KSCE Journal of Civil Engineering (2014) 18(1):142-148 Highway Engineering

Copyright ⓒ2014 Korean Society of Civil Engineers

DOI 10.1007/s12205-014-0255-2 pISSN 1226-7988, eISSN 1976-3808
www.springer.com/12205
TECHNICAL NOTE

Critical Review of Flexible Pavement Performance Models

Ankit Gupta*, Praveen Kumar**, and Rajat Rastogi***
Received September 9, 2011/Accepted November 11, 2012

··································································································································································································································

Abstract

Maintenance and repair of the highway network system are major expenses in the state budget. For this reason various concerned
organizations are pointing out the need for developing an intelligent and efficient pavement performance model that can prioritize
pavement maintenance and rehabilitation works. Such models can forecast the remaining pavement service life and pavement
rehabilitation needs, and can help in the formulation of pavement maintenance and strengthening programmes which will reduce the
road agency and road user costs. The flexible pavement performance or deterioration models involve the complex interaction
between vehicles and the environment, and the structure and surface of the pavement. Performance models relating to the pavement
distress conditions like, cracking, raveling, potholing, and roughness are analyzed and developed by various researchers. But most of
these models are found applicable to a particular set of traffic or environment conditions, thus highlighting the need of model(s) that
can work in varied conditions satisfactorily. The paper presents a detailed review of various pavement performance models to
examine the role of factors related to pavement materials, environmental conditions, type of traffic and volume of traffic, and to
identify the limitations and gaps in the present knowledge on such models.
Keywords: pavement performance models, flexible pavements, age, traffic, Equivalent Single Axle Load (ESAL), environment
··································································································································································································································

1. Introduction Many models have been developed by various researchers for

modeling pavement performance, but before going into the
Pavement is a very sophisticated physical structure that responds details of these models, the factors that are found affecting the
in a complex manner to the external traffic loading and performance of the pavements, are listed in a categorized form.
environmental conditions. This is mainly due to the heterogeneous Traffic Loading Associated Factors: These include factors
composition of the asphalt mixture, aggregate and subgrade soil, associated with traffic namely, traffic volumes, axle loads,
and the vast variation in traffic and environmental characteristics number of Equivalent Single Axle Loads (ESAL’s), tire pressure,
from one region to another. A key component in making design, type of axles and their configuration, load application time and
construction, maintenance, and rehabilitation decisions for mechanism of load application.
pavements, consists of evaluation like, assessing and measuring Construction, Structural Composition and Material
surface distresses namely, cracking and rutting, or structural Properties Associated Factors: These include the main engineering
properties say, deflection and strain and forecasting the effect of properties of the materials used in pavement construction such as
such conditions on future performances. Condition forecasts are strength or bearing capacity of soil subgrade layer, gradation of
generated with performance models, which are mathematical soil and aggregates, physical properties of aggregates, aggregate-
expressions that relate condition data to a set of explanatory mix properties, elastic and resilience modulus of layers and
variables such as material properties, pavement design Poisson ratio.
characteristics, traffic loading, environmental factors and the Environmental Associated Factors: Factors like, moisture in
history of maintenance activities. soil, temperature of soil, pavement layers and surroundings, freeze
To measure and predict the performance of any pavement, a and thaw cycles, humidity and precipitation, and movement of
repeatable, well-established and field calibrated condition rating ground water, capillary water or surface water becomes the part
system need to be adopted (Shahin et al., 1984). AASHTO of environmental associated factors.
(1993) defines the pavement performance as the ability of a
pavement to satisfactorily serve traffic over time. It further 2. Pavement Performance Models
defines the serviceability of a pavement as the ability to serve the
traffic for which it was designed. A pavement performance model is an equation or relation that

*Assistant Professor, Dept. of Civil Engineering, NIT Hamirpur, Hamirpur 177005 (HP), India (Corresponding Author, E-mail: anki_ce11@yahoo.co.in)
**Professor, Dept. of Civil Engineering, IIT Roorkee, Roorkee, India (E-mail: pkaerfce@iitr.ernet.in)
***Associate Professor, Dept. of Civil Engineering, IIT Roorkee, Roorkee, India (E-mail: rajatfce@iitr.ernet.in)

− 142 −
 Critical Review of Flexible Pavement Performance Models

relates some extrinsic ‘time factor’ (age, or number of load traffic, which is manifested in various kinds of distress. HDM-4
applications) to a combination of intrinsic factors (structural roughness model contains rut depth component, traffic, strength of
responses, material properties, drainage, etc.) and performance pavement, age, etc.
indicators. Depending upon the inclusion of such attributes, the The Indiana Department of Transportation (INDOT) (Gulan et
performance models are categorized under three groups. These al., 2001) updated simple pavement performance prediction models
are: using least number of explanatory (independent) variables for
• Surface Characteristics based Models: These models mainly various pavement types to plan for future rehabilitation or
include the information on roughness, rut depth, raveling, pot- replacement needs. The study did not yield statistically strong
holes, etc. being generated as a result of traffic factors listed pavement performance prediction models more probably due to
above. Apart from these, the age of the pavement is also non-uniform construction and foundation of the test sections.
considered in the development of such models. The NCHRP 1-37A project (NCHRP, 2004) concluded that
• Models based on Environmental Factors: Such models permanent deformations, thermal cracking, and fatigue cracking
include the effect of various environmental factors, again dis- were the most dominant distresses affecting roughness. The local
cussed before, on the performance of the pavements. environmental conditions and material used in the base layer
• Pavement Performance Rating Models: These are the mod- enhanced these distresses. Models were developed for different
els, which allow defining the performance of the pavement base layer materials, namely granular, asphalt-treated, and
using certain arbitrary or weighted values, which varies cement stabilized. The roughness model for pavements with
within a certain range. Various indices are proposed by different granular base is given below:
researchers, like PSI (Present Serviceability Index), PCI (Pave-
age
ment Condition Index), PCR (Pavement Condition Rating), ---------
IRI = IRI0 + 0.0463SF⎛ e – 1⎞ + 0.00119TCL
20
etc. These are based on various characteristics of the pave- ⎝ ⎠ (1)
ment, as discussed under previous two categories of the + 0.1834COVRD + 0.0736BC + 0.00115LC
models and the formation of a composite index based on
those characteristics (Gupta et al., 2012). The Site Factor (SF) is computed using the equation:
( RSD ) ( P 0.075 + 1 ) ( PI )⎞ ⎛ ln ( FI + 1 ) ( P0.02 + 1 ) ( ln( Rm + 1 ) )⎞
3. Surface Characteristics Based Models SF = ⎛ -----------------------------------------------
-⎠ + ⎝ ---------------------------------------------------------------------------⎠
⎝ 2 × 10
4
10
(2)
The characteristics of surface are defined in terms of roughness,
rutting, cracking etc. These are discussed in the grouped form Prozzi and Madanat (2003) developed models for IRI and
below. rutting, separately. The model for IRI takes into consideration the
mix-design factors, layer thicknesses, traffic load and material
3.1 Roughness Models characteristics.
Simple parameter used to evaluate pavement performance is
pavement surface roughness, which reflects the user perspective. 3.2 Rutting Models
A rough pavement directly affects the ride quality. Any incremental Permanent deformation or rutting is a load-related distress
increase in surface distress causes surface roughness to increase. caused by cumulative applications of loads at moderate to high
For thin bituminous surfaced roads, the pavement may actually temperatures, when the asphalt concrete mixture has the lowest
fail by break up of thin surfacing before the terminal condition in stiffness. It can be divided into 3 stages. Primary rutting develops
terms of rutting has been reached. early in the service life and it is caused predominantly by
Roughness is usually expressed in terms of International densification of the mixture (compaction effort by passing
Roughness Index (IRI), which represents the cumulative vertical traffic) and with decreasing rate of plastic deformations. In the
movement of a single vehicle wheel in m/km. Quarter Car secondary stage, rutting increments are smaller at a constant rate,
Model (Sayers et al., 1986), is generally used to analyze profile and the mixture is mostly undergoing plastic shear deformations.
data to calculate the IRI for each wheel path. The tertiary stage is when shear failure occurs, and the mixture
Al-Suleiman et al. (1989) evaluated the effects of pavement flows to rupture. Usually the tertiary stage is not reached in in-
age and traffic loading on routine maintenance effectiveness. service pavements, mainly due to preventive maintenance and
Change in surface roughness was considered as a measure of carrying out of rehabilitation works. Permanent deformation is
pavement surface deterioration. Regression models were developed predicted using empirical models, which provide strain value in
to examine the effects of routine maintenance expenditure level, the pavement related to either primary or secondary stages. Some
pavement age, and traffic loading on change in surface roughness. of these models are given in Table 1.
HDM-4 (Kerali, 2000) road deterioration models include the Where, a, a1, b, b1, c, d, f, m, S, ε0, β, and ρ = positive regression
complex interaction between vehicles, the environment, and the (material) constants; m = permanent deformation parameter
pavement structure and surface. The road deterioration models representing the constant of proportionality between permanent
predict the deterioration of the pavement over time and under strain and elastic strain (i.e., permanent strain at N = 1); α = 1 – b

Vol. 18, No. 1 / January 2014 − 143 −

 Ankit Gupta, Praveen Kumar, and Rajat Rastogi

Table 1. Empirical Models Defining Rutting Behaviour

Type of Model and Source Model Rutting stage
ε p = a1 + b1 logN
Barksdale’s Semi-log model
b Primary
(Zhou et al., 1994) Or εpn = ----1 ( N > 1 )
N
Monismith et al. Power law model b
(Zhou et al., 1994)
εp = aN Primary
2 3
logεp = C0 + C1( logN ) + C2 ( logN ) + C3 ( logN )
Monismith et al. Log-log third
Three stage permanent
order polynomial For theprefailure (primaryand seconday) zone: εp = aNb
deformation
model (Zhou et al., 1994) b1 N
For the failure (teriary) zone: εp = A 1e
Kenis VESYS model –α
εpn = µ × εr N Primary
(Zhou et al., 1994)
Majidzadeh et al. Ohio State model 1–m
ε p = aN Primary
(Zhou et al., 1994)
ε
log ⎛ ----p⎞ = – 6.631 + 0.435logN + 2.76logT + 0.110logS +
Leahy Model (1989) ⎝ εr⎠ Primary
0.118logη70 + 0.930logVbeff + 0.5011logV a
Lytton et al. Superpave model
(Zhou et al., 1994)
log ε p = logεp ( 1 ) + S logN Primary

ε
Kaloush Model (2000) log⎛ ----p⎞ = – 3.15552 + 3.9937logN + 1.734logT Primary
⎝ ε r⎠

AASHTO 2002 - Witczak model ε

log ----p = logC + 0.4262logN Primary
(Zhou et al., 2004) εr
b
Zhou Model (2004) εp = εPS + c ( N – N PS ), NPS ≤ N ≤ NST, ε PS = aNPS Secondary
fN – N ST
Zhou Model (2004) εp = εST + d ( e – 1 ), N ≥ NST, εST = εP S + c ( NST – NPS ) Tertiary

= permanent deformation parameter indicating the rate of (predefined) axle load equivalencies and structural numbers and
decrease in incremental permanent deformation as the number of it incorporates the environment factor in term of thawing index.
load repetitions increases. Normally, α value is greater than 0. The estimation results show that the model replicates the
C = T2.02755 / 5615.391 is a function of temperature (°F) and A
β
pavement behavior well, that the inclusion of an environmental
ρ
= 1 / e is a constant and less than 1. variable is important to avoid biases in other parameters. It is
Total permanent deformation is the summation of rut depths also found that interactions between some parameters in the
from all layers: nonlinear specification lead to significant differences between
RDTotal = RDAC + RDBase + RDSubgrade (3) parameter estimates among the two wheel paths rutting models.
Fwa et al. (2004) have used the C–φ based approach for
This can also be written as: developing a rutting prediction model. The proposed model
N employs the widely adopted expression of power relationship as
RDT = ∑ ( ε p ) i × ∆h i (4) the basic equation for prediction of cumulative damage. The
i=1
basic expression is modified by incorporating the following
Where, effects: (1) the effect of load through a stress ratio computed
N = Number of sub layers using the C–φ concept, (2) the effect of temperature, and (3) the
(εp)i = Vertical plastic strain at mid-thickness of layer i effect of loading speed.
∆hi = Thickness of sublayer i
Das and Pandey (1999) developed a mechanistic design method 3.3 Discussion
by correlating the performance data of bituminous pavements The most important factors responsible for surface distress are
from different parts of India to the critical stress-strain parameters age of the pavement, the average annual daily traffic and the type
of the pavement composition leading to the failure of pavements. of traffic. With the regular maintenance, the roughness keeps on
Archilla and Madanat (2001) developed an empirical rutting decreasing so this should also be considered as one of the factors
progression model using experimental data. The salient features in the pavement performance models and this can be incorporated
of the model specification are that it has used conventional in the form of cost incurred during maintenance activities like

− 144 − KSCE Journal of Civil Engineering

 Critical Review of Flexible Pavement Performance Models

patching and sealing. Rutting is measured as a function of time the material is exposed to near saturation conditions, and
permanent strain in the pavement. Various types of rutting most agencies use drainage coefficient values of 1.0, relying
models are proposed by the researchers like semi-log model, mostly on the effective subgrade resilient modulus as the
power law model, exponential model, etc., which considers load climatic-sensitive input parameter.
repetition as the major cause of rut development. The most Jin et al. (1994) evaluated the seasonal variation of the resilient
common relevant finding in the empirical literature is the modulus of granular soils for flexible pavement design. Field
concave shapes of rut depth with cumulative number of load temperature and moisture content underneath the pavement were
repetitions. Such trends have been observed with heavy vehicle monitored using soil moisture-temperature cells over a period of
simulators by Maree et al. (1982) and by Harvey and Popescu one year. Regression equation was developed to determine
(2000) and in other experiments such as the AASHO Road Test resilient moduli under the site specific environment of Rhode
(HRB, 1962). Furthermore, most developed models specify such Island.
a concave shape (Lister, 1981; Paterson, 1987). A general finding NCHRP (2004) reported a mechanistic model of one dimensional
emerged was that the surface layer properties are primarily heat and moisture flow that simulates changes in the behavior
responsible for such distresses. Secondary factor includes the and characteristics of pavement and subgrade materials induced
properties and thickness of subbase, subgrade and base layers. by environmental factors. It is named as Enhanced Integrated
The research already done yields good performance prediction Climate Model (EICM). Daily and seasonal variations of
models but still they lack in some areas. Many roughness and temperature and moisture within the pavement structure are
rutting models developed have been limited to linear specifications induced using weather history at the project site.
and do not account for the effects of the environment. But now Diefenderfer et al. (2006) developed models for predicting the
many researchers like Paterson (1987), HDM-4 (Kerali, 2000), daily maximum and minimum pavement temperatures. It is
Prozzi and Madanat (2004), etc. developed nonlinear models anticipated that the models presented within this study will be of
with data from in-service pavements that included the effect of benefit to those who need to determine the pavement temperature
the environment. Statistical techniques like model estimation are profile in order to calculate in situ pavement engineering
only efficient when accounting for the correlation among the characteristics. By using ambient temperature trends from
models. historical records, these models can predict anticipated future
pavement temperatures and thus aid researchers in determining
4. Models Based on Environmental Factors the amount of time that pavements are subjected to critical
temperatures.
There are several studies in the literature about the environmental Zuo et al. (2007) observed seasonal temperature and water content
influences on the various properties of the pavement. In the variations and evaluated their effects on predicted pavement life.
AASHTO method many modifications has been done after The results of the parametric study showed that the temperature
considering the environmental factors. In the case of flexible averaging period, the temperature gradient in the asphalt, and the
pavements, three major environmental effects are of particular timing and duration of wet base and subgrade conditions all
interest: affect the estimation of pavement life.
• Temperature variations for the asphalt concrete. The dynamic Brahmajaree et al. (2009) investigated the effects of climatic
modulus of asphalt concrete mixtures is very sensitive to factors on flexible pavement behavior by evaluating seasonal
temperature. Temperature distributions in asphalt concrete variations of pavement layer properties, especially asphalt
layers are predicted and then used to define the stiffness of concrete elastic modulus and subgrade resilient modulus, which
the mixture throughout the sublayers. Temperature distribu- vary throughout a year depending on temperature and moisture
tions are also used as inputs for the thermal cracking predic- conditions. Nonlinear regression is used to characterize the
tion model. results showing that the relationship between time and material
• Moisture variation for subgrade and unbound materials. The properties could be represented by a sinusoidal pattern.
resilient modulus input of unbound materials is defined as
being at optimum density and moisture content. A correc- 4.1 Discussion
tion factor is defined to adjust the resilient modulus based on Environmental models consist of relation between the various
predicted moisture content. pavement structure properties with the environmental variables.
• Freezing and thawing for subgrade and unbound materials. A major finding among the models discussed above is that
AASHTO (1993) suggested a relation to compute the ESAL damage is the important parameter resulting due to change in
expected to be repeated during the design period. In this relation environmental conditions and this depends upon the subgrade
the drainage coefficient is related to the material’s permeability modulus and ESALs. Another significant point to be noted is
and the amount of time that the material is expected to be at or that they follow the periodical waveform pattern. All the models
near saturation conditions. Huang (1993) has given the drainage are site specific as the environmental conditions are different at
coefficients for unbound materials. However, in practice it is each and every place so calibration factors play an important role
difficult to assess the quality of drainage or the percentage of in deciding the accuracy of the model. Much of the work has

Vol. 18, No. 1 / January 2014 − 145 −

 Ankit Gupta, Praveen Kumar, and Rajat Rastogi

been done on this area but still the effect of nearby ground pavement maintenance and rehabilitation plans can be made
conditions of pavement, its moisture ingress phenomenon is not using a newly introduced performance indicator called relative
well defined. Additional research on the topic is needed and it performance. Performance is defined as the integral of the
may include an analysis of temperatures, solar radiation pavement performance curve. Therefore, the area falling under
measurements, moisture movement, freezing and thawing in the life-cycle curve is by definition an indication of performance
various materials used in different layers of a pavement and (Yoder and Witczak, 1975; Huang, 1993). Relative performance
effect of seasonal variations to produce real-time pavement is defined as the ratio of the area corresponding to a pavement
temperature profiles. Different material types have different life-cycle curve to that of a perfect performance. A perfect
responses to climatic variations. Temperature is also the cause of performance is the one represented by a hypothetical horizontal
thermal cracks, either from a single thermal variation or from straight line. The optimum pavement maintenance and rehabilitation
repetitive cycles of warm/cool temperatures. It is recommended plan is the one associated with the maximum life cycle relative
that an effective subgrade resilient modulus be used to represent performance value.
the effect of seasonal variations, especially for moisture-sensitive Prozzi and Madanat (2004) used incremental approach to
fine-grained soils or for locations with significant freeze-thaw develop pavement performance models using AASHO Road
cycles. The effective resilient modulus is the equivalent modulus Test data. Panel data modeling was adopted to account for
that would result in the same damage to the pavement as if unobserved heterogeneity in the serviceability data. Unobserved
seasonal modulus were used. heterogeneity refers to the presence of section-specific effects
that affect deterioration and are unexplained by the exogenous
5. Pavement Performance Rating Models variables, such as construction quality. If not accounted for,
unobserved heterogeneity can lead to biased and inefficient
Different pavement performance rating models have been given parameter estimates (Frees, 2004).
by various researchers over time. The pavement serviceability
performance concept was developed in a better way during the 5.1 Discussion
AASHO Road Test. The inclusion of serviceability as a factor of The main observation that follows from the results of the
pavement design was an outstanding feature of the AASHO disaggregate-level comparison based on the literature discussed
design methods. Regression models were developed linking the above is that the models, which account for heterogeneity
performance data to design inputs. (differences between pavement sections in the panel) have
Five fundamental assumptions were used in the development significantly better predictive capabilities than the models that do
of the concept of serviceability in AASHO Road Test (HRB, not. While attractive for statistical and managerial reasons,
1962). They were: (1) highways are for the comfort of the model characteristics such as latent performance, incremental
traveling user; (2) the user’s opinion as to how a highway should predictions, and serial dependence, do not display significant
perform is highly subjective; (3) there are characteristics that can impact on the predictive capabilities of the models. The modified
be measured and related to user’s perception of performance; (4) AASHO models are observed to have performed worse than
performance may be expressed by the mean opinion of all users; original AASHO model because they only updated the load
and (5) performance is assumed to be a reflection of serviceability applications to failure but not the rate of deterioration. Models
with increasing load applications. that adopt nonlinear regression formulation are found to give
The AASHO model has been criticized by various researchers much better accuracy than those using simple linear regression.
(Small and Winston, 1988; Paterson, 1987). The major critiques The incremental analysis of PSI is another important aspect
include poor fit to data, use of an inefficient sequential estimation discussed in the literature. It depends on the cumulative traffic
procedure, mismatched units, and mis-specified model. To loading, increments in traffic loads (ESWL) and on the frost
account for subgrade types and environmental conditions that are depth observed during the period. Further, the evaluation of
different from those in the original study, the above equations are potential maintenance and rehabilitation plans based solely on
modified as reported in AASHTO (1993). pavement life-cycle relative performance is not considered an
Prozzi and Madanat (2000) presented a stochastic duration model effective approach, since it does not take into consideration life-
where time to failures is assumed to follow a Weibull distribution, cycle cost.
an assumption that is often used in structural reliability. Similar
to Small and Winston (1988), the focus is to update the equation 6. Conclusions
for the loadings to failure, ñ, while rigorously accounting for the
censored observations that are present in the data. The result is This paper has examined the predictive capabilities of various
converted to the form of the AASHO model and is shown below. flexible pavement performance models proposed by different
5.28 6.68 2.62 researchers in the world. Following are the conclusions drawn
10 ( SN + 1 ) L2
E [ρ ] = ------------------------------------------------
3.03
- (5) from the review of the literature:
( L1 + L 2 ) 1. Various number and types of performance models are avail-
Khaled A. Abaza (2002) suggested that evaluation of potential able in the literature, which predicts the performance of the

− 146 − KSCE Journal of Civil Engineering

 Critical Review of Flexible Pavement Performance Models

pavement. The critical variables affecting pavement perfor- fic speed and material ageing.
mance, including structural components, material properties, 3. With the regular maintenance, the roughness keeps on
design approaches, environmental factors, and maintenance decreasing so this should also be considered as one of the
effects are integrated in these models, and the effects of factors in the pavement performance models and this can be
these factors on pavement deterioration are thoroughly incorporated in the form of cost incurred during mainte-
investigated and quantitatively evaluated. nance activities like patching and sealing.
2. These models incorporate various variables which define 4. Effect of moist soils around the pavement and the mecha-
structural, traffic, environmental, and other service condi- nism of moisture movement is still a field which is not dis-
tions of the roads. Maintenance of pavements also affects covered much. Much of the roads pass through agriculture
the predictive capabilities of the different models so this land and problem of moisture ingress is always associated
should also be considered as one of the variable in perfor- with it. The deterioration mechanism of moisture movement
mance models. need to be developed to account for its effect. Additional
3. By far for prediction of any surface distress parameter, the research on the topic is needed and it may include an analy-
age and traffic on the pavement plays the most important sis of temperatures, solar radiation measurements, moisture
role. These factors are directly proportional to the distress movement, freezing and thawing in various materials used
parameters of the pavements. in different layers of a pavement and effect of seasonal vari-
4. In case of rutting prediction models, the structural number of ations to produce real-time pavement temperature profiles.
the layer and number of load repetitions are the major fac- 5. It is recommended that an effective subgrade resilient modu-
tors responsible. The rutting models calculates the strain in lus be used to represent the effect of seasonal variations,
various layers as this is the only variable which can change especially for moisture-sensitive fine-grained soils or for
and this strain is multiplied with the thickness of the layer to locations with significant freeze-thaw cycles.
predict the rut depth. 6. Various techniques like Finite Element Method, Artificial
5. Performance index based models takes account of all the Neural Network, Genetic Algorithm, and Cellular Automata
performance affecting variables. These models help, in gen- can be used for developing performance prediction models.
eral, in deciding the maintenance schedule. As all the vari- Much more accuracy is expected to be achieved from these
ables are accounted in these models, there are chances of methods as compared to simple regression analysis.
error due to correlation. Most of the work has been done on 7. All the distress predicting models are developed for high-
improving the index models developed after AASHO road ways where traffic is the most important factor behind dis-
test. tress. Still the area of performance models for low volume
6. Among the various modeling techniques used for prediction roads where moisture, environmental and seasonal traffic
of pavement performance, non linear regression analysis are the major factors responsible for failure is unexplored.
gives more accurate results. A more promising approach is
the use of joint estimation from different data sources. The Notations
objective of joint estimation is to yield a more reliable model
of pavement rutting than those produced with either data BCT = Total area of block cracking as percent of total lane
source alone. area
7. Environmental factors affect the structural properties of the COVRD = Coefficient of variation in rut depth %
pavements which are responsible for the deterioration of the IRI = International Roughness Index
pavements. As these are the factors which are quiet uncer- IRI0 = Initial IRI expected within six months after
tain in nature and vary from places to places so they become construction in m/km
important in analyzing the performance of the pavement. LC = Length of sealed longitudinal cracks outside the
wheel path in m/km
6.1 Future Directions of Research L1 = Nominal Axle Load Weight (kip)
The review of number of pavement performance models has L2 = 1 for single axle vehicle and 2 for tandem axle
indicated certain gaps on which further research is needed. These vehicle
are: MR or MR = Resilient modulus
1. Mostly all the performance models are site specific and N = Number of load repetitions
biased to some specific environment, so proper site specific NPS = Number of load repetitions corresponding to the
constants should be selected to get the desired level of accu- initiation of the secondary stage
racy. NST = Number of load repetitions corresponding to the
2. Most of the models are limited to the materials used in the initiation of the tertiary stage
AASHO road test. The experimental data may not represent P = Patching (ft2/1000 ft2)
the true deterioration mechanism of in-service pavements P0.02 = Percent passing the 0.02 mm sieve
because of differences in factors such as traffic wander, traf- P0.075 = Percent passing the 0.075 mm sieve

Vol. 18, No. 1 / January 2014 − 147 −

 Ankit Gupta, Praveen Kumar, and Rajat Rastogi

PI = Plasticity index Highway Research Board (HRB) (1962). “The AASHO road test.
R= Rut depth after N number of load repetitions Report 5-Pavement research.” HRB, National Academy of Sciences
RD = Mean rut depth (in) -National Research Council, Washington, DC.
Huang, Y. H. (1993). Pavement analysis and design, Prentice-Hall,
SF = Site factor
Upper Saddle River, New Jersey, USA.
SN = Structural Number Jin, M. S., Lee, K. W., and Kovacs, W. D. (1994). “Seasonal variation of
TCL = Total length of transverse cracks in m/km resilient modulus of subgrade soils.” J. Transp. Eng., Vol. 120, No.
εp = Accumulated permanent strain 4, pp. 603-616.
εpn = Permanent strain due to a single load application Kaloush, K. E. and Witczak, M. W. (2000). Development of a permanent
(at the Nth application) to elastic strain ratio model for asphalt mixtures, NCHRP 1-37 A,
εPS = Permanent strain corresponding to the initiation of Inter Team Technical Report.
secondary stage Kerali, H. R. (2000). Overview of the HDM-4 system, Vol. 1, The
Highway Development and Management Series, International Study
εr = Resilient strain
of Highway Development and Management (ISOHDM), World
εST = Permanent strain corresponding to the initiation of Roads Association (PIARC), Paris.
the tertiary stage Leahy, R. B. (1989). Permanent deformation characteristics of asphalt
εt = Tensile strain at the bottom of the bituminous layer concrete, PhD Thesis, University of Maryland, College Park, Maryland.
εz = Vertical strain on the subgrade Lister, N. W. (1981). “Heavy wheel loads and road pavements-Damage
η70 = Viscosity of asphalt binder at 70 F relationships.” Symposium on Heavy Freight Vehicles and their
Effects, Organization for Economic Cooperation and Development,
Paris.
References Maree, J. H., Freeme, C. R., Van Zyl, N. J., and Savage, P. F. (1982).
“The permanent deformation of pavements with untreated crushed
AASHTO (1993). AASHTO guide for design of pavement structures stone bases as measured in heavy vehicle simulator tests”. Proc., 11th
1993, AASHTO, Washington, D.C. ARRB Conf., Part 2, pp. 16-28.
Abaza, K. A. (2002). ‘‘Optimum flexible pavement life-cycle analysis NCHRP (2004). Mechanistic-empirical design of new and rehabilitated
model.’’ J. Transp. Eng., Vol. 128, No. 6, pp. 542-549. pavement structures, National Cooperative Highway Research Program,
Al-Suleiman, T. I., Sinha, K. C., and Riverson, J. D. (1989). “Effects of NCHRP Project 1-37A, National Research Council, Washington,
Pavement Age and Traffic on Maintenance Effectiveness” J. Transp. DC.
Eng., Vol. 117, No. 6, pp. 644-659. Paterson, W. D. O. (1987). Road deterioration and maintenance effects:
Archilla, A. R. and Madanat, S. M. (2001). “Estimation of rutting Models for planning and management, Highway Design and
models by combining data from different sources.” J. Transp. Eng. Maintenance, Standard Ser., Johns Hopkins University Press,
Vol. 127, No. 5, pp. 379-389. Baltimore.
Brahmajaree, N., Lavansiri, D., and Tepsittitarakorn, S. (2009). “Effects Prozzi, J. A. and Madanat, S. M. (2000). “Analysis of experimental
of seasonal variations on the flexible pavements in thailand.” pavement failure data using probabilistic duration models,”
Proceedings of 6th ICPT, Sapporo, Japan, pp. 283-290. Transportation Research Record 1699, Transportation Research
Das, A. and Pandey, B. B. (1999). “Mechanistic-empirical design of Board, Washington, D.C., pp. 87-94.
bituminous roads: An indian perspective.” J. Transp. Eng., Vol. 125, Prozzi, J. A. and Madanat, S. M. (2003). “Incremental nonlinear model
No. 5, pp. 463-471. for predicting pavement serviceability.” J. Transp. Eng., Vol. 129,
Diefenderfer, B. K., Al-Qadi, I. L., and Diefenderfer, S. D. (2006). No. 6, pp. 635-641.
“Model to predict pavement temperature profile: development and Prozzi, J. A. and Madanat, S. M. (2004). “Development of pavement
validation.” J. Transp. Eng., Vol. 132, No. 2, pp. 162-167. performance models by combining experimental and field data.” J.
Frees, E. (2004). Longitudinal and panal data: Analysis and applications Infrastruct. Syst., Vol. 10, No. 1, pp. 9-22.
in the social sciences, Cambridge University Press, New York. Sayers, S. W., Gillespie, T. D., and Queiroz, C. A. V. (1986). The
Fwa, T. F., Tan, S. A., and Zhu, L. Y. (2004). “Rutting prediction of international road roughness experiment, World Bank Technical
asphalt pavement layer using C-model.” J. Transp. Eng., Vol. 130, Paper Number 45.
No. 5, pp. 675-683. Shahin, M. Y., Nanez, M. M., Broten, M. R., Carpenter, S. H., and Sameh,
Gulan, S., Zhu, K., Weaver, J., Shan, J., and Flora, W. F. (2001). A. (1984). ‘‘New techniques for modeling pavement deterioration.’’
Development of improved pavement performance prediction models Transp. Res. Rec. 1123, Transportation Research Board, Washington,
for the indiana pavement management system, Final Report, Indiana D.C., pp. 40-46.
Department of Transportation. Witczak, M. W. (2004). Assessment of the allowable threshold rut
Gupta, A., Kumar, P., and Rastogi, R. (2012). “A critical review of depths by layers in asphalt pavement systems, Transportation Research
flexible pavement performance models developed for indian Board, NCHRP 9-19, National Cooperative Highway Research
perspective.” Indian Highways, Journal of Indian Roads Congress, Program, Washington, D.C.
New Delhi, Vol. 40, No. 3, pp. 41-60. Zhou, F., Scullion, T., and Sun, L. (2004). “Verification and modelling
Harvey, J. T. and Popescu, L. (2000). Rutting of caltrans asphalt of three-stage permanent deformation behavior of asphalt mixes.” J.
concrete and asphalt-rubber hot mix under different wheels, Tyres Transp. Eng., Vol. 130, No. 4, pp. 486-494.
and Temperatures – Accelerated Pavement Testing Evaluation, Zuo, G., Drumm, E. C., and Meier, R. W. (2007). “Environmental effects
Pavement Research Center, Institute of Transportation Studies, on the predicted service life of flexible pavements.” J. Transp. Eng.,
University of California, Berkeley. Vol. 133, No. 1, pp. 47-56.

− 148 − KSCE Journal of Civil Engineering