Upload
m-m
View
215
Download
0
Embed Size (px)
Citation preview
http://ijd.sagepub.com/Mechanics
International Journal of Damage
http://ijd.sagepub.com/content/early/2011/11/28/1056789511431991The online version of this article can be found at:
DOI: 10.1177/1056789511431991
published online 21 December 2011International Journal of Damage MechanicsKaykha
Ali Nazari, Shadi Riahi, Gholamreza Khalaj, Hamid Bohlooli and Mohammad Mehdiand Rice Husk-Bark Ash by Gene Expression Programming
Prediction of Compressive Strength of Geopolymers with Seeded Fly Ash
- Oct 15, 2012version of this article was published on more recent A
Published by:
http://www.sagepublications.com
can be found at:International Journal of Damage MechanicsAdditional services and information for
http://ijd.sagepub.com/cgi/alertsEmail Alerts:
http://ijd.sagepub.com/subscriptionsSubscriptions:
http://www.sagepub.com/journalsReprints.navReprints:
http://www.sagepub.com/journalsPermissions.navPermissions:
What is This?
- Dec 21, 2011OnlineFirst Version of Record >>
- Oct 15, 2012Version of Record
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Prediction of Compressive Strengthof Geopolymers with Seeded Fly Ashand Rice Husk�Bark Ash by Gene
Expression Programming
ALI NAZARI,*,1 SHADI RIAHI,1 GHOLAMREZA KHALAJ,1
HAMID BOHLOOLI1
AND MOHAMMAD MEHDI KAYKHA2
1Department of Mechanical Engineering, Birjand Branch, IslamicAzad University, Birjand, Iran
2Department of Mechanical Engineering, University of Zabol,Zabol, Iran
ABSTRACT: In the present work, compressive strength of inorganic polymers (geo-polymers) made from seeded fly ash and rice husk�bark ash has been predicted bygene expression programming. Different specimens were subjected to compressivestrength tests at 7 and 28 days of curing. One set of the specimens were cured at roomtemperature until reaching to 7 and 28 days and the other sets were oven cured for36 h at the range of 40�90�C and then room cured until 7 and 28 days. A modelbased on gene expression programming for predicting the compressive strength ofthe specimens was presented. To build the model, training and testing using exper-imental results from 120 specimens were conducted. According to the input param-eters, in the gene expression programming models, the compressive strength of eachspecimen was predicted. The training and testing results in the gene expression pro-gramming models have shown a strong potential for predicting the compressivestrength of the geopolymer specimens.
KEY WORDS: compressive strength, fly ash, FTIR, gene expression programming,geopolymer, particle size, rice husk�bark ash, seeded mixture.
INTRODUCTION
GEOPOLYMER WHICH WAS developed by Davidovits (1991) contains bothsilica and alumina which can act as a binder for geopolymerization.
*Author to whom correspondence should be addressed. E-mail: [email protected]
Figures 2�5, 8, 9 and 11 appear in color online: http://ijd.sagepub.com
International Journal of DAMAGE MECHANICS, 2011 1
1056-7895/11/00 0001–19 $10.00/0 DOI: 10.1177/1056789511431991� The Author(s), 2011. Reprints and permissions:http://www.sagepub.co.uk/journalsPermissions.nav
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Various alkali activators also play a key role in producing geopolymers bydissolving silica and alumina from the raw material and forming alumino-silicate structures. Geopolymer is used for a variety of applications such asfor sculpture, building, repairing, and estoration. Numerous research pub-lications related to geopolymers have been released, with some reporting onchemical composition aspects or reaction processes, while others presentresults related to mechanical properties and durability (Wongpa et al.,2010). The general formula to describe the chemical composition of thesemineral polymers is Mn[�(SiO2)z�AlO2]n�wH2O, where z is 1, 2, or 3, M is analkali cation (such as potassium or sodium), and n is the degree of polymer-ization. Accordingly, in order to better describe the geopolymeric structures,a terminology has been proposed: poly(sialate) (�Si�O�Al�O�), poly(sia-late-siloxo) (�Si�O�Al�O�Si�O�), and poly(sialatedisiloxo)(�Si�O�Al�O�Si�O�Si�O�; Davidovits, 1991). The main properties ofgeopolymers are: quick compressive strength development, low permeabil-ity, resistance to acid attack, good resistance to freeze�thaw cycles, andtendency to drastically decrease the mobility of most heavy metal ions con-tained within the geopolymeric structure (van Jaarsveld et al., 1997). Suchproperties make them interesting structural products, such as concretereplacements in various environments, and immobilization systems forheavy metal containment (Alvarez-Ayuso et al., 2008).
The compressive strength of an inorganic polymer depends on both theratio of Si/Al and the types of the utilized raw material. Fly ash (FA) isrecently used as a source material to produce geopolymer because of itssuitable chemical composition along with favorable size and shape. TheFA is a by-product of coal-fired electric power stations. Literature surveyspecifies that FA is primarily composed of SiO2, Al2O3, and Fe2O3. Sincethe quality of FA depends on the type and the quality of coal along with theperformance of the power plant, difficulties sometimes remain to control itschemical composition. In order to achieve a suitable chemical compositionto produce geopolymers, the preferred method is to blend FA with anotherhigh-silica source (Wongpa et al., 2010).
Rice husk�bark ash (RHBA) is a solid waste generated by biomass powerplants using rice husk and eucalyptus bark as fuel. The power plant com-pany providing RHBA for this research reported that about 450 tons/day ofRHBA are produced and discarded. The major chemical constituent ofRHBA is SiO2 (about 75%; Sata et al., 2007; Tangchirapat et al., 2008).Therefore, blending FA and RHBA can adjust the ratio of Si/Al as required.
Genetic programming (GP) has begun to arise for the explicit formulationof the properties and the performances of engineering materials recently(Cevik and Sonebi, 2009; Milani and Nazari, 2011; Nazari et al., 2011).The GP offers many advantages as compared to classical regression
2 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
techniques. Regression techniques are often based on predefined functionswhere regression analyses of these functions are later performed. On theother hand, in the case of GP approach, there is no predefined functionto be considered. In this sense, GP can be accepted to be superior to regres-sion techniques and neural networks. The GP has proven to be an effectivetool to model and obtain explicit formulations of experimental studiesincluding multivariate parameters where there are no existing analyticalmodels (Cevik and Sonebi, 2009; Milani and Nazari, 2011; Nazari et al.,2011). There are several works (Cevik, 2007; Cevik and Guzelbey, 2007;Cevik and Sonebi, 2008) in the literature addressing the utilization of geneexpression programming (GEP) for engineering problems.
As authors’ knowledge, there are no works on utilizing a mixture of FAand RHBA with seeded distribution of particles to produce geopolymers. Aliterature survey shows that utilizing fine pozzolan materials could producehigh-strength concrete specimens (Chindaprasirt et al., 2004). However, fewworks have been conducted on this fact in geopolymers. In addition, sincethe concept of geopolymers is completely new and there are few works ontheir properties, application of computer programs like GEP to predict theirproperties is rarely reported. The aim of this study is to investigate thecompressive strength of geopolymers produced form seeded FA andRHBA mixture experimentally and presenting suitable model based onGEP to predict their compressive strength. Both FA and RHBA with twodifferent particle size distributions have been mixed with different amountsto produce four classes of geopolymers. Compressive strength of the pro-duced specimens has been investigated after specific times of curing. Totally120 data of compressive strength tests in different conditions were collected,trained, and tested by means of artificial neural networks. The obtainedresults have been compared by experimental ones to evaluate the softwarepower for predicting the compressive strength of the geopolymer specimens.
EXPERIMENTAL PROCEDURE
The cementitious materials used in this work were FA and RHBA. Theirchemical composition has been illustrated in Table 1. In addition, Figure 1shows scanning electron microscopy (SEM) micrograph of the cementitiousmaterials, respectively. The as-received ashes were sieved and the particlespassing the finesses of 150 mm and 33 mm were grinded using Los Angelesmill for 30 and 180min, respectively, which yielded two different samples foreach of FA and RHBA. The average particle sizes obtained for FA were75 mm (coarser FA named cF in this study) and 3 mm (finer FA named fF inthis study) with the BET specific surface of 31.3 and 38.9 m2/g, respectively.
Prediction of Compressive Strength of Geopolymers 3
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Figure 1. Scanning electron microscope (SEM) micrograph of (a) fly ash (FA) and (b) ricehusk�bark ash (RHBA) used in this study.
Table 1. Chemical composition of FA, RHBA, and WG (Wt.%).
Material SiO2 Al2O3 Fe2O3 CaO SO3 Na2O Loss on ignition
FA 35.21 23.23 12.36 20.01 2.36 0.36 0.24RHBA 81.36 0.4 0.12 3.23 0.85 — 3.55WG 34.21 — — — — 13.11 —
Note. FA: fly ash, RHBA: rice husk�bark ash, WG: water glass.
4 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
The average particle sizes obtained for RHBA were 90 mm (coarser RHBA
named cR in this study) and 7 mm (finer RHBA named fR in this study) with
the BET specific surface of 26 and 33.1 m2/g, respectively. The four pro-
duced samples were used in the experiment. Figure 2 shows the particle size
distribution of the four produced samples.Sodium silicate solution or water glass (WG) and sodium hydroxide
(NaOH) were used as the solution part of the mixture. The WG was used
without following modification, but the NaOH was diluted to different con-
centrations before using. The chemical composition of the utilized WG is
also given in Table 1.Totally four series of geopolymer specimens each contain two different
mixture of FA and RHBA as illustrated in Table 2 were prepared for com-
pressive strength tests. The mixed alkali activator of sodium silicate solution
and NaOH was used. The NaOH was diluted by tap water to have concen-
trations of 4, 8, and 12M. The solution was left under ambient conditions
until the excess heat had completely dissipated to avoid accelerating the
setting of the geopolymeric specimens. The sodium silicate solution without
preparation was mixed with the NaOH solution. The ratio of the sodium
silicate solution to NaOH solution was 2.5 by weight for all mixtures
because this ratio demonstrated the best properties for FA-based geopoly-
mer (Pacheco-Torgal et al., 2005, 2007). For all samples, the mass ratio of
alkali activator to FA�RHA mixture was 0.4. Pastes were mixed by shaking
for 5�10min to give complete homogenization. The mixtures were cast in
ø30mm� 60mm polypropylene cylinders. The mixing was done in an air-
conditioned room at approximately 25�C. The molds were half-filled,
vibrated for 45 s, filled to the top, again vibrated for 45 s, and sealed with
the lid. The mixtures were then precured for 24 h at room temperature (this
precuring time has been found to be beneficial to strength development;
Bakharev, 2005). Precuring time before application of heat induces
0102030405060708090
100
0 20 40 60 80 100 120 140
Per
cen
tag
e fi
ner
th
an
Particle size (µm)
Coarse FA
Fine FA
Coarse RHBA
Fine RHBA
Figure 2. Particle size distribution pattern of the different ashes used in this study.
Prediction of Compressive Strength of Geopolymers 5
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
significant dissolution of silica and alumina from FA and formation of a
continuous matrix phase, increasing, therefore, the homogeneity of the geo-
polymeric materials (Chindaprasirt et al., 2007; Zuhua et al., 2009). After
the precuring process, the samples and molds were placed in a water bath to
prevent moisture loss and the carbonation of the surface. One batch of these
samples was placed in an air-conditioned room at 25�C. The other batch was
put in the oven at elevated temperatures of 50�90�C for 36 h. To determine
the most effective alkali concentration on compressive strength, one set of
the specimens cured at 80�C for 36 h were subjected to compressive strength
tests. Afterwards, the other sets of samples were tested at 7 and 28 days of
curing (for the specimens cured in elevated temperature, the time of oven-
curing were also considered).The compressive strength results of the produced specimens were mea-
sured on the cylindrical samples. The tests were carried out triplicately and
average strength values were obtained.The microstructural characteristic of geopolymeric specimens, which was
made at the optimum condition and had a high compressive strength, was
analyzed using Fourier transform infrared spectroscopy (FTIR). The
powder samples were mixed with KBr at a concentration of 0.2�1.0wt%
Table 2. Mixture proportioning of the utilized FA and RHBA to producegeopolymeric specimens.a
Sampledesignation
Weightpercent of
fine FA(fF wt%)
Weightpercent ofcoarse FA(cF wt%)
Weightpercent offine RHBA(fR wt%)
Weightpercent of
coarse RHBA(cR wt%)
SiO2/Al2O3
ratio
fF�fR-1 60 0 40 0 3.81fF�fR-2 70 0 30 0 2.99fF�fR-3 80 0 20 0 2.38fF�cR-1 60 0 0 40 3.81fF�cR-2 70 0 0 30 2.99fF�cR-3 80 0 0 20 2.38cF�fR-1 0 60 40 0 3.81cF�fR-2 0 70 30 0 2.99cF�fR-3 0 80 20 0 2.38cF�cR-1 0 60 0 40 3.81cF�cR-2 0 70 0 30 2.99cF�cR-3 0 80 0 20 2.38
Note. FA: fly ash, RHBA: rice husk�bark ash, fF: fine fly ash, cF: coarse fly ash, fR: fine rice husk�barkash, cR: coarse rice husk�bark ash, WG: water glass, NaOH: sodium hydroxide.aAlkali activator (WGþNaOH) to FA�RHBA mixture ratio is 0.4.
6 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
to make the KBr disks. Then the disks were evaluated using a Perkin ElmerFTIR microscope.
EXPERIMENTAL RESULTS AND DISCUSSION
Compressive Strength
To determine the effect of NaOH concentration on compressive strengthof the specimens, one series of the specimens (fF�fR series) were randomlyselected, produced by different concentrations of NaOH (4, 8, and 12M)and subjected to compressive strength tests after 36 h oven curing. Figure 3shows the effects of NaOH concentration on compressive strength of thefF�fR series specimens cured for 36 h at 80�C. As Figure 3 shows, thecompressive strength by the geopolymers synthesized using the most con-centrated alkaline solution (12M NaOH) was the highest for mixtures.Zuhua et al. (2009) reported that the use of high molarities NaOH (suchas 12M) could accelerate dissolution and hydrolysis but obstruct polycon-densation. Thus, 12M NaOH can be considered as the suitable solution forpreparing geopolymer specimens. Therefore, as the way discussed in theexperimental section, the specimens were produced by 12M NaOH solutionand were cured at the mentioned time and temperatures.
The compressive strength of the produced specimens has been illustratedin Figures 4 and 5 for 7 and 28 days of curing. Figures 4 and 5 show that thebest strength has been achieved for fF�fR2 specimen cured at 80�C for 36 h
0
5
10
15
20
25
30
35
0 2 4 6 8 10 12 14
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Concentration of NaOH (M)
fF-fR-1
fF-fR-2
fF-fR-3
Figure 3. The effect of sodium hydroxide (NaOH) concentration on compressive strength offine fly ash (fF)�fine rice husk�bark ash (fR) series specimens cured at 80�C for 36 h.
Prediction of Compressive Strength of Geopolymers 7
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
05
101520253035404550
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of fF-fR series at 7 days of curing(a)
(b)
fF-fR-1
fF-fR-2
fF-fR-3
05
1015202530354045
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of fF-cR series at 7 days of curing
fF-cR-1
fF-cR-2
fF-cR-3
0 20 40 60 80 100
0 20 40 60 80 100
(c)
(d)
0
5
10
15
20
25
30
35
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of cF-fR series at 7 days of curing
cF-fR-1
cF-fR-2
cF-fR-3
0
5
10
15
20
25
30
0 20 40 60 80 100
0 20 40 60 80 100
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of cF-cR series at 7 days of curing
cF-cR-1
cF-cR-2
cF-cR-3
Figure 4. In all, 7 days compressive strength of (a) fine fly ash (fF)�fine rice husk�bark ash(fR), (b) fF�coarse rice husk�bark ash (cR), (c) coarse fly ash (cF)�fR, and (d) cF�cR seriesspecimens.
8 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
0
10
20
30
40
50
60
70
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of fF-fR series at 28 days of curing(a)
(b)
fF-fR-1
fF-fR-2
fF-fR-3
0
10
20
30
40
50
60
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of fF-cR series at 28 days of curing
fF-cR-1
fF-cR-2
fF-cR-3
0 20 40 60 80 100
0 20 40 60 80 100
(c)
(d)
05
101520253035404550
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of cF-fR series at 28 days of curing
cF-fR-1
cF-fR-2
cF-fR-3
0
5
10
15
20
25
30
35
40
0 20 40 60 80 100
0 20 40 60 80 100
Co
mp
ress
ive
stre
ng
th (
Mp
a)
Oven curing temperature (°C)
Compressive strength of cF-cR series at 28 days of curing
cF-cR-1
cF-cR-2
cF-cR-3
Figure 5. In all, 28 days compressive strength of (a) fine fly ash (fF)�fine rice husk�bark ash(fR), (b) fF�coarse rice husk�bark ash (cR), (c) coarse fly ash (cF)�fR, and (d) cF�cR seriesspecimens.
Prediction of Compressive Strength of Geopolymers 9
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
in both 7 and 28 days curing regimes. As Figures 4 and 5 show, the optimumcuring condition for the all mixtures is at 80�C. Curing temperature has asignificant effect on the compressive strength development because it affectsspecimens setting and hardening. Synthesized products are known to be verysensitive to experimental conditions (Fernandez-Jimenez et al., 2007).However, compressive strength begins to decrease after curing for a certainperiod of time at higher temperatures. Prolonged curing at higher temper-atures can break down the granular structure of geopolymer mixture. Thisresults in dehydration and excessive shrinkage due to contraction of the gel,which does not transform into a more semicrystalline form (van Jaarsveldet al., 2002).
On the whole, samples made with the fine RHBA and FA particles (fF-fRseries) showed considerably higher strength than the other series. This maybe due to production of more compacted specimens. Fine particles are capa-ble of filling the vacancies and producing more densified specimens, whichmake them stronger to the applied loads. This has been confirmed in someworks done on concrete specimens (Naji Givi et al., 2010), but as authors’knowledge there is not any reports to confirm this matter in geopolymers.On the other hand, the fine particle with high surface areas enabled the silicaand alumina leaching out to the solution and enhanced the geopolymeriza-tion (Chindaprasirt et al., 2011).
FTIR Results
Geopolymers are composed from Si�O tetrahedrons, which are connectedvia corner sharing bridging oxygen. The connectivity of the tetrahedrons isspecified by the number of bridging oxygen. Tetrahedrons with n bridgingoxygens are denoted Qn (n¼ 0, 1, 2, 3, or 4). Thus, silicon in Q3 configura-tion is surrounded by three bridging oxygen and a nonbridging oxygen.Amorphous SiO2 is assumed to consist of only Q4 species forming a contin-uous random network (Karlsson et al., 2005). The FTIR spectra of theinorganic polymers synthesized using different NaOH concentrations areshown in Figure 6. In Figure 6, the IR bands are identified as follows:Si�O stretching is located in the range 1000�1200 cm�1, Si�O bending isfound at 800 cm�1 and between 890 and 975 cm�1. The band at approxi-mately 1100 cm�1 is assigned to the Si�O stretching of Q4 units and the bandat 1050 cm�1 is assigned to Q3 units with a nonbridging oxygen(Si�O�NBO) per SiO4 tetrahedron (Zholobenko et al., 1997).
From Figure 6, it is seen that an increase in the fF particles shifts theposition of the maximum absorbance of Si�O bands toward lower wavenumbers, indicating the transformation of Q4 units to Q3 units. Moreover,the emerging of a new band centered on 900 cm�1 is observed, which is
10 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
assigned to Si�O stretching of Q2 unit with two nonbridging oxygen perSiO4 tetrahedron (Serra et al., 2002). The observed increase in the signal at2380 cm�1 can be assigned to CO2 which has reacted with increasingamounts of NaOH leading to the formation of HCO�3 .
GENETIC PROGRAMMING
The GP proposed by Koza (Rosenblatt, 1962) is an extension to geneticalgorithms (GA). Koza (Rosenblatt, 1962) defines GP as a domain indepen-dent problem-solving approach in which computer programs are evolved tosolve, or approximately solve, problems based on the Darwinian principle ofreproduction and survival of the fittest and analogs of naturally occurringgenetic operations such as crossover and mutation. The GP reproducescomputer programs to solve problems by executing the steps as shown inFigure 7. This figure is a flowchart showing the executional steps of a run ofGP. The flowchart demonstrates the genetic operations in addition to thearchitecture chancing operations. Also, this flowchart demonstrates a twooffspring version of the crossover operation.
The GEP software which is used in this study is an extension to GEP thatevolves computer programs of different sizes and shapes encoded in linearchromosomes of fixed length. The chromosomes are composed of multiplegenes, each gene encoding a smaller subprogram. Furthermore, the struc-tural and functional organization of the linear chromosomes allows the
Figure 6. The Fourier transform infrared spectroscopy (FTIR) results of the selectedspecimens: (a) fine fly ash (fF)�fine rice husk�bark ash (fR)-2, (b) fF�coarse rice husk�barkash (cR)-2, (c) coarse fly ash (cF)�fR-2, and (d) cF�cR specimens.
Prediction of Compressive Strength of Geopolymers 11
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
unconstrained operation of important genetic operators such as mutation,transposition, and recombination (Anderson, 1983; Rumelhart et al., 1986;
Liu et al., 2002). The two main parameters of GEP are the chromosomes
and expression trees (ETs; Anderson, 1983; Rumelhart et al., 1986; Liu
et al., 2002). Two languages are utilized in GEP: the language of the
genes and the language of ETs. A significant advantage of GEP is that it
Figure 7. Genetic programming flowchart (Karlsson et al., 2005).
12 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
enables to infer exactly the phenotype given the sequence of a gene and viceversa, which is termed as Karva language.
For each problem, the type of linking function, as well as the number ofgenes and the length of each gene, are a priori chosen for each problem.While attempting to solve a problem, one can always start by using a single-gene chromosome and then proceed by increasing the length of the head.If it becomes very large, one can increase the number of genes and obvi-ously choose a function to link the sub-ETs. One can start with additionfor algebraic expressions or for Boolean expressions, but in some cases,another linking function might be more appropriate (like multiplicationor if, for instance). The idea, of course, is to find a good solution,and GEP provides the means of finding one very efficiently (Rumelhartet al., 1986).
GEP Structure and Parameters
In this study, as seen in Figures 8 and 9, the ETs of two different GEPapproach models, namely GEP-I and GEP-II, were constructed for com-pressive strength values of geopolymers. The d0, d1, d2, d3, d4, and d5 inFigures 8 and 9 represent the values for input layers, which were the per-centage of fine fly ash in the ashes mixture (fF), the percentage of coarse flyash in the ashes mixture (cF), the percentage of fine rice husk bark ash in theashes mixture (fR), the percentage of coarse rice husk bark ash in the ashesmixture (cR), the temperature of curing (T) and the time of water curing (t),respectively. In the GEP-I and GEP-II, the number of genes used three andfour genes (Sub-ETs), and linking function used addition and multiplica-tion, respectively. In training and testing of the GEP-I and GEP-II approachmodels constituted with two different Sub-ETs and linking function fF, cF,fR, cR, T, and t as input data and compressive strength values of geopoly-mers as independent output data. Among 120 experimental sets, 94 sets wererandomly chosen as a training set for the GEP-I and GEP-II modeling andthe remaining 26 sets were used as testing the generalization capacity of theproposed models.
For this problem, firstly, the fitness, fi, of an individual program, i, ismeasured by:
fi ¼XCt
j¼1ðM� jCðij Þ � Tj jÞ ð1Þ
whereM is the range of selection, C(i,j) is the value returned by the individualchromosome i for fitness case j (out of Ct fitness cases), and Tj is the targetvalue for fitness case j. If jCðij Þ � Tj j (the precision) is less than or equal to0.01, then the precision is equal to zero, and fi¼ fmax¼CtM. In this case,
Prediction of Compressive Strength of Geopolymers 13
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
M¼ 100 was used, therefore, fmax¼ 1000. The advantage of this kind offitness functions is that the system can find the optimal solution by itself(Hopfield, 1982; Cevik and Sonebi, 2009).
Afterwards, the set of terminals T and the set of functions F to create thechromosomes are preferred, namely, T¼ {fF, cF, fR, cR, T, t} and fourbasic arithmetic operators (þ, �, �, and /) and some basic mathematicalfunctions (Sqrt, x3) were used.
Another major step is to choose the chromosomal tree, that is, the lengthof the head and the number of genes. The GEP-I and GEP-II approachmodels initially used single gene and two lengths of heads, and increased the
Figure 8. Expression tree with three genes for predicting compressive strength in geneexpression programming-I (GEP-I) model. C0¼ 11.21 and C1¼�6.47.
14 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Figure 9. Expression tree with four genes for predicting compressive strength in geneexpression programming-II (GEP-II) model. C0¼ 3.54, c1¼ 8.99.
Prediction of Compressive Strength of Geopolymers 15
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
number of genes and heads, one after another during each run, and moni-
tored the training and testing sets performance of each model. In this study,
for the GEP-I and GEP-II approach models observed the number of genes 3
and 4, and length of heads 10 and 12, respectively. In addition, for the GEP-
I and GEP-II approach models determined the linking function multiplica-
tion and addition, respectively.Finally, a combination of all genetic operators (mutation, transposition,
and crossover) was utilized as set of genetic operators. Parameters of the
Table 3. Parameters of GEP approach models.
Parameter definition GEP-I GEP-II
P1 Function set þ, �, �, /, sqrt, x3þ, �, �, /, sqrt, x3
P2 Chromosomes 30 40P3 Head size 12 14P4 Number of genes 3 4P5 Linking function Addition MultiplicationP6 Mutation rate 0.044 0.044P7 Inversion rate 0.1 0.1P8 One-point recombination rate 0.3 0.3P9 Two-point recombination rate 0.3 0.3P10 Gene recombination rate 0.1 0.1P11 Gene transposition rate 0.1 0.1
Note. GEP: gene expression programming.
Figure 10. Chromosome with two genes and its decoding in gene expression programming(GEP; Cevik and Sonebi, 2009).
16 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:46pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
training of the GEP-I and GEP-II approach models are given in Table 3.
For the GEP-I and GEP-II approach models, chromosome 30 and 40
were observed to be the best of generation individuals predicting compres-
sive strength. Explicit formulations based on the GEP-I and GEP-II
approach models for fs were obtained by:
CVN ¼ f fF, cF, fR, cR,T, tð Þ: ð2Þ
The related formulations could be obtained by the procedure shown in
Figure 10 (Cevik and Sonebi, 2009).
y = 0.9057x + 2.5473R² = 0.9624
GEP-I
y = 1.0024x + 0.1424R² = 0.9765
GEP-II
0
10
20
30
40
50
60
70(a)
(b)
Pred
icte
d C
ompr
essi
ve S
tren
gth
(MPa
)
Experimental Compressive Strength (MPa)
GEP-I
GEP-II
y = 0.9057x + 2.5473R² = 0.9624
GEP-I
y = 1.0024x + 0.1424R² = 0.9765
GEP-II
0
10
20
30
40
50
60
70
0 20 40 60 80
0 20 40 60 80
Pred
icte
d C
ompr
essi
ve S
tren
gth
(MPa
)
Experimental Compressive Strength (MPa)
GEP-I
GEP-II
Figure 11. The correlation of the measured and predicted compressive strength values ofgeopolymers in (a) training and (b) testing phase for GEP models.
Prediction of Compressive Strength of Geopolymers 17
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Tab
le4
.D
ata
sets
for
co
mp
ari
son
of
exp
eri
me
nta
lre
sult
sw
ith
resu
lts
pre
dic
ted
fro
mth
eG
EP
mo
de
ls.
Th
ep
erc
en
tag
eo
ffi
ne
fly
ash
inth
ea
sh
es
mix
ture
(fF
)
Th
ep
erc
en
tag
eo
fco
ars
efl
ya
sh
inth
ea
sh
es
mix
ture
(cF
)
Th
ep
erc
en
tag
eo
ffi
ne
rice
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(fR
)
Th
ep
erc
en
tag
eo
fco
ars
eri
ce
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(cR
)
Th
ete
mp
era
ture
of
cu
rin
g(T
)
Th
eti
me
of
wa
ter
cu
rin
g(t
)
Co
mp
ressiv
estr
en
gth
va
lue
so
bta
ine
dfr
om
exp
eri
me
nts
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-Im
od
el
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-II
mo
de
l(M
Pa
)
60
04
00
25
72
7.1
26
.82
5.7
70
03
00
25
73
73
6.1
35
.78
00
20
02
57
33
.43
2.5
32
.86
00
04
02
57
24
.72
4.1
24
.97
00
03
02
57
33
.72
9.4
34
.78
00
02
02
57
30
.53
0.4
30
.90
60
40
02
57
19
.72
2.9
19
.60
70
30
02
57
26
.92
7.4
26
.90
80
20
02
57
24
.32
4.4
20
.70
60
04
02
57
14
.81
5.3
17
.50
70
03
02
57
20
.21
9.8
24
.10
80
02
02
57
18
.21
8.5
18
.96
00
40
04
07
30
.42
9.7
29
.27
00
30
04
07
39
.23
8.7
39
.78
00
20
04
07
35
.43
4.5
35
60
00
40
40
72
7.7
26
.92
6.6
70
00
30
40
73
5.8
30
.23
7.2
80
00
20
40
73
2.3
30
.33
2.5
06
04
00
40
72
2.1
23
.22
0.8
07
03
00
40
72
8.5
28
.22
7.8
08
02
00
40
72
5.8
25
.22
2.9
06
00
40
40
71
6.5
17
17
.50
70
03
04
07
21
.41
9.5
23
.10
80
02
04
07
19
.31
9.3
19
.9
(co
ntin
ue
d)
18 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Ta
ble
4.
Co
ntin
ue
d
Th
ep
erc
en
tag
eo
ffi
ne
fly
ash
inth
ea
sh
es
mix
ture
(fF
)
Th
ep
erc
en
tag
eo
fco
ars
efl
ya
sh
inth
ea
sh
es
mix
ture
(cF
)
Th
ep
erc
en
tag
eo
ffi
ne
rice
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(fR
)
Th
ep
erc
en
tag
eo
fco
ars
eri
ce
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(cR
)
Th
ete
mp
era
ture
of
cu
rin
g(T
)
Th
eti
me
of
wa
ter
cu
rin
g(t
)
Co
mp
ressiv
estr
en
gth
va
lue
so
bta
ine
dfr
om
exp
eri
me
nts
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-Im
od
el
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-II
mo
de
l(M
Pa
)
60
04
00
60
73
1.9
32
.13
1.3
70
03
00
60
74
1.7
41
.34
2.1
80
02
00
60
73
7.2
37
37
.26
00
04
06
07
29
.13
1.3
28
.77
00
03
06
07
38
34
.44
18
00
02
06
07
33
.93
2.9
34
.20
60
40
06
07
23
.22
4.9
23
.40
70
30
06
07
30
.33
0.9
30
08
02
00
60
72
72
7.8
26
06
00
40
60
71
7.4
19
.11
7.3
07
00
30
60
72
2.7
21
.42
30
80
02
06
07
20
.22
0.4
20
.76
00
40
08
07
34
.83
2.8
34
.57
00
30
08
07
44
.24
2.7
43
.38
00
20
08
07
39
.63
8.3
39
.36
00
04
08
07
31
.83
1.9
32
70
00
30
80
74
0.3
35
.24
3.8
80
00
20
80
73
6.1
33
.33
70
60
40
08
07
25
.32
6.4
26
.30
70
30
08
07
32
.13
2.3
32
.40
80
20
08
07
28
.82
7.8
29
.10
60
04
08
07
19
19
.51
7.7
07
00
30
80
72
4.1
22
22
.30
80
02
08
07
21
.52
0.9
20
60
04
00
90
73
2.4
32
.33
2.3
(co
ntin
ue
d)
Prediction of Compressive Strength of Geopolymers 19
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Ta
ble
4.
Co
ntin
ue
d
Th
ep
erc
en
tag
eo
ffi
ne
fly
ash
inth
ea
sh
es
mix
ture
(fF
)
Th
ep
erc
en
tag
eo
fco
ars
efl
ya
sh
inth
ea
sh
es
mix
ture
(cF
)
Th
ep
erc
en
tag
eo
ffi
ne
rice
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(fR
)
Th
ep
erc
en
tag
eo
fco
ars
eri
ce
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(cR
)
Th
ete
mp
era
ture
of
cu
rin
g(T
)
Th
eti
me
of
wa
ter
cu
rin
g(t
)
Co
mp
ressiv
estr
en
gth
va
lue
so
bta
ine
dfr
om
exp
eri
me
nts
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-Im
od
el
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-II
mo
de
l(M
Pa
)
70
03
00
90
74
1.7
42
.14
1.3
80
02
00
90
73
7.7
37
.73
86
00
04
09
07
29
.63
2.1
29
.97
00
03
09
07
38
35
.14
0.2
80
00
20
90
73
4.4
32
.63
5.9
06
04
00
90
72
3.6
25
.52
6.9
07
03
00
90
73
0.3
26
.22
50
80
20
09
07
27
.42
6.8
29
.10
60
04
09
07
17
.72
01
7.7
07
00
30
90
72
2.7
22
.22
1.8
08
00
20
90
72
0.5
21
.51
8.7
60
04
00
25
28
36
.13
5.4
35
70
03
00
25
28
49
.34
7.8
48
.48
00
20
02
52
84
4.5
42
.64
7.7
60
00
40
25
28
32
.53
5.4
32
.57
00
03
02
52
84
4.4
38
.94
3.6
80
00
20
25
28
40
.13
4.8
40
.70
60
40
02
52
82
72
82
7.4
07
03
00
25
28
36
.93
6.2
38
.50
80
20
02
52
83
3.3
33
.43
50
60
04
02
52
82
0.8
23
.72
4.8
07
00
30
25
28
28
.42
6.5
34
.60
80
02
02
52
82
5.6
24
.62
7.7
60
04
00
40
28
40
.53
9.5
39
.57
00
30
04
02
85
2.3
52
.75
2.4
(co
ntin
ue
d)
20 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Ta
ble
4.
Co
ntin
ue
d
Th
ep
erc
en
tag
eo
ffi
ne
fly
ash
inth
ea
sh
es
mix
ture
(fF
)
Th
ep
erc
en
tag
eo
fco
ars
efl
ya
sh
inth
ea
sh
es
mix
ture
(cF
)
Th
ep
erc
en
tag
eo
ffi
ne
rice
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(fR
)
Th
ep
erc
en
tag
eo
fco
ars
eri
ce
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(cR
)
Th
ete
mp
era
ture
of
cu
rin
g(T
)
Th
eti
me
of
wa
ter
cu
rin
g(t
)
Co
mp
ressiv
estr
en
gth
va
lue
so
bta
ine
dfr
om
exp
eri
me
nts
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-Im
od
el
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-II
mo
de
l(M
Pa
)
80
02
00
40
28
47
.24
5.7
47
.46
00
04
04
02
83
6.5
35
.73
4.4
70
00
30
40
28
47
.14
0.7
47
.18
00
02
04
02
84
2.5
37
.74
3.4
06
04
00
40
28
30
.33
0.3
30
.40
70
30
04
02
83
9.1
38
.44
0.5
08
02
00
40
28
35
.33
53
6.4
06
00
40
40
28
23
.32
3.8
24
.30
70
03
04
02
83
0.1
28
.53
2.4
08
00
20
40
28
27
.22
5.6
28
.86
00
40
06
02
84
2.5
42
.64
1.9
70
03
00
60
28
55
.65
5.5
55
.38
00
20
06
02
84
9.6
48
.24
9.8
60
00
40
60
28
38
.33
9.4
36
.37
00
03
06
02
85
04
55
0.3
80
00
20
60
28
44
.64
4.3
45
.60
60
40
06
02
83
1.8
32
.63
4.3
07
03
00
60
28
41
.54
2.2
42
.70
80
20
06
02
83
73
9.6
37
.10
60
04
06
02
82
4.5
25
.12
5.2
07
00
30
60
28
32
30
.93
2.7
08
00
20
60
28
28
.52
9.1
29
.56
00
40
08
02
84
6.4
44
.24
4.7
70
03
00
80
28
58
.95
5.8
59
.3
(co
ntin
ue
d)
Prediction of Compressive Strength of Geopolymers 21
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Ta
ble
4.
Co
ntin
ue
d
Th
ep
erc
en
tag
eo
ffi
ne
fly
ash
inth
ea
sh
es
mix
ture
(fF
)
Th
ep
erc
en
tag
eo
fco
ars
efl
ya
sh
inth
ea
sh
es
mix
ture
(cF
)
Th
ep
erc
en
tag
eo
ffi
ne
rice
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(fR
)
Th
ep
erc
en
tag
eo
fco
ars
eri
ce
hu
sk
�b
ark
ash
inth
ea
sh
es
mix
ture
(cR
)
Th
ete
mp
era
ture
of
cu
rin
g(T
)
Th
eti
me
of
wa
ter
cu
rin
g(t
)
Co
mp
ressiv
estr
en
gth
va
lue
so
bta
ine
dfr
om
exp
eri
me
nts
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-Im
od
el
(MP
a)
Co
mp
ressiv
estr
en
gth
va
lue
sp
red
icte
db
yG
EP
-II
mo
de
l(M
Pa
)
80
02
00
80
28
52
.85
0.9
53
.36
00
04
08
02
84
1.8
41
.74
07
00
03
08
02
85
34
6.3
53
.28
00
02
08
02
84
7.5
43
.94
7.8
06
04
00
80
28
34
.73
4.4
35
.40
70
30
08
02
84
44
3.5
43
.80
80
20
08
02
83
9.4
40
.53
8.3
06
00
40
80
28
26
.72
6.2
25
07
00
30
80
28
33
.93
1.6
32
.10
80
02
08
02
83
0.3
28
.62
86
00
40
09
02
84
3.2
42
.74
3.5
70
03
00
90
28
55
.65
4.6
58
.48
00
20
09
02
85
0.2
50
.35
1.7
60
00
40
90
28
38
.94
1.9
37
.57
00
03
09
02
85
04
5.7
50
.18
00
02
09
02
84
5.2
41
.34
60
60
40
09
02
83
2.3
33
.33
3.7
07
03
00
90
28
41
.54
1.9
42
.10
80
20
09
02
83
7.5
38
.83
6.8
06
00
40
90
28
24
.92
5.8
24
.40
70
03
09
02
83
23
1.2
30
.90
80
02
09
02
82
8.9
27
.92
6.4
No
te.
GE
P:
ge
ne
exp
ress
ion
pro
gra
mm
ing
.
22 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
PREDICTED RESULTS AND DISCUSSION
In this study, the error that arose during the training and testing in GEP-Iand GEP-II models can be expressed as absolute fraction of variance (R2),which is calculated by Equation (3) [Topcu and Sarıdemir, 2008]:
R2 ¼ 1�
Pi ðti � oiÞ
2Pi ðoiÞ
2
!ð3Þ
where t is the target value and o is the output value.All of the results obtained from experimental studies and predicted by
using the training and testing results of GEP-I and GEP-II models are givenin Figures 11(a) and (b), respectively. The linear least square fit line, itsequation, and the R2 values were shown in these figures for the trainingand testing data. Also, inputs values and experimental results with trainingand testing results obtained from GEP-I and GEP-II models are given inTable 4. As it is visible in Figure 11, the values obtained from the trainingand testing in GEP-I and GEP-II models are very close to the experimentalresults. The result of testing phase in Figure 11 shows that the GEP-I andGEP-II models are capable of generalizing between input and output vari-ables with reasonably good predictions.
The performance of the GEP-I and GEP-II models is shown in Figure 11.The best value of R2 is 97.650% for training set in the GEP-II model. Theminimum value of R2 is 95.22% for testing set in the GEP-I model. All of R2
values show that the proposed GEP-I and GEP-II models are suitable andcan predict the compressive strength values very close to the experimentalvalues.
CONCLUSIONS
From the experimental procedure, the following results were obtained:
1. The compressive strength of the specimens depends on the particle size ofthe ashes, time of oven curing, and the time of room condition curing.The finer the ashes particle size results in the denser and hence the stron-ger specimen. On the other hand, oven curing of the specimens at 80�Cwas found to be the optimum temperature of curing in geopolymericspecimens.
2. The GEP can be an alternative approach for the evaluation of the effectof seeded mixture of FA and RHBA on compressive strength values ofgeopolymer specimens.
Prediction of Compressive Strength of Geopolymers 23
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
3. Comparison between GEP models in terms of R2 showed that GEP
models are capable of predicting suitable results for compressive strengthvalues of geopolymer specimens.
REFERENCES
Alvarez-Ayuso, E., Querol, X., Plan, F., Alastuey, A., Moreno, N., Izquierdo, M. et al. (2008).Environmental, Physical and Structural Characterisation of Geopolymer MatrixesSynthesised from Coal (Co-)Combustion Fly Ashes, Journal of Hazardous Materials,154: 175�183.
Anderson, J.A. (1983). Cognitive and Psychological Computation with Neural Models, IEEETransactions on Systems, Man, and Cybernetics, SMC-13: 799�814.
Bakharev, T. (2005). Geopolymeric Materials Prepared Using Class F Fly Ash and ElevatedTemperature Curing, Cement and Concrete Research, 35: 1224�1232.
Cevik, A. (2007). A New Formulation for Longitudinally Stiffened Webs Subjected to PatchLoading, Journal of Constructional Steel Research, 63: 1328�1340.
Cevik, A. and Guzelbey, I.H. (2007). A Soft Computing Based Approach for the Prediction ofUltimate Strength of Metal Plates in Compression, Engineering Structures, 29: 383�394.
Cevik, A. and Sonebi, M. (2008). Modelling the Performance of Self-Compacting SIFCON ofCement Slurries Using Genetic Programming Technique, Computers and Concrete, 5:475�490.
Cevik, A. and Sonebi, M. (2009). Genetic Programming Based Formulation for Fresh andHardened Properties of Self-Compacting Concrete Containing Pulverised Fuel Ash,Construction and Building Materials, 23: 2614�2622.
Chindaprasirt, P., Chareerat, T., Hatanaka, S. and Cao, T. (2011). High-Strength GeopolymerUsing Fine High-Calcium Fly Ash, Journal of Materials in Civil Engineering, 23:2264�2270.
Chindaprasirt, P., Chareerat, T. and Sirivivatnanon, V. (2007). Workability and Strength ofCoarse High Calcium Fly Ash Geopolymer, Cement and Concrete Composites, 29:224�229.
Chindaprasirt, P., Homwuttiwong, S. and Sirivivatnanon, V. (2004). Influence of Fly AshFineness on Strength, Drying Shrinkage and Sulfate Resistance of Blended CementMortar, Cement and Concrete Research, 34: 1087�1092.
Davidovits, J. (1991). Geopolymers: Inorganic Polymeric New Materials, Journal of ThermalAnalysis, 37: 1633�1656.
Fernandez-Jimenez, A., Garcia-Lodeiro, I. and Palomo, A. (2007). Durability of Alkali-Activated Fly Ash Cementitious Materials, Journal of Materials Science, 42: 3055�3065.
Hopfield, J.J. (1982). Neural Networks and Physical Systems with Emergent CollectiveComputational Abilities, Proceedings of the National Academy of Sciences, 79: 2554�2558.
Karlsson, C., Zanghellini, E., Swenson, J., Roling, B., Bowron, D.T. and Boorjesson, L. (2005).Structure of mixed alkali/alkaline-earth silicate glasses from neutron diffraction andvibrational spectroscopy, Physical Review B, 72: 064206.
Liu, S.W., Huang, J.H., Sung, J.C. and Lee, C.C. (2002). Detection of Cracks Using NeuralNetworks and Computational Mechanics, Computer Methods in Applied Mechanics andEngineering, 191: 2831�2845.
Milani, A.A. and Nazari, A. (2011). Modeling Ductile-to-Brittle Transition Temperature ofFunctionally Graded Steels by Gene Expression Programming, International Journal ofDamage Mechanics, DOI: 10.1177/1056789511406561.
24 A. NAZARI ET AL.
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from
XML Template (2011) [9.12.2011–5:47pm] [1–25]K:/IJD/IJD 431991.3d (IJD) [PREPRINTER stage]
Naji Givi, A., Abdul Rashid, S., Nora, A., Aziz, F. and Mohd Salleh, M.A. (2010). Assessmentof the Effects of Rice Husk Ash Particle Size on Strength, Water Permeability andWorkability of Binary Blended Concrete, Construction and Building Materials, 24:2145�2150.
Nazari, A., Khalaj, G. and Didehvar, N. (2011). Computational Investigations of the ImpactResistance of Aluminum-Epoxy Laminated Composites, International Journal of DamageMechanics, DOI: 10.1177/1056789511411739.
Pacheco-Torgal, F., Castro-Gomes, J.P. and Jalali, S. (2005). Studies About Mix Compositionof Alkali-Activated Mortars Using Waste Mud from Panasqueira, Proceedings of theEngineering Conference, University of Beira Interior, Covilha, Portugal.
Pacheco-Torgal, F., Castro-Gomes, J. and Jalali, S. (2007). Investigation About the Effect ofAggregates on Strength and Microstructure of Geopolymeric Mine Waste Mud Binders,Cement and Concrete Research, 37: 933�941.
Rosenblatt, F. (1962). Principles of Neuro Dynamics: Perceptrons and the Theory of BrainMechanisms, Washington, DC, Spartan Books.
Rumelhart, D.E., Hinton, G.E. and William, R.J. (1986). Learning Internal Representation byError Propagation. In: Rumelhart, D.E. and McClelland, J.L. (eds), Proceeding ParallelDistributed Processing Foundation, Vol. 1, Cambridge, MIT Press, pp. 318�362.
Sata, V., Jaturapitakkul, C. and Kiattikomol, K. (2007). Influence of Pozzolan from VariousByproduct Materials on Mechanical Properties of High-Strength Concrete, Constructionand Building Materials, 21: 1589�1598.
Serra, J., Gonzalez, P., Liste, S., Chiussi, S., Leon, B., Perez-Amor, M. et al. (2002). Influenceof the non-bridging oxygen groups on the bioactivity of silicate glasses, Journal ofMaterials Science: Materials in Medicine, 13: 1221.
Tangchirapat, W., Buranasing, R., Jaturapitakkul, C. and Chindaprasirt, P. (2008). Influenceof Rice Husk � Bark Ash on Mechanical Properties of Concrete Containing HighAmount of Recycled Aggregates, Construction and Building Materials, 22: 1812�1819.
Topcu, I.B. and Sarıdemir, M. (2008). Prediction of Compressive Strength of ConcreteContaining Fly Ash Using Artificial Neural Network and Fuzzy Logic, ComputationalMaterials Science, 41: 305�311.
van Jaarsveld, J.G.S., van Deventer, J.S.J. and Lorenzen, L. (1997). The Potential Use ofGeopolymeric Materials to Immobilise Toxic Metals. Part I. Theory and Applications,Minerals Engineering, 10: 659�669.
van Jaarsveld, J.G.S., Van Deventer, J.S.J. and Lukey, G.C. (2002). The Effect of Compositionand Temperature on the Properties of Fly Ash- and Kaolinite-Based Geopolymers,Chemical Engineering Journal, 89: 63�73.
Wongpa, J., Kiattikomol, K., Jaturapitakkul, C. and Chindaprasirt, P. (2010). CompressiveStrength, Modulus of Elasticity, and Water Permeability of Inorganic Polymer Concrete,Materials & Design, 31: 4748�4754.
Zholobenko, V.L., Holmes, S.M., Cundy, C.S. and Dwyer, J. (1997). Synthesis of MCM-41materials: an in situ FTIR study, Microporous Materials, 11: 83.
Zuhua, Z., Xiao, Y., Huajun, Z. and Yue, C. (2009). Role of Water in the Synthesis of CalcinedKaolin-Based Geopolymer, Applied Clay Science, 43: 218�223.
Prediction of Compressive Strength of Geopolymers 25
at UNIV TORONTO on August 12, 2014ijd.sagepub.comDownloaded from