You are on page 1of 17

The International Journal of Pavement Engineering, Vol. 4 (1) March 2003, pp.

3750

A Design Model for Geosynthetic-reinforced Pavements


S.W. PERKINS* and M.Q. EDENS

Department of Civil Engineering, 205 Cobleigh Hall, Montana State University, Bozeman, MT, 59717, USA

(Received 12 February 2002; In revised form 1 November 2002)

The first paper to this companion series of articles [Perkins, S.W and Edens, M.Q. (2003) Finite
elements and distress models for geosynthetic-reinforced pavements, International Journal of
Pavement Engineering, 3(4), 239 250] described a finite element response model and empirical
damage models that were combined to form a mechanistic empirical (M E) for geosynthetic-
reinforced flexible pavement systems. The model provides predictions of an extension of service life or
a reduction in unbound aggregate thickness for equivalent service life when the geosynthetic is placed
at the bottom of the aggregate layer. In this paper, the model is used in a parametric study involving the
analysis of over 465 cases to provide predictions of reinforcement benefit for a range of design
parameters. The parameters varied are those that are believed to be most influential on reinforcement
benefit and include thickness of the structural section, strength and/or stiffness of the subgrade, tensile
modulus in both principal directions of the geosynthetic and several other properties related to the type
and structure of the geosynthetic. Regression equations are used to relate reinforcement benefit results
from the parametric study to these pavement design parameters and form the basis of the design model
presented in this paper. This model is calibrated against the results of large-scale reinforced pavement
test sections. The design model allows for the design of geosynthetic-reinforced flexible pavements
using generic input parameters for the pavement system.

Keywords: Pavement; Geosynthetic; Reinforcement; Model; Design

INTRODUCTION The second represented the maximum amount of


reinforcement benefit that could be expected with a
The first paper of this companion paper series (Perkins and perfect reinforcement product and was accomplished
Edens, 2003) described a finite element (FE) response by restraining the lateral degree of freedom of the
model and empirical damage models developed for element nodes along the interface between the aggregate
geosynthetic-reinforced pavements. Response parameters and subgrade layer. The third type of model used a
extracted from the FE model were vertical compressive separate material layer for the geosynthetic inserted into
strain in the top of the subgrade and bulk stress within the the finite element mesh of the unreinforced pavement
base course aggregate layer. These response parameters model.
were used in empirical damage models to provide The mechanistic empirical (M E) model resulting
predictions of long-term pavement performance and from the combination of the FE response model and
definitions of reinforcement benefit. The damage models damage models was developed for use in a parametric
provided a means of describing reinforcement benefit in study from which a design model is developed. The M E
terms of a Traffic Benefit Ratio (TBR), which defines the model was used for the analysis of approximately 465
increase in service life of a reinforced pavement as pavement design cases. The parameters varied for these
compared to an equivalent unreinforced pavement. cases includes:
Expressions for TBR from each of the two damage
models provided descriptions of reinforcement effects for 1. Asphalt concrete thickness.
the aggregate layer and for the subgrade layer, and were 2. Unbound aggregate thickness.
combined together to form a TBR for the total system. 3. Material properties of the subgrade leading to different
Three types of FE response models were developed. subgrade strength and/or stiffness values.
The first was for a conventional unreinforced pavement. 4. Material properties of the geosynthetic.

*Corresponding author. Tel.: 1-406-994-6119. Fax: 1-406-994-6105. E-mail: stevep@ce.montana.edu

ISSN 1029-8436 print/ISSN 1477-268X online q 2003 Taylor & Francis Ltd
DOI: 10.1080/1029843031000097562
38 S.W. PERKINS AND M.Q. EDENS

The three FE response model types (unreinforced, Execution of models according to this approach minimizes
perfect-reinforced and geosynthetic-reinforced) were used the number of cases that need to be analyzed and provides
to provide a systematic means of identifying variations in for a systematic method for interpreting the results.
reinforcement benefit for the parametric study variables
used.
Variation of Parameters for Perfect Reinforced Models
Regression equations were developed from the para-
metric study results to relate reinforcement benefit to the Perfect-reinforced FE response models were created for
variables used. The design model was then based on these five values of asphalt concrete thickness, six values of
regression equations and was calibrated from the results of aggregate thickness and for six sets of subgrade properties,
large-scale pavement test sections. resulting in 180 cases analyzed. For each perfect-
reinforced FE model, an identical unreinforced FE
model was created and analyzed. The comparison of
PARAMETRIC STUDY perfect-reinforced and unreinforced response parameters
through the two damage models resulted in definitions of
The M E model described in the companion paper was reinforcement effects in the base aggregate (TBRB) and
used to analyze over 465 cases involving the variation of reinforcement effects in the subgrade layer (TBRS) for
seven parameters related to the geometry and materials in each combination of asphalt concrete thickness, aggregate
unreinforced and reinforced pavement cross-sections. For thickness and subgrade stiffness.
each case analyzed, the FE response model was used to The values of the parameters varied for this part of the
provide two response parameters, namely vertical study are listed in Table I. The constitutive model used for
compressive strain in the top of the subgrade and bulk the subgrade contained six parameters that were most
stress in a representative volume of the base course responsible for the materials strength and stiffness
aggregate. These response parameters were then used in behavior. Each of these six parameters was varied for
the two damage models described in the companion paper each subgrade material parameter set. Resilient modulus
to determine values of reinforcement benefit, defined in tests were simulated with each material parameter set,
terms of TBR, for each damage model. This process as described in the companion paper, and produced
involved the comparison of response parameters between the values shown in Table I. The approximate subgrade
reinforced and unreinforced pavement cross-sections. CBR strength for each of these resilient modulus values is
The three types of FE models developed were used to also listed in Table I.
describe the effects of reinforcement in a systematic way As an example of the results obtained from the 360
that isolates the effects of each variable or group of cases analyzed, Fig. 1 shows the variation in the two
variables on the system. The unreinforced models provide response parameters with aggregate thickness for the case
response parameters for pavements without any type of of AC3 and S4. Vertical strain in the top of the subgrade is
reinforcement. The perfect-reinforced models do not seen to decrease as aggregate thickness increases.
contain parameters associated with the geosynthetic. Subgrade strain is consistently less for the perfect-
Benefit from the perfect-reinforcement model cases is reinforcement case as compared to the unreinforced case,
then a function of the remaining pavement system however, the differences become insignificant for an
variables, which include AC and aggregate thickness, and aggregate thickness of 600 mm. Cases of perfect
subgrade stiffness. From these results, conclusions are reinforcement are seen to result in higher levels of bulk
made regarding the combinations of structural section stress in the aggregate layer.
thickness and subgrade stiffness for which perfect Figure 2 shows variations of TBRS, as determined by
reinforcement provides benefit. The third type of model use of the damage model for subgrade rutting, as a
contains a reinforcement layer with corresponding material function of aggregate thickness and AC thickness for the
properties. Variation of four basic geosynthetic properties case of S4. The results show a nominal effect of
allows for conclusions to be made regarding how those reinforcement at an aggregate thickness of 150 mm and for
properties modify or reduce the maximum benefit seen lower values of AC thickness, indicating that reinforce-
from the corresponding perfect reinforcement cases. ment is not significant for excessively thin pavement

TABLE I Parameters varied for perfect-reinforcement cases

AC thickness case name AC1 AC2 AC3 AC4 AC5


AC thickness (mm) 50 75 100 125 150

Base thickness case name B1 B2 B3 B4 B5 B6


Base thickness (mm) 150 200 250 300 400 600

Subgrade case name S1 S2 S3 S4 S5 S6


Subgrade MR (MPa) 5.26 10.3 20.4 42.1 83.6 154
Subgrade CBR 0.5 1 2 4 8 15
DESIGN MODEL FOR REINFORCED PAVEMENTS 39

FIGURE 3 TBRS versus SN for S4 cases.

companion paper (Perkins and Edens, 2002) demonstrated


the correlation of these layer coefficients to mechanical
properties. Parameters D1 and D2 are the thickness of the
FIGURE 1 Response parameters versus aggregate thickness for AC3 AC and aggregate layers, respectively.
and S4 cases.
SN 0:4 D1 0:14 D2 1

Figure 3 shows TBRS plotted against SN for the same


sections. This observation is supported by results from test
cases showed in Fig. 2. While some scatter exists, the
sections reported in the literature (Berg et al., 2000). For
results suggest a relationship between TBRS and SN. This
an aggregate thickness of 150 mm, as the AC thickness
relationship, and others for TBRB, are more closely
increases, TBRS increases. As the aggregate thickness
examined for all cases in later sections of the paper.
increases, TBRS increases to a point, where this point
Results from the parametric study for the 180 perfect
appears to be a function of the AC thickness. For an
reinforcement cases analyzed can also be used to assess
aggregate thickness of 600 mm, TBRS decreases and is not
Base Course reduction Ratio (BCR), or more specifically
appreciably influenced by AC thickness. These results
combinations of BCR and TBR for certain comparison
indicate a relatively complex relationship between TBRS,
sets. For each unreinforced case where perfect-reinforced
aggregate thickness and AC thickness.
runs were made for the same AC thickness, the same
To reduce the number of variables in the system, a
subgrade condition and for reduced base thickness, a value
structural number for each combination of AC and
of BCR is calculated from the base thickness used in the
aggregate thickness was computed from Eq. (1), which is
unreinforced and perfect-reinforced models. For instance,
adopted from the AASHTO pavement design method
if case AC1 B6 S1 for an unreinforced pavement is
(AASHTO, 1993) and assumes layer coefficients of 0.4
compared to AC1 B5 S1 for a perfect-reinforced case, a
and 0.14 for the AC and aggregate layers, respectively.
BCR of 33% is calculated from the two models. From the
Material modeling and calibration work described in the
360 model runs, 75 sets of unreinforced and reinforced
comparison runs are available for different values of BCR
and for each of 6 subgrade conditions, giving 450
comparisons. The response measures of vertical subgrade
strain and bulk stress in the base aggregate were compared
between these comparison runs to assess TBRS and TBRB.
Values greater than one indicate a remaining TBR for the
assumed BCR. Values less than one indicate that
the assumed BCR resulted in a reduction of life of the
reinforced pavement as compared to the reinforced section.

Variation of Parameters for Geosynthetic-reinforced


Models
The anisotropic linear-elastic model for the geosynthetic
permits the variation of parameters related to physical
material properties and material characteristics related to
the type and structure of the geosynthetic. These
FIGURE 2 TBRS versus base thickness and AC thickness for S4 cases. parameters include the in-plane elastic modulus, the ratio
40 S.W. PERKINS AND M.Q. EDENS

TABLE II Parameters for geosynthetic material model to examine TABLE III Pavement sections used to evaluate effect of reinforcement
effect of reinforcement modulus modulus

Exm Em En vxm2m vxm2n Gxm2m Gxm2n Pavement section Subgrade cases Geosynthetic cases
Case name (MPa) vm2n =Gm2n (MPa)
AC1B1 S1, S2, S3, S4, GM1, GM2, GM3, GM4,
GM1 400 0.5 133 S5 GM5
GM2 1000 0.5 333 AC1B2
GM3 2000 0.5 667 AC2B1
GM4 4000 0.5 1333 AC3B4
GM5 8000 0.5 2667 AC4B4
AC1B5
m, machine direction; xm, cross-machine direction; n, direction normal to material AC2B5
plane. AC5B5
AC3B6
AC4B6
between the elastic modulus in the two principal directions
of the material, in-plane Poissons ratio and in-plane shear
modulus. Ten pavement sections were analyzed with the case
The effect of modulus was first examined by using the name for the AC and base thickness given in Table III.
geosynthetic-reinforced FE model with isotropic linear- These cases were analyzed for the subgrade cases and
elastic material properties for the geosynthetic. Compari- geosynthetic material property cases shown in Table III,
son of results to those from similar cases with perfect resulting in 250 additional model runs. The three groups of
reinforcement allowed for an assessment of the influence pavement sections distinguished in Table III correspond to
of modulus on reinforcement benefit. Additional cases relatively thin, medium and thick sections. These cases
were created where an anisotropic linear-elastic model were compared to identical cases with perfect reinforce-
was used to examine the effect of the ratio of elastic ment. As discussed in a later section, results were run in
modulus in the two principal directions of the material. these groups in order to provide a means of modifying
Further cases were created to examine the effect of general trends in the data seen from the model runs for
Poissons ratio and shear modulus. perfect reinforcement.

Parameters to Examine Effect of Reinforcement


Parameters to Examine Effect of Reinforcement Modulus Anisotropy
Modulus
The geosynthetic-reinforced FE model was used to
The geosynthetic-reinforced FE model was used to examine the influence of direction dependent geosynthetic
examine the influence of the isotropic value of elastic elastic modulus values on reinforcement benefit. Different
modulus of the geosynthetic on reinforcement benefit. The values of elastic modulus were used in the machine and
material model for the geosynthetic was given linear- cross-machine directions. Values in the direction normal
elastic properties that represented an isotropic material. to the geosynthetic were also changed to provide stability
In particular, the three elastic moduli were given the same of the elastic matrix. Five different values of geosynthetic
value, the three Poissons ratios were given the same value modulus ratio, defined as the ratio of the minimum to
and the shear modulus was computed from the elastic maximum elastic modulus for the two principal in-plane
modulus and Poissons ratio as for an isotropic material. directions of the material, were used. These ratios were
Five different values of geosynthetic isotropic elastic used for two different values of maximum elastic
modulus were used. Table II gives the values for each of modulus. Table IV gives the values for each of the nine
the nine elastic constants contained in the material model elastic constants contained in the material model for the
for the geosynthetic. geosynthetic.

TABLE IV Parameters for geosynthetic material model to examine effect of reinforcement modulus anisotropy

Exm Geosynthetic modulus ratio Gxm2m Gxm2n Gm2n


Case name (MPa) Em(MPa) En(MPa) (GMR) vxm2m vxm2n vm2n (MPa)

GA1 400 400 400 1 0.5 133


GA2 400 300 300 0.75 0.5 133
GA3 400 200 200 0.5 0.5 133
GA4 400 100 100 0.25 0.5 133
GA5 400 40 40 0.1 0.5 133
GA6 8000 8000 8000 1 0.5 2667
GA7 8000 6000 6000 0.75 0.5 2667
GA8 8000 4000 4000 0.5 0.5 2667
GA9 8000 2000 2000 0.25 0.5 2667
GA10 8000 800 800 0.1 0.5 2667
DESIGN MODEL FOR REINFORCED PAVEMENTS 41

TABLE V Pavement sections used to evaluate effect of reinforcement TABLE VII Pavement sections used to evaluate effect of reinforcement
modulus anisotropy Poissons ratio and shear modulus

Pavement section Geosynthetic cases Pavement section Geosynthetic cases


AC2 B5 S1 GA1GA10 AC2B5 S1 GvG1 2 GvG8
AC1 B2 S3 AC1B2 S3
AC5 B6 S5 AC5B6 S5

Three pavement sections were analyzed with the case described above, through which a design model is
name for the AC, base thickness and subgrade type given developed.
in Table V. These cases were analyzed for the geosynthetic
material property cases shown in Table IV, resulting in 30
Design Equations for Perfect Reinforcement
additional model runs.
Equations for TBR for Perfect Reinforcement
Parameters to Examine Effect of Poissons Ratio and The illustrative results presented earlier in Figs. 2 and 3 for
Shear Modulus perfect reinforced cases suggested a relationship between
TBRS and combinations of AC and base aggregate
All FE model cases previously examined used values for
thickness through the use of Eq. (1). Equation (1) was used
Poissons ratio and shear modulus that would produce
to compute SN for the 180 perfect-reinforced cases
optimal values of reinforcement benefit with respect to
specified in Table I and plotted against TBRS in Fig. 4 for
these two parameters. This was done such that the effects
different subgrade types. In Fig. 4, TBRS is labeled as
of reinforcement isotropic modulus and reinforcement
TBRS PR to denote values pertaining to the case of perfect
modulus anisotropy could be systematically accounted for.
reinforcement. While some scatter in the results is seen, a
Additional FE model cases were created to examine the
reasonable trend of TBRS with SN between different
effect of reducing Poissons ratio and shear modulus to
subgrade types is observed. The results indicate that
their lowest possible values. Table VI shows values of the
reinforcement benefit decreases as subgrade CBR
elastic properties used for the geosynthetic material model
increases, with very little improvement for a subgrade
to examine the effect of reducing Poissons ratio and shear
CBR of 8 and essentially no improvement for a subgrade
modulus. These properties were used for each of the three
CBR of 15. Reinforcement benefit is seen to increase from
pavement sections listed in Table VII, resulting in an
low values of SN and reaches a maximum around a SN of
additional 24 cases created to examine these effects.
3.3 and then decreases with no benefit seen after a SN of 6.
All geosynthetic-reinforced cases described above used
General agreement is seen between the trends shown in
equivalenced nodes between the geosynthetic and the
Fig. 4 and those observed from currently available
surrounding base aggregate and subgrade and thereby
experimental test sections as summarized in Berg et al.
simulates a tied contact interface. As such, the effect of
(2000). The high values of TBR noted in Fig. 4 are
interface properties is not addressed in the above cases.
expected since these cases represent a geosynthetic
The influence of interface properties is addressed by
material with an infinite modulus and an infinitely stiff
comparison of design model predictions to test section
shear interface between the geosynthetic and the base
results in a later section.
aggregate. Reductions for actual geosynthetic properties
are introduced later.
Straight-line equations in the semi-log plot of TBRS PR
DESIGN MODEL DEVELOPMENT versus SN were used to approximate the trend in the data
for each subgrade CBR. For a subgrade CBR of 15,
The purpose of this section is to summarize, interpret and TBRS PR was taken as equal to one for all values of SN.
synthesize results from the parametric study cases Equations (2) (7) describe the straight-line portion of

TABLE VI Parameters for geosynthetic material model to examine effect of reinforcement Poissons ratio and shear modulus

Case name Exm (MPa) Em (MPa) En (MPa) vxm2m vxm2n vm2n Gxm2m (MPa) Gxm2n (MPa) Gm2n (MPa)
GvG1 400 400 400 0.5 0.5 0.5 133 133 133
GvG2 8000 8000 8000 0.5 0.5 0.5 2667 2667 2667
GvG3 400 400 400 0 0 0 133 133 133
GvG4 8000 8000 8000 0 0 0 2667 2667 2667
GvG5 400 400 400 0.5 0.5 0.5 0 0 0
GvG6 8000 8000 8000 0.5 0.5 0.5 0 0 0
GvG7 400 400 400 0 0 0 0 0 0
GvG8 8000 8000 8000 0 0 0 0 0 0
42 S.W. PERKINS AND M.Q. EDENS

FIGURE 4 TBRS PR versus SN for all perfect reinforcement model cases.

each curve, where the parameters used in these equations A similar assessment of the data from the 180 perfect
are identified in Fig. 5. reinforcement cases was made to relate TBRB for the case
of perfect-reinforcement to AC and base thickness and
1:6 # SN # 3:3 TBRS PR 10m1 SNlog yo1 2 subgrade CBR. In this case, changes in TBRB PR were
most appreciable for changes in subgrade CBR and AC
log yy21
m1 3 thickness. Changes in base thickness were not seen
1:7 to appreciably affect TBRB PR. Fig. 6 shows values of
y1 TBRB PR versus AC thickness for various subgrade CBR
yo1 1:6 m 4
10 1 values. Data from a given subgrade CBR and a given AC
thickness is plotted for each base thickness with the
3:3 # SN # 6:0 TBRS PR 10m2 SNlog yo2 5 exception of the base thickness of 600 mm, which
2log y2 produced comparatively high values of TBRB PR. Those
m2 6 results appear to be inconsistent with the other data and no
SNTBRS PR 1 2 3:3
experimental data is available in the literature that
y2 supports this observation. As will be seen below, the
yo2 7
103:3 m2 design equations formulated from the data in Fig. 6
excludes these results and thereby avoids what would most
With Eqs. (2) (7) used to describe the lines seen in likely be unconservative estimates of benefit. The lines in
Fig. 4, values of y1 and y2 were then related to the subgrade Fig. 6 have been extrapolated for values of AC thickness
CBR with the following expressions derived. The lines less than 50 mm. Since minimum AC thickness for paved
shown in Fig. 4 are plotted according to Eqs. (2) (9). roads are generally greater than 50 mm, most designs will
not utilize this portion of the curve.
y1 10log 82log CBR 8

y2 10log 10021:7 log CBR 9

FIGURE 6 TBRB PR versus AC thickness (D1) for perfect-


FIGURE 5 Identification of constants for Eqs. (2)(9). reinforcement model cases.
DESIGN MODEL FOR REINFORCED PAVEMENTS 43

More scatter is observed in Fig. 6 for results of thickness reduction is used. The new base layer
TBRB PR as compared to those observed in Fig. 4 for coefficient is then used to determine the base course
TBRS PR. The values of TBRB PR are, however, relatively thickness reduction such that equal life or ESALs is
modest in comparison to TBRS PR. Due to this scatter, predicted for the reinforced and unreinforced pavement
relatively simple and conservative estimates of the data (i.e. BCR when TBR 1). The method can also be
were developed by straight-lines in an arithmetic plot of used to calculate the remaining BCR for an assumed
TBRB PR versus AC thickness for various subgrade TBR, or to calculate the remaining TBR for an assumed
CBRs. The equation of these lines is given by: BCR.
An unreinforced pavement design cross section is first
D1 assumed. The AC and base thickness and layer
TBRB PR 1 2 b b 10
160 coefficients, and the subgrade CBR or resilient modulus
where D1 is the AC thickness in units of mm and b is the is required. With these parameters, the 1993 AASHTO
y-intercept of the lines. Values of b were related to pavement design equation is used to calculate the number
subgrade CBR through Eq. (11). The lines shown in Fig. 6 of ESALs for this unreinforced pavement (W182U). The
result from the use of Eqs. (10 and 11). structural number used to determine W182U is determined
from Eq. (12), where a1 and a2 are the layer coefficients of
b 4:45 CBR20:522 11 the asphalt concrete and base aggregate layers, respect-
ively, D1 and D2 are the thicknesses of the AC and base
The TBR for the total system (TBRT) is computed as the layers, respectively, and m2 is the drainage coefficient of
product of TBRS and TBRB. the base layer.

SN a1 D1 a2 D2 m2 12
Equations for BCR and TBR/BCR Combinations for A known value of TBR when BCR 0 then allows W18
Perfect Reinforcement to be calculated from Eq. (13) for a reinforced pavement
Use of the M E model for the definition of BCR or having identical pavement layer thicknesses and proper-
combinations of TBR and BCR was illustrated in the ties. In Eq. (13), the TBR used may correspond to TBRB,
companion paper (Perkins and Edens, 2002) for a TBRS or TBRT, where TBRT is computed as the product
particular perfect-reinforced case (i.e. case AC4 B5 of TBRB and TBRS.
S2). In this example, response measures from FE models
of an unreinforced pavement and several reinforced W 182R W 182U TBR 13
pavements of reduced base course thickness could be used
to assess TBR. Reduction of base course thickness beyond
The value of W182R from Eq. (13) is then used in
a certain limit produced TBR values less than 1. Such an
the AASHTO pavement design equation to calculate a
approach allowed for the generation of a curve describing
value of SN for the reinforced pavement, with this
combinations of TBR and BCR between the limits of
value used in Eq. (12) to determine a new value of a2
1 # TBR # TBRBCR0 and 0 # BCR # BCRTBR1 for a
for the reinforced pavement, with all other parameters
particular unreinforced pavement design configuration.
being equal to those used for the unreinforced
While this method appears to have merit, the introduction
pavement. This approach assumes that the TBR seen
of a new variable (reinforced base course thickness) made
for the comparative reinforced and unreinforced
it difficult to develop regression equations that reasonably
pavements can be accounted for by an increased
described the results obtained. A simple method was
structural base layer coefficient and requires an iterative
therefore, developed for relating combinations of BCR and
solution for SN in the AASHTO design equation. Once
TBR to pavement design parameters. The method is based
a value of a2 for the reinforced pavement is obtained
on existing pavement design principles and methodologies
(a22R), this value is used in Eq. (12) along with a
and is validated by comparison to results obtained from the
reduced base thickness for the reinforced pavement
perfect-reinforced parametric study cases for which
(D22R) such that a SN identical to the unreinforced
combined values of BCR and TBR are available.
pavement is obtained. The reduced base thickness is
The method relies upon the use of design equations
calculated from
contained in the AASHTO93 pavement design guide
(AASHTO, 1993) and relies upon knowledge of a
SNU 2 a1 D1
TBR for equivalent unreinforced and reinforced sections D22R : 14
(i.e. TBR for BCR 0). The AASHTO93 pavement a22R m2
design equation is then used to calculate an increased
structural layer coefficient for the base course aggregate This allows for the determination of a value of BCR
for the reinforced section that gives the increased when TBR 1 from Eq. (15). Depending on the value of
number of Equivalent Single Axle Loads (ESAL) that TBR used in Eq. (13) (TBRB, TBRS or TBRT), the
this section can carry for the case where no base course computed BCR then pertains to reinforcement effects on
44 S.W. PERKINS AND M.Q. EDENS

the base, subgrade or total system. the parametric study for the case of perfect reinforce-
ment where combinations of TBR and BCR could be
D22U 2 D22R simply computed. For these cases, a BCR was known
BCR 100 15
D22U from the thickness selected for the reinforced FE
response model. TBR was then determined from the
With this approach, values of TBR when BCR 0 damage models using the stress and strain response
and BCR when TBR 1 are known. If a value of TBR parameters from the reinforced and unreinforced
less than TBRBCR0 is used, it should be expected that sections. The method described above, using the
the remaining benefit can be expressed in terms of a AASHTO93 pavement design equation, was then
BCR. Conversely, if a value of BCR less than BCRTBR1 used to assess remaining TBR for the corresponding
is used, remaining benefit can be expressed in terms of a BCR. The value of TBRBCR0 used in the method
TBR. corresponded to that which was obtained from the FE
The same general approach described above is used to model comparison cases of unreinforced and perfect
determine combinations of BCR and TBR. Each approach reinforced sections of equivalent thickness. In many
uses the predetermined value of TBRBCR0 to determine cases, both the M E model and the AASHTO
the increased structural base layer coefficient of the equations gave remaining TBRs less than one because
reinforced pavement. Essentially, Eqs. (13 15) are the BCR was too great. For cases where remaining
worked in reverse. Thus, in the case where a desired TBR was greater than one, values of remaining TBR
BCR is specified, the reinforced base layer thickness is from the M E model were plotted against those from
determined from Eq. (16). the AASHTO equations given above and are shown in
  Fig. 7 for all subgrade cases except case S1, which
BCR
D22R 1 2 D22U 16 plotted above the line of equality.
100
The results seen in Fig. 7 indicate that the use of the
AASHTO equations is conservative with respect to that
In this case the structural number of the reinforced
obtained from the M E model for all subgrade cases
pavement is then determined from Eq. (17).
except for the CBR 0:5 (S1) case. In this case, the
SNR a1 D1 a22R D22R m2 ; 17 AASHTO equations gave unconservative estimates of
remaining TBR. Due to the apparent unconservative
where a22R is determined from the method described nature of this method for CBR 0:5; and the decreased
above and by knowing the value of TBRBCR0 : A value of margin of safety when designing for a BCR at low values
W182R is then determined using the AASHTO design of subgrade CBR, it is recommended that the method
equation using the value of SN from Eq. (17). A value of described above for calculating BCR or TBR/BCR
remaining TBR is then computed by Eq. (18), where this combinations be used only for designs with a subgrade
value should lie between 1 and TBRBCR0 ; provided the CBR $ 1:
desired value of BCR used was less than or equal to
BCRTBR1 :

W 182R
TBR 18
W 182U

For the case where a desired TBR is specified, Eq. (13)


is used to calculate the number of ESALs required of the
reinforced section. The AASHTO design equation is then
solved iteratively to give the required SN for the reinforced
section (SNR). Equation (19) is then used to determine the
base layer thickness necessary to produce this SN, where
the reinforced structural base layer coefficient is used in
this equation.
SNR 2 a1 D1
D22R 19
a22R m2

The remaining BCR is then determined from Eq. (15),


where the value obtained should lie between 0 and
BCRTBR1 ; provided the desired value of TBR was less
than TBRBCR0 :
To evaluate the reasonableness of this approach, the
above method was compared to the 450 cases from FIGURE 7 Remaining TBRs from ME model and AASHTO method.
DESIGN MODEL FOR REINFORCED PAVEMENTS 45

Modification of TBRPR for Influence of Geosynthetic


Isotropic Elastic Modulus
The methods described above provide values of
reinforcement benefit (TBR, BCR or combinations of
the two) for the case of perfect reinforcement. The 250
analysis cases presented in Table III allow for modifi-
cations to be made to benefit values from the case of
perfect reinforcement to account for values of isotropic
elastic (tensile) modulus of the geosynthetic. In particular,
separate methods for modification of TBRS PR and
TBRB PR are provided.
FIGURE 9 Reduction factor for geosynthetic isotropic elastic modulus
for SN 1:6:
Modification of TBRS PR
The equations developed for TBRS PR Eqs. (2) (9)
describing the lines shown in Fig. 4 indicate three
show a strong dependency of RFGM on geosynthetic
critical values of SN for the breakpoints of these curves,
isotropic modulus but show little effect from variations in
namely SN values of 1.6, 3.3 and 6. The analysis cases
subgrade CBR. A regression equation fitting the data shown
given in Table III were chosen to provide values of
in Fig. 9 yields the following expression for RFGM, where
TBRS for values of SN approximately equal to 1.6, 3.3
EG is the isotropic elastic tensile modulus of the
and 6. Three to four pavement design cross-sections at
geosynthetic in units of kPa.
each value of SN were examined in order to provide
redundancy.
The results indicated that the curves presented in Fig. 4 RFGM 49:8 logEG 2 261 21
tended to shift as shown in Fig. 8, with the degree of
shift being dependent on the geosynthetic modulus and Equations (20 and (21) are then used to modify Eq. (8)
the subgrade CBR. For each case analyzed, a reduction by recognizing that y1 in Eq. (8) is equal to TBRS PR in
factor for geosynthetic isotropic elastic modulus (RFGM) Eq. (20) (see Fig. 5) and that the value of y1 when
was computed according to Eq. (20), which describes the geosynthetic isotropic modulus is accounted for ( y12GM)
reduction in TBRS for the case of a specific geosynthetic is equal to TBRS GM in Eq. (20).
isotropic modulus (TBRS GM) as compared to that
obtained for the case of perfect reinforcement (TBRS PR). 49:8logEG 2 261 log82logCBR
y12GM 10 2 1 1 22
100
TBRS GM 2 1
RFGM 100 20
TBRS PR 2 1 Cases in Table III with a SN of approximately 3.3 were
used to develop an expression for y2 that accounts for
Separate expressions were developed for RFGM for each geosynthetic isotropic elastic modulus (y22GM). Fig. 10
of the three values of SN. For case AC1 B1 with a SN of shows values of RFGM calculated from Eq. (20) for cases
1.61, Fig. 9 shows values of RFGM for all of the subgrade with a SN of approximately 3.3 for all of the subgrade and
and geosynthetic isotropic modulus values. The results geosynthetic isotropic modulus values, where it is seen

FIGURE 10 Reduction factor for geosynthetic isotropic elastic


FIGURE 8 Influence of geosynthetic isotropic modulus on TBRS PR. modulus for SN 3:3:
46 S.W. PERKINS AND M.Q. EDENS

that RFGM is dependent on both geosynthetic modulus and CBR and reinforcement isotropic elastic modulus.
subgrade CBR. Regression equations were developed to

describe the data seen in Fig. 10 where the lines drawn in 55:8logEG 2288  D1
this figure were according to Eq. (23). TBRBGM 124:45CBR20:522
100 160

4:45CBR20:522 21 1 27
RFGM 1:8 1026 logCBR 1:57 1026  EG 23

Equations (20)and (23) are then used to modify Eq. (9)


by recognizing that y2 in Eq. (9) is equal to TBRS PR in Modification of TBRGM for Influence of Geosynthetic
Eq. (20) (see Fig. 5) and that the value of y2 when Modulus Anisotropy
geosynthetic isotropic modulus is accounted for ( y22GM)
is equal to TBRS GM in Eq. (20). Effects of anisotropy of the elastic modulus of the
geosynthetic between the machine and cross-machine
directions were evaluated by comparison of values of TBR
1:81026 logCBR1:571026
y22GM obtained from the model cases listed in Tables IV and V to
100
corresponding cases with isotropic modulus values. A
EG 10log10021:7logCBR 211 24 reduction factor for geosynthetic modulus ratio (RFGMR)
was defined according to Eq. (28) and evaluated for the
finite element model cases listed in Table V. Reduction
The effect of reinforcement isotropic elastic modulus factors were evaluated for corresponding TBR values for
on values of SNTBR1 were evaluated by first using subgrade and base reinforcement effects.
Eq. (24) to calculate the value of y22GM for each value
of isotropic elastic modulus and subgrade CBR given in TBRGMR
Table III. Each value of y22GM corresponds to TBRS2GM RFGMR 28
TBRGM
for a SN 3:3: For the corresponding cases for
AC5 B5 where SN 4:57; a value of TBRS GM is
For values of TBRS, values of RFGMR were found
known. Based on these two points, a straight line in a
to be approximated by straight-line relationships given by
semi-log plot of TBRS GM versus SN is used to estimate
Eq. (29), where m is the slope of the lines and was seen to
a value for SN when TBRS GM 1: Regression
be a function of TBRS GM, as given by Eq. (30) and GMR
equations between these values and EG and subgrade
is the ratio of moduli and is given by Eq. (31).
CBR were then developed, with Eq. (25) being the
result. RFGMR m GMR 1 2 m 29
SNTBRS GM 1 2:54CBR 2 0:280:028 EG 0:0504 25
m 0:62 log TBRS GM 30
Equations (22), (24) and (25) are then used in
Eqs. (2) (7) for values of y1, y2 and SNTBRS PR1 ;
respectively. Equations (2 7) then provide predictions of EG2min
GMR 31
TBR accounting for the influence of the structural section EG2max
thickness, subgrade strength and geosynthetic isotropic
elastic modulus. Combination of Eqs. (28 30) yields Eq. (32).

TBRS GMR TBRS GM 0:62 log TBRS GM GMR


Modification of TBRB PR
Reduction factors were also developed to modify values of 1 2 0:62 log TBRS GM  32
TBRB PR to account for values of geosynthetic isotropic
elastic modulus. Evaluation of all cases contained in A similar analysis for values of TBRB showed that the
Table III indicated that the reduction factor was dependent reduction factor was dependent primarily on GMR, which
most strongly on the reinforcement isotropic modulus. allowed Eq. (33) to be developed.
The reduction factor developed, having the same
general definition as that given in Eq. (20), is given by TBRB GMR TBRB GM 0:2 GMR 0:8 33
Eq. (26).

RFGM 55:8 log EG 2 288 26 Evaluation of TBR S GM and TBR B GM from


Eqs. (2 7), (22), (24), (25) and 27) for the case of
Combination of Eqs. (10), (11), (20) and (26) yields isotropic geosynthetic modulus values allows these values
Eq. (27) for determination of TBRB GM accounting for to be reduced by the use of Eqs. (32) and (33) when GMR
the influences of structural section thickness, subgrade from Eq. (31) is evaluated.
DESIGN MODEL FOR REINFORCED PAVEMENTS 47

TABLE VIII Reduction factors for reinforcement Poissons ratio and Modification of TBRGM for Influence of Interface
shear modulus Properties
RFGv RFGG RFGvG The computation of TBR accounting for all the effects
TBRS 0.83 0.92 0.75 described above implicitly assumes optimal interface
TBRB 0.96 0.99 0.95 properties. A reduction factor for interface properties is
introduced to account for the influence of interface shear
properties as given in Eq. (37), where TBRI is the TBR
when interface reduction effects are accounted for.
Modification of TBRGM for Geosynthetic Poissons Empirical calibration of this reduction factor is accom-
Ratio and Shear Modulus plished for two different types of geosynthetics by
comparison of predictions of the method to results from
The finite element model cases presented in Tables VI large scale test sections reported by Perkins (1999) and is
and VII were used to examine the effect of reducing described in the next section.
Poissons ratio and shear modulus to zero from the values
used for the case of isotropic material properties. TBRI
RFGI 37
A reduction factor was computed for the effect of TBRGM
reducing each variable according to Eqs. (34) and (35),
where TBRv and TBRG were values of TBR obtained
directly from the cases examined where either v or G was Application of Reduction Factors
reduced to zero and TBRGM was the TBR for the
corresponding model case where isotropic elastic values The reduction factors developed to describe the reduction
were used for the geosynthetic material. A reduction of reinforcement benefit for actual geosynthetic properties
factor was also computed for the combined effect of from the case of perfect reinforcement are designed to
reducing both v and G according to Eq. (36). Reduction be applied in series. Benefit values for TBRS GM and
factors for both values of TBRS and TBRB were evaluated. TBRB GM are determined from Eqs. (2 7,22,24,25
and 27) and account for geosynthetic isotropic
TBRv elastic modulus. These benefit values are then used in
RFGv 34 Eqs. (32 and 33) to account for geosynthetic modulus
TBRGM
anisotropy. Values of TBRS GMR and TBRB GMR from
TBRG Eqs. (32 and 33) are then multiplied by RFGv, RFGG and
RFGG 35
TBRGM RFGI as appropriate for the geosynthetic material used.
Values of TBR for the total system (TBRT) are still
TBRvG
RFGvG 36 computed as the product of the final values of TBRS and
TBRGM
TBRB and are used in the method described above for
evaluation of BCR or TBR/BCR combinations.
Table VIII summarizes average values of reduc-
tion factors obtained for each reinforcement effect
(i.e. base and subgrade). From these results, it appears
Calibration of Design Model
that the combined effect of reducing both Poissons ratio
and shear modulus to zero corresponds to the product of The model described above was calibrated by compari-
the two reduction factors for each individual effect. Based son to results from four large-scale reinforced pavement
on these results, reduction factors can be used either test sections reported by Perkins (1999). Pertinent design
individually or in series and applied to TBRS GM and model input properties for these sections are listed in
TBRB GM for reduction of these TBRs according to Table IX and include AC and base aggregate thickness
Eqs. (34) (36). and subgrade CBR. Three geosynthetics were used in

TABLE IX Summary input parameters for design model and test section comparisons (Perkins, 1999)

Test section CS5 CS6 CS10 CS11


AC thickness, D1 (mm) 75 75 75 75
Base thickness, D2 (mm) 300 300 375 300
Subgrade CBR 1.5 1.5 1.5 1.5
Geosynthetic Geogrid B Geotextile A Geogrid A Geogrid A
2% secant modulus, GSM22% (kN/m) 595 680 425 425
Geosynthetic modulus ratio, GMR 0.615 0.313 0.595 0.595
Reduction factor for interface shear, RFI 1 0.765 1 1
RF for Poissons ratio, RFGv, on TBRS 1 0.83 1 1
RF for Poissons ratio, RFGv, on TBRB 1 0.96 1 1
RF for shear modulus, RFGG, on TBRS 1 0.92 1 1
RF for shear modulus, RFGG, on TBRB 1 0.99 1 1
48 S.W. PERKINS AND M.Q. EDENS

TABLE X Summary output for design model and test section comparisons (Perkins, 1999)

Test section CS5 CS6 CS10 CS11


TBRBCR=0 Test section values Subgrade 2.95 1.66 2.23 2.61
Base 7.74 1.43 2.63 3.01
Total 29.7 2.90 9.82 13.5
Design model predictions Subgrade 3.02 1.66 2.45 2.54
Base 1.82 1.27 1.71 1.71
Total 5.51 2.10 4.19 4.34

these sections. Geogrid A and B were punched and are listed in Table X.
drawn biaxial products. Geotextile A was a woven slit
film product. Values of secant tensile modulus at 2% EG 4400 GSM22% 38
axial strain from wide-width tension tests per ASTM
4595 (ASTM, 2001) were provided by the manufac- Results from test section CS10 were then compared to
turers (AMOCO, 2001; Tensar, 2001) for both model predictions. Test section CS10 also used geogrid A
the machine and cross-machine directions of the and therefore the same values for this material were used
material. The maximum of the two values for each as those used for test section CS11. The thickness of the
product is reported in Table IX as GSM22%. The ratio base layer was increased to 375 mm. The model predicted
between the minimum and maximum values is reported a value of TBRS of 2.45 while a value from the test section
in Table IX as GMR. Values of TBRB, TBRS and TBRT of 2.23 was observed.
are available for each reinforced test section and are Results from test section CS6 were then used to
listed in Table X. calibrate a reduction factor for interface shear resistance.
Since suitable test methods for the definition of Equation (38) was used to calculate EG for use in the
geosynthetic Poissons ratio and shear modulus are not design model. The reduction factor for interface shear was
currently available, selection of reduction factors for reduced from 1.0 until a value for TBRS of 1.66 was
these two properties corresponded to either no reduction obtained. This resulted in a value for RFI of 0.765, which
(i.e. reduction factors of 1.0) or the full values given in is taken as also being applicable to values of TBRB.
Table VIII. For geogrid A and B, available experimental Table X lists test section and design model values of
results indicate that geosynthetic reinforcement products TBRB, TBRS and TBRT for the four sections used for
of this type tend to provide benefit values at the higher calibration of the design model. The method used for
range of those available for test sections reported in the calibration of the design model results in good predictions
literature (Berg et al., 2000), while benefit values for of TBRS. The predicted values of TBRB are considerably
geosynthetics similar to geotextile A tend to be at the less and conservative in comparison to those seen from the
lower range. Reduction factors of 1.0 for Poissons ratio test sections. This is due to the inability of the material
and shear modulus were therefore used for geogrid A model for the base aggregate to properly show the effects
and B while the full values given in Table VIII were of increases in bulk stress on base layer behavior and the
used for geotextile A, as displayed in Table IX. In conservative steps taken in development of expressions for
addition, the reduction factor for interface shear was set TBRB. The comparatively low and conservative predic-
a priori to 1.0 for geogrid A and B. The value used for tions of TBRT are due to the conservative values predicted
geotextile A arises from the calibration process for TBRB and the conservative procedure for multipli-
described below. cation of TBRB and TBRS to produce TBRT. This
The geosynthetic elastic modulus (EG) used in the multiplication procedure appears to be conservative based
development of the design method is a true elastic on the values from test sections where it is seen that
modulus with units of kPa. The 2% secant tensile modulus multiplication of TBRB and TBRS produce values less
(GSM22%) differs from EG in that it has units of kN/m, than TBRT determined experimentally.
meaning that a material thickness must be assumed to
convert GSM22% to EG. In addition, GSM22% is a wide-
width secant modulus calculated at a relatively high strain SUMMARY AND MODEL LIMITATIONS
of 2%, whereas, EG is strictly a low-strain initial elastic
modulus. Calibration of a conversion factor between A M E model for reinforced pavements has been
GSM22% and EG was, therefore, necessary. This was developed and used in a parametric study to develop a
accomplished by the use of test sections CS5 and CS11, design model expressed in terms of generic input
which used geogrid B and A, respectively. To match as parameters describing the geometry and properties of the
closely as possible the values of TBR for the subgrade for reinforced pavement cross section. The model is expressed
test sections CS5 and CS11, a factor of 4400 was applied solely in terms of equations relating reinforcement benefit
to GSM22% according to Eq. (38). Resulting values of TBR to pavement and geosynthetic input design parameters.
DESIGN MODEL FOR REINFORCED PAVEMENTS 49

These equations are based on the results of predictions of load can be approximated in terms of a standard axle wheel
reinforcement benefit from a finite element based M E load. The model may not be appropriate for either
model from over 465 analyzed cases. The M E model and lightly loaded traffic or for excessive heavily loaded traffic.
the design model were calibrated from reinforced test The design model is calibrated largely from results
section results. based on a permanent pavement surface deformation of
The design model has been shown to be sensitive to 12.5 mm. It is anticipated that roadways that can be
pavement design variables that have previously been designed for a rut depth greater than 12.5 mm will realize
identified as having a significant impact on reinforced reinforcement benefits greater than those predicted from
pavement performance. In particular, the model is this model, meaning that this model should be
capable of showing that reinforcement benefit increases conservative for roadways designed for a rut depth in
with decreasing subgrade strength/stiffness, increasing excess of 12.5 mm.
geosynthetic tensile modulus and is sensitive to the The model is based on asphalt concrete and base
pavement structural thickness. Furthermore, the model is aggregate properties that correspond to structural layer
shown to be sensitive to the ratio of geosynthetic tensile coefficients of approximately 0.4 and 0.14, respectively.
modulus in the weak to strong directions of the material. Recommendations are provided in Perkins (2001) for
The model is calibrated from the use of geosynthetic pavement designs with different materials yielding
secant tensile modulus values determined at 2% axial different layer coefficients.
strain according to ASTM 4595 (ASTM, 2001). All cases of reinforcement were modeled by placement
The design model accounts for geosynthetic properties of the geosynthetic at the bottom of the base aggregate
pertinent to the type and structure of the material through layer. As such, the design model cannot account for
the assignment of reduction factors for geosynthetic situations where it might be desirable to place the
material Poissons ratio, shear modulus and geosynthe- geosynthetic further up in the base aggregate layer.
tic/aggregate shear interaction. Geosynthetic material Further work for improvement of the design model
Poissons ratio and shear modulus are parameters existing should focus on a better description of reinforcement
in the anisotropic elastic material model used for the benefit for reinforcement effects in the base aggregate
geosynthetic and were shown to have a moderate layer. The design model as presented in this paper yields
influence on reinforcement benefit. Reduction factors conservative predictions of this reinforcement component.
for Poissons ratio and shear modulus correspond to one Improvements in the material model used for the base
of two value sets and were calibrated for the two types of aggregate layer would be required to improve upon
geosynthetics used in test sections to which the model predictions of this component of benefit. These improve-
was compared. Similarly, the reduction factor for ments would then also warrant examining the model for
interface shearing resistance was calibrated from these predictions of benefit for cases where the geosynthetic is
test section results. moved up into the base aggregate layer. Further work
The design model appears to yield mostly conservative should also be performed to allow assignment of reduction
predictions of reinforcement benefit when compared to factors for geosynthetic Poissons ratio, shear modulus and
results available in the literature. Conservatism in the geosynthetic/aggregate shear interaction for various
model results from several sources. The method used to geosynthetics types. Finally, future results from new test
calculate reinforcement benefit for the base aggregate sections should be used to further refine and update the
layer was shown to yield values of benefit that are model.
consistently less than those determined for test sections.
The equation used to calculate total TBR from the partial
Acknowledgements
values for the base and subgrade layers was shown to be
conservative as compared to test section results. Finally The authors gratefully recognize the financial support
the method used to calculate BCR and TBR/BCR and technical review provided by the Montana, Idaho,
combinations was shown to be conservative with respect Kansas, Minnesota, New York, Texas, Wisconsin and
to that predicted by the finite element model. In spite of Wyoming Departments of Transportation and the Western
this inherent conservatism, use of the model has been Transportation Institute at Montana State University.
shown to offer significant life-cycle cost benefits for The Amoco Fabrics and Fibers Company and Tensar Earth
several pavement design examples (Perkins, 2001). Technologies, Incorporated graciously donated geo-
The design model is based on certain conditions to synthetic materials for preceding projects leading up to
which the model should be limited. The model is based on this work and also provided geosynthetic product
results from a finite element model where pavement load information used in the development of the design model.
has been simulated by the application of a 40 kN load
applied over a 300 mm diameter circular plate. Experi-
mental test section results from which the model has been References
validated used a similar load. The design model is,
AASHTO (1993) AASHTO Guide for Design of Pavement Structures
therefore, designed to describe reinforcement benefit for (American Association of State Highway and Transportation
conventional roadway applications where the pavement Officials, Washington, DC).
50 S.W. PERKINS AND M.Q. EDENS

AMOCO (2001) Personal email communication, 4-23-2001. (Montana Department of Transportation, Helena, MT), Report No.
ASTM (2001) Annual Book of ASTM Standards, Soil and Rock (II): FHWA/MT-99/8106-1, p 140.
D4943-Latest; Geosynthetics, 04.09, Section 4 (American Society of Perkins, S.W. (2001) Mechanistic-Empirical Modeling and Design
Testing and Materials, West Conshohocen, PA). Model Development of Geosynthetic Reinforced Flexible Pavements:
Berg, R.R., Christopher, B.R. and Perkins, S.W. (2000) Geosynthetic Final Report (Montana Department of Transportation, Helena, MT),
Reinforcement of the Aggregate Base Course of Flexible Pavement Report No. FHWA/MT-01/002/99160-1A, p 156.
Structures, GMA White Paper II (Geosynthetic Materials Associa- Perkins, S.W. and Edens, M.Q. (2003) Finite element and distress
tion, Roseville, MN), p 130. models for geosynthetic-reinforced pavements, International
Perkins, S.W. (1999) Geosynthetic Reinforcement of Flexible Journal of Pavement Engineering 3(4), 239 250.
Pavements: Laboratory Based Pavement Test Sections Tensar (2001) Personal fax communication, 4-23-2001, p 37.
Copyright of International Journal of Pavement Engineering is the property of Taylor & Francis Ltd and its
content may not be copied or emailed to multiple sites or posted to a listserv without the copyright holder's
express written permission. However, users may print, download, or email articles for individual use.

You might also like