Deformation Properties of Rubberized

Engineered Cementitious Composites Using

Response Surface Methodology

Bashar S. Mohammed, Lee Wei Xian,

Sani Haruna, M. S. Liew, Isyaka
Abdulkadir & Nor Amila Wan Abdullah
Zawawi
Iranian Journal of Science and
Technology, Transactions of Civil
Engineering

ISSN 2228-6160

Iran J Sci Technol Trans Civ Eng

DOI 10.1007/s40996-020-00444-3

1 23
Your article is protected by copyright and
all rights are held exclusively by Shiraz
University. This e-offprint is for personal
use only and shall not be self-archived
in electronic repositories. If you wish to
self-archive your article, please use the
accepted manuscript version for posting on
your own website. You may further deposit
the accepted manuscript version in any
repository, provided it is only made publicly
available 12 months after official publication
or later and provided acknowledgement is
given to the original source of publication
and a link is inserted to the published article
on Springer's website. The link must be
accompanied by the following text: "The final
publication is available at link.springer.com”.

1 23
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering
https://doi.org/10.1007/s40996-020-00444-3

RESEARCH PAPER

Deformation Properties of Rubberized Engineered Cementitious

Composites Using Response Surface Methodology
Bashar S. Mohammed1 · Lee Wei Xian1 · Sani Haruna1,2 · M. S. Liew1 · Isyaka Abdulkadir1,2 ·
Nor Amila Wan Abdullah Zawawi1

Received: 2 July 2019 / Accepted: 18 July 2020

© Shiraz University 2020

Abstract
This study reports the effects of crumb rubber and polyvinyl alcohol fibre (PVA) on the deformation properties of rubberized
engineered cementitious composites (R-ECCs), including drying shrinkage, elastic modulus, and Poisson’s ratio. By utilizing
response surface methodology, two variables have been considered in developing R-ECC mixtures which are the amount of
crumb rubber replacement to fine aggregate by volume 0–5% and PVA fibres from 0 to 2% by volume of cementitious materi-
als. Experimental data show that the incorporation of crumb rubber into ECC results in decreasing its compressive strength
and elastic modulus. A significant increase in Poisson’s ratio and drying shrinkage was reported with the incorporation of
crumb rubber. Design–Expert software has been utilized to construct predictive models for the responses. The goodness
of fit between the measured and predicted values is validated using the coefficient of determination. Results of numerical
optimizations showed that the best mixture was obtained by combining 1.92% of crumb rubber with 1.86% of PVA fibres.
The optimization results of the prediction model were conducted to acquire the optimal solution variables. The variation
obtained between the predicted results and the validation results is less than 5%.

Keywords Engineered cementitious composites (ECCs) · Crumb rubber (CR) · Deformation properties · Response surface
methodology

1 Introduction Engineered cementitious composite (ECC) is a special type

of high-performance fibre-reinforced cementitious compos-
Ordinary cement concrete has been generally utilized as a ites (HPFRCCs) micromechanically designed to attain high
construction material with the benefits of toughness, fire ductility at moderate fibre content (usually not more than
resistance, energy effectiveness, and on-site production. 2%) when subjected to tensile and shear loading (Kong et al.
Conversely, it has the hindrances of low elasticity, low flex- 2003; Mohammed et al. 2019a). Usually, a 2% volume of
ibility, volume instability, and conflicting dependability polymeric fibres is incorporated into the matrix to bridge the
because of variable vibration application abilities hand in finer cracks which developed on loading (Mohammed et al.
hand. To enhance its performance, a few changed materials, 2017). The coarse aggregate in ECC has been avoided due
for example, high-performance concrete (HPC) and high- to its adverse effect on ductile behaviour (Khed et al. 2018).
performance fibre-reinforced concrete (HPFRC), have been High-performance fibre-reinforced cementitious compos-
created (Kong et al. 2003). HPC has higher compressive ites (HPFRCCs) are known for their higher tensile strength
strength and enhanced durability, but it is extremely brittle. and superior resistance to environmental aggressive attacks.
HPFRCC has become an ideal alternative to tackle the issue
of infrastructure deterioration. Unlike typical concrete, ECC
* Sani Haruna
sani_17000823@utp.edu.my strain-hardened by developing some micro-cracks (below
100 µm) after experiencing first cracking exhibits similar
1
Department of Civil and Environmental behaviour like a ductile material and achieves a strain capac-
Engineering, Universiti Teknologi PETRONAS, ity of 3–5%, which is 300 to 500 times more than ordinary
32610 Bandar Seri Iskandar, Perak Darul Ridzuan, Malaysia
concrete (Huang et al. 2013). ECC has currently experienced
2
Department of Civil Engineering, Bayero University Kano, an increasing demand in skyscrapers construction, and it
Kano, Nigeria

13
Vol.:(0123456789)
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

will not be a surprise to spark the fear of natural aggregates toughness of cement-based material. Previous studies also
insufficiency. Despite the promising application of ECC, it suggested that rubber might be the best possible material for
is discouraged due to its high embodied energy consump- the absorption of energy (Atahan and Yücel 2012; Moham-
tion, carbon dioxide emission, and expensive material cost. med et al. 2018). Also, adding rubber particles to concrete
Considering the advantages of rubber particles brought into will soften it, generating better plastic deformation when
ordinary concrete, conducting investigation on the effect of the concrete is being impacted. Rubberized concrete was
adding crumb rubber to ECC seems worthwhile. proven to be capable of improving structures performance
Waste rubber, especially disposed vehicle tyres, domi- in terms of load–deformation response, deformability, dura-
nated the total amount of solid waste over the past few dec- bility, etc. However, concrete’s brittleness will increase as
ades. Collected data from the literature estimated that in the the concrete’s compressive strength goes up (Mohammed
developed regions, one car tyre per person was disposed of and Adamu 2018). Lack of ductility which is prone to crack
every year; therefore, 1 billion scrap tyres will be thrown development has become a major restriction of high-strength
away annually worldwide. The number is expected to grow concrete when subjected to high service load, making it the
due to the forecasted traffic growth, and the yearly aban- main cause of spalling, fragmentation, and cracking under
doned waste tyres are possible to reach 1.2 billion by 2030 loading as well as deterioration problems (Gupta et al.
(Thomas and Gupta 2016). The formidable and still growing 2014). Induced by the intensified infrastructure deteriora-
amount of scrap tyres has become a major environmental tion conditions and failures, researchers and authorities were
concern in solid waste disposal. It is problematic to dispose forced to search for property improvement to concrete.
of waste rubber to landfills because they are difficult to bio- The spalling of concrete under high temperatures is a
degrade and burning them will emit toxic gas. Waste tyre worry due to the exposure of the reinforcement bars to fire,
stockpiles also pose multiple health, environmental, and eco- which leads to loss of strength of the concrete elements
nomic threats through the various sources of pollution (e.g. (Sahmaran et al. 2010; Zhang et al. 2014). Utilization of
air, water, and soil), destroying the landscapes, and encour- ECC in the construction industry makes it important to
aging pest breeding which is the cause of infectious disease totally comprehend the outcomes as it improves the effi-
such as malaria. Moreover, landfilling is only a temporary ciency and durability of structures (Zhang et al. 2014; Han
solution as more and more spaces are required to accommo- et al. 2007). ECC is known for its dense microstructure due
date the increasing accumulation of waste tyres. In France, to significantly low water/cement ratio, high cement and
it is illegal to build new landfills starting from July 2002, binder content, and removal of coarse aggregate in the ECC
and the country soon will be running out of space to collect mixtures. The dense microstructure of ECC limits the path
cast-out tyres (Kannan et al. 2014). Recent rapid infrastruc- for the vapour pressure to escape, hence resulting in spalling
ture development has induced the immense consumption of at high pressure (Wang et al. 2017).
natural aggregates. High usage of building materials cre- Despite the numerous advantages of ECC, like other
ates massive exploitation of natural resources to fulfil the high-strength concretes, it has one major drawback which is
demanding supply of building materials. Some developing its explosive spalling tendency (Bhat et al. 2014). R-ECC is
countries have already experienced a strain in maintaining considered as one of the possible solutions to ECC spalling.
a steady supply of natural sands. This has created a driving At higher stress, the rubber particles experience high pres-
force to explore the approach of recycling these wastes. A sure, and they melt and create channels for vapour to escape,
significant quantity of discarded tyres is used in the field which allows the outward migration of gas and results in the
of civil engineering, for instance, rail and road foundation, reduction in pore pressure and likely to spall. Incorporating
asphaltic concrete as well as embankment. Waste tyres are a high volume of crumb rubber into the ECC mix is reported
also recycled into rubber particles to be incorporated in to have a negative effect on the properties of ECC, includ-
sustainable concrete products, for example, roadside barri- ing a reduction in compressive and tensile strength proper-
ers (Elchalakani 2015), pervious concrete, rigid pavements, ties. Deformation may induce cracking, which increases the
etc. In most of the cases, crumb rubber is added as a partial exposure to environmental attack and subsequently reduces
replacement to coarse aggregate if the size of the rubber the durability of ECC. However, there are limited studies
particle is large enough (Liu et al. 2016). on the effect of crumb rubber on the deformation properties
The idea of incorporating crumb rubber particles in a of the R-ECC. This research aims to investigate the influ-
cement-based material has acquired worldwide acceptance, ences of crumb rubber on the deformation properties of the
making numerous researchers focus on investigating the R-ECC. R-ECC mixtures were produced with varied per-
modified concrete (Onuaguluchi and Panesar 2014; Yilmaz centages replacement of fine aggregates with crumb rubber
and Degirmenci 2009). Rubber particles were found to be (0% to 5%) by volume and PVA fibres (0 to 2%). The role of
elastic and deformable in nature, making them an attractive crumb rubber and PVA fibre was examined based on defor-
solution to the concrete’s brittle nature and the low loading mation properties and response surface methodology.

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

2 Experimental Programme

2.1 Materials

Ordinary Portland cement (OPC), river sand, fly ash (FA),

water, crumb rubber (CR), polyvinyl alcohol fibre (PVA),
and superplasticizer were used as the ingredients to produce
rubberized ECC. Oxides composition of OPC and fly ash
which conforms to the requirements of ASTM C150 and
C168, respectively, is presented in Table 1. The total sum
of ­SiO2, ­Al2O3, and F­ e2O3 is greater than 70%; accordingly,
the fly ash in this study satisfies the requirement of F-class
fly ash as per ASTM 618-10 (2015). Table 2 describes the Fig. 1  Particle size analysis of fine aggregate and crumb rubber
mechanical and geometrical properties of PVA fibres. River
sand with a specific gravity of 2.65, the maximum size of
4.75 mm, and a density of 1656.09 kg/m3 was used as fine 2.2 Test Methods and Sample Preparation
aggregate. Crumb rubber produced from waste tyre rubber
was used for the partial replacement of fine aggregate in Response surface methodology (RSM) is a statistical method
0%, 2.5%, and 5% replacement levels. The average size of used to develop and validate the relationship between more
the crumb rubber that is incorporated in the design mixes than one independent variable and responses (Rezaifar et al.
is ranging from 600 µm to 2.36 mm. The sieve examination 2016; Mohammed et al. 2019b; Haruna et al. 2019). This
was carried out based on ASTM D5644 criteria, and few study is intended to investigate the deformation properties
trials of blends in sizes were carried out until curve grada- affected by two independent variables, crumb rubber of 0%,
tion like that of the fine aggregate was acquired. The particle 2.5%, and 5% and the PVA fibres of 0%, 1%, and 2%. It is
size gradation of fine aggregate and crumb rubber is shown comprised of four steps: (1) design experiments using central
in Fig. 1. Superplasticizer content (%) that will be added to composite design (CCD), (2) collect responses (deformation
the mixture in order to achieve the desired flowability was properties and compressive strength) by conducting experi-
determined in accordance with ASTM C230/C230M-03. ments, (3) develop RSM numerical model, and 4) validate the
developed model and optimize the factors. A face-centred cen-
tral composite response (FCCCD) surface design with α = 1
and second-order polynomial equation for all responses was
Table 1  Chemical composition of OPC and fly ash adopted. FCCCD was conducted to suggest 13 reasonable
Chemical oxide Fly ash (%) OPC (%) trial runs that were tested for their deformation properties and
compressive strength. The 13 variable combinations together
SiO2 57.01 20.76
with the weight proportions of materials are given in Table 3.
Al2O3 20.96 5.54
The statistical software developed by Stat-Ease, Inc., is used
Fe2O3 4.15 3.35
to analyse the experimental design.
CaO 9.79 61.40
The correlation between the independent factors and their
MgO 1.75 2.48
responses was developed by the quadratic equation shown in
Na2O 2.23 0.19
Eq. (1).
K 2O 1.53 0.78
TiO2 0.68 – k
∑ k
∑ j=1
∑
MnO 0.03 – y = 𝛽0 + 𝛽i xi + 𝛽ii xi2 + 𝛽ij xi xj + 𝜖 (1)
LOI 1.25 2.20 i=1 i=1 i=1

Specific gravity 2.38 3.15 where y is the response which are compressive strength,
Blaine fineness ­(m2/kg) 1.092 325 elastic modulus, Poisson’s ratio, and drying shrinkage, xi

Table 2  Mechanical and geometrical properties of PVA fibres

Type Grade of fibre Specific gravity Length of fibre Diameter of fibre Aspect ratio Tensile strength Modulus of elasticity
(mm) (µm) (MPa) (GPa)

PVA REC 5-15 1.3 12 40 462 1600 41

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Table 3  Variable combinations and mix proportion of CR–ECC mixture

Run CR (%) PVA fibre Crumb rubber River sand (kg/m3) Fly ash (kg/m3) Water (kg/m3) OPC (kg/m3) SP (kg/m3)
(%) (kg/m3)

1 5 0 23.35 443.65 700 187 583 19.89

2 5 1 23.35 443.65 700 187 583 20.53
3 2.5 2 11.68 455.32 700 187 583 20.53
4 5 2 23.35 443.65 700 187 583 19.89
5 2.5 1 11.68 455.32 700 187 583 19.89
6 0 2 0 467 700 187 583 20.53
7 0 0 0 467 700 187 583 16.68
8 0 1 0 467 700 187 583 16.68
9 2.5 1 11.68 455.32 700 187 583 19.89
10 2.5 0 11.68 455.32 700 187 583 19.25
11 2.5 1 11.68 455.32 700 187 583 19.89
12 2.5 1 11.68 455.32 700 187 583 19.89
13 2.5 1 11.68 455.32 700 187 583 19.89

and xj are the coded values for the factors crumb rubber 3 Results and Discussion
and PVA fibres, i and j are the linear and quadratic coef-
ficient, respectively, β is the coefficient of regression, k is 3.1 Compressive Strength
the number of variables, and random error is represented by
ε (Mohammed et al. 2017). The desired flowability of rub- The compressive strength of the ECC was determined at
berized ECC was determined using flow table test in accord- the age of 7 and 28 days after water curing. The compres-
ance with ASTM C230/C230M-03 (2014). Trial CR–ECC sive strength profile is shown in Fig. 2 for the two param-
mixes with different percentages of superplasticizer were eters considered for all the 13 mixtures. As anticipated, the
tested until the specified flow is acquired. Superplasticizer compressive strength of all the mixtures increases with the
content is the weight percentage of binder material. Water age of curing. The compressive strength of the ECC speci-
was mixed with superplasticizer in accordance with ASTM mens was in the range of 32.3–65.6 MPa and 53.7–80.6 MPa
C1602. for 7 and 28 days, respectively. The compressive strength
T h e c o m p r e s s i ve t e s t wa s c o n d u c t e d o n decreases with an increase in crumb rubber. The strength
50 mm × 50 mm × 50 mm ECC samples according to BS 1881: reduction of rubberized ECC can be partly caused by the
Part 116:1983 at the age of 3, 7, and 28 days of curing. Three increase in porosity due to the hydrophobic nature of crumb
samples per each mixture were tested using a compression rubber, which repels the water and leads to air entrapment
testing machine conforming to the requirements of BS1610. in the ECC microstructure (Khed et al. 2018). Increasing the
To investigate drying shrinkage, prisms with dimensions of porosity of ECC with crumb rubber percentage replacement
75 mm × 75 mm × 300 mm were used. The drying shrinkage weakens the matrix. The most probable cause of this trend
was determined in accordance with ASTM C 596-01 (2009). of crack propagation is the poor interfacial bond between
Three specimens for each mixture were tested, and the aver-
age was taken. The elastic modulus and Poisson’s ratio of
CR–ECC were determined using cylindrical samples of
300 mm by 150 mm as per the requirement of ASTM C469-
02 (2002). The test was conducted to ascertain the stiffness of
the samples after a curing age of 28 days. Three specimens
for each mixture were tested using a compression machine of
2000 kN capacity. A loading rate of 241 ± 34 kPa/s was used.

Fig. 2  Compressive strength profile for ECC

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

crumb rubber and cement paste, which results in the forma- of crumb rubber. The 2D contour diagram in Fig. 3 indicates
tion of a small amount of cement hydration on the surface of the interaction between the independent variables. The con-
the crumb rubber aggregates (Huang et al. 2013). The weak tour plot shows that there is a fair interaction between the
bond of the interface between crumb rubber particles and the crumb rubber and PVA fibres.
surface of cement paste allows a crack to be formed around
the rubber particles easily. Observation of the ECC’s fracture 3.2 Modulus of Elasticity Test
surface after the compression test showed that cracks were
seen to be propagating along with the crumb rubber parti- The deformation properties of the developed ECC were
cles. This is consistent with the previous studies which stated measured with the aid of the elasticity modulus test. The
that cracks pass at the crumb rubber/cement paste interface test was carried out in accordance with the guidelines of
(Huang et al. 2013). These factors may initiate cracking and ASTM C469. The modulus of elasticity (MOE) of the ECC
encourage propagation when applying low loads, resulting has a direct correlation with its compressive strength. In this
in compressive strength reduction. ECC with crumb rub- study, the elastic modulus of the ECC was obtained from
ber experienced an increase in compressive strength with an 17.5 MPa to 25.6 GPa. As shown in Fig. 4, crumb rubber has
increase in PVA content. This could be due to the potential a negative effect on the MOE, contrary to PVA fibre which
of randomly distributed fibres to strengthen the matrix and significantly contributed towards its enhancement. It is worth
control cracking propagation, thereby enhancing the com- mentioning that the addition of 1% of PVA fibre improved
pressive strength (Caggiano et al. 2016). The model relation- the MOE of the ECC by 7.5% and subsequent addition to
ship for the compressive strength has been established using
the response surface analysis as shown in Eq. (2).
Compressive strength = +53.537 + 0.851CR
+ 13.824PVA − 2.325CR ∗ PVA.
(2)
The compressive strength can be determined using Eq. (2)
for the parameters considered within the defined range. Fig-
ure 3 shows the 3D and 2D surface plots which describe
the effect of crumb rubber as partial replacement to fine
aggregate and PVA content on the compressive strength. It
is interesting to note that the compressive strength decreased
with the increase of crumb rubber and increased with PVA
fibres increment. This might be attributed to the multiple
pores developed, mechanical flexibility, and non-polar nature Fig. 4  Modulus of elasticity of ECC

Fig. 3  Response surface analysis for compressive strength

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

2% increased the MOE by almost 13%. Increasing crumb 3.3 Poisson’s Ratio

rubber replacement level decreases the volumetric propor-
tions of fine aggregates which contributed to the reduction The Poisson’s ratio of all the ECC mixtures is shown in
of MOE. This can partly result in a lower elastic modulus Fig. 6. As presented in Fig. 6, it is noted that the Poisson’s
of crumb rubber compared to fine aggregate. Stiffness has a ratio of ECC increases with the increase of crumb rubber,
significant impact on strength reduction, which results in a while PVA fibres caused opposite effects towards the ratio
decreasing tendency of elastic modulus. However, an incre- of transversal expansion to axial compression. Figure 7
ment in porosity with expanding crumb rubber volume may illustrates the effect of crumb rubber and PVA fibres incor-
have a progressively predominant role in the decrease of poration towards the Poisson’s ratio of ECC. Crumb rub-
MOE (Khed et al. 2018). Hence, rubberized ECC is less stiff ber possesses a much lower elastic modulus compared to
with lower elastic modulus. As shown in Fig. 4, the PVA fine aggregate, and when subjected to compressive stress,
fibre does not affect the elastic modulus of ECC significantly its deformation resistance is considerably lesser, causing
as compared to crumb rubber because low fibre content is ECC to experience a larger axial compression. Similarly,
incorporated (0–2%). However, the elastic modulus of ECC the Poisson’s ratio of the conventional ECC is lower than
under compression is expected to decrease as PVA fibres that of the rubberized ECC. Thus, significantly large rela-
content increases. The model relationship for the modulus tive deformations between rubber and ECC can occur. PVA
of elasticity has been established using the response surface fibres, however, are capable of controlling sliding as well
analysis as shown in Eq. (3). as extending the micro-cracks, which provides restraint to
Modulus of elasticity = +25.98531 − 1.79823CR
− 4.68393PVA − 0.068350CR
∗ PVA + 0.18927CR2 + 1.78165PVA2 .
(3)
As presented in Fig. 5, a three-dimensional response sur-
face diagram and 2D contour plot represent the interaction
effect of crumb rubber replacement and PVA content on the
elastic modulus of rubberized ECC. The figures indicate that
the increase in crumb rubber replacement level and PVA
content reduces the ECC’s elastic modulus. The elliptical
shape of the contour plot in Fig. 5b indicates that there is an
excellent interaction between crumb rubber and PVA fibres.
Fig. 6  Poisson’s ratio of the ECC mixtures

Fig. 5  Response surface analysis for modulus of elasticity

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Fig. 7  Response surface analysis for Poisson’s ratio

lateral expansion and improves Poisson’s ratio compared

with composites without fibres (Haruna et al. 2019). The
elliptical shape in Fig. 6 indicates a very good interaction
between the crumb rubber and PVA fibres. The model equa-
tion for the Poisson’s has been established using the response
surface analysis as shown in Eq. (4).
Poisson’s ratio = + 0.28626 + 0.027185CR − 0.11500PVA
+ 1.85700E − 003CR ∗ PVA − 5.08779E
− 003CR2 + 0.037232PVA2 .
(4)

Fig. 8  Drying shrinkage of the ECC mixtures

3.4 Drying Shrinkage

The portion of fibre volume is likewise one of the main con- shrinkage of ECC will increase slightly when PVA fibres
siderations, which assumes a role in opposing the shrinkage content increases from 0 to 2%. One possible explanation
in high-performance concrete (Sun et al. 2001). Figure 8 of this is that the hydrophilic PVA fibres allow the fibres to
shows the empirical relationship between crumb rubber absorb water during the fresh state of ECC and release the
replacement level and PVA content (factors) and the dry- absorbed water as ECC was hardened after water in capillary
ing shrinkage (response) of ECC at 28 days. As shown in pores has been lost (Sun et al. 2001; Noushini et al. 2014).
Fig. 8, drying shrinkage of ECCs increased with increas- It is worth to mention that crumb rubber creates a more
ing crumb rubber content. The increase in drying shrinkage significant impact on drying shrinkage of ECC. The drying
with increasing crumb rubber content can be a result of a shrinkage of the model can be predicted using Eq. (5), and
reduction in the amount of rigid river sand, which provides the parameters considered are CR and PVA. As shown in
internal restraints to deformation due to drying shrinkage Fig. 9, the three-dimensional and two-dimensional surface
(Zhang et al. 2015). Another factor that attributes to the diagram and contour plot describe the behaviour of dry-
reduction of shrinkage restraint is that the modulus of crumb ing shrinkage of ECC composite with respect to the crumb
rubber is much lower than the fine aggregate; thus, rubber- rubber and PVA fibres. The model relationship for drying
ized ECC exhibits higher drying shrinkage. In addition, the shrinkage has been established using the response surface
increase in porosity which is contributed by its hydrophobic analysis as shown in Eq. (5).
nature may also lead to higher drying shrinkage. Drying

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Fig. 9  Response surface analysis for drying shrinkage

Drying shrinkage = +1130.91992 + 58.07762CR PVA and the responses: (a) elastic modulus, (b) Poisson’s
− 22.27359PVA − 3.32408CR ∗ PVA ratio, (c) drying shrinkage, and (d) compressive strength
were constructed and are presented in Eqs. (1)–(4). Table 5
− 0.44366CR2 + 27.16052PVA2 . displays the coefficient of determination for each RSM
(5)
model. It suggests that there is a good correlation between
the measured and predicted responses. R2 values measure
the goodness of fit for the models. It indicates the variation
4 Analysis of Variance in the response models with respect to input variables. The
models are accounted for 94%, 92%, 95%, and 98% of the
Analysis of variance (ANOVA) with a 5% level of signifi- variation for compressive strength, elastic modulus, Pois-
cance (P < 0.05) is introduced to quantify the significance son’s ratio, and drying shrinkage, respectively. Merely 6%,
of the second-order polynomial function. Table 4 shows a 8%, 5%, and 2% (compressive strength, elastic modulus,
summary of ANOVA. It was observed that the P values of Poisson’s ratio, and drying shrinkage, respectively) of the
all the RSM models were below 0.05, indicating the models variations failed to be accounted for by the models.
are significant at a 95% confidence level (CL). This sug- Moreover, the difference between adjusted R2 and pre-
gests that the models are capable of providing outstanding dicted R2 for each model is less than 0.2 (Montgomery
and accurate responses. There is only 0.03–1.43% chance 2001), implying that the values of the adjusted R2 and pre-
that F-value this size can occur due to noise. Likewise, dicted R2 were in good agreement. The coefficient of vari-
each term in the RSM model was validated at a 5% sig- ation (C.V.) for all models which were used to measure the
nificance level to ensure the statistical significance of the variability of the laboratory results to the overall mean is
terms. Equations (2)–(5) and data presented in Table 4 indi- low. Hence, the results matched the overall mean. Moreover,
cate that apart from the equation, all the RSM models are the adequate precision for every model exceeds 4. Therefore,
quadratic functions. For compressive strength, significant design space defined by central composite design (CCD) can
model terms include PVA, CR*PVA, and ­PVA2, while CR be navigated by the predicted models. The RSM models can
and ­CR2 were insignificant. ANOVA summary for the elastic also be validated through normality plots, which are shown
modulus model shows that all model terms were significant in Fig. 10. The normality plot is a graphical method used to
except for CR*PVA and ­CR2. CR, PVA, and ­PVA2 for dry- evaluate whether the data are normally distributed or not. All
ing shrinkage were significant at the stipulated level of 5%. the plots show that the points fall closely on a straight line,
However, CR*PVA and C ­ R2 were not significant. ANOVA thus implying that the data set is normally distributed, and
for Poisson’s ratio has significant model terms of CR, ­PVA2, the degree of randomness is the same for all fitted values.
and ­CR2.
In addition, all models were insignificant to lack of fit,
implying that the experimental results fit into the model
accurately. The correlations between the factors: a) CR, b)

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Table 4  ANOVA summary

Responses Source Sum of squares df Mean square F value p value Significance

Compressive strength (MPa) Model 601.74 3 200.58 18.51 0.0003 Yes

A-CR 81.48 1 81.48 7.52 0.00228 Yes
B-PVA 385.12 1 385.12 35.53 0.0002 Yes
AB 135.14 1 135.14 12.47 0.0064 Yes
Residual 97.54 9 10.84
Lack of fit 41.04 5 8.21 0.581 0.7189 No
Modulus of elasticity (GPa) Model 59.95 5 11.99 6.54 0.0143 Yes
A-CR 34.26 1 34.26 18.68 0.0035 Yes
B-PVA 10.80 1 10.80 5.89 0.0457 Yes
AB 0.12 1 0.12 0.064 0.8080 No
A2 6.76 1 6.76 3.69 0.0963 No
B2 15.34 1 15.34 8.36 0.0233 Yes
Residual 12.84 7 1.83
Lack of fit 1.01 3 0.34 0.11 0.9472 No
Poisson’s ratio Model 0.016 5 3.157E−003 32.10 0.0001 Yes
A-CR 5.254E−004 1 5.254E−004 5.34 0.0541 No
B-PVA 8.339E−003 1 8.339E−003 84.78 0.0001 Yes
AB 8.621E−005 1 8.621E−005 0.88 0.3803 No
A2 4.886E−003 1 4.886E−003 49.67 0.0002 Yes
B2 6.698E−003 1 6.698E−003 68.10 0.0001 Yes
Residual 6.886E−004 7 9.836E−005
Lack of fit 5.335E−004 3 1.778E−004 4.59 0.0876 No
Drying shrinkage Model 1.279E+005 5 25,570.80 67.12 0.0001 Yes
A-CR 1.117E+005 1 1.117E+005 293.13 0.0001 Yes
B-PVA 3647.96 1 3647.96 9.58 0.0175 Yes
AB 276.24 1 276.24 0.73 0.4227 No
A2 37.15 1 37.15 0.098 0.7639 No
B2 3564.67 1 3564.67 9.36 0.0184 Yes
Residual 2666.88 7 380.98
Lack of fit 2217.90 3 739.30 6.59 0.0501 No

Table 5  Model validation terms Response Compressive Elastic modulus Poisson’s ratio Drying
strength (MPa) (GPa) shrinkage
(με)

SD 3.29 1.35 9.918E−003 19.52

Mean 63.68 21.08 0.25 1287.99
C.V.% 5.17 6.42 3.95 1.52
R2 0.8605 0.9236 0.9582 0.9796
Adjusted R2 0.814 0.9076 0.9284 0.9650
Predicted R2 0.6822 0.7581 0.7377 0.8575
Adequate precision 15.140 7.905 18.400 24.071

4.1 Experimental Validation in this investigation has been utilized. The general response
surface test plan used to determine the ideal answer for the
For examining the factors of the multi-objective optimiza- two factors (crumb rubber = 0 to 5%; PVA = 0 to 2%) joined
tion concurrent strategy, including response surface meth- is the central composite design procedure. If the ideal incen-
odology as the root for determining the best arrangements tive for every response is limited in various areas, at that

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Normal Plot of Residuals Normal Plot of Residuals

99 99

95 95
90 90
Normal % Probability

Normal % Probability
80 80
70 70

50 50

30 30
20 20
10 10
5 5

1 1

-2.00 -1.00 0.00 1.00 2.00

-4.00 -2.00 0.00 2.00 4.00 6.00

Internally Studentized Residuals

Externally Studentized Residuals

Normal Plot of Residuals

Normal Plot of Residuals
99

99
95
90

Normal % Probability
95
90 80
Normal % Probability

70
80
70 50
50
30
30 20
20 10
10 5
5
1
1

-3.00 -2.00 -1.00 0.00 1.00 2.00

-4.00 -2.00 0.00 2.00 4.00

Externally Studentized Residuals

Externally Studentized Residuals

Fig. 10  Normal probability plot of all the response models

point, it will be increasingly hard to discover those crite- 1. The compressive strength of ECC decreases as the level
ria that meet every one of the responses. The trouble-level of crumb rubber incorporation increases. Reduction
increments due to these ideal fields become increasingly in strength is caused by the hydrophobic properties of
particular from one another and will not converge. Various crumb rubber, which promotes air entrapment on the
possible solutions with desirability equal to 1.0 are acquired surface of the crumb rubber and repel the water during
from the optimized results given in Table 6. Experimental the mixing process.
works were performed for the optimized mixtures, and the 2. The elastic modulus of CR–ECC decreases with the
variations of results were obtained to be less than 5%. The increase of crumb rubber in the mixture. This might be
scale of the desirability ranges from 0 to 1.0. Zero desir- due to the incorporation of flexible tyre rubber, which
ability means a completely undesirable response, while the has a lower elastic modulus compared to fine aggregate.
fully desired response has desirability of 1.0. 3. The Poisson’s ratio of rubberized ECC increases with
the increase of crumb rubber, while PVA fibres caused
opposite effects towards the ratio of transversal expan-
5 Conclusions sion to axial compression. Thus, significant large rela-
tive deformations between rubberized ECC and conven-
Deformation properties of ECC incorporating crumb rub- tional ECC occurred, which degrades the Poisson’s ratio.
ber were investigated for 13 different variable combinations 4. The drying shrinkage of CR–ECC increases with the
developed by response surface methodology (RSM). Based amount of crumb rubber in the mixture. This is due to
on the findings of this research, the following conclusions the high porosity in CR–ECC induced by the hydrophilic
were drawn: nature of crumb rubber.

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Table 6  Experimental validation

Sample Validation Crumb rub- PVA fibres (%) Compressive Modulus of Poisson’s Drying shrinkage
ber (%) strength (MPa) elasticity (GPa) ratio (με)

Rubberized ECC Optimization 2.5 1 63.68 24.04 0.313 1202.43

Experimental 65.2 24.63 0.320 1209.36
Difference (%) 2.52 2.45 2.24 0.5
Optimization 1.920 1.862 69.99 20.44 0.241 1281.64
Experimental 72.02 21.03 0.232 1292.03
Difference (%) 2.82 2.81 3.73 0.80
Optimization 4.481 1.650 63.94 18.34 0.231 1394.88
Experimental 65.03 18.20 0.225 1403.62
Difference (%) 1.68 0.76 2.60 0.62

5. RSM is proven to be a reliable technique to develop Han B-C, Kwon Y-J, Kim J-H (2007) Behavior of fire resistance engi-
an optimum design mix for CR–ECC with deforma- neered cementitious composites (FR-ECC) under fire temperature.
J Korea Concr Inst 19(2):189–197
tion properties as responses. The developed quadratic Haruna S, Mohammed BS, Wahab MM, Haruna A (2019) Compressive
equations can be used to predict the responses. The strength and workability of high calcium one-part alkali activated
interaction between two independent variables and the mortars using response surface methodology. In: 2nd international
responses can be observed through the 3D response sur- conference on civil and environmental engineering 2019. 20th–
21st November
face. Contour diagrams can also be used to predict the Huang X, Ranade R, Ni W, Li VC (2013) On the use of recycled tire
results effectively given a set of variable combinations. rubber to develop low E-modulus ECC for durable concrete
All models for deformation properties and compressive repairs. Constr Build Mater 46:134–141
strength are proven to be statistically significant. Kannan D, Diabat A, Shankar KM (2014) Analyzing the drivers of
end-of-life tire management using interpretive structural mod-
eling (ISM). Int J Adv Manuf Technol 72(9–12):1603–1614
Khed VC, Mohammed BS, Liew MS, Alaloul WS, Adamu M (2018)
Hybrid fibre rubberized ECC optimization for modulus of elas-
ticity. IJCIET 9(7):976–1928
Kong HJ, Bike SG, Li VC (2003) Development of a self-consolidat-
References ing engineered cementitious composite employing electrosteric
dispersion/stabilization. Cement Concr Compos 25(3):301–309
ASTM C 469 (2002) Standard test method for static modulus of elastic- Liu H, Wang X, Jiao Y, Sha T (2016) Experimental investigation
ity and Poisson’s ratio of concrete in compression. Annual Book of the mechanical and durability properties of crumb rubber
of ASTM Standards, vol 4 concrete. Materials 9(3):172
Atahan AO, Yücel AÖ (2012) Crumb rubber in concrete: static and Mohammed BS, Adamu M (2018) Mechanical performance of roller
dynamic evaluation. Constr Build Mater 36:617–622 compacted concrete pavement containing crumb rubber and
Bhat PS, Chang V, Li M (2014) Effect of elevated temperature on nano silica. Constr Build Mater 159:234–251
strain-hardening engineered cementitious composites. Constr Mohammed BS, Achara BE, Nuruddin MF, Yaw M, Zulkefli MZ
Build Mater 69:370–380 (2017) Properties of nano-silica-modified self-compacting engi-
C230 A (2014) Standard specification for flow table for use in tests of neered cementitious composites. J Clean Prod 162:1225–1238
hydraulic cement. ASTM. West Conshohocken, PA Mohammed BS, Khed VC, Nuruddin MF (2018) Rubbercrete mix-
C596-01 A (2009) Standard test method for drying shrinkage of mortar ture optimization using response surface methodology. J Clean
containing hydraulic cement. Annual Book of ASTM Standards, Prod 171:1605–1621
vol 4, no 1, pp 1–3 Mohammed BS, Achara BE, Liew MS, Alaloul WS, Khed VC
C618 A (2015) Standard specification for coal fly ash and raw or cal- (2019a) Effects of elevated temperature on the tensile proper-
cined natural pozzolan for use in concrete, vol 15. ASTM Inter- ties of NS-modified self-consolidating engineered cementitious
national, West Conshohocken, PA, United States composites and property optimization using response surface
Caggiano A, Gambarelli S, Martinelli E, Nisticò N, Pepe M (2016) methodology (RSM). Constr Build Mater 206:449–469
Experimental characterization of the post-cracking response in Mohammed BS, Haruna S, Liew MS (2019b) Optimization and char-
hybrid steel/polypropylene fiber-reinforced concrete. Constr Build acterization of cast in situ alkali-activated pastes by response
Mater 125:1035–1043 surface methodology. Constr Build Mater 225:776–787
Elchalakani M (2015) High strength rubberized concrete containing Montgomery DC (2001) Design and analysis of experiments, vol 64.
silica fume for the construction of sustainable road side barriers. Wiley, New York, p 65
In Structures. Elsevier Noushini A, Vessalas K, Arabian G, Samali B (2014) Drying shrink-
Gupta T, Chaudhary S, Sharma RK (2014) Assessment of mechanical age behaviour of fibre reinforced concrete incorporating polyvi-
and durability properties of concrete containing waste rubber tire nyl alcohol fibres and fly ash. Adv Civ Eng 2014:836173
as fine aggregate. Constr Build Mater 73:562–574

13
 Author's personal copy
Iranian Journal of Science and Technology, Transactions of Civil Engineering

Onuaguluchi O, Panesar DK (2014) Hardened properties of concrete Wang X, Xia J, Nanayakkara O, Li Y (2017) Properties of high-per-
mixtures containing pre-coated crumb rubber and silica fume. J formance cementitious composites containing recycled rubber
Clean Prod 82:125–131 crumb. Constr Build Mater 156:1127–1136
Rezaifar O, Hasanzadeh M, Gholhaki M (2016) Concrete made with Yilmaz A, Degirmenci N (2009) Possibility of using waste tire rubber
hybrid blends of crumb rubber and metakaolin: optimization and fly ash with Portland cement as construction materials. Waste
using response surface method. Constr Build Mater 123:59–68 Manag 29(5):1541–1546
Sahmaran M, Lachemi M, Li VC (2010) Assessing mechanical prop- Zhang Q, Ranade R, Li VC (2014) Feasibility study on fire-resistive
erties and microstructure of fire-damaged engineered cementi- engineered cementitious composites. ACI Mater J 111(6):651
tious composites. ACI Mater J 107(3):297 Zhang Z, Ma H, Qian S (2015) Investigation on properties of ECC
Sun W et al (2001) The effect of hybrid fibers and expansive agent on incorporating crumb rubber of different sizes. J Adv Concr Tech-
the shrinkage and permeability of high-performance concrete. nol 13(5):241–251
Cem Concr Res 31(4):595–601
Thomas BS, Gupta RC (2016) Properties of high strength concrete
containing scrap tire rubber. J Clean Prod 113:86–92

13