A DESIGN MODEL FOR WATERBOUND MACADAM BASED ON HEAVY

VEHICLE SIMULATOR AND LABORATORY TEST RESULTS

H L THEYSE, E SADZIK and J P NOTNAGEL

Transportek CSIR, P O Box 395, Pretoria, 0001

GAUTRANS, Private Bag X3, Lynn East 0039
SANRAL, Private Bag X5, Alkantrant 0005

INTRODUCTION

Transportek recently completed a draft guideline document on the selection, design and
construction of waterbound macadam pavements which was co-funded by the Gauteng
Department of Transport and Public Works (GAUTRANS) and the South African National Road
Agency Limited (SANRAL). One of the objectives of the project was to develop a rational design
method for waterbound macadam from existing Heavy Vehicle Simulator (HVS) data and
laboratory test results and to incorporate the new design method into the South African
Mechanistic-empirical Design Method (SAMDM). Previous publications on the SAMDM did
not include any information on the mechanistic-empirical design of waterbound macadam
pavement layers. The information presented in this document is additional to the SAMDM (1)
and should be used in conjunction with the SAMDM. The following items are covered:
recommendations on the load sensitivity of waterbound macadam base layers,
recommendations on the stiffness input parameters required by the SAMDM for
waterbound macadam material and
design transfer functions for the mechanistic-empirical design of pavements with
waterbound macadam base layers.

The data on which the above recommendations are based were obtained from Heavy Vehicle
Simulator (HVS) test data and triaxial laboratory test data. The design transfer functions were
developed from laboratory test results for two practical cases which may be encountered on
waterbound macadam construction projects:
The first case is for a WM2 (2) waterbound macadam base layer constructed on a granular
subbase or relatively thin cement-treated subbase. Because of the relatively weak
compaction platform provided by the subbase, the base layer is at 84 % of apparent
density. Because of the lower density, the saturation level of the base layer is about 30
% for normal moisture conditions.
 2

The second case is for a WM1 (2) waterbound macadam base layer which is constructed
on a sufficient compaction anvil consisting of at least 125 mm cement-treated material
and the base density is 88 % of apparent density. Because of the low voids content, the
saturation level of the base layer is about 45 % for normal moisture conditions.

The laboratory design models showed good correlation with field test results from the HVS but
seem to underestimate the load sensitivity of waterbound macadam. The results from HVS tests
were therefore used to investigate the load sensitivity of waterbound macadam base layers. HVS
test results also contributed to the general information available on waterbound macadam and
results from the following HVS tests were used in this investigation:
Gauteng HVS tests 400a4 and 402a4 done near Cullinan (3)
Northern Province HVS tests 404a4 and 405a4 done near Louis Trichardt (4)
KwaZulu-Natal HVS test 329a3 done near Umkomaas (5)
HVS tests 128a3 and 178a3 at Marianhill, Pinetown (6)
HVS tests 362a3 and 366a3 at Seaslopes, Margate (7)

A brief description of the results from the HVS tests at these locations is given with specific
emphasis on information that is pertinent to the objectives of the guideline document.

HVS TEST RESULTS ON WATERBOUND MACADAM

HVS tests 123a3, 124a3, 128a3 and 178a3 at Marianhill, Pinetown

These HVS test sections were situated on Main Route 85 (MR85) near Marianhill, Pinetown. A
crusher sand was used as the filler for the waterbound macadam base layer of these sections.
Initially, the only difference between sections 123a3, 124a3 and 128a3 and section 178a3 was the
thickness of the pavement layers. After the disappointing HVS test performance (6) of sections
123a2, 124a3 and 128a3 (which was constructed without slushing the waterbound macadam base
layer), it was decided to recompact the base layer of section 178a3 by using a slushing process.
More detail on the effect of the slushing process will follow later.

The structural detail of the two test sections are as follows:

Sections 123a3, 124a3 and 128a3 Section 178a3
50 mm continuously graded asphalt 65 mm continuously graded asphalt
150 mm waterbound macadam 235 mm waterbound macadam
150 mm C1 cement-treated crushed stone 125 mm C1 cement-treated crushed stone
150 mm C3 cement-treated sandstone 125 mm C3 cement-treated sandstone
600 mm dumprock and sandstone 600 mm dumprock and sandstone

The test programme for section 123a3 consisted of 890 000 load repetitions at 70 kN with a tyre
inflation pressure of 600 kPa of which the last 60 000 repetitions were applied with water entering
the base layer through 38 mm diameter perforated pipes which were inserted into the layer. 600
000, 100 kN load repetitions at 600 kPa inflation pressure was applied to section 124a3 with
water being applied from 480 000 load repetitions onwards. Water was applied to the base layer
of section 128a3 for the full 600 000 load repetition duration of the test. The trafficking load was
increased from 40 kN at 600 kPa inflation pressure to 60 kN at the same tyre pressure from 220
000 repetitions onwards. The trafficking load for test 178a3 on the thicker waterbound macadam
base layer was 100 kN at 600 kPa inflation pressure for 973 000 load repetitions with water being
 3

applied from 600 000 repetitions onwards.

The surface rut for test section 123a3, 124a4 and 128a3 increased rapidly so that all these sections
reached a rut of 20 mm within 400 000 load repetitions of their respective trafficking loads. In
all three cases the rapid deformation of these sections seems to be related to either rain water
entering the base layer or the water applied during the test. This indicates that although
waterbound macadam may act as a drainage layer, the performance of the layer will be affected
by the presence of excess moisture. The density of the waterbound macadam layer was also not
sufficient, in addition to the influence that the moisture condition had on the performance of these
test sections. The density outside the HVS test section for test 128a3 was 84 per cent of Apparent
Density (AD) and this density is believed to be representative of the density before testing. The
density increased to 89,5 per cent of AD after the HVS test was completed. The poor
performance of sections 123a3, 124a3 and 128a3 may therefore be ascribed to the combined
effect of low density and high moisture content.

Based on the poor performance of the HVS sections tested on this site, a decision was taken to
recompact section 178a3 before HVS testing (8). The additional compaction basically consisted
of rolling and vibration of the layer followed by the slushing of the layer. The density of section
178a3 before recompaction was 85 per cent of AD and increased to 86 per cent with additional
rolling and then increased further to 88,6 per cent after slushing. The additional benefit of
slushing is thus clearly illustrated by these data. The density of section 178a3 increased slightly
more to about 90 per cent of AD under HVS trafficking.

The increased density of section 178a3 had a marked effect on the water absorbed by the layer
through the perforated pipes installed in the layer (8). The average water absorption was 15
litre/hour/hole for the low density sections 123a3,124a3 and 128a3 and 0,15 litre/hour/hole for
the high density section 178a3.

The higher density and hence lower water absorption of section 178a3 resulted in significantly
better performance with the section only reaching 20 mm rut at about 1 million 100 kN load
repetitions (6). The waterbound macadam layer was, however, again the main contributor to the
permanent deformation of all four of the test sections.

Horak calculated the bearing capacity of sections 123a3 and 124a3 as ranging from about 3 to 12
million standard axles with that of section 128a3 being even lower at between 0,8 and 3 million
because of the higher moisture level in section 128a3. The bearing capacity of section 178a3 was
calculated as being between 10 and 20 million standard axles.

HVS test 329a3 near Umkomaas

Based on the encouraging results fromHVS test section 178a3 near Marianhill, the then Natal
Provincial Administration decided to build a 13 km length of dual carriageway on the N2 near
Umgababa during 1985, using a base course of waterbound macadam with a natural sand filler.
HVS test section 329a3 was located on this section of road. A second section with a crushed
stone base layer was also constructed close to the waterbound macadam section and tested with
the HVS to enable a comparison of the performance of the two base layer types.
 4

The pavement structure consisted of the following:

40 mm semi-gap graded wearing course combined with a 25 mm continuously graded
levelling course
150 mm waterbound macadam base layer
125 mm cement-treated crushed stone subbase
125 mm lime stabilised gravel subbase
150 mm G7 subgrade
150 mm G9 subgrade
G10 subgrade

The HVS test was started off with a 100 kN trafficking load for 794 000 repetitions after which
holes were drilled through the surfacing and into the base layer for water to enter the base layer
and the trafficking load was reduced to 40 kN. The trafficking load was again increased to 100
kN from just less than 1,3 million repetitions to the end of the test at 1,721 million repetitions.

Again, as was the case with the test sections at Marianhill, the waterbound macadam base layer
contributed the largest portion (between 45 and 60 %) of the total permanent deformation of the
pavement structure. Wright and Hess (5) calculated the bearing capacity of the HVS test section
at between 14 and 16 million standard axle loads. The rate of deformation of the G1 crushed
stone test section was lower than that of the waterbound macadam with the G1 section having half
the rut (11 mm against 22 mm) of the waterbound macadam section after completion of a similar
test programme.

HVS tests 362a3 and 366a3 at Seaslopes, Margate

The only difference between sections 362a3 and 366a3 is the 40 mm asphalt wearing course that
was placed on section 366a3 whereas only a 10 mm slurry wearing course was placed on section
362a3. Both sections have penetration macadam base layers which is in essence a waterbound
macadam base with about the top 25 mm of the layer being penetrated with slurry. The pavement
structures of these two test sections consist of the following:
10 mm slurry wearing course for section 362a3 and 40 mm asphalt wearing course for
section 366a3
25 mm penetration slurry
125 mm waterbound macadam base layer
150 mm C2 cement-treated subbase
no detail is available on the support layers

The bearing capacity of section 362a3 was calculated from the HVS test by Roux and Otte (7) as
3,7 million standard axles and that of section 366a3 as 7,5 million standard axles. Again the main
contribution to the total permanent deformation of the pavement section came from the
waterbound macadam base layer. The relatively poor performance of these two sections in
comparison with the other waterbound macadam sections cited in this report was not explained
by Roux and Otte. McCall et al (9) did, however, mention problems with construction water that
was trapped in the waterbound macadam base layer of the section of road on which the HVS test
sections were situated because of blocked side drains. A high moisture content may therefore
have been the cause of the relatively poor performance of these test sections.
 5

HVS tests 400a4 and 402a4 near Cullinan

Sections 400a4 and 402a4 had waterbound macadam base layers of 100 and 150 mm thickness
respectively. The base layers were constructed using a 53 mm nominal size coarse aggregate and
a natural sand filler. The detail of the pavement structures of these two test sections is given
below:
Section 400a4 Section 402a4
30 mm continuously graded asphalt 30 mm continuously graded asphalt
100 mm waterbound macadam 150 mm waterbound macadam
150 mm C4/G4 sandstone material 150 mm C4/G4 sandstone material
300 mm G4 sandstone 300 mm G10 ferricrete

The quality of the ferricrete selected subgrade of section 402a4 was of a much lower quality than
the sandstone selected subgrade of section 400a4. The density of the waterbound macadam base
layers of these two sections were in places as low as 78 % of AD with maximum values of 81 %
of AD. (3).

Both these sections were subjected to the same test programme with 200 000 load repetitions
applied at 40 kN and a tyre inflation pressure of 520 kPa. Thereafter the load was increased to 70
kN at 670 kPa inflation pressure for a further 200 000 repetitions. Water was added to one half
of the section for a further 100 000 load repetitions at 70 kN.

The difference between these two test and the other HVS test case studies cited in this report is
that the supporting structures of these two test sections were of a much lower standard than that
of the other test sections. Whereas most of the rutting of the other test sections originated from
the waterbound macadam base layers, the layers below the waterbound macadam base layers of
sections 400a4 and 402a4 also contributed significantly to the total permanent deformation
especially the weak ferricrete layer on section 402a4. It must, however, be kept in mind that the
pavement structures of test sections 400a4 and 402a4 are intended for provincial roads and not
national routes. A comparison of the bearing capacities of the full pavement structures of sections
400a4 and 402a4 with that of the full pavement structures on the other HVS test sections would
therefore underestimate the bearing capacity potential of the waterbound macadam. The
estimated bearing capacity of the base layers alone is therefore reported. The bearing capacity
of the base layer of section 400a4 is estimated at 13 million standard axles and that of section
402a4 at 11 million standard axles.

HVS tests 404a4 and 405a4 near Louis Trichardt

These two test sections were located on a section of the N1-28 with a waterbound macadam base
layer constructed by labour-intensive methods. A 73 mm nominal size coarse aggregate was used
with a crusher sand filler. The pavement structure of the two sections was the same and the detail
is as follows:
50 mm asphalt
125 mm waterbound macadam
300 mm C3 cement-treated weathered granite
300 mm G7 weathered granite selected subgrade
 6

The waterbound macadam base layers of these sections were well compacted using heavy rollers
and a slushing process. The density of the waterbound macadam base layer was 88 % of AD in
the vicinity of the HVS test sections.

Section 404a4 was tested for 500 000 load repetitions with a dual wheel-load of 70 kN at an
inflation pressure of 690 kPa. Section 405a4 was tested with a 40 kN trafficking load at 520 kPa
inflation pressure for 380 000 load repetitions after which the load was increased to 100 kN at
800 kPa inflation pressure until 900 000 load repetitions was reached. Water was then added to
the one half of the section for another 100 000 load repetitions.

Both the test sections performed exceptionally well under HVS testing. Although most of the
permanent deformation occurred in the waterbound macadam base layer of sections 404a4 and
405a4 there was very little permanent deformation in total. The pavement bearing capacity of
these two sections were estimated at about 37 million standard axle loads.

Although the general information presented above on the performance of waterbound macadam
from HVS testing does not allow detailed analysis, the influence of the density and moisture
content of the waterbound macadam on the performance of this material is mutual to all these
tests. This is typical of the performance of unbound materials and although waterbound macadam
may in some cases be used as a drainage layer, it is certainly not immune to the detrimental
effects of excess moisture.

RECOMMENDATIONS ON THE STRUCTURAL DESIGN OF PAVEMENTS WITH

WATERBOUND MACADAM BASE LAYERS

Load Sensitivity

Figure 1 shows the bearing capacity estimates (number of load repetitions to 20 mm rut) at
different wheel loads for the base layers of a number of HVS test sections. The load sensitivity
of waterbound macadam may be calculated from the data in Figure 1.

The base bearing capacity is determined by the amount of bedding-in deformation that occurs
during initial trafficking and the eventual linear rate of increase in deformation of the base layer.
These two parameters are shown in Figures 2(a) and (b) plotted against the applied wheel load
for the HVS test sections for which permanent MDD displacement data were available. Both the
bedding-in deformation and the linear deformation or plastic strain rate are expressed as a
percentage of the initial base layer thickness.

The load equivalency factors and load equivalency exponents for the load equivalency law given
by Equation 1 was calculated from the average of the base bearing capacity values shown in
Figure 1 for the respective dual wheel load conditions. The results are summarized in Table 1.
n
P
LEF = Eq. 1
80

Where LEF = Load Equivalency Factor

P = the applied axle load (kN) for which the LEF is calculated
n = the damage law exponent
 7

1.00E+08

Bearing Capacity

1.00E+07

1.00E+06
0 20 40 60 80 100 120
Dual w heel load (kN)

Figure 1: The load sensitivity of waterbound macadam base layers

as determined from several HVS test sections

1.6 2.5
Plastic strain rate (%/million repetitions)

1.4
2
1.2
Bedding-in (%)

1.5
1

0.8
1

0.6
0.5
0.4

0.2 0
0 20 40 60 80 100 120 0 20 40 60 80 100 120
Dual w heel load (kN) Dual w heel load (kN)

(a) Bedding-in deformation (b) Linear deformation rate

Figure 2: The bedding-in and eventual linear deformation rate calculated for waterbound
macadam from the permanent MDD displacement data of several HVS tests

Table 1: Load equivalency factors and exponents for waterbound macadam base
layers
Axle load Average Base Load Equivalency Load Equivalency
(kN) Bearing Capacity Factor Exponent
40 39,1 million
70 14,3 million 2,7 1,8
100 9,2 million 4,3 1,6
 8

The results from Table 1 agrees well with similar results calculated by Horak, (6). The low
values of n calculated for the waterbound macadam base layers are indicative of the ability of
waterbound macadam to withstand high axle loads. It is, however, not only the waterbound
macadam base layer that will determine the load equivalency of the total pavement structure. The
calculation of the parameters shown in Table 1 was repeated using the pavement bearing capacity
data of the HVS sections and a damage exponent between 1,8 and 2,0 was obtained.

The pavement structures of the HVS test sections used in this analysis are typical of the designs
that will be used for waterbound macadam pavements and the low damage exponent of 2,0
therefore seems to be a realistic value for pavement structures with waterbound macadam base
layers. It is therefore suggested that a value of 2,0 should be used for the damage exponent when
converting actual traffic to equivalent design traffic.

Stiffness Input Parameters

Table 2 shows a summary of back-calculated effective stiffness values for waterbound macadam
material from a number of HVS test sections for a 40 kN dual-wheel test load. The initial values
were recorded at the beginning of the test but the effective stiffness soon reduced to values similar
to the average values listed in Table 2. The effective stiffness results calculated at different
positions on the HVS test sections are given for the initial and average values to indicate the
spatial variability that may be expected in the effective stiffness of waterbound macadam. These
values are in general higher than the values that would be expected for crushed stone material.
Hefer (10) found similar values of stiffness for waterbound macadam material using Falling
Weight Deflectometer data. One of the interesting aspects of the resilient behaviour of
waterbound macadam material is that for thinner layer thicknesses, the effective stiffness of the
layer increases. This is because of the relatively fewer load transfer points in a thin layer with
large aggregate particles in the layer. This boundary condition phenomena is not readily repeated
in the laboratory unless squat samples are used.

Table 2: 40 kN test load back-calculated effective stiffness moduli for a number of HVS tests
HVS section Effective stiffness modulus (MPa) Comments
Initial Average Wet
329a3 873, 1144 628, 622 - 150 mm WM, reasonable support
366a3 1318, 1634 1162, 1250 - 125 mm WM, extremely good support
400a4 1517, 2031 1295, 1714 520 100 mm WM, reasonable support
402a4 481, 1104 441, 472 335 150 mm WM, poor support
404a4 465, 835 371, 629 - 125 mm WM, good support
405a4 802, 1117 480, 964 267 125 mm WM, good support

Theyse (11) did extensive static and dynamic triaxial testing of waterbound macadam material.
Dynamic triaxial secant stiffness modulus data from this work were used to developed the
effective stiffness (Eeff) model for waterbound macadam given in Equation 2. This model
incorporates the effect of relative density (RD), saturation (S), confining stress (3) and shear
stress condition (SR) on the effective stiffness response of waterbound macadam and yields
stiffness results comparable to those back-calculated from depth deflection data. The relative
 9

density and degree of saturation is expressed as a percentage of the apparent density and inter-
particle voids respectively (88 % and 30 % for instance). The quantification of the shear stress
condition by the stress ratio (SR) is explained in a subsequent section.

Eeff = 3932,71 + 54,96 RD 1,54 S + 5,87 3 3,07 SR Eq. 2

The correlation coefficient, R2 was 0,651. The parameters included in the model therefore explain
65,1 per cent of the variation in the effective stiffness response. The effect of compaction on the
stiffness of waterbound macadam is evident from Equation 2 with the coefficient for the relative
density being the highest.

Table 3 provides suggested stiffness input values for the mechanistic-empirical design of
waterbound macadam material based on the model given above with allowance made for
increased stiffness for thin (100 mm thick) layers.

Table 3: Suggested stiffness moduli for the mechanistic-empirical design of

waterbound macadam base layers
Material Layer Dry condition (30 - 45 % Moist condition (60 %
code thickness saturation) saturation)
(mm) Well supported Relatively Well supported Relatively
by an intact weak support by an intact weak support
cemented- provided by a cemented- provided by a
treated subbase granular or treated subbase granular or
creating equivalent creating equivalent
confinement of granular confinement of granular
the waterbound subbase layer the waterbound subbase layer
macadam layer macadam layer
WM1 100 1200 800 1100 640
> 100 1080 620 1010 550
WM2 100 1000 600 900 440
> 100 880 420 790 330

Design Transfer Functions

The transfer functions of a mechanistic-empirical design method contains the performance data
for a particular road-building material and these transfer functions need to be calibrated using
measured performance data. The performance in terms of permanent deformation of waterbound
macadam using crusher sand and natural sand fine aggregate was tested using the dynamic triaxial
test (11). Bearing capacity estimates obtained from this study agreed well with HVS bearing
capacity estimates for a standard design axle load of 80 kN. It was therefore decided to use the
laboratory derived design model for the mechanistic-empirical design of waterbound macadam
as it allow much wider variation in density and saturation levels than HVS test data do.

There are, however, two components to the design model. Firstly, the static shear strength
parameters of the material need to be known and secondly the performance of the material needs
to be known for a range of stress conditions. As waterbound macadam is basically an unbound
material, permanent deformation is the dominant mode of distress of this material as was
 10

highlighted by the HVS test results from the previous section.

Equation 3 gives the model that was obtained for the friction angle () of waterbound macadam
material as a function of the relative density (RD) and saturation (S) of the material. The relative
density and saturation is expressed as a percentage of the apparent density and inter-particle voids
respectively (88 % and 30 % for instance).

= 26,38 + 1,021 RD 0,171 S Eq. 3

The correlation coefficient (R2) for the model was 0,981. Again, the relative density has the
biggest influence on the value of the friction angle. The cohesion (C) of the waterbound
macadam material was found not to be a function of the density or saturation and an average
value of 74 kPa was obtained for the cohesion. Figure 3 illustrates the relationship between the
friction angle, density and saturation of waterbound macadam material.

Waterbound macadam friction angle as a function

of relative density and saturation

60

58
( )
o

56
Friction angle

54

52

50
0.25
0.30
%)

48
0.35
(

0.40
on

46
0.45
ati

44 0.50
tur

0.88 0.55
0.86
0.84
Sa

0.82 0.60
0.80
Relati 0.78 0.65
ve den
s ity (%)

Figure 3: The relationship between the friction angle, density and

saturation for waterbound macadam material

The friction angle values obtained for the waterbound macadam material is not any higher than
the friction angle values expected for crushed stone material. The values obtained by Theyse (11)
were verified with additional testing at Spoornets track testing facility. The previous assumption
that the friction angle of waterbound macadam should exceed that of crushed stone and that the
permanent deformation performance of waterbound macadam should therefore exceed that of
 11

crushed stone is therefore not valid. The work done by Theyse (11) has, however, also indicated
that static shear strength parameters are not unique indicators of performance under dynamic
loading. Materials with excellent static shear strength have performed poorly under HVS and
dynamic triaxial testing. The maximization of the static shear strength parameters for a particular
material type through achieving the maximum practical density and the lowest practical saturation
will, however, ensure the best permanent deformation performance for that material.

Equation 4 provides the formula for calculating the bearing capacity in terms of the number of
axle load repetitions (N) that can be sustained before a certain level of plastic strain (PS) is
induced in waterbound macadam material from the dynamic triaxial test results by Theyse (11).
The definition of RD, S and SR is as given previously. The model is based on the combined
dynamic triaxial test data for waterbound macadam material with crusher sand and natural sand
filler. The stress ratio (SR) is analogous to the previously used safety factor for unbound
materials and the definition of the stress ratio is given in Equation 5.

log N = 1,891 + 0,075 RD 0,009 S + 0,028 PS 1,643 SR Eq. 4

a a
1  3 1  3
SR


m
1  3 Eq. 5
3 tan 2 45E   1  2 C tan 45E 
2 2

Where = principal stress (kPa)

= shear stress (kPa)
= internal angle of friction (E)
C = cohesion (kPa)
1m = maximum allowable major principal stress given , C and 3 (kPa)
1 or 1a
w
= working or applied major principal stress (kPa)
3 = minor principal stress or confining pressure (kPa)

The correlation coefficient for the model given in Equation 5 is 0,822 and a graph of the predicted
bearing capacity values plotted against the observed values is shown in Figure 4.

Equation 4 provides a general design model for waterbound macadam. If the conditions for the
two design scenarios from the Introduction are substituted in Equation 4 and the plastic strain is
expressed in terms of a 20 mm rut which is the terminal condition for permanent deformation,
Equations 6 and 7 are obtained. These equations serve as the basis for the mechanistic-empirical
design of waterbound macadam base layers with thickness between 100 and 150 mm for the two
design scenarios. Draft TRH4 (12) stipulates the approximate design reliability for different road
categories and the South African Mechanistic-empirical Design Method incorporate design
reliability by using transfer functions at certain offsets from the basic model, based on the
distribution of the data around the basic model. The offsets obtained from the analysis of the
residuals of the model given in Equation 4, are given in Table 4 and should be used for the design
of pavements for different road categories.
 12

8
Predicted log(N)

5
5 6 7 8 9
Observed log(N)

Figure 4: Predicted bearing capacity plotted against observed

bearing capacity for waterbound macadam material with either a
crusher sand or natural sand filler

Conditions for the 1st design case:

Relative density: 84 %
Saturation: 30 %
Design model:

log N = 7,905 + 0,028 20 100 1,643SR Eq. 6
t

Conditions for the 2nd design case:

Relative density: 88 %
Saturation: 45 %
Design model:

log N = 8,086 + 0,028 20 100 1,643SR Eq. 7
t

Where t is the thickness of the base layer in mm.

Table 4: Offsets from the basic design models to be used for the
design of pavements for different road categories
Road Approximate Offset (to be subtracted from the
category design reliability constants in Equations 6 and 7)
A 95 % -0,29
B 90 % -0,23
C 80 % -0,16
D 50 % -0,03
 13

Figure 5 shows contour plots of the general design model from Equation 4 for relative density
values of 84 and 88 per cent and a saturation level of 45 per cent. The effect of an increase in
density on the bearing capacity of the waterbound macadam layer is illustrated by comparing the
bearing capacity results at a stress ratio value of 50 per cent for the two cases.

100 100
Plastic Plastic
strain (%) strain (%)
Bearing Capacity (Repetitions)

Bearing Capacity (Repetitions)

80 80

60 13 60 13
Millions

Millions
16 16
40 20 40 20

20 20

0 0
0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1
Stress Ratio (%) Stress Ratio (%)

(a) 84 % relative density, 45 % saturation (b) 88 % relative density, 45 % saturation

Figure 5: The effect of density on the bearing capacity of waterbound macadam

The bearing capacity for 13 per cent plastic strain which is equivalent to 20 mm permanent
deformation of a 150 mm layer doubled from about 20 million standard axle loads for the 84 %
relative density case to 40 million standard axle loads for the 88 % relative density case. This
comparison is done at the same stress ratio for both cases. The shear strength of the material will,
however, increase with an increase in density, resulting in a lower stress ratio for the high density
case which will lead to an even larger increase in bearing capacity.

CONCLUSIONS AND RECOMMENDATIONS

The design model for waterbound macadam material presented in this paper is based on a
combination of Heavy Vehicle Simulator data and laboratory test data. The laboratory data
supplemented the HVS test data by widening the range of the variables over which the data was
collected. The impact of this on the design model is that the effect of density and saturation could
be included in the design model for both the effective stiffness and permanent deformation
response of the waterbound macadam material.

The effective stiffness results for the waterbound macadam from both HVS back-calculation and
laboratory test results indicate that the stiffness of waterbound macadam is higher than typical
values expected for a crushed stone material.

The friction angle of waterbound macadam is not higher than the friction angle of crushed stone
as was expected previously. The permanent deformation response of unbound aggregate is,
however, not uniquely determined by the static shear strength parameters of the material although
the permanent deformation of the material will be minimized by maximizing the static shear
strength parameters. By achieving maximum practical density and the lowest degree of saturation
possible under field conditions, the permanent deformation of waterbound macadam will be the
least as it will be for other unbound aggregates.
 14

The rational design model for waterbound macadam presented in this paper was incorporated in
the SAMDM for the design of different road categories at different design reliability levels. A
design catalogue based on this design model was incorporated into a guideline document on the
selection, design and construction of waterbound macadam base layers.

REFERENCES

1. Theyse H L and Muthen M. Pavement Analysis and Design (PADS) Software based on
the South African Mechanistic-empirical Design Method. Paper accepted for publication
at the Annual Transportation Convention, Pretoria. 2000.

2. Committee of State Road Authorities (CSRA), 1985. TRH14 Guidelines for road
construction materials. Department of Transport. Pretoria. 1985.

3. Theyse H L. The Evaluation of Pavements with Labour-intensively Constructed Base

Layers, a Case Study on Road 2388. Proceedings of the 7th Conference on Asphalt
Pavements for Southern Africa, Victoria Falls, Zimbabwe. 1999.

4. Theyse H L. First level report for HVS testing on the N1-28 near Louis Trichardt.
Transportek, Transportek, CSIR, Pretoria. 1999. (Report no. CR-99/065).

5. Wright B and Hess R. Evaluation of a freeway pavement incorporating a waterbound

macadam base. Division of Roads and Transport Technology, Pretoria. 1988.

6. Horak E. Watergebindemacadamplaveisels. M Eng Thesis, University of Pretoria. 1983.

7. Roux P L and Otte J R. Coarse aggregate bases. Transportek, CSIR, Pretoria. 1983.
(RDAC report RR 88/027)

8. Horak E. The recompaction of the 250 mm thick experimental waterbound macadam

section at Marianhill, Pinetown. National Institute for Transport and Road Research,
Pretoria. 1982. (Technical Note: TP/72/82)

9. McCall J, Roux P and Currie R. Waterbound macadam in single carriageway highway

construction in Natal. Proceedings of the Annual Transportation Convention, Pretoria.
1990.

10. Hefer A W. Towards design guidelines for macadam pavements. M Eng Thesis,
University of Pretoria. 1997.

11. Theyse H L. Laboratory design models for materials suited to labour-intensive

construction, Volume I: Report. Transportek, CSIR, Pretoria. 1999. (Confidential
contract report CR-99/038)

12. Committee of Land Transport Officials (COLTO). Draft TRH4 1996: Structural design
of flexible pavements for inter-urban and rural roads. Department of Transport, Pretoria.
1996.