VOL. 110, NO.

3
MAY-JUNE 2013

ACI
Materials J O U R N A L

A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

 CONTENTS
Board of Direction ACI Materials Journal
President
Anne M. Ellis May-June 2013, V. 110, No. 3
a journal of the american concrete institute
Vice Presidents
an international technical society
William E. Rushing Jr.
Sharon L. Wood

Directors 259 Effect of Microcracking on Frost Durability of High-Volume-Fly-

Neal S. Anderson Ash- and Slag-Incorporated Engineered Cementitious Composites,
Khaled Awad by Erdoğan Özbay, Mustafa Şahmaran, Mohamed Lachemi, and Hasan
Roger J. Becker Erhan Yücel
Dean A. Browning
Jeffrey W. Coleman
Robert J. Frosch
James R. Harris 269 Evaluation of Concrete Drying Shrinkage Related to Moisture Loss,
Cecil L. Jones by Reza Abbasnia, Mohammad Shekarchi, and Jamal Ahmadi
Cary S. Kopczynski
Steven H. Kosmatka
Kevin A. MacDonald 279 Mechanical Energy Dissipation Using Cement-Based Materials with
David M. Suchorski
Admixtures, by Po-Hsiu Chen and D. D. L. Chung
Past President Board Members
James K. Wight
Kenneth C. Hover 291 Effect of Leaching on pH of Surrounding Water, by David W. Law and
Florian G. Barth Jane Evans
Executive Vice President
Ron Burg
297 Novel Cementitious Binder Incorporating Cement Kiln Dust: Strength
and Durability, by Piyush Chaunsali and Sulapha Peethamparan
Technical Activities Committee
Ronald Janowiak, Chair
Daniel W. Falconer, Staff Liaison
JoAnn P. Browning 305 Sustainable Processing of Cellulose Fiber Cement Composites, by
Chiara F. Ferraris Parviz Soroushian, Jong-Pil Won, and Maan Hassan
Catherine E. French
Fred R. Goodwin
Trey Hamilton
Kevin A. MacDonald 315 Effect of Curing Water Availability and Composition on Cement
Antonio Nanni Hydration, by Md Sarwar Siddiqui, Wesley Nyberg, Wilson Smith, Brett
Jan Olek Blackwell, and Kyle A. Riding
Michael Sprinkel
Pericles C. Stivaros
Andrew W. Taylor
Eldon G. Tipping 323 Detection and Characterization of Early-Age Thermal Cracks in High-
Performance Concrete, by David Hubbell and Branko Glisic
Staff
Executive Vice President
Ron Burg 331 Alkali-Activated Natural Pozzolan Concrete as New Construction
Material, by Dali Bondar, Cyril J. Lynsdale, and Neil B. Milestone
Engineering
Managing Director
Daniel W. Falconer
339 Impact Resistance of Blast Mitigation Material Using Modified ACI
Managing Editor Drop-Weight Impact Test, by John J. Myers and Matthew Tinsley
Khaled Nahlawi
Staff Engineers
Matthew R. Senecal 349 In ACI Structural Journal
Gregory Zeisler

Publishing Services
Manager Discussion is welcomed for all materials published in this issue and will appear in the March-April
Barry M. Bergin 2014 issue if received by December 1, 2013. Discussion of material received after specified dates will
be considered individually for publication or private response.
Editors ACI Standards published in ACI Journals for public comment have discussion due dates printed with
Carl R. Bischof the Standard.
Karen Czedik Annual index published online at www.concrete.org/pubs/journals/mjhome.asp.
Kelli R. Slayden
ACI Materials Journal
Denise E. Wolber Copyright © 2013 American Concrete Institute. Printed in the United States of America.
Editorial Assistant The ACI Materials Journal (ISSN 0889-325x) is published bimonthly by the American Concrete Institute. Publication
Ashley Poirier office: 38800 Country Club Drive, Farmington Hills, MI 48331. Periodicals postage paid at Farmington, MI, and at
additional mailing offices. Subscription rates: $161 per year (U.S. and possessions), $170 (elsewhere), payable in
advance. POSTMASTER: Send address changes to: ACI Materials Journal, 38800 Country Club Drive, Farmington
Hills, MI 48331.
Canadian GST: R 1226213149.
Direct correspondence to 38800 Country Club Drive, Farmington Hills, MI 48331. Telephone: (248) 848-3700.
Facsimile (FAX): (248) 848-3701. Website: http://www.concrete.org.

ACI Materials Journal/May-June 2013 257

 Contributions to
MEETINGS ACI Materials Journal
The ACI Materials Journal is an open
forum on concrete technology and papers
2013 9-12—2013 Airfield & Highway related to this field are always welcome.
Pavements Conference, Los Angeles, All material submitted for possible publi-
MAY CA, content.asce.org/conferences/ cation must meet the requirements of
pavements2013 the “American Concrete Institute Publi-
17-18—ACPA 2013 Education cation Policy” and “Author Guidelines
Conference, Scottsdale, AZ, www. 18-20—Risk Management of Corrodible and Submission Procedures.” Prospective
concretepumpers.com/content/2013- authors should request a copy of the Policy
Systems, Washington, DC, events.nace.org/
and Guidelines from ACI or visit ACI’s
education-conference conferences/Rish2013/index.asp website at www.concrete.org prior to
submitting contributions.
20-22—Seventh National Seismic 18-23—7th Annual International
Conference on Bridges & Highways, Structural Engineering and Construction Papers reporting research must include
Oakland, CA, www.7nsc.info a statement indicating the significance of
Conference, Honolulu, HI, www.isec-
the research.
society.org/ISEC_07
26-29—Twin International Conferences The Institute reserves the right to return,
on Civil Engineering Towards a Better JuLY without review, contributions not meeting
Environment and The Concrete Future, the requirements of the Publication Policy.
Covilhã, Portugal, www.uc.pt/en/iii/ 8-10—9th International Conference All materials conforming to the Policy
novidades/2012/twinconferencesucubi on Earthquake Resistant Engineering requirements will be reviewed for editorial
Structures (ERES 2013), A Coruña, Spain, quality and technical content, and every
27-29—International Conference on wessex.ac.uk/eres2013 effort will be made to put all acceptable
Concrete Sustainability (ICCS13), Tokyo, papers into the information channel.
10-13—2013 CFA Summer Convention, However, potentially good papers may be
Japan, www.jci-iccs13.jp returned to authors when it is not possible
Santa Ana Pueblo, NM, www.cfawalls.org
to publish them in a reasonable time.
June
24-26—ICSA 2013: Second International
Discussion
2-5—International Bridge Conference Conference on Structures and
2013, Pittsburgh, PA, www.eswp.com/bridge Architecture, Guimarães, Portugal, www. All technical material appearing in the
icsa2013.arquietectura.uminho.pt ACI Materials Journal may be discussed.
2-5—12th Canadian Masonry Symposium, If the deadline indicated on the contents
25-28—ASCC CEO Forum, West Palm page is observed, discussion can appear
Vancouver, BC, Canada, www.cms2013.ca/
in the designated issue. Discussion should
index.php/cms/2013 Beach, FL, www.ascconline.org
be complete and ready for publication,
including finished, reproducible illustra-
3-5—TotallyConcrete Expo 2013, AUGUST tions. Discussion must be confined to the
Johannesburg, South Africa, www. scope of the paper and meet the ACI Publi-
totallyconcrete.co.za 4-7—International Symposium of
cation Policy.
Climatic Effects on Pavement and
Geotechnical Infrastructure, Fairbanks, Follow the style of the current issue.
5-8—Construction Expo 2013, São Paulo, Be brief—1800 words of double spaced,
Brazil, www.construction expo.com.br AK, http://ine.uaf.edu/autc/2013/02/06/
typewritten copy, including illustrations
international-symposium-of-climatic- and tables, is maximum. Count illustrations
6-7—CSDA Summer Meetings, Westlake effects-on-pavement-and-geotechnical- and tables as 300 words each and submit
Village, CA, www.csda.org infrastructure them on individual sheets. As an approxi-
mation, 1 page of text is about 300 words.
Submit one original typescript on 8-1/2 x
UPCOMING ACI CONVENTIONS 11 plain white paper, use 1 in. margins,
and include two good quality copies of the
The following is a list of scheduled ACI conventions: entire discussion. References should be
2013—October 20-24, Hyatt Regency & Phoenix Convention Center, Phoenix, AZ complete. Do not repeat references cited
2014—March 23-27, Grand Sierra Resort, Reno, NV in original paper; cite them by original
2014—October 26-30, Hilton Washington, Washington, DC number. Closures responding to a single
discussion should not exceed 1800-word
For additional information, contact: equivalents in length, and to multiple
discussions, approximately one half of
Event Services, ACI
the combined lengths of all discussions.
38800 Country Club Drive, Farmington Hills, MI 48331 Closures are published together with
Telephone: (248) 848-3795 the discussions.
e-mail: conventions@concrete.org
Discuss the paper, not some new or
outside work on the same subject. Use
references wherever possible instead of
ON COVER: 110-M25, p. 298, Fig. 2—Particle size distribution. repeating available information.
Discussion offered for publication should
offer some benefit to the general reader.
Discussion which does not meet this
Permission is granted by the American Concrete Institute for libraries and other users registered with the Copyright requirement will be returned or referred to
Clearance Center (CCC) to photocopy any article contained herein for a fee of $3.00 per copy of the article. Payments
should be sent directly to the Copyright Clearance Center, 21 Congress Street, Salem, MA 01970. ISSN 0889-3241/98 the author for private reply.
$3.00. Copying done for other than personal or internal reference use without the express written permission of the
American Concrete Institute is prohibited. Requests for special permission or bulk copying should be addressed to the
Managing Editor, ACI Materials Journal, American Concrete Institute. Send manuscripts to:
The Institute is not responsible for statements or opinions expressed in its publications. Institute publications are not able http://mc.manuscriptcentral.com/aci
to, nor intend to, supplant individual training, responsibility, or judgment of the user, or the supplier, of the information
presented.
Send discussions to:
Papers appearing in the ACI Materials Journal are reviewed according to the Institute’s Publication Policy by individuals
expert in the subject area of the papers. Journals.Manuscripts@concrete.org

258 ACI Materials Journal/May-June 2013

 ACI MATERIALS JOURNAL
Title no. 110-M21
Effect of Microcracking on Frost Durability of High-Volume-
Fly-Ash- and Slag-Incorporated Engineered Cementitious
Composites
by Erdoğan Özbay, Mustafa Şahmaran, Mohamed Lachemi, and Hasan Erhan Yücel
This paper reports the durability performance of high-volume-
fly-ash (FA)- and slag (S)-incorporated engineered cementitious
composites (ECCs) when subjected to mechanical loading and
freezing-and-thawing cycles. Composites containing two different
contents of FA and slag as a replacement of cement (55 and 70%
by weight of total cementitious materials) are examined. To find out
the effect of mechanical preloading on the frost durability of ECCs,
prism specimens were preloaded up to a certain deformation level
under four-point bending loading to generate microcracks. Then,
the preloaded and pristine (sound) specimens were subjected to the
freezing-and-thawing test in accordance with ASTM C666/C666M.
Experimental tests consisted of measuring the change in mass and
ultrasonic pulse velocity (UPV) and residual flexural properties
of ECC specimens exposed to the freezing-and-thawing cycles up
to 300. Test results revealed that the frost resistance of ECCs was
significantly influenced by the mineral admixture type and amount
and preloading deformation. The deterioration with an increasing
number of freezing-and-thawing cycles was relatively more for
ECC mixtures with FA than for slag mixtures at the same replace-
ment level. In addition, an increase in the FA replacement rate was
observed to exacerbate the deterioration caused by freezing-and-
thawing cycles. Apart from some reduction in flexural properties
and UPV and an increase in mass loss and residual crack width,
the results presented in this study, however, confirm the durability
performance of ECC material under freezing-and-thawing cycles,
even in cases where the material experiences mechanical loading
that deforms it into the strain-hardening stage prior to exposure.
It is important to note that this durability of ECCs under freezing
and thawing was achieved without deliberate air entrainment, and
contrary to conventional concrete, no relationship of frost resis-
tance was found to the air-void structure of the ECC mixtures.
Keywords: cracking; engineered cementitious composites; frost resistance;
mineral admixture.
INTRODUCTION
Resistance to frost action is used as a durability indicator of
cementitious composites such as mortar, concrete, and fiber-
reinforced concrete. Generally, frost durability of cementi-
tious composites is defined in the laboratory conditions on
the pristine (sound) specimens. However, in in-service struc-
tures, as a result of various physical, chemical, and mechan-
ical interactions between structure and environment, cracks
may occur at different stages throughout the life of the struc-
ture. Therefore, the conclusions obtained from the pristine
specimens might not always be correct and can have insuf-
ficient reliability.1 To correctly define the frost durability of
the structure in the field, it is therefore necessary to study
not only pristine specimens but also the cracked specimens.
In the last decade, Li2 and Li et al.3 have investigated a
new class of high-performance fiber-reinforced cementitious
composites—namely, engineered cementitious composites
(ECCs)—which are designed to achieve high ductility under
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
tensile loading and high durability under normal service
conditions. The mixture proportions of ECC are optimized
through micromechanics-based material design theory to
satisfy strength and energy criteria to attain high composite
ductility.2,3 The type, size and amount of fiber, matrix ingre-
dients, and interface characteristics are tailored for multiple
cracking and controlled crack width in ECCs. These have
a substantial benefit in both high ductility and improved
durability due to this tight crack width under applied loads.
Recently, various investigations have focused on the dura-
bility-related performance (alkaline environment resistance,
transport properties, corrosion resistance, salt scaling, and
chloride ion diffusion) of standard—also known as ECC
M45—(with 55% fly ash [FA]) and high-volume FA (with
70% FA) ECC mixtures.4-8 Investigations were made on both
pristine and preloaded or precracked specimens, and it was
observed that ECCs have superior durability performance for
both cases. However, very little work was done on the frost
durability of ECC mixtures. Şahmaran et al.6 assessed the
freezing-and-thawing resistance on the pristine specimens of
a standard non-air-entrained ECC mixture that has a fly ash
(FA)-cement (C) ratio of 1.2 by weight. They concluded that
despite a slight reduction in ductility and flexural strength,
after 300 freezing-and-thawing cycles, ECC samples are
found to retain their tensile ductility more than a few hundred
times that of normal concrete and fiber-reinforced concrete
with no environmental exposure. Li et al.7 also conducted
a freezing-and-thawing test on pristine ECC specimens and
concrete prisms concurrently over 14 weeks. They observed
that after 5 weeks (110 cycles), the concrete specimens had
severely deteriorated, requiring removal from the test cabinet.
However, all ECC specimens survived 300 cycles with no
degradation of dynamic modulus. This performance results
in a durability factor of 10 for concrete compared to 100 for
ECC, as computed according to ASTM C666/C666M.
As mentioned previously, studies on the frost durability of
non-air-entrained (without an air-entraining admixture) ECC
have been conducted in the absence of any external load
in accordance with ASTM C666/C666M. For some engi-
neering applications, such as pavements and bridge decks
in cold regions, the damage might occur through the double
actions of mechanical loading and freezing-and-thawing
cycles.9 In this paper, therefore, the durability performance
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-262.R1 received February 3, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
259
Erdogĝan Özbay is an Associate Professor in the Civil Engineering Department at mens were then exposed to freezing-and-thawing cycles in
Mustafa Kemal University, Antakya, Turkey. His research interests include durability accordance with ASTM C666/C666M, Procedure A. Exper-
of concrete, usage of waste materials in concrete, and self-consolidating concrete. imental tests consisted of measuring the change in mass
Mustafa Şahmaran is an Associate Professor in the Department of Civil Engineering at
and ultrasonic pulse velocity (UPV) and residual flexural
Gaziantep University, Gaziantep, Turkey. His research interests include advanced materials properties of ECC specimens exposed to 300 freezing-and-
technology and composite materials development for sustainable infrastructure. thawing cycles. In addition, fresh air content and hardened
air-void parameters of ECC mixtures in accordance with
Mohamed Lachemi is a Professor in the Department of Civil Engineering at Ryerson
ASTM C457/C457M were studied.
University, Toronto, ON, Canada. His research interests include the use of high-
performance materials in the built infrastructure, including the development and use
of self-consolidating concrete in construction. RESEARCH SIGNIFICANCE
The use of ECCs has been steadily increasing in the past
Hasan Erhan Yücel is a PhD Student Research Assistant in the Department of Civil decade. The major areas of application include road and
Engineering at Gaziantep University. His research interests include composite
materials development for sustainable infrastructure.
airport pavements and bridge decks. ECCs also have excel-
lent potential for the rehabilitation of distressed structures.
Cracking in ECCs is fundamentally different from that which
Table 1—Chemical properties of cement, fly ash, occurs in concrete or reinforced concrete. ECCs develop
and slag multiple microcracking in a strain-hardening response
Chemical composition, % Cement Fly ash Slag as a result of mechanical loading. Due to multiple micro-
CaO 61.43 1.64 34.48 cracking, it is important to test the precracked ECCs under
freezing-and-thawing cycles, which is of great importance
SiO2 20.77 56.22 38.40 for the critical assessment of the durability of ECCs under
Al2O3 5.55 25.34 10.96 comprehensive conditions.
Fe2O3 3.35 7.65 0.81
EXPERIMENTAL STUDIES
MgO 2.49 1.80 7.14 Materials and mixture properties
SO3 2.49 0.32 1.48 Two groups of ECC mixtures were prepared. The first
group includes the standard and high-volume FA ECC
K2O 0.77 1.88 0.86
mixtures, in which C is replaced with 55% (FA/C ratio of 1.2)
Na2O 0.19 1.13 0.18 and 70% (FA/C ratio of 2.2) Class F FA, respectively. The
Loss on ignition 2.20 2.10 3.00 ECC mixture with an FA/C ratio of 1.2 (also called M45),
which has the largest experimental data set published to date,
Physical properties
is called the “standard ECC mixture” in the literature.4-8 The
Specific gravity 3.06 2.31 2.79 second group includes similar ECC mixtures as the first
Blaine fineness, m /kg2
325 290 425 group, except slag is used instead of FA in ECC produc-
Note: 1 m2/kg = 0.54 yd2/lb. tion. Therefore, a total four ECC mixtures (two with FA and
two with S) were produced. In the production of the ECC
mixtures, CEM I-42.5 cement (similar to ASTM Type I),
fine silica sand with a maximum aggregate size of 400 mm
(0.016 in.), water, polyvinyl alcohol (PVA) fibers, and a
polycarboxylic-ether-type high-range water-reducing admix-
ture (HRWRA) with a solid content of 40% were used. The
chemical composition and physical properties of C, FA, and
slag are presented in Table 1. The particle size distribution
curves for silica sand, C, FA, and slag are presented in Fig. 1.
The morphological characteristics of FA and slag were also
identified with a scanning electron microscope (SEM) and
the resulting photograph is presented in Fig. 2.
The mixture proportions of the two ECC groups are given
in Table 2. The PVA fiber content of 2% by volume was kept
constant in both groups of mixtures. The fiber used in this
study was a PVA fiber with a length of 8 mm (0.3 in.) and a
diameter of 39 mm (0.002 in.). The fiber surface was coated
Fig. 1—Particle size distributions of cement, fly ash, slag, and with 1.2% oil by weight to reduce the fiber-matrix chemical
silica sand. and friction bond.10
In all mixtures, the water-cementitious materials ratio (w/cm)
of ECC mixtures incorporating different mineral admix- was kept constant at 0.27. HRWRA was added to the mixture
tures when subjected to mechanical loading and freezing- until the desired fresh ECC characteristics (described in
and-thawing cycles is demonstrated. To find out the frost another study11) were visually observed. Based on this study,
durability of mechanically loaded ECC mixtures, an experi- it is recommended that a w/cm in the range of 0.25 ± 0.05 be
mental program was conducted based on two mineral admix- adopted in an ECC mixture with uniform fiber distribution
ture types (slag [S] and FA), two mineral admixture replace- and a high plastic viscosity (high Marsh cone flow time) and a
ment rates (1.2 and 2.2 FA-S/C ratio), and two preloading low yield stress (high mini-slump flow diameter) be achieved
deformation levels (preloaded to 0 mm [0 in.] (pristine) through adjustment of the amount of HRWRA.11 As seen in
and 2.5 mm [0.1 in.]). The sound and microcracked speci- Table 2, the ECC mixture with 55% slag (S/C ratio of 1.2)

260 ACI Materials Journal/May-June 2013

 Fig. 2—Particle morphology of fly ash and slag.
Table 2—ECC mixture proportions by weight
Mixture ID Cement w/cm Sand FA/C
FA_1.2 1 0.27 0.36 1.2
FA_2.2 1 0.27 0.36 2.2
S_1.2 1 0.27 0.36 — 1.2
S_2.2 1 0.27 0.36 — 2.2
Notes: 1 kg/m3 = 1.6845 lb/yd3; 1 MPa = 0.145 ksi.
had the highest HRWRA demand; however, increasing the
replacement rate of FA by C decreased the required amount
of HRWRA. The smooth surface characteristics and spherical
shape of the FA (Fig. 2) improved the workability character-
istics of ECC mixtures so that similar workability properties
at a constant w/cm were achieved by using a lower HRWRA
content at a higher FA replacement level.12
ECC mixtures were produced by using a mortar mixer.
Solid ingredients were first mixed at 100 revolutions per
minute (rpm) for 1 minute. Water and HRWRA were then
added into the dry mixture and mixed at 150 rpm for 1 minute
and then at 300 rpm for another 2 minutes to produce a
consistent and uniform ECC matrix. PVA fiber was added in
last and mixed at 150 rpm for an additional 3 minutes. After
that, the air content of freshly mixed ECC was measured
by the pressure method in accordance with ASTM C231/
C231M. The compressive strength was determined from the
average compressive strength of the six 50 mm (2 in.) size
cube specimens at the ages of 14 and 28 days. As seen from
Table 2 and as expected, at the same replacement level, the
slag produced significantly higher compressive strengths
than the FA. This finding was partially a result of the higher
rate in hydration and pozzolanic reactions of the slag due
to its large specific surface area (425 m2/kg [229 yd2/lb])
compared to that of FA (290 m2/kg [157 yd2/lb]). A high
surface area provides more nucleating sites and OH– ions
as well as alkalis into the pore fluid. Moreover, as seen
from Table 2, the compressive strength of ECC decreased
with increasing FA or slag content. All mixtures, however,
showed compressive strengths higher than 45 MPa (6.5 ksi)
at the age of 28 days. This value could significantly exceed
that of normal concrete strength (30 MPa [4.3 ksi]) and
fulfill engineering requirements in most projects.
Specimen preparation and testing
From each ECC mixture, fourteen 400 x 100 x 75 mm
(16 x 4 x 3 in.) prisms for the freezing-and-thawing test and
ACI Materials Journal/May-June 2013
Compressive strength, MPa
HRWRA, Fiber,
S/C kg/m3 kg/m3 14 days 28 days
— 4.1 26 43.4 69.3
— 3.7 26 31.1 46.1
5.8 26 62.1 88.5
4.7 26 53.5 76.3
determination of air-void characteristics were prepared. All
specimens were cast in one layer without any compaction,
demolded at the age of 24 hours, and moist-cured in lime-
saturated water at 23°C ± 0.2°C (73.4°F ± 0.4°F) for 13 days.
Fourteen days after casting, from each mixture, six prism
specimens with a span length of 355 mm (14 in.) and a height
of 75 mm (3 in.) were tested under the four-point bending
test up to failure, and the average ultimate flexural strength,
midspan beam deflection capacity, and flexural stiffness
of each mixture were determined. The deflection capacity
was defined as the deflection at which point the bending
stress reaches maximum. At the same age, six prisms were
preloaded up to a 2.5 mm (0.1 in.) deformation level and then
load-released; just after that, three of those preloaded beams
were reloaded up to failure to observe the effect of damage
by preloading. The deflection of 2.5 mm (0.1 in.) is below
the ultimate deflection capacity of both groups of mixtures
(>3.44 mm [>0.135 in.]; refer to Table 3) with no localized
fracture. Tests were conducted on a closed-loop controlled
material test system with 100 kN (22.5 kip) capacity under
displacement control at a rate of 0.005 mm/s (0.0002 in./s).
A linear variable differential transformer (LVDT) was fixed
on the test setup to measure the flexural deflection of the
specimen. The crack widths were measured just after being
load-released by using a video microscope. The widths of
the crack were measured on the tension surface of the beam
specimens along the span length. Then, from each mixture,
three pristine and three preloaded specimens were trans-
ferred into the freezing-and-thawing chamber in accordance
with ASTM C666/C666M, Procedure A, and subjected
to between five and six freezing-and-thawing cycles in a
24-hour period in the unloaded state. Changes in ultrasonic
pulse velocity (UPV) and mass loss were measured at each
interval of nominally 60 cycles up to 300 cycles. The travel
time for the UPV was measured in the longitudinal direction
(normal to the cracks). After completing the 300 freezing-
and-thawing cycles, beam specimens were also loaded up
261
Table 3—Flexural properties of ECC mixtures
Preloading deformation, Midspan deflection, Flexural strength, Residual crack width, Flexural stiffness,
Mixture ID mm mm MPa mm MPa/mm
0.0 5.36 11.06 ~60 11.1
FA_1.2
2.5 3.31 10.58 ~60 8.8
0.0 6.27 9.27 ~40 8.6
FA_2.2
Cured; 14 days of 2.5 4.40 9.11 ~40 7.6
moist curing 0.0 3.44 11.31 ~90 14.8
S_1.2
2.5 1.33 10.88 ~90 10.5
0.0 4.16 10.07 ~70 13.4
S_2.2
2.5 2.18 9.94 ~70 10.0
0.0 4.33 9.63 ~80 10.7
FA_1.2
2.5 3.18 8.53 ~80 8.0
0.0 4.23 7.18 ~80 7.8
After 14 days FA_2.2
of moist curing, 2.5 3.05 6.24 ~80 5.7
subjected to 300
freezing-and-thawing 0.0 3.08 10.86 ~120 14.6
cycles S_1.2
2.5 1.45 9.11 ~120 10.3
0.0 3.55 9.28 ~120 12.7
S_2.2
2.5 2.01 8.29 ~120 9.1
Notes: 1 mm = 0.039 in.; 1 MPa = 0.145 ksi.

to failure under flexural loading and decreases in flexural
properties were determined.
The air-void parameters of ECC mixtures on polished spec-
imens were determined by the Modified Point Count Method
according to ASTM C457/C457M. This measurement is
made on the hardened ECC mounted under a microscope
at 14 days. The air-void parameter is examined by scanning
along a series of traverse lines. To characterize the air voids
and air content, the spacing factor and the specific surface
are used. In general, a high-quality frost-resistant concrete
requires a spacing factor of less than 0.20 mm (0.008 in.) and
a specific surface greater than 25 mm–1 (1 in.–1).13
RESULTS AND DISCUSSIONS
Basic flexural properties
Table 3 presents the flexural properties of pristine ECC
specimens at 14 days. Properties of specimens that were
previously preloaded to a beam deformation level of 2.5 mm
(0.1 in.) specimens are also included in this table. As shown
in Table 3, the deflection capacity of the ECC mixtures is
influenced by the type and amount of mineral admixtures
(slag or FA) and applied preloading conditions. An increase
in both S/C and FA/C improved the deflection capacity of
the ECC mixtures considerably, while a slight reduction in
flexural strength was observed. The higher ductility with
the higher FA or slag content can be attributed to the lower
matrix toughness and the better fiber-matrix interface due
to the addition of more FA or S.4,14-16 When comparing the
deflection capacities of slag- and FA-incorporated ECC
mixtures, it was noticed that slag-ECC mixtures exhib-
ited lower deflection capacity than the FA-ECC mixtures.
Although slag-incorporated ECC mixtures provide smaller
deflection capacity, their deflection capacity was at approx-
imately or greater than 3.44 mm (0.135 in.) at the age
of 14 days. The reduced ductility (reflected by the deflec-
tion capacity measurements) can possibly be caused by
the higher fracture toughness, bond strength, and friction
between the slag-ECC matrix and the fibers compared with
FA-ECC.16 As seen from Table 3, the average ultimate flex-
ural strengths of pristine and preloaded ECC specimens
varied from 9.27 to 11.31 MPa (1.35 to 1.64 ksi) for pristine
specimens and from 9.11 to 10.88 MPa (1.32 to 1.58 ksi)
for preloaded specimens. slag-incorporated ECC specimens
developed somewhat higher flexural strength than the speci-
mens of FA-incorporated ECC mixtures for both replace-
ment levels (55 and 70%) and preloading conditions.
To study the frost durability of preloaded ECC mixtures,
some of the prism specimens were preloaded to a 2.5 mm
(0.1 in.) deformation value at the age of 14 days to achieve
several microcracks. Table 3 summarizes the flexural properties
of ECC specimens that had been precracked to a 2.5 mm
(0.1 in.) deformation level, then unloaded and immediately
reloaded. The flexural stiffness is defined as the secant of the
initial rising branch of the flexural stress-deflection curve,
in which the first point is chosen at 1 MPa (0.145 ksi) and
the second point at approximately 6 MPa (0.870 ksi) for
all mixtures. Between these two stress points, the slope is
almost linear (assumed in elastic stage) for all mixtures. As
seen from Table 3, there is a remarkable difference in the
initial flexural stiffness between the virgin specimen and
the precracked specimen under flexure. This is due to the
reopening of cracks within precracked specimens during
reloading.17 The opening of these cracks offers very little
resistance to load, as the crack simply opens to its previous
crack width before fiber bridging is fully re-engaged. Once
fiber bridging is re-engaged, however, the load capacity
resumes, and further tensile straining of the intact material
can take place. Moreover, as seen in Table 3, an increase in
the FA/C or S/C ratio from 1.2 to 2.2 decreased the slope,
thereby indicating a reduction in stiffness.
By using a microscope, all tested specimens were inspected
to measure the crack width. All crack width measurements
were conducted in the unloaded stage, and the widths of the

262 ACI Materials Journal/May-June 2013

 Fig. 3—Typical cracking patterns of ECC beam specimen after flexure load application.
crack were measured on the tension surface of the beam spec-
imens. With crack width measurement, it was indicated that
the average crack width was lower than 90 mm (0.0035 in.)
on average for slag-ECC mixtures and 70 mm (0.003 in.)
on average for FA-ECC mixtures. It was also observed that
the residual crack width reduces significantly as the S/C or
FA/C ratio increases. Tight and multiple microcracks (less
than 90 mm [0.0035 in.]) instead of one crack with large crack
width are the foremost property of ECC (refer to Fig. 3).
Crack width control is of primary importance for many
reinforced concrete applications because it is believed that
there is a close relationship between the mean or maximum
crack widths and the durability of the structure.4
Air-void characteristics
It is generally recognized that the air-void structure of
cement-based composites is a critical parameter for the
durability of those subjected to frost action.18 The air-
void parameters of the non-air-entrained high-volume
slag- and FA-incorporated ECC mixtures are presented in
Table 4 along with the fresh air content. For each mixture,
only the average values obtained from two specimens are
shown in the table. Although no air-entraining admixture
was added to the ECC mixtures, air contents of these ECC
mixtures in the fresh state as measured by ASTM C231/
C231M gave values between 6.3 and 7.3%, which seemed
to be adequate for freezing-and-thawing resistance. For this
reason, no air entrainment was applied to the ECC specimens
in the freezing-and-thawing cycle tests. As seen in Table 4,
the air content in fresh ECC and hardened ECC revealed
some differences. The total volume of air in hardened
ECC was found to increase approximately 1% compared
to the fresh air content of ECC mixtures. Moreover, the air
contents measured during the fresh and hardened states of
the ECC mixtures showed that increasing the amount of FA
or slag somewhat augmented both the fresh and hardened air
contents. For instance, hardened ECC mixtures with 55%
FA and slag had 7.8% and 7.0% air contents at a hardened
state, respectively; however, these values became margin-
ally high with an increment of FA or slag content to 70% and
reached 8.2% and 7.2%, respectively. These fresh and hard-
ened air contents seemed to be higher than those air contents
of non-air-entrained normal concrete, probably as a result
of the lack of coarse aggregate in ECC mixtures and the
higher viscosity of the mortar matrix during the fresh state;
the fine particles and high viscosity tend to prevent some
of the air bubbles from rising to the surface during placing
operations.6,19 Another reason for the higher air content may
be attributed to the proprietary hydrophobic oiling agent of
ACI Materials Journal/May-June 2013
Table 4—Air-void characteristics of ECC mixtures
FA_ECC slag_ECC
FA_1.2 FA_2.2 S_1.2 S_2.2
Fresh air content, % 6.9 7.3 6.3 6.6
Hardened air content, % 7.8 8.2 7.0 7.2
–1
Specific surface, mm 24.1 35.6 21.7 24.2
Spacing factor, mm 273 197 308 287
Note: 1 mm = 0.039 in.
1.2% by mass coated on the surface of the PVA fibers to
tailor the interfacial properties between the fiber and matrix
for strain-hardening performance. The proprietary coating
chemical used on the surface of fibers may also entrain very
small air bubbles, thus enhancing the freezing-and-thawing
resistance of ECC mixtures. Further studies should be
conducted to understand the reason for the increase in air
content with PVA fiber addition, which is beyond the scope
of this paper.
In the four non-air-entrained ECC mixtures, the specific
surface value fluctuated between 21.7 and 35.6 mm–1 (0.85 and
1.4 in.–1) and the spacing factor ranged between 197 and 308 mm
(0.008 and 0.012 in.). Although the spacing factor values
generally exceed the generally accepted value of 0.200 mm
(0.008 in.) and the specific surface values are less than the
recommended minimum value of 25 mm–1 (0.85 in.–1),20,21 this
lack of an apparently ideal air-void system has not adversely
affected the frost durability of ECC, as indicated in the
following section. As also seen in Table 4, increasing the
S/C or FA/C ratio from 1.2 to 2.2 increased the specific
surface while decreasing the spacing factor values signifi-
cantly. For frost resistance, the degree of hydration and thus
the state of the microstructure at the moment of freezing
is a decisive factor. However, it is well-known that the
degree of hydration of mineral admixtures (especially
FA) incorporated cement-based composites are signifi-
cantly low—especially at early ages—and depend on the
content and quality of mineral admixtures and fineness of
cement.22 Most probably, the degree of hydration of 55%
slag- or FA-incorporated ECC mixtures was faster than
that of the high-volume (70%) slag or FA-ECC mixtures,
and thus the microstructure clearly becomes denser, which
becomes apparent through a decrease in hardened air content
and specific surface.
263
 Fig. 4—Relative mass loss changes as function of number of freezing-and-thawing cycles.
(Note: 1 mm = 0.039 in.)
Fig. 5—UPV change after 300 freezing-and-thawing cycles.
(Note: 1 mm = 0.039 in.)
Properties after being subjected to freezing-and-
thawing cycles
Mass loss and pulse velocity measurement—After being
subjected to the 300 freezing-and-thawing cycles, damage
on pristine and mechanically preloaded ECC specimens was
evaluated by means of mass loss and UPV changes. Even
though the Resonant Frequency Method was suggested by
ASTM C666/C666M for the assessment of internal damage
due to frost action, in this study, the UPV Method was used
due to the lack of resonance frequency measurement equip-
ment. Previous studies have also shown that the UPV Test
Method can also be used to measure the deterioration of
specimens during freezing-and-thawing cycles.23
Cracking in ECC is fundamentally different from that
which occurs in concrete or reinforced concrete. ECC
develops multiple microcracking in a strain-hardening
response as a result of the bending preload (Fig. 3). Due to
multiple microcracking, it is important to test the precracked
ECC under freezing-and-thawing cycles. The mass of loss
versus the number of freezing-and-thawing cycles for
precracked ECC specimens is shown in Fig. 4. The total
mass loss of pristine ECC specimens exposed to freezing-
and-thawing cycles is also included in this figure. It is well-
known that mass variations during freezing-and-thawing
cycles are due to water movements in and out of the speci-
mens, surface scaling, and cracking.24,25 As seen in Fig. 4,
a slight mass gain was observed in all ECC mixtures up
to 60 freezing-and-thawing cycles; however, it was gener-
ally less than 1%. The reason behind the mass gain is prob-
ably due to the absorption of water and formation of reaction
products in voids.26 However, beyond the 60 freezing-and-
thawing cycles, a gradual decrease in mass started, and at
the end of 300 cycles, relative mass loss according to the
264
initial mass of the specimen varied between 0.9 and 9.4%
depending on the preloading condition, mineral admixture
type, and amount. It can also be observed from Fig. 4 that
the total mass loss after 300 cycles tends to increase with
the mechanical preloading. The increase, however, is fairly
insignificant for ECC specimens. This increase is somewhat
more marked for Mixture FA_2.2 (9.4% after 300 cycles).
Figure 5 shows the data for the relative UPV change
(100 × Vi/V0) due to the freezing-and-thawing action. The
variable Vi is the UPV after a certain number of freezing-
and-thawing cycles and V0 is the initial pulse velocity prior
to any freezing-and-thawing cycles. As seen from Fig. 5,
the amount of deterioration caused by frost action in the
precracked ECC specimens was higher than that of the pris-
tine specimens. Moreover, the higher amount of slag and FA
in the mixtures attributed to the increase in the amount of
damage of ECC (higher reduction in UPV) mixtures as a
result of frost action. As in the mass loss observation, this
effect is more pronounced in the case of FA-ECC mixtures.
As seen in the figure, the highest decrease in the UPV value
for both pristine and preloaded specimens was obtained in
the case of the ECC mixture with 70% FA content (FA_2.2).
Although FA-ECC mixtures show the highest deflection
capacity and lowest residual crack width before freezing-
and-thawing cycles (refer to Table 3) due to the four-point
bending test, the highest variations in both mass loss and
relative UPV after frost action were also observed on these
mixtures. Moreover, Mixture FA_2.2 had more deteriora-
tion than Mixture FA_1.2, although Mixture FA_2.2 had a
higher fresh air content (7.3%) and lower spacing factor than
Mixture FA_1.2 (6.9%). FA-ECC specimens also exhibited
some surface scaling at the conclusion of the freezing-and-
thawing cycling, and increased surface scaling was observed
on the preloaded specimens. Therefore, it is not possible to
determine the critical air content and spacing factor for ECC
because no relation was found between frost durability and
air-void characteristics of ECC mixtures. The scaling was,
however, clearly confined to the surface layers of the test
specimens and had no effect on the integrity of the ECC mass.
On the other hand, almost no surface scaling (almost no mass
loss) was observed on both pristine and preloaded slag-ECC
specimens after 300 cycles of freezing and thawing. More-
over, although slag-ECC mixtures had lower ductility and
air contents and higher spacing factors, at 300 cycles, they
had a higher average relative UPV and lower mass loss value
than FA-ECC mixtures at the same number of freezing-and-
thawing cycles. This was probably due to the greater matu-
rity of the slag-ECC test specimens. When the compressive
and flexural strengths before freezing-and-thawing cycling
(refer to Tables 2 and 3) and mass loss (refer to Fig. 4) of
the ECC mixtures due to the freezing-and-thawing cycling
ACI Materials Journal/May-June 2013
 Fig. 6—Variation in flexural parameters of ECC mixtures due to freezing-and-thawing cycles
(F-T). (Note: 1 MPa = 0.145 ksi; 1 mm = 0.039 in.)
are evaluated together, it can be concluded that there is a
powerful relation among them; high strength (especially
compressive strength) leads to a lower mass loss due to frost
action. This result is in agreement with the findings of Sun
et al.,27 in which different strength grades of concrete under
the simultaneous action of load and freezing-and-thawing
cycles were studied. The authors concluded that at the same
stress ratio, concrete of higher strength could undertake
more freezing-and-thawing cycles, and the dynamic elastic
modulus decreased more slowly with freezing-and-thawing
cycles. Therefore, FA-ECC needs sufficient time (maturity)
to develop beneficial characteristics. Curing to only 14 days
in accordance with ASTM C666/C666M prior to the first
exposure to freezing-and-thawing cycles may have been
insufficient for the hydration and self-desiccation to reduce
the quantity of freezable water sufficiently in the case of
FA-ECC mixtures. The benefits of using Class F FA in ECC
in terms of improved durability properties such as frost is
usually manifested at later ages with the continuous supply
of moisture.28
Apart from the slight increase in mass loss, the results
presented in this study largely confirm the durability
performance of high-volume FA/slag-ECC mixtures under
freezing-and-thawing cycles, even in cases where the mate-
rial experiences mechanical preloading that deforms it into
the strain-hardening stage prior to exposure. This high frost
durability can be attributed to the ECC’s superior tensile
properties and inherent air-void parameters. It is well-known
that upon freezing, water in capillary pores expands. If the
required volume is greater than the space available, the pres-
sure buildup could reach the tensile strength of the material,
resulting in local microcrack formation, brittle rupture, and
scaling. Therefore, the high tensile strength and ductility—
and particularly fracture resistance in ECC—could lead to
its improved frost-resistance characteristics.6
Flexural performance—Table 3 shows the test results of
pristine and preloaded ECC specimens in terms of ultimate
deflection capacity, flexural strength, stiffness, and residual
crack width after being subjected to 300 freezing-and-
thawing cycles. Each reading represents the average value of
three specimens. As seen from Table 3, it is noticeable that
both of the flexural strength and deflection capacities of all
ECC mixtures reduced due to the 300 freezing-and-thawing
cycles; however, the impression of frost action is consid-
erably low. After being subjected to freezing-and-thawing
cycles, under flexural loading, all ECC specimens demon-
strate strain-hardening behavior and multiple cracking.
As in the non-frost-deteriorated specimens, the first crack
started inside the midspan at the tensile face. The flexural
stress increased at a slower rate, along with the development
ACI Materials Journal/May-June 2013
of multiple cracks with small crack spacing and tight crack
widths. Microcracks developed from the first cracking point
and spread out in the midspan of the flexural beam. Bending
failure in the ECC occurred when the fiber-bridging strength
at one of the microcracks was reached, resulting in localized
deformation once the modulus of rupture was approached.
Figure 6 demonstrates the variation of flexural strength
and deflection capacity of ECC mixtures before and after
being subjected to frost action. As seen in the figure,
contrary to expectations, the drops in flexural strength and
deflection values for the pristine and preloaded specimens
were not distinctive from each other; mechanical preloading
of the ECC specimens did not significantly influence the
residual mechanical properties of the frost-deteriorated ECC
specimens. The percentage decrease in the deflection and
flexural strength values for the pristine and preloaded speci-
mens due to the frost action was immediate. For instance, the
percentage decreases in the deflection values of pristine and
preloaded FA_2.2 specimens due to frost action were 32.5%
and 30.6%, respectively, and the percentage decreases in
the flexural strength of pristine and preloaded specimens for
the same mixture were 13.0% and 19.4%, respectively. This
result suggests that between the time of inducing precracking
and the time of testing, after freezing-and-thawing cycles,
slight healing of the microcracks also occurred in the ECC
specimens. This can be attributed primarily to the high
cementitious material content and relatively low water-
binder ratio within the ECC mixture. As a result of the
formation of microcracks due to mechanical loading, unhy-
drated cementitious particles are easily exposed to the water
during the thawing period, which leads to the development
of further hydration processes. Finally, microcracks under
conditions of a damp environment were partially healed by
newly formed products. Therefore, the self-healing phenom-
enon is only valid for the precracked specimens. These
observations are in good agreement with those discussed
in the literature.29,30 These investigations indicated that the
formation of rehydration products in microcracks is possible
during freezing-and-thawing cycles. In ECC, the rehealing
process is especially aided by the innately tight crack width.
Table 3 summarizes the average deflection capacity of
ECC specimens exposed to 300 freezing-and-thawing
cycles. The deflection capacity reported for these specimens
does not include the residual deflection from the precracking
deformation. By neglecting this residual deflection, the
large variability in material relaxation during unloading is
avoided, and a conservative estimation for ultimate defor-
mation capacity of the material is presented. The deflection
capacity of virgin and precracked ECC specimens exposed
to 300 freezing-and-thawing cycles ranges from 1.45 to
265

Fig. 7—Variation of relative flexural stiffness due to
freezing-and-thawing cycles and mechanical preloading.
4.33 mm (0.06 to 0.17 in.). The effect of preloading on the
deflection capacities before freezing-and-thawing cycles
was more prominent than the effect of frost action. For
instance, Mixture FA_2.2 had 6.27 and 4.40 mm (0.25 and
0.18 in.) midspan beam deflection capacities for the pris-
tine and 2.5 mm (0.1 in.) preloaded specimens at the age
of 14 days; however, after 300 freezing-and-thawing cycles,
the 6.27 mm (0.25 in.) deflection capacity of the pristine
specimen was decreased to 4.23 mm (0.17 in.). In the case of
flexural strength, as a result of 2.5 mm (0.1 in.) preloading,
the average flexural strength of pristine specimens decreased
from 9.27 to 9.11 MPa (1.35 to 1.32 ksi) for the same mixture,
but after being subjected to 300 freezing-and-thawing cycles,
the average flexural strength of the pristine specimens
dropped to 7.18 MPa (1.04 ksi). The reduction in flexural
strength was approximately 2% due to the preloading and
23% due to 300 freezing-and-thawing cycles for Mixture
FA_2.2. This can be attributed to the effects of damage on the
fiber-matrix interface and matrix microcracking as a result
of frost deterioration.31 As expected, when the dual effect
of preloading and freezing-and-thawing cycles was exam-
ined, the flexural strength of Mixture FA_2.2 decreased from
9.27 MPa (1.35 ksi) (average flexural strength of pristine
specimens before freezing-and-thawing cycles) to 6.24 MPa
(0.9 ksi) (average flexural strength of preloaded specimens
after freezing-and-thawing cycles). The rate of reduction
was approximately 33%.
As mentioned previously, the slope of the initial part of
the load-deflection curve represents the flexural stiffness
of the beams. Table 3 shows that the slope decreases with
the frost deterioration, thereby indicating a reduction in the
stiffness of the ECC mixtures. Figure 7 shows the reduc-
tion in the relative stiffness of both pristine and preloaded
specimens due to frost action. In this figure, the stiffness
value of pristine specimens at the age of 14 days (just before
freezing-and-thawing cycling) is accepted as 100%. As seen
in this figure, the reduction in relative stiffness due to frost
action was slight (not more than 9% at extreme deteriora-
tion) and at a tolerable range for pristine specimens but
was detrimental for preloaded specimens. As in the other
measured properties, the highest reduction was monitored
in a high-volume-FA-incorporated ECC mixture (FA_2.2),
followed by a high-volume-slag-incorporated ECC mixture
(S_2.2). The relative stiffnesses of the preloaded specimens
after 300 freezing-and-thawing cycles were very close to
each other at approximately 70%. These results revealed that
the presence of microcracks due to mechanical stresses or
266
time-dependent effects (for example, shrinkage and creep)
is one of the major factors that can influence durability
and serviceability of cement-based composites in terms of
freezing-and-thawing cycles.30
Table 3 shows the residual crack width of ECC mixtures
after being subjected to the 300 freezing-and-thawing cycles.
The residual crack width of virgin and precracked specimens
exposed to 300 freezing-and-thawing cycles is wider than
that of non-frost-deteriorated virgin specimens but remains
less than 120 mm (0.005 in.). As in the non-frost-deterio-
rated ECC specimens’ residual crack width, slag-incorpo-
rated ECC mixtures—after being subjected to freezing-and-
thawing cycles—had wider residual crack widths than the
FA-incorporated ECC mixtures. Before exposure to frost
deterioration, although a clear diversity of the residual crack
width was monitored by the increase of the mineral admix-
ture replacement rate, it disappeared after being subjected
to freezing-and-thawing cycles. In other words, the residual
crack width of the preloaded and pristine specimens became
equal for the different replacement rates of FA or slag, and
the beneficial effect of high-volume slag or FA usage on
residual crack width dissipated. More experimental studies
on a micromechanical scale are necessary to clearly under-
stand the mechanisms behind the increased crack width and
are beyond the scope of this paper.
It should be noted that the flexural properties of frost-dete-
riorated ECC beams provide a conservative estimate of their
residual flexural properties in actual structures. These accel-
erated deterioration periods are equivalent to a time span of
many years in real structures, even those located in regions
with harsh winters. This difference in accelerated and
normal frost deterioration periods should have a significant
influence on the residual flexural properties of ECC because,
in the long term, deterioration in ECC as a result of freezing-
and-thawing cycles can more effectively be closed due to
a self-healing process.5 Thus, the flexural performances of
ECC summarized in Table 3 are underestimated.
CONCLUSIONS
In this study, the frost durability of pristine (sound) and
preloaded high-volume slag- and FA-incorporated ECC
mixtures was investigated. ECC mixtures with two different
amounts of slag and FA as a replacement of cement (55 and
70% by weight of total cementitious material) were prepared.
To find out the effect of mechanical preloading on the frost
durability of ECC, prism specimens were preloaded up to a
2.5 mm (0.1 in.) deformation level under four-point bending
load to generate microcracks. Then, the preloaded and
pristine (sound) specimens were subjected to the freezing-
and-thawing test in accordance with ASTM C666/C666M.
Experimental tests consisted of measuring the change in
mass and UPV and residual flexural properties of ECC
mixtures exposed to the freezing-and-thawing cycles. The
air-void characteristics of mixtures were also studied. From
the results obtained, the following conclusions can be drawn:
• ECC specimens exposed to freezing-and-thawing cycles
show decreases in UPV, flexural strength, deformation
capacity, and stiffness, and an increase in the mass loss
compared with the non-frost-deteriorated specimens.
Furthermore, the crack width increases to approxi-
mately 1.5 times compared with that of air-cured speci-
mens. This phenomenon suggests a possible change in
the fiber-matrix interface bond properties after acceler-
ated freezing-and-thawing cycles.
ACI Materials Journal/May-June 2013
• For the preloaded ECC specimens, the applied
preloading deformation was found to have a detri-
mental effect on the frost resistance of ECC mixtures.
This effect is more pronounced in the case of FA-ECC
mixtures. Moreover, an increase in FA content in ECC
production can further exacerbate the deterioration
caused by freezing-and-thawing cycles.
• Unlike conventional concrete, the air-void parameters
are less important for the frost resistance of ECC. In
the case of ECC, the presence of micro PVA fibers
and superior ductility appears to be responsible for
enhancing the frost resistance, as revealed in this series
of tests.
In conclusion, apart from the reductions in UPV, ultimate
midspan beam deformation, flexural strength and stiffness,
and higher mass loss and residual crack width, the results
presented in this study largely confirm the durability perfor-
mance of ECC material under accelerated freezing-and-
thawing cycles, even in cases where the material experiences
mechanical loading that deforms it into the strain-hardening
stage prior to exposure.
ACKNOWLEDGMENTS
The authors gratefully acknowledge the financial assistance of the Scien-
tific and Technical Research Council (TUBITAK) of Turkey provided
under Project MAG-108M495, Gaziantep University Scientific Research
Centre provided under Project MF.10.09, and The Council of Higher
Education of Turkey.
REFERENCES
1. Miao, C.; Mu, R.; Tian, Q.; and Su, W., “Effect of Sulfate Solution on
the Frost Resistance of Concrete with and without Steel Fiber Reinforce-
ment,” Cement and Concrete Research, V. 32, 2002, pp. 31-34.
2. Li, V. C., “On Engineered Cementitious Composites (ECC)—A
Review of the Material and Its Applications,” Advanced Concrete Tech-
nology, V. 1, 2003, pp. 215-230.
3. Li, V. C.; Wang, S.; and Wu, C., “Tensile Strain-Hardening Behavior
of Polyvinyl Alcohol Engineered Cementitious Composite (PVA-ECC),”
ACI Materials Journal, V. 98, No. 6, Nov.-Dec. 2001, pp. 483-492.
4. Şahmaran, M., and Li, V. C., “Durability Properties of Micro-Cracked
ECC Containing High Volumes Fly Ash,” Cement and Concrete Research,
V. 39, 2009, pp. 1033-1043.
5. Şahmaran, M., and Li, V. C., “De-Icing Salt Scaling Resistance of
Mechanically Loaded Engineered Cementitious Composites,” Cement and
Concrete Research, V. 37, 2007, pp. 1035-1046.
6. Şahmaran, M.; Lachemi, M.; and Li, V. C., “Assessing the Durability
of Engineered Cementitious Composites under Freezing and Thawing
Cycles,” Journal of ASTM International, V. 6, 2009, pp. 1-6.
7. Li, V. C.; Fischer, G.; Kim, Y. Y.; Lepech, M.; Qian, S.; Weimann, M.;
and Wang, S., “Field Demonstration of Durable Link Slabs for Jointless
Bridge Decks Based on Strain-Hardening Cementitious Composites,”
Report for Michigan Department of Transportation RC-1438, 2003, 147 pp.
8. Şahmaran, M.; Li, V. C.; and Andrade, C., “Corrosion Resistance
Performance of Steel-Reinforced Engineered Cementitious Composite
Beams,” ACI Materials Journal, V. 105, No. 3, May-June 2008, pp. 243-250.
9. Mu, R.; Miao, C.; Luo, X.; and Sun, W., “Interaction between
Loading, Freeze-Thaw Cycles, and Chloride Salt Attack of Concrete with
and without Steel Fiber Reinforcement,” Cement and Concrete Research,
V. 32, 2002, pp. 1061-1066.
10. Li, V. C.; Wu, C.; Wang, S.; Ogawa, A.; and Saito, T., “Interface
Tailoring for Strain-Hardening Polyvinyl Alcohol-Engineered Cementi-
tious Composite (PVA-ECC),” ACI Materials Journal, V. 99, No. 5, Sept.-
Oct. 2002, pp. 463-472.
11. Yang, E.; Şahmaran, M.; Yang, Y.; and Li, V. C., “Rheological
Control in the Production of Engineered Cementitious Composites,” ACI
Materials Journal, V. 106, No. 4, July-Aug. 2009, pp. 357-366.
ACI Materials Journal/May-June 2013
12. Şahmaran, M.; Keskin, S. B.; Ozerkan, G.; and Yaman, I. O., “Self-
Healing of Mechanically-Loaded Self-Consolidating Concretes with High
Volumes of Fly Ash,” Cement and Concrete Composites, V. 30, 2008,
pp. 872-879.
13. Siebel, E., “Air-Void Characteristics and Freezing and Thawing
Resistance of Superplasticized Air-Entrained Concrete with High Work-
ability,” Superplasticizers and Other Chemical Admixtures in Concrete,
SP-119, V. M. Malhotra, ed., American Concrete Institute, Farmington
Hills, MI, 1989, pp. 297-320.
14. Wang, S., and Li, V. C., “Engineered Cementitious Composites with
High-Volume Fly Ash,” ACI Materials Journal, V. 104, No. 3, May-June
2007, pp. 233-241.
15. Zhou, J.; Qian, S.; Beltran, M. G. S.; Ye, G.; van Breugel, K.; and Li,
V. C., “Development of Engineered Cementitious Composites with Lime-
stone Powder and Blast Furnace Slag,” Materials and Structures, V. 43,
2010, pp. 803-814.
16. Şahmaran, M.; Yucel, H. E.; Demirhan, S.; Arik, M. T.; and Li, V. C.,
“Combined Effect of Aggregate and Mineral Admixtures on Tensile Ductility
of Engineered Cementitious Composites,” ACI Materials Journal, V. 109,
No. 6, Nov.-Dec. 2012, pp. 627-638.
17. Yang, Y.; Lepech, M. D.; and Li, V. C., “Autogenous Healing of
Engineering Cementitious Composites under Wet-Dry Cycles,” Cement
and Concrete Research, V. 39, 2009, pp. 382-390.
18. Elsen, J., “Automated Air Void Analysis on Hardened Concrete
Results of a European Inter Comparison Testing Program,” Cement and
Concrete Research, V. 31, 2001, pp. 1027-1031.
19. Powers, T. C., “Frost Resistant Concrete,” Journal of the PCA Devel-
opment Laboratories, V. 6, 1964, p. 19.
20. Powers, T. C., “Topics in Concrete Technology 3: Mixtures
Containing Intentionally Entrained Air,” Journal of the PCA Development
Laboratories, V. 6, 1964, pp. 19-42.
21. Powers, T. C., “Topics in Concrete Technology 3: Characteristics of
Air-Void Systems,” Journal of the PCA Development Laboratories, V. 7,
1965, pp. 23-41.
22. Stark, J., and Ludwig, H. M., “Freeze-Thaw and Freeze-Deicing
Salt Resistance of Concretes Containing Cement Rich in Granulated Blast
Furnace Slag,” ACI Materials Journal, V. 94, No. 1, Jan.-Feb. 1997,
pp. 47-55.
23. Cohen, M. D.; Zhou, Y.; and Dolch, W. L., “Non-Air-Entrained
High-Strength Concrete—Is it Frost Resistant?” ACI Materials Journal,
V. 89, No. 4, July-Aug. 1992, pp. 406-415.
24. Pigeon, M.; Gagne, R.; and Foy, C., “Critical Air Void Spacing
Factors for Low Water-Cement Ratio Concretes with and without Silica
Condensed Silica Fume,” Cement and Concrete Research, V. 17, 1987,
pp. 896-906.
25. Cohen, M. D.; Olek, J.; and Dolch, W. L., “Mechanism of Plastic
Shrinkage Cracking in Portland Cement and Portland Cement-Silica
Fume Paste and Mortar,” Cement and Concrete Research, V. 20, 1990,
pp. 103-119.
26. Duchesne, J., and Bérubét, M. A., “Effect of Supplementary
Cementing Materials on the Composition of Cement Hydration Products,”
Advanced Cement Based Materials, V. 2, 1995, pp. 42-53.
27. Sun, W.; Zhang, Y. M.; Yan, H. D.; and Mu, R., “Damage and
Damage Resistance of High Strength Concrete under the Action of Load
and Freeze-Thaw Cycles,” Cement and Concrete Research, V. 29, 1999,
pp. 1519-1523.
28. Şahmaran, M.; Yaman, İ. Ö.; and Tokyay, M., “Development of High
Volume Low-Lime and High-Lime Fly-Ash-Incorporated Self-Consoli-
dating Concrete,” Magazine of Concrete Research, V. 59, 2007, pp. 97-106.
29. Loukili, A.; Richard, P.; and Lamirault, J., “A Study on Delayed
Deformations of an Ultra High Strength Cementitious Material,” Fourth
CANMET/ACI/JCI Conference: Advances in Concrete Technology, SP-179,
V. M. Malhotra, ed., American Concrete Institute, Farmington Hills, MI,
1998, pp. 929-949.
30. Granger, S.; Loukili, A.; Pijaudier-Cabot, G.; and Chanvillard, G.,
“Experimental Characterization of the Self-Healing of Cracks in an Ultra-
High Performance Cementitious Material: Mechanical Tests and Acoustic
Emission Analysis,” Cement and Concrete Research, V. 37, 2007,
pp. 519-527.
31. Şahmaran, M.; Özbay, E.; Yücel, H. E.; Lachemi, M.; and Li, V. C.,
“Frost Resistance and Microstructure of Engineered Cementitious Composites:
Influence of Fly Ash and Micro Poly-Vinyl-Alcohol Fiber,” Cement and
Concrete Composites, V. 34, 2012, pp. 156-165.
267
 ACI MATERIALS JOURNAL
Title no. 110-M22
Evaluation of Concrete Drying Shrinkage Related
to Moisture Loss
by Reza Abbasnia, Mohammad Shekarchi, and Jamal Ahmadi
Drying shrinkage, defined as the volumetric change of concrete
induced by moisture loss, can change characteristics of concrete
such as durability, stress distribution, and deformation.
Considering the importance of shrinkage effects and the absence
of any comprehensive method to estimate shrinkage value, this
paper attempts to integrate numerical and experimental methods
to predict shrinkage strain based on internal moisture loss.
The hypothesis of volume changes being proportional to internal
moisture loss is used to evaluate the shrinkage strain, where the
proportionality coefficient, called the “shrinkage factor,” is to be
determined experimentally and is a function of the material proper-
ties and humidity conditions.
Results show that the proposed method has acceptable accu-
racy for estimating the shrinkage and evaluating the shrinkage
strain distribution.
Keywords: adsorption; diffusion; drying; humidity; shrinkage; shrinkage factor.
INTRODUCTION
Widespread use of concrete justifies the importance of
studying various features of this material. Specifically,
studying the durability of concrete, which is directly affected
by the shrinkage, is of particular interest because of its
significant effect on the service life of concrete members.
The effects of concrete shrinkage go beyond the concrete
durability, as in some codes (AASHTO) stresses induced by
shrinkage strain are identified as new loads, which should be
considered in addition to structural loads for the design of
concrete members.
Although concrete shrinkage has been the main subject
of many studies, due to the complexity of the problem,
researchers have not been able to provide an accurate method
for prediction of this phenomenon.
In previous studies, beside the detection of different mech-
anisms (capillary tension, disjoining pressure, and surface
tension), which govern shrinkage creation in a certain range
of humidity, the effects of various factors, such as envi-
ronmental conditions, specimen dimensions and geometry,
material mixture properties, additives effect, and curing
conditions,1-5 have been studied on shrinkage.
Increasing information about the effects of different
factors on the shrinkage has provided enough knowledge to
develop initial models, such as BP,6 for shrinkage predic-
tion. Further accumulation of empirical data has evolved
these models and enabled them to provide more precise
prediction of the shrinkage. In the B3,6 ACI 209R-92,7 and
CEB-FIP MC908 models, the shrinkage strain is calculated
as a function of environmental conditions, concrete proper-
ties, and geometry of the members.
Despite the evolution of shrinkage prediction models,
the predictions of these models are not reliable because of
the significant difference between observed and calculated
results; inability of models to predict strain distribution;
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
negligence interaction of parameters and basic factor effects
(that is, surface features of aggregate); and also the signifi-
cant limitations of models.9
Alongside the development of shrinkage prediction
models, other methods were developed based on different
principles and assumptions to determine concrete’s volume
changes. The determination of concrete volume changes
using internal moisture loss is among these methods. The
main features of this method, such as correlation of shrinkage
phenomenon to concrete microstructure and the ability to
evaluate shrinkage strain distribution, has made this method
more effective. However, the difficulty and complexity
of moisture flow determination in concrete and also the
unknown relation between moisture and volume changes are
the main obstacles for development of this method.
As mentioned previously, the determination of the mois-
ture flow in concrete is a fundamental step in this method
that has been studied since the 1930s.10 In that era, the main
assumption about this phenomenon was the independence of
the moisture flow rate (diffusion coefficient) from the mois-
ture content in the absence of active ions.11,12
This point of view was initially challenged by Carlson10
and then Pickett.13 Following that, Pihlajavaara and Vaisa-
nens’s14 observation showed a strong dependency of the
moisture flow rate to the moisture content. Eventually,
this relation was adopted by the scientific community after
Bažant and Najar’s11,12 seminal papers were presented.
These papers initiated an extensive study of water transmis-
sion in concrete, which affects many properties of concrete,
including electrical properties and active ion penetration.
Despite these efforts, the relation between concrete moisture
loss and volume changes has not been studied enough.
RESEARCH SIGNIFICANCE
The main objective of this paper is to develop a method
for the evaluation of the value and distribution of the
shrinkage strain caused by concrete desiccation. This study
is composed of two main steps: first, determination of the
internal concrete moisture changes (using the Finite Element
Analysis Method and using modified SDB sensors [Shek-
archi, Debicki, Billard moisture meter]15) and secondly,
development of the relationship between these changes and
shrinkage strain.
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-263.R2 received August 4, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
269
Reza Abbasnia is an Associate Professor at the Iran University of Science and Tech-
nology, Tehran, Iran. He received his BS, MS, and PhD in civil engineering from
Southern Illinois University, Carbondale, IL, in 1983, 1987, and 1992, respectively.
His research interests include nonlinear analysis of structures, rehabilitation of
concrete structures by fiber-reinforced polymer, and rehabilitation of structures and
performance-based design.
ACI member Mohammad Shekarchi is an Associate Professor at the University
of Tehran, Tehran, Iran. He received his MS and PhD in civil engineering from the
University of Tehran and Institut National des Sciences Appliquées de Lyon, Villeur-
banne cedex, France, in 1985 and 1999, respectively. His research interests include
shrinkage behavior of concrete, use of impact echo as a nondestructive test method in
concrete, study of alkali-aggregate reaction in concrete, self-consolidating concrete
characteristics, and lightweight aggregate concrete.
Jamal Ahmadi is an Assistant Professor at Zanjan University, Zanjan, Iran. He
received his BS, MS, and PhD in civil engineering from Zanjan University and the
Iran University of Science and Technology in 2000, 2003, and 2009, respectively. His
research interests include shrinkage behavior of concrete and concrete technology,
durability of concrete, and concrete structure repair.
Table 1—Functional forms of relations between
moisture loss and volume changes of concrete
Author Proposed relation
Witasse et al.23 De = asDhI
Kim and Lee26 De sh (t , t0 ) = ksh fs (h ) = ksh (1 − h ) and ksh = e 0s gs (t )
Hyo et al.27 esh = esh∞{fs(hn + 1) – fs(hn)}
MOISTURE FLOW IN CONCRETE
The transfer of water molecules occurs in concrete and
other porous materials where a driving force, such as a
concentration gradient or an external hydrostatic pressure,
is present.
Water molecules’ transfer due to a concentration gradient
is divided into vapor transfer and capillary transfer. Vapor
transfer occurs in non-water-filled pores and is further
divided into condense diffusion, surface diffusion, and capil-
lary diffusion.16 Diffusion is a net flow of molecules due to
random motion from a region with higher concentration to a
region with lower concentration.
Condense diffusion is diffusion in very small pores where
the water molecules’ and pore walls’ contact (character-
istics of contact area) have a significant influence on the
rate of diffusion. Surface diffusion is the transfer of bound
water on the structure of materials and capillary diffusion
occurs in water-filled pores due to pressure differences in
water menisci.
In the conducted researches in this area, there has been
agreement on the use of nonlinear diffusion equations for
macroscopic moisture transfer. In the macroscopic mois-
ture transfer, the diffusion coefficient is used as a coef-
ficient, including all thermodynamic mechanisms of fluid
flow within the porous materials. Although the use of such
a general factor has obstacles, because of the exact process,
the determination involves the complex interaction between
fluid and gas phases within the pore system.
In this paper, Eq. (1), which is the differential form of the
modified Fick’s second law, is used to determine the mois-
ture loss in isotherm conditions (ambient temperature)17
∂h
= div Cgrad (h )
∂t
270
The diffusion coefficient is defined as a function of initial
diffusion coefficient C1, relative humidity (F1(h)), tempera-
ture variation (F2(T)), and development of hydration process
(F3(te)), as shown in Eq. (2).
C = C1 F1 (h ) F2 (T ) F3 (te ) (2)
In this study, a finite element code has been developed
for simulation of two-dimensional (2-D) moisture diffu-
sion through the concrete. This program uses the Galer-
kin’s Weighted Residual Method18 (Eq. (3)) to estimate the
moisture changes in the concrete. The Galerkin’s Weighted
Residual Method requires that
∫ Wi ( x ) R ( x ) dx = 0
(3)
∂h
where R( x ) = − div [Cgrad(h)]
∂t
where Wi(x) is the weighting function. In the Galerkin’s
Weighted Residual Method, the residual R(x) is multiplied
by a weighting function (Wi(x)), and the integral of the
product is required to be 0.
RELATION BETWEEN VOLUME CHANGES AND
INTERNAL MOISTURE LOSS
Although the exact relation between moisture loss and
volume changes of concrete is not known, most past studies
have assumed a linear relation (proportionality hypothesis)
between these factors. For example, Baroghel et al.19 used
the proportionality hypothesis for shrinkage estimation in
their studies about the construction of nuclear power plants.
According to Neville,20 changes in the volume of the
concrete are not equal to the volume of the lost water
but just are proportional to it. In this regard, Ferretti and
Bažant21 introduced Eq. (4) as the functional relation
·
between shrinkage strain and the rate of moisture loss h (ksh
is a constant in Eq. (4))
e sh = ksh h (4)
Benboudjema et al.22 also used a similar relation in the 50 to
100% relative humidity (RH) range (refer to Eq. (5))

e ds = kds hI (5)
where I is the unit matrix.
Researchers such as Witasse et al.,23 Heum and Jason,24
Zhen et al.,25 Kim and Lee,26 Hyo et al.,27 and Zhengwu et
al.28 used the same hypothesis in their studies. The functional
forms of the relations that were used in these references are
summarized in Table 1.
Experimental verification of this hypothesis was conducted
by Hak29 through locating strain gauges and moisture sensors
inside the concrete.
Note that there are other methods with different assump-
tions for predicting drying shrinkage from moisture losses.
For example, the Munich Model30 relates the concrete
volume changes to the changes in the surface energy and
(1) moisture content (for moisture values less than 40% in
colloidal material) through the Bangham-Gibbs equation.
ACI Materials Journal/May-June 2013
 Laplace and Kelvin equations were suggested to calculate
the curvature of the water layer, which affects the volume
changes of the concrete, as a function of the internal moisture
loss.2 These equations are applicable to the humidity condi-
tions where the capillary tension is a dominant mechanism.19
Mackenzie31 established a relationship between volume
changes and changes in hydrostatic pressure within the
capillary pores by combining Kelvin and Laplace equations.
Bentz et al.32 modified the Mackenzie31 relation for unsatu-
rated materials.
Baroghel et al.19 established a direct relationship between
the volumetric strain and changes of internal moisture by
the combination of Kelvin-Laplace equations with the elas-
ticity laws.
Among the other conducted researches in this area, the
Bažant1 investigation pointed out that the mean strain
changes in any cross section of specimens is proportional Fig. 1—Concrete’s volume change due to moisture loss.
to the average amount of moisture in the concrete pores.
However, despite considerable researches in this regard,
the overall relation for determination of the shrinkage
strain and its distribution with internal moisture loss is not
completely established.
However, the model that is used in this paper is based on
the proportionality between volume and moisture changes,
as shown in Eq. (6). The proportionality coefficient or
shrinkage factor (k is the microstrain/RH%) is a function of
material properties and concrete moisture content

e ∝ dh 
→ e = k dh, dh = (100 − he ) (6)
In Eq. (6), k represents the volume change in concrete
caused by one unit of moisture loss. Because water exists in
different phases of the concrete, it is clear that the effect of
moisture loss on the volume changes depends on the phase
of the water donor. In other words, the volume changes
caused by moisture loss from the macropores, capillary
pores, or from the absorbed water on the calcium silicate
hydrate (C-S-H) particles will be different.
At the beginning of the drying process, water evapo-
rates from the large pores of concrete without any signif-
icant effect on the volume of the concrete. Therefore, the
shrinkage factor is not considerable at this stage.
With further reduction of moisture, large pores become
empty and thermodynamic imbalance induces a water
gradient from the capillary pores and adsorbed water layers
on the C-S-H particles toward the free surface of concrete.
Evaporation of this water causes considerable tensile stress
in the wall of the pores. This tensile stress causes shrinkage
in the volume of concrete, which is far more than the free
water withdrawal from the large pores.
These two mechanisms work simultaneously and act on
the overall volume of concrete, which includes the capillary
pores’ network and external absorbed water layer on C-S-H
particles; thus, the shrinkage factor has its highest value in
the range of 40 to 95% RH. When the RH reaches approxi-
mately 40%, the value of the shrinkage factor reduces due to
the elimination of the capillary pore phase effect (the diffu-
sion coefficient reaches a constant value of approximately
60% RH11).
At an RH lower than 40%, the effect of the outer layers of
absorbed water molecules on the surface of C-S-H particles
becomes less significant. In this phase, because of the elim-
ination of initially absorbed water layers and the contrac-
Fig. 2—Volume change of specimen due to moisture loss
along x-axis.
tion of the C-S-H particle surface, maximum deformation is
expected. But in practice, because this mechanism acts only
on C-S-H particles, the k factor is less than its value in the
previous humidity range.
Therefore, it is necessary to determine the value of the
shrinkage factor in a wide range of humidity to accurately
predict the concrete’s volume changes.
The concrete specimen shown in Fig. 1 is used to deter-
mine the relation between the concrete’s volume changes
and moisture loss under the isotherm condition. After
achieving moisture equilibrium in all parts of the specimen,
the volume of the specimen changes from V = A × B × C to
V′ = a × b × c (refer to Fig. 1). The ratio of volume changes
to initial volume is given by Eq. (7),33 where ex, ey, and ez are
shrinkage strains along the main axes.
DV
V
(
= ex + ey + ez ) (7)
Therefore, using Eq. (7), volume changes of the specimen
can be obtained by calculating the shrinkage strain along the
main axes.
In the following, the relation between volume changes and
moisture loss is determined for one-dimensional (1-D) mois-
ture loss, and then the method is generalized to 2-D problems.
Figure 2 shows a concrete specimen that was located in
the saturated condition (h1 = 100). The specimen is then
placed in a new environment with the same temperature but
different humidity h2 (h2 < h1). The specimen is kept in the

ACI Materials Journal/May-June 2013 271

 The specimen’s length variation and shrinkage strain
along the x-axis are given by Eq. (9)

(
da = k1dh1 a1 + k2 dh2 a2 ⇒ ) (9)
(
e x = da A = k1dh1 a1 + k2 dh2 a2 ) A

Therefore, the shrinkage strain for a specimen with n

moisture contour lines is obtained from Eq. (10) (Fig. 4)

n da 1 n
Fig. 3—Moisture distribution within concrete for 1-D mois- da = ∑ ki dhi ai ⇒ e x = = ∑ ki dhi ai (10)
ture loss (k1 and h1 shrinkage factor and relative humidity at i =1 A A i =1
a1) (t << t ∞ ) .
As shown in Fig. 4(a), this equation is extended for
discrete changes of moisture in n parallel layers; however,
the moisture change within the specimen is continuous. To
extend Eq. (10) to a continuous situation, the variation of a
length element of dx (Fig. 4(b)) due to moisture changes of
dh is given by

1 n
d ( dxi ) = ki dhi dxi ⇒ e i = ∑ ki dhi dxi (11)
A i =1

In the limit dx → 0, Eq. (11) can be rewritten in the form

of Eq. (12) (refer to Fig. 4(b))

1A
e= ∫ k ( x )dh ( x ) dx (12)
Fig. 4—Distribution of moisture with n moisture contour A0
lines (kn and hn shrinkage factor and relative humidity at an).
Equation (12) shows the shrinkage strain as a function of
moisture changes along the x-axis.
To extend Eq. (12) for the 2-D state of moisture loss, a
concrete specimen similar to Fig. 5 is considered.
The shrinkage strain along the x- and y-axes can be calcu-
lated from Eq. (12). Shrinkage strain can now be determined
along the arbitrary line cd, which is considered to be the sum
of the strains along the x-axis, and the induced strain along
cd by the moisture difference between the x-axis and the line
cd (Fig. 5). For moisture distribution similar to Fig. 5, the
shrinkage strain along cd is given by

Fig. 5—2-D moisture loss in concrete specimen. e m = e m −1 + De m , ( e cd = e m )

new condition long enough (t → t∞) to ensure that all parts or

of the specimen acquire the new moisture value.
This change in the value of the moisture induces a ∆V
change in the volume of the specimen. According to Eq. (7), e m = e m − 2 + De m + De m −1
the ratio of ∆V/V is equal to the sum of the shrinkage strain
along the main axes. Considering this relation and Eq. (6), 
the induced strain along the x-axis is (k is the constant)
e m = e1 + De m − ( m −1) + De m − ( m − 2) + ⋅⋅⋅ + De m
e x = k (h1 − h2 ) = k (100 − h2 ) (8) (13)
= e1 + ∑ mj=1 De j
At the next step, the shrinkage strain is calculated at
t << t∞, where the moisture distribution inside the specimen where e1 = ex.
is assumed to be similar to Fig. 3. According to Eq. (11)

272 ACI Materials Journal/May-June 2013

 1 n 1 A
De j =
A
( ) ( )
∑ i =1 k hi , j − hi , j −1 a i = ∫0 k hi , j − hi , j −1 dx
A
1A
and e x = ∫ k ( x )dh ( x ) dx
A0

Then

m 1 
m n
(
∑ j =1 De j = ∑ j =1  ∑ i =1 k hi , j − hi , j −1 ai 
 A 
)
(14a)
 1 
A
( )
or ∑ mj =1 De j = ∑ mj =1  ∫0 k hi , j − hi , j −1 d x 
A 

With multiplication of dy/dy (dy = yj – yj – 1) to ∆ej

m
∑ De j =
(
1 m  A hi , j − hi , j −1
∑ j = 1  ∫0 k
 )
dx dy (14b)
j =1 A  dy 

In the limit, Eq. (14b) can be rewritten as Eq. (15), which

is the final form of the shrinkage equation for 2-D moisture
losses (when dy → 0, then (hi,j – hi,j – 1)/dy → ∂h(x,y)/∂y).33

1 A 1 B A ∂h( x, y) Fig. 6—Shrinkage factor for w/c of 0.5 and 0.6.

e cd = ∫0 k ( x,0)[he − h( x,0)]dx + ∫0 ∫0 k ( x, y) dxdy (15)
A A ∂y

Determination of shrinkage factor

In laboratory experiments, the shrinkage factor was calcu-
lated for two different humidity ranges: 0 to 50% and 50 to
100% RH. Variations of the shrinkage factor obtained for two
different values of 0.5 and 0.6 water-cement ratio (w/c) are
presented in Fig. 6. These graphs are created by measuring
linear strain along the main axis in a cylindrical block of
concrete with a 1-D moisture flow.
As can be seen from Fig. 6, the shrinkage factor value
presented based on Fig. 6 increases at w/c = 0.6 for the
humidity ranges of 50 to 100%.

Construction and calibration of SDB

moisture meters
The modified SDB moisture meter15 is used to determine
the water changes in the concrete specimens of this study.
This sensor is also used to control the accuracy of the results
and model verification. This moisture meter consists of
two parallel conductive stainless steel plates (0.39 x 0.39 x
0.04 in. [10 x 10 x 1 mm]). The mortar mixture proportion
(between two plates) has been designed in such a way that it
gives the same absorption kinetic as testing concrete15 (refer
to Fig. 7(a)).
After calibration, these sensors are placed inside the
concrete members to measure the moisture content. Measure-
ments were conducted by passing alternating current in a
particular frequency range (1 KHz and 1.0 Vrms) to measure
electrical resistance.15
The hydration process changes the main features related
to the water movement inside the concrete. To preserve the
structure of the specimens after removal from the molds, they
are cured inside an autoclave under high temperature and
pressure (the X-ray diffraction [XRD] test was carried out to Fig. 7—Modified SDB moisture meter: (a) SDB sensor; and
determine the hydration degree of the autoclaved specimens). (b) calibration curve. (Note: 1 mm = 0.0394 in.)

ACI Materials Journal/May-June 2013 273

Specimen preparation and mixture design
To measure shrinkage strain, concrete specimens in the
form of slab (23.62 x 23.62 x 5.90 in. [600 x 600 x 150 mm]),
cylinder (23.62 x 11.81 in. [600 x 300 mm]), and prismatic
(15.75 x 7.87 x 7.87 in. [400 x 200 x 200 mm]) shapes with
a 2-D state of moisture loss were prepared.
The cement for the concrete specimens was ordinary port-
land cement (Type I). The concrete mixtures consisted of
1:3:3.5 cement:sand:gravel, respectively (proportions by
weight). The cement content was 21.84 lb/ft3 (350 kg/m3),
and the w/c was 0.5. The fine aggregate was river sand
and the coarse and fine aggregates were crushed, and the
maximum size of coarse aggregate was 0.39 in. (10 mm). In
the concrete mixtures, a high-range water-reducing admix-
ture was used to maintain suitable workability. After the
molds were removed, the specimens were cured under the
saturated condition for 48 hours. Then, they were exposed to
the laboratory condition (25°C [77°F] and 50% RH).
The shrinkage strain is measured at a distance of 75, 150,
225, and 300 mm (2.95, 5.91, 8.86, and 11.81 in.) from the
center of the slab along the x-axis. In the cylindrical spec-
imen, the shrinkage strain is measured along the axis parallel
to the cylindrical axis. In the prismatic specimen, in addi-
tion to measuring the shrinkage along the main axes, it is
also measured along the horizontal and vertical lines at the
distances of 50 and 100 mm (1.97 and 3.94 in.) from the
x-axis and 150 and 200 mm (5.91 and 7.87 in.) from the
y-axis.
Determination of moisture
flow coefficients
The functions that are used in this study for the initial
diffusion coefficient and also F1 (h) (Eq. (2)) have four unde-
fined factors: C1, hc, a0, and n (Eq. (16)).
 
 
1 − a0
C1F (h) = C1  a 0 +  (16)
  1− h 
n
 1+   2-D or radial moisture loss are shown in Fig. 10.
  1 − hc  
In this equation, C1 is the diffusion coefficient in the satu-
rated concrete. For the range of 0.45 to 0.8 w/c, the numer-
ical values for C1 are reported to be between 1.79 × 10–13 to
7.18 × 10–11 in.2/day (1.157 × 10–10 to 4.630 × 10–8 mm2/
day).17 However, in this research, the linear approximation
of C1, Eq. (17), is used.
C1 = (1.157 + 9.92 ( w c − 0.45) ) × 10 −10 (17)
The parameter hc and the coefficient a0 were chosen to be
75% and 0.05, respectively.11,12 Also, values of 6 and 9 for n
were obtained by the curve-fitting method for w/c of 0.5 and
0.6, respectively. As expected, with increasing the ratio of
the w/c, the value of n, which represents the physical concept
of the concrete moisture removal rate, increases. Increasing
pore size (resulting in a more continuous network of pores),
which reduces barriers against moisture movement within
the concrete, can be cited as the main reason for the observed
variation of n.
Using a similar method, the value of Cs (surface diffu-
sion coefficient) was inferred to be 0.58 and 0.55 for w/c of
274
0.5 and 0.6, respectively. Also, based on the results obtained,
the England-Ross34 Model was used to define the relation
between the RH (moisture) and concrete water content.
Results and discussion
In Fig. 8(a), moisture changes within the slab
at 20 and 150 days are shown to calculate shrinkage strain
by applying Eq. (15). The displayed moisture on the contour
lines of this figure represents the average RH between two
adjacent lines. Due to symmetry, only a quarter of the slab
is shown.
In Fig. 8(b), calculated shrinkage values from Eq. (15)
are compared with obtained values from thte laboratory
at 20 and 150 days.
Despite the observed discrepancy between calculated
and measured results, the calculated values are valid for the
estimation of shrinkage strain (note that the values obtained
from standard models have a significant approximation9).
Furthermore, according to Fig. 8(b), the proposed model
could be used to estimate the distribution of shrinkage strain.
However, it must be noted that the model needs to be opti-
mized for better estimation of shrinkage.
Figure 8(b) shows that the distribution of shrinkage strain
in concrete is not only uniform but also that shrinkage vari-
ations along an axis are not usually constant. To illustrate
this point, shrinkage variation along the x-axis (Fig. 8(a)) is
shown in Fig. 9. Note that these calculations are carried out
assuming that the moisture value between two consecutive
lines is constant, which implies that the shrinkage variations
along an axis are linear. However, to achieve more accuracy,
linear or even nonlinear variations of moisture between two
adjacent lines can be assumed.
Based on the observed results and the steep moisture
gradient near the surface areas, higher shrinkage strains are
expected in these areas. This causes the formation of micro-
cracks near the free surface of specimens, which is in full
agreement with laboratory observations.
Measured and calculated shrinkage strains around a cylin-
drical (23.62 x 11.81 in. [600 x 300 mm]) specimen with
Figure 10 shows that the shrinkage strain distribution in
the cylinder is not uniform and increases from the center to
the border areas. In this case, because of the radial moisture
transfer due to the displayed surface conditions, shrinkage
variations along any axis parallel to the cylinder’s main
axis are constant. The measured and calculated shrinkage
strains on the circumference of the cylinder are shown in
Fig. 10 to compare between the obtained results and the
ACI 209R-927 standard model.
Calculated and measured shrinkage strains at different
points of a prismatic specimen (15.75 x 7.87 x 7.87 in.
[400 x 200 x 200 mm]) with predicted shrinkage values from
the ACI 209R-927 model are shown in Fig. 11.
As can be seen from Fig. 11, moisture distribution is
not uniform along the x- and y-directions. Furthermore,
minimum shrinkage is formed along the main axis and the
amount of shrinkage strain increases near the corners of
the specimen.
CONCLUSIONS
In this paper, the combinations of experimental and
numerical results are used for the construction of a model
to predict the drying shrinkage and its distribution in the
concrete specimens.
ACI Materials Journal/May-June 2013
 Fig. 8—(a) Humidity contour lines in one quarter of slab with 23.62 x 23.62 x 5.90 in.
(600 x 600 x 150 mm) dimensions; and (b) predicted and measured values of shrinkage
along main axes. (Note: 1 mm = 0.0394 in.)

Fig. 10—Measured and calculated shrinkage strain in cylin-

drical specimen. (Note: 1 mm = 0.0394 in.)

Fig. 9—Shrinkage variation along x-axis of slab (150-day).
(Note: 1 mm = 0.0394 in.)
For this purpose, the moisture gradient in concrete speci-
mens is determined by solving a differential form of the modi-
fied Fick’s second law (using the Finite Element Method) and
also using the results of the humidity moisture meter.
The hypothesis of volume changes being proportional
to internal moisture loss is used to provide the relation-
ship between shrinkage and variations of internal moisture
(Eq. (15)). However, based on the obtained results from the
proposed method, the following can be concluded:
The assumption that volumetric changes are proportional
to the internal moisture loss where the shrinkage factor is a

ACI Materials Journal/May-June 2013 275

 Fig. 11—Shrinkage strains in different points of prismatic specimen. (Note: 1 mm = 0.0394 in.)
function of moisture content is accurate enough to provide
reasonable estimates of shrinkage strain.
The calculated shrinkage strain using the presented rela-
tion is in good agreement with laboratory observations.
Furthermore, the suggested model in comparison to standard
models such as the ACI 209R-927 model has the ability to
determine the strain distribution inside the specimens and
their surface.
The results obtained from the proposed model confirm
the well-known fact that the distribution of shrinkage
strain within the concrete is not uniform. Furthermore, the
shrinkage variations along an axis are not constant and the
rate of changes is an order greater than the changes in the
internal moisture.
This model predicts a higher probability of cracking and
crack propagation into thin concrete members than thicker
ones due to the severe moisture gradient.
NOTATION
as = approximately constant factor and dependent on material
properties
C = diffusion coefficient
C1 = initial diffusion coefficient (diffusion coefficient in saturated
concrete)
Cs = surface diffusion coefficient
gs(t) = ratio of elastic modulus with time—that is, gs(t) = E(t0)/E(t)
h = relative humidity
·
h = rate of moisture loss
ha = average humidity
hc = humidity at which diffusion coefficient drops halfway between
its maximum and minimum values
he = environment humidity
I = unit matrix
k = shrinkage factor
kds = moisture compressibility factor of the material
n = spread of the drop in diffusion coefficient (C)
T = temperature
t = time
te = time-equivalent hydration period
276
V = volume of specimen
wi(x) = weighting function
a0 = representing ratio min C/max C
e = shrinkage strain
e′ds = rate of drying shrinkage strain
e′s = ultimate shrinkage on complete drying
e′sh = rate of shrinkage strain
esh∞ = ultimate shrinkage
REFERENCES
1. Bažant, Z. P., “Prediction of Concrete Creep and Shrinkage, Past,
Present and Future,” Nuclear Engineering Design, V. 203, 2002, pp. 27-38.
2. Ferraris, C. F., and Wittmann, F. H., “Shrinkage Mechanisms of
Hardened Cement Paste,” Cement and Concrete Research, V. 17, 1987,
pp. 453-464.
3. Holt, E., “Early Age Autogenous Shrinkage of Concrete,” PhD disser-
tation, University of Washington, Seattle, WA, 2001, 209 pp.
4. Justnes, H.; Clemmens, F.; Depuydt, P.; Van Gemert, D.; and Sell-
evold, E. J., “Correlating the Deviation Point between External and Total
Chemical Shrinkage with Setting Time and Other Characteristics of
Hydrating Cement Paste,” Proceedings of the International RILEM Confer-
ence, V. 1, 2000, pp. 57-73.
5. Xi, Y., and Bažant, Z. P., “Moisture Diffusion in Cementitious
Materials Adsorption Isotherms—Moisture Capacity and Diffusivity,”
Advanced Cement Based Materials, V. 1, No. 6, Nov. 1994, pp. 248-257.
6. Bažant, Z. P., and Baweja, S., “Creep and Shrinkage Prediction Model
for Analysis and Design of Concrete Structure—Model B3,” Materials and
Structures, V. 28, No. 6, July 1995, pp. 357-365.
7. ACI Committee 209, “Prediction of Creep, Shrinkage, and Tempera-
ture Effects in Concrete Structures (ACI 209R-92) (Reapproved 2008),”
American Concrete Institute, Farmington Hills, MI, 1992, 47 pp.
8. CEB-FIP MC90, “Design of Concrete Structures,” CEB-FIP Model
Code 90, Thomas Telford, London, UK, 1993, pp. 52-58.
9. Jian, P. L., “Evaluation of Concrete Shrinkage and Creep Prediction
Models,” master’s thesis, San Jose State University, San Jose, CA, 2002,
pp. 12, 38-52.
10. Carlson, R. W., “Drying Shrinkage of Large Concrete Members,” ACI
Journal, V. 33, No. 1, Jan. 1937, pp. 327-336.
11. Bažant, Z. P., and Najjar, L. J., “Drying of Concrete as a Nonlinear
Diffusion Problem,” Cement and Concrete Research, V. 1, No. 5, Sept.
1971, pp. 461-473.
12. Bažant, Z. P., and Najjar, L. J., “Nonlinear Water Diffusion in Nonsat-
urated Concrete,” Materials and Structures, V. 5, No. 1, 1972, pp. 3-20.
ACI Materials Journal/May-June 2013
 13. Pickett, G., “The Effect of Change in Moisture Content on the Creep
of Concrete under a Sustained Load,” ACI Journal, V. 36, No. 2, Feb. 1942,
pp. 333-355.
14. Pihlajavaara, S. E., and Vaisanen, M., “Numerical Solution of Diffu-
sion Equation with Diffusivity Concentration Dependent,” Publication
No. 87, State Institute for Technical Research, Helsinki, Finland, 1965,
pp. 1-22.
15. Shekarchi, M.; Debicki, G. Y.; Billard, Y.; and Briot, R., “Nonde-
structive Monitoring Technique Based on Electrical Resistivity for Mois-
ture Condition in Containment Structures,” SMIRT 16, Washington, DC,
2001, pp. 1409-1414.
16. Yunping, X.; Bažant, Z. P.; Molina, L.; and Jennings, H. M., “Moisture
Diffusion in Cementitious Materials Moisture Capacity and Diffusivity,”
Advanced Cement Based Materials, V. 1, No. 6, Nov. 1994, pp. 258-266.
17. Martín-Pérez, B., “Service Life Modeling of RC Highway Structures
Exposed to Chlorides,” PhD dissertation, Department of Civil Engineering,
University of Toronto, Toronto, ON, Canada, 1999, pp. 65-74.
18. Mohammed, A., “Computational Modeling of Shrinkage in Repaired
Concrete,” master’s thesis, King Fahd University, Dhahran, Saudi Arabia,
1995, pp. 32-34.
19. Baroghel, V.; Mainguy, M.; Lassabatere, T.; and Coussy, O., “Charac-
terization and Identification of Equilibrium and Transfer Moisture Proper-
ties for Ordinary and High-Performance Cementitious Materials,” Cement
and Concrete Research, V. 29, No. 8, Aug. 1999, pp. 1225-1238.
20. Neville, A. M., Properties of Concrete, John Wiley & Sons, Inc., New
York, 1963, pp. 412-481.
21. Ferretti, D., and Bažant, Z. P., “Stability of Ancient Masonry Towers:
Stress Redistribution due to Drying, Carbonation, and Creep,” Cement and
Concrete Research, V. 36, No. 7, July 2006, pp. 1389-1398.
22. Benboudjema, F.; Meftah, F.; and Torrenti, J. M., “Interaction
between Drying Shrinkage, Creep and Cracking Phenomena in Concrete,”
Engineering Structures, V. 27, No. 2, Jan. 2005, pp. 239-250.
23. Witasse, R.; Georgin, J.; and Reynouard, J. M., “Nuclear Cooling
Tower Submitted to Shrinkage; Behavior under Weight and Wind,”
Nuclear Engineering and Design, V. 217, No. 3, Sept. 2002, pp. 247-257.
ACI Materials Journal/May-June 2013
24. Heum, J. M., and Jason, W., “Estimating Residual Stress in the
Restrained Ring Test under Circumferential Drying,” Cement and Concrete
Composites, V. 28, No. 5, May 2006, pp. 486-496.
25. Zhen, L.; Miguel, A.; Perez, L.; and Bolander, J. E., “Restraining
Effects of Fibers During Non-Uniform Drying of Cement Composites,”
Cement and Concrete Research, V. 36, No. 9, Sept. 2006, pp. 1643-1652.
26. Kim, J. K., and Lee, C. S., “Prediction of Differential Drying
Shrinkage in Concrete,” Cement and Concrete Research, V. 28, No. 7, 1998,
pp. 985-994.
27. Hyo, G. K.; Soo, J. H.; and Kim, J. K., “Non-Structural Cracking
in RC Walls Part I. Finite Element Formulation,” Cement and Concrete
Research, V. 36, No. 4, Apr. 2006, pp. 749-760.
28. Zhengwu, J.; Zhenping, S.; and Peiming, W., “Autogenous Rela-
tive Humidity Change and Autogenous Shrinkage of High-Performance
Cement Pastes,” Cement and Concrete Research, V. 35, No. 8, Aug. 2005,
pp. 1539-1545.
29. Hak, C. S., “Early Age Behavior of Bonded Concrete Overlays due
to Shrinkage and Thermal Changes,” PhD thesis, University of Illinois at
Urbana-Champaign, Urbana, IL, 2000.
30. Wittmann, F. H., “The Structure of Hardened Cement Paste: A Basis
for a Better Understanding of the Material Properties,” Proceedings of the
Conference on Hydraulic Cement Pastes: Their Structure and Properties,
Sheffield, UK, 1976, pp. 69-117.
31. Mackenzie, J. K., “The Elastic Constants of Solid Containing Spher-
ical Holes,” Proceedings of the Physical Society. Section B, V. 683, 1950,
pp. 2-11.
32. Bentz, D. P.; Garboczi, E. J.; and Quenard, D. A., “Modeling Drying
Shrinkage in Reconstructed Porous Materials: Application to Porous Vycor
Glass,” Modelling and Simulation in Materials Science and Engineering,
V. 6, 1998, pp. 211-236.
33. Ahmadi, J., “Evaluation of Drying Shrinkage Strain Associated to
Moisture Changes in Concrete,” PhD thesis, Iran University of Science and
Technology, Tehran, Iran, 2009, pp. 135-141.
34. England, G. L., and Ross, A. D., “Shrinkage, Moisture and Pore Pres-
sures in Heated Concrete,” Proceedings of the American Concrete Institute
International Seminar on Concrete for Nuclear Reactors, Special Publica-
tion, West Berlin, Germany, 1970, pp. 883-907.
277
 2013 HONORS AND AWARDS
PROGRAM NOMINATIONS
The American Concrete Institute is seeking your nominations for the 2013
Honors and Awards Program. Nominations are now being accepted for ACI’s
Honorary Member and Fellow of the Institute honors, as well as for ACI’s
personal awards for individual or organizational merit, the ACI Young
Member Award for Professional Achievement, the ACI Certification
Award, the ACI Concrete Sustainability Award, and the
Walter P. Moore, Jr. Faculty Achievement Award.

ACI selects the winners

of its annual awards
through an open
nomination process.

You can participate in the

Honors and Awards Program
by nominating worthy candidates
for award consideration.

Nomination forms can be found on thehe ACI website, www.concrete.org,

or by contacting Diane Pociask at diane.pociask@concrete.org.
 ACI MATERIALS JOURNAL
Title no. 110-M23
Mechanical Energy Dissipation Using Cement-Based
Materials with Admixtures
by Po-Hsiu Chen and D. D. L. Chung
Silica fume and a novel graphite network (8 vol.%) cementitious
admixture are effective for enhancing the mechanical energy dissi-
pation of cement-based materials, as shown under small-strain
dynamic flexure at 0.2 Hz. The fraction of energy dissipated reaches
0.26, 0.58, and 0.22 for cement paste, mortar, and concrete, respec-
tively, as provided by silane-treated silica fume and the cementi-
tious admixture, which cause steel-reinforced concrete to increase
the dissipation, loss modulus, loss tangent, and storage modulus
by 16,000%, 450,000%, 16,000%, and 170%, respectively. The
highest loss tangent and loss modulus obtained are 0.14 and
3.5 GPa (20.3 and 507.5 ksi), respectively. Silane-treated silica
fume alone causes steel-reinforced concrete to increase the dissi-
pation by 9900%; untreated silica fume alone gives an 8000%
increase. Without steel or admixtures, the dissipation decreases
from cement paste to mortar and concrete. With steel and/or the
admixtures, the dissipation increases from paste to mortar and
decreases from mortar to concrete. The dissipation decreases with
increasing frequency, such that the presence of silica fume reduces
the frequency effect.
Keywords: cement; compressive strength; damping; energy dissipation;
flexural strength; graphite; mortar; silica fume.
INTRODUCTION
Mechanical energy dissipation involves the elimination of
mechanical energy by conversion of the energy to another
form of energy, which is commonly heat. It is relevant to
vibration damping. This is passive damping. In contrast, active
damping is expensive because it uses sensors and actuators in
a synchronized fashion to suppress vibrations. Vibrations are
undesirable for bridges, railroads, pipes, and buildings and are
often associated with noise, which is also undesirable.
For passive damping effectiveness, materials that provide
the conversion of mechanical energy to heat are necessary.
These materials, known as damping materials, are to be
distinguished from mechanical isolation materials, which
are for isolating two objects so that the vibrations of one
object are deterred from propagating to the other object. An
example is the isolation of a building foundation from the
ground below it, as it is attractive for earthquake protection. A
mechanical isolation material acts like a cushion that spreads
the mechanical energy to a limited extent. Thus, mechanical
isolation materials are relatively soft; an example is rubber.
However, due to the softness, the mechanical energy absorption
is low, thus making the material ineffective for damping.
This paper concerns damping materials.
The performance of a damping material is described by: 1)
the loss modulus, which is the imaginary part of the complex
modulus and relates to the viscous modulus; and 2) the loss
tangent (also known as tand, where d is the phase angle
between the stress and strain waves), which describes the
degree of viscous character. The behavior is purely elastic
(no energy dissipation) when d = 0 degrees. The behavior is
purely viscous when d = 90 degrees. The degree of viscous
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
character increases with increasing d. The storage modulus
(the real part of the complex modulus) measures the stored
energy, representing the elastic portion; the loss modulus
(the imaginary part of the complex modulus and equal to
the product of the storage modulus and the loss tangent)
measures the energy dissipated as heat, representing the
viscous portion. In the small-strain elastic regime such that
the dynamic stress/strain is sinusoidal, the loss tangent is
equal to twice the damping ratio.
Concrete is inherently inadequate for damping; however,
due to its large volume in a concrete structure, an increase of
concrete damping can make a big difference to the structure.
Concrete is also attractive for its high durability compared to
materials such as rubber.
Large-amplitude vibrations that involve plastic deforma-
tion are encountered in earthquakes, but small-amplitude
vibrations without plastic deformation are encountered
during normal structural operation. This paper is concerned
with the latter.
Silica fume is fine noncrystalline silica produced by
electric arc furnaces as a by-product of the production of
metallic silicon or ferrosilicon alloys. It is a powder with a
particle size 100 times smaller than that of anhydrous port-
land cement particles—that is, a mean particle size between
0.1 and 0.2 mm (4 × 10–6 and 8 × 10–6 in.). The SiO2 content
ranges from 85 to 98%. Silica fume is pozzolanic1 due to the
cementitious character resulting from its surface reactivity,
which relates to its amorphous structure.
Silica fume used as an admixture in concrete has signifi-
cant effects1-4 on the strength, modulus, ductility, vibration
damping capacity, sound absorption, abrasion resistance, air
void content, shrinkage, bonding strength with reinforcing
steel, permeability, chemical attack resistance, alkali-silica
reactivity reduction, corrosion resistance of embedded steel
reinforcement, freezing-and-thawing resistance, creep rate,
coefficient of thermal expansion (CTE), specific heat, thermal
conductivity, defect dynamics, dielectric constant, and degree
of fiber dispersion in mixtures containing short microfibers.
However, silica fume addition degrades the workability.
Damping enhancement of cement paste under flexure
was achieved by using silica fume as an admixture; the loss
modulus at 0.2 Hz is increased by 1500% for untreated silica
fume and by 2100% for silane-treated silica fume.5-7 Hence,
the silane treatment increases the loss modulus by 40% for
cement paste. In the case of mortar, the silane treatment
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-272.R1 received April 5, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
279
Po-Hsiu Chen is a PhD Candidate in mechanical engineering at the University at
Buffalo (UB-SUNY), Buffalo, NY.
D. D. L. Chung is a Professor of mechanical and aerospace engineering and Director
of the Composite Materials Research Laboratory at UB-SUNY. She received her PhD
in materials science from the Massachusetts Institute of Technology, Cambridge, MA,
in 1977.
increases the loss modulus by 210% at 0.2 Hz and by 150%
at 2.0 Hz.6 By adding silane-treated silica fume to mortar,
the loss modulus is increased by 340% at 0.1 Hz and by
810% at 2.0 Hz.7 The silanes used in References 5 through
7, which are prior works of the second author, are not
the same as those used in References 8 and 9, although
References 5 through 9 all involve dynamic flexural testing.
Silica fume enhances the energy dissipation due to an
interfacial friction mechanism associated with the inter-
face between silica particles and cement. In contrast,
latex10 enhances the dissipation due to a bulk viscoelastic
mechanism associated with each latex particle. Compared to
silica fume, latex is expensive.
A graphite network (8 vol.%) formed in cement from exfo-
liated graphite was shown to be effective for increasing the
loss tangent to the level of rubber.11 The network formation
is enabled by the mechanical interlocking between pieces of
exfoliated graphite, each of which is obtained by the exfolia-
tion of an intercalated graphite flake and is known as a worm
(due to its morphology). The high value of the loss tangent
is due to the interfacial friction at the interfaces between the
multiple graphite layers that make up a wall of the cellular
structure of exfoliated graphite. This damping mechanism
is hereby referred to as a nanoscale multi-layer mechanism.
In contrast, silica fume—whether silane-treated or not—
enhances the loss tangent by a much lower degree, although
it significantly increases the storage modulus.5-7 Because the
graphite network mainly serves to enhance the loss tangent,
while silica fume mainly serves to enhance the storage
modulus, their synergistic use is promising and its investiga-
tion is thus the primary objective of this work.
Prior work11 involving the graphite network cementi-
tious admixture addresses the cement-matrix composite
containing the graphite network, such that the admixture is a
single monolith (a bulk material). However, it is more versa-
tile and convenient to use this material as an admixture. This
admixture is capable of cementitious bonding because it is
mostly cement and is incorporated in the mixture when the
degree of hydration of the cement in the admixture is low.
The cementitious nature is attractive for achieving a good
bond between the admixture and the cement matrix outside
the admixture. The second objective is to investigate the
effectiveness of this admixture as an admixture rather than
a monolith.
The fraction of input mechanical energy dissipated per
unit volume is a quantity that is practically meaningful.
Prior work using silica fume in cement,5-9 polymers in
cement,10,12 steel in cement,13 and a graphite network in
cement11 reports the loss modulus, storage modulus, and
loss tangent without evaluating this fraction. (The loss
modulus values in Reference 12 are incorrectly calculated;
the correct values are obtained by multiplying the reported
values by 0.01.) Prior work on the use of carbon nanotube
in cement14 reports the damping ratio without evaluating the
fraction of energy dissipated. Thus, the third objective is to
provide an evaluation of the fraction of energy dissipated.
280
Prior work involving the graphite network11 is limited to
cement paste without aggregates or steel reinforcing bar. The
fourth objective is to study the effects of aggregates and steel
on the energy dissipation—that is, the effects of the admix-
tures in the presence of aggregates and steel.
Prior work1,5,6 has shown that silica fume, particularly after
silane treatment, is effective for improving the static mechan-
ical properties of cement-based materials. The fifth objective
is to evaluate the effect of the combined use of silica fume
and the graphite network on the static mechanical properties,
specifically the flexural and compressive strengths.
RESEARCH SIGNIFICANCE
This work investigates the mechanical energy dissipation
ability of cement-based materials under small-strain low-
frequency dynamic flexure, as indicated by the loss modulus.
It addresses the synergistic effects of silica fume (including
its silane treatment) and a novel graphite network cementi-
tious admixture on the dissipation ability and the static
mechanical properties. It provides a comparative evaluation
of cement-based materials with and without admixtures,
with and without aggregates, and with and without steel
reinforcing bars. Cement-based materials that exhibit up to
60% flexural energy dissipation are provided. Guidelines are
provided for the design of materials for both energy dissipa-
tion and static performance.
EXPERIMENTAL METHODS
Materials
Portland cement (Type I, ASTM C150) was used. Silica
fume (if used) was at 15% by mass of cement, as in prior
work1; it had a particle size ranging from 0.03 to 0.5 mm (1 ×
10–6 to 2 × 10–5 in.) with an average size of 0.2 mm (8 × 10–6 in.);
it contained >93 wt.% SiO2, <0.7 wt.% Al2O3, <0.7 wt.%
CaO, <0.7 wt.% MgO, <0.5 wt.% Fe2O3, <0.4 wt.% Na2O,
<0.9 wt.% K2O, and <6 wt.% loss on ignition.
The silane coupling agent was a 1:1 (by weight) mixture
of Z-6020 (H2NCH2CH2NHCH2CH2CH2Si(OCH3)3) and
Z-6040 (OCH2CHCH2OCH2CH2CH2Si(OCH3)3).5,6 The
amine group in Z-6020 served as a catalyst for the curing
of the epoxy and consequently allowed the Z-6020 mole-
cule to attach to the epoxy end of the Z-6040 molecule. The
trimethylsiloxy ends of the Z-6020 and Z-6040 molecules
then connected to the –OH functional group on the surface
of the silica fume. The silane was dissolved in ethylacetate.
The concentration of silane in the ethylacetate solution
was varied from 0.5% to 5.0 wt.%. Surface treatment of
the silica fume was performed by immersion in the silane
solution, heating to 75°C (167°F) while stirring, and then
holding at 75°C (167°F) for 1.0 hours, followed by filtration
and drying. After this, the silica fume was heated at 110°C
(230°F) for 12 hours.
The fine aggregate was natural sand with a particle size of
0.1 to 2.36 mm (4 × 10–3 to 9 × 10–2 in.) with a typical size of
approximately 0.3 mm (1 × 10–2 in.). The sand-cement ratio
was 1.00. The water-cement ratio (w/c) was 0.35 for cement
pastes without silica fume, 0.40 for cement pastes with silica
fume, and 0.45 for mortars and concretes. A high-range water-
reducing agent was used at 1.0% by mass of cement. The
defoamer was used at 0.13% (percent of specimen volume).
The coarse aggregate had an average size of 3 and 1.5 mm
(0.11 and 0.06 in.) for static and dynamic testing, respectively.
The coarse aggregates were smaller than those convention-
ally used for concrete due to the small specimen size. The
ACI Materials Journal/May-June 2013
mass ratio of cement:sand:coarse aggregate was 1:1.5:2.5.
Therefore, the cement-matrix composites containing sand
and coarse aggregate were, strictly speaking, not concrete.
Nevertheless, the term “concrete” is used.
Exfoliated graphite (worms) is obtained by rapid heating
of expandable graphite flake (graphite flake intercalated
with sulfuric acid and nitric acid in the presence of catalysts
with a flake size of 300 mm [0.012 in.]) at 900°C (1650°F)
for 2 minutes with flowing nitrogen. During exfoliation,
each flake expands by hundreds of times along the direction
perpendicular to the plane of the flake, resulting in a shape
(like a worm) that is long along the direction of exfoliation.
The worms had a length of 2 to 4 mm (0.08 to 0.16 in.). During
exfoliation, the vast majority of the intercalate desorbed so
that the intercalate that remains after exfoliation is low in
concentration and is strongly held to the graphite.15,16 For
fabricating the graphite network cementitious admixture,
the particle size of the cement is reduced by ball milling
for 24 hours using ceramic cylinders as the grinding medium
contained in a ceramic container. The milling reduces the
typical cement particle size from 50 to 30 mm (2.0 × 10–3 to
1.2 × 10–3 in.).
For fabricating the graphite network cementitious admix-
ture, the exfoliated graphite is mixed with the milled cement
particles in the dry state for 24 hours using a ball mill without
any grinding medium. The weights of the exfoliated graphite
and cement in the mixture are controlled. The compression
of the mixture is conducted in the dry state in a cylindrical
mold with a length of 450 mm (18 in.) and an inner diam-
eter of 31.8 mm (1.25 in.) by applying a uniaxial pressure
of 5.6 MPa (810 psi [1000 lb]) via a matching piston. The
entire thickness of a composite specimen is obtained in one
5.6 MPa (810 psi) compression stroke. Each resulting spec-
imen is a disc with a diameter of 31.8 mm (1.25 in.) and a
thickness ranging from 2.0 to 3.0 mm (0.08 to 0.12 in.).
After this, the disc is exposed to water for curing the
cement. This involves exposure to moisture for 2 days,
followed by immersion in water for 26 days. After 7 days of
water immersion (when the sheet is not hard yet), the disc
is removed from the water and cut by using a knife along
four diameters (45 degrees apart) to make eight 45-degree
pie-shaped pieces. After the cutting, the pieces are immersed
in water for 1 day and are immediately used at this curing
age as an admixture. Thus, the longest in-plane dimension
of each piece is 16 mm (0.63 in.).
The graphite network cementitious admixture amounts
to 62 vol.% of the cement-matrix composite (excluding
the steel volume). Within the admixture, the graphite
network amounts to 8 vol.% and the cement matrix amounts
to 92 vol.%.9
All ingredients except the graphite network admixture (if
applicable) are mixed in a rotary mixer with a flat beater. In
the case of specimens for dynamic mechanical testing under
flexure, the pouring of the mixture into a 150 x 25 x 4 mm
(6.0 x 1.0 x 0.16 in.) steel mold to form a beam-shaped
flexural specimen is conducted in two steps, such that the
amount of the mixture is the same in each step and the
graphite network admixture plates are placed manually
between the two pours, with the admixture plane preferen-
tially in the plane of the beam. In the case that the flexural
specimen for dynamic testing contains steel reinforcing bars,
two reinforcing bars with a diameter of 3 mm (0.12 in.) are
placed in the 150 x 25 x 4 mm (6.0 x 1.0 x 0.16 in.) steel
mold prior to pouring the cement mixture into the mold.
ACI Materials Journal/May-June 2013
Fig. 1—Schematic illustration of 40 x 40 mm (1.6 x 1.6 in.)
cross section of specimen containing steel for flexural testing
up to failure. Specimen is 160 x 40 x 40 mm (6.3 x 1.6 x
1.6 in.), perpendicular to plane of drawing. Steel is shown
by black areas. (Note: 1 mm = 0.0394 in.)
The cement covering the reinforcing bar in the plane of the
beam is much smaller than the conventional thickness of
one reinforcing bar diameter due to the size limitation of the
testing equipment. Nevertheless, the comparative study of
specimens with and without reinforcing bars provides useful
information on the possible effect of reinforcing bars on the
effectiveness of the admixtures.
In the case of specimens for static mechanical testing
under flexure or compression, all ingredients except the
graphite network cementitious admixture (if applicable)
are mixed in a rotary mixer with a flat beater. Then, the
graphite network cementitious admixture (if applicable) is
added and mixing is conducted manually. The specimens
have dimensions of 160 x 40 x 40 mm (6.3 x 1.6 x 1.6 in.)
and 51 x 51 x 51 mm (2 x 2 x 2 in.) for flexural and compres-
sive testing, respectively, as provided by using steel molds.
Due to gravity, there is a degree of preferred orientation of
the graphite network cementitious admixture plates in the
horizontal plane during curing—that is, the plane of the flex-
ural beam and the plane parallel to the stress in compressive
testing. In the case of static flexural specimens containing
steel, four parallel steel reinforcing bars with a diameter
of 3 mm (0.12 in.) are positioned symmetrically in the mold
along its length prior to pouring (Fig. 1). The compressive
specimens do not contain steel.
For all specimens, after filling the mold, an external
vibrator is used to facilitate compaction and diminish the
air bubbles. The specimens are demolded after 24 hours and
then cured at a relative humidity of nearly 100% for 28 days.
Testing
Dynamic testing (per ASTM D4065-94) at a controlled low
frequency (0.2 to 2.0 Hz, below any resonance frequency)
was conducted at room temperature (20°C [68°F]) using a
dynamic mechanical analyzer in the form of a forced reso-
nance analyzer under stress control in the small-strain regime.
A sinusoidal force (stress) was applied and the displace-
ment (strain) was measured, thus allowing the measurement
of the phase angle d. A low frequency was chosen due to
the decrease in value and increase in inaccuracy of the loss
tangent (tand) with increasing frequency. Based on prior
related work, increasing the frequency to 2 Hz does not
281
Table 1—Dynamic flexural properties of cement-based materials at 0.2 Hz showing effects of silica fume
and its silane treatment (dissipation refers to fraction of energy dissipated)
No silica fume Untreated silica fume Silane-treated silica fume
Property Paste Mortar Concrete Paste Mortar Concrete Paste Mortar Concrete
Storage
1.81 ± 0.12 9.01 ± 0.31 10.83 ± 0.48 4.18 ± 0.20 19.23 ± 0.84 27.82 ± 0.87 7.13 ± 0.54 28.91 ± 2.22 31.02 ± 1.12
modulus,
(262 ± 17) (1310 ± 40) (1570 ± 70) (606 ± 29) (2788 ± 121) (4034 ± 126) (1030 ± 80) (4192 ± 322) (4498 ± 162)
GPa (ksi)

0.030 ± 0.00041 ± 0.00023 ± 0.055 ± 0.026 ± 0.016 ± 0.061 ± 0.046 ± 0.021 ±

Loss tangent
0.009 0.00009 0.00005 0.011 0.003 0.003 0.008 0.008 0.005
Without
steel 0.00359 ±
Loss 0.054 ± 0.00025 ± 0.230 ± 0.499 ± 0.440 ± 0.435 ± 1.331 ± 0.653 ±
0.00003
modulus, 0.001 0.00003 0.002 0.002 0.004 0.004 0.018 0.007
(0.521 ±
GPa (ksi) (7.8 ± 0.1) (0.036 ± 0.04) (33.4 ± 0.3) (72.3 ± 0.3) (63.8 ± 0.58) (63.1 ± 0.58) (193 ± 3) (94.7 ± 1.0)
0.004)

0.032 ± 0.0028 ± 0.0013 ± 0.104 ± 0.242 ± 0.124 ± 0.256 ± 0.461 ± 0.148 ±

Dissipation
0.006 0.0002 0.0003 0.011 0.029 0.007 0.042 0.083 0.010

Storage
1.61 ± 0.09 8.01 ± 0.34 10.36 ± 0.48 4.59 ± 0.27 18.81 ± 1.01 24.94 ± 0.71 6.33 ± 0.17 23.81 ± 0.95 29.83 ± 0.51
modulus,
(233 ± 13) (1161 ± 49) (1502 ± 70) (666 ± 39) (2727 ± 146) (3616 ± 103) (917 ± 25) (3452 ± 137) (4325 ± 74)
GPa (ksi)

0.028 ± 0.00065 ± 0.00024 ± 0.061 ± 0.041 ± 0.014 ± 0.068 ± 0.046 ± 0.019 ±

Loss tangent
0.013 0.00011 0.00006 0.018 0.012 0.004 0.012 0.013 0.007
With
steel 0.00519 ± 0.00024 ±
Loss 0.045 ± 0.275 ± 0.771 ± 0.351 ± 0.430 ± 1.088 ± 0.568 ±
0.00003 0.00004
modulus, 0.001 0.005 0.007 0.005 0.002 0.012 0.009
(0.753 ± (0.035 ±
GPa (ksi) (6.5 ± 0.1) (39.8 ± 0.7) (112 ± 1) (50.9 ± 0.7) (62.3 ± 0.3) (157.8 ± 1.7) (82.3 ± 1.3)
0.004) 0.005)

0.033 ± 0.0071 ± 0.0014 ± 0.123 ± 0.291 ± 0.112 ± 0.163 ± 0.321 ± 0.141 ±

Dissipation
0.008 0.0004 0.0004 0.077 0.006 0.006 0.072 0.005 0.011

affect the relative performance of the various cement-based
materials.10 This method of damping testing is as used in
prior related work.5-13 Consistency between results obtained
by the Forced Resonance Method and the Free Resonance
Method has been previously shown.17
The dynamic flexure (three-point bending) specimens are
simply supported beams with dimensions of 150 x 25 x 4 mm
(5.91 x 0.98 x 0.16 in.) with spans of 115 mm (4.53 in.).
The loads are large enough for the deflection amplitude to
range from 5 to 10 mm (2 × 10–4 to 4 × 10–4 in.) (greater
than the minimum value of 5 mm [2 × 10–4 in.] that the
equipment requires for accurate results). The static strain
ranges from 0.03 to 0.10%; the dynamic strain—that is,
the amplitude of the variation of the strain around the static
strain—ranges from 0.04 to 0.07%. The static stress ranges
from 0.14 to 0.31 MPa (20 to 45 psi) and the dynamic stress
ranges from 0.11 to 0.21 MPa (16 to 30 psi), such that both
stresses increase from cement paste to mortar and concrete.
All the strains are in the regime of elastic deformation, with
no damage inflicted, as confirmed by the reproducibility of
the results upon repeated testing of the same specimen. Six
specimens of each type were tested.
For a sinusoidal stress wave in the small-strain regime, the
energy loss per unit volume E is18
E = πγ 2 G ′′
where g is the dynamic flexural strain amplitude, and G″ is
the loss modulus. The quantity E is the area of the stress-
strain hysteresis loop centered at the static stress-strain
point. The energy input per unit volume corresponds to the
area under the loading (upper) part of the loop. The fraction of
energy dissipated is the energy loss per unit volume divided
by the energy input per unit volume.
For static flexural testing up to failure, the midspan deflection
rate is 0.5 mm/min (0.02 in./min) and the span is 120 mm
(4.7 in.). The specimen is a simply supported beam. A
hydraulic material testing system is used. Static compres-
sion testing up to failure according to ASTM C109-80 is
performed using a hydraulic material testing system. The
crosshead speed is 0.5 mm/min (0.02 in./min). Six speci-
mens of each composition were tested.
RESULTS
Effects of silica fume and its silane treatment
Table 1 shows the effect of silica fume addition on the
dynamic flexural properties of cement pastes, mortars, and
concretes in the absence and presence of steel. The silica
fume addition increases the storage modulus—whether the
material is cement paste, mortar, or concrete—such that the
effect is stronger when the silica fume is silane-treated. This
is expected because the silica particles are stiff and serve
to reinforce. Moreover, the small particle size of the silica
causes refinement of the pore system. The silane treat-
ment helps because it increases the hydrophilicity, thereby
improving the spatial distribution of the silica particles.6
The storage modulus increases from cement paste to
the corresponding mortar and the corresponding concrete,
(1) whether silica fume is present or not and whether the silica
fume is silane-treated or not. This is expected because the
sand and stones serve to reinforce.
The loss tangent, loss modulus, and fraction of energy
dissipated are all enhanced by the silica fume addition, such
that the effect is stronger when the silica is silane-treated.
Without silica fume, these three quantities all decrease from

282 ACI Materials Journal/May-June 2013

Table 2—Fractional change in dynamic flexural properties (0.2 Hz) of cement-based materials due to silica
fume (dissipation refers to fraction of energy dissipated)
Untreated silica fume (S)
Property Paste Mortar
Storage modulus, % 130 110
Loss tangent, % 510 6200
Without steel
Loss modulus, % 30 14,000
Dissipation, % 230 8500
Storage modulus, % 180 130
Loss tangent, % 120 6200
With steel
Loss modulus, % 510 15,000
Dissipation, % 270 4000
cement paste to the corresponding mortar and corresponding
concrete. However, with silica fume (whether silane-treated
or not), all three quantities increase from cement paste to
the corresponding mortar but decrease from mortar to the
corresponding concrete.
Comparison of cement paste and the corresponding mortar
shows that the addition of sand decreases the loss modulus
in the absence of silica fume (due to the decrease of the loss
tangent), but increases the loss modulus in the presence of
silica fume (due to the increase of the storage modulus).
With silica fume, mortar has a higher loss modulus than
the cement paste due to the high storage modulus and the
moderately high loss tangent of the mortar.
The addition of sand to cement paste decreases the loss
tangent, whether silica fume (silane-treated or not) is present
or not. Similarly, the addition of stones to mortar, as shown
by comparing mortar and concrete, decreases the loss
tangent, whether silica fume (silane-treated or not) is present
or not. This effect of silica fume is attributed to the interfa-
cial friction mechanism of energy dissipation that is accen-
tuated by the large area of the interface between the small
silica particles and the cement matrix.1 Comparison of the
loss tangent of cement paste and the corresponding mortar
shows that the addition of sand to cement paste decreases by
orders of magnitude the loss tangent in the absence of silica
fume and substantially decreases (not by orders of magni-
tude) the loss tangent in the presence of silica fume. This
effect of sand is due to the homogeneity within each sand
particle in contrast to the heterogeneity inside cement paste.
The heterogeneity helps enhance the energy loss. The sand
particles do not have a sufficiently small size to provide a
large enough interface area for increasing the loss tangent.
Consistent with this notion is the fact that carbon fiber with a
diameter of 15 mm (5.9 × 10–4 in.) as an admixture decreases
the loss tangent.6 In contrast, the small particle size (approx-
imately 0.1 mm [4 × 10–6 in.]) makes the silica fume effective
for enhancing damping. The effect of sand addition is less
dramatic in the presence of silica fume because the silica
fume enhances the loss tangent, thus making the negative
effect of sand less significant. For similar reasons, the addi-
tion of stones to mortar decreases the loss tangent.
Silane-treated silica fume is more effective than untreated
silica fume for increasing the loss tangent because of: 1) the
superior spatial distribution of the silica particles in the pres-
ence of the silane, which renders the silica hydrophilic1;
and 2) the silane coating providing a viscous mechanism
of energy loss that is in addition to the interfacial friction
mechanism provided by the silica particles.
ACI Materials Journal/May-June 2013
Silane-treated silica fume (S′)
Concrete Paste Mortar Concrete
160 290 220 190
6900 80 11,000 9000
180,000 710 37,000 260,000
9100 710 16,000 11,000
140 290 200 190
5700 140 7000 7800
150,000 860 21,000 240,000
7800 380 4400 9900
The effects described previously for the effects in the
absence of steel also occur in the presence of steel (Table 1).
With steel, mortar with silane-treated silica fume again gives
the highest loss modulus and the highest fraction of energy
dissipated. However, the values are lower than the corre-
sponding values without steel because the storage modulus
is slightly decreased by the steel addition. The effect of steel
on the loss tangent is negligible because the steel is in the
elastic deformation regime and the diameter of the steel
reinforcing bar is too small to provide adequate interface
area for enhancing the loss tangent.
Although the loss tangent decreases from cement paste to
the corresponding mortar and the corresponding concrete
(Table 1), the fractional increase in loss tangent due to
the silica fume addition increases from cement paste to
the corresponding mortar or concrete (Table 2). The frac-
tional changes in loss tangent, loss modulus, and fraction
of energy dissipated are much higher for mortar than the
corresponding cement paste, but the fractional change in
the storage modulus is slightly lower for mortar than the
corresponding cement paste. The larger fractional increases
in loss tangent, loss modulus, and fraction of energy dissi-
pated for mortar and concrete compared to cement paste are
because mortar and concrete exhibit lower loss tangents than
cement paste. The fractional increase in the loss modulus
due to the silica fume addition increases greatly from cement
paste to the corresponding mortar and the corresponding
concrete because the loss modulus without silica fume
decreases greatly from cement paste to mortar and concrete
and the storage modulus increases from paste to mortar and
concrete. For concrete, the addition of silane-treated silica
fume increases the loss modulus by 260,000%. The effect of
silane-treated silica fume on the loss tangent is smaller in the
presence of steel (Table 2). Thus, the effect of silane-treated
silica fume on the dissipation is smaller in the presence of
steel (except that it is comparable in the case of cement paste
with untreated silica fume).
Among all the materials in Table 1, mortars with silane-
treated silica fume (with and without steel) give the two
highest fractions of energy dissipated. This is consistent
with the fact that they give the highest values of the loss
modulus—a consequence of high values of both loss tangent
and storage modulus.
Table 3 shows that the fractional increase in storage
modulus due to the silane treatment decreases from paste
to mortar and concrete. However, the fractional increase
in loss tangent due to the silane treatment increases from
paste to mortar and decreases from mortar to concrete, thus
283
Table 3—Fractional change in dynamic flexural properties (0.2 Hz) of cement-based materials without
steel due to silane treatment of silica fume, as shown by comparing data for S and S′ in Table 1
Property Paste Mortar Concrete
Storage modulus, % 71 51 10
Loss tangent, % 11 77 30
Without steel
Loss modulus, % 89 170 50
Fraction of energy dissipated, % 150 400 30
Storage modulus, % 41 27 20
Loss tangent, % 11 77 40
With steel
Loss modulus, % 56 41 60
Fraction of energy dissipated, % 30 10 30

Table 4—Dynamic flexural properties (0.2 Hz), showing effects of silica fume (S is untreated silica
fume and S′ is silane-treated silica fume), graphite network admixture (A), and steel reinforcing bar (R)
separately and in combination
Additional ingredient(s) Storage modulus, GPa (ksi) Loss tangent Loss modulus, GPa (ksi) Fraction of energy dissipated
None 1.81 ± 0.12 (262 ± 17) 0.030 ± 0.009 0.054 ± 0.001 (7.83 ± 0.15) 0.032 ± 0.006
S 4.18 ± 0.20 (606 ± 29) 0.055 ± 0.011 0.230 ± 0.002 (33.4 ± 0.3) 0.104 ± 0.011
S′ 7.13 ± 0.54 (1034 ± 78) 0.061 ± 0.008 0.435 ± 0.004 (63.1 ± 0.58) 0.256 ± 0.042
A 1.92 ± 0.15 (278 ± 22) 0.128 ± 0.014 0.245 ± 0.002 (35.5 ± 0.3) 0.121 ± 0.017
S′ + A 3.88 ± 0.35 (562 ± 51) 0.125 ± 0.027 0.485 ± 0.009 (70.3 ± 1.3) 0.184 ± 0.021
S′ + A 4.92 ± 0.13 (713 ± 19) 0.128 ± 0.019 0.629 ± 0.002 (91.2 ± 0.3) 0.248 ± 0.019
Cement paste
R 1.61 ± 0.09 (233 ± 13) 0.028 ± 0.013 0.045 ± 0.001 (6.53 ± 0.15) 0.033 ± 0.008
S+R 4.59 ± 0.27 (666 ± 39) 0.061 ± 0.018 0.275 ± 0.005 (39.8 ± 0.7) 0.123 ± 0.077
S′ + R 6.33 ± 0.17 (917 ± 25) 0.068 ± 0.012 0.430 ± 0.002 (62.3 ± 0.3) 0.163 ± 0.072
A+R 1.91 ± 0.12 (276 ± 17) 0.096 ± 0.010 0.183 ± 0.001 (26.5 ± 0.1) 0.072 ± 0.006
S+A+R 4.41 ± 0.32 (639 ± 46) 0.117 ± 0.029 0.516 ± 0.009 (74.1 ± 1.3) 0.222 ± 0.005
S′ + A + R 5.81 ± 0.26 (842 ± 38) 0.119 ± 0.033 0.697 ± 0.008 (101 ±1) 0.258 ± 0.008
None 9.01 ± 0.31 (1306 ± 45) 0.00041 ± 0.00009 0.00359 ± 0.00003 (0.521 ± 0.004) 0.0028 ± 0.0002
S 19.23 ± 0.84 (2788 ± 121) 0.026 ± 0.003 0.499 ± 0.002 (72.3 ± 0.3) 0.242 ± 0.029
S′ 28.91 ± 2.22 (4192 ± 322) 0.046 ± 0.008 1.331 ± 0.018 (193 ± 3) 0.461 ± 0.083
S+A 21.01 ± 0.83 (3046 ± 120) 0.134 ± 0.015 2.821 ± 0.012 (409 ± 2) 0.541 ± 0.062
S′ + A 26.71 ± 0.83 (3872 ± 120) 0.139 ± 0.021 3.543 ± 0.017 (513 ± 2) 0.577 ± 0.042
Mortar
R 8.01 ± 0.34 (1161 ± 49) 0.00065 ± 0.00011 0.00519 ± 0.00003 (0.753 ± 0.004) 0.0071 ± 0.0004
S+R 18.81 ± 1.01 (2727 ± 146) 0.041 ± 0.012 0.771 ± 0.007 (111.8 ± 1.0) 0.291 ± 0.006
S′ + R 23.81 ± 0.95 (3452 ± 137) 0.046 ± 0.013 1.088 ± 0.012 (157.8 ± 1.7) 0.321 ± 0.005
S+A+R 22.14 ± 1.42 (3210 ± 205) 0.092 ± 0.018 2.223 ± 0.026 (322 ± 4) 0.488 ± 0.033
S′ + A + R 25.44 ± 1.13 (3688 ± 163) 0.102 ± 0.015 2.587 ± 0.016 (375 ± 2) 0.512 ± 0.028
None 10.83 ± 0.48 (1570 ± 70) 0.00023 ± 0.00005 0.00025 ± 0.00003 (0.036 ± 0.04) 0.0013 ± 0.0003
S 27.82 ± 0.87 (4034 ± 126) 0.016 ± 0.003 0.440 ± 0.004 (63.8 ± 0.58) 0.124 ± 0.007
S′ 31.02 ± 1.12 (4498 ± 162) 0.021 ± 0.005 0.648 ±0.007 (94.7 ± 1.0) 0.148 ± 0.010
S+A 27.34 ± 0.67 (3964 ± 97) 0.028 ± 0.006 0.763 ± 0.004 (110 ± 1) 0.193 ± 0.011
S′ + A 30.56 ± 1.43 (4431 ± 207) 0.042 ± 0.013 1.279 ± 0.011 (185 ± 2) 0.231 ± 0.024
Concrete
R 10.36 ± 0.48 (1502 ± 70) 0.00024 ± 0.00006 0.00024 ± 0.00004 (0.035 ± 0.005) 0.0014 ± 0.0004
S+R 24.94 ± 0.71 (3616 ± 103) 0.014 ± 0.004 0.351 ± 0.005 (50.9 ± 0.7) 0.112 ± 0.006
S′ + R 29.83 ± 0.51 (4325 ± 74) 0.019 ± 0.007 0.568 ± 0.009 (82.3 ± 1.3) 0.141 ± 0.011
S+A+R 26.11 ± 1.11 (3785 ± 160) 0.031 ± 0.009 0.813 ± 0.012 (11.8 ± 1.7) 0.194 ± 0.022
S′ + A + R 27.81 ± 2.01 (4032 ± 291) 0.039 ± 0.008 1.079 ± 0.035 (156 ± 5) 0.219 ± 0.041

causing the loss modulus and the dissipation to follow the
same behavior of increase followed by decrease. The posi-
tive effects of the silane treatment on the loss modulus and
the fraction of energy dissipated are less for concrete than
paste due to the relatively small positive effect of silane
treatment on the storage modulus of concrete. The positive
effects of silane treatment on the loss modulus and the frac-
tion of energy dissipated are greatest for mortar.
Table 3 also shows the effects of the silane treatment in the
presence of steel. The effects are all positive, but they tend
to be less than those in the absence of steel, particularly in
relation to the fraction of energy dissipated.

284 ACI Materials Journal/May-June 2013

Effects of silica fume, graphite network cementitious
admixture, and steel in various combinations
Table 4 shows the effects of silica fume, graphite network
cementitious admixture, and steel in various combinations
on the dynamic flexural properties of cement paste, mortar,
and concrete. Silica fume (whether silane-treated or not) is
much more effective than the graphite network cementitious
admixture (A in Table 4) for increasing the storage modulus
but is much less effective than the cementitious admixture
for increasing the loss tangent.
The combined use of silica fume and the graphite network
cementitious admixture (SA in Table 4) provides a syner-
gistic effect so that both storage modulus and loss tangent
are high. When the silica fume is silane-treated, the effect
is even stronger, mainly due to the increase in storage
modulus due to the silane treatment. As a consequence of
the synergistic effect, the loss modulus is higher than what
can be provided by any of the admixtures used as the sole
admixture. This synergistic effect is due to the silica fume
enhancing the storage modulus, while the graphite network
cementitious admixture enhances the loss tangent.
The cement paste with the graphite network admixture as
the sole admixture exhibits a storage modulus of 1.9 GPa
(276,000 psi), which is close to that of the cement paste
without admixture (1.8 GPa [260,000 psi]; Table 4) and is
much lower than the value of 9.3 GPa (1,400,000 psi) for
the graphite network cementitious admixture itself (not as an
admixture).13 This means that the cementitious admixture is
not effective for stiffening, but the admixture itself is much
stiffer than plain cement paste.
The cement paste with the graphite network admixture as
the sole admixture exhibits a loss tangent of 0.13 (Table 4),
which is higher than the value for plain cement paste (0.03;
Table 4) and that for cement paste containing silica fume
as the sole admixture (0.055 or 0.061; Table 4) but is lower
than the value of 0.81 for the graphite network admixture
itself.11 This means that the graphite network admixture
is effective for enhancing the loss tangent, although the
enhanced loss tangent remains low compared to the value of
the admixture itself.
The cement paste with the graphite network cementitious
admixture as the sole admixture exhibits a loss modulus of
0.25 GPa (36,000 psi) (Table 4), which is comparable to
or lower than the value for cement paste containing silica
fume as the sole admixture (0.23 or 0.44 GPa [33,000 or
64,000 psi]; Table 4) and is much lower than the value of
7.5 GPa (1,100,000 psi) for the graphite network cementi-
tious admixture itself11 but is much higher than the value for
plain cement paste (0.05 GPa [7000 psi]; Table 4).
Figure 2 shows an optical microscope photograph of
the mechanically polished surface of the cement paste
containing untreated silica fume and the graphite network
cementitious admixture. Due to the cementitious nature of
this admixture, the interface between this admixture and the
silica fume cement paste matrix shows no pore or gap. The
cement matrix is continuous from the silica fume cement
paste region to the graphite network admixture region.
Table 4 also shows that steel has little effect on the
storage modulus or the loss tangent. The addition of the
graphite network admixture to steel-reinforced cement paste
increases the loss tangent without much effect on the storage
modulus. The further addition of silica fume increases the
storage modulus significantly and increases the loss tangent
slightly, such that the effects are stronger when the silica
ACI Materials Journal/May-June 2013
Fig. 2—Optical microscope photograph of polished surface
of cement paste containing untreated silica fume and
graphite network cementitious admixture. Part of edge of
piece of this admixture is shown at right lower portion of
photo. Bright regions are graphite, which has probably been
smeared by polishing, as suggested by slightly shiny appear-
ance of admixture edge. Dark regions are cement.
fume has been silane-treated. The highest fraction of energy
dissipated among the cement pastes in Table 4 is 0.26, as
provided by steel-reinforced cement paste containing silane-
treated silica fume and the graphite network admixture.
The effects of silica fume, graphite network admixture, and
steel for cement pastes, mortars, and concretes are similar
(Table 4). The storage modulus is much higher for mortar
than the corresponding paste due to the stiffening by the sand.
The storage modulus is higher for concrete than the corre-
sponding mortar due to the stiffening by the stones. The loss
modulus and dissipation are particularly high for the mortars.
The highest fraction of energy dissipated among the mortars
is 0.58, as provided by steel-reinforced mortar containing
silane-treated silica fume and the graphite network admix-
ture. However, the loss tangent is lower for concrete than the
corresponding mortar, thus resulting in a low loss modulus
and low dissipation for the concretes. The highest fraction of
energy dissipated among the concretes is 0.23, as provided
by concrete containing silane-treated silica fume and the
graphite network admixture; the value is essentially the same
whether the concrete contains steel or not. Silane-treated
silica fume and the graphite network admixture together
cause steel-reinforced concrete to increase the energy dissi-
pation, loss modulus, loss tangent, and storage modulus by
16,000%, 450,000%, 16,000%, and 170%, respectively.
Comparison of all cement-based materials studied
in this work
Table 5 shows the fraction of energy dissipated for cement
pastes, mortars, and concretes in comparison. Without addi-
tional ingredients (the “none” case), the fraction of energy
dissipated decreases from paste to mortar and concrete. With
additional ingredients (the rest of the columns in Table 5), the
fraction of energy dissipated increases from paste to mortar
and decreases from mortar to concrete. Thus, the highest
fraction of energy dissipated occurs in mortars. This fraction
increases from paste to mortar because of the high storage
modulus of mortar. This fraction decreases from mortar to
concrete because of the low loss tangent of concrete.
285
Table 5—Fraction of energy dissipated (0.2 Hz) for cement pastes, mortars, and concretes in comparison.
Abbreviations for additional ingredients are as explained in Table 4
None S′ + A R S+R S′ + R S′ + A + R
Pastes 0.032 ± 0.006 0.248 ± 0.019 0.033 ± 0.008 0.123 ± 0.077 0.163 ± 0.072 0.258 ± 0.008
Mortars 0.0028 ± 0.0002 0.577 ± 0.042 0.0071 ± 0.0004 0.291 ± 0.006 0.321 ± 0.005 0.512 ± 0.028
Concretes 0.0013 ± 0.0003 0.231 ± 0.024 0.0014 ± 0.0004 0.112 ± 0.006 0.141 ± 0.011 0.219 ± 0.041

Table 6—Fractional increase in dissipation (0.2 Hz), which refers to fraction of energy dissipated.
Abbreviations for additional ingredients are as explained in Table 4
Additional ingredients S′ + A R S+R S′ + R S′ + A + R
Pastes, % 680 0 280 400 710
Fractional increase in dissipation
Mortars, % 21,000 150 10,000 11,000 18,000
relative to “none” case
Concretes, % 18,000 0 8400 11,000 17,000
Pastes, % — — 270 380 690
Fractional increase in dissipation
Mortars, % — — 4000 4400 7100
relative to “R” case
Concretes, % — — 8000 9900 16,000

Table 7—Static flexural and compressive strengths of cement pastes, mortars, and concretes, showing
effects of silica fume (S is untreated silica fume and S′ is silane-treated silica fume), graphite network
admixture (A), and steel reinforcing bar (R) separately and in combination
Compressive strength without steel, Flexural strength without steel, Flexural strength with steel,
Additional ingredient(s) MPa (psi) MPa (psi) MPa (psi)
None 54.2 ± 2.1 (7860 ± 300) 3.84 ± 0.16 (556 ± 23) 7.44 ± 0.20 (1080 ± 30)
S 63.4 ± 3.2 (9190 ± 460) 5.15 ± 0.44 (747 ± 63) 9.71 ± 0.54 (1410 ± 80)
S′ 67.1 ± 2.5 (9730 ± 360) 5.30 ± 0.25 (768 ± 36) 9.98 ± 0.39 (1450 ± 60)
Paste
A 38.9 ± 3.1 (5640 ± 450) 3.21 ± 0.23 (465 ± 33) 7.03 ± 0.66 (1020 ± 100)
S+A 58.1 ± 4.5 (8420 ± 650) 4.23 ± 0.31 (613 ± 45) 9.66 ± 0.71 (1400 ± 100)
S′ + A 61.7 ± 2.6 (8950 ± 380) 4.67 ± 0.47 (677 ± 68) 9.36 ± 0.64 (1360 ± 90)
None 56.2 ± 5.2 (8150 ± 750) 5.58 ± 0.41 (809 ± 59) 8.30 ± 0.23 (1200 ± 30)
S 66.7 ± 3.8 (9670 ± 550) 7.13 ± 0.37 (1030 ± 50) 10.73 ± 0.43 (1556 ± 62)
Mortar S′ 69.8 ± 3.2 (10,100 ± 500) 7.68 ± 0.77 (1110 ± 110) 10.91 ± 0.41 (1582 ± 59)
S+A 60.2 ± 4.5 (8730 ± 650) 6.21 ± 0.56 (900 ± 81) 8.06 ± 0.61 (1170 ± 90)
S′ + A 63.5 ± 1.8 (9210 ± 260) 6.46 ± 0.62 (937 ± 90) 9.31 ± 0.66 (1350 ± 100)
None 65.2 ± 1.9 (9450 ± 280) 8.80 ± 0.11 (1280 ± 20) 9.93 ± 0.35 (1430 ± 50)
S 71.4 ± 2.2 (10,400 ± 300) 9.29 ± 0.24 (1350 ± 30) 11.33 ± 0.41 (1643 ± 59)
Concrete S′ 73.1 ± 3.0 (10,600 ± 400) 8.81 ± 0.33 (1280 ± 50) 11.45 ± 0.31 (1660 ± 45)
S+A 66.9 ± 4.7 (9700 ± 680) 8.39 ± 0.29 (1220 ± 40) 9.87 ± 0.40 (1430 ± 60)
S′ + A 69.2 ± 5.0 (10,000 ± 700) 8.68 ± 0.25 (1260 ± 40) 10.33 ± 0.39 (1498 ± 57)

Table 6 shows that the fractional increase in the fraction
of energy dissipated due to the additional ingredient(s) is
much higher for mortar and concrete than the corresponding
cement paste, except for the case in which steel is the only
additional ingredient (that is, the “R” case, in which the frac-
tional increases are small for paste, mortar, and concrete).
The fractional increase tends to be slightly higher for mortar
than the corresponding concrete.
Table 6 also shows that the fractional increase in the
fraction of energy dissipated for cement-based materials
containing steel due to the additional ingredients other
than steel (that is, relative to the “R” case) increases from
paste to the corresponding mortar and the corresponding
concrete. The highest fractional increase of 16,000% is
obtained in concrete by using silane-treated silica fume with
the graphite network admixture. The effect of steel on the
loss modulus is negligible for all the formulations studied
(Table 4) whether the silica fume and graphite network
admixture are used or not.
Table 7 shows that the flexural and compressive strengths
are increased by the silica fume addition, whether aggregates
are present or not and, in relation to the flexural strength,
whether steel is present or not. The effects are stronger
when the silica fume is silane-treated, except for the flexural
strength of concrete without steel. However, the addition
of the graphite network admixture (A in Table 7) decreases
the flexural and compressive strengths, as shown for cement
pastes with and without steel under flexure and for cement
pastes without steel under compression. The resulting
flexural strength is much lower than the value of 15 MPa

286 ACI Materials Journal/May-June 2013

Table 8—Ranking of energy dissipation performance (0.2 Hz) of cement-based materials of this work with
consideration of cement pastes, mortars, and concretes (with and without steel). Top four performers are
all mortars. Ingredient abbreviations are as explained in Table 4
Dissipation ranking 1 2 3 4
Overall ranking 3 4 1 2 Baseline mortar Baseline concrete
Ingredients S′ + A S+A S′ + A + R S+A+R R R
26.71 ± 0.83 21.01 ± 0.83 25.44 ± 1.13 22.14 ± 1.42 8.01 ± 0.34 10.36 ± 0.48
Storage modulus, GPa (ksi)
(3872 ± 120) (3046 ± 120) (3688 ± 163) (3210 ± 205) (1161 ± 49) (1502 ± 70)
Loss tangent 0.139 ± 0.021 0.134 ± 0.015 0.102 ± 0.015 0.092 ± 0.018 0.00065 ± 0.00011 0.00024 ± 0.00006
3.543 ± 0.017 2.821 ± 0.012 2.587 ± 0.016 2.223 ± 0.025 0.00519 ± 0.00003 0.00024 ± 0.00004
Loss modulus, GPa (ksi)
(513 ± 2) (409 ± 2) (375 ± 2) (322 ± 4) (0.753 ± 0.004) (0.035 ± 0.005)
Fraction of energy dissipated 0.5877 ± 0.042 0.541 ± 0.062 0.512 ± 0.028 0.488 ± 0.033 0.0071 ± 0.0004 0.0014 ± 0.0004
6.46 ± 0.62 6.21 ± 0.56 9.31 ± 0.66 8.06 ± 0.61 8.30 ± 0.23 9.93 ± 0.35
Flexural strength, MPa (psi)
(937 ± 90) (900 ± 81) (1350 ± 96) (1169 ± 88) (1203 ± 33) (1430 ± 51)
Compressive strength 63.5 ± 1.8 60.2 ± 4.5 63.5 ± 1.8 60.2 ± 4.5 56.2 ± 5.2 65.2 ± 1.9
without steel, MPa (psi) (9210 ± 260) (8730 ± 650) (9210 ± 260) (8730 ± 650) (8150 ± 750) (9450 ± 270)

Table 9—Effect of frequency of dynamic flexural properties of cement pastes

Fractional increase Fraction decrease Fraction Fractional decrease
Additional Frequency, Storage modulus, in storage modulus Loss in loss tangent Loss modulus, of energy in dissipation
ingredient(s) Hz GPa (ksi) relative to 0.2 Hz tangent relative to 0.2 Hz GPa (ksi) dissipated relative to 0.2 Hz
1.81 ± 0.12 0.030 ± 0.054 ± 0.001 0.032 ±
0.2 — — —
(262 ± 17) 0.009 (7.8 ± 0.1) 0.006
1.96 ± 0.08 0.008 ± 0.016 ± 0.001 0.009 ±
None 1.0 8.3% 73% 72%
(284 ± 12) 0.002 (2.3 ± 0.1) 0.001
2.11 ± 0.14 0.0018 ± 0.004 ± 0.001 0.002 ±
2.0 17% 94% 94%
(306 ± 20) 0.0006 (0.6 ± 0.1) 0.001
4.18 ± 0.20 0.055 ± 0.230 ± 0.002 0.104 ±
0.2 — — —
(606 ± 29) 0.011 (33.4 ± 0.3) 0.011
Untreated 4.22 ± 0.20 0.022 ± 0.093 ± 0.002 0.066 ±
1.0 1.0% 60% 37%
silica fume (611 ± 29) 0.008 (13.5 ± 0.3) 0.011
4.46 ± 0.13 0.016 ± 0.071 ± 0.001 0.044 ±
2.0 6.7% 71% 58%
(646 ± 19) 0.004 (10.3 ± 0.1) 0.008
7.13 ± 0.54 0.061 ± 0.435 ± 0.004 0.256 ±
0.2 — — —
(1030 ± 80) 0.008 (63.1 ± 0.58) 0.042
Silane-
7.25 ± 0.36 0.032 ± 0.232 ± 0.003 0.110 ±
treated silica 1.0 1.7% 48% 57%
(1050 ± 50) 0.005 (33.6 ± 0.44) 0.035
fume
7.29 ± 0.48 0.019 ± 0.139 ± 0.003 0.069 ±
2.0 2.2% 69% 73%
(1060 ± 70) 0.004 (20.2 ± 0.44) 0.021

(2200 psi) for the graphite network admixture itself11 and the
resulting compressive strength is much lower than the value
of 280 MPa (41,000 psi) for the graphite network admix-
ture itself.11 This suggests that continuity in the graphite
network, which occurs in the admixture itself,11 is needed
for the network to be able to reinforce. On the other hand,
the combined use of silica fume and the graphite network
admixture results in flexural and compressive strengths that
are higher than those of the corresponding materials without
the admixtures, with the exception of concretes, for which the
strengths are comparable to those of the corresponding mate-
rials without the admixtures. Silane-treated silica fume tends
to be more effective than untreated silica fume, although the
difference is negligible when the data scatter is considered.
For all the pastes, mortars, and concretes, whether steel is
present or not (in the case of the flexural strength because
steel is not used in the compressive testing), the combined
use of silica fume (whether silane-treated or not) and the
graphite network admixture gives flexural and compressive
strengths that are comparable to or higher than those of the
corresponding materials without these admixtures. For the
same admixture(s), the flexural and compressive strengths
increase in this order: cement paste, mortar, and concrete,
but the effect on the flexural strength is relatively small when
steel is present.
A comparison of all the cement pastes, mortars, and
concretes—with and without steel—is discussed in the
following, taking into consideration the loss tangent, loss
modulus, and fraction of energy dissipated. The top four
performers—all mortars—are listed in Table 8. The material
that gives the highest energy dissipation performance is
mortar (without steel) containing silane-treated silica fume
and the graphite network admixture. Second in the dissipa-
tion ranking is the corresponding material with untreated
silica fume. Third is the material with silane-treated silica
fume, the graphite network admixture, and steel. Fourth
is the corresponding material with untreated silica fume.
However, the flexural strength is higher in the presence of
steel, such that the strength is higher when the silica fume
is silane-treated. Thus, the flexural strength is highest for

ACI Materials Journal/May-June 2013 287

the material that is ranked third in terms of the dissipation.
Because the fractional difference in dissipation performance
among the four materials in Table 9 is small compared to the
fractional difference in flexural strength, the material ranked
third in terms of the dissipation is the most highly ranked
material when both the static and dynamic mechanical prop-
erties are considered.
Compared to the baseline mortar and concrete (without
the admixtures), Table 8 shows that the top-performing
material according to the overall ranking is much superior in
the storage modulus, loss tangent, loss modulus, and fraction
of energy dissipated to the baseline materials. Moreover, it is
superior to the baseline mortar and comparable to the baseline
concrete in terms of the flexural and compressive strengths.
Due to the silane treatment of the silica fume, the material
of Overall Rank 1 is more expensive than that of Overall
Rank 2. The material of Overall Rank 1 provides dissipation
of 51%, whereas that of Overall Rank 2 provides dissipation
of 49%. Advantages of the material of Overall Rank 1 over
that of Overall Rank 2 include a higher storage modulus,
loss modulus, and flexural strength. The loss tangent and
compressive strength are comparable between Overall
Ranks 1 and 2. The fact that 50% of the input mechanical
energy is dissipated by the cement-based material is attrac-
tive. The dissipation provided by the baseline mortar or
concrete is below 1%.
The effect of the presence of steel on the loss modulus is
small. For example, the material of Overall Rank 1 dissi-
pates 51% of the energy, while the corresponding material
without steel (Overall Rank 3) dissipates 58% of the energy.
The fact that the presence of steel reduces the loss modulus
is related to the fact that the presence of steel reduces the
loss tangent from 0.14 to 0.10 and is because steel takes up
some volume that would have been occupied by the high-
damping cement-based material. Nevertheless, the presence
of steel increases the flexural strength from 6.5 to 9.3 MPa
(940 to 1350 psi), thus making the steel valuable.
The recommended design of the mixture to be poured to the
graphite network admixture and/or steel is a mortar mixture,
with the weight of the ingredients expressed relative to a
volume of 1 yd3 (0.76 m3): 1430 lb (649 kg) cement, 1430 lb
(649 kg) sand, 642 lb (291 kg) water, and 214 lb (97 kg)
silica fume. The total weight per yd3 is 3710 lb (1683 kg),
which corresponds to a density of 2.20 g/cm3 (137 lb/ft3).
This paper addresses materials rather than structures and
evaluates small-sized specimens. For practical applica-
tions, further work is necessary to evaluate the behavior of
larger specimens.
Effect of frequency
Table 9 shows the effect of frequency on the dynamic flex-
ural properties for cement pastes. As the frequency increases,
the storage modulus increases, the loss tangent decreases,
and the loss modulus decreases. These effects are consis-
tent with the viscoelastic nature and the associated strain
rate dependence. The fractional increase in storage modulus
due to the frequency increase is small compared to the frac-
tional decrease in loss tangent due to the frequency increase.
Hence, the decrease in dissipation with increasing frequency
is mainly due to the decrease in the loss tangent.
The fractional decreases in loss tangent and dissipation due
to the frequency increase are much smaller when silica fume
is present (Table 9). This means that the presence of silica
fume helps to reduce the detrimental effect of increasing
288
frequency, thereby allowing dissipation to remain substantial
at relatively high frequencies. This beneficial effect of silica
fume is attributed to the interfacial mechanism of damping
provided by the silica fume being able to operate at rela-
tively high frequencies. The interfacial mechanism domi-
nates in the presence of silica fume due to the small particle
size of the silica fume. Without silica fume, the interfacial
mechanism is less significant so that the conventional visco-
elastic mechanism dominates. The conventional viscoelastic
mechanism is apparently more sensitive to the frequency
than the interfacial mechanism. The effect of the silane treat-
ment of the silica fume on the frequency dependence is small
compared to the effect of the presence of silica fume.
Comparison with prior work
The highest loss tangent obtained is 0.14, which is
for mortar containing silane-treated silica fume and the
graphite network admixture. This is higher than the value
of 0.015 previously reported at 1 Hz for cement containing
Li5La3Ta2O12 particles19 and higher than the value of
0.062 for concrete containing silica particles of an average
diameter of 30 nm (1.2 × 10–6 in.).17 It is the same as the
value of 0.14 for cement containing carbon nanotubes.14 In
spite of the small size of the silica particles17 and the carbon
nanotube14 and the consequent large area of the interface
between the filler and the cement matrix, the loss tangent
is not higher than the highest value obtained. This reflects
the multi-layer structure in each ligament of the graphite
network, which provides interfaces inside the multi-layer.
However, the highest loss tangent obtained is smaller than
the value of 0.25 for cement mortar containing styrene-
acrylate admixture12 and the value of 0.81 at 0.2 Hz for the
graphite network admixture itself11; it is also small compared
to the value of 0.67 for rubber at the same frequency.20 It is
expected that the loss tangent of this work is smaller than
that of the graphite network admixture itself because the
material of this work contains this admixture in a limited
amount. Rubber and styrene-acrylate enhance the energy
dissipation by the viscoelastic deformation mechanism
rather than the interfacial mechanism. The two mechanisms
are competitive in relation to the loss tangent enhancement,
but the viscoelastic deformation mechanism tends to cause
a loss in stiffness.
The highest loss modulus obtained is 3.5 GPa (5.1 ×
105 psi), which is for the case of mortar containing silane-
treated silica fume and the graphite network admixture. This
value is higher than the value of 1.4 GPa (2.0 × 105 psi) for
cement containing styrene-acrylate admixture12 due to the
negative effect of the polymer admixture on the stiffness. The
value of this work is also higher than the value of 2.85 GPa
(4.1 × 105 psi) for concrete containing silica particles of an
average diameter of 30 nm (1.2 × 10–6 in.)17 due to the rela-
tively low value of the loss tangent provided by the silica
particle addition. On the other hand, the value of this work is
lower than the value of 7.5 GPa (1.1 × 106 psi) for the graphite
network admixture itself,11 as expected because the material
of this work contains this admixture in a limited amount.
Effective damping requires high values of both the loss
tangent and loss modulus. The graphite network admixture
itself gives loss tangents and loss moduli that are higher than
those of this work, but its elastic modulus is low.19 Concrete
containing silica particles with a diameter of 30 nm (1.2 ×
10–6 in.)17 gives loss tangents and loss moduli that are lower
than the highest dissipation material of this work. Cement
ACI Materials Journal/May-June 2013
mortar containing styrene-acrylate admixture12 gives a
higher loss tangent but a lower loss modulus than the highest
dissipation material of this work.
CONCLUSIONS
Silica fume (15% by mass of cement) and a graphite
network (8 vol.%, formed from exfoliated graphite) cementi-
tious plate admixture (with preferred orientation in the plane
of the specimen beam, 62 vol.% with the steel volume
excluded) are effective for enhancing the energy dissipa-
tion of cement-based materials under small-strain dynamic
flexure. The dissipation decreases with increasing frequency,
but silica fume reduces the frequency effect.
The fraction of energy dissipated reaches 0.26, 0.58, and
0.22 for cement paste, mortar, and concrete, respectively,
as provided by silane-treated silica fume and the graphite
network admixture, which cause steel-reinforced concrete
to increase the dissipation, loss modulus, loss tangent, and
storage modulus by 16,000%, 450,000%, 16,000%, and
170%, respectively.
The highest loss tangent and loss modulus obtained are
0.14 and 3.5 GPa (2.0 × 104 and 5.1 × 105 psi) respectively,
which are for the case of mortar containing silane-treated silica
fume and the graphite network admixture. These admixtures
cause steel-reinforced mortar to increase the dissipation by
7100% and cause steel-reinforced cement paste to increase
the dissipation by 690%. Without steel, these admixtures
cause mortar to increase the dissipation by 21,000%.
Silane-treated silica fume without the graphite network
admixture causes steel-reinforced concrete to increase the
dissipation by 9900%; untreated silica fume gives a corre-
sponding 8000% increase.
Without steel or admixtures, the fraction of energy dissi-
pated decreases from cement paste to mortar and concrete.
With steel and/or admixtures, this fraction increases from
cement paste to mortar and decreases from mortar to
concrete. As a result, the highest fraction of energy dissipated
occurs in mortars. This fraction increases from cement paste
to mortar because of the relatively high storage modulus
of mortar. This fraction decreases from mortar to concrete
because of the relatively low loss tangent of concrete.
Silica fume addition increases the flexural and compres-
sive strengths, whereas the graphite network admixture addi-
tion decreases these strengths. However, for all the pastes,
mortars, and concretes—with or without steel—the combined
use of silica fume (whether silane-treated or not) and the
graphite network admixture gives flexural and compressive
strengths that are comparable to or higher than those of the
corresponding materials without these admixtures.
Based on static and dynamic properties, the most highly
ranked cement-based material obtained is steel-reinforced
mortar containing silane-treated silica fume and the graphite
network admixture. This material exhibits high energy dissi-
pation compared to the mortar or concrete without these
admixtures and exhibits flexural and compressive strengths
ACI Materials Journal/May-June 2013
that are superior or comparable to those of the mortar or
concrete without the admixtures.
REFERENCES
1. Chung, D. D. L., “Improving Cement-Based Materials by Using Silica
Fume,” Journal of Materials Science, V. 37, No. 4, 2002, pp. 673-682.
2. Pourkhorshidi, A. R.; Najimi, M.; Parhizkar, T.; Hillemeier, B.; and
Herr, R., “A Comparative Study of the Evaluation Methods for Pozzolans,”
Advances in Cement Research, V. 22, No. 3, 2010, pp. 157-164.
3. Saje, D.; Saje, F.; and Lopatic, J., “Compressive Strength of Concrete
Containing Silica Fume,” Journal of the Mechanical Behavior of Materials,
V. 19, No. 6, 2009, pp. 355-364.
4. Daou, F., and Piot, B., “Cement-Slurry Performance and Set-Cement
Properties vs. Microsilica Densification,” SPE (Society of Petroleum
Engineers) Drilling & Completion, V. 24, No. 4, 2009, pp. 590-598.
5. Xu, Y., and Chung, D. D. L., “Improving Silica Fume Cement by Using
Silane,” Cement and Concrete Research, V. 30, No. 8, 2000, pp. 1305-1311.
6. Xu, Y., and Chung, D. D. L., “Cement-Based Materials Improved
by Surface Treated Admixtures,” ACI Materials Journal, V. 97, No. 3,
May-June 2000, pp. 333-342.
7. Xu, Y., and Chung, D. D. L., “Silane-Treated Carbon Fiber for
Reinforcing Cement,” Carbon, V. 39, No. 13, 2001, pp. 1995-2001.
8. Liu, T., and Ou, J., “Effects of Silane-Treated Silica Fume on Damping
Property of Cement Mortar,” PRO 32: International Conference on
Advances in Concrete and Structures, V. 1, 2003, pp. 168-176.
9. Ou, J.; Liu, T.; and Li, J., “Analysis of the Damping Behavior and
Microstructure of Cement Matrix with Silane-Treated Silica Fume,”
Journal of Wuhan University of Technology, Materials Science Edition,
V. 21, No. 2, 2006, pp. 1-5.
10. Fu, X., and Chung, D. D. L., “Vibration Damping Admixtures for
Cement,” Cement and Concrete Research, V. 26, No. 1, 1996, pp. 69-75.
11. Muthusamy, S.; Wang, S.; and Chung, D. D. L., “Unprecedented
Vibration Damping with High Values of Loss Modulus and Loss Tangent,
Exhibited by Cement-Matrix Graphite Network Composite,” Carbon, V. 48,
No. 5, 2010, pp. 1457-1464.
12. Liu, T.-J., and Ou, J.-P., “Effects of Styrene-Acrylate Emulsion on
the Damping Ability and Hydration of Cement Matrix,” Key Engineering
Materials, V. 302-303, Environmental Ecology and Technology of
Concrete, 2006, pp. 550-555.
13. Wen, W., and Chung, D. D. L., “Enhancing the Vibration Reduction
Ability of Concrete by Using Steel Reinforcement and Steel Surface
Treatments,” Cement and Concrete Research, V. 30, No. 2, 2000,
pp. 327-330.
14. Duan, Z., and Luo, J., “Effect of Multi-Walled Carbon Nanotubes on
the Vibration-Reduction Behavior of Cement,” International Conference on
Smart Materials and Nanotechnology in Engineering, S. Du, J. Leng, and
A. K. Asundi, eds., 2007, pp. 64230R/1-64230R/6.
15. Anderson, S. H., and Chung, D. D. L., “Exfoliation of Intercalated
Graphite,” Carbon, V. 22, No. 3, 1984, pp. 253-263.
16. Chung, D. D. L., “Intercalate Vaporization during the Exfoliation
of Graphite Intercalated with Bromine,” Carbon, V. 25, No. 3, 1987,
pp. 361-365.
17. Zou, D.; Liu, T.; Teng, J.; Leng, J.; Asundi, A. K.; and Ecke, W.,
“Improving the Damping Ability by the Addition of Nano SiO2 to the
Concrete Materials,” Second International Conference on Smart Materials
and Nanotechnology in Engineering, 2009, pp. 74933C/1-74933C/9.
18. Malkin, A. Y., and Isayev, A. I., Rheology—Concepts, Methods, and
Applications, ChemTec, Toronto, ON, Canada, 2006, pp. 47-54.
19. Kan, Z. P.; Li, C.; Wang, X. P.; Lu, H.; and Fang, Q. F., “Damping
Properties of Li5La3Ta2O12 Ceramic Particulates Reinforced Cement
Composites,” Materials Science and Engineering A, V. 528, 2010,
pp. 780-783.
20. Fu, W., and Chung, D. D. L., “Vibration Reduction
Ability of Polymers, Particularly Polymethylmethacrylate and
Polytetrafluoroethylene,” Polymers & Polymer Composites, V. 9,
No. 6, 2001, pp. 423-426.
289
 Notes:

290 ACI Materials Journal/May-June 2013

 ACI MATERIALS JOURNAL
Title no. 110-M24
Effect of Leaching on pH of Surrounding Water
by David W. Law and Jane Evans
When concrete structures—such as pier supports—are placed
in water, they can have a detrimental effect on the surrounding
environment by causing the pH to rise. This rise in pH can harm
and kill animal and plant life. The concentration of hydroxyl ions
leached from concrete can be affected by a number of factors,
including cement type, shape of structure, ratio of surface area
and volume, and the flow of the water. This paper presents the
results of a research project that investigated three mixtures: 100%
ordinary portland cement (OPC), 30% pulverized fly ash (PVA),
and 65% ground-granulated blast-furnace slag (GGBS). Tests
were conducted in both stagnant and flowing water using a range
of specimen geometries and sizes. The results showed that the
mixture, volume/surface area, and geometry of the specimen can
affect both the rate of leaching and the cumulative number of moles
of hydroxyl ions leached.
Keywords: cement type; durability; geometry; leaching; pH; volume/
surface area.
INTRODUCTION
It is estimated that over 10 billion tonnes (11 billion tons)
of concrete is produced worldwide each year.1 A signifi-
cant portion of this concrete is used in the construction of
structures that are built in water, such as bridge supports
and piers. Construction in water can cause environmental
concerns, including the releases of ions—such as aluminum,
calcium, chromium, magnesium, sodium, potassium, and
zinc—into the surrounding water and the increase of the
pH in the water, due to leaching from the concrete. When
fresh water becomes highly alkali, it can become toxic to
fish. It injures the skin and gills of the fish by the produc-
tion of mucus on the gills, which interferes with respiratory
gases and oxygen uptake. Even small rises in pH can have
an adverse effect on aquatic life. For example, a change from
pH 7 to pH 8 will result in ammonia becoming 10 times more
toxic to fish.2,3 In the open sea and flowing river systems,
the released hydroxyl ions are rapidly diffused and diluted;
however, in small, slow-flowing streams, this may not be the
case, with possible adverse impacts on local aquatic species.
The pH of water will increase to a peak value as the
leachate diffuses into the water and then decreases again
as the released hydroxide ions react with chemicals in the
water. Setunge et al.4 showed that the pH of stagnant water,
which is subject to concrete leaching, will rise to approxi-
mately 11.5 and then fall to a steady state of approximately
9. Research by Sagues et al.5 in a small volume of water
showed that as calcium ions begin to leach out of the
concrete, the pH of the water rose until a terminal pH was
reached. Alonso et al.6 and Vernet et al.7 also reported a rise
in pH when conventional concretes, as well as high- and
ultra-high-performance concrete, were exposed to water.
Research has shown that a number of factors can affect
leaching and pH. Jain and Neithalath8 showed that leaching
of calcium ions was reduced by the use of blended cements,
compared to ordinary portland cement (OPC). It has been
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
shown by Torrenti et al.9 that there is a direct relationship
between creep and leaching and by Kamali et al.10 that the
water-cement ratio (w/c) is particularly important, as the greater
the ratio, the more water there is available in the concrete to
aid the leaching process. The volume-to-surface-area ratio
has also been shown to affect the rate of leaching because
the surface area will affect the specific solid/liquid exchange
surface, as shown by Tiruta-Barna.11 Geankoplis12 has stated
that as the surface area of the concrete increases, so would the
rate of leaching. However, most of the research to date13-15 has
focused on long-term effects, particularly with regard to the
containment of waste materials and the leaching of metal ions
rather than monitoring the pH directly.
If the leaching takes place in a river or stream as opposed
to stagnant water, then the pH would be expected to reach
a lower peak value because the leachate would disperse
through the water. However, the hydroxyl ions leached may
still have a detrimental effect on the local environment—the
flora and fauna. In flowing water, the speed and type of flow,
laminar, or turbulent can also affect the rate of leaching,
which will, in turn, affect the pH of the water.16 Hence, a
consequence of the flow being affected by the geometry is
that the geometry of the structure affects the rate of leaching
and the pH concentration.12
Extensive research has been carried out to investigate
the effect of leachate on ground quality and, as a result,
several test methods have been established. There are
three commonly used methods to investigate the impact of
concrete leaching into groundwater and soil16:
• Batch contact test—solid mass mixing into groundwater;
• Tank leaching test—solid mass submerged in ground-
water; and
• Column percolation test—seepage of water containing
leachate into soil.
Tank leaching tests can be adapted to investigate concrete
leaching in rivers and streams because this test considers a
solid mass, such as concrete, submerged in water. It has been
shown that the precision of tank leaching tests is good and
they are able to be reproduced to an acceptable level.17
There are two types of tank leaching tests: static and dynamic.
Static leaching tests are used to ascertain how different intrinsic
properties of the material affect leaching. Dynamic leaching
tests are used to determine how leaching is affected by the
materials interaction with a changing environment.
RESEARCH SIGNIFICANCE
When concrete is placed in rivers and streams, there are
significant risks of pollutants being released through the
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-273.R2 received April 22, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
291
David W. Law is a Lecturer at the Royal Melbourne Institute of Technology (RMIT Table 3—Chemical composition GGBS, as provided
University), Melbourne, Victoria, Australia. His research interests include the dura- by manufacturer
bility and performance of reinforced concrete structures, particularly electrochemical
monitoring and protection, the use of sustainable materials, and material properties Chemical composition Percentage in slag
of concrete. SiO2 35.43
Jane Evans is a Graduate Structural Engineer in the oil and gas industry in Aber- Al2O3 13.00
deen, Scotland. She received her master’s degree in civil engineering from Heriot-
Watt University, Edinburgh, UK. Her research interests include the effects of concrete Fe2O3 0.37
leaching on the pH of rivers and streams. CaO 40.52
MgO 8.00
Table 1—Mixture design of OPC and blended MnO 0.55
cements, kg/m3
Mn2O3 (calc) 0.61
OPC PFA 30% GGBS 65%
TiO2 0.50
OPC 400 280 140 4-
S 1.03
Sand 660 660 660
S2- 0.83
Aggregate 0.394 in. (10 mm) 560 560 560
SO3 0.50
Aggregate 0.787 in. (20 mm) 700 700 700
LOI 0.88
Water 173.8 160 160
IR 0.85
PFA — 120 —
C 0.13
GGBS — — 260
Cl 0.02
High-range water-reducing admixture — 1.54 1.54
Glass count Not available

Table 2—Chemical composition PFA, as provided

by supplier
Chemical composition Percentage in PFA
Sulfate
Chloride
CaO
Na2O
MgO
Al2O3
Fe2O3
SiO2
leaching processes that take place. If these pollutants build
up, then they may have an adverse effect on the aquatic life.
Previous research has shown that a number of factors can
affect the rate of leaching, including geometry, flow, and
mixture design. This paper reports an experimental program
investigating each of these factors. Three mixture designs
are investigated: 100% PC, 30% pulverized fly ash (PFA),
and 65% ground-granulated blast-furnace slag (GGBS).
Tests were conducted in both stagnant and flowing water
using a range of specimen geometries and sizes.
EXPERIMENTAL PROCEDURE
Mixture design
Experiments were undertaken on three mixture designs:
100% OPC concrete, 30% PFA blended concrete, and 65%
GGBS blended concrete (Table 1). The OPC concrete satis-
fied the requirements of BS EN 197-118 CEM 11/BV 32.5R,
which means that the sum of the reactive CaO and SiO2 in
the cement was at least 50% of its mass. The standard regu-
lated the MgO content of the cement to less than 5% of the
cement, and also guaranteed that the ratio of CaO/SiO2 was
less than 2.0. The PFA had a free CaO value of 8%. The
chemical composition of the PFA is given in Table 2. The
chemical composition of the GGBS is shown in Table 3. The
0.41
GGBS met the requirements of BS EN 15167-1:2006.19
A w/c between 0.35 and 0.43 was used for the OPC
0.00
concrete and 0.4 for the PFA and GGBS concretes. A high-
2.18 range water-reducing admixture was added to the PFA and
2.08 GGBS mixtures to achieve a suitable workability and a target
1.53
slump of 50 to 60 mm (1.97 to 2.36 in.). The water was varied
as necessary in the OPC specimens to achieve the required
22.42
slump. The OPC concrete had a target strength of 40 MPa
7.6 (5800 psi) and the blended concretes had a target of 32 MPa
56.02 (4640 psi). All specimens were tested in duplicate.
The samples were left to cure for 24 hours under damp
burlap and then demolded. Once the molds were removed,
the samples were placed in plastic tanks that were filled with
tap water covering the specimens. The tanks contained 45 L
(11 gal.) of water and were sealed with lids to keep envi-
ronmental pollutants from entering the water, although some
dissolving of carbon dioxide in the water would be expected,
as the tanks were opened during the measuring cycles.
The pH of the water in the tanks was measured using a pH
meter. The measurements were taken at fixed intervals over a
3-day period to observe the change in pH caused by leaching
during the first few days after casting. Readings were taken
at 15-minute intervals initially, and as the rate of pH change
reduced, at longer intervals. The hydroxyl ion concentration
was calculated from the pH and volume of water. The labo-
ratory temperature was monitored throughout and observed
to vary between ±1.5°C (±2°F) over the course of the
experiments. Kamali et al.10 demonstrated that a significant
temperature change of the order of 60°C (108°F) can make
a significant difference to the rate of leaching, but a change
of the order recorded herein is not expected to have an effect
on the rate of leaching.

292 ACI Materials Journal/May-June 2013

Table 4—Specimen dimensions
Series
Dimensions 1 2 3 4 5
Cube size, in. (m) 4.016 (0.102) 7.560 (0.192) 8.740 (0.222) 11.575 (0.294) 5.984 (0.152)
Volume (V), in.3 (m3) 64.8 (0.0012) 432.1 (0.0071) 667.7 (0.0109) 1550.1 (0.0254) 214.3 (0.0035)
Surface area (S), in.2 (m2) 16.13 (0.062) 57.15 (0.221) 76.39 (0.296) 133.98 (0.519) 35.81 (0.139)
V/S 0.017 0.032 0.037 0.049 0.025

Table 5—Compressive strength, MPa

Series
Series Series Series Series 5 and
1 and 2 3 and 4 2 and 3 Series 2 cylinder
Mixture stagnant stagnant stagnant flowing flowing
OP 48.0 ± 3.3 45.6 ± 2.3 — 43.6 ± 1.1 42.5 ± 1.0
PFA — — 36.0 ± 1.1 — —
GGBS — — 35.3 ± 0.7 — — Fig. 1—Schematic of gravity feed for flowing water
Note: 1 MPa = 145 psi. experiments.

Volume/surface area
A total of five sets of specimens were investigated. The
specimens were cubic, with different volume/surface ratios
(Table 4). Series 1 to 5 were cast using OPC concrete, while
PFA and GGBS specimens were cast for Series 2 and 3 only.

Flowing water
A gravity feed system was used to provide a water flow rate
of 1 L/h (Fig. 1). Testing was undertaken with Series 2 type
specimens using the OPC concrete mixture. A w/c of 0.5 was
used to achieve the required slump. The pH of the water in
the tank was again monitored at fixed time intervals. The
hydroxyl ion concentration was calculated from the pH, the
volume of water, and the rate of flow. Fig. 2—pH versus time, Series 2 and 3, OPC and blended
cements.
Geometry
The effect of geometry was investigated using cylindrical
specimens with a diameter of 5.433 in. (138 mm) and length
of 9.252 in. (235 mm). These cylinders had the same volume
and surface area of concrete as the Series 5 cubes. These
tests were undertaken in flowing water conditions. The
cylindrical specimens were cast with OPC.

EXPERIMENTAL RESULTS AND DISCUSSION

Mixture design
The 28-day compressive strength for each mixture is given
in Table 5. The results show that each mixture had achieved
the target strength.
The pH-versus-time plots for the Series 2 and 3 OPC, PFA,
and GGBS concrete specimens are shown in Fig. 2. The results
show that the OPC concrete specimens display the largest
initial increase in pH, reaching a value of 10 within 1 hour. Fig. 3—Moles OH– versus time, Series 2 and 3, OPC and
The pH then steadily rises with time before reaching a final blended cements.
pH of between 11 and 11.5, in agreement with the work of
Setunge et al.4 The PFA and GGBS specimens display a To account for slight variations in the volume of water
smaller initial rise in pH, the PFA specimens reaching a pH within each tank, due to variations in sample size, the
of 9 to 9.5 and the GGBS a value between 8.5 and 9 after an hydroxyl ion concentrations with time is shown in Fig. 3.
hour. Both the blended concrete samples increased further The data again show that the OPC specimens had a higher
with time, such that after 3 days, the PFA had achieved a initial leaching of hydroxyl ions. However, by the conclu-
similar pH to that of the OPC specimens, while the GGBS sion of the test, the PFA specimens had actually leached
specimens gave a final pH value just below 11. more hydroxyl ions than the OPC specimens. The GGBS

ACI Materials Journal/May-June 2013 293


Fig. 4—pH versus time, V/S specimens, OPC specimens.
Fig. 5—Moles OH– versus time, OPC specimens.
specimens had leached fewer hydroxyl ions than both the
OPC and PFA concrete specimens.
These results contrast with previous research,18 which had
shown that for long-term exposure, both PFA and GGBS
had a beneficial effect on reducing leaching. The variations
can be explained by the hydration process. In the long-
term leaching from concrete, the more-soluble Na+ and
K+ are leached, followed by dissolution of the portlandite
(Ca(OH)2) before, finally, the C-S-H gel. However, when
concrete is immediately exposed to the water, the hydra-
tion reaction is still ongoing, with different reaction mecha-
nisms for the OP, PFA, and GGBS mixture designs. In OPC
concrete, the cement undergoes hydration on addition of the
mixture water to form CSH gel and Ca(OH)2.20 As for long-
term leaching, it is anticipated that the more soluble Na+ and
K+ are leached first, leading to the high initial pH observed,
and then the Ca(OH)2 is subsequently leached from the
concrete. However, in the blended cements, the PFA and
GGBS are activated by the Ca(OH)2 produced by the hydra-
tion of the OPC.
The PFA is a pozzolanic material and is activated in a two-
stage process. The OPC is hydrated by the water and the PFA
activated by the Ca(OH)2 produced. The GGBS is a latent
cementitious material and the hydration is a three-stage
process.21,22 Immediately after mixing, the slag particles
are coated by aluminosilicate hydrates, which are imperme-
able to water. As the hydration of the OPC takes place, the
hydroxide ions generated break down the coating and acti-
vate the hydration of the slag. Finally, the pozzolanic reac-
tion takes place, in which calcium hydroxide is consumed to
form secondary calcium silicate hydrates.
294
As such, it is hypothesized that in the OPC concrete,
the Ca(OH)2 initially produced by the hydration can be
leached on exposure to water. In the blended concretes, the
Ca(OH)2 available is reduced due to the activation reactions
with the PFA and GGBS, the result of which is a greater
amount of Ca(OH)2 available initially in the OPC concrete,
which results in a larger quantity of hydroxyl being leached
from the concrete on exposure and, hence, a more rapid rise
in the pH.
The total number of hydroxyl ions leached is dependent on
the chemical composition of the cements. Both the PFA and
GGBS have a lower CaO and higher SiO2 content than OPC;
however, the PFA has a higher equivalent Na2O content,
while GGBS Na2O content is similar to the OPC.23 Roy and
Idorn24 have also reported that on hydration, GGBS gives a
lower quantity of soluble Ca(OH)2 compared to 100% OPC.
Thus, for a 65% replacement of the cement with GGBS,
it could be expected that the three-stage hydration mecha-
nism would account for a lower initial rise in pH and the
lower CaO coupled with the 65% replacement would give
a lower total amount of hydroxyl ions leached. For the 30%
PFA, the two-stage hydration accounts for the lower initial
rise in pH compared to OPC, but higher than the three-stage
GGBS hydration process. While the reduction in CaO would
be expected to lead to a lower total quantity of hydroxyl ions
leached, this may be compensated by the higher equivalent
Na2O content, resulting in the higher quantity of hydroxyl
ions recorded.
Volume/surface area
The pH-versus-time for the four OPC volume:surface area
(V/S) specimens are plotted in Fig. 4. The data show that for
the three largest cubes—V/S of 0.032, 0.037, and 0.049—
the pH rises to between pH 11 and 11.5, again in agreement
with Setunge et al.4 For the smallest cube—V/S of 0.017—
this rise is to just over pH 10. This cube has a significantly
lower total volume and hence a significantly lower quantity
of hydroxyl ions available for leaching. The fall in pH noted
after 30 hours can be attributed to carbonation of the water
from atmospheric CO2.
Figure 4 shows that the cubes with a V/S of
0.017 and 0.032 display a rapid rise—with a peak after
approximately 20 hours—while those of a V/S of 0.037 and
0.049 show a later peak between 50 and 80 hours. The data
would indicate that the V/S is affecting the rate at which
the hydroxyl ions are able to leach from the concrete. It is
hypothesized that the availability of the hydroxyl ions at the
surface is restricted due to the rate of diffusion of the ions
to the surface. Leaching ability is governed by the volume
of the specimens, the surface area, and the distance the ions
need to travel to be completely removed from the concrete
through leaching,11 which is in turn is dependent on the pore
structure.12 The volume will control the total quantity of ions
that can be leached. However, the geometry and surface area
may also impact on the rate of leaching if these restrict the
ability of the ions to be removed.
Assuming the pore structure is similar in all specimens
than for those specimens with a low V/S, the path length the
ions need to travel to the surface is shorter compared to the
larger blocks. In addition, the total number of hydroxyl ions
available will increase proportional to the volume of speci-
mens as the block sizes increase. The effect of the V/S ratio
is illustrated in Fig. 5 and 6—plots of the moles of OH– and
moles OH–/(V/S) versus time. As would be expected when
ACI Materials Journal/May-June 2013
Fig. 6—Moles OH–/(V/S) versus time, OPC specimens. Fig. 8—pH versus time, stagnant and flowing specimens.

Fig. 7—Moles OH–/(V/S) versus time, blended cement Fig. 9—Cumulative moles OH– versus time, stagnant and
specimens. flowing specimens.

plotting OH– versus time, the larger the specimen, the greater
the quantity of OH– ions (Fig. 5). However, for OH–/(V/S)
versus time (Fig. 6), the 0.037 specimen gives the highest
rate, followed by the 0.049 specimen, with the 0.032 and
0.017 giving similar results to Fig. 5. The results indicate
that the V/S has an impact on the rate of leaching, with the
rate of leaching from the 0.049 specimen being reduced
compared to the 0.037 specimen, indicating the V/S has an
impact on the overall rate of leaching. The blended cement
specimens (Fig. 7) show the same trends, with the 0.037,
Series 3, specimens giving a higher peak than the 0.032,
Series 2, specimens.

Flowing water
A comparison between same-sized specimens in stagnant
and flowing conditions is given in Fig. 8—pH versus time—
and Fig. 9—cumulative moles OH– versus time. The results
show that the pH of the stagnant system is always greater than
that of the flowing system. This would be expected due to
the renewal of the water in the flowing system. Both systems
show an initial rise in pH, as ions are leached from the speci-
mens, followed by a decrease. For the stagnant system, this is
due to carbonation of the water from the CO2 from the atmo-
sphere. For the flowing system, this is due to the hydroxyl
ions being removed from the system. The plot of the cumula-
tive moles, accounting for the rate of renewal in the flowing
system, shows that, initially, there is a higher rate of leaching
from the stagnant system, but after approximately 24 hours,
the greater quantity comes from the flowing system. Alonso
et al.6 also found that a higher rate of leaching was evident in
flowing tests over a prolonged period of time.
Fig. 10—Cumulative moles OH– versus time, cylinder
and cube.
Geometry
The cumulative moles of hydroxyl ions versus time are
shown in Fig. 10 for a cube and cylinder of the same surface
area and volume. The data show that the cylinder has a
higher concentration leached compared to the cube. Over
the first 24 hours, the difference is relatively small, with the
cylinder having a concentration of 0.029 moles and the cube
a value of 0.025. However, after 75 hours, this has changed
to 0.07 and 0.046 moles, respectively. These data would indi-
cate that the geometrical shape has a definite impact on the
number of moles leached. This variation may be due to two
factors, assuming that the pore structure and rate of diffusion
through the concrete is the same in both specimens. The first

ACI Materials Journal/May-June 2013 295

factor is the mean distance the hydroxyl ions must diffuse
to the surface. In the cylinder, this distance is less than in
the cube. The rate of leaching has been shown to be depen-
dent on the distance the ions need to travel to be completely
removed from the concrete through leaching.11
The second factor is the flow pattern of water around the
specimen. In the cylinder, more of the surface would be
exposed to water flowing at a higher speed compared to the
cube. It is surmised that the flowing water removes hydroxyl
ions from the region around the outer surface of the concrete,
maintaining a higher concentration gradient between the
hydroxyl ion concentration in the concrete pore water and
the external tank water, resulting in higher rates of leaching.
CONCLUSIONS
Based on the results of this experimental investigation, the
following conclusions are drawn:
1. The concrete mixture affects the rate and the total
concentration of hydroxyl ions leached. In OPC concrete, a
high initial rate is observed, while in PFA and GGBS blended
concrete, a lower initial rate is observed. After 3 days, the
total concentration of hydroxyl ions leached for PFA is higher
than OPC concrete and GGBS is lower than OPC concrete. A
similar peak pH is observed for both OPC and PFA concrete,
while the GGBS concrete yields a lower peak pH.
2. The volume, surface area, and V/S affects the rate of
leaching. The highest rate of leaching is observed for speci-
mens with V/S of 0.037.
3. A higher pH is observed for specimens in stagnant water
than for those in flowing water. The cumulative quantity of
hydroxyl ions leached is also higher for the stagnant speci-
mens in the initial 24 hours, but after 3 days of exposure, a
higher cumulative rate is observed for those specimens in the
flowing water.
4. The geometry of the specimen can affect the total
number of hydroxyl ions removed and, to a lesser degree,
the rate of leaching. Cylinders displayed higher total rates
of leaching and cumulative concentration of hydroxyl ions
leached cubes.
ACKNOWLEDGMENTS
The authors would like to acknowledge ScotAsh for the supply of the
PFA from Longanett Power Station and the Cementitious Slag Makers
Association for the supply of the GGBS from the Scunthorpe works.
REFERENCES
1. Meyer, C., “The Greening of the Concrete Industry,” Cement and
Concrete Composites, V. 31, 2009, pp. 601-605.
2. Alabaster, J. S., and Lloyd, R., “Water Quality Criteria for Freshwater
Fish,” European Inland Fisheries Advisory Commission Report, Butter-
worth, 1980.
3. Dourdoroff, P., and Katz, M., “Critical Review of Literature on the
Toxicity of Industrial Wastes and Their Components I: Alkalis, Acids
and Inorgainic Gases,” Sewage and Industrial Wastes, V. 22, 1950,
pp. 1432-1458.
296
4. Setunge, S.; Nguyen, N.; Alexandar, B.; and Dutton, L., “Leaching
of Alkali from Concrete in Contact with Waterways,” Water Air and Soil
Pollution Focus, V. 9, 2009, pp. 381-391.
5. Sagues, A. A.; Moreno, E. I.; and Andrade, C., “Evolution of pH
during the In-Situ Leaching in Small Concrete Cavities,” Cement and
Concrete Research, V. 27, No. 11, 1997, pp. 1747-1759.
6. Alonso, C.; Castellote, M.; Llorente, I.; and Andrade, C., “Ground
Water Leaching Resistance of High and Ultra High Performance Concretes
in Relation to the Testing Convection Regime,” Cement and Concrete
Research, V. 32, 2006, pp. 1283-1594.
7. Vernet, C.; Alonso, C.; Andrade, C.; Castellote, M.; Llorente, I.;
and Hidelgo, A., “A New Leaching Test Based in Running Water System
to Evaluate Long Term Resistance of Concretes,” Advances in Cement
Research, V. 14, 2002, pp. 157-168.
8. Jain, J., and Neithalath, N., “Analysis of Calcium Leaching Behaviour
of Plain and Modified Cement Pastes in Pure Water,” Cement and Concrete
Composites, V. 31, 2009, pp. 176-185.
9. Torrenti, J. M.; Nguyen, V. H.; Colina, H.; Le Maou, F.; Benboud-
jema, F.; and Deleruyelle, F., “Coupling between Leaching and Creep of
Concrete,” Cement and Concrete Research, V. 38, 2008, pp. 816-821.
10. Kamali, S.; Moranville, M.; and Leclerqc, S., “Material and Environ-
mental Parameter Effects on the Leaching of Cement Pastes: Experiments
and Modelling,” Cement and Concrete Research, V. 38, 2008, pp. 575-585.
11. Tiruta-Barna, L., “Using PHREEQC for Modelling and Simulation of
Dynamic Leaching Tests and Scenarios,” Journal of Hazardous Materials,
V. 157, 2008, pp. 525-533.
12. Geankoplis, C., Transport Processes and Separation Process
Principles, fourth edition, Prentice Hall, Upper Saddle River, NJ, 2003,
1056 pp.
13. Andac, M., and Glasser, F. P., “Long-Term Leaching Mechansims of
Portland Cement-Stabilized Municipal Solid Waste Fly Ash in Carbonated
Water,” Cement and Concrete Research, V. 29, No. 8, 1999, pp. 179-186.
14. van der Sloot, H. A., “Comparison of Characteristic Leaching
Behaviour of Cements Using Standard (EN 196-1) Cement Mortar
and an Assessment of Their Long Term Environmental Behaviour in
Construction Products during Service Life and Recycling,” Cement and
Concrete Research, V. 30, 2000, pp. 1079-1096.
15. Clifford, N.; French, J.; and Hardisty, J., Turbulence: Perspectives
on Flow and Sediment Transport, John Wiley & Sons Ltd., Surrey, UK,
1993, 360 pp.
16. Chai, J.-C.; Onitsuk, K.; and Hayashi, S., “Cr(VI) Concentration from
Batch Contact/Tank Leaching and Column Percolation Tests Using Fly Ash
with Additives,” Journal of Hazardous Materials, V. 114, 2008, pp. 67-73.
17. Hohberg, I. G.; De Groot, A. M. H.; van der Veen, A.; and Wassing,
W., “Development of a Leaching Protocol for Concrete,” Waste Manage-
ment, V. 20, 2000, pp. 177-184.
18. BS EN 197-1:2001, “Cement. Composition, Specifications and
Conformity Criteria for Common Cements,” British Standards Institution,
London, UK, 2001.
19. BS EN 15167-1:2006, “Ground Granulated Blast Furnace Slag for
Use in Concrete, Mortar and Grout. Definitions, Specifications and Confor-
mity Criteria,” British Standards Institution, London, UK, 2006.
20. Roziere, E., and Loukili, A., “Performance-Based Assessment of
Concrete Resistance to Leaching,” Cement and Concrete Composites,
V. 33, 2001, pp. 451-456.
21. Neville, A., Properties of Concrete, third edition, Longman Group
Limited, Essex, UK, 1981, 792 pp.
22. Lea, F. M., Lea’s Chemistry of Cement and Concrete, P. C. Hewlett,
ed., fourth edition, Elsevier Butterworth-Heinmann, Oxford, UK, 2004,
1092 pp.
23. Canham, I.; Page, C. L.; and Nixon, P. J., “Aspects of the Pore Solu-
tion Chemistry of Blended Cements Related to the Control of Alkali Silica
Reaction,” Cement and Concrete Research, V. 17, 1987, pp. 839-844.
24. Roy, D. M., and Idorn, G. M., “Hydration, Structure and Properties
of Blast Furnace Slag Cements, Mortars and Concrete,” ACI Journal, V. 79,
No. 6, Nov.-Dec. 1982, pp. 445-457.
ACI Materials Journal/May-June 2013
 ACI MATERIALS JOURNAL
Title no. 110-M25
Novel Cementitious Binder Incorporating Cement Kiln Dust:
Strength and Durability
by Piyush Chaunsali and Sulapha Peethamparan
Fresh and hardened properties of a suite of cementitious binders
with cement kiln dust (CKD) as the main binding component (70%
by weight) are evaluated in this study. Two CKDs with different
chemical and physical properties were used in formulating CKD-fly
ash (FA) and CKD-slag mixtures without portland cement. The
setting time, workability, and strength development of these pastes
were evaluated first and the best-performing binders were then
used as a component in making concrete. The strength and dura-
bility of such concrete is also evaluated. The CKD (I)-slag combi-
nation outperformed all other mixtures with respect to the strength
and durability performance under various curing conditions. The
elevated temperature curing was found to be essential for the
property development of the CKD-FA binder. The performance of
CKD containing mortar mixtures with respect to the delayed ettr-
ingite formation (DEF) and the alkali-silica reaction (ASR) was
also good. The CKD that exhibited better performance contained
5% free lime, 3% Na2Oeq alkali, and 10.6% SO3 with an average
particle size of 4 mm (0.00016 in.).
Keywords: alkali-silica reaction; cement kiln dust; delayed ettringite
formation.
INTRODUCTION
Cement kiln dust (CKD) is a by-product generated during
the cement manufacturing process that contains significant
amounts of alkali, chloride, sulfate, and free lime. Many
studies involving the use of CKD as an alternative binding
material have been undertaken in the past.1-7 The alkalis
present in CKD have been thought to activate alumino-
silicate-rich industrial by-products—for example, fly ash
(FA) and ground-granulated blast-furnace slag (GGBFS)—
similar to how commercial alkalis do in traditional alkali-
activated binders.5,6 An external supply of alkalis increases
the pH of the pore solution, which plays an important role in
dissolving the vitreous phases of pozzolanic and supplemen-
tary cementitious materials.8 A wide variation in the chemical
composition of CKD has been reported in previous studies
with regard to alkali, sulfate, and free lime contents.5,9 In
addition to a specific CKD’s alkali content, its free lime and
sulfate contents may also influence its interaction with FA
and slag. A secondary calcium-silicate-hydrate gel may also
be formed through a pozzolanic reaction between lime and
pozzolans.10 Furthermore, the glassy phase of alumina in
FA and slag reacts with external sulfates to form ettringite
(AFt), which can also contribute to early-age strength.11
Significant variation observed in the chemical composi-
tion of CKDs poses a serious limitation to their real-world
application. The free lime, sulfate, and alkali (Na2Oeq)
contents of CKDs have been found to vary in the ranges of
1% to 30%, 4% to 16%, and 1% to 8%, respectively.12 The
two CKDs used for this study have free lime, sulfate, and
alkali content within the reported limits. This study reports
the properties of CKD-FA and CKD-slag binders developed
by incorporating two different types of CKDs. The influence
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
of CKD composition on early-age properties and mechanical
strength development of CKD-based binders is investigated.
Moreover, durability performance of CKD-based binders in
concrete with respect to delayed ettringite formation (DEF)
and alkali-silica reaction (ASR) is also evaluated.
RESEARCH SIGNIFICANCE
The use of CKD as an alternate binding material for devel-
oping sustainable concrete has the advantage of saving the
land and landfilling costs. However, the variation in the
chemical composition of CKDs, in general, is a concern
when it comes to the application of this material. This study
attempts to evaluate the performance of two different types
of CKDs in the property development of CKD-incorpo-
rated binders with additional studies on the durability of
these binders. The results from this study will be useful
in assessing the potential of a particular CKD as a binder
component based on its chemical composition.
EXPERIMENTAL INVESTIGATION
Materials
The chemical compositions of two CKDs along with
Class F FA and GGBFS are given in Table 1. CKD (I) and
CKD (II) were generated in cement plants that use long
dry- and wet-process kiln technologies, respectively. This
difference in the processing technology is also reflected
in the chemical composition. The concentration of most
of the reactive oxide components was slightly higher in
CKD (I), except for CaO and K2O. A significant difference
in the alkali (Na2Oeq) (3.1% versus 6%) and sulfate (SO3)
(10.62% versus 7.69%) contents can be noted from the table
between the chemical compositions of CKD (I) and CKD
(II), respectively. Moreover, CKD (II) had a considerable
amount of chloride, which was nearly absent in CKD (I).
The total CaO content of both CKDs is close to the upper-
bound values reported in the literature for the CKD composi-
tion.12 However, the free lime contents of CKD (I) and CKD
(II) were determined to be approximately 5% and 1.5%,
respectively. Figure 1 presents the difference in mineralogy
of two CKDs used in this study. Calcium carbonate (CaCO3),
quartz (SiO2), anhydrite (CaSO4), and free lime (CaO) were
the major minerals present in both the CKDs. Further-
more, CKD (II) contained a significant amount of alkali in
the form of sylvite (KCl) and syngentie (K2CaSO4·2H2O),
as shown in Fig. 1. The particle size distribution of all the
materials used in the study is shown in Fig. 2. It is evident
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-299.R1 received April 29, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
297
ACI member Piyush Chaunsali is a Graduate Research Assistant in the Department from the figure that CKD (I) had an average particle size of
of Civil and Environmental Engineering at the University of Illinois at Urbana- approximately 4 mm (0.00016 in.), whereas CKD (II) and slag
Champaign, Urbana, IL. He received his BTech in civil engineering from the National had similar particle size distributions, with an average particle
Institute of Technology Warangal, Warangal, India, and his MS in civil engineering
from Clarkson University, Potsdam, NY. He is a member of ACI Committees 223,
size of approximately 7 mm (0.00027 in.) each. Class F FA
Shrinkage-Compensating Concrete, and 236, Material Science of Concrete. His was the coarsest among them all, with an average particle size
research interests include the development of sustainable cementitious materials and of approximately 11 mm (0.0004 in.), and CKD (I) was finer
early-age behavior of concrete. than CKD (II).
ACI member Sulapha Peethamparan is an Assistant Professor in the Department of
Civil and Environmental Engineering at Clarkson University. She received her BS from
Sample preparation
the Mahatma Gandhi University, Kerala, India; her MS from IIT Madras, Chennai, A preliminary set of experiments was conducted to
India; her MEng from the National University of Singapore, Singapore; and her PhD determine the optimal curing temperature conditions and
from Purdue University, West Lafayette, IN. She is a member of ACI Committees 123, mixture proportions for obtaining the maximum compres-
Research and Current Developments; 130, Sustainability of Concrete; 201, Durability
of Concrete; and 236, Material Science of Concrete. Her research interests include
sive strength of the CKD-based binders. The optimum
development and characterization of sustainable binders such as alkali-activated curing temperature and mixture proportions thus obtained
systems and high-volume slag and fly ash concrete, and hydration kinetics of portland were used for the rest of the study. Class F FA/GGBFS was
and oil-well cement. homogenized with CKD in dry conditions prior to mixing
with water. A constant water-binder ratio (w/b) of 0.40 was
Table 1—Chemical composition of materials used to prepare the CKD-FA/CKD-slag paste following the
ASTM C305 procedure. Immediately after preparing the
Chemical composition, % CKD (I) CKD (II) Slag Class F FA CKD/CKD-FA/CKD-slag paste, workability was measured
SiO2 14.55 11.69 36.00 50.20 using the flow test in accordance with ASTM C1437.
Al2O3 4.46 2.20 10.53 28.70 Furthermore, the setting time of the CKD-based binders
(pastes) was also determined using the Vicat needle, as
Fe2O3 2.11 2.93 0.67 5.72 described in ASTM C191. For determining the compres-
CaO 61.15 66.11 39.80 5.86 sive strength (in accordance with ASTM C109/C109M),
MgO 3.84 0.87 7.93 1.74
the pastes were poured into cubical molds with dimensions
of 50 x 50 x 50 mm (2 x 2 x 2 in.) and compacted using
SO3 10.62 7.69 2.11 — a table vibrator. The samples were then kept at an ambient
K 2O 3.45 7.24 0.15 — temperature for 24 hours before demolding. Afterward, the
samples were wrapped in aluminum foil and subjected to
Na2O 0.80 0.87 0.27 0.96
heat curing in a laboratory oven for 48 hours. At the end of
Cl — 0.62 — — heat curing, the samples were allowed to cool down to an
Na2O *
eq 3.10 6.00 0.32 0.96 ambient temperature for at least 30 minutes. The compres-
sive strength after 48 hours of heat curing was used to deter-
Free lime 5.00 1.50 — —
mine the optimal curing temperature and binder propor-
*
Na2O + 0.658K2O.
tions. Additional samples with optimum binder proportions
were also prepared to investigate the influence of saturated
lime water curing on the strength development of the CKD-

Fig. 1—X-ray diffraction patterns of CKD (I) and CKD (II)

powders (A is anhydrite; CC is calcite; L is free lime; Q is
quartz; S is sylvite; and Sg is syngenite). Fig. 2—Particle size distribution.

298 ACI Materials Journal/May-June 2013

Table 2—Initial and final setting times for different
binders (pastes)
Initial setting time, Final setting time,
Binder composition minutes minutes
100% CKD (I) 90 280
70% CKD (I)-30% slag 240 680
70% CKD (I)-30% FA 360 800
100% CKD (II) >1440 (24 hours) —
70% CKD (II)-30% slag 900 1250
70% CKD (II)-30% FA >1440 (24 hours) —
based binder. Two different curing regimes were selected:
1) heat curing and subsequent saturated lime water curing;
and 2) only saturated lime water curing. In addition to paste
samples, this study also investigated the mechanical prop-
erties of concrete, consisting of CKD-FA and CKD-slag
as the binder. The CKD-(I)-based binder was selected for
preparing concrete samples due to the higher strength of its
paste compared to the CKD-(II)-based binder, as discussed
in a later section. Cubes with dimensions of 50 x 50 x 50 mm
(2 x 2 x 2 in.) were prepared using CKD (I)-FA (70:30) and
CKD (I)-slag (70:30) as the sole binders with fine aggregate
(river sand) and natural gravel coarse aggregate (maximum
nominal size of 9.75 mm [3/8 in.]). The fine aggregate-to-
coarse aggregate ratio and w/b were taken as 1.0 and 0.4,
respectively, for all concrete mixtures.
RESULTS AND DISCUSSION
Early-age properties
Setting time and workability—The workability and setting
time of the CKD-based binders were determined to under-
stand the binders’ early-age behavior. Table 2 shows the
setting times for various CKD-based binders. The optimal
CKD fraction—that is, 70% by weight of the total binder,
resulting in the maximum compressive strength (as discussed
in a later section)—was used in preparing CKD-FA/CKD-
slag pastes. As can be seen in Table 2, CKD-(I)-based binders
exhibited shorter setting times compared to CKD-(II)-based
binders. Because CKD (I) had considerably higher amounts
of free lime and sulfate, it exhibited the shortest setting times
among them all, which can be attributed to finer particle size
distribution, the immediate conversion of available free
lime to calcium hydroxide, and formation of ettringite crys-
tals.6,13,14 Between the CKD (I)-FA and CKD (I)-slag pastes,
the latter had a shorter setting time.
It can be noted that the setting time for neat CKD (II)
paste, which had very small amounts of free lime and
sulfates, exceeded 24 hours. Moreover, CKD (II)-FA pastes
exhibited a similar setting time to that of the neat CKD (II)
paste by registering a setting time that exceeded 24 hours.
Only CKD (II)-slag mixtures resulted in a final setting time
of less than 24 hours. X-ray diffraction (XRD) studies (not
presented herein) showed an increase in ettringite forma-
tion in CKD-(I)-based binders compared to CKD-(II)-based
binders. It is believed that the low free lime and sulfate
content of CKD (II) delays the onset of setting in CKD-(II)-
based binders.
A flow test was conducted to assess the workability of
CKD-based binders. The CKD-(II)-based binders (pastes)
exhibited larger flow values than those of CKD-(I)-based
binders (Fig. 3). The rheology of the fresh paste is directly
influenced by the physical and chemical characteristics of
ACI Materials Journal/May-June 2013
Fig. 3—Workability (flow value) of CKD and CKD-based
binders (pastes).
its constituents. Physical characteristics include particle size
and shape. There was not a distinctive difference in particle
shape (both were irregular), but CKD (I) was finer than
CKD (II). The higher surface area of CKD (I) is expected to
reduce the flow by increasing the interparticle interactions.
Moreover, the higher proportion of free lime and sulfate
content in CKD (I) results in the precipitation of calcium
hydroxide and ettringite crystals, reducing the flow value.
In addition to the physical and chemical characteristics of
CKD, the spherical shape of Class F FA particles and the
dilution effect also result in the increase of workability (flow
value) of CKD (I)-FA and CKD (II)-FA pastes.
Compressive strength
Paste—Figure 4(a) shows the variation of compressive
strength of the CKD-(I)-based binder with curing temper-
ature for a constant curing duration of 48 hours. Initially,
the optimization of curing temperature and CKD fraction
was performed for CKD (I). As shown in Fig. 4(a), the
maximum compressive strength for CKD (I)-FA and CKD
(I)-slag pastes was obtained at 75°C (167°F). This curing
temperature was chosen for the rest of the experiments. It
was observed that temperatures higher than 75°C (167°F)
caused thermal cracking in the paste and resulted in lower
compressive strength of the binder. Once the optimum curing
temperature was determined, the optimum CKD fraction was
found through testing the samples having a variable CKD
dosage after subjecting them to heat curing at 75°C (167°F)
for 48 hours. The 70% CKD by weight in CKD (I)-FA and
CKD (I)-slag pastes achieved the maximum compressive
strength after 48 hours of heat curing (Fig. 4(b)). Except
at a 70% CKD fraction, CKD (I)-slag paste showed higher
compressive strength than CKD (I)-FA paste. Interestingly,
CKD (I)-FA had higher strength at the optimal CKD frac-
tion. Both of the binders showed a compressive strength of
more than 25 MPa (3623 psi) at the optimal CKD fraction.
To further examine the strength development of CKD-(I)-
based binders (pastes), all samples were cured in saturated
lime water. Figure 5 shows the strength development of
CKD-(I)-based binders (pastes) for different curing regimes.
It is evident that the effect of saturated lime water curing was
299
Fig. 4—Effect of: (a) curing temperature; and (b) CKD fraction
on compressive strength of CKD-(I)-based binders (pastes).
Fig. 5—Effect of curing regimes on strength of CKD-(I)-
based binders or pastes (H is heat curing; WH is without
heat curing; L is saturated lime water curing).
300
Fig. 6—Effect of: (a) CKD fraction; and (b) curing tempera-
ture on compressive strength of CKD (II)-FA binder (paste).
more pronounced in CKD (I)-slag binder than in CKD (I)-FA
binder. Furthermore, the CKD (I)-slag binder, which had not
undergone initial heat curing, also showed steady strength
development. Overall, the hydraulic nature of slag seems
to have contributed more toward the strength development
compared to FA.
Similar to CKD-(I)-incorporated mixtures, the optimum
CKD (II) content and the elevated temperature of curing in
CKD (II)-FA mixtures was found to be 70% (Fig. 6(a))
and 75°C (167°F) (Fig. 6(b)), respectively. It is acknowl-
edged that 70% may not be the optimum CKD fraction
for CKD (II)-slag binder. However, to reduce the variation
among mixtures, a common CKD proportion (that is, 70%),
temperature (75°C [167°F]), and curing duration (48 hours)
were used for CKD-(I)- and CKD-(II)-based binders. The
strength of CKD-(II)-based binders was found to be low
compared to CKD-(I)-based binders (Fig. 7). Similar to the
trend observed in CKD-(I)-based binders, CKD (II)-slag
paste showed a higher strength than CKD (II)-FA paste,
but all had compressive strengths of less than 20 MPa
ACI Materials Journal/May-June 2013
Fig. 7—Strength development of CKD-(II)-based binders or Fig. 8—Influence of paste volume on strength of concretes
pastes (H is heat curing; L is saturated lime water curing). incorporating CKD (I) after 48 hours of heat curing (water-
cement ratio [w/c] of 0.4; fine aggregate-to-coarse aggre-
gate ratio of 1.0).
(2900 psi). It appears that the low free lime and sulfate
content are responsible for the lower strength develop-
ment of CKD-(II)-based binders as compared to CKD-(I)-
based binders. Surprisingly, the neat CKD (II) paste had a
compressive strength of approximately 4 MPa (580 psi) after
the heat curing, but the specimens cracked and disintegrated
on subsequent saturated lime water curing, which prevented
the strength determination at later ages.
The free lime content of CKD plays an important role in the
pozzolanic reaction that forms a secondary calcium-silicate-
hydrate (C-S-H) gel.5,13,14 Moreover, sulfate content contrib-
utes significantly toward early-age strength by forming the
ettringite (AFt) phase.6,13,14 In addition, the fineness of CKD
also affects its reactivity. CKD (I), having more fineness
along with higher amounts of free lime and sulfate content,
seems to have developed higher strength compared to the
CKD (II) used in this study.
Concrete—After determining paste strength, additional
concrete test samples were prepared to evaluate the perfor-
mance of concrete containing these binders. Only the
CKD-(I)-based binders were used for making concrete due
to the higher strength development observed in the pastes Fig. 9—Influence of binder type and curing condition on
incorporating CKD (I). Figure 8 shows the variation of compressive strength of concrete (paste volume of 40%; w/b
compressive strength with the paste volume. The concretes of 0.40; fine aggregate-to-coarse aggregate ratio of 1.0).
incorporating CKD (I)-slag binder showed higher strength
compared to the concretes incorporating CKD (I)-FA binder ment was noticed in CKD (I)-FA binder concrete after 56 days
after 48 hours of heat curing. Figure 8 also indicates that of lime water curing. The CKD (I)-slag binder concrete
the compressive strength decreases with an increase in the attained a compressive strength of approximately 32 MPa
paste volume. This has also been reported in the case of (4638 psi), whereas the CKD (I)-FA binder concrete exhibited
conventional concrete.15,16 It is usually attributed to a longer approximately 18 MPa (2609 psi) at the end of 56 days. The
crack path when the paste volume is lower or the aggregate concretes incorporating CKD (I)-slag were observed to have
volume is higher, which forces the crack to move around a gained their strength without undergoing heat curing, which
large number of aggregates. This results in higher energy can be useful from the field application viewpoint.
absorption or higher strength. When the volume of the paste
is higher, the length of the crack path becomes smaller and, Microstructural examination
correspondingly, the amount of energy absorbed decreases.17 A morphological examination was performed on both
Figure 9 shows the effect of subsequent saturated lime types of CKD-based binders in a lower electron imaging
water curing on the strength development of concrete made (LEI) mode using scanning electron microscopy (SEM).
with two different types of binders (40% paste volume). Figure 10(a) shows a typical microstructure of CKD (I)-slag
Because the CKD (I)-slag paste had higher strength than paste. The microstructure appears to be dense and shows the
the CKD (I)-FA paste, a similar trend was observed in the presence of C-S-H gel, which has been discussed in detail
strength of concrete made with these two binders. Further- by the authors elswhere.14 However, the microstructure of
more, there was a 50% enhancement in the compressive strength CKD (II)-slag had more open reticular structure morphology
of CKD (I)-slag binder concrete, while only a 30% enhance- (Fig. 10(b)), indicating an inadequate formation of the

ACI Materials Journal/May-June 2013 301


Fig. 10—SEM micrographs of: (a) CKD (I)-slag; and (b)
CKD (II)-slag pastes after 48 hours of heat curing.
(Note: 1 mm = 0.0394 in.)
Fig. 11—Expansion of CKD-(I)-based binders after heat
curing in accordance with ASTM C1038/C1038M.
302
binding gel products compared to the microstructure of the
CKD (I)-slag system. As discussed in a previous section,
the compressive strength of CKD-(II)-based binders was
lower than that of CKD-(I)-based binders. It is noted that
the morphological observations discussed in this paper hold
for 75°C (167°F) curing temperature. The variation in the
curing temperature may influence the morphology and was
not investigated in this paper.
Durability studies on CKD-based binders
Both the CKDs used for formulating the cement-/clinker-
free binder contained significant amounts of sulfate and
alkali. Hence, it is important to assess the durability of
the new cementitious binder before it can be used as an
infrastructural material. In this regard, two main potential
issues—DEF and ASR—were investigated in CKD-(I)-
based binders. The efficacy of CKD-(II)-based binders could
not be examined due to slow strength development in these
binders, preventing the demolding of the mortar bars even at
the end of 7 days, indicating that CKD (II) may not be used
as a binding component in making concrete.
DEF—Concrete that had undergone high tempera-
ture curing (usually above 70°C [158°F]) was found to
be more susceptible to DEF on subsequent exposure to a
moist environment due to recrystallization of ettringite at
later ages.18,19 The sulfate content of the clinker has been
shown to have an influence on the expansion caused by
DEF.8 Therefore, the DEF susceptibility of CKD-based
binders was tested as per ASTM C1038/C1038M for two
main reasons: 1) heat curing (>70°C [158°F]); and 2) high
sulfate content of the CKD used in this study. In particular,
CKD (I)-FA and CKD (I)-slag binders, having 70% CKD,
were assessed. Three samples of mortar bars were prepared
in accordance with ASTM C1038/C1038M for each type of
binder and were cured at 75°C (167°F) for 48 hours. After
heat curing, mortar bars were immersed in saturated lime
water and the length measurements were taken in accor-
dance with ASTM C490/C490M.
Figure 11 shows the expansion of mortar bars incorpo-
rating CKD (I)-FA and CKD (I)-slag binders. At the end of
a 500-day exposure period, less than 0.02% expansion was
measured in CKD-(I)-based binders, exhibiting a low poten-
tial for DEF. The authors have found the presence of ettrin-
gite even after heat treatment in CKD-based binders.13,14 The
stability of ettringite at the elevated temperatures used in this
study might have reduced the susceptibility for DEF damage
on subsequent exposure to saturated lime water. Further-
more, the incorporation of FA and slag also appears to have
controlled the DEF expansion due to the presence of reac-
tive alumina in these mineral admixtures, which results in a
reduction of the SO3/Al2O3 ratio.20,21
ASR—A high alkali content of CKD may influence its
susceptibility to ASR, which is a predominant damage
phenomenon in portland-cement concrete. In this study,
two standard tests were performed in accordance with
ASTM C1260 and ASTM C227 to evaluate the potential
of ASR in CKD-based binders. Although ASTM C1260 is
intended to evaluate the potential reactivity of aggregates
for ASR, it can also be used to compare the performance of
plain and modified mixtures. The performance of CKD (I)
was evaluated in comparison with ordinary portland cement
(OPC). Figure 12(a) shows the expansion of mortar bars
prepared with reactive siliceous aggregate and immersed
in 1 N NaOH solution at 80°C (176°F). The expansion of
ACI Materials Journal/May-June 2013
OPC mortar bars after 14 days was observed to be more than
the threshold value—that is, 0.2%—which clearly confirms
the reactive nature of the aggregates. It is also evident
that CKD (I)-slag did not undergo rapid expansion, as expe-
rienced by the OPC mortar. Moreover, CKD (I)-FA did not
undergo any expansion, even at the end of the 14-day period.
It can be concluded that the CKD (I)-FA binder was more
effective in inhibiting the expansion during the accelerated
test method. Because ASTM C1260 is intended to evaluate
the reactivity of aggregates, another test (ASTM C227)
was performed to evaluate the potential reactivity of
CKD binder-aggregate combination. The binder used for
ASTM C227 was CKD (I)-slag because it had undergone
some expansion when tested according to ASTM C1260.
Figure 12(b) shows the expansion of mortar bars made with
CKD (I)-slag and reactive siliceous sand in a 38°C (100.4°F)
and 100% relatively humidity (RH) environment. At the end
of 6 months, the expansion of CKD (I)-slag mortar bars was
observed to be approximately 0.03%, which is less than
0.1% (a threshold limit in ASTM C33/C33M). It seems that
the potential for ASR is significantly lower in CKD-based
binders in spite of the higher alkali content of CKD.
The presence of pozzolan or slag has been suggested to
have controlled the pore solution alkalinity due to increased
alkali binding by the hydrates.22,23 The authors have also
observed the alkali binding in CKD-(I)-based binders
reported elsewhere.13,14 Furthermore, some previous studies
have shown that consumption of calcium hydroxide by
the pozzolanic reaction may also inhibit the expansion
due to ASR.24,25 In CKD-based binders, the portlandite is
consumed during heat curing due to the pozzolanic reac-
tion.13,14 Therefore, the consumption of calcium hydroxide
might also have played a role in inhibiting the expansion
in CKD-based binders; however, it appears that the alkali
binding and consumption of calcium hydroxide enhance the
resistance of these binders to ASR.

CONCLUSIONS Fig. 12—Expansion of mortar bars in accordance with:

This paper evaluates the performance of CKD-based
cement-free binders and the influence of the physico-chem-
ical characteristics of CKDs on the property development and
performance. Early-age properties, such as flow and setting
time, were significantly influenced by particle size distribu-
tion of CKD and the amount of calcium hydroxide and ettr-
ingite formation during CKD hydration. CKD (I), which had
higher amounts of lime and sulfate content, showed reduced
setting time and flow value, which can be attributed to the
higher amounts of calcium hydroxide and ettringite present in
CKD-(I)-based binder compared to CKD-(II)-based binder.
The addition of both FA and slag increased the workability of
CKD-based binders. The strength of CKD-(I)-based binders
was significantly higher than that of CKD-(II)-based binders.
Higher strength of CKD-(I)-based binders was a result of
higher calcium hydroxide and ettringite formation in the
CKD-(I)-based system. The formation of ettringite enhanced
the early-age strength, whereas the calcium hydroxide
resulted in the formation of C-S-H during the pozzolanic
reaction. The addition of slag resulted in significant improve-
ment in the strength of CKD-based binders. The insignificant
expansion (<0.02%) on exposure to moist curing indicates
its resistance to DEF, probably due to the thermal stability
of ettringite at 75°C (167°F). Furthermore, reduced calcium
hydroxide and alkali binding appear to have contributed to
the ASR resistance of CKD-based binders.
(a) ASTM C1260; and (b) ASTM C227.
This study reveals that the free lime and sulfate content
contributes to the formation of C-S-H (through pozzolanic
reaction) and ettringite. Both of these reaction products
contribute to the strength of CKD-based binders. The study
on CKD-(I)-based binders indicates their resistance against
DEF as long as ettringite is stable during the heat curing,
which limits the maximum temperature used for activation.
Furthermore, the high alkali content of CKD does not neces-
sarily make it a bad choice, probably due to the consumption
of calcium hydroxide and alkali binding when FA or slag is
used as an alumino-silicate source. This research has impli-
cations on understanding the novel binders incorporating
CKD which, in the future, may potentially be used in making
an environmentally friendly concrete.
REFERENCES
1. Pierce, C. E.; Tripathi, H.; and Travis, W. B., “Cement Kiln Dust in
Controlled Low-Strength Materials,” ACI Materials Journal, V. 100, No. 6,
Nov.-Dec. 2003, pp. 455-462.
2. Bhatty, M. S. Y., “Use of Cement Kiln Dust in Blended Cements,”
World Cement, V. 15, No. 4, 1984, pp. 126-134.
3. Abo-El-Enein, S. A.; Hekal, E. E.; Gabr, N. A.; and El-Barbary, M. I.,
“Blended Cements Containing Cement Kiln Dust,” Silicates Industrials,
V. 59, No. 9-10, 1994, pp. 265-269.

ACI Materials Journal/May-June 2013 303

 4. Lachemi, M.; Hossain, K. M. A.; Shehata, M.; and Thaha, W.,
“Controlled Low Strength Materials Incorporating Cement Kiln Dust from
Various Sources,” Cement and Concrete Composites, V. 30, No. 5, 2008,
pp. 381-392.
5. Konsta-Gdoutos, M. S., and Shah, S. P., “Hydration and Properties
of Novel Blended Cements Based on Cement Kiln Dust and Blast Furnace
Slag,” Cement and Concrete Research, V. 33, No. 8, 2003, pp. 1269-1276.
6. Wang, K.; Shah, S. P.; and Mishulovich, A., “Effects of Curing
Temperature and NaOH Addition on Hydration and Strength Develop-
ment of the Clinker-Free CKD-Fly Ash Binders,” Cement and Concrete
Research, V. 34, No. 2, 2004, pp. 299-309.
7. Wang, K.; Konsta-Gdoutos, M. S.; and Shah, S. P., “Hydration,
Rheology, and Strength of Ordinary Portland Cement (OPC)-Cement Kiln
Dust (CKD)-Slag Binders,” ACI Materials Journal, V. 99, No. 2, Mar.-Apr.
2002, pp. 173-179.
8. Taylor, H. F. W., Cement Chemistry, second edition, Thomas Telford,
London, UK, 1997, 459 pp.
9. Peethamparan, S.; Olek, J.; and Lovell, J., “Influence of Chemical and
Physical Characteristics of Cement Kiln Dusts (CKDs) on their Hydra-
tion Behavior and Potential Suitability for Soil Stabilization,” Cement and
Concrete Research, V. 38, No. 6, 2008, pp. 803-815.
10. Shi, C., “Studies on Several Factors Affecting Hydration and Proper-
ties of Lime-Pozzolan Cements,” Journal of Materials in Civil Engineering,
ASCE, V. 13, No. 6, 2001, pp. 441-445.
11. Xu, A., and Sarkar, S. L., “Microstructural Study of Gypsum Based
Fly Ash Hydration in Cement Paste,” Cement and Concrete Research,
V. 21, No. 6, 1991, pp. 1137-1147.
12. Peethamparan, S., “Fundamental Study of Clay-Cement Kiln Dust
(CKD) Interaction to Determine the Effectiveness of CKD as a Potential
Clay Soil Stabilizer,” PhD dissertation, Purdue University, West Lafayette,
IN, 2006, 14 pp.
13. Chaunsali, P., and Peethamparan, S., “Microstructural and Mineral-
ogical Characterization of Cement Kiln Dust-Activated Fly Ash Binder,”
Journal of the Transportation Research Board, V. 2164, 2010, pp. 36-45.
14. Chaunsali, P., and Peethamparan, S., “Evolution of Strength, Micro-
structure and Mineralogical Composition of a CKD-GGBFS Binder,”
Cement and Concrete Research, V. 41, No. 2, 2011, pp. 197-208.
15. Stock, F. S.; Hannant, D. J.; and Williams, R. I. T., “The Effect of
Aggregate Concentration upon the Strength and Modulus of Elasticity
304
of Concrete,” Magazine of Concrete Research, V. 31, No. 109, 1979,
pp. 225-234.
16. Popovics, S., “Analysis of the Concrete Strength versus Water-
Cement Ratio Relationship,” ACI Materials Journal, V. 87, No. 5, Sept.-
Oct. 1990, pp. 517-529.
17. Kolias, S., and Georgiou, C., “The Effect of Paste Volume and of
Water Content on the Strength and Water Absorption of Concrete,” Cement
and Concrete Composites, V. 27, No. 2, 2005, pp. 211-216.
18. Heinz, D., and Ludwig, U., “Mechanism of Secondary Ettrin-
gite Formation in Mortars and Concretes Subjected to Heat Treatment,”
Concrete Durability: Proceedings of Katharine and Bryant Mather Inter-
national Symposium, SP-100, J. M. Scanlon, ed., American Concrete Insti-
tute, Farmington Hills, MI, 1987, pp. 2059-2071.
19. Diamond, S., “Delayed Ettringite Formation—Processes and Prob-
lems,” Cement and Concrete Composites, V. 18, No. 3, 1996, pp. 205-215.
20. Ghorab, H. Y.; Heinz, D.; Ludwig, U.; Meskendahl, T.; and Wolter,
A., “On the Stability of Calcium Aluminate Sulfate Hydrates in Pure
Systems and in Cements,” Proceedings of the 7th International Congress
on the Chemistry of Cement, V. 4, Paris, France, 1980, pp. 496-503.
21. Tishmack, J. K.; Olek, J.; and Diamond, S., “Characterization of
High-Calcium Fly Ashes and Their Potential Influence on Ettringite Formation
in Cementitious Systems,” Cement, Concrete and Aggregates, V. 21, No. 1,
June 1999, pp. 82-92.
22. Thomas, M. D. A., and Innis, F. A., “Use of the Accelerated Mortar
Bar Test for Evaluating the Efficacy of Mineral Admixtures for Controlling
Expansion due to Alkali-Silica Reaction,” Cement, Concrete and Aggregates,
V. 21, No. 2, Dec. 1999, pp. 157-164.
23. Diamond, S., and Lopez-Flores, F., “Comparative Studies of the
Effects of Lignitic and Bituminous Fly Ashes in Hydrated Cement Systems,”
Proceedings of the MRS Symposium, Boston, MA, 1981, pp. 112-123.
24. Bleszynski, R. F., and Thomas, M. D. A., “Microstructural Studies
of Alkali-Silica Reaction in Fly Ash Concrete Immersed in Alkaline Solu-
tions,” Advanced Cement Based Materials, V. 7, No. 2, 1998, pp. 66-78.
25. Chatterji, S., “Chemistry of Alkali-Silica Reaction and Testing of
Aggregates,” Cement and Concrete Composites, V. 27, No. 7-8, 2005,
pp. 788-795.
ACI Materials Journal/May-June 2013
 ACI MATERIALS JOURNAL
Title no. 110-M26
Sustainable Processing of Cellulose Fiber Cement Composites
by Parviz Soroushian, Jong-Pil Won, and Maan Hassan
The aim of this study was to develop an efficient approach to the
processing of cellulose fiber-reinforced cement composites, which
makes value-added use of carbon dioxide. Two categories of cellu-
lose fiber cement boards were evaluated: pressed and unpressed.
Comparisons were made between the physical and mechanical
properties of cellulose fiber cement products fabricated with
conventional and CO2 curing using different processing conditions.
This paper describes the results of an attempt at the sustainable
processing of cementitious composite products—preferably envi-
ronmentally friendly—that incorporate cellulose fiber. Response
surface analyses of experimental results are used to identify
optimum curing conditions. Analysis results yielded the preferred
processing conditions of cellulose fiber cement boards. In both
pressed and unpressed boards, oven temperature and duration
have significant effects on the performance of the end product. CO2
curing in some conditions yielded better results when compared
with conventional curing, even at half the autoclave duration.
Keywords: accelerated curing; carbon dioxide gas; cellulose fiber; cement
composites; flexural strength.
INTRODUCTION
Cellulose fiber cement composites offer desirable longevity,
fire resistance, and life-cycle economy, as well as strength
and toughness characteristics.1-3 Cellulose fibers derived
from softwoods or hardwoods offer a highly cost-effective
means of reinforcement in thin cementitious products. Indus-
trial developments in this area have focused on the use of the
chemical (kraft) softwood fibers, while other fiber types (for
example, kraft hardwood or thermomechanical pulp) have
also performed satisfactorily in cement composites. Cellu-
lose fibers possess adequate stiffness, strength, and bonding
capacity to cement-based matrixes for substantial enhance-
ment of their flexural strength, toughness, and impact resis-
tance. These improvements are achieved through arrest of
cracks propagating in brittle cementitious matrixes by cellulose
fibers. The desirable technical qualities and low cost of cellu-
lose fibers have made them the reinforcing materials of choice
to substitute asbestos fibers in broadly used thin cementitious
products.4 Cellulose fiber-reinforced cement composites are
sensitive to moisture effects5,6; saturated composites possess
substantially increased toughness characteristics, while flex-
ural strength tends to be reduced upon wetting. They present
desirable dimensional stability and durability characteris-
tics.5,7 Upon aging, these composites actually gain strength
and stiffness but tend to lose ductility with time. Cellulose
fibers are also sensitive to calcium hydroxide; the high alka-
linity of the cementitious matrix causes partial dissolution
of chemical components of the fibers and hence reduces its
reinforcement capacity.8,9
The processing and properties of cellulose fiber-reinforced
cement composites are sensitive to the specific wood species
and the cellulose fiber type, among other factors. The setting
and hardening of cementitious materials is a relatively slow
process. Tremendous reduction in the set time of cementi-
tious materials can be achieved through exposure to carbon
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
Table 1—Experimental design (pressed board)
Carbonation chamber storage time: 1 hour (20% CO2
concentration) at 50°C (122°F) and 95% RH
Sample No. Oven temperature, °C (°F) (1 hour) Autoclave time, hours
1 50 (122) 4
2 50 (122) 8
3 100 (212) 4
4 100 (212) 8
dioxide (CO2) gas,10-14 which could represent great impact
on the sustainable development of the building industry due
to the consumption of the polluting CO2.15 The predominant
chemical reaction occurring in CO2 curing involves the reac-
tion of the Ca(OH)2 resulting from hydration of cement with
CO2 to produce CaCO3. This rapid carbonation reaction leads
to accelerated setting time and increased materials hardening,
which may help reduce the accelerated curing time.
Manufacturing of cellulose fiber-reinforced cement
composites, mixture proportioning, and processing methods
are quite different from normal concrete or mortar. The
process includes the refining of fibers in slurry and the mixing
of all constituents in the slurry. The slurry has a low solid
content (20%) to uniformly disperse the fibers; a vacuum is
then applied as the sheet is built up in laminates to extract the
excess water. The composite may then be compacted under
a press, and curing is usually achieved using high-pressure
steam (approximately 8 to 12 hours) for accelerated strength
gain in prefabrication facilities.14,16-18
RESEARCH SIGNIFICANCE
The main purpose of this study was to develop efficient
processing systems for cellulose fiber-reinforced cement
composites that use the advantages of CO2 curing, which
could represent great impact on the sustainable develop-
ment of the building industry due to the consumption of the
polluting CO2. The performance characteristics were evalu-
ated through flexural testing of composites. Value-added use
of carbon dioxide in this application benefits the produc-
tivity of cement panel production plants and also reduces
greenhouse gas emissions.
EXPERIMENTAL INVESTIGATION
A full factorial experimental design designated in this work
is presented in Table 1 (pressed board) and Table 2 (unpressed
board). Each case (pressed or unpressed) has two different
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-304.R1 received August 19, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
305
Parviz Soroushian, FACI, is a Professor of civil and environmental engineering
at Michigan State University, East Lansing, MI. He is a member of ACI Committee
549, Thin Reinforced Cementitious Products and Ferrocement, and Joint ACI-ASCE
Committee 408, Development and Splicing of Deformed Bars. He received his PhD
from Cornell University, Ithaca, NY, in 1983. His research interests include concrete
materials and technology.
Jong-Pil Won is a Professor of civil and environmental system engineering at Konkuk
University, Seoul, Republic of Korea. He received his MS in civil engineering from the
University of Missouri-Columbia, Columbia, MO, in 1988, and his PhD from Mich-
igan State University in 1995. His research interests include structural mechanics and
concrete materials.
ACI member Maan Hassan is an Assistant Professor of building and construction
engineering at the University of Technology, Baghdad, Iraq, where he received his PhD
in 2007. His research interests include concrete materials and technology.
The influential variables14,17,19 in the pressed case were the
oven temperature used for the predrying of the young sheets
prior to CO2 precuring and the autoclave duration. This
research investigated the effects of these variables through
a full factorial experimental design implemented with the
following variables: oven temperature, oven duration,
CO2 chamber duration, and autoclave duration. After drying
in the humid environment of the CO2 chamber, the pene-
tration of CO2 (possibly with water vapor) into the board
accomplishes the carbonation reaction, which provides the
board with improved qualities for autoclave curing. The low
processing duration associated with CO2 curing used in this
study yields significant economic benefits.

Materials and manufacturing procedures

The cellulose fiber selected for this investigation was
Table 2—Experimental design (unpressed board) southern softwood kraft pulp. The average length of this
Factor level Control fiber is 3.0 mm (0.118 in.). Type I portland cement with
Factor Code – + 1 2
the chemical composition shown in Table 3 was used in the
mixtures of this investigation. The fiber mass fraction and
50 100 50 matrix mixture proportions used in this phase of the study
Oven temperature, °C (°F) A
(122) (212) (122)
are introduced in Table 4. In the slurry-dewatering method
Duration in oven, hours B 1 2 1 of manufacture, thin-sheet specimens were formed from
Duration in CO2 chamber at 20% diluted slurry of approximately 20% solids. A relatively
1 small dosage of diluted flocculent (flocculent/cement =
CO2 concentration, hours (50°C C 1 4
(0% CO2)
[122°F] and 95% RH) 0.001 by weight) was added to achieve agglomeration of
Autoclave duration, hours D 4 8 4 8
cement particles to prevent passing of particles through the
filtering screens during dewatering. Fiber mass fraction, in
the slurry-dewatering method of manufacturing, is generally
Table 3—Chemical composition of Type I portland defined as the ratio of fibers to the dry constituents of the
cement used in this investigation matrix by weight.
Chemical composition Percent by weight Thin-sheet cellulose fiber-reinforced cement composites
Tricalcium silicate (C3S) 43.3 were manufactured as follows: 1) a weighted dry lap of
cellulose fiber was soaked in water for a minimum period
Dicalcium silicate (C2S) 26.3 of 4 hours; 2) a laboratory-scale pulp disintegrator was used
Tricalcium aluminate (C3A) 11.0 at a speed of 3000 revolutions per minute (rpm). The beating
Tetracalcium aluminoferrite (C4AF) 8.6
time used in this study was 10 minutes (yielding refined
cellulose fibers with a Canadian standard freeness of 500 ±
Insoluble residue 0.12 50); 3) the fibers, sand, flocculating agent, and cement were
proportioned; and 4) the ingredients were mixed in water
Table 4—Composition of cellulose fiber-reinforced to produce a slurry of 20% solids by weight; a high-speed
cement composite mixer was used to achieve a uniform dispersion of cellu-
Fiber type Softwood kraft pulp lose fibers and other mixture ingredients in the slurry. The
flocculating agent was the last solid constituent to be added,
Fiber mass fraction, % 8 which improves the binding of cement particles to cellulose
Sand-binder ratio (by weight) 0.75 fibers and controls the escape of cement particles during
Flocculent-cement ratio (by weight) 0.001
vacuum-dewatering of the slurry (when the excess water is
extracted). The extraction of water was actually performed
Silica fume-cement ratio (by weight) 0.75 in two stages. First, the excess water on top of the settled
slurry was withdrawn manually, and then the settled slurry
was put in a vacuum box 305 x 305 mm (12 x 12 in.) in planar
control conditions: Control 1 and Control 2. In the pressed dimensions. The slurry was evenly spread onto the screen of
case, one control condition involves: 1) making the board; the vacuum box and then the vacuum was applied at 381 mm
2) storing it for 1 day in a plastic bag and then removing (15 in.) Hg of mercury. The sheet was then removed on the
it from the plastic bag; and 3) autoclave-curing the board filter screen. The sheet, together with the screen, was stored
for 8 hours. The second control condition of the pressed case between two steel plates. Two sheets were made in this
was similar to the first, except the 24 hours of storage in a manner, with the first one being stored temporarily with its
plastic bag was replaced with 1 hour of storage in an oven at screen between steel plates in a sealed plastic bag. The screen
50°C (122°F), followed by 1 hour of steam curing at 50°C was then removed and the bottom face of the second sheet
(122°F) and 95% relative humidity (RH). For the unpressed was placed against the top face of the first sheet, as shown in
case, the two control conditions are introduced in Table 2. In Fig. 1. The sheets were then slightly pressed for 5 minutes at
both the pressed and unpressed cases, oven drying was used a pressure of 0.24 and 3.2 MPa (35 and 464 psi), simulating
to lower the moisture content of the board to the point where unpressed and pressed production conditions, respectively.
CO2 penetration and reaction would be facilitated. One board (consisting of two sheets) was produced each

306 ACI Materials Journal/May-June 2013

Fig. 1—Schematic of formation of two-ply cellulose fiber-
reinforced cement composite sheet.

Fig. 3—Typical appearance of cellulose fiber cement.

Fig. 2—Time-pressure cycles in autoclave.

time with a final thickness of 7 mm (0.3 in.). The prepara-

tion was completed within 1 hour of starting the mixture.
After the completion of processing, the screen was carefully
removed from the sheets, which were then stacked flat in a
precuring oven or carbonation chamber. After precuring, the
sheets were autoclaved. Fig. 4—Photograph of flexural test setup for cellulose fiber
The autoclave time-pressure cycle pattern used in this cement composite.
study is shown in Fig. 2. The typical appearance of the
resulting cellulose fiber cement boards is shown in Fig. 3.

Specimens and test procedures

Flexural tests were performed according to ASTM C1185-
03.20 The flexural test samples have a clear span of 254 mm
(10 in.), a width of 152.4 mm (6 in.), and a thickness of 10 mm
(0.4 in.) for unpressed boards and 7 mm (0.3 in.) for pressed
boards. Flexural specimens were cut from each sample
using a diamond saw. A displacement rate of 2.8 mm/
min (0.11 in./min) was used in the flexure tests (which were
conducted in a displacement-controlled mode). A computer-
controlled data acquisition system was used to record the
test data. The load-deflection curves were characterized by
flexural strength, toughness (total area underneath the load-
deflection curve), and initial stiffness (defined herein as the
stiffness obtained through linear regression analysis of the
load-deflection points for loads below 15% of maximum
load). Figure 4 shows a photograph of the three-point flexural
test setup used for cellulose fiber-reinforced cement compos-
ites. The flexural performance was evaluated in a wet condi-
tion. The specimens were immersed in water at a temperature
of 23 ± 4°C (73 ± 7°F) for a minimum period of 48 hours.
The averages of three specimens were tested for flexural
performance. The specimens were tested immediately upon
removal from the water. All tests were conducted in an equilib-
rium condition at a room temperature of 22 ± 2°C (72 ± 3°F) Fig. 5—Flexural test results (pressed); C1 is Control 1,
and 50 ± 10% RH. C2 is Control 2.

ACI Materials Journal/May-June 2013 307

 Table 5—Analysis of variance of flexural test
results for pressed boards
Flexural strength Flexural toughness Initial stiffness
F-ratio P-value F-ratio P-value F-ratio P-value
A 14.1 0.006* 3.81 0.087 0.188 0.676
B 4.54 0.066 3.11 0.116 1.39 0.272
* *
A×B 15.34 0.004 6.703 0.032 1.575 0.245
*
Statistically significant difference at 95% level of confidence.
Notes: A is oven temperature; B is autoclave duration.

Table 6—Analysis of variance of flexural test

results for unpressed boards
Flexural strength Flexural toughness Initial stiffness
F-ratio P-value F-ratio P-value F-ratio P-value
A 4.57 0.039* 32.05 0.000* 13.55 0.001*
B 56.96 0.000* 72.96 0.000* 49.48 0.000*
C 1.69 0.203 17.03 0.000* 12.98 0.001*
D 1.65 0.206 6.99 0.012* 8.23 0.007*
A×B 0.06 0.795 0.47 0.495 11.21 0.002*
A×C 1.86 0.18 7.98 0.008* 2.74 0.106
A×D 0.05 0.813 3.27 0.079 13.14 0.001*
B×C 26.32 0.000* 19.28 0.000* 4.77 0.035*
* *
B×D 13.22 0.001 33.85 0.000 65.72 0.000*
C×D 12.09 0.001* 33.89 0.000* 26.72 0.000*
*
Statistically significant difference at 95% level of confidence.
Notes: A is oven temperature; B is oven duration; C is autoclave duration; D is CO2
Fig. 6—Flexural test results (unpressed); C1 is Control 1, chamber duration.
C2 is Control 2.
EXPERIMENTAL RESULTS AND DISCUSSION
The flexural test results are presented in Table A1 and
Fig. 5 (pressed) and Table A2 and Fig. 6 (unpressed). The
methodology adopted for the analysis of data is illustrated in
References 21 and 22. The results of analysis of variance for
the pressed and unpressed cases are shown in Tables 5 and 6,
respectively. The trends caused by the main effects and inter-
actions based on statistical analysis are shown in Fig. 7 to 9.
Based on the statistical analyses discussed previously,
all variables (oven temperature, oven duration, autoclave
duration, and CO2 chamber duration) in the unpressed case
are statistically significant at the 95% level of confidence,
and these effects should be considered in the optimization
process in the next phase of this investigation. However,
to reduce the size of the optimization test program, it was
decided to keep the oven temperature constant. To select
this temperature, the authors experimentally conducted
a series of tests at four different oven temperatures: 20,
35, 50, and 100°C (68, 95, 122, and 212°F). The first two
represent the normal curing condition, while the last two
represent the elevated temperature curing condition. A
big gap from 50 to 100°C (122 to 212°F) was intended to
raise the temperature within the matrix to the critical limit
of 60 to 70°C (140 to 158°F). Serious deleterious effects,
such as microcracks or delayed ettringite formulation, were
expected to occur.23 The fixed variables used were an oven
duration of 1 hour, a CO2 chamber duration of 1 hour,
and an autoclave duration of 8 hours, which yielded the
highest flexural strength. Among the four temperatures,
50°C (122°F) yielded a better balance of properties on
Fig. 7—Trends in flexural strength. (Note: °C = [°F – 32]/1.8.) flexural performance.

308 ACI Materials Journal/May-June 2013


Fig. 8—Trends in flexural toughness. (Note: °C = [°F – 32]/1.8.)
Because a limited number of variables were investi-
gated in the pressed case, based on the results produced,
the selected processing condition comprises 1 hour of oven
drying at 50°C (122°F), 1 hour of CO2 curing, and 4 hours
of autoclave curing.
Optimization of processing conditions
Three influential variables identified in the previous phase
of the study (oven duration, CO2 chamber duration, and
autoclave duration) were selected to be optimized in the
initial phase based on performance and processing consider-
ation. The optimization experimental design was formulated
based on the statistical theory of response surface analysis.
The objective of the optimization process was to maximize
flexural performance characteristics (strength, toughness,
and stiffness) and minimize processing duration.
Once the optimized manufacturing variables were identi-
fied, the mechanical and physical performances of CO2-cured
composites were compared with those of conventional
composites made without CO2 curing. Table A3 shows the
experimental program for optimization through response
surface analysis. Various combinations of the three statisti-
cally influential variables are considered in this experimental
program. The background (fixed) variable was an oven
temperature of 50°C (122°F).
The response surface analysis shows the effects of vari-
ables on the results. Increasing flexural strength and tough-
ness values were obtained by increasing oven, CO2 chamber,
and autoclave duration. In stiffness, decreasing the oven
ACI Materials Journal/May-June 2013
Fig. 9—Trends in initial stiffness (unpressed). (Note: °C =
[°F – 32]/1.8.)
Fig. 10—Optimum manufacturing conditions of unpressed
cellulose fiber-reinforced cement composites (based on
6 hours of autoclave duration).
time increased the stiffness values, while increasing
CO2 chamber and autoclave duration produced better stiff-
ness performance.
Optimization plots were then generated (refer to Fig. 10)
for achieving maximum flexural strength, toughness, and
stiffness of the lowest possible manufacturing duration.
In Fig. 10, the nonshaded region corresponds to optimum
conditions for the manufacturing of unpressed cellulose
fiber-reinforced cement composites using CO2 curing. The
optimum levels of the variables derived from the aforemen-
tioned process are as follows: an oven temperature of 50°C
(122°F), an oven duration of 1 hour; a CO2 chamber duration
of 1 hour, and an autoclave duration of 6 hours.
The flexural performance of the optimized unpressed
composite produced through CO2 curing is compared in
Fig. 11 to the two control boards manufactured without
309
 Table 7—Percentage difference in flexural
performance of CO2-cured boards versus
controls (unpressed)
Control 1 Control 2
Flexural strength, % +51 +35.1
Flexural toughness, % +2.7 –20.7
Initial stiffness, % +18.3 +29.4

Table 8—Percentage difference of flexural

performance of CO2-cured boards versus
controls (pressed)
Control 1 Control 2
Flexural strength, % +19.29 +10.15
Flexural toughness, % –32.29 –17.53
Initial stiffness, % +23.39 +5.28

Table 9—Water absorption and specific gravity

Fig. 11—Flexural performance of optimized CO2-cured
unpressed cellulose fiber-reinforced cement composite
(equilibrium condition) versus those of controls: Control 1
(1-1-6); Control 2 (1-1-8); and CO2 (1-1-6). (Note: 1 mm =
0.03937 in.)
Fig. 12—Flexural performance of pressed cellulose fiber-
reinforced cement composites (equilibrium condition):
Control 1 (1-1-4); Control 2 (1-1-8); and CO 2 (1-1-
4). (Note: 1 mm = 0.03937 in.)
test results
Unpressed Pressed
Control Control
1 2 CO2 1 2 CO2
Specific gravity 1.11 1.18 1.11 1.18 1.22 1.26
Water absorption, % 51.42 48.39 45.04 32.8 33.06 31.55
CO2 curing. The strength of the CO2-cured composite was
higher than that of the control boards at a 95% level of confi-
dence. Table 7 shows the percentage difference in the flex-
ural properties of the CO2-cured composite versus those of
the control boards.
CO2 curing seems to have yielded better matrix and board
qualities. The unpressed board seems to have some negative
effects on initial toughness. After aging, however, toughness
may also be improved with CO2 curing.
The pressed boards considered were CO2-cured with 1 hour
of oven drying at 50°C (122°F), 1 hour in the CO2 chamber,
and 4 hours in autoclave. Two control conditions were also
considered with 1 hour of oven drying at 50°C (122°F), 1 hour
in the CO2 chamber but with 0% CO2, and 4 and 8 hours in
autoclave. The flexural test results are presented in Fig. 12.
The effects of CO2 curing were significant at a 95%
level of confidence. Compressed boards associated with
CO2 curing seem to have some negative effects on toughness
characteristics. Table 8 shows the percentage differences in
the flexural performance of the CO2-cured versus the control
pressed cellulose fiber-reinforced cement composites.
Any improvement in composite toughness will depend on
whether fibers bridging the cracks are able to support the
load previously carried by the matrix and whether the fibers
break or pull out of the matrix. Hannant24 mentioned that
improving the bond between the fiber and the matrix (as
a result of CO2 curing, particularly in the pressed case in
this study) leads to an improvement in the contact area and
frictional force at the interface. The strain in the composite
at a given stress depends on the length of debonded fibers
and, hence, a greater bond leads to smaller failure strain and
fibers are broken rather than pulled out. This behavior
probably interprets the reduction in flexural toughness
associated with CO2 curing.

310 ACI Materials Journal/May-June 2013

 Fig. 13—Water absorption of cellulose fiber-reinforced cement composite:
(a) unpressed; and (b) pressed.

It is interesting to notice that, 2 hours after mixing, the

conventionally processed boards were still plastic, while
the CO2-cured ones could be hardened without breakage.
Water absorption and specific gravity are indirectly
related to density, in that both are dependent on the void
volume of the sample. The water absorption capacity
results are presented in Table 9 and Fig. 13 for CO2-cured
and control boards. The CO2 curing on unpressed and
pressed composites shows reduced water absorption
when compared with the control (non-CO2) curing
composite. The denser structure could be responsible for
this phenomenon.
Wood fiber consists of long chains of glucose molecules
attracted to each other by hydrogen bonds. They contain
voids in excess of 50% of their total volume. When these
voids are filled with moisture, the dimensional stability
of the material is affected. The diameter change of cellu-
lose is large and can affect the bonding of the fiber to the
matrix.9,25-28 Dimensional stability is measured in terms
of dimensional movements expressed as the percentage
changes in length, thickness, and mass as RH is increased
from 30 to 90%, following the BS 5669-428 test proce-
dures. The results are expressed as a percentage increase
of the original value. Figure 14 shows the results of tests
on the dimensional stability of CO2-cured and conven-
tional composites. CO2 curing is observed to yield major
improvements in the dimensional stability of boards.
Fig. 14—Dimensional stability test results: (a) unpressed;
SUMMARY AND CONCLUSIONS and (b) pressed.
An experimental study was conducted to assess the effects
of CO2 curing on the physical and mechanical characteris- 2. In the case of unpressed boards, oven duration had
tics of cellulose fiber-reinforced cement composites. Two statistically significant effects at a 95% level of confi-
categories of cellulose fiber cement boards were evaluated: dence on flexural strength. In the case of toughness, oven
pressed and unpressed. The processing parameters investi- temperature, oven duration, interaction of oven duration
gated were oven temperature and autoclave duration (both with CO2 chamber duration, and interaction of CO2 chamber
with constant concentration of CO2 curing) for pressed duration with autoclave duration were statistically significant
boards and oven temperature, oven duration, CO2 chamber at the 95% level of confidence. In the case of stiffness, inter-
duration, and autoclave duration for unpressed boards. The action of oven duration, oven duration with CO2 chamber
test results indicated that: duration, and interaction of CO2 chamber duration with
1. In pressed boards, all variables (oven temperature and autoclave duration were statistically significant at a 95%
autoclave duration) had statistically significant effects on the level of confidence.
end product at a 95% level of confidence on flexural perfor- 3. In both pressed and unpressed boards, an oven tempera-
mance. CO2 curing in some conditions yielded better results ture of 50°C (122°F) was chosen as the preferred temperature.
when compared with conventional curing, even at half the Response surface analysis of results yielded the preferred
autoclave duration. processing conditions of cellulose fiber cement boards.

ACI Materials Journal/May-June 2013 311

 REFERENCES
1. Kurpiel, F. T., “Rapid Growth of Cement-Cellulose Fiberboard (CFB),”
Inorganic-Bonded Wood and Fiber Composite Materials, A. A. Moslemi,
ed., V. 6, University of Idaho, Moscow, ID, 1998, pp. 55-60.
2. Soroushian, P.; Won, J.-P.; and Hassan, M., “Durability Characteristics
of CO2-Cured Cellulose Fiber Reinforced Cement Composites,” Construc-
tion & Building Materials, V. 34, 2012, pp. 44-53.
3. Soroushian, P.; Elzafraney, M.; Nossoni, A.; and Chowdhury, H.,
“Evaluation of Normal-Weight and Light-Weight Fillers in Extruded Cellu-
lose Fiber Cement Products,” Cement and Concrete Composites, V. 28,
2006, pp. 69-76.
4. Soroushian, P.; Marikunte, S.; and Won, J., “Wood Fiber Reinforced
Cement Composites under Wetting-Drying and Freezing-Thawing Cycles,”
Journal of Materials in Civil Engineering, ASCE, V. 9, No. 4, 1994,
pp. 595-611.
5. Kuroki, Y.; Nagadomi, W.; Kaneko, S.; Tachi, M.; Kawai, S.; and
Sasaki, H., “Transition of Markets for Cement Exterior Sidings and Devel-
opment of New Technology in Japan,” Inorganic Bonded Wood and Fiber
Composite Materials, A. A. Moslemi, ed., V. 6, University of Idaho,
Moscow, ID, 1998, pp. 61-98.
6. Habighorst, C., “New Cement Board Technologies,” Proceedings of
the International Particleboard/Composite Materials Symposium, Pullman,
WA, 1998, pp. 129-134.
7. Reed, J., Introduction to the Principles of Ceramic Processing, John
Wiley & Sons, Inc., New York, 1988, 371 pp.
8. Claramunt, J.; Ardanuy, M.; Garca-Hortal, J. A.; and Filho, R. D. T.,
“The Hornification of Vegetable Fibers to Improve the Durability of Cement
Mortar Composites,” Cement and Concrete Composites, V. 33, 2011,
pp. 585-595.
9. Ardanuy, M.; Claramunt, J.; García-Hortal, J. A.; and Barra, M.,
“Fiber-Matrix Interactions in Cement Mortar Composites Reinforced with
Cellulosic Fibers,” Cellulose, V. 18, 2011, pp. 281-289.
10. Tolêdo Filho, R. D.; Ghavami, K.; England, G. L.; and Scrivener, K.,
“Development of Vegetable Fibre-Mortar Composites of Improved Dura-
bility,” Cement and Concrete Composites, V. 25, 2003, pp. 185-196.
11. Hsu, W., “Blast Furnace Slag Flake-Board,” Inorganic Bonded
Wood and Fiber Composite Materials, A. A. Moslemi, ed., Forest Products
Society, Madison, WI, 1993, pp. 48-50.
12. Geimer, R. L.; Souza, M. R.; Moslemi, A. A.; and Simatupang, M. H.,
“Carbon Dioxide Application for Rapid Production of Cement Particle-
board,” Proceedings of the 3rd International Inorganic Bonded Wood and
Fiber Composite Materials Conference, A. A. Moslemi, ed., Forest Prod-
ucts Society, Madison, WI, 1993, pp. 31-41.
13. Simatupang, M. H.; Habighorst, C.; Lange, H.; and Neubauer, A.,
“Investigation on the Influence of the Addition of Carbon Dioxide on the
Production and Properties of Rapidly Set Wood-Cement Composites,”
Cement and Concrete Composites, V. 17, 1995, pp. 187-197.
312
14. Qi, H.; Cooper, P. A.; and Wan, H., “Effect of Carbon Dioxide Injec-
tion on Production of Wood Cement Composites from Waste Medium
Density Fiberboard (MDF),” Waste Management, V. 26, No. 5, 2006,
pp. 509-515.
15. De Silva, P.; Bucea, L.; Moorehead, D. R.; and Sirivivatnanon, V.,
“Carbonate Binders: Reaction Kinetics, Strength and Microstructure,”
Cement and Concrete Composites, V. 28, 2006, pp. 613-620.
16. Lange, H.; Simatupang, M. H.; and Neunauer, A., “Influence of
Latent Hydraulic Binders on the Properties of Wood-Cement Composite,”
Inorganic Bonded Wood and Fiber Composite Materials, A. A. Moslemi,
ed., University of Idaho, Moscow, ID, V. 1, 1989, pp. 48-52.
17. Soroushian, P.; Won, J.-P.; Chowdhury, H.; and Nossoni, A., “Devel-
opment of Accelerated Processing Techniques for Cement-Bonded Wood
Particleboard,” Cement and Concrete Composites, V. 25, 2003, pp. 721-727.
18. Andrade Silva, F.; Tolêdo Filho, R. D.; Melo Filho, J. A.; and Rego
Fairbairn, E. M., “Physical and Mechanical Properties of Durable Sisal
Fiber-Cement Composites,” Construction & Building Materials, V. 24,
2010, pp. 777-785.
19. Al-Khras, N. M., and Abu-Alfoul, B. A., “Effect of Wheat Straw Ash
on Mechanical Properties of Autoclaved Mortar,” Cement and Concrete
Research, V. 32, 2002, pp. 859-863.
20. ASTM C1185-03, “Standard Test Methods for Sampling and Testing
Non-Asbestos Fiber-Cement Flat Sheet, Roofing and Siding Shingles, and
Clapboards,” ASTM International, West Conshohocken, PA, 2003, 9 pp.
21. Basheer, P. A. M.; Montgomery, F. R.; and Long, A. E., “Factorial
Experimental Design for Concrete Durability Research,” Proceedings of the
Institution of Civil Engineers: Structures and Buildings, V. 104, London,
UK, Nov. 1994, pp. 449-462.
22. Montgomery, D., Design and Analysis of Experiments, seventh
edition, John Wiley & Sons, Inc., New York, 2009, 680 pp.
23. Taylor, H. F. W., Cement Chemistry, second edition, Thomas Telford,
London, UK, 1997, 459 pp.
24. Hannant, D. J., Fibre Cements and Fibre Concretes, John Wiley &
Sons, Inc., New York, 1978, 219 pp.
25. Claramunt, J.; Ardanuy, M.; and García-Hortal, J. A., “Effect of
Drying and Rewetting Cycles on the Structure and Physicochemical Char-
acteristics of Softwood Fibers for Reinforcement of Cementitious Compos-
ites,” Carbohydrate Polymers, V. 79, 2010, pp. 200-205.
26. Page, D. H., and Tydeman, P. A., “Transverse Swelling and Shrinkage
of Softwood Tracheids,” Nature, V. 199, No. 4892, 1963, pp. 471-472.
27. Weise, U., and Paulapuro, H., “Effect of Drying and Rewetting
Cycles on Fibre Swelling,” Journal of Pulp and Paper Science, V. 25, No. 5,
1999, pp. 163-166.
28. BS 5669-4, “British Standard: Particleboard,” British Standards Insti-
tution, London, UK, 1989, 14 pp.
ACI Materials Journal/May-June 2013
 APPENDIX A

Table A1—Flexural performance of cellulose fiber-reinforced cement composites (pressed)

Duration of CO2 chamber: 1 hour (20% CO2 concentration)
Oven temperature (1 hour)
50°C (122°F) 100°C (212°F)
Autoclave hours Autoclave hours Control
4 8 4 8
Run order 1 2 3 4 1* 2†
Flexural strength, MPa (mean
10.3 (0.67) 10.9 (0.54) 10.3 (0.77) 8.3 (0.23) 9.8 (0.84) 7.9 (0.29)
[standard deviation])
Flexural toughness, N-mm (mean
2305.1 (29.92) 2449.5 (366.87) 2417.0 (482.15) 1653.5 (33.14) 2974.7 (232.54) 2110.1 (189.55)
[standard deviation])
Initial stiffness, N/mm (mean
101.8 (8.14) 134.0 (26.65) 112.7 (14.79) 111.7 (32.96) 125.1 (8.44) 115.3 (1.52)
[standard deviation])
*
Control 1: 1 hour in oven at 50°C (122°F), then 1 hour in steam box at 50°C (122°F) and 95% RH and, finally, 8 hours in autoclave.
†
Control 2: 24 hours inside plastic bag and then 8 hours in autoclave.
Notes: 1 lb/in.2 = 6.89 × 10–3 N/mm2; 1 lb-in. = 113 N-mm; 1 lb/in. = 1.7512 × 10–5 N/mm.

Table A2—Flexural performance of cellulose fiber-reinforced cement composite (unpressed)

Strength, MPa (mean Toughness, N-mm (mean Stiffness, N/mm (mean
Run order A B C D [standard deviation]) [standard deviation]) [standard deviation])
1 – – – – 5.513 (0.366) 2653.687 (336.959) 139.856 (2.086)
2 + – – – 5.423 (0.5) 1976.639 (46.850) 146.976 (35.688)
3 – + – – 5.330 (0.394) 1872.826 (82.812) 80.535 (1.024)
4 + + – – 4.289 (1.189) 906.617 (256.946) 38.900 (7.125)
5 – – + – 7.056 (0.022) 2461.818 (81.413) 113.554 (2.311)
6 + – + – 6.728 (0.842) 2700.567 (85.459) 162.983 (3.791)
7 – + + – 2.525 (0.033) 827.480 (11.847) 54.831 (4.200)
8 + + + – 2.226 (0.216) 781.881 (172.512) 38.413 (17.051)
9 – – – + 5.402 (0.264) 2154.723 (175.897) 105.873 (2.063)
10 + – – + 3.131 (0.049) 807.713 (74.521) 43.101 (3.057)
11 – + – + 3.726 (0.230) 1727.695 (185.601) 115.188 (5.077)
12 + + – + 3.886 (0.346) 1523.974 (111.879) 88.802 (3.655)
13 – – + + 5.340 0.731) 2860.914 (354.019) 147.448 (6.367)
14 + – + + 6.292 (0.198) 2775.181 (176.401) 145.769 (16.953)
15 – + + + 5.320 (±0.863) 2748.388 (497.734) 170.762 (35.041)
16 + + + + 3.945 (0.490) 1573.622 (220.180) 92.284 (20.947)
Control 1 3.641 (0.692) 1706.078 (422.787) 103.577 (5.963)
Control 2 5.039 (0.707) 2101.850 (211.945) 150.930 (8.316)
Notes: A is oven temperature; B is oven duration; C is autoclave duration; D is CO2 chamber duration; 1 lb/in.2 = 6.89 × 10–3 N/mm2; 1 lb-in. = 113 N-mm; 1 lb/in. = 1.7512 ×
10–5 N/mm.

ACI Materials Journal/May-June 2013 313

 APPENDIX A, cont.

Table A3—Optimization experimental design and test results

Strength, MPa (mean Toughness, N-mm (mean Stiffness, N/mm (mean
Run order A* B* C* [standard deviation]) [standard deviation]) [standard deviation])
1 0.5 4 4 4.0 (0.2) 1609.0 (177.1) 117.0 (4.3)
2 1 2.5 6 5.5 (0.07) 2080.9 (307.5) 117.0 (42.41)
3 1.5 1 4 3.9 (0.3) 1935.3 (83.2) 77.3 (7.8)
4 1 2.5 6 5.1 (0.06) 1642.1 (147.9) 140.3 (48.1)
5 0.5 1 4 5.1 (0.05) 2037.3 (381.1) 154.0 (58.8)
6 0.5 1 8 5.07 (0.9) 2069.3 (132.6) 143.6 (9.4)
7 1.5 1 8 6.7 (0.003) 2595.9 (309.8) 141.0 (17.1)
8 1 2.5 6 5.4 (0.15) 2039.8 (313.6) 137.6 (11.3)
9 1.5 4 4 4.6 (0.07) 1737.3 (90.0) 107.8 (3.1)
10 1.5 4 8 4.8 (0.13) 2153.3 (92.6) 161.1 (6.2)
11 1 2.5 6 5.2 (0.25) 1778.2 (140.5) 129.8 (50.7)
12 0.5 4 8 4.8 (0.10) 1631.3 (134.0) 115.7 (4.5)
Notes: A is oven duration; B is CO2 chamber duration; C is autoclave duration; all units for A, B, and C are in hours; 1 lb/in.2 = 6.89 × 10–3 N/mm2; 1 lb-in. = 113 N-mm; 1 lb/in. =
1.7512 × 10–5 N/mm.

314 ACI Materials Journal/May-June 2013

 ACI MATERIALS JOURNAL
Title no. 110-M27
Effect of Curing Water Availability and Composition on
Cement Hydration
by Md Sarwar Siddiqui, Wesley Nyberg, Wilson Smith, Brett Blackwell, and Kyle A. Riding
Curing can help concrete reach its full strength and durability
potential. The effect of sealing the concrete with plastic or form-
work, use of a liquid curing compound, wet curing, and internal
curing with saturated lightweight aggregates on the cement degree
of hydration (DOH) development with time was examined using
isothermal calorimetry. Curing water amount, curing water ionic
concentration, and sample thickness were varied. Finally, curing
application timing was studied by comparing strength development
of concrete cylinders sealed, placed in a moist room after 24 hours
sealed, and immersed in a water bath immediately after finishing.
Increasing the height of curing water decreased the height of heat
of hydration rate peaks. Curing water ionic concentration affected
the setting time and heat of hydration rate peak heights. Strength
results show delayed curing can result in significant strength loss
because of the difficulty for water to penetrate the already-hard-
ened concrete.
Keywords: curing; early hydration; internal curing; ionic concentration;
pore solution.
Introduction
Concrete curing can improve strength, durability, dimen-
sional stability, freezing-and-thawing resistance, resistance
to surface wear, and reduces permeability and ion penetra-
tion. For Type I cement, curing can increase the strength
by more than 58% after 6 months (Price 1951). To achieve
complete cement hydration, a water-cementitious material
ratio (w/cm) greater than 0.42 is theoretically needed (Powers
and Brownyard 1948; Jensen et al. 1999). In practice,
however, complete hydration is rare. Studies have shown
that once the relative humidity within the capillary drops
below 80%, hydration virtually stops (Powers 1947). To
keep the concrete pore relative humidity high and continue
hydration, an additional supply of water is needed.
Curing is even more critical for concrete with low w/cm
for significant improvement of cement hydration. Heteroge-
neous microstructures develop in sealed low w/cm cement
pastes, with zones of high and low density not normally seen
in samples that are water-cured (Jaouadi 2008). Zones of
low density can create zones of weakness to lower strength.
Lower w/cm also results in a less continuous pore network
and a more rapid decrease of pore relative humidity under
sealed conditions. The use of low w/cm also contributes
to the development of autogenous shrinkage of concrete
(Jensen and Hansen 1996). Autogenous shrinkage occurs
because of a vapor-water interface that develops when
water is used up. This forms a surface tension and capil-
lary under-pressure that creates tensile stresses on pore
walls (Grasley 2006).
Conventional curing techniques, such as water ponding,
water spraying, fogging, plastic film, wet burlap, and liquid
membranes improve curing by reducing moisture loss and
providing an additional supply of water to the near-surface
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
layers. This method generally depends on one-dimensional
movement of water and has a very limited depth of influ-
ence. Internal curing has been developed to help water reach
cement grains on the concrete interior where ponded water
has not been able to reach (Bentz et al. 2005; Klieger 1957;
Mather 2001; Weber and Reinhardt 1997). Internal curing
systems store water inside the concrete in highly absorbent
material, such as saturated fine lightweight aggregate as
internal reservoirs (FLAIR) (Weber and Reinhardt 1997;
Hoff and Elimov 1995; Kovler and Jensen 2005; Kovler
et al. 2004; Villarreal 2005) or super-absorbent polymers
(Kovler and Jensen 2005; Jensen and Hansen 2001; Jensen
and Hansen 2002). Water is kept inside the internal reser-
voirs during mixing, which later slowly releases into the
cement matrix. The concrete diffusivity decreases as the
hydration increases. The curing water influence zone corre-
spondingly decreases with time from 20 mm (0.79 in.)
at 1 day to 0.25 mm (0.01 in.) at 28 days (Bentz et al. 2007).
The effectiveness of concrete curing will be dependent on
when the curing is applied. It was found with thin (2 to 5 mm
[0.079 to 0.197 in.]) cement paste samples that the degree
of hydration (DOH) at 92 days was similar for samples
with water curing started at 4 hours and 7 days (Bentz and
Stutzman 2006). Recently, some contractors and material
suppliers have advocated for beginning curing of concrete
cylinders immediately after finishing for high-performance
concrete. The influence of the timing of curing water appli-
cation on strength is important because after setting it might
be more difficult for water to penetrate the concrete in larger
samples and aid in curing. Supplying water to the concrete
during the initial curing period could increase the DOH
throughout the concrete cylinder and increase strength in the
bulk. The water, however, could also increase the effective
w/cm at the top surface. A weak plane at the top could cause
lower compressive strength tests.
Isothermal calorimetry was recently used by Lura et al.
(2010) to study the difference in hydration development as
measured using chemical shrinkage and isothermal calorim-
etry with and without water curing. It was found that the
water curing altered the aluminate reactions during the first
day and increased the overall hydration after 24 hours. Only
one specimen height and water amount was used, as the
purpose was to compare the effects of the curing used in the
two measurement methods.
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-315.R1 received March 5, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
315
ACI member Md Sarwar Siddiqui is a PhD Candidate at the University of Texas at curing water amount on the cement DOH development was
Austin, Austin, TX. He received his MS from Kansas State University, Manhattan, KS, studied. Distilled water, lime-saturated water, and cement
in 2010. He is a member of ACI Committee 221, Aggregates. His research interests pore water (CPW) was also used to investigate the effects of
include early-age properties of concrete, influence of aggregate on concrete properties,
sustainable construction materials, and pavement concrete. ionic concentrations of curing water on hydration. Finally,
the strength development was measured on cylinders made
Wesley Nyberg received his BS and MS from Kansas State University in 2010 and from a 0.32 w/cm concrete mixture under different curing
2012, respectively.
methods. The cylinders were cured by immersion in water
Wilson Smith is a Staff Engineer with Kleinfelder. He received his BS and MS from immediately after finishing, sealed for 24 hours followed by
Kansas State University in 2009 and 2012, respectively. curing in a 100% relative humidity room, or sealed.
Brett Blackwell is an Assistant Structural Engineer at Burns & McDonnell. He
received his BS from Kansas State University in 2010.
Research Significance
Cement hydration test methods require different types of
ACI member Kyle A. Riding is an Assistant Professor in the Department of curing, giving potentially different results. This study examined
Civil Engineering at Kansas State University. He received his MSE and PhD from the effects of curing methods commonly used in practice and in
the University of Texas at Austin in 2004 and 2007, respectively. He is Secretary of
ACI Committee 231, Properties of Concrete at Early Ages, and a member of ACI
hydration experiments using the same hydration measurement
Committee 236, Material Science of Concrete. He received the Wason Medal for technique: isothermal calorimetry. Additionally, changes in
Materials Research in 2011. concrete cylinder curing procedures for high-strength concrete
have been recently considered to allow for access to water.
This study examined the effects of curing application time on
Table 1—Physical and chemical properties of
low w/cm concrete strength development.
cementitious materials: percentages given are
by mass
Materials
Properties Type I cement Slag Silica fume An ASTM C150/C150M (2002) Type I cement,
Specific gravity — 2.5 — ASTM C1240 (2011) silica fume, and ASTM C989/C989M
SiO2, % 21.34 33.83 90.4
(2005) slag cement were used in this study. The physical
and chemical properties of cementitious materials that were
Al2O3, % 4.74 11.45 0.14 used in this study are presented in Table 1. A mid-range
Fe2O3, % 3.29 0.49 0.87 water-reducing admixture (WRA) used at 87 mL/100 kg
Si + Al + Fe, % 29.37 45.77 91.41 (12 oz/cwt) was used in mortar mixtures for both siliceous
and lightweight fine aggregate to improve workability. An
CaO, % 62.94 38.52 0.64
ASTM C33/C33M (2011) siliceous fine aggregate and a
MgO, % 1.69 11.63 2.53 lightweight fine aggregate were used in the mortar mixtures.
SO3, % 2.68 2.55 0 The bulk specific gravity (SG) and percent absorption of
siliceous fine aggregate used in the calorimetry experiments
Na2O, % 0.14 0.28 0.15
were 2.5 and 0.42%, respectively. Lightweight fine aggre-
K2O, % 0.53 0.38 0.82 gates were sieved and recombined in the laboratory to have
Total alkalis as Na2O 0.49 0.53 0.69 the same gradation as the siliceous fine aggregate. The light-
2
Blaine fineness, m /kg 360.1 — —
weight fine aggregate bulk SG and absorption after 48 hours
of soaking were 1.49 and 16.87%, respectively. The bulk
C3S, % 66.96 — — SG and absorption capacity of the coarse and fine aggregate
C2S, % 16.49 — — used for the concrete mixture were 2.58, 2.38% and 2.58,
C3A, % 2.92 — — 0.54%, respectively.
The w/cm of cement pastes was varied from 0.275 to
C4AF, % 9.29 — —
0.35 to study the interplay between curing method and w/cm.
Gypsum, % 2.31 — — The mortar fine aggregate-cementitious material ratio was
Hemihydrate, % 1.68 — — 2.75 by weight for siliceous aggregate. Mortars with a w/cm
of 0.45 were used to study the effects of FLAIR on cement
Arcanite, % 0.23 — —
hydration. Ten percent, 20%, 30%, 40%, 50%, and 100%
of siliceous aggregate was replaced by volume with light-
Table 2—Concrete mixture proportions weight aggregate presoaked for 48 hours to determine the
effectiveness of concrete internal curing with FLAIR. The
Coarse Fine
Water, Cement, aggregate, aggregate, WRA, concrete mixture was made using the mixture proportions
kg/m3 kg/m3 kg/m3 kg/m3 mL/100 kg shown in Table 2.
Content (lb/yd3) (lb/yd3) 3 3
(lb/yd ) (lb/yd ) (oz/cwt) A water-based liquid curing compound was used in this
181 594 1214 374 36.3 study. Distilled water, saturated lime water, and CPW were
(305) (1001) (2046) (631) (5) used to investigate the effects of water curing and ionic
concentration of curing water on hydration. Fresh CPW was
This study was conducted to evaluate the ability of water collected by extracting water from cement paste made with
curing, curing compounds, and FLAIR to increase the cement the same cement used in this study and with a w/cm of 1.5.
DOH. FLAIR is fine lightweight aggregate with internal The schematic diagram of the vacuum filtration setup used
porosity, which can store water for later release when the to extract the pore solution from the cement paste is shown
internal relative humidity of concrete drops. The effect of in Fig. 1.

316 ACI Materials Journal/May-June 2013

Fig. 1—Schematic diagram of cement paste pore water vacuum
filtration system.
Experimental Methods
In this study, the ASTM C305 (2011) mixing procedure was
followed for both cement paste and mortar mixing. Both the
cement paste and mortar were mixed at room temperature and
placed in the isothermal calorimeter within 15 minutes from
the addition of water to the cementitious materials. Concrete
was mixed according to ASTM C192/C192M (2007).
Distilled water, lime-saturated water, CPW, and curing
compound were used to cure the cement paste samples. These
curing substances were applied at the top of the cement paste
or mortar samples. These liquids were gently applied to the
sample surface so that the curing liquid did not disturb the
cement paste and affect the w/cm. The samples were sealed
in their containers after the curing liquid was placed on top
of the sample. Two layers of curing compound were sprayed
on the sample surface to ensure complete cover.
The concrete compressive strength was measured using
a 100 x 200 mm (4 x 8 in.) cylinder following ASTM C39/
C39M (2011). Concrete samples were cured in three different
ways: sealed by keeping the concrete inside the plastic molds
with the plastic lids firmly attached, immersed in water
at 23 ± 2°C (73 ± 3.5°F) immediately after finishing without
a lid, and keeping the concrete inside of the plastic molds
with the plastic lids firmly attached for 24 hours followed
by demolding and storage in a 100% relative humidity room
at 23 ± 2°C (73 ± 3.5°F).
An eight-channel isothermal calorimeter was used in
this study to measure the heat evolution of the cementi-
tious systems studied (Broda et al. 2002; Wadsö 2010). The
cement and mortar samples were mixed, weighed, curing
added when needed, and placed in the isothermal calorim-
eter within 15 minutes after beginning mixing. The DOH
was then calculated from the cumulative heat of hydration
using Eq. (1) through (3) (D’Aloia and Chanvillard 2002;
Kada-Benameur et al. 2000; Van Breugel 1998; Copeland et
al. 1960; De Schutter and Taerwe 1996)
H (t )
α (t ) = (1)
Hu
where a(t) is the DOH at time t; H(t) is the cumulative heat
of hydration from time 0 to time t; and Hu is the total heat
available for reaction. The variable Hu is a function of the
ACI Materials Journal/May-June 2013
Fig. 2—Effect of curing water on rate of heat evolution for
100% Type I cement paste (w/cm = 0.35) at 23°C (73°F):
(a) effect of curing height for 10 mm (0.39 in.) thick sample;
and (b) effect of sample depth for 1 mm (0.03937 in.)
curing water.
cement composition, amount, and type of supplementary
cementitious materials (SCMs) and may be approximated as
follows (Schindler and Folliard 2005)
Hu = H cem ⋅ Pcem + 461 ⋅ Pslag + 1800 ⋅ PFA −CaO ⋅ PFA (2)
where Pslag is the slag to total cementitious content mass
ratio; PFA is the fly ash to total cementitious content mass
ratio; PFA–CaO is the fly ash CaO to total fly ash content
mass ratio; Pcem is the portland cement to total cementitious
content mass ratio; and Hcem is the available heat of hydra-
tion of the cement. The value Hcem can be approximated as
shown in Eq. (3) (Schindler and Folliard 2005)
H cem = 500 ⋅ PC3 S + 260 ⋅ PC2 S + 866 ⋅ PC3 A +
(3)
420 ⋅ PC4 AF + 624 ⋅ PSO3 + 1186 ⋅ PFree Ca + 850 ⋅ PMgO
where Hcem is the total heat of hydration of portland cement
at a = 1.0; and Pi is the ratio of the mass of the i-th compo-
nent to total cement mass. Because the total heat available
for reaction was calculated instead of measured, small errors
in the DOH could result. Because comparisons for experi-
ments with curing were made on the same cementitious
materials, any small errors in the total heat available for
reaction would not affect the overall curing method compar-
isons and conclusions.
317
Fig. 3—Effect of lime-saturated water curing on rate of heat
evolution for 100% Type I cement paste (w/cm = 0.35) at 23°C
(73°F): (a) effect of curing height for 10 mm (0.39 in.)
thick sample; and (b) effect of sample depth for 1 mm
(0.03937 in.) lime-saturated water.
Results
Figure 2 shows the effect of curing water thickness and
sample thickness on the rate of heat evolution of the cement
used in this study. As the curing water thickness increased,
the maximum rate of heat evolution slightly decreased. The
third hydration peak also decreased, confirming the decrease
in aluminate hydration found by Lura et al. (2010). Increasing
the ratio of the sample thickness to curing water thickness
increased the maximum rate of heat evolution. This could be
because the changes in ionic concentrations in the pore solu-
tion are lower for the larger amount of cement paste.
Figure 3 shows the effect of lime-saturated curing water
height and sample thickness on the rate of heat evolution
of cementitious materials. Lime-saturated water accelerated
hydration, with the time of the second peak occurring over
an hour before the sealed sample; however, sample thick-
ness does not show any significant effect on the acceleration
of the samples cured with lime-saturated water. The lime-
saturated water samples show a change in the ratio of the
second and third peak heights, just as the samples cured with
distilled water. Increasing sample height-to-curing water
height ratio decreases this effect, most likely because of the
lower pore water dilution from the decreased curing water-
to-cement paste ratio. This trend is the same one as seen
when curing with distilled water.
318
Fig. 4—Effect of cement paste pore water curing on rate of
heat evolution for 100% Type I cement paste (w/cm = 0.35)
at 23°C (73°F): (a) effect of curing height for 10 mm
(0.39 in.) thick sample; and (b) effect of sample depth for
1 mm (0.3937 in.) cement paste pore water.
Figure 4 shows the effect of extracted CPW thickness as a
curing liquid and sample thickness on cementitious material
heat evolution rate. Cement pore water did not accelerate
the hydration reaction such as was seen with lime-saturated
water. This may be from the presence of sulfate or other
ions in the CPW solution. The effects of CPW and sample
thickness on the maximum rate of heat evolution were not as
significant as distilled water and lime-saturated water. The
ratio of the second and third peak heights did not change
significantly with the change in sample thickness because of
the more similar ionic concentration between the cure liquid
and the pore solution. This shows that changes in sulfate
concentration from curing liquid can affect the hydration of
cementitious materials.
To quantify the effects of water ponding on the DOH
of cementitious materials, calorimetry was performed on
samples with w/cm of 0.275, 0.3, 0.325, and 0.35 at 23°C
(73°F) and w/cm 0.275 and 0.35 at 38°C (100°F). For cement
pastes, sample depths of 10, 20, and 30 mm (0.39, 0.78, and
1.18 in.) were used at 23°C (73°F), whereas sample depths
of 10, 15, and 20 mm (0.39, 0.59, and 0.78 in.) were used
at 38°C (100°F) to determine the effect of curing on the
sample thickness. Three mL (18 in.3) of curing water was
applied to the top of the paste samples to provide a thin
film of approximately 1 mm (0.039 in.) of curing water.
Figure 5 shows the increase in DOH of cement pastes at
ACI Materials Journal/May-June 2013
 Fig. 6—Effect of w/cm on average DOH of water-cured
and sealed samples for 100% Type I cement paste at 23°C
(73°F) after 7 days.

Fig. 5—Increase in percent degree of hydration of water-

cured sample of Type I cement at 23°C (73°F): (a) w/cm
= 0.275; and (b) w/cm = 0.35. (Note: 1 mm = 0.03937 in.)

23°C (73°F) for a 0.275 and 0.35 w/cm. The decrease in the
DOH seen during the first day of hydration with distilled
water is likely because of dilution of the calcium ions in
solution, making it take slightly longer to reach the calcium
supersaturation limit.
Water-cured cement paste samples showed a higher
cumulative heat of hydration than sealed cement paste
samples after 7 days. The average DOH of different sample
thickness for water-cured and sealed samples after 7 days at
23°C (73°F) is presented in Fig. 6. The DOH for the sealed
samples increased almost linearly with the increase in w/cm.
Water-cured samples, however, showed a nonlinear response.
Figure 7 shows the effect of sample depth and w/cm on the Fig. 7—Effect of sample depth and w/cm on increase in
increase of DOH, compared with the sealed sample of degree of hydration than sealed sample of equal thickness
equal thickness after 7 days at the two different tempera- after 7 days for 100% Type I cement paste: (a) 23°C (73°F);
tures tested. For w/cm of 0.325 and 0.35, sample thickness and (b) 38°C (100°F). (Note: 1 mm = 0.03937 in.)
did not have any noticeable effect on the increase in DOH,
indicating that the water diffusion during the first week was increased the amount of available water for hydration, which
high enough for the water to penetrate to the bottom of the increased the cement hydration.
sample. Water-cured samples at a w/cm of 0.275 showed Very little difference was seen in the DOH of mortar
a lower increase in DOH than the 0.3 w/cm sample at all samples with internal curing, as would be expected at w/cm
sample thicknesses, possibly because of more limited space greater than 0.42. Figure 10 shows the effect of presoaked
for hydrated products to grow. lightweight fine aggregate to improve the hydration of
The use of curing compounds in a sealed condition cementitious materials. Even at a w/cm of 0.45, the use
increased the hydration of cement paste, as shown in of FLAIR improved the DOH of mortar much more than
Fig. 8. Water-cured samples and the curing compound external water curing. An increase in the FLAIR provided
samples showed very similar increases in DOH, as shown increased the cement DOH, although with diminishing
in Fig. 9. The water-based curing compound probably returns as shown by the percent DOH increase from the use

ACI Materials Journal/May-June 2013 319

Fig. 8—Effect of curing compound on hydration of cementi-
tious materials for 100% Type I cement w/cm of 0.275 at
23°C (73°F). (Note: 1 mm = 0.03937 in.)

Fig. 9—Comparison between water curing and curing

compound at sealed environment for 100% Type I cement
w/cm of 0.275 at 23°C (73°F).

of FLAIR. This is because once enough FLAIR was used

to fill the space left vacant by chemical shrinkage, adding
additional FLAIR would not provide any additional benefit
(Henkensiefken et al. 2009).
To understand the effect of curing timing on the strength
development of concrete, concrete cylinders were cured
using three different curing conditions. Figure 11 shows the
concrete strength development for the three curing condi-
tions used. The concrete immersed in water immediately
after finishing showed a significant increase in strength
compared to sealed curing and delayed wet curing.
Discussion
For the three types of curing solutions used, the cement
pore solution affected the hydration rate the least, with only
a small decrease in the peak rates observed. This small
decrease is likely because the pore solution in the paste
changes with time, with only a small concentration gradient
between that above the sample and in the paste. The distilled
water affected the peak rates more than that observed with
the cement pore solution, with a further decrease in the third
peak usually associated with the aluminate and sulfate reac-
tion. The lime-saturated water changed the cement pore
solution the most, giving a significant acceleration and
decrease in the ratio between the second and third hydra-
tion peak heights. For hydration studies on cement paste
samples, the curing solution chosen can impact the C-S-H
Fig. 10—Effect of FLAIR on DOH of mortar (w/cm =
0.45) at 23°C (73°F): (a) cumulative heat of hydration; (b)
increase in %DOH with respect to 100% siliceous aggre-
gate; and (c) increase in cumulative heat of hydration after
7 days.
nucleation and growth rate by significantly altering when
calcium supersaturation is reached and the sulfate dissolu-
tion rate and availability.
Both the sealed and water-cured samples generated an
equal cumulative heat of hydration up to a certain period
of time, after which water availability became a limiting
factor. At the time between the sealed and water-cured
sample hydration divergence, a significant amount of water
still remained in the pores, equivalent to a w/cm of 0.09 to
0.13 for the samples tested at 23°C (73°F). The large amount
of water available at the time of divergence shows that the
hydration is probably limited by water movement to the

320 ACI Materials Journal/May-June 2013

unhydrated cement grains and not necessarily by the total
amount of water remaining.
The importance of the available water location on curing
was especially seen in the samples with FLAIR. The 100%
FLAIR dosage accelerated the DOH increase because of the
better dispersion of available water, reducing the limiting
effects of water transport. The DOH increase with FLAIR
dosages between 50 and 100% converged to similar values
at 7 days because the extra water provided was more than
enough to compensate for all of the chemical shrinkage.
The sealed concrete cylinders had 10% higher strength
than the standard moist-room-cured cylinder. The dry
state of the cylinders during testing is most likely respon-
sible for the increased strength, even though the sealed
specimens should have a lower DOH than the moist-room-
cured cylinders. The concrete immersed in water imme-
diately after finishing had 15% higher strength than the
sealed cylinders and 27% higher strength than the standard
moist-room-cured cylinders. This illustrates the difficulty
of getting cure water into the concrete after hardening
because of the low water diffusivity. One of the concerns
about immersing the concrete in water immediately after
finishing was that there would be an effective increase in
the w/cm at the top surface; however, this problem was not
seen in the tests performed. Although this curing proce-
dure did increase the strength, it is doubtful that it would be
representative of in-place concrete strength. Water ponding
cannot normally be applied until after setting or penetrate
very large distances. These results reinforce the idea that
cylinders should be treated according to ASTM C31/C31M
(2010) to avoid misleading high compressive strength
results. This is because this procedure provides a standard
method of judging whether or not the concrete delivered is
what was ordered and not because the specified standard
curing conditions help the concrete achieve the maximum
possible strength.
Conclusions
The effects of different curing techniques were studied
using isothermal calorimetry and compressive strength
experiments. From these experiments, the following conclu-
sions can be drawn:
• Ionic concentration of curing water affects the rate of
hydration. Lime-saturated water accelerated hydration
probably because of the earlier time when the calcium
supersaturation limit was reached. Using a simulated
pore solution as the curing medium would appear to
have the smallest effect on the hydration. This has appli-
cation to hydration kinetics studies such as chemical
shrinkage experiments that require some form of curing
or liquid above the sample.
• FLAIR improved the hydration of cementitious mate-
rials more than water curing at the specimen surface.
This further emphasizes the importance of water avail-
ability during curing. Higher dosages of FLAIR than
just that required for chemical shrinkage may not be
necessary for complete hydration, but will increase the
rate of hydration.
• Curing compounds provided a similar increase in
hydration to water curing in a sealed environment
when compared to sealed specimens without curing
compound or water ponding, probably because of high
water content in water-based curing compounds.
ACI Materials Journal/May-June 2013
Fig. 11—Concrete cylinder strength development under
different curing conditions. (Note: 1 psi = 0.006894 MPa.)
• Wet curing immediately after finishing resulted in 27%
higher 28-day cylinder strength than that of cylinders
sealed for 24 hours followed by curing in a moist room.
Acknowledgments
The support of Kansas State University’s Civil Engineering Department
and Kansas State University’s Transportation Center is acknowledged.
References
ASTM C31/C31M-10, 2010, “Standard Practice for Making and
Curing Concrete Test Specimens in the Field,” ASTM International, West
Conshohocken, PA, 6 pp.
ASTM C33/C33M-11, 2011, “Standard Specification for Concrete
Aggregates,” ASTM International, West Conshohocken, PA, 11 pp.
ASTM C39/C39M-10, 2010, “Standard Test Method for Compressive
Strength of Cylindrical Concrete Specimens,” ASTM International, West
Conshohocken, PA, 7 pp.
ASTM C150/C150M-02, 2002, “Standard Specification for Portland
Cement,” ASTM International, West Conshohocken, PA, 8 pp.
ASTM C192/C192M-07, 2007, “Standard Practice for Making and
Curing Concrete Test Specimens in the Laboratory,” ASTM International,
West Conshohocken, PA, 8 pp.
ASTM C305-11, 2011, “Standard Practice for Mechanical Mixing of
Hydraulic Cement Pastes and Mortars of Plastic Consistency,” ASTM
International, West Conshohocken, PA, 3 pp.
ASTM C989/C989M-05, 2005, “Standard Specification for Ground
Granulated Blast-Furnace Slag for Use in Concrete and Mortars,”
ASTM International, West Conshohocken, PA, 5 pp.
ASTM C1240-11, 2011, “Standard Specification for Silica Fume Used
in Cementitious Mixtures,” ASTM International, West Conshohocken, PA,
7 pp.
Bentz, D. P., and Stutzman, P. E., 2006, “Curing, Hydration, and Micro-
structure of Cement Paste,” ACI Materials Journal, V. 103, No. 5, Sept.-
Oct., pp. 348-356.
Bentz, D. P.; Koenders, E. A. B.; Mönnig, S.; Reinhardt, H. W.; van
Breugel, K.; and Ye., G., 2007, “Materials Science-Based Models,” Internal
Curing of Concrete—State-of-the-Art Report of RILEM Technical Committee
196-ICC, RILEM Publications S.A.R.L., Bagneux, France, pp. 29-43.
Bentz, D. P.; Lura, P.; and Roberto, J. W., 2005, “Mixture Proportioning
for Internal Curing,” Concrete International, V. 27, No. 2, Feb., pp. 35-40.
Broda, M.; Wirquin, E.; and Duthoit, B., 2002, “Conception of an
Isothermal Calorimeter for Concrete-Determination of the Apparent Acti-
vation Energy,” Materials and Structures, V. 35, No. 251, pp. 389-394.
Copeland, L. E.; Kantro, D. L.; and Verbeck, G., 1960, “Chemistry of
Hydration of Portland Cement,” 4th International Symposium of the Chem-
istry of Cement, Washington, DC, pp. 429-465.
D’Aloia, L., and Chanvillard, G., 2002, “Determining the ‘Apparent’
Activation Energy of Concrete: Ea—Numerical Simulations of the Heat
of Hydration of Cement,” Cement and Concrete Research, V. 32, No. 8,
pp. 1277-1289.
De Schutter, G., and Taerwe, L., 1996, “Degree of Hydration-Based
Description of Mechanical Properties of Early-Age Concrete,” Materials
and Structures, V. 29, No. 7, pp. 335-344.
Grasley, Z., 2006, “Measuring and Modeling the Time-Dependent
Response of Cementitious Materials to Internal Stresses,” PhD dissertation,
321
Department of Civil and Environmental Engineering, University of Illinois
at Urbana-Champaign, Urbana, IL, 238 pp.
Henkensiefken, R.; Bentz, D. P.; Nantung, T.; and Weiss, J., 2009,
“Volume Change and Cracking in Internally Cured Mixtures Made with
Saturated Lightweight Aggregate under Sealed and Unsealed Conditions,”
Cement and Concrete Composites, V. 31, No. 7, pp. 427-437.
Hoff, G., and Elimov, R., 1995, “Concrete Production for the Hibernia
Platform,” Supplementary Paper in Second CANMET/ACI Interna-
tional Symposium on Advances in Concrete Technology, Las Vegas, NV,
pp. 717-739.
Jaouadi, I., 2008, “Etude numérique et expérimentale du retrait endogène
de la pâte de ciment au jeune âge,” doctoral thesis, Laboratory of Materials
of Construction, Swiss Federal Institute of Technology at Lausanne, Laus-
anne, Switzerland, 191 pp.
Jensen, O. M., and Hansen, P. F., 1996, “Autogenous Deformation and
Change of the Relative Humidity in Silica Fume-Modified Cement Paste,”
ACI Materials Journal, V. 93, No. 6, Nov.-Dec., pp. 539-543.
Jensen, O. M., and Hansen, P. F., 2001, “Water-Entrained Cement-Based
Materials: Principles and Theoretical Background,” Cement and Concrete
Research, V. 31, No. 4, pp. 647-654.
Jensen, O. M., and Hansen, P. F., 2002, “Water-Entrained Cement-
Based Materials: II—Experimental Observations,” Cement and Concrete
Research, V. 32, No. 6, pp. 973-978.
Jensen, O. M.; Hansen, P. F.; Lachowski, E. E.; and Glasser, F. P., 1999,
“Clinker Mineral Hydration at Reduced Relative Humidities,” Cement and
Concrete Research, V. 29, No. 9, pp. 1505-1512.
Kada-Benameur, H.; Wirquin, E.; and Duthoit, B., 2000, “Determination
of Apparent Activation Energy of Concrete by Isothermal Calorimetry,”
Cement and Concrete Research, V. 30, No. 2, pp. 301-305.
Klieger, P., 1957, “Early High-Strength Concrete for Prestressing,”
Proceeding in World Conference on Prestressed Concrete, San Francisco,
CA, pp. A5(1)-A5(14).
Kovler, K., and Jensen, M., 2005, “Novel Technique for Concrete
Curing,” Concrete International, V. 27, No. 9, Sept., pp. 39-42.
322
Kovler, K.; Bentur, A.; Lange, D. A.; Bentz, D. P.; Van Breugel, K.;
Lura, P.; Zhutovsky, S.; and Souslikov, A., 2004, “Autogenous Curing
of High-Strength Cementitious Materials by Fine Uniformly Distrib-
uted Lightweight Aggregate,” Research Report to U.S.-Israel Binational
Science Foundation, Technion, Haifa, 78 pp.
Lura, P.; Winnefeld, F.; and Klemm, S., 2010, “Simultaneous Measure-
ments of Heat of Hydration and Chemical Shrinkage on Hardening Cement
Pastes,” Journal of Thermal Analysis and Calorimetry, V. 101, No. 3,
pp. 925-932.
Mather, B., 2001, “Self-Curing Concrete, Why Not?” Concrete Interna-
tional, V. 23, No. 1, Jan., pp. 46-47.
Powers, T. C., 1947, “A Discussion of Cement Hydration in Relation to
the Curing of Concrete,” Bulletin 25, Research Laboratories of the Portland
Cement Association, Chicago, IL, pp. 178-188.
Powers, T. C., and Brownyard, T. L., 1948, “Studies of the Physical
Properties of Hardened Portland Cement Paste,” Bulletin 22, Research
Laboratories of the Portland Cement Association, Chicago, IL, 892 pp.
Price, W. H., 1951, “Factors Influencing Concrete Strength,” ACI
Journal, V. 47, No. 2, Feb., pp. 417-432.
Schindler, A. K., and Folliard, K. J., 2005, “Heat of Hydration Models
for Cementitious Materials,” ACI Materials Journal, V. 102, No. 1, Jan.-
Feb., pp. 24-33.
Van Breugel, K., 1998, “Prediction of Temperature Development in
Hardening Concrete,” R. Springenschmid, ed., Prevention of Thermal
Cracking in Concrete at Early Ages, E&FN Spon, London, UK, pp. 51-75.
Villarreal, V. H., 2005, “Building Better Pavements through Internal
Hydration—A Work in Progress,” Proceedings of the R. Villarreal Inter-
national Symposium on Durability of Concrete, Monterrey, México, 13 pp.
Wadsö, L., 2010, “Operational Issues in Isothermal Calorimetry,”
Cement and Concrete Research, V. 40, No. 7, pp. 1129-1137.
Weber, S., and Reinhardt, H. W., 1997, “A New Generation of High-
Performance Concrete: Concrete with Autogenous Curing,” Advanced
Cement Based Materials, V. 6, No. 2, pp. 59-98.
ACI Materials Journal/May-June 2013
 ACI MATERIALS JOURNAL
Title no. 110-M28
Detection and Characterization of Early-Age Thermal
Cracks in High-Performance Concrete
by David Hubbell and Branko Glisic
Streicker Bridge is a new pedestrian bridge built on the Princeton
University campus. Structural health monitoring (SHM) is applied
with the aim of transforming the bridge into an on-site laboratory
for various research and educational purposes. Two fiber-optic
sensing technologies are permanently deployed: discrete long-
gauge sensing technology based on fiber Bragg gratings (FBG)
and truly distributed sensing technology based on Brillouin optical
time domain analysis (BOTDA). The sensors were embedded in
the concrete during construction. This paper describes the real-
time detection and characterization of early-age thermal cracks
in the high-performance concrete deck of Streicker Bridge. The
deployed monitoring strategy and the monitoring systems, which
successfully detected cracking, are described. The observed crack
propagation trajectories are presented. Details of a simple finite
element model (FEM) of the bridge are given, and the analysis
procedure used to demonstrate the formation of thermal cracks
using this model is outlined.
Keywords: crack detection and characterization; early-age thermal loads; fiber-
optic sensors; high-performance concrete; structural health monitoring.
INTRODUCTION
Cracking of concrete under restrained conditions during
early age may become a major durability problem, especially
for bridge decks. The early-age cracking of concrete can
significantly increase the vulnerability of structures to noxious
environmental influences. The cracks form “open doors” to
the infiltration and penetration of noxious substances, such
as sulfate water and chlorides.1 These substances attack
the concrete and reinforcing bar and damage the structure,
thereby reducing its long-term capacity and durability. The
early-age cracking of concrete under restrained conditions
occurs due to early-age stress caused by various internal and
external influences, such as autogenous shrinkage, plastic
deformation, loading, and temperature.2 The thermal effects
can be particularly challenging due to the rapid increase of
temperature during the intense hydration process and vari-
able environmental temperature that can significantly influ-
ence the cooling of the concrete and create conditions for
early-age cracking. The high strength and accelerated setting
time of high-performance concrete (HPC) are accompanied
by an increased risk of cracking due to thermally induced
stresses in restrained structural members.3 Early detection of
the early-age cracks is important, but this can be challenged
by the small size of the cracks that are invisible to the human
eye or the concealment of cracks behind formwork.
This paper describes the detection and characterization of
early-age thermal cracking observed in real time by fiber-
optic sensors embedded in the HPC deck of Princeton Univer-
sity’s Streicker pedestrian bridge. The applied monitoring
systems and strategies are presented and a global mechanism
by which the cracks formed is explained and duplicated in
a simplified finite element analysis. The focus of this paper
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
is not to present a detailed modeling of the crack growth
and crack distribution but rather to describe and analyze the
dynamics of cracking on a real structure in real conditions,
and to study, determine, and present the global circum-
stances that led to the crack occurrence.
RESEARCH SIGNIFICANCE
Understanding the mechanism through which early-age
cracks form is important in devising strategies for their
prevention. To the best of the authors’ knowledge, there are
no reported examples where the dynamics of the early-age
crack occurrence were measured in real time on a real struc-
ture. The direct observation of thermal cracking in Streicker
Bridge allows for a full characterization of the cracking and
explanation of its cause—specifically, the time of crack
occurrence, the dynamics of crack growth, the maximum
size of cracks, and the distribution of cracks along the bridge.
The applied fiber-optic method demonstrates how the cracks
could be detected in real time and characterized. This is
useful both in gaining confidence with HPC construction
as well as for validating numerical models of crack forma-
tion and propagation in HPC. The highlights of this research
are: 1) it is performed on a real structure in real on-site
conditions; 2) it uses embedded sensors to characterize the
cracking, thus allowing crack measurement at various depths
of the cross section; and 3) it investigates early-age cracking
on elements subjected to bending as opposed to traditional
laboratory approaches, where early-age cracking is investi-
gated on axially loaded elements.
STREICKER BRIDGE: DESIGN, CONSTRUCTION,
AND INSTRUMENTATION
Streicker Bridge design and construction
Streicker Bridge, pictured in Fig. 1, is a 300 ft (104 m)
long deck-stiffened arch pedestrian bridge over Washington
Road on the Princeton University campus. Conceptual
design was done by Swiss engineer Christian Menn. The
main deck span was poured in August 2009, while the south-
east approach span was poured in October 2009. The bridge
was officially opened for use in June 2010.
Menn’s design for Streicker Bridge is for a 114 ft (35 m)
steel tubular arch stiffened by a post-tensioned concrete deck.
In plan, the deck is comprised of two intersecting arches,
which provide lateral stability to the slender steel arch,
creating a deck that splits at each end of the main span into
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2011-317.R1 received October 11, 2011, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
323
David Hubbell is a PhD Candidate at the University of Toronto, Toronto, ON,
Canada. He received his BASc in engineering science from the University of Toronto
and his MSE from the Department of Civil and Environmental Engineering at Princeton
University, Princeton, NJ.
ACI member Branko Glisic is an Assistant Professor of Civil and Environmental
Engineering at Princeton University. He received his civil engineering and mathe-
matics degrees from the University of Belgrade, Belgrade, Serbia, and his PhD from
the Swiss Federal Institute of Technology, Lausanne (EPFL), Lausanne, Switzerland.
His research interests include structural health monitoring (SHM) and structural
analysis, SHM methods and strategies, fiber-optic sensors and advanced sensory
systems, and model-based and model-free data analysis.
Fig. 1—Streicker Bridge.
Fig. 2—Streicker Bridge plan and elevation. (Note: 1 ft =
3.28 m; 1 in. = 25.4 mm.)
Fig. 3—Location and naming of sensor pairs in southeast leg.
four approach legs. These legs are continuous post-tensioned
concrete girders of constant cross section, supported by
Y-shaped steel columns. Figure 2 shows the major dimen-
sions of the bridge, as well as the naming convention used to
refer to locations on the bridge.
The concrete deck was cast in place on wooden formwork.
The concrete mixture was designed to meet the New Jersey
Department of Transportation (NJDOT) specifications for
Class A High-Performance Concrete (NJDOT CLASS A
324
HPC4). The design class of concrete was 6000 psi (41.4 MPa)
in 28 days. The concrete was made using a superplasticizing
powdered admixture (Super P). The slump of the three
different delivered concrete mixtures used in the deck ranged
between 7.50 and 7.75 in. (190 and 197 mm) and the air
content ranged between 5.6 and 6.3%. Compressive strength
was determined on standard 4 in. (101 mm) cylinders at ages
of 3, 4, 7, 14, and 28 days. Compressive strength at 3 days
was for different mixtures between 3705 and 3725 psi
(25.5 and 25.7 MPa). The temperature of concrete at the
pouring varied for different mixtures between 66 and 68°F
(18.9 and 20.0°C), while the temperature of the air was
between 55 and 56°F (12.8 and 13.3°C). The pouring was
performed on a cloudy day. Curing mats with running water
between the mats and the deck were used to keep the surface
of the deck moist while curing.
Streicker Bridge instrumentation
During construction of the bridge, approximately
100 discrete long-gauge fiber-optic strain and temperature
sensors and approximately 87.5 yd (80 m) of distributed fiber-
optic sensing cable were embedded in the concrete deck by
the Structural Health Monitoring group at Princeton Univer-
sity’s Department of Civil and Environmental Engineering.
The bridge thus acted as an on-site laboratory demonstrating
the effectiveness of fiber-optic structural health monitoring
(SHM) technology5 and allowing the direct observation of
the structural behavior of a bridge in service.
Sensors were concentrated in the eastern half of the main
span and in the southeast leg. Because the presented research
is related to the southeast leg, only this portion of the moni-
toring system will be described in detail. Each of the four
approach legs of the bridge are continuous, curved post-
tensioned concrete girders with a constant cross section.
They are fixed to the deck of the main span on one end and
simply supported on neoprene bearings at the abutments at
the other end. The geometry of each of the legs is similar;
thus, only the southeast leg was chosen for monitoring, as
it contains the longest clear span in the bridge at 62.5 ft
(19.05 m).
Two types of sensors were installed in the southeast leg:
discrete long-gauge sensors based on fiber Bragg gratings
(FBG) placed at strategic points throughout the leg and
a distributed sensor cable based on stimulated Brillouin
optical time domain analysis (BOTDA) installed throughout
Span P10-P11.6 The FBG sensors, being more accurate than
the distributed BOTDA sensor, provide the main source of
monitoring data used in this research, with the BOTDA sensor
used as an independent source for verifying observations.
Pairs of discrete FBG sensors parallel to the centerline
were installed close to the axis of symmetry of the deck
at the nine locations shown in Fig. 3. These sensors were
placed in sets of two parallel sensors: one near the top of
the deck and the other near the bottom, as shown in Fig. 4.
The gauge length of the FBG sensors is 23.6 in. (600 mm).
The distributed BOTDA sensing cable is installed close to
the discrete sensors in Span P10-P11 to allow comparison
between the two systems.6 The parallel topology used
(shown in Fig. 4) measures curvature and axial deformations
in the deck through the equations
eb − et
φ= (1)
h
ACI Materials Journal/May-June 2013
 Fig. 4—Deck cross section and sensor location.

y  y 
ec = eb  t  + et  b  (2)
 h  h

where φ is curvature; ec is the strain at the centroid; eb is

the strain measured by the bottom sensor; et is the strain
measured by the top sensor; h is the distance between the
top and bottom sensors; yt is the distance from the centroid
to the top sensor; and yb is the distance from the centroid to
the bottom sensor.
Each FBG sensor consists of a pretensioned strain sensor
immediately adjacent to a second sensor that is sensitive to
temperature effects only. The temperature sensors are used Fig. 5—Photographs taken during sensor installation.
to compensate strain measurements for temperature effects, Clockwise from upper left: student installing sensor, parallel
as well as monitor the temperature distribution in the deck, discrete and distributed sensors installed onto reinforcing
especially during the early age of the concrete. Photos taken bar, and student installing extension cable.
during the installation of sensors are shown in Fig. 5.

OBSERVED CRACKING
As the sensors were embedded in the concrete, they moni-
tored early-age deformation of the deck. A sample of results
is presented in Fig. 6. Registered maximum expansion
due to hydration occurred 8 to 12 hours after the pouring
(depending on location and thus the sequence of pouring),
which is comparable with laboratory experiments performed
on cured specimens with embedded sensors.7,8 The maximum
expansion ranged between 50 and 100 me (depending on
location). The contraction due to cooling and shrinkage
relative to maximum expansion at the age of 3 days ranged
between –450 and –200 me (depending on sensor loca-
tion—that is, on thermally generated bending stress; refer
to further sections), which is also comparable with results
found in literature9,10 for similar types of cured concrete.
It is important to highlight that comparison with results of
laboratory tests found in literature was possible only quali-
tatively because specimens tested in the laboratory were not
subjected to bending. For example, a simplified comparison
can be performed as follows: assuming that the autogenous
shrinkage at 3 days is between –75 and –150 me7,9 and the
thermal contraction due to a temperature change of approxi-
mately –77°F (–25°C) is –225 to –240 me, the strain change
is expected to be between –300 and –390 me, which is
comparable with the range of measured strains in the deck
of the Streicker Bridge. The difference is attributed to differ-
ences in the concrete mixture and thermally induced bending
in the deck (refer to further sections).
Nevertheless, strain data from the southeast leg show
unexpected strain changes in the time histories of several of
the sensors. Three locations—P10h11, P11, and P11h12—
all show a distinct upswing in strain approximately 3 days
Fig. 6—Strain evolution showing cracking.
after the concrete was poured, with Location P10 showing a
similar behavior 5 days later. Figure 6 shows the strain data
from these four sensor pairs, with the change after 3 and 8 days
clearly visible.
The possibility of the anomalous changes in strain
observed at some sensors being caused by a sensor or
reading unit malfunction rather than an actual change in
strain in the bridge deck must be considered. However, the
BOTDA sensor independently detected similar behavior, as
shown in Fig. 7. A more detailed comparison between the
two systems can be found in the literature.6 Furthermore,
Fig. 6 shows that these changes are always present either in
both the upper and lower sensors at a location or in neither,
with each lower sensor showing a larger strain change than
its upper partner (refer to Fig. 8(a) and (b)). This temporal
and geometrical orderliness in the observations further indicates
that the changes are not the result of a sensor malfunction but
do indeed indicate unusual behavior in the concrete. The
preliminary explanation of this behavior is cracking.
Figure 9 summarizes the observed cracking, detailing the
locations of cracks, the order in which they formed, and the

ACI Materials Journal/May-June 2013 325


Fig. 7—Comparison between strain measurements obtained
with discrete (FBG) sensor and distributed (BOTDA) sensor
at cross section above Column P11.
Fig. 8—(a) Crack formation at Locations P10h11, P11, and
P11h12; and (b) crack formation at Location P10.
Fig. 9—Crack locations, order of formation, and sizes. (Note: 1 mm = 0.03937 in.; 1 ft =
3.28 m; 1 in. = 25.4 mm.)
326
approximate sum of crack widths at each location, calculated
by multiplying the observed strain change by the sensor
gauge length. The long-gauge sensor cannot distinguish
between the occurrence of only one crack and the distri-
bution of several smaller cracks along the gauge length;
therefore, the sum of crack openings at each location is
calculated. However, the long gauge length made the
sensor more likely to detect cracking. All crack openings
were wider on the underside of the deck than on top, which
suggests that the deck was subjected to positive bending (the
bottom of the cross section in tension, the top in compres-
sion). Figure 8 shows the evolution of measured strain for
each observed crack location.
Figures 8(a) and (b) shows that the crack dynamic has
three phases: 1) the tension is relatively slowly generated;
this phase may take from less than an hour to several hours;
2) the crack is quickly opened (within less than 10 minutes
[big jump in the diagram]); and 3) there is again slow stabili-
zation of the crack opening that may take from a few tens of
minutes to a few hours, depending on the external influences
and the position of the involved cross section on the bridge.
It is very difficult to accurately compare the determined
crack properties with results found in literature due to
different loads and boundary conditions. In the labora-
tory tests described in the literature,10,11 the specimens
were restrained in the axial direction and cracking was
induced by average temperature change and shrinkage,
while in the case of the Streicker Bridge, the deck was not
significantly restrained axially and cracking was induced
by bending caused by thermal gradients. In addition, the
size of cracks (and spacing) for axially restrained (and
thus axially loaded) specimens strongly depend on the
concrete mixture (development of hydration tempera-
ture and autogenous shrinkage) and reinforcing bar cage
properties (number and size of bars). For example, crack
openings found in Reference 10 ranged between 0.04 and
0.05 in. (0.9 and 1.2 mm), while in Reference 11, they
were between 0.001 and 0.005 in. (0.02 and 0.12 mm). The
results obtained in this research are comparable to those
found in Reference 11 if observed at a local scale.
Because the cracks did not propagate to the upper surface
and the lower surface was covered by formwork, visual
inspection of the concrete during construction could not detect
any cracking. Post-tensioning of the deck, performed 10 days
after the pouring of concrete, while the formwork was still
under the deck, visibly closed all cracks. An ultrasonic
pulse velocity test conducted in March 2010 showed a pulse
ACI Materials Journal/May-June 2013
velocity 1640 ft/s (0.5 km/s) slower in the vicinity of the
sensor at Location P11h12.12 This suggests that although the
crack was visibly closed by post-tensioning, some disconti-
nuity remains.

ANALYTICAL INVESTIGATION
The cracking in the deck of the southeast leg is the conse-
quence of restrained early-age stresses. As the bridge deck
was supported by formwork at the time of cracking—that is,
no applied load—the main sources of stress were tempera-
ture changes and early-age shrinkage. A simple finite element
model (FEM) was created with the goal of understanding
the mechanisms that led to the cracking. The model was Fig. 10—FEM geometry.
created using the finite element analysis program SAP2000.
The boundary conditions and geometrical properties of the
model are shown in Fig. 10.
The first important finding was that the slender columns
offer little axial restraint to the deck. The combined
stiffness of both columns in the direction of the deck axis is
approximately 57 kip/in. (10 MN/m [10 N/mm]), and the tensile
stress generated by the restraining action of the columns is, for Fig. 11—Moment diagrams for uniform temperature
the maximum measured total strain changes in the deck, less than gradient: light gray with P10 fixed and dark gray with P10
14.5 psi (0.1 MPa) (calculated for strain changes of 390 me in pinned (drawn on tension side of structure.)
Span P10-11 and 270 me in Span P11-12 measured with respect
to peak expansion 8 to 12 hours after the pouring). Although
this stress increases vulnerability to cracking, it is actually much
lower than the cracking strength of concrete. Consequently,
uniform temperature change and early-age shrinkage do not
cause significant stresses in the deck and thus cannot be the
primary cause of the cracking.
The second important finding was that the thermal stresses
aroused in the deck were primarily due to vertical restraint
from the columns. If the columns were not present, the greater
thermal expansion of the upper surface of the deck would
cause an upward, “hogging” deflection. The restraint of the
columns causes positive bending, resulting in tensile stresses
that are greatest at the bottom surface of the deck, while the
top surface is under compression. The FEM confirmed this
conclusion (refer to Fig. 11). Consequently, the cracks open
the most at the bottom of the cross section and penetrate
toward the upper surface until they reach the compres- Fig. 12—Coefficient of thermal expansion. (Note: °F =
sion zone of the concrete. As the top surface of the 1.8°C.)
concrete is in compression, the cracks do not reach the
upper surface of the deck. This mechanism is in keeping
with the detected cracks, which all show a greater change The measured contraction in this span due to post-tensioning
in strain in the lower sensor. The FEM showed that the shows that E ≈ 3750 ksi (26 GPa) at 10 days after casting.
thermomechanical bending stresses induced in the deck These values of E and a were taken as constant properties,
by the heat of hydration and the ambient temperature neglecting their very-early-age evolution, a simplification
were sufficiently large to cause cracking. The details of the that is justified for the purposes of this study and supported
analysis are described in the following. by results published in literature2,13 that show small variations in
E and a for HPC over the observed period of time.
Model geometry and mesh The cracking strength of the concrete can be estimated
The model geometry is shown in Fig. 10. The deck and from the compressive strength by empirical formulas based
supports are modeled with beam elements. The deck is meshed on either a direct tensile test or the modulus of rupture
into 72 elements, each approximately 1.7 ft (0.5 m) long. (MOR) test. The MOR is approximately twice the direct
cracking strength. In Streicker Bridge, the vertically asym-
Material properties metrical cross section (refer to Fig. 4) with complex geom-
The stresses induced by thermal changes in the deck etry subjected to a combination of bending strain (due to
are dictated by the coefficient of thermal expansion of the vertical restraints imposed by columns) and a small axial
concrete a, the modulus of elasticity E, and the support tensile elastic strain (due to average temperature change
fixity. Analysis of the measured thermal change and and early-age shrinkage) combine to produce a strain state
associated contraction in the unrestrained span between that is somewhere between being accurately represented by
Piers P12 and P13 shows that a ≈ 5 to 5.3 me/°F (9 to the MOR and direct tensile tests. The analysis described in
9.5 me/°C) at 1 to 3 days after casting, as shown in Fig. 12. the following proceeded assuming the direct tensile stress,

ACI Materials Journal/May-June 2013 327


Fig. 13—Force method analysis of southeast leg.
Fig. 14—Average temperature change applied to spans.
(Note: °F = 1.8°C.)
and it will be shown that the analysis based on this cracking
strength was able to explain the mechanism of crack forma-
tion, while the analysis based on MOR would show no risk of
cracking. Because cracking was observed by three indepen-
dent tests (FBG sensors, BOTDA sensor, and the ultrasonic
pulse velocity test), the authors concluded that cracking did
occur. The analysis showed that the direct tensile strength
seems to be a better representation of the true cracking
strength of the concrete in this particular case. Studying in
detail the reasons for the mismatch between the monitoring
results and the MOR strength prediction model is out of the
scope of this research. Preliminary analysis points to two
potential reasons:
• The cross section of the deck is quasi-triangular with
void areas (refer to Fig. 4), and the lowest and most
loaded area of the cross section in the case of bending
is actually a point (refer to Fig. 4), while in the case of
the MOR test, it is a line (rectangular cross section). The
complex geometry of the deck cross section may be a
source of stress concentration (especially at the lowest
point of the cross section), which can increase vulner-
ability to cracking compared with the MOR test.
• The tensile stress in the deck generated by uniform
temperature change and shrinkage is very small;
however, combined with bending stresses, it shifts the
neutral line above the centroid, and this increases the
area of the cross section that is in tension. Moreover,
due to the quasi-triangular shape of the deck cross
section, even a small shift in the neutral axis results in a
significant added cross-sectional area in tension, which
is not the case in the MOR test.
The direct tensile strength of the concrete is determined
from the compressive strength by the ACI 318-0214 equation
for normalweight concrete
fcr = 4 fc′ psi (0.33 fc′ MPa) (3)
328
Observed cracking occurred at a lower stress than that
given by Eq. (3). Because the forms were in place and the
deck was covered with mats and cured with water, the evap-
oration of water was prevented and could not contribute to
cracking. Shrinkage and uniform temperature change were
estimated to have a very small effect. Creep was also small
and helped decrease the stresses in the deck. A possible
explanation for the lower cracking resistance of the concrete
is the large size and unusual shape of the cross section.
The cross section of Streicker Bridge is roughly 20 times
larger than a typical direct tension test specimen, the width
being approximately 36 times larger and the depth being
approximately six times larger (the triangular shape and
voids make the area ratio smaller than 36 x 6). The studies in
References 15 and 16 were made on directly tensioned spec-
imens with a dogbone shape, and they show that a change
in one dimension of the specimen from 4 in. (0.1 m) (diam-
eter of the cylinders tested for determining deck concrete
strength) to 144 in. (3.6 m) reduces the cracking strength
by a factor between approximately 0.55 and 0.60, while the
change in size from 4 to 24 in. (0.1 to 0.6 m) reduces the
cracking strength by a factor of 0.75 to 0.80. Combining
these two factors, the authors obtain a cracking strength for
the bridge reduced by 0.41 to 0.48. A reduction factor of
0.45 (fourth root 1/20, which is also in accordance with other
sources17) is reasonable and gives good agreement between
the predicted stresses and the observed onset of cracking.
Using this reduction factor, the cracking strength of the
concrete deck is taken as
fcr = 1.8 fc′ psi (0.15 fc′ MPa) (4)
Boundary conditions
The southeast leg is supported at four points:
• P13: At the abutment, the deck rests on a neoprene
bearing, which can be modeled as a roller, free to trans-
late and rotate in all directions.
• P11 and P12: The two Y-shaped steel piers are bolted all
around their perimeters to their footings, and are thus
modeled as fixed at the base. The slenderness of the
piers themselves effectively creates a roller support for
the deck, as discussed previously in the text.
• P10: The boundary between the southeast leg and the
main span is at Pier P10, making the fixity at this point
unknown. To account for this uncertainty, upper- and
lower-bound analyses were performed with P10 entirely
fixed and entirely free. Figure 11 shows the shape of
the moment diagram for these two cases. The maximum
moment for each case is similar, and a significant differ-
ence is noticeable only between P10 and Midpoint
P10h11 of Span P10-P11. As the first crack occurred
at P10h11, it is likely that the cross section at P10 was
fixed—that is, it behaved as designed.
Loading
The temperature variations and the shrinkage at early age
generate stresses in the concrete only after it is hardened.
A time-varying temperature load caused by cooling of the
concrete after the initial peak in temperature due to the heat
of hydration was applied to the model in two components:
a vertical temperature gradient (causing curvature) and a
temperature change at the centroid (causing axial deformation).
Shrinkage was not separately analyzed, as its global effect is
ACI Materials Journal/May-June 2013
similar to the effect of temperature change at the centroid.
Temperature loading was applied to the model assuming
that the concrete hardened when the peak temperature was
reached. This assumption is validated by laboratory tests on
full-scale specimens2 that show concrete hardening occur-
ring approximately 1 to 2 hours before the peak tempera-
ture was reached. A sensitivity study performed on the data
demonstrated that the shift of 1 to 2 hours in the reference
point for temperature changes will have a negligible effect
on the analysis results, making the peak temperature a
convenient and rational point at which to assume concrete Fig. 15—Linearization of temperature gradient.
hardening.
To further simplify the analysis, temperature loads were
averaged over each of the three spans. This can be done
without significant loss in accuracy, as demonstrated by
a force method analysis of the southeast leg, as shown in
Fig. 13. The moments in the structure are dictated by the
corrective forces applied at P11 and P12. The magnitudes
of these forces are determined by the displacement at
Locations P11 and P12. These displacements are in turn
calculated as the double integration of curvature. If an
average curvature is assumed over each span, the double
integral is unchanged, meaning that the displacements,
corrective forces at P10 and P11, and moments in the struc-
ture are all unchanged. Therefore, averaging the tempera-
ture changes over each span does not reduce the accuracy
of the analysis. Figure 14 shows the centroidal and gradient
temperature changes applied to each span.

RESULTS AND DISCUSSION
The analysis confirmed that the slender columns of bridge
behave as rollers to the deck—that is, the deck is not signifi-
cantly restrained in the axial direction. Consequently, the
centroidal temperature change and the early-age shrinkage
did not generate important stresses. However, the tempera-
ture gradients did generate important stresses that caused the
observed cracking in the concrete.
Applying a temperature gradient of 1.8°F/ft (3.3°C/m) to
the modeled bridge deck produces maximum tensile strains in
the deck of 13 me when P10 is pinned and 15 me when P10 is
fixed. The maximum observed gradient averaged over the
three spans is approximately 10.8°F/ft (19.7°C/m). Scaling
up the FEM strains for this gradient gives a maximum tensile
strain in the deck between 78 and 90 me. This is well within
the range of cracking strains for mature concrete, typically
assumed to be between 60 and 200 me.18
To confirm the reasonability of this result, a rough
comparison with the observed strains can be carried out as
follows. Figure 14 shows that the temperature change at the
time of cracking was approximately 25°C (77°F), corre-
sponding to a thermal strain of 225 me. The total post-set
strain change observed just prior to cracking—comprising
thermal strain, elastic strain, and shrinkage strain—was,
from Fig. 6, between 250 and 450 me. Subtracting the
thermal strain from the total gives the sum of elastic and
shrinkage strains as between 25 and 225 me. This is within
the same range as the strains predicted by the simplified
FEM analysis described previously.
It should be noted that in this calculation, as well as in
the more detailed analysis to follow, the temperature distri-
bution measured by the parallel sensors is assumed to vary
linearly between the two sensors. In reality, this will not be
the case.19 Figure 15 shows an example of how the actual
and assumed temperature gradients could differ. Including a
Fig. 16—Maximum deck stress at Locations P10h11 and
P11 over time, showing predicted cracking 3 to 4 days
after casting.
third temperature sensor in between the two parallel sensors
would allow for a more accurate measurement of the actual
temperature distribution in the deck. For the purpose of
determining the mechanism of cracking, however, a linear
approximation is sufficient.
As a more detailed analysis, the maximum stress in the
deck is compared to the evolving cracking strength of the
concrete, as shown in Fig. 16. It can be seen that the stresses
in the deck approach the cracking strength of the concrete
on the afternoon of October 26, roughly 3 days after the
concrete was poured. This is close to the same time that
the sensors observed an unexpected change in strain, as
shown in Fig. 6. The discrepancy between the model and
measurements can be explained by inaccuracy in the model
introduced by simplifications (in particular linearity). As
the purpose of the model was to understand global mecha-
nisms that led to cracking, rather than to accurately model
the behavior of the concrete at crack locations, the results
obtained by the model are considered acceptable. The crack
that formed at Sensor P10 appears to have been caused by
particularly warm weather on October 31 and the size and
dynamics of this crack were different from the cracks that
occurred on October 26.
CONCLUSIONS
The real-time observation of early-age cracking in the
HPC deck of Streicker Bridge has been presented. The order
and timing of crack formation throughout the deck and the
observed propagation of each individual crack demonstrate

ACI Materials Journal/May-June 2013 329

the time scale on which thermal cracking occurs. The main
conclusions of the study presented herein are:
1. The monitoring system based on embedded long-gauge
fiber-optic sensors was capable of successfully detecting
and characterizing early-age cracking in the deck, which was
invisible to the eye and concealed by formwork. The presence
of cracks has been confirmed by an independent distributed
sensing system and an ultrasonic pulse velocity test.
2. Simplified linear finite element analysis helped to
understand the global mechanism behind the cracking; based
on the timing and magnitude of the heat of hydration, the
finite element analysis predicted early-age thermal cracking,
with the accuracy of predicted stresses and the time of crack
occurrence within acceptable limits.
3. The cracking was found to be predominantly a conse-
quence of bending moments in the vertically restrained
deck caused by thermal gradients due to hydration heat and
ambient temperature variations. The axial temperature varia-
tion component and the early-age shrinkage did not signifi-
cantly contribute to the cracking occurrence, as the deck was
only minimally restrained in the axial direction, although
these effects could contribute to the size of crack openings
once the concrete was cracked.
4. The first three cracks occurred at 3 days after the pouring
as a consequence of uneven cooling of the upper and lower
parts of the deck. These cracks propagated eastward from the
center of Span P10-P11. The first at Location P10h11 took
approximately 2 hours to stabilize and the second at Loca-
tion P11 took approximately 3.5 hours to stabilize. It stabi-
lized after the crack at Location P11h12, which opened
last but stabilized after only 1 hour. All three cracks have
approximately the same size within ±0.02 mm (±0.8 mils)
limits (upper crack openings are at 0.04 ± 0.02 mm [1.6 ±
0.8 mils] and lower crack openings are at 0.09 ± 0.02 mm
[3.5 ± 0.8 mils]).
5. The fourth crack occurred 8 days after the pouring
and this was the consequence of unusually high ambient
temperature variation; it took approximately 5 hours for this
crack to stabilize and the size of the lower crack opening was
bigger than in the case of the other three cracks.
6. Significant thermal gradients can form during the early age
of a concrete bridge deck. If the deck is restrained at columns
or construction joints, then thermal cracking is a concern.
The research was performed on a real structure and in real
on-site conditions. It used embedded sensors to characterize
the cracking and it presents early-age cracking on elements
subjected to bending as opposed to traditional laboratory
approaches, where early-age cracking is investigated on
axially loaded elements.
In the case of Streicker Bridge, post-tensioning has largely
closed the cracks. This, coupled with the fact that the cracks
were widest on the underside of the deck, means that water
penetration is not expected, and the cracks do not pose a
significant threat to the performance of the structure.
ACKNOWLEDGMENTS
The following individuals and companies contributed to the Streicker
Bridge SHM project: J. P. Wallace at Princeton Facilities; S. Hancock and
Turner Construction Company; T. Zoli and R. Woodward at HNTB; D. Lee
and A.G. Construction Corporation; S. Mancini and T. R. Wintermute,
330
Vollers Excavating & Construction, Inc.; SMARTEC SA, Switzerland;
Micron Optics, Inc., Atlanta, GA.; G. Gettelfinger; M. Hersey; P. Prucnal;
Y. Deng; and M. Fok. The following students helped to install the sensors:
M. Wachter, J. Hsu, G. Lederman, J. Chen, K. Liew, C. Chen, A. Halpern,
M. Neal, D. Reynolds, K. Bakis, and D. Schiffner.
REFERENCES
1. Nagesh, M., and Bhattacharjee, B., “Modeling of Chloride Diffusion
in Concrete and Determination of Diffusion Coefficients,” ACI Materials
Journal, V. 95, No. 2, Mar.-Apr. 1998, pp. 113-120.
2. Glisic, B., “Fibre Optic Sensors and Behaviour in Concrete at Early
Age,” PhD thesis, École Polytechnique Fédérale de Lausanne (EPFL),
Lausanne, Switzerland, 2000, 154 pp.
3. Nagy, A., “Comparative Study of Early Age Cracking in HPC and NPC
Structures,” Proceedings of RILEM Conference on Early Age Cracking in
Cementitious Systems (EAC’01), Cachan, France, 2002, pp. 319-326.
4. NJDOT Specification 2007, “Standard Specification for Road and
Bridge Construction, Division 900 Materials,” http://www.state.nj.us/trans-
portation/eng/specs/2007/spec900.shtm#s903. (last accessed May 14, 2011)
5. Glisic, B., and Adriaenssens, S., “Streicker Bridge: Initial Evalua-
tion of Life-Cycle Cost Benefits of Various Structural Health Monitoring
Approaches,” Bridge Maintenance, Safety, Management and Life-Cycle
Optimization—Proceedings of the 5th International Conference on
Bridge Maintenance, Safety and Management, Philadelphia, PA, 2010,
pp. 2112-2118.
6. Glisic, B.; Chen, J.; and Hubbell, D., “Streicker Bridge: A Comparison
between Bragg-Grating Long-Gauge Strain and Temperature Sensors and
Brillouin Scattering-Based Distributed Strain and Temperature Sensors,”
Proceedings of SPIE—The International Society for Optical Engineering,
V. 7981, No. 79812C, 2011, 10 pp.
7. Viviani, M.; Glisic, B.; and Smith, I. F. C., “Separation of Thermal
and Autogenous Deformation at Varying Temperatures Using Optical
Fiber Sensors,” Cement and Concrete Composites, V. 29, No. 6, 2007,
pp. 435-447.
8. Viviani, M.; Glisic, B.; and Smith, I. F. C., “System for Monitoring the
Evolution of the Thermal Expansion Coefficient and Autogenous Deforma-
tion of Hardening Materials,” Journal of Smart Materials and Structures,
V. 15, 2006, pp. N137-N146.
9. Habel, K.; Charron, J.-P.; Braike, S.; Hooton, R. D.; Gauvreau, P.; and
Massicotte, B., “Ultra-High Performance Fibre Reinforced Concrete Mix
Design in Central Canada,” Canadian Journal of Civil Engineering, V. 35,
2008, pp. 217-224.
10. Shah, S. P., and Weiss, W. J., “High Strength Concrete: Strength,
Permeability, and Cracking,” Proceedings of the PCI/FHWA Interna-
tional Symposium on High Performance Concrete, Orlando, FL, 2000,
pp. 331-340.
11. Hughes, B. P., and Mahmood, A. T., “Laboratory Investigation of
Early Thermal Cracking of Concrete,” ACI Materials Journal, V. 85, No. 3,
May-June 1988, pp. 164-171.
12. Hubbell, D., “Determining the Structural Behaviour of Streicker
Bridge Using Fiber-Optic Sensors,” MSE thesis, Princeton University,
Princeton, NJ, 2011, 80 pp.
13. Østergaard, L.; Lange, D.; and Stang, H., “Early-Age Stress-Crack
Opening Relationships for High Performance Concrete,” Cement and
Concrete Composites, V. 26, No. 5, July 2004, pp. 563-572.
14. ACI Committee 318, “Building Code Requirements for Structural
Concrete (ACI 318-02) and Commentary,” American Concrete Institute,
Farmington Hills, MI, 2002, 443 pp.
15. Van Vliet, M. R. A., and Van Mier, J. G. M., “Effect of Strain Gradi-
ents on the Size Effect of Concrete in Uniaxial Tension,” International
Journal of Fracture, V. 95, 1999, pp. 195-219.
16. Vorechovsky, M., “Interplay of Size Effects in Concrete Specimens
under Tension Studied via Computational Stochastic Fracture Mechanics,”
International Journal of Solids and Structures, V. 44, 2007, pp. 2715-2731.
17. Collins, M. P., and Mitchell, D., Prestressed Concrete Structures,
Response Publications, Toronto, ON, Canada, 1997, 766 pp.
18. Neville, A. M., Properties of Concrete, fourth edition, Pearson
Education, Harlow, UK, 1995, 846 pp.
19. Gasch, I.; Alvarado, Y. A.; and Calderón, P. A., “Temperature Effects
on Load Transmission between Slabs and Shores,” Engineering Structures,
V. 39, 2012, pp. 89-102.
ACI Materials Journal/May-June 2013
 ACI MATERIALS JOURNAL
Title no. 110-M29
Alkali-Activated Natural Pozzolan Concrete as New
Construction Material
by Dali Bondar, Cyril J. Lynsdale, and Neil B. Milestone
In the near future, geopolymers or alkali-activated cementitious
materials will be used as new high-performance construction
materials of low environmental impact with a reasonable cost. This
material is a good candidate to partially replace ordinary portland
cement (OPC) in concrete as a major construction material that
plays an outstanding role in the construction industry of different
structures. Geopolymer materials are inorganic polymers based on
alumina and silica units; they are synthesized from a wide range of
dehydroxylated alumina-silicate powders condensed with alkaline
silicate in a highly alkaline environment. Geopolymeric materials
can be produced from a wide range of alumina-silica, including
natural products—such as natural pozzolan and metakaolin—or
coproducts—such as fly ash (coal and lignite), oil fuel ash, blast
furnace or steel slag, and silica fume—and provide a route toward
sustainable development. Using lesser amounts of calcium-based
raw materials, lower manufacturing temperature, and lower
amounts of fuel result in reduced carbon emissions for geopolymer
cement manufacture up to 22 to 72% in comparison with portland
cement. A study has been done by the authors to investigate the
intrinsic nature of different types of Iranian natural pozzolans to
determine the activators and methods that could be used to produce
a geopolymer concrete based on alkali-activated natural pozzolan
(AANP) and optimize mixture design. The mechanical behavior
and durability of these types of geopolymer concrete were inves-
tigated and compared with normal OPC concrete mixtures cast by
the authors and also reported in the literature. This paper summa-
rizes the main conclusions of the research regarding pozzolanic
activity, activator properties, engineering and durability proper-
ties, applications and evaluation of carbon footprint, and cost for
AANP concrete.
Keywords: alkali-activated cementitious materials; geopolymer concrete;
natural pozzolan; sustainable development.
INTRODUCTION
Concrete is a mixture of portland cement, water, aggre-
gates, and additives that is used as a durable construction
material. Thus, revising the ingredients and production
method of conventional concrete is important and makes a
route toward making this material more sustainable.
Portland cement is used as a binder in the production of
concrete and high consumption of concrete as a construc-
tion material causes vast requirements of cement produc-
tion. It is produced by mixing selected raw materials
(clay and lime) with a given proportion, and then grinding
and heating them at 2732°F (1500°C). In the process of
producing 2000 lb (1 ton) of cement, 27.5 gal. (125 L) of
fossil fuel and 118 kW·h of electricity is consumed.1 Higher
amounts of calcium-based raw materials, higher manufac-
turing temperature, and higher amounts of fuel increase
carbon emissions in portland cement production. More than
65% of the earth’s crust consists of Al-Si minerals; using
lesser amounts of calcium-based cement does not release
large quantities of CO2, so it is useful to understand how
these minerals convert to cement.2-6 For geopolymeric
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
cement production, no temperature higher than 1382°F
(750°C) is ever needed. The aforementioned item means
only one-third of the fuel requirement is needed for this kind
of cement production. Geopolymeric cements do not rely
on the calcinations of calcium carbonate and do not require
high-temperature kilns with a large expenditure of fuel; thus,
the chemical process and the consumption of significantly
less fuel reduces the carbon dioxide emission for geopoly-
meric cement production. Further, a good solution for
producing an environmentally friendly type of concrete is
to reduce the amount of portland cement and replace it with
a material that has a lower manufacturing temperature, such
as geopolymer cement. This paper briefly describes the use
of natural source materials to produce geopolymer concrete,
including its engineering and durability properties based on
the previous investigation by the authors, and finally evalu-
ates the carbon footprint and cost of this new type of concrete
while mentioning some of its applications.
RESEARCH SIGNIFICANCE
Unfortunately, the production of portland cement as
a major construction material worldwide releases large
amounts of CO2 into the atmosphere (production of 2000 lb
[1 ton] ordinary portland cement [OPC] releases 2000 lb
[1 ton] CO2), and this gas is a major contributor to the green-
house effect and the global warming phenomenon.7 The main
benefit of geopolymeric cement is the reduction in environ-
mental impacts and a move toward sustainable development,
which is explained as the optimum use of basic and natural
resources with correct and efficient operation to provide the
requirements of the future generations. This includes less
emission of carbon dioxide, more efficient use of mineral
and metal resources, and increased use of recycled materials.
The issue of long-term durability was studied in relation to
archeology analogues—namely ancient Roman cement. The
research data presented in this paper are useful to under-
stand the engineering and durability properties, evaluation
of carbon footprint, and cost of alkali-activated natural
pozzolan (AANP) concrete as a new sustainable construc-
tion material in comparison with OPC concrete.
MATERIALS
Five natural pozzolans, which are currently used to produce
portland pozzolan cement by Iranian cement factories, have
been considered in this work. These are Shahindej and
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2012-009.R1 received June 28, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
331
Dali Bondar is a Research Engineer at the Natural Disasters Research Center, Power
and Water University of Technology (PWUT), Tehran, Iran. She received her BS from
Amirkabir University of Technology (Tehran Polytechnic), Tehran, Iran; her MS
from the University of Science and Technology, Tehran, Iran; and her PhD from the
University of Sheffield, Sheffield, UK. Her research interests include alkali activation
of natural pozzolan as sole binders for concrete and its properties.
Cyril J. Lynsdale is a Lecturer at the University of Sheffield in cement and concrete
materials and technology. He received his MS and PhD from Leeds University, Leeds,
UK. His research interests include studying concrete materials and the use of various
cementitious materials for the development of environmentally friendly concrete
promoting sustainable construction.
Neil B. Milestone was a Senior Lecturer of cement chemistry at the University of
Sheffield and now runs his own consultancy business in New Zealand. He received his
MS from Victoria University, Wellington, New Zealand, and his PhD from the Univer-
sity of Waikato, Hamilton, New Zealand. His research interests include the materials
chemistry of silicates.
Sahand dacites from the northwest, and Sirjan and Rafsanjan
dacites and Taftan andesite from the southeast of Iran. All
were prepared with a particle size of 100% less than 246 mf
(75 mm) and used to produce portland pozzolan cement by
Ourmia, Ardebil, Kerman (using two sources of pozzo-
lanic materials, including Sirjan and Rafsanjan dacites), and
Khash cement factories, respectively. Because most of these
raw materials contain zeolites and clay minerals and have a
high loss on ignition (LOI), they were heated at 1292 and
1472°F (700 and 800°C) to improve their pozzolanic prop-
erties. X-ray diffraction (XRD) and X-ray fluorescence
(XRF) were carried out to determine chemical compositions
Fig. 1—Comparison of XRD of five untreated pozzolans.
Table 1—Mineralogy of investigated pozzolans
Material Shahindej pozzolan Sahand pozzolan
Clinoptiolite Albite
Albite Hornblende
Quartz Quartz
Minerals Calcite Muscovite
— —
— —
— —
332
and different minerals in each pozzolan. The mineralogy,
chemical composition, and physical properties of these five
pozzolans in their natural state and after calcining were
compared to determine the pozzolanic reactivity. The results
are presented in Fig. 1 and Tables 1 through 4, respectively.
The XRD patterns of the minerals in Fig. 1 showed that
all of them are mixtures of minerals with various degrees
of crystallization. All of these pozzolans contain albite
as the main mineral phase. Four of them—Shahindej,
Sahand, Sirjan, and Rafsanjan—contain both zeolites and
clay minerals, including clinoptiolite, muscovite, mont-
morillonite and biotite, and three—Sirjan, Rafsanjan, and
Taftan—contain approximately 30% amorphous materials.
All five pozzolans contain SiO2 and Al2O3 with the
SiO2 content varying from 61.67 wt.% in the Taftan andesite
to 70.13 wt.% in the Shahindej dacite. The Al2O3 content
varies from 11.11 wt.% in Shahindej dacite to 15.90 wt.% in
Taftan andesite, while the CaO content varies from 2.52 wt.%
in Shahindej dacite to 7.99 wt.% in Taftan andesite. The
calcium content is an important factor affecting the setting
time and final strength in concrete and there are indica-
tions that it also affects the properties of geopolymers.8-10 In
geopolymerization, the dissolution reaction and polycon-
densation steps involve alkali metals, which implies that the
alkali content of the reacting minerals could have a signifi-
cant effect on strength development.11
Potassium hydroxide (KOH) pellets were dissolved
to produce the alkaline solutions for geopolymeric paste
production. Sodium silicate was provided in the form of
Sirjan pozzolan Rafsanjan pozzolan Taftan pozzolan
Albite Montmorillonite Albite
Montmorillonite Albite Hornblende
Quartz Hornblende Quartz
Biotite Biotite —
— Calcite —
— Quartz —
Amorphous Amorphous Amorphous
ACI Materials Journal/May-June 2013
Table 2—Chemical composition (oxide percent) of materials used in this investigation
Material LOI SiO2 Al2O3 Fe2O3 CaO MgO TiO2 K2O Na2O
Shahindej dacite 10.28 70.13 11.11 1.27 2.52 0.92 0.14 2.25 1.01
Sahand dacite 5.15 64.67 11.85 3.03 6.79 1.11 0.537 4.26 2.3
Sirjan dacite 6.14 68.51 11.84 3.73 2.90 1.43 0.366 3.19 1.62
Rafsanjan dacite 4.41 68.31 12.59 2.70 3.88 1.37 0.263 3.26 2.40
Taftan andesite 1.85 61.67 15.90 4.32 7.99 2.04 0.438 2.12 3.21

Table 3—Chemical composition (oxide percent) of calcined materials used in this investigation
Material LOI SiO2 Al2O3 Fe2O3 CaO MgO TiO2 K2O Na2O
Shahindej dacite, 800°C 5.78 73.44 11.88 1.30 2.55 0.98 0.147 2.30 1.10
Sahand dacite, 800°C 2.90 67.40 11.87 3.10 6.87 1.18 0.564 3.41 1.90
Sirjan dacite, 700°C 2.20 68.36 13.40 4.41 3.90 1.20 0.508 2.66 2.78
Rafsanjan dacite, 800°C 2.48 71.53 12.46 2.76 3.93 1.46 0.276 2.53 1.87
Notes: °F = 1.8°C +32.

a solution (water glass). The chemical composition of the Table 4—Physical properties measured by
solution provided by the manufacture was: manufacturer
(a) 12.6% of sodium oxide (Na2O), 26.5% of silicon oxide Material Specific gravity Specific surface, cm2/g
(SiO2), and 60.9% of water; pH = 13;
(b) 11% of sodium oxide (Na2O), 26.5% of silicon oxide Shahindej dacite 2.2 10,621
(SiO2), and 62.5% of water; pH = 12.2; and Sahand dacite 2.7 6331
(c) 8.5% of sodium oxide (Na2O), 26.5% of silicon oxide Sirjan dacite 2.28 6348
(SiO2), and 65% of water; pH = 11.4.
Rafsanjan dacite 2.08 4870
ACTIVATION OF NATURAL POZZOLANS FOR Taftan andesite 2.22 3836
PRODUCTION OF GEOPOLYMER BINDER Note: 1 cm2/g = 70.37 in.2/lb.
The primary aim of this study was to investigate the
characteristics of five pozzolans from Iran (Table 1),
both in their natural state and after calcination at different
temperatures, as sources for the preparation of geopolymer
binders. Elevated temperature curing of pastes containing
the pozzolans was considered. This study used alkali solu-
bility and compressive strength as the indicators for pozzo-
lanic activity and showed that the highest reactivity and best
behavior resulted for pozzolans such as Taftan andesite with
a low LOI and a high soluble calcium content, which can
be activated directly from raw material without calcination.
For pozzolans containing sodic zeolites, such as Shahindej
dacite, on calcination at 1472°F (800°C), its clinoptilolite
minerals convert to amorphous opaline materials, which
react readily with aqueous alkali, making it suitable for
producing geopolymers at room temperature12,13 (Fig. 2).
The most important finding of this work was that Fig. 2—Different calcination and curing temperatures with
geopolymer binders can be synthesized by activating natural related compressive strengths of investigated pozzolans.
pozzolans, condensed with sodium silicate in a highly alkaline (Note: 1 MPa = 0.145 ksi; °F = 1.8°C + 32.)
environment.11,12 A new model is presented that allows for
prediction of the AANP strength versus their alkali solubility, was tried combining the first three parameters as inputs
chemical composition, and crystallinity.12,14 to find compressive strength using the least-squares tech-
A model has been proposed for assessing the pozzolanic nique with the effect of curing temperatures. This model
reactivity of pozzolans in both natural and calcined forms was improved with a nonlinear model that considered three
in terms of compressive strength of the pozzolanic-based further parameters,including LOI, the ratio of ([SiO2 + Al2O3
geopolymer binder to save both resources and time. The + CaO] solution/[SiO2 + Al2O3 + CaO]mineral) from ICP
correlation between compressive strength and relevant tests and the quartz percentage, thus obtaining better corre-
parameters includes the alkali solubility, alkali content, lation (R > 0.93) between the predicted and observed values
activity index, LOI, the ratio of (SiO2 + Al2O3 + CaO) in of compressive strength of pozzolanic-based geopolymer
solution to (SiO2 + Al2O3 + CaO) mineral (obtained from binder.12,14
inductively coupled plasma [ICP] measurements), and This investigation has shown that the most efficient
quartz percentage, investigated one by one. A linear model activator for activating natural pozzolans is a combina-

ACI Materials Journal/May-June 2013 333

Table 5—Concrete mixture proportion
Mixture Pozzolan, OPC, KOH, Na2SiO3, Water,
No. kg/m3 kg/m3 kg/m3 cc/m3 kg/m3
CM1 — 403 — — 180
ATAF1 391 — 66.27 34.18 157.78
ARSH 417.43 — 66.91 31.83 159.31
ACSH 417.43 — 66.91 31.83 159.31
Note: 1 kg/m3 = 0.06 lb/yd3.
tion of potassium hydroxide and sodium silicate solution.
The optimum dosage of activators is determined by a new
method that draws the strength contours versus different
molarities of alkaline hydroxide and various ratios of alka-
line hydroxide to alkaline silicate, allowing the islands of
ideal compositions to be defined. The results show that for
pozzolans containing high soluble silicate with low alka-
linity, a sodium silicate with lower SiO2 to Na2O ratios
gives higher strengths, but when a calcined or a natural form
with higher alkalinity is used, the reverse is true.11,14 Using
mineral additives, including kaolinite, other calcined pozzo-
lans such as Shahindej pozzolan, and lime when added to
Taftan pozzolan as a solid precursor is found to give approx-
imately the same strength, although it seems that the gel
obtained is more impermeable than when the pozzolan is
activated without mineral additives.12,16
MIXTURE DESIGN, PROCEDURE, AND
CURING TEMPERATURE
This research has successfully used geopolymer binders
made from AANP to produce concrete instead of using an
OPC cement paste. To determine the compressive strength of
pozzolanic materials, 5.2 × 10–6 yd3 (4 mL) of 7.5 M KOH
solution was added to 0.033 lb (15 g) pozzolan powders and
blended for 5 minutes before the addition of 6.5 × 10–7 yd3
(0.5 mL) of sodium silicate solution with a solid content
equal to 2.1%.11 The ratio of total dry mix (g)/total solution
(mL) was 3.3. Each of the subsequent slurries was mixed
for 5 minutes and transferred to a polyvinyl chloride (PVC)
mold measuring 20 x 20 x 20 mm (0.79 x 0.79 x 0.79 in.).
The samples were removed from the mold after 24 hours
and cured at 20, 40, 60, and 80°C (68, 104, 140, and 176°F)
for 27 days to study ambience and elevated temperature
curing of pozzolan pastes and the compressive strengths for
these samples were measured. The maximum compressive
strength resulted for each pozzolan at 28 days and the corre-
sponding calcination and curing temperature were reported.
To produce the best paste and ongoing mixture designs,
the effect of water content was studied and it was found that
the minimum water-binder ratio (w/b) for concrete mixtures
made with AANPs should be at least equal to the sum of
plastic limits, which is determined by the falling cone test
and aggregate water absorption. This amount would prepare
a low-workability concrete that achieves the maximum
strength when hardened. The proportioning of the concrete
mixture was based on the BRE Method of mixture design
considering the approaches for determining the minimum
w/b. Then, the amount of cement was substituted with the
same quantity of natural pozzolan plus the solids in water
glass with the water in the activator also taken as part of
the total mixture water. The details of the different mixtures
are presented in Table 5 and the notation for the mixtures is
as follows:
334
Total water, Total binder, Fine aggregate, Coarse aggregate, w/b,
kg/m3 kg/m3 kg/m3 kg/m3 %
181.3 403 577.82 1228 0.45
180 403 578.17 1229 0.45
180 428.57 498.8 1283 0.42
180 428.57 498.8 1283 0.42
• CM1: OPC control mixture;
• ATAF1: activated Taftan pozzolan;
• ARSH: activated raw Shahindej pozzolan mixture; and
• ACSH: activated calcined Shahindej pozzolan mixture.
To achieve the best results in geopolymer concrete, mixing
should be done in three stages by adding the hydroxide alka-
line solution to the natural pozzolan first, followed by an
alkaline silicate solution, and then adding the mixed paste
to aggregates.
This type of binder usually needs a temperature higher
than room temperature for curing to be activated because
pozzolans such as Taftan, which contain feldspars as main
mineral phases, seem to need a higher activation energy to be
activated with alkali. For Shahindej, it was shown that when
pozzolans that contain zeolite minerals such as clinoptilolite
are calcined, they can be activated at room temperature.12,17
FRESH PROPERTIES OF AANP CONCRETE
A geopolymer mixture can be timed to set either fast or
slow by adjusting the mixture components. Depending on
the synthesis conditions, structural integrity and reasonable
strength were attained in a short time—sometimes in as little
as 60 minutes.18 Using granulated blast furnace slag as the
source material with the addition of metakaolinite, Cheng
and Chiu19 found that the setting time of the geopolymer
paste was affected by curing temperature, type of alkaline
activator, and the composition of source material. They stated
that the setting time of the aforementioned geopolymer paste
was between 15 to 45 minutes at 140°F (60°C).19 The time
available between the end of mixing and the start of casting
of fresh geopolymer concrete may be more relevant in prac-
tical applications and measuring the setting time at elevated
temperatures may not be appropriate. The laboratory experi-
ence by Hardjito et al.20 showed that the fresh geopolymer
concrete could be handled for at least 120 minutes after mixing
without any sign of setting and degradation in compres-
sive strength.20 These results depended on the composition
of the source material, with higher CaO contents allowing
for faster setting. The presence of components other than
Al2O3 and SiO2 in the source material may also delay the
setting. In materials of pure geological origin (for example,
calcined kaolin), the dominant chemical contents are only
Al2O3 and SiO2, whereas by-product materials such as fly
ash may contain other compounds (for example, Fe2O3).
Therefore, it appears that pure geological materials may be
more reactive with alkaline activators and induce a reduc-
tion in initial setting time.21-23 In AANP concrete, this can
be classified as “sticky” concrete—the higher the amount of
silicate concentration, the longer the setting time—unless
the silicate source consists of minerals such as opal, which
reacts rapidly with an aqueous alkaline solution. Calcination
of Shahindej pozzolan increases the slump and results in a
ACI Materials Journal/May-June 2013
more workable geopolymer concrete. Pozzolans with higher
silicate content and aluminate may delay the setting time of
AANP, which decreases when the curing temperature and
dosage of alkaline hydroxide increases.12
ENGINEERING PROPERTIES OF AANP CONCRETE
There are many different views regarding the main parame-
ters that affect the compressive strength and other mechanical
properties of geopolymer concrete. Palomo et al.24,25 stated
that the significant factors affecting the compressive strength
are the type of alkaline activator, curing temperature, and
curing time. However, other researchers11,26,27 have reported
that the important parameters for satisfactory polymeriza-
tion are the relative amounts of Si, Al, K, Na, molar ratio
of Si to Al present in the solution, the ratio of alumina
silicate mineral to kaolinite (when kaolinite is added), the
type of alkaline activator, the water content, and the curing
temperature.The presence of silicate ions in the alkaline
solution substantially improves the mechanical strength
and modulus of elasticity values but has a slightly adverse
effect on the otherwise very strong matrix/aggregate and
matrix/steel bond.28 Experimental results show that the H2O/
M2O molar ratio in the mixture composition significantly
affects the compressive strength of geopolymer concrete,
whereas the influence of the Na2O/SiO2 molar ratio is less
significant.21-23 An increase of the H2O/M2O molar ratio and
water-to-geopolymer solids ratio decreases the compressive
strength of geopolymer.21-23
Geopolymeric concrete mixtures based on AANPs have
mostly shown lower strength and modulus of elasticity than
OPC mixtures at early ages, but they reach the same and
even higher strength and modulus of elasticity than OPC
mixtures after long-term curing (Fig. 3). It is concluded that
concrete made with an AANP develops moderate-to-high
mechanical strength and modulus of elasticity and shrinks
much less than ordinary OPC. The drying shrinkage strains
of AANP concrete are extremely small and one explanation
for this behavior may be found in the microstructural char-
acteristic of the new binder which, in the alkali activation
of pozzolans, can form a zeolite-type phase. Zeolite proper-
ties and microstructure are widely known to be unaffected
by the loss of the water incorporated during their synthesis
because water loss is not only reversible in most zeolites but
they are also able to absorb water from the humidity in the
atmosphere.28 The total w/b also has a significant effect on
the shrinkage properties of geopolymer concretes but seems
to be contrary to the behavior of OPC concrete, where lower
drying shrinkage resulted from higher w/b. For the lower
w/b, the cross linking may not be as efficient, as it must occur
within a restricted space, so it may well not be completed. In
concrete made from AANP, the higher the curing tempera-
ture, the lower the amount of drying shrinkage measured.
This is possibly because further cross linking is made as
a result of water removal by the higher temperature and
follows the pattern reported by Wallah and Rangan.29 All
of the AANP concrete mixtures show lower ultrasonic pulse
velocity (UPV) than OPC concrete mixtures, even though
they have higher compressive strengths despite lower densi-
ties. In this investigation, an attempt has been made to
formulate the equations to describe the relationship between
splitting tensile strength, static modulus of elasticity, and
UPV of AANP concrete and its compressive strength.12,17
ACI Materials Journal/May-June 2013
Fig. 3—Comparison of compressive strength of different
investigated AANP concrete mixtures and OPC concrete
control mixtures. (Note: 1 MPa = 0.145 ksi.)
DURABILITY PROPERTIES OF AANP CONCRETE
This type of concrete has shown that oxygen permeability
lies in the same ranges as OPC concrete. The oxygen perme-
ability in the AANP concretes reduced as the w/b reduced.
This effect was 20% greater in the geopolymer concrete than
in OPC mixtures cured at the sealed condition30 because
excess free water in AANP concretes gives rise to a more
open microstructure with higher permeability than OPC
concrete. Curing at higher temperature decreases the perme-
ability of the AANP geopolymer concrete. This is possibly
because the water in the geopolymer structure acts as a
medium that promotes geopolymerization, with further cross-
linking (Si-O-Al or Si-O-Si bonding) occurring as a result of
water removal at higher temperature, resulting in pore space
blockage and a stronger structure with lower shrinkage.
The rapid chloride permeability test (ASTM C1202)31 was
not found to be applicable for geopolymer concrete because
of the conductive pore solution that exists due to the presence
of alkaline activators in the pore solution. A reduced voltage
was used and the results of long-term chloride absorption
shows that the resistance of AANP concrete, when subjected
to chloride attack, was moderate. The long-term chloride
ponding results indicate that chloride ion penetrability of the
geopolymer concrete is moderate to high.30
On exposure to sulfate solution, while the compressive
strength development does not show significant reduction
in most samples, XRD traces show that sulfate compounds
were observed in the core of AANP concrete; thus, further
study is needed to confirm the resistance of AANP concrete
to sulfate attack for longer periods of time. A general rela-
tionship was found to predict the oxygen permeability of
AANP concrete from its compressive strength (Fig. 4).
EVALUATION OF CARBON FOOTPRINT AND
COST FOR AANP CONCRETE
Two potential advantages of concrete made with AANPs
compared with other binders are its carbon footprint and
cost. Increased pressure to improve sustainability within
the concrete industry makes these factors very important.
The relation between the CO2 footprint and cost of AANP
concrete and its compositions in comparison with portland-
based cements is roughly quantified.
ENVIRONMENTAL BENEFITS
Decarbonation of lime and calcination of cement clinker
release CO2 as a reaction product in OPC concrete, although
the use of an alkaline hydroxide or silicate as activating
solution rather than water for cement hydration does diffuse
some CO2. Production of these activators needs tempera-
335

Fig. 4—Comparison of oxygen permeability of different
AANP concrete mixtures and OPC concrete control
mixtures. (Note: 1 m2 = 10.76 ft2; 1 MPa = 0.145 ksi.)
ture similar to decarbonation of lime in OPC manufacture.
The CO2 emission of AANP concrete can be quantified in
terms of its compositions. Referring to the research done
by the authors,12 242.5 lb (110 kg) of activator is needed
to be mixed with 881.85 lb (400 kg) pozzolan to produce
1.31 yd3 (1 m3) of AANP concrete, which has the CO2 emis-
sion equal to 27.5% of the same amount of OPC when
pozzolan is used in its natural state (it is estimated that the
production of 2000 lb [1 ton] of OPC results in the release
of 2000 lb [1 ton] of CO2). If the calcined form is used, the
CO2 emission of AANP concrete would be the summation
of CO2 emission due to producing the required activators
and the amount related to the calcination procedure. Because
the temperature required for the calcination of these materials
is half of that needed to produce OPC, the CO2 emission for
calcinations of these materials can be considered 50% of equal
OPC production. Therefore, in this case, the CO2 emission of
AANP concrete increases to 77.5% of the amount emitted by the
same weight of OPC. Hence, the AANP concrete manufacture
is liable to reduce CO2 emission from 22.5 to 72.5%, compared
to OPC production.
SUPPLY AND COST OF ACTIVATORS
Referring to the research done by the authors,12 145.5 lb
(66 kg) of potassium hydroxide and 101.41 lb (46 kg)
(considering density equal to 2.28 lb/yd3 [1.35 kg/m3])
of water glass as activator is needed to be mixed with
881.85 lb (400 kg) of pozzolan to produce 1.31 yd3
(1 m3) of AANP concrete. Although these activators
have different prices in different markets, the cost of
industrial potassium hydroxide (55.12 lb [25 kg] KOH
flake made in Korea costs £52 [$81.20]) and water-glass
(25 lb [11.35 kg] water-glass solution is sold in the UK
for £7.69 [$12]) is considered equal to £2.1 ($3.28) and
£0.73 ($1.14) per kg (lb), respectively. Hence, calculating
as a rule of thumb the activators required for activating
881.85 lb (400 kg) pozzolans costs £173 ($269.89) and
considering 55.12 lb (25 kg) of pozzolan to be compa-
rable to the price of OPC in 55.12 lb (25 kg) bags, the
price of required activators would be £11 ($17.16) per
55.12 lb (25 kg).
APPLICATION ASPECTS OF AANP CONCRETE
Geopolymer concrete made from AANP is a new type of
concrete that needs a very good control on-site to be used
as a comparable alternative to OPC concrete. However, its
drawbacks, such as loss of workability, quick setting time,
336
and the health and safety implications of working with strong
alkali solutions, can easily be adapted in applications such as
precast concrete and mass concretes as in dam construction,
where roller-compacted AANP concrete may be a viable
construction method. This type of concrete, especially in
countries with greater resources of natural pozzolan, can help
decrease energy consumption and environmental impacts.
CONCLUSIONS
1. AANP concretes develop moderate-to-high mechanical
strength with a high modulus of elasticity and a shrinkage
much lower than with OPC.
2. The results show that compared to OPC concrete,
AANP concrete has oxygen permeability in the same
ranges as OPC concrete but it shows moderate-to-high
chloride ion penetrability.
3. The AANP concrete manufacture is liable to
reduce CO2 emission from 22.5 to 72.5% compared to
OPC production.
4. Geopolymer concrete can be produced with the same
cost of OPC concrete and comparable properties.
ACKNOWLEDGMENTS
This research described has been led by the Department of Civil and
Structural Engineering, University of Sheffield, Sheffield, UK, with experi-
mental work conducted in the concrete technology laboratory in the Civil
Department of PWUT, Tehran, Iran. The authors express their gratitude to
the Research Centre of Natural Disasters in Industry (RCNDI) in PWUT
for its support rendered throughout the research program. X-ray fluoresence
(XRF) analysis and X-ray diffraction (XRD) was detected in the Kansaran
Binaloud X-ray laboratory in Tehran, Iran.
REFERENCES
1. Vazinram, F., and Khodaparast, M. M., “Concrete and Renovation of
Consumption Pattern Considering Environmental Impacts,” Third National
Conference on Operation and Maintenance of Water and Waste Water,
Tehran, Iran, 2009, 10 pp.
2. Davidovits, J., “Geopolymers, Man-Made Rock Geosynthesis and the
Resulting Development of Very Early High Strength Cement,” Journal of
Materials Education, V. 16, No. 2-3, 1994, pp. 91-139.
3. Davidovits, J., “High-Alkali Cements for 21st Century Concretes in
Concrete Technology, Past, Present and Future,” Concrete Technology:
Past, Present, and Future—Proceedings of V. Mohan Malhotra Sympo-
sium, SP-144, P. K. Mehta, ed., American Concrete Institute, Farmington
Hills, MI, 1994, pp. 383-397.
4. Davidovits, J., “Properties of Geopolymer Cements,” First Interna-
tional Conference on Alkaline Cements and Concretes, SRIBM, Kiev State
Technical University, Kiev, Ukraine, 1994, pp. 131-149.
5. Davidovits, J.; Davidovits, M.; and Davidovits, N., “Process for
Obtaining a Geopolymeric Alumino-Silicate and Products Thus Obtained,”
U.S. Patent No. 5,342,595, 1994.
6. Barbosa, V. F. F.; MacKenzie, K. J. D.; and Thaumaturgo, C.,
“Synthesis and Characterisation of Materials Based on Inorganic Poly-
mers of Alumina and Silica: Sodium Polysialate Polymers,” International
Journal of Inorganic Materials, V. 2, 2000, pp. 309-317.
7. Bilodeau, A., and Malhotra, V. M., “High-Volume Fly Ash System:
Concrete Solution for Sustainable Development,” ACI Materials Journal,
V. 97, No. 1, Jan.-Feb. 2000, pp. 41-48.
8. Van Jaarsveld, J. G. S., and Van Deventer, J. S. J., “Effect of Alkali
Metal Activator on the Properties of Fly Ash Based Geopolymers,” Indus-
trial & Engineering Chemistry Research, V. 38, 1999, pp. 3932-3941.
9. Van Jaarsveld, J. G. S.; Van Deventer, J. S. J.; and Schwartzman, A.,
“The Potential Use of Geopolymeric Materials to Immobilise Toxic Metals:
Part II: Material and Leaching Characteristics,” Minerals Engineering,
V. 12, 1999, pp. 75-91.
10. Van Jaarsveld, J. G. S.; Van Deventer, J. S. J.; and Schwartzman, A.,
“The Potential Use of Geopolymeric Materials to Immobilise Toxic Metals:
Part II: Material and Leaching Characteristics,” Minerals Engineering,
V. 12, No. 1, 1999, pp. 75-91.
11. Xu, H., and Van Deventer, J. S. J., “The Geopolymerisation of Alumi-
nosilicate Minerals,” International Journal of Mineral Processing, V. 59,
2000, pp. 247-266.
ACI Materials Journal/May-June 2013
 12. Bondar, D., “Alkali Activation of Iranian Natural Pozzolans for
Producing Geopolymer Cement and Concrete,” PhD dissertation, Univer-
sity of Sheffield, Sheffield, UK, 2009, 300 pp.
13. Bondar, D.; Lynsdale, C. J.; Milestone, N. B.; Hassani, N.; and
Ramezanianpour, A., “Effect of Heat Treatment on Reactivity-Strength
of Alkali-Activated Natural Pozzolans,” Construction Building Material
Journal, V. 25, No. 10, 2011, pp. 4065-4071.
14. Bondar, D.; Lynsdale, C. J.; and Milestone, N. B., “A Simplified
Model for Prediction of Compressive Strength of Alkali-Activated Natural
Pozzolans,” Journal of Materials in Civil Engineering, V. 24, No. 4, 2012,
pp. 391-401.
15. Bondar, D.; Lynsdale, C. J.; Milestone, N. B.; Hassani, N.; and
Ramezanianpour, A. A., “Effect of Type, Form, and Dosage of Activators
on Strength of Alkali-Activated Natural Pozzolans,” Cement and Concrete
Composite Journal, V. 33, No. 1, 2011, pp. 251-260.
16. Bondar, D.; Lynsdale, C. J.; Milestone, N. B.; Hassani, N.; and
Ramezanianpour, A. A., “Effect of Adding Mineral Additives to Alkali-
Activated Natural Pozzolan Paste,” Construction Building Material
Journal, V. 25, No. 6, 2011, pp. 2906-2910.
17. Bondar, D.; Lynsdale, C. J.; Milestone, N. B.; Hassani, N.; and
Ramezanianpour, A. A., “Engineering Properties of Alkali Activated
Natural Pozzolan Concrete,” ACI Materials Journal, V. 108, No. 1, Jan.-
Feb. 2011, pp. 64-72.
18. Van Jaarsveld, J. G. S.; Van Deventer, J. S. J.; and Lorenzen, L.,
“The Potential Use of Geopolymeric Materials to Immobilise Toxic Metals:
Part I: Theory and Applications,” Mineral Engineering, V. 10, No. 7, 1997,
pp. 659-669.
19. Cheng, T. W., and Chiu, J. P., “Fire-Resistant Geopolymer Produced
by Granulated Blast Furnace Slag,” Minerals Engineering, V. 16, 2003,
pp. 205-210.
20. Hardjito, D.; Wallah, S. E.; Sumajouw, D. M. J.; and Rangan, B. V.,
“Geopolymer Concrete: Turn Waste into Environmentally Friendly
Concrete,” International Conference on Recent Trends in Concrete Tech-
nology and Structures (INCONTEST), Kumaraguru College of Technology,
Coimbatore, India, 2003.
21. Hardjito, D.; Wallah, S. E.; Sumajouw, D. M. J.; and Rangan, B. V.,
“On the Development of Fly Ash-Based Geopolymer Concrete,” ACI
Materials Journal, V. 101, No. 6, Nov.-Dec. 2004, pp. 467-472.
ACI Materials Journal/May-June 2013
22. Hardjito, D.; Wallah, S. E.; Sumajouw, D. M. J.; and Rangan, B. V.,
“The Stress-Strain Behaviour of Fly Ash-Based Geopolymer Concrete,”
ACMSM 18, A.A. Balkema Publishers, Perth, Australia, 2004, pp. 129-140.
23. Hardjito, D.; Wallah, S. E.; Sumajouw, D. M. J.; and Rangan, B.V.,
“Factors Influencing the Compressive Strength Fly-Ash Based Geopolymer
Concrete,” Jurusan Teknik Sipil, Fakultas Teknik Sipil dan Perencanaan –
Universitas Kristen Petra, V. 6, No. 2, 2004, pp. 88-96.
24. Palomo, A.; Blanco-Varela, M. T.; Granizo, M. L.; Puertas, F.;
Vazquez, T.; and Grutzeck, M. W., “Chemical Stability of Cementitious
Materials Based on Metakaolin,” Cement and Concrete Research, V. 29,
1999, pp. 997-1004.
25. Palomo, A.; Grutzeck, M. W.; and Blanco, M. T., “Alkali-Activated
Fly Ashes, A Cement for the Future,” Cement and Concrete Research,
V. 29, 1999, pp. 1323-1329.
26. Barbosa, V. F. F.; MacKenzie, K. J. D.; and Thaumaturgo, C.,
“Synthesis and Characterisation of Materials Based on Inorganic Poly-
mers of Alumina and Silica: Sodium Polysialate Polymers,” International
Journal of Inorganic Materials, V. 2, 2000, pp. 309-317.
27. Rowles, M., and O’Connor, B., “Chemical Optimisation of the
Compressive Strength of Aluminosilicate Geopolymers Synthesised by
Sodium Silicate Activation of Metakaolinite,” Journal of Materials Chem-
istry, V. 13, 2003, pp. 1161-1165.
28. Fernandez-Jimenez, A. M.; Palomo, A.; and Lopez-Hombrados, C.,
“Engineering Properties of Alkali Activated Fly Ash Concrete,” ACI
Materials Journal, V. 103, No. 2, Mar.-Apr. 2006, pp. 106-112.
29. Wallah, S. E., and Rangan, B. V., “Low-Calcium Fly Ash-Based
Geopolymer Concrete: Long-Term Properties,” Research Report GC2,
Faculty of Engineering, Curtin Technology, Perth, Australia, 2006,
pp. 52-58.
30. Bondar, D.; Lynsdale, C. J.; Milestone, N. B.; and Hassani, N.,
“Oxygen and Chloride Permeability of Alkali Activated Natural Pozzolan
Concrete,” ACI Materials Journal, V. 109, No. 1, Jan.-Feb. 2012, pp. 53-62.
31. ASTM C1202-97, “Standard Test Method for Electrical Indication
of Concrete’s Ability to Resist Chloride Ion Penetration,” ASTM Interna-
tional, West Conshohocken, PA, 1997, 6 pp.
337
 One Click... One Entire Journal

Introducing
a new feature on ACI’s website!
VOL. 108, NO. 6 VOL. 108, NO. 6
NOVEMBER-DECEMBER 2011 NOVEMBER-DECEMBER 2011

ACI ACI
MATERIALS STRUCTURAL J O U R N A L
J O U R N A L

A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE A JOURNAL OF THE AMERICAN CONCRETE INSTITUTE

215 Long-Term Behavior of Cracked Steel Fiber-Reinforced 215 Long-Term Behavior of Cracked Steel Fiber-Reinforced
Concrete Beams under Sustained Loading Concrete Beams under Sustained Loading

Now, with one click of the mouse, subscribers of the

ACI Structural and Materials Journals digital editions can
download an Adobe Acrobat PDF of an unabridged
issue at www.concrete.org/pubs/electronicjournals/
esjindexnew.asp and ww.concrete.org/pubs/
electronicjournals/emjindexnew.asp.
Digital Download Journals Full Page.indd 1 5/2/2013 2:56:06 PM
 ACI MATERIALS JOURNAL
Title no. 110-M30
Impact Resistance of Blast Mitigation Material Using
Modified ACI Drop-Weight Impact Test
by John J. Myers and Matthew Tinsley
This experimental program was conducted to evaluate the effec-
tiveness of a high-volume fly ash-wood fiber material at resisting
repeated low-velocity drop-weight impact. A modified version of
the ACI drop-weight test proposed by ACI Committee 544 was
conducted on seven groups of specimens. Each group was unique,
consisting of various quantities of materials, including cement, fly
ash, and wood fibers. Testing procedures for the drop-weight test
were consistent with test procedures listed in a previous research
study on drop-weight impact testing of polypropylene fiber-rein-
forced concrete. However, a few modifications were made to the
previous test setup to better accommodate this study. These modifi-
cations are discussed herein.
Test results revealed that the fly ash-wood fiber specimens have
significant resistance to drop-weight impact as compared to
unreinforced concrete specimens. Although results show that poly-
propylene fiber-reinforced concrete has the highest drop-weight
impact resistance, the fly ash-wood fiber material is a likely candi-
date for use in a barrier system due to its low stiffness and ductile
failure mode. Impact test results also coincided with results from
previous drop-weight impact tests.
Keywords: ACI drop test; blast mitigation material; high-volume fly ash;
impact resistance; wood fiber.
INTRODUCTION
Due to the lack of a standard impact test for concrete,
designers are not yet able to use impact resistance as a
design parameter. The lack of a standard impact test is due
to high variations in the results. However, ACI 544.2R-89
(ACI Committee 544 1989) proposed a drop-weight impact
test that is becoming a more common method to determine
the relative brittleness and impact resistance of fiber-rein-
forced concrete (FRC). This study uses a modified version of
the ACI 544 drop-weight test proposed by Badr and Ashour
in 2005. The results produced from their modified version of
the ACI test proved to be more statistically accurate. There-
fore, the procedures documented by Badr and Ashour (2005)
were followed closely to determine the relative impact
resistances of several types of concretes, including a high-
volume fly ash-wood fiber concrete (FA-WF). Five distinctly
different mixture designs were investigated in this study, and
impact tests were conducted at 7, 14, and 28 days of age.
Two materials used in this study have environmental
benefits. For example, fly ash is a by-product of the combus-
tion of pulverized coal. Its pozzolanic characteristics make it
a beneficial ingredient in concrete and its use has increased
with increasing energy costs in cement production. Another
main ingredient in several mixture designs was wood fiber.
The United States is currently facing the serious challenge of
disposing of wood-based waste in many landfills throughout
the country that are already near capacity (Joshi and Myers
2005). The recycled wood waste material used in this study
was milled oak wood fibers, which were milled from old
loading pallets.
ACI Materials Journal/May-June 2013
TECHNICAL PAPER
RESEARCH SIGNIFICANCE
This research effort investigates the effectiveness of a
high-volume FA-WF material at absorbing energy from
repeated low-velocity impact. The alternative construction
material developed at Missouri S&T can be used as an alter-
native masonry product (Joshi and Myers 2005) or as a sacri-
ficial energy-absorbing material for blast mitigation (Carey
and Myers 2011). The study also investigates the statistical
benefits of using the modified ACI 544 drop-weight test
(Badr and Ashour 2005) with slight modifications rather
than using the current method proposed by ACI 544.2R-89.
EXPERIMENTAL INVESTIGATION
This experimental program was conducted to evaluate the
effectiveness of a high-volume FA-WF material at resisting
repeated low-velocity drop-weight impact. No experimenta-
tion on the impact resistance of this material has taken place
in the past. However, numerous studies have been carried out
to analyze the drop-weight impact resistance of other types
of FRCs (Badr and Ashour 2005; Nataraja et al. 1999).
Drop-weight impact test proposed by ACI 544.2R-89
The design of FRC elements relies heavily on the statis-
tical consideration of their properties because of the signifi-
cant influence of the fabrication method, sampling, and
testing. The variations in mechanical properties should be
taken into account in deciding the minimum number of tests
required to determine the properties when testing specimens
of the same mixture design. The observed coefficient of vari-
ation (COV) in impact resistance is often greater than 50%.
Therefore, many specimens are needed to develop reliable
conclusions about the normality of the distribution of impact
resistance test results (Nataraja et al. 1999).
Several types of tests have been used to measure the
impact resistance of FRC, and the repeated drop-weight
impact test is the simplest. This test yields the number of
blows necessary to cause prescribed levels of damage in the
test specimen. The number of blows recorded serves as a
qualitative estimate of the energy absorbed by the specimen
at the levels of distress specified. Therefore, the relative
impact resistances of various materials can easily be estab-
lished using the test. The impact test equipment and proce-
dure are outlined by ACI 544.2R-89.
The current ACI 544.2R-89 drop-weight test is carried out
by dropping a 4.54 kg (10 lb) compaction hammer from a
ACI Materials Journal, V. 110, No. 3, May-June 2013.
MS No. M-2012-057 received February 16, 2012, and reviewed under Institute
publication policies. Copyright © 2013, American Concrete Institute. All rights
reserved, including the making of copies unless permission is obtained from the
copyright proprietors. Pertinent discussion including author’s closure, if any, will be
published in the March-April 2014 ACI Materials Journal if the discussion is received
by December 1, 2013.
339
John J. Myers, FACI, is an Associate Professor at Missouri University of Science and
Technology, Rolla, MO. He received his BAE from the Pennsylvania State University,
University Park, PA, and his MS and PhD from the University of Texas at Austin,
Austin, TX. He is Chair of ACI Committee 363, High-Strength Concrete, and ACI
Subcommittee 440-L, FRP-Durability. He is also a member of the Education Activi-
ties Committee and the Student and Young Professional Activities Committee, as well
as ACI Committees S801, Student Activities; S802, Teaching Methods and Educa-
tional Materials; S803, Faculty Network Coordinating Committee; S804, Student and
Young Professional Awards; 201, Durability of Concrete; 342, Evaluation of Concrete
Bridges and Bridge Elements; 440, Fiber-Reinforced Polymer Reinforcement; and
Joint ACI-ASCE Committee 423, Prestressed Concrete. His research interests include
advanced concrete materials for structural applications, fiber-reinforced polymers in
structural repair, and strengthening applications and sustainable materials.
ACI member Matthew Tinsley is a former Graduate Research Assistant at Missouri
University of Science and Technology. He received his BS from Arkansas State Univer-
sity, Jonesboro, AR, in 2005 and his MS from Missouri University of Science and
Technology in 2007. His research interests include material development for improved
blast resistance and reduced fragmentation.
height of 457 mm (18 in.) on a hardened steel ball with a
diameter of 63.5 mm (2.5 in.). The steel ball, which is centered
on top of a concrete specimen that is 63.5 mm (2.5 in.) thick
and 152 mm (6 in.) in diameter, transfers a single-point
impact to the surface of the specimen. The number of blows
to first crack (FC) and ultimate resistance (UR) are recorded.
ACI 544.2R-89 recommends that specimens be tested
at 7 and 28 days of age. ACI 544.2R-89 reports that results of
these tests exhibit a high variability and may vary consider-
ably with different types of mixtures and fiber contents.
Modified ACI 544 drop-weight impact test for
concrete (Badr and Ashour 2005)
A modified ACI 544 drop-weight impact test for concrete
was proposed by Badr and Ashour (2005) in an effort to reduce
the large variations in the existing ACI 544.2R-89 drop-
weight impact test. There are several sources of error in the
current ACI 544.2R-89 drop-weight test. First, the likeli-
hood of erroneous results is increased by using a single point
of impact because the impact is localized on such a small
amount of material. Second, the subjectivity of the test is
increased when cracks are allowed to occur at any location
on the specimen. Furthermore, by declaring failure when the
pieces of the specimen touch the lugs, more impacts may be
applied than necessary to break the specimen. Lastly, speci-
mens must all be prepared in the same manner (Badr and
Ashour 2005). Several changes were proposed by Badr and
Ashour (2005) to reduce the amount of variation in the results
of the drop-weight impact test. These include the following:
1. Notch the specimens to force cracks along a predefined
path;
2. Use a 50 mm (2 in.) line of impact from a steel bar
rather than a single-point impact;
3. Ensure that the specimens have similar faces whether
cut or troweled;
4. Declare ultimate failure when the specimen is separated
completely into halves;
5. Only accept specimens that crack through the line of
impact and reject any other failure locations; and
6. Ensure that the specimen thickness is reduced from
63.5 to 50 mm (2.5 to 2 in.) to reduce testing time.
Test setup
The test setup and procedures follow those proposed
by Badr and Ashour (2005) with a few minor differences.
All specimens were finished using trowels, and the impact
hammer was manual so it could be easily carried to any location.
340
Specimen molds were constructed from polyvinyl chloride
(PVC) pipe 6.4 mm (0.25 in.) thick with an inner diameter
of 150 mm (5.9 in.). The pipe was cut into lengths slightly
over 51 mm (2 in.) to create 54 round molds. Each mold was
reduced to 50.8 ± 1.5 mm (2 ± 0.06 in.) using a lathe. Once
cut to the proper thickness, a small hole was drilled through
opposite sides of each mold using a drill press. Then, each
mold was cut down its length so specimens could be easily
extracted once hardened. Two triangular wood pieces were
screwed to the inside of each PVC mold to create notches in
the drop-weight specimens.
Instead of using an automated soil compactor, in an effort
to reduce the overall cost of the device, a manual device was
fabricated to conduct the tests. The test fixture consisted
of three steel plates, four steel angle sections, a handheld
4.54 kg (10 lb) modified proctor hammer, and an impact
piston. Both the top and bottom plates were 254 x 229 x
6.4 mm (10 x 9 x 0.25 in.). The middle plate was 254 x 149 x
6.4 mm (10 x 5.9 x 0.25 in.) but had two 50 mm (2 in.) diam-
eter holes cut into it, along with a center hole of 63.5 mm
(2.5 in.) for the impact piston. Additionally, two partial
holes were cut in the middle plate above the steel notch loca-
tions in the bottom plate to allow for visual inspection of
the test specimen (refer to Fig. 1(b) and A1), and the impact
piston was placed in the center hole, preventing the piston
from rotating or moving laterally. The steel notches in the
bottom plate were intended to prevent the specimen from
moving during repeated impact testing. Figure 1 shows the
test fixture and impact piston. The weight of the piston was
0.85 kg (1.87 lb), and the weight of the entire test fixture was
20.4 kg (45 lb).
Mixture designs
Materials used in the preparation of the drop-weight speci-
mens include Class C fly ash, Type I portland cement, wood
fibers, limestone, sand, high-range water reducers (HRWRs),
and potable water. ASTM Class C fly ash was required to
fabricate two of the five groups of drop-weight specimens
investigated in the study. The material properties of the fly
ash are listed in Table 1. Commercial-grade Type I portland
cement was used to fabricate all five groups of drop-weight
specimens. The portland cement used for this study complied
with ASTM C150-05 specifications.
The wood fiber used in this study was recycled by milling
old oak wooden loading pallets in an effort to enhance the
sustainability aspects of the FA-WF. Previous studies by
Joshi and Myers (2005) indicated that wood fibers from recy-
cled oak wood pallets performed similarly to new growth
oak limbs used from recent forestry cuts. Initial work under-
taken in the development of the FA-WF material empha-
sized reuse of wood products that are traditionally discarded
during forestry cuts (that is, upper limbs) or disposed of (that
is, damaged wood pallets) after use in landfills. The grada-
tion of the milled wood fibers ranged from a maximum size
of 12.7 mm (0.5 in.) to fine particles of dust. The majority
of the fibers were approximately 6.4 mm (0.25 in.) in length.
In every group that wood fibers were used, the fibers were
added to the other contents of the mixture in a fairly dry
state with a moisture content of roughly 10%, determined in
accordance with ASTM C566-97(2004). The dry density of
the compacted wood fibers was determined to be 275.5 kg/m3
(17.2 lb/ft3).
Gasconade dolomitic limestone was also used. The lime-
stone had a nominal maximum size of 12.7 mm (0.5 in.) and a
ACI Materials Journal/May-June 2013
Fig. 1—Impact test fixture.
Table 1—Chemical properties of Class C fly ash Table 2—Properties of polypropylene fibers
Parameters Results Fiber parameters
Sulfur in ash, % weight 0.7 Fiber type Homopolymer polypropylene
Sulfur in ash (SO3), % weight 1.75 Fiber length 19 mm
Fineness (325 sieve), % weight (passing) 81.9 Color White/clear
Moisture, % weight 0.03 Ultimate elongation 20% at 21°C
Loss of ignition (LOI), % weight 0.17 Melting point 165°C
Density, g/mL 2.73 Ignition point 594°C
Specific gravity 2.74 Specific gravity 0.91
Titanium oxide as TiO2, % weight 1.23 Tensile strength 379.2 MPa
Aluminum oxide as Al2O3, % weight 17.06 Modulus of elasticity 4.14 GPa
Notes: 1 mm = 0.0394 in.; 1 MPa = 145 psi; 1 GPa = 145 ksi; °C = (°F – 32)/1.8.
Calcium Oxide as CaO, % weight 26.12
Iron oxide as Fe2O3, % weight 6.31
did not exceed 1% in any of the groups. Furthermore, an
Potassium oxide as K2O, % weight 0.39
HRWR was added in various quantities to each mixture as
Magnesium oxide as MgO, % weight 5.77 needed. The HRWR used meets ASTM C494/C494M-05a
Sodium oxide as Na2O, % weight 1.87 requirements for Type A and Type F admixtures. The
HRWR ratio for Group 1 was selected to meet workability
Silicon oxide as SiO2, % weight 35.43
and placement requirements for the FA-C-WF material. The
Phosphorus oxide as P2O5, % weight 0.90 HRWR ratio for the concrete series Groups 2 through 5 was
Slagging index 0.34 kept consistent.
Collated polypropylene fibrillated fibers were used in
Slagging classification Low the Group 3 mixture. The fiber is made to mechanically
Fouling index 1.41 bond with the concrete matrix, thus enhancing the impact
Fouling classification Severe
resistance and durability of the concrete. The polypro-
pylene fibers used in this study comply with ASTM C1116/
Available alkali, %Na2O, equivalent 1.59 C1116M-06 specifications, as well as with other national
Note: 1 g/mL = 0.06506 lb/oz. building codes. The mechanical properties of the fibers are
listed in Table 2. Five groups of specimens were evaluated
bulk specific gravity of 2.70 in an oven-dried condition. The for impact resistance using the modified ACI drop-weight
absorption of the aggregate was measured to be 1.2%. The test. The mixture proportions for each group are shown in
moisture content of the limestone in each group remained Table 3.
below 0.1% during the mixing processes. The sand used in The first group of drop-weight specimens, Group 1, was
the mixture designs had a bulk specific gravity of 2.54 in an fabricated from Class C fly ash, Type I portland cement, wood
oven-dried condition, and the absorption was calculated to fibers, and potable water. This mixture was designated as the
be approximately 1.4%. The moisture content of the sand FA-C-WF mixture. A total of 0.085 m3 (3 ft3) of material

ACI Materials Journal/May-June 2013 341

Table 3—Mixture proportions by weight, kg/m3
Group 1, Group 2, Group 3, Group 4, Group 5,
Component FA-C-WF concrete PFRC WF concrete concrete layer
Class C fly ash 462.9 — — — —
Type I portland cement 278.7 410.1 410.1 451.7 410.1
White oak wood fibers 112.1 — — 136.2 —
12.5 mm limestone — 999.6 999.6 808.9 999.6
Sand — 799.3 799.3 647.1 799.3
Water 281.9 144.2 144.2 136.2 144.2
HRWR, mL/kg 2.6 5.2 5.2 5.2 5.2
Polypropylene fibers — — 3.20 — —
Water-cement ratio 0.38 0.35 0.35 0.30 0.35
Wood fiber by weight, % 10.0 0.0 0.0 6.2 0.0
Slump, mm 76 178 127 51 153
Notes: 1 kg/m3 = 62.43 lb/ft3; 1 mm = 0.0394 in.; 1 mL/kg = 0.01543 oz/lb.

was needed to fill all 54 forms and produce enough cylinders
for compression, split cylinder, and modulus tests. Because
the material was fairly difficult to mix, the total volume was
broken into two separate batches of 0.0425 m3 (1.5 ft3). A
conventional rotary drum mixer was used for mixing. First,
the mixer was washed to dampen the inside of the drum.
Then, all of the fly ash and cement and half of the water
were added to the drum and allowed to mix for 1 minute.
The HRWR was added to the other half of the water. After
approximately 1 minute of mixing, the inside of the drum
was scraped to release any material stuck to the sides or back
of the drum. Then, half of the remaining water was added
to the mixer, along with the wood fibers. The fibers were
added slowly to prevent balling of the material. Finally, the
remaining portion of water was added to the mixer. Mixing
continued for another 5 minutes after all the materials were
added to the drum.
Once a uniform distribution of wood fibers was achieved,
the material was placed into the molds. After the molds were
filled, each one was vibrated to ensure there were no air
voids or segregation in the specimens. A portion of the mate-
rial was used for a slump test to check workability according
to ASTM C143/C143M-05. The slump was determined to
be 75 mm (2.95 in.). The specimens were finished using
trowels and then covered with a sheet of plastic. Compres-
sion, split cylinder, and modulus specimens were made in
compliance with ASTM C192/C192M-06. After 24 hours,
the specimens were removed from the molds and placed in
a moist cure room until the time of testing. Specimens were
tested at 7, 14, and 28 days. Eighteen specimens were tested
at each test age.
Mixing the plain concrete and polypropylene fiber-
reinforced concrete (PFRC) was simple. First, the coarse
aggregate, sand, and cement were mixed in a dry state for
approximately 1 minute before adding roughly half of the
mixing water. After 3 minutes of mixing, the remaining
portion of water was added along with the HRWR. Mixing
continued for 5 minutes to achieve a homogeneous mixture.
In the case of the PFRC, the fibers were added after all
the other components were allowed to mix for 2 minutes.
The slump values for the plain concrete and PFRC
were 178 and 127 mm (7 and 5 in.), respectively.
The specimens in Group 4 were batched using a normal-
weight concrete mixture with the addition of wood fibers.
The wood fibers were added in hopes of increasing the
impact absorption capabilities of the mixture. The mixing
procedures for Group 4 were the same as those used for the
previous groups. The slump was determined to be 51 mm
(2 in.).
Group 5 was a layered construction of Groups 1 and 2. A
plain concrete mixture was batched first and allowed to cure
for 24 hours. The molds were only filled halfway (approxi-
mately 25.4 mm [1 in.]). The following day, the FA-C-WF
mixture used for Group 1 was batched to fill the remaining
portion of the molds. Therefore, the specimens had a
25.4 mm (1 in.) plain concrete base and a 25.4 mm (1 in.)
FA-C-WF layer set on top. Each layer was mixed according
to the procedures mentioned previously for Groups 1 and
2. The bottom layer of plain concrete was left in a rough-
ened state to provide some amount of shear interlock
between the two layers. Specimens were removed from the
molds 24 hours after the top layer was placed and then put
in a moist cure room until testing age. Slump values for the
FA-C-WF layer and the concrete layer were 64 and 153 mm
(2.5 and 6 in.), respectively.
Mechanical properties
Several mechanical properties were determined for each
group of specimens. These properties include the material’s
compressive strength, indirect tensile strength, and modulus
of elasticity (MOE).
The compressive strength of each group was determined
using either a standard cylinder size of 150 x 300 mm (6 x 12 in.)
or 100 x 200 mm (4 x 8 in.). The specimens were prepared in
the lab according to ASTM C192/C192M-06 specifications.
The specimens were removed from the molds after 24 hours
and placed in a moist cure room until testing age. Before
testing, specimens were capped at the two bases to ensure
parallel surfaces according to ASTM C617-98 specifica-
tions. The cylinders were tested according to ASTM C39/
C39M-05. A load rate of 4.5 kN/s (1000 lb/s) was used for
the 150 mm (6 in.) diameter specimens, while the 100 mm
(4 in.) diameter specimens were loaded at a rate of approxi-
mately 2.25 kN/s (500 lb/s). A compression testing machine
was used to carry out the compression tests. The compressive
strengths for all five groups are shown in Table 4. Group 3 had
the highest compressive strength, with an average 28-day
strength of 61.3 MPa (8890 psi). The polypropylene fibers

342 ACI Materials Journal/May-June 2013

added to Group 3 increased the compressive strength of the Table 4—Properties of base materials
plain concrete by approximately 8.2%. Group 2 consisted Group No. 1 2 3 4
of plain concrete with an average compressive strength of 3
56.7 MPa (8220 psi) at 28 days of age—732% greater than Density, kg/m 1698 2467 2467 1852
the compressive strength of Group 4. Modulus of elasticity, GPa 2.93 40.68 43.78 10.34
Groups 1 and 4 had promising results, considering that Split cylinder strength, kPa 1068 4375 4840 948
each group contained a large percentage of wood fibers—
10% and 6.2% by unit weight, respectively. Results from the Compressive strength, MPa 7.52 56.67 61.31 7.74
compressive tests show that the addition of wood fibers in the Notes: 1 kg/m3 = 62.43 lb/ft3; 1 GPa = 145 ksi; 1 kPa = 0.145 psi; 1 MPa = 145 psi.

mixture considerably reduces the strength of the hardened

material. Group 1, which had a 3:5 ratio of portland cement
and fly ash, had an average 28-day strength of 7.5 MPa
(1090 psi). The compressive failure modes of the wood fiber
specimens of Groups 1 and 4 were much different than those
of the plain concrete and FRC specimens. The specimens
of Groups 2 and 3 fractured at strain levels slightly over the
strain at peak load, whereas specimens from Group 1 held
together at strains well beyond the maximum stress. Although
the strength of the FA-C-WF material is much lower than
that of plain concrete, the ductile nature of the material is an
added benefit in impact and blast resistance.
Group 5 consisted of two parts: the mixture designs of
Groups 1 and 2. The top half of the specimens had the same
mixture design as Group 1. The material had an average
compressive strength of 6.5 MPa (940 psi), slightly less than Fig. 2—Distribution of 28-day UR test results for Group 1.
that of Group 1. The 28-day compressive strength of the
plain concrete mixture of Group 5 was 53.3 MPa (7730 psi).
Split-tension tests were performed on 100 x 200 mm
(4 x 8 in.) cylinders to indirectly measure the tensile strength
of the concrete according to ASTM C496/C496M-04. The
specimen was loaded along its vertical diameter at a constant
rate of 556 N/s (125 lb/s) until failure. Table 4 presents the
tensile strengths determined from the split cylinder tests.
Split cylinder tests were not conducted for Group 5. It was
assumed that the tensile strengths of the materials used in
Group 5 were similar to those of Groups 1 and 2.
The MOE of each group was determined using a universal
testing machine in accordance with ASTM C469-02. It was
assumed that the top and bottom layers of Group 5 had the
same MOE values as those of Groups 1 and 2, respectively.
Group 3 had the highest stiffness, followed closely by Group 2.
The stiffness of Group 1 was much lower than those of the Fig. 3—Distribution of 28-day UR test results for Group 2.
plain and PFRC specimens, which was primarily due to the
addition of wood fibers and the removal of aggregates from nique determined that approximately 20 to 25 specimens are
the mixture. The reduction of aggregate in the mixture had required to keep the error below 10% at a 90% level of confi-
an adverse effect on the material stiffness. Table 4 summa- dence (Badr and Ashour 2005).
rizes the mechanical properties of the five base materials In this study, five unique groups of specimens were
investigated in the drop-weight impact test. tested at 7, 14, and 28 days of age. Once the specimen
was centered properly on the bottom plate with the line of
EXPERIMENTAL RESULTS AND DISCUSSION impact parallel to the notched plane, the hammer was raised
Impact resistance
to its maximum height of 457 mm (18 in.) and dropped.
As previously discussed, the repeated drop-weight impact
test characterizes the impact resistance of FRC by measuring The hammer impacted the piston, which transferred the
the number of blows required to achieve a prescribed level load directly to the specimen. The process of dropping the
of distress. The results of the test are useful for comparing hammer was continued until a crack developed in the spec-
the relative merits of different fiber concrete mixtures. Large imen. The number of blows to reach that level of damage
variations in the test are common, which indicates that the was recorded, and the process was continued until ultimate
dependability of the test is uncertain. Therefore, several failure occurred. Once all of the data were recorded for each
specimens must be tested to keep the error under 10%. group, the mean, standard deviation, and coefficient of varia-
Statistical studies on impact resistance of concrete using the tion (COV) were calculated for each set of data. Histograms
ACI 544.2R-89 repeated drop-weight test have determined and normal probability plots were also used to represent the
that a minimum of 35 to 40 specimens must be tested to distribution of impact data. Once the mean and the variance
keep the average percent error under 10% at a 90% level of of the data were calculated, the entire normal distribution
confidence. One study conducted using the modified tech- was characterized. Histograms illustrating the distribution

ACI Materials Journal/May-June 2013 343

Fig. 4—Distribution of 28-day UR test results for Group 3.
Fig. 7—Group 1 specimens’ post-testing image.

Table 5—Statistical impact test results

7-day strength 14-day strength 28-day strength
Group
No. Specimen FC UR FC UR FC UR
Standard
2.1 2.8 2.1 3.8 2.4 4.0
deviation
1
Mean 7.3 15.5 9.5 22.5 13.3 26.7
COV 28.5 17.9 22.0 17.0 17.9 14.9
Standard
1.0 1.2 1.5 1.9 1.9 2.1
deviation
2
Mean 2.8 3.7 4.7 5.3 5.0 5.6
COV 37.1 32.8 30.6 35.9 37.7 38.1

Fig. 5—Distribution of 28-day UR test results for Group 4. Standard

4.0 6.3 6.6 7.2 5.6 7.0
deviation
3
Mean 12.1 24.6 15.2 26.6 18.1 32.4
COV 33.0 25.6 43.0 27.1 31.0 21.6
Standard
2.4 3.3 3.5 4.7 6.0 6.1
deviation
4
Mean 8.9 12.5 10.6 15.3 15.3 19.2
COV 26.8 26.1 33.3 30.7 39.1 32.0
Standard
4.9 7.4 5.9 6.2 9.5 10.1
deviation
5
Mean 12.3 16.1 17.0 18.5 23.2 24.7
COV 40.0 46.0 34.5 33.2 41.2 40.7

Furthermore, the average density of the hardened material

Fig. 6—Distribution of 28-day UR test results for Group 5.
of the 28-day ultimate impact resistance of each group are
shown in Fig. 2 through 6.
Group 1 (FA-C-WF)—Group 1 was represented with a
total of 54 specimens tested in three groups of 18 specimens
at 7, 14, and 28 days of age. The ultimate failure of Group 1,
consisting primarily of fly ash, cement, and wood fibers, was
somewhat subjective. Compared to unreinforced concrete or
FRC specimens, the stiffness of the material was relatively
low (approximately 2758 to 3447 MPa [400 to 500 ksi]).
was roughly 1698 kg/m3 (106 lb/ft3). Therefore, the impacts
caused localized damage to the specimen (refer to Fig. 7).
Ultimate failure was declared when the specimen separated
into two halves or the 12.7 mm (0.5 in.) diameter bar was
completely submerged into the specimen and the bottom
of the cylinder began penetrating it. Normally, failure was
declared due to the latter of the two.
Table 5 presents the impact resistance results for the 54 spec-
imens in Group 1. In the table, FC denotes the number of
blows to first crack and UR signifies the number of blows to
failure. The average COVs were 22.8% and 16.6% for the FC
and UR values, respectively. The COV remained approxi-
mately the same—approximately 20%—for each testing age.
The average standard error of the mean was 0.68 blows. For

344 ACI Materials Journal/May-June 2013

the 28-day test age, the 95% confidence interval about the
UR mean was 26 to 28 blows. Impact resistance increased
as the material gained strength from curing. The mean UR
at 7 days of age was 15.5, and the mean UR at 28 days was
26.7—a 72.3% increase in resistance. The frequency distri-
bution for the 28-day UR data follows the superimposed
normal distribution curve well. A normal probability curve
for the 28-day test results is presented in Fig. 8 to analyze the
normality of the distribution of the impact data. The normal
probability curve indicates that approximately 70% of the
specimens tested from Group 1 will crack within 15 blows
and fail within 30 blows. Group 1 had the largest average
percentage increase in the number of blows from FC to UR
at 28 days of age—roughly 103%.
Group 2 (concrete)—Group 2 was composed of 54 speci-
mens of unreinforced concrete with an average density Fig. 8—Standard normal distribution for Group 1.
of 2467 kg/m3 (154 lb/ft3) tested in three groups of 18 speci-
mens. A statistical summary for Group 2 is shown in Table 5.
The impact resistance of Group 2 was very low compared to
the other groups. The low resistance was partly attributed to
the brittle nature of the concrete. Unlike Group 1, no local
deformation occurred around the impact piston, and there
was almost no postcracking resistance. The average resis-
tance increased only slightly from 3.67 blows at 7 days of
age to 5.61 blows at 28 days of age. The average COVs were
34.9% and 35.6% for the FC and UR values, respectively.
The average standard error of the mean was 0.4 blows. The
histogram in Fig. 3 presents the distribution in impact resis-
tance of Group 2 for the FC and UR at 28 days of age.
The normal probability plot for Group 2 is shown in Fig. 9.
The plot represents the data recorded for the impact resistance
at 28 days of age. According to the trend for the ultimate
resistance, there is a 70% probability that a plain concrete Fig. 9—Standard normal distribution for Group 2.
specimen will fail within eight blows. Although the standard
deviations are small, the COV for Group 2 is comparable to
results from previous studies (Badr and Ashour 2005). The
COVs for Group 2 are 37.7% and 38.1% for the FC and UR
values, respectively. Figure 10 displays a failed specimen
from Group 2. Notice that the crack develops through the
notches. Also, no local deformation from the impact piston
is visible in the figure.
Group 3 (PFRC)—Group 3 consisted of 54 specimens of
the same exact mixture design as Group 2 with the addi-
tion of 19 mm (0.75 in.) polypropylene fibers. The average
density of the mixture was 2467 kg/m3 (154 lb/ft3). The
impact resistance of Group 3 was the highest of all five
groups investigated in the study. The addition of polypro-
pylene provided the specimens with significant postcracking
strength. The average number of blows to FC and UR
increased from 12.1 and 24.6 at 7 days of age to 18.1 and
32.4 at 28 days of age. The addition of polypropylene fibers
basically increased the FC strength and UR of Group 2 by
260% and 471%, respectively. Fig. 10—Representative Group 2 specimen post-testing image.
Table 5 presents a summary of the data recorded for
Group 3 for each testing age. The average COVs of the
impact test results were roughly 35.6% and 24.7% for the which are skewed to one side. The data set does not follow
FC and UR values, respectively. The COVs of the FC results the overlapping normal curve. In both cases, the resistance
were generally higher than those for UR. The average stan- values for the majority of the specimens are above the
dard error of the mean was approximately 1.5 blows. The mean resistance values. According to the normal distribu-
95% confidence interval about the mean FC resistance tion plot in Fig. 11, 70% of the fiber-reinforced specimens
at 28 days of age was 15 to 21 blows, and the corresponding from Group 3 will crack within 20 blows and will fail within
interval for the UR was 29 to 36 blows. Figure 4 presents approximately 37 blows. The COV for Group 3 was 31.0%
the scatter of the impact test results for the 28-day impact for the FC values and 22% for the UR values. These values
test. The histogram reveals a poor distribution of the data, are comparable to the COV values reported in the previous

ACI Materials Journal/May-June 2013 345


Fig. 11—Standard normal distribution for Group 3.
Fig. 12—Standard normal distribution for Group 4.
Fig. 13—Representative Group 5 specimen post-testing image.
study—39.4% and 35.2% for FC and UR, respectively (Badr
and Ashour 2005).
Group 4 (WF concrete)—Group 4 consisted of 32 concrete
specimens with the addition of wood fibers. The average
density of the material was 1852 kg/m3 (115.6 lb/ft3).
The wood fibers accounted for 6.25% of the weight of the
mixture and increased the 28-day ultimate impact resis-
tance of the plain concrete by 250%. With the addition of
wood fibers, Group 4 had an average FC and UR of 15.3 and
19.2 blows, respectively. However, the compressive strength
346
of Group 2 was approximately eight times greater than that
of Group 4. As shown in Table 5, the impact resistance of
Group 4 increased from an average FC of 8.9 blows and a
UR of 12.5 blows at 7 days of age to 15.3 and 19.2 blows,
respectively, at 28 days of age. The average COVs for
Group 4 were 33.1% and 39.5% for FC and UR, respec-
tively. The COV values for each testing age remained low
compared to the COV values obtained in previous studies
(Badr and Ashour 2005; Nataraja et al. 1999). A histogram
showing the distribution of impact test results at 28 days of
age is presented in Fig. 5. The UR distribution has a mean of
approximately 20 blows, but the distribution has the highest
frequency between 13 to 15 blows. Figure 12 presents a
normal probability plot for the 28-day test results of Group 4.
The normal probability plot, which is not close to a straight
line, indicates a poor normality of the distribution. The plot
also shows a 70% probability that a Group 4 specimen will
crack within 20 blows and fail within 24 blows.
Group 5 (bilayered)—Group 5 was unique in that it
involved two separate materials. The specimens were
constructed of two layers of material, each approximately
25.4 mm (1 in.) thick. The base material was plain concrete,
and the FA-C-WF mixture was placed on top of the concrete.
A representative sample of Group 5 is shown in Fig. 13. The
fresh concrete was roughened to enhance the bond character-
istics between the layers. Table 5 presents the impact results
for Group 5. Results show that the average number of blows
to achieve the FC and UR of the specimens increased from
12.3 and 16.6 blows at 7 days of age to 23.2 and 24.7 blows
at 28 days of age. Although the soft layer increased the
impact resistance of the plain concrete, results show signifi-
cantly less postcracking strength than solid FA-WF speci-
mens. The COV values for Group 5 are slightly higher than
the COV values of the other four groups.
The average COVs for Group 5 were 38.6% and 40.0%
for FC and UR, respectively. The average standard error
about the mean was roughly two blows. The 95% confidence
interval about the mean was 20 to 26 blows for FC values
and 22 to 27 blows for UR values. The COV values for
Group 5 at 28 days of age were 41.2% and 40.7% for the FC
and UR, respectively. These results, along with results from
the other four groups, confirm that the modified drop-weight
impact test provides reasonable COV values for FRC and
high-volume FA-WF materials.
Figure 6 presents the distribution of the 28-day UR resis-
tances for Group 5. The frequency does not follow the normal
curve well, indicating poor normality of the distribution. The
UR results are evenly spread below the mean value, while
frequencies above the mean more closely resemble a normal
distribution. Higher frequencies occur at the upper and lower
bounds of the curve rather than at the mean. However, the
appearance of the histogram could change depending on the
bin size. A normal probability plot representing the 28-day
impact test data for Group 5 is shown in Fig. 14.
SUMMARY
Figure 15 illustrates a bar graph reporting the mean FC and
UR values for all five test groups at 28 days of age. Group 3,
the PFRC group, had the highest mean impact resistance with
a mean of roughly 33 blows. Group 1 (FA-C-WF) had the
second-highest values with a mean impact resistance of
roughly 27 blows, followed closely by Group 5 (bilayered)
with a mean of 25 blows. The bilayered specimens also
had the largest standard deviation of 10 blows, followed
ACI Materials Journal/May-June 2013
by Group 3 with seven blows. Group 2 (concrete) had an
average impact resistance of five blows. The bilayered
system had the highest FC resistance.
The minimum number of tests n required to guarantee that
the percent error in the measured average is below a speci-
fied limit e at a certain level of confidence was calculated
using Eq. (1).

( )
n = t 2 v 2 e2 (1)

Table 6 shows that if the error is to be kept under 10%,

the minimum number of tests should be roughly 24 for the
plain concrete at a 90% confidence interval. Slightly more
tests are needed for the bilayered construction of Group 5.
However, the FA-C-WF material of Group 1 had much less Fig. 14—Standard normal distribution for Group 5.
variation and only four to 12 tests were needed to keep the
error under 10% at a 90% level of confidence.

CONCLUSIONS
A modified version of the current ACI 544.2R-89 drop-
weight impact test was used to determine the relative impact
resistances of six unique groups of specimens. Comparing
the results presented herein to previous studies using the
ACI 544.2R-89 repeated drop-weight impact test, one might
conclude that the modified drop-weight test used in this
impact study is a more reliable measure of impact resistance.
The modified drop-weight impact test was developed in a
previous impact study (Badr and Ashour 2005).
Several conclusions can be drawn from this study:
1. Due to the large variations in impact resistance test
results, a larger sample size is necessary to obtain more
dependable conclusions regarding the normality of the distri- Fig. 15—Mean UR to impact at 28 days of age.
butions. However, the COV values for the six groups investi-
gated in the study are considerably lower than those reported
in previous studies using the current ACI 544.2R-89 drop- Table 6—Minimum number of tests required for
weight test method. 10% error
2. Based on the variations of the impact test results,
95% confidence 90% confidence
a proposal can be made as to the number of tests needed
to prevent the error in results from rising beyond a certain Group No. FC UR FC UR
percentage. Results from this study show that at least 16 tests 1 9 6 5 4
are required to keep the error below 10% for PFRC specimens 2 38 39 23 24
based on a 90% level of confidence. Likewise, 12 FA-WF
specimens should be tested to keep the error below 10%. 3 26 13 16 8
3. PFRC had the highest drop-weight UR, but the composite 4 41 28 25 17
layer of FA-WF material applied to Group 5 provided the 5 46 45 28 27
best FC resistance values. The FA-WF material proved to
have excellent energy absorption characteristics, although
the material has much less strength and stiffness than plain finished with a trowel. This preparation creates smooth
concrete and FRCs. This was attributed to the energy absorp- finished surfaces and provides more accurate results.
tion characteristics and more ductile nature of the wood fiber
compared to their more brittle counterparts (that is, plain ACKNOWLEDGMENTS
concrete and FRCs). The plain concrete had a low impact The authors wish to express their gratitude and sincere appreciation to
resistance and a brittle failure mode. the Air Force Research Laboratory (AFRL) for funding this research work.
4. The impact test results of the FA-WF specimens of The authors would also like to thank the support staff in the Center for
Infrastructure Engineering Studies (CIES) and the Department of Civil,
Groups 1 and 4 had lower COVs than the results for the Architectural and Environmental Engineering at Missouri S&T for their
other groups. valuable contributions.
5. The FA-C-WF material had the highest level of impact
ductility with a value of 2 compared to PFRC, which had an REFERENCES
impact ductility of 1.79. The bilayered specimens had the ACI Committee 544, 1989, “Measurement of Properties of Fiber
lowest level of ductility with a value of 1.07, just below that Reinforced Concrete (ACI 544.2R-89) (Reapproved 2009),” American
Concrete Institute, Farmington Hills, MI, 11 pp.
of plain concrete. ACI Committee 544, 1996, “Report on Fiber Reinforced Concrete
6. Regarding specimen preparation, specimens should be (ACI 544.1R-96) (Reapproved 2009),” American Concrete Institute, Farm-
cut to the right thickness using a diamond saw rather than ington Hills, MI, 66 pp.

ACI Materials Journal/May-June 2013 347

 ASTM C39/C39M-05, 2005, “Standard Test Method for Compressive Nataraja, M. C.; Dhang, N.; and Gupta, A. P., 1999, “Statistical Variations
Strength of Cylindrical Concrete Specimens,” ASTM International, West in Impact Resistance of Steel Fiber Reinforced Concrete Subjected to Drop
Conshohocken, PA, 7 pp. Weight Test,” Cement and Concrete Research, V. 29, No. 6, pp. 989-995.
ASTM C143/C143M-05, 2005, “Standard Test Method for Slump of
Hydraulic-Cement Concrete,” ASTM International, West Conshohocken, APPENDIX A
PA, 4 pp.
ASTM C150-05, 2005, “Standard Specification for Portland Cement,”
ASTM International, West Conshohocken, PA, 8 pp.
ASTM C192/C192M-06, 2006, “Standard Practice for Making and
Curing Concrete Test Specimens in the Laboratory,” ASTM International,
West Conshohocken, PA, 8 pp.
ASTM C469-02, 2002, “Standard Test Method for Static Modulus of
Elasticity and Poisson’s Ratio of Concrete in Compression,” ASTM Inter-
national, West Conshohocken, PA, 5 pp.
ASTM C494/C494M-05a, 2005, “Standard Specification for Chemical
Admixtures for Concrete,” ASTM International, West Conshohocken, PA,
10 pp.
ASTM C496/C496M-04, 2004, “Standard Test Method for Splitting
Tensile Strength of Cylindrical Concrete Specimens,” ASTM International,
West Conshohocken, PA, 5 pp.
ASTM C566-97(2004), 1997, “Standard Test Method for Total Evapo-
rable Moisture Content of Aggregate by Drying,” ASTM International,
West Conshohocken, PA, 3 pp.
ASTM C617-98, 1998, “Standard Practice for Capping Cylindrical
Concrete Specimens,” ASTM International, West Conshohocken, PA, 5 pp.
ASTM C1116/C1116M-06, 2006, “Standard Specification for
Fiber-Reinforced Concrete and Shotcrete,” ASTM International, West
Conshohocken, PA, 7 pp.
Badr, A., and Ashour, A. F., 2005, “Modified ACI Drop-Weight Impact
Test for Concrete,” ACI Materials Journal, V. 102, No. 4, July-Aug.,
pp. 249-255.
Carey, N. L., and Myers, J. J., 2011, “Full-Scale Blast Testing of Hybrid
Barrier Systems,” Behavior of Concrete Structures Subjected to Blast and
Impact, SP-281, G. Thiagarajan, E. Williamson, and C. Conley, eds., Amer-
ican Concrete Institute, Farmington Hills, MI, pp. 1-20. (CD-ROM)
Joshi, N., and Myers, J. J., 2005, “Investigation of an Alternative Wood
Fiber-Fly Ash Material for Infill Wall Systems (CIES 06-60),” Center for
Infrastructure Engineering Studies, University of Missouri, Rolla, MO, Fig. A1—Detailed base plate modifications to proposed ACI
123 pp. drop-weight test.

348 ACI Materials Journal/May-June 2013

  In ACI STRUCTURAL Journal
The American Concrete Institute also publishes the ACI
Structural Journal. This section presents brief synopses of
papers appearing in the current issue.
From the May-June 2013 issue
110-S29—On the Effectiveness of Steel Fibers
as Shear Reinforcement
by Fausto Minelli and Giovanni A. Plizzari
An experimental study on steel fiber-reinforced concrete (SFRC) beams
subjected to shear loading tested at the University of Brescia is presented
and discussed. Eighteen full-scale experiments were carried out to investigate
the effect of randomly distributed steel fibers within the concrete matrix
on shear behavior. The focus was on the parameters influencing the shear
response of members. All tested members contained no conventional shear
reinforcement. All SFRCs used were characterized in tension according to
the provision included in the fib Model Code 2010. A useful database was
developed linking the shear strength of members to the codified residual
strengths of the corresponding fiber-reinforced concrete (FRC) materials.
Results show that a relatively low amount of fibers (Vf < 0.7%) can
significantly increase the shear strength and ductility of concrete beams
without transverse reinforcement.
A critical discussion of two recent analytical models for calculating the
shear strength of FRC materials is also provided.
110-S30—Crack Protocols for Anchored Components
and Systems
by Richard L. Wood and Tara C. Hutchinson
The opening and closing of cracks in concrete (“crack cycling”), can be
induced during an earthquake and has been shown to significantly influence
the tension behavior of anchors. This paper presents a protocol for testing
anchors in cycling cracks in an effort to represent earthquake loading of
anchored components and systems. In the protocol, cyclic cracking is simu-
lated while a representative tension or shear load history is simultaneously
applied. The protocol is developed by performing rainflow counting of the
curvature histories extracted from nonlinear history analyses of a suite of
building models. Herein, the focus is on development of the crack histories;
loading histories are outside the scope of this paper. The selection of earth-
quake motions is guided by results from a probabilistic seismic hazard analysis
of a seismically active region in Southern California. The resulting cyclic crack
protocol is presented in statistical terms and an example application is provided.
110-S31—Cyclic Loading Test for Beam-Column Connection
with Prefabricated Reinforcing Bar Details
by Tae-Sung Eom, Jin-Aha Song, Hong-Gun Park, Hyoung-Seop
Kim, and Chang-Nam Lee
A Prefabricated Reinforcing Bar Construction (PRC) Method was developed
for fast construction and cost savings. In this study, a prefabricated reinforcing
bar connection method for the earthquake design of beam-column connections
was developed. Three interior and one exterior full-scale beam-column connec-
tions were tested under cyclic loading. The test specimens were designed to
satisfy the requirement of the special moment frame specified in ACI 318-08.
The connection method used reinforcing bar welding, coupler splice, and headed
bar anchorage considering the PRC Method. The test results showed that the
story drift ratio of the PRC beam-column connections exceeded 3.5%—a
requirement of ACI 374.1-05. The load-carrying capacity, yield stiffness, and
energy dissipation capacity of the PRC specimens were comparable to those of
the conventional RC specimen. The major failure modes are discussed.
110-S32—Shear Strength of Reinforced Concrete Walls for
Seismic Design of Low-Rise Housing
by Julian Carrillo and Sergio M. Alcocer
In the last decade, the construction of low-rise housing made of reinforced
concrete (RC) walls and slabs in Latin America has increased consider-
ably. These box-type structures commonly have large lateral stiffness
ACI Materials Journal/May-June 2013
PDF versions of these papers are available for download at
the ACI website, www.concrete.org, for a nominal fee.
and strength, thus exhibiting low lateral displacements and shear forces
demands. The low level of seismic response has prompted designers to
use concrete strengths of 15 to 20 MPa (2175 to 2900 psi), as well as
of 100 mm (4 in.) thick walls with web steel reinforcement ratios smaller
than the minimum prescribed by most design codes. Considering these
particular wall characteristics, design requirements in current codes are
not directly applicable. Moreover, a blind application of current require-
ments may lead to an unjustifiable excessive cost of a housing unit, espe-
cially because of the web steel ratio required. To improve design methods
for this type of construction, a behavioral model and equations capable
of estimating the peak shear strength of walls for low-rise housing were
developed and calibrated from test results.
110-S33—Reducing Steel Congestion without Violating
Seismic Performance Requirements
by Gerasimos M. Kotsovos, Emmanuel Vougioukas,
and Michael D. Kotsovos
This paper focuses on an experimental investigation of the behavior
of reinforced concrete (RC) beam-column specimens subjected to cyclic
loading combined with a constant axial force. The specimens have the same
geometry and longitudinal reinforcement but differ in that the portion of
the longitudinal reinforcement within the critical regions is debonded from
concrete in certain specimens and in the transverse reinforcement arrange-
ment; the latter is designed either in compliance with the earthquake-resis-
tant design clauses of the current European codes or in accordance with the
method of the compressive force path. The results indicate that adopting
the latter method yields significantly less transverse reinforcement in the
critical regions without compromising the code performance requirements.
Moreover, when the application of this method is combined with debonding
the longitudinal reinforcement within these regions from concrete, a signifi-
cant improvement in structural behavior is achieved.
110-S34—Recommended Procedures for Development
and Splicing of Post-Installed Bonded Reinforcing Bars in
Concrete Structures
by Finley A. Charney, Kamalika Pal, and John Silva
The use of post-installed anchors for connections in both new and existing
structures is commonplace and growing. A related type of post-installed
connection that has been employed for decades is the bonding of rein-
forcing bars into holes drilled into concrete to facilitate structural extensions and
strengthening of reinforced concrete structures. Although ACI 318-11 includes
provisions for the design of adhesive anchors in concrete (anchor rods bonded
with adhesive in a drilled hole), it does not address the design of post-installed
reinforcing bars. This paper addresses the issue by providing background into
the adhesive anchor design and development length provisions of ACI 318 as
well as the provisions for post-installed reinforcing bars available in international
standards. The paper makes recommendations for the development of a new
procedure that is applicable to the design of post-installed reinforcing bars.
110-S35—Two-Parameter Kinematic Theory for Shear Behavior
of Deep Beams
by Boyan I. Mihaylov, Evan C. Bentz, and Michael P. Collins
This paper presents a kinematic model for deep beams capable of
describing the deformed shape of such members in terms of just two
primary parameters. The kinematic model is combined with equilibrium
equations and stress-strain relationships to form a theory to predict the shear
strength and deformations patterns of deep beams at shear failure. These
deformation patterns include crack widths, maximum deflections, and the
complete displacement field for the beam. The kinematic theory predicts
the components of shear strength of deep beams and how they vary with
349
a/d ratio. These components indicate a significant size effect for the shear
strength of deep beams, even for members with transverse reinforcement. The
theory has been validated against a large number of experimental results.
110-S36—Design Formulas for Cracking Torque and Twist
in Hollow Reinforced Concrete Members
by Chyuan-Hwan Jeng, Hao-Jan Chiu, and Sheng-Fu Peng
Hollow concrete members are in wide use and are becoming increasingly
important in terms of energy savings and carbon reduction. To date, the
prediction of cracking torque in hollow reinforced concrete (RC) members
still relies on simplistic empirical formulas, and the accuracy of the existing
formulas is rarely examined. In this study, four existing formulas are used to
calculate the cracking torques Tcr and twists qcr of 44 hollow RC beam spec-
imens currently available in the literature. An existing empirical formula for
Tcr is modified to fit the test values. Based on the latest rational formula for
solid members, a new formula for the Tcr and qcr of hollow RC members
is developed to form a unified formula for solid and hollow sections. The
predicting accuracies of these formulas are compared and evaluated. It is
shown that the proposed formulas produce significantly more accurate and
reliable predictions than the existing formulas.
110-S37—Breakout Capacity of Headed Anchors in Confined
Concrete: Experimental Evidence
by Roberto Piccinin, Sara Cattaneo, and Luigi Biolzi
While provisions are available in current design codes that account for
the presence of tensile fields causing concrete cracking, no provisions are
available for anchors embedded in compressively prestressed concrete.
Because of this lack of information, a series of experiments were performed
to evaluate the tensile breakout capacity of headed anchors embedded in
confined and unconfined concrete. To simulate the confinement, uniform
biaxial compression (prestress) was applied to concrete specimens in the
directions orthogonal to the anchors’ stems.
This paper discusses the influence of embedment depth and confinement
on the behavior of headed anchors in terms of strength, ductility, and failure
pattern. A comparison between actual code provisions for unconfined
concrete and the experimental results reported herein is provided. Proposals for
the extension of the actual provisions to take the effects of confinement into
account are provided and are contingent on completion of further investigation.
110-S38—Cracking Behavior of Steel Fiber-Reinforced
Concrete Members Containing Conventional Reinforcement
by Jordon R. Deluce and Frank J. Vecchio
Uniaxial tension tests were conducted on 12 plain reinforced concrete
(RC) and 48 large-scale steel fiber-reinforced concrete (SFRC) specimens,
each containing conventional longitudinal reinforcement, to study their
cracking and tension-stiffening behavior. The test parameters included fiber
volumetric content, fiber length and aspect ratio, conventional reinforce-
ment ratio, and steel reinforcing bar diameter. “Dog-bone” tension tests and
bending tests were also performed to quantify the tensile properties of the
concrete. It was found that the cracking behavior of SFRC was significantly
altered by the presence of conventional reinforcement. Crack spacings and
crack widths were influenced by the reinforcement ratio and bar diameter
of the conventional reinforcing bar, as well as by the volume fraction and
aspect ratio of the steel fiber. Details and results of the experimental inves-
tigation are provided and discussed.
110-S39—Performance of AASHTO-Type Bridge Model
Prestressed with Carbon Fiber-Reinforced Polymer
Reinforcement
by Nabil Grace, Kenichi Ushijima, Vasant Matsagar, and Chenglin Wu
Carbon fiber-reinforced polymer (CFRP) composite material has been
widely studied and applied in bridge engineering as an alternative solution
to the corrosion-related problems posed by steel reinforcement. Never-
theless, adoption of CFRP reinforcement to replace conventional steel
reinforcement in highway bridges has not been fully realized yet in the
field. Therefore, large-scale experimental investigations on bridges with
CFRP reinforcement are essential to encourage its widespread application
in highway bridges. This paper presents an experimental investigation
conducted on a one-third-scale AASHTO-type bridge model prestressed
350
with carbon fiber composite cable (CFCC) strands. The bridge model was
designed, constructed, instrumented, and tested to thoroughly investigate
its flexural behavior, strain response, and ultimate load failure. A separate
one-third-scale single AASHTO-type I-beam was also constructed and
tested to study its flexural and shear behavior as a control beam. In general,
both the control beam and the bridge model experienced compression-
controlled failure as anticipated. Significant cracking and deflection were
experienced prior to failure. The ultimate strength of the control beam and
the bridge model were in close agreement with the values estimated using
the Unified Design Approach.
110-S40—Testing of Normal- and High-Strength Concrete
Walls Subjected to Both Standard and Hydrocarbon Fires
by Tuan Ngo, Sam Fragomeni, Priyan Mendis, and Binh Ta
Ten large-scale concrete wall panels were tested in this study. Four
walls were of normal-strength concrete (NSC)—two of which were axially
loaded at an eccentricity of 10 mm (0.39 in.) and two with no load—and
exposed to either standard or hydrocarbon fires. Four identically dimen-
sioned high-strength concrete (HSC) walls were also tested using these
variables. An additional two HSC walls, with polypropylene fibers added,
were tested under hydrocarbon fire conditions only. All walls were tested
in a vertical position in a large furnace and supported at the top and
bottom only. The results indicate that all concrete wall panels exposed
to the standard fire tests survived the 120-minute fire period, with low to
moderate spalling evident. The NSC walls exposed to hydrocarbon fires
also survived the 120-minute test, whereas the HSC walls experienced
severe spalling under these fire conditions with failure at 31 minutes. The
addition of polypropylene fibers in the concrete improved the fire resis-
tance of HSC walls in hydrocarbon fire to 65 minutes.
110-S41—Effect of Washout Loss on Bond Behavior of Steel
Embedded in Underwater Concrete
by Joseph J. Assaad and Camille A. Issa
Limited studies have been undertaken to investigate the bond prop-
erties of reinforcing steel bars embedded in underwater concrete (UWC).
Approximately 60 pullout tests were carried out to evaluate the effect of washout
loss (W) on residual compressive and bond strengths. Washout was determined
using the CRD C61 test and by simulation using a newly developed air-pressur-
ized tube. Reference mixtures sampled in dry conditions were also tested.
Test results showed that bond between steel and UWC is affected by
a combination of parameters that complement those documented in the
literature for concrete cast and consolidated above water. These include the
level of W, degree of segregation, hydrostatic water head (H), and interfa-
cial concrete-water velocity. The bond-stress-versus-slip behavior of UWC
is remarkably different from the one obtained using reference mixtures.
Initially, the linear response is less stiff due to a coupled effect related to
lower strength and increase in the relative coarse aggregate concentration.
The slip at ultimate bond strength was found to decrease for UWC mixtures
exhibiting higher levels of W.
110-S42—Experimental Evaluation of Disproportionate
Collapse Resistance in Reinforced Concrete Frames
by Stephen M. Stinger and Sarah L. Orton
Reinforced concrete frame structures may possess an inherent ability to
withstand collapse through the use of alternative resistance mechanisms.
These mechanisms include Vierendeel action, catenary action, compressive
arch action, and contributions from infill walls. This research tested a series
of three one-quarter-scale, two-bay-by-two-story frames. The column
between the two bays was removed to simulate a collapse scenario. Flexural
analysis of the frame with discontinuous reinforcement indicated little load
capacity; however, it reached a load of 2.34 kip (10.4 kN) under compres-
sive arch action and 8.19 kip (36.4 kN) under catenary tension before the
top bars ripped out of the stirrups. The frame with continuous reinforcement
also reached 8.3 kip (36.9 kN) in catenary action before the longitudinal
bars fractured under the tensile load. A frame with partial-height infill walls
showed only a minor increase in strength in the compressive arch phase.
The results show that both compressive arch and catenary action are viable
resistance mechanisms in frames under a collapse loading.
ACI Materials Journal/May-June 2013
ACI Materials Journal/May-June 2013 351
CALL FOR ACTION
ACI Invites You To...
Do you have EXPERTISE in any of these areas?
• BIM
• Chimneys
• Circular Concrete Structures Prestressed by Wrapping
with Wire and Strand
• Circular Concrete Structures Prestressed with
Circumferential Tendons
• Concrete Properties
• Demolition
• Deterioration of Concrete in Hydraulic Structures
• Electronic Data Exchange
• Insulating Concrete Forms, Design, and Construction
• Nuclear Reactors, Concrete Components
• Pedestal Water Towers
• Pipe, Cast-in-Place
• Strengthening of Concrete Members
• Sustainability

Then become a REVIEWER for the

A Structural Journal or the ACI Materials Journal.
ACI
How to become a Reviewer:
1. Go to: http://mc.manuscriptcentral.com/aci;
2. Click on “Create Account” in the upper right-hand corner; and
3. Enter your E-mail/Name, Address, User ID and Password, and
Area(s) of Expertise.

Did you know that the database for MANUSCRIPT

CENTRAL, our manuscript submission program,
is separate from the ACI membership database?
How to update your user account:
1. Go to http://mc.manuscriptcentral.com/aci;
2. Log in with your current User ID & Password; and
3. Update your E-mail/Name, Address, User ID and Password,
and Area(s) of Expertise.

QUESTIONS?
E-mail any questions to Journals.Manuscripts@concrete.org.
ACI
Materials
J O U R N A L
J O U R N

This journal and a companion periodical,

ACI Structural Journal, continue the publishing
tradition the Institute started in 1904.
Information published in ACI Structural Journal
includes: structural design and analysis of concrete
elements and structures, research related to
concrete elements and structures, design and
analysis theory, and related ACI standards and
committee reports.

38800 Country Club Drive

Farmington Hills, MI 48331