Vibratory Compaction
of Base Course Aggregates
Pritesh Karan, Tam Larkin, and Douglas James Wilson
The focus of this paper is on laboratory compaction of unbound granu-
lar base course material for road pavement construction. The primary
variables evaluated in this research were particle size distribution and
dry density. This research addresses the applicability of current labora-
tory compaction techniques and outlines the changes needed to mini-
mize the gap between laboratory and field compaction. The target density
achieved in laboratory studies with the New Zealand standard test meth-
odology, which is then used to target field compaction levels, is often
too high to realistically achieve in the field. It was found that there is a
ratio of maximum aggregate average least dimension (ALD) to speci-
men dia­meter and specimen height that must be adhered to when using
laboratory testing as a means to evaluate, design, or construct pavement
base course layers. In addition, the power input of laboratory vibratory
hammers used for compaction must be controlled to minimize aggregate
degradation, especially for test molds that do not satisfy the minimum
diameter to ALD ratio. The ultimate goal of laboratory compaction
should be to accurately simulate field compaction and provide realistic
targets that can be achieved in the field. If unrealistic field compaction
density targets are specified, then overcompaction can occur and cause
aggregate breakdown. The research concluded that changes are required
for test standards to better reflect field conditions, such as less lateral
constraint, and better reflect achievable target field densities.
To justify expenditure on the construction or rehabilitation of a pave-
ment, the predicted performance over the pavement life and its asso-
ciated returns need to be known. The California bearing ratio (CBR)
and the repeated load triaxial (RLT) tests are two common methods
that are used in the laboratory to predict granular material behavior
in the field. The CBR test is an index test for stiffness and fails to
predict the effects of cyclic loading. The RLT test involves cyclic
loading and is a more appropriate test to predict the rutting and resil-
ient response of the pavement. For the RLT test to reasonably predict
in situ performance of a particular layer, the test conditions need to
accurately simulate field conditions. The variables that need to be
controlled include saturation levels, aggregate type and size distri-
bution, fines content, underlying material, confining and deviatoric
stress, cyclic testing rate, number of cycles, and method and degree
of compaction achieved (1).
To have confidence in the application of RLT test results, the
compaction level achieved both in the laboratory and field need to
be realistic and similar. The 2005 New Zealand Specification for
Department of Civil and Environmental Engineering, University of Auckland, Private
Bag 92019, Auckland 1142, New Zealand. Corresponding author: P. Karan,
pkar020@aucklanduni.ac.nz.
Transportation Research Record: Journal of the Transportation Research Board,
No. 2655, 2017, pp. 45–53.
http://dx.doi.org/10.3141/2655-07
45
Construction of Unbound Granular Pavement Layers [Transit New
Zealand (TNZ) B/02] states that the mean density shall not be less
than 98% and the minimum value shall not be less than 95% of the
laboratory density of the material. Often when field density targets
are set, little consideration is given to the degree of difficulty in
achieving a target density without causing aggregate breakdown.
The motivation for this research was the realization that more atten-
tion needs to be paid to laboratory base course compaction. Of par-
ticular interest was New Zealand Standard (NZS) 4402 Test 4.1.3
(1986) and its applicability to current pavement compaction tech-
niques and whether it produces realistic density expectations. This
realization came about after considerable discussion with industry
engineers and technicians in New Zealand and overseas whose
daily activities relate to the field and laboratory compaction of base
course aggregates.
After these discussions and a review of NZS 4402 Test 4.1.3,
the objective of this research was formulated. The objective is to
determine the influence of compaction duration, test mold size, and
energy exerted on particle size distribution (PSD) and dry density
during laboratory vibratory compaction of unbound base course
aggregate.
Vibratory Compaction
The popularity of laboratory vibratory compaction is mainly because
of its ease of use and efficiency in effectively compacting specimens
for a relatively low cost (2– 4). In addition, because vibratory com-
paction is also the primary in-field method of choice for compacting
granular base course materials, it allows an appropriate comparison
of the results of the two processes in two different environments.
This approach provides for field validation of laboratory compaction,
if the laboratory compaction is representative of field compaction.
Overcompaction
However, there can be inherent but unintended flaws within the test
methodology that can cause specimens to be overcompacted in the
laboratory. Overcompaction can be characterized by the resulting
density achieved in the laboratory being higher than what is fea-
sible in the field. The discrepancies between densities from field
and laboratory can be attributed to the test conditions and equip-
ment used. In the laboratory, compaction is achieved in a rigid mold
under K0 conditions, whereas in the field there is often limited lat-
eral restraint. Trying to target an unrealistically high density in the
field can lead to overcompaction and result in material breakdown.
Material breakdown may cause dedensification and result in the loss
of base course structure and premature pavement failure.
46
Methods and Materials
Vibratory compaction is predominantly used in the compaction of
near cohesionless base course material in the field. The material used
was M/4 graywacke aggregate, which is classified as a premium base
course aggregate in New Zealand for state highway road construc-
tion. Graywacke is the most commonly used rock for aggregate in
New Zealand and is a sandstone that consists primarily of quartz
dominating over feldspar. There are no indications of zeolites in
the aggregates. The smectite–chlorite ratio in the rock aggregate is
approximately 30:70, which is considered a very minor inclusion of
smectite (5).
Three compaction molds of varying geometry were used. The
small mold had a diameter of 150 mm and a height of 125 mm.
The medium mold used had a diameter of 243 mm and a height of
175 mm. The large mold had a diameter of 250 mm and a height
of 625 mm. All diameters are the internal diameter and have a toler-
ance of ±2 mm. Three hammers were used to compact the speci-
mens. The Kango 900kV and Kango 950K have input power ratings
of 1,050 W, whereas the Bosch GSH 16-30 has a much higher input
power rating of 1,750 W (Table 1).
For this research, the procedures for vibratory compaction and
determination of optimum moisture content (OMC) were adopted
from the New Zealand vibratory hammer compaction standard
(NZS 4402 Test 4.1.3). Water content was determined in accordance
with NZS 4402 Test 2.1 directly on the aggregate from the quarry.
Additional water was used to reach OMC for laboratory tests. The
OMC for M/4 material lies between 4% and 5%, with little variation
in dry density between these values (6).
Once the sample was at OMC, specimens were compacted in
layers. The number of layers depended on the height of the speci-
men mold. In general, the layer height is at least twice the maximum
particle average least dimension (ALD) (7, 8). This approach allows
sufficient space for particle reorientation during compaction for all
particles. Each layer was scoured to a depth of half the maximum
aggregate ALD after compaction to ensure interlock with the layer
to be compacted above.
Three repetitions of each test were carried out to assess the repeat-
ability and variation of results. If the three replicate results were
not within 10% of each other, the tests were repeated until at least
three tests yielded PSD and density results within tolerance. More
than 75 compaction tests were conducted. Five tests were repeated
because the results of PSD or density were not within tolerable lim-
its. The variation in water content was kept within the OMC tolerance
of 4% to 5% for most of the specimens. For one set of three replicates,
the average water content was 5.1%.
TABLE 1   Vibratory Hammer Specifications
Input Energy Frequency
Hammer (W) (Hz)
Kango 950K 1,050 36.7
Kango 900kV 1,050 33.3
Bosch GSH 16-30 1,750 21.7
Transportation Research Record 2655
Vibratory Hammers
The three hammers used for this research were chosen because of
their commonality in the pavement engineering industry in New
Zealand. The specifications of the vibratory hammers are presented
in Table 1. NZS 4402 Test 4.1.3 does not specify a definitive figure
for hammer energy. The standard specifies that hammers with a
range of power consumption of 600 to 1,200 W can be used. The
Bosch hammer has a high input energy rating of 1,700 W, but the
power consumption of the hammer is not clear. It is assumed that
the term “power consumption” refers to the input power rating
of the vibratory hammer because the output power rating is more
difficult to ascertain and can deteriorate with age.
Table 1 also presents the energy per second applied by the hammer
assembly to the top of the specimen. Each specimen mold and
hammer combination has a different load shaft and compaction setup,
which results in a different combined weight for each setup. The
differences between the Kango 900kV and Kango 950K setups are
minor. The Bosch hammer setup is significantly heavier but has a
lower frequency and results in lower energy output.
Compaction Duration
The results discussed in this section concern two specimens that were
compacted using vibratory compaction in a mold of the same geom-
etry for two durations. The compaction durations were 60 s and 180 s.
NZS 4402 Test 4.1.3 states that 180 s of compaction is required. A
comparison was made with the British standard for vibratory com-
paction (BS 1377-4:1990), which specifies 60 s of compaction. The
three hammers were each used in the three molds. Fifty-seven valid
compaction tests were analyzed to determine the correlation between
compaction duration and PSD and dry density for each hammer. For
the tests involving the large mold (250-mm diameter × 625-mm
height), compaction durations of 40, 60, and 180 s were used. There
was very little variation in PSD between the three durations and
slight variations in density (approximately 1.5%). The density and
postcompaction PSD were recorded for each test.
Pumping of Fines
When specimens were compacted for a duration of 180 s, pump-
ing of fines was exhibited in specimens compacted at OMC for all
hammers. To calculate the dry density to an accuracy that is satisfac-
tory, all the fines that have pumped up and out of the specimen and
splattered around the mold need to be collected. These fines are then
Energy Applied to Specimen by
Specimen Size (J/s)
Weight Small Medium Large
(kg) (two layers) (three layers) (eight layers)
11.4 74.68 87.35 87.35
10 65.43 77.42 77.42
16.5 50.78 60.11 61.19
Karan, Larkin, and Wilson
added to the specimen, which is to be dried in the oven before PSD
and density analyses are carried out.
Pumping of fines was more evident in the small specimens
(150 mm × 125 mm). Very little pumping was witnessed in the
medium specimens (243 mm × 175 mm). There was no pumping
in the large specimens (250 mm × 625 mm).
Examples of Variations in Density
The upper and lower limits outlined in Figure 1 are the respective
upper and lower aggregate size limits for M/4 base course (TNZ M/4).
Figure 1 shows the PSD results of the two durations of compaction
for a 950K hammer assembly in the small mold size (150 mm ×
125 mm). There is no significant evidence that compaction dura-
tion has an effect on PSD when a hammer with a lower power rating
is used.
Figure 2a shows PSD results using the heaviest hammer assem-
bly in the smallest mold. Comparison with Figure 1 shows a signifi-
cant difference in results between the low-energy and high-energy
hammer assemblies. Use of the Bosch hammer resulted in aggregate
degradation in particles retained postcompaction on the 2.36-mm
sieve and above. This phenomenon was evident only in the 150-mm
× 125-mm specimens. There was no evidence of aggregate degrada-
tion for any of the hammer assemblies with the medium and large
molds. These results suggest that aggregate degradation is at its
peak when hammers with high energy ratings are used to compact
base course specimens for a prolonged duration in a small mold
size. The extent of degradation is not severe enough to cause the
PSD of the base course to exceed the upper M/4 limit as specified in
TNZ B/02, but is on the verge of doing so.
Figure 2b shows the standard deviation for each sieve size for the
two durations employed when three compaction tests were carried
100.0
90.0
80.0
70.0
Cumulative % Passing
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0.001 0.01
60-s average 180-s average
FIGURE 1   Influence of compaction duration (specimen size: 150 mm ë 125 mm; hammer: Kango 950K).
47
out for each duration using the small mold (150 mm × 125 mm). The
same test sequence was carried out with the medium mold (243 mm
× 175 mm), which showed smaller variation in test results.
The variation in PSD test results was further reduced when the
larger mold (250 mm × 625 mm) was used. The small mold showed
the greatest variation, which was further increased when the Bosch
hammer assembly was used. Use of the Bosch hammer generally
resulted in the variation in PSD being larger when the specimen
was compacted for 180 s compared with 60 s. The small specimen
size resulted in the compaction being erratic, resulting in unpredict-
able and varied results. The presence of large aggregates in a small
specimen is very influential in determining the fabric and structure
of the specimen postcompaction.
Influence of Mold Size
It is clear from Figure 2a that compaction duration increases the
effects of aggregate degradation on base course aggregate specimens
when the Bosch hammer assembly and the small mold are used. In
this section, a comparison is made between test results from the large
and medium-size molds with the Kango 950K and Bosch hammer
assemblies and 180 s of compaction for each lift. The results of
54 tests were analyzed to determine if the trend shown in Figure 2a
was present for the medium and large mold for the two hammers.
Figure 3 shows how the PSD changes with mold size when the
hammer and compaction duration are kept constant. Compaction
duration of 180 s was chosen because it leads to the greatest like-
lihood of aggregate degradation. The results show that when a
hammer with a power rating that is in accordance with the limits
specified in NZS 4402 Test 4.1.3 is used, there is very little aggre-
gate breakdown, regardless of the mold dimensions. The specimens
compacted in the small mold with the Kango hammer assemblies
0.1 1 10 100
Sieve Size (mm)
Upper limit Lower limit Precompaction
48 Transportation Research Record 2655

100.0

90.0

80.0

70.0
Cumulative % Passing

60.0

50.0

40.0

30.0

20.0

10.0

0.0
0.001 0.01 0.1 1 10 100
Sieve Size (mm)
60-s average 180-s average Upper limit Lower limit Precompaction
(a)

5
Standard Deviation (%)

0
37.5 19 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 0.001
Sieve Size (mm)
60 s 180 s
(b)

FIGURE 2   Influence of compaction duration: (a) PSD and (b) standard deviation (specimen size: 150 mm ë 125 mm;
hammer: Bosch GSH 16-30).

did show the beginnings of some particle breakdown. This result
is again attributed to the small size of the mold. The results of the
tests conducted with the Kango 900kV hammer are similar to those
shown in Figure 3. There is very little variation in PSD curves for
60-s compaction tests with either of the Kango hammers.
In contrast to Figure 3, Figure 4a shows significant aggregate
degradation caused by compacting the base course specimen with
the Bosch hammer assembly (with the higher energy rating) in
the smaller mold. This result may be because of the presence of
larger 37.5-mm ALD aggregates in the small 150-mm diameter
mold. This result did not occur for the medium or large molds; Fig-
ure 4a shows no changes in PSD after compaction. These results
agree with published work by Araya et al. (2) and CBR and RLT
standards that specify that only aggregates smaller than 19-mm
ALD can be used in a 150-mm diameter mold [AGPT-T053-07
(Australia), BS 1377-4:1990 (Britain), NZS 4402 Test 6.1.1
(New Zealand), and ASTM D1883-07e2 (United States)]. The
Australian standard for RLT testing (Method 6.4.1.) specifies a
specimen diameter that is six times the maximum aggregate ALD,
similar to that found in work by Toan (7) and Dawson et al. (9). The
literature is consistent in specifying that a ratio of aggregate ALD
to mold diameter to mold height of 1:6:2 should be followed to
avoid aggregate breakdown and inconsistent results (1). The results
shown in Figure 4a reinforce this stance.
Karan, Larkin, and Wilson
100.0
90.0
80.0
70.0
Cumulative % Passing
60.0
50.0
40.0
30.0
20.0
10.0
0.0
0.001 0.01 0.1
Sieve Size (mm)
150 × 125
Upper limit
FIGURE 3   Influence of mold size on PSD (compaction duration: 180 s; hammer: Kango 950K).
The breakdown of aggregates can be related to compacting over-
sized aggregate particles in a small mold with a hammer that has a
high power rating. Because of the confined area of the small mold,
the larger aggregates do not readily reorient during compaction and
tend to interlock with other aggregates and form a matrix with high
residual stresses. This process increases the likelihood of particle
fracture at points of contact. The amount of aggregate degradation
depends on the crushing resistance of the rock. The crushing resis-
tance test result for the aggregates used in this research is 0.8% fines
produced at a load of 130 kN. The M/4 specifications indicate that
the maximum amount of fines passing the 2.36-mm sieve during the
crushing resistance test must be less than 10% [BS 1377-4:1990 and
NZS 4407 Test 3.6 (1991)]. Considering the high crushing resis-
tance of the material, particle breakdown should theoretically be
very difficult to achieve. This study established that particle break-
down is possible through heavy compaction undertaken in a mold
that does not comply with the industry-established size criteria.
Figure 4b shows that the variation between PSD tests with the
same testing conditions is greatest in the small mold, less in the
medium mold, and least in the large mold. A larger specimen size,
as is the case for the large mold, reduces the likelihood of compac-
tion variation (largely as a result of particle constriction), so that
intentionally induced test variables can be studied with confidence.
Consistency in specimen formation is necessary to permit the rig-
orous identification of the effects of intended variation in testing
conditions.
Energy
It is evident that vibratory hammer energy ratings play a significant
role in the nature of the specimens produced. The triplet test speci-
mens compacted for 60 s had lower variation in degradation and
reproducibility than the specimens compacted for 180 s. In addi-
tion, the test specimens varied depending on the hammer assem-
49
1 10 100
243 × 175 250 × 625
Lower limit Precompaction
bly used in the compaction process. To understand the relationship
between hammer power rating and postcompacted PSD, the results
of 54 compaction tests were evaluated.
Figure 5a shows that there was very little difference between
postcompacted PSDs for the three hammers when the large mold
was used. The primary factor for the consistent set of results was
the large dimension of specimens. The larger specimen allowed
reorientation of the aggregate particles, perhaps similar to particle
reorientation in the field. In the field, a typical base course layer
thickness in New Zealand of approximately 150 mm would allow
particle reorientation to some extent. It is, however, the lateral con-
finement conditions that most heavily influence particle reorienta-
tion in the field. In the laboratory, lateral constraint is significantly
enhanced because the mold provides a rigid boundary.
Figure 5b shows the variation in PSDs with the use of the three
hammers. Although there is very little variation between PSDs, the
general trend is that the Bosch hammer provides greater variation
than the Kango hammers. Because these tests were done in the large
mold, it can be concluded with confidence that on average the Bosch
hammer provides the greatest variation in PSD curves when other
variables are kept constant. This result is understandable because
the Bosch hammer has the highest energy input and hence provides
the greatest opportunity for aggregate reorientation and degradation.
In contrast to Figure 5a, Figure 6 shows that when the small mold
is used, the effects of hammer energy on PSD are amplified, lead-
ing to a greater variation in results across the hammers. Because of
the small specimen size, aggregate degradation is more evident in
Figure 6 (i.e., the percentage passing a sieve increases with ham-
mer energy for particles larger than approximately 2.36 mm). The
greatest amount of postcompacted PSD change results from use of
the Bosch hammer. The results of this study agree with work done
by Shahin (10), which showed that some of the variability attrib-
uted to vibratory compaction is related to the energy output of the
vibratory hammer. Hammers with greater energy outputs have a
higher likelihood of causing aggregate degradation (10).
50 Transportation Research Record 2655

100.0

90.0

80.0

70.0
Cumulative % Passing

60.0

50.0

40.0

30.0

20.0

10.0

0.0
0.001 0.01 0.1 1 10 100
Sieve Size (mm)
150 × 125 243 × 175 250 × 625
Upper limit Lower limit Precompaction
(a)

5
Standard Deviation (%)

0
37.5 19 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 0.001
Sieve Size (mm)
150 × 125 243 × 175 250 × 625
(b)

FIGURE 4   Influence of specimen size: (a) PSD and (b) standard deviation (duration: 180 s; hammer: Bosch GSH 16-30).

Density of the laboratory densities is questionable. The use of the smaller

The results of the PSD analyses give an idea of the amount of deg-
radation that is caused by various factors during vibratory com-
paction. To understand the effectiveness of vibratory compaction,
the density of specimens must be evaluated. The density of a base
course layer significantly affects the outcome of traffic loading on
a pavement. Poor compaction, and in some cases overcompaction,
can result in rutting and eventually lead to water ingress in the pave-
ment layer, which may result in early failure of the pavement (11).
Typically, laboratory densities are used as a target for field densi-
ties. As discussed in previous sections, the method of determination
mold, which is the mold specified by NZS 4402 Test 4.1.3 for deter-
mining laboratory base course densities, can lead to breakdown of
aggregate particles and contribute to inconsistent results. The prob-
lem is amplified when particles with larger than appropriate ALDs
are used.
Table 2 shows the density relationships between hammers and
molds. At least three tests were done for each test condition to pro-
vide statistically robust results. The variation of water content and
density for the three tests is also presented in Table 2. In general,
Table 2 shows that the small mold results in the highest densities,
on average 200 to 300 kg/m3 greater than the densities obtained
Karan, Larkin, and Wilson 51

100.0

90.0

80.0

70.0
Cumulative % Passing

60.0

50.0

40.0

30.0

20.0

10.0

0.0
0.001 0.01 0.1 1 10 100
Sieve Size (mm)
Kango 900kV Kango 950K Bosch GSH 16-30
Upper limit Lower limit Precompaction
(a)

2
Standard Deviation (%)

0
37.5 19 9.5 4.75 2.36 1.18 0.6 0.3 0.15 0.075 0.001
Sieve Size (mm)
Kango 900kV Kango 950K Bosch GSH 16-30
(b)

FIGURE 5   Influence of hammer rating: (a) PSD and (b) standard deviation (specimen size: 250 mm ë 625 mm;
duration: 180 s).

with the medium and large molds. The densities of specimens of all
three molds are typically highest when the highest-energy hammer
is used for compaction. The variation in specimen characteristics is
largest for the small mold because of the unpredictable nature of the
structure of the particle volume, as a result of excessive confinement
during compaction. In contrast, the medium and large molds generally
provided a consistent set of results.
The New Zealand base course field construction specification
TNZ B/02 gives as a target a mean value of at least 98% of the
laboratory determined density. For example, if a laboratory deter-
mined density of 2,400 kg/m3 is used, which is typical for the tests
conducted in the small mold, then the field density must be on aver-
age 2,350 kg/m3. This target is unrealistic if the densities that are
achieved by the use of the medium and large molds are a reliable
indicator. The typical density for specimens from the medium and
large molds is approximately 2,200 kg/m3, with the exception of the
specimens compacted with the Bosch hammer, for which slightly
higher densities are found for all three specimen sizes. This density
52 Transportation Research Record 2655

100.0

90.0

80.0

70.0
Cumulative % Passing

60.0

50.0

40.0

30.0

20.0

10.0

0.0
0.001 0.01 0.1 1 10 100
Sieve Size (mm)
Kango 900kV Kango 950K Bosch GSH 16-30
Upper limit Lower limit Precompaction

FIGURE 6   Influence of hammer energy on PSD (specimen size: 150 mm ë 125 mm; duration: 180 s).

is consistent for both 60 and 180 s of compaction in the medium and
large specimens, indicating that no matter how long the compaction
duration, the density reaches a limit at which it no longer increases.
However, particle breakdown does not occur when the medium and
large molds are used because the specimen size is greater and the
effects of the solid base are minimized. The same is not true for the
smaller mold; the longer the specimen is compacted, the greater the
density becomes because of particle breakdown.
Because of the confining nature of small molds, the results of
PSD and density analysis show that the small mold is not suitable
for use for predicting target densities for base course layers in the
field (see Table 2). Use of the small mold and adherence to good

TABLE 2   Average Dry Density of Compacted M/4 AP40 Base Course, Triplets

Type of Average Water Average Dry Variation in Water Variation in

Hammer Mold Size (mm) Duration (s) Content (%) Density (kg/m3) Content (%) Density (kg/m3)

Vibratory 250 × 625 40 4.4 2,054 0.2 35

(1,050 W) 60 4.4 2,060 1.2 20
Kango 900kV 180 4.9 2,129 0.3 37
243 × 175 60 4.1 2,129 0.2 56
180 4.7 2,141 0.4 53
152 × 125 60 4.5 2,292 0.4 56
180 4.5 2,335 0.3 177
Vibratory 250 × 625 60 5.0 2,126 0.3 28
(1,050 W) 180 4.8 2,153 0.1 25
Kango 950K 243 × 175 60 4.7 2,151 0.2 19
180 4.7 2,163 0.1 44
152 × 125 60 5.1 2,409 0.6 102
180 4.7 2,386 0.2 48
Vibratory 250 × 625 60 4.7 2,231 0.4 60
(1,750 W) 180 4.8 2,303 0.4 46
Bosch 243 × 175 60 4.5 2,126 0.1 91
180 4.6 2,204 0.4 11
152 × 125 60 4.6 2,423 0.2 54
180 4.5 2,506 0.2 29

Note: 1 in. = 25.4 mm; 1 ft = 0.305 m; and 1 lb = 0.454 kg.

Karan, Larkin, and Wilson
TABLE 3   Dry Density Results for M/4 AP20, Triplets: Scalped Base Course
Water Content
Type of Hammer Mold Size (mm) Duration (s) (%)
Vibratory 152 × 125 60 4.6
(Kango 900kV) 180 4.8
Vibratory 152 × 125 60 4.6
(Kango 950K) 180 4.5
Vibratory 152 × 125 60 4.7
(Bosch GSH 16-30) 180 4.2
compaction practice (mold diameter greater than or equal to six
time the maximum particle ALD) would necessitate scalping of
any aggregate greater than 19 mm (NZS 4402 Test 4.1.3, BS 1377-4:
1990, AGPT-T053-07, and ASTM D1883-07e2). However, the
aggregate would then not be representative of that used in the field.
Further testing was done to establish the densities obtained with
the three hammers in the small mold after scalping aggregates that
were greater than 19 mm in accordance with various international
standards (BS 1377-4:1990, NZS 4402 Test 6.1.1, AGPT-T053-07,
and ASTM D1883-07e2). The results of these tests are displayed in
Table 3. In general, the average density is slightly reduced from that
of the unscalped tests. The variation is also significantly reduced for
the scalped test, barring one set of tests that yielded unpredictable
results. The result may be because a different operator conducted
the scalped tests. As explained by Shahin, operator technique can
significantly contribute to the variability of the test results (10). In
the initial period when an operator is learning consistent techniques,
it is often found that there is 20% more variability in dry density
than occurs at later stages when the operator is more experienced.
This effect was noted in this work. Similar variability was exhibited
in the medium and large molds when the unscalped material was
tested. This result emphasizes that the recommended values of the
ratio of maximum aggregate ALD to specimen diameter to speci-
men height as published by several researchers are reliable and must
be complied with to avoid unnecessary variability (1, 2, 7).
Laboratory test methods need to reasonably simulate field condi-
tions such that laboratory values are reliably transferrable to the
field. For this reason, field testing should be done to determine the
maximum density achievable in the field before aggregate degrada-
tion starts to occur. PSDs must be performed to establish correlation
between field and laboratory densities.
Conclusions
The following conclusions can be drawn from the research:
• The use of vibratory compactors with high energy results in
densities that are much greater than what may be achieved in the
field.
• The detrimental effects of hammer energy are amplified when
recommended ratios of aggregate to specimen size are ignored.
• Aggregates with lower crushing resistance would be expected
to demonstrate greater particle breakdown and therefore the varia-
53
Average Dry Variation in Water Variation in
Density (kg/m3) Content (%) Density (kg/m3)
2,274 0.3 56
2,314 0.4 148
2,266 0.3 26
2,325 0.3 34
2,398 0.2 14
2,435 0.4 20
tion in results caused by compactor energy and mold size would be
amplified.
• The influence of hammer energy is negligible when an appro-
priate mold size, according to specimen ALD, is adopted.
• It is recommended in the international literature, and also shown
in this research, that a minimum ratio of aggregate ALD to specimen
diameter to specimen height ratio of 1:6:2 results in more consistent
laboratory densities and realistic field target densities.
References
  1. Karan, P., W. Li, D. J. Wilson, and T. J. Larkin. Understanding Repeated
Load Triaxial (RLT) Tests. University of Auckland, New Zealand,
2016.
  2. Araya, A., A. Molenaar, and L. Houben. Characterization of Unbound
Granular Materials Using Repeated Load CBR and Triaxial Testing.
In Geotechnical Special Publication 203: Paving Materials and Pave-
ment Analysis—Proceedings of Sessions of GeoShanghai 2010, ASCE,
Reston, Va., 2010.
  3. Hoff, I., L. J. Baklokk, and J. Aurstad. Influence of Laboratory Compac-
tion Method on Unbound Granular Materials. In Pavements Unbound:
Proceedings of the 6th International Symposium on Pavements Unbound
(A. Dawson, ed.), Taylor & Francis Group, London, 2004.
 4. Chilukwa, N. N. Vibratory Hammer Compaction of Granular Materials.
MS thesis. Stellenbosch University, South Africa, 2013.
 5. Fisher, G. Petrological Assessment of the Source Rock at Drury Quarry.
Grant Fisher Industrial Geology, New Zealand, 2008.
  6. Karan, P., D. J. Wilson, and T. J. Larkin. Methods of Compaction of
Basecourse Aggregate for Large Scale Repeated Load Triaxial (RLT)
Testing. Presented at IPENZ Transportation Group Conference, Harding
Consultants, Wellington, New Zealand, 2014.
 7. Toan, D. V. Effects of Basecourse Saturation on Flexible Pavement
Performance. PhD dissertation. University of Auckland, New Zealand,
1975.
  8. Thom, N. Principles of Pavement Engineering. Thomas Telford Ltd.,
London; University of Nottingham, 2008. https://doi.org/10.1680
/pope.34808.
  9. Dawson, A. R., P. J. Galjaard, and J. L. Paute. Recommendations for
Repeated Load Triaxial Test Equipment and Procedure for Unbound
Granular Materials. In Flexible Pavements: Proceedings of the European
Symposium Euroflex 1993, A. A. Balkema, Rotterdam, Netherlands,
1996.
10. Shahin, A. W. Investigation of the Variability in the Results of the NZ
Vibrating Hammer Compaction Test. MS thesis. University of Auckland,
New Zealand, 2011.
11. Papagiannakis, A. T., and E. A. Masad. Pavement Design and Materials.
John Wiley & Sons, New York, 2008.
The Standing Committee on Aggregates peer-reviewed this paper.