Вы находитесь на странице: 1из 27

Incorporation of Environmental Effects

in Pavement Design

C.E. Zapata* D. Andrei** M.W. Witczak*


W.N. Houston*

* Department of Civil and Environmental Engineering


Arizona State University
Tempe, AZ 85287-5306, USA
czapata@asu.edu
witczak@asu.edu
sandy.houston@asu.edu
** Fugro Consultants LP
8613 Cross Park
Austin, TX 78754, USA
DAndrei@fugro.com

ABSTRACT. Environmental conditions have a significant effect on the performance of both


flexible and rigid pavements. External factors such as precipitation, temperature, freeze-thaw
cycles, and depth to water table play a key role in defining the bounds of the impact the
environment can have on the pavement performance. As part of the new US Mechanistic-
Empirical Pavement Design Guide (MEPDG) being developed under the overall project
sponsored by the National Cooperative Highway Research Program (NCHRP project 1-37A),
a climatic modelling tool called the Enhanced Integrated Climatic Model (EICM) was
implemented to incorporate the changes in temperature and moisture of unbound materials
into the design process. Currently a new independent review project (NCHRP 1-40) is
reviewing this model to correct errors and to develop further enhancements to produce a final
methodology ready for approval/disapproval vote by AASHTO in 2006. This paper reflects
the methodology used for the MEPDG and present the models incorporated by Arizona State
University into the EICM, the input needed and the outputs generated by the program. A
discussion on how EICM determines the temperature and moisture distribution within the
pavement system is also presented.
KEYWORDS: Pavement Design, Environmental Effects, Climatic Model, Design Guide.

DOI:10.3166/RMPD.8.667-693 2007 Lavoisier, Paris

RMPD 8/2007. Water in Pavements, pages 667 to 693


668 RMPD 8/2007. Water in Pavements

1. Introduction

1.1. Importance of climate in mechanistic-empirical design

Environmental conditions have a significant effect on the performance of both


flexible and rigid pavements. External factors such as precipitation, temperature,
freeze-thaw cycles, and depth to water table play a key role in defining the bounds
of the impact the environment can have on the pavement performance. Internal
factors such as the susceptibility of the pavement materials to moisture and freeze-
thaw damage; drainage of paving layers, and infiltration potential of the pavement,
define the extent to which the pavement will react to the applied external
environmental conditions.
In a pavement structure, moisture and temperature are the two environmentally
driven variables that can significantly affect the pavement layer and subgrade
properties and, hence, its load carrying capacity. Some of the effects of environment
on pavement materials include:
Modulus values can vary from 13,800 to 20,700 MPa (2 to 3 million psi) or
more during cold winter months to about 690 MPa (100,000 psi) or less during hot
summer months.
Temperature and moisture gradients particularly in the top Portland cement
concrete (PCC) layer can significantly affect stresses and deflections and
consequently pavement damage and distresses.
At freezing temperatures, water in soil freezes and its resilient modulus can
rise to values 20 to 120 times higher than the value of the modulus before freezing.
All other conditions being equal, the higher the moisture content the lower the
modulus of unbound materials; however, moisture has two separate effects: First, it
can affect the state of stress, through suction or pore water pressure. Second, it can
affect the structure of the soil through destruction of the cementation between soil
particles (Lekarp et al., 2000).
Excessive moisture can lead to stripping in asphalt mixtures or can have long-
term effects on the structural integrity of cement bound materials.
Freeze-thaw effects are experienced in the underlying layers but eventually
lead to distresses in the pavement surface.
A new US Mechanistic-Empirical Design Guide for Pavements (MEPDG) was
developed under the sponsorship of the National Cooperative Highway Research
Program (NCHRP). All the distresses considered in the MEPDG are affected by the
environmental factors to some degree. Therefore, diurnal and seasonal fluctuations
in the moisture and temperature profiles in the pavement structure brought about by
changes in ground water table, precipitation/infiltration, freeze-thaw cycles, and
other external factors are modelled in a very comprehensive manner by a climatic
model called the Enhanced Integrated Climatic Model (EICM).
Environmental Effects in Pavement Design 669

1.2. Scope of the paper

The scope of this paper is to present the models implemented into the EICM, the
input needed and the outputs generated by the program. A discussion on how EICM
determines the temperature and moisture distribution within the pavement system is
also presented.
This paper reflects the methodology at the time NCHRP 1-37A was completed
and provided to the sponsors. It should be recognized that further independent
review is currently being undertaken and it is anticipated that several further changes
in the methodology herein described are expected.

2. Considerations of climatic effects in design

The EICM is a one-dimensional coupled heat and moisture flow program that
simulates changes in the behavior and characteristics of pavement and subgrade
materials in conjunction with climatic conditions over several years of operation.
The original version of the EICM, referred to as the Integrated Climatic Model, was
developed for the Federal Highway Administration (FHWA) at Texas A&M
University, Texas Transportation Institute in 1989 (Lytton et al., 1990). This version
coupled an infiltration and drainage model (ID model) to a climatic-materials-
structural model (CMS model) (Dempsey et al., 1985) and a frost-heave and thaw
settlement model (CRREL model) (Guymon et al., 1986), to develop an integrated
environmental predictive methodology. The original version was then modified and
released in 1997 by Larson and Dempsey as ICM version 2.0 (Larson and Dempsey,
1997). Additional modifications in the moisture prediction model were performed in
1999 by Arizona State University, leading to ICM version 2.1. Further
improvements were made at Arizona State University as part of the MEPDG
development to further improve the moisture prediction capability of ICM version
2.1, leading to the version referred to as EICM. In developing the EICM, data from
the Long Term Pavement Performance (LTPP) Seasonal Monitoring Program
(SMP) test sections were used (Witczak et al., 2000a, 2000b, 2000c, 2000d).
The EICM software has been made an integral part of the MEPDG procedure.
The user inputs to the EICM are entered through interfaces provided as part of the
MEPDG software and EICM processes these inputs and feeds the processed outputs
to the three major components of the Guide materials, structural responses, and
performance prediction.
The EICM model can be applied to either asphalt concrete (AC) or Portland
cement concrete (PCC) pavements. The EICM records the user supplied resilient
modulus, MR, of all unbound layer materials at an initial or reference condition.
Generally, this will be at or near the optimum water content and maximum dry
density. Subsequently, the EICM computes and predicts the following information
throughout the entire pavement profile:
670 RMPD 8/2007. Water in Pavements

Evaluates the expected changes in moisture content, from the initial or


reference condition, as the subgrade and unbound materials reach equilibrium
moisture condition. Also evaluates the seasonal changes in moisture contents.
Evaluates the effect of changes in soil moisture content with respect to the
reference condition on the user entered resilient modulus, MR.
Estimates a set of adjustment factors for unbound material layers that account
for the effects of moisture content changes, freezing, thawing, and recovery from
thawing. This factor, denoted Fenv, varies with position within the pavement
structure and with time throughout the analysis period. The Fenv factor is a
coefficient that is multiplied by the MRopt to obtain MR as a function of position and
time.
Makes use of time-varying MR values in the computation of critical pavement
response parameters and damage at various points within the pavement system.
Evaluates changes in temperature as a function of time for all asphalt bound
layers.
For rigid pavements, the following additional information is generated:
Temperature profiles in the PCC and underlying layers used for thermal
gradients in PCC; and joint and crack openings and closings.
Effective linear temperature gradient used to model slab curvature and thermal
stresses.
Probability distribution of effective linear temperature gradients.
Freezing index and the number of freeze-thaw cycles for the selected location.
Mean monthly relative humidity values used in the estimation of moisture
warping of the PCC slabs.

3. Inputs required to model thermal and moisture conditions

The approach for selecting or determining material inputs in the MEPDG is a


hierarchical (level) system. This approach is based on the philosophy that the level
of engineering effort exerted in the pavement design process should be consistent
with the relative importance, size, and cost of the design project. Level 1 is the most
current implementable procedure available, normally involving comprehensive
laboratory or field tests. In contrast, Level 3 requires the designer to estimate the
most appropriate design input value of the material property based on experience
with little or no testing. Inputs at Level 2 are estimated through correlations with
other material properties that are measured in the laboratory or field.
The inputs required by the climatic model fall under the following broad
categories: General initialisation information; weather-related information; ground
water related information; drainage and surface properties; and pavement structure
and materials.
Environmental Effects in Pavement Design 671

The specific inputs required under each of the mentioned categories and the
recommended procedures to obtain them at the various hierarchical input levels are
summarized below. The data presented covers both new and rehabilitation design.

3.1. General information

In order to initialise the climatic model, the following information is required:


base/subgrade construction completion date for new pavement design; existing
pavement construction date for rehabilitation design; pavement construction date;
traffic opening date; and the type of design (new or rehab and AC or PCC).

3.2. Weather information

The MEPDG damage accumulation approach requires five weather-related


parameters on an hourly basis over the entire design life for the project being
designed: Air temperature, precipitation, wind speed, percentage sunshine, and
relative humidity.
The air temperature is required by the heat balance equation in the EICM to
define the frozen/thawing periods within the analysis time frame, and to determine
the number of freeze-thaw cycles. Precipitation is needed to compute infiltration for
rehabilitated pavements and aging processes. Wind speed is required in the
computations of the convention heat transfer coefficient at the pavement surface.
The percentage sunshine is needed for the calculations of heat balance at the surface
of the pavement and particularly the net long-wave radiation. Last, the relative
humidity is used in computing the drying shrinkage of JPCP and CRCP and also in
determining the crack spacing and initial crack width in CRCP.
The weather-related information is primarily obtained from weather stations
located near the project site. The software accompanying the Design Guide has an
available database from nearly 800 weather stations throughout the United States.
The climatic database can be tapped into by simply specifying the latitude,
longitude, and elevation of the project site. Once the co-ordinates and elevation are
specified, the Design Guide software will highlight the six closest weather stations
to the site from which the user may select any number of stations deemed to be most
representative of the local climatic conditions. After the appropriate number of
representative weather stations is chosen, interpolation of climatic data from these
stations is done and the interpolated data is made available for storage as a virtual
weather station.
672 RMPD 8/2007. Water in Pavements

3.3. Groundwater table depth

The groundwater table depth is intended to be either the best estimate of the
annual average depth or the seasonal average depth. At input Level 1, it could be
determined from profile characterization borings prior to design. At input Level 3,
an estimate of the annual average value or the seasonal averages can be provided. A
potential source to obtain Level 3 estimates is the county soil reports produced by
the National Resources Conservation Service (Schoeneberger et al., 1998).

3.4. Drainage properties

Drainage properties include the following parameters: Infiltration potential,


drainage path length, and pavement cross slope.
The net infiltration potential of the pavement over its design life is a qualitative
parameter. The infiltration can assume four values none, minor (10 percent of the
precipitation enters the pavement), moderate (50 percent of the precipitation enters
the pavement), and extreme (100 percent of the precipitation enters the pavement).
Based on this input, the EICM determines the amount of water available on top of
the first unbound layer.
The drainage path length is the distance measured along the resultant of the cross
and longitudinal slopes of the pavement. It is measured from highest point in the
pavement cross-section to the point where drainage occurs. This input is used in the
EICMs infiltration and drainage model to compute the time required to drain an
unbound base or subbase layer from an initially wet condition. This parameter is
computed by DRIP, a microcomputer program apart from the EICM. Finally, the
cross slope is the slope of the pavement surface perpendicular to the direction of
traffic. This input is used in computing the time required to drain a pavement base or
subbase layer from an initially wet condition.

3.5. Pavement structure materials inputs

3.5.1. Layer thickness


The layer thickness of each material in the pavement structure should correspond
to layers that are more or less homogeneous. EICM internally subdivides these
layers for more accurate calculations of moisture and temperature profiles.

3.5.2. AC and PCC material properties


Several bound material properties are required for the design of flexible and rigid
pavements, and AC or PCC overlays. Those that control the heat flow through the
pavement system and thereby influence the temperature and moisture regimes within
Environmental Effects in Pavement Design 673

it are the surface shortwave absorptivity, the thermal conductivity, K, and the heat or
thermal capacity, Q. The surface long-wave absorptivity is not required, as the
model makes use of the percent sunshine to calculate cloud cover, which is tightly
related to the albedo.
The surface short wave absorptivity directly correlates with the amount of
available solar energy that is absorbed by the pavement surface. Lighter and more
reflective surfaces tend to have lower short wave absorptivity and vice versa. At
Level 1, it is recommended that this parameter be estimated through laboratory
testing. At Level 3, default values can be assumed for various pavement materials:
Weathered asphalt (gray) 0.80 0.90
Fresh asphalt (black) 0.90 0.98
Aged PCC layer 0.70 0.90
Thermal conductivity, K, is the quantity of heat that flows normally across a
surface of unit area per unit of time and per unit of temperature gradient; while the
heat or thermal capacity is the actual amount of heat energy Q necessary to change
the temperature of a unit mass by one degree. Direct measurements of both
parameters are recommended under input Level 1 (ASTM E1952 and ASTM
D2766, respectively). For Level 3, it is recommended to use design values based
upon agency historical data or from typical values shown in Table 1.

Table 1. AC and PCC materials inputs required for EICM calculations

Material property AC materials PCC materials

Thermal conductivity, K 0.44 - 0.81 Btu/(ft)(hr)(oF) 1.0 to 1.5 Btu/(ft)(hr)(oF)


Heat Capacity, Q 0.22 to 0.40 Btu/(lb)(oF) 0.2 to 0.28 Btu/(lb)(oF)

3.5.3. Compacted unbound material properties


3.5.3.1. Mass-volume relationships
The parameters needed in this category are the maximum dry density (d max),
specific gravity (Gs), and the optimum gravimetric moisture content (wopt) of the
compacted unbound material in question. From these three inputs, all other mass-
volume parameters can be computed. At Level 1, it is required that the d max, wopt,
and Gs be carefully measured in the laboratory in accordance with standard
protocols for each unbound layer: AASHTO T180 for base layers and AASHTO
T99 for other layers; and AASHTO T100. If the user chooses not to measure d max,
wopt, and Gs, then it is suggested that Level 2 inputs be adopted. At input Level 2, the
user enters the effective grain size corresponding to 60 percent passing by weight,
D60, the percent passing the U.S. No. 200 sieve, P200, and the plasticity index, PI.
674 RMPD 8/2007. Water in Pavements

From these parameters, the EICM will compute dmax wopt, and Gs using the
following correlations (Witczak et al., 2000d):

P200 * PI
WPI = [1]
100
To compute Gs:

Gs = 0.041(WPI)0.29 + 2.65 [2]

where Gs = Specific gravity of the solids (dimensionless).


To compute wopt:
If WPI > 0:
wopt = 1.3 (WPI)0.73 + 11 [3]

where wopt = Gravimetric water content at optimum (%).


If WPI = 0

wopt (T99) = 8.6425 (D60)-0.1038 [4]

If layer is not a base course:

wopt = wopt (T99) [5]

If layer is a base course:

wopt = 0.0156[wopt(T99)]2 0.1465wopt(T99) + 0.9 [6]

wopt = wopt (T99) - wopt [7]

To compute d max:

Sopt = 6.752 (WPI)0.147 + 78 [8]

where Sopt = Degree of saturation at optimum condition (%).

G s water
d max comp = [9]
wopt G s
1+
S opt

where water = Unit weight of water (in consistent units).


Environmental Effects in Pavement Design 675

If layer is a compacted material:


d max = d max comp [10]

If layer is a natural in-situ material:

d = 0.90 d max comp [11]

EICM uses d for d max.


Level 3 inputs are not applicable for this category.

3.5.3.2. Equilibrium gravimetric moisture content


Equilibrium gravimetric moisture content is a required input for rehabilitation
design. However, it is not required for new pavement design. It is recommended that
this parameter be estimated from direct testing of bulk samples retrieved from the
site or through other appropriate means.

3.5.3.3. Saturated hydraulic conductivity


Saturated hydraulic conductivity, ksat, is required to determine the transient
moisture profiles in compacted unbound materials and to compute their drainage
characteristics. At Level 1, a direct measurement using a permeability test
(AASHTO T215) is recommended. At Level 2, the following correlations are
available:
If 0 WPI < 1:

1.1275(log D + 2 )2 + 7.2816 (log D + 2 )11.2891


60 60
k sat = 10
(cm/s) [12]

Equation [12] is valid for D60 < 0.75 mm. If D60 > 0.75 mm, set D60 = 0.75 mm.
If WPI 1:
0.0004 ( P PI )2 0.0929 ( P PI ) 6.56
200 200
k sat = 10
(cm/s) [13]

3.5.3.4. Dry thermal conductivity and dry heat capacity


To determine the dry thermal conductivity (K), a direct measurement is
recommended at Level 1 (ASTM E1952). At Level 3, recommended values for each
soil type are available. They are presented in Table 2.
To obtain the dry heat capacity (Q), a direct measurement is recommended at
Level 1 (ASTM D2766). At Level 3, the user selects design values based upon
676 RMPD 8/2007. Water in Pavements

agency historical data. Typical values range from 0.71 to 0.84 kJ/(kg K) (0.17 to
0.20 Btu/lbmoF).

Table 2. Dry thermal conductivity, K, for unbound compacted material

Range Range Recommended Recommended


Soil type
W/(m K) Btu/(ft hroF) W/(m K) Btu/(ft hroF)

A-1-a 0.38 0.76 0.22 0.44 0.52 0.30

A-1-b 0.38 0.76 0.22 0.44 0.47 0.27

A-2-4 0.38 0.42 0.22 0.24 0.40 0.23

A-2-5 0.38 0.42 0.22 0.24 0.40 0.23

A-2-6 0.35 0.40 0.20 0.23 0.38 0.22

A-2-7 0.28 0.40 0.16 0.23 0.35 0.20

A-3 0.43 0.69 0.25 0.40 0.52 0.30

A-4 0.29 0.40 0.17 0.23 0.38 0.22

A-5 0.29 0.40 0.17 0.23 0.33 0.19

A-6 0.28 0.38 0.16 0.22 0.31 0.18

A-7-5 0.16 0.29 0.09 0.17 0.23 0.13

A-7-6 0.16 0.29 0.09 0.17 0.21 0.12

3.5.3.5. Soil water characteristic curve parameters


The soil water characteristic curve (SWCC) is defined as the variation of water
storage capacity within the macro- and micro-pores of a soil, with respect to suction
(Fredlund et al., 1995). This relationship is generally plotted as the variation of the
water content (gravimetric, volumetric, or degree of saturation) with soil suction.
Several studies have been conducted on comparing the different equations available
to represent the SWCC (Leong and Rahardjo 1996, Zapata 1999). Those studies
have generally shown that the equations proposed by Fredlund and Xing (1994)
showed good agreement with an extended database.
At Level 1, direct measurement of suction (h) in kPa, and volumetric water
content (w) pairs of values are required. Based on a non-linear regression analysis,
the user needs to compute the SWCC model parameters af, bf, cf, and hr using the
Equations [14] and [15], proposed by Fredlund and Xing (1994), and the (h, w)
pairs of values:
Environmental Effects in Pavement Design 677




sat [14]
w = C( h ) cf

bf
h
ln EXP( 1 ) +

af

h
ln1 +

C (h) = 1 hr [15]
1 106
ln1 +
h
r

where sat = Saturated volumetric water content or porosity, which is computed by:

wopt d max
opt = [16]
water

opt
S opt = [17]
d max
1
water G s

opt
sat = [18]
S opt

For Level 2, the EICM will compute the SWCC model parameters af, bf, cf, and
hr by using the following correlations with WPI and D60 (Zapata, 1999; Zapata et al.,
1999). WPI and D60 parameters have been previously defined:
If WPI > 0

a f = 0.00364(WPI ) 3.35 + 4(WPI ) + 11 (kPa) [19]

bf
= 2.313(WPI ) 0.14 + 5 [20]
cf

c f = 0.0514(WPI ) 0.465 + 0.5 [21]

hr
= 32.44e0.0186(WPI ) [22]
af
678 RMPD 8/2007. Water in Pavements

If WPI = 0

a f = 0.8627( D60 ) 0.751 (kPa) [23]

b f = 7.5 [24]

c f = 0.1772 ln( D60 ) + 0.7734 [25]

hr 1
= [26]
af D60 + 9.7e 4

The SWCC will then be established internally using Equation [14] and [15] as
shown for Level 1. Equations [16], [17] and [18] are used to compute sat based on
direct measurements of dmax, wopt, and Gs.
For Level 3, the EICM will compute the SWCC model parameters af, bf, cf, and
hr by using correlations with WPI and D60, as shown for Level 2. However, direct
measurements of dmax, wopt, and Gs are not required and such parameters are
estimated by EICM based on Equations [1] to [11]. Figure 1 summarizes the results
obtained for both groups of soils.

1.2
WPI = % Passing #200 * PI/100
1.0
Degree of Saturation

0.8

WPI = 0.1
0.6 3
5 10 15
20 30 40 WPI = 50
0.4
D60=0.1 mm

0.2 D60=1 mm

0.0
1E-1 1E+0 1E+1 1E+2 1E+3 1E+4 1E+5 1E+6
Matric Suction (kPa)
Figure 1. Predicted SWCC based on D60 and WPI

3.6. Uncompacted/natural unbound material properties

A lower level of effort is generally sufficient to characterize the


uncompacted/natural unbound materials when compared to the properties of the
Environmental Effects in Pavement Design 679

overlying compacted materials. Therefore, Level 1 inputs are generally not required
for in-situ materials. It is recommended that only PI, P200, P4, and D60 be measured
for the in-situ layers (where P4 is the percent passing the U.S. No. 4 sieve; all other
parameters have been defined previously).

4. Environmental effects on resilient modulus of unbound pavement materials

To evaluate the resilient modulus, MR, of unbound materials used in the


MEPDG, several factors influencing the modulus need to be considered: Stress state,
moisture/density variations, and freeze/thaw effects.
Although the stress sensitivity is only considered if Level 1 inputs are used in the
MEPDG, the impact of temporal variations in moisture and temperature on MR are
fully considered at all levels through the composite environmental adjustment factor,
Fenv. The EICM deals with all environmental factors and provides soil moisture,
suction, and temperature as a function of time, at any location in the unbound layers
from which Fenv can be determined. The resilient modulus MR at any time or position
is then expressed as follows:

MR = Fenv MRopt [27]

The factor Fenv is an adjustment factor and MRopt is the resilient modulus at
optimum conditions and at any state of stress. It is obvious in Equation [27] that the
variation of the modulus with stress and the variation of the modulus with
environmental factors (moisture, density, and freeze/thaw conditions) are assumed
independent. Although this is not necessarily the case, recent studies support the use
of this assumption in predicting resilient modulus without significant loss in
accuracy of prediction. The adjustment factor Fenv, being solely a function of the
environmental factors, can then be computed by the EICM, without actually
knowing MRopt.
The development of predictive equations and techniques that address the
influence of changes in moisture and freeze/thaw cycles on the resilient modulus of
unbound materials is described in the following two subsections.

4.1. Resilient modulus as function of soil moisture

An intensive literature review study was completed with the objective of


summarizing existing models that incorporated the variation of resilient modulus
with moisture (Witczak et al., 2000a). Using these published models from the
literature (Li and Selig, 1994; Drumm et al., 1997; Rada and Witczak, 1981; Santha,
1980), it was possible to select a model that would analytically predict changes in
modulus due to changes in moisture. This model is presented in Equation [28].
680 RMPD 8/2007. Water in Pavements

MR ba
log =a+ [28]
M Ropt b
1 + EXP ln ( )
+ k m S S opt
a

where MR/MRopt = resilient modulus ratio; a = minimum of log(MR/MRopt); b =


maximum of log(MR/MRopt); km = regression parameter; and (S Sopt) = variation in
degree of saturation expressed in decimal.
Based on the available literature data, maximum modulus ratios of 2.5 for fine-
grained materials and 2 for coarse-grained materials were adopted. The values of a,
b, and km for coarse-grained and fine-grained materials are given in Table 3. Fine-
grained soils label refer to those with passing U.S. No. 200 sieve greater than 50%

Table 3. Values of a, b, and km for coarse-grained and fine-grained materials

Coarse-grained Fine-grained
Parameter Comments
materials materials

a -0.3123 -0.5934 Regression parameter

Corresponding to modulus ratios


b 0.3 0.4
of 2 and 2.5, respectively

km 6.8157 6.1324 Regression parameter

Resilient moduli for frozen/thawed unbound materials


To study the behavior of unbound materials under freezing/thawing conditions, a
significant number of sources were consulted and salient values of moduli, MR, and
ratios of moduli were extracted (Witczak et al., 2000). The objective of the search
was to obtain absolute values of moduli for frozen material, termed MRfrz, and the
ratio of MR just after thawing, termed MRmin, to the MR of natural, unfrozen material,
termed MRunfrz. The ratio is used as a reduction factor, termed RF. Because some of
the data from the literature produced RF values based on MRunfrz as a reference and
some were based on MRopt as a reference, it was decided to adopt the conservative
interpretation of using the smaller of MRunfrz and MRopt as a reference as shown in
Equation [29].

RF = MRmin/smaller of (MRunfrz, MRopt) [29]


The average values reported in the literature for MRfrz were found to be 20,685
MPa ( 3*106 psi) for coarse-grained materials; 13,790 MPa ( 2*106 psi) for fine-
grained silt and silty sands; and 6,895 MPa ( 1*106 psi) for clays.
Environmental Effects in Pavement Design 681

For thawed materials, the degree of MR degradation upon thawing was found to
correlate with frost-susceptibility, or the ability of the soil to sustain ice lens
formation under favourable conditions. Frost-susceptibility in turn can be estimated
from the percent passing the U.S. No. 200 sieve, P200, and the Plasticity Index, PI. In
Tables 4 and 5, the RF values recommended in the MEPDG are given for coarse-
grained and fine-grained materials as a function of P200 and PI.
Recovering materials experience a rise in modulus with time, from MRmin to
MRunfrz, that can be tracked using a recovery ratio (RR) that ranges from 0 to 1:
RR = 0 for the immediately after thawing condition, when excess water
makes the suction go to zero, MRrecov = MRmin.
RR = 1 when the suction is equal to the suction dictated by the depth to the
ground water table i.e. equilibrium is achieved, MRrecov = MRunfrz.
t
RR = [30]
TR

where RR = recovery ratio; t = number of hours elapsed since thawing started; and
TR = recovery period (number of hours required for the material to recover from the
thawed condition to the normal, unfrozen condition).
The recovery period, TR, is noted as a function of the material type/properties, as
follows: TR = 90 days for sands/gravels with WPI < 0.1; 120 days for silts/clays with
0.1 < WPI < 10; and 150 days for clays with WPI > 10.

Table 4. Recommended values of RF for coarse-grained materials (P200 < 50%)

Distribution
P200 (%) PI < 12% PI =12% - 35% PI > 35%
of Coarse Fraction*

<6 0.85 - -
Mostly Gravel (P4 < 50%) 6 12 0.65 0.70 0.75
> 12 0.60 0.65 0.70

<6 0.75 - -
Mostly Sand (P4 > 50%) 6 12 0.60 0.65 0.70
> 12 0.50 0.55 0.60

* If it is unknown whether a coarse-grained material is mostly gravel or mostly sand, assume


sand.
682 RMPD 8/2007. Water in Pavements

Table 5. Recommended values of RF for fine-grained materials (P200 > 50%)

P200 (%) PI < 12% PI = 12% - 35% PI > 35%

50 - 85 0.45 0.55 0.60

> 85 0.40 0.50 0.55

4.2. Computation of environmental adjustment factor, Fenv

To obtain the composite moduli for layers in which two or more states of the
material coexist and/or the resilient modulus varies with depth and time, the
environmental adjustment factor, Fenv is calculated. The resilient modulus MR at any
time or position is determined as a product of the composite environmental
adjustment factor, Fenv, and the resilient modulus at optimum conditions MRopt.
The environmental adjustment factor, Fenv is a composite factor, which could in
general represent a weighted average of the factors appropriate for various possible
conditions:
Frozen: frozen material FF (factor for frozen materials)
Recovering: thawed material that is recovering to its state before freezing
occurred FR (factor for recovering materials)
Unfrozen/fully recovered/normal: for materials that were never frozen or are
fully recovered FU (factor for unfrozen material)
The Fenv factors are calculated for all three cases, at two levelsat each nodal
point and for each layer.

4.2.1. Adjustment factors at node level


In the EICM the pavement structure is characterized by an array of nodes at
which the values of moisture, suction, and temperature are calculated at any time t.
The value of FF, the adjustment factor for frozen materials, is computed at each node
at which a freezing temperature occurs using the following procedure:
MRopt is either a direct user input or can be estimated from other engineering
properties such as CBR, R-value, structural layer coefficients (ai), Penetration Index,
or from gradation parameters. The estimation of MRopt is out of the scope of this
paper.
Assign values for the Frozen Resilient Modulus, MR frz : If WPI = 0; use MR frz =
17,238 MPa (2.5 x 106 psi); if WPI > 0; use MR frz = 6,895 (1 x 106 psi).
Compute the frozen adjustment factor, FF (Witczak et al. 2000c):
Environmental Effects in Pavement Design 683

M R frz
FF = [31]
M R opt_est

The adjustment factor for recovering materials, FR, is computed at each node at
which freezing temperatures do not occur and the recovery ratio RR is < 1. The
procedure to estimate FR is as follows:
Compute Recovery Ratio, RR as per Equation [30].
Compute Sopt as per Equation [17] for Level 1 or Equation [8] for Level 2.
Compute Sequil from the SWCC equation in terms of degree of saturation
(Fredlund and Xing, 1994):
1 [32]
S equil = C( h ) cf

bf

ln EXP( 1 ) + h
af

h
ln1 +
C ( h) = 1 hr [33]
1 106
ln1 +
hr

where: h = yGWT * water , in kPa; yGWT = groundwater table depth; af (kPa), bf, cf, and
hr (kPa) are fitting parameters defined earlier.
Compute Requil as (Witczak et al., 2000a):

M Re quil ba
log Requil = log =a+ [34]
b
M Ropt
( )
1 + EXP ln + k m S equil S opt
a
where: a, b, and km are constants defined in Table 3.
Compute the RF value from Tables 4 and 5.
Compute the factor for recovering material, FR (Witczak et al., 2000c):
If (Sequil Sopt) < 0: FR = RF + Requi l * RR RR * RF
If (Sequil Sopt) > 0: FR = Requil (RF + RR RR * RF)

To estimate the adjustment factor for unfrozen or fully recovered materials, FU,
the following equation is used (Witczak et al., 2000a):
684 RMPD 8/2007. Water in Pavements

MR ba
log FU = log =a+ [35]
b
M Ropt
(
1 + EXP ln + k m S S opt
a
)

where: a, b, and km are constants from Table 3, and S is the estimated degree of
saturation at any node.

4.2.2. Composite adjustment factors, Fenv, for structural layers


For a given layer (base, subbase, subgrade), frozen, thawed, and never frozen
materials can coexist within a single layer, and hence a composite adjustment factor
that can handle all possible cases is needed. The calculation of a composite
adjustment factor is useful even when the material in a layer is all at the same state
(unfrozen or recovering). This is because the adjustment factors vary from node to
node (with moisture or suction) and an equivalent factor for the whole layer is
needed.
To obtain an equivalent modulus, an elastic spring series analogy was
considered. Using the analogy, the elements of a column (corresponding to Hour 1,
for example) of a node/time matrix, are considered as elastic moduli of a series of
springs (one spring per node). If the stress applied to this model is , then the
displacement in one spring at a given node and time increment can be computed.
To get the average displacement, average, over the whole analysis period (2 weeks
or 1 month), Equation [36] is used:

1
t total n hnode
average =
t total
t =1

node =1 M Rnode ,t



[36]

where: t = time (corresponding to the column in the matrix being considered);


hnode = length of the spring assigned to the node being considered; MRnode,t = modulus
for the node; and ttotal = total number of t time increments (EICM uses 1 hour) over
which the composite modulus is calculated (number of columns in the matrix).
Then the composite (equivalent) modulus can be obtained by finding a
composite modulus, MRcomp, which produces the same average over the total layer
thickness for the same applied . Equating average for the composite model to average
from Equation [36] and cancelling which appears on both sides:

htotal 1
t total n
hnode
=
M Rcomp t total



[37]
t =1 node =1 M Rnode ,t

where: htotal = total height of the considered layer/sublayer.


Environmental Effects in Pavement Design 685

Because the resilient modulus at any node/time can be expressed as the product
of an adjustment factor times the resilient modulus at optimum, Equation [37] can be
replaced with Equation [38]. A composite adjustment factor, Fenv, for the considered
sub-layer (sub-matrix) can be obtained from:

htotal 1
t total n hnode
=
Fenv M Ropt t total
t =1

node =1 Fnode ,t M Ropt



htotal t total
Fenv = [38]
ttotal n hnode




t =1 node =1 Fnode ,t

where: Fenv = composite adjustment factor for the considered sublayer, and Fnode,t =
adjustment factor at a given node and time increment (which could be FF, FR, or FU,
depending on the state of the material).
The procedure should be applied for the entire design period (e.g. 20 years
divided into months or 2-week periods) since the adjustment factors vary from node
to node, even within a layer (or sublayer) in which all material is at the same state
(frozen, unfrozen, or recovering).

5. Determination of the temperature throughout pavement systems

The effect of moisture is more significant on unbound materials than on bound


materials. On the other hand, temperature affects both the bound (asphalt and
cement) and unbound layers significantly. At very cold temperatures, the asphalt
stiffness is close to that of PCC, whereas at very warm temperatures, its stiffness is
closer to an unbound material.
The durability of PCC materials is affected greatly by the freeze-thaw
environment it operates under. Temperature and moisture related curling and
warping phenomena play a significant role in defining the PCC pavement fatigue
behaviour. Temperature and moisture also play a role in the opening and closing of
joints in JPCP and cracks in CRCP, which affect performance.
In unbound materials, cooler temperatures result in frost formation and a
subsequent increase in modulus. On the other hand, warmer temperatures cause
thawing, resulting in increased moisture contents and a subsequent decrease in
modulus values. During the thawing process, the resilient modulus of unbound
materials may go well below the optimum value (0.5 to 0.85 times MRopt).
The CMS and CRREL models in the EICM are primarily responsible for most
the temperature calculations. The CMS model was originally developed at the
University of Illinois (Dempsey et al., 1985). It is a one-dimensional, forward finite
686 RMPD 8/2007. Water in Pavements

difference heat transfer model to determine frost penetration and temperature


distribution in the pavement system. The model considers radiation, convection,
conduction, and the effect of latent heat. It does not consider transpiration,
condensation, evaporation, or sublimation. These latter effects are neglected because
of the uncertainty in their calculations and because their omission does not create
significant errors in the heat balance at the surface of the pavement. Heat fluxes
caused by precipitation and moisture infiltration are also neglected.
For pavement layers (AC or PCC), the EICM assumes that the user input heat
capacity and thermal conductivity do not vary over time. However, for unbound
layers (base courses and soils), as the moisture and frost contents change with time,
so do the heat capacity and thermal conductivity. The user input dry heat capacity
and dry thermal conductivity, which along with the water and ice content predicted
by the EICM, are used to calculate the wet heat capacity and wet thermal
conductivity. In this manner the heat/temperature calculations of the EICM are
coupled with the EICMs moisture predictions.
Once the thermal properties that define the heat flow through the pavement and
unbound layers have been established and the boundary conditions have been
identified, it is necessary to determine the amount of heat inflow/outflow at the
pavement surface. The two processes by which heat is added or subtracted from the
pavement surface are convection and radiation.
The second model used in the MEPDG is the CRREL model (Guymon et al.,
1986). It is a one-dimensional coupled heat and moisture flow in the subgrade soil at
temperatures that are above, below and at the freezing temperature of water. The
model predicts the depth of frost and thaw penetration. It also estimates the vertical
heave due to frost formation and vertical settlement when the soil thaws. The
CRREL model uses the temperature profiles through the asphalt layers established
by the CMS model to compute changes in the soil temperature profile, and thus frost
penetration and thaw settlement.

5.1. Heat flux boundary conditions for CMS model

Temperatures throughout the pavement structure are dominated by atmospheric


conditions at the surface. While it is easy to measure the air temperatures, there is
not a direct correspondence between the air temperatures and pavement surface
temperatures. To estimate the pavement temperature, the energy balance at the
surface used in the CMS model is described by Equation [39] (Scott, 1964; Berg,
1968):

Qi Qr + Qa Qe Qc Qh Q g = 0 [39]
Environmental Effects in Pavement Design 687

where: Qi = incoming short wave radiation; Qr = reflected short wave radiation;


Qa = incoming long wave radiation; Qe = outgoing long wave radiation;
Qc = convective heat transfer; Qh = effects of transpiration, condensation,
evaporation, and sublimation; and Qg = energy absorbed by the ground.
The net all-wave length radiation at the surface is Qn.
Qn = Qs Ql [40]

where: Qs = net short wave radiation; and Ql = net long wave radiation.
Qs = Qi Qr [41]

Ql = Qa Qe [42]

Qs has been given by Baker and Haines (1969):

S
Qs = a s R * A + B c [43]
100

where: as = surface short wave absorptivity of pavement surface;


R* = extraterrestrial radiation incident on a horizontal surface at the outer
atmosphere, which is a function of the latitude of the site; A, B = constants that
account for diffuse scattering and adsorption by the atmosphere; Sc = percentage of
sunshine which accounts for the influence of cloud cover.
In Equation [42], Qa, the long wave incoming radiation, and Qe, the outgoing
long wave radiation, are given by Equations [44] and [45]:

(
Qa = Q z 1 NW
100
) [44]

Qe = Q (1 NW
x ) [45]
100

Thus Ql in Equation [46] is:

(
Ql = (Q z Q x ) 1 NW
100
) [46]

where: Qz is the incoming long wave radiation given by Equation [47]; and
(
1 NW
100
)
represents the cloud cover correction:

Q z = sb Tair G J p [47]
10

where: N = cloud base factor (0.9 to 0.80 for cloud heights of 305 m to 1,830 m
(1000 ft to 6000 ft) (Geiger, 1959); W = 100-Sc (average cloud cover during day or
night); Tair = air temperature in oR; sb = Stefan-Boltzmann constant, 5.67 x 10-8
688 RMPD 8/2007. Water in Pavements

W/(m2 K4) (0.172 x 10-8 Btu/(hr ft2 oR4)); G = 0.77; J = 0.28; = 0.074; p = vapor
pressure of the air (1 to 10mm Hg); and Qx = outgoing long wave radiation without a
correction for cloud cover.

Q x = sb Ts4 [48]

where: = emissivity of the pavement (1 albedo) which depends on pavement


color, texture and temperature (A typical value is 0.93); and Ts = surface temperature
in oR.
The rate of heat transfer by convection, Qc, is given by:

Qc = H ( Tair Ts ) [49]

with Tair and Ts expressed in oF; and H = convection heat transfer coefficient.
The convection heat transfer coefficient, H, can be expressed as follows
(Dempsey et al., 1985; Vehrencamp, 1953):

H = 122.93 [ 0.00144Tm0.3U 0.7 + 0.00097( Ts Tair )0.3 ] [50]

where: Ts = surface temperature, in oC; Tair = air temperature, in oC; Tm = average of


surface and air temperature, in oK; and U = average daily wind speed in m/s.
The maximum value of the heat transfer coefficient is partly controlled by the
stability criteria established for the finite difference approach in computations within
the EICM. The suggested maximum value is 17 W/(m2K) (3.0 Btu/(hr ft2 oF)). The
effects of transportation, condensation, evaporation and sublimation (Qh) have been
neglected in the formation because they are either too small to be significant or the
effects cancel each other out in the energy balance.
The above calculations determine the surface temperature and thus control the
temperature throughout the underlying materials. The depth of frost is established by
comparing the computed temperatures with the freezing temperatures of the soil.
The depth of frost penetration has been identified as the position of the 30oF
isotherm.
After the amount of heat inflow/outflow due to convection and radiation at the
pavement surface is determined, this amount of heat is added/subtracted from the
quantity of heat at the upper boundary. The EICM iterates a single time step,
calculating a new temperature profile for the pavement system. This updated
temperature profile is used for convection and radiation calculations at the next time
step.
Environmental Effects in Pavement Design 689

5.2. Temperature data for flexible pavement analysis

For the purpose of the MEPDG, a base unit of one month is used for incremental
damage computations. In situations where the pavement is exposed to freezing and
thawing cycles, the base unit of 1 month is changed to 15 days duration to account
for rapid changes in the pavement material properties during frost/thaw period. The
EICM provides 0.1 hours (6 minutes) temperature over the analysis period. This
temperature for a given month (or 15-days) can be represented by a normal
distribution with a certain mean value () and the standard deviation (), N(,).
While the EICM calculates temperature on a relatively small time step of 0.1
hours, temperatures are output to the MEPDG summary files in two formats for
flexible pavement analysis. One of them is used for rutting and fatigue analysis
while the other is used for thermal fracture.
The EICM generates five quintile temperature values for each interval and at
each selected depths (i.e. for rutting and fatigue, temperature values are required at
the surface of the pavement structure and at mid-depth of all asphalt bound sub-
layers). Since the first sub-layer for the asphalt is always 1.27 cm (0.5 in), the
temperatures are provided at 0.64 cm (0.25 in) from the surface. No temperature
information is generated for any other type of layer, as it is not required for the
analysis. The surface temperature and the temperature at 0.64 cm (0.25 in) are used
to estimate the fatigue at the surface (top down cracking).
Thermal fracture analysis requires hourly temperature data. Temperature values
are required at the surface, at 1.27 cm (0.5 in), and at every 2.54 cm (1 in) within the
asphalt layer. This defines the temperature-depth relationship within the asphalt
layer. In addition to developing a temperature-depth profile for thermal fracture
module to predict cracking, temperatures at the surface and at 1.27 cm (0.5 in) are
used for estimating tensile strains.

5.3. Temperature data for rigid pavement analysis

For rigid pavement design, the main temperature data of interest is the
temperature profile through the PCC layer. EICM is configured to produce hourly
temperature profiles for a minimum of one full year. For most sites, EICM climatic
database provides 5 years of hourly data. The data are used in the prediction of
faulting and fatigue cracking in JPCP and punchouts in CRCP.
In the JPCP design module of the MEPDG software, the output from EICM is
further processed to obtain monthly distributions of hourly temperature gradients
through PCC. In this process the nonlinear temperature distribution is first converted
to equivalent linear temperature gradient based on stress equivalence. The
equivalent linear temperature gradient is the linear temperature gradient that would
690 RMPD 8/2007. Water in Pavements

produce the same curling stress as that produced by the actual nonlinear temperature
profile.
In addition, parameters such as number of freeze-thaw cycles, mean annual
precipitation, and mean annual freezing index are also computed from the
temperature information for use in the various JPCP and CRCP structural distress
models. Other uses of temperature data include the JPCP joint opening/closing
model and the CRCP crack width model.

6. Use of output generated by the EICM

Once the EICM generates the aforementioned information, the following outputs
are generated for use by other components of the MEPDG software, as explained in
this paper:
Composite environmental adjustment factors, Fenv, are computed for every
sublayer at each node. These Fenv factors are sent forward to the structural analysis
modules where they are multiplied by MRopt to obtain MR as function of position and
time.
Temperatures at the surface and at the midpoint of each asphalt bound sublayer
are subjected to statistical characterization for every analysis period. The mean,
standard deviation, and quintile points are sent forward for use in the fatigue and
permanent deformation prediction models.
Values of hourly temperature at the surface and at a set depth increment within
the bound layers for use in the thermal cracking model.
An average value of moisture content for each sublayer is reported for use in
the permanent deformation model for the unbound materials.
Temperature profile in the PCC hourly values are generated for use in the
cracking and faulting models for jointed plain concrete pavement (JPCP) and the
punchout model for continuously reinforced concrete pavement (CRCP).
Number of freeze thaw cycles and freezing index are computed for use in JPCP
performance prediction.
Relative humidity values for each month are generated for use in the JPCP and
CRCP modeling of moisture gradients through the slab.

7. Concluding remarks and calibration to local conditions

Any agency interested in adopting the design procedure described in this paper
should prepare a practical implementation plan. The plan should include training of
staff, acquiring of needed equipment, acquiring of needed computer hardware, and
calibration/validation to local conditions.
Environmental Effects in Pavement Design 691

The mechanistic-empirical design procedure represents a major improvement


and paradigm shift from existing empirical design procedures (e.g. AASHTO 1993),
both in design approach and in complexity. The use of mechanistic principles to
both structurally and climatically (temperature and moisture) model the
pavement/subgrade structure requires much more comprehensive input data to run
such a model (including axle load distributions, improved material characterization,
construction factors, and hourly climatic data. Thus, a significant effort will be
required to evaluate and tailor the procedure to the highway agency. This will make
the new design procedure far more capable of producing more reliable and cost-
effective designs, even for design conditions that deviate significantly from
previously experienced (e.g. much heavier traffic).
It is important to notice that the flexible pavement design procedures have been
calibrated using design inputs and performance data largely from the national LTPP
database, which includes sections located throughout significant parts of North
America. Whatever bias included in this calibration data is naturally incorporated
into the distress prediction models and; therefore, the national calibration may not be
entirely adequate for specific regions of the country or for other country; and a more
local or regional calibration may be needed.
The following is the recommended calibration/validation effort required to
implement the MEPDG:
Review all input data.
Conduct sensitivity analysis.
Conduct comparative studies.
Conduct validation/calibration studies.
Modify input defaults and calibration coefficients as needed.

Disclaimer and acknowledgment


This study was funded by the National Cooperation Highway Research Program
(NCHRP) under the Project 1-37A: Development of the 2002 Guide for the Design
of New and Rehabilitated Pavement Structures.
The findings, conclusions or recommendations either inferred or specifically
expressed in this document do not necessarily indicate acceptance by the National
Academy of Sciences, the Federal Highway Administration, or by the Association
for State Highway and Transportation Officials (AASHTO).

8. Bibliography

Baker D.G., Haines D.A., Solar Radiation and Sunshine Duration Relationship in the North-
Central Region and Alaska, Experiment Station, University of Minnesota, MN, Technical
Bulletin 262, 1969.
692 RMPD 8/2007. Water in Pavements

Berg R.L., Energy Balance on Paved Surface, U. S. Army Terrestrial Sciences Center,
Hanover, NH., 1968.
Dempsey B.J., Herlach W.A., Patel A.J., The Climatic-Material-Structural Pavement
Analysis Program, FHWA/RD-84/115, Vol.3, Final Report, 1985, Federal Highway
Administration, Washington D.C.
Drumm E.C., Reeves J.S., Madgett, M.R., Trolinger, W.D., Subgrade Resilient Modulus
Correction for Saturation Effects, ASCE Journal of Geotechnical and Geoenvironmental
Engineering, Vol. 123, No. 7, 1997, p. 663-670.
Fredlund D.G., Xing A., Equations for the Soil-Water Characteristic Curve, Canadian
Geotechnical Journal, Vol. 31, No. 4, 1994, p. 521-532.
Fredlund D.G., Xing, A., Fredlund, M., Barbour, S.L., The Relationship of the Unsaturated
Shear Strength to the Soil-Water Characteristic Curve, Canadian Geotechnical Journal,
Vol. 32, 1995, p. 440-448.
Geiger R., The Climate Near Ground, Harvard University Press, Cambridge, MA, 1959.
Guymon G.L., Berg R.L., Johnson T.C, Mathematical Model of Frost Heave and Thaw
Settlement in Pavement, Report, 1986, U.S. Army Cold Region Research and
Engineering Laboratory.
Larson G., Dempsey B.J., Integrated Climatic Model, Version 2.0, Report No. DTFA
MN/DOT 72114, 1997, Newmark Civil Engineering Laboratory, University of Illinois at
Urbana-Champaign, IL..
Lekarp, F., Isacsson U., Dawson A., State of the Art I: Resilient Response of Unbound
Aggregates, ASCE Journal of Transportation Engineering, Vol. 126, No. 1, 2000, p. 76-
83.
Leong, E.C., Rahardjo H., A Review on Soil-Water Characteristic Curve Equations,
Geotechnical Research Report, NTU/GT/96-5, 1996, Nanyang Technological University,
NTU-PWD Geotechnical Research Center, Singapore.
Li D., Selig E.T., "Resilient Modulus for Fine Grained Subgrade Soils", ASCE Journal of
Geotechnical Engineering, Vol. 120, No. 6, 1994, p. 939-957.
Lytton R.L., Pufahl D.E., Michalak C.H., Liang H.S., Dempsey B.J., An Integrated Model of
the Climatic Effects on Pavement, Report No. FHWA-RD-90-033, 1990, Federal
Highway Administration, Texas Transportation Institute, Texas A andM University
McLean, VA.
Rada G., Witczak M.W., Comprehensive Evaluation of Laboratory Resilient Moduli Results
for Granular Material, Transportation Research Record 810, TRB, National Research
Council, Washington, D.C., 1981, p. 23-33.
Santha, B.L., Resilient Modulus of Subgrade Soils: Comparison of Two Constitutive
Equations, Transportation Research Record 1462, TRB, National Research Council,
Washington, D.C., 1980, p. 79-90.
Schoeneberger Wysocki D.A., Benham E.C., Broderson W.D., Field Book for Describing and
Sampling Soils, (Version 1.1, refined, revised, and compiled), National Soil Survey
Environmental Effects in Pavement Design 693

Center, Natural Resources Conservation Service, U.S. Department of Agriculture,


Lincoln, NE., 1998.
Scott R.F., Heat Exchange at the Ground Surface, US Army Material Command, Cold
Regions Research and Engineering Laboratory, Hanover, N.H., 1964.
Vehrencamp J.E., Experimental Investigation of Heat Transfer at Air-Earth Interface,
Transportation, American Geophysical Union, Vol. 34, No. 1, 1953.
Witczak M.W., Andrei D., Houston W.N., Resilient Modulus as Function of Soil Moisture
Summary of Predictive Models, Development of the 2002 Guide for the Development of
New and Rehabilitated Pavement Structures, NCHRP 1-37 A, Inter Team Technical
Report (Seasonal 1), 2000a, Tempe, AZ.
Witczak M.W., Houston W.N., Andrei D., Resilient Modulus as Function of Soil Moisture
A Study of the Expected Changes in Resilient Modulus of the Unbound Layers with
Changes in Moisture for 10 LTPP Sites, Development of the 2002 Guide for the
Development of New and Rehabilitated Pavement Structures, NCHRP 1-37 A, Inter Team
Technical Report (Seasonal 2), 2000b, Tempe, AZ.
Witczak M.W., Houston W.N., Andrei D., Selection of Resilient Moduli for Frozen/Thawed
Unbound Materials, Development of the 2002 Guide for the Development of New and
Rehabilitated Pavement Structures, NCHRP 1-37 A, Inter Team Technical Report
(Seasonal 3), 2000c, Tempe, AZ.
Witczak M.W., Houston W.N., Zapata C.E., Richter C., Larson G., Walsh K., Improvement
of the Integrated Climatic Model for Moisture Content Predictions, Development of the
2002 Guide for the Development of New and Rehabilitated Pavement Structures, NCHRP
1-37 A, Inter Team Technical Report (Seasonal 4), 2000d, Tempe, AZ.
Zapata C.E., Uncertainty in Soil-Water Characteristic Curve and Impacts on Unsaturated
Shear Strength Predictions, Ph.D. Dissertation, 1999, Arizona State University, Tempe,
AZ.
Zapata C.E., Houston W.N., Houston S.L., Walsh K.D., Soil-Water Characteristic Curve
Variability, Advances in Unsaturated Geotechnics, Shackelford C.D., Houston S.L., and
Chang N-Y (eds). ASCE GEO Institute Geotechnical Special Publication, No. 99. Also
Proceedings of Sessions of Geo-Denver 2000, August 5-8, 2000, Denver, CO, p. 84-124.

Received: 22 March 2006


Accepted: 13 October 2006