Beruflich Dokumente
Kultur Dokumente
o Attribution You must give appropriate credit, provide a link to the license, and indicate if
changes were made. You may do so in any reasonable manner, but not in any way that
suggests the licensor endorses you or your use.
o NonCommercial You may not use the material for commercial purposes.
o ShareAlike If you remix, transform, or build upon the material, you must distribute your
contributions under the same license as the original.
STRUCTURAL CLASSIFICATION OF
GRANULAR BASE PAVEMENTS USING
MEASURED DEFLECTION BOWL PARAMETERS
by
STRUCTURAL CLASSIFICATION OF
GRANULAR BASE PAVEMENTS USING
MEASURED DEFLECTION BOWL PARAMETERS
by
1992
SUMMARY
The: structural classification of granular base pavements by means of resilient deflection bowl
parameters measured by the Impulse Deflection Meter and application of measured deflections at
network level were investigated. The Investlgatlon consisted of a mechanistic analysis and an analysis
of observed field results. Curves for structural classiflcation of granular base pavements obtained by
regression are presented. Thecurves obtained by the anal ysis of field observations compare well with
local experience and theoretical failure criteria. It was found that the mechanistic model used in this
study did not yield satisfactory results when simulating deflection behaviour through linear elastic
programs. The parameters recommended for future use are Maximum Deflection (Y -rnax), Base
Layer Index, or DLl (formerly SCI), Middle Layer Index, or MLl (formerly DOl) and Radius of
Curvature. The use of deflection measurements at network level was also investigated. A method
for calculating a Structural Stiffness Index to be incorporated into Pavement Management Systems
as a Structural Indicator is proposed.
identification of roads for testing at network level are also made. Several factors influencing
deflections and structural bearing capacity are discussed and some recommendation." are made. The
main factors influencing deflections and predicted pavement structural capacity were shown to be
Temperature, Seasonal Influences and Condition of Surfacing. Further research into the specific
influence of these factors on deflections and hearing capacity is needed.
\I
The Director General, South African Department of Transport for his permission to use the
work done under RDAC project 91/325 for this thesis,
Prof. T.r.H. Legge for his guidance and support during the preparation of this thesis.
Dr J. H. Maree for his guidance, support and expertise offered throughout the preparation of
this thesis, and without whom this work would not have been possible,
All my colleagues at the Pavement Engineering & Materials Division of Stewart Scott Inc.
for the advice and help offered during the preparation of this thesis, and for enduring my
monopolizing of the typist.
Stewart Scott Incorporated for the bursary and opportunity to undertake this study.
\}
CONTENTS
Pagl:
SYNOPSIS
LIST OF CONTENTS
LIST OF TABLES
(i)
LIST OF FIGURES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . ..
(iv)
CHAIYfER I: INTRODUCTION
1.1
BACKGROUND
"
1-2
1.2
1-5
1.3
1-6
INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2-2
2.2
2-2
2.3
2-8
2.4
2.5
2-9
2-14
3-2
3.2
3.3
3-5
3.4
3-5
3.5
3.6
39
312
CONTENTS (continued)
PAGE
CIIAIJTER 4: ANALYSIS 0 ...... EU> RE..4iULTS
4.1
42
4.2
4-3
4.3
4-7
4.4
EMPIRICAL ANALYSIS
4-14
4.~
420
EVAI~UATION
RESUI.TS
5.1
5.2
52
52
5-6
5.4
5-15
5.5
RECOMMENDATIONS
5-16
INTRODUcrION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
6.2
6-5
6.3
69
6.4
6.S
\~
615
617
CONTENTS (continued)
PAGE
6.6
6.7
6.8
6-19
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 6-22
6-23
INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7-2
7.2
7.3
7-S
7-12
APPENDIX B:
APPENDIX C:
REFERENCES
\!
s-t
(i)
LIST OF TABLES
PAGE
CHAPTER 2
Tabl!; Nu.
2.1
. . . . . . . . . . . . . . . . . . . .. 2-11
CHAIYfER 3
Table No.
3.1
. .... 3-5
3.2
3-11
3.3
3-14
CHAPTER 4
Table No.
4.1
4.2
4-4
4.3
\.)
"
4-16
(ii)
PAGE
4.4
4.5
4.6
4.7
4.8
4.9
4.10
CIIAfYfER 6
Table No,
6.1
6.2
6-13
6.3
6.4
,
,
1. . . . .
6-13
: .....
6-18
, . . . . . . . . . . . 6-20
(iii)
PAGE
CIIAIYfER 7
Tabh: No.
7. I
7.2
7.3
78
79
7-14
7.4
7.5
7.6
7.7
, 7-15
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7-15
, 7-18
\ ,I
(iv)
r.rsr OF FIGURES
PAGE
CIIAIYfER I
EiI:un: No,
1.1
1.3
114
, . . . . . . . . .. 18
CIIAIYfER 2
Eil:urc No,
2.1
2-3
2.2
2.3
2.4
2.5
2-10
, . . . . . . . . . . . . . . . . . . . . . . . .. 2-13
CIIAPTER 3
Fi~\Ire
No.
3.1
3.2
load Simulation
3.3
3.4
, . . . . . . . . . . . . . . 33
3-4
37
\ J
:lCI1.Y01(fo1')
(v)
L1~"
OF FIGURES (continued)
PAGE
3.5
3.6
. 314
3.7
. 315
313
CIIAIYfER 4
Fi~ure
No.
4.1
4.2
. . . . . . . . . . . . . . . . . ..
E-Suhgrade = 25 (N/mm2) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
4.3
45, 6
4-10
4.4
4.5
= 150 (N/mm2)
. . . . . . . . . . . . . . . . . . . . . . .
4-10
4.6
4-11
4.7
4-12
4.8
4.9
(vi)
PAGE
CIIAI'TER 5
fj~lIn:
Nil,
5, I
, , , .. , , ,
, ,,,,,
, , .. , , , . , , . . . . .
5-3
5.2
5,3
5-4
5.4
5-4
5.5
, , .. ,
, 5-3
, , , . , , . . . . . . . . . 5-5
5.6
5.7
5.S
5.9
5. 10
5.11
5-7
,,
5-7
,. . ... . . . . .
5-S
. .. .
5-S
"
5-9
, . . . . . . .. 5-12
5.12
5.13
5.14
Comparison of Y-max
~1.VOI(Yo'l'l
V5
5-13
,,
5-14
(vii)
CHAPTER 6
Fh:lIn: No.
6. I
6-3
6.2
6.3
6-4
6.4
6-4
6.5
6-6
6.6
6.7
6.8
6-\0
. . . . . . .... . ... . . . . .
6-11
6-16
6.9
6-19
6.10
CHAPTER 7
Fi~ure
No,
7.1
7.2
, . . . . . . . . . . . . . . . . . . . . . . . . 7-3
. :lilt. YOI(Vo?)
I -I
CIIAI'TER I: INTRODUCfION
PAGE
1.1
BACKGROUND
1-2
1.2
1.3
1-6
\ ,I
I 2
1.1
BACKGROUND
The need fur accurate determination of pavement structural condition has become more urgent
recently as the emphasis in the pavement engineering industry has shifted from new
pavements to the maintenance and rehabilitation of existing pavements.
In the Republic of South Africa (SA), this shift ha.s been caused by various rea.s ons, The
most important of these is the limited amount of capital available for the con.struction of new
roads due to extensive cuts in government expenditure on roads. This has 1tX.! to a situation
where engineers in SA are faced with the problem of maintaining a fairly old road system to
a reasonable standard of serviceahility, This problem wa.s further aggravated by the large
increases in traffic volumes on existing roads, as well as the number of heavy vehicles and
the type of loads carried (I). Thus there originated the urgent need for a method by which
an existing pavement could be accurately assessed in terms of not only its surfacing condition
and other functional parameters but also in terms of the structural soundness of the pavement,
which ha.s a dlrect bearing on certain defects such as rutting and deformation as well as the
formation of cracks, caused by various factors such as consolidation and fatigue induced by
accumulated traffic. In order to assist in determining the structural capacity of a pavement
certain non-destructive testing (NDn techniques such as the Dynamic Cone Penetrometer'
(DCP) and Surface Deflection Measuring devices have been developed (2,3). More recent
NOT developments include Ground Penetrating Radar (GPR), which iscurrently in advanced
stages of development (4).
Of the NOT techniques currently available, the measurement of surface deflections under an
applied load has been acknowledged as one of the most cost effective and useful (5). Various
deflection measurement devices are currently used locally and overseas. These devices can
bebroadly categorised in terms of the type of load the device imposes on the pavement. The
three most well known modes of loading are (5): Static or Slow Moving Load,
~tcady
State
Vibratory load and the Transient Impulse load (3,6,7,8,). The transient impulse load device
known a.'l the Falling Weight Deflectometer (FWD) or Impulse Deflection Meter (10M) has
been shown to he the most efficient and versatile type of deflection measuring device (3).
Advantages of the FWD or 10M devices include (5):
I Althooah the ocr it no llrictly lJ'f&kinll _ NOT device. lhe clamAllC to the ,,-,.d uulCd loy the ocr p mini",.t ..tlcn
the tCJIC it nnI conducted with ~inll CORll and tUl riu.
I 3
i)
ii)
iii)
iv)
Traffic loads expected on the pavement can be simulated in terms of magnitude and
duration;
v)
vi)
Disadvantages of the 10M/FWD systems arethe complexityof their designs and subsequent
high initial cost.
10M measurements can be taken without damaging the road surface or excessive disruption
of traffic. The versatlllty and speed of, and information to be gained from the 10M, make
the 10M an invaluable tool for rehabilitation design. Figure 1.1 is adapted from the TRH
12 Draft document of 1991 (9), and shows the various steps involved in the rehabilitation
design process. II isclear from Figure 1.1 that deflection measurements form an integral part
of the rehabilitatlon design process.
Various methods currently exist for interpreting and assessing 10M measurements. All these
methods are aimed at providing information about the strength of the individual pavement
layers, or of the relative strength of the pavement as a whole. This information can be used
10
estimate the remaining life of a pavement or provide some indication of the failure
mechanism in cases where pavements are already showing signs of distress. One of the
methods of evaluating deflection measurements is through the use of deflection bowl
parameters(lO, I I). These parameters are used to characterise the shape of the measured
deflection bowl, and provide useful information on the stiffness ofdifferent pavement zones.
A computer program, called IDMP (10M Parameter) was developed in order to process
deflection data and evaluate pavement structural capacity using deflection bowl parameters
(II). Although the IDMP Program has been used with much
available on the ranges of deflection bowl parameters to c1a.'i.'iify pavement structures was
derived mostly from experience in this field (II). Apart from the work done by Horale (10),
very little analytical work has been done to establish the ranges ofdeflection bowl parameters
that would classlfy pavements of differing structures accurately.
1 - 4
INSPECTION
.. 1.
--------..-l .- . -..
(lE~~iie~~~~~~~~~~~ ~ ~~~N~~~A~~~~~T~~I:~-1
__
UNIFORM SECTIONS?
NO
TESTING
-_.
_-_...
-------I ~::-
[ --- .__.. - !
--~~~~~ PAVEMENT
INTO VIABLE
UNIFORM PAVEMENT SECTIONS
[-
STRUCTUR~~CA;;~I;~
ANALYSIS
- ._ J
----------- ---_
I ..
....
IDM
USED
\ I
I -S
A need also exists to investigate the validity of some parameters currently used, such as the
Middle Layer Index (formerly Base Damage Index, or BOI) and Lower Layer Index
(formerly Base Curvature Index, or BCI), as previous work by Maree and Bellekens (12) has
shown that these parameters are not always applicable. This research has contributed much
toour understandfng of some of the IDM Parameters, but more work needs to be done In this
regard.
Deflection measurements do not only find applications in project level investigations.
Currently, most highway authorities employ some sort of Pavement Management System
(PMS)( 13). These systems are designed to provide continuallyupdated information regarding
the condition of all road links within a road network. PMS data is used on a network level
to assist in planning, programming and budgeting of the maintenance plan of a network,
Currently, the evaluation of a road Iink is done by using mostly visual assessment and riding
quality information. Because these types of information do notprovide detailed information
on the structural capacity of the pavement, a need exists to incorporate deflection
measurements at network level in order to obtain a picture of the structural condition of a
pavement as related to the design traffic on that road. There is a definite need to incorporate
some sort of structural evaluation parameter into local Pavement Management Systems. The
directness and simplicity of the direct structural evaluation approach using deflection bowl
parameters make this method ideally suited for incorporation into PMS. However, specific
research should bedone to devise a method whereby deflectionbowl parameters can be used
to construct a Structural Stiffness Index, which can be directly incorporated into existing
PMS's.
ORJECfIVES OF TIlE STUDV
1.2
\ I
The objective of this study is to investigate the relationship between pavement structural
capacity and measured deflection howl parameters. The knowledge gained will be used to
calibrate and if necessary adapt the currently used criteria and methodology of direct
structural classlflcatlon by means of measured deflection bowl parameters a.o; utilised in the
IOMP Program. Specific objectives of this study include the following:
261.\IIlI(wp)
I 6
lnvestigatlng the factors affecting the measured deflections and predicted structural
capacity. Also, investigating the use of deflection measurements in a multi-criterion
approach to pavement capacity assesment,
Ind~lt
Systems.
1.3
This study can roughly bedivided into 3 phases of work, namely (i) background. (ii) analysis
and evaluation of results, and (iii) recommendations on the implementation of results, The
basic structure of the thesis is as shown in Figure 1.2. Chapter I contains the background
and objectives.
evaluation is given.
The relationships hetween the various deflection bowl parameters and pavement structural
capacity were investigated in two ways:
i)
ii)
The results obtained through each of the above-mentioned methods are compared and
discussed in Chapter S. In Chapter S, the results obtained in this study arc also compared
to other proven methods and recommendations are made.
In Chapter 6 the (actors which can have an influence on the measurement of deflections as
well 3.' on the predicted structural capacity are identified and discu!.\ed. The use of other test
methods in a multl-criterlon approach to pavement capacity 3.\.,es.\ment ls also discussed.
I -7
In Chapter 7 the use of deflection measurement at network level is discussed. This chapter
includes proposals on the density of testing at network level, as well as a proposed scheme
for the calculation of a Structural Stiffness Index (551) for incorporation into current PMS's.
I
IMPIRICAl
ANAlY818
(CHAPTER 41
NICHANIITIC
ANALY818
(CHMTER :tI
I
-
- C O M PA RI SON AND
II!VA.l UATION 0'
RII!SUUS
(CHAPTII!R 81
------r'---'
I
L
'
~-----r---
-~';~~ ~T '-O-H-N
c:;u-,,-.
M-.-N-T ...,
AT HIlTWOAI< LIY.l
(C.CAPTIR 11
]
[-~ ..-~g::;~;.;~.:
I~_~~~.p~~~.~_.
--.
__
_________.
-.J
2-I
PAGE
2.1
INTROI)UCTION
2-2
2.2
2-2
2.3
2.4
2-8
2.5
. . . 2-9
2-14
\ I
2 -2
2.1
INTRODUCTION
111~ use and
evaluation of deflection, the majority of methods can be categorised into one of two basic
approaches (14):
i)
li)
2.2
MECHANISTIC OR BACK-CALCULATION
2.2.1
GENERAL
Mechanistic methods, also known as hack-calculation methods are widely accepted methods
of deflection evaluation. The majority of mechanistic design procedures arc based on the
assumption that the pavement layers and supp<lrting suhgrade can he modelled a.'l multilayered elastic or visco-elastic materials, In this way the model of the pavement can he used
to calculate stresses, !Ilrain.' and deflection.' under loading at any point within the pavement
structure.
2 3
The computer simulations needed in order to perform a mechanistic analysis. as well as !he
material modelling and methodology to he folio woo has been the subject of active research
over the past few years and is well reported (17 to 21).
The underlying princtple of all mechanistic deflection evaluation techniques is that there is
a unique deflection bowl for each pavement configuration. Because the deflection bowl is
known (measured), the pavement configuration (l.e. layer moduli) can he determined.
Deflections are therefore used to 'back-calculate' the elastic moduli of each pavement layer.
Once the modulus of each pavement layer is known, a model of the pavement can be
constructed. This model is now used to calculate theoreticalstresses and strains under typical
loading conditions. The calculated stresses and strains arc finally used with transfer functions
to determine the expected structural life of each layer.
The actual hack-calculation process can be done manually through the use of programs such
asELSYM5 (22), CHEVRON (23) and BISTRO (24). The manual process, however, is slow
and requires a skilled operator. For this reason back-calculation programs were developed
which uses raw deflection data as input and automatically back-calculates the layer moduli.
MODULUS (25) and ELMOD (26) are well known examples of this type of program.
Although the actual methodology used in the back-calculation of moduli differs from program
to program, the basic concept used in back-calculations is as follows (5):
Referring to Figure 2.1 below:
E Elaille modulul
, Po I. Ion 'I rallo
P WhMI laad
I Laytr Ihlc"n... (mm)
4 : Oe"tcllon(mm)
FIGURE 2.1
2 -4
The known values ofapplied load P and the associated deflections are applied to estimate I;
and/or ti using an iterative procedure.
distribution throughout the layered system. As is shown in Figure 2.1 a limit or boundary
to the stress is defined. often in terms of the slope z/x of a line with the origin at the load
P. Any material above this line is said not to contribute toward carrying the applied load.
TIle layer thicknesses tt are assumed known. The moduli E, of the layers arc: then estimated
as follows:
i)
Arhitrary moduli for the layers are chosen. and the computer program calculates
initial estimates of deflections (the locations at which the deflections arc calculated
arc chosen so as to correspond with the intersection of the layers and the assumed
limit of stress distrlbution). For instance. 6) would he calculated at the intersection
of layer 3 with the stress limit line shown.
ii)
Calculated deflections (6) arc then interpolated from the measured deflections (6') at
these intersection points.
iii)
OJ and 6,' at the outermost intersections are then compared. All deflection will thus
he attributed to the lowest layer since the point at the outermost section coincides with
the lowest layer. The stress limit line assumption assign all stress to the material
below that line and so also all strain and deflection. The moduli chosen in (i) is then
altered. usually hy multiplication using some variant of the ratio 0,/6j
iv)
This procedure is then repeated using the newly calculated modulus until 0.,' and 6;
corresponds to a specific tolerance. In the layered system shown in Figure 2.1 this
would he 0; and 0,. so that the modulus of the 3rd layer isthen assumed known.
v)
The analysis now moves inward to thenext layer using the known values of E-moduli
of the lower layers as calculated above, until all layer moduli are calculated.
The method described above is utilised in the ELMOD program. Not all back-calculation
program.~
follow this methodology. In practice. good results have also been obtained with
MODULUS which uses a linear elastic subroutine to generate a data base of deflection bowls
and then compares the measured and calculated howls until an acceptable match is obtained,
2-5
Finite: dements have: also been used in the calculation of layer moduli from deflection data
(27), and although its use: is mostly limited to research, there is great potential in the usc of
finite: elements to overcome: problems such as non-linear material behaviour. The problem
of non-linear material behaviour is discussed in more detail below.
2.2.2
Each layer has a finite thickness except for the lower layer and all layers are
infinite in the lateral directions.
(ii)
(iii)
(iv)
(v)
While thislast assumption (v) is more or less true of asphalt, it is not in the case with
granular materials.
Hoffman and Thompson (29) have shown that when a deflection bowl calculated by
means of a linear elastic program is fined to a deflection bowl calculated with a nonlinear finite clement programme, different strengths (E-moduli) for the two pavements
analysed arc ohtained. That
i~.
model, differ instrength from those modelled with the linear elastic programme, even
i~
2-6
Ullidtz (30) states thatdetlectlon profiles derived by means of linear elastic pavement
programmes will differ from those found in non-linear materials due to the increase
of modulus with distance from the load that takes place in non-linear clayey materials.
Ullidtz hasalso shown (30) that the non-linearity of materials was found to be without
influence on the vertical stress of the load, if the non-linearity of the material is of
the type given by equation:
E = C(O)/o')"
2.1
0'
is a reference stress.
Much research has been conducted into thesimulation of granular type materials, and
the modulus of granular materials hasbeen shown to he dependent on the bulk stress
of the material (31, 32). The relationship between modulus and stress for granular
materials is of the following form:
2.2
Where E
Elastic Moduli
Material Constants
bulk stress value
The relationship given in equation 2.2 is used in most finite element programs, but,
as previously mentioned, the majority of mechanistic analysis programs assume the
layer to behave in a linear elastic fashion.
Apart from the linear elastic approximation, any pavement system responds to loads
according to a multitude of factors such
a.c;
!Iliff layers, etc.. When taking all these factors into account it is understandahle that
any computer model of a pavement is at best an approximation, and results obtained
through such a model should he Interpreted with sound engineering judgement
guardian.
a.c;
27
(ll)
Shupe
or contuct area
Programs such as ELSYM5 assumes the: shape of the contact areato he circular, with
the pressure uniformly distributed (22).
simulationof the actual loading area of the 10M, the shape: of the actual contact area
of a wheel load can he circular, elllptical or even rectangular, depending on factors
such as load, tyre pressure and tyre characteristics (33).
Uzan and Sides (33) have found that the effect of contact area shape and pressure
distribution is less pronounced as depth increases. Thus the upper layers are hound
to he most susceptible to the influence of contact area and pressure,
It was also
shown by Uzan and Sides (33) that the assumption of a circular contact area was
adequate for design purposess in the case of thick pavements, although results tended
to he slightly conservative.
(ill)
Dynamic rrrect'i
The 10M submits the pavement to a dynamic load with a duration of approximately
25 to 30 milliseconds. Programs such as ELSYM5 and CHEVRON, however, only
simulate a static load (22, 23). Hoffman and Thompson (34) have pointed out that
there is a need to include the inertia effects of the pavement in a theoretical analysis.
Because of the dynamic nature of the 10M loading, the deflection basin measured by
means of the 10M differs from the basin assumed in a static analysis. Thus the
assumption of a static load made in the mechanistic design method is another of the
shortcomings of the mechanistic design method.
the assumption of a static load hll'i on the mechanistic design method cannot be
accurately assessed at this stage, Sebaaly
et
2-8
2.3
2.3.1
GENERAL
As previously mentioned, the: method investigated in this study is related to the direct capacity
predlctlon technique.
empirically established criteria. Direct capacity prediction techniques all differ in complexity
and in the method the failure criteria is derived. Popular examples of this evaluation method
is the Transport and Roalls Research Laboratory (fRRL) method (35) and the Asphalt
Institute method (36). The failure criteria used in these methods is mostly presented as a
curve, with deflections on the abscissa and the observed number of repetitions to failure on
the ordinate axis. These curves are often derived hy means of test sections of pavements on
which full-scale testlng is performed over a period of time (35). By measuring the surface
deflection before and during testing, and observing the performance of the pavement under
repeated loading, a relationship is obtained between deflection, and the number of repetitions
(axle loads) to failure.
Originally, in the United States, Britain and Europe, the use of deflections in this method was
limited to maximum deflection only (14,35,36). Later developments, such as the method
developed at the University of Illinois (37) use deflection bowl parameters, which also
characterises the shape of the deflection howl. In the Illinois method the deflection howl
parameters are not used
III
stiffnesses, which are used to predict the required thickness of asphalt overlay when doing
rehabil itation design.
2.3.2
Although the simplicity of the direct capacity prediction technique makes it sulrahle for some
application." it has the disadvantage of providing result' which arenot
a.~
a.~
deflection is used.
ca.~cs
accurate or detailed
2-9
One of the most important disadvantages of direct capacity prediction methods is that the
failure criteria is normally derived from field testing on a Iimlted range of pavement types and
for specific loading and subgrade conditions (35, 36). The design curves cannot be used
outside of these ranges without some form of verification (35).
2.4
2.4.1
These
parameters are:
Maximum Deflection (Y-rnax), providing an indication of subgrade stiffness and
overall pavement capacity.
BlI: Base Layer Index, (formerly surface curvature index, or SCI), providing an
indicationofthe stiffness of the surfacing, base and sometimes thesubba.se depending
on the layer thicknesses.
MlI: Middle layer Index, (formerly base damage index, or 801), indicating the
stiffness of the subbase and upper selected subgrade.
Ll.l: lower layer Index, (formerly base curvature index, or
ncn,
indicates the
~J
10
610
91~
1000
1200
STRONG SUBGRADE
(LOW 07)
STIFF BASE
(LOW BLU
DEFLECTION
__
WEAK BASE
(HGH BLI)
POSITION OF SENSORS
200305
(mm)
1200
610
eoo
07
DEFlECTION Y MAX
Lli (FORMERLY
Bel)
1800 2000
2 - II
TABLE 2.1:
FORMULA
PARAMETER
1.
2.
Maximum
deflection
00
Radius of curvature
Benkelman beam
Lacroix deflectograph
= ----C
200(00/6,-1)
3.
Spreadability
MEASURING DEVICE
= 127 mm,
200 mm
[!6..tla+ 6,+6,l/51100
Dynaflect, 10M
00
0, '" 6, spaced 305 mm
4.
Area
= (60 - oJ/6.
= (0. - 0,)/6,
BLI = Do - 0,. where
r = 305 mm
or r = 500 mm
5.
Shape factors
6.
7.
LLI
8.
9.
Deflection
ratio
Q,
Bending
index
BI
a
Slope of
deflection
SO = tan' I(6o"iS,)/r
where r = 610 mm
10.
II.
:1&1.WI (WP)
F.
F,
=6
6 10 -
10M
+ 6)/00)
69 "
= 6'/60 where
10M
Benkelman beam
Road rater
10M
Road rater, 10M
10M
6, - 00/2
= 6 la
c
where
Deflection basin
Benkelman beam
Benkelman beam
2 12
The: above-mentioned parameters BLI, MU, LLI were previously known as SCI, DOl and
BCI respectively. Experience and previous research conducted by Maree and Bellekens (12)
have: however shown that BDI does not apply to the basecourse stiffness, but to the stiffness
of the middle pavement layers, being mostly subbase and upper selected layers, depending
on the: thickness of the: layers. This research has also shown thaI DCI is related to selected
layers and subgrade. The terms SCI, DOl and DCI were therefore: changed to the more
general but also more appropriate descriptionsof DU, MU and LLI.
2.4.2
calculated from the number of E80s the pavement carries and expected growth rate in E80s
during the design period.
In order to perform these calculations the IDMP computer program was developed (II). This
program makes use ofconverted 10M raw data toproducean output as shown in Figure 2.4.
As can be seen from this figure the program provides information on:
Deflections (01 07) normalised to 40kN test load.
Deflection Bowl Parameters (layer stiffness), Y-max, SCI, BDI, BCI.
Equivalent 80 kN axle Traffic Class according to each Deflection Bowl Parameter.
Estimated life expectancy according to each deflection bowl parameter and the
expected cumulative traffic with the current expected future growth rate.
The program output shown in Figure 2.4 presently still use the terms SCI, 1301 and BCI to
denote BLI, MU and LLI respectively. The range of deflection bowl parameters according
to which the pavement is classified depends on the type of flerible pavement. Currently
ranges have been established for the three most common Ilcxihle pavement types, namely:
:> - 13
FIGURE 2.4:
91.0J.2S
PAc;( I
ROAD: SlOlllANE.IWI
PAV[HCNT IYPE: QlAHUlAA IlAS[
W1IHl lOAD: 40.0k"
6.0\
H[ASUIlEO: 91.03.05
I 80 I
AUMO lRAfflC
I CLASS I
(80 a 106)
t--t
I COL I
I
I
I
I
I
I
ru I
III I
[2l I
[211 I
E3l
0.05 0.1
0.1- 0.2
0.2 0.4
0.4 0.8
0.8
1.6
1.G 3.0
J.D
6.0 12.0
12.0 24.0
24.0 50.0
SO.O 100.0
[011
s.e
I [31t I
1
I
I
E4l
E411
[5
LIST
I
I
I
I
I
I
I
1
I
I
10SDIJOO
810IOSO
100 810
500 100
410!lIlO
J?O.410
320 J90
260 320
210 2GO
110 210
O 110
100 900
540- 100
420- 540
320- 420
2SD- 370
200 2!i0
160 200
120- 160
90 120
10 90
0- 70
STATION
30.035
30.045
30.055
30.065
30.075
30.065
30.095
30.105
30.115
I
I
nCI
410
310
230
110
130
100
1555
40
300-
I
I
I
I
I
I
I
I
I
I
464
484
]96
429
29]
420
42]
429
425
294
]01
234
263
242
JI) 258
]24 259
326 256
m Z6S
J6J
]16
290
326
255
I m
Il38SlIVT
19
10
135
1]5 13
108 61
120 66
110 110
121 63
171 64
113 62
122 64
11111
t,---t---I------
I
I
I 30.025 I 486 392 316 ISO
I
I
SCI
I
I
UHIIS (1'1'\)
YlMa
I OCfUCTlOH
I YIl\U SCI
I
48
42
44
40
39
62
40
42
41
44
35 1 486
30 1 464
)2 I 484
26 I )1J(j
301 429
)9 I 293
30 1 420
)) I 42)
)) I 429
)S I 4ZS
44
1
S3
]3 I 4ZS
) 1 5S
31 I 4?6
170
18)
IG2
166
51
162
164
113
160
159
166
126
I4l
12
131
138
143
141
12
IS
91
-.
[STIKAHO W[ (yr)
[60 CLASS
I
I
SCI
801
DCI
001 OCI I YlMa SCI 001 ocr I Yna.
1
110 16G
11 I E2l [3l E2l [2t I 2- 3 6-10 2 3 2 J
IlOlII. PAIW1.
IS6
33
70S
IJ?
11
114
65
62
41
1
I
~I
601
SOl
57 I
51
~I
[211 Ell
Ell [ll
[211 [ll
[211 [ll
[)II [5
[211 [ll
[211 [ll
[211 Ell
E2l [211
E2l 711
[211 [ll
E2L [ll
Oil [211
E2l [211
E2l [711
[ll
2" nil m
)l
I
I
I
I
I
I
2 3 J. 6 I
J. 6
6-10
2 )
6-10 2- ] J. 6 I
I )- 6 6-10 ]. 6 610 I
I 3- 6 6-10 2 3 610 I
I 10IS IS 1015- J. 6 I
I )- 6 6-10 2 ] J. 6 I
I ). 6 6-10 2 ] 3 6 I
1 ). 6 610
610 I
I 3- 6 10IS, 2 ) 61<1 I
. -t
I
581 [211 [)It [lL [211 I
. I
7 I
61 I 2L [ZIt (lH [2M I
z )
2 - 14
i)
ii)
iii)
The 10M parameter criteria currently used in the 10M Program are shown in Figure 2.5.
As can he seen from Figure 2.5, the remaining life is expressed as a traffic class and not a
number of axle loads. This is because 10M parameters are at this stage only used as a
general indication ofpavement strength, The criteria shown in Figure 2.5 cannot be applied
to pavement structures with different configurations without some loss of accuracy in the
number of 80 kN axles predicted, Therefore the IOMP Program provides the designer with
a relative description of the pavement structural capacity, without implying an accuracy that
can only be obtained by means of, for instance, hack-calculation programs when used by an
experienced user.
The IOMP Program has been used with success for a number of projects. In a recent project
involving the M2 motorway south of Johannesburg, the program output clearly indicated
sections of differing expected life and traffic class. Further investigation by means of cores,
dynamic cone penetration and test pits confirmed these results.
2.S
l~)
a:
LLJ)(
9z
Oz
..J_
wi>j
cno
~!:
i~
,
I :
LLJ
..J
o~
~-
~~
t-r+rr;
I _ __
1.1_'
ill;
.J
W
.J)(
.J
-.J
:;:)t;
~w
a:
~-
a:
Z
0
11,r
"
11/
I I i 1/ /,' I v IA" I /
I
I ,1 // IX lA' i )~/}' ,i/,'''
I
I I I
V
,
v,j' 1/
i / /1/ / jl I It ' V
I ! I I' / VIII 1./ ' I "
I
(/)
a::
.-
<t
a::
<t
a..
II
Y /
-oJ
W
Will
111<
< III
III
0'"
o
z
.u
W
.....I
I..L.
a::
o
01 Z
:::
II
<t
.a:
u
'I
I / V
I 'I
JI
I
I
V,
fA
I/l, I / //
I ,
/"
1
i
.....I
'
to
i/f
N
W
0::
, "I,
I
1/.'
1/"
~
'
"
"
,'
I 1/ '
I ,
&
I I
II '
/ II
/Ii I
8
W901
>
<r
...J
II
d :E
::>
::>
'
,I
Ls..
CD
.~
0:
W
.J
ow
V
"I
,
.:::>
. ~'"
S~313VV\1~\1d
CO
II
x:
V V~
I
,
J ,
I
Jlil
'"
t(?
..J
1/,
(f)
1'\
' / I'
<{
_L Y
,,
1/ /
I , " /1 I I
I
"
1/ / ,
. I,' 1/(
1/
I.
II /
I' 1/ " I
I ~V
II
~
.-
u,
II
1//" I I
I~j, ,I/1/;
'
/
/1
/
, !I' 1/, f
/~ Ii! II ~/
,
,
(!)
(!)
1/
Ii,
II
/
I 1/ 1/
It'
If
)(
, II ,
I I
/,' /f' l V
I
I
11
I' I
II' I
v,
I , II I I
I / / I I II
/
Ji
I
I
I
I
a::
" I l l III
II
v,
~W:r oJ
III
o:!:~
/.' I / /
I/'//
01 ;)
I
I
I,
I I I/ /~'
w .... >
I
I
1/ ,
I I
oJ
0:1:
,II
I..L.
::>
~l'l
w
w
ID
I I
~
(SNO~:::>IW)
NOIl:)3"B30 03~nS'13W
:E
o::>
2 16
i)
ii)
Errors in calculations can occur if incorrect layer thicknesses and material types are
used. The mechanistic approach therefore requires fairly detailed information on the
existing pavement structure. This type of information can in most instances only be
obtained bydestructive testing methods such as trail pits and cores.
The factors discussed in (i) and (ii) above currently renders the mechanistic approach
unsuitable for applications where limited expertise is available, or in network. appltcations
where often very little information regarding the pavement structure is available. The need
was therefore created for a fast and easy to usc method, which would ultimately he suitable
for both network and project level investigations.
By using deflection bowl parameters in the direct capacity prediction technique, more
information can be obtained from a single test point than what can be gained from only
maximum deflection. At the same time the method in which the deflection bowl parameters
are utilised is simple enough to he applied to network applications.
It istherefore clear that both mechanistic and direct capacity prediction methods have definite
rehabilitation designer which method to apply to each type of project. This decision is most
likely to he influenced by the resources available, such as time, finances, qualification and
expertise of personnel as well as material information available.
3- I
3.1
3-2
3.2
3-2
3.3
3-5
3.4
3-5
3.5
3.6
3-9
3 2
3.1
The relationship between the various deflection howl parameters and the structural capacity
of individual pavement layers is investigated in this chapter. Previous research (12) has
shown that some detlection howl parameters such as MLI and LLI have anunacceptable high
scalier, which leads to a high degree of uncertainty when making capacity predictions based
on these parameters, In this chapter these findings arc investigated further and reported on.
An attempt is also made to confirm the relatlonshlps between the different pavement layer
capacitiess and different detlection howl parameters. Also, the usefulness ofparameters which
are currently not widely used, such as radius of curvature (RoC), shape factors f I and F, and
spreadability (SPD) are investigated and reported on. finally, the analysis results are used
to derive parametercriteria to he appl ied to granular base pavements.
3.2
The analysis was done through the use of the ELSYM5 computer program. This program
uses the layered linear elastic theory in order to calculate displacements, as well as stresses
and strains at given depths in the pavement (22). In order to obtain a relationship between
the 10M parameters and the number of axle loads, the approach showed in figure 3.1 was
adopted.
Figure 3.1 shows that 2 simulations were necessary. One of these simulations (STEP A) had
to resemble the loading of the pavement ali for a typical 10M test, that is, using a 40 leN load
applied through a 300 mm diameter base plate. The other simulation (STEP B) was that of
a 40 kN double wheel load (80 leN axle load).
Figure 3.2 shows the dimensions and details of the 2 load situations. The only information
U500
from the 10M simulation was the surface deflections as would normally be measured
hy the velocity transducers of the 10M. The 40kN double wheel load simulation was used
to obtain litre...sc..o; and strains directly under onewheel a' well a., in the centre of the 2 wheels.
These stresses and strains were then analysed according to the South African Mechanistic
design method proposed hy Marce and Freeme (39), in order to evaluate the structural
capacity of each layer.
3 - 3
STEP B
STEP A
AND COMPOSITION
(BASE T Y P E ,E -MODULI,POISSON)
SIMULATE 40 KN
SIMULATE 10M LOAD
WITH ELSYM6
-1 ~ ~ .~
(FIGURE 2.4 a )
OBTAIN STRESSES
AND STRAINS AT
OBTAIN DEFLECTIONS
--I
DIFFERENT DEPTHS
01 TO 07
ESTABLISH
CALCU~ATE
REMAINI NG LI FE
10M PARAMETERS
THROUGH MECHANISTIC
DESIGN METHOD
:t - .,
/
610 mm
916 mm
1200 mm
1600 mm
x
SENSOR POSITIONS
176 mm
x
175 mm
()
3-5
Table 3.1 shows the positions where critical parameters were calculated as well as thll
relevant parameter.
TAIlLE 3.1:
LAYER
Asphalt Surface
Granular Basil
Select,.o. d Layers
Subgrade
PARAMETER
Horizontal Strain
Safety Pactor>
Safety Factor"
Vertical Strain
POSITION IN LAYER
Bottom
Centre
Centre
Top
For this study, only granular base pavements were analysed. Structures with a 'cemented
subbase were included in the analysis. A number ofstructures of varying base thickness and
material qualities were simulated to obtain a set of results to be used for the regression
equations.
For example: A base thickness of ISO mm was used with varying material
qualities for the different pavement layers, after which the material strengths of all the layers
were again changed for a base thickness of 125 mm. This procedure made it possible to
obtain a set of 43 values relating remaining life to each 10M Parameter. These values were
used to obtain regression equations for each 10M Parameter. The relationships between the
simulated deflection bowls and calculated life of each layer was also obtained from the
results.
3.4
3.4.1
GENERAL
An important assumptlon made in the choice of material strengths W;L\ to regard the structure
3.\
'structurally balanced'. That is to say that all layers of the pavement have settled through
the action of traffic. and that the strengths of the different pavement layers have changed in
order to obtain a state of balance between the load and pavement load carrying capacity.
Referring to Figure 3.3, this process consists hasically of 6 steps (2):
3-6
Step I: The: load is spplied to the: pavement. The overall pavement structure stays unchanged
if no pavement layer is overstressed,
Step 2: If certain layers of the: pavement an: overstressed, these: layers will undergo further
compaction, or will shear. This will also lead to an accompanying deformation in the
overstressed layer.
Step 3: Because of the compaction/shear in the pavement layer, the Elastic Moduli of the
pavement layers change. This also leads to a change in overall pavement strength
(pavement becomes either weaker or stronger).
Step 4: Because of the change in Elastic Moduli of the different materials in the pavement
layers, the stress distribution due to the applied load will change (according to
classical soil mechanics). For example: if the increase indensity and modulus of a
certain layer leads to a higher load carrying capacity for that layer, that layer will
also tend to absorb more of the total stress,
Step S: The compaction and/or shear of the material as noted in Step 2 causes larger bending
of the overlying materials.
Step 6: If the material can safely accommodate the traffic loading and increased stress, the
pavement will notundergo any further deformation and layer Elastic moduli will stay
more or less unchanged. The effect is that the overlying materials will duplicate the
deformationof theunderlying material toa limited degree, depending on the stiffness
of the overlying layers.
Thus the structures used for the mechanistic analysis were assumed to he compacted and
settled to a relatively stable state under the action of the traffic. This implies that certain
limiting criteria had to he applied to the input characteristics of the materials.
3 - 7
NO
J_
'-
PAVEMENT STABLE :
NO CHANGE IN LOAD
( IS MATERIAL OVERSTRESSED?
~-'---"--
CARRYING CAPACI T Y
AND LAYER MODULI
-. 0------ ----'------=r---------------~:~::;TRS~~:~::SA::ER~~~M
G;
BALAN~~;E~l
PAVEMENT
RE-ORIENTATE
---------_.------------
OVERLYING MATERIALS
r.-.
-------------r------- -. -------
---------
IS OVERLYING MATERIAL--)
------- _._---_._-----._._-
OVERSTRESSED?
YES
(6'
.-------
NO
PAVEMENT STABLE:
PAVEMENT HAS DEFORMED
AND LOAD CARRYING
CAPACITY I SENSITIVITY
AND LAYER MODULI HAVE
CHANGED
j
'------------- --
__ ..
_.
-_.
-._---. _.-
_.-
-- _.---
._.-
-_.
3-8
3.4.2
I)
To comply with the: condition that all material were already in a steady state, ali a
general rule: no layer was given a stiffness more or less than twice: that of an adjacent
pavement layer l.e.: E, not greater than 2 x E, ,
or E,
n()tll:.~s
than 'h E, ,
where E1 is the Young's modulus or stiffness of the ith layer. This condition did not
apply in the case where a cemented layer overlies or uaderlles a relatively weak
layer, or in the case of a stiff asphalt layer overlying a weaker basecourse,
II)
The mechanistic design method proposed by Maree and Freeme (39) uses values
(such as angle of friction of material, 41; and cohesion, c) dependent on the type of
material classified according to the TRH 14, Manual for Classification of road
constructionmaterials (40), in order to calculate the safetyfactor related to allowable
axle loads. Because of the uncertainty involved in the classification of materials
already recompacted under traffic, it was decided to use the following classification
throughout the analysis:
a)
h)
3-9
III)
Cemented material'!
a."
high
Iv)
The relationship between a.sphalt tensile strain and remaining life isshown for asphalt
surfacings in Figure 3.4.
v)
A'iphnlt propertles
As can he seen from Figure 3.4 three curves exist for asphalt surfacing relationships
between tensile strain and remaining axle loads to failure. In the case of asphalt
surfacings, the percentage of voids in asphalt is needed. In all cases the average
curve was adopted. That is, for asphalt surfacings the percentage voids was taken as
5%.
3.5
BASIN
3.5.1
,.1.Wl(1tt1')
GRADED ASPHALT
t.:
;1.
I
. I
103
I
i
.I
10
10 6
1 I
I I
107
10 8
-..I
109
3 - II
II was found that a clear relationship between Ih.: deflection bowl parameters and structural
capacity (or remaining life) was not achieved for most of the deflection basin parameters,
In th.: following cases, however, a deli nite relarlonshlp was observed:
3.5.2
i)
Ii)
iii)
Iv)
v)
REGRESSION RESULTS
Alinear regression wall done for each of the cases mentioned in (i) to (v)in paragraph 3.5.1.
The results of these regressions are shown in Table 3.1. For each regression outliers were
first removed before the regression was performed (a maximum of two outliers were
removed).
TABLE 3,2:
10M
PARAMETER
LAYER
CONSIDERED
NO. OF
OBSERVATIONS
CORRELATION
COEF. (r)
Bli
41
0,82
0,67
Y'mal
40
0,84
0,71
Mll
40
O,IS
0,73
Lli
40
0,17
0,75
41
0,94
0,89
Rad o(Cu/\,.
(.)
COEF.OF
DETERMINATION (R~
Note: i)
R' explains how much the tot.1I variation in the data is explained hy the regression
line. When R'
ii)
III
The correlation coefficient (r) b equal to 1,0 when all the x and y value." faJl on a
straight line. When r == 0,0 there b generally no associatlon between y and x (41).
3 - 12
3.6
The observations and conclusions drawn from the mechanistic analysis are in line with
previous research. The most important observations are as follows:
a)
The best correlation obtained was that of the Radius of curvature (with r taken as 200
rnm) versus the remaining life of the surfacing. It can thus be concluded that the
Radius of curvature is a good indicator of relative surfacing life, expressed in
allowable number of E80s. TIle fact that the correlation of this comparison (Rad. of
Curv. vs Surfacing Life) is the highest, is most probably due to the effect of nonlinear materials ondeflections, which is less pronounced closer to the load. Because
the radius ofcurvature is calculated using only the first two deflections, the effect of
the non-linearity of the suhgrade and/or basecourse does not have a large influence.
h)
It was found that the Base Layer Index (OLl) is strongly related to the E-moduli of
the uppermost pavement layer. The life of the surfacing is also related to the Emoduli of the basecourse as can he derived from Figure 3.5. This finding is in line
with the research conducted on the effect of Overlays on deflection howl parameters
(12). One oftheconclusions drawn from this research was that the E-mooulus of an
overlay influenced mostly the BLI, while other parameters are only affected to a very
limited degree.
c)
Y-max showed a relation with the remaining life of the selected suhgrade. Previous
research (12) has also shown that the maximum deflection (y-max) is highly
dependent on the Ernodulus of the subgrade and selected layers. The association
between LLI and selected laycrlsuhgrade life as well as that of MLI and selected
layerlsuhgrade life i5 also a.o; expected.
)61. YOI(I\1')
3 - 13
FIGURE 3.5: RELATIONBETWEEN BASECOURSE
MODUU AND REMAINING UFE OF SURFACING
1000,----
---,
---.
--- ..
l00t:---r---r-.....-r--r...,..T'"T...,.---r--r-...,...........
..,.-,~
10oo
1000ooo
ALLOVv'ABLE EOO'.
d)
10ooooo
The observation that no association was obtained between any deflection howl
parameter and thestrengths of layers 2 or 3 can most likely be attributed to the nonlinearityof these layers, which is not taken into account with the ELSYM5 program.
As already mentioned in paragraph 2.2,2 (i) there are some indications that the nonlinearity of the material does not significantly influence the stress induced by any
load. Thus the overall behaviour of the pavement as a system is accurately assessed
by the ELSYMS program, although no clear picture is developed ofthe behaviour of
any granular layer alone.
e)
The regression analysis done was used 10 obtain equations relaling the remaining life
of a pavement layer
parameter relatea
10
10 a zone
3 14
TABLE 3.3:
PAVEMENTSX
10M
PARAMETER
RELATED
ZONE
COEFFICIENT
CONSTANT
(K)
(C) ..
Selected + Subgrade
Surfacing + Basecourse
Selected Layers
S~I~ctcd Layers
Surfacing
-179,8
-144,0
-39,8
-27,1
22252
2191,S
1284,7
458,4
283,3
-84774
Y-max
ALI
MU
LU
Rad. of Curvature
The equations shown in Table 3.3 were used to obtain the curves shown in Figures 3.6 and.
3.7. In Chapter 4 similar curve..s are derived from the analysis of observed field results, The
results obtained by means ofthe mechanistic analysis are compared with those of the analysis
of field ress ults in Chapter S.
FIG 3.6: MECHANISTICAllY OERNED 10M
PARAMETER CRITERIA (GRANULAR BASES)
1400,-------
--.
1200
MAXIMUM DEFLEcnON
1000
MIOOlE
LAYER
INDEX
=\====----~:=::::::::
LOWER LAYERIN:OX:7::::::i,.------------====~~
1 .05
1 .08
1 .07
CUMUL NUMBER OF EOUIVAlENTlIOt<N AXlES
1 .08
:I - 1!J
80
70
RADIUS OF CURVATURE
~ 80
i=
0(
it
a 50
!a
40
30
20
10
lE+04
I +05
I +06
CUMUL NUMBER OF EQUIVALENT 80KN AXLES
I +07
I +08
4-1
4-2
4.2
4-3
4.3
4-7
4.4
4.5
420
4-2
4.1
In Chapter 2 the limitations of the mechanistic design method was discussed in some detail.
To summarise, the main disadvantages of the mechanistic design method by means of
programs such as CHEVRON and ELSYMS are as follows:
i)
ii)
failure of the procedure to take account of the dynamic effects induced by the 10M
loading.
Apart from these factors, there is also a large number of factors such as moisture content,
drainage conditions, presence of cracks, etc. that are not incorporated in the program
calculations. Because of these limitations the results obtained by way of the mechanistic
analysis method should he interpreted with care and sound engineering judgement.
One of the most logical ways of overcoming the limitations inherent in the mechanistic design
method is to go hack to the field situation and observe and analyse the measurements taken
under actual loading conditions. This step provides the designer with a valuable relationship
between the computer model and the actual field situation. The analysis ofthe observed field
results will then either verify and calibrate, or totally discredit thecomputer model, in which
case the analysis results of the field data alone will be adopted. The knowledge gained can
also he used to improve the computer model for future use.
The aim of this chapter is to describe the analysis of field results obtained by 10M testing
done on a number of roads of varying types of constructions. The results obtained by this
"Empirical- analysis was then used to determine as accurately as possible the observed ranges
ofvalues of 10M parameters applicable to different types of pavements. Finally, these results
are compared with the results of the mechanistic analysis in Chapter S.
recommendations are also made.
Some
43
4.2
The 10M data base consists of a relatively large number of observations recorded by field
testing with the 10M where one observation isrepresented by the 10M Parameters calculated
from a field measurement. The main constraint in developing a data base of this type is the
general lack of information on the pavement construction. This means that some use has to
be made of "as built" data, which has often In the past proved 10 be inaccurate. However,
in some cases fairly accurate information exists on pavement layer thicknesses and material
Iype-li from test pi! data, cores and DCP's.
The basic information recorded in the data base is as follows:
i)
Road Name
ii)
iii)
Base Type
iv)
v)
The data base is continuously being upgraded as more testing is done and more information
is received. The number of records currently available in the dala base for granular base
pavements far outweighs that of asphalt and cemented base pavements. Currently the total
number of granular base pavements in the data base numbers approximately 30, depending
on how and whether outliers are identified and removed.
The main problem with the granular base pavement data base is the lack of data on high
Iraffic design constructions, such a'i those falling in the E4 traffic class according to TRH 4
(38). Table 4.1 shows all granular base pavements currently in the 10M data ba.se.
1. 'IIOI(fo?l
Figure
4-4
TABLE 4.1:
P206/1
MI. Johannesburg
N3 Section 12
PISS/I
N2 Sections 7X & 8X
Falck St, BFN
Monument SI, nFN
Dohsonvllle Rd
M2, Johannesburg
PIIII
HP8/6
John Meinen SI, WHK
Crompton SI
HPI/9
PI54
HP8/3
HPI/3
Bach St, WHK
Knudsen St, WHK
Paul Kruger St, PTA
Main St, East London
Parson St, WHK
HPI/8
Bulow St, WHK
Silwer St, JHB
Dobsonville Rd
Rural Rd, Mozambique
TRH 4 DESIGN
TRAFFIC CLASS
E4
E4
E3
E3
E3
E3
E3
E2
E2
E2
E2
E2
E2
EI
EI
EI
EI
EI
EI
EI
EI
EI
EO
EO
EO
EO
EO
CONSTRUCTION
TYPE (.)
I
I
3
3
2
4
4
5
6; 7
8
8
7
7
10
10
10
10
10
10
10
10
10
II
13
13
12
12
Refer to Figure 4. I
It should be noted thai thepavement structures shown in Figure 4.1 are founded on subgrade
10 the
recompacting of subgrade material to a depth determined by the CDR tested on the suhgrade
material. Details of subgrade preparation can be found in the TRH 4 d~ign manual (38).
~1.YOI('Io?l
4 - 5
CONSTRUCTION TYPE 2
40 A
160 01
150 01
126 C3
150 C3
126 C4
150 C4
CONSTRUCTION TYPE 4
CONSTRUCTION TYPE 3a
50
30-40 A
150 03
260 01
150 06
100 C3
TREATED LAYERS
CONSTRUCTION TYPE 6
CONSTRUCTION TYPE 3b
30-40 A
150 02
150 02
150 06
150 C4
0-_- - - - - _ . _ -
S INDICATES S2 OR 84
.... -
4 - 6
CONSTRUCTION TYPE 8
S OR 30 A
125 G2
150 G3
150 G5
150 G5
CONSTRUCTION TYP E 9
S OR 30 A
125 G2
CONSTRUCTION TYPE 10
S
150 G3/G4
125 C4
150 G5
CONSTRUCTION TYPE 11
CONSTRUCTION TYP E 12
100 G4
150 G5
150 G5
150 G5
CONSTRUCTION TYPE 13
S
100 G4
MATERIAL TYPES REFER TO TRH 14
A INDICATES AG,AC OR AS
125 G5
S INDICATES 82 OR S4
4-7
4.3
4.3. I GENERAL
In order to obtain a relationship between structural capacity and measured deflection, the
structural capacity corresponding to each deflection bowl in the data base had to be
detennlned. It should be kept in mind that it is neither practical nor possible to predict
pavement structural capacity with absolute accuracy, given the many variables that cannot be
accurately quantified. For this reason, the normal design capacity as stated In TRH 4 is
expressed in 400% intervals, which does not imply a high level of accuracy (38). Design
traffic is also normally presented on a log scale.
The structural capacity assigned to each deflection bowl was based on two fundamental
approaches:
i)
Structural capacity is directly related to the structure and material types of the
layerworks above the suhgrade.
ii)
Firstly, a
48
4.3.2
the upper end of the allowable EgO range of the design class assigned to the pavements on
which they were measured. For example: If a pavement structure indicated that it was an
E3 Class pavement, all dellection howls measured on that pavement would be a.ssigned a
structural capacity of 12 million E80s. (Upper end of E3 rangeof 312 million E80s). The
reasons why the upper end of the dess ign class was chosen as opposed to the middle or lower
end is a.s follows:
i)
All de.signs in the TRH 4 document were analysed mechanistically, and designed to
accommodate theupper end of the range of E80 values assigned to each design (38).
The TRH 4 document press cribes 300 mm of selected subgrade layers (38).
ii)
In
practice, these layers arc often constructed to a higher standard or thickness, which
leads to improved structural capacity, and lower deflections being measured. Maree
(32) has shown that typical granular base pavements most often fail in the upper
layers and not due to lack of cover over the subgrade. It is therefore realistic to
a.ssign an upper end of range value for the traffic class estimated on the basic
information.
The factors mentioned in (i) and (ii) above were shown by experience to apply to pavements
in SA. In other countries, different design philosophies, local condition.\ and construction
practice might also have justifiec.l using a middle, or even lower end of the allowable range
of ERO!>.
4-9
4.3.3
ADAPTION
FOR SUBGRADE
CONDITIONS.
VARIATIONS IN
MATERIAL
Because
overall pavement structural capacity is best indicated by maximum deflection. this parameter
was used as an indicator of possible deviation from the assigned structural capacity due to
variations in subgrade conditions. material properties and layer thicknesses. This was done
by considering the average maximum deflection and the distribution of all maximum
~)
class higher or lower than the cla.ss a.ssigned on the basis of pavement structure alone.
Therefore. the average maximum deflection of each class was assigned the upper end of the
range of E80s allowed for that specific class. Maximum deflection values which deviated by
one standard deviation from the average for that class were assigned thecapacity of the class
just higher (for lower dellection.s) or lower (for higher deflections). Using the average
maximum deflection and average standard deviation values with their assigned structural
capacitiess a.s data points, a function relating maximum deflection or structural capacity was
derived fur each traffic class, Because this adaption process wa.s applied toeach traffic class
separately, the adapted structural capacity was still realistic with respect to the actual
pavement structure. The exact way In which the function relating y-max tostructural capacity
wa., derived for each class, is illustrated by example below. Figures 4.6 and 4.7 illustrate
the procedure graphically.
~I.
YQI(Wr')
4 - 10
I I-r-, IT I I
....
d,p'h
b.low
..
...
...
...
IU~'AC.'"
a.I'
'url.c.
'
." ...
""counl
... ...
""
.... . ..
II
.. ,
.lHlMC.'"
d.p.h
b.low
lurl.c,
..
It"
aA . . c_tI
..
.ut....
U .....
...
IILlCTID
...
L"'~'
- - ' III.
'U.OUDI
--------
.ut.UDI
....
...
i-rtr I
II"
....
.. ,
d.p'h
b.low
d.p'h
b.low
IUrlac,
.urlac.
,"
tI
...
..
11
...
...
..
utt~"1
.:.-.
,.
,., "
1
..
_.....
- .__
..... ...
4-11
f- --- ---_ ..--. -.--- '-..' -- ..-- -- FIGURE 4.6: ALLOWABLE E80 CLASSIFICATION PROCEDURE
EXAMPLE: E2 CLASS PAVEMENTS
(.0
Pi\'/~
CIA~;~;"
I'AVIMlNfS
'j
/[- - - - -.- -.- -.- - '.- - ' -, -.- -,._- - -_._- - . . _-_ .,.-J
('~
POINT~~
ALL TEST
IN I:)
-"
....,,_....
--_._._~----_._-"'-_
(3\
CIA~~~;
PAVEMENTS
- . -,_ __ ._. ....
OF EOO flANGE
_.
__ .- _._."... _.
--~,,------_._--_._._._ ..
or
UPI'
i,; E
or f:'
------~----_
..
..
--------
~;;_.J
..._...._L
--_._~---~_
I ---------.---
I....-._-------.. ---.---
'-----_ ..
_-_._-~--
Y M,\
It,) (Ii)
10
AS~~I(jN
I~)
AL.LOWAll[ f
Ct. A~;S
... _.".
'-------~_
l-
r.no
~'.-_.-
._..... , . .
_-----
...
_ _ ..... J
4 - 12
MEASURED V-MAX
POINT A
POINT
(UPI'I I: I Ill)
1)1
11 \:1
AS~;.
:;II'!lI V)
AVG
ron
Y MAX
~)
1()
nr v )
1000
AVG STD,DE V
AVG
AVG,-
s t n DEV
0.1
LLU
.n
r.,
ALLOWABLE E80
'----------_._----_ _- ..
tS
J....J...UUL..!_L..1-J-Lj
10
(MILLIONS)
100
4-13
For example, for the E2 Class pavements the procedure was as follows:
i)
Step I :
i1)
St~p
iii)
Step 3
The average calculated in Step 2 was now plotted against the upper
end (on a log scale) of the range of E80s allowed for the E2 class
pavements (3 million) to obtain point A in Figure 4.7.
iv)
Step 4
v)
Step 5
The standard deviation was now subtracted from the average and this
value plotted against the upper end of the range of E80s allowed for
the classjust higher than E2. That is, the upper end of the E3 class
(12 million E80s). This is point C in Figure 4.7.
vi)
Step 6
The points A, 8 and C were now connected and the equation of the
line connecting these points was derived. This provided a function
or the form:
Nm A(Y-max)l ta
......
Where N
III
Allowable EROs
and A, 8
III
Constants
4.1
4-14
vii)
Step 7
Equation 4.1 was now used to assign values of allowable EROs to all
denection bowl parameters of pavements judged 10 be wllbin lbe E2
class.
The procedure described above was repeated for each class of lraffic. Thus a very low
maximum deflection could causethe allowable E80s for lbat mcasuremenlto be that of an EI
class pavement, even though the pavement nructure Is that of an EO. This procedure Is
particularly useful in compensating for roads with weak pavement layer designs but that are
founded on very sliff subgrade or bedrock.
4.4
EMPIRICAL ANALYSIS
4.4.1
GENERAL
As previously mentioned, the purpose of theempirical analysis was to verify and/or correct
the results of the mechanlstlc analysis, as well as to attempt to calibrate the current criteria
applied to the 10M parameters. The most important information needed from the empirical
analysis could therefore be summarised as follows:
i)
A verages, maximum and minimum, and other statistical Information for each
Different band widths of lbe different 10M parameten In order to estimate the
amount of -scatter- of each parameter, as well as the type of distribution of the
measured parameters.
III)
Relationships between the number of allowable ESOs of a pavement and the different
10M parameten.
4-IS
After the measured deflection parameters were dasslfied as described in the previous section,
the data was lmponed into a computer spreadsheet facility and a statistical analysis computer
program for analysis. The statistical computer program (Number Cruncher Statistical System,
or NeSS, Ver S.03 (42 enables the user to do detailed statistical analysis of up to 2S0
different variables and 32 000 observations. The results were then summarised numerically
and graphically, and analysed. Finally, a regression was performed on parameters showing
good correlation with structural capacity. The information obtained by analysing all 10M
parameters falling In a particular traffic class provided an idea of the numerical values of
10M parameters associated with each particular design traffic dass. The results of these
The units of all 10M parameters shown in Tables 4.2 to 4.7 are micron,
except for Radius of Curvature, which is given in metres.
4.4.2
4-16
TABLE 4.2:
10M
PARAM
MEAN
MEDIAN
COEFP
ICIENT
OP
VARI
ANCE
ETER
Ymu
BLI
MLI
LLI
212
9S
44
20
194
91
39
17
0,50
R.orCulV.
275
261
TABLE 4.3:
MAXIMUM
VALUE
MINIMUM
VALUE
UPPER
LOWER
CONP.
UMrr
CONP.
LIMIT
90"
90"
0,51
4S9
203
172
61
104
'0
IS
3
301
123
74
34
131
66
22
I
0,27
SSG
67
367
192
0.32
0,26
10M
PARAM
ETER
Ymu
BLI
MLI
LLI
R.orculV.
TABLE 4,4:
10M
PARAM
ETEI
Ymu
BLI
MLI
LU
R O(CUIV.
MEAN
MEDIAN
COEPPICIENT
OF
VARIANCE
MAXIMUM
VALUa
LIMIT
41
10
31
669
461
205
13
193
0,44
675
292
132
215
13
LOWER
90"
90S
CONP.
0,32
0,32
0,44
0,46
312
131
92
35
UPPER
MINIMUM
VALUE
CONP.
LlMrr
454
119
149
57
199
93
45
16
24
330
130
97
MEDIAN
COEFF
ICIENT
OF
VARIANCE
MAX
VALUE
MIN
VALUE
UPPER
.90.
CONP.
UMrr
LOWER
90"
CONP.
LIM1T
954
570
314
100
1(1)
55
17
4
134
])0
lOS
39
0,36
0,43
0,46
0,47
606
116
108
36
131
120
0.56
517
32
2)4
414
203
116
403
I"
66
...
17
63
4-17
TABLE4.5:
10M
PARAM
MEDIAN
ETEI
COEFF
ICIENT
OP
VARIANCE
MAX
VALUE
MIN
VALUB
Ymu
5n
BLI
MLI
LLI
293
134
42
415
2j7
121
39
0,42
0,50
0,47
0,55
1431
710
451
174
113
69
76
II
0,65
471
R. or
Curv.
TABLE 4.6:
10M
PARAM
ETER
LOWER
901
90_
CONP.
UMrr
CONP.
LlMrr
256
21
4
114
499
213
66
19
179
36
135
57
17
MEDIAN
COEFp
ICIEton"
OF
VARIANCE
51
165
45
0,31
0,39
0,41
0,54
57
49
0,51
Ymu
700
BLI
MLI
LLI
413
177
R orculV.
UPPER
669
396
MAX
VALUE
MIN
VALUE
UPPER
LOWER
901
90_
CONP.
L1Mrr
CONP,
LIMrr
222
21
10M
636
213
II
22
19
91
28
1m
210
910
561
179
92
257
393
so
The same statistical information obtained for each parameter according to each traffic class,
was also obtained for thedata base as a whole. This information is shown in Table 4.7.
TABLE 4.7: STATISTICAL DATA OF MEASURED IDM PARAMETERS: GRANULAR
PAVEMENTS: ALL DESIGN TRAmC CLASSES
IDM
PARAM
MEAN
MEDIAN
ETER
COUF
ICIENT
OP
MAXIMUM
VALUE
MINIMUM
VALUE
VAAl
UPPER
LOWER
901
90_
CONP,
UMrr
CONP.
UMrr
ANCE
Ymu
BU
MU
W
Rad. or Curv.
421
219
III
37
In
313
174
100
34
133
0,52
0,64
0,59
0,57
0,14
1677
910
561
179
67'
97
41
10
3
19
727
421
19S
64
219
117
II
31
..IJ,
4-18
After the statistical information on the averages, distribution and spread of the parameters
were obtained, the relationship between each 10M Parameter and the allowable number of
EgOs was studied. This was done by means of the statistical analysis package's Multiple
Regression facility.
It was observed that all parameters are normally distributed with a degree of positive
skewness. In order toensure that this skewness was not too high for the data to be assumed
as normally distributed, Bowley's Coefficient of Skewness was calculated. The calculated
coefficients are shown In Table 4.8.
TABLE 4.8:
DISfRIBtTrION:
OF
PAVEMENTS
10M PARAMETER
BOWLEY'S COEFFICIENT
OF SKEWNESS
Y-Max
BLI
MLI
LLI
Radius ofCurvature
0,06
0,30
0,11
0,11
0,12
In Table 4.8, a Bowley coefficient larger than O,S would Indicate that caution must be applied
when drawing conclusions from parametric tests or performing regressions. As can be seen
from the figures inTable 4.8, the degrees of skewness in the distributions of the observed
10M parameters are well within acceptable limits.
As previously mentioned, aclear straight line relationship wasobserved between some 10M
Parameters and theallowable number of EROs accorded to those parameters when these two
variables are plotted on a 10i-Iog scale. Thisstraliht line relationship was used to perform
a least-squares regression for each of these 10M Parameters. This procedure provided a
function relating each 10M parameter to the allowable number of EBOs. This function Isof
the following form (41):
4-19
.. . ... . . . . . . .
(4.2)
Where:
YJ
The Correlation Coefficients and Coefficients ofDetermination obtained from the regressions
are shown In Table 4.9. For each regressionthat was performed, the residuals (that Is. the
difference between the actual Y -value and the Y-value estimated by the regression line) were
studied In order to ensure that no patterns were present. as a pattern In the residuals could
mean that there were Influences present that the regression line did not take Into account. In
no case did a plot of the residuals indicate a need to discard the regression lines generated.
Because least-squares regression is highly sensitive to outliers, the residuals of each
regression were studied in order to identify outliers, which were removed. After this the
regression was performed again. The correlations shown In Table 4.9 are those of the final
regression. The regression constants obtained by the least squares regression are shown In
Table 4.10.
TABLE 4.9:
DOs
10M PARAMETER
V-Max
BLI
MLI
LLI
Radius of Curvature
CORRELATION
COEFFlCfENT (r)
4J,97
4J,93
4J,89
4J,74
+0,86
COEFFICIENT OF
DETERMINATION <R')
0,9S
0,88
0,80
O,SS
0,74
4-20
TABLE 4.10: REGRESSION CONSTANTS
OBTAINED BY
LEAST
SQUARES
10M PARAMETER
INTERCEPT
(C)
V-Max
BLI
MLI
LLI
Radius of Curvature
4,089
3,960
3,736
2,894
0,223
NOTE:
SLOPE
(K)
.(),23S
.(),263
.(),274
.(),216
+0,288
a
Units for Radius ofCurvature arc metres, all other parameters in micron
The regression coefficients shown in Table 4.10 were used to generate the curves shown in
Figures 4.8 and 4.9. The curves shown in Figures 4.8 and 4.9 indicate which number of
E80s could be allowed for a particular value of, say, V-max or BLI.
4.5
All the 10M Parameters are normally distributed to an acceptable degree. This was
found both in the analysis of measured data within a traffic class as well as in the
analysis of the data base as a whole (all traffic classes included). The analysis of the
observations falling In all traffic classes showed that there was a mwure of positive
skewness. Bowley's coefficients of skewness indicated thatthe data can stili be safely
classified as normally distributed (Table 4.8).
b)
An analysis of the spread of each 10M parameter Indicated that Ihe spread of LLI
was the largest. The respective spreads of Radius of Curvature, MLI and BLI were
very similar, with BLI showing a slightly narrower distribution In most cases. The
spread and variation of Y-max was thesmallest. although this can partly be attributed
to the fae:t that Ymax was ecnsldered In the classification of the pavements according
to allowable ESOs.
4 - 21
FIGURE 4.8: 10M PARAMETERS va ALLOW.
Eoo's : OBSERVED FIELD RESULTS
100
1( H - - - r - - - r.........- rn!'TT""-..,.--,.--r-lr"T"T'TTT"-..,--r--r-r"T'T",..,.,...-..,.-.,--,r-T"",...,..,"T"4
1 +07
1 +015
1 +015
1Et04
AllOWABLEEGO'S
RADIUS OFCURVATURE
lo+---,.--r"I'""T'TTT-rr--"'-"--r'T"T'T"TTT'-"--"--T"T'T'T"1I'T'T"-"'-,..,..,..,rTrM
lE+04
1 +05
1 +08
AU.OWABLE E8O'S
1 +07
1 +08
4-22
The curves obtained by linear regression and shown in Figures 4.8 and 4.9 showed
very good correlations between the structural strength of a pavement (measured in
allowable E80s) and the different 10M Parameters. An exception was the curve
obtained for Ll.l, This parameter showed a R' value of 0,55 (Table 4.9) which can
be seen as an unacceptably low degree ofcorrelation.
d)
It can be concluded that the analysis of field results yielded useful Information regarding the
characteristics of deflection bowl parameters as measured on different types of granular base
pavements. Not all deflection bowl parameters could be correlated to structural capacity.
Previous research (12) has also indicated that Ll.l values vary to an unacceptable degree and
cannot be trusted as M Indicator of structural capacity of a pavement. It is therefore doubted
whether the useof Ll.l should be continued. Some parameters which do correlate well with
structural capacity can serve as very useful indicators of pavement structural capacity. The
results of the analysis offield results are compared with the results ofthe mechanistic analysis
in Chapter S. The curves obtained in this chapter are also compared to other rehabilitation
design curves used Internationally and locally, In Chapter S.
S- I
5.2
5.3
S-2
S-2
5.4
50S
RECOMMENDATIONS
000000
00'
5-6
5-IS
5-16
S-2
s.i
In Chapters 3 and 4, relationships betweencenaln bowl parameters and the structural capacity
of granular base pavements were derived. This was done, In Chapter 3, by means of a
mechanistic analysis, and In Chapter 4, by means of an analysis of field results. In this
chapter, the results of these two analyses will be compared and commented on. Also, the
derived parameter-eapacily criteria will be compared to other rehabilitation criteria. The
purpose of this is to attempt to verify some of the assumptions made In the derivation of the
criteria In Chapter 4.
S.2
5.2.1
OBSERVATIONS
In order to study the agreement between the mechanistic and empirical analysis results, curves
were generated for Y-mar, BLI, MLI, LLI and RoC by plotting the curves generated In
chapters 3 and 4 on the same set of axes. This comparison is shown in Figures S.1 to 5.S
As can be seen from Figures S.1 to 5.5 there Is a substantial difference between the curves
obtained by empirical analysis and those obtained by mechanical analysis.
The curves
obtained by means of the mechanical analysis are clearly less conservative than the
empirically derived curves. For example, the mechanistic analysis curve for Y-max (Figure
S.I) Indicates that for an allowable number of ESOs of SO million (upper limit of E4 design
class), a maximum deflection of 800 micron is still acceptable. Table 4.2 indicates that the
Illib1 value measured on pavements with E4 type designs is 4S9 micron, with an average
maximum deflection of I mm
Indicates a very poor lubgrade and is in general wociated only with EO clus pavements.
The mechanistic analysis line however, shows that for I Y-max of I mm, E3 Trame can stili
be accommodated.
~1.YOI(wp)
5 - 3
-- -
to-
'-t--
lEt04
i'--. I'--
1 +07
1 +08
ALlOWA8lfEaO'S
I +05
+os
.........
10-
'"
r-- f-
-t--.. t~
10
lE+04
I tOll
...... t-.
r-, "I"-t'-
1 +08
ALLOWABlE EaO'I
I-
r-, ""-
1 +07
f'o
....."'' '
5 - 4
"'- ........
~
~
<; ........
"'-......J
--.....
...... .....
10
IE+04
1 +08
AllOWABLE eeo's
1 +05
I'"
II
1 + ee
1 +07
...
...
--
...
............ ........
...............
~
10
lE+04
1 +05
1 +08
AUOWABlE Eeo'S
, __ FlElD088RVATIOH8 .... folECH. AHAlYIJS
1 +07
II
oe
L- ~
...
I.-""
t-- ~
.~
...
.....
L--
~I--
--
l/v
'"
IE+04
1 +Cle
1 +08
ALlOWABlE EaO'S
, __ FIElDOBSERVATIOHS
-+-
1 +07
1 + os
MECH.ANAlYStS
The mechanistically derived curves for MLI and LLI show the same discrepancies when
compared with actual field measurements. In the case of BLI however, a closer relationship
was obtained between the mechanistic and empirically derived curves, as is shown in Figure
5.2. The reasons for this finding can most probably be attributed to the faet that BLI was
found to be closely related to the E-rnoduli of the surfacing, Thiswas indicated in Chapter
3 by the good correlation found between the remaining life of the surfacing and BLI. In
general, asphalt surfacings show a linear-elastic behaviour, and as mentioned in Chapter 2,
ELSYM5 uses Iinear.eJastic theory in its calculations.
Figures 5.6 to 5.9 show the curves as derived in chapter 4 plotted on the same set of axes IS
the lines currently used for wessment of FWD Denecdon basins in South Africa (9).
Flpres 5.6 to 5.9 indicate that these two curves are very similarinslope and absolute values.
It can therefore be concluded that the results of the empirical analysis compare well with
previous local experience on deflections when measured on granular base pavements In SA.
S-6
S.2.2
CONCLUSIONS
The general conclusions that can be drawn from the comparison of the field results to the
comparing the derived criteria for parameters related to non-linear pavement layers.
Ii)
The values obtained by using observed field results correlae well with previously
used criteria and local experience with deflections on granular base pavements.
II can thus be concluded that some caution should be applied when simulating pavement
structures under load by using layered linear elastlc theory. Absolute values of deflections
obtained by linear elastic programs such as ELSYMS can be inaccurate. Field results and
experience should be used 10 cal ibrate the computer model.
5.3
As indicated in the previous section, the curves for the different 10M Parameters which were
derived in this chapler Is very much in line with the experience of 10M deflection bowl
measurements in SA. Various other methods exist for determining the allowable number of
axles to failure from deflection measurements (21, 35, 36, 37).
In most instances the only parameter used is the maximum deflection. Widely used methods
using maximum deflec:tlon 10 predict equlvalenl ESOs are the TRRL and Asphalt Institute
methods (3S, 36). However, both the TRRL and Asphalt Institute methods use Benkelman
&un deflections.
5 - 1
FIGURE 6.8: YMAX VI. ALLOWABLE E80'S
OBSERVED va. PREVIOUS CRITERIA
1OOOO.,.---,,.........,--r-'r'"T'"i"'!"!'r--...---r--,....,..,...,..,.,....---,..--,..-r-,.........."I"!""'""-,.--.--.-"T"""""~
.:::::::::::I:l:::::::i::::l::rtll:tt::::::::::::t:::::::!::::l:::t::;:t:l:t.::::::::t::::::t:::j:::t::l:t::tr::::::.::t::::::~::::t:::t::l;:ll:t:
............. ,.!.....'.I... TTJ'
,l' . f......
,!.. f.. i!rH . r
I ..,I ,'"TTP
+!.. ht1r.
. . . t !j....!t.. "1+ '
...............
ITn
1" ..""]" .. r..1!tTt ..
,.... 1 ,.. t .... ;./.!.1
............., ..j-... ilrf . "f'+i . t'/'fiH T..1 +1 t .. . .+ i.. ~ . ~H+
........................I . pH
-T\IJI,TI,I r[...]
lll ,..
"'j'T .
l\..I'1
'1'
~i looo-::::::f,..~
~~JI[lr:n!lllrll,,l,
'~I ."
jtll
tr'
"If'
1"1'1
..
..
"
~~
..
,.".
..
""~f'
r-<1r:.: .
i"-r-.
1OO-l----jf--i-++++++t--+-+-H-H+1I+--+--+--HH+++t--i---i-++iiH+i
1E+04
1 +08
1 t07
1 +08
1 t05
AU.OWABLE Eeo'S
I'"
fIELD COURY"TlONS
.............
:I..t..J't
t , ..
+H
~ .. l
I.. t'..
..
:::::::::::::I:::::::I':::r:+:t:rh~::::::::::::1:::::::I::::: :::h:lH:l:::::::::::::\::::~l::"
~ .. l..tt
1
l.
. t 't t.. ..'..
f.. t1 .
:::..::t:\ t ::::::::::F::: .:: :::..:: :Ji
J'1
~
000;
. 'J'
1..
~~
.S
.::.:.:~:::
.........
'1'
.a.., .
"
/ .
..
;~
-";"'4.1..r~
:.:::::::" ::::::::: :'1111 H' .~ :.: r" ".1':tt!! 1~:."j .: ~.: '~f.'1 t!+t, .:~"" t . .i f tHf
t
, ..j
:
-lid j
ie
-.----"1.. 11, .
lE+04
r r fir Illl- -, Tn I,
lI!+OS
' 1'+08
ALLOWABlE Eeo'8
I_
FILD OOKRVATlONS
-+ PRVlOUI CATERIA
'-'1
'lIi~07
'
1'E +08
5 - 8
..~10rrif:llljl:l,"l.t-I!II!1'1:~r:l1rllll::ll11'lrIH
l.. .iTn.. . . r... jTTiT
1_~, ! 1] ',I'1]11.1
r.. T ..
. :::::::::::rs;;;fGE ::
100. .
..............
. . . . . . . . . . . .. ..... . . . .m. . . . . . . . .. . . . . .tt... .. +..
......................1 .., .. . .. ..
mm
r1
IIII
rllm
t'-.f"t': ~i- J . ~
......m
~~
lo+---+-+-+++-T+i+--i-r--+-H-+l-H--l-+-t-l-++H1f---+--+-+++++H
lE+04
l'E+05
' lE+oe
ALLOWABLE EGO'S
1E+07
1 +08
I
3
1O+--~H--+-HHt----1~H--t+H+t-4-+-+-1+-HH-+-++++-t+H
1E+08
1E+05
lE+04
ALLOWABlE EIO'S
5-9
Although the TRRL and Asphalt Institute methods have been used with adegree of success,
these methods were derived from experience and analysis of pavements overseas.
As
discussed in paragraph 2.3.2, this holds the drawback that the observations on these
pavements cannot be extrapolated to pavements used under different environmental, material
and loading conditions. However, both the TRRL and Asphalt Institute curves were also
These methods were therefore seen as good choices for comparison with the
empirically derived curve for Y-max. Figure S.IO shows the regression curve obtained by
analysis of the 10M Data Base compared with the proposed lines ofthe Asphalt Institute and
TRRL. (The plotted curve of the TRRL is the one proposed for pavements with granular
bases not showing any cementing actlon.)
I
~
1 +05
1 +08
AUOtVA8l EeO'9
1 +07
1 +01
5 10
As can be derived from Figure 5.10, the regression curve is similar In slope to the Asphalt
Institute line but differs In absolute values from both the TRRL and Asphalt Institute lines.
The most probable reasons for this are as follows:
a)
The TRRL and Asphalt Institute lines were derived for overseas soil conditions.
Jordaan (35,36) states that these curves cannot be extrapolated for different physical
conditions without verification. In the case of the TRRL curve, the experimental
work was done onsubgrade material with CDR values ranging between 2,5 and 15.
The TRRL curve wu also derived for a traffic loading between 0,3 and 10 million
E80s (3S). Once again the curve cannot be extrapolated for traffic figures beyond
The TRRL and Asphalt Institute lines were derived by using protection of the
subgrade as design criteria. Experience has shown that pavement failure In South
Africa is mostly attributed to the pavement layers themselves and not to the subgrade
(32).
c)
The TRRL and Asphalt Institute lines make use of Benkelman Beam deflections,
while the regression equation uses 10Mdeflections. Lacante (43) hu shown that an
averageconversion factor of 0,89 should be applied when convening 10Mdeflections
to Benkelman Beam deflections. Paterson (44) has stated that the ratio of Benkelman
Beam to FWD deflections ranges from 0,8 to 1,35 and is highly dependent on
pavement structure. The difference in the curves due to the different measuring
devices is thus not easy to quantify at this stage.
It Is thus clear that the TRRL and Asphalt Institute lines cannot be applied tolocal pavement
conditions without making some son of adaptlon for local traffic and lubgrade conditions u
well as for the type ofdevice used in testing. Another method more closely related to the
South African Mechanistic design method was thus also chosen for verlfylna the results
obtained by the empirical analysis.
5-11
The structural capacity predictions mode by the 10M parameters were verified by comparing
them to the asphalt tensile strain predictions. The approach adopted was to calculate the
tensile strain at the bottom of the asphalt layer by making use of the following analytical
equation proposed by Jung (21):
.....................................
Where S
(4.3)
La
Yo
YA
For 10M measurements, YA would be the deflection at an offset of ISO mm, as the radius of
the load plate (r) Is also ISO mm. Because the 10M only does not measure the deflection at
ISO mm, the deflection at a 150 mm offset was calculated by Interpolation between the
deflections at 0 and 200 mm. The calculated strain was then used with the TRRL asphalt
strain criteria in order to predict the remaining life.
Itwas decided to use the TRRL criteria because it Is used internationally and also because it
Is nearly identical to the asphalt strain criteria proposed in the South African mechanistic
design method. The TRRL strain criteria for asphalt layers is as follows (30):
f
:I
224
Where
lit (N)~,%)
f -
.......... . . . . . . . . . . . . . . . . . .
(4.4)
5 - 12
.-
-----_._----- ----------,
------1
. . -.--
MEASURED
DEF LECTION 1 - - - - - - - ,
BOWL
------_._,--
CALCULATE DEFLECTION
AT 160mm OFFSET BY
INTERPOLATION BETWEEN
CALCULATE
10M
PARAMETERS
-.
,------_.&.--------,
--
CALCULATE
TENSILE STRAIN
AT BOTTOM OF ASPHALT
LAYER (EQUATION 4.3)
-CALCULATE
ALLOWABLE
E80s (~)
BY APPLYING
CRITERIA DERIVED
IN THIS CHAPTER
Ne
I
1
Nt
....
.
.
10M PARAMETER .
E80 PREDICTION .
.
CALCULATE ALLOWABLE
E80s (Nt) BY
APPLYING TRRL CRITERIA
TO CALCULATED STRAIN
(EQUATION 4.4)
... ..
,
..
_1.11 J ,1.11 I
'------------_._._-
5 - 13
FIGURE 5.12: COMPo OFRAD. OF CURV.
VS.
I
I
~ '.OE+~
I.OE+04
UNE OFEQUALITY
-~. ./
I
i
1.0E+lle
.,0
5 14
UNE OF EaUALITY
. .-...: p
........
1\01"
11"..
1.0EtOS
1.0Et04
Figures 5.12 to 5.14 indicates that the predictions made from the IDM parameters compare
very well with the more theoretical strain prediction. The following observations can also
be made:
a)
Figure 5.12 shows the narrowest band of predictions. This confirms that Radius of
Curvature Is most closely related to the remaining lifeof the surfacing. Figure 5.13
Indicates that Ollis also a very good indicator of surfacing life. The observation that
BLI correlates well with surfacing strain predictions Is most probably due to the
relationship between asphalt Itraln and the ratio of surfacing modulus to Basecouno
modulus, IS Ollis also an Indicatorofthe relative stiffness of the buccourse.
b)
5 - IS
accurate when failure occurs in the surfacing and pavement layers. The poorer
correlations with the TRRL and Asphalt Institute curves Indicate that the 10M
parameter predictions are not accurate when lubgrade failure (I.e. runlng due to
subgrade strains) Is the primary mode offailure.
All the Figures 5.12 to S.14 show some outliers. The Interesting observation can be
c)
made that each of the parameters Y-max, BLI and Radius ofCurvature has outliers
In different places on the graph. For Instance. the two outliers observed in Figure
S.12 cannot be seen In either Figure 5.13 or Figure S.14. These observations
indicate that these parameters are Independent indicators and not just different
magnitudes ofthe same trend.
Itcan thus be concluded that the comparisonofthe empirical regression curves compares very
well with more theoretical predictions of remaining life when failure occurs In the pavement
layers, as Is mostly observed In South African conditions.
5.4
In this chaptera comparison was made between the mechanistically and empirically derived
criteria and other curves that have been used locally and internationally.
The main
conclusions that can be drawn from the analysis and comparisons described above are as
follows:
a)
The correlation between the empirically and mechanistically derived curves has
indicated that the absolute vaJues of the curves obtained by the mechanistic model
used In this study are not reaJistic. Theempirically derived curves are very much in
line with local experience In this field.
b)
The empirically derived curves for predlctlna allowable ESOs were tested aaainst the
more theoretical strain prediction. This comparison showed that the allowable E80
predictions made by the regression curves compare very well with the theoretical
strain prediction.
, - 16
5.5
RECOMMENDATIONS
i)
b)
Foran indication of basecourse and surfacing stiffness the Base Layer Index
(BLI) is recommended.
c)
For an indication of middle layer stiffness (subbase and selected layers) the
Middle Layer Index (MLl) is recommended.
d)
Ii)
in the IDMP program, be replaced with the criteria derived in Chapter 4 and shown
in Figure3 4.8 and 4.9. The rqresslon equations shown In FlaurtS 8 and 4.9
are the IVftlge regression CUrTes. These curves were derived from observations
of many different types of granular base pavements. All these pavements had also
carried some traffic. When analysing the structural capacity of anyone pavement
alone, it is strongly recommended that the allowable ESOs be expressed as a traffic
class (i.e. EO to 4) and not as a fixed figure (say 1,3 million ESOs). Also. when the
pavement has carried large volumes of traffic. this should be taken into account by
the designer. In the absence of detailed past traffic and other information the usc of
the lower 90 percentile line is recommended. The 90 percentile curves are shown
and discussed In Chapter 6. with further recommendations on temperature
6- I
PAGE
6.1
INTRODUCTION, . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6-2
6.2
6.3
6.4
6.S
6.6
6.7
6.8
6-2
6.1
INTRODUCTION
In the preceding two chapters It has been shown how various curves were derived relating the
different 10M Parameters to the structural condition of a pavement. The structural condition
ofthe pavement was expressed as the amount of 80 kN axle loads the pavement can carry to
failure.
It was concluded that the curves shown In Figures 4.8 and 4.9 were very much In line with
the experience gained locally in the field of 10M testing. The curves were also shown to
correlate well with the remaining life prediction made by asphalt tensile strain. The curves
can therefore be seen as a good first estimate of the structural capability of a pavement.
There are, however, several factors that influence measured deflection and structural capacity.
The most important factors to be accounted for are (30):
i)
Influence of temperature
II)
Iii)
Condition ofsurfacing
6 - 3
FIGURE 6.1: YMAX VI. AllOWABLE EOO'.
90THPERCENTILE RANGE
IOOOO-r-----------------------__,
"'-yPPER 90 % LINE
1 +05
toe
1 +07
AllOWABlEEIIO'S
LOWER go % LINE
1O+--r-r""I""TTn~-_r_"T'""T""I""1"TTTT'-..,..."T'""T"TTT1I'TT"-.,__r_T"T'1r'TTrt
1E+04
1 +05
, toe
AllOWABlE EIO'8
, +07
, .08
6 - 4
UPPER 90 % LINE
__
1 t05
1 tOll
1 +07
ALLOWABlE EeO'S
UPPER 90 % LINE
1 t05
1 tOll
ALLOWABlE EeO'S
1 +07
1 +08
6-5
Adetailed literature study was conducted In order to try and make a qualitative estimate of
the Influences of seasonal and surfacing conditions.
information on theway these factors influence deflection measuremenlJ and pavement bearing
capacity in South Africa. Because of this, only the resullJ of the literature study are
presented here. Although some tentative recommendations are made, it should at this stage
be left up to the experience of the designer to determine the magnitude ofthe various facton
to be taken Into account. A tentative scheme for adapting deflections due to the influence of
temperature Is also proposed.
It has been mentioned that the factors affecting the structural capacity of pavements are
numerous. It was noted in Chapter 4 that the allowable E80s obtained through the use of
deflection bowl parameters should be expressed as a traffic class, such as E2. according to
TRH 4 (38). Bringing temperature correction factors and seasonal varlalions into account
might indicate that a pavement which was classified as E3 might be classified as an E2 when
taking intoaccount the possible effects of. say. seasonal variations. Thus, taking temperature
and seasonal effecIJ or surfacing condition into account does not imply a more accurate
estlmatlon, but it ismeant to probe the sensitivity of a pavement classification (in terms of
class of traffic) by also taking into account those factors most likely to have an influence on
deflections and bearing capacity.
6.2
Very often the overall stiffness of a Sed ionof road isobtained by considering the average of
a series of deflection measurements. However, it often happens that various different subsections having different material properties and bearing capacities exist within a section of
road. For this reason deflection measurements should first be subdivided Into uniform subsections where an acceptable degree of variation is obtained.
Various methods exist for lubdivlding deflections into uniform subsections. Some of these
methods are described below:
6-6
The most simple and often used method for Identifying uniform sections by the use of
deflection measurements Is by plotting the maximum deflection against the kilometre
distances. This enables the designer to spot If and where a large variation In deflections Is
present. Figure 6.5 shows an example of deflectlons taken on I road section where two
uniform sections can be Identified.
After a uniform subsection has been Identified the designer should check whether the
coefficient of variation of the subsection Is within acceptable limits. The TRH 12 (1985)
specifies a maximum coefficient of variationof0,25 (45).
SOO"T"'""---------------.. . . - ----------..,
~
AVEAAQE (,All. DEfLECTIONS): 21. toIaon
COfFlClENl Of VARIATION
: 0..AVEAAQE ON IECTlON I
: 137 MIcron
COfflClENT Of VARlATlON : 0.17
AVEAAQE SECTION 2 : 333 Micnln
COEfFICIENT Of VARlATlON : 0.12
SUBSECTION 1
10
15
SUBSECTION 2
20
25
30
KIlOM'TEA OR 8TAT10N
67
6.2.1
A veryeffective method for subdivision ofdeflections isthe method used by the Cape
Provincial Administration and developed by Dumas (46). Because of its complexity
this method is used mainly for computer applications and only a brief outline of the
method is therefore presented. The method consists of two steps:
First thedata values are extracted from adata base at evenly spaced points and stored
In an array. This data set Is then smoothed In order to diminish the effect of sharp
spikes or dips.
After the data values have been smoothed a stepping algorithm Is applied to the
smoothed data to divide the road into uniform subsections. The algorithm uses the
areas between successive points to determine which area falls within a band width
specified by the user.
Although the C.P.A. method is relatively complex, it is ideally suited for computer
applications and has the added advantage that the user can specify the accuracy of the
division.
II)
~~I rt
(6.1)
6-8
[R
2_
(6.2)
u (1 - r) - -
"-2
" . I
0,5
with r
xl
_.I.:'=-I=--_~_-
(6.3)
, -I
is calculated using the total number of points measured. For agiven confidence level
(say 10%), u Is compared to the limits U. and u, . of thet-dlstributlon. If u. < u
< u,.. then the population belongs to a normal distribution and the zone Is
homogeneous. If u > u,.. then the data shows slow periodic variations and Is
divided where g(i) (Equation (6.1 Is a maximum.
If u < u, the data contains rapid fluctuations and cannot be subdivided.
Both methods (i) and (II) have been applied with success to computer programs
involving deflection measurements. The ELMOD back-ealculatlon program is one
example of a program that uses the method (Ii) liven abovo (26).
6-9
6.3
The influence of temperature on deflections has long been Incorporated into rehabilltation
design methods such as the TRRL and the Asphalt Institute methods (3S, 36). Temperature
has a direct Influence on the Elastic Modulus ofthe asphalt layer, and therefore also on any
deflection measured. There is also some evidence that temperature can cause expansion In
granular layers to cause "lock-up" which will stiffen the aranular layers and lead to lower
deflections (16). However, very limited research has been conducted in this field, and
temperature adjustments are therefore In most cases applied only to the asphalt layers of the
pavement.
Whilst temperature can affect the deflections measured an asphalt base pavements to some
degree, the effect of temperature on deflections on granular base pavements are minimal.
Furthermore. as indicated in Figures 4.2 to 4.5, the percentage of the deflection taken up by
the surfacing is minimal when compared with that of the other layers.
Currently very little information exists on temperature adjustment factors In SA. Lacante (43)
has conducted some work on the effects of temperature on deflections. However, this
Information Is limited as his worle was primarily aimed at investigating comparisons between
different types of deflection measuring devices. The curves proposed in the AASHTO
pavement design manual (14) have been incorporated into various NOT methods. These
curves are shown In an adapted form in Figure 6.6. As can be seen from Figure 6.6. the
asphalt base pavements have the largest temperature correction factors. while the pavements
with cement bases have the smallest.
Figure 6.7 is adapted from Figure 6.6 and shows the curve proposed for pavements with
ISOmm of aranular material for a basecourse. As can be seen from Figure 6.7. the
temperature factors range from 0.75 at SOC to 1.8 at OC.
50
45l40
~.-
r-
35r
A
B
-/
F
&&.I
Go
CURVE
i~
a: 30
2
...-\\ \\
\.
41
&&I
\\ \\ I
25
BASE
THICKNESS
Asphalt
(Full depth)
Asphalt
(Deep strenott\)
Granular
All thicknesses
100mm Granular
Subbase
150mm
Cemented
(sound)
100mm
Cemented
hound)
200mm
l-
~ 20
BASE
MATERIAL
::E
CI
1&.1
~
~ 15
&&.I
10
o '~---;~'=---:~'~---:~'~---::'L:;:--~:l::--_~~~
__'L_":~.l~~_-lc::~
l'
,
,
'>
>'...""
'..............._.c::::::::.,
'
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
J_J
"
2.4
50 ,
45
40
u
'"
a: 35
:;:)
t-
~ 30
""a.2
I&J
~
25
tZ
'"
:::I
e-
20
......
I
>
'"
~ 15
:3
2
10
5
0.0
0.2
0.4
0.6
0.8
1.0
1.2
1.4
1.6
1.8
2.0
2.2
2.4
6 12
Research work conducted by Scullion (17) has yielded results that compare favourably with
the AASHTO design guide's correction factors, but only on thicker pavements. The effect
oftemperature onsurface treated pavements have not yet been established, but Indications are
that the factors proposed In the AASHTO Guide arc not applicable to this type of pavement
(16).
From the above discussion It canbe concluded that temperature correction factors arc minimal
when applied to pavements with thin surfaclngs. Also, no accurate temperature adjustments
are currently available for surface treated pavements, such as those with single or double
seals.
6.3.1 PROPOSED TEMPERATURE CORRECI10N FACTORS
From the preceding discussion it can be gathered that temperature adjustment factors should
be applied with care. The factors used in the United States and elsewhere should not be
adopted for South African conditions without taking account of the fact that pavements
overseas are very different in structure from local pavements. Even pavements referred to
as those with thin asphalt surfacings often have asphalt surfacings of 100 mm (4 inches) or
more.
The measurements shown In Table 6. 1 were taken on a thin asphalt pavement at the Texas
Transportation Institute, and shows the typical effects of temperature on deflections.
6 13
TABLE 6.1:
TIME
DATE
15:24
271f12
271f12
7:3'
10:56
14:20
13101
13101
13101
9:4'
(16
DEPLECI10N (MICRON)
SURFACE
TEMP (OC)
05
D6
UXZ
71
33
66
'3
51
..3
104
"I
30
..
56
61
51
......
33
33
D2
03
13
592
20
630
221
221
47'
476
14
35
173
171
16
46
455
170
14
07
D4
01
46
"I
"I
38
30
The dotted line inTable 6.1 divides measurements taken during two different seasons. It can
be concluded that principally the maximum deflection is affected by temperature.
Other
LESS
CORRECTION FACTOR
TEMPERAnJRE
> 3SC
IS3SC
10 ISc
< IOC
01
02 AND 03
0,9
1,0
1,1
1,2
0,95
1,0
I,OS
1,1
6 14
The following considerations were taken into account in choosing the facton shown in Table
6.2:
I)
as
a~phalt
moisture conditions and many more. Because of the uncenalnty Involved, the use of
one factor for a given range of temperature values Ispreferred to afunction or graph,
IO~
was used on either side of unity, and then only when temperatures were well beyond
the normal range of values. In conditions where very low temperatures might have
prevailed a 20~ deviation was applied, as the possibility offrost In the lower layen
might Influence the deflections to a more significant degree. (The tendency of lower
temperatures to have a larger influence on deflections than high temperatures is also
clearly evident from all curves in Figure 6.6)
The method for applying the temperature correction factors Is simply to multiply the
maximum deflection by the appropriate factor. For example: If a maximum deflection of
530 micron was measured at an asphalt surface temperature of 40C, the corrected maximum
deflection would be:
Adjusted 01
477 micron
6 - IS
6.4
seasonal influences on deflections in this country. Work done byEmery (48) did investigate
seasonal influences on moisture content of pavement layers, but this work makes no reference
to seasonal influences on deflections.
The purposeof this section is to point to various tendencies and factors that need to be taken
into account when considering the effect of seasonal variations. It Is at this stage left up to
the designer to consider the relevant factors and the magnitude of their influence on remaining
life.
6.4.1 CLIMATIC ZONES
Figure 6.8 shows the influence of seasonal variations on two pavements situated in different
geographic zones in the USA (14). It is clear from Figure6.8 thatscasonal effects can differ
drastically between different climatic zones. It should however, be borne in mind that the
effect of the spring thaw which explains thedrastic variation in deflections of the Rochester
test section in Figure 6.8 is rarely found in South Africa.
In most parts of South Africa, rainfall will be the most imponant factor governing the effect
of seasonal variations on pavement performance. This leads to the conclusion that only
pavements in high rainfall regions will be susceptible to seasonal Influences.
the detenninatlon ofclimatic zones can be found In TRH 4 (38).
Guid~llnes
for
3
TEST SECTION NEAR ROCHESTER, MINN.
e1966-67 Data)
E 2
Z
0
tu
I&J
-I
&L
O!
-'"
I
o '
,
DEC
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
TIME
6 - 17
The effect of rainfall on subgrade modulus depends on the type of subgrade. It can be
expected that pavements with non-clayey subgrades will only experience minor changes in
subgrade E-modull with a change in moisture content. Subgrades with a high clay content
can conversely beexpected to undergo more radical changes with changing moisture content.
6.4.3 DRAINAGE
Drainage can possibly be seen as the one factor deciding the overall effect of seasonal
variations on pavement performance. Sufficient drainage conditions might lead the designer
to exclude the effects of seasonal variations altogether. Again. it is left to the designer to
provide the input from his knowledge of specific conditions applicable to the pavement under
consideration.
6.5
INCORPORATION
OF
VISUAL
RESULTS
INTO
TIlE STRUCTURAL
CLASSIFICATION PROCESS
Once the effects of temperature and seasonal variations have been considered. the designer
has an indication of the structural capacity of the pavement. This structural capacity Is
expressed as the allowable number of ESOs the pavement can carry to failure. This figure
can be used to describe the structural capability In terms of the traffic class (according to
6 18
One of the most importanl factors that could affect thc structural condilion of a pavement is
the condition of the surfacing. The TRRL method states that exterslve deterioration of the
road surface (with low measured deflections) will normally lead 10 premature failure (3S).
A brinle and dry surfacing could also lead 10 premature failure while a surfacing which
exhibits signsof bleeding might Indicatea longer than expected life. The age of the surfacing
can also be used to forecasl surfacing life.
The Incorporation of visual survey informalion should not be built Inlo the structural
classification process, bUI should rather be used as a separate unit ofInformation that warns
the designer of a possible reduction in pavemenilife due to surfacing conditions. Table 6.3
serves as an Indlcalion of the effect of various surfacing conditions on predicted structural
life.
TABLE 6.3:
NEGATIVE INFLUENCE
(REDUCTION IN PAVEMENT LIFE)
POSmVE INFLUENCE
(INCREASE IN PAVEMENT LIFE)
Figure 6.9 illustrates how the condition of the surfacing could possibly be used to adapt the
structural capacity prediction made by 10M deflection bowl parameters. FllUre 6.9 Is used
only to illustrate how surfacing condition can be used to adapt the predicted structural
capacity. The extent to which the predicted structural capacity is adapted may be determined
by the severity of the respective surface condillon. It is Impossible, It this stile, to quantify
the extent to which predicted structural capacity should be adapted, but designers should only
exceed the 90th percentile limits shown In Figures 6.1 to 6.4 In extreme cases.
6 19
FIG.6.9: AOAPTION OF PREDICTED CAPACITY
FOR SURFACE CONDITION (EXAMPLE)
10001. , . . --
--,
UPPER 90 % LINE-~"""'"
MeASURED VAW!
LOWER 90 % LINE
tOI+--~'T""'1""TT'TTT"r____r__r_._T"I""I'T"I'"r___r..,..."T"""I""T"I'.......-...,._"T"""I......~
tE+04
t +07
t +08
t +05
6.6
6.6.1
GENERAL
The Dynamic Cone Penetrometer (OCP) is a fairly cheap and reliable device for the
prediction of relative strengths of individualpavement layers. Very often OCP measurements
are available in addition to 10M measurements. Incorporatin& OCP results intothe structural
classification process will lead to a better overall understanding of the relative strengths or
quality of individual zones within a pavement structure. OCP results can also provide
valuable information regarding the type of failure mechanisms present In cases where a
pavement is showlna sians of distress.
6.6.2
Both the OCP and the 10M can be used to forewt the number of ESOs the pavement can
carry
to failure.
devices. 10M and OCP measurements liken on a number of roads were used to obtain
correlation between the numbers of E80s forewt by each method.
620
The roads used for the analysis were limited. This is mainly because of the difficulty of
finding roads where both DCP and 10M measurements were taken. Furthermore. data could
only be used where DCP and 10M measurements were taken in very close proximity to one
another.
The roads used for this correlation are shown in Table 6.4.
TABLE 6.4:
N312
M2 JHB
Pilanesberg Alrport
18
OSlO Bophuthatswana
31
Only maximum deflection and DSV100 wereused in this correlation. The correlation is shown
in Figure 6.10.
621
. - .
1.0+07
UNE OF EQUALITY
....
..,.
. .. . ..-:~
1 +05
1 +08
DCP (OSNllOO) PREDICTIONS
6.6.3
--..
1 +07
1 +08
It Is clear from Figure 6.10 that maximum deflection predicts a much lower structural
capacity than theDSN. of the OCP. The reasons for this observation can be as follows:
I)
The failure criterion used for OSN.ls based on runing, whereas that of the IDM is
based on cracking of the upper pavement layers and terminal riding quality.
II)
K1eyn (2) has shown that the wo of DSN IDO alone cannot accurately predict tho
structuraJ c:apachy of a pavement system. The overall capacity of the pavement
system depends on whether the pavement system is balanced with respect to the
different strengths of the pavement layers (2).
Tho correlation between IDM and DCP capachy prediction and resulting observations as made
In this study cannot be seen as conclusive. It Is the author's opinion that more meaningful
results will be obtained when correlating tho number of DCP blows obtained In each layer
with the capacity predictions made by the different 10M parameters mated to each pavement
ZODe.
The dati available for this study, however, do not allow this typo ofdetailed anaIysL,
at this stage.
622
6.7
From the discussions of the previous sections It is clear that there He many facton that affect
the structural condition of any pavement. No method used to usess the condition of a
pavement can accurately account for all thC$e factors.
TEST METHOD
INFORMATION PROVIDED
VISUAL
SURVEYS
Surface Condition/Capacity
Structural Capacity based on:
I. Shear Strength
10M
DCP
vv
vV
------------------------- -----
---- ----
vv
..,j
Vo/
V
V
2. Layer Stiffness
...............................
3. Visible Distress
vv Good Indication
Some Information
No Information
623
From Table 6.S It is clear thai none of the three test medlods alone provide all the
Information needed to carry out a detailed rehabilitation design. However, when the results
of all three methods arc considered together, not only Is a more complete picture of the
pavement condition or structural state provided, but the results ofIndividual tests can also be
verified to some extent. Furthermore, when two test results arc In clear contradiction with
one another, this may Indicate the need for
materials testing), which would often show the presence of other facton which would most
often Influence the overall rehabilitation design.
In conclusion:
The factors that affect pavement performance are numerous and cannot be quantified
or detected by anyone test method.
The results of some tests can be used to verify or dispute the results of other tests.
Simultaneously considering the results of visual surveys, 10M and DCP
measurements should provide a much more complete picture of the pavement
condition, as each test method provides specific Information on the pavement
condition.
6.8
positive or
negative Influence onthe remaining life predictions of 10M deflection bowl parameters. The
effects of temperature and seasonal variations on deflections and structural capacity
predictions were also discussed.
624
It was concluded that 10M measurements should always form part of a multl-crttenon
analysis. That ls, 10M results should be interpreted, verified and enhanced with the results
ofother tests such as Depts and visual surveys.
Due to the lack ofavailable information the effects of seasonal variations, temperature and
surfacing condition on structural capacity cannot be quantified with confidence at this stage.
lt lsstrongly recommended that further research be conducted inorder to attempt to quantify
the influence of these factors on pavement structural capacity. More specifically, research
should be aimed at investigating the following:
i)
The influence ofsurface conditions, such as brittleness, surface cracks, bleeding and
age on deflections and structural performance;
Ii)
iii)
iv)
7 I
CIIAYTER 7
PROPOSED SCIIEME FOR TIlE INCORPORATION OF A DEFLECTION BASED
STRUCTURAL STIFFNESS INDEX INTO PAVEMENT MANAGEMENT SYSTEMS
PAGE
7.1
INTRODUCTION. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
7.2
7.3
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. 7-12
72
7.1
INTRODUCTION
I)
Physical Distress:
Assessment of physical distress such as cracking and deformation is currently done
through visual assessments. Visual assessment can be seen as one of the key factors
in maintenance and rehabilitation decision making In PMS.
Ii)
Functional Condition:
This aspect deals mainly with the condition of the pavement with regard to safety and
service to the road users. This is determined by skid resistance and ridina quality
measurements.
7-3
FIGURE 7.1: MAJOR PMS ACTIVITIES (49)
NETWORK UPDATE
CONDmON ASSESSMENT
(Physical Distress, Functional and Structural Condition)
I
DATA ANALYSIS
(Data Processing, Identify and Prloritise Need, etc.)
I
PANEL INSPECTIONS
(Confirm High Priority Proposals)
I
PLANNINO
(performance of Network, Identifies Roads
to be Constructed or Maintained)
I
PROGRAMMES AND BUDGETS
(List of Proposed Projects, Time Frame and Funding
I
DESION
(Consider Alternatives, Life Cycle Costs)
I
IMPLEMENTATION
(Drawing up of Specifications, Tendering Construction
and Quality Control)
I
FEEDBACK
(Recording of As-built- Dati, Monitorina, etc.)
7-4
Iii)
Structural Capacity:
The structural capacity of the pavement can be defined as the ability of the pavement
to accommodate traffic loadings with lillie or no cracking ordeformation. Structural
capacity is normally assessed by means of deflection measurements. Currently most
network level PMS's do not include the use of NDT methods such as deflection
measurements.
The information obtained by visual assessments and riding quality and skid resistance
measurements areused to devise a VisualCondition Index (Vel). Normally the vel is used
together with aspects such as Road User Costs (l.e. costs Incurred by the road user, such as:
fuel, vehicle depreciation, maintenance, etc.) to Identify and prioritise roads for maintenance
and rehabilitation.
Pavement Management Systems in the United States and elsewhere have for some years
already used deflection measurements at network level in order to obtain an indication of the
structural capacity ofthe pavement (16). Deflection measurements arethen used to calculate
a Structural Strength, orStructural Stiffness Index (SSI) which can be used with the VCI and
User Costs to Identify road sections for maintenance and rehabilitation. The Structural
Strength Index can also be used in the Identification of the appropriate rehabilitation
measures. For example, a road section indicated as "poor" by the VCI, but having a
load
structural condition will require a reseal rather than major structural rehabilitation.
One of the major constraints in the incorporation of a Structural Strength Index into PMS's
Is the cost involved In conducting deflection measurements at network level. Various
factors come Into play here, such as available funding, method ofIdentification of road links
for non-dcstructlve tcslin, and required density of testing.
7S
It is the purpose of this chapter to address all the major issues Involved In lncorporatlng I
Structural Stiffness Index into Pavement Management Systems. More specifically. this
chapter is aimed at the following:
i)
Ii)
Iii)
Credit should be given to the work done by Scullion (16) at the Texas Transponation
Institute. of which much of the methodology followed was adopted in this study.
7.2
As was mentioned insection 7.1, the purpose ofa Structural Stiffness Index is to serve as an
indication of the structural capacity of the pavement. A SSI can be used in addition to the
Visual Condition Index as an Indicatorof the road link condition. As such, it is essential that
the SSI be expressed In a way that is compatible with most current PMSs. Currently. the
VCI is expressed as a figure between 0 and 100, from very poor to very good. respectively.
In order to devise an index that can simultaneously be used for prioritlsatlon of roads and
serve as an Indication ofthe actual pavement strength. the following guide-lines were used:
7-6
I)
In orderto be compatible with current PMS's the SSI should preferably be expressed
as a figure between 0 and tOO, from very poor to very &ood, respectively.
II)
The calculation of the SSI should take Into account the traffic for that specific link.
The stnaetural capacity will therefore be expressed Interms of the required structural
need for the traffic under consideration.
III)
The method for determining the SSI ofa road link should be robust and simple. The
SSI is primarily used for road priorltlsatlon and is nottherefore Intended to have the
accuracy that Is needed at project level analysis.
The use of 10M deflection bowl parameters Is Ideal to fulfil the last requirement stated above.
Scullion (16) has found that a SSI based on statistical data is more suitable for PMS's than
the more complex mechanistic approach.
Itshould be noted that the SSI is a structural parameter. The SSI should be Implementedonly
as a second-order Indicator after prlorltlsatlon of roads has been done using the VCI and
Riding Quality. The SSI can in this way assist in further prioritisation of roads for
rehabilitation. Also, the SSI can assist in deciding on the correct rehabilitation measure, as
pointed out in section 7.1.
7.2.1 CHOICE OF 10M PARAMETERS FOR INCORPORATING INTO A STRUCTURAL
STIFFNESS INDEX
In Chapter 4 a number of parameten were shown to have acceptable correlations with
expected pavement life, namely Y-max, BU, RoC and MLI.
calculations of theSSI as simple as possible It was decided to base the SSI on only two of
these parameters. Y-mu and Bli were chosen for this purpose. Thereasons for choosing
these parameters are as follows:
I)
Y-max and BLI showed the highest correlations In the rearcsslons performed In
Chapter" (See Table 4.9).
7-7
ii)
III)
BLI has been shown to be an excellent Indicator of basccourse and surfaclna stiffness
(and possible cracking of the surfacing).
Iv)
The use ofonly BLI and Y -max ensures that the requirement of simplicity is met.
At the same time these two parameters give a good Indication ofboth the upper and
lower zone stiffness within the pavement structure.
The use of the Radius of Curvature and MLI may provide more detailed information
regarding Individual pavement zones. However, such detailed Information Is more useful at
project level than at network level. Also, bearing In mind the stated purposes of the SSI, it
Is doubtful whether the inclusion of more parameters for the sake of obtaining a better SSI
can be justified.
7.2.2 PROPOSED METHOD FOR CALCULATION OF STRUCTURAL STIFFNESS INDEX
The methodology and criteria adopted for the calculation of theSSI follows directly from the
results of the empirical analysis in Chapter 4. The parameter criteria shown in Figure 4.8
was used as the basis upon which the SSlls calculated. In order to calculate the SSI. use was
made of the following functions:
I)
The functions rdating Y-max. and BU to Allowable E80s (N.......,.) which was
derived In Chapter 4:
Ny_
NIU
(1,17 It IOI')(BLI)'),1D7
(7.1)
(7.2)
78
Ii)
The function relating remaining life to Cumulative Design ESOs Equivalent Daily
E80s, Traffic growth and Lane distribution. This function wu derived from the TRH
4: 1988 (38) and Is of the following form:
I01([
N,.(O.OI.i)
E.B,.36S(1 +0,011)
+ 1
(7.3)
Where N.
E
B,
y
Iii)
v.... Y
IU
A function relating the 551 to the remaining life period (in yean). This function was
derived as follows. Firstly different remaining life periods were assigned to different
REMAINING LIFE
(YRS)
SSI
STRUCTURAL
RATING
> 15
10- IS
S 10
2 5
< 2
> 8S
708S
SO70
30 SO
<30
Very Good
Good
Fair
Poor
Very Poor
79
This classification and rating system are closely related to the rating system used in PMS
systems for the rating ofthe Visual Condition Index (Vel).
The values shown in Table 7.1 were used todo a non-l inear regression. This resulted in an
equation of the following form:
.....................................
SSI
Where SSI
IS
(7.4)
Regression Constants
a, b, c, d
VALUE
a
b
O,81S9
-0,0202
448,S408
0,0011
R2 0,99
7 - 10
The procedure for the calculation of the Structural Stiffness Index for I specific road link Is
shown schematically In Figure 7.2. The basic procedure Is u follows:
I)
From10M field measurements, calculate maximum deflection and Base Layer Index
(OLl)
II)
Uslni Equations 7.1 and 7.2, calculate the Allowable Number of 80 kN axle loads
for Y-max and BLl (Nv.... and NIU ) .
iii)
Uslni the values of Nv.... and Nw determined In (II), and dle figures of AOT,
Percentage Heavy Vehicles, Number of Lanes and traffic irowth rate for the road
link, calculate the remaining life predicted by Y-max and BLl by using Equation 7.3
(Y _ I and Yw).
Iv)
Calculate SSl y ... and SSlau, which Is the Structural Strength Indices for Y-max and
BLI respectively, by using Equalion 7.4.
v)
In Table 7.1 to obtain the overall structural rating for the road link.
In the cases where no Information regarding the traffic growth race Is available, a value of
31 Is recommended. The number of test points needed to accurately assess the structural
condition of a road link Is discussed in Section 7.3. A worked example of a calculation of
the SSIIs presented In Section 7.S.
7 - II
FIGURE 7.2: SCIIEMATIC REPRESENTATION OF CALCULATION OF STRVCTtJRAL
STIFFNESS INDEX
FROM PMS
DATA BASE:
10M FIELD
MEASUREMENTS
OBTAIN V-MAX
AND BLI
Equations 7.1
and 7.2
OBTAIN ALLOWABLE
E80sFOR
YMAX AND BLI
Equation 7.3
CALCULATE
ESTIMATED
LIFE (YRS) FOR
YMAX AND BLI
Equation 7.4
CALCULATE FINAL
SSI AS AVERAGE
OF SSly ._ and SSIItJ
ADT, , HEAVY
AND GROWTH
RATE FOR
LINK
7 12
7.3
test points.
II)
The average ofrhe maximum deflections and BLI's of each section is then calculated.
III)
Step(II) Is then repeated for Incrwlngly less tcst points, by disregarding every other
test point. Thus If 30 points were measured, the averqe Ymax and BLI would be
calculated for 30 test points, then for IS, 7. 4, etc.
7 13
Iv)
The pavements are now ranked according to the average maximum deflections. The
pavement having the lowest average Y-max would be ranked first, and that with the
highest V-max would be ranked last.
v)
Assuming the highest density ranking to be correct, the ranklnas ofthe lower density
testlngs arc compared by using the Spearman rank correlation coefficient. The
Spearman coefficient is calculated as follows:
(7.S)
r, 1 - .....;-----"(,,2 _ 1)
XI
VI
Where r.
The hypothesis Is
Ito :
H. :
When r, Is greater than a critical value, the null hypothesis Ito is rejected and the two
ranklngs are related. The critical value is dependent on the number of pairs that are ranked,
and the chosen level of confidence (a r, value of I would Indicate thai the two rankings are
identical).
7 14
Scullion (16) hasperformed this analysis usina 8different roads, and usln8 a hi8hest density
test of 40 points as a correct ranking (XJ. The results of this study is shown in Table 7.3.
Critical values of r, fordifferent confidencelevels and n 8 areshown in Table 7.4. Table
7.4 Indicates thai, for a 90% Confidence level, tho value of r. should be 8reater than 0,524
for tho rankings 10 be related.
TABLE 7.3:
NO. OF
TEST POINTS
40
20
10
7
5
4
2
TABLE 7.4:
2
2
2
7
2
4
3
3
4
6
8
8
3
4
4
2
3
1
1
1
1
1
6
6
I
I
2
3
5
5
7
5
5
3
5
8
8
8
8
8
7
7
7
7
6
6
4
4
2
5
4
5
3
7
6
6
SPEARMAN
COEFF. (rJ
1,000
0,976
0,905
0,571
0,833
0,476
0,357
0,01
0,05
0,10
Critical r,
0,833
0,643
0,524
The highway network in Texas is divided into sections of approximately 3,2 km (2 miles)
lon8 (16). Table 7.3 would thus indicate that for I
90~
section would be acceptable. It could therefore be concluded thll in order for the results to
be stltistically ICcClnable, the study at the Texu Transportation Institute showed that 5
7 - 15
An identical study was performed by using 8different local roads which were selected from
the data base on the basis of their deflections and construction type. Usina the approach
outlined above, rankings were obtained on the basis of both Ymax and BLI, as these were
the parameters selected for the calculation ofthe Structural Stiffness Index. The sections of
road chosen differed inlength from 1,0 km to 4,2 km. Tables 7.5 and 7.6 show the results
ofthis study.
TABLE 7.S:
NO. OF
TEST POINTS
2
2
2
3
8
8
8
8
I
I
I
I
S
6
O,9S2
7
7
O,S24
8
8
0,333
0,000
32
17
10
6
3
6
4
4
S
S
6
3
3
6
5
3
2
TABLE 7.6:
SPEARMAN
COEFF. (rJ
2
2
1
1,000
0,786
NO. OF
TEST POINTS
32
17
10
6
6
8
8
8
3
2
7
6
7
7
8
7
6
6
6
6
1
1
1
1
2
3
S
S
2
3
3
2
1
1
4
S
4
4
4
2
3
3
8
8
S
S
7
SPEARMAN
COEFF. (rJ
1,000
o,90S
O,87S
O,90S
O,90S
0,810
7 16
The results of the study of local roads show that 6 test points per road section are needed in
order to do an accurate ranking of sections. Below 6 measurements an erratic variation of
ranking occurs for maximum deflection although BLI shows an acceptable rank correlation
even at two testpoints per section.
Pavement Management Systems in South Africa normally divide the network into sections
having lengths of between 2 and S kin. It Is recommended thai at last 6 test points per
nellon be used torsfdlons havlna lenaths ofup to S km. More lests should be performed
on longer sections where the spatial variation of materials is greater. On these sections
testing should be done at a maximum spacing of 850 m. Sections not having a uniform
construction throughout should first be subdivided Into uniform subsections.
7.3.2 IDENTIFICATION OF ROAD SECTIONS FOR DEFLECTION MEASUREMENTS AT
NElWORK LEVEL
Due to budget constraints. deflection measurements cannot be taken on all road sections
within a network. Therefore. sections would first have to be Identified for testing, and the
percentage or the length of the network to be tested has to be established. Once again. this
percentage Is determined mostly by budget constraints.
In the United States. many road authorities test approximately
3O~
annum. The roads to be tested are identified according to the VCI as well as the rate of
change In the vel over the past analysis period.
For South African conditions it is recommended that the roads on which deflection
measurements be taken, be Identified according to:
I)
Ii)
7 - 17
These two factors should be assessed, with the funds available, In order to determine the
portion of the network that has to be tested.
Condition Index of 43 (poor). 10M testing was performed at 6 evenly spaced positions on
the section. Traffic Information for the section Is as follows:
Annual Daily Traffic
% Traffic Growth
3200 v.p.d
iii
4~
% Heavy Vehicles
9~
No. of lanes/direction
625 micron
Avg BLI
31S micron
N..,
IOI~(62S)04.m
(2,43 x
(1,17 x
IOI~(3IS)u:n
.................
(7.1)
(7.2)
7 18
S1m.l: Calculate the Equivalent number of ESOs (E). Assuming an E80lheavy of 1.5, and
using the supplied traffic filUres, the value of E is calculated u follows:
E
432
(7.6)
Determine the lane distribution factor from Table II ofTRH 4: 1988 (38).
Table 7.7 is adopted from Table II of TRH 4: 1988 (38) and is shown
below:
TABLE 7.7:
TOTAL NO. OF
TRAFFIC LANES
1,00
0,95
0,7
LANE 2
LANE 3
0,30
0,60
0,25
The valuesof Ny... and NIU are now substituted in Equation 7.3 toobtain the remaining life
predicted by Y-max (Yy..J and BLI (Y..,).
r,..
(7.3)
1,86 Years
(0,36
rill
lot)(O,04)
loa 432(I,OO)(36S)(I,04)
101 1,04
2.15 Yean
. . . . . . . . . . . . . . . .
(7.3)
719
S1m..J:
Using the values of Yv... and YIU calculated in Step 2, calculate the values of SSI".
-.II
(0.81~9
+ 448~8(1.86
. . . . . . . ..
(1.4)
. . . . . .
(7.4)
(l - 0.0202(1,86) + 0.0011(1,86)2
III
29,40
(0.8159 + 448,5408(2,15
(I -0.0202(2,15) +0,00112.15)2
Stm....!: Calculate the final 551 value as the average of 551".... and SSIIU obtained in Step 3.
SS/ _S5_'.;,;;r-=-_+_S_S-:/JU=
2
29.4
31,68
8 I
CIIAYI'ER 8
CONCLUSIONS AND RECOMMENDATIONS
82
8.
The structural evaluation of pavements through the use of an empirically based method has
been outlined and described in this thesis. A cost-effective method for determining the
structural capacity ofgranular base pavements by usinglDMIFWD measured deflections was
presented. Theproposed method can be applied directly to deflections measured on granular
base pavements in order to obtain a reasonably accurate estimate ofthe structural capacity of
the pavement.
The deflection response of pavements are complex and most pavement capacity analysis
procedures based on deflection measurements can at best offer to provide an estimate of the
structural behaviour of any pavement over the design period. The reason why no analysis
method can claim to provide exact answers regarding the remaining life of any pavement is
clear when one considers the many changing variables, such as moisture content, subgrade
stiffness, temperature effects, etc. which influence the behaviour of a pavement over time.
These variables can vary considerably over the design period ofany pavement, and someof
these are also very difficult to quantify satisfactorily. It is therefore the opinion of the author
that any pavement structural analysis using measured deflections should conform to the
following broad guide-lines:
The analysis should always form part of a multi-criterion analysis using Dynamic
Cone Penetrometer results, visual surveys and laboratory testing, amongst others.
It has been stated that road building Is an anform and not an exact science. The
results obtained by any type of analysis should therefore always be correlated with
practical experience gained in the field.
83
Advances in the field ofmicrocomputers have resulted in thedevelopment of mechanistically
based methods for evaluating the structural capacity of pavements. These methods, which
take into account the various possible mechanisms of failure which can occur, have been
widely recognised to bethe most advanced and technical methods ofpavement analysis. This
ismainly dueto the detailed information provided by most of these methods. Unfonunately
these methods require skilled personnel inorder to be applied correctly and effectively. Also,
these methods require fairly detailed input variables, making It unsuitable for applications at
network level where only limited information is available in most instances.
The methodology which was described in this study is aimed at providing a means whereby
agencies which do not dispose over the expertise required to do a detailed mechanistic
analysis, can make a direct estimate of the structural capacity of a pavement being analysed.
This method, which is based on the results of a detailed analysis of field results as well as
mechanistic analyses, uses the full deflection bowl in order to establish the relative capacity
of each individual pavement zone within the pavement system. The results of the method are
of necessity expressed as a range of 80 kN axles, and not a fixed value, which would imply
a much higher accuracy.
The main features of the proposed method are as follows:
The method iseasy to use and relatively fast whencompared to mechanistic methods.
Due to the simplicity of the method, the obtained structural capacity is expressed in
terms of a range of 80 kN axles, which the pavement should be able to accommodate
before the formation of cracks or runin" leadin, to terminal riding quality.
The results obtained through the use of the proposed method were shown to have an
acceptable correlation with results based on more theoretical methods, such u those using
tensile strain at the bottom' of the a,phalt layers as failure criteria.
84
The method proposed In this study was extended and adapted In order to enable Its
Incorporation Into Pavement Management Systems currently In use In South Africa. In
particular, the extension of the proposed method wu aimed at devcloplng a Structural
Stiffness Indcx which provides an Indication of the structural capacity of a pavement as
related to the traffic the structure has to accommodate. This study wu also aimed at
Investigating the optimum testing density required for network Icvellnvestlgatlons.
The study presented Included a limited Inveatlaatlon Into the factors affecting the deflection
response of a pavement system. Due to the lack of available Information, however, only
broad guide-lines were provided for talclna account such factors as surface condition,
temperature and seasonal variations.
There Is definite scope for further research which cannot only be used to enhance the results
ofthis study, but which can also be of use topavement engineers In South Africa in general.
The main areas Identified for research are:
condition assessment.
Further field testing should be undertaken to verify and, If necessary. calibrate, the
results ofthis ltudy.
APPENDIX A
AI.
AI. I GENERAL
The Impulse Deflection Meter (10M), also known as the Falling Weight Deflectometcr
(FWD). Is a device used for measuring the bending of the road surface under the application
of an Impulse test load. The 10M is a cost-effective device which canmeasure up to 300 tcst
points per day. The tesr load Is applied bydropping a mass from apredetermined height onto
a load plate with a diameter of 0.3 m. The 10M Isshown schematically in Figure A I. Photo
AI shows the 10M and towing vehicle.
A1.2
LOADING
Both the mass and thedrop height can be altered in order to vary the magnitude of the test
load. In this way the range of the test load can be varied between 7 and 120 kN. The load
pulse is essentially half-sine shaped with a duration of 25 to 30 milliseconds (SO). This load
pulse accurately simulates the load induced by a vehicle with a speed of60 to 80 1anIh. The
applied load Is measured by means of a load cell situated above the loading plate.
A1.3
DEFLECTION MEASUREMENT
Deflections are measured by means of velocity transducers directly under the load as well as
at 6 points away from the centre of the load (200. 300. 600. 900. I 200 and 1 SOO mm).
Theseoff-sets can also be varied. The Information from the transducers isconvened through
integrating. amplification and rectifying to a direct current signal that Is directly proportional
to deflection. Photo A2 shows a close up ofthe censor beam and load plate.
I zht d 11 lion
the pe k value
nd
e: h tr
ducer, which
tor
In men! r r
ub
u nt
r arc entered by me ns
mput r
mput r
y tern Pro
I'll
yb rd.
or an b dl pi yed on th ml r
I.
the ml ro mput
'I'l
~~ading
Plate
Deflection Sensors
FIGURE AI
DEFLECTOMETER (FWD)
PliO
II
A2:
rill
Al.
TECIINICAL INFORMATION
Load range
7 to 120 kN
Duration of load
2S to 30 msec.
4S seconds
S to 40 C
12 VDC, 20 Amps
12 VDC, 2 Amps
2~
0,1 Kn
0,2 Kn
I microns
2~
2 microns
I micron
APPENDIX B
RESULTS OF MECHANISTIC ANALVSIS
I.
Table BI:
2.
Figures B1 to B19:
usm
(a)
STRucruRE
40.\
3500
XlOO
2SOO
XlOO
3000
3000
3000
3000
3000
3000
3000
3000
3000
2SOO
2SOO
3000
3000
3000
125m
230
230
230
ISO
300
400
230
230
230
2SO
2SO
2SO
300
2JO
2SO
300
2SO
110
100
ISO
2JO
200
200
200
200
ISO
ISO
ISO
130
140
125m
100
100
100
100
200
200
125 ScJecI
120
110
110
120
120
120
120
120
120
SO
100
ISO
ISO
7S
90
100
90
70
70
70
70
70
70
70
70
70
70
70
40
)()
40
SO
30
70
Q~
70
(b)
ELASIlC MODULUS OF MATERIAL IN LAVER (MPa)
STlucruRE
40.\
J500
JOOO
2SOO
JOOO
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
3000
2SOO
3000
3000
ISO 02
zse
:so
250
ISO
250
300
400
250
250
250
250
250
250
250
JOO
250
200
200
ISO 01
100
200
200
200
100
100
200
100
ISO
100
250
200
200
200
ISO
ISO
100
ISO
I~
120
120
120
120
120
120
120
120
120
120
120
SO
100
ISO
70
70
SO
100
70
70
70
70
70
70
70
70
70
70
70
70
70
70
)()
)()
)()
40
ScJIld
Q~
(c:)
1h.~
----....
STRUCTlJR.E
40,41.
2SOO
JOOO
3500
JOOO
JOOO
JOOO
3500
125 0'2
ase
zse
200
200
210
210
1~C4
500
500
500
400
400
400
300
125 Sded
120
100
100
120
100
10
10
60
SO
30
60
SO
40
30
8 - 1
lGOO
t- ...AX
GOO
Bli
II
1400
..
1100
II
1200
1000
GOO
It
II
... ~
III
~ III II.
II
.
tA.
" "fa.......
400
lit
II
.J
II
300
..
"
GOO
200 .
400
100 .
200 .
o ._ ..,_.
10
110
l_._I-..~~l-J.U _ _ -l_-'--_j.--l.-1-U.l
0.1
10
0.1
SHAPE FACTOR I
- - - - - - . - ..
1 -.--.- - - - - - -
..
III
"
70
110
110
..
I
0.11
I
0.8
..
40
30
,.
II
I ..
0.4 .
.-".
wi'
. ..
.
..
II
20
0.2
10
o __..
~
"
Ii
0.1
ALlO'MBLI!
no',
10
(Million.)
n- 2
I -
1800
--------. .
Y-MAX
..
1400
DO ,
0.8
II
.. ..
0.8
It
DO
~
II
DO
tl
It
II
I~OO
'\.
l'l
II
... .
II
1000
r-l
II
II
..
~j1
800
...
N
0.4
II
..
..
II
II
II
. J4 ........
..
800
400
o.~
~OO
0.1
ALLOMBLE
10
no'. (Million.)
..
300
..",
It It
.1"
..
0.8
III
'11"
111M
I U
.a:-
..
-.
"
..
0.4
..
200
LlllW
.. .,
0.8 -
"
10
'00
'000 10000
ALlO'NABlE E80'1 (Million.)
ri"
. t.
I
400
0.1
SHAPE fACTOR I
I -------- - - - - - . - - - - - - - - - - ,
Bli
500
0.01
800~-------------------
-UU.a_L..U WA' , . . . .
0.00'
0.2 .
_...L~,jM._LUU.iL_~.j'ua._.&_.i~&.'.~&1
0.001
0.01
tOO
1000
ALL~ILII!IO'.,Million.)
0.1
10
'0000
-_
...
_-------- -----
8 - 3
----1
1 ----.------------------
.. /I
..
O.S
280
H
.;01\0
0,11
It ,
00 09
H
0.4
"-
DlIII.." \ II
II
180 .
..
DI
..
200
..
00
00
II
DI
00
..,.
..
00
II
DO
...
..
It
II
100
0.2 '
80
.....
0.001
I .1111.
0,1
0.01
pUM
10
100
VI
.~JU,ll...__.'-.I...&UM
I pUM
I'
1000
10000
"'IIl
...-L.i.&WIL-J-J,.I''le
0.001
0.01
0.1
1111I8
1"11I"
10
100
,all. .
1111
1000 10000
REMAINING LIFE OF
Y-MAX
11100.--------------------
1400 .
1200
1000 ~
liDO fI
III
..
..
Ol
... ~
eoo
..
400 .
200 .
ow.u~
0.001 0.01
....
o . --.6-'.I,uua-l.-l..
U U---...-&J1U"
0.001
I'
"',ee
0.01
0.1
,
~
ALLOWULII8O'. (Million.)
I ' .,.,.
'00
B- 4
LLI
100
300
10
140
..
260
1IO
200
II
It
.....
..
100
It
.. 11
160 II
120
.. .
II
II
so
If.
II
.. !\e
.. W
80
..
100
40
60
20
J..IUII_-,-L!.UUu.-..&_JJ.~'...L.lluw
0.001
J_ .11.1.1&1
0.01
0.1
I
10
ALLOWABLE ESO'S (lAillionl'
100
o
o
0.001
07
300,-----------------,
laOO
'-
1000
..
1110
..
aoo
100
aoo
II
II
100
---------------,
1200 .
10
V-MAX
1400
200
I'
2110
0,01
0.1
1
ALl0WA8LE Eao's (Million
....
-1Cbs
....
..
400
80
200
_.L".ujUll-_,~,oI..UIlll._J....A..JUUlL,'" ,'~,6-,"IU""
0.001
0.01
0,1
AlL~8LI
10
.I
.Itu
100
,~L'.I
161M
0,01
!e0'1 (lAilNonl'
.
'
0,1
..
I
10
ALLO'MILI 180'8
1000
D - 5
300
MU
tL
2110
200
III
It
II
140
II
..
..
1ft
'"1
It
120
It
II
II
II
100
- .~:
1110
..
1111
..
.1
II
It
- -..
.
..
II
It
II
aD
III
II
"b-:- "
"
00
100
40
110
20
o _...J41iUU&l_
0.01
0,1
J_,..&,!.I.J,UJL--l.,1..&lUU1,_.l~..j~1..&"'''
__ .~''
I
10
ALLO'MBl! no's
jl'
1000
100
_,~Io_,L.'UW_-4-"JlUW..._L1.U.l~..L.LUuu..~ ..i.J..lUI
0.01
0.1
I
10
AllO'MBll! !SO's
100
1000
2110
200
..
100
..
..
..
lao
- .
._-
60
O_',,JUI""_~....L"-'U""_'"_JJU,,,,,-,,'-','.'NM~
0.01
0.1
Allo.-Sll!
10
nO's
tOO
.. ,
1000
I
_I
APPENDIX C
FllUres CI CO C9:
2.
C- 1
FIGURE C 1: YMAX VI. AllOWABLE E8O"
OBSERVED FIELD RESUL ra
lOOOO:r----------------------------,
1000
1100
iI
,.
AUOWA&! rw.
"0
C - 2
"T"-----------------------------,
1000
lOOO:r------------------------------,
..
....... -
",:~
-c: ,.
~~~~~~!,I&IIt1...:f...
.
'
C- 3
1000.,-----------------------------..,
l00-y-----------------------------,
C - 4
100.,-----------------------------,
10+--rTTTT'I1rlT"'""-r-rTTTTrTr-"-r-rrTT'l1m-""T"T'T'1"TTTTr-""'T"""T"T1rrmn--'1'-r"TT1m1T-r-T"TT'TTT1t
1,0
.
:.::
.
1'... .. . - .
. . -....::_-\.
~.-
.......
FI.I
.-.
. . 1 _ _ . .
-.,.
I
'._11
... .
. 1,__
- 1__
. . . . . . . . '1
. . . . 4....
' 111
. _
'"11
II I . a . . .
.. 111_.............. .
_ _ 1.11
_
._11. _
.............
. ..
C - 5
. .......
.
- .....-.-...................
..--..
.........
.....-......
a...r:.~.:..ui.:. ~_..
..-.
...
.....
.11
PI
.. .
.-...
... .....
~
.... .
.
._ .. 1_'
...
..._ . . ..
,
..
..
..
C - 6
00
u:u...uw.
.~
00
40
~
30
I~ll
Ifl'
3
4
I~I
'.
~.
'Ull
II
n~
/41l
I
/I
~4"
~,
~I
Il
Itl
;'Il:l
/UJ
III
III
'.'11
14 I
.ln4
:11111
II
./
:Ill 4
:11111
41/
4:1t>
13
14
In
10 "
"
30
II
80
~'
tWiW,.'
III
1
J
;>~
:Ifl
411
4n
III
II
II I
el
1I
I
II
II I
III
011
18
nn
""-
II,
II
Gil
10
"
"
'3
'4
'011
1:10
130
14'
1111
III
Ie'
eo -
IIlJ
tn1
"'
'WI
111:1
"
'0
lila 1411
--------.:.:..---,,-----~------aNiG[
GG
100
1:10
130
141
3
I
11
'I
III
III
III
:12
:Ie
30
'II
"
:Ill
:JO
J4
:JII
I
It
40
liP
"p
1/:1
,
:I
eII
30
10
34
38
11
1:1
'3
14
I tl
4:1
4e
4G
113
II I
42
411
411
113
III
III
I
I
20
~
~
~~
~
~I\,Imm_.,"_,.
u
~ ~.
,
.~
v,
10
.'
~ "
.~, ~ ~
1111 HUlllbet
.'
Ill:l
le:l
11:1
00
<
14:1
.-
;>t>
'If'
II
"
'0
131
IU
Itl1
IOJ
_._-_.,-._--'OJ
:s ,
20
II
111
1:1:1
Numb.r 01 M...u,.m.nl.
~,
,~
Bin Numb.r
llU1.til1.
~,
lv'
III
II'
Bin Numb.r
Numb., 01 ......ur."'.nU
40
80
III
101
13
14
III
10
bl
1/1
,:I
4 :1t>
41>1l
no
II
.. 1;1
':S
1O
II'
!)
40
Ilbmm!\lU,.
11O
1
:J
4
10
--~
~
I
60
11)
10
:s
_ _ ._ _ . ,. _ _ _ _ _ . _
00
20
'}
20
I~
I til)
11~
.14 ,
1;ltj
104
\\
00
UA:iUL.
10" 11 II
"
1_.
r~ ~u.w
U.l
&1
lIUJ!U
\
m1
mm=
m:~
I
I
t:
'-1
'0"
11 II U ,
C - 7
eo
~"
I
2
III
:1
111
4
6
II
.~
.~
eo
II
Q
'0
II
12
.:1
14
40
'"
"14
In4
1'.1"
2~"
lUll
12ft
.,110
2llO
2Q7
:l21l
~Ql
:I~7
311U
:lIIQ
421
46J
41111
~ III
421
41lJ
41111
III
-----_.'._"
30
'N
II\)
a2~
j~7
~III
...... tl
~-
20
10
ol.U\lm.l
I
:I
,111m.."
m ii:l
10 II
II 1:1
14
..
Bin Numb"
'------------_._----
. - -' ._--_.--,,--_ ..
_-----
C - 8
eo
Number 01 Me ure",enl.
~
tl:tLlw.
I
60
~o
;-11
I I
.''"H'
~,
~'Ml
.lOU
II
21111
:.J}t1
/
/I
:11"
~"4
II
"0:.'
"TM'"
HO
un
41ft
!),II
12
!IIII
t,~.,
IJ
Mh
/>11:1
14
!)QJ
11:11
":11
_._-_._
II
80
_-
II0Q
...
20
a a
10 "
II 13
'4
O~'
II
IIIJ 1/11
1"1' '011
,OQ'
~~~:
I'
80
I~~' :;~
e,
10
10
100
40
4"'
20
of
J<l4
.. O~
II
I!I
eu
II I
IJ~
1/.1
:1
4
10
30
I t'"j
')1
n/
'03 3'1
3' I . 340
I'
34g 311
'3
3'"
14
III
40&' 433
433 4111
40&
.JWJ~.r.I~---4-_.......~___''__'.....1aJ
4
'0"
II 13 14 "
Bin Number
Bin Numbe,
Numb.r 01 Me ..ure",enl.
80
80r----:..-------:===-=
60
7J
311
40
Il~
4Q
II~
,.,
/&
116
1
II
1111
10'
'0
114
",
"
183
13
14
'110
"0
Ig,
II
30
1&
140
,
I
JIl
60
II
II
1
101
114
rr
140
'!l3
11111
'/0
.o~
IlAtillf.
IWi.J:ILQ.
23
40
10
0l11.lilI.
I
ur
II
g
10
~o
.,
I'
'3
30
14
'1
;>o~
II 13
13 III
III '3
'3 . 'II
33
:llI
30 43
43 411
411 &3
113 &11
1111 03
113 011
011 13
73
111
711 113
'11
33
20
a a
1 0 " " 13 , .
II
1111I ""mb.,
IIln Number
1..-
.. -
-- ... - . - - - - . - - - - - - - - . - - - - - . - ..
C - 9
,--------
eo
14
111
III
dl
2
J
4
Ib4
10'1
lUI'
141
~f'.
II
10
II
eo
"
13
14
16
14 I
2114
~/II
:111
411>
4611
601
641>
611lJ
1I:!7
I "
1M
.':'11
JfI
41'
4 ~It
Olll
tl4~
bllll
11:11
1I1b
40
20
01-.=....",
'0 II
II 13
I.
la
81n Hu"'b"
'--------------_.--~._-- ...
"
----_._------ -
C - 10
cn
FIGURE
DiSTRIBUTION OF MEASURED
Bli VALUES. E2 TRAFFIC CLASS
'00
Number of ......ur.m.nll
~--~- - - - .~----~. ----- --UAUW.
WfUig
80
Numb.r 01 "'.lIur.m.nl.
-- - ---------------=====--"-~
wtUtll IlAI:iQf;
I(}O
In!>
I~!i
'1
2n
4
e
II
;'/11
'l'n
2111
:t:14
:II)
3J4
I
441
HI
"0.1
I
~
"n"
/lill
~11l
til'}
12
1112
11U
I;'U
'"11
13
14
1/\
1141
nUll
'n"
,00
;JUI
"OJ
"110
III
II
eO'
,20
'UI
/lUll
U"4
Il
II
II
10
11
1'1
13
14
t&
110
~'
~
\'
110
'~
,x
\\
40
~.
1111
&&
114
IIll
1&8
124
IIlIl 1111
1112 ;/11
7111
12'
7Ub
2111
:l.lO
;0 Oil
:Ml4
aJQ
allft
3C4
4n
3110
437
4'"
IlOI
41l'
1101
1l:l4l
IIJ(!
11'0
----------
\,
~
40
~\'
~\
20
20
I ILdL
x
o
,
:t
10 "
..
I:t
'4 ..
Bin Numb.,
-- - "'::.,-=-==-~-:": .--
BinNumb.,
r
'0 "
II I:t 14 ,.
:t
,00
Numb.r 01 "'.lIur.m.nll
~~
,
I
3
4
&
110
e
1
Il
II
10
eo
II
\~
&
40
I'
13
'4
1&
17
3'
!il
III
IIIl
111l
13C
I &II
11&
11l&
'1&
73/\
'/\!i
"4
'1l4
3'
Po,
ltl
Otl
I III
13C
I !ill
eo
Il
II
Il
11/\
'Il"
'1&
7:)"1
'&11
eo
,,.
13
III
314
40
~ .~.
,~
~ x .~
20 ~ ~' ~
.~
,~
~~
\~
20
~,\,
.~
'-------------------------------
.'
10
"12
14
;>04
------
4 '0
'0 . I'
,r .
'3
'13 30
3O.3Il
311 47
4:1 411
41l 1111
611 II'
1I,
1111
1I11 r4
14 III
1I1 8'
1I1 114
114 tOO
C - 11
FlaURE
cn
DISTAlBUTlOH Of UEASunCD
H~M'
110
llo.!4.!U llA4...o.L
e.
)/
e.
1
li'
1'0
",
I: "
'0 ,
, ::8
'.'
,I!',"
II
'u
"
"I )
..
IIl/IHvaMI
10 II
''''
1:0
:~n
1>J'
>/1
l~~
>a.
t: ..
l~"
t"O
-:u
1]1
''.. ..
. ,.
II I I
jeI.
.,,!
."
'b~
II
"
C - I;:
~:w.
:V":""'"
'" t
.~n
60
e
I
0
t'n
'M
-,
~~
..
u.'t1
",;itt
"
"
I Oct"
,t/,
./ft
tIft.
tVltl
I ''it)
'4
,~
..e
.0
&
I
04
"
., I
'"0
i:i:
I'
r'
mm
i;
\ ,~ ~ rnm~mmfll
IIIIH"",",
,
",
, t,
''0
.0
I;
,~
.0
""
'J
11<1
""
)oIHt
"J
10
I I
1
J
'00
'''w
"p
::~
10
"1-1
'0
1U
"
,, ,.'1I-. , :'"!.
'" ,.,
,.,
I
I))
111<1
".a
"
I~
;~
..
.>0
I j
"
II "
cn
.. ., II
"
tOl
J.llJ
FIGURE
DISTRIBUTION OF MEASURED
LLI \'ALUES. E1 TRAFFIC CLASS
:0
.-J4
11
II
,.~
1
J
I: rI:
lW~06L
lOll
'01
tn",,"
"/J
"'
'u
III
I.'
a
.>H
;">e
""
, ."".,,
" 'M 1110
alkl ,.
,
":\.l
rJlt.
c.:LlJ,ll.
..
e'j
III
I.U
... 141
"I~
hI
II
...
.. IIII
II....
11
il.A:Wt
~~
ICI
"
.,.."
,.,
'~~)
,In
.r.,
o'\to)
fl'
,I
.J 10d
:)oftt!
It
-"'l\
"J
.:,
,"
"
'.... "
"
,. .,"
.. I
11
)I
I'
'2 ",ot
,oe
'"
,"
I" ..'"0
110
'"
I.'
II
'"
"J
~
\
\
'0
I
\
.~
. -. ....
""
I
I
"
~m
........
WW:t'd' Wm.mm, __ ,
II II
"
:t,
I ' I. f
C - 1:1
120
l1JlUtll. lI.AUUl.
:'
J
IU
.. 0
10
4\1
It)
10\1
,.
IOU
&
G
I
140
141,)
, Itl
1'0
:l\)()
;'00
'1.10
It
1:)()
~"o
'Ino
10
~oo
I~tl
II
JZO
:''\0
'11"
~!I()
1'1
:.~o
l:l
14
~fn'
.. "'
410
III
441
H'
4/1
40
20
o,
:a
bm_AU_~_'_
7
'0 t t l l t : J
t. It
Bin Nultb.r
-_ _----------_
..
"
C - 14
~a
NUlIIber of "'uttlll.nl.
---- --
- -------
--------;:::=====::;1
:sa --
llltUUI. Il.W1L
l1ltWll. UAUW.
~O
201
.KllI
:JOU
4011
:I
4
tl
e
I
o
o
10
la
60J
:so
COl
CUO
10/
IIUI)
nOli
UO~
211-
I/O;.>
1000
1;0
1000 I 11111
11011' I;oM
120"
I :II"
13
'4
16
1:lIH' 14/11
14111
1/1/0
11)/0' 1011
II
20
40"
tlOl
1101
1I01l
101
:I
4
II
I
0
0
10
II
20 .
1"1
13
14
Itl
III
loe
201
26e
310
306
10
.,.
620
603
lUI
002
,.0
eOI
8&6
010
10
10
]1
a
o
llW.lill. IlAtiaf.
I
;.>
:I
4
e
0
I
20
I
0
10
~~
~
~
.~
II
~~
11
1:1
14
III
K~\
~ ~ ~
.~ ,~~ ~
L.:-~
;'>11
04
04
00
00
131>
I/O
131>
I/O
;ooc
"lOll
11"/
;04 1
"111
:111
H/
3411
:11"1
34ft
3113
410
:Ill:l
410
4'\4
464
400
400
a.?~
Iln-_ . ._ /1111-__
, .S\ ~
~.
~,
'~
10
~.
~,
"
~.~
~
~'\
.~
'\
"
.~
~'
~.
'\
~~
\'
\\
~
~ " \"-'.\ ~
~ @ I. ~
~\
'\
\~
'l
11111 Hu",ber
ldml
10 1 I
II If II
NUtrlber of "'...urem.nl.
COI--------';::::===~
~ IlA.!iQf.
NUlIIber of "'lIfelll.nl.
~O
Bin Humb.r
Bin NUlIIb.r
CO
02 .
"I .
"101
260
310
3tl6
410
4 I .
11111
llft3
03/
002
'.0
1101
lIee
I
"I
10 I I
I' II
It
II
I
2
:I
4
II
30
7
II
10
II
12
13
14
III
I III
III :JO
:1041
41 e3
113 110
O. - III
70 III
III 00
00 110
110 112
171 133
133 U1>
l4e 11>11
1011 1M
1111 110
C - 15
,--- _._----------.-
_.....
---
_-----_._... __._-----
70
LUll-bll.
,
eo
:J
4
I)
,
II
110
o
10
II
12
13
14
III
40
tww..
.I~
10
:11\
'> I
I)'
"'
tl,
II;'
Ull
114
I :10
14ft
Itl:'
/ll
VII
114
IJO
1411
1111
1 '"
104
1711
11/4
7011
:'011
n~
nil
741
74 I
7&1
II 1:1
,. II
:10
20
10
:I
10 II
8111 Number
__
"---------------_ ..
._--------..... -.-.
REFf:RENCES
I.
Bosman, J. Flow of ~oods OD South African roads and the influence thers;of on road plaDnjnK
and dcsilm. PhD Th~is (Afrikaans), University of Pretoria, May 1988.
2.
Kleyn, E.G. Aspekte van PlavelseleyaluerioK eD =Ontwer:p 500$ bepul met behulD yaD die
Dinamiclic
3.
Ke~dpenetrometer.
Hudson, W.R., Elkins, G.E., Uddlng, W. and Reilley, K.T. EvaluatioD of Pavement
Deflection MeasurinK EguiameDt. USDOT, Federal Highway Administration, Report No.
FHWA -TS-87-208, WashiDgton DC, 1987.
4.
Lau, C.L., Scullion, T., Chan, P. Modelling ofground penetrating radar wave propagation
in pavement systems, Proceedings: 2Dd Annual MeeUne. Transportation Research Board.
Wa.~hiDgton
S.
Epps, J.A. and Monismith, C.L. Eguipment for obtainjoe pavement condition and traffic
loading data. NCHRP synthesis of Hiehway Practice No. 126. Transportation Research
Board: Washington D.C, 1986.
7.
Bush, AJ. Non.<JestcuCliye testing for Iiehl aircraft a3yemeDls Phase I, EyaluatioD of non
destructiyetesljoedeyices. USDOT, Federal AviatioD Administration, Report No. FAA-RD
8(,-9: Washington D.C, 1980.
8.
Smith, R.E. and Lytton, R.L. Synthesis study of non-destructive lest10e devices for use In
overlay thicknesS design Q( flexible DayemenlS. USDOT, Federal Highway Administration,
Report No. FHWAlRD83/097:
.
'
Wa.~hington
D.C, 1984.
9.
10.
II.
Stewart Scott Inc. Structural c1assificalion ofpavement structures using measured deflectloD
bowl parameters. The IOMP Program, Internal Report, Sandton, May 1990.
12.
Deflection Bowl response of Typical PavemeDt Structures. Project Report PR90/I02, South
African Roads Board, Research and Development Advisory Committee, Pretoria, April 1991.
13.
Joubert, P.B. QWs!e-ljnes for the Cost-Effective Design. ImplemeDlatjon and Operation of
a Pavement Management System. Draft MSc Thesis, University of Pretoria, 1992.
14.
AASHTO Guide for Design of Pavement Structures. American Association of State Highway
and Transportation Officials, Washington, D.C., 1986.
IS.
Sebaaly, B., Davis, T.G. and Marnlouk, M.S. Dynamics of falling Weight Deflectometer,
Journal of Transportation Eneineering. Vol. III, No.6, November 1985.
16.
Scullion, T. Incomoraling a Structural Strength Index IntQ the Texas pavement Evaluation
SnwJ]. Texas Transportation Institute, April 1988.
17.
Stock, A.F., Yu, J. Use of surface deflection for pavement design and evaluation,
TransoonatlQn Research Record. TRR 954, Wa.,hlngton D.C., 1984, pp, 64 - 69.
18.
All, N.A., Khosla, N.r. Determination of layer moduli using a failina weight deflectomerer,
IrwoonatiQn Research Record. TRR 1117, Washington D.C., 1987, pp. 1 - 10
'
19.
Roesset, J.M., Shao, K. Dynamic lnterpretetlonof dynaflect and falling weight defleetometer
tests. Transponatlon Research Record, TRR 1022, Washington D.C., 1985. pp. 7 - 16.
20.
Bush, AJ . Alexander, D.R. Pavement evaluation using deflection basin measurements and
layered theory. Transportation Research Record. TRR 1022, Washington D.C. 1985. pp.
16 - 28.
21.
Jung, F.W. Direct calculation of maximum curvature and strain In asphalt concrete layers
of pavement from load deflection basin measurements. Transportation Research Record.
TRR 1196. Washington D.C. 1988. pp. 125 - 132.
22.
D.C., 1986.
23.
1963.
24.
Peutz, M.O.F. Van Kempen. H.P.M. Jones, A. Layered systems under normal surface
loads. Highway Research Record No. 228. Highway Research Board. 1968.
25.
based Procedure to Back-ealculate Layer Moduli from FWD Data. Res. Report 2-1887
1123-1. Texas Transportation Institute 1988, College Station. Texas USA.
26.
27.
loannldes, A.M. Barenberg. E.J . Thompson, M.R. Finite element model with stress-
'
28.
Yoder, EJ., Witczak, M.W. Prjnciplesofpayement desjen. 2nd edition, New York, Wiley'
lnterscience, 1975.
29.
Hoffman, M.S. and Thompson, M.R. Back-ealculatlng non-llnear Resilient Moduli from
Deflection Data, Transportatjon Research Record. TRR 852, 1982, pp. 42 - 5 I.
30.
1987.
31.
Rada, G., Witczak, M.W. Comprehensive evaluation of laboratory resilient moduli results
of granular material, Transoortation Research Record. TRR 810, Washington D.C., 1981,
pp, 23 - 33.
32.
Maree,
J.H.
Aspekte
van
dje Ontwerp
Uzan, J. and Sides. A. The Effect of Contact Area Shape and Pressure Distribution on
Multi-layer Systems Response. Transpooatjon Research Record. TRR 1117, 1987, pp. 21
24.
34.
35.
Jordaan, GJ.
Rehabiljlatjon Design. RIT. Research Report DPVT 54, Pretoria. February 1989.
36.
Jordaan, OJ. Guidelines towards the use of the AsphaIJ Institute Method for Payement
Rehabilitation Desizn. RIT. Research Repoo DPVT 54, Pretoria, February 1989.
37.
38.
Division of Roads and Transport Technology. Structural desiKn of pavements for inJerurban
and rural roads. TRH 4. Council for Scientific and Industrial Research, Pretoria, 1988.
39.
Pavement Structures in the Cataloeue of the Draft IRH 4 1980. National Institute for
Transport and Road Research, Pretoria 1980.
40.
National Institute for Roads and Transport Technology. Guide-lines for road construction
materjals, IRH 14. Council for Scientific and Industrial Research, Pretoria, 1985.
41.
Wonnacott, T.H., Wonnacott, R.J. Regression: a second course In statistics. New York,
John Wiley & Sons, 1981.
42.
Number Croncher Statistjcal System. Version 5.03 5/90 Copyright Jerry L. Hintze 1990,
Kaysville Utah.
43.
Lacante, S.C. Comnaratjye Study of Deflection Basins Measured on Road Structures with
various Non-destOJetjve Measurine Deyices. Masters Diploma Thesis, Pretoria Technikon,
November 1991.
44.
Paterson, W.D.a. Road Deterioraljon and Majntenance Effects. The Highway Design and
Maintenance Series, Baltimore and London, John Hopkins University Press, 1987.
45.
46.
CPA Roads and Traffic Administration. Qltjmal Multj-lerm oon-djstQuine smoothing and
grouping of data. Developed by N. Dumas.
47.
.
'
Trondhclm,
48.
Emery, SJ. PrOOiction of Pavement Moisture CODlent in Southern Africa. National Institute
(or Transport and Road Research. Research Report RR391, Pretoria, 1984.
49.
SO.
I,YD1(WPI
e '