Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
Advances in
Concrete Technology
Proceedings
Second CANMET/ACI International Symposium
Las Vegas, Nevada, USA, 1995
Editor
V. M. Malhotra SP-154
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 Obb2949 0522392 9LT
The papers in this volume have been reviewed under Institute publication
procedures by individuals expert in the subject areas of the papers.
Copyright O 1995
AMERICAN CONCRETE INSTITUTE
P.O. Box 19150, Redford Station
Detroit, Michigan 48219
All rights reserved including rights of reproduction and use in any form or
by any means, including the making of copies by any photo process, or by any
electronic or mechanical device, printed or written or oral, or recording for
sound or visual reproduction or for use in any knowledge or retrieval system
or device, unless permission in writing is obtained from the copyright
proprietors.
PREFACE
iii
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-354 95 m Obb2949 0522394 792 m
Helayne Beavers of AC1 and her staff deserve very special mention
for their assistance in getting this publication ready on time for distribution
in Las Vegas, in view of the fact that the final revised manuscripts were
delivered to AC1 Headquarters only on March 9, 1995.
V. M. Malhotra, Editor
Chairman, Second CANMET/ACI International
Symposium on Advances in Concrete Technology
June 1995
June 11-14,1995
INTERNATIONAL SYMPOSIUM
ORGANIZING COMMITTEE
Chairman Secretary-Treasurer
Mohan Malhotra Harry Wilson
CANMET Consultant
Ottawa, ON Ottawa, ON
Members
iv
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-1.54 95 m 0662949 0522395 629 m
CONTENTS
V
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 m O662949 0522396 565 m
vi
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 W 0662949 0522397 4TL W
viii
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 75 IObb2747 0522377 274
SP 154-1
1
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 IOb62949 0522200 816 =
2 Bremner and Holm
INTRODUCTION
What is new and purported to be better does not always stand the test
of time. However, with lightweight concrete most of the desirable properties
Those fired clay pieces differ from those produced in a modern rotary
kiln producing expanded clay aggregates in the degree of expansion
developed by differing firing temperature regimes. Both expansion processes
would normally use essentially the same maximum temperature. With brick
manufacture, if the clay is heated too fast a vesicular expanded dimension
brick not suitable for masonry construction is produced. It took almost two
millennia before a materials pioneer in Kansas, USA recognized the potential
of this supposedly inappropriate material. The suitability and long term
stability of vitrified clay in concrete construction is verified by the success
of this work which should be reassuring to modern engineers. Also,
knowing of this good performance would have been of particular interest to
the founder of the rotary kiln expanded shale industry some eight decades
ago, for no doubt this question was raised when his product was first put on
the market.
Prior to World War I, concrete ships had been built and proved
functional and durable. The first concrete boat reinforced with steel was a
small rowboat built by LaInbot in 1848; one of his two row boats is on
display in the museum at Brignoles in the south of France, the other is in
storage at the Musée d'Orsay in Paris. These concrete row boats, though
functional, served no commercial role. In World War I, the U.S. Marine
Commission, charged with finding an alternative to the steel ships, found that
only structural grade lightweight concrete could be a viable material. Initial
construction experience showed that only a rotary kiln produced aggregate
with high cement content concrete incorporating chemical admixtures could
meet their needs. Just such a ship, the U.S.S. Selma, was launched in June
1919. This ship although subject to various marine disasters such as being
run aground on a stone sea wall while fully loaded with oil, had a
satisfactory service record and, when inspected in 1953 and again in 1988,
the concrete was found to be in exceptionally good condition. This is in
marked contrast to the Selma's predecessor, the USS Atlantis that was
launched some nine months previous. The Atlantis was made of relatively
low strength concrete (20-30 MPa) and with a lower quality aggregate made
in a bee hive kiln rather than in a rotary kiln. The Selma's rotary kiln
produced aggregate, plus enhanced concrete strength (50 MPa) as well as
special attention paid to the consistency of the concrete (the "slump cone"
was developed and first used during the construction of the Selma) resulted
in an advancement in technology equal to what is now being claimed for
HPC in concrete bridge deck construction (8).
As in the first world war, all ships built during war time were found
unsuitable for peace time use. The steel ships were broken up for scrap and
the concrete ships were usually used for piers and breakwaters. The authors
have visited many of these ships in the last decade and have found them all
in essentially good condition considering the use to which they have been
put.
NATURE OF HPLC
The closed cell foam is characterized by being rigid with low water
permeability. With an open cell foam, a flexible and highly permeable
material results. Unlike cellular plastics, the vesicules in aggregates are
formed at very high temperature and upon cooling the gases contract inside
the pores with the result that with a normal degree of expansion there appears
to be some connectivity between individual pores. These conduits are not
easily seen in a scanning electron microscope even at magnifications in
excess of 3000 indicating that if there are pores, they either self-seal at high
temperature or are of dimensions less than 30 nm (14).
effect on slump loss during mixing and placing and renders most concrete
mixtures pumpable. Achieving this level of aggregate moisture content is
best accomplished at the production plant and is not a field problem.
Water contained within the aggregate increases the early age concretes
vulnerability to damage due to freezing and thawing and to spalling in a fire.
Special precautions need to be taken with HPLC as the cement paste is
especially impermeable as the result of very low W/B and the possible use
of silica fume, fly ash and ground granulated blast furnace slag cement.
Concretes containing high moisture content lightweight aggregates need to be
allowed to dry for at least 7 days before being exposed to freezing and
thawing. Longer times are needed for the case of fire resistance where air
drying of both normal weight and lightweight full sized structural elements
can take several years if explosive spalling is to be avoided. The
incorporation of a small amount (1-2%) of polypropylene fibers has been
found to essentially eliminate the spalling tendency of highly saturated high
strength lightweight concrete when subjected to rapid temperature increases
such as is likely to be experienced by oil production facilities (13).
gas and chloride ion permeability of the concrete. Normal weight concrete
can approach the desirable condition where there are no stress concentrations
at the aggregate cement paste interface, however this is possible only at ultra
high strength levels. HPLC that is exposed to effective moist curing can be
programmed to be elastically compatible by either controlling the degree of
expansion of the aggregate or by proportioning concrete mixtures so that the
matrix will have a stiffness at maturity that matches the aggregate inclusion.
PROPERTIES OF HPLC
C
Strength and Stiffness of a Vesicular Aggregate
E = 0.008 pz
Holm (16) reports that, for typical North American structural grade
lightweight aggregates, the dry particle densities range from 1.2 to 1.5 (1200
to 1500 kg/m3). At these densities, the modulus of elasticity of the individual
particles of lightweight aggregates “approaches that measured on the mortar
insulating wall made with NC can be replaced with LC masonry units thus
significantly reducing heat flow and resulting in a high performance type of
constmction.
PRODUCTION OF HPLC
FUTUREROLEOFHPLC
CONCLUDING REMARKS
From the above it may be observed that the current, commonly held
assumption that the expression High Strength and High Performance are
synonymous is an unduly restrictive and sometimes erroneous linkage.
Indeed High Performance need not be high strength and high strengths
achieved without adequate concern for other qualities may not be high
performance (e.g. cracking).
REFERENCES
9. "Bridge Deck Survey", Expanded Shale, Clay and Slate Institute, Salt
Lake City, Utah, p. 24, 1960.
10. Strongin, N.S. and Baulin, D.K., "Lightweight Concrete Structures for
Large-Panelled Residential Buildings", published by Stroyizdat, 184 p . ,
1984. (In Russian)
11. Hoff, G.C., "High Strength Lightweight Aggregate Concrete for Arctic
Applications - Parts 1, 2 and 3", Structural Lightweight Aggregate
Concrete Performance, (Editors: T. A. Holn and A.M. Vaysburd),
AC1 SP-136, pp. 1-245, 1992.
13. Jakobsen, B., "The Status of LWA Concrete and its use in Offshore
Concrete Floating Platforms", presented at the FIP Symposium in
Kyoto, Japan, and published in FIP Notes No. 3, pp. 10-14, 1994.
15. Bremner, T.W. and Holm, T.A., "Elastic Compatibility and the
Behaviour of Concrete", Journal of the American Concrete Institute,
Vol. 83, No. 2, pp. 244-250, MarchiApril 1986.
17. Gibson, L.J. and Ashby, M.F., Cellular Solids - Structure and
Properties, Pergamon Press, p. 357, 1988.
SP 154-2
SYNOPSIS
Thirteen beams, 8 in. x 12 in. (200 x 300 mm) in cross-section and having a
9.0 ft ( 2.74 m) span were tested to failure. The embedded prestressed prisms
had a length of 9 ft. 6 in. (2.90 m) and cross-sectional dimensions ranging
between 1.5 in. x 3.0 in. ( 38 mm x 76 mm) and 4.5 in. x 3.0 in. (1 14 mm x
76 mm). The prisms were prestressed with 7-wire 3/8 in. (10 mm) diameter, 270
ksi (1860 MPa)tendons. Concrete strength in both the prisms and the beams was
in excess of 14,000 psi (100 MPa) using silica fume as a partial cementitious
replacement as well as a high range water reducer (superplasticizer) in order to
attain the desired workability and compressive strength.
21
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-354 95 I I 0 6 6 2 9 4 9 0 5 2 2 2 3 9 992 I
I
22 Nawy and Chen
INTRODUCTION
Crack control has long been considered a critical issue in concrete construction
(1-15). Use of precast pretensioned prisms as tension reinforcement can
effectively improve the cracking behavior of concrete members. Previous works
in this area was limited. Burns (5) described a testing program using prisms to
control cracking in continuous members at negative moment regions. Hanson
(10 ) studied crack control in T-beam flange. Bishara and Almeida(4) used
embedded prisms as tension reinforcement in normal strength simple and
continuous rectangular beams. All these tests showed a consistent improvement
of flexural performance at working load levels and also an adequate plastic
rotation capacity up to failure. Since 1976, this type of concrete construction was
also experimentally studied by a number of foreign (Paramewaran (16). Two
major advantages have emerged from using prestressed prisms as main
reinforcement in concrete members using normal strength concrete. These are
higher cracking capacity and the ability of cracks to close with unloading. Hence,
it is beneficial to investigate the use of such composite construction in high
strength and ultra high strength beams and develop expressions for evaluating the
crack width and spacing in such members as an aid to the designers in controlling
serviceabilty behavior.
Fiber optic sensors technology was developed and used in this investigation in
order to provide an accurate measurement of steel reinforcement strain and mean
crack width in the tested specimens. As a result, a mathematical model and
expressions were developed for crack control evaluation as well as a new
approach for long-term and remote monitoring of civil engineering structural
systems.
TESTING PROGRAM
The testing program included thirteen high strength concrete rectangular beams
with prestressed prisms as main reinforcement. The beams were 8 in. x 12 in.
(200 mm x 300 mm) in cross section, were simply-supported over a 9 ft (2.74
m) span and subjected to the third point loading. All the precast prestressed
prisms had the same length of 9.5 ft (2.90 m). Typical section details are given
in Fig: 1 and Table 1.
Weight
Component
ib/yd3 Kg/m3
Uncoated stress-relieved 270 ksi (1860 MPa) 7-wire 318 in. (10 mm) diameter
tendons conforming to the ASTM A416 were used to prestress the precast
prisms. No. 5 ( 15.9 mm dia.) deformed bars conforming to ASTM A616 were
used as additional tension reinforcement in beams A-1 through A-5. No.3 (9.5
mm dia.)deformed bars were used as web stirrups for all the test specimens.
At load levels corresponding to the service loads, all crack widths in the
constant-momentregion were measured with an illuminated microscope of 0.001
in. (0.025 mm) accuracy. Crack spacings of all the developed cracks in this
region were also recorded.
Fiber Optic Bragg Grating (FOBG) sensors were used to measure the average
strain developed in the reinforcement and in the concrete. The FOBG sensor was
either internally mounted along the reinforcing bars or externally mounted on the
beam surface with a gage length of 16 in. (406 mm) as shown in Fig. 1. Five to
seven electric strain gages were mounted inside each beam on the main
reinforcement within the constant moment region. A group of LVDTs were also
installed at the the maximum moment region.
TEST RESULTS
In beams reinforced with prisms only (beams A-6 through A-13), the first visible
crack initially propagated to a depth essentially equal to the thickness of the
prism concrete cover. Thereafter the cracks extended beyond the prism level
with further increase in load even though the prisms remained uncracked. This
was an important characteristic which indicated that the beam would retain a
much higher flexural rigidity after initial cracking of the concrete cover. At load
levels causing the prism concrete to crack, the maximum crack widths ranged
between 0.006 to 0.010 in. (O. 15 to 0.25 mm) at the beam bottom surface and
0.003 to 0.006 in. (0.076 to O. 15 mm) at the tension reinforcement level. These
crack widths fell within the tolerable crack width ranges given by AC1
Committee 224 (i). Once prism cracking occurred, a sudden increase in width
and height of the beam crack was observed corresponding to a sharp increase in
beam deflection. At ultimate loading, only one or two major wide cracks
propagated into the compression zone.
All thirteen test beams showed that the loads causing the first visible crack were
practically at the same level and corresponded to a relatively low stress level in
the reinforcement in the range of 3500 psi (24 MPa) . This meant that the
prisms did not result in a substantial increase in the cracking capacity of beam
cover. Because the prisms continued to maintain some of the initial
precompression, they remained uncracked. However, the restraining effect of
the prisms on the cracking' behavior of the cover concrete seemed to be
insignificant.
Figure 2 shows the variation of average crack spacing with the applied load
(bending moment ratio) for beams A-4 and A-5. This type of composite member
is similar in behavior to a regularly reinforced concrete member in that crack
spacing stabilizes at particular overload levels. Crack spacing in a member with
combined prism and bar reinforcement stabilized after cracking of the embedded
prisms at 30 to 60 % of ultimate load. For members reinforced with embedded
prisms only, the stabilized crack spacing was usually reached before cracking of
the embedded prisms at 25 to 45 % of ultimate load. The magnitudes of
stabilized mean crack spacing, ac; , for all test specimens are presented in Table
3. Other observations and findings from the test results are summarized below.
The stabilized mean crack spacing in the test specimens with deformed bars as
additional reinforcement ( beams A-2 through A-5 ) was normally less than that
reinforced only with prestressed prisms. Also, cracks in beams A-3 through A-5
stabilized at a steel stress level of 28,000 to 36,000 psi (193 to 248 MPa) and
were more evenly distributed than those of the rest of test beams, indicating the
importance of mild deformed bar reinforcement in crack control.
For beams reinforced only with prisms, the crack spacing stabilized at a
reinforcement stress level of 4000 to 7000 psi (27.6 to 48.3 MPa). In addition,
crack distribution in beam cover showed more randomness in irregular spacing.
This can be due to the difficulty of controlling the bond between the precast
prisms and the surrounding situ-cast beam concrete . The stabilized mean crack
spacing as developed by Nawy and Huang (13) using normal strength concrete
beams seems to be equally accurate for the high strength composite beams in this
investigation:
f:
a c s = c --
A,
/r/ Eo
In this study, factor c was found to have a mean value of 0.097 with standard
deviation of 0.016. The ratio (f ' t / df 'J has a mean value of 5.783 with
standard deviation of 0.657 ,as indicated in Table 3. Hence, the expression for
predicting the stabilized average crack spacing for this type of member becomes
as follows:
A, = [ h p + c,,] x b (3)
or
Eq. 5 shows that mean crack width w,,, is proportional to the product of E, and
ucs, experimentally evaluated as shown in Figure 3. It should be noted that
this linear relationship applies even before cracking of the prisms, indicating that
directly correlating the average strain in the reinforcement encased in the prisms
to the cracks on beam surface is reasonable.
By mounting the FOBG sensor at the beam extreme tension fibers for direct
measurement of the crack width, the mean crack width is expressed as:
where tg is the effective gage length and II,. is the number of cracks covered by
the FOBG sensor, w',,,, represents the mean crack width obtained from the
readings of a hand-held microscope. The mean crack width, w',,,, is plotted
against w',,,~"~ for beams A-9, A-IO, A-12 and A-I3 as shown in Figure 4. it
is observed from this plot that the slope of the regression line is 1. This again
indicates that factor a can be approximated by a value of to I .
it has been demonstrated in this work that the FOBG sensor gives a considerably
more accurate estimate of the mean crack width at the tensile face of the concrete
than other means as long as good bonding is developed between the optical fiber
and the concrete face. This is because the optical fiber is flexible and can
closely adjust itself to the deformed beam surface.
where the stabilized mean crack spacing a,.,<is given by Eq. 2, and hence,
Eq. 10 is verified from the experimental test results of beams A-3 through A-13
as given in Table 4. In general, there is good agreement. The average
reinforcement strain can be defined as.
E, = p -f s
E,
By substituting Eq. 11 into Eq. 9, the mean crack width, w,, , is correlated
with the steel stress, f , , giving
Once the mean crack width w,, is determined, the maximum crack width w,,,
under short-term loading can be determined by applying a multiplier 7 to the
Wmean, 1.e.,
The 7-value was statistically evaluated for the ratio of the test values of the
maximum to the mean crack widths in beams A-3 through A-13. These ratios are
shown in the normal probability plot of Fig. 5, based on 69 observations. The
mean value of 7 = 1.501 with a standard deviation of 0.328. Introducing the
multiplier 7 into Eq. 12, the maximum crack width becomes
w,, = 7 p f
2 a,
E*
Taking Young's modulus as 29 x 10 ' psi and using the results of the statistical
w , , , ~( m m ) = 3. 6 x A* f ,
CO
The relative frequency histogram in Figure 6 for the ratio of the test to theoretical
maximum crack from Eq. 15 shows a confidence interval R, = wtm,/w,,, of
about 80%. This scatter in crack width evaluation is expected due to the random
behavior associated with cracking [Nawy (1 i)]. Tests by Billah and Nawy (3) on
high strength normally prestressed non-composite partially prestressed beams
gave a multiplier in Eq. 15 with a range of 2.63 and 3.82 X and a mean
value of 2.75 x for non-composite prestressed beams having a compressive
strength, f ’ c , in the range of 10,000 psi.
The maximum crack width at tensile beam surface can be predicted by,
I A
w”,ax 2. 45 x 10-5 Ri f f ,
E O
where Ri is the ratio of distance from neutral axis to tension face to the distance
from neutral axis to centroid of reinforcement. For test beams A-3 through A-5,
R,=1.39 and for test beams A-6 through A-13, R,=2.28. Analytical results
obtained from Eq. 17 are compared with measured maximum crack width at the
beam tensile face as indicated in Fig.7. The ratio R, = (w’,J/ w’,,, is shown
to have a confidence interval of approximately 90%.
CONCLUSIONS
2. From these tests, the maximum crack width (in.) at the level of the
reinforcement for composite reinforced concrete beams can be
predicted from
A
w,na = 2 . 4 5 x 10-5 f,y
w;, = 2. 45 x 10" R, E o f,
REFERENCES
13. Nawy E. G. and Huang P., "Crack and Deflection Control of Pretensioned
Prestressed Beams", Proceedings. PCI Journal, Precast/ Prestressed
Concrete Institite, May-June 1977, pp. 131-147.
ACKNOWLEDGEMENT
This study was supported by NSF Grant No. MSM-9018572 and the
authors are indebted to Dr. Ken P. Chong, NSF Program Director, for his
advice. This project was conducted at the Rutgers Concrete Research Laboratory
under the direction of the first author.
1 in. = 25.4 mm
&
All Crack widths are measured in the unit of inch x 10” (1 in. = 25.4 mrn),
Ur and w are the observed and calculated values respectively.
6.0-l 108.0.
4 *
A-5 A-6 A-7
A-8
A-11
n7w:
B
-600:
ep - Beam A-4
Beam A-5
Nomai Scorie
Fig. &Normal frequency plot for the ratio (w',,,~~ / w',, ) for Beams A-3 through A-13 at
the reinforcement level
3
Qa
1.5 iI
8
z
0.5
0.0
0.6 1.o 1.4
Ratio It2
2.5 I
.91
2.0
I
h
1.5
a
8
r=: 1.0
0.5
0.0 1. o
0.G 1.4
Ratio R3
41
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 S P - 1 5 4 75 0662747 0522237 780
42 Long and Basheer
AC1 member Professor Adrian Long is the Director of the School of the Built
Environment at the Queen’s University of Belfast, U.K. He has nearly 35 years
of research experience, mostly relating to structural concrete, both reinforced and
prestressed, and concrete technology. He has more than 150 publications to his
credit on concrete and concrete structures.
Dr. Aim Sha’at was a research assistant at the Deparunent of Civil Engineering,
The Queen’s University of Belfast, Northern Ireland, U.K. The work reported
in this paper was carried out by bim for his PhD thesis.
AC1 member Dr. Muhammed Basheer is a lecturer at the Department of Civil
Engineering, The Queen’s University of Belfast, Northern Ireland, U.K. He
has over 7 years of research experience in the general area of concrete durability,
particularly on inspection, assessment, and protection of concrete and concrete
structures. This has resulted in nearly 60 technical publications.
INTRODUCTION
The durability of exposed concrete structures to a large extent depends on
the quality of the cover concrete (“covercrete”as it is often described) because
this provides the first line of defence against various aggressive substances.
Thus, the quality of the covercrete is extremely important and for this reason a
range of different curing techniques can be applied to improve its durability.
Typically, in order to maintain a satisfactory moisture content in the concrete
after the removal of the formwork, the surface is sprayed with water at regular
intervals or a curing compound/plastic sheet is applied. This ensures that
adequate hydration of the cementitious materials takes place but these curing
processes are expensive and time-consuming. Therefore, in the 1980s, Japanese
engineers developed a permeable formwork system (1) which had the objective
of reducing the water-cement ratio in the near to surface region of the concrete.
A similartechnique was used in 1955 for the main draw off chamber of the Glen
Shira dam. In this case, water absorbing Celoten soft boards were attached to
the formwork and these were found to be very effective in reducing the near to
surface water-cement ratio (W/C>. In spite of the fact that the resulting concrete
surfaces have proven to be very durable, the process was discontinued as it was
difficult to remove the board from the hardened concrete.
Within the last few years, a fabric formwork liner has been developed
which gives comparable benefits to the Japanese permeable formwork system at
a significantly reduced cost. The idealised behaviour of this controlled
permeability formwork (CPF) liner is illustrated in Fig. 1. During the vibration
of the concrete, the entrapped air and the surplus mixing water escape, via the
CPF liner which acts as a vertical conduit. Provided the excess water is allowed
to drain freely at the bottom of the formwork then a surface with improved
properties and an absence of blow-holes is produced. As this product has only
been commercially available for four years, relatively little detailed technical
information has been available up until recently on the effectiveness of this liner
TEST METHODS
The following tests, all of which are related to concrete durability, were
applied to the test specimens:
Permeation Tests using the ‘AUTOCLAM’
The air permeabilityand sorptivity tests were carried out at the three
locations shown in Fig. 2 using the ‘Autoclam’ (2) developed in the Department
of Civil Engineering, Queen’s University, Belfast. The results obtained using
this equipment have been shown to correlate well with longer established but less
convenient test methods such as the ISAT (3). Both types of test were carried
out at an age of 42 days with the average of the three test results being reported.
Air Permeabilitv Test: The principle of this test is that a pressure of 500
mbar is applied to the test area using the ‘Autoclam’ and its decay is monitored
over a period of 15 minutes. The natural logarithm of pressure bears a hear
relationship with time and the slope is reported as an air permeability index.
Sorqtivitv Test: These tests were carried out at the same test location as
the air permeability tests, at least one hour later. The test allows the volume of
water penetrating into the concrete at a constant pressure of 10 mbar to be
accurately monitored during a 15 minute period. A plot of the volume absorbed
against the square root of time has been found to be linear, and the slope is
reported as the sorptivity index (4).
g J
The four samples for the carbonation tests were removed from the
locations shown (Fig. 2) at an age of 60 days. Four faces of the samples were
coated with an epoxy resin emulsion thus leaving the two opposite test surfaces
free for carbon dioxide to diffuse. The samples were then subjected to a 15%
carbon dioxide concentration at 16 I 2 OC and 70% RH for 10 days.
Carbonation depths were measured after spraying the freshly broken surfaces
with phenolpthalein indicator solution.
Chloride Ineress: Normal Diffusion Test
Two slices 10 mm thick were obtained from the ends of the 100 mm
diameter cores taken at an age of 60 days from each of the two locations
indicated in Fig. 2. The curved surfaces of these slices were coated with epoxy
emulsion leaving a 74 mm diameter effective test area of the cast and the cut
surface. These samples were then placed in a diffusion cell (Fig. 3). The tank
environment was maintained at 17 i 3 OC and 55% RH throughout the period of
testing. The chloride concentrationin the diffusion cell was measured using a
chloride ion selectiveelectrodeat intervalsof two weeks for the first few months
and at one monthly intervalsthereafter. From a graph of concentration against
time, the diffusion coefficient was calculated using Fick’s f i t law of diffusion.
CONCLUSIONS
As a result of this extensive experimental study the following conclusions
can be drawn:
1. Relative to concrete cast against normal formwork the application of a CPF
liner results in:
o a significantly reduced W/C near to the surface and especially for
concrete with higher W/C
o a greatly reduced incidence of blow holes
o much greater resistance to carbonation and chloride ingress
o enhanced surface strength
o greatly improved performance under freezing and thawing conditions.
2. The benefits of applying a range of typical methods of curing to concrete
cast against normal formwork are modest relative to that achieved by using a
CPF liner.
ACKNOWLEDGEMENT
The financial support of EPSRC (UK) and Du Pont de Nemours
(Luxembourg) S A for this research which was carried out in the
Civil Engineering Department, Queen's University, Belfast, is gratefully
acknowledged.
REFERENCES
1. Harrison, T., “Introducing controlled permeability formwork”, Concrete
Quarterly, British Cement Association, i990, pp 6-7.
2. Basheer, P.A.M. “A Brief Review of Methods for Measuring the
Permeation Properties of Concrete In-Situ”, Proc. ICE, Buildings and
Structures, London, February 1993, pp. 74-83.
3. Montgomery, F.R., Basheer, P.A.M., and Long, A.E. “A Cornpaxison
between the Autoclam Permeability System and the Initial Surface
Absorption Test”, Proc. Structurai Faults and Repairs - 93, Extending the
Life of Bridges, Civil and Building Structures, University of Edinburgh,
29 June - 1 July 1993, pp 71-77, Editor: M. Forde.
4. Basheer, P.A.M., Long, A.E., and Montgomery, F.R. ‘nie ‘Autoclam’ for
Measuring the Surface Absorption and Permeability of Concrete on Site”,
Proc. CANMET/ACI Intl. Conf. on Advances in Concrete Technology,
Athens, Greece, 11-13 May 1992, pp. 107-132
5 . Long, A.E., and Murray, A.McC. “The Pull-off Partially Destructive Test
for Concrete”, Proc. Intl. Conf. on In-SitdNon-Destructive Testing of
Concrete, Ottawa, Canada, October 1994, AC1 SP-82, pp. 327-350, Editor:
V.M. Malhotra.
6. ASTM, “Standard test methods for resistance of concrete to rapid freezing
and thawing”, Annual Book of ASTM Standards, Vol. 04.02, Philadelphia,
C 666-84, 1989, pp 312-317.
7. Sha’at, A.A. “Assessment of Methods of Improving the Durability of
Surface Concrete”, PhD Thesis, The Queen’s University of Belfast. U K ,
April 1994,415 pages.
- CPF face
-Plywood face
- 1ûOrnm
950 diameter core
0 Pull-off
SEM, 10XlOX10
Nuusedregion
dl dim. in mm
Note: Autoclam test locations at 275,575 and 875 m m from the top
Volume of solution
80'
L
60-
9M
a
7, 40-
g
.I
ul
o
U
20'
O+
0.45 0.55 0.65
WIC
WIC
S AC
CC1-HC
cc2-WEI
la cc3-AB
7
0.45 0.55 0.65
WIC
50
40
30
20
10
O
0.45 0.55 0.65
WIC
2o
60 -fl
7"
30
20
10
O
0.45 0.55 0.65
WIC
SP 154-4
55
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-L5q 95 m Ob62949 0522253 050 m
56 Ogawa, Sakata, and Tanaka
INTRODUCTION
than that of ordinary concrete with the same w/c. And as shown
later, the drying shrinkage was also less than that of ordinary
concrete.
Generally, it is considered that highly-flowable concrete
is suited for horizontal members because it is self-leveling.
However, trowelling is difficult due to the high viscosity of
the concrete. Therefore, for monolithic concrete floor finish,
the use of ordinary concrete is preferable. The appearance of
concrete structures made with the highly-flowable concrete
incorporating limestone powder is beautiful; the concrete looks
whitish due to the color of limestone.
REFERENCES
Moisture o. 12%
SiO2 content O. 24%
Ca0 content 55.16%
(Cab, purity) 98.50%
Specific gravity 2. 72
Fineness modulus by Blaine method 2,920 cm'/g
00.50 52.5 1.20 640x650 180 360 154 8.0 3.0 40.8 48.8MPa
00.55 51.0 1.80 615x620 180 327 162 8. O 3.0 34.2 43.8
@ I 50.0
0.60 1.30 610x630 180 300 174 8.0 3.0 32.3 39.8
* SP: HRWRA, * * Viscosity-increaqing agent
*.
- Suffix indicates type of expansive additive,1)CSA-type
Ex:Expansive additive, Shrinkage-reducing agent
--*
2)CAO-type
(%I (%I
W C Lp EA SRA HRKRA VIA
~
O 90 180 270
Quantity of 1 imestone powder (kg/m3)
N: Ordinary c o n c r e t e
E: Expansive c o u c r e t e
ES: Concrete with EA and SRA
FL FL
i1
O 2 4 6 8 10 12
Age (month)
300 t " " ' / " ' 1 ' ' ' 1 ' 1 ' ~ " !
Y-- 200
O
4
x 100 .... . .. ... ... . . . . . . ..
C
3 -300
-400
O 2 4 6 8 10 12
Age (month)
I Vessel
200
O ....... ..........................................
HF*EA+SRA (No.4)
-200 .........
-400 .................
i :
-600 .............. ...............
1
L
. 4
-800 L. I . , . , I I
(A- t e s t )
, . . , , , , , .
O 14 28 42 56
Age (day)
HF+EA (Na.?)
O +HF+EA+CRA (Na41
\y-------
-500
1000
! , , , . . ! , , 1
O 14 28 42 56
0.0
-0.5
-1.0
-1.5
-2.0
-2.5
O 28 56 84 112 140 168
Age (day)
Fig. 7-Weight variation in drying shrinkage test
-1.0
aJ
M
r: -2.0
cb
c
V
c -3.0
M
.-
aJ
-4.0
73
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 75 W Obb2747 0522273 O76 =
74 Han and Walraven
INTRODUCTION
by AC1 Committee 209. The ageing effect (in particular, at early ages) is
discussed, whereas the influences of various factors on ageing are discussed.
EXPERIMENTAL PROGRAM
AU the HSC mixtures were prepared with rapid hardening portland cement
(class C ) . River sand was used as fine aggregate, whereas three different coarse
aggregates were used, i.e. crushed gravel, granite and limestone. The fine and
coarse aggregates were totally dried in air. therefore, they contained no moisture
before mixing. The maximum size of the coarse aggregate was 16 mm. Silica
fume was added in the form of dry powder. For one type of high performance
concrete (HSC-cg-n, with crushed gravel as coarse aggregate, for the creep test
with loading at a normal age), a superplasticizer containing lignosulfonate with a
little Na-gluconate, lignosulfonate and naphthalene was used. For three other
types of high strength concretes (HSC-cg-e, HSC-gr-e and HSC-1s-e, with
crushed gravel, granite and limestone as coarse aggregates respectively, for the
creep tests with loading at an early age), a melamine-based superplasticizer was
used. The slumps of HSC mixtures were higher than 200 mm. The details of the
mixture proportions are given in Table 1.
The test specimens were prisms which were cast III horizontal moulds of
100 x 100 mm in cross-section and 400 mm in length. Additionally, cubes of 100
x 100 x 100 mm were used for reference-tests. Immediately after casting, the
open surface of the specimens was covered with a plastic foil to avoid moisture
losses. Au specimens were then placed in a curing room with 65% relative
humidity and a temperature of 20 k 3 OC. Two hours before testing, the plastic
foil was removed in order to prepare the specimens (i.e. to glue the measuring
devices on the specimens). For the creep tests at normal ages, 65% relative
humidity and a temperature of 20 OC were kept constant during the whole test
period, whereas 50% relative humidity and the same temperature were applied for
the creep tests at early ages. AU the reference specimens were cured in the same
condition as the test specimens.
Creep tests with loading at a concrete age of 28 days were carried out for
only one type of HSC. Three stresslstrength ratios were adopted, i.e. 15%, 35%
and 50%. For any loading level, two specimens, which were placed on the top of I
one another, were used. Also, two specimens were stored in the same condition I
for observing the shrinkage behaviour. For the creep tests with loading at an early
age (16 hours), three loading levels were used, i.e. 30%, 50% and 70% of the ~
strength at the moment of loading. Just before the test, three cubes were tested
for the strength control of concrete, and three prisms were tested to obtain the
average value of the prism-strength which was used to figure out the loading
level. In addition, the development of strength and E-modulus was determined at
16 hours, 3 days, 7 days, 28 days, 90 days and 180 days after casting of the i
concrete.
Concrete is an ageing material, which implies that its properties change with
time. In fact, a lot of influencing factors are involved in this process. Among
others, the type of cement used, the curing conditions including the relative
humidity and temperature play a role. Basically, the strength and the E-modulus
develop very fast in the first period after casting (about several days). From then
on, the rate of development decreases. Fig. I and Fig. 2 show the rate of
development of strength and E-modulus. It indicates that the E-modulus develops
faster than the strength. It has to be pointed out that the test data of the strength
and E-modulus development come from measurement on reference specimens
which are free of external load. Actually, ail the specimens in the creep tests, to
some extent, are subject to pressure. Consequently, two contradictory tendencies
appear [2, 31. At a low loading level, the external pressure may result in a
beneficial effect to the strength as well as the E-modulus, whereas high initiai
stress may lead to a detrimental effect, as far as the later development of
mechanical properties of concrete is concerned. This is considered by the CEB-
FiP Model Code (MC90) as the strength development under sustained load.
In Fig. 1 and Fig. 2 the development of the strength and the E-modulus
versus time from the reference tests are compared with the predictions by MC90
and ACI-209 [4,5]. In both approaches, the type of cement, the curing
temperature and the relative humidity are taken into account. In MC90, the
cylinder compressive strength of the concrete at a time t is estimated as follows:
where E, is the modulus of elasticity at 28 days. Other parameters are the same as
those in eq. (1). In the ACI-209 model, the strength at any time t is predicted by
the following equation:
wheref, is the 28-day strength; a and ß are constants which are functions of the
type of cement used and the curing conditions, respectively. The modulus of
elasticity can be estimated as follows:
where g,, is taken as 0.043; w is the unit weight of the concrete in kg/m3;fcm(t) is
calculated from eq. (3).
From the comparison, it can be clearly seen that MC90 underestimates both
strength and E-modulus at an age earlier than 28 days. Afterwards, it slightly
overestimates tdl the age of 90 days. In comparison with MC90, the ACI-209
approach shows a better late-age estimation, whereas the early-age prediction
remains poor. It has to be mentioned that, although MC90 covers the strength
range up to 80 N/mm2 (28-day characteristic cylinder strength), the contribution
of silica fume to the strength development is not considered. The ACI-209
regulation does not cover high strength concrete.
difference was found between aggregates used, especially at later ages. At early
ages, crushed gravel shows a slightly faster increase of the strength than the two
other aggregates, whereas limestone shows a higher rate of development of the E-
modulus.
In Fig. 3 the shrinkage of HSC with crushed gravel exposed to drying from
an age of 28 days is presented. From Fig. 4 to Fig. 6, the shrinkage strains
measured for HSC with different aggregates exposed to drying at very early ages
are shown. At the mean time, results from three shrinkage prediction models are
also included in these figures.
For the sake of comparison, three expressions [4, 5 , 71 for predicting the
shrinkage of concrete exposed to drying are examined by using the test data The
factors considered in the expressions are, among others, the compressive strength
of the concrete, the type of cement, the relative humidity, the size and the shape
of the specimen, the age of the concrete and the time of drying. In addition, the
ACI-209 model also takes the composition of concrete into account. However, it
has to be pointed out that ail these expressions do not involve the use of silica
fume. From Fig. 3 to Fig. 6, it can be seen that the CEB-FIP Model Code 1990
(MC90) and the modified MC90 model (from Yue et.al) underestimate the
shrinkage of the HSC at normal age (28 days), whereas, for early ages, they
generally overestimate the shrinkage, except for the first few days after the start
of the drying process. The relation given by ACI-209 appears to be better in
predicting the shrinkage of HSC at a normal age than the two other models.
From Fig. 7 to Fig. 10, creep values for three different types of HSC loaded
at three different stresdstrength ratios at an early and a normal age are shown. In
Table 2, the data for creep, shrinkage, the creep coefficient, etc., after a loading
duration of about 210 days for HSC loaded at an age of 28 days, are summarized.
From Table 3 to Table 5, the early creep behaviour of HSC up to 90 days is
summarised.
Regarding the test results, a first impression is that the creep strain for the
HSC used is quite small in comparison with that of ordinary concrete (see Fig. 7).
The creep deformations generally occur very fast, i.e. the rate of creep for HSC is
higher. This is specially true for HSC loaded at early ages. The large part of the
creep strains occurs during the first 20 days and the increase thereafter is rather
small as is shown in Fig. 8, Fig. 9 and Fig. 10. This is also confirmed by other
researchers [8]. The influence of the aggregate used on the creep deformation is
less significant, although slightly smaller creep strains for HSC with limestone
were observed, especially at a lower stresdstrength ratio. This might be attributed
to the higher early development of the E-modulus.
In Fig. 15 and Fig. 16 the strains upon introducing the load, related to the
different stressktrength ratios for various HSC mixtures are shown. For both
HSC loaded at 16 hours and 28 days, the short term strains (elastic strains) are
basically linear till a stress level of 70% of the strength. This is in agreement with
previous observations [9, 101. The long-term strains of HSC loaded at 28 days
show a linear relation to a stress/strength ratio up to 70%, whereas the long-term
strains of HSC loaded at 16 hours reveal such a linear relation only up to about
SO% of the stresdstrength ratio. Previous works revealed that the non-linear
relation between stress and deformation originates from the development of
microcracks inside the concrete [ l i ] . Because of the differences of the material
structures between HSC and ordinary concrete, the starting point of this non-
linear relation (limit of linearity) varies significantly. From this investigation, it can
be seen that early age HSC behaves in a similar way as ordinary concrete loaded
at a normal age. From Table 2 to Table 5 , the comparisons of the creep
coefficient and the specific creep between HSC and early age HSC c o n f i i this
conclusion. For the same type of concrete, the creep coefficient of early age HSC
can be two to three times larger than that of mature HSC, whereas the specific
creep can reach about 8 to 10 times higher values. This defitely has to be
considered when early age HSC is used in practice.
CONCLUSIONS
The development of both strength and stiffness of HSC at early ages is much
fast than that of ordinary concrete. To predict this increased development, the
existing expressions have to be modified. No significant effects of the
aggregate type on the development of strength and stiffness were found.
The drying shrinkage at a normal age of HSC is rather smali, whereas the
chemical shrinkage of early age HSC is very large. Among three types of early
age HSC, the type with limestone shows a smaller shrinkage than the other
ones, due to its higher early stiffness. Because of the fast initial development
of the shrinkage, the expressions used to predict the shrinkage clearly
underestimate the shrinkage at early ages, whereas, at later ages,
overestimation is found.
Creep of early age HSC and mature HSC is significantly different. The rate of
creep of early age HSC is much higher. The type of aggregate seems to have
no considerable influence on the creep behaviour. Just as in the case of
shrinkage, the expressions used generally underestimate the creep at early
ages. In comparison with more mature HSC, early age HSC has a relatively
lower limit of linear behaviour as far as the stresdstrength ratio is concerned.
Due attention to the time-dependent behaviour has to be paid when early age
HSC is used in practice.
ACKNOWLEDGEMENTS
REFERENCES
1. N. Han, Creep of High Strength Concrete at early ages - state of the art
report, Stevin Laboratory Report, September 1994, Delft University of
Technology, The Netherlands, 56 pp.
9. N. Han, The behaviour of high strength concrete under sustained and slow
increasing load, 29. Forschungskolloquium, Deutscher Ausschuss für
Stahlbeton, Delft University of Technology, The Netherlands, 1994, pp. 13-
20. (in German)
11. N. Han and J. C. Walraven, The Effect of Strain Rate on the Compressive
Properties of High Strength Concrete, Progress in Concrete Research, Vol.
2, 1991, Delft University of Technology, pp. 31-43.
in
9
o
a)
7
L
5n
v,
U
I
cz
o,
U
2
c
w
c
o-
w
Ly
cz
U
U
O
z
s
2
v,
I
r+
-1
Y
m
i=
P
m
00
(u
0 crushed gravel
0 limestone
O 20 40 60 80 100
Age ( days)
0.80 I r I I
l I I
0.60
n
J
Y
a
0.40
&
.-
C
L
c
cn
0.20
0.00
O 400 800 1200 1600 2000 2400
Time ( hours )
0.80 1 I I I I
0.60
n
ìJ
Y
al
Y
3 0.40
.-
C
2
u)
0.20
0.00
O 400 800 1200 1600 2Ooo 2400
I
Time ( hours )
0.80
0 limestone
0.60
n
b%
Y
Q,
Q 0.40
Y
.-
C
L
æ
u)
0.20
0.00
O 400 800 1200 1600 2000 2400
Time ( hours )
0.60
n
3 0.60
v
p.
Q)
L
o 0.40
0.20
0.00
O 1o00 2000 3Ooo 4Ooo 5Ooo 6000
Time ( hours )
2.00
1
h
J
Y
p 1.00
!oi
0.00
O 600 1200 1800 2400
Time ( hours )
Fig. &Creep of HSC with crushed gravel (cg) loaded at an age of 16 hours
2.40
1B O
h
J
Y
Q 1.20
a3
2
o
0.60
0.00
O 800 1200 1600 2000 2400
Time ( hours )
2.00
1.50
n
J
v
Q 1.00
a3
L
o
0.50
0.00
O 400 800 1200 1600 2OOo 2400
Time ( hours )
2.00
1.50
r
.-.-a3CI
5CI
1.00
Q
a3
2!
o 0.50
0.00
O 1000 2Ooo 3Ooo 4000 5000 6000
Time ( hours )
Fig. 11-Comparisons of creep coefficients obtained from tests with different approaches
(loaded at an age of 28 days)
Fig. 12-Creep coefficients of HSC with crushed gravel loaded at an age of 16 hours
3.00
w
C
.E 2.00
u
g
n
aa
2 1.00
o
0.00
O 400 800 1200 1600 2Ooo 2400
Time ( hours )
3.00
U
C
.g 2.00
5
L)
8
n
0,
1.00
o
0.00
O 400 800 1200 1600 2000 2400
Time ( hours )
il .-
31
-
---)--
I
I
cg-n t=o
cg-n ta10 days
l l
A
'CI2 -cg-et=û
Q ,.A'
-
O
...A.. . cg-et=90days I*
,I
m
O
c L
Q)
3 I ,
'CI
.E 1 ~~
L
z
O
O 0.2 0.4 0.6 0.8 1
Stresdstrength ratio
d
Y
O
c
0
O 0.2 O .4 0.6 O .8 1
Stressktrength ratio
0.80
) I Cg 30%
.g
w
L
5
0
0.60
0.40
-
-
-
....A ... cg 50%
cg 70%
e
u)
À.
‘A- ---.
u) --.-a------
p!
fj 0.20 -&=--+-- ---
------a-
0.00
O 20 40 60 80 1 O0
Age of concrete ( days)
SP 154-6
Svnopsis: The resistance of rice hull ash concrete to freezing and thawing in saline
environment was studied in the laboratory, for non air entrained high performance
and normal concrete. The Swedish standard test for concrete resistance to freezing
and thawing in saline environment was used.
Although the number of tests was limited, the results were very promising for
the use of RHA in non air entrained normal or high performance concrete. The
laboratory salt scaling for concrete with 1 5 3 0 % replacement of portland cement
with RHA indicated that RHA concrete without air entrainment would be fairly
resistant to freezing and thawing in most applications except for in very severe
climates. No indications on an accelerated scaling rate over time was observed for
RHA concrete, as opposite to the accelerated scaling rate found for a non air
entrained high performance silica fume concrete tested.
95
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 m 0662949 0522293 737 m
96 Sandberg
P. Sandberg has been working with cement and concrete research since receiving a
M.Sc. in Chemical Engineering at Lund Institute of Technology, Sweden, in 1988.
His major research interests include chloride initiated reinforcement corrosion in
concrete, general durability of concrete, and cement chemistry. He is currently
employed at Cementa AB, and studying at the Division of Building Materials, Lund
Institute of Technology, Sweden.
INTRODUCTION
In recent years there has been an increasing interest in the use of rice hull ash
(RHA) as a mineral admixture in concrete /1,2/. RHA is produced by burning rice
hulls from the dehusking of paddy rice. If produced properly, RHA is a highly
reactive mineral admixtures due to its cellular structure and its high amorphous silica
content. The cellular structure provides a surface area of 50-100 m2/g. RHA is
frequently available in rice producing countries. 'Ihus it is a pozzolanic material of
high potential value for the concrete industry.
This study focuses on the use of RHA in concrete exposed to cyclic freezing
and thawing in saline environment (salt scaling). Traditionally, air entraining
admixtures are used in concrete to achieve resistance to salt scaling. The use of air
entraining admixtures however, causes a strength loss in the order of 5 percent for
each percent of air voids introduced into the concrete /3/.
The use of RHA in concrete would be even more beneficial, if RHA would
help to improve the resistance to salt scaling, without the use of any air entraining
admixture resulting in strength loss. Besides the general decrease in capillary
porosity associated with the pozzolanic reaction, the cellular structure of RHA may
also have some positive influence on the resistance to salt scaling. The cellular
structure may to some extent serve as an air void system, acting as a pressure relief
when subjected to freezing and thawing. RHA is in this sense unique compared to fly
ash, slag and condensed silica fume.
In this study, the Swedish standard method for laboratory testing of concrete
resistance to freezing and thawing in saline environment /4/ was used. This method
has been suggested to be implemented as an European standard method within the
EU counties. The testing is known to be severe, but concretes performing well in the
test have also been found to perform well in situ /5/.
The standard curing and conditioning of the concrete involves 7 days water
curing at room temperature (2OoC), followed by 14 days of conditioning at 50%
RH, 2OoC and air velocity < 0.1 m/s. Concrete specimens are usually 50 mm thick,
and cut from >lo0 mm thick concrete cylinders or cubes.
The exposed surface is usually a cut surface, in order to avoid any "skin"
effects and bleeding effects in the original concrete surface. After sample
preparation the concrete is conditioned for another 7 days at 50% RH, 20oC and air
velocity < O. 1 m/s. After conditioning, the concrete surface to be tested is saturated in
fresh water for 3 days. The fresh water is then reJhced by saline (3% NaCl) water,
and the testing is done by exposing the concrete surface through a minimum of 56
cycles of freezing and thawing /Y. One cycle of freezing and thawing is illustrated in
Fig. 1. The test setup is shown in Fig. 2.
In the present study, each test was tun until the tubber sealing was broken,
resulting in drainage of the tested concrete surface (see Fig. 2).
EXPERIMENTAL STUDY
Materials
Cement-A Swedish sulfate resisting portland cement was used. This cement
is usually specified for high performance concrete and for concrete in severe
environments.
Rice hull a&--RHA from a commercial U.S. source was provided by
Professor P.K. Mehta at UC Berkeley. The silica in the RHA was present in
amorphous form.
Silica fume--A Norwegian highly reactive silica fume supplied as a 50%
slurry was used in a high performance concrete mixture, as a comparison to the RHA
concrete.
Aggregates--Local natural sand (SG=2.69) and 16 mm maximum size
aggregate (MSA) crushed limestone (SG=3.11) was used.
uoerDlasticizer--A sulphonated naphtalene formaldehyde condensate was
used a s a 42% slurry, having a density of 1200 kg/m3. If needed, superplasticizer
was added to reach a 10-12 cm slump concrete 5 minutes after mixing was
completed
The materials used are specified in Table 1.
Concrete Mixtures
A high performance concrete having a W/(C+RHA) of 0.35 with 15%
replacement of portland cement with RHA was compared with 2 reference concrete
mixtures having a W/(C+SF) of 0.35, one without mineral admixtures in the binder
and one with 10% silica fume in the binder. In addition, a medium quality concrete
having a W/(C+RHA) of 0.50 with 30% replacement of portland cement with RHA
was compared to a reference concrete having a WIC of 0.50 without mineral
admixtures in the binder. No air entrainment was used in any concrete.
The concrete mixture proportions and compressive strengths at 28 days are
given in Table 2.
The concretes were mixed for 3 minutes and cast in 150 mm cubes at 2OoC,
followed by standard laboratory moist curing as described.
SUMMARY
Although the number of tests was limited, laboratory testing of the resistance
of RHA concretes to freezing and thawing in saline environments resulted in very
promising results for the use of RHA in non air entrained normal or high
performance concrete. The laboratory salt scaling for concrete with 1530%
replacement of portland cement with RHA indicated that RHA concrete without air
entrainment would be fairly resistant to freezing and thawing in most applications
except for in very severe climates. No indications on an accelerated scaling rate over
time was observed for RHA concrete, as opposite to the accelerated scaling rate
found for a non air entrained high performance silica fume concrete tested.
REFERENCES
kg dry materiallm
30 I I I I I
- 20
10
E
G o
-10
-20
O 6 12 18 24
Time (hours)
/@-
Fig. 2-Test setup for Swedish standard method "the Bods Method" for testing of concrete
resistance to freezing and thawing in saline environment. A) thermocouple, B) lid, C) 3 percent
NaCI solution, D) concrete specimen, E) rubber sealing, and F) heat isolation (5)
O
Number of freeze-thaw cycles
Fig. ?Salt scaling results for high performance concrete having a W/(C + RHA) or W/(C +
SF) of 0.35, with 15 percent and O percent replacement of portland cement with RHA or 10
percent replacement of portland cement with silica fume, and for "normal" concrete having a
W/(C + RHA) of 0.50 with 30 percent and O percent replacement of portland cement with RHA
SP 154-7
High-Performance Concrete
Floor Construction and
Performance Assessment
by R. L. Munn and I. Dumitru
Synopsis: During 1994 a new 50,000 n,* warchouse and similar area of
external pavement was constructed at Ingleburn near Sydney, Australia. The
client required that the warehouse meet very onerous performance criteria that
required the construction of a very flat, prestressed concrcte floor with excellent
abrasion resistance, is crack free and has a minimal number of joints. The
design required that the concrete base provide the wearing surface for the floor
without application of a surface topping.
103
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 m Obb2949 0522300 7 T 7 m
104 Munn and Dumitru
INTRODUCTION
Many concrete floors have been poorly cured, and the combination of
inadequate curing and concretes of lower strength grades has led to many floors
exhibiting dusting, cracking and poor abrasion resistance subject to normal
loading and traffic. Even when metallic floor shakes have been applied to
concretes of modest strength grades, despite the improvement in abrasion
resistance, numerous cases of delamination and local scaling of the surface have
occurred, which have adversely affected the serviceability of the industrial floor.
In addition, dry shake application has been reported to adversely influence
flatness. (4)
The concrete was pumped from the edge of each slab to the point of
placement using three pumps simultaneously to minimize the length of the
leading edge during placement but placing delays ensured that the leading edge
was rarely linear. The concrete was compacted using internal vibrators followed
by a vibrating beam over the full width of the slab. Frequent level checks were
provided using laser levelling during the screeding operation to ensure that
surface tolerances were being met. The concrete was finished sequentially
using single headed, then double headed and finally triple headed power
trowels. Levels were again checked by laser at frequent intervals to ensure that
the final surface was being h i s h e d within specified tolerances. Curing was
The plant floors are divided into 13 different locations, each having its
own concrete specification. The majority of the concrete is specified to have a
strength grade of 40 MPa with a drying siirinkage of less than 600 LE after
56 days drying, howcver two areas were specified to have strength grades of
80 MPa and 100 MPa respectively, with drying shrinkage of less than 450 PE
after 56 days drying. The abrasion resistance was specified for assessment
insitu using the Chaplin abrasion machine with a maximum permissible loss of
0.1 mm.
Before beginning the work, a series of laboratory and field trials were
conducted on 60-MPa, 80-MPa and 100-MPa strength grade concretes to refine
the mixture proportions and the finishing practices needed to achieve the
specified flatness and abrasion resistance.
PERFORMANCE ASSESSMENT
24 hours, and compressive strength was also determined at 24 hours using 1O0 x
200 mm cylinders to provide early warning of potential quality variation and for
initial stressing. All concrete tests were performed in accordance with
-
Australian Standard 1012 Methods of Testing Concrete, which contains
procedures similar to relevant ASTM standards.
where fcl is the 28-day flexural strength and fc is the 28-day compressive
strength.
E = 0.043 p ' S x
Test results from the 28-day Chaplin abrasion resistance surveys on the
Ingleburn and Chullora iria1 slabs arc reported in Tables 2 and 3. The results of
the Chaplin tests for each Ingleburn and some Chullora concrete placements are
comparcd with average 28-day compressive strengt11 of concrete and the average
Schmidt Ilammer Rcbound Index in Tables 5 and 7. Thesc test results indicate
that the specified abrasion resistance has been met for the Inglebuni project with
an averagc depth of wcar of 0.05 mm.
Complete data are not yet available for the Chullora project but initial
trials and field results todate indicate that the depth of wear recorded using the
Chaplin abrasion tester will be equal to or better than that achieved on the
Ingleburn project. The rebound index numbers are significantly higher for the
Chu1lora prqject concrete which tends to confirm that the surface hardness
increased with increasing concrete strength.
In this respect high range water reducing admixtures werc used to ensure
that the slump at the time of placement was substantially higher than the normal
slump of 80 nim conimon in Australia, without increasing the water content.
The nominal slumps adopted were i n the range 120 to 150 mm, which provided
increased workability and extended windows for finishing without adversely
allecting other properties.
The only concrete deliveries which were not pumped were placed at the
end of each placement represented by thermocouple location 12. ‘The initial
temperature of non-pumped concrcte was found to be up to 4°C lowcr than
identical concrete which has been pumped, and its temperaturehime profile is
significantly different. The non-pumped concrete also suffered a finishing time
delay of approximately 2 hours.
Figures 2,3 and 4 show that isochrones for placement and each stage of
finishing on one of the 60 m x 35 m Inglebum project slabs, and include the
location of the thermocouples. Contrary to expectations, each of the finishing
operations did not follow the same pattern as the placing operations duc to local
variations in concrete behaviour. These finishing time variations, with the
exception of the non-pumped concrete, appear to result from very small
variations in mixture proportions and site conditions. These random variations
in turn require a very high degree of skill on the part of the placing team to
successfully finish to strict tolerances over a large floor area.
The Chaplin abrasion testing machine was used for the first time in
Australia on these projects. Correlation with test data for other grades of
concrete and other niethods of assessing abrasion resistance and surface
hardness were carried out on field triai slabs to obtain a better understanding of
the significance of the test results. Prior to assessing the wear resistance of
floors which have curing compounds on the surface, it has been found essential
to remove the curing compound with a chemical solvent. Where the
curing compound has not been effectively removed, the Chaplin abrasion tester
skidded on the surface and a zero wear depth was recorded. Similarly, Chaplin
abrasion testing of the ground surface of the high-perfomiance concrete also
showed zero wear depth.
Chaplin abrasion wear data were obtained for some lower strength grade
concretes on each project under similar conditions to provide some
understanding of concrete strength sensitivity. On the Ingleburn project,
25-MPa concrete placed in the roof of a fire tunnel was assessed io have a mean
abrasion wear depth of 1.8 1 mm. On the Chullora project a concrete floor of
40-MPa concrete was assessed and found to have an average wear depth of
0.80 mm. These data confirm that the abrasion losses of normal concretes are
an order of magnitude greater than those of the higli-performance floors on these
projects.
Review of the test data derived from testing each concrete by different
methods was made. Given the limited range of concretes in this comparison, no
clear correlation between results reported for different methods could be
identified. The rebound index can provide useful information about the surface
hardness in the absence of abrasion testing equipment, although it does not
provide a direct measure of abrasion resistance in Fig.5. The rebound index
data from the Ingleburn and Chullora projects has been superimposed on data
reported in the United Kingdom by Kettle and Sagegzade. (7) Data obtained
from both projects classified the concrete floors as very highly wear resistant
and showed that the abrasion rcsistancc for a given rebound index was better
than U.K. concretes.
CONCLUDING REMARKS
The tcst data generated during the trial construction periods for these
projects confirmed the excellent properties which may be consistently achievcd
with appropriately designed high-performance concrete, including high early
strength, low drying shrinkage and flexural strength and elastic modulus
exceeding 10 MPa and 50 GI%, respectively. Several useful com parisons
between methods for assessing surface hardness and abrasion resistance were
made during cxccution of these projccts. It was concluded that the Chaplin
abrasion tcst technique provided the most appropriate method to assess the
potential durability of floors subject to wheeled traffic.
ACKNOWLEDCEMEN’I’S
iIEFERENCES
TABLE 1 - TEST DATA ON DENSITY AND STRENGTH FOR 60-MPo CONCRETE USED IN
INGLEBURN PROJECT
Placed
Number
23.2.94 PSI9
1.3.94 PSI6
7.3.94 PSI8
11.3.94 PSI3
*
15.3.94 PS17
17.3.94 PSI5
29.3.94 PSI2
I
~~
7.4.94 PSI4
11.4.94 I PSI0 I 2540 17.5 39.5 55.5 I 69.5
-14.4.94 I PSI 1 I 2530 17.5 40.5
20.4.94 I I
~~
'*
PS9 2.530 18.5 39.5 69.0
I
m
in
vi
iA
‘?
iA
vi
rd
in
m
3
d
vi
rnx
$2
‘cio0
r-N
@, 28 Days
23.2.94 PSI9 59
1.3.94 PS16 62.5
7.3.94 PSI8 62.5
11.3.94 PSI3 64.5
15.3.94 PSI7 67
17.3.94 PS15 66.5
29.3.94 I PSI2 I 66.5
7.4.94 I PS14 I 70.5
11.4.94 I PSI0 I 69.5
14.4.94 I I
PSI 1 I
I
69
20.4.94 1 ~ I’S7 I 69
~ ~ ~
22.4.94 PS9 69
27.4.94 PS4 64
2.5.94 PS8 66
3.5.94 PS3 67.5
I I
~~
5.5.94 PS6 65
13.5.94 [-PS2 I ~ 67
430 6.35
390 6.1
Ln
L
a,
n
s
I
c
E
c
U
n
E
U
<n
-<nm t m w -
u
o)
o)
S
u
127
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 m 0662949 0522324 130
128 Tritthart
INTRODUCTION
Figure 1 shows the face wall of the hall. As can be seen, the
concrete surface, which contained numerous spots of corrosion
products, was partly still covered by white paint. The area that
was selected for chloride removal was about 36 m2 and was located
behind the scaffolding. After the drilling cores to determine the
initial distribution of chloride, the closure of the holes with
mortar, and other preparatory operations, the reinforcement in
the area to be repaired was exposed at four spots to wich
electric cables were attached. Wooden slats of 20-nun thickness
were then fastened onto the concrete.These slats were used to
support a commercially available titanium mesh as anode.
Following this, another set of wooden slats were nailed as
spacers onto the previously fastened slats. Finally, pulp was
injected to fill up the 40-mm space between the concrete surface
and the top edge of the slats. Throughout the test, the pulp was
sprinkled with tap water almost daily to keep it wet and thus to
ensure electrical contact with concrete acting as electrolyte.
Average current density applied to concrete surface was 1 A per
m'. The necessary voltage was continually readjusted by hand.
Figure 2 shows the wall after the preparations had been
completed.
TESTS
The core sections of 50-mm diameter were first dried and then
ground to fineness c0.09 mm. After the samples had been dried
again over-night at 105OC, determinations were made on: (i) the
total chloride content after dissolution in nitric acid by
potentiometric titration with 0.1 molar or 0.01 molar AgN03, and
(ii) the Si02 content, after dissolution in diluted hydrochloric
acid (1:l) gravimetrically. From the soluble Si02 content, the
cement content of the samples was calculated. As no information
on the Si02 content of the cement was available, it was assumed
that the soluble Si02 content of the cement was 20% (typical
value for portland cement) and that the aggregates did not
contain any soluble silica (in natural Austrian aggregates it is
normally not more than a few tenths of a percent). It was
necessary to determine the cement content in each individual
section because concrete is a highly heterogeneous material and,
therefore, the cement content in such small samples may deviate
considerably from the average cement content of the concrete; yet
the chloride content of each sample must be related to the cement
content.
From the zone between the titanium mesh and the concrete
surface, a pulp sample was taken at weekly intervals and examined
for pH-value and chloride concentration. After 40 days of
migration, a first control core of 50-mm diameter was extracted
from near sampling point 4, and after 83 days of migration,
another one of 200-mm diameter from close to the extraction point
of the first core of the same diameter. After 96 days of
migration the flow of current was interrupted for about 3 months
( 9 8 days). Before the current was restored, another core of 200-
mm in diameter was taken to analyse the liquid phase and the
solids. The reason for the interruption was to determine whether
bound chloride would dissolve during such a break. It has been
reported (12) that C1- removal becomes inefficient with time,
because the dissolved chloride has already been removed, and
bound chloride dissolves very slowly, therefore, C1- removal can
be continued more efficiently after a pause during which bound
chloride has been dissolved.
evaluated by the time this paper was written, hence they are not
discussed here.
Cathode (reinforcement):
CONCLUSIONS
ACKNOWLEDGEMENTS
REFERENCES
c
c-
2c2
- - * - -Core4
s - - -x-. . Core 5
0-10 15-25 30-40 45-55 60-70 75-85 90-100 105- 120- 135- 150- 165-
115 130 145 160 175
Distance from Surface, mm
- - .fi. - - Core 4a
- -0- - Core 6
- - C- - - Core 6a
-X- Core6b
0-10 15-25 30-40 45-55 60-70 75-85 90- 105- 120- 135- 150- 165- 180-
100 115 130 145 160 175 190
Distance from Surface, mm
Fig. &Chloride content (i) before chloride removal (cores 4 and 6), (i¡)after 40 days of
ig. migration (core 44, (iii)after 83 days of migration (core 6a), and (iv) after 96 days of
migration and subsequent interruption of migration for 98 days (core 6b)
2,50
I
before migration
-
1
2,oo 0 after 83 days of migration I
Y
o 0,510
0,oo
0-1o 15-25 30-40 45-55 60-70 75-85
Distance from Surface, mm
3,OO
before migration
2,50 0 after 83 days of migration
-
5
O
after 83 days of migration and a
E 2,oo
e- break of 98 days
O
._ ~
2
S
1,50
a
ol
c
q 1,oo
o
0,50
0,oo
0-1o 15-25 30-40 45-55 60-70 75-85
Distance from Surface, mm
SP 154-9
INTRODUCIION
Construction
Records state3 that the initial W/C in the first crib was
approximately 0.5, subsequently increased as work progressed
up to about 0.6. At no time, it is stated, did the water-cement
ratio exceed 0.625.
Cold water was poured over the form work during the initial
hardening of each lift. However the surface temperature did
reach 4OoC within eight hours after casting.
Figure 5 shows two typical core sections, (from core SS-2, LHS;
core SS-1, RHS, 200 mm long x 100 mm diameter). The magni-
fication is 0.625.
T (N, E, S) - E to T (N. E, S) - T,
S (N, E, S) - 1 to S (N, E, S ) - 8,
From Table 3 it is evident that all concrete from the cope wall
above high water was manufactured with normal portland
cement. We know from the method of construction of the pier
that all concrete below low water was made using .CAC.
Table 3 shows that the inner concrete in the tidal zone was
made using CAC as expected but that the outer layer of facing
concrete can be of either normal portland cement or CAC.
Based on this information the remaining core drilling was
organized.
Mineralogy
Two samples, a and ß from core TE-T were examined and the
results are shown in Figures 8.1. and 8.2. It can be seen that
pore diameters are predominately clustered in the range
0.1 pm to about 1.0 pm and this is quite typical of a CAC
concrete5. The total intruded porosities of 41.7 mm3/g, (a) and
38.4 mm3/g, (ß) in Figure 8.2. correspond to volume porosities
of 10.0% and 9.2% respectively assuming a density for the
concrete of about 2.4 Mg/m3. About 70% of this porosity lies
between 1.0 pm and 0.1 pm while a further 25% lies between
0.1 pm and 0.01 pm. The curves have not leveled off even at
0.005 pm, the smallest pore diameter accessible, (maximum
Hg pressure, 150 MPa). This combination of capillary plus
micro-porosity is more typical of unconverted rather than
converted CAC5, in agreement with the result given in Table 3.
CONUUSIONS
ACKNOWiEDGh4ENTS
4. See reference 1.
I I
Visual and electronical charac-
Instituto Eduardo Torroja (ET), terization of reinforcement
Spain condition, chemistry, mineralogy,
physical properties
t I
Carbonation, chloride and sulfate
Building Research Establishment ion penetration, oxygen
(BR), United Kingdom permeability, optical micro-
scopy, electrochemical tests
W
L C
-'E
LL
I
c
.-
I
m L
.s O
E c
-
r
a
m
-
A
Lo
N
O
r
I
a
o
óo
50
m
a
Gi
N
c?
W
I
mO
I
z w
- I -
Building Research
1 After sampling this core has been transferred ni = non identified origin
to BRE
km+
í 300'4"
)
Fig. ?Schematic diagram of cribs. Interior compartments (not shown) were cast
integrally with the walls and base. Wall thickness of cribs is 533 mm
No. L W H location
1 1: 1
c2
2: ~
23.55 m
(77’3”)
32.7 m
(1 07’3”)
1 1 1
15.8 rn
(52’)
12.8 m
(42’)
(53‘)
16.15 m
(53’)
13.1 m
(43’)
3 N
5 s
E
4 inlanc
o f ten
very weak
o r no bond
?St
-_- ---------
tldal OE
zone
- --------_-_- L.W.O.S.T.
submerged
01 0 2 0 3
5 th top of pler th
------------- H.W.O.S.T.
tldal OL
zone
------------- O T
L.W.O.S.T.
submerged
Fig. 8.1-Pore size distribution from MIP: two samples, a and /3 from core TE-T
Fig. 8.2-Total intruded pore volume from MIP: two samples, a and /3 from core TE-1
SP 154-10
169
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 S P - 1 5 4 95 Ob62949 0522364 ô17
170 Marchand and Gérard
INTRODUCTION
Cement-based composites (hydrated cement pastes, mortars, plain and fibre-
reinforced concretes) are used in the construction of a wide range of structures.
During their service life, many of these structures are exposed to various types of
aggression where, in most cases, the deterioration mechanisms involve the
transport of fluids andor dissolved chemical species within the porosity of the
material. This transport of matter (in saturated or unsaturated media) can either be
due to a pressure gradient (permeation), a concentration gradient (diffusion) or the
application of an electrical field (migration). In many cases, the durability of a
cement-based composite is controlled by its ability to act as a tight barrier that can
effectively impede, or at least slow down, the mass transport process.
Given their direct influence on durability, mass transport processes have been the
object of a great deal of interest by researchers. Although the existing knowledge
of the parameters affecting the mass transport properties of cement-based materials
is far from being complete, the research done on the subject has greatly contributed
to improve the understanding of these phenomena. A survey of the numerous
technical and scientific reports published on the subject over the past decades is
beyond the scope of this report, and comprehensive reviews can be found
elsewhere [ 1-51.
The assessment of the mass transport properties by laboratory or in-situ tests is
often difficult and generally time-consuming [6,7].For this reason, a great deal of
effort has also been made towards developing microstructure-based models that
can reliably predict the transport properties of cement-based composites. A critical
review of the most pertinent models proposed in the literature is presented in this
paper. Some of these models have been previously reviewed by other authors [ 5 ,
8-10]. The purpose of this review paper is evidently not to duplicate the works
done by others, but rather to complement them. In the present survey, emphasis is
therefore placed on the most recent developments in the modeling of mass
transport processes. The review is limited to the models developed to predict the
diffusivity and the permeabilityof saturated cement-based composites.
Theoretical considerationspertaining to the calculation of the transport properties
of a porous medium are briefly reviewed in the fiist part of this paper. This part
also includes a survey of the various mathematical concepts developed to
characterize the physical structure of porous media. The second section of the
paper is devoted to the critical review of the main models found in the literature.
Empirical, physical and numerical models are reviewed in separate sections.
Special attention is paid to the recent innovations in the field of numerical and
digital image analysis based modeling.
Permeation
Permeation is usually defined as the process by which a fluid goes through a
material due to the action of a pressure gradient. Fig. 1 shows the simple case of
two pores connected by a pipe. Pore 1 is at a pressure Pi and pore 2 is at a
pressure P2, Pi being higher than P2. Under this pressure gradient, the fluid in the
first pore moves into the second through the pipe until Pi is equal to P2. In the
case where P1 and P2 remain constant and allow a laminar flow (Le. Reynolds
number < 2 0 0 ) , a steady flow is recorded and, neglecting the gravity acting on
the particles, the flow rate in the pipe can be calculated using Darcy's law:
- k
J = - - grad (P)
9
where is the volumetric flow rate of the fluid (m/s), that is analogous to a speed,
k is the intrinsic coefficient of permeation (m2) which is independent of the fluid,
q the dynamic viscosity of the fluid (kg/m/s) and P the pressure (Pa) acting at
coordinates x, y, z.
A modified version of Darcy's law has been specifically derived for saturated
porous materials:
where k' is the modified permeation coefficient (m3.sBg). Equation (2) was first
developed for water transport in soil, but is now commonly applied to
consolidated materials such has hydrated cement paste and concrete. It should be
noted that Darcy's law is a phenomenological equation, derived from empirical
observations, that describes the bulk flow of a fluid through a porous medium
without any reference to the microstructural characteristics of the material. In that
respect, Darcy's equation is analogous to Fourier's law for heat conduction and
Ohm's law for elecmcity conduction.
For practical purposes, equation (2) can be easily modified using the following
relation:
P=pg h (3)
where h is the hydraulic head (m), and p is the density of the fluid (kglm3). For an
incompressible fluid, p is constant and independent of the pressure applied, and
equation (2) thus becomes:
-
J = -K grid(h) (4)
where K is the permeation factor ( m / s ) that is both a function of the porous
material characteristicsand of the permeating fluid viscosity and density. For water
at 20' C,k/K = q/p/g= 10-7 (m.s).
The law of mass conservation applies to any mass transport process and provides
a differential equation when solved gives the head pressure and the flow current in
all points of the structure. For a steady flow in saturated porous media, Laplace's
equation (issues from the law of mass conservation) should thus be satisfied in all
points of the solid:
For a compressible fluid, the relationship existing between the applied pressure
and the fluid density can be derived from the law of ideal gases:
P=pRT (7)
where T is the temperature (Kelvin) and is the molar gas constant (J/moi/K).
Under isothermal conditions, equation (5) becomes:
E +-+--
a2h2 - div[h grãd(h)]= div[PgrGd(P)]= O (8)
ax2 ay2 az2
The applicability of Laplace's equation to saturated Darcy flows implies that most
permeation processes in porous media should be modeled using solutions derived
for similar heat flow problems[4].
Diffusion
Diffusion is usually defined as the transport of matter due to a concentration
gradient. Thus, while the permeation process is Concerned with the transport by
bulk flow, the diffusion process is concerned with the motion of individual
molecules or ions. As for the permeation process, the phenomenological equations
developed to describe the diffusion process, known as Fick's first and second
laws, have been derived from empirical observations. It was only many years after
their introduction that these laws were explained from theoretical considerations.
It has been demonstrated that the general kinetic theory of ideal gases can also be
applied to the perpetual motion of molecules in a solution. At a given temperature,
the molecules (or ions) in solution move with a uniform velocity until they collide
with one another. At the collision, momentum is transferred and particles lose
speed. Since it is extremely difficult to follow the motion of each individual
particle in solution, the transport process is generally studied from a macroscopic
standpoint i.e., the movement of molecules is monitored by measuring changes in
concentration.The movement of a species of molecules (or ions) in solution can be
expressed in terms of their free energy. In an ideal solution (which in
electrochemistry means a very dilute solution), the free energy of a molecular
species is equivalent to its chemical potential (p), and can be calculated using the
following equation:
p = p o + R T in(c) (9)
where p.0 is the standard chemical potential, R the ideal gas constant (J/moVK), T
the temperature (K) and c the concentrationof the species in solution.
The chemical potential (or the free energy) is directly related to the concentrationof
the species. Thus according to equation (9), if two points in a solution are at
different concentrations,there exists between them a gradient of free energy (Fig.
2). This gradient is the driving force of the molecular diffusion from the point of
higher concentration to the point of lower concentration. Applying the second
principle of thermodynamics (the energetic dissipation due to the movement of
ions is always positive or null), the diffusion flow of a molecular or ionic species
in an ideal solution can be expressed by the following relationship:
-
J = -B c grad@)
where B is a transport coefficient (it can be a scalar for isotropic media or a vector
if the material has a heterogeneous structure). According to equation (lo), the
steady-state diffusion flow is proportional to the gradient of concentration.
Replacing equation (9) in equation (lo), one fmds the well known empirical Fick's
first law of diffusion flux :
d = - D gräd(c) (1 1)
diny
D=BRT (1+-)
dlnc
where y is the activity. To add to the complexity of the problem, the activity
coefficient is also a function of the concentrations of all the ionic species in
solution. Several expressions have been developed to calculate the value of the
activity coefficient. A description of these models is beyond the scope of this
review, and comprehensive surveys on the subject can be found in any good text
book of electrochemistry [i 1, 121.
As for the permeation process, the Laplace's equation should also be satisfied for
the transport by diffusion. For a steady-flow (the concentration is independent of
time), Laplace's equation can be solved to determine the concentration of a species
in all points of the solution:
Equation (14) is often referred to as Fick's second law. Knowing the initial and
boundary conditions, it can generally be solved using the Laplace or Fourier
transform technique.
As previously mentioned, when diffusion is solely taking place in a solution, the
diffusion coefficient (D) is only a function of the type of diffusing species and of
their concentrations. The diffusion coefficient measured under these conditions is
generally termed the free liquid diffusion coefficient (Da). However, when
passing through a porous medium, the path of the diffusing particle is not only
affected by the presence of the other particles in solution but is also affected by the
characteristics of the solid pore structure. In the case of an inert material where the
diffusing particles do not interact with the solid surfaces (which is in most cases an
over simplification),the resulting diffusion coefficient is often termed the effective
diffusion coefficient (De& or the intrinsic diffusion coefficient (Dint). If the
diffusing particles interact physically or chemically with the material, the measured
coefficient is then called the apparent diffusion coefficient (Dapp) [13, 141.
Phvsical characteristics of gorous media
As previously mentioned, equations (1) to (14) have been developed from a global
standpoint. They do not, link the values of the transport coefficient to the
microstructural characteristics of the porous medium. In their efforts to establish
such a relationship, many authors have defined various parameters to characterize
the physical (or geometrical) properties of porous materials. These geometrical
parameters are often used by researchers to develop mathematical models. The
most commonly used of these parameters are briefly reviewed in the following
paragraphs.
Among the most commonly used parameters, the porosity of a material (0)
is
usually defined as the ratio of void to total volume:
Another commonly used geometrical parameter is the specific internal area (Sa>.
This is the ratio of the internal area of the voids to the total volume and is
expressed as a reciprocal length (m2/m3):
void area
sa=
total volume
The specific internal area provides information on the refinement of the pore
Sa1 > S n means that the pores of medium 1 are
structure. For a given value of 0,
smaller than those of medium 2.
Some prefer to use the effective specific internal area (So) which is the ratio of the
internal area of the voids to the volume of solids. The effective specific internal
area is also expressed as a reciprocal length (m2/m3), and can be determined by the
following relationship:
void area
so =
volume of solids
The total porosity and the specific internal area are useful parameters that can be
(more or less) easily measured, but they are only rough estimates of the pore
structure characteristics of a material. A model that would aim to predict the
transport properties of a given material could hardly be based solely on these two
parameters. In order to provide additional information on the properties of porous
media, researchers have tried to assign mathematical definitions to develop
concepts such as the connectivity and the tortuosity of a porous network. While
these definitions can be used effectively to describe simple porous media, their
applications to intricate systems, such as hydrated cement paste and concrete, is
more problematic and generally less successful [15].
The connecrivis, factor (CF) is often qualitatively defined as the level at which a
porous structure is connected. Several mathematical defimitionsof this parameter
can be found in the literature. One of them is given in Fig. 3 for a simple porous
network. The porous medium is imagined as nodes connected by links. The
connectivity factor is determined as the subtraction of the number of links and the
number of nodes plus one. Unfortunately, such a definition can hardy be applied
to complex porous systems, and the connectivity parameter has never been
accurately determined for cement-based composites.
The rortuosify (~1/2)is another concept that has been introduced, at the beginning
of this century, in an attempt to characterize the complexity of a porous network.
Several mathematical definitions of tortuosity can be found in the literature. It
seems, however, that the definition proposed by Epstein [16] is used most
commonly:
tortuosity 71’2 =
L
where Le is the average pore length and L is the apparent length (which one can be
measured). Epstein also defines the rormsis,factor (2)as:
tortuosity factor z =
(t1
The tortuosity factor appears in equations of transport when considering the real
speed of the fluid. The use of these definitions in the modeling of transport
processes in porous materials is correct only in the case of ideal (incompressible
and non-viscous) fluids. Furthermore, the concept of tortuosity, as defined by
equation (18), can hardy be applied to characterize intricate pore systems.
To overcome this difficulty, various experimental techniques have been proposed
to indirectly measure the tortuosity of complex porous systems. Carniglia [171, for
instance, developed a procedure where the tortuosity factor is deduced from
mercury intrusion pore size distributions and diffusion measurements. Others have
deduced tortuosity (or the tortuosity factor) from electrical conductivity
measurements. These methods are based on the assumption that, for saturated
systems, the electrical tortuosity is equal to the hydraulic tortuosity which is also
equal to the diffusion tortuosity. Willie and Spangler [181 proposed calculation of
the tortuosity factor according to the following equation:
T = @ P (20)
where F is the formation factor and corresponds to the ratio of the electrical
conductivity of free electrolyte (oa)to the electrical conductivity of the entire
saturated porous solid (oeff):
F = afl (21)
(Jeff
The formation factor (F) is often referred to as the Archie factor (F) or the
Macmuilin number (Nm)in the chemical engineering literature [19,20].
For unsaturated media, Willie and Spangler [i81 also proposed the calculation of
Te and Fe according to the following relationship:
The coefficient I is termed the indexed resistivity and is a function of the degree of
saturation (0) of the porous medium. From numerous experimental data, Willie
and Spangler deduced the following relationship:
I = 0-n (23)
where n is a constant depending on the material and its level of damage.
Although, as will be seen in the following section, tortuosity factors calculated
from electrical conductivitymeasurementshave been regularly used by researchers
in the development of transport property models, the validity of these parameters
has been consistently questioned. In his comprehensive review of the flow
mechanisms in porous media, Scheidegger [ 151 mentions that no consistent
correlation can be found between the electrical and the geometrical properties of
porous materials. The application of tortuosity factors to cement-based composites
has also been criticized recently by Chatterji [21].
Instead of using parameters such as connectivity and tortuosity, many researchers
have chosen to base their models on information deduced from pore size
distributions. These distributions are generally obtained by mercury intrusion
porosimetry, capillary condensation techniques or image analyses. An example of
a pore size distribution obtained for a hydrated cement paste by mercury intrusion
porosimetry is given in Fig. 4 [3]. Results can either be expressed in dVldP (P =
mercury intrusion pressure) versus pore diameter plots or cumulative intruded
volume versus pore diameter plots. Parameters such as the total porosity, the
threshold diameter (noted dt in the figure) theoretical pore diameter or the
maximum continuous pore diameter (noted d,) and the mean pore diameter (noted
dm)can be deduced from these distributions (Fig. 4).
Although all of these methods generally yield reproducible results, none of them is
believed to give the "true" pore size distribution of complex systems such as
hydrated cement paste and concrete. As emphasized by Diamond [22], each
method covers a limited portion of the pore structure, and none of them can tally
the whole range of possible pore sizes found in cement-based composites. Results
given by mercury intrusion porosimetry, which appears to be the most commonly-
used method, have also been criticized. in addition to the fact that the distributions
obtained by this method are based on numerous simplistic assumptions [23], the
pore filling by mercury intrusion has been reported to significantly alter the pore
structure of hydrated cement paste systems [24,25]. Numerous studies have also
clearly indicated that the results obtained are strongly influenced by the sample
preparation procedure [26,27].
Empirical models
Numerous empirical models have been proposed over the past decades. All of
them have been developed using the same approach. An equation linking the value
of a given transport coefficient to the material properties are deduced from a certain
number of experimental data. In most cases, the mathematical relationship is
derived from a (more or less refined) statistical analysis of the experimental
results. Since the development of empirical equations can not be truly considered
as an innovation in the field of microstructural modeling, only a few selected
models are reviewed in the following paragraphs. Obviously, many valuable
contributions had to be set aside in the process.
Most of the empirical models found in the literature have been developed to predict
the permeability of hydrated cement pastes and concretes. The very limited number
of models specifically devoted to the prediction of the diffusivity of cement-based
materials can probably be explained by the fact that diffusion experiments are
much more time-consuming than permeation tests.
Danyushevsky and Djabarov [28] were among the f i s t researchers to develop an
empirical equation linking the porosity and the permeation coefficient of hydrated
cement pastes. Derived from the statistical analysis of a large body of experimental
data, the authors proposed the following equation:
where k is the permeation coefficient (m2) of a given cement paste, @ its porosity
and Rm its mean pore radius (pm) deduced from a mercury intrusion porosimetry
curve. The authors also developed a set of empirical relationships to determine the
values of @ and Rm from the mixture proportions. The most interesting feature of
the model proposed by Danyushevsky and Djabarov is that it relies on at least two
empirical parameters (@ and Rm)to predict the permeation coefficient.
5
K=2.8.10-10(F)
where w/c is the mixture waterhement ratio. This equation is only valid for
concrete mixtures prepared without any supplementary cementing materials. For
concrete mixtures containing fly ash, equation (27) becomes:
where f stands for the fly ash content of the mixture. Apparently, no equation has
been specifically developed for other supplementary materials such as silica fume
and slag.
As previously mentioned, there exists relatively few empirical models specifically
devoted to the diffusion process in cement-based composites. Most of the
published models are presented under the form of equations (25) and (26).For
instance, Numata et al. [37] proposed the following equation to link the diffusion
coefficient of tritium to the total porosity of cementitiouscomposites:
D = 10-10.45 a0.947 (29)
Another type of empirical models has been developed by Hansen et al. [38]. The
value of the diffusion coefficient is then directly linked to the mixture proportions.
The following equation has been derived for the diffusion of chloride ions in
blended cement materials:
--
D = 1.7 1 O ( C + F + 7.0
Although the above equations generally have fair correlation coefficients, their
ability to accurately predict the permeation or the diffusion coefficients of a wide
range of cement-based composites are quite limited. As explained by Feldman [i],
the application of models like the one of Nyame and illston [29] or that of Mehta
and Manmohan [30] is restricted to normal portland cement pastes and can not be
used to predict the transport coefficients of materials prepared with fly ash or those
of mixtures containing aggregate particles or fibres. Similar comments can be
made for the most recent models, despite the fact that more sophisticated statistical
analyses were generally used to derive the empirical equations.
The intrinsic problem of these empirical models lies in the application of statistics-
based methods to the analysis of multi-parameter phenomena. Given the number
of factors having a direct bearing on the transport properties of cement-based
materials, its is practically impossible to carry out an experimental program that
would encompass all the required test conditions. Furthermore, if such a program
were to be completed, the resulting empirical model would most probably have a
low predictive potential considering that, for instance, permeation coefficients of
cement-based composites might easily extend over more than 8 orders of
magnitude.
where q is the dynamic fluid viscosity (kg/m/s) and VF is the volume fraction of
solid material. According to Powers, VF can be deduced from the porosity of the
paste (a):
VF = 1 - (@ + 0.26) (32)
Although, equation (31)was found to give a fair agreement between observed and
calculated values for a range of temperatures, it can not be used for most cement-
based composites. Equation (31) was developed to predict flow through very fine
"gel" pores, and can not be applied to materials where the permeation process also
includes flow through larger "capillary" pores.
Following the work of Powers et al., most of the subsequent permeation models
where developed by adapting Poiseuille's law to porous media. Solving the
Navier-Stokes hydrodynamic equation for viscous and incompressible fluid flow,
Poiseuille developed a mathematical relationship where the volumetric flow rate
measured between the two pores in Fig. 1 can be linked to the pore geometry*:
k = u JIVdV (33)
surface
CxLRi - 2 xN(R)L
Sa=2 -- (35)
Vtot "lot
where Vtot is the total volume of the specimen (pores+solids), Vpi is the volume
of pore i and L is the apparent pore length. The angular brackets indicate an
average over the random values of radius distribution.
As pointed out by Garboczi [lo], the artificial parameters in these equations are N,
and L. Since, for intricate porous media, N is essentially unmeasurable, many
researchers have modified equation (33, and expressed the permeation coefficient
as a function of a,Sa and a correction factor that accounts for the complexity of
the material pore structure. The most well-known of these relationships is
probably Carman's daptation [45] of the Kozeny analysis [46] of fluid flow in
granular materials. In the simplest case where Ri is equal to a constant R for all
pores, the prediction for k then becomes:
Although Carman-Kozeny type equations can generally predict fairly well the
permeation coefficients of powdered materials (where the assumption of a
monodimensionalpore size is more reasonable), their application to high-surface
area materials has been found to be much less successful. Obviously, the
prediction of the permeation process of complex solids cannot be based solely on
such global parameters as CD and Sa.
For a given pore size distribution, assuming that the flow rate in each pore may be
calculated by Poiseuille's law, Reinhardt and Gaber [47] demonstrated that, for
incompressiblefluids, the square equivalent pore radius h2 can be estimated by:
where ,Vpi is the volume of the pore size i and Vp is the total pore volume. Then,
the permeation coefficient is equal to :
Req2
k=@-
8
Using a Carman-Kozeny approach, Reinhardt and Gaber [47] also proposed the
following equation to calculate the water permeation factor of cement-based
materials:
where R,q2 is the equivalent radius derived from a mercury intnision porosimetry
experiment and equation (37), and u is an empirical parameter obtained from a
statistical analysis of the experimental results. The coefficient n characterizes the
tortuosity and the connectivity of the porous network. From an analogy to
electrical networks, the authors found by numerical simulations that the value of n
should be equal to 8 for mortars.
Reinhardt and Gaber [47] developed another equation to predict the oxygen
permeation coefficient (compressiblefluid):
similar description of the material pore structure had been used previously by one
of the authors to predict the mechanical properties of hydrated cement pastes [50].
The Luping and Nilsson model also includes an empirical coefficient (b) which
accounts for the connectivity and tortuosity of real porous media. According to the
authors, the water permeation factor can be predicted using the following
equations:
log (K)= 2.994 - 1.091 log (Pk) (41)
1 dP
Pk = -(-)o
b o dx
dP
where (-)o is the threshold pressure head beyond which the flow rate increases
dx
linearly with the applied pressure. Pk is a parameter dependent of the material pore
structure. According to the Luping and Nilsson model, the surface forces acting on
flowing fluid particles govern the dissipated energy of transport and cannot
therefore be neglected for low pressure gradients (Fig. 7). After a threshold
pressure gradient, the viscosity forces are greater than the electrical and friction
forces.
As in previous cases, the ability of the Luping and Nilsson model to predict the
permeation properties of cement-based materials is reduced by the fact that it relies
on a statistically-determinedcoefficient. The value of the coefficient b has to be
adjusted for each type of material (hydrated cement paste, concrete, etc.).
However, the theoretical analysis, from which the model is derived, contributes to
the understanding of the permeation process.
Certainly, the most interesting information provided by this model is that the
threshold pressure gradient is very high compared with the usual pressure applied
in most laboratory experiments. For water, a pressure gradient of more than one
hundred atmospheres per centimeter (1 atmosphere = O. 113 MPa) is required to
pass the threshold. Since it decreases with the fluid viscosity, the threshold
pressure of gas should be approximately equal to a few atmospheres. Thus,
according to this model, the water permeation factors deduced from most
laboratory experiments should not be calculated using Darcy's law!
More research is needed to investigate the influence of the pressure threshold in
water permeation experiments. If the analysis made by Luping and Nilsson is
confirmed, the large test result scatter reported in most studies might be partially
explained by this pressure-threshold effect. This also implies that one should
always mention the pressure applied to measure the permeability properties.
Without this information, one cannot truly compare the results of a given series of
tests with those obtained in other studies.
Aware of the intrinsic limitations of the Carman-Kozeny approach, researchers
looked for other avenues to model the transport properties of refined porous
media. In the past decade, many of these researchers saw the percolation theory as
a very promising alternative. According to Brown and Shi [9], the term
"percolation" was first introduced in the late 1950s. Since then, percolation theory
has been used in many fields of theoreticai and applied research [51].
The percolation theory deals with disordered media in which the disorder is
characterized by a random variation in the degree of connectivity. It finds
numerous applications in fields like catalysis, capillary phenomena, single or
multiphase fluid transport and ionic diffusion. Imagine a mesh composed of
square "sites" (Fig. 8). Each site can be either occupied by a black square
(probability Pr) or is empty (probability l-Pr). If one considers two conductivities,
o, for the occupied sites and O for the empty sites, there exists a probability (Prc)
to have a path (i.e. of occupied sites) connecting plane A to plane B. PX is termed
the percolation threshold. Usually, percolation studies treat infinite problems and
Prc are thus defined for infinite paths. For Pr less than Prc only isolated, non-
spanning clusters can exist. For Pr greater than Prc, there is always a spanning
cluster. Here, Pr also represents the probability of finding a neighboring occupied
site. Using the simple example of Fig. 8, it can be shown that the formation factor
F of this mesh can be written as a function of the Probability Pr:
Rm2
k=FaQ- (44)
32
where FQ is the fraction of connected pores and Rm is the mean pore radius. For
cracked structures, the following equation should be used:
where F, is the fraction of connected cracks and wm the mean crack width.
The Gueguen and Dienes model has been severely criticized for the inherent
difficulty of measuring the values of Fw, FQ, 0 and Rm. Since, in practice, these
parameters can not be determined accurately, the use of this model is quite limited.
Relying also on the percolation theory, Katz and Thompson [53-551developed a
model where the hydraulic conductivity (permeation coefficient) of a porous rock
is a function of the maximum continuous pore radius and the electrical conductivity
of the system:
9 (P+1IP
Pf =- (47)
32 (p + 3)P+2
where F is the ratio between the diffusion coefficient of a given ionic species in
free electrolyte (Da) and the effective diffusion coefficient of the same species in
blended cement systems (Derf), the blended cement paste porosity, and Rm its
median pore size (as determined by a mercury intrusion porosimetry test). The
coefficients u and v are empirically fitted constants. KUMARand ROY established
that u and v were respectively equal to 10.3 I 2.6 and -0.41 1 0 . 1 3 for C1-
diffusion and equal to 4.6 I 2.5 and -0.62 I 0.08 for Cs+ diffusion.
The predictive value of the Kumar and Roy model remains quite limited. To extend
such a model to other ionic species, one would have to carry out a large number of
Equation (50) implies that the pore structure only offers geometric constraint to the
particle motion. The equation neglects any physical and chemical interactions
between the ionic species and the pore walls.
Equation (50) has been used successfully to predict the effective diffusion
coefficients of porous rocks [65]and of cementitious systems [14,66]. The good
correlation is quite interesting from an experimental point of view since D,ff can
readily be calculated from the values of oeff,on and Dfl which are easier to
measure. From a modeling point of view, the validity of equation (50) is limited
since it only relates one property to an other, and no link is established with the
material's microstructure.
As for the diffusivity, there exist few publications on the relationship between the
Conductivity of cement-based composites and their microstructure. In a recent
paper, Christensen et aí. [56Jreviewed the ability of various models to predict the
conductivity of hydrating cement pastes. Among the models considered, they
evaluated the validity of Archie's law 1671:
where is the porosity and u and n are constants related to the pore structure.
Equation (51) is the generalized form of Archie's law. In most cases, the constant
u is set equal to 1, and all the information on the pore structure is treated via the
exponent n.
Although Archie's law has been successfully applied to porous rocks. Christensen
et al. [56]demonstrated that it could not be used to predict the diffusivity of mature
cement paste systems. Their conclusion is in good agreement with the results of
Atkinson and Nickerson [14] (see Fig. 9) who could not find any correlation
between the measured diffusion coefficients of various cement pastes and those
calculated according to Archie's law. As a matter of fact, none of the physical
models reviewed by Christensen et al. [56] were found to have any real predictive
potential. The Garboczi and Bentz approach was the only model that seemed to
yield interesting results [68]. This model is reviewed in the following section.
A very interesting model developed to predict the apparent diffusion coefficient of
metallic porous materials has been presented recently by Carniglia [ 171. Assuming
that the pore structure is a collection of cylindrical pores, the author proposes a
method to determine the tortuosity factor from mercury intrusion porosimetry and
nitrogen vapor adsorption. In Carniglia's model, the "ink bottle'' effect and the
random orientation of pores are taken into account. His simulation gives very
encouraging results and should be tested for cementitious materials.
Networks: Networks are made of elements (cylinder, spheres, etc.) tied together
according to simple geometrical rules and whose properties are postulated by
applying basic laws. The macroscopic property (conductivity, diffusivity,
permeability, etc.) is deduced from computer simulations made on the network.
This type of approach has been widely used by geophysicists, petroleum
geologists and physicists studying the properties of homogeneous materials [74-
771. Some of the most commonly-used networks are illustrated in Fig. 10.
Holly et al. [78] used a two-dimensional tube network to predict the apparent
permeation factor of porous materials (Fig. 11). The construction of the network is
based on a random procedure using a mercury intrusion pore size distribution.
Pores are assumed to be cylindrical, and a linear relationship between the length ~
I
(distance between two nodes) and the pore radius is used in the model. At any
node, the total inflow must be equal the total outílow. Between any two nodes, the
total pressure drop is independent of the path taken. Furthermore, the sum of the
pressure drop around each loop is zero. Using Darcy's law and the Poiseuille
permeation coefficient for a cylinder, the authors calculated the apparent
permeability for various tube angles, For an angle of 45", they compared their
results to those obtained by Mehta and Manmohan [30]. The model was found to
yield higher permeation factors for hydrated cement pastes having a w/c < 0.6 and
lower values for w/c > 0.6. Globally, the errors ranged from 40 % to 1100 %.
Whereas such a network model cannot provide very accurate results, it is useful to
explain some experimental observations. The model indicates, for instance, that
the permeation factor cannot be defined in terms of an average pore size and
porosity without taking into account the pore size distribution. Furthermore, the
flow in a pore is found to be affected not only by the pore radius but also by the
radius of the surrounding pores. Using mathematically-derived pore size
distributions, the authors studied the influence of each pore category on the global
flow.
Daïan and Saliba [79] modeled the porous structure of cement-based materials
using cubic network structures (Fig. 12). Using a mercury intrusion porosimetry
curve and a random procedure, the model represents the material pore structure by
glueing together an array of cubes. In order to simulate the sorption and desorption
mechanisms, and water transport by diffusion or viscous flow, the Laplace law is
used in the model as a local law which describes the intrusion of pores, and the
Kelvin-Laplace law is used to calculate the state of the fluid (condensed of
vaporized) in each pore as a function of the relative humidity. Considering the state
of the fluid, a diffusion or a permeation law is applied to describe the mass
transport process. The equivalent conductivity of the network can also be
calculated since it corresponds to the macroscopic or apparent permeability of the
material.
Although the results of Daïan and Saliba's model have not been compared to
experimental data, the authors simulated various experiments such as mercury
intrusion porosimetry or sorption and desorption isotherms. Fig. 13 shows an
example of a simulation of an isothermal sorption-desorption experiment [80]. As
can be seen, the model simulates properly the hystheresis phenomenon also called
the "ink bottle" effect. From similar simulations, the model can rebuild the "real"
pore- size distribution of a material according to some assumptions on the
probability that larger pores can only be invaded if the smaller pores connected to
them are already invaded. If it is found to yield satisfactory results, such a model
could certainly be used successfully for various applications. It could, for
instance, provide pore size distributions that could be used as input data for other
predictive models.
Euclidean network models, such as the two previous models, are limited due to the
high number of elements required to describe the multi-scale nature of cement-
based composites. All these elements are stored in the computer memory.
Although such a procedure facilitates the modeling of non-stationary phenomena
For mass transport processes, the elements are represented by cylinders. As can be
seen in Fig. 14 and Fig. 15, the various sizes of the four different cylindrical
pores of a given element are established from a mercury intrusion porosimetry
curve. The size of each pore is randomly attributed. The various transport
phenomena are reproduced using classical laws (such as the Kelvin-Laplace
equation, Darcy and Fick's laws) as suggested by Daïan and Saliba [79] and
Quénard and sallée [84]. This basic relation provides a good description of
physical phenomenon such as sorption (Fig. 16). In the specific case of viscous
flow, the model calculates a "global" permeation factor (Knet) for the entire
network. The following relationship has been established between Knet and the
permeation factor of the porous medium:
K = <D Knet
where n is an empirical parameter. A value of 6 < n < 8 tends to yield satisfactory
results for concretes and mortars. For hydrated cement pastes, n has been found to
be a linear function of the w/c. The physical meaning of n has still to be
established. On-going research on the subject should provide more information of
the subject in a near future [83].
VF being the solid volume fraction (the number of pixels considered as solids at
the initial state).
The hydration process is simulated by moving pixels in random directions (Fig.
17). The model considers various scenarios to reproduce the precipitation of C-S-
H and CH crystals within the water filled pores, and can account for the influence
of various parameters (such as the w/c, the degree of hydration and the type of
binder) on the hydrated cement paste microstructure. The model has also been used
to simulate the formation of cement-aggregate interfacial zones and to study the
effect of supplementary cementing materials (such as silica fume) on the porosity
in these areas [88, 891.
Relying on the percolation theory, Garboczi and Bentz could calculate the
connectivity of the system [90]. When a spanning cluster exists, the pixels which
belong to it are termed "burned pixels". The total number of burned pixels is
considered as representing the connectivity of the porous medium at a certain level
of hydration. This connectivity is compared with the total pore space (porosity).
Immediately after mixing, the fraction of burned pixels is 1. As hydration proceeds
on, the model indicates that there exists a threshold porosity (pc) for which the
number of burned pixels is zero. In such a case, the porosity is closed and no mass
transport can take place through the capillary porosity. For a 0.45 w/c paste, pc is
equal to O. 18.
Garboczi and Bentz also used their model to simulate the development of some
mass transport properties during the hydration process [68, 91, 921. They
reproduced, for instance, the effect of connected paste-aggregate interfacial zones
on the mercury injection during a porosimetry experiment [91]. The authors also
studied the diffusion mechanisms in hydrated cement systems. In agreement with
the conclusions of Atkinson and Nickerson [14] and Christensen et al. [56], they
showed that, if the diffusion properties of young and porous pastes could be
estimated by Archie's law, such a model could not reliably predict the diffusion
coefficient of more mature systems. Garboczi and Bentz proposed the following
equation to calculate the diffusion coefficientof these systems:
CONCLUDING REMARKS
As emphasized by Hall [4], no property more smkingly emphasizes the diversity
of cement-based materiais than permeation. The same remark could be made for
the diffusion process. The extreme complexity of the microstructure of
cementitious materials clearly complicates the modeling of the mass transfer
properties. While some models can predict reasonably the mass transport
mechanisms in simpler structures, such as porous rocks and soils, their direct
applications to the prediction of the properties of cement-based composites is
clearly less successful.
The recent development of models devoted specifically to cement-based
composites has yielded promising results. It appears clearly that the most
promising of these models are those which consider the multi-scale nature of the
pore structures. The combinations of these analytical and numerical models with
other mathematical tools, such as fractal analysis, will certainly contribute to
significantly improve their predictive potential. The application of fractal analysis
has proven to be extremely fruitful in the development of transport property
models in other fields [93,94].
ACKNOWLEDGEMENTS
The authors are grateful to Électricité de France (EDF) for their financial support.
The authors also gratefully acknowledge M. Pigeon and D. Breysse for their
precious collaboration
REFERENCES
Feldman, R.F., "Pore structure, permeability and diffusivity as related to
durability", Eighth International Congress on the Chemistry of Cement,
Rio de Janeiro, Brazil, Vol. 1, Theme 4, 1986, pp. 336-356.
Young, J.F., "A review of the pore structure of cement paste and concrete
and its influence on permeability", AC1 Special Publication SP- 108, 1988,
p. 1-18.
Buil, M., Ollivier, J.P., "Conception des bétons: La structure poreuse", in
La Durabilité des Bétons, Edited by J. Baron and J.P. Ollivier, Presses de
1'Ecoles Nationales des Ponts et Chaussées, Paris, France, 1992, pp. 57-
106.
Hall, C., "Bamer performance of concrete: A review of fluid transport
theory", Materials and Structures/Matériaux et Constructions, Vol. 27,
1994, pp. 291-306.
Beaudoin, J.J., Feldman, R.F., Tumidajski, P.J., "Pore structure of
hardened cement pastes and its influence on properties", Advanced Cement
Based Materials, Vol. 1, No5,1994, pp. 224-236.
Chatterji, S., Kawamura, M., "A critical reappraisal of ion diffusion
through cement based materials - Part 1: Sample preparation,
Portland cement and ticalcium silicate pastes -Part III: Specific surface
area and permeability", Cement and concrete research, Vol. 21, No 6,
1991, pp 975-982, 1991.
Li, S., Roy, D.M., "Investigation of the relation between porosity, pore
structure and C1- diffusion of fly ash and blended cement pastes", Cement
Concrete Research, Vol. 16, 1986, pp. 749-756.
Graf, H, Setzer, M.J., "Influence of water cement ratio and curing on the
permeability and structure of hardened cement paste and concrete, in Pore
Structure and Permeability of Cementitious Materials", Edited by L.R.
Roberts and J.P. Skalny, Materials Research Society, Vol. 137, 1988, pp.
337-347.
Costa, U., Massaza, F., "Permeability and pore structure of cement
pastes", Materials Engineering, Vol. I, N "2, 1989, pp. 459-466.
Hedegaard, S.E., Hansen, T.C., "Water permeability of fly ash
concretes", Materials and Structures/Matériauxet Constructions, Vol. 25,
1992, pp. 381-387.
Numata, S . , Amano, H., Minami, K., "Diffusion of tritiated water in
cement materials", Journal of Nuclear Materials, Vol. 171, 1990, pp. 373-
380.
Hansen, T.C., Jensen, J., Johannesson, T., "Chloride diffusion and
corrosion initiation of steel reinforcement in fly-ash concrete", Building
Materials Laboratory, Technical University of Denmark, 1986, 13 p.
Powers, T.C., Copeland, L.E., Hayes, J.C., Mann, H.M., "Permeability
of Portland cement paste", Journal of the American Concrete Institute, Vol.
51, No 3, 1955. pp. 285-298.
Powers, T.C., Copeland, L.E., Mann, H.M., "Capillary continuity or
discontinuity in cement pastes", Journal of the Portland Cement
Association Research and Development Laboratories, Vol. 1, No 2, 1959,
pp. 38-48.
Powers, T.C., Mann, H.M., Copeland, L.E., "Flow of water in hardened
cement paste", Highway Research Board, Special Report 40, 1958 , pp.
308-323.
Stienour, H.H., "Rate of sedimentation: Nonflocculated suspensions of
uniform spheres", Industrial and Engineering Chemistry, Vol. 36, 1944,
pp. 618-624.
Stienour, H.H., "Suspensions of uniform angular particles", Industrial and
Engineering Chemistry, Vol. 36,1944, pp. 840-847.
Stienour, H.H., "Concentrated flocculated suspensions of powders",
Indusmal and Engineering Chemistry, Vol. 36, 1944, pp. 901-907.
i491 Luping, T., Nilsson, L.O., "A study of the quantitative relationship
between permeability and pore size distribution of hardened cement
pastes", Cement and Concrete Research, Vol. 22, 1992 pp 541-550.
i501 Luping, T., "A study of the quantitative relationship between strength and
pore-size distribution of porous materials", Cement and Concrete Research,
Vol. 16, No 1, 1986 pp. 87-96.
De Gennes, P.G., "La percolation: un concept unificateur", La Recherche,
Vol. 7, No 72, 1976 pp. 919-927.
Gueguen, Y., Dienes, J., "Transport properties of rocks from statistics and
percolation", Mathematical Geology, Vol. 21, 1989pp. 460-463.
Katz, A.J., Thompson, A.H., "Fractal sandstone pores: Implications for
conductivity and pore formation", Physical Review Letters, Vol. 54, No
12, 1985 pp. 1325-1328.
Katz, A.J., Thompson, A.H., "Quantitative prediction of permeability in
porous rock", Physical Review B, Vol. 34, No 11, 1986 pp. 8179-8181.
Katz, A.J., Thompson, A.H., "Prediction of rock electrical conductivity
from mercury injection measurements", Journal of Geophysical Research,
Vol. 92, No B1, 1987 pp. 599-607.
Christensen, B.J., Coverdale, R.T., Olson, R.A., Ford, S.J., Garboczi,
E.J., Jennings, H.M., Mason, T.O., "Impedance spectroscopy of
hydrating cement-based materials: Measurements, Interpretation, and
Application", Journal of the American Ceramic Society, Vol. 77, No 11,
1994 pp. 2789-2804.
Hornain, H., Regourd, M.," Microcracking of concrete, Eighth
International Congress on the Chemistry of Cement", Rio de Janeiro,
Brazil, Vol. V, Theme 4, 1986 pp. 53-59.
Ollivier, J.P., "A non destructive procedue to observe the microcracks of
concrete by scanning electron microscopy", Cement Concrete Research,
Vol. 15, 1985 pp. 1055-1060.
Bier, T.A., Ludirdja, D., Young, J.F., Berger, R.L., The effect of pore
structure and cracking on the permeability of concrete, in Pore Structure
and Permeability of Cementitious Materials, Edited by L.R. Roberts and
J.P. Skalny, Materials Research Society, Vol. 137, 1989 pp. 235-241.
Vuorinan, H.," Applications of diffusion theory' to permeability tests on
concrete - Part II: Pressure saturation test on concrete and coefficient of
permeability", Magazine of Concrete Research, Vol. 37, No 137, 1985 pp.
153-161.
Gérard, B., Breysse, D., Ammouche, A., Houdusse, O., "Cracking and
permeability of concrete under tension", Accepted for publication in
Materials and Structures/Maténauxet Constructions, 1995.
Arsenault, J., " Caractérisation de l'état endommagé dun matériau et mise
au point d'un essai biaxial", Mémoire de DEA-MAISE, Ecole Normale
Supérieure de Cachan, Cachan, France, 30 p, 1994.
Daian, J.F., Ke, X., Quénard, D., "Invasion and transport processes in
multiscale model structures for porous media", COPS III IUPAC
Symposium on the characterization of porous solids, Marseille, France,
May 9-12, 12 p, 1993.
Kumar, A., Roy, D.M.," Pore structure and ionic diffusion in admixture
blended portland cement systems", Eighth International Congress on the
Chemistry of Cement, Rio de Janeiro, Brazil, Vol. V, Theme 4, 1986 pp.
73-79.
Brace, W.F., "Permeability from resistivity and pore shape", Journal of
Geophysical Research, Vol. 82, No23, 1977 pp. 3343-3349.
Ping, G., Ping, X., Fu, Y., Beaudoin, J.J., "Microstructural
characterization of cementitious materials: Conductivity and impedance
methods, in Materials Science of Concrete", Vol. 4,Edited by J. Skalny
and S. Mindess, 1994.
Archie, C.E., "The elecmcal resistivity log as an aid in determining some
reservoir characteristics", AIME Transactions, Vol. 46, 1942 pp. 54.
Garboczi, E.J., Bentz, D.P., "Computer-simulations of the diffusivity of
cement-based materials", Journal of Materials Science, Vol. 27, 1992 pp.
2083-2092.
Huet, C.," Application of variational concepts to size effect in elastic
heterogeneous bodies", Journal of Mechanics and Physics of Solids, Vol.
38, No 6, 1990 pp. 813-841.
Suquet, P.M.," Méthodes d'homogénéisation en mécanique des solides, in
Comportement rhéologique et structures des matériaux", Quinzième
Colloque du Groupe Français de Rhéologie, Paris, France, 1980 pp. 87-
128.
Pore 2
Fig. 1-Principle of moss transfer by permeation. The driving force is a pressure gradient
Concentration 2
Concentration 1
C l > > c2
Fig. 2-Principle of moss transfer by diffusion. The driving force is a concentration gradient
w
CF = N - m + 1 N: number of links m: number of nodes
Link
CF=l CF=5
1 o-6
1 O-’
-- o-8
o)
1
E -9
y 10
-
c
z
g löl(
E
k 10-l1
-0
a
3 10-12
2
1ö14
-15
10
(days) (months)
T i m e of hydration
Fig. %Influence of hydration time and wlc on water permeation factor (from Ref. 29)
_-
32 12
a) Cylinder b) Two parallel planes
Fig. bApplicotion of Poiseuille's equation for two simple cases. Permeation coefficient
is given as a function of pore geometry
al
P
c
m
L Darcian
I domain
-
O
U
Fig. 7-Influence of pressure gradient on permeation process. Above the threshold pressure
gradient, friction and electrical surface forces can be neglected (from Ref. 49)
A B
loB
-1 o -7
7y>
N E
s2
7o -8
10-1;
Waterkernent ratio
Fig. 9-Diffusion coefficient as a function of ionic species and w/c ratio (from Ref. 14)
f J J1 + J Z + J 3 + J 4 = 0
Link
Site
Fig. 12-Cubic network model for cementitious materials (from Ref. 79)
Relative humidity
Fig, 13-Simulation of sorption and desorption isotherms of a mortar with a cubic network
(from Ref. 79)
Fig. 14-Random pore diameter, network pore unit (from Ref. 81)
d2
Level 1->2
Level 2->3
Level 3-74
O 20 40 60 80 100 O 20 40 60 80 100
RH (x 100) RH (x 100)
1.c
0.8
B
2 0.e
o
O
c
.-c
o 0.4
0
2
0.2
0.16
wic ratio O
8
.-5
u)
0.12 - O 0.45
O 0.60
u
E
U
2
.- 0.08
d
-
ia
al
0.04
/? O
,G/
O 1 5
APPENDIX A
NOTATION
B: diffusion transport coefficient
C: ionic concentration (movi)
C: cement Content (kg/m3)
CF connectivityfactor
dequ: equivalent pore diameter (m)
dm: mean pore diameter (m)
dt: threshold diameter (m)
d& maximum continuouspore diameter (m)
D diffusion coefficient (m2/s)
Dam: apparent diffusion coefficient (mZ/s)
J&fG effective diffusion coefficient (m2/s)
Dfl: diffusion coefficient in free fluid (m2/s)
Dint: intrinsic diffusion coefficient (m2/s)
f: fly ash content (kg/m3)
F: formation factor
Fe: unsaturated medium formation factor
FQ: fraction of connected pores
Fw: fraction of connected cracks
g: acceleration due to gravity (= 9.8lm/s2)
h hydraulic head (m)
I indexed resistivity
J volumetric flow (m3/s/m2 or moie/s/m2)
kb: Boltzman number
k: intrinsic permeation coefficient (m2)
k': modified permeation coefficient (m3 s k g )
IC permeation factor ( 4 s )
L: apparent pore length (m)
Le: mean pore length (m)
n: empirical pore structurefactor
Nm: Macmuliin number
P: percolation exponent
Pf: Katz-Thompson constant
P: pressure (Pa)
pr: probability of occupancy
Pcr: percolation threshold or critical probability of occ:upancy
Pk: threshold pressure factor (Pa/m)
R:
- ideai gas constant (J/mole/K)
Fi: molar gas constant (J/kg/K)
%: equivalent pore radius (m)
R,: mean pore radius (m)
Rt: threshold radius (m)
L: maximum continuous pore radius (m)
Sa: specific internai surface (m2/m3)
So: effectiveinternal surface (m2/m3)
t: time (s)
T: temperature (K)
empiricai constants
volume fraction of the solid material
total pore volume
volume of the pore of size i (m3)
total volume (pores+solids) (m3)
crack width (m)
mean crack width (m)
water content (kg/m3)
coordinates (m)
valency of ionic species i
APPENDIX B
MATHEMATICAL OPERATORS
Gradient of a scalar
grad(f) is termed the gradient of the scalar f(x, y, z). grad(f) is a vector :
ai j + -kaf
ar. + -
grad(f)= -I
ax ay az
a . -aI +. - ak
V = -i+
ax ay az
grad(f) = Vf = -i
ar. + -
ar j.+ -k
af
ax ay az
Divergence of a vector
Let s(x, y,z) be a differentiable vector, where x, y, z are Cartesian coordinates, and
Ji, J2, J3 be the components of J(x,y,z). Then the divergence is defined :
- aJ ,
div(J) = 1
aJ
A+ aJ3
-
ax ay az
Divergence of a gradient
If f(x, y, z) is a twice differentiable scalar function then :
SP 154-11
211
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 S P - 1 5 4 95 =
Ob62949 0522406 OTb
212 Chatveera and Nirnityongskul
INTRODUCTION
NOTATION
G,., c;,= shear inodulus i n the uiicracked range of composite aiid ceineiit
paste, irspcct ively;
y,,,v = total volume fraction of cenient paste matrix aiid pulped îiber,
" respectively:
EXPERIMENT
Materials
(a) Modified Rice Husk Asli--The major characteristics of RHA are its high
water demaiid aiid low fineness as compared with coiidensed silica fume. Another
important problem is the dispersion of RHA particles uniformly in the inix. To
increase fineness, the rccornrnended grinding time for RHA was 45 ininutes (3,4).
Ilowcver, the degree of fineness achieved was riot sufficient to make high-sireiigth
concrete. To solve this problem, rice husk ash must be ground in the gtinding
machine fora long durationoftitile i n order toachieve avery high degree of fineness.
A grinding time uf 1 hour and 15 ininutes was proposed and this was maintained
throughout thc whole testing program.
(b) Sisal Pulp Fibers--A chemical pulping process called "kraft process" is
introduced in this study by using a boiler having temperature of 100 "C to treat the
fibers. NaOH and N,S having concentrations of 48 and 32 gm/L respectively were
combined and used as chemical agents for the treatment of fibers. The boiling time,
beating time and casting pressure were fixed at 8 hours, 30 minutes and 0.28 MPa,
respectively.
Since the mixture of sisal pulp and mortar is a lightweight coiiiposite material,
the casting process should be done under pressure. The formwork was made o í
wood and a I .6 min-thick steel sheet. The formwork consisted of the upper and
lower parts which werecoriiiected together with I6 min diameter bolts. The details
of the formwork are shown in Fig. I .
The pulp was beaten for 30 minutes to disintegrate the pulp which was mixed
with the cementitious inaterial for 5 ininutes. The mixing was done in such a way
that all the puips were coated with mortar uniformly. The mixing was done in two
operations. When the mixing procedure was íinished, the mixture was then placed
i n the formwork and pressed by hands i n order to check the uniformity. The casting
process had to bedoiie under pressure by applying load through a hydraulic cylinder.
In this study, the load corresponded to the required casting pressure. All bolts were
fastened tightly. The specimens were kept foi 24 hours before being removed from
the formwork and cured in the fog room at 100% relative humidity at 28 "C for 28
days. Subsequently, specimens were cut into small pieces according to the type of
testing. For durability test, the specimens were subjected to simulated aging cycles
until the time oí testing.
The physical and mechanical properties of sisal fiber and pulps were
determined, and are shown in Table I . With a natural organic fiber, it is rather
diîficult to obtain the exact value of the ultimate tensile strength or modulus of
elasticity of sisal fiber and pulps because of the variation in cross section, moisture
content, age of fiber and pulp and some defects existing i n them.
Regarding tlieeffect of pulp type on the flexural strength (Fig. 3), for simulated
aging cycles the durability of the composite containing 5% bamboo pulp was found
to be better than that of composites containing 5% pine and sisal pulps. After 48
simulated aging cycles, the increase i n strength of the bamboo pulp-mortar
composite was reinarkably higher than that the control composite. It was noted
that in this respect the bamboo pulp-mortar composite and pine pulp-mortar
composite showed a similar behavior.
From Fig. 5, it can be observed that the moduli of resilience of the sisal
pulp-mortar compositc boards increased with the pulp volume fraction. In addition,
the rate of decrease i n the modulus of resilience after being subjected to simulated
aging cycles was ratlier low compared to the control. This cati be explained by the
fact that more tlie pulp content, the less the cemeiit mortar, and consequently less
alkalinity in the system. This means that the pulp suffered less attack from alkaline
pore water.
The results are graphically presented in Fig. 7. It was obscrvcd that the effect
of using clieinical fiber treatineiit resulted iii higher toughness than the control when
subjected to siiniilated aging cycles. The results indicated that fiber treatineiit using
some cheiiiical agents (NaOl-I arid N,S) and MRHA us cement replacement were
eífective in improving the ductility of the pulp-mortar coinposites subjected to
siinulatcd aging cycles. After O- I2 simulated aging cycles, the sisal pulp-mortar
composite coiitaiiiiiig 5% pulp volume fraction showed the highest modulus of
toughness. The toughness of the sisiil pulp-mortar composite containiiig 7% pulp
volume fractioii was similar to that of the sisal pulp-mortar composite containing
9% pulp volumc fraction. For both the control composite and the sisal pulp-mortar
composite coiitziiiiing 3% pulp volume fraction, it was found that the toughness was
the lowest. This caii be explained by the fact that the inore the cemeiit mortar, the
less the pulp content, aiid consequently, tlie more tlie alkalinity. This means that
tlie alkaline attack was relatively stronger.
identical size and matrix composition, tested under similar conditions. The resultant
nondiinensional index, I,, represents the relative improvement iii the energy
absorption capacity due to the inclusion of fibers, as shown in Fig. 4. It is an index
for comparing the relative energy absorption of different fiber mixtures.
The data from the nondiinensional index study are presented in Table 2. It
was also observed that the effect of using pulp resulted in higher index than the
control when sub,jected to simulated aging cycles. A decrease in the rate of
nondimensional index was also observed in the case of sisal pulp-mortar composites
containing 5% pulp volume fraction. Regarding the effect of piilp type, it was also
found that, when subjected to simulated aging cycles, the composites containing
5% bamboo pulp showed a 9% increase in the nondiinensional index. This index
was compared to that of the control composite which was not subjected to any
simulated aging cycle.
In the expansion test, it was observed that most of the expansion occurred
during the first 24 hours of submersion and, i n general, the expansions were quite
low. The maximum expansion for all tested specimens was found to be much less
than 0.5%. as specified by ASTM C208-72. The composite containing 5% sisal
pulp resulted in a higher expansion than that containing the same amount of pine
pulp. The expansion for the bamboo pulp-mortar was one of the lowest.
The results from tlie impact resistance test showed that an increase in pulp
volutne fraction caused an increase in impact resistance. The maximum index of
impact resistance for the sisal pulp-mortar composites was 33.25, especially at 9%
pulp volume fraction when very tiny cracks appeared on the unloaded face. The
index for the control composite was 26.25. A ductile failure for tlie pulp reinforced
composites was observed in contrast to the brittle failure exhibited by the control
composite. Regarding the effect of pulp type, it was found that the coniposite board
containing 5% sisal pulp exhibited a higher index value than that containing 5%
bamboo pulp. The index values for the pine and bamboo pulps composites were
lower than the control.
E, = E,,,,V,,,+O.06IE,,Vp (2)
I, = '3",,,V!"
0, (3)
E, =E,,,~Vn,+0.061E,,V,, (4)
CONCLUSIONS,
REFERENCES
I . ßentur, A. and Akers, S.A.S. (1989) The Microstructure aiid Aging of Cellulose
Fiber Reinforced Cement Composites Cured in a Normal Environment.
International Journal of Cement Composites and Light-Weight Concrcte, I I (2),
99- 109.
2. Akers, S.A.S. and Studinka, J.B. (1989) Aging Behavior of Cellulose Fiber
Cetnent Composites in Natural Weathering and Accelerated Tests. International
Journal of Cement and Light Weight Concrete, I1(2), 93-97.
3. Islam, M.S. (198 I ) Grinding Methods and Its Effect on Reactivity of Rice Husk
Ash. M.Eng. Thesis No. ST-8 1-7, Asian Institute of Technology, ßangkok,
Thai land.
4.We, A.B. ( 198 I ) Production of RHA and Its Applications in Mortar and Concrete.
M.Eng. Thesis No. ST-81-20, Asian Institute of Technology, Baiigkok, Thailand.
6. Wu, M.Z. ( I 989) Use of Sisal as Fiber Reinforcement in Rice Husk Ash Mortar.
M.Eiig. Thesis No. ST-89-20, Asiaii Institute of Technology, Bangkok, Thailand.
Il
Diameter, min
Descripiion Sisal Fiber
O. 15-0.26
Sisal Pulp
I Bainhoo
Pub
Pille Pulp
0.0305
Length, inin 1200-1 500 2.48
Bulk Specific Gravity 0.38 0.41
Bulk Specific Gravity (SSD) 0.82 I .O9
Apparent Spcciric Ci-avity 0.69 1.46 i .27
Moisture Content. 7'0 1 1.00 0.69
Water Absorpiion, 70 119.0 52.8 189.5 168.8
Uliirnaie Tensile Strengili, MPa 297.83
Modulus of Elasticity, MPa I 1366
Equilihriuin Iiiipsci
No’ I 5.394
Shrinkagc
(%)
0.0842
Resisinnce
(Indcx Nuinhcr
26.25
5.225 5.784 0.0 I33 0.0883 24.00
Sisal Pulp 7.430 8.306 0.0205 0.0867 32.00
Sisal PiiII> 7.637 8.089 0.0135 0.0840 32.50
Sisal Pulp 9.160 o. I0 15 33.25
Haiiihoo Pulp 7.544 8.340 0.0135 0.0907 25.50
Pille I’ulp 5.070 5.680 0.0172 0.06 I 2 17.00
i
NO. (Ml’a) ( x 1 0 4 MPo)
Expt. Expi./llicory
-
-
Tlicoi-y ïïicory EX]”. Ilx])l.lflicoi-y
3 Sisal Pulp I .61 I. I 4 0.708 1.944 2.174 1.1 I X
5 Sisal Pulp 2.69 I .78 0.662 1.906 2.000 1 .rw
7 Sisal Pulp 3.76 I .64 0.436 1.869 2.000 I .O70
9 Sisal I’ulp 4.84 0.85 o. I76 1.83 I 1.181 0.645
5 Baiiihoo Pulp 6.74 0.88 0.131 I ,938 2.000 I,037
-
-
-
5 Pinc Pulp
8 8
6 6
4 4
2 2
O O
Number of Simulated Aging Cycles (Cycle)
B Nunc Ed Sisal Pulp D Dnniliuo Pulp E3 Pine pulp
100
80
60
40
20
%
SP 154-12
Synopsis: This paper reports the results of an investigation of the effect of using
ground granulated blast-furnace slag to prevent alkali-aggregate reaction damage to con-
crete.
229
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 W 0662949 0522424 O11 W
I
230 Nishibayashi, Kuroda, and Okawa
INTRODUCTION
The Japan Concrete Institute (JCI) recommends that one of the following
measures be taken to prevent AAR in new conmtc structures.
1. Use aggregates that have been found to be nonreactive through onc of the
following: chemical method (JC1 AAR- i ) , mortar-bar method (JCI AAR-Z), and the
concrete specimen method (JCI AAR-3). ’)
2. Usc low-alkali portland cement (JIS R 5210), ¡.c. portland ccment containing
less than 0.60% of Na,O cquivalcnt (Na,Ocg=%Na,O + 0.658 x %,K,O).
The mechanism for the prevention of AAR by using blast-furnace slag can be
explained both in terms of the dilution effect of blast-furnace slag on harmful alkali
and in terms of the physical o r stabilizing effects of the slag on the alkali. I-’’ It is
believed that the slag to prevent AAR damge is a combination of these mechanisms.
EXPERIMENT
Q u t h of Testing
Test M e t M
1. Mortar: The mixture proportion of mortar was as follows: C:S=I :2.25 (by
mass) with unit water content selected by a trial test to obtain a flow valuc of 170 to
190 mm. Slag replacements were O, 25, 40, 50, 65,and 75%. Total d k a l i contents
(Na,O equivalent) were 0.5, 1 .O, and 1.5%. The dimensions of specimen5 were 4í) x 40
x i60 mm.
The blending ratios of the cement and thc slag were 100:Ot 50:50, and
30:70, and of reactive coarse aggregates were O , 50, and 100%. Total alkali contcnts
(Na,O equivalent) were 1.0, 1.5, 2.0, and 2.5%. The dimensions of spccimcris wcrc
100 x 100 x 400 cm.
blast-furnace slag is evaluated on the basis of the reduction in the dcgrcc of cxpan-
sion [Re) defined by the following equation:
RESULTS
at the age of 12 months at each alkali level is shown in Fig. 1. Solid lines indicate
alkali contents in the total binding material (portland ccmcnt t slag) when the slag
contains no alkali. Dashed lines indicate alkali contents in portland cement.
Fig. 1 shows that when normal portland ccmcnt is replaced with slag. the degrcc
of expansion of mortar is greatly reduced. In terms of the alkali content in the total
binding material (solid line), as the slag content incrcascs, the dcgrcc of expansion of
mortar demases linearly.
Fig. 2 shows the relationship between slag content and Re, and indicates the
reduction in the degree of expansion of mortar containing slag vs. standard cernent
without slag.
cement alone, and concrete in which 50% of the normal portlmd cement has been
replaced by slag, respectively. Fig. 3 shows that expansion of concrete increases as
much as 0.7% when the ratio of RCíRF is 0/100 at the age of 12 months. On the other
hand, when using reactive fine aggregate (RF) and reactive coarse aggregate (RC)
with various mixing ratios (RC/RF=50/100, 100/50, 100/0, and lOO/lOO), the dcgrcc
of expansion falls to below 3% at thc age of 12 months. in addition, when cement is
replaced with slag, the degree of expansion at all ratios is greatly reduced to below
0.3%. However, in the case of RC/RF=I00/0, the pcssimum value is discerned re-
gardless of the slag replacement.
On the other hand, ignoring the alkali content in thc blast-furnace slag,
the total alkali content is 3.4 kdm’. In other words, even if ccmcnt is replaced yith
slag, and thc total alkali content excccds the regulation-limited value of 3.0 kgím’ to
prevent AAR, the expansion docs not cxcecd the regulation-limit value of O. 1 %.
Therefore, it can be presumed that the dominating factor for expansion is not only thc
alkali-ion concentration.
Fig. 8 shows the changc in rclativc clastic modulus over timc. Relative clastic
modulus is given bascd on thc value at the age of 0.3 months as 100. The changc in
the relative clastic modulus varied with the mixing ratio of reactive-aggregate. When
50% cement is replaced with slag, all rclativc elastic moduli increase to 85% or inorc.
When the mixing ratios of concrete are RCíRF=100/0 and lOO/lOO, the relative clastic
moduli increase to approximately 100%.
When reactive fine aggregate is used (Fig. 13), and 50% of normal por-
tland cement is replaced with slag, and the alkali content is 2.074, o r less. expansion
can be reduced by 60%. When the alkali content is 2.5%, expansion is reduced by
35-4095.
When the mixing ratios are RC/RF=100/100 and RC/RF=50/50 (Fig. 14)
and the alkali content is 1.5% or less, the rcduction in the degree of expansion is
almost loo%, ¡.c. expansion can be sufficiently prevented. However, as thc alkali
content increases, the reduction of expansion decreases.
The relationship between the Re and the slag content in concrete is given in Fig.
15 which shows that the absolute volume of binding material in concrctc is smaller
than that in mortar. Therefore, the Re values are generally smallcr. When the alkali
content is 1.0-1.5% and with 50% slag replacement, the Re value is 80% o r more.
When the alkali content is 2.0-2.5%, 70% or more slag replacement is presumably
required.
CONCLUSION
cement with blast-furnace slag, as well as the expansion characteristics of mortar and
concrete containing blast-furnace slag, and the influence of the mixing ratio of reac-
tive-agregate and the alkali content.
2. The expansion-reducti«n effect due blast-furnace varies with thc mixing ratio
of the reactivc-aggregate. On the other hand, when blast-furnacc slag is used and
only cxpansion-reduction effects duc to the alkali-dilution effect of the slag are taken
into account, expansion in concrete will not occur eve? if the total alkali value ex-
ceeds the regulated value for AAR which is 3.0kgim.’. It is presumed that blast-
furnace slag exhibits strong stabilizing effects on the alkali-ions, an adsorbing fixa-
tion effect on water and ions, as well as an alkali-dilution effect.
3. Even if ccmcnt is replaced with blast-furnace slag, the alkali content greatly
affects the dcgrcc of expansion which varies with reactive-aggregatc content. Except
in cases where reactive-aggregate is used alone, a 50% slag replacement can reduce
the degree of expansion t o a level at which deleterious expansion will nut occur.
Whcn reactive fine aggregate is used, the degree of expansion becomes greater than
that at other mixing ratios. Whcn a 50% ci€ the cement is replaced with slag with
i .O% or more alkali content, the degree of expansion bcccimes 0.1% or morc.
4. With a 50% slag rcplacemcnt, the limiting value of thc total alkali content
becomes roughly 8.0kg/m3 when reactive coarse aggregate is used. Taking account
of the alkali contcnt in slag, the total alkali content becomcs roughly Y.0k@m3.
REFERENCE
4. Nixon, P. J. and M. E. Gaze., The Effectiveness of Fly Ashes and Granulated Blast
Furnace Slags in Preventing AAR, Proc. 6th Int. Conf. on AAR in Concrete, Copen-
hagen, 1Y83, pp. 61-68
5. Sinis, i . “The Influence of Ground Granulated Blast Furnace Slag on thc Alkali
Reactivity of Flint Aggregate Concrete in the United Kingdom, Proc. 6th Int. Ccinf. on
AAR in Concrete, Copenhagen, 1983, pp. 69-84
* entrupped air
W : water, C:cement, Sl:slag, S:fine agg., G:coarse agg.
n
$?
W
c
O. 8
o.6
-
- 0.5
0.5
1.0
OPC
OPC+slag
OPC
O
-;
o.4 _c_ 1.0 OPC+slag
t: -cr_ 1.5 OPC
m
P
o.2
O. O
O 20 40 60 80
Slag contents (%)
Fig. 1-Relationships between slag contents and expansion (alkali content of slag:
O percent, mortar)
1 O0
80
n
6 0
8
v
a, 40
U
20
O
O 20 40 6 0 80
Slag contents (%)
;0 .
.-
v)
-
---I-
100/100
Oll00
50/100
K 0. 2
m .....................
P
i 0. 0 S l a g cont. : 0%
A l k a l i : 1.5%
Rc : reactive coarse agg.
Rf : r e a c t i v e f i n e a g g .
- 0 . 21 ' I ' I I , I
(The s y m b o l of
0 4 8 12 1 6 2 0 f u r t h e r figures are s a m e . )
Age (months)
-
0.8
-8
-
h
-
A l k a l i : 1.5% 100/0
t
0.4 100/100
.-u)
0
-2- O/lOO
t 50/100
OJ 0.2
0. 100/50
x
0.0
-0.2
O 4 8 12 16 20
Age (months)
Rc/Rf : 0/100
u 0.6
5
c 50%
O
'j; 0.4 _j) 70%
C
(d
P
x o.:
W
0.0
O 4 8 12 16 20
A g e (months)
9
Yo.6.
.-OUI 0.4
Rc/Rf : 100/0
A l k a l i : 2.0% - Slag cont.
I
0%
50%
I=
m
0.2
W
0.0
O 2 4 6 8 1 0 1 2 14
Age (months)
-
0.4
-
A
O. 3
s.
Y
50/50
SOI50 50%
c 0.2 100/100 0%
.-Ou)
c 0.1
o
P
x
0.0
Alkali : 1.5%
-0.1
O 4 8 12 16 20
A g e (months)
Fig. 7-Effea of slag on expansion
1 1 01 I
8
100-
-
- Rc/Rf Slag c o n t .
-
v
9 0- 100/0 o %
. 100/0 50
n
-
i 80- 100/100 o ye
100/100 50 %o
a 0/100 0 %
7 0-
0/100 50 %
6 O-,
O 4 8 12 16 20
A g e (months)
C
0.8
0.6
-
- Rc/Rf
100/0
100/0
100/0
o
50
70
Slag c o n t .
%
Ya
Y'
.-O
u) 0.4
C
Q
Q.
;0 . 2
Age: 1 2 m o n t h s
0.0 I I
.-
u)
0.4-
i'
C
Q
Q. Age: 1 2 m o n t h s
X
w 0.2-
0.0
0.5 1 .o 1.5 2.0 2.5 3.0
A l k a l i content (%)
O.
n
8
<y
e
.-O
g
o.
o.
-
- Rc/Rf
0/100
0/100
0/100
S l a g cont.
o
50 %
70 %
m
P
X
u o.
O.
0.5 1.0 1.5 2.0 2.5 3.0
Alkali content (%)
2 0-
o --
0.5
1 ,,
:8':8:B 1.o
RcIRf
1.5
Slag cont.
A
O
.
100 I A g e : 12 m o n t h s
ci
Q Oll00 70
-
4 0-
Rc/Rf Slag cont.
2 0-
7 50/50 50 %
I
o ! I I l I
i;;/ A l k a l i : 1.0 o r
1.5%
ii
1 Alkali :
8
v
60 A l k a l i : 2.0%
!
$ 1 2.0%
2.5%
al
d A l k a l i : 2.5% 4i
20
0 1 ' I ' 1 ; I i I *
O 2 0 40 60 80 100
Slag cont. (%)
SP 154-13
An analysis of some research data of the RA0 and MIP tests is presented to illustrate practical
usefulness of the techniques. Results of long-tem observations of concrete blocks subjected to
outdoor conditions and the results of the new laboratory tests of the aggregates previously used
in the blocks are compared. The new tests seem to provide means for more successful evaluation
of coarse aggregates for purposes of diagnostics, design, and prediction of service life of
concrete.
245
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-I154 95 m O662949 0522440 264 m
246 Rusin
ìI"RODUCïI0N
The deterioration of concrete structures due to durability problems is extensive and widespread.
One of the main problems is caused by poor physical characteristics of aggregates in concrete
structures exposed to freezing and thawing. Unintentional use of poor quality aggregates may
occur due to the lack of appropriate laboratory tests for their evaluation. A growing shortage of
good quality aggregates accentuates the problem.
It is known that certain aggregates show an apparent propensity for quick absorption and
retention of water in their pores, while others have pores which cannot be filled at ali in cement
matrix. Generally, aggregates with a large number of small size pores, are considered to be of
questionable quality for use in concrete structures. However, laboratory tests, including the
determination of properties such as, microporosity, specific surface, water adsorption and
absorption, Expected Durability Factor (EDF)[I], Iowa Pore Index Test (iPIT)[2] and other non-
standard tests which reflect an opinion about the strong relation between pore characteristics of
the aggregate and its performance in concrete structures, do not always yield satisfactory results
to correlate with field performance [3,4,5,6]. Therefore, the prediction of performance of
aggregates in concrete under field conditions is sometimesvery difficult and questionable.
A study was undertaken to develop a new, more reliable and rapid laboratory test, called the
RAO-Method (RA0 in Polish, stands for RÓWiicowa Analiza Odksztaken, which means
differential analysis of strains), to meet engineers' expectations in the area of aggregate
classification [7].
A new Pore Size Distribution Index (PSDIndex) based on mercury intrusion porosimetry (ME')
analysis is proposed as an additional tool for aggregate classification too.
Resuits of the RA0 and MIP tests and their correlation with concrete field performance are
discussed in the paper.
The method is based on theoretical analysis which relates the MO-Index directly to the
mechanism of frost action in aggregate particles as well as indirectly to the mechanism of water
absorption prior to freezing. I have proposed [SI that the destructive mechanism of frost action
is directly connected with the crystallization of water in two qualitatively dinerent stages in
aggregate pores. In the first stage there occurs crystallization of water supercooled in larger
pores. This is accompanied by a flow of unfrozen water to areas with a lower pressure and by a
d e a s e of a relatively large amount of latent heat of crystallization. In the second stage, which is
much slower, a part of water is frozen in smaller pores. This phenomenon is accompanied by a
considerable increase in pressure (expansion) with a relatively small amount of measurable heat
of crystallization.
Phenomena such as osmosis, adsorption, diffusion have a considerable significance for the
conditions in aggregate pores prior to freezing. Similarly to the capillary phenomena they exert
an influence on a general balance of water in accessible voids. Hydrated cations and adsorbed
water molecules exert an influence on the viscosity of water and its capacity to be displaced from
areas with a higher pressure to areas with a lower pressure. They have a m-ng effect on the
process of crystallization but they do not change the character of the destruciive mechanism.
Generally, the larger the part of water in aggregate pores which freezes in the second slower
part of crystallization, the poorer is the physical quality of aggregate in terms of concrete
durability. Sound rocks display freezing which occurs as a strong single puise at the first stage
of crystallization.The hypothesis is an extension of Powers' hydraulic-pressure theory [9].
It is assumed that most of the aggregates with fully saturated pores, including those of low
porosity, will damage concrete when exposed to freezing temperatures. An exception can be
aggregates of total porosity below 0.3 %. According to Verbeck and Landgren, they represent
particles with the ability of elastic accomodation [3].
Thus, the usefulness of aggregate for concrete depends on the capacity of aggregate grains to
absorb and hold water while confined in the cement matrix. The rationale behind the method is
that the character of the process of water ciystallization during freezing in fully saturated pores
indirectly reflects the characteristics of the pores. The size and the distribution of the pores
essentially controls the rate of water absorption prior to freezing. Diíferent aggregates have
different pore characteristics and, therefore, different rates of crystallization of water. I believe
that a measure of the rate of crystallization of water in an aggregate sample could be used to
predict the performance of aggregates in concrete, exposed to freezing and thawing conditions.
The RA0 method depends on observations of the relative changes in volume and temperature of
coarse aggregate particles during cooling. Measurements are made in two identical dilatometers.
The capacity of the dilatometer is about 250 cm3. One of the dilatometers contains aggregate
particles saturated in water, while the other dilatometer contains the identical aggregate
saturated in carbon tetrachloride (CC14). The advantage of using CCl4 is that it does not freeze
until cooled to -23OC and does not mix with water. Both dilatometers are filled with CC14.
The dilatometers are cooled from approximately +5OC to about -15OC. The volume changes are
measured by reading a scale on a glass tube attached to the dilatometers. The temperature
changes are recorded automatically by data logger from thermistor probes placed inside the
dilatometers. The difference in volume changes and that of the temperature changes are
represented by the following notations:
RA0 parameters of aggregate (dolomite, 2% abs.) of a good field performance record are shown
in Fig 1 and 2. In the first rapid phase of ciystallization, nearly 100% of the water contained in
aggregate pores is frozen. This phenomenon is observed in the form of a jump in the values of
both parameters.
For contrast Fig.4 and 5 represent parameters of an aggregate of very poor field performance in
concrete (dense limestone, 0.9% abs.). This test was camed out under identical conditions, with
the same rate of cooling (Fig.3). In this aggregate in the first phase of crystallization a small
part of the total water volume is frozen (about 20%). The most of the water is frozen in the
second, slow phase, which can be clearly seen on the diagram in the form of an increasing
parameter delta V.
Fig. 6 and 7 represent parameters of an aggregate (0.7% abs.) of a very good field performance
in which in the first phase of crystallization equally a small amount of water is frozen as in the
case of the aggregate in Fig. 4, but the phenomenon of water freezing and related expansion of
grains in the second phase are not observed. The character of the parameter deita V is identical
as in the case of the aggregate in Fig. 1.
In Fig. 9 analogous parameters of four aggregates of a decidedly poor field performance are
shown. As can be inferred the character of the curve deita V is more significant for the needs of
aggregate diagnostics and classification than the estimation of absolute values of the parameter
deita V.
The experiments [6,7] confrm earlier observations that the relationship between total aggregate
porosity or absorption and its applicability to frost-resistant concretes is not unequivocal.
Aggregates with absorption of 0.5% can be dangerous, causing Dcracking whereas aggregates
with absorption of 2% and above can be completely safe.
Reference (71 presents an example of the application of the records of RA0 parameters to
developing a qualification table on the basis of the MO-Index. The MO-Index is a quantitative
index which characterizes aggregate. It is calculated as a ratio of two values. In the numerator is
wTitten a number equal to the difference of the value of the parameter delta V between the
beginning and the 5th minute of crystallization. In the denominator is written a number equal in
value to the theoretical increase in the volume of water contained in aggregate pores assuming
100% crystallization. Aggregate absorption in vacuum was used as the other applicable index;
Fig. IO illustrates their relationship. The information about aggregate field performance was
obtained from analysis of 30-year-long observations of concrete blocks made from the
aggregates. Haif-immersed in water the blocks were exposed to the effect of atmospheric
conditions on the test ground of Ontario Hydro in Toronto [6,7].
For comparison purposes Fig. ll,i2,13 and 14 show results performed according to ASTM C
127, ASTM C 666 @roc. A), ASTM C 88 and EDF, on the same aggregates as the ones
presented in Fig. 10 [ 101.
Figure I I shows absorption values of the aggregates. The values roughly indicate the pore
volume. Earliest studies on the properties of nondurable aggregates showed that unsound
aggregates are characterized by a low specific gravity (a high porosity) and a high degree of
saturation. Thus the absorption test has been recognized as an indicator of aggregate frost
sensitivity. The value of absorption about 3% is frequently recommended as an acceptance
criterion. However, as Fig. 11 shows, aggregates of much lower absorption can be harmful in
concrete structures exposed to frost. A little more information can be obtained by comparing the
results of absorption and vacuum absorption tests. As presented in Fig.11 most cases of good
quality aggregates have lower values of absorption in relation to values of vacuum absorption,
than the poor aggregates. The phenomena gave support to the degree of saturation hypothesis,
but there is no laboratoiy test based on the idea which could be easily adopted to predict
aggregate performance in concrete.
In Fig.12 are shown the ASTM C 666 (pr0c.A) results versus vacuum absorption. The
commonly used ASTM C 666 method is based on the assumption that rapid freezing and
thawing tests on concrete beams can be used to differentiate those aggregates which yield good
field performance from those having poor performance. Almost since the beginning the test was
criticized due to often unrealistic conditions [ 11). The ASTM C 666 method may yield results of
little importance to actual concrete performance, especially among so called "marginal"
aggregates. Anyhow the ASTM C 666 tests as well as other tests on concrete specimens are
time-consuming. The results in Fig. 12 reflect above remarks.
Sulfate soundness tests (Fig. 13) have been widely used for many years. I share the opinion of
Verbeck and Landgren [3] that the sulfate test results have rough and unreliable empirical
correlation with aggregate performance and the mechanism of disruption is different from that
resulting from the freezing of water. However, as Fig.13 presents, it is generally a g r d that a
low sulfate soundness loss is usually, but not always, evidence of good durability, whereas a high
loss places the aggregate in a questionable category.
There is no question that the results presented in Fig. 1412 and 13 are strongly related to the
pore characteristics of aggregates, especially absolute pore volume, size of the pores, and their
continuity. Many studies have confirmed the importance of these properties. On the basis of
such studies it can be qualitatively concluded that nondurable aggregates have a large volume of
small pores giving rise to a stronger tendency to absorb and retain water. In Fig. 14 the results of
mercury intrusion porosimetiy (MIP) analysis are presented in the form of the so called
Expected Durability Factor (EDF) [i]. A good correlation between pore size distribution of an
aggregate and its durability factor @F) based on the ASTM C 666 test, was pointed out as a way
to predict the frost durability of an aggregate. The following limits to distinguish between
potentially good or poor aggregates were proposed [ 121:
The geometrical properties of aggregate pores differ depending on rock origin. The largest pores
can be seen with the unaided eye or under an optical microscope. The smallest pores are usually
larger than gel pores in cement paste. The total pore volume depends on the kind of rock from
which aggregate originated. Generally, pores in grains of crushed aggregates and gravels are
open. This is directly due to the process of rock formation. Pores in sedimentary rocks were
formed as non consolidated free spaces between components, whereas in igneous rocks pores
were formed mainly as cracks between individual crystals. In both cases pores are accessible to
water.
Pore sizes have an essential importance from the point of view of the evaluation of aggregate
applicability to frost-resistant concretes. Particulary, presence of so-called dominant pores is
essential, i.e. pores whose geometrical sizes differ little and their summary volume constitute the
essential part of the total porosity of an aggregate. Fig. i 5 represents a graphical record of pore-
size distribution with a distinct group of dominant pores. The more geometrically approximate
are pores to one another, the more similar is their capacity of water absorption and uptake. This
conception is illustrated by simplified diagram in Fig. 16 [13].
Ap=f(l/r) (1)
If, for simplification, we neglect the problem of adsorption and diffusion of water vapor, Fig. 16
illustrates the successive phases of water saturation of aggregate grain. With a considerable
differentiation of the values of pore radii (R>>r), pores of largest sizes can remain empty even
under a prolonged contact of concrete with water. This mechanism is certainly effective in air-
entrained cement pastes and it exerts a significant positive influence on concrete frost
resistance. Therefore, a question arises why it should not be effectivein coarse aggregate.
On one hand, differentiation of pore sizes within one grain of aggregate exerts influence on its
capacity to absorb and keep water in pores, and on the other, the geometrical similarity of
aggregate pores and the surrounding cement paste exert influence on the capacity of water
displacement between aggregate and cement matrix. Pores in cured cement pastes of W/C in the
range of 0.4 to 0.5 have radii between 0.05 and 0.005 pm. The more pores in an aggregate of
sizes much larger than those of pores in cement paste (air pores in cement paste of sizes 1000 to
10000 times larger than capillary pores remain empty), the greater are chances that the
aggregate grain will have a reserve of air voids suEicient for the preservation of multiple cycles
of concrete water saturation and freezing.
The idea of the PSD-Index is based on the assumption that poor quality aggregates, have pores
of size similar to or smaller than the sizes of pores in the cement paste. Such pores visibly
dominate in total porosity of aggregate (Fig.15). It means that the graphical record of any pore
size distribution contains the relatively narrow range of pore sizes where one can observe
relatively high cumulative pore volume. The range of pore sizes is far below 1 pm. This kind of
pore structure causes an aggregate to absorb water easily, in some cases more readily than
surrçunding cement paste. The presence of pores with these characteristics rather than the total
porosity controls the field performance of aggregate in concrete structures.
in Fig. 15 the idea of PSDIndex is presented. The normalized value of the range of pore sizes is
expressed by:
where:
Z is the size of an "active window" of pore radii fok searching the PSD record to
find the range where the maximum change of porosity occurs,
rn - the largest pore radius of the "window",
rn+l - the smallest pore radius of the "window".
Scanning the pore size distribution data with the specific "window Z" makes it easy to find the
value of rn and rn+l for which the expression 3 has the maximum value:
-
where: AP the change of porosity inside the "window Z"
-
P the total porosity of an aggregate.
For rn and rn+l related to the maximum value of PSD-Index (3) the value of (average pore
radii of pores inside the "window Z" ) is calculated from the expression 4:
Calculated values of for the aggregates presented earlier in Fig. 10,11,12,l3 and 14 are
plotted in Fig. 17. The size of the "active window Z", for establishing the PSD-Index, was chosen
arbitrarily as Z = 0.25.
CONCLUDING REMARKS
The RA0 method as well as PSD-Index seem to be relatively convenient tools for testing
aggregates. They provide rapid and accurate information about the physical characteristics of
coarse aggregate, which are correlated with field performance of the aggregate in concrete
structures exposed to freezing and thawing. As was mentioned earlier, the physical quality of an
aggregate and its usefulness for frost-resistant concrete depends not only on the behavior of
water during cooling in pores, but also on the ability of an aggregate particle to achieve a
critical saturation state while the concrete structure is being exposed to wet conditions.
From both points of view, the size, kind, and number of pores in aggregate have a decisive
importance. The RA0 method enables investigation of these properties of pores. As was
represented in Fig. 8 and 9, it is possible to make a rapid classification of aggregate exclusively
on the basis of the qualitative analysis of the character of water freezing. The shape of the curve
delta V informs about the potential properties of aggregate. It also results from Fig. 10 that it is
possible to represent aggregate properties in the form of single numerical parameter. It should
be noted that aggregates of very low absorptions (below 0.5 %), in spite of an unfavorable
structure of pores can constitute a relatively safe component of concrete.
A direct analysis of pore size distribution enables the character of aggregate pores to be
represented in the form of two numerical values:
The PSD-Index is related to the pore characteristics of the aggregate, which controls the
behavior of an aggregate particle during freezing and thawing. The idea of the PSD-Index is
based on the same theoretical philosophy as the RA0 method. However, the PSD-Index is of
questionable accuracy for predicting field performance of aggregates with lower absorption due
to the lower precision of mercuiy intrusion porosimetiy with low porosity aggregates.
ACKNOWLEDGEMENTS
The author would like to express his appreciation for the funding and the opportunity
provided by the Ontario Hydro Research Division, Toronto, to carry out the research activities
presented in this paper, as well as the Committee of Scientific Research, for the financial
support of the paper by the Research Project No 77227 92 03.
REFERENCES
1. Kaneuji, M., Winslow, D.N., and Dolch, W.L., "The Relationship Between an Aggregates
Pore Size Distribution and Its Freeze Thaw Durability in Concrete," Cem. Con. Res., Vol.10,
1980, pp. 433-441.
2. Myers, T.D., and Dubberke, W., "Iowa Pore Index Test," Interim Report, Highway Division,
Iowa Department of Transportation, 1980.
3. Verbeck, G., and Landgren, R., "Influence of Physical Characteristics of Aggregates on Frost
Resistance of Concrete," Proceedings, Am.Soc.Testing Mats, Vo1.60, 1960, pp.1063-1079.
5. Williams, F.M., Trefny, A., Paxton, J.T., and Davis, H.D., "Development of Laboratory
Methods for Determining D-Cracking Susceptibility of Ohio Gravel and Limestones Coarse
Aggregate in Concrete Pavements," Ohio Depariment of Transportation, Bureau of Testing,
Final Report OHIO-DOT-07-74, 1974.
7. Rusin, Z., and Mukhejee, P.K., "Test for Prediction of Aggregate Performance in Concrete
Structures Exposed to Freezing and Thawing," International Conference CONCETE 2000,
Dundee, U.K.,1993, pp. 1395-1407.
8. Rusin, Z., "A Mechanism of Expansion of Concrete Aggregate Due to Frost Action," Cement
and Concrete Research, V01.21, 1991, pp.614-624.
9. Powers, T.C., "Freezing Effects in Concrete,'' Durability of Concrete, AC1 Special Publication
SP-47-1, 1975, pp.1-Il.
10. Rusin, Z., "Prediction of Aggregate Field Performance in Concrete Exposed to Freezing,"
Fourth International Symposium on Brittle Matrix Composites, Warsaw, 1994, pp. 447-454.
1I. AC1 Committee 201, "Guide to Durable Concrete," AC1 Manual of Concrete Practice. 1990,
p.201.2R-7.
12. Winslow, D.N., Lindgren, M.K., and Dolch, W.L.,"Relation Between Pavement D-Cracking
and Coarse Aggregate Pore Structure," Transportation Research Record 853, 1982, pp. 17-20.
13. Fagerlund G., "Frost Resistance of Concrete with Porous Aggregate," Cement and Concrete
Research Institute at the Institute of Technology, Stockholm, 1978.
1.2
I
0.8
E
0.6
t
>- dV5- change in volume within
5 minutes of
- 0.4 crystallization of water
0.2
O 1 I
tI
-0.2 I
I
[ aggregate sample i
O
o
2.5
2
--
--
If dT5- change in temperature
within 5 minutes of
5
I aggregate sample i
O
o O
E!
3
ci
Cu -5
àj
E
i!?
-10
-15
O 50 1O0 150 200 250
time, min
0.3
0.2
c3
E
o
>- 0.1
-0.1 I I
1
I aggregate sample 2 I
0.8
L- 0.4
e
a,
= 0.2
O
-0.2
O 50 1O0 150 200 250
time, min
0.2
m
E
o
0.15
>- 0.05
0.1
1 aggregate sample 3
-m
ci
a
U o
-0.05
-0. I
O 20 40 60 80 IO0
time, min
O 20 40 60 80 IO0
time, min
0.8
> 0.6
*
-
c,
(II
0.4
0.2
O
O 50 1O0 150 200 250
time, min
Fig. &Records of Delta V (aggregates of good field performance)
0.2
O
O 50 1O0 150 200 250
time, min
Fig. 9-Records of Delta V (aggregates of poor field performance)
2
good
-
L
1.5
LI
11”1
O 2 4 6 8 10
vacuum absorption, %
Fig. I L P r o p o s e d aggregate classification for concrete structures exposed to
freezing and thawing, based on RA0 index and vacuum absorption[A
i I field performance:
o. I 1 10
vacuum absorption,%
relative
acceptance
criterion
o 110
w Ò 40
o
Co
LL
20
-
+
.-
)r
0
e3 o
0 0.1 I 10
vacuum absorption,%
s
A 100
ao
o +a
=
I-
(B
10
n
U
a
v
cn o
m l U
a,
D
S o
C
8
m
0.1
u)
o. 1 I IO
vacuum absorption,O
h
-1000
5
f
.-
27
- A acceptance
2 100 !: m criterion
n
L
3 dl
U
a,
c,
o
&-- o
n
X
m
10 1 p ;=
o. 1 1 10
vacuum absorption,%
CJ) 35
3
E 30 - lOgrn+l
E
ai 25
E
=> 20
O
PSD-Index = ( APP)
2 15
O
Y
W O
a,
W
3 5L
ci
.G 0
‘n+i ‘n
I 10 IO0 IO00 1E4
pore radius, nm
1
mop-outs
E
1
v
Ini-
o. I very
tn Poor
.-3
U
E v
al
L 0.01 v
8 poor field performance
0.001
o. 1 1 10
vacuum absorption,%
-
Fig. 17-Pore radii R related to maximum values of PSD index versus vacuum
absorption of aggregates
263
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 Ob62949 0522457 568 =
264 Stroeven et al
INTRODUCTION
Portlaiid ceineiit is tlie iiiost iiiiportaiit iiiorgaiiic biiider iii tlie coiistruc-
tioii iiidustry. Various bleiidecl cemeiits iiicrease the versatility of ceiiienti-
tious binders for coiistructioii purposes. A need is felt however to search for
cheaper b i d e r s , wliicli (additioiially) lielp to save eiiergy aiid precious raw
materials, aiid to protect the eiiviroiiiiieiit. Moreover, iii iiiost areas iii de-
veloping couiitries, portlaiid ceiiieiit is not readily available or only at liigli
cost. Iii sucli cases, use has to be iiiade of locally available raw iiiaterials aiid
waste coinpoiieiits. Rice husk asli (RHA) is reported to be tlie most proinis-
ing biiider coiiigoiieiit. It is placed iii tlie sanie category of liiglily pozzolanic
materials as coiideiised silica fuiiie (CSF) by RILEM (1). CSF is however
aii expeiisive by-product, siiice it is oiily available to a liiiiited extent. Com-
pared to tlie world’s CSF produc,tiori (iii 1989) of 0.5 iiiillioii toiis (a), the
world’s productioii of rice husks iii 1990 was 100 iiiillioii tons (3). Cook (4)
lias estiiiiated that about 6 iiiillioii tons of RHA c,ould be produced for use
as (partial) replaceiiieiit of portlaiid ceiiieiit.
Approximately 3.6 iiiillioii tons of rice liusks are produced per year iii
Vietriain (3), wliicli could possibly be coiiverted into 0.G iiiillioii tons of RHA.
This aniouiits to 20% of the aiiiiual ceiiieiit productioii iii tlie couiitry. Tliere-
fore, this iiiiiieral ‘waste’ could be eiripliasized for produciiig, iii coiiibiiiatioii
with lime, lower qualities of coiic.rete, to be applied iii tlie rural sector of this
country (5-11). High coiicrete qualities are additioiially required for tlie re-
building of tlie larger cities iii tlie iiortlierii part of Vietiiaiii (sucli as Hanoi),
or for coastal structures, aiid liarbour works. To that eiid, replaceiiieiit of
CSF by RHA sliould be empliasized (12-15). This forms tlie major theine of
this paper.
SCOPE OF INVESTIGATIONS
The tests were executed iii tlie Steviii Laboratories of Delft University
of Tecliiiology. Vietiiaiiiese aiid Taiizaiiiaii types of rice liusks aid a sinall
sa~iipleof fine Red River saiid were transported to Tlie Netlierlaiids. Tlie
inorpliological characteristics of tlie fine sa.iid were aiialysed. Tlie fiiieiiess
modulus of tlie smooth-textured saiid graiiis aiiioiiiited to 1.72. Further,
tlie said coiitaiiied some silt aiid relative large aiiiouiits of mica. For tlie
experiments, liowever, use was made of Dutch river aggregate, of wliicli the
saiid reseiiibled as close as possible tlie Vietiiaiiiese one. For tlie maximum
grain size of tlie gravel, 16 inin was clioseii.
For a detailed description of the oveii and of its operation, see (10,17,22).
Various provisions liad to be taken to fulfil environmental and safety regula-
tions in the Netherlands. Tlie oven is filled froiii tlie top. Initial ignition is
performed by a torch fueled with natural gas. It will be extinguished when the
pyrolysis process of the husks produces enough gas to maintain the burning
process. Temperatures were measured by tlierinocouples at three locations
in tlie oven, ie. in the firing zone, outside the coinbustion chamber, and in
the ash-pan. Tlie temperature was regulated by tlie rate of air flow and ash
removal. The air circulation was accomplislied by a fan and controlled by a
system of stopcocks. Temperatures in the oveil were kept well below 75OoC.
After gasification tlie residue was mechanically removed from the bed into the
ash-pan. This was accomplislied by means of a very slowly rotating motor-
driven vane. Tlie asli was kept liere for 15 hours. The chemical composition
of the ainorplious silica asli produced in this oven aiid used for tlie investi-
gation is given in Table 3. Tlie X-ray diffraction pattern is shown iii Fig. 2.
The aslies froiii the oven were ground to yield two different qualities of
RHA. Tlie first was produced by grinding tlie ash for 18 hours in a ceramic
laboratory ball mill; this quality is denoted by RHA(18). Merely to study the
effect of griiiding 011 particle size, an asli sample was takeii after 14 hours; this
is indicated by RHA(14). Tlie second quality of RHA was produced similarly,
but tlie ash was pre-mixed with a iiaplitliale~ie-basedsuperplasticizer (VN-
BETONMIX 415). This quality is denoted by RHA(18)+. These aslies were
used for blending with portland ceiiient. As a reference, an uiigrouiid asli
was analysed (denoted by RHA(0)). Data 011 BET specific surfam area of
the different ash samples 'obtained by nitrogen adsorption are presented in
Table 4.
MIXTURE PROPORTIONS
TEST SERIES
Mortar
Gap-graded concretes
RESULTS
The inortar reference streiigtli data confirm tlie general trends reported
iii tlie literature (13, 23). Since tliis is not tlie major subject discussed iii this
DISCUSSION
arrived at for the ultra fine ash! Influence of grinding tiiiie on BET specific
surface area can be expected to be sinal1 for tlie coarser particles. Tliis was
confirmed by earlier investigations (17,24). Using tliis fine asli, the present
experiments demonstrate that even for cement contents between 240 aiid 300
kg/in3, a 7-day strength exceeding 50 MPa can be obtained for very differ-
ent workability conditioiis. The last two iiiixtures in Table 10 differ only in
blending percentage aiid show tlie same liigli compressive strength
. level at a
slulllp of 200 111111.
CONCLUSIONS
REFERENCES
1. Final Report of RILEM TCJ 73-SBC, ”Siliceous by-products for use iii
concrete,’’ Journal of Material Structures, Volunie 21, No. 121, 1988,
pp. 69-80.
4. Cady, P.D. aiid Groiiey, P.R., ”Hydraulic cement from rice husks,” Ce-
ineiit Tecliiiology, Volume. 7, No. 6, 1976, p. 215.
5. Cook, D..]., Raina, R.P. aiid Paul, B.K., ” Rice liusk asli-lime cemeiit
iiiixtures for ilse iii inasoiiry uiiits,” Building aiid Eiiviroiiiiieiit, Volume
12, 1977, pp. ‘281-288.
7. Dass, A. aiid Rai, M., ”Prospects aiid probleins iii tlie production of
ceiiieiititious inaterials froin rice liusk,” Proceedings, UNIDO/ESCAP/
RCTT Worksliop 011 Rice Hiisk Asli Ceineiit, Pesliawar, Pakistan, 1979.
Regioiial Centre for Tecliiiology Transfer, Bangalor, Iiidia, 1979, pp. 49-
56.
10. Salmiii, E.L., ”H.esearcii into the poteiitialities of rice husk ash ceiiieiit
for application iii rural Taiizaiiia,” Report 03.21.1.32.05, arid ”Pliysico-
clieiiiical iiivestiga,tiorisof Taiizaiiiaii rice husks aiid of rice liiisb asli for
use in rural Taiizaiiia,” Report 03.21.132.21, Faculty of Civil Eiigiiieer-
irig , Delft IJiiiversity of Tecliiiology, 1995.
11. Melita, P.K. aiid Pitt, N., ”Energy aiid iiidustrial inaterials froiii crop
residues,” Resources Recovery aiicl Coiiservatioii, Voluine 2, No. 1,
1976, pl). 23-28.
12. Cook, D.J., Paina, R.P. aiid Dainer, S.A., ”Rice liusk ash a pozzolanic
iiiaterial,” New Horizoiis iii Coiistruction Materials, Volume 1, Eiivo
Publ. Co. Iiic,. Lehigli Valley, 1976, pp. 431-442.
13. Melita,’P.K., ”Properties of bleiided ceiiieiit made froin rice husk ash,”
AC1 Jouriial, Voliiine 74, No. 9, 1977, pp. 440-442.
14. Baoteiig, A.A. aiid Sltrete, D.A., ”Iiiciiieratioii of rice liull for use hs a
16. Cook, D.J., ”Rice liusli ash,” Ceiiieiit Replacement Materials, ed Swainy,
R.N., Surrey Uiiiv. Press, 1986, pp. 171-196.
19. Taylor, W.H., ”Coiicrete techiiology aiid practice,” Elsevier, New York,
1965.
22. Bui, D.D., ”Gap-graded coiicrete with very fine sand using portlaiid
RHA blended ceiiieiit,” Report 03.21.1.32.04, Fac.ulty of Civil Eiigi-
iieeriiig, Delft IJiiiversity of Teclinology, 1994.
23. Melita, P.K., ”Studies 011 iiiecliaiiisiiis by wliicli coiideiised silica fuine
iiiiproves tlie properties of coiicrete,” Proceedings, Worksliop Coiideiised
Silica Fuiiie iii Coiicrete, CANMET, Ottawa, 1987, pp. 1-17.
24. Al-Klialaf, M.N. aiid Yousif, H.A., ”Use of rice liusk asli iii coiicrete,”
Jouriial of Ceiiient Coinposites and Lightweight (‘:oiicxete, Voluine 6,
NO. 4, 1084, 1 ) ~ .241-248.
25. Melita, P.K., ”The clieiiiistry aiid tecliiiology of ceiiieiit inade froiii rice
husl< asli,” Proceedings, IIN I DO/ ESC AP/RCTT Worksliop Rice Husk
Asli Cemeiit, Pesliawar, Pakistan, 1979. R.egioiia1 Centre for Tecliiiol-
ogy Transfer, Bangalor, India, 1979, pp. 113-122.
Coinposition
iii % by weight
68.5
0.23
0.15
0.33
0.01
2.83
1.59
1LOI 23.0
RHA-qriali ty
RHA(0)
RHA(14)
RH A( 18)
RHA(18)+
Sieve size Rest on each sieve Average rest per sieve Cumulative rest
111111 gram
. .
graiii % %
1 1st 2nd I
16 I 113 79 I 96.0
I
1.92 I 1.92
8 4,424 4,226 4,325.0 86.50 88.42
4 436 625 530.5 10.61 99.03
2 27 70 48.5 0.97 100.00
W/C+RHA Flow
inin
0.470 110
Pc 0.340 110
RHA( 18) 0.365 110
RH A( 18) 0.370 110
RHA(18) 0.400 110
RHA( l8)+ 0.354 112
RH A( 1S)+ 0.370 110
RHA( 18)+ 0.360 113
0.376 110
- - -
- __ - -
c: RHA/C: s/s+c:
_.
RHA SP W S G W/RHA+C
kg kg kg -
kg kg kg
-
306 169 557 1297 0.55 O 0.30
300 120 383 1635 0.40 O 0.19
438 175 521 1215 0.40 O o3 0
500 175 590 1095 0.35 O 0.35
10 20 22.2 30.8
5 40 41.3 49.8
30 20 43.9 56.1
20 20 47.9 61.2
10 20 29.8 47. I
O 50 44.4 52.7
10 3O 53.5 72.3
5 40 49.1 64.1
25 20 44.7 64.3
15 20 50.4 69.9
6 20 47.3 70.2
15 20 53.8 69.1
200 20 50.3 68.9
8 LO 56.9 73.7
200 20 51.6 70.1
200 20 52.5 70.5
Fig. 1-Oven used for incineration of rite husks, simulating 'field' conditions in Vietnam
SP 154-15
279
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 75 = 0662747 0522473 700 D
280 Tsukinaga et al
INTRODUCTION
MATERIALS
TEST METHODS
Series I
Series II
Where
B ;z= diffusion coefficient (crnZ/sec:)
Dm = depth of water p e n e t r a t i o n (crn)
t = time required (sec)
cy = coefficient corresponding t o time applying p r e s s u r e
(81.5 when 8 hours)
E = c o e f f i c i e n t corresponding t o p r e s s u r e of water
(0.905 when 0.49 MPa )
CONCLUSIONS
increasing tendency.
4. The d e g r e e of improvement of c o n c r e t e q u a l i t y varied with
height of model c o n c r e t e , perhaps caused by t h e degree of
consolidat ion.
5. The improvements in t h e r e s i s t a n c e t o t h e f r e e z i n g and
thawing a c t i o n , t h e carbonation and t h e i n t r u s i o n of chloride-
ion were experimentally confirmed.
6. The improvement of water t i g h t n e s s was confirmed from t h e
d i f f u s i v i t y d a t a in t h e water permeability test using t h e input
method.
REFERENCES
1. Nakajima,M., Sato,Y., and Segami,M., "Study on analysis of
water cement r a t i o of f l e s h concrete", Transaction of N.M.B.
c e n t r a l r e s e a r c h l a b o r a t o r y , Nisso master b i l d e r s limited
central r e s e a r c h l a b o r a t o r y , Japan, No.2, pp.G3-88, 1979.
2. Tsukinaga,Y., Shoya,M., and Kasai,Y., "In-situ test methods
assessing t h e q u a l i t y of t h e s u r f a c e l a y e r of concrete", Proc.
of I n t e r n a t i o n a l Symposium on NDT&SSM, FENDT'92, vol. 1,
pp. 477-484, 1992.
3. Nasser,K. W. and Al-Manaseer,A.A., "New Nondestructive Test",
AC1 Concrete 1nternational:Design & Construction, Vol. 9, No. 1,
pp.41-44, 1987.
4. Murafa, j., "Studies on t h e permeability of concrete", TanS.of
JSCE, N0.77, pp.89-103, 1961
5. Hall,C., "Water S o r p t i v i t y of Mortars and Concretes", A Review,
Magazine of Concrete Research, 41, No. 147, pp.51-G1, 1989.
Oi-d i na i-v 1. 7 2. 1 I. 4
High cai.lv s t r e n g t h I. 7 3. 2 I. 3
28 day
Sri-ic compressive
strength
I
II
0.65
0.62
80
120
3.0
4.0
42.5
17.4
164
162
~ 252
261 '
1 807
885
~
'
1222
1015
~
!
O. 50
2.78
22. 7
25. 7
.&onCrete *
‘o . ‘o’
o -
.,@. * o
Concrefe Side
o .
I Excess Water I@
ri{/y
Fig. 1-Schematic sketch explaining function of permeable sheet
300
M
~~~ .$
N
4
O cu .4
4
Lower part 0 .2
Y,
CI
k-
1 1200
1
i GOO 1
1
900 * &
1100 1
Series I Series n
Fig. Z-ûutline of model concrete used in test
1 - 5 0 4
Circular steel probe
with hollow cylinder
x
W
't
o
C+ -lilon
01 I l I I I l I l I
O 10 20 30 40 50 60 70 80 90
Distance from surface (mm>
1200
2 1000
W
U
--
O
800-
1 I Lower p a r t I
A
+Sheet
O- -Non
3
L
' A
.-
~u
'+
6 a o ~ ~ - ~ - . A ~ -_ A
A
_ .-_
A
- - --- - -AA
O A
L 40Q-A A AAA A A A
rli
0
A
2 200-
O I I I I I I I l 1
I Upper part. I
I Lower part. I
+Sheet,
Q- -Ilon
+-
(u
E
GS
._
-5
A A
50 -
25 I I I I I I I l I
n in 20 30 40 50 60 7n 80 8n
Distance from surface (mni)
1
0. 10 4
3
L +Sheet.-upper part
AT- Sheet-l ower p a r t
-
c
t’ 0- Non-upper p a r t
t-
3
i+ -Non-l owr p a r t
-- -Non O. O947
r-
H
:-iI
;-flI r---1
I I I
-o om-
r - i I I
I I I i-,
I l I I I
I I I I I
3 r-i I I I I
I I _ I
L I
I
I
1
I
L-
2 0.~05- r-i - -
1-r-1
I I
r-i
I
I
-
l
L
- -
L-1
l
b-1
I
L Sheet-upper p a r t
\ t-Sheet-l o w r p a r t
\
1
e
\
w \
c
a,
't
o
-
-3
a
4-
A
rn
z
w
w
c 3-
a,
L
c, A
m
a,
.-
m o++ \ \
e
'
'
t
t
Qi
0
2- A--+ -- - - - _ _ -A
-
-!
1. I l l I I I l I I I I
k 40
L I
E
3
c
30
3
o
22
lz 20
h
tl
._
0
-o
$ 4
a)
-3
(0
Q
3
Lower part. Middle part. Upper part,
*Sheet-verti cal p a r t
(o +Sheet-sl ant i ng p a r t
-
3
3
O=-Non-vert i cal part
73
O h Non-sl ant i ng p a r t
E
105-
1ou
9 5 l , , , , , , , , , , , , , , , , , , ,
0 60 120 i 80 240 300
Freezing and thswi ng cycl es
-Sheet-vert i cal p a r t
&-Sheet-slanting Part
A
x
W
ai
o)
t
m
-c
o
+
-c
m
.-
Qi
3
- 2 1 , , , , , , , , , , , , , , , , , , , ,
o 60 120 i 80 240 300
Freezing and thawing cycles
%
O
30
c
2
t)
20
m
5 10
2
,T
SI ant i ng p a r t
SP 154-16
Self-Compacting Property of
Highly Flowable Concrete
by S. Nagataki and H. Fujiwara
Svnopsis: In recent years, there has been an increasing demand for high-
performance concrete with better workability, higher strength and greater durability
to meet current structural design needs. In Japan, studies of highly flowable
concrete with self-compacting properties have been undertaken with the goal of
improving reliability of concrete compaction in forms having complicated shapes or
densely arranged reinforcement. To produce highly flowable concrete, it is
necessary to create high-fluidity concrete by adding a superplasticizer and to
eliminate segregation by adding a viscosity-controlling admixture or a large volume
of powdered material. It is also necessary to provide the concrete with the ability to
pass between the steel reinforcing bars in order to make it self-compacting; this is
achieved by controlling the rheological properties of mortar and volume of coarse
aggregate. in this paper, the properties of self-compacting concrete are described.
301
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-I154 95 m Obb2949 0522494 435 m
302 Nagataki and Fujiwara
INTRODUCTION
Recent years have seen concrete structures increase in height and overall size,
resulting in an increasing demand for a wider diversity of types of high-strength,
high-performance concrete. In an effort to meet these needs, high-strength, high-
durability and high-fluidity concretes, among others, have been studied.
In Japan, high-fluidity concrete has been receiving particular scrutiny in recent
years, and application of highly flowable concrete is approaching the stage of
practicality. With highly flowable concrete, the use of a vibrator for compacting at
site is unnecessary because the concrete is self-compacting within the form. That
there is an increasing demand for improving the reliability of concreting work in
Japan, can be seen from the following information.
The topography of Japan comprises many precipitous mountains and rugged
valleys; thus it is often the case that concrete placement must be camed out under
extremely difficult circumstances. Further, the increasingly complex shape of
concrete structures is making it more difficult to use a vibrator, while the more
densely arranged reinforcing bars resulting from the increasing height of concrete
structures make consolidation more difficult to carry out.
i n addition, the following reasons may be cited for the increasing demand for
self-compacting, highly flowable concrete: a) there are not enough workers to carry
out compacting work at construction sites; b) vibrator compaction of concrete is
extremely noisy and deleterious to the health of workers, as well as an annoyance to
people in the surrounding neighborhood; and c) it is costly and time-consuming.
Highly flowable concrete not only alleviates these problems, but also improves
the efficiency of construction. in other words, the use of highly flowable concrete
requires only the setting up of forms and placement of concrete by pumping; the
previously required hard, labor-intensive work of compacting and the related setting
up of scaffolding are eliminated. For these reasons, it is expected that the use of
highly flowable concrete will become widespread.
Outline of Exueriment
Mixture Proportions
Fluiditv of Concrete
The slump flow, that is the diameter of the slump test base after release of the
slump cone, was measured and fluidity of concrete evaluated. Since slump flow is
measured when the concrete is in a static condition, yield value can be considered as
the controlling factor (4). Further, the yield value of concrete can be considered as
being determined mainly by the volume of coarse aggregate and the yield value of
mortar (5). Fig. 2 shows the relationship between the slump flow and yield value of
mortar for various volumes of coarse aggregate.
In Fig. 3, the Y-axis indicates volume of coarse aggregate and the X-axis
shows the rheological properties of mortar; based on the results obtained from
experiments, approximate lines were drawn to show slump flows of 500,600 and
700 millimeters. From this figure, it can be seen that the yield value of mortar
increases, while the slump flow value of concrete decreases as the volume of coarse
aggregate increases.
Resistance to SeFerration
Fig. 4 shows the relationship between yield value of mortar and the SI of
concrete. in each case, the SI falls sharply (solid line in center of figure) as the
mortar yield value increases; when the 'sfexceeds a certain level, it progresses to a
value near zero (broken line in center of figure). This is because the volume of
mortar adhering to the coarse aggregate, which accompanies the increase in q,
causes the SI value to decrease.
As shown in Fig. 5 , highly flowable concrete, with a SI of 4% or 8%, was
poured from the top of the form and left for ten minutes, after which samples'were
taken from the upper, middle and lower sections. Ten liters of concrete was
measured and taken from these samples; coarse aggregates were washed out using
water and a five-millimeter mesh screen. The results of measurement of volume of
coarse aggregate are shown in Fig. 6. From these results, it was found that, when
the SI was about 5% or lower, no sinking of coarse aggregate or other segregation
occurred. Thus, for each volume of coarse aggregate, the yield value of mortar
giving a SI of approximately 5% was obtained from Fig. 4 and superimposed on
Fig. 3 , resulting in Fig. 7 . The area to the right of the line (boundary of
segregation) 'chows the range in which sufficient resistance to segregation is
demonstrated, while the area to the left is the range in which segregation occurs.
The ability of concrete to pass between the steel reinforcing bars was evaluated
as follows. First, concrete was poured into the left side of the vessel shown in Fig.
8 until the concrete reached the upper lip of the vessel. Next, the gate was opened
and the concrete permitted to flow into the right side of the vessel. When the flow
stopped, the difference between the height of concrete in the left and right sides of
the vessel was measured and the pressure difference was calculated from the specific
gravity of the concrete. From this type of experiment, it can be roughly determined
if the concrete will have the ability to pass between the steel reinforcing bars or if it
will be stopped immediately after the gate is opened; however, in this experiment,
concrete having a pressure difference of 5,000 Pa or less was evaluated as having
sufficient ability to pass between the steel reinforcing bars.
In order to achieve sufficient ability to pass between the steel reinforcing bars,
the mortar must have sufficient ability to carry the coarse aggregate with it when it
passes between the bars. Other factors thought to have an effect on the ability of
concrete to pass between the steel reinforcing bars are space between the bars,
volume of coarse aggregate contained in the concrete, maximum size of the coarse
aggregate and viscosity of the mortar (7). In this experiment, the space between the
steel reinforcing bars was 37 millimeters and maximum size of coarse aggregate was
20 millimeters; within these conditions, concrete was evaluated as having the ability
to pass between the steel reinforcing bars for each volume of coarse aggregate
contained in the concrete.
Fig. 9 shows the relationship between viscosity of mortar and pressure
difference. The results of this experiment may be divided roughly into two
categories: when the pressure difference is 2,000 Pa or less, most concrete is able to
pass between the steel reinforcing bars; however, when the pressure difference is
7,000 Pa or more, the concrete is unable to pass between the bars. In other words,
when the volume of coarse aggregate is 24.5%,except under conditions where the
concrete was unable to pass between the bars, even when a viscosity-controlling
admixture was not used, most concrete was able to pass between the steel
reinforcing bars; when the volume of coarse aggregate was 34.5%,concrete was
unable to pass between the bars under all conditions. Further, when volume of
coarse aggregate was set between these levels at 29.5%, only concrete having a
viscosity within the range from 7,000 to 9,000 mPa*s was able to pass between the
bars.
in Fig. 10, conditions under which concrete was able to pass between the steel
reinforcing bars (see Fig. 7) are indicated by a white circle and those where concrete
failed to pass between the bars by a black circle. It is thought that concrete failed to
pass between the reinforcing bars when the viscosity of mortar was low, because of
segregation or because the mortar lacked the ability to cany the coarse aggregate
between the bars, thus permitting it to accumulate and clog the spaces between the
bars; failure to pass between the bars when viscosity was high is thought to be due
to insufficient fluidity.
RESULTS OF EXPERIMENT
CONCLUSION
With highly flowable concrete, the need for consolidation is eliminated as the
concrete is self-compacting within the form. In order to achieve this property, the
concrete must satisfy all of the following three conditions: high fluidity, resistance
to segregation and the ability to pass between the steel reinforcing bars. It was
found through these experiments that concrete having the volume of coarse
aggregate and rheological properties of mortar falling within the range shown in Fig.
1 I meets these requirements.
REFER ENCE S
45.5
I
420
390
Unit: kdm3
892
I 178 I 829 I 910
778 i+
I
I
(%o)
Viscosiiy-controllingadmixture p (%)
ímm) +~,=24.5%
-A- ~,=29.5%
-1 1x,=34.5%
500 1
I I I I I
3,000 6,000 9,000 12,000 (mPa-s)
Coefficient of viscosity
0 SI=4.0%
SI=4.8%
A SI=8.5%
20 , Lower
section
Middle
section
Upper
section
Yield value
I l I I I
3,000 6.000 9,000 12,000 (h4Pa.s)
Coefîicicnt of viscosity
/Gate n
O000 k
5000
5000
O
3000
RI5 6000 9000
~,=24.! 6
xv= 2 9.5%
x,=34.5%
12000 (mPa
Cocfficicnt of viscosity (ilpi)
I
0Pass Don'tpass
".$ e e e "\.\ O
ßoucdary line of segrcgation 5& (qm)
M
i'.,
O
I I
10 20 30 40 50 60 (Pa)
Yield value
I I I I I
3,000 6,000 9,000 12,000 (mPa-s)
Cocfficicnt of viscosity
35.0 O
u
+
w
m
ki
2
w
0
u
30.0
c-
O
-$
s
25 .O
10 20 30 40 50 60 (F
Yield valuc
I I I I I
3,000 6,000 9,000 12,000 (mpa-s)
Cocnicicnt of viscosily
Fig. 11-Range in which requirements for highly flowable concrete are satisfied
SP 154-17
Under tension and flexure, increases both in strength and strain capacity are
reported as a result of fiber reinforcement. Carbon fiber reinforced cement
composites are also much more impact resistant than the parent matrix. Under
compression, however, no increases either in the compressive strength or in the
elastic modulus are noticed.
315
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-I154 95 E 0662949 0522508 85T W
316 Banthia and Genois
INTRODUCTION
PRODUCTION OF CFRC
Given their small size (Table 1) and hence their high specific surface
areas, mixing carbon fibers in cements using conventional means is usually
difficult. Carbon fibers tend to ball and disperse non-uniformly. Beyond 1% by
volume of fibers, a suitable dispersing agent (carboxyl methyl cellulose,
condensed silica fume or granulated blast furnace slag) and appropriate quantities
of superplasticizer (about 3 % by weight of cement) h e needed. The availability
of a mixer with a higher shear capability, such as the omni-mixer, may facilitate
uniform mixing and dispersion even at high fiber volume fractions. The
workability of CFRC is usually characterized by the “flow number” measured
using a “flow table” as described in JIS R 5201 (5,13). For carbon fiber reinforced
composites, the “flow number” is found to be proportional to the fiber diameter,
inversely proportional to the fiber volume fraction but independent of fiber length
(5313).
For a good fiber-matrix bond, a dense packing of hydration products
around the fibers is essential. This is often achieved by using cements with a
maximum particle size less than 45 pm and mixtures with low waterkement ratio
and silica fume. Fig. 1 shows the dense packing of hydration products around
carbon fibers.
Tensile Behavior
Some stress-strain curves for CFRC under uniaxial tension are given in
Fig. 2,3. The somewhat stiffer response of fiber reinforced composites over the
plain, unreinforced matrix may be noted. In general, for fiber volume fraction
higher than I%, there are strength and modulus increases due to fiber
reinforcement and the composites can be expected to support much higher strains
at peak load than the parent matrix. Notice also that curves in Fig. 3 obtained for
mortars in a closed-loop test environment become distinctly bilinear at a fiber
content of about 2% by volume and significant increases in the fracture energy
(total area under the curve) can be expected. The behavior of CFRC in tension is
influenced not only by the length of the fibers used but also by other factors such
as matrix strength, fiber modulus, fiber-matrix bond, extent of fiber dispersion,
etc.
Flexural Behavior
Compressive Behavior
Similar effects were noted by Ohama and Amano (4) but only &er
exceeding a fiber volume fraction of 3%. While minor improvements in the
compressive strength are possible by using the shorter 3 mm fiber as compared to
the longer 10 mm fiber (4), on the whole one may conclude that the compressive
strength is almost unaltered by carbon fiber inclusion. This is equally true for
other types of fibers (2).
Elastic moduli for carbon fibers reinforced mortars in compression are
given in Fig. 8 (23). Notice a minor decrease in the elastic modulus due to fiber
addition. Overall, no substantial change in the elastic modulus due to pitch-based
carbon fibers, as in the case of other fibers, can be expected.
Bend Over Point, BOP, is defined as the point on the stress-strain curve
at which first cracking in the matrix occurs somewhere in the composite and the
non-linear constitutive behavior commences. An interesting feature of the stress-
strain curves in tension (Fig. 2) as well as the load-displacement curves in flexure
(Fig. 4) at higher fiber volume fractions is the pseudo-strain hardening that occurs
between the BOP and the peak load. Pseudo strain-hardening implies that once
the first matrix crack occurs at the BOP, further cracking or extension of existing
cracks will occur only if additional energy is supplied from the loading source.
Naturally, this behavior is quite desirable from the field performance point of
view. In the pseudo strain-hardening regime, the capacity of the composite to
carry the imposed far-field stresses and strains is controlled entirely by the
conditions at the tip of the critical crack which has nucleated at the BOP and wiil
propagate first in a stable manner but eventually attain unstable dimensions. With
the section having developed cracks, the global stresses and strains measured in a
test in the pseudo strain-hardening regime are at best only a crude estimate of the
composite performance and are not related to the fundamental behavior of the
material. For a fundamental characterization, one needs to study the nucleation
and propagation of cracks in these composites through Controlled Crack Growth
Studies. Such studies are also usefùl in predicting the sensitivity of CFRC to
strain-rate and to estimate the response under fatigue loads.
A test set-up for Controlled Crack Growth Studies on CFRC using
Contoured Double Cantilever Beam Specimens is shown in Fig. 11 (23). During
the test, the load is applied by means of wedges and the horizontal component of
the load (the opening load) as well as the crack mouth opening displacement
(CMOD) are monitored. The crack advances during the test, and the objective is
to characterize the stress field at the crack tip as it advances. Typical opening load
vs. CMOD curves obtained for mortars reinforced with different volume fractions
of carbon fibers are shown in Fig. 12a. Notice that for plain mortar the maximum
opening load and corresponding CMOD are low while carbon fiber reinforced
mortars carry much higher opening loads and exhibit much greater crack opening
displacements at the peak load.
These opening load-CMOD curves can be further analyzed to obtain
fundamental crack growth resistance curves, R-Curves (22). Typical R-Curves
plotted in terms of the stress intensity factor, K,, at the crack tip and the effective
crack length are given in Fig. 12b (23). The effective crack length, u@ is
determined using a compliance calibration while the stress intensity factor, KR,at
the tip of the corresponding effective crack is calculated using the available linear
elastic fracture mechanics equation for a contoured double cantilever beam (23).
This leads to
where P is the opening load, B is the width of the cantilever beams, b is the
cracked width and Hois the depth of the beam at the tip of the effective crack
length aeF
These curve indicate that mortars reinforced with carbon fibers can
withstand a substantially higher stress intensity at the crack tip (indicating a
significantly higher fracture toughness) and that the cracks can grow in a stable
manner to much greater lengths compared with unreinforced specimens. It is also
worth noting that the maximum values of the stress intensity factor do not occur
at the peak load, but rather soon after the peak load (Fig. 12b).
DURABILITY OF CFRC
Superior chemical inertness of carbon fibers over almost ali other high
performance fibers is one of the important benefits. Pitch-based carbon fibers are
more than 90% elemental carbon and, are not corroded in the alkaline
cementitious environment like the ordinary glass fibers. Although the fibers
themselves are attacked by strong oxidizing agents such as nitric and sulfùric
acids (S), their composites have been found to have no appreciable retrogression
either in the strength or in toughness when subjected to cyclic exposures of weak
acids (pH = 4.0) for up to 90 days (14). Carbon fibers tolerate heating of up to
250OOC in inert atmospheres but oxidize slowly at 3OO0C or more in air.
Limited data with regard to the fieezing-thawing resistance of non air-
entrained CFRC with tests performed in accordance with ASTM C666 &er 300
cycles have indicated that the diminution of dynamic modulus of elasticity is
insignificant (9). In Fig. 13, it may be observed that, under restrained shrinkage
conditions in a drying environment, carbon fibers lead to a substantial reduction in
the maximum crack widths, thus improving the durability (18). Notice also in Fig.
13 that a longer 10 m m fiber is more effective in restricting crack widths that
shorter 3 mm fiber.
APPLICATiONS OF CFRC
ACKNOWLEDGMENTS
REFERENCES
4. Ohama, Y. and Amano, M., “Effects of Silica Fume and Water Reducing
Agent on Carbon Fiber Reinforced Mortar”, Proc. 27th Japan Congress
on Materials Research, Society of Materials Science, Kyoto, 1984, pp.
187-191,
7. Akihama, S., Kobayashi, M., Suenaga, T., Nakagawa, H. and Sumki, K.,
“Mechanical Properties of Carbon Fiber Cement Composites and the
Application to Buildings (Part 2)”, Kajima Inst. of Construction Tech.,
Report No. 65, Tokyo, Oct. 1986, 54 pp.
10. Banthia, N. and Ohama, Y., “Dynamic Tensile Fracture of Carbon Fiber
Reinforced Cements”, Fiber Reinforced Cements and Concretes: Recent
Developments (Eds. R.N. Swamy and B. Barr) Elsevier Applied Science
Publishers, 1989, pp. 251-260.
11. Ohama, Y., “Carbon-Cement Composites”, Carbon, 27(5), 1989, pp. 729-
737.
13. Ando, T., Sakai, H., Takahashi, K., Hoshijima, T., Awata, M. and Oka,
S., “Fabrication and Properties for a New Carbon Fiber Reinforced
Cement Product”, in Thin Section Fiber Reinforced Concrete and
Ferrocement (Eds. J.I. Daniel and S.P. Shah) American Concrete Institute,
SP-124, 1990, pp. 39-60.
15. Soroushian, P., Aouadi, F. and Nag¡, M., “Latex Modified Carbon Fiber
Reinforced Mortar”, AC1 Materials Journal, 88(1), 1991, pp. 11-18,
19. Banthia, N. and Dubeau, S., “Steel and Carbon Micro-Fiber Reinforced
Cement-Based Materials for Thin Repairs”, ASCE J. of Materials in Civil
Engineering, Vol. 6, No. 1, February 1994, pp. 88-99.
20. Banthia, N., Moncef, A., Cholu-i, K. and Sheng J., “Uniaxial Tensile
Response of Micro-Fiber Reinforced Cement Composites”, RTLEM,
Materials and Structures (in press), 1995.
S/C = O 5, SF/C = O 2
8 Pitch Based Carbon Fiber:
18 p m dia, 3 mm long,
3% fiber -1
t
\
\
I/gL
95 rnm (15 mm
x 7 rnrn)
3 I
I
I
2 c
Strength (MPa) E (GPa) Strain at Peak Load (%
1
4 75 12 50 O 065
2% 6 44 13 94 O 126
3% 14 80 o 220
O
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9
Strain (%)
Fig. 2-Stress-strain curves in uniaxial tension for carbon fiber reinforced cement mortars (20)
L.
4 . 3 . 1 8
--
Paste w 1 c = 0.30
S I c = 0.40
A: Aggregate
S: Silica fume
ví : Fiber volume content
Fig. 3-Stress-strain curves in uniaxial tension for carbon fiber reinforced cement
pastes (5) and carbon fiber reinforced cement mortars (6)
1 .c
0% 6 08 1 57
0.e
I
0.6
n Mortar: WIC = O 35,
SIC = o 5, SFIC = o 2
H Pitch Based Carbon Fiber:
m
V 18 prn dia, 3 rnrn long,
3
0.4
0.2
0.0
0.0 0.3 0.6 0.9 1.2 1.5
Mid-span displacement, mm
Fig. 4-Load-displacement curves for carbon fiber reinforced cement mortars in flexure
45
40
35
I +BanthiaandSbemg(I990)
Beamr: 1 5 x l S x l S O m
-f Ohana et al. (1985) W/C=0.3;
SF=40% ofC; Beams: 40~10x160mm
W/C=0.3;
’
I
+ I I “ II 1)
30
25
20
15
10
5
SF=Silica Fume
O
O 1 2 3 4 5
Volume Fraction of Carbon Fiber (%)
..
0.4 -
O 3-
: Experiment No.3
__
(Third point
- -loading)
._._ ... ..
3,,, 40 -
OW/B=0.35, S/B=0.5
AW/B=0.35, S/B=l
OW/B=0.35, S/B=l, SF=lO% of C
V HW/B=0.35, S/B=l, SF=20% of C
I
O 1 2 3
Volume Fraction of Carbon Fiber (5%)
Fig. &Variation in modulus of elasticity os a function of carbon fiber volume fraction for
reinforced cement mortars
10.0
- 7.5
a
E
In
5.0
-0)
.-
'
In
2.5
O
O 500 1000 1500 2000 2500 30W 3500 4000 4500 51
Strain (x
Fig. 9-Fracture energy increases under impact as a result of carbon fiber reinforcement (10)
%
2000
h
i1500
500
O
O 200 400 600
47 67 87 107
CMOD (pm)
Effective Crack Length (mm)
Fig. 12-0) Opening load versus crack mouth opening displacement curves for some carbon fiber
reinforced mortars from the test shown in Fig. 11. b) Typical R-curves for CFRC (23)
3.5
3
Mortar ( W l M . 6 ; SF40 %of C,SIC%)
2.5
2
1.5
1
0.5
O
O 0.2 0.4 0.6 0.8 1
Volume Fraction of Carbon Fiber (%)
Fig. 13-Variation in maximum crack width as a function of carbon fiber volume fraction for
reinforced pastes and mortars (18)
SP 154-18
Development of Superworkable
Concrete for MuIti- Functional
Port Str uctu res
by T. Fukute, A. Moriwake,
K. Sano, and K. Hamasaki
Svnoosis: The Japanese economy has been highly developed through foreign trade. Port facilities
have been supporting this econoinic growth and inany concrete port structures have been
constructed and maintained during the past few decades.
Recently, various social and cconomical demands have required port facilities to he inulti-
functional. New facilities are being constructed to meet this trend. These changes include new
types of breakwaters, revetment and under-sea tunnels which iinprove aesthetics, and reduce cost,
labor and construction lime.
Fresh concrete used in the construction of these new types of structures is often required to have
high flowability and be sclf coiiipactable because of thc complicated shape and densely
arranged reinforceiiients of these structures.
In ordcr to ineet these demands, the authors have developed super workable concrete using
viscous admixture (ccgregation reducing admixture) and superplasticizer.
In this paper, the ink design and material properties of this super workable concrete and
exaniples of its application to new port concrete structures are presented.
335
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
336 Fukute et al
T. Fukute is the chief of Materials Laboratory, Structural Engineering Division, Port and Harbour
Research Institute, Ministry of Transport, Yokosuka, Japan. He received his Dr. Eng. from Nagoya
University in 1984. He has been engaged in research related to durability of concrete and
development of new type materials for marine environment.
A. Moriwake is the chief of Materials Laboratory, Technical Research Institute, Toa Corporation,
Yokohama, Japan. He has been closely associated with durability of concrete structures.
K. Sano is the chief of Materials Laboratory, Naruo Engineering Research Institute, Toyo
Construction Co., Ltd., Nishinomiya, Japan. He has been engaged in research related to
anti-washout underwater concrete and durability of reinforced concrete structures.
The super workable concrete developed here have three basic characteristics, namely, high
flowability, high segregation resistance and high self compactibility. To provide these properties
in concrete, mixture proportions and materials are proposed as shown in Fig.l(').
Flowability, an important property of the super workable concrete, is commonly evaluated by
its slump flow value. This value is determined by measuring the diameter of concrete base instead
of the slump of the concrete top. When the slump flow value is large, the flowability is evaluated
to be high.
To achieve satisfactory self compactibility as for a super workable concrete, high segregation
resistance is as essential as flowability. To evaluate the segregation resistance, several test
methods such as the V-shape funnel test, the box test and others are proposed. In the V-shape
funnel test, as shown in Fig.2, the falling time of the concrete through the 75mm square outlet
of the funnel is measured. For the box test, a transparent box separated at the center is used as
shown in Fig.3. The test results are evaluated by the difference of concrete levels in two
chambers after the concrete has flowed from one chamber into another through the obstacles of
reinforcement with 50mm opening.
These test results are explained in connection with rheology parameters. The yield value and
plastic viscosity are expressed as Fig.4 when concrete is assumed to be a Bingham liquid('). The
relations of the slump flow values and yield value are indicated by some researchers. The falling
time of V-shape funnel test and the slump flow speed are suggested to be related to the plastic
viscosity of concrete. The slump flow speed is represented by flow time which is obtained by
measuring the time of the concrete reaching a size of 50cm in diameter in slump flow test. It is
understood that these several tests are essential to evaluate the super workable concrete from the
viewpoint of rheology.
The segregation resistance and self compactibility of this super workable concrete is provided
by a viscous admixture which is made from "water-soluble cellulose ether" and flowability is
provided by superplasticizer. This conaete technology is an extension of anti-washout underwater
concrete in which a high volume of the viscous agent and superplasticizer are dosed to get
segregation resistance and self compactibility in water. The outline of mixture proportions of super
workable concrete developed here is shown in Table 1. The properties of fresh concrete are
dominated not only by the viscous admixture and/or superplasticizer but also by water content,
sand-aggregate ratio and content of coarse aggregate in the concrete.
The initial and final time of setting of super workable concrete show delay by the influence
of viscous admixture and superplasticizer when compared with those of conventional concrete. The
difference of time of setting measured for super workable concrete and for conventional concrete
mixed with blast-furnace slag cement type WBB) specified by JIS R 5211, for example, are two
hours and 3.5 hours respectively as shown in Fig.5. The delay of initial setting time of super
workable concrete leads to the increase of lateral pressure against the formwork. It should be
assumed that the lateral pressure of concrete is distributed hydrostatically if the rising rate of
concrete placing is high(3).
The specimens for compressive and splitting tensile strength test of super workable concrete
are cast with rocking (without rodding) the cylindrical molds of 10cm in diameter and 20cm in
height. The compressive strength and splitting tensile strength are measured according to JIS A
1108 and JIS A 1113 respectively after 3,7 and 28 days of aging.
The relations between compressive strength and Young's modulus are shown in Fig.6. Young's
modulus of super workable concrete is a little less than that of conventional concrete (solid line
in the figure, as specified by JSCE)(4). The relations of compressive and tensile splifting strength
are shown in Fig.7. The splitting tensile strength of super workable concrete seems to be larger
than that of conventional concrete indicated by solid line when the compressive strength is high.
Drying shrinkage
Specimens for drying shrinkage were cast in accordance with JIS A 1129 and removed from
the mold the day after casting. After 7 days of curing in water, the measurements of drying
shrinkage of the specimens cured in a room of 20°C temperature and 60% relative humidity are
started.
The test results are shown in Fig& The drying shrinkage of super workable concrete is
observed to have close relation to the volume of mixing water. The shrinkage of concrete mixed
with 200kg/m3 of water is larger than those of concrete mixed with 180kg/m3 and 190kgim3. To
decrease the drying shrinkage of the super workable concrete with a viscous admixture and
superplasticizer, it is important to decrease the amount of mixing water.
To measure the chloride penetration depth, specimens are prepared with the same procedure
as those for the compressive strength test and are cured in water for 91 days before a testing. The
penetration depth is evaluated by soaking the cylindrical specimens in sea water for specified
periods. The specimens are split and are sprayed with 0.1% fluorescent natrium solution and with
0.1N silver nitrate solution to evaluate the penetration depth of chloride. The test results are
expressed by the average depth of the fluorescent area where the chloride concentration is
evaluated to be more than 0.25% by weight of concrete. The comparison of penetration depth
between super workable concrete(W/C=50%,,55%,60%) and conventional concrete(W/C=55%)
are shown in Fig.9.
The results show that penetration depth into super workable concrete is less, regardless of
W/C, than that of conventional concrete. The reason is considered to be the decrease of defects
induced by bleeding in super workable concrete. The difference of the penetration depth of
chloride ions is not obsened in super workable concrete, regardless of its W/C.
Carbonation
The specimens for carbonation test are prepared with the same procedure as for the test of
penetration depth of chloride ions. After 91 days of curing in water, specimens are placed in a
room of 20°C temperature and 60% relative humidity for specified terms. Carbonation depth
from the surface is evaluated by spraying of 1% phenolphthalein alcohol solution. It is seen that
carbonation depth of super workable concrete (WiC=50%,55%,60%) is less than that of
conventional concrete (W/C=55%) as shown in Fig.10.
High flowability and segregation resistance make it possible for the super workable concrete
to flow a longer distance. Before applying the super workable concrete to actual structures, the
influence of flow distance on the quality of concrete must be assessed.
The compressive strength, Young's modulus and coarse aggregate content are shown
respectively in Fig.l I,i2,13 versus the flow distance. The fluctuation of compressive strength
(expressed by the ratio of compressive strength of core to that of specimen cured in water of
20°C) is not observed noticeably along the flow distance. The same tendency can be observed in
Young's modulus ratio (Young's modulus of core is divided by that of specimen cured in water
of 20°C). The segregation resistance or uniformity of concrete along the flow distance is evaluated
by coarse aggregate area ratio (the area of coarse aggregate appeared on the core surface is divided
by total area of core surface). From the measurements, uniformity of concrete along flow distance
of 7.2111 is maintained.
From these assessments, it is suggested that the super workable concrete shows high
reliability when it is applied to actual structures in which the reinforcement is densely arranged.
Outline
From the previous studies, the super workable concrete using a viscous admixture and
superplasticizer is confirmed to have good flowability, segregation resistance and self
compctibility. In this section, an example of the super workable concrete applied to a new type
of breakwater is described. The qualities of concrete are traced with sampling at the batching
plant, at delivery site, at the top of the conveying pipe and as it is placed in-situ through the
reinforcement.
The double cylindrical caisson type breakwater, as shown in Fig.14, is adopted for Shibayama
port area. The location of the port is shown in Fig.15. In this area, the depth of the seabed is about
30in and wave conditions are severe in winter seasons. This breakwater is developed for the
structures suitable for the deep sea and the arca of rough waves with the double cylindrical walls
dissipating wave energy effectively.
This caisson has a densely reinforced floor in which 14 layers of reinforcement is arranged
crosswise in 1 . h of thickness. As it is evaluated that compaction of conventional concrete is
difficult, the super workable concrete is applied.
Materials and inixture proportions used for the super workable concrete are shown in Table
3 and 4. The concrete has been dosed with 0.3% viscous admixtures by weight relative to
mixing water so as to get enough segregation resistance. The super plasticizer made with "poly
carboxylic ether" and "crosslinked polymer" is used to maintain the workability during the placing.
WiC of 48% is adopted for the concrete mixture to attain a coiiipressive strength of iiiorc than
30MPa at the age of 28 days.
As the super workable concrete is placed without coiiipaction, properties of fresh concrete
(flowability, segregation resistance and filling ability) definitely dorninate the reliabili~y and
durability of concrete structures. Thus, the quality control of super workable concrete at the inixing
plant is very iinportant. The inking tiiiie, as shown in Fig.16, is sufficiently long s o aï to disperse
the siiiall voluiiie of viscous adiiiixture to obtain uni forin concrete.
One of the inost iiiiportant control itcnis to inix this type of concrete at a inking plant is to
ineasure the inoisturc content of fine aggregate. To satisfy this requireinent, a inicrowave nioisture
sensor is inounted just under the sand loading gate (just before the weighing bucket) as shown in
P h o t o 2 The iiioisturc content nf sand is inonitored at every hatch by the sensor. In accordance
with JIS A i I I l , iiieasureiiient of iiioisture content is perforiiicd at every tenth batch iiianually.
As shown in Fig.17, both ineasurciiients are coincident with each other. Froin these results, it is
confiriiied that continuous monitoring by microwave sensor is useful in stabilizing the quality of
fresh concrete. Averages of four or five ineasureiiients by the sensor are carried over to the next
inixture.
Delivery
It takes about 30 iiiinutes for concrete delivery froiii the batching plant to the placing site by
truck agitators. The changes of properties of fresh concrete during transportation are traced.
The changes of slump flow ineasured at the batching plant and at the placing site are shown
in Fig.18. This figure shows that the slump flow is increased by about 5cin during the
transportation. This characteristic of concrete is considered to coine froin the superplasicizer. The
ineasuteinents of sluinp flow at the placing site are satisfactorily controlled within the specified
value of 60 2 5ciii. The 50ciii flow t h e and falling tiiiie of V-shape funnel are shown in Fig.19
and Fig.20. Froin these test results, noticeable changes during the transportation are not
inoni tored.
Placing
After the concrete is delivered, mobile concrete pumps, shown in Photo.3, are used for placing.
The concrete is conveyed to the proper position with flexible hoses connected to a turn table as
shown in Photo.4. Birds'-eye views during concrete work are shown in Photo.5. It takes about 12
hours to place about 760m3 of super workable concrete without the noisy vibrator. For the placing
work, eight skilled workers (two for operating the mobile pump, four for top of flexible hose and
two to operate the turn table) are engaged.
Due to the viscosity of super workable concrete, pumping resistance is observed to increase
in comparison with conventional concrete. As shown in Table 5, the pumping pressure reduced
the slump flow by 5cm. The 50cm flow time, the falling time of V-shape funnel and the air
content in concrete increased a little during the pumping process.
Curinp
As bleeding of super workable concrete is negligible, curing of the concrete surface should
immediately follow the completion of concrete placing. In this case, the concrete surface is
covered with a thin layer of potable water and cured for seven days.
The test results of compressive strength of concrete specimens made during the work are
shown in Fig.21. The test results show satisfactory strength and small deviation in strength
through the placing are observed. The compressive strength of specimens sampled at mixing plant,
after delivery, after pumping, after flowing through reinforcements are shown in Fig.22. There is
little difference in compressive strength between the specimens. From these measurements, it is
obvious that the work process affected the concrete strength just a little.
The spacing factor of concrete sampled at each step of work (mentioned above) has
been measured in accordance with ASTM C457-94. As shown in Fig.23, the spacing factor of
concrete becomes smaller (less than 15Oflm) as the work proceeds. The effective change of
spacing factor is observed during the pumping work. As advantageous changes are observed in the
spacing factor, the concrete is assumed to have enough resistance against freezing and thawing
action.
The pore size distribution measured with a mercury intrusion porosimeter is shown in Fig.23.
The pore volume of around IO4 angstrom in pore diameter is observed to increase during the
pumping process. It is assumed that the pumping pressure affects air-void system in concrete.
Though these microscopic changes are negligible, these test results tell the importance of
appropriate work process which is suitable for the properties of concrete in constructing reliable
concrete structures.
CONCLUSIONS
The authors have developed super workable concrete with viscous admixtures and have applied
it i o actual concrete structures in which reinforcement is densely arranged. The following
conclusions are made from the above work:
(1) Concrete with viscous admixtures and superplasticizer shows high flowability, segregation
resistance and self compactibility.
2) The initial and final time of setting of super workable concrete tend to delay when compared
with those of conventional concrete. This delay is due to the viscous admixtures and super-
plasticizer.
(3) Mechanical properties of the super workable concrete such as compressive, splitting tensile
strength and Young's modulus are similar to conventional concrete.
(4) To reduce the drying shrinkage of this super workable concrete, it is important to reduce the
mixing water.
(5) Penetration depth of chloride ions and carbonation of the super workable concrete are less than
that of conventional concrete.
(6) From the application of super workable concrete to a large-scale structure in which the
reinforcement is densely arranged, it is confirmed that few skilled workers are needed to place
concrete without the noisy vibrators.
(7) Judging from the fresh and hardened concrete sampled at each execution process, change in
concrete properties during execution is ncgligible.
ACKNOWLEDGEMENTS
The authors would like to express sincere thanks to the Third District Port Construction
Bureau of Ministry of Transport and the persons involved in the breakwater construction. The
developinent of super workable concrete was performed and is being coniinued by "The
Research Group of Super workable concrete for Marine Environment" which consists of 12
organizations, Oinotoguini Co.,Ltd., Japan Industrial Land Development Co.,Lid., Sack¡
Kensetsu Kogyo Co.,Ltd., Daito Logyo Co.,Ltd., Honina Corporation, Mitsui Harbour and
Urban Construction Co.,Ltd., Rinkai Construction &.,Lid., Wakachiku Construction Co.,Ltd
and authors.
REFERENCES
Mixture p r o p o r t i o n s
Water content Sand Volume of Visoxity ageit
per unit volume percentage uiarse aggregates dosage
Type of cement (kglm31 (%) (l/m3) (W*%)
180 190 200 46 48 50 52 300 325350 0.25 0.35
I I I I l I I I I I I I I
Blast-hace slagcement
type-B*) !U
I
I
I
I
,
I
I
I
I
I
I
I
,
,
u
,
I
‘i!I
I
j
I
I I
I
I
I I
Blast-hace slagcement
type-BOow heat type)’’
, I ,
I
; i
,
I
,
I I ,
i; :
<
ai , , I I I
l ’ r < ~ i i r i i i c i ipcriod
i Slwiip ilow value 50cni Sliinip Ilow iiiiie V-fuiinnel test value Air content value
(Ciil) (sec) (sec) (%)
A i i u deli\ cry
(heliirc piiriiliiiig) 64 5 11.4 14.6 4. I
,\llcr [iwiipiiig
io0 13.5 17.3 4.8
500
u75 75
(in mm>
35
30
25
20
15 . .
10 ANormal concreteV P C)
ANormal concrete(BB)
5
O Supenvorkable concretc(BB)
A
O
Fig. L T i m e of setting
35
30
h
a 25
$?9
-
-0
1
20
2 15
in
-9 10
5
O 5 10 15 20 25 30 35 40 45
Compressive strength fc(MPa)
(
5 I I 1 I l 1 I I l
- 03days A -
M
fi
e
+-a 3 -
v>
-
M -
Conventional concrete -
-
O 5 10 15 20 25 30 35 40 45
Compressive strength fc(MPa)
I . , , . I . . ,
O 5 10 15
Drying t¡me(weeks)
15 I I I 6 I I I
Exposure(months)
151 I I I I l I I
I
Exposure(rnonths)
u I I I l I I 1
1. o 20 30 0.4 I O 6.0 1. o
Flow distance (m)
.................... ____________________----
8 --o----
O 0 O 0 .
O .
O
_--_________________-----__-__-----_-___----------_
O : upper layer
O : middle and lower layer
I l I I l I l
.9.00; , , InsideofPort
c w c r smsrc11
P 12
:
L
2 8
10
.-
v)
L
F 61
- Sensor
g
a 4 +JISA1111
' t 2
% 50 100 150 200 250 300
Number of batch
Fig. 17-Results of moisture measurement of fine aggregate
70
I
i3
? 50
o-o after delivery
cr, 45
Hafter mixing
4% 20 40 60 80 1001201401~
Number of the agitating truck
8
10 5 o-o afterdelivery
H after mixing
=: Haíter mixing
% 20 40 60 80 1001201401~0
Number of the agitating truck
3 50
'
Q
5 40 B
0
- - + ~ - #
E 30
+
0
.-9
(B
2
(B
!?? 10 o-e
7days
E" H 28days
40
S
al
a
7 d w 4)
20
10
O
after mixing after delivesr after pumping after placing
Step of works
n
300
E 250
I
W
+0 200'
'o
0-l
.-
!=
150
D
2 100
50
O
after mixing after delivery after pumping after placing
Step of works
Fig. 23-Results of measurement of spacing factor
O.OSO(
0.01oc
h
M
\ 0.0100
o
W
o
W
-5>
O
0.0200
W
8 0.0100
a
Pore radius(Angstrom)
I - I - __ -
Photo 5 - Bird’s-eye views during concrete work
SP 154-19
Synopsis: Over the last 50 years, many researchers have investigated the
subject of sulfate attack of concrete and special cements have been formulated
to combat the problem. However, recent site investigative work and
laboratory-based studies carried out by the Building Research Establishment
have shown that a particular form of sulfate attack can proceed even in some
concretes which were specifically designed to provide good sulfate resistance.
In this type of attack, the main mechanism of deterioration is the breakdown
of the calcium silicate hydrate phases in the hardened cement paste in the
presence of an available supply of sulfate and carbonate ions to produce the
minerai thaumasite (CaSi03.CaSOq.CaC03.15H20).A microstnicturai
overview incorporating three examples of the thaumasite form of sulfate
attack is presented in this paper. The combination of optical and electron
microscopy has proved to be a very powerful technique for examining the
processes of deterioration in carefully selected site samples.
357
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 Ob62949 0522547 372
358 Crammond and Halliwell
INTRODUCTION
The chemical reactions responsible for causing sulfate attack of concrete have
been studied by many researchers over the years. The two main reactions
which have been identified are the transformation of portiandite (Ca(OH)2) to
gypsum (CaS04.2H20) and calcium aluminate hydrate to ettringite
(3CaO.Al203.3CaSO4.31H20). These findings resulted in the formulation of
sulfate-resisting portland cement (SRYC), in which the quantity of the
sulfate-susceptible tricalcium aluminate is kept at a low level. SRPC was
introduced into the UK in the 1960’s and has since been widely used in
concrete construction where sulfate resistance has been specified. During this
era the instances of sulfate attack in the field were mainly restricted to ground
floor slabs and other forms of concrete with at least one evaporating surface to
encourage the throughput of sulfate ions and therefore the initiation of sulfate
attack. A large-scale UK project was launched in 1970 in order to simulate the
performance of foundation concrete with no evaporating surfaces by burying
concrete specimens in a wet, sulfate-rich clay site. Thames Valley flint
aggregate was the type of aggregate used exclusively in the preparation of
these buried concretes. The analyses of the specimens recentiy excavated after
15 years exposure have shown that the SRPC concretes have performed better
than those made with normal portland cement. In practical terms, however,
the general performance of the normal portland-cement concretes was
satisfactory because, even though some deterioration due to sulfate attack was
identified in the outer layers of the weaker concretes, the interiors were found
to be still sound (l).
Despite these reassuring results, BRE have identified two recent cases in the
UK where sulfate attack has lead to rapid deterioration in foundation concretes
with a different composition to those studied in the above long-term
investigation as they contained a source of finely divided calcium/magnesium
carbonate. The first case (referred to as Case A) occurred in the tops of
foundation piles made using SRPC concrete (2) and the second (referred to as
Case C) in normal portland-cement concrete strip foundations (3). A third
case (referred to as Case B) has also been reported in the Canadian Arctic (4),
where concrete columns and slab-on-grade containing carbonate aggregates
deteriorated so rapidly that some had to be replaced two years after
construction.
The main difference in the degradation process found in these recent case
studies is that, in the presence of available carbonate and sulfate ions, a third
chemical reaction can occur in which the calcium silicate hydrates (CSH) in
the hardened cement paste react to form the mineral thaumasite
(CaSi03.CaC03.CaS04.15H20). SRPC concretes are just as susceptible to
this form of sulfate attack as normal portland-cement concretes because they
each contain similar quantities of CSH (2,576,778).The Building Research
Establishment ( B E ) has reproduced this form of sulfate attack in laboratory-
cast concrete and mortar specimens made using both sulfate-resisting and
normal portland cements and a variety of aggregates containing a readily
available source of calcium carbonate. ( 2 3 . Deterioration has been observed
in test specimens exposed to a range of different sulfate solutions and is more
rapid in the samples stored at cold temperatures below 15OC. A promising
observation from these preliminary experiments has indicated that the use of
slag cement could prevent thaumasite from forming in a concrete containing a
readily available supply of carbonate ions.
This paper presents an overview of the three recent case studies A, B and C.
The paper concentrates on the microstructural aspects of these investigations
because a great deal has been learnt of the reaction mechanisms by studying
the changes in microsîructure encountered during the deterioration process of
the concretes. Particular attention is given to the most recent investigation
(Case C) in which the technique of using a combination of opticai microscopy
and scanning electron microscopy on the same polished sample proved to be a
very powerful analytical tool.
Case A
In July 1990, BRE investigated the cause of severe deterioration in the tops
of 300 mm diameter concrete piles below a 4-year-old housing estate (2). The
piles had been cast on site using SRPC and an oolitic limestone aggregate.
They had been exposed to extremely wet, moderately severe Class 3 sulfite
conditions (analysis of the 2:l clay extract in accordance with BRE Digest
363 (9) was 2.5 g/L SO42-,0.56 g/L Caz+, 0.18 g/L Mg2+ and 0.30 g/L
Na+). Sulfates had penetrated to a depth of several tens of millimetres into
the tops of the piles, transforming the high quality pile concrete into an
incohesive, white pulpy mass containing abundant thaumasite. The
requirements for sulfate resistance current at the time of construction had been
fulfilled as sound parts of the concrete were well compacted and contained a
recommended cement type and content. This was the first recorded case
where deterioration due to sulfate attack had occurred in concrete foundations
specifically designed to provide good sulfate-resistingproperties.
Sound samples of the pile material and groundbeam concrete (SRPC concrete
containing dolomitic and oolitic limestone coarse aggregate and a quartz sand)
were exposed to various laboratory-prepared sulfate solutions at 5OC and
deterioration as a resuIt of sulfate attack was reproduced in both MgS04 and
Na2SO4 solutions. The deterioration was so extreme that the groundbeam
sample stored for three years in strong MgSO4 deteriorated into a sludge
which could be poured into a perspex box as shown in Fig. 1. These
laboratory studies provided conclusive evidence that the carbonate ions
needed for thaumasite formation were derived from the concretes themselves.
Case B
In the early 199O’s, another case of the thaumasite form of sulfate attack was
discovered by Bickley et ai (4) during an inspection of columns supporting a
building and parts of a slab-on-grade of a hangar at Resolute Bay in the
Canadian Arctic. Expansion and premature loss of structural integrity had
occurred to such an extent in some of the bases to the columns that they had
to be replaced two years after construction. Samples of the deteriorated
concrete had a mush-like consistency and contained the reaction products
gypsum and thaumasite. The base of the piers were embedded in permafrost
for 10 months of the year and for the remaining 2 months, they stood in the
ponded melt water. The levels of sulfates detected in the groundwater ranged
in severity from ’positive’ to ’considerable’ and the elevated levels of Mg and
Na in the deteriorated concrete samples suggested that both these cations were
present in the groundwater. The concrete was composed of portland cement
(both sulfate resisting and normal) with a coarse aggregate comprising both
dolomite and calcite and a fine aggregate dominated by the same two
minerals.
Case C
A large crack 10 to 20mm deep had opened up along the damp proof course
as shown in Fig. 2,3,4 with a follow-on crack passing downward through the
brickwork footings (normally above ground level) and the underlying
blockwork (normally below ground level) into the concrete foundations. This
cracking was first noticed between two and five years ago and has become
progressively worse ever since. A gulley or ditch which frequently becomes
water logged runs close by the defective wall and the back garden often
floods as a result. The site is therefore normally very wet.
The 2:l clay extract results (1.88 g/L S042-, 0.64 g/L Ca2+, 0.07 g/L
Mg2+ and 0.09 g/L Na+) showed a composition characteristic of Class 2
conditions, which is very close to that of a saturated gypsum solution. The
recommendations in BRE Digest (9) permit the use of normal portland-
cement concrete in Class 2 conditions provided it contains a minimum cement
content of 330 k,g/m3 (approximately 14%). A selection of samples were
collected from the strip foundations, all showing different degrees of
deterioration. Their cement contents ranged from between 5 % to 11% and
were therefore all lower than the recommended value. However, none of the
samples examined were completely sound as they all showed some evidence
of sulfate attack. This may have affected their cement contents as the low
values found in the most damaged concretes were a result of severe leaching
of the cement paste (see later section). It is therefore possible that leaching
has accompanied sulfate attack to a lesser degree in the other samples and so
all the cement contents analysed could be lower than the actual value of the
concrete as placed.
BRE carried out a geotechnical assessment of the site which concluded that
significant shrinkage of the clay soil had occurred under the defective rear
garage wall resulting from moisture removal by the nearby vegetation. Had
the concrete been sound, the foundation would possibly have been able to
span over the area of ground subsidence (since the loads on the wall are
small), resulting in littie or no damage to the wall. However, given that the
foundation concrete had been weakened by deterioration, it appeared to have
been unable to span the necessary distance resulting in the localiseú damage.
There was also significant lateral movement of the wall outwards from the
rest of the garage and this could be explained if the sulfate attack of the
foundations was accompanied by expansion.
MICROSTRUCTURALINVESTIGATION
Case A
-
.Case B
The swelling and softening of the cement paste in the severely deteriorated
Resolute Bay concretes was due to sulfate attack in which the main reaction
products were gypsum and thaumasite. The authors deduced that the gypsum
formed as a consequence of 'acid' attack in which reaction between CSH and
magnesium sulfate produced gypsum, brucite and silica gel. It follows that,
in a concrete containing an available supply of carbonate ions, the formation
of thaumasite is the next logical step in the reaction sequence. The most
obvious source of these carbonate ions was found to be the fine grained
calcite and dolomite present within the concrete itself. Microanalysis using
SEM was carried out on both polished and fracture surfaces which detected
that the normal hydrated portland-cement microstructure was only present in
isolated areas. The aggregate/paste boundary was found to be very weak and
aggregate pieces were frequently surrounded by empty peripheral voids.
However, it was also observed that the most distressed areas were the least
cracked. Ettringite was not observed in any of the samples examined.
Case C
In case C, the technique of studying the same area of sample using both
petrography and SEM assisted in identifying the following Ave stages of
deterioration (called stages 1 to 5) in a selection of samples coilected from
site.
Stage 1 The cement paste matrix of the strip foundation concrete contained
abundant spherical pores and in the first stage of the deterioration process,
many of these were found to be lined with ettringite as shown in Fig. 6
(photomicrograph taken under plane polarised light and x50 magnification).
In addition, the surrounding cement paste shown in Fig. 7 contained degraded
cement clinker grains (one example shown at the centre of Fig. 7), small
accumulations of ettringite approximately 10 to 20 pm across (pale-colored,
cracked areas spread throughout the paste in Fig. 7) and sulfur-bewing
calcium silicate hydrate (CSH).
The cement pastes in these areas still contained the ettringite accumulations
and the calcium silicate phases were found to be more enriched in Si and
depleted in Ca compared with their normal composition. Crumbie et al (12)
have shown that excess Ca is sometimes needed to satis@ the formation of the
ettringite and gypsum formed during sulfate attack and that under certain
conditions, this Ca can be provided through progressive dissolution of CSH
until only hydrated silica remains. It therefore follows that the breakdown of
the CSH by preferential removal of calcium ions could readily lead to the
formation of thaumasite in a concrete undergoing sulfate attack and containing
an available source of carbonate ions.
Stage 3 The microsbucture changed yet again in this stage of the reaction
sequence as the calcium silicate phases, calcium aluminate phases (including
the ettringite accumulations) and fine carbonate particles less than about 150
gim in diameter present within the cement paste matrix disappeared and were
replaced with the low birefringent 'type II' thaumasite. This thaumasite can
be seen in Fig. 10 as the pale-colored material spreading in from the left hand
side of the photomicrograph (taken under plane polarised light and x125
magnification). Ferrite was often the only one of the original cement phases
remaining but even it disappeared in the most severely attacked areas. Fig. 11
represents another view of thaumasite replacing the cement-paste matrix.
Three circular features can be seen which were originally pores but were then
filled with the denser, more highly birefringent type I thaumasite. This is the
view as seen using an SEM and in the accompaning X-ray map, lower
amounts of Al appear to be present in the type 1 thaumasite compared with
the type II variety.
The samples taken from the positions marked as '3' in Fig. 4 have reached
stage 3 of the deterioration process within their surface layers and show
evidence of stages 1 and 2 at depth. These samples were still relatively robust
in hand specimen, although their outer surfaces, which contain abundant,
soft, white inclusions were more friable.
Stage 4 The amount of void space found at this stage of the deterioration
process was in the order of 20% making it considerably higher than the
previous stages. It was found by means of optical microscopy that these pores
were not spherical as shown in Fig. 6 but were more interlocking. The
concrete had in fact become more porous through the leaching of the
spherical accumulations of type I thaumasite. In the worst areas, the
aggregate pieces were being loosely held together with type II thaumasite.
Isolated areas, usually adjacent to aggregate fragments, contained remnant
cement paste with ferrite being the only cement phase remaining. Large
crystals of gypsum were observed in places intermixed with the type II
thaumasite but no evidence was found of ettringite or portíandite.
The sample which was taken fiom the position marked '4' in Fig. 4, just
below the subsidence crack, had reached 'stage 4' in the deterioration
process. This sample was much more friable throughout and easily
disintegrated in the hand. The spaces between the aggregate pieces were filled
with a soft, white material identified as thaumasite by XRD.
Stage 5 The final stage of the deterioration process involved severe leaching
of the remaining type II thaumasite from the concrete leaving a very porous,
extremely friable concrete as shown in Fig.12 (photomicrograph taken under
crossed polars and x25 magnification). This concrete had no compressive
strength and would be capable of a great deal of compaction as in places there
was hardly any cementitious material holding it together. The cement content
was found to be 5 % and the sulfate content was very low (order of 1.4% by
mass of cement). Care must therefore be taken when diagnosing similar cases
of sulfate attack in which the reaction product has come and gone.
The final sample which had reached stage 5 in its reaction sequence was
collected from position '5' in Fig. 4, 100mm below position '4'. This sample
was also very friable throughout but lacked the white, interstitial thaumasite.
DISCUSSION OF MICROSTRUCTURALANALYSIS
In these three case studies, a particularly rapid form of sulfate attack was
identified in concretes in which the cement-paste matrix containing calcium
silicates, calcium aluminates and a finely divided source of calcite, dolomite
or both was replaced by the mineral thaumasite
(CaSi03.CaSOq.CaCQ. 15H20). The onset of this type of deterioration
relies on the stability of the CSH under the specific sulfate conditions found
on site. If the CSH in a concrete reacts with sulfate and carbonate ions under
cold, wet, conditions, the formation of thaumasite will be initiated. Providing
external conditions do not change, this reaction will continue to completion
because once thaumasite forms, its precipitation acts as a powerful driving
force drawing in all the relevant reactants. At 5OC, thaumasite i s in the order
of a hundred times less soluble than at 2oOC (13) which accounts for the
increased rate of deterioration at colder temperatures.
The study of the microstructure of carefully selected site samples has revealed
valuable information on the reaction mechanisms involved in the deterioration
of concrete as a result of thaumasite formation. Examination of the surface of
a polished thin section using both optical and electron microscopy proved to
be particularly useful This technique was used on the normal portland-cement
concrete samples from the 'Case C' investigation, in which five stages of
deterioration were observed and analysed. In the first stage, ettringite was
found in pores and in minute accumulations within the paste. The ettringite in
the pores gradually disappeared and was replaced by a highly birefringent
variety of thaumasite. ln the second stage of the reaction sequence this
thaumasite growing in the pores could be clearly seen coexisting with the
small ettringite accumulations in the adjoining paste. The whole cement paste
and very fine aggregate matix, including the ettringite accumulations, was
replaced by a more typical form of thaumasite in the third stage. The final
two stages of deterioration involved severe leaching of one type of thaumasite
Iollowed by the other leaving a porous, extremely friable and weak concrete.
It was this final leaching process which contributed to the subsidence seen on
site. In other parts of the site, where it was only the surface concrete (down
to a depth of several tens of millimetres) which contained aggregate pieces
Ioosely held together with thaumasite, evidence of movement above ground
was not detected.
There are a number of different reaction mechanisms which can lead to the
thaumasite form of sulfate attack, some involving ettringite or gypsum as
initial reaction products. In the deteriorated tops of the SRPC piles found in
Case A, thaumasite was the only deterioration product identified. However,
this was not so in the other two studies. Ettringite was found to form first
before thaumasite formation in Case C and the samples from Case B
contained large amounts of gypsum as well as thaumasite.
CONCLUSIONS
The thaumasite form of sulfate attack was found to be the cause of severe
deterioration in the three case studies reported in this paper, where portland-
cement concretes containing calcium/magnesium carbonate dust had been
exposed to moderately severe sulfate-bearing groundwaters for a period of
two to twelve years. Microstructural examination has shown that in these
concretes, the calcium silicate and calcium aluminate cement paste minerals
and the dusty carbonate material can be completely replaced by thaumasite.
ACKNOWLEDGEMENTS
The authors would like to thank Andrew Baxter and Mike Crilly for their
assistance on site and Kelvin Pettifer, David Rayment and Mike Bowley and
their respective analytical units for the materials analyses. We also appreciate
the invaluble discussions with the Canadian Researchers working in the field
and others elsewhere including Lars Hjorîh, Torben Knudsen and Mario
Collepardi.
REFERENCES
Fig. 5-Case A: Thaumasite replacing the cement paste and fine aggregate matrix
Fig. &Case C: Ettringite lining spherical voids (center left and center right)
Fig. 9-Case C: Electron photomicrograph and spot analyses of same pore as shown in Fig. 8
Fig. 9 [ont.-Case C: Electron photomicrograph and spot analyses of same pore as shown in
Fig. 8
Fig. 1 L C a s e C: Type II thaumasite replacing cement paste and fine aggregate particles
c 150 prn across
Fig. ll-Case C: Electron photomicrograph and spot analyses for two types of thoumasite
Fig. 12-Case C: Aggregate grains mainly surrounded by void space (black areas)
SP 154-20
381
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-I154 95 = Obb2949 0522573 691
382 Hayakawa, Matsuoka, and Yokota
INTRODUCTION
Filling Work
The types and plan of concrete filled steel tubular columns are shown in
Fig. 1 and Table I. The height of the fill is 39.4m or 36.0m. Pipes and tubes are
utilized as steel tubular columns. The pipe columns have diaphragms only outside
of the pipe, and the tube columns have two diaphragms inside of the tube at each
joint of column and beams. The diaphragm in the tube has a center hole with
diameter of 180mm, and four air release corner holes with diameter of 30mm.
Basically, concrete was placed in the steel tubular columns from bottom of the
columns with a concrete pump.
Super-workable Concrete
FILLING EXPERIMENT
Experiment I
After the concrete had hardened, the steel tube was removed to inspect the
placed concrete. The concrete had fully filled the underside of the diaphragms and
the corners of the column without any voids. Some air bubbles were seen on the
side surface of the concrete, and the number of the bubbles increased with the
height. It was considered that the flow characteristics of the concrete decreased as
concrete went up in the steel tube, and the entrapped air remained on the side
surface of the concrete.
Ex-piment 2
side surface of the column. After removal of the steel tube it was confirmed that
the pumping of the mortar was effective in reducing the air bubbles on the side
surface of the column.
Pressure gauge was set up on the pumping pipe near the column specimen
and concrete pressure was measured and is reflected in F i g 5 To evaluate the
influence of diaphragm on pressure, pressure for pumping was calculated by the
formula shown below:
P=P1+ a WH/98 (1)
where
P = pressure for pumping ( M a )
Pi = pressure from pipeline ( M a )
a = factor from diaphragms
W = unit weight of concrete (tonne/m’)
H = height (m)
Experiment 3
In experiment 3, concrete was placed from the top of the columns. The
purpose of experiment 3 was to examine the filling ability of the super-workable
concrete when the concrete was placed from the top of the column without any
consolidation. The 6.6-m high specimen with 4 diaphragms was used in the
experiment (Fig.2).
Concrete was delivered through a pipe and hose to the top of the specimen
and dropped. After the concrete has hardened, the steel tube was removed and the
filled concrete was inspected. The concrete was placed well under the diaphragms
and corners of the column similar to when the concrete was placed from the bottom
with a pump (experiment 1 and 2).
FILLING WORK
Production of Concrete
Properties of Concrete
The average slump flow of the super-workable concrete used for filling
the columns was 710mm, and the average air content was 2.5%. Compressive
strength of the concrete was shown in Table 3.
A video camera was hung from top of a column to inspect concrete flow
in the tubular column. It was confirmed by observing with the video camera that
the concrete went through the center hole and four comer holes of a diaphragm
simultaneously just as observed in experiment 1.
Some concrete samples were picked from the top of the columns just after
the placing had been completed. Slump flows of the concrete before placing, the
concrete from the top of the columns, and the concrete stored in a container during
the placing are shown in Table 4. The slump flow of the filled concrete decreased
to 360mm, but the slump flow of same concrete stored in a container decreased
much less. The slump flow loss of the filled concrete is due to a reduction of
m o m during the placing of concrete. Higher aggregate content of filled concrete
was also observed (Table 4).
The settlement of the placed concrete is shown in Table 5. When there are
diaphragms in the column, concrete settled about 2mm, and concrete settled about
4mm in the column with no diaphragm. These settlement were considered very
small compared with the filling height of over 36m.
In the actual concrete filling work, concrete pressure was almost equal to
the product of unit weight of the concrete and the height. This means that a in
Strain gauges were set on one of the columns, and the strain in the steel
column was measured. The result is shown in Fig.8,9. Fig.8 shows strains at
third floor, and Fig.9 shows strains at each floor.
A 6-m high removable column was set on one of the 40-m high column to
check the quality 'of filled concrete. Concrete was pumped from the column to the
removable column.
No voids and gaps were seen in the removable column. The ratio of the
total area of the bubbles on the surface of concrete in contact with the diaphragm is
shown in Fig.10. The ratio becomes higher along the height, but the total area of
the bubbles was small. Compressive strength in the column is shown in Fig.11
and Table 6 . Every core specimen showed higher strength than the specified
design strength of 35MPa.
CONCLUSION
REFERENCE
7Symbol
8I
for One
I
Volume
96.9 I
156.5
124.7
C5A 10-850x40-60 1 I 17.8 I 17.8 I
36.0
9.0 215.3
2 I 9.4 I 18.8 I
Total 66 885.2
Settlement (mm)
Type of column No.
1 hour 24 hour
1 1.5 2.0
2 1.o 2.0
3 0.0 3.0
Tube 4 0.5 2.0
5 1 .o 1.o
6 1 .o 2.0
7 2.0 2.5
Pipe 1 I 4.0
~~
Elsvaiinn i
Plan of %eBuilding
-I
O
O
Center Hole: $ 180
i
Corner Holes: $30
o b
i
!
Pump
:$'
- +
O
O
\o
i
O
oi
O
m
O
-L Unit: mm
Experiment 1 Experiment 2 Experiment 3
Fig. 2-Specimens for filling experiments
0.3
E?M 0.2
VY
e!
a
O
O 1 2 3 4 5 6 Height (m)
1 2 3 4 5 6 7 8 9 1 0
I I I I I I I I I I
Number of Diaphragms
Fig. 5-Concrete pressure in Experiment 2
1.o
h
2 0.5
E
: t
80
20
O
O 0.5 1 1.5 2 2.5 3 3.5
Time (hour)
80
60
Y
W
20
O
O 6 12 18 24 30 36 42 48
Time (hour)
Fig. 9-Strains in steel column
I
0.6m
n
E 4 e-a Center
v
o.--- 0-0 Corner
t1
~ 50 60 70 80
Compressive Strength (MPa)
399
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C I SP-154 95 m O L L ~ ~ V0522590
I 770 m
400 Xie and Beaudoin
Ping Xie served as a process engineer in a cement factory, China, 1982-1983, and
is currentiy a postdoctoral fellow, Dept of Civil Eng., Univ. of Ottawa, Canada.
He obtained his Ph.D. in Civil Engineering from the Univ. of Ottawa, 1992. He
has authored or co-authored more than 50 research papers in cement & concrete
technology. His current research focuses on new materials and their applications.
INTRODUCTION
There is a need for an effective solution to problems with ice and snow on
road surfaces without a negative impact on the environment. It is specially
important for the highway and municipal transportation systems in cold regions.
Current de-icing methods usually involve the use of de-icing chemicals (salts). In
spite of the effectiveness of de-icing chemicals (usually containing chlorides), it is
weil recognized that their use can result- in corrosion of reinforcing steel in
highway concrete bridges. Rehabilitation of deteriorated concrete bridges is usually
very costly. Use of de-icing chemicals is also becoming a major environmental
concern. Research and development of novel and environmentally friendiy de-icing
methods are required.
It is apparent that there is still a need to develop simpler and less expensive
alternatives based on electrical de-icing concept. A novel technique using
electrically conductive concrete has been developed at the National Research
Council of Canada (NRC). The new technique is based on the following concept:
A layer of electrically conductive concrete is cast between the road foundation and
the pavement. A few electrodes are embedded in the conductive concrete layer.
Heat is uniformly produced by the conductive concrete and transferred to the road
surface for de-icing when the system is activated by an external power supply. This
is illustrated in Fig. 2. The conductive concrete (i.e. the heating element) is a
primary component of this system. Some technical requirements for this material
are as follows:
and stable conductivity value than normal concrete. A brief comparison of the
characteristics of conductive concrete and normal concrete is given in Table 1.
í
PROPERTIES OF THE NOVEL CONDUCTIVE CONCRETE
1. Mixture 1 (M-F): This system consists mainly of portland cement, silica sand as
fine aggregate and carbon fibers.
2. Mixture 2 (M-G): This system consists mainly of portland cement and granular
carbonaceous material.
(1). Very low resistivity: Low resistivity values, especially for the systems M-G
(about 1.2 Rcm) and M-GF (about 2.5 Qcm), have been obtained. For the
system M-F which contains conductive fibers only, the resistivity value is about
16 Q-cm. The M-GF-S system containing non-conductive coarse aggregate
(about 34% (wt.) by composite mass) has a relatively high resistivity value of
about 20 Qcm. It is however still superior to that reported in the previous
works.
(2). High mechanical strength: The compressive strength is usually above 35-40
MPa at 28 days hydration. A value of 55.2 MPa at 90 days is obtained for the
M-F system. The flexural strength for the conductive fiber-reinforced systems
is significantly enhanced in comparison to the system without fibers (M-G), i.e.
8-1 1 MPa at 28 days hydration. A value of 15.6 MPa was obtained at 90 days
hydration for the system M-F.
(3). Stable resistivity values: The resistivity value for the four composites remains
essentially constant during hydration.
1. &glen'mental
A conductive concrete slab, 12" x 9.5" x 2", was made by casting M-GF-S
concrete formulation into a plastic mould. Bolts were embedded (about 2 cm in
depth) on opposite sides for attaching electrodes. The slab was demoulded after 24
hours curing at lûû% R.H.and 23OC. Curing continued under the same conditions
for an additional 27 days. The slab was then placed in a room under ambient
conditions for a week. Two electrodes, 9.5" x 2" aluminum plates, were
subsequently mounted on the two opposite sides of the slab, Fig. 9(a). A thin layer
of conductive paste was applied at the electrode-slab interface to maintain good
contact.
The experimental setup for the electrical heating test at different air
temperatures below the freezing point is illustrated in Fig. 9(b). The electric
heating element is placed in a soil mass with only one surface exposed to the
environment. This is to simulate practical situations. The heating element and the
soil mass were placed in a freezing room. The temperature of the freezing room is
controllable to -40°C. Temperature at three locations, air, the heater surface and
core, were recorded using thermocouples and a digital thermometer. Electricity, 60
Hz,is supplied by an electric transformer with adjustable output potential. Electric
current intensity, I, through the heater and potential, V, crossing the heater were
monitored by a multimeter.
The power output required for de-icing for a given air temperature can be
adjusted by changing the potential V so that the heater surface temperature is
maintained above the freezing point. Economical electrical de-icing is a process
during which heat loss is in equilibrium with heat supply at the freezing point, 0°C.
In practice, however, the surface temperature control setting was set to be about
3°C to maintain high system efficiency (1). The minimum heater power output in
this experiment was that required to maintain the surface temperature at 2-3OC.
The total heater power output, Pt, can be calculated using the following
equation:
V2
Pt=I.V, or pt= -
R
where, R is the resistance of the heating element. The heat loss occurs
mainly through the upper and lower surfaces of the slab-like heater. It is
proportional to the road surface area and the temperature difference between the
road surface and air. It is therefore more convenient to express the heater power
output in terms of unit surface area, denoted by Pout ,i.e.
Pt
Pout = -
A
CONCLUSIONS
3. The new material is usually lightweight. The density usually ranges from
1450 to 1850 kg/m3. It increases to about 1900-2250 kg/m3 if non-
conductive normal aggregate is used.
5. There is a linear relationship between the air temperature and the mjnimum
heater power output required to maintain the heater surface temperature
above the freezing point. The minimum power output value ranges from
150 to 855 W/m2 of concrete surface as air temperature varies from -5OC
to -3OOC.
REFERENCES
(2). Whittington, H.W., McCarter J., Forde M.C., "The Conduction of Electricity
through Concrete", Mag. Concr. Res., 3(114), 48-60 (1981).
Ratliff J.L., Hawkins H.F., "A Study of Cathodic Protection for Corrosion
Control of Reinforced Concrete Bridge Pilings", Contract No. 99700-7 142-
010, Office of Materials and Research, College of Engineering, Universi9 of
South Florida, Tampa, FX,1975.
(7). Banthia N., Djendane S., Pigeon M., "Electrical Resistivity of Carbon and
Steel Micro-Fiber Reinforced Cements", =(5), 804-8 14 (1992).
(8). Chiou J.M., Zheng Q., Chung D.D., "Electromagnetic Interference Shielding
by Carbon Fibre Reinforced Cement", Composites, 20(4), 379-381 (1989).
Foundation 5
Fig. 1-Schematic of British technique of electrical deicing of road surfaces
I Foundation * Electrode I
Fig. 2-Schematic of electrical deicing of road surfaces using electrically conductive concrete
80
h
E
Y
v 2o16 I Resistiviîy
M-F system
60 h
40 3
20
s
2 4- 8
w
O' 10
O 20 40 60 80 100
Hydration time (days)
1.6 80
h M-G system
E
Y
g 1.2 60
i;j
o
.d
>
.d
0.8
Compressive strength
v
5;
.d
v1
40 c,
M
i2 5
'oE 0.4
s
r/)
20
o
d
w Flexural strength
A
v Q
O n
O 20 40 60 80 100
Hydration time (days)
E9 M-GF system
Resistivity
s 3- - 60 n
2- - 40
O' 'O
O 20 40 60 80 100
Hydration time (days)
80
'=.
25 M-GF-S system
Resistivity
s 20 i 60 n
a
Compressive strength
+ E
Flexural strength
n O
O' 10
O 20 40 60 80 100
Hydration time (days)
2% 2026
-
1635 1579
Power
Power orúoff ?
Power output = ?
i
Tsurface = ?
I
Electric Control
heating system
element
Tair=?
icelsnow
Environmental on road ?
I
12" J
m
Thermometer
Freezing room
5 O
U
v
o 1 Pout = 812 W i d
* Surface
g-20-lob tCore
t
.I
O 2 4 6 8 10
Time (days)
t Air
Pout= 265 Wim'
$
Y -4: 6 Surface
a -8-
I t core
o, -
-12
--
-161 ' ' ' ' ' ' ' ' '
O 2 4 6 8 lo
Time (days)
'
Fig. 11-Surface temperature of powered specimen at air
temperature around -10 C
10
Pout = 1063 W/m '
Pout = 606 W/mz
E
s Air temperature
g -20
5 19
4.9 18.5 3
h
4.9 :
c
O
v
4.8 18
1
<-
3
.+
.-
>
c>
.+
CA
2 4.7 17.5 2
-ri" I ,
-20 -10 O 10 20 30
Temperature of specimen (OC)
1200 -
N
Applied electric
potential -e V=lOv t v=20v
g 800-
9
0 60Oe',-
I I - -
U
-15 -5 5 15 25
Temperature of Specimen ( OC)
SP 154-22
Synopsis: For years the concrete industry has used ultimate compressive
strengthand elastic modulus as principal design and analysis tools. This can be
very misleading when cracking and failure are evaluated. With modern concretes
that include Roller Compacted Concrete (RCC) and lower strength mass
applications, cracking that is serious may not occur until the concrete is strained
well beyond the elastic region.
ILNTRODUCTION
A very high creep rate, which can relax stresses caused by thermal
contraction and foundation deformation, is also associated with lower strengths
that are still quite adequate for dams and other massive applications.
Direct tende strength tests of concrete are difficult and expensive, so they
are seldom performed. The indirect, or B d i a n , splitting tensile test of concrete
cylinders is easy and economical, but it does not indicate the true direct tensile
strength. A conversion factor is needed that depends on the strength of the I
~
Figure 1 shows the typical range of modulus values as a function of age for
most conventional concretes. The purpose of Figure 1 is to demonstrate that
RCC can have very low elastic modulus, but that care must be taken to check this
property because it also can have a very high modulus if the cementitious content
is increased and/or a large amount of pozzolan is used. It should be noted that the
RCC mixtures shown are extreme cases. Generally, concrete with cementitious
contents in the range of about 50 to 100 kg/m3 will have elastic modulus values
on the order of about 5 to 15 GPa.
The modulus for the mix labeled RCC (AGOS) on Figure 1 is for a
concrete that used clean aggregates and was designed to give better strength. It
had a 1 year strength of 14 MPa. The mix had a cementitious content of 120
kg/m3, with 25 % of that being added pozzolan. However, the aggregate
included about 5 % fines (smaller than 75 microns) which were a combination of
natural and freshly ground siliceous sand. These fines were considered to be
pozzolanic, and resulted in the very high long term elastic modulus, despite the
fact that they did not cause a similar increase in strength.
A better look at the stress-strain behavior of the low strength mix in Figure
2 is shown in Figure 3. It helps explain what is happening. The stress-strain
curve is elastic only in the very early stages of stress. It then takes on a toughness
attributed to microcracking, creep, and plastic flow. When these types of
mixtures are tested in compression, they define a similar stress-strain behavior.
The problem with very low tensile strength concrete is that it tends to have
substantial variability when samples are taken. This needs to be acknowledged,
but it is not a major problem for large mass application. Large mass applications
such as dams usually use large aggregate. It is difficult to extract core samples
that use a pcticai and affordable nominal core diameter of 150 mm. A nominal
150 mm diameter core hole actually yields a smaller core diameter. Also, it is
very easy to damage the cores during drilling and handling. Figure 4 shows the
results of direct tensile strength tests on cores drilled from a low tensile strength
RCC dam in Honduras and sent to the USA for evaluation. The nominal diameter
of these cores was about 100 mm and the maximum aggregate size was a nominal
76 mm. This concrete used a poor quaiity and highly variable cement and natural
pozzolan, with a pozzolan percentage that ranged from about 15% to 25 % . The
aggregate was a poor quality ignimbrite with a specific gravity of about 2.3 and
an absorption ranging from 5 % to 11% . The cement plus pozzolan content of the
cores tested was either 90 or 95 kglm'. Although the variability can be criticized,
the overall result is that even with the worst of construction materials, less than
ideal coring, and significant handling of the specimens, the overall in-situ material
has been shown to have substantial strain capacity.
Figure 5 also shows a new and useful value referred to as the "ultimate"
modulus. It is the slope of the secant drawn from the origin of the stress-strain
curve to the average peak compressive strenph of companion cylinders. It is clear
in this typical example that it îakes much more deformation and substantial energy
to fail a lower strength mixture when compared to a higher strength mixture. A
lower ultimate modulus is associated with more deformation before cracking. In
a cracking analysis, it is the ultimate condition that should be studied, not just the
elastic condition. As the mix becomes stronger, the elastic modulus and ultimate
modulus approach the same value, so it is not as important with those types of
concrete. With low strength mixtures, use of the elastic modulus alone in a
cracking analysis can be extraordinarily overconservative, by many orders of
magnitude. Initial work has indicated that the ultimate modulus from compression
testing is a reasonable indicator of the modulus value that can be used for tensile
cracking analyses. Additional work to verify this is underway.
It is clear from these graphs that the elastic modulus decreases dramatically
at compressive strengths below about 10 MPa, and that it is fairly constant above
that level of strength.
Figures 8 and 9 show the relationship between the ultimate modulus and
compressive strength for the same respective projects and broad range of mixtures.
The San Rafael data for ultimate modulus includes data from two different labs.
The mixtures and tests were supervised by the same person, but the work was
done by totally different people who were new to the ultimate modulus. It is clear
from the results that the test can be done with confidence as a standard.
The difference in scale between the graphs showing elastic modulus and
those showing ultimate modulus should be pointed out. At lower strengths, on the
order of 5 MPa, the ultimate modulus is generally about one tenth of the value
of the elastic modulus. As the strength increases, the ultimate modulus gets closer
to the elastic modulus. Data not included in this paper indicates that the values are
essentially the same for very high strength concrete. A large difference between
elastic and ultimate modulus indicates an inherent toughness and improvement in
ultimate strain capacity. The test provides a method of evaluating this very
desirable property when selecting a mix for a project where toughness and ultimate
crack resistance are important.
Another useful tool to help better understand how the elastic and ultimate
modulus are effected by strength is to look at the ratio of modulus to compressive
strength as a function of strength. This is shown in Figures 10 and 11 for the San
Rafael mixtures. The scatter of data in the elastic modulus for the strength level
from about 2 to 10 MPa has been observed at other projects. The peculiar higher
average ratio of elastic modulus to compressive strength within this stress level as
compared to higher strength concretes is also typical. Figure 11 shows that the
large scatter of data is not present for the ultimate modulus test. It is easier to
determine and reproduce than the elastic modulus.
CREEP
A dramatic increase in stress relief due to creep is also associated with low
strength concretes. Creep essentially has the effect of further reducing the
sustained modulus over time. In addition to the benefits from decreased initial
elastic and ultimate moduli, creep has the effect of further reducing the modulus,
referred to as a sustained modulus. In low strength mixtures it can significantly
reduce the long term potential for cracking. Figures 12 and 13 demonstrate the
radical increase in creep when the strength at the time of loading is below about
15 MPa. As shown on the Figures, this relationship appears to be independent of
pozzolan content, whether the mixture is conventional or roller compacted, or the
age at the time of loading. Data for these figures comes from mixtures with a
wide range of cementitious contents, pozzolans ranging from 0% to 70% of the
cementitious content, and ages of loading from 7 to 180 days.
The splitting tensile strength increase with age similar to what can be
expected with compressive strength. Figure 14 shows the relationship between
age and splitting tensile strength for relatively low strength mixtures having a
cementitious content of 89 kg/m3, with pozzolan contents ranging from 0% to
50 % . Figure 15 shows the relationship for the same aggregates but with mixtures
having no pozzolan and a range of cement contents from 59 to 133 kg/m3.
More importantly, the ratio can be very different from project to project.
The tendency is for higher strength mixtures to have splitting tensile strengths that
are a lower percentage of the compressivestrength. However, a trend is not being
shown here because of exceptions. To demonstrate how much the ratio of split
tension to compressive strength can vary from project to project, the following
examples are offered for some typical RCC projects: Upper Stillwater 4 %-7%,
Willow Creek 7%-12%,Monksville 9%-13%,San Rafael 10%-16%,Urugua-I
10%-18%,Concepcion 12%-17%,Big Haynes 12%-18%,and Middle Fork 13%-
18%.
When the adjusting factor is applied to split cylinder strengths, and the
resulting direct tensile strength calculated from appropriate factors, it typically
plots as a straight line function of compressive strength as indicated Figure 20.
This agrees with results from direct tensile tests, at least within the range of about
O to 20 MPa.
CONCLUSIONS
Low strength concretes can have significantly more crack resistance than
indicated by using the elastic modulus as determined in compression tests. An
indication of how resistant the concrete wiii be to cracking, and its ultimate tensile
strain capacity can be obtained by continuing the stress-strain test to failure, and
calculating the "ultimate modulus."
Low strength RCC mixtures can be designed 'to have extremely low
ultimate modulus values, as well as low elastic modulus values.
The split cylinder indirect tensile test is a useful tool for easily obtaining
an indication of direct tensile strength. Special care and adjustment factors are
needed when converting the split tensile strength to direct tensile strength. This
is particularly important at lower strengths.
ACKNOWLEDGEMENTS
The data used in this article was developed from a variety of projects,
involving a number of organizations. Among those who should be recognized are
the Federal Electric Commission of Mexico, Jordan Jones and Goulding, ATEC,
Bechtel, CONCIL, and the North Pacific Division Lab of the U. S. Army Corps
of Engineers.
30 -
25 -
RCC (BG)
1 10 1O 0 1O00
AGE (DAYS)
I 3 240-33-4.3
---l
1
2t d
*I-
f
%
C 4 3 4 . 6
95-25-4.4 .
I I 1 I I
0.4
Q.3
0.2
0.1
-
A C 1 SP-I54 95 I0662949 0 5 2 2 b L 8 b8T I
+CORE R2-25-95A
Advances in Technology 429
o
O 50 1O0 150 200 250 300 350
STRAIN [MILLIONTHS)
0.5 25
* AVG-SDEV
0.4 20
28 Cores, 90 & 95 K g h 3 Type IP cernent
Resulis are grouped by strain capacity
0.3 15
0.2 10
0.1 5
O
O 50 100 150 200 250 300 350 400 450 500
STRAIN (MILLIONTHS)
STRESS iMPa)
20 I I
O
2
O 2 4 6 8 10 12 14 16
STRAIN (millionths x 1000)
O 5 10 15 20 25
COMPRESSION iMPa1
30
MIX
25 - ' RCCNOASH
+ RCCWITHASH
10
9-
8 -
7-
6 -
5-
4-
3-
2-
1-
0-p
15
l4 - MIX
13 -
NO ASH
12 -
a
11 - + WITHASH
3.5
2.5
1.5
0.5
O
O 2 4 6 8 10 12 14 16 18 20
COMPRESSIVE STRENGTH (MPa)
3.5
MEXICO CITY
3 -k SANRAFAELUB
2.5
1.5
0.5
O I
.l I
I
O 2 4 6 8 10 12 14 16 18 20
COMPRESSION iMPa)
35
30 f 23%TO30%
25
Specific Creep Imüiiomths/Mpal = FIK) inlt + 1)
20
15
10
O
O 5 10 15 20 25 30 35 40
STRENGTH AT INITIAL LOADING (MPa)
40
CONCRETE TYPE
35 -
30 - + CONVENTIONAL
25 -
Specifc Creep Imillionths/MPa) = FíK) Ln It+ I ]
20 -
15 -
10 -
.. ++.
-+
5-
+
+, ‘ Y
O’ I I I I I I I
O 5 10 15 20 25 30 35 40
STRENGTH AT INITIAL LOADING (MPa)
Fig. 13-Specific creep versus type of concrete and strength when loaded
% Pozzolan
2.5 0 % + 17% * 33% 50%
1.5
0.5
o I I I I I I 1 1 1 I I I I I I I I I I I I 1 1 1 1 1
1 10 1O 0 1O00
AGE (DAYS)
"I
*-: c
1
1.51
0.5 -
I
-
/------
I I I I I l I I I I I I I I I I 1 I I I I I I I I
30
25
I
+-
MEXICO CITY LAB SANRAFAELLAB
20 -
15 -
10 -
5-
O 5 10 15 20
COMPRESSION (MPa)
0.3
0.25
0.2
0.15
o. 1
A C 1 SP-154
[
Schrader
SPLITTING TENSION/COMPRESSION
. .. ..a*
-
95 m 0bb2949 0522625
0.05
I I I I
O
O 5 10 15 20 25
COMPRESSION iMPa1
0.3
-
' NOASH + WITHASH t WITHRETARDER
0.25
0.2 -
0.15 -
0.1 -
0.05 -
O 5 10 15 20 25
COMPRESSIVE STRENGTH (MPa)
0.5
0.4 1
0.11 p-”
O b I I I I I IlII I I I I I III
o. 1 1 10 1O 0 1O00
COMPRESSIVE STRENGTH (MPa)
1-
f =
d/
0.5 - t
f
I I I I
SP 154-23
Case 2: A 10-year-old precast box bridge beam was removed due to corrosion
of three comer tendons and spalling of the concrete which resulted in an
asymmetrical tendon pattern and cross section. Modal testing was again used
to detect damage in the beam. The beam was then tested to destruction.
Failure was brittle and occurred before the AASHTO nominal moment was
reached. This was probably the result of a lateral instability caused by the
asymmetry of the beam and the damage to the web concrete.
Case 3: Three slab bridges with light, moderate and heavy deterioration of the
concrete over the pier areas were tested using truck loading before, during and
after repairs were made. It was found that removing the deteriorated concrete
from around the top reinforcing bars during repair caused a temporary loss of
continuity which may have resulted in a redistribution of dead load moments.
After repair, the continuity and stiffness were restored in two of the three
cases, but the redistributed dead load moments were probably locked into the
structure.
439
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-I154 95 m 0662947 0522628 529 m
440 Miller, Shahrooz, and Aktan
INTRODUCLION
3) Three slab bridges with various states of deterioration were tested with
nondestructive truck load testing before, during and after repair.
In each case, the deterioration was evaluated and then the effect on structural
performance assessed.
RESEARCH SIGNIFICANCE
In the slab areas under the driving lanes, no deterioration was found.
Core samples showed no signs of corrosion, damage from freezing and
thawing, alkali-silica reaction or chloride intrusion. It is postulated that during
the winter season, chloride laden snow and water was pushed from the
roadway areas to the shoulder areas of the bridge and allowed to remain there.
This chloride laden water soaked through the asphalt and saturated the
concrete slab in the shoulder areas. The saturated slab areas were then
vulnerable to freezing and thawing and alkali-silica reaction. The chlorides
penetrating into the slab accelerated the corrosion.
Modal NDT
For case study 1, modal testing was done before removal of the
asphalt overlay (so the shoulder damage could not yet be seen). The resulting
flexibility matrix revealed abnormalities in the shoulder areas. Thus, the
damage to the shoulders was detected even though it could not be seen. Also,
this first test provided a baseline for a second modal test which was
conducted during the destructive test. This second modal test was conducted
just before the failure load was reached (see next section). Comparisons
between the first and second modal tests showed a significant change in
flexibility in the loaded region, thus revealing damage which was not yet
Destructive Testing
After modal testing was complete, the bridge was tested to failure.
This was accomplished by drilling small core holes in the deck and passing
posttensioning cables through the holes. One end of the cables was securely
grouted into the bed rock below the bridge and the other end was attached to
a 1 555 kN capacity hydraulic cylinder. A total of 4 cable and cylinder
assemblies were used providing a total load capacity of 6 220 kN. The
cylinders were mounted in two concrete blocks cast onto the bridge deck (Fig.
2). The size of the blocks was chosen to simulate a truck tandem axle.
The test took three days to complete. The first day of loading
consisted of loading in the linear range. On the second day of loading, the
bridge was loaded to a total load of 2840 kN and unloaded. At this time, the
bridge was inspected and no additional or severe damage was noted. As
previously noted, a modal test was conducted at this time. Unfortunately, the
data could not be processed before the next loading took place. Subsequent
analysis of the modal test data revealed that there was additional, severe
damage in the deteriorated shoulder regions which occurred due to load
application, but. this damage was not visible.
For the third day of loading, the load was reloaded to 2840 kN in 71
kN increments and additional 71 kN increments were then applied. At 3200
kN, the slab failed suddenly in shear (Fig. 2). Note that this load far
exceeded any reasonable load which may have actually been applied to the
bridge, but it was below the calculated flexural capacity and well below the
calculated punching shear capacity (capacities calculated with f,' = 54 MPa and
f, = 330 m a ) . Although previous research (3) indicates that slab bridges tend
to fail in punching shear, the AASHTO (4) and AC1 (5) provisions for shear
are generally conservative for bridge structures. The fact that the bridge failed
well below the calculated punching shear capacity indicates that deterioration
reduced the shear capacity of this bridge.
The bridge in this study was a skew bridge (Fig. 1). According to
finite element analyses, such bridges should carry loads perpendicular to the
skew. However, at first, this bridge carried load parallel to the traffic lanes.
At a total load of 1000 kN, there was a sudden change in rotational stiffness
at the abutment followed by a sudden increase in rotation at the pier line.
Analysis of instrument responses indicated that the load path had shifted from
parallel to the traffic lanes to the expected path perpendicular to the skew. It
appears that the abutment had been locked or frozen by deterioration and
rotation was restrained. This prevented the bridge from assuming the normal
load carrying path perpendicular to the skew. However, when the load was
sufficient to overcome the resistance at the abutment, the abutment rotated and
the bridge assumed the skewed load path.
The shift of load path to the skew path had an important impact on the
structural response. Once the skew load path was assumed, the damaged
shoulder near the point of load application was no longer in the load path so
the damage had no effect on the structural response.
At a total load of about 2840 kN, the bottom rebar under the point of
load application began to yield. As the bar yielded, the instrumentation
showed that the load path shifted back to a path parallel to the tranic lanes.
After this point, the reinforcing bar over the pier in the damaged shoulder
region (point "A" in Fig. 1) began to show increased strain and yielded. At
a total load of 3200 kN, the damaged shoulder failed in shear and the failure
propagated around the loading blocks (Fig. 2).
At the time of testing, one strand in the beam was missing and one
strand was corroded, but it appeared intact and was still partially embedded.
A third strand was corroded so badly that each of the individual wires were
broken at several places along the length. Damage to the concrete varied
along the length, but the area of missing concrete was approximately 75 mm
x 203 mm. As a result of the deterioration, both the concrete cross section
and the strand pattern were asymmetrical. This asymmetry caused some out
of plane bending; there was a lateral deflection of about 13 mm at the
beginning of the test.
Modal Testing
Destructive Testing
of the testing frame. It carried a load of 260 kN per point and did not fail.
As expected, this beam showed no lateral deflection as the cross section was
symmetrical.
Analysis of Results
The experimental results showed that the beam remained linear elastic
before cracking and precracking behavior of the damaged beam was explained
reasonably well using linear elastic theory. Load-deflection and load-strain
calculations made using linear elastic methods (including FEM) matched the
experimental results quite closely in the precracking region if a correction for
loss of cross section and strands was made. The calculated cracking moment
(using a calculated 18% loss of prestressing force) was 1410 kN-m compared
with the 1440 kN-m cracking moment actually measured.
However, the post cracking behavior was not explained by any simple
theory. Part of the reason for this is that there is no simple and/or accepted
method for post cracking analysis of asymmetrical prestressed beams. A
predicted failure moment of 1930 kN-m was obtained using the AASHTO
Specification (4) (which assumes symmetry) for the damaged beam. When
compared with the actual failure moment of 1805 kN-m, it can be seen that the
AASHTO Specification was not conservative for this beam, even though a
correction was made for loss of cross section and strands.
For this beam, damage again had a significant impact on the structural
behavior. The damage caused an asymmetrical cross section, which resulted
-
CASE STUDY #3 THREE BRIDGES STUDIED UNDER REPAIR
Three bridges were selected for this study; one with light damage, one
with moderate damage and one with heavy damage. Each bridge was a
straight (no skew), three-span concrete slab bridge and had one lane and a
shoulder in each direction.
The first bridge, located in Logan County, Ohio, was chosen because
it had very light damage to the edges and only slight damage to the concrete
over the piers (the main repair of this bridge was removal of an asphalt layer
which was replaced with microsilica concrete). It had spans of 9.1 m, 10.7
m and 9.1 m. The second bridge was located in Mercer County, Ohio and had
spans of 6.1 m, 7.6 m and 6.1 m. This bridge had moderate damage to the
shoulder area and some deterioration of the concrete over the pier lines. The
third bridge, located in Clinton County, Ohio, was severely damaged. Large
areas of deteriorated concrete over the piers, some as large as 6 m across, had
been patched with asphalt. The shoulders and edges of the bridge were also
deteriorated. This bridge had spans of 10 m, 12.2 m and 10 m.
Each bridge was repaired one half at a time with the other half being
open to traffic. The repair method for each bridge was roughly the same:
1) The asphalt surface was removed and the deck was scarified to a depth
of 6 mm (bridges #1 and #2 only, bridge #3 had no asphalt surface)
2) For bridges #2 and #3, a 1 m wide area of the shoulder was removed
along the entire length of the bridge (the transverse steel was left to tie
the new shoulder to the old bridge). A new shoulder was then cast.
3) All unsound areas of concrete on the deck were removed and repaired.
Testing Program
The bridges were loaded using dump trucks loaded with gravel. Each
truck weighed approximately 142 kN (each truck was weighed before the test).
Combinations of 3 or 4 trucks were placed on the bridge at any one time.
The trucks were placed in up to 8 different load positions, where the positions
were chosen to create either maximum moment over the piers, maximum
deflection in the end span or maximum deflection in the middle span (See Fig.
-
6 8). Tests were conducted before repair, after the shoulder was removed
(bridges #2 and #3), after the shoulder was replaced and the deteriorated deck
concrete was removed, and after all repairs were complete.
Test Results
Fig. 7 shows the deflection profiles for bridge #2 for the load
configuration which caused maximum deflection of the center span. The
deflection of the bridge after the removal of the shoulder shows an increase
in deflection near the shoulder (grid line 1), but less of a change away from
the shoulder (grid line 2). This shows that the concrete over the piers was still
effective in carrying the load. The data also show deflections near the center
of the bridge hardly changed between prerepair, under repair 1 (removal of
the shoulder) and under repair 2 (replacement of the shoulder accompanied by
removal of the unsound concrete). After the shoulder was repaired, the bridge
stiffness greatly increased. Once the shoulder was replaced, removal of the
unsound concrete over the piers affected the bridge only slightly. As with
bridge 1, the actual measured deflections were small.
Strain measurements indicated that during repair the portion of the live
load -moment resisted by the positive moment areas increased 60%. After
repair, the portion of the moment resisted by the negative moment regions
increased 16% over the prerepair phase. Again, the strains were small
indicating the associated moments were small. The repair was successful in
restoring stiffness and continuity.
Fig. 8 shows the deflection profiles for bridge #3 for the loading
configuration which causes maximum deflection of the center span. As with
bridge #2, removal of the shoulder and unsound deck concrete caused a loss
of stiffness. Unlike bridges #I or #2, after all repairs were complete the final
deflections of bridge #3 were only a few percent less than the prerepair
deflections. Although all three bridges were repaired in the same manner, the
repairs did little to improve the stiffness of bridge #3.
stage. The problem with the repair of this bridge seemed to be the sequence
of the repair. For bridge 2, the worst deterioration (the shoulder) was
removed first and the remainder of the bridge was able to hold the load during
repair without significant degradation. However, for bridge 3 the worst
deterioration was repaired last. Thus, while the shoulder was being repaired
the deteriorated areas over the piers had to resist the load (especially since the
only sound concrete over the piers was on the shoulder, and this was
removed). Had the deterioration over the piers been repaired first, the repair
might have been more effective. A more complete explanation of this study
is found in the literature (IO).
CONCLUSIONS
After the steel yielded, the load path again shifted to parallel to the
traffic lanes and this brought the deteriorated shoulder into the load
path. The deteriorated shoulder over the pier could not resist the load
and the bridge failed in punching shear at a load 40% lower than that
predicted by AASHTO or AC1 provisions. Thus the deterioration
clearly affected the structural performance and capacity of the bridge.
3) The third case study examined the repair of three continuous slab
bridges to determine if continuity is lost between spans during repair
and to see if repairing the bridges was effective. The tests revealed
that during repair, there is some redistribution of moment from the
negative moment regions over the piers to the positive moment regions.
For two of the three bridges, the repairs improved the bridge stiffness
and restored continuity. For the third bridge, the repairs did not
increase stiffness or restore full continuity. It appears the problem with
the third bridge was sequence of repair. The worst deterioration was
repaired last and the badly deteriorated regions had to carry more load
while lesser deterioration was repaired.
ACKNOWLEDGEMENTS
This research was conducted under various contracts from the Ohio
Department of Transportation (ODOT) and the Federal Highway
Administration. The authors wish to thank W.Edwards and V. Dalal of
ODOT R&D, D. Hanhilammi and L. Welker of the ODOT Bridge Bureau,
District Bridge Engineers R. Eltzroth of ODOT District 8 and J. King of
ODOT District 7.
REFERENCES
CENTER OF L O A D A P P L I C A T I O N
S I T E PLAN
f.4PPROACH SLAB
/,
APPROX. TOP
F BEDROCK
SITE ELEVATION
Corroded Strands
127 mm
838 m m
76 mm
r 76 mm
J
y O
TL
0 ' 0 o o o o o o o ,
L
-I 203 mm
13 m m dia. strand
Damaged Zone
Missing Strand
+DIRECTION OF L A T E R A L B E N D I N G
h
E
O 5 10 15 20
Length (rn)
300
250
o 150
a
13 100
50
O
O 1O0 200 300 400 500
DEFLECTION (mm)
Fig. &Load versus deflection results - Case Study #2
A 4
4-
-
I’
0 - ----D
-
4
-
I II K’
c
C
Ea
+
7 hl
d
. L
L
.-a
W L
.- 2
n L
._ .-W
2 a 4 a a
Distance ( f t )
. I
Distance (ft)
I I .
-
- BEFORE REPAIR
DURING REPAIR
_ - - AFTER REPAIR
Inrbument Giid
Abutment Pier 1 Pier 2
N I
1
2
-C
e
E L
01
3
.-
0 P,
._
8 a a
Distance ( f t )
-
- BEFORE
._
--..._
___ DURING
REPAIR
DURING REPAIR 1
REPAIR 2
AFTER REPAIR
2 I= -
C C
E .... N E c N
L L
3
n ._
0 ._
0
a- a
-- BEFORE REPAIR
DURING REPAIR 1
.-.....- DURING REPAIR 2
- - - AFTER REPAIR
Kevwords: Blast furnace slag; C,AH,; fly ash; high alumina cements; pH;
pozzolan cements; silica fume; sodium silicate; sodium sulfate;
strätlingite;
- x-ray diffraction
461
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-159 95 IObb2949 O522649 253
462 Fu, Ding, and Beaudoin
AC1 member Yan Fu is a Ph.D. candidate in the Department of Civil Engineering, University of
Ottawa, Ottawa, Canada. His research is carried out at the Institute for Research in
Construction, National Research Council of Canada. His research interests also include high
-
alumina cement based expansive cement, delayed ettringite formation in portland cement
-
concrete and zeolite based lightweight concrete products.
Jian Ding is a graduate research associate in the Department of Civil Engineering, University of
Ottawa, Ottawa, Canada. Her research is carried out at the Institute for Research in
Construction, National Research Council of Canada. She received her first degree in chemistry
in 1983 from Fudan University in Shanghai. Before entering the MASc program, she was a
materials engineer at the Shanghai Research institute of Building Sciences. Her research
interests are concrete admixtures and additives.
James J. Beaudoin is a Principal Research Officer and Head, Materials Laboratory, Institute for
Research in Construction, National Research Council of Canada. He is the author of more than
200 research papers and two books on concrete materials. His current research focuses on
micro-fibre reinforced cement systems and the application of a. c. impedance spectroscopy in
cement and concrete science.
INTRODUCTION
High early strength and good durability in aggressive environments have encouraged
the use of high alumina cement (HAC) concrete in certain engineering applications. However,
retrograde strength associating with conversion of hexagonal phases, CAH,, or CfiH,, to
C,AH, and AH, in hydrated HAC concrete, under certain environmental conditions, has been a
major problem limiting its use. The conversion process can result in significant reduction of
strength during the service life of the concrete. The presence of a minor amount of C&SH,
(strätlingite) in HAC at later ages, was reported by Midgley 21. It was suggested that
strätlingite formation might be responsible for the strength recoveiy after conversion as it has
relatively high strength. Microsilica has been reported to favour the formation of strätlingite l31.
Granulated blast-fumace slag (ggbs) can also react with high alumina cement phases resulting
in the formation of strätlingite [4-81. Although strength reduction was limited in HAC/silica fume
products water-cured at 40 OC, hydrogamet (C,AHJ was still formed 14]. Minimization of
strength reduction in this system was attributed to strätlingite formation in preference to C$H,.
Bentsen et al reported that strätlingite could crystallize as a stable phase in the temperature
range, 20 to 70 "C. Fentiman et al [lo] suggested that in an HAC-based system the optimum
curing temperature for strätlingite formation was about 40 OC. Strätlingite has also been
observed in the form of thin hexagonal plate crystals Ilr].
EXPERIMENTAL
Materials
1. High alumina cement (HAC), Ciment Fondu, produced by Lafarge Calcium Aluminates,
Virginia, USA;
2. Silica fume, supplied by the SKW Co., Montreal, Canada:
3. ASTM Class F Fly ash;
4. Ground granulated blast-fumace slag (ggbs);
5. Sodium silicate, reagent grade;
6. Sodium sulfate, reagent grade.
The oxide compositions of HAC and siliceous materials are listed in following table:
Sample Preparation
HAC
- -S , ,.
Two cement paste samples, an HAC paste and an HAC paste containing 1% sodium
silicate, having a w/c of 0.60 were made. The cement pastes were cast in plastic bottles (25
mm in diameter) and rotated on rollers for 24 hours at 25 "C. The hardened cement paste
samples were water-cured after demoulding at 23 "C. Samples cut from the cylinders were
ground in acetone and then scanned by XRD at designated ages.
Four cement paste samples, one plain HAC, one HAC containing 4.7% sodium
sulfate, one HAC containing 30% silica fume, and one HAC containing 30% silica fume and
4.7% sodium sulfate, having a w/s (s = HAC + silica fume) of 0.60 were made. The cement
pastes were cast in plastic bottles (25 mm in diameter) and rotated on rollers for 24 hours at 38
"C. The hardened cement paste samples were water-cured at 38 OC after demoulding. The
sample cut from the cylinders was ground in acetone and then scanned by XRD at designated
ages.
HAC
..
-1
Twelve HAC slumes were made. They contained different siliceous materials in
amounts of O, 20, 50 and 100% by mass. Siliceous materials included silica fume, fly ash, and
ground granulated blast-fumace slag (ggbs). In addition O and 4.7% sodium sulfate was added
to each system. The w/s (s = HAC + siliceous material) was 2.75. The cement slurries were
placed in plastic bottles (100 mL) and rotated on rollers (to prevent hardening) at 38 "C. The
pH value and XRD analyses were carried out at designated ages.
Tests
(1) A Rigaku X-ray Diffractometer System Geigerflex DMax -B was used for X-ray
studies. Copper Ka radiation was employed. The peak heights were calculated using Rigaku
Standard Data Processingsoftware. The major peaks associated with the strätlingite (CJSHJ
phase (d=l.258 nm) and the hydrogarnet ( C Y J phase (d=2.04 nm) were monitored for each
sample. The test was designed for comparison betweenparallel samples.
(2) The pH value of the HAC/siliceOus materials slurry system was measured at 23 OC
using an Orion pMSE meter, Model 720A.
(3) Scanning electron microscopy (SEM) was used to obtain micrographs of fracture
surfaces using a Cambridge S t e r e m n S250.
RESULTS
-
HAC sodium silicatesysiem
The effect of sodium silicate on the formation of C y , in HAC paste is shown in Fig.
1. It is apparent that CAH, formation in plain HAC paste occurred mainly in the first day of
hydration. The Cfl, formation rate in the HAC paste containing 1% sodium silicate was
greatly retarded. The Cfl, peak intensity gradually increased during the first five days
hydration and then decreased. Corresponding to the decrease of C y , peak intensity, the
CASH, peak intensity increasedcontinuously with hydration time.
The strätlingite phase could be detected in HAC pastes after 1-day hydration. Its peak
intensity in plain HAC paste reached a maximum value at 3 days and then remained nearly
constant. The strlltlingite peak intensities in the HAC pastes containing sodium silicate greatly
increased after 5 days hydration. The intensity reached maximum value at about 10 days age.
- -
HAC silica fume sodium suifate system
The effect of silica fume and sodium sulfate on strãtlingite and hydrogarnet formation
in HAC paste with hydration time is shown in Fig. 2. Hydrogarnet in plain HAC paste, cured at
38 OC, mostly formed during the first day of hydration. Little change was found in hydrogarnet
peak intensities in plain HAC paste at later ages. Silica fume and sodium sulfate delayed
hydrogarnetformation for about 3 days.
-
HAC silica fume, fly ash or ggbs -sodium sulfate system
The effect of silica fume and sodium sulfate on strätlingite and hydrogarnet formation
in HAC slunies is shown in Fig. 4. Large peaks of hydrogarnet were detected by XRD in the
plain HAC paste, HAC paste containing 4.7% sodium sulfate and HAC paste containing 50%
silica fume. Strätlingite was not found in these samples. The presence of hydrogarnet was
greatly reduced in the HAC paste containing both 20% silica fume and 4.7% sodium sulfate. It
disappeared in this system when the silica fume content increased to 50%. Correspondingly,
strätlingite formed in the samples containing silica fume in combination with sodium sulfate. In
this case, the strätlingite peak intensities in the XRD spectra increased with sillca fume content.
The effect of fly ash on strätlingite and hydrogarnet formation in HAC slunies was
similar to that of silica fume (Fig. 5). The use of 50% fly ash alone appeared to have no effect
-
on the strätlingite and hydrogarnet formation. Addition of sodium sulfate in the HAC fly ash
system made the fly ash more effective in promoting strätlingite formation and preventing
hydrogarnet formation.
The effect of content of ggbs, fly ash or silica fume in HAC slunies on the pH value of
the system is shown in Fig. 7. Addition of ggbs increased the pH value of the HAC slurry from
11-73 (plain HAC) to 12.17. Addition of 20% fly ash increased the pH value of the HAC slurry
from 11.73 to 11.95. This was followed by a slight decrease with the increase of fly ash
content. Silica fume generally reduced the pH value of the HAC slurry. The pH value
decreased with silica fume content. At high content of these siliceous materials, only ggbs can
-
sustain a pH value higher than 12. Correspondingly, for the HAC silica fume, fly ash or ggbs
systems, XRD results indicated that only the samples containing ggbs presented strätlingite
formation (Fig. 6.).
The effect of siliceous materials content of HAC slunies containing 4.7% Sodium
sulfate on the pH value of the system is shown in Fig. 8. The pH value of HAC paste increased
from 11.73 to 12.75 as sodium sulfate was added. The pH value of HAC slurry containing
sodium sulfate increased from 12.75 (HAC + sodium sulfate) to 12.92 with 20% ggbs addition.
This was followed by a slight decrease as ggbs content increases from 20 to 100%. The pH
value of the HAC slurry containing sodium sulfate increased from 12.75 to 12.91 with fly ash
content up to 50%, and then decreased greatly at 100% fly ash content. Silica fume generally
reduced the pH value of the HAC slurry containing sodium sulfate. The pH value decreased
-
with silica fume content in the HAC sodium sulfate system similar to that in plain HAC slurry.
The pH values of all the samples containing sodium sulfate were high than 12. XRD analyses
indicated that strätlingite formed in all these samples.
DISCUSSION
-
Previous studies on the hydration of the HAC siliceous materials system have
indicated that the reaction between siliceous materiais and HAC results in increased formation
of strätlingite in preference to hydrogarnet. The sodium ion can attack the silica ciystal lattice
and cause the pH value of the reaction system to be elevated. It plays a significant role in the
-
mechanism of strätlingite fornation in the HAC siliceous materiais system. Silica reacts to
form silicates in a high pH environment. This reaction is accelerated in presence of alkali ions.
The alkali ions probably act as a catalyst for reactions involving siliceous materials. Dissolved
silicate anions then react with hydrated calcium aluminates ( C p , or CAH,J to produce
strätlingite. The reaction, in general, may proceed by a through solution mechanism. XRD
analysis of HACisilica fume paste samples shown in Figure 2 appear to confirm this
hypothesis. A large amount of strätlingite formed in the HAC paste only when siliceous material
was added in combination with sodium sulfate. The XRD peak intensities for strätlingite in this
sample were three times as high as in the other samples.
Test results indicate that HAC slurries with pH value higher than 12 are conducive to
increased strätlingite formation if siliceous material is added in sufficient amounts. This may be
attributed to the reduction of AH3 in the HAC paste when strätlingite forms in preference to
hydrogarnet. Note that AH3 is also one of the products of conversion reaction:
AH3, ¡.e. alumina gel, is an amorphous compound; it has a high capacity to stabilize the pH of
the system at a relatively low level. Research has shown that the pH value of the HAC H,O -
system will decrease after the conversion reaction take places and more gibbsite is formed.
The pH value of HAC slurry (mainly comprising CAH,J cured at O OC was 12.57. The same
sample was then placed at 38 'C for 24 hours. The pH value of the sample dropped to 11.95. It
-
is postulated that (i)a pH value higher than 12 in the HAC siliceous materials system is a
possible signal of preferential fornation of strätlingite; and (2) strätlingite can stably exist in an
environment with a pH value higher than 12. Once strätlingite forms from the reaction between
hexagonal calcium aluminates and activated siliceous materials, the conversion reaction to
form cubic hydrogarnet is inhibited.
CONCLUSIONS
1. Addition of sodium silicate to HAC paste appears to promote the formation of
strätlingite.
2. Strätlingite nucleation and crystallization appears to be dependent on dissdved silica
-
concentration in the HAC siliceous materials system.
3. Sodium ions appear to play a significant role in strätlingite formation in the HAC -
siliceous materials system. It is postulated that crystallization of strätlingite results
from the reaction between CAH,, or C y , and dissolved silica. Sodium ions promote
increased dissolution of silica required for strätlingite formation.
4. Addition of siliceous materials, such as silica fume, fly ash or ground granulated blast-
fumace slag, in combination with sodium sulfate in HAC paste can prevent conversion
from hexagonal calcium aluminates to cubic hydrogarnet when the paste is water
cured at 38 "C.
5. -
Addition of sodium sulfate increases the pH value of the HAC siliceous materials
hydration system to values above 12.
6. A pH value higher than 12 is a potential indicator of stratlingite formation in the HAC -
siliceous materials system.
ACKNOWLEffiMENTC
Financial support from NSERC and the Nelwork of Centers of Excellence on High
Performance Concrete is gratefully acknowledged. Thanks are extended to Mr. E. G. Quinn for
the preparation of SEM micrographs.
REFERENCES
1. Midgley, H. G., 'The mineralogy of set high alumina cement", Trans. üdt. Ceram.
Soc.,Vol. 66, p.161, (1967).
2. Midgley, H. G. and Rao, P.E., "Formation of strätlingite 2Ca0.Si02.A120,.8H,0, in
relation to the hydration of high alumina cementTem. Concr. Res., Vol. 8, p.169
(1978).
3. Bentsen, S.,Seltveit, A. and Sanderg, E, "Effect of microsilica on conversion of high
alumina cement', Calcium Aluminate Cements, Chapman and Hall, London, UK,
p.294, (1990).
4. Edmonds, R. N. and Majumdar, A. J., "The hydration of mixtures of monocalcium
aluminate and blast-fumace slag', Cem. Concr. Res., Vol. 19,779, (1989).
5. Majumdar, A. J., Singh, B. and Edmonds, R. N.,'Hydration of mixture of 'Ciment
Fondu' aluminous cement and granulated blast-fumace slag', Cem. Concr. Res., Vol.
20, 197, (1990).
6. Majumdar, A. J., Edmonds, R. N. and Singh, E., "Hydration of calcium aluminates in
presence of granulated blast-fumace slag', Calcium Aluminate Cements, Chapman
and Hall, London, UK,, p. 259, (1990).
7. Singh, B. and Majumdar, A. J., 'The hydration of calcium dialuminate and its mixtures
containing slag", Cem. Concr. Res., Vol. 22, p.1019, (1992).
a. Majumdar, A. J. and Singh, B., 'Properties of some blended high-alumna cements",
Cem. Concr. Res., Vol. 22, p.1101, (1992).
9. Rayment, D. L. and Majumder, A. J., 'Microanalysis of high-alumina cement clinker
and hydrated HAC/Slag mixtures', Cem. Concr. Res., Vol. 24, p.335, (1994).
1o. Fentiman, C. H., Rashid, S., Bayoux, J. P., Bonin, A. and Testud, M., "The effect of
curing conditions on the hydration and strength development in Fondu : Slag",
Calcium Aluminate Cements, Chapman and Hall, London, UK,, p. 272, (1990).
11. Taylor, H. F. W., 'The Chemistry of Cement', Academic Press, London, Vol. 2, p. 400,
(1964).
12. Fu, Y., Ding, J. and Beaudoin, J. J., 'Conversion-preventing additive for high alumina
cement products", U.S. Patent Application, (1994).
13. Ding, J., Fu, Y. and Beaudoin, J. J., "Study of hydration mechanisms in the high
-
alumina cement silicaisodium silicate system", submitted to Cem. Concr. Res. for
publication, (1994).
14. Ding, J., Fu, Y. and Beaudoin, J. J., 'Strätlingite formation in high alumina cement -
silica fume systems: significance of sodium ions', submitted to Cem. Concr. Res. for
publication, (1994).
Fig. 1-Effect of sodium silicate on GAH, and strätlingite (GASH,) formation in HAC paste
during hydration (wk = 0.6)
O PiaintLîC o HAct30%slicafum:
0 HAC + 4.7%sodium sulphate 0 HAC + 30% silica fume +4.7% sodium sulphaie
--o
o
o
- -
IO
Hydration Time (days)
100 100
'
IO
Hydration Time (days)
I
Fig. 2-Effect of silica fume and sodium sulfate addition on strätlingite and hydrogarnei
formation in HAC paste water-cured at 38 C for 120 days (w/s = 0.6)
Fig. 3-SEM micrographs of HAC/silica fume paste, with or without sodium sulfate, water-cured
at 38 C for 120 days (wls = 0.6)
Fig. &URD patterns for HAC/silica fume slurry, with or without sodium sulfate, cured at 38 C
for 21 days (w/s = 2.75)
Plain HAC
s . ~ , ~ ~ u . m . ~ . m . a . ~ . m .
X-ray Diffraction Degree (29)
Fig. Z X R D patterns for HAC/fly ash slurry, with or without sodium sulfate, cured ot 38 C for
21 days (w/s = 2.75)
MC + @E + 4.7%Sodium Sulphate
liAC+w%ggbS
Plain HAC
. . . . . . .
m . w . u . æ . a . 3 o . m . a . m . m .
X-ray Diffraction Degrw (a)
Fig. XRD-51 patterns for HAC/ggbs slurry, with or without sodium sulfate, cured at 38 C for 21
days (wls = 2.75)
O 20 40 60 80 100
fly ashJ
.Y
NazS04
12.4-1 cilira fume content: 4.7%
I I
O 20 40 60 ao 100
Pozzolan Content (% by mass of HAC)
473
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 = Obb2949 0 5 2 2 6 6 0 914 =
474 Wang and Dilger
INTRODUCTION
Although the ingredients used to produce HPC and NSC are basically
the same, HPC is normally made with a very low water-cement ratio (W/C)
and a high dosage of water reducers and/or superplasticizers. Since the
hydration of portland cement depends on the availability of water, the release
of heat of hydration is affected by the W/C. When the W/C is very low as in
HPC, the maximum possible degree of hydration can not be achieved, resulting
in a lower amount of total heat of hydration per unit mass of cement. Because
of the rapidly diminishing water supply, the hydration of cement will terminate
earlier in a HPC. Also, water reducers and superplasticizers normally have
some retarding effects on the setting of fresh concrete. When a high dosage of
these admixtures is used in HPC, the apparent beginning of cement hydration
after placing is delayed, that is, the so-called dormant period of the cement
hydration process is prolonged. As a result, the release of hydration heat at
early ages could be significantly different from that in NSC.
The whole process of cement hydration can be divided into four phases:
initial, dormant or induction, acceleration and deceleration ( S ) , as shown i n
Fig. 1. Depending on cement type and temperature, the initial and dormant
periods take one to a few hours, and the acceleration period takes several
hours, followed by a long-lasting deceleration period. I n practical construction
with NSC, the initial and dormant periods of cement hydration normally occur
during the mixing, agitating, transporting and placing of the concrete. That is,
after the casting of concrete into forms, the hydration of cement i n a NSC is
about to start its acceleration stage. In HPC, the dormant period is norinally
extended as a result of the high dosage of superplasticizers.
It is apparent that Eq. 2 gives the peak heat rate at the beginning of
hydration and is continuoiisly decreasing thereafter, which fails to recognize
the acceleration stage occurring in a real cement hydration (Fig. I). Therefore,
m. 1 is not an accurate description for the real hydration curve ( 3 ) .
If the heat rate is derived from h.3 , the shape is close to the real rate
of hydration curve after the dormant period (Fig. 1).
Based on the fact that the heat liberation rate depends basically only on
the amount of heat that has already been released (the degree of hydration or
the maturity of concrete) and the current temperature. Wang and Dilger (1,3)
proposed a standard curve to represent the rate of hydration heat at Z O T ,
which can be applied to different temperatures by introducing a temperature
function. For normal portland cement concrete, the peak heat rate is assumed
to be 2.2 W/kg and it occurs at the maturity of 10 hours. Therefore, the heat
rate can be described as follows:
q(M) = A +BM0.5 forM 5 10 hours (4a)
M = )H(lJdt
O
where, H is the temperature function, T the temperature and t the concrete age.
Eq. 4 describes the heat rate after the placing of NSC, and the dormant
period for NSC is assumed to occur before finishing.
Although, theoretically, a W/C of 0.4 can provide enoirgh water for the
complete hydration of cement, in reality cement can never be completely
hydrated even in water, and the ultimate degree of hydration is affected by the
W/C (9,10,11,12). Based on the data of Mills ( I i), the relationship between
U = 0.256 + 1.547(~/~)
- 1.071(~/~)~
Eq. 6 is applicable for W/C between 0.2 and 0.6, as shown in Fig. 2.
Fig. 3 presents the results of the computed total heat after 7 days of
hydration for various W/C's. The theoretical model developed here seems to
agree with the experimental results (12) well. I t can be seen froin Fig. 3 that
W/C has a significant effect on the final total heat of hydration per unit mass
of cement. Failing to recognize the effect of very low W/C will lead to a
significant overestimation of total heat of hydration in HPC.
As expected, W/C also affects the rate of cement hydration. In the early
hours of hydration, the water supply is ainple to perinit a full speed reaction
even with a very low W/C. At low W/C, as hydration progresses, the free
water is consuined rapidly and as a result the hydration rate decreases. Test
results by Taplin (9), Danielsson ( I O ) and Jonasson et al. (13) show that the
hydration rate is basically not affected by the W/C up to IO to i2 hours, which
corresponds to the time of peak heat rate on the standard hydration curve.
Thus, the rate of hydration can be treated as the saine for any W/C up to the
peak heat rate, but the hydration rate subsequently decreases faster for lower
W/C, as illustrated in Fig. 4. Therefore, Eq. 4a remains the same, and the
empirical constant C in Q. 4b is obtained by fitting the following function:
D = 23.5 -83.8(~/~)
+74.5(~/~)'
with 7 to 8% silica fume was used. Further details of the tests can be found in
reference 14. Although the cement content for the 120-MPa concrete is 15%
higher than for the 90-MPa concrete, the temperature rise in the former is not
higher than in the latter as a result of the lower W/C. The retardation effect
of high dosage of superplasticizers is also clearly shown by the measured
results.
5. The rate of cement hydration at early ages and the total amount of heat
of hydration are cement dependent. In order to systematically study the
early-age properties of hardening concrete with various cement types
and mixture proportions, an adiabatic test system is being built at the
University of Calgary to measure the heat of hydration, and the results
will be used to further expand and refine the model presented here.
ACKNOWLEDGEMENT
REFERENCES
14. Cook, W.D., Miao, B., Akin, P-C, and Mitchell, D., "Therinal
Stresses in Large High-Strength Concrete Coluiiins," AC1 Murcriu6.s Journul,
V01.89, No.1, 1992, pp. 61-68.
17. Elbadry, M.M. and Ghali, A., "User Manual for Coinputer
Prograin FETAB: Finite Element Thermal Analysis of Bridges," Research
Report, Dept. of Civil Engg., The University of Calgary, Canada, Oct. 1982,
Revised 1984.
--
I
Time
400
Tests by Smeploss and Maage
Type I cement (with & without 157. silica fume)
Accumulated heat of hydration after 7 days
350
m
Y
‘5.
1
u
-300
8
-
I
O
c
I-O
250
I I I
010 0.30 0.40 0.50 ( IO
Woter-Cement Ratio
Decrease W/C
A
B
Time
0.045
0.025 i I I I
1.00
Data from Danielcson (1960)
and RILEM (1981)
0.80 -
c Type I cement
.-
O
+
0.60 -
I"
L
O
Q>
a
b
0.40 -
2
0.20 1
0.K
100 10' 102 li
Moturity (hours)
O I I I
O 10 20 30
Time (hours)
O
Water-Cement Ratio
80
I I I 1
O 24 4a 72 96 1
Time (hours)
I
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 75 81 0662749 0522b75 3 4 5 E
SP 154-26
489
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 S P - I I 5 4 95 E 0662949 O522676 2öL
490 Hobbs
INTRODUCTION
Although it is generally assumed that alkali-silica
reaction (ASR) is a reaction between disordered silica and
the alkalis released by a portland cement, alkalis from
other sources, such as the aggregate, additions, mixing
water and sources external to the concrete can also
participate in the reaction. Of particular concern is salt
present in the mix water and aggregate, salt ingress from
seawater, airborne salt and the use of deicing salt.
When sodium chloride is present in the aggregate or
mix water, the tricalcium aluminate in portland cement may
react with the chlorides taking some of the chlorides out
of solution into the compound C,AlCaCl,.10H,O (1) and sodium
hydroxide is formed which increases the hydroxyl ion
concentration. Alternatively, the sodium chloride may
simply react with calcium hydroxide forming calcium
chloride, which accelerates the hydration of the portland
cement, and sodium hydroxide(2). Clearly the presence of
sodium in the aggregate or mix water is likely to increase
the expansion due to ASR and several investigators have
shown that this is so for mortars stored at 2OoC and
38OC(3-6).
Several investigators(7-9) have shown that the
immersion, at an early age, of mortar and small concrete
specimens in a salt solution maintained at 20 and 38OC, can
increase the expansion induced by ASR. The effect upon the
risk of expansion due to ASR of exposing mature concrete or
mortar to a sodium chloride solution is less clear. It can
be argued that the effect will depend upon the following:
1. Concentration of sodium chloride. At low
concentrations, hydroxyl ions will diffuse out of the
surface layers of the concrete at a greater rate than
they are produced by ingress of sodium ions into the
concrete. At higher concentrations, the ingress of the
sodium ion into the surface layers of the concrete will
raise the hydroxyl ion concentration and this may
increase both the level of overall expansion induced by
the reaction and the risk of expansion due to ASR.
2. Concrete member size. Since salt ingress is restricted
to the surface layers of the concrete, its effect upon
member expansion, if any, will decrease as the member
F l v ash
TEST DETAILS
Specimen arenaration
Mixtures emloved
Details relating to the various concrete mixtures
employed are given in Tables 2 and 3. Original portland
cement alkali contents of Bkglm', for cristobalite, and
7kg/m3,for the Somerset aggregate, were known from previous
tests to be sufficiently high to ensure that abnormal
expansion occurred, whilst the lower original alkali
contents of 2 . 7 and 5 . 2 kg/m3 were considered to be
borderline alkali contents for the particular aggregate
combinations employed. Some cristobalite mixtures were
also included in the programme, which contained 2 5 per cent
fly ash by mass of binder and 40 per cent slag by mass of
binder (see Table 2 ) . These mixtures were proportioned to
have alkali contents, for the particular aggregate
combination employed, close to the lower level necessary to
induce abnormal expansion.
Exposure conditions and expansion measurements
At an age of 2 4 *
1 hours, the prisms from each
concrete mixture were wrapped in a moist paper towel or
cotton cloth with the ends of the prisms exposed(l7). The
wrapped prisms were placed into polythene tubing of the
same length as the prisms, the tubing being held secure
with rubber bands(l7). Each wrapped prism was then placed
in a polythene bag(l7). The bagged prisms were placed in
sealed containers(l7) and 8 prisms from each mixture were,
stored at 2 0 f 2"C, 38 k 2°C and externally in the grounds
of the British Cement Association at Wexham Springs,
Slough, UK. Periodically the prisms were measured using a
dial gauge comparator after they had been allowed to
equilibrate overnight within their closed container to
20°C. When necessary, additional water was added to
maintain the towel in a moist state and each wrapped prism
within its sealed container returned to its chosen
temperature environment.
To establish the extent to which exposure to deicing
salts, seawater or airborne salt may effect expansion due
to-ASR, duplicate prisms from each mixture were, at each
'temperature',immersed in a l/lOM salt solution at 1 month
and a 1M salt solution at ages of 1, 3 and 6 months.
(Seawater has a salt concentration of about 0 . 5 M ) .
Periodically the length of the prisms was measured after
they had been allowed to equilibrate overnight in closed
containers to 20°C. The prisms were then returned to their
chosen temperature and salt solution environment. Details
relating to the concretes which were immersed in the salt
solutions are given in Tables 2 and 3.
RESULTS
Cristobalite
The expansion data obtained on concretes
containing cristobalite which have exhibited abnormal
expansion is shown plotted in Fig. 2 to 9 . Results
DISCUSSION
Portland cement concretes
fly ash used in this study did not prevent expansion due
to ASR in a borderline concrete exposed to a 1-M sodium
chloride solution. The effectiveness of fly ash and
slag in preventing expansion possibly depends upon their
alkali content and the percentage of portland cement
replaced.
ACKNOWLEDGEMENTS
The author is grateful to Mr M Sehmar for his
assistance in carrying out the test programme. The work
described forms part of BRITE EüRAM Project 4 0 6 2 entitled
'The residual service life of reinforced concrete
structures'. The partners involved in the project are:
British Cement Association, Crowthorne, UK; Torroja
Institute, Madrid, Spain; GEOCISA, Madrid, Spain; Lund
Institute of Technology, Lund, Sweden; Swedish Cement and
Concrete Research Institute, CBI, Stockholm, Sweden;
Cementa AB, Danderyd, Sweden.
REFERENCES
1. Mehta, P. K. "Effect of cement composition on
corrosion of reinfoxcing steel in concrete", American
Society for Testing and Materials (ASTM), Special
Technical Publication 629, 1 9 8 0 , pp 1 2 - 1 9 .
2. Kawamura, M. Private Communication, Kanazawa
University, Japan, March 1 9 8 7 .
3. Hobbs, D. W. "Mix design. Quality of mixing water.
w/c ratio. Homogeneity. Sem Le Beton and L'eau",
International des Sciences de la Construction, Saint-
Remy-les Chevrause, France, 1 8 - 2 0 June 1 9 8 5 , Conseil
Internationale de la Langue Francais, Paris 1 9 8 7 , pp
46-67.
m 00 mmmmmm
m m mmmmmm
ffl
Li
3- o o 0 0 0 0 0 0
[r)V N N NNNNNN
O 0
a-
wX ..
d rl rlrldrlrld
u u u4Juuuu
x
ffl
x
w
xf f l fxf l fxf l wxf f xl f f xl
W w w
- . .
w
m
rl
r(
r
. - .
mrimrtmri
l 4 4
d . ni
o o 0 0 0 0 0 0
o o 0 0 0 0 1 D w
P
o
- o- - . . - . .
P ****drl
ni N
0 0 0 0 0 0
t-POO**
u
m m N N ~ ~ N N
d
ffl
E
ffl
U
a
d
ffl ld
u rl
ld Li
rn k
k
rn a
z Lu
i5
-4
5
u
ffl
Li
ffl
n2
-d
L4 k
u
2O c
O
V
-d
6-
ld
IC
N
. . N *r *- m m w w Fi
4
rl Ti mm
*
$ 3
I Curing period prior
c to immersion
O
'¿j2.5 mth then 1/10Molar
c
m
Q
+
l m t h then 1 Molar
w 2
*
3mth then 1 Molar
1.5 4
6mth then 1 Molar
U
1
Maintained moist
+
O .5
01
5 10 15 20
Age - months
o .2 +
O
O 5 10 15
Age - months
Fig. 3-Influence of salt immersion on expansion-age relationship for concretes
containing tristobalite. 4 kg/m3, 20 C. Mix 4
$ 0.8
I
Curing period prior
C
to immersion
O
.-
mth then 1/10 Molar
+
(I)
c
a 0.6
X l m t h t h e n 1 Molar
w
+
3mth then 1 Molar
O .4 +
6111th then 1 Molar
&
o -2
O
O
A g e - months
$ 2
I
Curing period prior
c t o immersion
.-O
(I) mt h then 1/10 Molar
c
a 1.5 +
X
l m t h then 1 Molar
W
*
3mth then 1 Molar
1 +
6 m t h then 1 Molar
--o-
0.5
A g e - months
$ 0.7
I Curing period prior
C to immersion
O
.z 0.6
1mth then 1/10 Molar
C
<d
Q --t-
w 0.5
X
l m t h then 1 Molar
*
0.4 3 m t h then 1 Molar
+
0.3 6111th then 1 Molar
--o-
Maintained moist
o .2
(3kg/m3)
+
0.1
$ 0.7
I
Curing period prior
to immersion
.-8 0 . 6 -
03 l m t h then 1/10 Molar
C
m -t-
a
x 0.5 - l m t h then 1 Molar
w
-0-
0.4 - 3mth then 1 Molar
$ 1.0
I Curing period prior
c t o immersion
.-O
o Imth then 1/10Molar
C 0-8
m
L1
-+-
X
u] I m t h then 1 Molar
0.6
+
3 m t h then 1 Molar
--f-
6mth then 1 Molar
O *4 &
Maintained moist
-a-
o .2
O I
O 5 10 15
Age - months
O 5 10 15
Age - months
K
I
L
a
x 0.5
w 1 mth then 1 Molar
&
0.4 - 7-
wr
3mth then 1 Molar
-0-
0.3
- Gmth then 1 Molar
-0-
0.2
- Maintained moist
I , Maintained moist
0.1
O I
Age - months
$ 0.7
I
Curing p e r i o d p r i o r
c to immersion
.G 0.6
Lo 1 mth then 1/10Molar
C
(3 -c
!? 0.5
w l m t h then 1 Molar
-a-
0.4 3mth then l Molar
-e
0.3 Gmth then 1 Molar
U
Maintained moist
o *2 -E-
Maintained moist
0.1 -o-
$ 0.7
I Curing period prior
to immersion
.-6 0.6
v)
S
1 mth then 1/10 Molar
a -f-
a
x 0.5
l m t h then 1 Molar
W
*
O -4 3mth then 1 Molar
-e-
0.3 6 m t h then l Molar
--o-
Maintained moist
0.2
-#-
O.!
O
O 5 10 15 20 25 30
Age - months
SP 154-27
Svnomis: This paper presents the results of a study on the effect of curing
conditions on the air-entrained, superplasticized high-volume fly ash concrete
made with ASTM Types I and III cements and silica fume. For the four
concrete mixtures made, the total cementitious materials content was about 370
kg/m3, and the water-to-cementitious materials ratio was kept at 0.3 1IO.01, The
proportion of ASTM Class F fly ash in all the mixtures was 58% by weight of
the (cement + fly ash) content. Two mixtures incorporated silica fume at the
dosage rate of 8.5% of the total cementitious materials content.
The use of ASTM Type III cement instead of Type I increased the early-
age strength significantly without affecting the long-term strength development
under moist curing conditions. Under air-curing conditions, the concretes
incorporating the Type III cement achieved significantly higher strengths at
every test age up to one year. The use of silica fume resulted in only marginal
improvement in the strength properties under the two curing regimes.
V. Sivasundaram was a Research Engineer with Canada Centre for Mineral and
Energy Technology (CANMET), Natural Resources Canada. She received her
B.Sc. in Civil Engineering from the University of Sri Lanka and an MS degree
from MIT. She worked in the area of durability of concrete incorporating
supplementary cementing materials. V. Sivasundaram left CANMET in
November 1993.
INTRODUCTION
SCOPE
Four high-volume fly ash concrete mixtures involving eight batches were
made in this investigation. ASTM Type I and Type III portland cements were
used in two mixtures each; within each mixture series, one mixture was made
incorporating 8.5% silica fume. The water-to-cementitious materials ratio was
kept at 0.3110.01 for all the concrete mixtures. The concretes were air-
entrained, and the properties of the fresh concretes were determined. The
autogenous-temperature rise was monitored on 152x305-mm concrete cylinders.
A number of specimens were cast, subjected to moist-curing and laboratory air-
curing regimes, and tested at various ages for compressive strength and
resistance to chloride-ion penetration (RCP). Tests for the flexural strength and
resistance to freezing and thawing cycling were also performed on the moist-
cured concrete prisms.
CONCRETE MATERIALS
Cement
ASTM Type I, normal portland cement and ASTM Type III, high-early
strength portland cement were used. The physical properties and chemical
analyses of the cements are given in Table 1.
Flv ash
ASTM Class F (low-calcium) fly ash from Atlantic Canada was used. The
physical properties and chemical analyses of the fly ash are given in Table 1.
Silica fume
ApgrePates
The coarse aggregate was crushed limestone with nominal maximum size
of 19.0 mm, and the fine aggregate was local natural sand. Both aggregates
were separated into different size fractions, and then recombined to a specified
uniform grading. The grading and physical properties of the coarse and fine
aggregates are given in Tables 2 and 3.
Superplasticizer
Air-entraining admixture
MIXTURE PROPORTIONS
Four concrete mixtures involving eight 0.1 m3 batches were made in this
program. Mixture 1 was a high-volume fly ash concrete made with ASTM Type
I cement (155 kg/m3) and ASTM Class F fly ash (215 kg/m3) at 58% by mass
of the total cementitious materials content. In mixture 2, in addition to the
above Type I cement and fly ash, silica fume was incorporated at 8.5% by
weight of the total cementitious material. The proportion of the fly ash in this
concrete mixture was 58% of cement+fly ash. In effect, the total cementitious
material in mixture 2 was the same as that in mixture 1. Mixtures 3 and 4 were
repeats of mixtures 1 and 2, respectively, except that ASTM Type III cement
was substituted for the Type I cement. The water-to-cementitious materials ratio
was kept at 0.31k0.01 for all the concretes. The concretes were air-entrained
with a target air content of 5kl%, and the superplasticizer was used to give
increased slumps. The concrete mixtures were made in a laboratory counter
current mixer with fly ash and silica fume added as separate ingredients.
The properties of the freshly mixed concrete, Le., temperature, slump, unit
weight, and air content are given in Table 5.
Two 0.1 m3 batches of concrete were made from each mixture in order to
cast the required number of specimens.
Batch A
The cylindrical and prismatic specimens were cast in two layers; in the
case of the 152x305-mm cylinders, an internal vibrator was used for compaction;
the rest of the specimens were compacted on a vibrating table.
After casting, all the moulded specimens were covered with plastic sheets
and a water- saturated burlap, and were left in the casting room for 24 h.
Afterwards, they were demoulded and transferred to the moist-curing room at
100% relative humidity for curing. After three days of moist-curing, eight
152x305-mm cylinders and four 102x203-mm cylinders from Batch B were
transferred to a room at ambient temperature (22 to 30°C) and at about 40 to
50% R.H. The rest of the specimens were moist-cured until needed for testing.
TESTING OF SPECIMENS
subjected to the RCP testing according to the standard, and the total electrical
charge passed through the specimens, in coulombs, was used as an indicator of
the resistance of the concrete to the penetration of chloride ions.
TEST RESULTS
Concrete Mixtures
S e t t h Time
For high-volume fly ash concrete made with ASTM Type III cement
(mixture 3), and silica fume high-volume fly ash concrete made with Type I
cement (mixture 2), the setting times were reduced substantially in comparison
with that of the high-volume fly ash concrete made with Type I cement (mixture
1). The final time-of-set for concrete mixtures 1, 2, and 3 were 10 h:45 min.,
7 h:40 min., and 6 h:40 min., respectively. Because of the high early hydration
of the cement, the final time-of-set of the high-volume fly ash concrete made
with Type III cement was shorter than that of the others. When silica fume was
incorporated in the concrete made with Type III cement (mixture 4), the setting
times were reduced further, resulting in times of initial and final set of 3 h:25
min. and 5 h:25 min.
Autorrenous-TemDerature Rise
The use of silica fume in both high-volume fly ash concretes made with
ASTM Type I and III cements did not contribute to any major changes in the
autogenous-temperature rise. In general, the maximum temperature rise in all
four concretes was found to be much lower than that of a comparable portland
cement concrete without silica fume.
Strenhh Properties
The compressive strength results are shown in Table 7, and are illustrated
in Figs. 2 and 3. The 1-day strengths of the high-volume fly ash concrete made
with ASTM Type I cement with and without silica fume (concretes 1 and 2),
were 6.5 and 6.9 MPa; for the two concretes with Type III cement (concretes
3 and 4), the 1-day strengths were 11.6 and 12.0 MPa, respectively.
Between the ages of 3 and 91 days, the use of silica fume resulted in only
marginal improvement in the strengths of the moist-cured, high-volume fly ash
concrete made with Type I and Type III cements. The incorporation of 8.5%
silica fume in the high-volume fly ash concrete did not yield any significant
benefits in the long-term strength development of the moist-cured, high-volume
fly ash concretes made with both Types I and III cements. Although the
strength development rate was not the same for these concretes, all four
concretes reached about 55 MPa at 365 days under moist-curing conditions.
Under the above curing conditions, ASTM Type III cement concretes
showed strengths 7 to 10 MPa higher than that of the corresponding ASTM
Type I cement concretes at each test age. Also, following one year of air
curing, both concretes made with the Type I cement reached only about 55 to
60% of the corresponding strength for the moist-cured concrete; the two
concretes made with the Type III cement reached about 70 to 74% of the
corresponding strength for the moist-cured concrete. Hence, the use of ASTM
Type III cement in high-volume fly ash concrete appears to be beneficial both
for strength development at early ages, and in terms of strength development
when prolonged moist curing cannot be provided.
When high-volume fly ash concrete is made using ASTM Type III cement,
the concrete attains a larger proportion of its potential strength at the early ages.
It is possible that the higher strengths developed in the above concrete after
three days of moist-curing results in a dense pore structure, which prevents
excessive drying with time; this is probably not so in the case of high-volume
fly ash concretes made with the Type I cement. Further research is needed in
this area.
As in the case of moist-cured concretes, the use of silica fume showed only
marginal strength improvement in the air-cured concretes made with ASTM
Type I and III cements. The reason for this could be that the high-volume fly
ash concrete mixtures were lean in cement, with only 140 kg/m3 of portland
cement and this inhibited the pozzolanic reaction of silica fume due to the
The 28-day flexural strength of the moist-cured prisms are also given in
Table 7. The available data show slight improvement in the flexural strength of
high-volume fly ash concrete when Type III cement was used instead of Type
I cement. The improvement was also noticed when silica fume was incorporated
in the high-volume fly ash concrete.
When ASTM Type I cement was replaced by Type III cement in the high-
volume fly ash concrete without silica fume, the resistance to chloride-ion
penetration increased markedly for air-cured specimens as compared with moist-
cured specimens. This pattern was also observed in the strength development.
Under moist-curing conditions, the RCP results for the ASTM Type III cement
concrete (742 coulombs at 28 days and 301 coulombs at 91 days) were about
the same as that of the ASTM Type I cement concrete. The RCP values of the
ASTM Type III cement concrete subjected to air curing (2277 coulombs and
2833 coulombs at 28 and 91 days, respectively) were higher when compared
with that of the moist-cured Type III cement concrete. Nevertheless, the results
obtained for the air-cured Type III cement concrete were about 50% of the
values obtained for the air-cured Type I cement concretes.
cement concretes incorporating silica fume, the RCP results for moist-curing
conditions were 172 and 200 coulombs at 28 days and 103 and 87 coulombs at
91 days, respectively; these low values are associated with negligible chloride-
ion penetration in the concretes. Even with the air curing of Type I and III
cement concrete incorporating silica fume, the RCP results at 28 and 91 days
were significantly lower than those obtained for the high-volume fly ash
concretes without silica fume. The RCP values for the Type I and III cement
concretes with silica fume for air-curing conditions were 843 and 622 coulombs
at 28 days and 1010 and 969 coulombs at 91 days, respectively.
There is some concern, however, that the RCP values obtained for air-
curing conditions are slightly higher at 91 days when compared with that
obtained at 28 days. This may be due to the experimental variances. These
tests were not carried out on one-year old specimens to verify this observation.
Further research is necessary in this area.
The results of the freezing and thawing tests given in Table 9 indicate that
all four concretes performed well after 300 cycles of freezing and thawing, with
durability factors above 95%. The weight loss due to surface scaling during
testing appeared to be slightly higher in the Type I cement concretes, both with
and without silica fume, when compared with those of the two Type III cement
concretes. Surface scaling during rapid freezing and thawing cycling has also
been observed in previous CANMET investigations on high-volume fly ash
concretes made with Type I cement (5,7,15). The residual flexural strengths of
the test specimens after 300 cycles of freezing and thawing were in the range
of 50 to 80% of the reference moist-cured specimens. The substitution of Type
III cement in the concrete or the use of silica fume did not appear to make any
significant difference in the freezing and thawing durability of high-volume fly
ash concrete.
CONCLUSIONS
1. When ASTM Type III cement was used instead of Type I cement in the
high-volume fly ash concrete, the superplasticizer and air-entraining
admixture dosage requirements increased slightly. These were increased
further by the incorporation of silica fume in the concretes.
2. The setting times of the high-volume fly ash concrete were reduced by the
use of silica fume in the concrete as well as by the replacement of ASTM
Type I cement by Type III.
4. When the concretes were subjected to air curing after three days of moist
curing, those made with the Type III cement achieved higher strengths at
all ages as compared with the concretes made with the Type I cement.
Therefore, the use of ASTM Type III cement in high-volume fly ash
concrete appears to be beneficial when prolonged moist curing is not
available; this is in addition to its high early-age strength development.
The use of silica fume in the air-cured concretes made with ASTM Type
I and III cements showed only marginal strength improvement.
5. When the high-volume fly ash concrete made with the ASTM Type I and
III cements was subjected to air curing after three days of moist curing, a
drop in the resistance to chloride-ion penetration was observed in both
concretes in comparison with those of the respective moist-cured concretes.
Nevertheless, the RCP values obtained for the concretes made with Type
III cement were about half of those obtained for the concrete made with
Type I cement indicating increased resistance to chloride-ion penetration.
REFERENCES I
1. Sivasundaram, V. "Thermal Crack Control of Mass Concrete"; MSL
Division Report MSL 86-96 (IR), Energy, Mines and Resources Canada,
Ottawa, Canada, 1986, 32 pp.
15. Bilodeau, A., Carette, G.G. and Malhotra, V.M. "Investigations of High-
Volume Fly Ash Concrete: Final Report"; MSL Division Report MSL 93-
30 (CR), Energy, Mines and Resources Canada, Ottawa, 1993, 95 pp.
ACKNOWLEDGEMENT
This paper has been edited for consistency by P.R.A. Andrews, Resource
Utilization Laboratory, CANMET.
TABLE 1 - PHYSICAL PROPERTIES AND CHEMICAL ANALYSES OF CEMENTS, FLY ASH, AND
SILICA FUME
Chemical Anaivsis. %
Silicon dioxide (Sioz) 21.95 21.21 42.2 93.6
Aluminum oxide (Alzo,) 4.13 5.19 21.6 0.06
Ferric oxide (Feo,) 2.88 2.35 27.6 0.45
Calcium oxide (Cao) 61.73 63.42 1.87 0.50
Free calcium oxide 0.48 __ -- __
Magnesium oxide (MgO) 3.68 3.03 0.92 0.67
Sodium oxide (Na,O) 0.25 1.50 0.66 O. 16
Potassium oxide &O) 0.44 0.60 2.55 0.85
Phosphorous oxide (P205) 0.07 0.08 -- 0.09
Titanium oxide (Ti03 0.20 0.26 __ 0.01
Sulphur trioxide (SO3) 2.81 3.49 1.10 O. 32
Loss on ignition (LOI) 1.41 2.34 1.85 2.26
Borne Potential Comwund. k
Tricalcium silicate C,S 42.8
Dicalcium silicate C$ 30.7
Tricalcium aluminate C,A 6.8
Tetracalcium aluminoferrite C,AF 8.8
'Crushed limestone
"Natural sand
I I I
in IC
t-
w r
U
CI h
rr) e-
CI CI
c? c?
O O
I -
1
Concrete -
W' Temp., Slump, Unit Weight, Ait
TYpe CM "C mm kgim' Content, %
Type I cement 0.31 26 150 2370 5.5
HVFC
without
silica fume 0.31 26 150 2350 5.6
Type I cement 0.31 26 100 2345 5.6
HVFC
with
silica fume 0.31 28 65 2370 5.5
Type III cement 0.31 28 65 2385 5.5
HVFC
without
silica fume 1 0.31 I 28 I 125 I 2410 I, 4.9
Type III cement 0.32 30 115 2360 5.3
4A HVFC
with
4B silica fume 0.32 30 90 2360 5.5
-- -- 29.7
-
29.8 18.2 26.5 30.9
-- --
-
36.8 26.7 36.7 40.8 392
t
...
...
rn
O
d
...
m
rn v)
m
O
O
O
*O
+o
o
+o
o
+o
m O
m
[I
+o m
+o m
o 0
o m
o m
0
o m v)
N
o m
I
I
o m
o m
o cb
O
m O
N
a I
v)
3
s?
v)
60 I
O '
1 10 1O0 1O00
Age, days
Fig. 3-Strength developed under both curing conditions at 28 days and 1 year
SP 154-28
531
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 Sp-354 95 IOb62949 0522738 5b0
532 Mehta and Folliard
INTRODUCTION
In 1992, a state-of-the-art report entitled, "Rice Husk Ash - A Unique
Supplementary Cementing Material," was presented by the first author and
published in the CANMET publication, Advances in Concrete Technology (i).
The report focused on the physical-chemical characteristics of rice-husk ash
(RHA) and reviewed the research findings from several countries on the influence
of burning conditions on strength-contribution potential of RHA. Not much
information was available on the effects of RHA on concrete durability.
In 1990, comprehensive research was undertaken at the University of
California at Berkeley to study durability aspects of concrete and mortar containing
RHA. The results of several of these studies are presented here, including
investigations on the resistance of cement products containing RHA to
hydrochloric-acid attack, sulfate attack, alkali-silica expansion, and frost action.
In addition to the work at UC Berkeley, the results of a study on freezing and
thawing of RHA concrete in saline environment, performed at Lund Institute of
Technology, Sweden, are also reviewed.
The rice-husk ash used in the various research studies reported here
consisted essentially of amorphous silica (90% silica content), with a surface area
of 50-60 m*/g as measured by BET nitrogen adsorption. The average size of
RHA pyticles was 8 pm and the carbon content was 5% as measured by loss on
ignition, unless otherwise noted. Ali of the ash samples were procured from
commercial sources within the United States.
152.4-mm) cylinders from each of the mixtures, after 7 days of moist-curing (20
OC, 100% relative humidity), was determined in the saturated, surface-dry (SSD)
condition, before immersion in the HCl solution. The solutions were kept in
constant motion through the use of magnetic stirrers.
On a weekly basis, the test cylinders were lightly scrubbed with a wire
brush and the mass determined in a SSD condition. The HCl solution was also
changed on a weekly basis before resubmerging the cylinders. Fig. I shows the
mass loss for the cylinders. A review of published literature shows that in this
test, most authors consider a mass loss of 25% as the failure criterion, which is
shown by the horizontal line in Fig. 1
The concrete containing RHA was found to be much more resistant to the
HCl solution than the concrete without RHA, as shown by the reduced mass loss.
The concrete with RHA had not reached the failure limit at the conclusion of the
28-week test program, whereas the concrete without RHA reached the failure limit
in 10 weeks. Fig. 2 shows a photograph of the concrete cylinders after 28 weeks
of immersion in the HCl solution.
The bars, 6 specimens for each mixture, were measured for initial length
and submersed in a sodium sulfate solution (50 g Na2S04per litre of water).
Average values of length-change measurements are shown in Fig. 3. At the end of
* Standard cement chemistry abbreviations are used : C=CaO, S=SiO2,
A=Al2O3, F=Fe203, H=H20.
the 16-week test, there was a 82% reduction in expansion for the mortar
containing RHA, compared to the reference mixture without RHA. Further, the
expansion of the RHA mortar had virtually stabilized by the end of the test,
whereas the mortar without RHA showed a tendency to expand further at the end
of the 16-week test period.
ALKALI-SILICA EXPANSION
A portland-RHA cement was made by blending together RHA with a high-
alkali portland cement (1 .O% alkalies as NaZO) in a laboratory ball mill. The
blended cement contained 90% high-alkali portland cement and 10% RHA. The
effectiveness of RHA in reducing expansion due to alkali-silica reaction was
investigated according to ASTM C 441. Crushed Pyrex glass was used as the
standard reactive aggregate. A standard consistency, characterized by a flow value
of 110 t 5 according to ASTM C 109, was maintained for each of the mixtures.
Table 3 shows the mixture proportions for the two mortars studied.
Standard bars, 1-in by 1-in by 11 %-in (25.4-mm by 25.4-mm by 285.8-mm),
were moist-cured for 24 hours, then sealed in special containers meeting ASTM C
227, according to which the specimens are held at a temperature of 38 OC for 14
days. The results of length change measurements, shown in Fig. 4, are the
average of three beams for each mixture. There was a 95% reduction in expansion
for the mortar containing RHA, which far exceeds the 75% minimum criterion
proposed in ASTM C 441 for assessing the effectiveness of pozzolans in reducing
the alkali-silica expansion.
FREEZING AND THAWING
As part of a comprehensive research program focusing on the frost
resistance of RHA concrete, a series of high-performance concretes were designed
and tested according to ASTM C 666, Procedure A. Only some of the results
from this study are discussed next, with emphasis on non-air-entrained concrete
having water-binder ratios of 0.30 and 0.35. For each water-binder ratio, three
concrete mixtures were tested; one reference concrete (REF30 or REF35), one
concrete containing silica fume (SF30 or SF35), and one concrete containing rice-
husk ash (RHA30 or RHA35).
The silica fume used in the study had a surface area of 22 mVg. The RHA
had a carbon content of 15% and was obtained from a commercial source in the
United States. Among the RHA commercially available in the United States, this
RHA had the highest carbon content. Previous work at UC Berkeley has found
that the mechanical properties of RHA concrete are not adversely affected by RHA
with a high carbon content.
Table 4 shows the mixture proportions used in the study. For the mixtures
containing silica fume or RHA, the admixture was used as a 15% by mass
substitution for portland cement. The coarse aggregate used was a crushed
limestone with 12.5 mm MSA, and the fine aggregate was a natural sand with a
fineness modulus of 3.1. A naphthalene sulfonate formaldehyde high-range
water-reducing admixture was used in each mixture to achieve a slump of 150-200
mm.
Fig. 5 shows the durability factor (DF) for each of the mixtures, with the
values shown being the average of three prisms. The two concretes containing
RHA were the only mixtures to endure 300 cycles without reaching the failure
criterion (a durability factor less than 60). However, both the silica-fume
concretes exhibited poor frost resistance, with durability factors less than 10.
Freezing-and-thawing cycles were terminated for SF30 prisms after only 101
cycles, and for SF35 prisms after 67 cycles. The reference concretes also fared
poorly in the testing, although they did perform better than the silica-fume prisms.
REF30 yielded a durability factor of 43 after 214 cycles, and REF35 had a
durability factor of 28 after 144 cycles.
The generally poor performance of the silica fume concretes could possibly
be attributed to essentially an "impermeable cemenvsilica fume matrix with very
small discontinuous pores that, in turn, adversely affects the movement of water,"
as proposed by Malhotra, et. al (3). In their investigation, non-air-entrained silica
fume concrete showed extremely low durability factors, especially for higher
levels of silica-fume replacement. The present study used a 15% (by mass)
substitution of cement by silica fume. The chloride permeability values of the
silica fume mixtures were found to be very low, with less than 400 coulombs
passed for each concrete. It is possible that the impermeability imparted by the use
of silica fume restricted the movement of water within the matrix, thus developing
high hydraulic pressure, even though the amount of freezable water in the two
mixtures was very small.
The frost resistance of RHA concrete has not been documented in past
literature, so the present study is perhaps the first to report the behavior. The
reasons for the superior performance of RHA concrete in freezing-and-thawing
testing are not fully understood. Perhaps, the unique form of RHA particles, with
their highly microporous structure, results in the formation of pozzolanic reaction
products that have a structure similar to RHA particles. This microporous
structure of the reaction products may give rapid chloride permeability values that
are higher than those for the corresponding silica-fume concrete mixtures. On the
other hand, the microporosity may be sufficient to provide escape paths for the
freezing water inside the concrete, thus reducing internal stresses. Any unreacted
RHA particles obviously would have contributed to this effect. It was confirmed
that the chloride permeability values for the RHA concretes were somewhat higher
than those registered by the silica-fume concretes. More research is needed in this
area to fully understand the reasons behind the improved frost durability of
concrete containing RHA.
CONCLUDING REMARKS
From the results presented in this report, it is concluded that the use of
R H A in cement products greatly improved their durability characteristics. It is
shown that R H A helped to substantially reduce the mass loss of concrete exposed
to a hydrochloric acid solution, and greatly decreased the expansion due to sulfate
attack and alkali-silica reaction. Non-air-entrained concrete containing R H A
exhibited significantly higher frost resistance than similar mixtures containing silica
fume when tested according to ASTM C 666, Procedure A. In addition, the dura-
bility of R H A concretes subjected to freezing and thawing in saline environment
is reported to be considerably higher than control mixtures without RHA.
REFERENCES
1. Mehta, P.K., "Rice Husk Ash - A Unique Supplementary Cementing Material,"
Advances in Concrete Technology, Edited by V.M. Malhotra, CANMET
Publication, 1992, pp. 407-432.
2. Nilsen, U., Sandberg, P., and Folliard, K., "Influence of Mineral Admixtures on
the Transition Zone in Concrete," Interfaces in Cementitious Composites,
Proceedings of the RILEM International Conference, Toulouse, France, 1992,
pp. 65-70.
3. Malhotra, V.M., Painter, K.A., and Bilodeau, A., "Mechanical Properties and
Freezing and Thawing Resistance of High-Strength Concrete Incorporating
Silica Fume," Cement, Concrete, and Aggregates, CCAGDP, Vol. 9, No. 2,
Winter 1987, pp. 65-79.
Mixture
#
28-day I 28-day
compressive chloride
I Air Content (%) Specific Spacing
Surface,a Factor,
strength permeability
Fresh Hardened
(ma) (cou'ombs) Concrete Concrete (m-9 (w)
REF30 77.6 2730 1.9 1.4
SF30 96.5 250 0.9 0.8 14.1 1
MA30 81.2 520 1.4 1.1 14.37
REF35 69.6 3680 1.8 2.1 9.76
SF35
-
82.7 370 0.9 1.4 11.21 774
MA35 78.5 900 0.6 1.o 9.53 1152
1O0
90
cw
O 40
i
Y
E
8 30 i ............................................. j
o-*
..........................................
..-e I
i
e.D
.................
*-i,
LI
8 20 -. ....... ....... ....................... i........................ / ...................... 1
10 -. ....... ..............
I ~ I , l , , , l , , , i , , , i , , , ~ , , / -
O
Fig. 2-Photograph showing concrete cylinders after 28 weeks immersion i n 1 percent HCI
solution. The cylinder on-the left is concrete without RHA, and the cylinder on the right contains
RHA (13 percent by weight of portland cement)
0.20
.............................................................................. +................
I
‘i
............................... i
:.............................................
82% Reduction ...............
‘ i in Expansion
i #
Y
8
E
Q)
pc 0.05 .............................. ................
c
0.00
O 4 8 12 16 20
Time (weeks)
0.50
0.40
0.30
0.20
0.10
0.00
Mortar without RHA Mortar with RHA
100
80 ..........................................................................................................................................
W
g8 60 ............................................... ..................................................
&
.-
,h
a
40 ........................... ..................................................
s
20 ........................... ..........................
fi
O
Fig. 5-Durability factors of concrete subjected to 300 cycles of freezing and thawing according to
ASTM C 666, Procedure A
SP 154-29
543
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 M 0662949 0522729 346
544 Langley, Gilmour, and Tromposch
INTRODUCTION
The main bridge will consist of 44 spans of 250 meters each. The
piers for the main span will be cast on land in the Prince Edward Island
staging facility, moved to a jetty by an innovative hydraulic slider
technique, and then transported to the site by a marine crane vessel.
These piers will be founded on bedrock in water depth up to 35 meters.
The main span girders will be match cast and post tensioned at the .
staging facility and then transported to the site. The spans will be
constructed using two types of components; a cantilever unit and a drop
in span. The completed cantilever units weigh upwards of 8000 tonnes.
The main span units will require a total of 183 individual precast
sections.
The main span cantilever girders are 190 meters long that
cantilever 95 meters in each direction over the piers. The cantilever
girder depth varies from 14 metres at the piers to 4.5 metres at the
cantilever tip.
A typical pier foundation, pier shaft, main span girder and drop-in
section is shown in Fig. 2 . Some of the design challenges are discussed
next.
DURABILITY OF CONCRETE
ProDedies of Materiais
Properties of Concrete
Class A Concrete
The results of the creep tests up to January 31, 1995, and drying
shrinkage are shown in Fig. 5 and Fig. 6 respectively. The results of
other long-term tests are summarized in Table 12.
Class C Concrete
Class F Concrete
CONCRETE CURING
QUALITY CONTROL
to occur, and to set safe limits on the concrete temperature at the time of
placement.
PIacement Temperature
protection to maintain steel passivity. Also, low C3A content lowers the
heat of hydration. The low alkali content (maximum of 2.7 kg/m3
contribution of cement) will ensure a greatly reduced potential for alkali
aggregate reaction even though non-reactive aggregates (confirmed by
extensive testing) have been chosen.
CONCLUSION
REFERENCES
7. Berke, N.S, Scali, J.M. Regan, J.C., and Shen, D.F., "Long-
Term Corrosion Resistance of Steel in Silica Fume and/or Fly
Ash Containing Concretes", Proceedings, Durability of Concrete:
Second CANMETIACI International Conference, Montreal,
Canada, 1991, Publication SP-126, American Concrete Institute,
Detroit, Michigan, 1991, pp 393-422. Editor V. Mohan
Malhotra.
Compound Percent
Si02 27.1
A1*0, 4.7
%O, 3.0
Ca0 59.5
MgO 1.1
so3 2.5
LOI 1.o
Na,O (equivalent) 0.45
GA 7.4
Property Results
, Fineness 45pm 89.5% Passing
Specific Surface 443 m2/kg
Setting Time (initial) 118 minutes
Autoclave Expansion 0%
Compressive Strength
TABLE 7 - CHARACTERS
I TC
I S OF CLASS A CONCRETE
NEW BRUNSWICK
NOVA SCOTIA
CONCRETE GIRDER
C W S 'F CONCRETE
A WE AGCREGX,~E
o COARSE AGGREGATE
Fig. &Accelerated mortar bar test for coarse and fine aggregate (CSA A 23.2-25A)
q
-
x
. 0.03
1:iiE. o*-i
Fig. A o n c r e t e prism test for coarse and fine aggregate ((SA A23.2-14A)
o
Age Under Load (days)
600 7 A 4
MOIST CURED FOR 28 DAYS @ 2 3 C
a THEN AT 5 0 % RH Q 2 3 C
b 500
e
x MOIST CURED FOR 28 DAYS O 23’C
THEN AT 75% RH O 23C
.-
E 400
O i i i i i i i r
O 1 O0 200 300
600 -s
o , , , , , , , , , , , , , , , ,
O 1w 200 300
SP 154-30
Corrosion test data have been used to show the validity of this analytical
procedure. The data show that the procedure can be used to rank the effectiveness
of corrosion-inhibiting admixtures in delaying the time-to-corrosion of steel in
concrete, and that passive and passive-active inhibitor systems can delay the
onset of corrosion longer than active inhibitor systems. In combination with
corrosion rate data, the relative overall effectiveness of corrosion-inhibiting
admixtures can be determined.
565
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 0662947 0 5 2 2 7 5 1 872 =
566 Nmai
INTRODUCTION
ACTIVE SYSTEMS
It is generally accepted that chloride ions react with ferrous ions to form a soluble
complex which upon reaction with hydroxyl ions lead to the formation of ferrous
hydroxide, Fe(OH)2, a form of rust. The chloride ions are then released back
into solution and become available for further reaction with ferrous ions. A
typical reaction between ferrous and chloride ions in concrete is as given in Eq. 2
c101.
However, half-cell potential and chloride content åata from concrete deck
specimens subjected to daily salting indicated that chloride-to-nitrite ratios
greater than 1.6 to 1 by weight will result in corrosion [9]. Nondestructive data
collected over 7 years and visual survey of test slabs in a study conducted by the
Federal Highway Administration (FHWA) indicated effective performance of the
calcium nitrite inhibitor up to a chloride-to-nitrite ratio of 0.9 to 1 by weight [5].
Admixed chlorides were used in this study.
Because of the competing reaction with chloride ions, the level of protection
provided by active systems is usually a function of the amount of chloride ions
expected at the level of the reinforcing steel over the period for which protection
is required. Therefore, increasing amounts of an active system will typically be
required to protect against higher levels of chloride ions. Published chloride
protection levels for a 30 percent by mass solution of calcium nitrite [i i] are as
shown in Table 1. The data show that the higher the dosage of calcium nitrite ,
used, the higher the chloride ion threshold for the protected steel, and
conceivably the longer the delay in time-to-corrosion of the treated concrete.
The effective chloride diffusion coefficient for concrete treated with a passive
system inhibitor will often be considerably lower than that for its companion
untreated concrete. For example, after two years of cyclic ponding with chloride
solution, concrete containing 350 kg/m3 (590 lb/yd3) of cement and a 15 percent
silica fume addition by mass of cement was reported to have an effective chloride
diffusion coefficient of 0.7 x 10-8 cm2/s, compared to 11.0 x 10-8 cm2/s for the
companion plain portland cement concrete [16]. The concretes had a water-
cement ratio of 0.48. At a water-cement ratio of 0.38, effective chloride
diffusion coefficients of 0.2 x crn% and 2.0 x 10-8 crn% were reported for
the 15 percent silica fume concrete and its companion plain portland cement
concrete, respectively [ 161.
Prior to the initiation of corrosion, comparisons between two active systems can
be based solely on their effect on the chloride ion threshold at the steel surface,
which can be determined relatively quickly from solution testing. However,
comparisons between active and passive or passive-active systems are not as
straightforward and cannot be based solely on chloride ion threshold because of
the inherent differences in protection mechanisms described earlier.
This means that an active system with a significantly higher chloride ion
threshold may not necessarily delay corrosion initiation longer than a passive or
passive-active system with a lower chloride ion threshold. Likewise,
comparisons cannot be based solely on degree of chloride screening because of
Note that if [Cl'],i@t is greater than [Cl']ref@t, as can occur in calcium nitrite-
treated concretes because of the accelerating effect of this admixture, the
screening efficiency will be negative.
Using Eq. 5 , a family of curves that relate screening efficiency and chloride ion
threshold to the chloride ion content in untreated concrete can be generated for
each corrosion-inhibiting admixture. This implies that for a given level of
screening efficiency and chloride ion threshold, an "equivalent chloride ion
threshold" in the untreated concrete can be determined. This provides a common
basis for direct comparison of corrosion-inhibiting admixtures. The following
examples will be used to illustrate the use of curves that can be developed by
using Eq. 5.
Fig. 1 shows a chloride ion content versus screening efficiency curve which was
generated assuming a chloride ion threshold of 0.9 kg/m3 (1.5 lb/yd3). This
curve is applicable to passive systems such as silica fume concrete, because it has
a chloride ion threshold similar to that for untreated concrete, that is, in the range
of 0.7 to 0.9 k@m3 (1.2 to 1.5 ib/yd3). The curve is also applicable to
dampproofing admixtures that do not increase the chloride ion threshold of
reinforcing steel.
vertical axis directly against the chloride protection levels reported in the
literature by the manufacturer. As will be shown later, these chloride protection
levels may not be completely accurate.
A pair of chloride ion content versus screening efficiency curves for a passive-
active system such as the water-based organic corrosion-inhibiting admixture are
shown in Fig. 3. The curves were generated using chloride ion threshold values
of 2.4 and 3.0 kg/m3 (4 and 5 lb/yd3), respectively. Limited time-to-corrosion
test data indicate that the chloride ion threshold for reinforcing steel in concrete
treated with 5 Um3 (1 gal/yd3) of the water-based organic corrosion inhibitor is
probably between these two values [7].
The validity of the analytical procedure presented in the sections above was
verified by comparing the predicted time-to-corrosion performance with that
obtained from accelerated corrosion evaluations. Two comparisons are
presented. The first is between a passive system (silica fume concrete) and an
active system. The second is between a passive-active system (water-based
organic corrosion inhibitor) and an active system. In both cases, the active
system was a calcium nitrite-based corrosion inhibitor.
The data shown in Table 2 were extracted from Reference 18 and will be used for ’
the comparison between concrete containing calcium nitrite (active system) and
silica fume concrete (passive system). The data were obtained after 5 years of
corrosion testing of lollipop and minideck specimens. At a water-cement ratio of
0.48, corrosion data for minideck specimens with 35 mm (1.38 in.) of cover
indicated that the silica fume concretes were performing better than concretes
containing 10 and 20 Um3 (2 and 4 gaVyd3) of calcium nitrite.
The corrosion rate data for the lollipop specimens are also shown in Fig. 4-6.
These figures show that at the lower water-cement ratios of 0.38 and 0.43, the
concretes containing 3.75 percent silica fume were performing better than those
containing 10 Um3 (2 gaVyd3) of the calcium nitrite inhibitor. The data also
suggest that, at a water-cement ratio of 0.43, the concrete containing 7.5 percent
silica fume were performing better than that containing 20 Um3 (4 gaVyd3) of
the calcium nitrite inhibitor. Note that 1/Rp values greater than 25 pS/cm2 have
been reported to be an indication of severe corrosion, whereas values less than 15
pS/cm2indicate passive corrosion conditions [ 181.
After 5 years of ponding, chloride ion contents of 19.9, 11.1 and 6.7 kg/m3 (33.5,
18.7 and 11.3 lb/yd3) in the vicinity of the steel were reported for lollipop
specimens containing O, 3.75 and 7.5 percent silica fume, respectively [18].
These concretes had a water-cement ratio of 0.48. The data indicate that, at the
time of sampling, screening efficiencies of 44 and 66 percent were obtained with
the concretes containing 3.75 and 7.5 percent silica fume, respectively.
These screening efficiencies are lower than those predicted by Fig. 1. The
“apparent” discrepancy can be attributed to several factors including the fact that
the reported chloride ion contents were obtained from 19 mm (0.75 in.) thick
concrete rings in the vicinity of the reinforcing steel, and therefore, may not be
representative of those at the steel surface at the time of corrosion initiation in the
silica fume concretes.
Because the chloride ion contents reported for the lollipop specimens containing
silica fume are well above the chloride ion threshold for reinforcing steel
embedded in silica fume concrete, “equivalent chloride ion threshold” values
cannot be determined.
Powder samples for the chloride analyses were obtained at the level of the
reinforcing steel, several weeks after the onset of corrosion in the treated concrete
slab. Therefore in this evaluation, the “equivalent chloride threshold” for the
treated concrete may have been slightly less than 8.2 kg/m3 (13.9 lb/yd3), which
would be consistent with the value of 7.1 kg/m3 (12.0 lb/yd3) reported in
Reference 7.
The chloride data indicates that a screening efficiency of about 67 percent was
obtained with the water-based organic corrosion inhibitor treated-concrete at the
level of the top reinforcing steel at the time of sampling [8]. Chloride data from
a companion evaluation of precracked beams also indicated that, at a distance of
about 1.5 in. (38 mm) away from the crack at the level of the top reinforcing
steel, an average screening efficiency of 68 percent was obtained with the water-
based organic corrosion inhibitor-treated concrete.
Assuming a chloride ion threshold of 2.4 kg/m3 (4 lb/yd3) for the water-based
organic corrosion inhibitor-treated concrete and a screening efficiency of about
67 percent, from Fig. 3 performance similar to that obtained with concrete treated
with 20 Um3 (4 gaVyd3) of the calcium nitrite inhibitor would be expected.
However, as stated earlier, concrete treated with 5 Um3 (1 gal/yd3) of the water-
based organic corrosion inhibitor performeå as well as concrete treated with 30
Um3 (6 gaVyd3) of the calcium nitrite inhibitor in the accelerated macrocell
corrosion evaluations.
The recalculated chloride protection levels for the calcium nitrite inhibitor are
also shown in Fig. 3 to the right of the manufacturer's recommended levels.
Using a screening efficiency of 67 ercent and the curve representing a chloride
!i
ion threshold of 2.4 kg/m3 (4 Ib/yd ) as before, Fig. 3 now predicts that concrete
treated with 5 Um3 (1 gaYyd3) of the water-based organic corrosion inhibitor
would give a level of performance similar to that obtained with concrete treated
with 30 L/m3 (6 gaYyd3) of the calcium nitrite inhibitor. This is consistent with
the results of the accelerated time-to-corrosion evaluations.
percent lower than those predicted earlier using the chloride protection levels
published by the manufacturer.
From the above, it can be deduced that for concretes treated with passive-active
systems such as the water-based organic corrosion inhibitor, continued screening
effectiveness and/or the ability of the protective film to reduce the availability of
oxygen and moisture at the steel surface can reduce the corrosion rate. For
concretes treated with an active system such as the calcium nitrite inhibitor,
possible migration of the active ingredient(s) to the steel surface may influence
the corrosion rate for a period of time. However, if the active ingredients of the
inhibitor gets completely overwhelmed by chloride ions, corrosion would most
likely proceed at a rate equal to, if not greater, than that in untreated concrete.
Scenario No. 1: assume that Inhibitor Y delays the onset of corrosion longer and
also corrodes at a lower rate than Inhibitor X. Without any computation, it is
clearly obvious that inhibitor Y will provide better corrosion protection
compared to Inhibitor X.
Scenario No. 2: assume in this case that Inhibitor Y delays the onset of corrosion
longer, but corrodes at a faster rate compared to Inhibitor X. In this situation, a
comparison of total effectiveness cannot be made without some amount of
computation. Analytical models for chloride ion diffusion into concrete and life
prediction for corroding structures will be required for such computations, taking
into account factors such as the expected service life of the structure and the level
of damage that would be considered unacceptable.
CONCLUSIONS
The following conclusions can be drawn from the information presented in this
Paper.
2. Passive and passive-active inhibitor systems can delay the onset of corrosion
longer than active inhibitor systems, depending on the level of screening
efficiency that can be obtained.
REFERENCES
2. Rosenberg, A.M., Gaidis, J.M., Kossivas, T.G., and Previte, R.W., “A Corrosion
Inhibitor Formulated with Calcium Nitrite for Use in Reinforced Concrete,”
Chloride Corrosion of Steel in Concrete, ASTM STP 629, American Society for
Testing and Materials, 1977, pp. 89-99.
3. Berke, N.S., “The Effects of Calcium Nitrite and Mix Design on the Corrosion
Resistance of Steel in Concrete (Part 2, Long-Term),’’CORROSION OF METALS
IN CONCRETE, Proceedings of the CORROSION/87 Symposium on Corrosion of
Metals in Concrete, National Association of Corrosion Engineers, 1987, pp. 134-
144.
10. Hime, W. and Erlin, B., ”Some Chemical and Physical Aspects of Phenomena
Associated with Chloride-Induced Corrosion,“ CORROSION, CONCRETE, AND
CHLORIDES - Steel Corrosion in Concrete: Causes and Restraints, SP-102,
American Concrete Institute, 1987, pp. 1-9.
11. Berke, N.S., and Rosenberg, A., “Technical Review of Calcium Nitrite Corrosion
Inhibitor in Concrete,” Concrete Bridge Design and Maintenance: Steel Corrosion
in Concrete, Transportation Research Record 1211, 1989, pp. 18-27.
12. Metha, P.K., “Pozzolanic and Cementitious Byproducts as Mineral Admixtures for
Concrete - A Critical Review,” Fly Ash, Silica Fume, Slag & Other Mineral By-
Products in Concrete, Malhotra V.M. ed., SP-79, Vol. I, American Concrete
Institute, Detroit, 1983, pp. 1-46.
13. Rose, J., “The Effect of Cementitious Blast-Fumace Slag on Chloride Permeability
of Concrete,”CORROSION, CONCRETE, AND CHLORIDES - Steel Corrosion in
Concrete: Causes and Restraints, SP-102, American Concrete Institute, 1987, pp.
107-125.
14. Ohama, Y., “Principleof Latex Modification and Some Typical Properties of Latex-
Modified Mortars and Concretes,”AC1 Materials Journal, Nov.-Dec. 1987, pp. 511-
518.
15. Rixom, M.R., and Mailvaganam, N.P., Chemical Admixtures for Concrete, Second
Edition, E. & F.N. Spon Ltd, London,l986,306 pp.
16. Berke, N.S., “Resistance of Microsilica Concrete to Steel Corrosion, Erosion, and
Chemical Attack,” SP-114, Fly Ash, Silica Fume, Slag, and Natural Pozzolans in
Concrete, Vol. 2, V.M. Malhotra ed., American Concrete Institute, Detroit, 1989,
pp. 861-886.
18. Berke, N.S., Dallaire, M.P., and Hicks, M.C., “Plastic, Mechanical, Corrosion, and
Chemical Resistance Properties of Silica Fume (Microsilica) Concretes,” SP-132,
Fly Ash, Silica Fume, Slag, and Natural Pozzolans in Concrete, Vol. II, V.M.
Malhotra ed., American Concrete Institute, Detroit, 1993, pp. 1125-1149.
19. Annual Book of ASTM Standards, Volume 04.02 - Concrete and Aggregates,
American Society for Testing and Materiais, Philadelphia, PA, 1990, 804 pp.
20. Clear, K.C., “Measuring Rate of Corrosion of Steel in Field Concrete Structures,”
Concrete Bridge Design and Maintenance: Steel Corrosion in Concrete,
Transportation Research Record 121 1, pp. 28-37.
21. Andrade, C., Castelo, V., Alonso, C., and Gonzalez, J.A., “The Determination of the
Corrosion Rate of Steel Embedded in Concrete by the Polarization Resistance and
AC Impedance Methods,” Corrosion Effect of Stray Currents and the Techniques
for Evaluating Corrosion of Rebars in Concrete, ASTM STP 906, V. Chaker, Ed.,
American Society for Testing and Materiais, Philadelphia, 1986, pp. 43-63.
22. F e h , S., Gonzalez, J.A., Feliu, S., Jr., and Andrade, M.C., “Polarization Resistance
in Reinforced Concrete,” AC1 Materials Journal, Vol. 87, No. 5, Sep./Oct. 1990, pp.
457-460.
23. Matsuoka, K., Kihira, H., Ito, S.. and Murata, T., Paper No. 121, Presented at
NACE Corrosion87, National Association of Corrosion Engineers, Houston. TX,
1987.
24. Andrade, C., Rz-Maribona, I., F e h , S., and Gonzalez, J.A., “Macrocell versus
Microcell Corrosion of Reinforcements Placed in Parallel,” Paper No. 194, NACE
Corrosion92, National Association of Corrosion Engineers, Houston, TX., 16 pp.
TABLE 2 - CORROSION DATA FOR SILICA FUME AND CALCIUM NITRITE-TREATED CONCRETES
(SOURCE: EXTRACTED FROM TABLE 1 IN REF. 18)
5-year Total
Corrosion Corrosion
Trial Cement Water- Silica Fume Calcium Rate for for Minidecks
No. Factor Cement Content Nitrite Lollipops with 35 mm
(kg/m3) Ratio (%by mass Dosafe YRp Cover
ofcement) (Um) gLs/cmz) (months*pA)
i2
-Chioride Threshold
10 30
25
BsY%
2o 30
15 25
20
10 15
10
2
Fig. 1-Chloride ion content versus screening efficiency curve for silica fume concrete
70
O
O 3.75 7.5
Silica Fume Percentage
Fig. 2-Effect of silica fume on chloride ion ingress after 5 years of ponding (18)
CNI
12
Dosage
(LW)
M.5
10
30
25
15 25
20
10 15
10
2
0
100 90 80 70 60 50 40 30 20 10 0
Screening Efficiency (%)
(CM - Calam rumte inhibitor; MR - Mandacrura's recammended dosage; Rev - R-alculaid dmagc)
Fig. 3-Chloride ion content versus screening efficiency curves for water-based organic corrosion
inhibitor-treated concrete
Fig. 4-5-year corrosion rate data for 0.38 water-cement ratio concretes (18)
45 r 41
40
35
E 30
$ i
-
a
25
20
g l5
10
5
O
Plain 3.75% 7.5% 15% CNI CNI
SF SF SF @ 10 @ 20
LIm3 Llm3
Fig. 5-5-year corrosion rate data for 0.43 water-cement ratio concretes (18)
Fig. 6 5 - y e a r corrosion rate data for 0.48 water-cement ratio concretes (18)
SP 154-31
Synopsis:
This paper presents the results of laboratory and
field tests conducted, by the authors over a twelve year
period, on the long-term durability performance of silica
fume based concretes, shotcretes, grouts, and concrete
slab overlays. The silica fume mixture proportions
varied, from 2 to 2 2 % of silica fume (by weight of
cement) and included normal portland cement, shrinkage
compensating cement, normal weight and light weight
aggregates. Laboratory and field specimens were tested
for compressive strength, bonding strength, chloride
permeability, electrical resistivity, and freezing and
thawing durability. The exhibited long-term performance
characteristics indicate that silica fume based
concretes, shotcretes and grouts provide excellent
protection to embedded reinforcing steel in chloride
environments. A s the materials age, they become
stronger, lower in permeability and higher in
resistivity. Silica fume modified materials produce one
of the best cementitious products available for adverse
concrete environments.
587
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
A C 1 SP-154 95 m 0662949 0522772 517
588 Wolsiefer and Clear
INTRODUCTION
Since the mid ~O'S, silica fume has been used in the
United States as a means of enhancing the durability of
concretes and grouts. The initial uses were in the
chemical industry, where increased resistance to acid was
required. Beginning in about 1980, the authors began to
consider silica fume (dry, uncompacted, in concert with
a superplasticizer) as a means of significantly
decreasing the chloride permeability of concrete and
increasing its electrical resistivity. The need for
improved concrete properties, is related to corrosion of
the reinforcement, as a result of exposure to deicing
salt or marine environments. Conventional concretes and
grouts exhibit relatively high permeability and low
resistivity. The chlorides, once present at the
reinforcing level at a concentration of about (0.77
kg/m3) 1.3 lbs/cy or 0.035% by weight of concrete,
destroy the passive film on the steel and act as a
catalyst, resulting in concrete cracking, delaminating
(undersurface fractures at the level of the steel) and
spalling. In some structures, the metal loss caused by
the corrosion can be significant and lead to structural
performance problems. Millions of dollars have been
spent in the last two decades to rehabilitate existing
bridge and parking decks and to provide protective
systems for new structures. This paper summarizes our
findings in laboratory, outdoor exposure and field
efforts during the last 15 years.
INITIAL LABORATORY TEST PROGRAMS
In the early 1 9 8 0 ' ~ ~the authors initiated
laboratory test programs to evaluate the potential of
S I L I C A FUME SHOTCRETE
c; '3
%-
--
.-E
=>
L
aa
.--z-.
w
E
aa
U
&
c
L
L
a
.-c
O
YI
e
L
O
U
-
vL
U
B
N
I
m
i=
al
Q
e
SP 154-32
607
COPYRIGHT ACI International (American Concrete Institute)
Licensed by Information Handling Services
AC1 SP-I54 95 m 0662949 0522791 479 m
608 Holland
INTRODUCTION
ASTM EFFORTS
AASHTO EFFORTS
Austral ia
The Australian standard [ i 7 1 is the newest of the
standards reviewed for this paper. The Australian
authors have obviously taken advantage o f the work
that predated their efforts.
The requirements for Si02 content, moisture
content, and loss on ignition are typical. Both the
Cio2 content and the loss on ignition can be violated
"providing the supplier o f that silica fume
demonstrates that the silica fume behaves
satisfactorily as a supplementary cementitious
material with regard to water demand and compressive
strength as appropriate."
This standard does include a mandatory maximum
limit on SO3 of 3 percent. Alkalies are to be
reported if "alkali-aggregate" reaction is considered
Canada
The Canadian Standards Association (CSA) issued a
standard, "Supplementary Cementing Materials" [18], in
1986 that includes silica fume. CSA elected t o cover
all supplementary cementing materials in one document,
with silica fume being included as "Type U." This
document defines silica fume as material coming from
the production of silicon or a ferrosilicon alloy
containing at least 75 percent silicon. However,
there is a provision that allows for the use of silica
fume from furnaces producing alloys with a lower
silicon content if the performance of that silica fume
is shown to be acceptable ttirough appropriate testing.
A s can be seen in Table 3 , there is little in this
document that is unusual. Much of what originally
appeared in this Canadian standard later was adopted
by AASHTO dnd ASTM.
Norway
At least two standards covering the use of silica
fume in concrete have been developed in Norway. The
first covers the requirements for silica fume itself
while the second covers the actual use in concrete.
Values from the first standard [i91 are included in
Table 3.
This standard does address the product form by
stating that three forms are available: natural state,
densified, and slurry. Unlike the Australian
standard, no minimum bulk density for the densified
form i s given.
The properties and limits in the Norwegian
standard are similar to those seen in other documents.
One interesting new wrinkle is the limit on the amount
of chloride which is handled as follows: if the silica
fume has a chloride content by mass of less than 0 . 1
percent, the material may be termed "chloride free."
If the chloride content is greater than this limit,
the actual value must be declared.
The Norwegian standard does recognize the
importance of surface area and require the use of a
surface adsorption techn'que to make the measurement.
The value selected, 12 m 5 /g is somewhat lower than
other values that have been cited.
The most significant aspect of the Norwegian
specification is the way in which the various physical
and chemical limits (termed "characteristic values")
are handled. "Compliance with a characteristic value
is assessed by means of a statistical control
procedure." It further states statistical parameters
foi- cornpl iance:
a. "The characteristic values specified shall
correspond to the 10 percent fractile."
b. "The confidence level shall be 95 percent for
all of the properties dealt with in this standard."
RILEM
~-
South Africa
The South African Standard [ 2 0 ] also contains no
surprises. The properties specified and the limits
shown are fairly typical, as can be seen in Table 3.
There i s no mention of surface area in this standard.
This standard does include ari interesting
reference to product form: “This part of the
specification covers, in addition to the material in
powder form, only densified material which will, when
tested in accordance with the relevant test methods
given in this part of the specification, so readily
disperse in water or in an aqueous solution that it
deconglomerates to a powder form.“ Dry compacted
silica fume will not typically disperse simply by
putting it in water.
A limit is established for equivalent sodium oxide
content when the silica fume is to be used to reduce
alkali-silica reaction. A discussion in an appendix
t o the standard outlines additional testing and
recommendations for this application.
The South African Standard does include a
requirement t o test for pozzolanicity using a lime-
reactivity test method. An absolute requirement for
compressive strength of 5 MPa is used rather than a
comparison with a control. Apparently, “relatively
little use is made o f the test by the local supplier
of silica furne, having established that their source
meets the specified requirements.” 1
DISCUSSION
REFERENCES
TABLE 1 - EXCERPTS FROM ASTM C 1240-93, STANDARD SPECIFICATION FOR SILICA FUME
FOR USE IN HYDRAULIC-CEMENT CONCRETE AND MORTAR (i)
Physical Requirements
Oversize:
Amount retained on 45-prn (No. 325) sieve,
maximum o f 1 0 percenta
Acce 1 erated pozzo 1 an i c act i v i t y i ndex :
With portland cement at 7 days, minimum 85
percent o f control
TABLE 1 (coont.) - EXCERPTS FROM ASTM C 1240-93, STANDARD SPECIFICATION FOR SILICA
FUME FOR USE IN HYDRAULIC-CEMENT CONCRETE AND MORTAR (i)
Uniformity requirements:
The density and oversize of individual
samples shall not vary from the average established by
t t w ten preceding tests or by all preceding tests if
the number is less than ten, by more than:
density, maximum variation from average. 5
percent
oversize, percent retained on 45-ym ( N o . 325)
sieve, maximum of 5 percentage points from average
TABLE 1 (tont.) - EXCERPTS FROM ASTM C 1240-93, STANDARD SPECIFICATION FOR SILICA
FUME FOR USE IN HYDRAULIC-CEMENT CONCRETE AND MORTAR (1)
Suifate resistance expansion,c
Chemical Requirements.
Silicon Dioxide (Sioz), minimum 8 5 percent
Sulfur trioxide ( S O 3 ) , maximum 3.0 percent
Loss on ignition, maximum 7.0 percent
Physical Requirements
Fineness: Amount retained when wet sieved on NO.
5 0 0 sieve, O percent
i 5
z
9-
%5
z
II
i
2
5 i
z
z
a.- i
zz z
U
O
Rz &z.
9-
cr)
I
!
O
i
s
z 5
z s
z P z
II
c
P-
O
II i i
Z z
i i i ai-
I
r
Z Z z SE
E
Y
c
Y
v,
i i i i i
z Z z z z
i
z i
z oer^
O-
s
Z
v,
æ
E
i
z
s
z
i
z
Oe,
zz.
U
O
a r
h:
i
z z i
z
n o
E
E- 0 0
m 3 z z
0
z z
E
x
.e
8;
U
C
EE
sz 8
0 i 2
z z
i
z 0
Z
h
C
.-O
In
i
Z
s
z i
Z 9
0
z
5
Z
5 i
Z z
U
al
c
E
E
i
z s
Z nH z -
.-05m
al
'FI
L
.9
al
s
8
v)
i
z
m>p =05
.-5
5
m
C'FI
5 i
z z
-O -- No guidance provided.
--
South Africa -- No guidance provided.
Norway (1) A 5 12
Characteristic limit Yo
L
c5 >95
m'lg
>12
-jI.l
I:
Length
Area
Volume (capacity)
Force
Mass
Temperatures
This selected list gives practical wnvenion facton ofunits found in wncrete technology. The reference sourDesfor information
on SI units and more a c t wnversion facton are ASIU E 380 and E 621. Symbols of metric units are given in parenthesis.
t E Indicates that lhe factor given is exact.
+ One liter (cubic decimeter) equals 0.001m3or loo0 cm3.
f These equations wnvert one temperature reading to another and include the necessary scale corrections. To convert a
differen- in temperahwe from Fahrenheit degrees lo Celsius degrees, divide by 1.8 oniy, i.e, a change from 70 to 88 F represents
e change of 18 For 18fl.û = 10 C deg.
INDEX