Early age properties of self-compacting concrete -...

Early age properties of self-compacting concrete – Effects of fine aggregate and limestone filler

OSKAR ESPING Department of Civil and Environmental Engineering Building Technology CHALMERS UNIVERSITY OF TECHNOLOGY Göteborg, Sweden 2007

THESIS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY

Early age properties of self-compacting concrete

- Effects of fine aggregate and limestone filler

OSKAR ESPING

Department of Civil and Environmental Engineering

Building Technology

CHALMERS UNIVERSITY OF TECHNOLOGY

Göteborg, Sweden 2007

II CHALMERS, Civil and Environmental Engineering

Early age properties of self-compacting concrete - Effects of fine aggregate and limestone filler OSKAR ESPING ISBN 978-91-7291-890-0 © Oskar Esping, 2007 Doktorsavhandlingar vid Chalmers tekniska högskola Ny serie nr 2571 ISSN 0346-718X Department of Civil and Environmental Engineering Building Technology Chalmers University of Technology SE-412 96 Göteborg Sweden Telephone: +46 (0)31-772 1000 http://www.chalmers.se Chalmers Reproservice Göteborg, Sweden 2007

IIICHALMERS, Civil and Environmental Engineering

Early age properties of self-compacting concrete - Effects of fine aggregate and limestone filler Oskar ESPING Department of Civil and Environmental Engineering Building Technology Chalmers University of Technology

ABSTRACT

Self-compacting concrete (SCC) is a sensitive mix, strongly dependent on the composition and the characteristics of its constituents. It has to possess the incompatible properties of high flowability together with high segregation resistance, a balance made possible by the dispersing effect of water-reducing admixture combined with cohesiveness produced by a high concentration of fine particles. These fines and their effects on the early age properties of the SCC have been in focus in this present dissertation.

The effect of the specific surface area of gravel and limestone filler on the rheology of SCC was evaluated. Performed experiments clearly demonstrated that traditional methods for geometric characterization of the fines (size distribution, water absorption, fineness modulus, etc.) are not sufficient to ensure consistent quality of SCC. By measuring the specific surface area with a simplified gas adsorption method, BET(H2O), it was found that the specific surface area of a normal gravel, accepted by traditional methods for production of SCC, can vary up to 7000 m2/kg. A model is proposed, based on an assumption that 30 full molecular layers of water covering the particle surface are required to provide lubrication sufficient to create flowability, where a change in specific area is translated to a change in water demand for the concrete mix. It is suggested that an increase in BET(H2O)-area of 1000 m2/kg corresponds to an increase in water demand by approximately 0.85% by mass of the filler or gravel content for constant flowability.

Furthermore, the influence of mix design and fines BET(H2O)-area on the SCC’s early-age deformation was demonstrated. The autogenous (sealed) deformation was measured with a specially developed concrete dilatometer, together with capillary pore pressure and temperature. It is suggested that that autogenous shrinkage and rate of evaporation are the main factors promoting the risk of plastic shrinkage cracking. An increased particle surface decreased the rate and magnitude of evaporation, and consequently reduced the plastic cracking tendency, despite an increase in autogenous shrinkage.

Key words: self-compacting concrete, rheology, autogenous deformation, plastic shrinkage, pore pressure, limestone filler, specific surface area, BET

IV CHALMERS, Civil and Environmental Engineering

APPENDED PAPERS

The work presented in this doctoral thesis is based on the following papers, referred to in the text by their Roman numbers:

I. Slump flow values vs. Bingham parameters for high flowable mortars and concretes Esping O., Accepted for publication in 5th International RILEM Symposium on Self-Compacting Concrete, 3-5 September 2007, Ghent, Belgium, 2007.

II. Methods for characterisation of fillers and fines for self-compacting concrete Esping O., 3rd International RILEM Symposium on Self-Compacting Concrete, PRO 33, 17–20 August, pp 208-219, Reykjavik, Iceland, 2003.

III. SCC flowability: Effect of changes in particle surface area, and how to compensate for this Esping O., Accepted for publication in 5th International RILEM Symposium on Self-Compacting Concrete, 3-5 September, Ghent, Belgium, 2007.

IV. Investigation of autogenous deformation in self-compacting concrete Esping O., International RILEM conference on Volume Changes of Hardening Concrete, 20-23 August, pp 273-282, Lyngby, Denmark, 2006.

V. Investigation of early age deformation in self-compacting concrete Esping O., Löfgren I., 2nd International Symposium on Advances in Concrete Ccience, 11-15 September, Quebec, Canada, 2006.

VI. Effect of limestone filler BET(H2O)-area on the fresh and hardened properties of self-compacting concrete Esping O., Submitted for publication in Cement and Concrete Research, 2007

VCHALMERS, Civil and Environmental Engineering

CONTENTS

ABSTRACT III

APPENDED PAPERS IV

CONTENTS V

NOTATIONS VIII

1 INTRODUCTION 1

1.1 Background 1

1.2 Objective and limitations 2

1.3 Disposition of the thesis 3

1.4 Original features 3

2 RHEOLOGY 5

2.1 Introduction 5

2.2 Particle suspensions 7

2.3 Measuring techniques 22

2.4 Concluding remarks 27

3 EARLY-AGE DEFORMATION 29

3.1 Introduction 29

3.2 Plastic shrinkage cracking 31

3.3 Mechanisms of early-age deformation 32

3.4 Measuring techniques and methods 42


4 EXPERIMENTAL WORK 49

4.1 Introduction 49

4.2 Materials and mix design 50

4.3 Particle quantifications 53

4.4 Mortar and concrete quantifications 62


VI CHALMERS, Civil and Environmental Engineering

5 FINAL DISCUSSION AND CONCLUSIONS 73

5.1 Discussion 73

5.2 General conclusions 75

5.3 Suggestions for future research 76

6 REFERENCES 79

APPENDIX A: BET(H2O) 91

APPENDIX B: CONCRETE DIGITAL DILATOMETER 99

APPENDIX C: CONCRETE CRACKING RING TEST 105

APPENDIX D: GLOSSARY 111

PAPER I: SLUMP FLOW VALUES VS. BINGHAM PARAMETERS FOR HIGH FLOWABLE MORTARS AND CONCRETES

PAPER II: METHODS FOR CHARACTERISATION OF FILLERS AND FINES FOR SCC

PAPER III: SCC FLOWABILITY: EFFECT OF CHANGES IN PARTICLE SURFACE AREA, AND HOW TO COMPENSATE FOR THIS

PAPER IV: INVESTIGATION OF AUTOGENOUS DEFORMATION IN SCC

PAPER V: INVESTIGATION OF EARLY AGE DEFORMATION IN SCC

PAPER VI: EFFECT OF LIMESTONE FILLER BET(H2O)-AREA ON THE FRESH AND HARDENED PROPERTIES OF SCC

VIICHALMERS, Civil and Environmental Engineering

ACKNOWLEDGEMENT

The work presented in this doctoral thesis is one of the two outcomes of the project “Industrial building with in-situ cast concrete – New concepts”, which was made possible by a donation from Thomas Concrete Group / AB Färdig Betong to Chalmers University of Technology. The work has been conducted at the Department of Civil and Environmental Engineering (Chalmers University of Technology) and at the research centre of Thomas Concrete Group in Göteborg, during the period 2001-2007. During the first years this thesis was carried out under the supervision of Professor Lars-Olof Nilsson now at Lund University of Technology, and during the remaining years under Professor Tang Luping at Chalmers University of Technology. Professor Per-Erik Petersson at the Swedish National Testing and Research Institute has acted as assisting supervisor. I am most grateful for their support and assistance.

Grateful thanks are also addressed to:

Annika Wirje, Färdig Betong AB Jan-Erik Lindqvist, Swedish National Testing and Research Institute Martin Hansson, Sika Mats Karlsson, Thomas Concrete Group AB Mette Geiker, Technical University of Denmark Olafur Wallevik, The Icelandic Building Research Institute Peter Billberg, Swedish Cement and Concrete Research Institute Sten Rodenstam, Nordkalk AB

Special thanks are due to Thomas Concrete Group and AB Färdig Betong for their generous donation, support and confidence in me.

I would like to extend my special thanks to Ingemar Löfgren and Professor Tomas Kutti at Thomas Concrete Group AB for their involvement in this work, and for their indispensable help, guidance, encouragement and always valuable advice.

Finally, I would like to thank my future wife, Lotta, and children Albin and Elin, for supporting and having patience with me during my doctoral studies.

It is my hope that this paper will be reviewed critically. Any viewpoints, comments and suggestions about the text should be directed to me.

By the nature of an experimental investigation, each question is often answered with many new questions. And with knowledge arises a question whether the experimental course of action could be made in a more proper way. It ought to be noted that the purpose of a PhD study is educational, and that the work presented in this doctoral thesis is mainly intended as a personal learning process, not to produce unique or useful results. Yet it has to be performed with a scientific approach and to fulfil certain requirements of quality, which I hope to have achieved for the benefit of future research.

Göteborg, February 2007

VIII CHALMERS, Civil and Environmental Engineering

NOTATIONS

Symbol Description Unit A area [m2] [mm2] A Arrhenius constant of the liquid [Pa·s]

mA monolayer adsorbate molecule area [m2] C BET-constant related to the heat of adsorption [-] d nominal size [mm]

maxd maximum grain size [mm]

id mean particle size from two contiguous sieves [m] D diameter [m] [mm]

circleD circle diameter with the same area as measured area [m] De Deborah number [-]

maxD length of the major axis [m]

minD length of the minor axis [m] E evaporation [kg/m2]

0E fluid activation energy [J/mol]

pE deformation coefficient [-] F cumulative retained material on specified sieve size [weight-%]

cf compressive strength [MPa]

0cf compressive strength “without” pores [MPa] Fm Feret diameter [m] FM fineness modulus [-] k constant for the effect of porosity on comp.strength [-] k constant for the effect of particles >0.5 mm [-] l length [mm]

0l specimen initial length [mm] M torque [N·m] M molecular weight of adsorbate (~0.018 kg/mol for water) [kg/mol] m mass weight [kg] [g]

cm mass of container [kg] [g]

drym mass of oven dry specimen [kg] [g] na number of objects per unit area [pcs/m3]

O2Hn number of layer water molecules adsorbed [pcs] N Avogadro´s number (6.022·1023 molecules per mol) [mol-1]

wp partial water vapour pressure [Pa]

sp partial water vapour pressure at saturation [Pa]

op/p relative pressure [-] P perimeter [m]

wP pore water pressure [Pa] P pressure [Pa]

IXCHALMERS, Civil and Environmental Engineering

P porosity [-]

op partial pressure at saturation [Pa] Q volumetric flow rate [m3/s] r correlation coefficient [-] r meniscus radius [m] r interparticle distance [Å] R radius [m] R universal gas constant (R=8.314) [J/(mol·K)]

iR radius of inner cylinder [mm]

oR radius of outer cylinder [mm]

pR relative flow area [-] RH relative humidity [%][-] S specific surface area [m2/kg], [m2/m3] SF slump flow spread diameter [mm]

OBETHS 2 specific surface area by BET(H2O) [m2/kg]

O2BETH'S approximated SBETH2O-area for the whole mass fraction [m2/kg]

2NS specific surface area by BET(N2) [m2/kg]

blaineS specific surface area by blaine [m2/kg]

SEMS specific surface area by image analyze from SEM [m2/kg]

sizeS specific surface area by size distribution [m2/kg] SP superplasticizer content [kg/m3] t time [sec] t time from water addition to mix [hours] T temperature [°C], [K] TX thixotropy [Pa/s]

50T time to 500 mm spread [sec] u moisture content [kgw/kgmtrl] V volume [m3]

c/w water cement ratio [weight%] w average crack width [mm2]

eW evaporable water [kg/m3] x fitting parameter (2.5 for spheres) [-]

mx monolayer capacity [-]

05x mass fraction passing 0.5 mm sieve [-] X adsorption isotherm [kg/kg] y BET adsorption isotherm [-]

α angle [rad] or [º]

pβ retained water ratio [-] l∆ specimen change in length [µm] m∆ sample weight loss due to drying [kg]

δ phase lag (loss angle) [rad] ε strain [10-6m/m]

wγ surface tension of air-water interface (~0.074 n/m) [N/m]

X CHALMERS, Civil and Environmental Engineering

γ& shear rate [1/s] η viscosity [Pa·s]

appη apparent viscosity [Pa·sA]

cη viscosity of the fluid phase [Pa·s]

plη plastic viscosity [Pa·s]

∞η constant viscosity near infinite shear rate [function dependent]

0η constant viscosity near zero shear rate [function dependent] ρ density [kg/m3] σ shear stress [Pa]

appσ apparent yield stress [Pa]

sσ static yield stress [Pa]

yσ yield stress [Pa]

0σ Bingham yield stress [Pa]

0σ stress amplitude [Pa] *0σ apparent Bingham yield stress [Pa]

0γ strain amplitude [-] ϕ volume fraction of solid particles [-]

mϕ maximum possible volume fraction of solid particles [-] ϑ air velocity [m/s] ω angular frequency [rad/s] Ω angular velocity [rad/s] Abbreviation Description A aggregate ACC accelerator C cement BET Brunauer, Emmet, and Teller CDD concrete digital dilatometer CH calcium hydroxide C-S-H calcium silicate hydrate COV coefficient of variance F filler H2O (see W) LOI loss of ignition N2 nitrogen P powder RE retarder SCC self-compacting concrete SEM scanning electronic microscopy SF silica fume Size size distribution SP superplasticizer SRA shrinkage reducing admixture W water

C h a p t e r 1 – I n t r o d u c t i o n

1CHALMERS, Civil and Environmental Engineering

1 INTRODUCTION

1.1 Background Cement-based materials are the most abundant of all man-made materials and are among the most important construction materials, and it is most likely that they will continue to have the same importance in the future. However, these construction and engineering materials must meet new and higher demands. When facing issues of productivity, economy, quality and environment, they have to compete with other construction materials such as plastic, steel and wood. One direction in this evolution is towards self-compacting concrete (SCC), a modified product that, without additional compaction energy, flows and consolidates under the influence of its own weight.

The use of SCC offers a more industrialised production. Not only will it reduce the unhealthy tasks for workers, it can also reduces the technical costs of in situ cast concrete constructions, due to improved casting cycle, quality, durability, surface finish and reliability of concrete structures and eliminating some of the potential for human error.

However, SCC is a sensitive mix, strongly dependent on the composition and the characteristics of its constituents. It has to possess the incompatible properties of high flowability together with high segregation resistance. This balance is made possible by the dispersing effect of high-range water-reducing admixture (superplasticizer) combined with cohesiveness produced by a high concentration of fine particles in additional filler material [1]. The main mechanisms controlling this fine balance are related to surface physics and chemistry; hence, SCC is strongly dependent on the activity of the admixtures, as well as on the large surface area generated by the high content of fines [2]. Fresh SCC, like all cementitious materials, is a concentrated particle suspension with a wide range of particle sizes (from 10-7 to 30 mm for concrete). The particles are affected by a complex balance of interparticle forces (i.e. interlocking, frictional, flocculation, colloidal, van der Waals, and electrostatic forces), generating a time dependence and viscoplastic non-Newtonian behaviour. In addition, the concretes have irreversible time-dependent properties as a result of the chemical reactions [3].

Since the ratio of surface area to volume increases exponentially with particle irregularity and decreased size, this area has a predominant effect on fresh and hardened concrete [4]-[7]. Particles with apparently similar grading can provide significantly different effects on concrete properties. These differences can be related to the fineness, shape, surface texture, porosity, etc. If the particles provide a large specific area, a large quantity of water will be adsorbed on the particles’ surfaces and less water will be available to lubricate and disperse the particles, which is needed to provide flowability [8]. This adsorption will also influence the development of negative capillary pore pressure, causing the paste to contract [9]. These contracting capillary forces are inversely proportional to the meniscus radius, and hence the capillary tension stresses increase with decreasing pore sizes and interparticle spaces. As a consequence of the large surface area generated by the high concentration of powder material, and the retarding effect of superplasticizer, the SCC may develop a large autogenous shrinkage and thus a high plastic shrinkage cracking tendency [10].


2 CHALMERS, Civil and Environmental Engineering

As aggregates represent 60-80% of the total volume of concrete, it is obvious that their influence on the fresh and hardened concrete is substantial. Properties of fine particles (<63 µm) and their influence are often neglected, because traditional standard methods for characterizing aggregates are deficient. The content of fine particles, by volume or weight, is often small but the number of particles and their total surface area are large. A content of 3 vol-% of particles finer than 63 µm (EN 933-1, or 75 µm according to ASTM C33) is considered harmful, but these particles and their surface properties may have a considerable influence on the quality of the concrete, in either fresh or hardened state [8].

Powder material (filler, cement, etc.) is traditionally characterized by its size distribution and specific surface area by Blaine. Due to the methodology, where the surface area is determined by air permeability, based on packed spherical particles, information about the shape, texture and surface porosity is neglected [5][11]. An irregular and porous particle can provide a considerably larger surface area, accessible to water, than a spherical one (see Figure 1). Under real conditions, a particle external surface can be as much as 1000 times greater that that of a sphere with equal volume [12]. Other, possibly more correct, methods for geometric characterization, such as the BET with nitrogen gas or image analysis, are more seldom used due to their complexity and high cost.

Hence, to be able to offer a high-quality product, it is imperative to have understanding and knowledge of how the constituents affect the concrete’s fresh and hardened properties, and how to characterize the concrete and its constituents’ properties.

Figure 1. The enlarging effect of size (logarithmic) and shape on specific surface area, illustrated with a spherical particle and an irregular one (with same section area but 5 and 10 times larger perimeter).

1.2 Objective and limitations The main objective of this study was to gain more understanding and knowledge regarding basic mechanisms that govern the rheological and time-dependent characteristics of cementitious materials. A further aim has been to investigate the possibility to find a tool for quality control of the constituents in order to detect, and to compensate for, variations in the materials that adversely affect the flow behaviour and early-age deformation of self-compacting concrete. Combinations of cement, fillers and aggregates normally used in Sweden, together with a superplasticizer, were used. The effect of limestone fillers and the fine part of aggregates has been in focus. In this work, the properties studied were mainly particle characteristics, the rheology of suspensions, and autogenous deformation. Thus,

Specific surface area - Particel geometry

0

10000

20000

30000

40000

50000

60000

70000

0.001 0.01 0.1 1Size, D [mm] (log)

Are

a, S

[m2 /m

3] sphere

Irregular (5x)Irregular (10x)



admixtures (especially the superplasticizer) predominate for the properties of fresh SCC, and their effects have not been examined in this work. Nether have the thermal effects on early-age deformation been considered. Methods for evaluation of specific surface area, and correlation of measurement results from these methods with flow behaviour of mortars and concretes, comprised a main topic. The concrete is structurally divided into two stages: early age and long term, where early age is limited to 24 hours from mixing. Rheologically, the concrete is also divided into two stages: fresh and hardened, where the concrete (or mortar) is considered to be fresh as long as the matrix of suspended particles is, or can be turned into, a liquid.

1.3 Disposition of the thesis This thesis consists of an introductory text and six papers (Papers I to VI). The introductory text, divided into three main parts (Chapters 2 to 4), gives a more comprehensive description of the subjects treated and measuring methods used in the appended papers.

First, an introduction to the rheology of particle suspensions and the factors affecting it is presented in Chapter 2. Rheological characteristics, models, and methodology for measuring the rheology are defined here, with cementitious materials in focus.

In Chapter 3 an introduction to early-age volume change of cementitious materials is presented. Basic mechanisms and some general methods and principals for evaluating the unrestrained early age shrinkage are described.

The experimental methods and materials used in the appended papers are presented in Chapter 4. A more detailed description of the experimental setups is given here, complemented with some analysis of the methods. Further description of the methods, unique for this work, can be found in Appendices A, B and C.

In Chapter 5, discussions, conclusions and suggestions for future research are presented.

1.4 Original features The main original features of this doctoral thesis may be summarized as follows:

o The particles’ specific surface area by BET has been shown to provide a good correlation with mortar and concrete flowability (slump flow, yield stress and viscosity), whereas other traditional methods characterizing the particle size, geometry and surface area did not.

o The simplified BET method, using water vapour as adsorbate, was proven high potential means of geometrical characterization for fillers and fine aggregates.

o It was shown that a change in fillers’ or fine aggregates’ BET(H2O)-area can be used to calculate the change in mixing water for constant flowability of SCC.

o A potential method for determining the concrete’s linear autogenous deformation, with the ability to start measuring in the liquid state before set, has been presented.

o The effect of mix design and BET(H2O)-area on the concrete’s capillary pore pressure, autogenous deformation, and plastic shrinkage cracking tendency was demonstrated in experiments.

C h a p t e r 2 – R h e o l o g y


2 RHEOLOGY

2.1 Introduction Rheology is defined as the science of deformation and flow of matter, and describes the interrelation between force, deformation, and time [13]-[15]. It is normally applied to more or less fluid materials, or materials that exhibit a time-dependent response to stress. An important issue of rheology is the definition and classification of materials. One way to characterize a material is by its relaxation time, i.e. the time required to reduce a stress in the material by flow (see Table 1). Glass, for instance, is usually defined as a solid material, but over a longer times perspective it flows like a liquid, although extremely slowly.

Table 1. Classification of materials state in terms of their relaxation time (at 20ºC).

Gases Liquids Solids

Relaxation time [seconds] <10-6 10-6 – 100 >100

Another way of defining materials rheologically is by their flow characteristics, with a classification into viscous, elastic and viscoelastic. Gases and many liquids are defined as viscous. An ideal fluid is unable to store any deformation energy; when subject to stress, it flows and the deformation energy is dissipated as heat (resulting in a rise of temperature). Solids, on the other hand, can be described as elastic materials. Ideal elastic material stores all imposed deformation energy, and will consequently recover totally upon a stress release. But a large number of materials are viscoelastic (both viscous and elastic), i.e. they store some of the deformation energy in their structure while some is lost by flow. The simplest model describing an ideal fluid’s flow characteristics, and the basis for classical fluid mechanics, is Newtonian – with a linear relationship between shear stress (σ) and shear rate ( γ& ), and with zero stress at zero rate. This behaviour is analogous to Hookean behaviour for a solid (see Figure 2). However, many important fluids, such as polymers, blood, foods, cosmetics, and concretes, show non-Newtonian plastic behaviour, in which flow only initiates above some level of stress (i.e. yield stress), and once flow begins the relationship between shear stress and shear rate is linear or non-linear. The non-linear materials can be either shear-thinning or thickening, with or without yield stress. Materials with thickening behaviour are not very common, especially among particle suspensions.

Shear strain, ε [-]

Shear stress, τ [Pa]

Elasticity, G [Pa]

Hookean

Shear rate, γ [1/s]

Shear stress, σ [Pa]

Viscosity, η [Pa s]

Newtonian

•

Figure 2. The Hookean behaviour in which stress is proportional to strain versus the Newtonian behaviour in which stress is proportional to shear rate.



Some materials also have a time dependence due to a flocculated microstructure, where the flow characteristics are influenced by the shear history of a material. These types of materials are usually defined as thixotropic or anti-thixotropic. In general, equilibrium measurements are desired, and the time dependence is minimized by pre-shearing the material and destroying the sample’s structure in the test [13]-[15].

A suspension with a wide range of particle sizes comprises a complex balance of interparticle forces (i.e. interlocking, frictional, flocculation, colloidal, van der Waals, and electrostatic forces), which generates this time dependence and viscoplastic non-Newtonian behaviour [4][13][16]-[18].

Other factors which affect the flow behaviour are the temperature, pressure, and pH. It is generally desirable to maintain these at constant levels throughout an experimental procedure, and this is normally done. The effect of pressure can be neglected, as it is very small at atmospheric pressure [13].

The viscosity of different materials varies in a wide range (Table 2) as do the shear rates they may be subjected to in different industrial processes (Table 3). To be able to characterize a fluid, a proper model to mathematically describe the rheology must be selected, together with a suitable instrument for measurements under realistic conditions. In the literature, there are several theoretical models, instruments and measuring techniques proposed to evaluate the flow behaviour of non-Newtonian fluids.

Table 2. Viscosity of some familiar materials at room temperature [13][19].

Liquid Viscosity η [Pa·s]

Glass Asphalt Molten polymers SCC Syrup Mortar Olive oil Water Air

1040 108 103 102 10 1 10-1 10-3

10-5

Table 3. Typical range of shear rates of some familiar processes [13].

Process Shear rate γ& [s-1]

Lubricating Rubbing Spraying and brushing Mixing and stirring Chewing / swallowing Pumping / pipe flow Extruding Levelling Sedimentation

103 - 107 104 - 105 103 - 104 10 - 103 10 - 102 1 - 103 1 - 102 10-2 - 10-1 10-6 - 10-3

The rheological properties of cementitious materials (cement paste, mortar and concrete) are mainly influenced by the following factors [5][20]-[24]:

a) Water content.

b) Amount of filler and aggregate, its particle shape, surface texture, porosity, size, and size distribution, and consequently the particle packing and specific surface area.

c) Type and amount of cement and cement replacements

d) Amount of entrained air, and bubbles’ size.

e) Presence and efficiency of admixtures affecting the properties of adsorbed water layer and particle dispersion.



f) Temperature.

g) Degree of hydration.

h) Time elapsed after mixing.

i) Mixing procedure.

Here, the water content and specific surface area are the two single most important factors influencing the rheological properties of cementitious materials [5][22].

The objective of this chapter is to present, through a literature review, an introduction to the rheology of particle suspensions and the factors affecting it. Flow behaviours, rheological characteristics, models, and methods for measuring the rheology of fresh cementitious materials are in focus.

2.2 Particle suspensions Because suspensions of solid particles in a liquid generally behave as fluids, it is often useful to characterize their rheological behaviour [13]-[15]. The two main factors affecting the rheology of particle suspensions are the volume fraction of solid particles in the suspension, and the extent to which the particles are flocculated [20]. The effect of flocculation is most pronounced for colloidal particles (i.e. particles in size range 20µm - 1ηm). The forces are fairly weak, so by applying a stress the flocculated network can easily be broken and the suspension will start to flow. For very dilute suspensions, the interparticle forces are negligible, and Newton's law can be applied. When the effect of a small increase of the volume fraction of solids (φ) merely increases the viscosity (η), it can be defined by the Einstein equation as follows:

)5.21(c ϕηη ⋅−⋅= (Eq. 1)

where ηc is the viscosity of the suspending medium. However, the Einstein equation and Newtonian model are not suitable for fluids with a high concentration of particles, such as paste, mortar, and concrete. As the volume fraction is increased above a few percent, viscosity increases progressively. Several equations have been proposed to better describe the full relationship between volume fraction and viscosity, taking into consideration the maximum packing fraction of solids (φm) and the shape of the solids by applying a fitting parameter (x), where the Krieger-Dougherty relationship (Eq. 2) is commonly used for colloidal suspensions [20][25].

mxmc )/1( ϕϕϕηη ⋅−−⋅= (Eq. 2)

For highly concentrated colloidal suspensions with a mix of particles with different properties (size, shape and surface properties), such as cementitious materials, the viscosity tends to be more complex when described mathematically. For these types of materials there is a fine balance between the interparticle forces (e.g. van der Waals attraction, electrostatic and steric repulsion, interlocking, friction, etc.) causing a non-Newtonian viscoplastic flow behaviour [13][19].

The effect of w/c on paste viscosity is usually defined by the Krieger-Dougherty equation (Eq. 2), whereas the effect of w/c on yield stress (σy) can be described by the power law:

By A ϕσ ⋅= [Pa] (Eq. 3)



where φ is the volume fraction of solids, and A and B are a constants (typically in the range of A is 2–4). Another way to express the yield stress is to consider also interparticle forces that are responsible for flocculation. The effect of w/c on yield stress (σy) can then be calculated as:

'max2

B

y rA Φϕσ ⋅⋅= [Pa] (Eq. 4)

where r is the particle radius, and Φ’max is the maximum interparticle force (maximum slope of the pair potential) [25][26].

2.2.1 Interparticle forces Particle suspensions’ rheology is affected by macro-level factors like particle size, size distribution, shape, texture, density, water content, etc., and micro-level forces such as capillary, flocculation, colloidal and true solution forces [4][13][16]-[18]. The forces interacting between suspended particles of different sizes can be categorized as shown in Figure 3.

Mechanical Capillary Flocculation Colloidal Solution Forces: Size [mm]: 30 1 0.1 2·10-4 10-6 10-7

Repulsion:

Attraction:

Liquid-need:

Flowability:

Coherence:

Effect of particle forces

Figure 3. Illustration of how particles of a certain size group under forces (x axis) are acting in a suspension, in terms of coherence, flowability, need of liquid, and attraction/repulsion (y axis). The magnitude is only relative within each property [18].

At macro-level there are interlocking and friction forces, and at micro-level there is a fine balance between attractive and repulsive forces. The main attractive forces are collectively called the van der Waals force (sometimes called London dispersion force), a short-range electromagnetic and relatively weak force interacting between two bodies. The van der Waals force arises from the polarization of molecules or particles into dipoles. This includes forces that arise from fixed or angle-averaged dipoles (Keesom forces) and free or rotation dipoles (Debye forces) as well as shifts in electron cloud distribution (London forces) [4][27]. The main repulsive force is the force which the electrical double layer gives rise to, and which acts between charged bodies with a potential difference.

The net result of these attractive and repulsive electric forces of suspended particles creates an energy barrier as they approach each other [4]. The consequent potential energy, as a



function of interparticle distance (r), is illustrated in Figure 4. At a small separation distance (r=r0), a deep potential energy trough is encountered, where the net force is zero and the particles are at an equilibrium distance apart, causing the matrix to flocculate. To make r smaller or larger than ro a dispersing or compressing force must be applied. Depending on the balance of forces, a much less deep secondary minimum, with zero net force, can exist at a larger interparticle distance (r=r0´) [14][28]. At this point the flocculated particles can more easily be separated by an applied shear. Still, if this secondary minimum is deep enough, relatively strong flocs can be formed. This is considered to be the basis for the thixotropic behaviour of particle suspensions [29].

Since the ratio of surface to volume is inversely proportional to linear dimension, the effect of these attractive and repulsive forces also increases as the particles become smaller and their specific surface area increases.

Cementitious materials are highly concentrated composite systems with a wide range of particle sizes (from 10-7 to 30 mm for concrete), where the fresh mix comprises a complex balance of interparticle forces generating a suspension with viscoplastic non-Newtonian behaviour. All these interacting forces depend on the shape of the particles, their size, concentration, and surface properties, but also on the water content of the suspension [4][18][19][28]. For cement-based materials the balance between the flocculation and colloidal forces acting between particles in the size range from 0.1 mm down to 1 ηm is predominant.

Interparticle distance, r

-

Pote

ntia

l ene

rgy

+

Repulsing forces

r0

Primary minimum

Secondary minimum

r0’

Attracting forces

r

Figure 4. Illustration of the potential energy as a function of interparticle distance.

In a matrix of the smallest particles, good coherence is accompanied by good fluidity (see Figure 5). In a system with a wide range of particle sizes and all the forces working together, the smallest particles with their good flowability now cause sedimentation of the larger particles in the fluid matrix, and thus a bad coherence of the total mass. Hence, for optimum coherence with strong flocculation, flowability is poor. This flowability can be increased by adding more water as well as by increasing the amount of the smallest particles, but at the price of decreased coherence. Thus, there is a compromise between fluidity and coherence. Figure 5 illustrates how each group of particle sizes is affected by another, in terms of coherence, flowability and need for water [4][18][24].

Hence, for a given mortar or concrete mix, the flowability can be adjusted on macro-level by grading, shape and water content etc., and on the micro-level by the ratio between the flocculation and colloidal forces. Stronger colloidal forces and weak coherence generate better flowability, a smaller compaction energy need and a larger separation tendency [18].



Mechanical Capillary Flocculation Colloidal Solution Forces: Size [mm]: 30 1 0.1 2·10-4 10-6 10-7

Water-need:

Flowability:

Coherence:

Interacting forces of composite particle systems

Figure 5. When a suspension incorporates a mix of particle sizes, the forces are acting together. The figure illustrates how this balance of different size groups forces (x-axis) affects coherence, flowability and need for water (y-axis). The magnitude is only relative within each property [18].

To increase a fresh cementitious material’s flowability, regarding the interparticle forces, one can [4][18][24]:

Choose a better total weighted grading curve (more densely packed structure).

Choose more rounded and smoother particles (with smaller interlocking and surface area).

Increase the water content (which separates the particles).

Vibrate the material (which de-flocculates it temporarily).

Add de-flocculation agent with high zeta-potential (which increases the double-layer effect).

Add very fine particles (<10-5 mm) with positively charged zeta-potential, e.g. silica fume and fines in fly ash (colloidal repelling particles).

Add air-entraining agent (small air bubbles acting like ball-bearings).

Add a surface-active agent (giving electrostatic and/or steric hindrance).

Add admixtures that decrease the water’s surface tension (which decreases flocculation).

Add large molecules which are adsorbed onto the cement particles (giving steric hindrance).

There is also sometimes a need to increase the coherence in order to create robustness and segregation resistance; therefore flocculating or thickening admixtures are used (to de-charge the double layers or immobilize the water phase).



2.2.2 Flow characteristics

Suspensions and solutions, under steady shear flow conditions (equilibrium state), may exhibit certain behaviours over a limited range of shear rates. Additionally, some materials may exhibit more than one distinct behaviour over different shear rate regions of the flow curve. Several types of behaviour can be classified according to their characteristic shape (see Figure 6). The following classification system [30] covers the most frequently encountered flow types, divided in terms of Newtonian, general non-Newtonian, and viscoplastic non-Newtonian fluids:



•

σapp

σ0

σ0*

1

3b)

3c)

2a)

3a)

2b)

Figure 6. Flow types illustrated as curves and their characteristic shape.

1. Newtonian: Viscosity is constant with shear rate changes.

2. General non-Newtonian:

a) Shear-thickening: Viscosity increases continuously as shear rate grows.

b) Shear-thinning (pseudoplastic): Viscosity decreases continuously with shear rate.

3. Viscoplastic non-Newtonian fluids:

a) Shear-thinning with yield-response: Viscosity decreases continuously with shear rate once the apparent yield stress (σapp) has been exceeded.

b) Bingham plastic (ideal): Above the Bingham yield stress (σ0) the differential viscosity is constant and is called the plastic viscosity.

c) Bingham plastic (non-ideal): Above the apparent yield stress the viscosity decreases continuously and approaches a constant value with increasing shear rate. Extrapolation of the flow curve from the linear region (plastic region) to the stress axis gives the apparent Bingham yield stress ( *

0σ ) and the plastic viscosity.

Newtonian Fluids In the simplest type of flow, termed Newtonian, the material’s viscosity is constant and independent of the shear rate [13][14]. That is, when shear stress (σ) is plotted against shear rate ( γ& ) at a given temperature, the plot shows a straight line with a constant slope that is independent of shear rate (see Figure 6). This slope is called the Newtonian viscosity (η) of the fluid. The simplest constitutive equation is Newton’s law of viscosity:

γησ &⋅= [Pa] (Eq. 5)

From this it may be seen that a single point measurement, with a corresponding pair of values of σ and γ& , would serve to rheologically define a Newtonian fluid.

All gases and common liquids such as ethanol, water and most oils are Newtonian. Also, liquids with low molecular weight and solutions of substances with low molecular weight in liquids, e.g. aqueous solutions of sugar or salt, are usually Newtonian.



General non-Newtonian Fluids Any fluids that do not obey the Newtonian relationship between shear stress and shear rate are non-Newtonian. The subject of rheology is generally devoted to the study of the behaviour of such fluids. Suspensions of fine particles are usually non-Newtonian [13][14].

In the case of general non-Newtonian fluids, the slope of the shear stress versus shear rate plot is not constant. When the viscosity of a fluid decreases with increasing shear rate, the fluid is called shear-thinning (pseudoplastic). In the opposite case where the viscosity increases as the fluid is subjected to a high shear rate, the fluid is called shear-thickening. The term dilatant is commonly used for shear-thickening, although this usage is strictly incorrect. Dilatancy is a property that usually is associated with suspension of irregularly shaped particles, in which the liquid exhibits an increase in volume while being sheared.

Viscoplastic non-Newtonian Fluids The other important class of non-Newtonian fluids is a viscoplastic fluid. This is a fluid which will not flow when a very small shear stress is applied. The shear stress must exceed a critical value known as the yield stress for the fluid to flow [13][14]. For example, when opening a tube of toothpaste, an adequate force has to be applied in order to make the toothpaste start to flow. Therefore, viscoplastic fluids behave like solids when the applied shear stress is less than the yield stress. Once the applied shear stress exceeds the yield stress, the viscoplastic fluid flows just like a normal fluid. Examples of viscoplastic fluids are blood, paint, mayonnaise, toothpaste, grease, foods, and mortar and concrete.

Once the yield stress has been exceeded, the slope of the shear stress versus shear rate curve may be either constant or non-constant. If the fluid obeys a yield value and has a constant viscosity it is called Bingham-plastic, and if the viscosity is non-constant it is called shear-thinning and shear-thickening with yield response.

2.2.3 Rheological models By using a mathematical relationship, a fluid can be rheologically characterized by a small number of coefficients (e.g. viscosity and yield stress).

The models that are presented in this section are used to characterize the non-Newtonian behaviour of fluids in equilibrium, under steady shear flow conditions at constant pressure and temperature. A large number of models have been reported in the literature, but only those which are applicable to the study of the rheology of particle suspensions have been included [13]-[15][26][28][31]-[35].

In the literature, several theoretical models used to describe the non-Newtonian flow behaviour can be found. The most frequently used fundamental models are the Power-law, Bingham and Herschel-Bulkley.

These models are well suited to studying materials over a small shear range or where only a simple relationship is required. But many materials will start to deviate from these relationships over a sufficiently large shear range. To be able to give a more realistic prediction of flow over a wider range of conditions, as enhancements to the fundamental models, other relationships have been developed. More detailed description and alternative expressions can be found in the cited literature.

For mortar and concrete, the Bingham model is considered by many researchers to be the most useful. This can be explained by its simplicity (with only two parameters), and by the



fact that materials of this type, over a limited range of shear rates, correspond quite closely to the model [24][28]. However, it ought to be noted that there are no fluids which behave like an ideal Bingham material. This is only a generalized model to be used for non-ideal Bingham materials [36]. For cement paste, there are disagreements as to which model best describes the flow behaviour. These types of highly concentrated suspensions, incorporating a high ratio of colloidal particles (i.e. particles smaller than approx. 1µm), tend to show a more pseudo-plastic shear-thinning behaviour at equilibrium flow. In the literature, different models have been proposed but the Bingham model and the Herschel-Bulkley model tend to be best suited for cement pastes [25][28][34][37][38]. The Bingham model is better suited for pastes with high water-to-cement ratio (w/c) than for those with low w/c [25]. When superplasticizer is added, paste tends to show Newtonian behaviour at low concentrations and viscoplastic behaviour at high concentrations [26]. SCC, however, has been suggested to a have a shear-thickening behaviour with a yield, and the rheology parameters are to be defined by the Herschel-Bulkley model (with A>1) [39]. Others consider the apparent shear-thickening behaviour of SCC to be an artefact from the experimental setup, caused by a lack of steady state during the measurement, where the structure of flocculated colloidal particles has not been broken [40][41].

Typical values of yield stress and plastic viscosity for some cementitious materials, defined by the Bingham model, are shown in Table 4. For SCC, it cannot be at the lower limits of yield stress and viscosity simultaneously, since it would separate.

Table 4. Typical range of yield stress (σ0) and plastic viscosity (ηpl) for cement paste, mortar, “normal” concrete (NC and flowing NC), and self-compacting concrete (SCC), defined by the Bingham model [5] [19] [42].

Paste Mortar NC Flowing NC SCC 0σ [Pa]: 10−50 10−100 400−2000 100-400 10−100

plη [Pa s]: 0.01−1 1−5 50−100 20-100 20−150

Power-law (or Ostwald-de Waele) Many non-Newtonian materials undergo a simple increase or decrease in viscosity as the shear rate is increased. One of the most widely used forms of the general non-Newtonian constitutive relation is a power-law model, a two parameter model expressed as:

Aapp γσ &⋅= η [Pa] (Eq. 6)

where σ is the shear stress, ηapp is a non-Newtonian apparent viscosity defining the consistency of the fluid, and A is the power-law model constant indicating the degree of non-Newtonian behaviour (the greater the departure from unity, the more pronounced the non-Newtonian properties of the fluid).

The power-law model does not handle viscoelastic materials with a yield stress. If A<1, a shear-thinning fluid is obtained, which is characterized by a progressively decreasing apparent viscosity with increasing shear rate – whereas if A>1, it is a shear-thickening fluid, and when A=1 it is a Newtonian fluid. These three types of power-law models are illustrated in Figure 7. The power-law model is good for describing a material’s flow within a small range of shear rates, but fails to describe the viscosity of many non-Newtonian fluids in very low and very high shear rate regions (e.g. for suspensions, η goes to infinity at a very low shear rate and becomes constant at a very high shear rate), nor does the model handle viscoelastic materials with a yield.





•

1) Newtonian (or Power-law when A=1)

3b) Bingham (or Herschel-Bulkley when A=1)

2a) Power-law, A>1 (shear-thinning)

3a) Herschel-Bulkley, A<1 (shear-thinning with yield)

2b) Power-law, A<1 (shear-thinning)

σapp σ0


Viscosity, η [Pa s]

•

1) 3b)

2a)

3a) 2b)

Figure 7. Flow curves of Power-law, Bingham and Herschel-Bulkley models.

Bingham Some materials exhibit an infinite viscosity until a sufficiently high stress is applied to initiate flow. Above this stress the material then shows simple Newtonian flow. The simplest model covering these types of viscoplastic fluids with a yield response is the ideal Bingham model, and is expressed by the following two parameter equation:

γσσ &⋅+= pl0 η 0σσ > [Pa] (Eq. 7)

where ηpl is the model coefficient of plastic viscosity, and γ& is the shear rate. σ0 is the constant which is interpreted as yield stress, and which often is referred to as the Bingham yield stress or simply the yield stress of the material. The Bingham model can describe the viscosity characteristics of a fluid with yield stress whose viscosity is independent of shear rate, as shown in Figure 7. Therefore, the Bingham plastic model does not have the ability to handle the shear-thinning characteristics of general non-Newtonian fluids.

Many concentrated particle suspensions and colloidal systems, such as mortar and concrete, show Bingham behaviour at limited ranges of shear rate.

All rheological measurements presented in the appended papers (Paper I, II, III and VI), on both mortars and concretes, were performed with the Bingham model.

Herschel-Bulkley The Herschel-Bulkley model is a three-parameter model used to describe viscoplastic materials exhibiting a yield response with a shear-thinning relationship above the yield stress. It is a combination of Power-law and Bingham, and is expressed as:

Aappapp η γσσ &⋅+= appσσ > [Pa] (Eq. 8)

where σapp is the constant that is interpreted as apparent yield stress, ηapp is the model coefficient of apparent viscosity, γ& is the shear rate, and A is the model constant indicating the degree of non-Newtonian behaviour (the greater the departure from unity, the more



pronounced the non-Newtonian properties of the fluid). When A=1, Herschel-Bulkley is reduced to the equation for Bingham (see Figure 7). The Herschel-Bulkley model tends to give a more realistic prediction of flow over a wider range of conditions than the Bingham model. It is often applied to industrial fluids (e.g. biological fluids, food and cosmetics), where it is used for specifying conditions in the design of process plants.

2.2.4 Yield stress measurement Whether yield stress is a true material property or not is a controversial issue [14]. However, there is generally an acceptance of its practical usefulness in engineering design and operation of processes where handling and transport of industrial suspensions are involved. For example, the minimum pump pressure required to start, the resistance to initiating flow, the segregation of aggregates, and entrapment of air in cementitious materials are typical problems where the knowledge of the yield stress is essential.

Numerous techniques have been developed for determining the yield stress, both directly and indirectly [14][15][20], based on the general definition of the yield stress as the stress limit between flow and non-flow conditions. Indirect methods simply involve the extrapolation of shear stress vs. shear rate data to zero shear rate, with or without the help of a rheological model. Direct measurements generally rely on some independent assessment of yield stress as the critical shear stress at which the fluid yields or starts to flow.

Various techniques have been introduced for measuring the yield stress directly and independently of shear stress vs. shear rate data [14][15][20]. Although the general principle of the yield stress as the stress limit between flow and non-flow conditions is often used, the specific criterion employed for defining the yield stress seems to vary among these techniques. Constant shear stress rheometers provide the most direct method for making the measurement, where the stress is progressively increased, yield is reached, and flow initiates. Furthermore, each direct technique appears to have its own limitations and sensitivity, so that no single technique can be considered versatile or accurate enough to cover the whole range of yield stress and fluid characteristics.

The value obtained by the extrapolation of a flow curve is known as “apparent” (or “extrapolated”) yield stress, whereas yield stress measured directly, usually under a near-static condition, is termed a “static” (or “true”) yield value. In a strain-stress measurement, the static yield stress (σs) can be evaluated either from the point where the elastic response starts to diverge from its linearity, or at the peak stress just before the material’s elastic structure breaks and turns into a plastic state (see Figure 8) [29]. However, it can be problematic to find the exact point where the response starts to diverge from linearity, as it usually is not a distinct point. Moreover, in a strain-stress measurement the value of static yield stress depends on the applied rate of shear strain (or shear stress when measuring the strain).





Peak level

σs

σs

Elastic response

Time, t [sec]


Load Respons

Figure 8. Determination of static yield stress (σs), by a strain-stress measurement, either from the point where the elastic response starts to diverges from its linearity, or at the peak stress.

The indirect determination of the yield stress simply involves the extrapolation of experimental shear stress vs. shear rate data to zero rate (see Figure 9). The extrapolation may be performed graphically or numerically, or can be fitted to a suitable rheological model representing the fluid.

Direct data extrapolation is a procedure to extend the flow curve at low shear rates to zero shear rate, and take the shear stress intercept as the yield stress value. The technique is only relatively straightforward if the shear stress vs. shear rate data are linear and there is no plug flow. With non-linear flow curves, as shown in Figure 9, the data may have to be fitted to a polynomial equation followed by the extrapolation of the resulting curve fit to zero shear rate. The yield stress value obtained obviously depends on the lowest shear rate data available and used in the extrapolation. Thus, it is imperative that some checking procedure should be carried out to ascertain the reliability of the low shear rate data before extrapolation is made. Thixotropic materials with a high yield stress tend to provide non-steady shear flow and plug flow due to structural build-up at low shear rate.


Curve fit

Shear rate, γ [1/s] •

Bingham

σ0

σappx x

x x

x x

x x

x x x

Figure 9. Determination of apparent yield stress (σ0 and σapp) by extrapolation from a curve fit, and by applying a rheological model. In this illustration a 3rd-order polynomial curve fit and the Bingham model are applied to measurements in a flow curve.

A more convenient extrapolation technique is to approximate the experimental data with one of the viscoplastic flow models. Many researchers appear to prefer the Bingham model which postulates a linear relationship between shear stress and shear rate. However, since a large number of yield stress fluids, including suspensions, are not Bingham-plastic except at very high shear rates, the use of the Bingham plastic model can lead to unnecessary over-prediction of the yield stress, as shown in Figure 9 [14][43].

Extrapolation by means of the nonlinear Casson model can be used from a linear plot of γ& versus σ . The application of the Herschel-Bulkley model is less certain although

systematic procedures for determining the yield stress value and the other model parameters are available. Even with the most suitable model and appropriate technique, the yield stress value obtained cannot be regarded as an absolute material property because its



accuracy depends on the model used and the range and reliability of the experimental data available. Several studies [14] have shown that a given fluid can be described equally well by more than one model and, hence, can have different yield stress values.

It has been shown by many investigators that the empirical tests to some extent provide information about the concrete rheology. Slump has for a long time been considered to be a unique function of yield stress (Bingham), without influence by plastic viscosity [44]-[60]. More recently the horizontal flow spread, rather than the vertical slump, has been used to calculate the yield stress [24][61]-[63]. Usually these correlations concern concretes in a small range of flowability (e.g. HPC, SCC), and are not valid for other or wider ranges. Often the correlation is purely empirical, but there are also some physical models proposed to explain the relationship. The viscosity, on the other hand, is generally considered to correlate with empirical tests recording the flow time. For SCC, this connection is usually made with the slump flow spread time to 500 mm (T50) or V-funnel time [64]-[66].

In Paper I experimental results are presented, addressing the interaction between the slump flow spread, flow time (T50), yield stress (σ0) and plastic viscosity (ηpl). It was shown that the slump flow spread is not a unique function of yield stress, but rather a more complex function of both yield stress and viscosity. However, the spread proved to be much more closely related to the yield stress than to the viscosity, especially at high viscosity, whereas the T50 time was connected more equally with yield stress and viscosity. These relations are illustrated in Figure 10.

y = -124.65Ln(x) + 1071.82R2 = 0.89

200

400

600

800

1000

0 200 400 600Yield stress [Pa]

Slu

mp

flow

[mm

]

200

400

600

800

1000

0 50 100 150 200Viscosity [Pa s]

Slu

mp

flow

[mm

]

0

5

10

15

20

25

0 50 100 150Yield stress [Pa]

T50

[s]

0

5

10

15

20

25

0 50 100 150Viscosity [Pa s]

T50

[s]

Figure 10. Measures of slump flow values vs. rheological parameters, individually, for ~550 mixtures with a wide range of consistency.



The following approximate models for calculating the relative slump flow area (Rp) and the slump flow spread time T50 (in seconds) for flowable concretes (and mortars), over a wide range of consistencies, was proposed:

0pl

2

p )2ln()6/H(R

ση ⋅+≈ (Eq. 9) 1)D/D(R 2

0p −= (Eq. 10)

pl0002.050T ησ ⋅⋅≈ (Eq. 11)

where ηpl is the plastic viscosity, σ0 the yield stress, H the cone height (300 mm for Abram’s cone), D the measured spread diameter in millimetres, and D0 the flow cone bottom diameter (200 mm). The models are purely empirical and are deficient in units. They are only intended to roughly illustrate the interaction between the rheological parameters and slump flow values.

In the literature it is not generally agreed that slump flow is a unique measure of concrete yield stress [67]-[73]. It has been noted that the correlation is weak, or that relation between yield stress and slump flow, without taking viscosity into account, does not show a satisfactory fit. In Figure 11, typical values for SCC, presented by Nielsson and Wallevik [67], are plotted in the diagram from Paper I showing the correlation between the measures of slump flow values and rheological parameters.

0

30

60

90

120


Yie

ld s

tress

[Pa]

>700 mm

600-700 mm

500-600 mm

500-600 mm

600-700 mm

>700 mm

550 mm 650 mm 700 mm600 mm

Figure 11. The slump flow spread (mixtures with a spread <500 mm are excluded) and Bingham rheology parameters (yield stress and plastic viscosity) for SCC from Paper I, complemented with typical values for SCC from [67].

2.2.5 Time dependence Some materials have a reversible time dependence, where the flow characteristics are influenced by the shear history of a material. At rest the material regains its original state. These types of materials are usually defined as thixotropic or anti-thixotropic. If there is a reversible recovery of viscosity with time, it is referred to as thixotropy – and if the recovery results in a decrease in viscosity, it is termed anti- or negative thixotropy (sometimes referred to as rheopexy, although this is not the preferred term). The effects of thixotropy and anti-thixotropy are illustrated in Figure 12.

Cementitious materials generally shows time-dependent reversible flow behaviour, referred to as thixotropy. Other well-known examples of thixotropic materials are: yoghurt, ketchup, mayonnaise, wall paint, and clay.



The phenomenon of thixotropy rises from the microstructure of the matrix system [14], due to aggregation and flocculation of suspended particles and the time taking to change this microstructure. In such a system, interaction occurs between the particles as a result of the attraction due to van der Waals forces and repulsion due to electrostatic and steric effects. The stability of the structure depends upon the existence of a potential energy barrier which prevents the particles from approaching close to one another. In this way, the weak physical bonds between the particles give rise to aggregation. As the suspension is sheared, the weak physical bonds among particles are ruptured, and the network among them breaks down into separate aggregates, which can disintegrate further into smaller fragments (small aggregates or flocs), whereas if the suspension is at rest the particle aggregation can re-flocculate and form again. These shear-induced changes in microstructure take time and, if fully reversible, are referred to as thixotropy [4][14][19][32][74][75]. The time for structural breakdown is normally shorter than the time for structural build-up [29].

Shea

r rat

e, γ

[1/s

] S

hear

stre

ss, σ

[Pa]

•

Time, t [s]

Shea

r stre

ss, σ

[Pa]

Thixotropic

Anti-thixotropic

Load

Figure 12. When gradually ramping down, the structure breaks down or rebuilds, whether it is thixotropic or anti-thixotropic, and whether at high or low load. The figure illustrates a controlled rate; with controlled stress, the effect is the opposite.

It should be noted that these changes are purely time related and the materials flow characteristics need to be studied as well. It is possible that a material could be both thixotropic and shear-thickening or shear-thinning.

In order to characterize the rheology of cementitious materials and other time-dependent material, one has to be aware of these effects. It is necessary to study the time dependence of the material in order to design a conditioning regime. To be able to compare measurements in common references, it is important to take steps to pre-condition the sample. The best method for this purpose is to pre-shear the sample for a time sufficient to destroy any structure (flocculated colloidal particles) [41]. One way to do this is to start at a high shear rate for a sufficient time, followed by measuring stress while decreasing the strain rate from high to low. Another method is to initiate each measurement with a constant shear, applied for a time sufficient to reach equilibrium (within acceptance) [14][28]. At low strain rates a suspension may be very slow to reach equilibrium.

There are many methods proposed, based on rheological measurements, for how to characterize the time dependence. However, the quantification of thixotropy in absolute or fundamental terms is problematic, as it depends on the shear history of the material as well as the method used to measure and evaluate the thixotropy; hence the thixotropic value is often only relative. A commonly used technique to determine the thixotropy is to measure the time needed for a structural breakdown [28]. This is usually done by applying a constant shear rate (or stress), and recording the time for the measured shear stress (or rate) to reach an equilibrium state (within acceptance); see Figure 14. As the breakdown time depends on the size of the applied shear, the test can be performed at different specific



levels of rate (or stress), respectively. A method to characterize the material’s thixotropy with a single value [76][77], when using different levels of shear, is to determine the area between the curves of initial shear stress and equilibrium shear stress (see Figure 13).

Time, t [sec]




Thixotropic area [Pa/s]

Initial shear stress

Equilibrium shear stress

Initial shear stress

Equilibrium shear stress

Figure 13. Thixotropy represented by the evaluated area between the curves of initial shear stress and equilibrium shear stress Another popular method, especially for cementitious materials, is to quantify the time dependence by the area of a flow hysteresis. This area represents the energy needed to break down the structure [78]. In a typical rheometric test, shear stress or shear rate is ramped at a fixed speed up to a maximum value, then ramped back down at the same speed to the beginning (see Figure 14). In hysteresis, one flow curve lies above the other, forming a continuous loop whose internal area depends on the shear and thermal history of the material, and on how rapidly the stress or shear rate was ramped. Not all researchers agree that the area of a flow hysteresis in measurements of stress as a function of rate represents any material characteristics. The first method, where the stress is measured directly as a function of time at some strain rate, is considered to better reflect the thixotropic behaviour [28].

Thixotropicarea [Pa/s]

Time, t [s]

She

ar s

tress

, σ [P

a]


Shea

r rat

e, γ

[1/s

]

•

She

ar s

tress

, σ [P

a]

Time, t [s]

Shea

r rat

e, γ

[1/s

]

•

Shea

r stre

ss, σ

[Pa]

Thixotropic time [s]

Figure 14. Thixotropy quantified by the area of the flow hysteresis when the shear rate is ramped up and down, or alternatively by the time to steady-state (equilibrium) flow at constant shear rate. In Paper II, a thixotropic measurement on mortars was performed, using the latter method with the area of the flow hysteresis.

There is no general agreement that the thixotropy can be evaluated by measuring structural breakdown, nor that methods based on the calculated area represent any material characteristics [28]. The structural build-up at rest, rather than the breakdown, is proposed to better represent the material’s true thixotropic behaviour [29], and it is to be monitored without breaking down the structure. Furthermore, the elastic region of a viscoelastic material is proposed to represent the structure; hence it is to be characterized by a strain-



stress measurement where the structure corresponds to the peak stress (the static yield stress), just before the material’s elastic structure breaks and turns into a plastic state (see Figure 8). However, it can be problematic to avoid a structural breakdown when measuring this peak level. Moreover, the peak level is dependent on the applied rate of shear strain (or shear stress when measuring the strain).

Another non-destructive method to measure the material’s structure, represented by its elastic region, is by an oscillating rheology test. The deformations are small and kept within the linear viscoelastic region, so the structural build-up can be monitored over a time period without destroying the structure. In a typical oscillation test, illustrated in Figure 15, the applied periodic stress (σ) and the resulting strain (γ) can be described as:

)tcos(0 ⋅⋅= ωσσ [Pa] (Eq. 12) )tcos(0 δωγγ −⋅⋅= [-] (Eq. 13)

where σ0 is the stress amplitude, γ0 is the strain amplitude, ω is the angular frequency, t is the time, and δ is the phase lag (loss angle) [14][79]. For an ideal solid the phase angle will be δ=0, whereas for a Newtonian fluid with a purely viscous response it will be δ=2π (=90°). The materials rheological characteristics are described in terms of the shear storage modulus G’ (representing the elastic response of the material), and the shear loss modulus G” (representing the irreversible viscous response) can be calculated as:

)cos(/'G 00 δγσ ⋅= [Pa] (Eq. 14) )sin(/"G 00 δγσ ⋅= [Pa] (Eq. 15)

Time, t [sec]

Strain, γ [-] Phase angle, δ [rad]Stress, σ [Pa]

Resulting strain, γ [-]

Applied stress, σ [Pa]

σ0 γ0

Figure 15. Principle of oscillatory measurement of a viscoelastic fluid. To quantify thixotropy in absolute or fundamental terms is problematic, as the methods are dependent on the shear history of the material as well as the method used to measure and evaluate the thixotropy. Therefore, the thixotropic value will only be relative.

For suspensions subjected to chemical reactions, particles absorption, etc., causing an irreversible time dependence, it is important to exclude these effects from those which are reversible. Cementitious materials have a time-dependent irreversible behaviour, which is not to be confused with the time-dependent reversible flow behaviour (i.e. thixotropy). Hydration produces a type of hysteresis, where the stress during the subsequent measurement is higher than during the former measurement [26][28]. These changes are not reversible. The same phenomenon can be observed in the time-dependent water adsorption on particles’ surfaces and the absorption into the particles. A porous particle with a rough surface texture generates a more delayed adsorption than a smooth non-porous particle. Also the loss of water due to evaporation will have the same effect as the adsorption (and absorption). Another feature causing an irreversible behaviour is the loss of dispersing efficiency of the superplasticizer and other water reducing admixtures. A feature also producing a hysteresis, observed with cementitious materials, is an irreversible structural breakdown on first mixing. This is not thixotropy (it is not reversible), but it could easily be confused with thixotropy if the reversibility were not assessed. Usually these effects are classified as irreversible structural breakdown.



There is no general agreement that the true thixotropy can be evaluated by measures of structural breakdown, but rather methods using structural build-up at rest are to be used [29]. This is proposed to be characterized by a non-destructive method, for example by the peak stress (see Figure 8) in a strain-stress measurement [29]. To exclude the time-dependent irreversible behaviour from the reversible (thixotropic), the static yield stress is to be related to the dynamic yield stress (e.g. by Bingham, without thixotropic effects). The reversible part of the structure can then be evaluated from the area between the curves of static and dynamic yield stress versus time (see Figure 16).

Time, t [sec]

Yield stress, σo and σs [Pa]

Thixotropic area [Pa·s]

Static yield stress (σs)

Non-reversible structure

Dynamic yield stress (σo) Reversible

structure

Figure 16. Principle for de-termining the thixotropic area of the reversible (time-depending) structure.

2.2.6 Temperature dependence Temperature has a dramatic effect on the viscosity of any liquid, including particle suspension as a whole, and its suspending media. The viscosity of fluids is usually found to decrease with an increase in temperature [13], assuming that no physical/chemical changes are being induced by the applied heat energy. Liquids with higher viscosities generally are more temperature-sensitive. By using the Arrhenius relationship, a Newtonian liquid’s change in viscosity with temperature can be calculated approximately as:

TRE0

eA ⋅⋅=η (Eq. 16)

where A is a constant, E0 is the fluid activation energy in J/mol, R is the gas constant, and T is the absolute temperature in Kelvin. For water at 20°C, the equation can be approximately written as:

)T/2520(e5.5 −⋅=η (Eq. 17)

To maintain ±1% accuracy in the measurements of viscosity, at 20°C, the sample then must hold the temperature to ±0.3°C. For particles suspended in water, the viscosity rises with concentration and the temperature dependence increases. Consequently, precise control of the sample temperature is necessary to measure viscosity accurately.

2.3 Measuring techniques To establish the fundamental rheological measurements for non-Newtonian fluids, a rheometer is to be used. Numerous types of this instrument have been proposed for evaluation of the dynamic viscosity and yield stress of such materials. A rheometer is used to measure both viscosity and yield stress of a material, whereas a viscometer measures only the viscosity since these instruments do not measure the shear stress and shear rate simultaneously of a fluid at a given point. In order to determine real values of stress and rate (at the same point) an absolute instrument is needed, where the determination is made on the basis of a known mathematical description of flow occurring in the instrument and on the basis of its geometry.



The measuring techniques can be divided into four principal ones, as shown in Figure 17.

Shearing parallel plates

Rotating concentric cylinder

Rotating parallel plates

Capillary tube

Figure 17. Illustration of the four main rheological techniques.

Numerous more or less empirical techniques have also been proposed. The most basic type is the cup method, where the time that it takes to empty a cup with a defined hole in the bottom gives a measure of the viscosity. Other types use a penetration rod, ball, etc., where a falling objects with a defined density penetrates the samples. These types of kinematic viscometers are only appropriate for making comparative studies, as the shear rate is not fully controllable, and they are excluded from this study [14][13][28][80][15].

In order to determine the rheological properties of a non-Newtonian fluid, a multipoint flow curve has to be measured. A single point on the flow curve does not describe this material correctly. Two fluids with completely different rheological properties may generate the same value of viscosity at a given value of shear rate if the flow curves intersect at this point (see Figure 18).

A schematic illustration of the principal techniques for measuring the viscosity and yield stress of non-Newtonian fluids is shown in Figure 19.

She

ar s

tress

, σ [P

a] Herschel-Bulkley


Viscosity (Newton)

Bingham

Figure 18. Two flow curves intersecting, generating the same value of viscosity at a single measurement point.

Rheometers

Viscosity measurements

Drag flow Pressure driven flow

Rotating rheometer

Capillary tube rheometer

Concentric cylinder

Parallel plates

Yield stress measurements

Indirect methods Direct methods

Data extrapolation

Model extrapolation

Cone & plate

Figure 19. Rheometers’ principal techniques for determining the viscosity and yield stress. Most of these rheometers can produce viscosity measurements at a specified, constant shear rate or shear stress (normally a constant shear rate is applied). Therefore, in order to measure the viscosity over a range of shear rates, one needs to repeat the measurement by varying either the pressure in the reservoir tank of the capillary-tube rheometer, or the



rotating speed (or torque) of the cone or cup in the rotating rheometers. In addition, since these tests are to be performed at a stationary state, the capillary-tube rheometer is not very applicable to a thixotropic fluid like fresh cementitious materials where the samples are to be pre-sheared before measuring. For the yield measurement of viscoplastic materials, the indirect methods are preferred rather than direct methods, for practical reasons [43]. Thus, the direct methods will not be rigorously discussed.

In a rotational rheometer, the fluid sample is sheared as a result of the rotation of a cylinder or cone. The shearing occurs in a narrow gap between two surfaces, usually one rotating and the other stationary. Three frequently used geometries are: the coaxial (concentric) cylinder, the cone-and-plate, and parallel plates. The coaxial cylinder is the most commonly used [14].

In a capillary-tube rheometer, the fluid is made to flow through a capillary tube and the rates of flow under known pressure differences are measured. This method is commonly used in the process industry, where the apparatus can be installed in series with the flow.

Standardized measuring geometry for the different rheometers can be found in DIN 53018, DIN 53019, ISO 2884 and ISO 3219.

2.3.1 Rotational rheometer, concentric cylinder In a concentric (coaxial) cylinder system the inner cylinder is often referred to as a bob, and the external one as a cup. The shear rate is determined by geometrical dimensions and the speed of rotation. The shear stress is calculated from the torque and the geometrical dimensions. By changing the speed of the rotating element, different torques can be obtained (or different rotation when changing the torque), which are used for the determination of the shear stress / shear rate curve. Figure 20 shows a typical coaxial cylinder system that has a fluid confined within a narrow gap (Ri/Ro≥0.99) between the inner cylinder rotating at an angular velocity ( Ω ), and the stationary outer cylinder (normally it is the inner one that rotates).

With the concentric (coaxial) cylinder system, the shear profile over the gap will not be linear; thus, ambiguity arises over the best value to use for the effective mean radius (Ra) in the calculation of shear stress and shear rate. The shape of the profile is governed not only by the geometrical dimensions, but also by the flow characteristics of the fluid [15]. For a known flow characteristic an approximate solution can be found and the analysis of data will not involve differentiation calculations. E.g. for a Bingham fluid, the so-called Reiner Riwling equation can be used, provided that there is no plug flow over the profile and that the gap is sufficiently narrow.

H

Ri

Ra

Ro

Ω

gap

Figure 20. Principle of a con-centric cylinder rheometer. In this example the bob is rotating.

For particle suspensions, such as cementitious materials, there are well-established rules for the geometry of apparatus and sample to ensure that rheological measurements are reliable. The chief rules are that any dimension should be at least 10 times the size of the



largest particles (Ro-Ri≥10·Dmax) and that the ratio of outer to inner cylinder radius should be less than 1.2 [14][19][28]. It can be noted that this is not the narrow gap that is required to generate a known shear profile (Ro/Ri≤1.01), but in order to have manageable equipment the maximum ratio of radii is set to 1.2 without generating too large errors. E.g. if 0.5 mm particles and a narrow gap are to be used, the diameters of the inner and outer cylinders would have to be approximately 1 metre. Wide-gap (Ro/Ri>1.2) concentric cylinder rheometers exist, but the calculation of shear stress and rate requires broad assumptions concerning these parameters over the range of shear across the gap. Thus, small-gap (Ro/Ri≤1.2) rheometers are preferable because they do not require large assumptions, and they also decrease the risk of plug flow [20][26][28].

For a viscoelastic fluid, at low shear rate, there is always a risk of plug flow. If the shear stress falls below the yield value (see Figure 22). The velocity gradient over the profile will change, and there will be a part of the material without velocity. The gap where flow occurs will then be reduced to an indeterminable value, so the calculations of angular velocity and torque into shear rate and shear stress will not be valid. With a rotating inner cylinder, the plug of solid material will appear at the outer cylinder and at the inner cylinder when the outer is rotating. In a flow curve (shear rate versus stress), plug flow will be manifested as loss of shear stress in the low shear rate region. For a Bingham fluid, the measured values at low shear rates will diverge from the linear curve (similarly to the non-ideal Bingham-plastic behaviour in Figure 6, curve 3a); hence these values are to be excluded. The lower limit of the angular velocity (Ω), where all material is sheared in the gap, for a Bingham fluid, can be calculated as [28]:

[ ] pl02

io /1)R/R( ησΩ ⋅−= (Eq. 18)

The end-effects, where shear is generated at the bottom of the inner cylinder, can be tackled by entrapping air at the bottom of the inner cylinder if it has a concave shape, or else by making the end of the inner cylinder bottom into a cone-and-plate geometry and incorporating this in the equations. The wall-slip, where a lubricating slip-layer is formed at the surface of the inner cylinder (see Figure 23), can be avoided by grooving the surface or designing the bob as a vane (see Figure 21). However, a vane may introduce a complex flow (e.g. turbulence).

Figure 21. Photo showing cylinders (bob) with different geometries. From left: cone-and-plate, concave, grooved and vane.

For mortars and pastes, the grooved inner cylinder and the vane have been shown to provide stable measurements and equal results, whereas the smooth (conical and concave in Figure 21) gave deficient measurements with a shear-thinning behaviour (indicating wall-slip) and with a larger scatter.



Incompleteflow

Ro

Ri

Complete flow

σ < σ0 σ

γ ·

Shear stress

Shear rate (velocity)

Ω

”plug”

Figure 22. Illustration of plug flow, with a section of solid material at the outer cylinder (when the inner rotates), where the shear stress is below the yield value.

Ω

Lubricating slip-layer

Figure 23. Illustration of wall-slip, with a lubricating slippage layer of suspending media formed at the surface of the inner cylinder.

For cementitious materials, the concentric (coaxial) cylinder is the most commonly used system for rheological measurements, usually with some type of grooved or ribbed surface on the inner or outer cylinder (or both). Also the vane type is commonly used for cementitious materials. A well-known commercial apparatus using this system is the BML rheometer for concrete. In the appended papers, the rheological measurements on mortars and concretes were performed using a concentric system.

2.3.2 Rotational rheometer, cone-and-plate

The common feature of a cone-and-plate rheometer is that the fluid is sheared between a flat plate and a cone with a low angle (see Figure 24). The cone-and-plate system produces a flow in which the shear rate is very nearly uniform everywhere in the liquid, provided that the gap angle (α) is small. Typically, this angle is not to be larger than 4º [13][15][80]. For cementitious materials, the cone-and-plate system is seldom used, as the narrow gap close to the centre will make the particles cause interference.

α

R

Ω

Figure 24. Principle of a cone-and-plate rheometer.

2.3.3 Rotational rheometer, parallel-plates

The rotating parallel-plates rheometer is similar to the cone-and-plate system. The fluid is sheared between a stationary and a rotating flat plate (see Figure 25). In this system the gap between the plates can be adjusted to fit the fluid, which is preferable for particle suspensions. However, in contrast to the cone-and-plate system, the parallel plates do not provide a shear rate that is uniform over the radius, which consequently makes it inappropriate for non-Newtonian fluids.

h

R

Ω

Figure 25. Principle of a rotating rheometer with parallel-plates.



A well-known commercial apparatus using the rotating parallel-plates system is the BTRHEOM rheometer for concrete.

2.3.4 Capillary-tube rheometer In a capillary-tube rheometer (see Figure 26), the fluid is forced through a cylindrical capillary tube with a smooth inner surface. The flow parameters have to be chosen in such a way that the flow may be regarded as steady-state, isothermal, and laminar. The ratio of the capillary length (L) to its inner radius is to be large so the end-effects occurring in the entrance and exit regions of the capillary tube can be neglected. Knowing the dimensions of the capillary tube (i.e., its inner diameter and length), one can determine the functional dependence between the volumetric flow rate and the pressure drop (∆P) due to friction. By applying various values of pressure drop (or flow rate), the flow curve of the fluid can be determined.

The principle of a capillary-tube rheometer is based on the so called Hagen-Poiseuille equation, valid for Newtonian fluids. Since the viscosity of a Newtonian fluid does not vary with flow or shear, only one measurement at any flow velocity is needed. However, for non-Newtonian fluids, it is more complicated as the viscosity may vary with flow velocity (or shear rate). Consequently, one needs to vary the pressure in the reservoir in order to change the shear rate, a time-consuming procedure. After each run, the pressure should be reset to a new value to obtain the relation between flow rate and pressure drop.

2·R

∆P

L

Figure 26. Principle of capillary-tube rheometer.

2.4 Concluding remarks Some concluding remarks concerning the rheology of fresh cementitious materials (paste, mortar and concrete) can be made as follows:

The most important factors affecting the rheological properties are the water content, and the particles’ specific surface area and packing. Other influential properties are the particle size distribution, shape, and surface texture.

Cementitious materials are widely recognized to show a viscoplastic non-Newtonian behaviour due to the complex balance of interparticle forces. The flow characteristic mainly depends on concentration of particles and on the extent to which content particles are flocculated. The particles cover a very broad size range, from micron-sized cement and silica fume particles to centimetre-sized aggregates, but flocculation is only important for the finer particles.

It is important to be aware of the time-dependent, reversible and irreversible, phenomena that these materials show (e.g. thixotropy, hydration, mixing, adsorption),



and rheological tests are to be performed at steady shear (equilibrium conditions). When evaluating the development of thixotropy over time, the irreversible part must be measured and excluded from the reversible.

In order to determine the rheological properties of a non-Newtonian/viscoplastic fluid, a multipoint flow curve has to be measured; hence a single point on the flow curve does not describe this material completely. It is preferable to use a controlled multipoint shear rate technique.

The Bingham model is the most frequently used, and probably the best suited, model to describe the rheology of these types of materials. It is simple (with only two parameters), and over a limited range of shear rates it is confirmed quite well. For highly concentrated colloidal systems, such as cement paste with low w/c, the Herschel-Bulkley model might be more suitable.

The slump flow spread and T50 time are not a unique function of yield stress and viscosity respectively, but rather a more complex function of both where neither yield stress nor viscosity can be neglected. The spread has proved to be more closely connected with the yield stress than the viscosity (~3 times), especially at high viscosity, whereas the T50 time was connected more equally with yield stress and viscosity.

C h a p t e r 3 – E a r l y - a g e d e f o r m a t i o n


3 EARLY-AGE DEFORMATION

3.1 Introduction When a porous body loses water to the ambience, contraction will take place. For cementitious materials there will also be an additional shrinkage caused by the chemical reactions (hydration, carbonation, etc.). Since the rigidity, as well as the effect of different chemical reactions, changes with time, it is convenient to distinguish between early-age and long-term behaviour, where early age is limited to 24 hours from mixing and long-term refers to the time beyond. However, there is not a well-defined and distinct transition but rather a continuous transformation from a plastic to a rigid state. Shrinkage of a cementitious material will start as soon water is added, due to the chemical reaction and also due to the loss of water by evaporation (if not prevented). In the initial fresh state, when the concrete is still a fluid, the volumetric shrinkage will, fully or partly, result in a vertical displacement (settlement) [81][82]. This deformation does not result in any stresses and is often ignored in the design of concrete structures, since it is considered to be less than the subsequent deformations or is considered not important. Moreover, in the fresh state, shrinkage measurements are difficult to perform without involving the effect of gravity (bleeding, segregation, etc.) which, in addition, is not isotropic.

Yet in order to elucidate the mechanism of early-age deformations, and in view of the fact that they are greatest up to an age of 12 hours, these deformations should be investigated and may have to be taken into account. The governing mechanisms of the total shrinkage, distinguishing between early age and long term, are illustrated in Figure 27. It ought to be noted that there are also structural and environmental factors that simultaneously can cause expansions, e.g. alkali silica reactions, delayed ettringite formation, expansive cements, sulphate attack, thermal changes and reabsorption of water. There is generally an interplay between the mechanisms of contraction and expansion, but often the contracting part is predominant. However, the mechanisms of expansion have not been considered in this work. Neither are the long-term deformations further explored.

Total shrinkage

Early age (<24 hours) shrinkage Long term (>24 hours) shrinkage

Plastic shrinkage Thermal dilatation Drying shrinkage

Evaporation Autogenous shrinkage External absorption Carbonation shrinkage

Chemical shrinkage Hydration

Figure 27. Illustration of the driving mechanisms of the total shrinkage, not considering the mechanisms of expansion, or including the structural restrain.



Shrinkage, especially at early age, has become of great concern, due to recurrent problems with cracking in concrete structures. Conditions such as reduced maximum aggregate size, increased amount of fines, presence of retarding admixtures, increased binder content, and deficient covering and curing all contribute to these problems. In recent research there has been a growing interest in the area of autogenous and plastic shrinkage, as well as how to control and counteract the early-age shrinkage. A “hot topic” today is e.g. the use of superabsorbent polymer for internal curing [83]-[88].

There are two main driving forces for early-age shrinkage [81]: (1) evaporation shrinkage (or early-age drying shrinkage) due to loss of water from the fresh concrete, and (2) shrinkage due to hydration and chemical reactions producing autogenous shrinkage. Yet under sealed conditions, if evaporation is prevented, the early-age shrinkage is a result of the autogenous shrinkage. However, temperature dilation may also contribute (e.g. when cooling the surface), as well as the external suction of water (absorption) by the formwork material or by the sub-base. Accordingly, cracking in the initial phase is a product of several factors. Thus, it is important to separate the mechanisms into those that are dependent on environmental conditions (e.g. evaporation of water from the surface) and those that occur without moisture transfer to the ambient and which are associated with the autogenous deformation of the material.

During the early age, here defined as the firs 24 hours from mixing, the concrete undergoes three more or less distinct structural phases [89]-[91]:

1. Plastic: a liquid state where the concrete is plastic with viscoelastic behaviour, and is still workable.

2. Semiplastic: after initial setting, the concrete will undergo a transition stage of early stiffening from fluidity to rigidity, where a self-supporting skeleton starts to form.

3. Rigid: at the final hardening stage, initiated by the point of final setting, the concrete’s strength starts to develop from the continuing hydration.

Water

Cement grains

Solid grains (aggregate/filler)

1. Plastic

2. Semiplastic

3. Rigid

Solid hydrates

Gas voids/pores

Figure 28. Illustration of structural development.

The term “plastic shrinkage” refers to the periods (1 and 2) until the point of final set [22][23][89]. However, neither the initial nor the final set is a distinct and well-defined physical state, but merely a gradual change of structure due to hydration. The points of setting are often arbitrarily chosen, referring to measures from empirical methods. As the point of final set is arbitrarily determined, so is the period of plastic shrinkage. However, this period is often simply given a general time (e.g. 12 or 24 hours). Commonly used methods for determination of setting (initial and final) are the Vicat needle method (EN 196-3) and the penetration resistance method (ASTM C403). There are also ultrasonic techniques proposed for better prediction of setting time [92]-[94].



3.2 Plastic shrinkage cracking Plastic shrinkage cracking is not a physical deformation, but an effect of a rather complex sequence of overlapping deformation mechanisms. The main underlying mechanisms for early-age deformation follow this section.

Plastic shrinkage is defined in the literature as the shrinkage of fresh concrete, exposed to drying, that takes place during the time to final set, while the concrete is “plastic” [23]. The duration is usually short and ends when the concrete has reached its final set. If the shrinkage is non-uniform, or if there is restraint, tensile stresses develops which may result in cracking as the concrete has a poorly developed tensile strength and strain [95]; see Figure 29. These cracks are usually referred to as “plastic shrinkage cracks”. The prevalent theory is that plastic shrinkage is directly caused by evaporation of water from the concrete surface; i.e. if the amount of water lost to the ambient exceeds the amount of water brought to the surface by bleeding, and if it is large, plastic shrinkage cracking may occur [22][23]. Care is to be taken to protect the surface against drying, although experience in the use of concretes with low w/c, or with silica added, has revealed that severe cracking may occur in spite of proper protection (curing membrane, plastic sheets, water spray, etc.), testifying to a large autogenous shrinkage [10][96]. Hence, most likely the cracking caused by plastic shrinkage is also a result of the autogenous shrinkage and thermal dilatation. Moreover, it can be difficult to distinguish between the plastic shrinkage cracks and cracks caused by thermal dilation, vertical displacement (settlement), early-age evaporation or autogenous shrinkage. Generally, cracks due to autogenous shrinkage develop uniformly through the concrete member, whereas cracks from evaporation are initiated at the outside surface and progress inward. For concretes where the paste generates a large chemical shrinkage (e.g. low w/c, addition of silica fume), the autogenous part of the plastic shrinkage can be significant, whereas when the chemical shrinkage is small (e.g. high w/c) the shrinkage due to evaporation tends to be predominant [91].

The environmental factors that promote early-age cracks are wind speed, air temperature, solar radiation and the relative humidity. These factors, together with the temperature of the concrete, govern the rate of evaporation from the concrete (see section 3.3.6). It is generally suggested that if the evaporation exceeds 1.0 kg/m2/h there is a significant risk of cracking, but cracks may occur also at evaporation rates lower than 0.5 kg/m2/h. Moreover, for low w/c or w/b concretes cracking may develop in spite of proper protection. It has therefore been suggested that if the early-age shrinkage magnitude exceeds 1000 µm/m, there is a high risk of cracking [97][98]. This corresponds well to the American Concrete Institute guidelines (ACI 209-92) where it is stated that cracks may occur for shrinkage of about 400–1000 µm/m. Moreover, this seems to be consistent with the actual tensile strain capacity of the concrete when cracking usually occurs; see Figure 29.

Age [hours] at 20ºC

Tens

ile st

rain

at f

ailu

re, ε

tu, [

10-6

m/m

]

1 10 10010

100

1,000

10,000

Figure 29. Relationship between tensile strain capacity and age of concrete, based on Kasai et al. [95].



3.3 Mechanisms of early-age deformation 3.3.1 Hydration and development of microstructure The hydration of Portland cement is a complex sequence of overlapping chemical reactions between clinker components, calcium sulphate and water, leading to setting and hardening. The process that occur can be divided in five stages as follows [22][26]; see Figure 30. As soon as C3S (in the cement) and water come in contact, reactions start to occur, releasing calcium and hydroxyl ions into the solution, generating an outburst of heat (Stage I). After this initial stage an induction period (or dormant period) is entered (Stage II), where the dissolution continues and pH reaches a high value (~12.5). During this induction period not much hydration takes place, but this does not mean that the paste is “dormant” with respect to volume changes. After a certain critical value of calcium and hydroxide ions is reached, there is a rapid crystallization of CH (calcium hydroxide) and C-S-H (calcium silicate hydrate) followed by a rapid reaction (Stage III). In this acceleration stage there is an onset of transformation to rigidity, with the initial set somewhat after the transition from Stage II while the final set occurs just before the peak between Stages III and IV. In the following stages (IV and V), there is a continuous formation of hydration products, yet the rate of reaction is slowing down. In the final stage (V) there is only a slow formation of products and the reaction is diffusion-controlled.

The degree of hydration refers to the amount of cement reacted relative to the total amount of cement. To obtain full hydration in a closed cement-paste system, the water and space needed corresponds to a w/c ratio of 0.42 [100]. However, this level can vary from 0.36 to 0.48, depending on cement type [8]. At lower w/c ratio, hydration stops due to lack of water and the remaining cement will be unhydrated. E.g. at w/c 0.21 only 50% of the cement will be fully hydrated [4]. However, if water is allowed to penetrate into an open system of hardening cement paste, the w/c ratio for full hydration is reduced to approximately 0.36.

I II III IV V

Degree of hydration

Rat

e of

hea

t evo

lutio

n Dissolution: ettringite formation

Initial set

Final set

Induction period: increase in Ca2+ concentration

Rapid formation of C-S-H and CH

Formation of monosulfate

Diffusion-controlled reactions

Time of hydration

~2% ~15% ~30%

~4 hours ~10 minutes ~12 hours ~24 hours

Figure 30. Schematic representation of heat evolution during hydration of cement and water, based on Gartner et al. [101]. In Figure 31, the products associated with the stages of cement hydration and the change of porosity and permeability can be seen. The setting process is the consequence of a change from a concentrated suspension of flocculated particles to a viscoelastic skeletal solid capable of supporting an applied stress; see Figure 32. Setting is controlled primarily by the hydration of C3S and occurs when the induction period is terminated by a rapid hydration of C3S leading to a fast temperature rise of the concrete (Stage III) [22].



As the hydration proceeds, capillary pores are gradually emptied and hydration products progressively fill the void spaces in the cement paste. As the pores will be smaller, or consumed, the total volume of pores (porosity) will decrease. This leads to decreased permeability and to increased strength and durability of the cement paste [90]. The pores in the paste are typically in the size range of 2·10-9 to 1·10-5 metre, where the smallest are distinguished as gel pores and the larger as capillary pores [99]. Larger pores are associated with air voids.

0 5 30 1 2 6 1 2 7 28 90

minutes hours days

C-S-H (short fibres)

CH

C4(AF)H13

Monosulphate

Ettringite

C-S-H (long fibres)

Time of hydration V III & IV I & II

Permeability

Porosity

Figure 31. Products associated with the stages of hydration, based on Locher et al. [102].

However, not only the w/c ratio will influence the hydration, but also a number of parameters such as: the composition and fineness of the cement (the main clinker components); addition of supplementary materials (such as fly ash and silica); admixtures (super plasticizer, accelerator, retarder, etc.); and the temperature.

Formation of a layer of ettringite or ‘AFt’ and ‘CH’. Possible particle agglomeration

Nucleation and early formation of ‘C-S-H’. Continued formation of ‘AFt’. Particle agglomeration

First minutes (Stage I): wetting and mixing

Induction period (Stage II):agitation, transport, placing, and finishing

Rapid formation of ‘C-S-H’ leading to solidification and creation of capillary pores

Acceleration period (Stage III):setting and early hardening

Concentrated suspension of flocculated particles

water

cement

Gel

AFt and CH

Outer C-S-H

AFt Outer C-S-H C-S-H needles

Figure 32. Schematic representation of the early microstructural development [91].

3.3.2 Chemical shrinkage In the literature, chemical shrinkage is also referred to as Le Chatelier contraction, named after the French scientist who in 1900 first observed the phenomenon [90]. Chemical shrinkage is the result of the reactions between cement and water, which lead to a volume reduction; the reactions of reactive fillers (pozzolanic) are also considered. The basic principle of chemical shrinkage is that the volume occupied by the hydration products is less than the sum of the volumes of the constituents (cement, pozzolan and water) before hydration [82]. According to Powers & Brownyard [100], when water reacts with the



cement ~25% of its initial volume is lost, and 1 gram of cement will consume 0.23-0.25 grams of water when fully hydrated. Hence, the chemical shrinkage amounts ~6 ml per 100 gram of cement reacted.

The basic reactions of cement clinker are generally defined by four reactions of C3S, C2S, C3A and C4AF. Each of these reactions requires water, is exothermic, and results in a different decreased volume of the reaction products [97]. The contribution from each of these four phases to the chemical shrinkage is given in Table 5.

Chemical shrinkage is a microscopic internal volume reduction, not considering the air in the capillary pores (porosity) [82]. At the stage where the paste is liquid-like, without a restraining skeleton and empty pores, the chemical shrinkage will be equal to the external bulk deformation [103][104]. However, at the stage when the self-supporting skeleton of hydration products starts to form (initial set) the external and internal chemical shrinkage will diverge, since the rigidity of the paste restrains the volume change [90]. This external deformation (for a sealed system) is referred to as autogenous shrinkage (see Figure 33).

Table 5. Approximate chemical shrinkage for the individual reacted phases in cement [89][103].

Vcs [ml/100g]

C3A 17.85 C4AF 11.13 C3S 5.32 C2S 4.00

Empty pores

I. Placing/casting

Chemical shrinkage

Chemical shrinkage

Autogenous shrinkage

II. Initial set

III. Hardening

WaterCementHydration products

Shrin

kage

Time

Chemical shrinkage

Autogenous shrinkage

Empty pores

Initial set

I. II. III. Figure 33. Principle of autogenous and chemical shrinkage for a cement paste under sealed conditions. As mentioned, the chemical shrinkage begins immediately when mixing cement and water and the rate of shrinkage is highest during the first hours. Chemical shrinkage is suggested to be almost linearly proportional to the degree of hydration (except for the very initial hydration) [105][106]. It has also been shown that the chemical shrinkage correlates linearly with the strength development, once the paste has set [107]. Factors such as temperature, w/c, cement fineness, C3A and C4AF content, and the efficiency of cement dispersion will have a significant effect on the rate [23]. Also additives such as chemical and mineral admixtures and pozzolans will influence the chemical shrinkage [22].



3.3.3 Autogenous shrinkage Autogenous shrinkage was first described by Lyman in 1934 [112] as a factor contributing to the total shrinkage, yet difficult to access. Davis [113] reported in 1940 measurements of autogenous shrinkage in the range 50-100 µm/m after 5 years of hardening. Compared with the thermal deformations and drying shrinkage this was considered to be relatively small, which is why historically little attention has been paid to the autogenous shrinkage [114]. However, many concretes today tends to develop a substantial autogenous shrinkage, due to the usage of chemical admixtures, low w/c, and high content of binder and filler material (e.g. high strength and self-compacting concrete).

Autogenous deformation may be defined as the unrestrained external bulk deformation taking place under isothermal and sealed conditions [104][108]., in other words with (1) no external forces or restraint acting on the material, (2) no exchange of matter with the surroundings (e.g. no evaporation), and (3) no variation in the temperature.

In the literature, a large number of terms for autogenous shrinkage can be found, such as: bulk shrinkage, external volume change, total or external chemical shrinkage, hydration shrinkage, and self-desiccation shrinkage [89][104][109]. The deformation may be either a shrinkage or an expansion (in length or volume). In this work, shrinkage is defined as a negative strain and expansion as a positive strain. Mostly, autogenous deformation is shrinkage, but any re-absorption of bleed water and/or absorbed water in the aggregates, as well as some early chemical reactions (e.g. ettringite formation), may cause an autogenous swelling [110].

The autogenous shrinkage originates from the chemical shrinkage, though it also involves the effect of the internal restraining structure and the contracting empty capillary pores. Whether the macroscopic bleeding (and re-absorption) is to be taken in account, or whether it only concerns homogeneous materials, is not clear [104].

As long as the concrete is fluid, autogenous shrinkage is considered to be equal to the chemical shrinkage. But once the self-supporting skeleton starts to form, the autogenous will diverge from the chemical [81][103][108][114][115]. Autogenous shrinkage before setting is not necessarily isotropic. It may be converted, fully or partly, to a vertical displacement. In a horizontal beam or slab, these volume changes will show up as vertical settlement without any linear horizontal deformation [105][116]. This part of the autogenous shrinkage is therefore sometimes referred to as setting shrinkage [111]. However, at the stage where the concrete starts to behave like a visco-elastic material, with a rigid self-supporting skeleton, the deformation of the specimen will become isotropic, and a linear contraction can be observed [81][103][117]. From being equal, the rate of autogenous shrinkage will be approximately 1/10 of the rate of chemical shrinkage [110].

There is no general agreement on whether the autogenous deformation is to be referred to the full period of hardening starting from water addition [117][118] or to the period in solid state from the time where a structure starts to form i.e. initial or final set. The latter point of set is sometimes referred to as the time zero (t0) [119]-[122]. Neither is there general agreement on how to define or measure the point of set. However, as this point is not a well-defined physical state but rather a continuous transformation from a liquid to a solid state [104], and as the shrinkage is highest up to an age of 12 hours [118], it is suggested that autogenous measurement is to be started as soon as possible after mixing. Moreover, autogenous deformation refers to the deformation under autogenous conditions (unstrained, sealed and isothermal), and does not refer to whether the system has set or not [104]. In literature the reported magnitude of autogenous deformation varies enormously,



which reflects the difficulties in the interpretation of the results based on different techniques [81][111][123].

The pattern of autogenous deformation, exemplified in Figure 34, comprises three distinct stages which can be defined as: (1) plastic, (2) semiplastic and (3) rigid, separated by the time to initial and final set [91]. In literature, other terms for these stages can be found, e.g. liquid / skeleton formation / hardening [89]. Also the four stages: initial / induction / acceleration / deacceleration, connected with the chemical reactions, can be found [125]. It has been suggested [81] that, when measuring the autogenous shrinkage, the setting (initial and final) will be manifested as a change of the slope of the deformation. Initially the autogenous shrinkage originates from the chemical shrinkage, whereas once the concrete sets the development of a contracting capillary pore pressure will contribute to the subsequent deformation (see Figure 34) [116].

~12 Time (from mix) [hours]

0 ~6 ~18

Def

orm

atio

n, P

ress

ure

& Te

mp

Plastic Rigid

~24

Semiplastic

Pore pressure (measured)

Self-supporting skeleton starts to form

Initial set

Final set

Autogenous deformation (measured)

Chemical shrinkage (assumed)

Temperature (measured)

Figure 34. Illustration of early-age autogenous deformation and the corresponding development of temperature, capillary pore pressure (based on measurements of a SCC with w/c 0.45, from [91]). There are numerous factors which have an impact on the autogenous shrinkage. Factors increasing the chemical shrinkage need not cause a larger autogenous shrinkage, since some of these will shorten the time to set and/or generate a more rigid restraining structure. E.g. the effect of cement fineness, content of C3A and C4AF, and temperature is not a matter of course. In addition to the chemical shrinkage, the autogenous shrinkage will be influenced by the internal restraint from aggregates. However, when the rate of chemical shrinkage is high, so is the rate of autogenous shrinkage, especially at early age. The autogenous shrinkage tends to increase with [23][91]:

o high binder content, o low w/c o low content of aggregate (especially the coarse fraction), o addition of silica, and o presence of chemical admixtures with a retarding effect.

The autogenous shrinkage may also be reduced by addition of accelerator, shrinkage-reducing admixture, or internal curing (saturated aggregates, superabsorbent polymer). The effect of different mix designs on the autogenous deformation is exemplified in Figure 35.



-1 200

-800

-400

00 6 12 18 24

Time [hours]

Aut

ogen

ous

defo

rmat

ion

[10-6

m/m

]

w/c 0.38

w/c 0.45

w/c 0.67

w/c 0.55

-800

-600

-400

-200

00 6 12 18 24

Time [hours]A

utog

enou

s de

form

atio

n [1

0-6m

/m]

0% Silica5% Silica

10% Silica

-600

-400

-200

00 6 12 18 24

Time [hours]

Aut

ogen

ous

defo

rmat

ion

[10-6

m/m

]

50% (8-16)

40% (8-16)

30% (8-16)

20% (8-16)

(a) (b) (c)

-600

-400

-200

00 6 12 18 24

Time [hours]

Aut

ogen

ous

defo

rmat

ion

[10-6

m/m

]

0.6% SP

0.8% SP

1.0% SP

-600

-400

-200

00 6 12 18 24

Time [hours]

Aut

ogen

ous

defo

rmat

ion

[10-6

m/m

]

2% SRA

1% SRA0% SRA

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Aut

ogen

ous

defo

rmat

ion

[10-6

m/m

]

Accelerator

Ref.Retarder

(d) (e) (f)

Figure 35. Measures of autogenous deformation on SCC (from [124]) and the effect of (a) w/c, (b) silica, (c) content of coarse aggregate of total aggregate, (d) superplasticizer dosage, (e) shrinkage-reducing admixture, and (f) accelerator and retarder.

3.3.4 Capillary pore pressure

When a concrete dries out due to evaporation, if the rate of evaporation exceeds the rate of free internal water transported to the surface, the concrete surface will dry out. This leads to the formation of numerous water menisci (with a curved surface) between the particles near the surface; see Figure 36.

qev

Water

Particle

γ

Pw

Pa γ

Air

Figure 36. Illustration of the formation of menisci between particles near the surface due to evaporation.

Since the fluid pressure is less on the convex side of a meniscus than on the concave side, the creation of water menisci results in a negative pressure in the pore water and contraction forces between particles [4]. Consequently, the average distance between the particles tends to be reduced and the paste contracts [9]. These contracting capillary forces are in inverse ratio to the meniscus radius, and hence the capillary tension stresses increase with decreasing pore sizes and interparticle spaces (e.g. with low w/c, high cement



fineness, additions of fines, etc.). For a concrete where evaporation is prevented, a negative capillary pressure will also start to develop, but only once the hydration commences and the concrete sets [89]. As a solid skeleton starts to form the chemical shrinkage is not totally transformed into external volume change and, if the water supply is restricted, empty pores will be formed inside the paste, and contracting water meniscus are created (see Figure 33 and Figure 34). At the stage where the concrete starts to set (initial set) both the capillary pore pressure and the temperature reaches an accelerating phase (see Figure 34 and Figure 37), which indicates that the dormant period is ended and that the cement hydration accelerates [91]. Initially (before setting), when the concrete is liquid-like, without restraining skeleton and empty pores, the pore pressure is zero [116]. According to Wittman [9] the action of capillary pore pressure the cause of plastic shrinkage.

Typical development of pore pressure is shown in Figure 37. As can be seen, the pore pressure develops more rapidly in an evaporating concrete than in a sealed one. The accelerating phase of pore pressure starts at 5-6 hours from mixing for a sealed concrete, whereas for an evaporating one it accelerates at 3-4 hours [91]. It can also be noted (in Figure 37) that the initial cracking, due to plastic shrinkage, which coincides with the accelerating phase of the pore pressure (3-4 hours from mixing). Furthermore, a pressure close to the exposed surface is ahead of a lower pressure. However, for SCC’s with a high w/c (>0.55) these differences are smaller, and for standard concrete the pressure at the different distances from the surface will develop almost equally in time [91]. This can be explained by the concrete’s ability to transport water to the surface in relation to the evaporation [116]. Thus, for a denser and less permeable matrix the internal transport of water to the area close to the surface, having a larger capillary tension (due to evaporation), will be restricted. Whereas for a more porous and permeable matrix the balancing of the differences in pore pressure, and the transport water to the surface, will be faster and the pressure will be more uniform through the section.

0 5 10 15 20Time (from mix) [hour]

Max

Min

Pore pressure

Temp.

Autogenousdeformation

Initial set

Final set

0 5 10 15 20

Time (from mix) [hour]

Max

MinInitial crack

Temperature

Strain

Pore pressure (high)Pore pressure (low)

Evaporation

Figure 37. Typical measurementss of pore pressure and temperature development, for a SCC with w/c 0.55. To the left, pore pressure and temperature together with the autogenous deformation, and to the right with the strain and evaporation (ring test). The axes have been normalized against maximum and minimum values. The capillary pore pressure (Pc) that is generated can be described by the Gauss-Laplace’s equation (Eq. 19) which relates the pressure to the capillary radius, and by the Kelvin’s equation (Eq. 20), which relate the pressure to the relative humidity (RH). The capillary pore pressure is, by definition, the differences between the pressures in the fluid that lies



on the concave side of the surface (in this case the air pressure, Pa) and in the other fluid (in this case the water pressure, Pw) [116]; see Figure 36.

⎟⎟⎠

⎞⎜⎜⎝

⎛+−=

⋅−=−=

21w

wwac r

1r1

r2

PPP γγ

[Pa] (Eq. 19)

( ) ⎟⎟⎠

⎞⎜⎜⎝

⎛+⋅

⋅⋅⋅

−=⋅⋅⋅

⋅⋅−=

21w

ww

w

ww

r1

r1

TRM2

rTRM2

RHlnργ

ργ

[-] (Eq. 20)

where γw is the surface tension of the capillary water (~0.074 N/m); r is the mean curvature of the surface; r1 and r2 are the principal radii of curvature in metre (see Figure 39, R is the ideal gas constant (8.314 J/mol K); T is the absolute temperature in Kelvin (20°C≈293K); Mw is the molar mass of water (~0.018 kg/mol); and ρw is the density of water. By combining the two equations, and setting Pa equal to zero (= atmospheric pressure) and r1 equal to r2, the following relationship is obtained:

( )RHlnM

TRP

w

wc ⋅

⋅⋅=

ρ [Pa] (Eq. 21)

This equation (Eq. 21) shows that a small reduction in RH will lead to very large capillary pressure; see Figure 38. Yet, the pressure can only develop if there are partly empty pores available in the system. Once the matrix sets and the pores lose their internal water (by hydration, evaporation, etc.) the pores loses their internal water, starting with the largest pores, and RH will start to drop [100]. In addition, the dissolution of salts into the pore water will increase the alkalinity (i.e. concentration of OH- ions, mainly from KOH and NaOH) and consequently reduce the RH. This reduction will be 1-3%, depending on the cement composition and w/c ratio [126].

1

10

100

1 000

10 000

100 000

1 000 000

0 0.5 1Relative humidity, RH [-]

Cap

illar

y po

re p

ress

ure

[kP

a]

Figure 38. Relation between RH and capillary pore pressure (at 20°C), based on the Kelvin equation. Note, pressure is negative and logarithmic.

Based on the assumption that the capillary pores in the paste are typically in the size range of 0.01 to 5 µm [90], the pore pressure calculated with Kelvin’s equation (Eq. 20) will roughly be in the rage of -10 kPa and -10 MPa. If the gel pores, down to 2 nm [99], were included, the pore pressure would theoretically be as low as -100 MPa.

r1

r1 r2 r2

r1

r1r*

2 r*2

∆Pc Figure 39. Schematic representation of liquid filled joint between two spherical particles at the surface.



According to Powers [4] the maximum negative capillary pore pressure, Pc,max, developed in a cement paste mixture is indirectly proportional to the w/c:

cwS

101P w3max,c

⋅⋅⋅−=γ

[Pa] (Eq. 22)

where S is the mass specific surface area of the cement [m2/kg] measured with Blaine. Assuming S = 350 m2/kg, and w/c ratio 0.55, the capillary pressure calculated with equation (Eq. 22) will be -45 kPa, in a fully hydrated cement paste. However, this is not near the theoretical lower limit of -100 MPa [128] in the smallest pores in the gel.

Many factors have an impact on the pore pressure development, related to the evaporation as well as to the hydration. That the rate of hydration (apart from the rate of evaporation) plays an important role for how the capillary pressure develops is shown in Figure 40, where concretes with accelerator and retarder, and with different superplasticizer dosage, change the time of the pressure drop. It should be noted that the time to, and magnitude of, maximum underpressure are often not possible to evaluate, due to a collapse (loss) of measured pressure at random times. This effect is usually referred to as the breakthrough pressure [9][116], and is the effect of air penetrating the pressure transducer, as the surrounding water menisci cannot find stable positions in the region adjacent to the measuring point.

-80

-60

-40

-20

0

0 2 4 6 8Time (from mix) [hours]

Por

e pr

essu

re [k

Pa]

Accelerator

Ref.

Retarder

-40

-30

-20

-10

0


Por

e pr

essu

re [k

Pa]

0.6% SP

0.8% SP

1.0% SP

-100

-80

-60

-40

-20

0


Por

e pr

essu

re [k

Pa]

Sealed

Evaporating

(a) (b) (c)

Figure 40. Measurements of capillary pore pressure on SCC with w/c 0.67 (from [91]), and the effect of (a) accelerator and retarder, (b) superplasticizer dosage, and (c) being sealed/unsealed.

3.3.5 Disjoining pressure

Water will be adsorbed in the narrow spaces and on the particles and the C-S-H surfaces at all relative humidities. The thickness of the adsorbed water film increases with an increasing RH. In locations where the distance between two particles/surfaces is hindered, the adsorption of water molecules may induce pressure and cause expansion (see Figure 41).

Hindered adsorption

Menisci

Pd

Figure 41. Illustration of disjoining pressure



This pressure, called disjoining pressure, will rise with an increase in RH of the system and lead to an expansion. Consequently, when the system dries out and RH is reduced the particles will contract [129][130]. The disjointing pressure concerns aggregations in which the smallest interparticle distance is much smaller than the particles [4], and is active in the area of hindered absorption where the distances between the surfaces are smaller than two times the thickness of free adsorbed water layer (~20Å at 100%) [22] [130]. The disjointing pressure develops steady in the range between 100% and around 40% [131], and shows an RH dependency similar to the capillary stress. It increases with increasing thickness of the adsorbed water between the particles (i.e. increased RH) until it exceeds the van der Waals’ attraction, when the particles will be forced apart. Therefore, on first drying and lowered RH, the decreasing disjoining will case the particles to be drawn together by the van der Waals’ forces, and there will be net volume shrinkage [22]. However, there is not a general agreement that the disjointing pressure has a significant influence on the autogenous shrinkage of cementitious materials [125].

3.3.6 Evaporation shrinkage The evaporation shrinkage (or “early-age drying shrinkage”) referees to the volume reduction due to the loss of water to the surroundings, without involving the autogenous shrinkage. Although it is not included, the effects are the same as the shrinkage caused by external suction of water (absorption) by the formwork material or by the sub-base. At the initial stage where the concrete is liquid-like, without a restraining skeleton and empty pore, the loss of water due to evaporation will be equal to an external bulk volume reduction, and fairly constant. Moreover, the evaporation is close to the evaporation from a free water surface, yet dependent on the rate of bleeding. However, for SCC and other high performance concretes, containing a large quantity of fines (and binders), the ability to bleed is restricted.

Once the concrete sets the transport properties changes and slow down the evaporation. During the drying, the larger pores are the first ones to lose their internal water [127]. Moreover, due to the contraction and the development of hydration products the pores will be smaller, or consumed, and the total volume of pores (porosity) will decrease, and so will the permeability [89]. This effect is illustrated in Figure 42 where the effect of w/c on the evaporation is shown.

Parallels can be drawn to the development of autogenous deformation, but instead of the internal consumption of water and chemical shrinkage, the water is externally lost by evaporation. However, the evaporation also has a cooling effect by water transforming to vapour (“evaporate cooling”) which lowers the concrete’s temperature near the surface (depending on the temperatures, vapour pressure, evaporation, etc.). When comparing the temperature development in sealed concrete specimens with evaporation, a difference of approximately 2-5°C can observed during the first hours from mixing (see Figure 43) [91].



0

1

2

3

4

0 6 12 18 24Time [hours]

Eva

pora

tion

[kg/

m2 ]

w/c 0.38

w/c 0.45

w/c 0.55

w/c 0.67Water

Figure 42. Example of evaporation from SCC with w/c from 0.38 to 0.67, exposed to air with 4.5 m/s, 35% RH and 23°C.

15

20

25

30


Con

cret

e te

mpe

ratu

re [C

°] Sealed

Evaporatin

Figure 43. Temperature development for a sealed SCC (w/c 0.55) exposed to air at 4.5 m/s, 35% RH and 23°C [91].

The concrete’s surface temperature and its ability to transport water to the surface in early age are dependent on a rather complex sequence of interconnecting mechanisms changing with time, and are not often considered. Neither are the concrete’s surface texture and the geometry of the pores at the surface considered [116]. Normally, and for practical reasons, it is the environmental factors (air velocity, air temperature and RH) and the concrete temperature that generally are used to evaluate the evaporation in early age (Figure 44). In addition, recommended limits of evaporation rate are sometimes given (see section 3.2) in order to lower the risk of plastic shrinkage cracking.

Figure 44. Commonly used “tool” to estimate the rate of evaporation from the concrete surface at early age (ACI 305).

3.4 Measuring techniques and methods 3.4.1 General Only the basic techniques and general methods for evaluating the unrestrained deformations at early age are described here. An overview is given of some methods and principles for measuring:

1. Chemical shrinkage 2. Autogenous shrinkage 3. Plastic shrinkage

Methods limited to measurement in the hardened state, i.e. only to be conducted after setting or demoulding, have not been considered. Nor have the restrained shrinkage test methods (e.g. ring tests, plate tests, longitudinal tests etc.), or the methods associated with deformations due to temperature dilation. An inventory of testing techniques and



methodologies of restraint shrinkage and early-age cracking can be found in TC 181-EAS (RILEM report 25), and in an article by Bentur & Kovler [132]. Moreover, the restraint shrinkage test methods can not measure prior setting, as the sample must have a certain strain capacity in order to be externally restraint.

The shrinkage due to evaporation cannot be measured alone, since shrinkage measurements on cementitious material will involve the chemical shrinkage from hydration. However, at the initial stage where the concrete is liquid-like, without a restraining skeleton and empty pores, the lost volume of water due to evaporation will be equal to an external bulk volume reduction (manifested as a vertical displacement). Moreover, the evaporation shrinkage, as well as the plastic shrinkage, is only relative as it strongly depends on a number of factors related to the sample’s geometry and environmental conditions. A possible method to evaluate the evaporation shrinkage could be to use the differences between measured plastic and autogenous shrinkage, together with measurements of the settlement.

There are a large number of proposed methods for determining the early-age unstrained deformation. Yet, no standardized methods are available, except for the chemical shrinkage of cement paste (e.g. ASTM C1608). However, a new standard test method for autogenous linear deformation (strain) of cement paste and mortar is proposed as an ASTM standard (C157/C157M). This method, presented by Jensen & Hansen [133], uses a corrugated mould system, developed in order to start measurement when the mortar (or paste) is fresh.

To measure the deformation in early age is a sensitive task, and many of the methods suffer from large disadvantages and errors. Moreover, in early age the deformation is often dependent on a rather complex sequence of interconnecting mechanisms, changing with time. In literature the reported magnitude of early-age deformation varies enormously, which reflects the difficulties in the interpretation of the results based on different techniques and methods [81][111][123]. Moreover, these differences are promoted by the fact that there is no general agreement on when to start measuring or how to define or verify this starting point. This has been pointed out in section 3.3.3.

The methods can be divided, by their principal measuring techniques, into (1) linear deformation measurements and (2) volumetric deformation measurements [81][104]. Both kinds of methods ought in theory to give analogous results. However, even under carefully controlled conditions, volumetric measurements can yield up to 5 times higher results than the linear techniques [81][104]. In the literature, there is no general agreement clarifying this problem [123].

3.4.2 Chemical shrinkage measurement Chemical shrinkage is a microscopic internal (absolute) volume reduction, not considering the air in the capillary pores (porosity) [82]. Hence, only volumetric measurements can be applied. Water must be allowed to fully penetrate into the material, preventing any pores from being emptied, so the thickness of the sample must be relatively small (maximum t=10 mm [134]). If the specimen is too thick, intrusion of water (on which the measurement is based) will be worsened and measurement faults appear [135]. However, for more diluted cement pastes, larger samples can be used [136]. The chemical shrinkage is directly proportional to the volume of water that the cement paste will absorb during its hydration [137]. The shrinkage can be measured in two fundamentally different ways: by dilatometry or by reduced buoyancy [89]. Both methods include an external water source



in contact with the cement paste sample. Bovin et al. [106] have shown that there is a high correlation between the two ways of measuring.

The dilatometric method was initially developed by Le Chatelier [138] . This method consists of immersing a given mass of cement paste in a glass flask, e.g. of “Erlenmeyer flask” type, filled with water and surmounted by a graduated capillary tube (see Figure 45). The flask is to be placed in a thermostable bath at constant temperature (usually at 20°C). The decrease in volume of water, gradually consumed by the hydration reaction, is manually read on the capillary tube at regular intervals [139].

The buoyancy method is based on the same principles as the dilatometric method, but instead of following the volume of consumed water directly, the sample’s reduced buoyancy is recorded (see Figure 45). Based on Archimedes’ principle, the water submerged sample will register a volume reduction by weight [106][136][140][141]. An advantage of this method is that the reading can be logged.

Water

t Cement paste

Flask Plug Graded tube

0.001 g

Water Cement paste

Membrane (oil) Cup

Balance

Dilatometry Reduced buoyancy

Thermo stable bath

Figure 45. Principles of measuring techniques, by dilatometry or by reduced buoyancy, for chemical shrinkage of cement paste.

3.4.3 Autogenous shrinkage measurement Methods for autogenous shrinkage measurements can be of either linear type or volumetric type. However, not many linear test methods are able to handle anisotropic volume changes, i.e. on cementitious materials in fresh state. Moreover, most methods are designed to be used on mortar and cement paste, and not on concrete. Measurements of autogenous deformation, both linear and volumetric, are very sensitive to bleed water in the sealed samples. Since the bleed water may be reabsorbed after setting by the cement paste, the linear shrinkage will reduce or even cause expansion [116]. For volumetric shrinkage, the effect of reabsorbed bleed water on the external volume will be the opposite, since the effect of loss of water is much larger than the expansion from absorption [133][142]. Hence, the presence of bleed water will overestimate the autogenous shrinkage when measuring volumetrically, and underestimate it when measuring linearly. In order to avoid bleeding, rotating the sample prior to setting has been proposed [111][143][144].

Volumetric measurement of autogenous shrinkage is generally performed on samples, mainly cement paste and mortars, in a tight flexible rubber/latex membrane submerged in a thermally stable bath of water (or e.g. silicone oil); see Figure 46. The sample’s change in volume is measured by the amount of water displaced by the immersed sample, usually by measuring the change in weight (reduced buoyancy, according to Archimedes’ principle) [143][145]. The membrane type is selected to withstand alkali and to be tight enough not to allow water to penetrate. The advantage of this method is that continuous measurement can



be commenced immediately after casting. However, in addition to the effect of bleeding water, entrapped air between the membrane and the sample will, once set, be sucked back into the cement paste, causing errors in the measurement (e.g. overestimated shrinkage) [114]. Care must be taken when filling the membrane, to avoid the entrapment of air. Furthermore, the pressure excreted on the sample by the elastic membrane and by the surrounding liquid may brake the growth of crystals and weaken the initial skeleton of the cement paste, and consequently increase the shrinkage during setting [125][133].

It has been shown that, before the formation of self-supporting skeleton and gas-filled pores, there is good agreement between the chemical and autogenous shrinkages when determined by the reduced buoyancy technique [81][108][145][144]. Still, once hydration products start to form (initial set), the chemical and autogenous shrinkages will diverge. This will verify the model for cement pastes, presented in section 3.3.2, Figure 33.

0.001 g Balance

Thermo stable bath

Sample

Flexible tight membrane

Figure 46. Illustration of volumetric autogenous shrinkage measurement by reduced buoyancy technique.

Linear autogenous measurements, at early age, are generally performed on samples in a rigid mould with low friction [81][108][111]. Usually the mould is of prism or beam type, but also slabs or cylinders type occur. An overview of different methods for linear measurements of autogenous shrinkage can be found in literature [111][123][146]. The size of the sample is dependent on the maximum grain size (i.e. if paste, mortar or concrete). The mould is to be temperature-controlled (e.g. circulating cooling liquid) or placed in a thermostable bath. However, for large samples it can be problematic to achieve isothermal conditions (constant temperature). The length change of the sample is usually recorded by displacement transducer, or by non-contact transducers (e.g. laser technique), at the end of the specimen or at rods/bars imbedded in the fresh cast (see Figure 47). Also imbedded strain gauges have been proposed for linear measurements [125][147]. It is important that the exposed surface is sealed/covered (e.g. by diffusion-tight plastic or aluminium foil) immediately after finishing, preventing any evaporation. However, none of these linear methods have the ability to measure before setting, as the material must have some degree of rigidity; hence it is proposed that the tests should be complemented with a transducer (contact/non-contact) recording vertical displacement (settlement) [145]. This displacement can be measured on a small plate placed on the surface of the specimen [148]. By simultaneously measuring the vertical and horizontal displacement, the volumetric shrinkage can be calculated for the whole period of early-age shrinkage.



Displacement transducer

l0

Specimen

Thermo stable bath

Mould Measuring rod

Cover/sealing (e.g. plastic foil) Low friction surface (e.g. teflon foil)

Figure 47. Principle of linear autogenous shrinkage measurement on prism/beam with displacement transducers recording the change in length at the end of the sealed sample. In addition, linear measurement of autogenous shrinkage can be performed on samples cast in a corrugated tight tube. The corrugation permits the tube to deform easily in the longitudinal direction, whereas the geometry of the cross section is well defined. Before set, the tube will transform the volumetric deformation into a linear deformation. However, as the cementitious material undergoes transition from a fluid to a rigid state, the deformation will become isotropic. Thus, continuous measurement can be commenced immediately after casting. The tube is to withstand alkali and be tight enough not to allow water/air to penetrate (e.g. PVC, polyethylene). To measure under isothermal conditions, the sample is to be submerged in a thermally stable bath of water (or e.g. glycol, silicone oil); see Figure 48. Linear autogenous deformation using a corrugated tube mould has been proposed for cement paste, mortar and concrete [133][149]-[151]. For pastes and mortars, smaller tubes are to be used, whereas for concrete the tube must have a larger diameter (~5 times the maximum grain size). Moreover, measurements can be performed in vertical or horizontal position, although linear might be preferable as the vertical position can introduce errors from bleeding and segregation. A setup for concrete is exemplified in section 4.4.3.


l0

SpecimenEnd cap

Corrugated tight mould

Thermo stable bath

Figure 48. Illustration of a linear autogenous shrinkage measurement on corrugated tube mould with displacement transducers, recording the change in length at the end. The autogenous shrinkage measurements, presented in the appended papers (Paper IV, V, IV and VI), were performed using the latter linear technique with a corrugated tube mould.

3.4.4 Plastic shrinkage measurement Plastic shrinkage incorporates most of the mechanisms of early-age deformation. The plastic shrinkage will mainly be governed by the autogenous shrinkage and the shrinkage due to evaporation [82][91][108]. However, when measuring the plastic shrinkage, thermal dilatation from hydration and the evaporation cooling (by water transforming from liquid



to gas) will also contribute to the total deformation. Moreover, the plastic shrinkage is only a relative measure as it strongly depends on a number of factors related to the sample’s geometry and environmental conditions [23]. Thus, these factors are to be controlled and well defined. Methods proposed for plastic shrinkage measurements (e.g. ASTM C157/C157M and C426) need a certain level of rigidity in order to measure the shrinkage, and usually do not consider the deformation during the first 24 hours after casting [118]. For early-age plastic shrinkage, measurement is to be started prior to demoulding. For concrete, a method has been proposed [145][152][153] for measuring the linear plastic shrinkage on slab-type samples in a rigid mould. This is performed analogously to the linear autogenous deformation, but the top surface is exposed to drying at controlled ambient condition (air velocity, RH and temperature). As the material must have some degree of rigidity, tests are complemented with a transducer recording vertical displacement (settlement). By simultaneously measuring the vertical and horizontal displacement, the volumetric shrinkage can be calculated for the whole period of early-age shrinkage. Furthermore, complementary measurements of weight loss can be recorded by placing the sample on a balance, in order to evaluate the evaporation (see Figure 49).


Specimen

Mould

Measuring rod Low friction surface (e.g. teflon foil)

0.001 g Balance

Displacement transducer (vertical)

v, T, RH

Fan Thermo couple

Figure 49. Principle of linear plastic shrinkage measurement on a slab exposed to drying, with displacement transducers recording the vertical and horizontal deformation.

3.5 Concluding remarks Some concluding remarks concerning the early-age shrinkage of fresh cementitious materials (paste, mortar and concrete) can be made as follows:

o There are two main driving forces for early-age shrinkage: (1) drying shrinkage due evaporation, and (2) chemical shrinkage due to hydration. However, temperature dilation may also contribute, as well as the external suction of water by the formwork material or by the sub-base. Accordingly, cracking in the initial phase is a product of several mechanisms.

o To elucidate the mechanism of early-age deformations, the underlying mechanisms, such as capillary pore pressure, temperature and structural development, are to be investigated and taken into account.

o At the stage where the paste is liquid-like, without restraining skeleton and empty pores, the chemical shrinkage will be equal to the external bulk deformation. However, at the stage when the self-supporting skeleton of hydration products starts to form (initial set) the external and internal chemical shrinkage will diverge, since the rigidity of the paste restrains the volume change. This external deformation (for a sealed system) is referred to as autogenous shrinkage.



o Early-age shrinkage can either be measured linearly or volumetrically, both techniques having advantages and disadvantages. The main difference is that linear techniques usually need a certain level of rigidity of the sample on order to start measuring. Moreover, even under carefully controlled conditions the volumetric technique produces up to 5 times higher measuring results.

o Bleed water may introduce large errors to the shrinkage measurements, and care are to be taken in order to avoid segregation and bleeding. Volumetric measurements are especially sensitive for bleed water and entrapped air, causing an overestimated shrinkage. Whereas, in linear measurements the reabsorbed bleed water may lead to underestimation of the shrinkage.

o Autogenous deformation refers to deformation under autogenous conditions (unstrained, sealed and isothermal) and do not refer to whether the system have set or not. Moreover, as the point of set is not a well-defined physical state but rather a continuous transformation from a liquid to a solid state, often arbitrarily chosen. And as the shrinkage is highest up to an age of 12 hours, it is suggested that autogenous deformation is to be referring to the whole period of hardening from mixing. Also this offers the possibility to register the time of setting.

o Linear measurements, started prior setting, are a sensitive task. It is not only associated with large scatter, it also somewhat ambiguous since the length is not defined in a fluid system. At the initial stage where the concrete is liquid-like, without a restraining skeleton and empty pore, the volumetric shrinkage will manly turn into a vertical displacement (settlement).

o By simultaneously measuring the vertical and horizontal displacement, the volumetric shrinkage can be calculated for the whole period of early-age shrinkage. Alternatively, linear measurement can be performed on samples cast in a flexible mould with a well defined cross section. Hence, before set the tube will transform the volumetric deformation into a linear deformation.

C h a p t e r 4 – E x p e r i m e n t a l w o r k


4 EXPERIMENTAL WORK

4.1 Introduction Complementary information concerning the experimental methods and materials used in the appended papers is presented in this chapter. Further, a more detailed description is provided, and some further analysis of the methods and the results is made.

In Paper I the influence of Bingham rheology parameters on the slump flow values of self-compacting concrete and mortar is evaluated, and it addresses the complex connection between the slump flow spread, flow time (T50), yield stress and plastic viscosity. A large number of more or less self-compacting mortars (~200 mixtures) and concretes (~550 mixtures) with a wide range of consistency were used. The mortar rheology was measured by using a Bohlin CVO200 rheometer with concentric setup, and for concrete a ConTec Visco5 rheometer was used.

In Paper II the properties of eight different fillers (seven limestone and one glass filler) and their influence on the rheology of self-compacting mortar are evaluated. The mortar rheology and slump flow were measured using the same setup as in Paper I. Traditional methods for determining the surface area of filler material were used, such as the area from size distribution (laser diffraction), by image analysis (SEM), Blaine, and by BET(N2). In addition, a simplified method called BET(H2O) was used. Geometrical properties such as F-shape, F-circle and Compactness were determined from image analysis (SEM), and the fillers’ retained water ratio and deformation coefficient were measured.

In Paper III the effect of the specific surface area of gravel and limestone filler on the rheology of self-compacting concrete is evaluated. The area was measured with the BET(H2O) method, as in Paper II. The differences in BET-area were used as a basis for calculation of the water demand for the concrete mixes. The concrete rheology and slump flow were measured using the same setup as in Paper I. The mix design and its constituents comprised typical materials and compositions for SCC in Sweden. A 0-8 mm natural gravel, collected at five different occasions, was used, as well as five limestone fillers with different geological origins.

In Paper IV the influence of fineness of concrete constituents on early-age (<24 h) autogenous shrinkage is investigated. The experiments were conducted on SCC, incorporating the same five limestone fillers as in Paper III. Tests were also conducted on corresponding mixes without cement, i.e. without chemical shrinkage. A newly developed concrete dilatometer was used, generating measurements of linear autogenous (sealed) deformation of the concrete cast in a vapour-proof flexible tube mould. Temperature and pore pressure were simultaneously measured on the sealed samples together with the deformation.

In Paper V the influence of mix design on autogenous deformation (with the same method as in Paper IV) and crack tendency due to plastic shrinkage are evaluated. For the crack tendency test, a modified ring-test method was used where the concrete specimens were exposed to early drying conditions while restrained by an inner steel ring. A large number of different SCC constituents and mix compositions were investigated: e.g. w/c-ratio from



0.38 to 0.67, silica fume, and different admixtures. For comparison, tests with standard concrete were also made.

Paper VI summarizes Papers II-V, with the focus on the five limestone fillers with different BET-area and their effect on the fresh and hardened properties of self-compacting concrete (SCC). Complementary tests of compressive strength (28 d) were made.

4.2 Materials and mix design The mix design comprised typical compositions for SCC (in Sweden), and the constituents used are all available products on the Swedish market, with some exceptions. However, in Paper I, some materials used are unique. In this case the mixture’s composition was not considered, as only the measures of flowability were of interest. Fore instance, different cements, mineral and chemical admixtures were used. Neither were all mixes self-compacting, but they still had the ability to generate a measurable slump flow spread.

In Paper II one of the seven limestone fillers, the one named “L180X” from Nordkalk, is not a commercial product. Furthermore, the filler named “MicroFiller” (ground recycled glass), from SGÅ, is no longer produced.

The SCC mixes were designed to have a water-to-filler ratio (w/f) by volume close to 1.0 [154]. In this calculation, all solid material smaller than 0.125 mm was included as filler.

The following two materials, typical for all mixtures (except for Paper I), were used:

o Cement, type CEM II/A-LL 42.5R, “Byggcement Std PK Skövde”, Cementa. o Superplasticizer, polycarboxylate ether-based type, “Sikament 56”, Sika.

The mortar recipes in Papers I and II are based on “normal” SCC, where the proportions were scaled down to a maximum grain size of 0.5 mm. This scaling was made, thus with some modifications for appropriate consistency, according to the following factors:

o 95% water o 60% superplasticizer o 100% cement and filler o 50% natural gravel (0-8 mm), with dmax = 0.5 mm. o 0% coarse aggregate (8-16 mm)

The limestone fillers used, all of ground type, were selected for their differences in size distribution and geological origin, thus providing variations in size, surface area, shape, surface porosity and texture. The location of the quarries where the limestone fillers originated from is illustrated in Figure 50, and their chemical composition is given in Table 6. One of the coarsest fillers, the “L70”, was also the one with the largest specific surface area by BET. An explanation for the large area may be the apparently porous structure of the particles available for water to be adsorbed (see Figure 51). Compared with the filler “L40”, used as reference, the area by BET(H2O) for “L70” was ~3.4 times larger.

G100 G200 (Filipstad)

L25 L40 L180X (Köping)

L10 L70 L180 (Uddagården)

L15 L190 (Ignaberga)

Figure 50. Location of the limestone quarry’s in Sweden.



Figure 51. Pictures of limestone fillers, taken with SEM. The pictures edge length corresponds to 0.5 mm. To the left, the relatively coarse “L70” filler with a large specific surface area. To the right, the “L40” filler which was used as a reference.

Table 6. Chemical composition of the limestone fillers, in percentage. L10 L15 L25 L40 L70 L180 L180X L190 G100 G200

CaO 48.7 49.1 51.2 50.2 46.8 47.4 50.0 51.5 52.9 52.5 SiO2 6.8 9.9 3.7 5.2 8.3 8.0 5.5 5.4 3.8 3.5 Al2O3 2.3 0.63 0.89 1.2 2.7 2.8 1.2 0.42 0.83 0.85 Fe2O3 1.10 0.25 0.78 1.00 1.4 1.40 0.95 0.19 0.9 0.7 MgO 0.63 0.44 1.40 1.60 0.61 0.68 1.54 0.45 6.70 4.50 K2O 0.55 0.33 0.23 0.30 0.63 0.65 0.33 0.19 0.19 0.15 LOI 38.7 39.1 41.1 40.5 37.7 37.6 40.6 41.2 40.0 40.0

In Paper III, a natural gravel (0-8 mm) named “Hol” was used. Samples (~100 kg) were collected over a period of 4 months at regular intervals from the daily production at a concrete plant. The deviations of standard quantification measures for natural gravel (size distribution, water absorption, density, petrography, modulus of fineness and silt content according to EN standards) were within acceptance. For each sample, the specific surface area was measured by BET(H2O). Five representative samples were then chosen, having a wide range of BET-area (see Figure 52). Geometrical properties such as F-shape, F-circle, Compactness and surface area were determined by image analysis on SEM pictures. No significant differences between the five samples were detected (see Figure 53). Neither did the petrography analyse show significant differences (see Table 7)

Table 7. Petrography analysis of the natural gravel “Hol”, for particles in the size range 0.125-0.5 mm (values in percentage of total number of particles analyzed).

Granite Quartz Feldspar Mica Amphibolite Porphyry Clay slate Sandstone

Opaque minerals

F(ref) 17 40 36 3 3 0,4 0.8 0.6 G 15 49 29 1 3 0,2 0.6 0.4 H 15 43 33 2 3 0,8 1 0.8 I 17 37 38 2 5 0,2 0.8 0.8 J 19 39 35 1 4 0,6 0.8 0.6



2 200

2 400

2 600

2 800

3 000

3 200

3 400

1 2 3 4 5 6 7 8 9 1011 12 1314 1516 17 1819 20

BE

T(H2

O) [

m2 /k

g]

H F(ref)

G

I

J

0.5

0.6

0.7

0.8

0.9

1.0

F(ref) G H I J

Com

pact

ness

, F-s

hape

, F-c

ircle

[-]

50 000

55 000

60 000

65 000

Spe

cific

sur

face

are

a [m

2 /m3 ]

CompactnessF-shapeF-circleSurface area

Figure 52. Regular measures (4 months) of specific surface area by BET(H2O) for the 0-8 mm natural gravel “Hol”, and the five selected samples (F to J).

Figure 53. F-shape, F-circle, Compactness and surface area determined by image analysis (SEM), for particles of sizes up to 0.125 mm

Each mortar mix was prepared in batches of 0.5 to 1.0 litres, using a 4.73 litre Hobart laboratory paddle mixer (see Figure 55). The mixing sequence is based on ASTM C305, and is illustrated in Figure 54.

The concrete mixes were made in batches of 30 to 60 litres, mixed in a BHS-60 twin-shaft paddle mixer (see Figure 55) for four minutes after all water was added to the premixed dry materials. The chemical admixtures (e.g. superplasticizer) were added directly after the mixing water.

30 sec I. 140 rpm

stop/start

0:00CEMENT

stop

0:30

1:00

(15 sec)

start 3:00

1:30

SAND & FILLER

SP

(SCRAPE)

stop 4:00

WATER

3

2

1

30 sec I. 140 rpm

30 sec II. 285 rpm4

5

90 sec 0 rpm

6

60 sec II. 285 rpm

7

Figure 54. Mixing sequence (mortar).

Figure 55. Photo showing the mixers. To the left is the Hobart mixer used for the mortars, and to the right the BHS mixer used for the concretes.



4.3 Particle quantifications 4.3.1 Surface area from size distribution The specific surface area (S) is a single parameter to describe particles’ total area available for surrounding media. One way to evaluate S is to assume the particle geometry and calculate the area from the size distribution (grading curve). Hence, the surface area generated by the particles’ shape, porosity and texture will not be taken into consideration. Moreover, the contribution from the smallest fraction will be underestimated. The content of fines might be small, but the number of particles can be large and, consequently, so will the total surface area (see Figure 1).

In Paper II the sample’s specific surface area from size distribution was determined, both on the basis of results from the laser diffraction analysis (Ssize) and from the image analysis with scanning electronic microscope (SSEM).

The Ssize was calculated (BS 4359 standard) [4][155]-[127] by assuming consistently cubical or spherical and smoothed particles with uniform density (ρ), as:

=⋅

=V

ASsize ρ

∑∑

⋅

⋅⋅=

)dm()dm(6

3ii

2ii

ρ [m2/kg] (Eq. 23)

The particle size ( id ) is represented by an average value from the two contiguous sieves´ nominal sizes (di and di+1), as:

1+⋅= iii ddd [m] (Eq. 24)

The mass of material (mi) in this fraction is represented by the percentages of retained material by weight on the smaller contiguous sieve (see Figure 56).

Size [mm]

Sta

ying

[wei

ght-%

]

10

20

30

40

50 Size distribution

0m1

0.25 0.125 0.50d1 d2 d3 d4

m2

m3

m4

0.063

Figure 56. Volume retained material (mi) and its average size ( id ), in a linear grading curve, for calculation of specific surface area (Ssize) from size distribution.

The size distribution from the image analysis (SSEM) was determined according to the Nordtest method NT BUILD 486, a method based on measurements of the maximum Feret diameter (Fm) of 2D intersected 3D particles, where the total number of 2D objects per unit area in size class i (nai) is calculated as:

∑=

⋅⋅=k

1jjjji Hnpna [pcs/m2] (Eq. 25)

where k is the number of classes and pj is the probability that a 3D particle of size jH will

be observed as a 2D object in size class i. nj is the number of 3D objects and jH is the geometrical mean size in class j.

The SSEM was calculated, assuming cubical particles (shape factor 0.668) with uniform density (ρ), as [4]:



∑∑⋅=

⋅=

)()(6

3,

2,

icircle

icircleSEM D

DV

ASρρ

[m2/kg] (Eq. 26)

where Dcircle is the diameter of a circle with the same area as measured (see Figure 57), and was computed as:

π/A2Dcircle ⋅= [m] (Eq. 27)

where A is the particle cross-section area.

Area, A

Dcircle

Figure 57. Dcircle.

Even though the laser diffraction and the scanning electronic microscope (SEM) recorded relatively small particles, the calculated area was still 10 to 100 times smaller than the area measured with the BET method. This proves the significance of the particles’ shape, porosity and texture of the surface area. Furthermore, in Paper II it was shown that the correlation between the specific surface areas from size distribution (Ssize and SSEM) and by BET was poor, whereas the correlation with Blaine was strong. The relationship with the rheology measures was also poor.

4.3.2 Specific surface area by BET(N2) A commonly used method to determine the specific surface area of powders is by gas adsorption methods. By using BET theory (named after its originators Brunauer, Emmett, and Teller [158]), the surface area can be determined from knowledge of the size and number of the physically adsorbed gas molecules. This is a more direct method of determining the specific surface area (S), which is deduced from the area occupied by the volume of gas required to cover and form a monolayer on the surface of the sample. This area is normally much larger than the surface area determined by air permeability (i.e. Blaine) [127][157].

In Paper II the fillers specific surface area by BET(N2) was determined by the multipoint method, using nitrogen gas as adsorbate at different relative pressures and constant temperature (-196°C). The test was performed according to ISO 9277 standard, by Norkalk AB (limestone filler producer) (see Figure 58) and by SP (Swedish National Testing and Research Institute) both using a Micromeritics BET analyzer.

Figure 58. Photo showing a Micromeritics analyzer, to determine the multipoint BET surface area by flowing gas technique.

The results, presented in Paper II, show that the effect of specific surface area by BET(N2) for different filler materials on the mortar’s rheology was significant, and that the BET-area (with N2 as well as with H2O) was the parameter that corresponded best to the rheological parameters.



4.3.3 Specific surface area by BET(H2O) Nitrogen is the most commonly used operating gas when determining the specific surface area by gas adsorption methods. But other gases can also be employed, provided that they are physically adsorbed by weak bonds at the surface of the solid (i.e. van der Waals forces): e.g. argon, benzene, krypton, butane, toluene, paraffin, freon, as well as water [127][157].

By applying BET theory with water vapour as adsorbate, and with the approximation that the relative pressure (p/p0) has the nature of relative humidity (RH) at 20°C, the BET equation can be presented as [159]:

RHCx

)1C(Cx

1)RH1(u

RH

mm

⋅⋅−

+⋅

=−⋅

[kg/kg] (Eq. 28)

where xm is the monolayer capacity (i.e. the weight of the gas required for monolayer coverage), C is a constant related to the heat of adsorption, and u is the moisture content (i.e. the weight of the adsorbed gas) defined as:

dry

dry

mmm

u)( −

= [kg/kg] (Eq. 29)

where m and mdry are the sample weight in adsorbed and dry condition.

y [k

g/kg

]

a

RH [-]

b

0.113 0.331

Figure 59. BET-plot with the intersection point (a) and slope (b) calculated from the linear relation-ship between y and RH.

By using a so-called BET-plot (see Figure 59), where the adsorption isotherm (y) is plotted against RH, a linear relationship is created with the intersection point (a) and slope (b) as:

RHbay ⋅+= [kg/kg] (Eq. 30)

)RH1(uRHy−⋅

= [kg/kg] (RH 0;1) (Eq. 31)

Cx1a

m ⋅= [kg/kg] (Eq. 32) Cx

)1C(bm ⋅

−= [kg/kg] (Eq. 33)

From this BET-plot, a and b can be determined (graphically or by linear regression), and the two unknowns xm can be derived as:

ba1xm +

= [kg/kg] (Eq. 34)

The specific area (S) is directly proportional to the amount of monomolecular adsorbate, as [156]-[157][160]:

mm AM

NxS ⋅

⋅= [kg/kg] (Eq. 35)

where N is Avogadro´s number (6.022·1023 molecules per mol), M is the molecular weight of the adsorbate (for water 18.016·10-3 kg per mol), and Am is the area occupied per molecule of adsorbate in a monolayer. For water, Am is approximately 10.6·10-20 m2 per molecule [26][156][127][159]. As a perspective, one litre of water in one molecule layer



will cover 3,546,000 m2. From this, the specific surface area by BET, with water as adsorbate, can be calculated as:

ba1054.3S

6

O2BETH +⋅

= [m2/kg] (Eq. 36)

It should be noted that adsorption is not the same for moisture as for nitrogen gas. The water molecule is smaller than the nitrogen molecule (10.6 Å2 for water, and 16.2 Å2 for nitrogen), and has larger affinity forces (of attraction) due to its two-polar covalent bonds (see Figure 60). Consequently, the water molecule can penetrate into more narrow spaces (pores and texture) and will be more strongly adsorbed to the particles surfaces, relative to the nitrogen molecule. A larger quantity of absorbed molecules, in a single layer, will generate a larger BET-area. However, using water (moisture) is probably closer to real conditions for materials for concrete production than using nitrogen gas.

H

H H2O

O

+

+ N N

N2

- Figure 60. The polar water (H2O) molecule versus the non-polar nitrogen (N2).

Mono-layer (~3.5Å for water)

Adsorbed molecules

Figure 61. Molecules adsorbed onto a surface

Usually a second layer is formed before the monolayer is complete (see Figure 61), but in the area of 0.05<p/p0<0.40 the BET-plot usually shows a linear relation and a monolayer theory can be adopted [127][160]. Moreover, no appreciable capillary condensation can take place below 0.4, which is the criterion for the BET theory [161]. Thus, to be able to use BET theory with moisture as a monomolecular adsorbate, the RH must be restricted to approximately 5-40% (at 20ºC). In Figure 62 the measurements of the adsorption isotherm for different fillers and natural gravel at RH up to 40% are given. It can be seen that there is a linear relation within 10-40%, which indicates a monolayer adsorption in this region.

0

0.002

0.004

0.006

0.008

0.01

0 10 20 30 40 50RH [%]

moi

stur

e co

nten

t [kg

/m3 ]

L180

L40

Natural gravel 0-1

Microfiller

Figure 62. Adsorption isotherm for two limestone filler (L40 and L180), a glass filler (Microfiller) and a natural gravel (0-1 mm).

The technique of using the BET method with moisture (water) instead of nitrogen gas has been applied by several researchers [4][156][127][159][162]-[165]. In the literature, it can be seen that during the 1960s and early 1970s there was a vigorous debate on whether water vapour should be used as BET-adsorbate. Traditionally, when using BET with water, the focus has mainly been on internal porosity (e.g. to determine the pore size distribution in a C-S-H gel) and not on external particle surface. Neither has the simplified method, with different relative humidity instead of pressurized vapour, been used for particles.



The sample particles’ specific surface area by BET(H2O) was determined gravimetrically by the simplified multipoint BET method, using water (as moisture) as adsorbate at different relative humidities (RH). Initially, the test was performed with a “two-pressure” humidity generator (see Figure 63) where the specimens were conditioned in air vaporized at several levels (from 6 to 50% RH). Later, all tests were performed with climate boxes at 11.3% and 33.1% RH, achieved with LiCl and MgCl2 saturated salt (binary) solutions (see Figure 64). To ensure linearity, KC2H3O2 can be used for 23.1% RH as a third climate.

The sample size was approximately 100 g, the weight was measured by a precision balance with 0.001 g readability, and the sample was oven dried at 105ºC. The samples were conditioned in each climate for approximately four days, at 20ºC.

Figure 63. Thunder 2500, a "two-pressure" humidity generator, for climates with high accuracy.

Fan

Saturated salt solution

Box with tight lid

Net

Sample

Figure 64. Photo of climate boxes with RH 11.3% (LiCl) and 33.1% (MgCl2) RH (left) and their principle design (right).

The time for the samples to be conditioned and reach steady state within acceptance was evaluated to be 3-5 days, depending on how large an area the material has. In Figure 65, it is shown how the oven dried samples, exposed to 50% RH, increase in moisture content with time. However, this is a relatively high RH.

0.0%

0.1%

0.2%

0.3%

0.4%

0.5%

1 10 100 1000 10000Time [hour]

Moi

stur

e co

nten

t [%

]

L10

L15

L25L40

L180

L180X

L190MF

B15

Figure 65. Changes of moisture content with time, for seven limestone fillers: a glass filler and a natural gravel (0-1 mm), conditioned at 50% RH. The time scale is logarithmic.



The repeatability was evaluated, using two different natural sands (1 and 2). Ten repetitions were made, with double samples. New samples (2x100 g) were taken between the runs. The results are presented in Figure 66. For each sample the coefficient of variance (COV) was lower than 1.82%. A tendency can be observed, indicating that there is a systematic error, yet not very large. There are numerous sources that can cause this error, such as the conditioning time, weighing of oven-dried samples, measuring order, and room temperature. No further investigation of this was made.

3100

3200

3300

3400

3500

3600

S(B

ETH

2O) [m

2 /kg]

1a, COV 1.82% 1b, COV 1.55%

2b, COV 1.45% 2a, COV 1.73%

1(average) COV 1.65%

2(average) COV 1.67%

2(average) 3298 m2/kg

1(average) 3349 m2/kg

Figure 66. Repeatability for the BET(H2O)-area, using double samples of two different natural sands (0-0.5 mm). New samples (2x100 g) were taken between runs.

For gravel/aggregate, the removed particles not passing the 0.5 mm sieve can be compensated for, but only very approximately. The BET(H2O) area (S´BETH2O) per unit specimen mass is then calculated for the whole size fraction and approximated as:

( ))x1(kxSS 0505O2BETHO2BETH' −⋅+⋅= [m2/kg] (Eq. 37)

tot

0505 m

mx = [-] (Eq. 38)

where k is a factor compensating for the removed particles not passing the 0.5 mm sieve, x05 is the mass fraction passing the 0.5 mm sieve, m05 is the mass passing 0.5 mm, and mtot is the total mass. If there is only a small amount passing the 0.5 mm sieve, the calculation of S´BETH2O is not valid. In a series of measurements on four different 0-8 mm gravels (naturals, sea sand and crushed), the area was measured on the fraction passing 0.5 mm as well as on the whole fraction. The test was repeated three times, in order to ensure correct measurements. The gravel’s size distribution is shown in Figure 67, and the measured BET-area and factor is given in Table 8. For gravels with a maximum grain size up to 8 mm the factor was approximated to k ≈ 0.5.

0%

20%

40%

60%

80%

100%

1684210.50.25

0.125

0.0630

Size [mm]

Pas

sing

[%]

Natural 1Natural 2Sea sandCrushed

Figure 67. Size distribution

Table 8. Measures of BET(H2O)-area for the fraction passing 0.5mm and the whole fraction, used for the evaluation of factor k.

Gravel (0-8mm)

<0,5mm [%]

SBETH2O [m2/kg]

S’BETH2O [m2/kg]

k [-]

Natural1 26% 4142 2375 0.42 Natural2 50% 8162 6320 0.55 Sea sand 65% 1287 1101 0.59 Crushed 34% 4815 3102 0.46 Average 0.51



In Paper II it was shown that the granular characteristics with the highest single correlation was found between the BET(N2) and BET(H2O) measures (specific area). Although the measured values are not equal, the linear relationship is good.

One has to be aware of the differences in size and polarity of the water and nitrogen molecules. Tests have also been made on fines (<1.0 mm) sieved from different aggregates, with the same good relationship. It was found that the BET(H2O) was 1.55 times larger than BET(N2), see Figure 68.

For binders of cement type, the BET(H2O) method did not supply reliable measures due to the nature of hydration, and is therefore not recommended.

y = 1.554xR2 = 0.994

0

10 000

20 000

30 000

40 000

0 5 000 10 000 15 000 20 000SBET(N2) [m

2/kg]

SBE

T(H

2O) [m

2 /kg]

Figure 68. The linear relation between measures of BET-area with N2 and H2O.

The results from the flowability measurements in Papers II and III for mortars and concretes show significant differences between the samples, which were strongly reflected by the differences in BET(H2O)-area. There was a strong connection between the BET(H2O)-area and the rheological parameters and slump flow values.

When there is a change in the dry material’s specific surface area, and as this area has to be covered with water in order to create mobility, the concrete water demand will change. One way to compensate for these variations in filler or fine aggregate, presented in Paper III, is to consider the specific area by BET(H2O) and translate this to a change in water demand of the concrete mix. With the assumption that 1 litre of water in one molecule layer covers 3,546,000 m2, the change in mixing water (∆W) due to a change in specific surface area (∆SBETH2O) can be calculated as:

mtr6O2H

20BETH m10546.3

nSW ⋅

⋅⋅= ∆∆ [litre/m3] (Eq. 39)

where mmtr is the mass of the dry material of the mix, and nH2O is the number of full water molecular layers covering the particle surfaces required to provide sufficient dispersion for flowability. Based on numerous experiments, a suitable value was found to be nH2O ≈ 30 layers. This can be compared to the 5-20 molecular layers of water adsorbed at high relative humidity, proposed in literature [4][156][127][159][166][167] or as much as 60 molecules in thickness [20]. Furthermore, it was shown that the free water content, available to provide flowability, in 1 m3 SCC mix can vary by several litres due to differences in apparently equivalent fine aggregates. In [5] it was found that a natural gravel collected at different spots in a gravel pit, with small differences in grading, can vary in BET(H2O)-area by ~7000 m2/kg. A change in BET-area of 1000 m2/kg corresponds to approximately 0.8% moisture content of the gravel. E.g. a SCC with 1000 kg gravel (0-8 mm) that increases by 10000 m2/kg in area will need 8.5 litres of extra water, in order to retain its consistency.

In Papers IV and IV the BET(H2O)-area was shown to influence – apart from the flowability and water requirement – the pore pressure development, autogenous deformation, plastic shrinkage and compressive strength. Limestone filler with larger BET-area increased the rate and magnitude of autogenous shrinkage, primarily in the plastic region, without affecting the temperature development and the times to initial and final set.



Increased particle surface area decreased the rate and magnitude of evaporation, and reduced the plastic crack tendency. In addition, a larger BET-area gave rise to a higher compressive strength. When adding extra water to the mix, compensating for the loss of flowability due to increased particle surface area, the crack tendency increased significantly, the autogenous shrinkage decreased and the compressive strength was lowered.

For more details and a methodology description of the BET(H2O)-method, see Appendix A.

4.3.4 Specific surface area by Blaine

The surface area of powder can be determined by measuring the pressure drop of fluid flow through a packed powder bed. The most frequently used apparatus is the Blaine, which is a standard test method for characterizing the fineness of hydraulic cement by air permeability.

In Paper II, the fillers’ fineness, in terms of specific surface area, was determined by the Blaine method (Sblaine). The test was made according to the SS 134222 (similar to ASTM C204) standard, with a Tonindustrie instrument (see Figure 69).

Comparative measurements on the fillers were performed by Norkalk AB, using a ToniPerm instrument.

Figure 69. The Blaine apparatus for determining powder fineness, in terms of specific surface area.

A strong correlation was found between the calculated specific area (from size distributions) and the area measured with Blaine, whereas the correlation with the BET-area was weak. This indicates that Blaine is mainly a measure of size and size distribution, based on theoretical spherical particles. However, it should be noted that great variation was observed between values measured on the same specimen, especially between those performed by different persons (and equipment used).

Furthermore, the results presented in Paper II, shows that the effect of specific surface area by Blaine for different filler materials showed a poor correlation with the mortar’s rheological parameters.

4.3.5 Image analysis Image analysis refers to a computer analysis of digital images, where a large number of particles, each with numerous parameters, are generated. The image analysis unit has the capability to generate an array of chords across the particle to arithmetically define geometrical quantification, e.g. of shape and size.



The image processing and analysis were made, in Paper II, with the software “UTHSCSA Image Tool”. The samples were epoxy-moulded and polished. The images were made by a low-vacuum scanning electron microscope (SEM), type Jeol 5310LV (see Figure 70), at Swedish National Testing and Research Institute (SP). For a representative sample, approximately 15,000 particles from four different digital pictures with 2016·2016 pixels size were used. The following shape factors were calculated as (Eq. 40) to (Eq. 42).

Figure 70. Images generated by SEM.

F-shape: the ratio of the length of the minor axis (Dmin) to the length of the major axis (Dmax). If the elongation is 1, the object is roughly circular or square (see Figure 71). The F-shape is computed as:

maxmin DD [-] (Eq. 40)

F-circle: the ratio of measured area to the area of a circle with the same perimeter as measured. At 1, the object is circular (see Figure 71). The F-circle is computed as:

2

4Perimeter

Area⋅⋅π [-] (Eq. 41)

Compactness: the ratio between the diameter of a circle with the same area as measured (Dcircle) and the maximum elongation (Dmax) of the measured area. At 1, the object is circular (see Figure 71). The Compactness is computed as:

max

/2DArea π⋅ [-] (= Dcircle / Dmax) (Eq. 42)

Dmax

Dmin a)

Area (pe.) Perimeter

Area (ci.)

b)

Dmax

Dcircle

c)

Figure 71. Geometrical measures for calculation of F-shape (a), F-circle (b) and Compactness (c).

The results, in Paper II, show that the correlations between the shape factors (F-shape, F-circle and Compactness) are good. This was expected since the values used for calculation have much in common. Thus it ought to be pointed out that F-shape is more a measure of shape, F-circle is more a measure of texture, and Compactness is somewhere in between. The correlation with other granular characteristics (calculated surface area, BET-area and Blaine) was poor, and so was the connection with the mortar’s rheological parameters.

The quantification by image analysis can be problematic, as the method is strongly dependent on a large number of factors, such as: sample preparation, microscope performance, photo quality (sharpness), resolution, number of particles, analyzing software, threshold technique, etc. Since the threshold was made manually, the results can be somewhat arbitrary.



4.3.6 Retained water and water sensitivity When a sufficient amount of water is provided to the powder, the interacting forces decrease, the particles start to disperse, and deformation arises by its own gravity. Okamura et al. [168] found that there was a linear relationship between the water-to-powder ratio and the relative spread flow, which can be used to empirically characterize the powder’s response to water addition.

In Paper II, this method was used to quantify the different fillers’ retained water ratio (βp) and deformation coefficient or sensitivity (Ep). A mini-slump test was used to measure the flow spread. The relative flow area (Rp), at different water-to-powder ratios by volume (Vw/Vp), was calculated as:

1DDR 2

0

2

p −= [m2/m2] (Eq. 43)

where D is the average spread diameter, calculated in four directions and D0 is the base diameter of the cone (see Figure 72).

The measure of retained water ratio (βp) can be considered as the amount of water, adsorbed on the particle surfaces and filling the voids in the particle system, needed to initiate flow. The deformation coefficient (Ep) is a measure of the sensitivity to a change in water content [169]. βp is represented by the interception point and Ep the slope, from the linear relationship between Vw/Vp and Rp (see Figure 73).

Mini-slump cone ØD=700/100 mm

H=60 mm

Sample Table (non-adsorbent)

Spread diameter

Figure 72. The mini-slump and the measurements of spread flow diameter (in four directions).

Vw/V

p [-]

βp

Rp [-]

Ep

Figure 73. The evaluation of βp and Ep from the linear relationship of the ratio of water to powder (Vw/Vp) and relative spread flow (Rp).

Results, in Paper II, show a strong correlation between the fillers’ retained water ratio (βp) and the mortar slump flow spread, whereas the deformation coefficient (Ep) correlated with the specific surface area from size distribution and with the Blaine area. Theoretically, the fillers’ retained water ratio (βp) should reflect the ability to adsorb water, and correlate with the BET(H2O)-area; but this correlation was poor. In addition, the fillers’ water absorption (with particles passing the 0.063 mm sieve removed) was determined in according to EN 1097-6, a method mainly designed for aggregates. But as the coefficient of variance was high (~25% when repeated four times) these measurements are not presented. Still, the water absorption showed a correlation with the BET(H2O)-area.

4.4 Mortar and concrete quantifications 4.4.1 Slump flow test The oldest and most frequently used test today is the slump cone test, which, associated with the Abram’s cone, has its origin in the USA around 1910 [170]. For high-flowable concretes and mortars, the slump cone is used to measure the spread (slump flow), rather than the vertical drop.



For the mortars, in Papers I and II, slump flow measurement was carried out by using a mini-slump cone (see Figure 74). The average spread flow diameter, from two perpendicular directions, was measured (in millimetres). No compaction was needed (as in EN 1015-2 and ASTM C230 standard), as the mortar mixes were highly flowable.

For the concretes, all more or less self-compacting, slump flow measurement was carried out using a traditional slump cone (EN 12350-2 or ASTM C143, but without compaction; see Figure 75). In addition to the slump flow spread (average), the time from lifting the cone to when the flow spread reached a 500 mm circle was recorded, referred to as T50 and stated in seconds.

The concrete slump flow test (spread and T50) was used in Papers I, III and VI. Moreover, in Paper I slump flow values (concrete and mortar) were used for evaluation of the relation to the rheology parameters (yield stress and viscosity).

Figure 74. Mini-slump flow test with mortar (spread).

Figure 75. Slump flow test with concrete (spread and T50).

4.4.2 Rheology

For a more scientific approach, a rheometer can be used (see section 2.3). Unlike the measurements of qualitative tests (e.g. slump flow), the rheological parameters are fundamental physical quantities, mutually independent and not dependent on operator or equipment. The most established parameters used to define mortar and concrete rheology are the yield stress and plastic viscosity in the Bingham equation [24][28] (see section 2.2.3) .

For the mortars (in Papers I and II), a Bohlin CVO200 rheometer was used, with a fixed outer cylinder (cup) of diameter Do=30 mm and a measuring rotating inner cylinder (bob) of diameter Di = 20 mm, providing a gap of 5 mm (see Figure 76). For the concretes, (in Papers I, III and VI) a ConTec Visco5 was used, with a rotating outer cylinder of diameter Do=290 mm and a measuring fixed inner cylinder of diameter Di=200 mm, providing a gap of 45 mm (see Figure 77).

Figure 76. The Bohlin CVO200.

Figure 77. The ConTec Visco5.



The two rheological parameters of plastic viscosity (ηpl) and yield stress (σ0) were evaluated in accordance with the Bingham model, see (Eq. 7), at a controlled shear rate (γ& ) at 20ºC. In addition, the experimental setup for mortars was initiated with a linear up/down loop for a thixotropic evaluation, represented by the area in hysteresis (see Figure 14). The experimental geometry and measuring sequence is illustrated in Figure 78. Thixotropy is not strictly a rheological definition, nor an absolute or fundamental term, but a quantification of reversible time-dependence (see section 2.2.5). A typical measure of mortar rheology is shown in Figure 79. The measure of segregation was used as a criterion for acceptance (measures rejected when >10%).

0

15

30

45

60

0 10 20 30 40 50 60 70 80 90Time [s]

She

ar ra

te [1

/s]

No loggingLogging

Ro=15 mmRi=10 mm

H=37.5 mm

Rotating and measuring inner cylinder

Fixed outer cylinder

Mortar sample

0

2

4

6

8

0 10 20 30 40 50 60Time [s]

She

ar ra

te [1

/s]

No loggingLogging

Ro=145 mmRi=100 mm

H=140 mm

Measuring inner cylinder Rotating outer cylinder

Concrete sample

Figure 78. The measuring sequence for the Bingham evaluation (followed by segregation estimation) and schematic illustration of the rheometers. For mortar the Bohlin CVO200 was used, and for concrete the ConTec Visco5. The sequence for mortars begins with a linear up/down loop for a thixotropic evaluation.

Thixotropic area

Yield stress and plastic viscosity (Bingham)

Segregation


Thixotropy [Pa/s]

Viscosity, ηpl [Pa·s]

2

1

3

Segregation [%]

Shea

r stre

ss, σ

[Pa]

Yield stress, σo [Pa]

Figure 79. A typical measurement with mortar and the methodology evaluating the Bingham parameters and thixotropy (and segregation).



4.4.3 Autogenous deformation Autogenous deformation is the unstrained bulk deformation of a closed system (under sealed conditions) at a constant temperature (isothermal), i.e. without exchange of matter with the exterior (e.g. loss of water) [104]. Fundamental measurements of autogenous deformation can be made with volumetric or linear techniques, the latter having been used in this study. More general details are given in section 3.3.3.

In Papers IV, V and IV, autogenous linear deformation was monitored by a concrete digital dilatometer (CDD), developed in order to start measurement before setting, when the concrete is fresh. The method is a modification of the CT1 digital dilatometer for pastes and mortars; see Jensen & Hansen [133]. The CDD sample consists of a concrete specimen, cast in a steel coil-reinforced vapour-proof flexible polyurethane tube with inner diameter of ~82 mm and specimen length ~400 mm, sealed with hose clamps and O-ring-equipped PVC end-caps. The mould is placed in a mechanically stable measuring rig and the unrestrained, time-dependent, linear bulk deformation is recorded by a digital gauge (Mitutoyu 543-450B, resolution 1 µm and accuracy 3 µm). The test, containing three complete CDD setups, was performed in a thermostable room at 20±1ºC, where the recording was started at 30 minutes from water addition and logged with an interval of 5 minutes for a period of 24 hours or more. The equipment and apparatus for a typical test are shown in Figure 81.

Due to greater stiffness of the mould in the radial direction than in the longitudinal, the flexible mould transforms most of the volumetric deformation into a linear deformation when the concrete is in the fluid state. Based on an experimental evaluation, using water as matter, the ratio between linear and volumetric deformation for a liquid was determined to be approximately 0.8 for the equipment (see Figure 80). As the concrete undergoes transition from a fluid to a rigid state, the deformation becomes isotropic. But as the setting point is not a well-defined physical state but rather a continuous transformation from a liquid to a solid state [104], no correction for this was made. The experiments were supplemented with measurements of temperature and pore pressure placed in the centre of the core.

y = 0.796xR2 = 0.9999y = x

y = 0.338x

-5%

-4%

-3%

-2%

-1%

0%

-5%-4%-3%-2%-1%0%Volume change

Leng

th c

hang

e (s

train

)

3-D tube (calculated)Fluid (measured)1-D tube (calculated)

Figure 80. The ratio between linear and volumetric deformation for the mould. Filled with a Newtonian fluid the ratio follows the measured ratio, but once rigid the ratio will follow the calculated 3-D deformation.



0 001

Fixed end-cap Moving end-capFlexible PU-tube Hose-clap Measuring rig

0 001

Mould length 460 mm

Specimen length 400 mm Didital gauge1/1000 mm

Adjustable fixture

Mould diameter 82 mm

Figure 81. Photo and illustration of the Concrete Digital Dilatometer (CDD), with supplemented measurements of temperature and pore pressure.

For more details and a methodology description of the CDD-method, see Appendix B.

The result from a test is presented graphically and the development of deformation (mean value from two or more tests) is plotted against time. Moreover, a number of deformation factors were evaluated with a model presented in Figure 82. For concrete, the deformation pattern comprises three distinct stages which can be defined as [91]: plastic, semiplastic and rigid period. In the plastic (1) and semiplastic (2) periods, the rate of deformation (dε/dt1 and dε/dt2), period time (t1 and t2) and deformation (ε1 and ε2) are evaluated from the deformation/time graph. In the rigid period (3), the rate of deformation (dε/dt3) is evaluated between 48 and 72 hours from water addition. The time (t) is related to the time when the mixing water was added to the premixed solid material (including the cement). As the measuring was started after 30 minutes, and as the initial deformation tends to be linear up to 2 hours from mix, the curve was extrapolated to the time zero.

For the SCC mixes used in this work, the transition from plastic to semiplastic (t1) takes place at ~6 hours after water addition and the transition to rigid period (t1 + t2) at ~12 hours. Compared to traditional concrete without addition of superplasticizer, the self-compacting concrete (SCC) tends to be retarded for about one hour [91].



t1+t2 t1 Def

orm

atio

n (s

train

) [10

-6 m

/m]

Time, t [hour]

ε1

ε2

dε/dt1

dε/dt2

dε/dt3

1. Plastic 2. Semi- plastic 3. Rigid

ε1+ε2

ε1

Initial set

Final set

Figure 82. Model of how the deformation parameters are evaluated from the linear autogenous deformation measurements, where the deformation is divided into plastic/semiplastic/rigid periods.

It should be pointed out that linear measurement of autogenous deformation on a concrete mix before setting, when the concrete is fresh, is a very sensitive task and must be made with great care. A mean value from two or more tests, each with three complete CDD setups, was used. A typical measurement (with three setups) generated a coefficient of variance (COV) below 10% (see Figure 83). What can also be noted in Figure 83 is the pattern of autogenous deformation, comprising the three distinct stages (plastic, semiplastic and rigid), separated by the time to initial and final set, and manifested by a plateau in the curve. Mortar, on the other hand, showed a different behaviour. In Figure 84, no semiplastic region and distinct point of initial set could be observed, but merely a plastic and a rigid period separated by a single point of set. Moreover, mortars generated more stable measures with a smaller deviation (COV <5%), but then the magnitude of deformation was also approximately two times that of the concretes.

It might seem that the development of autogenous shrinkage ends at final set, but it is merely slowing down. In long-term measurements, following final set the autogenous shrinkage after 9-12 months was approximately equal in magnitude as prior setting.

-600

-400

-200

0


Def

orm

atio

n [1

0-6 m

/m]

0%

4%

8%

12%

CO

V

COV

Deformation (average)

Figure 83. Autogenous deformation measures and COV for a SCC (w/c 0.55).

-1200

-900

-600

-300

0


Def

orm

atio

n [1

0-6 m

/m]

Figure 84. Autogenous deformation for a mortar (w/c 0.45 and dmax 8 mm).



For standard concrete (i.e. without additional filler and superplastisizer) a different behaviour was also shown. In Figure 85 it can be seen that the initial shrinkage is approximately half the shrinkage the SCC generates. And between initial and final set (semiplastic region), the standard concrete generated much less shrinkage than the SCC. These differences might be explained by the effects of the limestone filler and superplasticizer. When limestone is added to a concrete, the initial induction period (or dormant period) is shortened and an extra peak of hydration will occur [171]-[175]. This phenomenon is explained by the surfaces of the filler material acting as nucleation sites for the early reaction products of CH and CSH, which accelerates the hydration of cement clinkers (especially C3S) [176]-[179]. In addition, if the cement has a significant amount of tricalcium aluminate (C3A), calcium carboaluminate will be produced from the reaction between calcium carbonate (CaCO3) from the limestone and the C3A [180]-[186], which will also accelerate the hydration. On the other hand, the superplasticizer used for the SCCs retarded the mixes by approximately one hour.

In Figure 83 and Figure 85 it can be noted that slightly ahead of the final set (12-13 hours) there is a local dip in the deformation, which can be explained by the fact that this point coincides with the peak in the temperature development (see Figure 37). The temperature, in the core of the sample, changed by approximately 5°C. Compensations for this are possible, but are not an easy task as the thermal expansion coefficient (TEC) for concrete is changing with time, especially at early ages. The TEC for concrete is roughly 20·10-6 1/°C up to and including initial set, and 10·10-6 1/°C at final set [187]-[190], whereas for the rig it is 17.3·10-6 1/°C (stainless steel). Hence, no compensation for the change in temperature on the deformation was made.

-600

-400

-200

0

0 6 12 18 24Time [hour]

Def

orm

atio

n [1

0-6 m

/m] Standard concrete

SCC

Figure 85. Autogenous deformation measures for a standard concrete and a SCC (both with w/c 0.55).

When developing the equipment, complementary tests were performed, evaluating the effect of gravity on the measurements of autogenous deformation. Measurements with the sample placed in 0 (horizontal), 30 and 60 degree positions show no significant differences (see Figure 86), so the horizontal position was chosen for the method. However, to ensure unrestrained conditions, a double layer of low-friction paper (teflon type or baking-tray paper) was placed between the rig and the sample. In a series of tests, using a similar dilatometer, and thus with a smaller tube, Mounanga et al. [150] evaluated the effect of the mould’s position on the linear autogenous deformation on self-compacting mortars. It was found that a vertical position generated more than three times the shrinkage that a horizontal one did. This effect was mainly explained by sedimentation and bleeding of the mortars.



-600

-400

-200

0


Def

orm

atio

n [1

0-6 m

/m]

0° (horizontal)

30°60°

Figure 86. The effect of gravity on the measurements, by placing the rig in different angles.

4.4.4 Plastic shrinkage cracking tendency For a general description of plastic shrinkage cracking, see section 3.2.

In Papers IV, V and VI a ring-test method was used for the determination of the cracking tendency of concrete at early ages, as shown in Figure 87. The test method is a modification of the NORDTEST-method NT Build 433 developed at NTNU/SINTEF by Johansen and Dahl [191]. The main deviations from the Nordtest method are: the samples thickness, the environmental conditions, and the method for evaluating the crack tendency.

The test method consists of a ring-shaped specimen of concrete cast between two concentric steel rings with diameter 300 and 600 mm, and with a thickness of 80 mm. The steel rings have ribs attached to provide crack initiation and are fixed to a stiff base plate with a smooth surface (coated with a thin layer of oil). After casting, the ring specimens are placed in an environmental stable room with 23±2ºC air temperature and 35±5% RH. Air funnels are positioned over samples, exposing the surfaces to a constant air velocity of 4.5 m/s. The shape (angle) of the funnel is designed to generate a uniform velocity over the whole surface. For each test a set of three ring specimens was used.

ø300 mm

ø600 mm

80 mm

Fan

Scale

Pressure gauges

PT100

Strain gauges

20 mm

20 mm

Air velocity 4.5 m/s

Concrete specimen

Base plate

Steel rings

Steel ribs

Climat conditions:

23±2°C 35±5%RH

Figure 87. Illustration and photo, showing the test arrangement for the determination of the cracking tendency.



The measurements start 60 minutes after mixing and the temperature, the weight loss and the restraint strain are continuously recorded. After 20 hours of drying, the rings are taken out of the rig and the cracking tendency is evaluated as the average total crack area (crack length x crack width) on the concrete surface of each of the three specimens. The crack width was measured with a crack microscope (to an accuracy of 0.05mm) and the crack length was measured with a digital measuring wheel (to an accuracy of ±1mm). The weight change was recorded, for one of three samples, using a scale (load cell based) with accuracy better than 0.03% for a weight of 100 kg (minimum detectable weight change 20 g). The concrete temperature was measured using both a thermo thread, placed in the specimen, and a PT100 sensor mounted on the inner ring. To monitor the stress build-up of the concrete and the time when the cracks would appear, strain gauges were mounted (in full bridge) on the inner steel ring. The air temperature and relative humidity in the room were measured using a Visala sensor. In addition, the capillary pore pressure was measured (at 20 and 60 mm depth). Plastic shrinkage is considered to be related to the capillary pressure in fresh concrete [9][89][116]. The data (strain, pore pressures, weight, temperatures and RH) were logged every minute using a data logger (PC-logger3100i from INTAB).

For more details and a methodology description of the ring-test method, see Appendix C.

The method is intended for laboratory use and the test results are purely relative. They cannot be directly transferred to predict the extent of cracking which will occur in practice under the prevailing field conditions. Still, the underlying mechanisms can partly be quantified in terms of temperature and pore pressure development. In order to verify the ring-test method, Löfgren & Esping [192] conducted a field study in which the cracking tendency was evaluated on the ring-test specimens, and on larger slab elements that were both cast outside and exposed to the environment. A series of one reference and six different SCC’s were used. The experimental results (see Figure 88) showed a strong correlation (close to unity) between the ring-tests performed in the laboratory and the field study (both the ring-test and the slab specimens).

0

1

2

3

Mix 1(ref) Mix 2 Mix 3 Mix 4 Mix 5 Mix 6 Mix 7

Rel

ativ

e cr

ack

area

[-]

Ring (field)Slab (field)Ring (lab)

Figure 88. Comparison of the relative crack areas between the field study (the ring-tests and the slabs) and the ring-tests of the laboratory study.

4.4.5 Capillary pore pressure When the concrete dries out due to evaporation, the loss of water from the paste generates negative capillary pressure, causing the paste to contract [9], which in turn will lead to an external shrinkage. For a cementitious material where evaporation is prevented, a



contracting negative capillary pressure will also develop, thus uniformly through the member, but only once the hydration commences and the structure sets [89]. As long as the concrete is plastic the capillary pore pressure undergoes only small changes. However, at the stage where the concrete starts to set (initial set), and a restraining skeleton of hydration products are formed, the capillary pore pressure reach an accelerating phase.

In Papers IV, V and VI, pore pressure was measured with pore pressure transducers (Model AB 0-15 PSIG from Data Instruments 0-100 kPa / overload 200 kPa), connected to a de-aired water filled system with a 50 mm long needle (cannula type) with an external/internal diameter of 0.7/0.4 mm (see Figure 89). The test was performed according to a procedure developed by Radocea [116]. The data were logged every minute using a data logger (PC-logger3100i from INTAB). The measurements were performed on sealed samples (in specimens for autogenous deformation), and on samples exposed to drying (ring-test).

Pressure gauge

De-aired water filled system

Needle (cannula) L 50 / Ø 0.4 mm

Sample

Figure 89. Picture and schematic illustration of pressure gauge and the needle used to measure the capillary pore pressure in the concrete.

It should, however, be pointed out that this test technique is highly sensitive to local conditions. This sensitivity manifests itself in, for example, a loss of pressure at a random point in time. This point was named by Wittman [9] as the breakthrough pressure. The breakthrough pressure probably occurs with local disturbances at the needle tip, such as destruction of the water menisci by air pores, which relieves the pressure. As a result, the maximum recorded pressure may not be representative for the concrete. In most of the pore pressure measurements a breakthrough point was reached, usually at negative pressure in the range of 60 to 100 kPa. Occasionally measurements as low as 110 kPa were recorded without showing a breakthrough (see Figure 90). Lower pressures may not be possible, due to the fact that water at 20°C starts boiling at 101.3 kPa. If possible, a media with a higher boiling point, instead of water (e.g. silicone oil), can be used in the measuring system.

-120

-90

-60

-30

0


Por

e pr

essu

re [k

Pa]

breakthrough

Figure 90. Example, showing the development of pore pressure for a SCC (w/c 0.55).



4.5 Concluding remarks Some concluding remarks concerning the experimental methods and materials used in the appended papers can be made as follows:

o The quantification by image analysis was found to be complicated. The method is strongly dependent on a large number of factors, such as: sample preparation, microscope performance, photo quality (sharpness), resolution, number of particles, analyzing software, threshold technique, etc.

o The shape factors, F-shape, F-circle, and Compactness, are mainly a measure of particles’ circularity, and the correlation with rheology was poor.

o The different fillers’ retained water ratio (βp) and deformation coefficient or sensitivity (Ep) show weak correlation with mortars’ rheology.

o Specific area calculated from size distribution and measured by the Blaine method is somewhat deficient. It is mainly a measure of size and size distribution, deduced for theoretical spherical particles, and shows a weak correlation with rheology.

o The BET(H2O) method for characterization of specific surface area has proven large potential. It is simple and inexpensive, and has provided stable and reliable values. It has also shown a strong correlation with the rheological properties and with the traditional BET(N2).

o The concrete dilatometer (CDD) is a potential method for linear measurements of autogenous deformation for concrete at early ages. It has the advantage that measurements can be started before setting, when the concrete is fresh. It is also relatively simple and has been shown to provide stable and reliable results.

o From measurements of temperature and pore pressure development, the underlying mechanisms of autogenous shrinkage could be verified.

o A high correlation was found between the results of the ring-tests performed in the laboratory and the field study, and between the ring-test specimens and the slab specimens.

C h a p t e r 5 – F i n a l d i s c u s s i o n a n d c o n c l u s i o n s


5 FINAL DISCUSSION AND CONCLUSIONS

5.1 Discussion The distinguishing characteristic of self-compacting concrete (SCC) is its ability to flow and consolidate under the influence of its own gravity weight. Yet it is a sensitive mix, strongly dependent on the composition and the characteristics of its constituents. Moreover, it has to possess the incompatible properties of high flowability together with high segregation resistance – a balance made possible by the dispersing effect of high-range water-reducing admixture (superplasticizer) combined with cohesiveness produced by a high concentration of fines. However, the large specific surface area of the additional filler material and the retarding effect of the superplasticizer will also promote the autogenous (sealed) shrinkage and increase the risk of early-age cracking.

In this thesis, a literature review and experimental work were carried out in order to study the rheology and early-age deformation of cementitious materials and factors affecting them. The main conclusion from this work is that, due to the large surface areas involved, quantification of geometrical and surface properties (size, shape, porosity and texture) of fillers and the fine part of the aggregates is essential for the ability to control early-age properties, such as workability and plastic shrinkage cracking, in the production of SCC.

In addition to the size and size distribution, the particles’ external characteristics were found to be of major importance, especially their shape and texture. These characteristics of three-dimensional bodies are rather difficult to describe, and it is therefore convenient to use the total surface area as a single parameter. The surface area strongly reflects the water requirement that is needed to create fluidity for the concrete. A commonly used method to determine the specific surface area is the BET method, by which the area is deduced from the amount of monolayer-adsorbed gas. The main difference between the BET method and other traditional methods (e.g. that of Blaine) to determine the particles’ specific surface area is that the BET provides more of a “real” surface area, as it includes shape and surface properties, while most others are based on the assumption that all particles are spherical and non-porous. For example, a coarser filler or gravel can provide a larger BET-area than a finer one, due to differences in surface texture and accessible porosity. However, not only the surfaces are to be covered with suspending media to generate fluidity; the voids in between the particles are to be filled, and hence the particles’ packing will also have a major influence on the flowability.

A simplified adsorption method, the BET(H2O) with water vapour as adsorbate, for characterization of particles’ specific surface area was introduced, and is proposed as a potential means of geometrical characterization for fillers and gravels. It is simple and low-cost, and has provided stable and reliable values. It has also shown a strong correlation with the traditional BET with nitrogen gas, as well as with the rheological properties. Furthermore, it was shown that it is possible to calculate the extra water needed to compensate for changes in filler and fine aggregates’ BET(H2O)-area, in order to produce SCC with small variations in flowability. An increase in BET(H2O)-area of 1000 m2/kg corresponds to a need for additional water of approximately 0.85% by mass of the gravel (or filler) content for constant flowability. E.g. a SCC with 1000 kg gravel (0-8 mm) that increases by 1000 m2/kg in area will need 8.5 litres of extra water, in order to retain its



flowability. In cases where traditional methods for geometrical characterization (size distribution, water absorption, fineness modulus, etc.) are kept within acceptance, a normal gravel for concrete can vary by up to 7000 m2/kg in surface area.

In the presented work, the influence of the BET(H2O)-area and the mix design on the early-age deformation was investigated. The linear autogenous deformation was measured with a specially developed concrete dilatometer (the CDD), with the ability to measure before setting, when the concrete is still fresh. The plastic shrinkage cracking tendency was evaluated by a modified ring-test method. To verify the underlying mechanisms of early-age deformation (autogenous and plastic shrinkage), the development of capillary pressure and temperature was recorded. In addition, to follow the concrete’s stress development and the time of cracking in the ring-test, the restraint strain on the inner ring was monitored.

It was found that, as long as the concrete is plastic, the autogenous shrinkage developed rapidly and almost linearly with time. During this period the temperature and capillary pore pressure underwent only small changes. However, the rate of the shrinkage was temporarily slowed down, indicating an initial setting of the concrete. For the SCCs used in this work, this plateau was reached at approximately 6 hours. At this point in time, both the capillary pore pressure and the temperature reached an accelerating phase, which indicated that the dormant period was ended and that the cement hydration accelerated. At about 12 hours, a final setting was reached which, for the shrinkage, was manifested in a second plateau slightly ahead of the temperature peak. After this break point, the following shrinkage developed much more slowly.

For the concretes exposed to drying, it was found that the plastic shrinkage cracking usually started at, or slightly after, initial set. The formation of plastic shrinkage cracks is attributed to the development of capillary pore pressure due to the movement or loss of water. However, the rate of evaporation strongly depended on the mix composition and was not always the governing factor for the cracking tendency. Depending on the characteristics of the concrete, different mechanisms were found to be responsible. For concrete with a high w/c-ratio, evaporation was the governing cause, whereas for concrete with a low w/c-ratio (or with the addition of silica) the autogenous shrinkage proved to be the dominant cause. A minimum in crack tendency was found at w/c 0.55. For the mixes with retardation (e.g. retarder or high superplasticizer dosage), both the autogenous shrinkage and evaporation increased, and consequently so did the crack tendency.

With an increased particle surface area (by BET method), the rate and magnitude of autogenous shrinkage increased, primarily in the plastic region, without affecting the temperature development and the times to initial and final set. It also decreased the rate of evaporation, lowered the plastic cracking tendency, and generated a higher compressive strength. When adding extra water to the mix, compensating for the loss of flowability due to increased particle surface area, the effect was the opposite (i.e. increased crack tendency, decreased autogenous shrinkage and lower compressive strength). This indicates that that the BET-area is not only characterizing the particle’s external surface area, but partly also the internal area (porosity). If the mixing water is absorbed or sufficiently bound to particles the effective w/c ratio will decrease, increasing the compressive strength. Moreover, with an increased area, the increased nucleating will also contribute to a higher strength.

Finally, the corresponding mixes without cement appeared to generate a large “autogenous” shrinkage, despite the absence of chemical reaction of cement and water (hydration). However, in these mixes, the cement was replaced with equal volume of filler, why the effect of filler was enlarged. At 24 hours the shrinkage was, after a small initial



swelling, approximately in the same magnitude after 24 hours as for the SCC’s with cement. However, the SCC’s generated no swelling and almost all shrinkage before final set (~12 hours), whereas for the mixes without cement the rate of shrinkage was lower in this period. Moreover, once set the rate of deformation for the SCC’s significantly slowing down. This indicates that not only the hydration (chemical shrinkage) is the driving mechanism of autogenous shrinkage for concrete, but also the constituent’s adsorption/absorption will contribute, especially in the fresh state before setting. For cement paste, prior initial set, the chemical shrinkage might be equal to the autogenous shrinkage. But once more or less porous particles (filler and aggregate) are introduced to the mixture the adsorption/absorption will properly introduce an additional shrinkage.

5.2 General conclusions The main conclusion of this dissertation can briefly be summarised as follows:

o Due to the importance of their large surface area, geometrical quantification of filler and the fine part of aggregate is essential for the ability to control physical properties, such as workability and plastic shrinkage, in the production of SCC.

o The specific surface area, quantified by BET technique, has been shown to be the single most important granular measure concerning the mortar and concrete flowability (rheology and slump flow) and its time dependency (loss of consistency and thixotropy), which can be explained by the fact that BET surface area comprises important factors such as particles fineness, shape, porosity and surface texture.

o Other geometrical characterizations, such as shape factors (F-shape, F-circle, and Compactness), fineness modulus and specific area calculated from size distribution and measured by Blaine was somewhat deficient and showed consistently small differences between the samples. It was difficult to draw reliable conclusions on the basis of these results, and their correlation with flowability was generally poor.

o The BET (H2O), a simplified gas adsorption method for characterisation of fillers and aggregates containing fines for SCC, was introduced. The BET (H2O) has proven high potential. It is simple, low cost, and has provided stable and reliable values. It has also shown a strong correlation with the rheological properties and with the traditional BET (Nitrogen).

o It was shown that the free water content, available to provide flowability, in 1 m3 SCC mix can vary by several litres due to differences in apparently equivalent fine aggregates. Moreover, the specific surface area of normal gravel, accepted by traditional methods for production of SCC, can vary up to 7000 m2/kg.

o It was demonstrated how an increase in BET(H2O)-area of 1000 m2/kg corresponds to an increase in water demand by approximately 0.85% by mass of filler or gravel content for constant flowability. E.g. a SCC with 1000 kg gravel (0-8 mm) that increases by 1000 m2/kg in area will need 8.5 litre extra water, in order to retain its consistency.

o A specially developed concrete dilatometer for linear measurements of autogenous (sealed) deformations was introduced. It has the advantage that measurements can be started before setting, when the concrete is fresh. It is also relatively simple and has been shown to provide stable and reliable results. The method and the underlying mechanisms were verified by parallel measurements of capillary pore pressure.



o Increased particle surface area did not only decrease the SCC flowability, it also decreased the rate and magnitude of evaporation and consequently reduced the plastic crack tendency, despite an increased autogenous shrinkage. Moreover, an increased area resulted in a higher compressive strength. However, with additional water for constant flowability, compensating for the differences in BET(H2O)-area, evaporation and plastic cracking tendency was increased significantly, and strength was reduced.

o For concrete with a high w/c-ratio, evaporation was shown to be the main governing mechanism for plastic shrinkage cracking, whereas for concrete with a low w/c-ratio the autogenous shrinkage proved to be dominating. For mixes with retardation, both the autogenous shrinkage and evaporation increased, and consequently so did the crack tendency.

o It is proposed that not only the hydration (chemical shrinkage) is the driving mechanism of autogenous shrinkage for concrete, but also the constituent’s adsorption/absorption will contribute, especially in the fresh state before setting.

5.3 Suggestions for future research Materials science is a fundamental field within the area of civil engineering. Although the research and literature relating to early-age properties of cementitious materials are extensive, there are still questions to be answered and further research is needed. In general, if the early-age behaviour and properties are to be better understood, it is necessary to increase the knowledge and understanding of the mechanisms of interparticle forces and how they are acting in this complex composite mix.

For measurements of flowability and early-age deformation there is a need of general guidelines, standards and recommendations. Methods, methodologies, apparatus, models, etc. have to be defined. For flowability measurements of SCC there are well-established empirical test methods, whereas there is no general agreement on how to rheologically characterize the concrete. Surprisingly, for mortars and pastes there are even wider disagreements. Neither is there general agreement on how to define and characterize the time-dependent reversible (thixotropic) and irreversible behaviour. In the area of early deformation, there is even greater disagreement on what and how to measure. As several of these measures are only relative, general guidelines and standards are needed. Furthermore, the terminology and definitions are somewhat inconsistent and deficient. In the area of particle characterization, there are well-established guidelines and standardized methods, but still there is a lack of simple methods for quality control of the properties of filler and fines that have a clear connection with the fresh concrete behaviour.

Focusing on the matters identified in this work, the following suggestions for future research are made.

(1) Develop a methodology to perform measurements with the BET(H2O) method in a shorter time (normally a measurement takes at least 6-8 days). The working effort is not large, but it might be preferable to speed up the procedure. A suggested method is to do the conditioning at lower pressure. By placing the samples in the desiccator with the saturated salt solution and evacuating the air, the process of adsorption will be faster and the time to equilibrium shorter. Another suggested method, to speed up equilibrium formation, might be to perform the conditioning of samples at a higher temperature. A third method could



perhaps be to follow the sample’s weight as a function of time, frequently and long enough to produce a smooth curve with a clear tendency, and extrapolate to an equilibrium state.

(2) Evaluate the effect of particles’ (filler, gravel and aggregate) adsorption/absorption with time on the autogenous shrinkage of mortars and concretes. The hydration is, in general, considered to be the single driving force of autogenous shrinkage. But the measurements (in Paper IV) on mixtures without cement indicated a substantial “autogenous” shrinkage, after an initial swelling. Consequently, the contribution of adsorption/absorption to the autogenous shrinkage ought to be significant, especially in the fresh state, before initial setting. More measurements need to be made, using different compositions and materials, and to be complemented with pore pressure measurements on mixtures without cement.

(3) Develop methodology for measuring the plastic shrinkage, using the concrete dilatometer (CDD) with an open system where the samples are exposed to evaporation. If the measurements are made simultaneously with the sealed system, the evaporation shrinkage can be extracted from the plastic shrinkage. Yet the plastic shrinkage will only be relative, and so will the evaporation shrinkage, as these measures will depend on the geometry and diffusion coefficient of the mould, exposed area, air velocity, RH, temperature, etc. Within this work, an attempt was made to measure the plastic shrinkage with the CDD, by replacing the PE-tube with a felted cloth (geotextile/fabric; see Figure 91). But as this mould did not have the radial stiffness (e.g. from corrugation and steel wire coil), the shrinkage in the plastic stage was mainly transformed to vertical displacements and almost no linear measurements were recorded in this stage. Hence, the open (diffusion) mould needs to be designed with a relation between the volumetric, radial and longitudinal deformations, equal to the sealed mould.

Figure 91. Measurements with the CDD: (a) under sealed conditions, and (b) when exposed to evaporation (plastic shrinkage).

In the present study, the fundamental mechanisms affecting the rheology of cementitious materials were theoretically reviewed. However, to be able to practically use and apply these theories in “reality”, a large experimental study needs to be made and the results compared with those in the literature. The study should focus on the mix design of high-flowable concretes (e.g. SCC), in which the proportions of the constituents and their properties are of major importance. Recommendations or instructions for how the proportions and properties affect the workability/rheology of concrete should be given, e.g. by designing a flow diagram with orientation (see Figure 92). Also how different measures of the constituents, by specific methods, affect the workability/rheology are to be given. To be able to ensure the quality and hopefully lower the cost in the production of SCC, such a guideline could be a helpful tool.

(b)

(a)



Yie

ld st

ress

, σ0 [

Pa]

Viscosity (plastic), ηpl [Pa·s]

+Cement fineness

FLOW DIAGRAM

+Silica/Cement +Cons. loss +Sand/Stone +Cement

+Filler & fines

+Irregularity

+Elongation

+Paste volume

+Slagg/Cement+Flyash/Cement +Water

+Air

+SP

Figure 92. Illustration of a flow diagram, showing how changes in proportions and properties of constituents may affect the flow behaviour (rheologically in terms of the Bingham parameters σ0 and ηpl) of fresh concrete. This is a potential aid in mix design, showing how to best orientate towards a desired point of consistency.

C h a p t e r 6 – R e f e r e n c e s


6 REFERENCES

[1] Okamura H., Ouchi M., “Self-compacting concrete. Development, present use and future”,

Proceedings of the 1st International Symposium on Self-Compacting Concrete, pp. 3-14, Stockholm, 1999,

[2] Rilem Technical Committee, “Final report of RILEM TC 188-CSC casting of self-compacting concrete”, Materials and Structures, 2006.

[3] Skarendahl Å., Petersson Ö., “Self-compacting concrete: State-of-the-art report of RILEM technical committee 174-SCC”, RILEM report 23, France, 2000.

[4] Powers T.C., “The properties of fresh concrete”, New York, 1968.

[5] Esping O., “Rheology of cementitious materials: effects of geometrical properties of filler and fine aggregate”, Chalmers University of Technology, Göteborg, 2004.

[6] Ferraris, C.F. Karthik H.O., Russell H., “The influence of mineral admixtures on the rheology of cement paste and concrete”, Cement and Concrete Research, Vol. 31, pp. 245-255, 2001.

[7] Gallias J.L., “The effect of fine mineral admixtures on water requirement of cement pastes”, Cement and Concrete Research, Vol. 30, pp. 1543-1549, 2000.

[8] Neville A.M., “Properties of Concrete”, London, 1963.

[9] Wittmann, F.H., “On the Action of Capillary Pressure in Fresh Concrete”, Cement and Concrete Research, Vol.. 6, pp. 49-56, 1976.

[10] Bjøntegaard Ø., Hammer T.A., Sellevold E.J., “Cracking in high performance concrete before setting”, Proceedings of the International Symposium on High Performance and Reactive Powder Concretes, Sherbrooke, 1998.

[11] Potgier J.H., Strydom C.A., “An investigation into the correlation between different surface area determination techniques applied to various limestone-related compounds”, Cement and Concrete Research, Vol. 26, pp.1613-1617, 1996.

[12] Lowell S., Shields J. E., “Powder surface area and porosity”, Chapman & Hall, London, 1991.

[13] Barnes H.A., Hutton J. F., Walters K., “An introduction to rheology”, Elsevier, Amsterdam, 1989.

[14] Ferguson J., Kemblowski, “Applied fluid rheology”, Elsever Applied Science, Chapman & Hall, 1991.

[15] Whorlow R.W., “Rheological techniques”, Ellis Harwood, Chichester, New York, 1992.

[16] Flatt R., “Towards a prediction of superplasticized concrete rheology”, Materials and Structure, Vol. 37, pp. 289-300, 2004.

[17] Gotoh K., Masuda H., Higashitani K., “Powder technology handbook”, 2nd Ed., New York, 1997.

[18] Kreijger P.C., “Plasticizers and dispersing admixtures”, Proceedings of the 1st International Congress on Admixtures, Concrete Society, London, 1980.

[19] Banfill P.F.G., “The rheology of fresh cement and concrete - A review”, Proceedings of the 11th International Cement Chemistry Congress, Durban, 2003.



[20] Eirich F.R., “Rheology – Theory and applications”, Vol. 3, New York, 1960.

[21] Glanville W.H., Collins A.E., Matthews D.D., “The grading of aggregates and workability of concrete”, Road Research Technical Paper, No. 5, London, 1947.

[22] Mindess S., Young J. F., Darwin D., “Concrete”, 2nd Ed., Prentice Hall, New Jersey, 2003.

[23] Neville A.M., “Properties of concrete”, 4th Ed., Prentice Hall, London, 2000.

[24] Tattersall G.H., “Workability and quality control of concrete”, E&FN Spon, London, 1991.

[25] Struble L.J., “The rheology of fresh cement paste”, Advanced Cement Materials, American Ceramic Society, pp. 7-26, 1990.

[26] Ramachandran, V.S., Beaudoin J.J., “Handbook of analytical techniques in concrete science and technology - Principles, techniques, and applications”, New York, 2001.

[27] Russel W.B., Saville D.A., Schowalter W.R., “Colloidal Dispersions”, Cambridge University Press, 1992.

[28] Tattersall G.H., Banfill P.F.G., “Rheology of fresh concrete”, Pitman, London, 1983.

[29] Billberg P., “Form pressure generated by self-compacting concrete - Influence of thixotropy and structural behaviour at rest”, Division of Concrete Structures, Royal Institute of Technology, Stockholm, 2006.

[30] Hackley V.A., Ferraris C.F., “Guide to rheological nomenclature: Measurements in ceramic particulate systems”, NIST special publication, No 946, National Institute of Standards and Technology, Washington, 2001.

[31] Atzeni C., Massidda L. and Sanna U., “Comparison between rheological models for Portlant cement pastes”, Cement and Concrete Research, Vol. 15(3) , pp. 511-519, 1985.

[32] Barnes H.A., “A handbook of elementary rheology”, Elsevier, Amsterdam, 1999.

[33] Lapasin R., Papo A. and Rajgelj S., ”Flow behaviour of fresh cement pastes - A comparison of different rheological instruments and techniques”, Cement and Concrete Research, Vol. 13(3), pp. 346-349, 1983.

[34] Papo A., “Rheological models for cement pastes”, Materials and Construction, Vol. 21, pp. 41-46, 1988.

[35] Wallevik O., “Den ferske betongens reologi og anvendelse på betong med og utan tilsetning av silikastov”, The Norwegian Institute of Technology - NTH, Division of Building Materials, 1990.

[36] Norberg J, “The rheology of fresh concrete - tests on ready mixed concrete”, CBI report 2:94 (in Swedish), Stockholm, 1994.

[37] Atzeni C., Massida L., Sanna U., “Comparison between rheological models for Portland cement pastes”, Cement and Concrete Research, Vol. 15, pp. 511-519, 1985.

[38] Banfill P.F.G., “Rheological methods for assessing the flow properties of mortar and related materials”, Construction and Building Materials, Vol. 8(1), pp. 43-50, 1994.

[39] de Larrard F., Feraris C., Sedran T., “Fresh concrete, a Herschel-Bulkley material”, Materials and Structures, Vol. 31, pp. 494-498, 1998.

[40] Bager D.H., Geiker R.M., Jensen R.M., “Rheology of self-compacting mortars, influence of particle grading”, Nordic Concrete Research, Vol. 26, pp. 1-16, 2001.

[41] Geiker R.M., Brandl M., Thrane N.L., Bager D.H., Wallevik O., “The effect of measuring procedure on the apparent rheological properties on self-compacting concrete”, Cement and Concrete Research, Vol. 31, pp. 1791-1795, 2002.



[42] Wallevik O.H., “Rheology – A scientific approach to develop self-compacting concrete”, Proceedings of the 3rd International RILEM Symposium on Self-Compacting Concrete, PRO 33, 17–20 August, pp. 208-219, Reykjavik, 2003.

[43] Nguyen Q.D., Boger D.V., “Yield stress measurement for concentrated suspensions”, Journal of Rheology, Vol. 27, pp. 321-349, 1983.

[44] Domone P., “The slump flow test for high-workability concrete”, Cement and Concrete Research, Vol. 28(2), pp. 177-182, 1998.

[45] Tattersall G.H., “The rationale of a two-point workability test,” Magazine of Concrete Research, Vol. 25(84), pp. 169-172, 1973.

[46] Chidiac S.E., Madaani O., Razaqpur A.G., Mailvaganam, N.P., “Controlling the quality of fresh concrete - A new approach”, Magazine of Concrete Research, Vol. 52(5), pp. 353-364, 2000.

[47] Clayton S., Grice T.G., Boger D.V., “Analysis of the slump test for on-site yield stress measurement of mineral suspensions”, Journal of Mineral Processing, Vol. 70, pp. 3-21, 2003.

[48] Larrard D.F., “Concrete mixture proportioning, a scientific approach”, F&FN Spon, New York, 1999.

[49] Larrard D.F., Hu C., Sedran T., Szitkar J.C., Joly M., Claux F., Derkx F., “A new rheometer for soft-to-fluid fresh concrete”, ACI Materials Journal, Vol. 94(3), pp. 234-243, 1997.

[50] Ferraris C.F., de Larrard F., “Testing and modelling of fresh concrete rheology”, NISTIR 6094, NIST, USA, 1998.

[51] Hu C., de Larrard F., Sedran T., Boulay C., Bosc F., Deflorenne F., “Validation of BTRHEOM, the new rheometer for soft-to-fluid concrete”, Materials and Structures, Vol. 29(194), pp. 620-631, 1996.

[52] Murata J., “Flow and deformation of fresh concrete”, Materials and Structures, Vol. 17(98), pp. 117-129, 1984.

[53] Murata J., Kikukawa H., “Viscosity Equation for Fresh Concrete”, ACI Materials Journal, Vol. 89(3), pp. 230-237, 1992.

[54] Norberg J., Peterson O., Billberg P., “Effects of a new generation of superplasticizers on the properties of fresh concrete”, Proceedings of the CANMET/ACI International Conference, pp.. 583-598, Italy, 1997.

[55] Pashias N., Boger D.V., Sumers J., Glenister D.J., “A Fifty Cent Rheometer for Yield Stress Measurements”, Journal of Rheology, Vol. 40(6), 1996.

[56] Saak A.W., Jennings H.M., Shah, S.P., “A generalized approach for the determination of yield stress by slump and slump flow”, Cement and Concrete Research, Vol. 34, pp.. 363-371, 2004.

[57] Schowalter W.R., Christensen G., “Toward a rationalization of the slump test for fresh concrete -Comparisons of calculations and experiments”, Journal of Rheology, Vol. 42(4), pp. 865-870, 1998.

[58] Tanigawa Y., Mori H., “Analytical study on deformation of fresh concrete”, Journal of Engineering Mechanics, Vol. 115(3), pp. 493-508, 1989.

[59] Tanigawa Y., Mori H., Watanabe, K., “Computer simulation of consistency and rheology tests of fresh concrete by viscoplastic”, Proceedings of the RILEM Symposium on Properties of Fresh Concrete, pp. 301-308, Hanover, 1990.

[60] Wallevik J.E., “Relationship between the Bingham parameters and slump”, Cement and Concrete Research, Vol. 36, pp. 1214-1221, 2006.



[61] Flatt R.J., Domenico L., Roussel N., “Linking yield stress measurements: Spread test versus Viscomat”, Cement and Concrete Research, Vol. 36(11), pp. 99-109, 2006.

[62] Rossel N., N’Guyen H.T.L., Coussot P., “Yield measurements using stoppage tests”, Proceedings of the RILEM Symposium on Self-Compacting Concrete, pp. 575-582, Chicago, 2005.

[63] Saak A.W., Jennings H.M., Shah S.P., “A generalized approach for the determination of yield stress by slump and slump flow”, Cement and Concrete Research, Vol. 34(3), pp. 363-371, 2004.

[64] Domone P.L. and Jin J., “Properties of mortar for self-compacting concrete”, Proceedings of the RILEM Symposium on Self-Compacting Concrete, pp. 109-120, Stockholm, 1999.

[65] Grünewald S., Walraven J.C., “Rheological measurements on self-compacting fibre reinforced concrete”, Proceedings of the RILEM Symposium on Self-Compacting Concrete, PRO 33, pp. 49-58, Reykjavik, 2003.

[66] Jin J., Domone P.L., “Relationships between the fresh properties of SCC and its mortar component”, Proceedings of the North American Conference on the Design and Use of Self-Consolidating Concrete, pp. 33-38, Chicago, 2002.

[67] Nielsen I., Wallevik O.H., “Rheological evaluation of some empirical test methods – Preliminary result”, Proceedings of the RILEM symposium on Self-Compacting Concrete, pp. 55-68, Reykjavik, 2003.

[68] Emborg M., “Rheology tests for self-compacting concrete - How useful are they for the design of concrete mix for full scale production?”, Proceedings of the RILEM symposium on Self-Compacting Concrete, pp. 95-105, Stockholm, 1999.

[69] Cauberg N., Desmyter J., Dieryck V., “Rheology of Self-Compacting Concrete - Validation of Empirical Test Methods”, Proceedings of the RILEM Symposium on Self-Compacting Concrete, pp. 765-773, Chicago, 2005.

[70] Smeplass S., “Applicability of the Bingham model to high strength concrete”, Proceedings of the RILEM Workshop on Special Concretes – Workability and Mixing, pp. 145-151, Paisley, 1993.

[71] Takada K., Tangtermsirikul S., “Testing of Fresh Concrete”, Self-compacting concrete: State-of-the-art report, RILEM Technical Committee 174-SCC, Report 23, pp. 25-39, France, 2000.

[72] Ferraris C., Obla K., Hill R., “The influence of mineral admixtures on the rheology of cement paste and concrete”, Cement and Concrete Research, Vol. 31(2), pp. 245-255, 2001.

[73] Utsi S., Emborg M., Carlswärd J., “Relation between workability and rheological parameters”, Proceedings of the RILEM Symposium on Self-Compacting Concrete, PRO 33, pp. 154-164, Reykjavik, 2003.

[74] Bartos P., “Fresh concrete: Properties and tests”, Department of Civil Engineering, Paisley Collage, Scotland, 1992.

[75] Thielen G., Kordts S., Spanka G., “Rheological and setting behavior of mortar and concrete”, Proceedings of the 1st fib Congress on Development of New Materials, pp. 343-350, Osaka, 2002.

[76] Assaad J., Khayat K.H., Mesbah H., “Assessment of thixotropy of self-compacting concrete”, ACI Material Journal, Vol. 100(2), pp. 99-107, 2003.

[77] Papo A., “The thixotropic behavior of white Portland cement pastes”, Cement and Concrete Research, Vol. 18(4), pp. 595-603, 1988.



[78] Wallevik O.H., Nielsson I., “Self-compacting concrete - A rheological app.roach”, Proceedings of International Workshop on SCC, Kochi, 1998.

[79] Hackley V.A., Ferraris C. F., “The Use of Nomenclature in Dispersion Science and Technology”, NIST publication No 960-3, 2001.

[80] Macosco, C.W., “Rheology Principles, measurements and applications”, VCH Publishers, New York, 1994.

[81] Barcelo L., Boivin S., Rigaud S., Acker P., Clavaud B., Boulay C., “Linear vs. Volumetric autogenous shrinkage measurement: Material behaviour or experimental artefact?”, Proceedings of the 2nd International Seminar on Self-desiccation and its Importance in Concrete Technology, pp. 109-125, Lund, 1999.

[82] Bentur A., “Introduction: Overview of early age cracking”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 1-11, 2002.

[83] Geiker, M.R., Bentz, D.P., and Jensen, O.M., “Mitigating Autogenous Shrinkage by Internal Curing”, ACI SP-218, High Performance Structural Lightweight Concrete, American Concrete Institute, pp. 143-154, Farmington Hills, 2004.

[84] Igarashi S.I., Watanebe A., “Experimental study on prevention of autogenous deformation by internal curing using super-absorbent polymer particles”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 77-86, Lyngby, 2006.

[85] Lura P., Durand F., Jensen O.J., “Autogenous strain of cement pastes with superabsorbent polymers”, Proceedings of the international RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 57-65, Lyngby, 2006.

[86] Mechtcherine V., Dudziak L., Schulze J., “Internal curing by super absorbent polymers (SAP) – effects on material properties of self-compacting fibre-reinforced high performance concrete”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 87-96, Lyngby, 2006.

[87] Mönnig S., Reinhardt H.W., “Results of a comparative study of the shrinkage behaviour of concrete and mortar mixtures with different internal water sources” Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 67- 75, Lyngby, 2006.

[88] Piérard J., Pollet V., Cauberg N., “Mitigating autogenous shrinkage in HPC by internal curing using superabsorbent polymers”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 97-106, Lyngby, 2006.

[89] Holt E.E., “Early age autogenous shrinkage of concrete”. Technical Research Centre of Finland, VTT Publications 446, 2001.

[90] Mehta P.K., Monterio J.M., “Concrete: Structure, properties and materials”, 2nd Ed., Prentice Hall, 1993.

[91] Esping O., Löfgren, I., “Cracking due to plastic and autogenous shrinkage – Investigation of early age deformation of self-compacting concrete – Experimental study”, Publication 05:11, Department of Building Technology, Chalmers University of Technology, Göteborg, 2005.

[92] Haas D.G.D., Kreijger P.C., Niel E.M.M.G., Slagter J.C., Stein H.N., Theissing E.M., van Wallendael M., “The shrinkage of hardening cement paste and mortar”, Cement and Concrete Research, Vol. 5, pp. 295-320, 1975.

[93] Justnes H., Clemmens F., Depuydt P., Gemert, V.D., Sellevold, E.J., “Correlating the deviation point between external and total chemical shrinkage with setting time and other



characteristics of hydrating cement paste”, Proceedings of the International RILEM Workshop on Shrinkage of Concrete, pp. 57-74, Paris, 2000.

[94] Reinhardt, H.W., Große, C.U., Herb, A.T., “Ultrasonic monitoring of setting and hardening of cement mortar – A new device”, Materials and Structures, Vol. 33, pp. 580-583, 2000.

[95] Kasai Y., Yokoyama K., Matsui I., “Tensile properties of early age Concrete”, Mechanical Behavior of Materials, Society of Materials Science, Vol. 4, Japan, pp. 288 – 299, 1972.

[96] Mihashi H., Paulo J., Leite D.B., “State-of-the-art report on control of cracking in early age concrete”, Journal of Advanced Concrete Technology, Vol. 2(2), pp. 141-154, 2004.

[97] Holt E., "Contribution of mixture design to chemical and autogenous shrinkage of concrete at early ages", Cement and Concrete Research, Vol. 35(3), pp. 464 - 472, 2005.

[98] Uno P.J., “Plastic shrinkage cracking and evaporation formulas”, ACI Material Journal, Vol. 95(4), pp. 365-375, 1998.

[99] Bazant Z.P., Wittman F.H., “Creep and shrinkage in concrete structures”, John Wiley & Sons, New York, 1982.

[100] Powers T.C., Brownyard T.L., “Studies of physical properties of hardened Portland cement past”, Journal of American Institute, Vol. 43, Bulletin 22, Research Laboratories of Portland Cement Association, Chicago, 1948.

[101] Gartner E.M., Young J.F., Damidot D.A., Jawed I., “Hydration of Portland cement”, in Structure and Performance of Cements (Ed. Bensted & Barnes) Spoon Press, London, 2002.

[102] Locher F.W., Richartz W., Sprung S., “Erstarren von Zement” (in German), Zement Kalk Gips, Vol. 29(10), pp. 435-442, 1976.

[103] Justens H., Van Gemart, A., Verboven F. and Sellevold E.J., “Total and external chemical shrinkage of low w/c ratio cement pastes”, Advances in Cement Research, Vol. 8(31), pp.121-126, 1996.

[104] Jensen O.M., “Autogenous phenomena in cement-based materials”, Department of Building Technology and Structural Engineering, Aalborg University, Denmark, 2005.

[105] Bentur A., “Terminology and definitions”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 13-15, 2002.

[106] Bovin S., Acker P., Rigaud S., Clavaud B., “Experimental assessment of chemical shrinkage of hydrating cement pastes”, Proceedings of the International Workshop on Autogenous shrinkage of Concrete, pp. 77-88, Hiroshima, 1998.

[107] Knudsen T., Geiker M., “Chemical shrinkage as an indicator of the stage of hardening”, International RILEM Conference on Concrete at Early-Ages, Vol. 2, pp. 163-166, Paris, 1982.

[108] Hammer T.A., “Test methods for linear measurements of autogenous shrinkage before setting”, Autogenous Shrinkage of Concrete, (Ed. Tazawa) E & FN Spon, pp. 143-154, London, 1999.

[109] Justnes H., Sellevold E.J., Reyniers B.R., “Chemical shrinkage of cementitious pastes with mineral additives”, Proceedings of the 2nd International Seminar on Self-desiccation and its importance in concrete technology, pp. 73-84, Lund, 1999

[110] Hammer T.A., “Early age chemical shrinkage and autogenous deformation of cement paste”, Proceedings of the International Seminar on Self-desiccation and its importance in concrete technology, pp. 27-33, Lund, 1999.

[111] Lura P., “Autogenous deformation and internal curing of concrete”, Ph.D. thesis, Delft Institute of Technology, Delft, 2003.



[112] Lyman C.G., “Growth and movement in Portland cement concrete”, Oxford University Press, pp. 25-45, London, 1934.

[113] Davis H.E., “Autogenous Volume changes of concrete”, Proceedings of the American Society Tests Materials, Vol. 32(40), pp. 1103-1112, 1940.

[114] Jensen O.M., Hansen P.F., “Autogenous deformation and RH-changes perspective”, Cement and Concrete Research, Vol. 31, pp. 1859-1865, 2001.

[115] Bjøntegaard Ø., “Thermal dilation and autogenous deformation as driving forces to self-induced stresses in high performance concrete”, PhD thesis, Division of Structural Engineering, The Norwegian University of Science and Technology, Trondheim, 1999.

[116] Radocea A., “A Study on the mechanisms of plastic shrinkage of cement-based materials”, PhD thesis, Chalmers University of Technology, Göteborg, 1992.

[117] Tazava E.I., Miyazawa S., “Influence of constituents and composition on autogenous shrinkage of cementitious materials”, Magazine of Concrete Research, Vol. 49(178), pp. 15-22, 1997.

[118] Aïctin, P.C., “Autogenous shrinkage measurement”, in Autogenous shrinkage of concrete (Ed. Tazawa), E & FN Spon, pp. 257-268, 1999.

[119] Hammer T.A., Bjøntegaard Ø., “Testing of autogenous deformation (AD) and thermal dilation (TD) of early age mortar and concrete – Recommended test procedure”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 341–346, Lyngby, 2006.

[120] Liu J.P, Tian Q., Sun W., Miao C.W., Tang M.S., “Study on the self-desiccation effect in early-age concrete and the determination of ‘time-zero’ of self-desiccation shrinkage”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 401 – 410, Lyngby, 2006.

[121] Sant G., Lura P., Weiss J., “A discussion of analysis approaches for determining ‘time-zero’ from chemical shrinkage and autogenous strain measurements in cement paste”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp.. 375–383, Lyngby, 2006.

[122] Weiss J., “Experimental determination of the ‘time zero’ t0 (maturity-zero M0)”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 195-206, 2002.

[123] Hammer T.A., Bjøntegaard Ø., Sellevold E.J., “Measurement methods for testing of early age autogenous strain”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 234-245, 2002.

[124] Esping O., Löfgren I., “Investigation of early age deformation in self-compacting concrete”, Proceedings of the 2nd International Symposium on Advances in Concrete Science, Quebec, 2006. (Note, attached paper V in this thesis)

[125] Nawa T., Horita T., “Autogenous shrinkage of high-performance concrete”, Proceedings of the International Workshop on Microstructure and Durability to Predict Service Life of Concrete Structures, Sapporo, 2004.

[126] Hedenblad H., Janz M., “Inverkan av alkali på upp.mätt RH I betong”, Report TVBM-3057, Department of Building material. Lund University of Technology, 1994. (in Swedish)

[127] Gregg S. J., Sing K. S. W., “Adsorption, surface area and porosity”, Academic Press, New York, 1967.

[128] Schere G.W., “Structure and properties of gels”, Cement and Concrete Research, Vol. 29, pp. 1149-1157, 1999.



[129] Bazant Z.P., “Thermodynamics of hindered adsorption and its implications for hardened cement paste and concrete”, Cement and Concrete Research Vol. 2(1) pp. 1-16, 1972.

[130] Lura P., Jensen O.M., Breugel V.K., “Autogenous shrinkage in high-performance cement paste: An evaluation of basic mechanisms”, Cement and Concrete Research, Vol. 33, pp. 223-232, 2003.

[131] Ferraris C.F., Wittman F.H., “Shrinkage mechanisms of hardened cement paste”, Cement and Concrete Research, Vol. 17, pp. 453-464, 1987.

[132] Bentur A., Kovler K., “Evaluation of early age cracking characteristics in Cementitious systems”, Materials and Structures, Vol. 36, pp. 183-190, 2003.

[133] Jensen O.M., Hansen F., “A dilatometer for measuring autogenous deformation in hardening Portland cement paste”, Materials and Structures, Vol. 28(181), pp. 406-409, 1995.

[134] Hammer T.A., “Is there a relationship between pore water pressure and autogenous shrinkage before and during setting?” Proceedings of the 3rd International Research Seminar on Self-desiccation and its Importance in Concrete Technology, pp. 27-38, TVBM-3104, Lund, 2002.

[135] Persson B., “Experimental studies of the effect of silica fume on chemical shrinkage and self-desiccation in Portland cement mortars”, Proceedings of the International Seminar on Self-desiccation and its Importance in HPC Technology, TVBM-3075, pp. 116-13, Lund, 1997.

[136] Gangé R., Aouad I., Shen J., Poulin C., “Development of a new experimental technique for the study of autogenous shrinkage of cement paste”, Materials and Structures, Vol. 32, pp. 635-642, 1999.

[137] Geiker M., “Studies of Portland cement hydration: Measurement of chemical shrinkage and a systematic evaluation of hydration curves by means of the dispersion model”, PhD Thesis, Technical University of Denmark, Lyngby, 1983.

[138] Charron J.P., Marchand J., Bissonnette B., Pigeon M., Zuber B., “Comparative study of the effects of water/binder ratio and silica fume on the Volume instability of hydrating cement pastes”, Proceedings of the 3rd International Research Seminar on Self-desiccation and its Importance in Concrete Technology, pp. 39-49, TVBM-3104, Lund, 2002.

[139] Justnes H., Sellevold E.J., “The influence of cement characteristics on chemical shrinkage”, Autogenous Shrinkage of Concrete, (Ed. Tazawa) E & FN Spon, pp. 71-92, London, 1999.

[140] Geiker M., Knudsen T., “Chemical shrinkage of Portland cement pastes”, Cement and Concrete Research, Vol. 12, pp. 603-610, 1982.

[141] Paulini P., “A weighing method for cement hydration”, Proceedings of the 9th International Congress on Chemistry of Cement, pp. 248-254, New Delhi, 1992.

[142] Bjøntegaard Ø., Hammer T.A., Sellevold E.J., “On the measurement of free deformation of early age cement paste and concrete”, Cement and Concrete Composites, Vol. 26, pp. 427-435, 2004.

[143] Justnes H., Gemert V.A.., Verboven F., Sellevold E.J.,, “Total and external chemical shrinkage of low w/c ratio cement pastes”, Advances in Cement Research, Vol. 8(31), pp. 121-126, 1996.

[144] Sellevold E.J., Bjøntegaard Ø., Justnes H., Dahl P.A., “High performance concrete: Early Volume changes and cracking tendency”, Thermal Cracking in Concrete at Early Ages, E&FN Spon, 1994.

[145] Holt E., “Very early age autogenous shrinkage: Governed by chemical shrinkage or self-desiccation?”, Proceedings of the 3rd International Research Seminar on Self-desiccation and its Importance in Concrete Technology, TVBM-3104, pp. 1-25, Lund, 2002.



[146] Hammer T.A., Bjøntegaard Ø., Sellevold E.J., “Measurement methods for testing of early age autogenous strain”, Proceedings of the RILEM International Conference on Early Age Cracking in Cementitious Systems, pp. 217-228, 2001.

[147] Tomosawa F., Noguch T., Park K.B., “Experimental determination and analysis of stree and strain distribution of reinforced high-strength concrete column caused by self-desiccation and heat of hydration”, Proceedings of the 1st International Research Seminar on Self-desiccation and its Importance in Concrete Technology, TVBM 3075, pp. 99-115, Lund, 1997.

[148] Radocea A., “Autogenous Volume change of concrete at very early age - Model and experimental data”, Proceedings of the 1st International Research Seminar on Self-desiccation and its Importance in Concrete Technology, TVBM 3075, pp. 56-70, Lund, 1997.

[149] Hansen K.K., Jensen O.M., “Equipment for measuring autogenous RH-change and autogenous deformation in cement paste and concrete”, Proceedings of the 1st International Research Seminar on Self-desiccation and its Importance in Concrete Technology, TVBM 3075, pp. 27-30, Lund, 1997.

[150] Mounanga P., Loukiki A., Bouasker M., Khelidj A., Coué R., “Effect of setting retarder on the early age deformations of self-compacting mortars”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 311-320, Lyngby, 2006.

[151] Esping O., “Investigation of autogenous deformation in self-compacting concrete”, Proceedings of the RILEM Conference on Volume Changes of Hardening Concrete, pp. 273-282, Lyngby, 2006. (Note, attached paper IV in this thesis)

[152] Leivo M., Holt H., “Autogenous Volume changes at eraly ages”, Proceedings of the 1st International Research Seminar on Self-desiccation and its Importance in Concrete Technology, TVBM 3075, pp. 88-98, Lund, 1997.

[153] Kronlof A., Leivo M., Sipari P., “Experimental study on the basic phenomena of shrinkage and cracking of fresh mortar”, Cement and Concrete Research, Vol. 25(8), pp. 1747-1754, 1995.

[154] Okamura H., “Self-compacting high-performance concrete”, Concrete International, Vol. 19(7), pp. 50-54, 1997.

[155] Allen T., “Particle size measurement – Surface area and pore size determination”, Chapman & Hall, London, 1997.

[156] Brunauer S., “The adsorption of gases and vapours”, Vol. 1, Oxford, 1944.

[157] Ruthven D. M., “Principles of adsorption and adsorption processes”, Wiley-Interscience, New York, 1984.

[158] Brunauer S., Emmet H. P., Teller., “Adsorption of gasses in multimolecular layers”, American Chemical Society, 1938.

[159] Ahlgren L., “Moisture fixation in porous building materials”, Division of Building Technology, Lund Institute of Technology, Report 36, Sweden (in Swedish), 1972.

[160] Rouquerol F., Rouquerol J., Sing K.S.W., “Adsorption by powders and porous solids: Principles, methodology and applications”, Academic Press, 1999.

[161] Hansen T.C., “Physical structure of hardened cement paste”, Materials and Structures, Vol. 19(9), pp. 423-436, 1986.

[162] Fuglsang Nielsen L., “Moisture sorption in porous material - A modified BET-description”, Proceedings of the Symposium on Building Physics in Northern Countries, pp. 719-724, Vol. 2, Copenhagen, 1993.



[163] Odler I., “The BET-specific surface area of hydrated Portland cement and related materials”, Cement and Concrete Research, Vol. 33, pp. 2049-2056, 2003.

[164] Robens E., Benzler B., Büchel G., Reichert H., Schumacher K., “Investigation of characterizing methods for the microstructure of cement”, Cement and Concrete Research, Vol. 32, pp. 87-90, 2002.

[165] Ramachandran V.S., Malhotra V.M., “Concrete admixtures handbook, Properties, science and technology”, New Jersey, 1995.

[166] Badmann R., Stockhausen N., Setzer M.J., “The statistical thickness and the chemical potential of adsorbed water films”, Journal of Colloid and Interface Science, Vol. 82(2), pp. 534-542, 1981.

[167] Hagymassy J., Brunauer J.R., Mikhail R.S., “Pores structure analysis by water vapour adsorption”, Journal of Colloid and Interface Science, Vol. 29(3), pp. 485-491, 1969.

[168] Okamura H., Ozawa K., “Mix design for self-compacting concrete”, Concrete Library of the JSCE, No 25, pp. 107-120, June, (Translation from Proceedings of ISCE, No 496/V-24, 1994.8), 1995.

[169] Domone P. L., Chai H., “Testing of binders for high performance concrete”, Cement and Concrete Research, Vol. 27(8), pp. 1141-1147, 1997.

[170] Bartos P., Sonebi M., Tamimi A.K., “Workability and rheology of fresh concrete: Compendium of tests”, RILEM, France, 2002.

[171] Billberg P., “Influence of filler characteristics on SCC rheology and early hydration”, 2nd International RILEM Symposium on Self-Compacting Concrete, Japan, pp. 285-294, 2001.

[172] Poppe A.M., Schutter D.G., “Modelling of the heat development during cement hydration in self-compacting concrete”, Proceedings of the 4th International RILEM Symposium on Self-Compacting Concrete, pp. 271-277, Chicago, 2005.

[173] Sari M., Prat E., Labastire J.K., “High strength self-compacting concrete - Original solutions associating organic and inorganic admixtures”, Cement and Concrete Research, Vol. 29(6), pp. 813-818, 1999.

[174] Sharma R.L., Pandey S.P., “Influence of mineral additives on the hydration characteristics of ordinary Portland cement”, Cement and Concrete Research, Vol. 29, pp. 1525-1529, 1999.

[175] Ye G., Lui X., Schutter D.G., Poppe A.M., Taerwe L., “Influence of limestone powder used as filler in SCC on hydration and microstructure of cement pastes” Cement and Concrete Composites, Vol. 29, pp. 94-102, 2007.

[176] Lawrence P., Cyr M., Ringot E., “Mineral admixtures in mortars effect of type, amount and fineness of fine constituents on compressive strength”, Cement and Concrete Research, Vol. 35, pp. 1092-1105, 2005.

[177] Nomat A., “Interactions between chemical evolution (hydration) and physical evolution (setting) in the case of tricalcium silicate”, Materials and Structures, Vol. 27, pp. 187-195, 1994.

[178] Sari M., Prat E., Labastire J.K., “High strength self-compacting concrete - Original solutions associating organic and inorganic admixtures”, Cement and Concrete Research, Vol. 29(6), pp. 813-818, 1999.

[179] Soroka I., Setter N., “The effect of fillers on strength of cement mortars”, Cement and Concrete Research, Vol. 7, pp. 449-456, 1977.

[180] Bonavetti V.L., Rahhal V.F., Irassar E.F., “Studies on the carboaluminate formation in limestone filler-blended cements”, Cement and Concrete Research, Vol. 31, pp. 853-859, 2001.



[181] Ingram K.D., Daugherty K.E., “A review of limestone addition to Portland cement concrete”, Cement and Concrete Composites, Vol. 13, pp. 165-170, 1991.

[182] Péra J., Husson S., Guilhot B., “Influence of ground limestone on cement hydration”, Cement and Concrete Research, Vol. 21, 99-105, 1999.

[183] Soroka I., Stern N., “Calcareous fillers and the compressive strength of Portland cement”, Cement and Concrete Research, Vol. 6, pp. 367-376, 1976.

[184] Taylor H.F.W., “Cement Chemistry”, 2nd Edition, Thomas Telford Publications, London, 1997.

[185] Tsivilis S., Chaniotakis E., Badogiannis E., Pahoulas G., Ilias A., “A study on the parameters affecting the properties of Portland limestone cements”, Cement and Concrete Composites, Vol. 21, pp. 107-116, 1999.

[186] Vuk T., Tinta V., Gabrovek R., Kaui V., “The effects of limestone addition, clinker type and fineness on properties of Portland cement”, Cement and Concrete Research, Vol. 31, pp. 135-139, 2001.

[187] Hedlund H., “Stresses in high performance concrete due to temperature and moisture variations at early ages”, Licentiate Thesis, Luleå University of Technology, 1996.

[188] Ozawa M., Morimoto H., “Estimation method for thermal expansion coefficient of concretes at early ages”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp. 331-339, Lyngby, 2006.

[189] Sellevold E.J., Bjøntegaard Ø., “Thermal expansion coefficient of cemen paste and concrete - Effect of moisture content”, Proceedings of the Conference on Advances in Cement and Concrete, pp. 27-37, Colorado, 2003.

[190] Schutter D.G., “Thermal properties”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 121-125, 2002.

[191] Johansen R., Dahl P.A., “Control of plastic shrinkage of cement”, Proceedings of the Conference on Our World in Concrete and Structures, Singapore, 1993.

[192] Löfgren I., Esping O., “Early age cracking of self-compacting concrete”, Proceedings of the International RILEM Conference on Volume Changes of Hardening Concrete: Testing and Mitigation, pp.. 251-260, Lyngby, 2006.

Standards, recommendations and other description of test methods: (Note, standards are not referred to by a given index number, but merely by their ID.) ACI 116R-90, “Cement and concrete terminology”, American Concrete Institute, ACI Manual of

Concrete in Practice, Part 1, 2000.

ACI 209-92, “Prediction of creep, shrinkage, and temperature effects in concrete structures”, Farmington Hills: American Concrete Institute , 1997.

ACI 305, “Hot weather concreting”, ACI Manual of Construction Practice, Part 2, Farmington Hills: American Concrete Institute , 2002

ASTM 04.02, “Construction: Concrete and Aggregates”, Annual Book of ASTM Standards, Vol. 04.02, 2000.

ASTM C33, “Standard specification for concrete aggregates”, Annual Book of ASTM Standards, Vol. 04.02, 2000.

ASTM C157, Standard test method for length change of hardened hydraulic-cement mortar and concrete”, Annual Book of ASTM Standards, Volume 04.02, 2002.



ASTM C230-90, “Standard Specification for Flow Table for Use in Tests of Hydraulic Cement”, Annual Book of ASTM Standards, Volume 04.02, 2002.

ASTM C204-00, “Standard test method for fineness of hydraulic cement by air permeability apparatus”, 2002.

ASTM C403, “Test method for time of setting of concrete mixtures by penetration resistance, Annual Book of ASTM Standards, Vol. 04.02, 2002.

ASTM C426-06 Standard test method for linear drying shrinkage of concrete masonry units, Annual Book of ASTM Standards, Vol. 04.05, 2006.

ASTM C1608-06, “Standard Test Method for Chemical Shrinkage of Hydraulic Cement Paste”, Annual Book of ASTM Standards, Vol. 04.01, 2006.

BS 4359, “Determination of specific surface area from particle size distribution”, British standard, 1973.

BS 5168:1975, “Glossary of rheological terms”, British Standard Institution, 1975.

DIN 53019, “Viscometry; Determination of viscosities and flow curves using standard design rotary viscometers with a standard geometry measuring system”, 1976.

EN 196-3, “Methods of testing cement - Part 3: Determination of setting time and soundness”, 2005.

EN 933-1, “Tests for geometrical properties of aggregates - Part 1: Determination of particle size distribution - Sieving method”, 2004.

EN 933-2, “Tests for geometrical properties of aggregates - Part 2: Determination of particle size distribution - Test sieves, nominal size of apertures”, 1996.

EN 1015-2, “Bulk sampling of mortars and preparation of test mortars - Part 2: Bulk sampling of mortars and preparation of test mortars”, 2006.

EN 1015-3, “Methods of test for mortar for masonry - Part 3: Determination of consistence of fresh mortar (by flow table)”, 1999.

EN 1097-6, “Tests for mechanical and physical properties of aggregates - Part 6: Determination of particle density and water absorption”, 2000.

EN 12350-2, “Testing fresh concrete - Part 2: Slump test”, 2000.

ISO 2884, “Paints and varnishes: Determination of viscosity using rotary viscometers - Cone-and-plate viscometer operated at a high rate of shear”, 1999.

ISO 3219, “Plastics; Polymers/resins in the liquid state or as emulsions or dispersions - Determination of viscosity using a rotational viscometer with defined shear rate”, 1993.

ISO 9277, “Determination of specific surface area of solids by gas adsorption using BET method”, 1996.

NT BUILD 433, “Concrete: Cracking tendency – Exposure to drying the first 24 hours”, Nordtest, 1995.

NT BUILD 486, “Aggregates: Size distribution”, Nordtest, 1998.

RILEM report 25, “Early Age Cracking in Cementitious Systems, TC181-EAS, RILEM Technical Committee, 2002.

SS 132126, “Concrete testing - Aggregate - Modulus of fineness”, 1986.

SS 134222, “Methods of testing cement - Determination of fineness”, 1990.

A p p e n d i x A - S p e c i f i c s u r f a c e a r e a b y B E T ( H 2 O )


APPENDIX A: BET(H2O) DETERMINATION OF SPECIFIC SURFACE AREA OF FILLER AND FINE PART OF AGGREGATE BY BET METHOD WITH WATER ADSORPTION.



BET(H2O): DETERMINATION OF SPECIFIC SURFACE AREA OF FILLER AND FINE PART OF AGGREGATE BY BET METHOD WITH WATER ADSORPTION Key words: surface area, BET, adsorption, powder, filler, aggregate.

1 SCOPE This description provides guidelines for determination of specific surface area by the BET(H2O) method. The principle is based on the method of Brunauer, Emmet and Teller [1], where the area is given by the number of physically adsorbed gas molecules formed in a monolayer. For a general description of the BET method see ISO 9277, and for the physics and applications of gas adsorption see Gregg and Sing [2]. The measurements of BET area are normally conducted volumetrically with nitrogen gas (N2), but in this case it is gravimetrically with water vapour (H2O) as adsorptive gas. By using the relationship between the relative humidity and the adsorbed weight per unit specimen mass the surface area is calculated (in m² per kg). The BET(H2O)-area is generally much larger than the area determined by air permeability (e.g. Blaine), and slightly larger than the BET(N2)-area due to the differences in polarity and molecular size.

For concrete, particle concentration and size distribution strongly influence the workability since its matrix consists of a suspension of smaller particles filling the hollow spaces between the coarser ones. To decrease the internal friction and generate mobility, the particle surfaces are to be covered with water and the voids in the system to be filled. Other important properties affecting the workability are the particle shape, porosity and surface texture. Since surface area/volume ratio increases exponentially with particle irregularity (shape, surface texture and porosity) and decreased size, quantification of geometrical properties of fillers and the fine part of the aggregates is essential for the ability to control physical properties, such as workability.

2 FIELD OF APPLICATION The BET(H2O) test method is only to be applied for laboratory use on fillers and fine part of aggregates. Cement and other powders reacting with water are not to be analyzed. Maximum particle size is 0.5 mm. If the material incorporates particles up to 8 mm, these are to be removed and compensated for. For materials with particles larger than 8 mm, the BET(H2O) method is not recommended.

3 REFERENCES [1] Brunauer S., Emmet H. P., Teller., “Adsorption of gases in multimolecular layers”, American Chemical Society, 1938.

[2] Gregg S. J., Sing K. S. W., “Adsorption, surface area and porosity”, New York, 1967.

Standard: ISO 9277, “Determination of the specific surface area of solids by gas adsorption using the BET method”, 2002.

Standard: EN 933-1, “Tests for geometrical properties of aggregates - Determination of particle size distribution - Sieving method”, 2004.

Standard: EN 1097-5, “Tests for mechanical and physical properties of aggregates - Determination of the water content by drying in a ventilated oven”, 2004.

Standard: EN 932-1, “Tests for general properties of aggregates - Methods for sampling”, 1997.

Standard: EN 932-2, “Tests for general properties of aggregates - Methods for reducing laboratory samples”, 2004.

4 NOTATIONS 4.1 Definitions Adsorbate; A substance that is adsorbed at the interface of a substrate material.

Adsorbent; The substrate material onto which a substance is adsorbed.

Adsorption; Enrichment of adsorptive at the external and accessible internal surfaces of a solid.

Adsorptive; A measuring gas to be adsorbed.

Adsorption isotherm; A graphical representation of the relationship between the bulk activity of adsorbate and the amount of adsorbate at constant temperature.

BET; A model to determine the surface area from the physical adsorption of a gas on a solid surface, named after Brunauer, Emmett & Teller.

Specific surface area; The ratio of total (external and corresponding internal) surface area per unit weight or unit volume.



4.1 Symbols

dmax maximum grain size, in millimetres [mm]

k factor compensating for the particles not passing the 0.5 mm sieve (k ≈ 0.5)

m mass of specimen, in grams [g]

mc mass of container, in grams [g]

mdry mass of oven dry specimen, in grams [g]

RH relative humidity, in the range of 0-1 [-]

SBETH2O specific surface area by BET(H2O), in square meters per kilogram [m2/kg]

S’BETH2O approximated BET(H2O)-area for the whole mass fraction (up to 8 mm)

u moisture content, in kilograms of water per kilogram dry of material [kg/kg]

x05 mass fraction passing 0.5 mm sieve [-]

5 METHOD OF TESTING 5.1 Principle The specific surface area is determined by deducing the area from the volume of gas required to form a monolayer on the surface of the sample. By using BET theory, the area can be calculated from knowledge of the size and the number of physically adsorbed gas molecules.

With the BET(H2O), the specific surface area of fillers and fine part of aggregate is determined gravimetrically by a multipoint method, using water vapour (moisture) as adsorbate at different relative humidity (RH). The test is to be performed at 20ºC with climate boxes with 11.3% and 33.1% RH, achieved with LiCl and MgCl2 saturated salt solutions (see Figure 1a). The oven-dried double samples, approximately 2x100 g, are conditioned in 11.3% RH followed by 33.1% for minimum 4 days in each climate. The weight of the dry and conditioned samples is measured by a precision balance with 0.001 g readability. By plotting the BET(H2O) adsorption isotherm as a linear function of the RH, and evaluating the intercept and slope, the specific surface area (SBETH2O) can be calculated (see Figure 1b).

5.2 Equipment Required equipment for a typical test:

• Thermally controlled room or chamber with an air temperature of 20±2ºC.

• A desiccator or airtight cabinet with desiccant, to be used for the cooling stage. Silica gel with colour indicator is a suitable desiccant.

NOTE 1: If a desiccator or cabinet with desiccant is not available, thermal insulation can be used to protect the balance from the sample heat. A cork mat can be suitable.

• Precision balance of suitable capacity (100 g sample + container), with 0.001 g readability and 0.005% accuracy of the sample mass.

• Ventilated drying oven, thermostatically controlled to a temperature of 110±5°C.

• 2 sets of airtight boxes, equipped with a tight lid, fan, supporting grid/net and shallow tray that occupies most of the bottom for the saturated salt solutions (see Figure 1a).

• Salts, lithium chloride (LiCl) for 11.3% RH and magnesium chloride (MgCl2) for 33.1% RH. The amount of salt needed is dependent on the size of the box, where the shallow tray is to be filled to ~3 mm depth.

• 1 set (2 off) of containers for each material, large enough to store the ~100 g double sample. The container must be open, heat-resistant and made of non-adsorbent material. E.g. a 300 ml food tin of thin alumina without lid.

5.3 Preparation of equipment and test specimen 5.3.1 Saturated salt solutions The dry salt is to be spread to about 3 mm depth in the shallow tray. More than one tray can be used in order to occupy most of the bottom of the climate box. Water is added to moisten the salt, but not more than to make it look damp. If all salt is dissolved (i.e. no visual salt) too much water has been used, in which case add more dry salt.

5.3.2 Sample The recommended specimen size is approximately 100 g. For gravel/aggregate the sample is to be collected in accordance with EN 932-1, and reduced according to EN 932-2. All grains not passing the 0.5 mm sieve (see EN 933-1) are to be removed, and the masses of the passed and retained materials are to be weighed and recorded. The sample is to be made in pairs, to ensure precision. The maximum number of specimens to be tested simultaneously depends on the size of the climate boxes and the sample containers.



5.4 Procedure 5.4.1 Running the test a) Weigh and record the mass of the empty containers (mc). The containers must be clean and dry, and numbered with a marker.

NOTE 2: Two containers for each material are needed as the sample is to be made in pairs, to ensure precision.

b) If the materials contain grains larger than 0.5 mm these are to be removed by sieving. The masses of the passed and retained materials on the 0.5 mm sieve (see EN 933-1) are to be weighed and recorded.

c) The ~100 g specimens (two per material) are to be placed in each container.

d) The samples and the desiccant are to be placed in the ventilated drying oven with a temperature of 110±5°C. The time to dry depends on the sample’s size, fineness, porosity and moisture content. A minimum time of 4 hours is recommended.

e) The samples and the desiccant are to be placed in the desiccator or airtight cabinet. The time required for the cooling stage depends on the size and number of samples, the tightness of the cabinet, balance heat sensitivity, etc. Time less than 2 hours is recommended.

NOTE 3: It is also possible to weigh a sample immediately after removing it from the oven if the balance is sufficiently protected from the sample heat (with e.g. a cork mat).

f) Weigh the samples and determine the dry mass of the specimen (mdry) by subtracting the mass of the tray (mc) from the weight. Directly place the sample in the climate box with 11.3% RH (LiCl). The time for the test portion to be conditioned is a minimum of 3 days, preferably 4-5. The box is to be placed in a thermally controlled room or chamber with an air temperature of 20±2ºC.

NOTE 4: The time when the sample is exposed is crucial, and is to be kept as short as possible.

NOTE 5: In some cases where the particles are very fine (e.g. silica fume) or very porous, a longer time is needed for conditioning. The time required to achieve constant mass for a specific type of material, or particle size, can be determined by successive weighing at least 12 h apart and seeing that the mass does not differ by more than 0.005%.

g) Weigh the sample and determine the mass (m1) by subtracting the mass of the tray (mc) from the measured weight. Place the sample in the climate box with 33.1% RH (MgCl2). The conditioning procedure is the same as for the 11.3% RH conditioning.

h) Weigh the sample and determine the mass (m2) by subtracting the mass of the tray (mc) from the measured weight.

NOTE 6: Keep the lids of the boxes as closed as possible between the weighing of samples. The saturated salt solution can be reused a large number of times. Just ensure that the salt is sufficiently moistened, but not fully dissolved. Adjust by adding extra salt or water if needed.

NOTE 7: It is also possible to determine the dry mass of the test portion (see d to f) as a final step instead. For higher precision and lower error, the dry mass can be determined both initially and finally (and the mean value reported). The dry mass should not differ by more than 0.01%.

5.4.2 Finishing the test The measured masses (mdry, m1, and m2) for each test portion are to be transferred to a computer for calculation and evaluation.

5.5 Expression of results The BET(H2O) surface area per unit specimen mass is calculated as:

ba1054.3S

6

O2BETH +⋅

= [m2/kg] (Eq. 1)

where a is the interception point and b the slope of the curve from the linear relationship between the BET(H2O) adsorption isotherm and relative humidity (see Figure 1b). The adsorption isotherm (y) is calculated as:

)RH1(uRHy−⋅

= [kg/kg] (Eq. 2)

where RH is the relative humidity (0.113 and 0.331), and u is the moisture content of the conditioned specimen. The moisture content is calculated as:

dry

dry

mmm

u)( −

= [kg/kg] (Eq. 3)

where mdry is the mass of the oven-dry specimen and m the mass of the conditioned specimen (m=m1 for 11.3% and m=m2 for 33.1%).



For gravel/aggregate, the removed particles not passing the 0.5 mm sieve are to be compensated for. The BET(H2O) surface area per unit specimen mass calculated for the whole size fraction is approximated as:

( ))x1(kxba1054.3S 0505

6

O2BETH' −⋅+⋅

+⋅

= (Eq. 4)

where k is a factor compensating for the removed particles not passing the 0.5 mm sieve. For gravels/aggregates with a maximum grain size up to 8 mm, k ≈ 0.5 [-]. The mass fraction passing the 0.5 mm sieve is calculated as:

tot

0505 m

mx = [-] (Eq. 5)

where m05 is the mass passing the 0.5 mm sieve and mtot is the total mass.

Note 8: If there is only a small amount passing the 0.5 mm sieve, the calculation of S´BETH2O is not valid. For x05<0.1 the method is not recommended.

It is not necessary to graphically plot the relationship between the isotherm (y) and RH in order to evaluate the variables a (interception) and b (slope). By using Eq. 2, Eq. 3 and the linear relationship:

aRHby +⋅= [-] (Eq. 6)

a and b can be calculated as:

⎟⎟⎠

⎞⎜⎜⎝

⎛

−−

−⋅=

)mm(2565.0

)mm(1934.0ma

dry2dry1dry (Eq. 7)

⎟⎟⎠

⎞⎜⎜⎝

⎛

−−

−⋅=

)mm(5844.0

)mm(2696.2mb

dry1dry2dry (Eq. 8)

The specific surface area (SBETH2O) is to be presented as a mean value based on the double specimen results, together with a calculated deviation.

Note 9: Adsorption is not the same for moisture as for nitrogen gas. The water molecule is smaller than the nitrogen molecule (10.6 Å2 for water, and 16.2 Å2 for nitrogen), and has larger affinity forces (of attraction) due to its two-polar covalent bonds. However, using water (moisture) is probably closer to real conditions in concrete production than using nitrogen gas.

6 TEST REPORT The report is to include necessary information from among the following:

a) Document ID (name, no., test method, etc.).

b) Date and time.

c) Performer ID (name and address).

d) Material ID, supplier and characteristics.

e) Date, location and method of sampling.*

f) Mass fraction passing 0.5 mm sieve (x05), expressed in percentage [%].*

g) Maximum grain size (dmax), expressed in millimetres [mm].*

h) Mass of the analyzed specimen, expressed in grams [g].

i) Specific surface area (SBETH2O), expressed in square metres per kilogram [m2/kg].

j) Specific surface area for the whole size fraction (S´BETH2O), in square metres per kilogram [m2/kg].*

k) Deviation in surface area, based on the double specimen results, expressed in percentage [%].

l) Time of conditioning in each climate, expressed in days.

m) Any deviation from the test method.

n) Relevant visual observations and personal judgments and interpretations.

o) Date and signature.

Note 10: * only for gravel/aggregate, where also both SBETH2O and S´BETH2O are to be reported.

7 EXAMPLES To give an idea of the magnitude of the specific surface area by BET(H2O) for different materials, the following approximate values are given:

Material SBETH2O [m2/kg]

S´BETH2O [m2/kg]

Limestone filler Fly ash Silica fume Natural gravel (0-8 mm) Sea-sand (0-4 mm)

~2 000 ~3 000

~20 000 ~4 000 ~1 500

~2 500 ~1000



Fan

Saturated salt solutionBox with tight lid

Net

Samples (2x)

y [k

g/kg

]

a

RH [-]

b

0.113 0.331

(a) (b)

Figure 1. (a) Schematic illustration of climate box, and (b) the BET-plot with adsorption isotherm as a linear function of RH.

A p p e n d i x B – C o n c r e t e d i g i t a l d i l a t o m e t e r


APPENDIX B: CONCRETE DIGITAL DILATOMETER LINEAR MEASUREMENT OF AUTOGENOUS DEFORMATION IN FRESH AND HARDENING CONCRETE.



CONCRETE DIGITAL DILATOMETER (CDD): LINEAR MEASUREMENT OF AUTOGENOUS DEFORMATION IN FRESH AND HARDENING CONCRETE Key words: Concrete, shrinkage, autogenous

1 SCOPE This test method covers determination of concrete autogenous deformation. It has the ability to start measurements before setting, when the concrete is fresh. The method is a modification of the CT1 digital dilatometer for pastes and mortars.

As the use of high-performance and self-compacting concrete has increased, problems with early-age shrinkage and cracking have become significant. Conditions such as reduced water-cement ratio, reduced maximum aggregate size, increased amount of fines, and increased binder content all contribute to these problems. Compared with drying shrinkage, which generally occurs from the outer surface of the concrete inward, autogenous shrinkage develops uniformly through the concrete member, but can be more likely to cause cracking, because it develops more rapidly and occurs when the cement paste is young and has poorly developed mechanical properties.

2 FIELD OF APPLICATION With the concrete digital dilatometer test, the fresh and hardening concrete linear autogenous deformation is evaluated.

The test is only to be applied for laboratory use.

Maximum coarse aggregate size is 16 mm.

A concrete digital dilatometer (CDD) test is preferably combined with:

• Cracking tendency ring-test (NT BUILD 433) for concrete.

• Digital dilatometer (CT1) test for cement paste.

3 REFERENCES CT1 digital dilatometer: Jensen O. M., Hansen F., “A dilatometer for measuring autogenous deformation in hardening Portland cement paste”, Materials and Structures, Vol. 28(181), pp. 406-409, 1995.

Sampling procedure: EN 12350-1, ASTM C172 or NT BUILD 191.

Cracking tendency test: NT BUILD 433.

4 NOTATIONS 4.1 Definitions Linear deformation; the relative change in length due to shrinkage or expansion, referred to the specimen’s original length.

Autogenous deformation; the unrestrained, time-dependent bulk deformation of fresh and hardening sealed concrete at a constant temperature.

Shrinkage; when the deformation is a contraction, it may be referred to as shrinkage, e.g. autogenous shrinkage.

Chemical shrinkage; under sealed conditions, the cement paste hydration products occupy less space than the original reactants. Most factors causing the autogenous shrinkage are chemical.

4.1 Symbols

l0 length of specimen at time of casting (when measurements start), in millimetres [mm].

∆l change in length of specimen at time t, in micrometres [µm].

t time after mixing (when measurements start), in hours and minutes [hh:mm].

ε relative change in length, defined positive for shrinkage (negative for swelling), in microstrain [µm/m].

T concrete cross section temperature, in degrese Celsius [ºC]

5 METHOD OF TESTING 5.1 Principle The concrete digital dilatometer is a test method for evaluation of linear autogenous deformation of fresh and hardening concrete at sealed conditions and constant ambient temperature.

The concrete is cast in a vapour proof steel coil reinforced flexible polyurethane (PU) tube mould with inner diameter 82 mm and specimen length ~400 mm.



The mould is placed in a mechanically stable measuring rig, where the unrestrained, time-dependent bulk deformation is recorded by a digital gauge (1/1000 mm) at thermally controlled ambient conditions (20ºC) for a period of 24 hours or more.

The method is illustrated in Figure 1.

5.2 Apparatus Required equipment and apparatus for a typical test:

• Thermally controlled room or chamber with an air temperature of 20±1ºC.

• 2 or 3 complete CDD setups (mould consisting of PE-tube, end-caps, fixture and hose-clamps and measuring rig), as shown in Figure 1.

• 1 set (2 off) of recording temperature sensors.

• Table vibrator.

• Ruler or gauge graded in 1/1 mm.

• Stopwatch, measuring 1/1 sec (if manual reading of change in length).

5.3 Preparation of test specimen 5.3.1 Mould The first end-cap (moving end) is to be fitted to the PE-tube with a house-clamp. The mould is to be placed in a vertical support rack (e.g. a piece of duct) to ensure stability during casting.

5.3.2 Mix Concrete constituents’ volume is to be calculated to 1/1000 litre. Dry aggregate is preferable.

5.3.3 Mixing A suitable concrete mixing method, mixer and volume are to be selected and documented. No standards are available.

5.3.4 Sample The sample size is 2.2 litres and is to be collected in accordance with EN 12350-1, ASTM C172 or NT BUILD 191.


5.3.5 Casting The specimen must completely fill the mould. If needed, compaction by vibration.

If temperature sensors - place one in the centre and one at the surface of the mould cross-section.

The second end-caps (fixed end) are to be fitted to the PE-tube with a house-clamp.

5.4 Procedure 5.4.1 Starting the test a) The specimen is to be placed in the measuring

rig.

b) The mould is to be adjusted and fixed to the rig so that the digital gauge contact point is in starting position.

c) Reference length l0 (mould length excluding the end-caps, i.e. specimen length) is to be measured to 1/10 mm.

d) The digital gauge is to be re-zeroed and its recording phase (time t and length change ∆l) started.

e) If temperature sensors: start temperature T recording.

The test must be performed in a thermally controlled room or chamber with an air temperature of 20±1ºC.

The starting procedure (a-d) is to be performed within 20-30 minutes from adding water to the mix. This time is to be noted. The initial measurements are sensitive and it is recommended to let the sample set (or rest) in the rig for 10 minutes before starting the measurements.

5.4.2 During the test If manual reading, continuously record the time t, length change ∆l and concrete temperature T. Enough sampling frequency is to be selected for smooth curve of shrinkage development.

Visual observation of specimen straightness; possible curviness is to be measured and noted.

5.4.3 Finishing the test Recording phase (t, ∆l and T) stops. Values are to be transferred to computer for calculation and evaluation.



5.5 Expression of results The shrinkage, ε, for each test specimen is to be calculated as:

6

0

10⋅∆

=l

lε [in parts per million, e.g. µm per m]

The results are to be calculated as a mean value of two or three tests, and are to be given in microstrain with two significant digits.

The shrinkage must be corrected for the recorded concrete cross section temperature (T). Adequate thermal expansion coefficient is to be applied.

The development of shrinkage (ε) with time (t) is to be presented graphically. 6 TEST REPORT The report should include necessary information from among the following:


b) Date and time.


d) Test object ID.

e) Purpose of test.

f) Concrete ID (producer, recipe, etc.).

g) If relevant, concrete and/or air temperature.

h) Method of sampling.

i) Time from water addition to sampling start.

j) Reference length (l0), i.e. specimen length.

k) Storage conditions.

l) Test results (t, ∆l and T).

m) Relevant visual observations (e.g. sample curviness) and personal judgments and interpretations.

n) Any deviation from the test method.

o) Graphic presentation of shrinkage (ε) development with time (t).

p) Inaccuracy and/or uncertainty of test results.

q) Date and signature.

0 001

Fixed end-cap Moving end-capFlexible PU-tube Hose-clap

0 001

Mould length 460 mm

Specimen length 400 mmDidital gauge1/1000 mm

Adjustable fixture

Mould diameter 82 mm

Measuring rig

Figure 1. Principe of the Concrete Digital Dilatometer (CDD) for autogenous measurement.

A p p e n d i x C - C o n c r e t e c r a c k i n g r i n g t e s t


APPENDIX C: CONCRETE CRACKING RING TEST CRACKING TENDENCY MEASUREMENT DUE TO DRYING DEFORMATION THE FIRST 24 HOURS



CONCRETE CRACKING RING TEST: CRACKING TENDENCY MEASUREMENT DUE TO DRYING DEFORMATION THE FIRST 24 HOURS Key words: Concrete, cracking, shrinkage, drying

1 SCOPE This test method covers determination of concrete cracking tendency at early ages. The test is performed on 3 restrained ring-shaped specimens, exposed to an air stream of defined velocity, temperature and relative humidity, for the first 24 hours after casting.

The principle of the test is that the concrete sample is cast around a restraining inner steel ring, causing a development of tangential stresses, which if sufficiently high may lead to cracking. The evaluation is based on characterization of the cracks in terms of average total area in the three samples.

The method is a modification of the Nordtest method “NT BUILD433”.

Problems with early-age shrinkage and cracking have become significant. Conditions such as reduced maximum aggregate size, increased amount of fines, presence of retarding admixtures, increased binder content and deficient covering and curing all contribute to these problems.

Most probably the plastic shrinkage also consists of autogenous shrinkage. Compared with autogenous shrinkage, which generally develops uniformly through the concrete member, shrinkage due to evaporating occurs from the outer surface of the concrete inward, causing cracks that develop rapidly, and occurs when the cement paste is young and has poorly developed mechanical properties.

2 FIELD OF APPLICATION With the “Concrete cracking ring test”, the plastic and hardening concrete cracking tendency can be used for an evaluation of different types of concrete exposed to early drying.

The test is only to be applied for laboratory use, where the method information is relative and cannot predict the extent of cracking which might occur under prevailing conditions.


The concrete cracking ring test is preferably combined with:

• Volume or linear measurement of autogenous deformation, e.g. Concrete Digital Dilatometer (CDD) test for concrete.

3 REFERENCES

Sampling procedure: EN 12350-1, ASTM C172 or NT BUILD 191.

Cracking tendency test: NT BUILD 433

4 NOTATIONS 4.1 Definitions Shrinkage: when the deformation is a contraction, it may be referred to as shrinkage, e.g. autogenous or drying shrinkage.

Autogenous shrinkage: the unrestrained, time-dependent bulk deformation of fresh and hardening sealed concrete at a constant temperature.

Chemical shrinkage: under sealed conditions, the cement paste hydration products occupy less space than the original reactants. Most of the factors causing the autogenous shrinkage are chemical.

Evaporation shrinkage: when water evaporates from the fresh and hardening concrete, tensile stress builds up in the capillaries, causing the concrete to contract. In early stages; drying shrinkage can be defined as plastic shrinkage.

4.1 Symbols

n number of cracks of each specimen.

l length of each crack, in millimetres [mm].

w each crack average width, in millimetres [mm].

A average total crack area calculated from two ore more samples, in sq. millimetres [mm2]

t time after mixing, in hours [h].

∆m sample weight loss due to drying, in kilograms [kg].

E sample evaporation, in kilograms per sq. metre [kg/m2].

v air velocity, in metres per second [m/s].

RH air relative humidity, in percentage [%].

Ta air temperature, degrees Celsius [ºC].

Tc concrete temperature, degrees Celsius [ºC].



5 METHOD OF TESTING 5.1 Principle When water evaporates from the fresh concrete, the concrete tends to contract, and as contraction is restrained, tangential tension develops across a section of the ring specimen. The extent of cracking depends both on the magnitude of the tensile forces and on the strain capacity of the concrete.

The concrete is cast between two concentric steel rings with diameters 300 and 600 mm, and with a depth of 80 mm; see Figure 1. The steel rings have ribs attached to provide crack initiation and are fixed to a stiff base plate with a smooth surface. After casting, the ring specimens are positioned under air funnels with a 10 mm opening between the concrete surface and the funnel along the circumference of the outer ring. The funnel is shaped to provide equal wind velocity across the concrete surface of 4.5 m/s. The test is to be performed at stable and constant environmental conditions, where one of the samples is to be placed on a balance.

The cracking tendency is evaluated from crack length and average width measurements on the concrete top surface.

5.2 Apparatus Required equipment and apparatus for a typical test:

• Thermal and humidity-controlled room or chamber with a constant air temperature of T=20±1ºC and relative humidity of RH=35±5%. The conditions’ magnitudes are not absolute, but these are preferable.

• 3 complete mould setups for ring specimen, as shown in Figure 1.

• 3 sets of air funnels, including fan and ducting, as shown in Figure 1.

• Balance (load cell) with manual reading or automatic recording of weight changes.

• Devices and instrumentation for manual reading or automatic recording of: - air velocity - temperature and relative humidity of air - concrete temperature.

• Ruler or measuring wheel graded in 1/1 mm.

• Water level.

• Stopwatch, measuring 1/1 sec.

5.3 Preparation of test specimen 5.3.1 Mould The inner and outer steel rings are fixed on a stiff base steel plate with a smooth surface. The steel rings are to be covered with a thin layer of form oil and the base plate is not to be oiled.

5.3.2 Mix Concrete constituents by dry weight are to be recorded. Max. coarse aggregate size is 16 mm.

5.3.3 Mixing A suitable concrete mixing method, mixer and volume are to be selected and documented. No standards are available.

5.3.4 Sample The sample size is 17 litres (x3) and is to be collected in accordance with EN 12350-1, ASTM C172 or NT BUILD 191.

5.3.5 Casting The concrete is to be sampled from the mixer immediately after the end of mixing period. The casting of 3 ring specimens must be accomplished within 30-45 minutes after water addition.

The specimen must completely fill the mould (to the top of the inner and outer steel rings). If needed, compact by vibration. Top surfaces are to be smoothed in a corresponding manner.

5.4 Procedure 5.4.1 Starting the test a) One of the specimens is to be placed on the

balance.

b) Each specimen’s horizontal position is to be ensured (if unsure, use a water level).

c) One concrete temperature sensor is to be placed in the centre of the mould cross-section.

d) The air funnels are to be placed in such a way that the opening between the concrete surface and the funnel edge is uniform (10 mm) along the whole circumference of the outer steel ring.



e) The fans are to be started 55 minutes after water addition. Velocity over the concrete surface is to be ensured (4.5 m/s)

f) The recording phase (time t, concrete temperature Tc, specimen weight loss ∆m, air temperature and humidity Ta and RH) is to be started 60 minutes after water addition.

The test is to be performed in a thermal, humidity-stable and controlled room or chamber with an air temperature of 20±1ºC. and relative humidity of 35±5%.

The starting procedure (a-d) is to be performed within 60 minutes from water addition, if not this time is to be noted.

5.4.2 During the test If manually reading, continuously record the time t, concrete temperature Tc, specimen weight loss ∆m, air temperature and humidity Ta and RH. Enough sampling frequency is to be selected for a smooth curve of temperature and weight loss development (less than 1/2 hours interval).

Visual observation of specimens throughout the 24 hours can be made in order to describe the crack development. The type, orientation and time of occurrence of cracks can then be noted.

5.4.3 Finishing the test The recording phase (t, Tc, ∆m, Ta and RH) stops. Values are to be transferred to computer for calculation and evaluation.

5.4.3 Crack measurements The ring specimens should normally be examined after 24 hours’ exposure, and surface cracks with an approximate radial orientation should be identified and marked in an adequate way. The average width (w) and length (l) of each crack are to be measured and recorded. The width measurements shall be performed by the use of the magnifying glass and readings by interpolation to the nearest 0.02 mm. It is recommended that a lower crack limit is 0.05 mm. The main crack pattern of each ring can be recorded by photo or sketched by drawing.

The standard procedure also includes recording of weight loss and temperature development. These parameters give useful information about the evaporation of water, and serve as a control for identical tests as well.

5.5 Expression of results The total crack area for each ring is to be calculated as the accumulated sum over each average crack width multiplied by its length. The average total crack area (A) is expressed by the average area for 3 rings, rounded to the nearest 0.1 mm, as:

3

))wl((A

3

1j

n

1iii∑ ∑

= =

⋅= [mm2]

The water evaporation (E) is calculated by the quotient of weight loss (∆m) and ring surface area:

212.0mE ∆

= [kg/m2]

The development of concrete temperature (Tc) and evaporation (E) with time (t) is to be presented graphically.

6 TEST REPORT The report is to include necessary information from among the following:


b) Date and time.


d) Test object ID.

e) Purpose of test.

f) Concrete ID (producer, recipe, etc).

g) Identification of the test equipment and instruments used.

h) Method of sampling.

i) Time from water addition to sampling start/stop and crack measuring.

j) Air velocity (v), temperature (ta) and humidity (RH) during the test.

k) Test results: number of cracks (n), total crack area (A) and graphical presentation of concrete temperature and evaporation (t, Tc and E).

l) Relevant visual observations and personal judgments and interpretations.

m) Any deviation from the test method.

n) Inaccuracy and/or uncertainty of test results.

o) Date and signature.



Duct ø100 mm

Min

100

0 m

m

Adjustable damper

Fan

Base plate with smooth steel surface

Transparent air funnel

Distances 10 mm

Steel rings ø600/300 mm

Air flow 310 m3/h


Test arrangement

51

Transparent, e.g. Plexiglass, t=4 mm

ø100

ø612

11°

Distances h=10 mm

80

ø600 (inner)

120°

30°

ø300 (outer)

6

6

25 Steel ribs 80 x 25 x 3 mm, 12 pcs on outer & 3 pcs on inner ring

Transparent air funnel Ring mould setup

Figure 1. Principe of the concrete cracking ring test for cracking tendency measurement.

A p p e n d i x D - G l o s s a r y


APPENDIX D: GLOSSARY

The following terminology, related to the subject matter in this work, has been adopted. Thus some modification from its origins is made (see references at end of this appendix).

This glossary is intended as a reader’s aid. The explanations given here are not rigorous definitions and should not be regarded as such. Formal definitions may be found in documents such as:

o ASTM Standard C125 “Standard terminology relating to concrete and concrete aggregates”

o British Standard BS5168 “Glossary of rheological terms”

o ACI Manual 116R “Cement and concrete terminology”

o European standard EN 197-1 “Cement”, EN 206-1 “Concrete”, EN 934-2 “Admixtures for concrete”, EN 12620 “Aggregates for concrete”

Definitions regarding fresh concrete and mortar and its constituents are fairly well defined in literature and standards. However, an established and standardized terminology is somewhat more deficient regarding the subject of early-age deformation, and especially autogenous deformation.

admixture: A material other than water, aggregates, and cement, that is used as an ingredient of concrete and is added to the bath before and during the mixing operation. adsorbate: A substance that is adsorbed at the interface of a substrate material. adsorbent: The substrate material onto which a substance is adsorbed. adsorption hysteresis: The adsorption and desorption curves’ deviation from one another. Generally explained as due to the ink-bottle effect. adsorption isotherm: A graphical representation of the relationship between the bulk activity of adsorbate and the amount of adsorbate at constant temperature. adsorption: Development at the surface of a solid of a higher concentration of a substance than what exists in the bulk of the medium. For aggregate, concrete and cement, adsorption is the formation of a layer of water at the surface of a solid held by either physical and/or chemical forces. aerosol: Droplets or particles dispersed in a gaseous phase. aggregate: Granular material, such as sand, gravel, crushed stone, or crushed concrete, used with a hydraulic cementing medium to produce either concrete or mortar. aggregation: A general term defined as any process by which particles collect to form a cohesive mass or cluster structure. anti-thixotropy: See negative thixotropy. apparent density: The density of loose or compacted particulate matter determined by dividing actual weight by volume occupied; apparent density is always less than true density of a material comprising the particulate matter, because volume occupied includes the space devoted to pores or cavities between particles (see apparent particle volume). apparent particle volume: The total volume of the particle excluding open pores, but including closed pores. apparent viscosity, The quotient of shear stress divided by shear rate when the this quotient is



dependent on rate of shear ( γση &∂∂= /app ).

apparent yield stress: For non-ideal viscoplastic materials (e.g. concrete), where the yield stress is indefinite. An apparent yield stress can be defined by extrapolation from the linear shear rate of the flow curve to the shear stress axis (see Bingham model). autogenous shrinkage: The unrestrained, time-dependent bulk deformation of fresh and hardening sealed homogeneous cement paste, mortar or concrete at a constant temperature (isothermal). BET: A gas adsorption technique for determination of the specific surface area of porous materials, named after Brunauer, Emmett and Teller. The BET surface area is the area on which gas molecules, such as N2 or O2, can adsorb. Bingham model: A mathematical model used for describing viscoplastic flows exhibiting a yield response. The ideal Bingham material is an elastic solid at low shear stress values and a Newtonian fluid above a critical value called the yield stress. The plastic viscosity region exhibits a linear relationship between shear stress and shear rate, with a constant plastic viscosity ( plηγσσ ⋅+= &0 ). Fresh cement paste, mortar or concrete, under certain conditions, and the Bingham model are traditionally used for defining the rheological properties. Blaine fineness: The fineness of granular materials such as filler, cement and pozzolan, expressed as total surface area in square centimetres per gram, determined by the Blaine air-permeability apparatus and procedure. bleeding: The separation of water or paste from a fresh mix, usually caused by the settlement of the solid constituents. Bleeding is sometimes considered a form of segregation. Brownian motion: The random movement of small particles in a dispersed phase due to bombardment by molecules of the surrounding medium (see colloidal). bulk density: The mass of a material per unit volume (including solid particles, voids and pores). Compaction increases bulk density by reducing pore spaces. Bulk density is used interchangeably with unit weight. cement paste: Constituent of concrete, consisting of cement and water. cement: A powdery substance made by burning, at a high temperature, a mixture of clay and limestone – producing lumps called “clinkers” which are ground into a fine powder consisting of hydraulic calcium silicates. chemical shrinkage: Under sealed conditions, the absolute volume reduction associated with the hydration reactions in cementitious materials. The products of hydration occupy less space than the original reactants. Most of the factors causing the autogenous shrinkage are chemical. coarse aggregate: A general term for aggregate of such size that it is substantially retained on a sieve of specified size, commonly 8.0, 4.0 or 4.75 mm. cohesiveness: Ability of mix to resist segregation during transportation and placement. colloidal: State of subdivision implying that the particles have at least in one direction a dimension roughly between 1 nm and 1 ?m. Colloids are significantly affected by Brownian motion when suspended in a liquid. compactness: A shape measure, calculated as the ratio between the diameter of a circle with the same area as measured and the maximum elongation of the measured area. At 1, the object is circular. concrete: A composite material that consists essentially of a binding medium in which are embedded particles or fragments of relatively inert material filler. In Portland cement concrete, the binder is a mixture of Portland cement and water; the filler may be any of a wide variety of natural or artificial aggregates. consistency: The relative mobility or ability of fresh concrete or mortar to flow, usually measured with free flow techniques (e.g. slump for concrete and slump flow for mortar). continuous phase: Constituting the medium, a phase that exhibits continuity throughout the dispersion. E.g. the liquid in a suspension.



creep: The response of a material to the instantaneous application of a constant stress. Deborah number: The ratio of a characteristic relaxation time of a material to the duration of the observation. In equilibrium flow, the effective duration of the experiment is infinity, and De=0. deformation coefficient: A measure of the sensitivity of increased water content. deformation: Movement of parts or particles of a material body relative to one another such that the continuity of the body is not destroyed, resulting in a change of shape, volume or both. density: Mass per unit volume. For concrete and mortar density, the bulk density is normally substituted, but for the constituents it is the density of solids or the apparent density (interchangeably with specific gravity). desorption: The process by which the amount of an adsorbed substance is reduced. dilatant: A property often associated with suspension of irregularly shaped particles, in which the liquid exhibits an increase in volume while being sheared. The term 'dilatant' is commonly used in practice to indicate shear-thickening, although this usage is strictly incorrect. disjointing pressure: Repulsing pressure due to adoption of water molecules acting in locations where the distance between two surfaces is restricted. The forces acts in the range of 40-100% RH, increasing with increased RH. dispersed phase: In a dispersion, the phase that is distributed in the form of discrete discontinuities (particles, droplets or bubbles). dispersion: A two-phase system in which discontinuities of any kind (solid, liquid, gas) are dispersed in a continuous phase of a different composition or state. In the field of cementitious materials, dispersion is used to describe a suspension of solid particles in a liquid medium (see suspension). drying shrinkage: Contraction caused by the loss of water from the hardened cementitious material. Similar to the term 'evaporation shrinkage', but reserved for hardened state. elastic: A conservative property in which part of the mechanical energy used to produce deformation is stored in the material and recovered on release of stress. electrical double layer: A model for the situation at a charged solid surface in solution, where ions of opposite charge are attracted to the surface but try to diffuse away and remain uniformly distributed in the solution, resulting in two plates of an electrical condenser ? the charged surface and a balancing charge in solution spread over a short distance from the surface. elongation: A shape factor, defined by the ratio of length to width of a rectangle with sides parallel to the longest dimension of particles. emulsion: A dispersion consisting of two or more liquid phases. equilibrium (steady-state) flow: Condition under which constant stress or shear rate is maintained for sufficient time to allow a steady state to be achieved in a fluid containing time-dependent structure (such as concrete). An equilibrium flow curve can be used to characterize the time-dependent flow properties of a material (e.g. thixotropy). error: Result of a measurement minus the value of the measurand. evaporable water: Water set in cement paste present in capillaries or held by surface forces; measured as the water removable by drying under specified conditions. evaporation shrinkage: When water evaporates from the fresh and hardening cementitious materials, tensile stress builds up in the capillaries, causing the cement paste, mortar or concrete to contract. Evaporation shrinkage is generally reserved for the loss of water at early ages, whereas drying shrinkage is reserved for the long-term loss. F-circle: A shape measure, calculated as the ratio of measured area to the area of a circle with the same perimeter as that measured. At 1, the object is circular. Feret diameter: Geometrically defined as the perpendicular distance between parallel lines tangential to the perimeter at opposite sides of a 2D object in a certain direction. When measuring the Feret diameter in several predetermined directions, the obtained maximum value is the maximum Feret diameter.



filler: A general term for very fine granular material of solid material, usually inert and less than roughly 0.125 mm, which occupies space and may improve physical properties or lower costs. Filler can either be an additive or a fine fraction of aggregate (see powder). filling ability: The concrete's ability to flow into and fill the mould. final setting: Structural phase of the cementitious system matrix transforming to a true rigid, controlled by its hydration. The final setting is manifested by an increasing strain capacity and a distinct decreased rate of autogenous deformation. At the transitional period between initial and final set, the matrix changes from true fluidity to true rigidity state. fine aggregate: A general term for aggregate of such size that it substantially passes a sieve of specified size, commonly 8.0, 4.0 or 4.75 mm. fineness modulus: An index of fineness or coarseness of an aggregate sample. An empirical factor determined by adding total percentages of an aggregate sample retained on each of a specified series of sieves, and dividing the sum by 100. fines: Material in aggregate finer than a given sieve, usually the 0.125 or 0.074 mm sieve (see powder). Used for expressing the amount of fine aggregate in a concrete mixture as a percent by absolute volume of the total amount of aggregate. flocculation: The process by which particles in a suspension are mutually attracted by a combination of van der Waals and electrostatic forces, stick together to form agglomerates and either settle out of the suspension as flocs or form a three-dimensional network throughout the suspension. flow curve: A graphical representation of the material's flow behaviour in which shear stress is related to shear rate. The flow curve is a type of rheogram. flow hysteresis: A condition resulting from differences in the rate of energy dissipation due to shear history. In a typical rheometric test, shear stress or shear rate is ramped at a fixed speed up to a maximum value, and then ramped back down at the same speed to the beginning. In hysteresis, one flow curve lies above the other, forming a continuous loop whose internal area depends on the shear and thermal history of the material, and on how rapidly the stress or shear rate was ramped. If the down-curve lies below the up-curve, then it is referred to as a thixotropic loop, whereas if the down-curve lies above the up-curve, then it is called a negative thixotropic loop. The area enclosed in this loop is a commonly used measure of thixotropy/anti-thixotropy. flow: Time dependent irrecoverable deformation. flowability: A measure of the consistency of freshly mixed concrete, mortar, or cement paste. Usually expressed in terms of spread diameter of a slump flow test (see workability). fluidity: The reciprocal of viscosity. free moisture: Moisture having essentially the properties of pure water in bulk; moisture not absorbed by aggregate. See also surface moisture. F-shape: A shape measure, calculated as the ratio of the length of the minor axis to the length of the major axis. If the elongation is 1, the object is roughly circular or square. grading: See size distribution. grout: A mixture of cementitious material and water, with or without aggregate, proportioned to produce a pourable consistency without segregation of the constituents; also, a mixture of other composition but of similar consistency. Herschel-Bulkley: A three-parameter rheological model used to describe viscoplastic materials exhibiting a yield response with a shear-thinning relationship above the yield stress. hydration: Chemical reaction which takes place as a result of combining cement and water. Maximum hydration equals maximum potential strength of cementitious materials. initial setting: Transition phase of cementitious system when the matrix changes from a concentrated suspension of flocculated particles to a viscoelastic skeletal solid capable of supporting an applied stress. At initial setting, capillary pore pressure rises significantly. Before initial setting, the chemical shrinkage is equal to the autogenous shrinkage, and furthermore, the total shrinkage will transform into



settlement. linear deformation: The relative change in length due to shrinkage or expansion, referred to the specimen origin length. liquid phase: Consisting of a condensed fluid, e.g. the dispersion or suspending media in a suspension.mixing water: The water in freshly mixed sand-cement grout, mortar, or concrete, exclusive of any previously absorbed by the aggregate (e.g., water considered in the computation of the net water-cement ratio). See also surface moisture. mobility: An ability of a fresh mix to flow and fill formwork or other container spaces. moisture content of aggregate: The ratio, expressed as a percentage, of the weight of water in a given granular mass to the dry weight of the mass. monolayer adsorption: Adsorption in which only a single layer of molecules becomes adsorbed at an interface. In monolayer adsorption, all adsorbed molecules are in the position of closest approach to the substrate surface. mortar: Concrete with essentially no aggregate larger than about 4 mm, i.e. a mixture of cement, sand and water. negative thixotropy: A reversible time-dependent increase in viscosity at a particular shear rate. Shearing causes a gradual growth in structure over time (see flow hysteresis). Newtonian: A flow model of fluids in which a linear relationship exists between shear stress and shear rate, where the coefficient of viscosity is constant with shear rate ( ηγσ ⋅= & ).

particle phase (or solid phase): The particles in a suspension, gel or aerosol. particle size distribution: The division of particles of a graded material among various sizes; for concrete materials, usually expressed in terms of cumulative percentages larger or smaller than each of a series of diameters or the percentages within certain ranges of diameter, as determined by sieving. particle: Any condensed-phase three-dimensional discontinuity in a dispersed system may generally be considered a particle. The term is normally used in reference to solid materials. An aggregate may also be regarded as a particle. passing ability: The concrete's ability to flow through narrow spaces and prevent blocking around reinforcements. paste: The fraction of concrete comprising powder (including cement), water, air and admixtures. plastic shrinkage: Net sum of early-age volume changes, including autogenous shrinkage and shrinkage due to loss of water caused by evaporation, the aggregates' and filler materials' adsorption andabsorption, and suction of water by sub-base or formwork material. plastic viscosity: The excess of shear stress over yield stress divided by the shear rate (see Bingham model). In a flow curve, the viscosity is represented as the slope of the curve. For concrete, mortar or other non-Newtonian materials, the term is often used synonymously with viscosity. plastic: The property of a solid body that is in the elastic state when the stress is below a critical value, termed the yield stress, and in plastic state when this value is exceeded. plasticizer: An admixture which increases workability/flowability of freshly mixed mortar or concrete without increasing water content. Also known as water-reducing admixture because it can maintain the same workability/flowability of the mix with reduced water content. plug flow: Flow in a pipe of a material that possesses a yield value (see yield stress). Because of the existence of the yield value, flow starts near the surface of the pipe and the material flows forward as a solid plug (i.e. the core of the material has a zero velocity gradient). porosity: The ratio, usually expressed as a percentage, of the volume of voids in a material to the total volume of the material, including voids. Portland cement: A finely pulverized clinker produced by burning mixtures containing lime, iron, alumina, and silica at high temperature and in definite proportions, and then intergrinding gypsum to give the properties desired.



powder: A general term for relatively dry, undispersed accumulation of very fine granular material of solid material including cement and those having a fineness equal to cement. Maximum dimension in any given direction is less than roughly 0.125 or 0.074 mm. It will also include this size fraction of aggregate. precision: The degree of agreement within a set of observations or test results obtained as directed in a test method. pseudoplastic: See shear thinning. relative humidity: The ratio of the quantity of water vapour actually present to the amount present in a saturated atmosphere at a given temperature; expressed as a percentage. relaxation time: A time characterizing the response of a viscoelastic material to the instantaneous application of a constant strain. repeatability: Variability of independent test results obtained on the same material, with the same equipment, and by the same operator. reproducibility: Variability of the replicated results of same test method, obtained on identical material performed by different operators using different equipment. retained water ratio: The amount of water, adsorbed on the particle surfaces and filling the voids in the particle system, needed to initiate flow. retardation time: A time characterizing the response of a viscoelastic material to the instantaneous application of a constant stress. rheology: The science of the deformation and flow of matter. rheometer: An apparatus for measurement of rheological properties. rheopexy: An effect by which a material recovers some of its pre-sheared viscosity at a faster rate when it is gently sheared compared to when it is allowed to stand. The term 'rheopexy' is commonly used in practice to indicate negative thixotropy, although this usage is strictly incorrect. robustness: The concrete's ability to resist deviation in the constituents' proportions (water, cement, aggregate and additives) and properties (gravel moisture content, aggregate grading, etc.). roundness: The ratio of the average radius of curvature of the several edges of corners of a particle to the radius of curvature of the maximum inscribed sphere. sand: The fine granular material (usually passing the 4 mm sieve and predominantly retained on the 0.125 mm) resulting from the natural disintegration of rock, or from the crushing of friable sandstone. saturation: The condition of coexistence in stable equilibrium of either a vapour and a liquid or a vapour and solid phase of the same substance at the same temperature. Applied to aggregate or concrete, the condition such that no more liquid can be held or placed within it. segregation: A loss of uniformity of fresh concrete mix. A separation of one or more of the constituents of the mix with consequent non-uniform proportions. self-compacting concrete: A modified standard concrete that, without the influence of additional compaction energy, flows, de-airs and completely fills the formwork only under the influence of its own gravity weight. It has to combine high fluidity with a high segregation resistance and has to be able to carry coarse aggregate grains through narrow spaces. self-desiccation: Lowering of internal relative humidity in a closed isothermal cement paste system due to the consumption of capillary water and the chemical shrinkage from progress of cement hydration. separation: The tendency, as concrete is caused to pass from the unconfined ends of chutes or conveyor belts, for coarse aggregate to separate from the concrete and accumulate at one side; the tendency, as processed aggregate leaves the ends of conveyor belts, chutes, or similar devices with confining sides, for the larger aggregate to separate from the mass and accumulate at one side; the tendency for solids to separate from the water by gravitational settlement. See also bleeding and segregation. setting: Onset of solidification in a plastic cement paste system. The beginning of solidification, referred to as initial set, marks the point in time when the paste loses its plasticity, becomes unworkable



and is capable of supporting an applied stress. settlement: Vertical displacement in cementitious materials before initial setting, which is caused by bleeding, chemical shrinkage, etc. shear rate: The change of shear strain with time (dγ/dt). For fluids, the shear rate, rather than the strain, is generally used in describing flow. shear strain: The relative deformation of shear divided by the separation distance of shearing forces. Alternatively, the shear strain can be defined as tan(θ), where θ is the angle of deformation. shear stress: The component of stress that causes successive parallel layers of a material body to move, in their own planes, relative to each other. shear thickening: An increase in viscosity with increasing shear rate during steady shear flow (not to be confused with dilatant). shear thinning: A decrease in viscosity with increasing shear rate during steady shear flow. shear: The relative movement of parallel adjacent layers. shrinkage: When the deformation is a contraction, it may be referred to as shrinkage, e.g. autogenous or drying shrinkage. sieve analysis: The classification of particles, particularly of aggregates, according to sizes as determined with a series of sieves of different openings. size distribution: The distribution of particles of granular material among various sizes, usually expressed in terms of cumulative percentages larger or smaller than each of a series of sizes (sieve openings) or the percentages between certain ranges of sizes (sieve openings). Also named grading. slump flow test: A commonly used consistency test method for determination of vertical slump flow spread and velocity of highly flowable concrete and mortar, expressed in terms of average spread diameter and flow time. Also named spread flow cone test. slurry: A concentrated cement-based particulate suspension. sol: A liquid dispersion containing particles of colloidal dimensions. specific surface area: The total (external and corresponding internal) surface area per unit weight or unit volume. sphericity: The degree to which a particle approaches a spherical shape, defined as the ratio between the diameter of a sphere with the same volume as the particle and the diameter of the circumscribed sphere. stability: The concrete's capability of retaining homogeneous, i.e. resisting segregation and bleeding during handling or placing of the mix. Also named segregation resistance. stress growth: An increasing stress vs. time or modulus vs. time function is termed stress growth, when an instantaneous and constant strain (or shear rate) is applied to a material while stress is measured over time. stress relaxation: A decreasing stress vs. time or modulus vs. time function is termed stress relaxation, when an instantaneous and constant strain (or shear rate) is applied to a material while stress is measured over time. stress: Force per unit area (σ=F/A). structure: In rheology, structure is a term that refers to the stable physical bounds between particles in a fluid. These bounds result in aggregates, flocs or network structures, which impact the fluid and provide elastic and plastic properties. subsidence: See settlement. superplasticizer: A high-efficiency plasticizer (see plasticizer). surface moisture: Water retained on surfaces of aggregates capable of mixing with Portland cement in concrete; distinguished from absorbed moisture, which is contained inside the aggregate particles. surface texture: Degree of roughness or irregularity of the exterior surfaces of aggregate particles or



hardened concrete. surface voids: Cavities visible on the surface of a solid. See also porosity and surface texture. suspension: A liquid in which solid particles are dispersed (see dispersion). thixotropy: A reversible time-dependent decrease in viscosity at a particular shear rate. Shearing causes a gradual breakdown in structure over time. The thixotropy is a measure of applied work needed to break down the structure. In a flow curve, the thixotropy causes a flow hysteresis effect (see flow hysteresis). Note that thixotropy is a reversible phenomenon; the term is often used wrongly in concrete technology to describe an irreversible change. van der Waals force: A short-range electromagnetic force interacting between two non-polar molecules (or atoms), where the force arises when the molecules become polar for a short time because their electrons are in constant motion. This motion is usually balanced or symmetrical around the non-polar molecule, but if the electrons are briefly disturbed their negative charge may increase at one part of the molecule, creating a positive charge on another part of the molecule. The molecule becomes a dipole for a short time until the electrons rebalance, disturbing electrons in neighbouring molecules, turning them into dipoles as well. The force also acts between macroscopic bodies, as particle-particle and particle-surface, where if two solid surfaces are brought close together, all the atoms in the surface zones will contribute to the forces of attraction. The van der Waals forces are much weaker than the force that arises from electrostatic charged particles, but are strong enough to hold some particles together. vapour pressure: The pressure exerted by a vapour that is calculated from relative humidity and temperature. The higher the humidity, and the higher the temperature in degrees Celsius, the greater the vapour pressure exerted. viscoelastic: A time-dependent property in which a material under stress produces both a viscous and an elastic response. A viscoelastic material will exhibit viscous flow under constant stress, but a portion of mechanical energy is conserved and recovered after stress is released. viscoplastic: A hybrid property in which a material behaves like a solid below some critical stress value, the yield stress, but flows like a viscous liquid when this stress is exceeded. Often associated with aggregated particle suspensions (e.g. mortar and concrete). viscosity: The property of a Newtonian material to resist increased deformation increasingly with increasing rate of deformation. It is determined as the ratio of shear stress to shear rate ( γση &/= ) in a steady flow. For concrete or other non-Newtonian materials, often used synonymously with plastic viscosity (or apparent viscosity). viscous: The tendency of a liquid to resist flow as a result of internal friction. water-cement ratio: The ratio of the amount of water, excluding only that absorbed by the aggregates, to the amount of Portland cement in a concrete or mortar mixture; preferably stated as a decimal by weight. workability loss: The amount by which the workability of freshly mixed concrete changes during a period of time after the initial mixing. Traditionally expressed in terms of the decrease in spread diameter of a flow cone test. workability: A general term relating to behaviour of fresh concrete, comprising a number of characteristics of parameters obtained from different tests. A commonly used definition is: “that property of freshly mixed concrete or mortar which determines the ease with which it can be mixed, placed, consolidated and finished”. Usually expressed in terms of spread diameter of a slump flow test (see flowability). yield stress: A critical shear stress value below which an ideal plastic (Bingham-plastic) or viscoplastic material behaves like a solid (i.e., will not flow). Once the yield stress is exceeded, a plastic material yields (deforms plastically) while a viscoplastic material flows like a liquid. For non-ideal viscoplastic materials (e.g. concrete), where the yield stress is indefinite, an apparent yield stress can be defined by extrapolation from the linear shear rate of the flow curve to the shear stress axis (see Bingham model). Z-potential: A suspended charged particle in a solution attracts ions of opposite charge to those at its



surface, where they form the Stern layer. To maintain the electrical balance of the suspending fluid, ionsof opposite charge are attracted to the Stern layer. The potential at the surface of the part of this diffuse double-layer of ions that can move with the particle when subjected to a voltage gradient is the zeta-potential. The magnitude of the Z-potential is approximately proportional to the surface charge on the particle.

References: Bartos P., “Fresh concrete: Properties and tests”, Department of Civil Engineering, Paisley College, Scotland, 1992.

Bentur A., “Terminology and definitions”, RILEM report 25: Early Age Cracking in Cementitious Systems, RILEM TC181-EAS, pp. 13-15, 2002.

EFNARC, “Specifications and guidelines for self-compacting concrete”, European Federation for Specialist Construction Chemicals and Concrete Systems, Farnham, 2002.

EFNARC , “European guidelines for self-compacting concrete: Specification, production and use”, European Federation for Specialist Construction Chemicals and Concrete Systems, Farnham, 2005.

Hackley V. A., Ferraris C. F., “Guide to rheological nomenclature: Measurements in ceramic particulate systems”, NIST special publication, No 946, National Institute of Standards and Technology, Washington, 2001.

Hackley V. A., Ferraris C. F., “The Use of Nomenclature in Dispersion Science and Technology”, NIST publication No 960-3, Washington, 2001.

Holt E.E., “Early age autogenous shrinkage of concrete”. Technical Research Centre of Finland, VTT Publications 446, 2001.

Jensen O.M., “Autogenous phenomena in cement-based materials”, Department of Building Technology and Structural Engineering, Aalborg University, Denmark, 2005.

Jillavenkatesa A., Dapkunas S.J., Lum L-S.H., “Particle Size Characterization”, NIST, Puplication No 960-1, Washington, 2001.

Lura P., “Autogenous deformation and initial curing of concrete”, Technical University of Delft, Netherlands, 2003.

Mehta K.P., Monteiro P.J.M., “Concrete: Structure, properties, and materials”, 3rd Ed., McGraw-Hill, 1993.

Mindess S., Young J.F., Darwin D.,”Concrete”, 2nd ed., Pearson Education Inc., 2003.

Swedish Concrete Association, “Self-compacting concrete -Recommendations for use”, Report No. 10, Svenska Betongföreningen, 2002.

Tattersall G.H., Banfill P.F.G., “Rheology of fresh concrete”, Pitman, London, 1983.

Tattersall G.H., “Workability and quality control of concrete”, E&FN Spon, London, 1991.

Tazawa E-I., ”Autogenous Shrinkage of Concrete”, Taylor and Francis, 1999.

P a p e r I - S l u m p f l o w v a l u e s v s . B i n g h a m p a r a m e t e r s f o r h i g h f l o w a b l e m o r t a r s a n d c o n c r e t e s

PAPER I: SLUMP FLOW VALUES VS. BINGHAM PARAMETERS FOR HIGH FLOWABLE MORTARS AND CONCRETES

Esping O., “Slump flow values vs. Bingham parameters for high flowable mortars and concretes”, Accepted for publication at 5th International RILEM Symposium on Self-Compacting Concrete, 3-5 September 2007, Ghent, Belgium, 2007.


1

SLUMP FLOW VALUES VS. BINGHAM PARAMETERS FOR HIGH FLOWABLE MORTARS AND CONCRETES

Oskar Esping

Dep. of Civil and Environmental Engineering, Chalmers University of Technology, Sweden.

Abstract In the present work, the influence of Bingham rheology parameters on the slump flow

values of self-compacting concrete and mortar has been evaluated. The objective was to present experimental results, without any physical validation, addressing the complex connection between the slump flow spread, flow time (T50), yield stress and plastic viscosity. A large number of more or less self-compacting mortars (~200 mixtures) and concretes (~550 mixtures) with a wide range of consistency have been used for the evaluation.

The mortar rheology was measured using a Bohlin CVO200 rheometer with concentric setup, and the slump flow spread was measured using a mini-cone. For the concrete a ConTec Visco5 rheometer was used together with a traditional Abram´s cone.

The results showed a large scatter and poor correlation, but a clear tendency was found. The results indicate that the slump flow spread and T50 time are not a unique function of yield stress or viscosity, respectively, but rather a more complex function of both. The spread proved to be more closely connected with the yield stress than with the viscosity, especially at high viscosity, whereas the T50 time was more dependent on both viscosity and yield stress.

1. INTRODUCTION

To quantify the workability of fresh concrete, numerous tests of different types, more or less empirical, have been developed. The oldest and most frequently used test today is the slump cone test, which, associated with the Abram’s cone, has its origin in the USA around 1910 [1]. For high-flowable concrete, such as self-compacting concrete (SCC), the slump cone is used to measure the spread of the concrete (slump flow) and the time for the concrete to reach a spread of 500 mm (T50). For a more scientific approach, a rheometer/viscometer can be used. The most established parameters used to define the concrete rheology are the yield stress and plastic viscosity by the Bingham equation [2]. These rheological parameters, unlike the measures of qualitative tests (e.g. the slump), are fundamental physical quantities, mutually independent and not dependent on operator or equipment. But rheometer/viscometer tests are often expensive, immobile, and mainly for laboratory use. In the field, e.g. at concrete plant or construction site, fast, simple and cheap tests are preferred; thus slump/slump flow and other empirical tests can hardly be replaced by a rheological test, but should rather be a complement. To describe the flow properties of concrete, especially for types with high flowability, a minimum of two independent parameters is required [2]-[7]. Concrete mixtures with identical measures from an empirical test (e.g. slump flow) can be very different in their flow behaviour. These differences can be detected by a rheological test. However, in order to determine the properties of a non-Newtonian fluid, as all


2

cementitious materials are, a multipoint flow curve has to be measured. A single point measurement does not describe these materials correctly (see Figure 1).

Shear rate, γ ·

Viscosity (plastic), ηpl

She

ar s

tress

, σ

Shear rate, γ ·

Yield stress, σ0

She

ar s

tress

, σ

Shear rate, γ ·

She

ar s

tress

, σ

(a) (b) (c)

Figure 1: Illustration of concrete with: (a) same viscosity but different yield stress, (b) same yield stress but different viscosity, and (c) same empirical measure but different viscosity and yield stress (from [31]).

It has been shown by many that the empirical tests to some extent provide information about the concrete rheology. Slump has for a long time [3][6]-[22] been considered to be a function of yield stress, without influence by plastic viscosity. When analysed, the density of the concrete is often incorporated as well as the geometry of the cone. More recently the horizontal flow spread rather than the vertical slump has been used to calculate the yield stress [2][23]-[25]. Usually these correlations concern concretes in a small range of flowability (e.g. HPC, SCC, etc.), and are not valid for other or wider ranges. The increased usage of high-flowable concretes has also raised the interest in more dynamic measures such as flow time and viscosity. The flow time can be measured by e.g. a V-funnel or L-box, but also by measuring the time for the concrete to spread 500 mm (T50). For SCC, the slump flow spread is generally considered to correlate with the yield stress and T50 or V-funnel with the viscosity [26]-[28]. Often the correlation is purely empirical, but there are also some physical models proposed to explain this relationship. Examples of physical models linking yield stress (σ0) to slump flow spread diameter (D) are:

20 DVg2

⋅⋅⋅⋅

=πρσ 2D/a= (Eq. 1) [14]

20 D3Vg4

⋅⋅⋅⋅⋅

=π

ρσ 2D/b= (Eq. 2) [29]

52

2

0 D4Vg225

⋅⋅⋅⋅⋅

=πρσ 5D/c= (Eq. 3) [30]

where ρ is the material density, V the sample volume and g the gravity, while a, b and c are constants for a given sample volume and density.

The objective of this paper is not to find a physical relationship between slump flow values and Bingham parameters, but merely to present experimental results addressing the interaction between the slump flow spread, flow time (T50), yield stress and plastic viscosity.

2. TEST METHODS

2.2 Slump flow test The slump flow measurement was carried out using a traditional slump cone (see Figure

1), for concrete the Abram’s cone (EN 12350-2 or ASTM C143) and for mortar the mini-slump cone (EN 1015-3 or ASTM C230), but without compaction. The slump flow spread diameter was measured in two perpendicular directions, presented as a mean value in


3

millimetre. In addition to the slump flow, the time from lifting the cone to when the flow spread reaches a 500 mm circle was recorded; referred to as T50 and stated in seconds.

Slump cone (Abram) ØD=100/200 mm

H=300 mm

Concrete sample Flow table (non-adsorbent)

Spread diameter T50 time

Figure 2: Illustration of slump flow test used for mortar and for concrete (from [31]).

2.2 Rheology For mortar a Bohlin CVO200 rheometer was used and for concrete a ConTec Visco5, both

using a rotating concentric setup at a controlled shear rate (γ& ) at 20ºC. The experimental geometry and measuring sequence is illustrated in Figure 3. The two rheological parameters plastic viscosity (ηpl) and yield stress (σ0) were evaluated in accordance with the Bingham model: pl0 ηγσσ ⋅+= & (Eq. 4).

0

20

40

60

0 10 20 30 40 50 60Time [s]

Shea

r rat

e [1

/s] No logging

Logging

Ro=15 mmRi=10 mm

H=37.5 mm

Rotating and measuring inner cylinder

Fixed outer cylinder

Mortar sample

0

2

4

6

8

0 10 20 30 40 50 60Time [s]

She

ar ra

te [1

/s]

No logging

Logging

Ro=145 mmRi=100 mm

H=140 mm

Measuring inner cylinder Rotating outer cylinder

Concrete sample

Figure 3: The measuring sequence for the Bingham evaluation (followed by segregation

estimation) and schematic illustration of the rheometers. For mortar the Bohlin CVO200 and for concrete the ConTec Visco5 (from [31]).

The mixes and their constituents, as used in this study, comprise typical materials for self-compacting concrete (SCC) in Sweden. For mortar the maximum grain size was 1.0 mm, and for concrete 16 mm.

The consistency in some cases was very stiff, but as all mixtures incorporated both filler and superplastisizer they were considered to be self-compacting. All mixtures have their origin in other projects whose primary purpose was to examine the effect of different constituent properties and proportions on the rheology. Examples of this are the effect of: w/c, cement, silica, fly ash, different limestone fillers, coarse and fine aggregate content, superplasticizer, air entrainer, viscosity modifier, etc.

Mini-slump cone ØD=70/100 mm

H=60 mm

Mortar sample

Flow table

Spread diameter


4

The mortar samples were prepared in batches of 0.5-1.0 litres, using a 4.73 litre Hobart laboratory paddle mixer. The mixing sequence was based on ASTM C305 standard. The concretes were prepared in batches of 30-60 litres, mixed in a BHS60 twin-shaft paddle mixer for 4 minutes after water was added to the premixed dry materials. The admixtures were added directly after the water.

4. RESULTS AND DISCUSSION The presented results are based on a large number of more or less self-compacting

mortars (~200 mixtures) and concretes (~550 mixtures) with a wide range of consistency. The density is also considered to influence the slump and slump flow (see Eq. 1-3), but as the difference in density of the mixes was relatively small, it was not taken into account. Concretes and mortars with high tendency of visual segregation were excluded, as these are considered to give misleading measures and thus a poor correlation between the slump flow and rheology. For concrete slump flow spread (D) and time to 500 mm (T50) were recorded and correlated with the Bingham parameters (yield stress and plastic viscosity). For mortar only the spread was recorded and correlated with the Bingham parameters. The presented correlations are not intended as a “true connection” of slump flow values and Bingham parameters, but merely to show the more complex connection between the slump flow spread, flow time, yield stress and plastic viscosity.

4.1 Rheology vs. slump flow spread The experimental results of slump flow spread diameter and Bingham rheology

parameters (yield stress and plastic viscosity) for the mortar and concrete samples are presented in Figure 4. The spread flow is grouped into intervals of 50 mm for mortar and 100 mm for concrete, starting with spread equal to the cone bottom diameter (D0=100 mm respectively 200 mm). For each spread interval, a logarithmic trend-line is plotted, as this was the tendency that was best fitted over a wide range of spreads. It can be noted that there is a large scatter and rather poor correlation (R2=0.1-0.9) but still a clear tendency can be observed that the slump flow is affected by both yield stress and viscosity. At high viscosity or low yield stress, where the slope of the curve is lower, the effect of yield stress is more pronounced. If slump flow were to be independent of viscosity the curve would be horizontal. The slump flow can seem to be independent of viscosity for the higher flowable mortars and concretes (with slump flow >250 respectively >500 mm), as the curve looks horizontal. When excluding mixtures with lower spread flow it can be observed that also here the spread is affected by both yield stress and viscosity.


5

0

10

20

30

40

50

60

70

0 2 4 6 8Viscosity [Pa s]

Yie

ld s

tress

[Pa]

>350 mm300-350 mm250-300 mm200-250 mm150-200 mm100-150 mm

100-150 mm

150-200 mm

200-250 mm

250-300 mm300-350 mm

>350 mm

0

100

200

300

400

500

600

0 50 100 150 200Viscosity [Pa s]

Yie

ld s

tress

[Pa]

>700 mm

600-700 mm

500-600 mm400-500 mm

300-400 mm

200-300 mm

200-300 mm

>700 mm600-700 mm500-600 mm

400-500 mm

300-400 mm

Figure 4: The slump flow spread (in intervals of 50 resp. 100 mm) and Bingham rheology

parameters (yield stress and plastic viscosity) for the mortar and concrete.

Based on all the investigated mixes (> 750 mixes), the following approximate model for calculating the relative slump flow area (Rp) for both mortar and concrete is proposed:

0pl

2

p )2ln()6/H(Rση ⋅+

≈ (Eq. 5) 1)D/D(R 20p −= (Eq. 6)

where ηpl is the plastic viscosity, σ0 the yield stress, H the cone height (60 resp 300 mm), D the measured spread diameter in millimetre and D0 the flow cone bottom diameter (100 resp 200 mm). The model is purely empirical and is deficient in its units. In Figure 5 the model is plotted for mortar and concrete.

0

10

20

30

40

50

60

70

0 2 4 6 8Viscosity [Pa s]

Yie

ld s

tress

[Pa]

100-150 mm

>350 mm300-350 mm250-300 mm

150-200 mm200-250 mm

0

200

400

600

0 50 100 150 200Viscosity [Pa s]

Yie

ld s

tress

[Pa]

200-300 mm

>700 mm600-700 mm500-600 mm400-500 mm

300-400 mm

Figure 5: Calculated slump flow spread diameter as a function of Bingham rheology

parameters for mortar and concrete.

In Eq.5 the yield stress is predominant, where the effect of a change in yield stress corresponds to ~3 times larger change in viscosity on the slump flow. These differences are smaller at low viscosity, and are independent of the magnitude of yield stress.

In the literature, not all studies agree that slump flow spread is a measure of concrete yield stress. For example, Nielssen & Wallevik [32] explain how a low viscosity will increase the downward speed, overcome the yield stress and thereby generate a large slump flow

Mortar Concrete


6

spread. For high viscosity the effect will be the opposite. Cauberg et al [33] state that the relation between yield stress and slump flow, without taking viscosity into account, does not show a satisfactory fit, and that these relations usually are formed for a certain range of fluidity. Furthermore, a variable Z0, comprising both yield stress (σ0) and viscosity (ηpl), is proposed for a better fit to the slump flow spread (D) in millimetres.

pl00 /baZ ησ +⋅= (Eq. 7) D586.0381Z0 ⋅−= (Eq. 8) where a and b are constants. In the region of spread D=400-600 mm, and with a=1 and b=1000, Eq. 7 and Eq. 8 generate similar results as Eq. 5 and Eq. 6. For spread D>650 mm Eq. 8 is not valid. Furthermore, the shape of the curve in a ηpl-σ0 diagram is independent of the spread. Smeplass [34] has noted that the slump is more related to the yield stress than to the viscosity of concrete, while the flow table spread is more related to the viscosity than the yield stress. If the viscosity varies, at constant yield stress, so will the flow measures.

4.2 Rheology vs. slump flow time T50 The experimental results of slump flow spread time T50 and Bingham rheology

parameters for the concrete samples are presented in Figure 6. Concretes with slump flow spread <500 mm have been excluded as they have no T50 measure (~380 mixtures with slump spread >500mm). The T50 times are grouped into intervals of 2 seconds. For each T50 time interval a trend-line is plotted. It can be noted that there is a large scatter and poor correlation (R2=0.1-0.4). Still the tendency can be observed that the T50 time is affected by both yield stress and viscosity. If the flow time were independent of yield stress the curve would be vertical.

0

40

80

120


Yie

ld s

tress

[Pa]

>10 s8-10 s6-8 s4-6 s2-4 s< 2

>10 s8-10 s

6-8 s4-6 s2-4 s

<2 s

Figure 6: The experimental results of slump flow spread time T50 (in intervals of 2 s) and

Bingham rheology parameters (yield stress and plastic viscosity) for the concrete.

Based on the results, the following empirical model for calculating the slump flow spread time T50 in seconds is proposed:

pl0002.050T ησ ⋅⋅≈ (Eq. 9) In the literature, there is no general agreement that T50 is a measure of concrete plastic

viscosity (ηpl). For example, Tedaka et al [35] state that the slump flow spread corresponds well to the yield stress (σ0), whereas the T50 flow time does not alone represent the viscosity since slump flow also contributes. Thus, T50 can be used to evaluate the viscosity only when the slump flow is constant. Furthermore, Ferraris et al [36] show that the slump flow spread times of cement pastes with a wide range of yield stress have no correlation at all with the viscosity. Emborg [37] has also noted a weak correlation, but then for the T50 vs. viscosity


7

of SCC. Utsi et al [38] have observed that T50 correlates not only to the viscosity but also to the yield stress, and proposes that both the yield stress and viscosity influences the T50 and slump flow respectively. By dividing the slump flow into different intervals the correlation between T50 and viscosity was shown to be acceptable, which further fortified this theory.

5. CONCLUSIONS It has been shown by many investigators that the results of empirical tests are strongly

related to the concrete rheology. For SCC, the slump flow spread is generally considered to be a function of yield stress and T50 or V-funnel of viscosity. Often this correlation is purely empirical, but there are also different physical models explaining this relationship. In the presented work, the influence of Bingham rheology parameters on the slump flow values of a large number of more or less self-compacting concrete and mortar has been evaluated. Experimental results addressing the connection between the slump flow spread, flow time (T50), yield stress and plastic viscosity are presented. The results clearly indicate that the slump flow spread and T50 time are not a unique function of yield stress and viscosity respectively, but rather a more complex function of both where neither yield stress nor viscosity can be neglected. The spread proved to be more closely connected with the yield stress than the viscosity (~3 times), especially at high viscosity, whereas the T50 time was connected more equally with yield stress and viscosity. Consequently, T50 can only be used to estimate the viscosity for mixture with constant yield stress, and slump flow can only be used to estimate the yield stress with constant viscosity.

ACKNOWLEDGEMENTS Financial support from Thomas Concrete Group and Färdig Betong is greatly appreciated.

REFERENCES [1] Bartos P., Sonebi M., Tamimi A. K. (2002): “Workability and rheology of fresh concrete:

Compendium of tests”, RILEM, France. [2] Tattersall G.H. (1991): “Workability and quality control of concrete”, E&FN Spon, London. [3] Domone P. (1998): ”The slump flow test for high-workability concrete”, Cement and Concrete

Research, vol 28(2), pp 177-182. [4] Domone P.L.J., Xu Y., Banfill P.F.G. (1999): “Developments of the two-point workabiliy test

for high-performance concrete,” Magazine of Concrete Research, Vol 51(3), pp 171-179, 1999. [5] Ferraris C.F., “Measurement of the rheological properties of high performance concrete: State

of art report”, Journal of NIST, vol 104(5), pp 461-478. [6] Tattersall G.H. (1973): “The rationale of a two-point workability test,” Magazine of Concrete

Research, vol 25 (84), pp 169-172. [7] Tattersall G.H., Banfill P.F.G. (1983): “Rheology of fresh concrete”, Pitman, London. [8] Chidiac S.E., Madaani O., Razaqpur A.G., Mailvaganam, N.P. (2000): “Controlling the quality

of fresh concrete - A new approach”, Magazine of Concrete Research, vol 52 (5), pp 353-364. [9] Clayton S., Grice T.G., and Boger D.V. (2003): “Analysis of the slump test for on-site yield

stress measurement of mineral suspensions”, Journal of Mineral Processing, Vol 70, pp 3-21. [10] de Larrard F. (1999): “Concrete mixture proportioning, a scientific approach”, F&FN Spon,

New York. [11] de Larrard F., Hu C., Sedran T., Szitkar J.C., Joly M., Claux F., Derkx F. (1997): “A new

rheometer for soft-to-fluid fresh concrete”, ACI Materials Journal, vol 94(3), pp 234-243. [12] Ferraris C.F., de Larrard F. (1998): “Testing and modelling of fresh concrete rheology”,

NISTIR 6094, NIST, USA. [13] Hu C., de Larrard F., Sedran T., Boulay C., Bosc F., Deflorenne F. (1996): “Validation of

BTRHEOM, the new rheometer for soft-to-fluid concrete”, Materials and Structures, vol 29(194), pp 620-631.


8

[14] Murata J. (1984): “Flow and deformation of fresh concrete”, Materials and Structures, vol 17(98), pp 117-129.

[15] Murata J., Kikukawa H. (1992): “Viscosity Equation for Fresh Concrete”, ACI Materials Journal, vol 89(3), pp 230-237.

[16] Norberg J., Peterson O., Billberg P. (1997): “Effects of a new generation of superplasticizers on the properties of fresh concrete”, CANMET/ACI International Conference, pp. 583-598, Italy.

[17] Pashias N., Boger D.V., Sumers J., Glenister D.J. (1996): “A Fifty Cent Rheometer for Yield Stress Measurements”, Journal of Rheolgy, vol 40(6).

[18] Saak A.W., Jennings H.M., Shah, S.P. (2004): “A generalized approach for the determination of yield stress by slump and slump flow”, Cement and Concrete Research, Vol 34, pp. 363-371.

[19] Schowalter W.R., Christensen G. (1998): “Toward a rationalization of the slump test for fresh concrete -Comparisons of calculations and experiments”, J. of Rheology, vol 42(4), pp 865-870.

[20] Tanigawa Y., Mori H. (1989): Analytical study on deformation of fresh concrete”, Journal of Engineering Mechanics, vol 115(3), pp 493-508.

[21] Tanigawa, Y., Mori, H., Watanabe, K. (1990): “Computer simulation of consistency and rheology tests of fresh concrete by viscoplastic”, RILEM symposium on properties of fresh concrete, pp 301-308, Hanover.

[22] Wallevik J. E. (2006): “Relationship between the Bingham parameters and slump”, Cement and Concrete Research, vol 36, pp 1214-1221.

[23] Flatt R.J., Domenico L., Roussel N. (2006): “Linking yield stress measurements: Spread test versus Viscomat”, Cement and Concrete Research, vol 36(11), pp 99-109.

[24] Rossel N., N’Guyen H.T.L., Coussot P. (2005): “Yield measurements using stoppage tests”, RILEM symposium on SCC, pp 575-582, Chicago.

[25] Saak A.W., Jennings H.M., Shah S.P. (2004): “A generalized approach for the determination of yield stress by slump and slump flow”, Cement and Concrete Research, vol 34(3), pp 363-371.

[26] Domone P.L. and Jin J. (1999): “Properties of mortar for self-compacting concrete” RILEM symposium on SCC, pp 109-120, Stockholm.

[27] Grünewald S., Walraven J.C. (2003): “Rheological measurements on self-compacting fibre reinforced concrete”, RILEM Symposium on SCC, PRO 33, pp 49-58, Reykjavik.

[28] Jin J., Domone P.L. (2002): “Relationships between the fresh properties of SCC and its mortar component”, North American Conference on the Design and Use of Self-Consolidating Concrete, pp 33-38, Chicago.

[29] Chidiac S.E., Maadani O., Razaqpur A.G., Mailvaganam N.P.(2000): ”Controlling the quality of fresh concrete - a new approach”, Magazine of Concrete Research, vol 52(5), pp 353-363.

[30] Roussel N., Stefani C., Leroy R.,” From mini-cone test to Abrams cone test: measurement of cement-based materials yield stress using slump tests”, Cement and Concrete Research, vol 35(5), pp 817-822.

[31] Esping, O. (2004): “Rheology of cementitious materials: effects of geometrical properties of filler and fine aggregate”, Report 04:3, Chalmers University of Technology, Göteborg.

[32] Nielsen I., Wallevik O.H. (2003): “Rheological evaluation of some empirical test methods – Preliminary result”, RILEM symposium on SCC, pp 55-68, Reykjavik.

[33] Cauberg N., Desmyter J., Dieryck V. (2005): “Rheology of Self-Compacting Concrete - Validation of Empirical Test Methods”, RILEM symposium on SCC, pp 765-773, Chicago.

[34] Smeplass S. (1993): “Applicability of the Bingham model to high strength concrete”, RILEM workshop on special concretes – Workability and mixing, pp 145-151, Paisley.

[35] Takada K., Tangtermsirikul S. (2000): “Testing of Fresh Concrete”, Self-compacting concrete: State-of-the-art report, RILEM Technical Committee 174-SCC, Report 23, pp 25-39, France.

[36] Ferraris C., Obla K., Hill R. (2001): “The influence of mineral admixtures on the rheology of cement paste and concrete”, Cement and Concrete Research, Vol 31(2), pp 245-255.

[37] Emborg M. (1999): “Rheology tests for self-compacting concrete - How useful are they for the design of concrete mix for full scale production?”, RILEM symposium on SCC, pp 95-105, Stockholm.

[38] Utsi S., Emborg M., Carlswärd J. (2003): “Relation between workability and rheological parameters”, RILEM Symposium on SCC, PRO 33, pp 154-164, Reykjavik.

P a p e r I I - M e t h o d s f o r c h a r a c t e r i s a t i o n o f f i l l e r s a n d f i n e s f o r S C C

PAPER II: METHODS FOR CHARACTERISATION OF FILLERS AND FINES FOR SCC

Esping O., “Methods for characterisation of fillers and fines for self-compacting concrete”, 3rd International RILEM Symposium on Self-Compacting Concrete, PRO 33, 17–20 August, pp 208-219, Reykjavik, Iceland, 2003.


1

METHODS FOR CHARACTERISATION OF FILLERS AND FINES FOR SELF-COMPACTING CONCRETE Oskar ESPING Chalmers University of Technology, Sweden ABSTRACT: Due to the large surface area, quantification of fillers and the fine part of aggregates is essential for the ability to control physical properties, such as workability of self-compacting concrete. A new simple method for measuring the surface area has been investigated and compared with other traditional methods. The properties of eight different fillers (seven limestone and one glass filler) and their influence on the rheology of self-compacting mortar have been evaluated. KEYWORDS: self-compacting concrete, mortar, filler, rheology, specific surface.

1. INTRODUCTION The rheology of a particle suspension is a function of the suspending media. For concrete, optimized size distribution influence the workability positively since the suspension consists of a matrix where the suspension of smaller particles fills the hollow spaces between the coarser. To decrease the internal friction of fresh concrete, the particle surface is to be covered with water and voids in the system to be filled. More water will reduce the volume fraction of particles and thereby eases the particle motion. Consequently, when the contact zone between the particles decreases there will be a significant reduction of plastic viscosity and yield stress [1,2,4]. Besides particle concentration and size distribution, other important properties affecting the workability are particle shape, porosity and surface texture [3]. Since surface area/volume ratio is inversely proportional to linear dimension, as size decreases surface properties become more important. This effect can be exemplified by studying how the surface area changes for a cube with the edge length 10 mm when it is gradually divided into cubes with half the edge length (see Figure 1). This dividing process (showed in Table 1) leads to a drastically enlarged surface area.

1 ½

Figure 1. By dividing a cube to 8 cubes with half the edge length, the surface area will be doubled.

Table 1. How the surface area increases with the dividing process of the exemplified cube.

dividings edge length area [m2] 0 10 mm 0.0006

10 9.8 µm 0.6 25 1.19 nm 20000

During real conditions, the surface area also comprises an enlarging effect from the shape, porosity and surface texture. An irregular and porous particle can provide a considerable larger surface area, accessible to water, than a spherical. So, due to the large surface area, quantification of geometrical properties of fillers and the fine part of the aggregates is essential for the ability to control physical properties, such as workability, in the production of self-compacting concrete (SCC).


2

In this study, the properties of eight different fillers (seven limestone and one glass filler) and their influence on the rheology of self-compacting mortar have been evaluated. Traditional methods for measuring the surface area has been investigated and compared with a new simple method, called BET (H2O).

2. THE BET (H2O) METHOD

2.1 Introduction The cement and filler particles fineness is traditionally quantified with Blain measure. Due to the methodology where the specific surface area is determined from air permeability, based on packed spherical particles, information of the shape, texture and surface porosity is neglected. The Blaine measure is mainly an indication of particle concentration and size distribution. Other possibly more correct methods, such as the BET (Nitrogen) and image analysis is more seldom used due to its complexity and costly equipment. Concrete and mortar fluidity strongly depends on the size of the internal surface area that has to be covered with water to create mobility. A true value of this area, especially concerning the fines, is probably one of the quantification methods necessary for predicting the rheological properties. A methodology for this quantification is the BET (H2O).

2.2 Methodology The BET (H2O) is based on BET theory (named after Brunauer, Emmett and Teller) [8] where the specific external surface area is estimated from determining the quantity of a specific gas that adsorbs as a single layer of molecules, a so-called monomolecular layer, under controlled conditions. But instead, as normally, of using nitrogen gas at controlled temperature and pressure, moisture at two ore more relative humidity (RH) levels are used. The technique by using the BET-relation with moisture instead of nitrogen gas has been applied by Fuglsang Nielsen [6] and Ahlgren [7], but the focus was on internal porosity and not on external particle surface. It ought to be noted that moisture adsorption is not the same as for nitrogen gas. The water molecule is slightly smaller than the nitrogen molecule, and the water molecule has larger affinity forces (attraction) due to its two-polar covalent bonds (see Figure 2 and Figure 3). However, using water (moisture) is probably closer to real condition in concrete production than using nitrogen gas. To be able to use the BET theory with moisture as an adsorbent restricted to a monomolecular layer, the vapour pressure must be low. Therefore the RH is limited to approximately 30% (at 20ºC).

Mono-layer (~3.5Å for water)

Adsorbed molecules

Figure 2. Molecules adsorbed to a surface

H

H

H2O

O

+

+ N N

N2 -

Figure 3. The polar water (H2O) molecule versus the nonpolar nitrogen (N2)

When evaluating the BET (H2O)-method, a "two pressure" humidity generator (Thunder 2500, see Figure 4) where the specimens were conditioned in air vaporized at several levels


3

(6, 12, 16, 20, 25, 30 and 50% RH) with high accuracy. This was then compared with a method where the specimens were conditioned in two separate climate boxes (11.3 and 33.1% RH with LiCl and MgCl2 salt, see Figure 5). The climate box (salt) method showed adequate results, and are due to its simplicity a recommended method.

Figure 4. The Thunder 2500, a "two pressure" principle humidity generator, producing atmospheres with high accuracy.

Figure 5. Two climate boxes with RH 11% (LiCl) and 33% (MgCl2) For stable atmospheres, the box must be tight and a mechanical fan are to be rotate the air inside. The temperature must be kept constant at 20ºC.

By plotting the adsorption isotherm STP (see Eq.1 and Eq.2) as a function of the relative humidity RH, the variables a and b can be calculated by a linear relationship (see Figure 6). The specific area S is then calculated according to Eq.3. [7]

))1(( RHuRHSTP−⋅

= [kg/kg] (RH 0;1) (Eq.1)

Moisture content dry

dry

mmm

u)( −

= [kg/kg] (Eq.2)

)(1054.3 6

baS

+⋅

= [m2/kg] (Eq.3)

STP

[kg/

kg]

a

RH [-]

b

Figure 6. The intersection point (a) and slope (b) calcu-lated from the linear relation-ship between STP and RH.

Recommended equipment for the BET (H2O)-analyse is:

2 salt boxes (11% RH with LiCl and 33% with MgCl2), see Figure 5. 105ºC oven Scale (1/1000 g)

3. EXPERIMENTAL SETUP

3.1 Quantification methods

3.3.1 Surface area from size distribution The surface area of the fillers was calculated from the particle size distribution, both from the laser diffraction analyze (given in Figure 12) and from the image analysis (see section 3.3.5.), assuming spherical particles. The size distribution from the image analyze was determined according to the Nordtest method NT BUILD 486.

3.3.2 BET (N)


4

The method is described in chapter 2.2. The test was performed both by Norkalk AB (lime-stone filler producer) using a Micromeritics BET analyzer and by SP (Swedish National Testing and Research Institute) also using a Micromeritics.

3.3.3 BET (H2O) The method is described in chapter 2.2. The sample size was approximately 100 g. The test was performed both with humidity generator and climate boxes (11% and 33% RH).

3.3.4 Blaine The method determines fineness of powders in terms of specific surface. The test was made according to ASTM C204 standard, with a Tonindustrie instrument. Norkalk AB also performed this test, using a ToniPerm instrument.

3.3.5 Image analysis The image analysis was made with the "UTHSCSA Image Tool" on pictures (2016·2016 pixels) taken on epoxy-molded and polished specimens in a low vacuum scanning electron microscope (SEM). Following factors was measured and calculated:

F-shape: the ratio of the length of the minor axis (Dmin) to the length of the major axis (Dmax). If the elongation is 1, the object is roughly circular or square (see Figure 7). The F-shape is computed as: Dmin / Dmax [-] (Eq.4)

F-circle: the ratio of measured area and area of a circle with the same perimeter as measured. At 1, the object is circular (see Figure 7). The F-circle is computed as:

2

4Perimeter

Area⋅⋅π [-] (Eq.5)

Compactness: the ratio between the diameter of a circle with the same area as measured (Dcircle) and the maximum elongation (Dmax) of measured area and. At 1, the object is circular (see Figure 7). The Compactness is computed as:

max

/2DArea π⋅ [-] (= Dcircle / Dmax) (Eq.6)

Dmax

Dmin a)

Area p. Perimeter

Area c.

b)

Dmax

Dcircle

c)

Figure 7. a) F-shape; b) F-circle; c) Compactness.

3.3.6 Retained water and water sensitivity The retained water ratio βp and the deformation coefficient (sensitivity) Ep, was measured, according to a method presented by Okamura et al. [9]. The method, using mini-slump test (see 3.4.1), is based on the linear relationship between the relative flow area Rp (see Eq.4) and the water by powder ratio by volume Vw/Vp (see Figure 8).


5

20

20

2 )(D

DDRp

−= [m2/m2] (Eq.7)

D is the average spread diameter, calculated in four directions and D0 is the base diameter of cone (Figure 8). The measure of retained water ratio βp can be considered as the amount of water, adsorbed on the particle surfaces and filling the voids in the particle system, needed to initiate flow. The deformation coefficient Ep is a measure of the sensitivity of increased water content [10].

V w/V

p [-]

βp

Rp [-]

Ep

Figure 8. βp is represented by the interception point and the Ep the slope, calculated from the linear relationship between Vw/Vp and Rp.

3.2 Rheological test

3.4.1 Mortar flow test (mini-slump)

A traditional slump flow measure was carried out using a mini-slump cone (see Figure 9), mea-suring the spread flow diameter D (average value from two directions) The slump was measured 7 minutes after adding water to cement.

60 m

m

D2

D1

100 mm

70 mm

Figure 9. Mini slump flow test.

3.4.2 Rheology

A Bohlin CVO200 rheometer with rotating coaxial cylinder was used to measure plastic viscosity (ηpl), yield stress (σ0) and thixotropy at controlled shear rate (γ& ). The measuring device consisted an outer cylinder (cup) with diameter Do=30 mm and a grooved inner cylinder (bob) with diameter Di=20 mm, providing a gap of 5 mm (see Figure 10). The temperature was controlled and kept constant at 20±0.1°C. The evaluation of viscosity and yield stress was made according to the Bingham model (see Eq.8) and the thixotropy was evaluated from a linear up/down loop where the area in the flow curve represented the thixotropic measure.

plηγσσ ⋅+= &0 (Eq.8) The measuring sequence setup was made after consultation with Dr O. Wallevik [11] and is shown in Figure 11.

Figure 10. Bohlin CVO200 rheometer with a grooved inner cylinder (bob).

D4 D1

D0 = 100 mm

D3 D2

60 m

m

70 mm


6

Time [s]

Shea

r rat

e [1

/s]

0

10

20

30

40

50

0 10 20 30 40 50 60 70 80 90

Logging

No logging

1 2 3

Figure 11. The measuring sequence setup with a first loop (1) measuring the thixotropy, a second (2) for the Bingham evaluation and a third (3) for segregation estimation.

3.3 Materials and mix design The materials used in the experiments, are listed in Table 2. All have particles smaller than 0.5 mm. One coarse and one fine limestone filler, from three different time epochs, were selected. The idea with the differences in age is to get variations in shape, surface porosity and texture. Two additional fillers were selected, the Limus 40 (limestone) and the Mikrofiller (glass), representing "standard" (in Sweden) fillers for SCC. The fillers size distributions are presented in Figure 12 and SEM-photos as illustration shown in Figure 13. Baskarpsand 15, an industrial washed and sieved natural sand smaller than 0.5 mm, are selected to represent the fine part of concrete aggregate. All of the materials, except the L180X-limestone filler, are commercial Swedish products. The density was measured with pycnometer and the size distribution with a Malvern laser diffraction instrument, both performed by Nordkalk AB (limestone filler producer). Table 2. Materials. Material Type Name Supplier Age

[milj years] Density [kg/m3]

Cement: CEMII/A-LL42,5R Byggcement (Bygg) Cementa 2980 Aggregate: Sand Baskarpsand 15 (B15) Baskarpsand 2650 Filler: Limestone Limus 10 (L10) Nordkalk 500 2680 Limestone Limus 15 (L15) Nordkalk 80 2710 Limestone Limus 25 (L25) Nordkalk 1900 2710 Limestone Limus 40 (L40) Nordkalk 1900 2670 Limestone Limus 180 (L180) Nordkalk 500 2740 Limestone L180X Nordkalk 1900 2730 Limestone Limus 190 (L190) Nordkalk 80 2700 Returned glass MicroFiller SGÅ 2510 Plasticizer: Polucarboxylate Sikament-56 Sika 1100 The mortar recipe is based on a "normal" (in Sweden) SCC, where the proportions were scaled down due to the maximum grain size 0.5 mm. The volume of aggregate, cement and plasticizer were kept constant, while the w/c was varied in tree levels 0.45, 0.50 and 0.55. The volume distribution of the recipes is shown in Figure 14. Due to different filler density, totally 24 recipes were produced. Each mortar mix was prepared according to the ASTM Standard C 305 at 20˚C.

Yie

ld s

tress

, σ0 [

Pa]


·Viscosity,ηpl [Pa s]


1

2

3

Thixotropy [Pa/s]


7

Size distribution

0

20

40

60

80

100

0,5000,2500,1250,0630,0310,0160,0080,0040,002Size [mm]

Pas

sing

[%]

L10

L15

L25

L40

L180

L180X

L190

MF

Bygg

B15

B15

L180X

L190

MF

L40

L180Bygg

L15 L10

L25

Figure 12. Size distribution of the seven limestone fillers and the glass filler measured with a Malvern laser diffraction instrument.

Figure 13. Pictures of the fillers, taken with scanning electron microscope (SEM). The edge length of each picture responds to 0.5 mm. The specimens are epoxy-mould and polished.

w/c 0.45

Plasticizer0,3%

Water33,5%

Aggregate25,0%Filler

16,1%

Cement25,0%

w/c 0.5

Cement25,0%

Filler12,4%

Aggregate25,0%

Water37,2%

Plasticizer0,3%

w/c 0.55

Plasticizer0,3%

Water41,0%

Aggregate25,0%

Filler8,7%

Cement25,0%

Figure 14. Mix proportion by volume.

L10 L15 L25 L40

L180 L180X L190 MF


8

4. RESULTS AND DISCUSSION

4.1 Quantification methods The results from the geometrical quantifications are presented in Figure 15, Figure 16, Figure 17 and Figure 18. The quantifications incorporates the methods:

Specific surface area calculated from size distribution (laser diffraction) and from image analysis (SEM)

Blaine BET (N) and BET (H2O) Image analysis (F-shape, F-circle, Compactness) The retained water ratio βp and the deformation coefficient (sensitivity) Ep

0

100

200

300

400

500

600

700

L10

L15

L25

L40

L180

L180

XL1

90 MF

Spe

cific

sur

face

are

a [m

2/kg

] Calculated (Size distr)

Calculated (SEM)

Blaine

0

2000

4000

6000

8000

10000

L10

L15

L25

L40

L180

L180

XL1

90 MF

Spe

cific

sur

face

are

a [m

2/kg

]

BET (N)BET (H2O)

Figure 15. Specific area calculated from size distribution and measured with Blaine.

Figure 16. Specific area measured with BET (N) and BET (H2O).

Geometry

0,5

0,6

0,7

0,8

0,9

L10L1

5L2

5L4

0L1

80L1

80X

L190 MF

F-sh

ape,

F-c

ircle

and

C

ompa

ctne

ss [-

]

F-shapeF-circleCompactness

Retained water and sensitivity

0,02

0,03

0,03

0,04

0,04

0,05

0,05

0,06

L10

L15

L25

L40

L180

L180

XL1

90 MF

Def

orm

atio

n co

eff E

p [-]

0,20

0,40

0,60

0,80

1,00

reta

ined

wat

er ra

tio β

p [-]

Ep

Bp

Vw/Vp

R

Ep

βp

Figure 17. Roundness factors as F-shape, F-circle and Compactness, measured and calculated from Image analysis (SEM).

Figure 18. Retained water and sensitivity. Note that a high deformation coefficient indicates a sensitive material.


9

4.2 Rheological test

The results, listed in Figure 19, Figure 20, Figure 21 and Figure 22, are based on a large number oft test (~120 for the rheometer test and ~60 for the slump). Each test is presented as a mean value.

Rheology

1

10

100

1000

L10

L15

L25

L40

L180

L180

XL1

90 MF

Yie

ld s

tress

σ0

[Pa]

w/c 0,45w/c 0,50w/c 0,55

Figure 19. Yield stress σ0 for the w/c 0.45, 0.50 and 0.55 mortar mix.

Rheology

0

1

10

L10

L15

L25

L40

L180

L180

XL1

90 MF

Pla

stic

vis

cosi

ty η

pl [

Pa

s]

w/c 0,45w/c 0,50w/c 0,55

Figure 20. Plastic viscosity ηpl for the w/c 0.45, 0.50 and 0.55 mortar mix.

Rheology

1

10

100

1000

10000

L10

L15

L25

L40

L180

L180

XL1

90 MF

Thix

otro

py [P

a/s]

w/c 0,45w/c 0,50w/c 0,55

Figure 21. Thixotropy for the w/c 0.45, 0.50 and 0.55 mortar mix.

Rheology

100

150

200

250

300

350

400

L10

L15

L25

L40

L180

L180

XL1

90 MF

Slu

mp

flow

D [m

m]

w/c 0,45w/c 0,50w/c 0,55

Figure 22. Slump flow diameter D for the w/c 0.45, 0.50 and 0.55 mortar mix.

4.3 Correlation A correlation analyze was performed with the software tool AXUM 6.0. The results are listed in Table 3, The correlation coefficient is represented by the Pearson product moment corre-lation r, calculated as:

( ) ( )∑∑∑∑∑∑∑

−⋅⋅−⋅

⋅−⋅⋅=

2222 )()(

)(

yynxxn

yxyxnr [-] (Eq.9)

x and y are the variables and n is the number of pairs that are to be compared. The greater value of r the stronger correlation, as follows:

0.8 - 1.0 very strong 0.6 - 0.8 strong 0.4 - 0.6 moderate < 0.4 weak


10

There are also negative relations, but the important quality of correlation coefficients is not their sign, but their absolute value. Table 3. Correlation matrix within geometry and between geometry and rheology. Results with "very strong" (>0.8) correlation are typed bold and the strong (0.6 - 0.8) are underlined. Asize ASEM BETN BETH2O Blaine F-shape F-circle Comp. Ep βp Asize 1.00 ASEM 0.84 1.00 BETN 0.49 0.21 1.00 BETH2O 0.32 0.08 0.98 1.00 Blaine 0.94 0.77 0.47 0.30 1.00 F-shape 0.23 -0.12 0.62 0.58 0.43 1.00 F-circle 0.25 -0.22 0.68 0.65 0.34 0.87 1.00 Comp. 0.49 0.02 0.79 0.72 0.54 0.85 0.95 1.00 Ep 0.88 0.60 0.76 0.62 0.90 0.57 0.54 0.75 1.00 βp 0.44 0.35 0.73 0.75 0.54 0.47 0.41 0.45 0.61 1.00 Thixotr. 0.13 0.00 0.81 0.89 0.15 0.41 0.49 0.47 0.38 0.86 σ0 0.35 0.07 0.86 0.84 0.42 0.54 0.66 0.70 0.67 0.76 ηpl 0.34 0.17 0.80 0.82 0.40 0.43 0.55 0.55 0.56 0.90 D -0.26 -0.19 -0.70 -0.75 -0.33 -0.38 -0.44 -0.40 -0.42 -0.94 Thixotr. 0.08 -0.02 0.65 0.71 0.18 0.33 0.43 0.37 0.34 0.83 σ0 0.37 0.08 0.95 0.93 0.41 0.60 0.69 0.75 0.71 0.75 ηpl 0.33 0.19 0.81 0.83 0.41 0.45 0.50 0.52 0.58 0.92 D -0.30 -0.19 -0.80 -0.84 -0.37 -0.47 -0.50 -0.50 -0.52 -0.94 Thixotr. 0.35 0.30 0.52 0.55 0.41 0.12 0.20 0.22 0.43 0.88 σ0 0.42 0.24 0.87 0.84 0.47 0.41 0.49 0.58 0.69 0.82 ηpl 0.46 0.37 0.62 0.64 0.48 0.18 0.28 0.32 0.52 0.92 D -0.41 -0.37 -0.74 -0.75 -0.44 -0.23 -0.27 -0.34 -0.56 -0.93 The analyze shows on strong correlation between retained water ratio βp and mortar slump flow D (see Figure 23). This can be explained by the fact that the methods are based on the same methodology and equipment. The correlations between the roundness factors (F-shape, F-circle and Compactness) are good, as expected.

Correlation

0,5

0,6

0,7

0,8

0,9

1,0

100 150 200 250 300 350 400Slump D [mm]

Ret

aine

d w

ater

Ep

[-]

w/c 0,45w/c 0,50w/c 0,55

Figure 23. Correlation between retained water ratio βp and mortar slump flow D for the w/c 0.45, 0.50 and 0.55 mix.

But it ought to be pointed out that F-shape is more a measure of shape, F-circle more a measure of texture and Compactness is somewhere between. This is exemplified in Figure 24 with four different geometrical shapes (ellipse, rectangle, triangle and hexagon) and their impact on the roundness factor by changed size ratio (d/D).

w/c

0.5

5 w

/c 0

.50

w/c

0.4

5


11

The quantification by image analysis was problematic. The method is strongly dependent of a large numbers of factors, such as: sample preparing, microscope performance, photo quality (sharpness), resolution, number of particles, analyzing software, threshold technique, etc. The method might be better if a larger part of the procedure were automatized and standardized.

0,0

0,2

0,4

0,6

0,8

1,0

1:1 1:2 1:3 1:4d/D

Fsha

pe [-

]

EllipseRectangleTriangleHexagone

0,0

0,2

0,4

0,6

0,8

1,0

1:1 1:2 1:3 1:4d/D

Fcirc

le [-

]


0,0

0,2

0,4

0,6

0,8

1,0

1:1 1:2 1:3 1:4d/D

Com

pact

ness

[-]


a)

d

D

b)

d

D

c)

d

D

d)

d

D

Figure 24. The roundness factors F-shape, F-circle and Compactness and their geometrical dependency exemplified by a) ellipse, b) rectangle, c) triangle and d) hexagon. Another very strong correlation was found between the calculated specific area (from size distribution) and the measured with Blaine (see Figure 25). This indicates that Blaine mainly is a measure of size and size distribution, based on theoretical spherical particles. Thus, it should be noted that great variation was noted between values measured on the same specimen, especially between those performed by different persons (and equipment).

Correlation

0

100

200

300

400

500

0 200 400 600 800Sarea (Blaine) [m2/kg]

Asi

ze (s

ize

ditr.

) [m

2 /kg

]

Figure 25. The linear relationship between the calculated area and Blaine.

The highest single correlation is to be found between the BET (N) and BET (H2O) measures (specific area). Although the measured value is not equal, the linear relationship is good (see Figure 26). One has to be aware of the differences in size and polarity of the water and nitrogen molecule. This test has also been made on fines (<1.0mm) sieved from different aggregates, with the same good results. For binders, type cement, the BET (H2O) method has not supplied reliable measures, and therefore not recommended.

Correlation

0

2000

4000

6000

8000

0 2000 4000 6000 8000 10000 12000BET (Nitrogen) [m2/kg]

BE

T (H

2O) [

m2 /k

g]

Figure 26. The linear correlation between the BET (N) and BET (H2O) measures.

Both BET (N) and BET (H2O) show strong correlation to the rheological measures. Why the BET-method is better connected with rheological properties than the other surface area quantifications could be explained by importance of shape, surface texture and porosity. If


12

the spread flow diameter D is recalculated into flow area (see Eq.4), the correlation with BET gets even better. This is probably due to an exponential relation.

5. CONCLUSIONS Due to the importance of the large surface area, quantification of fillers and the fine part of aggregates properties is essential for the ability to control physical properties, such as workability in the production of self-compacting concrete. This study has proven the importance of particle size and geometry (shape, surface porosity and texture). The BET (H2O), a simple characterisation method for fillers and fines for self-compacting concrete, has been introduced. The BET (H2O) has proven high potential. It is simple, low cost, and has provided stable and reliable values. It has also shown on a strong correlation with rheological properties and with the traditional BET (Nitrogen). Other quantification methods have been carried out, but the specific area, quantified with BET-technique, has shown to be the single most important measure concerning the mortar rheology.

REFERENCES 1. Barnes H. A., Hutton J. F., Walters K., (1989), "An introduction to rheology", Elsevier,

Amsterdam. 2. Bartos P., (1992), "Fresh concrete: Properties and tests", Elsevier, Amsterdam. 3. Tattersall G. H., (1991), "Workability and quality control of concrete", E&FN Spon,

London. 4. Billberg P., (1999), "Self-compacting concrete for civil engineering structures - the

Swedish experience", Swedish Cement and Concrete Research Institute, CBI report 2:99, Stockholm.

5. Persson A-L., (1996), "Image analysis of fine aggregates: Characterisation of shape and grain-size parameters", Licentiate thesis, Royal Institute of Technology, Stockholm.

6. Fuglsang Nielsen L., (1993) "Moisture sorption in porous material - A modified BET-description", 3rd Symposium on building physics in norden contries, pp 719-724, Vol 2, Copenhagen.

7. Ahlgren L., (1972), "Moisture fixation in porous building materials", Devision of building technology the Lund institute of technology, Report 36, Sweden (in Swedish)

8. Brunauer S., Emmet H. P., Teller., (1938) "Adsorption of gasses in multimolecular layers", American Chemical Society.

9. Okamura H., Ozawa K., (1995), "Mix design for self-compacting concrete", Concrete Library of the JSCE, No 25, pp 107-120, June, (Translation from Proc. of ISCE, No 496/V-24, 1994.8).

10. Domone P. L., Chai H., (1997), "Testing of binders for high performance concrete", Cement and Concrete Research, Vol 27 (8), pp 1141-1147.

11. Personal communication with Dr O. Wallevik (Head of Concrete Division at the Icelandic Building Research Institute), 2001.

P a p e r I I I - S C C f l o w a b i l i t y : E f f e c t o f c h a n g e s i n p a r t i c l e s u r f a c e a r e a , a n d h o w t o c o m p e n s a t e f o r t h i s

PAPER III: SCC FLOWABILITY: EFFECT OF CHANGES IN PARTICLE SURFACE AREA, AND HOW TO COMPENSATE FOR THIS

Esping O., “SCC flowability: Effect of changes in particle surface area, and how to compensate for this”, Accepted for publication at 5th International RILEM Symposium on Self-Compacting Concrete, 3-5 September, Ghent, Belgium, 2007.


1

SCC FLOWABILITY: EFFECT OF CHANGES IN PARTICLE SURFACE AREA, AND HOW TO COMPENSATE FOR THIS

Oskar Esping

Dep. of Civil and Environmental Engineering, Chalmers University of Technology, Sweden.

Abstract Due to the large particle surface area, quantification of fillers and the fine part of aggregates is essential for the ability to control physical properties, such as workability of self-compacting concrete. In the presented work, the effect of the specific surface area of gravel and limestone filler on the rheology of self-compacting concrete has been evaluated. The surface area was measured with a simplified BET-method, using water as adsorbate. A model is presented, where a change in specific area by BET(H2O) is translated to a change in water demand for the concrete mix. The model is based on an assumption that 30 full molecular layers of water, covering the particle surface is required to decrease interactions and provide lubrication, sufficient to create flowability. The results indicated that a change in BET(H2O)-area of 1000 m2/kg corresponds to approximately 0.8% moisture content of the gravel. E.g. a SCC with 1000 kg gravel (0-8 mm) that increases by 1000 m2/kg in area will need 8.5 litre extra water, in order to retain its consistency. In cases where traditional methods for geometrical characterization (size distribution, water absorption, modulus of fineness, etc.) is kept within acceptance a normal gravel for concrete can vary by up to 7000 m2/kg in surface area.

1. INTRODUCTION For concrete, particle concentration and size distribution strongly influences the workability since its matrix consist a suspension of smaller particles filling the hollow spaces between the coarser. To decrease the internal friction and generate mobility, the particle surfaces are to be covered with water and the voids in the system to be filled. More water will reduce the volume fraction of particles and thereby eases the matrix motion. Consequently, when the contact zone between the particles decreases there will be a significant reduction of plastic viscosity and yield stress. Other important properties affecting the workability are the particle shape, porosity and surface texture. Since the ratio of surface area to volume increases exponentially by particle irregularity (shape, surface texture and porosity) and decreased size, quantification of geometrical properties of fillers and the fine part of the aggregates is essential for the ability to control physical properties, such as workability. A common problem at the job site or plant is that fresh concrete, of a given mix design, can vary in workability from batch to batch and can have a large variation in consistency over time (i.e. consistency loss). This is especially problematic for the self-compacting concrete (SCC), due to its great complexity and because of its sensitivity to variation in the constituents’ properties (gravel moisture content, aggregate grading, etc.). Contrary to ordinary concrete with insufficient flowability on site, SCC is not to be adjusted with extra water or plasticizer, or given extra energy by vibrating.


2

As the aggregate represents 60-80% of the total volume of concrete, it is obvious that its influence on the fresh and hardened concrete is substantial. It is well known that two aggregates, with apparently similar grading, can provide significant differences in concrete workability (see e.g. [6] and [9]). These differences can be referred to the aggregate fineness, surface texture, shape, porosity, mineralogy, etc. Properties of fine particles (<63 µm) and their influence are often neglected because traditional standard methods for characterizing aggregates are deficient or too complicated. Even though the content by volume or weight is often small, the number of particles and their total surface area are large. A content of up to 3 vol-% of particles finer than 63 µm (EN 933-1, or 75 µm according to ASTM C33) is normally considered harmful, but these particles and their surface properties have a considerable influence on the quality of the concrete, either fresh or in hardened state [8]. The gravel fineness is traditionally quantified by modulus of fineness from the size distribution, a method underestimating the contribution of the fines. The cement and filler particles fineness is traditionally quantified with Blain measure. Due to the methodology where the specific surface area is determined from air permeability, based on packed spherical particles, information of the shape, texture and surface porosity is neglected. Other possibly more correct methods, such as the BET with Nitrogen gas and image analysis is more seldom used due to its complexity and costs. The aim of this experimental study was to investigate the possibility to detect variations in raw material properties that are deleterious for the fresh concrete quality, and how to compensate for these.

2. TEST METHODS

2.1 BET(H2O) By using BET theory, the specific surface area (S) can be determined by deducing the area from the adsorbed volume of gas molecules required to cover and form a monolayer on the surface of the sample [3]. The filler and fine part of aggregate specific surface area was determined gravimetrically by the BET multipoint method, using water (moisture) as adsorbate at different relative humidity (RH). The test was performed at 20ºC with climate boxes with 11.3% and 33.1% RH, achieved with LiCl and MgCl2 saturated salt solutions (see Figure 1a). The sample size was approximately 100 g, the weight (m) was measured by a precision balance with 0.001 g readability and the sample was dried by oven at 105ºC. The samples were conditioned in each climate for 4 days.

Fan

Saturated salt solution

Box with tight lid

Net

Sample

y [k

g/kg

]

a

RH [-]

b

0.113 0.331

Figure 1: Schematic illustration of climate box, and the BET-plot with adsorption isotherm as a linear function of RH (from [6]).

By plotting the adsorption isotherm (y, see Eq.1) as a function of the RH, the variables a (interception) and b (slope) were determined by the linear relationship in a BET-plot (see Figure 1). The specific surface area (SBETH2O) was then calculated according to Eq.2 [1]-[6].


3

)RH1()mm(mRH

ydry

dry

−⋅−

⋅= [kg/kg] (RH 0;1) (Eq.1)

ba1054.3S

6

O2BETH +⋅

= [m2/kg] (Eq.2)

2.2 Slump flow test A traditional slump flow measurement was carried out by using a slump cone) and measuring the average spread flow diameter D (in two perpendicular directions). The test was performed at 7 and 60 minutes after adding water to cement. No compaction was needed (as in the vertical slump test EN 12350-2 or ASTM C143 standard for concrete).

2.3 Rheology The rheology was evaluated from the measurements, carried out at 7 and 60 minutes after adding mixing water, where the sample was let to stand between the runs. The mortar test measurement was performed using a Bohlin CVO200 rheometer and the concrete a ConTec Visco5, both using rotating coaxial cylinder at a controlled shear rate (γ& ) at 20ºC. The experimental geometry is illustrated in Figure 2. The rheological evaluation of plastic viscosity (ηpl) and yield stress (σ0) was made according to the Bingham model:

pl0 ηγσσ ⋅+= & [Pa] (Eq. 3).

0

2

4

6

8

0 10 20 30 40 50 60Time [s]

She

ar ra

te [1

/s]

No loggingLogging

Ro=145 mmRi=100 mm

H=140 mm

Measuring inner cylinder

Rotating outer cylinder

Concrete sample

Figure 2: The experimental setup, and the schematic illustration of the ConTec Visco5

rheometer for concrete (from [6]).

3. BET(H2O)-MODEL When there is a change in the dry material specific surface area, and as this area has to be covered with water in order to create mobility, the concrete water demand will change. One way to compensate for these variations in filler or fine aggregate, is to consider the specific area by BET(H2O) and translate this to a change in water demand of the concrete mix. With the assumption that 1 litre of water in one molecule layer covers 3 546 000 m2, the change in mixing water (∆W) due to a change in specific surface area (∆SBETH2O) can be calculated as:

mtr6O2H

20BETH m10546.3

nSW ⋅

⋅⋅= ∆∆ [liter/m3] (Eq. 4)

where mmtr is the mass of the dry material of the mix, and nH2O is the number of full water molecular layers covering the particle surfaces required to providing sufficient dispersion for flowability. Based on numerous experiments, a suitable value was found to be nH2O=30 layers. This can be compared to the proposed, in literature, 5-20 molecular layers of water adsorbed at high relative humidity [1][4][7][9] or as much as 60 molecular thickness [5].


4

4. MATERIAL AND MIX DESIGN The mix design and its constituents, as used in this study, comprise typical materials and compositions for self-compacting concrete in Sweden. The properties of the materials are listed in Table 1. Five limestone fillers with different geological origin and five natural 0-8 mm gravel named “Hol” (sample F to J) were used. The “Hol” gravel was collected at five different occasions over a two month period at a concrete production plant. The size distribution and specific surface area by BET(H2O) for the limestone fillers are presented in Figure 3, and for the “Hol” gravel in Figure 4.

Table 1: Properties of dry materials (cement, fillers, aggregates and admixture).

ID Type Name Supplier Density [kg/m3] C CEM II/A-LL 42.5R Byggcement, Skövde Cementa 3 080 F Ground limestone L40, L15, L70, G100 & G200 NordKalk & Karbonater 2 700

A 0-8 Natural aggregate Hol, E and F to J Färdig Betong 2 650 A 8-16 Crushed aggregate Kungälv Färdig Betong 2 700

SP Super plasticizer (PCE) Sikament 56 SIKA 1 100

0

20

40

60

80

100

0.50.250.1250.0630.0320.0160.0080.0040.002Size [mm]

Pas

sing

[%]

L40L15L70G100G200

L15G100

L40(ref)G200

L70

18082308

6148

1894 1822

0

2 000

4 000

6 000

8 000

L40 L15 L70 G100 G200

SBE

TH2O

[m

2 /kg]

Figure 3: Size distribution and spec.surface area by BET(H2O) for the five limestone

fillers.

0

20

40

60

80

100

84210.50.250.1250.063Size [mm]

Pas

sing

[%] F(ref)

GHIJAverage

23412570

2808 30193348

0

1 000

2 000

3 000

4 000

F(ref) G H I J

SBE

TH2O

[m

2 /kg]

Figure 4: Size distribution and specific surface area by BET(H2O) for the five “Hol”

gravel.

The concretes were prepared in batches of 40 litres, mixed in a BHS60 twin-shaft paddle mixer for 4 minutes after water was added to the premixed dry materials. Two self-compacting mixes were investigated (see Table 2): (1) with the five different limestone fillers and (2) with the five 0-8 mm natural gravels named “Hol”.


5

Table 2: Recipe of the w/c 0.55 SCC mixes, in kg/m3.

ID Product name 1. 2. W Mixing water 187 187 C Byggcement 340 340 F L40 220 L40/L15/L70/G100/G200 220

A 0-8 Hol E 873 Hol F to J 873

A 8-16 Kungälv 728 728 SP Sikament 56 4.08 4.08

In order to retain the consistency, relative the mix with the reference material, complimentary mixes were made with extra water (see Table 3) compensating for the difference in specific area by BET(H2O). This was made according to Eq. 4.

Table 3: Changes mixing water (∆mw) due to the differences in specific surface area for the limestone filler and “Hol” gravel.

Limestone filler ID

∆SBETH2O [m2/kg]

∆mw [kg/m3]

“Hol” gravel

ID

∆SBETH2O[m2/kg]

∆mw [kg/m3]

L40 (ref) - - F (ref) - - L15 +499 +0.68 G +229 +1.69 L70 +4 340 +5.89 H +468 +3.45

G100 +86 +0.12 I +679 +5.01 G200 +13 +0.02 J +1007 +7.44

4. RESULTS AND DISCUSSION All test results are represented by a mean value based on two or more tests. The slump flow and rheology (plastic viscosity and yield stress) were measured at two timings (7 and 60 minutes from water addition). The results from the flowability measures for the mix with the five different limestone fillers are presented in Figure 5. Both slump flow and rheological evaluation (yield stress and plastic viscosity), show significant differences between the samples, which is reflected by the differences BET(H2O)-area (see Figure 3). It can be noted, that thus the “L70” filler was the coarsest it generated the largest surface area and in the SCC mix it reduced the flowability most. All mixes with extra water achieved nearly the same consistence as the reference mix with limestone filler “L40”.


6

0

200

400

600

800

L40

(ref)

L15

L15

(+0.

68)

L70

L70

(+5.

89)

G10

0

G10

0 (+

0.12

)

G20

0

G20

0 (+

0.02

)

Spr

ead

[mm

]

7 min 60 min

0

100

200

300

400

L40

(ref

)

L15

L15

(+0.

68)

L70

L70

(+5.

89)

G10

0

G10

0 (+

0.12

)

G20

0

G20

0 (+

0.02

)

Yie

ld s

tress

[Pa] 7 min 60 min

507 Pa

0

20

40

60

L40

(ref)

L15

L15

(+0.

68)

L70

L70

(+5.

89)

G10

0

G10

0 (+

0.12

)

G20

0

G20

0 (+

0.02

)

Vis

cosi

ty [P

a s]

7 min 60 min

Figure 5: Flowability measures for the SCC mix with the limestone fillers, both without

and with the extra water added to compensating the BET-surface.

The results from the flowability measures for the mix with the five “Hol” gravel are presented in Figure 6. Similar to the results for the limestone filler (see Figure 5), the flowability with the “Hol” gravels shows significant differences between the mixes, which can be addressed to the differences in BET(H2O)-area. All mixes with extra water achieved nearly the same consistence as the reference mix with gravel “F”, which indicates that the differences in surface area can be compensated with water according to the proposed BET(H2O)-model, see Eq. 4. It can be noted, that thus the gravel size distributions is similar (see Figure 4), there is up to ~1000 m2/kg difference in BET(H2O)-area. This is ~¼ of the differences for the limestone filler, but then the content of gravel in the mix is ~4 times larger then the limestone filler (see Table 2). In [6] it was found that a natural gravel collected at different spots in a gravel pit, with small differences in grading, can vary in BET(H2O)-area by ~7000 m2/kg.


7

0

200

400

600

800

F (re

f) G

G (+

1.69

) H

H (+

3.45

) I

I (+5

.01) J

J (+

7.44

)

Spr

ead

[mm

]

7 min 60 min

0

50

100

150

F (re

f) G

G (+

1.69

) H

H (+

3.45

) I

I (+5

.01) J

J (+

7.44

)

Yie

ld s

tress

[Pa]

7 min

60 min

0

20

40

60

80

F (re

f) G

G (+

1.69

) H

H (+

3.45

) I

I (+5

.01) J

J (+

7.44

)

Vis

cosi

ty [P

a s]

7 min 60 min

Figure 6: Flowability measures for the SCC mix with the “Hol” gravels, both without and

with the extra water added to compensating the BET-surface.

Question arises whether this extra water added for constant flowability, based on an increased BET-area, is fixed to the aggregates or fillers and does not contributes to the concrete hydration. Is it possible to add this extra water without affecting the w/c, in the same approach made when using the particle water absorption (EN 1097-6 standard)? To compensate for an increased BET-area by adding extra superplasticizer (SP) to the mix will not have the same effect as extra water, as the SP (PCE based) normally only reduces the yield stress and not the viscosity [10]. In Figure 7 the results of the measured compressive strength for the SCC mix with the different limestone fillers, both without and with the extra water added to compensating the BET-area, is given. The result shows that the strength is promoted by increased area, whereas with additional water for constant flowability the effect is in reverse ratio. This indicates that approx. 50% of this water can be added without affecting the w/c.

45

46

47

48

L40 L15 L70 G100 G200

Com

p. S

treng

ht, 2

8d [M

Pa]

Additional waterNo compensation

Figure 7: Compressive strength (28 days) for the SCC mix with the limestone fillers, both

without and with the extra water added to compensating the BET-area


8

6. CONCLUSIONS An experimental investigation of the effect of the specific surface area of different gravels and limestone fillers and their influence on the rheology of SCC was carried out. The area was measured with a simplified BET-method, using water (as moisture) as adsorbate. A model was presented, where the effect of change in specific area by BET(H2O) is translated to a change in water demand for constant flowability of the concrete mix. The following observations and conclusions were made:

Due to the importance of their large surface area, quantification of fillers and the fine part of aggregates’ properties is essential for the ability to control physical properties, such as workability, in the production of SCC.

For better quality control and prediction of fresh concrete behaviour, the specific surface area by BET is proposed as a potential means of geometrical characterization for fillers and aggregates containing fines.

A coarser material can provide larger surface area than a finer, due to differences in surface texture and porosity, and thereby negatively influence the flowability.

Addressing to the changes in BET(H2O)-area, a gravel with apparently consistent grading over time can provide significant differences to the concrete workability.

The proposed model showed that it was possible to calculate the extra water needed to compensate for changes in filler and fine aggregates BET(H2O)-area, in order to control and produce SCC with small variations in flowability.

Approximately 30 molecular layers of water, covering the particle surface, are required to decrease interactions and provide lubrication, sufficient to create flowability.

It was shown that the free water content, available to provide flowability, in 1 m3 SCC mix can vary by several litres due to differences in apparently equivalent fine aggregates. A change in BET(H2O)-area of 1000 m2/kg corresponds to approximately 0.8% moisture content of the gravel. E.g. a SCC with 1000 kg gravel (0-8 mm) that increases by 1 m2/g in area will need 8.5 litre extra water, in order to retain its consistency.

ACKNOWLEDGEMENTS

Financial support from Färdig Betong and Thomas Concrete Group is greatly appreciated.

REFERENCES [1] Ahlgren L. (1972): “Moisture fixation in porous building materials”, Division of building

technology the Lund institute of technology, Report 36, Sweden (in Swedish). [2] Allen T. (1997): “Particle size measurement – Surface area and pore size determination”,

London. [3] Brunauer S., Emmet H.P., Teller E. (1938): “Adsorption of gasses in multimolecular layers”,

American Chemical Society. [4] Brunauer S. (1944): “The adsorption of gases and vapours”, vol 1, Oxford. [5] Eirich F. R. (1960): “Rheology – Theory and applications”, vol 3, New York. [6] Esping, O. (2004): “Rheology of cementitious materials: effects of geometrical properties of

filler and fine aggregate”, Chalmers University of Technology, Göteborg. [7] Gregg S. J., Sing K. S. W. (1967): “Adsorption, surface area and porosity”, New York. [8] Neville A. M. (1963): “Properties of concrete”, London. [9] Powers T. C. (1968): “The properties of fresh concrete”, New York. [10] Wallevik O. (2003): “Rheology – A scientific approach to develop self-compacting concrete”,

3rd International Symposium on SCC, Reykjavik.

P a p e r I V - I n v e s t i g a t i o n o f a u t o g e n o u s d e f o r m a t i o n i n S C C

PAPER IV: INVESTIGATION OF AUTOGENOUS DEFORMATION IN SCC

Esping O., “Investigation of autogenous deformation in self-compacting concrete”, International RILEM Conference on Volume Changes of Hardening Concrete, 20-23 August, pp 273-282, Lyngby, Denmark, 2006.


1

INVESTIGATION OF AUTOGENOUS DEFORMATION IN SELF-COMPACTING CONCRETE

O. Esping

Chalmers University of Technology, Göteborg, Sweden

Abstract

For modern concretes, such as high-strength and self-compacting concrete (SCC), increased amount of fines and binder content are a common explanation for the large autogenous shrinkage and early-age cracking. In order to investigate the influence of fineness of concrete constituents on autogenous shrinkage, tests with SCC, incorporating limestone fillers with different specific surface area (by BET), were conducted. A dilatometer was used, generating measures of linear autogenous (sealed) deformation of the concrete cast in vapour-proof flexible tube mould. Temperature and pore pressure were simultaneously measured with the deformation. Tests were conducted on mixes both with and without cement, i.e. with and without chemical shrinkage. The results indicated that increased surface area generated an increased magnitude and rate of autogenous shrinkage, without influencing the times to initial and final set. The mixes without cement appeared to generate a large “autogenous” shrinkage, without the presence of chemical reaction of cement and water (hydration). The shrinkage was approximately 0.8 mm/m at 24 hours, after an initial swelling up to 4 hours. This is the same magnitude as for the mixes with cement, but then these mixes generated no swelling and almost all shrinkage before final set (at ~11.5 hours from mixing). Increased particle surface area also decreased the rate and magnitude of evaporation, and consequently reduced the plastic crack tendency, despite increased autogenous shrinkage. Adding extra water to the mix, compensating for the loss of flowability due to increased particle surface area, increased the crack tendency significantly.

INTRODUCTION

Early-age shrinkage and cracking have become a recurrent problem in concrete construction. Conditions such as reduced maximum aggregate size, increased amount of fines, presence of retarding admixtures, increased binder content, and deficient covering and curing all contribute to this problem.

At early age, when the cement paste is young and has poorly developed mechanical properties, autogenous and drying shrinkage, both incorporated in the plastic shrinkage, are the two main driving forces for cracking. When the concrete dries out due to evaporation, the loss of water from the paste generates negative capillary pressure, causing the paste to contract (see Wittmann [1]), which in turn can lead to cracks. These contracting capillary forces are in reverse ratio to the meniscus radius, and hence the capillary tension stresses increase with decreasing pore sizes and interparticle spaces. Care has to be taken to protect the surface against drying, although experience in the use of concretes with low w/b has revealed that severe cracking may occur in spite of proper protection (curing membrane,


2

etc.); see Bjøntegaard et al. [2]. For a concrete where evaporation is prevented, a contracting negative capillary pressure will also develop, but only once the hydration commences and the concrete sets (see Holt [3]). As long as the concrete is fluid, autogenous shrinkage is considered to be equal to the chemical shrinkage; but once the self-supporting skeleton starts to form, the autogenous will diverge from the chemical (e.g. Hammer [4], Jensen & Hansen [5]). The pattern of autogenous deformation, exemplified in Figure 1, comprises three distinct stages which can be defined as (I) plastic, (II) semiplastic and (III) rigid, separated by the time to initial and final set; see Esping & Löfgren [6]. It has been suggested (e.g. by Barcelo [7]) that, when measuring the autogenous deformation, the setting will be manifested as a change of the slope of the deformation. Furthermore, once the internal voids are created, it leads to the development of a capillary pore underpressure in the skeletal structure (see Figure 1), which causes an external deformation of the hardening concrete (see Radocea [8]).

Water

Cement grains

Solid grains (aggregate/filler)

Plastic

Semiplastic

Rigid

Solid hydrates

Gas voids/pores


0 ~6 ~18

Def

orm

atio

n, P

ress

ure

& Te

mp

Plastic Rigid

~24

Semiplastic



Initial set

Final set

Autogenous deformation (measured)



Figure 1: Schematic representation of the early-age linear autogenous deformation of

concrete and the corresponding development of temperature, capillary pore pressure and microstructure (based on measures of SCC with w/c 0.45, from Esping & Löfgren [6]).

It can be argued that the deformation occurring when the concrete is plastic mainly causes setting and has little consequence for the risk of cracking, while the shrinkage taking place when the concrete is semiplastic is considerably more detrimental as the concrete at this stage has poorly developed tensile strain capacity; see Kasai et al. [9].

This paper aims at demonstrating how changes in particle specific surface area affect the early-age deformations, primarily the autogenous.

In this study, the change in area is represented by limestone fillers with different specific surface area quantified with the BET(H2O) method. Furthermore, early age is considered as the time from mixing up to 24 hours after mixing. The term ‘deformation’ is preferably used, since expansion also may occur. Autogenous deformation is considered to be the bulk deformation under sealed conditions, and is referred to the full period of hardening from the time when water was added to the mix. Linear displacement was used for measuring autogenous deformation. No corrections were made for the differences between linear and volumetric changes during the fluid state (before initial set). Moreover, ideal isothermal conditions are attributed to autogenous deformation, and this was not strictly the case. No compensation for thermal changes was made.


3

TEST METHODS

Autogenous shrinkage test method In this study, autogenous linear deformation was monitored by a concrete digital

dilatometer (CDD), developed in order to start measurement before setting, when concrete is fresh. The method is a modification of the CT1 digital dilatometer for pastes and mortars; see Jensen & Hansen [10]. The CDD sample consists of a concrete specimen, cast in a steel coil-reinforced vapour-proof flexible polyurethane tube with inner diameter of 82 mm and specimen length ~400 mm, sealed with hose clamps and O-ring-equipped PVC end-caps. The unrestrained linear deformation was recorded with a digital gauge (0.003 mm accuracy). The test, containing three complete CDD setups, was performed in a thermostable room at 20ºC, where the recording was started at 30 minutes from water addition. In order to elucidate the mechanism of autogenous deformation, and as it is highest up to an age of 12 hours, it is necessary to measure the deformation immediately after mixing (see Aïtcin [11]). The equipment and apparatus for a typical test are shown in Figure 2. Due to greater stiffness in the radial than in the longitudinal direction of the mould, when the concrete is in the fluid state, the flexible mould transforms most of the volumetric deformation into a linear deformation. As the concrete undergoes transition from a fluid to a rigid state, the deformation becomes isotropic. The ratio between linear and volumetric deformation is shown in Figure 2. No correction for this was made, as the setting point is not a well-defined physical state but rather a continuous transformation from a liquid to a solid state (see Jensen & Hansen [5]). The experiments were supplemented with measurements of temperature and pore pressure placed in the centre of the core. The pressure transducers were connected to a de-aired water-filled system with a needle (internal diameter of 0.4 mm and 50 mm length).

0 001

Specimen length 400 mmDigital gauge (1/1000 mm)

Adjustable fixture Mould diameter 82 mm

y = 0,796xR2 = 0,9999y = x

y = 0,338x

-5%

-4%

-3%

-2%

-1%

0%

-5%-4%-3%-2%-1%0%Volume change

Leng

th c

hang

e (d

efor

mat

ion)

3-D tube (calculated)Fluid (measured)1-D tube (calculated)

Figure 2: To the left, illustration of the Concrete Digital Dilatometer (CDD) for linear

autogenous deformation measurement (from Esping & Löfgren [6]). To the right, the ratio between linear and volumetric deformation for the mould. Filled with a Newtonian fluid the ratio follows the measured ratio, but once rigid the ratio will follow the calculated 3-D deformation (from [12]).

Plastic shrinkage cracking test method To evaluate the plastic crack tendency of concrete at early ages exposed to drying, a

modified ring-test method, originally developed by Johansen & Dahl [13], was used. The crack tendency was represented by a mean value of total crack area (crack length × crack width) on the concrete surface of each of three specimens. The setup was equipped with measurements of evaporation (by a scale), temperature and pore pressure (at depth 20 and 60 mm). Further description of the method can be found in Esping & Löfgren [6] and [14].


4

MATERIALS AND MIX DESIGN The mix design and constituents used comprise typical materials and composition for

self-compacting concrete (SCC) in Sweden, listed in Table 1. Previous tests have shown that at a w/c of about 0.55 the plastic crack tendency is at a minimum point, and that the interplay between evaporation and autogenous deformation is at equilibrium (Esping & Löfgren [6]). The limestone fillers’ size distribution and specific surface area by BET(H2O) are given in Figure 3. It can be noted that, despite L70 being the coarsest limestone filler, it showed the largest surface area. Normally, increased fineness generates a larger surface, but L70 is probably more porous due to differences in geological origin and thereby larger in area.

The concretes were prepared in batches of approximately 40 litres, and mixed in a twin-shaft paddle mixer for 4 minutes after water was added to the premixed dry materials. The admixtures were added directly after the water.

Table 1: Properties of constituents and SCC composition (w/c 0.55).

ID Type Product name Supplier Density Mass content C CEM II/A-LL 42.5R Byggcement, Skövde Cementa 3 080 kg/dm3 340 kg/m3 W Tap water - - 1 000 187

F Ground limestone L40, L15 & L70 G100 & G200

NordKalk SMA Karbonater ~2 700 160

A 0-8 Natural aggregate Hol NCC Ballast 2 650 875 A 8-16 Crushed aggregate Kungälv NCC Ballast 2 700 730

SP Superplasticizer (PCE) Sikament 56 SIKA 1 100 4.08

0

20

40

60

80

100

0.50.250.1250.0630.0310.0160.0080.0040.002Size [mm]

Pass

ing

[%

L40L15L70G100G200

18102310

6150

1890 1820

0

2 000

4 000

6 000

8 000

L40 L15 L70 G100 G200

Spec

ific

surfa

ce a

rea,

BET

(H2 O

) [m

2 /kg]

Figure 3: The five limestone fillers. To the left is the size distribution, and to the right the

specific surface area measured with the BET(H2O) method.

Mixes were also made without presence of cement, in order to evaluate the effect (if any) of particle surface area on the autogenous deformation without influence of chemical shrinkage. The cement was replaced with an equal volume of limestone filler, i.e. the filler content was increased from 160 to 458 kg/m3. Concrete flowability strongly depends on the size of the particle surface area which has to be covered with water to create mobility. When the specific surface area increases, so will the water demand of the mix for the same consistency. Using L40 as reference, extra water was added to the mixes, compensating for the differences in surface area by BET(H2O) and regaining constant flowability. A model was used where 30 molecular layers of water covering the particle surface are required to decrease interactions and provide lubrication sufficient to create flowability; see Esping [15].


5

In Table 2 the changes in recipe, due to differences in specific area of limestone filler, are given. This was only applied to the plastic shrinkage ring test, as the effect of water content on autogenous deformation has previously been demonstrated by Esping & Löfgren [12].

Table 2: Mixing water with compensations for the differences in specific surface area of limestone fillers. L40 is used as reference (smallest surface area).

ID L40 (ref) L15 L70 G100 G200 W 187 liter/m3 187.68 (+0.68) 192.89 (+5.89) 187.12 (+0.12) 187.02 (+0.02)

RESULTS

Autogenous deformation (with cement) The results of the autogenous deformation measurements for the SCC incorporating the

five limestone fillers with different specific surface area are presented in Figure 4(a), while Figure 4(b) and Figure 5(a) show the rate of the deformation. As can be observed, increased surface area (by BET method) increased both the magnitude and rate of shrinkage, primarily in the plastic region. The time to initial and final set, evaluated from the pattern of autogenous deformation when the rate is close to zero, was 6.1 respective 11.6 hours from mix; see Figure 5(b). The differences in surface area had almost no effect on times to set (shorter than ±0.1 hours). This is confirmed by the measured temperature development; see Figure 7(a-e).

Previous tests show that increased water content decreases the rate and magnitude of autogenous deformation, primarily as plastic, without affecting the time to set, which is equivalent to the effect that decreased surface area has (see Esping & Löfgren [12]).

-1200

-900

-600

-300

0


Def

orm

atio

n [1

0-6m

/m]

L40

L15

L70

G200G100

-400

-300

-200

-100

0

100


Rat

e of

def

orm

atio

n [1

0-6/h

]

L40

L15L70

G200G100

(a) (b)

Figure 4: SCC w/c 0.55 containing limestone filler with different specific area: (a) the autogenous deformation and (b) the rate of deformation.


6

-400

-300

-200

-100

0

L40

L15

L70

G10

0

G20

0

Rat

e of

def

orm

atio

n [1

0-6/h

our]

PlasticSemiplastic

0

5

10

15

L40

L15

L70

G10

0

G20

0

Tim

e to

set

(fro

m m

ix) [

hour

]

Initial setFinal set

(a) (b)

Figure 5: Evaluated values from autogenous deformation in the plastic and semiplastic period: (a) the rate of autogenous deformation and (b) the rate of deformation.

Furthermore, Figure 6(a) shows the development of pore pressure, and Figure 6(b) the evaluated maximum rate of pore pressure which occurred at approximately 7-8 hours from mixing. The time to and magnitude of maximum underpressure could not be evaluated due to the loss of pressure at random times. This effect is usually referred to as the breakthrough pressure; see Wittman [1].

-100

-80

-60

-40

-20

0


Por

e pr

essu

re [k

Pa]

L40

L15

L70

G200

G100

-60

-40

-20

0L40 L15 L70 G100 G200

dP/d

t [kP

a/ho

ur]

(a) (b)

Figure 6: SCC w/c 0.55 containing limestone filler with different specific area: (a) the development of capillary pore pressure during the first 12 hours from mix, and (b) the maximum rate of pore pressure (at approximately 7-8 hours from mix).

Finally, the relationship between the development of autogenous deformation, temperature, and pore pressure for the concretes with the five limestone fillers is shown in Figure 7. As can be seen, as long as the concrete is plastic the deformation develops rapidly with an almost linear relationship, while during this period the temperature and capillary pore pressure undergo only small changes. However, at one point (at ~5 hours) the rate of deformation is slowed down, indicating ‘setting’ of the matrix. At the following knee point (at ~6 hours), indicating the initial set, both the capillary pore pressure and the temperature reach an accelerating phase, which indicates that the cement hydration accelerates (see Esping & Löfgren [6]). The final set (at ~12 hours) is manifested by a plateau in deformation, slightly ahead of the temperature peak.


7


Max

Min

Pore pressure

Temp.


L40

0 5 10 15 20


Pore pressure

Temp.


L15


Pore pressure

Temp.


L70


Max

Min

Pore pressure

Temp.


G100

0 5 10 15 20


Pore pressure

Temp.


G200

Figure 7: Development of temperature, pore pressure, and autogenous deformation (normalized against maximum values), for SCC with respective filler.

“Autogenous” deformation (without cement) The driving force of autogenous deformation is generally considered to be the hydration

process (generating chemical shrinkage); hence the term “autogenous” in the case of mixes without cement is incorrect. The results of the “autogenous” deformation measurements for the mixes without cement (same composition as SCC w/c 0.55, but cement is replaced with equal-volume limestone filler) are presented in Figure 8. Even though no cement is present, the mixes generated an “autogenous” shrinkage of approximately 0.8 mm/m at 24 hours, counted after an initial swelling up to 4 hours. This is the same magnitude as for the mixes with cement, but those generated no swelling and almost all shrinkage before final set. The results indicate that larger surface area increases the deformation (both swelling and shrinkage). The expansion can be explained by water being absorbed by the filler (and aggregate) and by the disjoining pressure, i.e. the adsorption of water molecules in locations where the distance between two surfaces is restricted, inducing pressure and expansion (see Nawa & Horita [16]). Once lack of water occurs, this causes a restraining matrix of particles, and there will be pores with meniscus-generated capillary tension and thereby a contraction (see Hammer [17]).


8

-800

-400

0

400

800


Def

orm

atio

n [1

0-6m

/m] L70

L40G100G200

L15

-200

0

200

400

600


Rat

e of

def

orm

atio

n [1

0-6/

h]

L70 L40L15G100G200

(a) (b)

Figure 8: The mix without cement (same composition as SCC w/c 0.55, but cement is replaced with equal-volume limestone filler): (a) the “autogenous” deformation and (b) the rate of deformation.

Plastic shrinkage cracking A representative measure showing the relationship between the development of strain,

temperature, and pore pressure (at 20 and 60 mm depth) for one of the concretes exposed to drying is shown in Figure 9(a). As can be seen, the pore pressure develops more rapidly in comparison to sealed samples (autogenous). Here the maximum underpressure is reached at 5-6 hours from mix, with approximately 30 minutes between the two depths, whereas the development of temperature is similar to the sealed samples. The measured strain indicated that cracking was initiated at 3-4 hours from mixing. The average crack area (from three samples) is presented in Figure 9(b), where it can be seen that the crack tendency is lower for concrete incorporating filler with a high specific surface area.


Max

MinInitial crack

Temperature

Strain


Evaporation

0

20

40

60

80

L40(ref) L15 L70 G100 G200

Aver

age

crac

k ar

ea [m

m2 ] No compensation

Extra water

(a) (b)

Figure 9: In (a) is the development of pore pressure, evaporation, temperature and strain (normalized against maximum values), exemplified by the concrete with limestone filler L40. In (b) is the crack tendency for the respective filler without, and with, extra water compensating for the differences in surface area.

When adding extra water to compensate for the loss in flowability, the crack tendency increased significantly. The differences in crack tendency are most likely a consequence mainly of evaporation, which is verified by the measured evaporation in Figure 10(a) and its


9

initial rate (up to 4 hours from mix, before initial setting) in Figure 10(b). A large particle surface area lowers the evaporation, whereas extra water increases the evaporation.

0

1

2

3


Eva

pora

tion

[kg/

m2 ]

L15L70G100G100(+0.12)G200G200(+0.02)

L70(+5,87)L15(+0.68)

L40(ref)

0,3

0,4

0,5

L40(ref) L15 L70 G100 G200

Rat

e of

eva

pora

tion

[kg/

(m2 ·h

)] No compensation

Extra water

(a) (b)

Figure 10: SCC w/c 0.55 containing limestone filler with different surface area and extra water compensating for this: (a) the measured evaporation and (b) the evaluated initial rate (<4 hours from mix).

CONCLUSIONS An experimental investigation of early-age (<24 hours) deformation was made on a series

of self-compacting concretes with w/c 0.55 containing limestone filler with different specific surface area measured with BET(H2O). Autogenous linear deformations were measured and additional investigation of plastic crack tendency was carried out, both complemented by simultaneous measurements of pore pressure and temperature development.

The results show that: • Increased particle surface area (by BET method) increased the rate and magnitude of

autogenous shrinkage, primarily in the plastic region, without affecting the temperature development and the times to initial and final set.

• The relationship between autogenous deformation, temperature, and pore pressure shows that, as long as the concrete is plastic, the deformation develops rapidly and almost linearly while, during this period, the temperature and capillary pore pressure undergo only small changes. However, at one point (at ~5 hours) the rate of deformation is slowed down, indicating ‘setting’ of the matrix. At the following knee point (at ~6 hours), indicating the initial set, both the capillary pore pressure and the temperature reach an accelerating phase, which indicates that the cement hydration accelerates. The final set (at ~12 hours) is manifested by a plateau in deformation, slightly ahead of the temperature peak.

• The mixes without cement appeared to generate a large “autogenous” shrinkage, despite absence of chemical reaction of cement and water (hydration). The shrinkage was approximately 0.8 mm/m at 24 hours, after an initial swelling up to 4 hours. This is the same magnitude as for the mixes with cement, but then these mixes generated no swelling and almost all shrinkage before final set (at ~11.5 hours from mixing).

• Increased particle surface area decreased the rate and magnitude of evaporation and consequently reduced the plastic crack tendency, despite increased autogenous shrinkage. Adding extra water to the mix, compensating for the loss of flowability due to increased particle surface area, increased the crack tendency significantly.


10

ACKNOWLEDGEMENTS Financial support from AB Färdig Betong is greatly appreciated.

REFERENCES [1] Wittmann, F.H., ‘On the Action of Capillary Pressure in Fresh Concrete’, Cement and Concrete

Research, Vol. 6, pp. 49-56, 1976. [2] Bjøntegaard Ø., Hammer T. A., Sellevold E. J., ‘Cracking in High Performance Concrete before

Setting’, Symposium on High-Performance and Reactive Powder Concretes, Sherbrooke, 1998. [3] Holt, E. E., ‘Early age autogenous shrinkage of concrete’. Technical Research Centre of Finland,

VTT Publication 446, Finland, 2001. [4] Hammer T. A., ‘Test method for linear measurements of autogenous shrinkage before setting’, in

Autogenous shrinkage of concrete (ed. Tazawa E.), E & FN Spon, London, pp. 143-154, 1998. [5] Jensen O. M., Hansen F., ‘Autogenous deformation and RH-change in perspective’, Cement and

Concrete Research, Vol. 31, pp. 1859-1865, 2001. [6] Esping, O., Löfgren, I., ‘Cracking due to plastic and autogenous shrinkage – Investigation of

early age deformation of self-compacting concrete’, Publication 05:11, Chalmers, Sweden, 2005. [7] Barcelo, L., Moranville, M., and Clavaud, B., ‘Autogenous shrinkage of concrete: a balance

between autogenous swelling and self-desiccation’, Cement and Concrete Research, Vol. 35, pp. 177-183, 2005.

[8] Radocea A., ‘A Study on the Mechanisms of Plastic Shrinkage of Cement-Based Materials’, Doctoral dissertation, Chalmers, Sweden, 1992.

[9] Kasai, Y., Yokoyama, K., and Matsui, I., ‘Tensile Properties of Early Age Concrete’, Mechanical Behavior of Materials, Society of Materials Science, Vol. 4, pp. 288-299, Japan, 1972.

[10] Jensen O. M., Hansen F., ‘A dilatometer for measuring autogenous deformation in hardening Portland cement paste’, Materials and Structures, Vol. 28 (181), pp. 406-409, 1995.

[11] Aïctin, P.C., ‘Autogenous shrinkage measurement’, in Autogenous shrinkage of concrete (ed. Tazawa E.), E & FN Spon, pp. 257-268, 1999.

[12] Esping, O., Löfgren, I., ‘Investigation of early age deformation in self-compacting concrete’, Conference advanced cement-based materials, DTU, Denmark, 2005.

[13] Johansen, R., Dahl, P.A., ‘Control of plastic shrinkage of cement’. 18th Conference on Our World in Concrete and Structures, Singapore, 1993.

[14] Löfgren, I., Esping, O., ‘Early age cracking of self-compacting concrete’, RILEM Conference of Volume Changes of Hardening Concrete, Denmark, August 2006.

[15] Esping O., ‘Rheology of cementitious materials – Effects of geometrical properties of filler and fine aggregate’, Licentiate thesis, Publication No. P-04:3, Chalmers, Sweden, 2004.

[16] Nawa T., Horita T., ‘Autogenous shrinkage of high-performance concrete’, Proceedings on microstructure and durability to predict service life of concrete structures, Japan, 2004.

[17] Hammer T. A., ‘Is there a relationship between pore water pressure and autogenous shrinkage. Before and During Setting?’, in Self-desiccation and its importance in concrete technology, Lund University of Technology, TVBM-3104, pp. 27-38, 2002.

P a p e r V - I n v e s t i g a t i o n o f e a r l y a g e d e f o r m a t i o n i n S C C

PAPER V: INVESTIGATION OF EARLY AGE DEFORMATION IN SCC

Esping O., Löfgren I., “Investigation of early age deformation in self-compacting concrete”, 2nd International Symposium on Advances in Concrete Science, 11-15 September, Quebec, Canada, 2006.


1

INVESTIGATION OF EARLY AGE DEFORMATION IN SELF-COMPACTING CONCRETE

Oskar Esping (1) and Ingemar Löfgren (2)

(1) Dept. of Civil and Environmental Engineering, Chalmers University of Technology, Sweden, and AB Färdig Betong.

(2) Thomas Concrete Group AB, Sweden

Abstract

In the presented work, early age (< 24 h) autogenous deformation was measured and crack tendency due to plastic shrinkage was evaluated; see Esping and Löfgren [1]. For the autogenous deformation, a specially developed digital dilatometer was used which generated accurate measurements of the linear displacements of the concrete, which was cast in a vapour-proof flexible tube mould. The plastic shrinkage cracking tendency was evaluated by exposing concrete specimens to early drying conditions while these were restrained by an inner steel ring. A large number of different SCC constituents and mix compositions have been investigated: e.g. w/c-ratio from 0.38 to 0.67, silica fume, and different admixtures. For comparison, tests with standard concrete were also made. The influence of different constituents and mixes on the autogenous deformation and plastic shrinkage crack tendency was observed. The results indicated that a high crack tendency arose when there was: large autogenous shrinkage (silica addition, low w/c, high fineness); high water evaporation (high w/c, low fineness); retardation (retarder or high superplasticizer dosage); low content of coarse aggregate. Minimum crack tendency was found at w/c 0.55.

1. INTRODUCTION

At early age, when the cement paste is young and has poorly developed mechanical properties, autogenous and drying shrinkage – both incorporated in the plastic shrinkage – are the two main driving forces for cracking. Generally, plastic shrinkage is usually defined as the shrinkage of fresh concrete, exposed to drying, that takes place during the time when the concrete is ‘plastic’, the duration is usually short (< 8 to 12 hours) and ends when the concrete has reached its final set. In traditional concretes plastic shrinkage cracking is mainly caused by the loss of water from the fresh concrete, e.g. by evaporation of water, which generates negative capillary pressures; this cause the paste to contract (see Wittmann [2]), which in turn can lead to cracks. These contracting capillary forces are in reverse ratio to the meniscus radius, and hence the capillary tension stresses increase with decreasing interparticle spaces. For a concrete where evaporation is prevented, a negative capillary pressure will also develop, but only once the hydration commences and the concrete sets. To avoid this type of cracks, care has to be taken to protect the surface against drying. However, experience in the use of concretes with low w/b has revealed that severe cracking may occur in spite of proper protection (curing membrane, etc). Conditions such as reduced maximum aggregate size, presence of retarding admixtures, increased binder content, and deficient curing all contribute to this problem. Early cracking is usually observed in the period soon


2

after casting up to 6-8 hours later, depending on the concrete temperature, material composition, weather conditions and the degree of retardation. In this early phase, the rheology of concrete changes dramatically as the concrete sets, i.e. it changes from a liquid to a solid behaviour within some hours. At the same time the tensile strain capacity goes through a minimum. The cause of this change in rheology is the hydration of Portland cement, which is a complex sequence of chemical reactions leading to setting and hardening. Immediately after mixing cement and water, reactions start to occur and these generate an outburst of heat (Stages I in Figure 1). After these initial stages an induction period, or dormant period, is entered (Stage II). During the induction period not much hydration takes place, but this does not mean that the paste is ‘dormant’ with respect to volume changes.

Setting (during Stage III) is defined as the onset of rigidity in fresh concrete, and the

period of fluidity, preceding setting, corresponds to the induction period (Stage II); see Figure 1(a). As long as the concrete is fluid, there will be a linear relationship between the linear shrinkage and the volumetric chemical shrinkage. However, once the self-supporting skeleton starts to form, the chemical shrinkage will mainly result in internal voids and the linear deformation diverges from the chemical shrinkage. It has been suggested (e.g. by Barcelo [4]) that, when measuring the linear deformation, the setting will be manifested as a change of the slope of the deformation; see Figure 1(b). Furthermore, once the internal voids are created it leads to the development of a capillary underpressure in the skeletal structure, see Figure 1(b), which causes an external deformation of the hardening concrete. It can be argued that the deformation occurring when the concrete is plastic may have little consequence for the risk of cracking, while the shrinkage taking place when the concrete is semiplastic is considerably more detrimental as the concrete at this stage has poorly developed mechanical properties (low tensile strength and strain capacity).

I II III IV V

Rat

e of

hea

t evo

lutio

n

Dissolution: ettringite formation

Initial set

Final set

Induction period: increase in Ca2+ concentration

Rapid formation of C-S-H and CH Formation of

monosulfate

Diffusion-controlled reactions

Time of hydration ~4 h ~12 h ~24 h~10 m

0 ~6 ~12 Time (from mixing) [hour]

~18

Final set

Def

orm

atio

n, P

ress

ure

& T

empe

ratu

re

I & II V III & IV

Plastic Semiplastic Rigid

~24

Temperature Pressure

Deformation Initial set

(a) (b)

Figure 1: (a) Heat evolution during hydration (based on [3]) and (b) early-age linear autogenous deformation of concrete and the corresponding development of temperature and capillary pore pressure (from [1]).

2. TEST METHODS

2.1 Autogenous shrinkage test method In this study, autogenous linear deformation was monitored by a new test method, the

concrete digital dilatometer (CDD), developed in order to start measurement before setting, when concrete is fresh, as well as for the hardening concrete. The method is a modification of the CT1 digital dilatometer for pastes and mortars [5]. The CDD sample consists of a concrete specimen, cast in a steel coil reinforced 0.4 mm thick vapour-proof flexible polyurethene (PU) tube with inner diameter of 82 mm and specimen length ~400 mm, sealed


3

with a pair of hose clamps and 30 mm thick PVC end-caps equipped with O-ring sealing. The mould was placed in a mechanical stable measuring rig, where a digital gauge recorded the unrestrained linear deformation with an accuracy of 0.003 mm. The test was performed in a thermostable room at 20ºC, where the measurement recording was started at 30 minutes from water addition. The equipment and apparatus for a typical test are shown in Figure 2, and for each test three complete CDD set-ups were used. The experiments have also been supplemented with measurements of temperature and pore pressure in sealed specimens.

0 001

Specimen length 400 mm Didital gauge 1/1000 mm Adjustable

fixture Mould diameter 82 mm

0 001

Fixed end-cap Moving end-cap Flexible PU-tubeHose-clap Measuring rig

Figure 2: Test arrangement of the Concrete Digital Dilatometer (CDD) for linear

autogenous deformation measurement (from [1]).

2.2 Plastic shrinkage cracking test method The method used is intended for determination of the cracking tendency of concrete at

early ages (developed by Johansen and Dahl; see [6] and [7]) but does not separate between cracks formed before or after final setting. The variability (or scatter) of the test method is relatively high; for one test series (three samples) the coefficient of variation is in the range of 20 % to 50 % and the repeatability is in the same order (for three repeated series). In the test, the fresh concrete is cast between two concentric steel rings and the specimen is then placed under an air funnel, creating an air velocity of 4.5 m/s over a fresh concrete surface; see Figure 3. The crack sensitivity is expressed as a crack index, see Figure 3, and both the temperature and the weight loss of the specimens is continuously recorded. The measurements started 60 minutes after mixing. After 20 hours of drying, the rings were taken out of the rig and the crack index was measured as the average total crack area (crack length × crack width) on the concrete surface of each of the three specimens. The crack width was measured with a crack microscope (to an accuracy of 0.05 mm) and the crack length was measured with a digital measuring wheel (to an accuracy of ±1 mm).

Ø 300 Ø 600

80

Fan

Scale


Stress raisers

Base plate (steel)

Steel rings (Ø300/600)

Climat conditions: 23±2 ºC

35±5% RH

Sample

Air funnel

10 Sample

( )3

∑ ×=

widthcracklengthcrackareacrackAverage

Figure 3: Test arrangement for the determination of the cracking tendency (from [1]).


4

3. EXPERIMENTAL PROGRAM

3.1 Materials The mix design and its constituents, as used in this study, comprise typical materials and

compositions for self-compacting concrete (SCC) in Sweden. The properties of the dry materials and admixtures are listed in Table 1 and Table 2. The concretes were prepared in batches of 40 or 60 liters, mixed in a twin-shaft paddle mixer for 4 minutes after water was added to the premixed dry materials. The admixtures were added directly after the water.

Table 1: Properties of dry materials (cement, silica, filler and aggregate).

ID Type Product name Supplier Density [kg/dm3]

C CEM II/A-LL 42.5R Byggcement, Skövde Cementa 3 080 SF Silica fume Microsilica ELKEM 2 250 F Ground limestone Limus 40 NordKalk 2 670

A 0-8 Natural aggregate Hol Färdig Betong 2 650 A 8-16 Crushed aggregate Kungälv Färdig Betong 2 700

Table 2: Properties of admixtures.

ID Type Product name Supplier Density [kg/dm3]

Dry content [weight%]

SP Super plasticizer (polycarboxylate ether) Sikament 56 SIKA 1 100 37 % ACC Accelerator (sodium) SikaRapid-1 SIKA 1 200 37 % RE Retarder (polyalkyl ether) SikaRetarder SIKA 1 200 27 %

SRA Shrinkage-reducer (polymeric glycol) SikaControl-40 SIKA 1 000 – In order to evaluate the effect of constituent type and dosage, a number of different mixes

were investigated. The following mixes are presented in this paper (for the full experimental study see [1]): 1. Reference concretes: w/c 0.38, 0.45, 0.55 and 0.67 (see Table 3). 2. Silica fume: 5% and 10% SF by cement weight (SF replaced equal C volume). 3. Coarse aggregate content: 20%, 30%, 40% (REF) and 50% A 8-16 of total volume of

aggregate. The changes were replaced by equal volume of A 0-8. 4. Superplasticizer dosage: 0.6%, 0.8% (REF) and 1.0% SP dosage of C weight. 5. Accelerator and retarder: 1.5% ACC and 0.2% RE by C weight. 6. Shrinkage-reducing admixture: 1.0% and 2.0% SRA by C weight. 7. Conventional concrete: w/c 0.55 with 345 kg cement and 60/40% (0-8 mm/8-16 mm)

aggregate; and w/c 0.67 with 330 kg cement and 60/40% (0-8 mm/8-16 mm) aggregate.

Table 3: Recipe of SCC reference mixes in kg/m3. ID w/c 0.38 w/c 0.45 w/c 0.55 w/c 0.67 C 420 380 340 300 W 160 171 187 200

A 0-8 1021 998 960 926 A 8-16 694 678 651 628

F 40 100 160 220 SP 7.6 (1.8%C) 5.7 (1.5%C) 4.1 (1.2%C) 2.4 (0.8%C)


5

4. RESULTS

4.1 Autogenous deformation The results of the autogenous deformation measurements for the reference concretes and

one conventional concrete are presented in Figure 4(a), while Figure 4(b) shows the rate of the deformation. As can be observed, increased cement content (lower w/c) increased both the total chemical shrinkage and its rate, and thereby the total autogenous deformation. Increased cement content (lower w/c) also increased the autogenous deformation and its rate in the semiplastic period (between initial and final set). Finally, increased superplasticizer addition had a delaying and retarding effect on the hydration and thereby extends all the time to the initial and final set.

Furthermore, Figure 5 shows the relationship between the development of deformation,

temperature, and pore pressure for the concretes with w/c 0.45 (a) and w/c 0.67 (b). As can be seen, as long as the concrete is plastic the deformation develops with almost a linear relationship and, during this period, the temperature and capillary pore pressure undergo only small changes, which are linear. However, at one stage (at about five hours for w/c 0.67 and seven hours for w/c 0.45) the rate of the deformation is slowed down, indicating ‘setting’ of the concrete, and a knee point is reached. At this point in time, it can be seen that both the capillary pore pressure and the temperature reach an accelerating phase, which indicates that the dormant period is ended and that the cement hydration accelerates. Final set is reached at about 10 hours for w/c 0.67 and 13 hours for w/c 0.45, which for the deformation is manifested in a plateau slightly ahead of the temperature peak.

-1 200

-1 000

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6m

/m]

conventional w/c 0.55

SCC w/c 0.38

SCC w/c 0.45

SCC w/c 0.67

SCC w/c 0.55

-125

-100

-75

-50

-25

0

250 6 12 18 24

Time [hours]

Rat

e of

def

orm

atio

n [1

0-6 m

/m/h

our]

SCC w/c 0.38

SCC w/c 0.45

SCC w/c

SCC w/c 0.67

conventional w/c 0.55

(a) (b)

Figure 4: Autogenous deformation of self-compacting concrete with w/c from 0.38 to 0.67 (a) and (b) rate of deformation (from [1]).


6

0 5 10 15 20Time [hours]

Pres

sure

& D

efor

mat

ion

[-]

Tem

pera

ture

[-]

Deformation

Pressure

Temp.

Initi

al s

et

Fina

l set

w/c 0.45

0 5 10 15 20Time [hours]

Pres

sure

& D

efor

mat

ion

[-]

Tem

pera

ture

[-]

Deformation

Pressure

Temp.

Initi

al s

et Fina

l set

w/c 0.67

(a) (b)

Figure 5: Development of temperature, capillary pore pressure, and autogenous deformation (the axes have been normalized against maximum values): (a) for concrete with w/c 0.45; and (b) for concrete with w/c 0.67 (from [1]).

It is well known that silica fume increases the autogenous shrinkage, as well as accelerating the early hydratation and initiating the stiffness at earlier age. The results of the autogenous deformation measurements for the concretes with silica fume are presented in Figure 6. These results indicate that addition of silica fume increased the magnitude and rate of autogenous shrinkage in all periods (in the plastic, semiplastic, and rigid periods), and the effect increased with increased silica dosage. Silica fume decreased the plastic time period, and the effect increased with increased silica dosage. No changes in semiplastic time could be observed.

-1 200

-1 000

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6m

/m]

w/c 0.55 (ref)

w/c 0.55 Si 5%w/c 0.55 Si 10%

-125

-100

-75

-50

-25

0

250 6 12 18 24

Time [hours]

Rat

e of

def

orm

atio

n [1

0-6 m

/m/h

our]

w/c 0.55 (ref)

w/c 0.55 Si 5%

w/c 0.55 Si 10%

(a) (b)

Figure 6: Effects of silica fume (Si) on: (a) the autogenous deformation of self-compacting concrete with w/c from 0.55 and (b) the rate of deformation (from [1]).

The results for the concrete with varying coarse aggregate content are presented in Figure 7. It should be noted that the content of coarse aggregate had a large impact on the rheology, which can be deduced from the aggregate packing. Good packing improves the flowability as more water is made available for dispersing the particles. The results indicate that increased coarse aggregate content decreased both the magnitude and rate of autogenous


7

deformation. The effect was more apparent at higher content of coarse aggregate, which might be explained by the coarser particles’ ability to create a restraining matrix. No significant changes in times for the plastic and semiplastic periods could be observed, but an increased content of coarse aggregate tended to delay the rigid period.

-600

-400

-200

00 6 12 18 24

Time [hours] (w/c 0.55)

Def

orm

atio

n [1

0-6m

/m]

w/c 0.55 50% (8-16)

w/c 0.55 (ref) 40% (8-16)

w/c 0.55 30% (8-16)

w/c 0.55 20% (8-16)

-125

-100

-75

-50

-25

0

250 6 12 18 24

Time [hours] R

ate

of d

efor

mat

ion

[10-6

m/m

/hou

r]

w/c 0.55 20% (8-16)w/c 0.55 30% (8-16)

w/c 0.55 (ref) 40% (8-16)

w/c 0.55 50% (8-16)

(a) (b)

Figure 7: Effects of coarse aggregate content (8-16) on: (a) the autogenous deformation and (b) the rate of deformation (from [1]).

The results of the autogenous deformation measurements for the concrete with shrinkage reducing admixture (SRA) are presented in Figure 8. The results indicate that addition of SRA decreased the magnitude and rate of autogenous shrinkage for all concretes, and the effect increased with increased SRA dosage. In the semiplastic period the effect of SRA was not as pronounced as in the plastic and rigid periods. SRA showed no effect on the times of the plastic and semiplastic periods.

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6

m/m

] w/c 0.45 SRA 2%w/c 0.45 SRA 1%

w/c 0.45 (ref)

0 6 12 18 24

Time [hours]

w/c 0.55 SRA 2%

w/c 0.55 SRA 1%

w/c 0.55 (ref)

0 6 12 18 24

Time [hours]

w/c 0.67 SRA 2%w/c 0.67 SRA 1%

w/c 0.67 (ref)

(a) (b) (c)

Figure 8: Effect of the addition of shrinkage-reducing admixture (SRA) on autogenous deformation: (a) for w/c 0.45, (b) for w/c 0.55, and (c) for w/c 0.67 (from [1]).

It is well known that SP retards the concrete, as it delays the early hydration and initiates the stiffness at higher age. The results of the autogenous deformation measurements clearly


8

verify these phenomena, as can be seen in Figure 9(a). The results indicate that with increased SP dosage the magnitude of autogenous shrinkage increased, which can be explained by improved cement dispersion; see Holt [8]. An increased SP dosage also increased the rate of shrinkage in the plastic and semiplastic periods. After setting, the effect was the opposite, where the rate of shrinkage decreased with SP dosage. Moreover, an increased SP dosage prolonged the plastic time period, while no changes in the semiplastic time could be observed.

The results of the autogenous deformation measurements for the concrete with accelerator (ACC) and retarder (RE) are presented in Figure 9(b). The results indicate that in the plastic period (before initial setting), the accelerator increased the magnitude and rate of shrinkage while the retarder hade the opposite effect, which could be expected. In the semiplastic period, the accelerator and retarder showed no significant effect on magnitude and rate of shrinkage. In the rigid period, both the accelerator and retarder increased the rate of shrinkage. The accelerator tended to shorten the time to initial setting while the retarder tended to delay it. According to the supplier (SIKA), RE will have a small delaying effect at early hydration (to initial setting), and ACC will have no effect in the same period. The results in this study showed the same tendency.

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6m

/m]

w/c 0.67 SP 0.6%

w/c 0.67 (ref) SP 0.8%

w/c 0.67 SP 1.0%

-800

-600

-400

-200

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6m

/m] w/c 0.67 ACC (1.5%)

w/c 0.67 (ref)

w/c 0.67 RE (0.2%)

(a) (b)

Figure 9: Autogenous deformation of concrete with w/c 0.67: (a) effect of superplasticizer dosage (SP) 0.6%, 0.8%, and 1.0%; and (b) effect of accelerator (ACC) and retarder (RE) (from [1]).

4.2 Plastic shrinkage cracking The results for the average crack area for the reference concretes are presented in Figure

10(a). As can be seen, it is evident that the concrete with the high w/c-ratio (0.67) had the highest crack area and that the concrete with w/c 0.55 had the smallest crack area. Furthermore, there seems to exist an optimum w/c-ratio for the investigated reference mixes, which indicates that the w/c-ratio should be in the region of 0.55. The conventional concrete had a higher crack area for both the tested w/c values. Evaporation curves for the concretes are presented in Figure 10(b) where it can be seen that the evaporation is higher for a concrete with high w/c and lower for low w/c. Moreover, the conventional concrete had significantly higher evaporation, which could explain the increased cracking tendency. The initial rate of evaporation (before initial set) for the investigated concretes varied between


9

0.37 and 0.44 kg/m2/h (the low values for w/c ≤ 0.45); this can be compared with the evaporation from a free water surface of 0.50 kg/m2/h under same conditions.

0

50

100

150

0.35 0.45 0.55 0.65 0.75

w/c

Ave

rage

cra

ck a

rea

[mm

2 ]

Conventional

SCC

0

1

2

3

4

0 6 12 18 24Time [hours]

Evap

orat

ion

[kg/

m2 ]

conv. w/c 0.55Water

SCC w/c 0.38

SCC w/c 0.45

SCC w/c 0.55

SCC w/c 0.67

conv. w/c 0.67

(a) (b)

Figure 10: Influence of w/c on: (a) the crack tendency and (b) the evaporation (from [1]).

When silica fume replaced 5% and 10% of the cement in the concrete with w/c 0.55, the average crack area, which can be seen in Figure 11(a), increased dramatically. The effect that silica fume had on the evaporation can be seen in Figure 11(b). Silica reduced the evaporation. A possible explanation for the increased crack area is an increased amount of small particles, which increased the autogenous shrinkage, as could be seen in the CDD experiments.

0

50

100

150

0% 5% 10%Silica [weight% of Cement]

Ave

rage

cra

ck a

rea

[mm

2 ]

w/c 0.55

0

1

2

3

4

0 6 12 18 24Time [hours]

Evap

orat

ion

[kg/

m2 ] w/c 0.55 (ref)

w/c 0.55 Si 5%

w/c 0.55 Si 10%

(a) (b)

Figure 11: Influence of silica fume (Si) on: (a) the crack susceptibility and (b) the evaporation (from [1]).

That the aggregates are important is well known, and it is beneficial to have large aggregates and a high aggregate content. The effect of the coarse aggregate content (8-16 in relation to total aggregate content) on crack area can be seen in Figure 12(a). As the coarse aggregate was reduced, the crack area increased. Interestingly, however, an increased amount of fine aggregate content also reduced the evaporation, as can be seen in Figure 12(b). This indicates that it is not only the evaporation which determines the cracking behaviour; also the autogenous deformation plays an important role.


10

0

10

20

30

40

50

30% 40% (ref)Coarse agg. content [vol% of total]

Ave

rage

cra

ck a

rea

[mm

2 ]

w/c 0.55

0

1

2

3

4

0 6 12 18 24

Time [hours]

Evap

orat

ion

[kg/

m2 ]

w/c 0.55 (ref) 40% 8-16

w/c 0.55 30% 8-16

(a) (b)

Figure 12: Influence of coarse aggregate content (8-16) on: (a) the crack susceptibility and (b) the evaporation (from [1]).

The shrinkage-reducing admixture (SRA) had a positive effect on the crack area. As can be seen in Figure 13(a), for both the investigated concretes (w/c 0.55 and 0.67) the crack area was reduced with SRA. The effect of SRA on the evaporation can be seen in Figure 13(b). This effect starts to be notable at about three hours, after which point the evaporation and its rate were significantly lower for the concretes containing SRA. The main effect of the SRA is that it reduces the surface tension of the water (or pore solution), which has a positive effect on shrinkage as it reduces the capillary tension caused by a reduction in pore radius. However, the SRA also influences the rate of drying; the concretes containing SRA had a significantly lower weight reduction than the reference concretes, and the mechanism is notable as soon as the hydration starts.

0

20

40

60

80

100

0% 1% 2%Shrinkage reducer [weight% of Cement]

Ave

rage

cra

ck a

rea

[mm

2 ]

w/c 0.55

w/c 0.67

0

1

2

3

4

0 6 12 18 24Time [hour]

Evap

orat

ion

[kg/

m2 ]

w/c 0.55 SRA 2%

w/c 0.55 (ref)

w/c 0.67 (ref)

w/c 0.67 SRA 1%

w/c 0.67 SRA 2%

(a) (b)

Figure 13: Influence of shrinkage-reducing admixture (SRA) on: (a) the crack susceptibility and (b) the evaporation (from [1]).

The superplasticizer dosage seemed to have a large influence on the average crack area. As can be seen in Figure 14(a), when the SP-dosage was reduced to 0.6% the crack area was significantly reduced, and with the high SP-dosage (1.0%) the crack area increased. For the case with a delayed additional SP-dosage (0.2% after 30 min) the crack area increased even


11

more. The effect that the superplasticizer (SP) dosage had on the evaporation is presented in Figure 14(b). An increased dosage, as in the case with 1.0% and with a delayed dosage of 0.2%, resulted in an increased evaporation. With a reduced SP-dosage, 0.6%, the evaporation was reduced. That the evaporation increases is probably a result of the prolonged setting time.

0

50

100

150

200

0,6% 0.8% (ref) 1,0%SP dosage [weight% of Cement]

Ave

rage

cra

ck a

rea

[mm

2 ] w/c 0.67 SP 0.8% +0.2% delayed addition (30min)

w/c 0.67

0

1

2

3

4

0 6 12 18 24Time [hours]

Evap

orat

ion

[kg/

m2 ]

w/c 0.67 (ref) SP 0.8%

w/c 0.67 SP 0.6%

w/c 0.67 SP 1.0%

w/c 0.67 SP 0.8+0.2%

(a) (b)

Figure 14: Influence of superplasticizer (SP) dosage on: (a) the crack susceptibility and (b) the evaporation (from [1]).

Similarly to the effect of the SP-dosage, accelerating and retarding the concrete had a considerable effect on the crack area, as can be seen in Figure 15(a). For the concrete with accelerator the crack area was reduced, while for the concrete with retarder the crack area increased. The effect that the accelerator and the retarder had on the evaporation can be seen in Figure 15(b), and is comparable to the effect that the SP-dosage had. For the concrete with accelerator the evaporation was reduced, and for the concrete with retarder the evaporation increased.

0

50

100

150

ACC (1.5%) Reference RE (0.2%)

SP dosage [weight% of Cement]

Ave

rage

cra

ck a

rea

[mm

2 ]

w/c 0.67

0

1

2

3

4

0 6 12 18 24Time [hours]

Evap

orat

ion

[kg/

m2 ]

w/c 0.67 (ref)

w/c 0.67 ACC (1.5%)

w/c 0.67 RE (0.2%)

(a) (b)

Figure 15: Influence of accelerator (ACC) and retarder (RE) on: (a) the crack susceptibility and (b) the evaporation (from [1]).


12

5. DISCUSSION Based on the results of this study, the following paragraph contains a general discussion

of the mechanisms leading to the formation of plastic shrinkage cracks, and a brief description of possible actions to counteract them is presented. The formation of plastic shrinkage cracks is governed by the development of capillary pore pressure. However, depending on the characteristics of the concrete, different mechanisms are responsible. For concrete with a high w/c-ratio, evaporation is the governing mechanism, while for concrete with a low w/c-ratio (or with the addition of silica) the autogenous deformation is the driving force. This is presented in Figure 16, which also indicates that the optimum concrete has a w/c-ratio of 0.55. These observations also suggest that, in order to avoid plastic shrinkage cracking, different actions need to be taken depending on the type of mechanism. A concrete with a high w/c needs to be protected against early evaporation (e.g. by a curing membrane: see [1]). For a concrete with a low w/c-ratio, protection against evaporation may not help, as experience shows (see [9]), and the autogenous deformation needs to be reduced. This can be achieved for example with a shrinkage-reducing admixture (which also reduces evaporation) or by an increased coarse aggregate content.

w/c < 0.55 Autogenous deformation

0

1

2

3

4

0 6 12 18 24Time [hours]

Evap

orat

ion

[kg/

m2 ] Water

w/c 0.38w/c 0.45w/c 0.55

w/c 0.67

w/c > 0.55 Evaporation -1 200

-800

-400

00 6 12 18 24

Time [hours]

Def

orm

atio

n [1

0-6m

/m]

w/c 0.38w/c 0.45

w/c 0.67w/c 0.55

w/c 0.55optimal

0

2

4

6

0.38 0.45 0.55 0.61 0.67w/c

Rel

ativ

e cr

ack

area

[-]

Figure 16: Separation of mechanisms governing the formation of plastic shrinkage cracks.

6. CONCLUSIONS An experimental investigation of early age deformation and cracking tendency was made

on a number of self-compacting concretes, having w/c-ratio between 0.38 and 0.67, and the influence of various mix parameters was investigated. Autogenous deformations were measured and the cracking tendencies were investigated in a restraint ring specimen.

The results from the measurements of linear autogenous deformation by the CDD show that: − Increased cement content (lower w/c) increased the rate of and total chemical

shrinkage and thereby the autogenous deformation. − Increased coarse aggregate content decreased the magnitude and rate of autogenous

deformation. − Addition of silica fume increased the magnitude and rate of autogenous shrinkage and

the effect increased with increased silica dosage. Silica fume decreased the time of the plastic period.


13

− Addition of shrinkage-reducing admixture (SRA) decreased the magnitude and rate of autogenous shrinkage for all concretes, and the effect increased with increased SRA dosage.

− Increased SP dosage increased the magnitude of autogenous shrinkage. SP dosage increased the rate of shrinkage at the plastic and semiplastic periods. After setting, the effect was the opposite, where the rate of shrinkage decreased with SP dosage.

− Accelerator (ACC) increased the magnitude and rate of shrinkage, and retarder (RE) had the opposite effect, in the plastic period (before initial setting). ACC and RE showed no significant effect on the time of the periods, but RE tended to delay the time to initial setting.

The conclusions that can be drawn from the restraint ring tests are that: − The rate of evaporation was not always the governing factor for the cracking tendency. − Silica fume led to increased autogenous deformation (the CDD experiments) and

increased crack area in the ring test, though the evaporation was reduced. − Reduced aggregate content increased autogenous deformation, while the evaporation

was reduced. − A shrinkage-reducing admixture (SRA) proved to be very effective in reducing the

cracking tendency. SRA reduced the autogenous deformation as well as the evaporation.

− Delaying/retarding the hydration, with increased SP-dosage or by adding a retarder, increased both the autogenous deformation (the CDD experiments) and the crack area.

− Accelerating the hydration, e.g. by adding an accelerator, decreased both the autogenous deformation (the CDD experiments) and the crack area.

− The concrete with the high w/c-ratio (0.67) had the highest crack area. Furthermore, there seems to exist an optimum w/c-ratio for the investigated mixes, which indicates that the w/c-ratio should be in the region of 0.55; see Figure 10(a).

− The conventional concrete showed a similar tendency as the SCC (minimum cracking tendency at w/c 0.55) but, due to higher evaporation, it was found to be more susceptibile to cracking.

ACKNOWLEDGEMENTS

Financial support from SBUF (the Swedish construction industry’s organisation for research and development) and AB Färdig Betong is greatly appreciated.

REFERENCES [1] Esping, O. and Löfgren, I., 'Cracking due to plastic and autogenous shrinkage – Investigation of

early age deformation of self-compacting concrete – Experimental study', Publication No 05:11, Department of Building Technology, Chalmers University of Technology, Göteborg 2005, 95 pp.

[2] Wittmann, F.H., 'On the Action of Capillary Pressure in Fresh Concrete', Cement and Concrete Research, Vol. 6 (1976), pp. 49–56.

[3] Gartner, E.M., Young, J.F., Damidot, D.A., and Jawed, I., 'Hydration of Portland cement'. Chapter 3 in Structure and performance of cements (ed. Bensted and Barnes), Spoon Press, London, 2002.


14

[4] Barcelo, L., Moranville, M., and Clavaud, B., 'Autogenous shrinkage of concrete: a balance between autogenous swelling and self-desiccation', Cement and Concrete Research, Vol. 35 (2005), pp. 177–183.

[5] Jensen O. M., Hansen F., 'A dilatometer for measuring autogenous deformation in hardening Portland cement paste', Materials and Structures, Vol. 28 (181), pp. 406-409, 1995.

[6] Johansen, R. and Dahl, P.A., 'Control of plastic shrinkage of cement'. Paper presented at the 18th Conference on Our World in Concrete and Structures, Singapore, 1993.

[7] NORDTEST NT BUILD 433, 'Concrete: Cracking Tendency – Exposure to Drying During the First 24 Hours'. NORDTEST (Espoo, Finland, 1995).

[8] Holt, E.E., 'Early age autogenous shrinkage of concrete'. Technical Research Centre of Finland, VTT Publication 446 (2001), Espoo, 184 pp.

[9] Bjøntegaard Ø., Hammer T.A., and Sellevold E.J., 'Cracking in High Performance Concrete before Setting', Proceedings of the Int. Symposium on High-Performance and Reactive Powder Concretes, Sherbrooke, Aug. 1998.

P a p e r V I - E f f e c t o f l i m e s t o n e f i l l e r B E T ( H 2 O ) - a r e a o n t h e f r e s h a n d h a r d e n e d p r o p e r t i e s o f S C C

PAPER VI: EFFECT OF LIMESTONE FILLER BET(H2O)-AREA ON THE FRESH AND HARDENED PROPERTIES OF SCC

Esping O., “Effect of limestone filler BET(H2O)-area on the fresh and hardened properties of self-compacting concrete”, Submitted for publication in Cement and Concrete Research, 2007.


1

Effect of limestone filler BET(H2O)-area on the fresh and hardened properties of self-compacting concrete

O. Esping*

Department of Civil and Environmental Engineering, Chalmers University of Technology, Sweden

Abstract

This paper presents a study on the use of limestone fillers with different specific surface area and their effect on the fresh and hardened properties of self-compacting concrete (SCC). The surface area was determined by a simplified BET method using water vapour as adsorbate. A rheometer and a slump flow test were used to measure the flowability of fresh concrete. A concrete dilatometer was used to measure the autogenous shrinkage, and a ring-test for the plastic cracking tendency. The compressive strength was determined at 28 days. It was found that the measure of BET(H2O)-area can be used to evaluate the water requirement for constant workability of the SCC, where a change in BET(H2O)-area of 1 000 m2/kg corresponds to approximately 0.8% in moisture content. The results showed that filler with a large area will result in an increased autogenous shrinkage, decreased evaporation, lower plastic cracking tendency, and a higher compressive strength. With additional water the results was the opposite. Keywords: Concrete; Filler; Surface Area; Rheology; Autogenous; Shrinkage 1. Introduction

Ground limestone has been used in concrete production for the last 25 years [1], for the main purposes of lowering the costs and environmental load of cement production, but also to increase the concrete durability. More recently ground limestone is also used as a filler material to improve the workability and stability of fresh concrete. For a high flowable concrete, such as self-compacting concrete (SCC), limestone filler is added to increase the packing of the granular skeleton, bind excess water and increase the volume of the continuous phase of lubricating paste. SCC has to possess two incompatible properties: high flowability and high segregation resistance. This balance is made possible by the dispersing effect of high-range water-reducing admixture (superplasticizer) combined with cohesiveness of high concentration of fine particles in additional filler material [2]. The main mechanisms controlling this fine balance are related to surface physics and chemistry; hence SCC is strongly dependent on surface activity of the admixtures together with the high specific surface area generated by the fines [3]. A consequence of the high concentration of powder material and the

retarding effect of superplasticizer, the concrete may develop a large autogenous shrinkage and plastic cracking tendency [4]. At early age, when the cement paste is young and has poorly developed mechanical properties, evaporation and autogenous shrinkage, both incorporated in the plastic shrinkage, are the two main driving forces for cracking. When the concrete dries out, the loss of water from the paste generates negative capillary pressure, causing the paste to contract, which in turn can lead to cracks [5]. These contracting capillary forces are in reverse proportional to the meniscus radius, and hence the capillary tension stresses increase with decreasing pore sizes and interparticle spaces. For a concrete where evaporation is prevented, a contracting negative capillary pressure will also develop, but only when the hydration commences and the concrete sets [6]. As long as the concrete is fluid, autogenous shrinkage is generally considered to be equal to the chemical shrinkage; but once the self-supporting skeleton starts to form, the autogenous will diverge from the chemical [7,8]. The setting will be manifested as a change of the slope of the autogenous deformation, and once the internal voids are created the development of capillary pore


2

underpressure in the skeletal structure will cause an external deformation [9]. The pattern of autogenous deformation, exampled in Fig. 1, comprises three distinct stages which can be defined as plastic, semiplastic and rigid, separated by the time to initial and final set [11,12].


0 ~6 ~18

Def

orm

atio

n, P

ress

ure

& T

emp

Plastic Rigid

~24

Semiplastic



Initial set

Final set

Autogenous defor-mation (measured)



Fig. 1. Typical early-age linear autogenous deformation and corresponding development of temperature and capillary pore pressure (based on measures of a SCC w/c 0.45, from [12]).

The physical nature of better packing, by addition of filler material, will not only govern the concrete flowability, but also the compressive strength due to the denser matrix and the better dispersion of cement grains [13-16]. Furthermore, the surfaces of the filler material will act as nucleation sites for the early reaction products of CH and CSH, which will accelerate the hydration of cement clinkers (especially C3S) and consequently increase the early age compressive strength [17-20]. The effect of nucleation on the strength is dependent on the filler’s affinity to cement hydrates, and it increases with fineness and specific surface area of the filler. Limestone filler is not pozzolan, but nor fully inert as it reacts with the alumina phases of the cement. If the cement has a significant amount of tricalcium aluminate (C3A), calcium carboaluminate will be produced from the reaction between calcium carbonate (CaCO3) from the limestone and the C3A [21-27]. This reaction, accelerating the hydration and increasing the compressive strength, increases with the C3A content of the cement and the fineness and specific surface area of the filler.

Since the ratio of surface area to volume increases

exponentially with particle irregularity (shape, texture and porosity) and decreased size, and as this area has a predominating effect on fresh and hardened concrete [28-31], quantification of geometrical properties of fillers and fines is essential. Powder material (filler, cement, etc.) is traditionally characterized by the size distribution and the specific

surface area by Blaine fineness. Due to the methodology where the specific surface area is determined from air permeability, based on packed spherical particles, information about the shape, texture and surface porosity is neglected [28,32]. Other possibly more correct methods, such as the BET with nitrogen gas and image analysis, are more seldom used due to their complexity and costs.

In this work, the effect of the specific surface area of different limestone fillers on the fresh and hardened properties of self-compacting concrete (SCC) has been experimentally evaluated. The investigated properties were rheology, autogenous deformation, plastic cracking tendency and compressive strength. The surface area was determined by a simplified BET method, using water vapour as adsorbate. The change in BET(H2O)-area was translated to a change in water demand for the concrete mix to maintain a constant workability, based on the assumption that 30 full molecular layers of water covering the particle surface are required to decrease interactions and provide sufficient lubrication for flowability. 2. Materials and mix design

The mix design and its constituents, as used in this study, comprise typical materials and compositions for self-compacting concrete with w/c 0.55 in Sweden. The cement was a type CEM II/A-LL 42.5R, the fine aggregate a 0-8 mm natural ground gravel, the coarse aggregate a 8-16 mm crushed granite stone and the superplasticizer a polycarboxylate ether-based type. Five ground limestone fillers, with similar chemical composition but different size distributions and specific surface areas, were used. The area was determined by BET(H2O); A simplified method, based on the BET theory [33], where the area was determined gravimetrically on conditioned samples at different relative humidities and deduced from the adsorbed volume of water molecules required to cover and form a monolayer on the surface of the sample [28,34-37]. The measured BET(H2O)-area (SBETH2O), the area by Blaine (SBlaine) and the calculated area from size distribution (SSize) [34,37] for the five limestone fillers are given in Fig. 2. The size distribution, measured with a Malvern laser diffraction instrument, is presented in Fig. 3. The limestone filler named L40 has the smallest BET-area, and is used as a reference. It can be noted that the filler named L70 is the coarsest, but has the largest BET-area.


3

The BET(H2O)-area for the gravel is ~2800 m2/kg, and the BET(N2)-area for the cement is ~2000 m2/kg. For cement the BET(H2O) method is not valid. Due to the differences in size and polarity of the water and nitrogen molecule the area by BET(H2O) is larger than by BET(N2) [37]. With a linear relationship of 1.55 [12], the BET(H2O)-area for the cement can be estimated to 3100 m2/kg.

0

2 000

4 000

6 000

8 000

SBE

TH2O

[m2 /k

g]

0

200

400

600

800

SBl

aine

& S

Size

[m2 /k

g]

BET(H2O) 1808 2307 6148 1894 1821Blaine 357 565 254 345 353Size 259 458 160 165 154

L40 L15 L70 G100 G200

Fig. 2. The five limestone fillers’ specific surface area by BET(H2O) and by Blaine, and calculated from size distribution.

0

20

40

60

80

100

0.50.250.1250.0630.0320.0160.0080.0040.002Size [mm]

Pas

sing

[%]

L40L15L70G100G200

L15G100

L40(ref)G200

L70

Fig. 3. Size distributions for the five limestone fillers’.

Corresponding mixtures were made without presence of cement, in order to evaluate the effect (if any) of fillers’ surface area on the autogenous deformation without influence of chemical shrinkage. The cement content (340 kg/m3) was replaced with an equal volume of limestone filler (i.e. filler content increased from 160 to 458 kg/m3), whereas the other constituents was kept constant.

Finally, mixtures were made with additional water, compensating for the fillers’ differences in BET(H2O)-area in order to regain constant flowability. A model was used where 30 molecular layers of water covering the particle surface are required to decrease interactions and provide lubrication sufficient to create flowability [28]. An increase in BET(H2O)-area of 1000 m2/kg corresponds to a need for additional water of ~0.8% by mass of filler (or gravel) content for constant

flowability. In Table 1 the changes in mixing water due to differences in specific area of limestone filler, using L40 as reference, are given. Table 1. Mixing water, in litre/m3, with compensations for the differences in specific surface area of limestone fillers. The L40 with the smallest BET area is used as reference. L40 (ref) L15 L70 G100 G200 187.00 187.68

(+0.68) 192.89 (+5.89)

187.12 (+0.12)

187.02 (+0.02)

The concrete was prepared in batches of 40 litres,

and mixed in a BHS-60 twin-shaft paddle mixer for 4 minutes after water was added to the premixed dry materials. The admixtures were added directly after the water. 3. Experimental methods 3.1. Flowability test methods

A slump flow test was carried out by using a traditional slump cone (EN 12350-2), measuring the spread flow diameter (average from two perpendicular directions).

Concrete rheology measurements were performed using a ConTec Visco5 with a rotating concentric setup at a controlled shear rate (γ& ). The experimental geometry and measuring sequence are illustrated in Fig. 4. The rheological parameters, plastic viscosity (ηpl) and yield stress (σ0), were evaluated in accordance with the Bingham model:

pl0 ηγσσ ⋅+= & (Eq. 1). Both slump flow and the rheology test were

performed at 7 and 60 minutes after adding water to the mix.

0

2

4

6

8

0 10 20 30 40 50 60Time [s]

She

ar ra

te [1

/s]

No loggingLogging

Ro=145 mm Ri=100 mm

H=140 mm

Measuring inner cylinder

Rotating outer cylinder

Concrete sample

Fig. 4. The ConTec Visco5 geometry, and the measuring sequence for the Bingham evaluation and segregation estimation (from [28]).


4

3.2. Autogenous deformation test method

The autogenous linear deformation was monitored by a concrete digital dilatometer (CDD), able to measure before setting when concrete is fresh [12]. The method is a modification of the CT1 dilatometer for pastes and mortars by Jensen & Hansen [38]. The CDD setup consists of a concrete specimen cast in a steel coil-reinforced vapour-proof flexible PE-tube with tight end-caps, a measuring rig in stainless steel, and a digital gauge (3 µm accuracy) recording the unrestrained linear deformation; see Fig. 5. The test, containing three complete CDD setups, was performed at 20±1ºC, and was started at 30 minutes from water addition.

0 001

Specimen length ~400 mm Digital gauge(1/1000 mm)

Adjustable fixture Mould diameter 82 mm

Fig. 5. The Concrete Digital Dilatometer (CDD) for linear autogenous deformation measurement (from [12]).

Due to greater stiffness in the radial than in the longitudinal direction of the mould, in fluid state, the flexible mould transforms a large part of the volumetric deformation into a linear deformation. As the concrete undergoes transition from a fluid to a rigid state, the deformation becomes isotropic. Based on empirical evaluation, the ratio between linear and volumetric deformation for a liquid is approximately 0.8 [39]. No correction for this was made as the setting point is not a well-defined physical state but rather a continuous transformation from liquid to a solid state.

The experiments were supplemented with measurements of temperature and pore pressure placed in the centre of the core. The pressure transducers were connected to a de-aired water-filled system with a needle (internal diameter of 0.4 mm and 50 mm length). 3.3. Ring- test method for plastic shrinkage cracking

The plastic cracking tendency of concrete at early ages exposed to drying was evaluated, using a modified ring-test method [12], originally developed by Johansen & Dahl [40]. In the test, the fresh concrete is cast between two concentric steel rings and then exposed to drying; see Fig. 6. The measurement was started at 60 minutes after water

addition, and the sample was exposed to an air velocity of 4.5 m/s with 23±2ºC and 35±5% RH. Temperature, pore pressure and weight loss (evaporation) in the concrete specimens, and the strain in the inner ring, were continuously recorded. After 20 hours of drying the crack index was measured as the average (of three specimens) total crack area (crack length × crack width) on the concrete surface.

Ø 300 mmØ 600 mm

80 mm

Fan

Scale

Air velocity (4.5 m/s)

Stress raisersBase plate (steel) Steel rings (Ø300/600)

Sample

Air funnel

10 mm

Sample Cracks

Fig. 6. The ring-test method for evaluation of plastic

shrinkage crack tendency (from [12]). 3.4. Compressive strength test Compressive strength was measured on 150 mm

cubes after 28 days, in accordance with EN 12390-1,-2 and -3. All presented values of compressive strength (fc) are normalized to a porosity of 4%. Due to the differences in limestone fillers’ BET-area the flowability was not consistent between the mixtures; hence also the air content was not equal. A series of corresponding mixes with different air content, in the range of 0-5%, were made in order to evaluate the effect of air content (porosity) on the compressive strength. This was made in accordance with the general equation for ceramic materials [41]:

Peff k0cc ⋅⋅= (Eq. 2). The porosity, P, was

determined by the measured density in fresh and hardened states (according to EN 12350-6 and EN 12390-7), together with the measured air content by pressure test (EN 12350-7). The compressive strength “without” pores, fc0, was 55.0 MPa for the concrete de-aired in a vacuum desiccator for 30 minutes before moulding. The constant k was empirically evaluated to 6.5.


5

4. Results and discussion 4.1. Flowability

The results from the concrete flowability measurements (slump flow, yield stress and plastic viscosity) are presented in Fig. 7. All test results are represented by a mean value based on two or more tests.

0

200

400

600

800

Spr

ead

[mm

]

7 min

60 min

0

150

300

450

600

Yiel

d st

ress

[Pa]

7 min

60 min

30

40

50

60

0 2000 4000 6000 8000

SBETH2O [m2/kg]

Vis

cosi

ty [P

a s]

7 min

60 min

Fig. 7. Limestone fillers BET(H2O)-area vs. slump flow,

yield stress and plastic viscosity for the SCC mixtures (at 7 and 60 minutes).

The results shows that the limestone filler BET(H2O)-area have a strong influence on the SCC flowability. The larger area, the higher yield stress and viscosity, and consequently the lower slump flow. The loss in flowability by time is significant regarding slump flow and yield stress, and increases with BET(H2O)-area, whereas the change in viscosity by time is small.

Furthermore, results indicated that the differences in BET(H2O)-area can be compensated for with a

calculated change in water content in order to regain constant flowability. All mixtures with additional water showed nearly the same rheological measures as the reference mix with “L40”. At 7 minutes the slump flow for the reference SCC (L40) and the mixtures with additional water was 595±5 mm, the yield stress 71±12 Pa, and the viscosity 35±2 Pa·s. The deviations were similar for the measures at 60 minutes. The results are presented in [42].

The specific area by Blaine and from size distribution (see Fig. 2) showed a poor correlation with the flowability, which can be explained by the differences in size distribution (see Fig. 3). Thus, despite the fact that “L70” is coarser, it generates a larger specific area and thereby a stiffer consistency, and has a larger water demand. This might be explained by the filler’s geological origin and age, where “L70” has a rougher and more porous particle surface. 4.2. Autogenous deformation

The results (from [43]) of the autogenous deformation measurements are presented in Fig. 8, and the evaluated rate of deformation in Fig. 9. As can be observed, increased surface area by BET(H2O) increased both the magnitude and rate of shrinkage, primarily in the plastic region. The time to initial and final set, evaluated from the pattern of autogenous deformation when the rate reaches a minimum, was 6.1 and 11.6 hours from mix, respectively. The differences in BET-area had almost no effect on times to set (shorter than ±0.1 hours). This was confirmed by the measured temperature development, where the measures showed almost no differences.

Previous tests show that increased water content decreases the rate and magnitude of autogenous deformation, without affecting the time to set, which is equivalent to the effect decreased particle surface area [12].

-1200

-900

-600

-300

0


Def

orm

atio

n [1

0-6 m

/m] L40

L15

L70

G200G100

Fig. 8. Autogenous deformation for the SCC with different limestone fillers (from [43]).


6

-400

-300

-200

-100

00 2 000 4 000 6 000 8 000

SBETH2O [m2/kg]

Rat

e of

def

orm

. [10

-6/h

our]

Plastic

Semiplastic

Fig. 9. The limestone filler’s BET(H2O)-area and the evaluated rate of autogenous deformation from autogenous deformation in the plastic and semiplastic period.

Furthermore, Fig. 10 shows the development of the

measured pore pressure, and Fig. 11 the evaluated maximum rate of pore pressure which occurred at 7-8 hours from mixing [43]. The point of maximum underpressure could not be evaluated due to the loss of pressure at random times. This effect is usually referred to as the breakthrough pressure [5]. It can be noted that the rate of pore pressure, and rate and magnitude of autogenous deformation, showed to correlate with the limestone fillers’ BET(H2O)-area.

-100

-80

-60

-40

-20

0


Por

e pr

essu

re [k

Pa]

L40

L15

L70

G200

G100

Fig. 10. Development of capillary pore pressure during the first 12 hours from water addition (from [43]).

-60

-40

-20

00 2 000 4 000 6 000 8 000

SBETH2O [m2/kg]

Rat

e of

por

e pr

es. [

kPa/

hour

]

L40 & G200

L15 L70

G100

Fig. 11. The limestone filler’s BET(H2O)-area and the evaluated maximum rate of pore pressure (at 7-8 hours from mixing).

The results of the measured “autogenous”

deformation for the mixtures without cement (same composition, but cement is replaced with equal-volume limestone filler) are presented in Fig. 12. There was no separation observed, thus no cement was present, and the consistencies were similar to the corresponding SCC mixtures.

-800

-400

0

400

800


Def

orm

atio

n [1

0-6m

/m]

L70

L40G100G200

L15

Fig. 12. Sealed (“autogenous”) deformation for the mixtures where the cement was replaced with equal volume filler (from [43]).

The hydration process (generating chemical shrinkage) is generally considered to be the driving force of autogenous deformation; hence the term “autogenous” in this case might be incorrect. Even though no cement is present, the mixes generated an “autogenous” shrinkage of approximately 0.8 mm/m at 24 hours, counted after an initial swelling up to 4 hours. This is the same magnitude as for the mixes with cement, but those generated no swelling and almost all shrinkage before final set. At 48 hours, the development of shrinkage levels out. It ought to be noted that the effect of differences in surface area is augmented, as the content of limestone filler is larger relative the SCC mixtures (i.e. 458 instead of 160 kg/m3). The results indicate that larger BET(H2O)-area increases the deformation (both swelling and shrinkage). The expansion can be explained by water being absorbed by the filler (and aggregate) and by the disjoining pressure, i.e. the adsorption of water molecules in locations where the distance between two surfaces is restricted, inducing pressure and expansion [44]. Once lack of water occurs, this causes a restraining matrix of particles, and there will be pores with meniscus-generated capillary tension and thereby a contraction [45].


7

4.3. Plastic shrinkage cracking A representative measure (from [43]) showing the relationship between the development of strain, temperature, and pore pressure (at 20 and 60 mm depth) for one of the concretes exposed to drying is shown in Fig. 13. As can be seen, the pore pressure develops more rapidly in comparison with the sealed samples, in Fig. 10. The maximum underpressure is reached at 5-6 hours from mix, with approximately 30 minutes between the two depths. The measured strains indicate that cracking was initiated at 3-4 hours from mixing [43]. The average crack area is presented in Fig. 14, where it can be seen that the crack tendency is lower for concrete incorporating filler with a high BET(H2O)-area, and that, when adding extra water to compensate for the loss in flowability, the crack tendency increased significantly. An increase in filler’s BET(H2O)-area of 1000 m2/kg results in ~20% less cracking tendency.


Max

MinInitial crack

TemperatureStrain


Evaporation

Fig. 13. Development of pore pressure, evaporation, temperature and strain (normalized against maximum values), for the SCC with limestone filler L40 (from [43]).

0

20

40

60

80

0 2 000 4 000 6 000 8 000SBETH2O [m2/kg]

Ave

rage

cra

ck a

rea

[mm

2 ]

Additional water

No compensation

Fig. 14. The limestone filler’s BET(H2O)-area and the plastic shrinkage crack area, without and with extra water compensating for the differences in BET-area (“L40” as reference).

The differences in crack tendency are most likely a consequence mainly of evaporation, which is verified by the initial rate of measured evaporation (up to 4 hours from mix, before initial setting) in Fig. 15. A large particle surface area lowers the evaporation, whereas extra water increases the evaporation.

0.38

0.40

0.42

0.44

0 2000 4000 6000 8000SBETH2O [m2/kg]

Rat

e of

eva

por.

[kg/

(m2 ·h

)]

Additional water

No compensation

Fig. 15. Initial rate (<4 hours from mix) of evaporation, without and with extra water added to compensate for the BET(H2O)-area. 4.4. Compressive strength

The results (from [42]) of the measured compressive strength are presented in Fig. 16. Due to the differences in air content, the presented values are adjusted to an equal porosity of 4%, in accordance with Eq. 2. All test results are represented by a mean value based on two or more mixtures.

As can be observed, the compressive strength increased with BET(H2O)-area of the filler. And for the mixtures with additional water, compensating for the loss in flowability, the effect was the opposite. The increased strength with particle surface area can be attributed to the filler effect (denser packing and accelerated hydration due to nucleation). But as the filler with the largest area (“L70”) also was the coarsest, the increased strength can also be attributed to the filler’s porosity increasing its absorption of water and consequently reducing the effective water/cement ratio. A change in BET(H2O)-area by 1000 m2/kg proved to correspond to ~0.15 MPa in compressive strength for a SCC with 160 kg limestone fillers (i.e. ~1 kPa/kg filler).

That the porosity (accessible to water absorption) is incorporated in the BET-area is confirmed by the measured “autogenous” deformation on the mixtures without cement (Fig. 12), where the mixture with “L70” generated the largest shrinkage (and swelling).


8

45

46

47

48

0 2000 4000 6000 8000SBETH2O [m2/kg]

Com

p. S

treng

ht, 2

8d [M

Pa]

Additional water

No compensation

Fig. 16. The limestone filler’s BET(H2O)-area and the SCC compressive strength (28d), both without and with the extra water added to compensate for the BET(H2O)-area. 5. Conclusions

This paper has presented an experimental investigation of the effect of the specific surface area of different limestone fillers and their influence on the fresh and hardened properties of self-compacting concrete (SCC). The area was measured with BET(H2O), a simplified BET method using water (moisture) as adsorbate. Based on the experimental conditions and results in this study, the following observations and conclusions were made: o For better control and prediction of fresh and

hardened concrete behaviour, the specific surface area by BET is proposed as a potential means of geometrical characterization for fillers.

o Coarser filler can provide larger surface BET-area than a finer filler, due to differences in surface texture and accessible porosity.

o Limestone filler with larger BET(H2O)-area will decrease the SCC flowability, increase the autogenous shrinkage, decrease the evaporation and thus lower plastic cracking tendency, and generate a higher compressive strength.

o An increase in BET(H2O)-area of 1000 m2/kg corresponds to a need for additional water of ~0.85% by mass of filler content for constant flowability.

o For a SCC with 160 kg limestone filler, an increase in filler’s BET(H2O)-area of 1000 m2/kg results in a ~20% lower plastic cracking tendency.

o With additional water for constant flowability, compensating for the differences in the filler’s BET(H2O)-area, evaporation and plastic cracking tendency was increased, and strength was reduced. The effect of BET(H2O)-area on the cracking tendency and strength, with additional

water, are in reverse ratio to the effect without compensation.

o The mixtures where the cement was replaced with limestone filler generated an “autogenous” shrinkage of ~0.8 mm/m at 24 hours, counted after an initial swelling up to 4 hours. The magnitude of deformation (both swelling and shrinkage) corresponded to the limestone filler’s BET(H2O)-area.

References [1] M. Nehdi, S. Mindess, P.C. Aïtcin, Use of ground

limestone in concrete – A new look, Build. Res. J. 43 (1995) 245-261.

[2] H. Okamura, M. Ouchi, Self-compacting concrete. Development, present use and future, Proc. of 1st Int. Symp. on SCC, Stockholm, 1999, pp. 3-14.

[3] Rilem Technical Committee, Final report of Rilem TC 188-CSC Casting of self-compacting concrete, Mat. Struct. 2006).

[4] Ø. Bjøntegaard, T.A. Hammer, E.J. Sellevold, Cracking in high performance concrete before setting, in Proc. of Int. Symp. on High Performance and Reactive Powder Concretes, Sherbrooke, 1998.

[5] F.H. Wittmann, On the action of capillary pressure in fresh concrete, Cem. Concr. Res. 6 (1976) 49-56.

[6] E.E. Holt, Early age autogenous shrinkage of concrete, Tech. Res. Centre of Finland, VTT Publ. 446, Finland, 2001.

[7] T.A. Hammer, Test method for linear measurements of autogenous shrinkage before setting, in Autogenous shrinkage of concrete, E & FN Spon, London, 1998, pp. 143-154.

[8] O.M. Jensen, F. Hansen, Autogenous deformation and RH-change in perspective, Cem. Concr. Res. 31 (2001) 1859-1865.

[9] L. Barcelo, M. Moranville, B. Clavaud, Autogenous shrinkage of concrete: a balance between autogenous swelling and self-desiccation, Cem. Concr. Res. 35 (2005) 177-183.

[10] A. Radocea, A Study on the Mechanisms of Plastic Shrinkage of Cement-Based Materials, Doctoral thesis, Chalmers, Göteborg, 1992.

[11] L. Barcelo, M. Moranville, B. Clavaud, Autogenous shrinkage of concrete - a balance between autogenous swelling and self-desiccation, Cem. Concr. Res. 35 (2005) 177-183.

[12] O. Esping, I. Löfgren, Cracking due to plastic and autogenous shrinkage - Investigation of early age deformation of self-compacting concrete, Publ. 05:11, Chalmers Univ. Tech., Göteborg, 2005.

[13] P. Billberg, Influence of filler characteristics on SCC rheology and early hydration, in Proc. 2nd Int. Symp. on SCC, Japan, 2001, pp. 285-294.

[14] V.L.. Bonavetti, H. Donza, G. Menéndez, O. Cabrera, E.F. Irassar, Limestone filler cement in low w/c concrete: A rational use of energy, Cem. Concr. Res. 33 (2003) 865-871.


9

[15] M. Nehdi, S. Mindess, P-C. Aïtcin, Rheology of high-performance concrete: Effect of ultrafine particles, Cem. Concr. Res. 28 (1998) 687-697.

[16] A. Yahia, M. Tanimura, Y. Shimoyama, Rheological properties of highly flowable mortar containing limestone filler-effect of powder content and W/C ratio, Cem. Concr. Res. 35 (2005) 532-539.

[17] P. Lawrence, M. Cyr, E. Ringot, Mineral admixtures in mortars effect of type, amount and fineness of fine constituents on compressive strength, Cem. Concr. Res. 35 (2005) 1092-1105.

[18] A. Nomat, Interactions between chemical evolution (hydration) and physical evolution (setting) in the case of tricalcium silicate, Mat. Struct. 27 (1994) 187-195.

[19] M. Sari, E. Prat, J.K. Labastire, High strength self-compacting concrete - Original solutions associating organic and inorganic admixtures, Cem. Concr. Res. 29(6) (1999) 813-818.

[20] I. Soroka, N. Setter, The effect of fillers on strength of cement mortars, Cem. Concr. Res. 7 (1977) 449-456.

[21] V.L. Bonavetti, V.F. Rahhal, E.F. Irassar, Studies on the carboaluminate formation in limestone filler-blended cements, Cem. Concr. Res. 31 (2001) 853-859.

[22] K.D. Ingram, K.E. Daugherty, A review of limestone addition to Portland cement concrete, Cem. Concr. Comp. 13 (1991) 165-170.

[23] J. Péra, S. Husson, B. Guilhot, Influence of ground limestone on cement hydration, Cem. Concr. Comp. 21 (1999) 99-105.

[24] I. Soroka, N. Stern, Calcareous fillers and the compressive strength of portland cement, Cem. Concr. Res. 6 (1976) 367-376.

[25] H.F.W. Taylor, Cement Chemistry, 2nd Edition, London, 1997.

[26] S. Tsivilis, E. Chaniotakis, E. Badogiannis, G. Pahoulas, A. Ilias, A study on the parameters affecting the properties of Portland limestone cements, Cem. Concr. Comp. 21 (1999) 107-116.

[27] T. Vuk, V. Tinta, R. Gabrovek, V. Kaui, The effects of limestone addition, clinker type and fineness on properties of Portland cement, Cem. Concr. Res. 31 (2001) 135-139.

[28] O. Esping, Rheology of cementitious materials: effects of geometrical properties of filler and fine aggregate, Chalmers Univ. Tech., Göteborg, 2004.

[29] C.F. Ferraris, H.O. Karthik, H. Russell, The influence of mineral admixtures on the rheology of cement paste and concrete, Cem. Concr. Res. 31 (2001) 245-255.

[30] J.L. Gallias, The effect of fine mineral admixtures on water requirement of cement pastes, Cem. Concr. Res. 30 (2000) 1543-1549.

[31] T.C. Powers, The properties of fresh concrete, New York, 1968.

[32] J.H. Potgier, C.A. Strydom, An investigation into the correlation between different surface area determination techniques applied to various

limestone-related compounds, Cem. Concr. Res. 26 (1996) 1613-1617.

[33] S. Brunauer, H.P. Emmet, E. Teller, Adsorption of gases in multimolecular layers, Am. Chem. Soc., 1938.

[34] L. Ahlgren, Moisture fixation in porous building materials, Lund Univ. Tech., Rep. 36, Sweden (in Swedish), 1972.

[35] T. Allen, Particle size measurement – Surface area and pore size determination, London, 1997.

[36] S. Brunauer, The adsorption of gases and vapours, Vol. 1, Oxford, 1944.

[37] S.J. Gregg, K.S.W. Sing, Adsorption, surface area and porosity, New York, 1967.

[38] O.M. Jensen, F. Hansen, A dilatometer for measuring autogenous deformation in hardening Portland cement paste, Mat. Struct. 28 (1995) 406-409.

[39] O. Esping, I. Löfgren, Investigation of early age deformation in self-compacting concrete, Conf. Advanced Cement-based Materials, DTU, Copenhagen, 2005.

[40] R. Johansen, P.A. Dahl, Control of plastic shrinkage of cement, in 18th Conference on Our World in Concrete and Structures, Singapore, 1993.

[41] A.M. Neville, Properties of Concrete, 4th Ed., Pearson Educion Limited, 2000.

[42] O. Esping, SCC flowability: Effect of changes in particle surface area, and how to compensate for this, 5th Int. RILEM Symp. on Self-Compacting Concrete, Ghent, 2007.

[43] O. Esping, Investigation of autogenous deformation in self-compacting concrete, RILEM Conf. on Volume Changes of Hardening Concrete , pp 273-282, Lyngby, 2006.

[44] T. Nawa, T. Horita, Autogenous shrinkage of high-performance concrete, Proc. on microstructure and durability to predict service life of concrete structures, Japan, 2004.

[45] T.A. Hammer, Is there a relationship between pore water pressure and autogenous shrinkage – Before and During Setting?, in Self-desiccation and its importance in concrete technology, Lund Univ. Tech., TVBM-3104, pp. 27-38, 2002.


10

Early age properties of self-compacting concrete -...

Documents

Transcript of Early age properties of self-compacting concrete -...