Materials and Design 64 (2014) 227–236

Contents lists available at ScienceDirect

Materials and Design

journal homepage: www.elsevier .com/locate /matdes

Mechanical properties of roller compacted concrete containing rice huskash with original and recycled asphalt pavement material

http://dx.doi.org/10.1016/j.matdes.2014.07.0720261-3069/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author. Tel.: +98 9111163215.E-mail addresses: a.modarres@nit.ac.ir, amirmodarres2003@yahoo.com

(A. Modarres).

Amir Modarres ⇑, Zeinab HosseiniDepartment of Civil Engineering, Babol Noshirvani University of Technology, Babol, Iran

a r t i c l e i n f o

Article history:Received 15 June 2014Accepted 31 July 2014Available online 9 August 2014

Keywords:Roller compacted concreteRice husk ashReclaimed asphalt pavementModulus of ruptureEnergy absorbencyFatigue life

a b s t r a c t

This study focused on the effects of rice husk ash (RHA) on the mechanical properties of roller compactedconcrete (RCC) designed with original and reclaimed asphalt pavement (RAP) materials. The RCC mixeswere produced by partial substitution of cement with RHA at varying amounts of 3% and 5%. Four aggre-gate combinations including the mix with original aggregate, coarse RAP + fine original aggregate, coarseoriginal aggregate + fine RAP and total RAP were considered. The main experimental design consisted ofthe compressive strength and three points bending tests. Bending test was used to measure the modulusof rupture, material’s energy absorbency and analyse the fatigue response of RCC mixes. All tests wereperformed after 7, 28 and 120 days curing except the fatigue test that performed on 120 days specimens.Adding RHA resulted in higher optimum moisture content (OMC) and lower maximum dry density. Fur-thermore, adding RAP with different dimensions reduced the OMC and maximum dry density. The mate-rial’s flexibility improved upon replacing 3% cement by RHA. However, the energy absorbency reduced byincreasing the RHA content to 5%. The fatigue life of RCC mixes containing RAP material was lower thanthe conventional one. Furthermore, replacing the coarse aggregate by RAP led to higher fatigue life thanthe fine aggregate. There was a strong relationship (R2 > 0.90) between the energy absorbency and fatigueresponse of RCC mixes. At higher stress ratios of 0.72, the mix with higher energy absorbency behavedbetter under repeated loadings. Besides, a reverse relationship was found between the fatigue life andmaterial porosity. Adding 3% RHA reduced the porosity especially after 120 days curing and improvedthe fatigue resistance. However, the addition of RHA to 5% resulted in higher porosities and lower fatiguelives.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

The construction and maintenance of roads pavement should belong-lasting due to their major impact on the economy of coun-tries. Regarding the environmental perspective pavement engi-neering has selected novel techniques that could result in achange that is friendlier to the environment. Some of these envi-ronmental friendly techniques include the use of recycled wastematerials, pavement recycling and warm mix techniques thatcould reduce the pollutant gases emissions. In this regard the useof recycled materials in Portland cement concrete (PCC) pavementshas become more and more popular in recent years. Roller com-pacted concrete (RCC) is a type of concrete which developed fromstandard PCC coupled with experience gained by using compacted

mixtures of soil and gravel materials stabilized with Portlandcement.

RCC is a zero-slump concrete which composed of dense-gradedaggregate, sand, Portland cement and water which is usuallyplaced with an asphalt paver and compacted with conventionalvibratory roller compactors. Based on economical analysis the ini-tial construction costs of RCC pavements are about 30% lower thanthe conventional asphalt pavements and about 10–20% lower thanthat of the conventional PCC. However, the hardened RCC pave-ment behaves like a conventional PCC [1–3].

According to the literature, RCC was firstly used in a timbermanufacture plant site in Vancouver during the initials of 1970.The performance of RCC in this site which was under heavy loadingtraffic and severe abrasive effects was reported to be successful.Since then, RCC pavements have been extensively used in theindustrial pavement areas in Canada [3].

In Europe, RCC was initially used in low traffic roads of Spain.Also since 1984 many parking lots and heavy duty military campswere paved by RCC in Texas State of America [3]. After the oil crisis

228 A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236

during the 1970 decade, due to higher construction costs manyconventional asphalt pavements were widely replaced by RCCpavements [3]. In comparison to flexible pavement a reduced con-struction costs of 30% has been reported in the literature [3].

The main parameters affecting the properties of PCC such aswater to cement ratio (w/c) and density are also considered asthe major criteria in designing RCCs. However the compactionproperties of RCCs have been recognized as the key parametersin achieving the proper load bearing capacity [3–6]. Followingthe idea of the maximum density, most of RCC design instructionshave recommended to compact this layer at optimum moisturecontent [3,4].

RCC could be used as the final surface of pavement especially inheavy duty low speed pavements such as terminals, and parkingareas. However at higher traffic speeds usually a hot mix asphaltoverlay is constructed to promote the skid resistance of pavement[1]. Because of lower water to cement ratio, RCC has lower bleed-ing potential than the traditional PCC pavements. Therefore theproblem of the constitution of weak layer over the pavement sur-face is not the case in RCC, except for inordinate compacted layers.As a result, the higher surface quality results in lower permeabilityand higher durability of RCC pavements [1,3].

In recent years due to high construction costs and many environ-mental aspects, various recycled materials were used in PCC andRCC mixtures [7–9]. Recycled or waste materials have been utilizedto replace the coarse and fine aggregates or used as cementitiousadditives. For example crushed PCC, recycled asphalt pavement(RAP) and crushed waste glasses are some of the recycled wastematerials that have been used as coarse or fine aggregates in con-crete pavements. Moreover, different natural and artificial pozzo-lans have been utilized as cementitious material especially forPortland cement substitution in PCC and RCC pavements [10–13].

During a research study crushed concrete slabs were used asaggregate materials in RCC pavement [12]. Based on the obtainedresults the technical properties of RCC containing crushed concreteslabs were comparable to conventional RCC with original highquality aggregates. However, the compressive strength of RCC withoriginal aggregate was higher than that of designed with recycledconcretes [12].

In a separate study, steel furnace slag was used as aggregatematerial in RCC mixture [14]. It was specified that replacing 25%of natural aggregate by steel slag will even improve the strengthproperties of mixture. In contrast, any further increase had adverseeffects on the mechanical properties of RCC [14].

During a laboratory research study the effect of RAP materialswas investigated as aggregate replacement in PCC mixture. Labora-tory fabricated RAP materials were used in coarse and fine aggre-gate fractions. Results showed a systematic reduction in bothcompressive and indirect tensile strengths for the RCC mix madewith RAP material. In addition it was shown that the RAP contain-ing concrete had a much higher toughness than the conventionalPCC [7].

A similar study was conducted at the university of Florida, inwhich RAP materials were added to PCC mixture [10]. RAP materi-als with different percentages of 0%, 20%, 40%, 70% and 100% wereadded to mixture. Results of laboratory testing indicated that com-pressive strength, modulus of elasticity, flexural strength and indi-rect tensile strength decreased by increasing the percentage of theRAP materials. However, the reduction in flexural strength waslower than that of the compressive and indirect tensile strengths.Furthermore, it was found that the addition of RAP materialincreased the coefficient of thermal expansion and drying shrink-age of PCC mixture [10]. By comparing the stress–strain curves ofvarious studied mixes it was realized that adding the RAP contentwill increase the material’s flexibility and the failure strain of PCCmix [10].

The mechanical properties of rubberized PCC were also reportedin the literature. Similar to RAP, rubber reduced the strength andincreased the flexibility and energy absorbency of PCC mix[15,16]. However, it was reported that compared to rubber, RAPhad better chance of replacing aggregate in concrete mix especiallyat higher percentage of 10% by total mass of mix [7].

There are several reports that addressed the effects of fly ash as asubstitute material for Portland cement and sand in PCC and RCCmixtures [17–19]. During a laboratory study circulating fluidizedbed combustion (CFBC) ash was used to replace fine aggregates ofRCC. This material is a waste or by-product of petroleum coke com-bustion power stations and has a high content of CaO and SO3. Testresults showed that CFBC ash can increase the water absorptionand effectively reduce the initial surface absorption. Meanwhile,CFBC ash had a positive effect on compressive strength, splitting ten-sile strength and sulphate attack resistance of hardened RCC [20].

In a similar research study the strength properties of high-vol-ume fly ash (HVFA) RCC and superplasticised workable concretecured at moist and dry curing conditions were evaluated. Concretemixtures made with 0%, 50% and 70% replacement of ordinaryPortland cement (OPC) with two different low-lime class F flyashes were prepared. The study showed that producing highstrength concrete was possible with high volume fly ash content.HVFA concrete was found to be more vulnerable to dry curing con-ditions than was the OPC concrete. Finally, it was concluded thatHVFA concrete was an adequate material for both structural andpavement applications [2].

During a laboratory study the strength properties of a high vol-ume fly ash RCC was investigated analysing the rate of strengthchanges at various curing times. It was concluded that at early agesof curing the strength of HVFA was poor, while the fly ash effectwas low or negative. Following its curing age, the strength ofRCC increased rapidly; meanwhile, the fly ash effect graduallyimproved and was more beneficial to raising the flexural strength.Furthermore, it was found that at long curing ages its effect on highvolume fly ash RCC becomes more remarkable by increasing the flyash proportion [21].

In a similar study the influence of fly ash on the fatigue perfor-mance of RCC mix was investigated. It was found that if added atthe rate of 15–30% of the cement content, fly ash improves the porestructure and increases the fatigue life of RCC mix [22]. Apart fromthe fly ash content and dimension, knowledge about the Portlandcement and fly ash hydration mechanisms is of high importancein the practical usage of fly ash in cement mixtures. The quantityof CaO in the fly ash can determine the course of hydration andinfluence the strength of cement paste [23].

Rice husk ash (RHA) is a powder that affords from combustionof rice husk. This ash is a potential source of amorphous reactivesilica which has a variety of applications in materials science. Mostof the ash is used in the production of Portland cement. Whenburnt completely, the ash can have a Blaine No. of as much as3600 cm2/gr compared to the Blaine No. of cement between 2800and 3000 cm2/gr, meaning it is finer than cement. In recent yearsthe potential use of RHA has been investigated in various pave-ment applications. For example it has been used as active filler inhot mix asphalt and as cement complement in conventional andspecial PCC mixtures [6,24,25].

The main objective of this research is to investigate the effectsof RHA on the mechanical properties of RCC mixture. In this regarddifferent mix designs were considered including RCC mixture con-taining original and RAP materials.

Using RAP material in RCC has several advantages. Due tobitumen coating, RAP particles have lower porosity and waterabsorption than the natural aggregates. Therefore, higher moisturewill be available which could react with cementitious materials.Hence, the use of RAP material will help to achieve higher hydra-

Table 1Physical properties of original aggregates and RAP materials after the bitumenextraction.

Test Standard Unit Result

Original RAP

Compressive strength BS 812-3 kg/cm2 800 611Fractured particles ASTM: D5821 % 83 57Los angles abrasion STM: C131 % 20 35Soundness ASTM: C88 % 1.8 3.7Water absorption ASTM: C127 % 2.2 1.4

Table 2Chemical compositions of studied type II cement and RHA.

Component(%)

Type IIcement

RHA after ignition at (�C)

500 600 700 800 900

SiO2 21.84 85.21 86.11 87.53 87.37 87.49AL2O3 4.72 0.51 0.44 0.48 0.55 0.55Fe2O3 3.77 0.18 0.19 0.18 0.21 0.19MgO 1.15 0.28 0.29 0.30 0.31 0.32CaO 61.14 1.11 1.18 1.22 1.24 1.25Na2O 0.44 1.21 1.27 1.29 1.28 1.29SO3 2.23 1.33 1.39 1.38 1.40 1.38K2O 0.65 2.95 3.08 3.15 3.17 3.18LOI 1.63 6.52 5.25 3.69 3.62 3.56Other 2.43 0.70 0.80 0.78 0.85 0.79Total 100 100 100 100 100 100

A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236 229

tion progression at a constant water/cement ratio. In addition theangularity of the RAP materials is lower than that of original aggre-gates. The latter property along with the lubricating effects of thebitumen coating could increase the workability and compactibilityof the RCC mix at lower water/cement ratios.

Likewise, there are limited data about the fatigue behavior of RCCmixes which is considered as a main structural failure in rigid pave-ments design. Portland cement association (PCA) utilizes the fatiguecracking as a major criterion in its pavements design procedure [3].

Therefore this study was designed to achieve the followingobjectives:

– Investigate the effects of RHA on RCC mix properties at differentages. Intended mechanical properties consisted of the compac-tion properties, compressive strength, modulus of rupture,energy absorbency and fatigue response of RCC mix.

– Evaluate the effects of various RAP sizes as replacement of ori-ginal aggregate on the mechanical properties of RCC mix.

2. Experimental design

2.1. Materials

The different materials used to prepare RCC specimens in thisstudy were water, cement, RHA, RAP and original aggregate mate-rials. Type II Portland cement was utilized with the specificationsin accordance with ASTM: C150 standard method. The aggregategradation was selected according to PCA instruction [26]. Fig. 1,depicts the design aggregate gradation and the upper and lowerlimits determined in PCA instruction. As shown in this figure themaximum nominal size of aggregates was as equal to 19 mm. Itshould be mentioned that the same gradation was selected for allstudied mix proportions. The physical properties of the originalaggregates and RAP materials after the bitumen extraction havebeen presented in Table 1.

2.2. RAP processing

RAP materials were prepared from the west deposit of Mazand-aran province of Iran. Large pieces of RAP were put in an oven withthe temperature of 110 �C for about 30 min. After softening, RAPmaterials were crushed by means of a rubber hammer. Then mate-rials were sieved to achieve the desired gradation. During the laststage materials were separated into two fractions of coarse (i.e.retained on No. 4 sieve) and fine (i.e. passed No. 4 sieve) graded RAP.

2.3. Cement

Type II cement was used in this research. The chemical composi-tions of this cement which were determined by X-ray fluorescencespectrometry according to ASTM: E1621 are presented in Table 2.

Fig. 1. Gradation of aggregates used to prepare the RCC mixes.

2.4. Rice husk ash (RHA)

During the RHA preparation first the rice husks were burnt infree air condition in a special furnace for about 2 h. Then the result-ing black powder was transferred to an electric arc furnace withthe capability of discharging the CO2 content of RHA. Table 2 pre-sents the chemical compositions of RHA after different ignitiontemperatures. As presented the loss on ignition (LOI) continuouslyreduced upon increasing the ignition temperature. Since thechanges in LOI at higher ignition temperatures of 700 �C were neg-ligible, this temperature was determined as proper ignition tem-perature for RHA production.

2.5. Mix combinations

In order to attain the research objectives, 12 mix proportionswere prepared. Table 3, presents the details of studied mix propor-tions. According to the literature and published mix design instruc-tions the amount of total cementitious materials (i.e. cement + flyash content) in RCC has been usually recommended between 12%and 16% by dry mass of aggregates [1,3,26]. Therefore, in thisresearch the average of these quantities was selected and for allmix proportions the amount of cementitious materials (i.e.cement + RHA) was fixed to 14%. The reference RCC mix contained14% cement and original aggregates in both coarse and fine frac-tions. In mix No. 2 the coarse part of aggregates replaced withRAP materials. Mix No. 3 contained fine RAP and mix No. 4 com-pletely prepared with RAP materials. It should be mentioned thatthe gradation of all mix combinations were the same as that ofshown in Fig. 1. Mixes No. 5–8 contained 11% cement and 3%RHA and the same aggregate proportions that expressed forprevious mixes. Similarly as presented in Table 3, mixes No.9–12 contained 9% cement and 5% RHA.

2.6. Mixing and curing

The main experimental design consisted of unconfined com-pressive strength test which was performed on cylindrical speci-

Table 3Different RCC mix proportions.

No. Coarse aggregate(retained on No. 4sieve)

Fine aggregate(passed No. 4sieve)

Additive (%)

Cement RHA

1 Original Original 14 –2 RAP Original 14 –3 Original RAP 14 –4 RAP RAP 14 –5 Original Original 11 36 RAP Original 11 37 Original RAP 11 38 RAP RAP 11 39 Original Original 9 5

10 ssRAP Original 9 511 Original RAP 9 512 RAP RAP 9 5

230 A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236

mens and the flexural strength, the energy absorbency and fatiguetests which were performed on beam shape specimens using threepoints bending procedure.

For compressive strength specimens were compacted withstroke hammer compactor (i.e. proctor compaction). In this proce-dure RCC mixes were compacted with various water contents andthe maximum dry density (cdmax) was determined. Fig. 2, showsthe effect of water content on the dry density of mixes No. 1, 2, 9and 10. As seen in this figure the dry density of compacted mixesfirst increased by increasing the water content up to an optimummoisture value. After that the dry density drastically reduced byincreasing the water content. Incorporating RHA increased theoptimum water content and reduced the maximum dry density.The optimum water content was determined based on the maxi-mum density criterion for all RCC mixes. Table 4 presents theresults of proctor compaction test. The relationship between thedry density and water content was developed by a polynomialfunction. Afterward, the optimum moisture content (OMC) andthe ratio of water to total cementitious materials (OMC/C) wasdetermined. According to this table for all mix proportions theOMC reduced by the addition of RAP materials into RCC mixtures.This could be related to lower water absorption of RAP materialsthan the original aggregates. According to Table 1, the waterabsorption of original aggregates was as equal to 2.2%, whereasRAP materials had a water absorption of 0.6% and 1.4% beforeand after of bitumen extraction, respectively. For RCC mixescontaining 3% RHA (i.e. No. 5 to No. 8 specimens) the OMC wasmore than the reference specimen. This could be related to higherspecific surface of RHA compared to cement. Studied RHA had aBlaine number of 3450 cm2/gr which was higher than that of typeII cement (i.e. 3050 cm2/gr).

Further studies have been carried out on RCC mixes preparedwith the optimum water content. The mixing proportions of RCCs

Fig. 2. Dry density–water content curves of mixes No. 1, 2, 9 and 10.

which determined on the basis of the OMC value are given inTable 5. These mixing proportions were utilized to prepare themain experimental design RCC specimens.

Flexural strength test was accomplished using beam shapespecimens. In order to achieve a uniform density throughout thebeam, specimens were compacted by a vibratory table.

In this method, mixtures were poured into the prism shapemetal molds in three layers and each layer was vibrated for1 min while compacting by a 20 kg surcharge. Fig. 3, depicts animage of the beam specimens after the compaction process.

Before removing from the mold, both beam and cylindricalspecimens were cured at laboratory temperature for 24 h whilecovered by plastic envelopes. Then, samples were removed fromthe mold and curing continued by putting them into a water bath.In order to investigate the effect of curing time, all experimentaltests except fatigue were accomplished on 7, 28 and 120 dayscured samples. Since fatigue cracking is considered as a long termpavement failure, this test was conducted on 120 days specimens.

2.7. Testing procedures

72 Cylindrical specimens with 100 mm diameter and 200 mmheight were prepared to perform the compressive strength test.This test was accomplished based on the ASTM: C39 standardmethod. Specimens were tested after each curing period at labora-tory temperature of 25 �C.

Three points bending test was carried out using a universal test-ing machine. Upon performing this test the modulus of rupture wasmeasured. Moreover, the energy absorbency or flexibility of RCCmixes was evaluated using the force–deformation diagram. Totally72 beam specimens, 300 mm in length and 50 by 50 mm in crosssection were prepared to perform the modulus of rupture test.According to Fig. 4, during the loading period the load–deformationdiagram was automatically drawn by related software. At the end ofthe test, the modulus of rupture was calculated by Eq. (1).

MR ¼3PL

2bd2 ð1Þ

where MR is the modulus of rupture (MPa), P is the maximumapplied load (N), L is the span length (mm), b is the specimen width(mm), and d is the specimen depth (mm).

As mentioned, the load–deformation diagram was used to mea-sure the energy absorbency (Ea) of RCC mixes. This parameter rep-resents the material flexibility or the capability of material todeform without fracturing. According to definition given in JCISF4standard method, the energy absorbency of beam shape specimenin three points bending test is equal to the total area under theload–deformation diagram [27]. Eq. (2), presents the mathematicalfunction used to calculate the Ea parameter.

Ea ¼Z Df

0FðDÞdD ð2Þ

where F(D) is the load magnitude at midspan deflection of D, (N), Dis the midspan deflection (mm), and Df is the maximum midspandeflection (mm).

Fatigue test was conducted by three points bending method onspecimens with similar dimension to MR test. This test was per-formed at loading frequency of 1.0 Hz. Stress control method wasused in which the stress ratio is defined as Eq. (3):

SR ¼ rf

MRð3Þ

where SR is the stress ratio, rf is the flexural strength (MPa), and MR

is the modulus of rupture (MPa).This test was carried out at three stress ratios of 0.65, 0.75 and

0.85. At each condition the test repeated three times and the

Table 4The OMC of RCC mixes based on the maximum dry density criterion.

Mix No. Dry density (cd)–water content (W) relationship R2 Optimum moisture content (OMC)% (OMC/C) Ratio

1 cd = �27.36W2 + 322.41W + 1441.8 0.93 5.9 0.422 cd = �8.13W2 + 87.31W + 2080.3 0.85 5.4 0.393 cd = �10.54W2 + 116.11W + 1998.5 0.87 5.5 0.394 cd = �10.54W2 + 109.82W + 1974.4 0.91 5.2 0.375 cd = �16.48W2 + 203.78W + 1666.7 0.93 6.2 0.446 cd = �13.38W2 + 150.16W + 1805.4 0.96 5.6 0.407 cd = �8.96W2 + 108.91W + 1919.2 0.97 6.1 0.448 cd = �6.29W2 + 70.29W + 2038.7 0.90 5.6 0.409 cd = �10.63W2 + 147.92W + 1693.6 0.91 7.0 0.50

10 cd = �11.27W2 + 150.48W + 1660.5 0.94 6.7 0.4811 cd = �11.0W2 + 142.46W + 1687.5 0.85 6.5 0.4612 cd = �11.9W2 + 148.36W + 1645.3 0.92 6.2 0.44

Table 5Laboratory mix design of RCC mixes (kg/m3).

Mix No. Coarse RAP Fine RAP Coarse aggregate Fine aggregate Cement RHA Water

1 – – 1084 952 285 – 1202 1084 – – 952 285 – 1113 – 952 1084 – 285 – 1114 1084 952 – – 285 – 1055 – – 1084 952 224 61 1256 1084 – – 952 224 61 1147 – 952 1084 – 224 61 1258 1084 952 – – 224 61 1149 – – 1084 952 183 102 143

10 1084 – – 952 183 102 13711 – 952 1084 – 183 102 13112 1084 952 – – 183 102 125

Fig. 3. Beam specimens after compacting by vibratory table.

Fig. 4. Test setup in modulus o

A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236 231

diagram of mid-span deflection was drawn by related software.The fatigue life was determined as the total number of load appli-cations that led to complete cracking of specimen near themid-span point. Fig. 5, shows the deformation curve of mix No. 5which was tested at stress ratio of 0.65. As shown in this figure fati-gue life is obvious from the loading–deflection curve. According tothis figure, the deflection curve has a vertical asymptote at thepoint of complete fracturing. Fig. 6, shows an image of the mixNo. 5 at the end stage of the fatigue test.

Finally, in order to investigate the porosity of different mixesand its relationship to material’s behavior, the RCC porosity wasmeasured as presented by Eq. (4).

P ¼ 1� W2 �W1

V :qw

� �� �� 100 ð4Þ

where P is the RCC total porosity, %, V is the sample volume, cm3,qw is the water density, which assumed as equal to 0.001 kg/cm3,W2 is the specimen’s mass after drying for 24 h in an oven at105 �C, kg, and W1 is the under water specimen’s mass, kg.

f rupture and fatigue tests.

Fig. 6. Fractured specimen at the end stage of fatigue test.

Fig. 7. Comparison between the compressive strength of RCC mixes after differentcuring times.

Table 6Effect of curing time (D) on the compressive strength (rc) of RCCmixes.

Mix No. Function R2

1 rc = 4.07 ln (D) + 25.60 0.932 rc = 4.14 ln (D) + 14.65 0.893 rc = 3.54 ln (D) + 10.58 0.914 rc = 2.74 ln (D) + 6.47 0.975 rc = 6.85 ln (D) + 13.89 0.946 rc = 5.68 ln (D) + 3.66 0.877 rc = 5.46 ln (D) + 0.51 0.928 rc = 4.47 ln (D) � 1.51 0.959 rc = 6.23 ln (D) + 9.01 0.93

10 rc = 5.88 ln (D) � 0.19 0.9011 rc = 5.45 ln (D) � 0.87 0.9812 rc = 4.76 ln (D) � 3.64 0.88

232 A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236

3. Results and discussion

3.1. Compressive strength

Fig. 7 compares between the compressive strength of studiedmix proportions after different curing times. The minimumrequirement for compressive strength relates to the layer in whichthe RCC is used. As the main structural layer, at least a compressivestrength of 27.6 MPa is needed [3]. Based on this figure only mixesNo. 1 and 5 satisfied this criterion after 7 days of curing. However,mixes No. 2, 5 and 9, satisfied the criterion after 28 days of curing.

Results of Fig. 7 indicate a systematic reduction in compressivestrength of RCC mixes upon replacing various RAP dimensions. Asit can be seen the coarse RAP mix had the least strength decrease,whereas, the strength reduction of the fine RAP mix was inbetween the coarse and the whole (i.e. both coarse and fine) RAPmix.

In order to better investigate the effect of curing time especiallyfor those mixes containing RHA, the relationship between the cur-ing time and compressive strength was developed by logarithmicfunctions. Table 6 presents the related functions of various RCCmixes. As presented, for RHA containing mixes curing time had aprominent effect on the compressive strength of specimens. Itcan be realized from the higher slopes of the functions thatobtained for mixes No. 5 to No. 12 compared to mixes No. 1 toNo. 4. At the beginning of the curing period the calcium hydroxideconcentration is low and pozzolan has least activity. At longer cur-ing times the calcium hydroxide concentration increases as a con-sequence of hydration progression. For an appropriate curingcondition the pozzolanic reaction starts to develop after about 7–14 days and during the progression of this reaction the calciumhydroxide concentration will continuously reduce [23,28]. Thisreaction leads to the production of secondary cementitious compo-sitions which have higher volume than the preliminary composi-tions. Therefore after a proper curing, the mix containingpozzolanic material has usually lower void contents and the moreintegrated structure. Formation of secondary cementitious compo-sitions along with the reduction of the pore sizes have been recog-nized as the main factors influencing the long term strengthgaining of the pozzolanic materials [23,28,29].

3.2. Modulus of rupture (MR)

Results of MR test at different curing times have been shown inFig. 8. Based on this figure, the coarse RAP mix showed higher MR

value than the fine RAP. Furthermore, the addition of RHA contentup to 3% had beneficial effects on MR of 120 days RCC mixes. How-ever, further increase of this additive to 5% resulted in lower MR

values. Results of MR test are to a large extent similar to the com-

Fig. 5. Deformation curve and fatigue life definition in three points bending fatiguetest.

pressive strength test. Based on the previous studies the differencebetween the calcium hydroxide concentration during the curingperiod is the main reason of different behavior for pozzolanicmaterials such as fly ash and ordinary Portland cement (OPC). Inthe case of OPC the calcium hydroxide concentration which hasnegligible contribution on strength gaining steadily increases byincreasing the curing time [28]. According to Table 2, the amountof free CaO in RHA was negligible. If the chemical composition offly ash is compared to cement, fly ash with high CaO content ismuch more similar to cement than that of containing a low CaOcontent [23,28]. For RHA with low CaO content sufficient calciumhydroxide must be produced from hydration reaction which couldactivate the pozzolanic compounds of RHA. Based on the obtainedresults it could be presumed that for 5% RHA, the amount of cal-cium hydroxide that produced from the hydration reaction wasnot sufficient to activate all available pozzolan particles. In this

Fig. 9. Load–deformation curves of RCC mixes in modulus of rupture test.

A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236 233

case the excess RHA remained inactive and constituted weak zonesin concrete matrix which caused a reduction in MR of RCC mix.

3.3. Energy absorbency (Ea)

With reference to Eq. (2), the energy absorbency was evaluatedby calculating the area under the load–deformation curve obtainedfrom the three points bending test. Fig. 9, shows the load–deforma-tion curves of mixes No. 1 to No. 4 after 28 days of curing.

The material with higher energy absorbency is more flexibleand expected to behave better under the effects of repeated load-ings. As seen in Fig. 9, mix No. 1 completely fractured after about1.3 mm bending deformation, while, other mixes containing RAPmaterials tolerated considerably higher deformations. The maxi-mum deflection (Df) of mixes No. 2, 3 and 4, was as equal to 2.7,2.1 and 2.2 mm, respectively. The mechanism of crack propagationin RCC mix with original aggregate is different from that of con-taining RAP material. Due to high adhesion between the aggregateand cement mortar in conventional RCC, crack propagates throughaggregate particles. In contrast, for the RAP incorporated mix, bitu-men forms a thin film at the interface of cement mortar and aggre-gate which may cause to arrest the crack propagation. Hence, crackdevelops around rather than go through aggregate particles andthis phenomenon leads to increase of the total deformation ofspecimen before complete fracturing [7].

Fig. 10, compares between the energy absorbency of studiedmixes. As it can be seen the inclusion of RAP increased the Ea ofmixes. Furthermore, RCC mixes containing coarse and whole RAPexhibited higher Ea than the fine RAP mix.

As seen in Fig. 10, the energy absorption increased by increasingthe curing time. After 7 and 28 days curing, the Ea of those mixescontaining RHA was to some extent less than the reference mix.The energy absorption of these mixes increased by a lower ratecompared to the conventional RCC. However, for 120 days curedspecimens the Ea of RCC mixes containing 3% RHA was even higherthan the conventional mix. Although the increase of RHA contentto 5% was not efficient and the energy absorbency of 120 daysRCC containing 5% RHA was lower than the reference one.

3.4. Fatigue

The relationship between the fatigue life and stress ratio (SR) ofmixes No. 1 to No. 4 has been shown in Fig. 11. It should be men-tioned that in these curves each point represents the average ofthree fatigue tests that performed at each SR. Therefore, each fati-gue curve was established based on the results of 9 fatigue tests.According to this figure mixes No. 2 and 3 behaved similarly underrepeated loadings. However, the mix No. 2 that contained coarseRAP exhibited slightly higher fatigue resistance than that of theNo. 3 mix containing fine RAP. In contrast, mixes No. 1 and No. 4

Fig. 8. Results of modulus of rupture test after different curing times.

revealed considerably different results at various SRs. At higherSRs than 0.75 mix No. 1 had higher fatigue life whereas at lowerratios the reverse was true. Fig. 12 depicts the fatigue laws thatobtained for RCC mixes containing 3% RHA. As seen for these mixesthere was a similar trend between the fatigue life and SR value. AtSR of about 0.75 the mix No. 5 with original aggregate showed bet-ter fatigue response than other mixes, whereas at lower SR quan-tities RAP containing mixes showed to some extent higherfatigue life. As illustrated, at SR of 0.65 the fatigue life of mix No.5 was about one half of other mixes.

Table 7 presents the obtained fatigue curves of different studiedmixes. As presented the slopes of fatigue curves increased byincreasing the RHA content. Similarly, for RAP incorporated mixesthe slopes were noticeably more than the conventional RCC.Fig. 13, compares between the fatigue curves of the RCC mixes con-taining original aggregates and various RHA contents (i.e. mixesNo. 1, 5 and 9). According to this figure the mix No. 5 revealed sim-ilar fatigue response to mix No. 1. At SR of o.65 the average fatiguelife of mix No. 5 was about 84% of mix No. 1. However the differ-ence between the fatigue lives reduced upon increasing the SRvalues so that at SRs of 0.75 and 0.85 the fatigue life ratios wereas equal to 87% and 114%, respectively. It means that at higherSR values RCC mixes containing 3% RHA even behaved better fati-gue response than the conventional RCC. The addition of RHA con-tent to 5% caused a considerable reduction in fatigue resistance ofRCC mixes. According to this figure the fatigue life of mix No. 9 atSRs of 0.65%, 0.75% and 0.85% was equal to 38%, 84% and 79% of mixNo. 1, respectively. Laboratory observations revealed similar trendfor other RCC mixes that contained 5% RHA. Fly ash can react with

Fig. 10. Comparison between the energy absorbency of different RCC mixes.

Fig. 11. Fatigue curves of No. 1 to No. 4 mixes.

Table 7Fatigue equations of studied RCC mixes.

Mix No. Fatigue equation R2

1 SR = �0.025 ln (Nf) + 0.965 0.942 SR = �0.028 ln (Nf) + 0.993 0.913 SR = �0.028 ln (Nf) + 0.990 0.954 SR = �0.034 ln (Nf) + 1.041 0.905 SR = �0.026 ln (Nf) + 0.971 0.946 SR = �0.029 ln (Nf) + 0.997 0.887 SR = �0.029 ln (Nf) + 0.998 0.968 SR = �0.036 ln (Nf) + 1.063 0.919 SR = �0.028 ln (Nf) + 0.973 0.87

10 SR = �0.030 ln (Nf) + 0.992 0.9611 SR = �0.030 ln (Nf) + 0.996 0.9512 SR = �0.036 ln (Nf) + 1.037 0.94

Fig. 13. Fatigue curves of No. 1, 5 and 9 mixes.

234 A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236

the cement hydration product calcium hydroxide. At high calciumhydroxide concentrations the RHA shows considerable pozzolanicreactivity. This phenomenon leads to a reduction in the calciumhydroxide concentration and producing more CSH gel. As a resultthe pore structure is improved and the matrix is densified. Other-wise, too fine RHA particles act as inert material which has nocementitious property [20–22]. Therefore, it could be inferred thatin RCC mix containing 5% RHA a part of pozzolan remained inactiveand this issue resulted in formation of weak zones throughout theconcrete structure.

Improvement of pore structure was recognized as an importantfactor in increasing the material’s flexibility and resistance againstthe fatigue damage [22]. In order to investigate the effects of RHAcontent on the porosity of material this parameter was determinedas presented by Eq. (4) Fig. 14, depicts the changes of the RCC poros-ity after different curing times. As it can be seen, for mixes contain-ing 3% and 5% RHA the mix porosity reduced by increasing the curingtime. At initial curing period the porosity of mixes No. 1 to No. 4 wasapproximately the same as those of mixes No. 5 to No. 8. In continuethe porosity of mixes No. 5 to No. 8 reduced by increasing the curingtime. The latter result indicated the pozzolanic reaction progressionwhich led to a more integrated structure and lower void contents inRCC mixes. In contrast, the addition of RHA content to 5% in mixesNo. 9 to No. 12 led to an increase in porosity of mixes. It could beinferred that in these mixes a part of RHA remained unused and onlybehaved like an inactive filler material.

3.5. Fatigue life (Nf) – energy absorbency (Ea) relationship

Comparison between the results of the fatigue and Ea testsrevealed a meaningful relationship between the slopes of fatiguecurves and the energy absorption of RCC mixes. The relationshiphas been presented by Eq. (5).

Fig. 12. Fatigue curves of No. 5 to No. 8 mixes.

Ea ¼ �120367SV� 287:6;R2 ¼ 0:668 ð5Þ

where Ea is the is the energy absorbency of 120 days specimens aspresented in Fig. 10, N mm, and SV is the slope of fatigue curves asestablished in Figs. 11–13.

According to Eq. (5), the absolute slope values of fatigue curvesreduced by increasing the Ea of RCC mixes. As a result there weresome intersection points between the fatigue curves of differentRCCs. As shown in Figs. 11 and 12, fatigue curves intersected atSRs between 0.72 and 0.75. Therefore, it could be concluded thatat higher than these SRs the mix with higher Ea revealed higherfatigue life whereas at lower SR levels the reverse was true.

A regression analysis was carried out using SPSS statistics soft-ware for determining the relationship between the energy absor-bency (Ea) and fatigue life (Nf) at various SRs. This correlationmeasures how variables or rank orders are related. In this analysisa correlation coefficient is determined which varies between �1and 1. The absolute value of correlation coefficient indicates how

Fig. 14. Effect of curing time on the porosity of RCC mixes.

Table 8Results of correlation analysis for RCC mixes.

Range of SR Parameters SR Ea

SR 6 0.72 Nf �0.797** �0.352*

SR > 0.72 Nf �0.811** 0.334*

* Correlation is significant at the 0.05 level.** Correlation is significant at the 0.01 level.

A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236 235

two analyzed parameters are perfectly related. Furthermore, thepositive or negative quantities signify the straight or reverse rela-tionship between two compared parameters, respectively.

In this analysis the SR of 0.72 was selected as the referencevalue. Table 8 presents the results of this analysis for RCC mixes.With regard to this table the correlation between fatigue life (Nf)and SR was significant at the 0.01 level and the correlation coeffi-cients were negative. It means that at the confidence limit of 99%there was a reverse relationship between the SR and fatigue lifewhich is even obvious from the fatigue curves of Figs. 11 and 12.However, the correlation between the Ea and Nf was significantat the 0.05 level. At SR levels of 0.72 and below the correlationcoefficient was negative which indicated the reverse relationshipbetween the Ea and fatigue life. Therefore at the confidence limitof 95% at these SRs, the RCC mix with lower Ea exhibited predom-inant fatigue response. Vice versa, for the more flexible materialwith higher Ea the degree of fatigue damage reduced at higherSRs of 0.72.

3.6. Fatigue life–porosity relationship

Fig. 15, shows the relationship between the porosity and fatiguelife of RCC mixes at different SRs. As shown in this figure althoughthe R2 values were moderately strong but at all SRs there was areverse relationship between the mix porosity and fatigue life. Itmeans that the RCC mix with the more integrated structure orlower porosity exhibited better fatigue behavior compared to amix with high air void content.

During a research study related to porous concrete, differentRHA contents between 0% and 12% of cement were added to mixand the porosity of mixes were measured [25]. It was found thatthe porosity of porous concrete increased by increasing the RHAcontent up to 8%. After that any increase in RHA content led todrastic increase of the mix porosity. Furthermore, a similar rela-tionship was attained between the RHA content and compressivestrength of porous concrete. Finally, it was presumed that thestrength of the material was strongly (R2 > 0.9) related to concreteporosity or void content [25]. An earlier study showed similar rela-tionship between the fatigue life and porosity of RCC mix contain-ing less than 30% fly ash by total weight of total cementitiousmaterials [22].

Fig. 15. Relationship between the porosity and fatigue life of RCC mixes.

Based on the obtained results and abovementioned analysis itcould be concluded that if RHA is added to mix at an optimum con-tent, it will reduce the mix porosity and improve the mechanicalproperties (e.g. strength and fatigue resistance) of RCC mix.

4. Conclusions

In this study the mechanical properties of RCC mixes made withthe original and RAP materials containing rice husk ash as cementreplacement were investigated. Based on the test results and per-formed analysis the following conclusions can be drawn:

(1) The compaction properties of RCC were investigated byproctor and vibratory table methods. Incorporating RAPmaterials resulted in lower optimum moisture content andlower maximum dry density than the conventional aggre-gate. Furthermore, due to high specific surface adding RHAto RCC increased the optimum moisture content andreduced the maximum dry density.

(2) Using both coarse and fine RAP reduced the compressivestrength of RCC. However, substituting the coarse partresulted in higher compressive strength than the fine RAP.

(3) Due to progressive nature of the pozzolanic reaction thecompressive strength, modulus of rupture and flexibility ofRCC mixes improved by increasing the curing time.

(4) Incorporating the RAP materials in RCC increased the energyabsorbency. Moreover, substituting 3% RHA had beneficialeffects on the material’s flexibility. However, the energyabsorbency of material reduced by increasing the RHA con-tent up to 5%.

(5) The fatigue life of RCC mixes containing RAP materials waslower than the conventional RCC. However, replacing thecoarse part led to lower detrimental effects than the fineRAP.

(6) The RCC mix with 3% RHA had comparable fatigue life toconventional RCC. Increasing the RHA content to 5% had con-trary effects on the fatigue behavior of mixes.

(7) There was a meaningful relationship between the energyabsorbency and fatigue response of RCC. At higher stressratios of 0.72, the mix with higher energy absorbencybehaved better under repeated loadings.

(8) There was a reverse relationship between fatigue life andmaterial porosity. The addition of 3% RHA reduced theporosity especially after 120 days curing and improved thefatigue resistance of material. However, the addition ofRHA to 5% resulted in higher porosity and lower fatiguelives.

References

[1] Wahhabh A, Asi IM. Optimization of roller compacted concrete for localapplication. Transp Res Rec 1999;1458:1–7.

[2] Duran AC. Strength properties of high-volume fly ash roller compacted andworkable concrete and influence of curing condition. Cement Concr Res2005;35(6):1112–21.

[3] ACI 325.10. State of art Report on Roller Compacted Concrete Pavements.American Concrete Institute report ACI 325. 10-95, R2001.

[4] Yerramala A, Ganesh Babu K. Transport properties of high volume fly ash rollercompacted concrete. Cement Concr Compos 2011;33(10):1057–62.

[5] Habert G, Choupay N, Montel JM, Guillaume D, Escadeillas G. Effects of thesecondary minerals of the natural pozzolans on their pozzolanic activity. CemConcr Res 2008;38(7):963–75.

[6] Rahman ME, Muntohar AS, Pakrashi V, Nagaratnam BH, Sujan D. Selfcompacting concrete from uncontrolled burning of rice husk and blendedfine aggregate. Mater Des 2014;55:410–5.

[7] Huang B, Shu X, Li G. Laboratory investigation of Portland cement concretecontaining recycled asphalt pavements. Cem Concr Res 2005;35(10):2008–13.

[8] Al-Oraimi S, Hossam F, Hago HA. Recycling of reclaimed asphalt pavement inPortland cement concrete. J Eng Res 2009;6(1):37–45.

236 A. Modarres, Z. Hosseini / Materials and Design 64 (2014) 227–236

[9] Huang B, Shu X, Burdette EG. Mechanical properties of concrete containingrecycled asphalt pavements. Magaz Concr Res 2006;58(5):313–20.

[10] Okafor A, Fidelis O. Performance of recycled asphalt pavement as coarseaggregate in concrete. Leonardo Electron J Pract Technol 2010;17:47–58.

[11] Torkaman J, Ashori A, Sadr Momtazi A. Using wood fiber waste, rice husk ashand limestone powder waste as cement replacement materials for lightweightconcrete blocks. Constr Build Mater 2014;50:432–6.

[12] Courard L, Michel F, Delhez P. Use of concrete road recycled aggregates forroller compacted concrete. Constr Build Mater 2010;24:390–5.

[13] Li G, Stubblefield MA, Garrick G, Eggers J, Abadie C, Huang B. Development ofwaste tire modified concrete. Cem Concr Res 2004;34(12):2283–9.

[14] Karimpour A. Effect of time span between mixing and compacting on rollercompacted concrete (RCC) containing ground granulated blast furnace slag(GGBFS). Constr Build Mater 2010;24(11):2079–83.

[15] Neyisi EG, Gesoglu M, Zturan TO. Properties of rubberized concretescontaining silica fume. Cem Concr Res 2004;34(12):2309–17.

[16] Huang B, Li G, Pang SS, Eggers J. Investigation into waste tire rubber-filledconcrete. J Mater Civ Eng 2004;16(3):187–94.

[17] Eleftherios K, Papayianni I, Papachristoforou M. Behavior of self compactingconcrete containing ladle furnace slag and steel fiber reinforcement. Mater Des2014;59:454–60.

[18] Shaikh FUA, Supit SWM, Sarker PK. A study on the effect of nano silica oncompressive strength of high volume fly ash mortars and concretes. Mater Des2014;60:433–42.

[19] Marandi A, Andik-Cadir O, Ramyar K. Freeze–thaw resistance and transportproperties of high-volume fly ash roller compacted concrete designed bymaximum density method. Cement Concr Compos 2013;37:259–66.

[20] Chi M, Huang R. Effect of circulating fluidized bed combustion ash on theproperties of roller compacted concrete. Cement Concr Compos2014;45:148–56.

[21] Cao C, Sun W, Qin H. The analysis on strength and fly ash effect of roller-compacted concrete with high volume fly ash. Cem Concr Res2000;30(1):71–5.

[22] Qian B, Jin Z, Zhang Y, Qin H, Liu J, Sun W. Fatigue performance and equationsof roller compacted concrete with fly ash. Cem Concr Res 1998;28(2):309–15.

[23] Cojbasic L, Stefanovic G, Sekulic Z, Heckmann S. Influence of the fly ashchemical composition on the portland cement and fly ash mixture hydrationmechanism. Facta Univ Ser Mech Eng 2005;3(1):117–25.

[24] Kannan V, Ganesan K. Chloride and chemical resistance of self compactingconcrete containing rice husk ash and metakaolin. Constr Build Mater2014;51:225–34.

[25] Hesami S, Ahmadi S, Nematzadeh M. Effects of rice husk ash and fiber onmechanical properties of pervious concrete pavement. Constr Build Mater2014;53:680–91.

[26] Portland Cement Association. Structural design of roller-compacted concretefor industrial pavements. USA: PCA, 2003.

[27] JCI-Sf4. Standards for test methods of fiber reinforced concrete. Japan concreteinstitute 2003.

[28] Chengzhi Z, Aiqin W, Mingshu T. The filling role of pozzolanic material. CemConcr Res 1996;26:943–7.

[29] Ghafari E, Costa H, Júlio E, Portugal A, Durães L. The effect of nanosilicaaddition on flowability, strength and transport properties of ultra highperformance concrete. Mater Des 2014;59:1–9.