Record: Highway Research

HIGHWAY
RESEARCH RECORD
Number 307
Synthetic Aggregates
and
Granular Materials
5 Reports
Subject Areas
31 Bituminous Materials and Mixes
32 Cement and Concrete
35 Mineral Aggregates
HIGHWAY RESEARCH BOARD

DIVISION OF ENGINEERING NATIONAL RESEARCH COUNCIL
NATIONAL ACADEMY OF SCIENCES-NATIONAL ACADEMY OF ENGINEERING
WASHINGTON , D.C. 1970
Standard Book Number 309-01805-6
Price: $1.80
Available from
Highway Research Board

National Academy of Sciences
2101 Constitution Avenue
Washington, D.C. 20418
Department of Materials and Construction
R. L. Peyton, Chairman
State Highway Commission of Kansas, Topeka
R. E. Bollen and W. G. Gunderman

Highway Research Board Staff
GENERAL MATERIALS DIVISION

John L. Beaton, Chairman
California Division of Highways, Sacramento
COMMITTEE ON MINERAL AGGREGATES

(As of December 31, 1969)
F. E. Legg, Jr., Chairman
University of Michigan, Ann Arbor
William P. Chamberlin Donald R. Lamb R. D. Shumway

John C. Cook Thomas D. Larson Nor man G. Smith
J. T. Corkill W. B. Ledbetter Travis W. Smith
Karl H. Dunn D. W. Lewis G. W. Steele
R. D. Gaynor J. F. McLaughlin Richard D. Walker
Joseph E. Gray Kenneth A. Nelson E. A. Whitehurst
Hans I. Hansen Robert G. Pike Milton H. Wills, Jr.
Eugene Y. Huang F. A. Renninger D. O. Woolf
Foreword
Of all the materials used in the highway system, aggregate by
far represents the largest quantity hauled to the job. Hundreds
of millions of tons are utilized annually for all types of pur-
poses, including high-quality, skid-resistant aggregates for
surfaces, aggregates for portland cement concrete and asphaltic
concrete, and aggregates for bases and subbases. These needs
place tremendous demands on the nation's supply of this re-
source. Coupled with this demand is a strong need to better
characterize the aggregates so that they may be utilized more
effectively.
As a result of the tremendous use of aggregates and the
rapid urbanization of our population, natural aggregates are
being rapidly depleted in many areas of the country. Thus, a
demand is being created for suitable aggregate replacements,
and the materials engineer must look to new sources of ma -
terials to supply this demand. In this RECORD four reports
deal with such aggregate replacements, termed "synthetic"
aggregates. Malisch, Day, and Wiksondeal with domestic waste
glass used as an aggregate in bituminous concrete. Research
results indicate that bituminous concrete mixes made entirely
with domestic waste glass as the aggregate can be designed to
satisfy the Marshall design criteria. Another aggregate re-
placement described by Hendrickson and Lund is volcanic
cinders. The successful utilization of these cinders as a wear-
ing surface for unpaved forest access roads is fully documented.
A third synthetic aggregate discussed is sintered fly ash, and
Minnick's paper describes the properties and performance of
this material when made into a structural lightweight aggregate
for portland cement concrete. Information is also presented
on typical mix designs for use in structural concrete. A fourth
synthetic aggregate, described by Moore, is that produced in
a rotary kiln from raw clay. This aggregate is nonbloated and
similar in properties to brick particles. Research results,
coupled with field evaluation, indicate that this material per-
forms very well as an aggregate inflexible bases.
Kandhal and Lee discuss the improved characterization of
aggregate, in particular the bulk specific gravity. The standard
ASTM methods for determination of a saturated surface-dry
specific gravity for both coarse and fine aggregates are com -
pared with five other methods. Results indicate that the stan-
dard ASTM tests underestimate the bulk specific gravity and
may not be as reproducible as other methods suggested.
-W. B. Ledbetter
Contents
USE OF OOMESTIC WASTE GLASS
AS AGGREGATE IN BITUMINOUS CONCRETE
Ward R. Malisch, Delbert E. Day, and Bobby G. Wixson . . . . . . . . . . . . 1
Discussion: K. 0. Anderson . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Closure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
CONSTRUCTION SPECIFICATIONS FOR VOLCANIC

CINDERS USED AS ROAD-SURFACING AGGREGATE
Larry G. Hendrickson and John W. Lund . . . . . . . . . . . . . . . . . . . . . . . 11
LIGHTWEIGHT CONCRETE
AGGREGATE FROM SINTERED FLY ASH
L. John Minnick. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
FIRED-CLAY AGGREGATES
FOR USE IN FLEXIBLE BASES
William M. Moore . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
AN EVALUATION OF THE BULK

SPECIFIC GRAVITY FOR GRANULAR MATERIALS
Prithvi S. Kandhal and Dah-Yinn Lee. . . . . . . . . . . . . . . . . . . . . . . . . . 44
Use of Domestic Waste Glass as
Aggregate in Bituminous Concrete
WARD R. MALISCH, DELBERT E. DAY, and BOBBY G. WJXSON,
University of Missouri-Rolla
The increasing amounts of waste glass being generated in the United

States are necessitating the development of new methods for disposal
of this refuse component. The use of waste glass as an aggregate in
bituminous concrete was investigated in this study. The objectives
were to determine whether a mixture using glass aggregate could be
designed that would meet Marshall test design criteria, to investigate
the amount of degradation occurring in these mixtures, and to determine
the water-resistance of such mixtures.
Waste glass was crushed and screened into several size fractions,
which were then combined to give a maximum density gradation. Stan-
dard Marshall design methods were used to determine stability, flow,
and void parameters at several asphalt contents, and to determine an
optimum asphalt content satisfying Marshall design criteria. Extrac-
tion tests were conducted on some of the specimens tested, and mech-
anical analyses of the recovered aggregate were used to 'assess the
amount of degradation of the aggregate. Static stripping tests were
conducted on several combinations of glass aggregates and bituminous
materials.
Bituminous mixtures satisfying Marshall design criteria can be de-
signed using aggregates composed entirely of crushed glass. Some deg-
radation of the glass aggregate does occur under laboratory mixing,
compacting, and testing conditions, with greater degradation appearing
to be associated with gradations deviating from those giving maximum
density. Severe stripping occurs when the glass-aggregate bituminous
concrete with no antistripping agent is exposed to water. Glass aggre-
gates coated with asphalt cements containing antistripping agents or
coated with cationic emulsions do not strip in the static stripping test.
•SOLID WASTES generated in the United States each year total 3.5 billion tons accord-
ing to the U. S. Public Health Service. This waste includes about 30 billion glass bottles
and jars, with increasing amounts of this component expected as the use of "one-way"
bottles continues to increase at a rapid rate (1). Glass is a particularly troublesome
component of waste because it will not burn, rust, or decay . A study of the composi-
tion and characteristics of municipal inc"ine1·ator residues (2) indicates that glass con-
stituted the major fraction of incinerator residues and accounted for nearly half of the
residue by weight. Thus, even after incineration, there is still a considerable amount
of glass remaining to be disposed of.
Other means of disposal of glass include sanitary landfills and salvage and reclama-
tion operations. In some a reas, sanitary landfills are doubtful solutions to the problem
because of the lack of available land. As these landfills are moved further and further
from the urban centers, transportation costs mount and this method becomes less
Paper sponsored by Committee on Mineral Aggregates and presented at the 49th Annual Meeting.
1
2
appealing. The bulk of glass is also a problem because bottles take up an inordinate
amount of space in relation to the glass present, unless the bottles are thoroughly crushed.
An alternate method of disposing of waste glass is to reuse the material, perhaps in
a different form. Economic considerations have generally dictated whether or not sal-
vage and reclamation operations are feasible, and, in the past, processing and trans-
portation costs have usually ruled out this possibility. For instance, waste glass could
be returned to a glass manufacturer to be melted and reused, but transportation costs
and the costs of grading the glass into different categories by color, type, and so forth,
are generally prohibitive. Furthermore, particularly in large cities, public works of-
ficials are justifiably apprehensive of reliance on a disposal method that depends on
fluctuating and sometimes nonexistent markets for salvage by-products. If the markets
collapse, the city may find itself with quantities of refuse and no method of disposal ex-
cept on an emergency landfill distant from the city (3). Thus, in order to maximize
potential benefits from salvage and reclamation operations, it is necessary to minimize
processing and transportation costs while ensuring a steady market for the salvaged
refuse.
By using waste glass as aggregate in bituminous concrete mixtures for maintenance
of city streets, transportation costs would be minimized because the glass would be
used in the city where it was discarded. Separate collections of glass refuse, although
increasing collection costs, would make costly hand-sorting unnecessary, and all pro-
cessing (cleaning, crushing, and screening) could be mechanized. A steady market
would be available because of the continuous street maintenance programs of cities, and
the savings from purchasing less aggregate would partially offset the increased costs
associated with separate collections. A final benefit from this utilization of glass as
an aggregate may materialize later as present sources of conventional aggregates are
depleted a_nd new sources are sought.
OBJECTIVES
This paper describes work done in investigating the use of crushed waste glass as
an aggregate in bituminous mixtures. No previous investigations of this nature could
be found in the literature, and therefore the following initial objectives were defined:
1. Design a dense-graded bituminous mixture using glass aggregate that would meet
Marshall design criteria specified by The Asphalt Institute;
2. Determine the amount of degradation that occurs during laboratory mixing, com-
paction, and testing of the mixtures; and
3. Determine the water-resistance of the mixtures.
M.ATERLL\.LS
All aggregates used in this investigation were obtained by crushing waste glass con-
sisting primarily of one-way beer and soft-drink bottles. The bottles were first washed
in hot water to remove labels and any other foreign material. After drying, they were
passed through first a jaw crusher and then a roller mill for finer crushing, after which
the crushed material was dry-sieved into nine different size fractions ranging from ma-
terial passing a 1/2-in. sieve and retained on a %-in. sieve to material passing the No.
200 sieve. A washed sieve analysis was performed on samples from each of the size
fractions, the results of which were used in combining the fractions to obtain a desired
gradation.
The crushed material contained a relatively large amount of flat or elongated parti-
cles or both in nearly all fractions. Samples from several of the larger size fractions
were tested for percentage of flat and elongated particles using Corps of Engineers
Methods CRD-C 119-53 and CRD-C 120-55. Each size range sample was subdivided
until approximately 100 to 300 particles were obtained, with the larger sample size be-
ing used for smaller particles. The length, width, and thickness of these particles were
measured and the particles were classified according to the ratios of length to width
and width to thickness. A flat particle had a width-to-thickness ratio of 3 or greater,
whereas an elongated particle had a length-to-width ratio of 3 or greater. Results of
3
TABLE 1
FLAT AND ELONGATED PARTICLE COUNT
Percent in Each Class

Sieve Size
Particles
Counted Elongatedb Flat and Not Flat or
Passing Retained
Elongated Elongated
'h in. '/s in . 101 93 0 4 3

'lain. No. 4 117 48 3 2 47
No . 4 No. 8 300 9 19 0 72
No. 6 No. 16 306 25 2 0 73
No. 16 No. 30 305 49 3 1 47
8 Width to thickness ratio greater than 3 .0 . blength to width thickness ratio greater than 3.0.
these particle measurements are given in Tables 1 and 2. As was expected, nearly all
particles in the 1/2- to %-in. size range were flat, but the percentage of flat and elongated
particles decreased as the size of the sieve opening on which the particles were retained
approached the wall thickness of the bottles. In the material passing the No. 16 sieve,
the percentage of flats began to increase again , and microscopic investigation of the ma-
terial finer than the No. 30 sieve indicated the presence of a significant number of flat
and elongated particles. Photomicrographs of the finer frac tions a re shown in Figure 1.
Specific gravity and Los Angeles abrasion tests (gradation C) were conducted on the
glass, and a hydrometer analysis was conducted on the material passing the No. 200
sieve. The bulk specific gravity of the crushed glass aggregate was 2.50, the absorp-
tion was 0.01 percent, and the Los Angeles abrasion loss was 41 percent.
The asphalt used was an 8 5-100 penetration asphalt cement furnished by the Shell Oil
Company and produced from a West Texas crude. Its properties are given in Table 3.
TEST PROCEDURES
Marshall test procedures, as specified by ASTM Method D 1559, were followed with
the following exceptions:
1. Immediately after mixing the specimen in a Hobart Model N-50 mixer for 2 min-
utes, the bituminous mixture was placed in the compaction mold and then the mold was
placed in an oven maintained at 275 F for 30 minutes. It was then removed and spaded
and compacted as specified in ASTM Method D 1559.
2. In the first trial mix series, rather than placing the specimens directly in a 140
F water bath for 30 minutes prior to testing, they were first placed in a plastic bag and
then in the water bath. This was done to prevent surface stripping, which would alter
the results of mechanical analyses of the extracted aggregate. Some leakage occurred
in most of the bags, however, and in the second trial mix series all specimens were
placed in an oven at 140 F for 2 hours prior to testing.
After the completion of stability testing, bitumen content was determined using a re-
flux extractor, and a mechanical analysis of the extracted aggregate was performed us-
ing AASHO Method T 30.
TABLE 3
TABLE 2
PROPERTIES OF ASPHALT
SIZE DISTRIBUTION OF MINUS-200 MATERIAL
Property Value
Particle Particle
Percent Percent
Size Size Specific gravity at 60 F 1.011
Finer Finer
(microns) (microns) Penetration at 77 F 92
Viscosity, Saybolt Furol at 275 F, sec 143 . 5
74 100 20 17 Flash, Cleveland open cup, deg F 595
60 62 10 6 Solubility in CC1 4, percent 99.9
40 50 5 2 Ductility at 77 F , cm 150+
4
Figure 1. Photomicrographs showing particle shape of fine crushed glass fractions : (a) plus-100 mesh size
fraction ( X 27), and (b) minus-200 mesh size fraction ( X 67) .
MIX DESIGN
The first trial mix used aggregat e gr adation No. 1 as given in Table 4. This (!;r ada-
tion is one s uggested by Goode and Lufsey (4) for maximum density and is calculated
from the r elations hip -
P = (~)o.45
,o, x 100
where P percent passing a sieve having an opening of d inches,

D maximum size of the aggregate, and
0.45 an empirical constant.
It was realized that the exponent 0 .4 5 for maximum density was based on reasonably
equidimensional particles and thus might not yield maximum density for the flat and
elongated glass particles. However, it was taken as a starting point and modified as
noted later.
Six asphalt contents were chos en in 0. 5 percent increments from 4. 5 to 7 .0 percent
(total weight basis) and five specimens were made fo r ea ch as ph alt content. Six s peci -
mens were made each day for 5 days , with one specimen of each of the 6 asphalt con-
tents being made each day. A hand-operated Mar shall compaction 11amm e r was used
to apply 50 blows to each end of the specimen. Bulk specific gravity of each specimen
was determined using ASTlVI Method D 2726-68 T, afte1' which the specimens were tested
for stability and flow.
Results from this first trial are given in Table 5 and shown in Figure 2. As can be
seen from the graphs in Figure 2, there is no asphalt content that satisfies the Marshall
design criteria suggested by The Asphalt Institute and given in Table 6. Although sta-
bility and flow are adequate at the lower as-
phalt contents, air content is above the mini-
TABLE 4 mum specified level only at 4. 5 percent and
GRADATION OF AGGREGATES voids in the mineral aggregate are below
minimum allowable values at all asphalt con-
Gradation No. 1, Gradation No. 2,
Sieve Size
Percent Passing Percent Passing
tents between 4.5 and 6.0 percent. The data
indicated that a gradation deviating from that
1
/z in. 100 100 giving maximum density was necessary.
'le in. BB 88
No. 4 65 67 A second series of specimens was molded
No. 8 47 48 using aggregate gradation No. 2 given in Table
No. 16 35 37
No. 30 26 28 4. This gradation had less material passing
No. 50 18 18 the No. 100 and No. 200 sieves . Six asphalt
No. 100 14 11
No. 200 9 ,7 6.3
contents were chosen in 0.5 percent incre-
ments from 4.0 to 6.5 percent and, once
5
TABLE 5
MARSHALL METHOD DATA
Percent Unit Percent

Bitumen Percent Voids in Stabil!ty Flow
Weight
(TWB) (pcf)
Air Voids Mineral (lb) (1/100 In.)
Aggregate
First Trial Mix-Aggregate Gradation No . 1
4.5 140.7 3.88 13.89 820 15 .8

5.0 141.4 2.67 13.95 744 16.4
5.5 141.7 1.82 14.14 729 12.6
6.0 141.3 1.37 14.85 596 19.4
6.5 140.9 0.98 15 .59 536 19.4
7.0 139.9 1.14 16.61 429 20 .8
Second Trial Mix-Aggregate Gradation No . 2
4.0 137 .6 6.57 15 .36 1061 7 .6

4.5 137 .9 5.74 15.55 736 7.4
5.0 138.7 4.50 15.57 839 7.4
5.5 139.4 3.39 15.52 767 8.0
6.0 139 .5 2.62 15 .93 520 10.4
6.5 139.5 1.97 16.44 492 11.0
Note: All values are averages of 5 specimens.
llOO 24
- ~
---
--
- ) ~
._..._ 16 ~
............
~-
...,...._ /
--- 12
.~
5
ASPHALT CONTENT - %
6 " 7
8
4 5
ASPHALT OONTENT· %
6 7
u.: 143 4
~
--
~ "'
~142
~14 I ,...
/ ..........
,3
(/)
0
~2
" ...
........
...
I-
~ 140
" .....
"
~
!!:
<f
I
,~
0
4 5 6 7 4 5 6 7
ASPHALT CONTENT - % ASPHALT CONTENT- 01.
17
J
~ 16 L
c:i.
IL
/
~ 15
~
~.,,...
> 14
.-----
13
4 5 6 7
ASPHALT CONTENT - %
Figure 2. Marshall test property curves, first trial mix.

6
TABLE 6 again, 5 specimens were made for each as-

MARSHALL DESIGN CRITERIA a phalt content using 50-blow Marshall com-
Test Property Minimum Maximum
paction.
Results from this second trial are given
Stability 500 in Table 5 and shown in Figure 3. The re-
Flow B 18
Percent air voids, surfacing 3 5 duced fines content resulted in higher voids
Percent voids in mineral aggre-
1 contents and permitted the choice of an as-
gate, /2 in. mn.xfmunl size 15 phalt content satisfying all of the design
a Recommended by The Asphalt Institute for medium traffic (50-blow com - criteria suggested by The Asphalt Institute.
paction).
DEGRADATION STUDIES
One specimen was chosen at random
from the five specimens representing each
asphalt content for both trial mix series. The bitumen was removed from each of these
by use of a reflux extractor with benzene as the solvent. Washed sieve analyses were
conducted on the recovered aggregate (AASHO Method T 30); the results are given in
Table 7.
In order to express the degradation that occurred in terms of a single number rather
than a multiplicity of percentages , a factor lmown as the Hudson A was calculated ( 5).
Hudson A is one-hwidredth of the sum of the percentages passing the ten U. S. Standard
IC:.\)\, 14
-
1'..
-- ........... ~
,,
~,....
/
SX>
"" .....
/
~·
" ,.........._
~
o ..... )
.._.
400 6
4 5 6 7 4 5 6 7
ASPHALT CONTENT-% ASPHALT CONTENT - %
141 8
- ~6
.......
~
/ """"":l
39
;' ...... '-
138 __,,. ./ ~-
137 0
4 5 6 7 4 5 6 7
ASPHALT CXJNTENT - % ASPHALT CONTENT - %
18
~17
/
L--'"
14
4 5 6 7
ASPHALT CONTENT - %
Figure 3. Marshall test property curves, second trial mix.

7
TABLE 7
RESULTS OF SIEVE ANALYSES OF EXTRACTED AGGREGATES
Sieve Uncompacted Gradation of Ext ra cted Aggregate,

Size Percent Passing Percent Passing for Indicated Asphalt Content
First Mix Design Series 4.5 5.0 5.5 6.0 6.5 7 .o

1/2in. 100 100 100 100 100 100 100
3/.in. 88 94 94 94 93 92 93
No. 4 65 68 68 69 67 67 66
No . 8 47 49 49 49 49 48 47
No. 16 35 35 36 35 35 34 34
No. 30 26 26 26 26 27 23 25
No . 50 18 18 18 18 19 18 18
No. 100 14 14 14 14 14 13 14
No. 200 9 .7 9 ,0 9.3 8.7 8.9 8.8 9 .5
Hudson A 5.03 5. 13 5.14 5.14 5.13 5.04 5.07
Second Mix Design Series 4.0 4.5 5.0 5.5 6.0 6.5
1/2in. 100 100 100 100 100 100 100
'!.in . 88 95 95 94 94 93 92
No. 4 67 69 70 69 70 68 68
No. 8 48 50 50 52 51 51 50
No. 16 37 39 39 38 39 37 39
No. 30 28 28 28 28 28 27 28
No. 50 18 19 19 20 19 18 19
No. 100 11 13 12 13 13 12 12
No. 200 6.3 10.0 6.6 7 .6 7.8 6 .0 7 .8
Hudson A 5.03 5.23 5.20 5.22 5.22 5.12 5.16
sieves starting with the 11/2-in. sieve and including the No . 200 sieve . It has been found
that, with asphaltic concrete aggregates in the usual range of A from 4.00 to 6.00, a
change of 0.50 in the value of A would change the asphalt demand by about 1 percent by
volume (5).
The results from the fir st trial mix indicate that little degradation of the aggregate
occurred. There is an anomaly here in the percentages of material passing the No. 200
sieve. As can be seen in Table 7, the percentage of minus-200 mesh material actually
decreased after the specimens had been mixed, compacted, and tested for stability.
This loss of mate1·ial is thought to be caused by stripping and loss of surface material
on the specimens during the period in the water bath prior to testing for stability. How-
ever, it is quite apparent that very little new minus-200 mesh material was produced
during mixing, testing, and compaction of these specimens and that, in general, degra-
dation was very light as indicated by changes in Hudson A.
Degradation was more pronounced in the second trial mix series. The Hudson A in-
creased by as much as 0.20 and there was a greater percentage increase in the minus-
200 mesh material. The fact that more air voids were present indicates that perhaps
the asphalt filler matrix was not as effective in cushioning the compactive blows, thus
resulting in the formation of more fines.
WATER-RESISTANCE STUDIES
Static stripping tests (ASTM Method D 1 ()64) were conducted on several combinations
of glass aggregates and asphalt. Stripping was noted in the combination of glass and
asphalt cement with no antistripping compound added. However, mixtures of glass and
either a slow-setting cationic emulsion or an asphalt cement with a proprietary anti-
stripping compound exhibited no stripping.
One set of 6 specimens utilizing gradation No. 2 and an asphalt content of 5. 5 percent
was molded to determine the percentage stability lost after immersion in water at 120
F for 4 days. However, stripping was so severe that no stability testing was possible.
Tests involving the use of asphalt cements and several antistripping compounds are in
progress.
8
DISCUSSION OF RESULTS
The previous data have indicated that a mix satisfying the Marshall design criteria
suggested by The Asphalt Institute can be designed using crushed glass aggregates and
a conventional gradation. The fact that a large percentage of the particles were flat or
elongated also suggests that the stability values obtained by the Marshall testing method
may have been on the conservative side because of particle orientation effects. Puzin-
auskas (6) observed the fact that, in mixtures containing elongated or flattened aggre-
gates, the agg1•egate particles tend to become axially aligned in a direction perpendicular
to the direction of the compaction force. This effect was most pronounced in specimens
compacted by intermittent impact-type compactive forces, such as are applied by the
Marshall compactor. Specimens tested such that the compressive force was parallel
to the compactive force always exhibited higher strengths than specimens tested with
the force perpendicular to the compactive force. Thus, because the specimens were
tested in a direction perpendicular to the direction of compaction and because a large
number of flat particles were present, it is likely t hat the Marshall stabilities are con-
servative.
The data also indicate that degradation of the aggregate may be a problem in mixes
deviating from a maximum density gradation. Nijboer (7) reports that a German inves-
tigator, Herrmann, has stated that the crushing of aggregate under traffic is dependent
on the grading, with maximum density grading resulting in less crushing. This would
appear to be the case with the two trial mixes investigated in this study, as deviations
from the maximum density gradation resulted in greater increases in the minus-200
material.
The extent to which the degradation noted in the second trial mix would impair mix
performance is open to question. Data obtained by Goode a d OWings (8) in a laboratory
and field study of degradation occurring in hot asphaltic concrete wearing course mix-
tures were used to obtain values for changes in Hudson A associated with field compac-
tion and traffic compaction. Although this parameter was not used in their paper, the
data they presented permitted calculation of it for purposes of comparison. An increase
in the Hudson A value of 0.25 resulting from field compaction and an increase of 0.39
resulting from field compaction and h'affic were both considered to be minor and insuf-
ficient to affect the service behavior of the pavement. These changes in Hudson A, how-
ever, were associated with no more than a percent, more or less, increase in the
minus-200 mesh material.
The use of impact-type compaction with the glass aggregate may also have resulted
in more degradation than would occur under field-rolling conditions. This could be
more thoroughly studied in a field installation using the glass mixture.
CONCLUSIONS
Based on the laboratory work completfild thus far, the following conclusions can be
drawn:
1. Bituminous mixtures satisfying Marshall design criteria recommended by The
Asphalt Institute can be designed using penetration-grade asphalts and aggregates com-
posed entirely of crushed glass.
2. Some degradation of the glass aggregate does occur under laboratory mixing,
compacting, and testing conditions. The degradation appears to increase as the original
gradation deviates from a maximum density curve.
3. Severe stripping occurs when a bituminous concrete using dense-graded glass
aggregates and asphalt cement with no additives is exposed to water at 120 F. Stripping
is also exhibited in mixtures of asphalt cement and glass tested in the ASTM static
stripping test (ASTM Method D 1664).
4. Mixtures of glass aggregates and asphalt cement treated with an antistripping
compound did not strip when subjected to the static stripping test (ASTM Method D 1664).
Mixtures of glass aggregates and a slow-setting cationic emulsion were also resistant
to stripping in the static test.
9
ANTICIPATED FURTHER RESEARCH

Further studies of the water-resistance of bituminous mixtures using glass aggre-
gates are planned including immersion tests. A field installation consisting of patches
or overlays of these mixtures is planned to assess several mixture parameters that are
difficult to evaluate in the laboratory. These include workability of the mixture during
placement as well as skid resistance, tire wear, and reflectivity properties of the
surface.
ACKNOWLEDGMENT
This investigation was supported by the Bureau of Solid Waste Management of the
U. S. Public Health Service. The laboratory work was carried out by Charles Foster,
graduate research assistant in civil engineering, at the University of Missouri-Rolla.
REFERENCES
1. Solid Wastes Pile Up While Laws Crack Down and Engineers Gear Up. Engineering
News-Record, Vol. 182, No. 24, June 12, 1969.
2. Kenahan, C. B., Sullivan, P. M., Ruppert, J. A., and Spano, E. F. Composition
and Characteristics of Municipal Incinerator Residues. Bureau of Mines, U. S.
Dept. of the Interior, Rept. of Investigations 7204, Dec. 1968.
3. Municipal Refuse Disposal. American Public Works Assn., 1966.
4. Goode, J. F., and Lufsey, L.A. A New Graphical Chart for Evaluating Aggregate
Gradations. Proc. AAPT, Vol. 31, 1962, pp. 176-207.
5. Effects of Different Methods of Stockpiling and Handling Aggregates. NCHRP Rept.
46, 1967.
6. Puzinauskas, V. P. Influence of Mineral Aggregate Structure on Properties of As-
phalt Paving Mixtures. Highway Research Record 51, 1964, pp. 1-18.
7. Nijboer, L. W. Plasticity as a Factor in the Design of Dense Bituminous Road Car-
pets. Elsevier Publishing Co., New York, 1948.
8. Goode, J. F., and Owings, E. P. A Laboratory-Field Study of Hot Asphaltic Con-
crete Wearing Course Mixtures. Papers on Road and Paving Materials, ASTM
STP 309, 1962.
Discussion
K. O. ANDERSON, The University of Alberta-The concern expressed by the authors
regarding the disposal of increasing amounts of waste glass and development of new
methods for disposal of this waste component is shared by many investigators. It so
happens that a similar study, in which I had a small part, has just been completed at
our University (9). Although the investigation dealt with several possibilities of con-
version of city refuse to useful products, including the possible recovery of metallics
and heat as well as the nonmetallics, the potential use of this latter fraction of inciner-
ated refuse as an aggregate for asphalt concrete was investigated.
In support of the authors' statement, no previous investigations of this nature were
found by the discussant in the literature. As a result, a similar, although not as thor-
ough, investigation was undertaken to evaluate the potential of this aggregate as a con-
stituent of a hot-mix asphalt paving mixture. Standard mix design procedures, following
ASTM Method D 1559 but modified slightly to include a measure of loss in stability re-
sulting from the action of water, were also used.
The nonmetallic fraction of the incinerated refuse accounting for approximately 65
percent of the total was crushed and separated into seven fractions and recombined to
meet requirements for Type IVb Asphalt Institute gradation. A 200-300 penetration
grade of asphalt cement was used without any additives. A mechanical compactor was
used to impart 50 blows per face at the recommended compacting temperature based
on viscosity. Initial stability and flow values were obtained on specimens brought to
10
140 Fin an air bath, and comparative

AVERAGE values were obtained after soaking in
VALUES water at 140 F for 24 hours. Results ob-
143.8 pd tained are shown in Figure 4.
Although insufficient tests were per-
formed to establish optimum conditions
for this mix, it was apparent that sur-
prisingly high stability values could be
attained, even after the period of immer-
(b) sion. Even though the retained stability
2000 values were 69 and 59 percent for the 9
and 10 percent asphalt series respec-
-" -
-1500
>
AFTE2 SCA~l~~G
FOR 24 HRS AT
tively and below a generally accepted
minimum of 7 5 percent retained stability,
140°F the absolute values were still above the
!::
-'
iii minimum stability requirements for me-
:! IOOO dium traffic as suggested by The Asphalt
V>
Institute. The flow values were very er-
ratic; however, some were within the
500 prescribed limits.
On the basis of these preliminary
(c)
24 tests, we were able to state that it would
V>
w MAX
appear possible to use this material in
~o 18 an asphalt paving mixture. This would
g~
_
....
0
14 VRECOMM
LIMITS
ENDED be in support of the first conclusion by
the authors. 'l'he less sever e stripping
0 MIN.
8 in our tests, in contrast to the third con-
4 clusion of the paper, may be due in part
to the fact that the incinerated refuse
4 6 8 10 12 contained minor amounts of porcelain,
PERCENT ASPHALT BY WT. OF MIXTURE rock, some metallics, and fly ash that
may have increased the resistance to
Figure 4 . Results of tests on asphalt paving mixture us- stripping of our mixture. It will be in-
ing nonmetallic incinerator refuse as aggregate. teresting to hear of further studies on
this very timely topic.
Reference
9. Lilge, E . 0. Preliminary Report
on the Conversion of City Refuse t o Dsefui Products, Dept . of Mining and Metallurgy,
Univ. of Alberta, Edmonton, Dec. 1969.
WARD R. MALISCH, DELBERT E. DAY, and BOBBY G. WIXSON, Closure-The authors

appreciate the supporting comments and the additional information provided by the dis-
cussant. The possibility that incinerator residue can be used with asphalt to produce
a reasonably water-resistant mixture without the use of antistripping agents is further
indication that this may be a feasible means for reutilization of solid waste materials.
The higher stabilities that were obtained by Anderson would also seem to indicate a
difference in surface texture or shape between the crushed bottles used in our investi-
gation and the incinerator residue. Continuing research should further clarify this
point.
Construction Specifications for Volcanic
Cinders Used as Road-Surfacing Aggregate
LARRY G. HENDRICKSON, U.S. Forest Service, Portland, Oregon; and
JOHN W. LUND, Oregon Technical Institute, Klamath Falls, Oregon
The purpose of this study was to determine parameters that in-

fluence the performance of volcanic cinders when used as a
wearing surface for unpaved forest access roads. The steady
depletion of good, high-type aggregate sources has brought
about renewed interest in other aggregate types, making this
study particularly timely. A literature review, correspondence,
and personal interviews were conducted to determine existing
use and performance of cinders. A road rating system was
developed, similar to the AASHO Road Test method, to deter-
mine which were "good roads" and which were "poor roads"
based on a 0 to 5 numerical rating scale. Thirty of the rated
roads and the corresponding cinder pits were sampled and the
material was tested in the laboratory. The physical properties
of the cinders were then statistically correlated with the road
rating. It was found that density, gradation, durability, and
plasticity were the most significant independent variables.
Based on these results, specifications for untreated cinder sur-
face courses have been developed. These include (a) a mini-
mum of 100 percent compaction relative to AASHO T 99; (b)
gradation limits of 100 percent passing the 1-in. sieve, 80 to
95 percentpassing the %-in., 35 to 60 percentpassing the No. 4,
22 to 45 percent passing the No. 10, 8 to 25 percent passing the
No. 40, and 3 to 12 percent passing the No. 200; (c) a plastic
index value between 2 and 10; and (d) a maximum Los Angeles
abrasion value of 50 prior to processing. In most cases this
will require crushing of the harder (purple, gray, and black)
cinders.
•VOLCANIC CINDERS occur in many areas of the world and are particularly common
in central Oregon, which is the location of this study. The Forest Service of the U. S.
Department of Agriculture builds many miles of untreated aggregate-surfaced roads in
the states of Oregon and Washington. In the central Oregon area, the use of volcanic
cinders to surface those roads is very common. The performance of cinder-surfaced
roads has varied, with both good and poor results. The general trend among many
highway agencies is to avoid their use. However, the steady depletion of good, high-
type aggregate sources has brought about renewed interest in other aggregate types,
including volcanic cinders. Present use of volcanic cinders is primarily for low-class
roads, either untreated or with a light asphaltic surface treatment.
PURPOSE OF STUDY
The purpose of the study was to determine the parameters that influence the per-
formance of volcanic cinders when used as a wearing surface and base course for forest
development roads. The approach to this problem suggested an evaluation of the
11
12
performance of existing untreated cinder roads to determine which were "good roads"
and which were "poor roads" based on a numerical rating scale. This, in turn, was
related to the physical properties of the cinders in the roadway and from the borrow pit.
Included in the study was a survey to determine current practice and associated prob-
lems involved in the use of cinders as road aggregates. These relationships, together
with other field observations, were used to recommend specifications for selection of
borrow material and for techniques of construction quality control.
This paper is concerned with the relationship between cinder properties and road-
surfacing performance. The details of the entire study, giving the background of the
road-rating system and the results of the survey to determine current use, are pre-
sented in a U. S. Forest Service report (!).
BACKGROUND !NFOPJV!AT!ON
Cinders, as referred to in this report, are a pyroclastic material associated with
volcanic activity (2, 3). Fragmented material of this type is generally classified ac-
cording to size. Pieces larger than approximately 10 in. (256 mm) in diameter are
called blocks or breccia if angular and ejected in a solid state and are called bombs if
ejected in a plastic state, which produces a roundish or ellipsoidal shape with twisted
ends. Pieces between 10 in. and 'ilia in. (4 mm) in diameter are called lapilli or cinders.
Particles finer than approximately '!ia in. in diameter are called ash or dust (4).
A cinder, unlike most rocks, does not have a set mineral or chemical composition
because it is mainly a textural classification of rock. It may be either acidic (light-
colored matter consisting principally of quartz and feldspars) or basic (dark-colored
matter consisting principally of ferromagnesian minerals with no quartz). Color re-
flects the conditions prevalent during formation (i.e .• temperature or oxygen available)
and weathering. Oxygen-deficient cinders are darker incolor, whereasredorbrownish
cinders generally indicate weathering and are located in the upper zone of a deposit.
Minor accessory minerals, such as iron, may also contribute to the coloring of the cinder.
Scoria is a term used to describe any volcanic ejecta that is rough, sharp, and ve-
sicular, either pyroclastic material or the upper surface of some lava flows. It is
commonly black or reddish and principally basic in composition. The term is often
used to include cinders.
Cinders are located throughout the world. The main areas in the United States are
central Oregon east of the Cascades, northern and eastern California, Arizona, New
Mexico, and Hawaii. Notable deposits are found in New Zealand, Japan, Turkey, Mexico,
Central America, and the Caribbean. In other countries the terms scoria, scoriae,
volcanic agglomerate, or grits are used more often than cinders, especially by British
Road Research Laboratory reports (~, ~' J_).
Current Practice and Experience

According to an earlier publication (§), almost all agencies will avoid the use of
cinders if other sources of mineral aggregate are available. When used, cinders are
applied as surfacing material primarily on unpaved roads, such as logging, forest ac-
cess, and low-traffic roads. On higher type roads cinders are used as subbase and
base material and for asphalt-stabilized surfacing.
The best results with pit-run cinders have been with the softer types (generally the
red-colored types; average Los Angeles abrasion in the 40s). The harder types (purple,
gray, and black; average Los Angeles abrasion in the 30s) lack fines, and thus are un-
stable (Fig. 1). When crushed, the harder types perform more satisfactorily, because
the necessary fines are obtained and thus the surface better withstands the abrasive
effects of traffic. In addition, the rough riding quality resulting from the pit-run over-
sized material is eliminated.
Grid rollers are popularly used for breaking the larger pieces and for compaction.
The effectiveness of this type of roller for compaction is questionable. Steel wheel,
rubber-tired, and vibration rollers appear to achieve better results, as does compac-
tion from logging and other heavy truck traffic.
13
Figure 1. Sycan Road, North of Bly, Oregon: Note contrast between washboard surface
using purple cinders (foreground) and smooth surface using red cinders (background).
With increasing traffic, new road construction and pavement thickness upgrading,
and the lack and depletion of good mineral aggregate sources, the use of cinders is being
investigated in more detail. Many agencies feel that cinders can be used effectively
with proper gradation and other quality controls. Asphalt treatment will probably give
the best road surface and the most economical maintenance over an extended period.
CORRELATION OF CINDER PROPERTIES WITH

SURFACING PERFORMANCE
Development of the Rating System
A road-rating system was developed similar to the AASHO Road Test methods. A
major differentiation was that for this study the roads were rated over a period of time;
thus, the roads were given a performance index (Pl) rating rather than a serviceability
rating. Forty-seven roads were rated on a scale from 0 to 5 with performance increas-
ing with increasing rating values (Figs. 2 and 3 ). This rating was performed by people
familiar with the long-term serviceability of the road. After the rating was obtained,
it was correlated with maintenance frequency, rutting, watering, speed, and traffic vol-
ume by means of a multiple linear regression computer program. The resulting re-
gression equation had a multiple correlation coefficient of 0. 812. This equation was
then used to determine the performance index of the various roads, which was, in essence,
the performance rating with the individual rater's bias removed. Complete details of
this rating technique can be obtained by referring to the U. S. Forest Service report (!).
Road- and Pit-Sampling Program

Once the road-rating system was established using the PI, the next step was to select
typical examples of roads within each rating group and to sample the surface material.
In all, 30 roads and their corresponding borrow-pit sources were sampled, including
15 in the Winema National Forest, and 5 each in the Fremont, Rogue River, and Deschutes
National Forests. The general location of the sample sites is shown in Figure 4.
The purpose in sampling these roads and pits was to determine physical properties
that could be identified with roads of high performance (Pl) and those associated with
14
Figure 2. Kirk Road, Northeast of Chiloquin, Oregon: Rough surface contains black, uni·
formly graded cinders (Pl: 2.1).
Figure 3. Twincheria Road, East of Butte Falls, Oregon: Cinders cemented with iron oxide
produce a hard, compact riding surface (Pl : 3.1 ).
15
Figure 4. Study location map.
low performance, or more specifically the correlation between performance and phys-
ical property. All roads were constructed of pit-run material.
The testing program consisted of two parts, field sampling and testing, and labo-
ratory testing. The field sampling and testing portion involved first selecting a straight
sectio!l of the rated road and obtaining approximately 150 lb of material (2 sample
sacks). This was obtained from a trench dug at least halfway across the road and
through the majority of the sul'face course (approximately 4 to 6 in. deep). In addition,
the following field tests were performed: (a) in-place density using the Washington
densometer; (b) in-place moisture content using the Speedy moisture tester; (c) color
identification; and (d) penetration test of surface and base material (later discontinued
because there appeared to be no correlation).
The pit site was located next and a 75-lb sample (1 sample sack) was obtained from
the approximate area in the pit where the road sample was obtained during construc-
tion. Obtaining a related sample was sometimes difficult because of the mixing of pit
material and subsequent use of the pit for other construction projects.
The samples were sent to the Forest Service Regional Office in Portland, and in
turn sent to the Federal Highway Administration laboratory in Vancouver, Washington,
or to the Oregon State Highway Department (OSHD) laboratory in Salem (Table 1 ).
16
TABLE 1 Splitting of the work was necessary be-
TESTS PERFORMED ON ROAD SAMPLES cause of the volume of testing involved and
Tests Performed on the tim e required. Some additional test-
Test
Test Method
ing was also performed at the Forest Ser-
Description Road Pit vice Engineering Materials Laboratory in
Samples Samples
Portland, Oregon.
Los Angeles In addition to the testing, the following
abrasion AASHO T96 x x
calculations were performed: (a) coeffi-
Specific !(ra vity
of fines California 208B X cient of uniformity (Cu) for both road and
Unit weight,
pit gradation; (b) coefficient of curvature
dry-rodded (Cc) for both road and pit gr adation· and
Unit weight, (c ) relative compaction of road sample (as
loose compared to maximum density). Several
x points shouid be made m regard to the test
Liquid limit AASHO T89 x x results:
Sand equivalent AASHO T 176 x
1. Los Angeles abrasion test-The ac-
Oregon air
degradation OSHD x tual values for the Los Angeles abrasion
Moisture -density appear to be very good (low values) and
relations hip AASHO T 99 D x this is due in part to the volume of ma -
Gradation (dry terial used in running the test. The stan-
sieve ana lysis) AASHO T2 7 x x <lard charge is 5 ,000 grams of material.
aThese dete rm inations were made only on samples sent to the Oregon State With the low bulk specific gravity of cin-
Highway Depa rtment (OSHD) ,
ders, a greater volume of material is used
than would be normally utilized (9). Be-
cause this study considers only cinder ag-
gregate , relative values between va rious cinder samples should be acc eptable for use
in the analysis.
2. Specific gravity test-Unfortunately this test was run in a Le Chatelier flask,
(Calif. 208B); thus, only material passing the No. 4 sieve was used and the apparent
specific gravity was determined. As a result, the values were very uniform, whereas
values for the coarse portion would probably have been more meaningful.
3. Plasticity index and liquid limit-In most cases the cinders had very few plastic
fines, thus giving no plastic limit. However, the liquid limit could be determined for
most samples (often only by one point and extrapolation) and was thus reported because
it appeared to give good correlation.
Correlation With Performance Index

The results of the field sampling and laboratory testing were compared with the per-
formance index (road rating) by means of the multiple regression program. Nineteen
independent variables from road tests and 12 independent variaule1> frum pit tests were
considered. These variables are listed in table 2 together with their average values,
standard deviations, and correlations with the performance index.
According to the regress ion program, the following are the most significant vari-
ables (with cori·elation coefficients above 0. 3) in decreasing order of importance. The
values for the percent passing the 11h-and % - in. sieves are results from pit source
samples.
Variable Correlation With Performance Index
Relative compaction 0.395
Plasticity index 0.392
Percent passing No. 4 sieve -0.341
Percent passing 11/:i-in. sieve 0.328
Coefficient of curvature 0.323
Oregon air degradation-H 0.319
Percent passing %-in. sieve 0.310
17
TABLE 2
CORRELATION OF ROAD AND PIT TEST VARIABLES
Standard Correlation With

Variable Mean
Deviation Performance Index
Road Test Variables
Sand equivalent 49.4 24.0 -0 .130

Los Angeles abrasion, percent 43 .8 10.4 -0.099
Maximum density, pcf 102.5 7.6 -0 .058
Relative compaction, percent 102.4 5.7 0 .395
Specific gravity 2.72 0.12 -0 .154
Plasticity index 0.1 0.4 0 .392
Liquid limit 23 .8 4.2 0.196
Oregon air degradation,
sediment height, in. 3.5 4.7 0 .319
Oregon air degradation, per-
cent passing No. 20 sieve 26 .4 9.8 -0.024
Color (red =2, black =1) 1.8 0.4 -0 .183
Coefficient of uniformity 52 .6 31.4 0.269
Coefficient of curvature 0. 72 0.56 0.323
Gradation, percent passing
sieves
1
1 / , in. 96.1 5.6 -0 .119
3/. in. 88,9 6.9 -0 .117
3
/a in. 76.2 7.4 -0 .213
No. 4 61.4 8.6 -0.341
No. 10 50 .1 9.5 -0 .325
No. 40 31.1 8.0 -0 .312
No. 200 9.6 2.8 -0.243
Pit Source Variables
Los Angeles abrasion, percent 46.7 10.6 -0 .029

Liquid limit 28. 5 3.3 -0.082
Unit weight, loose, pcf 67.1 7.5 -0 .073
Coefficient of uniformity 38.1 22.6 -0.071
Coefficient of curvature 1.60 1.21 -0 .127
.G radation, percent passing
sieves
1 1/, in. 96. 1 2.8 0 .328
'/,in. 86 .0 6.5 0.310
3/ain. 69 .0 9.8 0 .286
No. 4 50 .5 12.8 0.247
No. 10 37 .1 13.4 0.252
No. 40 18 . 5 8.4 0 .090
No. 200 4 .6 2.5 0.095
The multiple regression equation is as follows:
PI= - 5.65 + 0.06(P) + 0.05(RC) + 0.08(0A-H) - 0.04(0A-20) + 0.06(PP1.5)
- 0.07(PP4) - 0.24(Color) + 0.09(PP10) - 0.02(PUW) + + 0.30(CC)
where
PI performance index;
p plasticity index;
RC relative compaction;
OA-H Oregon air degradation, sediment height;
OA-20 Oregon air degradation, percent passing No. 20 sieve;
PPl.5 percent passing 1%-in. sieve from pit;
PP4 percent passing No. 4 sieve from pit ;
Color cinder color (red = 2, black = 1);
PP10 percent passing No. 10 sieve from pit;
PUW loose unit weight of material from pit; and
cc coefficient of curvature from road.
18
This equation has a multiple correlation coefficient of 0. 854. Individual equations were
alsu determined for each of the 9 most significant variables. The equations and the in-
dividual correlation co fficients indicate the relationship between the physical test prop-
erty and the performru1ce index (rating). Unfortunately, 24 of the 3 2 roads had a per-
formance index within the relatively narrow range of 2.1 to 3. 5, with only 2 roads rated
below this range. Thus, the regression analysis gave the greatest weight to this range,
with the result that most of the regression lines have very steep slopes.
Several important trends were noted when comparing the test results with the per-
formance index. They are summarized in the following:
1. Compaction-In the tests of road samples, the relative compaction is very im-
portant, whereas the maximum density is not.
2. Gradation-In the tests of road samples, the coefficient of curvature varies di-
rediy with the performance index (i.e., gradations with a concave-shaped curve on a
semilog graph are associated with the higher performance index). This relationship
is also reinforced by the trend showing less material passing the No. 4, No. 10, and
No. 40 sieves with a higher performance index; the value !or the other sieves appears
not to be significant. In the tests of pit samples, the increased percentages passing
the 1~-in. and 3/,.-in. sieves is desirable (i.e., less large-size material).
3. Plasticity-In the road sample tests, the significant characteristics of the fine
material appear to be increased plasticity index, liquid limit, and sediment height in
the Oregon air degradation test, and decreased sand equivalent with increasing per-
formru1ce index (i.e., cohesive binder is desirable).
4. Abrasion-In the road sample tests, the Los Angeles abrasion value varies in-
versely with the performance index (i.e., harder material with less abrasion is desir-
able). In the pit sample tests the decreased Los Angeles abrasion value is desirablP..
Recommended Specifications and Field Quality Controls

Based on the physical test results, the cinder characteristics that appear to be sig-
nificant are compaction, gradation, plasticity, and resistance to abrasion. Using these
characteristics, the following specifications are recommended for the construction of
cinder wearing surfaces.
Compaction-In all cases the final constructed surface should be compacted to at
least 100 percent of standard compaction (AASHO T 99 ). This should extend to a depth
of 6 in. below the surface. In most cases, it is recommended that a vibratory com -
pactor be utilized, or as an alternate a pneumatic roller. Sheepsfoot and grid rollers
should be avoided. Steel-wheel rollers can be used for finish or proof rolling.
Gradation-The maximum-sized particle should be 1 in. in diameter in the finished
surface. In certain cases, this size could be increased to 1Ji in. at the pit source, as
some degradation will take place in the construction process. The following in-place
road gradation is recommended (based on dry-sieving similar to AASHO T 27, with a
maximum shaking time of 5 minutes):
Sieve Size Percent Passing

1 in. 100
% in. 80 to 95
No. 4 35 to 60
No. 10 22 to 45
No. 40 8 to 25
No. 200 3 to 12
These limits will provide a dense gradation.
This gradation is similar to AASHO M 147-55, grading C, and U.S. Forest Service
Item 151, grading B, Item Rl51, grading C, and Item Rl50, grading C-1. The gradation
recommended in this report differs somewhat in that it has a smaller percentage pass-
ing in the No. 4 to No. 40 sieve range to allow for subsequent degradation of the cinders.
Local practice may indicate that the standard specifications listed previously are
adequate.
19
It is realized that in-place road gradations are often difficult to estimate on the basis
of pit-sampling. The best technique for sampling cinder pits is to obtain as representa-
tive a sample as possible (i.e., for the portion of the pit to be used, or a blended sample).
Based on results from this study, the f.,it gradation should have approximately 5 percent
less passing a given sieve (below the ~-in. size) than is desired for the finishedproduct
on the road . This a llows for degradation in transporting and compacting the cinders.
This value should be approximately 10 percent for the softer red and brown cinders.
The recommended gradation requirements could be met by selected blending and
scalping at the pit source. However, in some cases it may be necessary to run the ma-
terial through a primary crusher with some selective screening.
Control of gradation is felt to be extremely important because too much large ma-
terial (above 1-in. size) causes a rough riding surface, and too much fine material will
cause excessive and early washboarding and rutting.
Atterberg Limits-The material passing the No. 40 sieve should have the following
characteristics: The liquid limit should be 35 at the maximum, and the plasticity index
should range from 2 to 10.
Based on the trends illustrated, sand equivalent test results ranging from 50 to 30
would give an indication of the required plasticity. This value should not, however, be
used as a specification requirement because it measures somewhat different properties
of the fine material.
This requirement of plasticity is felt necessary to replace evaporated surface mois-
ture through capillarity. In most cases, cinders do not produce this plasticity, and thus
the addition of plastic binder is necessary. This binder may be available as a cap of
weathered cinders over the cone, or can be generated in some cases by watering, mix-
ing, and compacting the softer brown and red cinders. A pulvemixer attached to a small
tractor would perform an effective job of blending the cinders and plastic borrow ma-
terial. In some pits, the presence of an iron oxide binder will give a satisfact ory bind-
ing action in the road. If this is the case, the plasticity requirements could be waived.
If immediate paving of the cinder surface with an impervious mat is anticipated, then
the plastic matel"ial should be avoided. If paving is contemplated at a later date, then
lime, asphalt, cement, or other additives should be considered to reduce or eliminate
the plastic characteristics at this time.
Resistance to Abrasion-Maximum pit Los Angeles a brasion of 50 should be used to
prevent excessi ve degradation of the material in the road surface, because excessive
fines will cause early washboarding and rutting. With pit-run material, this degrada-
tion was necessary to generate sufficient fines for binder. However, with the recom-
mended limits on gradation, this will no longer be necessary or desirable.
CONCLUSION
In most cases, the harder cinders (purple-, gray-, and black-colored ones) should
have less degradation and provide a better riding surface over a longer period of time.
However, this type of cinder usually lacks the necessary fines, and thus some crushing
is required to meet the 1·ecommended gradation limits. The softer cinders (red- and
brown-colored ones) sometimes have excessive degradation; thus, care must be ex-
ercised in selecting the necessary gradation. The Los Angeles abrasion requirement
will eliminate the use of most of the softer types. In some pits, where several cinder
colors are available, blending of the harder and softer types may produce the necessary
gradation and plasticity to provide an adequate riding surface.
SUMMARY
The foregoing analysis shows that cinders behave much as any other granular ma-
terial when used as an untreated road surface. The study revealed that several modi-
fications were desirable to standard specifications to better fit the properties of cinder
aggregates. The test data indicate that a Los Angeles abrasion value of 50 would be
more appropriate than the usual limit of 40. The mean of all Los Angeles abrasion
values is a 46. 7 percent loss, with a standard deviation of 10. 6 when sampled from the
pit. Therefore, a maximum allowable limit of 40 is too restrictive. The test data show
20
that both the plasticity index and the gradation are very important for good performance.
The graualiun should ideally be somewhat on the coarse side of the gradation band to
allow for subsequent degradation under traffic. Relative compaction was also found to
be very important, which further emphasizes that cinders behave much as any other
granular material used as a road aggregate.
Reference to Table 2 shows a minor correlat ion (low value ) between Los Ang eles
abrasion and p erformance index (PI) and a positive, strong correlation be tween the
Oregon air degradati on sediment height (H) valne and Pl. However, it is recommended
that the Los Angeles abrasion test be used as a control t es t r ather than the Or egon air
degr ada tion tes t. Thi s is because high -qua lity aggrega te gives l ow Oregon air H values.
B eca use this s tudy s howed a positive correlation belween H and PI, the s tatistics in-
dicate that l'1e poorer quality material yields higher PI value s . This must, however,
be considered in light of the fact that ; for ~ 11 the r n;:i_d.s stl!d.ied, the surfacing was pfa.ccd
with almost no gradation control and the surfacing was almost always lacking in fines.
Therefore, when the material had a high H value, the cinders tended to break down
under watering and traffic to produce the needed fines. Thus, there was a strong cor-
relation between H value and PI. The recommendations in this report are to control
the gradation before placing the material; consequently, the desirability of having a high
H value is no longer valid. There was a negative correlation between the Los Angeles
abrasion and PI; therefore, a specification for Los Angeles abrasion is recommended.
If a similar study were performed on roads where cinders were applied under a con-
trolled gradation specification, it is felt by the authors that the Los Angeles abrasion
test would show a much stronger negative correlation. For this reason, and because
of the obvious need to specify a material that will tend to hold its gradation, the Los
Angeles abrasion test is recommended. The maximum allowable limit was based on
t.'1e mean and the standard deviation of the test results.
It is believed that the specifications recommended are reasonable, based on test re-
sults of typical cinders in central Oregon and the proven, accepted practice for aggre-
gate surfacing control. The typical specifications when inserted into the regression
equation yield a performance index over 5. 0, which is an indication that the criteria are
consistent with the test results.
REFERENCES
1. Lund, J. W., and Hendrickson, L. G. Road Performance and Construction Specifica-
tions Using Volcanic Cinders. U. S. Forest Service, Pacific Northwest Region,
Portland, Oregon.
2. Bullard, F. M. Volcanoes in History, in Theory, in Eruption. Univ. of Texas Press,
1962.
3. Emmons, W. H., Theil, G. A., Stauffer, C. R., and Allison, I. S. Geology Principles
and Processes. McGraw-Hill, New York, 1!:160.
4. Twenhofel, W. H. Principles of Sedimentation. McGraw-Hill, New York, 1950.
5. Laing, P. L., and Major, N. G. Civil Engineering Raw Material in New Zealand. The
Institute of Civil Engineers, Paper No. 8, London, 1964.
6. Carroll, D. M., and Beaven, P. J. Roadmaking Material in the Caribbean-V: The
Volcanic Antilles. Road Research Laboratory Note No. LN/744/DMC, PJB.,
London, Jan. 1965.
7. Study on Slags (Volcanic Scoria) Proposed for Base Cours e on P r oject CIA-31. U.S.
Bureau of Public Roads, Region 19, Guatemala, Ma rch 15, 1963.
8. Lund, J. W., and Hencfrickson, L. G. The Use of Volcanic Cinders for Low Class
Roads. Proc. Seventh Annual Engineering Geology and Soils Engine ering Sym-
posium, Idaho Dept. of Highways, April 1969.
9. Harra, W. G. Volcanic Cinders in Highway Cons truction. Proc. Weste1·n Associa -
tion of State Highway Officials, 40th Annual Conf., Las Vegas, Nevada, April 1961.
Lightweight Concrete Aggregate From
Sintered Fly Ash
L. JOHN MINNICK, G. and W. H. Corson, Inc., Plymouth Meeting, Pennsylvania
Lightweight concrete aggregate is being produced commercially from pulverized

coal fly ash by several plants. The products are of good quality, although they
vary somewhat with respect to shape and size. The type of fly ash used in the
manufacturing process is generally that produced from bituminous coal. The
characteristics of fly ash that are required in the forming and sintering process
are described. A description of the physical properties of the fly ash aggregate
is presented. The evaluation of these aggregates by the use of ASTM methods
gives a good indication of the performance of the various materials in structw·al
concrete. Comparisons are provided between other types of lightweight aggre-
gate concrete and, in all cases, the fly ash concrete shows a favorable range
of properties. Concrete strengths in excess of 6,000 psi (420 kg/cm 2 ) can be
produced with fly ash aggregate. Of significance in making comparisons is the
high rate of absorption of the aggregate, the residual pozzolanic activity, and
the relatively low cement factors required for lean concrete. Information is
presented on typical mix designs for use in structural concrete.
•THE PRODUCTION of lightweight aggregate from fly ash has received the attention
of investigators throughout the world (1,2). Although a number of early attempts to
produce a commercial product proved"Unsucessful, recent developments give every
indication that the manufacture of fly ash aggregate is now possible. Furthermore,
the product quality is quite acceptable for use in structural concrete, for insulation-
type concrete, and as an aggregate for masonry units.
Fly ash serves as an exceptionally good raw material because it is available in
large quantities in geographic areas where lightweight aggregates are in demand. The
commercial methods in use today produce aggregate by first preparing an unfired
pellet or granule of the fly ash followed by firing on some type of sinter strand. Be-
cau.s e fly ash usually contains some unburned carbon, there is an opportunity to utilize
the available heat generated by burning of residual combustible material during process-
ing. Fly ash aggregates, therefore, are properly classified as "sintered" products
rather than "bloated" materials, such as those produced by heating clay or shale in
rotary kilns.
Historically, coal ash has often found application as an aggregate. Originally this
was in the form of "cinders" from traveling grates or stoke1·-fed coal burners. These
systems have now been largely replaced by pulverized-coal burners and the resulting
fly ash is collected in a very finely divided condition in various types of dust-collecting
systems, such as cyclone collectors or electrostatic precipitators.
It is the purpose of this paper to describe the requirements for fly ash as used in
the sintering processes and the properties of the fly ash aggregates that are produced.
Although it is not intended to present the techniques involved in the production of aggre-
gates, some information is included relating to the effect of the manufacturing process
on the characteristics of the aggregate.
21
22
TABLE 1
TYPICAL FLY ASH ANALYSES FOR VARIOUS TYPES OF COAL
Sample Sourc e LOI Si02 Al,O, Fe20:i Cao MgO so,
Bituminous
Illinois 2 41 19 19 s 4
Kentucky 2 42 20 25 4 <l 2
Ohio 10 41 23 21 1 1 <l
West Virginia 6 41 24 21 4 <l <l
Pennsylvania
Source 1 10 47 26 11 2 1 <1
Source 2 5 38 24 26 4 I 1
Source 3 6 40 24 20 6 l 2
Source 4 20 40 23 12 <l <1
England 2 48 26 10 4 2
TI~l~- ...l
~ UV 26 12 l
West Germany 44 27 12 3 <l
Lignitic
Minnesota 23 12 14 32 10 3
Montana <l 28 10 11 30 17 1
North Dakota <1 31 13 11 24 11 4
North Dakota 4 23 16 8 36 9 4
Canada 1 39 21 7 19 5 2
Czechoslovakia 3 58 11 11 15 2 2
France <l 58 9 22 6 2 2
Greece 2 38 18 9 28 2 3
East Germany <l 39 7 21 21 2 10
Poland 2 34 6 38 4 9
RAW MATERIALS
The types of fly ash available cover a broad spectrum of chemical and physical
properties . Table 1 gives some typical analyses of fly ashes from various parts of
the world. There are several factors that affect the quality of ash as used in the manu-
facture of aggregate. Of prime importance is the type of coal used; thus, ash produced
from bituminous coal is usually quite low in calcium and magnesium oxides, whereas
ash produce d from lignite coal is usually high in these lwo constituents . The presence
of excessive amounts of free calcium oxide in a fi r ed aggr ega te can h ave deleterious
effects; among these is hydration of the oxide, resulting in the formation of "pits" or
"pops" in the hardened concrete.
A second factor affecting the quality of the fly ash is the ash content of the coal. The
majority of coals used in the United States is relatively low in ash, usually approxi-
mately 10 percent. The coals in Europe are frequently quite high in ash content (up to
about 45 percent). European practice, therefore, requires that the coal be burned more
thoroughly to recover as much fuel as possible . The resulting fly ash is low in residual
carbon-in many instances less than 1 percent. Carbon contents in the United States
are rarely this low and it is not uncommon for the carbon content of the ash to be in ex-
cess of 10 percent. It is essential that fly ash contain not less than about 4 percent and
prefer ably not more than 8 per cent carb on to be an acceptable material for aggregate
production .
A third factor affecting the fly ash quality is the type of ash collection systems in
use. This can directly influence the fineness of the ash. Fineness varies not only
between different coal-burning operations but also within a given operation. For ex-
ample, in a typical collection system used in this country there may be as many as 20
hoppers removing ash from the furnace. The location of these hoppers in relation to
the gas flow results in substantial differences in the particle size of the ash.
The variation of properties of the ash influences to a consi der able degree the quality
control of the sinter strand operations. Thus, if a fly ash chang·e s in its fuel content
(resi dual carbon) from a low to a high value as it is being fe d into th.e firing process,
serious difficulties can be experienced in controlling the operation. Variation in iron
oxide content is of considerable importance because this ingredient serves as a primary
23
flux and substantially influences the

N operation of the plant process as well
!i as the product quality.
'e n As a result of an intensive study
~ 4001--~~.i-.~~-1-~~-+-~~-i..-<---~~ BO a
0
made in the author's laboratory, it
~
has been found that the most impor-
tant requirements for suitable fly
~ ~ ash are the carbon content (or loss
& JOOt--f-r~+-~----:,.....,'--~+-~~-t-~~~60 l on ignition), the fineness (as mea-
-August
"'
---September
sured with a No. 325 mesh sieve),
and the iron oxide content. It has
also been found from statistical
200L-~~ .........
~~-1-~~-'-~~ ~~~40......
IOO 200 JOO 400 500 600 studies that these individual charac-
per cent Carbon • per cent thru 325 Mesh Sieve
teristics do not reliably predict the
resultant aggregate properties, but
that when these characteristics are
Figure 1. Effect of carbon content and sieve fineness on
pellet strength and percentage of high-strength pellets. used as multiple factors they give
signific ant relationships between the
fly ash and the properties of the final
product.
Figure 1 shows the effect of carbon and No. 325 mesh sieve fineness on both the
percentage of high-strength pellets and the fir ed pellet strength on a large number of
samples (at a relatively constant Fe2 Os content) taken at different periods of time from
one power plant. Although carbon content and sieve fineness do not give a high degree
of correlation individually, the product of the percent of carbon times the percent
through the No. 325 mesh sieve does, as shown in Figure 1. Based on studies such as
these, limits can be established to provide adequate quality control of the raw fly ash
for use in the sintering processes.
The interrelationship between fineness and carbon (also Fe20a content) can be use-
ful in arriving at the selection of fly ash material at the power station. Although fly
ash from both mechanical and electrostatic collectors can be used, there are situations
that require the rejection of a portion of the fly ash that does not conform to the gen-
eral requirements of the lightweight aggregate process.
In addition to the control of properties of the raw material, an important factor in
the preparation of the aggregate is the water content used in the formation of the green
pellets prior to firing. The water is necessary not only to develop adequate physical
strength of the pellet but also to control the porosity or absorption of the fired aggregate.
SHAPE AND SIZE OF FLY ASH AGGREGATE

Although there are a few exceptions, most fly ash aggregates develop the ultimate
shape and volume of the individual particles at the time the material is formed in the
unfired (green) state. Several devices are currently used to form the unfired pellets
or granules. Pelletizing drums or pans are1used in a few plants. These particular
devices form sphe rically shaped particles usually of a very uniform diameter. In the
process developed by the Corson Company, an extrusion principle is used that results
in particles of varying shapes and sizes, depending on the selection of the forming
equipment. Figures 2, 3, and 4 show typical shapes produced by the different forming
processes.
One significant factor resulting fro.m the choice of the forming method is the final
density that is achieved in the aggregate. Furthermore, this density also contributes,
to a substantial degree, to the absorption oJ~ the aggregate. The commercial operations
are all engineered to produce a product that retains the discrete shape of the unfired
pellet without formation of large clinkers, which are often produced from other types
of raw materials.
From the standpoint of producing aggregate for sh·uctural concrete, the range of
sizes of the pellets are set to meet the basic requirem ents of the ASTM specification
for concrete aggregates. In order to get the proper gradation, usually several methods
24
Figure 2. Extruded fly ash aggregate (crushed and graded).
Figure 3. Unfired and fired fly ash aggregate made by Figure 4. Unfired and fired fly ash aggregate made by
pelletizing process. extrusion process.
25
are used; either the material is made in different sizes and then blended, or, as in the
case of extruded products, the distribution of the size may be established in the green
bodies prior to the firing cycle. To produce finer aggregate, it is customary to crush
the material, screen it, and proportion it so that it meets the requirements of the ap-
plicable specification for size gradation. For this reason, the fly ash aggregate plants
usually incorporate crushing, screening, and blending operations in the manufacturing
process.
PROPERTIES OF FLY ASH AGGREGATE

There is as yet no standard method to measure the strength of the individual gran-
ules, although several methods are used in individual laboratories. A study reported
by the Central Electricity Generating Board (3) provides some results of tests of pellet
strength made on aggregate utilizing differentbu1·ni11g techniques. This report shows
that the strengths are quite good for those pellets produced by the sintering grate, as
well as those produced experimentally in rotary or shaft kiln equipment. In general,
the aggregate produced from fly ash is evaluated on the basis of ASTM Specification
C 330 and the reference test methods as called for in this specification. The tests in-
clude measurements of concrete-making properties, that is, strength and unit weights,
drying shrinkage, popouts, and durability. Also of importance is the test for stai.n ing.
The firing process results in some sintering of individual particles of fly ash on the
outer surface of the aggregate coupled with a more general fusion action in the interior.
The sintered exterior normally represents less than 10 percent of the total volume.
Figure 5 shows a close-up view of the typical structure of a cross section of fly ash
aggregate. The structure of the exterior sintered portion of the aggregate results
from the individual particles of the fly ash adhering to adjacent particles by partial
melting or softening. The structure of the interior of the aggregate represents a
Figure 5. Close-up of a sawed particle of fly ash aggregate enlarged 10 times.

26
TABLE 2
FLY ASH LIGHTWEIGHT AGGREGATE PROPERTJES
ASTM
Normal Range for Fly
Test Specification
Ash Aggregate
c 330
Unit weight, lb/ft' (kg/m')
At F. M . = 6. 5
(nnminol 3/, in. to No. 4) 37 to 48 (593 to 769) 55 maximum
At F. M. = 5.8
(nominal 'le in. to No. 8) 41 to 51 (657 to 817 ) 55 maximum
At F. M. = 3.5
(nominal 'le in, to O) 53 to 61 (849 to 977) 65 maximum
At F. M . = 3.0
(nominal No. 4 to 0) 56 to 63 (897 to 1,009) 70 maximum
24 -hour absorotion, percent 14 to 24
Staining · None to very light
Loss on ignition, percent 0 to 3.4 5.0maximum
NOTE: F.M ~ means Ummosi modulus.
multicellular structure of a more completely melted product. In both cases the voids
in the aggregate are caused by evaporation of the mixing water and elimination of car-
bon during the sintering process.
Table 2 gives the normal ranges of properties on a number of fly ash aggregates.
It is ta be noted that the values for absorption are quite high. The rate of absorption
is also hi gh and the effect that this has on compositions such as concrete is believed to
be more beneficial than delete1·ious. It has been reported (!) that the water absorbed
by this aggregate when preparing a batch of concre te is continuously available during
the curi ng process. The aggregate can thereby provide curing water r ather than ab-
sorbing water out of the cement matrix of the concrete mix. This may explain in part
why these materials ofteo provide somewhat superior performance when used in con-
crete. F ield experience has indicated that plastic shrinkage 0f the concrete i s also
substantially reduced (presumably as a result of the highly saturated condition of the
aggregate). The relative absorption of the different portions of the fly ash aggregate,
as well as the uncr ushed and crus hed aggr egate mate r ials, is shown in Figure 6. This
graph is representative of materials produced commel'cially at the Corson Plant at
Plymouth Meeting, Pennsylvania.
25
20
A
=:_-_-_·
~
------ ----t--
- -t - -
-----~
~ ;-:~---·-
A
---
~ c
v - - -!-- - - - - -- - - -----l
~-~r_!!s~d-~xtruded Porlictes_(:3!!! :~:.:'!! - - - ----l
B- Extruded Whole Pellets
0
0 5 to
::·~::·I~~-- _-_-~----
15 Minutes 5 Hours 24 Hours
Time as indicated
Figure 6. Moisture absorption of different fly ash aggregate products made from a
single source of fly ash.
27
TABLE 3 With respect to the bulk density of aggre-

P O ZZO LANIC STRENGTH OF FINELY gate produced from fly ash, some of the
GROUND FLY ASH AGGREGATE 3
variation depends on the raw material from
Blaine Pozzolanic which the aggregate was produced. For
Material Under Test Fineness, Index,
Percent of
instance, fly ash with very high fineness
cm'/gm Control and lOW Carbon COntent Will USUally TeSUlt
in a product having the higher densities.
Fly ash from source A 2,450 121
Ground lightweight aggregate The data given in Table 2 illustrate the
made from source A fly ash 2,250 197 effect of aggregate size on bulk density.
Fly as h from source B 5,420 130
Ground lightweight aggregate
Any residual fine fly ash particles that
made from source B fly ash 5,330 151 are either carried through the burning
aM odified test accord ing to ASTM Method C 3 1 1·63 T .
process or are formed by abrasive action
of the aggregate during handling are still
highly pozzolanic and will react in much
the same manner when mixed with portland
cement as the unfired fly ash. This results in the usual long-term improvements in
strength and dimensional stability of the concrete. When the lightweight aggregate pro-
duced from fly ash is pulverized to pass a No. 200 mesh sieve, virtually all of the
OJ:iginal pozzolanic activity is recovered. Table 3 gives data on this effect as measured
with two lightweight aggregate materials from different sources of fly ash. Figure 7
is a close-up of a fractured specimen of concrete containing fly ash aggregate. The
excellent bond established at the interface between the aggregate and the mortar matrix
is typical of results with fly ash material.
Figure 7. Fractured specimen of lightweight concrete containing fly ash aggregate.

28
kg I cm2
250 300 350 400 250 300 350 400
1000 600 600
l
u
0
800
l
....
iii
c'"'.:. '"'e
N
e
u
~ ~ 60 0
u ,s '~ ~
c
'....
"'
c
"'
8.
~- 201) ~ 20
Vf
200 200 15
600 350 900

60
Normal
c
c
.. 300 ~
.;l
a. N
0
u,.,
.... ~400
250 '"'e
'.... -..
"
a:
0
'
e
u
...."'
i :!' 200 "' ..."
..2
0
JOO 20
~
::E
150 Moist
10
200 100
160 1200 1200

:E
"' 2400
"i
;r:
·;:
140 ------ 2200 0
"
"'
1000
Normal Cure
1000
=> :EJ: '°o "'g
... ...
... u 120
-
c a.
' "' " 800
Ii
.!:
< ....
"' ~
0
·"' 'Ii
...
0
100
c
c
0
"
c
m 1400
N
80 400
120 120
....
u
€U) "''g ~0
'el 'o N " "-
0 -
iii Ii
'O ..."a. '....
.?-
:5
2 1.5 "' c
~ 60 Dry Aggregate 60
-60
Air Enlrained
::E 1.0
1.0 40 40
3 4 5 6 3 4 5 6
28 Doy Compressive Slrenglh, ksi
~Sand · Fly Ash Lighlwelghl ~ Sand · Lighlweighl - - --Rtferenco Concrelo
Concrelo ' Cone role
Figure 8. Comparison of lightweight concretes using sand as the fine aggregate.

29
Another important difference between aggregates produced from fly ash is in the
shape of the individual granule. Uncrushed materials, such as those produced by the
Corson process (which makes an extruded shape), have been demonstrated to have some
advantage over the round pellets in improved workability and pumpability of the con-
crete. Crushed materials have been blended with sand and have been found to be effec-
tive in packaged concrete. The overall balance of properties, i.e., density, strength,
and workability, in these latter mixtures is quite good.
PROPERTIES OF HARDENED CONCRETE

Considerable structu~ral concrete has been produced with lightweight aggregate made
from fly ash . In general, the concrete has demonstrated satisfactory performance and
compares very favorably with other forms of lightweight aggregate concrete . A recent
report of Committee 213 of the American Concrete Institute (5) provides a good descrip-
tion of lightweight aggregate concrete and includes fly ash aggi·egate in the total evalua-
tion made.
Figure 8 shows graphs describing concrete properties using fly ash agg1·egates pro-
duced in various plants throughout the world. The Wlshaded bands shown in the graphs
are taken directly from the ACI repol't, as are the dotted lines representing a typical
normal weight concrete. The shaded bands, as superimposed on this backgroWld, rep-
resent the fly ash lightweight aggregate concrete.
The graphs in Figure 8 are based on the use of normal weight natural sand as the
fine aggregate. Most of the companies producing concrete commercially use this type
of composition because of difficulties in workability experienced when lightweight aggre-
gate is used in both the coarse and fine fractions .
Figure 9 shows data on the latte1· type mixtures. Because the source of this infor-
mation is limited to fly ash aggregate c oncrete produced at the Corson Company plant,
the pertinent data a.J:e given in the form of a single line rather U1an in a band, as is the
case in Figure 8. For comparative purposes, Figure 9 also includes the backgi·ound
information as presented by ACI; t11e shaded band l'epresents all lightweight aggregates
and t11e dotted line again represents normal concrete.
A study recently reported by the Portland Cement Association Laboratories (6) pro-
vides additional information on fly ash aggregate concrete. The Portland Cement
Association program has utilized material that has been supplied from a number of
processing plants in this country. An interesting report of studies carried out in
Yugoslavia (7) gives information on the feasibility of using fly ash from lignite coal to
produce acceptable lightweight materials. A cooperative investigation is currently
Wlder way (8) to determine the fire-resistance of concretes produced with fly ash light-
weight aggregate. An early report by Pearson and Asce (9) states that fire tests on
fly ash aggregate concrete conducted for New York City, Using ASTM Method E 119-62,
shows that a nominal-size specimen of 4 by 12 by 60 in. has a fire-resistance of 3
hours.
Figure 10 provides some data on the thermal values for fly ash lightweight aggregate
concrete. Included in this figu1·e are data supplied tlwougb the courtesy of the Centre
Scientifique et Technique du Batiment (10) . The spread in the K- values as plotted re-
flects the conditions present at t11e time the measurements are made. The higher value
of K at each density results primarily from tests of concrete ma.de in a moist condition.
There is considerable interest, in France as well as in other parts of the world, in the
use o! this type of concrete in building prefabricated structures. The very favorable
K-values that are obtained, together with the high strength and other properties of the
fly ash aggregate concrete, enable U1e fabricator to produce exterior walls and other
units of minimum thickness and weight but of adequate strep.gth.
Table 4 gives design criteria that a.re cu1Tently in use for commercial mixtures in
this country. Comparison of t11ese designs with the recommended practice, as developed
by AC! 613A-59, indicate that t11e fly ash aggregate falls well within the ranges of that
study. These design criteria provide adequate factors of safety when compared with
the strength results as shown in Figures 8 and 9.
30
kg I cm 2
250 300 350 400
...,
E
~"' JOO Moist 20

Vf
20 0 ~--~--~----- 200 ~--~--~--__..15
900
60
Normal Cure
c ~
c0" ..., g. N
-
a: 0 E
u..., E u
~400 0
"
L
;; '
3: ' "'
-'<
~ "'
-'<
"O
2 ~
0 300
~ 150 Moist
10
200 100
160 1200 ......--...- --.--- --.----.--.1200

::;:-
"' 2400
---- -- ----- 2200
·; Normal Cure
3:
140
"'"
·;:
,.,
:::> _ 2000 ...,E ~
0
~"'
Q
L U 120 Vl BOO
Cl Q.
' L
"c
c
< ~
·"'
::!
100
""c
Cl
CD
6"
N
BO 400
5.0 120 ....--.---~-~--r--.t20

u
UJ
,., 4.0 u;>
Q
L
0
100
~'°
~ Q
c N " ...
u
c
-
w
0
~
"
·;;
Q.
3.0 ~
"'
~
"'.a"
ao
~
-'<
"
"3
"O
2.0 60
0
~
Air Entrained
1.0
1.0 40 ~--~~--~----"40
3 4 5 6 3 4 5 6
28 Doy Compressive Strength, ksi
o:s::s::s::sl All lighlweight Concrete All Fly Ash Aggregofe - - --Reference Concrete
Concrete
Figure 9. Comparison of concretes using all lightweight aggregate.

31
160
Normal Woighl Concrete
~-- -----
140
M
;:::
',a 120
~
-~
CJ
..
., 100
~
.3 <E-Fly Ash lightweighl Aggregale Concrete
80
60
0 2 4 6 8 10 12
Thermal Conduclivily (K Factor)
Figure 10. Range of thermal conductivities for fly ash aggregate

concrete.
TABLE 4
CONCRETE MIX DATA
Quantities Per Cubic Yard

Air - Weight of Weight of
Aggregate (lb)
Slump Entraining Percent Plastic Concrete,
Design Cement Watera (in.) Agent Air Concrete 28 Days
Mix Corson-
Concrete (oz/sack) (lb/cu ft) (lb/cu ft)
P ounds
Lite,
Sacks Sand, Pounds Gallons
%-in.Size,
S. S. D. c
Bone Dry0
1000 3.35 3 15 950 1,320 441 53 4 l. 50 G 112 106

2000 5.00 470 875 1,300 433 52 4 l. 75 •6 114 109
3000 6.00 564 875 1,240 437 52.5 4 l. 75 6 115 110
4000 7.00 658 875 1,180 441 53 4 l. 75 G 116 111
5000 8. 00 752 875 1, 120 441 53 4 2.00 6 117 112
aTotal water, including water of absorption , ho·as.od on Corson -Lite, 44 lb/cu ft density , cs.S.D. = satu rated su rfa ce dry.
SUMMARY
Fly ash aggregate is being successfully used in structural concrete. Because of the
light weight of the aggregate, it finds application for structural concrete in the range
of 95 to 116 pcf (1,520 to 1,860 kg/ cm 2 ) and may also be used for insulating conc1·ete
at lower densities. With proper quality fly ash, commercial sintering processes
produce various shapes of aggregate similar in properties to other stn1ctural light-
weight aggregates . However, the high rate of absorption and residual pozzolanic prop-
erties provide some advantage over other materials. Tests of concrete using fly ash
aggregate give results falling well within the range of values typical of lightweight con-
crete in general use.
REFERENCES
1. Short, A., and Kinneburgh, W. Lightweight Concrete. John Wiley and Sons, New
York, 1963.
32
2. Minnick, L. J. Corson Company Researches Fly Ash Lightweight Aggregate.

Rock Products, Sept . 1964, pp. 74-77.
3. Sintered Lightweight Aggregate. Central Electricity Generating Board, England,
Sept. 1967.
4. Corson, B. L., and Pfau, J. H. Lightweight Aggregate. U.S. Patent No. 3,341,341,
Sept. 12, 1967.
5. Gujde for Structural Lightweight Aggregate Concr ete . American Concrete Institute
Committee Rept, Jour. American Concrete Ins titute, Proc., Vol. 64, No . 39,
Aug . 1967 , pp. 433-469 .
6. Pfeifer, D. W. F ly Ash Aggregate Ligh tweigh t Concrete . P o1·tland Cement As sn .
Res. and Dev. Bull. , 969.
7. Brzakovie ; Prerlrag . Properties of Aggregate of Fly Ash From Yugoslav Power
Stations and Some Types of Lightweight Concretes. Second Fly Ash Utilization
Symposium, Pittsburgh, Penn., March 1970.
8. National Ash Associa tion. Private communication.
9. P earson, A. S., and As ce , F. Lightweight Aggregate From Fly Ash. Civil Engi-
neer ing, Vol. 43, No . 9, Sept. 1964, pp. 51-53.
10. Croiset, M. Centre Scienlif.ique et Te chnique du Batiment. Private communication .
Fired-Clay Aggregates for Use in Flexible Bases
WILLIAM M. MOORE, Texas Transportation Institute, Texas A&M University
This is the second report issued under a research study dealing

specifically with nonbloated synthetic aggregates produced from
naturally occurring clays. The first report described the re-
sults of a laboratory investigation that indicated that most (if
not all) highly plastic soils could be used to produce high-quality
synthetic aggregates. The present report describes a later in-
vestigation directed toward the development of acceptance cri-
teria that can be used to evaluate aggregates of this type. The
investigation included the production of aggregates in the rotary
kiln, the evaluation of these aggregates, and a study of several
synthetic aggregate flexible bases that have been in service in
the vicinity of Houston, Texas, for several years.
•THIS PROGRESS REPORT is the second part of a study entitled "Synthetic Aggregate
Research" being conducted by the Texas Transportation Instit ute, sponsored by the
Texas Highway Department and the U.S. Bureau of Public Roads. This phase deals
specifically with synthetic aggregates for use in flexible bases, and its primary objec-
tive is to develop acceptance criteria for such aggregates. It was initiated in 1966 after
nonbloated, fired-clay aggregates had been used experimentally in flexible bases on
several Texas highways. To the author's knowledge, fired-clay aggregates are, at
present, the only synthetic aggregates economically feasible for use in flexible bases.
Therefore, all research efforts so far have been directed toward evaluating this type of
aggregate.
The first study done under this phase of the reseaz·ch was an investigation into the
chemical and physical stability of laboratory-produced aggregates. The results were
reported in a paper by Moore et al. (1), which included the following tentative
conclusions: -
1. The clay minerals, montmorillonite, ii lite, and kaolinite, will not rehydrate under atmo-
spheric conditions once they have been completely dehydrated (dehydroxylated); therefore,
once they have been completely dehydrated, they become chemically stabilized for use as high-
way construction materials. Complete dehydration is accomplished by heating the clay and
holding it at the elevated temperature for sufficient time to allow the dehydration to occur. A
period of 15 minutes at 1400°F. was sufficient to completely dehydrate the clay present in the
small, oven dry laboratory specimens made from the Texas soils investigated.
2. Incomplete dehydration of aggregates made by dehydrating clay-type soils can be detected
by a relatively simple laboratory test.
3. Most (if not all) clay-type soils having a relatively high strength when air dried can be fired
to produce hard, durable aggregates suitable for use in flexible base and asphaltic concrete.
The conclusions listed above were based almost entirely on an evaluation of cylin-
drical particles ( h in. in diameter by 11h in. long) that had been fixed in a laboratory
1
muffle furnace (Figs. 1 and 2). Therefore, t he next investigation initiated under this
phase of the research was an evaluation of aggregates produced in a rotary kiln, and
the development of acceptance criteria for such aggregates. The investigation included
the production of aggregates in the Texas Transportation Institute rotary kiln, their
33
34
Figure 2. Specimens being removed from a laboratory

muffle furnace after firing.
Figure 1. Molded specimens ready for firing in a
laboratory muffle furnace.
evaluation, and a study of several synthetic aggregate flexible bases that had been in
service in the vicinity of Houston, Texas, for several years.
All research results to date are favorable to the use of fired-ciay aggregates in
flexible bases. The results indicate that these aggregates have great potential for use
in highway construction in the many areas of the world where high -quality aggregates
no longer exist, but where the clays required for the production of synthetic aggregates
are plentiful.
MATERIALS
As a step in the development of a preliminary recommended criterion for the accep-
tance of flexible base aggregates, the Texas Transportation Institute' s research rotaTy
kiln (Fig. 3) was used to produce several samples of graded aggregates. T11e develop -
ment of th research kiln and its capabilities are des-c.ribed lsewh re (2, 3).
Aggregate samples were produced from three plastic soils obtained from different
sources in B1·azos County, Texas. The Atterberg limits and gradations of the soils
used are given in Table 1. Standard Texas Highway Department test procedures were
tollowed in these determinations. Aggregates were made by firing each of the three
soils at 8 different temperatures; thus, 24 aggregate samples were available for testing.
Each sample consisted of about 2 50 lb of graded material, approximately 9 5 percent of
which was sized between the 1-in. sieve and the No. 10 sieve.
In addition to aggregates made by the Texas Transportation Institute, 4 aggregate
samples that haci bP.P.n produced in commercial rotary kilns for use in flexible bases
were obtained. Two of the 4 aggregates have been successfully used in flexible bases
by the Texas Highway Department. These are referred to as the Hopkins aggregate used
by the Paris district and the Wharton aggregate used by the Houston distl'ict. Two other
samples, designated as the Madison 1 and Madison 2 agg1·egates, were obtained. These
aggregates had been investigated by the Bryan district for possible future use in flexible
bases. The results of laboratory tests conducted by the Bryan district, which were
Figure 3. Texas Transportation lnstitute's research rotary kiln.

35
TABLE 1
MATERIALS USED FOR RESEARCH AGGREGATE PRODUCTION
Graclationa
Raw Liquid Plastic Plasticity
Material Limit Limit Index Percent P e rcent Percent
Sand Silt Clay
Red clay 74 28 46 4 24 72
Gray clay 67 23 44 -{;· 41 54
Black clay 53 19 34 13 34 53
8
Determined by hydrometer analysis (MIT classification: sancl, 2 to 0 06 mm; silt, 60 to 2 microns; and clay,
smaller than 2 microns) .
reported by Long (4), indicated that either of the two aggregates would be suitable for
use in flexible bases.
None of the above aggregates can be classified as lightweight synthetic aggregates.
However, 5 samples of lightweight aggregate-i.e., having a wut weight of less t han 55
pow1ds per cubic foot (pcf )-produced commercially in Texas primarily fo1· use in struc-
tural concrete were obtained. These aggregates are currently under investigation for
porlland cement concrete application in the first pa.rt of this study.
In summary, 33 aggregate samples were investigated; 24 of them were produced in
the TTI research kiln, 2 have been used successfully in flexible bases on Texas high-
ways, 2 have been investigated for possible similar future use, and the r emaining 5 are
lightweight aggregates that are available commercially in Texas.
EVALUATION OF AGGREGATES
The simple testing procedure developed during the initial laboratory studies is shown
in Figures 4 and 5. Basically it consists of the cooking of specimens under water in a
common kitchen-type pressure cooker, and then observing the effect of such treatment
on the particles. This procedure works quite well for laboratory-produced cylindrical
specimens because deterioration caused by the treatment is readily apparent on the
smooth surface of the particles. It was initially thought that the same procedure would
yield significant changes in the gradation of kiln-produced aggregates if they were not
sufficiently dehydroxylated; however, this was found not to be the case. Only minor
changes in gradation wer e observed aitm· the cooking treatment for many aggregates
obviously unsuitable for use in flexible bases. Arter the cooking treatment, these ag-
gregate samples could b e easily crumbled with the finger s .
Several variations of the initial test were tried. A procedure found to be adequate
is given in Appendix A. It is called the pressure-slaking test to distinguish it from the
s laking test (now considered obsolete) described elsewhere (1) and from the slaking
procedures often used in the preparation of soil samples for Tabo1·atory testing. Basi-
cally the pressure-slaking test consists of the same underwater pressure-cooking (Fig.
6 ), but an additional Li:eatment of severe
agitation in water is inserted between the
cooking and the gradation-change measure-
ment. The agitation is accomplished using
Figure 4. Specimens prepared for cooking in a

pressure cooker. Figure 5. Specimens after cooking in pressure cooker.
36
Figure 7. Kiln-produced aggregates subjected to severe

agitation in water by a heavy-duty shaker.
Fiyuri; 6. Bottles containing kiln -produced aggregates

after cooking in pressure cooker. TABLE 2
TEST RESULTS FOR RESEARCH AGGREGATES AND
AGGREGATES PRODUCED IN COMMERCIAL KILNS
Pres sure Los Angele s

Aggregate Sa mple SlakingP Abrasionc
Type Designationa (percent (percent
l oss) wear)
Red clay, RC-1000 66.6 64 .2

nonbloated, RC-1100 38 .9 61.3
produced by RC-1205 20.5 56.6
TT! RC-1295 6.2 50.2
RC-1390 4.5 42.8
RC-1585 2.2 36.1
RC-1800 2.5 36.8
RC-1910 2.8 33.9
Gray clay, GC-1030 66.4 62.2
nonbloated, GC-1095 61.7 66.6
produced by GC-1200 25.6 55 .2
TT! GC-1305 13 .8 61.5
GC-1400 9.8 62.8
GC-1600 6.1 49.4
GC-1800 5.8 52 .7
Figure 8 . Complete samples immediately after cook· GC-1930 4.0 39,6
ing and subsequent agitation in water (numbers refer Black clay, llC-1010 55.9 71.3
to approximate firing temperature, hundreds of nonbloated, BC-1100 37.4 67 .0
degrees F). produced by BC-1205 16.8 64 .9
TT! IJC-1310 9 .3 60.1
BC-1395 6.7 56 .8
BC-1610 6.5 51.7
BC-1800 4.2 37.4
BC-1940 4.2 37 .0
N onbloated, Hopkins 8_3 35_0
produc ed Wh ~ rt.nn 9_7 43.6
commercjally Madison 1 6.9 33.3
Madison 2 7 _7 38 .3
Bloated, R 2.7 27 .5
lightwe ight, s 4.9 22_5
produced c 5.7 40 .4
c ommercially E 2.5 25 .2
D 5.0 23 .1
8The 4-d lgi t numbers ~hc.rwn in some :s:ample des1gmu a ns refer to the maxi -
mum ki ln [O:fll(K'rn u,u o n~·.1 sured c(urlht) sampl e 1uoduction.
bEach prc.')$Uf0-5lak ihg IOS5 ¥alue is lhct trlerage Qf 1hr tests performed in
accordance with Appendix A.
Figure 9. Fractions of samples retained on the No. 40 cEach Los Angeles abrasion va lue is th e result of a single test (ASTM Method
sieve. c 131)
37
70 a standar d labor atory heavy - duty shaker
., RED CLAY AGGREGATES
m GRAY CLAY AGGREGATES (Fig. 7). Typical results of cooking and
~ 60 A BLACK CLAY AGGREGATES gradation are shown in Figures 8 and 9.
One can observe in these figures the rela-
U)
U) 50 tive amounts of disintegration resulting
g from the test.
C> Results of the pressure-slaking test
z 40
~
and the Los Angeles abrasion test for the
<(
...J
24 aggregates produced by TTI as well as
U) 30 the 9 aggregates produced commercially
w are given in Table 2 and shown in Figure
a::
~ 20 10. From the figure it is clear that the
U)
w research aggregates made at about 1,000
a:: F would not be suitable for use in a flexi-
a..
ble base, whereas all those made at firing
temperatures of 1, 400 F and higher com -
pare favorably with the synthetic aggre-
KILN TEMPERATURE (°F) gates produced commercially.
Figure 10. Pressure-slaking loss versus maximum kiln EVALUATION OF FIELD SAMPLES
temperature for research aggregates made from threeDuring 1963 and 1964, the Houston dis-
clays. trict of the Texas Highway Department
constructed several projects totaling about
15 miles in length utilizing synthetic ag-
gregate flexible base. The base material
consisted of a mixture of appro~imately 70 percent aggregate produced from clay in a
rotary kiln and 30 percent field sand. The Wharton aggregate (given in Table 2 for
comparison with the research aggregates) was used in at least one of these projects.
The field sand, taken from several sources, was required to have a liquid limit of less
than 35 and a plasticity index of less than 10. In one of the projects, the base material
was stabilized with lime. According to local engineers, all of the projects are still in
good shape. One short section has been reworked -(necessitated by a bridge grade
change) and the synthetic aggregate flexible base in this section was salvaged andreused.
In the summer of 1968, at 6 different locations several miles apart, sections of these
synthetic aggregate flexible bases were sampled and field-tested (Fig. 11). Samples
from 5 of these locations were prepared for laboratory testing in accordance with stan-
dard Texas Highway Department testing procedures.
As may be seen from the gradations given in Table 3, the amount of soil binder (frac-
tion finer than No. 40 sieve) in the 5 samples prepared for laboratory testing varied
from 22 to 34 percent and averaged 28 per-
cent. When one considers that the kiln-
produced aggregates had from 2 to 5 per-
cent soil binder, and that the construction
specif:i,_cations required that they be mixed
with 30 percent of a field sand with 99 per-
cent soil binder, it appears that there has
been no significant disintegration of syn-
thetic aggregates during the 5 to 6 years
they have been in service.
The pressure-slaking tests (Table 3)
were made on aggregate samples separated
from the base samples by washing over
the No. lOsieve. The loss values obtained
varied from 8.3 to 15.5 percent and aver-
aged 11.0 percent, as compared to a range
of 6.9 to9.7 percent and an average of 8.2 Figure 11. Synthetic flexible bases sampled for
percent for the 4 commercial flexible base laboratory testing after several years of service in the
aggregates sampled from stockpiles Houston area.
38
TABLE 3
'!'i,;8'!' tti,;8UL'l'8 UH'l'ALNi,;lJ ON 8ALVlf'Li,;S Ffi.OLVl HIGHWAYS NEAR HOUSTON
Mois tur e
Dr y Percent Retained by Sieve s Pressure-
Secti on
De ns ity Slaking
No . (percent)
·1~ in.
(pc() No . 4 No . 10 No . 40 Test
2 13 . 5 111. 2 0 29 50 66 12.6
7 15.8 105.9 1 32 60 78 10 .0
8 15.9 90 .6 0 28 54 71 8.7
9 15.8 90 .8 2 34 58 73 15.5
14 15.7 11 1.4 24 49 72 8 .3
15a 16.4 107 .0
a Base material was stabilized with lim e and co uld not be tested
(Table 2). Thus, the material samples from the roadway had a somewhat greater range
and a higher average loss than was the case for the materials sampled from stockpiles .
According to the general trend shown in Figure 10, it appears that the synthetic aggre-
gates taken from the roadway probably were produced at temperatures slightly lower
than those taken from stockpiles.
In summary it can be said that all of the synthetic aggregate bases appear to have
performed satisfactorily to date. The laboratory tests made on the samples that had
been in service for several years indicate that these aggregates had not undergone any
significant disintegration. Thus, it can be concluded-at least for the Houston environ-
ment-that synthetic aggregates suitable for use in flexible base can be produced by fir-
ing highly plastic clays. Such aggregates may or may not be suitable for use in more
severe environments.
In the opinion of the author, 10 percent loss determined by the pressure-slaking test
(Appendix A) is a safe and reasonable upper limit for an acceptance cr iterion for flexible
base synthetic aggregates in Texas to ensure that the aggregates have been fired suf-
ficiently to remain stable in their intended use.
PRESSURE-SLAKING TEST, MOD. 1
In order to utilize existing equipment in Texas Highway Department district labora-
tories, a modification of the pressure-slaking procedure described in Appendix A was
undertaken. The modified procedure is called the "Pressure-Slaking Test (Mod. 1)" and
is described in Appendix B. Basically it consists of the same procedure used in the
pressure-slaking test except that it uses a Tyler sieve shaker (Figs. 12 and 13) for ac-
complishing the agitation in water instead of the heavy-duty shaker (Fig. 7). Comparative
Figure 12. A Tyler sieve shaker used in the modified

pressure-slaking test instead of heavy-duty shaker Figure 13. Container designed to fit shaker for
shown in Figure 7. agitating 5 samples simultaneously.
39
TABLE 4
COMPARATIVE RESULTS OF PRESSURE-SLAKING TEST
VERSUS PRESSURE-SLAKING TEST (MOD . 1)
Pr essure-Slaking Test, Pressur e -Slaking 1'csL (Mod. 1),

Aggregate Percent Lossa Percenl Lossb
De signation
Test 1 Test 2 Test 3 Average Test 1 Test 2 Test 3 Average
RC-1000 60 .3 60.4 79.0 66.6 53 .6 47 .3 43 .3 48.1

RC-1100 37 .4 33.2 46 .0 38 .9 21.9 26 .7 23 .1 23.7
RC-1205 22 .9 18 .7 20.0 20.5 9.4 7 .6 9 .8 8.9
RC-1295 5.9 5.6 7.0 6.2 2 .1 2 .2 2. 5 2.3
RC-1390 4.0 4.4 5.0 4.5 1.5 1.6 1.3 1.5
RC-1585 1.8 1.9 3 .0 2.2 1.3 1.0 0 .8 1.0
RC-1800 2.4 2.1 3 .0 2.5 1.2 0.8 1.0 1.0
RC-1910 2 .3 2.1 4.0 2.8 0 .9 1.2 0.6 0.9
GC-1030 66.6 65.5 67 .2 66.4 54 .3 60 .3 52.6 55.7
GC-1095 62 .7 60.7 61.7 51.6 46 .0 44 .6 47.4
GC-1200 27 .5 26.2 23.1 25.6 13 .0 12.8 14.1 13 .3
GC-1305 12.7 14.9 13.8 5. 8 4 .8 5.6 5.4
GC-1400 10.2 9.2 10 .0 9.8 3 .9 3 .7 4.5 4.0
GC-1600 6 .9 5.5 6 .0 6 .1 1.9 1.9 2 .5 2.1
GC-1800 6 .4 5.1 6 .0 5.8 2 .0 2.1 2 .1 2.1
GC-1930 4.0 3 .1 5.0 4.0 1.9 1.4 1.0 1.4
BC-1010 51.6 52 .2 64 ,0 55.9 35. 2 36.1 32 .4 34.6
BC-1100 36 .0 36 .3 40 .0 37 .4 23 .9 22 .7 19. 5 22.0
BC-1205 17.4 16.1 17 .0 16.8 7 .0 7.4 9 .0 7 .8
BC-1310 8.9 10.0 9.0 9.3 4.1 3.6 3.2 3.6
BC-1395 6 .5 7 .0 6.5 6.7 2 .2 2 .2 2.8 2.4
BC-1610 6. 1 6.4 7 .0 6.5 2 .5 2. 7 2 .6 2 .6
BC-1 800 3 .3 4.3 5.0 4 .2 1.4 1.6 1.8 1.6
BC-1940 4 .1 4.6 4.0 4 .2 0 .9 1.1 1.4 1.1
aAnalysis of variance: within sampl e standard deviation, 3.1 ; within sample CV, 16.0 percent.
b Analysi s of variance : within sample standard deviation , 1.8; within sample CV, 14 .5 percent.
70
60
:.!!
la.-
VI 50
VI
g
(!>
40
z
:;;:
<t
_J
VI 30
w
a:
:::l
VI
VI 20
w o RED CLAY AGGR .
a:
a_ IOGRAY CLAY AGGR.
BLACK CLAY AGGR.
10 "'
20 30 40 50 60
PRESSURE SLAKING (MOD I) LOSS (%)
Figure 14. Comparison between original and modified pressure·

slaking test results.
40
test results obtained using the two different procedures on the 24 aggregate samples
produced by TTI are given iu Table 4. It can be seen from these data that results ob-
tained using the modified test are always lower than those obtained using the original
procedure.
Comparative analyses of variance for the data obtained from the two test procedures
indicate that the repeatability of the two is about the same . A plot of comparative val-
ues is shown in Figure 14. From this it is clear that there is a strong correlation
between the tests; thus, the two are essentially measuring the same property. Specif-
ically, it can be said that the recommended acceptance criterion of 10 percent loss
based on the pressure-slaking test is equivalent to a loss of 4 percent based on the modi-
fied test.
CONCLUSION AND RECOMMENDATIONS

The following findings were reached as a result of the investigation described in this
report:
1. Synthetic aggregates suitable for use in flexible bases in Texas can be produced
in rotary kilns from highly plastic clays. These aggregates may or may not be suitable
for use in locations having a more severe environment than Texas, such as locations
subjected to deep fr ost penetration.
2. Incomplete dehydration of synthetic aggregates made by firing clays can be de-
tected by two relatively simple tests. The procedures for these tests are given in Ap-
pendixes A and B.
3. Test results and service records to date indicate that these aggregates have great
potential for use in highway construction in the many areas of the world where high-
quality aggr egates no l onger exist but wher e the clays required for the production of
synthetic aggregates are plentiful.
Based on the results of this investigation, it is recommended that synthetic aggre-
gates for use in Texas flexible bases have a loss measured by the pressure - slaking test
(Appendix A) of less than 10 percent or a loss measured by the modified pressure-slaking
test (Appendix B) of less than 4 percent.
ACKNOWLEDGMENTS
This research was done by the Texas Transportation Institute, Texas A&M Univer-
sity, in cooperation with the Texas Highway Department, and sponsored jointly by the
Texas Highway Department and the U.S. Bureau of Public Roads. The author wishes to
thank all members of the Institute who assisted in this research. He would like to ex-
press special appreciation to Frank H. Scrivner for his advice and assistance. His
help throughout the study was particularly valuable. Special gratitude is also expressed
to C. H. Michalak and Rudell Poehl for their assistance in data reduction and report
preparation and to Gene Schlieker for his assistance during the testing phases.
The author wishes to acknowledge the guidance and assistance given by the advisory
committee for this study. The committee members include Texas Highway Department
per sonnel Kenneth D. Hankins, Study Contact Repr es entative and Res earch Area Rep-
resentative; H. A. Sandberg, Jr., Materials and Tests Division Representative; and
Louis White, Bridge Division Representative; and Bureau of Public Roads personnel
Edward V. Kristaponis, Division Representative; and W. J. Lindsay, Regional Repre-
sentative.
The opinions, findings, and conclusions expressed in this paper are those of the
author and not necessarily those of the Bureau of Public Roads.
REFERENCES
1. Moore, W. M., Van Pelt, R., Scrivner, F. H., and Kunze, G. W. Suitability of
Synthetic Aggregates Made From Clay-Type Soils for Use in Flexible Base.
Research Rept . 81- 5, Texas Transportation Institute, Texas A&M Univ., Feb.
1969.
41
2. Houston, J. T ., and Ledbetter, W. B. Development of a Rotary Kiln for Synthetic

Aggregate Research. Pit and Quarry, Nov. 1967, pp. 110-112.
3. Houston, J. T. Development of a Research Rotary Kiln and a Study of the Effects
of Degree of Thermal Transformation on Selected Properties of Synthetic Ag-
gregates. PhD dissertation, Texas A&M Univ., Jan. 1969.
4. Long, R. E. A Summary of Base and Subgrade Testing for IH45. Unpublished
report, District Laboratory, District 17, Texas Highway Department, Bryan,
1963.
Appendix A
PRESSURE-SLAKING TEST
Scope
The test method described here is intended to be used to evaluate the amount of de-
hydration that has occurred in the production of synthetic aggregates fired in a rotary kiln.
This procedure is a modification of the procedure previously reported elsewhere (!)·
Apparatus
The apparatus shall consist of the following: (a) pressure cooker (common kitchen-
type with 6-quart capacity and 15-psi pressure regulator); (b) centrifuge bottles, 500
ml Pyrex; (c) balance with 3,000 gram capacity having a sensitivity of 0.1 gram; (d)
heavy-duty i·eciprocating laboratory shaker (Precision Scientific Cat. No. 5855 01· equiv-
alent); (e) sieves (3/4-in., No. 10, and No. 40 sieves that meet the requirements of ASTM
Specification E 11); and {£) drying oven capable of heating to 105 C (220 F).
Sample
An unwashed representative sample of sufficient volume to half fill the centrifuge
bottle should be chosen. The sample material is that which passes a %-in . sieve and
is caught on a No. 10 sieve. Any material retained on the %-in. sieve s.hould be crushed
to pass this sieve using a minimum amount of crushing. Because synthetic aggregates
vary widely as to specific gravity, a volumetric measure of the sample is used rather
than weight.
Procedure
1. Place the sample into the centrifuge bottle, and add 200 ml of distilled water. It
is not necessary to determine the initial weight of the sample. (Repeat for any number
of samples up to as many as can be conveniently placed in the pressure cooker.)
2. Place the centrifuge bottles containing the aggregates in the pressure cooker, add
1
approximately /2 in. of distilled water to the pressure cooker, and seal the lid tightly.
3. Heat the pressure cooker with a large Bunsen burner until full pressure is indi-
cated by the pressure regulator.
4. Adjust flame to allow only a slight escape of steam and maintain pressure for 15
minutes. Remove the Bunsen burner, release the pressure, and remove the centrifuge
bottles.
5. After cooling to approximately 100 F, place corks in the centrifuge bottles and
place the bottles in the laboratory shaker. Shake the aggregates for 15 minutes.
6. After removing the bottles from the shaker, wash the sample over a No. 40sieve,
taking care not to lose any of either -40 or +40 material.
7. Dry both -40 and +40 materi:ll to a constant weight at 105 C (220 F). Because of
rehydration, the final total weight of the sample may be greater than the initial weight.
42
Calculations
The pressure-slaking loss is expressed as the percent passing the No. 40 sieve and
is calculated by the following equation:
Percent Loss = _W_e_i"""g_h_t_o_f_m_m_·_u_s_-_4_0_m_e_sl_1_m_.:..:.at'-e'--r_i...:.c.al x 100

Total weight of material
Appendix B
PRESSURE-SLAKING TEST (MOD. 1)
Scope
This procedure is a modificati on of the pressure-slaking test p1·ocedure given in
Appendix A. The modification was made to better utilize existing equipment in the
Texas Highway Department district laboratories.
Apparatus
The apparatus shall consist of the following: (a) pressure cooker (common kitchen-
type With 6-quart capacity and 15-psi pressure r egulator); (b) centrifuge bottles, 500
ml Pyrex; (c) Tyler sieve shaker (Soiltest Model No. C1-305A or equivalent, cpm =
285 ± 10, throw = 13/.i ± 1/., in.); (d) balance with 3,000 gram capacity having a sensitivity
of 0 .1 gram; (e) stainless steel bucket (fits Tyler sieve shaker) consisting of Bain Marie
pot with cover and beake1· without a pouring lip (8% quart, 8-in . body diameter, 93/.i-in.
depth, 8%-in. over bead diameter, available from Texas Hi~lnvay Department D- 4 stock;
(f) spacer (7%-in. diameter by 2in. thick), rubber cushion (7%-in. diameter by /e in.
1
thick), and miscellaneous rubber sheeting or rags; (g) sieves (%-in ., No. 10, and No. 40
sieves that meet the requirements of ASTM Specification E 11); and (h) drying oven
capable of heating to 105 C (220 F).
Sample
An unwashed representative sample of sufficient volume to half fill the centrifuge
bottle should be chosen. The sample material is that which passes a 3/..-in. sieve and
is caught on a No. 10 sieve. Any material retained on the %-in. sieve should be crushed
to pass this sieve using a miuimum amount of crushing. Because synthetic aggregates
vary widely as to specific gravity, a volumetric measure of the sample is used rather
than weight.
Procedure
1. Place the sample into the centrifuge bottle, and acid 200 ml of distilled water. It
is not necessary to determine the initial weight of the sample. (Repeat for any number
of samples up to as many as can be conveniently placed in the pressure cooker.)
2. Place the centrifuge bottles containing the aggregates into the pressure cooker,
add approximately 1/2 in. of distilled water to the pressure cooker, and seal the lid tightly.
3. Heat the pressure cooker with a large Bunsen burner until full pressure is indi-
cated by the pressure regulator.
4. Adjust flame to allow only a slight escape of steam and maintain pressure for 15
minutes. Remove the Bunsen burner, release the pressure, and remove the centrifuge
bottles.
5. After cooling to approximately 100 F, place corks in the centrifuge bottles and
place the bottles vertically in the stainless steel bucket. (The rubber cushion should be
43
placed beneath the bottles and the rubber sheeting or rags inserted between the bottles
to press them firmly against the side of the bucket.)
6. Place the spacer over the rubber corks in the bottles and fasten the cover to
press the bottles against the bottom of the bucket.
7. Lock the stainless steel bucket in the Tyler sieve shaker and shake the aggre-
gates for 15 minutes.
8. After removing the bottles from the s haker, wash the sample over a No. 40sieve,
taking care not to lose any of either -40 or +40 material.
9. Dry both -40 and +40 material to a constant weight at 105 C (220 F). Because of
rehydration, the final total weight of the sample may be greater than the initial weight.
Calculations
The pressure-slaking (Mod. 1) loss is expressed as the percent passing the No. 40
sieve and is calculated by the following equation:
Percent Loss = Weight of minus - 40 mesh material x 100

Total weight of material
An Evaluation of the Bulk Specific Gravity
for Granular Materials
PRITHVI S. KANDHAL, Berger Associates, Inc., Harrisburg, Pennsylvania; and
DAH- YrnN LEE, Engine el'ing Research Institute, Iowa state University , Ames
Accurate determination of bulk specific gravity of aggregates is of

paramount importa.11c~ Ior successful design of bituminous paving mix-
tures as well as portland cement concrete. The complicating element
in specific gravity determination is the achievement of saturated surface-
dry conditions. The various modifications suggested by several inves-
tigators either offer little improvement or are too elaborate to be of
practical value in the field or common laboratory. In this study, the
bulk specific gravities of crushed aggregates and cylindrical rock cores
from six limestone quarries, one slag, one crushed trap rock, and one
synthetic aggregate (Synopal) were determined by five methods. The
methods are standard ASTM, geometrical measurement, mercury dis-
placement, and two chemical indicator methods proposed by the authors.
The results by the standard ASTM methods, in terms of reasonableness
and repeatability, are compared and evaluated in relation to results by
the other four methods. The procedures of the proposed chemical indi-
cator methods (cobalt chloride and fluorescein sodium salt) are de-
scribed. The repeatability and reproducibility of the cobalt chloride
indicator method are presented.
•BULK SPECIFIC GRAVITY is the ratio of the weight in air of a given volume of a per-
meable material (including both permeable and impermeable voids normal to the mate -
rial) at a stated temperature to the weight in air of an equal volume of distilled water
at a stated temperature. The bulk specific gravity of an aggregate, as defined by the
ASTM, equals the oven-dry weight of the aggregate (A) divided by the sum of the aggre-
gate volume (Vs), the volume of the permeable voids (Vp), and the volume of the im-
permeable voids (Vi) times the unit weight of water (yw):
or
bulk specific gravity = (Vs + v: + Vi) "YVi
A
B - C
where B is the saturated surface-dry weight (in grams) of the material in air and C is
the weight (in grams) of saturated material in water. Those voids that cannot be filled
with water after a 24-hour soaking are referred to as impermeable voids. Voids that
can be filled with water after a 24-hour soaking period are referred to as permeable.
Percent voids is widely used as one of the criteria for the design of bituminous pav-
ing mixtures. The exact determination of bulk specific gravity of the various constitu-
ents of the paving mixture is necessary to calculate the voids properties accurately.
44
45
Bulk specific gravity is also used in design calculations of concrete mixtures. The
specific gravity of aggregates must be known to enable the designer to compute the
amounts of the various ingredients in the mixture. Because almost all batching is now
done on a weight basis, the importance of correctness of its value is readily apparent.
Experience has shown that, for some materials, tJrn results of bulk specific gravity
are ve1·y difficult to l'eproduce. This is mainly caused by the persc:mal element involved
in judging the point at which the w.et aggregate has dried out sufficiently to reach the
"saturated surface-dry condition," which is theoretically the point at which all the sur-
face moisture has gone from the particles but with the permeable pores remaining com-
pletely filled.
ASTM Standard Methods C 127 and C 128 outline means for determining absorption
and bulk specific gravity of aggregates. These standards call for immersion of mate-
rial in water for 24 hours, followed by drying until the surface-dry state is attained.
Coarse aggregates are rolled in an absorbent cloth until all visible films of water are
gone.
Fine aggregates are spread on a pan and exposed to a gentle current of warm air
until a free-flowing condition is reached. The aggregate is then lightly tamped into a
conical mold. If the cone stands when the mold is removed, the fine aggregate is as-
sumed to carry moisture on its surface, and it is dried further. When the cone just be-
gins to slump after removal of the cone, it is assumed to be in a saturated surface-dry
state. In addition to variation caused by individual judgment, the fact that large particles
tend to dry more quickly than small particles may lead to overdrying of the larger par-
ticles, unless the precaution is taken of recognizing and handling these particles sepa-
rately. These variations become significant in the case of aggregates having a rough
and porous surface.
For natural, well-graded fine aggregates, the saturated surface-dry condition is
usually reproducible. The end point is more erratic for crushed fine aggregates be-
cause the angularity of the particles does not permit a definite slump condition as do
the rounded surfaces of natural sands. Besides this, the higher percentage of material
passing the No. 100 sieve also poses a problem in achieving slump condition.
Various attempts have been made in the past to pinpoint the saturated surface-dry
condition of the aggregates to improve the reproducibility of the bulk specific gravity
test results. These include Howard's glass jar method (1, 2), Martin's wet and dry
bulb temperature method (3)' Saxer's absorption time curve procedure (4) I and Hughes
and BaJ1ramian' s saturated-air-drying method ( 5). However, the various modifications
either offer little improvement or are too elaborate to be of practical value in the field
or average laboratory.
Bulk specific gravity has also been determined by simple mensuration; for example,
by use of a micrometer or a dial gage to determine the volume 0f regular shape samples.
When the sample is irregular or particulate it has been determined by mercury dis-
placement under a given pressure, provided the mercury does not wet the surfaces of
the sample. Both methods have been included in this study to assess the comparative
values of bulk specific gravity as obtained from ASTM standard methods. Rock cores
from the same block samples have been used for both these methods to eliminate errors
caused by sample variation.
A new chemical indicator method has been tried in this study. Its adoption is sug-
gested for determining the saturated surface-dry condition of both coarse and fine ag-
gregates. This seems very practical and eliminates most of the personal judgment in-
volved in standard ASTM tests.
OBJECTIVES
The purposes of this investigation are (a) to evaluate the reasonableness and repeat-
ability of the standard ASTM methods for bulk specific gravity, as compared with other
alternative methods; (b) to develop new, simple methods to determine the bulk specific
gravity or the saturated surface-dry condition for granular materials more reliably;
and (c) to evaluate these new methods as compared to the ASTM methods.
46
TABLE 1
LIMESTONE AGGREGATES STUDIED
Number County Quarry Beds or Ledges Geological Formation
Adair Menlo 3-6-Argentine Missourian series,

Pennsylvanian system
Blackhawk Pints Rapid Cedar Valley formation,
Devonian system
Hardin Alden Gilmore City formation,
Mississippian age
4 Scott Linwood Davenport Devonian system
Story Cook St. Louis formation,
Mississippian series
6 Washington Keota Beds 14 Osage series,
through 22 Mississippian age
MATERIALS
Crushed .L\.ggregates
.
Six limestone aggregates that were obtained from various areas in Iowa and that range
from most absorptive to least absorptive in character have been included in the study
for both coarse and fine aggregates. Mercury-intrusion porosity (21-0.05 µ.pore radius
range) varied from 0.90 percent in Linwood limestone to 17.44 percent in Cook limestone,
with va1·ying pore characteristics and surface texture. Besides these, one trap- rock
aggregate and one blast-furnace slag were included to provide variation in water absorp-
tion, pore characteristics, and specific gravity. Subsequently, at a later stage, Synopal
(a synthetic aggregate manufactured in Denmark) was also included because it is sup-
posed to have uniform structure and other characteristics and it could be used in the
comparative study on bulk s pecific gravity. The details of limestone aggregates studied
are given in Table 1. For coarse aggregates, the fraction passing the %-in. sieve and
retained on the No. 4 sieve was used so as to have uniformity in results. For fine ag-
gregates, the fraction passing the No. 4 sieve and retained on the No. 100 sieve was used
(Table 2).
Rock Cores
Rock cores % in. in diameter were drilled from blocks obtained from the respective
limestone quarries to be used for determination of bulk specific gravity. They do not
necessarily represent the beds fro m which the aggregates were procu r ed but have been
used to ensure material homogeneity and to assess the relative values of ASTM bulk
specific gravity in comparison with that obtained from other methods. For uniformity,
the rock cores were drilled at right angles to the natural bedding planes. For geomet-
rical measurements, the ends of the rock cores were trimmed so as to have a true cyl-
inder for accurate measurement of volume.
TABLE 2
GRADATION OF FINE AGGREGATES
Percent Passing
Sieve No.
Menlo Pints Alden Linwood Cook Keota
4 100 100 100 100 100 100

8 20 67 53 60 70 86
30 20 15 12 25 40
50 0.6 8 .9 7.4 4 .9 10 .8 10 .0
100 0.3 0 .5 0.4 0.4 1.1 0 .3
47
Chemical Indicators
For the new methods to determine satur ated surface-dry condition , four chemic al
indicators were t ried: (a) cobalt chloride, (b) fluorescein disodium salt, (c) cupric
chloride, and (d) copper sulfate. Of these, only the first two gave encouraging results
and have been included in this study.
METHODS FOR DETERMINING BULK SPECIFIC GRAVITY
Rock Cores
ASTM Method-ASTM Standard Method C 127 was adopted for 1/2-in. diameter rock
cores to determine bulk specific gravity. The rock cores could be rolled uniformly on
absorbent cloth to achieve a saturated surface-dry condition as defined by ASTM.
Geometrical Mensuration Method-In the geometrical mensuration method each rock
core was weighed to 0.0001 gram and its dimensions were measured with a caliper up
to 0.01 cm. Average diameter and length were calculated after measuring at three or
four places. Duplicate determinations were made for each rock type. Bulk specific
gravity was calculated as weight divided by measured volume.
Mercury Displacement Method-A mercury porosimeter , as used in determination
of mercury-intrusion porosity and pore-size distribution of aggregates, was used to
determine the bulk volume of the rock core at 5 psi. The theory and operation of the ap-
paratus are described elsewhere ( 6, 7). The authors have devised a simpler apparatus,
using a modified 250-ml burette coiuiected to a sample chamber at one end and a mer-
cury reservoir at the other end, to measure bulk specific gravity of aggregates by the
mercury displacement method, but it will not be discussed here.
Crushed Aggregates
ASTM Methods-ASTM Standard Methods C 127 and C 128 were followed for deter-
mination of bulk specific gravity of coarse and fine aggregates respectively, except that
Chapman flasks were used for fine aggregates. The bulk specific gravity (Gb) was cal-
culated as follows:
A
Gb = V - 200
where A = weight in grams of oven-dry sample in air, and

V =combined volume in millimeters of saturated surface-dry sample and water .
Mercury Displacem ent Method- The procedure used is the same as described for
rock cores except that crushed aggregates were used instead in the sample chamber.
Chemical Indicator Methods-With the salt cobalt chloride there is a color change
from the- r ed hexahydrate through the violet monohydrate to the blue anhydrous salt; that
is, anhydrous CoCh is blue whereas C0Cl2 · 6H20 is red. Use is made of this property
of the indicator to pinpoint the saturated surface-dry condition. A 5 percent solution of
cobalt chloride is prepared and the aggregate (coarse or fine) is immersed in this solu-
tion. After 24 hours of immersion, the aggregates are removed from the solution and
spread on a table with a white top or in any wide white enameled tray. The aggregate
may or may not acquire a pink tint, but as soon as drying proceeds and its surface is
dry, it attains a distinct bluish color. This is assumed to be the saturated surface-dry
condition. The remaining procedure is the same as indicated in ASTM Standard Methods
C 127 and C 128. The detailed procedure is described in the Appendix.
In the other chemical indicator method , a solution (0 .5 percent) of fluorescein sodium
salt is prepared and the aggregate (coarse or fine) is immersed in this solution for 24
hours. When the aggregate is taken out of the solution, it acquires a distinct light yellow
color. As the aggregate is spread on a white porcelain surface and is subjected to dry-
ing, the color of the aggregate changes from light yellow to distinct orange, thus achiev-
ing the saturated surface-dry condition. The remaining procedure is the same as inASTM
Standard Methods C 127 and C 128. The detailed procedure is described in the Appendix.
48
TABLE 3 RESULTS AND DISCUSSION

BULK SPECIFIC GRAVITY OF ROCK CORES OF INVESTIGATIONS
Built Specific Gravity by Rock Cores
Core
ASTM
Geome tric al The bulk specific gravity of rock cores
Mer c ury
Me nsu rationwas determined by ASTM Standard Method
Displacement
Method
M!!thod
Me Urnd
C 127, the geometrical mensuration method,
1-Menlo 2 .637 2.728 2.710 and the mercury displacement method. Re-
2- Pints 2.271 2.337 2.312 sults are given in Table 3. It is evident
3-Alden 2 .510 2.58 6 2 .527 from the data that the ASTM standard test
4-Linwood 2.636 2.734 2.694 gives a minimum value whereas the geomet-
5-Cook 2.565 2.634 2.622 ric al mensuration method gives a maximum
6-Keota 2.489 2.592 2. 549 vaiue of bulk specifil: gi·avlty. The values
obtained by three methods have excellent
correlations with each other, which indi-
cates that consistent results have been ob-
tained by using rock cores.
Plots are made of values obtained by the ASTM standard test and those obtained by
geometrical mensuration in Figure 1. The correlation coefficient is 0.9969, which is
significant at the 1 percent level. The equation of the regression line is
Y = -0.0735 + l.0625X
where X = bulk specific gravity by the ASTM method, and
Y = bulk specific gravity by geometrical mensuration.
Bulk specific gravity by geometrical mensuration is greater than that by the ASTM
method. Differences range from 0.07 for light aggregate to 0.09 for heavy aggregate.
A possible reason for this almost constant difference is the moist surface of the rock
cores even after being rolled in absorbent cloth. Because of this, the saturated surface-
dry weight observed is high, resulting in reduced bulk specific gravity. However, bulk
specific gravity by geometrical measurements may give slightly higher results because
of a slight undulation irregularity that is neglected in the measurement of dimensions.
However, it seems that the moist surface plays a predominant role in causing this sig-
nificant difference.
8 2.800 0
0
2. BOOr-- - -- - -- - -- --..- --,
i:
w ...
:i::
~ ~
2.700 z 2.700
z
w 0
~~ 2.600
::E
g 2 .600
z
!E0
~
~ 2.SOO :i 2 .500 r=0 . 9969
:::i
0w y = -0 .0735 + 1.0625 x
~ C>
w
~ 2,400 r = 0.9915 g 2.400
Y=-0.1154+1.0661X
0
!d.
"'<-' e;
a:
~ 2 . 200~-~--~-~--.....__ _,__ __, ~ 2,200..___~_ _ .....__ _,__ _.....__ _,__ __,
5 2.200 2.300 2.400 2.500 2,600 2.700 2.800 ~ 2.200 2.300 2.400 2,500 2.600 2,700 2.800
"' BULK SP. GR. (CORES) ASTM METHOD BULK SP. GR. (CORES) ASTM METHOD
Figure 1. Bulk specific gravity of cores, ASTM Figure 2. Bulk specific gravity of cores, ASTM
method versus geometric mensuration method. method versus mercury displacement method.
49
The correlation coefficient for values obtained by the ASTM standard method and
those by the mercury displacement method is 0.9915 (Fig. 2), which is significant at the
1 percent level. The equation for the regression line is
Y = -0.1154 + 1.0661X
Y = bulk specific gravity by the mercury displacement method.
Specific gravity obtained by the mercury displacement method is higher than that of
the ASTM standard method. The difference ranges from 0.04 for light aggregate to 0.06
for heavy aggregate. Again, a possible reason is the moistness over the surface in the
ASTM standard test.
From the results of these three methods, it thus appears that ASTM standard tests
C 127 and C 128 may underestimate the bulk specific gravity as they take into account ab-
sorption as well as adsorption of the particles. This is indicated by the moistness that
can be seen visually. This discrepancy had been pointed out by Shergold {8) 1 who r ec -
ommended that the aggregate should be exposed to the atmosphere for at least 10 min-
utes before being weighed after the aggregate has been wiped dry with absorbent cloth.
However, this 10-minute time period cannot be adopted in all circumstances because
of the varying air temperature and humidity conditions, but it does indicate the required
improvement of the test.
Hughes and Bahramian (5) also obtained results showing that water absorption values
obtained from the ASTM method were significantly higher than those obtained by their
saturated air method. This further supports the underestimation of bulk specific gravity
by standard test.
TABLE 4 COARSE AGGREGATES
BULK SPECIFIC GRAVITY OF COARSE AGGREGATES Bulk specific gravity of coarse aggre-
BY VARIOUS METHODS gates was determined by four methods:
CoCl 2 FSS
Mercury ASTM Standard Method C 127, chemical
ASTM
Aggregatea
Method Method Method
Displacement indicator method using cobalt chloride,
Method
chemical indicator method using fluores-
Cook cein sodium salt, and mercury displace-
1-S-B 2.567 2.601 2.615 2.560 ment method. Average results based on
1-L-A 2.613 2.556
1-L-B 2.603 2. 526 duplicate determinations are given in
Pints
Table 4.
2-A 2.348 2.321 Bulk specific gravity values determined
2-B 2. 333 2.356 2.337 2.364 by the ASTM method have been plotted in
Alden Figure 3 against those obtained by the mer-
3-S-B 2.4 75 2.537 2.487 2.496
3-L-A 2.5 17 2. 546 cury displacement method for 15 coarse
3-L-B 2.508 2. 555 aggregates. Correlation is very good-
Linwood value of the correlation coefficient is
4-A 2.6 13 2.644 0.9924, which is significant at the 1 per-
4-B 2.582 2.620 2.631 2.658
cent level. The equation of the regression
Cook line is
5-S-B- 2.397 2.445 2.437 2.480
5-L-A 2.426 2.447
5-L-B 2.408 2.481 Y = 0.1738 + 0.9358X
Keota where X = bulk specific gravity by the
6-A 2 .3 26 2.322
6-B 2.263 2.270 2.250 2.257
ASTM method, and
Y bulk specific gravity by the
Slag
A 2 .668 mercury displacement method.
B 2.656 2.639 2.654
As in cores, bulk specific gravity val-
Trap
A 2.8 22 ues obtained by mercury displacement are
B 2.857 2.900 2.903 higher than those obtained by the ASTM
Syn opal method.
B 2.149 2.180 Values of bulk specific gravity obtained
a A = passing l in. sieve and retained on 3/e in . sieve. 8 = passing 3/8 in. sieve by using cobalt chloride (Table 3) as the
and retained on No, 4 sieve. indicator of saturated surface-dry condition
50
0
0 2 .9(1(1
i= 2 .700 ~---------------..
:E.... 0 "'
0
0 2,800 _,....coCl2 METHOD
z ....
:I: r ~ 0,9913
~ 2.600 ~
..... 2.700 - <:>of SS METHO D
~ r = 0 . 9224 0
l5 rR0 , 9946
~ 0
~
~ 2 .500 ~ 2.600
u
""
::> I
j;1
-::. 2,500
:EI 2.400
y = 0.1738 + 0. 9358 x !:l
< ,,;. 2.400
('.)
!:l a..
~ 2.300 ;'.: 2.300
.....
::>
e; "' 2. 200 =--=-*""-----='""""'~=-::-+.=-.,,..-.!:""'""...,,....,"="~
~ 2.200 '----'----~--~--~-~ 2.200 2 . 300 2 .400 2 ,500 2. 600 2.700 2 .800 2 .900
"' 2.200 2.300 2.400 2.500 2.600 2.700 BULK SP, GR. (C. A.) -ASTM METHOD
BULK SP. GR. (C. A .)-ASTM METHOD
Figure 4. Bu!k specific gravity of coarse aggregates.

ASTM method versus chemical dye method.
Figure 3. Bulk specific gravity of coarse aggregates,
ASTM method versus mercury displacement method.
are higher than those obtained by the ASTM method. The difference is about 0.03 for
the entire range of aggregates . Linear correlation is excellent (Fig. 4)-the correlation
coefficient is 0.9913, which is significant at the 1 percent level.
The striking constancy of difference in the entire range of light to heavy aggregates,
except slag, seems to point out that the chemical method takes care of the adsorbed
water on the surface of the aggregates, because only anhydrous C0Cl2 can have a blue
color while CoCb · 6H20 is red. Because of the disappearance of adsorbed water, the
weight of saturated surface-dry aggregate in air is less, which gives an increased value
of bulk specific gravity. It seems that the chemical method gives realistic bulk specific
gravity values. It is further supported when these values obtained from the C0Cl2
method are compared with those obtained
bythe mercury·displacement method(Fig.
5). Correlation is excellent-the value of 2 . 700 ~-------------~
the correlation coefficient is 0. 9 671, which -o-CoCl2 METH OD

is significant at the 1 percent level. The "' r = 0 . 9671
Q
0
discrepancy in values of bulk specific :I:
~ 2.600 _,,..F SS METHOD
gravity in the case of slag obtained by the r = 0.9688
ASTM and chemical methods can probably
be attributed to the cellular nature of slags, ~ 2, 500
5
w
which -retain internal water not held by :I:
u
capillary forces. I
As in the case of cobalt chloride, flu- < 2 ,400
orescein sodium salt has also indicated
the saturated surface-dry condition of the
aggregates fairly well. The values of bulk e; 2 , 300
specific gravity obtained by using this in-
dicator are shown. in Figure 4 against those "'::>.....
obtained by the ASTM standard method. "' 2.200 ~-~!----o-~--~--"'"=---z---o'
2.200 2 .300 2 .400 2.500 2.600 2.700
The correlation coefficient is 0.9946, !ULK SP. (C. A ,)-ME~CURY DISPLACEMENT METHOD
which is significant at the 1 percent level.
The values obtained from the fluores-cein Figure 5. Bulk specific gravity of coarse aggregates,
sodium salt (FSS) method also have excel- mercury displacement method versus chemical dye
lent linear correlation with those obtained method.
51
TABLE 5 2.700 . - - -- -- - - - - - -- ------,
BULK SPECIFIC GRAVITY OF FINE AGGREGATES CoCl2 METHOD - o -
c FSS METHOD --6--
0
ASTM CoCL, FSS :i::
Aggregate :;; 2.6()0
Method Method Method ::E
Cook 2.551 2.553 2 .535
5
P ints 2.284 2.2 88 2.295 ~ 2. 500
:i::
Alden 2.498 2.480 2.492 u
I y ~ 0.1831+0,9237 x
Linwood 2.534 2. 516 2. 536
Cook 2.395 2.420 2.408 <. 2 , 400
r = 0,9862
!:-
Keota 2.426 2.409 2 .401
""l!>
e; 2. 300
by mercury displacement. The linear cor- "...--'

~
relation coefficient is 0.9688, which is sig- 2.200'='::---=-~---=-"::':----:-"---'----'

nificant at the 1 percent level. Comparison 2,200 2.300 2.400 2.500 2.600 2.700
BULK SP. GR. (F. A.}-ASTM METHOD
of the two chemical methods (Fig. 4) shows
that the two indicators give practically
identical results for an average weight ag- Figure 6. Bulk specific gravity of fine aggregates,
gregate. ASTM method versus chemical dye method.
Fine Aggregates
Results of bulk specific gravity tests
on fine aggregates (average value based on duplicate determinations) are given in Table
5. Figure 6 shows the linear correlations of ASTM bulk specific gravity with values ob-
tained by using the two chemical methods.
With cobalt chloride, the correlation coefficient is 0. 98 62, which is significant at the
1 percent level, and the equation for the regression line is
Y = 0.1831 + 0.9237X
where X bulk specific gravity by the ASTM method, and
y bulk specific gravity by the CoClz method.
As in the case of coarse aggregates, values obtained by the CoClz method are higher
than those obtained by the ASTM method. The difference is about 0.01 for the entire
range of aggregates and is smaller than the 0.03 observed in the case of coarse
aggregates.
TABL E 6
DIFFERENCES BETWEEN DUPLICATE DETERMINATIONS OF BULK
SPECIFIC GRAVITY OF COARSE AGGREGATES
Coarse Aggregate ASTM Method CoCl, Method FSS Method
Menlo 0 .022 0.005 0 .00 8

Pints 0 .001 0.005 0 .003
Alden 0 .005 0.006 0 .004
Linwood 0 .022 0.008 0 .011
Cook 0.006 0.003 0.007
Keota 0 .019 0.009 0 .012
Trap 0 .014 0.004 0 .019
Slag 0 .020 0.022 0 .015
Average difference 0 .014 0.008 0 .010
Range 0 .001 to 0.022 0.003 to 0.022 0.00 3 to 0.019
Average difference,
excluding slag 0 .012 0.006 0.009
Range, excluding slag 0.001 to 0.022 0.003 to 0.009 0 .003 to 0.019
52
With fluorescein sodium salt, the correlation coefficient is 0.9897, which is signifi-
cant at the 1 percent level. The equation for the regression line is
Y = 0.1801 + 0.9250X
Y = bulk specific gravity by the FSS method.
The relationship is similar to that observed in the case of cobalt chloride.
Repeatability and Reproducibility

Coarse Aggregate-Table 6 gives the differences between duplicate determinations
for each coarse a!'.!'.regate for three methods, that is, the ASTM standard method a.TJ.d
the two chemical methods . It ii, evirient from Table 6 that probably both the ASTM
method and the chemical methods are not suited for determination of bulk specific grav-
ity of slag.
From the data in Table 6 obtained from limited study, it appears that both the CoC1 2
method and the FSS method give better duplicate results than the ASTM standard test.
The C0Cl2 method in particular gives duplicate determinations that check within 0.01,
compared with the 0.02 provided in the case of the ASTM standard test, and is recom-
n1ended for fu1ther investigations to establish its repeatability a...'1.d reproducibility.
To compare the reproducibility of the standard ASTM method and the proposed CoCb
method of bulk specific gravity determination, experiments were conducted on 6 mate-
rials involving three experienced operators. Duplicate determinations were made on all
materials by all operators and with both methods. The results are given in Table 7.
Average values based on the duplicate determinations have been calculated. Between
operators, the maximum difference observed in value of bulk specific gravity by the
CoCh method is 0.006, whereas for the ASTM method it is 0.026 (Table 7). Average
variation between operators is 0.005 in the case of the CoCh method and 0.019 for the
ASTM standard test. In all cases, bulk specific gravity by the ASTM method is less
than that by the CoCh method The difference is greater, in the case of absorptive ag-
gregates (2-B and 3-S-B).
Fine Aggregate-Table 8 gives the differences between duplicate determinations for
each fine aggregate for the ASTM standard method and the two chemical methods. The
data given in Table 8 also indicate the better performance of chemical methods, which
TABLE 7
RESULTS OF REPRODUCIBILITY STUDIES FOR COARSE AGGREGATES,
VALUES OF BULK SPECIFIC GRAVITY
Maximum
Aggregate Operator L Operator D Operator K Variation
Between
Operators
1-S-B
2.615 2.608 2.605
CoCI, 2.600 > 2.608 2. 599 > 2.604 2. 598 > 2.602 0.006
2. 576 2. 598 2.590
ASTM 2. 577 > 2.577 2.588 > 2. 589 2.592 > 2. 590 0.013
2-B
2.404 2.402 2.401
CoC! 2 2.417 > 2 .4 11 2.415 > 2.409 2.414 > 2 .4° 8 0.003
2.318 2.334 2.335
ASTM 2.317 > 2.318 2.339 > 2.337 2 . 3~1 > 2.333 0.019
3-S-B
CoC! 2 2. 529 2.530 2. 518
2. 522 > 2.526 2.524 > 2. 527 2.523 > 2. 521 0 .006
ASTM 2.446 2.475 2.472

2.452 > 2 .4 49 2.474 > 2.475 2.475 > 2.474 0 .026
5-S-B
CoC! 2 2. 554 2. 560
2. 550 > 2.552 2. 552 > 2.556 Not done 0.004
ASTM
53
TABLE 8
DIFFERENCES BETWEEN DUPLICATE DETERMINATIONS OF BULK
SPECIFIC GRAVITY OF FINE AGGREGATES
Difference
Fine Aggregate
ASTM Me thod CoC1 2 Method FSS Method
Menlo 0.014 0.007 0.002

Pints 0 .000 0.008 0.001
Alden 0.004 0 .009 0 .002
Linwood 0.007 0.014 0.013
Cook 0 .008 0.005 0.006
Keota 0.021 0 .004 0.008
Average difference 0.009 0.008 0.005
Range 0 .000 to 0 .021 0 .004 to 0.014 0 .001 to 0 .013
TABLE 9
RESULTS OF REPRODUCIBILITY ST tJ OIES FOR FINE AGGREGATES ,
VALU ES OF BULK SPECIFIC GRAVITY
Maximum
Aggregate Ope rator L Opera tor D Operator K Variation
Between
Operators
1- S-C
2.574 2.637 2.552
CoC1 2 2. 573 > 2. 574 2.634 > 2.636 2 .551 > 2.552 0.084
2.571 2.569 2.560
ASTM 2. 561 > 2. 566 2. 568 > 2. 569 2. 557 > 2. 559 0.010
3-S-C
2.466 2.522 2. 510
CoC1 2 2.464 > 2 .4 55 2.519 > 2. 521 2 .507 > 2. 50 9 0.056
2.481 2. 553 2.491
ASTM 2.486 > 2 .4 84 2.542 > 2. 548 2 .486 > 2.489 0 .064
may give duplicate determinations checking within 0.01. Because CoCb is very inex-
pensi ve in comparison with fluorescein s odium salt, the former is preferred .
Between operators, the maximum difference obser ved in value of bulk specific grav-
ity by the CoC1 2 method is 0 .084, wher eas for the ASTM method it is 0.064 (Table 9).
It seems that over drying occur r ed in some of the tests, resulting in higher bulk specific
gravity.
SUMMARY AND CONCLUSIONS

The following conclusions can be drawn from this study, wherein a new method has
been proposed:
1. Investigations have shown that the ASTM standard tests underestimate the bulk
specific gravity of aggregates because these tests measure adsorption as well as ab-
s orption of particles . Results obtained in determination of bulk s pecific gravity by othe·r
methods, such as the mer cury displacement method, the geometrical mensu ration method,
and the chemical methods appear t o confirm this statement.
2. Experiments on rock cores indicate that the mercury displacement method gives
realistic values in between those obtained fro m the ASTM standai·d method and t he geo -
metrical mensuration method. Further inves tigations are needed t o establish a proper ly
specified pressure at which the measurements of volume displaced by mercur y s hould
be taken for consistent results with all aggr egates .
3. Chemical methods eliminate the hum:an element to a great extent in obs erving the
saturated surface-dry condition of the aggregates because color change is quite apparent.
54
4. Chemical methods achieve the elimination of the moistness over the aggregates
and thus tend to give realistic values of bulk specific gravity. Allainuuml of satu1·ated
surface-dry condition in the case of fine aggregates by the chemical method is not af-
fected by the interlocking of the particles.
5. Results of chemical methods agree well with those obtained from the mercury
displacement method.
6. The cobalt chloride method in general appears to be the preferable of the two
chemical methods investigated, considering the data in this study and the relative cost.
However, fluorescein sodium salt is better suited when dealing with dark-colored ag-
gregates.
7. Data in this limited study seem to indicate that most duplicate determinations
check within 0.01 in the case of the chemical method compared with the 0.02 specified
in the ASTM standard test, although extensive tests should be made in various labora-
tories to establish the reproducibility and limitations of this method.
ACKNOWLEDG:MENTS
This report is the partial result of a study at the Engineering Research Institute,
Iowa State University, under the sponsorship of the Iowa Highway Research Board and
the Iowa State Highway Commission.
REFERENCES
1. Howa1·d, E. L. Discussion, Proc. Am. Concrete Inst., Vol. 54, 1958, p. 1215.
2. Peai-son, J. C. A Simple Titration Method for Determining the Absorption of Fine
Aggregate. Rock Products, Vol. 32, 1929, p. 64.
3. Martin, J. R. Two Years of Highway Researcl1 at Oklahoma A. and M. Proc. AAPT,
Vol. 19, 1950, pp. 41-54.
4. Saxer, E. L. A Direct Method of Determining Absorption and Specific Gravity of
Aggregates. Rock Products, Vol. 87, 1956, pp. 77-79.
5. Hughes, B. P., and Bahramian, B. An Accurate Laboratory Test for Determining
the Absorption of Aggregates. Materials Research and Standards, 1967 1 pp.
18-23.
6. Lee, D. Y. The Relationships Between Physical and Chemical Properties of Aggre-
gates and Their Asphalt Absorption. Proc. AAPT, Vol. 38, pp. 242-275, 1969.
. "
7. Hiltrop, C. L., and Lemish, J. Relationship of Pore_-Size Distribution and Other
Rock Properties to the Serviceability of Some Carbonate Aggregates. HRB Bull.
239, 1960, pp. 1-23.
8. Shergold, F. A. Test for the Apparent Specific Gravity and Absorption of Coarse
Aggregates. Jour . Appl. Chem., Vol. 3, Pt. 3, 1953, pp. 110- 117.
Appendix
DETERMINATION OF BULK SPECIFIC GRAVITY OF AGGREGATES
BY CHEMICAL INDICATOR METHODS
Cobalt Chloride Method
The procedure is the same as for the ASTM Standard Methods C 127 and C 128 ex-
cept for the following points:
1. The sample is immersed in a 5 percent solution of cobalt chloride instead of plain
water for 24 hours.
2. The sample is removed from the solution and put on a table, which should have
a white top, or it can be put in a wide, white enameled tray. The aggregate may not
appear to be colored, but the visible water films on the aggregate will be pinkish in
color.
55
3. The sample is spread out to have a layer of individual particles exposed to a

~ently moving current of warm air. It is stirred frequently to secure uniform drying.
Some of the solution sticking to the porcelain will appear to be pink.
4. As drying proceeds, the aggregate attains a bluish color. The aggregate should
be turned over frequently by gentle hand action. If there is still some moisture on the
surface of the aggregate, the bluish color on the porcelain will change back to pink.
Keep turning the sample, while exposing it to warm air, until all the aggregate is bluish
in color and the bluish spots on the white porcelain no longer turn pink. This is assumed
to be the saturated surface-dry condition.
5. The saturated surface-dry coarse aggregate is immediately weighed in air and
water as outlined in ASTM Method C 127, while the fine aggregate is introduced into the
Chapman's flask, w1'ich has been filled with water to the 200 ml mark.
Fluorescein Sodium Salt Method

The procedure is essentially the same as the cobalt chloride method. In this case
a 0. 5 percent solution of fluorescein sodium salt is used. The sample is dried in the
same manner as in the cobalt chloride test. The sample acquires a yellowish color
when taken from the solution after 24 hours of immersion. On drying, an orange color
appears on the aggregates; the porcelain also changes to a distinct orange. This is as-
sumed to be the saturated surface-dry condition, and the sample is further treated ac-
cording to the ASTM C 127 and C 128 test procedures.

Record: Highway Research

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Record: Highway Research

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Record: Highway Research

Uploaded by

Copyright:

Available Formats

HIGHWAY

HIGHWAY RESEARCH BOARD

Highway Research Board

R. E. Bollen and W. G. Gunderman

GENERAL MATERIALS DIVISION

COMMITTEE ON MINERAL AGGREGATES

William P. Chamberlin Donald R. Lamb R. D. Shumway

CONSTRUCTION SPECIFICATIONS FOR VOLCANIC

AN EVALUATION OF THE BULK

The increasing amounts of waste glass being generated in the United

Percent in Each Class

'h in. '/s in . 101 93 0 4 3

where P percent passing a sieve having an opening of d inches,

Percent Unit Percent

First Trial Mix-Aggregate Gradation No . 1

4.5 140.7 3.88 13.89 820 15 .8

Second Trial Mix-Aggregate Gradation No . 2

4.0 137 .6 6.57 15 .36 1061 7 .6

Figure 2. Marshall test property curves, first trial mix.

TABLE 6 again, 5 specimens were made for each as-

Figure 3. Marshall test property curves, second trial mix.

Sieve Uncompacted Gradation of Ext ra cted Aggregate,

First Mix Design Series 4.5 5.0 5.5 6.0 6.5 7 .o

ANTICIPATED FURTHER RESEARCH

140 Fin an air bath, and comparative

WARD R. MALISCH, DELBERT E. DAY, and BOBBY G. WIXSON, Closure-The authors

The purpose of this study was to determine parameters that in-

Current Practice and Experience

CORRELATION OF CINDER PROPERTIES WITH

Road- and Pit-Sampling Program

Figure 4. Study location map.

Correlation With Performance Index

Standard Correlation With

Road Test Variables

Sand equivalent 49.4 24.0 -0 .130

Pit Source Variables

Los Angeles abrasion, percent 46.7 10.6 -0 .029

The multiple regression equation is as follows:

PI= - 5.65 + 0.06(P) + 0.05(RC) + 0.08(0A-H) - 0.04(0A-20) + 0.06(PP1.5)

- 0.07(PP4) - 0.24(Color) + 0.09(PP10) - 0.02(PUW) + + 0.30(CC)

Recommended Specifications and Field Quality Controls

Sieve Size Percent Passing

Lightweight concrete aggregate is being produced commercially from pulverized

Sample Sourc e LOI Si02 Al,O, Fe20:i Cao MgO so,

flux and substantially influences the

SHAPE AND SIZE OF FLY ASH AGGREGATE

Figure 2. Extruded fly ash aggregate (crushed and graded).

PROPERTIES OF FLY ASH AGGREGATE

Figure 5. Close-up of a sawed particle of fly ash aggregate enlarged 10 times.

~-~r_!!s~d-~xtruded Porlictes_(:3!!! :~:.:'!! - - - ----l

B- Extruded Whole Pellets

TABLE 3 With respect to the bulk density of aggre-

Figure 7. Fractured specimen of lightweight concrete containing fly ash aggregate.

600 350 900

160 1200 1200

Figure 8. Comparison of lightweight concretes using sand as the fine aggregate.

PROPERTIES OF HARDENED CONCRETE

~"' JOO Moist 20