Professional Documents
Culture Documents
Aci sp-234-2006
Aci sp-234-2006
Durability of Concrete
Proceedings
Seventh CANMET/ACI International Conference
Montreal, Canada 2006
V. M. Malhotra
Editor
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete
Editor
V.M. Malhotra
<e(j!'))
American Concrete Institute®
Advancing concrete knowledge
SP-234
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The Institute is not responsible for the statements or opinions expressed in its
publications. Institute publications are not able to, nor intended to, supplant individual
training, responsibility, or judgment of the user, or the supplier, of the information
; presented.
The papers in this volume have been reviewed under Institute publication procedures
by individuals expert in the subject areas of the papers.
Copyright© 2006
AMERICAN CONCRETE INSTITUTE
P.O. Box 9094
Farmington Hills, Michigan 48333-9094
All rights reserved, including rights of reproduction and use in any form or by any means,
including the making of copies by any photo process, or by any electronic or mechanical
device, printed or written or oral, or recording for sound or visual reproduction or for use
in any knowledge or retrieval system or device, unless permission in writing is obtained
from the copyright proprietors.
May 2006
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SP-234-19: Carbonation of Concrete Incorporating High Volumes of Fly Ash ......... 283
by N. Bouzoubaa, B. Tamtsia, M.H. Zhang, R.L. Chevrier, A. Bilodeau,
and V.M. Malhotra
SP-234-43: Durability of Lightweight Concrete Containing High Volume Fly Ash and
Highly Porous Bottom Ash ..................................................................................... 703
by M. Nisnevich, G. Sirotin, andY. Eshel
SP-234-46: ASR-Free Dry Shake Hardeners for Concrete Industrial Floors .............. 747
by M. Collepardi, E.N. Croce, G. Fazio, J.J. Ogoumah Olagot, and R. Troli
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete -
The Zigzag Course of Progress
by P.K. Mehta
Synopsis: Concrete design and construction practices today are essentially strength-
driven. However, due to escalation in the repair and replacement costs of structures
and a growing concern about sustainability of the concrete industry, more attention is
being paid now to durability issues. An overview of the state of the built infrastructure
in the world shows that the current methods for achieving considerable enhancement
of durability of concrete structures are proving inadequate, especially with concrete
members exposed to severe weather conditions. It seems that inhomogeneities in
the microstructure of concrete are responsible for microcracks which, when subjected
to tensile stress from weathering and loading effects during the service, grow into
macrocracks. When the macrocracks, voids, and microcracks become interlinked
there is a sudden increase in the rate of transport of water, carrying harmful ions and
gases from the surface into the interior of concrete. This point marks the initiation
and progressive deterioration of the material from one or more causes. A three-stage
concrete damage process is discussed to show what changes in concrete technology
are needed for radical enhancement in the durability of structures to be built in the
future.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
There are compelling reasons why the concrete construction practice during the 21st
century will have to be driven by durability rather than by the strength considerations.
For instance, increasing carbon emissions, global warming, and a credible threat of
climate change have brought to public attention the issue that sustainable economic
development cannot occur without reducing the wasteful consumption of natural
resources through radical enhancement of durability of manufactured products. Note the
following excerpts from a March 2001 report, " Vision 2030: A Vision for the U.S.
Concrete Industry", issued by the Strategic Development Council, an organization of
senior concrete industry executives dedicated to promoting the development of new
technologies for which there is near-term commercial demand2•
In this paper, I have used a holistic approach that involves a search for the root causes
behind commonly known durability problems. This approach has led to identification of
the three stages of concrete damage process. For radical enhancement of durability of
concrete structures, it is proposed that the damage process must not be permitted to go
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
It was nearly 20 years ago when the U.S. National Materials Advisory Board, in 1987,
sounded an alarm bell by reporting that concrete bridge-decks, built since the 1970's,
were suffering from an epidemic of durability problems that would require billions of
dollars to fix. Meanwhile, more cases of serious and premature deterioration of concrete
infrastructure have been reported from around the world. To site a few examples,
Khanna et a!. 3 in 1988 reported the results of an investigation of premature deterioration
in reinforced concrete piles of the Rodney Terminal at St. John in New Brunswick,
Canada. The investigation revealed that thermal cracks during the pile manufacture were
precursors to serious cracking from cycles of freezing and thawing, and from
reinforcement corrosion. Gerwick4 in 1989 reported similar cases of premature cracking
and deterioration in reinforced concrete lining of undersea tunnels in Dubai, Hong
Kong, and Japan. In the 1990's, Shayan and Quick5 from Australia, and Collepardi6
from Italy reported cases of premature cracking and deterioration in prestressed concrete
railway sleepers.
A 1995 report on the condition of 29 bridges in Kansas stated that there was
twice as much cracking with 6400 psi (44 MPa) concrete than with 4500 psi
(31 MPa) concrete. In 1997, the high-performance concrete deck in the Louetta
Overpass - a showcase bridge in Texas - cracked more than the conventional
concrete deck in the aqjoining lane. In Denver, the high-strength concrete in
the 23rd Street Viaduct cracked even before construction was finished. This
cracking was due to very high thermal contraction and autogenous shrinkage
resulting from the use of a high cement content (wlc = 0.31), and a fast-
hydrating Type II cement. The cement fineness was 391 m2/kg and the C~
plus-C3S content was 72%. The cracking tendency of this concrete mixture
might have been exacerbated by silica fume, which tends to increase the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
In the Aprill998 issue, the ASCE News -a publication of the American Society of Civil
Engineers issued a report card on the state of durability of concrete structures, in
general, at the close of the 201h century. This report card assigned a "D" grade, (i.e. very
poor grade) to durability. It was reported that a huge amount of money, some 1.3 trillion
dollars were needed for repair and replacement.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
That concrete progress is not linear, but moves in a zigzag course was demonstrated
when FHWA and several other U.S. organizations stopped advocating the use of high-
early strength HPC mixtures for cast-in-place, massive, bridge elements. In 1998, a
FHWA report contained the following guidelines with regard to HPC for building
bridges in the 21st Century 10 :
The Port Authority ofNew York and New Jersey maintains many transportation
facilities in the New York City metropolitan area, including bridges, tunnels, and
airports. The concrete used at these facilities is subjected to high traffic volume and
harsh weather conditions. According to Bognacki et al. 11 :
"In the past, as with other specifications, our concrete specifications only
emphasized a compressive strength for a mixture proportion, and did not give
the same emphasis to concrete permeability and durability. It is of little
consolation to know that the compressive strength met the required contract
requirement when the concrete is cracked, spalled, and delaminated. This
necessitates significant repair work and causes delays at these transportation
facilities, inconveniencing the public. "
Learning from experience, the Port Authority of New York and New Jersey set an
example that offers a step in the right direction for concrete durability enhancement. The
Agency is now specifying performance-based concrete mixtures, for instance a
There is another important reason why the goal of achieving a radical enhancement in
durability of yet-to-be-built concrete structures will have to be pursued with a sense of
urgency. Among the manufacturing industries, the concrete industry is the largest
consumer of materials in the world. It is estimated that today, globally, we are producing
concrete at the yearly rate of some 12,000 million tonnes. The concrete industry
therefore requires a very large volume of natural materials like sand, gravel, and crushed
stone. Furthermore, it is consuming approximately 1, 700 million tonnes/year cementing
materials composed mostly of portland-cement clinker - a manufactured product that
not only is highly energy-intensive but also responsible for large emissions of carbon
dioxide, which is the most voluminous global-warming gas. Portland-cement clinker
production also consumes large amounts of limestone and fossil fuels. Due to
unsatisfactory durability of a large number of concrete structures, considerable volumes
of concrete are being used for repair and replacement of structures. In short, the
resource productivity of the concrete industry is not high, and the industry is not
sustainable at its current global rate of consumption of natural resources and
production ofgreen-house gas emissions.
Field experience with deteriorated concrete structures shows that, in order of decreasing
frequency, the apparent causes of deterioration generally are the corrosion of steel
reinforcement, frost action, alkali-silica reaction, and sulfate attack. In practice, the
deterioration of concrete is seldom due to a single cause; at the advanced stages of
material's degradation, more than one deleterious process is found at work. Usually the
physical and chemical processes of deterioration are so closely intertwined and mutually
supporting that separation of cause from effect becomes impossible. If we examine the
mechanisms of deterioration separately, they appear to be different from each other.
However, irrespective of the apparent cause of deterioration, it is generally observed that
the material failure occurs through a succession of expansion and cracking cycles in
which the presence offree water is always necessary.
It is well known that ordinary concrete mixtures, with moderate cement content (300 to
350 kg/m3) and moderate w/c (0.45 - 0.60), when properly consolidated and cured, are
essentially watertight. So, what is the source of free water in hardened concrete? For an
answer to this question let us examine the four causes of deterioration of concrete
individually:
Frost Action
Commenting on the damage to concrete by frost action, Valenta 14 suggested how the
material lost its watertightness:
Reinforcement Corrosion
The authors proposed that, as a result of heavy reinforcement and differential cooling
rates, microcracks must have occurred in the massive beams (8 by 3.7 by 1.8 m) during
the steam-curing process. These pre-existing microcracks later enlarged and became
continuous on exposure to more severe weathering action on the windward faces.
Thereafter, penetration of the salt water to the surface of the reinforcing steel set the
stage for the corrosion-cracking-corrosion type of chain reaction, which led to serious
damage. A diagrammatic presentation of the Mehta-Gerwick hypothesis is shown in Fig.
1.
Sulfate Attack
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
microcracks.
Alkali-Silica Reaction
In regard to concrete damage from alkali-silica reaction, Swamy 16 made the following
comments:
Funny things can then happen in real life - the interior columns of an exposed
bridge, sheltered from direct sunlight and rain, may show no cracking whilst
the exterior columns may develop extensive cracking. A structural member,
when partly sheltered and partly exposed by the nature of the structure, showed
very different crack patterns, with extensive cracking in the exposed faces and
little or no cracking on the sheltered parts .....
Exclude water - and one can almost have a trouble-free reaction even if
concrete contains reactive aggregates and mobile alkalis. Marked
deterioration due to the alkali-silica reaction, in field practice, therefore,
occurs under wet environmental conditions.
Among the key features of the holistic approach to concrete deterioration, the point
worthy of special attention is that during the first stage, little or no apparent damage to
the structure takes place as long as it remains watertight. The second stage marks the
initiation of the damage which occurs at a slow rate at first, and then proceeds rather
rapidly as the system becomes more permeable. It is suggested that during the latter part
of this stage the hydraulic pressure of the pore fluid in saturated concrete rises due to
one or more expansive phenomena (e.g., freezing of water, corrosion of the reinforcing
steel, and swelling of microcrystalline ettringite or alkali-silica gel). If, at the same time,
the hydroxyl ions from the cement paste are being leached away and replaced by acidic
ions, such as chloride or sulfate, the calcium silicate hydrate would suffer a loss of
adhesion, and as a consequence the concrete strength and elastic modulus are reduced.
The damage to the structure accelerates when the microcracks have grown into
macrocracks as a result of the two mutually reinforcing processes of damage. These
features of the holistic approach emphasize the common role of microcracks in the
initiation and propagation of concrete damage, and provide the basis for the three-
stage concrete damage process that will be discussed later.
For a variety of well-known reasons, even with structural members designed for
compressive loads, tensile stress and tensile cracking are unavoidable. The
reinforcement of concrete with steel does not eliminate tensile cracks but it does restrict
the crack-widths to 0.15 mm or less. The fine cracks, known as microcracks, are too
small to be visible and quantified, and are therefore ignored in the structural design and
construction practice. It is generally accepted that microcracks caused by settlement,
plastic shrinkage, restrained thermal shrinkage and drying shrinkage, and accidental
overloads would not have any adverse effect on the static behavior of concrete
structures. However, due to the important role played by microcracks in determining the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Also, when the growing microcracks link up with existing macrocracks and voids in
concrete, this is the point in time that marks a sudden increase in the permeability. The
internal damage begins when the water carrying corrosive ions and gases starts to
penetrate more readily into concrete. As described earlier, once concrete becomes
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
saturated and any one of the four above-described expansive phenomena is initiated, the
hydraulic pressure in the pore fluid would increase, enabling the microcracks to grow.
The end result is a further loss of watertightness of concrete followed by expansion,
cracking, spalling, and loss of mass. Obviously, for a holistic approach to durability,
the microcracks and their growth under service conditions is another common
denominator that must be controlled if we want to achieve a radical enhancement in
concrete durability. To achieve this objective, we must understand and control the
microstructure of concrete.
The concept of "no damage" state is useful for achieving a radical enhancement in
concrete durability. Clearly, without the growth in microcracks, there would be no
increase in the permeability of concrete. Any method that successfully prolongs Stage 1
of the damage process (Fig. 3), would have the effect of delaying exposure of the
structure to subsequent stages of damage. Thus, by holding the structure in Stage 1 for a
very long period of time, it is possible to achieve a radical enhancement in durability.
The use oflow w/c, as mentioned by the ACI Building Code and other codes for durable
concrete, enables a reduction in the volume and size of capillary voids, but this alone is
not sufficient to reduce the cement paste content of concrete which is the source of
microcracking from thermal shrinkage and drying shrinkage. To reduce the cement paste
content, both the water content and the cement content must be reduced as much as
CONCLUDING REMARKS
responsible for concrete deterioration. This has led to the development of the concept of
a three-stage sequence of concrete damage process summarized as follows:
• A high degree of water saturation is one of the three root causes for
durability problems because, in almost every case, it precedes the
occurrence of any visible damage to concrete structures. Water is not only
the primary vehicle for transport of ions and gases into concrete but also
is implicated in expansion and cracking mechanisms in solids.
• Microcracks in concrete are a second root cause of durability problems
because their growth and connectivity with macrocracks and voids marks
the point in time when the stage of "no damage" ends, and the stages of
initiation and propagation of damage begin as a result of loss in water-
tightness. Being discontinuous and invisible to the human eye, microcracks
are ignored by the strength-driven design and construction practice in the
concrete industry.
• The first root cause of the concrete durability problems is the presence of
inhomogeneities in the hydrated cement paste (e.g. large capillary voids
and oriented layers of crystalline calcium hydroxide). These
inhomogeneities serve as potential sites for microcrack formation when
concrete experiences tensile stress from mechanical loading and
environmental exposure conditions.
• By a judicious selection of concrete-making materials and mix-
proportions, and by proper construction practice, the microstructural
There are a number of indicators showing that the era of lack of progress in
durability of concrete structures may finally have come to an end.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
GRQWTH IN MICROCRACKING
• HUMIDITY AND
TEMPERATURE GRADIENTS
• REPEATED LOADS AND
OVERLOADS
• CHEMICAL ATIACKS,
LEACHING OF CEMENT PASTE
AND FREEZE-THAW CYCLES
SEAWATER
CONTAINING
HIGHLY
PERMEABLE
o ,C0
2 2 AND cr
CONCRETE
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INITIATION AND
PROPAGATION OF
CORROSION OF
EMBEDDED STEEL
I
I
Penetration I
towards ------1
reinforcement I
I Acceptable depth
Initiation Propagation
i -----------
LIFE
I
No Loss in
Water-Tightness Water-Tightness Water-Tightness
GROWTH OF DAMAGE
NO DAMAGE MICROCRACKING PROPAGATION
Fig. 3- A holistic approach illustrating the three stages of concrete damage process.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: Several ageing models are available for the prediction of the mechanical
consequences of rebar corrosion. They are used for service life prediction of
reinforced concrete structures. Concerning corrosion diagnosis of reinforced concrete,
some Non Destructive Testing (NDn tools have been developed, and have been in
use for some years. However, these developments require validation on existing
concrete structures. The French project "Benchmark des Poutres de la Rance"
contributes to these aspects. It has two main objectives: (i) validating of mechanical
models to estimate the influence of rebar corrosion on the load bearing capacity
loss of a structure, (ii) qualifying the use of the NDT results to collect information on
steel corrosion within reinforced-concrete structures. This project started in 2004
for 2 years. Ten French and European institutions from both academic research
laboratories and industrial companies are contributing. This paper presents the
project that was divided into several work packages: (i) the reinforced concrete
beams are characterized from non-destructive testing tools, (ii) the mechanical
behavior of the beams are experimentally tested, (iii) complementary laboratory
analysis are performed and (iv) numerical simulations will be finally compared to the
experimental results obtained with the mechanical tests.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
these developments need to be validated on existing concrete structures. Moreover, it is
important to take into account the recent progress in the use of non-destructive testing
methods. The development of NDT applied on materials and condition assessments of
bridges has become a major subject of interest for the structural engineering community.
The French project "Benchmark des poutres de la Rance" contributes to these items
through a study of 20 reinforced concrete beams exposed for 40 years in a marine
environment.
OBJECTIVES
The two main objectives of this project are:
• To validate existing mechanical models for estimating the influence of rebar
corrosion on the load-bearing capacity loss of a structure by taking into account
the reduction of steel cross section and/or the decrease of steel-concrete bond,
• To establish the reliability of the use of non-destructive results for collecting
information on steel corrosion within reinforced-concrete structures.
Finally, this project will allow:
• To build up a reference: by establishing a large database from characterization
tests of prestressed beams after 40 years exposure period in a marine
environment (mechanical, electrochemical and physico-chemical properties) for
improving the knowledge on the influence of corrosion on mechanical
performance of reinforced concrete elements (reduction of steel cross section,
loss ofbond, etc.);
• To reduce the uncertainties due to simulations and measurements: by selecting
adapted mechanical models for the design or for evaluating the security margins
of reinforced concrete structures exposed to corrosion , and by promoting the
validation of some future mechanical models based on the experimental
database obtained during this project;
• To increase the development of non destructive methods for inspection and
control;
PARTNERS
The partners in the "Benchmark des poutres de Ia Rance" are:
• For the experimental part: Centre d'Expertise de Recherches et d'Etudes du
Batiment et des Travaux Publics (CEBTP, France), Electricite de France (EDF
R&D, France), Commissariat a l'Energie Atomique (CEA, Saclay, France),
Laboratoire Central des Ponts et Chaussees (LCPC, Paris), Laboratoire
Materiaux et Durabilite des Constructions (LMDC, Toulouse), Laboratoire
d'Etudes et de Recherches sur les Materiaux (LERM, France), OXAND
(France), SP institute (Sweden).
• For the modelling part: EDF R&D (France), CEA Saclay (France)), LCPC
(Paris, France), LMDC (Toulouse, France), OXAND (France), Laboratoire de
Mecanique et Technologies (LMT, France), Laboratoire de Mecanique de Lille
(LML, France), Laboratoire GeM (Nantes, France), Institute Eduardo Torroja
(Madrid, Spain).
DESCRIPTION OF THE BEAMS
This project is based on a long-term experimental program started in 1962 by the "Union
Technique Interprofessionnelle des Federations Nationales du Batiment et des Travaux
Publics". Initially 40 prestressed concrete beams were cast and stored in marine
environment (Rance dam) (Figure 1). In 1976, these beams were moved to the Sainte
Anne de Portzic harbour (IFREMER, Brest), in a tidal zone. Finally, since 2000, 20
beams have been stored on a land site (CEBTP, Saint Remy les Chevreuse). The
dimensions of 20 beams are 2500x200x200 mm. The ends (on 250 mm) are protected by
a bituminous binder. Each beam is identified by three numbers: the first one corresponds
to the lay-out configuration (see Figure 2 for more details), the second one to the
granulometry of concrete mixture and the last one to the cement content (Table 1). Figure
2 illustrates the layout details of the beams. The beams are reinforced with passive plain
carbon steel bars (~6 mm) and 10 stirrups (~6 mm) spaced at about 250 mm. The
concrete cover is 16 or 41 mm, depending on layout configurations. For each type of
beams, 4 concrete designs are studied (Table 1): two different cement mixtures (300 and
400 kg/m 3 of concrete) and two gradings (continuous and discontinuous). The concrete is
prestressed with wires, 7 mm in diameter, placed under a plastic sheath, 12 mm in
diameter, and anchored at the beam ends. For beams of Types 1 and 2, prestressing cables
are centered along the axis of the beam. For Types 4, 6 and 9, they are in the lower part of
the beam, as shown on Figure 2. So, in these cases, the concrete prestressing reproduces
cracking pattern similar to that in a structure during a long-term exposure. Table 2
presents the initial longitudinal stresses within the concrete (in upper and lower parts of
the beam, respectively) due to presence of prestressing cables.
METHODOLOGY AND DETAILED WORK PROGRAM
This project is structured into seven steps (Figure3):
0 Step 1 - A survey of structural damage is performed from non-destructive
testing methods;
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
20 Poupard et al.
D Step 2 - The global mechanical behavior of the reinforced beams is
experimentally estimated by one partner (CEBTP). The results will be kept
secret until step 5.
D Step 3 - Complementary laboratory analysis are carried out for
characterizing passive reinforcing steels, prestressing cables, corrosion
products and concrete. These data are transmitted as input data to the
modeling teams;
D Step 4 - Modeling teams apply their mechanical models to the global and
local mechanical behavior of the beams, the input data are the experimental
results obtained during steps 1 and 3;
D Step 5 - Numerical simulations results (Step 4) are compared to the
corresponding experimental results (Step 2).
Conclusions are focused on the validity of the different mechanical models, and on the
reliability of the non destructive· methods. Each step is detailed below (partners are
indicated in bold letters).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Visual inspection
First, all the beams are visually observed. It aims to survey the beams in order to detect
which type of structural damage is being developed in the structures. Regarding
reinforced corrosion problems, three types of visual inspection results are given:
• Maps of cracking positions: cracks due to reinforcement corrosion are drawn
taking into account the position of the visible cracks and their widths,
• Corrosion maps: the locations of the rusted areas on the beams are drawn,
• Information on spalling, delamination and loss of concrete cover are also
reported.
Reinforcement geometry
The reinforcement details is obtained from magnetic measurements 1 to control:
• concrete cover thickness,
• rebar position : transverse and longitudinal reinforcing steel bars.
Electrochemical methods
Electrochemical measurements are performed on the beams to evaluate in a non-
destructive way the rebar corrosion activity. First, half-cell potentials are measured. The
principle of this method is based on the measurement of the potential difference between
rebars and a reference electrode. A multi-electrode device 2 is used to collect data (three
measurement lines per side and one reading per em along the beam length). The reference
electrode is a copper/saturated copper sulfate (CSE = +318mV /NHE) electrode.
Measurements are performed according to RILEM recommendations (9).
1
Ferroscan®, Pachometer
2
Canin®
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
-`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 21
The measurement of half-cell potential is a qualitative way to mainly locate corroding. To
characterize in a quantitative way the corrosion activity, it is essential to evaluate the
steel corrosion rate. The corrosion rate, v com is deduced from the polarization resistance
(Rp) measurement (10). The current density, !carr' is then obtained from Stem and Geary
equation (Eq. 1) (11):
I
corr
=!!_
R (Eq. 1)
p
with B a constant parameter which varies from 13 to 52mV depending on the studied
system. In this study, B is considered as equal to 26mV (if Rp and Icorr are expressed
respectively in k.Q and 11A), according to Rilem recommendations (10).
The corrosion rate can be deduced from corrosion current from Faraday equation. If
corrosion current is expressed in 11A, corrosion rate (in llm.year- 1) is given by the relation
(Eq. 2) assuming uniform corrosion across the rebar surfaces:
1
vcorr = 11.6 ";' (Eq. 2)
where A is the steel surface area (in cm2 ).
Table 3 summarizes the different techniques for corrosion rate and resistivity
measurements used in this project.
3
A SIR-2000 GPR developed by Geophysical Survey System Inc. (GSSI®). SIR-2000 is
a GPR data acquisition system and the GSSI 1.5 GHz antennas provide high quality
waveforms allowing simpler signal processing.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
22 Poupard et al.
Characterization of the constitutive materials (reinforcing steel, prestressed cables,
corrosion products and concrete) (Step 3)
After the non-destructive inspection of the beams, 10 beams are used to characterize the
materials properties and to estimate the aggressive species penetration (chloride front and
carbonation depth). The carbonation depths are measured by a phenolphthalein test. For
this purpose, freshly broken concrete surface are sprayed with a phenolphthalein solution
at various locations along the beam. Sound concrete appears red while the color of the
carbonated concrete remains unchanged. This change occurs in the pH range 8.3-10.
Powdered concrete samples are taken for determining contents of "total" and "free"
chloride and estimating chloride penetration profiles at different locations on the beams
according AFPC-AFREM recommendations (15).
Samples are collected for each concrete type to characterize the material microstructure
features (porosity, pore size) which are in strong relationship with transport properties.
MIP tests and water absorption measurements are performed.
Cores are also taken from specimens to assess the mechanical properties of concrete at 40
years (compressive strength, elastic modulus and tensile strength).
Mechanical measurements are also performed to estimate the mechanical properties of
prestressing cables and passive reinforcements. Tests are also performed to evaluate the
residual prestressed value within beams
The nature of corrosion products is well identified, with scanning electron microscope
observations, X Ray diffractometry analysis and J.l-Raman method. The localized
corrosion (pits) areas and depth are also measured, as well as the section loss due to
generalized corrosion. Finally, the weight loss of the steel is also determined.
Tensile test
Figure 4 shows a detail of instrumentation for tensile tests and a photograph of the system
layout. This test is only performed on beams type 1 and 2 (where prestressing cables are
centered in beam). The experimental results are used for determining the global and local
behavior of the beams. For the global behavior, the load-displacement curve is recorded
and the ultimate tensile strength is determined. For the local behavior, the concrete local
strains (curvatures and mean deformations in concrete) in non-cracked and cracked cross-
sections are determined.
At the end of these mechanical tests, maps are drawn showing the locations of cracks.
The results from this task are being kept secret, and they will be disclosed only at the end
of the project for a comparison with simulation model results.
Visual observations
Before removing concrete cover, a visual observation gives a global overview of the
corrosion state of prestressed concrete beams after 40 years in marine environment.
Figure 6 presents crack maps for beams 122 and 622.
After removing concrete cover, the actual corrosion damage of each rebar is observed. As
expected in chloride contaminated concrete, corrosion occurs in localized areas (pitting
attack). For beams type 1 and 2, similar corrosion damage is observed between upper and
lower rebars. This can easily be explained by the fact that similar loads are applied in
concrete (in upper and part). For beams type 4, 6 and 9, the details of corroded areas
show corrosion initiation along the tensile rebars (upper part). Except for some cases,
rebars under compression (lower part) in these beams are weakly corroded. It is observed
that the highest corrosion risk is in the upper face of type 6 beams with a thin concrete
cover of 16mm and prestressing cables highly eccentric. This result agrees with the paper
by Francois and Arliguie (17). Microcracks in tensioned concrete enhance chloride and
oxygen penetration into concrete and damage at steel/concrete interface.
There are large differences in corrosion behaviour depending of concrete mixture. For a
given layout configuration, concrete of type 1.2 (continuous granulometry, cement
content 400 kg per m3 of concrete) always shows the lowest percentage of corroded
areas. Conversely, beams with low cement content (300kg. per m 3 of concrete) show the
worst behaviour towards corrosion risk. This is in good agreement with previous studies
that have shown that the W/C has a major role on the corrosion process within reinforced
concrete (18, 19). When this parameter increases, porosity also increases (see Table 1).
So aggressive species (chloride, oxygen) can more easily penetrate into concrete and
develop high-level corrosion at a rebar surface.
After this global synthesis, remarks can be made on several parameters influencing
corrosion rates. Figure 8 presents the corrosion rates values measured for the beams type
4 and 6 (prestressed on the lower part of the beam, and concrete cover of 16mm on the 4
faces). Figure 9a shows the results for the concrete beams containing 300 kg cement per
m3 of concrete, whereas the Figure 9b presents results obtained for the concrete
containing 400 kg cement per m3 of concrete.
For the two cases, graphs point out an important scatter of corrosion activity in the
considered concrete zone. The measured corrosion rate is globally more important in the
points located on the upper face (tensile area) than in points located on the lower part of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the beam (lower face, compressive area). So, the corrosion rates are different in tensioned
and in compressed -faces of a beam, but this is not the case for half-cell potential values.
These results totally agree with the visual observation presented previously.
Figure 9 presents the influence of concrete design (cement content) on the corrosion
activity. Figure 9(a) shows the corrosion rates and half-cell potential results of types 1
and 2 beams centrally prestressed, concrete cover 4lmm), whereas Figure 9(b) presents
these values for the types 4 and 6 beams (prestressed in the lower part, concrete cover
16mm). Moreover, for beams prestressed in their lower part, only measurements of the
upper face are reported (tensioned areas).
The data of the beams type 1 and 2 (Figure 9(a)) clearly show that:
• Concrete with high content cement (400 kg) and a continue granulometry
presents the lowest corrosion activities: these conclusions are drawn from
corrosion rates values and from half-cell potentials measurements.
• Concretes with low content cement (300 kg) present highest corrosion rates and
most negative half-cell potentials.
A similar analysis performed on type 4 and 6 beams (Figure 9(b)) does not point out any
effect of the concrete design on corrosion activities. There, the cement content and the
granulometry are not the determining parameters on the corrosion behaviour of beams.
FigurelO illustrates the correlation between these factors and the percentage of corroded
areas. On these graphs are distinguished values estimated on lower and upper sides for
each beam.
Except for "crack" factor, a good accordance is observed between corrosion diagnosis --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
from NDT factors and corroding rebar area. As expected, correlation is observed for
electrochemical parameters (gradient potential and corrosion rate). Concerning radar
results, a correlation is found between the DW amplitude value and the corroded areas at
rebar surface for a beam side. That is in good agreement with previous laboratory results
(15): DW amplitude decreases with increasing percentage of corroded areas. The
limitations of "crack" factor is easily explained: longitudinal cracks enhances chloride
penetration at rebar surface. For types 1 and 2 beams, the concrete cover is thicker and
chloride penetration is not sufficient to induce crack damage at surface concrete. No
crack appears before that chloride content at steel surface reaches a threshold. It is
interesting to notice that in the case of a result non-expected, NDT indicators confirm
actual corrosion risk. For example, due to its concrete design (low W/C), beam 412
should show a good corrosion behaviour. However a high corrosion level is visually
observed in upper part of this beam compared to the others beams type 4. NDT indicators
values measured on the upper face of this beam are in good agreement with the real
corrosion state (high potential gradient and corrosion rate values and low DW amplitude).
Finally, in the case of beams of type 4, 6 and 9, NDT factors confirm the difference of
activity, for a same beam, in the upper and lower parts. As confirmed by visual
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 27
observations after removing concrete, higher corrosion activities are observed for these
beams in tensile rebars (upper part).
CONCLUDING REMARKS
The French national project "Benchmark des poutres de la Rance" has been proposed to
evaluate the influence of corrosion on mechanical behavior of reinforced concrete
structures after a long-term exposure in marine environment. At this time, only results
from Tasks 1 and 3 are presented, because the project is not yet fully achieved. The final
results and a scientific report will be published at the end of 2005. The information about
these beams will be structured as follows:
A database characterizing their initial state.
A database characterizing their corrosion state after 40 years in a marine
environment (mechanical, electrochemical and physico-chemical properties).
The comparison between the corrosion diagnosis performed from non
destructive testing tools and actual corrosion damage (destructive inspection).
The comparison between numerical simulations (from various mechanical
models) and the experimentally observed behavior.
Concerning the corrosion diagnosis on 16 prestressed concrete beams after 40 years
exposure in marine environment, the results are summarized, as follows:
physico-chemical measurements have confirmed that chloride ions are the only
responsible of corrosion attack.
crack maps gives a global overview of the corrosion area (but only when the
corrosion process reaches a critical issue).the tensile/compressive stresses within
concrete structure have a significant effect on corrosion of passive reinforcing
steel. Conclusion on an "active" corrosion in the tensile part of the beam and a
"low" corrosion in the compressive part can be made with confidence.
the thickness of concrete cover and W/C are important factors determining the
ability of aggressive species (chloride and oxygen in our study case) to reach the
rebar surface and to start the corrosion process.
the NDT indicators proposed in this study are well-correlated with the real
corrosion state of rebars within concrete structures and will have to be of a first
relevance to establish a diagnostic of corrosion.
These initial results are encouraging to support confidence in such NDT tools (radar,
half-cell potential (using spatial gradient values) and corrosion rate measurements) to
control and evaluate in a semi quantitative way (for electrochemical indicators at least)
the corrosion risk in a real reinforced-concrete structure.
ACKNOWLEDGMENTS
The authors acknowledge the French Ministry of National Education, Research and
Technology for its financial support. The study is part of a RGCU (Urban and Civil
Engineering Network) program, "Benchmark des poutres de La Rance". The authors are
grateful to all the participants of this project.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
10. RILEM TC 154-EMC, "Test methods for on-site corrosion rate measurement of
steel reinforcement in concrete by means of the polarization resistance method";
Materials and Structures, vol.37, 2004, pp. 623-643.
12. Clemena G.G., "Short-pulse radar methods", in: Malhotra, Carino, eds. CRC
Handbook of non destructive testing of concrete: CRC Press, 1991.
13. Daniels D.J., "Surface-penetrating radar", in: London, UK: Institute of Electrical
Engineers, 1996.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 29
14. Laurens S. Aptitude de la technique radar a la caracterisation du beton
d'enrobage-Aide au diagnostic de la corrosion des armatures, Doctoral Thesis.
INSA Toulouse (FR), Sherbrooke University (CA); 2001.
17. Fran<;ois R. and Arliguie G., "Effect of microcracking and cracking on the
development of corrosion in reinforced concrete members"; Magazine of
Concrete Research, vol.51, 1999, pp. 143-150.
19. Schiess! P. and Raupach M., "Laboratory studies and calculations on the
influencer of crack width on chloride-induced corrosion in concrete"; ACI
Materials, vol.94, No. I, 1997, pp.56-62.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
20. "Standard Test Method for Half-Cell Potentials of Uncoated Reinforcing Steel in
Concrete", ASTM C 876-91, 1991.
"
1
SP institute Galvanostatic pulse Rapicor®
f LREP Gecor6®
! SP institute Rapicor®
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
....,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Crack factor'
Potential facto" Potential gradient ractor"
(in%) {in%)
Corrosion rate
{in ~.year' I
RADAR factor (arbitrary
unity)
Percentage of corroded
areas
(in%)
.~
"'I"
Designation beam
Typa1
In uppar
face
In lower
face
In uppar
face
In lower
face
In upper
face
In lower
face
In upper
faca
In lower
face
In uppar
face
In lower
face
In upper
face
In lower
face f
...:!.
121 0.0 0.0 99.7 99.3 11.0 9.6 6.5 8742 9124 39.3 13.6 !.
~
122 2.4 0.0 100.0 100.0 3.8 4.0 3.0 7181 8267 31.6 13.3
Typa2 211 0.0 0.0 99.5 100.0 17.9 5.8 2.9 7708 7808 30.3 28.9 a-
~
212 0.0 0.0 6.5 24.7 9.5 17.4 1.5 10828 9894 8.1 5.5
Typa4 -~
3~
411 9.3 0.0 84.4 100.0 15.9 5.0 2.5 1.3 8042 8185 28.3 22.1
412 21.4 0.0 71.0 76.8 30.3 3.3 8.5 2.1 8112 10129 36.9 2.0
:.-l
421 0.0 1.6 94.2 98.3 20.9 8.8 8.5 2.1 8442 9240 ~ S'
=- .....
422 13.5 12.0 77.6 100.0 35.3 4.5 12.8 3.2 7183 8167 16.6 12.0 ~ ~
I» Q
Typa6 ...
811 10.1 0.0 75.8 90.7 44.1 5.5 17.1 2.4 8179 9164 84.4 6.3
;;: g, c
C"r»
612 31.2 0.0 84.3 32.5 15.6 14.8 6.0 0.9 9556 9855 39.9 2.1
621 7.9 0.0 84.7 95.4 24.6 11.1 11.3 4.7 7349 8810 55.2 9.2 e, ;c
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
622 52.9 0.0 69.0 14.3 20.5 5.8 5.6 4.6 8723 9329 50.0 14.6 ~
Typa9 911 21.9 16.3 100.0 96.2 8.7 9.3 7.8 6.2 8583 9268 24.2 ~ E:
912
921
0.0
0.0
0
17.9
0.0
99.3
0.0
99.8
5.0
5.4
2.5
7.2
3.8
3.5
4.4
0.6
9589
8601
9998
9132
5.9
9.4 ~ ~
g. ....,
922 88.5 0 80.4 50.7 16.5 4.2 8.2 0.9 7948 8957 36.9 ~ 0
N
L length (i) X width (i) where N is the number of cracks existing on the considered face ot the beam. ..
I» g
• data not yet available crack filctor =
=
1
s n
-
i-1
• no passive rebars in lower face
3
percentage of potential values lesser than the critical threshold -350 mV/CSE (according ASTM)
percentage ot gradient potential values higher to a cr111cal threshold. In this study, a critical threshold of 8 mV/cm will be considered.
i& ~
tD
Daneshlink.com
w
....
32 Poupard et at.
~~[bd
s~:_;:=oJ~ LEiJ~Dm
Prestressed
e9
wi='12mm
Fig. 2- Schematic view of the layout details for each beam type
Step 1
Non destructive c:hcracterizations of beams
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Discussions
Comparison Simulations I Experiment
Reliability of NDT results
..
F
measurement Bituminous
Strain area
gauges
(b)
Fig. 4- Tensile test (a) schematic view of the instrumentation (b) experimental setup
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Strain i !
gauges Ii i!
extenso meter
Deflection
measurement
(ii) View according 2•d lateral side
(b)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
B
.
I
.
I I
.
I
B . . . . . . . . ! . .
. -.-+.- ·!-.- !-. -!·-.-!-.- ·!-. -~-. -!·- ·!-. -!-. -!·-.-
L
H!
!
I
. -.!.- ~-- ;~-·
I..........
!~!...I-!
I
j".-
.........
I
•i I I
"Si-. !"""'"!.
! ....
!
I
~-
I I
",:j;-.-
· - · "T...,... r- · - 1- • --. • ..,..,.... ~ ··- • - r - · ., · - · ~ · - , - · .,.. - · -
R ! ~..,._ ~ i i'-'~ i ~ ! ! !
0
Fig. 6- Some examples of corrosion-induced crack maps (a) Beam 122 (b) Beam 622
10r---
X Type 1 +Type2 ~Type4 I:. TypeS DType9
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
High
I
J
Low D D
+
0.1 ~~~--~--------
1:. D
+ D
Negligeable D
0.01
0 -100 -200 -300 -400 -500 -600
Half-cell potential (mV/CSE)
. a a
"E
~
. .. .... ..
.a, a .t
J 0.1
.. " "~ "a B
a
0.01
-200 -250 -300 -350 -400 -450 -500
Half-cell potential (mV/CSE)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
(b)
10 .& Tensile area a Compressive area
.. .
. ..
I
.a, all
'i.
.........
.. aa
a a
J 0.1 "
"
0.01
-200 -250 -300 -350 -400 -450 -500
Half-cell potential (mVICSE)
!
:;_
1, A D aa a
:lb: a
U A t "S adb a A
·- 0.1 t-"'-------------A-----,ora.-------
A
0.01 +--~--~--~--~-~--~-~
-200 -250 -350 -400 -450 -500 -650
Ha~-cell poten~al (mV/CSE)
(b)
3 3
10 a 300 kglm A400 kglm
I •
~ 0.1 t - - - - - - - - - - - - - - - - - -
a A a a
0.01 +--~-~--~-~--~-~-~-~
-200 -250 -300 -350 -400 -450 -500 -550 -600
Hall-cell potential (mVICSE)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~ so+-~----------------------------
0 D
;;80
.§. e
I 60 c
c
~ 80 D
D
i 40~----~0
=--a------------------=-- ~ ~+---------"c"c--~--~-----------
~
8
I" c
20~--~~c_________________________ ~ 20t-----":~~--aJLv~C~----------------
XC >f(DX':,Xa~ a
o+---~----~----~~x~L-------~
0~----~----~----~------~----~
6000 7000 8000 9000 10000 11000 12000
0 20 40 60 60 100 Radar factor (in a.u.)
crack factor (in arbitrary unit (a.u.))
(d) 100 DUpper X LONer
I 60
a a
m 60 I m D
c
"il 40 .:.--------lli---<I:I~-----------------
"!!
"8
40
Dx
.., c
1! li r::IC § cC
.,; c
~20\ X D D
20
X D
o ~ xXJC' ~ x >si eX X
10 15 20 10 20 30 40 50
1
corrosion rate (In 11m.year" ) potentlal gradient factor (in%)
Fig. to- Comparison between corroded area of rebars and NOT factors
(a) crack factor (b) radar factor (c) corrosion rate (d) potential gradient factor
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Daisuke Mori is a research scientist of Cement Chemistry Team, Research & Development Center,
Taiheiyo Cement Corp., Japan, Ms. Eng.
Bjorn Johanneson is a research scientist, Cement Chemistry Team, Research & Development
Center, Taiheiyo Cement Corp., Chiba, Japan (now in Lund University of Technology, Sweden),
Ph.D. He has interests in the theoretical aspect of elements movement. He is trying to establish a
physically stringent model to estimate material transport.
INTRODUCTION
A major method for performance verification of reinforced concrete structures in
chloride bearing environments is the prediction of chloride ion (Cr) concentration at the depth of
reinforcement at a certain age by using Fick's second law as shown in Eq. 1 with an apparent
diffusion coefficient Da·
where,
C(x, t) = Cr Concentration at depth X and time (
c, = surface cr concentration
C; = initial cr concentration,
erfc = error function.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
In this study, a new method that involves estimating De and non-linear binding
parameters by reverse analysis is developed and the validity of this method is examined by
comparison with experimental results and information from previous studies.
EXPERIMENTS
Factors and mixture prooortion of concrete
Examined factors and the levels are indicated in Table 1. The cement type and w/c were
varied because they are expected to affect chloride binding and De significantly. The cement types
are normal portland cement (N), low heat portland cement (L), Japanese type ecocement (E), and
blended blast furnace slag cement type B (BB). Every cement conforms to the corresponding ns
standard. Although basic mineral composition and characteristics of Type E are similar to type N,
the essential difference is that type E must be produced by using waste material more than 50 % as
raw material. Type BB was made by replacing 40 mass% of type N with ground granulated blast
furnace slag having a Blaine value 400m2/kg. Mixture proportions of concrete are shown in Table
2. The w/c was selected as 0.30, 040 % and 050. The fine aggregate was Ogasa land sand having a
density of2.60 glcm3 and the coarse aggregate was Fujiwara crushed limestone having a density of
2.65 glcm3. The maximum aggregate size was 20 mm.
Exposure condition
After water curing, every surface of a prism specimen except the exposure surface T
the cast face) was coated by epoxy resin and immersed into 3 %-NaCl solution at 20 oC for 91
days. For cylinders, the top 2 ern was cut off and this surface was exposed to the same solution
mentioned above. Other faces were coated by epoxy resin.
Evaluation
cr concentration profile -- cr concentration profiles were measured by area analysis
using EPMA. After exposure to NaCl solution, a plate specimen having a surface parallel to the
penetrating direction of cr from the most surface of original concrete prism was cut out and the
surface was ground flat to be suitable for EPMA analysis. Sample preparation procedures,
quantification of cr concentration, and processing of the measured area data to obtain the cr
concentration profile only in cement paste by eliminating the effect of aggregates were done in
accordance with a previous study (3). The spatial resolution of measurement was 0.1 mrn.
JSCE-G571-2003 (4). The porosity of each specimen cured in water at20 oC for 28 days was
evaluated according to ASTM C 642.
ac Eq.2
at
Eq.3
Eq.4
where,
c = free cr concentration,
Cb = bound Cr concentration,
2
De= effective diffusion coefficient (m Is),
3 3
&=porosity of concrete (m f!li m cone),
a, f3= parameters obtained by fitting.
In this study, the non-linear binding behaviors of cr were assumed to have a form of
Fruendlish-type adsorption equilibrium as shown in Eq. 3. Upper case symbols indicate the mass
concentration per unit volume of concrete (kg/m3cone) and lower case symbols indicate the mass
concentration per unit volume of pore solution (kg/m3so1).
The numerical analysis used the Crank-Nicolson implicit finit-difference method (5).
Boundruy conditions are given by Eq. 5.
where,
Cs = free Cr concentration at the surface,
C~,; = initial total Cr concentration.
When SOlving Eq. 2, an initial Cr COncentration has to be given as C; (free Cr concentration).
Therefore, the following Eq. 6, obtained from Eqs. 3 and 4, is solved by using the Newton -
Raphson mehod (6) and c; is calculated from Ct,i.
Input parameters and conditions of numerical calculation - The input parameters for
the numerical calculations are set as follows. The initial condition C~,; is determined by the EPMA
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
44 Yamada et al.
measurement of each specimen. The boundary condition Cs is set as 18.8 kg/m3sol of Cr
concentration that corresponds to 3 %-NaCl solution. Exposure time t is set as 91 days. The
porosity of the concrete cwas detennined experimentally, as mentioned previously.
N
F(De,a,p) = ~)c~b (xi) -Cf (xi,De,a,p)f
1 Eq.7
where,
N =number of data,
xi= depth of measured point i,
Ct'Jb =measured values of total cr concentration,
etc/= numerical calculation results of total cr concentration.
RESULTS AND DISCUSSIONS
Fitting for Concentration Profiles
Some typical examples of fitting results for cr concentration profiles by using the
reverse analysis method are shown in Fig. 1. Hereafter, total cr concentration is converted and
expressed as mass ratio in terms of cement content. As shown in Fig. 1, good fitting results were
obtained in every case. In Fig. 1, fitting results assuming linear cr binding using Eq. 1 are also
indicated. In some cases fitting ofEq. 1 gave good results but in others unsatisfactory. As an index
for evaluating the degree of fitting, a square root of variance s is defined by Eq. 8 and the
calculation results are shown in Table I.
Eq.8
where,
m = the number of parameters estimated by reverse analysis,
The values of s are significantly different depending on the mixture. In every case
except 50E, non-linear cr binding gives smaller s, which indicates better fitting than the case of
linear cr binding. In Table 3, the binding parameter /]is also shown. Poor fitting results using Eq. 1
correspond to the cases showing estimated f3 less than about 0.6 and this indicates strong
non-linearity ofCr binding. On the other hand, the estimated /]in the case of type E are close to 1
in both W/C. This indicates that for type E, cr binding can be considered as linear.
The reason why the non-linearity ofCr binding is detected successfully as shown in Fig.
1 is the high spatial resolution of EPMA measurements. Compared with the traditional method
evaluating cr concentration profiles, EPMA has much higher resolution, on the order of 10 J.ID1
rather than 1mm as for grinding methods. For this kind of reverse analysis, more data points are
preferable for more precise estimation, and EPMA is a suitable method.
Estimation ofDe
Estimation results of De by reverse analysis are shown in Fig. 2. A higher W/C gives a
larger De, and at the same W/C, type BB gives a smaller De around 1/4 of that of type N. These
kinds of tendencies are in agreement with the results of electrical migration tests reported
previously (7). Regarding type L and E cements, although electrical migration tests have not been
accomplished, because immersion tests with type E showed faster cr penetration than with type N
(8) and type L showed similar cr penetration speed as type N (9), the estimated values in this study
are judged to be reasonable.
However, in this study, the time dependence of De, a, and fJ is not considered. De is
reported to decrease with age (2), and a and /]change with the progress of hydration as explained
before. These parameters are doubtlessly time dependent. Therefore, for the prediction of long term
cr penetration, the time dependence of these parameters must be considered. This is the subject of
future research.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUSIONS
The fitted profile of cr concentration by numerical calculation based on non-linear
chloride binding shows better fit to precisely measured data by EPMA than that by the
conventional model assuming linear binding. Therefore it is described to be useful to estimate an
effective diffusion coefficient De and non-linear binding parameters a and fJ by reverse analysis
from a cr concentration profile. In this study the estimation method is developed and the validity is
examined. The following conclusions are cited:
(1) EPMA revealed the feature of non-linearity of cr binding in the cr concentration profile that
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 47
is different from simple diffusion. By assuming a non-linearity of cr binding, excellent fit to
data are possible for various kinds of cement and w/c.
(2) The estimated De shows a comparable tendency with w/c as in previous studies, and that
shows a good linear positive correlation with the De measured by electrical migration tests.
(3) The estimated values of a and j3 show comparable tendencies as in previous studies.
Especially, it is confirmed that a has a positive linear correlation with alumina content in
concrete.
(4) By using estimated parameters from data fro immersion to 91 days, cr profiles for immersion
up to 182 days were predicted and compared with experimental results. Good agreement was
obtained.
(5) The proposed method for estimating the parameters required for cr penetration prediction,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
1. Hosokawa, Y., Mori, D., Yamada, K., A study on the estimations of chloride penetration and
service life of RC structures by using an effective diffusion coefficient by electrical migration
tests, Proceedings of the annual meeting of JCA, No. 58, pp. 116 - 117, 2004 (in Japanese)
2. Hosokawa, Y., Yamada, K., Takami, M., Sugiyama, T., A study on the chloride diffusion
coefficients evaluated by immersion tests and electrical migration tests, Proceedings of the
Japan Concrete Institute, Vol. 25, pp. 725 - 730, 2003 (in Japanese)
3. Mori, D., Hosokawa, Y., Yamada, K., Yamamoto, M., Application of EPMA for the
determination of chloride ion concentration profile in concrete, Proceedings of the Japan
Concrete Institute, Vol. 26, No. 1, pp. 867 - 872, Jul., 2004 (in Japanese)
4. Sub-Committee on Test Methods and Specifications for Concrete, Committee on Concrete,
JSCE standards ''Test method fro effective diffusion coefficient of chloride ion in concrete by
migration", JSCE Journal, No. 767 /V-64,pp.l-9, 2004 (in Japanese)
5. Yamada, Y., Oshiro, T., Masuda, Y., Analytical study on chloride penetration into concrete
exposed to salt-laden environment, J. Struct. Constr. Eng., Architectural Institute of Japan, No.
501, pp. 13 - 18, Nov., 1997 (in Japanese)
6. Press, W. H., Teukolsky, S. A., Vetterling, W. T., Flannery, B. P., Numerical Recipes inC,
Cambridge University Press, 1988
7. McGrath, P., Development of test methods for predicting chloride penetration into high
performance concrete, Ph.D Thesis, University ofToronto, 1996
8. Nagashio, Y., Yamaji, T., Hamada, H., Tochigi, T., Application of normal Eco-cement to
concrete in marine environment, Proceedings of the Japan Concrete Institute, Vol. 26, No. 1, pp.
1479- 1484, Jul., 2004 (in Japanese)
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
48 Yamada et al.
9. Hosokawa, Y., Yamada, K, Mori, D., Kim, D. S., Evaluation of chloride penetration into
concrete using electron probe micro analysis, Concrete under Severe Conditions: Environment
and loading, B. H. Oh eta/. (eds), pp. 473-480, 2004, CONSEC '04, Seoul, Korea
10. Zibara, H., Hooton, D., Yamada, K, Thomas, M. D. A, Roles of cement mineral phases in
chloride binding, Cement Science and Concrete Technology, No. 56, pp. 384 - 391, 2002
11. Tang, L., Nilsson, L. 0., Chloride Binding Capacity and Binding Isotherms of OPC Pastes
and Mortars, Cement and Concrete Research, Vol. 23, No.2, pp. 247-253, 1993
Mixture
p Square root of error variance s
(predicted) Non-linear binding Linear binding
SON 0.431 0.040 0.132
SOL 0.299 0.041 0.079
SOE 0.990 0.055 0.054
SOBB 0.501 0.021 0.077
40N 0.306 0.028 0.123
40L 0.251 0.029 0.079
40E 0.800 0.023 0.024
40BB 0.413 0.016 0.066
30N 0.603 0.008 0.032
c~
:~:D
o_
~~ 2 ' 50N
c:e
55
c~
8~1
u~
~~ ~
E;JD
1--0
0 10 20 30 0 10 20 0 10 20 30
g~
_ce E" 2 0 Meas.
-Ana. (Non-Linear) 2 2
• - -Ana. (Unear)
U·.
c
.. :
~.5.ool-..__wlllmillll2o-iliilllil3o o 10
«£
20 30
:
o 10
-
20 30
Depth [mm] Depth [mm] Depth [mm]
Fig. 1 Some examples of fitting results for concentration profiles by using the reverse
analysis method
w/c
• 0.30
.t:::. 0.40
• 0.50
2 4 6 8 10
(X 10·12]
0 8 estimated by reverse analysis[m2/s]
Fig. 4 Estimated ct· binding isotherm using a and ~obtained by reverse analysis
w/c
l:l. 0.40
• 0.50
oL-..L.....I.....L......J~~~
0 10 20 30 40 50
3
Amount of AI20a [kg/m cond
Fig. 5 Relationship between a and alumina content in concrete for each w/c
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
30 40
BB
0 oL-~Jo~•2•o.-•3•o-.•~
Depth (mm)
Fig. 6 Total n concentration profiles for 182 days of immersion measured by EPMA
and predicted with parameters estimated by reverse analysis
by H. justnes
INTRODUCTION
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Chloride induced corrosion of reinforcing steels is perhaps the most common
deterioration of reinforced concrete. The chloride intrusion in high performance concrete
1
(e.g. bridges) exposed to marine environment can in practice be much faster than
expected from laboratory trials. Concrete road structures also suffer corrosion damage
due to the use of de-icing salts. The addition of an inhibitor to the concrete mixtures for
structures prone to chloride attack is one measure that may enhance service life or time to
necessary expensive repair.
A corrosion inhibitor should (a) raise the level of chlorides necessary to initiate
corrosion or (b) decrease the rate of corrosion after it has started or (c) both. Thus an
inhibitor could delay corrosion initiation time or slow down corrosion propagation rate or
both. Since it does not necessarily prevent corrosion from happening altogether, some
prefer to refer to such compounds as corrosion retarders rather than inhibitors. The term
inhibitor will be used in this paper.
A sketch of corroding reinforcing steel in concrete is reproduced in Fig. 1. The
corrosion process consists of a cathodic half-reaction reducing oxygen to hydroxyl ions;
(1)
Ferric hydroxide in different forms may have molar volume 2-4 times that of the
corroded metallic iron. The corrosion process (i.e. oxidation) will therefore after
sufficiently propagation lead to cracking of the concrete cover and eventually spalling.
CORROSION INHIBITORS
Corrosion inhibitors may either affect the cathodic (Eq. 1) or the anodic (Eq. 2)
half-reactions, and are consequently denoted cathodic or anodic inhibitors, respectively.
The April, 2004, issue of the journal "Cement and Concrete Composites" focused on
Review -- Among anodic inhibitors, inorganic nitrites (N0 2-) are well known to be
efficient. Calcium nitrite, Ca(N02) 2, is preferred over sodium nitrite, NaN02, since high
dosages of sodium salts will lower long term concrete strength and also increase the risk
of alkali-aggregate reactions. Calcium nitrite has been used for more than 25 years as an
inhibitor against chloride induced corrosion of reinforcing steels. It was promoted first in
Japan 10 in the 1970s and has been extensively studied as a corrosion inhibitor in the last
decades (e.g. Berke and Hicks 11 ).
The anodic corrosion inhibitor effect of inorganic nitrates (N0 3-) is not as well
known as nitrite (N0 2'), since the inhibitor effect was only discovered in 1994. Note that
since there is only one letter (a vs. i) or a number (3 vs. 2) between nitrate and nitrite,
there has been some confusion in the literature. Hopefully this confusion is sorted out
here.
It is generally accepted among scientists that nitrates do not promote steel corrosion
(Alonso and Andrade 12 , D'yachenko et al. 13 and Goetz 14). Meland et al. 15 investigated the
reinforced concrete of a 22 year old calcium nitrate (containing some ammonium)
production plant where the concrete was deteriorated due to external calcium
nitrate/ammonium attack (probably since concentrated calcium nitrate will form
expansive double salt with calcium hydroxide). However, concrete with up to 1.5 % N0 3-
showed no damage. They found that the reinforcement in severely nitrate-polluted areas
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
was in a perfect condition with only surface rust and no signs of brittleness. Holm 16 noted
an inhibiting effect of calcium nitrate with respect to steel corrosion, but subscribed the
effect to other minor components in this commercial admixture. Vogelsang 17 proved that
calcium nitr~te and nitrite had identical corrosion inhibiting properties by
cyclovoltametry, and he explained this inhibiting effect by the precipitation of a calcium
hydroxide layer in the alkaline solution (i.e. this would only occur in an alkaline
environment like concrete).
Alonso and Andrade 18 have related the values of corrosion current, icorr, to the
corrosion state of steel in concrete. icorr < 1-2 mA/m 2 indicates negligible corrosion, while
greater values mean active corrosion. In terms of service life, icorr > 10 mA/m2 indicates
high corrosion rate, while icorr > 100 mA/m 2 is severe corrosion rate. Justnes and
Nygaard showed by long-term macro cell studies that nitr~tes in an alkaline
19
The reduction of nitrite may proceed all the way to nitrogen because the reduction
potential of N 20 is so high in an acidic environment (the reduction potential of N 20 in
alkaline solution is not given in the electrochemical series). Rosenberg and Gaidis 25
confirmed the reduction mechanism for the nitrite inhibitor, but claimed that the
reduction does not proceed further than to NO (Eqs. 7 and 5 combine to 8), perhaps
because of kinetic reasons;
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
If the proposed mechanism (Rosenberg and Gaidis 25 ) for nitrite inhibition of steel
corrosion in concrete summed up in Eq. 8 is correct, Eq. 10 implies that nitrate, N0 3-,
should be an even better inhibitor than nitrite, N0 2-, because three more moles of ferrous
iron are oxidised to ferric per mole nitrogen oxide added (i.e. nitrate is three times as
effective at an equimolar dosage). Unlike nitrite, nitrate does not appear to work when
tested in a rapid solution test (e.g. Brown et al. 26 ). Even though reaction 10 has a greater
driving force (i.e. potential) than reaction 8, the kinetics may be slower and a longer-term
test is required. However, corrosion in practice is a slow process so this does not matter
in a real structure. Another hypothesis is that nitrite acts as an oxygen scavenger since
N0 2- may be oxidised in an alkaline environment, as seen from the two half-reactions Eq.
11 and the reverse ofEq. 9;
E0 = +0.39 V (12)
However, this latter mechanism was proven (Justnes et al. 27 ) to only take place in
acidic environment.
Visual inspection of reinforcing steels embedded in concrete with anodic inhibitor--
Concrete cylinders of 100 mm diameter and 200 mm height were cast (11-12 September
1996) with a 20 mm reinforcing steel in the centre, leading to an effective concrete cover
of 40 mm. The concrete consisted of ordinary portland cement (CEM 1-42,5-R), tap
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
water, natural sand and crushed gravel and had water-cementitious materials ratio (w/cm)
= 0.54. There were two concrete series consisting each of three mixtures with 0, 2 and 4
% calcium nitrate. One series was cured 56 days in water before exposure to 3 cycles of
drying at 105°C for one week and suction of 5 % NaCl for one week, while another series
was intermixed with 3.2 % sodium chloride (NaCl) of the cement weight. The concrete
cylinders were stored at 38°C and 90 % RH with air access until inspection for surface
28
cracks. Justnes reported the results from these series, but some main points are repeated
here.
After 3 years of storage, one reference cylinder from the series with intermixed
chlorides cracked because of severe corrosion of the steel and the cylinder could be
ripped open by hand. Cylinders from this series with 2 and 4 % Ca(N03) 2 were then
broken (set-up as a concrete splitting tensile test), and the exposed steel removed. The
reinforcing steels removed from these three cylinders are compared in Fig. 2. The
From tables 3 and 4 it can be seen that although 28 day compressive strengths vary
from 44.7 (mixture 7) to 67.9 MPa (mixture 13), the capillary porosity only varies from
9.1 (mixture 4) to 10.1 vol% (mixtures 8, 12 and 14) for the 14 mixes. The relative low
variation in capillary porosity is comforting because then it can be assumed that the
permeability will be comparable and the true difference in chemical environment on
corrosion can be evaluated.
The 24 cylinders of the 7 mixtures without intermixed chlorides (1, 2, 3, 7, 8, 9 and
13) were grouped in 2: 12 cylinders were stored in open boxes in the marine tidal zone so
they were totally submersed and exposed for air 2 times every 24 h, while 12 cylinders
got chlorides intruded by 3 drying and wetting cycles with seawater followed by storage
at 38°C and 90% RH with air access.
The 24 cylinders of the 7 mixtures (4, 5, 6, 10, 11, 12 and 14) that had intermixed
3.2 % NaCl of cement weight were also grouped in 2: 12 cylinders were exposed for the
changing weather of Trondheim, Norway, on the roof top, while the 12 other cylinders
was stored at 38°C and 90% RH with air access. In the reference on the roof top it was
observed that minor cracks caused by corroding reinforcing steel had appeared after 3
years. The cylinder was split open and a picture of the reinforcing steel is shown in Fig.
5. This reinforcing steel had actually spots of "green rust" (ferrous hydroxychloro
complex) that oxidised to ordinary ferric red rust after 3 days in laboratory air.
The test programme involves casting of instrumented concrete wall elements
(dimensions 1550x1350x250 mm = 523,125,000 mm 3 = 523 L) with two layers of
reinforcement (20 and 50 mm concrete cover). The elements were placed in the tidal zone
of the sea just a few days after casting to mimic real life and to get some chloride
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 61
intrusion while the concrete was still young. Self-compacting plain concrete mixtures
(600 L) were made with w/cm = 0.50. The three mixtures were a reference without
inhibitor, with 4% calcium nitrate, and with a molar equivalent dosage of commercial
calcium nitrite. The concrete properties like compressive strength ( crc), density (p ),
specific resistivity (Pr) and parameters from fitti.ng chloride profiles after 16.5% NaCl
exposure to Fick's 2"ct law of diffusion; surface concentration (Co, ct) and diffusion
coefficient (Dct), are all listed in Table 5.
The reinforcement grids and installed instrumentation are depicted in Fig. 6. There
are two kinds of electrodes (ERE20 and MMO) placed in three different levels (bottom,
middle and top) to monitor potential differences. In addition, a special device is installed
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
in every element to monitor chloride induced corrosion at different depths. The
reinforcement nets are placed in a manner that 100-mm diameter cores can be drilled out
and still maintain maximum concrete cover. This will facilitate the possibility of making
periodical chloride ingress profile and nitrate/nitrite leaching profiles. Chloride profiles in
the concrete elements after 2 years exposure in the tidal zone are plotted in Fig. 7. The
chloride content at 20 mm (about 0.25% of concrete mass) is sufficient to initiate
corrosion (limit ~ 0.1%) for the reinforcement net with this cover, but not for the net with
50 mm cover (level~ 0.01% Chloride of concrete mass).
The ERE20 electrode is developed by the Danish FORCE Institute and is based on a
sintered manganese oxide (Mn0 2) ring in contact with an alkaline solution (NaOH).
Typical values measured in saturated calcium hydroxide (pH = 12.55) at 23°C are+ 160
mV versus saturated calomel (SCE) and + 105 mV vs. saturated copper sulphate electrode
(CSE). The ERE20 electrode has been tested for the temperature range -10 to +40°C with
excellent results and withstands periods of frozen conditions. The expected service life is
more than 20 years. A set of simpler MMO (mixed metal oxide on titanium metal)
electrodes was mounted as a backup in case something goes wrong with the ERE20
electrodes. The MMO electrodes are simply pieces of mesh used for cathodic protection
systems in concrete.
The devices for monitoring chloride ingress have thinner reinforcement steels of
different length and their tips will be located at different distances from the concrete
surface when it is cast into the element. Their potentials and current passing will be
measured relative to the thicker reinforcement bar that will be laying parallel to the
concrete surface and furthest away from it. In this way intruding chlorides initiating
corrosion will be spotted and the time to reach this depth recorded for the reference
concrete element, while the inhibitors will delay corrosion at the same time.
The wall elements are placed in the marine tidal zone of the Trondheim fjord. All
the wires from the electrodes and devices are gathered in a tight coupling box at the top.
This simplifies the measurements at low tide with a simple multimeter. Thus, all
measurements are discrete in time, but the system is more robust against harsh weather in
this way.
The potential developments as a function of time for the ERE20 electrodes versus
the reinforcement grid with 20 mm concrete cover are plotted in Fig. 8. The general
trends are that the potential decreases as a function of time for both grids, and somewhat
steeper for the grid with highest concrete cover. Note that the electrodes are closest to the
grid with least cover, so the IR drop is less. The general downward trend in potential has
two reasons: I) the densification of the concrete matrix due to increased hydration and
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
62 justnes
thereby increasing IR drop (steeper drop due to larger distance in the 50 mm cover case).
2) Generally increasing temperature in the first period from late autumn/early winter
2001 until late summer 2002. The last data point indicates that corrosion has started for
the net with 20 mm cover in concrete without inhibitor, which is confirmed by the high
chloride content (about 0.25% cr of concrete mass) for all 3 elements. This means also
that the inhibitors are working.
Al-Amoudi et al. 30 compared also the performance of calcium nitrate and nitrite as
inhibitors against corrosion of steel embedded in concrete. They found that nitrate and
nitrite performed equally well against chloride induced corrosion.
Note that nitrates are known to lead to stress corrosion cracking in high strength
steels in other applications than concrete, which is why nitrates are prohibited in injection
grout for cable ducts and should not be used in pre-stressed concrete elements. However,
in addition to require reinforcing steel of low carbon steel (< 0.1% C), the rate of stress
corrosion is strongly pH dependent, and the corrosion rate is reduced to 25 % when pH 2
9.5, according to Niimberger31 • Based on his measurements, Niimberger31 concluded with
respect to working conditions that it may be assumed that stress corrosion cracking is
only able to occur in situations with extreme constructional conditions (limited cover)
and high wear due to intensive chemical exposure. Thus, the process of stress corrosion is
not believed to take place in sound concrete with pH 12.5-13.5. Furthermore, stress
corrosion will not occur if there is an active surface corrosion at a low rate. In spite of
this, the three bends at the top of each element (see Fig. 6) were placed there in order to
be able to drill them out at a later stage and inspect them for stress corrosion cracks.
CONCLUSIONS
10. Soeda, K. and Ichimura, T., 2003, "Present State of Corrosion Inhibitors in Japan",
Cement and Concrete Composites, Vol. 25, No. 1, pp. 117-122.
11. Berke, N.S. and Hicks, M.C., 2004, "Predicting Long-term Durability of Steel
Reinforced Concrete with Calcium Nitrite Corrosion Inhibitor", Cement and
Concrete Composites, Vo. 26, No.3, pp. 191-198.
12. Alonso, C. and Andrade, C., "Estudio de la effiacacia del Nit-Cal como inhibidor de
la corrosion de armaduras del hormig6n". Report no. 16.878, Instituto de ciencias
de la construccion Eduardo Torroja, Madrid, Spain, 8th September, 13 pp (Spanish
title, but English content available).
13. D'yachenko, Yu.K., Enisherlova, S.G., Kucheryaeva, G.S., Ratinov, V.B. and
Semenova, S.D., 1987, "Influence of Chloride, Nitrate and Nitrite on the Corrosion
of Pre-stressed Reinforcing in Heavy Concrete" Moscow Automobile Road
Institute, Translated from Zashchita Metallov, Vol. 23, No 4, July-August, 1987, pp
655-659.
Mix 1 2 3 4 5 6 7 8 9 10 11 12 13 14
A-Ca(N03}z 0 3 4 0 3 4 0 0 0 0 0 0 0 0
K-Ca(N0 3) 2 0 0 0 0 0 0 0 3 4 0 3 4 0 0
Ca(NOz)z 0 0 0 0 0 0 0 0 0 0 0 0 2.7 2.7
NaNOz 0 0 0 0 0 0 2 0 0 2 0 0 0 0
NaCl 0 0 0 3.2 3.2 3.2 0 0 0 3.2 3.2 3.2 0 3.2
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Mix 1 2 3 4 5 6 7
Slump(mm) 170 180 180 180 180 190 >140
Air(vol%) 3.7 3.2 4.5 6.1 3.1 4.7 6.5
3 2,390 2,390 2,320 2,395 2,355 2,320
Pr(kgfm ) 2,360
cr. (MPa) - - - - - 27.7
cr,.zs (MPa) 53.3 55.0 51.9 48.9 56.0 52.3 44.7
o.:..Jvol%) 9.9 9.7 9.7 9.1 9.6 9.6 9.9
~(vo1%) 1.9 2.1 2.4 2.6 1.8 1.9 2.6
3
P• {kg{m ) 2,184 2,187 2,177 2,187 2,201 2,195 2,177
Mix 8 9 10 11 12 13 14
S1ump(mm) 180 180 180 190 190 180 165
air(vo1%) 3.1 4.9 6.6 4.8 4.8 4.2 5.0
Pr(kWm 3) 2,395 2,355 2,320 2,365 2,360 2,375 2,355
cr (MPa) 24.2 22.4 24.8 27.4 24.5 20.0 27.4
cr,2s (MPa) 54.5 57.3 47.3 54.5 55.2 67.9 58.8
&.:ao (vo1%) 10.1 10.0 9.8 10.0 10.1 9.3 10.1
~:m.. (vo1%) 1.4 2.1 2.7 1.9 2.2 2.3 1.4
_,o,Ji<g/ml) 2,201 2,178 2,166 2,181 2,174 2,191 2,167
Fig. 1- Sketch of the corrosion process on the surface of steel. (a) Reactions at
anodic and cathodic sites and electric current loop. (b) Flow of electrical charge in the
electric current loop during the corrosion process. See equations 1 and 2 in text for
full cathodic and anodic half-reactions.
Fig. 2- Reinforcing steels removed from concrete cylinders with 3.2% NaCl intermixed
after 3 years storage at 38 oc
and 90% RH. Upper reinforcing steel from reference,
middle reinforcing steel from concrete with 2% Ca(N0) 2 and lower reinforcing steel
from concrete with 4% Ca(N0) 2 •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 3 -Steels removed from the two reference concrete cylinders (2 at top) from
the series with intruded chlorides, cylinder with 2% Ca(N0) 2 (no. 3 from top) and
cylinder with 4% Ca(N0) 2 (bottom) after 4 years in storage at 38 °C and 90% RH.
0 20 40 60 so 100 120
Time (days)
Fig. 4- Currents in the circuit of the "lollipop" test of concrete without admixtures
(reference) and with equal molar addition of calcium nitrate and calcium nitrite.
Fig. 5 ·Corroded steel (lower part) and imprint in concrete (upper part) for reference
concrete with intermixed 3.2% NaCl stored at roof top for 3 years.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 6 - Reinforcement and placement of electrodes (next to white stickers) for the
wall elements. The device to monitor chloride induced corrosion at different depths is
mounted at the upper left corner. The 3 bends close to the top can be drilled out later
to inspect for potential stress corrosion cracking.
0.0
5 10 15 20
Depth (mm)
Fig. 7 ·Chloride profiles for the concrete wall elements (w/cm = o.so) exposed in the
tidal zone for 2 years after they were placed there at 3 days of age.
Fig. 8- The potential development for the average of the 3 ERE2o electrodes versus
the reinforcement grid with 20 mm cover as a function oftime.
Mr Finnegan is a graduate student of the School of the Built Environment at Heriot Watt
University.
Mr Linfoot is Senior Lecturer in the School of the Built Environment at Heriot Watt
University with specialist interest in mathematical modelling ..
Dr Chrisp is Senior Lecturer in the School of the Built Environment at Heriot Watt
University. His research interests are in the repair and rehabilitation of structures and the
performance of concrete.
INTRODUCTION
It is an inescapable fact that all structures deteriorate with time, although the rate
at which they do so depends on many factors. Reinforced concrete is no exception. The
initiation and propagation stages of reinforced concrete deterioration processes result
from a complex interaction of physical, chemical and electrochemical phenomena.
Prediction of the field performance of reinforced concrete requires numerous data inputs,
in particular, the response of the cover-zone (covercrete) to the changing micro-
environment in the vicinity of a specific structural element or part of a structure.
Covercrete transport mechanisms, their variation through the cover-zone, drying
characteristics and the temporal change in these properties due to on-going hydration and
pozzolanic activity, all influence reinforced concrete durability.
water) is considered with the environment (e.g. tidal, splash and spray zones); in all, 18
exposure classes are defined in EN206. British Standards BS811 0-1:1997 2 and BS5328-
1: 1997 3 define exposure conditions in terms of five classes ranging in severity from mild
to most severe (with a further class for abrasion). The limitations of this necessarily
restrictive classification system is recognised with the recent withdrawal of BS5328 and
introduction of BS8500-l :2002-1 4, which adopts, with one or two exceptions, those
exposure environments defined in EN 206 with the informative examples tailored to UK
conditions. Satisfactory guidelines for ensuring adequate reinforced concrete durability
can only be developed by monitoring concrete performance under field exposure
conditions over an extended period of time. Only then can there be a move from
prescriptive, durability specifications (minimum grade, maximum water-binder ratio, and
minimum cement content) to performance-based methods which are now defined in BS
EN 206-1: 2000 Annex J 5•
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 73
The need to limit the amount of chloride entering reinforced concrete structures has
led to the development of several methods to alleviate the problem. This paper concerns the
performance of two of these techniques used in a marine environment:
(i) the application of a surface coating (silane) to the finished structure; and,
EXPERIMENTAL
Many studies on chloride ingress into concrete have been undertaken using small-
scale samples manufactured and tested under laboratory conditions, or placed at outdoor
exposure sites. In comparison, there are relatively few long-term studies/results on full-size
structures. In the current work, full-size bridge pier stems are fabricated under site
conditions and tested for chloride ingress on a regular basis. A period of seven years data is
presented in this current work.
Pier stems
After the construction ofthe road bridge across the Dornoch Firth in North-East
Scotland in 1991 (Fig. 1), nine pier-stems were cast and placed on the east side of the
southern causeway leading to the bridge. Each pier stem is 2 metres high (see Fig. 2(a))
and fabricated using cut-down sections of the formwork used in the construction of the
bridge piers. The pier-stems are octagonal in cross-section with each face being
660mm; the reinforcement is identical to that used in the bridge piers and comprises a
combination of 32 mm and 40 mm reinforcing in the form of a circular cage. The cover
to the reinforcement varies due to the circular cage arrangement and the pier stems
having an octagonal shape; this gave a minimum cover of 65 mm at the centre of each
face.
Three of the pier stems were constructed with concrete containing caltite
waterproofing admixture, the remaining six used normal Portland cement (ASTM Type
I I CEM I:42.5N) concrete to the same mix design as the actual bridge piers (see Table
1). Three of these six were treated with silane•; the remaining three were left untreated.
In order to simulate the full-height piers of the bridge, where water absorption can
occur only through the sides of the piers, a waterproof coating was cast on top of all the
pier stems. This was undertaken to ensure that water and ionic ingress would occur
mainly through the sides and thus the three similar pier stems (at each location) would
act as a 'continuous' column.
Fig. 2(b) shows a schematic of the positioning of the pier stems at three different
heights on the side of the causeway to provide several exposure conditions typical of a
marine environment. One pier stem of untreated concrete, caltite and silaned concrete
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
• Fosroc Nitocote SN511
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
,,,,```,-`-`,,`,,`,`,,`---
74 McCarter et al.
was placed at each location; two sets of three pier stems were placed at locations which
are exposed to sea water during each tidal cycle. The remaining set of three was placed
clear of high water. None of the pier stems were permanently submerged. The different
exposures discussed in this paper are (see also Fig. 2(b )):
Tidal -'- submerged each tidal cycle and equivalent to XS3 conditions in EN206.
These are defined as the low-level pier stems.
Splash- the upper levels of pier stems subjected to moderate splash from waves
breaking on and around the pier stems. This condition would also fall within to XS3
conditions. These are defined as the mid-level pier stems.
fum!y- clear of the splash zone, subject to wind blown spray (XS 1 conditions).
These are defined as the upper level pier stems.
Note: XS2 represents the fully submerged condition and is not applicable to the
current exposure conditions.
Sampling
The main purpose of the experimental work is to obtain a detailed picture of the
performance of the concrete and the effectiveness of caltite and silane with respect to
chloride ingress and environment. Dust samples were obtained by drilling with a 30mm
diameter bit at selected locations on each pier stem. A total 117 sampling locations
were used for this study. Samples were obtained from the following locations (refer to
inset on Fig. 2 for pier-stem face numbering: face 1 was the seaward face and face 5
the landward face):
(a) Low-level pier stems: faces 1, 2, and 5, at 0.2, 1.0 and 1.8m above base;
(b) Mid-level pier stems: faces 1, 2, 4, 5, 6, and 8, at 0.2, 1.0, and 1.8m above the
base.
(c) Upper-level pier stems: faces 1, 2, 4, 5, 6, and 8, at 0.5 and 1.5m above the base.
At each of the above locations dust was collected using a drilling tool with depth
increments: 0-5, 5-15, 15-25, 25-35, 35-45, 45-55, 55-65mm from the surface. The
drilling tool and drill hole were cleaned between depth increments to reduce the
possibility of cross-contamination of samples from different depths. After drilling, the
drill-hole was plugged with mortar containing Ronafix (a waterproofing polymer). On
average, 400 to 500 samples were collected every 6 months.
The dust samples were analyzed for total chloride content (i.e. acid soluble) in
accordance with BS 1881 :Part 124 6 and BS 633 7:Part 4 7 •
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 75
RESULTS AND DISCUSSION
Using the original cement content of the mix (460 kg/m 3) the results obtained
from analysis of the dust drillings were expressed as percentage chloride ion
concentration by weight of cement. The initial chloride content of the concrete was
measured by drilling cubes manufactured from the same batches of concrete as the pier
stems, which were found were negligible (<0.02% by weight of cement). Generally, a
chloride concentration of between 0.2% and 0.4% is considered as a chloride threshold
for corrosion, although it is clear that there is no single value that represents the wide
range of concreting materials and exposure conditions. As a consequence, a value of
0.4% is used as a reference to study the effectiveness of protection afforded by treated
and untreated concrete surfaces.
Due to the large amount of sampling only typical data are presented for
illustrative purposes. Chloride concentration profiles are presented for face #l(seaward
face) at l.Om for the low- and mid- level pier stems and 0.5m for the upper-level pier
stems. To highlight the effect of the different treatments and location, the same vertical
scale is used throughout.
At this location, the pier-stems were submerged most of the tidal cycle and
subjected to a maximum hydrostatic head of approximately 2.0m at high tide. Chloride
gradients are presented in Figs. 3-5. Fig. 3 indicates that, over the seven year period
presented, chlorides have penetrated to a significant depth, with the threshold value of
0.4% cr being exceeded at 50mm for all pier stems. The pier stems are constructed of
good quality concrete - 66MPa for the untreated and silaned pier stems and 44MPa in
the case of the caltite pier stem. Such strengths would, traditionally, be regarded as
having a good resistance to the ingress of water and dissolved salts. High chloride
levels have been attained in the untreated pier stem in a relatively short time scale; in
addition, the caltite (Fig. 4) and silane (Fig. 5) treated pier stems have not prevented
the ingress of chloride for this particular exposure.
Silane is a hydrophobic pore liner repelling liquid water but allowing the passage
of water vapour and water under pressure to enter the pores. Since the pier stem is
subjected to intermittent hydrostatic pressure, this would account for ineffectiveness of
silane at this location. Caltite, on the other hand, is an integral hydrophobic and pore-
blocking ingredient 8 which is dispersed throughout the entire volume of concrete. Whilst
the hydrophobic action could, like silane, be rendered ineffective under hydrostatic
pressure, the pore-blocking effect should stop the ingress of sea-water under pressure.
Fig. 4 clearly indicates that the latter mechanism is not working as chloride levels are
similar to the untreated pier stem.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Results for three mid-level pier stems are presented in Figs. 6-8. The sampling
points are approximately 0.8m above high water mark. Although still classified as XS3
environment, the main difference is the lack of direct contact with sea-water under
hydrostatic pressure.
Chloride concentrations for the untreated concrete (Fig. 6) are, for a given depth
from the surface of the pier stem, approximately 50% less that the equivalent pier stem
in the tidal zone (Fig. 3). Over the test period, the threshold value of 0.4% chloride has,
in this instance, advanced to a depth of between 25-30mm from the surface. From Fig.
7, the chloride concentrations in the caltite concrete are approximately 70-80% lower
than its low-level counterpart. It is apparent, however, that when compared to Fig. 6,
although chloride concentrations in the caltite concrete are maintained at slightly lower
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
levels than the untreated concrete, the threshold value of 0.4% chloride has still
advanced to a depth of between 25-30mm from the surface.
Fig. 8 presents the chloride concentration profiles for the silane treated concrete.
The silane has maintained lower chloride levels at all depths when compared to the
untreated and caltite concrete at this exposure location. For the silane concrete, the
0.4% threshold value has penetrated to a depth of between 15-20mm from the surface.
The results from the mid-level pier stems would suggest that at depths is excess of
5mm, silane is capable of maintaining lower chloride concentrations within the cover
zone than either the untreated concrete or caltite concrete.
Chloride profiles are presented in Figs. 9-11. In this position, any sea-water contacting
the surface will be as wind-blown spray. The untreated concrete displays a small
concentration of chloride at the surface, whereas caltite and silane have almost
eliminated the ingress of chloride.
The plots in Figs. 3-5 display the same general trend with chloride
concentrations increasing with time and decreasing with depth. It is also apparent that
the chloride profiles are a function of exposure conditions - tidal, splash and air-borne
spray. Although chloride ingress involves a complex interaction of mechanisms, it is
commonly assumed that it can be approximated to a diffusion process. The most common
approach is to apply the error function solution to Fick's diffusion equation and derive an
effective diffusion coefficient. The error function equation is of the form:
(1)
Using equation (1) to obtain a best-fit solution to the data in Figs. 3-5, the effective
diffusion coefficients (Deff) obtained for the lower pier stems are presented in Fig. 12(a).
Over the period presented, all values fall within a relatively small range (1.5-4.5xl0- 12
m2/s) and it is difficult to ascertain a general trend for the silane and caltite concretes. The
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
untreated Portland cement concrete, however, displays a decrease in Deff with time. The
surface chloride concentration, C,, is displayed in Fig. 12(b). The value of C, for the
treated and untreated concretes converges to a value of between 6-7% after 7 years. The
trend for all pier stems is increasing C, with time over this period.
CONCLUDING COMMENTS
Although the tidal zone and splash zone are collectively called XS3
environment according to EN206, there are significant differences between these two
zones with respect to chloride ingress. The treated and untreated pier stems within the
tidal zone indicate that, over the seven year period presented, the threshold
concentration of 0.4% chloride by weight of cement has advanced to a depth of 50mm.
The treated pier stems were unable to prevent chloride ingress under hydrostatic
pressure. That part of the pier stem within the splash zone indicated that chloride
concentrations are noticeably reduced, resulting from the lack of direct contact with
sea-water.
The diffusion coefficient obtained for the untreated pier stem within the tidal
zone decreased with time although values obtained from the treated pier stems were
more erratic. Over the test period presented, the surface chloride concentration
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
78 McCarter et at.
increased with time for both treated and untreated pier stems up to a value of
approximately 6.5% chloride by weight of cement.
It appears that neither of the treated concretes are performing in a way that
would suggest they are significantly better than untreated concrete in cases where the
sea water comes into direct contact with the surface of the pier stems and is subjected
to hydrostatic pressure. Generally, in cases where the concrete is subjected to splash
and spray, the treated concrete pier stems are effective at maintaining lower chloride
levels than the untreated concrete.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACKNOWLEDGMENTS
The work presented forms part of the concrete durability programme into
monitoring and improving the performance of structural concrete in bridges undertaken
for the Scottish Executive Development Department (SEDD) in collaboration with the
Transport Research Laboratory (TRL). Funding from the SEDD is gratefully
acknowledged. The views expressed in this paper are those of the Authors and not
necessarily those of the TRL or SEDD.
REFERENCES
6. British Standards Institution, BS1881: Testing Concrete: Part 124 'Methods for
analysis of hardened concrete', British Standards Institute, London, 1988.
9. Bamforth, P. B., 'Definition of exposure classes and concrete mix requirements for
chloride contaminated environments', Proc. SCI Conference on Corrosion of
Reinforcement in Concrete Construction, Cambridge, 176-190, 1996.
11. Nagataki, S., Otsuki, N., Wee, T-H. and Nakashita, K., 'Condensation of chloride
ion in hardened cement matrix materials and on embedded steel bars', ACI
Materials Journa~ 90, 4, 323-332, 1993.
*SikaFR
# Cementaid Superplastet XL 1.3% w/w PC
+ Includes 24kg water from 83% by volume of caltite
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 1- Showing the location of the marine exposure site on the Dornoch Firth,
adjacent to the road bridge.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
(a)
5 Seaward
-· High-Water
Tidal
, _ . Sampling heigh3
(b)
Fig. 2- (a) The nine pier stems at the marine site, and (b) a schematic showing the
positioning of the pier stems.
0
0 10 20 30 40 50 60
Depth(mm)
Fig. 3- Chloride concentration profiles for untreated concrete in the tidal zone
(low-level).
6
Ca~lte
5
G---El 1.17 years
E'
., __..... 2.67 years
E
4 ------ 4.84 years
tl /!r----6 6.67 years
0 3
~
~
2
u
0
-------
0 10 20 30 40 50 60
Depth (mm)
Fig. 4- Chloride concentration profiles for caltite concrete in the tidal zone
(low-level).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0
0 10 20 30 40 50 60
Depth(mm)
Fig. s- Chloride concentration profiles for silane concrete in the tidal zone
(low-level).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Untreated
5
~E 4
8 er--e 1.17 years
0 13-----El 3.17 yea,.;
j 3 ~4.17yea,.;
...,__..... 6.17 years
~ 2
(3
10 20 30 40 50 60
Depth (mm)
Fig. 6- Chloride concentration profiles for untreated concrete in the splash zone
(mid-level).
Cal lite
5
~
4
"
E
1! G----E> 1.17 years
0 3 G---El 3.17years
~ ?---"'i 4.17years
c 2
....--... 6.17years
13
10 20 30 40 50 60
Depth(mm)
Fig. 7- Chloride concentration profiles for caltite concrete in the splash zone
(mid-level).
6
Silane
5
E'
4
"1!E
0 G----E> 1.17 years
3 G---El 3.17 years
~ ?---"'i 4.17 yeras
c 2 ....--... 6.17 years
(3
20 30 40 50 60
Depth (mm)
Fig. 8 - Chloride concentration profiles for silane concrete in the splash zone
(mid-level).
~ 4
1l
0 e---El 2.17 years
3 G---El 3.17 years
~ ?-----'11' 5.34 years
,e
~
2
'u
10 20 30 40 50 60
Depth (mm)
Fig. 9 - Chloride concentration profiles for untreated concrete in the air-borne spray
zone (upper level).
6
Cal tile
5
'E
Q)
E
4
1l
0 3 e---El 2.17 years
~ ~3.67years
?-----'11' 6.17 years
~
2
ti
Depth (mm)
Fig. 10- Chloride concentration profiles for caltite concrete in the air- borne spray
zone (upper level).
Silane
5
iE 4
!'!
0
~ 3 G-----El 2.66 years
13----€1 5.34 years
c 2
,......._...... 6.67 years
(3
Depth (mm)
Fig. 11- Chloride concentration profiles for silane concrete in the air- borne spray
zone (upper level).
10
(a)
•...
0
Pier 1: Untreated
Pier 2: Callite
I
I Pier 3: Silane
5
,..!!!.
E • 0
';"~
... •
D
•... ...
.
Cl
2
0
...
0
•
1
1 2 5 10
Time (years)
7
(b) ... ~
6
1?
5 ... •
0
"
~ 4 • •
0
..g
e;.
{)
. 3
~
0
... Pier 1: Untreated I
I • 0 Pier 2: Caltite
Pier 3: Silane
2
1 2 5 10
Time (years)
Fig. 12- Temporal change in (a) effective diffusion coefficient, o.ff' and
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Characterization of Portland
Cement Concrete with Carbon
Material Admixtures
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
steel have been studied. Using a constant water/cement ratio of 0.42 the flexural and
compression strengths of concrete with different amounts of carbonaceous materials
and different curing periods have been determined. Also, the effect of adding some
amount of silica fume to the mix formulation has been considered.
The addition of small quantities of carbonaceous materials to the mix
produces an increase of mechanical strengths and a reduction of the concrete
permeability. Due to this smaller permeability the corrosion levels of embedded steel
are lower as compared to the ones in an admixture-free mix, in spite of the higher
electrical conductivity of the composite.
2. EXPERIMENTAL PROGRAM
2.1 Materials. The following basic materials were used in sample preparation:
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Also partially densified silica fume (SF) supplied by Sika SA was used as
admixture in some of the specimens.
2.2 Mechanical strength. Bending and compression tests were carried out in laboratory
ambient conditions. Cylindrical specimens 15 em diameter and 30 em long were used for
compression tests. Prismatic 1Ox 1Ox40 em size specimens were prepared for bending
(flexural) tests. Specimens were stored in a vapor saturated curing chamber for 7, 28 and
90 days, depending on the test. Al tests were carried out in accordance with Spanish-
European norms.
2.3 Corrosion tests. Specimens were similar to those reported in previous works [5]. In
this work specimens were cylinders of 10 em diameter and 10 em height. Two embedded
corrugated steel bars were used as working electrodes and a stainless steel bar in the
cylinder axis was the counter-electrode. The active corrosion area in each of the working
electrodes was 28.26 cm2 . The electrochemical technique used to assess the corrosion
rate was the Polarization Resistance (Rp) [6,7].
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Rp and the corrosion potential Ecorr were periodically measured along the
experimental time. The result data in the following figures refer to average values of the
two working electrodes of each specimen. In order to confirm the Icorr values, obtained
through Rp values, the calculated corrosion product weight of each bar was checked
against the gravimetric weight obtained at the end of the experiments.
2.4 Mercury porosimetry. Porosimetry characterization of the cement paste was carried
out by mercury intrusion porosimetry technique. So both the total porosity volume and
the pore size distribution in the 0.007-400 11m diameter range were obtained. An
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
when the rate of carbon admixture increases. In general this trend can be observed both
with GP admixtures (figure la) and also with GPCF admixtures (Figure lb), although
with smaller differences. From this data it can be drawn that with GP admixtures there is
a clear increase of total porosity when the admixture rate is increased. In the case of
GPCF this trend is not so clear but in none of the specimens the porosity have decreased
when the rate of admixture have been increased.
3.2 Mechanical properties. Figures 2 and 3 show the results of compression strength
tests on 90-day aged specimens fabricated with admixtures of GP and GPCF
respectively, in content ratios ofO% (control), 0.5%, 1%,3% and 5%. It can be observed
that the strength decreases when the admixture content increases, except for GP at 1%
rate.
These results are consistent with the porosity results of Table 1. The higher the
admixture content ratio the higher is the porosity and the lower is the compression
strength.
Bending tests results on 90-day aged specimens with GPCF admixtures in
different ratios are shown in figure 4.
The higher the admixture ratio the higher is the flexural strength. In this case the
reduction of the filler effect (higher porosity) is compensated by the contribution of the
fibers located in the tension part of the specimen.
Figure 5 shows the the results of compression tests on 28-day specimens with
admixtures of HPCF of different aspect ratios (length/ diameter). Only the ones with 3
mm long fibers show an improvement of strength due to the extra strength that the fibers
provide in the transversal direction of the specimen. Short carbon particles do not have
enough bonding and relatively long ones (10 mm) adopt a folded shape, produce voids
and increase porosity.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 91
Bending test data for 28-day specimens with HPCF admixtures of different
aspect ratios are shown in figure 6. Under this type of load all fiber lengths provide some
strength increase with respect to the control specimen. Again the most efficient happen
to be the 3 mm long fibers, for the same reasons as above. They are long enough to get
anchored in the cement paste and short enough not to fold when they are in the wet mix.
behavior should be an increase of the corrosion level when the content ratio of carbon
material admixture is increased. That is the conductivity of the specimen increases, the
effect of carbon-steel galvanic couple increases and the specimen porosity increases and
therefore the mean corrosion level should also increase.
The recorded result data happen to be totally different of the expected results. In
fact the joint action of the above three factors does not imply the growing of Icorr values
when the proportion of carbon material in the mix increases but just, and surprisingly,
the other way around.
The explanation to this unexpected behavior should be searched in other aspects
different to the ones so far considered: a) on one hand the steel bars and the carbon
particles are in contact and consequently the Rp of the carbon particles should be
considered in addition to the Rp values of the steel bars. The kinetics of both materials
act jointly and the contribution of steel bars on Icorr values diminishes; b) porosity has
very little or no effect on the Icorr values when the bar is in passive state, because it does
not control the oxygen diffusion to the cathode surface and presumably the galvanic
effect contributes to a faster growing of the passivating layers. But after the immersion,
when active layers are already present and porosity can control the oxygen cathodic
reduction, the adsorption of oxygen particles by the carbon particles reduces the oxygen
diffusion so counteracting and even overcoming the effect of a higher porosity. The
contibution of these two effects on the reduction of Icorr values will be higher when the
proportion of carbon materials within the paste is increased and that explains the results
shown in figures 9 a) and b).
On the other hand when the steel bar is in contact with carbon fibers, a shift of
potential to positive, less negative, values occur as a consequence of the fact that in basic
media carbon has potential more positive than steel. The higher the proportion of carbon
material the higher this shift will be. In fact this is what can be observed in figure 9 b)
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 93
with the registered corrosion potential. This argument explains the formation of a more
compact passivating layer on the already passive steel surface and consequently a lower
faradaic current and lower Icorr values are registered as can be observed in figure 9 a).
4. CONCLUSION
The addition of carbon material to concrete may provide improvements in the
mechanical properties without damaging the corrosion state of the reinforcing steel bars.
ACKNOWLEDGMENTS
The authors would like to acknowledge financial support received from the
Generalitat Valenciana (Spain) (CTIDIB/2002/164) from Ministerio de Ciencia y
Tecnologia (MAT 2003-06863) ( MAT2002-0621) and from Ministerio de Fomento
(Spain).
REFERENCES
l. CHUNG, D.D.L. (2000) Cement reinforced with short carbon fibers: A multifunctional
material, Composites Part B: Engineering, 31, pp. 511-526
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2. CHUNG, D.D.L (2003) Multifunctional cement-based materials. (Ed. Michael Meyer
editor. Marcel Deeker, Inc.). New York
3. HOU, J. Y CHUNG, D.D.L. (2000). Effect of Admixture in concrete on the Corrosion
Resistance of Steel Reinforced Concrete, Corrosion Science, 42, pp. 1489-1507.
4. GARCES, P., FRAILE, J. Y VILAPLANA-ORTEGO, E. (2005) Effect of carbon
fibres on the mechanical properties and corrosion levels of reinforced portland cement
mortars, Cement and Concrete Research, 35, pp.324-331.
5. ANDION, L.G., GARCES, P. Y CASES, F. (2001) Metallic corrosion of steel
embedded in calcium aluminate cement mortars, Cement and concrete research, 31,
pp. 1263-1269
6. STERN, M Y GEARY, AL. (1957) A theorical analysis of1he shape ofpolarization cmves, Joumal of
Electrochemical Society, pp. 56.
7. ANDRADE, C. Y ALONSO, C. (1996) Corrosion rate monitoring in the laboratoty and
on-site, Construction and Building Materials, 10, pp. 315-328.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
94 Garces et al.
Table 1. Total porosity (% of specimen volume) of concrete with
GP and GPCF admixtures
Fiber 0% 0,5% 1% 3% 5%
type
GP 468 4,98 5 21 577 6,53
GPCF 4,68 5,48 4,84 5,43 5,60
0,030
(a) -•-Control 0,030
-•-GP0,5%
-•-GP1% (b) -•-Control
0,025 o.... -•-FCUG0,5%
-•-GP3%
-•-FCUG1%
f 0,020
-•-GP5%
-:gQQ2(1
.s.'
-•-FCUG3%
-•-FCUG5%
1 0,015 ! 0,015
~ ~
"
-~ 0.010
~
I
!i
0,010
0,005
Fig. 1. Results of mercury intrusion porosity tests in concrete samples with different
ratios of GP admixtures (a) and GPCF admixtures (b).
GP ()'% GP 0,5% GP 1% GP 3% GP 5%
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 4· Bending test (flexural strength) results on 90-day aged specimens with GPCF
admixtures in different ratios.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 6. Bending test data (flexural strength) for 28·day specimens with HPCF
admixtures of different aspect ratios.
Fig. 8. Bending test results on 28·day specimens with both HPCF and SF admixtures.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
i 0,1l~....,~tel~~~~-·
~ 0,01
3~--~~r-~~~~--~~r-~~~--~~=
10 20 30 40 50 60 70 80 90
Time(days)
Fig. 9· Evolution of Icarr and Ecorr values in steel bars embedded in concrete samples
with GP admixtures in different proportions and different ambient conditions.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: In the framework of radioactive waste storage, the French atomic energy
agency needs to design a concrete container with a service life of 300 years.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
50°C, using a Carlo Erba 2000WS porosimeter. The pore size distribution curve (Fig. 4)
(mean value for three samples) enables two porous modes to be distinguished [6]. The
first porous mode (centred on pore radius of 90 nm) corresponds to the porous volume
located between the grains of anhydrous cement. The second porous mode (centred on a
pore radius between 10 and 20 nm) corresponds to the porous volume located between
hydrates.
The studied concrete shows a bimodal pore size distribution. The additions (silica
fume and fly ashes) according to their filler effect and pozzolanic properties densities the
macrostructure and obtains a porosity evaluated by mercury intrusion of 8.9 %.
The value of the porosity evaluated by mercury intrusion is lower than the value
obtained by the water opened porosity technique which is around 20 %. This
experimental observation can be explained by the fact that the measurement by mercury
intrusion under-evaluates the volume of large size voids and the volume of small radius
pores linked to the C-S-H. This last volume is partially accessible by this technique. With
water opened porosity techniques all pores are accessible for measurement.
The high performance concrete used on the worksite at Sens presents a mercury
intrusion porosity of 7.3 % [4]. The values of porosity more important for the concrete
studied than those obtained for the high performance concrete used on the worksite of
Sens can be explained by a higher W/C.
DURABILIY INDICATORS
The use of a predictive guideline for the durability of reinforced concrete
structures is based on the determination of durability indicators coupled with
mathematical modelling for the prediction of the service life. These durability indicators
are fundamental parameters that influence the physical, chemical and mechanical
mechanisms which induce degradation of concrete. The estimation of the durability of the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 103
high performance concrete chosen in this study was performed by measuring the
durability indicators proposed by the French group AFPC-AFREM [2]. Before the
measurement of these indicators, compressive and tensile strengths were tested at six
months. The results show that the mean compressive strength is 92.9 MPa and that the
mean tensile strength is 7.0 MPa. The Young's modulus has a mean value of38.2 GPa.
Capillary adsomtion
The capillary adsorption gives the capacity of concrete to recover water. The three
samples are preliminary dried at 80°C until constant mass, then put in air-tight bags and
placed in a drying room for 10 days. The lateral surfaces of samples are coated with resin.
The test consists in immerging the base (the first three millimeters) of cylindrical
concrete samples (diameter 110 mm, high 40 mm) in water. It seems difficult to retain
only a particular value of the capillary adsorption coefficient because the variations of
this coefficient versus the square root of time are not linear (Fig. 5). For concretes of the
C80 class the capillary adsorption coefficient is around 2.70 kg/m2 at 24 hours [2]. The
tests carried out during this study show that the value of the capillary adsorption
coefficient is 1.25 kg/m2 at 24 hours. The material absorbs a small quantity of water; it is
of better quality that concretes of the same class.
Transfer properties
The presence of chloride ions drastically modifies the service life of reinforced
concrete constructions. The chloride ions play a major role in the corrosion of steel. In a
saturated concrete, transfer is governed by the diffusion of ions under a gradient of
concentration and their fixation on hydrates of the cement paste. In a partially saturated
concrete (case of the containers), the chlorides are able to migrate with the interstitial
liquid phase by convection. This phenomenon is more important in that it is very fast.
Nevertheless, even when the movement of chloride ions is coupled with humidity
transfers, the knowledge of diffusion coefficient remains necessary in order to realize a
good prediction of the penetration of chlorides. Currently, there is no experimental
procedure proposed by the French work group. The experimental procedure established
by the LMDC [7] was chosen to realize the measurement of the diffusion coefficient. A
sample of concrete is placed between two electrodes and an electrical field is applied
across the specimen. The upstream compartment contains chloride ions solution and the
downstream compartment contains a solution without chlorides. At regular intervals the
concentration of chlorides in the upstream compartment is measured by titration of the
solution. The curve of cumulated chloride ions quantity leaving the compartment versus
time is plotted. The slope of the straight line permits to determine the effective diffusion
coefficient De. The obtained value for two tests carried out 0.62 10- 12 m2/s. In order to
compare the studied concrete with another concrete of the same class, the effective
chloride diffusion coefficient of the high performance concrete used on the bridge of
Sens obtained by the same procedure was 0.24 10- 12 m2/s.
The gas permeability of concrete is very important with regard to its durability. It
influences the penetration of aggressive agents like C0 2 and water vapour transfer during
the drying of concrete. The hydrous state of the material is important because according
to the saturation rate different pore sizes are accessible. The gas permeability is one of the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
104 Devillers et al.
durability indicators defined by the AFPC-AFREM French workgroup [2]. However, for
this indicator preconditioning was carried out according to the RILEM procedure [8]. The
preconditioning proposed by the AFPC-AFREM consists in an air vacuum saturation
followed by a drying at 80°C during 28 days then at 105°C until a constant mass of
samples is obtained. The chosen preconditioning is a drying at 50°C in a drying room
until the saturation rate corresponds to the equilibrium at 75% of relative humidity. This
phase is followed a phase of redistribution of gradients of relative humidity in the sample.
A first measurement of permeability is realized. Then samples are dried at 105°C until a
constant mass is obtained. A second measure of permeability is performed. This
preconditioning has been chosen because on the one hand it limits the risks of cracking
due to high temperature gradients and, on the other hand, it obtains the most
homogeneous possible saturation rates of samples.
The three cylindrical samples were 15 em diameter and 5 em high. Their
circumference was coated with sticky aluminium so that gas transfers, during drying or
permeability measurements, would take place by the two plane faces of the sample. The
experimental apparatus used is the permeameter proposed by CERILH. The gas used for
these tests is oxygen. The apparent permeability is measured under a pressure gradient
from 0.1 to 0.6 MPa in a stationary flow of gas. The mean values obtained for the oxygen
permeability (Fig. 6) are 1.3 10· 17 m 2 for preconditioning at 75% of relative humidity
(corresponding to a saturation rate of 62%) and 5 10· 17 m2 at drying state. The apparent
permeability of the high performance concrete used on the bridge of Sens obtained at
drying state is 6.7 w-l? m2 •
Carbonation
Whether or not in combination with chloride ions, concrete carbonation is an
important factor regarding the durability of reinforced concrete structures. The concrete
carbonation is due to the reaction of carbon dioxide contained in the air with calcium
compounds of concrete. The formation of calcite leads to a consumption of portlandite
and a decrease in pH. In a safe concrete, the pH of the interstitial solution is 13, placing
steel in passivity zone. The pH of the interstitial solution of a carbonated concrete is
around 9, placing then steel in the corrosion zone. It is therefore necessary to study
carbonation in order to guarantee the service life of structures.
The French work group AFPC-AFREM [2] has defined a procedure for an
accelerated carbonation test of concrete. The two samples used are cubes 7 em high
obtained by sawing of cylindrical samples moulded during the construction of the
containers. The preconditioning consists in an air vacuum water saturation, weighing and
drying at 40±2°C during two days. After this preconditioning the samples are placed in a
controlled hygrometric room (RH=65±5%) including a gas mixing distribution system
(50% of C02 and SO% of air). The carbonation depth is measured at 7, 14 and 28 days.
Nevertheless the tests can be continued until 90 days, when durability criteria are
required for long period of time (120 years for the Tagus Bridge or 300 years for
radioactive wastes storage). The measurements of the carbonation depth are realized after
humidification of the surfaces and phenolphthalein solution spraying.
The results show that the concr~te studied has a very low carbonation depth (1.7
mm at 90 days). This result must be linked to the low porosity of the material. The
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
[ 1] Commissariat a 1'Energie Atomique, "Cahier des charges projet MAVL :
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2'
ta
=:: ~:::: ::::::::: :::~::: ::::::: :::~::: ::::::: _ :::~::: ~::: :::~::: 0,100
t -9 ~_,f-+~d-+-l----+-+-+--f----,-+-+-t-H~++-l----++-+--1 ~
-~::: ~~::: :::~::: t---f--1----:--+=-~L=t-~'--,::: J,__= ~
~ -12
a ::::::: :::;:::
-' ll - -',- _ll - y-::=r::_ -},()0/a - - -',- I -,- --, - 5
~ -r- -,- -r- -:~~ ~~- -,- :::r::: -- --- a
-24
-:- -~- ~~~~:\- -~-t4= -~-
-I---_:__Jf--.:.-+_:_-l--;_~-i-'*\:...:.,q:.._;_f--.:.--Jli"'"'""""""~~-4--..:'_--J-
:1- I
0,000
0 100 200 300 400 500 600 700 800 900 1000 llOO 1200
temperature (OC)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A
0,7
~~ 0,6
r--
~ 0,5 r--
-;:- 0,4
~
§ 0,3 I--
~ D,2 I--
0,1 I--
1-
10 100 1000 10000 100000
Pore radius (um)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1,4
1,2
,
I
i
!
0,8
i
0,6
I 0,4
0,2
7,00
0,00 SJO%
I-- r---
'f r-- r-- r---,
.... 5,00
:s
iii
E
..8.
4,00
3,00
2,00
--
~
1,00 S.-62%
0,00
I
~00 0,05 0,10 0,15 0,20 0,25 0,30 0,35 0,40
Mean absolute pressure (MPa)
1,6
ld_
./;
A
e
t.
1,4
1.2
1
-
-- v
~ /
/
J :::
0,4
"" ..........__
v
0,2
5 6 10
Square root of time (day••')
100 5000
90 4500
80
~ I, 4000
70
i\ II _~
\; 3500 .....
80
40
II .;
~
II I ~\
/"
3000
2800
2000
l
:
fl/ ..
30 1800
20 1000
10
~'I 500
..# ~
20 40 60 80 100 120 140 160
number of cycles
Fig. 8. Deformation and resonance frequencies ratio versus number of freezing and
thawing cycles.
0,8
" 0,8
~ 0,4
f 0,2
/ ./'-
~ ---~
~ 0 ...........
! .j),2
-...
I
c: -0,4
[ .j),6 ~ -1-----
~ .j),8
IV
·1
-1,2
-60 ·50 -40 ·30 -20 -10 10
Temperature ("C)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: In this paper the results of an investigation into the frost salt scaling
resistance of cement mortars bearing extra-fine slag (EFS) with Blaine fineness of soo
kg/m 2 is reported. The water to cement ratio of the mortar specimens is kept as 0.45.
EFS was used as a replacing material for both European CEM 111/B blast furnace slag
cement and CEM I normal portland cement. Within the framework of this study, the
carbonation resistance and pore structure of the cement mortars are also studied.
Carbonation resistance is measured by phenolphthalein method after accelerated
carbonation test in 3% C02 chamber. Pore structures of the samples are evaluated
by means of mercury intrusion porosimetry. The microstructures of the samples are
further characterized by environmental electron microscope. It appears that EFS
replacement in CEM Ill/Band CEM I decreases the frost salt scaling durability of
mortars.
J Bijen has been director ofiNTRON, the institute for quality assessment in the building
industry in The Netherlands for 20 years. Nowadays he is involved as a consultant. He is
an emeritus professor from Delft University of Technology and director of FEMMASSE
BV.
TECHNICAL BACKGROUND
It has been known for a while that the effect of carbonation leads to a more open
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
pore structure in slag cement pastes bearing high amounts of slag, and increases the
vulnerability of the system to external aggressive environment [1, 2]. It has also been put
forward by several authors that severe carbonation leads to poor frost salt scaling
resistance of concrete containing high levels of ordinary ground granulated blast furnace
slag [3-5]. However, very little research has been conducted on the carbonation and/or
frost durability of concrete bearing slag utilization compared to normal portland cement
(NPC) based materials. There is also a limited amount of research on the effect of fine
slag replacements on the carbonation and/or frost salt scaling resistance.
Zhang et al. [6] carried out research on the carbonation of hardened binder
pastes containing super-pulverized blast-furnace slag. In that study, 50% NPC 32,5- 50%
ordinary slag (with fineness of 346 m 2/kg) and 50% NPC - 50% super-fine slag blends
(with fineness of 800 g/cm 2) were used. They concluded that the carbonation resistance
of cement paste with finer slag is superior to that of original one. This finding was
attributed to two aspects; I, that the addition of super pulverized slag helps to form a
dense structure and II, that lower amount of Ca(OH) 2 is produced in the finer slag cement
paste.
Bilodeau and Malhotra [7] presented the frost salt scaling performance of
concretes containing cement with 25% and 50% granulated blast furnace slag
replacements. Three different Blaine fineness levels were chosen for the slag (608, 540,
418 m2/kg). They concluded that 50% slag replacement showed more scaling than 25%
replacement. However, their results did not reveal a relationship between slag fineness
and frost salt scaling performance in the concrete used. They also did not investigate the
effect of carbonation.
This study is designated to determine the effect of EFS on the carbonation and
frost salt scaling resistance of mortars at different replacement levels, and for two types
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 113
of cements; a CEM III IB blast furnace slag cement and a CEM I normal portland cement
will be investigated. Due to economical and ecological concerns, the performance of very
high-level replacements was of particular interest. Fineness of cement has been an
important factor in increasing the early strength of concrete. Therefore, utilization of a
finer slag might be influential with respect to the improvement of the general resistance
of slag cement systems, especially including frost salt scaling.
Granulated blast furnace slag with Blaine fineness of at least 500 m 2/kg is
considered as super-fine slag [8]. However, as the slag used in this study is at the lower
limit, it is named as extra-fine slag.
EXPERIMENTAL DATA
In this experimental series, it was intended to increase the slag content further as
well as increasing the EFS portion in the cement blend. In the various mixtures 10%,
25% and 50% (by weight) of CEM III/B cement was replaced by fine slag. The obtained
cement blends are presented in Table I. The EFS portions became 14% (M2), 33% (M3)
and 60% (M4) of the total slag content. The slag cement with no EFS replacement (Ml)
was the control, and its name according to the European Cement standard NEN EN 197 is
CEM III/B 42,5 N HSR LH (see Table 2).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The replacement levels were 0% (PMl), 25% (PM2), 50% (PM3) and 75%
(PM4) (by weight) of the normal Portland cement (NPC, CEM 1). The chemical and
physical properties of this cement are presented in Table 3 and the obtained cement
blends are presented in Table 4.
The results reveal that increasing the average fineness and total amount of slag
leads to higher degrees of water uptake with regard to the control sample. The only
exception was the very low level replacement (10%, M2) of EFS (Fig. 1). This was
probably due to the better particle packing effect in the system. Carbonation is another
important factor for water suction. For the cements with high slag content, carbonation
increases the water suction rate. Increased carbonation depth strongly contributes to the
water uptake. Again the tendency for highly (or accelerated) carbonated samples was
similar to the normal carbonated ones.
Depths of carbonation were measured at 10 random locations according to
phenolphthalein spray method right after the carbonation period on highly carbonated
broken surfaces. The average depths were 6.5, 6.5, 8.0 and 13,5 mm for M1, M2, M3 and
M4, respectively. For normal carbonated samples the carbonation depths were not more
than 1 mm.
After the conditioning period, the samples were tested according to the standard
CDF test. The results are presented in Fig. 2. The test had to be stopped after 14 cycles
due to the severe disintegration of highly carbonated M3 and M4 samples. The results
showed that M2 had almost identical frost salt resistance to M1 control samples even
though it contained higher amount of the total slag.
In Fig. 3 and Fig. 4 the pore size distribution of the samples in non-carbonated
and carbonated conditions are presented. In non-carbonated samples, the majority of the
pores were identified in 0.01 - 0.1 J.lm zone (micro pores). It was found that the
increasing amounts of EFS leads to more microporosity, which is likely to be favourable
for frost salt scaling resistance, because of small meso- and macroporosity, which are
related to frost resistance. However, the situation appears to be the other way around
when the samples are carbonated. Generally speaking, the microporosity was shifted to
macroporosity by the effect of carbonation reactions. According to the MIP results, EFS
bearing samples are more affected compared to the control samples. More coarsening in
the pore structure is obtained, where the EFS content increases. One should bear in mind
that the total slag content is also significantly effective on the frost salt scaling resistance.
As mentioned earlier, Zhang et al [6] concluded that the good carbonation
performance of their super fine slag pastes could be attributed to the lack of calcium-
hydroxide (CH). However, lack of CH does not seem to be favorable in terms of
resistance against C0 2 attack. It is very important to consider the two different
carbonation reactions affecting the carbonation of the cement-based materials in opposite
directions. Carbonation is a slow through-solution process that is a reaction of dissolved
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
As a result of the first part of this experimental study, it was claimed that the
critical amount of normal Portland cement clinker (or CH content) is of importance with
regard to carbonation and frost salt scaling durability. In this second part of the
experimental study the w/c 0.45 cement mortars with EFS replacement levels of 0%
(PMl), 25% (PM2), 50% (PM3) and 75% (PM4) were studied in order to explore the
issue further.
In Fig. 5 the carbonation resistance of the mortars with various fine slag
replacement levels is presented. The mortars show faster rate of carbonation as the
amount of EFS in the cement increases. The increase of depth of carbonation is
particularly slow up to 50% replacement level. It was observed that at 75% replacement
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
level the depth of carbonation is almost double amount for NPC with 50% fine slag
replacement.
The utilization of fine slag in NPC had also an impact on the water uptake
properties of the mortars. In Fig. 6, the amount of 3% NaCl solution taken up during one
week is presented. It can be noticed that the tendency of the graph is quite similar to the
carbonation performances. The amount of brine taken up by PM 1, PM2 and PM3 is
between 10-15 grams after 7 days while PM4 showed considerably higher uptake of 20
grams and 37 grams for lab carbonated PM4 and 3% C02 carbonated PM4, respectively.
The effect of carbonation on the mechanical properties of PM I, PM2, PM3 and
PM4 are presented in Table 5. It can be seen that the high level replacement caused a
significant drop in the tensile and compressive strengths compared to PMl, PM2 and
PM3 under current curing conditions. This is most likely due to the insufficient degree of
hydration. It is noteworthy that 50% fine slag replacement presented a competitive
performance compared to the control mortar (PMl). Carbonation generally leads to a
decrease in tensile strength and an increase in compressive strengths in the mortars with
up to 50% slag. In mortars with 75% slag, both tensile and compressive strengths
diminish because of carbonation.
It could be expected for PM4 mortars to show a much lower tensile strength
after accelerated carbonation. However, the decrease after accelerated carbonation seems
to be rather small considering the difference in scaled mass between lab-carbonated and
carbonated samples. The worse scaling performance of accelerated carbonated samples
may be attributed to the coarsening of pore structure due to carbonation (Fig. 9 and Fig.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
116 ~opuroglu et al.
10). In case of accelerated carbonation, a coarser, more interconnected pore structure is
obtained and consequently a much higher amount of water is taken up by the mortars (Fig
6, PM4/NC vs. PM4/HC). This should cause higher amount of materials to be affected by
the ice expansion and scaling action.
The curing procedure in this study seems to be insufficient for slag hydration.
However it was intended to replicate a typical on-site practice, which demonstrates water
curing hardly longer than a couple of days. It is reasonable to assume a better
performance of similar specimens in case of prolonged initial water curing prior to
carbonation. A reasonable explanation to the current results could be the inefficiency of
activation of the PM4 cement blend. Utilization of slag in CEM I brings gypsum
(CaS0 4 .2H20) into the system as an activator instead of anhydrite (CaS0 4) which is
generally used in slag cement blends (such as CEM III /B) as the activator. This explains
why more scaling is seen in Fig. 2 than in Fig. 7.
The frost salt scaling performances of mortars with NPC containing various
portions ofEFS is presented in Fig 7 and Fig. 8. Up to 50% EFS replacement, no inferior
effect of carbonation was seen on the frost salt scaling resistance of the mortars. It is
interpreted that due to the existence of abundant CH in the system, the carbonation
reactions resulted into mainly higher volume-stable calcite minerals. However, in case of
very high replacement level (PM4), small amount of calcite is produced along with meta-
stable carbonate minerals, which can be further decomposed by the combined attack of
frost and de-icing salts. This action is reported to be one of the factors in decreasing the
surface durability of slag cement based materials [4].
A critical result obtained from the frost salt scaling tests is the comparison of
performance ofPM4 (CEM I+ 75% EFS) with M3 (CEM III /B + 25% EFS, 75% slag in
total). PM4 showed relatively lower scaling resistance with respect to M3 even though it
was possessing 75% EFS slag against 33%EFS+42% normal slag of M3. Again the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
activation of slag might play an important role here. Even though the slag content is
identical, M3 is a blend, possessing CEM III /B which complies with the cement standard
requirements. In M3, the CEM III /B clinker and the original slag are interground
resulting in a relatively high fineness of the clinker because it is softer. That means that
the fine clinker might react faster in the CEM III/B and will activate the slag also faster.
On the contrary, PM4 blend with 75% slag is not optimized according to the cement
standards. According to the results, the activator type and content seems to be more
crucial than the EFS portion.
The MIP measurements in Fig 9 and Fig 10 implied that in case of EFS
replacements up to 50% the intrusion volume peaks at 01-1 11m move slightly to the right
side after the carbonation (for PM1, PM2 and PM3). However in case of very high
replacement (PM4), the peak moves from 0.01-0.1 11m to 0.1-1 11m zone indicating a
significant coarsening of the microstructure.
The MIP observations were confirmed by the ESEM photomicrographs of the
non-carbonated and carbonated mortars (Fig. 11 and Fig. 12). In non-carbonated samples,
CH phase was visible in up to 50% slag content. However in PM4 with 75% slag, the
insufficient degree of hydration is distinguishable. The unreacted slag particles have
pronounced interfaces with the matrix showing hardly any sign of hydration. In case of
carbonation these zones become wider due to the dominant progress of equation (2) in the
lack of CH and this leads to a less resistant skin. In Fig. 12 the negative effect of
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 117
carbonation on PM4 is presented at lower magnification. The coarsening of the pore
structure is visible even at 125 times magnification. Similar microstructural disintegration
was observed in certain locations of PM3 as well. However, 50% of normal portland
cement assures a reasonable quality, as can be confirmed by the frost salt scaling test
results. PM1 and PM2 showed visually good microstructural integrity after the
accelerated carbonation.
CONCLUDING REMARKS
addition.
CH content of the system, the hydration rate of clinker and slag and the related
activation of the slag appear to be factors of major interest, which should be investigated
further.
REFERENCES
1. Matala, S. The changes in pore structure of carbonated grabulated blast furnace slag
and ordinary Portland cement concretes. in Service Life and Maintenance of
Buildings. Oulu, Finland.l994, pp. 144-160.
2. Matala, S., Effects of Carbonation on the Pore Structure of Granulated Blast Furnace
Slag Concrete. 1995, Helsinki University ofTechnology. p. 161.
3. Stark, J., Ludwig, H.M. Freeze-Deicing Salt Resistance of Concrete Containing Blast
Furnace Slag Cement. in Freeze-Thaw Durability of Concrete. Sainte-Foy, Canada
and Lund, Sweden: E&FN Spon.1997, pp. 107-120.
4. Stark, J. and Ludwig, H.M. Freeze-Deicing Salt Resistance of Concrete Containing
Cement Rich in Slag. in Frost Resistance of Concrete. Essen: E&FN Spon
London.1997, pp. 123-138.
5. Utgenannt, P., The influence of ageing on the salt-frost resistance of concrete, Ph.D.
thesis. 2004, Lund University.
6. Zhang, X., Wu, K., Yan, A., Carbonation Property of Hardened Binder Pastes
Containing Super-Pulverized Blast-Furnace Slag. Cement and Concrete Composites,
26 (2004) 371-374.
7. Bilodeau, A. and Malhotra, V.M. Deicing salt scaling resistance of concrete
incorporating supplementary cementing materials: CANMET research. in Freeze-
Thaw Durability ofConcrete.1997, pp. 121-156.
8. Niu, Q., Feng, N. and Yang, J., Effect of superfine slag powder on cement properties.
Cement and concrete research, 32 (2002) 615-621.
9. CDF, CDF, Capillary Suction of Deicing Solution and Freeze Thaw Test. Materials
and Structures, 28 (1995) 175-182.
35
days
Fig. 1-- 3% NaCl solution suction of lab/accelerated carbonated mortar samples Mt,
M2, M3 and M4. (NC: Lab carbonation, HC: 3% C0 2 carbonation)
200~~----------------------~
180 -!1_,,~1-------------------.j
100r-----------~
~1~r------------
.!l
~1~r-------------
~1oor--------------
~
sor---------
'iii
E so r - - - - - - - - -
~ 40
20
M1 M2 M3 M4
Fine slag replacement
Fig. 2-- Frost salt scaling performances of mortars with CEM 111/B 42,5 N HSR LH
cement containing various portions of extra fine slag. (NC: Laboratory carbonation,
HC: 3% C0 2 carbonation). Total slag content in the cements is 67%, 70%, 75% and
84% for M1, M2, M3 and M4, respectively.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
: ::~' : ::: I I
I
111
Ill
0.2
: '1•::::
I 1111
I
l
I
I II
Ill
Ill
Ill Ill
Ill Ill
Ill ill
i 0.15
Ill
Ill
Ill
Ill
Ill
ill
ill
ill
I
II
II
II
Q1 II
0.05
0.3
0.25
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.2
i
c 0.15
.8'
~ 0.1
0.05
0
0.001 0.01 0.1 1 10 100 1000
pore diameter (~m)
camonamd mortars wlc=0.45 --M1C --M3C --M4C --M2C
Fig. 4 --Pore size distribution of carbonated mortars with CEM 111/B 42,5 N HSR LH
cement with various portions of extra fine slag (w/c of 0.45).
Fig. 5 --Carbonation depth of NPC mortars w/c 0-45 with various slag contents in
3% CO 2 environment
40
35
30
i 25
c::
.Q
t) 20
:::>
"'Gi
115 ___ : .. : .. :: .. : .... ---·
10
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
l
;:'!;
80
60
~
"'
01)
.5
40
Oi
~
20
0
0% 25% 50% 75%
Fine slag replacement
Fig. 7-- Effect of slag content on the frost salt scaling resistance of mortars wfc 0-45.
(NC: lab carbonation, HC: 3% C0 2 carbonation.)
Fig. 8 ·-Surface conditions of the lab carbonated (upper row) and 3% C02 carbonated
(lower row) samples from the second series after 14 freezing and thawing cycles.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
I I I I I !I I I I I I
I I I I I II I I I I II I I Ill II II I I I I II
I I I I I II I I I I II I I Ill I I I II I I Ill
I'
0.14 jPM4:- f5o/~ I I I
I
I
I
I
I I I
' I
I I I II
I !Ill
I
~ : : : : : : : I I ill
I I I I Ill I I
II
I I II I I I I I Ill I I 'I
0.12 1-i 'I I I I I Ill I I I
I I I I II I I II I I I I I I I I
I I I II I I I II I I I I I I I
) I I ' I I I
~ 0.1 I'
1\ I I ill I I
II I ' I Ill! I
I ""I I
'I I I
.§. I \1 I I I I
i
I
~
I
: I I I I
0.08 ~~ I I
I I I I' I I II
:!2 I I I I I II
~ 0.06 I
I'
I ill
Ill!
I I II
Ill
I
I
I
0.04 I I I
I I I I
I I' I I
I I I I
I I
0.02 I I
I 111
I I \\I
o-J.I____Llt~lJ1l~___L_jjL_L~~~~Id~DI~IU~bL~~~~~~~j__LI_LI
I I llll~
Fig. 9 -- Pore size distribution of non-carbonated CEM I 32,5 cement mortars with
various portions of extra fine slag for carbonated mortars w/c 0.45.
- 75°/o
I
Stag : : : ::
I I I I ill
0.14 I I I I ill
I I I ill!
I I I I ill
I I I fill
I I Ill!
0.12 I I Ill
I I ill
I I Ill
I 'I Ill
I I Ill
~ 0.1 I I I II
I Ill
s I Ill
i
1 II II
I I ill
0.08 I I ill
~
~ 0.06
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.04
0.02
o+.--~~~~~--~~~~~~~~
Fig. 10 --Pore size distribution of carbonated CEM I 32,5 cement mortars with various
portions of extra fine slag for carbonated mortars w/c 0-45.
INTRODUCTION
The durability of concrete has increasingly become the focal point of research in
the concrete community throughout the world. This is due to the fact that deterioration in
concrete severely affects the service life, safety and maintenance of concrete structures 1•
There are many factors that contribute to the deterioration of concrete, such as the alkali-
aggregate reaction (AAR), which is one of the most common causes of concrete
deterioration in Eastern Canada2 • This mechanism leads to cracking of the concrete,
which often results in a "map pattern" on the surface, as well as global swelling of the
concrete structure leading to severe operational problems.
RESEARCH SIGNIFICANCE
There are few studies that have applied pattern recognition techniques, such as
texture analysis, on concrete imagery to obtain deterioration information, such as crack
density, and there have so far been even fewer efforts to combine two approaches, such
as statistical methods and signal processing methods 9 • The aim of this research is to
develop an imaging method that characterizes concrete damage through the use of the
statistical texture analysis approach of the Grey Level Cooccurence Matrix 10 and the
signal processing technique of Haar's Wavelet Transform 11 • This combined method is
applied to grey scale images of three concrete blocks, exposed outdoors for ten years,
and exhibiting various levels of AAR damage. Texture information is used to recognize
and quantify surface deterioration, such as the total amount of crack damage, and the
length and width of cracks in the concrete blocks.
Grey scale images of the blocks were taken by digital camera at a pixel resolution
of 0.26 mm; the images were then cropped to an image matrix of 512 x 512 pixels in
order to obtain better representation of damage.
Image texture
There are three major aspects that characterize digital image data: spatial --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
resolution, spectral resolution and radiometric resolution. The spatial resolution of the
data is the equivalent in kilometres, meters or even centimetres of the target object to its
smallest components available for processing from the original image: discrete picture
elements, known as pixels, the size of which varies according to the sensor system. The
radiometric resolution of an image refers to the number of binary digits, or bits, needed
to represent the range of available discrete brightness levels, known as digital numbers
(DN), which are the quantized radiance values recorded for each pixel by the sensor.
The spectral resolution of the image data corresponds to the wavelength bands, or
channels, in the electromagnetic spectrum, in which the image is acquired. This is
usually a measurement of the spatial distribution of reflected, or emitted, radiation in the
ultraviolet, visible and near-to-short wave infrared range of wavelengths, known as the
solar spectrum. Sometimes, it can be a measurement of the spatial distribution of energy
emitted by the earth itself in the thermal infrared wavelength region. It can also be a
measurement of the relative backscatter from the earth's surface of energy actually
emitted from the remote sensor itself in the microwave band ofwavelengths 12. 14 •
One of the most important aspects of texture is scale 16 . This is significant because
texture is a complex multiscale phenomenon 17 ; it has a recursive nature. A texture seen
at one scale may not be visible at a different scale. For example, consider the texture
represented in a brick wall. At a low resolution, the texture is perceived as formed by the
individual bricks in the wall; the interior details in the brick are lost. At a higher
resolution, when only a few bricks are in the field of view, the perceived texture shows
the details in the brick 17 •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Haar's discrete wavelet transform
In signal analysis, there are a number of different functions that can be performed
on a signal in order to translate it into different forms that are more suitable for different
applications. One of the most popular functions in the engineering world is the Fourier
transform (FT); it is used to determine the frequency content of a signal in digital signal
processing applications. In imaging applications, the FT allows inter-conversion
between the spatial domain representation and the frequency domain representation of
images 8 . Unfortunately, this transform does not provide good resolutions for both high
and low frequencies at the same time.
This problem, however, can be overcome through the wavelet transform, which
allows a signal to be converted and manipulated while keeping resolution across the
entire signal and still be based in time. Wavelets are finite windows through which the
signal can be viewed; in order to move the window about the length of the signal, the
wavelets can be translated about time in addition to being compressed and expanded.
Haar' s transform is the simplest form of wavelets; it is essentially a process of averaging
and differencing of values. For two-dimensional images, the transform is applied to the
pixel values in each row of the digital image, then to each column, thus decomposing an
image into low-frequency and high-frequency components 11 •
The low-resolution image can further be decomposed into the next level of low-
resolution and detailed images. Therefore, depending on the number of resolution steps
required I, 1+31 sub-band images will be produced, resulting in a multi-level
decomposition as shown in Fig. l(b). The LL image results from a low-low-pass filter,
the LH image from a low-high-pass filter, the HL image from a high-low-pass filter, and
the HH image from a high-high-pass filter. In a pyramidal decomposition, only the low-
resolution image (LL) is further decomposed.
One of the defining qualities of image texture is the arrangement of the grey
values in the image space, known as the spatial distribution. Statistical texture methods
analyze the spatial distribution of grey values by computing local features at each point
in the image, and deriving a set of statistics from the distributions of the local features.
Depending on the number of pixels defining the local feature, statistical methods can be
classified into first-order (one pixel), second-order (two pixels) and higher-order (three
or more pixels) statistics. The basic difference is that first-order statistics estimate
properties of individual pixel values, ignoring the spatial interaction between image
pixels, whereas second- and higher-order statistics estimate properties of two or more
pixel values occurring at specific locations relative to each other 12- 14 •
The success ofthe GLCM method depends on the choice of the distance and the
direction between the pixels, and the window size. The appropriate distance between
pixels depends on how fine or coarse the texture of interest is. Small distances are
usually used for fine textures since pixels close to each other will present enough
variation in their grey values to characterize these textures, whereas greater pixel
distances are generally used for more coarse textures because variations in the grey
values occur in pixels farther away from each other. It has been found that small
distances produce the best results 20 since they are appropriate for textures that are fine,
as well as for those that are coarse. A distance of 1 pixel means that the pixels in the pair
will be located right next to each other.
Concerning the direction between pixels, there are four directions that can be
used: 0° (horizontal), 45° (diagonal), 90° (vertical) and 135° (diagonal); however,
selecting which direction to use can be difficult. One method consists of calculating the
GLCM features for all four directions and then taking their averages 10 • The most
common choice for the direction between pixels found in literature, however, is 0°; this
means that the pixels in the pair will be located horizontally from each other.
The classification accuracy using texture features also depends on the size of the
window used. This refers to the size of the pixel neighbourhood around a central pixel
that represents a certain area in the image to be analyzed at a time; after analysis of this
area, the window is shifted by one pixel to the next area in the image for analysis. A 3x3
window for example, represents an area in the image that is 3 pixels x 3 pixels. If the
window is too small or too large relative to the texture structure, then real textural
properties will not be accurately reflected 12• 14 • One method that can be used to choose
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
134 Kabir et at.
the appropriate window size is based on the calculation of the variation coefficient for
each object class (texture) as a function of the size of the window, using any given
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
texture feature 21 • The appropriate window size will be that for which the variation
coefficients start to stabilize for the majority of the object classes, while having the
lowest value.
The MLP network (Fig. 2) is usually composed of three layers: an input layer, a
hidden layer and an output layer. In the input layer, the number of nodes corresponds to
the number of input features. The number of nodes in the output layer corresponds to the
number of target classes. Typically, an input signal travels from the input layer to the
hidden layer, where it passes through an activation function to compute the output from
the hidden nodes. This output signal from the hidden nodes becomes the net input signal
to the output nodes, where it again passes through an activation function to compute the
predicted output. Various activation functions can be used in an MLP network: logistic,
Gaussi'an, linear, hyperbolic and threshold. Although it is possible to use different
activation functions in calculating the output from the hidden layers and the output
layers, usually the same function is used. In an MLP network, the activation function
most commonly used is the logistic function, since it introduces non-linearity into the
network25 •
One of the significant advantages of neural networks lies in their ability to model
extremely complex functions, without the need of computer algorithms. They can model
non-linear functions with large numbers of variables, and extract patterns and detect
trends that are too complex for algorithms 28 • Another advantage of ANNs is that they are
more tolerant of imperfect data such as missing values or other data quality problems.
Also, due to their parallel distributed processing structures, ANN functions are carried
out simultaneously, thereby gaining enormous increases in processing speed 26 •
Three concrete blocks, referred to as Dl, D2 and 03, which were exposed
outdoors to the elements for over ten years, were used as the samples in this study. Grey
scale images of these blocks were taken and processed using Harr's wavelet transform
and the GLCM texture analysis methods and then classified through the ANN technique
in order to characterize and quantify crack damage. These different techniques can be
applied through a variety of image processing software; for the purposes of this study,
the Matlab software and ENVI image processing system were employed.
For further textural discrimination between the narrow and wide cracks in the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
concrete surface, the statistics of the wavelet detail coefficients can be employed. As a
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
136 Kabir et al.
result, first-order statistics were computed from the histogram of each reconstructed
image. Although a range of first-order texture features can be used, the following five
were derived for this study: data range, entropy, mean, skew and variance; these first-
order feature images represent the wavelet histogram signatures.
A further improvement of the detail image statistics may be obtained from the
wavelet co-occurrence signatures, which reflect the coefficients' second-order statistics.
In order to determine which window size to use for calculation of the second-order
statistics, the homogeneity feature was randomly chosen to establish the variation
coefficients; a 3x3 pixel window size was found to be the most appropriate. Thus, the
GLCM of each reconstructed image was computed based on a distance of 1 pixel and a
direction of oo between pixels, and a window size of 3x3 pixels. As many as fourteen
second-order texture features can be derived from the GLCM; in this study, however, the
following eight features were extracted: contrast, correlation, dissimilarity, entropy,
homogeneity, mean, second moment and variance. These second-order texture images
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
constitute the wavelet co-occurrence signatures.
Since many of the texture features obtained from the first-order and second-order
statistics are redundant and capture similar concepts, features that do not help
discrimination should be discarded 12. 13 • The most effective features are selected through
a process consisting of visual analysis, histogram analysis, and analysis of correlation
matrices. For this step, only the feature images produced for the D3 sample were used in
the analysis since this block presented the most textural variation.
Visual analysis of the feature images produced from the first-order statistics
revealed that the entropy and skew features did not provide enough visual information.
This was confirmed through analysis of the histograms and the correlation matrices, and
so the entropy and skew features were subsequently discarded. Thus, three first-order
features, data range, mean and variance, were selected for use in this study. The feature
images for the D3 sample can be seen in Figure 4; data range (Fig. 4(a)) presents the
details of the wide cracks, the texture of the narrow cracks are extracted in the image of
the mean feature (Fig. 4(b)), and the variance feature image (Fig. 4(c)) highlights the
background or no-crack area.
After extraction of the textural information from the images, the next step
consists of classifying and quantifying the different classes of texture. For the purposes
of this study, a supervised neural network classification approach was used to extract the
crack patterns from the concrete imagery. The MLP architecture used in this study was
composed of seven input nodes to represent the following seven input features: the
original grey scale image (Fig. 7(a)-(c)), the three first-order texture features and the
three second-order features. Three output nodes were used to correspond to the
following three target classes: wide crack, narrow crack and no crack. The logistic
activation function, which is the most commonly used in an MLP network, was used in
this study. For the training algorithm, the error back-propagation method is the most
commonly used for the MLP, which was used in this study also. The training rate
weights the change in the connections between the network layers. The training
momentum is a term that changes the connections in the direction of the average
gradient. The training threshold value indicates when the training should stop and the
number of training iterations indicates the number of iterations the training procedure
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
should perform. To select appropriate values for these variables, tests can be done by
varying the values and selecting the ones that provide the highest testing accuracies 29 . In
this study, a training rate of 0.2, a training momentum of 0.9, a training threshold of 0.9
and 1000 training iterations were used.
In order to train the MLP, training data and testing data were selected from the
original input images. After regions of the image were selected as training data and
separate areas were selected as testing data, the software implemented the MLP, which
then performed a classification using the seven input features. This was done for images
of each of the block samples D 1, D2 and D3.
The results of the classification are presented in two forms: a classified image,
also known as a thematic map (Fig. 8(a)-(c)), showing the spatial distribution of the
various classes in which each pixel is assigned a symbol or colour that relates it to a
specific class, and a table that summarizes the number of pixels in the whole image that
belongs to each class (Table 3). At the completion of a classification process, it is
necessary to assess the accuracy of the results obtained. The Kappa coefficient, which
was found to be the most appropriate index to provide classification precision 29 , was the
method adopted in this study. Table 4 shows the overall classification accuracies
obtained for each class, as well as the Kappa coefficients and overall accuracies for each
sample.
For the sake of evaluating the effectiveness of the combination of the first-order
and the second-order feature sets, the classifications were also carried out with only the
second-order features. This step demonstrated that discrimination is improved with the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
138 Kabir et al.
combination of the two feature sets as the classification accuracies were higher with the
combined set, which had overall accuracies of 86.3%, 84.2% and 87.5% forD 1, D2 and
D3 respectively compared with 72.3%, 76.5 %and 75.4% for the second-order feature
set only (see Table 4).
In order to quantify the total length of wide cracks, the number of pixels along
the length of each branch of the cracks was summed and the total multiplied by the pixel
resolution of 0.26 mm. Figure 9 shows the wide cracks present in the D3 sample for
which a total crack-length of 237.4 mm was calculated. For the D2 sample, the total
length of wide cracks was found to be 97.6 mm, and for the D 1 sample, the total length
of wide cracks was 0 mm.
These findings are supported by in-situ data recorded for the three blocks.
Laboratory data (Table 1) show that the D3 sample was prepared with the highest alkali
content, and the Dl sample with the lowest. This corresponds well with the higher
percentage of damage found in the D3 sample and the lowest percentage found for the
D 1 sample. The highest values for the total length of wide cracks as well as for the
average width of cracks found for the D3 sample also relate well to its having the lowest
average for the P-wave velocity tests (see Table 2), indicating the presence of the most
deterioration. The absence of wide cracks in the D 1 sample, which had a value of 0 mm
for the average width of cracks, as well as for the total length of cracks, corresponds
well to the lower percentage for average expansion levels measured on the blocks, where
the D2 sample showed a higher percentage, and with the D3 sample having the highest
percentage of expansion (see Table 2).
CONCLUSION
Acknowledgments -The authors wish to thank Dr. Benoit Fournier from CANMET for
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
allowing them to use the concrete specimens. This project is supported by a grant from
the Fonds quebecois de recherche sur Ia nature et les technologies (FQRNT). Special
thanks to Groupe de recherche sur !'auscultation et !'instrumentation (GRAI) for
supplying the image data as well as the laboratory and in-situ data.
REFERENCES
4. Chong, K.P., Scalzi, J.B. and Dillon, O.W., "Overview of non-destructive testing
evaluation projects and initiatives at NSF", Journal of Intell. Mater. Syst. Struct.,
1990, vol. 1, pp. 422-431.
7. Weil, G.J. and Rowe, T.J? "Nondestructive testing and repair of the concrete roof
shell at the Seattle Kingdome", NDT&E International, 1998, vol. 31, no. 6, pp.
389-400.
9. Livens, S., Scheunders, P., Van de Wouwer, G., Van Dyck, D., Smets, H.,
Winkelmans, J. and Bogaerts, W., "A texture analysis approach to corrosion image
classification", Microscopy, Microanalysis, Microstructures, 1996, vol. 7, no. 2,
pp. 143-152.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
11. Van de Wouwer, G., Scheunders, P. and Van Dyck, D., "Statistical texture
characterization from discrete wavelet representations", IEEE Trans. Image
Processing, 1999, vol. 8, no. 4, pp. 592-598.
12. Schowengerdt, R.A., "Remote Sensing: Models and Methods for Image
Processing", 2nd edition, San Diego, Academic Press, 1997.
13. Richards, J.A. and Jia, X., "Remote Sensing Digital Image Analysis: An
Introduction", 3rd edition, New York: Springer-Verlag, 1999,363 p.
15. Pickett, R.M., "Visual analysis of texture in the detection and recognition of
objects", Picture Processing and Psychopictorics (B.C. Lipkin and A. Rosenfeld,
eds.), Academic Press, New York, 1970, pp. 298-308.
17. Tuceryan, M. and Jain, A.K., "Texture analysis", Handbook of Pattern Recognition
and Computer Vision, 2nd edition, (C.H. Chen, L.F. Pau and P.S.P. Wang, eds.),
World Scientific Publishing Co., 1998, pp. 207-248.
18. Masad, E., Muhunthan, B., Shashidhar, N. and Harman, T., "Internal structure
characterization of asphalt concrete using image analysis", ASCE Journal of
Computing in Civil Engineering, 1999, vol. 13, no. 2, pp. 88-95.
19. Mallat, S.G., "A theory for multiresolution signal decomposition", IEEE Trans. on
Pattern Analysis and Machine Intell., 1989, vol. 11, pp. 674-693.
20. Karathanassi, V., Iossifidis, C.H. and Rokos, D., "A texture-based classification
method for classifying built areas according to their density", International Journal
ofRemote Sensing, 2000, vol. 21, no. 9, pp. 1807-1823.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 141
21. Anys, H., "Reconnaissance des cultures a l'aide des images radar: approche multi-
polarisation et texturale", These de doctorat, Universite de Sherbrooke,
Sherbrooke, QC, Canada, 1995,241 p.
22. Lek, S. and Guegan, J.F., "Artificial neural networks as a tool in ecological
modeling: An introduction", Ecological Modeling, 1999, vol. 120, pp. 65-73.
23. Peng, C. and Wen, X., "Recent applications of artificial neural networks in forest
resource management: an overview", Environmental Decision Support Systems
and Artificial Intelligence, (U. Corte and M. Sanchez-Marre, eds.), Tech. Rep. WS-
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
24. Sarle, W.S., "Neural networks and statistical methods", Proc. of the 191h SAS Users
Group Int. Conf., SAS Institute, Cary, NC, 1994.
25. Liu, C., Zhang, L., Davis, C.J., Solomon, D.S., Brann, T.B. and Caldwell, L.E.,
"Comparison of neural networks and statistical methods in classification of
ecological habitats using FIA data", Forest Science, 2003, vol. 49, no. 4, pp. 619-
631.
26. Awcock, G.W. and Thomas, R., "Applied Image Processing ",McGraw-Hill, New
York, 1996, 300 p.
27. Haykin, S., "Neural Networks ", 2"d edition, Prentice Hall, 1999.
28. Bishop, C.M., "Neural Networks for Pattern Recognition", Clarendon Press,
Oxford, 1995.
29. Fung, T. and Ledrew, E., "The determination of optimal threshold levels for
change detection using various accuracy indices", Photogrammetric Engineering
and Remote Sensing, 1988, vol. 54, no. 10, pp. 1449-1454.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
No Crack 234 278 89.37
'
H- LH2
HL,
HH2
HL,
LH 1 HH,
(a) (b)
w 00 ~
Fig. 3- (a) 512 x 512 Raw grey scale image of D3 (b) Complete decomposition
(c) Reconstructed image
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
(a)
Fig. 6 -Texture neo-channels (a) first-order DMV and (b) second-order MHO
Fig. 7- 512 x 512 Raw images of samples (a) 01 (b) 02 and (c) 03
Fig. 8 -Thematic maps of the classifications for samples (a) 01 (b) 02 and (c) 03
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 10 - Crack width measurements for 03; each square represents one pixel at a
resolution of 0.26 mm
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Alkali-Silica Reactions in
Concrete- Relationship between
Water Content and Observed Damage
on Structures
Synopsis: Water is generally accepted to be one of the main factors affecting Alkali-
Silica Reactions (ASR) in concrete. The water content in ASR affected structures is
normally expressed as relative humidity (RH). However, the measurement of RH
is notoriously very difficult and uncertain, particularly in the field. The degree of
capillary saturation (DCS) may be a more suitable parameter to characterize the water
content and the progress of damage on structures due to ASR. The relation between
the RH in concrete and the DCS varies depending on several factors, where the water/
binder-ratio is the most important one.
For more than 100 Norwegian concrete structures the DCS has been determined on
specimens cut from drilled concrete cores, according to a special procedure. The
water content has been determined at depths in the range from about 100 to 400
mm from the surface, depth ranges where it is considered to be rather stable and
uninfluenced by seasonal changes in the weather.
The results show a good correlation between the presence of ASR in a concrete
structure and the DCS. With only a few exceptions the degree of capillary saturation
of the concretes with pronounced ASR is higher than 90 %. The extent of damages
generally increases with increasing water content above this level.
Eva Rodum (M.Sc. Eng) is Research Engineer at SINTEF Concrete, Trondheim, Norway
Bard Pedersnen (PhD) is technical manager at one of the largest aggregate and concrete
producers in Norway (NorStone I NorBetong).
INTRODUCTION
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Description of the moisture conditions in concrete
Hedenblad2 reported that this chemical effect on the pore water can be as much as
5 % reduction of the RH, i.e. a fully saturated concrete only exerts 95 % RH. This effect
is present in Fig. 1, where the storing in chloride solution moved the sorption isotherm
curves to the left in the figure. Also carbonation and increasing temperatures affect the
sorption isotherm of concrete.
The degree of saturation (DS) as defined in eq. 2 is a direct measure of the degree
of pore filling and includes micro pores as well as macro pores, i.e. the whole concrete
porosity.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Weight after pressure saturation -Weight after drying at 105° C
A more common expression for the moisture level that does not include filling of
macro (air) pores, is the degree of capillary saturation (DCS) as defined in eq. 3.
It is known that severe drying both changes the pore structure and introduces
micro/macro-cracks. However, it has been shown that this does not influence pore
volums significantll.
In general the relation between the RH in concrete and the DCS varies a lot
depending on several factors, where the water/binder-ratio is the most important one.
Examples of first desorption isotherms for two concretes of different w/b-ratios are
shown in Fig. 24•5 • Note that the sorption isotherm is significantly higher for the low w/b-
ratio concrete compared to the high w/b-ratio concrete due to the difference in pore
structure. Also note that the RH in the range> 75 % is very sensitive to a change in the
moisture content for the concrete with low w/b-ratio.
between the sensors and the concrete often is the cause of unreliable results. There are
different types and brands of RH-sensors on the market, with great variations in quality.
Proper calibration is very important and will minimize the variations. A great advantage
of installing sensors in structures is the possibility of automatic monitoring seasonal
changes of RH. However, one does not know of any cases where this has proven to be
reliable.
Laboratory measurement of DCS on field samples is done by weighing
immediately after unwrapping, after saturation in water and after drying. DCS is then
calculated in accordance with eq. 3. The sampling is of course very important, and care
should be taken to ensure that the results are valid for the part of the structure and the
actual depth (see later discussion). DSC measures, in contradiction to RH measures, are
easy to perform accurately, but they are destructive and cumbersome to do, involving
cutting samples from the structure and taking them to a laboratory in an "undisturbed"
condition (see later discussion).
SCOPE OF WORK
This paper discusses the influence of water content, expressed as the degree of
capillary saturation (DCS), on the presence of ASR and extent of damage observed on
Norwegian concrete structures. As a basis, results from research dealing with the
following issues are presented:
c:tr How to take "undisturbed" samples from a structure for measurement of DCS (i.e.
samples describing the in situ moisture state in the concrete structure)
c:tr Experience from field survey of several concrete structures
c:tr Measurement of in situ moisture state in several Norwegian concrete structures,
mainly bridges, with- and without ASR
Results from some other Norwegian projects are also presented, among these:
o A student work at N1NU (The Norwegian University of Science and Techno-
logy, Trondheim, Norway) dealing with "undisturbed sampling" as basis for
measurement ofDCS 11
o Comprehensive laboratory investigations performed by SINTEF on two
Norwegian concrete dams with ASR (not published yet)
One method used in some Norwegian research projects the last ten years for
''undisturbed" sampling, is by dry cutting samples (prisms) using a diamond saw4 •
Normally the sawn prisms are about 40-50 mm deep, 150-200 mm long and 20-30 mm
thick, see Fig. 3. After cutting, the prisms (normally 3-4) are chiselled off and
immediately sealed tightly in plastic bags to prevent moisture loss. When the samples
arrive at the laboratory each piece of sample is unwrapped, dust is removed by brushing,
and then the samples are weighed. The samples are placed in water until equihorium
weight is achieved, before they are finally dried at 105°C. The equilibrium suction weight
is assumed to represent a state when all pores except the air/macro pores are filled. The
DCS is then calculated according to eq. 3. This procedure is often followed by the ''PF-
method"3, where the samples are weighed at different moisture states to characterize both
gel/capillary porosity and air/macro porosity.
After unwrapping the samples in the laboratory, parts of each sample are in some
cases crushed and placed in glass tubes for RH-measurements by use ofRH probes, thus
both the in situ DCS and the in situ RH are measured.
Sellevold1 found that the measured moisture contents were independent of the
ratios between the sawn surface areas and the specimen volumes, and thus concluded that
the samples did represent the "in situ" moisture content of the field concrete. In a student
work11 , however, it was concluded that the time of the cutting process should be
minimized in order to prevent loosing significant moisture content.
Some other disadvantages with this sampling method should be mentioned:
o The sampling method cannot be applied when it is raining (i.e. the concrete
surface should be dry and no water should be added to the concrete)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Norwegian coastal bridges show that the moisture state in only the outer few mm
of uncracked parts of a concrete will vary due to changing weather conditions.
Carbonation depths in these structures were also generally only a few mm, which
also indicates a stable and relative high moisture content deeper in than a few
mm'.4
o The moisture state in the interior of the concrete is not possible to measure by use
of this sampling method.
In the Norwegian research project regarding field experience with AAR7•8 cores
were drilled (with water cooling) from about 80 concrete structures for laboratory
investigations. To be able to measure the moisture state in the interior of the structures,
one decided to combine the sampling for measurement ofDCS with this core drilling.
error in the measured moisture state, i.e. to a less than 1 % DCS difference in the
measured moisture content in the samples compared to the estimated "true" in situ
moisture state in the concretes (based on a common evaluation of the moisture
state measured by use of the two sampling methods):
Concrete quality (w/b-ratios varying from 0.43 to 0.68)
Diameter of the drilled cores (70, 100 and 150 mm, respecively)
Time from sampling to unwrapping and measuring in the laboratory (varying
from some hours to about one month) - provided that the samples are sealed
well after sampling
o The other two parameters varied; drilling time (i.e. time for water exposure -
varying from about 2-15 minutes) and moisture content of the concretes (varying
from 80-100% DCS), lead in some cases to some larger errors:
The maximum difference found between measured DCS on samples collected
according to the two sampling methods was about 3 % (for a concrete with a
moisture content of about 80 % DCS and a period of water exposure during
drilling of about 15 minutes).
The overall conclusion from the work is that in the moisture range 80-100 % DCS
(relevant for most outdoor Norwegian concrete structures), the water cooling did not lead
to any significant increase in the water content. For most samples the increase in water
content due to water cooling was in the range 0.5-1 %in the inner part of the drilled cores
provided the above described procedure was followed. The possible increase in water
content due to water cooling is, however, supposed to increase with increased time of
drilling, decreased water content (below about 80-85 % DCS) and increased extent of
cracks along the outer surface of the drilled cores (for instance due to a high extent of
ASR).
Field objects
In the research project which produced most of the results presented here 7•8,
several hundred concrete structures were planned to be visually inspected in the field.
Project information and specifications for a large number of relevant structures were
collected and evaluated. Experience from preliminary studies showed, however, that such
information received from owners, concrete- and aggregate producers was too uncertain
to be used as documentation without performing supplementary laboratory investigations
on drilled cores.
To be able to fund investigations on such a large number of structures, a co-
operation was established with the Norwegian Public Road Administration in the South
Eastern part of Norway. Totally about 160 concrete structures, mainly bridges, were
visually inspected in the field. Most of the inspections and sampling were performed by
the personell from the Norwegian Public Road Administration in connection with their
regular inspections of their bridges (based on procedures prepared by the research
project).
Due to variation in damage and a possible variation in concrete recipes and
moisture content between different parts of the structures, only selected structure
members were chosen for further investigations. The selection of these structure members
was based on the following criteria, i.e. the following parameters were aimed to be
varied:
o Structures with - and without possible ASR
o Different structure parts (bridge towers, abutments, foundations, columns,
superstructures, retaining walls, culverts)
o Age of the structures (must be so high that a possible development of ASR can be
revealed during the investigations - the age of most of the structures selected were
in the range 25-45 years)
o Wide spread in concrete compositions, i.e. w/c-ratio, strength, concrete alkali-
content and aggregate type (i.e. different reactivity potential of the aggregates)
o Access to external water supply (rain, suction from the ground) or proteced from
external water supply.
Half of the about 160 visually inspected concrete structure members (normally
one from each concrete structure) were selected for sampling. Three cores (diameter 100
mm and length about 300 mm) were drilled out from the same part of each of these
structure members, all by use of water cooling.
The concrete cores from 46 of these structures members were examined in detail
in the laboratory, including the following methods/parameters:
o Visual inspection of the concrete cores (possible signs of ASR, i.e. depth and
widths of cracks in the concrete surface, alkali gel in pores or in connection with
aggregates, ..... )
o Degree of capillary saturation (DCS - see description of the method in an earlier
papagraph of the paper)
o Porosity (gel/capillary and air/macro porosity according to the PF-method3
o Degree of damage in the drilled cores (according to a method developed in the
research project. Based on counting of different cracks in the plane polished
sections of half cores a "Crack Index" was defined 12)
o Concrete compositon, i.e. cement content (basis for calculation of the alkali
content) and petrographic description of the aggregate types (basis for evaluation
of the alkali reactivity potential of the aggregates).
The concrete cores from the remaining structure members were only investigated
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
according to the first three methods, i.e. visual inspection, degree of capillary saturation
(DCS) and porosity.
The general impression from field surveys is that the extent of map cracking is
considerably higher on structure members with abundant water supply compared to
structure members periodically exposed to rain. For structure members totallysheltered
from water/rain no damaging ASR was detected. Examples of these findings are given in
Fig. 6-8. Fig. 6 shows a bridge column where map cracking only occurs on the outer parts
of the column which are periodically exposed to rain. Fig. 7 shows an abutment of a
bridge, where severe map cracking only are present on the part of the abutment which, in
addition to rain, is exposed to leakage water from the paving of the bridge (due to a poor
drainage system).
Fig. 8 shows a wide column on a plate dam about 40 years old, one part being
sheltered from rain (behind a frost wall - only exposed to the rather high humidity in the
air) and one part exposed to rain (outside the frost protection wall downstream). No map
cracking occurred on the sheltered parts of any of the columns, while map cracking due
to ASR in an early stage was detected on exposed parts of most of the columns. The
mean DCS measured in the interior of the concrete on the sheltered parts of the columns
(inside) and the exposed parts (outdoor) was 85 % and, 93.5 %respectively- see Table
2. Similar observations and measured DCS values were also detected in the main research
project for some culverts. The sheltered parts (well drained walls) had no map cracking
and DCS < 85 %, while the exposed wings often had map cracking due to ASR and a
considerable higher moisture content (DCS > 90 %).
These general findings in the field are confirmed by laboratory investigations on
the drilled cores. Figs. 9 and 10 present results from measurements of DCS versus extent
of cracks in the drilled cores, described by the "Crack Index" 12 • Increased "Crack Index"
correlates to an increased extent of ASR induced cracks in the cores. Fig. 9 presents
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
As seen in Fig. 9 several concretes with DCS > 90 % had not developed ASR even
after several decades of exposure. The main reason for this was due to use of a non
reactive concrete composition, i.e. either an alkali content below the threshold value
needed to initiate ASR and/or use of a non reactive aggregate type.
CONCLUSIONS
A review on laboratory work and reported field cases show that most researchers
tend to use relative humidity (RH) as the only parameter for moisture level, also with
respect to ASR. The authors of the current paper have long experience with measurement
of moisture content in Norwegian outdoor concrete structures, both in terms ofRH and in
terms of degree of capillary saturation (DCS). On this basis it is recommended to increase
the use of DCS to express the moisture state in the concretes in connection with condition
surveys, also for ASR affected structures. This is not the conclusion based on
fundamental considerations, but based on practical experience. The measurement of RH
is notoriously very difficult and uncertain, particularly in the field. DCS measures, in
contradiction to RH measures, are easy to perform accurately, but they are destructive
and may be more cumbersome to do.
The results show a good correlation between the presence and the extent of ASR
in a concrete structure and the DCS. With only a few exceptions the degree of capillary
saturation of the concretes with pronounced ASR is higher than 90 %. The extent of
damages generally increases with increasing water content above this level.
REFERENCES
Table 1- Water content measured as DCS in the interior(depth > 200 mm) of73
Norwegian concrete structure members, mainly bridges 7 ~.
DCS Stdev
Structure I structure member Number (%) (%)
Bridges
Abutments 23 91.4 4.7
Superstructures 6 90.8 5.7
Columns 14 90.2 5.5
Towers 4 89.0 3.3
Sidewalls 8 95.3 3.3
Dams
Walkways 4 92.5 3.9
Columns 3 90.2 3.1
Culverts 8 92.4 3.5
Retaining walls 2 100.0 0.0
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
DCS Stdev
Structure I structure member Number (%) (%)
Dams (with ASR)
Columns - exposed to rain 3 93.7 2.1
Columns - sheltered from rain 6 85.0 4.5
Plate- downstreams (some parts
exposed to rain in addition to water 21 93.2 4.5
suction through the plate)
Massive dam (frost damae;e)
Downstreams 7 97.6 1.0
Up streams 5 96.4 1.9
Table 3--Water content measured as DCS on dry cut samples (depth 0-40/50 mm) from
concrete foundations for transmission lines with steel towers 13 •
5
4
80
60
40
20
0
0 20 40 60 80 100
Relative Humidity(%)
Figure 2--Desorption isotherms for concrete discs by the degree of capillary saturation
(DCS) as a function of RH for two laboratory concretes of differenct w /c-ratio4.
(The concrete with w/c-ratio = 0.42 contained 5% silica fume).
Figure 3--"Undisturbed" sampling by dry cutting. The arrow shows two pieces of
concrete already sampled.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
100~----------------------------~~
95
90
85
80
75
70
65
60
55
50
Main project -mainly bridges
73 single specimens
Figure 4--0verview- DCS measured in the interior (depth >200 mm) of the 73 concrete
structure members, mainly bridges, included in the main research project?. 8 •
DCS (%)
100~------------------------------~
95
90
85
80
75
70
65
60
55
50
Three dams
42 single specimens
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Figure 7--Example of the influence of access to external water supply on the extent of
map cracking due to ASR. Note that severe map cracking only is present on the part
of the abutment with plentiful access to external moisture supply from leakage water
from the paving of the bridge (due to a poor drainage system) in addition to rain.
100
(~·
• •• •
I
95 ·-
~ 90
~
Q 85
~
,;;
.
~
~~ fa
1"". •
~
~
80
75
0 5 10 15 20 25 30 35 40 45 50 55 60
"Crack Index"
Figure 9--DCS versus extent of crack in the drilled cores, described by the
"Crack lndex"' 2 • Measurements performed on 46 concrete structure members,
mainly bridges, included in the main research projecF·8 •
95 •'fA• l
I I
~
·.. /
rn
g
90
85
A
•i
~I
80
I
75 I
0 5 10 15 20 25 30 35 40 45 50 55 60
"Crack Index"
• Mlderate-large extent of />SR & Small-moderate extent of />SR
+No-small signs of />SR
Figure 10--DCS versus extent of crack in the drilled cores, described by the
"Crack lndex" 12 • Measurements performed on two Norwegian concrete dams
partly map cracked due to ASR.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: Self compacting concrete (SCC) can be placed without any compaction,
avoiding some health risks as well as environmental problems. The two essential
properties of SCC are a high flowability and a high segregation resistance, obtained
by the use of either a large amount of fine particles (P) or a viscosity modifying
admixture and a superplasticizer. Already there is a lot of knowledge about
composition and workability of SCC, however there are questions regarding long·term
durability due to significant difference in the mix proportions of sec in comparison
to traditional concrete. The degradation mechanisms of cementitious materials are
greatly influenced by the penetration ability of aggressive fluids, and there is an
important relation between the 'pore structure' of solids, fluid transport properties
and degradation. lf the pore structure of sec turns out to be different from traditional
concrete, some changes in durability behaviour might be expected. An experimental
program was set up to study the pore structure of self·compacting concrete. Mercury
intrusion porosimetry (MIP) was used as a testing method. As MIP can use only
small-size specimens, it is customary to study hardened cement paste specimens
with similar W/C ratios and curing ages as in actual concrete. In this study tests were
performed on samples of hardened cement paste of several ages in order to evaluate
the relative influence of various parameters on both traditional and self compacting
concrete mixtures.
INTRODUCTION
Self compacting concrete (SCC) can be placed without any compaction, avoiding some
health risks as well as environmental problems. The two essential properties of SCC are a
high flowability and a high segregation resistance, obtained by the use of either a large
amount of fine particles (P) or a viscosity modifying admixture and a superplasticizer.
Already there is a lot of knowledge about composition and workability of SCC, however
there are questions regarding long-term durability due to significant difference in the mix
proportions of SCC in comparison to traditional concrete. The degradation mechanisms
of cementitious materials are greatly influenced by the penetration ability of aggressive
fluids, and there is an important relation between the 'pore structure' of solids, fluid
transport properties and degradation. If the pore structure of sec turns out to be different
from traditional concrete, some changes in durability behaviour might be expected.
EXPERIMENTAL PROGRAM
Mixture design
The mixture proportions and parameters of 11 cement pastes are presented in Tables 1a
an 1b. The mixture proportions of the cement pastes are based on the mixture design of
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 169
sec and traditional concrete on which. already many tests on durability and transport
properties have been performed (l, 2 ' 3l. The proportion of materials used for making the
cement paste listed in Table la correspond to the amounts of constituents in 1 m3 of
concrete. In Table 1b the proportions of cement paste are listed. All cement pastes were
made with Portland cement type CEM I 52.5 N according to EN-197-1 Standard
Specification. Pastes PTl and PS 1 are reference materials for traditional concrete and for
self-compacting concrete, respectively, with a cement content of360 kg/m3 and a W/C of
0.46. Limestone filler and a polycarboxylic ether superplasticizer (PCE) were used for the
cement pastes representing the pastes of self-compacting concrete. The amount of water
was varied to achieve W/C of 0.40 (PT2, PS2), 0.46 (PTl, PSI), 0.50 (PT3, PS3) and
0.55 (PT4, PS4), keeping the cement and powder content constant. The C/P varied from
0.50 (PSS), 0.60 (PSI), 0.67 (PS6) to 0.75 (PS7), keeping the powder content and water
content constant. The average particle size of limestone filler was slightly smaller than
that of the cement used. The minimum, maximum and average size is respectively 0.2
Jlm, 80 Jlm and 15 11m for cement and 0.2 Jlm, 100 11m and 10 11m for limestone filler.
The PT cement pastes were made by mixing cement and water at a low speed for 2
minutes and at a moderate speed for another 2 minutes. For the PS cement pastes, first
cement and limestone filler were blended together and then mixed at the low speed for 1
minute, and for one more minute after the addition of water. After adding the
superplasticizer, the paste was mixed at the moderate speed for 3 minutes. The cement
pastes were then cast into 500 ml plastic bottles and the bottles were rotated at a speed of
5 rpm for 24 hours and a constant temperature of 20°C. After rotating the bottles, they are
further stored until the time of testing in a control-climate room at 20°C. By rotating the
bottles with cement paste, the influence of bleeding on the pore structure of the cement
paste was avoided. However, there are certain limitations of this procedure that are
discussed in concluding remarks of this paper. At 7, 14 and 28 days, prisms of 12x12x20
mm 3 were sawn from the hardened cement pastes and dried in an oven of 105°C until
constant mass. During the experiments, one of the samples has been split and it was
observed that the mercury was well distributed over the whole sample.
Mercury intrusion porosimetry is used to determine the pore size distribution of cement
paste. Pressurized mercury, being a non-wetting fluid, enters the sample of cement paste.
The intrusion pressure gets increased and the corresponding volume of mercury injected
into the pores of a previously dessicated and degassified sample is recorded (intrusion).
This volume is obtained by measuring the variation of the mercury level in a calibrated
capillary tube called the dilatometer. During depressurizing, the change of volume is still
being recorded and gives data about the extrusion of mercury out of the sample
(extrusion). The classical method of analyzing mercury intrusion pressure volume curves
is based on the model of parallel cylindrical non-intersecting pores of differing radii. The
relation between the radius (r) of pores being intruded and the intrusion pressure (PHg) is
given by the Washburn equation
P = 2ycose
Hg r --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The volume pore size distribution Dv, defined as the pore volume per unit interval of pore
radius, can be deduced from the volume of mercury that intrudes within the pressure
range (P; P + 8P) corresponding to a pore radius range (r; r + 8r). (S) Instead of Dv, one
can calculate the volume log radius distribution Dv(log r), which can be expressed as
dV
Dv(logr)=--
dlogP
In this paper Dv (log d) is plotted. The equivalent pore size corresponding to the biggest
peak in the volume pore size distribution is referred to as the critical pore size, which is
the maximum pore diameter at which the continuous mercury intrusion begins. This
critical pore size might give an indication about the permeability of the cement paste <6l.
In Figs. 3, 5, 7 and 9 the graphs of the volume pore size distribution of all the cement
paste are plotted. As can be seen from these Figs., the pore diameters are mainly
distributed over a range from 10 nm to 1 1-1m. Based on a classification of pore size in
hydrated cement paste, this range covers the medium (10-50 nm) and large capillaries (50
nm - 1 )lm), both of which affect the strength and permeability <6l. As such, the
infonnation obtained on the pore structure falls within the field of interest.
Plotting the cumulative volume of mercury versus the pore diameter gives the intrusion
and extrusion curves (Fig. 1). In Fig. 1 the cumulative volume is plotted against the
pressure, which is related to the pore diameter. These curves make it possible to make a
rather good comparison between several cement pastes. From the intrusion curve the total
porosity can be deduced as the total volume of intruded mercury at the maximum applied
pressure per gram of the sample. The ink-bottle porosity (ml!g) can be calculated as the
cumulative extrusion volume minus the cumulative intrusion volume at the lowest
pressure. The effective porosity can be calculated as the total porosity minus the ink-
bottle porosity. (?)Also in this graph the critical pore size can be determined, as being the
pore size corresponding with the steepest slope of the intrusion curve.
Variation of C/P
The pore size distribution and the intrusion and extrusion curves of PS5, PSl, PS6 and
PS7 at 28 days curing are plotted in Figs. 2 and 3. When the C/P decreases, the total
porosity increases (from 0.1046 mllg for PS7 to 0.1235 ml/g for PS5). Also the ink-bottle
porosity and the effective porosity do increase. The critical pore size diameter shifts
towards higher values when the C/P decreases. Decreasing the C/P and keeping the
amount of water and powder constant, gives an increase in the W/C, inducing a much
coarser pore stucture. The biggest change can be noticed for the lowest value of the C/P
(0.50), this corresponds with the highest W/C (0.55).
Variation ofW/C
In Figs. 4 and 5, the pore size distribution and the intrusion and extrusion curves of PS2,
PSI, PS3, PS4, PT2, PTl, PT3 and PT4 (increasing W/C) at 28 days curing are plotted.
There is a clear difference in between the traditional cement pastes (PT) and the self
compacting cement pastes (PS). For every value ofW/C, the porosity and the critical pore
size diameter reach higher values for the traditional cement pastes which tend to have
more water-filled pores initially. With increasing W/C, the total porosity increases from
0.0958 ml/g to 0.1346 mllg for the self compacting cement pastes and from 0.1380 ml/g
to 0.2182 ml/g for the traditional cement pastes. The limestone filler and superplasticizer
induced a finer pore structure. The fact that limestone filler, used in this project, does not
participate in the chemical reaction during cement hydration has been prooved both from
thermal analysis and BSE image analysis (SJ. So one can assume that the limestone filler
only acts as a filler in the self compacting paste and concrete, although at later ages
carbo-aluminate hydrates may form. The lower porosity induced by the presence of filler
effect might be caused because of the creation of more nucleation sites in favour of the
CH produced. The effect of a higher amount of water and as such a higher W/C, and a
higher W/P, remains the same for both types of cement paste. Increasing the amount of
water for a constant amount of powder and cement, leads to a more accessible pore
structure. When the results of the previous paragraph (variation of C/P) are compared to
the results of this paragraph, it can be noticed that the effect on the pore structure gets
more influenced when the W/C is varied by means of changing the amount of water, then
by changing the amount of cement (Figs. 2-5). The shape of the pore size distribution is
different depending on the type of cement paste. When it concerns a traditional cement
paste towards the pore sizes smaller than the critical pore size diameter, a more important
increase in volume is observed in the curve.
The influence of the period of curing has been investigated for all mixtures. In Figs. 6-9,
the pore size distribution and the intrusion and extrusion curves of mixtures PS2, PT2
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUDING REMARK
The influence of adding limestone filler to the traditional cement paste is examined by
means of mercury intrusion porosimetry. Also the changes of the microstructure of
cement paste with additional limestone filler, due to variation of parameters like W/C,
C/P and W/P are investigated. Although the results on porosity, pore size distribution,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
mercury intrusion and extrusion on hardened cement paste produced by the method
described in this paper cannot be directly related to the hardened cement paste present in
a corresponding concrete mixture (due to the influence of the interfacial transition zone
between aggregate and cement paste), they do provide some better insight into the pore
structure of self-compacting concrete containing limestone fillers and its difference with
the pore structure of traditional concrete mixtures. The pore-size distribution of the
cement paste present in concrete is not as uniform as in the pure cement paste produced
without bleeding. The difference is mainly due to higher W/C and therefore higher
porosity of the interfacial transition zone in concrete. This phenomena is well described
in <91 .
The following specific conclusions can be drawn from the results ofthis study:
• Adding limestone powder as a filler and superplasticizer had an important
influence on the pore structure of the cement paste.
• Decreasing the C/P and keeping the amount of water and powder constant
induces a more porous cement paste.
• Increasing the W/C and keeping the amount of cement and powder
constant made the cement paste pore structure more coarse.
REFERENCES
(6) D.A. Silva, V.M. John, J.L.D. Ribeiro, H.R. Roman: 'Pore size distribution of
hydrated cement pastes modified with polymers', Cement and Concrete Research 31
(2001) 1177-1184
(7) G. Ye: 'Experimental study and numerical simulation of the development of the
Microstructure and Permeability of Cementitious Materials', Doctoral thesis Delft
2003
CEMI 52.5N 360 360 360 360 360 360 360 360 300 400 450
water 165 144 180 198 165 144 180 198 165 165 165
limestone filler - - 240 240 240 240 300 200 150
PCE - - 3.10 3.48 2.83 2.51 2.51 3.40 3.70
water/cement 0.46 0.40 0.50 0.55 0.46 0.40 0.50 0.55 0.55 0.41 0.37
cement/powder I I I I 0.60 0.60 0.60 0.60 0.50 0.67 0.75
water/powder 0.46 0.40 0.50 0.55 0.46 0.24 0.30 0.33 0.28 0.28 0.28
CEMl 52.5N 1.281 1.384 1.216 1.146 0.969 1.027 0.931 0.890 0.800 1.083 1.227
water 0.587 0.554 0.608 0.630 0.444 0.411 0.465 0.489 0.440 0.447 0.450
limestone filler - - - 0.646 0.684 0.621 0.593 0.800 0.541 0.409
PCE fg/11 - - - - 8.35 9.93 7.32 6.19 6.68 9.21 10.09
water/cement 0.46 0.40 0.50 0.55 0.46 0.40 0.50 0.55 0.55 0.41 0.37
cement/oowder I I I 1 0.60 0.60 0.60 0.60 0.50 0.67 0.75
water/powder 0.46 0.40 0.50 0.55 0.46 0.24 0.30 0.33 0.28 0.28 0.28
0.4
g; elllruslon cun.e
§. 0.35
..
c:
.9
0.3
~
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Ill 0.25 effectlw
.~ porosity
:!§ 0.2
::J
E
::1
u 0.15
0.1
0.05
0
1000 100 10 0.1 0.01 0.001
Pressure [MPa]
28days- C/P
o.25
1 ---------------;:::::::;;;:;;:;;::;l
-P$5(0.50)
-PS1 (0.60)
-+-P$6 (0.67)
_c::l"~?_iQ.J_SL
0+-------~------~-------.------~------~
0.001 0.01 0.1 10 100
Pore diameter (IJm)
1.2
-PS5(0.50)
-+- PS1
(0.60)
~PS6(0.67) CIP\,
-PS7 (0.75)
0.8
...•...
0.4 ·······················
0.2 .~
0
__,{ ~
0.001 0.01 0.1 10
Pore diameter d (~Jm)
28 days- W/C
0.25r.---------;::=~;;;:::===:::;;;:;:;;;::;;:::;-J
---PS2(0.40)
wtc! -+- PS1 (0.46)
-PS3(0.50)
-PS4 0.55)
0+-------~------~
0.001 0.01 0.1 ------~------~----~
10 100
Pore diameter (~Jm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1.2
--PT2
-+-PT1 WJC/
-PT3
-PT4
0.8
'0
g 0.6
c
0.4
0.2
0
0.001 0,01 0.1 10
Pore diameter d (llm)
W/C=0.40
0.18 ~------------;:======:;-]
0.16 +-----,-.::::------------1-PT2_7d -PS2_7d
c;; 0.14 -PT2_15d -PS2_14d
g 0.12 -PT2_28d -PS2_28d
~
~ 0.1
-~ 0.08 -l-----~~~~~-~~-----------1
1==============~~~~~~~~;;~;;::::::~
:; 0.06
3m 0.04 + - - - - - - - - - - \
0.02 + - - - - - - - - - - 1 - l l r - - - " - . : - - - - - - - - - - - - - l
0+-------~------~--~--~------~------~
0.001 0,01 0.1 10 100
Pore diameter (llm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1.6
-PT2_7d -PS2_7d
1.4
~PT2_15d -PS2_14d
1.2
- PT2_28d - PS2_28d
,
~ 1
g 0.8 --
0
0.6
~
0.4
0.2
~
0
0.001 0.01 0.1 10
Pore diameter d (IJm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
W/C=0.55
0~------~----~~~--~~----~----~
0.001 0.01 0.1 10 100
Pore diameter (IJm)
1.2
-PT4_7d -P$4_7d
~PT4_15d -PS4_14d
PT4 28d -P$4 28d
0.8
0.4 ...........
0.2 1---- ~
0
I
~
""\
0.001 0.01 0.1 10
Pore diameter d (!Jm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Alain Sellier is a Civil Engineering doctor at University Paul Sabatier and at Laboratoire
Materiaux et Durabilite des Constructions, in Toulouse, France. He holds a PhD in Civil
Engineering from Ecole Normale Superieure de Cachan in 1995. His research interests
numerical modelling of degraded concrete structures.
Ilie Petre-Lazar is a civil engineer, holding a PhD from University Laval in Quebec. He
works for Electricite de France Research Division. He is the project leader of an R&D
program dealing with the durability of concrete structures for nuclear and hydraulic
means of production.
Yann Le Pape is a civil engineer, holding a PhD from University Louis Pasteur in
Strasbourg. He works for Electricite de France Research Division. His research interest is
mainly focused on the safety and the long-term durability of concrete structures for
nuclear and hydraulic means of production.
INTRODUCTION
In order to evaluate the safety level and the maintenance costs of its dams,
Electricite de France (EDF) must get a better understanding and a better prediction of the
swelling phenomena. In this context, EDF is developing a numerical model based on the
finite element method in order to simulate AAR-affected structures.
these phenomena, basic creep in compression and tension play a significant role.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 181
MODEL DESCRIPTION
The AAR produces large concrete swelling. In absence of creep, the resulting
concrete state of strain would correspond to a very large damage state, however, damage
remains small [the mechanical characteristics decrease of 10% (D=O, 1) for a volumetric
swelling of 0.1% according to Sellier (1 )]). So we suggest to explain the compatibility
between the large AAR swelling and the associated moderate damage by long term creep
that allows a relaxation of self equilibrated stresses induced by AAR gel pressure .
Moreover, for a good long term prediction of strain induced by both the swelling
pressure of AAR and the stresses induced by other loading, it is necessary to use a
realistic creep model.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Acker (2) suggested that the basic creep of concrete is mainly due to the CSH
behaviour (CSH sliding and consolidation). This assumption is confirmed by recent
experimental evidences (3). Therefore, we develop a simple Visco-Elasto-Plastic (VEP)
orthotropic damage model including chemical AAR pressure. To model CSH sliding and
consolidation the model is separated in two levels :
An isotropic level (VEP in Fig. 1) containing a spherical part (module
VEP' in Fig. 2) and a deviatoric part (module VEPd in Fig. 2). The
spherical part is used to model the response of the CSH structure under
hydrostatic stress (consolidation) while the deviatoric part takes into
account the CSH sliding (without volumetric change) under shear stress.
A tensile stresses orthotropic relaxation level (VD' in Fig. 1), linked to an
orthotropic traction damage model of the concrete. This level has been
introduced into the model to simulate the very large orthotropic strains
noticed in free stress directions of concrete specimen submitted to AAR.
The detailed rheological model of the isotropic creep level is presented in Fig. 3
for the spherical part and in Fig. 4 for the deviatoric part. It accounts for the multiscale
porous structure of the cement paste: '1' for the micro capillary porosity and '2' for the
CSH inter layer porosity. Moreover water exchange between micro and macro capillarity
porosity is possible during the consolidation phase (link Pl-Pw in Fig. 3).
This VDt module (Fig. 5) takes into account an increase of basic creep when the
concrete is submited to tensile stresses. The tensile creep rate is given in a principal stress
direction of effective tensile stress by Eq. 1 :
.
&vp =1
- ( &o---&vp
D ) (Eq. 1)
-r0 1-D
(Eq. 2)
u = E(l-D)c (Eq. 3)
In this law, (7= the apparent stress, E =the Young's modulus and p:the deformation.
(Eq. 4)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
where:
Sr and Sr 0 are, respectively the current saturation degree, and the smallest
saturation degree necessary to allow the chemical reaction.
Concrete cylinders
Figure 6 shows the experimental principle. The parameters of the model were
selected for three experimental tests conditions (cylinder allowed to swell freely, cylinder
with uniaxial loading without confinement, and cylinder with uniaxial loading with steel
ring confinement). The Figure 7 shows the capability of this model to simulate
anisotropic swelling of confined cylinders affected by the AAR and submitted to various
axial loads.
Figure 8 shows the experimental principle. In the first 424 days of the test, the
upper face of the beam is subjected to drying conditions, this face is then saturated.
This boundary conditions lead to time dependent moisture content profiles along
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the beam's height. These profiles are estimated using a finite element formulation of the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
184 Grimal et at.
water mass balance equation (8). Results of the simulation are compared with
experimental data (see fig.9).
This moisture content profile is taken into account to compute the chemical
reaction advancement (Eq. 4). Next the pressure is evaluated (Eq. 3), and the rheological
model coupled with the orthotropic damage model gives the stress state and the damage
state. The numerical analysis was carried out using the finite element program
Code_Aster. We discuss bellow results obtained on two beams differentiated by their
reinforcement : a normally reinforced beam and a highly reinforced beam.
The highly reinforced beam (Fig. 10, bottom) has a smaller deflection during the
drying phase. In fact AAR occurs in the bottom where reinforcement limits the swelling.
Figures 11 and 12 showed the tensile damage fields 'dtx', 'dty' and 'dtz' predicted
of the normally and highly reinforced beams. We can see an anisotropic cracking pattern
(anisotropic damage field) in accordance with experimental observations reported by
Multon (7).
The highly reinforced beam shows a smaller longitudinal damage (dtx) and larger
transversal damage compared with the normally reinforced beam. In other words, the
crack anisotropy is amplified by reinforcement. This crack (or damage) anisotropy
induced an anisotropic AAR swelling, which explains the macroscopic behavior
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
differences noticed above.
CONCLUSION
The compatibility between the large strains encountered in AAR affected structure
and the moderated damage of affected concrete is explained through long-term relaxation
of tensile stresses and basic creep in compression. A rheological model and a damage
model are coupled and implemented in the finite element program Code_Aster to
simulate this phenomena. The model has been tested successfully on experimental data
for various anisotropic loading configurations and to describe the long-term behaviour of
beams subjected to AAR. The model allows especially to estimate damage and swelling
anisotropy induced by the combination of structural and environmental phenomena.
REFERENCES
3. Bernard 0., Ulm F.J., T. Germaine J., "Volume and deviator creep of
calcium-leached cement-based materials", Cement and Concrete Research,
33, 2003, 1127-1136.
4. Sellier A., Barry B., Capra B., Meftah F., "Endommagement anisotrope du
beton", Revue Franr;aise de Genie civil, Juin 2003, p. 607-620.
Total stress
r e
Isotropic level
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
kz Pw
0,20
vertical strain without vertical loading
..(),10 vertical strain with a 10 MPa compressive vertical loading nme (days
j___ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ __:_..::...:J
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
. .. _______.
-5 -4 -3 -l -I
.....
0
~
I l
0
0,05
~
,__ ;t'
0,1
,--·
- - 420"' days simulated ' .. ~
0,15
i....
O,l
·, \.• ~
• ...\
!
0,15
• 424"'days .
0,3
·~
.. 698"' days
'\
0,35
0,4
-700"' days simulated
'I
0,45
0,5
1,0
a 0,5
...
! 0,0
~ -0,5
= -1.0
-1,5
-2,0
Highly reinforced beam
j simulate • esperimentalj
1,0
0,8
0,6
I... 0,4
ll 0,2
~ 0,0
= -O,l
-0,4
-0,6
-0,8
0 0.5 z
yj_x
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Anwar Khitab received his PhD from the National Institute of Applied Sciences (INSA)
in Toulouse (France) in September 2005. His main research interest is chloride diffusion
through concrete.
INTRODUCTION
The objective ofthis work is to develop and document a method of prediction based on a
numerical model in which the impact of the surrounding ionic environment is accounted
for. Chloride is considered to be a part of an ionic solution containing mainly sodium,
potassium and hydroxide.
As in other numerical codes, input data are necessary. The special focus here is to be able
to derive the input data needed from a single sample of material, whatever its age. The
analysis of a single sample would provide a "package" of input data at a given age, while
the model can be used to determine the chloride content in the material after any time of
immersion into a chloride containing solution. The input data package consists of five
parameters, namely density, porosity, pore solution composition, chloride diffusion
coefficient, and chloride binding isotherm. In addition the specified boundary conditions
(the salinity of the exposure solution) are required.
In order to validate the model, experiments were conducted. They consisted of immersion
tests. The concrete was exposed to a sodium chloride solution (165 g/1) after a 28-day
curing period for some samples, and a 420-day cure for others, for the sake of checking
the influence of the age of concrete on chloride penetration. The effect of the exposure
time to the NaCl solution was also investigated.
EXPERIMENTS
The concrete specimens, 11-cm in diameter and 22-cm in height, were molded following
the French specifications NF P18-421 1 with a water-cement ratio of 0.4. The CEM-I
type cement chemical composition is given in Table 1. The cement, fine and coarse
aggregate (maximum diameter 8-mm) contents used are 560, 695 and 825 kg/m 3 of
concrete. The specimens were placed in a wet room for 24 hours before they were
demolded. Half of the specimens were then cured for 28 days in the same wet room. The
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
179 mol/m3 respectively for Na+, and K+ and OR. Calcium was also detected but its
concentration was lower than 0.1 mol!m3 .
After each curing period, the specimens were sawed into 50 mm thick samples from the
core of each specimen. The samples were exposed to 165 g/1 NaCl following the
specifications of the NordTest NTBuild 443 2 method (Fig. 1). The samples were kept in
immersion for 100 and 200 days. In each combination, namely 28-day cured or 420-day
cured concrete subjected to a 100-day immersion and a 200-day immersion, three
samples of concrete were tested for the sake of controlling the repeatability of the
experiments. The exposure solution was replaced periodically in order to keep the
boundary conditions constant.
At the end of the immersion period, the concrete samples were removed from the
immersion cells, and ground into powder in order to measure the chloride profiles. Both
total and free chloride profiles were measured. The standard AFREM (Association
Fran9aise de Recherche et d'Etude des Materiaux ) method 3 was applied to determine the
acid soluble chlorides defined as the total chlorides. The water-soluble chloride content
termed as the free chloride content was determined using the standard AFREM I RILEM
method 4 • The resulting solutions were then further analyzed in order to determine by
potentiometric titration the chloride concentration.
Figures 2 a-b show the free (water-soluble) chloride profiles measured after 100 days
(Fig. 2-a) and 200 days (Fig. 2-b) of immersion. Plotted on these figures are the results
corresponding both to a concrete aged of 28 days and 420 days. Whatever the time of
immersion in the NaCl solution, the penetration depth for a given time of exposure is the
same. This may be true because a one year difference in the age of the material does not
modify the material microstructure. In addition, it appears clearly from Figs. 2 a-b that
the free chloride concentration profiles are the same. Therefore no influence of the age of
concrete can be detected on the chloride penetration through the pore solution of
concrete.
In order to complete the analysis, the total chloride concentration profiles were measured.
The results are shown on Figs. 3 a-b, again for the two different times of immersion. Figs.
3 a-b illustrate the relationships between the total and the free chloride which is a way to
account for the bound chloride contents. Again, no significant difference seems to exists
between the concrete aged of 28 days at the moment of immersion and the 420 one, for
the same reason already given.
Therefore, an approximately one year difference on the age of concrete seems to have no
impact on the chloride penetration both for chloride in the pore solution and for total
chloride.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
194 Khitab et al.
MODELING
The model solves the continuity equation for the transport of ionic species (Eq. 1) written
at the scale of the material.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ac d"IVJ
p-=- (I)
at
where p is the porosity of the material, c is the concentration, and J the flux of species.
The objective of the model is to account for the electrical interactions existing among the
different ionic species present in the pore system. Consequently, the flux of species is
expressed by the Nemst-Planck equation. That kind of model belongs to the multi-species
approach models 5' 6 •
The model describes the one-dimensional transport of ionic species through a saturated
porous medium in the absence of pressure gradient. Because cementitious materials are
not inert, interactions occur between the ionic species and the solid phase. For a given
ionic species the bound amount of ion ( Cs,i ), which is obtained from experiments, is
usually expressed in mole per unit mass of material. Therefore the first term on the left-
hand side of Eq. 2 represents the variation in time of the ionic concentration in the pore
solution, while the second term is dealing with the ionic concentration bound on the solid
phase.
where the subscript i indicates the ionic species, D is the effective ionic diffusion
coefficient, z is the charge number, F is the Faraday constant, R is the ideal gas constant,
T is the absolute temperature, and <p is the electrical potential created by the interactions
among the ionic species. <p is also termed as membrane potential or liquid junction
potential.
The interactions with the solid phase are concentration dependent, hence
acB,i
- -= -acB,i
-- aci
- . In th e mo de 1, th e b"m d"mg o f one c hl on"de wn
. ts
. assume d to be
at acj at
balanced by the release of one hydroxyl ion 5. We omit the algebra and write Eq. 2 such
that two unknowns appear, the ionic concentration in the pore solution and the electrical
field:
ac acBi -1 ( a2 ci z - F a<pac.1 J
-
at
1
=D
I
(p+(l-p)p--'
aci J
-+
8x 2
---
I RT ax ax
(3)
0 = "z
L..i I c.I (4)
(5)
where J, the flux of each ionic species is given by the Nemst-Planck equation (6):
(7)
Equations 3 and 7 are solved using a finite-differences procedure. At every time step, Eq.
7 is solved first with a centered explicit scheme. The resulting electrical field is then
introduced into Eq. 3, which is solved with a second order procedure: Crank-Nicholson
scheme for the diffusion term, and centered Lax-Wendroff scheme for the electrical field
7
In addition to the material properties (density, porosity), the boundary and initial
conditions, the diffusion coefficient of each ionic species are necessary input data,
together with the binding isotherm. The diffusion coefficient of chloride is measured
from a migration test. It is assumed additionally that the ratio between the diffusion
coefficient of a species and the chloride diffusivity is the one found in an infinitely
diluted solution 8 :
D J _ Di J
~I
(8)
D material - D Cl diluted solution
The concrete properties are often measured 28 days after casting the material. Yet,
because the material is not still mature 9 , its resistance to the diffusion of ionic species
continues to evolve with time. It has been shown previously 9• 10 that the diffusion
coefficient of chloride is higher 28 days after casting than when the material is mature.
Figure 4 shows our experimental data together with others, also measured on CEM-I
based concrete 10 • The chloride diffusivities were measured 28 days and 330 days after
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
(J
tconst .
D( t) = D cons! - t - , If t< tconst (9-a)
( J
D(t)=Dconst, ift :::0: tconst (9-b)
Equations 9 a-b state that after a certain age, the diffusion coefficient becomes constant.
The age at which the effective diffusivity is no longer time-dependent varies with the
type of binder 9 . It is considered from Fig. 4 that the effective diffusion coefficient of
chloride reaches a constant value at the age, tconst = 70 days 10 • The exponent parameter cr
is equal to 0.5 10 , which is in good agreement with our own experiments also based on a
CEM-I type binder. Hence, the chloride diffusivity of the material can be calculated from
the chloride diffusion coefficient measured at 28 days after casting. Because the diffusion
coefficients of the other ionic species are linked to the chloride diffusivity, they are also
time-dependant.
The way the isotherm is obtained was described previously in Khitab et a!. 11 • It is based
on the view that during an immersion test a local equilibrium is reached between the
chloride in the pore solution, the bound chloride, and the solid phase, at every distance
from the exposed surface, regardless of the immersion time. The experimental data
published by Mohammed and Hamada 12 show indeed no time effect on the chloride
binding on concrete samples exposed to a marine environment for 10 to 30 years.
Once the total and water soluble chlorides at various depths from the exposed surface
have been determined, the amount of bound chloride is calculated as the difference
between the total chlorides and the water-soluble chlorides at each depth. In order to
obtain the binding isotherm, the amount of bound chlorides is plotted on the ordinates
against the water soluble concentrations. For this, only one sample of material is required.
The main objective is to obtain the total and water-soluble profiles, the latter being
considered to be similar to the free chloride profile. This objective has two consequences:
the direct consequence is to generate experimental data with which numerical results can
be compared with, while the indirect consequence is to determine the binding isotherm.
In this sense, the proposed method obtaining the binding isotherm belongs to the category
of inverse methods. From the experimental results on total chloride profiles and free
(water soluble) chloride profiles measured after the immersion tests, the resulted bound
chloride content was plotted on Fig. 5. The following isotherm equation is proposed (see
Khitab eta!. 11 ):
(10)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The data necessary for the comparison to the experimental results are summed up in
Table 2. They represent the package of input data described in the introductive part of this
work.
The experimental results obtained after 100, and 200 days of immersion are reported in
Figs. 6 a-b. The black squares are for the water soluble chlorides, while the empty
squares are the data for the total chloride profiles. All the experimental data are reported
in these figures. In this example, the concrete was placed in contact with the NaCl
solution after a 28-day cure. Recall that in the first part of this work, it was shown that the
age of the material (at least for the two cases tested) has no effect on the chloride
penetration. The simulations were made with a time-dependence following Eqs. 9 a-b.
The shape of the numerical concentrations profiles follows with a good accuracy the
experimental data. The penetration depths, which increase with the time of exposure, are
also in good agreement with the experimental results. It is important to stress that not a
single input data is an adjusted value. The input data are the result of experiments. They
are the five parameters needed in the "package" described in the introduction to this
article.
CONCLUSIONS
This work documented the effect of time on chloride penetration through concrete. The
influence of the time of immersion to a 165 g/1 NaCl solution was measured through
chloride profiles. The concrete was made of a CEM-I type binder. The effect of the age of
the material at the moment of the exposure to the salt solution was also investigated. It
appeared that the age of the material at the immersion (28 days after casting, or 28 days
plus approximately one year) does not impact the chloride transport, neither the chloride
penetration through the pore solution, nor the bound chloride.
In the work presented, a model dedicated to the ionic transport through cementitious
materials. The description of the chloride transport corresponds to the transport of ionic
species, and accounts for the ionic interactions (membrane potential), unlike in Pick's
laws-based models. It is proposed to predict the chloride transport through concrete with
a single package of experimental data: density, porosity, pore solution composition,
chloride diffusion coefficient and chloride binding isotherm. The binding isotherm is
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
5. Wang, Y., Li, L.Y. and Page, C.L., "A two-dimensional model of electrochemical
chloride removal from concrete"; Computational Material Science, 2001, 20,
pp.196-212.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
6. Marchand, J., "Modeling the behaviour of unsaturated cement systems exposed
to aggressive chemical environments"; Materials and Structures, 2001, 34, pp.
195-200.
10. True, 0., "Prediction of chloride penetration into saturated concrete - Multi
species approach"; PhD Thesis, Chalmers University, Goteborg (Sweden), and
INSA, Toulouse (France), 2000.
11. Khitab, A., Lorente, S. and Ollivier, J.P., "Predictive model for chloride
penetration through concrete"; Magazine of Concrete Research, in press.
12. Mohammed, T.U. and Hamada, H., "Relation between free chloride and total
chloride contents in concrete"; Cement and Concrete Research, 2003, 33, pp.
1487-1490.
CaO 6Q.63%
MgO 4.52%
SiO, 19.90%
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Al203 4.04%
Fe20, 2.81%
Na20 0.28%
K20 1.00%
so, 3.81%
Porosity 0.16
2
28 days
I 330 days
18x10'13
Dc1 (m /s)
I
K+
8 xlo-"
(mollm3)
Na•
T I I
011
Binding isotherm
23
T910-'c I
156 0
I
179
Ca + 1.0610-3c 0·526
Ca. in molikg of concrete, c in I +310-'c
mol/m3 ofso1ution
Concrete
II
Solution
/\ 1\
Fig. 1. Sketch of an immersion test.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3000
~
-c: 2500 •420days o28 days
0 0
~ 2000 •
'
:!!
0
.s...
1500
1000
•
£]
500
•
D •
D•
10 15 20 25 30 35
x(mm)
(a)
3500
-c: 3000 D
0
"'g 2500
~
• i•420da~ I
0 IJ28da~
"e
'a
.s
2000 •
u Iii
1500
[!
1000 D
D
500
• • 0
• D• ,..,.
-..... ~
~
10 15 20 25 30 35
x(mm)
(b)
Fig. 2. Free chloride profile after too (a) and 200 (b) days of immersion for a
concrete age of 28 days and 420 days at the moment of the immersion.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
oD
•420 days I
1
028 days I
(a)
"E
Q)
5 D
E
~ 4 D
q.
eif• •
0
;;e. D
.;:: 3
01
·a;
~
~ 2
iil(jD •
:9~
·~..::1~. 1•420 days I
~i··o
ul
1 028days
0
0 0,5 1 1,5 2 2,5 3 3,5
c free, weight % of cement
(b)
Fig. 3. Total chloride versus free chloride after 100 (a) and 200 (b) days of immersion
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
for a concrete age of 28 days and 420 days at the moment of the immersion.
120
D W/C = 0.32, ref [10]
- - D(t), W /C = 0.32, ref [1 0]
• W/C = 0.55, ref [10]
.., 100
- D(t), W/C = 0.55, ref [10]
~
,•
~~ 80
A W/C = 0.4, present work
;: --- D(t), present work
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
.! 60
.§., 40
:E
i5
Fig. 4· Variation of chloride effective diffusion coefficient with the age of concrete.
0,14
::.:
0,12
I... 00
u l5
"1:1 ... 0,1
§0 0 0
.8~
0 0,08
s
0,06
0
0,04
0 bound Cl, 100 days of immersion
0 lK bound Cl, 200 days of immersion
0,02 - - binding isothenn
0
0 500 1000 1500 2000 2500 3000 3500
3
c (mol/m of solution)
(a)
5000
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
4500
- free chloride, rmdel
4000 • total chloride, rmdel
c total chloride, exp
- 3500
• water soluble chloride, exp
6
~ 3000
~
'0 2500
..,
§ 2000
0
.§.
u 1500
1000
500
0
0 0,5 1,5 2 2,5 3 3,5 4
x(cm)
(b)
Fig. 6. Comparison between the experimental and the numerical results, profiles
after 100 days (a), and 200 days (b) of immersion.
Synopsis: This paper reports results of a study conducted to assess the effect of
chloride concentration on initiation and propagation of reinforcement corrosion.
Since it is expected that the tolerable chloride concentration will vary with the type
of cement, the combined effect of cement type and the chloride concentration on
reinforcement corrosion was evaluated. Concrete specimens were prepared with
Type I, Type V, and silica fume blended cements and they were exposed to sodium
chloride solutions with varying chloride concentration. Reinforcement corrosion was
monitored by measuring corrosion potentials and corrosion current density. After
two years of exposure, the reinforcing steel bars were removed from the concrete
specimens and they were examined for the extent of corrosion and the gravimetric
weight loss was determined. The electrochemical and gravimetric weight loss
measurements indicated a good correlation between the chloride concentration
in the exposure solution and the corrosion activity. The time to initiation of
reinforcement corrosion and its rate were influenced by the type of cement and the
chloride concentration in the exposure solution. Least reinforcement corrosion was
noted in the silica fume blended cement concrete specimens followed by Type I and
Type V cement concrete specimens.
INTRODUCTION
Portland cement concrete provides both chemical and physical protection to the
reinforcing steel. The physical protection is provided by the dense and relatively
impermeable structure of concrete that impedes the diffusion of moisture, oxygen,
chlorides, and carbon dioxide to the steel surface. The chemical protection is provided by
the highly alkaline nature of the pore solution (pH > 13). At this high pH, steel is
passivated in the presence of oxygen presumably due to the formation of a sub-
34
microscopically thin y-Fe 20 3 film - • Rime and Erlin 5 have suggested that the passivating
layer on the steel surface could have a composition other than that of y- Fe 2 0 3 •
6
According to Page , the lime-rich layer that is formed at the steel-concrete interface,
provides further protection to the steel. This was confirmed by Leek and Poole 7 who
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
tolerance in their specimens, compared with the BS 8110 limit of 0.4%, to the protective
nature of concrete produced under the well-controlled conditions of the laboratory 14 •
Research findings, however, have shown that the alkalinity of cement also significantly
influences its chloride binding capacity 15- 17 • Taking into account the combined effect of
chloride and alkalinity, Hausmann 18 suggested the critical C(/OH- to be around 0.6.
Gouda 19 , using pH values of the electrolyte representative of the concrete pore solution,
indicated that the threshold Cl-/OH- was 0.3. Lambert et al. 14 investigated the
relationship between CC!OH- and corrosion current density in various cements. Their
investigation indicated that the passive conditions of steel in concrete, characterized by a
corrosion current density Ocorr) substantially less than I 00 nA/cm2 were maintained till
a threshold Cf /OH- ratio of approximately 3 was exceeded. There was a considerable
scatter in the values of Icorr recorded at C(/OH- ratios in excess of 3 and even at
CC!OH- ratios as high as 15 to 20, there were instances of bars suffering no significant
Depassivation of steel occurs by the reduction of the pH of the pore solution, due to
carbonation, or by the ingress of chloride ions to the steel surface. A number of
mechanisms by which chloride ions break down the passive layer have been proposed,
e.g., the chemical dissolution of the film 24 , the build up of the metal holes at the
film/substrate interface 25 , and high chloride concentrations at the iron oxide/pore solution
interface, that leads to local acidification and pitting 26 • Leek and Poole7, based on SEM-
EDS studies of the passive film breakdown on steel in mortar prisms, have shown that
chloride ions initiate corrosion by breaking the bond between the film and the metal.
While the aforesaid studies, along with international codes, provide guidelines on the
allowable chloride in concrete, mostly contributed by the mix constituents, information is
lacking on the effect of chloride concentration in the service environment, such as
marine, soil and groundwater, on reinforcement corrosion. Such information will be
beneficial in adopting preventive measures, such as the use of appropriate cements or
other protective methodologies. The reported study was, therefore, conducted to assess
the effect of chloride concentration in the exposure solution on reinforcement corrosion
in plain and silica fume blended cement concretes.
METHODOLOGY OF RESEARCH
The concrete specimens were prepared with three types of cements, namely Type I, Type
V and Type I plus silica fume. In the silica fume blended cement concrete, 7% by weight
of cement was replaced with silica fume. All the concrete mixtures were prepared with a
cementitious material content of 370 kg/m 3 and an effective water-to-cementitious
materials ratio of 0.40. Table 1 shows the chemical composition of cements and silica
fume used in the preparation of the concrete specimens.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Cylindrical concrete specimens, 75 mm in diameter and 150 mm high, with a single 12-
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
mm diameter steel bar that was placed in the center were prepared. To avoid crevice
corrosion, the steel bars were coated with cement paste followed by an epoxy coating at
the bottom and at the concrete-air interface.
Exposure and Measurements
Three reinforced concrete specimens, with similar cement, were exposed to solutions
with chloride concentration of 500, 1,000, 2,000, 5,000, 10,000, 20,000 and 30,000 ppm.
In order to accelerate corrosion of the reinforcing steel only half the depth of the concrete
specimen was exposed to the chloride solution. The chloride solution in the exposure
tanks was changed periodically.
After 18 months of exposure, the concrete specimens were crushed to retrieve the
reinforcing steel bars. The steel bars were then examined for the degree of corrosion.
The degree of corrosion was identified on a qualitative scale varying from 0 to 5. The
corrosion rating and its description are detailed below:
Corrosion Potentials
The average corrosion potentials on steel in the concrete specimens exposed to solutions
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
with varying chloride concentration are plotted against the time of exposure to the
chloride solution in Figures 1 through 7. These curves were utilized to assess the time to
initiation of reinforcement corrosion according to ASTM C 876 criterion. This
information is provided in Table 2. Corrosion initiation was not noted in Type V cement
concrete specimens exposed to solutions with a chloride concentration of 2,000 ppm or
less. In Type I cement concrete specimens, the corrosion potentials were numerically
more than -270 mV SCE in the specimens exposed to solutions with a chloride
concentration of 5,000 ppm or less. Passivity was maintained in the silica fume blended
cement concrete specimens exposed to solutions with chloride concentration of 10,000
ppm or less. In cases where the corrosion activation was noted, the time to initiation of
corrosion decreased with increasing chloride concentration and generally corrosion
initiated earlier in the Type V cement concrete specimens than in the Type I and silica
fume blended cement concrete specimens. The time to initiation of reinforcement
corrosion in the silica fume blended cement concrete specimens was more than that in the
Type I cement concrete specimens.
The variation of corrosion current density (Icorr) with time in the concrete specimens
exposed to solutions with varying chloride concentration is plotted in Figures 8 through
14. The variation of Icorr after the completion of the exposure is plotted against the
chloride concentration in Figure 15. The Icorr increased with the period of exposure in
all the cement concrete specimens. Further, the Icorr values in Type I cement concrete
specimens were less than those in Type V cement concrete specimens, though the
difference between the two was not that significant. The Icorr on steel in the silica fume
blended cement concrete specimens was less than that on steel in Type V and Type I
cement concrete specimens.
The corrosion ratings, based on visual examination, for plain and silica fume blended
cement concrete specimens are summarized in Table 3. Corrosion was not noted on steel
in Type V cement concrete specimens exposed to solution with a chloride concentration
of 500 ppm while minor corrosion was noted on steel bars in this batch of specimens
exposed to solution with a chloride concentration of 1,000 ppm. In the concrete
specimens exposed to solution with chloride concentration of 2,000 and 5,000 ppm, the
intensity of corrosion was minor to moderate. In the concrete specimens exposed to
solutions with a chloride concentration of 10,000 ppm and above the intensity of
corrosion was moderate.
Reinforcement corrosion was not noted in the silica fume blended cement concrete
specimens exposed to solution with chloride concentration of up to 5,000 ppm while
minor corrosion was noted on the steel bars in the concrete specimens exposed to
solutions with a chloride concentration of 10,000 ppm or more.
The gravimetric weight loss in the steel bars, due to corrosion, is plotted against the
chloride concentration in Figure 16. The weight loss generally increased with the
chloride concentration. Minimum weight loss was noted in the silica fume blended
cement concrete specimens for all the chloride concentrations, followed by Type I and
Type V cement concrete specimens. While the weight loss increased with the chloride
concentration in the Type V cement concrete specimens, it was significant only for a
chloride concentration of 30,000 ppm in Type I cement concrete specimens. In the silica
fume blended cement concrete specimens, the weight loss was insignificant even for a
high chloride concentration of 30,000 ppm.
DISCUSSION
The corrosion current density on steel in Type V cement concrete specimens was more
than that in Type I and silica fume blended cement concrete specimens. However, it
should be noted that the corrosion current density on steel in the concrete specimens
exposed to solutions with varying chloride concentration was less than 0.3 J.1Aicm 2 , the
threshold value for corrosion initiation. This may be ascribed to the fact that concrete
specimens exposed to solutions become saturated with water and as a result the diffusion
of oxygen to the steel surface is reduced.
The data on visual ratings (Table 3) and gravimetric weight loss (Figure 16) also indicate
the superior performance of silica fume blended cement concrete over plain cement
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The data developed in this study, indicate that reinforcement corrosion is affected by the
chloride concentration in the exposure solution and the type of cement. The time to
initiation of reinforcement corrosion generally decreased with increasing chloride
concentration. Similarly, an increase in the rate of reinforcement corrosion was indicated
with increasing chloride concentration. The data also indicated that the type of cement
has a significant effect on both the time to initiation and corrosion current density. The
time to initiation of reinforcement corrosion in the silica fume blended cement concrete
specimens was more than that in Type I and Type V cement concrete specimens. Among
plain cements, Type I cement performed better than Type V cement. The rate of
reinforcement corrosion, evaluated in terms of both corrosion current density and
gravimetric weight loss, particularly the latter, was low in the silica fume blended cement
concrete for all the chloride concentrations (Figure 16), while a significant increase in the
corrosion rate was noticed in the Type I cement concrete specimens exposed to solution
with a chloride concentration of 30,000 ppm. The corrosion rate in the Type V cement
concrete increased almost linearly with the chloride concentration.
The superior performance of silica fume blended cement concrete compared to plain
cement concretes may be attributed to its dense microstructure that reduces the diffusion
of the chloride ions to the steel surface. Even after the chloride ions have diffused to the
steel surface and the steel passivity is destroyed, the rate of corrosion in the silica fume
blended cement concrete is low due to its high electrical resistivity compared to plain
cement concretes. The increased corrosion-resistance of Type I cement concrete over
Type V cement concrete may be attributed to the increased chloride binding capacity of
the former cement compared to the latter cement.
CONCLUSIONS
Corrosion initiation was noted in the Type V cement concrete specimens exposed to
solutions with a chloride concentration of 5,000 ppm or more while such a behavior was
noted in the Type I cement concrete specimens exposed to solutions with a chloride
concentration of 10,000 ppm or more. Corrosion initiation was noted in the silica fume
blended cement concrete specimens exposed to solutions with chloride concentration of
20,000 ppm or more. The time to initiation of reinforcement corrosion decreased with
increasing chloride concentration in the solution. Corrosion initiation was earlier in the
Type V cement concrete specimens followed by that in Type I and silica fume blended
cement concrete specimens.
The corrosion current density on steel in all the concrete specimens increased with the
period of exposure and the chloride concentration. The corrosion current density on steel
in the silica fume blended cement concrete specimens was less than that in Type V and
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The gravimetric weight loss data indicated that the intensity of corrosion generally
increased with the chloride concentration. However, the weight loss in the silica fume
blended cement concrete specimens was much less than that in plain cement concrete
specimens. Among plain cements, the weight loss in Type I cement concrete was less
than that in Type V cement concrete. The weight loss in Type V cement concrete
increased almost linearly with the chloride concentration in the exposure solution, thus
indicating the vulnerability of this cement concrete to reinforcement corrosion.
RECOMMENDATIONS
Since the results of this study have indicated that the type of cement significantly affects
the corrosion-resistance of concrete, the following chloride limits are suggested as
guideline for selection of cement appropriate for the exposure conditions.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Chloride concentration Suggested cement
Less than 5,000 ppm Type I or Type V
5,000 ppm to 10,000 ppm Type I
More than 10,000 ppm Type I plus silica fume
Since the corrosion-resistance of silica fume blended cement concrete was better than that
of Type V and Type I cement concretes it is recommended that Type I cement blended
with 7% silica fume should be utilized for substructures that are to be exposed to
chloride-sulfate environments. In addition to providing corrosion-resistance, silica fume
blended cement concrete would provide enhanced sulfate-resistance compared to Type V
cement concrete in such environments. However, good quality silica fume complying
with ASTM C 1240 should be utilized in addition to making sure that silica fume blended
cement concrete is cured for at least 14 days. Further, if sulfate ions are associated with
magnesium cations, the structural components need to be protected by the application of
a high quality protective coating. Epoxy and polyurethane coatings were found to be
useful for this purpose.
ACKNOWLEDGMENTS
Authors acknowledge the support provided by Saudi Arabian Oil Company and King
Fahd University of Petroleum and Minerals, Dhahran, Saudi Arabia.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Table 2: Time to initiation of reinforcement corrosion in plain and silica fume blended
cement concrete specimens.
Time to initiation of reinforcement corrosion
Chloride concentration, ppm according to ASTM C 876 criterion, days
Tvne V cement Type I cement Silica fume
500 Passive Passive Passive
1,000 Passive Passive Passive
2,000 Passive Passive Passive
5 000 335 Passive Passive
10 000 253 312 Passive
20,000 115 234 329
30 000 ND* 66 255
*ND: not determined
Table 3: Intensity of corrosion on steel bars in plain and silica fume blended cement
concrete soecimens.
Chloride Intensity of corrosion Corrosion rating
concentration, Silica Type Silica
Type I TypeV Type I
ppm fume v fume
500 None None None 0 0 0
1.000 None Minor None 0 1 0
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-200 +-----,----,------,-------,------,------1
0 100 200 300 400 500 600
Period of exposure, Days
~ 40+------*~~~-~~~~--~-~
~ ~0+----~~~-~-----~~--~
~ ~O+--~~L--~----------------1
i.. 1------::--;:z<------------;:=====;1
-a
-100
.~ -120 i-7--+----...,---A------l
~
0
-140 n~f---tt----tl~l------1t---~~~~_j
u-t~+-~~~-~~~~-~~~~---~---1
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~ ~+-~~~~~~~~--~~~~~~~---4
~
:l-100 ...,-f--------------------------='-------+.-4
!g,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
·~ -150
~
a -2oo t--~:!..~::__________ _!__________--c~;:::=:;-:=:;1
~00+-----.------.-----.----~----------~
0 100 200 300 400 500 600
Period of exposure, Days
~ -100 r::=;=:;;;;;;;;:~~~r::~--.---=---'\--1
>
a~oot------~b+-~-------~
f
B~H+-------~r--r-------~
&.
""'
~ -400 +----------'...-----==-~--~
~
-500
~
"&.
a~oo+---~--'~~----~------~--1
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~
0.09
0.08
~:--Type V cement/
I
~
~e --e- Type I cement
-.-Silica fume
~ 0.07
:I.
~
..:: 0.06
~ 0.05
_.----
~
!= 0.04
~
·~ 0.03
~ 7
l:l 0.02
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.01
N= 0,06 t-----------:--======-·---j
~
:1. o.os-j---<..-===-"-----------------1
f
~ 0.04 +---------~-==-------~----!
i =
0.03 +-------J.,.-:...----=..,..=-------------l
.2
~ 0.02 +---y"--____.~--------------1
e
0.09
__..__TYPe V cement
0.08
N -Type I cement
e -+-Sillca fume
~ 0.07
,e;; 0.06
·;;
""
..,. 0.05
=~ 0.04
"
.2 0.03
e~ o.o2
0.01
Figure 10: Corrosion current density on steel in the concrete specimens exposed to
2,ooo ppm chloride solution.
Figure 11: Corrosion current density on steel in the concrete specimens exposed to
s,ooo ppm chloride solution.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Nil 0,1
{
f·08
~ 0.06-j----------==~~-F'--------j
1l
" -t-----/-7"--::.,.L-'----------------l
·~ 0.04
E
8 0.02 t - - J 7 C - - - - - - - - - - - - - - - - - J
Figure 12: Corrosion current density on steel in the concrete specimens exposed to
1o,ooo ppm chloride solution.
Na o.u
~
"- 0.1 +---------------1'-+----1
{
-l:l 0.08 +-------------++-----,..----1
8
0.01 +-----:~-:::::::::;;:~~====~------~
Figure 13: Corrosion current density on steel in the concrete specimens exposed to
2o,ooo ppm chloride solution.
0.14
~--+-Type I cement I
Na o.12 ---silica fume
~ I
"- O.t
~-l:l 0.08 ;,
11l 0.06
II
=
·!
~ 0.04
)/
II
u=
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.02
,....-- :/
~
100 200 300 400 500 600
Period of exposure, Days
Figure 14: Corrosion current density on steel in the concrete specimens exposed to
30,ooo ppm chloride solution.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
5000 10000 15000 20001 25000 30000 35000
Chloride concentration in soil, Jllllll
Figure 15: Variation of corrosion current density with the chloride concentration in the
exposure solution.
0.8
;1.
:; 0.6
lO
•
.!f
.ld
1l
~
0.4
"
0.2
Figure 16: Gravimetric weight loss of steel in the concrete specimens exposed to
solutions with varying chloride concentration.
Synopsis: The alkali-silica reaction (ASR) was recognized by Stanton in 1940 and
is still today a world-wide problem. The basic reaction mechanisms are reasonably
well understood and in many cases a prevention of the ASR is possible. However,
there are still severe damages on concrete structures caused by alkali-silica reaction,
especially related to slow/late aggregates. This paper presents the experiences with
the German Alkali-Guideline during the past decades in Germany and introduces
latest approaches in developing forward-looking test methods to assess the
durability of concrete regarding ASR due to alkali-reactive slow/late aggregates.
The emphasis is particularly placed on assessing the influence of an external alkali
supply, i.e. the impact of de-icing solutions, on pavement concretes. For this purpose
the cyclic climate storage was used and proved as appropriate method.
Science. Main field of research work: Durability of concrete, especially related to ASR.
INTRODUCTION
Damages on concrete structures caused by alkali-aggregate reaction (AAR) occur all over
the world. In Germany, the first damage due to alkali-silica reaction (ASR) occurred on a
bridge, built in 1964 in Northern Germany. Already in 1968 the bridge had to be
demolished due to the loss of structural stability. In the following years, the ASR in
Germany was essentially understood as ASR of alkali-reactive aggregates containing
opaline sandstone and flint from Northern Germany. More and more damages appeared,
especially on bridges, roads and railway ties (Fig. 1). After intensive research work, the
very first Alkali-Guideline was published in 1974 by the German Committee for
Reinforced Concrete (DAfStb). Now it was possible to classify alkali-reactive aggregates,
which contain opaline sandstone and flint, in 3 different alkali-sensitive classes.
Depending on environmental conditions (moisture, external alkali supply) and cement
content, clear measures were proposed to prevent a deleterious ASR (1,2). One of the
most important countermeasures proposed in the Alkali-Guideline is the use of low-alkali
cements, if the aggregates contain alkali-reactive rocks and environmental conditions
(moisture, external alkali supply) that are beneficial to an ASR are expected (Table 1).
The background was to hold the total amount of alkalis in the concrete below 3.0 kg/m3
and the concentration ofOH- in the pore solution below 500 mmol/1 (pH= 13.7). Since
most alkalis in blast-furnace slag are not soluble and hence not beneficial to an ASR, a
higher Na20eq for slag containing cements is admissible. If an external alkali supply is
possible in combination with moisture and highly alkali-reactive aggregate, alternate
aggregate must be found.
However, the Alkali-Guideline had to be revised several times (1986, 1997 and 2001),
because the issue was finally more complicated than supposed. In Eastern Germany
ASR-re1ated damages occurred on concretes with precambrian greywacke from South-
East Germany but also with siliceous schist and porphyry. The greywacke showed a
different behavior than known from opaline sandstone or flint. Damages occurred at very
late ages and not every greywacke seems to be alkali-reactive. Thus, in 1997 the Alkali-
Guideline was supplemented by a special test for greywacke and other ASR-suspicious
aggregates, where 3 concrete beams (10x10x50 em) and 1 concrete cube (30x30x30 em)
are stored in a fog chamber for 9 months. Contrary, the alkali-reactive aggregates from
Northern Germany, containing opaline sandstone and flint, were only classified by
petrographic methods (3). Furthermore, owing to the more difficult issue, the fog
chamber test allows merely a "yes or no" -decision. That is, either the aggregate is
classified as innocuous or not. Additionally, the testing period is much too long for most
clients. But that was not the end of the story. Recently, some crushed aggregates from the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 227
upper reaches of the river Rhine (Southern Germany) caused severe damages, when used
in concrete roads (external alkali supply). Until that it was not known that in this region,
far away from the application area of the Alkali-Guideline, deposits with alkali-reactive
aggregates existed (4). It is often reported, that fly ash (FA) is effective in suppressing a
deleterious ASR (5). But there are no sufficient experiences in using fly ash to control
ASR in Germany. Consequently, the use of fly ash is not allowed for pavement concretes
and is not recommended in the Alkali-Guideline as well. There are inconsistent research
results which showed on the one hand that fly ash was able to prevent an ASR in concrete
specimens that were stored in the fog chamber (40 °C, 100 % RH), but fly ash does not
prevent an ASR in concrete specimens stored in the field at ambient temperatures and
moisture conditions as typical in Middle Europe (6,7). It is supposed, that the pozzolanic
reaction of fly ash is accelerated in the fog chamber (40 °C, 100 % RH), so that a denser
matrix is formed and portlandite is consumed (8). However, under the climatic conditions
in the field the pozzolanic reaction proceeds significant slower and beneficial effects may
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SLOW/LATE AGGREGATES
Present damages on concrete structures in Germany due to ASR are caused by slow/late
aggregates, which are so far not sufficiently treated by the Alkali-Guideline. During the 9
months of the fog chamber test slow/late aggregates do not cause necessarily critical
expansion (> 0.6 mm/m) on the concrete beams or large cracks (> 0.2 mm) on the
concrete cube. However, deleterious expansion and severe cracking may occur much
later because of the slower reaction mechanism (Fig. 2). Therefore, these kinds of
aggregates are designated as slow/late and many aggregates can be assigned to this group,
e.g. gravels with amounts of siliceous limestone, siliceous schist, greywacke, quartzite
and crushed aggregates made from rhyolite, greywacke, granite, crushed gravel and
recycled aggregates (1). Already in 1975 Mather (9) expressed it this way: "All
aggregates are alkali-reactive, they differ only in the kind of reaction and the degree and
rate". There is presently no reliable testing method to assess in a reasonable time the
damaging degree of slow/late aggregates.
Greywacke aggregates consist mainly of quartz, feldspar, mica, chlorite and partly of
clays. Quartz is mainly present as cryptocrystalline and stressed quartz, sometimes also as
chalcedony. By means of a light-optical microscope with crossed polarizers, the different
quartz polymorphs can be distinguished and a first statement regarding the alkali-
reactivity is possible by assessing grain size, texture and particularly the amount of fine
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
228 Stark and Giebson
quartz in the matrix (Figs. 4,5). Recent solubility experiments from Hunger and Hill (1 0)
with different greywackes revealed that the amount of dissolvable silica is not the only
crucial point but also is the amount of dissolvable aluminate. The dissolved aluminate as
species Al(OH) 4- are able to form alumosilicate complexes with silica. Hence, the
consumed silica is no longer available for alkali-silica reaction (1 0,11 ). This mechanism
might be an important reason, why some greywacke are highly alkali-reactive and some
are not. Another example is rhyolite which consists generally of a quartz matrix, where
greater quartz grains are embedded. Also inclusions of feldspars and intergranular pores
are possible, whereas the latter are able to support transport processes. Quartz can be
present as microcrystalline and cryptocrystalline polymorphs in the matrix. Since the
mineral composition of porphyry can fluctuate largely, an assessment of the alkali-
reactivity is considerably difficult. Also some quartzites, phyllites, andesites, mylonitic
granites as well as clay schists are considered to be alkali-reactive in some way. A recent
study by Freyburg and Berninger (12,13) showed enhanced possibilities to estimate the
alkali-reactivity of aggregates by using X-ray diffraction in combination with a
petrographic examination. The parameters crystallite size and root mean square-strain
(RMS-strain) are used to support the assessment of the alkali-reactivity for rocks like
siliceous limestone, siliceous schist or granite.
homogeneous supply of alkalis is available for the aggregate. After demoulding the
mortar bars, they will be stored at 70 °C above water. Test results become available after
28 days. If the aggregate shows critical expansion, that is when 1.5-2.0 mm/m are
exceeded, it will not be rejected necessarily, but the aggregate will be tested directly
within the intended concrete. Therefore, concrete beams (10x10x40 em) are stored in a
new climate simulation chamber.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the concrete beams (lOx 10x40 em). The resonance frequency is used to calculate the
dynamic modulus of elasticity. A deterioration of the concrete beams can be assumed,
when expansion exceeds 0.4 mm/m, which was defined after performing many
preliminary investigations and considering the expansion by moisture. Since all
measurements are done at 20 °C, thermal expansion is canceled out. If the expansion
exceeds 0.4 mm/m microscopic examinations (ESEMIEDX, light-optical microscopy on
thin sections) will be performed to identify the cause for the expansion and the damage
respectively. At least 4-6 cycles (12-18 weeks) are necessary to assess the durability of
the concrete regarding ASR.
Particularly pavement concretes are damaged by microcracks due to road traffic which
supports the penetration of water and de-icing solution. Therefore, an additional test
option is to simulate such damages by creating microcracks and flaws due to mechanical
stresses (pre-damage). For this purpose, the concrete beams will be loaded under a
compressive strength testing device until rupturing (16). The intention is to increases the
permeability of the concrete already before the cyclic climate storage starts in order to
accelerate chemical transport processes and to initiate latent damaging potentials faster. A
latest attempt in assessing the concrete regarding an external alkali supply is currently
done by applying a defined de-icing solution on the upper surface of the beams prior to
the cyclic climate storage. For that purpose, a cellular rubber-tape surrounds the upper
beam edges as border, so that a de-icing solution can be applied.
REFERENCES
1. Stark, J., Wicht, B. et al., "Durability of concrete"; Birkhauser Verlag,
Basel, 2001, pp. 289-330
4. Ottl, C., "The deleterious alkali-silica reaction of crushed aggregates from the
upper reaches of the river Rhine"; Dissertation, Otto-Graf-Institute Stuttgart,
Germany, 2004
5. Shehata, M. H., Thomas, M.D.A., "Use of ternary blends containing silica fume
and fly ash to suppress expansion due to alkali-silica reaction in concrete";
Cement and Concrete Research 32, pp. 341-349,2002
6. Heinz, D., Schmidt, K., "Fly ash in concrete to prevent an deleterious alkali-
silica reaction"; Centre for Materials and Materials Testing, University
Milnchen,Germany,2005
10. Hunger, K. H., Hill, S., Weidmiiller, H., ,The self-inhibition mechanism of
certain greywackes"; In Tang M. & Deng M. (eds), Alkali-Aggregate Reaction
in Concrete (1): pp. 125-134, Proceedings ofthe lth ICAAR, Beijing, 2004
14. Philipp, 0., Eifert, K., "Detennining the alkali reactivity of gravels and
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
chippings to be used in making concrete"; Betonwerk+Fertigteil Technik
10/2004, pp. 6-19 Bauverlag Giitersloh, Germany, 2004
15. Oberholster, R.E., Davies, G., "An accelerated method for testing the potential
alkali reactivity of siliceous aggregates"; Cement and Concrete Research,
Volume 16, Issue 2, pp. 181-189, 1986
16. Seyfarth, K., Stark, J., "Perfonnance testing method for durability of concrete
using climate simulation"; F.A. Finger-Institute for Building Materials Science,
Bauhaus-University Weimar, Germany, 2005
17. Stark J., Seyfarth K., Heinrich, U., ,Initial testing experiences of HPC under
climatic simulation conditions"; F.A. Finger-Institute for Building Materials
Science, Bauhaus-University Weimar, Germany, 2004
/
7 ( oncret with
tlow/lat aggre ate
/
e'E 2.0
s
a 1.5
I
./ Testin ~ Perio ~
(9 mor th ace. Alkali- ~uideli e) /
iii
c
3
ftl
1.0
I /
0.5
7 Limit .6mm/~ /
0.0
~~ ~
~
~
3 6 9 12 15 18 21 24 27 30
Duration of fog chamber storage (month)
Fig. 3- Greywacke grain with crack, optical-light microscope image of a thin section.
Fig. 4- Crossed polarizers with the optical-light microscope image of Fig. 3 shows
fine quartz.
u4o1ft-t-t-t-t-~~~~=t=t~=t=t=t=t~-t~~~
e... 30 -fM-+-+---+___._.--+-jJ Measuring expansion, mass and 1
~ lc..,---,---r-res--r-on_a,n_ce-rf_req--r-u-en,c..;..y-,---.--T"'l - ;-.r-
'§ 20
- ~
~10 ~~-+-+~--r-+-~-+-+_,--r-+-~-+-+~,-T-~flr~~
{! 0
-1o
• ······f·
+-~-+-+~--r-+-~-+-+_,--r-+--r-+-+--1---+-~~;~;H
!
·20 ~±:::±::::J;;::t:±:::;±;:±:::;±;:::::±::::::;i;::;:J::::::;t:::~:;l=:±;:±;:±~~~=:;-j
1 cycle= 21 days= 4 days drying+ 14 days fog+ 3 days freezing and thawing I
-30
0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Time (days)
Fig. 5- Cycle scheme for the climate simulation chamber (Stark and Seyfarth).
3.5
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
....._Graywacke 2-3 mm
3.0 --Graywacke 8·16 mm 1---------+--------------t---~~;,...-=r
---,!r- Gravel1, 2-8 mm
--e-Gravel1, 8-18 mm
-2.5
· • • Gravel 2, 2-3 mm
~ 2.0
.§.
· + · Gravel2, 8·16 mm
c
0
~ 1.5
ra
0.0
0 14 21 28
Age(days)
Fig. 6- Obtained test results from the new mortar bar test for 3 aggregates and
different grain size fractions.
-0.1
0 21 42 63 84 105 126 147 168 189 210 231 252 273
Age(days)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 7- Expansion for 2 concretes with greywacke, stored in the climate
simulation chamber.
0.5
CEM I 32.5 R (Na20eq = 1.1 ), 400 kg/m3 I -+- CEM I and gravel1 I
CEM 1118-S 32.5 R (Na 20 00 = 1.0). 400 kg!m• ..... CEM 11/B.S and graval1
OA W/C=0.45
30 % 0-2 mm (innocuous quartz sand)
~.§.
40 % 2-8 mm (gravel1, Middle Germany)
30%8-16 mm (gravel1, Middle Germany)
0.3
l ~I
c
0
'iii
~ 0.2 ........
....
re
~~ ......__ ~
lr:~
>C
w
1---
0.1
v ....-
0.0
j
21 42 63 84 105 126 147 168 189 210 231 252 273
Age (days)
.s~ 0.3
30%8-16 mm (grave12, Middle Germany)
c
0
"iii
i 0.2
! .. .... ......- ~
!i.. ..... ..... ....... ~
0.1
-........ ~ ~
If"....... r--- 1----"
0.0
J 21 42 63 84 105 126 147 168 189 210 231 252 273
Age [days]
700 , - - - - - - - , - - - - , - - - , - - - - - , - - - , - - - - , - - - - , - - , - - - - - - - , - 14.0
soo L-rl~-_J--k=~=i===t=~t:==t==l13.9
~e 500 pr---r--=:::+==t===t::;::==t===it=~+==:t==t
v r-- ........................... ·-·-·· 13.8
i§f400+~~~4--.~-~--r---+--4--~r---r--+---+ 13.7
:z:
:z:
~300"-~~~~~~·-··_·_·--·~·-··_·_·_··+·~-~~-----~·-·_··_·_·~-~·-·_··_•_·~·-·_·_··_·1---+ 13.6 "-
c .~-
~ ,• cEM 1(K+) l........ ......... cEM 11/B-s (K+)
0
a200 ·1-.~:·~-+--H-o- CEM I (Na+)
": CEM I (pH) I· • ·
-r.- CEM 11/B-S (Na+)•f------1---+ 13.5
- .. - CEM 11/B-S (pH)
13.3
4 7 10 13 16 19 22 25 28
Age (days)
Fig. to- Analyzed pore solution of the CEM I and CEM 11/B-5 until28 days.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.2i___JL~~~~~~~§f~~~=t:==t::=JI=:==t:==J~~_j
0.0 _/v-
.J-.,4;---1-~-.j.......,---1--~l-----1-~-----l---+--~..-+---1-~~
21 42 63 84 105 126 147 168 189 210
Age [days]
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACI member Michel Thibault is Engineer and technical advisor, Infrastructure Service,
Division of Bridges and Tunnels, City of Montreal, Quebec, Canada. His research interest
include DEF and durability of concrete. He is presently a Master's level student at the
Universite de Sherbrooke.
INTRODUCTION
EXPERIMENTAL PROGRAM
RESULTS
The temperature changes of concrete are presented in Figure 2 for the first 10
days. The maximum temperature at the center of a cubic sample of 72.2°C was developed
only through heat of hydration, 22 hours after casting for a sample made with a Type III
cement, designated as cement C, having a high fineness of 600m2/kg and the highest C3 S
content of all cements used in this experiment. The temperature of concrete at the center
of this sample remained above 70°C for a period of 38 hours, and above 60°C for a
period of 6 days.
After a period of 70 days of continuous storage in lime-saturated water, this
concrete started to expand significantly at an approximate rate of 30 J.lE per day, as shown
in Figure 3.
The maximum temperature of the second concrete that showed expansion was
67.9°C. The concrete temperature in the center of the sample was higher than 60°C for a
period of 4 days. This concrete was made with another Type III cement, designated as
cement A, having a fineness of 600m 2/kg. However, its C3S content was slightly lower
compared to that of cement C. A significant expansion of this concrete began after 200
days of continuous storage in lime-saturated water and at a rate of approximately 6 J.lE per
day, which is lower than the rate of expansion of specimen C. Moreover, specimen A's
ultimate expansion value of 3000 J.lE (0.3%) was also lower compared to specimen C's
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
DISCUSSION
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
It is believed by many researchers that DEF can cause expansion of concrete.
However, the mechanisms of ettringite formation and possible expansion of concrete
remain controversiae· 10 • Much research has been done to show the effect of various
factors on expansion due to DEF such as: cement chemistry and DEF in hydrated cement
paste 11 -21 , elevated temperature at early age 1• 10• 12 • 22 -24 , cement fineness 11 , pre-existing
microcracks, and increase ofporosity 15 • 16 , etc. However, to our knowledge, present codes
do not make any clear distinction between temperature effects caused by steam curing, or
other external sources, and heat of cement hydration. Under these circumstances,
concrete manufacturers do not take into consideration maximum temperature for cast-in-
place concrete. Several studies of deteriorated field concrete assume that the cause of
deterioration is DEF, but such assumptions are often contested because the expansion
caused by DEF in field concrete is easily masked by other causes of deterioration, such as
freezing and thawing, alkali-silica reactions, corrosion of steel, etc. Moreover, for field
concrete elements, the temperature history is often not available. Laboratory studies show
that the late sulfate release from hydrated Portland cement paste in the case of normally-
cured concretes does not occur. The deterioration of field concrete is not attributable to
DEF if laboratory studies do not show that the rise of temperature caused by heat of
cement hydration could have created the conditions for DEF.
In our study, we showed that the rise of temperature caused by heat of hydration
combined with other necessary factors 6 • 25 -26 can cause DEF and consequent expansion.
The factors for DEF are usually grouped in three categories: concrete composition,
thermal history, and service conditions 6• 27 . In the case of field concrete, it is often
possible that all circumstances are present except for elevated temperature at early age.
When the analysis of service conditions and materials show that there exists a real threat
for DEF, precautions must be taken to avoid the rise of temperature above 60°C.
In the studied cases, two concretes of identical composition, but made with
similar cements, showed expansion. The fineness of these two cements was high, at 600
2
m /kg. However, it is not very clear why in very similar circumstances, the concrete made
with cement TIII-B did not expand. The reason for this observation can be the slightly
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 243
lower temperature maximum of only 2°C less (66°C compared to 67.9°C) and slightly
lower cement fineness.
The SEM/EDXA analysis showed that the massive ettringite found in the gaps
surrounding aggregate particles has an excess of aluminate. In other words, it is possible
that the monosulfoaluminate was still present together with ettringite in the same zones,
and that its transition to ettringite did not occur. Such mechanisms of ettringite formation
from monosulfoaluminate have been proposed in the literature 10 •
CONCLUSIONS
The research conducted on self-heated concrete samples shows clearly that DEF
and consequent expansion are possible in field concrete when the maximum concrete
temperature rose above 65°C and when this temperature was maintained several hours. In
addition to elevated temperature during hardening, other factors are needed for DEF and
expansion, such as the use of cement sensitive to DEF and exposure to water or high
humidity.
REFERENCES
3. Divet, L., Guerrier, F., and Le Mestre, G., "Existe-t-il un risque d'attaque
sulfatique endogene dans les pieces de beton de grande masse?" (Is there a risk
of internal sulfate attack on massive concrete members?) (in French), Bulletin du
Laboratoire des Ponts et Chaussees 213, 1998, pp. 59-72.
5. Johansen, V. and Thaulow, N., "Heat Curing and Late Formation ofEttringite",
in "Ettringite the Sometimes Host of Destruction", ACI SP 177-4, 1999, pp. 47-
64.
6. Thaulow, N., Johansen, V., and Jakobsen, U.H., "What Causes Delayed
Ettringite Formation?", Ramboll Bulletin no.60, 1996, 8 pp.
11. Kelham, S., "The Effect of Cement Composition and Fineness on Expansion
Associated with Delayed Ettringite", Cement and Concrete Composites, Vol. 18,
June 1996,pp. 171-179.
13. Grattan-Bellew, P.E., Beaudoin, J.J., and Vallee, V.-G., "Delayed Ettringite
Formation: Effect of Clinker Particle Size and Composition on Expansion of
Mortar Bars", Materials Science of Concrete, Ed. The Sidney Diamond
Symposium, 1998, pp. 295-307.
15. Petrov, N. and Tagnit-Hamou, A., "Is Microcracking Really a Precursor to DEF
and Consequent Expansion?", ACI Materials Journal, Vol. 101, No.6, 2004, pp.
442-447.
16. Fu, Y., Xie, P., Gu, P., Beaudoin, J.J., "Significance of Pre-Existing Cracks on
Nucleation of Secondary Ettringite in Steam-Cured Cement Paste", Cement and
Concrete Research, Vol. 24, No.6, 1994, pp. 1015-1024.
17. Fu, Y., Beaudoin J.J., "Mechanisms ofDelayed Ettringite Formation in Portland
Cement Systems", ACI Materials Journal, Vol. 93, No.4, 1996a, pp. 327-333.
19. Maranville, M., Li, G., "The U-Phase - Formation and Stability", Material
Science of Concrete- Sulfate Attack Mechanisms, 1997, pp. 175-188.
20. Odler, 1., Gasser, M., "Mechanism of Sulfate Expansion in Hydrated Portland
Cement", Journal ofthe American Ceramic Society, Vol. 71, No. 11, 1988, pp.
1015-1020. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
22. Akman, M.S., Giilseren, H., "The Influence of Heat Treatment on the Durability
of Concrete", Proceedings Third CANMET-ACI International Conference,
Nice, France, 1994, pp. 1051-1066.
24. Scrivener, K.L., Lewis, M.C., "Effect of Heat Curing on Expansion of Mortars
and Composition of Calcium Silicate Hydrate Gel", in "Ettringite the Sometimes
Host of Destruction", American Concrete Institute, Farmington Hills, Mich., SP
177-1, 1999, pp. 93-104.
26. Mehta P.K., "Sulfate Attack on Concrete - Separating Myths from Reality",
Fifth International Conference on Durability of Concrete, Barcelona, Spain,
2000, pp. 1-12.
TIII
Cement TI
A B c
Si02 19.1 19.9 20.0 19.5
Ah03 5.0 5.3 5.1 4.7
F~03 3.4 2.3 2.4 2.0
Chemical
composition CaO 62.4 63.4 62.7 63.8
(%)
MgO 2.6 2.6 2.7 2.0
so3 3.8 4.4 3.8 3.6
Na20equ. 0.60 0.90 0.91 0.83
Blaine fmeness (m2/kg) 380 600 550 600
c3s 55 50 55 66
Phase c2s 14 20 16 6
composition
(%) C3A 8 10 10 9
C0-F 10 7 8 6
80
72.2 °C
70
u
!?_.,
60
~
. 50
~. 40 TIII-C
~
Till-A
S'
~
30
TIII-B
E-1
20
TI
10
0
0 2 4 6 8 10
Thne (days)
Fig. 2.- Recorded temperature at the centre of "self-heated" concrete specimens for
the first 10 days of hydration
5500
..-.
~ 4500
=
.Sl
."'
=
~
3500
2500
f;J;;l
1500
500 TIII-B
Time (days)
za:JD
2000
lEIID
1000
Ell
Fig. 4· - Energy dispersive X-ray spectrum of massive ettringite found in the gaps
surrounding aggregate particles (concrete A)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. sa.- Massive ettringite in the gaps surrounding aggregate particles of concrete A
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. sb.- Massive ettringite in the gaps surrounding aggregate particles, generally
dispersed in all interfacial zones of concrete A
Fig. sc. - Massive ettringite in interfacial zones and needle-like ettringite crystals in
large voids of concrete A
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
across the Maryland State Bridge districts. At each bridge several cores were drilled
for subsequent examination of fracture surfaces by Scanning Electron Microscope
(SEM) with energy dispersive X-ray fluorescence. At every bridge, ettringite was
detected, but alkali-silica reaction (ASR) gel was detected only very rarely. In more
heavily damaged locations, the occurrence of ettringite crystals was more frequent,
appearing in the rims around aggregates as well as in voids. Also, the morphology of
the ettringite crystals appeared to be more lamellar than acicular. The implications
are that: DEF is widespread geographically, map cracking is not diagnostic only
for ASR, and the onset of DEF may be associated with a change in ettringite crystal
morphology.
Since the early 1990's, there has been concern about the potential for damage to
concrete structures due to delayed ettringite formation (DEF). There have been several
thorough reviews of the literature on this subject 1-4 , which involve a number of ongoing
controversies. One issue is whether DEF only occurs in concrete that has been cured at
high temperatures, e.g. steam-curing. Another is the geographic extent of the incidence of
DEF since some have suggested that this problem is limited to only a few localities. A
third controversy concerns the role of ASR vs DEF in causing the damage. A fourth
issue is whether the primary ettringite that forms during early age hydration actually
transforms into monosulfate, as is often stated5•
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
was decided to undertake a more systematic approach to obtaining samples of concrete
for DEF investigation 7 • This has been done in collaboration with the Maryland State
Highway Administration, using the State's inventory of bridges
• To develop and execute a systematic sampling plan for a pilot field survey of
Maryland bridges for the presence of DEF.
• To evaluate two nondestructive test (NDT) methods that could provide indirect
evidence of DEF damage, and thus minimize the need for destructive drilling of
cores.
SURVEY METHODOLOGY
The DEF process involves the transformation of one crystalline or amorphous phase in
the concrete to the specific crystal structure of ettringite. Unfortunately, there is no
nondestructive method for detecting the presence of this mineral uniquely in concrete in
the field. Consequently, samples of the material have to be taken destructively and
prepared for analysis in the laboratory. One method for detecting mineral phases is X-ray
diffraction 8• This can be a quantitative method in some situations, but for concrete this is
usually not feasible because of sampling problems at the length scales of the fine and
coarse aggregates. The other main method involves the scanning electron microscope9
which is used to recognize the characteristic crystal morphology and elemental
composition of ettringite.
Since the direct identification of ettringite involves destructive core drilling, the number
of samples that can be obtained is limited by the costs of drilling and concerns about the
structural integrity of the bridge being sampled. Consequently, there is great interest in
nondestructive test methods that can at least indirectly indicate the presence of DEF
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 253
damage. One method involves measuring the level of potassium in the concrete, since
the probability of damage appears to increase with higher levels of potassium 10 • The
other method is a modification of the ultrasonic impact-echo technique to detect the
distributed micro-cracking associated with extensive damage. It is recognized that other
mechanisms such as alkali-silica reaction (ASR) or freeze-thaw cycles can also produce
micro-cracking so this method is not a unique measure of DEF-related damage. In the
absence of such as unique measurement, this method is being evaluated as a surrogate to
be used in conjunction with other methods.
The bridges selected for the survey sample were identified from the databases
compiled by the Maryland State Highway Administration (MDSHA) as part of the
Federally-mandated National Bridge Inventory System 11 . This inventory has a structure
of one record per bridge. However, Maryland, along with a number of other states, has
opted to add databases that include information down to individual bridge elements:
Pontis and Structural Inventory and Appraisal (SI&A). The SI&A database contains
supplemental information at the bridge level such as age, materials, type, length, height,
detour length, etc, and includes a general evaluation of the major components, such as the
deck, superstructure, substructure, channel, etc. Pontis concentrates on evaluating the
individual elements and provides numerical ratings of the condition level of each 12 •
Along with the numerical ratings, Pontis includes narratives by the bridge inspector that
give descriptions of the conditions encountered. These comments are intended mainly to
clarify the type of maintenance work that should be performed.
In the absence of any a priori information about the incidence rate of DEF in
Maryland bridges, a simplistic approach to obtaining an unbiased estimate of the rate
would be to select a sample of bridges at random from the entire population of concrete
bridges in the inventory. However, given the limited number of bridges in a practical-
sized sample (on the order of20) and the possibility of a low incidence rate ofDEF, there
is a significant probability of a false negative as shown in Fig. 113 . In other words, none
of the bridges in the sample may present any DEF symptoms, even though it does exist in
other bridges in the population.
Of the 2463 bridges owned by the MDSHA, 905 bridges have at least one element
with map cracks noted by bridge inspectors in the PONTIS database. However map
cracking alone was found to be too general as a description, since map cracking is often
associated with several possible causes including: overstress, improper stripping of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A representative number of bridges from this set were then selected from each
MDSHA district in proportion to the total number of bridges in that district. The
distribution is shown in the map in Fig. 3. This in effect created a sampling plan stratified
by geographic coverage. Fortuitously, the state of Maryland covers a variety of climates
from tidewater to the Appalachian Mountains. The geology, and hence the type of
aggregate, also varies from alluvial gravel near the Chesapeake Bay through limestone in
the central Maryland region to igneous and metamorphous silicate rock in the western
part of the state.
Finally, to provide a control set, two bridges were also selected that did not exhibit
map cracking.
Core analysis
Concrete cores 5.lcm (2 in.) in diameter and 7.6 em (3 in.) deep were taken with a
diamond drill from the bridges selected for the sample. For each bridge, several core
locations were specified based on an onsite inspection. Where possible, the locations
were chosen so that at least one core came from an apparently undamaged ("clean") area
of concrete and another from a damaged ("dirty") area. Examples of each type are given
in Figs. 4a and 4b.
The primary method of analysis was SEM imaging of fracture surfaces. Each core is
broken into small pieces with a 2.27 kg (5 lb) sledge hammer, and a sample was selected
which had a relatively flat side to allow mounting. After carbon coating to avoid charge
buildup while in the electron beam, the sample assembly is placed into the SEM chamber
with the top surface at an approximately 30 degree angle to produce the maximum
possible signal for the EDAX. The sample was searched in a raster scan pattern to
identify possible regions of DEF and/or ASR gel. Ettringite crystals were identified by
their acicular morphology and confirmed by elemental ratios determined by EDAX.
Alkali-silica gel was also identified by appearance and by the tendency to develop cracks
under prolonged exposure to the harsh environment of the SEM.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
fresh concrete surface. The surface was then washed with clean water and a solution of
dilute uranyl acetate is applied. After the sample had dried for at least 3 minutes, the
sample was placed under a short wave ultra-violet (UV) lamp (~250 nm). When viewed
under the UV light, the stained ASR gel appears yellow-green.
Potassium autoradiography
This NDT method takes advantage of the fact that potassium is naturally radioactive.
The radioisotope 4°K emits both beta particles and gamma rays with characteristic
energies. The approach used in this survey was to measure the radiation dose by storage
phosphor image plates, which are much more sensitive than conventional X-ray film 10 •
To separate the gamma-ray dose from the beta dose a pair of image plates was used with
a 1 mm lead plate in between. Since the gamma ray is much more penetrating than the
beta particle, its dose gives information about potassium in the bulk of the material, while
the beta dose arises from potassium concentrations within a few centimeters near the
surface of the concrete.
In practice, two pairs of image plates were exposed at each bridge site, one pair at a
"dirty"location, the other at a "clean" location. This thus leads to four exposure
conditions for each bridge: clean front, clean back, dirty front & dirty back. The plates
were left in position overnight, so exposure times were typically around 12 hours. After
exposure, the plates were taken to the laboratory for reading and then erased for re-use.
The output of plate is a digital image. Each pixel can be considered an independent
radiation detector. The image plate size is 20 x 25 em, so that at the minimum pixel size
of 25 Jlm, the number of detectors is about 80 million. The energy resolution depends
upon the number of greyscale levels. The maximum number is 65536 or 16 bits. The
actual data produced by the image plate is the amount of light in photostimulable units
(psi) emitted by the energy deposited by the radiation at each pixel. Calibration functions
are required to convert the psi to radiation dose units.
Impact-echo
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
echoes, and this can be fitted to an exponential decay function as shown in Fig. 5. The
amount of damping is quantified by the parameter a, the decay constant 17 •
RESULTS
The complete results of the field study are presented in the Maryland State Highway
Administration Report of the projed 8 . This section is intended to provide only an
overview of the main points. The data collection is summarized in Table 1.
Core analyses
A total of twenty-nine cores were taken and examined. After fracturing each core,
samples from the fragments were selected and prepared for viewing using SEM and
EDAX. Some form of ettringite was found in 26 of the 29 cores. Thus every bridge in
the sample, including the two control bridges, had some detectable ettringite.
Furthermore, ettringite was found in cores from cast-in-place as well as precast elements.
Compared to the ettringite, ASR gel occurred very infrequently. Nevertheless, it was
found in 7 of the bridges. The ASR gel particles tended to be isolated. They did not
appear to be associated with any coarse aggregate particles.
Potassium autoradiography
For technical reasons unrelated to the performance of the image plates, the number
of bridges examined by potassium autoradiography was limited to only seven. Figure 7
shows a typical pair of images from a clean and a dirty location at a specific bridge. The
images have been false colored to highlight the differences. Since there were no
The overall results for the autoradiography are summarized in Fig. 8. For each
bridge, the measured radioactivity dose for the extreme exposure conditions ranks in the
expected order: Clean back< Dirty front. For the other two conditions, in some bridges,
Dirty back < Clean front and in others, the ranking is reversed. This may indicate
reconcentration of potassium near the surface.
Impact-echo
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Subsequently, a more advanced method of processing the data was proposed20 . This
consisted of windowing the data to exclude the early transients. Next a bandpass filter
was applied with the central frequency of the filter chosen near the peak frequency of the
spectrum and with a bandwidth about 50% of the central frequency. The result is shown
in Fig. 9a. The series of echoes is much more visible. With the peaks clearly identifiable,
both positive and negative amplitudes were selected and were plotted in a semi-
logarithmic graph, Fig. 9b. The possibility of using an automated impactor rather than
the manual one for more reproducible results is also being investigated.
DISCUSSION
The primary finding is that some ettringite was detected in every bridge that was
sampled, regardless of whether or not map cracking was visually observed. This makes
the initial concerns about the probability of a false negative outcome of the survey
superfluous. The fact that ettringite was found all across Maryland over a range of
climatic conditions and geological provinces suggests that it is likely to be found in
concrete in other Northeastern states as well. Samples from neighboring states using the
same survey methodology would be highly desirable.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
258 Livingston et al.
Another significant finding is that ettringite occurs in both precast and cast-in-place
elements. Since the latter usually do not experience very high temperature excursions
during curing, this suggests that DEF may not be restricted to only steam-cured
elements. As a corollary, avoiding high temperatures during curing may not necessarily
prevent DEF.
The ubiquitous occurrence of ettringite in this sample of bridges raises the question
of whether all of it is actually due to DEF. There is the possibility that some of the
original primary ettringite persists under certain conditions rather than transforming into
monosulfate or becoming amorphous. The problem then becomes one of distinguishing
between primary ettringite and delayed ettringite in a concrete sample. This might be
done on the basis of crystal morphology. Laboratory studies have shown that the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ettringite crystal size and shape varies significantly as a function of solution pH 21 and
chemistry22 • As noted above, several characteristic crystal morphologies were identified
in this survey in the SEM investigations of the fracture surfaces of the cores. These
appear to be related to the crack patterns of the concrete. Table 2 presents some tentative
relationships.
Another significant finding of this survey is that the relationships between ASR,
DEF and map cracking are more complex than is often assumed. Ettringite was found in
every specimen with map cracking. On the other hand, ASR was not found in many of
them. Therefore, map cracking seems to be a better predictor of DEF than of the
presence of ASR. However, ettringite was also found in the control specimens, which
did not show map cracking. This suggests that map cracking may be a sufficient but not
a necessary condition for DEF. In any case, map cracking is not a reliable indicator of
ASR damage only.
CONCLUSIONS
The present knowledge of the DEF problem in concrete rests on a very small set of
data. To resolve some of the many controversies, it is necessary to obtain more data
from the field using a systematic survey methodology. Applying such a methodology to
the state of Maryland, it was found that for every bridge sampled, ettringite was detected,
but alkali-silica reaction (ASR) gel was detected only very rarely. In more heavily
damaged locations, the occurrence of ettringite crystals was more frequent, appearing in
the rims around aggregates as well as in voids. Also, the morphology of the ettringite
crystals appeared to be more lamellar than acicular. The implications are that: DEF is
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 259
widespread geographically, map cracking is not diagnostic only for ASR, and the onset of
DEF may be associated with a change in ettringite crystal morphology.
ACKNOWLEDGMENTS
The authors gratefully acknowledge financial support from the Maryland State
Highway Administration and the National Science Foundation. We would also like to
thank Joe Miller of MDSHA for his assistance with querying the Maryland bridge
inventory databases.
REFERENCES
21. Chartschenko, I., Volke, K. and Stark, J., "Investigation of the Influence of pH on
the Formation of Ettringite," Scientific Reports of the Weimar Institute for Architecture
and Construction, V. 39, No.3, 1993, pp. 171-176.
22. Cody, A.M., Lee, H., Cody, R.D. and Spry, P.G., "The effects of chemical
environment on the nucleation, growth and stability of ettringite
[Ca3Al(OH) 6]z(S04) 3 .26H20," Cement and Concrete Research, V. 34, No., 2004,
pp. 869-881.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Number of Number of
Number
Impact- Autoradio
Bridge!D# Location of
echo graphy
Cores
waveforms Locations
0208303 US 13 at ND346 4 -
0311100 CSX over I-695 6 2 -
01dCourtRd
0314200 6 2 3
overl-95
1-795 over
0333402
Franklin Blvd
4 -
I-795 Over
0333702° 0
Cockeys Mill Rd - 3
MD404over
osoo8oo•
Watts Creek - -
Ladiesburg Rd
0605800 3
over MD 632
MD 5 over Zekiah
0800201 3 3
Swamp
US 219 0\over
1102100 6 3
Deep Creek Lake
US 29 over Mid.
1301302•• 3
Patuxent River
1311000 MD 32 over US 1 6 4 3
Greentree Rd over
1511300 12 2 3
I-495
US 1 over Paint
1600400 9 2 3
Branch
1620003 MD 198 Over I-95 10 2 3
2111504 I-70 over MD 632 4 2
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Table 2: Possible Associations between Crack Patterns and Ettringite Crystal Structures
Extensive
Local Map
No Map Local Map Map
Crystal Structure Cracking
Cracking Cracking Cracking
WithASR
WithASR
Acicular in Voids X X X X
Acicular in Cracks X X X
Lamellar in Voids X X
Lamellar in Cracks X
Q)
> 0.8
~
Cl
Q)
z 0.6
Q)
VJ
m
u.
oDA
~
::c
_g 0.2
e
Q.
o.o -t---r---.-~"T""......,...--,r-~"T""--r---,--r--,
0.00 0.05 0.10 0.15 0.20 0.25 0.30
Incidence Rate
Figure 1: Probability of detecting zero cases of DEFina limited sample of 20 as a
function of actual incidence rate in the full population.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Figure 3: Maryland state bridges selected for sampling
(Filtefd wayeform)
"'"""""'"""''"T"""""'"'"'T'""""''''"'"'""""
I I
I I
I I
''"t"''''"""'""t"""'
l l
Decay constant
2.85[msY'. 0=16.5
Figure 7: Comparison of autoradiography images from clean and dirty areas of the
Maryland Route 198 overpass. The images have been smoothed and false
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
colored to highlight differences.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
266 livingston et al.
25 • Clean Back
• Clean Front
+ Dirty Dirty
• Dirty Front
Gi 20
-"!
><
·c.
15
••
~ •
">.. 10 ...
u ... •
I"CI
... •
0
•
~ 5
•• ...
0::: •• •• •• ... t
I
0~--~--~--~--,---,---,---,----
1.5
(\
0.5
...
0
'V\
/\ ,.. ~
(\ (\ (\ !\ /"'. ....
vvvv>,J--
-'3.5
-1
\
"
v
-1.5
0 0.5 Time, ms 1.5 2
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-0.5
"'-.
g
.5 -1.5
-1
• ':N
~ .
~
-2 ........
-2.5 •
-3
0 0.5 1 time, ms 1.5 2
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: The effect of cracking on chloride migration and gas permeability were
investigated on w/c = 0.49 high performance concrete. Cracks of designed widths
(ranging from so ~m to 700 ~m) were induced in the concrete specimen using
a controlled splitting tensile test. Gas permeability and chloride migration are
measured before and after loading. Chloride migration coefficients are measured
by migration test under steady state condition. Results indicate that the gas
permeability of cracked samples increased with increasing crack width. A relationship
was obtained between the chloride migration coefficient and the open area of
crossing cracks and a good comparison between the measurement and numerical
simulation was obtained. This result confirms the validity of the model of transfer
in parallel. A comparison between the gas permeability and chloride migration
coefficient was given. This indicates that the gas permeability is significantly more
sensitive than the chloride migration coefficient with respect to the crack widths used
in this study.
Djerbi Assia is a PhD Student at the Research Institute of Civil and Mechanical
Engineering- IUT, Technical Institute of Saint-Nazaire -France. She is working on the
influence of damage on mass transport in porous media.
INTRODUCTION
The objective of this research was to examine the influence of crack width on gas
permeability and chloride migration into concrete. Cracks of designed widths were
induced in the concrete specimen using a controlled splitting tensile test. After loading
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
EXPERIMENTAL PROGRAM
with nitrogen as the neutral percolating gas 7• The relative pressure (Pi- Patm) applied to
the sample ranged between 0.05 MPa and 0.3 MPa and was measured by a digital
pressure gauge with a 100 Pa precision. Permeability measurements were made in an air-
conditioned room (20 ± 1oc and RH 50± 5%). Each disc was tested with five differential
pressures: 0.05, 0.1, 0.15, 0.2 and 0.3 MPa for uncracked specimen and with five
differential pressures: 0.01, 0.015, 0.02, 0.025, 0.03 MPa for cracked specimen. The
establishment of steady state flow before an actual measurement required a significant
time (varying from 30 minutes to several hours). This condition was verified by taking
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
272 Djerbi et at.
two measurements separated by a 10 minutes interval. If two values differed by less than
2%, a steady state flow condition was assumed to be achieved. For each differential
pressure, the apparent coefficient of permeability kA (m2 ) is calculated from the Hagen-
Poiseuille expression (Eq 1) for laminar flow of compressible viscous fluids through a
porous bodl.
Q 2).! L P,tm
kA A (P;2 -P,~) (1)
where L is the thickness of the sample (m), A is the cross-sectional area (m2 ), Q is the
gas flow (m 3/s), ).lis the coefficient of viscosity (1.78.10- 5 Pa.s for nitrogen gas at 20°C),
Pi is the applied absolute pressure or inlet pressure (Pa), and P atm is the atmospheric
pressure (Pa).
In fact, the gas percolation through a fine porous body like concrete, can be
regarded as resulting from two flow modes: viscous flow and slip flow or Knudsen flow.
Various methods for the calculation of non-viscous flow exist. The most widely used is
the relation proposed by Klinkenberg 8 (Eq 2) introducing the concept of an intrinsic
coefficient of permeability kv relative to viscous flow only.
kA =kv(l+ ~J (2)
where Pm is mean gas pressure, Pm =(Pi+ Patm)/2, b is the Klinkenberg coefficient (Pa)
which is function of the porous body and the infiltrated gas, and Kv is the limiting value
of gas permeability when the mean pressure Pm tends towards infinity. The method of
determination of kv consists in measJ.Iring kA at different pressures (Pi) and plotting it
against the inverse of the mean pressure (1/Pm) (Fig 3). The slope of the line leads to the
empirical Klinkenberg coefficient b.
A polynomial regression was then carried out by assuming the relationship given in
Eq 7. The Fitted parameter is F, with F=~ y3• The adjustment between the relative
increase in permeability measured and the model is given in Fig 6. The values of the
correcting factor ~ can be evaluated from the adjusted value of y (2nd row of Table 2) or
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
274 Djerbi et al.
from the average width measurements of the main crack by videomicroscope w using the
relationship given in Eq 6 (3rd row of the Table 2). The values~ obtained are almost the
same.
Fig 9 shows the relationship between the open area of transecting cmcks and migration
coefficient. This last is increasing by a factor of about 3.5 for the most cmcked concrete by
comparing to the coefficient of uncmcked specimen. The migration coefficients obtained from
these experiments will be compared with the model of tmnsport in pamllel. This model could
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
D~(d)=A~'+D,0 (9)
where De"(d) is the migration coefficient in cracked specimen and Deo is the migration
coefficient in uncracked specimen. This result confirms the validity of the model of
transfer in parallel.
Numerical simulations of this model were carried out by combining Eq 3 and the
continuity equation written as following:
ac(x,t) 8J(x,t)
p--=--- (10)
at ax
where p is the concrete porosity. In this study, interactions between chloride ions and
concrete were neglected. Using Eq 3 and Eq 10, the migration problem can expressed as:
=~[8 c(x,t) _FE c(x
2
8c(x,t) t)]
at p axz RT , (11)
The flux within the crack Jc(x,t) and the migration equation problem within the crack can
be written in the same manner:
J (x t)=-D 8c,(x,t)+D FE
, , , ax , RT c, (x, t ) (12)
2
8c,(x,t) =D [8 c,(x,t) FE c (x t)] (13)
at ' ax 2
RT ' '
where cc(x,t) is the ions concentration in the crack. Eq 11 and 13 were solved using finite
differences with an implicit scheme (in order to avoid numerical instabilities). The limit
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 275
conditions are: c(O,t)=0.5 mol/1 and c(L,t)=O mol/1. The cumulative ions quantity in the
downstream compartment is obtained by time integration of J1(L,t) (Eq 8). Figure 10
shows the comparison of the numerical simulations and the measurements for various
crack areas. In spite of the relative simplicity of the parallel model, simulations are in
good agreement with measurements.
CONCLUSIONS
In this study controlled splitting tests were performed to obtain a series of concrete
specimens with different crack widths. The results indicate the relationship between the
crack recovery and the maximum crack opening displacement.
Gas permeability of concrete with different crack widths was evaluated. The results
suggest that permeability increases with COD cubed, as shown in theoretical models of
cracked rocks. A parameter y, accounting the tensile crack density and the crack-width
distribution was introduced to estimate the reduction factor ~- The value of this factor is
0.028.
The chloride migration coefficient was determined on cracked concrete using steady
state migration test. The results indicate that the chloride migration coefficient increased
with increasing crack width and the time to penetrate through the specimen was reduced
which is very important to evaluate the initiation corrosion. A relationship was found
between chloride migration coefficient and the open area of transecting cracks and a good
comparison was obtained between measured values and numerical simulations. The
analysis of the results confirms the validity of the model of transfer in parallel.
A comparison was given between the gas permeability and chloride migration
coefficient. The result showed that the gas permeability was significantly more sensitive
than chloride migration with respect to the crack widths used in this study. This can be
explained from the theoretical models that show that the rate flow of gas through cracked
concrete is proportional to the cube of the crack width, whereas the chloride migration
coefficient through cracked concrete is shown to be proportional to the crack area. The
different relationships obtained in this study have to be checked with other concretes.
ACKNOWLEDGMENT
This study is supported by the European Union in the frame of the EU program
Interreg III B. This EU project MEDACHS deals with the maintenance and reliability of
structures exposed to marine environment (Atlantic coast).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3. Picandet, V., Khelidj, A., Bastian, G., "Effect of Mechanical Damaging of Gas
Permeability of Concrete", Proceedings of the 2nd International Symposium on
Cement and Concrete Technologie, Istanbul, Turkey, vol. 2, 2000, pp. 322-331.
4. Wang, K., Jansen, D.C and Shah, S.P, "Permeability study of cracked concrete",
Cement and Concrete Research, vol. 27 No.3, 1997, pp.381-393.
5. Aldea, C., Shah, P., Karr, A., "Effect of cracking on water and chloride
permeability of concrete", Journal of materials in civil Engineering, vol. 11 No.3,
1999, pp. 181-187.
6. Olga, G., "Influence of cracks on chloride ingress into concrete", ACI Materials
Journal, vol. 100 No.2, 2003, pp. 120-126.
11. True, 0., "A new way for determining the chloride diffusion coefficient in concrete
from steady state migration test'', Cement and Concrete Research, vol. 30, 2000,
pp.217 -226.
HPC
F=E;( 3.8.10"3
_1,_value ify = 0.5 0.030
C, value detennined from the main crack width w 0.027
measurement
COD(!lm)
t~ag(hour)
After loading
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0 72.38
15.76 63.41
37.06 57.76
100.16 56.89
200.88 27.83
390.40 0
I I
30 ------~--------~---------1---------~--------
1 I I
25
~20
]...:1 15 I
I
I
10 l---------
1
I
I
·------~--------
5 _ _ _ _ _ _ _ _ 1_ _ _
30 40 0 60
COD after loading COD(J.lm) Wanted COD under
loading
6.E-17 ~-----~----------,-------.--------.---------,
I I
S.E-17 ---------T---------,--------
I kA
- !
I
-r---------
I I
I I
I
~ 4.E-17
.£
] 3.E-17
-
j 2.E-17
l.E-17
O.E+OO
0 2 4 6 8 10
tfPm (Mpa- 1)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
!Oil
600
I
1
I
1 I
I
I
I
I
I
-----+-----~------!------~-----+-----
I
I
I
I I
I
------1------
I
I I I 1 I
~ 500 ----- T----
I
-~----------- -~-Recowrcd.COD
I I
I
I
8 400
0 I
Jl.. 300
Q
8 200 I
I I I
100 - - - - -1- - - - - - t-- - - - - + - - - - -..,---- - -r-- - - - -
I I I I I
I I I I
I I I I
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0
0 100 200 300 400 500 600 700 800
COD under Loading (Jlm)
l.E+05
I
I
I
I
I I
l.E+04 L---,-----~-----,----------,--------
1
I I
I I I
I I I I I I I
l.E+03 ----~-----{-----L ---:-----~-----}-----:-----~-----
:l I
I
I
I
1 I
I
I
I
I
I
""'
';i
I
I I I
I I
I
I
I I
l.E+02 - ..l -
I
- - - - L - - - - -1- - - - - ...J - - - -
I I I
- l- - - - - -1- - - - - ...J - - - - -
I I I
I I I I I I I
I I I I
I I I I
I I I I I I I I
l.E+Ol ~-----~-----t------1-----~-----~-----l-----~-----
1 I I I I I I I
I I I I I I
I I I I I
I I
I
l.E+OO
0 50 100 150 200 250 300 350 400 450
COD(j1II1)
-+-COD=390.4 m : : 1
--------~--------r--------r--------r---- ---r--------
1 I I I I
I I I I I
________ L ________ L ________ L _____ _
I I I
I I I
I I I
--------~--------~-------
1 I
I
7E-12 - . - - - - - - - , - - - - - . . - - - - - - - - , - - - - - , - - - - - ,
r-----~--------~1
I
SE-12 ---------~-----
i
1
I I
I I
_________ l _______ _ I
,..._ 4E-12 I ---------,----------
~ 3E-12
I 1
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
I I I
--------- ---------~---------~----------'----------
1 I I
I I
I
2E-12 --------~---------~---------~---------~----------
1 I
I
IE-12 + - - - - - - t - - - - - + - - - - - - - - J - - - - - t - - - - - - - 1
O.OE+OO S.OE-06 l.OE-05 l.SE-05 2.0E-05 2.5E-05
A.,(m')
~ I
_____ L _____ L _____
I I
I_
..§, 0.12 I I I
I I
]0 I
I
I
I
I
I
6 o.o8
_____ L _____ L
I I
0.04
0 2 4 6 8 10 12 14 16 18
Time (Days)
1E-12 ~---.,...----.-------..----,------,,....•:----,
1
I
I
I I I I
!E-13 -------~--------L-------
1 I
-----1--------~--------
1 I
I I
I •
I
1E-14 -------~--------r------
~
.,; •
1E-15
I •
I
\E-16 .1
-------~-------- -------1--------~-------
.I
1 I
I I I
I I
I
IE-17~------~------~------~------~------~----~
D.(m'/s)
Fig. 11- Relationship between chloride migration coefficient and gas permeability
ACI member Alain Bilodeau is a research engineer, ICON, CANMET. His research
interests include the use of supplementary cementing materials and durability of concrete.
INTRODUCTION
The results of laboratory studies have shown that the depth of carbonation is greater when
the amount of Ca(OH)z is lower, which is generally the case of fly ash concrete (1 ).
However, published data have also shown that the rate of carbonation is strongly
influenced by the water-cementitious materials ratio (w/cm) and moist curing period, and
the level of fly ash in the range of 0 to 30% has relatively little impact (2). According to
Thomas et a!. (2), concrete containing 50% fly ash carbonates more rapidly than
equivalent grade concrete without fly ash. However, this effect can easily be
compensated for by small reduction in w/cm and increase in the design strength (e.g. 5
MPa) or by extending the period of moist curing (e.g. from 1 to 3 days). Nagataki eta!.
(3) reported a direct relationship between compressive strength at 28 days and depth of
carbonation irrespective of fly ash replacement in concrete. They found that the extent of
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`--- Durability of Concrete 285
carbonation decreased with an increase in compressive strength; the fly ash content
investigated was 30%, and the 28-d compressive strength of concretes ranged from less
than 20 to about 40 MPa. However, it should be mentioned that although the statement
that the rate of carbonation is a function of the strength of concrete is generally true, it is
an inadequate simplification. The rate of carbonation of concrete is a function of the pore
system of the hardened cement paste that is mainly affected by the type of cementitious
materials used, the w/cm, and the period of moist curing.
The results of the study by Ho and Lewis (4) suggest that with 7 days moist curing, the
fly ash concrete specimens (fly ash content ranging from 20 to 40%) tend to have greater
depth of carbonation, but as the curing period was extended to 90 days, the fly ash
concrete showed improved resistance to carbonation. According to Joshi et al. (5), after
90 days of curing the fly ash concrete (with fly ash content ranging from 40 to 60%)
exhibited less carbonation than that of the control concrete.
Also, most of the investigations on carbonation of fly ash concrete involved exposure of
relatively immature concrete to 50-60% relative humidity and 3-4% C0 2 atmosphere
which is at least 10 to 100 times that of exposure under normal field conditions. Fly ash
concretes exposed for long periods in field conditions have resulted in significantly
reduced depth of carbonation (6).
Based on the above, HVF A concrete as developed at CANMET by Malhotra and his
associates (7) that contains 55 to 60% fly ash and has a w/cm in the order of 0.30 (an
average 28-d compressive strength of35 MPa) should have high resistance to carbonation
due to its low w/cm and hence very low permeability, provided the concrete is well
cured. However, only limited published data on the carbonation of HVF A concrete are
available that support the above statement. Data on carbonation of a high-volume fly ash
concrete exposed to outdoor conditions in Ontario, Canada for 10 years were obtained
from a CANMET/Industry joint research project (8). This long-term investigation
included six non-air-entrained concretes. The HVFA concrete contained 150 kg/m 3 of
ASTM Cl50 (9) Type I portland cement and 200 kg/m 3 of ASTM C618 (10) Class F fly
ash, with a w/cm of 0.29 and a 28-day compressive strength in the order of 50 MPa. The
other five concretes all contained approximately 485 kg/m 3 cementitious materials with
w/cm ranging from 0.27 to 0.29; one mixture contained portland cement only, one
incorporated 35% ground granulated blast furnace slag, two incorporated 8 and 12%
silica fume, and one incorporated a combination of 7% silica fume and 28% slag. After
10 years, the HVF A concrete had a carbonation depth of only 3 to 5 mm, that is
considered negligible, and the other five concretes had carbonation depths of less than 1
mm.
In another CANME'F study (11), three large blocks, 3.05x3.05x3.05 m, were cast in
Halifax, Canada, using three concrete mixtures i.e. a control concrete made with 400
kg/m 3 of portland cement and a w/c of 0.33 (28-d compressive strength of 45 MPa), a
HVF A concrete made with 180 kg/m3 of portland cement, 220 kg/m 3 of Class F fly ash,
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
286 Bouzoubaa et al.
and a w/cm of 0.27 (28-d compressive strength of 44 MPa), and a lean HVF A concrete
made with 100 kg/m3 of portland cement, 125 kg/m 3 of Class F fly ash, and a w/cm of
0.49 (28-d compressive strength of 22 MPa). After casting, a wooden enclosure was built
to house the blocks to provide shelter. After 14 years, both the control and the HVFA
concretes had a carbonation depth ranging from 0 to 3 mm, and the lean HVF A concrete
had a carbonation depth of about 15 mm.
A recent study in India (12) on the use of HVF A concrete in structural applications has
shown that at 28 days the carbonation depth of HVF A concrete (25 mm) determined by
an accelerated carbonation test was higher when compared to that of ordinary portland
cement concrete (9 mm) having similar 28-d compressive strength, but at the ages of 56
and 91 days, HVFA concrete had only a marginally higher carbonation depth (11 - 12
mm) than that of ordinary portland cement concrete (7- 8 mm).
Present Investigation
In 1997, CANMET initiated a joint program with Hydro-Quebec and the Canadian
Electrical Association (CEA) to investigate the carbonation of high-volume fly ash
concrete. This program consisted of investigating three high-volume fly ash concrete
mixtures (HVF A) incorporating 58% of fly ash, each made with a different fly ash i.e.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
two ASTM C618 (10) Type F fly ashes, and one ASTM C618 (10) Type C fly ash, and
two control portland cement concrete mixtures, one with similar w/cm (0.32), and the
other with similar 28-d compressive strength as that of the HVF A concrete.
This paper presents the results on carbonation of the above concrete using an accelerated
testing method in which concrete specimens were exposed to a 3% C02 at 23 °C and 65%
relative humidity for up to 700 days. The carbonation of concrete specimens exposed for
more than seven years to natural environments (indoor and outdoor without shelter) was
also determined for comparison.
Cement
ASTM Cl50 Type I normal portland cement was used, and its physical properties and
chemical analysis are given in Table 1.
Fly Ash
Fly ashes from Point Tupper (Nova Scotia), Thunder Bay (Northern Ontario), and
Sundance (Alberta) were used. Their physical properties and chemical composition are
also given in Table 1.
The Point Tupper fly ash, an ASTM C618 Class F ash (CSA Type F), contained 2.8%
CaO and had a specific surface of236 m2/kg (Blaine).
The Sundance fly ash had a CaO content of 12.5%, but its combined (Si02 + Al203 +
Fe20 3) content was higher than 70%, thus meeting the requirements of ASTM C618
Class F fly ash (CSA Type CI). The ash had a relatively high specific surface of 408
m2/kg (Blaine).
Aggregate
The coarse aggregate used was crushed limestone with a maximum nominal size of 19
mm, and the fine aggregate was natural sand from the Ottawa region. The grading of the
aggregates is given in Table 2. The coarse and fine aggregates had a specific gravity of
2.7, and water absorptions of0.6 and 0.8%, respectively.
Superplasticizer
Air-Entraining Admixture
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A synthetic resin type air-entraining admixture was used in all the concrete mixtures.
MIXTURE PROPORTIONS
Five air-entrained concrete mixtures were studied in this program, and the mixture
proportions are summarized in Table 3. Three concrete mixtures incorporated 58% of
different fly ashes by mass of the total cementitious materials with a water-cementitious
materials ratio (W/(C+FA)) of0.32. Two control portland cement concrete mixtures were
made for comparison; that included one with the same water-cementitious materials ratio
of 0.32 and the other (w/cm of 0.42) proportioned to have a 28-day compressive strength
equivalent to that of the HVF A concrete.
The concrete was mixed in a laboratory counter-current mixer for 6 minutes. The
properties of the fresh concrete including the slump, air content, and unit weight were
determined immediately after the mixing according to relevant ASTM Standards, and are
given in Table 4.
For each of the mixtures, three 102x203-mm cylinders were cast for determining the 28-
day compressive strength, and twelve 100x100x400-mm prisms were cast for the
carbonation testing.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
288 Bouzoubaa et al.
All the test specimens were compacted on a vibrating table. After casting, the specimens
were covered and left in the casting room for 24 hours. The specimens were then
demoulded and cured in a standard moist curing room at 23±2 oc and 100% relative
humidity until specified ages.
CURING REGIMES
Prior to the accelerated carbonation test, the concrete was subjected to three different
curing modes; these are as follows:
Curing mode 1 - 1 day in mould, 6 days in moist condition (23 oc and 100%
relative humidity) and 84 days in laboratory air;
Curing mode 2 - 1 day in mould, 27 days in moist condition and day in
laboratory air;
Curing mode 3 - 1 day in mould, 90 days in moist condition and day in
laboratory air.
The carbonation of concrete was also determined on samples exposed to indoor and
outdoor natural environment. The concretes used for the carbonation test in indoor
environment were moist cured for 7 days, and then left in laboratory air. For those
exposed outdoor, the concrete was left in laboratory air for 49 days periodically wetted
after an initial moist curing for 7 days before they were moved to an outdoor exposure
site. The average temperature and humidity at the outdoor site ranged from -16 to 27 °C,
and from 53 to 90%, respectively. The monthly average precipitation ranged from 64 to
89 mm, and during the winter (end ofNovember till mid of April), the samples were most
of the time covered with snow.
TESTING OF SPECIMENS
The compressive strength was determined at 28 days in accordance with ASTM standard
C 39 (13). For the accelerated carbonation testing, two prisms each were preconditioned
as described previously. The specimens were then transferred to a chamber with 3% C02
at 23±2 oc and 65% relative humidity. The depth of the carbonation was determined at
frequent intervals up to the age of 700 days following RILEM Recommendation CPC-18
(14). A slice of concrete was sawn off from the concrete prisms for each test. The freshly
broken surface was cleaned of dust and loose particles and then sprayed with a
phenolphthalein indicator. Measurements ·of the carbonation depth were made at two
points along each side-casting surface of the concrete slices, and the four values obtained
were averaged. In this test, the depth of the uncoloured zone is designated as the
carbonation depth.
For carbonation testing of test prisms exposed to the natural exposure conditions, three
prisms were used for the indoor and three prisms were used for the outdoor exposure.
The depth of carbonation was determined at various intervals up to the age of 7 years
following the above procedure.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The properties of the fresh concrete including the slump, unit weight, and air content and
the compressive strength of the hardened concrete are summarized in Table 4. The slump
of the concrete mixtures ranged from 55 to 195 mm, and the air content ranged from 5.3
to 6.2%. The unit weight of the concrete ranged from 2300 to 2400 kg/m 3 . For the three
HVFA concrete mixtures, the 28-day compressive strength ranged from 35.4 to 44.0
MPa. The control portland cement concrete with a w/c of 0.42 had a 28-day compressive
strength of 42.8 MPa, that was similar to those of the HVFA concretes incorporating the
Thunder Bay and Sundance fly ashes, but approximately 20% higher than that
incorporating the Point Tupper fly ash. The 28-day compressive strength of the control
portland cement concrete with the same w/cm as the HVFA concretes was 57.1 MPa that
was higher than those of the HVF A concretes due to its much higher cement content.
Carbonation of Concrete
Accelerated carbonation test -- Figures 1 to 3 show the carbonation depth of the high-
volume fly ash concrete and control portland cement concrete cured under different
conditions. The results show that the carbonation depth of the HVF A concrete was
greater than that of the control portland cement concrete (w/c of 0.32 and 0.42). After
being moist cured for 7 or 28 days and followed by 700 days exposure to carbonation
(Figs 1 and 2), the concrete incorporating Point Tupper fly ash had greater carbonation
depth of 35 to 37 mm than concrete incorporating Thunder Bay (26 to 32 mm) or
Sundance fly ash (25 to 34 mm). This may be attributed to the slow reactivity of the
Point Tupper fly ash as shown by its lower strength activity indices (Table 1) and its
lower 28-d compressive strength. However, with the increase of moist curing to 91 days,
the concrete incorporating Point Tupper fly ash had the lowest carbonation depth,
whereas, the concrete incorporating Thunder Bay that showed the highest reactivity
(Table 1) had the highest carbonation depth.
The above results show that the effect of fly ash on the carbonation resistance of concrete
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Natural exposure -- Figures 4 to 6 show the carbonation depth of the concrete specimens
exposed to indoor (7 days moist curing) and outdoor natural exposure (7 days and 28
days moist curing). The above figures show the same trend as that observed with the
accelerated test, i.e. the HVF A concretes carbonated more than both control concretes,
and the depth of carbonation of the HVF A concrete made with Point Tupper fly ash was
generally more than that incorporating Thunder Bay or Sundance fly ash. However, it
should be mentioned that the depth of carbonation determined on specimens exposed to
natural environment were significantly lower than the corresponding specimens tested
following the accelerated test procedure. For example, the depth of carbonation of
HVF A concretes exposed to outdoor environment for seven years after being moist cured
for 7 days (Fig. 5) was about 7.5 mm for that incorporating Point Tupper fly ash, 6.5 mm
for that incorporating Thunder Bay fly ash, and about 3.5 mm for HVF A concrete
incorporating Sundance fly ash; the depth of carbonation of the control concrete with a
w/c of 0.42 was 2.5 mm and that of the control concrete with a w/c of 0.32 was less than
1 mm. This is due primarily to the facts that first the C0 2 concentration into the
atmosphere is significantly lower than that used in the accelerated test chamber, and
secondly, the humidity in the concrete specimens exposed outdoor increases from time to
time due to the rain and snow, which slows down significantly the ingress of C02 into the
concrete. Nevertheless, the above results show that although the accelerated test might
overestimate the carbonation depth of concrete it certainly gives a good idea on the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
relative carbonation resistance of different concretes.
For each given curing mode and exposure, data were fitted using the above equation and
the regression results are presented in Table 5. For the accelerated test, the carbonation
coefficient of the control concretes ranged from 0.06 to 0. 72 mm/d 0·5• The lowest value
was for the control concrete with lower w/c that had been moist cured for 28 days, and
the highest value was for the control concrete with higher w/c and lower moist curing
period. For the indoor exposure, the carbonation coefficient of the control concrete with
w/c of 0.32 was 1.14 mm/y 0·5, and that of the control concrete with w/c of 0.42 was 2. 74
mmJl·5 • For the outdoor exposure, the carbonation coefficient of the control concrete
with lower w/c could not be calculated since the carbonation depth was nil all the time.
For the control concrete with higher w/c, the carbonation coefficient was 0.78 mmJl· 5 for
the specimens moist cured for 28 days and 0.31 mm/y 0·5 for those moist cured for 7 days.
The carbonation coefficient of the HVF A concretes followed the same trend i.e. it
increased with decreasing moist curing period and decreasing compressive strength.
Also, the carbonation coefficients of the concrete subjected to outdoor exposure were
lower than those of the concrete subjected to indoor exposure due to the effect of rain and
snow that increased the humidity at the surface of the concrete, that in tum retarded
significantly the ingress of the C02 into the concrete.
Based on the carbonation coefficient values presented in Table 5, the time required for
the carbonation front to reach a certain depth can be predicted using the above equation.
According to CSA A23.1 04 Clause 6.6.7.5.1 (18), the minimum concrete cover for
structural elements exposed to conditions prone to carbonation-induced corrosion is 40
mm, provided the concrete is not cast against and permanently exposed to earth. For
example, for HVF A concrete incorporating Point Tupper fly ash and moist cured for 7
days, the carbonation front will reach 40 mm depth from the surface exposed to outdoor
environment in approximately 240 years. But it will take only 60 years for the
carbonation to reach the same depth from the surface exposed to indoor environment.
However, in indoor environment, although the carbonation front might reach the
reinforcing bars faster, the humidity is usually not high enough in most locations to
promote corrosion.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Even though extensive data have shown the substantial effect of composition and/or
curing on the carbonation of concrete, a universal relationship between accelerated test
and natural exposure remain unsatisfactory. However, according to Ho and Lewis (19),
results from accelerated carbonation, da. can be related to those from natural exposure, dm
by two factors a and ~, as follows
(2)
In the present study, a more direct relationship was used to correlate between the
accelerated carbonation coefficient and the natural exposure carbonation coefficient.
This relation is expressed as follows:
(3)
The values of yare presented in Table 6. The results show that for the control concrete
(w/c = 0.42), the value of yranges from 1.08 to 1.19. Whereas, for HVFA concrete the
value of yranges from 1.21 to 1.52, from 1.84 to 2.12, and from 0.79 to 1.45 for Point
Tupper, Thunder Bay, and Sundance fly ash, respectively. This means that for the
control concrete, the carbonation coefficient (in days) determined from the accelerated
test could be multiplied by a maximum value of 1.2 to reflect the carbonation coefficient
(per years) of a structure made with the same concrete and exposed to Canadian natural
environment. For HVFA concretes, this value ranges from 0.8 to 2.1 depending on the
fly ash used. It appears difficult to relate the parameter yto a given property of fly ashes.
However, when compared to Point Tupper and Thunder Bay concretes, HVF A concrete
incorporating Sundance fly ash with the higher specific Blaine surface tends to have
lower values of y.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1. For a given water-cementitious materials ratio, fly ash percentage, and total
cementitious materials content of the concrete, the type of fly ash has an effect on the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
depth of carbonation which seems similar to its effect generally observed on the
permeability of concrete.
2. For both the HVFA and the control portland cement concretes, the carbonation depth
of the concrete decreased with an increase in the duration of moist curing from 7 to 28
days. However, a further increase in the moist curing from 28 to 91 days did not
substantially affect the carbonation depth of the control concretes and the HVF A concrete
incorporating the highly reactive fly ash.
4. The depth of carbonation of control and HVF A concretes can well be approximated by
a linear function of the square root of time. The coefficient of correlation ranged from
0.71 to 0.98 for the concretes investigated. Based on this equation, the time required for
the carbonation front to reach 40 mm depth in the HVF A concretes exposed to outdoor
conditions was predicted to be minimum 240 years.
5. The results show that the carbonation is not an issue for well-cured HVFA concrete
due mainly to the low water-cementitious materials ratio, and hence very dense structure
and low permeability of this type of concrete.
ACKNOWLEDGEMENTS
REFERENCES
l. Bier, Th. A., "Influence of Type of Cement and Curing on Carbonation progress and
Pore Structure of Hydrated Cement Paste", Materials Research Society Symposium,
Vol. 85, 1987, pp. 123-134.
2. Thomas, M.D.A., Matthews, J.D., and Haynes, C.A., "Carbonation of Fly Ash
Concrete", ACI SP-192, 2000, pp. 539-556 (Editor V.M. Malhotra).
3. Nagataki S., Ohga H., and Kim E.K., "Effect of Curing Conditions on the
Carbonation of Concrete with Fly Ash and the Corrosion of Reinforcement in Long
4. Ho D.W.S. and Lewis R.K., "Carbonation of Concrete and its Prediction, Cement
and Concrete Research", Vol. 17, No.3, 1987, pp. 489-504.
5. Joshi R.C., Lohtia R.P., and Salam M.A., "Some Durability Related Properties of
Concrete Incorporating High Volumes of Sub-bituminous Coal Fly Ash", Third
CANMET/ACI international Conference on Durability of Concrete, Supplementary
papers, 1994, American Concrete Institute, Farmington Hills, Mich., pp. 447-464
(Editor V.M. Malhotra).
6. Tsukayama, R., Tagataki S., and Abe H., "Long Term Experiments on the
Naturalization of Concrete Mixed with Fly Ash and Corrosion of Reinforcement",
Proceedings, ih International Congress on the Chemistry of Cement, 1980, VII-
ISCC, Paris.
8. Malhotra, V.M., Zhang, M.-H., Read, P.H., and Ryell, J., "Long-Term Mechanical
Properties and Durability Characteristics oh High-Strength/High-Performance
Concrete Incorporating Supplementary Cementing Materials under Outdoor
Exposure Conditions", ACI Material Journal, Vol. 97, No.5, 2000, pp. 518-525.
10. ASTM C618-03, "Standard Specification for Coal Fly Ash and Raw or Calcined
Natural Pozzolan for Use as a Mineral Admixture in Concrete", ASTM International
West Conshohocken, Pa., 2003, 3pp.
12. SERC Report, "Demonstration of Utilising High Volume Fly Ash Based Concrete
for Structural Applications", Report prepared by Structural Engineering Research
Centre (SERC) for Confederation of Indian Industry, Project No. GAP-2041, 2005,
p. 46.
16. Tuutti, K, "Corrosion of Steel in Concrete", Swedish Cement and Concrete Research
Institute, CBI Fo 4:1982, Stockholm, Vol. 51, No. 469, 1982, p. 18.
17. Hobbs, D.W. (editor), "Minimum Requirements for Durable Concrete", British
Cement Association, 1998, Crowthome, Berks (U.K.).
19. Ho D.W.S. and Lewis R.K., "The Specification of Concrete for Reinforcement
Protection - Performance Criteria and Compliance by Strength", Cement and
Concrete Research, Vol. 18, No.4, 1987, pp. 584-594.
Table 1 - Physical properties and chemical analysis of the cements and fly ashes used
ASTM Flvash
Type I
Cement Point Tupper Thunder Sundance
Bav
Physical Tests
Specific gravity 3.13 2.67 2.43 2.00
Fineness
-passing 451J.m, % 93.1 90.2 89.7 90.9
-specific surface, Blaine, m 2/kg 400 236 240 408
-median particle size, IJ.m - 11.1 13.2 10.5
Compressive strength of 51 mm cubes, MPa
-7-day 34.8 - -
-28-day 41.2 - - -
Water requirement,% - 95.0 93.0 93.8
Strength Activity Index, %
-7-day - 75.2 85.1 80.0
-28-day - 92.8 104.3 106.9
Chemical Analyses %
Silicon dioxide (SiOz) 21.36 40.71 41.99 53.30
Aluminum oxide (Alz03) 3.98 17.93 21.44 23.63
Ferric oxide (F e203) 3.15 29.86 4.45 4.40
Calcium oxide (CaO) 62.41 2.80 15.81 12.45
Magnesium oxide (MgO) 2.57 1.09 3.18 1.15
Sodium oxide (Na20) 0.20 0.73 7.03 3.03
Potassium oxide (KzO) 0.80 1.56 0.36 0.42
Equivalent alkali (Naz0+0.658K20) 0.73 1.76 7.27 3.31
Phosphorous oxide (P20s) 0.21 0.17 0.58 0.12
Titanium oxide (Ti02) 0.19 0.85 1.05 0.71
Sulphur trioxide (S0 3) 3.43 1.27 1.79 0.20
Loss on ignition 1.72 1.95 0.75 0.71
Bogue Potential Compound Composition
Cumulative Cumulative
Sieve size percentage Sieve size percentage
(mm) retained (mm) retained
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
f --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
[ ~~~~~~~
(")(")(")(")(")<>!!
I""· >-'j>t:i I
i' g'
==a~«
e-
if 9,
!;' >-'j
>rj I
l
I I 'l> I
5 ~fn lable J - Mtxture proportions ot the concrete
.. :
I-t ("0 I
> <l tl)'tl cr' I
~ ~-a
~tities kg/m
3
)
t-------+-----11
8 ~; Cement Fly ash type Fly ash
Fine Coarse
00~~~~~;~ Mix type content,%
Water• Cement Fly ash agg. agg. SP"
A.E.A.·"
a~: No. W/C+SCM (mL/m3 )
~------~----~1 NPC 58
Cl Point Tupper 0.32 118 155 216 742 1113 5.1 421
00000 ~ i C2 NPC Thunder Bay 58 0.32 118 156 216 735 1104 2.1 161
~~~~~ 2* I C3 NPC Sundance 58 0.32 118 155 216 712 1070 4.0 501
1-------+----t!
1 C4 NPC - 0 0.32 121 373 0 755 1134 13.0 670 c
en C5 NPC - 0 0.42 156 373 0 723 1085 1.3 141 c
Vo----ss
Vl~v::~~e~
I
I NPC - normal Portland cement DJ
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
~Y.?'-~?'~8~~
<l I
b'
:::s
0
\Ht.Jotv0tv .!
n
-
[
(a
tD
.j:>.t.Jo.j:>..j:>.\H~a-N
~~~~Y.~=s~
"'
oo-wo.;..,~""
N
\0
Daneshlink.com
- .......
298 Bouzoubaa et at.
Table 5 -Effect of curing, exposure and fly ash source on the carbonation parameters k;
anddlO
Regression Parameters Time required
for a carbonation
MixNoo Curing Exposure
k; do If depth of 40 mm,
year
Accelerated 2o07 mm/d 112 - 3o74 Oo98 -
Mode 1* Indoor 4o90mmtv'12 1.44 Oo89 62
.5 ~
-uog. Outdoor 2051 mmtv'12 Oo79 Oo91 240
ll.E'-< Accelerated 1.32 mrnld 112 Oo51 Oo99 -
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Mode2**
Outdoor 2000 mrn/y112 0 Oo92 400
...
Accelerated 1.01 mm/d 112 4.41 Oo95 -
Mode 1 Indoor 5o04mmtv112 Oo85 Oo92 60
"
N"'>-.
U § as Outdoor 2o14mmly112 1.10 Oo82 330
!:l~ Accelerated Oo96mmld112 1.50 Oo98 -
Mode2 112
Outdoor 1.77 mm/y 0 Oo84 510
II)
Accelerated 1.41 mm/d 111 1.45 Oo96 -
0 Mode 1 Indoor 4o22mmlyll2 Oo81 Oo87 86
<')~ Outdoor 1.12 mm/y 112 Oo77 Oo71 1225
u§
rn Accelerated 1.03 mm/dl/2 Oo34 Oo98 -
Mode2
Outdoor 1.49 mm/y 112 0 Oo77 720
Accelerated 0.40 mm/d 111 1.28 Oo84 -
Mode 1 Indoor 1.14 mmly 112 Oo59 Oo77 1196
C)~ Outdoor - - - -
Accelerated Oo06mmld 112 0 Oo78 -
Mode2
Outdoor - - -
Accelerated Oo72 mm/d 111 Oo98 Oo97 -
Mode 1 Indoor 2o74mmlyl/2 Oo51 Oo89 208
V'lu
u~ Outdoor Oo78 mmly 111 0 Oo91 2630
Mode2
Accelerated Oo26mmldl/2 0 Oo85 -
Outdoor 0.31 mm/y112 0 Oo86 16645
00
• 1 day m mould, 6 days m a standard m01st cunng room (23 oc and 100% relative humidity) and
84 days in laboratory air
.. 28 days in the moist curing room and 1 day in laboratory air
Table 6- y Correlation values between the accelerated and the natural exposure test of the
investigated concretes
MixNoo ka kn y
(mm/day05) (mmfyear0·5)
Cl
1.32 to 2o07 2o00to 2o51 1.21 to 1.52
Point Tupper
C2
Oo96 to 1.01 1.77 to 2.14 1.84 to 2012
Thunder Bay
C3
1.03 to 1.41 1.12 to 1.49 0079 to 1.45
Sundance
C4
W/C=Oo32 NPC
Oo06 to 0.40 - -
cs 0026 to 0072 Oo31 to Oo78 1.08 to 1.19
W/C=0.42 NPC
15
'0 I
I
t..
I I I
10 T----~----,----~----
1 I I I
'tl I I I I I I
5 _____ L ____ l ____ i----~----~----~----
1 I I I I I I
I I I I I
I I I I I
Figure 1- Effect of fly ashes on the carbonation of the concrete specimens exposed to
3% C0 2 after 7 days moist curing and 84 days in laboratory air.
40
-+--- Fbint Tupper - HVFA
35 ------Thunder Bay - HVFA
__._ Sundance - HVFA
'E 30
- 0 - W/0=0.32- NPC
§. -11-- W/0=0.42 - NPC
c
0 25
iic
0 20
-e
5 15
'0
t.. 10
'tl
5
0
0 100 200 300 400 500 600 700 800
period of exposure (days)
Figure 2- Effect of fly ashes on the carbonation of the concrete specimens exposed to
3% C0 2 after 28 days moist curing and 1 day in laboratory air.
o~~~G-~-&~~~~~~~+---+---~
0 100 200 300 400 500 600 700 BOO
period of exposure (days)
Figure 3 -Effect of fly ashes on the carbonation of the concrete specimens exposed to
3% C0 2 after 91 days moist curing and 1 day in laboratory air.
20
I I I I I I I
1:
0
12
:;1:
0 10
-e
3 8
'0 6
..c:
1i
Gl
"0
4
2
0
0 2 3 4 5 6 7 8
period of indoor exposure (year&)
Figure 4- Effect of fly ashes on the carbonation of the concrete specimens exposed
indoors after 7 days moist curing and 84 days in laboratory air.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
i! 1 I I I I I I
3 8 -----~----~--~--:-----~----~----~-----'-----
1 I I I I I
"0 6
I I
----r----,---------- ---
I I
t
CD 4
1 I
"g
2
o+-a=~~~~-*--~--~~--~--~--~
0 2 3 4 5 6 7 8
period of outdoor exposure (years)
Figure 5- Effect of fly ashes on the carbonation of the concrete specimens exposed
outdoors after 7 days moist curing and 84 days in laboratory air.
20
I I I I I I I
i! 1 I I I I I I
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
I I I I I I I
3 8 - - - - T - - - - I - - - - -,- - - - -
I I I
r ----
I
T - - - - I - - - - -,- - - - -
I I I
"0 6
I I I I I I
.c ----r----,-----r----r----T----~----
..
~
"g
4
1
I
I
I
- - - - T - - - - "1-- - -
I I
I
I
-~-- -
I
2
0
0 2 3 4 5 6 7 8
period of outdoor exposure (years)
Figure 6- Effect of fly ashes on the carbonation of the concrete specimens exposed
outdoors after 28 days moist curing and 1 day in laboratory air.
..
1i
0 20
1
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
I
----r----,----,----,-----,-----r----r----
-e 1
I
I
I
I
I
I
I
I
I
I
I
I
I
5 15
____ L ____ J ____ j _ _ _ _ J _____ I _____ L_
I I I I I 1
0 I I I I I I
.c I I I I I I I
..
Q.
'0
10 ----L----~----
1
I
I
I I
-----1-----~----L----
I
I
I
I
I
I
l
I
I
5 -r----~----,----,-----~-----r----r----
1 I I I I
0~ ~:e~~~~~~~~~~~~~~~:t~~t:==~--~
0 100 200 300 400 500 600 700 BOO
period of exposure (days)
Figure 7- Effect of curing mode on the carbonation of the control portland cement
concrete (w/c=o.32) exposed to 3% C0 2 •
40
I I I I I I I
I t I I I I I
.2 25 ----r----~----~----~-----t-----~----r----
1 I I I I I I
1i I I I I I I I
c I I I I I I I
0 20 ----r----,----,----,-----,-----r----r----
f! 1
I
I
I
I
I
I
I
I
I
I
I
I
5 15
I I I
----~----~----~-----------
I I _______ L
I I
___ _
0 I I
.c I
..
Q.
'0
10
0
0 100 200 300 400 500 600 700 800
period of exposure (days)
Figure 8 ·Effect of curing mode on the carbonation of the control portland cement
concrete (w/c=o-42) exposed to 3% C0 2 •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
' I I
I I I
i' 30
I I I
g ----L----~----~--
1 I I
I
c I
0 25
i
I
0
.r.
20
15
I I
I
I
I
I
I
I
I
---~----~----~-----1-----~----
1 I
I
I
I
I
I
I
I
Figure 9- Effect of curing mode on the carbonation of the concrete incorporating 58%
(by weight of binder) Point Tupper fly ash exposed to 3% C0 2 •
40
--+--- 7d rroist curing - 84d lab air : : :
35 -e--28d rroistcuring- 1d lab air-:-----:-----~----
--tr-- 91 d rrolst curing - 1d lab air : :
i' 30 ----L----J----~----~-----1-----L ___ L ___ _
g I
I
I
I
1 I I I
I
c I I
0 25 ----t------t----
ic 1 I
I I I
'
a... 10 ----~----~----~-----1-----~----~----
1
I
I
I
I
I
I
I
I
I
I
I
"1:1 I I I I I I I
5 --~----~----~----~-----·-----~----~----
1 I I I I I
I I I I I I
I I I I I I
0
0 100 200 300 400 500 600 700 800
period of exposure {days)
Figure 10- Effect of curing mode on the carbonation of the concrete incorporating
58% (by weight of binder) Thunder Bay fly ash exposed to 3% C0 2 •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
'
I I
'
I I
0 25
ic:
0 20
€
IJ 15
0
I I I I' I
i
'
10 -~----+----~----4----~-----·-----
1 I I I
I I I I I
'CI I I I I I I I
5 - -r----T----1----,----,----~----~-----
'' '
'
'
'
Figure 11- Effect of curing mode on the carbonation of the concrete incorporating
58% (by weight of binder) Sundance fly ash exposed to 3% C0 2 •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
According to the new European standard EN 206-1 (1) and DIN 1045/1-4 (2), sufficient
concrete durability is guaranteed when the corresponding requirements are met as to:
- Examination and supervision of the concrete components
- Usage of the "description"-principle (EN 206-1) with examination and supervision o
concrete.
The usage of performance-oriented design processes (,performance"-principle), i.e. the
examination of the concrete's resistance against individual kinds of attacks, is so far only
allowed when special approvals exist. The presently standardized ways to achieve suffi-
cient durability are not always successful. On the occasion of the 15 1h ibausil 1 2003 in
Weimar was stated (3) that the number of damaged concrete buildings in Europe in-
creases worryingly. The problem of alkali silica reaction (ASR) in the concrete is not
sufficiently treated, even with the given measures in the guideline of the DAfStB 2 (4) and
the examination time of 9 months with fog chamber storage. This could be shown in
recent investigations (5) on damages that had occurred at relatively young buildings
caused by alkali-reactive slow/late aggregates, and also in earlier investigations on the
effects ofpozzolanic concrete additives (6).
When damage occurs it is difficult in most cases to retrace whether the damage occurred
although the standards and technical rules were followed or whether the damage is a
consequence of a partial non-compliance. This is especially the case when new developed
concrete components are used for the concrete production (cementitious materials,
aggregates, additives - and their combinations) and also when new kinds of concretes are
developed, descriptive concepts may fail because of lacking experiences. The resistance
test of concrete against single attacks (e.g. frost, frost-de-icing salt, sulfates, alkali silica
reaction) - as it is presently practiced - seems to be ineffective to produce reliable fore-
casts or judgments in the field of durability as the examination of specific properties that
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1
INTERNATIONAL CONFERENCE ON BUILDING MATERIALS
2
GERMAN COMMITTEE FOR REINFORCED CONCRETE
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 307
concrete corrosion as a consequence of chemical attack and initially insufficient frost-
and frost/de-icing salt resistance are not part of this investigation. On the one hand there
are effective testing methods available in the field of frost and frost/de-icing salt resis-
tance (7, 8), on the other hand outdoor elements in free weathering are exposed to com-
plex environmental influences (Fig. 2), i.e. not alone to frost or frost/de-icing salt.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
may cause swelling and expansion processes.
Water can penetrate the concrete either through an existing capillary pore system or
through existing cracks or internal flaws. This phenomenon will be designated as "pre-
damage" from hereon. Pre-damages in form of cracks and gaps around aggregates can
have numerous causes, especially high thermally or hygrically induced tensions, but also
mechanic or dynamic loads. Avoidable causes also have to be considered, especially
inappropriate concrete composition with a tendency for segregation; deformation or
movement of the forms after compaction and during the early stage of hardening; insuffi-
cient curing; premature straining; changed loading conditions and usage deviant from the
planning. In reinforced concrete constructions varying crack widths are acceptable
according to the standard ( 10). Cracks, however, even when they are not classified as
harmful by the standard can increase the permeability of the concrete. Even very fine
cracks have a capillary sucking capacity and should be considered as critical for the
durability when other conditions for harmful formation of phases in the hardened con-
crete can be found (changing moisture, reaction partners - Fig. 2). Pre-damages in ele-
ments exposed to free weathering can get worse during the service life (crack widening,
crack propagation, crack summarizing, widening of gaps around aggregates). Possible
causes are:
-deformations due to strains (static or dynamic strains)
- obstructive deformations independent of strain
-different hygric and thermal behavior of hardened cement paste and aggregates (thermal
expansion (11; 12); shrinkage and swelling deformation)
- frost- or frost/de-icing salt influence under moisture conditions
- harmful formation of phases in the hardened concrete
A worsened pre-damage increases the permeability of the concrete and encourages the
damage mechanisms such as a lower frost- or frost/de-icing salt resistance, harmful
formation of phases, reinforcement corrosion, etc. From these facts, it should be derived
that the "pre-damage effect" must be taken into account within "performance"-testing
methods.
"Performance"-testing methods to evaluate the durability of outdoor elements in free
weathering must include all essential influences, i.e. the intended concrete composition,
the processing technology, the pre-damage degree and the environmental conditions.
"Performance"-testing methods would be helpful in the context of planning, building and
usage of concrete structures:
-preliminary test phase: examination of the possible concrete durability with the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 309
intended composition, including curing by means of especially produced specimens
-quality control phase: examination of the actually achieved durability by means of
specimens that are produced during the execution, or by means of drilling cores that are
taken from the building
-structure test phase: examination of the current durability by means of drilling cores
taken from the concrete structure.
Should these examinations show hints concerning a reduced durability, appropriate
countermeasures could be taken to ensure the required service life.
Experiences with temperature and moisture changes that were carried out manually on
specimens in a conventional manner (drying, water storage, freezing, storage at standard
climate conditions) have been the subject of numerous publications on concrete durability
by the FIB (see also (14); (16) - (20)). The principal adequacy of cyclical storage
changes with alternating temperature and moisture conditions in order to achieve a time-
lapse effect in the simulation of weather conditions on concrete has been proven in
previous comparative investigations (Fig. 3).
3
Feutron Klimasimulation GmbH, Greiz
4
Materials Testing and Research Institution
5
Volume appr. 5m3, internal dimensions 1.6x2.0xl.5 m, 750 kg sample material possible
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
310 Seyfarth and Stark
sections and fractured surfaces by means of electron microscope (SEM/ESEM) and
phase-analytical investigations by means of XRD, DTA/TG, EDX are carried out. If
necessary, investigations on the carbonation progress, the change of the pore size distri-
bution (Hg-porosimetry) and on frost/frost-de-icing salt resistance (CIF- or CDF-Test)
are carried out parallel to be integrated in a complex assessment of the durability.
First results of investigations using the new climate simulation chamber
Since the climate simulation chamber became operative (Dec. 2001), different cyclic
storage programs were carefully and successfully optimized during an evaluation phase.
The cycling temperature and moisture gradients in the climate simulation chamber led -
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
as expected - to very intensive transport-, solution- and accumulation processes in the
concrete. This was also confirmed in the comparative investigations carried out during
the evaluation phase in the climate simulation chamber and in the fog chamber on the
influence of de-icing substances on ASR in the concrete. Remarkable were especially
investigations on concretes made with highly alkali-reactive (E III) aggregates (concrete
composition according to the German Alkali-Guideline (4): WIC = 0.45; 400 kglm 3 CEM
I 32.5 R- sodium equivalent 0.9 %) under usage of air entraining agents (fresh concrete
air content 5.5 %). The cyclic storage led to irreversible expansions of the specimens
after just one cycle. After approx. 60 days the limit given in the German Alkali-Guideline
(critical expansion value= 0.6 mm/m) was exceeded (Fig. 6, from (18)). The expansions
determined in the fog chamber at 313 K (40 oc I 104 °F) include the thermal expansion
of the concrete which has a magnitude of 0.1 mm/m at a temperature difference of 10 K,
that means that the critical value should be reduced from 0.6 to approx. 0.4 mm/m for
measurements at 293 K (20 oc I 68 °F). Latest results of an intensive research program
are presented in another paper from our institute (19).
In association with a research project on durability of high-performance concrete (HPC),
the first version of a program was tested which simulates the Central European climate
conditions in a time-lapsing manner (Fig. 7). Comparative investigations on the durability
of normal and high-performance concretes were carried out. A base-series compared two
high-performance concretes with basalt gravel > 5 mm but with different types of cement.
HPC I contained a CEM I 42.5 Rand HPC II a CEM I 42.5 R-HS (450 kglm3 , WIC =
0.35). The cement came from one cement works to avoid differences in the concrete
components. The investigations should clarify whether the high-performance concrete
produced with a normal portland cement with low C 3A content (2.5 %) shows a different
durability than the high-performance concrete with a normal portland cement with
normal C 3A content (10.3%). In a comparative series the influence of artificial pre-
damages on the durability of high-performance concrete was observed. Separate speci-
mens of equal composition, produced parallel, were pre-damaged after the first drying
phase on the 21st day after the casting. In order to do this, the specimens were loaded in a
compressive strength testing device (Fig. 8, Fig. 9), which can be adjusted very finely, so
that an early fracture recognition is ensured. The damage process during testing leads to
fine, hardly visible cracks and was monitored online and non-destructive.
An appropriate data recording and analyzing mode was already developed, so that the
reproducibility of the applied pre-damage procedure could be checked. On the same day,
the pre-damaged specimens (crack widths<< 0.01 mm) were stored again in the climate
simulation chamber and the fog storage cycle starts immediately.
6
KO 362/51-1; Sta 353/17-1: High performance concrete- influence of concrete composition and environ-
mental conditions on hydration and durability
7
AiF 12954 B I DBV 230: Research Program: Durability of High performance concrete for outdoor elements
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 313
OUTLOOK- FURTHER INVESTIGATIONS WITH NEW TYPE OF CLIMATE
CHAMBER
Further basic investigations will contribute to the development of a performance test for
the assessment of the durability of concretes for free weathered outdoor elements. Subse-
quent analytical examinations will be performed to reveal the cause for the insufficient
durability. Possibilities for adjusting especially the concrete composition and hardening
conditions will be checked and their effectiveness determined.
Special emphasis will be put on the cement composition to avoid a delayed ettringite
formation. Amount, reactivity and crystal structure of aluminate-containing clinker
phases as well as the proper adjustment of the sulfate content (amount, composition,
solubility, etc.) related to the properties of the aluminate-containing clinker phases are of
importance for the durability. Also the influence of the alkali content of the cement is
going to be investigated in recently started investigations. High pH-values in the pore
solution are not only a disadvantage in combination with alkali-reactive aggregates; they
also obstruct the primary ettringite formation. In conclusion, a late decrease of the alkali
content in the hardened concrete, e.g. caused by leaching or ASR could lead to harmful
DEF.
CONCLUSIONS
The state of development of performance testing method at the FIB is a result of exten-
sive research activities in the past years. For more than a decade basic research has been
carried out on very different aspects, in order to clarity the damage mechanisms and
reasons for deterioration and harmful formation of phases in the hardened concrete,
which can decrease or disable the load capacity and the protective function of the con-
crete during the service life. The developed performance test methods for assessing the
durability of concretes by using climate simulation are suitable to prevent concrete
damages in future. It shall be possible to identify concretes with insufficient durability by
comparative investigations with performance test methods. Comparative investigations
on selected specimens stored in the field and in the climate simulation chamber are
currently in progress (Fig. 29). By means of the new climate simulation chamber also a
new performance testing method for assessing the durability of concretes regarding ASR
has been developed, first results are reported in (19).
REFERENCES 8
I. DIN EN 206-1: Beton. Teil I: Festlegung, Eigenschaften, Herstellung und
Konformitlit. Deutsche Fassung EN 206-I :2000. Juli 2001 (Concrete- Part 1: Specifi-
cation, performance, and conformity;German version EN 206-1 :2000)
2. DIN 1045: Tragwerke aus Beton, Stahlbeton und Spannbeton. Ausgabe 2001-07
(Concrete, reinforced and prestressed concrete structures)
3. SchieBl, P.: Lebensdauermanagement von Betonbauwerken. 15. Internationale
Baustofftagung. 24.-27. 09.2003, Weimar. Tagungsbericht- Band 2. S. 2-1343-1362
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
8 not translated references are German publications which are not available in English
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
314 Seyfarth and Stark
4. DAfStB-Richtlinie- Vorbeugende MaBnahmen gegen schadigende Alkalireaktion im
Beton (Alkali-Richtlinie), Ausgabe 2001-05 (Preventive measures against deleterious
alkali-silica reaction in concrete - Alkali-Guideline)
5. Freyburg, E.; Berninger, A.-M.: Mikroprozesse beim Ablauf der Alkali-Kiesel-Saure-
Reaktion. 15. Intemationale Baustofftagung. 24. - 27. 09.2003, Weimar. Tagungs-
bericht- Band 2. S. 2-0719-0734
6. Siebel, E.; Sylla, H.-M.; Bokem, J.: Einfluss von puzzolanischen Betonzusatzstoffen
auf die Vermeidung einer schadigenden Alkali-Kieselsaure-Reaktion. 14. Intemation-
ale Baustofftagung, 20.-23.09.2000, Weimar. Tagungsbericht- Band 1, 2000, S. l-
0959- 1-0967
7. Setzer, M. J.: Die Mikroeislinsenpumpe - Eine neue Sicht bei Frostangriff und
Frostpriifung. 15. Intemationale Baustofftagung. 24. - 27. 09.2003, Weimar. Ta-
gungsbericht- Band 1. S. 2-1321-1336
8. Palecki, S.; Setzer, M. J.: Priifung des Frost- und Frost-Tausalzwiderstandes von
Hochleistungsbeton mittels des CDF/CIF-Tests. 15. Intemationale Baustofftagung.
24.-27. 09.2003, Weimar. Tagungsbericht- Band 1. S. 2-0799-0808
9. Stark, J.; Seyfarth, K.; Heinrich, U.: Betonpriifung in der Klimasimulationskammer.
Fachtagung betonbau.aktuell 2003. Weiterbildungszentrum Betonbau Apolda;
13.02.2003
10. DIN 1045-1: Tragwerke aus Beton, Stahlbeton und Spannbeton, Teil 1: Bemessung
und Konstruktion, Ausgabe 2001-07 (Concrete, reinforced and prestressed concrete
structures- Part 1: Design and construction)
11. Ziegelsdorf, S.; Kleiser, K.; Hilsdorf, K. H.: Vorherbestimmung und Kontrolle des
thermischen Ausdehnungskoeffizienten von Beton. DAfStB (1979), Heft 305
12. Dettling, H.: Die Warmedehnung des Zementsteins, der Gesteine und der Betone.
Technische Hochschule Stuttgart, 1962
13. Stark, J.; Wicht, B.: Zement und Kalk. Der Baustoff als Werkstoff. Basel, Boston,
Berlin. Birkhauser-Verlag 2000
14. Stark, J.; Wicht, B.: Dauerhaftigkeit von Beton. Der Baustoff als Werkstoff. Basel,
Boston, Berlin. Birkhauser-Verlag 2001
15. Reinhardt, H.W.; JooB, M.: Permeabilitat und Diffusion von Hochleistungsbeton mit
und ohne Risse zwischen 20 und 80 °C. 15. lntemationale Baustofftagung. 24. - 27.
09.2003, Weimar, Tagungsbericht- Band 1. S. 1-0001-0013
16. Bollmann, K.: Ettringitbildung in nicht warmebehandelten Betonen. Dissertation,
Bauhaus-Universitat Weimar, Fakultat Bauingenieurwesen, 2000, ISBN 3-89825-
115-2
17. Bellmann, F.; Stark, J.: Ein Beitrag zum Chemismus der Thaumasitbildung. 15.
Intemationale Baustofftagung. 24. - 27. 09.2003, Weimar. Tagungsbericht- Band 2.
s. 2-0659-0671 --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
20. Seyfarth, K.; Stark, J.: Zur Dauerhaftigkeit von Hochleistungsbetonen. In: Tagungs-
unterlagen, Bd. 2, 14. Internationale Baustofftagung. 20. - 23. September, Weimar,
2000, S. 2-1117 - 2-1132
21. CONLIFE Deliverable Report 6: "Durability data of lab-testing - public version",
available on the CONLIFE Internet homepage http://fasae.ibpmw.uni-
essen.de/euproject/, EU Research Project - 5th Framework 2004.
22. CONLIFE Deliverable Report 8: "Identification of damage mechanisms and creation
of theoretical models for failure phenomena", available on the CONLIFE Internet
homepage http://fasae.ibpmw.uni-essen.de/euproject, EU Research Project - 5th
Framework 2004.
23. Mehta, K; Monteiro, J. M.: Concrete -Microstructure, Properties and Material, Me
Graw Hill, 2nd Edition, 1993, 548pp.
..,
Drying out changing moisture Moisture penetration
solar radiation rain-/melting-/infiltrating water
high air humidity/fog
:.:T~ thaw-/condensate-/splash water
micro ice lanse pump
ascening moisture etc....
1. concrete permeability
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3 4
I -concrete 1 -concrete 2 Number of cycles
I I I I
3.5 ~--L- __ I ____ I ___ _j
I I I I
I I I I
3.0 - - - I - - -~- - - -~- - - I
t 2.5
I I
---:--------:----:
I I
.!: 2.0 I I
l
0.5
Fig. 6 --Results of first comparative investigations in the fog chamber and in the
climate simulation chamber.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~
70
60
50
40
Cyclic Climate storage· Program for HPC
molo1ullllg
• m:
---
.510
1:
! 0
·10
40
r
40
0 14 28 42
0,7
1 ...
e ...
.5
.§ 0,3
. ...
l!
~
0,1
OP~--~--~~.-~~~r-~~~--~--~
.0.1 ! !
Jf~
,..to,~./
....
A..,./.1'~,..//.""./.1'!,../.1'~~.//'1.,../.1';,""./.1'~,..//'~
'\o' '\t ":1· ..,. ... bo· ~· C)• ~· '0· '\· '\• <q· 'b· ~· 0)• ...~
0..
--:::::;--
"o--\ 1--
1120 ....
~ \ ~HPCD
·""-..
;-----
1100 \ -
i .. ~
; .. _\
\
...... _....
2d !d
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 11 --Concentration of sulfate ions in pore solution of hardened
cement paste, made from cement with different C A content
3
(calculated according to Bogue: HPC 1: 10.3 %, HPC II: 2.5 %)
Fig.14 -- HPC II FTW (pre-damaged), compact ettringite formation near micro crack.
Fig. 15 -- HPC ll FTW (pre-damaged), detail of Fig. 14: compact ettringite formation.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
T
i Sl
...
.... .... .... .... ... .... .... .... ........
K
0.40
Fig. 18 -- HPC II FTSW (pre-damaged), matrix overview: flaws filled with new
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
formed ettringite.
21100
I
I
..
.........
:I~
i
I 1~ ... ~.-:::.:.....
.....
I
···~---·
.
•
·•····
.
! • I
0 I
I"":W"......
..•. I
I
I
I
·1~
·7 0 7 14
dU!IItlon{dl
· + · OPC wlc 0.50 • • • OPC low C3A w/c 0.50 -OPCwlc0.35 --OPC lowC3Aw/c 0.35
.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
I
3500
I
..
3000
I
..- ..
.. . .
.
I
r-
s
I
I . . .....
~-
!
I
I
I
,_': ...
1!00
.~~
1000
soo
.•.. -:.,
.-~~·
.....
I
..
I
0
0 7 21 28
number of FTC
120
110
.,.._. II-14PTC I
I
14-28 FTC
100
-....... . . ":"..· . .. .
.
90
... '"'
"!'•."'-.
.. .
"'.......... ... .
I
i:70 I
to
I
"
I
150
I
40
.........
30
I .........
I
20 .........
I
10
I
0
0 7 14 21 28
number ot FTC
• • ·OPCwlc0.50 • • • OPC low C3A w/c 0.50 -OPCw/C0.35 --OPC low C3Aw/c 0.35
70~-~- -...
Cyclic Climate Storage- 101 Version of Shortened P r o g r a m .
--;;:;:c
I'
.s
.
so
30
f:• 'I-·
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-10
411
-so
211 35
~ 0.3 - 1 I ----+-
t 0.2 ·-~~ . k-=:=b-
=0.1 --1··
! ~ I
--~·------~-----
T ..
o.o~ ~1~~~-+~----~~----~----~
42 77 112 147
s.o
-r-·
r
<~.5 I --
H I
4.0
i< 3.5
~r--1
- I!----- -----1 !- - ..?
......-- -
c
i. 3.0
I
I /
-
12.5
I //
120
i //// ..,
....,
1.5
I /$' .~~ .H i H
1.0
0.5
! ff/ ~..-. .. ~Jj
0.0
v ·~
7 42 77 112 147
I ...... KBI ..... KBII ..... AB -<>-KW -e-BW I age or concntDin days
Fig. 23-- Comparison of the mass change of normal concretes during cyclic climate
storage (specimens without pre-damage).
5.0
=r-= Mass Change or Nonnal Conrot& Sped""""' with Pndamags
,
~r
~
.,.,-/
...1
----
c
;; 3.0 1··- ................. --
"'
12.5
I I
: 2.0
/
~ 1.5
I /ff
1.0
0.5
0.0
7
, v
/F
42
.~i
;r
77
J!
u~
'"
112
Jt
.
~I-->
147
., I
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
7 42 71 112 147
I .,._KBipd -KBGpd -o-ABpd -<>-KWpd ...,...sw I ageofconcretelndays
Fig. 26- ESEM/REM image, DEF in artificial air-voids of concrete KBI and cracking.
tiCI .. i
I
I
Is tzO &..!
...
I -
.9 'a..' ......
..
.. I
..
• J._..J_..J_...L_...L_..J__J
012:3 4 5614 7 1GU161t2l'Z&28313431<40
tht»ot#'zydniCfclnlnMunl tlmtofhyetr.dlonlndap
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-o-KW I
I•
16
• 0
0 f 2 3 4 SIt 4 710121fttU2128JtM3740.0
I!!MolhJdNfiOIJinhowl d!Mofbydratlonln--
Fig. 29 --Specimens stored in the field for comparison with laboratory specimens.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SP-234-21
Abstract:
In this laboratory investigation, the effects of limestone filler and granulated blast-furnace
slag (GBFS) additions on the mechanical and durable characteristics of concrete are
analyzed. The evolution of compressive and flexural strength were determined and, water
absorption and sorptivity were used to characterize the permeability of concrete. The chloride
penetration under continuous soaking was evaluated using the pounding tests and the
sulfate resistance of cement was determined using ASTM C1 012 test. The results show that,
the complementary behavior of limestone filler and GBFS additions permits to obtain
concrete with strength development similar to portland cement with 35 % less clinker, and
the incorporation of GBFS into the mixtures prevents and improves the inadequate
performance of limestone filler cement in chloride and sulfate environments. Consequently,
concretes made with ternary cement offer economic and ecological benefits, with similar
strength evolution and similar or better durable properties compared with binary and plain
concretes. However, to assure the reduction of permeability of concrete an adequate cured
time should be proportionate to assure the hydration progress.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
emissions per ton of cement produced. During the last decade, the production of blended
cements made with portland cement and two mineral admixtures, also called ternary or
composite blended cements, has increased because they present a better performance than
the those of binary blended cements. Ternary blended cements containing the combination
of fly ash-slag; fly ash-silica fume or slag-silica fume are commonly used and several studies
have been published (1-6).
This trend in cement production was leadership by European Standard (EN 197) and it can
also be observed in Latin-American countries (Brazil EB-2138/91, Mexico NMX C-414-0/99,
Argentine IRAM 50000/00). In Argentine, this cement can contain two or more additions
(slag, pozzolan and calcareous materials) ranged from 0 to 35 %. In USA, ASTM C 1157
standard has introduced performance-based hydraulic cements that does not limit the type
and the amount of mineral admixtures that can be blended with portland cement. Usually,
both mineral admixtures used may present a complementary effect on cement hydration.
Limestone filler addition produces several effects on the cement paste, which are arbitrarily
divided in chemical and physical. The chemical effects are related to the interference of
carbonate ions during the hydration of C.Y:... phase leading to the calcium carboaluminates
formation and delaying the ettringite-monosulfoaluminate transformation (7). The physical
effects (dispersion, nucleation, filling, dilution) caused by limestone filler enhance the
strength due to hydration acceleration of portland clinker grains at very early age (8, 9) and
the improvement of particle packing of the cementitious system (10). These effects can
compensate the reduction of clinker active grains produced by limestone filler replacement.
There is general agreement that the principal hydration product formed when GBFS is mixed
with portland cement and water is essentially CSH similar to the compound produced by the
hydration of calcium silicates of portland cement ( 11 ). The rate of hydration of GBFS is
initially lower than the corresponding to portland cement. Hence, portland cement containing
GBFS typically shows a reduction of strength at early age (7 to 20 days) and similar or
greater strength at later ages (12). Mostly, GBFS reduces permeability and ionic diffusion of
chloride regardless of composition and replacement level (13).
made with composite cements, it is very difficult to assure the durability limiting only the
w/cm, because the clinker composition, the type and proportion of mineral admixtures are
generally unknown in these cements. Besides, the drying shrinkage, the porosity and pore
network are modified by the combined use of supplementary materials.
The focus of this paper is on the determination of the role and effectiveness of combinations
of limestone filler and GBFS to produce concrete with similar strength development to plain
concrete, comparable strain due to drying shrinkage, and similar or better permeability that
provide a good resistance to aggressive environments. The evolution of hydration process is
also studied to justify the behavior of concrete and to determine the initial water curing that
should be taken into consideration for practical applications.
This experimental investigation was carried out using a normal portland cement (PC) and two
limestone portland cements (PC12L and PC18L) obtained from the same portland clinker by
an intergrinding process in the cement plant. According the data supplied by cement
producer, the potential composition of portland cement was C3S = 58 %, C2S = 18 %,
C~ =2 % and C4 AF =13 %, and the limestone content by mass was 0, 12 and 18% for PC,
PC12L and PC18L cement, respectively. They have the same strength class (CP40, f"c > 40
MPa at 28 days tested on ISO-RILEM prisms) and for this purpose a large specific surface of
limestone portland cement was required (321, 380 and 383m2/kg for PC, PC12L and PC18L
cement, respectively). The chemical composition and physical characteristics of cements
used are given in Table 1.
A ground granulated blast furnace slag (GBFS) was used for preparing binary and ternary
cements with PC and limestone portland cements. It had a chemical modulus (C+M+A/S) of
1.8 and it was classified as "high-activity GBFS" according to the EN 196-1 standard and it is
equivalent to Grade 100 GBFS in ASTM C 989. Its chemical composition and physical
characteristics are also given in Table 1.
Table 2 shows the composition of composite cements used for each test program.
The coarse aggregate was crushed granite from a very sound granite rock with a maximum
size of 19 mm (compressive and flexural strength, absorption and sorptivity test) and 12.5
mm (drying shrinkage and chloride penetration test). The fine aggregate was natural
siliceous river sand having a fineness modulus of 2.35.
All concrete mixtures were designed with a wlcm of 0.50 and a cementitious material content
of 350 ± 10 kg/m 3 and the ratio between weight of sand and the total weight of aggregate
was 0.45. Concrete mixtures have adequate workability, a slump of 70 ± 20 mm, good
finished, cohesiveness and moderate rate of bleeding.
TEST PROCEDURES
Concrete specimens were prepared in accordance with ASTM C 192. After casting, molds
containing the specimens were covered with plastic sheets and left in the laboratory
environment for 24 hours. Subsequently, specimens were demolded and cured in lime
saturated water until the time of test was reached.
Absorption and water sorptivity: The water absorption and the sorptivity test (15-17) were
determined on remaining fragments of specimen previously broken by flexural test. For each
concrete, three fragments of 100 x 150 x 120 mm in size were cut using a diamond saw and
then they were dried in oven at 105 ± 5°C for 24 hours. The water absorption was
determined in accordance with ASTM C 642.
In the sorptivity test, after temperature stabilization, specimen surfaces were coated with a
bituminous-paint layer limiting the absorption surface (100 x 100 mm) to the cast face. This
was done to prevent the surface absorption in the axial direction. Subsequently, the
specimens were exposed to water on one of plane ends by placing them in a plastic tank.
The fluid level was maintained at constant level throughout the experiment such that
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Non evaporable water: At each test-age, fragments of concrete specimen were used to
determine the amount of non-evaporable water according to the procedure proposed by
Powers (18). This value was used as a means to estimate the progress of hydration
reactions assuming that the filler is a chemically non-active constituent and all grains of
GGBFS are reactive. Results provided by this method are not accurate, but they are
sufficiently indicative of hydration reaction progress. The relative combined water was
calculated as the ratio between combined water in the blended cement and the combined
water in portland cement at the same test age.
Drying shrinkage: The length changes due to drying were measured on 100 x 100 x 400
mm concrete specimens according to ASTM C 157 (IRAM 1597). After remove from the
molds, they were cured in lime saturated water up to seven days age, and then specimens
were stored in drying cabinet at 20 ± 2 °C and relative humidity of 50 ± 5 %. The strain was
measured using a mechanical extensometer with a measuring length of 300 mm and a
precision of 1 1-Jm. Strain reading were made each two days up to 14 days, each week up to
seven weeks, and each two week up to end of test. Also, spacemen were weighed and the
water loss was determined as the difference between initial and actual weight.
Chloride profile: In this test, concrete specimens (100 x 150 x 530 mm) were cast and
cured 24 hours in the molds. After demolded, specimens were cured in wet regimen (6 days
in lime saturated water and 21 days in laboratory air) and in air regimen (27 days in
laboratory air). At 28 days, the surfaces of the specimens were epoxy-coated with exception
of one surface perpendicular to molding and then immersed in water during 24 hours for
saturation of specimens. Thereafter, the specimens were exposed to a 3% NaCI solution.
After 180 and 360 days of immersion, a 70 mm thick slice was sawed from the top of each
prism and the specimens were newly epoxy-coated and immersed in solution again for future
determination. Using a driller device, a powder sample was obtained each 5 mm depth and
then the total chloride content was determined and the percentage total chloride content with
respect to concrete mass was calculated.
Sulfate resistance: For each cement, mortar specimens (285 x 25 x 25 mm) were cast
according to ASTM C 1012 (sand-to-cementitious material ratio of 2.75 and w/cm of 0.485).
Mortar specimens were stored in moist cabinet during 24 hours, and then they were removed
from the mold and cured in saturated lime-water solution until 28 days, instead of curing
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
RESULTS
Flexural strength: Fig. 2 shows the evolution of the flexural strength. For PC12L and PC18L
concretes, it was lower than the corresponding plain concrete after seven days and the
strength loss was 9 and 4 % at 360 days, respectively. On the contrary, PC20S concrete
showed a flexural strength similar or higher than those of PC concrete at all tested ages and
it had an increase of 10 % at 360 days (Fig. 2a). For ternary concretes containing PC12L
cement and GBFS, it can be observed that the flexural strength increases when the GBFS
content increases. For PC12L20S concrete, it was 14 % higher at one year (Fig. 2b); while
flexural strength of PC18L20S was lower at 7 days and then it increased up to 12 %at one
year.
Non-evaporable water: The results of relative non-evaporable water (Wne) are illustrated in
Fig. 3. For PC12L and PC18L concretes (Fig. 3a), Wne increases when the filler content
increases especially at early age. At 28 days, Wne was 1.16 and 1.24 for PC12L and PC18L
concretes, respectively. At later ages, the progress of portland cement hydration reduces this
value (Wne trend to 1.0). For PC20S concrete, the combined water was lower than that
corresponding to PC concrete up to 90 days of curing and the Wne was 1.02 at 360 days. For
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 333
PC12L 1OS and PC18L20S concretes (Fig. 3b ), the Wne was greater than 1.00 at all studied
ages showing the highest value at 7 days. However, a reduction of Wne was found up to 3
days in PC12L20S concrete. At 360 days, the Wne is around 1.05 for ternary blended
cements.
Water absorption: Fig. 4 reports the 48 hours water absorption in concretes. These values
were in the range of 5.6 to 2.0 % and 5.8 to 2.0 % for binary and ternary concretes,
respectively. The water absorption of PC 12L concrete is lower than that of PC-concrete at all
ages (Fig. 4a), while this property was higher in PC18L concrete after 28 days. On the other
hand, PC20S concrete presented high water absorption up to seven days, but it showed the
lowest absorption for binary concrete at one-year water curing. Compared with water
absorption of plain concrete, this property of ternary concretes was similar until 28 days of
curing. However, it decreases at 90 days and remains constant up to one year (Fig 4b).
Sorptivity coefficient: The values of the sorptivity coefficient (S) are reported in Table 3.
For PC12L concrete, this coefficient was lower than that corresponding to PC-concrete at all
ages studied. However, the S-coefficient was higher (15 %) in concrete with a large
proportion of filler (PC18L), indicating its increased vulnerability. For PC20S concrete, the S-
coefficient is high up to seven days and then it presented a low coefficient after 28 days. The
addition of GBFS to PC12L cement produces an increase in the sorptivity coefficient until 7
days. Subsequently, this coefficient had a similar value, independently of the replacement of
GBFS used. However, the addition of GBFS to PC18L cement reduced considerably the S-
coefficient at later ages, enhancing its potential durability. The reduction of S-coefficient for
ternary cements was 24 to 42 % at 360 days.
Drying shrinkage: Fig. 5 shows the evolution of strain caused by drying shrinkage up to
300 days. It can be observed that a large proportion of shrinkage C 55%) occurs up to 28
days of drying. At the end of test, concrete containing mineral admixtures showed a final
strain of 0.92 to 1.33 times of the strain measured for PC-concrete (Table 4 ). In general
terms, final strain increases when increase the proportion of GBFS in the cement, and it
decrease when increase the proportion of limestone filler. A similar tendency was observed
in the water loss of specimens. After five days of drying, the water loss of specimens is more
than one half of water loss determined at end of test, while the final water-content in
specimens was 23 to 27% lower than the initial calculated value.
Chloride penetration: Fig. 6 shows the chloride profile obtained after 180 and 360 days of
specimens air (AC) and moist curing (MC). For all mixtures studied, the penetration of
chloride was deeper when the moist curing was reduced. For example, penetration of
chloride front in PC-concrete at 180 days (Fig. 6a) was 35 and 45 mm for moist and air
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Sulfate resistance: Data on expansion of mortar prisms in sulfate solution are shown in Fig.
7. For cement containing limestone, data indicate that PC12L mortar has the lowest
expansion during the test, while PC18L mortar has a higher expansion during duration of
test, instead of the low C~ content in this cement. Fig. 7b shows the expansion of mortar
containing 20% of GBFS and a similar evolution could be observed for these mortars after
two years in sulfate solution.
At 6 and 12 months, all cements had expansion lower than 0.05 % and 0.10 %, respectively,
and could be considered as "high sulfate-resistant". Beyond one year, PC18L showed an
increase of expansion and it has a expansion of 0.22% at two years; while expansion of
PC12L and PC cements was lower at all exposure time. After one year, it is noted that
PC12L expanded at low rate and there was an increase of expansion rate in PC mortar
beyond one year. For these cements, the addition of 20 % GBFS does not produce
detrimental effects on the sulfate performance of PC or PC12L and it produce a slight
reduction of later expansion on PC18L cement.
DISCUSSION
The results obtained in this research can be justified analyzing the kinetics of hydration and
the changes of macroporosity that cause each mineral admixture or their combinations.
Analyzing the results of Wne (Fig. 3), it can be observed that both concretes containing
limestone filler (PC12L and PC18L) showed an acceleration of hydration at early ages.
Therefore, the progress of hydration was little from 7 to 28 days, while the PC concrete
showed an increasing rate of hydration degree until 28 days. This increase could also be
attributed to the interaction between limestone filler with portland cement grains, and the
replacement of coarse grains by smaller grains of clinker fraction in portland limestone
cements (9).
However, the acceleration of hydration cannot compensate the reduction of active cementing
material produced by limestone addition, which affects the strength of concrete at later ages.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
In PC20S concrete, the GBFS does not react at early ages and the Wne was less than 1.00.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Consequently, the dilution effect of clinker grains is preponderant and PC20S concrete
presented a low compressive strength until 28 days. Thereafter, the GBFS reacts slowly
forming CSH and causing a Wne higher than 1.00 at 90 days. For a given portland cement,
the age at which the GBFS contributes to hydration depends on the particle size and its
reactivity. GBFS used in this work has only 7% of particles higher than 45 11m and 60% of
particles higher than 10 !Jm. For this particle size, the contribution to strength could be
expected after 7 days (21 ). Consequently, at early age the strength decreases by the dilution
effect. Low percentages of GBFS also produce a positive filler effect (22). However, the
dilution effect appears to be more important because the GBFS yet to react producing less
hydration products and consequently the Wne decrease. Then, this behavior is reverting when
the GBFS hydration progresses.
Summarized, filler effect appears fundamentally during the first days of hydration as
indicated by the Wne (> 1.00) and the contribution of GBFS to the cementing compounds
appears after 7 days of hydration. The high hydration degree caused by limestone filler can
only compensate some part of the dilution effect, waiting for the GBFS contribution at later
ages.
For mixtures with the same wlcm, this evolution of hydration implies that limestone filler
addition increases the strength of concrete at early age, while the GBFS contributes to
strength at later ages. For the ternary blended, the complementary effect of additions
produces a concrete with similar development of the compressive strength to PC-concrete.
On the other hand, the improvement of paste-aggregate interface that produces GBFS
addition causes an increase of flexural strength after seven days for binary and ternary
blended.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
336 lrassar et at.
Data on water absorption and water sorptivity coefficient indicate that these water transport
parameters decrease continuously with the progress of hydration up to 28 days. However,
the permeability of concrete can be reduced when the pore network becomes discontinous.
According to Powers eta/. (24), the time at which it occurs depend on the w/c used and the
degree of hydration developed. In the percolation theory proposed by Bentz and Garboczi
(25), the pores became discontinuous when the capillary porosity is 18 % or less, regardless
of wlc and the hydration degree. The experimental results obtained by Cook and Hover (26
using the MIP technique show that for plain cement paste the pore became discontinues at
around 65 % of hydration degree for a w/c of 0.50, and this value is around 74 %for a w/c of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.60.
Using the data on Wna. the hydration degree (a) can be obtained assuming the water needed
for full hydration is 0.19. Thereafter, the time estimated for the pore segmentation in plain
concrete is 28 days; while this time is around 14 days for PC12L and PC18L concretes.
Complementary, the presence of filler grains disseminated within paste can modify the
connectivity and tortuosity of the capillary network, increase the total porosity and produce a
marked shift in large pores of cement paste containing limestone filler (27). This increase of
early hydration of limestone filler cement lead to concrete less sensitive to a lack of moist
curing (20). Generally, the increase of the proportion of limestone filler in binary and ternary
cements produces an increase in the absorption parameters, consequently, is necessary to
restrict the limestone filler content in order to obtain durable concrete (15). Hence, the w/cm
and cement content proposed by the concrete codes may be insufficient for concretes with
large proportion of limestone filler.
Hence, PC12L mixture compared with PC concrete presented a low absorption and
S-coefficient at all ages and, it is attributed to the increase in hydration degree of
cementitious material that compensates the increase of effective wlc produced by limestone
filler addition. However, the increase of effective wlc is preponderant in PC18L concrete and
then its transport parameters are higher than the corresponding to PC-concrete at all ages.
This can be attributed to the large pore size owing to the decrease in the volume of
cementing materials through the effect of dilution (28). For this reason, it is recommended to
limit the content of limestone filler to reduce the permeability of concrete.
All water permeability parameters of PC20S concrete were higher than that of PC concrete
up to 28 days of curing, and later they presented lower values. Therefore, in concrete
containing GBFS these parameters are only reduced for well cured concrete (29). At 28
days, the progress of hydration reaction of GBFS produces a decrease of mean radius of
For ternary blended, the water permeability parameters were higher than those of PC-
concrete up to 7 days of curing, later, at 28 days they were decreased and remains until 360
days. After 7 days of water curing, this behavior could be attributed to the contribution of
GBFS reaction to pore size refinement and blocking the capillary network (30). As a
consequence, it will be necessary to increase over 7 days the time of moist curing in order to
improve the durability concrete (29). For these water permeability parameters, the
performance of combination of limestone filler and GBFS is complementary, too.
Regarding the drying shrinkage strain in concretes (Fig. 5), there is an increase of initial
strain in concrete containing mineral admixtures, which can be attributed to the filler effect,
especially to increase of hydration rate of clinker at early ages (see Table 4). The loss of
water contained in CSH produce a high drying strain that the loss of water contained in large
capillary pores, therefore the increase of hydration degree at early age may produce an
increase of shrinkage in concrete containing mineral admixtures. This acceleration of
hydration also produce a reduction a water loss for concrete containing limestone filler due to
an increase of chemical combined water and the increase of tortuosity of large capillary
pores (27). The dilution produce two opposite effects: the effective w/c increases and the
filler particles increase the restriction to paste deformation. Assuming that limestone filler is a
non-hydraulic addition, and the GBFS used reacts after 7 days, the effective w/c of concrete
containing mineral admixture increases at the early stages of drying causing more
deformable and permeable paste. Consequently, the drying shrinkage increases for mixture
containing blended cement. On the other hand, Powers (21 ), evaluating the factors that
affect the drying shrinkage, suggest that unhydrated cement grains could be considerer as
aggregates that restricting the shrinkage of paste. In the same way, slag and filler grains
were also computed as restriction for shrinkage of blended paste at early ages.
According the results of drying shrinkage, the filler effect and the increase of effective w/c
prevails on the restriction caused by filler and GBFS grains and, consequently the initial
shrinkage increase for concretes containing mineral admixtures. At later ages, the
importance of filler effect decays and the dilution effect and the GBFS reaction govern the
shrinkage. Hence, the shrinkage decrease when the content of filler increase in the mixture
due to the small volume of CSH and the large proportion of restrictive grains in the paste;
Chloride profiles (Fig. 7) show a practical application of the hydration and porosity evolution
for this mixtures. The increase of effective wlc in PC 18L concrete causes a deeper chloride
penetration for moist or air cured specimens, while the well hydrated GBFS concrete
(PC+30S) produce a less penetration due to a dense structure of paste, the reduction in the
pore size and the improvement of paste-aggregate interface. However, a poor curing of this
cement cause a large chloride penetration due to an increase of porosity and pore size (31 ).
Finally, the ternary blended (PC12L +20S} lead to a reduction of chloride penetration depth at
large time (360 days).
At large ages, the same behavior is observed for mortar exposed to sulfate solution. The
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
increase of effective w/c in cement containing large proportions of limestone filler cause a
more permeable mortar that may result in deeper penetration of sulfate ions and large
expansion. This expansion occurs instead of the low C3A of portland cement, because the
limestone filler did not impedes de gypsum formation due to the reaction between sulfate
ions and CH into the paste (32). According the recommendation (33), the combination of
portland cement with 40 and 70 % of low AI20 3 slag provides a good sulfate resistance to
concrete, and low GBFS proportions can reduce the sulfate resistance of cement. The low
C3 A cement with 20% of GBFS (PC+20S) provides a good sulfate resistance, reduces the
expansion of PC18L cement, and also maintains the high sulfate resistance of PC12L
cement. This behavior is related to the low permeability and the reduction of CH in the mortar
that cause the addition of GBFS in well cured mortar.
Based on the results obtained from this materials and proportions, the main conclusions from
this study are:
• The use of limestone filler in the ternary blended increase the early hydration of
portland clinker; while the GBFS contributes to the later hydration in this cement. This
complementary behavior of limestone filler and GBFS addition permits to obtain
concrete with similar strength development to normal portland cement with 34 % less
clinker.
• Concretes with ternary cements have less absorption and permeability by sorption
when the limestone filler content is limited (no more than 12%) and well curing
practices are used.
• For the same concrete proportions, the final drying shrinkage is lower when increase
the limestone filler content in ternary cement, while it increase when large proportions
of GBFS were used to formulate the ternary cement.
• Ternary cements containing a limited proportion of limestone filler (no more than
12%) and 20 to 30 % of GBFS provide a good resistance to chloride ingress and
good performance in sulfate environment of low C~ Portland cement.
REFERENCES
(1) Gopalan, M.K. "Sorptivity of fly ash concretes". Cement and Concrete Research, V. 26, N° 8, 1996,
pp. 1189-1197.
(2) Jianyong, L.; Pei. T. "Effect of slag and silica fume on mechanical properties of high strength
concrete". Cement and Concrete Research, V. 27, N° 6, 1997, pp. 833-837.
(3) Bagel, L. "Strength and pore structure of ternary blended cement mortars containing blast furnace
slag and silica fume". Cement and Concrete Research, V. 28, N° 7, 1998, pp. 1011-1020.
(4) Ozyildirim, C. "Laboratory investigation of low-permeability concretes containing slag and silica
fume". ACI Materials Journal, V. 92, N° 2, 1994, pp. 197-202.
(5) Thomas, M.D.A.; Shehata, M.H.; Shashiprakash, S.G.; Hopkins, D.S.; Gail, K. "Use of ternary
cementitious systems containing silica fume and fly ash in concrete". Cement and Concrete
Research, V. 29, N° 8, 1999, pp. 1207-1214.
(6) Ozyildirim, C.; Halstead. W.J. "Improved concrete quality with combinations of fly ash and silica
fume". ACI Materials Journal, V. 91, N° 6, 1994, pp. 587-594.
(7) Bonavetti, V.L.; Rahhal, V.; lrassar, E.F. "Studies on the carboaluminate formation in limestone
filler blend cements". Cement and Concrete Research, V. 31, N° 6, 2001, pp. 883-859.
(8) Soroka, 1.; Stern, N. "Calcareous fillers and the compressive strengh of portland cement". Cement
and Concrete Research. V. 6, N° 3, 1976, pp. 367-376.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Table 1: Chemical composition and physical characteristics of cements and GBFS used.
Portland Portland
Portland
limestone limestone Blast·fumace
cement
cement cement slag
PC
PC12L PC18L
Chemical composition, %
Silicon dioxide {Si02) 21.44 19.96 19.75 34.27
Aluminum oxide {AI20a) 3.40 3.28 3.14 12.68
Ferric oxide {Fe20a) 4.20 3.91 3.35 0.84
Calcium oxide {CaO) 63.45 61.77 61.42 40.58
Magnesium oxide {MgO) 0.57 0.56 0.57 9.75
Potassium oxide {1<:!0) 1.18 1.01 0.92 0.41
Sodium oxide {Na20) 0.04 0.02 0.04 0.05
Sulfur trioxide {SOa) 2.91 2.77 2.75 0.41
Loss on ignition 1.82 5.86 7.25 0.11
Physical characteristics
Specific gravity 3.15 3.09 3.06 2.80
Fineness {Blaine), m2/kg 321 380 383 459
Retained on sieve,%
75 iJm (#200) 3.9 3.8 9.7 0
45 1-1m {#325) 16.4 14.5 25.7 6.7
Position parameter x, 1-1m * 28.81 22.66 31.13 18.87
Homogeneity parameter, n ** 0.93 0.95 0.86 1.05
• Characteristic diameter of particle size distribution obtained at cumulative mass of 63.2 %.
.. Slope of particle size distribution curve representing the wide of distribution.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Initial drying
Drying strain Water loss
Mixture Hydration
Effective w/c 121 at 300 days, 1Jm/m at 300 days, %131
degree, %111
PC 0.61 0.50 493 25.7
PC+35S 0.89 0.77 654 26.8
PC+ 15L 0.66 0.59 456 24.9
PC+6L+ 225 0.86 0.69 623 24.0
PC+11L+ 115 0.76 0.64 477 23.0
PC+22L+65 0.81 0.69 471 22.9
(1) Calculated according to the model proposed by Powers.
(2) The limestone filler was discounted as oementitious material.
(3) Mass variation between the weight of specimen and the initial water oontent in the mixture.
tc 35
~
____.
___.,
----o
!II
~ 40
tc 35
=
----,6
~
<1>
'li: 30 30
·~
<1>
·~ 25
-+-PC </)
25
~ 20 -o-PC12L ~ 20 --o-PC12L10S
E --<>-PC18L E -<>-PC12L20S
0 0
(.) 15 (.) 15
-h-PC20S -h-PC18L20S
10 10
37 28 90 360 37 28 90 360
Age, days Age, days
-----<>
------.!>.
__.,
----11
!II
c..
::!!: 5
£Cl
--
==ll
~
!II
~5
c,
c c
<1>
~4 g4
</)
~ -+-PC ~ -+-PC
~
~
16 -o-PC12L X --o- PC12L105
u: 3 -<>-PC18L
..Q/3
u.. -<>- PC12L20S
-h-PC20S -{¥- PC 18L20S
37 28 90 360 37 28 90 360
Age, days Age, days
Fig. 2: Flexural strength of concrete with (a) binary cement and (b) ternary cement.
1.3-r-------------, 1.3 - r - - - - - - - - - - - - - - - - ,
~
~ 1.2 PC18L20S
<1>~
:c
~1.1 PC12L10S
Q. ~~
~ 1.0 ~ ____fQ_
g JPC12L20S
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~ 0.9
~
~ 0.8
0.7 _.,_-..,.......----.,.---11,-_,J A
2 a - - - - - 9....o_ _,~ 1v-
o. 7 ...,3,..7-... 1 ,--.,..J360
37 28 90 360
Age, days Age, days
Fig. 3: Relative non-evaporable water of concrete with (a) binary cement and
(b) ternary cement.
2 --
37 28 90 360 37 28 90 360
Age, days Age, days
Fig. 4: Water absorption (ASTM C 642) of concrete with (a) binary cement and
(b) ternary cement.
0~----------------------~
E -100 --PC
...,.... PC+15L
"E~ -200 -- PC+35S
<li
g> -300
-'£
·§ -4oo
-55
g> -500
·~
Cl -600
-700 L..__ _.___ _...___ _.....__ _.....__ _....__ _.....J
0 50 100 150 200 250 300
Drying time, days
0~----------------------~
.€ -100 --PC
...._ PC+6L+22S
~-200 _.,_ PC+11 L+11S
_._ PC+22L +65
&-300
.ll!
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
.§ -4oo
J::
~-500
c
·~-600
Cl
-700 L..__ _.___ _...___ _.....__ _....__ _....__ _....&.J
Fig. s: Drying shrinkage of concrete with (a) binary cement and (b) ternary cement.
8"
c:
3
~E
;:, 2
(.)
"'
"'"'. 1
5
.l!l
b4 ,--a..
PC12L+ 20S
~ 4
---
---:a~, 360 d -1 d AC
PC+ 30S
'~
8
c:
'',<'','Q 360 d -1d AC "
c:
8 ""
~E
3
'\ '\ ~E
3
'ft~,
:- \b.\ 180d-1dAC :- 2 '\,180 d -1dAC
(3 2
', (3
\'a.,
"' "'
',<::~:::..., "'
.:.:.
.• 1 .:.:. 1 b ',,
(3 (3
' ......:a_,
00 00 2 3 4 5 6
2
Depth, em Depth, em
Fig. 6: Chloride penetration profile for moist cured (MC) and air cured (AC) specimens
at 180 and 360 days: a) PC cement; b) PC18L cement, c) PC12L+20S cement and
d) PC+30S cement.
0.30 0.30
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Concrete" (1) published by the Architectural Institute of Japan (AIJ) in March 2004 is regarded
Concrete" (2) published similarly by AIJ in 1991. The new recommendations have been
worked out since 1999 primarily aiming for presenting a more advanced durability design
method.
economic scales, and service lives of buildings widely vary, all-inclusive specifications may not
performance-based specification has therefore arisen, leading to various steps in this direction,
such as the concept of performance-based design incorporated in the "Building Standard Law"
performance-based design of buildings in particular was proposed for the first time in the
"Development of Technology for Improvement ofBuilding Durability (1980 to 1985)" project --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
promoted by the Ministry of Construction, which was summarized as "Guideline for Design
and Practice for Durability of Reinforced Concrete Buildings" (3) in 1986. This work, setting
out the first guidelines for durability design of reinforced concrete structures in Japan, was
introduced to the world by Tomosawa et al. (4) and served as a precursor and model of similar
However, it is still difficult to determine the specifications for every single segment
of buildings by the performance-based method, while most building construction is carried out
practical details of guidelines. It was therefore decided to present both the performance-based
decide between them at their discretion depending on the situation. These recommendations
design," while defining "performance-based design - special approach" as a framework for the
case where either of the first two approaches leads to unreasonable results. These are regarded
performance requirements represented by the NKB Five-level Structure (Fig. 1) and follow the
design specifications including the concrete proportioning, bar arrangement, and placement
conditions are optimally determined to meet the requirements for the "service life" of the
building or its segments and the "performance" of the structure and members under the
approach" was established for the case where no models or verification methods specified in
these recommendations are used or the case where the specifications are to be determined by
designers select from presented standard specifications according to the service life class and
design - general approach" by the absence of awareness about the performance resulting from
the selected specifications. This does not mean that the specifications established as standards
will not assure the required performance. This method does not widely differ from
perfonnance-based design at the current level of technology, excepting the lower degree of
freedom of design, since the specifications are established based on sufficient theoretical
general design practice, a performance-based theoretical system for unreasonable cases, and a
verification method for ensuring durability in the case of limited technical information, such as --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
new technology. A more universal system of durability design was thus intended.
Fig. 3. By this method, a designer basically establishes the design service life to be aimed for,
design limit states, and design external deteriorative forces and verifies if these requirements
and conditions meet the assumed deterioration phenomena. The outline and procedure are
In contrast to the semi-permanent service life expected of civil structures, the service
life of a building that is a disposable asset of its owners is basically determined by their free will
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
and yet simultaneously makes up a part of the social capital and so has a public nature. The
designer should therefore determine the design service life appropriately in consideration of the
possible social impact of the building, such as its uses, scale, and importance. The present
(1)
performances serve as the basis for determination of the limit states of the structure and
members. The necessity of the concept of performance-based design has been discussed since
after World War II, resulting in a variety of standards for the classification, items, and definitions
Restombility (repairability)
Economy
However, it is pmctically difficult to carry out durability design for the purpose of
assuring the required performance throughout the design service life in regard to some of the
properties, such as aesthetics, heat insulation, and sound insulation, whose relationship with the
deteriomtion of concrete is not clear or is hard to establish based on technical data. Also, such
properties as fire resistance, heat insulation, and sound insulation do not affect the performance
of the structure or members until they are substantially lowered. There is no point in durability
design for the purpose of preserving these properties, as the lowering of these performances will
For actual durability design from the aspect of performances, the limit states of the
structure or members should then be defined from their relationship with the performances.
Also, the limit states for each performance ought to be affected by the maintenance activities
during the service life in addition to the environmental conditions of structures. Moreover, the
normal design method can be unreasonable in a number of cases. These include the case
economical to actively incorporate the effect of finishing materials. The design limit states and
maintenance limit states were therefore established according to the level of maintenance. The
performance indices which can be quantitatively evaluated and the limit states were established
for each performance as shown in Table 2. The relationship between the limit states and the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The method of establishing the performance items to be considered for structures
and members from the aspect of performance requirements for the buildings has been discussed
so far. Specific design and maintenance limit states are as a rule determined for the structure
and members as shown in Fig. 6 (a). However, the design limit states and maintenance limit
states are determined for the deterioration of component materials of the structure and members,
such as reinforcement, concrete, and finishing materials, as shown in Fig. 6 (b) in the following
cases: when techniques and data are not yet available for quantitative evaluation of the effect of
the deterioration of component materials on the performance of the structure and members; and
when the deterioration of component materials is deemed directly decisive for the service life of
the structure and members. It is therefore important in durability design to define how
Reinforced Concrete Structures - Based on Allowable Stress Concept -" published by AIJ
provides equations for calculating the allowable bending moment of beams with a rectangular
cross section, allowable axial force and allowable bending moment of columns subjected to
horizontal shearing force on shear walls, and required bond length and anchorage length of
reinforcement and the compressive strength and cross-sectional area of concrete are
incorporated as parameters in any of the equations for allowable axial force, allowable bending
moment, and allowable shearing force, explicitly or implicitly, regardless of the member type.
A method of calculating the bending strength of reinforced concrete members with corroded
reinforcement has also been investigated (6) as shown in Fig. 7. If the corrosion amount of
reinforcement and deterioration state of concrete (cracking, profile losses) at the time of the
design limit state of the structure and members can be calculated using the equations provided
by standards and guidelines or finite element analysis, then durability design can be carried out
by setting such states as the targets. On the other hand, the maintenance limit state for
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
structural safety is defined as a state in which the axial strength, flexural strength, and shear
strength of the structure and members are lowered by reinforcement corrosion or concrete
cover concrete or when repeated freezing and thawing cause cover concrete of the structure and
Methods of determining the design limit states for safety of use, leakage, and
(2) concrete deterioration due to cracking, delamination, spalling, surface deterioration, strength
losses induced by frost damage, alkali-aggregate reaction, chemical erosion, and other causes;
and (3) other deterioration phenomena. External deteriorative forces causing carbonation,
chloride attack, alkali-aggregate reaction, and chemical erosion and other external deteriorative
forces were then established as the design external deteriorative forces, and various external
Table 4 gives the deteriorative design forces for each deterioration phenomenon.
The deteriorative design forces are required to be determined based on continuous field
The deteriorative design forces for carbonation, i.e., the carbon dioxide
concentration, sulfur oxide concentration, nitrogen oxide concentration, and carbonic acid
concentration of soil and water in contact with the building, are known to widely vary
depending on many factors, such as whether the measuring point is on the inside or outside of a
building, in an urban or rural area, near a factory or road. It is also pointed out that these
concentrations tend to increase every year. According to indoor measurement data, the carbon
and 0.20% were therefore adopted for outdoor and indoor standards, respectively.
The deteriorative design forces for chloride attack deal with external chloride ions,
such as those conveyed by seawater droplets and sea salt particles and those in water in contact
with the building. The estimated airborne chloride contents at Om from the shoreline were
rearranged and set as a standard based on the relationship between the sea salt particle
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The deteriorative design forces for frost damage cover the annual number of
freezing and thawing cycles, minimum air temperature, and moisture acting on concrete. In
the verification process of performance-based design, however, these are evaluated in terms of
the ASTM-equivalent number of freezing and thawing cycles. This is the annual
ASTM C 666 in consideration of the annual number of freezing and thawing cycles and annual
(1) Carbonation
reinforcement fluctuates due to differences in the placement accuracy, and carbonation depth
does not proceed uniformly due to fluctuation in the concrete qualities. For this reason, a new
durability design technique was proposed in the present recommendations, by introducing the
technique of reliability design to incorporate stochastic expressions of the limit states. The
reliability design technique in regard to carbonation has already been adopted as one of the
5) and is deemed essential for formulating a more theoretical method of durability design.
Specifically, the process primarily comprises verifYing that the probability of reinforcement
corrosion to be described later in this paper does not exceed the allowable probability of
The limit state for carbonation was defined as the corrosion of outermost bars based
on a judgment that the reductions in the safety of use due to delarnination/spalling of cover
concrete resulting from the corrosion of outermost bars are of concern prior to the reductions in
the structural safety due to the corrosion of main bars. Specifically, the state at which 20% of
the outermost bars are at the onset of corrosion was defmed as the design limit state, while the
state at which carbonation reaches a depth of causing corrosion of any of the bars was defined
of corrosion to the total amount of reinforcement, i.e., the ratio of the part where carbonation
has reached a depth judged as causing reinforcement corrosion. When the depth of corrosion
judgment is the depth of reinforcement, for instance, the corrosion probability is defined as the
part where the scatter of carbonation depth overlaps the scatter of cover depth as shown in Fig.
8. The depth of reinforcement was defined as the depth of corrosion judgment for exterior
parts subjected to rainwater. For exterior parts not subjected to rainwater and general interior
parts, a depth 20mm deeper was defmed as the depth of corrosion judgment.
Assuming that the scatters of carbonation depth and cover depth are normally
distributed with the probability density functions of carbonation depth and cover depth put as C1
In simple terms, carbonation depth, Ct. is generally considered to follow the rule of
.Jt ,for which a large number of equations have been proposed. From among such equations,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
representative ones by Kishitani, Shiroyama, and Izumi were selected to examine the
consistency of parameters, and an equation (Eq. (3)) was newly proposed as a recommendation
by AIJ.
(3)
equation
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Each parameter is optimized, while keeping Pt under the allowable corrosion
probability, to determine the specifications for the cover depth, concrete type, cement type, and
attack, and the state at which 20% of the outermost bars are at the onset of chloride-induced
corrosion was defined as the limit state. The onset of corrosion under chloride attack was
defmed as the time when the chloride ion content of concrete at the reinforcement level exceeds
Fluctuations in the cover depth and the chloride ion content of concrete at the
reinforcement level were selected as the fluctuation factors affecting the progress of chloride
attack. However, few research data have been available in regard to fluctuation in the chloride
ion content at the reinforcement level, demanding future research. If fluctuation in the depth
of concrete exceeding the corrosion threshold of chloride ion content is selected instead of the
fluctuation in the chloride ion content, then reliability design similar to that for carbonation can
be conducted.
When chloride ions permeate through concrete surfaces, their content at the
(4)
t = age (years)
When evaluating fluctuation in the chloride ion content by fluctuation in the depth at
which the chloride ion content reaches the corrosion threshold, Eq (4) has to be solved with
respect to x. The use of approximative Eq. (5) was permitted in consideration of ease of
handling.
x J} +
2
The limit state for frost damage was defined as the time when significant
scaling/cracking occurs on the surfaces of the structure and members. The maintenance limit
state was defined as the time when deleterious scaling/cracking occurs on the surfaces of the
The amount of scaling resulting from freezing and thawing is expressed by the mass
loss in freezing and thawing testing. Surface and internal cracks are expressed directly by the
ratio of length gains in freezing and thawing testing. On the other hand, the relative dynamic
index to the degree of cracking of specimens and correlates with the ratio oflength gains.
result mostly from fine cracks on the surfaces. The relative dynamic modulus was therefore
When no reliable data are available for determining the relative dynamic modulus
causing scaling/cracking and when no test is conducted for verification, a relative dynamic
modulus of 60% was adopted as the design limit state. Similarly, a relative dynamic modulus
of 85% was selected as the maintenance limit state. The selection of 60% is based on a large
number of test reports indicating a relationship in which when the relative dynamic modulus
decreased to around 60% there was a cracking-induced expansion of the fabric of concrete of
approximately I x I o· (O.I% ).
3
It was also judged that such loosening of the fabric facilitates
the permeation of water and air into concrete, causing deterioration to the limit in terms of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
STANDARD-SPECIFICATION-BASED DESIGN
are selected according to the design service life class and environmental class as shown in Fig. 2.
The environmental classes include four area types: general environment areas, chloride-rich
environment areas, freezing environment areas, and chemically erosive environment areas.
The design service lives are 65 and 100 years for general environment areas and 30 and 65
years for chloride-rich and chemically erosive environment areas as a standard. Buildings in a
general environment area with a design service life of 100 years and those in a chloride-rich or
chemically erosive environment area with a design service life of 65 years are as a rule
SUMMARY
This paper has reported on the durability design set as a goal in the
published by AU.
These recommendations for the first time took a definite step toward
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
a prototype of such guidelines in this field. Much time was spent in discussion, particularly
about the framework of performance-based design (7), before forming the recommendations as
system. They are particularly significant in that they summarized the method of establishing
the limit states of the structure and members from the aspect of performance and clearly
discussed the limit states of individual materials including reinforcement, concrete, and
measured environmental conditions have ranged so widely from exhaustive to fragmentary that
they have obscured the essential standard. In the present recommendations, such models and
information were closely reviewed and restructured into a system usable for durability design.
The Architectural Institute of Japan has also been preparing for the establishment of
performance-based design and specification selection is adopted in the ongoing preparation for
REFERENCES
3. Guideline for Design and Practice for Durability of Reinforced Concrete Buildings, 1986
(in Japanese)
5. ISO : IS06241 Performance standards in building - Principles for their preparation and
Performance Verification, pp.59-66, Concrete Journal, Vol.40, No.5, 2002 (in Japanese)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Chapter Title
Chapter! General
Chapter2 Principle of dumbility design
Chapter3 Design e:xteroa1 deterioraJive furres
Chapter4 Design limit- and maintenance limit-
ChapterS Perfonnan<:e-based design- general approach
Chapter6 Standanl-specification-based design
Chapter? Pmctice and quality management
ChapterS MainJemmce and management
Annex Example ofdunlbility design
Performance Equations
M=C·b·d 2 , C=Min(C1.C2 )
Cz =~~(l-xnlX3-xnJl-r(n-lXxnl
3n L-xnJ
-dc~X3dcl-xnJl)
b :width
Allowable bending moment ofbeam d : distance from extreme compression fiber to centroid of tension reinforoement
with rectangular cross section (M) : tensile reinforcement ratio
Pt
lc : allowable tensile stress ofconcrete
f, : allowable tensile stress of reinfurcement
x. : distance from extreme compression fiber to neuttal axis
n : modular ratio of elasticity
r : ratio of compression reinforcement to tension reinforcement
d : distance from extreme compression fiber to centroid of comoression reinforcement
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Deteriorative Determination
Dereriorative design fun:es In case of no measurements
phenomenon method
Exterior: 0.05%
Atmospheric carbon dioxide (CO,) concentration
Interior : 0.20%
Atmospheric sulfur oxide (SO,J concentration
Carbonation Continuous field
Atmospheric nitrogen oxide (NO,J concentration measurement Reliable data
Carbon dioxide concentration of soil and water in
contact with the bnilding
Chloride ion content carried to the building
Continuous field
Chloride attack Chloride ion concentration of water in contact with Reliable data
measurement
the building
Annual number offt=ing and thawing cycles
Continuous field
Freezing and thawing Minimwn air temperature Reliable data
measurement
Moisture acting on concrete
Alkali ion content carried to the building
Reliable data
Alkali-aggn:gale Alkali ion ooncentralion of water in contact with the Continuous field Convert from Chloride ion concentration
reaction building measurement
Moisture acting on concrete Reliable data
Continuous field
Chemical erosion Atmospheric acidic substances concentration Reliable data
measurement
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
General enviromnent area The allowable crack widths in regan! to durability shall be 0.3mm in exterior segments exposed to
rainwater, 0.4mm in exterior segments not exposed to rainwater, and O.Smm in interior segments.
The allowable crack widths in regan! to leakage shall be 0.05mm in segments under constant
water pressure and 0.2nnn in segments without constant water pressure.
The design cover depth shllll be in accordance with Table 5.1 depending on the design service life.
Chloride-rich Quaskhloride-rich The allowable crack width in regan! to durability shall be 0.15nun as a target in chloride-rich
environment environment area envirorunent areas.
area The design cover depth for a design service life of 65 years shall be in accordance with Table 5.2
depending on the external deteriorative forces in regan! to chloride attack.
The maximum water..:ement ratio shllll be in accordance with Table 5.3 depending on the
external deteriorative force class in regard to chloride attack.
General chloride-rich
environment area Table5.3 Maximwn water-eement ratio in chloride-rich environment area
External deteriorative force class Maximwn W/C ("/o)
Heavily chloride-rich
environment area Heavily chloride-rich environment area 45
General chloride-rich environment area 50
Quasi..:hloride-rich environment area 55
Freezing Quasi-freezing The proportioning air conteut shall be 5%.
environment enviromnent area The maximum water..:ementratio shall be in accordance with Table 5.4 depending on the
area external deteriorative forces in regan! to frost damage.
General freezing
environmeut area Table5.4 Maximmn water-cement ratio in fu eziruz enviromnart areas
Heavily freezing Exteroal deteriorative force class Maximwn W/C C"/ol
environment area Heavily :th:ezing environment area 50
General fJeezing environment area
55
environment area
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Design condition ~
I -- Deaip1
Performance requirements
service life
n Deemed-to-satisfy
specifications Standard-specification-based design
Selection of atandard-apecificationa for
r-
structural dimensions, layout of reiaforcement,
covering depth, material property, quality of con"" ta,
mixture propoticms, type of reinforcement,
fiDisbiDg matarialo,etc.
J
+ Partial uoe of Standard-opecification
Shortening of
design service life
Enhancement of
maitenance
Review of
performance
No
Determination of specifications
Components, mixture propotions, covering depth, etc.(Ch.6)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
- 0%: &%perimental
u a - 14.0'Hi: Experimental
Cntsh or eoncret.e
~n weight 1088 0% •••• O%'FEM
---~ 14,,Q%:FEM
u· ~I 11)
"' ..
Spalling of concrete Deflection(mml
Corrosion weight loss 14.0%
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0 15 50 100 Time(year)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Hannah Schell
Manager, Concrete Section
Ms. Schell and is a professional engineer in the Province of Ontario. She joined the
Ontario Ministry ofTransportation in 1981, as a research engineer, and currently is
Manager of the Ministry's Concrete Section. Responsibilities include development of
technical policy and specifications for concrete and for related materials for highway
structures and pavements. Ms. Schell is active on national committees, including
Canadian Standards Association (CSA) Committee A23.1/23.2 on concrete materials and
methods of construction, and is a member of ACI Committee 201, on durability of
concrete.
Mr. Cautillo is the Senior Manager of the Materials Engineering and Research Office of
the Ministry of Transportation of Ontario. A professional engineer, he has been with the
Ministry of Transportation since graduating from the University of Toronto in 1974 and
has held positions in the areas of materials, geotechnical and construction engineering.
He currently manages a diverse group of engineers, researchers, geologists, chemists and
technicians involved in development of standards, research, forensic investigations and in
the provision of expertise in asphalt, concrete, soils, aggregates, foundations, pavements
and chemicals.
BACKGROUND
The first broadly circulated reports of corrosion of ECR occurred in 1986 in the
substructure of bridges in the Florida Keys after only 4 - 6 years service in a marine
environment, in the tidal splash zone regions of piers. The very short time period for the
corrosion distress to become apparent was attributed to the high temperatures, high
humidity and high atmospheric chloride concentrations on the Florida coast. Isolated
examples of corrosion of coated reinforcement were reported from about 1990 onwards
although many field investigations reported good performance (8).
Since the Ontario Ministry of Transportation (MTO) had used ECR in bridge decks since
1979 and in substructure components since 1981 as a standard specification requirement,
these reports of corrosion were a serious concern to the ministry. An MTO survey of
field performance was undertaken in 1988 in response to the reports of poor performance
of ECR in the Florida Keys. In view of the relatively short time that epoxy-coated
reinforcement had been used in Ontario, it was decided to adopt a worst-case
experimental design rather than a random survey. A deliberate attempt was made to
locate those components having the highest chloride-ion contents at the level of the
reinforcing steel. Since all MTO bridge decks are waterproofed, the investigation was
limited to barrier walls and specifically to those barrier walls that had been in service the
longest and were in locations subject to the greatest exposure to deicing salt. Two barrier
walls built in 1979 with shallow cover and which exhibited cracking were investigated;
one had been constructed with black steel, and one with epoxy-coated reinforcement.
The investigation of the wall containing ECR (Eglinton Avenue) is discussed in detail
below. In general, the study conclusions were positive (7).
In 1990, unexpected corrosion ofECR in precast concrete noise barriers was observed by
the ministry; the panels were adjacent to one of the most heavily travelled highway
sections in North America (Highway 401 at Keele Street) which is subject to deicing salt
application. There was very low concrete cover over the steel, and the concrete was of
relatively poor quality.
In 1992, a broader field and laboratory study was initiated by MTO to assess the
performance of ECR. The program was initiated in response to ministry and Ontario
industry concerns regarding the effectiveness of corrosion protection afforded by the
ECR. The study included condition surveys of twelve structures built in the period 1978-
1992, concentrating on barrier walls, and laboratory analysis of samples of ECR obtained
from concrete cores from the structures, from construction sites and from a coating plant.
The condition surveys included: assessment of the degree of reinforcement continuity,
measurement of corrosion potentials using a copper-copper sulphate half cell, corrosion
rate measurements using three electrode linear polarization, visual assessment of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Overall, the study found that the bridges were in good condition and no delaminated
concrete was identified in the structures studied. It was noted that, in some structures,
chloride content at the level of the steel had exceeded the corrosion threshold. There was
evidence of loss of coating adhesion from the steel substrate after approximately l 0 years
of exposure. The loss of adhesion was seen on ECR in both chloride-contaminated and
chloride-free concrete environments. The steel substrate of bars with debonded epoxy
was clean and shiny in chloride-free concrete, but was visibly corroded in concrete where
chloride levels had exceeded the corrosion threshold.
In 1993, due primarily to the findings of this study, several meetings were held between
representatives ofMTO and the epoxy-coating industry. The main purpose of the
meetings was to discuss a mutual concern, the performance and quality of the ECR. As a
result of the meetings, MTO specifications for ECR were altered to improve coating
adhesion by specifying a chromate treatment prior to coating, shorten unprotected storage
time on site, reduce allowable damage to coating, require repair of all visible damage to
ECR prior to placement of concrete, and make mandatory the use of plastic-coated
vibrators for consolidation of concrete containing ECR.
Currently, condition surveys are being conducted on structures that were constructed 20-
26 years ago, that is, structures that were built using "first generation" epoxy. It is too
early to establish what improvements to durability, if any, have occurred as a result of the
alteration of the specification.
In 1996, performance problems with ECR were discovered in two MTO structures
through accidental exposure; barrier walls on two different structures were struck by out-
of-control vehicles, exposing the reinforcement. On one structure (Highway 7,
Peterborough) the exposed ECR was found to be actively corroding (Figure 1).
On the other structure (Queen Elizabeth Way, Burlington Bay Skyway) no corrosion was
observed, however the epoxy showed little or no adhesion to the reinforcement (Figure 2
shows how easily the exposed epoxy could be removed by a knife blade).
During repair work (expansion joint replacement) in the end dams of five bridges, carried
out during this same time period, it was observed that ECR in all five of the end dams
was experiencing corrosion. It should be noted that, like barrier walls, end dams on
ministry structures have no waterproofing or other protective barrier to prevent ingress of
chlorides and moisture, and so are subject to a more aggressive environment than the
waterproofed portion of the bridge deck. Loss of coating adhesion, as well as actual
corrosion, was observed.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
In 1997 the ministry initiated the detailed investigation of the barrier walls of five
highway bridges constructed in Ontario in 1979, to assess the effectiveness of the epoxy
coating in preventing chloride-induced corrosion. From this investigation, it became
clear that the epoxy coating was not effective in protecting reinforcing steel from
corrosion over the long term. This was visually evident as significant areas of some of
the concrete barrier walls showed significant corrosion-induced delamination and
spalling. Figure 3 shows delaminations, determined by sounding, marked on the barrier
wall ofHighway 17's Madawaska River Bridge (Site 29-191), built in 1979 with ECR.
Analysis of cores removed from the structures verified that, where chloride levels
exceeded threshold levels, corrosion of the ECR was taking place. Based on the
structures studied, it appeared that while the epoxy coating may have delayed the onset of
corrosion somewhat, it did not prevent corrosion from occurring. Findings were
summarized in an internal report (9).
The purpose of the remainder of this paper is to provide an update on the ministry's
investigations since the 1997 barrier wall study, and to document the significant findings
with respect to ECR performance.
The following represents the most significant recent findings and observations by MTO
on the performance of ECR, based on both field studies and interaction with other
jurisdictions.
In 1998, a unique opportunity arose to study an Ontario bridge deck constructed with
ECR in both top and bottom deck mats. During discussions with industry, it had been
suggested that MTO's poor experience with ECR might be related to the MTO policy of
using coated steel in the top mat only. The Perley Bridge, a federally-owned structure,
was built in 1931, spanning the Ottawa River between Hawkesbury Ontario and
Grenville Quebec. A study commissioned in 1985 by Public Works Canada concluded
that the concrete deck of the Perley Bridge was in an advanced state of deterioration and
exhibited extensive areas of corrosion. In December 1998, a new bridge was completed
adjacent to the first and opened to traffic. The subsequent demolition of the earlier
structure allowed MTO, with the cooperation of Public Works Canada, to investigate the
performance of deteriorated spans of the old bridge.
Two full-depth slabs cut from Span 17 of the Perley Bridge were delivered to the Albany
Research Centre for examination. Slab Ml was cut from the eastbound lane, in which no
patches were recorded during the 20-year history of Span 17. Slab M3 was cut from the
westbound lane of the bridge, which had been patched extensively, over approximately
40% of its area, beginning early in its life. Measurements made by MTO on site
indicated that the concrete cover on the eastbound lane was almost twice that of the
westbound lane.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Results of Investigation
Testing indicated that the original concrete and the patch concrete used to rehabilitate the
span were of good quality, in terms of compressive strength and resistance to chloride ion
penetration. There was no evidence that the installed ECR failed to meet the applicable
standard at the time of installation; for example, the blast profile for the steel reinforcing
bar was measured and determined to be adequate.
The first recorded indication that the Perley Bridge was in distress, based on the
structural inspection records, was in 1986 (westbound lane section M3) and the first
known patches to the Span 17 deck were done in 1990. Using 1986 as the date that
corrosion damage was first noticed, the "worst-case" estimate of time for corrosion to
cause damage to the deck was 7 years from the date the span was rebuilt; a "best-case"
estimate of time to corrosion is obtained if the date of first repair in 1990 is used. In
other words, corrosion of the ECR in the deck was initiated and progressed to the extent
that damage to the concrete was noticed, in a time period of between 7 and 11 years.
There was inadequate (low) cover depth relative to the top mat of steel reinforcing bars
for slab M3; the researchers concluded that this resulted in chloride levels at the depth of
the top mat in M3 that exceeded the corrosion threshold for black steel reinforcing bar,
within 2. 7 years of deck installation. They suggested corrosion propagation time, based
on work by Sagues, of 3.5 years for bare steel reinforcing bars. Adding these two figures
(initiation time and propagation time) the total time for the first appearance of corrosion
in the slab, if black steel were used or if the epoxy provided no protection at all) would be
expected to be in the area of 6.2 years.
In summary the investigation concluded that the epoxy coating used in the westbound
lane of the Perley Bridge deck in Span 17 provided only 1 to 5 years of additional
protection over bare steel reinforcing bar, in terms of corrosion resistance.
Because of greater cover depth, the corrosion threshold for the top mat of bars in M1 was
estimated to have been reached only after 15.4 years. This resulted in a significantly
extended time to corrosion and consequently, appearance of corrosion damage in Slab
M1 was delayed until about the time of the bridge replacement.
The Eglinton Avenue Bridge over Highway 403, built in 1979, is of particular interest
because it was included in condition investigations conducted by MTO in 1988, 1992,
1997 and 1998. Several evaluation procedures were used, including determination of
chloride profiles, continuity of reinforcement, half-cell potential measurements, rate of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
corrosion and the condition of the steel and of the epoxy coating. The results of the 1988
investigation on the walls of this structure concluded that there was no evidence of
corrosion on the coated bars. There was evidence of poor bond between the epoxy and
the ribs on several bars, but the coating was well bonded to the steel between the ribs. A
similar barrier wall also constructed in 1979 but with uncoated reinforcement, was also
studied for purposes of comparison. The 1988 study concluded that, compared with
black steel, the epoxy coating had reduced corrosion activity substantially during the first
9 years in service but, while the initial indications were promising, the long-term
performance was uncertain (7).
In 1997 a 50-metre section of the barrier wall at the southwest comer of the Eglinton
Avenue Bridge was visually assessed and sounded for delaminations, and approximately
7% ofthe surface area was found to be delaminated. It appeared that the delaminations
had occurred since the 1992 investigation, corresponding to an average delamination
growth rate of approximately 1.4 % per year. In Figure 6 the wall section can be seen
with the delaminated areas marked on the surface. Figure 7 shows the type of corrosion
seen on the ECR in cores removed from delaminated areas of the structure (at the edges
of the photograph it can be seen that the colour of the epoxy-coating in this structure was
brown rather than green; this colour was used for a very brief period of time in Ontario).
An internal report (9) documented the extent of delaminations and corrosion of the ECR
found on the bridge.
The same barrier wall was again sounded for delaminations in 1998. It was found that
some of the previous de laminations had increased in size, while some new de laminations
had appeared. Overall, delaminated concrete covered approximately 9% of the surface
area investigated, a 2% increase from the previous year. Based on the ministry's past
experience, the delamination progression on this bridge, once delaminations were
detected, appears to be consistent with, and typical of, black reinforcement in a similarly
corrosive environment.
The Humber River Bridge (Highway 427- Site Number 37-633) was built in 1982; this
bridge carries heavy urban traffic and is located in northwest Toronto. The bridge deck
was waterproofed and paved at the time of construction. In 1998 the bridge was widened
as part of an upgrading project to increase the traffic capacity of the highway. In the
course of widening the bridge, the barrier wall and sidewalk were removed, and the end
portions of the epoxy-coated transverse bars in the bridge deck were exposed. The bars
were exposed with some care since they would remain in place to provide a structural
link to the new portion of the structure. This offered the ministry a unique opportunity to
examine the condition and performance of the coated reinforcement.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 381
There were many locations of visible deterioration of the epoxy coating. These included
areas with total loss of adhesion of the epoxy coating to the steel, brittleness of the
epoxy-coating and numerous defects and bare areas within the epoxy coating. It should
be noted that this damage was unrelated to any minor damage caused by the concrete
removal operations. Figure 8 shows the exposed ECR after removal of the surrounding
concrete; Figures 9 and 10 illustrate the extent of deterioration of the epoxy coating and
condition of the steel on this site.
The fact that most of the reinforcement was corroding but was not in a more advanced
state of deterioration was probably due to the effectiveness of the bridge deck
waterproofing membrane and the concrete cover in providing protection from chloride
ingress. (Along with waterproofing and paving of all bridge decks, it is standard ministry
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The previous ministry experiences with deteriorating ECR and related corrosion had been
restricted to barrier walls and end dam areas (where no protective waterproofing
membrane is applied), and the poor condition of the ECR in the bridge deck at this site
came as a surprise.
It had been assumed that the coating would be in sound condition and the contract had
originally required only that any areas of damage on the ECR be repaired before
placement of new concrete; the extent of coating and steel deterioration made this
approach impractical. Instead, abrasive blasting was used to remove any remaining
coating on the steel and to clean the steel surface before new concrete was placed, to
ensure a sound bond between the steel and concrete.
The Garden City Skyway (Queen Elizabeth Way, Site 18-111) is a high level bridge
2200m long and 21m wide constructed in 1963 over the W elland Canal. Due to chloride
contamination over the years, significant corrosion-induced deterioration had occurred.
Since original construction, partial rehabilitation of the superstructure had been carried
out including construction of new barrier walls with ECR in 1981. In 2002 a major
rehabilitation was carried out on the structure, including repair of the deck and barrier
walls. Epoxy-coated steel removed from the barrier walls was examined; some areas of
barrier wall had very little cover, resulting in relatively rapid penetration of chloride to
the level of the steel. Figure 11 illustrates an unusual type of corrosion observed; the bars
appear to show initiation and progression of corrosion beneath the coating, while the
coating remained relatively intact (but disbanded from the steel); in the figure rust
staining inside of the dis bonded epoxy can be seen. The corrosion products are
distributed along a length of bar in an even manner, rather than concentrated at a point.
Until recently the idea that corrosion could initiate beneath disbonded epoxy was
considered somewhat speculative. However observations on the steel from the Skyway
would appear to confirm this affect and this is significant to MTO in that this is the first
documented case of such corrosion on a ministry bridge.
As previously mentioned, early performance problems with ECR were first observed in
MTO structures through accidental exposure when in 1996, two different structures were
struck by out-of-control vehicles, exposing the reinforcement.
In 2003, two additional structures (Sites 37- 984 and 24-465) were struck by vehicles,
causing damage to the concrete and exposing the ECR. At Site 37-984 the exposed ECR
was found to be in poor condition. An examination of the entire barrier wall found that
large areas of the wall were cracked and delaminated. The de laminations and cracks
were related to corrosion of the reinforcement.
At Site 24-465 the vehicle collision damaged the concrete. Once the loose concrete was
removed, it was discovered that the reinforcement had large areas of epoxy missing; this
occurred because over time the epoxy had lost adhesion with the steel to the point that the
epoxy bond with the concrete was greater than with the reinforcement. When the
concrete was removed the epoxy was cleanly removed from the reinforcement. The two
incidents, in addition to the 1996 collisions, were considered significant in that in these
totally random events all four of the walls were found to contain actively corroding bars
or bars where the coating had disbonded. This suggests that the loss of bond is a
widespread phenomenon, not confined to any particular age of structure or type of steel.
When chloride is present in sufficient quantity to initiate corrosion, it proceeds.
Findings of recent work carried out by researchers in Florida dealing with ECR damage
and coating disbandment is relevant to the ministry's experience. A study completed by
Hartt (5), who assessed a total of76 salt contaminated concrete slabs that contained ECR
and were exposed outdoors for over six years, concluded that:
(a) Coating defect density was the predominant parameter that affected ECR
performance.
(b) The number of coating defects increases with exposure time in concrete with the
rate of this increase being constant and proportional to the initial defect density.
(c) Coating disbandment occurs and proceeds outward from coating defects.
It can be concluded, based on the above, the better the quality of the coating, the longer it
takes to develop corrosion-induced damage.
Work underway in the U.S. by Transportation Research Board is also relevant to ministry
experience and needs, in terms of the need to identify how to effectively repair structures
containing corrosion-damaged ECR.
The primary goal of a research project recently completed by the National Cooperative
Highway Research Program Project 010-37C (12) was "to address the anticipated need
for premature repair and rehabilitation of concrete bridge elements containing epoxy-
coated rebar (ECR)".
The report states that although concrete damage from ECR corrosion is not presently
widespread, this corrosion control alternative is not likely to provide the relatively
maintenance- free service life that was forecast earlier, and, that intensive repairs and
rehabilitation will ultimately be required during the anticipated service life of many
structures.
The report also states that cathodic protection is now generally recognized as one method
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
that can be used for control of ongoing corrosion of bare steel in chloride contaminated
concrete. However, ECR presents unique repair challenges in that treatments such as
cathodic protection and electrochemical chloride extraction may be compromised or
ineffective. For example, the usefulness of the techniques may be limited because of the
anticipated lack of electrical continuity between ECR bars, potential for coating
disbandment during application of current, and potential for lack of current flow to steel
beneath dis bonded but intact areas of coating.
The MTO first used ECR in bridge decks and barrier walls in 1979 and a number of the
early structures have now undergone repairs, as have those of other Ontario jurisdictions.
Table 1 presents a list of structures that have suffered deterioration due to corrosion and
have been repaired.
The majority of repairs to date have been largely limited to barrier walls. The use of
improved waterproofing materials on bridge decks has reduced their vulnerability to
corrosion damage, but no such supplementary corrosion protection is present in
components such as barrier walls, expansion joint dams or sidewalks.
The conflicting research findings cited by industry regarding the performance of ECR led
the ministry to convene an expert task force to provide some resolution of the issues
surrounding performance ofECR. It consisted of representatives from the ministry and
proponents of both "pro" and "con" epoxy-coated reinforcement camps, from North
American industry, government and academia.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The task group met for two days in August 1998 in Toronto, and a document was
prepared at the meeting which outlined areas of agreement, areas where there was lack of
consensus and areas where more work was needed. The document appears in appendix A.
It became clear that there were very significant areas of technical disagreement among
the experts within this industry, not necessarily between MTO and everyone else.
In summary the discussion of the task force was not conclusive and the ministry was
uncertain as to what its future direction with regard to the use ofECR should be. This led
to the Ministry to hire an independent research team to review the issues and to develop a
corrosion protection strategy for Ministry bridges.
In July 2000 the principal investigators for the project, Professors C. Hansson and R Haas
of the University of Waterloo, provided the final report "Corrosion Protection Strategies
for Ministry Bridges" (4) to the ministry.
(i) "When the lack of any suitable repair methods and non-destructive evaluation
methods for ECR is taken into account, it is clear that an extra service life of
only about five years does not warrant the use of ECR and it must be concluded
that ECR will not be viable for a design life of 75 years."
(ii) "On the basis of technical merit, the use of stainless steel is strongly
recommended, particularly for those components exposed to severe conditions,
including expansion joint dams, barrier walls and sidewalks."
(iii) "The use of high performance concrete as defined by MTO, (OPSS 904 HPC,
May 1998) is recommended for components exposed to severe conditions."
Since the submission of the above report, to the present day, the ministry's corrosion
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
protection strategy has continued to evolve. Use of stainless steel reinforcement is
now the norm for those structure components with the most severe exposure
(barrier walls, expansion joint dams, and sidewalks on structures), in most new
construction. In addition, significant use of high performance concrete to enhance
structure durability has been made in some areas of the province. Balancing the
increased cost of corrosion protection, against the value obtained, remains a challenge for
the ministry.
Currently MTO condition surveys involve carrying out a detailed visual inspection of the
structure as well as a condition investigation (both destructive and non-destructive)
including a half-cell potential survey to assess the corrosion condition of the steel. MTO
inputs the information obtained from the condition survey into a process to establish the
method of rehabilitation for the structure. Both the condition survey protocol and the
decision matrix process were developed based on experience with black steel
reinforcement.
Since the epoxy resin is not electrically conductive, half-cell potential measurements on
epoxy-coated steel do not provide useful information. Conventional condition surveys
conducted on structures containing ECR provide limited information and use to the
Ministry. It was recognized that a new approach for condition assessment of structures
built with ECR is required. This is an issue not only for MTO, but also for many
transportation agencies.
The recommended protocol consists of a two-phase investigation process, with the first
phase to be carried out after a minimum of 15-20 years of service. In the Phase One
investigation, standardized test methods are utilized to collect data on the condition of the
steel and concrete, that are then input to a service life prediction model to predict a time-
to-corrosion and service life for the structure. The use of bridge deck service life
prediction formula for both the corrosion initiation period and the propagation period (to
reach a specified level of structural damage), represents a new approach that takes into
account the influence of the waterproofing membrane on chloride penetration and state-
of-the-art knowledge with respect to the corrosion performance of epoxy-coated bars in
chloride-contaminated concrete. The findings of the Phase One investigation are used to
determine the timing of the second phase of investigation, based on the anticipated time
at which significant damage to the structure will have occurred. The primary goal of this
Phase Two survey is to collect sufficient data to support the rehabilitation process.
A full description of the test methods proposed for use, and the method of service life
prediction, is included in the report. The recommended protocol takes a rigorous
approach which is expected to yield much more useful information than is currently
available to the ministry, but this comes at a higher cost than a conventional survey.
Field trails of the new protocol are necessary in order to refine the process for future
ministry use, as well to validate the overall approach. A Phase One condition survey was
carried out on selected sections of the Burlington Bay Skyway in the fall of 2004, for this
purpose. Data from this survey is being used to fine-tune the protocol.
CONCLUSIONS
1. Based on the findings of field studies described in this report, which include a
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2. The fact that several MTO bridges originally built with ECR have already been
rehabilitated prior to achieving a 20-year service life, is cause for concern.
3. The number of structures across the province which have exhibited deterioration
is a strong indicator of the widespread nature of ECR corrosion. The
deterioration is not specific to a particular supplier of steel, to a specific area of
the province, or to unusual exposure conditions.
5. It is acknowledged that the use ofECR can increase the service life of structures
over those constructed with bare steel, however this increase is difficult to
quantify. The consultant investigation of the Perley Bridge concluded that the
epoxy coating provided only 1 to 5 years of additional corrosion protection over
bare steel reinforcing bars (for a concrete deck in which all steel was epoxy-
coated).
6. Since the Ministry's initial move away from ofECR in 1998 no new
information has become available which would either improve confidence in its
performance in the future or indicate that the ministry's experience is due to
unique conditions. A number of jurisdictions have now discontinued use of
ECR.
7. Because the prior use ofECR severely limits the investigative techniques and
repair methods that can be used, new approaches to condition assessment and
repairs will be needed for the large population of Ontario structures containing
ECR. A new protocol for condition surveys has been developed for use by
MTO on structures containing ECR, which utilizes steel and concrete condition
information and service life modelling; it is being refined through use on
selected trial structures.
REFERENCES
1. Clifton J.R., Beeghly, H.F. and Mathey, R.G., "Non-metallic Coatings for
Concrete Reinforcing Bars". Report FHWA-RD-74-18 Federal Highway
Administration Washington D.C., 1974.
2. Covino, B. S., Cramer, S. D., Holcomb, G. R., Russell, James H., Bullard, S. J.,
Dahlin, C., and Tinnea, J. S., "Performance of Epoxy-Coated Steel
Reinforcement in the Deck of the Perley Bridge", Final Report prepared for the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
10. Pike, R.G., Hay, R.E. Clifton, J.R. Beeghly, H.F. and Mathey, R.G. "Non-
Metallic Coatings for Concrete Reinforcing Bars" Public Roads 1973.
13. Virmani, Y.P., Clear, K.C. and Pasko Jr., T.J. "Time-To-Corrosion of
Reinforcing Steel in Concrete Slabs" Vol. 5 - Calcium Nitrite Admixture or
Epoxy-coated Reinforcing Bars as corrosion protection systems. Report no
FHWA/RD-83/012, Federal Highway Administration, Washington D.C. 1983.
15. Weyers, R. E., Pyc, W., and Sprinkel, M. M., "Estimating the Service Life of
Epoxy-Coated Reinforcing Steel", ACI Materials Journal, Volume 95, 1998,
Pages 546-557.
Summary of Discussion
Points of Agreement
1. When the chloride content of the concrete at the level of the steel reaches
threshold value, corrosion of ECR begins at damage sites; that is, the time
period required to initiate corrosion is the same for coated steel and for black
steel.
2. Bars in the "Real World" will always have some holidays or coating damage.
4. The length of time coated steel is exposed to the elements (prior to placing in
concrete) reduces long-term effectiveness.
9. Research tells us that poor quality material in a chloride environment will result
in corrosion damage.
10. Damage reduces the effectiveness of corrosion protection; the greater the
damage, the lower the effectiveness.
13. There have been no studies that specifically address what happens to corrosion
protection if coating loses adhesion before chloride reaches the steel.
1. There is disagreement on the cause of, and impact of, reduced coating adhesion
that has been observed in some studies. What is the cause of the poor/reduced
adhesion observed?
-the adhesion of the coating at the time of manufacture was poor, versus.
- loss of adhesion occurs over time in a concrete environment what is the impact
of a coating that has lost adhesion (once chloride reaches the steel)?
- still provides benefit in terms of corrosion protection/extension of service life.
- provides no benefit; once adhesion is lost, corrosion can progress.
- experts disagree on the role of corrosion itself in the reduction of adhesion of
epoxy-coating.
2. There is uncertainty as to how fast corrosion will progress once chloride reaches
the steel. Once corrosion is initiated, what is the impact of epoxy-coating on
the rate of corrosion?
-rate of corrosion is reduced significantly, versus
-no impact; rate of corrosion is the same as for black steel
the material that MTO examined in its field investigations was produced. There
is disagreement as to whether overall quality control (production and field) is
sufficient today to ensure an acceptable low degree of damage; there is
uncertainty as to what is an achievable degree of damage.
Although there was insufficient time to explore the following issues in detail during the
meeting, it is recognized that they are important and require consideration.
3. There is a need for more study of the assumptions used in life cycle cost
analyses.
Table 1
Structures Constructed with ECR Where Corrosion Damage has been repaired
Name Location MTOSiteNo. Date of
Installation ECR
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Figure 2 Accidental exposure of ECR, 1997, Burlington Bay Skyway (Site 36-o6o)
Figure 9 ECR exposed by construction operations, Humber River Bridge (Site 37-633)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: This paper proposes a model of the mechanical behavior of corroded
reinforced concrete members subjected to bending under service load. The model is
based on the formulation of a macro-element to be used in finite element analysis,
having a length equal to the distance between two consecutive flexural cracks and
a cross-section equal to the member cross-section. The mechanical formulation is
based on the concept oft he transfer length necessary for the transmission of tensile
load from the reinforcement to tensile concrete through bond. It is thus possible
to take into account the effect of reinforcement corrosion on the bond between the
steel and concrete by increasing the transfer length as a function of the intensity of
corrosion. The variation of the transfer length with degree of corrosion is expressed
using a scalar damage parameter. Experimental validation is performed on a 17-year-
old beam kept in a chloride environment under its service load.
INTRODUCTION
The prediction of changes in the mechanical behavior of reinforced concrete structures
during their ageing is an objective of major importance for building owners. It will help
in the decision to plan repair or reinforcement of the structure, set up a maintenance
program or, on the contrary, envisage the demolition and re-building of the structure.
The main cause of ageing damage in reinforced concrete structures is reinforcement
corrosion. Damage can be detected visually as coincident cracks along the reinforcement,
which is significant in terms of reduction of the bar cross-section and loss of bond
between reinforcement and concrete.
Building design standards neglect the tensile concrete located between flexural cracks
because it does not have any effect on the load-bearing capacity (Ultimate Limit State).
So, by using these standards, it is not possible to evaluate the effect of bond degradation
due to reinforcement corrosion on the serviceability of a reinforced structure [1,2]. In this
paper, a model of reinforced concrete behavior in the post-cracking state is proposed
based on a reduced-inertia approach, which takes tension stiffening and corrosion effects
into account.
In normal conditions, reinforced concrete elements subjected to bending are always
cracked in their tensile zones because the tensile strength of concrete is low.
Consequently, the usual mechanical models for reinforced concrete design do not take
into account the tensile concrete, since it does not significantly influence the load-bearing
capacity of a structure [3]. However the tensile concrete located between two flexural
cracks contributes to the flexural stiffness of the structural element. Indeed, the bond
between the reinforcing bars and the concrete is still active in these areas and leads to a
mechanical interaction between the steel bars and the concrete. This is called the
"tension-stiffening" effect, and is well known to structural engineers. To take this
phenomenon into account in order to correct the calculations based on neglecting
concrete tensile strength, two main approaches have been adopted.
In the first approach, tension stiffening is modelled by decreasing the concrete elastic
modulus based on the constitutive laws of concrete under tension, which include a strain
softening curve in the post-cracking phase [4,5,6]. Bond-slip based models are also
included in this approach. These models are based on the assumed bond stress
distribution along the tension zone and are constructed from the force equilibrium and
strain compatibility condition at the cracked concrete matrix [7-16].
The second approach is based on the moment-curvature relationship [17-20], which
takes into account the development of flexural cracks during loading. The tension-
stiffening effect is related to various experimental parameters: tensile strength of
concrete, creep, surface ratio between bars and concrete, bar diameter, type of bars
[21,22). The global behavior is then calculated by integration of the moment-curvature
law.
The use of the moment-curvature law is also the CEB-FIP approach [23] with two
alternatives. The first is empirical (Chapter 3.6 CEB-FIP Model Code [23]), and the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 401
second is based on the concept of an effective tensile member that replaces the tensile
concrete between cracks. The cross-section of the effective tensile member determined by
finite element analysis is 2.5*(h-d)*b where h is the cross-section height, d the effective
depth (distance between the centre of gravity of the bars and the compressive fibre of the
beam) and b is the beam thickness. The moment-curvature law is corrected to consider
the tension-stiffening effect by using the average strain in the effective tension member
resulting from the force equilibrium.
Recently, Kwak et al. [24] proposed an analytical approach to model the tension
stiffening effect in the case of tension members and beams. Unlike previous approaches
based on the assumed bond stress distribution function, this model represents the normal
strain distribution in the concrete by using a polynomial function on the transfer length,
which is the length necessary for the full transmission of the tensile load from steel bars
to concrete.
In the present paper, an approach based on a linear variation of concrete and steel
strain over the transfer length is adopted. This approach is the same as the CEB-FIP
model, which assumes a constant bond strength between reinforcement and concrete. It is
then possible to take into account the development of corrosion and the resulting de-
bonding by increasing the transfer length between concrete and steel bars. The
implementation of the model is based on the definition of macro-elements having a
length equal to the distance between two flexural cracks and the same height as the beam.
To establish the diagnosis of a corroded element, it is necessary to perform numerous
tests (NDT and visual) on the structures, so as to know the exact spacing between flexural
cracks. The reduced moment of inertia of these macro-elements then depends on the
transfer length and its increase due to corrosion. The moment of inertia of the macro-
element is then implemented in an elastic finite-element analysis that calculates the
global stiffness of corroded reinforced concrete members. The validity of this approach is
established by comparison with experimental results.
EXPERIMENTAL PROGRAM
h
and wen Lt -<x< Lelem
- 2
This average flexural curvature allows the average bending moment of inertia Im of
the macro-element to be determined.
_M
Xm-Eim (5)
The approach developed in this paper consists of increasing the transfer length as a
function of the intensity of the steel bar corrosion (Fig. 4). The new transfer length taking
into account corrosion is called Ltcor.
According to the intensity of the corrosion, the transfer length varies from the initial
value Lt (without corrosion) and tends to infinity for serious corrosion leading to total
debonding between steel bar and concrete.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
404 Fran~ois et al.
In order to be more in touch with other studies taking into consideration environmental
effects on the mechanical behavior of reinforced concrete [29,30,31], a scalar bond
damage parameter De is introduced that varies from 0 (no corrosion) to 1 (total
debonding due to corrosion). The relation between Ltcor and De is the following:
Ltcor=(_ll_)
1-0c
(7)
When Dc=O, there is no damage due to corrosion and Ltcor=Lt. When Dc=l, there is
total damage of the bond between steel bar and concrete, and Ltcor tends to infinity
(fig.5).
The deterioration of the bond due to corrosion of the reinforcement is linked to the
section loss of the bars. The appearance of cracks (parallels to steel bars) and their width
are directly correlated to the amount of corrosion products present due to the oxidation
process [32]. The loss of bond is correlated with the cracking process because of the loss
of confinement of the bars in the surrounding concrete. Then, the loss of bond varies
because of the development of corrosion cracks (length and width) but also because of
the increase in the amount of corrosion oxides which are weaker than steel [33]. The
bond damage can then be expressed as a function of the section loss as follows.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
where ~Aso is the section loss threshold that initiates the first crack
~As is the section loss of the bar
n is a empirical parameter describing the quantitative variation of progressive
debonding versus the section loss of the reinforcement.
Variation of the Bond Damage Parameter with the Section Loss due to Corrosion
It is difficult to evaluate the variation of the De parameter versus the section loss of
reinforcement experimentally. Mechanical tests could be done on corroded tension
members to measure the transfer length. But, unfortunately, the transfer length is affected
by corrosion once corrosion cracks have appeared. The use of strain gauges on such
cracked tension members would not provide significant measurements.
Much research has been conducted on corroded samples using the pull-out test to
estimate the bond strength versus corrosion. In most cases, the corrosion was not natural
corrosion but was accelerated by an electric field. Of course, doubts could be raised as to
whether such accelerated corrosion is representative, but these results allow the effect of
corrosion cracks on the bond to be estimated. Fig. 6 shows a synthesis of experimental
results obtained by some researchers [33-37]. In these research works, the progressive
debonding with increasing intensity of corrosion is expressed as a reduction of the failure
bond stress in a pull-out test. On Fig. 6, the experimental results are expressed using a
scalar damage parameter Dpo which represents the relative loss of failure bond stress in
pull-out tests:
and 'tuc =the failure bond stress in the pull-out test on a c% corroded specimen (c% is
the relative section loss of the reinforcement)
Of course, the use of a damage parameter calculated from pull-out tests to evaluate the
variation in transfer length for tension tests is open to question. Nevertheless, the results
of pull-out tests will be influenced by the re-bar confinement loss due to corrosion cracks
in the same way as the variation of transfer length will be influenced by the same loss of
confinement.
Fig. 6 shows a strong decrease of the failure bond strength (or increase in the damage
parameter) versus the intensity of corrosion. For relatively low percentages of section
loss, there is a large amount of bonding damage. This phenomenon is well fitted by using
n=5 in the Eq. 8 for the scalar damage De. The parameter De was calculated using the
section loss corresponding to the percentage of corrosion c%.
account the location of the flexural cracks. For each macro-element of beam AlCL, the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
406 --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`--- Fran~ois et at.
damage parameter De is calculated as a function of the section loss recorded at the
location of the macro-element. The average moment of inertia is then deduced from the
length of the macro-element, the damage parameter and the loss of section of the
reinforcement. When the macro-element is not located between two consecutive cracks,
for example at either end of the beam, only the loss of section of the reinforcement
reduces the moment of inertia. As a result, the moment of inertia is very close to that
corresponding to the non-corroded state.
Fig. 9 shows the comparison between the experimental behavior of the control beam
(A1T) and corroded beam (A1CL) and the proposed model. As mentioned previously, it
is the behavior of the pre-cracked beam that is compared. There is a good agreement
between experimental stiffness of both the control and corroded beams and the model.
Fig. 9 shows that the decrease of the stiffness between the corroded beam and the control
beam is about 37%. This variation is important and corresponds to only 11% of section
loss of the bars in the middle part of the corroded beam according to the map of section
loss (Fig. 8 and Table 2). An 11% section loss of tensile reinforcement will lead to only a
1% decrease in stiffness if the transfer length is not changed and only the section loss of
reinforcement is used to decrease the average moment of inertia of the macro-element.
This calculation then corresponds to bond that is not affected by corrosion. If the tension-
stiffening effect is neglected, the 11% of section loss of bars will lead to a 10% decrease
in the stiffness of the beam. Then in these two opposite cases (bond not affected by
corrosion or no tension-stiffening effect), the influence of the section loss due to
corrosion of the reinforcement is insufficient to explain the decrease in stiffness shown
by experimental data. As a result, the loss of bond due to reinforcement corrosion appears
to be the main parameter for describing the mechanical behavior of reinforced concrete
members under service loads. There is a lack of knowledge on the effect of corrosion on
the deflection behavior of RC members. Most test methods used in the literature are
performed by separating corrosion and mechanical tests as sequential processes [3,33-
37,39,40]. In this study, as in real structures, corrosion takes place while the beam carries
load and the two effects act synergistically. Furthermore, corrosion has not been
accelerated by impressing an electrical current on the steel. In an study assessing the
structural effects of reinforcement corrosion under simultaneous load and corrosion
CONCLUSION
This paper deals with a model to represent the mechanical behavior of reinforced
concrete members under their service load when damaged by reinforcement corrosion.
The model is based on the discretization of reinforced concrete members using macro-
elements whose length is the distance between two consecutive flexural cracks. Because
the model will be used to evaluate the behavior of existing corroded members, the
location of existing flexural cracks will be known during the diagnostic phase. The
tension-stiffening effect is taken into account by using a transfer length. The effects of
local intensity of the reinforcement corrosion is represented in the model as both a
reduction of the reinforcement cross-section and an increase of the transfer length to
account for the progressive debonding due to corrosion. Thus, this approach allows the
combined effect of the reduction of cross-section and the debonding due to corrosion to
be taken into account in the mechanical behavior of reinforced concrete members.
The use of the model requires knowledge of the transfer length, which could be
evaluated for a given concrete by experimental tests on tension members. The model
results in good fit to experimental results obtained on a non-corroded reinforced concrete
beam. The transfer length needs to be increased to account for the effect of corrosion on
the bond. This modification of the transfer length is modelled through a damage
parameter, the variation of which versus the intensity of corrosion is based on data from
pull-out tests on accelerated corrosion specimens found in the literature. The proposed
model was validated experimentally using a "naturally" corroded 17 -year-old beam. An
important point is that the experimental program allows for assessing the structural
effects of reinforcement corrosion under simultaneous load and corrosion conditions.
Good agreement was obtained between experiments and the model. This first result is
interesting, but it is necessary to test the model on other reinforced concrete members
with different geometries to draw firm conclusions about its reliability.
Results show that the variation of the mechanical behavior of reinforced concrete
members under service load is mainly caused by loss of bond due to reinforcement
corrosion.
Another benefit of the proposed model based on a local approach (at the scale of a
macro-element) for the corrosion is that it allows the variation of the intensity of
corrosion along the bars to be taken into account. Indeed, on a reinforced concrete
structure, the development of corrosion is random in terms of both location and intensity.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
[3] Rodriguez, J., Ortega, L. M., Casal, j., Diez, J. M., "Assessing structural conditions of
concrete structures with corroded reinforcements", 4th Int. Congress on concrete in the
service of mankind, Proceedings of an International Conference, Dundee UK, 1996
[4] Gilbert RI, Wanner RF. Tension stiffening in reinforced concrete slabs. J of Structural
Engineering ASCE 1978;104(12):1885-1900.
[5] Vecchio FJ. Non-linear finite element analysis of reinforced concrete membranes.
ACI Struct. J. 1989;86(1):26-35.
[6] Thomas TC, Zhang H, Zhang LX. Tension stiffening in reinforced concrete
membrane elements. ACI Struct. J. 1996;93(1):108-115.
[7] Chan HC, Chueng YK, Huang YP. Crack analysis of reinforced concrete tension
members. J of Structural Enginering ASCE 1992;118(8):2118-2132.
[8] Choi C-K., Cheung S-H., tension stiffening model for planar reinforced concrete
members, computers&structures Vol.59, No.1, 1996, 179-190.
[11] Hwang S-J., Leu Y-R., Hwang H-L, tensile bond strengths of deformed bars of high-
strength concrete, ACI Structural Journal, V. 93, No. 1, 1996, 11-20.
[12] Kaufmann W., Marti P., Structural concrete: cracked membrane model, Journal of
structural engineering, 1998, 1467-1475.
[13] Manfredi G., Pecce M., A refined RC beam element including bond-slip relationship
for the analysis of continuous beams, computers&structures 69 (1998), 53-62.
[14] Salem H., Maekawa K., spacially averaged tensile mechanics for cracked concrete
and reinforcement in highly inelastic range, Concrete library of JSCE No. 34, 1999, 151-
169.
[16] Yang S., Chen J., Bond slip and crack width calculations of tension members, ACI
Structural Journal V. 85, 1988, 414-422.
[18] Carreira JD, Chu K. The moment- curvature relationship of reinforced concrete
members. ACI Structural Journal1986;83(2):191-198.
[22] Prakhya GKV, Morley CT. Tension stiffening and moment curvature relations of
reinforced concrete elements. ACI Structural Journal 1990;87(5):597-605.
[23] CEB-FIP model code. Structural concrete. basis of design volume 2. Updated
knowledge of the CEB-FIP model code 1990. July 1999
[24] Kwak HG, Song JY. Cracking analysis of RC members using polynomial strain
distribution function. Engineering Structures 2002;24( 4 ):455-468.
[25] Fran~ois, R., Arliguie, G, Maso, J.C. (1994) Durabilite du beton arme soumis a
I' action des chlorures, Annales de l'ITBTP, no529, p.l-48.
[26] Fran~ois, R., Ringot E., Capteur de force sur chevetre de charge pour poutre en
beton arme, GAMAC INFO, n°2-3, pp. 21-28, 1988.
[27] Maldague JC. Contribution a l'etude des deformations instantanees des poutres en
beton arme. Institut Technique du Bi'itiment et des Travaux Publics 1965;n°213.
[28] Batoz, J-L., Dhatt, G., Modelisation des structures par elements finis, vol.2, poutres
et plaques, Ed. Hermes (1990)
[29] Carde C., Fran~ois R., « Aging damage model of concrete behavior during the
leaching process», Materials and Structures, vo1.30, pp.465-472, October 1997.
[30] Gerard, B., Pijaudier-Cabot, G., La Borderie, C. (1998) Coupled diffusion damage
modelling and the implications on failure due to strain localisation, Int J. Slids &
Structures, vol. 35, pp. 4170-4120.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
410 Fran~ois et al.
[31] Saetta A., Scotta R, Vitaliani, R. (1999) coupled environmental-Mechanical damage
model of RC structures, J. Engrg.Mech. ASCE, vol. 125, pp. 930-940.
[32] Vidal, T., Castel, A., Fran~ois, R., "Analyzing crack width to predict corrosion in
reinforced concrete", Cement and Concrete Research, Vol34/1 (2004), 165-174.
[33] Almusallam, A.A., Al-Gahtani, AS., Aziz, A.R., Rasheeduzzafar (1996); Effect of
reinforcement corrosion on bond strength, Construction and Building Materials, Vol. 10,
No.2,pp. 123-129, 1996
[34] Mangat, P.S., Elgarf, M.S.; (1999); Bond characteristics of corroding reinforcement
in concrete beams, Materials and structures, vol. 32, pp 89-97, march 1999
[35] Fang C., Lundgren K., Chen L., Zhu C., (2004), Corrosion influence on bond in
reinforced concrete, Cement and Concrete Research 34 (2004) 2159-2167
[37] Lee H.S., Noguchi T., Tomosawa F., (2002); Evaluation of the bond properties
between concrete and reinforcement as a function of the degree of reinforcement
corrosion, Cement and Concrete Research, 32 (202), pp. 1313-1318
[38] Ballim Y., Reid, J.C. "Reinforcement corrosion and the deflection of RC beams- an
experimental critique of current test methods", Cement and Concrete Composites 25
(2003) 625-632.
[39] Stanish K., Hooton R.D., Pantazopoulou S.J., Corrosion effects on bond strength in
reinforced concrete. ACI Struct J 1999; 96 (6), 915-921.
[41] Fran~ois R., Castel A. and Vidal T., A finite macro-element for corroded reinforced
concrete, Materials and Structures, (to be published).
AlT
Nodes Location(m) Lelem(m) Im(m')
1 0.000 0.575 2.899e-4
2 0.575 0.105 1.02e-4
3 0.680 0.210 1.02e-4
4 0.890 0.240 1.114e-4
5 1.130 0.230 1.085e-4
6 1.360 0.200 0.994e-4
7 1.560 0.250 1.142e-4
8 1.810 0.230 1.085c-4
9 2.0401 0.220 1.055e-4
10 2.260 0.105 1.02e-4
11 2.365 0.435 2.899e-4
12 2.800
Table 2 --Mesh and characteristics of the macro-elements for corroded beam AlCLl
AlCLl
Nodes Location(m) lelem(m) llAs(m') De 1m (m')
I 0.000 0.205 3.00E-05 0.296 2.899e-4
2 0.205 0.105 3.90E-05 0.378 0.791e-4
3 0.310 0.210 4.00E-06 0.014 l.OO?e-4
4 0.520 0.130 4.00E-06 0.014 0.842e-4
5 0.650 0.250 4.20E-05 0.403 0.812e-4
6 0.900 0.250 3.50E-05 0.342 0.851e-4
7 1.150 0.250 4.50E-05 0.427 0.798e-4
8 1.400 0.220 4.20E-05 0.403 0.785e-4
9 1.620 0.200 3.60E-05 0.351 0.793e-4
10 1.820 0.230 O.OOE+OO 0 0.108e-4
11 2.050 0.200 6.00E-06 0.038 0.963e-4
12 2.250 0.270 1.20E-05 0.109 1.093e-4
13 2.520 0.105 2.50E-05 0.248 0.805e-4
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
nbbedw8J{
14 stirrups <t6 mm:
ribbed ci>16
i---15ll----l
-------- &.me
befo~e cracking
~----------~~
Lei em
cracks
: ~ 43111 ~ :
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
macr<H!Iement
&; (x)
&inc
befoke cracking
~~---+----~~
Fig. 4 --Typical view of variation of transfer length to take into account the loss of
bond due to corrosion
X
Strain distribution
in re-bar
Fig. 5-- Variation of transfer length and bond damage parameter versus intensity of
bar corrosion
0.9
0.8
0.7
--n=5
0.6 o Mangat-Eigarf
x C.Fang et al.
0.5 t:.. Cabrera
o H-S. Lee et al.
0.4 <> A. Almusallam et al.
0.3
0.2
~
0.1 corrosion(%)
0
0 20 40 60 80 100
Fig. 6 --Fitting of the damage parameter model with the experimental results from
References 33-37
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
5
.JJV, X ":
)< ••••.
Fig. 8 --Variation of section loss in the tensile reinforcement due to corrosion along
beam AtCLt
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
12
10
8
o Control beam A1T
- • -Ltmodel
4
--Ltcormodel
2
Deflection (mm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: Most organic admixtures for mortar and concrete are based on mineral
oil derivatives. Future generations will need natural replacements that can secure
a sustainable development, the so-called bio-admixtures. This study focuses on
influence of vegetable oils on long term water repellency and other durability
aspects.
Vegetable oils from sunflower, olives, soya beans, peanuts, linseeds, corn
and rapeseeds were tested in 2002: Dosages were o.o, o.s and 1.5% vegetable oil by
cement weight. The flexural and compressive strength of 1:3 mortars with w/c = o.so
at 1 and 28 days were then measured together with the capillary water absorption
and water vapour diffusion.
Thereafter the specimens were stored in room temperature at 93%
relative humidity for three years before compressive strength, chloride intrusion,
carbonation, capillary water absorption, monolayer capacity and chemically bound
water on the mortar samples were measured to investigate long term durability of
the samples and how the water repellency changes with time. This last program is
reported in the present paper.
Dr. Harald Justnes is Chief Scientist at SINTEF Concrete, Trondheim, Norway. His field
of interest covers the chemistry of cement, concrete, admixtures and additives (including
polymers) from production, through reactivity, to durability. Justnes is also Adjunct
Professor in "Cement and Concrete Chemistry" at Institute of Materials Technology,
Norwegian University of Science and Technology (NTNU), Trondheim.
INTRODUCTION
Vegetable Oils
Natural fats consist primarily of glycerides (i.e. esters of glycerol and fatty
acids) but also other lipids in minor quantities. The fatty acids influence greatly both the
physical and chemical character of glycerides because of their preponderant weight in the
glyceride molecules and also because they comprise the reactive proportion of the
molecules. Most oils and fats are based on about a dozen fatty acids. These acids can be
divided into saturated and unsaturated acids. The saturated acids contain only single
bonds between carbon atoms and are characterized by relatively high melting
temperatures, while unsaturated acids have one or more pairs of carbon atoms joined by
double bonds and are low melting and chemically much more reactive.
Glyceride is not chemically stable in the highly alkaline interior of a cement
mortar. It will hydrolyse to glycerol, CH2(0H)CH(OH)CH 2(0H), and fatty acid anions
consuming three hydroxyl ions in the process. The carboxyl group, -Coo-, of the fatty
acid anion will coordinate strongly with calcium. The fatty acid will thus be trapped
inside the mortar and the hydrophobic part of the molecule will create water repellency.
EXPERIMENTAL
Samples
Capillary suction
of time. The samples were then finally dried to a constant weight at 105°C. Open macro
porosity (com), closed macro porosity (cern), capillary porosity (ccap), entrained air volume
(cair), average density of mortar solids (Psoi) and dry density of mortar (Pdry) were
calculated as described by Justnes et al.(3)
Chloride intrusion
Monolayer capacity
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A =PVm.Asp·N ~3812·V [2]
s M W A m
where p, M and W is the density, molecular weight of the adsorbent and weight of the
sample respectively, NA is Avogadro's number and Asp is the specific molecular area
(0.114 nm2 for water).
a= Wn/0.23 [3]
The weight loss during ignition at 1000°C in this case is however the sum of the
water lost froiJ.1 the cement phase, weight loss by decomposition of limestone
(CaC0 3 (s)---+CaO(s) + C0 2 (g)), water lost from the sand (assumed to be 2.3% due to the
fact that the sand used for the mortars was not available for tests of ignition loss) and
evaporation of oil. The amounts of chemically bound water for all mortar samples were
thus calculated by subtracting the weight loss subscribed to the limestone due to release
of C0 2, the amount of oil added to the cement and the water subscribed to the sand from
the total weight loss measured from the samples. The degree of hydration was thereafter
determined by equation [3].
Carbonation
The results from the compressive strength measurements done after 3 years of
curing are depicted in Fig. 1. Justnes et al. (2) found that both the 1-day and 28-day
strengths for mortars with oil were not significantly lower than the reference mortar.
Albayrak et al. C6l studied the effects of using sunflower oil as grinding aid on the strength
of cement and found a 50% decrease in compressive strength after 28 days of curing.
However, their experimental conditions are not comparable to the present study (i.e. oil
contact with dry cement versus fresh mortar). The 3-year compressive strengths in this
study are approximately 30 % lower for samples with oil compared to the reference
samples. The strength of the reference mortars has increased by 100% compared with the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
28 day strength, while samples with oil increased strength by 60-70% in the same period.
Similar strength increase were found by NevilleC7l. The strengths do however not show a
clear dependency on the oil dosage which indicates that even 0.5% vegetable oil is
sufficient to slow down strength development.
A reason for lesser strength gain might be that the transport of water from one
part of the microstructure to another may be obstructed when the hydration has proceeded
far enough to enrich the oil concentration to become an important phase in the liquid (i.e.
oil droplets may block pores). This could also offer an explanation to the lower strength
development from 28 days to 3 years versus I to 28 days. A second reason for this latter
finding might be that the oxygen over time diffuses through the sample and oxidizes the
liquid (i.e. mobile) unsaturated fatty acids to a solid saturated fatty acid.
In addition, it is well known that carboxyl acid derivates (occurring in solution
when the oils decompose) coordinate strongly with Ca-sitesC8l. The adsorption of
carboxyl acid derivates on the cement particles might alter the hydration both by delaying
it and changing the morphology of hydration products. Adsorption of organic additives
are known to change the morphology of various materials such as CaC03 , gypsum and
hydrated cementc9•10•1ll. Encapsulation of carboxyl acid derivates might therefore have
changed the morphology of the hydrated cement phase in a way which has weakened the
strength of the paste. The morphology of later products may be more affected than early
ones since the concentration of fatty acid increases as the water is consumed.
No trends between degree of hydration and amount of oil added or oil type were
found (Table 2) even though the compressive strength decreased by oil addition. This fact
indicates that the oils do not interfere with the hydration of the cement paste, but rather
weaken the mortars possibly by lubricating the sand grains or forming hydration products
of inferior strengths.
The inner surface of the hydrated cement measured by drying the samples at
50°C and 105°C can be found in Table 2. The results indicate that the hydration of the
mortars have not been influenced by the addition of the vegetable oils. There is a scatter
of the results which make it difficult to extract any trends (inner surface against oil
dosage or oil type (amount of mono-unsaturated acids)). It can be noted that the inner
surfaces are higher for many of the mortars with oil than the reference mixes. This
phenomenon might be caused by the oils dispersing the cement grains better than for the
reference samples. Morphology changes caused by carboxylic acid salts might also have
resulted in a higher surface, albeit in this case, not necessarily a higher degree of
hydration.
The deviation between the top (including skin) and bottom samples were
reasonable (less than 5%) except for the samples Ref I, Rape 0.5% and Rape 1.5% which
have a deviation in the range of 12-16%. The general trend of compatible results for the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 423
top (including skin) and bottom part (including old sawn surface) of the mortar samples
indicates that the surface skin do not form a membrane against water causing reduced
hydration.
Chloride intrusion
Carbonation
The addition of the vegetable oils to the mortars increased the carbonation depth
compared to the reference samples as illustrated in Fig. 8. The carbonation depth seems
to increase with increasing oil dosage in some instances. There was no clear connection
between the amounts of mono-unsaturated acids in the oils and the carbonation depth.
The carbonation depth seemed however to increase with increasing amount of apparent
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
macro pores in the samples as illustrated in Fig. 9. This could mean that the dryer the
interior, the faster the carbonation. The share of apparent macro pores in the samples is
found to increase with the amount of mono-unsaturated acids in the oils as illustrated in
Fig. 5.
CONCLUSIONS
Vegetable oils can be used as water repellents for mortars in rather small amounts (0.5-
1.5% of cement weight) and the effect last for more than 3 years at moist conditions. Oils
with high content of mono-unsaturated fatty acids seem to be most effective; like olive,
peanut and rapeseed. The water suction was found to be dependent on the oil
concentration. The strongest effect was seen between the reference and the concentration
of 0.5% oil. Further increase of oil concentration resulted in further reduction in the water
uptake, but this effect was not as strong.
Neither 1-day nor 28-day strengths were significantly lower than the reference
mortar. The 3-year compressive strength was, however, approximately 30 % reduced
compared to the reference samples, but had still increased from 28 days. The strengths
did not show a clear dependency of oil dose which indicates that the dose of 0.5% oil has
already resulted in this strength level. The lesser strength gain might be caused by
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
424 Vikan and justnes
oxidation of unsaturated fatty acids in the oil as air is diffusing through the mortar
samples with time, morphological changes of the hydrated cement paste.
No trends between degree of hydration and amount of oil added or oil type were
found even though the compressive strength decreased by oil addition. This fact indicates
that the oils do not interfere with the hydration of the cement paste.
Chloride intrusion was reduced to 35-66% of the reference sample, but no trends
were found in-between the oils. The chloride intrusion profiles were not altered much by
the increase of oil concentration from 0.5-1.5% which indicates that 0.5% oil addition is
sufficient in reducing chloride penetration.
The addition of oils reduced in a substantial increase of carbonation depth
compared to the reference samples. The carbonation depth seemed to be correlated with
the amount of apparently closed macropores in the sample which increased with
increasing amount of mono-unsaturated acids in the oil.
REFERENCES
I. Gunstone, F., Fatty Acid and Lipid Chemistry, Blackie Academic and
Professional, London, 1st edition, 1996
2. Justnes, H., 0stnor, T.A., Vila, N.B., "Vegetable Oils as Water Repellents for
Mortars", Proceedings of the International Conference on Recent Trends in
Concrete Technology and Structures, Coimbatore, India 2003
3. Justnes, H., Thys, A., Vanparijs, F., "Reasons for increase in long-term
compressive strength of concrete by the set accelarator calcium nitrate",
Proceedings from the Kurdowski Symposium: Science of cement and Concrete,
Krakow, June 20-21,2001, pp.265-278
4. Sellevold, E.J., "Cured Concrete: Determination of air/macro and gel/capillary
porosity together with relative content of binder", Report NBI, 1986 (in
Norwegian)
5. Powers, T.C., Brownyard, T.L., Studies of the physical properties of hardened
Portland cement pastes, Bull.22, J.Am. Co ncr. Inst. 43 (194 7)
6. Albayrak, A.T., Yasar, M., Gurkaynak, M.A., Gurgey, I., "Investigation of the
effects of fatty acids on the compressive strength of the concrete and the
grindability of the cement", Cern. Caner. Res. 35 (2005) 400-404
7. Neville, A.M., Properties of Concrete, Pitman, London, 1977, p.64.
8. Otha, A., Sugiyama, T., Tanaka, Y., "Fluidizing Mechanism and Application of
Polycarboxylate-based Superplasticizers", SP 173-19, Proceedings of the 51h
CANMET/ACI International Conference on Superplasticizers and Other
Chemical Admixtures in Concrete, October 10-13, 2000, Nice, France, pp.211-
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
228
9. Hekal, E. E., Kishar, E.A., "Effect of sodium salt of naphatlene-formaldehyde
polycondensate on ettringite formation", Cern Caner Res 29 (1999) 1535-1540
10. van der Leeden, M.C. and van Roosmalen, G.M., "Aspects of Additives in
Precipitation Processes: Performance of Polycarboxylates in Gypsum Growth
Prevention", Desalination 66 (1987) 185-2000
11. Shen, Q., Wei, H., Zhao, Y., Wang, D.-J., Zheng, L.-Q., Xu, D.-F.,
"Morphological control of calcium carbonate crystals by polyvinylpyrrolidone
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 425
and sodium dodecyl benzene sulfonate", Colloids and Surfaces A. 251 (2004)
87-91
12. Blankenhorn, P.R., Evans, V.A., Kline., D.E., Cady, P.D., "Chloride penetration
of concrete impregnated with various linseed oil/mineral spirits combinations",
Cern. Caner. Res. 8 (1978) 565-570
Oil Type Dose(%) Ea.p llo.n llem :EE, I:Et.tat Poolid Pdry
Compressive Strength
s= 120
c, 100
c
Cll
!:: 80
Ul-
Cll ..,
l!iil0.50%
>a.. 60 D 1.00%
·- ::E
:l!- B 1.50%
...a.
Cll 40
E 20
0
CJ 0
0.0 17 20 22 31 38 56 78
Mono-unsaturated acids (%)
Fig. 1. Compressive strength of mortar samples as a function of oil dose after 3 years
moist curing.
3.50
3.00
2.50 -+-Ref
'E 2.00 ---- 0.5% Olive
~ 1.50 ......_1.0% Olive
1.00 -+-1.5% Olive
0.50
0.00
0 200 400 600 800
Sqrt(t)
Fig. 2. Water suction measurements for mortar samples with o-1.5% Olive oil after
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3 years moist curing. All samples have been pre-dried at 50°C.
Bil0.5%0il
01.0% Oil
l!'.l1.5% Oil
0 17 20 22 31 38 56 78
Mono-unsaturated oils (%)
Fig. 3· Absorption rate of sawn upper mortar surfaces as a function of oil dosage and
content of mono-unsaturated acids.
25~------------------------~
0 0.5 1 1.5
Dose(%)
Fig. 4- Distribution of pores in mortar samples without oil and with 0.5-1.5% olive oil.
The total porosity is approximately constant for all samples, while the share of
closed macro pores increases and the share of suction pores decreases with
increasing oil dose.
10.0
8.0 .. .. . ..
--
~ 6.0 .. ..
~
-;;
0
0
4.0 ...• [• Eom[
.lEcm
0.. • •
2.0 • • •
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.0
0 20 40 60 80 100
Monounsaturated fatty acids (%)
Fig. 5- Correlation between open (Eom) and closed (Ecm) macro pores and amount of
mono-unsaturated acids in the oils.
1.2 . - - - - - - - - - - - - - - - ,
- 1 ~----------------------~
'$.
";;' 0.8 -H:Irr------------------------l --Ref
.2 --0.5% Olive
f 0.6 +---+~----··--------------·--·-----! ___._ 1.0% Olive
1:
8 0.4 +--~~-------------------l --1.5% Olive
c
8 0.2 +---~~"-<;;::::;:-----------l
Fig. 6. Chloride intrusion profile for mortar samples with o.s-1.5% Olive Oil.
ISI0.50%
.1.00%
rn 1.50%
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0 17 20 22 31 38 56 78
Mono-unsaturated acids (%)
Fig. 8. Carbonation depth (mm) of mortars as a function of oil dosage (%) and content
of mono-unsaturated acids in the oils.
1.5% Oil
~ 10
..I!!
~
8
0
Cl. 6
e
...u
E
4
..
...
"0
0
(j
2
0
40 60 80 100 120
Carbonation depth (mm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
KATO Hiroyuki
Manager
Special Additive International Dept., DENKI KAGAKU KOGYO Co.Ltd.
1. INTRODUCTION
There are a many published reports on the effectiveness of expansive additives in
preventing cracks due to dry shrinkage, autogenous shrinkage, and thermal stress. t)
In cold climates and areas with heavy snow, road and highway structures are
subjected to chloride based deicing chemical in order to prevent wheel slipping
during freezing conditions. While studies on the resistance of expansive concrete to
seawater have been performed2l, little research has been conducted on the effects of
deicing salts. This study investigates the damaging effects of such salts on concrete
prepared with low dosage of expansive additives. Both on ordinary portland cement
and a blast-furnace slag cement were investigated with and without the additive to
determine whether the addition of the expansive additive had any affect on the
chloride resistance of concrete.
2. EXPERIMENTAL
2.1 Materials and Mix Composition
0--Materials
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A normal portland cement and a blast-furnace slag cement were used in addition to an
expansive additive based on free CaO hydration as well as ettringite formation. Table 1
shows the chemical composition and physical properties of the expansive additive and
cement used. A crushed sand (2.55g/cm3), and crushed stone from the Hime River
(2.65g/cm\ a lignin based water-reducing agent and a sulfonic acid hydrocarbon AE
auxiliary were also used.
2)--Mix Proportions
Table 2 shows the mix proportions of concrete, while Table 3 shows the properties of
fresh concrete mixtures. All concrete mixture were designed to contain 0.5 water/cement
ratio and 4.5±1.5 air.
2.2 Curing
Concrete mixing and molding was according to JIS Standard A 1138:1998. The
specimens were cured at 20°C, 80% R.H. conditions for one day in molds and then
removed from the molds, and cured in 20°C water until 28 days. The top and bottom
spray), 88 days (60 days after curing in salt spray), and 119 days (91 days after curing in
salt spray) after initial curing.
5)--Measuring Pore Size Distribution
Mercury Penetration Porosimeter was used to measure pore size distribution at 56 days
(28 days after curing in salt spray), 88 days (60 days after curing in salt spray), and 119
days (91 days after curing in salt spray) after initial curing. To test the effect of salt attack,
an analysis was performed on the surface of those test pieces that were deemed to have
been affected. Those deemed not to have been affected were inspected in the middle.
6)--Microstructural Observation
Scanning Electron Microscopy (SEM) was used to determine the microstructural
changes. The SEM observation were made at 56 days (28 days after curing in salt spray)
after initial curing. Note that the SEM observations were only made on mixtures with
ordinary cement.
4. CONCLUSION
This study involved testing of the chloride resistance of expansive-cement concrete
mixtures made from both a normal portland cement and a portland blast-furnace slag
cement to which a low dosage (6 % of the total cementitious material) of an expansive
additive had been added. The additive derived its expansive characteristics foam both free
lime hydration and ettringite formation. Calcium chloride spray was used as a source of
chloride ions. Based on test results from measurement of changes in length, compressive
strength and chloride penetration depth, it was concluded that there was essentially no
adverse effect of the expansive additive on the chloride resistance.
Referrence
1. Japan Concrete Institute, Symposium Committee Report on Enhanced Functionality
and Durability of Expansive Concrete in Construction, 2003,
2. Yukio Sasagawa, Masaaki Mashimo, Tsutomu Fukute, Etsuo Sakai, Durability of
Expansive Concrete Dipping Sea-Water at Long Ages, Proceeding of the Japan
Concrete Institute, 1998, pp.l57-162, Vol.20, No.2
3. Nobuyuki Tanno, Tiong-Huan Wee, Etsuo Sakai, Yoshiharu Watanabe, Influence of
Various Mineral Admixtures on Penetration of Chloride Ions in the Tropics,
Proceeding ofthe Japan Concrete Institute, 1996, pp.375-380, Vol.l8, No.1
4. Akihiro Hori, Mitsuo Takahashi, Tsuji Yukikazu, Physical Properties of Concrete
with Low-Added Type Expansive Additive, Proceeding of the Japan Concrete
Institute, 2002, pp.261-266, Vol.24, No.1
5. Akihiro Hori, Hideaki Gomi, Takashi Torikoshi, Yukikazu Tsuji, A Study on
Durability of Concrete Wall with Low-Added Type of Expansive Additive,
Proceeding of the Japan Concrete Institute, 2003, pp.659-664, Vol.25, No.1
6. P.K. Melta and P.J.M. Monteiro, Concrete: Microstructure, Properties, and Material,
McGraw-Hill, pp.650, 2005
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
NC 310 20
817 1021 165 825 0.66
B 330 0
BC 310 20
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Photo t--Cross section of the test concrete ready for EPMA Analysis.
BC (slagcement+
B (slag cement)
expansive additive)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
N C (ordinary cement+expansive additive)
~
~100
14
Age (days)
~ 40
~ 30
·~ 20
~
~ 10
0
u
0~~~~--_.--~~~~
0 10 20 30 40 50 60
Age (days)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
g
.s 0.04
inner
0.01 ..........~~......-~.......~~......--~._~-...
0.001 0.01 0.1 10 100
Pore Diameter ( ~tm)
0.1 r---..-----.------.---.----.
Ordinary Portland Cement
+ Expansive Additive
g
E
~ 0.04 surface
"E
-8
>
~ 0.02
0.1
Pore Diameter C.um)
0.1 r----.----.-------.---.----,
Blast Furnace Slag Cement
! 0.04
surface
"§
0
>
~ 0.02
"-
innner
0.01
0.001 0.01 0.1 10 10(
Pore Diameter (~tm)
0.1
Blast Furnace Slag Cement
+Expansive Additive
!!!
s 0.04 surface
"E:>
0
>
~ 0.02
"-
innner
0.01
0.001 0.01 0.1 10 10(
Pore Diameter (~tm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-E'
l 0.04
!5" 0.02
~ 0, 1>., D
~ 0.01 •• A, •
o, • : 28days
: 60days
"'· ,A
0 . : 91days
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Ordinary Portland Cement
+ Expansive Additive
~
l 0.04
!5" 0.02
0
> 0, 1>., D
~ 0.01 •• A, •
0, e : 2Bdays
1>., A : 60days
0, • : 9ldays
l-E' 0.04
!5" 0.02
0
> 0, 1>., D
~ 0.01
e, A , .
0, e : 2Bdays
1>., A : 60days
D, • : 9ldays
Q>
l 0.04
!5" 0.02
~
~ 0.01
..
0, 1>.,
0, •
A, •
D
: 2Bdays
: 60days
"'·
D,. A : 91days
Fig. 4--Relationship Between Salt Spray Duration and Pore Size Distribution.
ACI member Jean Pera is a Professor of Civil Engineering at the National Institute of
Applied Sciences of Lyon, from which he received his Dr Eng and Doctor of Sciences
degrees. His research interests include innovative materials for Civil Engineering and
Waste Management.
INTRODUCTION
Some concrete slabs curl upward at their comers and perimeter. This condition is
particularly common with slabs on grade [1, 2]. When this occurs, comers and edges are
unsupported and tend to break under load. This phenomenon is specially marked in
screeds which are not bound to their support. As the acoustical regulations become more
and more stringent in Europe, this phenomenon will appear more often. The new
regulations assign the disposal of a layer of insulating material between the support slab
and the screed. Therefore, the bond between the support and the screed is broken, and
the last one is promoted to curl.
This phenomenon is mainly due to the moisture gradient that appears within the
thickness of the screed. Curling is caused by differential shrinkage between the top and
the bottom of a slab or a screed, mainly drying shrinkage. The top surface dries and
shrinks, while the bottom stays wet and undergoes little change in dimensions [3]. The
curling of screeds utilizing normal portland cement is very important and, therefore, the
development of these screeds is limited. One solution limiting curling is to use calcium
sulfoaluminate cement, as shown in a previous paper [4]. Such cement presents lower
drying shrinkage than normal portland cement.
EXPERIMENTAL
In a first step, an experimental study was undertaken, using an original device, and
compared the behavior of screeds based on normal portland cement (NPC) and calcium
sulfoaluminate cement (CSA), respectively. The mixture proportions of screeds are
shown in Table 1. OPC was a CEM I 52.5 Type according to the European Standard EN
197-1. The CSA cement was composed of 80% CSA clinker and 20% phosphogypsum.
The composition of CSA clinker was as follows:
The fluidity of the mortar was measured by means of the static spread a truncated
cone having the following dimensions: 0inf = 95 mm; 0sup = 55 mm; h = 70 mm. Mortar
was considered as self-levelling when the spread reached 250-270 mm.
In a second step, the influence of polyether polyol supplied by Bayer on the drying
and curling of CSA screed was investigated. Its typical physical properties are presented
in Table 2.
Measurement of Curling
mortar.
b) and, a balance following continuously the mass loss of the whole system.
Twenty-four hours after casting, the slab was equipped with four vertical LVDT
sensors placed in the comers and measuring the vertical displacement, and two horizontal
LVDTs measuring the shrinkage occurring at the center of the slab. The instruments
precision is 0.5%. These informations were automatically recorded each hour. The
equipment was stored at 20°C and 50% RH.
The internal relative humidity (RH) was measured by means of two sensors encased
in Gore-Tex sleeves respectively placed at 5 mm from the top and the bottom of the
screed (Fig. 2). The accuracy of the sensors is reported by the manufacturer to be± 0.5%
between I 0% and 95%. The Gore-Tex allows vapor transmission while preventing the
penetration of liquid moisture and ions that could invoke erroneous measurements.
Compressive Strength
The mass loss of screeds is presented in Fig. 3. At 28 days, the mass loss of CSA
screed was 25% lower than that of NPC one. This can be explained by the higher
consumption of mixing water necessary to produce ettringite according to the following
equation:
C4A3 + 8C H2 + 6CH + 74H Q 3C6A 3H32
with: C = CaO; A = Alz0 3 ; = S03 ; H= H 20.
This hydration reaction may explain the lower value of drying shrinkage observed in
the CSA screed. After 28 days of curing at 20°C and 50% R.H., the recorded values
were:
- NPC: 800 ~m/m,
- CSA: 480 ~m/m.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The curling of screeds is shown in Fig. 4. After 20 days of hydration at 20°C and
50% RH, the curling of CSA screed was 3 times lower than that of NPC one. This result
is mainly due to a smaller gradient in the distribution of internal relative humidity within
the thickness of the CSA screed.
RH measurements are reported in Fig. 5. RH gradient was always smaller than 13%
in the CSA screed, while it ranged from 19 to 37% in the OPC screed. Otherwise, the
CSA screed dried slowly than the OPC one.
Polyether polyol was introduced in the recipe of the CSA screed at a content of 2.5
kg/m 3, which is 0.63% of the cementitious content (CSA + NPC +limestone powder).
Influence on Drying
The addition of polyol has not any influence on the mass loss of the screed, as shown
in Fig. 6. The kinetics of drying is not modified and the value reached after 28 days of
drying is similar: 6.6% instead of 6.4%. The same gradient of RH within the thickness of
the screed was observed in presence or in absence of polyol: < 13%. The drying of the
CSA screed is not affected by the addition of polyol.
The addition of polyol considerably reduced the drying shrinkage of the screed, as
observed in Table 4. After 28 days, shrinkage is reduced by 40%. Fig. 7 shows that the
kinetics of shrinkage is drastically reduced by the addition of polyol.
Influence on Curling
The kinetics and final value of curling are reduced by the addition of polyol, as shown
in Fig. 8. After 20 days of drying, the vertical displacement measured at the comers is
reduced by 23%.
The pore size distribution was assessed by means of mercury intrusion porosimetry on
samples of screeds dried at 40°C until constant weight. The results obtained on the three
types of screeds (NPC, CSA, CSA + polyol) are shown in Fig. 9. Polyol addition did not
modifY the pore distribution of the CSA screed: this remained bi-modal. CSA screeds
contained more pores in the range ofO.l J..Lm to 3.15 J..Lm than NPC screed. Defining three
families of pores [6]:
-macropores:> 0.6 J..Lm,
- mesopores: 0.1 to 0.6 11m,
- micropores: 0.006 to 0.1 J..Lm,
allows to conclude that CSA screed, with or without polyol, contains more mesopores
and macropores than NPC screed (Fig. 10). These larger pores can explain the lower
gradient of moisture observed in CSA screeds: the movement of water is easier in such
system and more homogeneous. Such porosity is induced by interlocked needles of
ettringite as shown in Fig. 11. Some of these needles are hollow.
CONCLUSIONS
The results of the investigation work reported in this paper can be summarized as
follows:
1. It is possible to drastically decrease (by three times) the curling of screeds which
are not bound to their support, using calcium sulfoaluminate cement instead of normal
portland cement.
2. The drying of CSA screed is more homogeneous and the R.H. gradient within the
thickness of the screed is two or three times lower than in NPC screed.
3. The drying shrinkage is reduced by 50%.
4. The compressive strength of CSA screed is remarkable at early age and allows a
rapid utilization of the hardened surface.
5. When polyol is added to the mixture (0.63% of the cementitious material content),
curling is still reduced by 23%.
6. Polyol reduces drying shrinkage by 40%, but does not affect the mass loss of the
screed and the porous distribution.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
Shrinkage (J.trnlm)
Days
Withpolyol Without polyol
1 33 77
7 220 340
28 300 480
9
8
/
7
6
5
/
,P--
--a
/
?ft. 4 ,
3 1-NP:~
2 1-c:J -CSA
0
0 10 20 30
Days
0,45
0,4
0,35
I 0,3
i
0,25
0,2
6 0,15
0,1
0,05
-
~
:I:
80
60
a: 40
20
0
0 10 20 30
Days
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
NPC
-<>--bottom
120 - . - - - - - - - - - ! -·-top
100~-~------------L_-_ _
--_ _RH~g~
__d~~nt~
80
.
60 --.-..-• ...:--.-.::::~~-:-a:-.;;.-:;:-;-:;--~
0-~-,----T----r----r--4
0 5 10 15 20 25
Days
/
/
( -withoutpolyot } - -
0
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0 10 15 20 25 30
Days
10 15 20 25 30
Days
0,14
0,12
0,1
e.§.
0,08
i§ 0,08
...
.!!
Q
Q,G4
0,02
10 15 20 25
Days
0,009
0,008
~ 0,007
!§ 0,006
0 0,005
>
1 o,o04
~ 0,003
0,002
0,001
50
40
35
3D+----==---
.,. 25
20
15
10
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
In the present study, recycled fine and coarse aggregates were produced by
crushing demolished concrete. These recycled aggregates have particles of natural
aggregate to which hydrated cement paste adhere, and also hydrated cement paste particles
themselves. As compared with recycled coarse aggregate, recycled fine aggregate has a
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
greater amount of hydrated cement paste particles. These cement pastes influence the
quality of the recycled aggregates, and, in tum, influence the properties of concrete
containing the recycled aggregates [2, 3, 4]. However, the manner in which the quality of
recycled fine aggregate influences the properties of concrete remains unclear. For this
reason, recycled fine aggregate has been employed only for low-strength concrete.
Clarifying the influence may enable to increase the amount of recycled fine aggregate used
in concrete. When recycled concrete is for construction, understanding the influence of
recycled fine aggregate on durability of concrete is an important issue.
TEST PROCEDURES
Physical properties of the aggregate used in this study are summarized in Table 1
Three natural sands were used as typical natural fine aggregates; i.e. a river sand from
Japan (NR), a river sand from China (NRC), and a lime sand (NL). Four recycled fine
aggregates, called RSP, RS 1, RS2, and RS3, were used. The source of the RFP was various
demolished concretes brought into a recycling plant. The sources of RF 1, RF2, and RF3
were concrete produced by demolishing 45-, 70-, and 44-year-old buildings, respectively.
For all demolished concrete, particles having diameters less than 20 mm were produced by
crushing each demolished concrete with a jaw crusher and a cone crusher. Subsequently,
particles ofO.IS-2.5 mm in diameter were extracted from the produced particles and used
as RFP, RF 1, RF2, or RF3.
The ordinary cement, fine aggregate, and coarse aggregate were mixed in a
mixer for 30 seconds. Subsequently, the appropriate amount of water was added, and
mixing was performed for another 3 minutes. A suitable quantity of admixtures, such as a
polycarboxylic acid type superplasticizer, a ligninsulfonic acid type air-entraining agent, a
water-reducing admixture, and a supplementary air-entraining agent, were added to the
concrete in order to adjust properties of fresh concrete to the target values. The target
values were a slump of 120 mm and an air content of 5%.
The tests specimens for carbonation were 100 x 100 x 400-mm prisms that had
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
TW = W + s x Ds x ws + g x Dg x wg (1)
where, TW is total water content (kg/m 3), W is water content (kg/m 3), s is content of fine
aggregate per unit volume (lim\ Ds is density of fine aggregate in the oven-dried
condition (kg/1), ws is water absorption of fine aggregate, g is content of coarse aggregate
per unit volume (kg/1), Dg is density of coarse aggregate in the oven-dried condition (kg/1),
and wg is water absorption of coarse aggregate. Additionally cement to total water ratio
( CITW) is defined as the ratio of cement content to TW.
Figure 2 shows the relationship between compressive strength and C/TW. A plot
of all experimental data points approximates a straight line with a correlation coefficient of
0.96 (Fig. 2).
This result shows that the strength of concrete are same independently of the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
These results indicate that the amount of water that is not consumed in the
hydration reactions and physical absorption relates to shrinkage of cement paste in the
dried condition. Therefore, in the case of fine aggregate with high water absorption, the
shrinkage of the concrete becomes large, because the water absorbed to the aggregate
migrates to the paste and the unconsumed water were increased.
Figure 7 shows relationship between C/W and the depth of carbonation for
specimens maintained in an atmosphere of 5% C0 2 for 26 weeks. For concrete containing
natural fine aggregate, the depth of carbonation is 0 when C/W ;;;; 2.5. In the case of
recycled fine aggregate, the depth of carbonation is 0 when C/W = 4 too. The influences of
W/C and water absorption were found in the case where C/W ;;:; 2.5. Therefore, the
influence of C/TW on depth of carbonation was investigated, and the results are shown in
Figure 8. When CITW is less than 2, the relationship between carbonation depth and C/TW
approximates a straight line, independent of the type of fine aggregate, with a correlation
coefficient of0.953. Influence of the water absorbed into the aggregate on the carbonation
depth of concrete.
Trinker stated that pore volume of 0.2 - 2 Jlm diameter influences frost damage
of concrete [9]. Therefore, these results indicate that the water absorbed by the aggregate
influences the amount of pore volume in the paste and concrete is given a serious frost
damage independently of the kind of fine aggregate.
In addition, as the results, it was, also, found that the ratio between cement
content and TW, which is the water content added to water absorbed by the aggregate, can
be used as an effective index for estimating strength, shrinkage, carbonation depth, and
durability factor of concrete containing recycled fine aggregate.
Conclusions
In this study, the properties of the hardened concrete with natural and recycled
fine aggregate, at water to cement ratios (W/C) of0.25 to 0.7, were investigated in order to
determine the influence of quality of aggregate on the properties of concrete. The results
lead us to the following conclusions.
1) If total water content (TW) is defined as the water added to water absorbed by the
aggregate, the ratio between cement content and TW can be used as an effective index
for estimating strength, shrinkage, carbonation depth, and durability factor of concrete
containing recycled fine aggregate.
2) When recycled fine aggregate with high water absorption is used, the shrinkage of
concrete becomes large, because the water absorption of the recycled fine aggregate
contributes to the mixing water which is not consumed in hydration reactions and
physical absorption
3) For concrete with recycled fine aggregate, volume of gas permeating into concrete
increases, because pore volume of hard paste increases with migration of water
absorbed into aggregate. Here, pore volume in recycled fine aggregate does not relate
to gas permeability of concrete.
4) The water absorbed into aggregate influences pore volume in the paste, and durability
factor decreases drastically when C/TW ~ 1.2.
References
2. Sri Ravindrarajah, R. and Tam, C. T.," Methods oflmproving the Quality ofRecycled
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3. Kasai, Y., Hisaka, M. and Yanagi, K., "Durability of Concrete Using Recycled Coarse
Aggregate," Proceedings of the Second International RILEM Symposium on the
Demolition and Reuse of Concrete and Massonry, Tokyo, Japan, November 1988, pp.
623-632, 1988
4. Chen, H. J., Yen, T. and Chen, K. H., "Use of Building Rubbles as Recycled
Aggregates," Cement and Concrete Research, Vol. 33, Issue 1, pp. 125-132,2003
5. Uchikawa, H., Uchida, S. and Hanehara, S., "Research on the composition and
structure of hardened cement paste, mortar and concrete related to their performance,"
2nd International symposium on cement and concrete, Beijing, China, Vol. 1, pp.
314-330, 1989
9. Trinker, B.D., Shvirayev V.A., Volodina A.Y., Denissov A.S., Boot Y.M., Kolbassov
B.M., Berlin L.E., Mchedlov-Pertrosyan O.P., Melnichenko P.A., Olginsky A.G.,
Ataroselsky A.A., "Methods of Structure Research in Estimating the Life of Heavy
Concrete, RILEM Symposium Durability of Concrete, Part I, pp A-69- A-86, 1969
Density in
Water
oven-dried
Kind of aggregate Symbol absorption Source of raw material
condition
(%)
cm3
NR 2.52 2.74 River sand from lbi river, Japan
(weight ~ercent}
Supplementary
air-entraining agent•2 Cx0.0035 Cx0.0005 Cx0.0035 Cx0.0035 Cx0.0035
(weiggt ~ercent}
Target value of slump or slump flow 600±50m
18±2 em 12 ±2 em 12±2 em 12±2 em
(mm} tn
Target value of air content (%) 3±1 3±1 5±1 5±1 5±1
*I Ligninsulfonic acid type air-entraining agent *2 Polycarboxylic acid type super
plasticizer
Here, this mixture proportions are with NSI, a crushed stone, and an ordinary portland
cement.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Substitution rate of
Water to
Durability-test Used recycled fine
cement ratio
number aggregate
(%)
25 RF1,RF2 0, 50,100
35 RF1,RF2 0, 50,100
50 RF1,RF2 0, 50,100
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0.5 0 292
0.6 0 280
0.6 50 331
0.7 0 300
120
Age: 28 days in water
100 0
~
ONR eRFP
l'.NRC .A.RF1
~ 80 ONL .RF2 I
1
·~"
60
+RF3
~
0
!
!
i
u
40
:II
20
0
1 1.5 2 2.5 3 3.5 4 4.5 5
CIW
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~ 80
R=0.981
~
~ 60
.:::"
u
I 40
20
ONR 8RFP
.6-NRC .A.RFI
<)NL .RF2
+RF3
crrw
Fig. 2--Relationships between compressive strength and cement-to-total water ratio.
14~------~-------r-------,
Age: 7 months
0
eNR
ORF3
• 0
•
•
4~------~------~------~
I 2.5
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1.5 2
C/TW
eNR
0RF3
Correlation coefficient
R=0.948
4~------~------~~------~
1 1.5 2 2.5
crrw
15
.:::;--
Age: 48 weeks
"1--
~ in dried condition ...
~
..,
12
]
I
•
!: 9
~
I" e 0
•§ 6 I !;,.
8..., ~ ONR eRFP
'
.6NRC ARFl
f
00
ONL .RF2
0
20 30 40 50 60 70
W/C
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
f"' 3
0
ONL •RF2
20 30 40 50 60 70 80 90 100
TW/C
40
Age: 26 weeks in 5 % CO condition
2
Ig 30
• ....
LNRC .6.RF1
ONL .RF2
·~
~ 20 •e ....
0
'a ,0. I
i0 10 ~
....
I
0.__........_ _._-tt-........-..J----11---'-----1 ...
1 1.5 2 2.5 3 3.5 4 4.5 5
C/W
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
I 30 • ONR eRFP
.t,.NRC .a.RFl
0NL .RF2
§
'i Y = 63.8 - 31.6 X
c::
~t.>
20 Correlation coefficient
R=0.950
'S
-s
Q
fr
10
f
0
0.5 1.5 2 2.5 3 3.5 4
CITW
100
80
ewJc=50%
.A.W/C=60%
20 •w/C=70%
0 20 40 60 80 100
Substitution rate of recycled fine aggregate
(vol. %)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
5
100
80 ... •••• •
:g 60
-~ •
.&
J 40
• ewtc=50%
..&W/C=60%
20
• •wtc=70%
0
0.5 1.5 2
crrw
Fig. 10--Relationships between durability factor and cement-to-total water ratio.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
one more or less easy to implement, have been proposed to determine this important
point when dealing with low water to binder ratio (W /B) concrete. It is at this moment
that external water curing must be applied to control the detrimental effects of
the development of early autogenous shrinkage strains because the development
of early cracking can be very detrimental for concrete durability. Concrete is
continuously subjected to volumetric changes, particularly at an early age, when
hydration heat and autogenous shrinkage evolve very fast. It is difficult to separate
thermal and plastic shrinkage from shrinkage due to chemical contraction. In large
concrete elements, it is appropriate to calculate isothermal shrinkage to reach the
starting point of autogenous shrinkage. Neither the temperature criterion alone,
nor the penetration resistance test can predict the starting point of autogenous
shrinkage, and therefore, the risk of early-age cracking. In this research, the concept
of threshold of solidification was coupled with that of temperature rise to more
accurately determine the starting point of autogenous shrinkage.
An experimental study on the development of isothermal shrinkage of large concrete
elements made with different binders was carried out. In order to clarify the effect of
theW /B on the importance of the determination of the starting point of autogenous
shrinkage, concrete mixtures were made at three W/B: 0-45, 0.35 and 0.26.
Isothermal shrinkage was measured using vibrating wire gauges imbedded in a
large concrete element. Experimental results show the importance of the correct
determination of the starting point of autogenous shrinkage when the W/B is low.
ACI member Nikola Petrov is Adjunct Professor in the Department of Civil Engineering
at Universite de Sherbrooke, Sherbrooke, Quebec, Canada. His research interests include
expansion due to delayed ettringite formation in heated mortars and concretes.
INTRODUCTION
Field observations and laboratory studies have shown that concrete with a low
W/B concrete is more sensitive to early-age cracking than normal concrete. At early age,
the principal volumetric changes are: plastic shrinkage, thermal dilation, and chemical
shrinkage due to hydration reactions that can result in autogenous shrinkage (AS). ACI
116R standard defines AS as "change in volume produced by the continued hydration of
cement, exclusive of effects of applied load and change in either thermal conditions or
moisture content".
It is well known that AS of normal concrete (W/C > 0.40) is much lower than
drying shrinkage, so that it could be ignored for practical purposes. However, for high
performance concrete (HPC) having a low W/B, AS can be significant, especially with
binders containing silica fume (SF). Based on Powers' results on Portland cement
hydration, Ai'tcin et al. 1 have shown that a W/C of 0.42 can be considered the cut-off
value to determine concrete sensitivity to the development of AS. The lower theW/Cor
WIB value, the faster AS will develop.
Autogenous shrinkage can be measured in the laboratory on small samples under
isothermal conditions. However, from a practical point of view, this is not always
appropriate, in particular in the case of large structural elements. In fact, the volumetric
variations measured in such cases contain AS, but also plastic shrinkage and thermal
dilation. The effect of temperature during cement hydration can be problematic for an
adequate measurement of AS since volumetric variations due to the release of heat during
cement hydration must be evaluated and taken into account. Moreover, as the rate of
cement hydration is temperature-dependent, what is measured is actual shrinkage rather
than autogenous shrinkage that per definition requires isothermal conditions. The
monitoring of temperature development during the hydration process is therefore
essential to follow the evolution of shrinkage during hydration and calculate the
equivalent isothermal shrinkage.
It is well known that in the absence of internal or external water curing, concrete
starts to shrink as soon as hydration progresses. As a consequence, it is very important,
on the one hand, to begin shrinkage measurements immediately after the casting of
concrete, and on the other hand, the precise determination of the starting point of AS is
crucial in order to separate the different shrinkage strains that develop in the liquid phase,
from those that develop in the semi-liquid or even solid phase.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
EXPERIMENTAL PROGRAM
Materials
Three binders were used for this study. A Type 10 Canadian cement, equivalent
to ASTM Type I normal Portland cement (containing less than 5% limestone filler), was
selected as reference cement. A binary binder containing 8% SF and a ternary binder
containing 22% FA and 5% SF were also used. The chemical composition and physical
properties of these binders are given in Table 2.
A crushed limestone with a maximum size of 14 mm was used as coarse
aggregate. The fine aggregate used was natural siliceous sand having a fineness modulus
of 2.4. In order to obtain the targeted slump value, a polynaphthalene sulfonate
superplasticizer (PNS) and lignosulfonate plasticizer were used in all concrete mixtures
having a W/B of0.35 and 0.26. For concrete having a W/B of0.45, only a lignosulfonate
plasticizer was added.
200±20mm.
The 0.45 W/B concrete is a reference concrete that was made with a Type 10
cement. The cements used for the concretes having a W/B of 0.35 and 0.26 were the
Type 10 with SF and the ternary binder with SF and FA, as mentioned above.
The concretes were mixed in a mixer having as maximum capacity of0.40 m 3 •
Testing method
Shrinkage was measured on rather large concrete elements measuring 500 x 850
x 900 mm. Such dimensions were selected to represent a small wall. In this type of
element, the effect of thermal deformation cannot be overlooked as in the case of small
samples currently used in laboratory tests. In fact, in such elements the maximum
temperature recorded exceeded 60°C depending on the type of binder used, compared to
the 20°C initial temperature of concrete.
Temperature and strains were measured at the center of the element using a 127
mm-long vibrating wire gage embedded horizontally at the center of each element (Fig.
2). The vibrating wire gages used in this study, described in detail by Kada et al. 6 had a
thermocouple located at their center, making it possible to follow temperature variations
during hydration and calculate the deformation induced by the temperature change. The
vibrating wire used in this investigation can measure shrinkage strains from the plastic to
the hardened state. It was connected to a computer-controlled data logger immediately
after the casting of concrete. Temperature and shrinkage strains variation with time were
recorded.
All the concrete elements were demolded after 24 h and covered with a wet
burlap and a plastic foil in order to avoid early drying shrinkage and ensure a continuous
Importance of shrinkage
Figures 3a and 3b show the development of isothermal shrinkage over time
during the first 7 days, as measured at the center of the element for the studied concretes.
The influence of the cement type on the development of isothermal shrinkage is shown in
Figures 4a and 4b.
Aside from the lower WIB of HPC, when compared to conventional concrete
(Figures 3a and 3b), the importance of shrinkage of HPC increases significantly due to
the use of ultra-fine supplementary cementitious materials such as SF (Figure 4b ). Silica
fume refines the pore system of concrete, thus the amount of finer pores is much greater
then that of coarser ones. During the hydration process, water movement from capillary
pores to anhydrous particles leads to the creation of menisci in finer pores. As a result,
menisci appear in pores of smaller size. Due to the extremely small size of SF particles
(average diameter of 0.1 J..Lm), the pores within a cement paste containing this
supplementary material rapidly become disconnected by the filler effect of small
particles. Consequently, in low W/B concretes water transport through the network of
disconnected pores becomes practically impossible.
It can be seen that shrinkage is low in the case of conventional concrete (W/C =
0.44) compared with that of concrete with low W/B, due to a W/B greater than 0.40 and
the absence of ultra-fine supplementary cementitious materials (Figures 3a, 3b, and 4a).
Under the conditions described, the shrinkage of concrete made using portland
cement (TlO) develops during the first 24 hours and is ofthe order of 130 J..Lmlm for the
two W/B examined. Curing water, applied upon demolding, quickly saturates the
connected pores of such concrete and stops the development of shrinkage.
In concretes containing SF, 50% of shrinkage develops during the first 24 hours, so that
the residual shrinkage compared to the shrinkage at 7 days is only 80 J..Lm/m for 0.44 W/B
and 130 !!mlm for 0.35 W/B.
The autogenous shrinkage developed during the first 24 hours in the concrete
made with the ternary binder increases with the decrease in W/B. This increase of the
magnitude of shrinkage is estimated to be from 20% to 50% when the W!B decreases
from 0.43 to 0.26. The shrinkage that can be determined after demolding is equal to 100,
120 and 150 J..Lm/m in concretes having a W/B of0.43, 0.35 and 0.26, respectively.
CONCLUDING REMARKS
The results reported here on the ASS of concrete made with different binders
measured on large concrete elements using vibrating wires led to the following
conclusions:
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
3. Tazawa, E., Sato, R., Sakai, E., and Miyazawa, S., "Work of JCI Committee on
Autogenous Shrinkage", in "Shrinkage of Concrete, Shrinkage 2000",
Proceedings of the International RILEM Workshop -PR017-, edited by
Baroghel-Bouny, V. and Aitcin, P.-C, Paris, France, 16-17 October 2000, pp.21-
33.
5. Roussel, S., "Influence d'une cure humide a jeune age sur le developpement du
retrait endogene des BHPs", Rapport interne, Universite de Sherbrooke, 122 pp,
1996.
6. Kada, H., Lachemi, M., Petrov, N., Bonneau, 0., P.-C, Aitcin., "Determination
of the Coefficient of Thermal Expansion of HPC from Initial Setting", Material
and Structures, Vol. 35, 2002, pp.35-41.
8. Neville, A. M., "Properties of Concrete", 41h ed, Longman, London, 1995, 844
pp.
Specific Blaine
gravity (m2/kg)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Binary
0.44 357 157 748 1004 0.94 -
0.35 427 149 711 1046 0.98 4.48
0.43 378 162 752 1010 0.87 -
Ternary 0.35 418 146 696 1024 0.96 3.43
0.26 480 125 720 1100 1.0 6.0
*WR = water reducer
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
12 15 19 22
nme(h)
4.0
3.5
3.0 ::c
~ -100 2.5 i3
l,..
E
a -1so 2.0 ~
;::
t" -200
1.5
1.0 :I
...
.
i -250 0.5 ii "'
! -300
0.0 ~
.{),5 r:."
-350
-1.0
400~----------------------------~ -1.5
2.7 hours
(C)
00 3.0
e -1oo
'E
a -2oo
~
'.il
~-300
~ 400
-500
Fig. 1 - Determination of the threshold of solidification from the thermal flux and the
total shrinkage of concrete used in a large size element.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
482 Meddah et at.
Time (days)
0
4 5 6 7
-50
~
:0 -100
..= 0.44
·2 -150
~
-;; -200
.s.... -250
E
0.35
:l
-300 HSF
-350
Time (days)
0
4 5 6 7
-50
~
~ -100 0.43
~ -150 0.35
-;; -200
.s.... -250
e
:l 0.26
-300
Tm
-350
Figs. 3a and 3b- Development of shrinkage over time for two binder types and
various W/B.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
·c~ -150
..=
.., TIO
;; -200
..a
i-250
5l
..... -300
W/B=0.43
-350
Time(days)
0
2 3 4 5 6 7
~ -50
~-100 TIO
.=
~ -150 '
' TER
" -200 --------
;;
..a
~ -250 HSF
=
.,!;1-300
-350
W/B = 0.35
Figs. 4a and 4b- Development of shrinkage over time for two W/Band various
binder types.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-;
a initial setting: slightly over-estilmted
~ -200
i5
~
...... -250
Tl0-0.45
-300
Time (days)
3 4 5 6 7
~ -100
..
1011 ---demolding- '------
: under-estilmted
.=
] -200
-- 150
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-300 300
~
.=c -400
-SOO TER·0.26
Figs. sa and sb- Shrinkage ofT to and ternary concrete calculated from the onset of
setting, the onset of the threshold of solidification and the time of demolding.
Y. Monnin is a PhD student in civil engineering department of the Ecole des Mines
de Douai.
INTRODUCTION
The diagnosis of Alkali Silica Reaction (ASR) in concrete is difficult and now uses
different methods such as :
the ultraviolet fluorescence of gels, where calcium or alkalis have been
exchanged with uranyl ions, allows to identify the ASR gel [1],
petrographic and electron microscope examinations allow to identify the
reactive minerals, the altered aggregates and the reaction products [2],
the mineralogical analysis of concrete with the determination of the soluble
silica can give information on the ASR occurrence [3].
But these diagnosis methods of ASR are generally qualitative. In addition the
specific identification of ASR related expansion is often difficult, because of the
simultaneous occurrence of different causes of expansion. So the determination of
the potential expansion due to ASR from concrete cores [4) [5] could be misjudged
for example by other swelling processes like delayed ettringite formation (DEF).
Thus, the development of a more specific and quantitative method is needed.
Our previous papers [6, 7, 8] reported the development of a methodology to quantify the
reaction degrees of ASR and correlate these reaction degrees with the swelling of the
aggregate, mortar bars and concrete specimens. It was shown that :
ASR degrades reactive silica to form among other silanol sites by the siloxane
breaking up inside silica.
the silanol sites formation results in an increase of apparent mass volume of
silica. So, this apparent mass volume is a specific parameter of the ASR.
the swelling of the flint aggregate by increase in the apparent mass volume is
responsible for expansion of not only mortar bar but also concrete specimens.
From these results, the objective was to establish if the measurement of apparent mass
volume of silica present in different aggregates is a pertinent parameter to estimate ASR
reaction degree responsible for damaging of concrete. This paper reports observations
carried out on laboratory concrete specimens containing either flint like a coarse
aggregate or siliceous limestone like a sand or a coarse aggregate from different works
hence different experimental methods. First application on field concrete cores have also
been tested. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Concrete formulations
Case 1 : Concrete with flint aggregate -- The concrete composition with flint aggregate is
given in Table 1. The cement is normal portland cement CEM I 42.5 with an alkali
content of 0.99 % Na2 0 equivalent. The sand is a non-reactive limestone. The coarse
aggregate is composed of a reactive flint aggregate and a non-reactive limestone. Thanks
to the high content in cement, alkali content for the concrete is approximately 5 kg/m 3
without sodium or potassium hydroxide addition.
A complete characterisation of the reactive flint aggregate given by Bulteel [9] shows
that the aggregate has a composition close to 99% Si02. The mineralogy of the flint
aggregate is determined by optical microscopy. This flint aggregate is essentially
micro to cryptocrystalline (microquartz) and contains radial fibers (chalcedony).
Case 2 : Concrete with siliceous limestone aggregate -- The concrete compositions with
siliceous limestone coarse or fine aggregate are given respectively in Tables 2 and 3. The
cement is normal portland cement CEM I 42.5 with an alkali content of 0.804 % Na 20
equivalent. The sand of the first concrete is a non-reactive limestone whereas the coarse
aggregate is a reactive siliceous limestone. In the second concrete, the sand is a reactive
siliceous limestone whereas the coarse aggregate is a non-reactive limestone. Thanks to
sodium hydroxide addition, the alkali content for the concrete is 5.125 kg/m3.
A complete characterisation of the reactive siliceous limestone has been given by
Monnin et al. [IO]. It is composed of close to 7.8 % free Si0 2 • A petrographical
analysis shows that siliceous limestone is a micritic rock mainly constituted by
calcite associated with a smaller quantity of dolomite. Diagenetic silica is principally
observed under small particles (»::IOJ..Lm) scattered on the matrix. Sometimes,
automorph quartz, microcrystalline quartz or chalcedony are also visible. Fine silica
(<2J..Lm) is present and associated with the clay fraction. Several silicates are detected
: phyllosilicate identified as illite and clinochlore. Illite is dispersed into carbonated
matrix. K-feldspar and framboYdal pyrite are also present in smaller quantities.
Case 3 : Field concrete -- The field concrete is extracted by taking cores of the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
viaduct built in 1976 and altered by ASR. The concrete formulation of the viaduct is
unknown but petrographical analyses show that the coarse aggregate is composed of
a non-reactive limestone and the sand is essentially siliceous with a reactive part. It is
a mixture of mono and poly crystal lattice quartz, flint aggregate, fragments of
sandstone ...
EXPERIMENTAL METHODS
The expansion test on concrete is based on a part of the AFNOR PI8-587 Standard
specification [II].
Case 2 : Concrete with siliceous limestone aggregate -- In the case of the concrete
containing siliceous limestone like coarse or fine aggregate, after 24 hours of hardening,
70x70x280 mm concrete specimens are introduced in a container and autoclaved 35
weeks at 38°C and 100% Relative Humidity to accelerate A SR. After each chosen week,
a specimen is taken out of a container and placed at 20°C and 100% Relative Humidity to
cool down for 24 hours. After the cooling period, the linear extension of the specimen is
measured.
The volume characteristics are measured on silica contained in aggregate skeleton. First,
this silica must be extracted. For laboratory concrete specimens, a 70x70x100 mm sample
is cut off from the central area of the specimen. For field concrete, cores are extracted
from the viaduct. The samples are crushed, dried at 105°C and finally ground below 200
jlm. Selective chemical attack developed by Riche [12] to extract "silica" from aggregate
skeleton is used. This selective chemical attack removes the cement paste and carbonated
matrix. In the case 1 (Table 1), extracted silica come from a part of coarse aggregate
which is a flint aggregate. The efficiency of the treatment is controlled by X-ray
fluorescence and gives a silica content of at least 99%. In the case 2 (Tables 2 and 3), for
the concrete containing siliceous limestone like sand or coarse aggregate and in the case 3
of the field concrete containing siliceous sand, the residue of the attack called "silica"
corresponds to silica (Si0 2) but also different insoluble minerals (clay and feldspar
minerals). These insoluble minerals are presumed to be stable as ASR progresses and so
their volume parameters stay constant. On the other hand, "silica" is composed of a part
of reactive silica which presents variation of their volume parameters.
Second, the volume characteristics of "silica" extracted by the chemical treatment are
measured:
the porous mass volume "Ypor" expressed in cm 3/g is obtained by the BJH
method (Barrett, Joyner, Halenda) from the desorption curve measured on an
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Case 1 : Concrete with flint aggregate -- ASR alteration of the concrete with reactive flint
aggregate involves the swelling of the specimen (Fig 1) expressed by DVN:
[2]
The swelling curve increases during the first twelve weeks then reaches an asymptote.
Table 4 contains the relative volume variation of the specimen and the flint aggregate
volume characteristics for different weeks of autoclaving.
For ASR alteration, apparent mass volume of the flint aggregate increases. A
significant increase of the porous mass volume has been observed as it is multiplied
by 1.8. The absolute mass volume increases a little for the same period.
Case 2 : Concrete with siliceous limestone aggregate -- Fig. 2 represents the expansion
curves of the two types of concrete expressed by DLIL:
[3]
In the case of concrete containing siliceous limestone like coarse aggregate, two periods
are identified : a relative dimensional stability during the first seventeen weeks and an
increase in expansion between 17 and 35 weeks to reach 937 J.Hnlm.
For the concrete containing siliceous limestone like sand, swelling increases during the
35 weeks and especially between 4 and 21 weeks where the slope is higher. At 35 weeks,
the linear extension value reaches 2736 J..Un/m.
Tables 5 and 6 show volume characteristics variation of "silica" respectively extracted
from siliceous limestone coarse and fine aggregate increases between 0 and 35 weeks of
ASR. In both cases, the increase of apparent mass volume of "silica" is essentially due to
porous mass volume. In concrete with reactive coarse aggregate (Table 5), "silica" porous
mass volume increases to 33 % and absolute mass volume gains 0.0027 cm 3/g. In the
other concrete with reactive sand (Table 6), a significant increase of the porous mass
volume is observed. The value is multiplied by 2.2. The absolute mass volume progresses
3
substantially to 0.0104 cm /g.
Globally, volume characteristics variation of "silica" extracted from siliceous limestone
sand are higher than "silica" extracted from siliceous limestone coarse aggregate. This
result correlates with linear extension value of the concrete specimens.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Two types of laboratory concrete were made with different reactive aggregates : one
with flint aggregate and the other with siliceous limestone aggregate. The concrete
specimens were submitted to ASR accelerated tests. They were in saturated condition
and without restraint. All concrete specimens were swollen. The "silica" extracted
from different reactive aggregates showed an increase of its apparent mass volume as
ASR progressed. This increase was related to the porous mass volume increase which
is characteristic of ASR and could be a pertinent parameter for following ASR
reaction degree in field concrete.
Field concrete
In this study, a first application on field concrete altered by ASR is achieved. Concrete
cores were extracted from a viaduct afflicted by ASR. For viaduct life, these concrete
cores were subjected to degree variations of saturation of concrete and amount of
restraint. The details of sand and coarse aggregate used during the building of the viaduct
were not known. Sample A is extracted from an unaltered part of the viaduct (Table 7).
Sample A was used as a reference. Samples B and C are extracted from an altered part.
An apparent mass volume of "silica" which increases from unaltered to altered samples is
observed (Table 7). The absolute mass volume is constant for sample B and gains only
0.0017 cm3/g for sample C. On the other hand, the porous mass volume increases
substantially as it is multiplied by about 3 between unaltered sample A and altered
samples B and C.
The porous mass volume variation could be a pertinent parameter to measure the ASR
reaction degree responsible for the degradation of the concrete.
CONCLUSION
The objective of this study was to improve the quantitative estimation of ASR in
concrete by following new parameters. Two types of laboratory concrete made with
different reactive aggregates (flint aggregate and siliceous limestone) and a field
concrete cores altered by ASR were tested. The methodology used allows to extract
by a chemical method the "silica" of the aggregate contained in concrete and
measures its apparent mass volume from absolute and porous mass volume.
As ASR progresses, the increase of the apparent mass volume of "silica" contained in
aggregate was accompanied by the swelling of the concrete. This increase of the apparent
mass volume is related to the porous mass volume increase. To assess the progress of
ASR, this porous mass volume appears more pertinent than the absolute mass volume.
The apparent mass volume and more especially porous mass volume of "silica"
extracted from the aggregate are new pertinent parameters to follow ASR reaction
degree. Their advantage is to be specific of ASR. They may be used for quantitative
assessment of the alteration of "silica" from aggregate which permit to follow ASR
reaction degree responsible for the degradation of concrete. This method does not
permit to directly measure the damage of the structure which, among other factors,
depends on the degree of saturation of concrete and amount of restraint on the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACKNOWLEDGMENTS
REFERENCES
[5] Cyr M., Carles Gilbergues A, Normalized age applied to AAR occurring in
concretes with or without mineral admixtures, Cement and Concrete
research, 2002, 32, n° 11, 1771-1782.
[6] Bulteel D., Garcia-Diaz E., Vernet C., Zanni H., Alkali-silica reaction: a
method to quantify the reaction degree, Cement and Concrete research,
2002, 32, n°8, 1199-1206.
[7] Riche J., Garcia-Diaz E., Bulteel D., Siwak J.M., Vernet C., Mechanism of
damage for the Alkali-Silica Reaction : relationships between swelling and
reaction degree, Proceedings of the International Seminar : Repair,
Rejunevation and Enhancement of concrete, Thomas Telford, Dundee, 2002,
93-102.
[8] Garcia-Diaz E., Riche J., Bulteel D., A new method to improve the diagnose
of the alkali silica reaction in concrete, 1st International Conference on
Concrete Repair, Proceedings of International conference of Saint Malo,
2003, Vol1, 391-398.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Time (weeks) (DVMsoecimen V,,, (cm'Jg) Voo, (cm 3/g) v,•• (em /g)
0 0.00 0.3854 0.0045 0.3899
2 0.05 0.3857 0.0046 0.3903
4 0.16 0.3858 0.0047 0.3905
6 0.26 0.3860 0.0052 0.3912
8 0.34 0.3861 0.0056 0.3917
10 0.38 0.3862 0.0061 0.3923
12 0.48 0.3865 0.0076 0.3941
24 0.47 0.3868 0.0080 0.3948
0000~-------;F=~====~·
~4000 /
i 3000 /
'J2ooo 41
/
~ 1000+-~~~--------------------------
e. 0 ~
0 10 15 20 25
nmelnweeks
~ 500+---~~------~~~~----------
10 20 30
-500 _[_____________________________ __40
Time In weeks
Sulfate attack on concrete is not a new phenomenon. Eminent scientists, including Vicat,
Michaelis and Le Chatelier, recognized a century ago that adverse reactions occurred
between cement paste and aqueous solutions containing sulfate. Important contributions
to our understanding were made during the first few decades of the 20th century by
Thorvaldson and others: Skalny, et al., give a brief history [6].
These studies, including results from both field exposure and laboratory trials, led to the
development in the 1920's of ASTM type V cements. Type V cements specify an upper
limit to tricalcium aluminate cement contents, field exposure and laboratory trials having
shown a positive correlation between extent of deterioration and tricalcium aluminate
cement contents. In general the use of low C 3A cements has been successful in avoiding
sulfate attack - such that Harboe in the Verbeck symposium of 1982 wrote.
"When I am asked to show someone a structure that has deteriorated due to sulfate
attack, I am hard pressed to find good examples for them to see. For 40 years we
have been building structure to resist sulfate attack, and the older structures that
did deteriorate have now been replaced. When you add this to the fact that well
over 90% of the concrete in the United States is not exposed to sulfate attack, it is
not hard to understand why most people today have never seen sulfate attack. "
However, since standards for sulfate-resistant cement were first adopted, expectations of
concrete performance lifetimes have altered. Fifty to one hundred years ago
"performance lifetimes" could not have been an explicit target: cement concretes were a
relatively new and innovative product. Today, it is increasingly expected that major
infrastructure projects will have long performance lifetimes, perhaps a century or more,
with periodic inspection leading to only minor maintenance. The challenge of achieving
these lifetimes without entailing excessive costs and environmental impact is not well
addressed in the current literature and could not reasonably have been addressed by older
work. Moreover the design of satisfactory accelerated tests, to compress perhaps a
century of real exposure into, at most, a few years oftesting, remains elusive.
Responses to these developments have been disappointing. There are, at present, a series
of national prescriptive specifications without a clear rationale for the minor differences
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
````,,,,```,-`-`,,`,,`,`,,`--- Durability of Concrete 497
between them. Clearly, traditional approaches have little to contribute to a harmonized
approach, having reached if not exceeded the data base supposedly underlying their
construction: we discuss this in a subsequent section.
This paper considers the research needed to build such a generic approach. The basis of
this has been elaborated by the European NANOCEM* consortium for fundamental
research on cementitious materials. The NANOCEM concept has been to apply the fruits
of modem materials science to cementitious materials; to develop and implement a broad
interdisciplinary approach to durability by integrating relevant chemistry and physics and
to integrate observations on phenomena occurring across a broad range of length scales,
from macroscopic to nanoscale. The approach recognises that the course of sulfate attack
on concrete requires a holistic approach; attack is marked by both chemical and physical
changes - the two are concurrent and cannot be considered in isolation. The scale of
reaction is important because, while diffusion occurs at the atomic scale, after nucleation,
solids develop initially at the microscale and if stress accumulation becomes too great,
the macroscopic appearance and performance of concrete is affected. Thus visual
appearance, dimensional changes and reduction of engineering performance are
ultimately caused by phenomena occurring at the micro- and nanoscale.
Degradation Mechanisms
One of the most controversial areas related to sulfate attack is the nature of the
degradation mechanisms occurring. As already mentioned, many publications attribute
degradation purely to ettringite formation. Mehta [2] reviews various theories, giving
prominence to a mechanism whereby topochemical formation of ettringite occurs "with
www.nanocem.net
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
498 Glasser and Scrivener
directional crystal growth and swelling of ettringite by water absorption." More
generally three principal degradation mechanisms are discussed:
1. expansion due to ettringite formation
2. expansion due to gypsum formation
3. softening due to loss of portlandite and decalcification ofC-S-H
Before discussing these in more detail it must be stated that all three mechanisms can
occur and that their relevant importance will depend on the conditions of exposure,
whether in the field or in the laboratory. Moreover, mechanism 3 may to some extent
relieve stress resulting from exposure reactions by creating new space potentially
available for crystallisation. Mechanism 2 is discussed subsequently.
(j)
Ca 2 + +so~- + 2H ~ CS.Hz
The equations CD and @ above for the formation of ettringite from calcium alumino
monosulfate and external sulfate ions, and for the formation of gypsum indicate that in
both cases calcium ions are also required. Thus, the formation of both phases will
inevitably be accompanied by some loss of portlandite and eventually decalcification of
C-S-H, but additional leaching is also likely, depending on the external conditions.
Despite the clear importance of leaching, the focus on expansive processes may perhaps
be explained as much as anything else by the fact that this is one of the easiest properties
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
to measure in laboratory tests. The relationship between ettringite (or gypsum) formation
and expansion has been the subject of considerable debate. Recent work related to the
mechanism of heat induced internal sulfate attack (DEF) [3,4,5] gives new insights on
this question. Although there is no uptake of sulfate from the environment in the course
of internal sulfate attack, it is still not strictly speaking a closed system as the leaching of
alkali ions and the uptake of water may be necessary for expansion to occur [6].
Scherer considered the issue of expansion due to crystallisation pressure, drawing on the
seminal experiments of Lavalle [7] confirmed by Becker and Day [8] and Tabor [9]. This
work emphasises several important aspects of crystallisation pressure:
1. If empty pores, voids or cracks are available, crystals will preferentially grow in
these without exerting pressure
2. To exert pressure crystals must grow in confined spaces, where supersaturation
can be maintained locally
3. The pressure exerted is inversely proportional to the pore size
The feature which distinguished expanding from non expanding samples was the
formation of ettringite from calcium alumino monosulfate and sulfate absorbed by the C-
S-H, within the outer C-S-H products as inferred by microanalyses (Figure 3) and
recently observed directly in the TEMt (Figure 4). This finding was confirmed by the
study ofRamlochan eta! [1 0].
It may be inferred that a similar mechanism of expansion occurs during external sulfate
attack. Microanalyses of mature cement pastes show that outer C-S-H is finely mixed
with calcium alumino monosulphate. In this case the sulfate ions come from the exterior,
rather than the C-S-H itself and react with the confined monosulphate to form expansive
ettringite. This hypothesis is consistent with the microstructural observations of Gollop
and Taylor [11].
Recent work [12] on the exposure of sulfate resisting cements to sulfate solutions of high
concentration (29800 ppm) indicates that expansion can also be linked to the formation of
gypsum in small pores. After curing, and before exposure to the sulfate solutions,
microstructural analysis indicated that ettringite, rather than monosulphate, was
intermixed with the outer C-S-H product (Figure 5), which is consistent with the findings
of Gollop and Taylor for sulfate resistant cements [13]. After exposure to the sulfate
solution (Wittekindt test) all mortars expanded more than the limit of 0.5 mmlm at 70
days. There was no macroscopic evidence of gypsum, but microanalyses indicated fine
mixtures of this phase with C-S-H (Figure 5). Although these are only preliminary
results they further support the hypothesis that expansion only occurs when crystals form
in small confined pores, and that the massive deposits later seen in veins (e.g. Figure 6)
form as the result of recrystallisation.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The first requirement for deterministic models is knowledge of the staring point- i.e. the
chemical and physical structure of the cementitious material before exposure. The
constitution of hydrated cement paste has been the subject of intense research. But the
relationships between cement composition, water: cement ratio, paste mineralogy, curing
history, temperature and the amount and composition of blending agents remains at best
semi-quantitative. This is due to inadequate knowledge on the solid solution ranges for
the hydrate phase, especially those containing alumina (AFm and AFt phases) as well as
incomplete reaction of the anhydrous cementitious materials, in particular that of the
ferrite phase. Both these limits to knowledge are exacerbated in blended cements, where
the SCM component is more slowly reacting and usually introduces alumina (in reactive
or inert forms) as well as silica.
Laboratory studies should in theory afford better control of formulation and conditions of
exposure. In practice the design of tests rarely permits extraction of a full set of relevant
data and correlations. Firstly, in assessing the many manifestations of "sulfate attack", a
distinction must be made between "open" and "closed" systems. Table 1 shows some
important distinctions.
But many real concretes exchange mass with their surroundings and must be considered
in the context of open systems. A basic difference arises between our treatment of the two
types of attack: in closed systems, bulk composition remains constant and if sulfate attack
is to occur, it can only do so by redistribution of chemical matter between the constituent
solid and aqueous phases. On the other hand, open systems require for their description a
set of mass balances, showing mass loss as well as gain within the envelope defined by
the boundary between the concrete and its service environment. We can state at the outset
that, of the two types of system, closed systems are easier to treat than open systems. It is
of course fortunate that most open systems comprising concrete solids are limited in
openness: not all chemical species are free to exchange. An important approach to the
treatment of external sulfate attack therefore consists of defining the extent to which
chemical species have migrated. Needless to state, and depending on the openness of the
system, attack can follow many different pathways, even between the same initial and
final states.
Test Methods
Scrutiny of "standard" or "frequently used" test procedures reveals a number of concerns
about test protocols.
Table 2 illustrates some aspects of tests and test methods commonly in use
Thermodynamic studies
Systematic studies of sulfate phase stability in aqueous solutions lead to the conclusion
that mineralogical changes to the cement solids occur such that the free energy of the
system decreases. Cement is not stable in the natural environment and a series of
reactions ensue. Naturally we tend to be more concerned about the impact of reaction on
the cement but the local environment may also be affected although this is little studied.
At the outset, the composition contrast may be sharp but, as reaction progresses, the
differing diffusion rates, need to maintain local electrostatic charge balance and the
physicochemical nature of reaction in a polyphase material results in formation of zoned
structures. Diffusion is not occurring in a homogeneous system and for that reason,
Fick's laws are not directly applicable. This is equivalent to stating that a series of local
equilibria develop in response to composition gradients. The chemically and
mineralogically zoned structures frequently developed in the course of sulfate attack are a
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Once these studies are complete and integrated our ability to predict relations between
mineralogy and composition will take a big step forward. Progress can then be made in
interpretation of degradation and devising more satisfactory test protocols.
References
1 Skalny, J., Marchand, J., Odler, I. "Sulfate attack on concrete", (2002), Modem
Concrete Technology 10. Spon Press, London and New York.
2 Mehta, P.K. "Sulfate attack on concrete- a critical review", (1992). Materials Science
of Concrete (Eds. Skalny, J and Mindess, S.) American Ceramic Society, Westerville,
OH, 105-130.
3 Scrivener, K.L and Taylor, T.F.W. "Delayed ettringite formation: a microstructural and
microanalytical study", (1993). Adv. Cern. Res. 5, 139-146.
4 Famy, C., Scrivener, K.L., Brough, A.R., Atkinson, A., Lachowski E.
"Characterisation of C-S-H products in expansive and non-expansive heat-cured
mortars: an electron microscopy study". Proceeding of the 51h CANMET/ACI
International Conference of Concrete Durability, Barcelona, Spain, Vol1, 385-402.
5 Taylor, H.F.W., Famy, C and Scrivener, K.L. "Delayed ettringite formation" (2001).
Cern & Cone. Res. 31, 683-693.
6 Famy, C. and Scrivener, K.L., Atkinson, A. and Brough, A.R. "Influence of the storage
conditions on the dimensional changes of heat-cured mortars" (2002). Cern and
Concr. Res. 32, 269-278.
7 Lavalle, "Recherches sur Ia formation lente des cristaux a temperature ordinaire"(l853)
Compte Rend. Acad. Sci. 36,493.
8 Becker and Day, J., "Note on the linear force of growing crystals (1916).1. Geology 24,
313.
9 Taber S., (1916). "The growth of crystals under external pressure" Am. J. Sci. 41, 532.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
sulfate attack. I. ordinary portland cement paste", (1992). Cern. Cone. Res. 22, 1027-
1038.
12 Mauris, F. Lipus. K. and Scrivener, K.L. "Study of mortars made from sulfate
resisting cements subject to the Wittekindt test" (2005). Unpublished.
13 Gollop, R.S. and Taylor, H.F.W. "Microstructural and microanalytical studies of
sulfate attack. II. Sulfate-resisting portland cement: ferrite composition and hydration
chemistry", (1992). Cern. Cone. Res. 24, 1347-1358.
14 Idom, G.M. "Innovation in concrete research-review and perspective" (2005). Cement
and Concrete Research, 35, 3-10
15 Marchand, J., Samson, E., Maltais, Y., Beaudoin, J.J. "Theoretical analysis of the
effect of weak sodium sulfate solutions on the durability of concrete" (2002). Cern,
and Concr. Comp. 24, 317-329
16 Maltais, Y., Samson, E., Marchand, J. "Predicting the durability of Portland cement
systems in aggressive environments-laboratory validation" (2004). Cern. Concr.
Res. 34_1579-1589
17 Damidot, D., Barnett, S.J., Glasser, F.P. and Macphee, D.E. "Investigation of the
Ca0-Ab0 3-Si0 2 - CaS0 4-CaC03-HzO system at 25°C by thermodynamic
calculations" (2004). Advances in Cement Research. 16, 69-79.
OPEN CLOSED
Some components can exchange System mineralogy and microstructure can still
between cement and its service change although the bulk composition of the paste
environment. remains constant
Mineralogy changes inhomogeneously Mineralogy is comparatively constant provided
in re~onse to diffusiongradients temperature remains constant
Local changes in mineralogy depend on Mineralogy remains comparatively fixed and can
balance between mass Eains and losses. be predicted approximately from existing models
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Hiroshi Hirao is a research scientist of Cement Chemistry Team, R&D Center, Taiheiyo
Cement Corp., Dr. Eng. He is a specialist in the field of cement chemistry.
Daisuke Mori is a research scientist of Cement Chemistry Team, R&D Center, Taiheiyo
Cement Corp., Ms. Eng.
INTRODUCTION
Thaumasite form of sulfate attack (TSA) is a recent topic with major
implications on the durability of concrete. It is well known that TSA is serious in some
specific environments; sulfate-bearing conditions, existence of carbonate source, at low
temperature, and in high humidity.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Aggregate
For mortar, siliceous sand conforming to JIS R 5201-1997 was used. For
concrete, two kinds of fine aggregate, land sand 0 and crushed limestone G and two
kinds of coarse aggregate, crushed sandstone I and crushed and washed limestone G
were used as shown in Table 3. Coarse aggregate was washed before use to eliminate
fine powders.
proportions and the strength of each mixture by standard water curing are shown in
Table 4. The mixture proportion of N was selected as the same with that of standard
strength test, in which w/c is 0.50. For wlc of other cements, suitable w/c was selected to
show the similar strength level at 28 days that is 60 MPa by standard water curing.
According to JIS R 5201-1997, mortar was mixed and rectangular specimens of 40 X 40
X 160 mm were made and cured in water for 7 days at 20 aC and were applied for
immersion tests. As shown in Table 4, each cement showed similar strength at 28 days
under standard water curing. At 7 days when immersion started, LS showed higher
strength of 49 MPa. N and BL showed similar strength around 45 MPa. However, BB
showed lower strength of 34 MPa.
The volume of solution was 500 ml for one mortar specimen and the solution
was replaced every month. Two series of containers were kept at 20 oC and 5 °C.
Detailed analysis was carried out for mortar specimens at the age of 6 months. After that
only one specimen for each mixture proportion was kept in a container at each
temperature condition without replacement of solutions.
Evaluation
Mechanical behavior of the exposed specimens was evaluated by measuring
compressive strength and dynamic modulus. In order to study the mechanism of the
change of mechanical behavior, detailed analyses were carried out. Pore size distribution
was measured by using mercury intrusion porosimetry after D-drying. Ca(OH) 2 content
was measured for crushed whole specimen sliced from a mortar bar by using a
differential scanning calorimeter. Mineral composition was evaluated by using X-ray
diffraction. For this evaluation, samples just after grinding were measured in order to
avoid phase transformation by drying or carbonation. In order to distinguish ettringite
and thaumasite, 5 g of ground samples were treated by 100 ml of 5% Na2C03 for 1 hr
and the peak change was checked (8). Peaks of ettringite will disappear by this treatment
but those of thaumasite will remain. Chloride penetration depth was measured by color
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
at 20 °C. However, at 5 °C, slight increase was detected in Na2S04 solution except BB
mortar. BB mortar showed larger strength loss in Na2S04 solution. In the case ofMgS0 4
solution, strength decrease was detected except LS mortar. LS mortar showed slight
strength gain in MgS04 solution.
From these results, although strength was different depending on the kind of
solution, cement LS and BL including limestone powder showed no more degradation
compared to cement N and BB. In this range of experiments, limestone filler does not
affect the durability even in sulfate solution at low temperature.
6 months are shown in Table 7. In the cases of cement Nand LS penetration depths are
higher than 20 mm. The penetration depths of cement BB and BL are less than 20 mm
and BB shows less value because of the inhibiting effect of slag against cr movement.
In artificial seawater, larger penetration depth is observed and this is attributed to the
reducing effect of sulfate ion to the chloride binding capacity by cement hydrates (II).
ACKNOWLEDGMENT
Field exposure test has been carried out under the cooperation with Prof. K.
Ayuta, Kitami Institute of Technology, Japan.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
1. V. Ramachandran et al., Influence of seawater solution on mortar containing carbon
carbonate, Materials and Structures, Vol. 23, pp. 412-417, 1990.
2. P. Poitevin, Limestone aggregate concrete, usefulness and durability, Cement and
Concrete Composites, Vol. 21, pp. 129-137, 1999.
3. R. Ranc et al., Durability of cements with fillers, SP 126-66, 2nd Int. Conference,
Durability of Concrete, ACI, Montreal, pp.1239-1257, 1991.
4. Ted. Sibbick and Derek. Fenn, The occurrence of thaumasite as a product of
seawater attack, Proceedings of the First International Conference on Thaumasite in
Cementitous Materials, 2002.
5. H. Lee et al., An analysis of mineralogical and chemical changes occurred in
deteriorated concrete under influence of sear water, Concrete under Severe
Conditions: Environement & Loading, B. H. Oh et al. (eds), pp. 125-132,2004.
6. M. Ono et al., Mechanism of chemical erosion of cement hardened body by
seawater, Semento-Gijutsu-Nempo, No. 32, pp. 100-103, 1978 (in Japanese).
Table4
Mixture proportion of mortar and compressive strength
Unitweil!i t(kgfm Compressive strength (MPa)•••
Cement w/c
w c s P* 7 days 28 days 6month
N 0.50 256 512 1535 512 43.2 59.5 76.9
LS 0.41 256 621 1418 443 49.1 59.8 71.5
BL 0.44 256 583 1451 374 46.5 59.0 76.2
BB 0.51 256 513 1516 282 33.9 58.0 80.7
*) P: Normal portland cement**) Estimated value.***) Specunen stze was 40X40 X 160 mm.
NI 0/1 N 50.0 44.0 160 320 793 1029 12.0 3.6 2 340 47.4
LS 0/1 LS 36.0 41.2 180 500 652 948 13.5 3.3 2330 47.7
BL Ofl BL 40.0 42.0 165 413 713 1003 11.0 3.3 2 342 51.3
NG GIG N 50.0 44.0 160 320 812 1056 10.0 4.6 2 362 45.8
NF 0/G N 50.0 44.0 160 320 793 1048 11.5 4.0 2 353 45.1
BB 0/G BB 50.0 44.0 160 320 788 1042 18.0 4.5 2 346 46.9
*) Specunen stze was 150 X 300 mm cylmder.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Cl solution immersion 28 18 23 20 24 15 8 10
Dissolved Ca(Oflh Sea water immersion 4 12 9 8 4 16 3 9
3
(mg!cm ) Cl solution immersion 74 68 63 59 38 38 25 28
Depth of Cl ion Sea water immersion <;:20 ;;:2o <;:20 <;:zo 19 19 17 17
peoetration (mm) Cl solution immersion :?:20 <;:20 :?:20 ;;:zo 16 15 14 15
1.1
'i
:tJ
.t:l
"'
~
·u;
"'~ 0.9
~
0
<J
~
·;::
o:l 0.8
~
0.7
0.14 ,
.....•.....
1113-6111
06--!iOrm
om~----------------------~
~
0.12
-1--
• ·50~100rm
•>100rm
0.<1!5 ~--······--·····--· .................................................. ..
.g 0.1
li / irurl
~0.00
g Jo.rns c --
·~
l· g
~0.06 .· ... 1-
&. 8.0.01 f-----
..
.....·· 1- 1-- 1---
0.04
2--
L.-1~
0.02 -- -
~
L.,_ L__._ L__._ L.,_ L__._
20"C s"C 2o•c s•c 20"C s•c 20"C s"C 2o•c s"C 2o"C s"C
water sea water Cl water sea water Cl
0.02
b
00 I
li! O.o15 I
I
/j.
~ ? /j.
"" O.o1 I
"""'
0
u'cr
§~ •I
•N
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-"
~1
~~
0.005
0 ~
I
•• •Ls
[j.BL
OBB
0
~J -o.oo5
-20 0 20 40 60 80 100
Decrease ofCa(OH}2 content (mglcm3)
•
0 0.005 O.Gl O.G15 0.02
Increase of volume of pore larger than
100nm(cc/ ccl
Fig. 4 Relationship between relative increase of volume of pore larger than too nm
and relative compressive strength.* Values cured in water at 20 °C was assumed to
be x = o andy= 1.
(a) (b)
~ 1000
Me
2000
~
~
.£ ~ 500
5 1000
.E
8.0 10.0 12.0 14.0 16.0 8.0 10.0 12.0 14.0 16.0
28 28
(a) (b)
1500 ~ 1500
~
.e
1000 f 1000
ii 500 500
.E
8.0 10.0 12.0 14.0 16.0 8.0 10.0 12.0 14.0 16.0
28 28
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
10 12 14 16 18 20
26
6 10 12 14 16 18 20
20
Fig. 8 XRD patterns of LS and NG concrete. Upper and lower lines are before
and after Na 2 C0 3 treatment, respectively. 5: mineral phases from sand,
Et: ettringite, CH: portlandite, Fs: Friedel's salt.
M. Sc. Eva Rodum is Research Scientist at SINTEF Concrete, Trondheim, Norway. Her
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
fields of interest are durability of concrete, repair methods and condition surveys.
INTRODUCTION
In the last few years thaumasite has been in focus as an underestimated degradation
mechanism of concrete. This has resulted in the arrangement of a special conference 1
dedicated to thaumasite and a special issue of Cement and Concrete Composites2 on this
topic.
It is generally accepted that thaumasite; Ca 3Si(OH)6(S0 4 )(C0 3 )·12H 20, forms under
moist conditions below l5°C and taking silicate from the CSH binder in the concrete,
carbonate from, for example, limestone filler and sulfate usually from external sources.
Thaumasite is not a hydraulic binder, so the process will make the concrete crumble
while transferring the binder to a white mush.
There is considerable discussion currently among scientists as to the conditions
under which thaumasite may form, and since the present paper reports on three case
studies where thaumasite is detected, it may throw some light on this issue.
• The main cause of damage in the cellars with open connection is chloride induced
reinforcement corrosion. The damages are concentrated around the gratings, i.e. the
beams crossing beneath and the floor edges along the gratings.
• The main cause of damage in the cellar with trapdoor connection is concrete
deterioration due to sulfates (gypsum, ettringite and thaumasite). Mainly structural
elements that most of the time is above the manure level were damaged.
The high chloride level in concrete (up to 0.3% Cr) of concrete mass in cellars with
open connection was at first puzzling. However, it was found that the cows are licking
salt from blocks and that the excess salt is excreted in the urine that in this case is soaking
the concrete, probably with some drying and wetting cycles.
The carbonate source for the thaumasite formation can also be urine, or rather the
major component urea that through a reaction with calcium hydroxide will form calcium
carbonate and ammonia;
However, the carbon dioxide content in closed fermenting manure basements can be
quite high, so carbonation cannot be ruled out as a carbonate source;
The scanning electron microscopy (SEM) study of plane polished specimen taken from
the right beam in Fig. 2 revealed that the sulfate attack is external as gypsum layer at the
surface is shown by the back scattered electron (BSE) image in Fig. 3. The sulfate
content was analyzed as a function of distance from the surface for several beams and
plotted in Fig. 4. The sulfate content is much higher (up to 5% sol· of concrete mass)
than the expected background level due to contribution from portland cement (estimated
to 0.6-0.7 % of concrete mass). Sulfate can either be deposited after direct contact with
the manure or as hydrogen sulfide gas, H 2S, being chemisorbed on the concrete and
oxidized to sulfate by bacteria. As seen by SEM, the binder is full of cracks and partly
transformed into thaumasite, as identified by voids partially (Fig. 5) or fully (Fig. 6) filled
with pure thaumasite according to element distribution in energy dispersive spectra
(EDS). The binder in Fig. 6 is crazed while the aggregate appears as even but rough
surfaces. Energy dispersive spectra (EDS) of other crazed binder areas show that the
binder itself has a composition close to thaumasite. The temperature is believed to be
above 15 oc most of the year in this manure cellar with trapdoor connection, which is
supposed to be the upper limit for thaumasite formation.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
524 justnes and Rodum
CASE 2: SEAGULL GUANO AT CONCRETE BRIDGE
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
surface below the nest and somewhat increased sulfate content at 1-4 mm in all samples.
Chloride profiles plotted in Fig. 9 show that under the nest, the chloride content is
already above the limit for chloride induced corrosion ("='0.1 % cr of concrete mass) at
about 18 mm and as high as 0.39 % cr at 1-4 mm from the surface. 0 in Fig. 9 is side
with nests, while Vis other side without nests. P3 is core on sloping surface, while PI is
cores from vertical surface below slope. Thus, P3-0 with the highest chloride content is
directly under the nest. An example of the enhanced chloride load at the surface by the
seagull activity is shown in Fig. 10 where to air voids near the surface were filled with a
compound consisting of Ca, Cl and 0, probably as double salt of calcium hydroxide and
calcium chloride with crystal water (e.g. CaClz·3Ca(OH)2·12H20).
Scanning electron microscopy of concrete under nests showed strong chemical
alterations in the first 5 mm from the surface. Fig. 11 depicts fine crystals growing in an
air void, and that the composttton of these corresponds to thaumasite;
Ca3 [Si(OH)6-12H 20](S04)(C03 ), with contamination of aluminum probably due to
coexistence with ettringite; (Ca3 [Al(OHk 12H20]) 2 ·(S0 4)3-2H 20. Due to their similar
structure, thaumasite and ettringite may actually form solid solution. The average
temperature over the year is close to +5°C, so it is cold enough for formation of
thaumasite in this case. The paste is full of microcracks where often these originate from
the same point, typical for expanding compounds as in the image reproduced in Fig. 12
where the EDS of such a point indicates largely ettringite contaminated by some silicon.
Other similar cracked regions give EDS corresponding to thaumasite composition
contaminated by aluminum. Anyway, there must be an external sulfate source as sea
water or guano. Indeed, Fig. 13 reveals a pure gypsum crystal at an air void wall proving
excess sulfate in the system and supporting an external supply. The necessary carbonate
supply for thaumasite formation (in the absence of limestone filler in the concrete recipe)
can be urea from the guano (Eq. 1) or possibly soluble carbonates in the guano or sea-
water that is saturated with respect to calcium carbonate (i.e. low solubility). A number
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 525
of phosphate compounds were found as well; magnesium phosphate and calcium
phosphate with or without alkalis. Syngenite; K2Ca(S0 4) 2, along with magnesium
phosphate was found at a depth of 3 mm in cores under the nest. The penetration of
phosphate to this depth indicates the absence of soluble calcium (which would otherwise
be precipitated as calcium phosphate). Indeed calcium hydroxide was absent in the guano
affected zone, while calcium hydroxide crystals were found in abundance in voids at 40
mm indicating also water saturated conditions at this depth.
Due to public reaction, and environmental activists, it was not an option to hinder
the seagulls in their activities. As a result of the study, it was rather decided to coat the I-
beam with a polymer membrane to stop the chemical attack by guano and let the
kittiwakes nest at will.
Industrial cooling tubes made ofEtemite (i.e. asbestos fiber reinforced cementitious
paste) started to fail (i.e. cracked open) after 30 years in service. The material in the pipe
walls was investigated to see if the failure was due to the shock wave exceeding the
design strength when pumps where started after standstill or chemical or physical (e.g.
leaching) changes. The outer part of the tubes was affected by the tidal zone of the sea
(partly submerged), while the cooling medium was seawater that would affect the inner
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
wall. A failed and an intact tube are shown in Fig. 14. A plane polished thin section was
made across the tube after fluorescent epoxy impregnation that was suitable for both
optical and scanning electron microscopy studies. The wall thickness was 36 mm, and
cracks along the surface to a depth of 8 mm from the inner surface could be observed by
the naked eye. In addition, there was a 1.5 mm layer of another density/color at the inner
surface. Both surfaces had a brownish precipitation. Fig. 15 shows the width of cracks
(about 0.2 mm) running parallel with the inner surface as a photo from optical
microscopy.
The SEM investigation showed two types of asbestos fibers; one appearing dark
consisting of the elements Mg, Si and 0 and the other appearing bright consisting ofMg,
Ca, Si, Fe and 0. During the SEM study, a mapping of the elements Na, Mg, Al, Si, S,
Ca, Mn and Fe was performed close to both surfaces. The inner surface had a 30 J.lffi thin
precipitated layer rich in manganese (Mn) and iron (Fe). The following 100 11m was
washed out resulting in several pores. Thereafter a layer of 1,000 11m was rich in
magnesium (Mg). The latter is the well known seawater effect where magnesium
replaces calcium in hydroxides and silicates. The mapping of the outer area revealed a
magnesium enriched zone of 500-800 J.lm. The SEM also disclosed 60-70 11m bands of
more or less pure thaumasite running parallel to the surface as exemplified in Fig. 16.
The average temperature is in this case assumed to be well below 15 °C, perhaps + 10 °C.
A total of 15 mm (sum of inner and outer zone) of the 36 mm thick wall was
chemically altered, and together with the 0.2 mm cracks running parallel to the tube
length, this is likely to be the direct reason of reduced strength leading to pipe failure.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
which together with the cracks, lead to reduced strength and subsequent failure at high
loads (i.e. shock waves during pump start).
REFERENCES
Fig. 3- Back scattered electron SEM image at the surface of the right beam in Fig. 2.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Strong attack
~u 3.0 -o- Strong attack
c
0
u
0 2.0
~
~ 1.0
''
II) ' bo - - - -t:.- - - -fro - - -
0.0
0 5 10 15 20 25
Distance from surface (mm)
Fig. 5- 6ox BSE image of an area close to an exposed surface (seen in the upper part)
showing a large air void in the process of being filled with thaumasite.
Fig. 6- BSE (tsox) SEM image showing 2 air voids(, 200 IJm diameter) filled with
thaumasite (according to the lower EDS).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 8- Drilling out cores from the concrete beam under nests and below nests.
i 0.15
8 0.10
t 0.05
0.00 E.......-'--L-->--J~z:iliii!E~~"'*-....- ......,j~
0 5 10 15 20 25
Dapth(mm)
Fig. to - Air voids at a depth of 1 mm below vertical surface under nests filled with
an unusual compound consisting of Ca, Cl and 0 (probably CaCl 2 3Ca(OH) 2 t2H 2 0)
indicating an extreme chloride load at the surface.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 11- 1,ooox BSE image of an air void (d::::: 30 11m) with a hairy growth
from the surface inwards consisting ofthaumasite/ettringite, indicating
water saturated conditions.
Fig. 12- 0 4oox SSE 2.5 mm from the surface of sample P40 directly under a nest. The
microcracking is probably due to expanding paste.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 13- t,ooox SSE of an air void (dark area to the right) in sample P40 about 3 mm
from the surface.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 14 · Photo of parallel cooling pipes in Eternite where one has failed
(i.e. blown open, note the debris in the back).
Fig. 15 ·An image by optical microscopy over an area of 4.2x2.7 mm 2 showing cracks
of 0.2 mm width running parallel to the inner surface of the cooling tube.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 16- zoox BSE showing a 6o IJm broad band that has a composition that
according to the EDS complies with the composition of pure thaumasite.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Dr. Norah Crammond - Former Head of the Diagnostic Unit at the BRE (Concrete
Section). In this role she was heavily involved in all BRE longer-term investigations and
research into sulfate attack and other longer-term concrete durability problems. Dr
Crammond was the primary research advisor to the Thaumasite Expert Group that was
set in 1998 by the UK government.
INTRODUCTION
During the early 1990s, several investigations by BRE found that cast-in-situ
concrete foundations had been attacked by sulfates in the ground and that in each case the
principal reaction product was the mineral Thaumasite 1• This form of sulfate attack
(TSA) has been found to be more prolific in concretes containing carbonate aggregate.
The important factors external to the concrete appeared to be low temperature (less than
l5°C), very wet sites and readily available sulfate in permeable adjacent soil.
A field trial to investigate the performance of concretes containing limestone
aggregates in sulfate-bearing ground was instigated by BRE in 1998. The trial site was a
9m wide roadside verge at Shipston on Stour in Central England, adjacent to one of the
TSA field cases2 • The ground comprised Lower Lias Clay with wet, Sulfate Class 3 (1.4-
3.0 g/1 SO/) groundwater conditions at a reasonable working depth (2.5 m) for test
specimens3 • The field trial comprised the burial in 1998 of 196 test specimens in two
trenches, with half to be excavated for assessment after three years, and the other half to
be excavated for assessment after ten years.
In June 2001, three years after their burial, the 98 test specimens in the first
trench were excavated for visual and laboratory assessment. The preliminary findings on
the condition of 86 of these concretes were presented at the First International
Conference on Thaumasite in Cementitious Materials held at BRE in June 2002 4 • The
conference paper described the range of concrete specimens buried at the field trial site,
including their composition, method of manufacture and installation. The exposure
conditions, the excavation and preliminary visual assessment of the specimens were also
presented. After the preliminary on-site visual assessment, the excavated concretes were
transported to BRE, where they were examined in more detail using optical microscopy,
X-ray diffraction, chemical analysis and scanning electron microscopy. The surfaces of
the concretes were also re-examined and photographed under laboratory conditions. The
main findings from these laboratory analyses are presented in this paper, which also
includes improved groundwater analyses obtained from recently installed piezometers on
site. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Concrete mixtures
Three sources of limestone aggregate (Magnesian/dolomitic limestone,
Carboniferous limestone and Jurassic oolitic limestone), both coarse and fine fractions,
were included in the concrete mixtures. Also one source of flint gravel was included as a
control. The geological type and mineralogical compositions of these are given in Table
1 of 4 •
Eight types of binder were included: three portland cements (PC) with
tricalcium aluminate (C 3A) varying from 1 to 10%, and the 10% C3A PC blended with
ground limestone (PLC with 15% ground limestone), ground granulated blast furnace
slag (70% ggbs/30% PC and 40% ggbs/60% PC), pulverised fuel ash (30% fly ash/70%
PC), metakaolin (25% metakaolin/75% PC) and microsilica (10% microsilica/90% PC).
BRECEM (a 50:50 blend of high alumina cement and ggbs) was also included. The
compositions and appropriate British and European Standard notations are shown in
Table 2 of 4 .
The four aggregate types were combined with the 8 binder types to produce 39
different concrete mixtures, comprising 33 main and 6 outlier mixtures, as listed in Table
3 of 4 . The main group all had a nominal binder content of 320 kg/m 3 , a slump of 50± 20
mm and a range of water/binder ratios between 0.52 and 0.58. The outlier group of 6
mixtures was included in order to obtain data on the effect of sulfate on poor quality
concrete. These had a reduced nominal binder content of 290 kg/m 3 , an increased slump
of 75 ± 20 mm and a range of water/binder ratios between 0.72 and 0.78. All mixtures
were made in 30 litre batches, which provided sufficient material to cast two cubes for
placing on site, one in each of the two trenches.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
• Groundwater conditions on the Shipston site throughout the field trial have been wet
(below groundwater table) and Mobile at the 2.5 m depth of the concrete specimens.
• The pH of the ground and groundwater at all levels has been consistently neutral, at
about 7.5.
• During the period May 1998 to June 2001, sulfate concentrations in the 3-year trench
at the level of the concrete specimens were at the lower-half of Sulfate Class 33 . New
UK specifications are now in place, which classify sulfate conditions in terms of an
Aggressive Chemical (AC) Class (Building Research Establishment C55]). The AC
Class at the 2.5 m depth of the concrete specimens works out as AC-3 on the basis of
a Design Sulfate Class of 3, a natural soil site, a neutral pH of 7.5 and a Mobile
groundwater condition.
• The sulfate ions in the groundwater are associated primarily with calcium (Ca++),
magnesium (Mg++) and sodium (Na+) cations. The groundwater is near-saturated
with respect to gypsum (CaS0 4 ·2H2 0). Analyses of bicarbonate in groundwater from
the new piezometers installed at concrete specimen level has indicated an amount of
carbonate present of 0.24 to 0.29 g/1 co3, equivalent to 0.41 to 0.48 g/1 expressed as
CaC03.
• Ground temperatures have varied seasonally between about l0°C and 13°C at the 2.5
m depth of the concrete specimens. This is within the temperature range which
encourages the occurrence ofTSA in susceptible concrete.
• Though pyrite is probably present at depths approaching Sm, no contribution to the
chemistry of the groundwater at the 2.5 m depth of the concrete specimens would
appear to have been made from the (recent) oxidation of pyrite.
excavated from the trench and this has been presented in [4]. After transportation back to
BRE, each test specimen was subjected to a visual and physical inspection in the
laboratory. Photographs were taken of the two side faces, exposed to either in-situ clay or
backfill clay whilst buried. Samples were scraped from the surfaces of selected
specimens and these were analysed using X-ray diffraction. Nineteen of the cast-in-situ
test specimens then underwent further sample preparation enabling their surfaces and
• The depth of attack varied over the surface area of an exposed face and was often
concentrated at comers, along edges and in blisters.
• The surface of the in-situ clay side face of the cast-in-situ cubes can be particularly
irregular, which makes reading off a definitive depth measurement difficult.
• During excavation, soft surface material from many of the badly attacked cubes
remained behind in the walls and floor of the 3-year trench. This would mean that
the depth of attack is underestimated in some specimens.
• The onset of TSA is probably accompanied by expansion of the degraded surface
layers. This could lead to an overestimated depth of attack as the measurement
would not have been made from the original external surface.
• The deteriorated outer surface of a TSA-affected concrete will shrink in volume as it
dries out.
• For cast-in-situ cubes, the initial front to back dimension was not accurately known,
so this could not be used in the calculation of depth of attack.
halves were impregnated with a low viscosity epoxy resin containing a blue dye. The
half-impregnated core samples were then cut along their length so that the impregnated
and non-impregnated halves were separated. The non-impregnated halves were further
cut into a series of 7mm thick slices running parallel to the outer surfaces of the core and
these were analysed for total sulfate content.
The resin-impregnated halves of the cores were lapped to produce flat
surfaces, which incorporated the two original outer side faces of the cubes and any
attached materials. These lapped surfaces were then made into standard 30 micron thin
sections covered with temporary glass slips. The specimens were then ready for
examination in transmitted light using a polarizing microscope and, if required, by the
scanning electron microscope (SEM) at a later stage.
a) TSA
Four distinct zones of TSA have been well documented in the literature [ 1,6].
These range from innocuous Zone 1 TSA, in which thaumasite occupies areas of pre-
existing voidage to Zone 4, in which the cement paste matrix is completely transformed
into thaumasite mush. In addition, three main types of thaumasite were identified under
the microscope, a colourless, low birefringence variety, a yellow, higher birefringence
variety and a variety somewhere between the two. Whilst examining the thin sections, it
became apparent that the more yellow colouration was found in thaumasite deposition
closest to the clay interface.
b) PCD
An increasing amount of 'Popcorn Calcite Deposition' (PCD) has been
identified in deteriorated concretes over the last few years [7]. It comprises coarse
crystals of calcite and, like thaumasite, it does not possess any binding properties. It
forms under wet conditions in areas of a deteriorated concrete previously occupied by
cement paste. It can be regarded as the final degradation process associated with TSA.
PCD occurs as the pH of the cement pore fluid drops from around 10.5 towards 7 and it
has also been linked to the presence of available bicarbonate ions in surrounding
groundwater.
light deepens towards the concrete/clay interface. Typical consumption of the cement
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 547
paste and Magnesian limestone aggregate particles have been observed, in which
thaumasite is mostly present as Zone 2 to Zone 4 deposition with little evidence of Zone
1. A single phase of thaumasite was identified by XRD. Its crystalline structure is very
close to that of pure thaumasite as shown in Fig. 1.
Cast-in-situ concrete cubes made with PLC and all aggregate tvoes
Deleterious TSA was identified in the exposed surfaces of all eight PLC
concretes irrespective of aggregate type and concrete quality. Zones 1 and 4 were more
prevalent than Zones 2 and 3. Optical microscopy studies have shown there to be a sharp
degradation front between the attacked and the sound concrete. Neither ettringite nor
gypsum have been identified as void-fill. Out of all the cubes investigated during this
field trial, the outlier specimen comprising PLC and Magnesian limestone aggregate was
the most severely attacked (Table 1) and Fig. 4.
XRD analysis has shown that two reaction products tend to form in concretes
containing PLC and these are a thaumasite-like phase and lesser quantities of an
ettringite-like phase. PCD was identified within some Zone 4 TSA locations but only in
the concretes containing siliceous aggregate.
Cast-in-situ concrete cubes made with 30%fly ash/70%PC and all aggregate types
None of the concretes made with 30% fly ash/70% PC show any evidence of
deterioration. However, the level of sulfates in the surface layers is much higher than in
the bulk of the specimens. This phenomenon was found in other unattacked concretes and
is discussed in more detail in a later section.
Under the microscope, some minor Zone 1 TSA was identified but this was
insignificant in terms of deleterious TSA. Some secondary ettringite was also found
precipitating in voids. PCD was identified in the concrete specimens containing siliceous
aggregate. This points to the fact that PCD can form in the absence of TSA.
Cast-in-situ concrete cubes made with 70%ggbs/30%PC and all aggregate tvoes
A small amount of attack (up to lmm deep in places on three faces) was
identified in the 70%ggbs concrete containing a mixed aggregate of Magnesian limestone
and Carboniferous limestone. XRD analysis found that this was due to the presence of
gypsum and not thaumasite. Thaumasite was found in the surface of some of the cubes
but only as a non-deleterious void filling (Zone 1). Secondary ettringite was also
observed in some of the surface air voids. Small pockets of PCD were observed in the
surface of the 70%ggbs cube containing siliceous aggregate. This has occurred in the
absence of deleterious TSA.
Cast-in-situ concrete cubes made with remaining binders and Magnesian limestone
aggregate
There is no evidence of deleterious TSA occurring in any of these three
concrete specimens although a significant build-up of sulfates was discovered in the
outermost surfaces of the concretes made with BRECEM and microsilica (see later
section). Minor amounts of thaumasite were observed under the microscope but only as
Zone 1 TSA. A few air voids and adhesion cracks were also lined with gypsum and
ettringite. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
40%ggbs/60%PC, BRECEM and microsilica cement, sulfates must have penetrated but
have not reacted detrimentally with the cement paste phases. Work was carried out to
identify the sulfate-bearing species within the surface cement paste and the subsequent
results are discussed in the following section.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The thin sections already made and examined under the optical microscope were polished
further, coated with carbon and re-examined using SEM and EDX. A back scattered
electron image was taken for each of the samples examined and these are shown in Figs. 5
and 6. Ten separate spot analyses for CaO, Si02, A)z0 3 , S04, Na20, K 20, MgO and total
iron as Fe20 3 were then collected at depths of 0.2, 1, 2, 6, 10 and 20mm in from the
concrete surfaces of each specimen. Only the concentrations of CaO, Al 20 3 , Si02, and
SO/' are reported in this paper; the values for Na20 and K20 being consistently below
0.5%, for Fe20 3 below 5% and for MgO varying depending on closeness to aggregate
fragments. Accurate quantitative analysis of C0 2 and H20 is not possible using this
technique. Three distinct sulfate-bearing phases were identified; one in the 30%fly
ash/70%PC concrete (Fig. 5,A) and two in the PLC concrete (Fig. 6, B and C). The phase
in the fly ash concrete can easily be seen forming as high relief, cracked features evenly
distributed throughout the cement paste. A high relief, cracked phase has also been
identified in the PLC concrete but it is less abundant with sulfates tending to be in the
low relief, highly porous background material.
An average composition was determined for these three phases and these are
shown in Table 2. It can be seen from this table that the percentage total counts collected
by the SEM detector varies enormously from 18% for the low relief background material
up to 78% for the high relief phase seen in the fly ash concrete. There could be a number
of reasons for this difference but optical microscopy studies suggest that the high
capillary porosity seen in the background material could be a major factor resulting in the
low detection rates.
The oxide compositions for ettringite, thaumasite and gypsum are shown in Table 3 and
as both ettringite and thaumasite contain roughly 27% CaO, a third table (Table 4) has
been constructed, in which the oxide concentrations from Table 2 have been normalised
to 27% CaO. It can be seen from Table 4 that the high relief, cracked areas found in both
specimens contain more A)z0 3 than Si02 and an approximate sulfate content of 17 -18%,
which is closer to that of ettringite than thaumasite. On the other hand, the non-cracked
background material in the PLC concrete contains more Si02 than Al 20 3 and has a lower
sulfate content of 14.4%, more indicative of thaumasite.
In order to find out more about these phases, a representative sample from the
outermost 7mm of both concretes was analysed using XRD and the resultant traces are
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
550 Crammond et at.
shown in Fig. 7. It can be clearly seen that the high relief, cracked phase found in the
surface of the fly ash concrete is in fact amorphous, as signified by the absence of sulfate-
bearing mineral diffraction peaks. Under the optical microscope, these amorphous
inclusions are indistinguishable from the general isotropic cement paste phases. It may
follow that the similarly cracked areas in the PLC concrete may also be amorphous but
the background material certainly is not. Both XRD (Fig. 8) and optical microscopy
studies have identified the background phase to be a highly porous and expansive mesh
of thaumasite-like crystals with a structure, and to a lesser extent composition, closer to
that of pure thaumasite than pure ettringite.
The values with depth for weight % S032. of total counts are presented in
Table 5. It can be seen that there is a high concentration of sulfate in the outer 4mm of
the fly ash specimen, which drops to the background level at a depth of about 6mm.
Sulfates have penetrated farther in the case of the PLC cube, reaching a depth of over
lOmm before achieving background levels.
• the concrete is sent to a materials analyst who has had no input into the site
investigation or sample collection
CONCLUSIONS
The BRE Shipston field trial has proven to be very successful. The findings to
date have indicated that a timescale of three years is sufficient to differentiate between
TSA-resistant concrete mixtures and mixtures, which are susceptible. Unfortunately, the
concrete mixture proportions do not comply with the new UK recommendations [5] but
they probably relate to a significant percentage of the foundation concrete currently
buried in the ground. The conclusions from the physical, chemical and mineralogical
assessments of the field trial specimens, which have been exposed to Class 3 sulfate
conditions for three years are summarised below.
• Even after the relatively short period of exposure to Class 3 sulfates on site, the
surfaces of many of the test specimens were covered with a white-coloured reaction
product. X-ray diffraction analysis and microscopic studies have shown this reaction
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACKNOWLEDGMENTS
REFERENCES
10 Collett G, Crammond NJ, Swamy RN, Sharp JH: The role of carbon dioxide in
the formation of thaumasite. Cement and Concrete Research 34, pp 1599-1612
(2004).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
U'l
U'l
~
n
DJ
3
- --- - - -- -- .., ------- --------- -- -- --- ----- ------
d ---- -- ... - -- --------d -----·---... 3
Mixture type Approx. depth Average Bulk so,~ Surface reaction Surface layer of concrete in contact with clay - 0
sot ::s
-
of attack lmm) Surface in% by wt. products detected Optical Microscopy findings
Mean Max in % by wt. binder using X-Ray c.
binder Diffi'action
PC 10 +Sil <I <I 11.2 4.5 CC EfT,minor G
tD
Mainly TSA zones l-3.lsolated zone 4. PCD observed.
PC 10 +Maa 7 10 10.5 4.8 T minorG+CC Mainly TSA zones 2-4. Found below sound outer skin.
Dl
PC 7 +Sil
PC 7 +Mao
<1
7
<1
10
12.5
18.7
4.3
5.1
G CC minor E!f
T CC traceG
Mainly TSA zones 2 and 3. G 1rr0wino in voids
Mainly TSA zones 2 and 3. T and G to<rether in air voids
:-
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SRPC+ Sil 0 0 14.9 2.5 G, minor Eff+CC Deleterious TSA absent Zone 1 TSA only. G growing in
voids close to surface. PCD observed.
SRPC+Mao 6 10 12.5 3.4 T minorG+CC Zone 4 TSA -Smm deen) on surface with in-situ clav.
30%flv ash + Sil 0 0 7.3 2.8 Not analysed Deleterious TSA absent. Zone I TSA onlv. PCD found.
30%flv ash+ Mao 0 0 11.1 3.3 None found Deleterious TSA absent. Zone I TSA onlv. MinE voids.
40%_g_gbs+Sil 0 0 9.1 3.1 Not analysed Deleterious TSAabsent Zone I TSA onlv. PCD fuund.
40%ggbs+Mag <I I 12.0 3.1 Notan~d TSA zones 1-4 all present in thin outer surface laver.
70%_ggbs+Sil 0 0 4.5 2.3 Not analysed No TSA zones detected. Minor PCD and E linin• voids.
70%ggbs+Mag 0 0 6.1 1.9 Not analysed Deleterious TSA absent. Zone I TSA onlv. E in voids.
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
PLC+Sil 6 8 11.3 3.1 T trace G TSA zones 1-4. PCD in Zone 4 TSA locations.
PLC+Mag 8 10 16.4 4.3 T minorG TSA zones 1-4. Classic examole ofTSA deiUildation.
BRECEM+Mal! 0 0 9.2 1.4 Not analysed Deleterious TSA absent. Zone I only. G+E in voids.
Microsil + Mao 0 0 12.1 3.6 Not analysed Deleterious TSA absent. Zone I TSA onlv. Gin voids.
Metakaol + Mao 0 0 6.3 3.2 Not anal__Eed Deleterious TSA absent. Zone I TSA onlv. G+E in voids.
PC Outlier + Si1 5 8 13.9 3.0 PIT minorCC TSA zones 1-4. Minor Gin voids. Minor PCD found.
PCOutl.ier+Ma• 10 15 Not analvsed Not analysed
T minorCC Notani!!Y_sed
PLC Outlier + Sit 8 12 Not analysed T minorCC+E
Not analysed Not analysed
PLCOutl.ier+Mag II 20 16.3 3.8 T,minorCC Mainly TSA zone 4 acco~=ed by expansion. Sharp
d~on front at 20mm de G+E absent.
Key: PC(•) - PC with 10% or 7% C,A, Sll - siliceous aggregate, Mag - magnesian limestone aggregate, T - thaumasite, E - ettringtte, EfT -
mixed crystal, G -gypsum, CC - calc~e
Daneshlink.com
Durability of Concrete 555
Table 2 - % Oxide compositions for sulfate-bearing phases of the two Shipston
concretes selected for SEM microanalysis
PLC outlier+ Low relief, bland back- 27.0 3.5 8.2 14.4
Mag ground (Fig. 6, C)
Mixture Average wt.% S031- with depth from concrete/clay interface (mm)
type
0.2mm I.Omm 2.0mm 4.0mm 6.0mm !Omm 25mm
700
600
500
z.
-ii)
r::::: 400
.$
.E
300
200
100
15.7 16
2-Theta
Fig. 1- Four XRD traces showing their (110) peak positions in relation to the standard
peak positions for ettringite and thaumasite (labelled vertical lines). Highest XRD
peak relates to a PC concrete containing magnesian limestone aggregate, second
highest to a PC concrete containing siliceous aggregate, third highest to an SRPC
concrete containing magnesian limestone aggregate and shallowest peak to an SRPC
concrete made with siliceous aggregate.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 3-- Photograph of sliced core of concrete made with 30% fly ash and magnesian
limestone aggregate. Sound surface in contact with backfill clay is shown on the
right-hand side of the photograph. Image width 63mm.
Fig. 4-- Photograph of sliced core of concrete made with PLC and magnesian
limestone aggregate. Deteriorated surface in contact with in-situ clay is shown on
left-hand side of photograph. Image width 63mm.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Fig. 5 -- Back-scattered electron image of cube containing 30% fly ash at a depth of
tmm from the surface previously in contact with backfill clay. Circular features are fly
ash and cracked features marked with the letter A are rich in sulfates.
Fig. 6 - Back-scattered electron image of cube containing PLC at a depth of 2mm from
the surface previously in contact with in-situ clay. Both the cracked features (marked
B) and the low relief grey background (marked C) are rich in sulfates.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
600
Thaumasite
500 Thaumas~e
Ettringite
~400
"iii
c
2 300
.5
200
100
10 11 12 13 14 15 16
2-Theta
Fig. 7- XRD traces from surface samples of specimen containing PLC (upper trace)
and specimen containing 30% fly ash/70% PC (lower trace). The standard lines
for ettringite (powder diffraction no. 411451) and the standard lines for thaumasite
(powder diffraction no. 250128) are marked accordingly.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Jiirg Neuenschwander received the diploma degree in physics and the PhD degree in
natural sciences from the Swiss Federal Institute of Technology (ETH), ZUrich,
Switzerland. He is Senior Scientist in the Electronics/Metrology Laboratory ofEmpa,
Duebendorf, Switzerland. His research interests include ultrasonic material
characterization, ultrasonic and eddy current imaging and medical ultrasound.
Thomas Liithi has a Ph.D. in technical sciences by ETH (Swiss Federal Institute of
Technology, Zurich). His field of research is in general applications of non-destructive
evaluation techniques. He is head of the Center for Non-destructive Testing at Empa
(Swiss Federal Laboratories tor Materials Testing and Research, Duebendorf,
Switzerland) and affiliated to the Laboratory for Electronics I Metrology.
INTRODUCTION
In cases of sulfate interaction, the formation of secondary minerals like ettringite, gypsum
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
or thaumasite may lead to softening, cracking or spalling 1' 2 • Sulfate interaction always
leads to chemical gradients within the exposed cementitious materials 3• 4 . As a
consequence, in long time interaction (field conditions) the formation of distinctive zones
is often recognizable 5' 6 . The zones may be defined on the basis of specific phases like
portlandite, ettringite or gypsum being present or absent. A further complication of
sulfate attack is given through the fact that chemical interaction is never limited to
sulfates but is always accompanied by other mass exchange mechanisms involving
alkalis, hydroxyl and other soluble species 4 • The situations for accelerated laboratory
testing and for field exposures will always be much different. It is therefore very
important to develop reliable laboratory tests for clarifying specific factors of deleterious
reactions.
THEORY
The theory of surface sensitive Rayleigh waves 8 uses an incident beam near the leaky
Rayleigh angle 91 R from the liquid onto the surface of the solid sample. This leads to a
direct reflection of the beam (* in Fig. 1) and generates a leaky Rayleigh wave which
causes a trailing field which leaves the sample at the angle 91 R (arrows in Fig. 1). The
leaky Rayleigh or surface waves have a penetration depth of about one wavelength 9 • For
a velocity of the leaky Rayleigh wave of 2 km/s and for a frequency of 1 MHz the
wavelength 'A corresponds to 2 mm which is suitable to study the top surface layer of a
mortar.
Rayleigh waves in mortar have been investigated with conventional contact technique
using separate transmitter and receiver probes 10. Another technique using a non contact
mode is described by Titov et a!. 11 • They describe an immersion technique with a
transmitter and receiver probe in a liquid. This technique uses a pitch and catch
arrangement but due to the type of transducer (PVDF) and the high nominal frequency
(10 MHz) it is not suitable for the investigation of mortar.
The velocity of the leaky Rayleigh wave with immersion technique can be determined by
successively displacing the receiver probe by & and measuring the time difference .t1t of
the signal relative to the preceding position of the receiver 11 (Fig. 2):
(1)
Transducers should have a low frequency (0.5-1.0 MHz) in order to overcome the
attenuation of mortar and to facilitate longer measuring distances 12 •
EXPERIMENTAL
Mortar specimens were produced following the European standard EN 196- 1 13 using
portland limestone cement (Tables 1 and 2). Large slabs 40 mm x 160 mm x 240 mm
were cast and stored in lime water for 28 days. The cured slabs were cut into test
specimens 40 mm x 160 mm wide and approximately 12 mm thick. The freshly cut
specimens were carefully ground on both sides to a thickness of 10 mm and again stored
in lime water at 20 oc. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
could be moved horizontally (parallel to the sample surface). The whole assembly was
fixed on a 3-axis computer manipulator to adjust the probes relative to the samples.
After the sample was placed under water the probes (immersed in water) were laterally
set to a predefined position. The determination of the leaky Rayleigh wave velocity
contained the following steps:
After the ultrasonic investigation, the samples were cut perpendicular to the original
surface and prepared for microstructural examination in the scanning electron microscope
(SEM).
The measurement of the sand blasted sample with a comparable surface roughness to the
sample affected by 0.5 molar sulphuric acid did not result in a significant changed surface
wave in velocity compared to the reference sample.
The new technique using low frequency (1 MHz) ultrasonic surface waves has been
successfully verified for the investigation of degradation processes on mortar samples.
These first experiments demonstrate a significant reduction of the leaky Rayleigh wave
velocity by up to 7.5% when mortar is exposed to sulphate solutions of very low pH.
Increased surface roughness did not affect the velocity of the leaky Rayleigh wave.
Even though the exposure was rather short the examination of the microstructure gave
evidence of chemical alteration down to approximately 1 mm beyond the surface
(depletion of calcium, uptake of sulphate). The microstructure of the hardened cement
paste did not show signs of deterioration like increased porosity or microcracking. It is
therefore concluded that small scale alterations like dissolution of portlandite taking place
over rather large scales (mm) in the mortar correlate with the reduced physical integrity
detected by the new ultrasonic method.
The current PhD-project will use this non destructive ultrasonic method (leaky Rayleigh
wave) in parallel with other methods to monitor the physical properties of mortar samples
during exposure to sulphate solutions over periods of up to 2 years. The reliability of the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACKNOWLEDGMENTS
The authors thank G. Rossler for useful discussions and A. Gruskovnjak, W. Trindler, W.
Knecht and D. Rechenbacher for technical assistance. The project is financially supported
by cemsuisse, the Association of the Swiss Cement Industry.
REFERENCES
3. Taylor, H.F.W. and R.S. Gollop, "Some chemical and microstructural aspects of
concrete durability, in Mechansisms of chemical degradation of cement-based
systems", K.L. Scrivener and J.F. Young, Editors. 1997, E & FN Spon: London.
p. 177-184.
4. Skalny, J., J. Marchand, and I. Odler, "Sulfate attack on concrete." Modem concrete
technology, ed. A. Bentur and S. Mindess. Vol. 10. 2002, London: Spon Press. 217.
5. Romer, M., L. Holzer, and M. Pfiffner, "Swiss tunnel structures: concrete damage by
formation of thaumasite." Cement & Concrete Composites, 2003: p. 1111-1117.
7. Neuenschwander, J., Th. Schmidt, Th. Luthi and M. Romer, "Leaky Rayleigh wave
diagnostics of mortar." Ultrasonics (in preparation).
9. Viktorov, I.A., "Rayleigh and Lamb waves." Plenum, New York, 1967.
I 0. Piwakowski, B., "Generation of Rayleigh waves into mortar and concrete samples.''
Ultrasonics, 2004. 42(1-9): p. 395-402.
11. Titov, S.A., R.G. Maev, and A.N. Bogatchenkov, "V(x) Ultrasonic material
characterisation systems employing pairs of line-line and line-point focused
transducers." Review of Quantitative Nondestructive Evaluation, 2004.
23 (American Institute of Physics): p. 1382-1389.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 567
12. Goueygou, M. et al., "NDE of two-layered mortar samples using high-frequency
Rayleigh waves" Ultrasonics, 2004. 42: p. 889-895.
CEM 11/A-L 32.5 R 19.9 4.7 2.6 63.8 1.7 0.95 0.01 3.1
Mineralogical
c1s c2s C1A c~
composition
Leaky Rayleigh
37.2 37.5 40.8 37.5
angle 9LR [0 ]
Velocity ofleaky
Rayleigh wave cLR 2.45 2.39 2.27 2.44
[km/s]
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
liquid (water)
Fig. 1 Schematic situation of the mortar surface immersed in water and the
arrangement of the ultrasonic probes, modified from Reference 7.
Transmitter Receiver
y-axis
x-axis
mortar sample
Fig. 2 Determination of the velocity of the leaky Rayleigh wave by moving the
receiver along the x-axis.
10 15 20 25 30
Position (mm)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
:? 2.5
E
~
l:' 2.4
·c:;
0
Gi
> 2.3
.c
Cll
..
Gi
>.
a:
2.2
..
~
Cll
...I
2.1
Fig. 4 Leaky Rayleigh wave velocity cLR (mean value and standard deviation) for
mortar samples.
:g
.....
~
...co
0
E
.,..
0
Q.l
Q.
E
co
Ul
Q.l
u
c
e
J!!
e
A
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
FLOW DIAGRAM
High early strength portland cement was used; the physical properties and chemical
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
574 Hamada et al.
composition of the cement are given in Table 1. Mixing water was potable tap water.
River sand and gravel were used. The density, absorption, and fineness modulus of the
fine aggregate was 2.65 g/cm 3, 2.30%, and 2.84, respectively. The values were 2.75 g/cm 3
1.04%, and 6.80, respectively for coarse aggregate. Mixture proportions of concrete are
presented in Table 2. Cylinder specimens of 1OOmm in diameter and 200mm in length
were made. The compressive strength of the control specimens was 51 MPa at 3 days, 60
MPa at 7 days, and 69 MPa at 28 days. The mixture proportion was similar to the typical
mixture proportion used for prestressed concrete. Steam curing was applied. The process
of steam curing was: 1) Specimens were kept in the casting room for 3 h after making the
specimens; 2) Air temperature was raised at the rate of 15 °C/h, and kept constant at 60 °C
for 4 hours; and 3) Air temperature was lowered gradually to 20 °C.
After lowering the temperature to 20 °C, the specimens were cured for 28 days at
relative humidity more than 90%. Then the specimens were exposed to the marine
submerged, tidal, and atmospheric exposure zones. The latitude and longitude of the
exposure site were 35"N and 138°E, respectively. The amount of rainfall at the exposure
site was 1650 mm/year. The average monthly temperature varied from 6 to 30 °C at the
exposure site in a year. The exposure site was free from freezing-and-thawing effect. The
tidal pool is located beside the sea. Seawater was directly pumped into the pool from the
sea and drained out to the sea automatically at 6 hours intervals. The marine atmospheric
zone is located about lOrn from the sea and above 5m above the mean seawater level. The
physical properties and chemical composition of seawater are given in Table 3.
Methods of Evaluation
The appearance of the specimens was checked frequently to observe the formation
of cracks on the surface of the specimens. The compressive strength of the specimens was
tested at 3, 14, and 28 days. At the age of 30 years, pulse velocity and compressive
strength were tested. The ultrasonic pulse velocity was measured at the center of the
specimens (frequency of the pulse was 54 kHz).
The mineralogy of the mortar portion and aggregates of deteriorated and sound
concrete was investigated at 15, 27 and 30 years of exposure by X-ray diffraction (XRD)
analysis. The fractured surfaces of the sound and deteriorated regions (matrix and
aggregate-matrix interfaces) were investigated by scanning electron microscope (SEM).
The microstructure of the mortar portions at the outer and inner regions of the specimens
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
was also evaluated by Mercury Intrusion Porosimetry (MIP).
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 575
Appearance of the Specimens
Cracking over the specimens was identified after 5~6 years for tidal and submerged
specimens. The appearance of the specimens at 15 years of exposure is shown in Fig. 2.
The specimens exposed to the atmospheric exposure show minute hair cracking without
any advance stage of deterioration. In the tidal exposure, the extent of deterioration
varied from severe cracking on the specimens, to moderate cracking. In the submerged
zone, the extent of deterioration varied from complete disintegration of the specimens, to
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
moderate cracking. Lateral expansion as well as distortion of the specimens was clearly
found for some specimens. The submerged specimens were deteriorated very quickly
compared to the tidal specimens. At 30 years of exposure, some specimens of the tidal
exposure showed severe lateral expansion. Some tidal specimens showed moderate
deterioration. Interestingly, no significant deterioration was noted for atmospheric
exposure. The above-mentioned observations indicate that deterioration of the specimens
strongly was influenced by the type of marine exposure; severe for submerged exposure,
moderate for tidal exposure, and no significant influence for atmospheric exposure.
The cut surfaces of the specimens are shown in Fig. 3. Severe cracking in the
specimens was observed due to the expansion of aggregates with the presence of
seawater at 15 and 30 years of exposure. Cracking in concrete is not necessarily uniform
around the specimens (Fig. 3, Left). It is due to the variation of the distribution of
reactive aggregates in the specimens. The mortar portion of the deteriorated region
becomes relatively softer but not necessarily porous due to the deposition of the new
products, such as ettringite, calcite, brucite, etc. in the pore structures. The results related
to this deposit are explained later. White deposit around the aggregate at the deteriorated
region was clearly observed. The deposit was confirmed as ettringite, based on the SEM
investigation. The central portion of the specimens remained as normal concrete.
The remaining specimens of the tidal and atmospheric zones were tested by pulse
velocity at the age of 30 years. Some of them were crushed to evaluate the compressive
strength. Compressive strength of concrete varies widely for tidal exposure. In
atmospheric exposure, the average strength loss of the specimens was about 17%
compared to the 28-day strength. Contrary, in tidal exposure, the pulse velocity varies in a
wide range from 2.1 km/s to 5.09 km/s; 10% of the data show a strength value less than 4
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
576 Hamada et al.
MPa, 50% of the data about 17 MPa, and 90% of the data below 52 MPa. The strength
and pulse velocity data clearly indicate that the deterioration of the specimens was
strongly influenced by the exposure conditions.
Compressive Strength and Pulse Velocity (30 Years Marine Exposure Plus 4 Cycles
of Artificial Exposure)
The XRD chart of the aggregates (after ground to a powder) collected from the
severely deteriorated region of the specimens after 15 years of exposure is shown in Fig.
4. Peaks for laumontite mineral are clearly found. The SEM micrographs of the fractured
region of the aggregates are shown in Fig. 5. Bands of laumontite minerals are clearly
identified in the aggregates.
Aggregate-matrix interfaces are shown in Fig. 6 for the atmospheric and tidal
specimens. Densely populated ettringite crystals were found at the aggregate-matrix
interface at the outer region of the tidal specimens (b)-(d). No significant amount of
ettringite was found at the interfaces at the inner region of the tidal specimens as well as
the outer region of the atmospheric specimens (a) and (e). White deposit was clearly
found surrounding the aggregates. The deposits were confirmed as ettringite (d). The
aggregates at the outer region of the tidal specimens lost its adhesion with the matrix due
to the deposition of ettringite surrounding the aggregate.
Experimental Procedure
After 27 years of exposure, a plan was made to investigate the aggregates further.
Unfortunately, no source of fresh aggregates containing laumontite was found. Therefore,
coarse aggregates were collected from the concrete specimens of age 27 years and
crushed into fine aggregates. The experimental procedure is shown in Fig. 7. Aggregates
were collected both from severely expanded specimens as well as less expanded
specimens. Using the recycled reactive fine aggregates, mortar specimens (40x40x 160
mm) were fabricated and exposed under seawater and tap water at temperature 40°C (for
investigating expansion caused by the laumontite minerals). The length change of the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
specimens was measured till the age of 416 days. The mineralogy (by XRD) and
microstructure (by SEM and MIP) of the mortar portion were investigated after 416 days.
Ordinary portland cement and blast furnace slag cement were investigated. Curing
condition was normal curing and steam curing. Water to cement ratio of the mortar was
0.5. Table 5 presents the orthogonal array for this experimental work. The factors tested
are: 1) Type of aggregates collected from severe expansion or no significant expansion, 2)
Cement type, 3) Curing (Steam curing or water curing) and 4) Exposure condition
(Seawater immersion or tap water immersion). The alkali-silica reactivity of the
aggregates was tested by mortar bar expansion and chemical method. Uranyl acetate
solution was sprayed over the cut surface of 27 years old specimens and observed under
ultra violet ray. The presence of alkali-silica rim surrounding the aggregates was also
observed visually.
Experimental Results
No cracks over the specimens were found after 416 days of exposure. The clear
presence of white deposit on the specimens was found over the specimens exposed under
seawater. Fig. 8 shows the time-dependent length change of mortar specimens exposed
under tap water (bottom) and seawater (upper). Fig. 9 shows the "F-value" obtained
based on the analysis of variance. The expansion of the specimens at 416 days was about
0.055% for exposure under seawater and 0.01% for exposure under tap water. SEM and
mineralogical investigation confirm the formation of ettringite for exposure under
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
578 Hamada et at.
seawater. Based on the results of expansion tests, the influence of seawater was clearly
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
understood. The use of fine aggregates may help to develop a uniform expansion and
therefore eliminating the local distress. Relatively less pore volume was found in the
mortar for the exposure under seawater. No alkali-silica reactivity of the aggregates was
confirmed in any of the tests related to ASR. Due to the space limit, these results are not
included in this report.
It was clearly understood that the expansion of the specimens is not associated with
the alkali-silica reaction in concrete. Delayed ettringite formation (DEF) in concrete is
reported, due to the late formation of ettringite caused by the decomposition of ettringite
at higher temperature, or the slow hydration of sulfate in clinkers [6][7][8][9]. In this
study, it was found that the deterioration of the specimens varied in a wide range, from
complete disintegration to moderate deterioration of the specimens in the submerged zone
Therefore, it cannot be concluded that the expansion is associated with a problem related
to the cement. If the expansion is caused by the cement, the expansion should have been
observed for all of the specimens, more or less uniformly for each exposure zone. No
noticeable cracks were found in the atmospheric exposure, even though the total amount
of rainfall at the exposure site was 1650 mm/year (the specimens were under open sky).
Therefore, the possibility of expansion due to DEF cannot be involved.
Lack of CH and CSH at the outer region of the specimens indicates the possibility of
sulfate attack. Looking at the surface of the specimens, it was found that the expansion is
caused by the aggregates. Conventional chemical sulfate attack of the specimens after
cracking by expansion could not be neglected, as microcracking in concrete was
supposed to be a prerequisite for sulfate attack [7]. The expansion was not noticeable in
atmospheric exposure. The seawater accelerated the formation of ettringite in this special
case. Further studies are strongly recommended.
CONCLUSIONS
REFERENCES
6. Taylor, H.F.W., Famy, C., and Scrivener, K.L., Delayed Ettringite Formation, CCR,
31,200l,pp.683-693.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
7. Mehta, P.K., Sulfate Attack on Concrete: Separating Myths From Reality, ACI
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
580 Hamada et at.
Concrete International, August 2000, pp. 57-61.
9. Hime, W.G., and Mather, B., Sulfate Attack or Is It, Cement and Concrete Research,
Vol. 29, 1999,pp. 789-791.
Table 4 -- Compressive Strength and Pulse Velocity Before and After Accelerated
Exposure
Pulse Velocity and Compressive Strength
Specimen
Before Exposure After Exposure
Tidal 4.8 km/s, 32 MPa 4.9 kmls, 35 MPa
Atmospheric 4.5 kmls, 55 MPa 4.5 kmls, 55 MPa
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
No Significant
5 Normal Normal Seawater
Expansion
No Significant
6 Normal Steam Tap Water
Expansion
No Significant
7 Slag Cement (Type B) Normal Tap Water
Expansion
No Significant
8 Slag Cement (Type B) Steam Seawater
Expansion
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Mineralogy of the
aggregates
Conclusions
(c) Submerged
• Laumontite
o Quartz
5 10 15 20 25 30 35 40
29 (degree)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2. Grinded to fine.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
j 0.040
~CueS
.a 0.030
j' 0.020
<.I
0.010
0.000 -=--_.__ _.__ _._______.__....__..__.....__ _,
0 50 100 150 200 250 300 350 400 450
'I1me (Days)
0.070 . - - - - - - - - - - - - - - - - - - - - - - ,
0.060 - -.!r-Case 4-----------------------------------------
_ ~Case6 ---------------------------------
~Case7
50 ~ ~ ~ ~ ~ ~ ~ ~
'I1me (Days)
Fig. 8- Expansion of the Mortar Bar (Top: Seawater, Bottom: Tap water)
__ curing
---Cement
-A- Initial Curing
35
...II> -e-- Exposure
·2=
01
30
....;.. 25
=
.[!l
~
01
20
=
01
15
=
=
'"0
~ 10
01
.c
II>
5
..e01
...or 0
1 61 122 183 245 306 367
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
(CIC) at LEMIT, La Plata. He conducts several researches on durability of concrete and
is authored numerous papers, especially in the field of alkali-aggregate reaction.
INTRODUCTION
In the field exposition, the durability of concrete structures is determined by the coupled
action of various physical (i.e. drying, soaking, temperature cycles, overloads) and chem-
ical (i.e. carbonation, structure densification) processes; which are complex in nature and
progressive with the time. To analyze the coupled action of these parameters, long time
durability tests on concrete are needed because the accelerated laboratory tests under
controlled conditions can help to predict the behavior of mortar or concrete mixtures but
they cannot be directly extrapolated to predict the performance of concrete structure.
During the last decades, the cement composition has changed. There has been an increase
in reactivity of clinker at early ages due to high C 3S content, the use of mineralization
substance and large amount of particles less than 45 J..Lm. This increase permits the
incorporation of large proportions of mineral admixtures in blended cement composition;
which is promoted due to ecological and economical factors of concrete industry.
The sulfate performance of concrete containing high volume mineral admixture cement
could be affected because it has an increased sensitivity to magnesium solution and
physical salt attack. It is known that strong magnesium sulfate solutions were generally
more aggressive than the equivalent concentration of sodium sulfate; and damage is more
severe for concrete containing a high volume or highly reactive mineral admixtures due
to the drastic reduction ofCH content that accelerates the degradation ofC-S-H [3]. Tori
and Nakamura [4] reported a poor performance of concrete containing 50 and 70% of
low calcium fly ash. Kollek & Lumely [5] have reported that cements with more than
50% of slag-content are vulnerable to magnesium sulfate attack.
For permeable soils and arid climatic conditions, high volume mineral admixture cement
may also suffer the severe attack due to distresses caused by salt crystallization at the
soil-air interfaces or in slab on the grade. To mitigate this attack, Haynes et a!. [6]
propose low permeability mixtures (w/c = 0.45) and the use of mineral admixture such as
fly ash. However, Stark [7] reports that the combination of Type II or Type V cement
with 40% of pozzolan (fly ash and Monterrey shale) produced no significant
improvement of sulfate resistance of half-buried beams. He suggests that conversion of
thenardite to mirabilite caused the distress of specimens above the ground. Arabian
experiences, cited by Osorbone [8], showed that high slag cement-concretes (70% slag)
are more susceptible to scaling or salt weathering phenomenon observed along the Gulf
coast. Using a laboratory drying and wetting test, de Almeida [9] reported that high
strength concretes (w/c = 0.36 and 0.40) containing fly ash or natural pozzolan showed a
visual deterioration and retrogression of compressive strength after 120 cycles in sulfate
solution. <;olak [ 10] using a wetting and drying test, similar to proposal in ASTM C 88
for aggregate, found that the performance of cement paste decreased considerably when
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
This paper summarizes 15-years of results related to the effect of high volume mineral
admixture on the sulfate resistance of concrete. Based on the macroscopic and
microscopic behavior, a degradation mechanism of concrete is discussed with emphasis
on the physical and chemical attack produced by sulfate soils.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
A normal portland cement (OPC - Type I) and a low-C 3A portland cement (SRPC -
Type V) were used in this field test. The chemical composition and physical
characteristics of these cements are provided in Table 1. Low calcium fly ash, natural
pozzolan and granulated blast furnace slag were used to produce the blended cement.
Pozzolan cement was obtained by the combination of OPC with 40 % of fly ash or
natural pozzolan; while slag cement was formulated containing 80% of granulated blast
furnace slag. The principal characteristics of active mineral admixtures used are also
reported in Table 1.
In all concrete mixtures, the same set of aggregates was used. Coarse aggregate was
granite crushed with a maximum size of 25 mm, and fine aggregate was natural river
sand with a fineness modulus of2.75.
The composition of the different concrete mixtures is given in Table 2. For these
mixtures, the w/cm was 0.52±0.02 and the amount of cementitious material was
300 ± 20 kg/m 3 . The sand-coarse aggregate ratio was kept constant (0.42).
For each mixture, different types of specimens: cylinders (150 x 300 mm), prisms (70 x
100 x 400 mm) and beams (150 x 150 x 900 mm) were cast. Specimens were cured for
one-day (two-days for H7-0PC+80%S concrete) in molds and then stored in moist
cabinet for 27 days. At 28-days age, the initial properties of specimens were determined
and then they were transported to the experimental field.
MICROSTRUCTURAL ANALYSES
After fifteen years, the mineralogical and petrographical properties of concrete were
identified using optical microscope and XRD analysis. Optical microscope examination
was carried out on the polished thin section obtained from prismatic specimens using
transmitted light (plane polarized and cross polarized). For this purpose, a slice was cut at
the center of the prism with a diamond saw and then it was impregnated with low
viscosity resin. Analyzed thin section had a width of 20 mm and 70 mm depth from the
cast-surface to the center of specimens. To obtain the compound profile using XRD,
other slice of the same prisms was sawed using a diamond disk. Subsequently, samples at
0, 5, 10 and 20 mm depth were obtained using the wearing device. Powder samples were
obtained by wearing perpendicularly to the cutting surface and evading the material at the
same distance of both faces of the prism. Material from each layer was collected and
sieved through a 45 J.iffi and stored in desiccators to prevent carbonation before XRD
analysis. XRD measurements were performed on Philips X' Pert diffractometer equipped
with a graphite monochromator using CuKa radiation and operating at 40 kV and 20 rnA.
Step scanning was used with a scan speed of2°/min and sampling interval of0.02° 28.
Visual appearance
Fig. 1 shows the evolution of the visual aspect of cylindrical specimens cast with H1, H4,
H6, and H7 concrete. All half-buried specimens showed a progressive spalling of
concrete surfaces up to soil level and this deterioration is due to salt recrystallization
distress which is physical nature damage. At the first stages, the outer layer of paste came
off and the surface of specimens had a sandy appearance. Then, the mortar over the
aggregates was removed with a plane of weakness at the paste-aggregate interface
leading to expose coarse aggregates and the mass loss increased. Later, a deeper layer
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
594 lrassar et at.
was subjected to a similar attack and so the deterioration progressed. The extension of
damage caused by physical salt attack could be easily measured on the surface of half-
buried beams (Fig. 2). Exposed aggregates can be observed above 20 mm of soil level for
H8 concrete; while for concrete containing high volume of mineral admixtures the height
affected varied from 100 to 250 mm. In all cases, the most severe deterioration occurred
at soil-air interface (see detail in Fig. 2).
In the buried zone of cylinders (Fig. 1), HI-concrete specimens showed a progressive
deterioration leading to the complete destruction after ten years. For more detailed
observation, Fig. 3 shows the visual appearance of buried prisms of HI, H4, H6, H7 and
H8 concrete after I5 years. For HI-concrete, the deterioration process firstly showed
cracks running along the edge of the prisms and propagating to the comers at two years
inspection. Then, the pull out of comers and the swelling on surface of specimens due to
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the formation of expensive compounds could be observed, while the interior of
specimens remained cohesive. Finally, there was a large mass loss of concrete specimens
due to the softening and the loss of cohesiveness of concrete, leading to a granular mass
easily to break with the hand. On the other hand, concretes containing 40% of fly ash
(H4) or natural pozzolan (H6), high slag cement (H7) and SRPC (H8) showed no
evidence of cracking at the edges after 15 years in sulfate soils. For this type of
exposition, this behavior is a typical chemical attack caused by sulfate on the concrete.
Mass loss
The mass loss data measured on cylinders and prismatic specimens are showed in Fig. 4.
From these data, it appears that concrete with high volume of mineral admixture
presented a large mass loss until five years in accordance with the evolution of visual
appearance of top-level in cylindrical specimens; while a slight rate of mass loss was
measured in prisms specimens. On the other hand, HI-concrete (OPC) showed a drastic
mass reduction after 7 years due to the damage produced by chemical sulfate attack that
disintegrated specimens. For H4 and H6 concrete, the mass loss in these specimens was
produced at the face exposed to air.
Microstructural observation
Deterioration of concrete on the surface above the soil level is characterized by spalling
the concrete and the presence of white efflorescence. Fig. 7 shows XRD analysis on
samples of the outer layer of beams after seven years. It shows strong peaks of thenardite
as product of salt crystallization into the pores of concrete and gypsum was also detected
for all concretes. However, ettringite could not detect in these XRD-pattems. This
evidence demonstrated that the salts crystallization is the main attack on concrete up to
soil level. According to Novak and Conville [15], thenardite is the salt deposit that occurs
when concrete is exposed to air.
For buried concrete, the main findings observed on specimens of Hl (OPC) and
H4 (OPC+40%FA) concrete using the petrographic and the XRD analyses are presented.
Both concretes were selected as representative from macroscopic behavior of concretes
studied.
Fig. 8a shows the thin section of HI-concrete from the surface of the specimen to 2.5 mm
depth. It is observed that cement paste (Pc) is totally carbonated and it has a dark aspect,
in this region the cracks and the interfaces of pasta-aggregate are partially or totally filled
with a fibrous microstructure compounds (let- ettringite-thaumasite) and some air voids
are partially or totally filled by a smooth gray compound (gypsum). The calcite crystals
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
596 lrassar et al.
are also located on the wide cracks. As described by Daimond [16], the extensive
deposition of attack compound, especially ettringite, occurs inwards the carbonated zone
of concrete. The XRD analysis at surface region (Fig. 9) showed strong and well defined
peaks of ettringite (E) and thaumasite (T) compounds, the complete depletion of calcium
hydroxide (CH) and formation of calcite (CC). Thaumasite was clearly identified using
XRD by its characteristic peak at 28 = 19.35°, confirming as previous works [16-18] that
it would form at room temperature as the last state of sulfate attack, when the complete
depletion of CH paste produced the breakdown of C-S-H and led to thaumasite formation
in the carbonated layer.
From 2.5 mm to 16.5 mm, a grayscale gradient across the paste was observed, indicating
the progressive diminution of carbonation. At large magnification, the paste showed
morphology composed by dark and bright areas. This morphology, called "salt and
paper" by Diamond [ 16], corresponds to no-uniform decalcification of paste: darker areas
are depleted of calcium and the brighter areas have Ia large proportion of calcium. In this
region, voids (Po) were filled with compounds having a fibrous morphology (Fig. 8b).
From 2.5 to 11 mm (Fig. 8c), an extensively cracking (Mf) parallel to cast face of prism
was detected. These cracks ran along the paste (Pc) surrounding the fine and coarse
aggregates. Cracks were completely filled with small needle compound and crystals
growth perpendicular to their walls indicating that this compound filled the new space
created by sulfate attack [19]. At 16.5 mm depth, the paste showed a gradual increase of
brighter point attributed to CH, and its deterioration degree decrease significantly.
At 5 mm depth, XRD patterns of concrete showed the presence of ettringite and gypsum
(G) due to sulfate attack, but only some traces of calcite could be detected. At 10 mm,
ettringite was detected and the presence of calcium hydroxide (CH) was confirmed by its
main peaks. For the next layer sample, XRD analysis showed an increase of CH peaks
and main compounds of sulfate attack could not be detected. These observations are
consistent with the sulfate attack mechanism described by Irassar et a!. [20] on
ASTM C 1012 mortar prisms containing Type II cement. In this case, the ingress of
groundwater containing sulfate led first to the formation of ettringite, later to gypsum
deposition and finally to thaumasite formation when the decalcification of mortar leads to
the breakdown ofC-S-H provides reactive silica. In this case, the formation of thaumasite
occurred in the carbonated region from decomposition of the ettringite first formed at
ambient temperature.
For H4 concrete (OPC+40%FA), the thin section analysis showed an altered zone
(especially carbonation) from the surface of prism up to 8 mm depth (Fig. lOa). In this
region, the paste (Pc) was smoothly carbonated and it could be observed that voids and
paste-aggregate interfaces located near surface of prisms were filled with a compound of
low birefringency (slight gray) with morphology as small sheets, and it could not be
identified by petrographic analysis. According to XRD analysis (Fig 11 ), the powdered
sample of the surface showed only the formation of calcite due to carbonation, neither
ettringite nor gypsum was detected for this concrete. At 8 mm depth, the paste of
concrete was homogenous without signs of distress or alteration (Fig lOb) and brighter
compounds with small sheets morphology (HC) located at paste-aggregate interface
could be observed. For 5, 10 and 20 mm layers, XRD analysis showed the same pattern,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CH could not be clearly detected and only a weak wide band at 28 = 34.09° was observed
in the 20 mm - layer. These observations indicate that high volume fly ash concrete has a
low quantity and poor crystallized calcium hydroxide, which provides a good resistance
to sulfate attack because according to Mehta [21 ], the ettringite formation requires a high
calcium concentration to be expansive
1. Combination of portland cement with a high volume of appropriate fly ash, natural
pozzolan, or granulated blast-furnace slag provide an excellent performance for
concrete structures buried in very aggressive alkali sulfate soils. There was no sign or
evidence of sulfate attack after fifteen years, providing sulfate-resisting properties
like SRPC concrete.
2. Concretes containing high volume of mineral admixtures are more susceptible to the
physical salt attack caused by the cycles of wetting and drying that produces a
progressive scaling and mass loss. Most severe deterioration occurs for slag concrete
specimens (half buried specimens - H7) and the surface scaling of beams specimens
decrease when increases the height from soil-air interface; while the extension of
damage was less for high volume fly ash concrete. Vulnerability of this type of
concrete merits special attention for structures exposed to this action, such as
foundation or slab on grade. This factor will be taken into account for the design of
concrete structures providing a well curing practice, a protective barrier or using an
impermeable soil around the structure to diminish the capillary rise on the surface of
concrete.
REFERENCES
4. K., Torii, K., Taniguchi, and M., Kawamura, Sulfate Resistance of High Fly Ash
Content Concrete, Cement and Concrete Research, V. 25, No.4, pp. 759-768, 1995.
5. J.J. Kollek and J.S. Lumley, Comparative Sulphate Resistance of SRPC and Portland
Slag Cements, Proc of Fifth Intern. on Conf. Durability of Building Materials and
Components, J.M. Baker eta!. Eds., E.& F.N. Spon, London, pp. 409-420, 1991.
6. H. Haynes, R. O'Neill and P.K. Mehta, Concrete Deterioration from Physical Attack
by Salts, Concrete International, V. 18, No 1, pp. 63-68, 1996.
9. I.R. de Almeida, Resistance of high strength concrete to sulfate attack: Soaking and
drying test, Proc.Second International Conference on Durability of Concrete SP-126,
American Concrete Institute, Detroit, pp. 1073-1092 (1991).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
11. E. F. Irassar, A Discussion of the Paper "Characteristics of pastes from a Portland
cement containing different amounts of natural pozzolan" by Adnan <;olak. Cement
and Concrete Research, V. 34, No 7, pp. 1265-1266,2004.
12. E.F. Irassar, A. Di Maio, and 0. Batie, Durability of Concrete Containing Mineral
Admixtures in Sulfate Soil, Proc. Second International ACI-CANMET Conference on
Durability of Concrete, Supplementary papers, American Concrete Institute, Detroit,
1991.
13. E.F. Irassar, A. Di Maio, and 0. Batie, Sulfate Attack on Concrete with Mineral
Admixtures, Cement and Concrete Research, V. 26, No 1, 113c123, 1996.
17. P.W. Brown, R.D. Hooton, and B.A. Clark, The Co-Existence of Thaumasite and
Ettringite in Concrete Exposed to Magnesium Sulfate at Room Temperature and the
Influence of Blast-Furnace Slag Substitution on Sulfate Resistance, Cement and
Concrete Composites, V 25, pp. 939-945,2003.
18. E.F. Irassar, V.L., Bonavetti, M.A. Gonzalez & M:A: Trezza, Thaumasite Formation
in Limestone Filler Cement Exposed to Sodium Sulphate Solution at 20 °C, Cement,
Concrete and Composite, V 27, No 1, pp. 77-84, 2005.
21. P.K. Mehta, Sulfate Attack on Concrete -A Critical Review, Materials Science of
Concrete III, J. Skalny Ed., American Ceramic Society, Westerville, USA, pp. 105-
130, 1992.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Natural
OPC SRPC Fly Ash Slag
Pozzolan
Chemical composition,%
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
D.
z
~
~
c.>
0
.
D.
CD
::J:
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-200 -200
01 01
:i-300 :i-300 +40%NP
,g ,g
Ul -400 Ul-400
Ul Ul
Ill Ill
:E -500 :E -500 H1-0PC
H1-0PC -600
-600 H7-0PC
+80%Siag Prismatic specimens
-70~ -700
0 6 8 10 12 14 16 18 0 6 8 10 12 14 16 18
Age, years Age, years
1.10 1.10
Cylindrical specimens Prismatic specimens
Ul en
:::J :::J
:; :;
"0 1.00
0
"0 1.00
0
H8
E H8 E H4
u H6 H6
·e
Ill 0.90
H4 ·eu
Ill 0.90
H7
c:: c::
>- >-
"0 H7 "0
~ Gl
~
0.80 0.80
i
Gi
D:: H1 & H1
0.70 0.70
0 4 6 8 10 12 14 16 18 6 8 10 12 14 16 18
Age, years Age, years
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-..._-=:::a:-----.:::a.~~
H6
Prismatic specimens
4.4 ..___.__,____.._.__...__.__.__....___j
0 2 4 6 8 10 12 14 16 18 0 2 4 6 8 10 12 14 16 18
Age, years Age, years
5.2 . - - - - - - - - - - - - - ,
Beam specimens
H7
0 2 4 6 8 10 12 14 16 18
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Age, years
Fig. 6- Ultrasonic pulse velocity measured at the center of cylindrical, prismatic, and
beam specimens.
C) l! 8 :i u
~ .:t
<!I
IJ u
:!
~
~ ~ ~:f I)
H8-SRPC
~
8 10 12 14 16 18
.......
20
W!~ ~~ !I
22 24 26 28
H4-0PC+
40%FA
30 32 34
26, 0
Fig. 7- XRD pattern of samples take on surface of cylindrical specimens above the
soil level (Then a = Thenardite, G =Gypsum, Ag =Aggregate, CC =Calcite).
(c)
Fig. 8- Micrographs obtained from thin cut sample of Ht- concrete: a) Cement paste
with high carbonation (Pc) and space at interface filled with ettringite-thaumasite
compound (let); b) Air void filled with a fibrous morphology compound (Po); and
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
u
H1- OPC u
f-
ui
8 10 12 14 16 18 20 22 24 26 28 30 32 34
0
28,
(a) (b)
Fig. 10- Micrographs obtained from thin cut sample of H4- concrete: a) cement
paste (Pc) with high carbonation (Pc) and; b) Paste with dense (Pc) and calcium
hydroxide (HC) at paste-aggregate interface.
H4-0PC+40%FA !i !i
8 10 12 14 16 18 20 22 24 26 28 30 32 34
28, 0
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Penetration in Lightweight and
Normalweight Aggregate Concrete
Synopsis: This paper presents a comparative study on the water permeability and
chloride penetration in lightweight and normalweight aggregate concrete (LC and
NC). Water-cementitious material ratios (w/cm) of 0.35 and 0.55, and silica fume
content of o% and 10% as cement replacement were used to obtain concrete with
different strength levels. In concrete mixture proportioning, the volume of the coarse
aggregate for the lightweight concrete was the same as that in the corresponding
normalweight concrete.
The results indicated that at the strength level of 30-40 MPa, the water
permeability of the LC with a w/cm of 0.55 was lower than that of the NC with
equivalent w/cm when subjected to 4 MPa water pressure. However, at a
w/cm of 0.35 the water permeability of LC and NC was not significantly different.
The resistance of the LC to chloride ion penetration was similar to that of the
corresponding NC with the same w/cm and mixture proportion. As the LC had lower
compressive strength than that of the corresponding NC, the results indicated that
the LC would probably have lower water permeability and better resistance to the
chloride-ion penetration than the NC with equivalent 28-day compressive strength.
INTRODUCTION
are affected by the aggregate and the cement paste as well as the interfacial zone between
them. In concrete, it is generally the pore structure rather than total porosity that is
essential in establishing the water permeability and chloride penetration. In addition,
micro-cracks in the matrix may contribute significantly to the water permeability and
chloride penetration. Fig. 1 shows schematic differences between the porosity and
penetrability 1• From the figure it is clear that the connectivity of the pore system is
essential for water and chloride-ion penetrability. A material can be porous but still has
low permeability as long as the pores are not interconnected (i.e. closed pore system). On
the other hand, a material can have lower porosity but higher permeability if the pores are
interconnected. Soongswand et aP found that the permeability of concrete made with
porous limestone aggregate was about the same as that made with dense river gravel.
Over recent years much attention has been given to the influence of the interfacial
zone on various properties of the concrete, especially water permeability and penetration
of harmful substances such as chloride ions as they have direct impact on the long-term
durability of reinforced concrete structures. In a recent study 3 for normal weight aggregate
concrete (NC) with a water-cement ratio (w/c) of 0.45 exposed in a tidal pool for 15
years, it was reported that more chlorides were present at the interfacial zones when
viewed under a scanning electron microscope (SEM). This was probably related to a
more porous interfacial zone between the aggregate and cement paste matrix than the
bulk paste that is characteristic of NC at a strength level of -30-40 MPa for a given
cementitious system.
For lightweight aggregate concrete (LC), although lightweight aggregates (LA) are
porous, several investigations have shown that the LC is less permeable than NC<Hi. The
reasons are numerous. The pore system of the LA is partly closed, and some of the inner
core pores are not connected with the outer layer, which has easy access from the
7
outside • This partly open structure allows the absorption of water into the aggregate
particles, and this water can later be used for "internal curing" in the hardening process of
concrete. The result is a well-hydrated and dense interfacial zone around the porous
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 609
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
aggregates, and the rough surface texture of the LA further contributes to an excellent
bond to the matrix. Also, the frequency of micro-cracking in the interfacial zone caused
by drying shrinkage is relatively low due to the lower modulus of elasticity of the LA.
Six concrete mixtures were included in the study, and Table 1 gives the mixture
proportion, slump, density, and 28-day cube compressive strength of the concrete. At
each w/cm, the volume of the coarse aggregate in the LC and NC was the same. For
concrete with w/cm of 0.35, superplasticizer was added in a dosage of 1.5 liters per 100
kg of cementitious materials to improve workability. Ten percent of silica fume was used
in two concrete mixtures with the w/cm of 0.35 as cement replacement. All the concrete
mixtures were prepared by using a pan mixer. In addition, during the preparation of all
the LC mixtures, the amount of water required for 1-hour absorption by the LA was
added to the mixing water. The 1-hour water absorption ofthe LA was determined on as-
received LA with initial moisture of between 6-8% by mass of the oven-dry aggregates.
Prior to each concreting, samples of the as-received LA were used to determine the
amount of water absorption in one hour. The results were then expressed as a percentage
of the mass of the as-received LA samples, and used to calculate the water required for
that particular batch. The average 1-hour water absorption rate of the as-received LA was
about 5%.
ASTM Type I normal portland cement was used for all the concrete mixtures, and its
chemical composition and physical properties are given in Table 2. The silica fume was
dry uncompacted powder from the production of silicon metal with Si0 2 content of
93.6%. Its chemical composition and physical properties are also given in Table 2. The
superplasticizer was a mixture of naphthalene sulfonates and modified lignosulfonates
with a specific gravity of 1.2. The LA used in this study was expanded clay type
(commercially available) with a dry particle density of 1.2 and bulk density of ~650
kg/m3 . The aggregate particle size ranged from 4 to 8 mm. The water absorption of oven-
dry LA at 1 and 24-hour was 9 and 14%, respectively. Coarse normalweight aggregate
used was crushed granite with a maximum nominal size of 10 mm and a specific gravity
of 2.65. Natural sand with the same specific gravity as granite was used in all the
mixtures as fine aggregate. Grading of the sand and crushed granite is given in Table 3.
The slump of the LC determined immediately after mixing was higher than that of
the corresponding NC due to the additional water added in the LC mixtures for
Water Permeability
The water permeability of the concrete is given in Table 4. It was determined using
slightly tapered cylinders with a height of 180 mm. Prior to the water permeability test,
the concrete specimens with w/cm of 0.55 and 0.35 were cured in fog room for 28 and 70
days, respectively. During the test, one end of the cylinder with a diameter of 110 mm
was subjected to a water pressure of 4 MPa, while the other end with a diameter of 100
mm was exposed to atmospheric pressure. At the end of test periods, the water
penetration depth was determined by splitting the specimens, and the water permeability
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
kw = d2 I (2.P.t)
where kw is cofficient of water permeability (m/s); dis penetrated depth (m); Pis pressure
head of water (m); and tis time oftest (s).
The results showed that the water permeability coefficient of the LC with w/cm of
0.55 was almost one order lower than that of the corresponding NC. With the reduction
of w/cm the water permeability of both LC and NC was reduced. The incorporation of
silica fume further reduced the water permeability. However, the water permeability of
the LC and NC at w/cm of0.35 was not significantly different.
The lower permeability of the LC with w/cm of 0.55 compared with that of the
corresponding NC is consistent with the results from several other studies4--6. It may be
attributed to a dense interfacial zone between the aggregate and the cement paste matrix
in the LC9• Also, the bond between the LA and paste matrix appears to be better due to an
improved mechanical interlocking between the aggregate and the cement paste. The NC
with w/cm of 0.55 may have a more porous interfacial zone, which would lead to higher
water permeability.
With the reduction of w/cm and the incorporation of silica fume, the interfacial zone
in the NC improved significantly. Furthermore, similar water permeability of the LC and
NC at w/cm of 0.35 indicated that there was significant improvement of the
microstructure of the mortar matrix. Fig. 2 gives a curve indicating the maximum degree
of cement hydration that can be achieved under ordinary curing conditions 10- 11 .
According to the curve, the higher the w/c, the greater the maximum degree of hydration
is. In the same figure, another curve is presented indicating the degree of hydration
required for the pore structure of the paste changing from an open pore system into a
closed, or blocked pore system. From the figure it can be deduced that closed pore
structure can be achieved in mature pastes with a w/c less than about 0.5, but beyond that
an open pore structure is expected. Hence, in reducing the w/cm ratio from 0.55 to 0.35,
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 611
there is probably a change in the connectivity of the pore system in the mortar matrix, i.e.
from open pore to closed pore structure. The above explains why the water permeability
of the LC and NC was not significantly different.
Chloride Penetration
the concentration of free chloride ion was above 0.15% by mass of cement 12 • Fig. 4
shows the formation of silver chloride precipitate on the split-up specimens immersed in
the NaCl solution. The details of the tests can be found in reference 13.
The results from the ASTM C 1202 test presented in Table 5 indicated that the
average charge that passed through the LC was similar to the NC with the same w/cm and
similar cementitious materials.
From the results shown in Table 6, it is clear that the depth of chloride ion
penetration decreased with the decrease in w/cm, and the incorporation of silica fume
further reduced the chloride penetration. The LC and NC appear to have similar
resistance to chloride penetration. There was a discrepancy showing high penetration
depth for the LC with silica fume (Mix L35SF) as compared with the corresponding NC
at 30- and 60-day tests. At 90 days, however, the depth of chloride penetration was
similar for these two concrete. This might be attributed to higher localized porosity for
the LC specimens used for the 30- and 60-day tests.
As the LC had lower compressive strength than that of the corresponding NC, the
above results indicated that the LC would probably have lower water permeability and
better resistance to chloride ion penetration than the NC with equivalent 28-day
compressive strength.
ACKNOWLEDGMENTS
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3. Mohammed, T.U., Yamaji, T., and Hamada, H., "Microstructures and Interfaces
in Concrete after 15 Years of Exposure in Tidal Environment," ACI Materials
Journal, V. 99, No. 4, 2002, pp. 352-360.
6. Nishi, S., Oshio, A., Sone, T., Shirokuni, S., "Watertightness of Concrete
against Sea Water," Journal of Central Research Laboratory, Onoda Cement
Co., Tokyo, V. 32, No. 104, 1980, pp. 40-53.
9. Zhang, M.H., and Gj0rv, O.E., "Microstructure of the Interfacial Zone between
Lightweight Aggregate and Cement Paste," Cement and Concrete Research,
V. 20, 1990, pp. 610-618.
11. Mills, R.H., "Factors Influencing Cessation of Hydration in Water cured Cement
Pastes," Proc. Symp. on Structure of Portland Cement Paste and Concrete, ACI
SP 90, 1966, pp. 406-424.
13. Chia, K.S., and Zhang, M.H., "Water Permeability and Chloride Penetrability of
High-strength Lightweight Aggregate Concrete," Cement and Concrete
Research, V. 32,2002, pp. 639-645.
0\
~
~
n
:::r
Cii"
cu
:::s
c.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
N
:::r
cu
:::s
Table 1 - Mixture proportions, slump, density, and 28-day compressive strength of concrete CIQ
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0\
....
0\
n
:::r
iii"
Q)
=
Q.,
N
:::r
Q)
Table 4- Results for water permeability
Water permeability I
=
CIQ
28-day cube Length of
Coarse Silica Density of Average Standard Duration Water
w/cm compressive moist
Mix !D. aggregate (effective) fume 1-day cube strength penetration deviation oftest permeability
3 curing
type (%) (kg/m ) (days) coefficient
(MPa) (days) depth (mm)*
(m/s)
'if55 NA 0.55 0 2250 43 28 60 6 12 4.3 X 10· 2 I
1.4 x w- 13
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 617
Table 5- Results for rapid chloride penetrability determined according to ASTM C1202
28-daycube Average
Coarse Silica Density of
w/cm compressive charge Standard Chloride**
MixiD. aggregate fume 1-day cube
(effective) strength passed deviation penetrability
type (%) (kg/m3)
(MPa) coulombs)*
N55 NA 0.55 0 2250 43 5445 1033 High
55 LA 0.55 0 1770 34 5095 444 High
'11"35 NA 0.35 0 2310 78 2290 117 Moderate
35 LA 0.35 0 1835 50 2843 644 Moderate
'IT35SF NA 0.35 10 2295 92 421 42 Very low
35SF LA 0.35 10 1835 56 316 23 Vervlow
* Average from 3 specunens
**According to ASTM C1202
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0\
~
()0
n
:::r
Qi"
Ql
:::s
0.
N
:::r
Ql
:::s
(IQ
----- - ----------- -----.-- ------- ----------- --
28-daycube Immersion Salt nonding
Coarse Silica Density of
w/cm compressive Ave. penetration depth (mm) Ave. penetration depth (mm)
MixiD. aggregate fume !-day cube
(effective) strength Days Days
type (%) (kg/m')
(MPa) 15 30 60 90 15 30
60 90
NA 0.55 2250 -- ..
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~I
b. porous I high penetrability
I Q~o00 ol
c. porous I non-penetrability
m: (j
d. low porosity I high penetrability
Fig. 2- Maximum degree of hydration (amax) in concrete and threshold value for the
degree of hydration beyond which an open pore structure will remain.'a·u
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
r- Clearperplex
e
6
Concrete Specimen Epoxycoatin
PVC pipe
~
I
-'- .
Atmosphere 75 Vo R.H
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
by G.C. Hoff
INTRODUCTION
The benefits of using lightweight aggregates in concrete to help reduce cracking in slabs
and bridge decks has been intuitively known for decades by the lightweight aggregate
industry but the reasons as to why this occurred were not extensively examined and the
benefits were not widely promoted. It was believed, and correctly so, that the lower
modulus of the LWA and the improved transition zone around the LW A particles due to
their generally vesicular surface, helped reduce stress concentrations between the paste
and the aggregate and those reductions subsequently reduced the amount of early-age
cracking in the concrete. In the 1980's, the production of high-strength concrete (HSC)
became more common and, to accomplish it, came the use of higher cement contents,
supplementary cementing materials such as silica fume, fly ash and blast furnace slag
cement, and lower water-binder ratios as a result of the extensive use of high-range water
reducing admixtures (superplastizers). The term "high-performance" concrete (HPC)
also emerged with a focus on providing special properties of concrete above what would
normally be expected from concrete produced for general use. Most of the HPC was
directed at improved durability. The durability improvements came by reducing or
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
eliminating the transport mechanisms of the environment into the concrete and this
generally followed the same modifications of the mixture proportions that occurred for
HSC.
With the mixture proportion changes for HSC/HPC, came a shrinkage problem. Concrete
shrinkage, over time, induces cracking that can severely reduce the life expectancy of
concrete. Long-term drying shrinkage has typically been what is addressed in the
literature and what is considered in structural design. Lately, much more attention has
been paid to early-age shrinkage, as it can be responsible for cracking when the concrete
has not gained significant strength to withstand internal stresses. The components of both
early-age and long-term shrinkage are drying, autogenous, and thermal shrinkage, with
carbonation shrinkage also contributing to the overall long-term shrinkage 1•
The use of HSC/HPC in pavements further complicated the situation by the "fast track"
approach to paving where the concrete gets only minimal moist curing. ACI 308-922
notes that for moist curing pavements and other slabs on the ground at temperatures
above 40°F (5°C), "the recommended minimum period of maintenance of moisture and
temperature for all procedures is 7 days or the time necessary to attain 70% of the
specified compressive or flexural strength, whichever period is less." This curing period
may have to be extended when fly ash is included as part of the binder because of its
slower initial strength development. Also, when silica fume and other pozzolans are
used, the water demand is greater. For structures and buildings, ACI 308-92 2 notes that
the use of liquid membrane-curing compound not be authorized "when the concrete has a
water-cement ratio of 0.4 or less by weight." This implies that even by not allowing the
At early ages (the first few hours), before the concrete has formed a hardened skeleton,
autogenous shrinkage is the result of only chemical shrinkage. At later ages (> 1+ days),
the autogenous shrinkage can also result from self-desiccation since the hardened
skeleton resists the chemical shrinkage. Self-desiccation is the localized drying resulting
from a decreasing relative humidity (RH) 3.4. The lower humidity is the result of the
cement using available water for hydration.
In a HSC/HPC with a low w/c, the finer porosity causes the water meniscus to have a
smaller radius of curvature. These menisci cause a large compressive stress on the pore
walls, thus having a greater autogenous shrinkage as the paste is pulled inwards. Self-
desiccation is only a risk when there is not enough localized water in the paste for the
cement to hydrate; thus the water is drawn out of the capillary pore spaces between the
solid particles. At later ages, a strong correlation exists between internal relative
humidity and free autogenous shrinkage.
In 1991, Philleo 5 suggested incorporating saturated lightweight fine aggregate into the
concrete mixture to provide an internal source of water to replace that consumed by
chemical shrinkage during hydration of the paste. LWAs typically have 24-hour
absorptions in the range of 5% to 25% and, if properly preconditioned prior to their
introduction into the mixture, can provide additional internal water for curing the
concrete. His suggestion finally gained some recognition in the mid-1990's with a
considerable amount of work being done on the use of saturated L WA to alleviate
autogenous shrinkage. Many of these studies are described in later sections of this report.
The use of mineral admixtures, such as fly ash and silica fume, tend to refine the pore
structure towards a finer microstructure. If there is a finer microstructure, the water
consumption will be increased and the autogenous shrinkage due to self-desiccation will
be increased.
At the surface of the concrete, an additional humidity gradient occurs due to evaporation
from the concrete surface. This accelerates the appearance of the localized humidity
gradient. The water from the L W A near the surface is then used up faster than in the
interior of the concrete thus causing the near-surface layer of the concrete to become
denser in a shorter period of time. This helps reduce the amount of water that would
normally evaporate and contributes to improved internal curing of the concrete. It also
leads to reduced stresses or no stresses due to drying, and helps eliminate surface
cracking.
As early as 19487, it was shown that autogenous shrinkage due to self-desiccation occurs
when the w/c is below 0.42 as all the mixing water is consumed at this ratio. Other
investigations noted that the w/c limit can vary between 0.36 and 0.48 depending on
cement type. When the w/c is much lower than 0.42, and can no longer gain curing
water, the cement seeks extra water from the internal pores and thus lowers the relative
humidity. More recently, Mather and Hime 8 have noted that a lot of modem "high-
performance" concrete, made with water/cement ratios (w/c) below 0.4 (by weight) will
not have all the original mixing water-filled space filled with hydration product. In
concluding this, they too observed that the critical feature of the chemical reaction
between the constituents of cement and the mixing water is the ratio of the volume of the
water to the volume of the cement. In a simplified form, if that ratio is 1.2, then all the
water and all the cement can combine, and the hydration product can fill all of the space
originally occupied by the mixing water (the original mixing water-filled space) 8 • This
ratio is approximately equivalent to a water/cement ratio (w/c) of 0.4 (by weight). If the
wlc is lower than 0.4, some of the cement will always remain unhydrated. They further
noted that about 0.2 wlc goes into chemical combination with the cement and that the
other 0.2 wlc was the amount needed to fill the gel pores with the hydration product. It
was also observed that, in practice, externally available water on low w/c concretes does
not penetrate very far into the concrete surface.
Bentz and Snyder 10 developed equations to estimate the replacement level of saturated
LWA fines that would be needed to provide all of the water needed for the complete
curing of a HPC. Their calculations assumed that there was no water exchange between
the concrete and external environment (sealed curing). From computer simulations, they
addressed the proximity of the cement paste to the surfaces of the LWA fines using the
"protected paste volume concept" for air-entrained concrete where the interest is in the
volume of paste within a given distance of an air void surface. Using a previously
developed three-dimensional continuum micro-structural model for concrete, in which
they ignored the interfacial transition zone and focused only on a paste shell surrounding
the LWA particles, they were able to determine the relative proximity of the cement paste
to the additional water source contained in the LWA. From several simulations, using
various percentage replacements of the normal weight fine aggregate by its saturated
LWA fines counterpart and two aggregate gradations, they concluded that a well
dispersed system of small saturated LWA fines would be very beneficial to the curing of
field concrete, much in the same manner that a well dispersed system of small entrained
air bubbles is beneficial in protecting concrete from damage due to freezing and thawing.
SUPPORTING RESEARCH
The collective experiences for many of these variables are summarized in the following
sections.
Various types of lightweight aggregates have been used in these studies. They included
pozzolanic scoria and tuff20 , expanded, pelletized clay (LECA 21 or
6 11 12 13 14 15 16 17 18 19 18
Liapor · · · · · · · · · ), a pelletized, sintered fly ash (Lytag ), a finely ground perlite
with a binder and gas forming agent that is pelletized and coated 22 , expanded shale that is
crushed 22 ·23 ·24·25 ·29·30 and may be coated22 , and expanded slate 26 '27 . All of them, when
properly moisture preconditioned and used in sufficient amounts, can make a positive
contribution to the internal curing of the concrete.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
As noted previously, concrete made with a low water/binder ratio will experience self-
desiccation. This self-desiccation will be accompanied by a drop in relative humidity
(RH) within the concrete and by autogenous shrinkage. If normal weight aggregate is
replaced wholly or partially by saturated lightweight aggregates, when the RH in the
hardening paste begins to drop, a moisture flow begins from the LW A to the drying paste -
thus increasing the RH and reducing the self-desiccation and accompanying autogenous
shrinkage. The efficiency of the LWA for this behavior depends on the preconditioned
moisture content of the LW A at the time of hatching, the size and amount of the
aggregate and its subsequent distribution.
To better understand this moisture flow, the Dutch 17 looked at the transport mechanism of
the water in the aggregate into the paste using a black ink solution with white cement as a
binder. The preliminary findings indicated that the ink solution progressed at least 1-mm
from the aggregate within two weeks of casting. This is indicative that liquid transport
does occur, but because the ink molecules may not diffuse with the same speed and to the
same depth as water, it does not provide quantitative values for the water migration. If it
is assumed that the distance of water transport is on the millimeter scale, that is an
argument for using smaller, more closely spaced particles to achieve more uniform
internal curing. The size limitation for these particles to be effective is often a matter of
debate between researchers. Some researchers believe that LWA in sizes of 300 f.Lm (#50
sieve) or less have mostly surface water and therefore contribute very little to continued
internal curing.
The investigators 17 also observed that not all the water contained in the LWA is available
for internal curing. They concluded that only about 2/3 of the water present in the LWA
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
is readily available for transport to self-desiccating paste. Some water remains in the
LWA in the high RH range and only comes into play when the overall RH humidity in
concrete is significantly reduced.
The principal reason for making additions of lightweight aggregates to the high-
strength/high-performance concretes is to provide sufficient internal curing water to the
concrete so it is imperative that the aggregates used be capable of containing adequate
amounts of moisture. Fig. 4 shows the effects of varying aggregate moisture conditions
on autogenous shrinkage 18 . The lightweight aggregate was a pelletized expanded clay in
In the research noted above, when there was a sufficient amount of lightweight aggregate
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
in the concrete, and the aggregate had a high degree of saturation, the autogenous
shrinkage and its associated stresses were virtually eliminated from the concrete.
However, there was an influence of the aggregate size on the shrinkage reduction. Fig. 5
shows the effect of lightweight aggregate size fractions on the amount of autogenous
shrinkage 17 . The lightweight aggregate was a pelletized expanded clay. The binder
consisted of a blend of blast furnace slag cement (bfsc ), Portland cement (pc ), and silica
fume (sf) with a w/(bfsc+c+sf) = 0.35. All aggregates had the same degree of saturation.
While all the aggregate size fractions resulted in positive autogenous deformations, the
sand size (0- to 4-mm) gave the best performance.
Bentz and Jensen28 noted that the pore structure of aggregate particles may have a strong
effect on self-desiccation shrinkage and that aggregate particles containing water in
bigger pores may provide internal curing for the hydrating cement paste. In this case, the
formation of empty pores due to chemical shrinkage takes place first in the bigger
aggregate pores and does not involve the smaller pores in the cement paste. They also
noted that active use of this principle had been carried out with saturated lightweight
aggregate particles6•13 •19 and that it had been demonstrated by both computer simulation 10
and experiments 18 that self-desiccation shrinkage is best minimized by the use of smaller
size moisture preconditioned LW A as opposed to bigger ones. The fine saturated
lightweight aggregates result in a more uniform distribution of the needed curing water
throughout the microstructure. They suggest that by using this technique, self-
desiccation shrinkage could be completely eliminated.
The Dutch investigators 18 also concluded that using smaller fractions of saturated
aggregate particles resulted in a more homogeneous distribution of the water-containing
particles and shorter transport distances of the moisture from the aggregate to the drying
paste. They recommended that a partial substitution of normal weight aggregates with
LWAs be used to reduce autogenous shrinkage in HPC with low w/c ratio and that this
LWA be of small size and fully saturated to be most effective.
Work done in the United States23 •24 •25 used a crushed expanded shale aggregate to
promote the internal curing of paving concretes having a water-cement ratio of 0.4 to
0.45. In a series of tests performed in 1999-2000, a standard paving mixture had partial
normal weight aggregate replacements made by both lightweight fines and lightweight
coarse aggregate (9.5- to 2.4-mm). In separate mixtures, both the natural sand and the
normal weight coarse aggregate had L WA replacements of 59-, 119- and 178-kg/m3.
Table 2 shows the results of the tests on the freshly mixed concrete. The addition of
either of the LWAs had virtually no affect on the unhardened characteristics of the
concrete. Table 3 shows the results of tests on the hardened concrete. Improvements
were observed in concrete with the LWA additions for all the physical characteristics that
were measured. Of particular note are improvements in both the compressive and tensile
strength and also in the resistance to chloride ion penetration as measured by ASTM C
1202 (AASHTO T 277) "Standard Test Method for Electrical Indication of Concrete's
Ability to Resist Chloride-Ion Penetration". Experience with small additions of crushed
lightweight aggregate as a replacement for normal weight aggregates in concrete
typically shows some strength improvement because the vesicular surface of the LWA
allows penetration of the cement binder into the aggregate with a resulting improvement
in the transition zone 23 ' 24 • The improvements in the coulomb ratings indicate a denser
matrix that is attributed to the additional hydration products that formed because of the
availability of water from the LWA fines.
An additional test program was begun in 2001 to compliment the original Northeast
So lite work. This program examined the partial replacement of cement with fly ash and a
reduction in the water-binder ratio to produce a high-performance paving mixture. The
results 30 indicated improved performance with saturated lightweight fines additions.
Mitigation strategies for reducing or eliminating autogenous shrinkage cracking and self-
desiccation have also been proposed and discussed by the U.S. National Institute for
Standards and Technology (NIST) 28•29 • That work also suggested that self-desiccation
shrinkage is best minimized by the use of smaller size fractions of preconditioned LWA
as opposed to larger size fractions. The smaller size fractions result in a more uniform
distribution of the needed curing water throughout the microstructure.
In general, LWAs are not as stiff or strong as normal weight aggregates and thus may or
may not slightly reduce the concrete strength depending on a number of considerations.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Their contribution to strength depends on the type and quality of the LWA, the size
fraction used, the amount of aggregate used, and the type and quality of the binder in the
concrete. In general, crushed LWAs provide a better surface for binder interaction than
do LWAs with a sealed surface that often results from a pelletizing process. The
vesicular surface resulting from the crushing operation allows paste penetration and
provides more surface area for any reaction between the aggregate and paste to occur. It
is believed that the transition zone associated with a crushed aggregate has advantages
over a more smooth and sealed surface. The use of soft or friable lightweight fines may
result in some reductions of strength. Consequently, it is helpful to perform a mortar cube
strength test (ASTM C 109C/109M) of the lightweight fines being considered for use.
While the moisture contained in any size of saturated LWA will contribute to the
continued hydration of the binder, larger size aggregates are not as effective as small size
fractions because of the difference in spatial distribution. Excellent internal curing
results have been obtained by using saturated LWAs in the size 4. 75-mm with 2-4%
retained on the 425!lm sieve.
In the work conducted at NIST29 , it was found that at 28-days age, the mortars containing
the lightweight fines had slightly improved compressive strengths (up to 3.5 MPa) more
than the no-lightweight mortar. Similar improvements were found at coarse LWA
replacements up to 45% by volume of the coarse aggregate fraction 26 •27 •
PRACTICAL CONSIDERATIONS
Introducing LWA does not substantially change the mixing procedure or time of mixing
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
provided the L WA is adequately preconditioned. The amount of preconditioning
required at the time of hatching must be included in the job specifications and continually
verified during progress of the job. One economical way to do this is to have the
aggregates shipped SSD from the aggregate supplier, and then regaining any moisture
lost during shipping by sprinkling for 24-hours or until the SSD condition is regained.
Continued sprinkling may be necessary to maintain the needed moisture conditioning of
the aggregate during prolonged periods of concrete production.
CONCLUDING REMARKS
The amount of LWA used to achieve beneficial internal curing is a function of the type of
LWA used, its size and amount, the degree of moisture preconditioning the LWA
receives, the amount and type of binder(s) in the mixture, the water-binder ratio at
mixing, and the amount and duration of external moist curing provided to the concrete
element. In almost every study reviewed by the author, the use of adequate amounts of
near saturated LWA has completely eliminated self-desiccation and autogenous
shrinkage in the concrete. Most lightweight aggregates, if properly preconditioned so
that they contain near saturation levels of moisture, will contribute to the continual
internal hydration of the binder. Those aggregates with low absorptions will require
more LWA to be added to the mixture in order to get the total extra moisture into the
concrete. While any size L WA appears to be beneficial, most researchers have indicated
that the use of smaller size fractions is preferable because of a more uniform distribution
of the fines throughout the concrete and the improved proximity of the binder to these
particles. The need for internal curing is more critical when the binder contains finely
divided silica such as fly ash or silica fume. The addition of preconditioned LWA to the
concrete does not appear to have any adverse effects on the mechanical properties of the
concrete and, in most instances, small improvements have been noted. An optimization
between the need for internal moisture in the concrete to reduce autogenous shrinkage
REFERENCES
1. Holt, E.E., Early age autogenous shrinkage of concrete, VTT Publications 446,
Technical Research Centre of Finland, Espoo, 2001, 195 pp.
2. ACI 308-92 (Reapproved 1997), Standard Practice for Curing Concrete, ACJ
Manual of Concrete Practice, Part 2, American Concrete Institute, Farmington
Hills, Michigan, 2002.
5. Philleo, R., "Concrete Science and Reality," Materials Science of Concrete II, Eds:
J.P. Skalny and S. Mindess, American Ceramic Society, Westerville, Ohio, 1991,
pp. 1-8.
6. Weber, S., and Reinhardt, H.W., "A New Generation of High Performance
Concrete: Concrete with Autogenous Curing," Advanced Cement Based Materials,
Vol. 6, No.2, August 1997, pp. 59-68.
7. Powers, T.C., and Brownyard, T.L., "Studies ofthe Physical Properties of Hardened
Portland Cement Paste," Bulletin 22, Portland Cement Association, Skokie, Illinois,
1948,992 p.
8. Mather, B., and Hime, W.G., "Amount of Water Required for Complete Hydration
of Portland Cement," Concrete International, Vol. 24, No.6, June 2002, pp. 56-58.
9. Bonneau, 0., and Aitcin, P-C., "Importance of Water Curing to Control the Initial
Shrinkage Cracking and Durability of High Performance Concrete (HPC)
Structures," Proceedings, Third International Concrete on Concrete Under Severe
Conditions, CONSEC '01, Editors: N. Banthia, K. Sakai, and O.E. Gjorv,
University of British Columbia, Vancouver, Canada, 18-20 June 2001, Vol. 2,
pp. 1271-1278.
10. Bentz, D.P., and Snyder, K. A., "Protected Paste Volume in Concrete- Extension to
Internal Curing Using Saturated Lightweight Fine Aggregate," Cement and Concrete
Research, Vol. 29, 1999, pp. 1863-1867.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
13. Weber, S., and Reinhardt, H. W., "Manipulating the Water Content and Micro-
structure of High Performance Concrete Using Autogenous Curing," Modern
Concrete Materials: Binders, Additions and Admixtures, Editors: R.K. Dhir and
T.D. Dyer, Thomas Telford, London, 1999, pp. 567-577.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
14. Tasdemir M.A., Tasdemir, C., Grimm, R., and Konig, G., "Role of Aggregate
Fraction in the Fracture of Semi-Lightweight High Strength Concrete,"
Proceedings, 6th International Symposium on Utilization of High Strength/High
Performance Concrete, Ed: G. Konig, F. Dehn, and T. Faust, Leipzig, Germany,
Vol. 2, June 2002, pp. 1453-1466.
15. van Breugel, K., and de Vries, H., "Potential of Mixtures with Blended Aggregates
for Reducing Autogenous Deformation in Low Water/Cement Ratio Concrete,"
Proceedings, 2nd International Symposium on Structural Lightweight Aggregate
Concrete, Editors: S. Helland, I. Holand, and S. Smeplass, June 18-22, 2000,
Kristiansand, Norway, pp. 463-472.
16. van Breugel, K., Kaptijin, N., and de Bruijn, W. A., "Effect of Paste Composition
on Autogenous and Drying Shrinkage of HSC - Grade B65," Proceedings, 6th
International Symposium on Utilization of High Strength/High Performance
Concrete, Ed: G. Konig, F. Dehn, and T. Faust, Leipzig, Germany, June 17-19,
2002, Vol. 2, pp. 1477-1488.
17. Lura, P., and van Breugel, K., "The Influence of the Moisture Flow from the LWA
to the Paste on the Early-Age Deformation of LWAC," Proceedings, 6th
International Symposium on Utilization of High Strength/High Performance
Concrete, Ed: G. Konig, F. Dehn, and T. Faust, Leipzig, Germany, June 17-19,
2002, Vol. 2,pp. 1149-1160.
18. van Breugel, K., and Lura, P., "Effect oflnitial Moisture Content and Particle Size
Distribution of Lightweight Aggregates on Autogenous Deformation,"
Proceedings, 2nd International Symposium on Structural Lightweight Aggregate
Concrete, Editors: S. Helland, I. Holand, and S. Smeplass, June 18-22, 2000,
Kristiansand, Norway, pp. 453-462.
19. van Breugel, K., and de Vries, H., "Mixture Optimization of Low Water/Cement
Ratio, High-Strength Concretes in View of Reduction of Autogenous Shrinkage,"
Proceedings, International Symposium on High-Performance and Reactive Powder
21. Bentur, A., Igarashi, S., and Kovler, K., "Prevention of autogenous shrinkage in
high-strength concrete by internal curing using wet lightweight aggregates," Cement
and Concrete Research, Vol. 31, No. 11, November 2001, pp. 1587-1591.
22. Kohno, K., Okamoto, T., Isikawa, Y., Sibata, T. and Mori, H., "Effects of artificial
lightweight aggregate on autogenous shrinkage of concrete," Cement and Concrete
Research, Vol. 29, No.4, Aprill999, pp. 611-614.
23. Roberts, J. W., and McWhorter, Jr., J. F., "Internal Curing", Northeast Solite
Corporation, Saugerties, New York, 6 pp.
24. Test Report from Professional Service Industries (PSI), Sterling, Virginia to
Northeast So lite Corporation, Saugerties, New York, April 11, 2001.
26. Hoff, G. C., "The Use of Structural Lightweight Aggregates in Offshore Concrete
Platforms," Proceedings, International Symposium on Structural Lightweight
Aggregate Concrete, Sandefjord, Norway, June 20-24, 1995, pp. 349-362.
27. Walum, R., Weng, J.K., Hoff, G.C., and Nunez, R.A., "The Use of High-Strength
Modified Density Concrete in Offshore Structures," Proceedings, International
Conference on Concrete Under Severe Conditions, Sapporo, Japan, August 1-4,
1995, Vol. 2, pp. 1368-1377.
28. Bentz, D. P., and Jensen, 0. M., "Mitigation Strategies for Autogenous Shrinkage
Cracking," submitted to American Concrete Institute, 2001.
29. Bentz, D. P., and Geiker, M., "On the Mitigation of Early Age Cracking,"
International Seminar on Self-Desiccation and Its Importance in Concrete
Technology, Lund, Sweden, 15 June 2002,9 pp.
31. "The Use of Lightweight Fines for the Internal Curing of Concrete", Northeast
Solite Corporation, Richmond, Virginia, USA, August 20,2002,37 pp.
32. Bremner, T.W., and Holm, T.A. (1986), "Elastic compatibility and the behavior of
concrete, ACI Journal, Proceedings, Vol. 83, No.2, pp. 244-250.
33. Bremner, T.W., Holm, T.A., and deSouza, H. (1984), "Aggregate-matrix interaction
in concrete subjected to severe exposure," Proceedings, FIP-CPCI International
Symposium on Concrete Sea Structures in Arctic Regions, Calgary, Canada.
34. Bremner, T.W., Holm, T.A., and Stepanova, V.F. (1994), "Lightweight concrete- a
proven material for two millennia," Proceedings, Advances in Cement and
Concrete, Eds: S.L. Sarkar and M.W. Grutzeck, University of New Hampshire,
Durham, pp 37-41.
35. Holm, T.A., and Bremner, T.W. (2000), "State-of-the-Art Report on High-Strength,
High-Durability Structural Low-Density Concrete for Applications in Severe
Marine Environments," Report ERDC/SL-TR-00-3, U.S. Army Corps of Engineers
Engineer Research and Development Center, Vicksburg, Mississippi, 116 p.
36. Hammer, T.A. (1992), "High Strength LWA Concrete with Silica Fume- Effect of
Water Content in the LWA on the Mechanical Properties," Supplementary Papers,
Fourth CANMET/ACI International Conference on Fly Ash, Silica Fume, Slag and
Natural Pozzolans in Concrete, Istanbul, Turkey, pp. 313-323.
37. Hammer, T.A. (1993), "The Maturation of Mechanical Properties of High Strength
Concrete Exposed to Different Moisture Conditions," Proceedings, Utilization of
High Strength Concrete, Ed: I. Roland and E. Sellevold, Vol. 2, Lillehammer,
Norway, June 20-23, 1993, pp. 1084-1096.
38. Hammer, T.A., and Smeplass, S. (1995), "The Influence of Lightweight Aggregate
Properties on the Material Properties of the Concrete," Proceedings, International
Symposium on Structural Lightweight Aggregate Concrete, Ed: I. Roland, T.A.
Hammer, and F. Fluge, Sandefjord, Norway, June 20-24, 1995, pp. 517-530.
39. Hammer, T.A., Bjontegaard, 0., and Sellevold, E.J. (1998), "Cracking Tendency of
High Strength Lightweight Aggregate Concrete at Early Ages," Proceedings,
Fourth CANMET/ACIIJCI International Conference on Recent Advances in
Concrete Technology, Ed: V.M. Malhotra, Tokushima, Japan, pp 53-65. (Available
as SP-179, American Concrete Institute, Farmington Hills, Michigan).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
41. Geiker, M., Bentz, D.P., and Jensen, O.M. (2004), "Mitigating Autogenous
Shrinkage by Internal Curing," Proceedings, High-Performance Structural
Lightweight Concrete, SP-218, Editors: J.P. Ries and T.A. Holm, American
Concrete Institute, Farmington Hills, MI, 2004, pp. 143-154.
42. Bentz, D.P., Koenders, E.A.B., Monnig, S., Reinhardt, H-W., Van Bruegel, K., and
Ye, G. (2005), "Materials Science-Based Models in Support of Internal curing," to
appear in RILEM State-of-the-Art Report, Internal Curing of Concrete, 2005.
43. Roberts, J.W. (2004), "Internal Curing in Pavements, Bridge Decks and parking
Structures Using Absorptive Aggregates to Provide Water to Hydrate Cement not
Hydrated by Mixing Water," 83rct Annual Meeting of the Transportation Research
Board, Washington, DC., January 2004.
44. Lura, P. (2003), Autogenous Deformation and Internal Curing of Concrete, PhD
Thesis, Delft University, Delft, The Netherlands, 2003.
45. Lura, P., Bentz, D.P., Lange, D.A., Kovler, K., and Bentur, A. (2004), "Pumice
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Aggregates for Internal Water Curing," PRO 36: Proceedings, International
Symposium on Concrete Science and Engineering, RlLEM Publications S.A.R.L.,
2004, pp. 137-151.
46. Bentz, D.P., Koenders, E.A.B., Monnig, S., Reinhardt, H-W., Van Breugel, K., and
Ye, G. (2005), "Materials Science-Based Models in Support oflnternal Curing,' to
appear in RILEM State-of-the Art Report, Internal Curing of Concrete, 2005.
47. Villarreal, V.H. (2005), "Building Better Pavement Through Internal Hydration, A
Work in Progress," presented at the Raymundo Rivera International Symposium on
Durability of Concrete, Editor: Alejandro Duran-Herrera, Universidad Autonoma de
Nuevo Leon, Monterrey, Nuevo Leon, Mexico, May 12-13, 2005.
48. Hoff, G.C. (2003), "Internal Curing of Concrete Using Lightweight Aggregates,"
Proceedings, Theodore Bremner Symposium on High-Performance Lightweight
Concrete, CANMET/ACI International Conference on Durability of Concrete,
Thessaloniki, Greece, June 1-7,2003, pp. 185-203.
49. Hammer, T.A., Bjontegaard, 0., and Sellevold, E.J. (2004), "Internal Curing- Role
of Absorbed Water in Aggregates," Proceedings, High-Performance Structural
Lightweight Concrete, Eds: J.P. Ries and T.A. Holm, SP-218, American Concrete
Institute, Farmington Hills, Michigan, 2004, pp. 131-142.
""" "'---. -~
Fig. 1- Typical Autogenous Deformation vs Time Relationship for Concretes With and
Without Lightweight Aggregates.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--Normal Weight Concrete
• • • · AD Condition
- - SSD Condition
100 --.------,------,---"T'""'"----,
0 ~r-~_,-------r------r-----_,
-100 +-------t"""'-o;;;;;;::-----r-----..-..-c-r-----_,
-200 -1------+----+-----=----....±----1
-300 ...L..,__ ____,__ __ _ _ j_ _ _...L,__ ___,
l
- - N onnal Weight Concrete
• • • · AD Condition
=
1:1.
3.50
- - SSD Condition I
::;;
~
"'
Q,j
!1
C'll
OJ)
.e
3.00
2.50
2.00
1.50
1.00 ~
t'1 -.
.,.....,: -
--- ..,
r-.J
~- ........ 1'- - ..
.e 0.50
J~ _r---
....."'= 0.00
"'
Q,j -0.50
=
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0 50 100 150 200
Time, hours
0 50 100 150
Time, hours
e ----
.s~
100
k··
...r-- • • • ·4-8 mm
- -8-16mm
=
~
'-'
......
; ·!
80
60
fl ..._ r---
~ .!:I
~Clll "' 40
.s 20
<=
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
0
0 100 200 300 400
Time, hours
by T.A. Hammer
Synopsis: Early age cracking may be what is commonly known as "plastic shrinkage
cracking", which normally is cracking of horizontal surface before and during setting
(initial phase), and "thermal cracking" which normally is in the period of cooling
following the period of temperature rise due to heat of hydration (thermo phase).
In the initial phase, any mix water absorption by the LWA (i.e. like in concretes with
relatively dry LWA) may contribute to increased settlement, capillary tension of pore
water and shrinkage, and thus, an increased risk of cracking in typically the first
hour after finishing. However, in the early hardening age the absorbed water may
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
constitute a reservoir which contributes to swelling of the paste which counteract any
plastic shrinkage and/or any contraction due to cooling, and thus, to reduced risk of
cracking.
In the thermo phase the temperature rise in LWAC is potentially higher due to the
lower heat capacity of the LWA, which may result in larger contraction during cooling,
which again may generate higher stress, and finally to higher risk of cracking. On the
other hand the autogenous shrinkage is significantly reduced or even eliminated,
and theE-modulus is lower, which both contributes to lower stress, and thus, to
lower risk of cracking. The net result is often reduced risk of early age cracking.
INTRODUCTION
The whole field of stress predictions and crack limitation has been and still is quite active
in many countries, and many authors have reported on the topic. The present paper is not
a full literature review, but a modest attempt to identify the important parameters
controlling early age cracking and where examples from some authors are used to
demonstrate the influence of light weight aggregates.
Early age cracking is a problem associated with high strength concrete/high performance
concrete (HSC/HPC), mainly. Hence, high strength lightweight aggregate concrete
(LWAC) is given special attention to in the paper.
Early age of concrete may be divided in the initial phase, which is before and during
setting, and in the thermo phase which is during the heat generation and the subsequent
cooling.
In the initial phase the typical cracking problem is related to horizontal surfaces where
plastic settlement and plastic shrinkage (caused by insufficient protection against
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
evaporation) is the typical problem. For low water/binder concretes (HSC/HPC) the
problem is worsened and cracking may occur in spite of "proper" curing (curing
membrane, plastic sheets, water spray, etc ) 1• The author has not found any information
from field experience that gives reason to state that LWAC behaves significantly
different from normal weight concrete (NWC) in the initial phase.
thermal contraction due to the cooling following the hydration generated temperature
rise,
autogenous shrinkage (self-desiccation shrinkage) resulting from the chemical
shrinkage generated by the hydration
drying shrinkage
E-modulus
creep/relaxation,
The author has not found field experience that reveals considerable differences between
LWAC and NWC with respect to thermo cracking. A direct comparison is of course
difficult because the structures (and mortar compositions of the concretes) are different.
However, to our knowledge, LWAC have not been associated with thermo cracking
problems.
In the following the influence of LWA on the material parameters responsible for both
types of cracking are discussed. Furthermore, the resulting effect (i.e. the crack risk) is
demonstrated by results from testing of the restraint stress development under realistic
temperature conditions.
The driving forces to cracking in the initial phase may be divided in volume changes and
external mechanical impact. The latter may be unwanted movement of the concrete after
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
placing/finishing caused by e.g. formwork movement (settlement, vibration, etc.). Also, it
may be a combination of sloping structure and concrete with high viscosity and low yield
strength (e.g. some high strength concretes). Lighter concrete may reduce these external
loads, but this is not confirmed by tests.
In general, the cracking risk increases with decreasing w/b 1• This may disfavour LWAC if
compared with NWAC having the same compressive strength, because the LWAC often
has lower w/b.
Experience shows that cracking ofHSC may appear at two basically different ages 1:
From both an engineering and testing point of view it is opportune to consider the volume
changes taking place as settlement and shrinkage, i.e. as vertical and horizontal
components, respectively. As the concrete changes from a fluid to a solid in the initial
phase, the battle between gravity and stiffness evolution will influence the ratio between
shrinkage rate and settlement rate basically from 0 to 1, i.e., in the beginning, all volume
change is seen as settlement.
Autogenous deformation
Air evacuation
Water loss due to bleeding
Water loss driven by evaporation (in addition to any bleed water)
Contraction due to capillary tension of pore water
Thermal contraction (i.e. cooling due to colder surrounding air)
LW A, when not sufficiently pre-wetted, absorbs some mixing water in the time before
setting. In the case referred to later, the estimated mix water absorption of the LWAC (with
"Leca 700" 4 - 12 mm with a particle density of approx. 1300 kg/m 3) was about 40 litres per
m 3 of concrete, which corresponds to a volumetric shrinkage of 40 000 J..lStrain! Although
most of it occurs in the time before the start of setting and testing, it will certainly constitute
a potential additional shrinkage. This is demonstrated in Fig 1. It shows autogenous
settlement and shrinkage of LWAC and an equivalent NWC (w/b = 0.40) under sealed
conditions (i.e. no external drying). The tests were performed on 280 x 100 x 100 mm
beams in a steel mould with double plastic sheets between the mould and the concrete to
minimise friction 3 . As can be seen the autogenous settlement of the LWAC is
approximately 4000 11strain higher in the first hour, corresponding to 4 litres of water
absorption in this time period. Furthermore, the shrinkage of the L WAC starts earlier and
reaches a maximum value that is close to 1000 J..lStrain, while the NWAC show no
significant autogenous shrinkage in the period. However, if the LW A has a rather coarse
pore system a major part of the absorption may take place during mixing, transportation and
placing. Thus, the additional shrinkage may be negligible, even when the LWA is not
sufficiently pre-wetted before mixing.
Fig. 2 show results from testing of similar LWAC, plus one with pre-wetted LW A (24 hours
in water) and all exposed to moderate drying, i.e. in air with 50% RH (no wind). Again, it can
be seen that the settlement is much higher in the first hour, here about 4000 J..lStrain, probably
due to mix water absorption. The LWAC with pre-wetted LWA also show larger settlement
than that of the NWC. It has not been found reliable explanation for this. The shrinkage rate of
the one with dry LWA is slightly higher in the beginning, see Fig 2. The one with pre-wetted
LW A follows the NWC until initial set approximately. Thus, the internal water reservoir does
not seem to have any significant influence on the plastic shrinkage at moderate drying in
plastic age. Another results may however be seen at more severe drying i.e. if the evaporation
rate is very high compared to the maximum water flow to the surface (controlled by
permeability). Then the internal water may help by reducing the transport distance of water.
This is not yet confirmed by tests.
Note that both L WAC's expand in the time beyond approx. 8 hours, which is close to the
time of setting (seen from e.g. temperature measurements), while the NWC continues
shrinking. This demonstrates that even the internal water reservoir developed from mix
water absorption, is sufficient to counteract early age drying shrinkage (see also next
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Cracking in the thermo phase has traditionally been controlled by the use of a maximum
allowed temperature difference, which simply is the tensile strain capacity (which often is
not measured for the actual concrete and temperature history) divided by an average
thermal dilation coefficient. This principle may be adequate in many cases. However, the
use of "modem" concretes (HPC/HSC) which have other properties, in more advanced
structures, has shown that the traditional curing technology may not give the same
accuracy in predicting the crack risk: Sometimes, the crack risk has been overestimated
causing excess costs due to e.g. extra cooling, sometimes underestimated causing
cracking that needs expensive repair.
Potentially, the temperature rise in LWAC is higher due to the lower heat capacity of the
LW A. Thus, the contraction during cooling is probably larger, which contributes to
higher risk of cracking. On the other hand, the insulating capacity is higher which means
a slower cooling, which to some extent may compensate for the effect of larger cooling.
However, a longer period of high temperatures may give steeper temperature gradients in
a section and thus higher risk of surface cracking.
One may assume that the coefficient is dominated by the paste coefficient, since the
stiffness of the paste is higher than that of the LW A. The coefficient of a paste is
typically twice the coefficient of equivalent concretes, but depends strongly on moisture
content. On the other hand the water in the L W A grains may contribute to a higher
degree of water saturation in the paste and thus contribute to a lower coefficient. The net
result is therefore not given.
Tests on NWC demonstrate clearly that the autogenous shrinkage may give a substantial
contribution to increase the risk of early age cracking. Tests have shown that the L WAC
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
range of three times higher than the necessary amount needed to fill the contraction pore
volume. Thus, the results are as expected.
One investigation includes results from testing of the autogenous shrinkage of concretes
with a part replacement of the coarse normal aggregate by coarse L W A ("Liapor I 0" -
expanded clay with bulk specific density of approx. 1700 kg/m 3) with initial moisture
content of approximately 13 % by weight, see Fig. 5 8 • As can be seen, there is a distinct
effect of increased replacement, but 25 % is not enough to compensate for the autogenous
shrinkage. In this concrete, the water content in the L W A corresponds to about 16 litre of
water per m 3 of concrete, which is in the range of 50 to 100 % more than the amount
needed to fill the contraction pores at this maturity. Consequently, the availability of the
water is not sufficient, which may be due to the relatively dense structure Liapor 10
aggregate, and possibly the relatively large distance between the L WA grains.
Fig. 7 shows the combined effect of thermal strains and autogenous deformation of a
LWAC and an equivalent NWC (w/c = 0.38) both exposed to two temperature histories
with maximum temperatures of about 30 and 36 °C 8 . As can be seen, the total contraction
caused by the cooling (between about 12 and 24 and 12 and 72 hours for the two
histories, respectively) and the autogenous deformation, is much less for the LWAC. In
fact it can be shown that the LWAC has autogenous expansion: The maximum
contraction of the L WAC is about 60 microstrains, see Fig. 7, which is much less than the
thermal strains alone: (36- 24) °C X w-s I °C = 120 microstrain (approx.). Furthermore,
the LWAC in the period after the cooling does not show any autogenous shrinkage, but
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
rather a swelling. Note also the additional shrinkage in the initial phase (before 6 hours of
age), as discussed earlier.
In general the autogenous deformation is getting very unclear when the concrete is given
realistic temperature histories: Bj0ntegaard 12 showed that the temperature dependency is
quite unsystematic, and even that autogenous deformation mat tum from contraction to
expansion by time.
Note that the present concretes have rather low w/b, and thus, a potentially considerable
autogenous shrinkage. Normally, the autogenous shrinkage decreases with increasing
w/b. Consequently, the beneficial effect of the internal water reservoir of the LWA will
decrease with increasing wlb.
The drying shrinkage is normally not a major driving force for early age thermal
cracking. However, in any case the internal water reservoir may eliminate, or at least
significantly reduce the drying shrinkage in the first days 7 and thus contribute to an even
lower driving force compared with NWC.
Hammer and Kanstad 13 performed a comparison of LWAC and NWC with the same
recipe except for the aggregate 4 - 16 mm. The L WAC was made with initially dry "Leca
800" (expanded clay, particle density of 1400 kg/m 3) and the effective w/c was 0.38 in
both concretes (no silica fume). Both concretes were cured isothermally at 20 °C and at
semi-adiabatic temperature history with maximum temperatures of 40 and 45 °C. The
results show that the E-modulus of the LWAC is about 25 % lower and that the
development is quite similar to that of the NWC, see Fig. 8.
Creep I Relaxation - Creep and relaxation are directly related and one may assume
that differences in creep behavior between LWAC and NWC also apply for relaxation 12 •
Creep data on L WAC is found in the literature on relatively mature concrete. In those
cases the specific creep of LWAC is sometimes higher and sometimes lower than NWC,
but seldom considerably different 13 • The inconsistency may be explained by the
following factors:
LWAC and NWC are often compared based on compressive strength, i.e. the LWAC has
a higher paste strength. In those cases LWAC may experience lower creep/relaxation
which has a negative influence on the stress development and thus the crack risk. On the
other hand, L WAC has often higher paste content and higher moisture content, which
may give some higher relaxation/creep. Furthermore, when subjected to external drying,
the internal water reservoir of L WAC may contribute to somewhat lower drying
creep/relaxation (particularly at relatively early ages) compared to NWC, because it
prevents the drying of the paste.
Little data is found on early age creep on NWC and no data on LWAC. Determination of
creep or relaxation in early age concrete is difficult because of the rapid change of
properties during testing. The E-modulus development is quite rapid and relatively high
already after a day or two. Consequently, the general difference between LWAC and
NWC may be used in the early age too.
From the above, it follows that it is not obvious how creep/relaxation of LWAC
influences the risk of cracking compared to NWC. However, in most cases it is not
expected that creep/relaxation will play any significantly different role compared to
NWC.
The concretes referred to under the section "E-modulus" were tested on uniaxial tensile
strength, also. The results show that the relative tensile strength is lower for the L WAC
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The ductility of non-reinforced LWAC is much less, decreasing with increasing amount
of LWA and strength-density ratio 13 • This means that cracks more easily lead to global
failure. Thus, the lower ductility may contribute to somewhat higher crack risk.
The risk of early age cracking depends on many properties as discussed above,
complicating the determination, and there is yet no simple way of determining the crack
risk of one concrete compared to another. In order to measure the stresses being built up
due to the thermal and autogenous strains, a number of special test rigs, "Temperature
Stress Testing Machine" (TSTM), have been constructed during the last decade
throughout the world. Even these tests do not always give a general ranking of concretes
since the influence of temperature history, geometry, degree of restraint, etc., varies
between concretes.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The most sophisticated TSTMs measure the stress generation under full or partial
restraint (via a feedback system) from about the time of setting. The temperature of the
concrete is controlled to produce either isothermal conditions (to measure the
contribution of autogenous shrinkage alone) or any realistic temperature history to
simulate practical conditions. Thus, the TSTM may be used directly to characterize the
stress generation in a given concrete composition in a given structure, or to give the
answer to check the accuracy of the calculations of stress development based on the
materials properties (numerical simulation).
The test results referred to in the following are based on different LW A types and initial
water contents of the LWA. A common feature is that all concretes have pastes with
rather low w/b, i.e. with a relatively high autogenous shrinkage potential. Therefore, the
great influence that the LW A has on autogenous shrinkage dominates the results.
One HSNWC and one HSLWAC were tested. Both concretes have the same paste
composition: CEM type I cement and w/c = 0.38. The natural aggregate was typical
Norwegian glacifuvial gneiss/granite sand and gravel. The LW A used was "Leca 800"
(produced in Norway by a.s Norsk Leca), with the bulk specific gravity (i.e. "particle
density") of 1400 kg/m 3 . The LWA was oven dry at mixing. The mix water absorption of
the LWA corresponded to 4 7 litres/m3 . Both concretes were mixed and tested two times,
with two sets of temperature histories, see Fig. I 0. The two temperature histories were
obtained by using steel moulds (1 mm wall thickness) with and without insulation (20 mm
The stresses are recorded only from after setting (here at the age of five hours). As can be
seen in Fig. 11, the thermal expansion (until approximately 12 hours) gives a small
compression. Surprisingly, it is 40 - 50 % higher in the L WAC compared to the HSC, in
spite of a nearly equal temperature rise. The reason for the unexpected difference may be
a maturity difference: The temperature of the HSL WAC specimens at setting was 3 - 5 °C
higher than of the HSNWC specimen, see Fig. 5, causing higher maturity of the L WAC
in the heating period, and thus a more rapid E-modulus development.
However, the most interesting and important observation is in the critical cooling period
(i.e. beyond the age of about 12 hours). As can be seen the HSNWC build up very high
tensile stresses in this period. The one with the highest temperature failed just below
3 MPa at the age of nearly 60 hours. The failure stress is fairly close to the measured
uniaxial tensile strength of 3.1 MPa at this time and nearly the same temperature history
as here.
The LWACs showed quite different behavior, and the one with the lowest temperature
did not experience tensile stresses at all. The development in the period of almost
constant temperature (after 2 - 3 days) shows a small decrease of the tensile stress of the
"insulated" L WAC, correspondingly a small increase of the compressive stress of the
other LWAC. This reflects the small expansion of about 10 micros train seen in Fig. 7.
Assuming an average E-modulus of 20 GPa, no relaxation and 100 % restraint, a strain
change of 10 microstrain would give a stress change of 0.2 MPa. The measured stress
change in the period is of the same order, but somewhat lower, probably due to
relaxation.
The results demonstrate that L WAC is more robust against thermal cracking than NWC
with similar paste (mortar) compositions. The main reason is the lack of autogenous
shrinkage. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
General - The mixes were based on the following: Cement content of 475 kg/m 3, a
blend of 50% blast furnace slag cement (CEM IIV B52.5) and 50% portland cement (CEM
1152.5 R). The W/C was 0.37. An amount 50 kg/ m3 silica slurry, 50% water and 50% silica
fume, was added in order to enhance the final strength. In the reference mixture a siliceous
aggregate was used, consisting of 45% sand (1-4 mm) and 55% coarse aggregate (4-16 mm).
Two types of plasticizers were used, i.e. lignosulphonate and naphthalene sulphonate. The air
content was 2.5%. The target cube compressive strength was 85 MPa at 28 days.
Figs. 12 and 13 show some results. In the isothermal case (20 °C), the stresses in the concrete
with 25 % Liapor are substantially lower than in the concrete with dense aggregate, only.
This is mainly caused by the lower autogenous shrinkage shown in Fig. 7. In the concrete
without Liapor cracking occurred after 210 hours at a tensile stress of3.5 MPa. The mixture
with 25% Liapor did not crack even after 336 hours after casting. At that time the tensile
stress was 2.2 MPa.
Full Replacement of Dense Aggregate by LWA with Different Initial Moisture Content
- The lightweight aggregate considered was Liapor 10 and Lytag (expanded fly ash with
bulk specific gravity (i.e. "particle density") of approx. 1400 kg/m 3). The Liapor aggregate
was used with different degrees of saturation. The target values of the degree of saturation
were 100 % and 60 %. The exact degrees of saturation of the aggregate were determined on
samples taken just prior to mixing. In addition, one mixture with Lytag has been considered
with a target degree of saturation of 100 %.
In the isothermal case none of the mixtures experienced tensile stresses due to the lack of
autogenous shrinkage, see Fig. 14. The Lytag mix showed increased compression stress
practically the whole week, confirming the swelling observed in the autogenous shrinkage
test, see Fig. 6.
In the semi-adiabatic case, however, see Fig. 15, the concretes experienced high tensile
stresses and the one with Liapor failed at the age of 78 hours at a tensile stress of about
3.1 MPa. The stresses generated in the Lytag mixture were smaller and it did not fail, which
mirror the swelling seen from the autogenous shrinkage tests. The stress development seems
to flatten out after about 6 days.
CONCLUSIONS
The temperature rise in LWAC due to heat of hydration is potentially higher due to the
lower heat capacity of the L WA. Thus, the contraction during cooling may be larger,
which contributes to higher stress, and thus, risk of cracking. On the other hand, the
autogenous shrinkage is significantly reduced or even eliminated and the E-modulus is
lower, which both contributes to lower stress, and thus, lower risk of cracking. The net
result is often reduced risk of early age cracking.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1. Kempen, R.: High performance concrete: Field Observations of Cracking Tendency
at Early Ages. Proceedings of the International Rilem Symposium: Thermal
Cracking in Concrete at Early Ages, Munich, Rilem Proceedings 25, October 1995.
2. Hammer, T.A. and Fossa, K.T. and Bj0ntegaard, 0.: Cracking Tendency of HSC:
Tensile Strength and Self generated Stress in the Period of Setting and Early
Hardening. George C. Hoff symposium on High-Performance Concrete and Concrete
for Marine Environment, Las Vegas, May 2004, pp. 139- 150.
3. Hammer, T.A.: The Relationship between Settlement and Plastic Shrinkage of High
Strength Concrete, International RILEM Conference on Early Age Cracking of
Cementitious Systems, Haifa, Israel, March 12-14, 2001, pp. 89- 102.
6. Hammer, T.A.: High Strength Concrete with Silica Fume- Effect of Water Content
in the LWA on the Mechanical Properties; Proceedings of the Fourth CANMET/ACI
Conference on Fly Ash, Silica Fume, Slag and Natural Pozzolans in Concrete
(Supplementary Papers), Istanbul, 1992, pp. 313- 30.
8. van Breugel, K.: Mixture Optimization of Low water/Cement ratio High Strength
Concrete in View of Reduction of Autogenous Shrinkage. Proceedings of the
9. Hammer, T.A., Bj0ntegaard, 0., and Sellevold, E.J.: Cracking Tendency of High
Strength Lightweight Aggregate Concrete at Early Ages. Proceedings of the Fourth
CANMET/ACI Symposium on Advances in Concrete Technology, Tokushima,
Japan 1998, pp. 53-64.
10. Weber, S., and Reinhardt, H.W.: "Various Curing Methods Applied to High-Strength
Concrete With Natural and Blended Aggregates"; Proceedings of the Fourth
International Symposium on the Utilization of High Performance Concrete, Paris,
1996, pp. 1295-1304.
11. Bj0ntegard, 0.: Thermal Dilation and Autogenous Deformation as Driving Forces to
Self-Induced Streses in High Performance Concrete. Dr.ing theses 1999:21, Division
of Structural Engineering, The Norwegian University of Science and Technology,
Trondheim 1999.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
12. Hammer, T.A., and Kanstad, T.: Lightcon, Report 2.16 Cracking Tendency of
LWAC at Early Ages. SINTEF-report STF22 A98782. Trondheim, Norway 1998 (in
Norwegian).
: Shrinkage LWAC
, Shrinkage NWAC
10
Tlme after water addition (hours)
Fig. 1- Autogenous settlement and shrinkage of equivalent concretes with initially dry
LWA and normal aggregate (NWA), respectively.
654 Hammer
9000
8000
~ 7000
i 6000
5000
I 4000
I"' 3000
2000
1000
10
Time after waler addition (hours)
Fig. 2- Plastic settlement and shrinkage of equivalent concretes with initially dry
LWA, pre-wetted LWA and normal aggregate (NWA), respectively.
I I I I I I I I
e
i -1200
i -1600
I
- __ ..J __ - -
I I I I I I
_ , L _ - - __I _ - - - - .L - - - - - L - - - - .J - - - - _ l . _ - - - -
~
I I I I I I I
I I I I I
I I I I I
I I I I I
I I I I I
-2000
3 4 5 6 7 10
nme after water addition (hours)
Fig. 3- Pore water pressure at 5 mm depth of equivalent concretes with initially dry
LWA, pre-wetted LWA and normal aggregate (NWA), respectively.
18
14 -
.,._Staltto (pre-saturated) I
- 0 ·L700 (initially dry)
l
---- --
E -
.,
12
10
- - L700 (pre-saturate<l)
_y;
~
~.
~
.
8 --CJ"-" •
~ 6
50 60 70 80 90 100
RH(%)
Tlme[hOUIW)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
T=20'C T=200C
_/
Y/. :emperature iSotnermaJ
~=~9.3%satl
~po<100%sa1J
'NCr0.37
1
0.00
T=40'C ~~=:~1.2%satl
r-
lf
nme[hours)
..
,
'
I' 'I I I I
'
200 ---~-----~------1------~-----~-----
1 I I
~c: 100
·e
Cii - NDC, 29 degC
0 -LWAC, 30degC
/+----+-NDC, 36 degC
-LWAC, 36 degC
-100
0 24 48 72 96 120 144
---
0.8 -- g---~~--:f- ----- r------ f- ----- f- ----- '
5 '
'
'
'
'
'
-#~---I~------~------~------~
..,~
_____ L __ _
0.6 --- _0 ._ I•
I I I I I I
.
j •1
:
I I I
: '
r-'----.1..__----,
.~ 0.4 --- --.-~----- -~-- ----:------:----- • e •LWAC (20oC) -
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
' ''
0.8 - - - - - 1'- - ----1-------1-------1-------
' ' '
' '
·§
..
'
j 0.4
I I
'
'
'
I'
------:------:------:------:---- • •
I
'
•LWAC (20oC)
...~ ' ' ' ..._NDC(20oC)
ll. '
'
'
'
'
'
0.2 ------~------~------~------r---- •G •LWAC(40oC)
: : : ~NOC (40oC)
45,-,.-,----,-----,----.----~---.
I
I I
I I I --NDC I
~ 40 ~-----,----~-- -~-----
······LWAC
~
1 I I I
: : -NDC
I!! 35 -----,-----,-- ·•• LWAC
~
8. 30
E
~
25
20+---~-----r----~---+----,_--~
0 24 48 72 96 120 144
Time after water addition (hours)
Fig. 11- Stress development due to thermal and autogenous strain of the LWAC
and NWAC concretes tested in the TSTM at SINTEFB.
I I
8 H,~I?D'C
....,.o.a7 I
2.11
.-----
O%Liapor
--
2
1.5
/_ ........
/ ~l.lllpor f - -
/ ./
~
0.5
~ L../
-v
.0.5
0 48 72 ge 120 144
Tlmll (hotn]
Fig. 12- Stress development due to autogenous strain of concretes with 25%
replacement of the coarse NWA with coarse LWA (w/c = 0.37) 7•
3
,.__r
l-a.37
J
2
~I
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
·1 \A
\.1 :!BIJopcr
0 24 48 72 ge 1.20 144
1111111 (hou ..]
Fig. 13 -Stress development due to thermal and autogenous strain of concretes with
25% replacement of the coarse NWA with coarse LWA (w/c = 0.37)7.
0.00
-0.10
- [\ ,jY ~v~ ~Liapor 100% sat I
T=40°C
Jremperature isothermal
wcr 0.37
rt 1
II
-0.20
I' "'\J\,ill A ~~lapor69.3% sat
T=20"C
"''-' 111
1
-0.30
If . I
-0.40
)-- -~~)\
-0.50
·0.60
_...... ...... ~~ytag 91.2% sat 1 v v'V\J\)('- ha l ......
T=20"C
'-"""' '"-.N I.JII
-0.70
-0.80
-0.90
0
/'\...-
Liapor 97.7% sat
T=20"C
24
l -- 48
-- 72 96 120
Time [hours)
144
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
replacement of the coarse NWA with coarse LWA (w/c = 0.37). Effect of LWA type 7•
3.00
2.00
"-./ ~~
ILiapor 94.9% sat I 1-
v -
1.00 Temperature semi-adiabatic
v Iw/c=0.37
0.00
//
""
Y1Lytag 93.6% sat I
-1.00
-2.00
\:; -r
-3.00
24 48 72 96 120 144
Time [hours]
Fig. 15- Stress development due to thermal and autogenous strain of concretes with
too% replacement of the coarse NWA with coarse LWA (w/c = 0.37).
Effect of LWA type7•
BACKGROUND REVIEW
HPLC having compressive strength in the range 41 to 100 MPa has been
developed by several researchers using various mixture proportions [10, 13, 15, 19-26].
Extensive studies on HPLC durability and shrinkage have also been carried out [13, 17,
18, 27, 28]. However, creep ofHPLC and particularly the influence of the internal curing
on creep ofHPLC remain unknown.
There are two opposite arguments that can be examined in assessing creep and
shrinkage ofHPLC. The first argument is that HPLC possesses a low W/B (water-to-
binder ratio), so the long-term deformations such as creep and shrinkage should be
greatly reduced [ 1]. On the other hand, the use of a comparatively low stiffness
aggregate will reduce the aggregate restraining effect on the creep and shrinkage of the
cementitious matrix, so the overall long-term deformations may increase [29]. These two
competing effects make the estimation of creep and shrinkage in HPLC difficult.
In addition, the internal curing provided by pre-wetted lightweight aggregate in
low W/B mixtures have proved to improve mechanical properties [30-34] and to reduce
autogenous shrinkage. Nevertheless, to the authors' knowledge, no published studies
have reported on the influence ofthe internal curing on overall shrinkage and creep in
HPLC.
Normal weight HPC presents specific creep in the range of25 to 40 !lEIMPa [35,
36]. For comparison, SLC has shown specific creep as high as 270 11EIMPa for certain
lightweight aggregates [37]. Some have reported specific creep ofHPLC values as large
as twice that of normal weight concrete of the same strength [38] while others have found
specific creep values of several HPLC mixtures loaded at 24 hours to be about 75 11EIMPa
(39]. The latter results are greater than the values reported for normal weight HPC (36]
under similar conditions.
When investigating shrinkage of HPLC, some authors found that HPLC had a
lower shrinkage rate, but a higher ultimate shrinkage compared with HPC [28, 38]. The
lower shrinkage rate was thought to be related to the presence of water in the aggregate
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
which delays drying. Holm and Bremner [10] also concluded that high strength
lightweight concrete had reduced shrinkage at early ages, but found that one-year
shrinkage was approximately 14% higher than its HPC counterpart.
Lopez et al. [15, 19] compared total time dependent deformations ofHPLC
having 69 MPa compressive strength with those measured on a normal weight HPC of
similar strength. They concluded that after 900 days under load and drying conditions,
creep plus shrinkage ofHPLC was lower than that ofHPC. The rate of deformation was
also consistently lower for the HPLC mixtures through all the testing period.
RESEARCH SIGNIFICANCE
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
program allows for the isolation of the aggregate stiffness effect (HPLC-A vs. HPC-A)
and the aggregate internal curing effect (HPLC-A vs. HPLC-B).
In order to increase the statistical significance of the experimental results, all
three mixtures were cast in the laboratory on three occasions. Two batches of each
mixture were cast on each occasion, and specimens for compressive strength, modulus of
elasticity, creep, and shrinkage were obtained.
The following tests were performed:
(1) Relative density and absorption of the coarse aggregates and the fine
aggregate were measured according to ASTM Cl27, C128, respectively.
(2) Compressive strength was determined by testing l 00 x 200 mm cylinders
according to ASTM C39.
(3) The chord modulus of elasticity was found using 150 x 300 mm cylinders
loaded in compression according to ASTM C469.
(4) Creep and drying shrinkage were determined by testing 100 mm x 380 mm
cylinders according the ASTM C512 guidelines. However, there were three differences
with respect to the ASTM procedure. The first difference was that the diameter of the
cylinders was smaller than that recommended in ASTM because the bearing capacity of
the creep frames used was not enough for applying the required stress to 150 mm x 300
mm cylinders. The other two deviations were the age of loading (24 hours instead of 2
days or greater) and the use of an accelerated curing regime. The later changes were
adopted in order to match the actual conditions of typical precast prestressed bridge
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 665
girders and of specific, companion HPLC pretensioned girders which were stressed at
very early ages. The accelerated curing used insulated boxes for the first 24 hours in
order to maintain the heat generated during the hydration in order to match the
temperature within an AASHTO Type IV girder as demonstrated in previous research
[ 14]. After the initial curing, some of the specimens were placed in the environmentally
controlled room (50% relative humidity and 23°C) where the creep and shrinkage testing
started. The other specimens were placed in the fog room until the day of testing.
From previous research [19], it was concluded that for accelerated cured
specimens approximately 80% of the creep and 90% of the shrinkage developed after 620
days occurred in the first 120 days. For that reason, creep and shrinkage were measured
for a period of only 120 days. This short cycle allowed for the repetition of the creep and
shrinkage testing three times in one year for each mixture.
AGGREGATE PROPERTIES
Table 1 presents the relative density and absorption capacity of the aggregate
used in this study. Coarse aggregates, granite and expanded slate, had the same
maximum size aggregate. The granite from Georgia had a relative density in saturated
surface dry condition (SSD) and absorption of2.681 and 0.71%, respectively. The
expanded slate had a SSD relative density of 1.551 with absorption of 8.1% of its weight.
Among the different lightweight aggregate available for concrete, expanded slate presents
one ofthe lowest absorption values. Nevertheless, due to the relatively high volume of
coarse lightweight aggregate (37% of the volume of the concrete), the water in the
aggregate represents 42.71/m 3• The curing water needed to reach the maximum
attainable degree of hydration can be estimated from the cementitious materials content,
initial W/B and estimated chemical shrinkage of the system. Following the estimation
procedures proposed by other authors [30, 31], the required curing water would be
between 20.3 and 24.4 11m3• Therefore, the water within the lightweight aggregate is 1.91
times the required water for attaining the maximum degree of hydration.
Fresh Properties
Slump and fresh density (ASTM C143, and ASTM Cl38) were measured for all
the batches. From the workability results, the HPLC-A and HPLC-B slump was 75 mm
± 35 mm. The HPC-A mixture had a slump of90 mm ±50 mm. Fresh density ofHPLC
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
varied from 1950 kg/m 3 to 1980 kg/m 3, regardless of the type ofHPLC. The HPC-A
mixture averaged a fresh density of2420 kg/m3 • The density of the HPLC mixtures was
80% of the weight of the normal weight HPC. The air content was not measured, but
from previous research done with HPLC-A it was between 3.3 and 4.5%.
Compressive Strength
All specimens used for mechanical testing, creep and shrinkage measurements
were accelerate-cured in insulated boxes. Compressive strength for laboratory mixtures
was measured using 100 x 200 mm cylinders at 24 hours, and then at 7, 28, 56, and 365
days. A total of four specimens were tested at each age on every cycle of the
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
666 Lopez et at.
experimental work. The average compressive strengths measured at these ages for the
HPLC-A, HPLC-B, and HPC-A mixtures are presented in Table 3 and Fig. 1.
From previous research [15, 19], it was concluded that accelerated-cured
specimens exhibited a higher strength than the ASTM-cured specimens at early ages.
However, at 28 days and beyond, the ASTM-cured specimens exhibited the higher
compressive strength. For that reason some of the specimens in the current study were
cured following the ASTM recommendation, at room temperature for at least 24 hours,
and then they were placed in the fog room until the age of 56 days when they were tested.
The HPLC-A accelerated-cured specimens had an average strength at 24 hours
of73.5 MPa which was very similar to the strength of the HPLC-B at the same age of
72.8 MPa. On the other hand, HPC-A mixture had a 24 hour compressive strength of
90.5 MPa. Since the only difference between HPLC and HPC mixtures was the coarse
aggregate, it can be concluded that the strength of the HPLC was limited not by the
mechanical properties of the cementitious matrix but by the lightweight aggregate
intrinsic strength. As a consequence of this strength ceiling, the HPC-A mixture
presented a compressive strength 23 and 28% higher than the HPLC mixtures at 24 hours
and 28 days, respectively. This is a very well-known phenomenon reported by many
researchers [1, 8, 10, 13, 40].
At 56 days, the average strength of the HPLC-A was 80.6 MPa while the
strength of the mixture using the lightweight aggregate in dry condition (HPLC-B) was
77.0 MPa. The difference in strength between the two HPLC mixtures kept increasing
until one year, when the HPLC-A mixture had a strength of 87.5 MPa versus the 77.5
MPa of the HPLC-B. The greater gain in strength of the HPLC mixture having pre-
wetted lightweight aggregate compared to the one that used air-dried lightweight
aggregate clearly showed the influence of the internal curing provided by the pre-wetted
lightweight aggregate which assured a continuous hydration of the cement and
supplementary cementing materials.
The HPC-A mixture presented a compressive strength at 56 days and at one year
of 106.5 and 122.2 MPa, respectively. The higher strength of the HPC-A mixture shows
the influence of having a coarse aggregate with a higher intrinsic strength [1, 8, 10, 13,
40].
Modulus of Elasticity
Modulus of elasticity was measured using 150 mm x 300 mm cylinders made
according to ASTM C469. Specimens with accelerated curing were tested at 24 hours
and 28 days. The average 24-hour and 28-day elastic moduli are shown in Table 4.
Fig. 2 presents the average values ofthe modulus of elasticity for each ofthe
three mixtures. The HPC-A mixture had a modulus of elasticity 7 and 9% higher than the
HPLC mixtures at 24 hours and 28 days, respectively. This represents a very small
difference when it is compared with the variability between two batches of the same
mixture; in fact, at 28 days one HPLC-A specimen had a modulus of elasticity virtually
equal to one of the specimens from HPC-A.
It can be stated that for these high strength/high performance mixtures, the
influence of the lightweight aggregate on the modulus of elasticity is little, especially
when it is considered that the difference in strength between HPC-A and the HPLC
mixtures was around 25%. This small difference in modulus of elasticity might be
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Creep and Shrinkage behavior of the HPLC-A, HPLC-B and HPC-A mixtures
In order to compare creep of the various specimens, which exhibited varying
strength, and because the creep on each mixture was measured under different stresses
(equivalent to the 40% of the compressive strength of the particular mixture at the time of
loading), the results were first expressed as specific creep (creep strain divided by the
applied stress) and then multiplied by 30 MPa. The value of30 MPa roughly represents
the 45% of69 MPa, which was the design strength of the HPLC-A mixture.
Figs. 3 and 4 present the average creep for the adjusted applied stress of30 MPa
and average shrinkage measured on the unsealed specimens of the three mixtures under
study. Fig. 3 shows the creep and shrinkage measured when loading and drying started at
24 hours while Fig. 4 presents the same data when the loading and drying started at 28
days. The two figures present time under load and drying in a logarithmic scale, and
creep and shrinkage are presented in microstrains (mm/mm x 10"6).
Creep and shrinkage behavior with loading and drving starting at the age of 24 hours
Fig. 3 shows clear differences in the creep strain of the three mixtures. HPLC-A
showed consistently lower creep strains than its HPC-A counterpart. An analysis of
variance (ANOV A) revealed that there were statistically significant differences between
the two mixtures and that creep ofHPC-A was higher than that ofHPLC-A. This was
true even though the compressive strength of the HPC-A mixture at time ofloading was
23% higher than that ofHPLC-A.
When the air-dried lightweight aggregate was used (HPLC-B), the creep strain
increased 2.2 times than that ofHPLC-A after 120 days. These results show the large
effect of the internal curing on the creep behavior of the HPLC mixtures. When
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Creep and shrinkage behavior with loading and drying starting at the age of 28 days
Fig. 4 shows creep and shrinkage strains of the mixtures under study when the
loading and drying started at the age of 28 days. The specimens were accelerate-cured
for 24 hours and then placed in the fog room for the remaining 27 days, until the time of
testing. As expected, creep of the specimens loaded at 28 days was lower than that
observed for specimens loaded at 24 hours. After 120 days under load, creep strains
under an adjusted stress of30 MPa were 400, 537and 365 microstrains, for HPLC-A,
HPLC-B, and HPC-A mixtures, respectively. This represents a decrease with respect to
the specimens loaded after 24 hours of 497 and 177 microstrains for HPLC-B and HPC-
A, respectively. Creep of the HPLC-A mixture decrease only by 70 microstrains between
the two ages of loading.
It is believed that the fog room provided good curing conditions for the HPLC-B
and HPC-A mixtures, so their performance improved considerably. On the other hand,
HPLC-A did not show a great improvement after 27 days of external curing in the fog
room because it already had a good curing through the water in the aggregate.
Since creep performance ofHPC-A was improved by the curing in the fog room,
average creep strain after 120 days under load was less than that ofHPLC-A. This
difference was too small to be considered statistically significant; in fact, the creep strain
between zero and 56 days under load of the two mixtures was virtually the same. In
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Total time dependent deformations with loading and drying starting at the age of 24
hours and 28 days
Fig. 5 presents the difference between creep plus shrinkage measured between
HPLC-A and HPC-A, and the difference obtained between HPLC-A and HPLC-B. Fig.
Sa presents the results obtained when loading and drying started at the age of 24 hours
while data Fig. 5b is from when loading and drying started at the age of28 days. The
horizontal axis presents time in logarithmic scale, and the vertical axis presents the creep
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
plus shrinkage measured in the correspondent mixture minus the creep plus shrinkage
measured in HPLC-A.
Since all the data points from Fig. 5a are negative, it can be concluded that creep
plus shrinkage ofHPLC-B and HPC-A were always greater than that ofHPLC-A. After
120 days of measurements, this difference was almost 600 microstrains for the HPLC-B
mixture and around 150 microstrains for HPC-A. It is clear for Fig. 5a, that the
difference is not constant with time, but continuously increased for the HPLC-B mixture.
For HPC-A, it can be seen that the difference steadily increased up to 28 days under
drying and load, and then the difference reached a plateau at around 150 microstrains.
ANOVA revealed that all the differences presented in Fig. 5a were statistically
significant with a confidence level of 95%.
Fig. 5b shows smaller differences between the time dependent deformations of
the mixtures under study. The greater difference, of around 180 microstrains was
between HPLC-B and HPLC-A after 120 days under load. The difference in creep plus
shrinkage between HPC-A and HPLC-A produced positive and negative values meaning
that there was not a consistent difference between those two mixtures. In fact, the
ANOV A concluded that the difference was not statistically significant with a confidence
level of95%.
Since the main difference between the time dependent behaviors was in the
creep portion, the differences presented in Fig. 5 can increase importantly if a stress
greater than 30 MPa is applied.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
670 Lopez et al.
DISCUSSION, CONCLUSIONS AND RECOMMENDATIONS
creep plus shrinkage of HPLC-A was lower than that of a HPC mixture of similar
strength when the load and drying started at 24 hours of age. They also reported that the
difference continued to increase until 620 days under load and drying.
While these results show lower long-term deformation for HPLC, as compared
to HPC, it is emphasize that these results are preliminary and are limited to only the
reported concrete mixtures. The results shown herein include those obtained from
several batches after 120 days under load. The creep and shrinkage ofHPC mixtures
may vary widely depending on the aggregates, supplementary cementitious materials,
and admixtures, and laboratory testing ofHPC mixtures is recommended prior to their
application in the field.
Further research should focus on measuring changes on internal relative
humidity in sealed and unsealed specimens, produced with pre-wetted and air-dried
lightweight aggregate and normal weight aggregate, under loading and drying conditions.
Results from this research should be implemented and confirmed on full scale
prestressed concrete members.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Department of Transportation, or any other sponsoring or cooperating organization.
The authors would like to express their appreciation to the reviewers for their
thoughtful comments on the original manuscript.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
internal curing using saturated lightweight fine aggregate". Cement and
Concrete Research, 29(11): 1999. p. 1863-1867.
31. Jensen, O.M. and P.F. Hansen, "Water-entrained cement-based materials I.
Principles and theoretical background". Cement and Concrete Research, 31 (4):
2001. p. 647-654.
32. Jensen, O.M. and P.F. Hansen, "Water-entrained cement-based materials II.
Experimental observations". Cement and Concrete Research, 32(6): 2002.
p. 973-978.
33. Lura, P., D.P. Benz, and D.A. Lange. "Measurements of Water Transport from
Saturated Pumice Aggregates to Hardening Cement Paste." in Advances in
Cement and Concrete. Copper Mountain, Colorado: Engineering Conferences
International, 2003. p. 89-99.
34. Holm, T.A., O.S. Ooi, and T.W. Bremner. "Moisture Dynamics in Lightweight
Aggregate and Concrete". in Theodore Bremner Symposium on High-
Performance Lightweight Concrete. Tessaloniki, Greece, 2003. p. 167-184.
35. de Larrard, F., P. Acker, and R. LeRoy, "Shrinkage Creep and Thermal
Properties", in High Performance Concrete: Properties and Applications, S.P.
Shah and S.H. Ahmad, Editors. McGraw-Hill: New York, NY. 1994, p. 65-114.
36. Yang, Y., et a!. "Experimental Investigation on Shrinkage and Creep of High
Strength Concrete at Early Ages". in Fourth CANMET I ACI I JCI International
Symposium on Advances in Concrete Technology, SP-I79. Tokushima, Japan:
American Concrete Institute, 1998. p. 201-215.
37. Pfeifer, D.W., "Sand replacement in structural lightweight concrete-- Creep and
shrinkage studies". American Concrete Institute-- Journal, 65(2): 1968. p. 131-
140.
38. Berra, M. and G. Ferrada. "Normalweight and Total-Lightweight High-Strength
Concretes: A Comparative Experimental Study". in High-Strength Concrete.
Second International Symposium. Berkeley, California: American Concrete
Institute, 1990. p. 701-733.
39. Vincent, E. C., Compressive Creep of Lightweight, High Strength Concrete
Mixture, in Civil Engineering. Virginia Polytechnic Institute and State
University: Blacksburg. 2003, p. 137.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Development Bulletin RD 104 Portland Cement Association. 1994.
48. Malhotra, V.M. and P.K. Mehta, High-Performance, High-Volume Fly Ash
Concrete: Materials, Mixture Proportions, Properties, Construction Practice,
and Case Histories. 2nd ed. Ottawa: Supplementary Cementing Materials for
Sustainable Development Inc., 2005.
49. Ayers, M.E. and M.S. Khan. "Overview of Fly Ash and Silica Fume Concretes:
The Need for Rational Curing Standards". in Concrete Technology: Past,
Present, and Future, Proceedings ofV Mohan Malhotra Symposium: American
Concrete Institute, 1993. p. 605-617.
50. Jensen, O.M. and P.F. Hansen, "Autogenous deformation and RH-change in
perspective". Cement and Concrete Research, 31(12): 2001. p. 1859-1865.
LIST OF SYMBOLS
WIB: water-to-binder ratio by weight
HPLC-A: HPLC using pre-wetted coarse lightweight aggregate at the time of
mixing HPLC-B: HPLC using air-dried coarse lightweight aggregate at the time of
mixing HPC-A: HPC using granite as coarse aggregate
-
Table 2 -- Mixture proportions o fHPL C-A HPLC B andHPC A miXtures
. -
HPLC-A HPLC-B HPC-A
Ty]:le III cement kg/mj 442 442 442
class "F" fly ash kg!m' 89 89 89
silica fume kg/m3 59 59 59
12.7 mm aggregate kg/m' pre-wetted air-dried normal
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
expanded expanded weight granite
slate 571 slate 531 987
normal weight natural sand kgfm' 612 612 612
waterkg/mj 135 175 135
water reducer lim' 2.8 2.8 2.8
superplasticizer lim' 5.9 5.9 5.9
air entrainer agent 1Im3 0.5 0.5 0.5
water/cementitious ratio 0.23 0.23 0.23
cement paste content(%) 39 39 39
coarse/fine ratio by volume 1.5 1.5 1.5
theoretical air content (%) 3.5 3.5 3.5
density kg/m3 1918 1918 2334
- -
I
120 I
- -
~ 110
OS 'JI
~ 100 I
""
~
90
80 ...
-
....
.. -- I
!l .. ~
I
<ll 70
I
.::" 60 X HPLC-A
[IS
50 <> HPLC-B
!:l r-
S'
0
40 ::1( HPC-A
u 30 f- --Log. (HPLC-A)
I
20 • • • · Log. (HPLC-B)
I
10 I Log. (HPC-A)
0
10 100 1000
Maturity (days)
Fig. 1 --Compressive strength vs. maturity of HPLC-A, HPLC-B, and HPC-A mixtures.
28
Age (days)
Fig. 2 -- Modulus of elasticity vs. age of HPLC-A, HPLC-B, and HPC-A mixtures.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~
l
-800 c--
-600
+ Creep HPLC-B
II Creep HPC-A
<> Shrinkage HPJ r -~
+Shrinkage HPC-A
.. ~
·i -400 ••
~ -200
II ~I
II Ill'
~·
~ ~~
•=
0
0,01
I~ il
0.1 1
~
10 100 1000
Time under Loading and Drying (days)
Fig. 3 --Creep and shrinkage of HPLC-A, HPLC-B, and HPC-A mixtures loaded at the
age of 24 hours.
-1200
IIIIII II IIIII II
~ -1000 - I I Creep HPLC-A X Shrinkage HPLC-A
-;:;
_ +Creep HPLC-B o Shrinkage HPLC-B
~ -800
+ Shrinkage HPC-A
1
- -600
DCreep HPC-A
·t 400
•
....
r
Ill
i
-200
.: II
0
0.01 0.1
111111 ~ lfl 10
~ ~)
100 1000
Time under Loading and Drying (days)
Fig. 4-- Creep and shrinkage of HPLC-A, HPLC-B, and HPC-A mixtures loaded at the
age of 28 days.
l-300
·j -200 Ill ~~
j -100 •• •
Ill ...
_n I ill
il'
I
• I
0
lm I
I
0.01 0.1 10 100 1000
Time under Loading and Drying (days)
b -600
I I
• Difference in Creep + Shrinkage between I
~ -500 HPC-A and HPLC-A
0
';<
0 Difference in Creep + Shrinkage between
13 -400
HPC-A and HPLC-A
~§, -300
.S
.s -200
,4 ~
§ •
~ -100
0 [I'
- .~ .~ b
I I
0.01 0.1 10 100 1000
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Time under Loading and Drying (days)
Fig. 5 --Differences in time dependent deformations between HPLC-A and HPC-A, and
between HPLC-A and HPLC-B loaded at the age of 24 hours (a) and 28 days (b).
Synopsis: This paper describes the development and material testing results of
high strength lightweight concrete (HSLC) mixture proportions for use in precast,
prestressed concrete highway girders. As part of a research project at Georgia Tech,
it was necessary to develop concrete mixture proportions using slate lightweight
aggregate able to achieve design strengths of 8,ooo and 1o,ooo psi (55.2 and
69.0 MPa) with an equilibrium unit weight of approximately 120 pcf (1922 kg/m3) or
less. A methodical approach to proportioning mixture components was developed
and used to configure HSLC mixtures. Field testing was conducted to verify the
mixture proportions for use in a production environment. An extensive material
testing program based on specimens cast during all phases of the research project
produced a database of material properties to include strength, modulus of elasticity,
rupture modulus, coefficient of thermal expansion, and chloride permeability.
Brandon S. Buchberg, is an engineer with Simpson, Gumpertz & Heger, Inc. in Boston.
He earned his Master of Science Degree in Civil Engineering (Structures) from the
Georgia Institute of Technology in 2002. His research focused on the material properties
of high-strength lightweight concrete.
Lawrence F. Kahn, ACI Fellow, is a Professor in the School of Civil and Environmental
Engineering at the Georgia Institute of Technology. His research interests include
prestressed concrete and structural retrofit techniques. He is currently a member of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
Increases in the number of lanes on interstate highways coupled with
interchange site restrictions have been an impetus for extending the length of precast,
prestressed concrete highway girders. Advances in concrete quality and engineering
practices have enabled the design and construction of precast prestressed concrete bridge
girders that approach 200 feet in length. Problems occur when trying to move and erect
these long and heavy girders. In order to facilitate easier road movement and erection,
high-strength lightweight concrete (HSLC) can be used to reduce member self-weight
while still allowing a large load carrying capacity.
A research project was conducted at the Georgia Institute of Technology that
examined the use ofHSLC for prestressed highway girders. One goal of the project was
the development of HSLC mixture proportions using slate lightweight aggregate
achieving 28-day design compressive strengths (fc') of 8,000, 10,000 psi, and 12,000 psi
(55.2, 69.0, and 82.8 MPa) and having equilibrium unit weights less than approximately
3
120 pcf (1922 kg/m ). An analytical study conducted as part of the project showed that a
compressive strength of 10,000 psi (55.2 MPa) would be required to achieve a length of
150 ft using standard PCI Bulb-Tee girder configurations. 1 In addition to mixture
proportion development in the laboratory, field verification at a prestressed concrete plant
was performed to ensure commercial production of the mixture proportions was feasible,
and to fabricate Type II AASHTO girders for subsequent flexure, shear and strand
development evaluation.
This paper presents an overview of a methodical approach developed for
proportioning HSLC mixtures, an in-depth discussion of mechanical property testing on
both laboratory and field-produced slate HSLC specimens, and a discussion of some of
the challenges faced in producing high quality HSLC in a field environment.
Materials
To produce HSLC able to reach compressive strengths in excess of 10,000 psi
(69.0 MPa), slate lightweight aggregate (LWA) from North Carolina was selected based
on its performance characteristics. The 1/ 2 -in. and 3/ 8-in. (13-mm and 10-mm) slate LWA
used during the study had oven dry specific gravity values of 1.44 and 1.52, respectively.
Both aggregate sizes absorbed 6 percent moisture by weight from an oven dry state after
soaking for 24 hours and had maximum absorption values of about 10 percent. 3
Natural normal weight fine aggregate (NWA) with a fineness modulus of 2.36
and an average specific gravity of2.62 was used for the fine aggregate. The natural sand
was used in order to achieve the highest compressive strengths and because initial
mixtures showed that maintaining consistent water content when using expanded slate
lightweight sand was difficult in the laboratory and probably not possible in field
manufacturing.
Type III portland cement with a specific gravity of 3 .08 was used during both
the laboratory and field phases of the project based on its superior early strength gain as
compared to Type I cement.
Silica fume and Class F fly ash meeting requirements of ASTM C 618 4 were
used in all mixtures to assure low permeability.
To improve workability and durability, and to meet entrained air requirements, a
high-range water reducer (HRWR), low-range water reducer (LRWR), and air entraining
admixture (AEA) were incorporated in the mixture proportions. The LRWR was
important in slowing the cement hydration while the HRWR helped to increase
workability. Workability was a significant concern considering that the HSLC mixtures
generally had a W/CM less than 0.32. The use of an AEA helped achieve a target
4 percent air content. --`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
HSLC mixing was accomplished in the lab with a 1.75 cubic foot (0.05 cu. m)
capacity pan mixer having a shearing type mix action. Some trial mixtures were also
made in a 4 cubic foot (0.11 cu. m) rotary mixer; the latter mixtures showed consistently
lower strength.
Mixing in the field was accomplished at a local prestressed concrete
manufacturer using a 3 cubic yard (2.3 cu. m) auger-style mixer. The mixer was charged
using conveyor belts for the aggregate and a chute for the cement. Since the mix tower
did not contain silos for fly-ash and silica fume, those components were added to the
mixer manually.
Components were added and mixing was accomplished in the following order:
all aggregate and 2/3 of the mixture water plus chemical admixtures followed by 30
seconds of mixing; then all cementitious material and the remaining water followed by 3
minutes of mixing. Additional HRWR was added if needed to achieve a 4 to 6-in. slump.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the study in Table 6 indicated that the three different batches were not statistically
different and that based on 4 x 8 compression strength results the G2B batch series
satisfied the requirements for a 10,000 psi design strength mixture. Based on 6 x 12
results, the batch series did not satisfy the requirements for a 10,000 psi design strength
mixture, but did qualify for a design strength of9,988 psi.
The linear trend resulting from Equation 2 used to predict modulus of elasticity,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
normalized by (fc') 0 ·5 , is also plotted in Fig. 6. Equation 2 contains the term w/ 5 , which
emphasizes the effect of unit weight. The resulting linear trend does not compare
favorably with the trend indicated by experimental values.
Similar evaluations with other variable forms indicated that (fc') 0.s was the
variable that appeared most appropriate in predicting modulus of elasticity.
After examining several equation forms, Equation 4 was determined to be the
equation that best predicted the modulus of elasticity of the slate HSLC values examined.
Ec = 44,000JJ; we (4)
145
Overall, Equation 4 provided a better prediction of modulus of elasticity for
slate HSLC than the previously suggested Equation 1. Equation 4 gave modulus of
elasticity values about 0.2 percent less than the experimental data with a standard
deviation of 5.1 percent. The linear trend resulting from Equation 4 used to predict Ec
values is also plotted in Fig. 6. Since Equation 4 contains the term wc05 , which de-
emphasizes unit weight, the resulting linear trend compares more favorably with the trend
indicated by the experimental values.
Low, mean, and high unit weight values of 111.5 pcf, 118.2 pcf, and 134 pcf,
respectively, were determined from the 56-day experimental modulus of elasticity values
from this research project and modulus of elasticity values from previous research. 1 Fig.
7 shows a plot of the above-mentioned values vs. compression strength and also shows
lines plotting the predicted modulus of elasticity using Equation 4 based on the low,
mean and high unit weight values. Fig. 7 shows that at a given compressive strength, the
variation in modulus of elasticity can be substantial. At a compressive strength of
approximately 10,600 psi, the modulus of elasticity varied from approximately
3.65 million psi to 4.45 million psi.
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 689
Figs. 8 and 9 show Equation 3 and Equation 2, respectively, evaluated using the
low, mean, and high unit weight values plotted over the same modulus of elasticity
values shown in Fig. 7. Equation 3 unconservatively predicted modulus of elasticity for
slate HSLC; Equation 2 conservatively predicted modulus of elasticity for slate HSLC.
Chloride Permeability
To measure chloride permeability, 4-in. (102-mm) diameter x 8-in. (205-mm)
long cylindrical specimens were cast according to ASTM C 31. 9 At the time of testing,
the specimens were removed from the fog room and cut using a water-cooled diamond
saw into four pieces, each being 2 inches (52 mm) in length. Only the two inner 2-in.
(51-mm) disks from each cylinder were used for the chloride permeability tests. Once
cut, the 2-in. (51-mm) thick chloride permeability specimens were conditioned and tested
according to ASTM C 120223 using a PROOVE-IT system. Table 9 shows results for
rapid chloride permeability testing results. Each value listed is the average of four
specimens tested. According to ASTM 1202, results below 1000 coulombs are termed
"very low;" results below I 00 coulombs are termed "negligible."
The PROOVE-IT system results are based on 3.75-in. (95-mm) diameter
specimens. When using 4-in. (102-mm) diameter specimens as in this project, it was
necessary to correct the results using Equation 6.
Qs = (QJ * (3.75/ d)
2
(6)
Qs =charge (coulombs) passed through a 3.75-in. diameter specimen
Qx =charge (coulombs) passed through "d" in. diameter specimen
d = diameter (in.) of the nonstandard specimen, 4-in. in this project
Table 9 shows that as the silica fume percentage increased, the amount of charge passed
through the specimens decreased. Because of its fineness, the silica fume was able to fill
voids that exist in the structure of the cement paste. As these voids were filled, the ability
for water to travel through the cement paste structure, either by diffusion or absorption,
decreased. A silica fume content of 10% of total cementitious by weight provided the
lowest chloride permeability in this study. Table 9 also shows that the chloride
permeability decreases as the W/CM decreases. A lower W/CM provides for a denser
matrix, decreasing the chloride permeability. In general, the chloride permeability
decreased as the 56 day compressive strength increased.
CONCLUSIONS
Both 8,000 psi (55.2 MPa) and 10,000 psi (69.0 MPa) design strength high
performance lightweight concretes were developed using expanded slate Yz-in. (13-mm)
aggregate and natural sand. Mixtures made at the precast plant had slightly higher
strengths than those produced in the laboratory. In all cases, close attention to moisture
content in the lightweight aggregate was required. Accelerated cured cylinders, which
matched curing conditions of precast beams, showed about 25 percent higher one-day
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 691
strength than ASTM cured cylinders; at 56 days, the ASTM cured cylinders were about
4% stronger than the accelerated cured cylinders. A 12,000 psi design strength mix could
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
not be developed.
Equation 4 was developed to provide a better estimate of modulus of elasticity
for slate HSLC than previous relations for normal strength concretes. The ACI equation
for high strength concrete (Equation 2) gave Ec values about 9 percent less than those
found for slate HSLC.
The CTE for slate HSLC is consistent with CTE values for other L WC and
NWC specimens tested. Chloride permeability was reduced as the silica fume content
was increased, and as the W /CM decreased. Chloride permeability was consistently less
than 1000 coulombs.
Overall, the research demonstrated that high-strength/high-performance
concretes can be developed with equilibrium unit weights less than 122 lb/ft 3 for precast
prestressed bridge applications.
ACKNOWLEDGMENTS
This research was sponsored by the Georgia Department of Transportation under
Georgia DOT Task Order No. 97-22, Research Project No. 2004. Carolina Stalite
Company donated all lightweight aggregate for this project. Grace Construction Products
donated chemical admixtures and silica fume. Bora! Industries donated all Class "F" fly
ash. Lafarge Corporation (previously Blue Circle Materials) and CEMEX donated all
Type III Cement. Tindall Corporation, Jonesboro, Georgia, provided Georgia Tech
researchers with equipment, manpower, and materials for the field production study.
These sponsors and their advice and cooperation are gratefully acknowledged.
Mr. Adam Slapkus, Mr. Mauricio Lopez, Ms. Natalie Hodges, and Ms. Maria Wilmhoff
assisted in the experimental work; Professors Kimberly Kurtis and James Lai provided
valuable insights.
The conclusions and opinions expressed herein are those of the authors and do
not necessarily represent the opinion, conclusions or policies of the Georgia Department
of Transportation or of the other sponsors.
REFERENCES
1. Meyer, K. F., Kahn, L. F., "Lightweight Concrete Reduces Weight and
Increases Span Length ofPretensioned Concrete Bridge Girders," PC! Journal,
Vol. 47, No.1, January-February 2002, pp. 68-77.
6. ASTM C 138, Standard Test Method for Unit Weight, Yield, and Air Content
(Gravimetric) of Concrete, American Society for Testing and Materials, West
Conshohocken, P A, 2000, 3 pp.
7. ASTM C 1064, Standard Test Method for Temperature of Freshly Mixed Portland
Cement Concrete, American Society for Testing and Materials, West Conshohocken,
PA, 1999,2 pp.
8. ASTM C 173, Standard Test Method for Air Content of Freshly Mixed Concrete by
the Volumetric Method, American Society for Testing and Materials, West
Conshohocken, PA, 1994, 3 pp.
9. ASTM C 31, Standard Practice for Making and Curing Concrete Test Specimens in
the Field, American Society for Testing and Materials, West Conshohocken, PA,
1998, 5 pp.
10. ASTM C 39, Standard Test Method for Compressive Strength of Cylindrical
Concrete Specimens, American Society for Testing and Materials, West
Conshohocken, PA, 1999, 5 pp.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
11. ASTM C 469, Standard Test Method for Static Modulus of Elasticity and Poisson's
Ratio of Concrete in Compression, American Society for Testing and Materials,
West Conshohocken, P A, 1994, 4 pp.
12. ASTM C 78, Standard Test Method for Flexural Strength of Concrete (Using Simple
Beam with Third-Point Loading), American Society for Testing and Materials, West
Conshohocken, PA, 1994,3 pp.
13. ASTM C 1231, Standard Practice for Use ofUnbonded Caps in Determination of
Compressive Strength of Hardened Concrete Cylinders, American Society for
Testing and Materials, West Conshohocken, PA, 2000,4 pp.
14. CRD-C 39, Test Method for Coefficient ofLinear Thermal Expansion of Concrete,
U.S. Army Corps of Engineers Engineer Research and Development Center,
Vicksburgh, Ml, 1981,2 pp.
15. Building Code Requirements for Structural Concrete, ACI 318-05, and Commentary,
ACI 318R-05, American Concrete Institute, Detroit, 2005.
18. Shideler, J.J., "Lightweight-Aggregate Concrete for Structural Use," Journal of the
American Concrete Institute, Vol. 54, No. 10, October 1957, pp. 299-328.
19. Hanson, J.A., "Tensile Strength and Diagonal Tension Reinforcement of Structural
Lightweight Concrete," Journal of the American Concrete Institute, Vol. 58, No. 1,
July 1961, pp. 1-37.
20. ASTM C 496, Standard Test Method for Splitting Tensile Strength of Cylindrical
Concrete Specimens, American Society for Testing and Materials, West
Conshohocken, P A, 1996, 4 pp.
23. ASTM C 1202, Standard Test Method for Electrical Indication of Concrete's Ability
to Resist Chloride Ion Penetration, American Society for Testing and Materials,
West Conshohocken, PA, 1997, 6 pp.
Table l - 8,000, 10,000, and 12,000 psi (55.2, 69.0 and 82.8 MPa) Mixture
Proportions for Field Testing
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
4 x 8Cy1inders 6 x 12 Cylinders
Mean Required Standard Mean Required Standard
Batch# Compressive Mean Deviation Compressive Mean Deviation
Strength t'., s Strength t'., s
(psi) (psi) (psi} (psi} (psi) (psi)
G2B-1 10 717 10 730 528 10 930 10 300 224
G2B-2 10464 10 740 532 10 349 10428 319
G2B-3 10 703 10 214 160 10 106 10 557 416
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
12,000 12L 4.0S 4.25 4.24
Curebox G2A 3.55 3.93
G2B 3.91 4.10
12F 4.40
12,000 12L 4.33
ASTM G2A 4.06
G2B 4.05
Objective
Batch
Mean Modulus of Rupture !, Concrete
Strength
(psi)
Series
(H=Hours, D=Days)
(psi) ;_Jl Compressive
Strength (psi)
& Curing 24H 56D 56D 56D
8F 788 1,089 12.43 10,160
8,000 8L 649 1,077 12.40 10,430
Curebox GlA 1,042 12.86 9,080
~ 830 11.09 7,750
8F
8,000 8L 761 1,030 11.81 10,520
ASTM GlA 992 11.21 9,350
---rna- 743 9.50 8,460
10,000 10F 641 998 11.10 11,170
Curebox 10L 670 1,164 13.74 9,920
10,000 lOF
ASTM 10L 678 1,006 11.26 11,040
12F 761 1,014 11.10 11,550
12,000 l2L 645 926 10.45 10,860
Curebox G2A 1,283 14.79 10,420
~
G2B 1,038 12.06 10,250
12F
12,000 12L 678 918 10.07 11,480
ASTM I G2A I 1156 13.07 10 820
G2B 805 9.23 10,510
12,000 ~------------~
J 9,000 .J~-==---'<
- GIA-Curebox
-G1B-Curebox
g
-~ 8,000 ~~~~~-----::..;;::;:::::.:::::::.:·::::·::·
J~ =··=·::·=-·~ • ·•· ·8F-ASTM
• ·• • · 8L-ASTM
S"
..
7,000 • ·A • · GIA-ASTM
8 6,000 ~-------------1
.j:..:._.,_.=-. • ·•· ·GlB-ASTM
5,000 -h-~.,...,...~~..........~~-~.,...,...~,...,r...-r-~
0 10 20 30 40 50 60
Cylinder Age (Days)
Fig. 1- Compression Strength vs. Cylinder Age for 8,ooo psi Mixtures to 56 Days.
12,000
'j 11,000
•. -.-.-.--.-.-·..a
~--··
t
!
·I
10,000 ~
9,000
8,000
•"',"
.. ,.
I~:··#
- -10F-Curebox
-10L-Curebox
• ·• · · 10F-ASTM
• ·• • ·!OL-ASTM
7,000
S'
0
u 6,000 •
5,000
0 10 20 30 40 50 60
Cylinder Age (Days)
Fig. 2- Compression Strength vs. Cylinder Age for 10,000 psi Mixtures to 56 Days.
12,000
_ ......
~ 11,000
.. ...... .- ......
. .. :;;;
-12F-Curebox
.,g
!""
10,000
9,000
... .
1-..- .. : •• ---~ .· .·
-
-
121-Curebox
G2A-Curebox
~-··
-G2B-Curebox
"'=
-~
~
8,000 . • ·+··12F-ASTM
··•··12lrASTM
7,000
S'
0 • ··•··G2A-ASTM
u 6,000 • ·•· ·G2B-ASTM
5,000
0 10 20 30 40 50 60
Cylinder Age (Days)
Fig. 3- Compression Strength vs. Cylinder Age for 12,000 psi Mixtures to 56 Days.
~
~ 4.5E+06 +-------..----:;:--c-_....___ _ _ _ _ _ _ _-1
:i
~ 4.0E+06 f-f=::::::::;:;;;4.tj~~;:::=;~_:_!!~----;;;:-----~
i
0
+--.----.---r--.---.--.-----,,..---.--,--~
';;;' 5.0E+06 - , - - - - - - - - - - - - - - - - - - - - ,
-e.
·-E' 4.5E+06 +--------------.-----:--.-~
·~
~ 4.0E+{)6 +-----,--------- -:--.a.-~=:>rr.r-"--'-1
0
.E 3.5E+06 +----:-~,......::~:.;-=---__:_--.,------,lr----------1
~
::; 3.0E+06 +--,--.,...---,--,---....--.---.---r---r-~----1
6,000 7,000 8,000 9,000 10,000 11,000 12,000
Corqression Streq¢1 {pli)
• lli;~ I
- -Iirar (AnFlqJerirrmal Vahrs)
4.5E+04 +-----------------------1
3.0E+04 -1-----~--------~---~-~----1
110 115 120 125 130 135
Unit Weight (pcf)
5.75E+06 , . - - - - - - - - - - - - - - - - - - - - - - - - - - ,
]; 5.25E+06
2.75E+06 1---~---~---~---~------__j
6,000 7,000 8,000 9,000 10,000 11,000 12,000
Compression Strength (psi)
Fig. 7- Equation 4 Plotted Using Low, Mean, and High Unit Weight Values and
Test Results.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2.75E-Kl6 +----,.---..,.------,,...----,----...-----~
6,000 7,000 8,000 9,000 10,000 11,000 12,000
Compression Strength (psi)
Fig. 8- Equation 3 Plotted Using Low, Mean, and High Unit Weight Values and
Test Results.
5.75E-Kl6
<;;' 5.25E-Kl6
,e,
.~ 4.75E+06
·il
~ 4.25E-Kl6
... ... ...
.... "i.( .~t¥
~·
••••• 6
'0
]
.g
3.75E+06 . . 6
6 b. •
6
• ••
·---;:-
6
..
ft• ~--
• -
;::;: 3.25E-Kl6 ~-
~~
2.75E-Kl6
6,000 7,000 8,000 9,000 10,000 11,000 12,000
Compression Strength (psi)
Fig. 9- Equation 2 Plotted Using Low, Mean, and High Unit Weight Values and
Test Results.
Gregory Sirotin, Ph. Dr., is a researcher at the Research Institute of the College of Judea
and Samaria, Ariel, Israel. His research interests include behavior and properties of
concrete mixtures and technology of concrete. He holds a Ph. Dr. degree from the
Polytechnic University, Minsk, Belarus.
Ya'akov Eshel, Ph. Dr., is head of the Samaria and Jordan Rift R&D Center, Ariel, Israel.
His research interests include utilization of coal ashes from power stations in lightweight
concrete.
INTRODUCTION
The mains problems of utilizing large volumes of highly porous bottom ash as an
aggregate lie in the low strength of the bottom ash particles and the enhanced activity
concentrations (in a number of cases) of above mentioned radionuclides. Table 1 presents
the activity concentrations of these radionuclides in bottom ash produced at a power
station in Israel, from different sources of coal (used in our experiments). These features
of bottom ash currently limit its use for thermal insulating/structural and structural
lightweight concretes. To overcome these limitations, the following technological
solutions were suggested:
a) A combined use of bottom ash and fly ash. This combination results in the
formation of a dense matrix of the hardened cement/fly ash paste 14• 15 • The obtained
binary concrete (according the number of by-products used) shows potential as a thermal
insulating lightweight concrete based on coal ashes. However, due to the high total
volume of bottom ash and fly ash in the concrete, ashes with low concentrations of
radionuclides must be used 16 •
b) A combined use of bottom ash and fly ash with unprocessed crushed sand
(UCS). UCS is a widely available by-product of stone quarries, which is more dense than
bottom ash and contains a minimum of the undesirable radionuclides. Due to its very low
radioactivity level, UCS produced from carbonate rocks is most effective for use as a
diluting material for the active coal ashes. For example, the activity concentrations of
undesirable radionuclides (Bq/kg) in UCS from one source in Israel are: 226Ra- 16.8±0.9:
232
Th- 0.9±0.2: 4°K- 12.5±3.5 (data of the Soreq Nuclear Research Center, Israel). The
use of UCS as an additional aggregate, replacing part of the bottom ash, increases the
strength and reduces the radioactivity of the ternary lightweight concrete 17• 18 •
In both binary and ternary lightweight concretes, fly ash is used to create a solid
cement-fly ash paste matrix, which compensates for the low strength of bottom ash.
Incorporation of fly ash in producing lightweight concrete with highly porous bottom ash
as an aggregate is essentially different from normal weight concrete. Accordingly, in our
opinion, the role of fly ash in increasing the durability of these concretes can be explained
through the following features of its proportion and structure.
EXPERIMENTS
Materials
Cement- Portland cement similar to ASTM C150, Type 1, was used in all experiments.
The physical properties of the cement were as follows: relative specific gravity - 3.15;
Blain's finesse-370m 2/kg; 28 day compressive strength of standard prism- 41.5 MPa.
Coal ashes - The samples of fly ash and bottom ash were taken from the one of Israel's
power station. The ashes were produced by combustion a batch of coal supplied from one
of the South African sources. The sample of fly ash for the experiments was taken in dry
conditions from the power station silo. The bottom ash was conveyed to the power station
storehouse by hydraulic transport means. The wet sample of bottom ash was dried in the
laboratory to a constant mass.
Fly ash - Data of the chemical analysis of the fly ash used is shown in Table 2. Fly ash
met the requirements of ASTM C618, Class F. The sum of Si02 + Ah0 3 + Fe 20 3 was
81.57% (the standard requirement is min. 70%). The value of S0 3 was 2.98% (the
standard requirement is max. 5%). Loss of ignition was approximately 3% (the standard
requirement is max. 6%). The relative specific gravity of the fly ash was 2.25; Blain's
finesse was 385m2/kg.
Bottom ash- The grading of bottom ash is presented in Table 3. The loose bulk density
was 600 kg/m3 .
Unprocessed crushed sand (UCS) - The sample of UCS was taken from one of the
Israel's dolomite quarries. The grading ofUCS met the specifications of ASTM C33, with
the exception of a lower content of particles passing sieve 1.18 mm, and higher content of
particles passing sieve 0.15 mm. The percentage of the particles passing sieves according
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
For proportioning the concrete mixtures for the experiments, the following
considerations were taken in account:
1. The cement content complied with ASTM C330 (intended for the evaluation of
aggregates for structural lightweight concrete) was 335 kg/m 3 .
2. The content of unprocessed crush sand (UCS) was constant (435 kg/m3), in
order to exclude influence of its varying on the concrete properties.
3. Fly ash was used as a main varied component. The variation of fly ash content
was in the range ofO to 335 kg/m 3.
4. Bottom ash content was in the range of 590 to 360 kg/m 3 , according to the fly
ash content.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
5. Water content was such that the concrete mixture consistency was constant
(Vebe time 15 s).
Mixture proportions for the experiments are presented in Table 4. The concrete
specimens were prepared in a laboratory pan mixer. Casting the concrete specimens in
molds (lOOxlOOxlOO mm), compaction by vibration and curing in a moistening chamber
were performed using conventional methods.
Concrete density and the strength were determined by using standard methods.
The volume of pores and capillaries in the hardened cement or the cement-fly ash
paste was calculated as the difference between the volume of pores and capillaries in the
concrete, and volume of pores and capillaries in the aggregate accessible for filling with
water. The latter volume was determined from the data on water absorption of bottom ash
in the concrete mixture 14 • In our case, it was 17.5%. The water absorption ofthe UCS is
essentially lower and can be ignored.
For our experiment, the concrete prisms with dimensions of 50x50x1 00 mm were
sawed out from the above-mentioned cubic specimens with dimensions of lOOxlOOxlOO mm.
The prisms were cured in an oven at a temperature 100±5°C before the test. The cured
prisms were installed vertically over a container of water (temperature 20±5°C) and
immersed to a depth of 10 mm. The mass of absorbed water was determined on the
difference of the specimens mass before and after test.
As an index of capillary absorption, the mass of water absorbed per unit of the
suction surface (50x50 mm) during the time of test was accepted.
Fig. I presents the effect of fly ash content in the concrete on reducing the
capillary absorption index. The timing of test critically affects the value of index;
however, the relative reduction in indices for the time conditions varied slightly. When
FA/(FA+C)=0.33, the index decreased by a factor of 1.24-1.25; when FA/(FA+C)=0.5,
the index decreased by a factor of 1.53-1.55.
The current standard method for determining the resistance of concrete to varying
environmental humidity and temperature conditions is a freeze-thaw cyclic test.
However, our experiments were conducted with reference to conditions of the
Mediterranean climate. This climate is characterized by a very hot summer with
temperature extremes (especially in desert regions), and a very wet winter with prolonged
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Any test method for evaluating the failure of concrete must conform to the abov:e-
mentioned features of the considered climate. For this purpose, we selected a method for
evaluating the soundness of aggregate using sulfates crystallization (ASTM C88,
AASHTO Tl04), modified for testing concrete specimens. This method is based on an
application of alternative cycles of immersion of aggregates in saturated solutions of
Na2 S04 or MgS0 4 at a temperature of 25±5° C and drying at a temperature of 11 0±5° C.
Crystallization of sulfates causes pressure in the pores and capillaries. The method
simulates the expansion of water when frozen and is used in several countries for this
purpose. However, it is also accepted as a standard in India, characterized by a hot
climate, for evaluating soundness of aggregates (IS 383). This procedure is considered as
essentially more strict than the freeze-thaw cyclic test.
Using this method for evaluating the durability of concrete, we took in account
that the procedure does not allow us to distinguish damage by sulfate crystallization from
damage by sulfate attack. However, character of macro-cracks in the concrete after a
certain number of cycles can confirm the presence of a high pressure in pores of hardened
cement-fly ash paste. Such cracks resulting from an expansion of sulfate crystals play a
major role in the destruction of concrete. This reason allows using the crystallization
method for an approximate evaluation of exposure of the concrete to cyclic humidity-
temperature conditions with a simultaneous short-time action of sulfate attack.
The procedure of the crystallization test used complies with ASTM C88
(AASHTO Tl04), except for specimens that had a regular cubic shape. Specimen
dimensions were 50x50x50 mm, corresponding to the maximum size of the aggregates
used for the standard test. The specimens were sawed out from the lOOxlOOxlOO mm
concrete cubes. Twelve specimens were prepared from each sample of concrete, and 5
cycles were used. The quality evaluation of the concrete specimens and forming the
visible cracks was performed after each cycle of test.
The dependence of the strength of the tested and the corresponding reference
specimens on the number of wet-dry cycles is presented in Fig.3. The indicated essential
increase in the compressive strength of the reference specimens is due to the pozzolanic
effect of the fly ash.
The test of specimens which were exposed to wet-dry cycles influence showed no
loss of mass or formation of cracks. The absence of cracks was confirmed by application
of the ultrasonic pulses method. Moreover, findings point to an increase of the strength
of the specimens resulted from wet-dry treatment during the test. Thus, the tested
lightweight concrete containing fly ash can be accepted as a stable material under the
conditions of the Mediterranean climate.·
The effect of fly ash content on porosity and water permeabilitv of lightweight concrete
Thus, the increase of the volume of cement-fly ash paste and the decrease of the
volume of pores and capillaries in the cement-fly ash paste matrix (i.e. the increase of
density) are the main factors affected the properties of the studied lightweight concrete.
The following improvements in the properties of concrete were obtained:
(a) an increase of 4.5% and 11.5% of the density and increase of 35% and 90%
of the strength when FN(C+FA) = 0.33 and 0.5, accordingly;
(b) a reduction in the capillary absorption (Fig.1) of 16-17% and 33-36% when
FN(C+FA) = 0.33 and 0.5, accordingly (duration of test was 4 and 2 h).
The effect of fly ash on the soundness of lightweight concrete under cyclic crystallization
and wet-dry tests
Findings of a wet-dry test to determine the soundness of the specimens with the
cement-fly ash content FA/(C+FA) = 0.5, indicated that lightweight concrete containing
bottom ash and fly ash passed 50 wet-dry cycles without a reduction in strength and
change in shape. This test corresponds approximately to 50 annual environment cycles of
the Mediterranean climate.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Component Percentage
Si02 44.00
AbOJ 33.27
Fe203 4.30
CaO 10.32
MgO 2.37
Ti02 1.46
K20 0.52
Na20 0.26
so3 2.98
P20s 2.27
RS 0.15
RF 0.06
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
9.50 94.7
4.75 84.2
2.36 73.0
1.18 61.5
0.60 43.3
0.30 24.0
0.15 10.7
1 2 3
3
Cement, C, kg/m 335 335 335
Fly ash, FA, kglm3 165 335
Bottom ash, BA, 590 480 360
kg!m3 435 435 435
UCS,kg!m3 360 345 330
Water, 11m3
Volume of cement or
cement-fir ash 370 450 530
paste, lim
I 2 3
3
Density, kg/m 1415 1480 1575
Strength, MPa 12 16 23
Volume of pores and
capillaries in the
concrete,% 31 28 22
Volume of pores and
capillaries in cement
or cement/fly ash
paste,% 56 45 30
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
spalling was observed was observed
3 Development of No damage No damage
spelling was observed was observed
4 Beginning of No damage No damage
surface was observed was observed
destruction
5 Complete Cracks were Cracks were
destruction of observed in observed in
the specimens most of the single
surface specimens specimens
....,=
:8c.
Q
1.5
1.4
,Q
Ill 1.3
.....
......
~~~ a
1.2
au
t;'~ 1.1
...
u
Q
1
0.9
..
""'
~
-== 0.8
0.' 0.1 0.2 0.3 0.4 0.5
FA/(FA+C)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
..
~
34
::;: 32
.IIi
li l)
=
~ 28
~
.. 16
"'
=
C>
u 1A
zz
0 10 ll 30 41 Sl
Number of wet-dry cycles
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SP-234-44
Mr. Ilia Goldfarb, MEd is a Computer System Officer with NRC Institute for Information
Technology. His research interests are in the area of Multimedia and User Interface
Design. He was an interface designer and a Web Developer for the CANMET concrete
durability multimedia database and the U.S. Army Corps of Engineers Web based
information system.
INTRODUCTION
The need to reduce the environmental impact of concrete was recognized in a recent
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
report of the U.S. Strategic Development Council. According to this report, " ... concrete
technologists are faced with the challenge of leading future development in a way that
protects environmental quality while projecting concrete as a construction material of
choice" (Plenge, 2001 ). To achieve this goal, government research laboratories and
construction industry professional associations have a responsibility to effectively
disseminate the results of scientific studies on durability of environmentally friendly and
sustainable concrete, including lightweight concrete, to concrete practitioners.
Research on sustainable concrete is being done in a large number of countries and the
results are largely positive, but what seems to be lacking is the dissemination of this
knowledge to a broader audience (Bremner, 2001). Fortunately, the Canadian federal
government, and, in particular, Natural Resources Canada (NRCan) is starting to realize
the value of knowledge and information on sustainable development and is committed to
providing Canadians with the required knowledge: "Given the importance of natural
resources in our everyday decisions, NRCan must be able to create, share and
communicate the knowledge and information Canadians require" (NRCan, 2005).
Slowly, but steadily, it becomes widely recognized, within the Civil Engineering
community, that there is a growing need for an easily accessible and credible source of
knowledge on environmentally friendly concrete.
The research and scientific community is also in the process of moving away from the
old "information spread" model for dissemination of scientific information (Klein and
Gwaltney, 1991), where knowledge is normally channelled through refereed academic
journals and conference proceedings following the traditional Garvey-Griffith model of a
scientific communication system (Garvey and Griffitth, 1972). Researchers are becoming
increasingly involved in publishing their articles in online-refereed journals that provide
free or low fee access to scientific information (Case, 2001).
However, scientists have not yet fully embraced the power of the Internet. The Web was
invented so that researchers would have a network to collaborate, exchange documents,
discuss them, coordinate research, and create new knowledge. Yet, the use of the Internet
by scientists is still mostly at the stage of electronic publishing and not at the level of true
collaborative work and communication. There are some rare examples of collaborative
Internet-based research environments for knowledge exchange in the strategic research
areas such as nuclear and fusion energy research (Virtual Laboratory for Technology,
2005; Virtual Laboratory for Biokinetic and Dosimetric Research, 2005) but the majority
of scientific information is still posted, mostly in the "information spread" mode, as in the
case of online publications.
To evaluate the state of information dissemination in the area of lightweight concrete, the
authors of this paper thoroughly researched online resources on lightweight concrete and
found only few available. The most comprehensive source of information on lightweight
concrete and, particularly, on structural lightweight concrete found was a web site of the
Expanded Shale, Clay & Slate Institute - ESCSI (Expanded Shale, Clay & Slate Institute,
2005). The ESCSI site provides detailed information on structural lightweight concrete;
lightweight concrete aggregates, national and international companies-members, and has
links to some comprehensive state-of-the-art reports and publications on the long-term
performance of lightweight concrete. (Holm and Bremner, 2000; Holm and Ries, 2001).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
U.S. Anny Corps of Engineers Engineering Research and Development Centre (ERDC,
2005) offers free online access to an excellent Digital Archive of full text ERDC
publications and reports. This is a valuable source of infonnation for civil engineering
materials and concrete. The U.S. Anny Corps of Engineers Web site maintains another
valuable resource on concrete, and in particular, on lightweight concrete - a complete
Web based infonnation system on the concrete durability research at Treat Island
(Kondratova et al., 2002).
We found that even the most comprehensive of the· above-mentioned online resources on
lightweight concrete lack the depth and breadth of coverage of the topic. For a person
seeking infonnation on lightweight concrete, be it a homeowner, a concrete researchers,
or a representative from the industry, it would be beneficial to have a "one stop"
complete infonnation source on production, properties, use, and environmental
advantages of lightweight concrete - a Lightweight Concrete Portal. This Lightweight
Concrete Portal should contain repository of scientific reports, research papers as well as
raw testing data for lightweight and nonnal weight concrete. To demonstrate the
advantages of using raw digital experimental data for research and decision-making, the
success story of a computer database and an online resource for the long-tenn concrete
testing at Treat Island, Maine will be discussed.
The Treat Island natural weathering exposure site dates back to 1937 and is the most
comprehensive long-tenn concrete exposure site in the world. Today, approximately 40
·test programs are active at Treat Island. Some of the variables investigated include
lightweight aggregates made out of clay, shale and industrial by-products, supplementary
cementing materials, and blended cements. Treat Island programs are administered by the
United States Anny Corps of Engineers, Waterways Experimental Station. Sponsors of
the programs at Treat Island include the U.S. Bureau of Reclamation, the U.S. Anny Corps
of Engineers, the Canadian Centre for Mineral· and Energy Resources (CANMET), the
Construction Productivity Advancement Research program (CPAR), and private industry,
At the site, the test prisms are positioned on a rack at mid-tide level, so that they are
exposed alternatively to a marine atmosphere, and immersion in seawater twice daily.
The alternating cycles of immersion and exposure to air provide over 100 cycles of
freezing and thawing per year. Experience, acquired at the site over the past six decades,
has shown that concrete that successfully resist these exposure conditions will normally
provide good long-term performance in marine facilities.
All concrete specimens at the Treat Island site are inspected yearly. Inspection is carried
out by the Materials Group, Department of Civil Engineering, University of New
Brunswick (UNB) in Canada. Inspection includes photographing, visual examination,
visual rating and non-destructive testing (NDT) measurements for each concrete
specimen. NDT measurements include measuring the speed of sound through the
concrete and determination of the transverse resonant frequency of the concrete prisms -
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
a process that estimates the change in the relative dynamic modulus of elasticity of
concrete during exposure, thus giving a measure of concrete deterioration.
Until 1995, the findings of various research programs at Treat Island were not easily
accessible to the engineering community and industry practitioners. To present these
findings, visitations were scheduled on even numbered years. A few scientists and
engineers, invited by the Corps of Engineers, attended this event. The dissemination of
results of the studies completed in previous years was confined to annual research reports
and a relatively limited number of technical papers that needed to be brought to the
attention of practicing professionals and of the construction industry.
CANMET database
The first step in developing a new knowledge dissemination mechanism was taken in
1994 with the development of a multimedia database accommodating the results of more
than two decades of research for CANMET's concrete durability studies (Kondratova et
al., 1998). The researchers of the Materials Group, UNB, developed the database. To
provide for better data visualization capabilities, research data were presented as a
multimedia computer database. Multimedia data in the CANMET database comprises of
static media, like text and historical photographs of specimens, and of dynamically linked
Excel charts graphically representing annual results of non-destructive testing (NOT) and
visual evaluations. The CANMET database was developed as a stand-alone CD-ROM
database and is updated annually following the summer testing and inspection.
The database design allows side-by side comparison of historical photographs and testing
results for different concrete mixtures and supports decision-making on the choice of
environmentally friendly and durable concrete. Unfortunately, the database is not
In 1999 the U.S. Army Corps of Engineers were searching for better ways to disseminate
the results of Treat Island research on concrete durability and wanted to exploit
Information Technology as a knowledge dissemination tool. Owing largely to the
international success of the CANMET database the U.S. Army Corps of Engineers invited
UNB's Materials Group to submit a work proposal for the development of a similar
multimedia computer database for all Treat Island site's concrete durability programs.
Later, however, in consultation with the U.S. Army Corps of Engineers staff, it was
decided that a Web-based information system would be created instead. The decision was
based on the understanding that an online version would provide worldwide access to
research findings on concrete durability at the Treat Island exposure site and would also
be easier to use and maintain.
Currently the Treat Island Web-based durability information system (USACE, 200la)
contains information on 37 long-term research projects, including studies on the use of
supplementary cementing materials and lightweight aggregates for marine concrete
(CANMET); high strength and high performance concrete, supplementary cementing
materials to lower cement consumption, and high-performance repair materials for
concrete structures (U.S. Army Corps of Engineers); concrete corrosion inhibitors and
epoxy coatings (University of New Brunswick and Master Builders); and high
performance semi-lightweight concrete (Exxon Mobil).
The design of the U.S. Army Corps of Engineers site allows easy information navigation
through links to the individual research programs. The user can find information using a
keywords option; also experienced users can visually locate specific specimens using
maps of the exposure rack and the beach area. Each research program is presented using a
general description and a photograph, with a link to a data page, which includes
thumbnail photographs of the individual specimens involved in the program, which, in
tum, leads to the third and forth design layers, which provide information on the concrete
mixture design, non-destructive testing data, and a historical photographic record of test
specimens.
The authors will now demonstrate how CANMET and U.S. Army Corps of Engineers
information systems can be used to retrieve information on the long-term performance of
lightweight concrete in a severe marine environment. Data for all specimens in
CANMET database is presented as tables with all the essential information about the
concrete specimens, including the phase of the research project, year when specimens
were placed at the exposure site, age of the specimen when the photographs were taken
(5, 10, 15, 20 years etc.) and details of the mixture proportions. Using the CANMET
database, it is easy to pick particular specimens right from the table. This allows the
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The table query is displayed in a report form. The report can be viewed on a computer
screen and printed. In this paper we captured screen shots with specimen reports (Figures
1 - 4). Unfortunately, the CANMET database currently does not contain all the details on
specimens and, therefore, when used for performance analysis, should be supplemented
by data from the U.S. Army Corps of Engineers Treat Island online database that has more
details on concrete mixture design and compressive strength testing.
For a side-by-side comparison oflightweight and normal weight concrete of the same age
using the CANMET database, we chose specimens from two CANMET phases, Phase II
and Phase III. Phase II concrete prisms contain air-entrained concrete incorporating
different percentage of fly ash, and pelletized blast-furnace slag as cement replacement.
Phase II specimens were installed at Treat Island in 1979. Phase III specimens contain
air-entrained semi-lightweight concrete incorporating different percentage of pelletized
blast-furnace slag as a cement replacement. Phase III was installed in 1980. For Phase III
concrete, expanded shale was used as the coarse aggregate. In our example, in order to
account for the possible effects of cement replacement on concrete performance, for
comparison purposes we are using control specimens of Phase II series D (normal
weight) and Phase III series I (semi-lightweight) that contain 0% of cement replacement.
Normal weight concrete specimens Dl, D5, D9 have water-to-cement ratio of 0.40 and
cement content of 395 kg/m 3 . Semi-lightweight concrete specimens 11, IS and I9 have
cement content of 480 kg/m 3 • Based on the US Army Web-based information system
data, the average 28 days compressive strength for control concrete prisms of series D is
33 MPa. 28 day compressive strength for control lightweight concrete prisms of series I
is 35.7 MPa.
In addition, using the data in the CANMET database, we can compare not only historical
photographs of specimens, but also can retrieve and compare the results of nondestructive
testing for both types of concrete. Based on more than twenty years of nondestructive
testing results for concrete prisms made of normal weight and semi-lightweight concrete
(Figures 2 and 3), it is clear that for semi-lightweight concrete the relative modulus of
elasticity based on resonant frequency results is somewhat lower than for normal weight
concrete of comparable quality. However, the relative modulus of elasticity for
lightweight concrete based on pulse velocity testing data was steadily increasing for up to
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Similar conclusions about the excellent performance of structural lightweight concrete
could also be drawn from the test results posted on the Natural Weathering Exposure
Station Treat Island US Army Corps of Engineers Web site. For example, we can easily
examine the results of the Exxon Mobil program (US ACE, 2001 b). This program
contains Hibernia specified density concrete reference blocks and cylinders. Four large
concrete blocks (600 x 900 x 900 mm) were cast from the Hibernia production concrete
used for the splash zone of the outer perimeter of the structure and were installed at Treat
Island in October 1996.
These two examples of concrete evaluation based on the data contained in the CANMET
database and the US Army Corps of Engineers online information system demonstrate
the ease of information retrieval, use, and performance analysis using these IT tools.
However, the information contained in these two information systems is by no means
complete and, for the purpose of decision-making, could be and, we believe, should be
supplemented by other available experimental data from external research programs.
CONCLUSION
According to Mehta, " ... the greatest challenge that the concrete industry faces during the
21st century is to achieve a sustainable pattern of growth" (Mehta, 2001 ). The task is
tremendous, but we can accomplish this by making an industry-wide paradigm shift to
the culture of conservation of energy and materials. This paradigm shift will be supported
by another shift related to the use of information technology to utilize knowledge on
sustainable construction materials, including lightweight aggregates and lightweight
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 729
concrete. Knowledge utilization in this case includes technology transfer, information
dissemination and utilization, research utilization, innovation, and organizational change
(Backer, 1993) and, according to Paisley's research (Paisley, 1993): "Digital technology
brings the most significant new communication capabilities to knowledge utilization ... "
Internet is here to stay. After the dot. com bubble burst in 2000, the current usage of the
Internet in Europe is up by 161% and the number of Internet users in US and Canada in
2005 increased by about 107% as compared to the year 2000 (Internet World Stats,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
2005). The construction industry is ready to accept the Internet as the new knowledge
utilization mechanism. A survey of Canadian AEC (Architectural, Engineering and
Construction) Industry conducted in 2000 found that that 86% of the architectural firms,
97% of the engineering firms, and 83% of the contractors surveyed are all connected to
the Internet (Rivard, 2000). More recent survey of construction industry conducted in
Atlantic Canada (Desai, 2005) found that all companies that responded to a survey have
Internet connection in the office. Thus, the Internet has the capability to become an
accepted communication medium for efficient technology transfer to the construction
industry.
One of the proposed steps in utilizing the power of the Internet to promote lightweight
concrete is to set up an International Web Portal on lightweight concrete. This portal
could be based on an existing industry association Web site, for example the ESCSI site
that already incorporates some essential attributes of a vertical industry Web Portal such
as links to scientific information and to an industry forum. According to research studies
(Newell et a!., 1998), professional associations have great potential to play a mediating
role in the diffusion of knowledge. They provide a forum for the creation of inter-
organizational networks that, in tum, create the necessary channels for diffusion of
information, knowledge and ideas that enable companies and organizations to bring
technological innovation. However, individual members of the professional association
can benefit from these opportunities only if they attend workshops, branch meetings, etc,
which can be costly and time consuming.
Internet technology is removing these barriers. With a development of a full scale Portal,
ESCSI, as an international industry association, could become a worldwide source of
credible information and expertise in the area of lightweight concrete and will be in a
position to directly influence innovation and technology transfer to the industry. A
comprehensive Lightweight Concrete Portal, containing a repository of scientific reports,
testing data and, possibly, information from the CANMET database and links to the US
Army Corps of Engineers Treat Island information system, could also provide unique
collaborative work and e-leaming opportunities (Kondratova and Goldfarb, 2002).
The model of scientific communication in this case will come closer to the modem
scientific communication model (Hurd, 2000). The most important feature in this new
communication model, discounting modernized technology that supports traditional
publishing functions and e-publishing process, is to include repositories of raw data on
Internet servers, for use by researchers that wish to build on findings of others, or to
create new knowledge by combining their own data with the data of others. The
evaluation of the performance of lightweight concrete based on CANMET database and
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
730 Kondratova and Goldfarb
U.S. Army Corps of Engineers online system is an example of the possible use of this
technology.
The proposed Lightweight Concrete Portal can be based on this model. Some information
could be provided free of charge and some could be for members and subscribers only,
similar to the model adapted by the highly successful CorrosionSource.com Web Portal
that contains a knowledgebase of corrosion protection resources and a popular Discussion
Forum for corrosion community (Corrosion Source, 2005). The proposed model of an
online collaborative environment would also provide new opportunities for distance
education of graduate students and young researchers (NRC, 2001), and create a valuable
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
ACKNOWLEDGMENTS
The authors would like to acknowledge the support provided for the information systems
development projects by the Canadian Federal Department of Supply and Services,
CANMET, the US Army Corps of Engineers and the National Research Council Canada.
The authors also would like to acknowledge valuable advice and guidance of the
Honorary Research Professor Theodore W. Bremner in the design and development of
the CANMET and US Army Corps of Engineers concrete durability information systems.
REFERENCES
Case, M. M. (2001). Public access to scientific information: Are 22,700 scientists wrong?
College and Research Libraries News, 62 (7), 706-709.
Edquist, C. and Riddell, W. C. (2000). Role of Knowledge and Innovation for Economic
Growth and Employment in the Information and Communication Technology (ICT) Era,
Transition to the Knowledge Society. Eds. Ruberson, K. & Schuetze, H. G. UBC Institute
for European Studies, University of British Columbia, Vancouver.
ERDC (2005). U.S. Army Corps of Engineers Engineer Research and Development
Centre. http://www.erdc.usace.army.mil/.
Farny, J. (1996). Treat Island, Maine- the Army Corps' Outdoor Durability Test Facility,
Concrete Technology Today, PCA, Vol. 17, No.3, pp. l-3.
Garvey, W.D. and Griffitth, B.C. (1972). Communication and Information Processing
within Scientific Disciplines: Empirical Findings for Psychology. Information Storage
and Retrieval, 8, pp. 123-126.
Holm, T.A. and Ries, J.P. (2001). Benefits of Lightweight HPC, Expanded Shale Clay
and Slate Institute, HPC Bridge Views, Issue No. 17, September/October 2001,
http://www.escsi.org/Structur.htm.
Hurd, J.M. (2000). The Transformation of Scientific Communication: A Model for 2020,
Journal of the American Society for Information Science, 51 (14), pp. 1279-1283.
Industry Canada, Manufacturing Industries Branch. (2002). Input to the AMG Working
Group on the Competitiveness Analysis of the Impact of GHG Abatement Measures on
the Canadian Manufacturers of Cement,
http://nccp.ca/NCCP/national stakeholders/pdf/competitiveness bkg studies/FINALcem
ent.pdf.
Klein, S.S. & Gwaltney, M.K. (1991). Charting the Education Dissemination System:
Where We are and Where We Go from Here. Knowledge: Creation, Diffusion,
Utilization, 12 (3): 241-265.
Kondratova, I., Bremner, T. W., and Goldfarb, I. (2002). IT Providing a Path From
Research to Practice- Promoting Use of Environmentally Friendly Cement and Concrete
Malhotra, V.M. and Bremner, T.W. (1996). Performance of Concrete at Treat Island,
U.S.A; CANMET Investigations, ACI Special Publications, SP-163, pp. 1-52.
NRC. (2001). National Research Council: Issues for Science and Engineering Research
in the Digital Age, Office of Special Projects Policy and Global Affairs. National
Academy Press, Washington, DC.
Plenge, W.H. (2001). Introducing Vision 2030: Our Industry's 30-year Map to the Future,
Concrete International, March 2001, Vol. 23, No.3, pp. 25-34.
USACE. (200la). U.S. Army Corps of Engineers Natural Weathering Exposure Station
Treat Island, Engineer Research and Development Center Geotechnical and Structures
Laboratory, http://www.wes.army.mil/SLITREAT ISL!index.html.
USACE. (2001b). U.S. Army Corps ofEngineers Exxon Mobil Program, Natural
Weathering Exposure Station Treat Island, Engineer Research and Development Center
Geotechnical and Structures Laboratory,
http://www. wes.army.mil/SLITREAT ISL!Programs/ExonMobil8.html.
Virtual Laboratory for Biokinetic and Dosimetric Research. Oak Ridge National
Laboratory. (2005). http://homer.ornl.govNLab/.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
. . ., . .
~FA
't'oSF
'
!.
'
0
·.·::~:··:.·
0
I '""CEMENT TYPE
AIRCONTENT,~~
OOf
I ''"'
~ ·--~-~·
I
.....
PHASE DESCSUPTIOH:
D•lttmtfrtllfk>ttolttl•~-of~r-ffftf11inwleo~t•
IIICOrpotM111191ow-clllc/umfl)'llahlllfdp-'t.tl:ftd
blaM'r.Niflceal.;
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REPORT TABLE C
ID: 511 PHASE: Ill (IIIlO) SERIES: 11)1 PHASE DESCSUPTION:
~sLAO r- o ~~ wm AmlkQtma! O.-..mm#tlonoftllfpwf'wm-oflllf'.""trllkrH-1·
ligltf~teo.renrt.lt!eorpordtttP•IhttizH bJHt.fUI'II-.Jifl
't'ofA. {"~()~ CEMENTTYI"E ~
\'oSF L2_.j AIII:CONrENT,%~
Fig. 3-- NOT testing report for series I semi-lightweight concrete specimens.
Fig. 4·· NOT testing report for series H semi-lightweight concrete specimens.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Synopsis: For more than fifty years, a large number of bridge decks and parking
structures have been constructed using lightweight concrete. In most cases, the
lightweight concrete has been made using a lightweight coarse aggregate and a
normal weight sand for the fine aggregate. Some of the natural sands used have been
shown to be at least partially susceptible to alkali-silica reaction (ASR). Many of the
bridge decks have been in service for at least 30 years and show no outward signs of
deterioration due to ASR. A test program was undertaken to determine how concrete
made with lightweight coarse aggregate and reactive sand does not show the
effects of ASR. Properties measured include expansion, indirect tensile strength and
compressive strength. Test results indicate no significant difference in performance
between concrete mixes with lightweight coarse aggregate combined with either
reactive, non reactive, or lightweight fine aggregate.
Thomas A. Holm, P.E., FACI is the Director of Engineering of the Expanded Shale, Clay,
and Slate Institute, Richmond Virginia. He is also past-chairman of ACI Committees 213 .
and 122. He has published more than 40 papers on concrete, masonry, thermo-physical and
geotechnical issues, and is the co-author ofthe Corps of Engineers' State-of-the-Art Report
on High-Strength, High-Durability, Low-Density Concrete for Applications in Severe
Environments.
INTRODUCTION
The use of structural lightweight concrete (SLC) in bridges and other exposed
structures has grown significantly in recent years in both number and type of application.
Especially noteworthy is the widespread usage in the rehabilitation of existing bridges to
accomplish improved functionality, i.e. longer spans, more or wider lanes, upgraded load
capacity, and reduced seismic inertial forces 1• 2• 3 •
Currently, there has been increased attention directed at the reduced service
perfonnance caused by the damaging effects of expansion in concrete containing reactive
normal weight aggregates. SLC has performed entirely satisfactorily with respect to volume
stability in the presence of cements with high alkali contents, however concern has been
expressed when potentially reactive normal weight aggregates were used in combination
with structural lightweight aggregates. This program was designed to provide useful
information that will address this concern.
Alkali-silica reaction (ASR) occurs between the alkali hydroxides produced during
cement hydration and silica minerals found in certain aggregates. The alkalis slowly break
down the bonds that hold silicates together. These fragmented molecules then bind with
alkali-hydroxides to form a gel-like substance. This gel absorbs water and swells causing a
disruptive force inside the concrete. The effects of ASR result in expansion, cracking, and
subsequent strength loss. Concrete suffering the effects of ASR can show little or no outward
signs of distress when subjected to compressive stresses. Cores taken from ASR affected
structures and tested in compression usually show no significant loss of strength because the
cracks can close and transfer the load across the crack. However, in tension, the cracks
participate in an active way in the failure process. The cracks open up under tensile stress
and failure occurs at the crack. This results in a significant reduction in tensile strength as
ASR progresses. The cracking from ASR also lets deleterious elements like chloride ions
into the concrete to attack the reinforcing steel.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
TESTING PROGRAM
Mixture constituents
A Type I cement with an alkali content of0.89 percent was used for all the concrete
mixtures in this program. Care was taken to make sure the cement was fresh and free flowing
and was the same for all of the mixtures. Despite the well-known, proven performance of
high quality pozzolans and GGBF slag cements to mitigate or eliminate expansion due to
ASR, they were not used in this action in order to develop a scenario that identified the
behaviour of structural lightweight concretes (SLC) containing a potentially reactive natural
sand.
Water used for all the mixtures was ordinary tap water from the City ofFredericton
water supply that had been used extensively in the past and meets all the CSA requirements
for mix water.
Aggregates for this test program come from three sources. Coarse aggregate used
was Solite expanded slate, lightweight aggregate manufactured in Aquadale, NC. This
aggregate is produced under controlled conditions in a rotary kiln at 1200 °C. The particles
expand in the kiln and result in an aggregate with uniformly sized vesicles throughout. The
chemical composition ofthe expanded slate is mostly Silica (69%), and Aluminum (12%).
This aggregate has been used extensively for more than fifty years in the Eastern United
States and has remained unchanged over that time. This structural lightweight aggregate has
been used in bridge decks and parking garages exposed to harsh environmental conditions
and has performed very satisfactorily. Tests conducted on this aggregate in accordance with
ASTM C 1260, "Standard Test Method for the Potential Reactivity of Aggregate (Mortar Bar
Method)", showed an expansion of 0.07% which suggests innocuous behaviour. Fine
aggregates used included normal weight sand obtained from a ready-mix concrete plant that
is widely used commercially and has as its source, a mid-atlantic river deposit. Tests
conducted in accordance with ASTM C 1260 showed an expansion of 0.247%, which
suggests potential reactivity. This sand is referred to in the test program as PRS. A local fine
Mixture proportions
The concrete mixtures for this testing program were based on the CSA A23 .2-14A
M95 standard for Potential Expansivity of Aggregates which is similar to ASTM C 1293,
"Standard Test Method for Concrete Aggregates by Determination of Length Change of
Concrete Due to Alkali-Silica Reaction." The test procedures used accelerate the alkali-
silica reaction by increasing the alkali content from the normal 0.89% usually found in Type
I cement used to 1.25 percent by mass of cement. This increase is accomplished by adding
reagent grade NaOH to the mixture.
The two mixture proportions used in this test program were supplied by So lite Corp.
and are shown in Table 1. These proportions have been widely used commercially for more
than 50 bridge decks and a large number of parking structures. The intent of these mixtures
was to measure the expansion due to the effect of the ASR process on a typical commercial
lightweight concrete containing potentially reactive sand and to compare the results with a
concrete made with non-reactive sand. Both mixtures were air-entrained.
Results of tests on the fresh concrete are shown in Table 1. Air contents for both
mixtures were in the acceptable range as required for freeze-thaw resistance by specification
of mid-atlantic departments of transportations. Other than the different natural sands there is
no significant difference between the two mixtures ..
Testing procedures
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Tensile strength tests--Baseline 28-day tensile strengths were determined using the
procedures in ASTM C 496, "Test Method for Splitting Tensile Strength of Cylindrical
Concrete Specimens." These splitting tests at 28 days resulted in tensile strengths of2.99
MPa for the PRS concrete mixture and 3.17 MPa for the NRS concrete mixture. For the rest
RESULTS
This testing program was conducted in accordance with Canadian Standards which
were developed before and in principle, adopted into ASTM C 1293. The Canadian
Standards Association sets a limit of 0.04% expansion at one year to determine if an
aggregate is potentially reactive. The results of the expansion measurements for concrete
mixtures PRS (potentially reactive sand) and NRS (non-reactive sand) are shown in Figure 1.
as explained earlier, these mixtures have the same lightweight coarse aggregate. Expansion
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Figure 1 shows that the expansions of both mixtures were well below the usually
accepted limit of0.04% at one year and that there is no appreciable difference between the
two mixtures. Figure 1 also shows the expansion of both mixtures has levelled off. For
comparison purposes, the expansions of concretes containing reactive and non-reactive
normal weight coarse aggregates combined with the same non-reactive sand used in this
program are also shown. These results were reported in references 4 and 5.
Tensile and compressive strengths were measured up to 210 days. The results in
Figure 2 show there is little difference in the compressive strengths achieved, and also no
regression. In retrospect, it would have been desired to have compressive and tensile strength
results at later ages for this program, but the batches were limited in size with only enough
specimens to test up to 210 days.
To normalize the results of these two concretes with differing strength making
behaviour, results are shown on the basis of the ratio of tensile to compressive strength.
Figure 3 shows a small regression of this ratio in both the PRS and the NRS concretes up to
approximately 80 days. After this point, there is no regression of the ratio for the NRS
mixture but the ratio continues to decrease for the PRS concrete.
Expansion measurements
Compressive strength
Compressive strength of concrete may decrease after ASR occurs. This is because
the swelling of the reaction product may cause the aggregate to fracture. This disrupts the
aggregate matrix bonding and causes the concrete to deteriorate. The more severe the
reaction, the more severe the drop in compressive strength. As reported in reference 4,
replacing non-reactive natural sand with lightweight aggregate fines significantly reduces the
expansion and strength loss in concretes containing a highly reactive normal weight coarse
aggregate. As shown in Figure 1, SLC mixtures in this program containing a potentially
reactive natural sand had expansion not significantly different from parallel mixtures
containing normal weight non-reactive sand. While the early strength making properties of
the concrete used in the NRS mixture are superior to that of the PRS concrete, there does not
appear to be any strength regression up to 210 days.
Tensile strength
As shown by the authors in earlier works, the tensile strength of concrete suffering
from distress caused by ASR will be more indicative of the extent of damage than
compressive strength. This is because under a compressive load, cracks caused by ASR can
close and the load is effectively transferred across the crack. Thus the load carrying capacity
of the structure is not diminished and the extent of ASR may not be evident until it is too
late. Higher tensile strength will reduce the amount of cracking which occurs and allow the
concrete to resist the expansive forces of ASR. This in tum helps slow down the ingress of
substances like chloride ions, which can corrode the reinforcing steel.
The second mechanism is based on the fact that the pyro-processed surface of
expanded shale, clay and slate (ESCS) aggregates exhibit a certain degree of pozzolanic
behavior. Because of the mildly pozzolanic surface of the pyro-processed alumina-silicates
in ESCS aggregates, and considering the availability of a small amount of fines supplied
with a coarse aggregate, these sites of silica may react with the alkalis which would normally
react with the silicates in the reactive aggregates. By having this reaction take place at small
sites, the amount of reaction product formed may be less significant and therefore disruptive
forces may not build up. Further testing needs to be carried out to determine the exact
mechanism operative. Field investigations of a number of mature bridges containing these
combinations of aggregates reveal entirely satisfactory performance.
CONCLUSIONS
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
1. The major conclusion of this research program is that with the particular
potentially reactive natural sand used in this test program, there was
limited detrimental effects of ASR when used in combination with a
lightweight coarse aggregate. With this particular combination of
materials, there were small concrete prism expansions (ASTM C1293)
well below the generally accepted limit of0.04% at one year. No decrease
in compressive strength, no cracking, and a limited decrease in tensile
strength were observed.
2. Based on these concrete prism test results, it appears possible that using
concretes containing lightweight coarse aggregate may allow the use of
some natural sands which expand excessively when tested by the
RECOMMENDATIONS
REFERENCES
1. Brown, W.R., Larsen, T.J., and Holm, T.A., Long Term Service Performance of
Lightweight Concrete Bridge Structures. International Symposium on Structural
Lightweight Aggregate Concrete, Sandefjord, Norway, June 1995.
2. Holm, T.A., and Ries, J.P., Specified Density Concrete- A Transition for the Concrete
Industry. Second International Symposium on Structural Lightweight Aggregate
Concrete, Kristiansand, Norway, June 2000.
3. Stolldorf, D.W., and Holm, T.A., Bridge Rehabilitation Permits Higher Loads. Materials
for the New Millenium, 41h Materials Engineering Conference ASCE, Washington,
D.C., USA, November 1996.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
4. Bremner, T.W., Boyd, A.J., Holm, T.A., and Boyd, S.R., Indirect Tensile Testing to
Evaluate the Effect of Alkali-Aggregate Reaction in Concrete. Proceeding of Structural
Engineers World Conference 98, San Francisco, CA, USA, July 1998.
5. Boyd, S.R., Bremner, T.W., and Holm, T.A., Addition of Lightweight Aggregate
Reduces Expansion in Concrete Containing a Highly Reactive Normal Weight
Aggregate. 11th International Conference on Alkali-Aggregate Reaction, Quebec,
Canada, June 2000.
f A1 (REF 4)
I
,' •• ~ ·- ~ ~ "- :r--~P"R"s ••••••••• --
.02 I ; .-..,.. !:..___
:~ -t_NRS
Fig. 1 Expansion of concrete prisms with age according to procedures of ASTM C1293.
~ 40
[
~ 30
~
~ 20
~
8 10
J:
1-
(!)
z
w
0:
1-
en
15 -
w
>
iii
en
w
0:
0.
::;; 10
0
CJ
0
1-
~
iii
zw 5 -
1-
u.
0
..
0
>=
0:
DAYS
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
INTRODUCTION
Cement plants today are all based on the "dry process" with raw meal (moisture of
only 0.2%) feeding the kiln where about 220 kg of coal are fired to make one tonne of
cement, instead of the more expensive 125 litres of oil per tonne of cement used before
the 1980s in the "wet process" where a sort of slurry with 40-50% of water was used as
raw meal. As a matter of fact, a certain increase in the alkali content has been recorded
after the conversion to coal of most oil-fired plants. This means that many silica-based
aggregates, which were safely used before the 1980s, have become today more prone to
react because of the higher content of Na 2 0 and K20 in portland cement. Due to the
alkali increase in portland cement, the risk of pop-out and cracking in industrial concrete
floors was significantly increased during the last decades. In order to mitigate this effect
and to avoid such a risk, blended cements have been used by replacing portland cement
by natural pozzolan, fly ash and/or ground granulated blast furnace slag (GGBFS). This
change, from pure portland cement to pozzolanic or slag-cement, did not occur in the
cement mixtures used to manufacture dry shake hardeners for the reason explained later.
Dry shake hardeners are incorporated on the concrete surface of industrial floors for
additional abrasion resistance and impact protection. They are ready to use mixtures of 0-
4 mm carefully graded silica-quartz aggregates, workability agents, portland cement and
mineral oxide pigments, if any. These products are ready for application without further
mixing. When applied as a dry shake hardener on a still plastic concrete floor, they
become an integral part of the floor, producing very attractive colored surfaces with
improved durability and wear resistance characteristics. Applications of shake hardeners
are ideally suited for industrial plants, shopping centers, hospitals, schools, swimming
pools, parking garages, and service areas exposed to heavy duty traffic. Typically 3 to 20
kg/m2 of dry shake hardener are used for light traffic to extremely heavy duty with steel
wheeled traffic, respectively. Due to the relatively high amount of portland cement in the
dry shake hardener (about 40%) and the alkali content in the portland cement (0.7-0.8%
as Na20 equivalent) the concentration of alkali oxide in the top reach can reach up to 6
mg/mL which corresponds to 6 kg/m 3 ofNa20. This value is 3 to 4 times the threshold
A concrete mixture typically adopted for industrial floors was manufactured by using
320 kg/m 3 of normal portland cement with a relatively low alkali content (Na20 = 0.40%
and K20 = 0.20% corresponding to a Na20 equivalent of 0.53%). Table 1 shows the
mixture composition of the concrete substratum characterized by a slump of 160 mm and
a cube 28-day strength of about 35 MPa. Coarse aggregates used were identified as
potentially reactive due to the presence of amorphous silica embedded in a calcitic rock
(2).
Table 2 shows the composition of different mixtures of the dry shake hardeners; 5
kg/m 2 of dry shake hardeners were applied on the surface of the concrete substratum in
form of 1 m2 slab at about 6 hours after placing. Graded 0-2 mm quartz (60%) was mixed
with about 40% of normal portland cement or blended cements containing 50% of fly ash
or GGBFS to reduce the risk of ASR (3); two different types of slag were used, the
difference being the Blaine specific surface area of 400 or 650m 2/kg. Table 3 shows the
chemical composition of cement, fly ash and GGBFS used in this work. About 0.1% of
naphthalene-based admixture was used to improve the workability of the dry shake
hardeners. Accelerating admixtures in form of calcium formate (0.5%) was used in some
special formulations of dry shake hardeners to reduce the setting time and the time prior
to exposure of the finished surface to the pedestrian traffic.
RESULTS
According to the results shown in Table 4 the time for the exposure to pedestrian
traffic ("opening time") is significantly increased when 50% of the normal portland
cement is replaced by GGBFS and even more in the presence of fly ash. The dry shake
hardener A (100% of normal portland cement as binder) shows an opening time of less
than 12 hours, whereas the dry shake hardener B (50% of fly ash in the binder
composition) shows an opening time of more than 24 hours. In the presence of GGBFS
(50% of the binder) the opening time is 24 hours and even less when the specific surface
area of the slag is increased from 400 to 650m 2/kg. However, all these time- except that
of the hardener with only normal portland cement - are considered too long by the floor
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUSIONS
The replacement of normal portland cement by 50% ofF class fly ash or GGBFS as
binder in the dry shake hardeners completely removes the risk of ASR and that of pop-out
formation in the transition zone between the concrete substratum, where potentially
reactive aggregate are embedded, and the hardener on the top surface very rich in the
alkali content for the high cement factor in the top layer. Moreover, the abrasion
resistance is not changed in the dry shake hardeners with GGBFS replacing 50% of
portland cement.
However, this replacement, which is very advantageous for the durability of the
industrial floor (in terms of both alkali - and abrasion - resistance), causes some troubles
to the floor workers for the excessive extension of the time for exposure to the pedestrian
traffic (from less than 12 hours to about 24 hours). In order to remove this drawback in
the execution of the industrial floors, a finely ground slag (with a specific surface area of
650 m2/kg) and the addition of an accelerating admixture have been used. These
compositional options reduce the time for the exposure to the pedestrian traffic to less
than 12 hours and furtherly improve the abrasion resistance.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
REFERENCES
[1] Fournier, B., Berube, M.A., and Rogers, C.A., "Canadian Standard Association
(CSA), Standard practice to evaluate potential reactivity of aggregates to select
preventive measures against alkali-aggregate reaction in new concrete structures",
Proceedings of the 11th Conference on AAR, Quebec City, pp. 633-642,2001.
[3] Smolczyc, H.G., "Slag Cements and alkali reactive aggregates", Proceedings of the
6th International Symposium on Chemistry of Cements, Moscow, 1974.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Accelerator
* CEM I 52.5R: Na20 eq = 0.71%
** Class F Fly Ash: Na20 eq = 0.61%
*** GGBFS: NazO eq = 0.55%
F No Defect
Control No Defect
Transition
Fig. 1- Schematic picture of ASR at the transition zone between the hardener layer
and the concrete substratum.
LAYER A
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
c
EXPANDED REACTIVE
SILICA: BLACK AREA
CORRESPONDS
TO VOID SPACE
Fig. 3- Pop-out on the concrete slab treated with the A hardener on the top surface.
Synopsis: The purpose of this research work was to make a drying shrinkage-free
concrete (SFC) ,even in non-wet curing conditions. This concrete was produced by the
combined use of:
a) a water-reducing admixture, based on polycarboxylate (PA), in order to reduce both
the mixing water and cement, and increase the amount of aggregate;
b) a special polycarboxylate (PA/SRA) including, in its molecular structure, a
shrinkage-reducing admixtures (SRA) based on polyethylene glycol capable of
reducing the surface tension of liquid water filling the capillary pores;
c) an expansive agent based on a special calcium oxide (CaO) manufactured in a kiln
at relatively high temperatures (about 1000 °C).
Traditional shrinkage-compensating concretes are theoretically based on the
restrained expansion produced by portland-cement products containing either
calcium sulfa-aluminate or free CaO as expansive agent. However, in practice this
effect is cumbersome to achieve because these concretes must be wet-cured, for at
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
least 3-7 days after the final set.
On the other hand, with the concrete described in this paper, drying shrinkage
is completely compensated even in the absence of wet curing. The concrete is
demolded at 3 days and then exposed to air curing. Compressive strength and
restrained expansion of laboratory specimens as well field cured concrete are given.
INTRODUCTION
Concrete durability can be jeopardized by the presence of cracks even when a low
water-cement ratio (w/c) is used and a low capillary porosity of the cement matrix
envelops the aggregates as well the metallic reinforcements. Indeed, through these
interlinked cracks the aggressive agents, in form of gaseous or liquid products (water,
carbon dioxyde, chloride and sulfate ions, etc.), can quickly and easily penetrate the
concrete cover and damage the cement matrix and/or the metallic reinfocements. One of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
the most important causes of concrete craking is related to the restrained drying shrinkage
in environments with relative humidity lower than 90%. A tensile strength (cr1) is induced
by such restrained shrinkage ( E5) which can cause cracks when it becomes higher than the
tensile strength (ft) according to equation [l]:
[ l]
b) reduction in the surface tension (y) of the water filling the capillary pores so that the
capillary pressure (P), caused by the formation of menisci developed in the cement
matrix, is reduced according to the Laplace equation [2]:
where r is the pore radius and e is the wetting angle of water on the cement paste surface:
this reduction can be accomplished by using shrinkage-reducing admixtures (SRA) or
special superplasticizers which include this chemical group in their molecular structure;
c) addition of an expansive agent based either on calcium oxide (CaO) which can be
transformed into Ca(OH)2 under wet-curing conditions for at least 2 days, or calcium
sulfa-aluminate (4Ca0.3Al 20 3 .S03) which can be transformed into ettringite
(3CaO·Alz0 3·3CaS04 ·32H20) under wet-curing conditions for 7 days. Both the
processes occur with volume increase and, when adequately restrained by metallic
reinforcements, can compensate the drying shrinkage by producing a compressive
stress in the concrete and a tensile stress in the metallic reinforcements. To bring
about this effect wet-curing is essential during the expansion period (1-2 days for
calcium oxide and 5-7 days for calcium sulfa-aluminate) according to the equation
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
[3] and [4] respectively :
To make concrete mixture blended portland cement was used (70% of portland
cement and 30% of fly ash according to the European Norm EN 197-1) in combination
with natural sand (0-4 mm) and gravel (4-25 mm). A superplasticizer based on
polycarboxylate (P A) was used as water reducing agent in some concrete admixtures; in
others a special polycarboxylate (PA I SRA) was used in order to reduce both the mixing
water content and the surface tension (y) of the water due to the presence of a SRA
chemical group, based on polyethylene glycol in the molecular structure of the
polycarboxylate as schematically shown in Fig.1. This chemical group is gradually
released by the superplasticizer PA/SRA due to the alkaline environment in the aqueous
solution (1 ).
A commercial expansive agent was used in form of calcium oxide manufactured in a
kiln at 1000 °C with a particle size distribution in the range of 10-125 J..lm (2).
The following five concrete mixtures (Table 1) were manufactured all at the w/c of
0.50 and a slump level in the range of 170-200 mm at 30 minutes after mixing:
the control mix (A) without any chemical admixture contains 430 kg/m 3 of
cement, 215 kg/m 3 of water and an aggregate/cement ratio of 3.88;
due to the presence of the superplasticizer PA (1% by cement weight), the
amount of mixing water and that of the cement factor in the mixture B is
reduced to 154 and 308 kg/m 3 respectively; consequently the aggregate/cement
ratio (a/c) increases to 6.30;
Compressive strength of the five concrete mixtures were measured at 1-28 days on
cubic specimens cured at 20°C and at a R.H. of95%.
The length change of the metallic bar embedded in the prismatic specimens
(80x80x240 mm) was determined, according to the ASTM test method 845-90, with the
original length measured at 6 hours just after demolding the reinforced specimens (Fig.2).
Two different curing conditions were adopted:
according to the the first method, based on ASTM 845-90, the specimens after
demolding at 6 hours were cured under lime -saturated water; then after 3 days
they were exposed to air with R.H. of 50% at 20°C for 6 months;
according to a modified method, the demolded specimens at 6 hours were
protected from drying by a polyethylene sheet up to 3 days in order to simulate
the curing conditions occurring realistically into the formworks; then the
specimens were exposed to the air at 20°C and at a RH of 50% from 3 days up
to 6 months.
RESULTS
Figure 3 shows the compressive cube strength of the five concrete mixtures as a
function of the curing time. No significant difference in the compressive strength of the
five concrete mixtures at a given time can be observed, except at 1 day when there is
slight retardation in the cement hydration caused by the PA/SRA chemical admixture.
Moreover, in the presence of CaO, there is a slight increase in the compressive strength
probably due to the consumption of water by the CaO hydration according to the equation
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
a much higher restrained expansion with respect to the concrete mixture C (800
vs. 200 microstrains at 1-3 days);
a much lower shrinkage, as compared to the concrete mixture C, in the
subsequent drying period when the concrete is no longer protected by the
polyethylene sheet and is exposed to dry air.
In other words, a concrete mixture containing both CaO, as expansive agent, and the
chemical admixture PA/SRA, as water-reducer and surface- tension-reducer, is capable to
keep its original compressive stress even in the absence of wet-curing as it usually occurs
for concrete structures demolded at 1-3 days after their placement into the formwork.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUSIONS
Consequently, crack-free concrete can be made even in the absence of early wet-
curing which, in general is needed to manufacture shrinkage-compensating concrete.
REFERENCES
(1) Sugiyama, T., Ohta, A. and Tanaka, Y., "Shrinkage Reducing Type of
Advanced Superplasticizers", Proceedings of the CANMET-ACI Symposium on
Advances in Concrete Technology, pp. 189-200, 2000.
(2) Collepardi, M., Borsoi, A., Collepardi, S., Troli, R. and Ogoumah Olagot, J.J.,
"Effects of Shrinkage reducing Admixtures in Shrinkage-Compensating
Concretes Under Non-Wet Curing Conditions", Cement Concrete Composites,
Volume 27, Issue 6, pp. 704-708,2005.
(3) Lea, F. M., "The Chemistry of Cement and Concrete", Arnold, London, 1970.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
MIX A B c D E
Cement* (IIJB-V 42.5 N) (kg!m·) 430 308 310 308 312
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Water~m )
3
215 154 155 154 156
Sand (0-4 mm) (kg!m3 ) 669 776 774 766 764
Gravel (4-25 mm) (kgfm3 ) 1003 1164 1062 1149 1148
Polycarbo_xylate (PA): 1%** (kg!mj) - 3.1 - 3.1 -
Polycarboxylate (PA/SRA): 2.5%** 7.8
(k2/m)
- - 7.7 -
Expansive A2ent (CaO) - - - 30 30
w/c 0.50 0.50 0.50 0.50 0.50
ale 3.88 6.30 6.25 6.22 6.13
Slump (mm) at 30 min 170 190 200 180 190
*Blended cement: 70% OPC + 30% fly ash
**By weight of cement
CH,ti CHi
.tL l~lJ.. PA
"1l CH,H CH
M =METAL
±-! 1"
EO: ETHOXV GROUP;
PA!SRA
~
00 10 20 30
Time(days)
~ca.- 600
,~. ·~-
.ll j 4ooi!
g200~ ,.,,..~..._-
~] 0
·=~ -200
~ -400Lf+-:!:-----!~-~----::--.;:::'9----::--~
1 2 3 2
days
Fig. 4- Length change of the reinforced concrete specimens cured under water and
then exposed to air at R.H. of so%.
s
·~
4001~~~----~~~~~~------+1
p•otected air with R.H. of 50%
a_ Vi 200 bys~=tc
~ .5
l!!
PAISRA: MIX c
~-200
...
~ ]-400
~
c
·;: -600
.c
"' -800~-!o-:!:----+----,:---.:;::===~==:::::;;::===9
123 2 3 4 6
days Time (months)
-
./:
~ ·400~1~2~3~--~--~~--~3~--~4~--~--~6
d ays Time (months)
Fig. 8- Details of Fig. 7 showing a cracked area in the Concrete Mixture D and crack-
free area in the Concrete Mixture E both in the absence of wet-curing.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Numerical Modeling of
Concrete Carbonation based on
Durability Indicators
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Firstly, the methodology and the selected panel of universal Dis concerning
carbonation are presented. Secondly, with the purpose of protecting structures
against carbonation-induced corrosion, a model accounting for the coupled C0 2 -H 2 0-
ionic transports, the carbonation reactions of Ca(OH) 2 and C-S-H, the pH decrease,
and the microstructure evolution is described. In this model, the Dis porosity, initial
Ca(OH) 2 content, and liquid water permeability are introduced as major input data.
Complementary parameters are also used: Ca(OH) 2 crystal size, C-S-H content and
capillary pressure curve. The main numerical outputs are the carbonation kinetics,
the residual Ca(OH), and newly formed CaC0 3 content profiles, and the pH value. As
a first step, the model is validated with accelerated carbonation data obtained on a
cement paste and on three porous concretes. The carbonation depth and profiles,
measured by means of phenolphthalein spray test and thermal analysis respectively,
are in good agreement with the numerical simulations. The study is completed by
a sensitivity analysis. The model, together with the test methods required for the
assessment of the relevant Dis, could be included in a toolkit for durability evaluation
and prediction of carbonation-induced corrosion of real structures.
related to the concrete chemical evolution and the gas transport (drying, carbonation,
shrinkage, permeability and diffusion phenomena ... ) in the porous material.
INTRODUCTION
This paper focuses on numerically modeling carbonation. The model developed relies
upon a sufficiently detailed description of the carbonation chemistry in order to predict
the drop of pH in the pore solution, the transport of water and of ions through the porous
medium partially saturated, the reduction of porosity and the increase of the water
content induced by the chemical reactions of carbonation. These input data of the
modeling are consistent with the physically- and chemically-based durability indicators
(Dis) used in performance approaches such as the method presented by Baroghel-
Bouny1'2.
The Dis have been selected upon both their theoretical relevance in the quantification
and the prediction of durability, and their reliable and easy assessment by means of well-
defined and reproductible methods. 1' 2 For many durability issues, a single parameter is
not sufficient to characterize the behavior, owing to the various driving forces that are
involved in the transport of aggressive species and to the complex physical and chemical
processes that take place: thus, a whole set of parameters is required.
Many works have been devoted to the modeling of carbonation in concrete. The
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
More sophisticated carbonation models integrate the variations in porosity and liquid
water saturation because of carbonation. 6' 7 Bary and Sellier8 have presumed a gradual
dissolution of the set of hydrated phases (Ca(OH) 2, C-S-H and ettringite), in respecting
the thermodynamic equilibrium between the solid and the pore solution. The transfer of
Ca2+ ions by diffusion and by advection of the liquid water has also been incorporated
and induces a dissolution effect downhill from the carbonation front. Nonetheless, the
only kinetics introduced by these authors, besides for purposes of numerical
accommodations, is that of calcite precipitation. The speed constant introduced is such
that calcite formation is quasi-instantaneous. Otherwise, the other chemical reactions (i.e.
dissolution of portlandite and C-S-H) lie at equilibrium. The relative absence of
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
chemical kinetics is reflected in the general sharpness of carbonation fronts, which do not
correspond with the experimental data. 5
(1)
The reaction rate 1; 1 refers to an unit volume of the liquid phase of pores which are
partially filled with water and air. So, it must be multiplied by the volume ¢S of liquid
degree). k1 is the rate constant which has been calibrated (around 150 mor 1.L.s- 1). 10' 20
K= 1-
nCa(OH),
n~a(OH),
1/3
Rc(K) = R0 1-K+--'-K
Vcaco
S (K
3 i
) =RnCa(OH),VCa(OH), (1 - K )1/3
[ )
Vca(OH), 0
n;ca(OH)l is the initial concentration of Ca(OHh in moles per unit volume of concrete.
K is the carbonation rate of Ca(OH) 2, R the radius of the Ca(OH) 2 sphere after
carbonation, Rc the external radius of the coating of CaC03 (cf. Fig. 1), vk the molar
volume of crystal k and s is the surface of Ca(OH) 2 per unit volume of concrete. It can be
established that the rate of dissolution ofCa(OHh is qp (mol.L- 1.s- 1): 10•20
dnCa(OH),
(2)
dt
The parameter D characterizes the diffusion of ionic species through the CaC0 3
coating and is calibrated 10' 20 around 10- 13 mol.m- 1.s- 1 and h = 5,6.10-4 mol.m-2.s- 1 is a
kinetic constant.
The CaC03 formation contributes to the clogging of the pore system, causing a
decrease of porosity. 12 The change of porosity is described with a simplified analytical
model. Regarding the reduction of porosity L1¢ as a function of amounts of carbonated
Ca(OHh and C-S-H, we consider the following equation:
In the same way, it is necessary to link the build up of water released after
carbonation to the amount of hydrates carbonated. It experimentally seems that the
release ofwater is essentially due to the carbonation ofCa(OH)2?0 •13
Dgo is the diffusion coefficient of C0 2 in the air and .f(¢,S) is the resistance factor
describing the variation of space offered to the gaseous constituents and the tortuosity
effects. Numerous expressions have been proposed by researchers for the reduction factor
f The form derived from Millington's works 14 is retained: .f(¢,S)=f(I-St Thanks to
measurements of diffusion coefficients of gas 0 2 and C0 2 in mortars 15 , we fit parameters
a and b: a= 2.74 et b = 4.20. 10•20
The velocity of liquid water under the effect of capillary suction is expressed via the
Darcy's law:
11 2
K(¢) . lk(S) = .JS'[I-(1-S mt]
vl = ---k(S)gradp, With (6)
f.1
3
K = K; (¢I¢; ) [ (1- ¢;) /(1- ¢) t
where Pt and K( ¢) are the pressure of the liquid phase and the intrinsic permeability
coefficient; f.1 denotes the dynamic viscosity of the water. The macroscopic function k(S)
affects K by accounting for variations of saturation degree due to moisture transfer,
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 771
supply of water by carbonation and porosity changes. The form of k(S) is based on a
formulation proposed by Van Genuchten 16 • m is a scalar, set to 0.45 for numerical
applications. We assume that K(fJ depends on the total porosity ¢. 16 ¢; and K; are
respectively the initial porosity and the intrinsic permeability coefficient of the
noncarbonated material. Supposing that the pressure of the gaseous phase is negligible
with respect to the liquid one, we have Pcap = -p1 where Pcap is the capillary pressure
which is given as a function of saturation rate S by the desorption isotherm, characteristic
of the cement based material. We assume that the moisture transport is controlled by the
sole liquid phase, the contribution of the vapor transfer being neglected, as recently
proposed by Baroghel-Bouny et al. 17
The flux of ions in the pore solution is given by the extended Nemst-Planck-Einstein
relation with an advection term for the constituent i in solution:
zF
wi = -D;(¢,S) grad [i]--'-D;(¢,S)[i] grad 'P +[i]V.
RT
9,95¢
(7)
with D;(¢,S)=2.9x10-4D;o e
1+625(1-S) 4
where D; is the diffusion coefficient at the macro level through the porous medium
and D;o at the micro scale in solution. D; depends both on ¢and S via a relation proposed
by Bary et Sellier. 8 z; is the valence number of the species, F is the Faraday constant, R
the ideal gas constant, T the temperature and 'P the electrical potential which is chosen to
ensure that the electrolytic solution remains neutral, i.e. FL:z";[i]=O. Actually, it is possible
to show that if the system is initially neutral the electro-neutrality is accomplished by
enforcing zero total courant FLz;wFO. 10•20
The mass balance equation for each constituent i takes the general form:
ani= - d"lVW.
-
at , + "L...JV'·l..,.~
j j
(8)
with n; as the content of i per unit volume of material. V;J is the algebraic
stoichiometric coefficient of constituent i in chemical equation}. V;,J<O (resp. >0) if i is a
reactive (resp. a product). V;,J = 0 ifi does not appear inj.
As a first step, accelerated carbonation tests were conducted to validate the model
thanks to experimental results.
A first accelerated carbonation test was performed on a W/C = 0.45 cement paste C45
(cement CEM I 52.5). Cylindrical moulds of the fresh cement paste were rotated for 48
hours to avoid any segregation and sedimentation. After demoulding, cylindrical samples
were wrapped in two superimposed adhesive aluminum foil sheets to guarantee a suitable
water-tightness and were stored (T = 20° C and RH = 95%) during 8 months to ensure
complete hydration prior to exposure to carbonation. After this curing, samples were wet-
sawn by half, leaving a side face exposed to unidirectional drying and carbonation. The
program continued with a 1 month drying at 45° C and a 8 months storing into sealed
cells in which RH is controlled at RH = 53% and T = 20° C. The imposed RH = 53% is
chosen in such a way that the speed of carbonation goes through a maximum according to
Wierig?' One considers that the distribution of water in the specimen is relatively
homogeneous after such a pretreatment. Then, the accelerated carbonation took place
during 14 days in a gas chamber where a continuous flow of a 50% C0 2 - 50% air
mixture was imposed and where RH = 53% and T = 20° C were maintained. After 28
days of accelerated carbonation, the test was interrupted; a specimen was split to measure
the carbonation depth by spraying a pH-indicator (phenolphthalein) whose color changes
at pH= 9. Another specimen was sawn into 2-3 mm slices parallel to the C0 2-exposed
surface and a Thermo-Gravimetric Analysis (TGA) apparatus was used to quantify the
solid concentration ofCa(OH) 2 and CaC03 in each slice. 22
The initial characteristics of the material used are given in Table 2. The initial
permeability coefficient Ki is estimated by the Katz-Thompson theory. 18 The initial
hygral state after pretreatment is determined with the desorption curve (linking S to RH)
by assuming that specimens have attained moisture equilibrium at RH = 53%. The
desorption curve is taken from results obtained on a cement paste of same W/C but
containing a different CEM I 52.5 cement. 17 The average radius R 0 of masses of Ca(OH) 2
was linked to W/C ratios by Chaussadent et al. 19 The content of C-S-H is calculated by
assuming C3S2H3 as stoichiometry of C-S-H and a complete hydration of anhydrous
calcium silicates C3 S and C2 S.
Fig. 3 depicts depth-profiles of pH. Assuming that the depth of carbonation obtained
by spraying phenolphthalein corresponds to a pH value of 9, we add in Fig. 3 the depth of
carbonation, computed by the model, as a function of square root of time. A satisfactory
level of agreement is observed between model and experimental results. Furthermore, we
remark that a certain time (around 4 days) is necessary to sufficiently carbonate the
specimen surface, i.e. to make the pH value drop under 9. This phenomenon is directly
linked to the effect of the chemical reactions kinetics. Actually, if the chemical reactions
were at equilibrium, the pH value would have instantaneously decreased below 9.
The LCPC model developed herein is indeed consistent with the performance-based
approach given that the model input data correspond well with the so-called physically-
and chemically-based durability indicators (Dis) ofperformance. 1•2 Nonetheless, some of
these parameters exert much stronger influence than others on the carbonation
phenomenon. Hence, a model sensitivity study was conducted. We consider that the
different Dis are identified with a certain lack of precision; it induces some discrepancies
on the carbonation depth which can be computed. The results are presented in Fig. 5: they
illustrate that the key parameters are the porosity, the portlandite content and the
permeability, which remains consistent with the conclusions forwarded in the litterature. 1
The variations of the portlandite crystals size, and the initial C-S-H content have weak
influence on the carbonation penetration. Besides, Fig. 5 indicates that the LCPC model
is very sensitive to the external C02 concentration and relative humidity, which is of
course an expected result often mentioned in the literature.
The simulated depth-profile of Ca(OH)2 is shown in Fig. 6 for concrete M25 and is
consistent with TGA measurements. The depth of carbonation indicated by spraying
phenolphthalein is compared with the calculated concrete depth Xc on Fig. 7 for M25.
We note a good agreement. The variation in Xc vs. time only conforms with a square root
law once a sufficient time period has elapsed (14 days of accelerated carbonation) for a
carbonation depth of about 3 em, as detected by means of phenolphthalein projection. It
should be kept in mind that such a carbonation time frame may correspond with several
decades in the case of natural carbonation. For illustration purposes, the natural
carbonation depth of the same M25 concrete is equal to approximately 1 em after 4 years
of natural carbonation. 1
The study is completed with an analysis to investigate the sensitivity of the model to
variations in material parameters. Small variations of these input data have a great
incidence on output data such as depth-profiles of pH or Ca(OH)2 content. Parameters
assessment depends on the level of measurement precision. By taking into account the
variability of these indicators, we show on Fig. 7 a curve which underestimates and
another which overestimates the depth of carbonation. The experimental results,
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUSIONS
The approach for describing the carbonation of concrete takes into account C0 2 , ions
and water transfers. Chemical reactions are introduced using equilibriums or kinetics and
constitute source or sink terms in mass balance equations. The diffusion coefficients and
the water permeability are affected by the reduction of porosity and supply of water due
to carbonation.
In its current version, the proposed model gives interesting results, which permit to
improve the global comprehension of the accelerated carbonation processes. In particular
it emphasizes the impact of kinetics of chemical reactions on the shape of the carbonation
front. It actually shows that neutralization does not occur immediately, even on the
surface which is directly exposed to C0 2•
At the present stage, the model is rather devoted to simulate accelerated conditions,
but it also integrates the drying of the material and the transfers of ions in order to
simulate shortly in situ conditions of carbonation.
This model would constitute in the future a powerful prediction tool based on relevant
physico-chemical durability indicators and could be integrated in a more general toolkit
for an efficient evaluation and prediction of durability of real reinforced concrete
structures.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
o Experiments (TGA)
- Simulations
30 40
11
J:
a.
40.p%
7 38.4% I
41.6% ~~
I
2.10119 m2
6 8.10-20 m2 5.10-19 m2 Kl
I
I 1
6,1 moi.L-1 5,6 rmi.L- n'Ca(OH)
~5 5,1 moi.L- 1
2
I
.l!l
Q) 2,4 rrYoi.L- 1
E4 2,9 moi.L- 1 "'j"" 1,9 moi.L- 1 n~SH
~ I
tO 30\Jm
0..3 20Jlm -'--
I
40 Jlm Ro
I
55.%
2 65% 45% RH .,.
I
5~%
40% 60% [C0 2],,.
0 ~7~--~a----~9--~-1~0-----1~1----~1~2~
Xc(mm)
Fig. s-Sensibility analysis of the LCPC model. Identification of the key durability
indicators (input data of the model).
:r"'
..,
0
()
0::::
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
40
'E 30
.su '
X
,' '
20
, ,, '
,
,,
I ''
10 I
I
I
I
• Experiments (Phenol)
I
I
I _ Simulation: Xc
I
I ___ Simulation: Xc min and max
00 2 4 6 8
..Jt (..Jdays)
Fig. 7-Depth of carbonation Xc as function of square root of time for the concrete
M25 (comparison with the upper and lower Xc corresponding to the values of input
data which cause either an overestimation or underestimation of XJ
50rr===~~---~-----~----.
- M25
"""' M40
40 ••• M50
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~ 30
E
.su
::::~:
X 20
........ I
10 ··''I l f •L-
_.)·······''] •. --}------1
( I •' .-
%~~~~2~----~4~----~6~----~a
..Jt (..Jj)
Fig. 8-Depth of carbonation for the concrete mixtures M25, M4o, and Mso.
INTRODUCTION
Shrinkage is one of the main reasons of reinforced concrete structure failures (1, 2).
These failures consists of crack formation, in new building structures, and of repairing
mortars detachments from the substrate (the original structure), in restoration works. In
scientific literature the use of shrinkage reducing admixtures is seen as able to improve
significantly the dimensional stability of hydrating cementitious systems (3), thus
extending the service life of reinforced concrete structure. Moreover, some admixtures
are widely used, and others start to be appreciated by the market, in order to reduce the
shrinkage or to contain its negative effects, like the appearance of cracks or the
detachments of repair mortars from the substrate. In the range of products, expansive
agents, based on calcium oxide, magnesium oxide or 'aluminosulfate' (C 4A 3S) (4), are
widely used in the production of pre-blended mortars for compensating the shrinkage of
cement based materials. Shrinkage compensation occurs through an expansion, which,
properly restrained by the reinforcing rods, or by bonding to the substrate, is able to
create a slightly compression pre-stress (5). This compressive stress is able to resist the
tensile stress subsequently induced by the restrained shrinkage. Nevertheless, the extent
of the induced pre-compressive stress depends strongly on the conditions of the job-site
and, in particular, on the curing after the placing. In fact, it is noticed that if a wet curing
is not guaranteed the transformation of calcium oxide in the correspondent hydroxide is
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
EXPERIMENTAL DETAILS
The study has been performed by detecting mechanical properties of mortars with
variable composition: without admixtures, with 7% (by cement mass) of an expansive
agent (EXP), with 0.4%, 1.2% or 2% of a shrinkage reducing admixture (SRA), and
with different combinations of both admixtures. The mortar specimens have been
characterised by measuring their compressive strength development, unit weight, free
drying shrinkage, restrained shrinkage and restrained expansion.
Unit weight, slump, air content, compressive strength, restrained shrinkage and restrained
expansion of concrete have been also evaluated.
The influence of EXP and SRA on the hydration of cement paste samples has been also
studied by: ESEM-FEG (Environmental Scanning Electron Microscopy- Field Emission
Gun) and specific surface area measurements.
Raw Materials
A portland cement type I 52,5 R has been used for mortar and cement paste sample
preparation. Several cements have been employed to make concrete: type II/A-L 42,5 R;
type II/A-LL 32,5 R; type II/A-LL 42,5 R. All the cements are defined according to ENV
196/l. A normalised (ENV 196/1) silica sand (0-2.5 mm) has been utilized for mortar
preparation. Two natural silica sands and six different crushed gravels have been used for
concrete preparation.
A propylene glycol ether as shrinkage reducing admixture (SRA), a calcium oxide based
expansive agent (EXP) and an acrylic based superplasticizer have been employed. The
selected properties of the admixtures are shown in Tab. l.
Mechanical Test
Cement paste, mortar and concrete compositions are reported in Tab. 2 and Tab. 3. The
mortar specimens have been prepared and stored according to ENV 196/1 (40x40x160
mm). Compressive strengths have been determined at 24 hours, 7 and 28 days (each
result is an average of six measurements). Workability (EN 13395 -each result is an
average of three measurements) and unit weight (EN 1015/6- each result is an average
of three measurements) have been detected on fresh made mortar samples. Free drying
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
specimen and welded to two metallic plates) have been cured for 8 hours at 20 oc and
95% R.H.; the specimens have been demoulded and conditioned in water; the restrained
expansion data have been calculated with respect to the initial steel rod length and
measured from 24 hours till 28 days after mixing. Restrained shrinkage has been carried
out as already described for the restrained expansion test with the difference that, after 8
hours at 20 oc and 95% R.H., the specimens have been kept in a dry environment (23 °C
and 50% R.H.). Concrete has been mixed according to EN 480/1. Fresh made concrete
has been characterised by measuring unit weight (each result is an average of two
measurements), air content (EN 1235017- each result is an average of two measurements)
and slump (EN 12350/2). The dimensional changes of concrete specimens have been
followed by restrained shrinkage and restrained expansion (each result is an average of
two measurements) on 80x80x240mm samples (UNI 8148). Compressive strength (EN
12390/3 - each results is an average of three measurements) on 150x150x150 mm
samples has been also determined.
The investigations have been performed on cylindrical cement paste samples (135 g)
prepared at 23 °C and 50% R.H. and stored in sealed plastic containers.
An ESEM (Mod. XL30 ESEM-FEG- Philips) study has been performed after 2 hours of
hydration (vapour pressure of 6.0 Torr and 5° C by means of a Peltier stage).
The specific surface area (B.E.T. method - each result is an average of three
measurements) has been measured by an instrument type Coulter mod. SA 3100 after
curing for 24 hours.
Static surface tension was determined by a tensiometer kl2 Kruss (according to Lecomte
du Nouy method) on different saturated calcium hydroxide solutions containing several
dosages of SRA (each result is an average of two measurements).
RESULTS
Mechanical Test
The results of Tab. 3 and 4 (standard deviation: 3%) show that the specimens (both
mortar and concrete) added with SRA (for example MIXl,,. and MIXlexp+sra) are
characterised by a lower compressive strength development than the reference (MIXl,.r).
These data also suggest that EXP do not influence the compressive strengths (7% EXP
and 0% Admixture have similar values). The air content (or unit weight) among the same
group of samples is rather similar (Tab. 3 and 4) and, therefore, it cannot affect the
compressive strength values. Table 5 (free drying shrinkage data of mortar specimens)
points out that SRA promotes the reduction of mortar contraction. Moreover, 7% EXP
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 785
does not show a significant mortar shrinkage reduction. Likely, EXP acts during the first
24 hours of curing, and therefore, the test values do not take into consideration the initial
mortar expansion. An effective shrinkage reduction can be obtained by the combined use
ofEXP and SRA. In water, the samples containing the expansive agent show a significant
expansion (Tab. 7 and 9- standard deviation: 4%).Whereas, the samples admixed with
SRA are characterised by dimensional changes similar to the reference (0% Admixture).
After 28 days, the restrained shrinkage tests show that a large part of the mortar and
concrete bars (stored at 50% R.H.) are subjected to a significant dimensional contraction
(Tab. 6 and 8 - standard deviation: 4%). This shrinkage is less pronounced in the
specimens containing both admixtures (EXP and SRA). In order to verify if, at a time t
after mixing, the shrinkage reduction, due to the combined use of SRA and EXP, arises
from a synergistic interaction between the admixtures, the following considerations could
be carried out (for example, for the restrained shrinkage determination, assuming t=28
days and for concrete system MIXlexp+sra)
Srt= 28 days shrinkage ofMIX1rer: -380 f.un/m;
S51= 28 days shrinkage ofMIX1sra: -191 ~-trnlm;
Sex1=28 days shrinkage ofMIX1exp: 550 ~-trnlm;
S1=28 days shrinkage ofMIXlexp+sra: 913 ~-trnlm.
From these values, the following differences can be calculated:
Shrinkage difference between MIX1sra and MIX1reFL1SRA= ISscSrtl= 189 ~-tm/m;
Shrinkage difference between MIX1exp and MIX1rer=L1EX=ISexcSrtl= 930 ~-trnlm;
Shrinkage difference between MlXlexp+sra and MIXlrer=~S=IScSrtl= 1293 ~-trnlm.
Defining L1=L1EX+L1SRA=lll9 ~-trnlm, if L1=L1S, the shrinkage reduction is due to the sum
of the two individual admixture contributions. However, since L1S>L1 (1293~-trnlm
> 1119~-trnlm), the interaction between SRA and EXP could be considered synergistic.
This effect is not evident for the drying shrinkage test.
SRA causes an increase of the specific surface area of hydrating cement (Tab.IO -
standard deviation: 10% ). Therefore a significant effect, on the morphology of hydrating
cement phases, may be expected.
In Fig.l an ESEM micrograph of the reference paste (0% Admixture) after 2 hours of
hydration, is presented. A gel like material surrounding cement grains is evident. Typical
hexagonal plates of likely portlandite are visible when only the expansive admixture has
been employed (3% EXP - Fig. 2). A dramatic morphology change has been noted in
cement pastes added with SRA (Fig. 3): some long prismatic needles are present on the
material surface. Such crystalline formation is more pronounced when also EXP is
employed (Fig. 4). This effect could suggest that such prismatic needles could be crystals
of hydrated lime with a particular morphological shape which could arise by the presence
of SRA. These crystals are too thin to be analysed by EDS (Energy Dispersion System)
and their chemical composition has been not determined.
The results in Fig. 5 show that at increasing SRA concentration a static surface tension
reduction of saturated calcium hydroxide water solutions occurs.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Today the scientists generally agree that water evaporation causes material deformation
through several possible mechanisms depending on the environmental relative humidity
(2, 7, 8). For example, at a relative humidity higher than 40% and lower than 100%,
water, in the partially filled mortar pores, can attract, by capillary pressure, the pore walls
thus determining the material contraction. The lower is the capillary stress, the less
pronounced is the shrinkage. A simplified expression of the capillary stress (9) is:
P=2'Yir (1)
Where
P = capillary pressure;
y =surface tension of the pore solution;
r =radius of the meniscus.
High shrinkage may cause cracks and degradation of cementitious materials which is a
main problem of the building industry. Some admixtures, which are able to reduce
shrinkage, are today commercially available (3). Several studies (10, 11) pointed out that
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
SRA, being a non ionic surfactant, could reduce surface tension of cement pore solution.
According to equation (1 ), the decrease of surface tension (y) favours a capillary stress
reduction with a consequent decrease of drying shrinkage. Our experimental results (Tab.
5, 6, 8 and Fig. 5) confirm that SRA reduces static surface tension (of saturated calcium
hydroxide water solutions), drying and restrained shrinkage at 23°C and 50% R.H .. As a
matter of fact, at all curing ages and for drying and restrained shrinkage tests, the
specimens with added SRA shrink less compared to the reference samples. If the
specimens are cured in water (restrained expansion test - Tab. 7 and 9), SRA does not
affect the dimensional variations of the cementitious material (both mortar and concrete).
This behaviour suggests that, as the specimens are not subjected to evaporation, drying
shrinkage and capillary stress are not operative and, therefore, SRA is not effective.
The use of expansive agents such as, for example, alkaline earth metal oxides (in
particular calcium and magnesium oxide) (4), is another way commonly used from the
concrete admixture manufacturers, to reduce shrinkage. These chemicals reacting with
water slowly produce an expansion due to the formation of the corresponding hydroxide.
The transformation of calcium oxide to calcium hydroxide causes a molar solid volume
increase of about 90% (12) which, at proper rates, could compensate material contraction.
The mechanism of expansion has been already studied. Chatterji (12) described the
expansion in terms of delayed calcium hydroxide formation. This compound, having a
molar volume higher than calcium oxide, gives origin to a crystal growth pressure.
According to Chatterji, the expansion is strictly related to the solubility of calcium
hydroxide: the higher the solubility, the more diffused is the migration of calcium ions,
the higher is the distribution of calcium hydroxide and the lower is the expansion. The
experimental results (for example Tab. 8 in comparison with Tab. 9) show that EXP is
more effective for water immersed samples. Calcium oxide is strongly embedded in the
cement matrix and likely the mixing water alone could not be enough to complete its
reaction with water. Therefore, an external supply of water (through water immersion)
could allow a better calcium oxide hydration, which favours a full expansion. On the
contrary, at 23oc and 50% R.H., calcium hydroxide formation, due to the initial slight
calcium oxide hydration, could not completely compensate shrinkage. In fact, after 28
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 787
days (Tab. 6 and 8), the specimens containing only EXP have the tendency to shrink. At
23 oc and 50% R.H., when the admixtures are used together, a low shrinkage or an
expansion is evident (Tab. 5, 6, 8). As already mentioned, the effect of shrinkage
reduction, for restrained shrinkage test, seems due to a synergistic interaction between
EXP and SRA. A first hypothesis of this synergism could be the possible massive
formation of elongated crystals (Fig. 4) during the first hours of curing. These crystals
could favour an initial expansion of the cement system, which could compensate the
material contraction. The expansive effect due to crystal growth has been already studied
(12, 13, 14, 15). Furthermore, hypothesising that such crystals are mainly composed of
portlandite, Chatterji (12) stated that the lower is the solubility of calcium hydroxide, the
higher is the mortar expansion. SRA, being an organic hydrophobic molecule, could
reduce the ability of water to dissolve salts (16). Therefore SRA could reduce the water
solubility of calcium hydroxide, which is already low, promoting the material expansion
and compensating more efficaciously the shrinkage. The lack of synergism of drying
shrinkage test could depend on the fact that the measurements have been referred to the
length of the test sample after 24 hours of curing and the effect of expansive crystals,
formed during the first hours of curing, could be missed.
CONCLUSIONS
The use of an expanding agent (calcium oxide based) with a shrinkage reducing
admixture (propylene glycol ether based) allows to obtain mortars and concrete less
sensitive to drying. The results of restrained shrinkage test have showed a synergistic
interaction between the mentioned admixtures and some hypotheses have been proposed
to explain it.
REFERENCES
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
York, 1983, 328.
90Ca0;
Chemical Composition(%)
10 CaC03
EXPANSIVE (EXP)
Specific Gravity (kgll) 0,8
Material kept on a 8011m sieve(%) 98
99 Propylene glycol
Chemical Composition(%)
ether
26Acrylic
Chemical Composition(%)
SUPERPLASTICIZER Superplasticizer
Density (kg/1) 1,08
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Normalized
sand (g)
2700 2700 2700 2700 2700 2700 2700 2700 cc
I
EXP (g) I 63 I I 63 63 63 3 I 3
iiJ
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
~
--·
0
0"
"""'
n
0
:::s
-
n
-t
tD
tD
.......
co
Daneshlink.com
\0
No reproduction or networking permitted without license from IHS
Provided by IHS under license with ACI
Copyright American Concrete Institute
.......
\0
0
:=
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
cu
::::;
tD
U)
Table 3- Concrete compositions and some physical-mechanical properties
-
tD
Acrylic SRA 1'--ompressive strength
Cement Super- EXP dosage
Slump(mm) (MPa) tD
Aggregate type Water/ Unlt
Type and plasticizer dosage kgm"' kgm"'
aoddosage Cement weight Air
cu
dosage
(kgm"')
(kgm"') ratio
dosage kgm" (o/o by cement
(%by
cement mass
mass)
(%by
cement
mass)
(kgm"')
(%)
0
30 1 24 7
minutes boor boors days
28
days =-
Mixlrer Sand0-5mm: I I 2443 2.5 215 160 110 35.0 51.8 59.0
650;
Mix!"" Cem.II/A-L Gravel6-12mm: 30.0 (7.5) I 2440 2.2 220 165 100 33.9 50.0 56.3
Mix! 0.36 2.8 (0.7)
42,5R: 400 850; I 5.0 (1.25) 2441 2.4 230 180 110 26.7 42.4 52.5
Mix!~
Gravel!8-30mm:
Mix!.,.._ 380. 30.0 (7.5) 5.0 (1.2S) 2444 2.0 22S ISS · lOS 28.3 4S.6 SS2
Mix2,., Sand 0-4mm: I I 2336 3.7 235 220 180 18.7 37.2 43.7
1080;
Mix2"" Cem.JI/A- 24.0 (7.0) I 2350 3.5 230 190 140 18.7 36.9 43.6
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
Gravel4-16mm:
Mix2 LL 32,5R: 0.52 4.0 (1.2)
180;
Mix2~ 338 I 5.0 (l.S) 2340 3.1 233 195 135 16.2 36.4 39.5
Gravel8-20mm:
Mix2...- 550. 24.0 (7.0) 5.0 (1.5) 2368 2.5 225 185 140 17.5 36.9 40.9
Gravel8-15mm:
Mix3...- 660. 25.0 (8.6) 5.0 (1.7) 2385 2.0 135 80 I 12.9 28.2 40.6
Daneshlink.com
Durability of Concrete 791
Table 4- Mortar physical-mechanical properties
2.0%SRA 10 -380
0% Admixture 10 0 -5
0.4%SRA 15 20
1.2%SRA 7 -5 2
2.0%SRA 22 12 6
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Mixlrer 21 -8 8
Mixlsra 15 5 11
Mix2rer -13 4 0
Mix2sra 5 -5
P (3%EXP +3%SRA) 23
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
80
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
70
60 '\
50
_'\
~
E
40 ""-
30
20
10
0~--~--~------------~--~--~
0 0,5 1,5 2,5 3,5
% SRA
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
by M.D.A. Thomas
Synopsis: This paper reports the results of a study on the resistance of lightweight
aggregate concrete to the penetration of chloride ions. Concrete specimens were
fabricated with a blended silica fume cement at a water-cementitious materials ratio
ofW /CM = 0.40 or 0.30 and with combinations of aggregate as follows: (i) limestone
coarse aggregate and river sand, (ii) expanded slate coarse aggregate and river sand,
or (iii) expanded slate coarse and fine aggregate. A further series of mixes was made
using the latter combination of aggregates with the blended cement being partially
replaced with 25, 40 or 56% fly ash. Concrete specimens were subjected to a series
oftests including "rapid chloride permeability" (ASTM C 1202), and non-steady-state
diffusion (bulk diffusion test). Tests were conducted at 28 and 56 days, and 1 and 3
years.
The results up to one year clearly show the benefits of incorporating expanded
slate in the concrete, with permeability and diffusion coefficients being reduced
significantly. The improvements attributed to the presence of the lightweight
aggregate appeared to increase with the maturity of the concrete and, after 3 years
continuous curing, the reduction in the apparent chloride diffusion coefficient was
observed to be as much as 70%. As expected, the addition of fly ash produced
further reductions in permeability and diffusion.
The data developed in this study were used as input parameters for service life
predictions. Although, there are insufficient data to allow firm conclusions to be
drawn from these analyses, it is clear that the incorporation of lightweight aggregate
will lead to a significant extension of the service of life.
INTRODUCTION
HPC with silica fume and low W/CM develops low permeability and high resistance to
ion penetration at early ages, but generally exhibits relatively modest improvements with
age. This behaviour is attributed to internal self-desiccation of the low W/CM system,
which stifles the later age hydration of the cementitious system. Lightweight aggregate is
characterized by having high internal porosity and, if aggregate is prewet to an adequate
degree of saturation prior to mixing, it may provide an internal source of moisture to
offset the effects of self-desiccation (2).
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
This paper reports results from chloride diffusion and permeability studies of concrete
containing expanded slate LW A.
EXPERIMENTAL METHODS
Materials
A single source of blended cement, meeting the requirements of CSA A362-98 for a
Type JOE-SF blended Portland silica fume cement, was used for the study. The cement
contained 8% silica fume and its chemical analysis is presented in Table 1. Some mixes
contained fly ash, which met the requirements of CSA A23.5-98 for a Type F fly ash and
the properties of the fly ash are also shown in Table 1.
The expanded slate aggregate used was from a commercial source in North America and
was produced in a rotary kiln at approximately 1150°C (21 00°F).
Concrete Mixes
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Two series of concrete mixtures were produced and proportions are given in Table 2. The
first series of mixes were produced with blended silica fume cement and W/CM ratios of
either 0.30 or 0.40, and the following combination of aggregates:
The second series of mixes were all produced with a W/CM of0.30 and a combination of
lightweight coarse and lightweight fine aggregate. In this series the blended silica fume
cement was partially replaced with a CSA Type F fly ash to produce four mixes with 0,
25, 40 or 56% fly ash. The silica fume contents of these mixes was 8.0, 6.0, 5.6 and
2.6%, respectively.
In all mixes the lightweight aggregate was soaked for 24 hours prior to mixing. The batch
water content was corrected to allow for any additional "free" moisture above that
absorbed by the aggregate.
The dosages of water-reducing chemical admixtures were adjusted for each mix to
produce concretes with slump values in the range of 125 ± 25 mm. As freeze-thaw
durability was not part of this study, concrete mixtures were not air entrained.
All concretes were cured under wet sacking and polythene for the first 24 hours after
casting. After this initial period, concrete specimens (1 00 mm diameter x 200 mm long
cylinders) were cured either in a fog room at 23°C or over water in sealed containers at
The first test was ASTM C 1202, which actually measures electrical conductivity rather
than chloride penetrability. However, within limitations, the test does provide an indirect
measure of concrete's resistance to ionic diffusion and it has earned the sobriquet of the
"Rapid Chloride Permeability Test" or "RCPT". Tests were conducted after 28 or 56 days
storage at 23°C or 38°C, or after 1 or 3 years storage at 23°C.
The other test used was a bulk chloride diffusion test. For this test 50-mm thick test
specimens of various ages were cut from 100-mm diameter cylinders, the curved surfaces
and one flat face were sealed with epoxy and the specimens were then vacuum saturated.
After saturation, specimens were immersed in a chloride solution (165 g NaCl per litre)
in a sealed container at laboratory temperature (nominally 23°C). After immersion
periods of 40 or 90 days, specimens were retrieved from the salt solution and were
sampled by precision profile grinding. This technique involved grinding successive layers
from the exposed surface of the specimen in 1-mm depth increments down to a depth of
10 mm or greater. The resulting powder samples were then collected and subjected to a
total chloride analysis using nitric-acid digestion and subsequent titration against silver
nitrate solution. The resulting output of the test is a profile of chloride concentration
versus depth from the exposed surface.
The penetration of chlorides into saturated concrete has been shown to follow Fick's
second law, which states that, for a semi-infinite system:
Eqn. 1
Cx,· =1-erf(
- x J Eqn.2
Cs 2~Da ·f
The unknown values of Cs and Da are found by fitting Eqn. 2 to the experimental data
using commercially available curve fitting software. Figure 1 shows the output from the
curve-fitting program and the experimental data for 3-year-old samples from Mix #1
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Specimens were prepared for testing after 28 days, I year or 3 years storage in the fog
room at 23°C.
RESULTS
Results from the RCP Test are shown in Table 3 and Figs. 2 and 3. It is immediately
apparent from these data that all of the concretes tested are of low permeability and likely
to have a high resistance to chloride ingress. In the ASTM C 1202 test method there are
guidelines for interpreting the results with values below 1000 coulombs indicating
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
concrete of "very low chloride ion penetrability" and values below I 00 signifying
"negligible chloride ion penetrability". All nine concrete mixes tested gave values
representing very low penetrability when tested after 28 days curing at 23°C. There are
significant differences between the values recorded for concretes with W/CM = 0.40 and
those for concretes produced with the same materials, but a lower W/CM (= 0.30), as
expected. But, for concretes of the same W/CM, the differences between mixes after 28
days at 23°C are generally small, the exception being the relatively high result recorded
for the concrete with W/CM = 0.30 and the highest level of fly ash (56%). However, as
either the age of the concrete or the temperature of curing increases, in other words with
increases in maturity, differences between mixes become increasingly apparent. Concrete
containing L WA or fly ash (or both) show very significant improvements over the normal
density concrete without fly ash, especially when tested at later ages. Indeed RCP values
below 100 were measured in concrete with lightweight coarse and fine aggregate when
part of the blended silica fume cement was replaced for fly ash. The mix with LW A and
56% fly ash recorded the lowest reading of28 Coulombs.
The chloride diffusion data are presented in Table 4. The general trends in behaviour are
similar to those observed for the RCPT data, with concretes containing L W A and fly ash
becoming progressively more resistant to chloride ingress with aging compared to
concrete with normal density aggregate and without fly ash. Again the differences
between mixes of the same W/CM are not too significant at early ages.
DISCUSSION
The magnitude of the observed changes in both the RCP value and bulk diffusion
coefficient with time are significant for all mixes. The ratio of these values at I year and
3 years to the early-age values at 28 days are as follows:
Mix 2 3 4 5 6 7 8 9
where: R28, R365 and R 1095 are the RCP value at 28, 365 and 1095 days.
D 28, D 365 and D 1095 are the diffusion coefficients at 28, 365 and 1095 days.
The reductions in electrical conductivity and ionic diffusivity for the LW A mixes with fly
ash are considerable, especially at the higher levels of fly ash. Mixes with 40% fly ash
showed a reduction in these values from 4 to 8 times between 1 month and 1 year,
compared with reductions between 12 to 20 times for the mix with 56% fly ash. Changes
in chloride diffusion with time have a significant impact on the time taken for chlorides
to reach embedded steel in sufficient concentration to initiate corrosion and predictions
based solely on early-age measurements can lead to very large errors (8). A common
approach used in service-life modelling is to use a power law to describe the time-
dependent nature of diffusion; the following equation is an example (9):
Eqn.3
Some models use D 28 , the diffusion coefficient at t = 28 days, and this value together with
the exponent m will vary depending on the mixture proportions, the type of cement and
the curing and exposure conditions of the concrete (9). The bulk diffusion data at 28 days
and 3 years for mix series I, and data at 28 days and 1 year for mix series II were used to
calculate values of m and the results are shown below. In calculating these values it was
assumed that the diffusion coefficient determined from tests that were initiated at 28 days
actually corresponded to an equivalent age of 48 days since the test duration was 40 days
(i.e. the specimens were 48 days at the midpoint of the test). Also shown are calculated m
values applying Eqn. 3 to the RCPT data.
Mix 2 3 4 5 6 7 8 9
RCPT 0.05 0.35 0.07 0.07 0.31 0.38 0.79 1.01 1.37
Bulk 0.09 0.40 0.06 0.13 0.23 0.23 0.47 0.69 1.46
Diffusion
It should be noted that the m values calculated from the bulk diffusion data were based on
only two measurements per mix (i.e. 28 days and 3 years or 28 days and 1 year) and
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
The importance of the value of m on the service life can be demonstrated using a service
life model. In this case, the model Life-365 was used with the following input values:
In this model the diffusion coefficient D 1 at any time tis calculated from the values of D 28
and m using Eqn. 3, but the value only decreases until 25 years after which it remains
constant with time.
Fig. 6 shows the predicted time to corrosion as a function of the m value for three
different cover depths. It is apparent that the predicted life is highly sensitive to the value
of m and that ignoring the influence of time on diffusivity, i.e. making prediction based
solely on coefficients determined at early age, can lead to extreme errors. As stated
above, there are insufficient data from this study to allow a reliable calculation of m to be
made, but it is clear that the use of LWA or fly ash will have a significant impact on the
time to corrosion, especially when used in combination.
The ability of fly ash to reduce long-term diffusion is well established and understood
(10). The effect of LWA is almost certainly linked to its ability to provide an internal
reservoir of moisture for the continued hydration of the cement paste. All of the mixes
tested here were of relatively low or very low water to cement ratio (i.e. W/CM = 0.40 or
0.30) and contained blended silica fume cement. As such, internal desiccation of the
cement might be expected and this process eventually stifles the hydration of the cement
and pozzolanic reaction of the silica fume (and fly ash, if present). Additional water may
be supplied from external sources since these concretes were stored in the fog room, but
the process of moisture diffusion in these concretes is likely to be very slow, due to their
low porosity and permeability. The expanded slate aggregate used in this study contains 5
to 6% by mass of water when pre-wet prior to mixing and this water will be available to
the cement paste to replace that lost during hydration. The impact of this phenomenon is
probably more marked at the cement-aggregate interface, which is often considered to be
a weak zone in the concrete. Microstructural studies of the mature concrete specimens
(including studies of the cement-aggregate interface) are currently underway at the
University ofNew Brunswick.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
CONCLUSIONS
1. The use of LWA aggregate substantially reduced the electrical conductivity and
chloride penetrability of high-performance concrete containing blended silica fume
cement. The impact of the LWA became more apparent at later ages and this may be
explained by the aggregate supplying an internal reservoir of water, which promotes
the continued hydration and reaction of the cementing materials. The chloride
diffusion coefficient for L WA concrete after 3 years was between one third to one
half of that of normal density concrete of the same age and W/CM.
2. The combination of fly ash and L WA produced concrete with very low diffusion
coefficients, which decreased dramatically with continued curing.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
3. The "rapid chloride permeability test" does provide a reliable indication of the
concrete's ability to resist the diffusion of chloride ions.
REFERENCES
3. Kefeng, T., Zhang, M.H. and Gjorv, O.E. "Diffusivity of chlorides from seawater
into high-strength lightweight concrete." ACI Materials Journal, 91 (5), 1994, pp.
447-452.
5. Zhang, M.H. and Gjorv, O.E. "Characteristics of lightweight aggregates for high-
strength concrete." ACI Materials Journal, 88 (2), 1991, pp. 150-158.
9. Boddy, A., Bentz, E., Thomas, M.D.A. and Hooton, R.D. "An overview and
sensitivity study of a multi-mechanistic chloride transport model." Cement and
Concrete Research, Vol. 29, 1999, pp. 827-837.
10. Thomas, M.D.A. and Bamforth, P.B. "Modelling chloride diffusion in concrete: effect
of fly ash and slag." Cement and Concrete Research, Vol. 29, 1999, pp. 487-495.
00
0
0\
,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-t
::r
Table 2 Concrete Mixture Proportions 0
3
D)
Constituent Mixture Proportions (kglm3)
Series I Series II
i
"'
I 2 3 4 5 6 7 8 9
TIOSF Cement 350 350 450 450 450 450 337.5 270 198
Fly Ash - - - - - - 112.5 180 252
Water 140 140 135 135 135 135 135 135 135
W/CM 0.40 0.40 0.30 0.30 0.30 0.30 0.30 0.30 0.30
1
NDCoarse 1165 - 1121 - - - - - -
1
NDSand 771 742 742 - - - - -
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
2
LWACoarse - 659 - 634 634 634 621 613 604
2
LWAFines - 524 - - 504 504 493 487 480
3WRA 17.1 17.1 34.0 34.0 34.0 22.0 22.0 22.0 22.0
4
HRWA 25.5 43.8 34.0 66.0 57.8 57.0 52.3 45.8 42.5
- --- -- - --- - - - --
1
ND =Normal Density
2
LWA = Lightweight Aggregate
3
WRA =Water-Reducing Admixture (mL admixture per 100 kg cementing material)
4
HWRA =High-Range Water-Reducing Admixture (mL admixture per 100 kg cementing material)
Daneshlink.com
No reproduction or networking permitted without license from IHS
Provided by IHS under license with ACI
Copyright American Concrete Institute
Series I Series II
I 2 3 4 5 6 7 8 9
W/CM 0.40 0.40 0.30 0.30 0.30 0.30 0.30 0.30 0.30
Fly Ash(%) - - 25 40 56
Coarse/Fine'" NDIND LW!LW ND/ND LW/ND LW/LW LW!LW LWILW LW/LW LW/LW
Curing History
28 days @ 23°C 644 621 468 316 395 427 434 550 829 I
!=.
1 year@ 23°C 522 202 320 220 162 !54 63 44 28
3 years @ 23°C
-
530
- -
!53
- L__
343
----
24~ 112
--- - - -
- ___:____j ~
0
1
ND =Normal Density Aggregate
"""'
2
LW = Lightweight Aggregate ~
=
n
-
;
tD
00
Daneshlink.com
0
......
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
No reproduction or networking permitted without license from IHS
Provided by IHS under license with ACI
Copyright American Concrete Institute
co
0
co
;;!
0
3
C»
Ill
Table 4 Bulk Diffusion Test Results (x I o-" m"/s)
Series I Series II
1 2 3 4 5 6 7 8 9
W/CM 0.40 0.40 0.30 0.30 0.30 0.30 0.30 0.30 0.30
Fly Ash(%) - - - - - 25 40 56
Coarse/Fine1.z NDIND LWILW NDIND LW/ND LWILW LWILW LWILW LWILW LWILW
Curing Histon:
Not for Resale, 08/10/2015 01:24:04 MDT
Licensee=UNI OF NEW SOUTH WALES/9996758001
28 days @ 23•c 24.3 18.5 15.7 13.4 12.6 8.19 8.25 9.98 23.1
1 year@23•c - - - - 5.16 3.21 2.48 1.19
3 years @ 23•c
-
18.6
---
5.35
-
12.9
---
8.91
-
6.20 -- --·-··-
- - - -
-----
1
ND =Normal Density Aggregate
2
LW =Lightweight Aggregate
Daneshlink.com
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Durability of Concrete 809
3.5
l
~
2.5
2.0
• NDJND- W/CM=0.40
3.40
18.6 X 10" 13
6.20 x 10"
13
~
'-"' 1.5
:g"
~ 1.0
u
0.5
0.0
0 5 10 15
Depth(mm)
1000 ~-----------------..,
I~ 100
10+------~-----~----~
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
~}l~C }n·c
]
8 100
....
~
u
~
10
10 100 1000 10000
Age (days)
100
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
-'!;
"'s
-~ 10
.!$,
0
-+-- ND!ND- 0.30
--- LW/ND- 0.30
....._ LW!LW- 0.30
....._OFA
-+- 25 FA
::? ...,._40FA
"'e
--56 FA
"';'~ 10
~
cS
e Cover= 25 mm
+ Cover= 50 mm
~ 150
_;
-~
~ 100
u
£
~ 50
0~~~==~~----~--~----~
0.0 0.2 0.4 0.6 0.8 1.0
m Value
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
l.E-11
~
~ l.E-12
cS
••
l.E-13 •• A Previ>us Data (Uofl)
• + LWAStudy
l.E-14 +----.....,....----r------,-------1
10 100 1000 10000 100000
RCPT (Coulombs)
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Length
inch millimeter (mm) 25.4Et
foot meter(m) 0.3048E
yard meter (m) 0.9144£
mile (statute) kilometer (km) 1.609
Area
square inch square centimeter (cm 2 ) 6.451
Volume (capacity)
ounce cubic centimeter (cm 3 ) 29.57
gallon cubic meter (m 3)t 0.003785
cubic inch cubic centimeter (cm 3 ) 16.4
cubic foot cubic meter (m 3) 0.02832
cubic yard cubic meter (m 3Jt 0.7646
Force
kilogram-force newton (N) 9.807
kip-force newton (N) 4448
pound-force newton (N) 4.448
Pressure or stress
(force per area)
kilogram-force/square meter pascal (Pa) 9.807
kip-force/square inch (ksi) megapasca1 (MPa) 6.895
newton/square meter (N/m 2) pascal (Pa) l.OOOE
pound-force/square foot pascal (Pal 47.88
pound-force/square inch (psi) kilopascal (kPa) 6.895
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Temperature§
deg Fahrenheit (F) deg Celsius (C) lc = (IF - 32)/1.8
deg Celsius (C) deg Fahrenheit (F) tF = 1.8tc + 32
* This selected list gives practical conversion factors of units found in concrete technology. The reference
source for information on SI units and more exact conversion factors is "Standard for Metric Practice" ASTM E
380. Symbols of metric units are given in parentheses.
t E indicates that the factor given is exact.
:j: One liter (cubic decimeter) equals 0.001 m 3 or 1000 cm 3.
§ These equations convert one temperature reading to another and include the necessary scale corrections. To
convert a difference in temperature from Fahrenheit to Celsius degrees, divide by 1.8 only, i.e., a change from 70
to 88 F represents a change of 18 F or 18/1.8 = 10 C.
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
Index
A Bournazel, J.P., 99
Bouzoubaa,N.,283
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
curling, 445
Cussigh, F., 99 F
D Fazio, G., 747
field concrete, 239
damage, 1, 147 field trial, 539
damage analysis, 127 finite-difference method, 39
Dangla, P., 765 finite element modeling, 179
database, 721 Finnegan, L., 71
De Ia Varga, 1., 87 Finnerty, P.G., 251
De Schutter, G., 167 flexural strength, 327
DEF, 239 flint aggregate, 485
degradation, 495, 561 fly ash, 283, 589, 703, 747
degree of capillary saturation (DCS), 147 Fraaij, A., 111
Degrugilliers, P., 485 Fran~ois, R., 399
deicing salts, 373 freezing and thawing, 347
delayed ettringite formation, 305 frost salt scaling, 111
deterioration, 1, 239, 521, 571 Fumoto, T., 457
determination method, 473
Devillers, P., 99 G
Di Maio, A., 589
diagnose, 485 Garces, P., 87
diffusion, 797 Garcia-Diaz, E., 485
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
Durability of Concrete 817
Georgin, J.-F., 445 Johannesson, B., 39
Giebson, C., 225 Justnes, H., 53, 417, 521
Glasser, F.P., 495
Goldfarb, 1., 721 K
gravimetric weight loss, 205
grey level co-occurence matrix (GLCM), Kabir, S., 127
127 Kahn, l.F., 661, 681
Grima!, E., 179 Kanematsu, M., 347
guan0,521 Khelidj, A., 269
Khitab, A., 191
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---
H Kondratova, I.L., 721
Kurtis, K.E., 661
Hamada, H., 571
Hammer, T.A., 641 L
heat of hydration, 239
high performance concrete, 99, 269, 473, L'Hostis, V., 17
621,661 laumontite, 571
high-strength concrete, 621, 661 Laurens, 5., 17
high-strength lightweight concrete, 681 Le Pape, Y., 179
high-volume fly ash concrete, 283 leaky Rayleigh wave, 561
highway bridges, 373 lightweight, 661
Hirao, H., 507 lightweight aggregate(s), 607, 621, 797
Hiroyuki, K., 431 lightweight concrete, 681, 737
Hoff, G.C., 621 lightweight early age cracking, 641
holistic, 1 limestone, 507
Holm, T.A., 737 limestone filler, 167, 327
Honma, K., 507 limit state, 347
Hosokawa, Y., 39 Lindgard, J., 147
HPC, 681 Linfoot, B.T., 71
hydration, 327, 473 Livingston, R.A., 251
Lolli, A., 781
Lopez, M., 661
Lorente, 5., 191
Ibrahim, M., 205 low dosage of expansive additive, 431
Ichikawa, M., 507 LUthi, T., 561
lrassar, E.F., 327, 589
image analysis, 127 M
impact echo, 251
inhibitor, 53 macrocracks, 1
information technology, 721 Malhotra, V.M., 283
internal curing, 661 Maltese, C., 781
Internet, 721 manure, 521
marine environment, 71, 507
Masanobu, A., 431
Maslehuddin, M., 205
Copyright American Concrete Institute
Provided by IHS under license with ACI Licensee=UNI OF NEW SOUTH WALES/9996758001
No reproduction or networking permitted without license from IHS Not for Resale, 08/10/2015 01:24:04 MDT
Daneshlink.com
818 Index
mass transport, 765 ordinary expansive additive, 431
Masuda, Y., 347 Ormsby, C., 251
material properties, 681
mathematical modeling, 765 p
McCarter, W.J., 71
McMorris, N., 251 Pedersen, B., 147
mechanical behavior, 399 Pera, )., 445
mechanical properties, 641 performance-based design, 347
Meddah, M.S., 473 performance test, 225, 305
Mehta, P.K., 1 permeability, 269, 797
Menendez, G., 327 Petre-Lazar, 1., 17, 179
mercury intrusion porosimetry, 167 Petrov, N., 239, 473
Meyer, K.F., 681 Pianca, F., 373
microcracks, 1 Pistolesi, C., 781
microscopy, 539 plain and silica fume blended cement
microstructure, 1, 111, 589 concretes, 205
migration, 269 polyol, 445
mixture proportion, 681 Ponce, J.M., 589
modeling, 17, 191 Pontis, 251
Mohammed, T.U., 571 pop-out, 747
moisture, 147 pore structure, 167
moisture content, 641 pore volume, 457
Monnin, Y., 485 porosity, 445
Mori, D., 39, 507 potassium autoradiography, 251
mortar, 417, 561 Poupard, 0., 17
multi-resolution analysis (MRA), 127 prestressed concrete, 681
R
N
Raharinaivo, A., 17
natural pozzolan, 589 recycled fine aggregate, 457
NDT methods, 17 reinforced concrete, 17, 373, 399
Neuenschwander, j., 561 reinforcing corrosion, 87
Nisnevich, M., 703 reinforcement corrosion, 205
nitrate, 53 reverse analysis, 39
nitrite, 53 Rivard, P., 127
Noguchi, T., 347 Rodum, E., 147, 521
nondestructive test, 589 Romer, M., 561
nondestructive testing, 721
numerical calculation, 39 s
0 Schell, H., 373
Schlangen, E., 111
Ogoumah Olagot, J.J., 747 Schmidt, T., 561
Ollivier, J.P., 191 Scrivener, K.L., 495
--`,```,,``,,,`,,,``````,,,,```,-`-`,,`,,`,`,,`---