were considerably lower than the maximum dry unit weight (%

relative compaction80%). Also, most

eld moisture contents werehigher than the optimum moisture content (8.8%).

4.2. Stabilization of test section

Three biaxial geogrid types, henceforth referred to as BX1, BX2,

and BX3, were used to stabilize the shoulder section. The properties

of the geogrids are presented in Table 2. The purpose of using three

geogrid types was to compare performance as there are mechanical

and cost differences between them. The geogrids were placedat the

interface between the subgrade and a new 200 mm overlying

crushed limestone layer. As shown in Fig. 4, the test section was

approximately 310 m long (from milepost 220.60 up to milepost

220.40) and was about 2.4 m wide. The rst 60 m were left unsta-

bilized as a control section. Following the control section was

a 100 m long section stabilized with BX1. Two sections, each 75 m

long, followed the BX1 section, and were stabilized with BX2 fol-

lowed by BX3.

The shoulder test section was stabilized by rst removing and

discarding the contaminated granular layer (Fig. 5a). This layer was

replaced with about 450 tons of new crushed limestone material.

The subgrade was leveled and compacted using a skid loader fol-

lowed by a pneumatic roller (Fig. 5b). Using a power saw, the

geogrids were cut to 2.4 m wide to match the width of the stabi-

lized area. The geogrids were rolled over the soft subgrade starting

with BX1 followed by BX2 then BX3 (Fig. 5c). Beyond approxi-mately 300 m, the BX3 was damaged, which occurred while

transporting the geogrid to the site. The defective grid, denoted by

BX3*, was placed without alteration to evaluate the effect of

improper geogrid installation. A motor grader followed by a pneu-

matic roller were used to spread and compact the aggregate. The

edges of the geogrid were exposed at about 2.4 m from the pave-

ment edge at the end of construction (Fig. 5d). This was due to

tapering of the granular layer thickness during construction (i.e. the

thickness of the granular layer decreased with increasing lateral

distance from the pavement). The entire process of excavation of

contaminated material, geogrid placement, aggregate placement,

and compaction took approximately 5 h. Upon construction

completion, the section was opened to trafc. Unlike some chem-

ical stabilization, mechanical stabilization using geogrid is fast to

install and does not require curing time, which is a benet for

providing minimum disturbance to trafc.

4.3. Field monitoring

The section was continuously monitored using in situ tests over

a period of about one year. After one month from construction, an

edge rut of about 127 mm was measured at the control section as

Table 2

Geosynthetics engineering properties.

Property Test method Geosynthetic material

BX1 BX2 BX3 Woven geotextile Non-woven geotextile

Polymer type Polypropylene

Grab tensile strength (N) ASTM D4632 e e e 1400 712

Tensile strengtha (5% strain) (kN/m) ASTM D6637 11.8 8.5 8.0 e e

Elongation (%) ASTM D4632 e e e

15 50Flexural stiffness (mg-cm) ASTM D5732 750,000 250,000 250,000

Junction efciency (%) 93 93 93

Aperture stabilityb (kg-cm/deg) e 6.5 3.2 2.8 e e

Minimum rib thickness (mm) 1.27 0.76 0.76

Aperture dimensionc ( mm)/Appar ent openi ng siz e ASTM D4751 25 25 33 40 US S td. sieve ( 0. 425 mm) 70 US S td. sieve ( 0. 212 mm)

Resistance to installation damaged (%SC/%SW/%GP) 95/89/86 90/83/70 90/83/70

Resistance to long term degradation (%) 100 100 100

a Tensile Strength values are measured in the machine direction.b Measured in accordance with U.S. Army Corps of Engineers Methodology for measurement of torsional rigidity.c Reported aperture dimension are measured in the machine direction.d Resistance to loss of load capacity when subjected to mechanical installation stress in clayey sand (SC), well graded sand (SW), and poorly graded crushed stone (GP).

Fig. 4. Schematic diagram of the test section (a) plan view (b) cross section.


http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-http://-/?-


5/12

shown inFig. 5e. The stabilized sections showed no signs of rutting

(Fig. 5f).

Part of monitoring the performance of the test section was by

performing plate load tests (PLT). The PLT comprised of a 300 mm

steel plate, hydraulic jack for applying incremental loads, and three

linear variable differential transducers (LVDTs) to measure platedeection. PLT was conducted immediately after construction,

three months, and 10 months from construction at about 1.2 m

from the pavement edge where the granular shoulder is a nominal

200 mm. No tests were conducted near the shoulder edge where

the granular layer started to taper to avoid excessive deformation of

the granular material, which may bias the results by increasing the

magnitude of rutting. Immediately after construction, the highest

modulus of subgrade reaction for both the loading (K) and

reloading (KR) stages of the PLT was measured at the section

stabilized with BX1. K was determined by calculating the secant

modulus between 0 and 10 mm deection. K was slightly higher at

the BX2 section compared to the BX3 section (Fig. 6). This may be

attributed to the small difference in their aperture stability

modulus (the aperture stability modulus of BX2 and BX3 are 3.2

and 2.8 kg-cm/deg, respectively) as well as the small difference in

tensile stiffness (Table 2). As expected, the highest soil deection

was measured at the control section where the average K wasabout

60% less than the BX1K value for the rst loading stage (Table 3).

After three months, a 70% increase in K was measured for all geo-

grid sections compared to values after construction. The increase inK with time can be caused by progressive lateral connement of

aggregate under repetitive trafc loads. It is also possible that as the

section was loaded, the subgrade layer deformed applying tension

forces to the geogrid adding to the stability of the sections. The

increase in K at the control section was due to placing and com-

pacting new crushed limestone material to alleviate the rutting

developed after one month. PLT results obtained after 10 months

showed a reduction in K values for the control section and BX2

section by about 23% and 8%, respectively. The sections stabilized

with BX1 and BX3, however, continued to show increase in K with

time by 5% and 26%, respectively (Table 3).

Using a 20 kg hammer, CIV tests were performed every 15 m.

The tests were performed at different time intervals as shown in

Fig. 7. The CIVs measured after stabilization were about 3 times

Fig. 5. Shoulder reconstruction using geogrids (a) motor grader removing the contaminated granular layer (b) pneumatic roller used to compact the subgrade (c) rolling the geogrid

over the subgrade (d) exposed geogrid at the end of construction (e) rutting developed at the control section after one month (f) no rutting at the geogrid stabilized section after one

month.




6/12

higher than CIVs before stabilization. After three months, addi-

tional increase in surface stiffness was measured at the stabilized

sections evidenced by an average 20% increase in CIVs. No consid-

erable increase in CIVs was measured at the control section. Therewas no clear difference between the CIVs measured at the three

geogrid sections even though their aperture stability modulus

varied. It is possible that the CIV value obtained by dropping a 20 kg

Clegg hammer is inuenced by a relatively shallow depth, and

therefore unable to detect different degrees of connement. At 10

months, CIVs were reduced in all the stabilized sections by an

average of 4%. Relative loosening of the granular layer after the

freeze-thaw period may be a reason behind this reduction.

DCP tests were also used to evaluate and document changes in

the test section performance. DCP testing carried out before and

after stabilization for each geogrid section, and was used to calcu-

late CBRGL(Fig. 8). For the BX1 section, CBRGLincreased from 3 to 18,

whereas for BX2 and BX3 sections, CBRGLincreased from 4 to 19

and 3 to 17, respectively. As anticipated, the strength increase

occurred in the granular layer with no considerable change in CBR

values below the depth of the geogrid (i.e. subgrade layer). The

increase noted at 240 m (Fig. 8c) below about 400 mm is attributedmore to the variations in measurements and repeatability associ-

ated with the DCP test.

Parts of the defective BX3* were exposed after four months from

installation (Fig. 9a). The exposure was a result of severe rutting.

Use of defective geogrid demonstrates no reinforcement benet.

After 10 months, additional geogrid exposure was observed along

section BX3*. The exposed portions were along the edges of the

stabilized section where the grids were originally insufciently

covered with little crushed limestone (Fig. 9b).

In spite of the small reduction in strength parameters and

geogrid exposure, overall, geogrid stabilization improved the

shoulder performance and eliminated shoulder rutting. To prevent

geogrid exposure in future stabilization applications, the thickness

Stre

ss

(kN/m

2)

0

200

400

600

Deflection (mm)

0 10 20 30 40

Stress

(kN/m

2)

0

200

400

600

Deflection (mm)

0 10 20 30 40

Control

Section

BX1

BX2 BX3

K

KR

K

K

K

K

K

KR

KR

KR

KR

15 m, K=12.03 MN/m3, K

R=36.17 MN/m

3

30 m, K=7.18 MN/m3, K

R= 40.83 MN/m

3

322 m, K=12.58 MN/m3, K

R=42.6 MN/m

3

91 m, K=26.28 MN/m3, K

R=80.43 MN/m

3

122 m, K=26.41 MN/m3, K

R=69.67 MN/m

3

183 m, K=18.92 MN/m3, K

R=53.17 MN/m

3

213 m, K=26.23 MN/m3, K

R=79.03 MN/m

3

244 m, K=20.59 MN/m3, K

R=60.65 MN/m

3

274 m, K=15.73 MN/m3, K

R=47.74 MN/m

3

Fig. 6. Plate load test results immediately after construction.

Table 3

Modulus of subgrade reaction determined from plate load testing.

Section Distance (m) After construction 3 months after construction 10 months after construction

K (kN/m3) KR(kN/m3) K (kN/m3) KR(kN/m

3) K (kN/m3) KR(kN/m3)

Control section 15 12030 36170 32050 112260 28100 201470

30 7180 40830 28720 88690 25040 125400

322 12580 42600 7900 34600 10140 54670

BX1 91 26282 80427 28250 107500 40550 136510

122 26400 69670 55090 122430 35650 109470

BX2 183 18900 53170 33670 94340 27380 80680

213 26230 79030 18030 90270 28040 83070

BX3 244 20590 60650 28560 78500 35200 109160

274 15730 45730 25820 60300 25990 111930




7/12

of the overlying granular layer should be uniform. Based on the

experience from this test section, a nominal 150 mme

200 mmgranular layer thickness is recommended.

5. Laboratory box study

A laboratory testing program was developed to supplement the

eld study and evaluate the performance ofve geogrids and geo-

textilesand twogranularmaterialsin reducing shoulderrutting.The

selected materials were tested by designing a laboratory box model

to simulate a granular shoulder overlying a soft subgrade. The box

model was then stabilized with each geogrid and geotextile and

subjected to cyclic loading where the surface displacement was

recorded. The cumulative soil displacement and change in soil

properties afterthe test were used toassess theperformanceof each

geogrid and geotextile.

5.1. Test setup

The model setup consisted of a loading frame, reaction beam,

and a hydraulic actuator (Fig. 10a). The actuator had a maximum

force of 250 kN and a dynamic stroke of 150 mm. Twelve 200 mm c-

channels were assembled together to form a 0.2 m3 steel box

(0.6 m 0.6 m 0.6 m), used to contain the soil. The box was

loaded using a 150 mm diameter loading plate (Fig. 10b). To mini-

mize friction and stress concentrations, a compressible 12 mm

thick neoprene pad covered with a Teon sheet was placed at the

interface between the soil and the steel box.

The subgrade soil was placed at the bottom of the box and

compacted in 76 mm lifts by applying a static load to reach a nal

thickness of 300 mm.The reinforcement wasplaced at the interface

between the subgrade and the overlying granular layer. Similar to

the subgrade layer, the granular layer was compacted by applying

a static load to reach a nal thickness of about 150 mm (Fig. 11). To

ensure soft foundation conditions for all tests, the target range of

CBRSGvalues was selected to vary between 3 and 5, whereas the

target range of CBRGLvalues was 4e7.

The hydraulic actuator was used to apply a sinusoidal load pulse

to the 150 mm diameter loading plate. Three incremental cyclic

loads were applied tothe soil each sustained for5000 cycles (totalof

15,000 cycles) at a frequency of 1 Hz. The initial cyclic pressure was

275 kPa,which was then increased to550 kPa and then 827 kPa.The

frequency of one cycle per second was sufcient in sustaining the

applied load despite the large deections observed at some tests.

The hydraulic actuator control system was used to collectdisplacement data at predetermined load cycles. DCP and a 4.5 kg

clegg hammer were used to document the changes in CBR and CIV

for each soil layer before and after each test.

During the test, and for every loading stage, the deection of the

reaction beam was measured using a dial gauge. A linear relation-

ship was developed between the applied load and beam deection.

The measured beam deections were subtracted from the recorded

soil displacement at the corresponding loading stage to calculate

the nal soil displacement.

5.2. Materials

Three soil materials were used throughout the laboratorytesting program. The subgrade soil consisted of Paleosol clay with

a PI of 32 and classied as CH (fat clay; A-7-6). The materials used

for the granular layer were Class A crushed limestone and recycled

asphalt pavement (RAP). The crushed limestone was classied as

SP-SM (A-1-a) with optimum moisture content and maximum dry

unit weight of 6% and 21.9 kN/m3, respectively, whereas the RAP

CBR

1 10 100

Depth(mm)

0

200

400

600

800

CBR

0.1 1 10 100

CBR

1 10 100

Before stabilization After stabilization

180 m 240 m

a b cGranularlayer

Subgradelayer

60 m

Increase inCBRGLafter

stabilization

Fig. 8. DCP results before and after stabilization (a) BX1 (b) BX2 (c) BX3.

Distance (m)

0 50 100 150 200 250 300

CIV

0

2

4

6

8

10

12

14

16

18

20

Before stabilization

After stabilization

3 months

10 months

Controlsection

BX1 BX2 BX3

BX3*

Rutting after1 month.New aggregateadded

D

amage

dgri

d

Fig. 7. CIV prole with time at 1.2 m from the pavement edge.




8/12

material was classied as GP (A-1-a) with a ne content of about

0.5% compared to 10% for the crushed limestone (Table 1).

The laboratory study assessed the performance of three poly-

propylene biaxial geogrid types, polypropylene woven, and non-

woven geotextiles. The biaxial geogrids were the same type as theones used to stabilize the eld test section (i.e. BX1, BX2, and BX3).

The geotextiles were a woven geotextile lm used mainly for soil

separation and stabilization, and a needle-punched non-woven

geotextile ber. The properties of the selected geosynthetics mate-

rials are presented in Table 2.

5.3. Test results

A total of nine tests were executed and compared to determine

the performance differences for the selected materials. During all

tests, the subgrade moisture content for was kept constant at about

25% during placement. The changes in soil properties before and

after each test are presented in Table 4, whereas the soil cumulative

displacements recorded by the hydraulic actuator control systemare presented inFig. 12.

5.3.1. Test no. 1 e control

Therst test performed was a control test, which modeled eld

conditions where a granular shoulder overlies a soft subgrade layer.

At thebeginning of thetest, thedry unit weightfor thesubgradeand

granular layers were 19 kN/m3 and 13.4 kN/m3, respectively. The

CBRSGvalue, calculated from DCP tests, increased from 3 to 9 after

the test was completed as a result of subgrade soil densication. On

the contrary, there was no signicant change in the CBRGLvalues.

The nal soil displacement after 15,000 cycles was 284 mm. Visualinspection revealed about 50 mm of aggregate punching into the

subgrade layer.

5.3.2. Test nos. 2, 3, and 4e BX1

In test Nos. 2, 3, and 4, BX1 was placed at the interface between

the granular and subgrade layers. During test No. 1, the dimensions

of geogrid were 6 mm shorter than the dimensions of the box to

eliminate any interaction between the grid andthe box. The dry unit

weight of the subgrade and granular layers were 18.6 kN/m 3 and

14 kN/m3, respectively. After the test, the CBRSG value increased

from 5 to 7 indicating less soil densication compared to Test No. 1.

Further, the subgrade modulusdetermined by LWD (ELWD) increased

from 7.0 to 12.0 MPa. Almost no change in properties of the granular

layer was measured. The maximum measured soil displacement at15,000 cycles was 125 mm, which is about 55% less than the control

test. The measured soil displacement was compared to the predicted

soil displacement at the endof each loading stage. The predicted soil

displacement was determined using a semi-empirical method out-

lined by Giroud and Han (2004a,b). It should be noted that this

method does not account for staged loading so rut that developed

Fig. 9. Exposed geogrid after 10 months (a) BX3* (b) BX1.

Fig. 10. Schematic of the laboratory apparatus setup (a) Steel frame and hydraulic actuator used for loading the stabilized soil (b) steel box used to contain the soil.




9/12

during previous stages is not accounted for. Therefore, rut depths

predicted for stages 2 and 3 were based on loading and the number

of cycles only. This may explain some of the discrepancy between

measured and predicted rut depths. The predicted and measured

soil displacements were similar at the end of the rst loading stage

(i.e. at 5000 cycles). However, the predicted soil displacement was

considerably lower than the measured one at the end of the second

and third loading stages (Fig. 13). Examining the unanchored geo-

grid after the test revealed that the geogrid was deformed and

pulled to the center of the box (a phenomenon that would not occur

in the eld) under the effect of repetitive cyclic loading. This

observation may explain the difference between the measured and

predicted soil displacement. Similar to the control test, most of thesoil displacement occurred during the rst 500 cycles of each load

increment. The amount of aggregate punching through the subgrade

layer was reduced compared to Test No. 1 but was not eliminated.

Visual inspection revealed an aggregate punching depth of about

25 mm.

During Test No. 3, the four corners of the geogrid were anchored

using steel rods driven to the bottom of the subgrade layer. The dry

unit weight of the subgrade and granular layers were 19.2 kN/m3

and 14.2 kN/m3, respectively. As a result of soil displacement, the

CBRSGvalue increased from 5 to 8 and the ELWDincreased from 8.0

to 11.0 MPa. The results show no change in the properties of the

granular layer. Partially anchoring the geogrid further decreased

the soil displacement by about 10% compared to Test No. 2 (geogrid

was not anchored). Moreover, partially anchoring the geogrid

decreased the difference between measured and predicted soil

displacement (Fig. 13). At 15,000 cycles the difference between the

measured and predicted soil displacement was 47 mm for Test No.

2 and 35 mm for Test No. 3. Visual inspection showed punching of

aggregate through the subgrade soil to a depth of about 25 mm.

The setup of Test No. 4 was similar to the previous two tests

except that the entire perimeter of the BX1 geogrid was xed to

eliminate any geogrid movement. This was accomplished by xing

the geogrid edges between the c-channels used to assemble the

wallsof thesteelbox.The purpose of eliminating geogrid movement

is to better represent the geogrid eld behavior where movement isprevented yet the geogrid can still deform and put in tension under

repeated load. The dry unit weight of the subgrade and granular

layers were 18.7 kN/m3 and 13.8 kN/m3, respectively. The CBRSGvalue increased from 4 to 7 after the test. Also, the subgrade E LWDincreased from 8.0 to 10.0 MPa. By restricting the geogrid move-

ment, the measured soil displacement was further reduced

compared to Test Nos. 2 and 3. At 15,000 cycles, the soil displace-

ment was 35% lower than that measured in Test No. 3 (75% lower

than the control test). Also, the differences between the measured

and predicted soil displacement at all three loading stages were

reduced (Fig. 13). Since locking the geogrid yielded a more repre-

sentative behavior of a geogrid installed in the eld, other

Fig. 11. Laboratory box setup (a) applying a static load to compact the soil (b) applying cyclic loading through a 150 mm loading plate.

Table 4Change in soil properties before and after each test.

Test No. Test

description

Dry unit weight of

granular layer

(kN/m3)

Dry unit weight of

subgrade layer

(kN/m3)

CBRGL CBRSG CIVGranular CIVSubgrade

Before test After test Before test After test Before test After test Before test After test

1 Control 13.4 19.0 5 6 4 9 2.9 7.1 e e

2 BX1a 14.0 18.6 5 5 4 7 3.7 5.4 3.2 6.2

3 BX1b 14.2 19.2 6 6 5 8 4.8 4.4 4.2 8.1

4 BX1 13.8 18.7 5 6 4 7 4.3 5.3 4.1 7.8

5 BX2 13.5 18.7 4 5 5 8 4.7 3.6 4.0 6.6

6 BX3 14.2 19.2 4 5 5 9 3.7 4.1 3.8 8

7 Woven geotextile 13.4 19.0 5 6 4 10 3.4 6.1 4.1 6.3

8 Non-woven geotextile 14.0 18.9 6 7 4 9 4.9 5.8 3.6 6.5

9 BX1 with RAP 14.0 19.2 5 5 4 8 4.0 4.2 4.0 7.0

a Not anchored.b Partially anchored.


http://-/?-http://-/?-http://-/?-http://-/?-


10/12

mechanical reinforcements used later in this laboratory study were

xed to the steel box in a similar manner.

5.3.3. Test nos. 5 and 6e BX2 and BX3

The soft subgrade was stabilized with BX2 in Test No. 5 and BX3

in Test No.6. The engineering properties of both geogrids are

summarized inTable 2. The dry unity weight of the subgrade and

granular layers for Test No. 5 were 18.7 kN/m3 and 13.5 kN/m3,

respectively. Afterthe test,the CBRSG value increasedfrom5 to8 and

ELWD increased from 9.0 to 12.0 MPa. The soil displacement

measuredat 15,000cycleswas about 15%higherthanTest No.4 (BX2

geogrid). However, the soil displacement was still about 70% less

than the control test.

The dry unit weight of thesubgrade and granularsoils forTest No.

6, which was stabilized with BX3, were 19.2 kN/m3 and 14,2 kN/m3,

respectively. Dueto soil densication, the CBRSG increasedfrom5 to9

andELWDincreased from 9.0 to 16.0 MPa. Similarto previous tests,the

properties of the granular layer did not change. Test No. 6 resulted ina similar soil displacement as Test No. 5 (displacements overlap in

Fig.12). This may be attributed to the somewhat similar mechanical

properties of the geogrids. It is apparent from the cumulative

displacement of Test Nos. 4, 5, and 6 that the BX1 showed better

performance; nonetheless, the other geogrid types greatly reduced

the soil displacement when compared to the control test.

5.3.4. Test nos. 7 and 8e woven and non-woven geotextiles

For Test Nos. 7 and 8, the subgrade layer was stabilized with

woven and non-woven geotextiles (Table 2). The geotextiles are

used primarily for soil separation and stabilization. The soil

densication, which occurred during Test No.7, was reected in the

increase in CBRSGfrom 4 to 10 and the increase in E LWDfrom 9.0 to

14.0 MPa. After 15,000 cycles, the soil displacement was about

78 mm equal in magnitude to the displacement measured during

the BX2 and BX3 tests. One advantage of using woven geotextiles is

the complete elimination of aggregate punching by separation of

granular and subgrade materials.

A non-woven geotextile was used to stabilize the soft subgrade

soil during Test No. 8. Similar to previous tests, the CBRSGand ELWDincreased after the test, while there was no considerable change of

the granular soil properties. For the rst and second loading stages

(10,000 cycles), the non-woven geotextile showed better perfor-

mance and managed to reduce soil displacement by up to 25%

compared to the woven geotextile test. Also, the non-woven geo-textile outperformed all geogrids. However, at the third loading

stage, the soil displacement increased rapidly exceeding that

measured during the BX1 test (Test No. 4). The displacement at

15,000 cycles was about 82 mm (70% lower than the control test).

Visual observations showed that the non-woven geotextile elimi-

nated aggregate punching through the subgrade.

Number of cycles

0 2000 4000 6000 8000 10000 12000 14000

Soildispla

cement(mm)

0

50

100

150

200

250

300Control

BX1 not anchored

BX1 partially anchored

BX1

BX2

BX3

Woven geotextile

Nonwoven geotextile

BX1 (RAP)

Fig. 12. Measured soil displacement with increasing load cycles.

Number of cycles

0 2000 4000 6000 8000 10000 12000 14000 16000

Displacement(mm)

0

20

40

60

80

100

120

140 x= Predicted rut value based

on Giroud and Han (2004)

CBRGL= 5

CBRSG= 4

Unanchored

Partiallyanchored

Fully anchored

Fig. 13. (Color) Measured and predicted soil displacement for the BX1 tests.

Measured soil displacement (mm)

0 50 100 150 200 250 300 350

Predictedsoildisplacemen

t(mm)

0

50

100

150

200

250

300

350Control

BX1 not anchored

BX1 partiallyanchored

BX1

BX2

BX3

Woven geotextile

Nonwoven geotextile

BX1 (RAP)

Fig. 14. Measured versus predicted soil displacement.




11/12

5.3.5. Test no. 9e BX1 with recycled asphalt pavement

Due to the increasing use of recycled material in shoulder

applications, this test aimed to document the performance of RAP

material when used as a granular layer. Similar to the setup of Test

No. 4, the underlying subgrade was stabilized by placing the BX1 at

theinterface between both layers. After the test, the CBRSG andELWDvalues increase from 4 to 8 and from 8.0 to 12.0 MPa, respectively.

Compared to Test No. 4, where the granular layer comprised of

crushed limestone, the soil displacement was about 20% higher at

the end of the third loading stage. The authors attribute the differ-

ence in displacement to a less stable granular layer because of the

lower percentage ofnes content in the RAP material.

The measured and predicted soil displacements were compared

forall threeloading stages as shownin Fig.14. Overall,thereis a good

agreement between the measured and predicted soil displacement.

The design method over predictedthe soil displacement for six tests.

The range of over prediction ranged from 8% (control test) to 40%

(woven and non-woven geotextiles tests). The over estimation of

soil displacement was considerably lessfor geogrid tests(an average

of 23%) compared to 40% for the geotextiles. The design method

appears to be more applicable for geogrid reinforcements. The

design method underestimated the soil displacement for the

unanchored BX1, partially anchored BX1, and BX1 with RAP tests.

The range of under estimation ranged from 16% (BX1 with RAP) to

37% (Unanchored BX1). It is believed that the Giroud and Han

(2004a,b) method can, with an acceptable level of accuracy, esti-

mate the magnitude of rutting for granular shoulders.

6. Shoulder design charts

After demonstrated applicable, the Giroud and Han (2004a,b)

method was utilized to develop a shoulder design chart to help

design stable shoulders and mitigate rutting that occurs due to

bearing capacity failure of the subgrade. The magnitude of rutting is

predicted based on deformations in the subgrade layer only. In

other words, rutting that may occur due to degradation of the

granular layer is not accounted for. The chart can be a rapid tool for

designing new granular shoulders and provide basis for QA/QC

specications.Fig. 15shows an example of a design chart that was

developed for an unreinforced granular shoulder with a 150 mm

thick granular layer, 80 kN load, 550 kPa tire pressure, and a CBRGL

of 6, which are common parameters encountered eld values. To

use this chart, an allowable rut depth for a granular shoulder

section is selected and the corresponding CBRSG is computed for

a certain number of load cycles (N). Similar charts can be generated

for any set of shoulder parameters. Field and laboratory rut depth

measurements were compared to the design chart. Even with some

unknown eld parameterssuch as N and trafc loads,the measured

and predicted rut depths are in a relatively good agreement. The

chart was found to be simple yet practical for designing new

unreinforced granular shoulders, QA/QC, and predicting the

performance of existing granular shoulder.

7. Summary and conclusions

A common shoulder performance problem in Iowa is granular

shoulders overlying soft subgrade soils. Once developed, this

problem is both hazardous and difcult to maintain. Field obser-

vations of granular shoulder across the state of Iowa demonstrated

that many existing sections have a soft subgrade layer with a CBRSGof 10 or less. The maintenance alternative using mechanical stabi-

lization was evaluated by constructing a test section, where the soft

subgrade was stabilized using three biaxial geogrids. The shoulder

performance wasevaluated over a periodof 10 month using various

in situ tests. Overall, all three geogrid types managed to eliminate

shoulder rutting and improve the strength properties of the

shoulder section evidenced by the increase in CBR, CIV, and K with

time. Compared to chemical stabilization, geogrid stabilizationallows for a more rapidreconstruction. Further, the repaired section

can be opened immediately for trafc with no curing time required.

Other mechanical stabilization alternatives such as woven, non-

woven geotextiles, and the use of alternative shoulder granular

material like RAP were evaluated by conducting a laboratory box

study. Subjected to 15,000 cyclic loads, each stabilizer was evalu-

ated based on the cumulative measured soil displacement. The

following conclusions are withdrawn from the laboratory study:

Thehighest soil displacement was measured during the control

test, and was about 284 mm after 15,000 cycles.

To better represent eld behavior, movement of the mechan-

ical reinforcement was constrained by xing the geogrid

perimeter to the steel box. This reduced the soil displacement

Allowable rut depth (mm)

0 50 100 150 200

CBRSG

0

10

20

30

40

N = 10

N = 100

N = 1,000

N = 10,000

Field measurements

Lab measurements

CBRGL

= 6

Axle load = 80 kNGranular layerthickness = 150 mmRange of CBR values

for chemical stabilization

Range of in situCBR values

Fig. 15. Unreinforced shoulder design chart correlating CBRSG with allowable rut depth.




12/12

and resulted in smaller differences between measured and

predicted soil displacements.

The CBR and ELWD values of the subgrade layer always

increased after the test as a result of soil densication. There

was almost no change in the properties of the granular layer.

BX1 reduced soil displacement by about 75% compared to the

control test, whereas using BX2 and BX3 reduced soil

displacement by about 70%. All geogrids, however, did not

prevent aggregate punching through the subgrade layer.

The woven and non-woven geotextiles reduced the soil

displacement by about 70% compared to the control test. The

performance of the non-woven geotextile started to deterio-

rate after the third loading stage. Both geotextiles were

successful in separating the granular and subgrade layer and

eliminating aggregate punching.

Using RAP as a granular material resulted in a 20% increase in

soil displacement compared to using crushed limestone.

TheGiroud and Han (2004a,b)semi-empirical method slightly

over predicts the soil displacements at the end of each loading

stage, except for Test Nos. 2, 3, and 9 (unanchored, partially

anchored BX1, and BX1 with RAP).

The results of the eld and laboratory studies indicate thatmechanical reinforcement as is an effective method in reducing

rutting and repairing granular shoulder overlying soft foundations.

Additional monitoring of the constructed test section is needed to

verify the long term performance and evaluate the effect of higher

trafc load and freeze-thaw cycles.

To design or repair unreinforced granular shoulders, the

provided design chart can be a simple design tool in selecting

a minimum CBRSGvalue corresponding to an allowable magnitude

of subgrade rutting. The chart may also provide bases for QA/QC.

Acknowledgments

The Iowa Department of Transportation and the Iowa Highway

Research Board sponsored this study under contract TR-531. The

authors appreciate the help of the technical steering committee in

identifying shoulder sections for investigation, and in their assist in

rening the research tasks. The authors would also like to thank

Iowa DOT personnel, Jim Howely for providing geogrid materials,

Heath Gieselman, Mike Kruse, and Amy Heurung, for their assis-

tance with eld and laboratory testing.

References

ASTM D5874, 2007. Standard test method for determination of the impact value (IV)of a soil. American Standard Testing Methods, West Conshohocken, Phila-delphia, pp. 1e9.

ASTM D6951, 2003. Standard test method for use of the dynamic cone penetrom-eter in shallow pavement applications. American Standard Testing Methods,West Conshohocken, Philadelphia, pp. 1e7.

ASTM D698, 2000. Standard test method for laboratory compaction characteristicsof soil using standard effort (12,400 ft-lbf/ft3 (600 kN-m/m3)), West Con-shohocken, Philadelphia, pp. 1e11.

Berg, R.R., Christopher, B.R., Perkins, S., 2000. Geosynthetic Reinforcement of theAggregate Base/subbase Courses of Pavement Structures. Geosynthetics Mate-rials Association, Roseville, MN, pp. 1e176.

Fannin,R.J., Sigurdsson,O., 1996. Field observations on stabilizationof unpavedroadswith geosynthetics. J ournal of Geotechnical Engineering 122 (7), 544e553.

Giroud, J.P., Han, J., 2004a. Design method for geogrid-reinforced unpaved roads. I.Development of design method. Journal of Geotechnical and GeoenvironmentalEngineering 130 (8), 775e786.

Giroud, J.P., Han, J., 2004b. Design method for geogrid-reinforced unpaved roads. II.Calibration and applications. Journal of Geotechnical and GeoenvironmentalEngineering 130 (8), 787e797.

Giroud, J.P., Noiray, L., 1981. Geotextile-reinforced unpaved road design. Journal ofthe Geotechnical Engineering Division 107 (GT9), 1233e1254. Proceedings ofthe American Society of Civil Engineers.

Hufenus, R., Rueegger, R., Banjac, R., Mayor, P., Springman, S., and Brnnimann, R.2006. Full-scale eld tests on geosynthetics reinforced unpaved roads on softsubgrade. 24 1. 21e37.

Mekkawy, M.M., White, D.J., Jahren, C.T., Suleiman, M.T., 2010. Performance prob-lems and stabilization techniques for granular shoulders. Journal of Perfor-mance of Constructed Facilities 24 (2), 159e169.

Powell, W., Keller, G.R., Brunette, B., 1999. Application for geosynthetics on forestservice low-volume roads. Transportation Research Record 1652, 113e120.

Subaida, E.A., Chandrakaran, S., Sankar, N., 2009. Laboratory performance ofunpaved roads reinforced with woven coir geotextiles. Geotextiles and Geo-membranes 27 (3), 204e210.

U.S. Army Corps of Engineers. 2003. Use of geogrids in pavement construction.Technical Letter No. 1110-1-189, Washington, D.C., pp. 1e38.

Notations

The following symbols are used in the paper:

Nc:Bearing capacity factorJ:Aperture stability modulus

CIV:Clegg Impact ValueCBRSG, CBRGL:California Bearing Ratio of the subgrade and granular layersK, KR:Modulus of subgrade reaction during load and reload cyclesELWD:Modulus measured using light weight deectometerD10, D30, and D60:Sieve size through which 10%, 30%, and 60% of the particles would

passCu:Coefcient of uniformityCc:Coefcient of gradation

%P#4: Percent passing the No. 4 sieve%P#200: Percent passing the No. 200 sieve


Please cite this article in press as: Mekkawy M M et al Mechanically reinforced granular shoulders on soft subgrade: Laboratory and full scale

mechanically reinforced granular shoulders on soft subgrade laboratory and full

Documents