Table 7. Mean and variance of roughness and
gold characterized the angularity of all other ag-
roundness values from image analysis.
gregates as the difference between the percentage
of voids and 33%. Then, AN was found to range
Roughness
Roundness
between 0 and 12:
Aggregate type
Mean
Variance
Mean Variance
Crushed gravel
1.0306
0.000211
78.537
46.668
AN = percentage of voids - 33
Crushed stone
1.0500
0.000198
73.285
37.846
Pike crushed stone
1.0471
0.000115
72.365
66.589
100M
AN = 67
cGa
a difference between the crushed gravel and
crushed stone. There was no difference in the
where M = mass of standard volume of aggre-
mean roughness values between the crushed
gate (g)
stones. From these very limited tests on coarse
c = mass of water required to fill the
aggregates, and without additional strength/
same volume (g)
modulus related testing, it is not certain whether
Ga = oven-dried specific gravity of the
image analysis can clearly distinguish the shape,
aggregate.
angularity, and roughness of different aggre-
gates.
Several limitations of the angularity number
were pointed out by Lees (1964). Primarily, the
Determination of aggregate characteristics
AN was developed based on results from only six
from indirect methods
samples of coarse aggregate. Furthermore, the
It is clear from the above discussions that the
angularity as described by Shergold, "angular to
determination of aggregate shape, angularity,
rounded," was based on a consensus reached
and surface texture is a fairly lengthy and labori-
from visual examination of the test aggregate by
ous task. An alternate approach taken by engi-
25 observers. Another quirk with the AN was that
neers is to infer these characteristics from the
it was not to be applicable to all shapes (such as
mass properties of the aggregates. Several indices
regular geometric objects [spheres and cubes]),
for coarse aggregates, such as angularity number,
and the AN for perfect spheres was found to be
particle index, rugosity, uncompacted void, and
higher than for perfect cubes, which appears to be
time index have been identified in the literature.
contradictory. Furthermore, the heavy compac-
tion required may break the aggregates, causing
Angularity number
artificial changes in angularity.
The angularity number (AN) developed by
Gupta (1985) refined the AN model to account
Shergold (1953) is recommended by British Stan-
for the shape of the aggregate. The AN was calcu-
dards (BS 812 1975) for indexing the angularity of
lated in the same manner as before, with the ex-
natural and crushed aggregates used in concrete.
ception that Gupta defined the percentage of
Shergold found that when the aggregates were
voids in an aggregate mass as a function of the
compacted in a prescribed manner, the percent-
shape and the average size of the aggregate
age of voids in the aggregate mass decreased as
(mm). Based on test results of the three different
the aggregates became more rounded. He also
found that as the amount of round gravel in-
range), subangular crushed quartzite (6- to 50-
creased in a mixture of natural and crushed ag-
mm range), and rounded gravel (6- to 100-mm
gregates, the percentage of voids also decreased.
Based on his study of six aggregates, he found
and rounded gravel mixtures, he found that the
that the minimum percentage of voids in round-
percentage of voids (Fig. 12) can be expressed as
ed gravel was approximately 33%. The tests were
η = Cdn
conducted on 19-, 12.7-, 9.5-, 6.35-, and 4.76-mm
aggregates. The test procedure involved compac-
where η = percentage of voids
tion of individually sized aggregates in three lay-
C = shape factor
ers in a 2800-cm3 mold. Each layer was compact-
d = volume mean aggregate diameter
ed with a tamping rod that weighed between 900
(mm)
and 950 grams to 100 blows. The percentage of
n = exponent.
voids was calculated by using the net weight of
the aggregate in the mold.
The size (d) is determined by taking a known
Using round gravel as a reference point, Sher-
number of particles and soaking them in water
9