Академический Документы
Профессиональный Документы
Культура Документы
FAULTING PREDICTION
INTRODUCTION
Transverse joint faulting is one of the main types of distresses in jointed concrete pavements
(JCP) that affects serviceability (1). Joint faulting is defined as the difference in elevation
between adjacent joints at a transverse joint measured approximately 1 ft from the slab edge
(longitudinal joint for a conventional lane width), or from the rightmost lane paint stripe for a
widened slab. Significant joint faulting has a major impact on the life cycle cost of the pavement
in terms of early rehabilitation and vehicle operating costs.
Repeated heavy axle loads crossing transverse joints create the potential for joint faulting, as
shown in figure 1. Faulting is the result of excessive slab edge and corner deflections that cause
erosion and pumping of fines from beneath a loaded leave slab. These fines are then deposited
under the approach slab. The rate of progression of JCP faulting is increased significantly when a
given pavement exhibits a combination of poor load transfer across a joint or crack, heavy axle
loads, free moisture beneath the pavement, and erosion and pumping of the supporting base,
subbase, or subgrade material from underneath the slab or treated base. The following
conditions must exist for faulting to occur (2):
Significant differential deflections of adjacent slabs that impart energy to the underlying
pavement materials. These deflections cause the movement of the saturated underlying
pavement material as equilibrium is reestablished, resulting in erosion and pumping. The
differential energy across the joint or crack is amplified by several factors, including heavy
wheel loads and inadequate load transfer.
Underlying pavement materials that are erodible. An unstabilized or weakly stabilized
material with a high percentage of fines is a prime candidate for erosion.
The presence of free water in the pavement structure, which leads to the saturation of the
underlying materials at the slab/base or treated base/subgrade interface and provides
transportation for pumping and erosion of fines.
When excess moisture exists in a pavement with an erodible base or underlying fine-grained
subgrade material, repeated vehicle loadings typically cause the mixture of water and fine
material (fines) to be removed from beneath the leave slab corner and ejected to the surface
through the transverse joint or along the shoulder. This process, commonly referred to as
pumping, will eventually result in a void below the leave slab corner. In addition, some of the
fines that are not ejected will be deposited under the approach slab corner, causing the approach
slab to rise. This combination of a buildup of material beneath the approach corner and the loss
of support resulting from a void under the leave corner can lead to significant faulting at the
joint.
This paper describes the faulting mechanism as well as a mechanistic-empirical model for
prediction of faulting in jointed plain concrete pavements (JPCP) developed under the NCHRP
1-37A study, Development of the 2002 Guide for Design of New and Rehabilitated Pavement
Structures. The overall objective of the 2002 Design Guide is to provide the highway
community with a state-of-the-practice tool for the design of new and rehabilitated pavement
structures based on mechanistic principles.
STATE OF THE ART OF FAULTING MODELING
Transverse joint faulting has been the focus of various field (3-9) and laboratory investigations
(2). A comprehensive summary of the available joint faulting prediction models can be found in
other sources (1, 9). Analysis of the available mechanistic-empirical models for faulting
predictions led to the following conclusions:
There are currently no mechanistic procedures for predicting the effect of the moisture state
of the underlying materials on faulting. Indirectly, however, this effect can be accounted for
by including seasonal variation of subgrade stiffness into prediction of structural responses.
There are currently no mechanistic procedures for determining the erodibility of the
underlying materials. However, several material-related variables have been identified as
affecting erodibility (e.g., base, subbase, and subgrade type, gradation, percent fines, and
treatment or stabilizer type and amount).
The stresses, deflections, and work or power to which the underlying pavement materials are
subjected have been satisfactorily modeled using current technology based on mechanistic
principles.
The FHWA PAVESPAC 3.0 faulting model (9) was found to be the most advanced among the
models evaluated in this study and was selected as a basis for faulting model adaptation. The
model relates differential energy of subgrade deformation to faulting development. PAVESPAC
3.0 faulting prediction depends on the traffic volume, dowel diameter, portland cement concrete
(PCC) slab and base properties, subgrade support, and climatic conditions. Nevertheless, the
model has significant limitations:
The model uses equivalent single axle loads (ESALs) to characterize traffic. The 2002
Design Guide models are supposed to use axle spectrum distributions for traffic
characterization.
Like other available faulting models (3-8), the PAVESPAC 3.0 model uses average
pavement parameters (load transfer efficiency [LTE], PCC slab properties, subgrade
support conditions). The 2002 Design Guide models are supposed to use incremental
damage approach. Therefore, it is possible to account for deterioration of LTE and the
increase in PCC slab stiffness over time.
The model neglects seasonal and environmental effects on faulting development.
Incorporation of the Enhanced Integrated Climatic Models (EICM) into the 2002 Design
Guide permits more realistic modeling the effects of such factors as seasonal variation
in subgrade k-value, PCC slab warping, and curling.
To address these limitations, the PAVESPAC 3.0 model was further enhanced in this study. The
key elements of the 2002 design procedure faulting model are presented below.
The differential energy of subgrade deformation is defined as the energy difference in the elastic
subgrade deformation under the loaded slab (leave) and unloaded slab (approach):
k
k
= L UL
2
2
2
DE = EL EUL
(2)
where:
DE is differential energy of subgrade deformation
EL is energy of subgrade deformation under the loaded slab corner
EUL is energy of subgrade deformation under the unloaded slab corner
L is the corner deflection under the loaded slab
UL is the corner deflection under the unload slab
Equation 2 may be re-written in the following form:
DE = EL EUL =
k
( L UL )( L + UL )
2
(3)
The term ( L + UL ) is equal to the free corner deflection. It represents the total flexibility of the
slab. The higher the slabs flexibility, the greater the differential energy and the joint faulting
potential. The term ( L UL ) is the differential corner deflection between the loaded slab and
unload slab corner. It represents the relative movement between the loaded and unloaded slabs.
The greater the difference, the higher the joint faulting will be. Without any differential
deflections at the corner, there will be no faulting, as seen at the AASHO Road Test. The
differential corner deflection depends on the free corner deflection and the deflection load
transfer efficiency, LTE. If the latter is defined as
LTE =
UL
100%
L
(4)
(5)
One can see that joint load transfer efficiency has a major effect on the differential energy of
subgrade deformation.
MODELING OF JOINT LOAD TRANSFER EFFICIENCY
The main difference between the 2002 Design Guide faulting model and PAVESPAC 3.0 is that
the 2002 model accounts for incremental deterioration of transverse joints. Joint deterioration
reduces joint load transfer efficiency, increases the magnitude of differential PCC slab deflection
across the joint, and as a result, increases the magnitude of differential energy of subgrade
deformation for the same traffic level and faulting development. Since this concept has a
paramount importance on faulting prediction, it is described in a greater detail below.
The 2002 Design Guide joint model characterizes joint stiffness through deflection LTE
measured for a flat slab conditions (i.e., assuming that no significant separation between the
foundation and the PCC slab exists). Testing early in the morning (from about 6 to 8 a.m.)
usually closely resembles these conditions.
If a joint exhibits a poor ability to transfer load, then the deflection of the unloaded slab is much
less than the deflection at the joint of the loaded slab, and the LTE defined by equation 4 has
values close to 0. If a joints load transfer ability is very good, then the deflections at the both
sides of the joint are equal, and the LTE has a value close to 100 percent.
For transverse joints, the total deflection LTE includes the contribution of three major
mechanisms of load transfer:
(6)
where:
LTEjoint is total joint LTE, percent
LTEdowel is joint LTE if dowels are the only mechanism of load transfer, percent
LTEbase is joint LTE if the base is the only mechanism of load transfer, percent
LTEagg is joint LTE if aggregate interlock is the only mechanism of load transfer, percent
Modeling of each of these LTE mechanisms is presented below.
Aggregate Interlock LTE
The 2002 faulting prediction model adopted the Zollinger et al. aggregate interlock model (13).
This model states that the non-dimensional stiffness of an aggregate joint is a function of the load
shear capacity, S.
S e
where:
JAGG =(Agg/kl) is joint stiffness of the transverse joint for current increment
e is constant equal to 0.35
f is constant equal to 0.38
(7)
(9)
where:
sh,mean is PCC slab mean shrinkage strain
PCC is PCC coefficient of thermal expansion, in/in/0F
JTSpace is joint spacing, ft
is joint open/close coefficient assumed equal to 0.85 for a stabilized base and 0.65 for a
unbound granular base
Tmean is mean monthly nighttime mid-depth temperature, 0F
Tconstr is PCC zero-stress temperature at set, 0F, defined as the temperature (after
placement and during the curing process) at which the PCC layer exhibits zero thermal
stress.
Loss of shear capacity is also computed on a monthly basis. Each axle load application
contributes toward loss of shear. The cumulative loss of shear at the end of the month (i.e., in
the beginning of the next design month) is determined as follows:
end
b
stot
= stot
ni si
(10)
where:
b
is cumulative loss of sheer capacity at the end of the current month equal to sum of
stot
loss of shear capacity form every axle load application.
ni is the number of application of axle load i.
si is loss of capacity shear due to single application of an axle load i defined as follows:
6
i
0.005 *10
si =
5.7
jw
h
1
.
0
(
/
)
PCC
ref
i
0.068 *10 6
1.98
1.0 + 6.0 * ( jw / hPCC 3)
ref
if
if
(11)
where:
i is shear stress on the transverse joint surface from the response model for the load
group i
ref is reference shear stress derived from the PCA test results.
Figure 2a presents an example of predicted aggregate interlock component of joint LTE over the
pavement design life. One can observe two major trends in aggregate interlock LTE variation:
Seasonal variability
Reduction with time
Seasonal variation comes from the change in joint opening due to seasonal changes in mean PCC
temperature. LTE reduction with time comes from the loss of shear capacity and the increase in
joint opening due to shrinkage.
Doweled Joint Load Transfer
Ioannides and Korovesis identified the following nondimensional parameters governing dowel
joint behavior (14):
JD =
where:
D
skl
(12)
Ad
118,
> 0.656
if
h
PCC
A
A
Ad
0.4, if
< 0.009615
hPCC
(15)
where:
J0 is initial non-dimensional dowel stiffness
J*d is critical non-dimensional dowel stiffness
Ad is area of dowel cross-section:
d2
Ad =
4
(16)
where:
d is the dowel diameter
Dowel joint damage accumulated from an individual axle repetition is determined using the
following equation:
DOWDAM i = C8 *
where:
Fj,A
(17)
d f c*
where:
Jd is non-dimensional dowel stiffness at the time of load application
L is deflection at the corner of the loaded slab induced by the axle
U is deflection at the corner of the unloaded slab induced by the axle
DowelSpace is the space between adjacent dowels in the wheel path, in
Like aggregate interlock LTE, dowel LTE deteriorates with time, as shown in figure 2b.
(18)
10
Even if no dowels in the base and joints are open so widely that no aggregate interlock exists,
joint load transfer efficiency is not equal to zero. This indicates that a certain portion of load is
transferred from the loaded to unloaded slab through the base, subbase, and subgrade pavement
layers. The 2002 design procedure accounts for this effect by assigning a percentage of load
transfer efficiency of the base layer, LTEbase, depending on the base layer type. For the majority
of cases, the value of the load transfer efficiency can be determined from table 1 (as it is assumed
in the Design Guide software). As can be observed from table 1, the 2002 design procedure
assumes that a properly designed stabilized layer provides a better load transfer efficiency than a
granular base.
It is recognized, however, that if the pavement system is frozen, then the load transfer efficiency
of joints increases. Considering that joints are open the most in cold temperatures (i.e., the
aggregate portion of the load transfer efficiency is lower than in warm weather), this increase in
LTE should come from the increase of the load transfer efficiency of the base layer. To account
for this effect, the 2002 Design Guide software calculates the mean monthly mid-depth PCC
temperatures. If for a given month this temperature is less than 32 F, the LTEbase is set to 90
percent for that month.
STRUCTURAL RESPONSE MODELING FOR FAULTING
Determination of the differential energy of subgrade deformation and load transfer efficiency
parameters presented above requires prediction of deflections at the corner of loaded and
unloaded slabs from a single, tandem, tridem, or quad axle located close to the approach slab
corner, as shown in figure 1. While many of the parameters above remain constant throughout
the design period (e.g., slab thickness and joint spacing), others vary seasonally, monthly, or with
pavement age. For accurate faulting analysis results, all cases that produce significantly different
deflections must be evaluated separately for each month of the pavement life to account for the
following factors:
11
difference (top minus bottom). The equivalent temperature gradient for month m is
determined as follows:
Tm = Tt ,m Tb,m + Tsh ,m + TPCW
(19)
where:
Tm is the effective temperature differential for the month m.
Tt,m is the mean PCC top-surface nighttime temperature (from 8 p.m. to 8 a.m.)
for the month m.
Tb,m is the mean PCC bottom-surface nighttime temperature (from 8 p.m. to 8
a.m.) for the month m.
Tsh,m is equivalent temperature differential due to reversible shrinkage for the
month m for old concrete (i.e. shrinkage is fully developed).
TPCW is equivalent temperature differential due to permanent curl/warp.
The incremental design procedure requires thousands of deflection calculations to compute
damage monthly (for the different loads, joint stiffnesses, and equivalent temperature
differences) over a design period of many years. These computations would take hours (if not
days) using existing finite element programs. Thus, it is not practical to include a finite element
program with the design guide software at this time. To reduce computer time to a practical
level, neural networks (NNs) have been developed based on a large number of finite element
analysis runs made using ISLAB2000 (15). This makes it possible to conduct detailed
incremental analysis (month by month) to sum damage over time in a realistic way. The neural
networks reproduce the same deflections very accurately, given the set of required inputs.
Neural networks were developed separately for single, tandem, and tridem axles. A concept
developed specifically on this studythe Equivalent Pavement Structure was used extensively
to reduce the number of independent parameters of the neural network and speed up its training
(16). The proposed rapid solution provides a good match of the ISLAB2000 deflection values
for a small fraction of the computation cost. The neural networks directly incorporate all factors
listed above and closely match ISLAB2000 deflections for a wide range of input parameters, as
shown in table 2.
2002 Design Guide Faulting Model General Overview
The mean transverse joint faulting is predicted using incremental approach as illustrated in figure
3. A faulting increment is determined each month and the current faulting level affects the
magnitude of increment. The faulting at each month is determined as a sum of faulting
increments from all previous months in the pavement life since the traffic opening using the
following model:
m
Fault m = Fault i
(20)
(21)
i =1
(22)
P * WetDays
)
FAULTMAX 0 = C12 * curling * Log (1 + C 5 * 5.0 EROD ) * Log ( 200
ps
12
C6
(23)
where:
Faultm is mean joint faulting at the end of month m, in.
Faulti is incremental change (monthly) in mean transverse joint faulting during month i, in.
FAULTMAXi is maximum mean transverse joint faulting for month i, in.
FAULTMAX0 is initial maximum mean transverse joint faulting, in.
EROD is base/subbase erodibility factor
DEi is differential deformation energy accumulated during month i
EROD is base/subbase erodibility factor
curling is maximum mean monthly slab corner upward deflection PCC due to temperature
curling and moisture warping
PS is overburden on subgrade, lb
P200 is percent subgrade material passing #200 sieve
WetDays is average annual number of wet days (greater than 0.1 in rainfall).
C12 = C1 + C 2 * FR 0.25
C 34 = C 3 + C 4 * FR 0.25
(24)
(25)
FR is base freezing index defined as percentage of time the top base temperature
is below freezing (32 F) temperature.
C1 through C7 are calibration constants.
The functional form of the model reflects the hypothesis that faulting potential depends on the
amount of the PCC slab curling, base erodibility, and presence of fines and free water in the
subgrade. Faulting potential decreases with increase of overburden pressure on the subgrade.
The rate of faulting development depends on the faulting level and decreases when faulting
increases until it stabilizes at a certain level.
Prediction of transverse joint faulting in the 2002 design procedure involves the following steps:
1. Tabulate input data summarize all inputs needed for predicting JPCP transverse joint
faulting.
2. Process traffic data the processed traffic data needs to be further processed to determine
equivalent number of single, tandem, and tridem axles produced by each passing of
tandem, tridem, and quad axles.
3. Process pavement temperature profile data the hourly pavement temperature profiles
generated using EICM (nonlinear distribution) need to be converted to effective
nighttime differences by calendar month.
4. Process monthly relative humidity data the effects of seasonal changes in moisture
conditions on differential shrinkage is considered in terms of monthly deviations in slab
warping, expressed in terms of equivalent temperature differential.
5. Calculate initial maximum faulting.
6. Evaluate joint LTE.
7. Calculate current maximum faulting.
13
The JPCP transverse joint faulting model given in equations 19 through 24 is a result of
calibration based on performance of 248 field sections located in 22 states. The calibration
sections consist of 138 Long Term Pavement Performance (LTPP) program GPS-3 and SPS-2
sections and 110 sections from the FHWA study Performance of Concrete Pavements (6).
Time-series data were available for many of the sections, making the total number of field
cracking observations 535.
The process of calibration involved determination of the calibration parameters C1 through C7
from equation 20 through 25 and rate of dowel deterioration parameter, C8, from equation 17
which minimizes the error function, ERR, defined as
ERR(C1 , C 2 ,..., C8 ) =
Nob
(FaultP redicted
ob =1
ob
FaultMeasured ob )
(26)
where:
ERR is error function
C1 , C 2 ,..., C8 are calibration parameters
FaultP redicted ob is predicted faulting for observation ob in the calibration database
FaultMeasured ob is measured faulting for observation ob in the calibration database
Nob is the number of observations in the calibration database
Using non-linear optimization techniques, the following values of the calibration parameters
were found:
C1
C2
C3
C4
=
=
=
=
1.29
1.1
0.001725
0.0008
C5
C6
C7
C8
=
=
=
=
250
0.4
1.2
400
A plot of predicted versus measured faulting is shown in figure 4. The diagnostic statistics and
the plot verify that the model is effective for predicting transverse joint faulting. The R2 of 71
percent and SEE of 0.0267 in (0.68 mm) are very reasonable, given the large number of data
points (N = 534) used in model development.
14
A comprehensive sensitivity analysis of the key variables in the developed model was conducted
in this study to ensure the predicted faulting is reasonable for a wide range of input design
parameters. Several examples of this analysis are provided in figures 5 through 9. These plots
are drawn by fixing all other variables at a constant value and allowing the selected variables to
vary over a range of values. As shown in the graphs, the model predicts that faulting increases
with time (traffic).
The sensitivity of JPCP joint faulting to dowel diameter is shown in figure 5. One can see that
JPCP joint faulting is highly sensitive to dowel diameter. The use of properly sized dowels is
generally the most reliable and cost-effective way to control joint faulting. A slight increase of
diameter of the dowels (e.g., 0.25 in) will significantly increase joint shear stiffness and reduce
the mean steel-to-PCC bearing stress and thus the joint faulting.
Figures 6a and 6b illustrate the effect of the base erodibility on faulting prediction for nondoweled and doweled pavements, respectively. The treating of nonstabilized aggregate base
with adequate amounts of asphalt or cement will reduce the erosion potential of the base.
Stabilized dense and open graded bases have lower erodibility than aggregate bases. One can see
that lower faulting is expected if less erodible bases are used.
A tied PCC shoulder (especially one constructed monolithically with the mainline) provides
better edge and corner support than an asphalt concrete (AC) shoulder and reduces the deflection
of the slab and the potential for erosion and pumping, especially for non-doweled pavements.
Figures 7a and 7b illustrate the effect of shoulder type on faulting in non-doweled and doweled
pavements, respectively.
Widening the slab effectively moves the wheel load away from the slab corner, greatly reducing
the deflection of the slab and the potential for erosion and pumping. Studies have shown that
slab widening can reduce faulting. Figures 8a and 8b illustrate that lower faulting is expected if
widened slab is used for non-doweled and doweled pavements, respectively.
Generally speaking, shorter joint spacing results in smaller joint openings. Thus, aggregate
interlock has a more useful effect on maintaining higher LTE. Moreover, the effects of curling
and warping on corner deflections are less pronounced for shorter slabs. Figures 9a and 9b show
that predicted mean joint faulting is smaller for shorter joint spacing. However, the effects of
joint spacing on total faulting will be somewhat less due to increased number of joints.
CONCLUSION
This paper presents a summary of the procedures used to model the effects of transverse joint
faulting in the design of JPCP in the 2002 Design Guide. The developed performance
prediction model is a substantial enhancement of the FHWA PAVESPAC 3.0 faulting prediction
model. Like the FHWA faulting model, the new model relates differential energy of subgrade
deformation to faulting development. The 2002 Design Guide model retains all positive features
15
of the PAVESPAC 3.0 model. It is capable of accounting for the effects of traffic volume, dowel
diameter, PCC slab and base properties, subgrade support, and climatic conditions on faulting
prediction. In addition, the model has the following advantages:
It uses axle spectrum distributions for traffic characterization.
It uses an incremental damage approach that accounts directly for changes in LTE and
PCC stiffness over time.
It accounts directly for seasonal and environmental effects on faulting development by
considering seasonal variation in subgrade k-value, PCC slab warping, and curling.
The model was calibrated using a comprehensive performance database with sections located
throughout the United States.
ACKNOWLEDGMENTS
The research described in this paper was conducted by the ERES Consultants Division of
Applied Research Associates, Inc., under the NCHRP 1-37A study. Dr. Amir Hanna served as
NCHRP Program Officer. The authors would like to acknowledge the database assembly
contribution by Leslie Titus-Glover and integration of the model into the 2002 Design Guide
software by Gregg Larson from ERES.
REFERENCES
1. Khazanovich, L., M. Darter, and R. Bartlett, Common Characteristics of Good and Poorly
Performing PCC Pavements. Washington, DC: Federal Highway Administration, 1997
2. Permanent International Association of Road Congresses. Combating Concrete Pavement
Slab Pumping by Interface Drainage and Use of Low-Erodability Materials: State of the Art
and Recommendations, Permanent International Association of Road Congresses, Paris,
France, 1987.
3. Darter M. I., J. M. Beck, M. B. Snyder, and R. E. Smith, Portland Cement Concrete
Pavement Evaluation System-COPES, NCHRP Report 277, Transportation Research Board,
Washington, DC, 1985.
4. Wu, C. L., J. W. Mack, P. A. Okamoto, and R. G. Packard "Prediction of Faulting of Joints
in Concrete Pavements," Proceedings, Fifth International Conference on Concrete Pavement
Design and Rehabilitation, Vol. 2, Purdue University, West Lafayette, IN, April 1993.
5. Simpson, A. L., J. B. Rauhut, P. R. Jordahl, E. Owusu-Antwi, M. I. Darter, and R. Ahmad,
Early Analysis of LTPP General Pavement Studies Data, Volume 3: Sensitivity Analyses for
Selected Pavement Distresses, Report SHRP-P-393, Strategic Highway Research Program,
Washington, DC, 1994.
6. Yu, H.T., M.I. Darter, K.D. Smith, J. Jiang and L. Khazanovich. Performance of Concrete
Pavements Volume III - Improving Concrete Pavement Performance. Final Report, Contract
DTFH61-91-C-00053, Federal Highway Administration, McLean, VA, 1996.
7. Owusu-Antwi, E.B., L. Titus-Glover, L. Khazanovich, and J. R. Roesler. "Development and
Calibration of Mechanistic-Empirical Distress Models for Cost Allocation." Final Report,
Federal Highway Administration, Washington, DC, March 1997.
16
17
Midpoint
Traffic Lane
Tandem
Axle
Load
Shoulder
Critical location
(deflections across joint)
Figure 1. Critical load and structural response location for JPCP joint faulting analysis
18
50
45
40
35
30
25
LTE
20
15
10
5
0
0
50
100
150
200
a. Nondoweled LTE
100
90
80
70
60
50
LTE
40
30
20
10
0
0
50
100
150
200
b. Doweled LTE
Figure 2. Example of predicted LTE components (flat slab conditions)
19
Compute
y Corner deflections
y Shear stresses
y Dowel bearing stresses
All Time
Increments
Completed?
Figure 3. Flowchart showing the transverse joint faulting prediction process (computations will
be performed by a neural network program)
20
0.4
y = 0.9914x
2
R = 0.709
SEE = 0.0288 in
0.35
Measured faulting, in
0.3
0.25
0.2
0.15
0.1
0.05
0
-0.05
0.05
0.1
0.15
0.2
0.25
0.3
-0.1
Predicted faulting, in
0.35
0.4
21
0.3
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
Wet-freeze climate
0.25
AC shoulder
Faulting, in
0.2
no dowels
d = 1 in
d = 1.25 in
d = 1.375 in
d = 1.5 in
0.15
0.1
0.05
0
0
50
100
150
200
250
300
Age, months
22
0.3
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
No dowels
AC shoulder
0.25
EROD=5
EROD=4
EROD=3
EROD=2
EROD=1
0.15
0.1
0.05
0
0
50
100
150
200
250
300
Age, months
a. nondoweled pavement
0.12
Illinois
9-in JPCP
1-in dowels
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
1.25-in dowel
AC shoulder
0.1
0.08
Faulting, in
Faulting, in
0.2
EROD=5
EROD=4
EROD=3
EROD=2
EROD=1
0.06
0.04
0.02
0
0
50
100
150
200
250
300
Age, months
b. doweled pavement
Figure 6. Effect of base erodibility on predicted faulting
23
0.3
0.25
AC shoulder
PCC shoulder
PCC monolitic
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
Wet-freeze climate
0.15
0.1
No dowels
0.05
0
0
50
100
150
200
250
300
Age, months
a. non-doweled pavement
0.1
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
Wet-freeze climate
0.09
0.08
0.07
Faulting, in
Faulting, in
0.2
1.25-in dowels
0.06
AC shoulder
PCC shoulder
PCC monolitical
0.05
0.04
0.03
0.02
0.01
0
0
50
100
150
200
250
300
Age, months
b. dowel pavement
Figure 7. Effect of shoulder type on predicted faulting
24
0.3
0.25
Widened slab
Standard width slab
0.15
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
Wet-freeze climate
0.1
No dowels
0.05
0
0
50
100
150
200
250
300
Age, months
a. non-doweled pavement
0.1
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
Wet-freeze climate
0.09
0.08
0.07
Faulting, in
Faulting, in
0.2
WL +PCC
0.06
widened slab
standard width slab
0.05
0.04
0.03
0.02
0.01
0
0
50
100
150
200
250
300
Age, months
b. doweled pavement
Figure 8. Effect of slab widening on predicted faulting
25
0.4
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
No dowels
AC shoulder
0.35
0.3
Faulting, in
0.25
JTSPACE=12 ft
JTSPACE=15 ft
JTSPACE=18 ft
JTSPACE=20' ft
0.2
0.15
0.1
0.05
0
0
50
100
150
200
250
300
Age, months
a. non-doweled pavements
0.16
Illinois
9-in JPCP
6-in aggregate base
Design period = 20 years
ADTT = 2,000 (TTC 8)
4% compound growth
1.25-in dowel
0.14
0.12
AC shoulder
Faulting, in
0.1
JTSPACE=12 ft
JTSPACE=15 ft
JTSPACE=18 ft
JTSPACE=20' ft
0.08
0.06
0.04
0.02
0
0
50
100
150
200
250
300
Age, months
b. doweled pavements
Figure 9. Effect of joint spacing on predicted faulting
26
LTEBase
Aggregate base
20%
30%
LCB base
40%
27
Table 2. Ranges of input parameters for the neural networks computing corner deflections for
JPCP faulting analysis.
Input Parameter
Radius of relative stiffness a
Joint spacing
Transverse joint LTE
Shoulder LTE
Axle offset from the slab edge
Temperature difference (top bottom)
Axle weight, single axle
Axle weight, tandem axle
Axle weight, tridem axle
Axle weight, quad axle
Tandem and tridem axle spacing
a
b
c
Minimum Value
22.5 in
12 ft
0%
0%
0 in
0 F
0 lb
0 lb
0 lb
0 lb
40 in
Maximum Value
80 in
30 ft b
95%
90%
36 in
> 55 F c
45,000 lb
90,000 lb
135,000 lb
135,000 lb
70 in
The radius of relative stiffness of highway pavements typically fall between 22.5 and 80 in. Analyses
based on plate theory become increasingly inaccurate for the radius of relative stiffness values beyond the
limit shown above.
The results for 30-ft slab are given for the actual joint spacing greater than 30 ft. In general, long joint
spacing (20 ft or greater) is not recommended because of excessive curling stress.
Depends on PCC coefficient of thermal expansion, k-value, PCC unit weight, PCC thickness, and radius of
relative stiffness.