The Effect of Slab Curvature on IRI Values for Jointed

Concrete Pavements
Christopher R. Byrum, PhD, PE1

Abstract

This paper presents the results of a study showing the effect pavement slab curvature
has on International Roughness Index (IRI) values for jointed concrete pavements. A
set of idealized curved slab profiles were generated using connected circular arcs
having varying lengths and magnitudes of constant curvature covering the range of
curvature encountered in real pavements. IRI values are calculated for these profiles
and a simple regression equation that approximates IRI for varying slab length and
curvature magnitude is presented. The ideal slab curvature vs. IRI trends are then
compared to existing profile curvature estimates and IRI data for the LTPP GPS3 test
sections for over 1000 profiles. The IRI trends for the ideal profiles match well with
the curvature vs. IRI trends for the GPS3 data. A discussion regarding how to fairly
take into account varying slab length when analyzing slab curvature data for profiles
is provided.

Introduction

The Federal Highway Administration’s (FHWA) Long Term Pavement Performance

(LTPP) study is the largest and most comprehensive pavement study ever undertaken.
The pavement profile data within LTPP is an excellent large controlled data set of
pavement profiles and has great research potential. This study compares slab
curvature data from the LTPP GPS3 pavement profiles to hypothetical ideal profiles
comprised of repeating constant curvature arcs having varying slab length and
curvature magnitude. This data is used to quantify the effect of profile wheel path
slab curvature on International Roughness Index (IRI) values for jointed concrete
pavements.
Figure 1 shows a pavement elevation profile from the LTPP GPS3 pavement
group for a rough JPCP, which has unusually high locked-in curvature in the slabs. It
is the GPS3 test pavement identified as 55-3009, which was very rough and faulted at
less than ten years old. Site 55-3009 is the test pavement with the most rapid
development of roughness (International Roughness Index) in the FHWA’s LTPP
GPS3 jointed concrete pavement research data. The elevation difference between the

________________________
1
Project Engineer, Soil and Materials Engineers, Inc. 43980 Plymouth Oaks Blvd.
Plymouth, Michigan 48170 (phone: 734-454-9900).

23
Byrum

joint and center slab along the wheel path is as high as 10 mm (0.4 inches) for slab
lengths of about 5.5 m (18 feet). Temperature gradients (curling effects) required to
cause this much uplift at joints for a flat slab would be around +0.22 C°/mm (10
F°/in) from the slab surface downward. For a flat 254 mm (10-inch) thick slab, the
top of slab would have to be about 67 C° (120 F°) cooler than the bottom, an
unrealistic thermal gradient (Byrum, 2001a). It is a combination of long-term creep,
warping, and construction related locked-in curvature that caused such large overall
curvature to develop in the pavement slabs shown in figure 1. By the 1920’s,
Westergaard had noted that based on the general understanding of concrete shrinkage
existing at that time, warping curvatures equivalent to that caused by temperature
gradients of 0.11 C°/mm (5 F°/in) were possible (Westergaard, 1927). Obviously,
locked-in curvatures in slabs can be many times greater than curvature caused by
typical weather related temperature gradients, which are typically less than about
±0.07 C°/mm (3 F°/in). It is well known that high degrees of locked in curvature in
concrete pavement slabs will lead to premature deterioration of the slabs.
.

LW P
G P S 3 S it e 5 5 - 3 0 0 9 RW P
15

10
Elevation, mm

-5
10± mm
-1 0

-1 5
50 60 70 80 90 100
D is t a n c e , m
NOTES:
1. Highest IRI of all GPS3 pavements less than 10 years old.
2. High faulting at an early age.
3. Sandy clay subgrade (PI=14.5, w%=13), sand and gravel with 11% P200 base.
4. Built in 1984, 300-350 KESAL/yr, 220 mm PCC, no dowels.
Profile Date: 6/19/1992
ProfileTime: 3:51:44 PM

Figure 1. Profile data from a GPS3 JPCP with large curvature.

(average curvature for the 152.4 m profile ≈ 1.62 1/mx1000)

In the 1940’s, Hveem referred to slabs having significant locked-in upward warped
shape as “man-made pumping machines” susceptible to faulting (Hveem, 1951).
Although it has long been accepted that slab shape factors affect long-term
performance, clear statistical trends showing the relative contributions slab curvature
on IRI have not been established. This fact and because IRI is the roughness statistic

24
Byrum

used by FHWA to track the overall condition of the United States of America’s
freeways were the motivating forces for this research. This research uses the FHWA
LTPP GPS3 pavement profile data to perform one of the first known studies showing
the effect of average slab curvature on IRI values

Effect of Pure Curved Slab Wheel Paths on IRI Values

Figure 2 shows a set of idealized pavement profiles used for this study. The profiles
shown consist of repeating connected segments of circular arcs all having constant
curvature magnitude of 0.00231 m-1 (0.000065 ft-1) simulating up-warped pavement
slabs. It is important to note that for the same magnitude of slab curvature, longer
slab lengths result in more joint uplift and steeper slopes at joints. The mid-slab to
joint elevation differences for the 4.9 m, 6.1 m, and 7.3 m (16-ft, 20-ft and 24-ft)
slabs are 6.4 mm, 9.9 mm, and 14.2 mm (0.25-in, 0.39-in and 0.56-in), respectively,
increasing as a function of length squared. Slope values near the joints are about
0.50%, 0.63%, and 0.76%, respectively, increasing linearly with slab length. For this
study, profiles made of the three arc (slab) lengths were generated with varying
magnitudes of arc segment curvature. For each slab length, idealized repeating arc
segment profiles with curvature magnitudes of 0.000164, 0.000656, 0.001148,
0.001640, and 0.002133 m-1 (0.00005, 0.0002, 0.00035, 0.00050, and 0.00065 ft-1)
were developed. These three slab lengths shown all have a common repeating pattern
length of 73.152 m (240 feet). The ideal profiles analyzed were 146.304 m (480 feet)
in length, which includes two of the common repeating pattern segments.

Effect of Slab Length on Profile Data for Curved Slabs

7.32 m (24-ft)
16 4.87 m (16-ft) Slab Length = 6.1 m (20-ft)
14
Elevation, mm

12
10
8
6
4
2
0
0 10 20 30 40 50 60 70
Curvature Magnitude = 2.13 1/mx1000
Distance, m

Figure 2. Idealized profile elevations for jointed concrete roadways having

varying slab length with constant wheel path curvature magnitude.

IRI values were then calculated for all of the idealized profiles. Figure 3
shows the resulting IRI trends for the three slab length classes, for increasing
magnitudes of curvature. The IRI model responds linearly to increasing slab
curvature for a given slab length. IRI increases roughly exponentially with respect to
slab length at a constant curvature magnitude.

25
Byrum

The 15 data points shown in figure 3 were used to develop an equation to

predict IRI caused by slab curvature for slab curvature magnitude and slab (arc)
length combinations. The following simple regression expression can be used to
predict the idealized IRI for any combination of slab length (4m<length<8m) and
wheel path curvature magnitude:
0.1595(Arc Length)
IRIcurvature = 0.5945(Curvature Magnitude)e

Where,

IRIcurvature = International Roughness Index of idealized repeating arcs, m/km.

Curvature Magnitude = repeating arc (slab) curvature in units of 1000/m.
Arc Length = Length of the repeating arcs (slabs) in meters (4 m < L < 8 m).
e = 2.7182818245904, the base of the natural logarithm.

4.5
4.0 4.9 m Length Arcs
3.5 6.1 m Length Arcs
3.0 7.3 m Length Arcs
IRI (m/km)

2.5
2.0
1.5
1.0
0.5
0.0
0.0 0.5 1.0 1.5 2.0 2.5
Repeating Arc Curvature (1/m X 1000)

Figure 3. Resulting IRI trends for the idealized repeating arc profiles.

Curvature in the LTPP GPS3 Jointed Concrete Pavements

The author has developed a method for obtaining what is referred to as the Curvature
Index, CI, for road profile data (Byrum, 2001a; Byrum, 2001b; Byrum, 2001c;
Byrum, 2004). The CI provides an estimate of the average slab curvature present
within a 152.4 m (500-ft) pavement profile sample. The method used for establishing
the curvature index is described elsewhere and is based on establishing the frequency
distributions of finite difference arc curvature magnitudes for the continuous profile
segments between obvious cracks and joints, using many arc samples ranging from
0.305 m to 2.438 m (1-ft to 8-ft) in length. Curvature statistics were obtained for
most of the 152.4 m (500-ft) long LTPP GPS3 and GPS4 jointed concrete pavement
profiles obtained during the 1990’s. The overall curvature index for a given profile is
the average of several arc sampling methods for estimating the slab segment

26
Byrum

curvatures. These curvature statistics are summarized below and compared to the
reported IRI values for the profiles.
Figure 4 shows the frequency distributions for average slab curvatures
observed in the GPS3 pavement test sections. The plot is broken down into two
distributions, one for doweled pavements and one for pavements with no dowels at
joints. The addition of dowels at joints provides restraint to the development of slab
curvature and faulting. Pavements without dowels have a greater tendency to
develop an upward warped shape, where the joints are lifted relative to the interior
portion of the slabs. Notice that both of the frequency distributions do not have mean
values in the downward curved (negative) region even though the pavement profiling
was typically performed during daytime conditions. Given that thermal gradients
during daytime conditions are almost always warm on top, the frequency distributions
show that there is a general condition of locked in upward curvature for both doweled
and non-doweled jointed concrete pavements, because the average values are about
zero (flat) for doweled pavements, and about 0.00066 m-1 (0.0002 ft-1) for slabs

Effect of Dowels on GPS3 PCC Slab Curvatures

100
# Profiles in Category

75

No Dowels
50
Dowels

25

0
8

00

16

33

49

66

82

98

15

31

48

64

80

97
.9

.8

.6

.4

.3

.1

0.

0.

0.

0.

0.

0.

0.

1.

1.

1.

1.

1.

1.
-0

-0

-0

-0

-0

-0

Avg. Slab Curvature, 1/m x 1000

No Dowels LCI RCI Dowels LCI RCI

Minimum -0.68 -0.68 Minimum -0.54 -0.61
Maximum 1.62 1.70 Maximum 1.30 1.28
Average 0.21 0.22 Average -0.03 -0.02
Std Dev 0.34 0.35 Std Dev 0.27 0.26
Count 351 351 Count 218 218

LCI = Left wheel path average curvature

RCI = Right wheel path average curvature

Figure 4. Frequency distributions for average slab curvature in the wheel

paths for each of the GPS3 pavement profiles, along with summary
data for doweled and non-doweled jointed concrete pavements.

without dowels. Note how the negative curvature sides of the two frequency
distributions are similar, where as the positive curvature sides of the distributions are
significantly different. The primary effect that dowels have on slab curvatures is to

27
Byrum

reduce the chance of developing unusual upward slab curvature and lifted joints.
This effect along with the load transfer provided by the dowels will significantly
reduce the rate at which faulting develops at joints between slabs. This GPS3
frequency data can be used for reliability modeling of curvature in jointed plain
concrete pavements.

Results

Figure 5 shows the overall scatter plot for GPS3 IRI vs. the average 1.219 m (4-ft)
long moving arc based slab curvature estimates. A key point regarding figure 5, is
that pavement deterioration is represented by upward vertical movement within the
scatter data. When analyzing a scatter plot for pavement data, it must be understood
how aging is present within the plot trends. The most significant trend from this
scatter plot is the lower boundary trend, which represents smooth younger pavements
with low amounts of other types of age related roughness features. Fitting a trend
line through the center of mass of this data is not of much use. That trend would
constantly change as the pavements age and more data points are accumulated in the
higher IRI region of the plot. When a pavement has no other roughness except slab
curvature, the IRI values would be down near the lower boundary and roughly along
the ideal pure curvature lines. Then as faulting and other deformations develop in
addition to curvature, IRI will increase above the curvature only trend lines shown.
Please note that the lines representing the ideal slab curvatures shown in figure 5 all
initiate from an initial IRI of about 0.5 m/km (32 in/mi). This value of 0.5 m/km (32
in/mi) is the apparent lower bound for the initial IRI values for the LTPP GPS3 at a
zero curvature condition and was added to the trend lines shown in figure 3 prior to
plotting them in figure 5. This value of 0.5 m/km (32 in/mi) is the apparent GPS3
lower bound for the contribution to IRI from the typical jointed concrete pavement
texture and construction techniques used to build the test sites. Actually, each
individual roadway has its own unique initial “zero-curvature and zero-faulting”
roughness level, and the curvature vs. IRI trend lines branch off from that point
based on the effective slab lengths.
The lower boundary trend line generally follows the ideal 5-meter (16.4-ft)
arc trend for low values of curvature and eventually ends up following the ideal 4-
meter (13.1-ft) arc trend line for high curvature magnitude. This makes sense
because about 60% of the LTPP GPS3 pavement test sites have slab lengths between
4.26 meters and 5.18 meters (14 feet and 17 feet), with an average slab length of 5.1
meters (16.6 feet). The LTPP GPS3 test sections have slab lengths ranging from 3.47
m to 9.14 m (11.4 feet to 30 feet) with only 13% of test sites having joint spacing less
than 4.26 meters (14 feet). The observed lower boundary trend for the author’s 1.219
meter (4-ft) arc based LTPP GPS3 slab curvature estimates vs. IRI data set is what
would be theoretically expected given the slab lengths used in the GPS3 pavements.
Real pavement slab wheel paths wont respond to gradients quite like ideal
constant curvature arc segments. As thermal and moisture gradients develop that
would cause curvature, pure constant curvature over the slab area would only develop
in a zero-gravity and zero edge restraint condition. In the presence of gravity, the
slabs are pressed against the earth and the middle portions of the slabs have reduced

28
Byrum

curvature and increased internal stresses. The bending stress can be considered anti-
curvature in this case, where the flattened interior portions of slabs exposed to
gradients are stressed (Westergaard, 1927; Byrum & Hansen, 1994). These flatter
middle portions of slabs result in regions of lesser curvature and lesser joint uplift
compared to the ideal pure arc curvature profiles. The greater the curvature causing
forces become, the greater percentage of those forces that get expressed in the form of
stress rather than curvature. Only short slabs can maintain high levels of average slab
curvature. For longer slabs, a greater percentage of the slab area will be stressed and
flattened. This general behavior could explain the general downward curvature for
the lower boundary trend line observed for the real pavements compared to the ideal
arc curvature profiles. The lower boundary trend line shown is unique to the LTPP
GPS3 data set and is based on the slab lengths within.

4.5 IRI rises above the Lower Boundary as Faulting/Deformation Develops

8 m Arcs
4
7 m Arcs
3.5
6 m Arcs
3
5 m Arcs
IRI, m/km

2.5 4 m Arcs
3 m Arcs
2

1.5
Lower Boundary for GPS3 Data
1 0.0948x3 - 0.4585x2 + 1.5966x

0.5

0
0 0.25 0.5 0.75 1 1.25 1.5 1.75 2
x = Average Profile Curvature (24"), 1/m x 1000

Figure 5. Scatter plot for the LTPP GPS3 1.219 m (4-ft) arc length based
profile curvature magnitude compared to the profile IRI values and
showing the ideal arc curvature profile IRI trends and the lower boundary
trend for the LTPP GPS3 pavements.

Figure 5 demonstrates an important difficulty with accounting for slab

curvature for initial smoothness specifications and profile evaluations in general.
Pavement design engineers use a wide range of slab lengths. For a constant
magnitude of curvature, IRI values will be higher for longer slabs compared to
shorter slabs. If, for example, a contractor uses the same construction techniques on
two roadway segments having two different slab lengths, and then equal magnitudes
of construction related curvature are locked in to the slabs, the pavement segment
with longer joint spacing would have higher IRI than the one with short joint spacing.
This difference in IRI is not at all related to the contractor, but is related to the
designer’s joint spacing. Would it be fare to hold these two pavement segments, one

29
Byrum

with shorter slabs and one with longer slabs, to the same IRI smoothness
bonus/penalty scheme? Is it necessary to adjust the bonus/penalty scheme as a
function of the design joint spacing? It is not the purpose of this paper to answer
these questions, but to provide clear data regarding these issues.
Figure 5 also shows a key observation regarding loss of serviceability for a
roadway when significant slab curvature is present. Notice that regardless of the
magnitude of slab curvature present, there are no roadways with IRI greater than
about 4 m/km (250 in/mi), a typical limiting threshold for allowable roughness for
highways. Pavement segments with high locked in curvature in slabs will have
higher initial IRI values and, therefore, a smaller range of serviceability available
between initial construction and the end of useful service life. There is also much
evidence available that if large locked in slab curvature is present, premature cracking
and faulting will occur for the roadway segment indicating more rapid loss of the
already diminished available serviceability (Byrum, 2001a, 2001b; Westergaard,
1927; Hveem, 1951; Byrum & Hansen, 1994; Sixbey et al, 2001; Hansen et al 2002,
Byrum, 2004). Locked in slab curvature of high magnitude must be avoided.
To summarize this discussion on the effect that slab curvature has on IRI
values, consider drawing the general shapes of the positive sides of the frequency
distributions shown in figure 4, onto the ideal IRI vs. curvature trends shown in figure
3. This allows a view of the relative percentages of pavements being affected by slab
curvature of varying magnitudes. About 85% of the GPS3 pavements with non-
reinforced joints having no dowels have locked in curvature values less than about
0.00065 m-1 (0.0002 ft-1). The IRI due to slab curvature for these flatter 85% of test
sections without dowels will be between about zero m/km and 0.8 m/km (up to about
50 in/mi) .
Doweled pavements have reduced measured slab curvature values. About
85% of the GPS3 pavements with reinforced doweled joints have locked in curvature
values less than about 0.00035 m-1 (0.00011 ft-1). The IRI due to slab curvature for
these flatter 85% with dowels at joints will be between about zero m/km and 0.5
m/km (up to about 32 in/mi).
The IRI values from slab curvature for the two high values in the doweled and
non-doweled pavements (0.0013 m-1 (0.0004 ft-1) and 0.0017 m-1 (0.00052 ft-1),
respectively) are about 2 m/km and 2.5 m/km (125 in/mi and 160 in/mi), respectively.
For the flatter 85% of the GPS3 test sections, adding dowels at joints will apparently
reduce potential slab warping by about 40% to 50%. In extreme cases, IRI from slab
curvature can reach 2.5 m/km (160 in/mi). Extreme locked in curvature can develop
in both doweled and non-doweled slab systems.

Conclusions

This paper presents a detailed comparison of the International Roughness Index (IRI)
trends for ideal repeating arc profiles and the observed curvature vs. IRI trends for the
LTPP GPS3 jointed concrete pavement profiles. The study quantifies the effect that
slab curvature has on jointed concrete pavement IRI values.
Frequency distributions describing the magnitude of average slab curvature
encountered in the LTPP GPS3 jointed concrete pavements are provided, which can

30
Byrum

be used for reliability modeling of jointed concrete pavements. Practicing pavement

engineers and researchers can compare their measurements of slab curvatures from in
service pavements to these frequency distributions to assess the relative severity of
the curvature observed.
Ideal repeating arc profiles were developed covering the range of curvature
and slab lengths observed in typical GPS3 type jointed concrete pavements. IRI
values are calculated for the ideal profiles and a regression model developed from the
results, which can predict the IRI caused by typical combinations of slab curvature
and slab length. These ideal trends for IRI vs. slab curvature are compared to the
LTPP GPS3 scatter plot for 1.219 m (4-ft) arc based slab curvature (Byrum, 2001a)
vs. IRI. The ideal profile trends match well with the GPS3 data trends. The lower
boundary IRI vs. slab curvature trend from the LTPP GPS3 data matches the ideal
curved profile trends for 4-meter to 5-meter (13.1 to 16.4 feet) long constant
curvature arc profiles, which matches the joint spacing for the majority of the GPS3
test sections.
For any constant magnitude of slab curvature, IRI increases exponentially
with increasing slab length. Therefore, two pavement sections, one with short joints
and one with long joints, constructed in the same way under the same environmental
conditions could have significantly different initial IRI values if slab curvature
develops. This would not be the result of the contractor’s actions, but is the result of
a design decision put in place long before construction. This joint spacing effect
should be considered in initial smoothness specifications having very high
smoothness requirements that can be affected by typical locked in slab curvature
values.
Maximum slab curvature is different than wheel path curvature. A road
profiler moving at high speeds is measuring the “wheel path view” of slab curvature.
The curvature measured along a wheel path is a subdued and distorted image of the
maximum slab curvature. Slab curvature is generally a radial-symmetric
phenomenon occurring radial outward from the centroid of the slab area. In the
presence of gravity, curvature is generally never constant over a slabs area, as self-
weight stress will tend to keep the middle portions of slabs flatter than the edges,
especially free edges. Maximum slab curvature is best estimated by profiling
diagonally across an individual slab, from a slab corner, through the center of the slab
area, and across to the opposing corner. Wheel path curvature magnitudes will
generally be less than the maximum slab curvature magnitude. All the data in this
study is from high speed profiling devices and is, therefore, referring to the slightly
subdued “wheel path view” of slab curvature. However, this is the vector component
of the slab elevation that the moving vehicle will actually experience, and which is
contributing to the IRI values. The range of maximum slab curvatures is, therefore,
slightly greater than the wheel path curvature frequency distributions shown in this
paper.

References

31
Byrum

Byrum, C.R. , 2001a. A High Speed Profile Based Slab Curvature Index For Jointed
Concrete Pavement Curling And Warping Analysis. Ph.D. Dissertation. University of
Michigan.

Byrum, C.R. 2001b. The Effect of Locked-in Curvature on Pavement Performance,

Published in a book titled, “Long-Term Pavement Performance: Making Something
of It”. Edited by R.G. Hicks and J. B. Sorenson. American Society of Civil Engineers.

Byrum, C.R. 2001c. Analysis of LTPP JCP Slab Curvatures Using High Speed
Profiles. Transportation Research Board, Transportation Research Record 1719.
National Academy Press, Washington DC.

Byrum, C.R. 2004. A Detailed Breakdown of the IRI Statistic: The GPS3 Case Study
Proceedings of the 5th Symposium on Pavement Surface Characteristics. World
Road Association. Toronto, Canada.

Byrum, C.R. Hansen, W. 1994. Influence Function Approach to Analysis of Jointed

Portland Cement Concrete Pavements. Transportation Research Record 1449.
National Academy Press, Washington DC.

Hansen, W. Smiley, D. Peng, Y. Jensen, E. 2002. Validating Top-down Premature

Transverse Slab Cracking in Jointed Plain Concrete Pavement (JPCP). Proceedings
of the 81st Annual Meeting of the Transportation Research Board. Washington, DC.

Hveem, F.N. 1951. Slab Warping Affects Pavement Joint Performance. American
Concrete Institute. Vol. 22, no. 10.

Sixbey, D. Swandlund, M. Gagarin, N. Mekemson, J. 2001. Measurement and

Analysis of Slab Curvature in JPC Pavements Using Profiling Technology,
Proceedings of the 7th International Conference on Concrete Pavements. Orlando,
Florida. Sept. 11, 2001.

Westergaard, H.M. 1927. Analysis of Stresses in Concrete Roads Caused by

Temperature. Public Roads, Journal of Highway Research, USDA Vol 8., No. 3.

32