6th tntemational Symposium on
Heavy Vehicle weights and Dimensiol"lS
s.,,,katoon, Saskatchewan, C _ a June 'IS - 22, 2000
Investigation ofthe Relationship between Surface Roughness,
Truck Dynamic Loading and Pavement Distress
Using Field Data from In..setvice Pavements
Karim Chatti, Ph,D.
Assistaot Professor
DoseungLee
Gr!uiuste Student
Dept. of Civil & Environmental Engineering
3546 Engineering Building
Michigan State University
East Laming, MI 48824
In this paper, ten road sites in Michigan were selected and analyzed for a case study to
investigllte the relationship between road roughness and pavement damage. The selected
sites include five rigid pavements, three flexible pavements with rubb!ized concrete bases
and two composite pavements. The length of the sites varies from 6.5 to 13 km. Michigan's
Ride Quality Index (RQt) and Distress Index (D1) profiles were generated along the project
length, allowing for the identification of zones of high distress and high roughness and
those with high distress and low roughness. International Roughness Index (lRI) was also
calculated. Dynamic load profiles were genet'ated, and the corresponding DLC-values were
determined, using three truck configurations: (l) a two-axle single truck; (2) a three-axle
single truck; and (3) a five-axhi tractor-semi-trailer. Good fits were obtained between RQI,
00, and DLC values. Trends ofDI, RQT, IRl and DLC from year to year were generated.
These plots allowed for discerning (dynamic) load-related from non-load related distress
accumulation. The paper shows examples of different scenarios encountered for the three
pavement types. The coefficients of correlation (P) between dynamic loading and distress
points were determined for all the sections. Distress Poims (OP) were calculated at 3-m
intervals. Comparison of dynamic-load and distress-point profiles along the length of each
section enabled for the assessment of the reasonableness of p-values relating load and
distress profiles. The results showed that a reasonable match between load and distress
profiles could be obtained with relatively low values of p.
219
.........................................................--._...-- .....
,
-~----- ------
..............--------.....•,.- .. .
1.0 INTRODUCTION
Dynamic pavement loading due to the mix of commercial heavy trucks that make up the
national fleet has been steadily increasing in the US as the national highway network
continues to age. While this may be somewl:-.at counteracted by an increase in the use of
"road-friendly" vehicles, truck loading and its dynamic component remain a major cause of
road damage. Al! road surfaces are inherently rough, and they become increasingly rougher
with age. This process is the result of the interaction between vehicles and pavements.
Accordingly, the relationship between surface roughness and pavement damage should be
of great concern to pavement engineers. The relationship between load and distress is
compl~ The variability in pavement materials, environmental and traffic factors, coupled
with errors in field measurements and the subjectivity of the field distress surveys make it
extremely hard to relate roughness and dynamic loading to the accumulation of distress in
in-service pavements. Furthermore, some distresses are mainly material- or environmentrelated. Therefore any interpretation of field data must be made with extreme caution.
In this paper, measured roughness and distress index data from ten road sites were
analyzed to look at the interaction between road roughness and pavement damage. The
analysis is aimed at showing how one could use roughness and distress data for the
purpose of discerning cases of dynamic load-related distress accumulation.
2.0 ROAD SURFACE ROUGHNESS
Roughness is not only an indicator of road surface condition but also excites the dynamic
behavior of trucks, thus increasing damage. There are several ways for quantifYing road
roughness. One of the most widely used parameters is the International Roughness Index,
(IRI). The IR1 is calculated from a measured longitudinal road profile by accumulating the
output from a quarter-car model and dividing by the profile length to yield a summary
roughness index with units of slope. MDOT collects IRI data for the entire pavement
network in Michigan on a yearly basis. Another road roughness measure is the Ride
Quality Index (RQI), which was developed and is being used by MDOT. The RQI is
measured with a Rapid Travel Profilometer, and is calculated as (Darlington, 1995):
[Eq. 1]
RQI=3 *1n(Var I )+6*!n(Var2)+9*In(Var3)
where Varl, Var2 and Var3 are variances for (7.6-15.2 m), (1.5-7.6m) and (0.6-1.5 m)
wavelengths, respectively. An RQI value between zero and 30 indicates excellent ride
quality, 31 to 54 for a good ride quality, 55 to 70 for a fair ride quality and more than 70 is
considered poor pavement ride quality (Darlington, 1995).
3.0
PAVEMENT DAMAGE EVALUATION
The MDOT developed the Distress Index (Dl) to evaluate pavement surface deterioration.
Distress data is collected by videotaping 50 percent of the pavement network every year.
220
The tapes are reviewed in the main office, and each distress type is counted, documented,
then tabulated and stored in the PMS data bank. Each distress type is assigned certain
distress points (OP) depending on its severity and extent using pre-established distress
point values. The DI is the sum ofDP along I!. project normalized to the project length:
[Eq.2}
DI=LDPIL
where L is the length of the project in 161-m (D. I -mile) unit sections. The m scale starts at
zero for a perfect pavement and it increases without a limit. MDOT categorizes DI into
three levels: Low; <20, Medium; 20-40, and High; >40. Any project with a DI of more
LItan 50 is treated as a candidate project for rehabilitation.
4.0
A~ALYSIS
4.1 Site Selection
The study involves the analysis of roughness and distress data from a limited number of
actual pavement projects in Michigan. Ten sites with known performance records and
having exhibited some distress were selected by MOOT. Roughness and distress data were
available only in the time period between 1992 and 1997. Five of the ten sites were rigid
pavements, three were flexible pavements with rubblized concrete bases, and two were
composite pavements. The selected sites are listed in Table 1.
4.2 Data Collection
For the ten selected sites, DJ and RQI data as well as road surface profiles were obtained
from the MOOT PMS database. DI values were available for 1993, 1995 and 1997,
whereas RQI values and road profiles were available for the period between 1992 and
19%. The m and RQI data was available for 161 m long (0. I-mile) sections. Surface
profile data were converted to ASCII files containing surface elevations at 76-mm (3-in.)
intervals. Detailed distress data in the form of distress type, severity and extent was also
available at 3m (10-ft) intervals. The data was then converted to Distress Points (DP).
4.3 Data Analysis
Relationship between RQl, lRl and DLC - For each pavement section (total number of
sections from the ten projects is 168), OO-values were calculated using the RoadRuf©
computer program. RQI-values were plotted against the calculated OO-values, as shown in
Figure 1 (a). As expected, RQI and 00 have a good correlation, with the RQI increasing
asymptotically with increasing 00. The OO-value corresponding to an RQI of 70 is about
2.4 mm/m (ISO infmi) This value was chosen as the threshold 00 to be used in
normalizing 00 data. A standard I8-wheel tractor-semi-trailer with leaf-spring
suspensions was run over all sections at normal tra..'fic speed, and the TruckSim"" program
was used to generate dynamic loads. Dynamic Load Coefficients (DLC) were then
221
calculated. DLC is a dimensionless measure of dynamic variation of an axle load, and is
calculated as the ratio of the standard deviation over the mean static load. The DLC-values
reported herein correspond to those from the front drive axle. DLC versus RQI and IR1 values were plotted in Figure 1 (b) and (0), The figure shows good fits between these
indices, indicating that both roughness indices can be good indicators for dynawic loading
and future pavement deterioration. An RQI-value of 70 corresponds to a DLC-value of
12%, This value was chosen as the threshold DLC-value for normalization.
Distress and Roughness Profiles - DI and RQI data for all ten sites were normalized with
respect to their respective threshold values of 40 and 70, and piotted along the longitudinal
direction, The threshold values correspond to the lower limit of the high level range, as
categorized by the MDOT PMS. These plots which were generated for 1993, 1995 and
1997 give an overall view of Dl and RQI levels for the entire project length in a given
project. Three projects that exhibit relatively high distress levels were chosen for the
detailed analysis, which involves matching dynamic loads from truck simulations to
detailed distress data. They are CS! 1017 (rigid pavement), CS09!Ol (flexible pavement)
and CS38JOl (composite pavement). Figures 2,3, and 4 show the corresponding Dl and
RQI plots, From these three projects, three (161m) sections were selected and dynamic
loads were generated, A total of9 sections were selected (3 sections per project).
Correlation between Dynamic Loads from Different Axles and Trocks - The TruckSim~
program was used to generate dynamic loads from three truck types, Truck configurations
and suspension properties used in the simulation are shown in Tables 2 and 3, respectively.
Figure 5 shows coefficients of correlation (P) between profiles for each axle load and those
for the reference axle loan (2nd axle load in 5-axle semi-trailer) within each section. The
figure shows variable degrees of correlation among different axles. The best correlations
are among the drive axles while the worst is with the rear axle of the tractor seml"trailer,
indicating that the trailer is out-of phase with the tractor. The correlation between
aggregate axle loads for the different trucks is strong (see Figure 6) with coefficients of
correlation higher than 0,77. Aggregate axle loads for each truck and the reference axle
load show very good correlation with values around 0,7 (see Figure 7). This indicates that
the reference axle can be used to represent the aggregate load from all trucks.
Distress Point and Dynamic Loading Spatial Distributions - To determine whether distress
is dynamic load-related or not, one can plot the distribution of Distress Points (OP) along a
section against that of the dynamic load (DL). This was done for all sections, Both DF and
DL values have been normalized with respect to their respective 951h percentile value,
which should represent the more severe cases of distress accumulation and dynamic load
amplification, respectively. The coefficient of correlation (p) between DL and DP was
calculated for the 9 selected sections, Five different loads were uSed in the analysis: The
2nd axle load of the tractor semi~trailer, the aggregate load for each truck and the
cumulative aggregate load for all trucks, Figure 8 shows the corresponding p-values. As
expected from the correlation between axle loads, there is little difference among load
types. The figure shows that the 6 sections of Projects CS 11 0 17 and CS38! O! tend to have
222
higher p-values than those from Project CS09101. Section No .! in Project CS! 1017 has
the highest p-values ranging from 0.54 to 0.62, showing good correlation between DL and
DP. On the other hand, all three sections of Project CS09101 have very low p-values
ranging from 0.15 to -0.18, indicating very weak to negative correlation. This indicates that
surface distress fur these sections are not related to dynamic loading. Sections from Project
CS38101 have intermediate p-values ranging from about 0.2 to 0.4, showing weak to fair
correlation between DL and DP Figure 9 shows an example of a very good correlation
with p-value of 0.54. Figure 10 shows an example of very poor correlation with p-value of
-0.08, while Figure 11 shows an example of a good correlation with p-value of only 0.28 .
Distress, Rouglmess and Dynamic Loading Trends with Time - Normalized D1, RQI, 00,
and DLC values (with respect to their threshold values) fur each section were plotted at
consecutive years. Figures 12, 13 and 14 show examples of such plots for Projects
CSII017, CS09101 and C538101, respectively.
4.4 Discussion
Project CS 11017 - This project is a plain concrete pavement section with an open-graded
base course. Prior to its rehabilitation in 1995, the overall level of RQI values was very
high. However, the DI showed severe variability along the project, as shown in Figure 2
(1993 and 1995). Figure 12 shows that RQI and IRllevels are at or above the thresliold
value for rough pavements; consequently, the DLC values are also high. This apparently
caused the DI to increase from about 70% of its threshold value to about 100% in 199 5.
This prompted the rehabilitation action in 199 5. It can be inferred that the distress
accUlDlllation for the three sections selected in this case is mainly load-related. Figure 9
confirms that. Subsequent discussion with MOOT revealed that the pavement exliibited
numerous transverse cracks which may have been initiated as hairline cracks during initial
construction, and have propaga.ted under the load action. Some minor faulting had
developed and some major spalling had taken place, mainly due to load action and some
poor quality aggregate. These findings support the conclusion based OD the PMS data.
Project CS 09101 - This project was reconstructed in 1990 by rubblizing the old concrete
slabs and overlaying them with an asphalt concrete layer. Figure 3 shows a dramatic
increase in DI-values from 1993 to 1995, particularly in the sections from .MP 1.6 to MP
5. 1. DI-values were further amplified in ) 997 for some sections, while they were reduced
for others, presumably as a result of maintenance activities. The RQI level remained
relatively low throughout the period from 1992 to 1996. The increase in DI cannot be
explained by roughness or dynamic loading. Figure 13 shows that RQI and IRl levels are
significantly below the threshold value for rough pavements; consequently, the DLC
values are relatively l{)w. In all sections, DI-values greatly increased from 1993 to 1995,
exceeding the threshold value of 40. In Section I (MP 1.6 to MP 1.7), the DI further
increased in 1997 to reach I. 7 times the threshold value. Sections 2 and 3, on the other
hand, show a relative decrease in DI, presumably as a result of some maintenance activity.
It can therefore be inferred that the distress accumulation for the three sections selected in
223
tbis Case is not primarily load-related. Figure 10 confirms that. Subsequent discussion with
MOOT revealed that tbe pavement exhibited longitudinal cracks both inside am! outside
the wheel path as well as some transverse cracks, which are thought to be material-related
(low asphalt coment; high air voids). This supports the conclusion based on the PMS data.
Project CS 38101 - This project is a composite pavement section >vith 30-m (99-ft)
reinforced concrete slabs that were overlaid by a 1GO-mm (4·in.) asphalt concrete layer in
1985 . Figure 4 shows that the match between RQI and DI proG.les from year to year is
excellent The leveis of RQI and DI values are intermediate to high in some sections.
Figure 14 shows that RQI and IRl levels are at or above tbe threshold value for rough
pavements; consequently, the DLC values are also high. This apparently caused the DI to
increase from about 70% of its threshold value to about 100 to 120 % in 1997. It can be
inferred that the distress accumulation for the three sections selected in this case is mainly
load-related. Figure 11 confirms that. Subsequent discussion with MDOT revealed that the
pavement exhibited reflective cracking at all the joints, and that a deterioration of the
cracks due to load action forced a major maintenance action (milling and overlay) in 1998.
These findings support the conclusion that was made based on the PMS data.
5.0 CONCLUSION
The analysis in this paper showed that plotting roughness and distress profiles along the
project length allows for identifying zones of high distress andIor high roughness.
Combining the above information with the trends of distress am! roughness indices from
year to year allows for discerning dynamic load-related from non-load-related distress
accumulation. The paper showed examples of different scenarios eooountered for the three
pavement types, namely rigid, flexible and composite. Comparison of dynamic-load and
distress--point profiles along the length of a given pavement section enabled the assessment
of the reasonableness of the p-values relating the two profiles. The results showed that a
reasonably good match between load and distress profiles is possible with relatively low pvalues (between 0.25 and 0.5) whereas a p-value that is less than 0.2 is indicative of a poor
match. A higher p-value indicates dynamic load-related distress accumulation, and a lower
p-value indicates that the distress accumulation is not primarily related to dynamic loading.
Finally, distress/roughness/dynamic load trend with time can be a good indicator to see
whether distress is load related or not.
REFERENCES
I. Darlington, John. "The Michigan Ride Quality Index", MDOT Document, Dec. 1995.
2. Michigan Department ofT.ransportation. "Evaluating Pavement Surfaces: LISA and
RQf' Material and Technology Research Record, Issue Number 79, June 1996.
224
aJROI V8.Utl
Teblc 1. l.ist of Projects InvtJtiSlucd
(I milt-I.6lun)
Pavemft"a Type
BMP
ROIJ"/e
Control
EB US-ID
09101
33084
47014
~liZed pee bue) 0..924
76024
R;gW(iIlCP)
0.0
18024
18024
38101
Rigid(JPCp)
Rigitl(iRCP)
Rigid(JRCP/JPCP)
IUgid(JRCPIJPCP)
Composite
38131
C~)tnpc6ite
Scciion
E8 ]-96
NB US-23
EB ]-69
EB 1-94
NB]-69
EB US-ID
WO US-IO
E81-94
N8I1S- m
EMP
'T'j5ii-~-'---]7.491
1.165
3.827
Ul/93
10/91
7191
1.015
0.0
5.875
9<9$
5.0
11191
0.0
7.598
0.0
0.0
0.0
1.598
8.7
5.2
11176
11/76
1fJif!5
10181
Ftexible(llubbltzed pec base) 11.979
Fltxibk(Rubblir.oti pee ""..) 0.0
11017
23063
CO!'.stmctioR
50
lOO
150
iftl,lnlmllo
200
250
300
b)RQlvs.DlC
R' "'0.7500
Table 2. Truck Mztrb: Sizes Ilftd Weights
IV
IV
ot:;l
'"
oC:l
c;<;;::;;1
AxleLoi~fB~
Confiattra~jon
Truck
~lime _~_
ConfigurO:l.tC?'.n
__(kN)
--<!®.---~.
2 Axte Truck
125-
49176
4.3
3 Axle TruCK
ISO
56194
6.1
5 Axle Semi~
356
5411511151
3.6/11.0
TrQiler
i~§~
...
'"
'"
Table 3. Slapcnsion VertjcQl Properties
SUSPCUWII
LOC4rtioo
Sus~n
Eftve!ope
StifJhesa
(kN1l)l)
28.5
0.04
FI.. Leaf
47.8
41.3
Flal Leaf
47.8
Flat Leaf
47.8
Type
-'s...,.,w;- PI.. Leaf
S~R8le
Drive Axle
Ts.nd:errt
Drive Axle
Tandem
Smd-tmiter
..,
'20
"'"
cl fRS Y8. DLC
--* Linear
Lower
liP!'«
Envelope
Stiffitess
(kN/m)
28.5
'"
-u;;;prung
o.mptnl
Weigh'
CocfJ"tCietlit
(kN)
(!j:!-"m!
0.5
6.2
0.08
om
ILl
41.3
0.08
0.02
22.7
41.3
0.08
0.02
16.9
""
,oo
..
,
ERI, 1Mr.:It&
'"'"
"""
""'10
Flgur. 1. Re!»ll<mships belwelln Rat. tRI and DlC for All SuIls<lct!nno
Arlolyzed
300
~
is
• 1997 RQI is nol aveileble
"r--r
!
I ~-- ~
---,---~-;---~~~~:~
1.5 _._.- ----
I
2
,--
i
(lmilo~l
;
.6km)
I
0;97-
--R~~
1.5-- ~
'---.- .-. :- ' - '
1.5
j
J
I
I
;
r ", . .- '.'".
__ __,.,--'-___ .L_ .. ....
0.5
o
r-~~-J?!VfiI -J
,I
2
3
I
4
0.5
O.S
0
o
6
2
5
.
I
'
5
,'
!
'
6
2
Figure 2. Distress and Roughness Profiles along Project Length for EB 1-94 (CS 11017) M.P 1.015-5.875
~
is
1.5
1.5
n=~.;A~-J'
~-'rn, 97
! , . ~~
;
I
IV
IV
CI\
Iz
0.5
0.5
0
....
4
*'~
0
8
2
Figure 3. Distress and Roughness Profiles along Project Length for EB US-l0(CS 09101) M.P 0.924-7.356
__
~ 1.51 ___I_ ··: ~ ~+__IJ·_~i931
Q
2
1.5
~~
C;.._,~._~_L+-W~RQr95
1.5
I:'-!- ·-j~-{jf-""lj#\;ff--T-Il
0.5
2
i
11
&;!
~
~
0.5
~~~Jblr-' -..Jt:i'lI~
0.5
0
2
6
Mllepoot
0
3
4 5 6
Mllepost
7
8
Figure 4. Distress and Roughness Profiles along Project Length for EB 1-94 (CS 36101) M.P 0.0 - B.7
9
0
1
2
3
4 5 6
Mllepoat
7
8
9
3{)
o
00
00
120
150 (mj
111E--·"-:'-"A~-.j;~:~~-~-=--,- ~·P;;;-~=b::~1
hi
2AmTf\Id!
!:SAs!lI SIl'/'iii.T~
3 AIds Truck
Figure S. Coefficient of Correlation for Axle Load Compared with Reference
Axle Load (2'" Axle Load in &.Axle Seml-n a»er)
m~~
,....,"""
34\xtaTNd(
5-~T~r
"''''''''
Figure 6. CoeffICIent 01 Co....elion for Aggregate Axle Load 01 Each Truck
Comparod with Aggregate Axle load from All Trucks
""
.....
IV
0
I ~~ll--_-~
64W0T,...
AIITf'CW;S
~
_
_
Figure 9. N()rMaU~ad Dynamic Load and OisttoQS Points vorsus Longitudinal
Dlstence along Subsection #1 (MP 2.6 to 2.7) of Projm:l CS11 017
l::~~'::-~
lll.:.:,= m~-c ..• .•.. .• . •
.•.•.•..
o
3-AlOOTrudI
m
i)1!!1teI'W:e,'t
I ~J ~-.~ . , ",'
l!~~
"_
24v1l$Trotk
~
_
::=-= .~=~~~---::~: ~
~
~
_
~
~
DI~,ft
~of~
Flgure 10. Nonnallz!:ld Dynamic load QI'ld Ot'31roS8 Poillt.$ versus l ongltudtnm
Figure 7. CoeffICIent 01 Correlollon I<>!' Aggregot. Aw Load with Z" Axle Load
In &.Axie Somi-Trailer
:~..
~::::
iI~ j~, ,41. ~ :=:i1i11
"==~Am
allfI
-0.2 _ _
#2#3
CS 11011
#1
#2
cs 09101
Al3
81
"&1283
CS 38101
Figure 8. eo.ffioIent of Correletion botween AxIs Lood ond 01.".... Poin"
for Three 5eiect6d Projects
Distance along Subs<mtionl/2 (MP 3. t 10 3.2.) of Project CS09101
2.~
~
I
,.
,.:
III 0:
j :::
I
100
""
_.ft
300
"""
... ...
Figure 11 . NormaliZed Dynamic Load and Ollllr••• _
""""'" Loagitudl""l
DI.lonce along SubIulction 113 (MP 5.3 to M) of ProJoct CS381 01
:T~~L~
92
93
95
94
2~MJ!1&;I,7
1':~_-'
92
';F++-~ -~JI~~~I
1
...--.
0.5
- - - - - - --.
_ A ____
- ...... A
__• __
ADI
0 Dle
._~
o
o
97
96
-_ ._.
~
0.5
93
95
94
96
92
97
93
94
96
95
97
Figure 12. Year-ta-Year Trend of Rougljness. Distress and Dynamic Loading Indices for Three Subsections (160m)
along EB 1-94 Project (CS 11017)
~
•
2
1.5
,....
,....
00
1
---._'.
- -.-.
o
0
~_..il.__
0.5
0
92
...
,
"
93
94
95
2
M.P3.1-3.2
1'5~
---- .- ~-1
---
-- g .
0.5
i'
92
97
93
94
95
96
.01
0
- _..- - t r - ....- --- .ODLC
. _.--r--
0-
97
OIRI
ORQI
---.----~--
1
0.5
~-
0
96
l':Ef-"'~
_~,P5'~
age
92
93
95
94
96
97
Flgurel3. Year-ta-Year Trend of Roughness. Distress and Dynamic Loading Indices for Three Subsections (160m)
along EB US-1O Projact (CS 09101)
.P 1,.0-1.1
1.5
a
.....JL f!I
•
0.5
2~~M
'P4'4
- ---
.5
1.5
o
~ __
I
4
0.5 __........_
o1
92
93
95
94
Year
96
97
92
_
•
0.: ---- --..
,
95
94
M·E.u..~
96
97
<> iRI
--- -~ - ooJ
____.__ - -- - - -.--
-8------s-- .!Ii--..
1
.A..
i
93
..JJ.._
2 ~____
1.5 .
92
93
Year
Figure 14. Yea r-Io-Year Trend of Roughness, Distress and Dynamic Loading Indices for Three Subsections (160rn)
along EB 1-94 Project (CS 38101)
"
i
i
94
95
V\):lr
0 RQI
401
~
96
97