Professional Documents
Culture Documents
Determination of Aashto Layer Coefficients
Determination of Aashto Layer Coefficients
TE
5092
.M8A3
no.90-5
v.1 c.2 (SOUR! COOPERATIVE HIGHWAY RESEARCH PROGRAM
FINAL REPORT
9 0-5
Property of
Volume I:
BITUMINOUS MATERIALS
Study 90-5
Prepared for
by
DAVID N. RICHARDSON
JEFFREY K. LAMBERT
PAUL A . KREMER
ROLLA, MISSOURI
in cooperation with
December 1994
The purpose of this investigation was to determine layer coefficients for several
MHTD specified pavement materials. The coefficients are necessary as input to the
materials, and is reported herein. Volume II deals with unbound aggregate base and
Besides determining layer coefficients, the study also entailed the determination
gradation, testing temperature, aggregate source, and asphalt content within the limits
All materials were sampled and delivered to UMR by MHTD personal. Choice
of material sources was made by MHTD. The types of pavement materials were Type
C, Type 1-C, and bituminous base. The specific materials making up these types were
two grades of asphalt cement, two sources each of surface mix coarse aggregate and
base mix coarse aggregate, and one source each of natural sand, manufactured sand,
Routine index and specification tests were performed. For the asphalt cement,
the tests were: penetration at 38° and 77°F, kinematic viscosity, absolute viscosity,
specific gravity, and softening point. The aggregates were tested for gradation,
specific gravity, and particle shape/texture. Equipment was fabricated for the particle
shape/texture tests.
combinations was determined by use of the Marshall mix design method (75 blow,
manual flat-faced hammer). Use of AC10 and AC20 grades resulted in 24 mixes.
And, 24 additional mixes were made which had 0.5% asphalt added above optimum,
method, and 2) calculation from material proportions and specific gravities. Ninety-six
specimens were tested . A voids analysis was conducted to determine the effect of
the assumption that the effective specific gravities of low absorption aggregates is
midway between the bulk and the apparent specific gravities. The voids analysis
indicated that the estimation method correlated very well with results from Rice
method testing. However, for absorptive aggregates (.fill., the bituminous base
materials in this study), the estimation method underpredicted air voids by about 1 %.
Ten methods of characterizing gradation curve shape and position were used .
Two of these were original to this study. The first involved the area between the
gradation curve and the maximum density line as plotted on FHWA 0 .45 power paper.
The second method involved determination of the slopes of three portions of each
gradation curve . The method of determining the area between the 0.45 power
maximum density line (MDL) and the gradation line had only a fair (R 2 = 0. 79)
correlation with resilient modulus (MR). This was because the magnitude of the area
was not sensitive to relatively small differences in position of the gradation curve
relative to the MDL. The second unique method involved calculation of the slope of
111
three different parts of the gradation curve. This method was shown to be of
was not quite as helpful as merely including certain critical sieve sizes directly into the
regression equation .
Each mix was tested for indirect tensile strength. A regression model was fit
to the data, which included 96 test results. The regression model was relatively
strong (adj-R 2 = 0.840) and was a function of asphalt viscosity, effective asphalt
content, percent accumulative retained on the #4 sieve, and coarse aggregate particle
shape.
Each mix was tested for total resilient modulus (indirect tension) . Necessary
software and equipment were developed to perform the tests and to acquire, store,
and analyze the data. A total of 192 specimens were tested at three temperatures
for a total of 576 tests. The resilient, modulus test is sensitive to testing conditions
L VDT fixation, LVDT tip design, and resolution of both vertical and horizontal LVDT's.
Constant diligence is required by the operator to assure that the very small
strong (adjusted R2 = 0.946) regression model was fit to the UMR MR data.
variables that were significant to changes in MR. The analysis of the data indicated
that temperature was by far the most important variable that affects MR, followed by
IV
asphalt viscosity and whether the gradation was very fine or very coarse .
Increases in (-) #200 material and decreases in ( +) #4 material tended to increase MA.
Particle shape of coarse or fine aggregate did not seem to affect MA in a consistent
manner . It should be noted that both coarse aggregates were crushed limestones, and
that all mixes contained varying amounts of manufactured sand, so large ranges in
particle shape were not present. And, all other things held constant, decreasing air
Resilient modulus data from other studies found in the literature were merged
with the UMR data. A general regression model was fit to the overall data base. The
model was not as strong as the UMR model, but was deemed superior because it
passing the #200 sieve, percent accumulative retained on the 3/4 in sieve, and
air temperature data from 104 weather stations in Missouri were analyzed to produce
an air temperature contour map of Missouri. Pavement thickness data for MHTD and
Road Test flexible pavements were analyzed for mean pavement thickness. This
data was supplied by MHTD . This was converted to load dwell times and loading
V
frequency for MHTD pavements . The same was done for AASHO Road Test
pavements .
UMR, AASHO Road Test, and MHTD 1990 mix des ign data were used to
estimate resilient modulus , mixture stiffness (Shell method) , and dynamic modulus at
both laboratory conditions and field conditions of pavement temperature and loading
rate . This was done in order to see which type of modulus would be most useful for
layer coefficient determination . The Odemark equation was used to rate the three
of ability to predict resilient modulus, was: MR estimated from the above general
Five different methods of calculat ing mixture stiffness (Sm) were compared;
each varied in the manner of handling asphalt aging or source . Of the f ive d ifferent
methods , the method of Bonnaure , wh ich uses the Ullidtz asphalt aging
approximations, was found to be the most accurate for the purposes of this study .
Two options to obtain layer coefficients were presented for possible use . The
the proper AASHTO nomograph to obta in the corresponding layer coefficient . The
modulus, but the moduli must be converted to the pavement temperature conditions
in the locale of interest. Then, the layer coefficient is computed via the Odemark
Option One resulted in a fixed layer coefficient per material. For 1990 mixes ,
vi
Type Ca, = 0.42, Type IC a, = 0 .42, and bituminous base a 2 = 0 .34. Option Two
is a method for pavement designers with which to calculate layer coefficients for a
overall mixture quality. Only asphalt grade and extremes in gradation were significant
in their effect on MR. For asphalt layer thicknesses less than 4 in, the effect of
variations in the above two variables was small. However, as layer thickness
of layer coefficient was shown to be a function of mix variables and location of the
site within the state . Examination of the effect of choice of layer coefficient revealed
practical impact of the trends is that higher layer coefficients can be obtained by :
Excessively hard asphalts, highly angular aggregates, and low air void values can lead
to such problems as thermal cracking, lack of workability, poor durability, and rutting.
Thus, caution should be used when interpreting the results of the above analysis .
vii
Also, it must be kept in mind that the MR predictive equations are based on data that
in present use in order to update or replace the above equation by use of a more
representative data set of the materials. A greater degree of accuracy will also
probably be achieved. Then, both Options One and Two will render more
It should be remembered that this study is in the mold of the traditional method
asphaltic material problems with thermal cracking and rutting, for instance, are not
directly addressed. To address a wider range of material issues, creep testing and
gyratory shear testing may be in order, however, these kinds of tests were beyond the
scope of this project . Also, this project was conceived in 1989 and the bulk of the
testing was performed in 1991, before the SHRP project results became generally
known. In the future, it may be that some of the recommendations coming out of the
SHRP program can be used to update the quest for layer coefficient determination.
viii
EXECUTIVE SUMMARY
TABLE OF CONTENTS ii
LIST OF FIGURES V
LIST OF TABLES viii
INTRODUCTION 1
General 1
Objectives and Scope 4
RESILIENT MODULUS 27
LABORATORY INVESTIGATION 28
Mix Designs 28
Asphalt Cement 32
Penetration 32
Kinematic Viscosity 32
Absolute Viscosity 32
Specific Gravity 33
Ring and Ball Softening Point 33
Aggregate 33
Initial Gradation 33
Final Gradation 33
Particle Shape and Surface Texture 35
ix
Specific Gravity 36
Screening 38
Fabrication of Specimens 38
Marshall Mix Design 40
Maximum Theoretical Specific Gravity 40
Specimen Bulk Specific Gravity 41
Voids Analysis 41
Indirect Tensile Strength 41
Resilient Modulus 44
SUMMARY 150
CONCLUSIONS 154
RECOMMENDATIONS 158
ACKNOWLEDGEMENT 159
REFERENCES 160
APPENDICES 171
Appendix A: Fine Aggregate Particle Shape Determination 172
Appendix B: Coarse Aggregate Particle Shape Determination1 81
Appendix C: Determination of Indirect Tensile Strength and
Resilient Modulus for Bituminous Mixtures 189
Appendix D: Determination of Asphalt Type 206
Appendix E: Specific Gravity and Voids Data 209
Appendix F: Results of Marshall Mix Design 218
Appendix G: Industry-wide Database for Resilient Modulus
Regression Analysis 231
xi
LIST OF FIGURES
24 (Gmm - D) vs . Absorption 75
25 Comparison of Air Voids Methods of Determination for Types C
and IC Mixtures 77
LIST OF TABLES
INTRODUCTION
GENERAL
Highway and Transportation Department (MHTD) . The study is being made at the
request of the MHTD Research Advisory Committee. The project was executed by
Engineering.
Based on the results of the AASHO Road Test, a pavement design method
applying the method, the highway designer determines a "structural number" (SN)
reliability, and desired terminal serviceability. The magnitude of the SN reflects the
degree to which the subgrade must be protected from the effects of traffic . For
determine the manner in which the SN will be achieved, i.e ., what the required
thicknesses and quality of each pavement layer should be. This is done by solving
SN = structural number
layers, respectively
Drainage coefficients are essentially modifiers of the layer coefficients, and take
into account the relative effects of the internal drainage of the pavement structure
addressed in a second report submitted by UMR to the MHTD concurrent with this
study (2).
A preliminary review of the literature indicates that reported values for layer
The range of layer coefficients determined at the AASHO Road Test are shown in
Table 2 (5, 11 ).
3
It has been postulated that the magn itude of any layer coefficient is a
function of several factors. For example, the asphalt surface layer coefficient a 1 is
thickness, and compacted mix stiffness. For an unbound granular base, the layer
the layer, degree of saturation, compactive effort, aggregate properties, and base
layer thickness .
As originally used in the AASHO Road Test results, layer coefficients were
actually regression coefficients which were the result of relating layer thicknesses
to road performance under the conditions of the Road Test. The problem is to
4
translate the Road Test findings to other geographic areas where the construction
materials and climate are different. Layer coefficients must be determined in order
to use Eq. 1 for design purposes. In a pure sense, layer coefficients are abstract
tangible. Most commonly, layer coefficients are determined on the basis of relative
layer material strength or stiffness considerations. Over the years since the
AASHO Road Test, many methods have been used to determine values for layer
coefficients.
economy, accuracy, and length of study. In brief, the study entails determination
of stiffness values for several commonly used MHTD types of pavement materials .
The stiffness values were determined by both direct laboratory modulus testing
coeffic ients and then verified for reasonableness by comparing the resulting
method suitable for use in routine design which will enable the pavement designer
traditional one (5, 12), which is to take some measure of strength, stability, or
resilient modulus) for the counterpart AASHO Road Test material. The
5
comparisons are usually done by use of the AASHTO Design Guide chart or some
ratio of the two parameters. Thus, the influence of rutting is not directly
addressed.
The materials for which layer coeffi9ients were determined were limited to
two types asphalt surface mixes (Types C and 1-C), one type of bituminous base
mix, two types of cement treated base mixes, and one type of unbound granular
materials ( 13).
Layer coefficients were determined by use of two methods and the results
AASHTO NOMOGRAPHS
The moduli determined in the laboratory phase were used directly with the
layer coefficient nomographs in the 1986 AASHTO Guide, which reflect generic
moduli-layer coefficient relationships. There are charts for dense graded asphalt
cement treated base (ai), and bituminous treated base (a 2 ) . The relationships
between layer coefficients and moduli were developed by Van Til et al. (5). Thus,
can be determined.
6
The nomographs were used in a second manner. Data from all approved
MHTD mix designs for 1990 were used in a regression equation developed in this
study which estimates resilient modulus. The resulting estimated moduli were
EQUIVALENT STIFFNESS
the following equation which relates MHTD material properties to AASHO Road
a _ a J modulus, MHTD ] 1 / 3 • • • • • • ( 2)
n ,MHTD - n, AASH't_ modulus, AASHO
The moduli may be one of several types such as resilient modulus (MR),
dynamic modulus (IE* I), or mixture stiffness (Sm). The above equation, originated
by Odemark ( 14), was discussed by Carree and White ( 12) and is based on
All materials in the study were approved MHTD materials and were used in
the specific mixtures as normally intended by MHTD. The material sources were
Two types of asphaltic cement concrete were studied. These were MHTD
wearing course mixtures: Type C and Type 1-C. Included in the study were two
grades of asphalt cement, two sources of coarse aggregate; one source of natural
sand, two sources of manufactured sand, one so_urce of mineral filler, and one
source of hydrated lime. The materials, sources, and identification codes are
shown in Table 3. All sources are located in Missouri except as noted . The two
The plant mix bituminous base mixtures contained the same asphalt cements
and natural sand as did the Types C and 1-C mixtures. Two sources of coarse
aggregate were used, and are shown in Table 3. The coarse aggregates were
chosen by MHTD personnel to give a range of particle shape and surface texture .
DR-8 natural Missouri River sand St. Charles Sand Co. Bridgeton
#1
GENERAL
desirable. This reasoning has given impetus to the effort by the pavement industry
In order to determine layer coefficients via the Odemark equat ion, the
each mix. This section deals with determination of mixture stiffness. Subsequent
binder stiffness (Sb) , binder volume , and aggregate volume . While working with
method with which to estimate binder stiffness (Sb). The use of his method
involves a nomograph (Fig . 1 ). The required input data include penetration index
10 7 108 109 ;_,10 3 2.s,103
•1 $1,llnrn 111odulu1 !N/rn~) 106
•6
.. •5 •6
i
-
:;
.
,,i
]
•2
•S
:,
~ •I
~ 0 •I
cE -i 0
-1
-2
-)
' , . .
V11co111v. p
DWG . 69 .12 . 1164 1 I oS 1(r 10 3 102 10 T 1" 3" 10··30·1· 2' 5' 10' 30 "1h2h Sh10h1d 2d 7d 30d lyur 10 y 100 y
10 fi
r·
IO S
'
I o-l
1
I
I
I
1a J 10 2
I
I
r----1
10 ·' 10
, I I,,
I
10 2
t · .II I
I
10 3
1
1 It I ',,'•I
I
10
4
10 5
I
, l,1
I
10 6
,, ,
10 7 10 8 10 9 10 I 0
Ti mr of load ing, 1
<D
Fig. 1. Van der Poel Nomogra.ph.
10
(Pl), ring and ball softening temperature, binder temperature of interest, and
20 - 500A
Pipen/pen = (3)
1 + 50A
The parameter "A" is the slope of the penetration - temperature curve, a measure
approximately the softening point for most asphalts . Thus, if T, is 25 °C, then T 2
will be the ring and ball softening temperature. The calculation of A becomes:
11
The calculation of Pl from the results of two or more actual penetration tests
has been shown by Heukelom (17) to give more accurate results (Eq. 4).
Heukelom recommended use of a "Bitumen Test Data Chart" (BTDC) which allows
plotting of both penetration and viscosity versus temperature on the same graph ,
portion of the curve representing the temperature range of interest should be used,
penetration tests, and that the Pl should be calculated from this, rather than the
curve derived from Pfeiffer and Doormaal's original method. Secondly, Heukelom
penetration = 800 line, and the temperature at this point (T pen 800 ) should be used
in the Van der Poel nomograph instead of the T R&B· This is shown in Fig. 3. In
practice, the ring and ball temperature specificat ion is no longer commonly
running the ring and ball test. Thus, in this study, Pl was calculated based on
1000 .--.--.--.-----,-.--.--.--.---,-.--.--.------,,--.--.--.---
800
-aa
-
0
-100
AC-20
s:I
-~
0
+J
"....
+J
Cl)
s:I
cu
Pot
10
AC-10 : : AC-20
R&B : R&B
44 .5 °C: : 48.5 °C
1
0 25 50 75 100
Temperature (°C)
~')
~ r "'- "'-"'-·-" --,-,.---,---"'- ·-~---· --,.-.:.c
"'.....;•_...,
..;...._ "'- - - - , 9
":' __-,-~-"'....--•-,'"'--
(/)
0
a.. ·+H fl: '/I lf·H- li"II I :-+·i· I···IJ H-+~1i1+ tt··l:f1l h-+·-~ 1··· ~
1
/ 11 If
13
JT/ 1+-L~---f j l··j···l··· f'll··r'· . +. 1:tl fl::.:.ll:+,-··
>-
(/) +r :ij'ilr Tr/-=;.1:·I -~I ··t -rr+t·
·,_._J -·1 · ·+·· r ·i" ··HI ~7f·{Tr .1.11·
1··
/
1·· o
~mt
0
u
(/)
'=i 11 I Hll l• ''l!;J J
- ·-
··· - i
: - ..--.. --.. ·$· l·t~ .· . ~-
. "f ~
~ :IfI it!J· itrr-,
...'... ..,.
iilr· I i' i.i
~,ij-c£
--..:
'- -
-
:
- I•
i
:
..;.
.
.·.;
: I,
...;...., I
,. ,
.. . -t--
..........._.
I
I
I
..- -~ ·= -~=- ..._· ~ - - Ir- ~ ._· ·_·_ ~-- 1· "Hl H+·+··- ~
Ji
I!
i! ! ii I I
I-+++++;-+; if+++-r-+-+-.......;+++;.+-,f-+~,....,...++,i-++-it-,,tttt-t---+,----1
i i : I! I
l!I ,: : !! I ("'")
I:
~-~ ~
.. _ .__: _.:_,: ,: '.. ·I ·..___.1._
.. -'-r
1 r .. __.Li 1i11.. . . . . .. •! : 1u.1 iLI . ...
l!i l!!r 8
~ ..... ,...I: ~
I II UII ;! ! I i i I I ("'")
__._ : .. l~ . --~· - -~ ~- T. ;. - . ....... 1; ~~ .(.(. --t-·
l: ../. . . l !:.... .... :~ . ~~). ~~-i···
-. -t-
-~) ··· fi · -- .-
• P-• •. . .•• - - ~,-. ·· ~ • ••
-t-t-t-~-=-"'"''..,.
:, :
··~~ ..
~~
·1tt- l: -r
iI
0
U")
....0
I: I
::~,:~ Ij::
lo'i,, i! I (\j
Lh, ,1 1 ! ~ s:::
o 11"L, IEI i ! , I! i I w .....0
~ -r~ [ITI' ' I ~.
++-+----...... /r/" it :f. iii I
C
:J
+J
a:l
..__..,. I I I I; I I-
J
·-E
i...L., 0
s:::
~
0
I""! W' I ~ 1 N
1--
' ....I-......J.-.+.+.~,',.__.,1_......_..... l! ........-+-........ ~•;1 ~ V L-+L
_ _ _..... I -H+++-+-+-......... .,..I- :;-: -t--:...-
: ~ ~
I;:;
z
~-t=:..=j-~'"':..t·;:::;::~ ~ ~ ~ rn:P~-n::..::ii:::n~•
~
I I .,; ~ ii g-+-+---+rr+++---~1-1----H+++++-I~,....--
; I~++.....,.~~='-· '...,,..,.
'· !~ - 1 0 .....bD
0 r...
I : I 11 '~ -- ./. :;
: I ··· { ... -··· - . . - -- ~;. '........ ;:! _ ;11_1_:.Ll-~-
·.·.·-r- --~-1---- : _· ~-+-._._. ;;
····- ... . . -c.~ ;., ·}_-1 ~-1--- ·-r- 11
tt :_·. 1,·
. _,
- · - . . . .
. _,__ - - .
. ·- -· - ·· - - · . . - .. - -~- . .
1" .-
- . - - ... t, ..tit .t.t.· ~-Jt~.-
=E ii : !! O
E
0
Q
,-
~
_// I•~:-, I:-~. ~
.:.-'< . , . ._ _ ,._. .__.-.,;. . . . _ _ _ _ _ ._. __.______
=......
~ ~ ~,~
b - o
U")
N ... "°
<
c:: 0
c::
14
stiffness of the asphalt mixture (Sm). Most methods involved a narrow range of
Bonnaure fil fil. utilized the Van der Poel binder stiffness value (Sb), but on plant-
aged asphalts, to better represent asphalt stiffness in the field. This procedure has
been adopted in the Shell pavement design method (22). A nomograph depicting
where:
N Jm 1
8U 90
0
,o • (V
. ~ '-' "'' ' "" "'~~
I
,o
60' "
. ' "
~~--- ------ ~
~ ":::,,_ i'-..
[",.
,o"
..,,, ~r-c-- ·- ·~ I~
-~-"--
.... ' ~
"~ "
70 " " " "
' " "' " " ' ' ' I'-
' " "- '-{
' "' " "'' , '\_""-.f'I'--..' "I'- '" ' ~'
0 '),.,, '
~
'\
I'-. I'-. ....
~ .... i'-..
~~
"'
' I
'i'-.. ~
ol ~ bo1vrn.nout
1
lu1
..
I\ '
I\\ I
"' ~\\ hl "-['\_ !,()
-
- ~
~ 0. "~ ~ "~ ~ -
-
'\ .
"~ ~
~o
- 10 '
=
-
-- 10 -
-
i=
" ,'>
\\1'~
,,
M,nc1al
---
eo . . ....
"'19'~ 1r \\" ~
\' '\
10' ll
~ 8 ftu1nie-,, ~~ ~
,_ -
6. .__l!!!__
o,o.,
Jv.Jr ' "
\ \
'
'.
S 10 70 30 100
V o l u"'< oft,.no., V 13 I 0
10
moo.J i u,
o f O'c !THI
0
,a • t \' .I
v. = %volume of aggregate
Vb = %volume of binder.
Plant-aged asphalt penetration and ring and ball values can be obtained from
asphalt residue. In lieu of laboratory aging of asphalt, Coree and White ( 12) drew
upon the work of Ullidtz (23) to approximate aged asphalt penetration (pen,) and Pl
Pl , = Pl of recovered asphalt
Coree and White (12) drew on work by Witczak (24) to estimate the prevailing
at any depth:
T.., = 6 . 0 -
...
34
(z + 4)
+ TA [1 + ---- ]
(z
1
+ 4)
.... (16)
-
where: T A = mean air temperature
depth .
Alternatively, the mean-value theorem (25) can be employed to f ind the expected
-
mean layer temperature TP in a layer of z-thickness :
4
34 ln [ - -]
TP = 6. 0 - 4
_ _ _ _z_·--+TA [1. 0 - 4
(ln [ - - ) ) / 2) (17)
Z Z • 4
-
where: TA = average yearly air temperature, °F
The Ullidtz "Sb" equation is based on the Van der Peel nomograph. It is
orginally proposed by Lefebvre (26). McLeod felt that the Pl did not handle waxy
asphalts well because these asphalts tend to exhibit a false softening point. He
1 og V = 3 . 4 6 2 8 9 - o • 61o94 1 og P . . • . • . ( 19 )
asphalt of interest).
McLeod modified the Van der Poe I nomograph to allow substitution of PVN for Pl.
He also substituted a "base temperature" for the T R&B · To use his Sb nomograph,
one first needs to determine the difference between base temperature and the
penetration test temperature. This is done by use of Fig. 5. Then one enters
w 130
c::
::,
-~
1-
c::
w
125
120
c..
I I :i
w .l..
•- 110 /....._
IO:i '
~
$)
/....._
w 100
§1-- j
..... <.fl
<w ~
/
1--
c::
w z 90 ~ /
c..o
""C -
~ /
-
w <
I-
1--
c:: 80
;....._~..y /
~~
Wl-
<.llW
< - 75
~'
s~ .,..s /
/
:~ 70
A..$"cf .,.<;/
~~ 65 ;:::-
(S~
> ~
-
I-
I-
60
,._~
u.·
C) 0
er "'- A.. /"'<?
I.J~ 10
y u. 55
~~/I
zw
0
- >-
so '?c., .s. 0
5 1--
(.11
w
w O
u _.J
z: c..
45
s s I-
j~
~o :::
w :! .§>/ - 5 0
C:
w
w
35
::s I-
I... ~-:/ '.J EXAMPLE <
c::
I...
30 ~
c ~
I
, GIVEN 3
1--
w
l.,J
25 /~ , ,; PENETRATION AT 25°C = 9 0 :::
r.: 0 s PEN-VIS NUMBER= -1 ·0
20~ '<-
I c:'
THEN
..:. -
C::. C "+ '° ------~----..--~..J-------,,_..;..-:,-..-:..-:,_.'--:,_0:::...,,...-.....:=--:-_.. . c::, z::z
::f 2: .. ~ ----------"~----1---,--rL-:-....:......,,......:__:,--..:..,,...-.:::.."?'".-::.::,........ } . ~ z
( ,
'-.
-
...,_
< t-1 V')
-
-
<
:;, C:
,- 0
>I "
:.:z..,
'-' -, ....,2 2
(..
c.. '- '-
-2
,0 00 1,000 tOp::,o
v1!.COS1T Y, GP
I \
I \
I \
BAS( Tf .. PfRATUf<f
\
TEMPJl'u.TuR( ,11.:: EIUDw BAS( l[MPfRA11uR(
100 ::,o . O 10 20 !>O • O =,o 60 70 t10 9-0 IOO
I \
I
I
\
I
\
I
I
\
I \
I \
I \
I \
I I ko/cm2= 14 ·2psi = 9·81 X 104N;m2 \
I
\
I
I
\
,.,VISCOSJTY \
1"JV ,c, 1 , S[C \ ~ IMIII IHOUA \ 10.:.Y tW((I(
I ... f.. ~ 2 ,t, 6 4 2 , , /
' , < lt £ ~ 6e 2 "6 i,
1
2 ,; i;e' 2 ,; 6t 2 4 C>e 2 4 t>t
'I \
2 4 t,t
'1 2 4 U I
CK,~. .:,.,;,, O·t IQ 100 I po,:, 10,p()O 100;,<::,0 t,/XNp:::xJ
Lc.<.::m,c T11,,i(, SE.C
Kandhal and Koehler (27) found that PVN did not change with in -service
aging, while Pl did. They also found that PVN predicted mix stiffness better than
Pl. Thus, PVN was considered by Kandhal and Koehler to be a superior method in
study.
values.
values via Eq. 12, then use of these in calculation of Pl, in Eqs . 3 and
4.
22
25°C were developed in this study from a review of the literature (27-
30):
Sm values were then calculated via Eq . 6 by use of Sb values determined from each
of the above two ways. This resulted in five different ways to calculate mixture
SUMMARY
The mixture stiffness (Sm, smm' sm,aged' smrm,aged' smPVN) of the mixtures
examined in this study were estimated by six methods; five using an aged form of
Sb and one using SbPVN· The results were correlated with actual test results of
24
resilient modulus to judge which estimation method was best. Then the results
GENERAL
alternate equations. The choice of equation depends on the character of input data
available to the user. Both equations were used in this study to assist in
determining layer coefficients . The two equations are presented by Akhter and
report)
T = temperature, °F
complex modulus, which also takes into account the viscous nature of asphalt
E* = E' + iE"
i = imaginary number
occur for short, relatively light load applications, then the complex modulus is
26
reduced to one term, which is commonly called the dynamic modulus:
( 2 7)
cyclic pulses in the form of a compressive sine wave to 4 in ( 10.2 cm) diameter 8
in (20.3 cm) long cylinders. The test has fallen into disuse because of its
cumbersome testing technique and because it has been criticized for not being
appropriate for pavement design /analysis methods which use elastic layer
assumptions . It has been largely supplanted by the repeated load indirect diametral
tensile test, which is discussed later in this report . However, the Akhter and
Witczak equations are still well-known . Their usefulness was examined in this
study .
SUMMARY
The dynamic modulus concept is useful for this study as follows. The
available from Road Test data, UMR-study data, and MHTD mix design data.
Thus, use of Road Test, UMR-study, and MHTD dynamic moduli in the form of
estimated values can be used in the Odemark equation in the determination of layer
coefficients:
27
i: E• : UHR or KHTD]
113
a n. UMRorHHTD = a n ,MSH 'E • '
I I AAS HO
RESILIENT MODULUS
The resilient modulus is a repeated load test that is similar to the dynamic
modulus but with several important differences. The applied stress wave form is
usually in the form of a stress pulse followed by a rest period, rather than the
sinusoidal wave form (with no rest period) as used in the dynamic modulus test.
The test equipment is less complex. There are two different ways in which the
test can be performed. One method, AASHTO T-274 (35), utilizes compression
chamber. The second method (which is now the predominant method), ASTM
Marshall-type specimens . The latter test is more convenient to perform, and is the
method used in this study . It is the recommended test in both AA MAS (Asphalt-
substitution of mix characteristics and other readily accessible data into the
modulus data gleaned from the literature combined with data from this study.
28
Layer coefficients were determined by applying the resilient modulus data
developed in this study along with estimations of resilient modulus to the AASHTO
modulus data and estimated resilient modulus from AASHO Road Test mixtures:
3
- _ MR, UMR or MHTD]l/
an, UMR or MHTD - an, AASHO - t---.,.,M:-------
R,AASHO
LABORATORY INVESTIGATION
MIX DESIGNS
Mix designs were developed for Type C, 1-C, and plant mix bituminous base.
The main thrust of this portion of the study was to determine layer coefficients for
the three types of mixtures based on repeated load indirect tensile diametral tests.
This is the test recommended in the 1986 AASHTO pavement design method ( 1).
prediction of resilient modulus without having to actually perform the test in cases
grade, asphalt content, and indirectly, void content. The latest SHRP protocol (38)
29
has omitted test load frequency as a variable, therefore frequency was not varied
in this study.
The proposal for this study did not include the performance of m ix designs
for two reasons: 1) limited available funds and limited contract duration, and 2)
the thrust of the research was to study the effects of varying the aggregate
gradation across an acceptance band, with the interaction of asphalt content and
grade. However, to assist in determining optimum asphalt content during this part
of the study, Marshall mix designs were performed for all mixtures containing ~C-
20 grade asphalts . Some mixtures had AC-10 asphalt substituted for the AC-20
with no change in mix design because mixing and compaction temperatures were
Two gradations each were chosen for the Type C, 1-C, and bituminous base
mix materials, resulting in six gradations. These were picked by a process which
involved determining the coarsest and finest job mix formula (JMF) gradations that
were approved by the MHTD during the 1990 and partial 1991 seasons. Then, to
get an even wider separation of gradation, the coarsest JMF gradation was pushed
Likewise, the finest JMF was pushed to a finer gradation via the maximum
within the master specifications on some sieves, and at the - #200 sieve in order
realized that some of these mixes may not have been approved in routine work,
30
but they do approximate where some gradations may end up in the fie ld after
adjustment.
Two coarse aggregate sources each were used for the Type C, 1-C, and
bitum inous base materials . These were chosen by MHTD personnel to exhibit a
Two asphalt cement grades were chosen (AC- 10 and AC-20) to represent
analyses were made to determine optimum asphalt contents. Once these were
m ixtures in all were evaluated in the res ilient modulus testing: 16 mixtures each
for the Type C, 1-C, and bituminous base mixtures. These are shown in Table 5.
Coarse Gradation X X X X X X X X
Note : 1 . This chart applicable to Type C, 1-C, and bituminous base mixes . Thus 48
mixtures were used .
2. "X" denotes that this combination of parameters was
represented by a mix.
The choice of using the Marshall mix design method for determining
31
optimum asphalt contents was based on several factors . First, it is the most
commonly used method by state and federal agencies, private practice, the Asphalt
Institute, and the National Center for Asphalt Technology . Second, the MHTD
utilizes this method to a certain extent in its mix design evaluation. Third,
personnel communication with MHTD personnel indicates that the Marshall method
will be the preferred method if a contractor QC/QA program is initiated, and fourth,
the UMR Bituminous Laboratory is equipped with Marshall equipment and has
The optimum asphalt contents were chosen based on percent air voids as a
major criteria, but also were optimized in an attempt to satisfy MHTD requirements
for stability, voids filled, dust/asphalt, VMA, percent natural sand, inclusion of
hydrated lime, percent asphalt, and makeup of fines where applicable. Although
flow is not specified by the MHTD, this parameter was also used as a guidance
criteria.
It was decided that in order to make comparisons from mix to mix within a
given type (C, 1-C, or bituminous base) of mixture, the percent of aggregate
constituent (coarse, natural sand, manufactured sand, mineral filler, and hydrated
lime) per sieve would be kept constant. For example, for the Type C mixes, at the
#16 sieve, all 16 mixtures would retain 0% coarse aggregate, 79% natural sand,
The decision about the kind of materials going into each mix type was made
average, the Type C mixtures contained 48.0% coarse aggregate, 24.1 % natural
sand , 22 .9% manufactured sand, and 5.0% mineral filler . The percent retained of
each type of aggregate on a particular sieve had to be changed from sieve to sieve
in some cases in order to make the total contribution of each material type
reasonable. The gradations are given in Table 12 in the "Results" section of this
report.
The following are brief descriptions of the test methods employed in this
ASPHALT CEMENT
Penetration
Both grades of asphalt were tested for penetration as per AASHTO T49-89.
Test temperatures were 77°F (25°C) and 37.8°F (3.2°C) . The penetration at 77°F
(25°C) information was necessary for calculating PIR&B• Plpen/pen • PVN , and for use
in the Ull idtz aged penetration equation. Penetration at 37.8°F (3.2°C) was
necessary for calculating Pl pen/pen· Both were used for estimating T penBOO·
Kinematic Viscosity
AASHTO T201-90. This information was required for calculation of PVN, for
Absolute Viscosity
These data were necessary for determination of mixture mixing and compaction
Specific Gravity
(25°C). These data were used for volume calculations in the mix design process
and for use with the Van der Poe! nomograph. Fig. 7 depicts the ring and ball
AGGREGATE
Initial Gradation
capable of reading to the nearest 0.1 g. This information was useful for
Final Gradation
gravity and for making asphalt mixture specimens. Fig. 8 depicts the steel storage
w
..:,.
35
Numerous studies have shown that aggregate particle shape and texture
concrete mixtures. It is difficult to separate the effects of shape and texture. The
the following also increase: stability, resistance to rutting, VMA, and optimum
asphalt content. Opinions are somewhat mixed as to the effect of shape and
texture on static indirect tensile strength (IDT). In regard to IDT, Kalcheff and
Tunnicliff (40) found little difference between various particle shape /textures . The
explanat ion was that, in compression, particles attempt to slide past each other,
particles do tend to result in higher IDT values. Also, the literature indicates that
the characteristics of the fine aggregate are much more important than those of
Numerous test methods have been devised to quantify particle shape and /or
texture. These can be divided into direct methods (those that result in
methods (those that measure some sort of bulk aggregate property, such as void
methods were reported by Meier and Elnicky (43), Mogawer and Stuart (44), and
Kandhal et§.!. (45) at NCAT (National Center for Asphalt Technology) . It appears
36
that efforts are being concentrated in the area of fine aggregate evaluation, and
that there are several methods available which can be used in lieu of the standard
or B) for fine aggregate (47). Both of these are indirect methods of particle shape
determination.
In this study, the(-) #8 to(+) #100 sieve size material of each asphalt
mixture blend were tested using the NAA Method A. The method is given in
Appendix A of this report. For the ( +) #4 size, the blends were tested in
accordance with ASTM D 3398. This method is given in Appendix B. The results
of both methods were used in developing the indirect tensile strength and resilient
report. Photographs of the NAA test device and the D3398 equipment are shown
Specific Gravity
and #100 sieve sizes and tested in accordance with AASHTO T85-88 and T84-88
for the ( + )#4 and (-)#4 to ( + )#100 material, respectively. For the (-)#100
These data were used for voids analyses calculations. Weighing was performed on
w
'-.J
38
Screening
All aggregates were dry shaken through the appropriate screens on a Gilson
order to minimize problems with incorrect sizes of material being retained on any
given sieve.
Upon being shaken, the split material was stored in 20 gal plastic cans with
FABRICATION OF SPECIMENS
All asphalt mixture specimens were made in accordance with AAS HTO T
each face of the specimens. The specimens were Marshall-type pucks: 4 in ( 10.2
cm) diameter 2 .5 in (6.35 cm) high cylinders. Usually, two or three pucks per
asphalt content were made for mix design purposes, two for static indirect tensile
strength, and four for resilient modulus testing. Two uncompacted specimens per
asphalt content were used in the Rice method for maximum theoretical density.
centistokes and 250 to 310 centistokes, respectively. From the relationship in Fig.
11, mixing temperatures of 298 °F (AC-10) and 302 °F (AC-20) temperatures were
(j
@ 1~ 0 F Cst ~ I Poises / ( .98 x Sri cci l,c Gravily )! ~ 100
~
@275 F Cs t c I Poises/ ( .9 34 x Specific Gra vi ty )! x 100
~
2
- ·"·~
- -~
10 ~
6 l""' ' ' I
..:
' ' '\
'-
'\
I
I\.
3 '\ '\.
~--
2
"'\. "'\.
'\. '\
10 ..: '\l""-
e ' '
"" ""
' '\ '\
Cf)
w 6
~ I\.
0
I- ..: '\ '\
["'\ "'\
Cf)
I- !
2 '\. '\. I
w
0 2 '\. '\.
~ '\_' ~
>-
1-
vi
O
o
10 3
s
e
~'\ ' '
'\ '\
Cf)
> 6
7
'\. '\
'\
~
,.
~----- ------
8-
------ .... ----- i------- ------ ---~
COMPAC_T ION R, ~N GE
~ ____ J_ _____
"""" \J\.
iI"" " I----- ·
2
~----- ---------- - ------r------ ~------ L.. ----- t~1--~
I
A- Ml~ING Ri JNGE I, ~
11
I1
l,
~I I"'
\1
E I I
,1
7
1J I•
6 11 tl
;,
I
11
11 .I
/?~ 1.; o
- V 160 18 0 200 225 250 275 30J
h .. po, , c:! l ,
f ~f A ::i hH t l lf l~TllU l[ TEM PERAT UR E IN °F
( :.> t : I C, f P .L , • ,,. ._ P-. 1 £ , ~r,
279 °F (AC-10) and 282 °F (AC-20) were used. Details of specimen fabrication are
given in Appendix C.
Bituminous base mix aggregate and mineral filler were separated into 1 ",
3/4", 1/2", #4, #8, #30, #200, and -#200 size fractions. Type C and IC mix
aggregate and mineral filler were separated into 1 ", 3/4", 1 /2", #4, #8, #16, #30 ,
The Rice specific gravity specimens were made in a similar manner, with the
245-90 . Prior to testing, the specimens and breaking head were heated to 140°F
(60°C) in a water bath. The pucks were tested in a Pine Press Marshall device
which applied the load via a motor driven mechanical jack at a deformation rate of
2 in/min (0.084 7 cm/sec). The load was sensed by a 10,000 lb load cell. Flow
was measured by a Schaevitz LVDT (Model GCA-121-500 S/N 4427) which has a
Both signals were sent through signal conditioners; the output was recorded
that load were taken from the load-deformation trace. The test arrangement is
The maximum theoretical specific gravity was determined for every mixture
41
A water bath dedicated to the Rice system was used for both the
pycnometer calibration and test procedures. The electronic balance had a capacity
of 12, 100 g and was readable to 0 . 1 g. During the deairing step, a vacuum of 30
accordance with AASHTO T 166-88. The data were necessary for determination
VOIDS ANALYSIS
An analysis of the voids system in the mixtures was made utilizing the
results of the specific gravity testing . Air voids, voids in the mineral aggregate
(VMA) and voids filled (VF) were calculated. Air voids were calculated based on
above, and the results were compared . The voids analysis was necessary for
specimen with a static compressive load. Loads are applied along and act parallel
42
to the specimen's vertical diametral plane. Lower and upper 0.5 in wide steel
loading strips, which are curved at the interface to fit the radius of the specimen,
uniform tensile stress perpendicular to the plane of the applied load and along the
vertical diametral plane . This ultimately causes the specimen to fail by splitting
along the vertical diameter. The tensile stresses developed within the specimen
simulate the state of stress at the lower position of the asphalt layer in a pavement
structure, which is generally the critical area for fracture and fatigue cracking.
Procedures for the indirect tensile test have been developed and reported by
loads to be applied during the repeated load diametral resilient modulus test. A
percent of the total stress at failure is normally used . In the SHRP P07 (Strategic
mixture cores, repeated load level is tied to test temperature : 30% at 41 °F (5°C),
follows:
44
2 p
s = ( 28)
rthD
test press w ith a different break ing head . Load is sensed by a load cell; the signal
is condit ioned and maximum load is determined from an XY recorder plot . A more
RESILIENT MODULUS
The resilient modulus (indirect tensile) test is similar to t he ind irect tensile
strength test except the specimen is not loaded to failure; rather, it is cyclically
The equipment used for testing asphalt specimens for d iametral repeated
load resilient modulus was developed at UMR. Several modifications over a period
The total resilient modulus was calculated as follows (as per SHRP Protocol
P07):
0.859-0.0BRt
=
0.285Rt-0.04
Ht,, Ht 2 = total horizontal cyclic deformations from horizontal L VDT #1 and #2,
in
t = specimen thickness, in
Vt,, V 12 = total vertical cyclic deformation values from vertical LVDT's #1 and
#2.
of the horizontal LVDT's, vertical L VDT's, and the load cell are listed. Actual
minimum deformations and loads during the testing were kept at least ten times
ASPHALT CEMENT
The results of the asphalt cement testing are shown in Table 6. The number
of test replications are shown which were necessary to stay within AASHTO
47
precision guidelines.
performed as per AASHTO T49 with 1 OOg weight at 5 sec duration. The
penetration at the lower temperature was performed under the same conditions,
rather than the suggested 200g at 60 sec duration. A review of the literature
(28,51) indicated that researchers favor the 1 OOg 5 sec method when Pl is being
determined .
Sb values were calculated based on Eq . 14, which is based on the Van der
Poel nomograph (Fig. 1 ). Use of Fig. 1 assumes that the asphalt is an S-type of
48
bitumen, as opposed to a W-type (high wax content) or 8-type (high asphaltene
type). de Bats and Gooswilligen (52) give criteria for qualifying the asphalts as to
type. The method is given in Appendix D. From this analysis, both asphalts (AC-
10 and AC-20) used in this study were classified as S-types, therefore use of Fig.
1 and Eq. 14 is appropriate . Also, Heukelom defines S-type asphalts as those that
plot in an approximate straight line on the BTDC paper. Fig . 3 reveals this type of
behavior for both asphalts used in this study. Fig. 3 also indicates that the T 800
and T R&B are quite close , as would be expected for S-type asphalts.
mixture stiffness. In Table 7 are aged residue estimations of penetration and T R&B
respectively . Use of Van der Poel's Sb nomograph (Fig . 1) indicates that the
type S asphalts . Thus , it would seem that use of the penetration test at a second
temperature (as opposed to the ring and ball softening point) is appropriate.
Binder stiffness, Sb, was determined for loading times of 0.1 sec and '.).04 sec.
These loading times correspond to the load duration time of the resilient modulus
testing (0.1 sec) and to the estimated load duration time for MHTD pavements at
the average vehicle speed (56 .3 mph) and average asphalt pavement thickness
(8.33 in) in accordance with Barksdale (53) as explained later in the "Load
Aged residue data were used to better reflect actual pavement conditions . PVN
has been shown not to change with aging, therefore unaged penetration values
:::- 20000 50
- l:ll
0..
.c
rn 15000
l:ll
l:ll
-
Q)
s:=
:".:: 10000
_.J 0 AC-10, Sb(PI)r
I:'/)
e AC-10, Sb{PVN)
M
Q) t::. AC-20. Sb(Pl)r
'g 5000 • AC-20 , Sb{PVN)
•.-4
co
-5000 --~~--~~........~~~---~~--~~~
0 10 20 30 40 50
Temperature (deg. C)
Fig. 16. Binder Stiffness From PI and PVN for 0.04 sec
Loading Time. r
25000 --~~--~~--,.~~~.....-~~--~~--
• AC-20, Sb(PI)r @ 0.04 sec loading
-
l1l
0.. "' AC-20. SbPVN @ 0 .04 sec loading
-
Q.)
s:=
:-;: 10000
_.J
I:'/)
M
Q)
-a 5000
s:=
•.-4
-5000
0 10 20 3040 50
Temperature (deg. C)
Fig. 17. Binder Stiffness From Plr and PVN for AC-20.
51
were used in calculation of PVN. As can be seen , binder stiffness decreases with
increasing temperature and longer load duration (or slower vehicle speed), as
asphalt, the PVN method exhibited lower binder stiffness than the aged residue Pl
method.
approximated the resilient modulus test data generated in the present study. The
most accurate method would then be used in the Odemark equation to determine
Sm 0.914
SmPVN 0.910
Sm,m.aged 0.886
Sm . aged 0.880
smm 0.849
As can be seen, all methods produced good estimations. The most accurate was
the Bonnaure method utilizing the Ullidtz asphalt aging equations, and thus this
method was used in the rest of the study to calculate mixture stiffness.
52
AGGREGATE
Gradations
Gene ral. The results of the aggregate testing are shown in Tables 11 through 19
and in Figs. 18 through 24 . Table 11 shows the as-received gradations of the DR-
Percent Passing
Sieve size DR4 DR5 DR6 DR7 DRS DR9A DR9B MF HL
1 in . 100 100 100 100 100 100 100 100 100
3 /4 in . 100 100 100 100 100 100 100 100 100
1 /2 in . 94 99 78 84 100 100 100 100 100
3/8 in . 63 71 -- - - 100 100 100 100 100
#4 5 18 52 43 99 100 100 100 100
8 3 4 42 32 93 90 73 100 100
16 3 2 -- - - 81 54 44 100 100
30 3 2 30 20 56 28 26 100 100
50 3 2 -- -- 31 13 14 100 100
100 3 2 -- - - 14 8 8 98 99
200 3 1 11 8 1 6 6 88 99
Note : MF = mineral filler ; HL = Hydrated Lime
Table 12 shows the gradations of the six final blends: two Type C's, two Type 1-
C's, and two bituminous bases. Fig. 18 shows the Type C MHTD master
specification, and final blends (fine and coarse) . Likewise , Figs . 19 and 20 depict
Type 1-C and bituminous base mixes, respectively . As explained previously , the
53
100
,,
/
,,
-
/
/ I /
90
E-<
,, "' I
zt:z;:l //
,, I
I
I
I
80
u //
// I
I
I
~ //
//
I I
-
t:z;:l // /
10
0.. //
// /
/
I
/ / I
/
c., 60 / /
z
..... /
/
/
/
/
/
/
/
/
/
/
(/) / / / /
/ / / /
(/) 50 / / / /
< /
/
/
/
/
/
/
0.. / / /
/
....:l 40 / /
<
E-<
/
/
/
//
,
//
0 30 , ,
/,
,
//
E-< ,,/ /
/
20
/
/
/
, /
/
- /
~
/
V ,.
10 / I /
'II
0
100 10060 :,o 18 II . :9/11 ID 1/Z ID S/4 ID I ID
SIEVE SIZES
o Specification Limits
• Test Gradations
100
I I ,,,_
,N I ,,,_
I
110
-...
I
I I
FF 1
TI I
BO
:z: , ,
, ,
I
I
l':a:l 1 ·1 F
t.) , , I I
I
= 70
,, , I
I I
I
-
l':a:l
0..
u
eo I
I
I ,I
I
,7
I I
-
:z:
C7l
C7l
.:
50
,,
T
I
F
I
,
, ,
I
I
I
I
I
,
J
_,
,,
,,
0.. '40 ,, , ,
M /
,,/ ,,/ , ,
, ,, , ,,
...
.:
30 ,
, ,, ,,,
,
...
0
20
,,
II
...
10
0 _ _..____________.._____________,_______________......_ _ _ ___.
SIEVE SIZES
a Specifi c c1 lion Limi ls
• Tesl Grc1da lions
100
/
/
90
/
~ - /
/
/
/
-
/ /
/ I/
80 / /
E-< / /
z /
/
/
/
~ / /
u 70 / /
a:: /
/
/
~
-
,I /
r:i. 60 ,//
I ~
// /
C, // /
-
z
-
// /
/
50 -, /
Cl.l / /
Cl.l // /
< // /
-
/
r:i. 40 /
/
...:I II
<
E-< 30 I
j
/
/
0 ,, I
/
/
E-< , ~
n ./
20 / ~
ff ~
/
;'
10 II /
0
200 100 50 30 11, e • 3/8 1D, 1/2 lD 3/4 1D. 1 lD.
SIEVE SIZES
D Specification Limits
• Test Gradations
• Coincidence of Test
Gradation and Specification
Limit
Fig. 20: Final Bituminous Base Blend Gradations.
56
fine and coarse blends for each of the three types of mixtures represent the f inest and
coarsest gradation approved by MHTD during 1990 moved to the finest or coarsest
limit allowed by individual sieve tolerance. The m ixtures in this study may not be
totally realistic field mixes, but the wide spread in gradation was necessary to satisfy
one of the major criteria for this study-the examination of the effect of gradation on
% Passing
Type C Type 1-C Bit. Base
Sieve Fine Coarse Fine Coarse Fine Coarse
Size
1 in. 100 100 100 100 100 100
3/4 in . 100 100 100 100 -- --
1 /2 in . 100 99 100 99 90 60
3/8 in . 90 75 88 76 -- --
#4 62 42 60 44 65 35
8 40 26 39 28 47 25
16 32 22 29 18 -- --
30 23 18 23 13 35 15
50 15 13 17 9 -- --
100 9 4 10 5 -- --
200 5 2 5 4 9 5
aggregate, on a sieve by sieve basis . The idea was to try to keep the percentage
57
makeup on each sieve the same for the fine and coarse gradations. Then, if during
the analysis portion of the study it turned out that a particular sieve was critical to,
say, resilient modulus, the fine and coarse blends could be compared.
Size %Retained %CA %MS %NS %MF %Lime %Retained %CA %MS %NS %MF %Lime
3/4" 0 0
1/2" 0 1 100 0 0 0 0
#4 28 100 0 0 0 0 33 100 0 0 0 0
8 22 0 83.3 16.7 0 0 16 0 83.3 16.7 0 0
16 8 0 21 79 0 0 4 0 21 79 0 0
30 9 0 21 79 0 0 4 0 21 79 0 0
50 8 0 21 79 0 0 5 0 21 79 0 0
100 6 0 21 79 0 0 9 0 21 79 0 0
200 4 0 10 0 90 0 2 0 10 0 90 0
-200 5 0 33 0 67 0 2 0 33 0 67 0
3/ 8 12 100 0 0 0 0 23 100 0 0 0 0
#4 28 100 0 0 0 0 32 100 0 0 0 0
8 21 0 100 0 0 0 16 0 100 0 0 0
16 10 0 63 37 0 0 10 0 63 37 0 0
30 6 0 63 37 0 0 5 0 63 37 0 0
50 6 0 63 37 0 0 4 0 63 37 0 0
100 7 0 63 37 0 0 4 0 63 37 0 0
200 5 0 63 37 0 0 1 0 67 33 0 0
-200 5 0 0 0 20 80 4 0 0 0 25 75
58
Table 14 shows the overall contribution of each material to the final blends
compared to the 1990 MHTD average mixtures . As can be seen, when viewing the
mineral filler, and lime, the proportions of the UMR mixtures closely followed MHTD
field mixes.
59
gradation upon resilient modulus. The most promising methods were later tried in the
would represent the shape and position of the gradation curves. Eight different
methods were tried, and are described in the following paragraphs. The results of the
1) Fineness Modulus
(L cumulative % ret 1d on 1. 5 11 , 3/4 11 , 3/8 11 , #4, 8, 16, 30, 50, 100 sie v es )
F/:1=~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ ( 3 0)
1 00
drawback is that the effect of the minus #100 material is not accounted for, which
The coefficient of uniformity (Cul and the coefficient of skew (C 1 ) come from
the geotechnical f ield and are used to help classify soil particle size distributions as to
61
Cu = D 10 • • • • • • • • • • • (31)
DGo
(32)
A = (L %passing l. 5 11 , 3/4 11 , 3/8 11 , #4, 8, 16, 30, 50, 100, 20 0 sie ve s ) (33)
100
3) SF/SSF
sieve series is 1 1 / 2" , 3 /4", 3 / 8", #4, 8, 16, 30 , 50, 100, 200 for SF;
In asphalt work, the position of the gradation curve relative to the position of
the maximum density curve as plotted on the FHWA 0.45 power curve graph paper
has been used to assist in predicting asphalt mixture behavior . In this study it was
decided to use some measure of this relative pos ition to help in the estimation of
resilient modulus.
The method developed in this study was to determine the area between the
two curves. This can be done in a variety of ways . In this study, the use of the
plotting program AUTOCAD (57) was used. Use of a manual planimeter or just
To find the area between the maximum density line (MDL) and the gradation
curve of interest, the position of the MDL must first be determined. There are at least
four methods reported in the literature that have been presented for determining the
position of the MDL (58-61 ). The method of Goode and Lufsey (58) seems to offer
the most likely way to estimate the largest maximum particle size, and hence the
63
truest position of the MDL, which flows from the largest particle size . In their
method, a line is drawn from the origin of the 0.45 power paper up through the point
on the gradation curve that represents 90% passing, then on up to the 100% passing
line. If the point that is struck on the 100% passing line is to the left of the point of
100% passing of the actual gradation, the straight line is used as the MDL. If the
point is to the right, another straight line is drawn between the origin and the 100%
passing point of the actual gradation and this line is used as the MDL. These plots
Because the positions of the MDL of each of the six experimental gradations
were all different, it was decided to use a ratio of the area between the curves to the
area beneath the MDL and the 0% passing line, which is a large triangle. The area
calculation results are shown in Table 15. As can be seen, the C and 1-C fine mixes
hewed most closely to the MDL, with the bituminous base fine mix somewhat further
away. The coarse mixes were, predictably, significantly further away, with the C and
1-C coarse mixes exhibiting the greatest relative area under the curve.
These eight moduli were each correlated with the results of the asphalt mixture
resilient modulus testing. The results are shown at the bottom of Table 15 along with
the R2 factor, which is a measure of the strength of the correlation. It appears that
-
A correlates the best with resilient modulus. Consequently, this parameter was tried
in the predictive equation for resilient modulus as discussed later in this study. Fig.
Two other methods of quantifying the position of the gradation curves were
Type C - Fine 100 T "De C - C oa.rse
..,,
100 64
90 l1. 90
j I
f I ,J I
I
80 80
,/ I
'
I '
I
70 70
60
I 'J
I
I
60
./I I
I
I
50
I,'/ I
I
50 .I I
,J I • 7 I
40 . 40 , I
30
j[7 I
I
30 I
I,
I.J
I I
V L.,, I
20 I I
20 • I
lJ J I 1•1/ I
10
I
10 I
I IJ
0 .....
IY
Ill ..
II I
'
I
100 100 T
·,r :>e IC - C oarse
90 90 r:
bO 80 80 1 I
I
....c 70 70
/j I
ll'l
Ill
G.1 60
I I
60
'ti'/ I
I
'1-4
~ 50 I
' I
50 I, '
I
11
I .
I
cCl) I I
40 '
I
40 I
,...0
Cl) 30
I I
30 I
I,
I/ I
I
'1-4 I I 20 ' I I
20 ll I
I I
I•
10 10
I ~
! I
0 .....ltt 11
II I
' 1/1 1/1
I
1/4 l
0
I/'
111 I t
100 I t
II I 4 1/1 l/1 1/4 I
80
I
80
/, I ~ '/ I
70 70
VI
I
60
J
I
I I
;,
J /
60 I I
V
I
I . I
50
50
l/~ I
,
,/ I
I
40 --,
I 40 ,/
/ I
30 I
I .
I
30 I
'
/
/
I
I I/
20 . 20 1, L,; I
"
I
I I,, I
10 10 I
I
l/ [) I
0 I 0
. . . IO II I 4 1/1 1/1 1/4 I
Ill I t II I 4 1/1 1/1 1/4 I 100 10
Ill II
....
-,
100
, ,
... ...
II
. I -
,
-
•• I
90
,
,
~ I
I'
,,
,, ..
+1 A
,, ,, '
• I
i::: 80 I
QJ
c:, '-
.... , ,, . ,,
'QJ"' -· I
,,
-
70 ' - -I I
~
eo
,
,,
-
I-
7"-
,
..
'- I
,,
I
I
,
,,,,, ......
tlO . ,;
.....i:::11.1
I ,
/
I r_
,,, ,,I
.,...
,,,,,, ,
.•
11.1 :10 , :, / - ecreas1n A
' ' ,,
L
aj
,
/ -A
~ ,
-"
llf
.,
.,,
I I
40 .,
,
, ~
' 77
/.I n Hit H
"'·" -,;,., . fiTIP b Hi
+1
0 JO I , //
//.,
,,. A T .. ""., If
- A. ?P.
-
,I_
n'··-
- . .. ITr ,..,,,.- n o <I Q?
20 I
I·
,
, ,
~-
., .,,.., .. ~
r-1
~,
.
0
200 100 0030 1e e 3/8 1/2 3/4
Sieve Sizes
could each of the eight methods just discussed. These two methods are presented
below.
The first of these two methods was to simply use the cumulative retained
percent on certain key sieves. Akhter and Witczak (34) have previously found the #4
and 3/4 in sieves to be important, and also included the percent passing the #200
sieve . The use of this approach is explained later in the section dealing with the
breaking the line into several parts and determining the slopes of the portions. The
= % passing #4 - % passing #2 OO
(37)
Dt 45
- Diot 5
where: 04 = sieve opening, #4 sieve
The three slopes for each of the six test gradations are listed in Table 16.
Again, the results of using this approach are outlined in the section of the report that
sieve material, and by NAA Method A for the (-) #8 through ( +) #100 material for
each blend. Both are measures of void content of bulk aggregate which is related to
"Uncompacted Voids Percent" (U). The results are shown in Table 17.
Round, smooth particles give IP's of 6 or 7, while angular, rough particles result
in values of more than 15. The range of IP's of the combined aggregates in this study
was 9 to 12. The Particle Index was determined for the coarse aggregate fraction of
each blend and the Uncompacted Voids content was determined for the fine
aggregate fraction.
Looking at Particle Index values, the DR4 aggregate averaged 12. 7, indicating
that it was the most angular . The other aggregates, in descending order of angularity
were DR5 ( 10.6), DR7 ( 10.1), and DR6 (9.4). As it turned out, Particle Index tests
were performed only on one type of aggregate per test because the ( +) #4 sieve
68
BB 6F DR6 9 .1 DR6+DR8 42 . 5
BB 6C DR6 9.7 DR6+ DRS 43 .0
BB 7F DR7 10.3 DR7 + DRS 43 . 5
BB 7C DR7 -9..JL DR7 + DRS 43.5
Average 9.75 43 . 12
DRS 38.3
ORSA 44 . 1
DR98 44.0
DR4,5 , 6,7 = Coarse aggregate
DRS = natural sand
DR9A,9B = manufactured sand
"blends" only contained coarse aggregate. So, Particle Index values were measures
The NAA method was used on the fine aggregate fractions , thus the results
indicate the weighted average of the combined natural sand (DRS) and manufactured
sand (DR9A or DR98) . Additionally, particle shape was determined for each of the
three individual sand sources. The results verify that the manufactured sands were
more anglular than the natural sand (44 vs 38). The IC mixes averaged 42.85, the
bituminous bases 42.5, and the C mixes 39.85. These results were not unexpected
due to the smaller amounts of natural sand in the 1-C and bituminous base blends
compared to the C blends. The 1-C average was slightly more angular than the BB
average, possibly because the DR4 and DR5 aggregates were more angular than the
Looking at the data a little differently, the fine aggregate degrees of angularity
in descending order were 887 (43. 5), IC4 (43 .0), IC5 (42. 7), 886 (42. 7), C4 (40.0)
Specific Gravity
In general, each aggregate was split into three portions (if possible): the(+) #4,
(-) #4 to ( +) #100, and the (-) #100 . Then each portion was blended together to
equate to the final gradation of interest as per the percentage contributed by the
On the ( +) #4 and the (-) #4 to #100 portions, both the bulk and apparent
70
specific gravities were determined, as well as their average. This average represents
an estimation of effective specific gravity . Also, Rice spec ific gravities were
performed on loose asphalt mixtures, and effective specific gravities were calculated,
as shown in Table 18. For the (-) #100 material, only apparent specific gravities can
be determined . The combined bulk specific gravities (Gsb) for each blend are also
shown.
C5C 58 38 4 2 .568 2 .682 2 .625 2.611 2 .670 2 .640 2 .750 2 .591 2 .633 2 .636
IC4F 40 50 10 2 .616 2.710 2 .663 2.633 2 .702 2 .667 2 .644 2 .627 2.672 2 .663
IC4C 56 39 5 2 .629 2 .718 2 .673 2 .638 2 .712 2 .675 2 .619 2 .632 2 .669 2 .671
IC5F 40 50 10 2 .577 2 .701 2 .639 2 .608 2 .687 2 .647 2 .657 2 .600 2 .637 2 .645
IC5C 56 39 5 2 .572 2 .687 2 .630 2 .605 2.677 2 .641 2 .632 2 .588 2 .632 2.634
BB6F 35 30 35 2 .651 2 .795 2.773 2 .649 2.773 2 . 711 2 .709 2 .671 2 .753 2 .715
BB6C 65 20 15 2 .667 2 .790 2 .729 2 .661 2 .766 2 .714 2 .750 2 .678 2.726 2 .729
BB7F 35 30 35 2 .608 2 .694 2 .651 2 .576 2 .680 2 .628 2 .709 2 .633 2 .685 2 .664
BB7C 65 20 15 2 .592 2.688 2 .640 2.574 2 .680 2 .627 2 .696 2 .603 2 .670 2 .646
Estimation method:
100
D ==-------
1 p P2 ( 3 8)
1 + +
G1
where:
P3 ::
Percent of (-) #100 mtr 11.
G3 Combined specific gravity of (-) #100 mtr 11.
::
BSG 1 + ASG 1 of ( +) #4 mtr 1 1.
G1
2
Rice Method :
p mm - pb (mi x )
(39)
where :
Maximum theoretical specific gravities were calculated in two ways . The first
was based on an estimated effective specific gravity (as discussed above) and the
total asphalt content by weight of mix. The second way was to calculate the
maximum theoretical specific gravities based on the effective specific gravity of the
Estimation Method
100 + pb ( agg )
D=
pl
+
P2
+
p }
+
pb(agg ) . . . . . . . ( 40)
G1 G2 G} Gb
Rice Method
( 41)
Results of the two methods of both effective specific gravity and maximum
theoretical specific gravity are shown in Table 19 . Comparisons of G.e vs D,, and Gmm
vs Dare shown in Figs . 23 and 24, respectively. Statistical analysis indicates that the
(G.e-D,) and (Gmm-D) values for the C and IC mixes are not significantly different, but
the difference between either one of them and the bituminous base is significant at
the 0.05 level. It appears that for aggregates with low absorptions there is not much
VOIDS ANALYSIS
From the results of the specimen bulk specific gravities , the maximum
theoretical specific gravities, and the material percentages and specific gravities, the
air voids, VMA, and voids filled were calculated for the specimens made in the mix
design portion of this study. Results of the voids analysis are shown in Appendix E.
74
Table 19. Maximum Theoretical Specific Gravities and Effective Specific Gravity by Two
Methods.
Estimated Method:
V= - -
D-d * 100, , , • • . . . , . (42)
D
-_0 . 020
0
I
IP
m
t,
--0.010
0 . 000
-0 . 01 0 - - - - - - - - - - - - - - - - - - -
0.5 1 .0 1.5 2.0 2 .5
Absorption (%)
Fig 23. (G 5 E - D 1 ) vs . Absorption.
0 . 040
0 . 030
-
Q
1
E 0 . 020
8
t,
_.. 0.010
0.000
-0.010
D1 - d1
VMA = *100 . . . . . . . . . (43)
D1
d1 = 100d
(44)
100 + Pb(agg)
VF = VMA-V*
VMA 100 • • • • • • • • • (45)
Rice Method:
Gmrn - Gmb
Pa = * 100 . . . . . . . . . (46)
Gmrn
VMA - p
VF = a * 100 • • • • • • . • • (48 )
VMA
Air voids were computed by use of: 1) the calculated effective specific
gravities, and 2) the effective specific gravities derived from the Rice testing.
Comparisons of these are made in Figs. 25 and 26. The use of the estimated method
assumes that the effective specific gravity is midway between the bulk and apparent
77
10
i::
.....
0
..., 9
- ct1
::,
-
t)
Q
ct1
B
7
>.
.0
I'll 6
"O
.....
0
> 5
·-.....<
~
4
i::
Q) 3
Q
~
c:, Type C
Q)
p.. 2 o Type IC
0
0 1 2 3 4 5 6 7 8 9 10
Percent Air Voids by Rice Method
10
C:
....0
.J
9
.....::,a:, 8
-u CJ
a:,
7
Line of Equality----;;i~~/
/
>,
.0 6
rn
....0
"O
5
>
....
M 4
<
..,.J
C: 3
(IJ Regression Line
CJ
M
(IJ
2
0..
1
0
0 1 2 3 4 5 6 7 8 9 10
Percent Air Voids by Rice Method
specific gravities. It appears that th is gave an air void value quite close to that given
by the Rice method for the Type C/1-C mix aggregates. However, with the estimation
method, the absorptive bituminous base coarse aggregates predict lower effective
specific gravities and hence lower air voids than really exist. It seems that the more
Wagg
(49)
Vagg + Vpore
For a given aggregate particle weight, the more absorptive the aggregate is, the
smaller the denominator (smaller Vagg), and therefore, the higher the Gse· The G •• will
be closer to the apparent specific gravity which is still higher. Looking at maximum
1 00 + pb(agg )
D =
pl Pz P3 pb ( agg )
+ + +
G1 Gz G3 Gb
if, say, aggregate 1 is absorptive, than G, will really be greater than it is estimated .
underestimated . The data in Table 19 bear this out. If D is too low, then from
V = D -d * 100
D
air voids will be underpredicted. Again, this is indicated by the data in Fig. 26.
80
Although the estimation method gave comparable results to the Rice method
for the Types C and IC mixes, it must be remembered that this is for only two sources
of aggregate. It is recommended that the Rice method be the preferred method for
testing, consideration should be given to heating the mixture in an oven for several
recommendations.
Results of Marshall method unit weight, stability, flow, air voids, VMA, and
Optimum asphalt contents were chosen primarily on the basis of meeting the
value. This was tempered by other criteria, such as voids filled, dust/asphalt content
(by weight of aggregate), Marshall stability, percent asphalt, and VMA. The six
mixtures and the MHTD criteria are shown in Table 20. In general, the mixes met all
criteria except for dust/asphalt ratio and VMA. All four Type C mixes were on the
borderline of dust/asphalt content acceptance; VMA values for both Type C and 1-C
Type C
Parameter MHTD UMR Gradation
Type 1-C
natural sand, % < 15 12.2 8 .8
hydrated lime, % 1.0 1.0 1.0
all other (-) #200 = MF ------ yes yes
air voids, % 3-5 3.0,3.3 4 .0,4.0
stability, lbs (Nl 2: 1500 3000 ,3500 2250,2800
VMA,% 2: 15 11.2,11.E 12.1,12 .2
voids filled, % 60-80 72.5,72 68,67 .5
dust/asphalt, by wt. of aggregate 0.6-1.2 1.2, 1.2 0 .9,0.9
flow, 0.01 in (mm) ------ 8,8 8,8
design asphalt content, by wt. of mix ------ 4.0,4.0 4 .0,4.1
82
Table 20 cont'd .
Bituminous Base
fine aggregate, % =:; 30 12.4 6 .5
fine aggregate, % pass 3/8 in. (cm) 100 100 100
fine aggregate, % pass# 200 s 6 0 0
design asphalt content, %, by wt . of mix 3-6 3.9,3.9 3.5,3.5
air voids, % ------ 4.0,4.0 4.0,3.8
stability, lbs (N) ------ 3800,3850 3400,3500
VMA,% ------ 12.1,12.4 10.1,10.0
voids filled, % ------ 62,67 60,60
dust/asphalt, by wt. of aggregate ------ 2.2,2.0 1.4, 1.3
flow, 0.01 in (mm) ------ 6,7 8,8
As mentioned previously , these mixtures may not have been approved for field
use due to the low VMA, and the asphalt contents may be somewhat low (in an effort
to satisfy air voids criteria). It was decided to not adjust the mixture gradations or
aggregate source (particle shape) ratios because two of the major criteria of the study
were to determine the effect of gradation and particle shape on resilient modulus.
Also, after discussions with MHTD personnel it was felt that the voids would have
been greater had the compression method of compaction been used instead of the 75
blow Marshall compaction effort. In addition, the specimens were compacted with a
manual hammer. Studies at NCAT have shown that this tends to render higher
densities than what is usually achieved with flat-faced mechanical hammers. MHTD
35°F lower than for Marshall specimens. This may also contribute to a higher void
83
content. The gradations were actual approved gradations, but the MHTD mix designs
compaction .
Note that, besides the above 24 mixtures that were prepared at the design
asphalt contents, another 24 mixtures were prepared which had an asphalt content
Results
The results of the indirect tensile strength testing are shown in Table 21. The
values shown are averages of two specimens per mix, in accordance with the
recommendations given in NCHRP 332 (62) . The mixtures reflected the following
The tests were performed in order to determine the seating and testing loads
for the resilient moduli tests, which are based on certain percentages of indirect
Average 102
Average 103
Average 117
2) a decrease in asphalt content from 0.5% above optimum down to Marshall optimum
the cases,
The data also indicate that bituminous bases had greater tensile strengths than Type IC
mixes by about 14% on the average, which in turn had about equal strengths to the Type
C mixes. A statistical analysis showed that the difference in mean between the bituminous
base mixes and either of the C or IC mixes was significant at the 0.05 level. It appears that
the greater amount of fines and the more well-graded nature of the bituminous bases
These results are in general agreement with trends published in the literature.
Specifically, Hadley et Q.[. (41,42) and Abkemeier (63) report increasing IDT with increasing
asphalt viscosity. In regard to asphalt content, it must be remembered that optimum asphalt
contents tend to be lower for IDT than for Marshall stability (64), thus it is not surprising
that, for a given gradation, lowering the asphalt content resulted in higher IDT values.
Hadley et Q.[ . (41) indicated that an increase in gradation fineness with an accompanying
increase in asphalt content led to higher IDT. In the present study, this was true for the
87
bituminous base mixes, but for the C and IC mixes, the IDT increased with increasing
gradation fineness with constant asphalt contents. In a review of the literature, Abkemeier
(63) noted that particle shape does not seem to have a great effect upon IDT (which he
verified in his study), although at least two studies (41,42) indicate an increase in IDT with
increasing angularity. In the present study, the opposite was true in most cases. However,
the differences in particle shape were not great. The literature (42,48, 50) indicates that the
relationship of change in IDT due to change in voids is not conclusive - as was shown in the
present study . In general, interactions between variables makes it difficult to make sweeping
Multiple Regression
Many linear multiple regression equations were developed and analyzed to estimate
the indirect tensile strength from certain mixture data by use of the computer program
SYSTAT (65). Multiple regression equations were fit to the indirect tensile strength data .
Many combinations of variables were analyzed. The criteria for final selection included:
1. The highest adjusted squared multiple correlation (adj -R2 ) that met all the below
listed criteria. This statistic reflects the overall goodness of fit of the equation.
It describes how well the equation would predict a population of data, not the
sample data. Thus it usually is a little lower than the R2 value which is for the
sample data.
2. A !ow standard error of estimate compared to the mean resilient modulus (Y).
Data to develop the equations came from Tables 6, 12, 17, and Appendix F. The equation
of best fit had an R2 value of 0.854, an adjusted R2 = 0.846, and a standard estimate of
error (SEE) of 7. 1 51 .
angularity/surface texture
The relationship between the estimated and experimentally determined values of indirect
·-
.........
D'l
0..
.._..
160
..c::
+J
bD 0
s::(l)
1-,
140
+J
rn
(l)
.....
·-
rn
s:: 120
(l) 0
E-
+J
C)
(l)
·-s:: 100
1-,
'iJ
0 Q)
-
'iJ
(l)
OCID
oco
+J
80 co
·-
C)
'iJ
(l)
1-,
0...
60
60 BO 100 120 140 160
Observed Indirect Tensile Strength (psi)
RESILIENT MODULUS
Test Results
Table 22 presents the results of the resilient modulus testing . The values shown are
averages of four specimens per mix, in accordance with the recommendations given in
NCHRP 332 (62) . The tests were performed at 1 load cycle per second, with the load
applied for 0 . 1 sec in a haversine form, followed by a 0.9 sec resting period . Both horizontal
and vertical deformations were measured and Poisson's ratio calculated. The most current
SHRP protocol at the time of testing (38) specified that if the calculated Poisson's rat io
exceeds 0.50, a value of 0.50 should be assumed for resilient modulus calculations . This
procedure was followed in this study . A previous draft of the SHRP protocol had
recommended assuming values of Poisson's ratio of 0.20 at 41°F (5°C), 0 .35 at 77°F
(25°C) and 0.50 at 104°F (40°C). In the present study, the average Poisson ' s ratio at 41 °F
was 0.30, at 77°F it was 0.49, and at 104°F it was over 0.50 but was held at 0.50 as per
the SHRP protocol. Other investigators (42,66-71 l have reported values of Poisson's ratios
greater that 0.5, especially at higher testing temperatures, with the majority of values
between 0.1 and 0. 7. All of this is not surprising considering that , as asphalt becomes
Bit. Base
Average 1149825 0.33 398379 0.495 113163 0.50
Std. Dev. 253772 0.095 84759 0.007 30391 0.0025
Type C
Average 1117504 0.29 391113 0.49 117629 0.50
Std. Dev. 193551 0.070 92047 0.020 42016 0.010
92
Type IC
Average 1075032 o. 28 414073 0.48 110973 0.49
Std. Dev . 212402 0.106 109000 0.045 40110 0.042
I
relationship of each mixture, the moduli at 68°F were calculated. The moduli at 68°F
are necessary for entering the predictive nomographs in the 1986 AASHTO Design
Guide for choosing layer coefficients. At this temperature, the 1-C mixes were slightly
stiffer than the bituminous base mixes and the Type C mixes were the least stiff. A
regression equation was developed between resilient modulus at 77°F and indirect
tension at 77°F . The R2 coefficient was 0.536. This relationship is shown in Fig . 28.
temperature, using indirect tensile testing conditions as was done in the present
study. As can be seen, none of the UMR mixes were excessively stiff, although some
93
-
IQ
0
.....
7.0
><
·-
-
Cll
P. 6.0
~
0
r--
r--
@ 5.0
Cll
::,
......
::,
"O
0
~
4.0
~
~
.....
V
......
.....
Cll 3.0
V
~
, 02
I
I I
I I
~
---- ..............
"-... ........,,
Mo 1ulms
i
I
00 Hialh
- 1
I
'(·...... ___
"'. , 1
,..
. . '~ .... I I I
-. '-
""-
. . ' ., '
I
i
. ". r-,... ."
I
I ~
·. 1!-·,, 1'....,.
- "-
, I
I I
·~
.' I
I I
'~
~ I 'r, "'
Moc Uh.ljS T DO
I
uOW
II "f'k.!
i •
'
i
I
I
.....
' i
I I
I I
I
I
I •
. I
I I
I I
0 20 40 60 80 100 120
Temperature (°F)
of the mixes were insufficiently stiff at one or more temperatures. Analysis of the mix
characteristics indicated that at 104°F, the large majority of the deficient mixes were
the purposely overasphalted ones, and most were of the low-fines(coarse) gradation
and contained the softer grade of asphalt (AC-10). Particle shape was not a
significant factor.
AASHO Road Test asphaltic material reportedly had a modulus of 450,000 psi
at 68°F. It is interesting to note where the AASHO Road Test material plots on the
chart-barely stiff enough at 68°F. Yet, there is a statement in the Guide that
cautions against mixes stiffer than 450,000; mixes exceeding this value are said to
be prone to thermal and fatigue cracking. Perhaps one way to reconcile these two
here. Three types of elastic modulus have been presented in the literature to any
great extent over the last 20 years: dynamic modulus (IE* l ), resilient modulus -
compression (MR,comp), and resilient modulus - indirect tension (MR,idt). The AASHTO
Guide a 1 -chart comes from NCHRP 128 (5). It is not stated upon which modulus the
chart is based, but the references that are given deal with l E* J. However, the 1986
Guide states that either MR,comp or MR,idt should be used to enter the chart. l E * I is
not mentioned. AAMAS recommends using MR , idt' which the industry seems to be
doing. To further complicate the issue, MR itself can be calculated in two ways: by
use of total cyclic deflection in the denominator (MR = a/E) or instantaneous cyclic
deformation (a smaller value). Which should one use? The choice may give
There is very little in the literature by way of comparisons between the different
moduli. First, Von Quintus et fil. (72) found MR ,idt to be significantly higher than
MR,comp at 104 and 77°F, but equal at 40°F . The data appeared to be in terms of
total MR. Khosla and Ohmer (73) found essentially no difference between the two MR
vs IE* I, only two studies were found: Bonaquist et al. (74) indicated that IE* I was
greater than total MR. Kallas (75) found MR to be higher than IE* I at low
temperatures, lower at 104 ° , and the same at 68 ° F (IE* I was measured at 4 hz).
Von Quintus et fil. (72) estimated IE* I via an earlier version of the Witczak equation
and found that MR , idt was usually higher than IE* 1- In the present study, total MR,idt
greater than l E* I -
From all this, it would seem that one should enter the AASHTO Guide a 1 - chart
with MR,idt· If one uses total MR,idt' it should be recognized that the resulting a 1 may
be somewhat high, which would explain the low position of the Road Test mix in Fig.
29. And, if one uses instantaneous MR,idt• the resulting a 1 would be even more likely
Effect of Variables. As previously mentioned, resilient modulus results for each mix
reported in Table 15. The best correlation was with A. Two other methods, which
did not lend themselves to direct correlation, were evaluated in the multiple regression
97
-....l7l
0..
20
Data for 41. 77 , 104 °F
-
IO
0
0
0
>< , 5
.....,.
0
0
,:,
I'll
-
"C
::1
::l
0
~
o o?o
'€,0
88
0
0
0 0
0 o-V
~ 10 0
0 Oo
,.J
'-'oo
r:::: 0
........
Q.)
...... 0
(J" Oo
0
0
l7l
Q.)
~
"C
5
Q.)
<9
> <~/6
s... 0
Q.)
l7l
.0
0
0
0 5 10 15 20 25
• 5
Estimated Dynamic Modulus, I El (x 10 psi)
mixture (32-34). These are: temperature, asphalt viscosity at 70°F, effective asphalt
content, some measure of void content, gradation, and possibly particle shape and
texture.
1. lower temperature ,
sieves), and
4. avoidance of over-asphalting .
Loss of modulus with increasing temperature has been well documented in the
literature (37). Likewise, a loss in modulus with a decrease in asphalt viscosity is also
supported by other researchers (76, 77). Little has been reported in the literature
accumulative retained on the #4 sieve, all of which result in a finer, more densely
graded gradation.
Coarse and fine aggregate fraction particle shape, as shown in Figs. 36 and 37,
did not seem to exhibit a significant trend in their effect on resilient modulus.
-....
,0
0
6 .50
99
><
6.00
rz..
0
co
co 5 .50
0
5.00
...•
Bi tuminous Base
....
-
....
ll'l
QI
4.50
a
Type C
Type IC
0::
4 . 00
AC 10 AC 20
AC Grade
Fig. 31. Effect of Asphalt Cement Viscosity on Resilient Modulus .
- 7.00 - - - - - - . . . - - - - - - - - - - - -
-
,0
0
-rz.
0
co
6.00
0
co
_.
Ill
D'l
- ::,
::,
't:I
o 5.00 • Bituminous Base - coarse
::Iii
_. o Bituminous Base - fine
s::: • Type C - coarse
-- QI
ll'l
c.
•
Type C - fine
Type IC - coarse
cu
0:: o Type IC - fine
4 .00
3 4 5 6
A
Fig. 32. Effect of Gradation on Resilient Modulus.
IO -0
.....
6 . 00
•
A
81tuminous Base
Typ e C
100
c T\·pe IC
><
·-
-~
0
Ill
C.
cc
co
..,
a:,
5 .50
Ill
-==
"O
0
::s
..,
i::
·-·--
cu
ll'l
cu
~
5 .00
p (P optimum + O.S%)
optimum
-
IO
0
.....
7.00
><
-
~
0
cc
6.00
co
•
5.00 •
·o Bil Ba s e - Fine
Ty pe IC - Fine
-·-
0
c:, Type C - Fine
•• Bil Base - Coarse
Type IC - Coarse
• Type C - Coarse
4.00
0 10
% Passini #200 Sieve
Fig . 34. Effect of % Passing the #200 Sieve on Resilient Modulus.
7
O Bil Base - fine 101
D Type IC - fine
>< c, Type C - fine
• Bil Ba se - Coa rse
• Type IC - Coarse
• Type C - Coarse
tz... D
0 6
CD
co
5 •
-·-
4 ..__....._ _.....________..._________..
30 35 40 45 50 55 60 65 70
% Accumu l ative Reta ined on the #4 Si eve
Fig . 35 . Effect of Percent Accumulat ive Re t ained on t he # 4 Si e v e
on Resilient Modulus .
6.00 - - - - - - - . - - - . . - - - - - . - - - - - -...
--
IO
0 D
0 B1lummous Base - DR7
~
0
co
co
5 .50
0
5.25
....
s::Cl.)
-
5 . 00
8 10 12 14
Particle Index
Fig . 36. Effect of Index of Particle Shape of Coarse Aggregate
on Resilient Modulus.
102
- 6.00
-
IC
0 t::. Ty pe C - DR5
• Typ e C - DR4 0
- l1l
~
e
o
•
Type 88 -
Type IC -
Type IC -
DR6
DR5
DR4
•
~ 0 Type 88 - DR?
0
cc
cc
5.50
0
~ •
C:
........
-
G)
l1l
G)
~
5.00
38 39 40 41 42 43 44
Voids in Uncompacted Mxture (U), %
Adedilmila and Kennedy (67) indicated that there was no significant difference in
resilient moduli of gravel and crushed limestone mixes, although the gravel mixes
In this study, compactive effort for all specimens was held constant.
Therefore, any differences in VMA and air voids were a function of differences in
particle shape, gradation, and whether or not the mixture was purposely over
asphalted. Changes in void content were not, then, a function of level of compaction.
Being dependent variables, it was difficult to draw conclusions about their effect on
MR . For instance, air void content was lower in the overasphalted mixes, and so was
MR, but what really caused the MR loss? As Abkemeier (63) has summarized:
Generally, opinion and testing results have been mixed with respect to the effect of air
voids on asphalt mixture properties. Most test results from the static and dynamic indirect
tensile tests (31, 76, 78) indicate that with decreasing air void content of the asphalt mixture,
Other investigators (42,67, 79) have had less conclusive results. These investigators
have concluded that air voids are not directly related to fatigue life , tensile strength, and
resilient moduli and that possibly an optimum air void content exists for these properties .
Careful evaluation of mix factors would be required when evaluating the effect of air void
content, since air voids are highly dependent upon other factors. Such factors include
compactive effort, asphalt content, a change from open to dense gradation, or any combination
of these.
gradation changes were significant, both in the paired t-test and in the regression
analysis.
104
Statistical Analys is
An analysis of the data was undertaken using the statistical software package
SYSTAT. First, the means of the five independent variables were checked via Tukey's
HSD test and paired t-tests to compare the means in order to determine which
variables were statistically different from the others . Next, multiple regression was
used to provide a model for the estimation of resilient modulus from the index data .
Tests of Significance. The first question to be answered was, for each of the five
independent variables, was the difference in magnitude from high to low a significant
contr ibutor to changes in MR? For example, for the whole data set , did switching
from AC10 grade to AC20 significantly affect MR? To test this hypothesis, the MR
data were sorted into two groups (AC10 and AC20) and the means of both groups
were calculated . Then the two means were tested via Tukey's HSD to see if they
were significantly different at the 0.05 level. The same sort of analysis was done on
the other independent variables: two levels of asphalt content (optimum and
aggregate type (see Table 17). Two other variables that have been shown to be
important are percent passing the #200 sieve and percent accumulative retained on
-
the #4 sieve . These were, in effect, tested when A was tested .
The results are shown in Table 23. The analysis was broken into four parts .
First, as the top portion of Table 23 indicates, results from all 48 mixtures and all
three temperatures were lumped into one data set and examined. The results show
105
sign if icant at the 0.05 level. Of course, MR decreased with increas ing temperature .
The second portion of Table 23 shows the results when temperature was
removed as a variable. The trends are the same as in the upper portion of the table,
that is, greater stiffness came from mixtures that were comprised of higher viscosity
asphalt, finer gradation, asphalt contents near optimum (as opposed to being
and aggregate gradation (fine or coarse) were significant at the 0 .05 level , although
-
gradation (six levels of A) was significant at the 0 .089 level.
The bottom portion of Table 23 breaks the analys is down in regard to mixture
type. For the bituminous base and IC mixes, viscosity and gradation (only 2 levels are
possible) were significant at the 0.06 level; for the C mixes, only viscosity was
significant .
106
MR (psi)
Agg. blend, 886 and IC4 582 ,394 521 ,888 60,506 no
Gradation, fine to coarse 638 ,750 500,875 137,875 no (yes at 0.089 level)
Agg . blend, 886 and B87 576 ,375 529 .625 46,750 no
68°F, Type C
Viscosity, AC20 and AC10 613 ,250 458 ,750 154,500 yes
68°F, Type IC
Viscosity, AC20 and AC10 618 ,875 495 ,000 123,875 yes
Gradation , fine to coarse 605 ,875 508,000 97,875 no (yes at 0.061 level)
To examine the effect of coarse aggregate particle shape more closely, paired
t-tests were performed for each mix type at each test temperature. In this way, all
variables except particle shape were kept constant. For an example, refer to Table
24. Here, for C mixtures at 41 °F, each pa\r of MR values were for constant asphalt
grade, asphalt content, temperature, and gradation (nearly) . The mean of the
DR 4 DR 5
The coarse aggregates that were compared were 6 vs 7, and 4 vs 5 . Almost none
of the nine data sets that were examined showed any significant changes in MR due
A similar study was undertaken for fine aggregate. Here, IC mixes were
compared to C mixes, holding all other variables constant. In essence, this was a test
sand made any difference in MR. The results indicated that in most cases there was
no significant difference .
Finally, a paired t -test was performed on VMA results . Each pair of MR values
had a common aggregate blend, asphalt viscosity , asphalt content, and testing
temperature (68 °F). The only var iable different between the two MR values was
gradation, which caused a difference in VMA. The results indicated that VMA did
Multiple Regression . Multiple regression equations were fit to the resilient modulus
data. Many combinations of variables were analyzed . The criteria for final selection
an absolute error = 18 .0% . This is comparable to the results of Akhter and Witczak
(R 2 = 0.931 and 0 .934, SEE = 0.122 and 0 . 125). Absolute error of Akhter and
Witczak's equation was not available, but an earlier model based on the same data
Predicted MR-Measured MR
Absolute Error = *100 (51)
Measured MR
Data to develop the equations came from Tables 5, 10, 15 and Appendix F.
asphalt content. % .
variables were, in descending order: temperature, AR 4 , viscosity, P.ff, P200 '* (P 0 pteffv -
P.ffvl, and P8 ;,· Thus, by knowing relatively easily obtainable mix design data, the
The variables included in the model are temperature, P200 and AR 4 (measures
by volume, and the difference between the mix design effective asphalt content
(optimum) and the amount actually in the mix. These were identified through use of
the multiple regression analysis as being the most significant variables in their effect
upon resilient modulus. Many other variables were tried, but did not contribute
when two or more variables are highly correlated. When this happens, the regression
would enhance the model. In fact, inclusion of U (fine aggregate particle shape index)
and IP (coarse aggregate particle shape) did improve the model, but a t-statistic test
indicated that they were not significant contributors . This may be because of the
important. As it turned out, air voids had a slight edge over VMA.
Other variables that were tried unsuccessfully were percent absorbed asphalt
content, voids filled, dust ratio, aggregate gradation curve slopes (M 4 . 200 , M, ,2 . 4 , M3 14 _
-
112 ), percent retained on each sieve, and SF/SSF. Both Hudson's A and the gradation
curve slopes enhanced equation accuracy, but using individual sieves proved to be
somewhat better .
the relative importance of prediction cannot be gaged from the magnitude of the
coefficient because this is influenced by the scale of the variable. Also, the direction
of the association between the dependent variable (resilient modulus) and any
must be emphasized that regression equations are only valid for the ranges of data
111
temperature range where errors (scatter of data) is much higher . This is typical for
any kind of modulus testing. And, if one is tempted to play "what if" by substitution
of hypothetical mixture data into the equation , it must be remembered that certain
variables tend to be dependent upon each other in the real world (e.g., - #200 sieve
manner .
It can be seen that Eq. 52 is similar to Akhter and Witczak's Eq. 26 which
contains the variables temperature, (} 70 , P 200 , Peffv• Pa;,, AR 4 , and (Peffv - P 0 p 1 e11 vl as well
multicollinearity problems, testing frequency (f) was not a variable in this study, and
modulus is shown in Fig . 38. The high R2 value demonstrates a strong correlation.
Applications
practical impact of the above trends is that higher layer coefficients can be obtained
by:
10.00
~
,.,
)'
/,.,
-
,E
fl
1 .00
.~ "
::,
"'O
0 ~ )
2 04 ~.©
+1
,.Jd ~ 0
....=
-....
Cl)
fl
0.10 r ~h ~
" -
b
l
Cl) ·,
~
"'O
.,, .,
Cl) ,/
+1 V
....CJ
"'O V
V
Cl)
1-4
0..
0.01
0.01 0.10 1.00 10.00
8
Observed Resilient Modulus (psi x 10 )
Excessively hard asphalts, highly angular aggregates, and low VMA values can lead
to such problems as thermal cracking, lack of workability, poor durability, and rutting.
Thus, caution should be used when interpreting the results of the above analysis.
Also, it must be kept in mind that the MR predictive equations are based on data that
a broader data base by adding data from the literature into the data base developed
experimentally in the present study. Based on the work done on the UMR data, it was
decided that, as a starting point, the equation ought to reflect the following variables:
Not all variables used in the previously developed UMR equation were readily available
in the industry-wide data base. For instance, it is not common to measure particle
shape/surface texture, so it was necessary to omit this variable from the model. The
industry-wide data base included 84 sets of data (72, 74, 76, 78,80). The combined
sets for the general model numbered 229. Many regression models were determined.
where:
114
T = mixture temperature, °F
of variables were, in descending order : temperature, Pa;,, viscosity, P20 0 , AR 314, and
P.ttv· Again, the model was based on the highest adjusted R2 compatible with the
= 0 .842, SEE = 0 . 164, and an absolute error of 32 .9%. The model is less accurate
than the model developed from laboratory data developed in the present study, which
had an absolute error = 18.0%. This is not surprising , because the general data base
equipment, and, most likely, test methodology. And, to put things in perspective, it
should be remembered that the Shell method for Sm determination is based on Van der
Poel's work, which was claimed to have an accuracy within a factor of ± 2; it would
The method of cross-validation was used to determine the variables which were
significant . The whole data set was split into two parts in a random fashion. A
regression model was developed for the first portion as per the above criteria. Once
115
the variables were chosen, the second data set was applied to the model. The SEE
of each set were compared, and were shown to be close. Finally, the data sets were
recombined. A final model was calculated, using the same variables as were
The ranges of variables that the industry-wide portion of the data represented
are shown in Table 25. The industry-wide portion of the data set is included in
Appendix G.
Variable Range
Temperature, OF 24 - 104
Viscosity at 70°F, poise x 10 6 0.17 - 9.5
Range of loading time was very small (0.1 to 0.15 sec) with most of the data
remove loading time as a variable. Thus, the data set was limited to a loading time
of 0.1 sec. The fact that the equation does not reflect loading time or frequency
116
effects is, although unavoidable, unfortunate because the repeated modulus of asphalt
mixtures is to a certain degree a function of loading time . However, the standard time
of 0.1 sec is conservative because this is 3 to 10 times longer than dwell times under
Not all data were present in the desired form, and had to be converted from the
existing form. Witczak (32) faced the same dilemma when he developed his model,
peffv = (55)
0.483
where:
Even though the general model is less accurate than the UMR model, the data
on which Eq. 53 is based contains variables that exhibit a wider range of values.
Thus, it would seem that the general equation would be more generally applicable.
Data from 236 MHTD mixtures approved in 1990 were substituted into Eq. 53
re,- 10.00
0
.....fll
- i:i.
1.00
~
A
-
.....
Q)
.....
fll
Q> 0.10
~
"C
Q) o UMR De.ta
~
MHTD
Mix Min. Max . Mean
Type C 360,415 439,198 403 ,819
Type 1-C 370,998 425,394 404,973
Bit . Base 377,642 581,453 460,000
In regard to each variable's regression coefficients, it should be pointed out that the
relat ive importance of prediction cannot be gaged from the magnitude of the
coefficient, because this is influenced by the scale of the variable . Also, the direction
of the association between the dependent variable (res ilient modulus) and any
And it must be emphasized that regression equations are only valid for the ranges of
data studied. Extrapolation shou ld be done with caution , especially in the lower
temperature range where errors (scatter of data) are much higher . This is typical for
any kind of asphalt modulus testing . And, if one is tempted to play "what if" by
subst itution of hypothetical mixture data into the equation , it must be remembered
that certain variables tend to be dependent upon each other in the real world ( ~,
#200 sieve material and VMA), so any substitution of values must be done in a
realist ic manner.
PAVEMENT TEMPERATURE
Values of temperature are necessary for the estimation of dynamic modulus (Akhter
119
and Witczak), mixture stiffness (Shell), and resilient modulus (as developed in the
present study). Because layer coefficients are applied to field conditions, it becomes
by use of Witczak's method (24) as prev iously discussed. Use of Witczak's equation
(Eq. 16) requires input of mean air temperature and one representative depth in a
given layer.
order to calculate a mean monthly air temperature of 55.06 °F. This data is tabulated
data supplied by the MHTD were analyzed for mean thicknesses of asphalt surface,
binder, and base layers . Because surface and binder layers were considered as one
layer at the AASHO Road Test in computation of layer coefficients, the MHTD surface
and binder layers also were combined for temperature analysis . The Missouri data
indicated that for full-depth and asphalt-over-unbound base structures, the mean
surface-plus-binder thickness was 3.26 in and the mean bituminous base thickness
was 5.07 in. Typically, the temperature at a depth of one third the layer thickness
is taken as representative of the mean temperature of the entire layer. This can be
shown to be close to the mean pavement temperature by use of the mean value
theorem (25). Use of these points in the mean pavement structure in Eq. 16 yielded
representative MHTD pavement temperatures of 65.2 °F (18.4 °C) and 63.4 °F (17.4
120
In a similar manner, from AASHO Road Test data, pavement temperatures were
predicted for the surface/binder layer and the bituminous base as shown in Table 27 .
There is some question about the average pavement temperature at the Road
Test. In NCHRP 128 it is stated that this temperature was 67. 5 °F and in the 1986
is standardized at 68 °F. It is stated in the Road Test Specia l Report 5 that pavement
temperatures were recorded. Carree and White ( 12) state that in a search of the
special reports and computer data from the Road test, no such information could be
found . Personal commun ication with White revealed that some records have been lost
in a fire . A search by the authors has also been fruitless. NCHRP 128 states that the
standardized modulus of the surface layer was 450,000 psi. In NCHRP 291 , data
from Road Test asphalt mixture modulus testing are presented . An analysis of these
data indicates that the average pavement temperature at the Road Test would have
been 63°F in order to give a dynamic modulus of 450,000 psi. In the classic study
of asphalt pavement temperatures at College Park, Maryland, Kallas (81) found that
the average pavement temperature at all depths was about 63 to 64°F for an average
pavement temperatures in a similar climate, Croney (82) found that average pavement
temperatures were about 7°F above the average air temperature of 49°F . According
to data in Appendix C of Special Report 5, the average air temperature during the
Road Test was about 50°F. Using this temperature, the Witczak equation (Eq. 16)
121
average air temperature. Thus, it would seem reasonable that average pavement
60 °, at the Road Test. This is significantly lower than the 68 ° F stated in NCH RP
128, and is close to the 58.1 and 59.1 °F predicted by the Witczak equation. It also
then seems reasonable to use the Witzcak equation to predict Missouri pavement
temperatures of both MHTD and Road Test pavements . The use of the AASHTO layer
coefficient chart, which calls for modul i input at 68°F, is thus brought into question.
Use of this information will be discussed in the layer coefficient portion of the study .
Duration of load pulse (or frequency of application) has been shown to have an
effect upon modulus or stiffness of asphalt mixtures (15,31 ). Either pulse duration
or frequency is used as an input into the Shell method of stiffness estimation, and
At any given point in an asphalt layer, both pulse duration and frequency are
dependent upon vehicle speed and depth of interest. From data supplied by the
Planning Division of MHTD, the mean vehicle speed of all highways in the Missouri
state system is 56.3 mph. Barksdale (_53) has developed a relationship between
vehicle speed, depth, and pulse duration, as shown in Fig . 40. Using the mid-depth
of both the mean surface and bituminous layers (as determined above), and a vehicle
speed of 56.3 mph, load pulse durations of 0.020 and 0 .032 sec were determined for
122
,o 0
j - I --- -f-·- -----
- - - ,-
5111,jCl( OR OUAl WH(ll lOAOINC - --·- -----
-
~ ! ~ ~ £- V
-!lO:•~-t-- ~ --~~-~--=:,...~~---___. - ~ . ~ ~ --~
-- _L---
,,______ --
ID
~ - - -~ - ~ _-·_-__
·- ~ - t
------ ------i-----i-----t-----·+-----+------i
0.1
v -- - -
-~ -
~i:::-=-- -
- "·"·'"
T.,s;;
rAINCIPAl S1NU$0l0Al
-
ruLSE TIM(
-
rRINC1'Al T~IANGUlAR
,uu£ TIM£
o a,
12 u 20 2, 21
To convert load pulse duration (t) to load frequency (f), the relationship from
1
f = ( 56)
27t t
Thus, at the midpoints of the surface and base layers, the mean load frequency
In a similar manner, from AASHO Road Test data, times of loading and
frequencies were predicted for the surface/binder layer and bituminous base. The
Layer Veh. D t f Tp
Speed
mph in sec cps OF (OC)
MHTD
surface/binder 56.3 3.26 0.020 8 65.2(18.4)
base 56.3 5.07 0 .032 5 63 .4(17.4)
AASHO
surface/binder 35 4.0 0.035 4.5 58.5(14.7)
base 35 8.83 0.068 2 .3 57.1(14 .0)
D = layer thickness
These values were substituted into Eqs. 6,26, 52, and 53 to estimate mixture
stiffness and dynamic modulus, respectively. The results are discussed in the
following sections .
124
Mix stiffness via the Shell method for all 48 mixes used in this study were
estimated by use of Eq. 6. For each mixture, values of asphalt content , aggregate
content, asphalt cement penetration, time of loading (0.1 sec) and test temperature
(41, 77,104 °F) were substituted into Eqs. 3,4, 13, 15, and 16 in order to utilize Eq .6.
The resulting values of Sm were correlated with resilient modulus test data . The
Finally, using AASHO Road Test mixture data, Sm values for the surface/binder
and bituminous base layers were calculated and are shown in Table 28, using Road
Table 28 . Estimation of AASHO Road Test Mixture Stiffness Under Road Test
Conditions .
Layer Pen vb Va t T Sm
% % sec oc ps i
surface 91 12.67 83.74 0.032 15.0 687,320
binder 91 10.64 84.56 0.038 14.5 778,168
surface/binder - - - - - 740,194
base 91 11.36 82.40 0.068 14.0 565,276
Note : AASHO Road Test average surface = 1.67 in, binder = 2.32 in
Note that the Shell method considers six variables: three that address the
viscoelastic properties of aged asphalt cement, and three that address the effects of
mix properties. The gradation and particle shape/texture of the aggregate are ignored.
20
-....
rn
0.
Data for 41.77 . 104 °F
IC 0
0
~ 15
--
0
0
>< 0
rn 0
::J
::J
"C 0
~
~ 10 Oo
0
0 0
+J
c::
........rn
-
Q,)
0
Q,)
~ 5
"C
Q,)
>
i..,
Q,)
Ill
..c
0 0
0 10 20
5
Mixture Stiffness, s m (psi X 10 )
lower temperature
The resilient modulus testing program indicated that, on the average, the 1-C
mixes were slightly stiffer than the bituminous base mixes, and both were stiffer than
the Type C mixes, which was the result of gradation and sand shape/texture
differences. Because these variables are not considered in the Shell method, and
because temperature, time, and asphalt grade were held constant, the relative
rankings by the Shell method were the result of only relative material proportion
differences. Because of the slightly higher aggregate and lower asphalt proportions,
the predicted Shell stiffness of the C mixes averaged higher than the bituminous base
mixes, followed by the 1-C mixes. Table 29 shows the UMR, MHTD, and AASHO Sm
Table 29. Comparison of Shell Mixture Stiffness for AASHO, UMR , and MHTD Mixes.
Surface 315,980 - -
Type C - 576,671 488 , 629
Type 1-C - 523 ,054 459,597
Binder 374,836 - -
in this study, MHTD mix designs approved in 1990 were estimated for mixture
stiffness. Time of loading and pavement temperature were taken as 0.020 sec and
17.4°C for Type C and 1-C mixes, and 0.032 sec and 18.4°C for bituminous base
mixtures, respectively. The mean Sm values for Types C, 1-C, and bituminous bases
were 823307 , 771293, and 852051 psi , respectively . Note that the values in Table
The dynamic modulus of each of the 48 mixtures used in this study was
estimated by use of the Akhter and Witczak equation (Eq. 25). For each mixture,
values of air content , accumulative percent retained on the 3/4 in sieve, asphalt
128
viscosity at 70 °F, test temperature (41, 77, 104 °F), load frequency (1 .6 H2 ),
effective asphalt content by volume , mix design optimum effective asphalt content
the #200 sieve, and percent absorbed asphalt by weight of mix were substituted into
the equation .
The resulting values of dynamic modulus were correlated with resilient modulus
In a similar manner, using AASHO Road Test data, dynamic modulus values for
the surface/binder and bituminous base layers were calculated by use of Eq . 26 and
Table 30 . Estimation of AASHO Road Test Dynamic Modulus Under Road Test Conditions.
It was reported in NCHRP 291 that the Asphalt Institute performed dynamic
modulus tests on actual Road Test materials. An analysis of the data was performed
and dynamic moduli were determined for both the surface and binder mixtures, which
129
were 556,863 and 524,407 psi, respectively. A comparison of these laboratory test
values to the estimated values in Table 30 results in absolute relative errors of 19.1
and 19. 7%, which is comparable to Witczak's stated 20.6% absolute relative error.
Thus it appears that the method of estimation of the mixture's dynamic moduli is
acceptable.
Note that the Witczak equations not only consider the viscoelastic nature of the
asphalt and mix proportions, but also aggregate gradation. Thus the UMR 1-C and
bituminous base mixes are more correctly rated stiffer than the C mixes - an
improvement over the Shell method. Dust ratio (percent by weight passing #200
lower temperature
A comparison is made between AASHO, UMR, and MHTD mixes in Table 31,
Table 31. Comparison of Dynamic Modulus for AASHO, UMR and MHTD Mixtures.
The UMR bituminous base mixes are rated stiffer than the 1-C (and C) mixes due
to greater ( +) 3/4 in sieve and minus #200 sieve size materials. The effect of
gradation is still only partially accounted for if one considers the displacement of the
gradation curve from the 0.45 power maximum density curve as an important criteria.
The MHTD 1990 mixes were ranked from high to low: bituminous base, Type
C, and then Type IC. Bituminous bases were predicted stiffer because of lower air
contents, greater amount of ( +) 3/4 in material, lower effective asphalt contents, and
greater amount of minus #200 sieve material. Type C mixes were rated stiffer than
In comparing UMR mixes to MHTD mixes, UMR mixes were stiffer for all three
mixture types due to lower asphalt contents, lower air void contents, and, in the case
131
Looking at the AASHO Road Test mixes, in descending order of stiffness, the
rankings were bituminous base, binder, and then surface, although the spread of
values is not great. The difference in predicted values is primarily due to gradational
differences.
The AASHO surface and binder moduli are predicted to be greater than the
UMR (and MHTD) mixes primarily due to higher percent retained on the #4 sieve and
lower(-) #200 sieve material (and greater air void content in the case of MHTD data).
However, the AASHO bituminous base was less stiff than both the UMR and MHTD
averages because of a greater percent ( +) 3/4 in. material, lower air voids, greater (-)
#200 sieve material, and lower effective asphalt contents. These relationships of
UMR and MHTD to AASHO modulus values become important when estimating layer
In a similar manner to UMR mixture dynamic modulus, data from MHTD mix
designs approved in 1990 were used for estimation of each design's dynamic modulus
by use of Eq. 26. Frequency and pavement temperature were taken as 4.5 hz and
17.4°C for surface mixtures Type C and 1-C, and 2.34 hz and 18.4°C for bituminous
base mixtures, respectively. The mean dynamic modulus values for Type C,IC, and
bituminous base mixtures were 518,864; 509,020; and 600,821 psi, respectively.
Five different methods were used for determination of layer coefficients. These
132
AASHTO NOMOGRAPHS
Use of the layer coefficient nomographs (Figs. 42 and 43) available in the 1986
68°F . It is assumed that the Road Test conditions of loading time or frequency are
also in effect. The modulus values at 68°F for the 48 mixtures used in this study
nomographs, equations were developed to represent the curves for a 1 (asphalt mixture
a 1 =0.391071logE-1.77224 . • • • . . (57)
Each UMR resilient modulus at 68°F and at the test loading time of 0 . 1 sec for
the Types C, 1-C, and bituminous base were substituted into Eqs. 57 and 58 to obtain
layer coefficients. The results are shown in Table 32 in the column marked "AASHTO
Chart-UMR".
Secondly, the UMR-study mix data (48 mixes) were used in Eq . 53 (the General
Model) to estimate MR (68°F and 0.1 sec) . Then, these MR values were used in the
AASHTO nomograph to obtain layer coefficients. The results are also shown in Table
32.
133
05 I I I I I
-
~- - .... -
04 ./""
.E ~
-
C1)
-
/
V)
:...
:i
"' u
0
C
4) 4)
u
u 0.3
"':,
/
C1)
-
0 U'l
u
ai
C1)
C1)
-
> u
"' C0
...J
0 .2
"':::, u=
u:::, .i::."'
~
U'l
C.
"'
~
- -
0 .1
..... -
0.0 I I t I I
1800 4 0
0 JO 1600
1400 J .O
2.5
1200
1000
2.0
800
0.20
1.5
c 600 ·.;;
Q) ..ci Q.
u
::: 400 "'0
Q)
>-
0
u :0
a, "'
IJl :,
:, "C
0
u 200 ro
.r.
1.0
~
-
(f)
"'
<"O
0 . 10 -~---------
,......,
-
IC
0
1.0
><
......
Cl) 0.8
0..
'-"
Cl)
-
"O
~
~ 0.6
0
~
...>
c:: 0.4
-
CV
......
......
Cl)
CV
0:: 0.2
0.0 ............_._.............._._........................................................................................................
40 50 60 70 80 90 100 11 0
Temperature (°F)
EQUIVALENT STIFFNESS
Layer coefficients were also calculated by use of the Odemark equation for
stiffness ] 1 13
a = a
AASH
1 stiffness, AASHO
where layer stiffness in this study was represented by mixture stiffness (Sm), dynamic
Mixture Stiffness
M ixture stiffness (Sml values were determined as previously discussed for UMR
laboratory data and AASHO Road Test materials. Each mixture set of values were
substituted into Eq. 6. The temperature and loading time of Missouri conditions were
used for the UMR study dataset in the numerator, and AASHO Road Test conditions
were used for the AASHO Road Test mixes in the denominator :
S ] 1/3
a - a m
1. 2 - AASHO 1 , 2 [ Sm, AASHO
Dynamic Modulus
substituted into Eq . 6:
113
a
l. 2
=a
AASHO 1 '2
[
IE·IE·I ]
1AASHO
Resilient Modulus
Again, in a manner similar to the above, estimations of MR for UMR study data
(Missouri road conditions) and AASHO Road Test mixes (Road Test conditions) were
METHOD OF CHOICE
Based on a review of the literature, all of the above methods appeared to give
reasonable results with the possible exception of the Ode mark method using IE* I.
138
To determine the method of choice, the four methods which used estimated data
(IE* I, Sm, estimated MR) were correlated with the method which used real data
(laboratory MR) . R2 values were calculated for each correlation with the results from
laboratory data entered into the AASHTO nomograph . The results are as follows
shown in Table 33 .
Method R2
- Sm
-1;3 0.857
aAASHO
Sm,AASHO
- lE*i
-113 0.526
aAASHO
: E*: AASHO
Option One
For a given mix design, if laboratory MR data is available , layer coefficients should be
should be estimated from Eq. 53, and the results should be used in the AASHTO
In Table 26 are shown the MR values estimated from Eq. 53 for 1990 MHTD
approved mixes at 68°F . By use of the AASHO nomograph, the resulting layer
Option Two
The above option is recommended in the 1986 AASHTO Guide, which relies on
determination of a specific modulus at 68 °F. This ignores the fact that pavement
The method is easily implemented. The steps are shown in Table 35. The
temperature (See Fig. 45) and mix design data or MR data at three temperatures. As
indicated in Fig . 45, mean annual air temperatures range from 50 to 60°F in Missouri.
By use of Witczak's pavement temperature equation, this means that across the state,
pavement temperatures would be 59.8 to 72.2 and 58.9 to 70.8°F for the surface
and base courses, respectively. The 1990 MHTD mix designs were substituted into
the equations in Table 35 and the results are shown Fig. 46 . Note that the 1990
mixes straddle the lower boundary of acceptance in accordance with AAMAS criteria.
Step No . Action
1 Obtain mean annual air temperature (TA) from Fig . 45
~\
;. /
I
'~
-~
-........ "'-.
.......
~
.
I
" ~ ' ."' ...
~
I
~
I
I
'
I
'
'
i
! : ,:-,... '" "X
,"1, 1, 1 I
.J\..'
I I
'"\·'
'~ , '
, ~,
,_ ~
I I
I :' I
-
I
I I
0 20 40 60 80 100 120
Temperature, °F
Table 36. 1990 MHTD Mix Designs Resilient Moduli in Three Parts of Missouri .
Option Two gives somewhat lower coefficients than option One for the surface
mixes, but about the same for the bituminous base mixes. The reason for this is
found by looking at the MA predictive equation. First, the average air temperature in
Missouri is about 5 ° F warmer than that which was recorded at Ottowa during the
Road Test. This accounts for a great deal of the loss in MA when comparing MHTD
to Road Test material. Now, keeping temperature constant for both kinds of
materials, for average conditions of temperature and mix design, the Type C and IC
C and IC vs AASHO. Looking at Table 37, MHTD mix designs on the average have
higher air voids, lower P 200 , and higher asphalt contents, all of which diminish MR
according to the MR predictive equation. Asphalt viscosity and AR 314 are about the
same. Thus, one would expect the AASHO MR to be higher, even at the same
temperature.
I Material
I P.;,
I 'ho I P200 I AR314 I p effv I
AASHO* 4.3 1.49 5.0 2 .3 9.1
MHTD-C 5.3 1.49 3.5 0 10.0
MHTD-IC 6.0 1.51 4.3 0 9.6
AASHO 6.2 1.49 5.6 4.0 10.8
Base
MHTD-BB 4.8 1.51 7.7 10.6 9.1
*weighted average of surface plus binder courses
Bituminous Base vs AASHO. MHTD mix designs averaged lower air voids, higher P 200 ,
and lower effective asphalt contents, all of which would increase the predicted
moduli. MHTD did have more AR 314 than AASHO, but care must be taken when
interpreting the effect of changing variable magnitudes that are interactive with other
MHTD pavements and Road Test pavements, one might expect MHTD C and IC layer
coefficients to be somewhat lower, but bituminous bases about the same as AASHO
pavement layers.
145
SENSITIVITY ANALYSIS
As shown in Table 23, only asphalt viscosity and extremes in gradation were
significant in their effect on MR. An analysis was performed to ascertain the effect
Thickness of the surface layer was varied at 1.25 in and 4 in for both Type C and IC
mixes. For the bituminous base mixes, thickness was examined at 4, 8, and 14 in.
Three levels of viscosity or gradation change were looked at: most significant change
in the UMR data set, least significant change in the UMR data set, and average
change for all mixes of a certain type. For instance, for the Type C mixtures, the
mixture as gradation changed from the fine side (MR= 762,000 psi) to the coarse side
(MR= 562,000). The least change occured where the C4C20-4.25 mix changed from
the fine side to the coarse side (essentially no change in measured MR). The
"average" condition was simply a comparison of the average of the MR values for all
the Type C fine gradation mixtures to the average of the MR values for all the coarse
gradations at 68°F (as shown in Table 23). Gradation or viscosity was changed from
high to low with a resulting change in MR. Then the resulting change in layer
required change in thickness was computed as needed to maintain the initial structural
146
number rendered by the initial assumption of layer thickness. The results are shown
in Table 38. All data are based on 68°F . For example, what is the average change
in required surface layer thickness for an initial thickness of 1.25 in using a Type C
mix when gradation is changed from the fine side to the coarse side? Looking at row
3 in Table 38, the MR for the fine side Type C is 567,625 psi. Moving to the coarse
side results in a reduction to 504,375 psi. Using the AASHTO nomograph, the
corresponding loss in a, is 0.02. The structural number (SN) provided by the fine side
mix is SN = a,D, or SN = 0.48 • 1.25 = 0.60. When the layer coefficient is reduced
from 0.48 to 0.46, the new required thickness D 2 = SN/a, = 0.60/0.46 = 1.30, or,
From Table 38, it appears that wide swings in gradation are not significant for
thin ( 1.25 in) layers, and are marginally significant for 4 in layers. For thicker layers,
initial 8 in required thickness, changes varied from 0.02 to about 1. 5 in. At 14 in, the
variation was from 0.04 to 2.66 in (on the average, the required extra thicknesses
were 0.61 and 1.06 in for the 8 and 14 in initial thicknesses, respectively).
Changes in viscosity grade rendered the following results. For the C and IC
mixes, thin surface layer (1.25 in) thickness changes varied from 0.03 to 0.31 (0.15
average) in, while 4 in layer thickness changes were 0.11 to 1.00 (0.48 average) in.
For the bituminous bases, the variance for a 4 in layer ranged from 0.08 to 0.94 (0.38
average) in. An 8 in layer could see a variance from 0.16 to 1.88 (0. 76 average) in,
147
GRADATION CHANGE
TYPE C
1.25 AVG 567625 504375 63250 0.48 0.46 0.02 0.60 1. 30 0.05
4.00 567625 504375 63250 0.48 0.46 0.02 1. 91 4.18 0 . 18
TYPE IC
1.25 WORST 725000 500000 225000 0.52 0.46 0.06 0.65 1. 42 0.17
4.00 725000 500000 225000 0.52 0.46 0.06 2.08 4.55 0.55
1. 25 LEAST no change 0 0
4.00 no change 0 0
BITUMINOUS BASE
4 WORST 790000 494000 296000 0.41 0.35 0.07 1.66 4.76 0.76
8 790000 494000 296000 0.41 0.35 0.07 3.31 9.52 1. 52
14 790000 494000 296000 0.41 0.35 0.07 5.80 16.66 2.66
VISCOSITY CHANGE
1.25 WORST 626000 364000 262000 0.38 0.31 0.08 0.48 1. 56 0.31
4.00 626000 364000 262000 0.38 0.31 0.08 1. 53 5.00 1.00
1.25 AVG 613250 458750 154500 0.38 0.34 0.04 0.47 1.40 0.15
4.00 613250 458750 154500 0.38 0.34 0.04 1. 51 4.48 0.48
1.25 LEAST 562000 525000 37000 0.37 0.36 0.01 0.46 1.28 0.03
4.00 562000 525000 37000 0.37 0.36 0.01 1.47 4.11 0.11
TYPE IC
1.25 WORST 725000 480000 245000 0.40 0.34 0.06 0.50 1. 46 0.21
4.00 725000 480000 245000 0.40 0.34 0.06 1. 61 4.68 0.68
1.25 AVG 613250 458750 154500 0.38 0.34 0.04 0.47 1. 40 0.15
4.00 613250 458750 154500 0.38 0.34 0.04 l. 51 4.48 0.48
1.25 LEAST 533000 475000 58000 0.36 0.34 0.02 0.45 l. 31 0.06
4.00 533000 475000 58000 0.36 0.34 0.02 1.44 4.19 0.19
BITUMINOUS BASE
4 AVG 617375 488625 128750 0.38 0.35 0.03 1.52 4.38 0.38
8 617375 488625 128750 0.38 0.35 0.03 3.04 8.76 0.76
14 617375 488625 128750 0.38 0.35 0.03 5.31 15.33 l. 33
4 LEAST 494000 470000 24000 0.35 0.34 0.01 1.39 4.08 0.08
8 494000 470000 24000 0.35 0.34 0.01 2.79 8.16 0.16
14 494000 470000 24000 0.35 0.34 0.01 4.87 14.29 0.29
while a thick 14 in layer's required change might be from 0.29 to 3.29 ( 1.33 average)
in. So, for thicker pavements, using a harder grade of asphalt could lead to significant
Two can vary as a result of changes in temperature and mix design. The following
Three typical pavement sections were examined, and the change in required 0 2
(bituminous base thickness) was calculated. The three sections all contained a 1.25
in surface layer 0 1 , with 0 2 varying as 4, 8 , and 14 in. Using the AAS HO Road Test
layer coefficients (a, = 0 .44, a 2 = 0.34), structural numbers (SN) were calculated for
these pavements. Then, using a constant 0 1 = 1.25 in, 0/s were calculated based
on worst , average , and best temperature conditions and mix quality in Missouri. From
Table 35, layer coefficients for these conditions for the surface and base layers were
obtained. Finally, the 0/s required to maintain the computed SN's were calculated.
The results are shown in Table 39. As an example, using the AASHO layer
(temperature and 1990 mix quality), using a 1.25 in surface layer, the required 0 2 =
8.43 in, or a required increase of 0 .43 in. From examination of the table, several
things are apparent. First, average temperature and mix quality conditions lead to
rather small changes in required base thickness. Worst case sceneries of lower
quality mix used in the warmest part of the state lead to significant increases in
required base thickness. On the other hand, use of higher quality mixes in cooler
AASHO MHTD
o, D2 SN o, D,(in)
(in) (in) provided (in) Worst Average Best
a, = 0.44 a2 = 0.34 a, = 0.34 a, = 0.39 a, = 0.43
a2 = 0.28 a2 = 0.33 a2 = 0.38
1.25 4 1.91 1.25 5.30 4.31 3.61
1.25 8 3.27 1.25 10.16 8.43 7 . 19
1.25 14 5.31 1.25 17.45 14.61 12.56
SUMMARY
The purpose of this investigation was to determine layer coefficients for several
MHTD specified pavement materials. The coefficients are necessary as input to the
materials, and is reported herein. Volume II deals with unbound aggregate base and
Besides determining layer coefficients, the study also entailed the determination
gradation, testing temperature, aggregate source, and asphalt content within the limits
1. All materials were sampled and delivered to UMR by MHTD personal. Choice
were Type C, Type 1-C, and bituminous base. The specific materials making up
151
these types were two grades of asphalt cement, two surface mix sources of
coarse aggregate, two base mix sources of coarse aggregate, one source of
natural sand, two sources of manufactured sand, one source of mineral filler,
2. Routine index and specification tests were performed. For the asphalt cement,
viscosity, specific gravity, and softening point. The aggregates were tested for
combinations was determined by use of the Marshall mix design method (75
blow, manual flat-faced hammer). The same optimum asphalt content was
used for both the AC10 and AC20 grades because mixing and compaction
temperatures were adjusted to give equal mixing and compaction viscosities for
both grades. This resulted in 24 mixes. Finally, 24 additional mixes were used
which had 0.5% asphalt added above optimum. Thus, the total number of
7. Ten methods of characterizing gradation curve shape and position were used.
Two of these were unique to this study. The first involved the area between
the gradation curve and the maximum density line as plotted on FHWA 0 .45
8. Each mix was tested for indirect tensile strength. A regression model was fit
9. Each mix was tested for total resilient modulus (indirect tension) at three
programmed and developed to perform the tests, and to acquire, store, and
analyze the data. A total of 192 specimens were tested at three temperatures
10. The results of the MA testing were analyzed statistically to determine the
variables that were significant to changes in MA. The variables that were
12. Resilient modulus data from other studies found in the literature were merged
with the UMR data. A general regression model was fit to the overall data
base .
13. Air temperature data from 104 weather stations in Missouri were analyzed to
153
14. Pavement thickness data for MHTD flexible pavements were analyzed for mean
pavement thickness.
17. Mean vehicle speed data was supplied by MHTD. This was converted to load
dwell times and frequency for MHTD pavements. The same was done for
18. UMR, AASHO Road Test, and MHTD 1990 mix data were used to estimate
loading rate . This was in order to see which modulus would be most useful for
19. Five different methods of calculating mixture stiffness (Sm) were compared ;
20. Layer coefficients a, (for Types C and IC mixes) and a 2 (for bituminous base
nomographs to obtain a, or a 2 .
aU!1R = aAASHO
MR U!1R
1--'--
Jl/3
MR,AASHO
au11R
= a
AASHO
t:
, E*
1
E* : U!1R
,
IAASHO
]1/ 3
Sm,U!1R
aU!1R = aAASHO - - - -
Jl/3
sm,AASHO
temperatures (high, average, and low areas of the state) and the
Road Test .
21 . Two options were presented for the calculation of layer coefficients. Option
One resulted in a fixed layer coefficient per material. For 1990 mixes, Type C
22. A sensitivity analysis was performed which examined the effect of specific
· CONCLUSIONS
1. Ten methods were used to characterize gradation curve shape and position;
two of these were developed during the course of the study . The method of
determining the area between the 0 .45 power maximum density line (MDL) and
155
the gradation line had only a fair (R 2 = 0. 79) correlation with resilient modulus,
MR. This was because the area was not sensitive to relatively small differences
in position of the curve relative to the MDL. The second unique method
involved calculation of the slope of three different parts of the gradation curve.
merely including certain critical sieve sizes directly in the regression equation .
2. The voids analysis indicated that the method of assuming that the effective
specific gravities of low absorption aggregates is midway between the bulk and
the apparent specific gravities correlates very well with results from Rice
materials in this study), the estimation method underpredicts air voids by about
1%.
3. Of the five different methods for computation of mixture stiffness (Sm), the
method of Bonnaure, which uses the Ullidtz asphalt aging approximations, was
4. The regression model for indirect tensile strength was relatively strong (adj-R 2
where:
156
fixation, LVDT tip design, and resolution of both vertical and horizontal LVDT'S.
Constant diligence is required by the operator to assure that the very small
6. A relatively strong (adjusted R2 = 0.946) regression model was fit to the UMR
MR data:
log MR = 7 .137- 0.016 T - 0.005 AR4 + 0.088(}70 - 0.028 peffv - 0.016 peir -
4
0.006 p200 (JPopteffv - peffvl)
where :
T = mixture temperature, °F
content, by volume .
7. A statistical analysis of the data indicated that temperature was by far the most
important variable that affects MR, followed by asphalt viscosity and whether
lower MR, but was not statistically significant. Increases in (-) #200 material
should be noted that both coarse aggregates were crushed limestones, and that
particle shape were not present. All other things held constant, decreasing air
practical impact of the trends is that higher layer coefficients can be obtained
by:
Excessively hard asphalts, highly angular aggregates, and low air void values
durability, and rutting. Thus, caution should be used when interpreting the
results of the above analysis. Also, it must be kept in mind that the MR
9. The UMR data base was augmented with mix design/MR data from five
applicable, regression model was produced. The model was not as strong as
the UMR model, but was deemed superior because it represented a much wider
10. The Odemark equation was used to rate the three methods of obtaining mixture
resilient modulus, was: MR estimated from step 9, Shell mixture stiffness, and
dynamic modulus.
11. Two options to obtain layer coefficients were presented for possible use. The
determine the MR by test or to estimate the resilient modulus, but the moduli
interest. Then, the layer coefficient is computed via the Odemark equation
159
RECOMMENDATIONS
For Missouri pavement designers, two options are presented. Option One is the
simplest: if MR test data are available for a given mix (say, for three temperatures),
and 43) should be entered with the MR 68 value and the a 1 or a 2 value determined.
If MR test data are not available, the specific mix data should be substituted into Eq.
53, repeated here, and the above procedure followed to obtain a 1 or a 2 values .
keeping with the design location and pavement thickness, the procedure in Table 35
should be followed.
in present use in order to update or replace the above equation by use of a more
representative data set of the materials. A greater degree of accuracy will also
probably be achieved. Then, both Options One and Two will render more
It should be remembered that this study is in the mold of the traditional method
asphaltic material problems with thermal cracking and rutting, for instance, are not
directly addressed. To address a wider range of material issues, creep testing and
160
gyratory shear testing may be in order. These kinds of tests were beyond the scope
of this project . Also, this project was conceived in 1989 and the bulk of the testing
was performed in 1991, before the SHRP project results became generally known.
In the future, it may be that some of the recommendations coming out of the SHRP
program can be used to update the quest for layer coefficient determination .
ACKNOWLEDGEMENT
The authors wish to thank the MHTD for its sponsorship and support of this
research project . They also thank the UMR Department of Civil Engineering for its
support. Special thanks go to Mr. J.D. Stevenson and Mr . Kevin Hubbard for their
REFERENCES
p.
5. Van Til, C.J., B.F. McCullough, B.A. Vallerga, and R.G . Hicks, "Evaluation of
Missouri, 1988.
Granular Materials from Resilient Modulus, "Trans. Res. Rec. 852, Trans.
10. Jorenby, B.N. and R.G. Hicks, "Base Course Contamination Limits, Trans.
11. The AASHO Road Test - Report 5 - Pavement Research, Spec. Report 61 E,
12. Carree, B.J.D. and T.D. White, "The Synthesis of Mixture Strength
Distributions," Proc. of Assn. of Asphalt Paving Tech., Vol. 58, 1989, pp.
i---- -
162
109-141.
Coefficients , Vol. II : Unbound Granular Bases and Cement Treated Bases , "
15 . Van der Poe I, C. "A General System Describing the Viscoelastic Properties of
Bitumens and Its Relation to Routine Test Data," Shell Bitumen Reprint No .
.9_.
16. Pfeiffer, J.P. and P.M . Van Doormaal , "The Rheological Properties of
Asphaltic Bitumen , " J. Inst. of Petroleum Tech., Vol. 22, 1936, pp. 414 .
with the Aid of Their Mechanical Properties," Proc. Assn. of Asphalt Paving
19. Van Draat, F. and P. Sommer , "Ein Geratzar Bestimmung der Dynamischen
Elastizitats Modulu von Asphalt," Strasse und Autobahn, Vol. 35, 1965.
282.
21. Bonnaure, F., G. Gest, A. Gravois and P. Uge, "A New Method of Predicting
Pavement Design: The Shell Method," Proc. 4th lnt'I. Conf. on Structural
23. Ullidtz, P., "A Fundamental Method for Prediction of Roughness, Rutting,
24. Witczak, M.W., "Design of Full Depth Asphalt Airfield Pavements, " Proc.
Co., 1987.
26. Lefebvre, J.A., "A Modified Penetration Index for Canadian Asphalts," Proc.
99-117 .
28. Bell, C. A., "Use of the Shell Bitumen Test Data Chart in Evaluation of
Asphalt Data;" Proc. of Assn. of Asphalt Paving Tech., Vol. 52, pp., pp.1 ·
164
31 (1983) .
29. Button, J.W., J.A. Epps, D.N. Little, and B.M. Galloway, "Asphalt
Asphalt Concrete," Proc. of Assn. of Asphalt Paving Tech .• Vol. 38, 1969,
pp.140-178.
33. Miller, J.S., J. Uzan, and M.W. Witczak, "Modification of the Asphalt
Institute Bituminous Mix Modulus Predictive Equation," Trans. Res. Rec. 911
35. "Standard Test Method for Resilient Modulus of Subgrade Soils," AASHTO
Methods of Sampling and Testing, 14th ed. Part 111, Methods of Sampling
36. "Standard Test Method for Indirect Tension Test for Resilient Modulus of
165
11
Bitum inous Mixtures, ASTM D4123-82, Annual Book of ASTM Standards,
37. Von Quintus, H.L., J.A. Scherocman, C.S. Hughs, and T .W. Kennedy,
11 11
Asphalt-Aggregate Mixture Analysis System, NCHRP Rpt. 338, Trans .
38. Resilient Modulus for Asphalt Concrete, SHRP Protocol P07, Strategic Hwy.
Res. Pgm., Nat'I. Res. Council, Washington, D.C., 1992, pp. PB07B-PB07B-
17.
Sampling and Testing, 15th Ed., Part II, Tests, AASHTO, Washington, D.C.,
1990.
40. Kalcheff, I. V., and D.G. Tunnicliff, "Effects of Crushed Stone Aggregate Size
42. Hadley, W. 0., W.R. Hudson, and T. W. Kennedy, "Evaluation and Prediction
43. Meier, W.R., and E.J. Elnicky, "Laboratory Evaluation of Shape and Surface
Texture of Fine Aggregate for Asphalt Concrete," Trans . Res. Rec. 1250,
166
44. Mogawer, W.S. and K.D. Stuart, "Evaluation of Test Methods Used to
Quantify Sand Shape and Texture," TRB 71 st Annual Meeting, Trans. Res.
45. Kandhal, P.S., J. 8. Motter, and M.A. Khatri, "Evaluation of Particle Shape
and Texture: Manufactured vs. Natural Sands, "NCAT Rpt. No. 91-3,
46. "Test Method for Index of Aggregate Particle Shape and Texture," ASTM
4 7. "Standard Test Method for Particle Shape, Texture, and Uncompacted Void
49. Kennedy, T.W., and J.N. Anagnos, "Procedures for the Static and Repeated-
Load Indirect Tensile Test," Res. Report 183-14, Center for Trans. Res.,
51. May, R.W., and M.W. Witczak, "An Automated Asphalt Concrete Mix
Analysis System," Proc. of Assn. of Asphalt Paving Tech ., Vol. 61, 1992,
pp. 154-187.
53. Barksdale, R.D., "Compressive Stress Pulse Times in Flexible Pavements for
Use in Dynamic Testing," Hwy. Res . Rec . 345, Hwy. Res. Bd., 1971, pp.32
- 44 .
Procedures - Aggregate Variations and Effects," NCHRP Rpt. No. 69. Hwy.
56. Joel, R.N., "A Method for Controlling Concrete Workability Using Aggregate
1990, 318 p.
58. Goode, J .F., and L.A. Lufsey, "A New Graphical Chart for Evaluating
59. Mix Design Methods for Asphalt Concrete. MS-2. Asphalt Institute,
60. Kandhal, P.S., "Maximum Density Line: Which One Should Be Used?"
62. Anderson, D.A., D.R. Luhr, and C.E. Antle, "Framework for Development of
NCHRP Apt. No. 332. Trans. Res. Bd., Washington, D.C., 1990, 118 p.
63. Abkemeier, T.J., "Indirect Tensile Test Correlative Study," M.S. Thesis.
64. Adedimila, A.S., "A Comparison of the Marshall and the Indirect Tensile
66. Monismith, C.L. and K.E. Secor, "Viscoelastic Behavior of Asphalt Concrete
67. Adedimila, A.S. and T.W. Kennedy, "Fatigue and Resilient Characteristics of
Tensile Test," Res. Report 183-6, Center for Hwy. Res., Univ. of Texas at
Austin, 1975.
Indirect Tensile Test, "Proc . of Assn. of Asphalt Paving Tech., Vol. 46,
Analysis of Characterization for Design," Trans . Res. Rec. 821, Trans. Res.
72. Von Quintus, H.L., J.B. Rauhut, and T.W. Kennedy, "Comparisons of
Proc. of Assn . of Asphalt Paving Tech., Vol. 51, 1982, pp. 35-52.
73. Khosla, N.P., and M.S. Omer, "Characterization of Asphaltc Mixtures for
75. Kallas, B.F., "Elastic and Fatigue Behavior of Emulsified Asphalt Paving
1988,313p.
Fatigue," Hwy. Res. Rec. 404, Hwy. Res. Bd., Washington, D.C., 1972, pp.
1-7.
80. Schmidt, R.J. and P.E. Graf, "The Effects of Water on the Resilient Modulus
81 . Kallas, B.F., "Asphalt Pavement Temperatures," Hwy. Res. Rec. 150, Hwy.
82. Croney, D.C. and P. Croney, The Design and Performance of Road
83. Fairhurst, G.E., Y.R. Kim, and N.P. Khosla, "Resilient Modulus Testing of
0
171
APPENDICES
APPENDIX A
FINE AGGREGATE
FINE AGGREGATE
NAA METHOD A
APPLICATION
EQUIPMENT
1. Funnel
2. Funnel Stand
5. Spatula
6. Balance - readable to .± 0 .1 g
CALIBRATION OF MEASURE
Apply a light coat of grease to the top edge of the dry, empty measure. Weigh the
measure, grease, and a flat glass plate slightly larger than the diameter of the
C). Place the glass plate on the measure, being sure that no air bubbles remain.
Dry the outer surfaces of the measure and determine the combined mass of
V= W
0.998
where:
V = volume of cylinder, cm 3
PROCEDURE
into the combined gradation (fh9.:., the natural sand, the manufactured sand,
gradation.
2. Calculate the average specific gravity of the blend of the different types of
sand:
where:
for the ( + )#100 to (-)#16 size. For example, for a total sand
methods in ASTM C 117 and then dried and sieved into separate size
4. Weigh out and combine the following quantities of dry sand from each of
the sizes. These quantities are combined from the different types of sands
No. 8 to No. 16 44
No . 16 to No. 30 57
No. 30 to No . 50 72
190
5. Set the funnel stand in the pan. Place the funnel on the funnel stand.
6. Mix the test sample until it is homogenous. Using a finger to block the
opening of the funnel, pour the test sample into the funnel. Level the
material in the funnel with the spatula. Center the measure under the
funnel, remove the finger, and allow the sample to fall freely into the
measure.
7. After the funnel empties, remove excess heaped sand from the measure by a
single pass of the spatula with the blade vertical using the straight part of its
edge in light contact with the top of the measure. Until this operation is
176
grains from the outside of the measure and determine the mass of the
Note 3 -- After strikeoff, the measure may be tapped lightly to compact the
8. Collect the sample from the retaining pan and measure, recombine, and
repeat the procedure again. The results of two runs are averaged. See
9. Calculate the uncompacted voids to the nearest 0.1 percent for each
determination as follows:
U = V - (M/ G) x 100
V
V = volume of measure, cm 3
Note 4 -- For most aggregate sources the fine aggregate specific gravity does not
vary much from sample to sample or from size to size finer than the 2.36-mm (No .
routine specific gravity test of an as-rece ived grading of the fine aggregate. If
should be determined on material from the same field sample from which the
uncompacted void content sample was derived, Normally the as-received grading
can be tested for specific gravity, particularly if the 2.36-mm (No. 8) to 150-um
(No . 100) size fraction represents more than 50 percent of the as-received grading.
However , it may be necessary to test the graded 2.36-um (No. 8) to 150-mm (No.
100) sizes for specific gravity for use with the graded void sample, Method A or
the individual size fractions for use w ith the individual size method, Method 8, (not
included here). A difference in specific gravity of the 0.05 will change the
DATA SHEET
30
50
100
Total: A = B = C =
=
wt. Ret. on #16, sand 2
F2 , 16
Wt. Ret. on #16, both
=
Wt. Ret. on #30, sand 1
F1,30
Wt. Ret. on #30, both
=
Wt. Ret. on #30, sand 2
F2,30
Wt. Ret. on #30, both
179
=
Wt . Ret. on #100, sand 2
F2,1 00
Wt. Ret . on #100, both
180
DATA SHEET
PARTICLE SHAPE AND TEXTURE OF FINE AGGREGATE
Sample _ _ _ __
Date Tested _ _ __
Technician
Specific Gravity
G, 2 P, 2
Sand #1
Sand #2
STANDARD GRADATION
Size Fraction
Mass sand #2 = F2 • 44 g
Mass sand #2 F2 • 57 g
Mass sand #2 = F2 • 72 g
Mass sand #2 F2 • 17 g
TOTAL 190 g
UNCOMPACTED VOIDS
Trial 1 Trial 2
Weight measure, g
APPENDIX 8
COARSE AGGREGATE
APPLICATION
For asphalt mixture aggregate from ( +) #4 sieve to the (-) 1 in sieve size fractions.
EQUIPMENT
1. 6 in dia. Mold - inside height 7.00_±_ 0.01 in for testing the following
fractions:
( +) 3/4" to (-) 1 in
(+) 1 / 2" to(-) 3/4 in
CALIBRATION OF MOLD
1. Fill the mold with water at room temperature and cover with a piece of plate
glass in such a way as to eliminate bubbles and excess water.
3. Measure the temperature of the water and determine the volume of the mold
by multiplying the mass of the water by the corresponding specific volume
of water given in Table 8-1 for the temperature involved.
PROCEDURE
6 in mold 13 lbs
4 in mold 4 lbs
3. Wash the sample of aggregate by decanting the wash water through a sieve
at least one size smaller than that being used. Continue the washing and
decanting operation until the wash water is clear. Then flush the residue on
the sieve back into the aggregate sample. Dry the sample to constant
weight at a temperature of 230 .±. 9°F ( 110 .±. 5°C).
5. Place the cylindrical mold on a uniform, solid foundation. Gently place the
aggregate, from the lowest height possible , into the mold until it is
approximately one-third full. Level the surface with the fingers, and
compact the layer using 10 drops of the tamping rod evenly distributed over
the surface. Apply each drop by holding the rod vertically with its rounded
end 2 in (50 mm) above the surface of the aggregate (controlled by the slot-
and-pin arrangement) and releasing it so that it falls freely. Place a second
layer in the mold using the same procedure, filling the mold approximately
two-thirds full . As before, level the surface and apply the same compactive
effort , 10 drops of the rod. After the final layer has been compacted, add
individual pieces of aggregate to make the surface of the aggregate mass
even with the rim of the mold, with no projections above the rim. Determine
the mass of the aggregate in the mold to an accuracy of at least 4 g.
6. Repeat the filling of the mold using the same specimen and compaction.
Make a second determination of the mass of the aggregate in the mold as
described above . Use the average mass of the two runs in calculating the
percentage of voids at 10 drops for each size .
7. For the higher degree of compaction, follow the steps outlined in #5 and #6,
except use 50 drops of the tamping rod in compacting each layer. Again
average the masses from the two runs for use in computing the percentage
of voids at 50 drops for each size fraction.
where :
9. Determine the particle index (lal for each size fraction tested as follows :
For sizes represented by less than 10% in the grading, for which no particle index
data were obta ined, use the average particle index of the next coarser and finer
sizes for which data are available or the particle index for the next coarser or finer
size if a value is available only in one direction .
185
72 (22) 1.0022
75 (24) 1.0027
79 (26) 1.0032
82 (28) 1.0038
86 (30) 1.0044
90 (32) 1.0050
186
DATA SHEET
Index of Particle Shape and Texture (03398)
3/8" to 1/2"
I
#4 to 3 / 8"
I I
Total 100
PARTICLE INDEX
Mass of mold, g
Mass of aggregate, g
Average of 2 trials ( M 10 )g
Moss of mold , g
Mass of aggregate , g
Average of 2 trials (M 50 ), g
Mass of aggregate (M 10 ), g
Mass of aggregate (M 50 ), g
lal = 1 . 2 5 V 1 0 - 0 . 2 5 V 50 - 3 2. 0
187
PARTICLE INDEX
Mass of aggregate, g
Average of 2 trials (M 10 )g
Average of 2 trials (M 50 ), g
Mass of aggregate (M 10 ), g
Mass of aggregate (M 50 ), g
PARTICLE INDEX
Mass of aggregate, g
Average of 2 trials (M 10 )g
Mass of aggregate, g
Average of 2 trials (M 50 ), g
Mass of aggregate (M 10 ), g
Mass of aggregate (M 50 ), g
APPENDIX C
DETERMINATION OF
FOR
BITUMINOUS MATERIALS
190
1. INTRODUCTION
were tested using repetitive compressive haversine load pulses of 1 Hz frequency with
a 0 . 1 second load duration and a 0.9 second rest period. The thickness of the
specimens tested varied between 2.4 in and 2. 7 in. Prior to performing the resilient
modulus test, the indirect tensile strength was determined at 77 ± 2 °F for each
combination of mix type (bituminous base, Type C and Type IC), gradation (fine and
coarse), asphalt cement grade (AC-10 and AC-20), asphalt cement content (optimum
and 0. 5 % above optimum) and aggregate (DR-6 and DR- 7 for bituminous base mixes,
DR-4 and DR-5 for Type C and IC mixes) . Duplicate specimens were prepared for the
specimens were aged for one week before testing . The value of tensile strength
determined by this procedure was used to estimate the indirect tensile stress and
during subsequent resilient modulus testing. For resilient modulus testing, four
specimens were made for each mix. In all, 192 specimens were tested. The
1.2 Typically, the specimens were built on a Tuesday and then testing would
begin for those specimens on the following Tuesday at 41 °F, continue at 77 °Fon
Wednesday, and finish at 104 °Fon Thursday. The testing was done from lower to
specimens were brought to the test temperature 24 hours before testing began.
191
1 .3 Once the test was completed, plots were made of load, horizontal
deformation and vertical deformation with time. The measured total recoverable
Poisson's rat io . The resilient modulus values were then calculated using cyclic loads
2.1 Both the indirect tensile strength and the resilient modulus specimens are
made using the same procedure as that described in the Marshall Mix Design portion
of this report. In addition, all other specimen properties except flow and stability are
2.2 Specimens are marked and measured as follows after determination of the
2.2.1 For indirect tensile strength specimens use a paint marker and a
straight edge to draw a thin line along the diameter of the specimen face with the
smoothest texture. Use a 4 in filter paper with notches on opposite sides of its
diameter to mark the two points needed to draw the line. Avoid drawing the line
2.2.2 For resilient modulus specimens, mark a diametrical axis on the test
specimen as specified in 2.2.1 . Add an arrow tip to one end of the line. Using the
paint marker and a straight edge, draw a thin horizontal line at the mid-thickness of
2.2 .3 Measure the thickness and the diameter of indirect tensile strength
and resilient modulus test specimens to the nearest 0.01 in using dial calipers.
Determine the thickness by averaging a single center measurement with three equally
spaced measurements located 0.5 in from the test specimen edge. Determine the
diameter by averaging the diameter of the specimen at mid height along ( 1) the
192
diametrical axis drawn in 2 .2 . 1 and 2.2.2 above and the axis perpendicular to the axis
3 .2 Place the specimens to be tested in watertight buckets and then put the
3.3 Prepare the xy plotter , LVDT signal conditioner, and the strain gage signal
3.4 Place the test specimen in the test press loading apparatus and position
it so that the mid thickness of the test specimen is located in the line of action of the
load cell. Use the diametrical marking to ensure that the specimen is aligned from top
to bottom-the diametrical marking should be centered on the top and bottom loading
strips .
3 .5 Turn on the test press, check the zero point on the xy plotter, make sure
the plotter pen is touching the paper, and depress the UP button on the test press to
of the test press platen of 2 in/min. The platen will continue to move until the
maximum load is reached and the load decreases as indicated by the strain gage
indicator. The upper limit switch on the test press forces the platen to stop moving.
The upper and lower limit switches on the test press can be adjusted using the button
underneath the test press. Record the maximum load reading from the strain gage
conditioner on the chart paper. Note: if the beam holding the plotter pen binds during
the test, quickly turn off the power to the plotter and straighten the beam so it can
3 .6 Depress the DOWN button on the test press to lower the platen. The
lower limit switch on the test press will stop the platen. Remove the breaking head
193
from the platen, sweep any loose asphalt off of the platen, and save the test
specimen by placing it in a plastic bag or some other suitable container. Write the
specimen ID on the bag. Remove the recording chart from the xy plotter and fill in the
chart data block. Clean the breaking head loading strips and lubricate the guide rods
3 . 7 Make a sketch of the specimen failure on the data sheet and calculate the
where:
3.8 Prepare a table of average indirect tensile strength values for each mix
along with tensile stress levels of 30, 15, and 5 percent of the tensile strength to be
41, 77, and 104 ° F, respectively. Include specimen contact loads shown in Table 21
of 3, 1 .5, and 0.5 percent of the tensile strength to be used during resilient modulus
determinations at 41, 77, and 104 °F respectively. Contact loads and cyclic test
The system includes an MTS 810 closed loop electrohydraulic load system
specimen yoke, a yoke stand, and a data acquisition and load signal generation
computer/software system. The specimens were seated on the lower concave loading
194
strip which was affixed to the loading frame. Both upper and lower loading strips had
The specimen yoke was a small frame that held two horizontal L VDT'S which
were diametrally opposite. Originally, the LVDT's were fixed to a base. However,
specimens tended to rock under load; this could be discerned by both visual
observation and by examining the readout from each horizontal channel. Ideally, both
LVDT's should be compressed upon loading; however, sometimes one of the readouts
indicated that one LVDT was extending. It was thought that possibly one of the tips
of the L VDT could be slipping down into a surface void. So, the tips were replaced
with disk-type tips that threaded onto the LVDT cores. The idea was that the 1 /4 in
(0.64cm) diameter disks would bridge across surface voids. Next, the method of
transferring the load was changed . Originally, the upper load strip was not fixed; a
ball bearing was used to transfer the load from the load cell, which was fixed to the
MTS load ram, to the top of the upper load strip. This configuration was basically the
state of the art several years ago. Additionally, vertical deformation was measured
with an L VDT internal to the MTS system; the L VDT had a full range of 6in (2.4cm).
Data indicated that the specimen was unstable, and that the vertical L VDT was not
sensitive enough, even though the controls of the MTS were set to utilize 10 percent
L VDT to the yoke base, and fixing the upper loading strip to the load cell in order to
but there still seemed to be specimen movement, even under higher baseline static
loads. The yoke base still had to be centered manually to align the upper and lower
load strips .
In an effort to assure better vertical alignment of the upper and lower load
195
strips, a new base was purchased which was manufactured for the purpose of
performing repeated load and static load indirect tension tests. This device featured
an upper platen which slid vertically by way of two guide posts, thus aligning the
upper and lower load strips . The horizontal LVDT's were fixed to the base as well as
the vertical LVTD. However , some rocking was still evident, possibly due to slight
looseness in the guide bearings and to the difficulty of aligning the upper platen with
case, a light gauge steel frame holding the two LVDTs was mounted directly onto the
specimen, reminiscent of the original Schmidt device (80) and the newer MTS resilient
modulus system (83). This allows the LVDTs to follow the specimen should rocking
occur . This, coupled with the higher static background loads recommended by the
latest draft of the SHRP protocol, and coupled with the use of guide rods, seems to
the SHRP protocol , two vertical LVDT's were mounted to monitor the movement of
the upper platen, rather than the ram movement . This virtually eliminates machine
deflection from the vertical deformation measurement, and allows for a more accurate
calculation of Poisson ' s ratio, which is necessary for a truer calculation of resilient
modulus. The f inal setup is shown in Fig. D1. An auxiliary device was necessary in
order to mount the yoke accurately and consistently on the sample, as shown in Fig.
02.
41 °, 77°, and 104°F (4 , 25, and 40°C) plus or minus 2°F(1 °C) . A 1 .5 in (3.8cm)
thick styrofoam box was built to surround the yoke . Also surrounding the yoke on
three sides were 0 .25in (0 .64cm) copper coils through which was circulated a mixture
heat ing / re fr ige rat io n unit . A small fan w as mounted ins ide the env ironmental chamber
to d ist ribut e th e air . A dial gage thermomete r was mounted through the box to
monitor air tempe rature. The chamber was found to be able to maintain the
temperature within the ASTM 4123 specified 2°F ( 1 °C) . The env ironmental chamber
The data acquisition / load signal generation system consisted of the following
megabyte hard d isk , Data Translat ion OT 2801 A 12 bit analog /digital board , printer,
color mon itor, one strain gage conditioner, four L VDT conditioners , ASYST software,
a 2500 lb. Stra in-sert load cell , two Schaevitz LBB-375-TA-100 LVDTs with a range
of .±.0. 1 OOin (0 . 25cm) of travel (horizontal deformation), and two Schaevitz PCA-220-
100 LVDTs with a range of .±.0. 100 in (0.25cm) of travel (vertical deformation).
The m inimum value measurable with the system setup is calcu lated as fol lows .
The input voltage range of the horizonta l LVDT's was -10 to + 10 V. The precision
of the input signal measured by the 12 bit A/D board is 20V/2 12 which is 0.00488
V. When 0.1 in full scale displacement LVDT's were used, the horizontal transducers
were ranged 10% of full scale which calibrates the output to a finer scale travel and
allows higher resolut ion measurements. The vertical transducers were ranged 33%
of full scale . The min imum value of horizontal displacement measured was then (0.01
in/1 OV) * (0 .00488V) which is 0.00000488 in. The vertical LVDT's could be read to
(20/2 12 )*(0.033/10) =0.000016 in. The load cell could be read to (2500
lb/10V)*(0.00488) = 1 . 22 lbs .
temperature env ironment needed to bring them to the specified test temperature at
198
least 24 hour s pr ior to testing . A water bath is used to cool the spec imens to 41 °F.
The spec imen s are pla ced in 2 gal po lyethylene buckets before being put in the water
bath. Eac h bucket can hold 6 specimens. Use the hole in top of each bucket to
check the specimen temperature with a dial gage thermometer . Allow the specimens
to sit in the ambient air to come to 77 ° (the ambient conditions can be controlled by
opening the door of the testing room or by turning the room air conditioner on and
off), and use an oven to bring the specimens to 104 °F . Keep a dial gage or mercury-
maintain the testing temperatures of 41 °F and 104 °F during resilient modulus tests .
A 50-50 mixture of antifreeze and water is used as the circulating liquid . Check the
level of coolant in the circulator reservoir before turning the circulator on. The coolant
level should be above the coils in the reservoir at all times. Set the thermoregulator
(read to the top of the thermoregulator indicator). Check the temperature of the
cabinet us ing a dial gage thermometer and the thermometer access port on the top
left of the cabinet. Both the circulator heating and cooling switches should be "on"
to reach 41 °F and the "heating only" switch should be "on" to reach 104 °F . Move
the control valve lever forward on the right side of the circulator to circulate coolant
° F as specified in 4 . 1 . Place the loading head with the vertical L VDTs and the
horizontal frame with the horizontal LVDTs in the temperature cabinet 24 hours prior
to testing.
4 .3 Equipment calibration. Before each test, make sure that all LVDTs and the
load cell are securely mounted for testing and that there are no loose connections with
4 .3.1 Calibrate the LVDTs before any testing begins, and monitor their
calibration after every 48 resilient modulus tests using the micrometer calibration
block . Monitor the calibration of the LVDTs at each test temperature since there is
a slight variation of L VDT response with temperature. Prior to each specimen test,
make sure that the shaft of each L VDT is not sticking by depressing and releasing the
LVDT tip. Apply silicone spray or WD-40 on a regular basis to the LVDT shafts. Also
4.3.2 Calibrate two load cells before performing resilient modulus tests.
One load cell is used as the MTS load cell during testing and the other is used to
check the calibration of the testing load cell every 48 resilient modulus tests at 77 °F
only.
4.4 Prepare testing log. For each combination of mix, gradation, asphalt
cement type, asphalt cement percentage and aggregate, a testing log is prepared
before testing. The log uses the maximum suggested seating load (P contact) and the
3.8, along with load cell calibration data to determine the necessary cyclic load
(P cyclic), MTS set point and haversine load pulse for testing the 4 specimens
Note: round P rnax and P contact down to the nearest 10 lbs and 1 lb respectively before
calculating P cyclic to avoid overloading the specimen. Round Pcyclic to the nearest 1
lb .
4.4. 2 The log also has space for recording the cumulative vertical
deformation and the number of preconditioning and load cycles used for each
201
specimen test. The cumulative deformat ion is measured using the dia l gage
magnet ical ly mounted to the hydraulic actuator shaft . Subtract the dial gage reading
in 4.7 from the read in g in 4.8 to get the cumulat ive vertical deformation. If total
cumulat ive vertical deformations greater than 0.025 inch for 41 °F or 0.050 in for
77° and 104 °F occur, reduce the applied load to the minimum value possible and still
4 .5.1 Take the loading head assembly from the temperature cabinet .
Place the lucite specimen on the lower section loading strip. Place the upper section
of the loading head on top of the specimen by sliding it over the guide rods. Rotate
the lucite specimen so that its d iametrical marking is coincident with the upper and
lower loading strips. Place a weight on top of the upper section and make sure that
the tips of the vertical LVDTs are touching the heads of the bolts clamped to the
upper section . Use the hose clamps to adjust the position of the LVDTs until their
respective L VDT conditioner reads about + 0 .4500 for the front LVDT and -0.4500
for the rear LVDT . Make sure that the engraved 14's on the upper and lower section
of the loading head are on the same side. After the vertical LVDTs are adjusted,
remove the lucite specimen and return the loading head assembly to the temperature
cabinet .
-----
4. 5 . 2 Get a specimen for testing from the appropriate controlled
temperature environment (make sure that the specimen and the MTS temperature
4. 5 .3 Take the horizontal L VDT frame from the temperature cabinet and
place it on top of the alignment loading strip. Place the specimen in the horizontal
L VDT frame so that the diametrical axis of the specimen is perpendicular to the plane
of the LVDTs . The arrow of the diametrical axis should be pointing down . Use the
202
th in piece o f ru bbe r to pro t ec t the unma rked face of the spec imen from the po inted
frame . The screw should make contact at the center po int of the specimen along the
marked d iametrical axis . Do not allow the screw to penetrate more than 1 / 16 in into
the face of the specimen . The pointed tip screw works well on specimens with
smooth faces , and the flat t ip screw works well on spec imens w ith rough faces . It is
helpful to hold the specimen and the frame with one hand while turning the screw
with the other hand . Make sure that there is no gap between the frame and the
that the ir respect ive LVDT condit ioner is read ing about + 0.2500 . Make sure that the
LVDT ti ps are not to uch ing any vo ids on the sides of the spec imen.
section of the load ing head inside the temperature cabinet. Move the frame and
specimen by grabbing hold of the top of the specimen . Rotate the specimen sideways
to clear the loading head guide posts when placing it in the cabinet . Place the upper
section of the loading head making sure that the engraved 14' s line up as spec ified
in 4. 5. 1. Place the steel contact ball in the center of the loading head upper sect ion .
coincident with the upper and lower loading strips, and the mid-thickness marks on
the specimen are located in the line of act ion of the actuator shaft . Use the peep hole
on the left side of the cabinet to center the specimen on the upper and lower loading
strips. Use the ruler to ensure that the horizontal LVDTs are at the same height above
4. 5 . 6 Remove the dial gage from the hydraulic shaft . Use the setpoint
knob (turn the knob clockwise for load ing) on the MTS 442 contro ll er panel to bring
down the load frame loading ram . Overshoot the set point specified in 4. 5 .2 to get
the ram moving, but be careful not to get the ram moving too fast, otherwise the
spec imen may be damaged if too large a load is applied. When the load cell on the
end of the loading ram is near the steel ball on top of the loading head, adjust the
setpoint to the exact value specified in 4 .4 . 1. The steel ball should seat uniformly in
both the loading head and the depression in the center of the load cell. Return the
peep hole cover and the front cover of the cabinet to bring the cabinet back to the
test temperature . No more than five minutes should elapse between removal of the
specimen from its controlled temperature environment and application of the contact
load on the spec imen in the loading frame . Return the dial gage to its position on the
hydraul ic shaft .
4 . 6 Configurin g data acquis iti on and control software. Turn on the computer ,
monitor and printer and wait for the DOS prompt C: >.
reading . Precondition the specimen along the diametrical axis prior to testing by
apply in g the repeated have rsine-shaped load pulse of 0 . 1 sec with a 0.9 sec rest
period for the specified number of cycles in 4.6.2. After each sequence of
preconditioning cycles, view the channel plots by using the Enter key to go to the
next plot . While inspecting the vertical and horizontal response curves, calculate the
vertical and horizontal deformation ratios to ensure that the ratios are less than or
equal to 1 . 5 . Calculate the ratios as Rv = Vmax / Vmin and Rh= Hmax/Hmin where Rv
is the vertical ratio and Rh the horizontal rat io. Vmax is the maximum total
20 4
deformat io n o f t he t wo vert ical L VDT curves and Vmin is the minimum total
deformat ion of the two curves. Likewise Hmax is the maximum total deformation of
the two horizontal L VDT curves and Hm in is the min imum deformation of the two
curves . The deformat ions are measured by taking the difference in height of the peak
and basel ine of the largest peak on each plot . When inspecting the plots, check that
successive deformation readings agree within 10 percent and that a line drawn
through the peaks on each plot would be slop ing upward for all plots or sloping
downward for all plots . Preconditioning can be stopped when the minimum number
of cycles specified in 4.4 .2 and the criteria outlined above are met . If the criteria
above are not met , make adjustments to the seating of the specimen in the loading
head without removing the contact load . In no case should the maximum number of
precondit ion ing cycles be exceeded . It the cr iteria above cannot be met within the
max imum number of precondi ti on ing cycles, remove the contact load from the
specimen , enter bye to get back to the DOS prompt on the computer and repeat steps
resilient modulus test ing program . "Resmod .run" will apply 37 load pulses to the
specimen as expla ined in 4. 7 . The last 7 load pulses and the resultant measured
deformations can be viewed in the channel plots after the 37 load cycles have
completed . To determ ine when load ing is completed , use a stopwatch to time the 37
load cycles (each cycle is 1 second). The criteria in 4.7 are used to accept or reject
the test except that the range in deformation values of five successive horizontal
deformation values mus t be less than 10% of the average of the five deformation
values. If the criteria are not met , remove the contact load from the specimen, enter
bye to get back to the DOS prompt on the computer and repeat steps 4.5.5 through
4. 7 above . Make a note of the hydraulic shaft dial gage reading and record the
205
APPENDIX D
To determ ine the type (S,B,or W) of asphalt, the Heukelom Bitumen Test Data
Chart (BTDC) is used (Fig . 3). The vertical axis is actually in consistency units "C".
range, c is:
applicabil ity as :
BTSvisc = (87)
'T'
-u. ooop (88)
where:
As can be seen, both criteria are m~t in regard to classification as "S" type
asphalts . Additionally , Heukelom defines "S" -type asphalts as those that plot in
approximately a straight line on the BTDC paper. Both asphalts used in this study
AC-10 60 135 301 1099 2.83 1.007 44.5 -0.92 2.1 0.24
APPENDIX E
APPENDIX F
219
!UNIT WEIGHT VS . %Aej !%AV BY RICE VS. %Aej
157
156
I I
~
V
"- 155
I I
•
:l:
~
"' 154 ' •
'g
::::,
153
152
' T .I •
I
I
151 1
T • i
~5 4 ~5 5 ~5 6 LS 7 ~5 4 ~5 5 ~5 6 LS 7
%AC By Wgl Of Agg. %AC By Wgt. Of Agg.
8
14
12
••
' ! ! ! ! !
;.::
! 10
8 •
6
T I 12
T
I I
4 ~5 5 ~5 6 LS 7
1
1.s 4 ~5 5 ~5 6 LS 7
'lf,AC By Wgl Of Agg . %AC By Wgl Of Agg.
1500 /i; 60
"E
i 1000 ~
~ 55
500 50
! I I
~.5 4 4.5 5 5.5 6 6.5 7 ~ -5 4 4.5 5 5.5 6 6.5
%AC By Wgl Of Agg . %AC By Wgt. Of Agg .
tpbbdr6f. wq 1
D:ite Printed: 10-Aug-93 Test Property Curves for BBDR6C Mix
220
!UNIT WEIGHT VS. %Ac\ l%AV BY RICE VS . %AC!
. ••
157
' i
156
I ' ~ ' •
~
:l:
~
Cl
155
15-4
T
! 't •
•
5
2
'
§ 153
152
5
1
;' -
151
3 3.5 , 4.5 5 5.5 6
' :u 4 46 5
' 6.6
%AC By Wgt Of Agg. 1"AC8yWglOf-
I !
' ,•
12
14
i:
...
.51 12
I
I
.
~
.. 11.,
10
8
I
T
i 11
10. g
• ' • •T
~
'
3.5 4 4.5 5
%AC By Wgt Of Agg.
5.5 6 -
T
&
'
.. , •
1"AC ByWgl O f -
iii
.t::
em
::E
2000
1500
I
• ·-
•
16
1000
3 3.5 , 4.5 5 5.5 6 4 4.6 6 u
%AC By Wgt Of Agg. 1"ACByWg<Of-
tpbbdr6c.wql
Date Printed : '.;]- Mar-91 Test Prope rt y Curves fo r BBDR 7F Mix
22 1
!UNIT WEIGHT VS. %AC! j%AV BY RICE VS . %ACj
15 1 75
150.5
I i I
i I I 0 ...
u
150 0 i I
Q.
149.5
i l I " I I i
.c
a, 149
I
i I i ~·
w
5
I I
;
~
!,2
,.8.5 I I : a:
~ 45
I
'le
::, 148
I ! I
I > • l
147 .5
I I I ;! 3S
!
I
147
3.5
I
4 45
i
%AC By Wgt. Of Agg .
5
I
5 .5 2.S
~· u
I
M C ByWgt O f -
I
~·
I i I ! 13
I
. 5
I
I
I
l .
0
ii 12.1
~
I
I
• !
C 12.S
I ,I
7.5
i
T
I I
l 12.4
I i
I
..••
7 12.2
i i ! I
5 12
. • •s 55 H 5.5
M C By Wgt Of Agg MC By Wgt Of.OW
5000
!STABILITY VS. %ACl
I I I
.. f'/FAC BY RICE VS. %Acj
I I I
I I
I i I I
•i I!
0
on I
•
. I E 10 I ! I
i
2
• i
I
I w l I
0 ..
I ii:
I I I
I I •I I
1ii ..
;
0
.
..,
I
!
I
I
I
I
I I
45
I I i
4. 5
I
5.5
M C By WQl Of.OW MC ByWQlOf.OW
tpbbdr 7f. wq I
D.1t e Pri nted : 08-Dcc- 92 T est Property Curves for BBDR7C Mix
222
.... !UNIT WEIGHT VS. %AC! l/oAV BY RICE VS. %AC!
0
,..,
I
i l
i
I I' I
I 5.5-----+----+--+----+---I
!
.---,-,---1--+---+---+---1
i
l
I
: i ! o"1 -
~ 4..5+---+---+----+--+----+---I
I I
I I
i
I i •+---+--_._---+--+----+---I
~ 3.l+---+---,..--•r---+---+---1
.
I
I I
I ! I ,.a:
I
! I I
• I ; 2.1
·---~----~----~-
151 21----t----1r---+-------t----1
I
150.5
2. 5
• I I35 ,
I
,s 5.5
U+---+----t-----+--+---+---1
2.5 u
......c ByWgt OfAw
i i
0
i ~ ,a.
I
2
•
I I I '"
~
.,. '
I I T
• • I . :.
10
• t
I
I
•I I I I
5
2.5 u
!! I
I I I
l ! I
• .t
I
I I 'I •I 2.5 3.1 4 4.!5 u
MCByWgt OfAw
tpbbdr7c.wq 1
Date Printed : 21 -Ma r-93 Test Property Curves for CDR4F Mix
223
JUNIT WEIGHT VS. %Aej J%AV BY RICE VS. %AC!
15J 5
15J
• •
I i ~ i j
i ! I
I I l I ;
152.5
I I
• T I •' ••
u i
152
Q.
1: 151.5
I ; I I I I
~
0,
15 1
i
•
I I
I
I
I
: I
~
:,
150 5
!
I I I i I I
149.5
150
i i i I i •~ I
I I
I I I
1'9
3 3 .5
'
4 4.5 5 5 .5
%AC By Wgt. OI Agg.
6 6.5 .,
3 H Ui 5
MC llyWg< OfAQI.
5..5 ...
8 16 I !
' E •• 5 I I I !
i I :I
•
J
0
.:.:
14
12
10
I
I .
~
"'u
.
ic 13. 5
>-
<
I
i
I
!
I
I
I
.•
I
e i
•
J l
~ ,~ 5
i I I I I I
6
'T T
~
I
!
I . • j i I I
! ! !
4
3 as • 45 s ~5
MC By Wgt. OI Agg .
s ~5 tt .5
...T 45
I
5
MCByWg< OfAQI;
5..5 ...
t
1i
2000
/
I .• .
i I
l
I
~
]
1800
1600
I
I !
i
!
:
I
-
•
I • I I I •
;•
~
; 1400
I ! I I ! ! I I I
I
1200
10003 3 .5
I
4
I
5
I
!
5.5
4 .5
i •
I
6 6.5
•... I !
i
I I
I
I
45 5 5..5
%AC By Wgl OI Agg. MC ByWg\ Of Aog
tpcdr4f.wql
D3le Printed: D- Mar -93 Test Property Curves for CDR4 C Mi x
22 4
!UNIT WEIGHT VS . %Aej !%AV BY RICE VS. %AC!
152
151
I i
I I
..• I ' ! i
I I
u
a.. 150
I
I
•
I
•. I 8,:: • .•• I
I
! I
i
•I
• I
."
t;; •
•T
. I I I
.c
°'
l
~
149
1.a
•'' ! I
0
ii'
:;
•
I I .. !
:,
14 7
i I I j 2
I I II
i I I'j
I I I
146
3 3.5 4 4.5 5
MC By Wgt Of Agg.
5.5 0
•
I
..
I
MCllyWr,.Of-
I
...
•
3.5 4 4.5 5
MC By Wgt Of Agg.
5.5 12
• :u • 4.5
MCBr Wr,.Of-
...I
. 2200 - ' I I I I
B 21 00 !
i
I
i
- I I
t 2000
:0
~
I
•i i . I I I
I I
"ii
.t:
l!!
1900
1800
I
I
•
I
I ;
•
2
1700
1600
'I I
1 • i I
I '' I iI I
1500
3 3.5 4 4.5
MC By Wgt Of Agg.
5 5.5 - ... ' 4.5
MC8y Wr,.Of-
o.s
tpcdr4c.wq l
Date Printed: 08-Dec-92 Test Property Curves for CDR5F Mix
225
IUNIT WEIGHT VS. %AC! !%AV BY RIC E VS. %AC!
152 9 I
i I
i i I I I I
I
' 8
~
151
150
I I I i r 0
0
I
7
I
I
!
I' 'I I ! tu 6
149 :I'
:l:
~ 1'8
I I ! ! w 5
u I
I I I I !
i
c
::,
U7
U6
i I I
I
a:> '
"'> 3
2
• ~
+
i
.!.
I
I I
145
14'1
U
I
3 U
I
,
I
I
U
MC By Wgt Of Agg
5 U
' ~ .5 3 3.5
'
i
I
, .5
MCByW gt Of Agg.
+
5 5 .5
11
0
ffi 12.5
! ! !
- 10
:I:
w
12
I
. 9
~ I
i
T
...
] 8 >
"'< 11 .5
7
8 •
I
~ 11 I
I
I
I
3
•
3 .5
I
4
I
, .5
I
5 5.5 10.~ .5 3 3.5 4
T I
, .5
I
5 5.5
%AC By Wgt Of Agg. MC By Wgt Of Agg.
i I •I
~ T
,Q 3000
80
l
t
I
2800
2600
•
' :I'
w
u
a: 80
70
I
-;; 2.00 >
..c
~ 2200 "'u 50
:I' <
~
2000
! '°
1 ~.5 3 3 .5 , , .5 5 5.5 ~ .5 3 3.5 4 4.5 5 5 .5
%AC By Wgt Of Agg. %ACByWgt Of Agg.
tpcdr5f.wql
D.lle Printed : 08 -Dec-92 Test Propert y Curves for CDR5C Mix
226
!UNIT WEIGHT VS . %AC! !%AV BY RICE VS . %Aej
151 •
. • i
•
: I I
.
150.5
•I I i I
u
150
I I
C
0
:,:
•5
: I
c.. 1"8.5
t;; I I
:c
;
"'
1'9
1.CS.5
I
I .
:,
u
ii'
•
:I. 5
I
I
I
I
~
:,
1'8
1'7.5
1•1
• I
I
I
I
iii
; 25
-
I
i
I
.- I
I
i
• I
i •
1~.5
3 .6 • .1 • .6
MC By Wgt Of Agg.
5 .1 1
... 41
I
4.1
'-'C 8y Wg< OI "4111
I
6.t
8
.!.!i ••
a:
11.
11 .
I
!
..:
I I ~ • 11 .
7 i 2 11 .
1
I
6
10.
• . I
•...
I
~1 ~6 ~1 10
4.1 4.1 1,. t
MC By Wgt Of Agg. '-'C lly Wgl Of "4111
2800
I -I 16
I I
I
a"
2600
• I
I I I . 0
0 80
I I
I
? I :,:
t;; ,.. I
.,
:;; 2.acl
I
!
.
:,
~
70
I I
ui
ci
£;
.,
~
2200
2000
I I ..,..
u
16
I
i
I
~
1800
I I !:; 80 I
I
16
I I
1 ~.6 • 1 • .6 5. 1 50
u ... , .... 6.1
MC By Wgt Of Agg. 'MCllyWg<0,"4111
tpcdr5c.wql
DJtc Printed: 02 -Jul- 9:< Test Propert y Curves for ICDR4F Mix
227
jUNI T WEIGHT VS. %AC J j%AV BY RICE VS. %AC!
1~
I I I I I 0 6
I I
I
u
Cl.
152
151
I I I 0
~ 5 . I I
l
I'
I
!
I I ~
u
4
I
I
I a: 3
>-
a, I I
148
I I
> 2
I I
147 I I ' 1
0
37 ~2
I
~7 5.2
3.2 37 ~2 ~7 ~2 3.2
%AC By W~ Of Agg. %AC By W~ Of Ag g.
~ ::,+ ---+----+--;--+--+~ 12
I
;,+ ---+---+--+----+-ti--+__. ~ 11 .8
I
1... 8t---+---+----tl------+----t lii< 11 .6 . I
7+---+----+--t----+---+----i ! !
st---+---.--+-<>-+---+~
5
I I
I
11 .4
11 .2
1
I
. I
I
I
~ -2 3.7 4.2
I
47
i
5.2
%AC By W~ Of Agg . %AC By W-:;t. Of Ag g.
tpicdr4f.wql
D:i lc Pr inl cd: O'.:-Ju J-9:; Test Property Curves for ICDR4C Mix
228
JUNIT WEIGHT VS . %Ac\ !%AV BY RICE VS. %AC!
15 1 5 7
151
!
' 6 .5
150.5
! I 0
6
I
0
I :i:
150 5.5
I I t;;
l: 149.5 ::. 5
149 •I I w
CJ>
!,2 4.5
I
i 148.5
! I
a:
>
m
4
148
I I > 3 .5
'
147.5 3
I I
147 2.5
146.5
I I
3 .5 " 4.5 5 ~ .5 4 4.5
%AC By Wgt Of Agg . %AC By Wgt Of Agg.
12
I I 13.6
8 13.4
E
8 11
! ~ 13.2
::Ii 13
10
w 12.8
~ •
...I
12.6
9 1ii 12.4
I <
~ 12.2
I 12
7
3.5 '
I
4 4 .5
%AC By Wgl Of Agg .
5 11.\.5
i
4 4.5
%AC By Wgt Of Agg .
5
. 2400
'I I 80
!J.
2300
I I i 75
~
.
:.; 2200
iii 2100
;;;
-!; 2000
!
II
i
'I
I
::Ii
w
u 65
If
>
m 80
70
•
•
;; I ~ 55
::.
1900
I I !; 50
18~.5
I' I ~ .5 4 4.5 5
4 4 .5 5
%AC By Wgl Of Agg. %AC By Wgt Of Agg.
tpicdr4c.wql
UJIL· Pri nt ed: :::<- Mar- 91 Test Propert y C urves for ICD R5F Mix
229
IUNIT WEIGHT VS . %AC! j%AV BY RICE VS. %AC!
151 55
150 5
i ! i •• 0
5
i
I I
I !
i
u 150
i I I' I 0
J:
4.5
• I I
4
Q.
149 5
•
I ! I I t;:;
::Ii
3.5
I I
.r:.
w ~
"'
~ , 49
I I \,l 3
a: 2 .5
~ 148 5
I I >
a,
> 2
I
T I '
'
1.5
I I
147.5
I I
1
0.5
I
3.5 4 4.5 5 5.5 6 3.5 4 4 .5 5 5.5 6
MC By Wgt Of ,t,gg. MC By Wgt Of Agg .
~ -5
T
I
4
I
4.5 5 5.5 8
11 .4
11 .~ .5 4
'
4 .5 5 5.5 6
MC By Wgt QI Agg. MC By Wgt Of Agg.
E
~
~ 3500
4000
i
i
I
I
0
0
J:
t;:;
::Ii
90
85
eo
'I
~
<ii i ~
I
w
\,l
a:
75 •
.; 3000 > 70
I I
a,
~ u 65
I I
i 2500 c(
\; T
II I 60
I i
~ -5 4 4.5 5 5.5 6 4 4.5 5 5.5
MC By Wgt Of Agg . MC By Wgt Of Agg.
tpicdr5f.wq 1
O:i te Prin ted : 08- D ec-9'.! Test Property Curves for ICDR5C Mix
230
!UNIT WEIGHT VS. %ACi !%AV BY RICE VS. %ACi
150 7
149 5
I
'
i • I
' .•
I Cl 8
~
149 I
u ' i I
I I
Q. 1.&a.5 5
I ~ I :E
~ 1'8
"' I ~ 4
~ 147.5
~ 147
•' '
if
~
::, 1'6.5 •' >
3
1'6
I I ~ 2
145.5
I
1
3.5 4 4.5 5 5.5 3.5 4 4.5 5 5.5
%AC By Wgt OI Agg. %AC By Wgt OI Agg.
~
15
. 3100 i 80
f 3000
I :E 75
T w
~ 2900
I ~
r
85
"i 2800
.
~ 2700
I, • >
m
() 80
:E 2800
! < 55
I I !i;
2500 50
i I
2 ~.5 4 4.5 5 5.5 ~ -5 4 4.5 5 5.5
%AC By Wgt OI Agg. MC By Wgt OI Agg.
tpicdr5c.wql
231
APPENDIX G
Boudreau CB 0 2 40
Boudreau CB 0 2 73
Boudreau 0 1 40
Boudreau 0 1 73
Boudreau 0 2 40
Boudreau 0 2 73
Boudreau 0 1 40
Boudreau 0 l 73
Boudreau 0 2 40
Boudreau 0 2 so
Boudreau 0 2 60
Boudreau 0 2 70
Boudreau 0 1 40
Boudreau 0 1 so
Boudreau 0 1 60
Boudreau 0 1 70
Boudreau 0 2 40
Boudreau 0 2 so
Boudreau 0 2 60
Boudreau 0 2 70
Boudreau 0 1 40
Boudreau 0 1 so
Boudreau 0 1 60
Boudreau 0 1 70
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
Schmidt & Graf G 0 9 73
237
TE
5092
.M8A3
no. 90-5
SOURI COOPERATIVE HIGHWAY RESEARCH PROGRAM
FINAL REPORT
9 0-5
- v.2
Property of
-~-...--·----··-.
I
I DETERMINATION OF AASHTO LA YER COEFFICIENTS
I VOLUME II:
I
I Prepared for
by
DAVID N. RICHARDSON
PAUL A. KREMER
I
I DEPARTMENT OF CIVIL ENGINEERING
UNIVERSITY OF MISSOURI - ROLLA
ROLLA, MISSOURI
I
I in cooperation with
U.S. DEPARTMENT OF TRANSPORTATION
FEDERAL HIGHWAY ADMINISTRATION
I December 1994
I
I The opinions, findings and conclusions expressed in this
publication are not necessarily those of the Federal
Highway Administration.
I
I
I EXECUTIVE SUMMARY
I
I The purpose of this study was to determine the layer (a) coefficients for
I several commonly specified materials used in MHTD designed pavements for use in
the 1986 AASHTO Guide pavement design method. The project entailed a review
I and compilation of published literature, laboratory testing, analysis of results, and
I preparation of this report. The study was divided into two parts. Volume I
I presented the determination of layer coefficients for several Type C and 1-C asphalt
surface mixes and plant mix bituminous bases. This Volume II presents the layer
I were shown to be a function of the resilient or static modulus for each material.
Resil ient moduli were determined for four sources of the unbound granular
I Types 1 and 2 base material (two crushed stones, two gravels) . In addition , static
Addit ional tests included sieve analyses, hydrometer analyses, specific gravities,
I Atterberg limits, aggregate particle shape/surface texture determinations, standard
I and mod ified proctor densities, and vibratory table relative densities.
modulus were bulk stress, degree of saturation, and compactive effort. Results
I
indicated that for granular materials, resilient modulus decreases with decreasing
regression model was developed for the prediction of resilient modulus of unbound
granular material which reflected the change in modulus in accordance not only
with the above variables, but with the base course position in the pavement
structure. Layer coefficients were then determined using the AASHTO Design
Guide nomograph method, and the Odemark method. The nomograph equations
(base course), and a 3 (subbase course) from resilient modulus results are
the base course and its position in the pavement structure, a sliding scale for layer I
coefficients is necessary, and use of the above-mentioned regression equation is
recommended.
Static compressive chord moduli were determined for the two soil sources in
the soil-cement portion of the study. The major testing variables were cement
models were developed for the prediction of chord modulus. Layer coefficients
were then determined from the cement stabilized base AASHTO Guide nomograph
designed using both the former MHTD method and the AASHTO method. The
layer coefficients developed in this study for unbound granular and soil-cement
I
I bases were used. This analysis verifed the use of the AASHTO nomographs for
iii
I A sensitivity analysis was also performed. The results indicated that for the
I
I
I
I
I
I
I
I
I
I
I
I
iv
TABLE OF CONTENTS
Page
EXECUTIVE SUMMARY • • • I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I I t
TABLE OF CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
GENERAL . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
OBJECTIVES AND SCOPE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
44
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
I k 1 and k 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Bulk Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
45
46
Corrected Bulk Stress . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
Resilient Modulus Regression Equation . . . . . . . . . . . . . . . . . . . 47
Granular Material k 1 and k 2 . . . . . . . . . . . . . . . . . . . . . . 50
Asphalt Modulus (E 1 ) . • . . . . . . • . . . • . . . . . . . . . . . . . 51
I Subgrade Modulus (E 5 g) . . . • . . • . . . . . . . . . . . . . . . . .
Model Statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
51
59
I Usage . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Effects of Pavement Cross-section and Material Variables . . . . . .
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
60
62
62
DETERMINATION OF LAYER COEFFICIENTS . . . . . . . . . . . . . . . . . . . 66
Matching Layer Coefficient with Layer Position . . . . . . . . . . . . . 66
AASHTO Nomographs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
I Equivalent Stiffness . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Road Test Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . .
69
69
70
NCHRP 291 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
Traylor . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
NCH RP 128 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
I KENLAYER . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Determination of MHTD Layer Coefficients . . . . . . . . . . . . . . . .
77
78
Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
CEMENT-TREATED BASE . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
INTRODUCTION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
I MATERIAL TYPES AND SOURCES . . . . . . . . . . . . . . . . . . . . . . . . . . 82
LABORATORY INVESTIGATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Soil Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
I Experimental Material Proportions and Moisture Density
Relationships . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Specimen Fabrication . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
I Unconfined Compressive Strength . . . . . . . . . . . . . . . . . . . . . . 89
Modulus of Elasticity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
RESULTS OF LABORATORY INVESTIGATION . . . . . . . . . . . . . . . . . . 90
ESTIMATION OF STATIC CHORD MODULUS . . . . . . . . . . . . . . . . . . . 98
DETERMINATION OF LAYER COEFFICIENTS . . . . . . . . . . . . . . . . . . 101
I VERIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
COMPARISON TO PUBLISHED DATA . . . . . . . . . . . . . . . . . . . . . . .
101
101
I DESIGN VERIFICATION . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Granular a 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
Granular a 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
101
104
104
I
I
vi
RECOMMENDATIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
ACKNOWLEDGEMENT . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
REFERENCES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
APPENDIX A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 130
DATA SET FOR RESILIENT MODULUS . ... . . . . . . . . . . . . . . 130
I
vii
I LIST OF FIGURES
Page
I 1.
2.
Pavement Section with Unbound Granular Base . . . . . . . . . . . . . . . .
MHTD-Middle and New Jersey Experimental Gradations . . . . . . . . . .
. 6
. 12
3. Vibratory Compaction Table . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
I 4.
5.
Resilient Modulus Testing Equipment . . . . . . . . . . . . . . . . . . . . . . .
Typical Vibratory Table Test Result . . . . . . . . . . . . . . . . . . . . . . . .
. 19
. 30
6. Moisture-Density Relationships for MHTD Middle Unbound Granular
I 7.
Materials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
Typical Resilient Modulus Test Results . . . . . . . . . . . . . . . . . . . . . . . 34
8. Relationship Between Experimentally Derived Factors (k 1 and k 2 ) for
I 9.
the Theta Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
Effect of Aggregate Source on Resilient Modulus . . . . . . . . . . . . . . . . 39
I 10.
11.
Effect of Degree of Saturation and Aggregate Source on Resilient
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
Effect of Gradation, Degree of Saturation, and Compactive Effort on
I 12.
13.
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
Definitions of Layer Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
Relationship of Subgrade Resilient Modulus and Deviator Stress for
Three Types of Subgrade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
14. Observed vs Estimated Resilient Modulus of Unbound Granular
Material . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
15. Effect of Asphalt Layer Thickness and Stiffness on Base Resilient
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
16 . Effect of Granular k 1 and Subgrade Stiffness on Base Resilient
I 17 .
Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
Effect of Base Thickness and Asphalt Layer Thickness on Base
Resilient Modulus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
I 18.
19.
AASHTO Unbound Granular Base Layer Coefficient (a 2 ) Nomograph ... 67
AASHTO Unbound Granular Subbase Layer Coefficient (a 3 )
Nomograph . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
I 20.
21 .
Average AASHO Road Test Cross-Section . . . . . . . . . . . . . . . . . . . . . 73
Soil Cement Material Moisture-Density Relationships . . . . . . . . . . . . . . 86
22. Static Chord Modulus Equipment . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
I 23.
24.
Effect of Cement Content on Unconfined Compressive Strength . . . . . . 92
Effect of Cement Content on Chord Modulus . . . . . . . . . . . . . . . . . . . 93
25. Effect of Sand Content on Chord Modulus . . . . . . . . . . . . . . . . . . . . 94
26. Relationship Between Compressive Strength and Chord Modulus . . . . . 95
27. Overall Relationship of Unconfined Compressive Strength and Static
Compressive Chord Modulus for Soil-Cement Mixtures . . . . . . . . . . . . 96
I 28. Relationship of Estimated and Observed Chord Modulus of Soil-
Cement Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
29. AASHTO Nomograph for Determination of a 2 for Cement-Treated Base 100
I
I
viii
LIST OF TABLES
Page
1. Reported Layer Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2. Range of Layer Coefficients at the AASHO Road Test . . . . . . . . . . . . . 3
3. Material Types and Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
4. Test Sequence for Granular Specimens of Base/Subbase Material . . . . . 21
5. Testing Variable Scheme . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
6. As-Received Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
7. Experimental Gradations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26
8. Gradation Shape Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
9. Experimental Gradation Slopes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
10. Specific Gravity and Moisture Density Data . . . . . . . . . . . . . . . . . . . . 28
11. Particle Shape/Texture Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
12. Atterberg Limits of the Base Materials . . . . . . . . . . . . . . . . . . . . . . . 33
13. Resilient Modulus Test Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
14.
15.
16.
Significance of Variables to Resilient Modulus . . . . . . . . . . . . . . . . . . 43
Estimated Values of k 1 . . • . . . . . • . . . • . . . • . • . . . . . . . . . . . . . . . 49
Typical Input Soil Constants for KENLAYER Analysis. . . . . . . . . . . . . . 53
I
,
17. Elastic Moduli/Equations for AASHO Road Test Materials . . . . . . . . . . 72
18. Seasonal Moduli/Bulk Stress of AASHO Road Test Materials . . . . . . . . 75
19.
20.
21.
AASHO Road Test Moduli from Traylor . . . . . . . . . . . . . . . . . . . . . . . 76
Road Test Data Input to KENLAYER . . . . . . . . . . . . . . . . . · · · · · · · · 77
Subgrade Rutting Prediction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
I
22. Backcalculation of a 2 Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
23. Comparison of AASHTO Nomograph and Odemark Stiffness Methods
of Layer Coefficient Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
24. Material Types and Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
25. Soil Classification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
26. Soil/Sand Mixture Characteristics . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
27. Soil-Cement Mixtures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
28. Compressive Strength and Chord Modulus Results . . . . . . . . . . . . . . . 97
29. Soil-Cement Layer Coefficients . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
30. Comparison of AASHTO Designs to MHTD Designs for a 2 . . . . . . . . . 105
31. Comparison of AASHTO Designs to MHTD Designs for a 3 . . . . . . . . . 106
32. Base Thickness Sensitivity to Changes in Compactive Effort. . . . . . . . 109
33. Subbase Thickness Sensitivity to Changes in Compactive Effort. . . . . 110
34. Thickness Sensitivity to Changes in Cement Content and Sand
Content . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
I
1
INTRODUCTION
GENERAL
layer coefficients for several highway materials commonly specified by the Missouri
I Highway and Transportation Department (MHTD). The study was made at the
request of the MHTD Research Advisory Committee. The project was executed by
I Engineering.
results of the AASHO Road Test. In this method, the highway designer determines
a "structural number" (SN) using such factors as design traffic level, subgrade
the SN reflects the degree to which the subgrade must be protected from the
effects of traffic. For example, a relatively high SN would indicate that a thick or
stiff pavement structure would be necessary to protect the subgrade from failing or
required thicknesses and quality of each pavement layer should be . This is done
where:
I SN = structural number
I
I
2
respectively I
D 1 ,D 2 ,D 3
drainage coefficients of the base and subbase, respectively
take into account the relative effects of pavement structure internal drainage on
A preliminary review of the literature indicates that reported values for layer
82 asphalt-treated base
cement-treated base
0.10 - 0.62
0.12 - 0.50
5-7,9
5-7,9
I
lime-treated base 0.12 - 0.26 4,6,7 I
unbound granular base 0.03 - 0.23 5,6,8-11
I
I
3
I a,
a2
0.33 1
0.12 2
0.78 1
0.23 2
0.44 1
0.14 2,0.07 3
I 1
a3 0.07 4 0.12 4 0.11 4
Examination of Eq. 1 indicates that the thickness of any particular layer is,
function of several factors. For example, the asphalt surface layer coefficient a 1 is
I thickness, traffic level, and compacted mix stiffness. For an unbound granular
base, the layer coefficients a 2 and a 3 have been shown to be dependent on the
As originally used in the AASHO Road Test analysis, layer coefficients were
actually regression coefficients which were the result of relating layer thicknesses
I to road performance under the conditions of the Road Test. The problem is to
translate the Road Test findings to other geographic areas where the construction
I
4
materials and climate are different. Layer coefficients must be determined in order I
to use Eq. 1 for design purposes. In a pure sense, layer coefficients are abstract
basis of relative layer material strength or stiffness considerations. Over the years
since the AASHO Road Test, many methods have been used to determine values
economy, accuracy, and length of study. In brief, the study entails determination I
of stiffness values for several commonly used MHTD types of pavement materials.
The stiffness values were determined by both direct laboratory modulus testing
structures designed with the AASHTO method to structures designed with the
former MHTD method. The report includes a method suitable for use in routine
design which will enable the pavement designer to solve Eq. 1 and hence obtain
The approach taken for layer coefficient determination was the traditional I
one (5, 12), which is to take some measure of strength, stability, or stiffness of a
I
particular mix or blend and compare it to the same type of measure for the
counterpart AASHO Road Test material. The comparisons are usually done by use
of one of the AASHTO Design Guide charts or some form of a ratio of the two
I In this study, the materials for which layer coefficients were determined
were limited to two types of asphalt surface mixes (Types C and IC), one type of
I bituminous base mix, two types of cement-treated base mixes, and one type of
Volume I( 12) covers asphaltic materials; Volume II deals with unbound granular
INTRODUCTION
The main thrust of this portion of the study was to determine base and
which are used as base material under an asphaltic surface layer or as subbase
under asphaltic surface and base layers. Fig. 1 shows the pavement section being
I modeled. The analysis was based on repeated-load triaxial resilient modulus tests
I modulus.
I
I
13.57 in. I ()
4.Sk 4.Sk
p = 100 psi - - -
AsphGl t
Bound
LGyer D1 El AJ 1
l 'nbound
I t
Gro.nulo.r
L0yer
It> \.J n \.J <'> \...! <'> .._., C-"'> '..J n \.J n '-' <'> .._., r> \.J C-"'> '-' n \.J n
D2
E2 /u 2
Subgro.de E3 AJ 3
given impetus to the effort by the pavement industry to produce a type of modulus
I 1) Direct use of the base and subbase layer nomographs in the 1986 AASHTO
guide ( 1). The necessary input is the resilient modulus of the granular
I modulus) of MHTD aggregate types was related to the modulus of Road Test
material at various stress states in a manner similar to that done with the
RESILIENT MODULUS
As can be seen from the above, the resilient modulus (E 9 ) is the key
parameter. The factors that most significantly affect unbound granular base
factor in regard to material stiffness. The higher the confining pressure, the higher
the modulus. The significance of the other parameters seems less well defined.
( 10, 18) . It appears that, as fines content increases, resilient modulus decreases
( 11, 13, 14, 19). Perhaps with higher fines content, the influence of a high degree
( 13) . Thus, there may be an interaction between the amount of fines and their
moisture content. This may be the reason that several studies have shown that
counterparts. However, several studies have shown the reverse( 13,20), or at least
that gradation is of negligible significance (17), and others have shown that there
exhibits a higher modulus (14,20), although, in some cases, the reverse seems to
minor significance ( 14, 20) and of major significance( 14). Thus the influence and
I
I 9
modulus were studied in this project. These include: the effect of stress state,
I gradation was not in the scope of this study. However, in the companion project
(2), a different gradation was used for the same materials which were used in the
present study. The results of modulus testing from both projects were examined
I jointly. Thus a study of the effect of gradation was accomplished and is reported
herein.
materials were selected and sampled by MHTD personnel. Two Type 1 crushed
stone base aggregates were studied. They were selected by MHTD personnel to
give a wide range of particle shape and texture. Additionally, in the companion
project (2), two Type 2 gravel materials were tested for resilient modulus. Test
I results for these two gravel materials are also included in this report. The
DR-15 Type 2 Black River gravel base Williamsville Stone Poplar Bluff
LABORATORY INVESTIGATION
General
I
The principal property determined for unbound granular base materials was
the resilient modulus. However, the performance of other tests and procedures
was necessary in order to ultimately conduct the resilient modulus testing and
analyze the results. These other procedures included sieve analyses, gradation
Gradations
curve situated midway between the upper and lower limits of the allowable
gradation specification band for MHTD Type 1 unbound base material. This
gradation was used for both the two crushed stones and the two gravels. At the
finer size end, the gradation was extended to include 8% passing the #200 sieve.
This value was chosen because 1) it matched one of the gradations used in the
drainage study which is the companion project to the present study, and 2) this
Type 1 aggregates supplied by MHTD for this project. The gradation plot is termed
open gradation (the so-called New Jersey gradation) was also tested for resilient
I 11
I modulus. The results of that testing is also herein reported. The New Jersey
resilient modulus. The most promising methods were later tried in the
to characterize the gradations so that a single parameter would represent the shape
and position of each gradation curve. Nine different methods to determine the
I gradation parameter were tried. These are described in detail in Volume I of this
Particle Shape/Texture
Numerous test methods have been devised to quantify particle shape and/or
J texture. These can be divided into two catagories: 1) direct methods (those that
indirect methods (those that measure some sort of bulk aggregate property, such
these methods were reported by Kandhal et al. (24) at NCAT (National Center for
Asphalt Technology). From the literature, it appears that efforts are being
concentrated in the area of fine aggregate evaluation and that there are several
methods available which can be used in lieu of the standard test, ASTM D 3398
(25) which is somewhat cumbersome to perform. Kandhal .fil fil. recommended the
I
100 ,,
,.
90 /I
,,.,
1
di
....-..
80
70
lY
T'\
O.A . J.J.J.l. . J.J.J.
! J. -
LJ ~J.J. C)J.\, 1
r,
T ! -
J..JJ.J.J. ~ l
I' l
·I '
I
I
, -
I
/
I ~
-
[)!~y
' .
Tl::::"Tr.:,rc;-v
.T~ ·r=
T:iT IJ.~
I
1\/ 1 IP // l -'
~ .. I / = ---.......----.-----~
..__, /1 /
(~ /
/
60 ,, /
tlJ) / .,
-,, /
.....C #/
/
/
[l'.l
[l'.l
50 "' /
/" 1
cx:l /I
/, I
0...
, ," .. I
40 ,
.-I / I
,
....,a:! [ l
/
I
I
0 30 /
, I
I
E-<
I
"' I
I
,-( I
y /
20
I
I
I
I I
,0 I
I
I
/
chn·'
,.
0
200 50 so Ie B 4 :1/B In S/4 in I In 1. 75 in :I in
100 1/2 in
Sieve Sizes
0 Middle
(0.45 Power)
D New Jersey
I
13
aggregate (26). Both the standard ASTM method and the proposed NAA method
I In this study the (-) #8 to ( +) #100 sieve size fraction of each aggregate
mixture blend was tested using the NAA Method A. The methodology is given in
Appendix A of Volume I of this study. For the ( +) #4 size, the blends were tested
Volume I of this study. The results of both methods were used in developing the
this report. Photographs of the NAA test device and the D 3398 equipment are
Specific Gravity
I and #100 sieve sizes and tested in accordance with AASHTO T85-88 (27) and
T84-88 (28) for the ( + )#4 material and the (-)#4 to ( + )#100 material,
respectively. These data were necessary for use in the degree of saturation
calculate the specific gravity for each gradation of each of the four aggregates as
follows:
100
G = --,------------ (3)
I % Passing #4 + % Retained #4 · · · · · · · · · · ·
ASG ASG
where:
Screening
All aggregates were shaken in an air dry state through the appropriate
Upon shaking, the split material was stored in 20 gal plastic cans with lids
specimens, standard and mod ified proctor tests were performed in accordance
with AASHTO T-99 (29) and T-180 (30). Additionally, the maximum density of
the open-graded gradation was determined via the vibratory table method (ASTM
D4253) (31) . For each of the four aggregates, a double amplitude vs dry density
curve was obtained in accordance with the dry method to obtain the optimum
power setting. This power setting was then used for the determination of density
Relative Density
density (32) of all Eg test specimens. The equation to calculate relative density is:
15
I
I
18
accordance with Claros fil fil. (35), a 1 /a3 ratios were not allowed to exceed three
acquisition system. Load was measured with an internal 1000 lb capacity load cell
and deformation was measured with two LVDT'S mounted externally to the cell.
and the load cell were in accordance with the AASHTO specifications. Actual
minimum deformations and loads during the testing were kept at least ten times
the minimum resolutions to assure confidence in the test results. Air was used as
the confining fluid instead of water in order to protect the internal load cell.
Triaxial cell pressure and back pressure were controlled via a Geotest control panel.
The Research Engineering triaxial cell that was used has several advantages. First,
the chamber cylinder wall can be placed after the loading piston is brought into
contact with the specimen. Also, end caps can be purged of air very easily by the (
shear strength, the more confined a granular material is, the higher will be the
modulus . In the field , confinement is supplied by: 1) the layer underneath the
granular material , 2) the granular material itself in the lateral (tangential and radial)
(a 2 and a 3 ) direction, 3) the overburden above the point of interest, and 4) the
momentary load from a vehicle. In a triaxial test, the difference between ·total
vertical stress (a 1 ) and the confining pressure (a3 ) is called the deviator stress or
stress difference (ad) . Cell pressure supplies the lateral confinement to the
specimen (a2 and a 3 ). A small static load (0.1 ad) supplies the "overburden"
pressure, and cyclic deviator stress (0.9 ad) supplies the "vehicle" momentary
stress. All of the stresses combined are known as the bulk stress .
0 = a1 + a2 + a3 where a 2 = a3 and a 1 = ad + a3
= (ad + a 3) + 2a3
= ad + 3a3
was determined at 15 stress states where effective confining pressure ranged from
2 to 20 psi and ad varied from 2 to 40 psi. This resulted in a range of bulk stress
from 8 to 100 psi. This was considered adequate to cover the range of stress
shown later , the weighted average 0 in the crushed stone base at the AAS HO
Road Test was about 14 psi. The testing sequence and stress state schedule is
shown in Table 4.
21
Road Test granular base materials suffered a marked increase in distress above 85
90 percent saturation (22). In the present study, each material was tested at two
with higher levels of compaction. Two levels of compactive effort were evaluated
for each material and gradation. For the dense gradation, specimens were
compacted to 100% standard and 100% modified proctor densities. For the New
density via vibratory compaction (wet method), while the lower level of density
proctor.
shape/surface texture is not well-defined. Two crushed stones and two gravels
from different geological formations were chosen for delineating the effect of
Testing Scheme. The testing scheme involved the following variables: two
sources of two particle shapes, two compactive efforts, two gradations and two
degrees of saturation for a total of 32 "tests". Each test was run with duplicate
CEL 0 S = 60 X X X X X X X X
0 S = 100 X X X X X X X X
CEH 0 S = 60 X X X X X X X X
os = 100 X X X X X X X X
Note: Mid. = middle of MHTD Type 1 gradation band
NJ = New Jersey gradation
CEL = lower compactive effort
CEH = higher compactive effort
0 S = 60 or 100% saturated
Test Procedure. The resilient modulus testing procedure involved the following
After the load application in the 100% saturation step, the specimen was tested
for permeability. The permeability results were used in the drainage project that
was concurrent with the layer coefficient project. As a final step, the specimens
hand-held air hammer. The material was compacted at about 60% saturation into
a split mold. After cell assembly and consolidation, the specimen was conditioned
24
with 1000 repetitions. Conditioning is used to eliminate the effects of any
minimizing the effects of initially imperfect contact between the end platens and
the test specimen. The various stress states and loads were then applied in
accordance with Table 4. The number of load applications varied from 50 to 200,
repeatability requirements.
Load and deformation data were taken for every load application over the
entire sequence, but only the last five repetitions were used for calculation of
resilient modulus.
The load duration for each repetition was 0.1 sec followed by 0.9 sec rest.
The stress pulse shape was haversine in nature. Repeated load equipment
deflection was determined through the use of an aluminum dummy specimen and
was subtracted from total deflections for each stress state. Initially, calibration of
the load cell and LVDT's was performed before each test, but the interval was
increased upon determining that the drift in calibration was insignificant. The
tests at 60% saturation were performed in a drained condition while the 100%
As-Received Gradations
The as-received gradations of the four granular materials are shown in Table
I
I
26
Table 7. Experimental Gradations.
different methods were explored. Sieve size data from both experimental gradation
curve types (Middle and NJ) were used for calculation of various parameters. The
parameters were then used in the development of the multiple regression model of
resilient modulus to see which method increased the accuracy of the model the
most. This was judged from the adjusted - R2 statistic of the equation. The nine
I 27
uniformity (Cu), coefficient of skew (Cz), surface fineness (SF), specific surface
-
factor (SSF), a combination of SF and SFF (SF/SSF), Hudson's A, slopes-of-
for these parameters are shown in Tables 8 and 9. The parameters are discussed
from that which was in Volume I to better match the natural break points of the
granular base experimental gradation curves. The slopes of each curve were
determined between the 1 in and #4 sieve, the #4 and #16 sieves, and the #16
Mid. NJ
FM 4.53 5.66
cu 82.6 5.29
CZ 1.19 0.21
SF 1588 1938
SSF 294.4 65.8
SF/SSF 5.40 29.4
- 4.55 3.36
A
Mn-n ----- -----
3/4, #4, 16, 200 ----- -----
The results of the regression model analysis indicated that it made very little
28
difference as to which gradation parameter was used (except A was somewhat
less effective, and no more than one "M" could be used at a time due to
collinearity problems).
determined in regard to the two test gradations for each of the four granular
I
DR- 12 MHTD M id 2 .69
DR - 14
NJ
MHTD Mid
2 .78
2.65
100.8
109 .9
..
132.5
--
7.8
135.2
134.5
8 .3
6.7
124. 1
..
13 .0
..
I
DR - 15
NJ
MHTD Mid
2.65
2.65
99.9
110.2
119 .0
134.4
8.0
7 .6
..
136.9
..
6.1
122 .8
..
12 .0
..
I
.. ..
I
NJ 2 .65 96.2 109 .5" • 8 .9 114.3 14.0
density tests for certain materials when graded according to the New Jersey
gradation, therefore, the vibratory table density (wet) method was also performed.
It is believed that this compaction method gives a density value more comparable
to that which will be achieved in the field. The tests were performed in a dry state
at different power settings to determine the optimum setting which would result in
the highest density. The test was then run with the material in a moisture state
which was more in line with field compaction conditions. A typical vibratory table
Relative Density
Minimum density test results are also listed in Table 10. The minimum and
maximum density values were used in Eq. 4 to calculate the relative density of
each E9 specimen, as listed later in Table 13. Relative density values give a wider
I spread of values than "percent of T-180 values", and therefore are more useful in
I analysis of behavior.
3398 for the ( +) #4 sieve material and by NAA Method A for the (-) #8 through
( + l #100 material for each aggregate source. Both are measures of void content
of bulk aggregate. Void content has been shown to be related to particle
shape/texture. D3398 results in a "Particle Index" {IP) while the NAA Method A
30
123
122
--
-
C')
~ 0
........__ 12 1
..... Wet
..0
.....
..__, 120 0
~
..c:
·-;,::
tll)
( l)
1 19 -
·- ~
i::
;:J
1 18
>. 1 17
i...
Q
1 15
0.008 0.010 0 . 012 0.014 0.016 0.018
-.... 1 40
T180
~140
~
t".)...,
...,
....
-
.........
..0
...J
..., 135
-
.........
..0
.:::. 135
TIIII
..c::
Oil
Q)
Q)
;a,: 1 30 )I:: 130
...,
.... ...,
c:: ....
I ::i
0
t' 125
Sample:
Optimum
Optimum
Max. Dry
DR-12 Middle
H2 0 Content: ? .Olt (T99)
H 2 0 Content: ?.O: (T\80)
Unit 'Wt.: 138.6 Lbt/tt 3 (T99) 0
Cl
::i
>. 125
I-.
Sample:
Optimum
Optimum
DR-13 Middle
H2 0 Content: 7 . 7'% (T99)
H2 0 Content: 5 .9% (T\BO)
Ma:,:. Dry Unit 'Wt .: 136.3 Lb! / !t 3 (T99)
I 120
5
Max. Dry
6
Unit Wt .: 13?.8 Lb!/!t (T180)
7 8 9
120
4
Mu. Dry
5
Unit 'Wt. : 141.0 Lb! / ft (TlBO)
6 7 B 9 10
Water Content (%)
Water Content (%)
I
145
-- 140
Optimum H 2 0 Content: 7 .8 (T99)
Opt imum H 2 0 Content: 6 .7 (T1BO)
-- 140
I
M..., M_,
.........
......0 -
.........
...J
'--'135 -
..., -..c
...J
..., 135
I ..c::
....
bO
Q)
T180 ..c::
....
QI)
Q)
I 0
Ma:,:. Dry Unit ll't.:
Max . Dry Unit ll't .:
s
132.5 Lbf/ft 3 (T99)
134.5 Lbf/ft (T160)
Q Optimum
Mu. Dry
Max. Dry
I
H 2 0 Content: 6 . 1 (Tl\O)
Unit wt .: 13'.4 Lbf/ ft 3 (T99)
Unit Wt.: 136.9 Lbt/ ft (TISO)
120 12 0
0.0 0.1 0.2 4 5 6 7 8 9
Water Content (%) Water Content (%)
gives an "Uncompacted Voids Percent" (U). The results are shown in Table 11.
Round, smooth particles give IP's of 6 or 7, while angular, rough particles result in
values of more than 15. The range of IP's of the aggregates in this study was
10.0 to 12. 5. The Particle Index was determined for the coarse aggregate fraction
of each gradation and the Uncompacted Voids content was determined for the fine
aggregate fraction.
Looking at Particle Index and especially the Uncompacted Voids values, the
crushed aggregates were somewhat more angular than the gravels, as expected,
Plasticity of Fines
The results of the Atterberg Limits testing are shown in Table 12. All four
Resilient Modulus
(DR-12 and DR-13) and two gravels (DR-14 and DR-15) at two degrees of
saturation (approximately 60 and 100%), two compactive efforts (low and high),
two gradations (MHTD "Middle" and New Jersey), and 14 stress states with
duplicate samples, for a total of 896 tests. Because the same 14 stress states and
the two degrees of saturation were used for each specimen, there were only 32
I specimens. The testing sequence for each specimen is shown in Table 5. Fig. 7 is
I a typical plot of bulk stress vs resilient modulus. Each data point is representative
of one stress state. As can be seen, modulus increases with an increase in stress
state. The classic equation of the line (known as the "theta model") is:
I or
34
+J
~
-
....
Q,)
....
I'll
Q,)
~
1 10 100
I Eg = k1 ek2 . . . . . . . . . . . . • . . . . . . . (6)
I where:
I The results of all resilient modulus testing are tabulated in Table 13.
I seen, the present study data falls in the range of data that has been reported
elsewhere.
Fig. 9 shows the effect of aggregate source on Eg. It appears that in the
effect of aggregate source on Eg. This may be a result of the limited range of
I curve slopes. The percent loss for the dense graded materials with low and high
compactive efforts was 9.8 and 11 .0, while the loss for the open graded materials
I
I
Table 13 . Resilient Modulus Test Data.
DR-12 Mid (low) 136.5 136.4 138.9 98 .2 92 . 1 61.4 3040 0.85 21,522
DR-12 Mid (low) 136.5 136.4 138.9 98 .2 92 .1 100 2958 0.86 21,428
DR-12 Mid (high) 137.6 138.4 138.9 99 .6 98.4 62.5 3828 0.82 25,295
DR-12 Mid (high) 137 .6 138.4 138.9 99 .6 98 .4 100 3758 0.80 23,437
DR-12 NJ (low) 121 .1 120.5 131 .2 91 .8 69 .3 59 .6 4307 0.72 22,865
DR-12 NJ (low) 121 .1 120.5 131 .2 91 .8 69 .3 100 1940 0.90 15,584
DR-12 NJ (high) 131.2 127.7 131.2 97.3 90 .5 53 .1 5134 0.74 27,890
DR-12 NJ (high) 131 .2 127.7 131 .2 97.3 90 .5 100 2706 0.92 22,508
DR-13 Mid (low) 138.3 138.2 141.0 98.0 91.3 63.8 4212 0 .76 24,238
DR-13 Mid (low) 138.3 138.2 141.0 98 .0 91.3 100 3606 0.80 22,752
DR-13 Mid (high) 141.0 139.5 141 .0 98 .9 95 .2 59.8 8312 0.58 31,967
DR-13 Mid (high) 141.0 139.5 141.0 98 .9 95.2 100 5918 0 .62 24,390
DR-13 NJ (low) 124.1 125.9 135.2 93.1 78 .3 63 .8 3470 0 .81 22,407
DR-13 NJ (low) 124.1 125.9 135 .2 93 . 1 78 .3 100 2824 0.82 18,658
DR-13 NJ (high) 135.2 134. 1 135.2 99 .2 97 .5 58.8 3164 0.90 25,428
DR-13 NJ (high) 135.2 134. 1 135 .2 99.2 97.5 100 2997 0.86 21,963
DR-14 Mid (low) 132.5 131 . 1 135.4 96 .8 85 .9 58 .6 4443 0.60 17,688
DR-14 Mid (low) 132.5 131 . 1 135.4 96.8 85.9 100 3401 0.68 16,278
DR-14 Mid (high) 134.5 134.4 135.4 99.3 96 .8 67.2 5468 0.64 23,869
DR-14 Mid (high) 134.5 134.4 135 .4 99 .3 96 .8 100 5793 0.67 27,096
DR-14 NJ (low) 122.8 121 .1 125.8 96.2 84 .7 58 .2 6618 0 .56 24,309
DR-14 NJ (low) 122 .8 121 .1 125 .8 96 .2 84.7 100 4504 0 .72 23,369 w
0)
DR -15 Mid (low) 134.4 131 .5 136.9 96 .0 83 .1 100 2645 0 .69 12,955
DR - 15 Mid (high) 136.9 135 .4 136 .9 98 .9 95.2 59.7 4498 0.65 20,090
DR-15 Mid (high) 136.9 135 .4 136.9 98 .7 94 .5 100 2702 0.75 15, 194
DR-15 NJ (high) 114.3 115.8 115.8 100 100 100 1338 0 .96 12,203
CE = Compactive effort
MADD = Maximum Attainable Dry Density
DR = Relative Density
*E 9 = k 1 ok 2
w
-....,J
38
...-...
.....
rn
0..
I
'-"
.....
~ 100
'
I ~ o©
O ~ 00
~ % 0
(f)~~O~o
0@ 0
0
10 0
0
I
I
1
0.0 0.2 0.4 0.8 1 .0 1 .2
I
10 5
I 0 DR12
I • DR13
V DR14
I ... DR15
..........
.....
I Cll
p.
.__...
Cll
I ;::J
......
;::J
'C
0 104
::;.g
I ~
i:1
Q)
.....
......
I .....
f/l
Q)
~
I
I
I 10 3 1..-~~~~~~--'~~~~~~~_.._~~~~~~~~
0 .1 10 100
I Bulk Stress, 8 (psi)
I
I
I
I
40
C"J~ 25
0
><
·~v.i
o.. 20
V
15
0
•
• DR14M
10 a DR13NJ
T DR12M
6 DR13M
• DR14NJ
5
30 40 50 60 70 80 90 100 11 0 120
Degree of Saturation (%)
was 17. 7 and 19.5, respectively. However, the effect on Eg of providing a drained
I base can also be seen from Fig . 11 by looking at the dashed lines. For both the
I low and high compactive effort cases, there is a benefit from changing from a
I dense graded material which will remain saturated for extended periods of time to
an open graded material which will remain in a drained state most of the time.
11 Additionally, the effect of compactive effort is apparent. For both the dense
I and open graded materials, an increase in compactive effort from close to 100% T-
independent variables on Eg. Paired-t tests were performed to see if there was a
I significant difference between the means of all Eg data of 1) a low degree vs a
high degree of saturation, 2) a low compactive effort vs a high one, and 3) a dense
I results and, if so, which source(s) were significantly different. The results are
shown in Table 14. As can be seen, degree of saturation, compactive effort, and
I aggregate source were significant to differences in Eg at the 0.05 level, and the
increasing the saturation from 60 to 100%, reducing the level of compaction from
about 100% T-180 to 100% T-99 density, and having a saturated, dense graded
I
I
42
30
-
a)
0
.....
25
><
·~ Ill
c..
'-"
Ill
- ::l
::l
'tj
0
20
:::s
,..J
A
-
Cl)
·~
.~Ill o Mid - High CE
QJ
0:: • NJ - High CE
15
v Mid - Low CE
"' NJ - Low CE
8 - 10 psi
Low High
E0 at 8 = 10 psi (psi)
Condition Maximum Minimum Difference Significance
at 0.05 level
I All Mixtures :
Saturation, low vs
0 22,706 19,452 3254 yes
I high
Gradation , dense vs 21,634 20,562 1072 no
I open
Comp . effort, high vs 22,679 19,518 3161 yes
low
I Agg. Source,
* DR12vsDR15 22,541 15,585 6956 yes
I DR 13 vs DR 15 23,975 15,585 8390 yes
DR 14 vs DR 15 22,330 15,585 6745 yes
Gradation and 22,704 20,442 2262 yes at 0 .088
saturation: open level
I graded drained vs
dense graded
undra ined
I * All other combinations: not significantly different
I source, the DR-15 had a significantly lower modulus than the other three sources
(which were all about the same). However, this does not appear to be a function
I of particle shape or plasticity of fines, because the U and IP values of the DR15
were close to those of the DR14, and the plasticity Index all four sources was
about the same. A possible explanation is that the as-tested density of the DR- 15
I specimens was somewhat low compared to the target density.
I
I
44
ESTIMATION OF RESILIENT MODULUS
Introduction
material are found from a relationship with resilient modulus, Eg. The resilient
modulus for a given material can be found by test using the "theta model", or by
I
use of an estimation regression equation. The development of the regression
where :
(} = bulk stress = a 1 + a 2 + a3
a1 = major principal stress
= ax +
I a2
a3 = av +
k 0 (v 1z 1
k0
+ V2Z2)
(v 1z 1 + v2 z2 )
I ax = horizontal stress in x-direction induced by wheel load, psi
testing.
I Thus, in a pavement structure:
k 2 , and 8 are necessary . The following discussion shows the methods by which
I k1 and k 2
were developed by use of the statistical software package SYSTAT (36) from the
I
I
46
test data produced in the present study. Numerous linear multiple regression
models were developed and analyzed to estimate k 1 from certain test data. Many
from Tables 7, 9, and 10. The equation of best fit had a low degree of accuracy
and is not shown here, but it was helpful in indicating trends and for estimation of
k 1 values to set up boundary conditions for other models in the study. A second
Thus, k 1 can be estimated from physical properties of the aggregate (as shown
Bulk stress in an unbound granular base layer is a function of applied load (P)
and contact pressure (p), stiffness (E 1 ) and thickness (D 1 ) of the overlying asphalt-
bound layer, stiffness (E 5 gl of the subgrade underlying the base layer, stiffness (E 9 )
and thickness (D 2 ) of the base layer, and unit weight of the overlying layers .
A load of 9000 lbs (one half of an 18k axle load) and a contact pressure of
I
100 psi (average U.S. truck tire pressure) was used.
I
I 47
pavement layers.
I stresses.
I computed.
I with 8 was not a separate process; rather, it was integral to determining the Eg
I was assigned.
I
I
I
48 I
I
A predictive equation of resilient modulus could be constructed from Eg test
data which reflected the effect of saturation, density, and bulk stress, 8.
I
Unfortunately, 8 is not normally known. Results of numerous runs of KENLAYER I
indicated that 8 could typically vary from 3 to 33 psi. This is too wide a variance
I
to merely assume a value for 8. Bulk stress could be calculated for a variety of
unavoidable.
I
In the formulation of Eq. 9, Witczak and Smith assumed a constant value for
k 2 . However, it has been shown in the present study and others (22) that k 2
substantial drop in k 1 and an increase in k2 (22). This was verified in the present
I
I
I 49
I approximated as shown in Table 15, which is based on the results of the present
study and a previous study (22). It must be emphasized that k 1 values vary
I considerably within an aggregate source class, and it is quite possible for a given
I Material SCE
k,
MCE
(100%) (100%)
I Gravel 4300 5000
I Crushed stone
Note:
4800 6100
I
I The modulus calculated by use of Eq. 10 will approximate the modulus from
Eq. 6 that has been reconciled with 8. The other terms in Eq. 10 are necessary to
I to the base layer, hence E9 decreases. As E59 decreases, the base layer is less
confined under loading, hence E9 decreases. Note that Eq. 10 can only be used for
I a single granular layer sandwiched between an asphalt layer and the subgrade. For
I more complicated sections, use of programs such as KEN LA YER are recommended
I
I
I
50
I
(Eg = k1e"z). In the regression equation development, resilient modulus was varied
I
by use of 237 combinations of k 1 , k 2 , and 9 in the program KENLAYER. These
D2 = 4, 12, 18 in
The following paragraphs show how the above values were determined.
three levels:
1) 60% saturation; 95% SCE; dense gradation with 35% minus #16
fines
For each of the 237 combinations, KENLAYER calculated the bulk stress 9 in I
I
I the granular base and the deviator stress ad in the subgrade soil from an applied
51
I load. The different cross-sections represented by the model are shown in Fig. 12.
Asphalt Modulus (E 11. The asphalt layer was characterized as linearly elastic, while
I the granular base and subgrade materials were characterized as non-linearly elastic.
I For the asphalt layer, three conditions of stiffness were characterized using the
I relationship of temperature and resilient modulus for MHTD 1990 mix designs as
I choosing the stiffest 1990 mix at the coldest individual Missouri weather station's
smallest asphalt modulus utilized the least stiff 1990 mix at the warmest individual
I weather station's average monthly temperature (T P = 92.3 °F). A middle value for
I modulus was found by using a 1990 mix of average stiffness with the overall
average monthly pavement temperature for all Missouri weather reporting stations
I +Tp = 64.2°F). This resulted in asphalt mixture moduli of 2,100,000; 500,000,
I and 130,000 psi. Pavement temperatures were calculated from air temperatures
type, degree of saturation, compacted density, and state of stress within the
I pavement structure. In KENLAYER, the soil modulus was characterized as being
I very soft, medium, or stiff. The soils were described as shown in Fig. 13. As can
I be seen, the curves of each of these three soils has the same general shape and
I
I
o.,, De Surface Mi x ture (e.g. Type C or [ C)
a.• , D' 52
n 2 = l.O Bitur-iinous Bose
Subgrode
a . Two Lo. yer Full Depth Aspho.l t Sect ion.
Subgr ode
DG Bo.se
Sub gro.de
Surf'o.ce Mixture
BituMinous Bose
a.•' D,
n 2
= 1.0
n , . a.,, D, DG Bose
I equal line slopes (K 3 and K4 ). The parameters that distinguish one soil's
consistency from another are the maximum and minimum moduli (boundary
I conditions on a possible spectrum of stiffness) and K 1 , except for the very soft
Input values to describe the three soils are shown in Table 16.
·I
Table 16. Typical Input Soil Constants for KENLAYER Analysis.
I
Soil K, K2 K3 K4 E5 g (max) E5 g (min)
I Consistency
very soft
(psi)
1000
(psi)
6.2 1110 0
(psi)
5662
(psi)
1000
I medium
stiff
7680
12,340
6.2
6.2
1110
1110
178
178
12,342
17,002
4716
7605
I From Fig. 13 it is seen that E5 g is also a function of stress state. Thus, there is an
I interaction between the stress transmitted to the base and to the subgrade, with
the modulus of both materials fluctuating with stress state. KENLAYER performs
I numerous iterations to reconcile the base, subase, and subgrade moduli and stress
states.
Table 16 is based on work by Thompson and Robnett (39). Note that with
the exception of very soft soils, the slopes of the lines in Fig . 13 are all the same.
The most significant variable is K 1 . K 1 is the input in KENLAYER which sets the
I curve position. KENLAYER computes deviator stress, ad, and E59 is thus
determined by moving along the curve in accordance with the point where ad
I
I
54
-
t'l
0
~
18
><
•.-4 16
- rn
Pot
r.:I
tlG
ll
14
Kl - 12,340
.
I'll
12
::::l
...-4
::::l 10 = Stiff
'O
0
~ 7630
+> 8
i:::: - 178 7605
•.-4
...-4
•.-4
Cl>
Cl>
rn
6 Medium
4716
I
~
4
'O
Cl>
Kl - 1000
cd
s.. 2 K2 - 6.2 Very Soft
.c
U)
bO
::::l 0
1000 I
0 5 10 15 20 25 30 35
Repeated Deviator Stress, ad (psi)
I K1 = 3.63+0.1239(PcLAv)+0.4792(Pl)+0.0031(Ps,LT)-0.3361(GI) · .. (11)
where:
I = resilient modulus of soil at ad = 6.2 psi, ksi
Pl = plasticity index
= + +
I (P 200 -35)(0.2 0.005 (LL-40))
=
I LL liquid limit.
I can be estimated. Again, K2 , K3 , and K4 are as shown in Table 15 for any fine-
grained soil.
The K 1 equation is based on a dry density equal to 95% standard proctor (T-
I 99) maximum and at optimum moisture content (OMC). For an increase in density
I to 100%, an increase of about 1.4 ksi is suggested (39). For densities between
from 95% to 100% T-99 maximum density on the dry side of OMC. If the in- I
service moisture content will be greater than OMC, the use of the Denscor should
I
be omitted.
MC,s
Satsvc = _ _ _6_2_.4_ _ _ _1__
. . . . . . . . . . . (14)
(Pcom,/1 OOXMDD) sp.grsv.
where moisture is in "%". Note the "dry density" may be different in Satsvc
I
I
57
I and SatoMc·
moisture contents above OMC. However, on the dry side of OMC, application of
I the correction is limited to only, say, zero to 2% below OMC. If the in-service
content of clays generally exceeds their plastic limits. The moisture content of
silty soils are equal to or just under their plastic limits, and the moisture content of
sandy loams are less than their plastic limits. Plastic limits are generally higher
I than OMC values. A large proportion of all fine-grained soils exhibit in-service
downward):
I This K 1 (corr) would then be input into KENLAYER. From the pavement cross-
I section and material moduli, KENLAYER will compute the deviator stress in the
subgrade, and knowing K 1 (corr), it will compute the resilient modulus of the
I
58
subgrade.
At some point during routine design, the E5 g must be determined, but the
ad. Several options for estimation of ad are open. In the performance of the 237
values were noted: 2.1 psi minimum , 12.2 psi maximum, and 5.1 psi average.
The 5.1 psi average ad is less than 6.2 (the "knee" in Fig. 13), thus the E5 g is
situated on the steep-sloped portion of the curve (Fig. 13) and E5 g would be
greater than K 1 • On the other hand, by use of ad max (12.2 psi), E5 g would be less
than K 1 .
where:
suggests that it is not even appropriate. Conversely, use of ad max = 12.2 psi
may be unduly conservative. Thus, it is suggested here that a value of 6.2 psi be
I used and thus E5 g should be set at the K 1 (corrected) value . This would be
considered the "normal" condition. The 1986 AASHTO Guide recommends that a
freezing, and length of season. This final weighted average Esg ("effective res ilient
regression model was developed by use of SYSTAT. The criteria for model
models that met all the below listed criteria. This statistic reflects the
equation will predict a population of data, not the sample data . Thus
it is usually a little lower than the R2 value, which is for the sample
excessively.
an asphalt surface layer and a bituminous base layer. If the latter is the case, E1
should represent a combination of the two asphalt bound layers (Eeql as follows:
where:
I
I
I
I -
(')
~
0 40
><
..... 35
I'll
A
.......
O'l
;:j
30
......
;:j
I re
::il
0 25
~
~ 20
.....
(L)
......
.....
O'l
I 0::
CL)
reCL)
15
10
I >
J.,
CL)
I'll
..c 5
0
0
0 10 20 30 40 50
I
62
temperature.
15).
As can be seen from Fig. 15, Eg increases with a decrease in asphalt layer
thickness or stiffness. This is because higher stress is transferred to the base layer
which increases Eg. Fig. 16 shows that as E5 g increases, bulk stress in the base
thus Eg increases. And in Fig. 17 it is seen that an increase in base thickness has
Summary
I
63
I
25 ---------------------------------------
Medium Subgrade
kl 3000 k2 = 0.653
D 2 = 4 in
-- 20
E 1 = 130,000 psi
-
t')
0
= 500,000 psi
><
......
P..
fl)
= 2,100,000 psi
........... 15
Cl.I
::l
......
::l
'"d
0
:::g 10
..,.)
~
(l)
......
......
......
Cl.I
(l)
0:: 5
I O · ..___,.......,__,__,i..-__,...1-__,--1.__,__,....__ _--.1.____.i.....__....1
0 2 4 6 8 10 12 14 16
D 1 (in)
I
64
Stiff Subgrade
20 Medium Subgrade
Very Soft Subgrade
,-..._
t'j
0
.......
>< 15
.....
rt.l
0..
..__,,
rt.l
;:1
...... D 1 = 8 in
;:1 10
't:l
0
:::s
+J
i::
.....
Q.)
D2 12 in
......
..... 5
rt.l El 500,000 psi
Q.)
0:::
o---------------------
0 2 4 6 8 10 12
3
k 1 (psi X 10 )
30
Medium Subgrade
El = 500,000 psi
25
,,..._
(")
kl = 3000 psi
0
..--t
~
..... 20
Ill D1 = 2 in
0.
.._
Ill
- ;j
15 D1 - 8 in
\
;j
'O
0
~
.+,J
I:: 10
• • •
~
-
.....Q)
.....
Ill
Q)
D1 - 15 in_/
0:: 5
0 5 10 15 20
say, 14 Eg tests at different 8 values and define the Eg-8 relationship (Fig. 7).
assigned. In this study, the development of layer coefficients was done with the
assignment of the definitions of layer variables, as shown in Fig. 12. Note that for
asphalt bound layers, a 1 and a 2 are determined from the procedures given in
Volume I of this study ( 12). The drainage coefficients m 2 and m 3 are determined
from the procedures given in the companion study to this report (2). The layer
coefficients a 2 and a 3 for unbound materials are determined in accordance with the
The nomographs available in the 1986 Guide (1) as shown in Fig. 18 and 19
relate layer coefficients a 2 and a 3 to base or subbase resilient modulus (Egb or Egsb)
I
I
0 20
I 0 18
40
0 . 16
..,.
(J 14 - - - - ,oo- - - - --35- -- - - 2 .0 - - -- - 30
·;;;
70 80 a.
0 . 12 C:
60
25 §
50
I "' M
u
40 70 2 .5
C/)
::;
ClJ _ ____ _ a: - rn-- -
0 10 CD - m ----- "S
3 30 u ;;,
a:
X
.!!'
u
0
I- 2
60
0 .08 3.5 C/)
u "'X
20
Cl)
t- 15
(f)
0 06 ------ -----50----- - --
4.0
0 .04
0 02
I 0
I 131
ljJ
Scale
Scale
derived
derived
by
on
aver aging co rr elat ions 01:ltained from Texcs
N CHRP pr o 1ec1 (3) .
0 .20
~
0 . 14 - - - - 100 - - - - - 90 - ::- - - - - - 2
M ·;;;
ro 70 ~ 20 C.
80
50 §
0 . 12 C
40 70 3
Cl>
a: :,
·"' CD
u 60
ro
~ ___ _
t- 15 "'
:,
0 10 _ a; _ "' _ _ _ _ _14
__ ro :,
-0
~ 13 0
50 t- 12 ~
~
0 .08 :,
4 11
u 10
:, 10
(/)
40
0 06 -
30
5
25 5
Once the Eg value for a given situation is known, either by test or by use of Eq .
nomographs.
a variation in degree of saturation are addressed through the use of drainage (m)
Equivalent Stiffness
Odemark's transformation. As was done in Volume I of the study with the asphalt
mixtures ( 12), stiffnesses (in the form of resilient modulus) of MHTD aggregate
types were related to the modulus of the Road Test base material at various stress
for unbound crushed stone base at the Road Test is reported in the literature as
0.14. 2) The value for Eg (MHTD) will vary with several factors. Eg can be
determined by test or from Eq. 10. Knowledge of the stress state (or pavement
I
70
geometry/load condition) is necessary to determine Eg. For the present analysis,
the stress state for the average Road Test section was determined and used in Eq.
2 for both MHTD and AASHO Eg values. 3) The Eg (AASHO) was determined as
follows.
Road Test Modulus. In the AASHTO Guide (1 ), Fig. 2.6 gives a relationship
between a 2 and resilient modulus. It states that the basis of the nomograph is a 2
= 0.14 at Eg = 30,000 psi. The question is, where did the 30,000 psi value
come from? First, a search of the literature was made. In the Guide, an equation
was given to be used in lieu of the nomograph which has been presented earlier in
this study. The reference given is Rada and Witczak ( 10). However, Rada and
Witczak state that the equation was based on the nomograph. Thus, it seems that
Rada and Witczak either found the source of the nomograph (but did not elaborate
on it) or they just calculated the equation from the nomograph. Apparently they
moduli varied from 15,000 to 30,000 psi. It does not state the origin of these
values for all the Road Test materials were assigned on the basis of four
references: Seed et al. (43), Shook and Fang (44), Coffman fil fil. (45), and Skok
and Finn (46). State DOT studies of layer coefficient derivation are described in
71
I Appendix C of NCHRP 128. Very few are actually referenced, indicating that they
testing. Of the four references mentioned above, Seed et fil. dealt with soil, Shook
and Fang reported index tests and so forth, but no resilient modulus testing, and
Coffman et fil. tested static triaxial properties of Road Test granular materials and
estimations of dynamic properties. No estimations of dynamic modulus exceeded
20,000 psi, and for Road Test vehicle speeds, from the authors' calculations it
appears that the base and subbase E9 would have been considerably lower than
20,000 psi. Skok and Finn summarized other studies. In their paper, it appears
that a value of 15,000 psi was assigned to the Road Test base based on plate load
tests. Skok and Finn emphasized that this was very approximate. So , references
tied to NCHRP 128 do not shed light on the matter. The question remained -
where did the 30,000 psi value come from? Several sources or methods of
NCHRP 291. Pursuing the question of the base modulus further, several
Guide, the value for the gravel subbase was given as 15,000 psi. It was stated
that E9 values came from NCHRP 291 (21 ). NCHRP 291 stated that the Road Test
base and subbase were tested by the Asphalt Institute, but no reference is given.
Apparently, this is another internal report. The testing was done under conditions
of low moisture, medium moisture, and high moisture. Although test data were
not given, Table 17 is a summary of the seasonal moduli equations that were
72
presented. For some unstated reason, September was omitted. Neither seasonal
moduli nor an overall single modulus were given. Thus, to compute the Road Test
moduli, in the present study elastic layer analysis was used to calculate the
seasonal moduli from the given equations. By use of the program ELSYM 5 (47),
numerous iterations were run for each season to reconcile the stress state (0) and
modulus values. ELSYM 5 allowed the use of the subgrade modulus model shown
From the Road Test reports (3), the average thickness for asphalt surface plus
binder, crushed stone base, and gravel subbase were calculated from the main
factorial study loop sections . The load application details ( 18 k SAL, 2 tires, tire
spacing, tire pressure) were taken from NCHRP 128 and Witczak (48). Poisson's
73
r- p - 70 ps i
4.00 in U 1 = 0.40
tr 2 = 5 .39
µ...
1
in ,i = 0 ,....)_!
~,c
2
I
I
D = 8 .75 in
}_J = 1].]5
.?
3
,U
4
= 0.45
ratios for each material were taken from Appendix DD of the AASHTO Guide.
From all this, the seasonal moduli and bulk stress values were calculated, and are
shown in Table 18. Winter moduli were assumed to be 50,000 psi for the base,
with October and November, the average moduli for each material (weighted for
the number of months per season) were calculated and are shown in Table 18. As
can be seen, even with the inclusion of September into the highest modulus
season and assuming a frozen base in winter, the base modulus is still only 25 to
26,000 psi, not 30,000. In this analysis, the subbase seems high at 24,000, and
the subgrade seems very high at 17,000. Most references place the AASHO Road
Test soil at 3000 to 5000 psi. Use of KEN LA YER puts it at 5 543 psi. So,
inclusion of frozen conditions seems questionable. Thus, the average moduli (and
stress states) were re-calculated, omitting the frozen months. These somewhat
lower values are shown in the right-hand column in Table 18. The subgrade (5800
psi) seems much more reasonable. The subbase (15,000 psi) now lines up with
that which was stated in the AASTHO Guide. However, the base is 18,000 psi,
The results of Road Test trench studies (50) indicated that the average
moisture content of the crushed stone base was 4.3% (0.1 % above T-99
opposed to "dry" or "wet") and from Table 17, "moist" test results indicate k 1 =
3600 psi. At 0 = 14 psi (see Table 17), this corresponds to an Eg = 18,000 psi.
The average moisture content of the gravel subbase was 5 .65%, which was
considerably higher (for a granular material) than the optimum of 3.8%. Thus this
or 5000 psi. The corresponding Eg values at 0 = 6.2 are 14,000 and 15,000 psi,
respectively.
I
76
Traylor. Traylor (49) tested the Road Test base material and reports moduli
relationships as shown in Table 19. Traylor measured the base and subbase
moisture contents as 4.3 and 5. 7%, respectively, which compares favorably with
Road Test trench studies. Thus, at these average conditions, from Table 19 it
'
appears that the average base modulus would be about 27,000 psi and the
subbase would be about 11,000 psi. Traylor also tested the subgrade for moisture
and Esg· The average moisture content was 15.8% and the average Esg was 3472
psi. The AASHO Road Test subgrade T-99 optimum moisture content was 13.3%.
Road Test trench studies showed the average subgrade moisture in the top 4 in to
be 16.6%.
Moisture contents, %
4.0 5.5 7.0 Road Test
Base
Equation 1 o,360 e0 ·35 11,795 e0 ·34 2s50 e 0 ·62 *10,647 e0 ·35
Egb' psi 26,815
Subbase
Equation 6840 e 0 ·32 6270 e0 ·30 4075 e 0 ·40 * *6270 e 0 ·32
E0 ~h· psi 10,839
* interpolated k 1 and k 2 at moisture = 4.3%; 0 = 14.0 psi
* * at moisture = 5.5%; 0 = 6.2 psi
NCHRP 128. Fig. 18 of NCHRP 128 indicates that, from an elastic layer analysis
for E1 = 450,000 (the given value) psi and 0 1 = 4 in., E9 b equals 30,000 psi.
77
The analysis is based on equivalent sections which render equal vertical subgrade
compressive strain.
KENLAYER. In the present study, KENLAYER was used to calculate Egb and Egsb·
Input included E1 = 656,800 psi (best estimation of Road Test average asphalt
layer, in accordance with Vol. I of this report), k 1 = 10,647 psi and k 2 = 0.35 for
the base (as per Traylor), k 1 = 6270 psi and k 2 = 0.32 for the subbase, and K 1 of
the subgrade = 5619 psi. K 1 (corr) was calculated from Eqs. 16 with the following
Parameter Value
minus #200, % 81
Liquid limit 29
Pl 13
GI (new) 8.6
amount clay, % 15.3
amount silt, % 27
average compaction, % 97.7
avg. max. dry density, pcf 116.4
97. 7% max. dry density, pcf 113.7
95% max dry density, pcf 110.6
optimum moisture content, % 15. 1
Specific gravity 2.71
average in-service moisture, % 16
Thus it is not clear as to the value of the resilient modulus of the Road Test
base material. As shown in the next section, several values were evaluated.
In the Road Test report, the layer coefficient of the crushed stone base
material was reported as 0.14, regardless of layer thickness. This implies that only
elastic layer analysis with Eg varying from 5000 to 30,000 psi. In this way (J also
varied. Vertical compressive strain (Ev) on the subgrade was calculated. Results
of this calculation were used to calculate subgrade rutting fatigue life as follows
(48):
The results are shown in Table 21. As can be seen, Eg is significant to pavement
D2
5.39 in 14.14 in
Egb (psi
Ev Ntr Ev Ntr
5,000 0.001162 18,802 0.0009997 36,884
I 10,000
30,000
0.001127
0.0009577
21,561
44,686
0.0007064
0.0003814
174,555
2,756,077
Note: D1 = 4 in, E1 = 450,000 psi
The basic design equation in the 1986 AASHTO Guide was used to back-
calculate SN values:
log [ fl.PSI
log W1a = ZR So + 9.36 log (SN+1 )-0.20 + - - - --4 -·~-~-·;-4~--
2
l (23)
0.40 + - - - - ~
(SN+1)5.19
+ 2.32 log E50 - 8.07
where:
50% reliability)
80
SN = structural number
ll. PSI = loss of Present Serviceability Index (4.2 to 2.5 was used)
Then, knowing Road Test average layer thicknesses, the following was used to
SN - a 1 D 1
a 2 = - - - - - - - . • • • . . . . . . . . . . . . . . (24)
D2
where:
I shows that layer coefficients calculated from the Odemark analysis do not drop off
decreases. According to Table 22, it appears that the nomograph coefficients are
more realistic than the Odemark coefficients. Although this brief analysis was for
only one type of distress (rutting) and one pavement section, it does seem to show
that the use of the AASHTO nomograph for prediction of a2 from resilient modulus
I
I
82
Conclusions
solution for E9 via Eq. 10 or by use of KENLAYER and then 2) solution for a 2 and
CEMENT-TREATED BASE
INTRODUCTION
The third part of this study was to determine layer coefficients for cement-
treated soil base. The analysis is based upon the static compressive chord
for the soils analyzed in this report. Once the modulus is determined, the layer
coefficient (a 2 ) can be obtained directly from the base layer nomograph contained
Thus, factors that affect compressive strength will likely affect compressive chord
finer than the #270 sieve, and material retained between #270 sieve and 0.005
I 17s). All materials were sampled and delivered to the UMR laboratory by MHTD
personnel.
I The materials, sources, and identification codes are shown in Table 24.
LABORATORY INVESTIGATION
Soil Classification
analyses and index testing. The results are shown in Table 25.
soil-cement mixture was tested at three cement contents. Additionally, the DR-17
soil was mixed with the DR-17s sand. Records supplied by the MHTD indicated
that a reasonable spread of soil-sand proportions would include 23, 30, and 37%
soil content. Because essentially all the DR-17 soil passed the #200 sieve and
I
84
virtually all of the DR-17 sand was retained on the #200 sieve, this meant that the
percent passing the #200 sieve for the three mixtures would be 23, 30 and 37%.
The gradations, Atterberg limits, and AASHTO soil classifications are shown in
Table 26. Also shown are the results of gradation curve characterization via
-
Hudson's A and the slopes-of-gradation curve calculations. These were used later
-
in the regression analysis portion of the study. Calculation of Hudson's A and the
slopes was discussed earlier in the unbound aggregate section. However, different
sieve sizes were used for the soil/sand mixtures to better reflect their gradation
I 85
I curve shapes.
The PCA Soil-Cement design manual (51) (tempered with MHTD experience)
indicated that for preliminary purposes, the amount of cement for the DR-16 and
DR-17 combination materials should be 9 and 8%, by dry weight of soil or soil-
-
(")
...,
135 Sample Type :
,: Cement
1,1:DD
DR-16
g,:
112 .9 pc!
-
(")
...,
.:::::_ 130
135
.:::::_ 130
......c OMC 10.B!i ......c
!:- !:- 1 25
...,
1 25 ...,
..d
..ct
.~120
-~ 120 Ill
Ill
ll::
ll::
..., ..., 11 5
....
1 15
c c Sample Type :
,: Cement
DR-17
e,:
::::, ::,110 ,: Send 77,:
=,... 11 0 =,...
... ... MOD 129 .7 pc!
Q OMC 6 .11 ,:
Q 105
105
100
100
4 6 8 10 12 14
6 8 10 12 14 16
Wat.er Content (%)
Water Content (%)
140 140
"' -.... 13 5
(") -...,
135
-
.:::::_130
..c
!:::-125
.:::::_ 130
....
..c
!:::-125
..., ...,
..ct ..d
.~120
ll::
Ill - ta0120
I ll
II::
..., 11 5 ..., 115
....
c c S ample Type : DR-17
;::::,110 Sample Type : DR-17 ::>110 ,: Cement 9,:
,: Sand 77,:
...=,... ,: Cement
,: Sand
g,:
10,:
I>'\
... WDD 122.3 pc!
o 105 MOD 126.7 pc! o 105 OMC 11.1,:
OMC 10 .5 :t
determined for the DR-16, DR-17 (23/77), DR-17 (30/70), and DR-17 (37/63)
I materials. The results are shown in Fig. 21. As expected, for soil/sand mixtures
I (DR17), the more granular the material, the higher the dry density and the lower
I indicates that the DR-16 soil should contain 9% cement. However, in the PCA
manual, Fig. 36 indicates 10%. MHTD records indicated that 6.9% has been
specified. Consequently, to cover this range, three sets of specimens were made
with cement contents of 6, 8, and 10%. For the DR-17 (23/77), DR-17 (30/70),
and DR-17 (37/63) mixtures, the PCA Table 2 indicates cement requirements of 5,
I 5, and 6% respectively. PCA Fig. 36 indicates 6, 6, and 7%, while MHTD records
show 4 to 10% recommended or specified, with 7 .4% as the most recent and
most similar in density to materials used in this study. Thus, mixtures were
I schedule.
I % Cement DR-16
(23/77)
DR-17
(30/70) (37 /63)
5 X X X
6 X
I 7
8 X
X X X
9 X X X
10 X
I
I
88
Specimen Fabrication
For the DR-17 (23/77) and DR-17 (30/70) materials, the material for four
bring the cement, soil (and sand) mixture to its optimum moisture content (as
mixer. Subsequently, for the rest of the specimens, the material was dry mixed in
the mixer for better mixture control. Then for each lift, the amount of water
Specimen size and method of compaction were two factors that had to be
are usually used; this was the case in this study. Specimens were 6 in in
the T-99 standard effort. This method was used in the present study, with
additional effort via vibration by air hammer and plate to bring each lift to its
required density. The specimens were compacted into steel split molds with a 5.5
lb hammer and a 12 in drop. Each layer received a varying number of blows which
determined by the T-99 analysis. After compaction, the specimens were placed in
a moist room at 73°F, then demolded. The DR-17 specimens were demolded after
I 89
two days, while the DR-16 specimens were demolded after three days because of
I their more fragile nature. After a total of seven days moist curing, the specimens
I were removed from the moist room and capped with a proprietary sulfur
compound. Two replicate specimens were made for modulus testing and one for
I compression testing, then one or both of the modulus specimens were also tested
I in compression.
90 (53). The results of the replicate specimens were used for determination of the
I modulus test load for each soil-cement mixture. Also, the results were used in the
Modulus of Elasticity
I (54). In brief, each specimen was capped with a proprietary sulfur compound. A
I compressometer yoke was mounted on the specimen which held an LVDT (linear
was then converted to load and displayed on a digital display. The load and
QUATTROPRO (55) where the load and deformation data was parsed and
I
I
90
converted to stress and strain. The stress and strain data were then loaded into
the data-fitting program TABLECURVE (56). A best-fit line was fit through the
data. The chord modulus was calculated by determining the x-y coordinates of
two points on the curve: 1) the load at 0.000050 strain, and 2) the load and
strain at 40% ultimate compressive strength. Two load runs were made and thus
the modulus was determined twice for each specimen ; then the results were
a 1 -a2
Ee = . . . . . . . . . . . . . . . . . . . . (25)
€1-€2
where:
E1 = strain at a 1 , in/in
E2 = 0.000050 in/in.
The equipment is shown in Fig. 22.
Results of the compressive strength and modulus testing are shown in Table
28. As can be seen, both strength and modulus increased with increasing cement
content. Also, as sand content of the DR-17 specimens increased, strength and
between strength and modulus is shown in Figs. 26 and 27. The best-fit equation
91
I
I
I
I
I
Fig. 22. Static Chord Modulus Equipment.
I
I
92
-
.....
I'll
0..
800
• DR17 23/77
0 DR17 30/70
.,_,,
..c:: 700 ...
DR17 37/63
DR16
~ D
bO
~
Q)
i...
~
Cl) 600
Q)
.....>
I'll
I'll
Q) 500
i...
0..
s
0
u 400
'O
Q)
~
.....
.....
~ 300
0
C.)
~
~
200 --~~~~---~~~~--~~~~---~~~---'
4 6 8 10 12
Cement Content (%)
I 12
I ....-,.
I[)
0 10
~
><
.....
C'/l
I -- 0.
C'/l
;:1 8
........
;:1
'O
0
~
I 'O
s...
0 6 0 DR17 23/77
I
.c:
C,)
• DR17 30/70
V DR17 37/63
.... DR16
I 4
4 6 8 10 12
I
94
12
9% Cement
........_
-
lO
0
10
><
.....
C/l
0-t
'-" 7% Cement
C/l
~ 8
......
~
"O
0 5% Cement
::.::s
"O
~
0 6
...c:::
u
4 ____....__ ___._____________
50 60 70 80 90 100
Sand Content (%)
I
I
I
I
I
.........
I[)
0
~ 10
><
.....
I 0..
[/J
..__,,
I
[/J
- ;=j
;=j
'O
8
0
~
'O
r,...,
0 6 0 DR17 77/23
I ...c::
u • DR17 70/30
'v DR17 63/37
T DR16
I 4
200 300 400 500 600 700 800
I Compressive Strength (psi)
Fig. 26. Relationship Between Compressive
Strength and Chord Modulus.
I
96
-
U')
0
~
....><
1:/)
14
0..
'-"
1:/)
:::l
...... 12
:::l
"O
0
:::::s
10 0
"O
~
0
0
,.q
t) 0
8
Q)
0
....>
1:/)
0
1:/)
00
Q)
~
6 0
0..
s
0
0
t)
4 oo
....
+J
C) 00
cd
+J
U) 2
200300 400 500 600 700 800
Unconfined Compressive Strength (psi)
Fig. 27: Overall Relationship of Unconfined Compressive
Strength and Static Compressive Chord Modulus
for Soil - Cement Mixtures.
97
This is similar to that which is reported by Felt and Abrams (52). For sand and
I soil/sand mixtures similar to the present study, they noted increasing modulus and
(-) #200 material seven day compressive strengths for 3, 6, and 10% cement were
100, 450 and 1000 psi, while their soil/sand mixture with 37% (-) #200 material
I
98
seven day compressive strengths at 6 and 10% cement were 400 and 600 psi,
respectively. The compressive moduli for the 10% (-) #200 material mixtures at 3,
6, and 10% cement contents were 0.3, 1.0, and 1.5 x 10 6 psi, respectively.
A multiple regression model was fit to the data . The criteria for model
acceptance was the same as presented earlier for Eg in Eq. 10. The R2 statistic =
0.601, the adjusted R2 = 0.562, and the SEE = 158,433. The best fit model is
as follows:
where:
C = cement content, %
Various parameters were tried in the model, including Pl, activity of fines, percent
-
passing #200 sieve and #270 sieve, A, slopes-of-curve, percent retained on #4
sieve, amount 0.005 mm material, and percent finer than 0.002 mm. None of
them improved the model nor were any statistically significant to the model. A
plot of the relationship of the estimated and observed soil cement chord moduli is
shown in Fig. 28. In comparing this relationship to the one delineated by Eq. 26, it
compressive strength.
I
I 99
I
I
I
I
12
I cu 0
.....> LD..--...
I fl)
I'll
cu
J-,
0
~
10
0 0
0
0
p... ><
I s
0
.....C/l
u p... 8
.._ 0
CJ
.....
I
I'll
...., ;::1
....,ro ......;::1 0 0
lf.1 "O 0 0
0 6 0
Oo
I "O
cu
+'
ro
~
"O
0
s J-,
0 0
I .....
....,
I'll
t:.J
..cl
u
4
I 2
2 4 6 8 10 12
I Observed Static Compressive Chord Modulus (psi X 10 5 )
I
I
I
I
100
28
10 0
.26
1(J()(J
.24 9 .0
>
0 .20
7.0
600 N
. 18
N
m
V,
'vi ::::,,,
. 16 c
------------- ------------- "'0
.~ 400 .c 6 .0
u
Cl
C V,
1<: CJ :::,
0 ~
:::,
u U'J "O
CJ 0
200 > 2
·.;;
0.12 u II)
5.0
CJ
(/)
~ ---- - ---
- - - - - - - - uo
0 . 10
::;
C
C
0
u
C
:::,
0
I 1l Scale derived by averaging correlations from Illino is . Louisiana and Texas .
121 Scale derived on NCHRP project (31.
I which is shown in Fig. 29. The equation for the relationship of a2 and modulus
I Substitution of elastic moduli into Eq. 28 resulted in the layer coefficients that are
I VERIFICATION
In Table 1 are listed unbound aggregate base and cement-treated base layer
I coefficients as reported by others. The layer coefficients shown in Table 29 seem
DESIGN VERIFICATION
I In order to further check the reasonableness of the layer coefficients, 18
I design situations were examined to compare standard MHTD designs vs. AASHTO
designs using the layer coefficients developed in this study. Subgrade soil and
I traffic conditions were varied to give a wide range of pavement thickness. The
I MHTD Flexible Pavement Thickness Design Chart was used to determine total
the unbound crushed stone base thicknesses were converted to the required
I
I
102
6 0.09
DR-16 -- 8 0.12
10 0.27
5 0.19
23/77 7 0.26
9 0 .29
5 0.16
DR-17 30/70 7 0.24
9 0.27
5 0.17
37/63 7 0.18
9 0.20
recommended in the MHTD method. Traffic levels were chosen so that each of
I
four design curves (2 through 5) in the MHTD chart could be used.
provided by Murray (57) was used to convert the MHTD 2 axle equivalents to
18kESAL as used in the AASHTO method . Also, the subgrade was characterized
to have a Group Index (old) of 6, 9, and 17. Eqs. 17 and 18 were used to
calculate E59 values. These values represent the soils predominant in Missouri.
Data from the MHTD Geology and Soils Manual (58) were analyzed and it was
determined that the average GI (old) of the soils reported was 10 and the standard
I
I 103
I deviation was 4, which indicates that about two thirds of the soil types have a GI
between 6 and 14. The asphaltic layer thickness was held constant for both
I design methods. In the AASHTO method, initial Present Serviceability Index (PSI)
I was assumed to be 4.2 and terminal PSI was chosen as 2.5 for a ~PSI = 1.7.
Reliability was taken as 50%, and standard deviation as 0.45. From the 1986
I AASHTO Guide design equation (Eq. 23) the resulting required structural numbers
I
I To solve for D3 in an asphalt-over-bituminous base-over-unbound subbase section:
I 03 = SN-(a1D1 +~D2 )
~
I The layer coefficient (a 1 ) for the asphaltic layers was taken as 0.42 and a 2 = 0.34
(as per Volume I of this report), and the layer coefficients a 2 and a 3 for the
I unbound base and subbase at various thicknesses were obtained from Eqs. 20 and
I 21. Eb values for the granular base and subbase were calculated using Eq. 10.
457,351 psi was used, which represents the average resilient modulus of 1990
I MHTD mix designs (see Volume I). Values of k 1 for the base and subbase were
I taken from Table 15: 6100 psi for the base and 4300 psi for the subbase. The
I
I
104
Granular a 2
In studying Table 30, the design base (0 2 ) thickness for the granular
material based on the AASHTO nomograph is similar to MHTD designs for the 6
I
and 17 GI soils for most traffic levels. For GI = 9, the nomograph a 2 - derived
designs are conservative. The Odemark a 2 -derived designs are conservative for
GI= 9, but are thinner than MHTD designs for Gl's of 6 and 17.
Granular a3
designs for soils of GI = 6, are conservative for GI = 9, and are thinner for GI =
17. The Odemark-based designs are conservative for GI = 9 and thinner than
Thus, two options are presented for possible use by MHTD. If an evaluation
by MHTD of its granular base design thicknesses reveals them to be either on-
base thicknesses are comparable for GI = 9, would be the same for GI =6 (if
- - - - - - ---------
Table 30 . Comp ariso n of AASHTO Design s to MHTD Des igns for a2.
GI Esg 2AE 18kSAL MHTD AASHTO Norn. Ode MHTD AASHTO MHTD AASHTO
(psi) (thous.) (in) (in) (in) (in) (in)
Old New Norn. Ode
(in) (in)
9 9 5619 100 100 2.2 3 3 0.10 0.12 6.0 9.1 7.6 4.o· 5.2
200 175 2.4 3 0.10 0.12 7.4 10.9 9.1 4.9
• 6.3
3
500 450 12.3 10.1 5.7
• 6.2
2.8 4 4 0.09 0.11 8.6
1000 875 3.1 4 4 0.09 0.11 11 .1 15.4 12.6 7.4 7.9
17 30 4579 100 100 2.4 3 3 0.10 0.12 9.5 11.4 9.4 6.3 6.3
200 175 2.6 3 3 0.10 0.12 12.1 13.3 11.0 8.1 7.4
500 450 3.0 4 4 0.09 0.11 14.2 15.0 12.2 9.5 7.3
1000 875 3.3 4 4 0.09 0.11 17.7 18.2 14.8 11.8 9.0
6 5 500 450 2.3 1.25 3.75 1.25 3.75 0.11 0.15 4.3 4.4 3.3
6 5 1000 875 2.6 1.25 3.75 1.25 3.75 0.12 0.15 6.4 6.9 5.1
9 9 500 450 2.8 1.25 3.75 1.25 3.75 0.11 0.14 5.7 9.3 7.0
9 9 1000 875 3.1 1.25 3.75 1.25 3.75 0.11 0.14 8.2 11.9 8.9
17 30 500 450 3.0 1.25 5.75 1.25 5.75 0.09 0.11 8.3 6.1 4.5
17 30 1000 875 3.3 1.25 5.75 1.25 5.75 0.09 0.12 11.8 9.3 6.9
I
I 107
thinner for two traffic levels on GI = 17 soil. On balance, the layer coefficients
I seem reasonable. However, the wide range in coefficients reported in Table 29
I emphasize the importance of testing the actual materials and assigning the proper
effect of saturation is examined in the companion study (2). For the a 2 coefficient
analysis, thickness of the surface (asphalt) layer was held constant at 4 in.
Thickness of the base layer was varied at 6 in and 12 in. Three levels of
I compactive effort (CE) were looked at: most significant change in the UMR data
set, least significant change in the UMR data set, and average change for all
blends . For instance, looking at Table 13, the largest change in E9 as a result of a
change of CE occurred with the DR-14 Mid high CE material as CE went from high
I (E 9 = 27,096) to low (E 9 = 16,278 psi). The least change occurred where the
I
I
108
E9 's for the four aggregate sources lowest CE to the average of the E9 's for the
four aggregate sources highest CE. In each case, CE was changed from high to
low with a resulting change in E9 • Then the resulting change in layer coefficient
(a 2 ) was calculated from the AASHTO nomograph. Finally, the required change in
rendered by the initial assumption of layer thicknesses. The results are shown in
Table 32. For example, what is the average change in required base layer
high to low? Looking at row 1 in Table 32, the E9 for the high CE (of the pair
which resulted in the most loss of modulus) is 27,096 psi. Moving to the low CE
layer coefficient is reduced from 0.127 to 0.072, the new required thickness 0 2 =
From Tables 32 and 33, it appears that, on the average, changes in unbound
base layer compactive effort are significant for both 6 in and 12 in layers. The
thicker the granular base or subbase layer, the more pronounced is the effect.
I
I 109
6 Worst 27,096 16,278 10,818 0.127 0.072 0 .055 2.44 10.6 4.6
12 27,096 16,278 10,818 0.127 0.072 0.055 3.20 21 .2 9.2
6 Avg 22,629 19,518 3111 0.107 0.091 0.016 2.32 7.1 1.1
I 12 22,629 19,518 3111 0.107 0.091 0.016 2.96 14.1 2.1
6 Least 25,884 24,309 1575 0.122 0.115 0.007 2.41 6.4 0.4
I 12 25,884 24,309 1575 0.122 0.115 0.007 3.14 12.7 0 .7
Note: a 2 from AASHTO nomograph
0 1 = 4 in, a 1 = 0.42
E9 at 8 = 10 psi
* = Data includes both gradations, both degrees of saturation, all aggregate sources
I
I Thus, it appears that compactive effort is important to the resulting layer
coefficient. The effect of degree of saturation goes beyond just the effect on k 1
and k 2 of the granular material because the available moisture adversely affects the
I subgrade modulus which reduces the bulk stress of the granular base which
reduces the Eg in addition to the reduction brought about by the change in k 1 and
I k 2. This will be addressed by use of m-coefficients as discussed in Reference 2.
I SOIL CEMENT
As shown in Eq. 27, only cement content and sand content were significant
I in their effect on Ee. An analysis similar to that for the granular material was
content or sand content. Thickness of the surface (asphalt) layer was held
I
I
110
~
12 17,926 16,299 1627 0.126 0.117 0.009 4.34 13.0 1.0
Note: a3 from AASHTO nomograph
0 1 = 1.25 in, a 1 = 0.42; D 2 = 6.75 in, a 2 = 0.34
Eg at 9 = 5 psi
* = Data includes both gradations, both degrees of saturation, all aggregate sources
I
constant at 4 in. Thickness of the base layer was varied at 6 in and 12 in. Three
levels of cement content or sand content were studied: most significant change in
the UMR data set, least significant change in the UMR data set, and average
change for all mixes. For instance, looking at Table 28, the largest change in Ee as
a result of a change of cement content occurred with the DR-1 6 soil as cement
content went from 6% (Ee = 435,700) to 10% (Ee = 1,001,600 psi). The least
change occurred where the 37/63 DR-17 blend went from 5% (Ee = 632,000) to
comparison of the average of the Ec's for the four soil blends' lowest cement
contents to the average of the E/s for the four soil blends' highest cement
contents. Cement or sand content was changed from high to low with a resulting
111
I change in Ec. Then the resulting change in layer coefficient (a 2 ) was calculated
from the AASHTO nomograph. Finally, the required change in thickness was
computed as needed to maintain the initial structural number rendered by the initial
assumption of layer thicknesses. The results are shown in Table 34. For example,
what is the average change in required base layer thickness for an initial thickness
I Table 34, the Ec for the high cement content (of the pair which resulted in the
most loss of modulus) is 1,001,000 psi. Moving to the low cement content
corresponding loss in a 2 is 0.18. The structural number (SN) provided by the fine
From Table 34, it appears that all changes in cement content and sand
I content are significant for both 6 in and 12 in layers. For thicker layers, the effect
becomes more important. Thus, it appears that optimizing sand and cement
The purpose of this report was to determine layer coefficients for Type 1
the various materials and then conversion to layer coefficients. The study included
I
Table 34. Thickness Sensitivity to Changes in Cement Content and Sand Content.
I
CEMENT CONTENT CHANGE
12
Worst 1,001,600
1,001,600
435,700
435,700
595,900
595,900
0.27
0.27
0.09
0.09
0.18
0.18
3.30
4.92
18
36
12
24
I
I
6 Avg 965,675 591,950 373 ,725 0.26 0.15 0.11 3.24 10.4 4.4
12
Avg 935,033
935,033
678.467
678,467
256,566
256,566
0 .25
0.25
0 . 18
0. 18
0 .07
0 .07
3 .1 8
4 .68
8 .33
16.67
2.33
4 .67 I
6 Lea st 698,400 601,300 97,100 0 .19 0 . 16 0 .03 2 .82 7. 1 1.1
gravels, and two soils (to be mixed with cement). Additionally, a concrete
sand was supplied for combining with one of the soils. One soil was a fine
sand, while the other was a silty clay. All materials were selected, sampled,
2. For the granular base study, two gradations of granular material were
acceptance band, and the other was the so-called New Jersey open-graded
gradation.
.,
113
I (T-99, T-180) were performed for the dense-gradation, while vibratory table
03398, while the (-) #8 to ( +) #100 fraction was tested using the NAA
I method. The measured angularities of the two stones were about the same,
and they were more angular than the two gravels, which also were about
6. Resilient modulus tests were run on all four aggregates using two
deviator stress were used for each specimen. Effective confining pressures
I
114
ranged from 2 to 20 psi and cyclic deviator stress ranged from 2 to 40 psi.
Thirty-two specimens were fabricated. The total number of tests run was
896. I
The results of the testing indicated that Eg increases with increasing
For the materials tested, the effects of particle angularity and gradation were
not significant.
saturation gave significantly different results at the 0.05 level, but gradation
did not. Tukey HSD analysis indicated that one of the two gravels gave
significantly different results than the other gravel and the two crushed
stones. The particle shape analysis indicated that the shapes of the two
gravels were about the same, but they were both different from the two
crushed stones. All of this taken together indicated that particle shape was
8. A multiple regression model was developed for the parameter k 1 but the
9. To assist the pavement designer who may not be able to obtain actual E9
material constant k 2 (0.341, 0.653, 0.776). The most accurate model was:
10. For comparison purposes, layer coefficients were computed in two different
I for unbound granular base and sub base to obtain a 2 and a 3 directly.
I 11. Because layer coefficients are directly affected by resilient modulus, the
drainage).
12. The Knox soil (A-6) was mixed with concrete sand in three proportions of
soil to sand: 23/77, 30/70 and 37/63. Specimens were made with 5, 7,
I
116
and 9% cement contents by weight. The Lintonia (A-3) soil was mixed with
13. Two regression equations were developed for estimation of static chord
modulus:
strength.
14. Layer coefficients were obtained for the soil-cement mixtures by use of the
AASHTO nomograph. Values ranged from 0.09 to 0.27. Most of these fit
designed by use of both the former MHTD method and the AASHTO method
using the layer coefficients developed in this study for unbound base and
soil-cement. This analysis tended to verify the choice of using the AASHTO
16. A sensitivity analysis was performed. The results indicated that a higher
the granular material, and to lower subgrade support, all of which act to lower the
coefficients, as developed in the companion study of this report (2). Thus, the
SN = a1 0 1 + m 2 a2 D 2 + m 3 a3 D 3
RECOMMENDATIONS
I more accurately:
recommended.
19:
E =
D1-.tE.1a)o.333 + D1°'.IE.1b)o.333r
eq [ D1a + D2a
2. Layer coefficients for cement treated soil base can be determined in the
following manner:
accurately
Ee = 132,772 + 1314.9 qu
or less accurately
where:
C = cement content, %
the effects of particle shape and gradation on Eg. This would include several
I 4.
the subgrade, the prediction of Esg needs to be better defined.
regression equation.
120
ACKNOWLEDGEMENT
The authors wish to thank the MHTD for its sponsorship and support of this
research project. They also thank the UMR Department of Civil Engineering for its
support. Special thanks go to Mr. Kevin Hubbard for his assistance in the figure
preparation portion of the study, and to Aswath V. Rao for assistance in the
laboratory work.
121
I
REFERENCES
I
I
122
3. "The AASHO Road Test, Rpt. 5-Pavement Research," Hwy. Res. Bd. Spec.
Equivalency Ratios", Trans. Engrg. Series No. 38, Illinois Coop. Hwy. and
6. Van Til, C.J., B.F. McCullough, B.A. Vallerga, and R.G. Hicks, "Evaluation of
Application to Design," Hwy. Res. Rec. 71, Hwy. Res. Bd., 1965, pp. 151-
I 171.
123 I
9. Walters, R., "Implementation of the New AASHTO Design Practice", filfil.
Annual UMR Asphalt Conference, Univ. of Missouri-Rolla, Rolla, Missouri,
1988.
Granular Materials from Resilient Modulus", Trans. Res. Rec. 852, 1982, pp.
15-21.
11. Jorenby, B.N. and R.G. Hicks, "Base Course Contamination Limits", Trans.
14. Barksdale, R.D., S.Y. Itani, and T.E. Swor, "Evaluation of Recycled
15. Jin, M., K.W. Lee, and W.D. Kovacs, "Field Instrumentation and Laboratory
Soils," Trans. Res. Bd. 71 st Annual Meeting, Trans. Res. Bd., Washington,
D.C., 1992, 30 p.
16. Kallas, B.F. and J.C. Riley, "Mechanical Properties of Asphalt Pavement I
124
I 17. Thom, N.H. and S.F. Brown, "The Effect of Moisture on the Structural
18. Hicks, R.G. and C.L. Monismith, "Prediction of the Resilient Response of
WES, 1979, 82 p .
Granular Bases," Trans . Res. Rec. 1278, Trans. Res. Bd., Washington, D.C.,
I 1990, pp . 7-17.
I Resilient Moduli Results for Granular Material," Trans. Res. Rec. 810, Trans.
I 23 . Haynes, J.H. and E.J. Yoder, "Effects of Repeated Loading on Gravel and
125
Crushed Stone Base Course Materials Used in the AASHO Road Test," Hwy.
Res. Rec. 39, Hwy. Res. Bd., Washington, D.C., pp. 82-96.
24. Kandhal, P.S., J.B. Motter, and M.A. Khatri, "Evaluation of Particle Shape
and Texture: Manufactured vs. Natural Sands," NCAT Rpt. No. 91-3,
25. "Test Method for Index of Aggregate Particle Shape and Texture," ASTM
26. "Standard Test Method for Particle Shape, Texture, and Uncompacted Void
MD, 1991, 12 p.
27. "Standard Test Method for Specific Gravity and Absorption of Coarse
28. "Standard Test Method for Specific Gravity and Absorption of Fine
29. "Standard Test Method for The Moisture-Density Relations of Soils Using a
I Testing, 15th ed, Part II, Tests, AASHTO, Washington, D.C., 1990, pp.
226-230.
455-459.
32. "Standard Test Method for Minimum Index Density of Soils and Calculation
33. Interim Method of Test for Resilient Modulus of Subgrade Soils and
35. Claros, G., W.R. Hudson, and K.H. Stakoe, II, "Modifications to the Resilient
Modulus Testing Procedure and the Use of Synthetic Samples for Equipment
Calibration," Trans. Res. Bd. 69th Annual Meeting, Trans. Res. Bd.,
I Washington, D.C., 1990, 27 p.
127
37. Huang, Y.H., Pavement Analysis and Design, Prentice Hall, Englewood
38. Witczak, M.W., and B.E. Smith, "Prediction of Equivalent Granular Base
39. Thompson, M.R. and Q.L. Robnett, "Resilient Properties of Subgrade Soils,"
Pavements," Hwy. Res. Bd. 25th Annual Meeting, V. 25, Hwy. Res. Bd.,
41. "Method for Determining the Potential Vertical Rise, PVR," Texas Hwy.
42. Elliot, R.P., "Selection of Subgrade Modulus for AASHTO Flexible Pavement
43. Seed, H.B., C.K. Chan, and C.E. Lee, "Resilient Characteristics of Subgrade
44. Shook, J.F., and H.Y. Fang, "Cooperative Materials Testing Program at the
AASHO Road Test," Hwy. Res. Bd. Spec. Rep. 66, Hwy. Res. Bd., 1961,
I
I 128
I pp. 59-102.
45. Coffman, B.S., D.C. Kraft, and J. Tamayo, "A Comparison of Calculated and
I Measured Deflections for the AASHO Test Road," Proc. of Assn. of Asphalt
46. Skok, E.L., Jr., and F.N. Finn, "Theoretical Concepts Applied to Asphalt
I Concrete Pavement Design," Proc. International Conf. on the Structural
I 50.
1978, 213 p.
"The AASHO Road Test, Rpt. 2-Materials and Construction" Hwy. Res. Bd.
52. Felt, E.J., and M.S. Abrams, "Strength and Elastic Properties of Compacted
I Soil-Cement Mixtures", Bulletin D16, PCA, 1957, 26 p.
54. "Standard Test Method for Static Modulus of Elasticity and Poisson's Ratio
58 . Saville , V.B. and W.C. Davis, MHTC Geology and Soils Manual, Von
I
I
I APPENDIX A
I
DATA SET FOR RESILIENT MODULUS
I
I REGRESSION EQUATION (Eq. 10)
I
I
I
I
I
I
I
I
I
I
I
-
Resilient Modu li for Gran ular Base Material (E 1 = 130,000 psi) .
E1 = 130,000 psi
---- - --
E0 (filename/base/subgrade), psi
D2(in .)
4 12 18
o, Esg k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k,= 3000 k,= 11,000 k, = 1800 k, = 3000 k, = 11 ,000
(in.) k2 = 0.776 k2 = 0 .653 k2 = 0.341 k2 = 0.776 k2 =0.653 k2= 0.341 k2 = 0 .776 k2 = 0.653 k2 = 0.341
115vs41 115vs42 115vs43 115vs1 21 115vs 122 115vs123 115vs181 115vs182 115vs183
very soft
1000-5662 6628 8255 14,450 6991 9076 17,960 7138 9419 19,290
3
[
D E 0 .333 + D E o.333
]
. . la la lb lb
Note: Eg values based on °S = 10, 60%; 0 3 = 0 m.; for multiple asphalt layers, use Eeq = ( } ( }
D1a + D1b
• statistical outliers due to non-convergence in iterative process.
- - ----------- - - - -
Resil ient Moduli for Granular Base Material (E 1 = 500,000 psi) .
D2(in .)
4 12 18
o, Esg k, = 1800 k, = 3000 k, = 11,000 k,= 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11 ,000
(in .) k2 = 0 .776 k2 = 0.653 k2 = 0 .341 k2 = 0 .776 k2 = 0 .653 k2 = 0.341 k2 = 0.776 k2 = 0.653 k2 = 0 .341
22vs541 22vsl 21 22vs122 22vs123 22vsl 81 22vs182 22vs183
very soft
5938 + +
1000-5662 18,370 21,160 30,500 19,610 22,470 31 ,610
3
D E 0.333 + D E 0.333
. . la la lb lb
Note: Eg values based on °S = 10, 60%; D3 = 0 m.; for multiple asphalt layers, use Eeq = ( ) ( )
[ D1a + D1b ]
• statistical outliers due to non-convergence in iterative process
--------
Resilient Moduli for Granular Base Material (E 1 = 2, 100,000 psi) .
E1 = 2,100,000 psi
-- -- ---
E0 (filename/base/subgrade). psi
4 12 18
o, Esg k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11,000 k, = 1800 k, = 3000 k, = 11,000
(in.) k2= 0.776 k2 = 0 .653 k2 = 0 .341 k2 = 0.776 k2 = 0.653 k2 = 0.341 k2 = 0.776 k2 = 0 .653 k2 = 0 .341
- - 3
D la(Ela )°"333 + D lb(Elb)° .333
Note: Eg values based on °S = 10, 60%; 0 3 = 0 in.; for multiple asphalt layers, use Eeq =
D1a + D1b