## Services on Demand

## Article

## Indicators

## Related links

- Cited by Google
- Similars in Google

## Share

## Journal of the South African Institution of Civil Engineering

##
*On-line version* ISSN 2309-8775

*Print version* ISSN 1021-2019

### J. S. Afr. Inst. Civ. Eng. vol.57 n.4 Midrand Oct./Dec. 2015

#### http://dx.doi.org/10.17159/2309-8775/2015/v57n4a3

**TECHNICAL PAPER**

**Effect of paste content on the properties of high-strength concrete pavements**

**M S Smit; E P Kearsley**

**ABSTRACT**

Ultra-thin continuously reinforced concrete pavement (UTCRCP) is an innovative pavement type that has the potential to fulfil South Africa's pavement repair strategy requirements. Premature failure in UTCRCP is linked to the formation of cracks that allow water ingress into supporting layers. Environmental conditions, as well as concrete properties, determine the concrete cracking tendency. Concrete properties are a function of the mixture proportions, and unlike environmental conditions, mixture proportions are controllable. The effect of mixture proportions on the properties of high-strength concrete (HSC) used in UTCRCP should be investigated.

The objective of this research was to study the influence of paste content on the properties of HSC used in UTCRCP. Two sets of concrete were tested. The paste content of the first set was varied from 23% to 37% by mass, using multivariable analysis in conjunction with superplasticiser (SP) dosage. The paste content of the second set was varied from 25% to 60% by mass, only varying SP dosage to control the workability.

The multivariable analysis revealed that, within the parameter range tested, paste content influenced early-age properties, but not long-term properties. Through variation of the paste content over a wider range during the second set it was found that paste content does influence both the early-age and long-term properties of HSC. From the results it could be seen that increasing the paste content of HSC generally has a detrimental effect. The paste content of HSC used in UTCRCP should be minimised, while maintaining a reasonable workability.

**INTRODUCTION**

Ultra-thin continuously reinforced concrete pavement (UTCRCP) is a developing pavement type and has the potential to fulfil many of South Africa's pavement repair strategy requirements (Kannemeyer *et al *2007). UTCRCP consists of an approximately 50 mm thin layer of 80 MPa high-strength concrete. It is reinforced continuously with small-diameter (5 mm), closely spaced, welded deformed steel bar mesh. Research by Kannemeyer *et al* (2007) showed that UTCRCP is sensitive to the development of cracks large enough to allow moisture ingress. Premature failures occur when moisture deteriorates the pavement substructure, resulting in loss of support in the overlay (SANRAL 2012).

Cracks are formed in concrete when restrained deformation induces tensile stress greater than the tensile strength at the time. Concrete shrinkage and thermal movement both influence the deformation of concrete (Giussani & Mola 2012). The materials and proportions of materials used in concrete control the deformation of concrete (Neville 1995).

It has been stated by SANRAL (2009) that the mixture proportioning of HSC is one of the main challenges in UTCRCP. The mixture proportioning of HSC is more complex than normal-strength concrete, because low water/cementitious ratios and a wider variety of materials are used (Zain *et al *2005). Workability and strength requirements are the main factors governing the mixture proportions used in concrete. Paste content consists of water, cement and admixtures, and is expressed as a percentage of the total concrete mass or volume. In the mixture proportioning of HSC there is an inclination to increase the paste content to facilitate workability (Alves *et al *2004). The high paste content and low water/cementitious ratio of HSC increase its cracking tendency. Crack formation in concrete can be controlled by taking cracking tendency into account during the mixture proportioning process (Khokhar *et al *2010).

The objective of this paper is to determine the effect of paste content on the properties of HSC, where the observed trends serve as a starting point for further research in minimising the cracking tendency in UTCRCP by mixture proportioning. The paper presents a literature review, the experimental programme and the analysis of the experimental data. Lastly, conclusions are drawn and recommendations are made.

**LITERATURE REVIEW**

Concrete undergoes a combination of load-dependent and load-independent deformation throughout its lifetime, and it is the cumulative contraction and expansion that induces a resultant stress in the presence of restraint. Load-independent deformation includes plastic shrinkage, chemical/ autogenous shrinkage, drying shrinkage, thermal deformation and carbonation shrinkage. Load-dependent deformation consists of elastic strain and creep. Load-independent deformation is a function of a combination of moisture and temperature effects. Load-independent strains induce stresses when the concrete is restrained, and lead to deformation that is affected by load-dependent deformation. Figure 1 is a schematic diagram which was compiled to clarify the load-dependent and load-independent deformation.

Cracks can start forming in concrete within the first 24 hours (Holt 2001). Figure 2 is a schematic illustration of crack development in concrete. It shows how a crack develops when the stress exceeds the tensile strength (Owens 2009). Figure 2 also shows how cracking tendency is affected by time. As concrete hardens the tensile strength increases, reducing the cracking tendency, while the modulus of elasticity increases and stress alleviation by creep decreases, increasing the cracking tendency.

HSC mix designs are often described as 'rich' because of their high cement content. The cement content is dependent on the water content, which has the tendency to be high to maintain the workability at low water/cementitious ratios. High cementitious content also ensures high early strength gain (Neville 1995). Conventional mixture proportioning methods have restrictions on the amounts of cement used to limit shrinkage cracking (Neville & Brooks 2010). These restrictions are not always implemented for HSC. Acceptable cement content ranges between 500 and 550 kg/m^{3}, while HSC has been made with cement content up to 1 000 kg/m^{3} (Alves *et al *2004).

In a study by Alves *et al *(2004) the mixture proportions obtained from four methods used to design HSC in Brazil were compared. Three of the four methods showed how the paste content increased as the water/cement ratio decreased. The only method that did not show this tendency was the mix design method based on the particle packing theory. The other mix design methods were a method used for the mix design of normal strength concrete and methods adjusted from normal strength concrete mix design methods for HSC. The research emphasised the tendency of replacing fine aggregate with binder material to maintain workability in HSC. The suitability of a mix was determined by its cost and strength, and the cracking tendency of the mix designs was not taken into account.

The liberation of heat, during the exothermic chemical reaction of cement hydration, causes a temperature rise in the concrete member. The temperature rise is dependent on various factors, including material properties, member geometry and environmental conditions that can be adiabatic or semi-adiabatic (RILEM 1997). Increasing the paste content of a concrete mix would increase the maximum temperature reached due to heat of hydration.

The combination of heat of hydration and thermal expansion can cause cracks (Neville 1995). The temperature of the concrete rises because of heat of hydration. As the heat dissipates, the magnitude of thermal contraction is dependent on the difference between the peak temperature and the ambient temperature. If the concrete is restrained the contraction causes tensile stress in the concrete, because the concrete hardens and gains strength at elevated temperatures. Tensile stresses sufficiently large to crack the concrete can develop (Domone & Illston 2010).

To obtain high early strength in HSC the cementitious content is often increased while the water/cementitious ratio remains constant (Bentz *et al *2011). The effect of the increased paste content on the long-term properties of HSC is not clear, but it has been shown for normal strength concrete that leaner mixes yield higher strengths after only seven days (Singh 1958). Stock *et al *(1979) found that, at a water/cement ratio of 0.5, compressive strength decreases as the aggregate content increases from 0% to 40% of total volume. With aggregate content exceeding 40% the compressive strength increases. This phenomenon can be explained by a combination of factors that include reduced total interfacial transition zone area for low aggregate content concrete and reduced porosity of high aggregate content concrete. A similar effect was recorded for the tensile strength of normal strength concrete. Because of the high strength of paste used in HSC concrete, it is possible that the trends found by Stock *et al *(1979) are not applicable to HSC.

Paste is the volumetrically unstable component of concrete. Deformation properties of concrete are significantly influenced by paste content. Autogenous shrinkage is caused by the withdrawal of water from the capillary pores in the hydrating paste. The magnitude of autogenous shrinkage increases as the water/cementitious ratio decreases. Autogenous shrinkage is important in HSC, because low water/cementitious ratios are used to obtain high strengths. The magnitude of autogenous shrinkage of concrete is also proportional to paste content.

Drying shrinkage is the volumetric contraction of concrete by the removal of water. Concrete with high water/cement ratios contain more water that can be removed from the paste. Similarly concrete with high paste content will also shrink more, due to the overall higher water content. When the aggregate content is increased from approximately 71% to 74%, drying shrinkage can be reduced by as much as 20% (Neville 1995).

The relative proportion of paste and aggregate influences the modulus of elasticity of concrete, but it is not usual to find significant variation due to this factor (Gutierrez & Canovas 1995). Contrary to this, Leemann *et al *(2011) found that an increase in paste content decreased the modulus of elasticity by as much as 20% for concrete containing 320 kg/m^{3} and 520 kg/m^{3} cement at a constant water/cement ratio. Aggregate volume concentration also affects the creep of concrete by restraining the movement of paste. An increase of aggregate content by volume from 65% to 75% can decrease creep by 10% (Neville 1995). Leemann *et al *(2011) found that concrete with high paste content, such as self-consolidating concrete, exhibits higher creep.

**EXPERIMENTAL PROGRAMME**

**Response Surface Methodology**

Design of Experiments (DoE) is the strategic planning and execution of experiments to reduce the experimental work required to obtain statistically relevant results. Response Surface Methodology (RSM) is a set of statistical and mathematical techniques (used in DoE) that assist in the modelling and analysis of responses that are influenced by a number of variables. The end goal of RSM is the optimisation of the response (Montgomery 2001).

RSM usually consists of three phases, where phase one focuses on the experimental study, phase two develops the response surface models and phase three uses the statistical models for optimisation (Lotfy *et al *2014).

Central Composite Design (CCD) is the RSM design used in this study. CCD is an augmented version of the factorial design with centre points (Stat-Ease Corporation 2014). CCD is useful in RSM, because it makes it possible to develop first- and second-order models.

True functional relationships can be approximated for a small range of variables by polynomials of higher degree. In a second-order model, curvature is represented by interaction and quadratic terms. A central composite matrix does not make it possible to plot a cubic model. Equation 1 is the characteristic form of a second-order model:

where:

= quadratic effects of a single variable

= interaction effect between two variables

= regression coefficients

= investigated factors

*k *= number of factors

*e *= observed noise or error

CCD consists of 2^{k} factorial points, 2k axial points and *n *centre runs. In this context *k *represents the number of independent variables. Figure 3 shows the three different types of points that define the region of interest for a two-factor design. The factorial points are situated on the corners of the square, the axial points are situated at distance alpha (α) away from the centre point on the negative and positive sides of each axis, and the centre points are situated at the intersection of the two axes.

The quality of prediction of a response surface design is improved when it is rotatable. Whether a response surface design is rotatable is dependent on alpha and the number of centre points. Alpha can be calculated as a function of the number of independent variables. When two independent variables are being tested, alpha will be equal to the radius of a circle fitted on the factorial points that form a square. (Note that the factorial points are also distance alpha away from the centre.)

General Linear Model Analysis of Variance (GLM-ANOVA) is used to determine the influence of a test parameter on the tested response. The higher-order terms that model curvature are excluded to develop a linear model of the response after which the ANOVA of the model is used to evaluate the effect of the test parameters. For a parameter to have a significant influence on a response the *F*-value has to be large and the probability associated with it, the *p*-value, has to be smaller than 0.05. Lotfy *et al *(2014) describes the p-value as a descriptor of the significance of the parameter on the test results. Similarly, the percentage contribution is a measure of the effect of the independent variables on the response.

**Materials**

Table 1 shows the relative density and Figure 4 shows the particle size distribution of the fine materials used in this study. The materials used were selected according to quality and availability. The particle size distribution of silica fume could not be obtained because of the material's tendency to agglomerate in suspension.

**Mix design**

The effect of paste content on the properties of HSC was determined using two sets of experiments. For the first set, multivariable analysis was used where the two parameters that control workability (paste content and superplasticiser (SP) dosage) were varied. CCD was used as the response surface design. The concrete set and its results are labelled as Set 1. When response surface design is applied, the range in which the parameters are varied must be such that the resulting concrete mixes must be useable. In using CCD the parameters could not be varied over a very wide range, because the concrete mixes were not allowed to be too dry to compact and the SP dosage was not allowed so high that it would cause segregation.

In the second set of experiments the effect of the paste content was of primary concern and the SP dosage was only varied to ensure good workability for each concrete mix. The base mix design was derived from a published mix design for UTCRCP (Mukandila *et al *2009). The concrete set and its results are labelled as Set 2. Figure 5 shows the range over which the parameters were varied for each set of concrete. The paste content is calculated as a percentage of the concrete mass and the SP dosage is calculated as a percentage of the cementitious content, which is all cement, fly ash and silica fume. The target compressive strength was 80 MPa.