Vous êtes sur la page 1sur 117

SUSTAINABLE

CEMENT
PRODUCTION
CO-PROCESSING
OF ALTERNATIVE FUELS
AND RAW MATERIALS
IN THE EUROPEAN CEMENT
INDUSTRY
CONTENTS
Executive summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
Cement and concrete manufacturing: the process . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Co-processing: from waste management to resource management . . . . . . . . . . . . . . 5
Co-processings contribution to CO
2
emission reduction . . . . . . . . . . . . . . . . . . . . . . 9
No increased environmental impact from co-processing . . . . . . . . . . . . . . . . . . . . .10
Impact on cement quality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Co-processing and health & safety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
Co-processing legislation & guidelines . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
The European cement industry: an overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
Cement in Europe . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
1
AVERAGE EU SUBSTITUTION RATES:
F
Alternative fuels: 18%
F
Alternative raw materials: 5%
F
Alternative constituents: 12%
A WIN/WIN/WIN SITUATION
Industry (PROFIT):
A cost-effective substitution of natural resources thereby
improving the competitiveness of the industry
CO-PROCESSING
Ecology (PLANET):
Environmentally sustainable
waste management and impor-
tant saving of natural resources
Society (PEOPLE):
A long term and sound solution for
the treatment of different types of
waste produced by society
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
EXECUTIVE
SUMMARY
Cement is an essential product, pro-
viding society with what it needs in
terms of safe, comfortable housing
and reliable modern infrastructure.
Co-processing in the cement industry
is the optimum way of recovering en-
ergy and material from waste. It offers
a safe and sound solution for society,
the environment and the cement in-
dustry, by substituting non renewable
resources with societal waste under
strictly controlled conditions.
The co-processing of alternative fuels
provides a solution in terms of reduc-
ing fossil fuel dependency as well as
a contribution towards the lowering of
emissions. The use of alternative raw
materials also has numerous benets,
including a reduced need for quarry-
ing and an improved environmental
footprint of such activities. Substitu-
tion of clinker in cement is an exam-
ple of the positive contribution of the
European cement industry to resource
management.
The use of alternative materials in the
cement industry lowers global CO
2

emissions and does not have a nega-
tive impact on production process emis-
sions, nor on the environmental and
technical quality of the nal product.
Furthermore, co-processing in the ce-
ment industry is carried out in a safe
and sound manner, thus not affecting
the health & safety of its workers or
neighbourhood.
2
Cement and concrete manufacturing: the process
The main component of cement is clinker.
Clinker is produced from raw materials, such
as limestone and clay, which are crushed,
homogenised and fed into a rotary kiln. The
clinker burning takes place at a material tem-
perature of 1450C which is needed to form
the new compounds. Clinker consists mainly
of calcium-, silicium-, aluminium- and iron-
oxides.
The next phase is handled in a cement grind-
ing mill. Gypsum and other additional mate-
rials (such as blast furnace slag, coal y ash,
natural pozzolanas, limestone, etc.) are add-
ed to the clinker. All constituents are ground
leading to a ne and homogenous powder:
cement.
In concrete production, cement, aggregates
and other suitable materials are mixed with
water. When water is added to cement, it re-
acts and forms a glue which binds together
the other main constituents of concrete.
Extraction / Grinding Grinding / Firing
Grinding Storage
Dispatch
THE CEMENT PROCESS
WORKING TOGETHER TO SOLVE AN IMPORTANT
SOCIETAL PROBLEM
In 1999, an urgent solution for the treatment of thousands of tonnes of
animal meal and fat from potentially contaminated animal products was
required in Belgium. At the time, there was major public concern about
this potential health hazard. The federal authorities identifed the co-
processing of the contaminated meat & bone meal in the cement indus-
try as the best way of resolving this crisis. Belgian plants were, therefore,
requisitioned to treat a large amount of potentially contaminated animal
meal. This process provided a safe and environmentally sound solution
as it allowed for the complete destruction of the contaminants in the kiln,
as well as reducing emissions as a result of fuel substitution.
F More information: www.febelcem.be
A similar situation arose in Italy in 2001. Italian cement plants, in agree-
ment with the public authorities, were requisitioned to treat a large
amount of potentially contaminated meat & bone meal.
CEMENT IS AN
ESSENTIAL PRODUCT,
PROVIDING SOCIETY
WITH WHAT IT NEEDS
IN TERMS OF
SAFE
AND COMFORTABLE
HOUSING
AND
RELIABLE MODERN
INFRASTRUCTURE
2
3
The European cement industry is committed
to ensuring that society has sufcient ce-
ment to meet its needs, whilst at the same
time reducing its fuel and raw material re-
quirements and subsequent emissions. Use
of waste materials in the cement industry,
also referred to as co-processing, contrib-
utes towards achieving these objectives. Co-
processing is the substitution, in industrial
processes, of primary fuels and raw materi-
als with suitable waste materials.
The co-processing of waste in the cement
industry provides a maximum substitution
of non renewable materials. The decision on
what type of waste can be nally used in a
certain plant cannot be answered uniformly.
As a basic rule, waste accepted as an alter-
native fuel and/or raw material must give an
added value for the cement kiln in terms of
the caloric value of the organic part and the
material value of the mineral part. Some alter-
native materials will often meet both of these
requirements, making it difcult to formulate
general criteria regarding the materials which
are co-processed in the cement industry.
Due to the characteristics of the production
process, the cement industry is capable of
co-processing:
+ alternative fuels, which have a signicant
caloric value (e.g. waste oils);
+ alternative raw materials, the mineral com-
ponents of which mean they are suitable
for the production of clinker or cement (e.g.
contaminated soil)
+ materials that have both a caloric value
and provide mineral components (e.g. pa-
per sludge, used tyres)
However, not all waste materials can be co-
processed in the cement industry. Several fac-
tors must be taken into consideration when
deciding on the suitability of the materials.
These include the chemical composition of
the nal product (cement) as well as the en-
vironmental impact of the cement production
process. Examples of waste which is not suit-
able for co-processing in the cement indus-
try include nuclear waste, infectious medical
waste, entire batteries and untreated mixed
municipal waste.
An adequate quality control system is strictly
adhered to for all the materials used. This
ensures that they are co-processed in an en-
vironmentally safe and sound manner, safe-
guarding the:
+ health & safety of the workers in the plant
and the people living in the neighbour-
hood;
+ environmental impact of the production proc-
ess;
+ high quality of the nal product;
+ correct and undisturbed functioning of the
production process;
The alternative materials used by the cement
industry are derived exclusively from select-
ed waste streams. They usually require pre-
treatment (e.g. drying, shredding, blending,
grinding or homogenisation) and an appro-
priate Quality Assurance. Pre-treatment is,
therefore, an integral part of the recovery
operation. Waste is often prepared for use as
an alternative fuel by outside suppliers and
waste-treatment experts.
USE OF SOLID WASTE
The treatment of municipal, commercial and some industrial waste re-
duces the need for landflling of organic and recyclable waste. In Aus-
tria, cement plants started to use solid waste (made up of recyclable
plastics, paper, textiles and composite materials) in 1993. All nine ce-
ment plants in Austria use solid waste to a certain extent. In addition,
several companies are working in partnership with waste management
companies in order to build pre-treatment facilities to provide alterna-
tive fuels which meet cement process specifcations.
The Austrian cement industry started to co-process tyres as early as
1980.
Co-processing: from waste management to resource
management
4
The co-processing of waste has been recog-
nised as a recovery operation under EU legis-
lation. Recovery operations in cement plants
are carried out in compliance with the provi-
sions of both the Directives on the Incineration
of Waste and Integrated Pollution Prevention
and Control (IPPC), and are recognised as a
Best Available Technique (BAT).
It should be borne in mind that the degree of
co-processing in Europe varies from country
to country as a result of:
+ national regulation/waste management
+ experience (in the cement industry)
+ market and local conditions
The use of alternative resources in certain
European countries is low and has a clear
potential for growth. The European cement
industry is, therefore, keen to collaborate in
developing this further.
CASE LAW
Case C-228/00. Judgment of the Court
(Fifth Chamber) of 13 February 2003. Eu-
ropean Commission v. Germany
Ruling: A ruling delivered by the European
Court of Justice holds that using waste as
a fuel in cement kilns should be classifed
as recovery.
Characteristics Temperature and time
Temperature at main burner >1450C: material
>1800C: fame temperature
Residence time at main burner >12-15 sec and >1200C
> 5-6 sec and > 1800C
Temperature at precalciner >850C: material
>1000C: fame temperature
Residence time at precalciner > 2-6 sec and >800C
THE AVANTAGES OF CO-PROCESSING
CO-PROCESSING IN THE CEMENT
INDUSTRY PROVIDES THE
OPTIMUM
WAY OF RECOVERING ENERGY

AND MATERIAL FROM WASTE.
ADDITIONALLY IT OFFERS A SAFE
AND SOUND SOLUTION FOR SOCIETY,
THE ENVIRONMENT AND THE CEMENT
INDUSTRY BY SUBSTITUTING NON
RENEWABLE RESOURCES WITH
SOCIETAL WASTE UNDER STRICTLY
CONTROLLED CONDITIONS
+ Excess of oxygen during and after combustion
+ Complete destruction of organic compounds
+ Total neutralisation of acid gases, sulphur oxides and hydrogen chloride, by the active lime in the kiln load, in large excess to
the stoechiometry
+ Embedding of the traces of heavy metals in the clinker structure with very stable links (metallic silicates formation)
+ No production of by-products such as ash or liquid residue from gas cleaning
Total recovery of energy and mineral content of waste (raw material and fossil fuel saving). The technical characteristics are
compliant with European Directives, i.e, the Waste Incineration Directive.
5
In 2006, the European cement industry used
an energy equivalent of about 26Mt of coal,
a non renewable fossil fuel, for the produc-
tion of 266Mt cement. Alternative fuels con-
stituted 18% of this across Europe, saving
about 5Mt of coal.
In terms of using waste as an alternative fuel
in the cement industry, there are two char-
acteristics of particular importance: burning
conditions (high temperature with a long res-
idence time and oxidising atmosphere) and
a natural alkaline environment of raw ma-
terials. These conditions are particularly fa-
vourable for insuring a high level of destruc-
tion of polluting substances such as dioxins
and furans. Indeed, the SINTEF
1
report states
that with material temperature at approxi-
mately 1450C and kiln gas temperatures up
to 2000C, long residence time up to 8 sec-
onds, or more, insures complete pyrolysis or
breakdown of organic waste.
In the cement plant, dedicated facilities are
built to receive, store and inject the alterna-
tive fuels, often complemented by waste pre-
treatment facilities. These facilities take into
account the characteristics of the alternative
fuel and the specic regulation related to
the handling of waste (especially in terms of
health & safety).
Energy provided by co-processing is entirely
used in the clinker manufacturing process.
The caloric content may vary depending
on the waste used. However, it is not only
the caloric value which counts as the min-
eral content of the waste (ash) can also be
used as an alternative raw material in clinker
production, meaning no residues are gener-
ated.
> Alternative fuels:
Saving of non renewable fossil fuels
CO-PROCESSING ALTERNATIVE FUELS PROVIDES A SOLUTION IN TERMS
OF
REDUCING FOSSIL FUEL
DEPENDENCY AS WELL AS A CONTRIBUTION
TOWARDS A LOWERING OF EMISSIONS
Co-processing of used tyres: a long lasting success
story
The clinker burning process offers the possibility of a simultaneous
energy and material recovery of the individual components of the
tyres. The high calorifc value of the rubber is used to substitute
primary fuels and the inert ingredients (mainly iron and alumina)
substitute the raw materials. Moreover, if the natural raw material
does not contain enough iron, the use of tyres helps directly to
meet the desired product requirements. Based on the long last-
ing, successful experience in the co-processing of tyres, Germany
has put tyres on a list of materials suitable for the clinker burning
process. It should be noted that tyres contain a signifcant amount
of biogenic carbon (about 27% due to the content of natural rub-
ber), thus leading to a direct reduction of fossil fuel related CO
2
.
Depending on where they are fed into the kiln, tyres can also de-
liver a signifcant contribution to the reduction of nitrogen oxide
emissions.
1
Formation and Release of POPs in the Cement Industry.
Second edition. SINTEF 26 January 2006
6
In 2006 about 5% of the raw materials
used in the production of clinker consisted
of alternative raw materials, totalling about
14.5Mt/ year.
Alternative raw materials can be used to re-
place the traditional raw materials extracted
from quarries, such as clay, shale and lime-
stone, which are used in the kiln. Examples
of alternative raw materials include contami-
nated soil, waste from road cleaning and
other iron-, aluminium-, and silica-containing
wastes, such as coal y ash and blast fur-
nace slag. The chemical suitability of alterna-
tive raw materials is important to ensure that
they provide the necessary constituents for
the formation of clinker.
Cement in Europe must be manufactured ac-
cording to the harmonised standard EN 197-1
which clearly indicates the 27 common ce-
ments according to their main constituent. As
mentioned, the main component of cement
is clinker. Depending on availability, part of
the clinker can be replaced with alternative
constituents. Two major examples are gran-
ulated blast furnace slag, a by-product of the
iron manufacturing process, and y ashes,
one of the residues generated from the com-
bustion of coal.
> Alternative raw materials:
Substitution of natural virgin resources
> Cement: Substitution of clinker
with alternative constituents
SUBSTITUTION OF CLINKER IN CEMENT IS AN
EXAMPLE OF THE
POSITIVE CONTRIBUTION

OF THE EUROPEAN CEMENT INDUSTRY TO
RESOURCE MANAGEMENT
THE USE OF ALTERNATIVE RAW MATERIALS PROVIDES
NUMEROUS BENEFITS,
INCLUDING A REDUCED NEED FOR
QUARRYING AND AN IMPROVED ENVIRONMENTAL FOOTPRINT OF
SUCH ACTIVITIES
7
Rest Waste
& Energy
Waste incineration
& cement manufacturing
Waste used as fuel in
cement manufacturing
CO
2
CO
2
CO
2
Mt
CO
2
Waste
+
Waste Fossil Fuels
Fossil
Fuels
Waste
Incinerator
Cement plant Cement plant
Cement Cement
Products
Resources
Emissions
THE USE OF WASTE & BIOMASS INSTEAD OF
FOSSIL FUEL IN THE CEMENT INDUSTRY
REDUCES EUROPEAN ABSOLUTE EMISSIONS
Co-processings contribution to CO
2
emission reduction
Co-processing offers a high poten-
tial for the cement industry to reduce
global CO
2
emissions. Without co-
processing, the wastes and by-prod-
ucts which make up these materi-
als would have to be incinerated or
landlled
2
with corresponding green-
house gas emissions.
In 1990, the percentage of energy
gained by the use of waste as a fuel
was 3%. In 2006, it stood at about
18%, resulting in a reduction of 8Mt of
CO
2
emissions each year whilst sav-
ing about 5Mt of coal.
CO
2
emissions from biomass are cli-
mate neutral. More than 20% of the
alternative fuels used by the Euro-
pean cement industry consist of pure
biomass, such as meat & bone meal
and sewage sludge. This could be
developed further provided biomass
remains accessible. In addition, the
alternative fuels used in the cement
industry contain a high level of bio-
mass due to its content of e.g. wood
or paper fractions.
The CO
2
emissions of manufacturing
cement result from the production of
the intermediate product, clinker. Re-
ducing the clinker content reduces,
therefore, the energy and carbon in-
tensity of the cement produced.
TACKLING INDUSTRIAL
WASTE: CEMENT KILNS
VERSUS INCINERATORS
AN ENVIRONMENTAL
COMPARISON
A study conducted by the Nether-
lands Organisation for applied Sci-
ence Research (TNO) compared
the environmental impacts of us-
ing waste as an alternative fuel
and raw material in the cement in-
dustry, and burning waste in haz-
ardous waste incinerators while
recovering electricity and steam.
This assessment took into con-
sideration the complete life-cycle
of the different waste streams and
all the environmental impact cat-
egories (Life-cycle assessment). It
concluded that, for the vast major-
ity of environmental impacts, using
industrial waste as alternative fuels
in the cement industry was better
for the environment than treating
them in waste incinerators.
Source: Summary by Greenfacts.
F More information:
www.coprocessing.info
2
Emissions from landll consist of about 60% methane, a gas with a global warming potential
21 times that of CO
2
.
Landll &
Leachate
Waste landlling &
cement manufacturing
Waste used as fuel in
cement manufacturing
CO
2
Methane
CO
2
CO
2
Mt
CO
2
Waste Waste Waste Fossil Fuels
Fossil
Fuels
Landll without are + Cement plant
Cement plant
Cement Cement
Products
Resources
carbon content
Emissions
THE USE OF SOLID WASTE IN A CEMENT PLANT
PREVENTS METHANE EMISSIONS IN A LANDFILL.
8
No increased environmental impact from co-processing
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
Until recently, the only solution was to dump or use sewage sludge in
agriculture. However, sewage sludge can now be used as both an al-
ternative fuel and raw material in the clinker manufacturing process. It
is important to note that there is a surplus of sludge, and therefore the
need for alternative forms of processing is becoming all the more acute.
The Netherlands and Spain are just two examples of countries where
the cement industry is providing a solution for sewage sludge.
Since March 2000, the ENCI cement plant located in Maastricht (Neth-
erlands) has been working together with the Limburg Purifcation Board,
receiving pre-treated sewage sludge from their sewage water treatment
plants (following further treatment in the Boards own thermal sludge
dryers). Today, 80.000 tonnes of dried sewage sludge are co-processed
annually in a kiln with a capacity of 865.000 tonnes of clinker per year
In 2005, the cement sector in Catalonia (Spain) reached an agreement
with the Catalan administration, trade unions and the local councils, to
launch a trial to monitor the environmental behaviour of thermally dried
sewage sludge from the Barcelona area as an alternative fuel in ce-
ment plants. The aim is to use more than 60.000 tonnes of dried sewage
sludge every year as a petcoke substitute, providing a solution for the
high amount of sewage sludge which cannot be used in agriculture.
Sewage sludge: a multifaceted solution
Emissions from the cement kiln come from the
physical and chemical reactions of the raw
materials and from the combustion of fuels.
The main constituents of the exit gases from
a cement kiln are nitrogen from the combus-
tion air, CO
2
from calcination and combustion,
water from the combustion process and the
raw materials, and excess oxygen. The exit
gases also contain small quantities of dust,
chlorides, uorides, sulphur dioxide, NO
x
,
carbon monoxide, and still smaller quantities
of organic compounds and heavy metals.
Impact of co-processing on kiln
emissions:
+ Sulphur oxides SO
2
: Alternative fuels have
no inuence on total SO
2
emissions.
+ Nitrogen oxides NO
x
: Alternative fuels do
not lead to higher NO
x
emissions in some
cases, NO
x
emissions can even be lower.
+ Total organic carbon TOC: there is no cor-
relation between the use of alternative fu-
els and emissions levels.
+ Polychlorinated dibenzo-p-dioxins and
polychlorinated dibenzofurans (PCDD/
PCDF): No difference has been found in di-
oxin emissions when alternative fuels are
used.
+ Hydrogen chloride HCI: HCl emissions
vary irrespective of the fuel used.
+ Hydrogen uoride HF: There is very little
difference in HF emissions when using al-
ternative fuels.
+ Heavy metals: Emissions vary irrespective
of the fuel used. However, nearly 100% of
them remain either in the cement clinker
matrix or the cement kiln dust as non leach-
able compounds. In any event, alternative
fuels undergo a rigorous acceptance and
inspection procedure before being used.
+ Dust: Dust emissions taken under both fuel
regimes indicate no difference between
the two.
In relation to the co-processing of alternative
fuels, the installations meet the same stand-
ards as dedicated waste incinerators.
THE USE OF SUITABLE
ALTERNATIVE MATERIALS
DOES NOT HAVE ANY
NEGATIVE IMPACT
ON EMISSIONS
9
All cement produced in Europe must meet
the applicable European standard, regard-
less of the materials used. Therefore, the ce-
ment industry carefully selects and monitors
all intake materials.
Depending on the amount of alternative raw
materials and fuels used, the concentra-
tion of individual elements in the nal prod-
uct may increase or decrease as a result of
waste processing. However, this is of little
consequence as these materials replace fos-
sil fuels and/or raw materials composed of
the same elements. As cement is blended
with aggregates, e.g. gravel and sand for
the production of concrete or mortar, it is
the behaviour of these trace elements in the
building material (concrete or mortar) which
is ultimately decisive for evaluating the en-
vironmentally relevant impacts of waste co-
processing in the clinker burning process on
the quality of the product.
Heavy metal releases from concrete and
mortar are low, as they remain rmly trapped
in the product. Independant tests on concrete
and mortar have shown that the leaching of
heavy metal concentrations is noticeably be-
low those prescribed, for instance, by nation-
al legislation. In addition, storage under dif-
ferent and partly extreme conditions has not
led to any environmentally relevant releases.
Impact on cement quality
Secil-Outo case study
Intertox performed an extensive risk
assessment of co-processing in Outo
(Portugal) to estimate the potential haz-
ards emissions may pose to both hu-
man health and the ecosystem. This
assessment was based on the worst
case scenario, simulating the effect of
cumulative less favorable occurrences,
in order to determine whether the emis-
sion levels would pose a risk to health
and the environment. It concluded that
co-processing is not expected to have
any signifcant impact on the health of
the local population or the local environ-
ment.
F More information: www.secil.pt
CO-PROCESSING IN THE CEMENT INDUSTRY
DOES NOT HAVE A NEGATIVE IMPACT
ON THE HEALTH & SAFETY OF ITS WORKERS
OR SURROUNDING NEIGHBOURHOOD
The European cement industry is committed to
the health & safety of its workers and neigh-
bours. As with all other fuels, raw materials and
components, the necessary health & safety
analyses have been conducted, with adequate
procedures put in place, including personal
protective equipment and storage facilities.
Co-processing and health & safety
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
: : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : : :
THE CO-PROCESSING OF WASTE HAS
NO NEGATIVE IMPACT ON THE
ENVIRONMENTAL AND TECHNICAL QUALITY
OF THE PRODUCT
10
Co-processing legislation & guidelines
Co-processing in the European cement is
strictly controlled. Guidelines, such as the
Guidelines for the Selection and Use of Fu-
els and Raw Materials in the Cement Manu-
facturing Process produced by the Cement
Sustainability Initiative, provide practical
guidance for cement manufacturers. In addi-
tion, legislation at European Union level also
applies to the use of AFR in the European ce-
ment industry.
+ Directive 2008/98/EC of the European Par-
liament and of the Council of 19 November
2008 on Waste and repealing certain Di-
rectives
+ Directive 2000/76/EC of the European Par-
liament and of the Council of 4 December
2000 on the Incineration of Waste.
+ Directive 2008/1/EC of the European Parlia-
ment and of the Council of 15 January 2008
concerning Integrated Pollution Prevention
and Control (Codied version) - Incorpo-
rates and repeals the Council Directive
96/61/EC of 24 September 1996 concerning
Integrated Pollution Prevention and Control
and all its amendments (Directive 2003/35/
EC (only Article 4 and Annex III), Directive
2003/87/EC (only Article 26), Regulation No
1882/2003 (only point (61) of Annex III) and
Regulation No 166/2006 (only Article 21(2))
into a single codied text.
+ Regulation (EC) No 1013/2006 of the Eu-
ropean Parliament and of the Council of
14 June 2006 on Shipments of Waste.
This information is correct at the time of publi-
cation. For the latest information, please visit:
www.cembureau.eu
The UKs Committee on the Medical Ef-
fects of Air Pollution, (COMEAP) has stat-
ed that the burning of waste-derived fuels
in cement kilns results in no changes in
emissions that would be of signifcance
to human health. Its latest (2008) report
on Meat & Bone Meal, Processed Sewage
Pellets, and Refuse Derived Fuel extended
the 2005 review of waste tyres and substi-
tute liquid fuels, and indicated that these
fndings were relevant to any combination
of these fuels.
F More information:
http://www.advisorybodies.doh.gov.uk/
comeap/
11
The European Cement Industry: an overview
Cement plays a key role in our lives: it is a basic
material for all types of construction, including
housing, roads, schools, hospitals, dams and
ports, and may even be used in decorative
items such as tables and bookcases.
Cement is a nely ground, non-metallic, inor-
ganic powder, which, when mixed with water,
forms a paste that sets and hardens. Its most
important use is in the production of concrete,
acting as the binder gluing the other key ingre-
dients of concrete sand and gravel. It typically
makes up about 12% of the concrete mix.
Concrete is the second most consumed sub-
stance on earth after water and is an essen-
tial, virtually irreplaceable product used in the
built environment. As a building material, it
contributes positively towards the energy ef-
ciency of buildings, and has excellent and
proven re resistance properties which deliver
protection of life, property and the environment
in the case of re. In relation to adaptation to
climate change, concrete also has much to of-
fer in terms of ood prevention and resilience
to adverse weather conditions.
CEMBUREAU CEMENT CONSUMPTION 2007
PER CAPITAL / KG
12%
88%
12%
88%
12
CEMBUREAU
WORLD CEMENT
PRODUCTION 2007
2.77 BILLION TONNES
CEMBUREAU, the European Cement
Association based in Brussels, is the
representative organisation of the ce-
ment industry in Europe. Currently, its
Full Members are the national cement
industry associations and cement
companies of the European Union
(with the exception of Cyprus, Malta
and Slovakia) plus Norway, Switzer-
land and Turkey. Croatia is an Associ-
ate Member of CEMBUREAU.
In 2007, CEMBUREAU Members pro-
duced a total of 325Mt of cement. The
average per capita consumption in
the CEMBUREAU member countries
in 2007 was 546kg. In the 27 Member
States of the European Union, produc-
tion reached an estimated 270Mt, ac-
counting for nearly 10% of world pro-
duction.
CEMENT
IN EUROPE
4
Rue dArlon 55,
BE-1040 Brussels,
Belgium
Tel: +32 2 234 10 45
www.cembureau.eu
Final January 2009
Design: GRAB IT
ACI Education Bulletin E4-03. Supersedes E4-96.
Copyright 2003, American Concrete Institute.
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 electronic or
mechanical device, printed, written, or oral, or recording for sound or visual reproduc-
tion or for use in any knowledge or retrieval system or device, unless permission in
writing is obtained from the copyright proprietors.

The Institute is not responsible for the statements or
opinions expressed in its publications. Institute publica-
tions are not able to, nor intended to, supplant individual
training, responsibility, or judgement of the user, or the
supplier, of the information presented.
E4-1
CHEMICAL ADMIXTURES FOR CONCRETE
Prepared under the direction and supervision of ACI Committee E-701
Materials for Concrete Construction
ACI Education Bulletin E4-03
This document discusses commonly used chemical admixtures for concrete
and describes the basic use of these admixtures. It is targeted at those in
the concrete industry not involved in determining the specific mixture
proportions of concrete or in measuring the properties of the concrete.
Students, craftsmen, inspectors, and contractors may find this a valuable
introduction to a complex topic. The document is not intended to be a state-
of-the-art report, users guide, or a technical discussion of past and present
research findings. More detailed information is available in ACI Committee
Report 212.3R, Chemical Admixtures for Concrete and 212.4R, Guide
for the Use of High-Range Water-Reducing Admixtures (Superplasticizers)
in Concrete.
CONTENTS
Chapter 1Introduction, p. E4-2
1.1History
1.2Definitions
Chapter 2Overview, p. E4-2
2.1Function
2.2Standards
Chapter 3Air-entraining admixtures, p. E4-3
3.1History
3.2Mechanism
3.3Use of air-entraining admixtures
Chapter 4Water-reducing and set-controlling
admixtures, p. E4-5
4.1Types and composition
4.2Type A, water-reducing admixtures
4.3Type B, retarding, and Type D, water-reducing and
retarding admixtures
4.4Type C, accelerating, and Type E, water-reducing
and accelerating admixtures
4.5High-range water-reducing admixtures
4.6Mid-range water-reducing admixtures
Chapter 5Corrosion-inhibiting admixtures, p. E4-9
Chapter 6Shrinkage-reducing admixtures, p. E4-9
Chapter 7Admixtures for controlling alkali-silica
reactivity, p. E4-9
Chapter 8Admixtures for underwater concreting,
p. E4-9
Chapter 9Effectiveness of admixtures, p. E4-9
Chapter 10Admixture dispensers, p. E4-10
10.1Industry requirements and dispensing methods
10.2Liquid admixture dispensing methods
10.3Accuracy requirements
10.4Application considerations and compatibility
10.5Dispensers for high-range water-reducing admixtures
10.6Dispenser maintenance
David M. Suchorski
Chair
James A. Farny
Secretary
Leonard W. Bell Tarek S. Khan Kenneth B. Rear
*
Richard P. Bohan Paul D. Krauss Raymundo Rivera-Villarreal
David A. Burg Colin L. Lobo Jere H. Rose
Darrell F. Elliot Stella L. Marusin Paul J. Tikalsky
James Ernzen Patrick L. McDowell Mark E. Vincent
J. Pablo Garcia Gerald R. Murphy Christopher H. Wright
Ramon F. Gutierrez Charles K. Nmai Kari L. Yuers
Morris Skip Huffman Anthony C. Powers Robert C. Zellers
Herb Johns
*
Chair of document subcommittee.
E4-2 ACI EDUCATION BULLETIN
Chapter 11Conclusion, p. E4-11
Chapter 12List of relevant ASTM standards,
p. E4-11
Chapter 13Glossary, p. E4-12
CHAPTER 1INTRODUCTION
1.1History
Admixtures have long been recognized as important com-
ponents of concrete used to improve its performance.
The original use of admixtures in cementitious mixtures is
not well documented. It is known that cement mixed with
organic matter was applied as a surface coat for water resis-
tance or tinting purposes. It would be a logical step to use
such materials, which imparted desired qualities to the sur-
face, as integral parts of the mixture. The use of natural
admixtures in concrete was a logical progression. Materials
used as admixtures included milk and lard by the Romans;
eggs during the middle ages in Europe; polished glutinous
rice paste, lacquer, tung oil, blackstrap molasses, and
extracts from elm soaked in water and boiled bananas by the
Chinese; and in Mesoamerica and Peru, cactus juice and
latex from rubber plants. The Mayans also used bark extracts
and other substances as set retarders to keep stucco workable
for a long period of time.
1.2Definitions
ACI 116R-00 defines the term admixture as a material
other than water, aggregates, hydraulic cement, and fiber rein-
forcement, used as an ingredient of a cementitious mixture to
modify its freshly mixed, setting, or hardened properties and
that is added to the batch before or during its mixing. In ACI
212.3R it is stated that chemical admixtures are used to
enhance the properties of concrete and mortar in the plastic
and hardened state. These properties may be modified to
increase compressive and flexural strength at all ages,
decrease permeability and improve durability, inhibit corro-
sion, reduce shrinkage, accelerate or retard initial set, increase
slump and workability, improve pumpability and finishability,
increase cement efficiency, and improve the economy of the
mixture. An admixture or combination of admixtures may be
the only feasible means of achieving the desired results. In cer-
tain instances, the desired objectives may be best achieved by
mixture changes in addition to proper admixture usage.
Chemical admixtures are materials that are added to the
constituents of a concrete mixture, in most cases, specified
as a volume in relation to the mass of the cement or total
cementitious materials. The admixtures interact with the
hydrating cementitious system by physical and chemical
actions, modifying one or more of the properties of concrete
in the fresh and/or hardened states.
Concrete is composed principally of aggregates, hydraulic
cement, and water, and may contain other cementitious
materials and chemical admixtures. It will contain some
amount of entrapped air and may also contain purposely
entrained air obtained by use of a chemical admixture or air-
entraining cement. Chemical admixtures are also frequently
used to accelerate, retard, improve workability, reduce mixing
water requirements, increase strength, improve durability, or
alter other properties of the concrete.
There are many kinds of chemical admixtures that can
function in a variety of ways to modify the chemical and
physical properties of concrete. This bulletin provides infor-
mation on the types of chemical admixtures and how they
affect the properties of concrete, mortar, and grout.
CHAPTER 2OVERVIEW
2.1Function
In ACI 212-3R, the reasons for the use of admixtures are
outlined by the following functions that they perform:
Increase workability without increasing water content
or decrease the water content at the same workability;
Retard or accelerate time of initial setting;
Reduce or prevent shrinkage or create slight expansion;
Modify the rate or capacity for bleeding;
Reduce segregation;
Improve pumpability;
Reduce rate of slump loss;
Retard or reduce heat evolution during early hardening;
Accelerate the rate of strength development at early ages;
Increase strength (compressive, tensile, or flexural);
Increase durability or resistance to severe conditions of
exposure, including application of deicing salts and
other chemicals;
Decrease permeability of concrete;
Control expansion caused by the reaction of alkalies
with potentially reactive aggregate constituents;
Increase bond of concrete to steel reinforcement;
Increase bond between existing and new concrete;
Improve impact and abrasion resistance;
Inhibit corrosion of embedded metal; and
Produce colored concrete or mortar.
2.2Standards
Air-Entraining Admixtures ASTM C 260
Standard Specification for Air-Entraining
Admixtures for Concrete AASHTO M 154
Standard Specification for Air-Entraining
Admixtures for Concrete CRD-C 13
Chemical Admixtures ASTM C 494
Standard Specification for Chemical
Admixtures for Concrete AASHTO M 194
Standard Specification for Chemical
Admixtures for Concrete CRD-C 87
Calcium Chloride ASTM D 98
Standard Specification for Calcium Chloride AASHTO M 144
Foaming Agents ASTM C 869
Admixtures for Shotcrete ASTM C 1141
Admixtures for Use in Producing
Flowing Concrete ASTM C 1017
Grout Fluidifier For Preplaced Aggregate
Concrete ASTM C 937
Pigments For Integrally Colored Concrete ASTM C 979
*
ASTMASTM International
AASHTOAmerican Association of State Highway and Transportation Officials
CRDArmy Corps of Engineers, Chief of Research and Development
CHEMICAL ADMIXTURES FOR CONCRETE E4-3
An admixture may conform to the requirements of one or
more of the above types, and may serve more than one of the
listed functions. Thus, combinations of two or more admix-
tures might be used in varying dosages so that optimum
results could be obtained using local materials. When using
combinations of admixtures, each admixture must be added
separately to the concrete mixture. Pre-job testing should be
conducted to ensure compatibility of the admixture system.
When using admixtures, and particularly combinations of
admixtures, testing for compatibility of the admixture sys-
tem requires knowledge of the rate of slump loss (that is the
relationship between slump and time) and the setting time
under relatively hot and cold conditions (in addition to labo-
ratory conditions). Test placements on-site are recom-
mended to verify proper workability, finishability, and
setting time of the proposed mixture.
CHAPTER 3AIR-ENTRAINING ADMIXTURES
3.1History
Air-entraining admixtures are primarily used to stabilize
tiny air bubbles in concrete, produced by mixing, and protect
against damage from repeated freezing-and-thawing cycles.
The dramatic effect of freezing and thawing on concrete is of
little surprise to those who live in climates with extensive
temperature cycling. Crumbling walls and scaled sidewalks
are evidence of the devastating effect that repeated exposure
to freezing and thawing can have on concrete proportioned
with an inadequate air content and bubble spacing or
improperly cured concrete.
During the l930s, certain concrete pavements were more
able to withstand the effects of freezing and thawing than
others. Investigation showed that the more durable pave-
ments were slightly less dense, and that the cement used had
been obtained from mills using beef tallow as a grinding aid
in the manufacturing of cement. The beef tallow acted as an
air-entraining agent, which improved the durability of the
pavements. Later, after rigorous investigation, air-entrained
concrete was specified where freezing-and-thawing resis-
tance was needed.
The incorporation of an adequate amount and distribution of
entrained air in properly proportioned concrete that contains
sound aggregates and is protected from cycles of freezing until
the compressive strength reaches about 28 MPa (4000 psi) can
render concrete resistant to freezing-and-thawing deterioration.
Until recently, the most commonly used air-entraining admix-
ture for concrete was a neutralized wood resin. Now, other for-
mulations that have some enhanced performance properties,
such as improved stability, have been introduced. Today, more
than 80% of the portland-cement concrete pavements in the
United States contain an air-entraining admixture to provide
resistance to freezing and thawing and impart better work-
ability, improved homogeneity, and decreased segregation
and bleeding.
3.2Mechanism
Entrained air should not be confused with entrapped air.
Air entrainment is usually the result of an addition of a liquid
admixture to the concrete during batching, but may be
obtained by using a cement blended with a powdered admix-
ture. As a result of the mixing action, these admixtures stabi-
lize air bubbles that become a component of the hardened
concrete. The resultant air-void system consists of uniformly
dispersed voids throughout the cement paste of the concrete.
These tiny voids (between 10 and 1000 micrometers in diam-
eter) must be present in the proper amount and spacing to be
effective at providing freezing-and-thawing protection. Con-
crete made with fine aggregate that is deficient in the smaller
particle sizes may benefit from air entrainment.
The space occupied by the mixing water in fresh concrete
rarely becomes completely filled with cementitious material
reaction product after the concrete has hardened. The
remaining spaces are capillary pores. Under saturated condi-
tions, these cavities are filled with water. If this water
freezes, the resulting expansion of water to ice creates tre-
mendous internal pressures. The expansion (approximately
9%) when water freezes produces a stress in a confined
space. This stress is far in excess of the tensile strength of
concrete. The result in non air-entrained concrete is crack-
ing, scaling, and spalling.
Entrained air voids make these capillaries discontinuous.
Because the air voids are generally much larger than the pas-
sageways, they form tiny reservoirs that act as safety valves
during ice expansion, accommodating the increased volume.
The importance is not only the amount of entrained air but
also the size and spacing of the bubbles. The level of air con-
tent recommended by ACI Committee 201 for normal-
strength concrete is listed in Table 1. It is based on different
exposure conditions and aggregate size. Adding air-entrain-
ment can also improve the finish of the surface of slabs and
reduce the occurrence of voids and sand streaking on wall
surfaces. Air entrainment, however, is not recommended for
interior steel troweled floors. Air content in excess of 3% can
cause blisters and delamination.
3.3Use of air-entraining admixtures
Air-entraining admixtures should be required to conform
to ASTM Specification C 260.
Air voids should not have a spacing factor larger than 0.2 mm
(0.008 in.) for adequate protection of water-saturated concrete
in a freezing-and-thawing environment. The term spacing
Table 1Recommended air contents for concrete
exposed to freezing and thawing (ACI 201.2R)
Nominal maximum
aggregate size, mm (in.)
Average air content, percent*
Severe exposure

Moderate exposure

9.5 (3/8) 7-1/2 6


12.5 (1/2) 7 5-1/2
19 (3/4) 6 5
25 (1) 6 4-1/2
37.5 (1-1/2) 5-1/2 4-1/2
*
A reasonable tolerance for air content if field construction is 1-1/2%.

Severe exposureOutdoor exposure in a cold climate where the concrete may be in


almost continuous contact with moisture before freezing and where deicing com-
pounds are used. Examples are pavements, bridge decks, and sidewalks.

Moderate exposureOutdoor exposure in a cold climate where the concrete will be


only occasionally exposed to moisture before freezing and where no deicing com-
pounds will be used. Examples are certain exterior walls, beams, bridge decks, and
slabs not in direct contact with soil.
E4-4 ACI EDUCATION BULLETIN
factor represents the maximum distance that water would
have to move before reaching the air-void reservoir or safety
valve.
Another factor that must be considered is the size of the air
voids. For a given air content, the size of the air voids cannot
be too large if the proper spacing factor is to be achieved with-
out using an unacceptable amount of air. The term specific
surface is used to indicate the average size of the air voids. It
represents the surface area of the air voids in concrete per unit
volume of air. For adequate resistance to repeated freezing and
thawing in a water-saturated environment, the specific surface
should be greater than 24 mm
2
/mm
3
(600 in.
2
/in.
3
).
With all the benefits a proper air-void system provides, there
may also be detrimental effects in concrete. Increasing the air
content will typically decrease the strength of concrete. An
increase of 1% in air content will typically decrease compres-
sive strength by about 5% in concrete mixtures with a compres-
sive strength in the range of 2l to 35 MPa (3000 to 5000 psi).
The air content of fresh concrete should be closely moni-
tored. Table 2 summarizes some of the factors that influence
the entrained air content of fresh concrete.
Additional dosage rates of the various air-entraining
admixtures generally range from about 15 to 130 mL per
100 kg (1/4 to 2 fl oz. per 100 lb) of cementitious material.
Equipment for dispensing air-entrainment admixtures is dis-
cussed in Chapter 9. The measurement of entrained air content
should be performed immediately before discharging the con-
crete into the forms. Samples for acceptance testing, however,
should be taken from the middle of the batch in accordance
with ASTM C 172. Unit weight should also be checked.
The methods and materials for performing air-content
tests on concrete are described in ASTM Standard Test
Methods C 231 and C 173. The gravimetric method (ASTM
C 138) is not generally used in the field because it requires
knowledge of the theoretical unit weight of the concrete on
an air-free basis. Unit weights, however, should be moni-
tored in the field to verify uniformity between batch mixture
proportions and air contents. Hardened cylinder weight
should be recorded on concrete test reports adjacent to com-
pressive strength. Cylinders should be weighed immediately
after demolding. Air content should be measured each time
concrete is sampled, and air meters should be calibrated reg-
ularly. Inexpensive devices for quickly calibrating air meters
are available. The Chace Air Indicator, a commonly misap-
plied device, does not provide the degree of accuracy and
precision necessary to measure air content.
For more than 20 years, testing agencies that record hardened
cylinder weights have been aware that measured air contents
may not be reasonable when compared with hardened cylinder
weight. Lower-than-expected strength is often associated
with a low hardened cylinder weight, which is not consistent
with the measured fresh air content. Subsequent petro-
graphic analysis of a companion cylinder often indicates an
air content significantly higher than the measured air con-
tent. The air content obtained from the petrographic analysis
adequately explains the lower-than-expected cylinder
strength, particularly if the air has coalesced around coarse
aggregate.
On-site control of air content of fresh concrete requires
coordination between the inspector-technician, the concrete
supplier, and the concrete contractor. Agreement on proce-
dures and timing of sampling should be made before the start
of concrete placement operations. All ingredients must be
added to the concrete before testing is initiated. A minimum
of 0.04 m
3
(1.0 ft
3
) of concrete from the middle of the batch
should be discharged into a suitable container, such as a
wheelbarrow or concrete buggy, in accordance with ASTM
C 172. The remainder of the testing techniques must be per-
Table 2Factors affecting the air content of concrete at a given dosage of admixture
Factor Affect on air content
Cement
An increase in the fineness of cement will decrease the air content.
As the alkali content of the cement increases, the air content may increase.
An increase in the amount of cementitious materials can decrease the air content.
Fine aggregate
An increase in the fine fraction passing the 150 m (No. 100) sieve will decrease the amount of entrained air.
An increase in the middle fractions passing the 1.18 mm (No. 16) sieve, but retained on the 600 m (No. 30) sieve and 300 m (No. 50)
sieve, will increase the air content.
Certain clays may make entraining air difficult.
Coarse aggregate
Dust on the coarse aggregate will decrease the air content.
Crushed stone concrete may result in lower air than a gravel concrete.
Water
Small quantities of household or industrial detergents contaminating the water may affect the amount of entrained air.
If hard water is used for batching, the air content may be reduced.
Pozzolans and slag Fly ash, silica fume, natural pozzolans, and ground granulated blast-furnace slag can affect the dosage rate of air-entraining admixtures.
Admixtures Chemical admixtures generally affect the dosage rate of air-entraining admixtures.
Slump
For less than a 75 mm (3 in.) slump, additional admixture may be needed. An increase in slump to about 150 mm (6 in.) will
increase the air content.
At slumps above 150 mm (6 in.), air may become less stable and the air content may decrease.
Temperature
An increase in concrete temperature will decrease the air content. Increase in temperature from 21 to 38 C (70 to 100 F) may
reduce air contents by 25%.
Reductions from 21 to 4 C (70 to 40 F) may increase air contents by as much as 40%. Dosages of air-entraining admixtures must
be adjusted when changes in concrete temperatures take place.
Concrete mixer
The amount of air entrained by any given mixer (stationary, paving, or transit) will decrease as the blades become worn or become
coated with hardened concrete buildup.
Air contents often increase during the first 70 revolutions of mixing then will hold for a short duration before decreasing. Air con-
tent will increase if the mixer is loaded to less than capacity and will decrease if the mixer is overloaded. In very small loads in a
drum mixer, however, air becomes more difficult to entrain.
CHEMICAL ADMIXTURES FOR CONCRETE E4-5
formed in strict accordance with ASTM test methods,
including proper remixing of the sample.
If the air content is outside specified limits, a retest should
be taken immediately. If the air content is found to be too low
or too high, the deficiency should be corrected in coordina-
tion with the producer, engineer, and testing agency.
Many factors are involved in the assurance of properly air-
entrained concrete. Improper concrete placement, consolida-
tion, and finishing techniques may decrease the air content.
The configuration of the boom on a pump may affect the air
content of the concrete. Tests have shown that there is often
more air loss when the boom is in a vertical position as
opposed to when the boom is extended in a more horizontal
configuration. Attention to proper selection of all materials
involved in the proportioning of the mixture is essential, as
compatibility problems may exist with other components of
the concrete mixture. Materials complying with relevant
specifications, including air-entraining admixtures meeting
ASTM C 260, and the adherence to proper proportioning
procedures is necessary. Proper testing according to standard
practices and proper placement and curing of fresh concrete
will make a major contribution to obtaining adequate dura-
bility and resistance to deterioration by freezing and thawing.
Careful consideration should be given to the need for air-
entrainment of steel-troweled slabs. Steel troweling of air-
entrained slabs can result in surface scaling. Maximum total
air content for interior steel-troweled slabs should normally
be 3% to reduce the possibility of scaling. If steel troweling
is required, it should be kept to a minimum. Overworking the
surface may decrease air entrainment at the surface where it
is needed. This also can result in sealing in a layer of water,
which will result in scaling.
CHAPTER 4WATER-REDUCING AND
SET-CONTROLLING ADMIXTURES
4.1Types and composition
Water-reducing, set-controlling admixtures are added to
concrete during mixing to increase workability, improve dura-
bility, provide easier placement, control the setting time, and
produce easier finishing with less segregation of the ingredi-
ents. This is accomplished while allowing a reduction of the
total water content and providing the ability to control the time
of setting to meet changing jobsite and climatic conditions.
The strength improvement resulting from water-reducing
admixtures is primarily a result of reducing the water-
cementitious materials ratio and increasing cement effi-
ciency. For a given air content, concrete strength is inversely
proportional to the water-cementitious materials ratio and,
therefore, the reduction in water needed to achieve the
desired slump and workability when a water-reducing agent
is used will effect an increase in strength. The result of water-
reducing admixtures in improving strength, however, often
exceeds the results of simply reducing the water content.
Proper use of admixtures should begin by gathering avail-
able information and comparing the different types and
brands that are available. Trial mixtures, with those admix-
tures under consideration, should be made to determine their
effect on strength, finishability, and other construction
requirements, such as rate of slump loss and setting time.
Consideration must be given to information such as uniformity,
dispensing, long-term performance, and available service.
These are points that cannot be assessed by concrete tests but
could determine successful admixture use.
The admixture manufacturer should be able to provide
information covering typical dosage rates, times of setting,
and strength gain for local materials and conditions. The
evaluation and application of the admixture should be made
with specific job materials using the construction procedures
under anticipated ambient conditions. Laboratory tests con-
ducted on concrete with water-reducing admixtures should
indicate the effect on pertinent properties necessary for the
construction project, including: water requirement, air con-
tent, slump, rate of slump loss, bleeding, time of setting,
compressive strength, flexural strength, and resistance to
freezing and thawing. Following the laboratory tests, field
test should be conducted to fully comprehend how the
admixtures will work in actual field conditions.
ASTM C 494, Standard Specification for Chemical
Admixtures for Concrete, classifies admixtures into seven
types as follows:
Type A Water-reducing admixtures;
Type B Retarding admixtures;
Type C Accelerating admixtures;
Type D Water-reducing and retarding admixture;
Type E Water-reducing and accelerating admixtures;
Type F Water-reducing, high-range, admixtures; and
Type G Water-reducing, high-range, and retarding
admixtures.
Each of the seven types of admixtures covered by ASTM C
494 is designed to function in a specific manner. ASTM C 494
outlines the physical requirements for performance of the
potential admixture to be qualified in the respective categories.
To be classified as a Type A (water-reducing) admixture,
a minimum water reduction of 5% must be obtained. Initial
and final time of setting must be no more than 1 h earlier and
not more than an 1.5 h later than the same concrete without
the admixture. Compressive strength requirements stated as
a percentage of reference are outlined at the various intervals
specified for testing. These are stated for both compressive
and flexural requirements. Drying shrinkage and freezing-
and-thawing resistance are also factors addressed for all
types of admixtures in ASTM C 494.
The requirements for compressive strength compared to a
control mixture allow no reduction of compressive or flex-
ural strengths for all types except B and C.
In discussing the commercially available water-reducing
set-controlling admixtures, it is appropriate to consider five
classes of admixtures. Categorized by basic or primary
ingredients, they are as follows:
1. Lignosulfonic acids and their salts;
2. Hydroxylated polymers;
3. Hydroxylated carboxylic acids and their salts;
4. Sulfonated melamine or naphthalene formaldehyde
condensates; and
E4-6 ACI EDUCATION BULLETIN
5. Polyether-polycarboxylates.
Persons involved in concrete construction should possess
a basic knowledge of proper use, application, benefits to
expect, and cautions to observe for each of these five classes.
Members of the first group, salts of lignosulfonic acids or
lignins, provide excellent water reduction and produce good
strength characteristics. Lignosulfonates tend to entrain air,
and can produce sticky, hard-to-finish concrete.
Hydroxylated polymer admixtures are widely used water-
reducing admixtures. Some of the benefits associated with
these admixtures are:
1. Improved mobility of the concrete. Improved workability
and easier placement, reducing the incentive to add water;
2. May reduce segregation, particularly at higher slump
ranges;
3. May improve finished appearance and reduce stickiness in
finishing flat surfaces, even when used at increased dosages;
4. Does not entrain air and can be used at increased dos-
ages without increasing air contents; and
5. May improve pumpability. This admixture, in a prop-
erly proportioned mixture, allows the concrete to be moved
with less pressure and provides improved lubrication.
Hydroxylated carboxylic acid-based admixtures, some-
times referred to as HC type admixtures, are designed to
reduce water also. Concrete containing HC admixtures is
mixed and placed with a higher water content than is the case
with other types of water-reducing admixtures. Some HC
admixtures promote rapid bleeding from the interior of the
concrete. This water must be removed from the surface and
the concrete revibrated to ensure proper density in the hardened
state. The benefits associated with this admixture type are
generally similar to those of the hydroxylated polymer
admixtures.
The fourth and fifth groups, sulfonated melamine or naph-
thalene formaldehyde condensates and polycarboxylates, are
discussed in Section 4.5.
4.2Type A, water-reducing admixtures
The purpose of water-reducing admixtures is stated by
ACI Committee 212.3R as: Water-reducing admixtures are
used to reduce the water requirement of the mixture for a
given slump, produce concrete of higher strength, obtain
specified strength at lower cement content, or increase the
slump of a given mixture without an increase in water con-
tent. They also may improve the properties of concrete con-
taining aggregates that are harsh, poorly graded, or both, or
may be used in concrete that may be placed under difficult
conditions. They are useful when placing concrete by means
of a pump or tremie.
Typically, the use of Type A water-reducers will decrease
mixing water content by 5 to 12%, depending on the admix-
ture, dosage, and other materials and proportions. Dosage
rates of water-reducing admixtures depend on the type and
amount of active ingredients in the admixture (that is, per-
cent solids content). The dosage is based on the cementitious
materials content of the concrete mixture and is expressed as
milliliters per hundred kilograms (fluid ounces per hundred
pounds) of cementitious materials. Typically the dosage rate
of Type A water-reducers range from 130 to 390 mL per 100 kg
(2 to 6 fl oz. per 100 lb) of cementitious materials. Higher
dosages may result in excessive retardation of the concrete
setting time. Manufacturers recommended dosage rates
should be followed and trial batches with local materials
should be performed to determine the dosage rate for a given
concrete mixture. Usually, the primary ingredients of all
water-reducing admixtures are organic, which tend to retard
the time of setting of the concrete. This retardation may be
offset by small additions of chloride or nonchloride acceler-
ating admixtures at the batch plant. Typically, Type A admix-
tures already contain some accelerators that offset this natural
retardation. Care should be taken to ensure that addition of
chloride does not exceed the ACI 318 limits for maximum
chloride-ion content in reinforced or prestressed concrete.
4.3Type B, retarding, and Type D, water-reducing
and retarding admixtures
4.3.1 Conventional retarding admixturesThese two
types of admixtures are used for the same basic purpose: to
offset unwanted effects of high temperature, such as acceler-
ation of set and reduction of 28-day compressive strength,
and to keep concrete workable during the entire placing and
consolidation period. Figure 1 indicates the relationship
between temperature and setting time of concrete and specif-
ically indicates why retarding admixture formulations are
needed in warmer weather.
The benefits derived from retarding formulations include
the following:
1. Permits greater flexibility in extending the time of set
and the prevention of cold joints;
2. Facilitates finishing in hot weather; and
3. Permits full form deflection before initial set of concrete.
As with Type A admixtures, their dosage rates are based
on the amount of cementitious materials in the concrete mix-
ture. While both Type B and Type D provide some water-
reduction, Type D is more effective in achieving this goal.
The amount of retardation depends upon many factors
including: admixture concentration, dosage rate, concrete
proportions, and ambient and concrete temperatures.
Different sources and types of cement or different lots of
cements from the same source may require different amounts
Fig. 1Relationship between temperature and setting time
of concrete.
CHEMICAL ADMIXTURES FOR CONCRETE E4-7
of the admixture to obtain the desired results because of vari-
ations in chemical composition, fineness, or both. The effec-
tiveness of the admixture seems to be related primarily to the
amount of tricalcium aluminate (C
3
A) and the alkali (Na
2
O
and K
2
O) content of the cement.
The time at which the retarding admixture is introduced
into the concrete may affect the results. Allowing the cement
to become totally wet and delaying admixture addition until
all other materials are batched and mixed may result in
increased retardation and greater slump increase.
Increased retardation may also be obtained with a higher
dosage of the retarding admixture. When high dosages of
retarding admixture are used, however, rapid stiffening can
occur with some cements, resulting in severe slump loss and
difficulties in concrete placement, consolidation, and finishing.
4.3.2 Extended-set admixturesRecent advances in
admixture technology have resulted in the development of
highly potent retarders called extended-set admixtures,
which are capable of stopping the hydration of portland
cement, thereby providing a means to control the hydration
and setting characteristics of concrete. The effectiveness of
extended-set admixtures has been attributed to their ability to
retard the reaction of all the major cement constituents,
unlike conventional retarding admixtures that only act upon
some of the cement constituents.
Extended-set admixtures are used in three primary appli-
cations: stabilization of concrete wash water, stabilization of
returned plastic concrete, and stabilization of freshly batched
concrete for long hauls. The use of extended-set admixtures
in stabilization of concrete wash water eliminates the dumping
of water that is used to wash out a ready-mixed concrete
truck drum while keeping the fins and inner drum clean. The
process is relatively simple and involves the addition of low
dosages of the extended-set admixture to the wash water to
control the hydration of concrete stuck to the fins and inside
the drum. The stabilized wash water may be included in the
mixing water for fresh concrete that is batched the next day
or after a weekend. The setting and strength development
characteristics of concrete are not adversely affected by the
use of stabilized wash water.
The use of extended-set admixtures to stabilize returned
unhardened concrete has made it possible to reuse such con-
crete during the same production day or the next day in lieu
of disposal. The dosage of extended-set admixture required
depends on several factors that include the ambient and con-
crete temperatures, the ingredients used in the manufacture
of the concrete, and the age of the concrete. Stabilized con-
crete is reused by batching fresh concrete on top of the stabi-
lized concrete. In overnight applications, an accelerating
admixture may be used to reinitiate the hydration process
before adding fresh concrete. Increasingly, extended-set
admixtures are being used for long hauls and to maintain
slump and concrete temperature during transit, especially in
warm weather. For this application, the extended-set admix-
ture is added during or immediately after batching, and the
required dosage is established based on the amount of retar-
dation desired.
4.4Type C, accelerating, and Type E,
water-reducing and accelerating admixtures
Accelerating admixtures are added to concrete to shorten
the setting time and accelerate the early strength develop-
ment of concrete. Figure 1, which shows the relationship
between temperature and setting time of concrete, specifi-
cally indicates why accelerating admixture formulations are
needed.
Some widely used and effective chemicals that accelerate
the rate of hardening of concrete mixtures, including calcium
chloride, other chlorides, triethanolamine, silicates, fluo-
rides, alkali hydroxide, nitrites, nitrates, formates, bromides,
and thiocyanates.
The earlier setting time and increased early strength gain
of concrete brought about by an accelerating admixture will
result in a number of benefits, including reduced bleeding,
earlier finishing, improved protection against early exposure
to freezing and thawing, earlier use of structure, and reduc-
tion of protection time to achieve a given quality. Accelera-
tors do not act as anti-freeze agents; therefore, protection of
the concrete at early ages is required when freezing temper-
atures are expected.
Although calcium chloride is the most effective and eco-
nomical accelerator for concrete, its potential to cause corro-
sion of reinforcing steel limits its use. ACI Committee 318
suggests that the water-soluble chloride-ion content should
be limited to the following levels for the conditions
described:
1. Prestressed concrete0.06% by mass of cementitious
material; and
2. Reinforced concrete0.15% by mass of cementitious
material.
Note that the amount of calcium chloride that may be used
is based on the cement content of the concrete mixture.
The following guidelines should be considered before
using calcium chloride or chloride-bearing admixture:
1. It should not be used in prestressed concrete because of
its potential for causing corrosion;
2. The presence of chloride ion has been associated with
corrosion of galvanized steel such as when this material is
used as permanent forms for roof decks;
3. Where sulfate-resisting concrete is required, calcium
chloride should not be used;
4. Calcium chloride should be avoided in reinforced con-
crete in a moist condition. In non-reinforced concrete, the
level of calcium chloride used should not exceed 2% by
weight of cementitious material;
5. Calcium chloride should be dissolved in a portion of
mixing water before batching because undissolved lumps
may later disfigure concrete surfaces;
6. Calcium chloride precipitates most air-entraining agents
so it must be dispensed separately into the mixture; and
7. Field experience and laboratory tests have demonstrated
that the use of uncoated aluminum conduit in reinforced con-
crete containing 1% or more of calcium chloride may lead to
sufficient corrosion of the aluminum to collapse the conduit
or crack the concrete.
E4-8 ACI EDUCATION BULLETIN
Non-chloride accelerating admixtures containing salts of
formates, nitrates, nitrites, and thiocyanates are available
from admixture manufacturers. These nonchloride accelera-
tors are effective for set acceleration and strength develop-
ment: however, the degree of effectiveness of some of these
admixtures is dependent on the ambient temperature and
concrete temperature at the time of placement.
Some formulations will give protection against freezing to
concrete placed in sub-freezing ambient temperatures. These
non-chloride accelerating admixtures offer year-round ver-
satility because they are available to be used for acceleration
purposes in cool weather and for sub-freezing protection.
The role water-reducing set-controlling admixtures play in
achieving control of concrete quality continues to grow as
the admixtures are improved. They are used in all types of
concrete construction to achieve a wide range of benefits.
4.5High-range water-reducing admixtures
The primary difference between these admixtures and
conventional water-reducing admixtures is that high-range
water-reducing (HRWR) admixtures, often referred to as
superplasticizers, may reduce the water requirement by more
than 30%, without the side effect of excessive retardation.
By varying the dosage rate and the amount of mixing water,
an HRWR admixture can be used to produce:
1. Concrete of normal workability at a lower water-cemen-
titious material (w/cm) ratio;
2. Highly flowable, nearly self-leveling concrete at the
same or lower w/cm as concrete of normal workability; and
3. A combination of the two; that is, concrete of moder-
ately increased workability with a reduction in the w/cm.
When used for the purpose of producing flowing concrete,
HRWR admixtures facilitate concrete placement and consol-
idation.
HRWR admixtures should meet the requirements of ASTM C
494 for classification as Type F, High-Range Water-Reducing,
or Type G, High-Range Water-Reducing and Retarding,
admixtures. When used to produce flowing concrete, they
should also meet the requirements of ASTM C 1017 Type 1,
Plasticizing, or Type 2, Plasticizing and Retarding Admix-
tures. HRWR admixtures are organic products that typically
fall into three families based on ingredients:
1. Sulfonated melamine-formaldehyde condensate;
2. Sulfonated naphthalene-formaldehyde condensate; and
3. Polyether-polycarboxylates.
HRWR admixtures act in a manner similar to conventional
water-reducing admixtures, except that they are more efficient
at dispersing fine-grained materials such as cement, fly ash,
ground granulated blast-furnace slag, and silica fume. The
most widely used HRWR admixtures do not entrain air but
may alter the air-void system. Concrete containing HRWR
admixtures, however, may have adequate resistance to freez-
ing and thawing even though the spacing factors may be
greater than 0.2 mm (0.008 in.). HRWR admixtures based on
polyether-polycarboxylate technology are different chemi-
cally and more effective than those based on sulfonated
melamine-formaldehyde and sulfonated naphthalene-form-
aldehyde condensates and, as a result, are typically added at
the batch plant. Polyether-polycarboxylate HRWRs also
retard less and develop strength faster compared to the other
HRWR formulations. Because of their increased efficiency,
polyether-polycarboxylate HRWRs are gaining widespread
acceptance, particularly in precast concrete applications and
in making self-consolidating concrete, a high-performance
concrete with high flowability that requires minimal or no
vibration for consolidation.
A characteristic of some HRWR admixtures is that their
slump-increasing effect is retained in concrete for only 30 to
60 min, by which time the concrete will revert to its original
slump. The amount of time that the concrete retains the
increased slump is dependent upon the type and quantity of
cement, the temperature of the concrete, the type of HRWR
admixture, the dosage rate used, the initial slump of the con-
crete, the mixing time, and the thoroughness of mixing.
Because of the limited workability time, HRWR admixtures
are typically added at the jobsite. With some HRWR admix-
tures, it is possible to redose the concrete to regain the
increased workability. Generally, the strength is increased
and the air content is decreased. HRWR admixtures that offer
extended slump life are also commercially available. These
HRWR admixtures are typically added at the batch plant.
HRWR admixtures can be used with conventional water-
reducers or retarders to reduce slump loss and stickiness, espe-
cially in silica-fume concrete mixtures. Because a lower dos-
age of HRWR admixture may be required in such instances,
there may be some savings. These combinations of admixtures
may also cause unanticipated or excessive set retardation.
The strength of hardened concrete containing HRWR
admixtures is normally higher than that predicted by the
lower w/cm alone. As with conventional admixtures, this is
believed to be due to the dispersing effect of HRWR admix-
tures on the cement and other cementitious or pozzolanic
materials. Because the w/cm of mixtures containing HRWR
admixtures are typically low, shrinkage and permeability
may also be reduced and the overall durability of the concrete
may be increased.
A good summary of benefits and limitations for this class
of admixtures can be found in National Ready Mixed Con-
crete Association (NRMCA) Publication No. 158. Briefly
outlined are eight advantages and six limitations as follows:
POTENTIAL ADVANTAGES OF HRWR:
1. Significant water reduction;
2. Reduced cement contents;
3. Increased workability;
4. Reduced effort required for placement;
5. More effective use of cement;
6. More rapid rate of early strength development;
7. Increased long-term strength; and
8. Reduced permeability.
POTENTIAL DISADVANTAGES OF HRWR:
1. Additional admixture cost (the concrete in-place cost
may be reduced);
2. Slump loss greater than conventional concrete;
3. Modification of air-entraining admixture dosage;
CHEMICAL ADMIXTURES FOR CONCRETE E4-9
4. Less responsive with some cement;
5. Mild discoloration of light-colored concrete; and
6. Air-void and color blemishes on exposed and formed
finishes.
Therefore, a fundamental knowledge regarding the action
of base materials will greatly assist in choosing the appropri-
ate admixture. The choice of specific functional type will
vary with jobsite conditions.
4.6Mid-range water-reducing admixtures
Water-reducing admixtures that provide moderate water
reduction without significantly delaying the setting charac-
teristics of concrete are also available. Because these ad-
mixtures provide more water reduction than conventional
water-reducers but less water-reduction than high range water-
reducers, they are referred to as mid-range water-reducing
admixtures. These admixtures can help reduce stickiness and
improve finishability and pumpability of concrete including
concrete containing silica fume, or manufactured or coarse
sand. Mid-range water-reducing admixtures are typically
used in a slump range of 125 to 200 mm (5 to 8 in.) and may
entrain additional air. Therefore, evaluations should be per-
formed to establish air-entraining admixture dosage for a
desired air content.
CHAPTER 5CORROSION-INHIBITING
ADMIXTURES
Reinforcing steel corrosion is a major concern with regard
to the durability of reinforced concrete structures. Each year,
numerous bridges and parking garage structures undergo
extensive rehabilitation to restore their structural integrity as
a result of corrosion damage. In addition to bridges and parking
structures, other reinforced concrete structures exposed to
chlorides in service are also at risk of corrosion attack. Chlo-
rides are one of the causes of corrosion of steel in concrete.
They can be introduced into concrete from deicing salts that
are used in the winter months to melt snow or ice, from sea-
water, or from the concrete mixture ingredients.
There are several ways of combating chloride-induced
corrosion, one of which is the use of corrosion-inhibiting
admixtures. These admixtures are added to concrete during
batching and they protect embedded reinforcement by delaying
the onset of corrosion and also reducing the rate of corrosion
after initiation. There are several commercially available
inhibitors on the market. These include an inorganic formu-
lation that contains calcium nitrite as the active ingredient
and organic formulations consisting of amines and esters. As
with all admixtures, the manufacturers recommendations
should be followed with regard to dosage.
CHAPTER 6SHRINKAGE-REDUCING
ADMIXTURES
The loss of moisture from the concrete as it dries results in
a volume contraction termed drying shrinkage. Drying
shrinkage tends to be undesirable when it leads to cracking
due to either internal or external restraint, curling of floor
slabs, and excessive loss of prestress in prestressed concrete
applications. The magnitude of drying shrinkage can be
reduced by minimizing the unit water content of a concrete
mixture, and using good-quality aggregates and the largest
coarse-aggregate size and content consistent with the partic-
ular application. Drying shrinkage can also be reduced sig-
nificantly by using shrinkage-reducing admixtures. These
are organic-based formulations that reduce the surface ten-
sion of water in the capillary pores of concrete, thereby
reducing the tension forces within the concrete matrix that
lead to drying shrinkage. Manufacturers recommendations
should be followed with regard to dosage and suitability for
use in freezing-and-thawing environments.
CHAPTER 7ADMIXTURES FOR CONTROLLING
ALKALI-SILICA REACTIVITY
Alkali-silica reactivity (ASR) is a reaction between soluble
alkalies in concrete and reactive silica in certain types of
aggregate that results in the formation of a water-absorptive
gel that expands and fractures the concrete. The reaction is
typically slow and is dependent on the total amount of alkali
present in the concrete, the reactivity of the aggregates and
the availability of moisture. ASR can be mitigated by using
low-alkali cement, sufficient amounts of pozzolans or
ground granulated blast-furnace slag, and if economically
feasible, non-reactive aggregates. Alternately, ASR can be
mitigated by using lithium-based chemical admixtures. Lithium
compounds are effective in reducing ASR because they form
a nonabsorptive gel with the reactive silica in the aggregates.
The high cost of lithium-based admixtures, however, has
greatly limited their use to date.
CHAPTER 8ADMIXTURES FOR UNDERWATER
CONCRETING
Placing concrete underwater can be particularly challenging
because of the potential for washout of the cement and fines
from the mixture reducing the strength and integrity of the
in-place concrete. Although placement techniques, such as
tremies, have been used successfully to place concrete
underwater, there are situations where enhanced cohesive-
ness of the concrete mixture is required, necessitating the use
of an antiwashout or viscosity-modifying admixture (VMA).
Some of these admixtures are formulated from either cellu-
lose ether or whelan gum, and they work simply by binding
excess water in the concrete mixture, thereby increasing the
cohesiveness and viscosity of the concrete. The overall ben-
efit is a reduction in washout of cement and fines, resistance
to dilution with water as the mixture is placed, and preserva-
tion of the integrity of the in-place concrete. Proper place-
ment techniques should be followed even with concrete
treated with an antiwashout admixture.
CHAPTER 9EFFECTIVENESS OF ADMIXTURES
The effectiveness of any admixture will vary depending on
its concentration in the concrete and the effect of the various
constituents of the concrete mixture, particularly the cement.
Each class of admixture is defined by its primary function. It
may have one or more secondary functions, however, and its
use may affect, positively or negatively, concrete properties
other than those desired. Therefore, adequate testing should
be performed to determine the effects of an admixture on the
plastic and hardened properties of concrete such as slump,
E4-10 ACI EDUCATION BULLETIN
air content, time of setting, rate of slump loss or relationship
of slump and time, and strength development. In addition,
testing should be performed to determine the effect of the
admixture on other concrete properties that may be of interest,
for example, drying shrinkage, modulus of elasticity, or
permeability. The final decision as to the use of any admix-
ture and the brand, class, or type, depends on its ability to
meet or enhance specific concrete performance needs.
Many of the improvements can be achieved by proper selec-
tion and application of specific admixtures. The selection pro-
cess should focus on the functional qualities required by
structural demands, architectural requirements, and contractor
needs.
Whatever the approach, be it a single water-reducing
admixture or a combination approach, the use of admixtures
can be beneficial. Admixtures provide additional means of
controlling the quality of concrete by modifying some of its
properties, however, they cannot correct for poor-quality
materials, improper proportioning of the concrete, and inap-
propriate placement procedures.
CHAPTER 10ADMIXTURE DISPENSERS
10.1Industry requirements and dispensing
methods
The subject of liquid admixture dispensers covers the
entire process from storage at the producers plant to intro-
duction into the concrete batch before discharge. Their oper-
ation may be separated into four functions:
1. The dispenser transports the admixture from storage to
the batch;
2. The dispenser measures the quantity of admixture
required;
3. The dispenser provides verification of the volume dis-
pensed; and
4. The dispenser injects the admixture into or onto the
batch.
These are the basic functions. In practice, some of the
functions may be combined, for example, measurement and
verification. For reliability, the functions may be interlocked
to prevent false or inaccurate batching of the admixture and
to dispense the admixture in the optimal sequence in the con-
crete production process.
The various systems of dispensing, their applications to
specific types of concrete production, and the practical limi-
tations of their operation and accuracy are the subjects that
will be discussed in this section. They are important because
the successful use of any chemical admixture stands on accu-
rate measurement and correct addition of the material to the
concrete batch.
10.2Liquid admixture dispensing methods
The three most commonly used dispenser systems at
ready-mix plants are Systems 1, 2, and 3.
System 1System 1 is a fully automatic dispenser system
for interfacing with the batch plants automation. It is
designed for the ready-mix plant operation, which already
has admixture dispensing control capabilities built into its
computerized batch control panel. These systems are capable
of controlling multiple admixtures at the same time with var-
ious interlocks for system compliance to regulatory require-
ments.
The basic components of this dispenser system includes:
A metering device for volumetric measurement;
A measuring unit for visual verification of admixture
being dispensing;
Air-/electric-operated valves for automatic control of the
flow of admixture in and out of the measuring unit; and
A storage tank with fill adapters for connection to the
admixture pump.
An interface sub-junction box and cabling for connection
to the computerized batch control panel and dispenser sys-
tem is used. This system may come preassembled in its own
protective enclosure or may be assembled at the plant in a
protected area.
System 2System 2 is designed for the ready-mix plant
operation that does not have admixture dispenser control
capabilities built into its batch panel automation and, there-
fore, will require a stand-alone admixture batch-control unit
to automatically and simultaneously control the dispensing
of multiple admixtures. These units often have the capability
of being remote started by a signal provided through the
plants automation system.
The basic components of this dispenser system include:
A metering device for volumetric measurement;
A measuring unit for visual verification of admixture
being dispensing;
Air-operated valves for automatic control of the flow of
admixture in and out of the measuring unit;
A storage tank with fill adapters for connection to the
admixture pump; and
A stand-alone control unit.
This system may come preassembled in its own protective
enclosure or may be assembled at the plant in a protected area.
System 3System 3 is a manual dispenser system for
ready-mix plants that are not automated. The operator con-
trols the quantity of admixture requirements by means of a
manual three-position pneumatic valve and visually verifies
the correct amount of admixture before dispensing.
The basic components of this dispenser system includes:
A measuring unit for manual verification of the amount
of admixture being dispensed;
Air-operated valves for manual control of the flow of
admixture in and out of the measuring unit;
A storage tank with fill adapters for connection to a
pump; and
A three-position pneumatic valve and miscellaneous
fittings that will be located in the batch control room of
the plant.
This system in most cases is assembled at the plant in a
protected area.
10.3Accuracy requirements
Standards of operation for admixture dispensers are specified
by scientific groups, concrete producers trade organiza-
tions, and government agencies with authority over concrete
production contracts.
CHEMICAL ADMIXTURES FOR CONCRETE E4-11
The NRMCA and ASTM C 94 specify a batching toler-
ance of 3% of the required volume or the minimum recom-
mended dosage rate per unit of cement, whichever is greater.
(1.3.5.3) ACI 212.1R recommends an accuracy of 3% of the
required volume, or 15 mL (1/2 fl oz.), whichever is greater.
10.4Application considerations and
compatibility
Admixture dispensing systems are complex, using parts
made of different materials. Therefore, the admixture dis-
pensed through this system should be chemically and opera-
tionally compatible with these materials.
The basic rules of application and injection are that the
admixtures should not be mixed together. This problem is
handled in several ways:
1. Injecting admixtures into the waterline at separate
points at least 3 ft apart and only when the water is running;
2. Placing the air-entraining admixture on the fine aggre-
gate and injecting the water-reducer into the stationary or
truck mixer along with water; and
3. Sequentially discharging the admixtures. The air-entraining
admixture is discharged first; and the water-reducer, or combi-
nation of water-reducers, is discharged later.
Recommended injection sequences for various admixtures
are as follows:
Generally, it is not necessary to distribute the admixtures
throughout the entire water batch to get good dispersion in
the mixture.
There is evidence that the timing of injection of water-
reducing retarders has important effects on the length of
retardation and, to a lesser extent, the slump and air content.
A delay of 1 to 5 min between the water addition and dispers-
ing of the retarder may result in a three-fold increase in set
retardation time with lignin and polymer retarding admix-
tures, a one and one-half to two-fold increase in entrained
air, and lesser increases in slump.
10.5Dispensers for high-range water-reducing
admixtures
Some high-range water-reducing admixtures have a short-
lived effect on the slump of concrete. Therefore, it is
expected that these materials will be dispensed as close to
placing time as possible. For ready-mix operations, this
might mean the use of a truck-mounted dispenser in the form
of a calibrated storage tank. The tank will be charged with
the admixture at the same time the concrete is loaded. The
user can request an increase in slump by injection of a
HRWR admixture, and the driver will dispense the required
amount into the turning drum. The volume dispensed will
be recorded on the delivery ticket. The injection should be
performed under pressure through a spray nozzle to thoroughly
disperse the admixture into the drum. Field dispensers,
consisting of a measuring unit and pump, can be used at the
job site.
10.6Dispenser maintenance
It is incumbent on the concrete producer to take as great an
interest in the admixture dispensing equipment as in the rest
of the batch plant. Operating personnel should be trained in
the proper operation, winterization, maintenance, and cali-
bration of admixture dispensers. Spare parts should be
retained as needed for immediate repairs. Regular cleaning
and calibration of the systems should be performed by qualified
internal personnel or by the admixture suppliers service
representative. Admixtures have too powerful an influence
on the quality of the concrete produced for their dispensing
to be given cursory attention.
CHAPTER 11CONCLUSION
Air-entraining and other chemical admixtures have
become a very useful and integral component of concrete.
Admixtures are not a panacea for every ill the concrete pro-
ducer, architect, engineer, owner, or contractor faces when
dealing with the many variables of concrete, but they do offer
significant improvements in both the plastic and hardened state
to all concrete. Continued research and development will
provide additional reliability, economy, and performance for
the next generation of quality concrete.
CHAPTER 12LIST OF RELEVANT ASTM
STANDARDS
C 94 Ready-Mixed Concrete
C 138 Unit Weight, Yield, and Air Content (Gravimetric)
of Concrete
C 143 Slump of Hydraulic-Cement Concrete
C 150 Portland Cement
C 173 Air Content of Freshly Mixed Concrete by the
Volumetric Method
C 231 Air Content of Freshly Mixed Concrete by the
Pressure Method
C 260 Air-Entraining Admixtures
C 494 Chemical Admixtures
C 869 Foaming Agents Used in Making Preformed
Foam for Cellular Concrete
C 937 Grout Fluidifier for Preplaced Aggregate Concrete
C 979 Pigments for Integrally Colored Concrete
C 1012 Length Change of Hydraulic-Cement Mortars
Exposed to a Sulfate Solution
C 1017 Chemical Admixtures for Use in Producing Flowing
Concrete
C 1144 Admixtures for Shotcrete
C 1157 Hydraulic Cements
D 98 Calcium Chloride
ADMIXTURES INJECTION SEQUENCE
Air-entraining admixture With early water or on sand
Water-reducing admixtures Follow air-entraining solution
Accelerating admixtures
With water, do not mix with air-entraining
admixture
High-range water-reducing
admixtures
Immediately before discharge for placement
or with the last portion of the water at the
batch plant
Polycarboxylate high-range
water-reducing admixtures
With early water or with the last portion of the
water at the batch plant
E4-12 ACI EDUCATION BULLETIN
CHAPTER 13GLOSSARY
AdmixtureA material other than water, aggregates,
hydraulic cement, and fiber reinforcement, used as an ingre-
dient of a cementitious mixture to modify its freshly mixed,
setting, or hardened properties and that is added to the batch
before or during its mixing.
Admixture, acceleratingAn admixture that causes an
increase in the rate of hydration of the hydraulic cement, and
thus, shortens the time of setting, increases the rate of
strength development, or both.
Admixture, air-entrainingAn admixture that causes
the development of a system of microscopic air bubbles in
concrete, mortar, or cement paste during mixing.
Admixture, retardingAn admixture that causes a
decrease in the rate of hydration of the hydraulic cement and
lengthens the time of setting.
Admixture, water-reducingAn admixture that either
increases slump of freshly mixed mortar or concrete without
increasing water content or maintains slump with a reduced
amount of water, the effect being due to factors other than air
entrainment.
Admixture, water-reducing high-rangeA water-
reducing admixture capable of producing large water reduction
or great flowability without causing undue set retardation or
entrainment of air in mortar or concrete.
Aggregate, reactiveAggregate containing substances
capable of reacting chemically with the products of solution
or hydration of the portland cement in concrete or mortar
under ordinary conditions of exposure, resulting in some
cases in harmful expansion, cracking, or staining.
Air, entrainedMicroscopic air bubbles intentionally
incorporated in mortar or concrete during mixing, usually
by use of a surface-active agent; typically between 10 and
1000 m in diameter and spherical or nearly so.
Air, entrappedAir voids in concrete that are not pur-
posely entrained and are significantly larger and less useful
than those of entrained air, 1 mm in diameter or larger in size.
Air contentThe total volume of air voids in cement
paste, mortar, or concrete, exclusive of pore space in aggregate
particles, usually expressed as a percentage of volume of the
paste, mortar, or concrete.
AlkaliSalts of alkali metals, principally sodium and
potassium, specifically sodium and potassium occurring in
constituents of concrete and mortar, usually expressed in
chemical analysis as the oxides Na
2
O and K
2
O.
Alkali-aggregate reactionChemical reaction in either
mortar or concrete between alkalies (sodium and potassium)
from portland cement or other sources and certain constituents
of some aggregates, under certain conditions, deleterious
expansion of concrete or mortar may result.
Alkali-carbonate reactionThe reaction between the
alkalies (sodium and potassium) in portland cement and certain
carbonate rocks, particularly calcitic dolomite and dolomitic
limestones, present in some aggregates, the products of the
reaction may cause abnormal expansion and cracking of
concrete in service.
Alkali-silica reactionThe reaction between the alkalies
(sodium and potassium) in portland cement and certain sili-
ceous rocks or minerals, such as opaline chert, strained
quartz and acidic volcanic glass, present in some aggregates;
the products of the reaction may cause abnormal expansion
and cracking of concrete in service.
Calcium chlorideA crystalline solid, CaCl
2
; in various
technical grades, used as a drying agent, as an accelerator for
fresh concrete, a deicing chemical, and for other purposes.
Cement, portlandA hydraulic cement produced by
pulverizing portland-cement clinker and usually containing
calcium sulfate.
CementitiousHaving cementing properties.
Sulfate attackEither a chemical or physical reaction or
both that occurs between sulfates usually in soil or ground-
water and concrete or mortar; the chemical reaction is primarily
with calcium aluminate hydrates in the cement-paste matrix,
often causing deterioration.
Sulfate resistanceAbility of concrete or mortar to with-
stand sulfate attack.
An expert system for mix design of high performance concrete
Muhammad Fauzi Mohd. Zain
*
, Md. Nazrul Islam
1
, Ir. Hassan Basri
2
Department of Civil and Structural Engineering, Faculty of Engineering, Universiti Kebangsaan Malaysia,
43600 UKM Bangi, Selangor Darul Ehsan, Malaysia
Received 5 March 2004; received in revised form 20 August 2004; accepted 14 October 2004
Available online 10 December 2004
Abstract
This paper describes a prototype expert system called HPCMIX that provides proportion of trial mix of High Performance Concrete (HPC)
and recommendations on mix adjustment. The knowledge was acquired from various textual sources and human experts. The system was
developed using hybrid knowledge representation technique. It is capable of selecting proportions of mixing water, cement, supplementary
cementitious materials, aggregates and superplasticizer, considering the effects of air content as well as water contributed by superplasticizer
and moisture conditions of aggregates. Similar to most expert systems, this system has explanation facilities, can be incrementally expanded,
and has an easy to understand knowledge base. The system was tested using a sample project. The systems selection of mix proportions and
recommendations regarding mix adjustment were compared favourably with those of experts. The system is user-friendly and can be used as
an educational tool.
q 2004 Elsevier Ltd. All rights reserved.
Keywords: High performance concrete; Mix design; Mix adjustment; Expert systems; Knowledge-based systems; Hybrid knowledge representation
1. Introduction
The selection of mix proportions is the process of
choosing suitable ingredients of concrete and determining
their relative quantities with the object of producing as
economically as possible concrete of certain minimum
properties, notably strength, durability, and a required
consistency [1]. Because the ingredients used are essentially
variable and many of the material properties cannot be
assessed truly quantitatively, selecting proportions for
concrete can also be dened as the process of nding the
optimum combination of these ingredients on the basis of
some empirical data as stated in relevant standards,
experience, and some rules of thumb [2].
Concrete mix design involves complicated issues, and
the correct ways to perform this can be achieved with
experts advice and experience [3]. Mix design of High
Performance Concrete (HPC) is more complicated because
HPC includes more materials, like superplasticizer and
supplementary cementitious materials (e.g. silica fume, y
ash, llers, etc.). In addition, maintaining a low water-
binder ratio with adequate workability makes the design
process more complicated. Traditionally, experienced civil
engineers, largely based on their experiential knowledge,
do the job of mix design [4]. However, experts are not
always available, nor do they always have time to consult
all possible references, review available data, and so on.
Some companies do not have personnel with the experi-
ence to make necessary decisions regarding concrete mix
design. The conventional computer programs are useful
only in manipulating the numerical data and providing
mathematical reasoning for the nal selection. They lack
the intuitive reasoning based on heuristic knowledge such
as experience and rules of thumb [5]. Many factors
inuence concrete mix design, and their mutual relation-
ship is so complicated that it is impossible to formulate
mathematical models to express their mutual actions and
reactions [6]. In addition, adjustments of trial mixes are
always performed by taking into account the information
0965-9978/$ - see front matter q 2004 Elsevier Ltd. All rights reserved.
doi:10.1016/j.advengsoft.2004.10.008
Advances in Engineering Software 36 (2005) 325337
www.elsevier.com/locate/advengsoft
* Corresponding author. Tel.: C603 89216223; fax. C603 89216147.
E-mail addresses: fauzi@vlsi.eng.ukm.my (M.F. Mohd. Zain),
nazrul@vlsi.eng.ukm.my (M. Nazrul Islam), drhb@vlsi.eng.ukm.my
(I. Hassan Basri).
1
Tel.: C603 89216819; fax: C603 89216147.
2
Tel.: C603 89216100; fax: C603 89216147.
from concrete quality tests, experts advice and experience.
It is believed that the problem of mix design and
adjustment of HPC can be alleviated if the engineers
knowledge can be augmented with some expert system
for afrming his judgment.
This paper describes a prototype expert system called
HPCMIX. The purposes of developing HPCMIX were to
improve the process of selecting and proportioning HPC
constituents and to make the knowledge of HPC easily
available to the concrete industry. It is also capable of
diagnosing causes of mix performance failure and giving
recommendations on corresponding performance
adjustment. However, like other expert systems, the
developed expert system will serve as a decision support
system; it will not replace completely human experts
decision making.
2. Expert systems and concrete mix design
An expert system is dened as a computer program
designed to model the problem-solving ability of a human
expert [7]. It utilises observed or available information to
produce high grade knowledge and solves problems by
qualitative reasoning using the heuristic knowledge of the
human expert [8]. Expert systems are most useful when the
knowledge is based on heuristics, which is often the case in
concrete mix design. Since concrete mix design and
adjustments are somewhat complicated, time-consuming
and tedious tasks, and also because it is not always possible
to be helped by the experts, there were some efforts to
develop expert system for concrete mix design. These
systems [2,3,9,10] give proportion of concrete mix
especially for normal concrete. A brief review of these
systems is available elsewhere [11]. Most of these systems
are rule-based systems, work using DOS operating system
and do not consider the cost of concrete mix selected. None
of these systems can diagnose causes of performance failure
of trial mix and give recommendations on corresponding
performance adjustment. Most importantly, they do not
consider the criteria of mix design for HPC such as
maintaining a low water-binder ratio and use of super-
plasticizer, silica fume, and so on. Therefore, these systems
cannot be used for mix design and adjustment of high
performance concrete.
3. Development of the HPCMIX
3.1. Knowledge acquisition
Knowledge for the HPCMIX was acquired from text-
books and manuals written by experts and related
professional institutions [1,1221], research papers from
journals and conference proceedings [3,2229] and experts
involved in concrete production. Thus knowledge was
acquired by text analysis (i.e. collection of knowledge from
the literature) and interviewing experts. Only unstructured
interviews of several experts involved in teaching, research
and consultation of concrete production were performed
in this project. Experts were asked to describe their
knowledge about the selection of concrete proportions,
diagnosing the problems in mix design and adjustments,
and their solutions. However, the main source of
knowledge was the literature mentioned above. By
analysing the knowledge from these sources, a more
objective perspective of the most appropriate expertise was
achieved, instead of being restricted to a single view
preferred by a particular expert. It may be relevant to
mention here that acquiring knowledge from these sources
was felt to be the most difcult and time-consuming task in
the prototype development process.
The mix design procedure for HPC developed by Aitcin
[15] was used as the mix design procedure for the HPCMIX
because of its wide acceptability among the experts in
Malaysia (Alternatively, in future, any other state-of-the-art
mix design method can be added to the system as a new
module without affecting overall performance of the
system). The Aitcin method follows the same approach as
ACI Committee 211 [18]. It is a combination of empirical
results and mathematical calculations based on the absolute
volume method. A ow chart of this method is presented
latter in this paper (Section 4.4). The procedure is initiated
by selecting ve different mix characteristics or materials
proportions in the following sequence: water-binder
ratio, water content, superplasticizer dosage, coarse
aggregate content and entrapped air content. The suggested
water-binder ratio is obtained from a compressive strength
vs. water-binder ratio graph for a given 28-day compressive
strength. The mixing water content is determined on the
basis of saturation point of superplasticizer. The super-
plasticizer dosage is deduced from the dosage at the
saturation point. The coarse aggregate content is obtained
as a function of the typical particle shape. The method
suggests using 1.5% as an initial estimate of entrapped air
content, and then adjusting it on the basis of the result
obtained with trial mix.
The ow diagram developed for the acquired knowledge
regarding mix performance adjustment is shown in Fig. 1.
The diagram shows that three criteria were considered for
judging the performance of a mix, i.e. strength, workability
and durability. For example, if a mix fails in strength
performance, then information is required about test results
regarding any of the following to nd out possible causes of
performance failure: (i) fracture pattern, (ii) bond failure
pattern, (iii) passage of fracture surface, and (iv) effect of
water-binder ratio. On the other hand, if the workability
performance is inadequate then it is required to know which
workability performance of the following is inadequate:
(i) rapid slump loss, (ii) low slump, or (iii) inadequate
workability. The ow diagram developed for the purpose of
cost estimation of a designed mix is shown in Fig. 2.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 326
The gure shows that unit cost of ingredients, mix
composition and batch size are the required information
for cost estimation of a concrete mix.
3.2. Knowledge representation
Through the efforts of researchers in articial intelli-
gence, a number of effective ways of representing knowl-
edge in a computer were developed [7]. The selection from
these knowledge representation techniques depends on the
nature of the expertise to be computerized, as well as the
practical capabilities and facilities of the expert system tool
used [30]. In developing HPCMIX, a hybrid approach of
knowledge representation (i.e. rule and frame systems) was
followed using Kappa-PC expert system shell [31]. Thus,
the domain of mix design of HPC was modelled using
object-oriented approach and production rules.
As an illustration of object-oriented approach, Fig. 3
shows the object hierarchy of Binder class. It consists of two
subclasses namely Cement and SupplCemMats (Supplemen-
tary Cementitious Materials). Each of these subclasses
includes several instances. For example, SupplCemMats
subclass consists of SilicaFume, FlyAsh, GGBS (ground
granulated blast-furnace slag), RiceHuskAsh and Others
instances. The attributes of objects were dened as object
slots. Slots can be thought of as descriptions of a particular
object. They add detail structure, list attributes or properties
which can be single or multiple-valued, textual strings or
numeric, or even Boolean. Slot values can be pre-dened,
restricted to a range or set of pre-specied possible values,
user-dened or determined from user consultations with
Fig. 1. Flow diagram of mix performance adjustment.
Fig. 2. Flow diagram of cost estimation.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 327
the system. For example, the slots of the instance FlyAsh
(Fig. 3) includes, among others, content in percent, specic
gravity and content in kg/m
3
(kilogram per cubic metre).
Table 1 shows these slots and the corresponding slot values.
Interactions among objects, instructions from one object
to another, and processes of objects were codied in the form
of methods, functions and rules. The rst approach involves
enhancing objects so that they represent the behaviour of the
things to which they correspond. Methods are written in KAL
(Kappa-PC Application Language) and stored within the
object. The next approach to representing processes involves
functions. Functions are also written in KAL and can either
be user-dened or built-in the system. Kappa-PC provides a
library of over 300 functions that allow for the manipulation
of its knowledge base. The third approach involves using
rules to represent the relationship between causes and effects
which species the conditions under which a particular
action or inference can occur. The syntax for methods,
functions and rules is identical; therefore, the same syntax
can be used to add an object, to write a method or to create a
rule. An example of a simple rule written for the coarse
aggregate content of the mix design module is shown below.
If: Shape of the coarse aggregate is cubic
Then: Coarse aggregate content of the mix should be
1100 kg/m
3
.
In KAL format, this rule was written as:
If: CoarseAgg: ShapeOfAgg #ZCubic
Then: CoarseAgg: ContentKgPerM3 #Z1100;
Where, CoarseAgg is an instance representing coarse
aggregate; ShapeOfAgg is a slot of CoarseAgg instance
representing the shape of the coarse aggregate; Cubic is the
value of the slot ShapeOfAgg representing that the shape of
the coarse aggregate is cubic; and so on. There are two
approaches for evaluating production rules: backward
chaining and forward chaining [7,3234]. The HPCMIX
uses a forward-chaining or data-driven inference
mechanism.
4. Knowledge base modules of the HPCMIX
The HPCMIX knowledge base consists of three design
related modules namely Mix Design, Mix Performance
Adjustment and Cost Estimation modules and an accessory
module named General Information module. The modules
can be accessed from the main interface window shown in
Fig. 4. A brief description of each of the above modules is
given in the following sections.
4.1. Mix Design module
The objective of Mix Design module is to proportion
HPC mixes. It consists of four submodules namely First
Trial Batch, Trial Batch for Laboratory, One Cubic Metre
SSD Composition (SSD stands for saturated-and-surface-
dry) and Batch Composition as shown in Fig. 5. The First
Trial Batch submodule computes proportions of concrete
mixes according to the data supplied by the user. It gives
composition of one cubic metre of concrete for eld
conditions of aggregates. The Trial Batch for Laboratory
submodule helps in computing quantities for making
concrete samples for laboratory testing. This gives the
user opportunity to test mix design results in a laboratory for
desired performance requirements for a small amount of
proportioned ingredients. The One Cubic Metre SSD
Composition submodule calculates proportions for one
cubic metre of concrete for SSD conditions of aggregates.
The Batch Composition submodule activates a function to
calculate amounts for batch quantities for eld conditions of
aggregates.
4.2. Mix Performance Adjustment module
The Mix Performance Adjustment module helps in
adjusting mix proportions after laboratory testing. The
interface window of this module is shown in Fig. 6.
Incorporating the knowledge that experts use in diagnosis,
the Mix Performance Adjustment module diagnoses
possible causes of performance failure of concrete mix
and recommends corresponding remedial measures. The
Quantitative Advice button (Fig. 6) of this module gives
specic quantitative recommendations on achieving various
performances of HPC mix. The module also displays
reasons for giving any recommendation.
Table 1
Description of an Instance using Slots
Instance: FlyAsh, Parent Class: SupplCemMats
Slot Name Slot Value Comment
ContentPercent 10.00 User dened value
SpecicGravity 2.50 User dened value
ContentKgPerM3 53.19 System derived value
Fig. 3. Object hierarchy of the Binder class.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 328
Fig. 4. Main interface window of the HPCMIX.
Fig. 5. Interface window of the Mix Design module.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 329
4.3. Cost Estimation module
If the cost estimation is felt to be necessary, it can be
carried out using the Cost Estimation module. The rst step
is to input unit cost of the proportioned ingredients. A
transcript image in the Cost Estimation interface window
helps the user by displaying information about the ranges of
unit cost of concrete ingredients. Although these values vary
from country to country, this information will help the user
in getting idea about approximate unit cost of ingredients.
After inputting the unit cost of ingredients, the user gets the
costs of one cubic metre of concrete and of a particular batch
size by pressing appropriate buttons.
4.4. General Information module
Fig. 7 shows the interface window of the General
Information module. The objective of this module is to
assist the user in the efcient utilisation of the HPCMIX
prototype. The module provides a user-friendly environ-
ment whereby the various guidelines and information are
available to the user. The User Guide button assists the user
in the efcient consultation process. The user gets advice on
the fundamentals of HPC and the approaches of mix design
and adjustment of HPC through HPC Technology and
Principle of Mix Design buttons of this module. A
knowledge dictionary is also available through Knowledge
Dictionary button to assist the user with unfamiliar technical
terms. This dictionary is also useful as an educational
feature. The dictionary includes: basic denitions of
concrete mix design, HPC and expert system; types of
tests for evaluation of fresh and hardened concrete; and
statistical measures used in assessment of concrete mix
design. In addition, a general conceptual ow diagram of
HPC mix design using Aitcin method (see Fig. 7) and
various photographs showing the testing of fresh and
hardened concretes are also accessible through appropriate
buttons of this module.
5. Case study and evaluation of the HPCMIX
The objective of this case study was to evaluate the
performance of the HPCMIX consultation process and
results when it was applied to a HPC mix. The case study
was carried out on an example mix design of HPC from a
classical textbook [15]. This example was selected because
the author of this textbook is considered as one of the well-
known experts in the domain of HPC. Another expert (Dr
Hilmi Mahmud, Associate Professor, University of Malaya,
Malaysia), who has been involved in teaching, research and
consultation in the production of concrete for past fteen
years, was asked to perform concrete mix design based on
the information of this textbook. The results of the design
obtained from HPCMIX were compared with those
reported in this textbook (referred to as Expert-1) and
those calculated by Dr Hilmi Mahmud (referred to as
Expert-2).
Fig. 6. Interface window of the Mix Performance Adjustment module.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 330
The user clicked on Mix Design button of the main
interface window (Fig. 4) for mix design purpose. This
opened the interface window of the Mix Design module
(Fig. 5). From that window, the user learned about the
functions of the buttons regarding mix design submodules
and got advice about each submodule by pressing on
appropriate Advice button. The input data, consultation
process and explanation of results of each submodule of the
mix design module as well as mix performance adjustment
module are described in the following sections. These
sections also include description about user-friendliness,
advantages, disadvantages and an overall evaluation of the
system.
5.1. First trial batch
Data for the rst trial batch composition of the present
case study is described in Appendix A. In this example, a
100 MPa concrete mix was designed with Type I Portland
cement, a naphthalene-type superplasticizer, a dolomitic
limestone, a siliceous natural sand and silica fume.
For each concrete ingredient of Appendix A, a related
window appeared one after another as the user continued
with data input process. For example, window that
appeared for data input related to silica fume is shown in
Fig. 8. The system provided information about each
concrete ingredient similar to that shown for silica fume
(Fig. 8). Each of these windows also displayed several
buttons namely Data Input, Advice, Main Screen, Mix
Design Screen, Back and Exit. If the user wanted to input
data about an ingredient, he pressed on the Data Input
button. As a result, the system prompted the User Request
form such as shown in Fig. 9. This form displayed the
limits of the value of the desired input parameter. After
entering the value of the parameter, the user pressed on OK
button and, as a consequence, the window for the next
input parameter automatically appeared. If the user was not
sure about the value of the parameter, he could press on
Comment button (Fig. 9) for usual range of values as
shown in Fig. 10, which also gave the sources of
information. For the beginners or students of concrete
technology who are not familiar about the parameter may
press Advice button (see Fig. 8). This operation gives
detailed information about the parameter in the second
transcript image (see Fig. 8 for advice on silica fume).
Moreover, the beginners can access to knowledge
dictionary through Knowledge Dictionary button
(see Fig. 7). These facilities make the system very
user-friendly and suitable for using as an educational tool.
At the end of data input session, the Summary of Input
Data window appeared. From that window, the user could
verify input data. If there was any inconsistency, the user
could go back and modify it. After verifying input data, the
user could see design values (i.e. First Trial Batch
composition) by pressing on Final Design button. The
design values after moisture adjustment as proportioned by
the HPCMIX are shown in Fig. 11. This gure also shows
the key input parameters that were input by the user. From
this window, the user might request for explanation about
the mix design process, might go back to main screen or
Fig. 7. Interface window of the General Information module showing mix design ow diagram.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 331
might exit the mix design session by pressing on appropriate
buttons.
Table 2 shows the comparison of the trial batching mix
design selected by the developed system and the experts. It
can be observed from Table 2 that the results of the three
mix designs are very close except cement and ne aggregate
contents. The cement content suggested by the system is the
lowest (i.e. 464.38 kg/m
3
) and that suggested by Expert-1 is
the highest (i.e. 470.00 kg/m
3
). This was because the system
maintained high precision in computation by considering up
to ten digits after decimal point. But Expert-1 rounded off
the gures to integer values for convenience. For example,
total binder content was rounded off from 518.5 to
520 kg/m
3
and silica fume content was rounded off from
51.85 to 50 kg/m
3
by Expert-1. On the other hand, cement
content suggested by Expert-2 was of intermediate value
(i.e. 466.67 kg/m
3
) because he considered only two digits
after decimal point in his calculation. Due to the variation in
cement content, the variation in ne aggregate content was
also observed among the three mix designs
(i.e. 777.51 kg/m
3
by HPCMIX, 772.00 kg/m
3
by Expert-1
and 775.17 kg/m
3
by Expert-2) in order to maintain the total
displaced volume according to absolute volume method.
However, Expert-2 indicated that the differences among the
three mix designs were not critical. Expert-2 also indicated
that the concrete proportions selected by the system were
accurate enough for the rst trial batching. It may be
relevant to mention here that, for a particular mix design,
there are always several answers, which can satisfy the
requirements of the specication [3]. Thus the concrete
proportions selected by the system were accurate enough for
Fig. 8. A typical window during data input process.
Fig. 9. A typical User Request form during data input (after pressing Data
Input button of Fig. 8.
Fig. 10. A typical explanation window during data input (after pressing
Comment button of Fig. 9.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 332
the rst trial batching. The developed system is, therefore,
capable of proportioning the rst trial batch of HPC mix
with acceptable accuracy.
5.2. Trial batch for laboratory
The calculated mixture proportions should be checked by
means of trial batches prepared and tested for the intended
performance requirements [15]. These tests and observation
on the designed mix are usually performed in the laboratory
by using a small portion of the designed mix, i.e. by
laboratory trial batch. For the purpose of validation of
the system, data used for an example laboratory testing is
described in Appendix B. The results computed by the
system and those of the experts are compared in Table 3. It
can be observed that the proportions selected by the
HPCMIX were close enough to those selected by
the experts. The highest variation was observed in
ne aggregate content, which was only 0.69 kg/m
3
(i.e. 72.4271.73Z0.69). According to the experts opinion,
the variations among the three mix designs were negligible.
Thus the present system can be used for the calculation of
laboratory trial mix of HPC with acceptable accuracy.
Fig. 11. Window showing composition and key data of First Trial Batch submodule.
Table 2
Comparison of results of First Trial Batch
Item HPCMIX Expert-1
a
Expert-2
b
Water-Binder
Ratio
0.27 0.27 0.27
Water (l/m
3
) 123.38 124.00 123.40
Cement (kg/m
3
) 464.38 470.00 466.67
Silica Fume
(kg/m
3
)
51.60 50.00 51.85
Coarse Aggre-
gate (kg/m
3
)
1066.40 1066.00 1066.40
Fine Aggregate
(kg/m
3
)
777.51 772.00 775.17
Superplasticizer
(l/m
3
)
10.66 10.70 10.72
Air Content (%) 1.50 1.50 1.50
a
Expert-1: Aitcin [15] (see reference).
b
Expert-2: Dr Hilmi Mahmud, University of Malaya, Malaysia.
Table 3
Comparison of results of Laboratory Trial Batch
Item HPCMIX Expert-1
a
Expert-2
b
Water-Binder
Ratio
0.27 0.27 0.27
Mixing Water
(litre)
11.49 11.53 11.42
Cement (kg) 43.26 43.70 43.18
Silica Fume
(kg)
4.81 4.70 4.80
Coarse Aggre-
gate (kg)
99.33 99.10 98.68
Fine Aggregate
(kg)
72.42 71.80 71.73
Superplasticizer
(litre)
0.99 0.99 0.99
Air Content (%) 1.50 1.50 1.50
a
Expert-1: Aitcin [15] (see reference).
b
Expert-2: Dr Hilmi Mahmud, University of Malaya, Malaysia.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 333
5.3. One cubic metre SSD composition
This submodule can be used to compute composition of
one cubic metre of concrete for SSD conditions of
aggregates using the data of laboratory trial batch. It can
also be used for conversion of concrete composition from
eld conditions of aggregates to SSD conditions of
aggregates. During computation, it takes into account
moisture conditions of aggregates in the bins as well as
water hidden in liquid superplasticizer. The data for a
sample example are presented in Appendix C. The results
are compared in Table 4 that shows good agreement
between the results of the HPCMIX and those of the
experts. The highest variation was only 0.88 kg/m
3
(i.e. 1074.511073.63Z0.88) in coarse aggregate content.
The system gave proportions closer to those of Expert-2.
Expert-2 was consulted and he agreed that the variation was
negligible and the concrete proportions selected by the
system were accurate enough to make the trial batching.
These results serve as a means of verication of the
accuracy of knowledge base of the developed system.
5.4. Batch composition
This submodule computes batch composition for a
concrete construction project for eld conditions of
aggregates. The system collects following data from the
user in order to compute batch composition: one cubic metre
SSD composition, moisture conditions of aggregates in the
eld, and the expected batch size. It was assumed, for
example, that a concrete batch plant was to produce 8 m
3
of
HPC on the basis of the SSD composition shown in
Appendix D. The user input those data by pressing
appropriate buttons of the Batch Composition interface
window. After the completion of data input, the user got
results of the batch composition for eld conditions of
aggregates. The results are compared with those of the
experts in Table 5. Again, it can be seen that the results of
the HPCMIX compared well with those of the experts.
The only variation with the result of Expert-1 was observed
in mixing water requirement (i.e. 875.08872.00Z3.08 l).
Expert-1 rounded off the water contributed by liquid
superplasticizer from 5.6 to 6.0 l in one cubic metre SSD
composition calculation. This excess water contributed by
superplasticizer (i.e. 6.05.6Z0.4 l) in one cubic metre
composition was multiplied by 8 for getting 8 m
3
compo-
sition (i.e. 0.4!8Z3.2 l). On the other hand, this excess
water (i.e. 3.2 l) was subtracted from the required mixing
water and hence the value was less (i.e. 872 l instead of
875.2 l). Again Expert-2 was consulted and he agreed that
the performance of the HPCMIX in computing batch
composition was satisfactory. This comparison also gives
a positive indication of the accuracy of knowledge base of
the system.
5.5. Mix performance adjustment
The purpose of this module is to diagnose possible causes
of performance failure of a HPC mix and to recommend on
corresponding remedial measures in order to achieve the
desired performance. These features are demonstrated in
this portion of the case study. It was assumed, for example,
that the strength performance of the mix was not achieved in
the laboratory testing of the rst trial batch. Careful
examination of the failure pattern of the test specimen
revealed that fracture surface passed through hydrated
cement paste. After getting these data from the user, the
inference engine of the HPCMIX forward chained through
its knowledge base and produced recommendation as stated
ahead (see Fig. 6): Use lower water-binder ratio. The
system also explained the reason for giving this recommen-
dation (see Fig. 6): If the fracture surface passes almost
entirely through the hydrated cement paste around the
aggregates, a stronger concrete can be made with the same
aggregates by lowering further the water-binder ratio. The
system also recommended, when the user pressed on
Quantitative Advice button, that increasing the 28-day
compressive strength of concrete by 1 MPa necessitate the
addition of approximately 8.66 kg/m
3
of extra cement to
Table 5
Comparison of results of Batch Composition
Item HPCMIX Expert-1
a
Expert-2
b
Water-Binder
Ratio
0.29 0.29 0.29
Mixing Water
(litre)
875.08 872.00 875.12
Cement (kg) 3600.00 3600.00 3600.00
Coarse Aggre-
gate (kg)
8400.00 8400.00 8400.00
Fine Aggregate
(kg)
6120.00 6120.00 6120.00
Superplasticizer
(litre)
64.00 64.00 64.00
a
Expert-1: Aitcin [15] (see reference).
b
Expert-2: Dr Hilmi Mahmud, University of Malaya, Malaysia.
Table 4
Comparison of results of One Cubic Metre SSD Composition
Item HPCMIX Expert-1
a
Expert-2
b
Water-Binder
Ratio
0.27 0.27 0.27
Water (l/m
3
) 144.61 144.88 144.57
Cement (kg/m
3
) 478.70 478.35 478.72
Fly Ash (kg/m
3
) 53.19 53.15 53.19
Coarse Aggre-
gate (kg/m
3
)
1074.51 1073.63 1074.57
Fine Aggregate
(kg/m
3
)
723.65 722.84 723.72
Superplasticizer
(l/m
3
)
9.57 9.57 9.57
Air Content (%) 1.50 1.50 1.50
a
Expert-1: Aitcin [15] (see reference).
b
Expert-2: Dr Hilmi Mahmud, University of Malaya, Malaysia.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 334
the mix. These recommendations and explanation exactly
matched with those of the experts [3,15] and these
recommendations were better than the recommendation
from any individual of these experts. Thus the performances
of the prototype to diagnose mix performance failure and to
give recommendation on corresponding performance
adjustment are considered to be satisfactory.
5.6. User-friendliness and the user interface
The HPCMIX user interface was designed for user
friendliness to enable its efcient utilization in the
following ways.
(i) Due to the exibility in determining ow of consul-
tation, the user may need to be advised on the most
appropriate sequence of design steps. In order to
achieve this goal, the User Guide button in the General
Information module gives brief information of the
overall consultation steps. On the other hand, the rst
Transcript Image in the beginning of each consultation
session provides information regarding the steps to
follow for that session (for example, see Fig. 5). In
addition, Advice buttons are available in most of the
modules (see Fig. 5, for example) to guide the user
about design procedure. Moreover, each consultation
window contains some buttons to guide the user for the
next design steps.
(ii) The advantages and disadvantages of each alternative
solution are made easily available to the user so that he
can interact with the HPCMIX with better perspective.
For example, the Advice button in the Binder Types
window during the data input for binder types opens
text information on binders (similar to Silica Fume
window in Fig. 8). It explains the advantages and
disadvantages of using cement only and cement and
pozzolans as binder. The user will then be able to have
a better understanding of the alternative use of these
binders.
(iii) Each User Request form of data input has a Comment
button attached to it as described earlier (see Fig. 9).
This helps the user by expanding the meaning of a
question and thus aids the user in responding more
efciently to the prompts of the design procedure.
(iv) One of the distinguishing characteristics of an expert
system such as the HPCMIX is the transparency of its
reasoning process and knowledge base. This advantage
is available to the user through Explain button (for
example, see Fig. 11), which displays the rules that
have been used by the inference mechanism, thus
explaining the reasons for arriving at a particular
recommendation. Another example of explanation
facility is shown in the second transcript image of the
Mix Performance Adjustment module (Fig. 6), which
explains the reasons for giving any recommendation
regarding mix performance adjustment.
(v) As mentioned earlier, the General Information module
contains a Knowledge Dictionary. Basic knowledge
about HPC, mix design and expert system is available
there. It also contains some statistical information. This
dictionary is very useful as an educational feature.
Moreover, concrete mix design ow diagram (see
Fig. 7) and several photographs are included in the
General Information module to enhance the edu-
cational performance of the HPCMIX.
5.7. Advantages, disadvantages and overall
evaluation of the system
It may be mentioned, by considering all these features,
that the consultation process of the HPCMIX is reason-
ably satisfactory and systematic. The ow of consultation
is exible, allowing the user to reset data, to go back for a
new consultation, to review input values and other
procedures until he is satised with the results. The
ability of the HPCMIX to run using Windows operating
system, to give recommendation on possible causes of
performance failure of the mix, and the facility of
knowledge dictionary make this system superior to similar
other systems in the domain. The system-user interaction
is very interactive with explanatory facilities available
throughout the consultation session. Moreover, the system
gives information of data at every stages of data input,
which makes it superior to conventional programs of
concrete mix design. It has facilities like Explain, Advice
and Comment buttons as well, which make the system
very user-friendly.
The main disadvantage of the developed system is that it
is applicable for concrete compressive strength from 40 to
160 MPa for which Aitcin method is valid. Another
limitation of the system is that it is not platform
independent. The user will need a runtime version of
Kappa-PC in order to use the system. However, in future,
these limitations can be handled by incorporating another
state-of-the-art mix design method and by using another
state-of-the-art expert system shell if they are proved to be
useful.
In order for expert systems not to become obsolete, they
must be nurtured and kept current [35]. All expert systems,
the HPCMIX included, cannot claim completeness in their
knowledge bases; they are always subject to upgrading,
modication and correction. It should be recognized that
HPCMIX is a research prototype; and hence, it must be
rened and tested further for commercial use. The existing
knowledge base of the prototype can be improved by
rening, expanding, and reinforcing its knowledge base
using new ndings as reported in literature or new
experience from domain experts. It must also be kept in
mind that, like other expert systems, HPCMIX will serve as
a decision support system; it will not replace completely
human experts decision making.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 335
6. Conclusions
This paper presented a prototype expert system called
the HPCMIX developed for mix design of HPC. The
HPCMIX is capable of selecting proportions of mixing
water, cement, supplementary cementitious materials,
superplasticizer and aggregates, considering the effects of
air content as well as water contributed by superplasticizer
and moisture conditions of aggregates. Using an expert
system shell, HPCMIX was developed according to the
mix design method proposed by Aitcin. Alternatively, in
future, any other state-of-the-art mix design method of
HPC can be added to the system as a new module without
affecting overall performance of the system. In addition to
proportioning concrete mixes, the system is also capable
of giving recommendations on mix performance adjust-
ment. It was found that it is feasible, efcient and
effective, to use an expert system approach for the
proportioning of HPC mix. The ability of the system to
give comments and advice about an input data, together
with the facility of knowledge dictionary, makes this
expert system very useful for the educational environment.
Additional knowledge to expand the scope of the system
can be added without major modication of the structure
of the program.
The best approach to making a mix design of HPC, of
course, is to use proportions previously established for
similar concrete using the same materials. In addition,
rules-of-thumb and past experience should also be used,
wherever possible. Where such prior information is limited
or unavailable, the HPCMIX can be used to assist the user in
the mix design of HPC.
Acknowledgements
The authors would like to express sincere gratitude to
Universiti Kebangsaan Malaysia for providing the fund for
the research and MBT (Malaysia) Sdn. Bhd for the supply of
materials and technical support throughout the research
program. The authors would like to thank Dr Hilmi
Mahmud (University of Malaya, Malaysia) for his
invaluable input and patience.
Appendix A. Data for rst trial batch
It was supposed that a 100 MPa concrete had to be made
with the following data [15]: a Type I Portland cement; a
naphthalene-type superplasticizer with a total solids
content of 40% and specic gravity of 1.21; a dolomitic
limestone having maximum sizeZ10 mm, specic gravity
(SSD)Z2.80, absorptionZ0.8%, eld moistureZ0%, and
the shape of the particles can be described as between
average and cubic; a siliceous natural sand of
specic gravity (SSD)Z2.65, absorptionZ1.2%, and eld
moistureZ3.5%. Silica fume at 10% replacement (of total
cementitious material) was to be used; its specic gravity
was 2.20. The dosage of solids superplasticizer at the
saturation point was 1.0%.
Appendix B. Data for laboratory trial batch
In order to test First Trial Batch of HPC mix, the
following specimens are needed [15]: three 100!200 mm
cylinders for tests at 1, 7, 28 and 91 days in compression;
three 150!300 mm cylinders for tests at 28 days in
compression; three 150!300 mm cylinders for tests
for elastic modulus at 28 days; and three beams 100!
100!400 mm for tests for modulus of rupture at 28 days. A
slump test, an air content test and a unit mass test will be
done on the fresh concrete. Except the air content test, the
concrete used for these tests will be recovered. Knowing
that an air test needs 15 kg, the amount of concrete to make
this trial batch needs to be calculated, assuming 10% extra
materials to compensate for losses.
Appendix C. Data for one cubic metre SSD composition
A trial batch with an adequate consistency and adequate
initial and nal slumps was made using the following
quantities of materials [15]: waterZ12 l, cementZ45 kg,
Fly ashZ5 kg, coarse aggregateZ100 kg, ne aggregateZ
70 kg, and superplasticizerZ0.9 l. The air content of this
trial batch was 1.5%. The materials used to make this trial
batch had the following properties: coarse aggregate-
specic gravity (SSD)Z2.75, absorptionZ1.0%, and eld
moistureZ0%; and ne aggregate-specic gravity
(SSD)Z2.65, absorptionZ1.0%, and eld moistureZ
3.9%. The y ash used had a specic gravity of 2.50. The
superplasticizer was naphthalene based one with a specic
gravity of 1.21 and a solid content of 42%. What is the
composition of 1 m
3
of such concrete?
Appendix D. Data for batch composition
A concrete batch plant is to produce 8 m
3
of HPC on the
basis of the following SSD composition [15]: w/cZ0.29,
waterZ130 l, cementZ450 kg, coarse aggregatesZ
1050 kg, ne aggregatesZ750 kg, and superplasticizerZ
8 l (liquid) and 4 kg (solid). The aggregates having the
following water contents are in the bins: coarse aggregate-
specic gravity (SSD)Z2.75, absorptionZ0.8%, and eld
moistureZ0.8%; and ne aggregate-specic gravity
(SSD)Z2.65, absorptionZ1.0%, and eld moistureZ
3.0%. The superplasticizer is a naphthalene superplasticizer
containing 42% solids and having a specic gravity of 1.21.
What are the masses of materials that must be weighed to
make 8 m
3
of concrete?
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 336
References
[1] Neville AM. Properties of concrete, 4th ed. Essex: Longman Group
Limited; 1995.
[2] Akhras G, Foo HC. A knowledge-based system for selecting
proportions for normal concrete. Ex Sys Appl 1994;7(2):32335.
[3] Bai Y, Amirkhanian SN. Knowledge-based expert system for concrete
mix design. J Cons Eng Mngt, ASCE 1994;120(2):35773.
[4] Islam MN, Al-Mattarneh HMA, Zain MFM, Basri HB. Towards an
expert system for HPC mix design. Proceedings of the world
conference on concrete materials and structures, Shah Alam,
Malaysia, May; 2002.
[5] Foo HC, Akhras G. Expert systems and design of concrete mixtures.
Conc Int 1993;15(7):426.
[6] Oh J, Lee I, Kim J, Lee G. Application of neural networks for
proportioning of concrete mixes. ACI Mat J 1999;96(1):617.
[7] Durkin J. Expert systems design and development. New Jersey:
Macmillan; 1994.
[8] Hickey JP, Aldridge AJ. An expert system for the prediction of
aquatic toxicity of contaminants. In: Hushon JM, editor. Expert
systems for environmental applications. Washington: American
Chemical Society; 1990.
[9] Smith LM. Interim report on COMIX: an expert system for concrete
mix design, Report No. M4.87/1, Central Laboratories, New Zealand;
1987.
[10] Malasri S, Maldonado S. Concrete mix designer. Comp Appl Conc
Tech, ACI 1988;SP113-3:3341.
[11] Islam MN, Al-Mattarneh HMA, Zain MFM, Basri HB. Expert
systems and concrete mix design: a review. Proceedings of the sixth
international conference on concrete technology for developing
countries, Amman, Jordan, October; 2002.
[12] Malier Y, editor. High performance concrete: from material to
structure. London: E and FN Spon; 1992.
[13] Shah SP, Ahmad SH, editors. High performance concrete and
applications. London: Edward Arnold; 1994.
[14] Nawy EG. Fundamentals of high strength high performance concrete.
London: Longman Group Limited; 1996.
[15] Aitcin PC. High-performance concrete. London: E and FN Spon;
1998.
[16] Sauddin M. Inuence of different curing methods on the mechanical
properties and durability of HPC exposed to medium temperature.
MSc Thesis, Universiti Kebangsaan Malaysia, 1998.
[17] Day KW. Concrete mix design, quality control and specication,
second ed. London: E and FN SPON; 1999.
[18] ACI Committee 211. Standard practice for selecting proportions for
normal, heavyweight, and mass concrete (ACI 211.1-91). Detroit:
American Concrete Institute; 1991.
[19] ACI Committee 363. State-of-the-art report on high-strength concrete
(ACI 363R-92). Detroit: American Concrete Institute; 1992.
[20] ACI Committee 211. Guide for selecting proportions for high strength
concrete with Portland cement and y ash (ACI 211.4R-93). Detroit:
American Concrete Institute; 1993.
[21] ACI Committee 201. Guide to durable concrete. ACI Manual of
Concrete Practice: Part 1 (ACI 201.2R-92). Detroit: American
Concrete Institute; 1995.
[22] de Larrard F. A method for proportioning high-strength concrete
mixtures. Cem Conc Agg 1990;12(2):4752.
[23] Mehta PK, Aitcin PC. Principles underlying production of high-
performance concrete. Cem Conc Agg 1990;12(2):708.
[24] Rougeron P, Aitcin PC. Optimisation of the composition of high-
performance concrete. Cem Conc Agg 1994;16(2):11524.
[25] Mehta PK, Aitcin PC. Microstructural basis of selection of
materials and mix proportions for high-strength concrete. Proceed-
ings of the one-day short course on concrete technology/HPC:
Properties and Durability, University Malaya, Kuala Lumpur,
May; 1997.
[26] Aitcin PC. Durable concrete: current practice and future trends.
Proceedings of the one-day short course on concrete technology/HPC:
Properties and Durability, University Malaya, Kuala Lumpur, May;
1997.
[27] Aitcin PC. The durability of high performance concrete. Proceedings
of the one-day short course on concrete technology/HPC: Properties
and Durability, University Malaya, Kuala Lumpur, May; 1997.
[28] Mahmud H. Cement replacement materials in concrete- for strength,
durability and economy. Proceedings of the one-day short course on
concrete technology/HPC: Properties and Durability, University
Malaya, Kuala Lumpur, May; 1997.
[29] Neville A, Aitcin PC. High-performance concretean overview.
Materials and Structures 1998;31:1117.
[30] Islam MN, Zain MFM, Basri HB. Development of an expert system
for high performance concrete mix design. Proceedings of the
international conference on robotics, vision, information and signal
processing, Penang, Malaysia, January, 2003.
[31] IntelliCorp Inc. KAPPA-PC Ver. 2.4 Users Manual, IntelliCorp Inc.,
CA, 1997.
[32] Basri HB. An expert system for preliminary landll design in
developing countries. PhD Thesis, Department of Civil Engineering,
University of Leeds; 1994.
[33] Basri HB. An expert system for landll leachate management. Env
Tech 2000;21:15766.
[34] Basri NEA. An expert systemfor the design of composting facilities in
developing countries. PhD Thesis, Department of Civil Engineering,
University of Leeds, 1999.
[35] Kaetzel LJ, Clifton JR, Klieger P, Snyder K. Highway concrete
(HWYCON) expert system user reference and enhancement guide.
Gaithersburg: National Institute of Standards and Technology
(BFRL); 1993.
M.F. Mohd. Zain et al. / Advances in Engineering Software 36 (2005) 325337 337
Durable fiber reinforced self-compacting concrete
V. Corinaldesi, G. Moriconi
*
Department of Materials and Environment Engineering and Physics, Marche Polytechnical University, Via Brecce Bianche, 60131 Ancona, Italy
Received 18 December 2002; accepted 28 July 2003
Abstract
In order to produce thin precast elements, a self-compacting concrete was prepared. When manufacturing these elements, homogenously
dispersed steel fibers instead of ordinary steel-reinforcing mesh were added to the concrete mixture at a dosage of 10% by mass of cement.
An adequate concrete strength class was achieved with a water to cement ratio of 0.40. Compression and flexure tests were carried out to
assess the safety of these thin concrete elements. Moreover, serviceability aspects were taken into consideration. Firstly, drying shrinkage
tests were carried out in order to evaluate the contribution of steel fibers in counteracting the high concrete strains due to a low aggregate
cement ratio. Secondly, the resistance to freezing and thawing cycles was investigated on concrete specimens in some cases superficially
treated with a hydrophobic agent. Lastly, both carbonation and chloride penetration tests were carried out to assess durability behavior of this
concrete mixture.
D 2003 Elsevier Ltd. All rights reserved.
Keywords: Durability; Fiber reinforcement; Self-compacting concrete
1. Introduction
Self-compacting concrete is considered a concrete that
can be placed and compacted under its own weight
without any vibration effort, assuring complete filling of
formworks even when access is hindered by narrow gaps
between reinforcement bars. Concrete that must not be
vibrated is a challenge to the building industry. In order
to achieve such behavior, the fresh concrete must show
both high fluidity and good cohesiveness at the same
time. As already outlined by Collepardi [1], some con-
cretes bearing these requirements were first studied in
19751976. At that time, neither the modern acrylic-
based superplasticizers nor the new viscosity-modifying
agents (VMAs) were available. Nevertheless, the simul-
taneous use of melamine or naphthalene-based super-
plasticizers with a relatively high content of powdered
materials proved to be very effective [2]. The present
work steams from that experience and aims at evaluating
the durability of this material. A self-compacting concrete
was designed for thin precast elements, which only
require steel reinforcement for dead and minor loads.
For this reason, the addition of steel fibers to the mixture
in order to counteract drying shrinkage of concrete was
also chosen. However, from a functional point of view, a
compressive strength class of 40 MPa was requested,
taking into account early flexural strength for lifting and
moving, as well as fatigue and creep effects.
2. Concrete preparation
2.1. Materials
A portlandlimestone blended cement, type CEM II/
A-L 42.5 R according to the European Standards EN-
197/1, was used. Its Blaine fineness was 0.363 m
2
/g and
its specific gravity was 3050 kg/m
3
. The chemical com-
position of the cement is reported in Table 1. A com-
mercial limestone powder, originating from marble, was
used as a mineral admixture. It was chosen bearing in
mind the suggestion on fineness reported in Ref. [3]; its
Blaine fineness was 0.610 m
2
/g. Moreover, its specific
gravity was 2650 kg/m
3
and its chemical composition is
reported in Table 1. Crushed limestone aggregate (10-mm
0008-8846/$ see front matter D 2003 Elsevier Ltd. All rights reserved.
doi:10.1016/j.cemconres.2003.07.005
* Corresponding author. Tel.: +39-071-2204725; fax: +39-071-
2204729.
E-mail address: moriconi@univpm.it (G. Moriconi).
Cement and Concrete Research 34 (2004) 249254
maximum size) and natural sand (6-mm maximum size)
were used. The gradation of both crushed aggregate and
sand are shown in Fig. 1, and their physical properties
are reported in Table 2. A 40% aqueous solution of a
melamine-based polymer was added to the mixture in
order to guarantee very high workability. Straight steel
fibers, with length and diameter of 11.0 and 0.4 mm,
respectively, thus implying an aspect ratio (AR) equal to
28, were employed at a dosage of 0.6% by volume. This
dosage was decided on the basis of some information
reported in the literature [46], and carrying out prelim-
inary tests, which confirmed that an increase in fiber
content from 0.5% to 1%, results in lower concrete
performance. Finally, a commercial product based on
siloxanic resins was used to make the concrete surface
hydrophobic.
2.2. Mixture proportions
The mixture proportions of the self-compacting con-
crete are reported in Table 3, whereas its composition,
expressed by volume percentage, is reported in Table 4.
With respect to a common concrete (Ref-0.40 in
Table 4), with the same water to cement ratio (0.40)
and the same maximum size of aggregate particles (10
mm), a higher content of very fine material (cement and
mineral additions) and a lower content of gravel (fraction
with grain size higher than 5 mm) had to be used for self-
compacting concretes. The exceeding paste is necessary to
minimize energy loss caused by coarse particle collisions
during concrete flowing through narrow sections [7], and
to counteract segregation when the concrete is already
placed. The water cement ratio chosen was equal to 0.40,
and the water fine material ratio was 0.34 due to the
addition of limestone powder at a dosage of 16% by mass
of cement. In this work, VMAs were not employed;
therefore, a very high content of materials passing through
ASTM nj100 sieve of 150 Am was necessary [8], up to
19% by volume of concrete (see Table 4). Other con-
ditions that should be respected in order to meet the self-
compactibility requirement concern the water fine mate-
rial ratio and the sand to mortar ratio; in this work, they
were equal to 1.08 and 0.50, respectively, both expressed
by volume percentage (see Table 4). In order to obtain a
workability measured as high slump flow, a dosage of
1.6% by mass of cement of a melamine-based super-
plasticizer was necessary.
3. Experimental program and discussion of results
3.1. Slump flow test
As a first step, the properties of the fresh concrete other
than slump were evaluated, since in this case the slump
value is not relevant being the concrete very fluid. There-
fore, the attention was focused on the measurement of the
slump flow, which is the mean diameter (U
fin
) of the cake
of concrete obtained after releasing of a standard slump
cone. Then, the elapsed time to gain the mean diameter of
Table 1
Chemical composition of cement and limestone powder
Oxide (%) Cement Limestone powder
SiO
2
29.67 38.70
Al
2
O
3
3.74 8.02
Fe
2
O
3
1.80 3.34
TiO
2
0.09 0.12
CaO 59.25 40.61
MgO 1.15 2.93
SO
3
3.25 1.20
K
2
O 0.79 1.37
Na
2
O 0.26 1.00
LOI 11.62 34.23
LOI =loss on ignition at 1000 jC.
Fig. 1. Grain size distribution curves of the aggregate fractions.
Table 2
Physical properties of the aggregate fractions
Aggregate
fractions
Specific
gravity (kg/m
3
)
Water
absorption (%)
Fraction passing
75-Am sieve (%)
Natural sand 2470 1.1 1.0
Crushed aggregate 2710 0.7 0.0
Table 3
Concrete mixture proportions
Water/cement 0.40
Water/fine material 0.34
Aggregate/cement 3
Ingredients Mixture proportions (kg/m
3
)
Water 200
Cement 500
Limestone powder 80
Natural sand 1080
Crushed aggregate 420
Steel fibers 50
Superplasticizer 9.4
V. Corinaldesi, G. Moriconi / Cement and Concrete Research 34 (2004) 249254 250
500 mm (t
500
) and the elapsed time to gain the final
configuration (t
fin
) were also recorded. The results obtained
are reported in Table 5. The presence of a halo of cement
paste around the cake was not observed. The concrete had
enough deformability under its own weight (strictly related
to the U
fin
value), and had quite a high viscosity (related to
the t
fin
value), which is necessary to avoid segregation of
coarse aggregate particles.
3.2. L-box test
The ability of the concrete of compacting itself under its
own weight was evaluated by means of the L-box with
horizontal steel bars. Both the difference in the concrete
level, between the beginning and the end of the box (DH
fin
),
and the elapsed time to establish the final configuration (t
fin
)
were detected. The results obtained are reported in Table 5.
The concrete showed satisfactory, but not excellent, results
in terms of mobility in narrow sections; in fact, the differ-
ence in the concrete level between the beginning and the end
of the box was significantly higher than the maximum
required, equal to about 30 mm. Nevertheless, for producing
the thin precast elements, this workability level resulted
adequate, due to the absence of reinforcing steel bars inside
the formwork.
3.3. Compression and flexure tests
Cubic specimens, 100 mm in size, and prismatic
specimens (100 100 500 mm) were manufactured
according to Italian Standards UNI 6130-72 for compres-
sion and bending tests, respectively; these specimens
were cast in stainless steel forms and wet cured at 20
jC. Compression tests according to Italian Standards UNI
6132-72 were carried out on cubic specimens, which
were tested at right angles to the casting position. The
specimens were loaded at a constant strain rate until
failure. Due to the lack of an Italian standard test for
measuring direct tension in concrete, it was preferred to
measure the tensile strength of concrete by subjecting the
concrete beam (100 100 500 mm) to flexure with a
span of 400 mm (according to B.S. 1881:1970 prescrip-
tions). The maximum tensile stress reached at the bottom
of the middle cross section of the tested specimen; in
other words, the Modulus of Rupture (MOR) was also
evaluated. At different curing times (1, 3, 7, 28 and 180
days), three specimens were used for each mechanical
test, according to Italian Standards. Fig. 2 shows both
compressive and flexural strengths as a function of curing
time up to 180 days. The requested compressive strength
level of 40 MPa, at 28 days of curing, was easily
achieved. Concerning the flexural strength, the contribu-
tion of the steel fibers did not appear to be evident, due
to the low dosage of fibers employed. No ductility
measurement was carried out.
3.4. Drying shrinkage test
Prismatic specimens (100 100 500 mm) were pre-
pared according to Italian Standard UNI 6555-73 Hy-
draulic Shrinkage Determination. After 1 day of wet
curing, the specimens were stored at constant temperature
(20 F2 jC) and constant relative humidity (50 F2%)
while measuring drying shrinkage at different curing times.
Fig. 3 shows the results obtained up to 180 days. A drying
shrinkage of 500 Am/m can be predicted after 1 year of
exposure to a relative humidity of about 50%. The
effectiveness of steel fibers addition in counteracting
drying shrinkage of concrete becomes evident because
for a concrete with the same mixture composition without
Table 4
Concrete composition by volume percentage (%)
Ref-0.40 SCC-0.40
Water 19.1 20.6
Fine material
a
15.6 19.1
Sand
b
20.6 40.9
Gravel
c
41.1 15.0
Steel fibers 0.6
Superplasticizer 0.6 0.9
Air 3.0 2.9
a
All materials passing through ASTM nj100 sieve of 150 Am.
b
All materials that can pass the sieve opening of 5 mm, keeping into
account that almost 10% of crushed aggregate passes this sieve opening, as
observable in Fig. 1.
c
All materials that cannot pass the sieve opening of 5 mm.
Table 5
Fresh concrete workability: results of slump flow and L-box tests
Slump flow Slump (mm) 290
U
fin
(mm) 650
t
500
(s) 2
t
fin
(s) 30
L-box DH
fin
(mm) 90
t
fin
(s) 30
Fig. 2. Compressive and flexural strengths of the concrete as a function of
curing time.
V. Corinaldesi, G. Moriconi / Cement and Concrete Research 34 (2004) 249254 251
fibers, a drying shrinkage of about 800 Am/m after 6
months can be expected on the basis of well-consolidated
data by Lea [9]. Moreover, this result is further confirmed
by other researchers independently of the kind of concrete,
such as lightweight or fly ash or recycled aggregate and so
on, as reported in the literature [1013].
3.5. Carbonation and chloride penetration depths
The carbonation depth was measured by the phenol-
phthalein test (RILEM CPC18) on cubic concrete speci-
mens, 100 mm in size, exposed to air at a temperature of 20
jC after demoulding for 1 day. No carbonation was detected
up to 6 months of exposure to air. Fig. 4 shows the chloride
penetration depth as a function of the time of exposure to a
10% sodium chloride aqueous solution after water saturation
of concrete specimens. Data are reported after 1-month
immersion, when instability phenomena due to chloride
interference with the cement matrix [14] can be considered
over. The chloride penetration into concrete was evaluated
through the silver nitrate and fluorescein test [15] on cubic
concrete specimens, 100 mm in size, exposed to a 10%
sodium chloride aqueous solution after a wet curing of 1
week and an air curing of 3 weeks at a temperature of 20
jC. Both solutions were sprayed on the two fractured
surfaces obtained by splitting concrete specimens. Colle-
pardi et al. [15] found that chloride penetration depth (x)
varies with the elapsed time (t), following equation obtain-
able from the solution of Ficks second law under unsteady-
state conditions for diffusion in a semi-infinite solid:
x 4

D t
p
where D is the diffusion coefficient of chloride ions into the
concrete pores filled with water, expressed in 10
8
cm
2
/s.
The value of D, obtained from the equation by interpolating
the results showed in Fig. 4, is 0.2510
8
cm
2
/s at a
temperature of about 10 jC. Fig. 5 shows two images
obtained by scanning electron microscopy observations on
concrete samples after 130 days of immersion in the sodium
chloride solution. In the first case (a), the selected region is
located 2 mm away from the external surface of the specimen
in direct contact with the aggressive solution, whereas in the
second case (b), it is placed 30 mm from its edge, and
consequently, the steel fiber no longer comes into contact
with chloride ions. As a matter of fact, from a morphology
point of view, there is slight evidence of corrosion in (a),
while a sound fiber shape is detectable in (b).
Fig. 4. Chloride penetration as a function of the time of exposure to a 10%
sodium chloride aqueous solution.
Fig. 5. Images of a fiber located at only 2 mm from the outer surface of a
specimen exposed for 130 days to a 10% sodium chloride solution (a) and
of a fiber located 30 mm inside the same specimen (b), both obtained by
scanning electron microscopy at a magnification of 50.
Fig. 3. Drying shrinkage measurements up to 180 days of exposure to 50%
relative humidity.
V. Corinaldesi, G. Moriconi / Cement and Concrete Research 34 (2004) 249254 252
3.6. Pore structure characterization
Three samples were picked up from the concrete speci-
mens after 90 days of wet curing, for testing them by means
of the mercury intrusion technique. These samples consisted
of only hydrated cement paste, after excluding both fibers
and aggregate particles. The mean results are included in
Table 6, where not only the total open porosity is reported
but also the porosities corresponding to certain pore size
ranges. In particular, attention was focused on those pore
size ranges that were indicated by Litvan [16,17] as detri-
mental to frost resistance or effective for it. The results
obtained showed a large amount of pores belonging to both
detrimental and effective ranges; therefore, no useful infor-
mation on frost resistance of concrete can be extrapolated.
3.7. Frost resistance
The frost resistance was measured according to the
Resistance of Concrete to Rapid Freezing and Thawing
Standard Test Method (ASTM C 666) following Procedure
B: Rapid Freezing in Air and Thawing in Water on
concrete cubic specimens (100 mm in size), wet cured for
90 days. Mass and dynamic elastic modulus of the concrete
specimens were measured every 30 cycles in order to
evaluate both the relative dynamic modulus (RDM) of
elasticity and the percentage of mass change.
Six concrete specimens were tested; three of them were
previously treated on their surface with an hydrophobic
agent, such as a commercial product based on siloxane
resins, in order to evaluate any decrease in water absorption
for treated specimens and the consequent beneficial effect
on vulnerability to freezing and thawing cycles. The results
obtained up to 300 cycles showed that after 200 cycles, a
significant trend towards low values of the RDM was
detected in every case, especially for the untreated speci-
mens (see Fig. 6). The durability factors result equal to 52
and 67, respectively, in the case of the absence and the
presence of a hydrophobic surface, respectively, thus indi-
cating a certain benefit related to the use of an hydrophobic
agent. The mass change of the concrete specimens was also
monitored, and the values obtained, expressed as a percent-
age with respect to the initial saturated condition, are
reported in Fig. 7. The mass of the concrete specimens
slightly increased up to 120 cycles (about 50% lower mass
change was detected in the case of surface-treated speci-
mens); subsequently, a fairly constant value was maintained.
This may be explained by water absorption in the first
period (less evident for hydrophobically treated specimens)
and by the progressive depletion of this phenomenon;
concrete scaling was never observed.
4. Conclusions
The concrete, which was prepared to manufacture thin
precast elements for nonstructural applications, met both the
self-compaction and mechanical requirements. The fiber
addition proved to be very effective in counteracting drying
shrinkage of self-compacting concrete, which is usually a
great problem for this material, rich in powders and poor in
the coarse aggregate fraction. Due to the very low porosity
of its cementitious matrix, the rate of chloride ion diffusion
was low. The resistance to freezing and thawing was
Fig. 7. Percentage of mass change of both untreated and hydrophobic
concretes.
Fig. 6. RDM of elasticity of both untreated and hydrophobic concretes.
Table 6
Pore size distribution of the self-compacting concrete after 90 days of wet
curing
Pore diameter (nm) Porosity (% by volume)
1100 0.65
1001000 8.84
100010,000 6.23
>10,000 0.98
Total 16.70
4300
a
5.39
3502000
b
3.44
a
Pore size range detrimental to frost resistance, according to Litvan
[16].
b
Pore size range effective for frost resistance, according to Litvan [17].
V. Corinaldesi, G. Moriconi / Cement and Concrete Research 34 (2004) 249254 253
moderate and can be improved by the superficial application
of a hydrophobic agent, which notoriously reduces water
ingress into concrete. However, reminding of the type of
applications to which the thin precast elements are devoted,
self-compacting concrete durability is most satisfactory and
it appears competitive with other materials in manufacturing
these elements.
References
[1] M. Collepardi, A very close precursor of self-compacting concrete,
in: V.M. Malhotra (Ed.), Suppl. Papers of the Third CANMET/ACI
Int. Symp., San Francisco, ACI, Farmington Hills, MI, USA, 2001,
pp. 431450.
[2] M. Collepardi, R. Khurana, M. Valente, Construction of a dry dock
using tremie superplasticized concrete, in: V.M. Malhotra (Ed.),
Superplasticizers and Other Chemical Admixtures in Concrete, ACI
SP-119, Farmington Hills, MI, USA, 2001, pp. 471492.
[3] B. Van Khanh, D. Montgomery, Drying shrinkage of self-compacting
concrete containing milled limestone, in: A. Skarendahl, O. Petersson
(Eds.), 1st International RILEM Symposium on Self-Compacting
Concrete, Stockholm, Sweden, RILEM Publications S.A.R.L., 1999,
pp. 509521.
[4] K.H. Khayat, Y. Roussel, Testing and performance of fiber-reinforced,
self-consolidating concrete, in: A. Skarendahl, O. Petersson (Eds.),
Self-Compacting Concrete, 1999, pp. 509521, Stockholm, Sweden.
[5] P. Grouth, D. Nemegeer, The use of steel fibres in self-compacting
concrete, in: A. Skarendahl, O. Petersson (Eds.), 1st International
RILEM Symposium on Self-Compacting Concrete, Stockholm, Swe-
den, RILEM Publications S.A.R.L., 1999, pp. 497507.
[6] S. Grunewald, J.C. Walraven, Self-compacting fibre-reinforced con-
crete, Heron 46 (3) (2001) 201206.
[7] M. Ouchi, Y.A. Edamatsu, Simple evaluation method for interaction
between coarse aggregate and mortar particles in self-compacting con-
crete, in: A. Skarendahl, O. Petersson (Eds.), 1st International RILEM
Symposium on Self-Compacting Concrete, Stockholm, Sweden,
RILEM Publications S.A.R.L., 1999, pp. 121130.
[8] K.H. Khayat, Z. Guizani, Use of viscositymodifying admixture
to enhance stability of fluid concrete, ACI Mater. J. 94 (4) (1997)
332340.
[9] F.M. Lea, The Chemistry of Cement and Concrete, 3rd ed., Edward
Arnold (Publishers) Ltd, London, 1970, p. 408 (Table 69).
[10] J.-C. Chern, C.-H. Young, Compressive creep and shrinkage of
steel fibre reinforced concrete, Cem. Concr. Compos. 11 (4)
(1989) 205214.
[11] A.A. Alekrish, S.H. Alsayed, Shrinkage of fiber and reinforced fiber
concrete beams in hot dry climate, Cem. Concr. Compos. 16 (4)
(1994) 299307.
[12] O. Kayali, M.N. Haque, B. Zhu, Drying shrinkage of fibre-reinforced
lightweight aggregate concrete containing fly ash, Cem. Concr. Res.
29 (11) (1999) 18351840.
[13] H.A. Mesbah, F. Buyle-Bodin, Efficiency of polypropylene and me-
tallic fibres on control of shrinkage and cracking of recycled aggregate
mortars, Constr. Build. Mater. 13 (8) (1999) 439447.
[14] M. Pauri, S. Monosi, I. Alvera`, M. Collepardi, Assessment of free and
bound chloride in concrete, Mater. Eng. 1 (2) (1990) 497501.
[15] M. Collepardi, A. Marcialis, R. Turriziani, Penetration of chloride ions
into cement pastes and concretes, J. Am. Ceram. Soc. 55 (10) (1972)
534535.
[16] G.G. Litvan, Pore structure and frost susceptibility of building materi-
als, in: S. Modry, M. Svata (Eds.), Pore Structure and Properties of
Materials, Proceedings of the RILEM/IUPAC International Symposi-
um, 1821 September 1973, Prague, vol. II, Academia, Prague,
Czechoslovakia, 1974, pp. F-17F-30 (Final Report).
[17] G.G. Litvan, Air entrainment in the presence of superplasticizers,
ACI J. 80 (4) (1983) 326331.
V. Corinaldesi, G. Moriconi / Cement and Concrete Research 34 (2004) 249254 254
INSTITUTE OF PHYSICS PUBLISHING SMART MATERIALS AND STRUCTURES
Smart Mater. Struct. 14 (2005) S39S51 doi:10.1088/0964-1726/14/3/006
Self-diagnosis of structures strengthened
with hybrid carbon-ber-reinforced
polymer sheets
Z S Wu
1
, C Q Yang
1,3
, T Harada
1
and L P Ye
2
1
Department of Urban and Civil Engineering, Ibaraki University, Nakanarusawa-cho 4-12-1,
Hitachi 316-8511, Japan
2
School of Civil Engineering, Tsinghua University, Beijing 100084,
Peoples Republic of China
E-mail: ycq@hcs.ibaraki.ac.jp
Received 2 April 2004, in nal form 28 February 2005
Published 26 May 2005
Online at stacks.iop.org/SMS/14/S39
Abstract
The correlation of mechanical and electrical properties of concrete beams
strengthened with hybrid carbon-ber-reinforced polymer (HCFRP) sheets
is studied in this paper. Two types of concrete beams, with and without
reinforcing bars, are strengthened with externally bonded HCFRP sheets,
which have a self-structural health monitoring function due to the electrical
conduction and piezoresistivity of carbon bers. Parameters investigated
include the volume fractions and types of carbon bers. According to the
investigation, it is found that the hybridization of uniaxial HCFRP sheets
with several different types of carbon bers is a viable method for enhancing
the mechanical properties and obtaining a built-in damage detection
function for concrete structures. The changes in electrical resistance during
low strain ranges before the rupture of carbon bers are generally smaller
than 1%. Nevertheless, after the gradual ruptures of carbon bers, the
electrical resistance increases remarkably with the strain in a step-wise
manner. For the specimens without reinforcing bars, the electrical behaviors
are not stable, especially during the low strain ranges. However, the
electrical behaviors of the specimens with reinforcing bars are relatively
stable, and the whole range of self-sensing function of the
HCFRP-strengthened RC structures has realized the conceptual design of
the HCFRP sensing models and is conrmed by the experimental
investigations. The relationships between the strain/load and the change in
electrical resistance show the potential self-monitoring capacity of HCFRP
reinforcements used for strengthening concrete structures.
(Some gures in this article are in colour only in the electronic version)
1. Introduction
Fiber-reinforced polymer (FRP) sheets, especially carbon-
ber-reinforced polymer (CFRP) as external reinforcements,
have been increasingly applied in repairing, strengthening
and upgrading the deteriorated and deteriorating civil
infrastructures. In contrast to the traditional strengthening
methods such as steel plate bonding, overlaying and jacking
3
Author to whom any correspondence should be addressed.
methods, the externally bonding of FRP sheets represents
a more applicable and effective technology because of its
benecial properties such as high strength and stiffness to
weight ratios, high corrosion resistance, easy operation and
other properties.
Meanwhile, FRP composite is a young group of materials
in comparison with metals and other materials, and has been
applied as external reinforcements to concrete structures just
recently. As a result, there is a lack of knowledge about the
0964-1726/05/030039+13$30.00 2005 IOP Publishing Ltd Printed in the UK S39
Z S Wu et al
long-term properties, load transfer modes, failure mechanisms
and failure criteria. Moreover, it has been realized that the
stressstrain behavior of FRP composites is linear-elastic up to
ultimate failures. Consequently, the energy release of CFRP
is mainly elastic and sudden, which often leads to a sudden
rupture of CFRP reinforced structures. Besides, the cost of
CFRP composite is also a concern for a wider engineering
application.
In general, two solutions may exist to reduce and even
avoid the inuence of such disadvantages. One is to make
FRP more ductile, and for this purpose bers with high
ductility, such as aramid and glass bers, can be used to
fabricate ductile hybrid FRP composites. An investigation
on hybrid FRP rods (Bakis et al 2001) indicated that the
hybridization of unidirectionally reinforced composites using
low elongation piezoresistive bers and high elongation
electrically inert bers was a viable method of obtaining
pseudo-ductile behavior with intrinsic capability to monitor
the mechanical health of the composites. Other hybrids
such as carbon/glass (Muto et al 2001), carbon/aramid (Qiu
and Schwartz 1993), carbon/polyethylene (Peijs and De
Kok 1993), and carbon/aramid/polyethylene (Somboonsong
et al 1998) had also been investigated to enhance the
mechanical properties to prevent the sudden ultimate ruptures.
The layers of sandwich hybrid composites were found to
be in favor of improving the smoothness of the stress
strain curves, due to the more graceful transfer of load
from one layer to another and gradual rupture of different
layers. However, for the applications in strengthening and
upgrading civil infrastructures, especially in improving the
serviceability behavior of concrete structures, the stiffness
of FRP reinforcements is also required highly in order to
gain a rational enhancement effect. To realize a certain
stiffness, for example to maintain a typical level such as steel
reinforcements, sometimes continuous bers with modulus
higher than that of steel reinforcement such as a high modulus
type of carbon bers may be applied.
The other solution is to apply structural health monitoring
technology or to devise more sensitive means to inspect the
FRP structures in real time for signs of abnormality. It is found
that the CFRP possesses intrinsically the damage detection
and health monitoring capacity by virtue of its electrical
conductivity and piezoresistivity, which may enhance the
value added of CFRP composites. Both dc and ac resistance
measurements (Kupke et al 2001, Abry et al 2001) may
be used as an indicator on the mechanical state of CFRP
structures. From the tensile experiments of unidirectional
HCFRP sheets and rods consisting of several types of carbon
bers with different moduli (Wu et al 2003a, 2003b, 2005), it
was revealed that the initiation and evolution of damage could
be detected by measuring the change in electrical resistance
without sophisticated discrete sensors and actuators. The
results also indicated that the electrical resistance of HCFRP
could change in a stepwise manner due to the gradual rupture
of different types of carbon bers. Thus, a useful warning of
the impending fracture could be provided.
The objective of this investigation is to develop high
performance concrete structures externally strengthened with
HCFRPsheets with intrinsic damage detection functions. First
a conceptual design of HCFRP reinforcements using different
types of carbon bers is outlined. Then an experimental
program has been carried out to characterize the electrical
and mechanical properties of HCFRP-strengthened structures
under three-point bending. The main variables include the
types and volume fractions of carbon bers, the structural
types and with or without steel reinforcement. During the
experimental program, some important ndings have been
obtained, among which for instance the different damage
stages, failure processes and ultimate failure modes of the
HCFRP-strengthened concrete structures can be identied
through the denite fractional increases in electrical resistance.
The whole range of sensing function of HCFRP-strengthened
RC beams can be realized due to the strain development and
gradual micro- and macro-ruptures of HCFRP sheets, and the
conceptual design of HCFRP sensing models is conrmed
by the experimental investigation of HCFRP-strengthened RC
structures. Moreover, for the specimens with larger numbers
of HM carbon bers the corresponding electrical resistance
becomes more and negative as the cyclic loading progresses,
while for the specimens with fewer HM carbon bers the
corresponding electrical resistance remains almost constant.
2. Conceptual design of HCFRP reinforcements
Extensive literature has been published on the mechanical
properties of hybrid FRP composites as addressed above. This
was mainly aiming at improving the mechanical properties
of FRP composites to avoid the brittle rupture rather than
developing self-health monitoring functions and concerning
the stiffness of FRP reinforcements. In the present study,
hybridizations with several types of carbon bers are employed
to enhance the mechanical properties especially the stiffness
behavior and develop structures with smart sensing function.
For the CFRPcomposites withonlyone kind of active material,
the distinctly identiable change in electrical resistance
accompanies the catastrophic ultimate failures of the CFRP
reinforcements and CFRP-strengthened structures, leading to
the loss of the self-health monitoring function.
Herein, the authors have designed a new type of HCFRP
reinforcements consisting of more than one kind of carbon
bers for strengthening effectively the concrete structures. In
general, the carbon bers can be classied into three types:
high modulus (HM), middle modulus (MM) and high strength
(HS) carbon bers. The Young modulus of CFRP composite
with HS carbon bers is considered to equate with normal
steel reinforcements. The HM and MM carbon bers are
expected to fracture in a lower strain range, and the fractures
will lead to a denite increase in electrical resistance. The
HS carbon bers are mainly used to carry further loading
after the macro-fracture of the HM and MM carbon bers.
Figure 1(a) schematically demonstrates the mechanical and
electrical models of HCFRPreinforcements consisting of three
or more types of carbon bers. It is shown that the mechanical
properties with pseudo-plastic behavior after the rupture of
HM bers can be improved, and the high stiffness, strength
and certain ductility can be realized through a rational design
on hybridization of proper types and amounts of bers. The
load carrying capability is lost and defunction develops step
by step due to the gradual rupture of different types of carbon
bers. As a result, the effectiveness of strengtheninga structure
S40
Self-diagnosis of structures strengthened with hybrid CFRP sheets
Strain
Gradual rupture of HM
and MM carbon fibers
Low strain range
Gradual rupture of HS carbon
fibers and ultimate rupture

R
/
R
0
Stepwise increase in resistance
due to the gradual macro-rupture
of different type of carbon fibers
Ultimate failure due to the
macro-rupture of HS fibers
Macro-fracture
of HM fibers
Strain softening
Macro-fracture
of MM fibers
Strain hardening
Micro-fractures
L
o
a
d
(a)
CFRP
Voltmeter (I) Low strain range before the
micro rupture of carbon fibers
Loading
(II) Micro fracture of carbon fibers
DC Generator
(III) Macro fiber rupture
Fractional increase in resistance due
to partial rupture of carbon fibers
Residual increase in resistance due to
the rupture of high modulus carbon
fibers
(b)
Figure 1. The electrical and mechanical models of hybrid CFRP composite (a); measuring and sensing principles (b).
under the serviceability state can be obtained and the sudden
ultimate failure may be avoided.
Due to the hybridization effects, the electrical signals
varying in a step-wise manner with the increase in strain due
to the fractures of carbon bers become easy to identify and
the strengthened concrete beams are still capable of carrying
further loading even after the rupture of the HM and MM
carbon bers. The measuring principles are based on the
gradual micro- andmacro-fractures of different types of carbon
bers, schematically shown in gure 1(b). In stage (I), the
electrical resistance varies slowly in proportion to the applied
strain due to the piezoresistivity of carbon bers. Stage (II)
corresponds to the gradual micro-ruptures of carbon bers,
where the fractures of carbon bers begin to nucleate and
the resistance increases at a rapid rate. In stage (III), the
electrical resistance increases denitely in a step-wise manner
due to the gradual macro-rupture of the HM, MM and HS
carbon bers, and the number, width and height of the steps
are in close relation to the types and volume fractions of
carbon bers. During this stage the warning of impending
catastrophic ruptures can be established. After the rupture of
HM and MM carbon bers, there still exists some residual
increase in electrical resistance after unloading. The residual
change in electrical resistance makes it possible for the HCFRP
composites to remember the peak strain or load during the
loading history.
Table 1. Properties of used continuous carbon ber sheets.
Tensile Youngs
Type of strength modulus Thickness Elongation Density
materials (GPa) (GPa) (mm) (%) (g cm
3
)
C1 3.40 230 0.111 1.478 1.80
C5 2.75 392 0.165 0.744 1.81
C7 1.90 540 0.143 0.352 2.10
C8 1.90 640 0.143 0.297 2.12
3. Experimental details
3.1. Fabrication of HCFRP sheets and preparation of
experimental specimens
Inthe experiments, different types of unidirectional continuous
pitch-based carbon ber sheets from Nippon Steel Composite
Co., Ltd were used for fabricating the HCFRP reinforcements.
The bers were chosen according to their tensile strength and
modulus. The variety and dispersion of the ultimate elongation
of different carbon bers were also taken into account in order
to obtain excellent mechanical properties as well as good self-
healthmonitoringfunction. Table 1lists the material properties
of the used carbon ber sheets according to the manufacturers
specications.
S41
Z S Wu et al
Table 2. Types and compositions of the HCFRPs. (The values
before and in (brackets) are number of layers and widths in cm,
respectively.)
Number of layers and widths of different sheets
Types C1 C5 C7 C8
A 1 (10) 1 (0.2)
B 1 (10) 1 (5)
C 1 (10) 1 (10)
D 1 (10) 1 (1) 1 (1)
E 1 (10) 1 (2) 1 (1)
F 1 (10) 1 (2) 1 (2)
G 1 (15) 1 (15)
H 2 (15 + 7.5) 1 (15)
I 3 (15 + 15 + 7.5) 1 (15)
Eight types of HCFRP reinforcements were designed and
used to study the inuence of volume fractions and types of
carbon bers on the mechanical and electrical performances
of HCFRP-strengthened concrete beams. Among them, ve
types of concrete beams are plain concrete beams, andthe other
three types are RC beams. The types and compositions of
the HCFRP reinforcements are shown in table 2. The cement
used was ordinary Portland cement and the concrete had a
compressive strength of about 24.5 MPa. The cement, water
and other admixtures were mixed by a rotary mixer and poured
into the oiled molds. An external electrical vibrator was used
to facilitate compaction and reduce the amount of air bubbles.
The specimens were demolded after 7 days and cured in air
at room temperature for 28 days. Before externally bonding
carbon ber sheets, the bottom surfaces of the specimens were
polished with a diamond sander and ller putty was used to
create a smooth substrate surface. Carbon ber sheets were
adhered to the bottom surface of concrete beams with epoxy
adhesive. Prior to the loading test, the HCFRP-strengthened
specimens were cured in air at room temperature for more
than one week. Figures 2(a) and (b) show the schematic view
and dimensions of the strengthened beams without and with
reinforcing bars, respectively. For simplicity, the specimen
strengthened with the HCFRP reinforcement of type X is
written as specimen Xs in the following sections.
3.2. Loading and measuring systems
Figure 2(c) shows the electrical resistance measuring assembly
and the distribution of strain gages on the bottom surface of
the HCFRP-strengthened beams. The electrical resistance (R),
along the loading axis, was measured by means of the two-
probe method. In the experiments, four thin copper foils were
used as electrical electrodes and silver paste was applied in
order to ensure a good electrical contact between the specimen
surfaces and copper electrodes. Both cut ends of the HCFRP
reinforcements were lled with conductive resin to improve
the electrical conduction among all the conductive bers in
cross-section. A small direct current (DC) of about 5 mA was
introduced into the specimens from the copper electrodes, and
the voltage across the two electrodes was measured by using a
digital voltmeter with an accuracy of 0.001 mV. According to
Ohms law, the electrical resistance of HCFRP reinforcements
can easily be obtained. Longitudinal strain was measured by
employing strain gages with a gage length of 5 mm. The
Table 3. Typical data of experimental results.
HCFRP Ultimate load R/R
0
(%) due to
types R
0
() (kN) the rst fracture
A 1.10228 25.1 0.84
B 0.51654 26.8 1.61
C 0.78998 27.6 2.12
D 0.72836 29.6 3.96
E 0.65724 25.4 5.53
F 0.56346 31.6 12.3
bending tests of the HCFRP-strengthened concrete beams
were conducted on a hydraulic universal testing machine (UH-
500KNI) under a load control mode at a loading rate of
1 kN min
1
. After every loading step of 3 kN, the specimens
were unloaded and reloaded to the next step. The loading rate
was lowered when the specimens approached fracture. During
the tests, the load, strain and DC electrical resistance were
recorded simultaneously. Sustained loading and unloading
tests were carried out to investigate the inuence of loading
history on strainresistance behavior of the specimens.
4. Results and discussion
The main purpose of the present investigations is to develop
a self-sensing and diagnosis function in concrete structures
externally strengthened with HCFRP sheets. Since a
structural health monitoring technology relies on identifying
the difference among the measured values at different times
to estimate the health state of the structures, the stability of
measurement is important. To keep the values of electrical
resistance relativelystable, the appliedcurrent was deliberately
kept low. Consequently, the compensation for the changes
in temperature due to electrical current was not taken into
consideration in the present investigation.
4.1. Concrete beams without reinforcing bars
The experimental results of the HCFRP-strengthened concrete
specimens without reinforcing bars are summarized as table 3.
From the table, it is seen that the initial values of electrical
resistance increase with the increase in the amount of MM
and HM carbon bers (C5 and C7). It is attributed to the
reason that the increase in amount of HM and MM carbon
bers results in an increase in the cross-section area of the
HCFRP reinforcements. One can also nd that the values of
the rst jumps in electrical resistance due to the rst rupture
of the HM carbon bers increase as the amount of HM carbon
bers increases. The jumps in electrical resistance are about
0.84%, 3.986%, 5.53% and 12.3% with respect to the initial
resistance (R
0
) for specimens As, Ds, Es and Fs, respectively.
4.1.1. Cyclic loading results of HCFRP-strengthened concrete
beams in low strain ranges.
Specimens with two types of carbon bers. Specimens As,
Bs and Cs with two types of carbon bers were designed and
fabricated to investigate the inuence of volume fraction of
HM carbon bers on the electrical and mechanical properties
of HCFRP-strengthened concrete structures. The fractional
increase in electrical resistance is indicated as R/R
0
in
S42
Self-diagnosis of structures strengthened with hybrid CFRP sheets
90cm
80cm
70cm
10cm
15cm
HCFRP
Concrete beam
Electrodes
sheets
(a)
210
2 D13
2 D13
24
Unit: cm
4
4
12 20
15
8
15 15 5 5
180
HFRP sheets
170
(b)
90cm
35cm
6cm 6cm 6cm 6cm 6cm
strain gages
5cm
Support
Support
5cm
DC generator
mV
(c)
10 (@8 23)

Figure 2. Schematic dimension of HCFRP-sheet-strengthened concrete beams: (a) without steel reinforcements, (b) with steel
reinforcements. (c) Typical assembly for measuring the strain and electrical resistance.
the following sections. Figures 35 show the evolutions of
average strain and R/R
0
for specimens As, Bs and Cs in low
strain ranges before the fractures of carbon bers. From the
gures, it can be seen that for specimen As with fewer HM
carbon bers (C7) the corresponding R/R
0
almost remains
constant as cyclic loading progresses in a low strain range
before the micro-fractures of carbon bers (as seen in gure 3),
while for specimens Bs and Cs with larger amounts of C7 the
corresponding R/R
0
becomes more and more negative as
the cyclic tension progresses. Especially for specimen Cs, the
maximum negative increase in electrical resistance is up to
about 5% (gure 5).
Specimens with three types of carbon bers. The evolutions
of strain and R/R
0
with time for specimens Ds, Es and Fs
are shown in gures 68, respectively. For specimen Ds with
fewer HMcarbonbers (C5andC8) the correspondingR/R
0
remains nearly constant as the cyclic tension progresses.
However, for specimens E and F with more HM bers,
the corresponding R/R
0
decreases gradually as the cyclic
tension progresses.
The comparison of the specimens consisting of three types
of carbon bers with those of two types of carbon bers reveals
the same conclusion, that the corresponding R/R
0
of the
specimens withmore HMcarbonbers becomes more negative
as the cyclic tension progresses. However, for the specimens
with fewer HM carbon bers, the corresponding R/R
0
remains relatively constant as the cyclic tension progresses.
This is probably attributable to the fact that the increase in HM
carbon bers enlarges the area of interface between different
S43
Z S Wu et al
0
50
100
150
200
250
300
350
400
-0.05
-0.04
-0.03
-0.02
-0.01
0
0. 01
0. 02
0. 03
0. 04
0. 05
Average strain
Time
FCR
R/R
0

R
/
R
0

(
%
)

S
t
r
a
i
n

(

)

Figure 3. Variations of strain and R/R
0
with time for specimen
As under cyclic loading.
0
200
400
600
800
1000
1200
1400
1600
1800
2000
-0. 3
-0. 2
-0. 1
0
0.1
0.2
0.3

R
/
R
0

(
%
)

S
t
r
a
i
n

(

)

R/R
0
Time
Average strain
Figure 4. Variations of strain and R/R
0
with time for type Bs
under cyclic loading.
0
500
1000
1500
2000
2500
-6
-5
-4
-3
-2
-1
0
1
2
3
4
Average strain
FCR R/R
0

R
/
R
0

(
%
)

S
t
r
a
i
n

(

)

Time
Figure 5. Variations of strain and R/R
0
with time for type Cs
under cyclic loading.
kinds of carbon bers, and the electrical contact will get better
as the cyclic tension processes. It is known that a good
electrical contact leads to a decrease in electrical resistance.
However, the details are still not clear. Such a phenomenon
was also observed in HCFRP sheets by the authors (Wu et al
2005) and short-carbon-ber-reinforced polymer matrix (Shui
and Chung 1996) subjected to unidirectional cyclic tension.
For the HCFRP composites, the electrical resistance became
relatively stable after about ten cycles.
4.1.2. Mechanical and electrical behaviors of HCFRP-
strengthened beams during the whole strain range. To
investigate the relationships between the mechanical and
-50
0
50
100
150
200
250
300
350
400
450
500
-0.04
-0.02
0
0.02
0.04
0.06
0.08
0.1
0.12
0.14

R
/
R
0

(
%
)

FCR R/R
0
S
t
r
a
i
n

(

)

Time
Average strain
Figure 6. Variations of strain and R/R
0
with time for type Ds
under cyclic loading.
-50
0
50
100
150
200
250
300
350
-0.25
-0.2
-0.15
-0.1
-0.05
0

R
/
R
0

(
%
)

S
t
r
a
i
n

(

)

FCR R/R
0
Time

Average strain
Figure 7. Variations of strain and R/R
0
with time for type Es
under cyclic loading.
0
400
800
1200
1600
2000
-1.5
-1
-0.5
0
0. 5
1
1. 5
2
2. 5
3
3. 5
4
Average strain
Time
R/R
0

R
/
R
0

(
%
)

S
t
r
a
i
n

(

)

Figure 8. Variations of strain and R/R
0
with time for type Fs
under cyclic loading.
electrical properties due to the gradual rupture of HM carbon
bers in large strain ranges, the specimens were loaded up to
the ultimate failure. The specimens used in this section are
the ones that had been tested in low strain ranges under cyclic
three-point bending tests in the above section.
The distributions of cracks in specimens AsCs are
schematically demonstrated in gures 9(a)(c), respectively.
From the gures, it can be seen that the ratio of shear cracks
S44
Self-diagnosis of structures strengthened with hybrid CFRP sheets
(a)
(b)
(c)
Photo 1. The ultimate failures of HCFRP-strengthened concrete beams without reinforcing bars: shear failure and debonding of HCFRP for
specimen Bs (a); concrete rupture for specimen Cs (b); the ultimate rupture of HCFRP for specimen Fs (c).
Concrete Cracks
Debonding
(a)
Concrete
Cracks
Debonding
(b)
Concrete
Cracks
(c)
Figure 9. A schematic illustration of crack distribution for
specimens As, Bs and Cs shown in (a), (b) and (c).
increases with the increase of HCFRP amount and the volume
fractions of HM bers (C7). Specimen As failed with the
propagation of exural cracks and the debonding of HCFRP
reinforcements fromthe endof a mainexural crackaroundthe
middle span of the beam. The ultimate fracture of specimen
Bs was caused by the propagation of shear cracks with the
debonding of HCFRP sheets from a dominant crack end.
Moreover, specimen Cs showed a complete shear failure of
HCFRP sheets near the end of the concrete beams. Typical
photos of the fractured specimens Bs and Cs are demonstrated
in photos 1(a) and (b). For specimens Ds and Es with three
types of carbon bers, the mechanical properties and failure
modes are similar to those of specimens As and Bs. Photo 1(c)
shows the ultimate failure of specimen Fs. The ultimate failure
of specimenFs was due tothe ruptures of HCFRPsheets around
the middle span, which was different from the failure modes
of the other ones.
The experiments indicate that the relationships between
R/R
0
and the applied strains are complicated and vary
in character from specimen to specimen partly due to
the deciency and property dispersion of concrete beams.
Consequently, the general features of the electrical and
mechanical behaviors are discussed in this section. Generally,
the relationships between R/R
0
and strains can be classied
intothree typical types, as showningures 1012, respectively.
Specimen Bs exhibits a linear fractional increase in R/R
0
with the average strain before 1900 (shown in gure 10(c)),
after which two sudden jumps in R/R
0
by about 0.5% and
25% take place. The decreases in load at 1900 and 3000
(shown in gure 10(b)) imply that the sudden jumps in R/R
0
S45
Z S Wu et al
0
1000
2000
3000
4000
5000
6000
0.0 5.0 10.0 15.0 20.0 25.0 30.0
-5
0
5
10
15
20
25
30
35
Maximum strain
Average strain
RE R/R0
M
i
c
r
o
-
s
t
r
a
i
n

(

R
/
R
0

(
%
)

R
/
R
0

(
%
)

Load (kN)
(a)
0
5
10
15
20
25
30
0 1000 2000 3000 4000
-5
0
5
10
15
20
25
30
35
Load
Macro-rupture of
FCER R/R0
L
o
a
d

(
k
N
)

Average strain ()
(b)

R
/
R
0

(
%
)

0.0
5.0
10.0
15.0
20.0
25.0
30.0
0 500 1000 1500 2000
-0.5
0
0.5
1
1.5
2
Load
FCER R/R0
L
o
a
d

(
k
N
)

Average strain ()
(c)
HM fibers (C7)
Figure 10. Evolution of R/R
0
and strain with load (a);
relationships between R/R
0
, load and average strain during the
whole strain range (b); in the low strain range (c) for specimen Bs.
are attributed to the bundle rupture of HM carbon bers. After
the ruptures, R/R
0
increases with the strain with a larger
slope. Such electrical and mechanical behaviors agree with the
HCFRP models as proposed in the second section. Specimen
Cs exhibits an unstable change in electrical resistance, and the
ultimate increase in R/R
0
is about 4%, much smaller than
those of specimens Bs and Fs. R/R
0
undulates during the
experiments; however, the sudden jumps in R/R
0
due to
the ruptures of HM carbon bers can be clearly identied in
gure 11. This kind of electrical behavior is rare during the
present experiments. Specimen Fs shows another electrical
behavior, the R/R
0
of which decreases linearly by about
2.5% before 1760 (shown in gure 12(c)). After this,
R/R
0
begins to increase after a jump in R/R
0
by about
12.3% due to the macro-rupture of C8 at a strain level of about
0
5
10
15
20
25
30
0 500 1000 1500 2000 2500
-6
-5
-4
-3
-2
-1
0
1
2
3
4
Load
Gradual ruptures of HM
and HS carbon fibers
FCER R/R0
L
o
a
d

(
k
N
)

R
/
R
0

(
%
)

Average strain ()
Figure 11. Evolution of R/R
0
and load with strain for specimen Cs.
1860 . This mode of electrical behavior for HCFRP-sheet-
strengthened plain concrete beams occurred frequently during
the experiments.
As mentioned, the electrical behavior of the HCFRP-
strengthened plain concrete beams is not very stable and the
change in electrical resistance is relatively small during the
low strain range. For specimen Bs and Fs, there exist big
jumps of electrical resistance due to the macro-ruptures of HM
and MM carbon bers. However, the HCFRP-strengthened
structures approach nal failure soon after the big jumps in
electrical resistance, which shortens the strain sensing range.
This is because the HCFRP sheets are the only reinforcements
in the HCFRP-strengthened concrete beams. Once the macro-
fracture of HCFRPsheets happens in the HCFRP-strengthened
structure, the structure itself becomes unstable. Consequently,
in order to obtain a stable electrical and mechanical behavior
of HCFRP-strengthened structures, it is necessary to ensure
that the HCFRP-strengthened structures are relatively stable
during the gradual macro-ruptures of HCFRP sheets.
4.2. HCFRP-strengthened RC beams
In order to assure a suitable stability of the HCFRP-
strengthened structures after the macro-ruptures of carbon
bers in HCFRP sheets, the HCFRP-strengthened RC beams
are fabricated and investigated. It is revealed that for the
HCFRP-strengthened RC beams the change in electrical
resistance is more stable than that of the ones without
reinforcing bars.
The distribution of cracks in specimen GsIs is shown in
gures 13(a)(c), respectively. The failure mode of HCFRP-
strengthened RC structures is dominated by mainly exural
failure with HCFRP ruptures, concrete crushing, partial
HCFRP sheet debonding (specimen Gs and Hs) and complete
debonding of HCFRP sheets from a dominant exural crack
endwithpartial HCFRPruptures (specimenIs). It is alsofound
that the exural cracks become slim, increase and distribute in
a wider span with the increase in the number of HS carbon
bers.
The relationships between R/R
0
, strain and the applied
load for specimens GsIs are demonstrated in gures 14
16, respectively. It can be seen that the reinforcing bars
improve greatly the deformation and load-carrying capability
of the HCFRP-strengthened RC beams. For these three
S46
Self-diagnosis of structures strengthened with hybrid CFRP sheets
0
1000
2000
3000
4000
5000
6000
7000
0.0 5.0 10.0 15.0 20.0 25.0 30.0 35.0
-10
0
10
20
30
40
50
Maximum strain
Average strain
R/R0

R
/
R
0

(
%
)

M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
(a)
0.0
5.0
10.0
15.0
20.0
25.0
30.0
35.0
0 1000 2000 3000 4000 5000
-10
0
10
20
30
40
50
load

R
/
R
0

(
%
)

R
/
R
0

(
%
)

FCER R/R0
Macro-rupture of C8
Macro-rupture
of C5 and C1
Macro-rupture of C8
L
o
a
d

(
k
N
)

Average strain ()
(b)
18.0 3
R/R0
load
16.0
2.5
14.0
1.5
12.0 2
L
o
a
d

(
k
N
)

10.0
8.0
6.0 1
4.0
0.5
2.0
0.0 0
0 200 400 600 800 1000
Average strain ()
1200 1400
(c)
Figure 12. Evolution of R/R
0
and strain with load (a);
relationships between R/R
0
, load and average strain during the
whole strain range (b) and in the lowstrain range (c) for specimen Fs.
specimens, the electrical resistance increases monotonically
with the applied load during the whole strain/load ranges,
whose sensing range is much wider than that of the HCFRP-
strengthened plain concrete beams. The fractional increases
in electrical resistance for specimens GsIs are about 0.7%,
1.19% and 0.6%, respectively, before the macro-ruptures of
carbon bers.
It is also shown in gures 1416 that the electrical
resistance is related to the mechanical behavior of the HCFRP-
strengthened RCstructures and changes in a distinct step-wise
manner, corresponding to the gradual ruptures of different
types of carbon bers in the HCFRP reinforcements. After
every macro-rupture of carbon bers, the micro-strain and
electrical resistance will change with the applied load with
a larger slope. From the results, one can nd that the electrical
and mechanical behaviors of the HCFRP-strengthened RC
structures agree well with the electrical and mechanical models
of HCFRP composites as proposed in gure 1. Therefore, the
variation in R/R
0
can be regarded as an indicator of the
state of integrity of the HCFRP-strengthened structures, and
the jumps in electrical resistance can also be used to give a
necessary warning of the impending failure.
The curves of R/R
0
versus average micro-strain(as seen
in gures 14(b)16(b)) reveal that after the distinct jumps of
electrical resistance due to the macro-rupture of HM carbon
bers in the HCFRP sheets the HCFRP-strengthened RC
beams can sustain the load in a broad range of strains. It is
the stability of the strengthened RC structures after the macro-
ruptures of HCFRP sheets that enlarges the sensing range of
the strain/load.
For the HCFRP-strengthened RC beams, the reinforce-
ments include the reinforcing bars and HCFRP sheets, so in
general the gradual ruptures of HCFRP sheets do not have a
fatal inuence on the structural stability. However, for the
HCFRP-strengthened plain concrete structures, the HCFRP
sheets are the only reinforcements, and the ruptures of HCFRP
reinforcements have great inuence on the structural stabil-
ity. After the gradual rupture of HCFRP sheets, the HCFRP-
strengthened RC structures can also sustain load stably. As
a result, the HCFRP sheets of the HCFRP-strengthened RC
beams can be fully tensioned and fail step by step due to the
strain development and gradual rupture of different types of
carbon bers in HCFRP reinforcement before the nal fail-
ure of the structures. These are the reasons why the electri-
cal behavior with a larger ultimate value of R/R
0
for the
HCFRP-strengthened RC beams is more stable than that for
the HCFRP-strengthened plain concrete beams.
4.3. Damage identication through resistance measurement
It is conclusive that the electrical resistance changes with the
strain development and gradual ruptures of different types of
carbon bers in HCFRP reinforcements. Consequently, the
damage degree and mechanical state of HCFRP-strengthened
concrete structures can be identied through measuring the
change in electrical resistance.
Generally, the electrical behavior of HCFRP-strengthened
RC structures with a larger ultimate value of electrical
resistance is more stable than that of the HCFRP-strengthened
plain concrete structures because there are two types of
reinforcements including reinforcing bars and HCFRP sheets
in the HCFRP-strengthened RC structures. The reinforcing
bars are in favor of maintaining the structural stability after
the gradual ruptures of HCFRP reinforcements, so the gradual
ruptures of HCFRP sheets do not have fatal inuences on
the structural stability. The HCFRP reinforcements can be
fully tensioned and the damage of HCFRP reinforcements
can be accumulated up to the ultimate ruptures, which can
be conrmed by the ultimate maximum strains within a range
from 12 000 to 17 000 for the HCFRP-strengthened RC
structures (as shown in gures 14(a)16(a)), much larger than
those of HCFRP-strengthened plain concrete beams. As a
result, the electrical and mechanical behaviors can experience
S47
Z S Wu et al
Concrete
Rupture with a small partial debonding of HCFRP sheets
Flexural cracks
HCFRP sheets
Crushing of concrete
(a)
Concrete
Rupture with a small partial debonding of HCFRP sheets
Flexural cracks
HCFRP sheets
Crushing of concrete
(b)
Concrete
Debonding with a small partial rupture of HCFRP sheets
Flexural cracks
HCFRP sheets
(c)
Figure 13. The schematic illustration of crack distribution for specimens Gs, Hs and Is shown in (a)(c).
every sensing stage as proposed in gure 1 and agree well
with the proposed conceptual sensing models of HCFRP-
strengthened structures.
However, for the HCFRP-strengthened plain concrete
beams, the HCFRPsheets are the onlyreinforcement. After the
macro-fractures of HCFRP sheets, the cracks will propagate
quickly and the structural stability will be impaired greatly.
Consequently, the HCFRP sheets are not fully tensioned and
fracture at much lower strain range. This is the reason why
the electrical behavior is less stable and the ultimate values of
strain and R/R
0
are much smaller than those of the HCFRP-
strengthened RCstructures. In this section, the ultimate values
of R/R
0
and strain are the values corresponding to the
maximum load during loading. It is apparent that in order
to get a stable and broad sensing range of strain, the structures
should be relatively stable after the gradual ruptures of HCFRP
sheets.
The HCFRP-strengthened structures may have a nal
failure mode of HCFRP rupture at the mid-span or HCFRP
debonding from a dominant crack end depending on the
amount of HCFRP reinforcements used and the volume
fractions of each type of carbon bers. In general, the rupture
of HCFRP sheets corresponds to a large ultimate value of
R/R
0
, and the debonding of HCFRP sheets corresponds to
a small ultimate value of R/R
0
. For example, the ultimate
values of R/R
0
are different for the specimens Bs, Cs and
Fs without reinforcing bars. For specimen Cs, the HCFRP
reinforcement was not severelydamaged because the specimen
failed in complete shear nearby one end of the concrete beam,
and it is reasonable that the ultimate value of R/R
0
is small.
For specimen Fs, the ultimate failure was due to the sudden
rupture of HCFRP reinforcements around the middle span,
which resulted in a large ultimate value of R/R
0
.
In cases of HCFRP-strengthened RC structures, the main
failures of specimen Gs and Hs are due to the exural rupture
of HCFRP sheets, while the main failures of specimen Is is
due to the debonding of HCFRP sheets from the end of a
dominant exural crack, as shown in photo 2. The ultimate
failure of HCFRP debonding for specimen Is corresponds to
a ultimate R/R
0
of about 100%; while the ultimate failure
of HCFRP occurs around the mid-span in specimens Gs and
Hs, generating an ultimate R/R
0
of about 250% and 390%,
respectively. The results imply that the ultimate failure mode
of HCFRP-strengthened structures can be identied through
measuring the variation in electrical resistance.
5. Conclusions
Through the experimental investigations, it is demonstrated
that a suitable hybridization of unidirectional HCFRP sheets
with several kinds of carbon bers is a viable method for
strengthening and self-monitoring concrete structures. The
hybridization is not only in favor of improving the mechanical
properties to prevent the sudden ultimate fractures of the
HCFRP strengthened structures but also in favor of providing
an intrinsic health monitoring function in a distinct indication
S48
Self-diagnosis of structures strengthened with hybrid CFRP sheets
0
2000
4000
6000
8000
10000
12000
14000
16000
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0 80.0
0
50
100
150
200
250
300
Average strain
Gradual rupture of C1
Rupture of C7

R
/
R
0

(
%
)

R
/
R
0

(
%
)

R/R0
M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
(a)
0
500
1000
1500
2000
2500
3000
3500
4000
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1
Average strain
R/R0
M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
(b)
0
50
100
150
200
250
300
0 1000 2000 3000 4000 5000 6000 7000 8000
Gradual rupture of C1
Partial macro-rupture of C1
Macro-rupture of C7

R
/
R
0

(
%
)

Average strain ()
(c)
Maximum strain
Maximum strain
Figure 14. Evolution of R/R
0
and load with strain: during the
whole strain range (a); in the low strain range (b); relationship
between R/R
0
and average strain (c) for specimen Gs.
for the strengthened structures. The DC electrical resistance
measurement proves to be an effective method in detecting
and identifying the damage of HCFRP-strengthened concrete
structures, without sophisticated sensors and actuators. In
addition, the following conclusions are drawn.
(1) The change in electrical resistance before the micro- and
macro-fractures of carbon bers are considerably small.
The value before the rupture of micro-fracture of HM
carbon bers is less than 1%.
(2) For the specimens with fewer HM carbon bers such
as specimens As and Ds, the corresponding R/R
0
remains almost constant for different loading cycles, and it
increases linearly as the strain increases. However, for the
specimens withmore HMcarbonbers, the corresponding
R/R
0
becomes more negative as the cyclic loading
progresses.
0
2000
4000
6000
8000
10000
12000
14000
16000
18000
0.0 20.0 40.0 60.0 80.0 100.0
0
50
100
150
200
250
300
350
400
Maximum strain
Average strain
Gradual rupture of C1
Rupture of C7
R/R0
Load (kN)
(a)
0
500
1000
1500
2000
2500
3000
3500
4000
4500
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
0
0.2
0.4
0.6
0.8
1
1.2
1.4
Maximum strain
Average strain

R
/
R
0

(
%
)

R
/
R
0

(
%
)

R
/
R
0

(
%
)

R/R0
M
i
c
r
o
-
s
t
r
a
i
n

(

)

M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
(b)
0
50
100
150
200
250
300
350
400
450
0 2000 4000 6000 8000 10000
Macro-rupture of C7
Partial macro-rupture of C1
Average strain ()
(c)
Figure 15. Evolution of R/R
0
and load with strain: during the
whole strain range (a); in the low strain range (b); relationship
between R/R
0
and average strain (c) for specimen Hs.
(3) The values of the sudden jumps in R/R
0
due to the
rst fracture increase with the increase of the amount of
HM and MM carbon bers (C5, C7 and C8). The values
are about 0.84% and 12.3% for specimens As with fewer
HM carbon bers and specimen Fs with more HM carbon
bers, respectively.
(4) The relationships between mechanical and electrical
properties for a broad strain range are clearly correlated
in HCFRP-strengthened concrete beams due to the strain
development and gradual micro- and macro-ruptures of
HCFRP reinforcements. The sudden growth of HCFRP
strains due to the macro-rupture of HM carbon bers can
lead to a sudden jump in electrical resistance. The number
of big jumps is related to the number of carbon ber types.
All of these are expected to be of practical use.
(5) Generally, the electrical response of the HCFRP-
strengthened RC beams during a broad strain ranges
S49
Z S Wu et al
0
2000
4000
6000
8000
10000
12000
14000
0.0 20.0 40.0 60.0 80.0 100.0 120.0
0
20
40
60
80
100
120
Maximum strain
Average strain
Gradual rupture of C1
Rupture of C7

R
/
R
0

(
%
)

R
/
R
0

(
%
)

R
/
R
0

(
%
)

R/R0
M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
M
i
c
r
o
-
s
t
r
a
i
n

(

)

Load (kN)
(a)
0
500
1000
1500
2000
2500
3000
3500
4000
0.0 10.0 20.0 30.0 40.0 50.0 60.0 70.0
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
Maximum strain
Average strain
R/R0
(b)
0
20
40
60
80
100
120
0 1000 2000 3000 4000 5000 6000 7000 8000 9000
Gradual rupture of C1
Macro-rupture of C7
Average strain ()
(c)
Figure 16. Evolution of R/R
0
and load with strain: during the
whole strain range (a); in the low strain range (b); relationship
between R/R
0
and average strain (c) for specimen Is.
is more stable than that of the HCFRP-strengthened
plain concrete beam because of the structural stability
of HCFRP-strengthened RC structures after the gradual
macro-ruptures of the HCFRP reinforcements. Moreover,
the ultimate value of electrical resistance is much larger
in RC structures than in plain concrete structures due to
the extent of damage growth in HCFRP. Therefore, the
whole range of self-sensing function up to the macro-
rupture of HS carbon bers (at a maximum strain within
a range of about 12 00017 000 ) can be realized in
the HCFRP-strengthened RC beams, the electrical and
mechanical behaviors of which prove that self-diagnosis
of HCFRP-strengthened structures is feasible with an
integrated monitoring-strengthening system.
Acknowledgments
The nancial support of partial work of this paper from
Advanced Technology Research in China (863 National
(a)
(b)
(c)
Photo 2. The ultimate failures of HCFRP-strengthened concrete
beams with reinforcing bars: exural failure with ruptures of
HCFRP reinforcements for specimen Gs (a); exural failure with
ruptures of concrete and HCFRP reinforcements for specimen Hs
(b); exural failure with debonding of HCFRP reinforcements for
specimen Is (c).
Program) under 2001AA336010 and the NSFCJoint Research
Fund for Overseas Chinese Young Scholars under 50228808
are gratefully acknowledged.
References
Abry J C, Choi Y K, Chateauminois A, Dalloz B, Giraud G and
Salvia M 2001 In-situ monitoring of damage in CFRP
specimens by means of AC and DC measurements Compos.
Sci. Technol. 61 85564
Bakis C E, Nanni A, Terosky J A and Koehler S W 2001
Self-monitoring, pseudo-ductile, hybrid FRP reinforcement
rods for concrete applications Compos. Sci. Technol. 61
81523
Kupke M, Schutle K and Sch uler R 2001 Non-destructive testing of
FRP by dc and ac electrical methods Compos. Sci. Technol. 61
83747
Muto N, Araki Y, Shin S G, Matsubara H and Yanagida H 2001
Hybrid composites with self-diagnosing function for
preventing fatal fracture Compos. Sci. Technol. 61 87583
Peijs A A J M and De Kok J M M 1993 Hybrid composite based on
polyethylene and carbon bers part 6: tensile and fatigue
behavior Composites 24 1932
S50
Self-diagnosis of structures strengthened with hybrid CFRP sheets
Qiu Y and Schwartz P 1993 Micro-mechanical of
Kevlar-149/S-glass hybrid seven-ber micro-composites: I
tensile strength of the hybrid composite Compos. Sci. Technol.
43 289301
Shui X P and Chung D D D 1996 A piezoresistive carbon lament
polymermatrix composite strain sensors Smart Mater. Struct.
5 2436
Somboonsong W, Ko F K and Harris H G 1998 Ductile hybrid ber
reinforced plastic reinforcing bar for concrete structure: design
methodology ACI Mater. J. 95 65566
Wu Z S, Yang C Q and Takahashi T 2003a Damage detection of
hybrid CFRP reinforcements by DC measurement technique
Int. Conf. on Advanced Technology in Experimental Mechanics
2003 (Nagoya, Japan) OS09W0167 (CD-ROM)
Wu Z S, Yang C Q and Tobe Y H 2003b Self-diagnosis of concrete
beams reinforced with hybrid CFRP rods Proc. 4th Int.
Workshop on Structural Health Monitoring (Stanford) ed
F K Chang (Lancaster, PA: DEStech Publications) pp 15562
Wu Z S, Yang C Q and Tobe Y H 2005 Electrical and mechanical
characterizations of hybrid CFRP sheets Compos. Sci. Technol.
at press
S51
Building and Environment 41 (2006) 262267
The effects of atmospheric exposure on the fracture properties of
polymer concrete
J.M.L. Reis
a,
, A.J.M. Ferreira
b
a
Laboratorio de Mecanica Teorica e Aplicada, Universidade Federal Fluminense, Programa de Pos-Graduac- ao em Engenharia Mecanica,
UFF, Rua Passo da Patria, 156, Bloco. E, sala 216, 24210-260, Niteroi, Rio de Janeiro, Brasil
b
DEMEGIDepartamento de Engenharia Mecanica e Gestao Industrial, Faculdade de Engenharia da Universidade do Porto, FEUP,
Rua Dr. Roberto Frias 4200-465, Porto, Portugal
Received 11 November 2004; received in revised form 7 December 2004; accepted 9 December 2004
Abstract
In this experiment the effect of atmospheric exposure of epoxy and ber-reinforced epoxy polymer concrete was investigated to
evaluate its fracture properties, such as stress intensity factor, K
Ic
; and fracture energy, G
f
: The deterioration and structural
performance of polymer concrete were investigated in a real situation of exposure during a year period and compared the same
formulation in laboratory conditions. The relationship between year period, exposure time and load-bearing capacity of deteriorated
polymer concrete is studied and fracture mechanics of the specimens are discussed. From the tests results and discussion it is clear
that the material studied, polymer concrete, suffers a high deterioration when subjected to aggressive environments.
r 2005 Elsevier Ltd. All rights reserved.
Keywords: Polymer concrete; Environmental exposure; Fracture mechanics
1. Introduction
The deterioration of a material depends on how and
to what extent it interacts with its surroundings. The
outdoor environment if considered in terms of sunshine,
temperature, rainfall and wind, varies widely in dura-
tion, intensity and sequence. As far as the durability of
materials is concerned, weight should be given to severe
climatic conditions and depends on the condence level
required in the performance of the material, but in
general it is the time-averaged climatic factors which
should be considered. It is thought the prevalence of
certain elemental features for prolonged periods has to
be the basis of the concept of climate involved. Various
climatic classications have been developed to meet
particular human requirements: in agriculture to dene
vegetation and crop distribution, in transport to make
safer travel by land, sea and air and in medicine to relate
mans physical well-being and social development to his
environment. There is as yet no accepted classication of
climates for the deterioration of man-made materials are
in its infancy. However, examples such as the hydrolic
reversion of polyurethanes, the ozone-induced stress
cracking of rubbers, the decoloration of polymers by
sunlight and the dimensional variability demonstrate the
interaction between the environment and polymers. The
main components of weather which cause degradation
are sunlight, temperature, moisture, wind, dust and
pollutants.
Because of their impact on mans social economic
welfare, features of the weather have been measured
for many years and detailed records go back to more
than a hundred years in certain regions. Today,
meteorological data are collected at numerous sites
throughout the world through a network of global
stations, but concentration of this varies widely over
regions.
ARTICLE IN PRESS
www.elsevier.com/locate/buildenv
0360-1323/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.buildenv.2004.12.017

Corresponding author. Tel.: +55 216295588; fax: +55 216295585.


E-mail address: jreis@mec.uff.br (J.M.L. Reis).
The durability of any material is related to the
environment that it is exposed to UV radiation,
temperature, rain and wind represent large variations
in durability and it is related to the characteristics that
are intended to be obtained from the material that is
being used.
There is no climatic classication in terms of material
degradation, although the principal clime components
which cause degradation are solar radiation, rain,
humidity, temperature, wind, dust and contamination
[1].
The high level of temperature, humidity and radiation
combined make the places hot and wet (tropical) the
hardest clime for composite materials.
There are several countries which have their atmo-
sphere data recorded for several years, in our case the
reference that is going to be used is a meteorological
tower that Aveiro University has nearby the place where
tests were performed.
In this paper, the fracture properties of epoxy
polymer concrete and ber-reinforced polymer concrete
were tested by three-point bending tests after exposure
to different weather conditions. Polymer concrete
consists, in essence of an aggregate blend mixed
with a polymer resin in convenient proportions
which were studied previously by authors [2,3]
and are used in repairing or in constructing layers or
sections in reinforced concrete structures with high
ductility [4].
To perform these tests, the TPFM is used to evaluate
two independent-sized fracture parameters, the stress
intensity factor, K
Ic
; and critical crack tip opening
displacement CTOD
C
according to RILEM recommen-
dations [5,6]. Another fracture parameter calculated
from these tests was the fracture energy G
f
: The fracture
energy also is calculated following RILEM recommen-
dations [7].
2. Experimental program
2.1. Environmental condition
The meteorological tower from Aveiro University,
Portugal, was used to collect the temperature and
humidity from Aveiro city, which is an excellent
indication of the local weather condition during the
year.
There is another signicant issue of these tests: the
weather condition in Aveiro is overall similar tempera-
ture and humidity every combined season, registering
the lowest temperatures in January, in winter, and the
hottest in the beginning of September, summer. Accord-
ing to the meteorological tower the lowest registered
temperature was 5.9 1C in January and the highest was
20.3 1C in early September. The air humidity on these
precise days was 77% and 66%, respectively. The air
humidity during the year varied from 43% to 103% [8].
2.2. Test samples
Polymer concrete formulations were prepared by
mixing an epoxy resin with siliceous sand with rather
uniform granulometry.
The aggregate used in this study was foundry sand
which consists in siliceous sand, designed by SP55, from
SIBELCO used in the foundry industry, with uniform
granulometry, and an average diameter of 245 mm in
80% of mass. The epoxy resin system used was EPOSIL
551, from SILICEM based on a diglycidyl ether
bisphenol A and an aliphatic amine hardener it was
processed with a maximum mix ratio of hardener of 2:1,
in 20% of mass.
The reinforcement used in this research was 1%, in
mass, of E-glass ber provided from by PPG with no
sizing. The bers were emerged in a solution of 2% of
Silane A174 for adhesion treatment. Carbon ber
provided by TENAX, which is a PAN HTA 5131 ber,
with epoxy sizing was added in the percentage of 2%, in
mass. All bers have an average length of 6 mm. The
ber percentage was also investigated by the authors in
their previous works [9,10], in order to obtain higher
reinforcement performance.
Both reinforced and plain polymer concrete, with
these binder formulations and mix proportions were
mixed and moulded to specimens of size
30 mm60 mm280 mm, and then notched with
20 mm depth using a 2 mm diamond saw according to
RILEM standard TC113/PC2 [11]. All specimens were
allowed to cure for 7 days at room temperature and then
post-cured at 80 1C for 3 h.
2.3. Experimental procedure
The specimens were placed in the roof of a house near
Aveiro, 60 km away from Porto, near the ocean.
The samples faced Southwest with the notch free from
touching any surface, for the fracture properties that
were studied to express a real situation. The material
deterioration and structural performance were investi-
gated in a real situation of exposure for a year period.
The relationship between year period, exposure time and
load-bearing capacity of deteriorated Polymer concrete
is studied and fracture mechanics of the specimens are
discussed.
After the environmental exposure the samples were
tested in laboratory conditions to determine their
fracture properties. The tests setup apparatus is
displayed in Fig. 1. A decomposition of CMOD is done
due to nonlinear effect.
The critical opening displacement of the original
precrack tip, CTOD
C
, is calculated from the maximum
ARTICLE IN PRESS
J.M.L. Reis, A.J.M. Ferreira / Building and Environment 41 (2006) 262267 263
load registered and the value of the effective critical
crack length, which is the initial notch depth plus the
stable crack growth at peak load.
The fracture toughness K
Ic
is calculated using the
equation:
K
I
C
3P
max
0:5W
Spa
c

1=2
Fa
2d
2
b
; (1)
in which
Fa
1:99 a1 a2:15 3:93a 2:7a
2

p
1=2
1 2a1 a
3=2
q ; (2)
where a ac=d; P
max
is the measured maximum load
(N), W W
0
S=L (N) and W
0
is the self-weight of the
beam (N).
The fracture energy G
f
of the polymer concrete can be
calculated by the following equation [7]:
G
f

W
0
mgd
max
A
lig
; (3)
where W
0
is the area under the loaddeection curve
(Nm), mg is the self-weight of the specimen between
supports (kg), d
max
is the maximum displacement (m),
A
lig
is the fracture area [db2a] (m
2
), b and d are the
height and width of the beam, respectively and a is the
depth of the notch.
3. Results and discussion
The application of the two-parameter model and
fracture energy to specimens exposed to 1 year
outside and in laboratory conditions are shown in
Figs. 2 and 3 for epoxy polymer concrete, in Figs. 4 and
5 for glass ber-reinforced polymer concrete and in
Figs. 6 and 7 for carbon ber-reinforced polymer
concrete. From Load vs. CMOD TPFM properties
ARTICLE IN PRESS
Fig. 1. Laboratory test setup apparatus.
3
2.5
2
1.5
1
0.5
0
0 0.1 0.2 0.3
CMOD (mm)
0.4 0.5 0.6 0.7 0.8
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 2. Load vs. CMOD test results for epoxy polymer concrete.
3
2.5
2
1.5
1
0.5
0
0 0.2
Mid-span Displacement (mm)
0.4 0.6 0.8 1
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 3. Load vs. mid-span displacement test results for epoxy polymer
concrete.
3
2.5
2
1.5
1
0.5
0
0 0.5
CMOD (mm)
1 1.5 2
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 4. Load vs. CMOD test results for glass ber epoxy polymer
concrete.
J.M.L. Reis, A.J.M. Ferreira / Building and Environment 41 (2006) 262267 264
are obtained and the Load vs. Mid-span displacement
curves are the reference to calculate the fracture energy,
G
f
: As can be seen from such results, all samples were
degraded.
As can be analyzed from Tables 13 after 1 year of
weather exposure the fracture toughness of polymer
concrete are deteriorated when compared to specimens
in laboratory conditions and reference values. By 1 year
weather exposure, epoxy polymer concrete has
deterioration values for the stress intensity factor of
K
Ic
; is 2.134 MPa m
1/2
(average). A decrease of 9.8%
compared to 2.365 MPa m
1/2
, after 1 year in the
laboratory is observed. The laboratory specimens
show no deterioration as expected. The fracture energy
results after 1 year outside is practically the same as
reference but a decrease of 6.9% is observed when
compared to the specimens subjected to laboratory
conditions. These results show that the material displays
ARTICLE IN PRESS
3
2.5
2
1.5
1
0.5
0
0 0.5
Mid-span Displacement (mm)
1 1.5 2
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 5. Load vs. mid-span displacement test results for glass ber
epoxy polymer concrete.
3.5
3.5
2.5
2.5
2
2
3
3
1.5
1.5
1
1
0.5
0
0 0.5
CMOD (mm)
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 6. Load vs. CMOD test results for carbon ber epoxy polymer
concrete.
3.5
3.5
3
3 4
2.5
2.5
2
2
1.5
1.5
1
1
0.5
0.5
0
0
Mid-span Displacement (mm)
L
o
a
d

(
k
N
)
Weather Exposure
Laboratory Exposure
1 Year Exposure
Fig. 7. Load vs. mid-span displacement test results for carbon ber
epoxy polymer concrete.
Table 2
Season exposure results for glass ber-reinforced polymer concrete
FVE Specimens K
ic
(MPa m
1/2
) CTOD
C
(mm) G
f
(N/m)
Reference 2.412 0.013 11.575
2.379 0.025 10.992
2.576 0.016 11.132
Average 2.456 0.018 11.233
1 year weather
exposure
2.273 0.025 16.781
1.930 0.032 12.934
2.348 0.033 14.452
Average 2.183 0.030 14.722
1 year laboratory
exposure
2.252 0.011 13.885
2.449 0.019 19.185
2.290 0.016 18.525
Average 2.330 0.015 17.198
Table 1
Season exposure results for epoxy polymer concrete
Epoxy specimens K
ic
(Mpa m
1/2
) CTOD
C
(mm) G
f
(N/m)
Reference 2.323 0.008 8.429
2.344 0.023 7.012
2.267 0.015 8.183
Average 2.311 0.015 7.875
1 Year weather
exposure
2.005 0.019 8.107
2.260 0.009 7.655
2.138 0.013 7.901
Average 2.134 0.014 7.888
1 year laboratory
exposure
2.285 0.009 8.192
2.262 0.003 8.392
2.547 0.026 8.844
Average 2.365 0.013 8.477
J.M.L. Reis, A.J.M. Ferreira / Building and Environment 41 (2006) 262267 265
no increase in the fragility, showing that even after
weather exposure the material can retain its energy to
stop crack propagation.
When reinforced polymer concrete is analyzed, the
reinforcement has a signicant effect on the results. The
glass ber reinforcement value for stress intensity factor,
K
Ic
; is 2.183 MPa m
1/2
, on average. After 1 year weather
exposure a decrease of 5.5% is seen compared to the
average reference result. The fracture energy, G
f
, has an
increase of 31.1% after weather exposure compared to
reference. The specimens exposed to controlled labora-
tory conditions has a higher increase compared to
reference specimens, 53.1%.
When carbon ber reinforcement is observed, the gap
in the results are higher than in the other or with no
reinforcement. The stress intensity factor, K
Ic
; decreased
by 27.9% after weather exposure when compared to the
same polymer concrete composition in the laboratory
condition for the same period of time. The Fracture
energy, G
f
; after the weather exposure period, has a
result close to specimens allowed in the laboratory, less
than 3%. In all formulations the fracture energy
increases its values, showing that after a period of time
the material becomes tough and less fragile to crack
propagation.
During the outdoor exposure, visual inspection was
performed. After exposure, the specimens were mea-
sured and no change, cracks shrinkage or expansion was
observed in the tested polymer concrete.
Specimens containing carbon ber reinforcement
presented a higher level of visual degradation since
these bers are UV radiations sensitive. Its color
changed from black to light brown after the tests
showing that this is one of the reasons why the carbon
ber polymer concrete has a lower comparative result
than laboratory specimens.
4. Conclusions
This was a good approach to verify the reliability of
the material when exposed to real conditions experi-
ment. The place where the specimens were allocated
proves to be a good region to make this experiment since
almost all the aggressors in real conditions are present,
sun, wind, dust, and marine close environment. From
the results analyzed after 1 year exposure polymer
concrete deteriorates, but not as much as expected,
knowing that hotter the environment involving polymer
concrete, higher deterioration that occurs due to the
intensity of the sun and UV radiation. The glass ber
reinforcement proved to be a better combination to
expose the specimens under such conditions, showing
almost no degradation, except on the surface erosion
observed. Neverless, the results conrm the fact that
moderate losses of toughness are possible in polymer
concrete.
The carbon ber reinforcement proved to be an
inappropriate reinforcement to use due to the loss of
properties of carbon when exposed to weather condi-
tions such as direct sunlight exposure, UV radiation.
Carbon ber these days are often used as reinforcement
in concrete structures, like steel or those involving the
concrete piles. These reinforcements are performed for
inside structures with a high rate of success but when
carbon bers are subjected directly, specially to UV
radiations, the performance decreases drastically.
Acknowledgements
The support of Fundac-a o para a Cie ncia e a
Tecnologia, under the project POCTI/EME/35955/
1999, Long-term Properties of Polymer Concrete, is
gratefully acknowledged.
References
[1] Davis A, Sims D. Weathering of Polymers. Essex, UK: Applied
Science Publishers; 1983.
[2] Ribeiro MCS, Tavares CML, Figueiredo M, Fernandes AA,
Ferreira AJM. Bending characteristics of resin concrete. Materials
Research-Revista Ibero Americana de Materiais 2003; 6(2).
[3] Ferreira AJM, Tavares CML, Ribeiro MCS. Flexural properties
of polyester resin concretes. Journal of Polymer Engineering
2000;20(6):45968.
[4] El-Hawary M, Al-Khaiat H, Fereig S. Effect of seawater on
epoxy-repaired concrete. Cement and Concrete Composites
1998;20:4152.
[5] RILEM, TC 50-FM. Determination of the fracture parameters
(K
SIC
and CTOD
C
) of plain concrete using three-point bend tests
on notched beams. Materials and Structures 23(138): 45760.
[6] Shah SP, Carpinteri A. Fracture Mechanics Test Methods for
Concrete. London: Chapman and Hall; 1991. pp. 310.
[7] RILEM. TC 50-FMC, Fracture mechanics of concrete, determi-
nation of fracture energy of mortar and concrete by means of
ARTICLE IN PRESS
Table 3
Season exposure results for carbon ber-reinforced polymer concrete
FCE specimens K
ic
(Mpa m
1/2
) CTOD
C
(mm) G
f
(N/m)
Reference 2.711 0.038 28.823
2.586 0.059 28.857
2.925 0.032 29.238
Average 2.741 0.043 28.973
1 year weather
exposure
1.443 0.059 21.687
2.614 0.019 36.887
2.192 0.034 34.551
Average 2.083 0.037 31.042
1 year laboratory
exposure
3.121 0.073 31.961
2.932 0.007 29.574
2.611 0.019 33.823
Average 2.888 0.033 31.786
J.M.L. Reis, A.J.M. Ferreira / Building and Environment 41 (2006) 262267 266
three-point bend test on notched beams, RILEM recommenda-
tion. Materials and Structures 1995;18:40713.
[8] Reis JML. Fracture mechanics of polymer concrete, PhD thesis,
FEUP, December 2003.
[9] Reis JML, Ferreira AJM. The Inuence of Notch Depth on the
Fracture Mechanics Properties of Polymer Concrete. Interna-
tional Journal of Fracture 2003;124(12):3342.
[10] Reis JML, Ferreira AJM. Fracture behaviour of glass bre-
reinforced polymer concrete. Polymer Testing 2003;V.22(2):
14953.
[11] RILEM TC/113 PC-2, Technical Committee 113. Method of
making polymer concrete and mortar specimens. Symposium on
properties and test methods for concrete-polymer composites,
Oostende, Belgium, 1995.
ARTICLE IN PRESS
J.M.L. Reis, A.J.M. Ferreira / Building and Environment 41 (2006) 262267 267
Steel encased polymer concrete under axial compressive loading:
Analytical formulations
Walter O. Oyawa
*
Dept. of Civil Engineering, JKUAT, P.O. Box 62000-00200, Nairobi, Kenya
Received 17 June 2004; received in revised form 25 July 2005; accepted 31 July 2005
Abstract
Composite structures and in particular steel encased concrete structures, are rapidly emerging as one of the inevitable structural sys-
tems for earthquake resistance, as they have been known to exploit the best attributes of steel and concrete, resulting in higher stiness,
strength and ductility. However, limitations imposed by the encased cement concrete, e.g., its brittleness, low tensile strength and dimin-
ished durability, further compounded by mounting environmental concerns since the 1997 Kyoto protocol and the need for sustainable
development, dictate that novel ecologically benign construction materials be sought for the 21st century. In this context, experimental
and analytical studies have been conducted on polymer-based materials (latex cement mortar and epoxy concrete) as supplementary and/
or complementary materials to ordinary cement concrete in composite tubular systems. Results from experimental work on the compres-
sive behaviour of steel encased polymeric materials reveal signicant increase in strength and/or ductility of polymer and polymer con-
crete lled steel stub columns. In addition, polymer formulations were found to present an array of basic properties, which could be
tailored to meet specic design requirements. In conjunction with the experimental results, existing models by several researchers are
evaluated and found to generally underestimate the conned strength of the ll materials. A proposed model, taking into consideration
key material properties such as shear and bulk moduli, is found to depict the conned strengths satisfactorily. Finally, forcestrain mod-
els are developed for the composite stub columns that adequately capture the elastic stiness and ultimate state, unlike existing models.
2005 Elsevier Ltd. All rights reserved.
Keywords: Composite columns; Polymer concrete; Compressive strength analysis
1. Introduction
Recent catastrophic eects of major earthquakes on
infrastructure, coupled with fast changing global needs re-
lated to the preservation of the environment and sustain-
able environment, have instigated renewed interest in
composite structures, particularly steel encased concrete
structures [18]. It is no wonder that concrete lled tubular
(CFT) system has been adopted in several landmark struc-
tures, e.g., 63-storey commerzbank building in Germany,
which is the tallest building in Europe at 300 m, the 57-storey
Two Union Square in Seattle, USA, and most recently a
sixty metre span space-truss pedestrian bridge in Quebec,
Canada [6]. In addition to providing a better reproong
and soundproong property than steel structures, and are
even easier, safer, and faster to construct than reinforced
concrete structures, concrete-lled structures exploit the
best attributes of both steel and concrete, thus allowing
the engineer to maintain manageable member sizes while
obtaining increased stiness, strength, energy absorption,
and ductility [4,8,21]. The composite action is such that
the steel tube, in addition to acting as a reinforcement,
connes the concrete resulting in signicant increase in
concrete compressive strength, while the conned concrete
not only relieves the steel tube of some load but also delays
and moderates buckling deformations in the steel tube.
Although the use of CFT continues to be popular in
seismic areas, their widespread use seems limited due to
0950-0618/$ - see front matter 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.conbuildmat.2005.07.013
*
Tel.: +254 734 691617; fax: +254 067 52164.
E-mail addresses: oyawaw@yahoo.com, oyawa@jkuat.ac.ke.
www.elsevier.com/locate/conbuildmat
Construction and Building Materials xxx (2005) xxxxxx
Construction
and Building
MATERIALS
ARTICLE IN PRESS
lack of information about their true strength and inelastic
behaviour [5] and the limitations of the inll cement con-
crete. Ordinary cement concrete, although preferred due
to its lower cost and relatively good strength in compres-
sion, has marked drawbacks in that it is unfriendly to the
environment, brittle, reactive, has very low tensile strength,
shrinks with time and because of its porous nature allows
for the ingress of deleterious substances that signicantly
compromise the durability of the structure so constructed.
Furthermore, the extremely disastrous eects of recent ma-
jor earthquakes on both steel and reinforced concrete
structures has prompted a change of seismic design per-
spective from the previous emphasis on structural strength
to structural ductility and energy absorption. In this con-
text, interest is now being directed towards evaluating the
suitability of polymers or polymer based materials, for
use as ll materials in lled steel composite members.
Polymers or polymer based materials are attracting in-
creased attention in the construction industry due to the
supplementary properties to concrete that they posses
viz higher tensile and adhesion capacity, lower weight
and shrinkage, high ductility and resilience, and resistance
to physical and chemical attack that ensures their longev-
ity or durability [11,13,22,24]. They were rst introduced
to hydraulic-cement system in 1923 because of an in-
creased need at that time for durable construction materi-
als. Although the cost of polymers may be comparatively
high at present, a situation attributed to their use in other
highly lucrative industries such as the defence and aero-
nautical industries, it is envisaged that continued promo-
tion of their unconventional advantages and research
directed towards seeking optimally cheap derivatives,
e.g., recycled plastic wastes [17] for use specically in
the construction industry will eventually lead to satisfac-
tory cost reduction. Moreover, proposals being made
for the evaluation of full life-cycle cost of structures
during design to include such parameters as maintenance,
repair, demolition and environmental degradation, rewar-
dingly place the durability of polymer based materials at
the forefront of cost analysis.
Previous experimental and nite element studies con-
ducted concentrated on the compressive behaviour of epoxy
polymer-lled and latex-cement mortar-lled steel compo-
site stub columns [14,15,23]. Results obtained indicate the
great potential of epoxy concrete and ber latexcement
mortar as alternative ll materials to cement concrete or
which could be combined with ordinary concrete for
enhanced strength and/or ductility of lled steel composite
members.
This present study extends on the observations drawn
from the aforementioned studies, and focuses on develop-
ing analytical models for conned strength of the ll mate-
rials, as well as the loadstrain relationship of the
composite stub column. In addition to providing improved
design quantities to augment existing codes that have large
discrepancies [2,9,10,20,21], the models may be of use in
the furtherance of studies pertaining to the use polymeric
materials in composite structures, e.g., nite element anal-
ysis and ber analysis.
2. Overview of previous work on composite stub columns
Experimental work on lled steel composite stub col-
umns comprised testing several types of circular steel com-
posite columns lled with various types of ll materials,
and gradually subjected to axial compressive load until fail-
ure, while recording load, biaxial strain in the steel compo-
nent and axial displacement measurements at suitable
increments (Figs. 1 and 2). The variables were the type of
ll material (some of which are shown in Table 1) and
the radius to thickness ratio of steel tube (D/2t). Fill mate-
rials investigated were special formulations by one of the
collaborating companies, and included bre latex cement
mortar (LCM), polymer concrete type 1 of high strength
and stiness (E1A), polymer concrete type 2 of low
strength and stiness (E2A) and normal concrete (CN). Fi-
bre latex cement mortar (LCM) was prepared from a mix-
ture of ne aggregate, Portland cement, carbon bres,
Styrene Butadiene Rubber latex and water in required ra-
tios. The polymer concretes consisted of a formulation of
Fig. 1. Filled steel stub column specimen.
Fig. 2. Stub column test set-up.
2 W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx
ARTICLE IN PRESS
epoxy polymer, coarse aggregate, water and admixture; the
dierence between E1A and E2A was in the type of epoxy
used, i.e., high strength epoxy and low strength epoxy,
respectively. It may be noted that the physical properties
of epoxy resin depend upon many factors including type
and amount of curing agent, cure history, type of ller
and other modiers.
The specimens were designated according to their com-
ponents and D/2t ratio, where S/LCM-30 is latex cement
mortar lled steel of D/2t = 30, S/E1A is epoxy concrete
type 1 lled steel, S/E2A is epoxy concrete type 2 lled steel
and S/CN is normal concrete lled steel column. Typical
response obtained is shown in Fig. 3 for a radius to thick-
ness ratio (D/2t) = 50, where the versatile response of poly-
mer-based stub columns is observed in comparison to the
empty steel tube (S) and the cement concrete lled steel
tube (S/CN). Most importantly, it was determined that
polymer-based materials present a wide array of properties,
unlike conventional concrete, which can be tailored to meet
specic design and construction needs, e.g., high ductility
low strength polymers (in the retrotting of columns and
for ductility-based design), high strength polymers (supple-
mentary to ordinary concrete), and low-weight low-noise
polymers (for elevated railway networks in residential
areas).
To further comprehend the behaviour of conned ll
materials, nite element modelling of the composite stub
columns was carried out using ABAQUS nite element
package in conjunction with experimental results. Paramet-
ric studies conducted employing the model demonstrate the
enormous eect of ll materials on the response of stub col-
umns under compression load. Especially, ll material
properties such as Poissons ratio (m), modulus of elasticity
(E), bulk modulus K = E/3(2 m), shear modulus G = E/
2(1 + m), and strength are demonstrated to have a marked
eect on stub column response. Accordingly, this paper
seeks to quantify these eects in formulating simple analyt-
ical tools that may be used in the design of composite stub
columns.
3. Evaluation of conned ll material strength models
3.1. Mechanics of connement
Ordinary concrete surrounded by an enclosing restrain-
ing medium and subjected to axial load will develop passive
lateral pressure as it expands under the inuence of axial
compression, creating a multiaxial state of stress. The con-
nement commences when the Poisson ratio of the concrete
ll exceeds that of the enclosing steel tube, and has the
eect of signicantly changing the structural response of
the concrete inll and the encasing steel tube.
Gardner and Jacobson [7] proposed that as the steel
tube restrains the dilating concrete core, and internal pres-
sure develops between the steel tube and concrete, creating
hoop stresses as illustrated in Fig. 4. Connement action is
thus dependent on the tendency of the concrete inll to
y
/
0 20 40 60 80 100
P
/
P
y
0
1
2
3
4
5
6
7 S/LCM
S/E1A
S/E2A
S/CN
S
S/LCM
S/E1A
S/E2A
S/CN
S
Fig. 3. Typical normalized load vs. strain (D/2t = 50).
Fig. 4. Interaction of concrete and steel in CFT column under compres-
sion (a) composite stub column (b) concrete stress (c) steel stress.
Table 1
Properties of some of the ll materials
Fill material
description
Fill material
label
Youngs modulus
E
ll
(KN/mm
2
)
Poissons
ratio t
f
Ultimate strength
(N/mm
2
)
Shortening strain at
peak strength (%)
Bulk-K and
shear-G moduli
(KN/mm
2
)
K G
Latex mortar LCM 15.2 0.222 21.1 0.7272 9.11 6.22
Polymer concrete-1 E1A 12.9 0.316 52.0 0.7208 11.7 4.90
Polymer concrete-2 E2A 3.0 0.480 19.0 4.845 25.0 1.01
Normal concrete CN 29.6 0.171 26.0 0.4035 15.0 12.6
W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx 3
ARTICLE IN PRESS
dilate when loaded, as well as the radial stiness of the con-
ning member to restrain the dilation. When this interac-
tion is active, it is rational to assume that there will be
geometric (strain) compatibility between the core and the
shell, and also the equilibrium of forces in the free-body
diagram for any sector of the conned section.
Equilibrium of forces acting on the steel tube is obtained
as summation of forces on the small element subtended by
angle oh from the centre, i.e.;

F
y
r
sh
2t
_
p
0
r
l
r
i
sin hdh; 1
()r
sh

r
i
t
r
l
; 2
()r
l

t
r
i
r
sh

2t
D
i
r
sh
. 3
The axial load in the composite stub column at any strain
may be considered to be the summation of forces in the
components (Eq. (4))
P P
c
P
s
. 4
3.2. Conned strength of ll materials
The strength of inll material with non-linear and non-
homogeneous characteristics under a multi-axial state of
stress may be dicult to establish theoretically [16], neces-
sitating the use of test data to develop empirical or semi
empirical approaches. Early investigators showed that the
strength (r
cc
) and the corresponding longitudinal strain
(e
cc
) at the strength of concrete conned by an active
hydrostatic uid pressure can be represented by the follow-
ing simple relationships:
r
cc
r
co
k
1
r
1
; 5
e
cc
e
co
1 k
2
r
1
=r
co
; 6
where r
co
and e
co
are the unconned ll material strength
and strain at ultimate strength, respectively, while k
1
and
k
2
are constants. Eq. (5) may be approximately formulated
by considering a cylindrical medium under a multi-axial
state of stress (Fig. 4(b)). The lateral strain (e
l
) at conned
strength may be expressed as below
e
l

r
l
E
m
r
cc
E
m
r
l
E
()
r
1
E
1 m m
r
cc
E
. 6a
The lateral strain may also be expressed in terms of uncon-
ned stress, i.e., (r
1
= 0) as
e
l
m
r
co
E
. 6b
Equating the above Eqs. (6a) and (6b) gives
r
1
E
1 m m
r
cc
E
m
r
co
E
()r
cc
r
co
r
l
1 m
m
_ _
.
6c
However, the modulus of elasticity (E) and Poissons ratio
(m) are dierent in the three directions and also vary with
loading due to material non-linearity. Hence, a general
expression as in Eq. (5) is adopted with a constant k
1
,
which is specied from experimental data. A similar expres-
sion is adopted for strains as in Eq. (6), with a constant k
2
.
Richart et al. [19], one of the earliest researchers in this
eld, found k
1
= 4.1 and k
2
= 5k
1
. Since then a number
investigators have basically modied this simple linear
model, to suit the various conditions of their tests. These
include Mander et al. [12], Saatci oglu and Razvi [19], Cus-
son and Paultre [3] and the Architectural Institute of Japan
(AIJ) [1], as is shown in Table 2. Expressions by Mander,
Saatcioglu et al. and Cusson mainly targeted connement
in reinforced concrete columns, while recent work by AIJ
deals with composite concrete lled tubular members.
In this study, the existing models and a proposed model
given in Table 2 are evaluated for suitability vis-a`-vis
experimental results. As already mentioned, previous
experimental work on composite stub columns under axial
load have been conducted, whereupon applied load, axial
shortening deformations as well as longitudinal and hoop
strains on the encasing steel were measured. The measured
strains enabled the estimation of hoop and longitudinal
stresses on the encasing steel at any applied load, by
employing Hookes law in the elastic range and Prandtl-
Reuss equations associated with perfectly plastic von
Mises yield function in the plastic range [14]. Accordingly,
experimental stress on conned ll-material (r
c
) at any
applied load (P) is deduced from the experimental data
by utilizing
r
cc

P r
s
A
s
A
f
; 7
where P is the applied axial compressive load on stub col-
umn, r
s
is the axial stress in encasing steel component, A
s
is
the sectional area of steel, A
f
is the sectional area of ll
material.
The maximum value of r
c
corresponding to the ultimate
load of composite stub is taken as the experimental con-
ned ll strength r
cc
.
A look at the Figs. 57 clearly illustrates that for con-
ned latex cement mortar (LCM) and normal concrete
(CN), existing models signicantly underestimate the con-
ned strength, whereas for conned polymer concrete
(E1A), the models are reasonable. This may be attributed
to the dierent response mechanisms of the conned ll
materials. The post-elastic loadstrain relationship of com-
posite S/LCM and S/CN stub columns show a gradual in-
crease in force up to the ultimate followed by a gentle
decrease, an indication that the ll materials increases in
strength even as the steel is failing. On the other hand,
the post-elastic stressstrain relationship of composite
S/E1A stub columns show a gradual increase in force up
to the ultimate followed by a sudden decrease, suggesting
a near-simultaneously failure of steel and ll materials. It
would seem that the existing models do not capture the
gradual strength increment in the latex cement mortar
4 W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx
ARTICLE IN PRESS
and normal concrete after the yielding of encasing steel, but
instead only depict the conned strength of ll materials
after the commencement of yielding of the encasing steel.
That makes the models more suitable for conned E1A,
where a near-simultaneous failure between the encasing
steel tube and the ll material is thought to occur.
Thus, a proposed model is specied that modies the
expression of lateral stress on ll material by taking into ac-
count additional unique factors such as the shear modulus
G = E/2(1 + m
f
), bulk modulus K = E/3(1 2m
f
), strain at
ultimate (e
cu
), Poissons ratio (m
f
) of the ll materials (Eq.
(15)). A quantity e
fb
is introduced which may be regarded
as ll material axial strain at yielding of steel (e
fb
has a
value of 1.3 for LCM, 0.35 for E1A and 1.40 for CN).
r
l

e
fb
m
f
K
e
co
G
2tr
sy
D 2t
. 15
Table 2
Existing and proposed conned strength and strain models
Researcher Conned strength Strain at conned strength
Richart et al. [19] r
cc
= r
co
+ 4.1r
1
e
cc
= e
co
(1 + 20.5r
1
/r
co
)
Mander et al. [12] r
cc
r
co
1:254 2:254

1 7:94
rl
rco
_
2
rl
rco
_ _
e
cc
e
co
1 5
rcc
rco
1
_ _ _ _
and r
l

2rcomt
D2t
Saatcioglu, et. al. [19] r
cc
= r
co
+ 6.7(r
l
)
0.17
r
l
e
cc
r
co
1
5rl
rco
_ _
Cusson [3] r
cc
r
co
2:1r
co
rl
rco
_ _
0:7
e
cc
e
co
0:21
rl
rco
_ _
1:7
AIJ [1] r
cc
r
co

4:10:192trsy
D2t
ecc
eco

1 4:7
rcc
rco
1
_ _
; when
rcc
rco
6 1:5
3:35 20
rcc
rco
1:5
_ _
; when
rcc
rco
> 1:5
_
_
_
Proposed Oyawa [15] r
cc
r
co
4:1r
1
where r
l

efbmf K
ecoG
2trsy
D2t
e
cc

e
co
1
20rl
rco
_ _
for LCM2 and CN
e
co
1
40rl
rco
_ _
for E1A
_
_
_
Nomenclature
r
cc
conned strength of ll material
r
1
lateral conning pressure on ll material
e
cc
strain at conned strength of ll material
e
fb
may be regarded as strain at yield of biaxial steel
K = E/3(1 2m
f
) bulk modulus of ll material
r
co
unconned strength of ll material
r
sy
steel yield strain
e
co
strain at unconned strength of ll material
m
f
Poissons ratio of ll material
G = E/2(1 + m
f
) shear modulus of ll material
Fig. 6. Conned strength of epoxy concrete (E1A).
Fig. 5. Conned strength of latex cement mortar (LCM).
W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx 5
ARTICLE IN PRESS
The conned strength is then obtained as
r
cc
r
co
4:1r
l
. 16
The inclusion of these unique parameters in the model is
inspired by previous experimental and nite element ana-
lytical work that indicated the signicant eect of these
properties on the response of composite stub columns,
especially the bulk modulus (regarded as the degree of
compressibility), the shear modulus (regarded as a mea-
sure of ow capability) and by extension Poissons ratio
and modulus of elasticity. In the previous studies, it was
noted that ll materials with high bulk modulus and
low shear modulus have characteristics tending towards
that of liquids, i.e., high degree of incompressibility and
high degree of outward ow. It is observed that the pro-
posed model closely simulates experimental results for all
the cases, i.e., LCM, E1A and CN. Thus, the proposed
conned strength model is viewed as a signicant leap to-
wards a comprehensive understanding of the behaviour of
unique polymer-based materials in composite construc-
tion.
4. Prediction of loadstrain relationships for stub columns
4.1. Concepts in model development
Usual loadstrain model adopted for composite stub
columns entails the summation of forces in the conned ll
material and the encasing steel tube, while utilizing avail-
able stressstrain relationships for the constituent materi-
als. Thus, at any given compressive strain, normalized
force is determined as given in Eq. (17).
P
P
y

r
c
A
c
r
s
A
s
P
y
where r
s

E
s
e
s
for e
s
6 e
sy
r
sy
for e
s
> e
sy
_
17
and
r
c

r
cc
e
ecc
_ _
r
r 1
e
ecc
_ _
r
; r
E
c
E
c

e
ecc
the values ofr
cc
; r
l
; e
cc
are as defined by AIJ in Table 2.
This existing model allows for distribution of applied
load between the conned ll material and the encasing
steel tube depending on their cross-sectional areas and
stiness. However, it is argued herein that this may not
be a rational assumption in view of likely stress concen-
trations and surface imperfections at the commencement
of loading. Moreover, the localized behaviour and buck-
ling of thin steel sections will almost certainly result in
load re-distribution.
Instead, it is presented that the degree of load sharing
between the encasing steel and ll material will mostly de-
pend on the level of connement of the ll material. In
other words, if the ll material is heavily conned, minimiz-
ing dilation and vertical deformation, then the conned ll
material will carry most of the applied load in the elastic
range. On the other hand if the ll material is less conned,
there will be signicant vertical deformation of the ll
material resulting in the re-distribution of some of the ap-
plied to the encasing steel tube.
To test the above hypothesis, an alternative model is
evolved and presented herein. The model is evolved
assuming that the composite stub column response is lar-
gely related to the conned ll material stressstrain res-
ponse, since the thin-walled encasing steel tube largely
acts as a conning medium by redistributing most of
its axial load to the ll material. That is to say that by
suitably factoring and modifying conned stressstrain
relationship of the ll material, the normalized load
deformation response of the corresponding lled steel
composite column may be obtained. Thus, a proposed
normalized loaddeformation model is derived as a modi-
cation of the conned concrete stressstrain relationship
developed by Mander et al. [12]. The modication in-
volves factoring the popular Mander equation by a quan-
tity dependent on several factors such as the radius/
thickness ratio and Poissons ratio of the steel tube.
Accordingly, the Mander conned stressstrain relation-
ship of the ll material is modied in the term r
m
as well
as the inclusion of an additional factor b which generally
denes the level of connement by including a term D/
(2t 2). For thin steel section, the term D/(2t 2) is
approximately equal to the ratio of cross-sectional areas
of the ll material and steel (A
f
/A
s
). Hence, when the
area of the steel tube A
s
reduces, there will be less con-
nement of ll material, permitting axial deformation
of the ll material and hence re-distribution of some of
the applied load to the steel tube. The vice versa is true.
On this premise or argument, the proposed model has
been developed.
Fig. 7. Conned strength of normal concrete (CN).
6 W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx
ARTICLE IN PRESS
P
P
y

br
cc
e
ecc
_ _
r
m
r
m
1
e
ecc
_ _
c
m
_ _
r
co
; the values of
r
cc
; r
l
; e
cc
are as proposed in Table 2 18
where b
r
co
D
2t
_ _
2r
sy

1 m
2
s
_ r
m

E
c
E
c
S
c
e
e
cc
and
S
c
log
10
D
2t
_ _
4.2. Verication of the proposed model vis-a` -vis existing
model
To assess the applicability of the proposed model (Eq.
(18)), analytical results are compared with experimental re-
sults, as well as with existing model that considers simple
summation of forces, while utilizing available stressstrain
relationships for the constituent materials (Eq. (17)). Figs.
813 show the analytical and experimental normalized
force vs. strain relationships. The immediate observation
is that the proposed closely simulates the experimental re-
sults in both the elastic region up to the ultimate load
point. On the other hand, the existing model is not able
to simulate the elastic region, as it gives a much higher elas-
tic stiness than the experimental results. The overestima-
tion of stiness by the existing model is due to the
assumption that the steel tube plays a signicant role in
load resistance, largely depended on the steel axial stiness.
It is hereby argued that this might not be the case due to
surface irregularities and stress concentrations on the thin
sy
/
0 10 20 30 40 50 60 70 80
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
Fig. 8. Normalized load vs. strain (S/LCM-30).
sy
/
0 10 20 30 40 50 60 70 80
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
Fig. 9. Normalized load vs. strain (S/LCM-40).
0 10 20 30
/
40 50 60 70 80
sy
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
Fig. 10. Normalized load vs. strain (S/E1A-30).
0 10 20 30 40 50 60 70 80
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
/
sy
Fig. 11. Normalized load vs. strain (S/E1A-40).
0 10 20 30 40 50 60 70 80
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
/
sy
Fig. 12. Normalized load vs. strain (S/CN-30).
W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx 7
ARTICLE IN PRESS
steel tube as loading commences, resulting in stress redistri-
bution to the ll material. In other words, perfect compos-
ite interaction may not be possible for thin steel-tubes, thus
causing the conned ll material to take up most of the ap-
plied load, with consequent reduced elastic stiness as sim-
ulated by the proposed model.
Considering S/LCM2 and S/CN composite stub col-
umns, it is noted that, unlike the proposed model, the exist-
ing model underestimates the strength of the composite
stub column, possibly due to assumed elasto-perfectly plas-
tic behaviour of the steel tube. This is safe as far as design
requirements are concerned, but perhaps conservative and
hence uneconomical.
Simulation of polymer concrete (S/E1A) lled steel col-
umns shows variations for both the proposed method and
existing method. Whereas the proposed method, unlike the
existing method, captures the elastic stiness well, it has
diculties capturing the ultimate strength of the stub col-
umns; sometimes underestimating, sometimes overestimat-
ing and sometimes giving close values. This may be
attributed to the variable failure characteristics of polymer
concrete. The polymer concrete used in this study had brit-
tle failure characteristics, which seemed to be initiated by
random bursting cracks within the polymer concrete. As
these cracks formed progressively within the matrix, there
seemed to be unpredictable drops and rises in strength of
the stub column as shown by the experimental results.
Nonetheless, the proposed method attempts to present ra-
tional average graphs. It also to be noted that for S/E1A,
the ultimate strengths depicted by the proposed method
and existing method are very similar.
5. Conclusions
This analytical study has extended on experimental
studies and evolved suitable models for the determination
of conned strength of conned polymeric materials, as
well as forcestrain relationship of the composite stub
columns made out of the polymeric materials. Is noted
that:
1. Proposed conned strength model for ll materials sim-
ulates the test results much better than several existing
models for the cases of conned latex cement mortar
and ordinary concrete. For the polymer concrete, both
the proposed model and existing models give rational
results.
2. Proposed forcestrain model best captures or simulates
the test results for composite stub columns, as it depicts
both the elastic stiness and ultimate capacity of the
composite stub columns better than existing method.
3. More work is still required on polymer concrete in
view of the fact that there are numerous polymeric
products or formulations in the market that have quite
dierent characteristics, and hence need to be assessed
separately.
Acknowledgements
Present work is under the support of Japan Society for
the Promotion of Science (JSPS), to whom the authors
are most grateful. The authors remain thankful to the Min-
istry of Education, Science, Sports and Culture for their
previous support.
References
[1] AIJ, Recommendations for design and construction of concrete lled
steel tubular structures, 1997.
[2] Brauns J. Analysis of stress state in concrete-lled steel column. J
Constr Steel Res 1999;49:18996.
[3] Cusson D, Paultre P. Stressstrain model for conned high-strength
concrete. J Str Eng ASCE 1995;121(3):46877.
[4] Elnashai AS, Broderick BM, Dowling PJ. Earthquake-resistant
composite steel/concrete structures. Struct Eng 1995;73(8):12132.
[5] Elremaily A, Azizinamini A. Behaviour and strength of circular
concrete lled tube columns. J Constr Steel Res 2002;58:156791.
[6] Fam A, Qie FS, Rizkalla S. Concrete-lled steel tubes subjected to
axial compression and lateral cyclic loads. J Struct Eng ASCE
2004;130(4):63140.
[7] Gardner NJ, Jacobson ER. Structural behaviour of concrete-lled
steel tubes. ACI J 1967;65:40413.
[8] Ge H, Usami T. Cyclic tests of concrete-lled steel box columns. J Str
Eng ASCE 1996;120(10):116977.
[9] Giakoumelis G, Lam D. Axial capacity of circular concrete-lled tube
columns. J Constr Steel Res 2004;60:104968.
[10] Han L. Flexural behaviour of concrete-lled steel tubes. J Constr
Steel Res 2004;60:31337.
[11] Kardon J. Polymer-modied concrete: review. J Mat Civil Eng ASCE
1997;9(2):8592.
[12] Mander JB, Priestley MJN, Park R. Theoretical stressstrain model
for conned concrete. J Str Eng ASCE 1988;114(8):18041826.
[13] Ohama Y. Principle of latex modication and some typical properties
of latex-modied mortars and concretes. ACI Mat J
1987;84(6):5118.
[14] Oyawa WO, Sugiura K, Watanabe E. Elasto-plastic behaviour of
axially loaded lled circular steel stub columns. J Struct Eng JSCE
1998;44A:14758.
[15] Oyawa WO, Sugiura K, Watanabe E. Polymer-concrete-lled steel
tubes under axial compression. J Constr Build Mat 2001;15:18797.
[16] Razvi1 S, Saatcioglu M. Connement model for high-strength
concrete. J Struct Eng ASCE 1999;125(3):2819.
0 10 20 30 40 50 60 70 80
P
/
P
y
0
1
2
3
4
5
6
7
Experiment
Proposed
AIJ
/
sy
Fig. 13. Normalized load vs. strain (S/CN-40).
8 W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx
ARTICLE IN PRESS
[17] Rebeiz KS, Fowler DW. Flexural strength of reinforced polymer
concrete made with recycled plastic waste. ACI Struct J 1996;93(5):
52430.
[18] Roeder CW. Overview of hybrid and composite systems for seismic
design in the United States. Eng Struct 1998;20(46):35563.
[19] Saatci oglu M, Razvi SR. Strength and ductility of conned concrete.
J Struct Eng ASCE 1992;118(6):1590607.
[20] SawHS, RichardLiewJY. Assessment of current methods for the design
of composite columns in buildings. J Constr Steel Res 2000;53:12147.
[21] Shams M, Ala SM. State-of-the-art of concrete-lled steel tubular
columns. ACI Struct J 1997;94(5):55871.
[22] Soroushian P, Aouadi F, Nagi M. Latex-modied carbon bre
reinforced mortar. ACI Mat J 1991;88(1):118.
[23] Sugiura K, Watanabe E, Oyawa WO. Assessment on non-concrete
lled steel bridge piers subjected to cyclic loading. J Struct Eng JSCE
2003;44A:393402.
[24] Zayat K, Bayasi Z. Eect of latex on the mechanical properties of
carbon reinforced cement. ACI Mat J 1996;93(2): 17881.
W.O. Oyawa / Construction and Building Materials xxx (2005) xxxxxx 9
ARTICLE IN PRESS
Fire Safety Journal 40 (2005) 669686
Stressstrain constitutive equations of concrete
material at elevated temperatures
Long-yuan Li

, John Purkiss
School of Engineering and Applied Science, Aston University, Birmingham B4 7ET, UK
Received 18 February 2005; received in revised form 20 May 2005; accepted 2 June 2005
Available online 27 July 2005
Abstract
The paper presents a critical review of the currently available models for the mechanical
behaviour of concrete at elevated temperatures. Based on these models and experimental data
a stressstraintemperature model is proposed which incorporates the effect of transient strain
implicitly. This model can be easily incorporated into existing commercial nite element
analysis software. A numerical example on a wall element heated on two opposite faces
indicates that at very early stages of heating transient strain does not play an important part,
but that as the exposure time increases the effect of ignoring transient strain progressively
increases and produces unconservative estimates of load carrying capacity.
r 2005 Elsevier Ltd. All rights reserved.
Keywords: Concrete; Stressstrain relation; Fire; Transient strain; Fire performance
1. Introduction
Over a long period the concept of critical temperatures was evolved for the design
of structural members in re. This was in part due to the imposition of loading
designed to generate the same stresses as under the loading applied to the structure at
working loads (or serviceability limit state). It should be observed that the critical
temperature approach is simplistic and can only be used where a single material
ARTICLE IN PRESS
www.elsevier.com/locate/resaf
0379-7112/$ - see front matter r 2005 Elsevier Ltd. All rights reserved.
doi:10.1016/j.resaf.2005.06.003

Corresponding author.
E-mail addresses: l.y.li@aston.ac.uk (L.-y. Li), j.a.purkiss@aston.ac.uk (J. Purkiss).
dominates the re response of the element. Thus, it may hold for steelwork or for
reinforced and prestressed concrete in exure where performance is dominated by
the behaviour of the reinforcing steel. It is not strictly applicable to reinforced
ARTICLE IN PRESS
Nomenclature
C
1
, C
2
, C
3
constants used in Schneiders model
E initial tangent modulus of se curve at T
E
0
initial tangent modulus of se curve at T
0
E

tangent modulus of se curve in descending branch at T


E

0
tangent modulus of full se curve in descending branch at T
0
E* initial tangent modulus of full se curve at T
E

av
secant modulus of full se curve at T
E

p
tangent modulus of full se curve in descending branch at T
f(T) empirical function used in Diederichs model
g function used in Schneiders model
k
p
constant used in Eq. (17)
k
tr
constant used in Anderberg and Thelanderssons model
n constant used in Schneiders model
T temperature
T
0
initial temperature
T
g
constant used in Schneiders model
t time
V
a
volume fraction of aggregate
w moisture content
a thermal expansion coefcient
b function used in Schneiders model
g
0
constant used in Schneiders model
g
w
parameter used in Schneiders model
j parameter used in Schneiders model
F function used in Schneiders model
e total strain
e
1
strain at a joint point used in Anderberg and Thelanderssons model
e
cr
creep strain
e
th
free thermal strain
e
tr
transient strain
e
ult
ultimate compressive strain
e
uT
strain corresponding to peak stress of se curve

uT
strain corresponding to peak stress of full se curve
e
s
instantaneous stress-related strain
s stress
s
1
stress at a joint point used in Anderberg and Thelanderssons model
s
F
compressive stress generated by external forces before heating
s
u0
peak compressive stress at T
0
s
uT
peak compressive stress at T
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 670
concrete columns where the re behaviour is controlled by the combination of the
concrete response and the reinforcement response (indeed the concrete response may
dominate). This effect is exacerbated due to the existence of thermal gradients. Even
for steelwork and concrete members in exure the concept of critical temperature
needs to be abandoned when the effect of level of loading is considered. In the
Eurocode2 [1] a variable load in the re limit state is allowed to be lower than that in
the serviceability limit state. Also, tests on the steel frame building at Cardington
indicated steelwork could achieve far higher temperatures than those predicted even
with allowance for load level [2,3].
The situation for concrete is made worse in that there is a greater variation in
material types than in equivalent steelwork, potential problems due to spalling and a
greater complexity in material modelling. For steel, the total strain has only three
componentsthermal strain, instantaneous stress-related strain and creep strain.
Using anisothermal temperature modied strengths the creep term may be omitted.
The total strain for concrete has an additional termtransient creep strain, which is
due to changes in the chemical composition of the concrete during heating. The
transient creep strain only occurs on heating during the rst heating and cooling
cycle.
It is important that transient strains are included in any analysis involving
concrete in compression [47]. The effect of not including transient strain in a full
stressstrain model can be shown to produce erroneous unsafe results for the
behaviour of columns heated in three sides, thus inducing a thermal moment, in re
[8], although the effect may be exacerbated due to the coexistence of thermal and
moment gradient.
The purpose of this paper is to present a review of the available models for the
mechanical behaviour of concrete at elevated temperatures and to provide the
comparisons between the existing models. Finally, it makes recommendations on
their validity and use of the existing models.
2. Thermal stress analysis
The analysis of thermal stress in concrete structures in general includes two
phases. One is the time-dependent heat transfer in which temperature distributions at
various time stages are calculated. The other is the thermal stress analysis, which uses
the temperature distributions obtained in phase one to carry out the stress analysis.
Since the temperature is time-dependent, the thermal stress is also time-dependent,
although the inertia and damping forces are not considered during this transient
analysis (so-called quasi static analysis). Coupled heat transfer and thermal stress
analyses can be performed nowadays in many commercial software codes such as
ANSYS [9] and FEMLAB [10] provided that the material constitutive equations and
the temperature-dependent material constants are predened.
The model used for heat transfer analysis for concrete is relatively simple if
moisture migration and phase changes within the concrete can be ignored. In this
case, the required temperature-dependent material constants are only the specic
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 671
heat and thermal conductivity coefcient, which are well dened and can be obtained
from books (for example, [8]) or from standard design codes [1]. In contrast, the
temperature-dependent material constants required for thermal stress analysis are
not very well addressed. There are only few publications relating to models for the
mechanical behaviour of concrete at elevated temperatures [1115].
In a thermal stress analysis using nite element methods, the unknowns are usually
the nodal displacements. The corresponding strain thus is the total strain, which
includes four parts, namely, the free thermal strain caused by the change of
temperature, the classical creep strain caused by the dislocation of microstructures of
the material, the transient strain caused by the change of chemical composition and
the instantaneous stress-related strain caused by externally applied stresses.
Mathematically, this is expressed as [8,16]

th
T
s
s; T
cr
s; T; t
tr
s; T, (1)
where e is the total strain, s the corresponding stress, T the temperature, t the time,
e
th
the free thermal strain that is temperature dependent, e
s
is the instantaneous
stress-related strain that is a function of the stress and temperature, e
cr
is the classical
creep strain that is a function of the stress, temperature and time and e
tr
is the
transient strain that is a function of the stress and temperature. Note that the
instantaneous stress-related strain is also stress history dependent, which means that
the stressstrain relation is elastoplastic. When a material becomes plastic the strain
states corresponding to two identical stress states may be different if the two stress
states have different loading histories.
Traditionally, the free thermal strain is expressed by a linear function of
temperature by employing a thermal expansion coefcient, that is,

th
aT T
0
, (2)
where a is the thermal expansion coefcient and T
0
is the initial temperature. For
concrete, a is temperature dependent and can be determined experimentally. For
normal weight concrete with siliceous aggregates a can be approximately expressed
as follows [17]:
a 0:008T 6 10
6
. (3)
Unlike the free thermal strain, which is a function of only temperature, the creep,
transient and instantaneous stress-related strains are functions of the stress,
temperature and time. This makes it quite difcult to separate them during an
experiment. Therefore, some of existing models use only two strains, one is the
stress-induced strain and the other is the transient creep strain, which includes both
the transient and creep strains [12,15]. While in some models all of the three strains
are taken together as a single strain [13].
2.1. Anderberg and Thelandersson model
Anderberg and Thelandersson [11] proposed a model in which the creep, transient
and stress-related strains are separated. Based on experimental data, the classical
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 672
creep strain was assumed as

cr
s; T; t 0:00053
s
s
uT
_ _
t
180
_ _
0:5
e
0:00304T20
, (4)
where s
uT
is the peak compressive stress at temperature T and the unit for time t
used in Eq. (4) is minutes. The classical creep strain is often very small compared to
the other three strains due to the short period of the re. Anderberg has indicated
that, practically, it may be neglected. The transient strain was assumed as:
For Tp550 1C

tr
k
tr
s
s
u0
_ _

th
. (5a)
For TX550 1C
q
tr
qT
0:0001
s
s
u0
_ _
, (5b)
where k
tr
is a constant between 1.8 and 2.35 and s
u0
is the peak compressive stress at
room temperature. Note that the transient strain is assumed to be the same for
loading and unloading as long as the stress is in compression. The instantaneous
stress-related strain was assumed as:
For 0p
s
p
1
s E
s


2
s
2
uT
_ _
. (6a)
For
1
p
s
p
ult
s s
1
E

s

1
(6b)
in which
s
1
E
1


2
1
2
uT
_ _
;
1

uT
1
E

E
_ _
;
uT

2s
uT
E
, (6c)
where E is the initial tangent modulus of the stressstrain curve at temperature T, E

is the tangent modulus of the stressstrain curve in the descending branch at


temperature T, and e
ult
is the ultimate compressive strain at temperature T. Note
that, the strain, e
uT
, at the peak point and the stress and strain, s
1
and e
1
, at the
intersection point between Eqs. (6a) and (6b) on the stressstrain curve can be
expressed in terms of E

, s
uT
and e
ult
. Thus there are only four independent
parameters in the model. If these four parameters are dened then the instantaneous
stress-related strain can be calculated using Eq. (6). Fig. 1 graphically shows the
stressstrain curve described by Eq. (6). In the Anderberg and Thelanderssons
model the tangent modulus of the descending branch was assumed as
E

880 MPa, while the other three parameters were assumed to be functions
of temperature. Eq. (6) is a modication of the conventional nonlinear stressstrain
equation of concrete at room temperature by taking the peak stress, initial tangent
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 673
modulus and ultimate compressive strain as the function of temperature, while Eqs.
(4) and (5) were developed from experimental data.
Khennane and Baker [14] proposed a similar model in which the creep and
transient strains were taken as the same form as those in Anderberg and
Thelanderssons model [11]. The only difference is the instantaneous stress-related
strain, which was split into two parts referred to as elastic and plastic strains.
However, this model does not appear to allow for the reduction of the stress after it
reaches the peak value.
2.2. Schneider model
In the Schneiders model [12], apart from the thermal strain, there are only two
other strain components. One is the transient creep strain, which combines the
transient and creep strains together and the other is the instantaneous stress-related
strain. The stressstrain constitutive equation was taken as follows:

th

s

tr

cr

1 b
g
s
E

F
g
s
E
, (7a)
where b, g and F are empirical functions dependent upon both the stress and
temperature and are dened as follows:
bs; T
1
n 1
s
s
uT
_ _
5
or bs; T
1
n 1

uT
_ _
n
, (7b)
gs
F
; T
1
s
F
s
uo
T 20
100
for
s
F
s
uo
p0:3;
1
0:3T 20
100
for
s
F
s
uo
40:3
_

_
, (7c)
ARTICLE IN PRESS
Fig. 1. Stressstrain curve in Anderberg-Thelanderssons model.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 674
Fs; T
gj
s
s
uT
T 20
100
for
s
s
uT
p0:3;
gj
0:3T 20
100
for
s
s
uT
40:3;
_

_
(7d)
j C
1
tanhg
w
T 20 C
2
tanhg
0
T T
g
C
3
, (7e)
g
w
0:3w 2:2 10
3
, (7f)
where s
F
is the compressive stress generated by external forces before the concrete is
heated, w is the moisture content, n, g
0
, T
g
, C
1
, C
2
and C
3
are constants dened by
Schneider [12] for various different concrete.
The interpretation of the use of functions b, g and F in Eq. (7a) is as follows. The
two terms in the right-hand side of Eq. (7a) represent the instantaneous stress-related
strain and the transient creep strain, respectively. The function b is to account for the
plastic magnication in the stressstrain curve, whereas the function g is to allow for
the increase in initial tangent modulus due to the pre-applied stress. Similar to
Anderberg and Thelanderssons model, Schneiders model also requires the
determination of E and s
uT
, both of which are only functions of temperature.
2.3. Diederichs model
The model proposed by Diederichs [13] is very simple, which can be expressed as
follows:

th

s

tr

cr

s
E
0
1
E
0
s
u0
f T
_ _
, (8a)
f T 3:3 10
7
T 20
3
1:72 10
4
T 20
2
0:0412T 20, 8b
where E
0
is the initial tangent modulus of the concrete material at room temperature
and f(T) is the empirical function obtained by tting the experimental data provided
by Diederichs [13].
2.4. Khoury and Terro model
Khoury et al. [5,18,20] and Terro [15,19] used a three term formulation for total
strain, which is similar to that used in Schneiders model [12], that is,

th

s

tr
. (9)
However, Khoury and Terro assumed e
s
as the elastic strain dened as s=E
0
. Thus
the transient creep strain at a temperature T and a stress level s, called load induced
thermal strain (LITS) is given by
LITST; s
th

s
E
0
_ _
. (10)
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 675
Khoury [20] reported only the experimental results for different concretes. Later,
Terro [15] developed an empirical formula by data tting and by assuming the LITS
to be a linear function of applied stress. The resulting equations are
LITST; 0:3s
u0
43:87 2:73T 6:35 10
2
T
2
2:19 10
4
T
3
2:77 10
7
T
4
10
6
. 11a
This equation is related to a stress level s equal to 0.3s
u0
. For other stress levels,
the strain was determined using the following approximate equation:
LITST; s LITST; 0:3s
u0
0:032 3:226
s
s
u0
_ _
. (11b)
For gravel (siliceous) aggregate concrete Eq. (11a) needs modifying to
LITST; 0:3s
u0
1:48 10
6
1098:5 39:21T 0:43T
2

1:48 10
9
2:44T
3
6:27 10
3
T
4
5:95 10
6
T
5
. 12
Additionally, the master curves were for concrete containing 65% aggregate by
volume. For concrete with a volume fraction of aggregate V
a
, the load-induced-
thermal-strain was given by
LITST; s

V
a
LITST; s

65%
V
a
0:65
, (13)
where V
a
is the volume fraction of aggregate in the concrete. It should be noted that
all of the equations in the model for calculating LITS only apply temperatures up to
590 1C.
3. Model comparisons
In order to compare the four models shown in the preceding section, reduction
functions for the initial tangent modulus, E, and for the peak compressive stress, s
uT
,
are assumed, which are plotted in Fig. 2 and are given by Eqs. (14) and (15),
respectively:
ET
E
0
for Tp60

C;
800T
740
E
0
for 60

CoTp800

C
_
, (14)
s
uT
T s
u0
0:00165
T
100
_ _
3
0:03
T
100
_ _
2
_
0:025
T
100
_ _
1:002
_
, (15)
where E
0
20 GPa and s
u0
25 MPa are the initial tangent modulus and
compressive strength of the concrete at room temperature, respectively. Eqs. (14)
and (15) are developed based on published experimental data [8] and the data
published in Eurocode2 [1]. The constants are assumed as, k
tr
2:35 in Anderberg
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 676
and Thelandersons model and C
1
2:6, C
2
1:4, C
3
1:4, g
0
0:0075, T
g
700,
w 0:3, n 3 in Schneiders model. For simplicity, the classical creep strain in
Anderberg and Thelandersons model and the increase of elasticity due to pre-
applied stress in Schneiders model are ignored (that is,
cr
0 and gs
F
; T 1).
The comparisons of the full stressstrain curves (where the strain does not includ
the free thermal strain) at various temperatures between the four models are plotted
in Fig. 3. Because of the assumptions used in Diederichs and Khoury and Terros
models, the stressstrain curves provided by these two models are straight lines.
Also, although the slopes of these straight lines decrease with the increase of
temperature, there are no descending branches when the stress reaches the maximum
compressive stress. In contrast, the models proposed by Schneider and by Anderberg
and Thelanderson not only provide the nonlinear stressstrain relationship but also
account the stress reduction after the stress reaches the maximum compressive stress.
It is noticed from Fig. 3 that, for a temperature of 40 1C all of the loading curves
are very close. However, as temperature increases, the loading curves provided by
the two linear models diverge from those provided by the two nonlinear models. In
contrast to the two linear models, the loading curves in the two nonlinear models are
reasonably close for the temperature range considered. The main difference between
the two nonlinear models is found in the descending branches where the stress
reduction in Anderberg and Thelandersons model is generally slower than that in
Schneiders model. Since there are no experimental data published in the literature
for describing the descending branches it is difcult to judge which model is more
accurate. Nevertheless, if the slope of the descending branch, E

, in Anderberg and
Thelandersons model is reduced the two models will become close.
ARTICLE IN PRESS
Fig. 2. (a) Reduction of the initial tangent modulus (test data from [8]). (b) Reduction of the peak stress.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 677
Fig. 4 is a re-plot of Fig. 3 for the two nonlinear models in which the stress and
strain are normalised by the stress and strain values at the peak point. It is interesting
to notice that, for the normalised stressstrain curves, the inuence of the
temperature becomes clearer, which can be summarised in two aspects. The rst
one is on the grade of nonlinearity of the loading part of the curve. The second one is
on the slope of the descending branch. It is seen from Fig. 4 that, with the increase of
temperature, the nonlinearity of the loading curve decreases and the slope of the
ARTICLE IN PRESS
Fig. 3. Comparisons of full stressstrain curves of four models at various temperatures.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 678
descending branch increases. This is attributed to the contribution of the transient
strain. The transient strain is a linear function of the stress. For high temperatures
the dominant strain becomes the transient strain and this is why the stressstrain
curve in the loading part becomes almost linear. On the other hand, when the stress
reduces the transient strain follows to reduce and therefore the slope of the
descending branch increases with the temperature. This is why the strain prole is
convex in the descending branch. Unfortunately, there are no available experimental
data to validate this strain behaviour as it would be difcult, not impossible, to carry
out experimental verication, although the explanation from the reduction of the
transient strain appears reasonable.
The dependence of the strain value at the peak point on temperature in the two
nonlinear models is plotted in Fig. 5, which are also compared with the values given
in Eurocode2 [1]. Interestingly, for temperatures up to 400 1C, the peak strain values
calculated from Anderberg and Thelandersons model are found to be very close to
the values published in Eurocode2. It should be mentioned here that in EN 1992-1-2
it is ambiguous as to whether the peak strain value includes the transient strain.
For the convenience of computation, an empirical formula for the peak strain is
developed here and is given by Eq. (16), which is based on Anderberg and
Thelandersons model for temperatures below 350 1C and on the average of the three
models shown in Fig. 5 for temperatures over 350 1C,

uT
T
2s
u0
E0
0:21 10
4
T 20 0:9 10
8
T 20
2
. (16)
ARTICLE IN PRESS
Fig. 4. Normalised full stressstrain curves for different temperatures.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 679
Based on the stress and strain values at the peak point, one can calculate the secant
modulus (that is, E

av
s
uT
=

uT
) or the initial tangent modulus if the nonlinear
stressstrain equation given in EN 1992-1-2 is employed (that is, E

1:5s
uT
=

uT
).
ARTICLE IN PRESS
Fig. 5. Temperature dependence of the strain at the peak stress point.
Fig. 6. (a) Reduction of secant modulus. (b) Reduction of initial tangent modulus.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 680
Fig. 6 plots the variations of the secant modulus and the initial tangent modulus with
temperature.
For highlighting the difference between the stressstrain curves with and
without considering the transient strain, the reduction function of the initial tangent
modulus of the stressstrain curve excluding the transient strain (that is, Eq. (14))
is also plotted in Fig. 6. The difference between E and E
*
clearly indicates
the importance of including the transient strain in the analysis when temperature
is high.
In many commercial nite element analysis codes, one can input temperature-
dependent stressstrain curve dened by the temperature-dependent peak stress and
temperature-dependent secant modulus or temperature-dependent initial tangent
modulus. It is clear that, if the kind of FEA codes are used, Eq. (15) can be used to
dene the peak stress and the secant modulus or the initial tangent modulus can be
calculated from Eqs. (15) and (16).
For the tangent modulus in the descending branch of the full stressstrain curve,
the following equation is suggested:
E

p
E

0
expk
p
T 20
2:15
, (17)
where E

p
is the tangent modulus in the descending branch at temperature T, E

0

E

880 MPa is the tangent modulus in the descending branch at room


temperature, and k
p
10
6
is a constant obtained by the best t of Eq. (17) with
Anderberg and Thelanderssons full stressstrain curves for various different
temperatures.
A comparison of the full stressstrain curves between Anderberg and Thelan-
derssons model, the present model using the peak stress, initial tangent modulus and
tangent modulus of the descending branch in the full-stressstrain curve, and EN-
1992-1-2 is shown in Fig. 7 for various temperatures. The present model is very
consistent with Anderberg and Thelanderssons model, which is not surprising as the
parameters in the present model are determined based on Anderberg and
Thelanderssons model. The only difference is the presentation of the stressstrain
curve. In the present model the stressstrain equation is expressed as the full
stressstrain equation, which is more convenient for use, while in Anderberg and
Thelanderssons model the strains are separated. For most temperatures the
stressstrain curves provided by EN 1992-1-2 are signicantly different from those
provided by Anderberg and Thelanderssons model. The main causes of the
difference are found to be the peak strain and the tangent modulus in the descending
branch. From Fig. 7 it is evident that the full stressstrain curves provided in EN
1992-1-2 for higher temperatures are unconservative.
At lower temperatures when transient strain may not be important there is little
difference between Anderberg and Thelandersson, Li and Purkiss or the curves in
EN 1992-1-2. However, as the temperatures increase EN 1992-1-2 predicts higher
strains at a given stress level or perhaps more critically predicts a higher peak strain
at a given temperature. This means that the EN 1992-1-2 curves will predict a higher
ductility.
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 681
Provided strains do not exceed values at the peak then the choice of transient
strain models may not be critical, but it should be noted that the models due to
Diederichs and Khoury (extended by Terro) give much lower values of stress at a
given strain. Because of the likely existence of unloading (or strain softening) in the
numerical simulation presented later in the paper it was considered inappropriate to
examine Diederichs or Khourys models.
4. Numerical example
In order to demonstrate the importance of including the transient strain in
analysing concrete structure when it is in re, the numerical example of a concrete
wall of 1000 mm wide and 400 mm thick, subject to double-sided re exposure is
discussed here. The temperature distribution within the wall cross-section is
calculated by solving the one-dimensional heat transfer equation using the standard
re temperature curve and temperature-dependent specic heat and temperature-
dependent thermal conductivity coefcient, which are taken from Eurocodes. The
temperature distributions along the wall thickness direction at three different times
together with the parametric values used are shown in Fig. 8. The wall is assumed to
have a compressive load and the compressive strain is uniformly distributed in the
cross-section although the corresponding compressive stress is the function of the
coordinate along the thickness direction because of the non-uniformly distributed
temperature.
ARTICLE IN PRESS
Fig. 7. Comparisons of full stressstrain curves at temperatures 40, 100, 200, 300, 400, 500, 600 and
700 1C.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 682
Fig. 9 shows the loaddisplacement curves of the wall, when it is exposed to a re
by 20 and 60 min, respectively. The two sets of results shown in the gure correspond
to two models; one ignores the transient strain, that is to use Eq. (15) as the peak
stress and Eq. (14) as the initial tangent modulus and E

880 MPa as the tangent


modulus in the descending branch; the other includes the transient strain, using
ARTICLE IN PRESS
Fig. 8. Temperature distribution in the cross-section at different times.
Fig. 9. Compressive loaddisplacement curves (a) at 20 min and (b) at 60 min (A cross-section area).
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 683
Eq. (15) as the peak stress, Eq. (14) as the initial tangent modulus and Eq. (17) as the
tangent modulus in the descending branch. The difference between the two sets of
results reects the inuence of the transient strain. It is seen from Fig. 9a that, for
short time exposure, the inuence of the transient strain on the loaddisplacement
curve is not very signicant and thus it has a little inuence on the re performance
of the wall. However, when the exposure time is not short, in which case the
temperature in the concrete is high, the inuence of the transient strain on the
loaddisplacement curve becomes remarkable and thus it has considerable inuence
on the re performance of the wall, as is demonstrated by Fig. 9b.
Figs. 10 and 11 show the compressive stress distributions in the cross-section of
the wall for three different compressive strains at two different exposure times. It is
seen that, except for the case where the strain is very large, for all other cases the
stress distributions obtained from the two models are signicantly different. This
demonstrates the importance to consider the transient strain in the analysis. It is
found from the gures that, for the same axial compressive strain, the model
excluding the transient strain predicts higher compressive stress than the model
including the transient strain. Thus, the loading capacity is over-predicted by the
model excluding the transient strain.
5. Conclusions
It is possible to formulate a stressstrain-temperature model for concrete
implicitly including transient strain.
ARTICLE IN PRESS
Fig. 10. Axial compressive stress distribution at 20 min.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 684
The original models due to Anderberg and Thelanderson or Schneider produce
very similar results as do Diederichs or Khoury.
If strain softening (or apparent unloading) occurs during the strain history it is
recommended that Anderberg and Thelandersons or Schneiders model is used.
It would appear that at high temperatures the stressstrain curves in EN 1992-1-2
are unsafe if there are high axial loads due to high peak strains.
It is observed that ignoring transient strain where there are signicant axial
compressive loads is unsafe.
References
[1] EUROCODE2. Design of concrete structures Part 1.2: general rules, structural re design. EN 1992-
1-2. Brussels: European Committee for Standardisation; 2003.
[2] Bailey CG, Lennon T, Moore DB. The behaviour of full-scale steel-framed buildings subjected to
compartment res. Struct Eng 1999;77(8):1521.
[3] Bailey CG, Moore DB, Lennon T. The structural behaviour of steel columns during a compartment
re in a multi-storey braced steel-frame. J Construct Steel Res 1999;52(2):13757.
[4] Sullivan PJE, Khoury GA, Grainger BN. Apparatus for measuring the transient thermal strain
behaviour of unsealed concrete under constant load for temperatures up to 700 1C. Masg Concr Res
1983;35(125):22936.
[5] Khoury GA, Grainger BN, Sullivan PJE. Transient thermal strain of concrete: literature review,
conditions within specimen and behaviour of individual constituents. Mag Concr Res 1985;
37(132):13144.
[6] Purkiss JA, Bali A. The transient behaviour of concrete at temperatures up to 800 1C. In: Proceedings
of the 10th Ibausil (Weimar 1988), Hochschule fur Architektur und Bauwesen, Werimar, Section 2/1,
1988. p. 2349.
ARTICLE IN PRESS
Fig. 11. Axial compressive stress distribution at 60 min.
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 685
[7] Nielsen CV, Pearce CJ, Bicanic N. Improved phenomenological modelling of transient thermal
strains for concrete at high temperatures. Comput Concrete 2004;1(2).
[8] Purkiss JA. Fire safety engineering, design of structures. Oxford: Butterworth Heinemann; 1996.
[9] Moaveni S. Finite element analysistheory and application with ANSYS. New Jersey, USA:
Prentice-Hall; 2003.
[10] Comsol. Femlab 3, Multiphysics Modelling. Stockholm, Sweden, 2004 (www.comsol.com).
[11] Anderberg, Y., Thelandersson, S., 1976. Stress and deformation characteristics of concrete,
2-experimental investigation and material behaviour model. Bulletin 54, University of Lund, Sweden.
[12] Schneider U. Modeling of concrete behaviour at high temperature. In: Anchor RD, Malhotra HL,
Purkiss JA, editors. Design of structures against re. New York: Elsevier; 1986. p. 5369.
[13] Diederichs, U., 1987. Modelle zur Beschreibung der Betonverformung bei instantionaren
Temperaturen. In Abschlubkolloquium Bauwerke unter Brandeinwirkung, Technische Universitat,
Braunschweig, p. 2534.
[14] Khennane A, Baker G. Uniaxial model for concrete under variable temperature and stress. J Eng
Mech, ASCE 1993;119(8):150725.
[15] Terro MJ. Numerical modelling of the behaviour of concrete structures. ACI Struct J
1998;95(2):18393.
[16] Buchanan AH. Structural design for re safety. Chichester: Wiley; 2001.
[17] Lie TT. Structural re protection. New York: American Society of Civil Engineers; 1992.
[18] Khoury GA, Dias WPS, Sullivan PJES. Deformation of concrete and cement paste loaded at
constant temperatures from 140 to 720 1C. Mater Struct 1986;19(110):97104.
[19] Terro MJ. Numerical modelling of thermal and structural responses of reinforced concrete structures
in re. PhD thesis, Department of Civil Engineering, Imperial College, 1991.
[20] Khoury GA. Strain components of nuclear-reactor-type concretes during rst heat cycle. Nucl Eng
Des 1995;156(1/2):31332.
ARTICLE IN PRESS
L.-y. Li, J. Purkiss / Fire Safety Journal 40 (2005) 669686 686
ROLE OF ADMIXTURES
ROLE OF ADMIXTURES
IN
IN
READYMIX CONCRETE
READYMIX CONCRETE
Chemical Admixtures
ASTM C 494
EN 934 2
IS 9103:2003
Chemical Admixtures
Type Description
A Water Reducing admixture (WRA)
B Retarding Admixture
C Accelerating Admixture
D Water Reducing & Retarding Admixture
E Water Reducing & Accelerating Admixture
F High Range Water Reducing Admixture (HRWRA)
G High Range Water Reducing & Retarding Admix
Main
Main
effect
effect
of
of
plasticizing
plasticizing
admixtures
admixtures
1 2
Higher workability at
constant w/c ratio
Increase of strength parameters through
reduction of w/c ratio at constant slump flow
S
l
u
m
p

f
l
o
w

[
c
m
]
with
Plasticizer
without
Plasticizer
Water/Cement Ratio
1
2
Types of concrete admixtures
Lignosulphonates LS
Beta-Naphthalenesulphonate - Polycondensate BNS
Melamine-Sulphonate-Poly-condensate MFS
Poly Acrylates PAC
Polycarboxylate Ethers - PCE
Traditional Plasticizers/Superplasticizers
BNS molecules
Electrostatic
Repulsion
Mode of Action
1950 1960 1970 1980 1990 2000 2010
V
o
l
u
m
e

o
f

A
d
m
i
x
t
u
r
e
s
Conventional AE water-
reducing agent [Pozzolith]
High water-reducing
agent [Rheobuild]
AE high water-reducing agent
Super plasticizer [Glenium]
Innovation History of Water Reducer Technology
Year
PCE Molecule
complex and flexible
molecules,
comprising of main chain
Side chains
functional groups.
PCE Based - GLENIUM Molecules
GLENIUM molecules
Electrostatic
and Steric
Repulsion
Glenium Concrete
High grades
Low Grades
SCC
Precast
Glenium Concrete for Higher Grades
Why higher grades? (M45 M 60)
Early destripping
Advantages in column size reduction
Longer spans and thinner decks
PT slabs 30 MPa in 3 days
General Practice for Higher Grade
Use of Higher Cement contents
Use of very low water /cement ratios
Use of Expensive Mineral Admixtures
Phenomenal increase in costs
Concrete very sensitive to changes in moisture content
Extra cement content due to retention effect
Critical Areas
Use of higher cement content Thermal Shrinkage
Use of silica fume Risk of Plastic shrinkage
Use of low w/c ratio Higher dosage of SNF admixture
(retardation effect-plastic shrinkage)
Higher cost of concrete!
Hydration Reactions
Glenium Concrete
GLENIUM
TM
molecules
Improved Dispersion
due to electrostatic and
steric repulsion
.low w/cm
How does Glenium Work?
Glenium Concrete
Excellent dispersion of binders
Water reduction upto 40%
High early strengths
Reduce/Eliminate SF
Reasonably lower w/c ratios
Better Hydration
Reduction in Cement content
Lower risk to Thermal Shrinkage
Faster de-stripping possible
Economical concrete
Tolerant to changes in moisture
content
GLENIUM helps in
Typical Glenium Mix M 60
Nil Nil Depth of water penetration
71.11 72.34 28 Days
58.2 58.47 7 Days
37.40 39.46 3 Days
18.30 19.02 1 Day
0.28 0.28 W/B Ratio
1.2% XXX Glenium Range
xxx 1.8% Traditional Admix
161 158 Water
1722 1782 Total Aggregates
575 565 Binder
0 35 Silica Fume
150 80 Flyash
425 430 Cement
Mix with Glenium Mix with SF
Economical Glenium Concrete
160
50
0.08
0.40
30
1.0
4.25
Rs./Kg
3778 4191 Material Cost/Cu.M
1120 0 Glenium Range
0 509 Trad. Admixture
13 11 Water
689 713 Total Aggregates
0 1050 Silica fume
150 80 Flyash
1806 1828 Cement
Mix With Glenium Mix with SF
Cost Saving ~ Rs. 400/M
3
Glenium Concrete for Lower Grades
Requirements in Lower Grade concrete
Early age strengths
Reduction in cement
Addition of Fly ash
Lower dosage of admixtures
Extended retention 2 to 3 hrs
Avoid retempering of Concrete
Tolerance to water (10 15 Litres)
Glenium Concrete
Excellent dispersion of binders
Water reduction
lower w/c ratios
Better Hydration
Reduction in Cement content
Early Strength gain
Addition of Flyash
Economical concrete
Tolerant to changes in water content
GLENIUM helps in
Typical Glenium Mix M 30
38.40 35.32 28 Days
28.92 26.42 7 Days
18.40 16 .23 3 Days
7.8 8.30 1 Day
0.42 0.46 W/B Ratio
0.6% XXX Glenium Range Admixture
xxx 1.0% Traditional Admix
152 162 Water
1820 1842 Total Aggregates
340 340 Binder
120 60 Flyash
220 280 Cement
Glenium Concrete Traditional M 30
Economical Glenium Concrete
130
28
0.08
0.40
1.0
4.25
Rs./Kg
2060 2095 Material Cost/Cu.M
265 0 Glenium Range
0 95 Trad. Admixture
12 13 Water
728 737 Total Aggregates
120 60 Flyash
935 1190 Cement
Mix With Glenium Traditional
Thank You!!

Vous aimerez peut-être aussi