Full Text 01

L ICE N T IAT E T H E S I S
Department of Civil, Environmental and Natural Resources Engineering

Division of Structural and Fire Engineering
Aggregates in Concrete Mix Design
Yahya Ghasemi Aggregates in Concrete Mix Design

ISSN 1402-1757
ISBN 978-91-7583-800-7 (print)
ISBN 978-91-7583-801-4 (pdf)
Luleå University of Technology 
Yahya Ghasemi
Structural Engineering
Aggregates in Concrete Mix Design
Yahya Ghasemi
Luleå 2017
Division of Structural and Fire Engineering
Department of Civil, Environmental and Natural Resources Engineering
Luleå University of Technology
SE-97187 Luleå, Sweden
Cover photo: 3D reconstruction from 2D X-ray images generated from Microtomography
data on Bro crushed sand (fraction 0-1 mm). Height and diameter = 1.7 cm.
Printed by Luleå University of Technology, Graphic Production 2017
ISSN 1402-1757
ISBN 978-91-7583-800-7 (print)
ISBN 978-91-7583-801-4 (pdf)
Luleå 2017
www.ltu.se
Academic thesis
For the Degree of Licentiate (Tech. Lic) in Structural Engineering, which by due of the
Technical Faculty Board at Luleå University of Technology will be publicly defended in:
Room F1031, Luleå University of Technology

Tuesday, March 28th, 2017, 13:00
Discussion leader: Dr. Hans-Erik Gram, Cementa AB

Principal supervisor: Prof. Mats Emborg, Luleå University of Technology
Assistant supervisors: Prof. Andrzej Cwirzen, Luleå University of Technology
Dr. Marin Nilsson, Luleå University of Technology
PREFACE
The project supporting this licentiate thesis was started in 2014 at the Department of civil,
environmental and natural resources engineering at the division of structural and fire
engineering of Luleå University of Technology in Sweden. The project was financially
supported by Formas.
I would like to use the opportunity to thank all the people and organizations that directly or
indirectly supported me and the project to this point. Particularly, I would like to express my
gratitude towards my supervisors, Professor Mats Emborg, Professor Andrzej Cwirzen, and
Dr. Martin Nilsson for their knowledgeable inputs, support and understanding.
I would also like to express my gratitude to my colleagues in the university and also the
technicians in the laboratory for the help and support.
Moreover, I should thank my parents for their unconditional love which has always been a
source of inspiration for me.
Yahya Ghasemi
Luleå, February of 2017
i
ii
ABSTRACT
The importance of studying the behaviour and properties of concrete can be highlighted by
considering the fact that concrete is the most used man-made material in the world. The very
first step in making concrete is its mix design and deciding the type and amount of constitutes
used in the production of concrete which should fulfil the requirements of the final product.
Mix design models are commonly used for the purpose of proportioning concrete ingredients
while anticipating the properties of the final product.
The current document deals with the commonly used principals in mix design models namely
particle packing theory and excess water/paste layer theories. The conducted studies includes
an investigation on accuracy of particle packing models (Toufar, 4C, CPM) and also tries to
address the issue with measurement of specific surface area of particles as an essential input
to water/paste layer theories.
It has been observed that the particle packing models can predict the packing density with
acceptable margin. However, it should be mentioned that the particle packing models by
themselves are not mix design models but should be rather used as a part of a mix design. In
addition, it was found that the accuracy of calculating the specific surface area of particles
based on their size distribution curve can be further improved by assuming angular platonic
solids as uniform shape of aggregate instead of traditional approach of assuming spheres for
aggregates’ shape.
Keywords: Particle packing, excess paste/water layers theories, specific surface area, mix
design, loose packing.
iii
SAMMANFATTNING
Vikten av att studera beteendet och egenskaperna hos betong kan belysas genom att betrakta
det faktum att betong är det mest använda konstgjorda materialet i världen. Det allra första
steget i att göra betong är proportioneringen inkluderande val av typ och mängd av olika
beståndsdelar som ska användas vid framställning av betong och som ska uppfylla kraven i
den slutliga produkten. Olika modeller används vanligen för att proportionera betongens
delmaterial och uppfylla egenskaperna.
Denna licentiatuppsats behandlar de vanligaste principerna för betongproportionering

nämligen partikelpackningsteori och teorier för överskott av vatten/cementpasta. De
genomförda studierna omfattar en undersökning av noggrannheten hos
partikelpackningsmodeller (Toufar, 4C, CPM) och försöker också att ta upp frågan om
mätning av partiklarnas specifika yta hos partiklar som en viktig input till teorier för överskott
av vatten/cementpasta.
Det har observerats att partikelpackningsmodeller kan förutsäga packningsdensiteten med

acceptabel marginal. Det bör dock nämnas att partikelpackningsmodeller i sig inte är
proportioneringsmodeller utan bör snarare användas som en del av proportioneringen.
Dessutom konstaterades det att noggrannheten hos beräkningen av den specifika ytan av
partiklar baserat på deras storleksfördelningskurva kan förbättras ytterligare genom att anta
platoniska kroppar som former på sand och grus i stället för det traditionella
tillvägagångssättet att anta sfärer.
Nyckelord: partikelpackning, teorier för överskott av vatten/cementpasta, specifik yta,

proportionering, lös packing.
iv
TABLE OF CONTENT
PREFACE ................................................................................................................................................ i
ABSTRACT ............................................................................................................................................iii
SAMMANFATTNING ........................................................................................................................... iv
TABLE OF CONTENT ...........................................................................................................................v
NOTATIONS ......................................................................................................................................... vii
ABBREVIATIONS ................................................................................................................................. ix
1. INTRODUCTION ............................................................................................................................... 1
1.1. Background .................................................................................................................................. 1
1.2. Research Objectives and Questions.............................................................................................. 3
1.3. Limitations ................................................................................................................................... 4
1.4. Approach ...................................................................................................................................... 4
1.5. Structure of the thesis ................................................................................................................... 5
1.6. Appended papers .......................................................................................................................... 5
2. AGGREGATES IN CONCRETE ....................................................................................................... 7
2.1. Purpose and role of aggregates ..................................................................................................... 7
2.2. Origin and classification of aggregates ........................................................................................ 9
2.3. Physical properties of aggregates ............................................................................................... 10
3. MIX DESIGN APPROACHES......................................................................................................... 13
3.1. Ideal size distribution curve ....................................................................................................... 15
3.2. Particle packing theory ............................................................................................................... 16
3.2.1. Toufar Model........................................................................................................................... 17
3.2.2. DTI 4C model.......................................................................................................................... 18
3.2.3. Compressible packing model (CPM) ...................................................................................... 18
3.3. Multi-phase approaches .............................................................................................................. 20
3.3.1. Particle-matrix model .............................................................................................................. 21
3.3.1.2. The flow resistance ratio ...................................................................................................... 21
3.3.1.3. The air voids modulus .......................................................................................................... 22
3.3.1.4. Workability function for concrete ........................................................................................ 23
3.3.2. Layer theories .......................................................................................................................... 23
3.3.2.1. Square-cube law ................................................................................................................... 26
3.3.2.2. Equivalent polyhedron shape ............................................................................................... 27
v
3.4. Concluding remarks ................................................................................................................... 29
4. METHODS........................................................................................................................................ 30
4.1. Test methods .............................................................................................................................. 30
4.1.1. Sampling of aggregates ........................................................................................................... 30
4.1.2. Sieve analysis .......................................................................................................................... 30
4.1.3. Specific gravity........................................................................................................................ 30
4.1.4. Packing methods...................................................................................................................... 31
4.1.5. Surface area measurements ..................................................................................................... 31
4.2. Theoretical methods ................................................................................................................... 32
4.2.1. Studied particle packing models.............................................................................................. 32
4.2.2. Layer theories .......................................................................................................................... 32
5. EXPERIMENTAL RESULTS .......................................................................................................... 34
5.1. Packing models study ................................................................................................................. 34
5.2. Computation of specific surface area ......................................................................................... 36
6. DISCUSSION AND CONCLUSIONS ............................................................................................. 38
6.1. Accuracy of particle packing models ......................................................................................... 38
6.2. Estimation of specific surface area ............................................................................................. 38
6.3. Future work ................................................................................................................................ 39
6.3.1. Estimation of SSA ................................................................................................................... 39
6.3.2. Water requirement of mixtures................................................................................................ 40
References ......................................................................................................................................... 42
APPENDIX 1 ........................................................................................................................................ 53
vi
NOTATIONS
Greek
Įi Actual packing density of class i [-]
Įagg Packing density of aggregates [-]
Įk Steepness of Kp function [-]
Įt Calculated packing density [-]
ȕi Virtual packing density of class size i [-]
ȕj Virtual packing density of class size i [-]
ȕti Calculated virtual packing of a mixture size i being dominant [-]
ȜQ Flow resistance ratio [-]
ȡ Density [kg/m3]
ȡb Bulk density [kg/m3]
ȡs Relative density [kg/m3]
Roman
a Polyhedron edge length [mm]
ac Circumsphered edge length [mm]
aij Coefficient presenting the loosening effect [-]
am Midsphered edge length [mm]
asph Spherical specific surface area [cm2/cm3][cm2/Kg]
Ap Specific surface area of particles [cm2/cm3][cm2/Kg]
bij Coefficient presenting the wall effect [-]
di Diameter of dominant particle size class i [mm]
di,arith Arithmetic mean of fractions i and i+1 [mm]
di,geo Geometric mean of fractions i and i+1 [mm]
vii
dഥన Mean diameter of fractions i and i+1 [mm]
dj Diameter of dominant particle size class j [mm]
D Diameter of particles [mm]
Dmax Maximum particle size of the aggregate in the mix [mm]
Dmin Minimum particle size of the aggregate in the mix [mm]
e Void content [-]
Fi Accumulated flow of an ideal fluid through FlowCyl [m3/s]
Fmp Fineness modulus for the coarse aggregate [-]
Fms Fineness modulus for the sand [-]
Fp Volume fraction of fluid phase of concrete [-]
Ft Accumulated flow of test material through FlowCyl [m3/s]
Hm The air void modulus [-]
Hp Air void ratio in the coarse aggregate [-]
Hs Air void ratio in the sand [-]
kd Factor that relates the packing density to the diameter ratio of particle classes [-]
ks Factor that deals with the placement of particles in relation to each other [-]
K Compaction index [-]
mi Mass percentage of the fraction between di and di+1 [Kg/Kg]
mp Mass of all particles in a mixture [Kg]
Ms Mass of the solid [Kg]
p Porosity [-]
P(D) Fraction that pass the sieve with opening D [-]
r Median radius of equivalent sphere [mm]
SSAi Specific surface area of particles in fraction i [cm2/cm3][cm2/Kg]
tep 7KLFNQHVVRIH[FHVVSDVWHOD\HU>ȝP@
viii
tew 7KLFNQHVVRIH[FHVVSDVWHOD\HU>ȝP@
Tp Aggregate parameter of the coarse aggregate [-]
Ts Aggregate parameter of the sand [-]
Vagg Volume of aggregates [m3]
Vco Volume fraction of coarse aggregate [-]
Vcp Volume of cement paste [m3]
Vep Volume of excess paste [m3]
Vew Volume of excess water [m3]
Vi Volume of fraction i [m3]
Vp Volume of internal pores [m3]
Vs Volume of the solid [m3]
Vsa Volume fraction of sand in the aggregate [-]
Vt Total volume of the solid [m3]
Vv Volume of the external granular voids [m3]
Vw Volume of water in a mixture[m3]
yi Volume fraction of size class i [-]
yj Volume fraction of size class j [-]
ABBREVIATIONS
CPM Compressible Packing Model
HSC High Strength Concrete
LALLS Low Angle Laser Light Scattering technique
LPDM Linear Packing Density Model
PMM Particle Matrix Model
PSD Particle Size Distribution
SCC Self Compacting Concrete
SSA Specific Surface Area
SSD Saturated surface dry
SSM Solid Suspension Model
ix
x
Introduction 1
1. INTRODUCTION
1.1. Background
Historical evidences indicate that cement was first used by ancient Macedonians around
eighth century. The knowledge of making hydraulic cement was later on documented by
French and British engineers in the 18th century. Construction with cement and also usage of
reinforcement in the structural design eventually led to making concrete the most used man
made material. As the fundamental knowledge of making cement and concrete developed and
was able to cover the basic questions about constitutes of concrete, researchers have been
continuously working with the ways of optimizing mix design recipes. Optimization can be
achieved by means of studying the ingredients of concrete mixes with the aim of maximizing
the performance of concrete in both fresh and hardened state while keeping a low cost of
production and limiting the pollutants released in the air due to cement production. As a
result, several attempts have been made on formulating the mix design of concrete.
Understanding the role of constitutes in fresh concrete is fundamental to the production of
high quality concrete at fresh state, during hardening and as a hardened structural material.
Fresh concrete can be characterized by several aspects among which workability is the most
important one and is chiefly influenced by the water requirement, which in turn is a function
of aggregates’ shape, grading, and fine content. As for the performance of the hardened
concrete, the crucial factors are water to cement ratio which influences strength and
permeability and cement characteristics and performance.
Among the components of concrete, aggregates have an important role especially in fresh
stage as 60% to 80% of concrete volume is occupied by them. Moreover, increasing the
amount of aggregates in volume of concrete corresponds to less usage of cement which has
several beneficial effects, e.g. reduction in the cost of producing concrete, decrease in some of
the durability problems of hardened concrete, reducing shrinkage and cracking, etc.
In addition, reduction in usage of cement leads to a decrease in pollution caused by its

production. The cement industry produces about 5% of global man-made CO2 emissions; the
amount of CO2 emitted by the cement industry can be as high as 900 kg of CO2 for every
1000 kg of cement produced (Mahasenan et al., 2003). It should be noted that the cement
industry worldwide and especially in Scandinavia and Europe takes its responsibility and
strong efforts are taken to reduce the CO2 emissions at production. Some companies (e.g.
Cementa) have formulated a zero-vision (“Carbon capture newsletter”, 2014) and was able to
reduce the CO2 emissions per ton of cement to lower than 700 kg. Others companies are
2 Introduction
engaged in carbon capturing of emitted gas (“Meeting the challenge through a zero vision”,
2014) describing a Heidelberg Cement supported project. Also, concrete producers worldwide
are now striving to reduce the amount of clinker and thus CO2 by replacements such as fly
ash, blast furnace slag, lime stone filler etc.
Currently, there are several models available for predicting the properties of concrete in both
fresh and hardened states. Most of these models are based on the assumption that the
properties of concrete in fresh state i.e. flow properties and workability are chiefly governed
by the particle size distribution (PSD) and the particle packing (Glavind and Pedersen, 1999).
The packing density concept can be used as a part of concrete mix design with the aim of
minimizing the inter-particle voids between the constituents of concrete in order to reduce the
paste demand. Packing density is the ratio of the volume of solids to the bulk volume of the
solid particles (Toufar et al., 1976; Quiroga et al., 2004). The date for one of the first articles
on particle packing goes as far as 1892 (Feret, 1892), further research were conducted mainly
concentrating on designing of an ideal aggregate size distribution curve (Fuller and
Thompson, 1907; Andreasen and Andersen, 1930). In 1929 the first analytical packing model
was designed to predict the void ratio of a mixture of two particle groups (Furnas, 1929).
Since then, plenty of researches were conducted on the subject of packing resulting in
development of several analytical models and computer-aided mix design software.
According to the above-mentioned models, particle packing can be increased by modifying

particle size distribution (PSD) which in turn usually leads to increasing the share of fines.
Packing theory assumes that adding fine particles to a particle structure helps fill up the voids
in between the particles and leaving only minimum space for water. In this way, adding fine
particles will reduce the water requirement (De Larrard, 1999; Kronlöf, 1994; Fennis, 2011).
However, the packing of aggregate is dependent also on the shape of the aggregate particles,
an effect that is more difficult to comprehend and it is indirectly accounted for by measuring
the packing of monosized fractions.
Another approach to compiling a mix design model is based on excess paste/water layer
theories first introduced by Kennedy (1940). A hypothesis by Brouwers and Radix (2005)
states that the relative slump of a water-powder mixture becomes a function of the specific
surface area (SSA) when sufficient water is present to flow. Based on the hypothesis, a thin
layer of adsorbed water molecules around the particles is necessary to assure the flow
characteristics of the hydrating system. It is reported that the thickness of this water layer is
related to sensitivity of the mix to changes in the water content and also the specific surface
area of the material used, as was later confirmed by Hunger (2010). Moreover, the layer
theories assume that the water demand of a mixture depends on the specific surface area of
the particles in that mixture. Increasing the surface area by adding small particles will increase
the water requirement (Hunger and Brouwers, 2009; Maeyama et al., 1998; Midorikawa et al.,
2009) which is in contrast to particle packing theory.
Introduction 3
Both approaches (Particle packing and Water/paste layer theories) strongly depend on the
shape of the aggregate in one way or another, that is especially more essential when it comes
to water/paste layer theories which require specific surface area as an input for the model.
While it is possible to directly measure specific surface area (SSA), the complexity of
instrument required for the measurement imposes issues such as the availability of the testing
instruments and the cost. It is also possible to estimate the SSA using the PSD data and the
assumption that particles have ideal spherical shapes (McCabe et al., 1993).
Although currently there are several advanced concrete mix design models, they are rarely
used by the concrete industry. One of the main reasons that these models are not used in
practice is the complexity of the advanced models and the number of empirical input data that
is required to use the models correctly. The input data for some of the models includes
extensive chemical and physical tests on the ingredients of concrete. Moreover, some of the
required data cannot be readily measured and/or in some cases there are no commonly
accepted methods for conducting the measurement, as an example measuring the specific
surface area of the particles. As a result of complexity and in cases lack of accuracy of the
tests required for measuring the specific surface area, in most cases the value is calculated
mathematically based on the size distribution curve and assumption of spherical shape for the
particles. However, even in case of computation of specific surface area, the effect of square
cube law is usually neglected.
The above mentioned issues emphasise the need for a comprehensive yet simple mix design
model that can be both used in practice in the industry as well as further developments of mix
design models. The thesis aims to lay a foundation for such a model by studying the role of
aggregates as they form most of the concrete volume.
1.2. Research Objectives and Questions
According to above, the main objective of the project is to formulate an approach to mix
design where the workability of the fresh concrete can be anticipated before actual mixing
takes place. The mix design method may be based on principles of particle packing and water
layer theory. Water demand, shape of aggregates and changes in size distribution curve, will
be studied as the mentioned parameters can greatly influence the properties of concrete in
fresh state.
The research is expected to answer the following questions:
1- Is it possible to formulate a simple yet efficient mix design approach?
2- Can optimizing concrete constitutes lead to a greener less pollutant production of concrete?
3- What is the role and significance of water demand in workability of fresh concrete?
4 Introduction
4- What are the main parameters that results in aggregate from one quarry to have different
performance properties comparing to another quarry? (shape, flakiness, size distribution
curve, mineralogy etc.)
1.3. Limitations
The studies were restricted by various limitations which bounds the scope of this document:
Firstly, the vast domain of interconnected parameters that affect the properties of fresh
concrete makes it difficult to isolate and study each parameter without considering the
possible effect of changes in other constitutes in concrete.
Secondly, it should be mentioned that all of the packing studies conducted were only based on
loose packing method and does not include other packing methods such as hard packing and
vibration compacted packing. Moreover, the used materials were taken from a limited number
of Swedish quarries.
Thirdly, the complexity and the randomness in properties of aggregates such as shape, surface
texture, size distribution curve, chemical components, etc. makes it almost impossible to
simulate their flow in fresh concrete.
Fourth, this document only deals with the role of aggregates in mix design as a part of the
overall research planned for completion of the PhD and does not include tests on concrete at
this stage.
Finally, the complexity and the high cost of conducting accurate specific surface
measurements limit the number of samples that can be tested.
1.4. Approach
Similar to general approach for scientific studies at Luleå University of technology, the
presented research started by a literature review on available mix design models and the role
and importance of concrete mix constitutes with the aim of understanding the current
knowledge on the subject and also to understand the research gaps. The research questions
were then formulated based on the revealed gaps and neglected aspects of the subject.
Available mix design approaches were divided in to three categories: a) Ideal size distribution
curve, b) methods that are based on particle packing theory and c) the ones that are based on
the concept of water demand and water/paste layer theories. The results from laboratory
measurements of loose packing and calculations based on particle packing models (see
Section 3.2) were presented and compared in Papers I and II. Papers III and IV deal with the
principals of defining shape of particles by utilizing specific surface area which directly
affects the water demand of mixtures. The last two papers were built on comparison of
measured specific surface area by Blaine test to theoretically calculated specific surface area
(3.3.2) based on size distribution curve and assumption of spherical shape for the particles.
Introduction 5
1.5. Structure of the thesis
The thesis consists of four main parts:
- Theoretical background (Chapters 2 and 3);
- Method (Chapter 4)
- Results and discussions (Chapters 5 and 6)
- Conducted studies and published papers (Appendix 1)
The theoretical background consists of an overview on the concepts and theories that were
utilized within the scope of the thesis which includes: General information on the role of
aggregates and the influence of their properties on fresh concrete (Chapter 2), an overview of
mix design approaches particularly particle packing theory and water and paste layer theories
(Chapter 3) and methods used for conducting measurements and calculations (Chapter 4). The
theoretical part should be viewed as background containing the definitions and theories that
are necessary for understanding the approaches taken in the studies in the last part of the
thesis (Appendix 1).
1.6. Appended papers
Paper 1
“Particle packing of aggregates for concrete mix design: Models and Methods.”, Ghasemi,
Y., Johansson, N. (2014), Published in proceeding of the XXII Nordic Concrete Research
symposium, Reykjavik, Iceland, Aug 13-15, 2014, pp. 109-112.
The paper deals with the concept of particle packing theory and explores different analytical
models compiled as a tool to be used in a mix design model, namely Toufar model, CPM and
4C.
Paper II
“Particle packing of aggregates for concrete mix design: Models vs. reality.”, Ghasemi, Y.,
Emborg, M. (2014), Published in Nordic Concrete Research, Vol. 51, No. 3, December
2014,pp. 85-94.
In continuation of the previous paper, second paper examines the reliability of the particle
packing models. Seven binary aggregate mixes were studied, the packing density for mixes
were measured in the lab and is compared to the results obtained from the models.
Paper III
“Quantification of the shape of particles for calculating specific surface area of powders.”
Ghasemi, Y., Emborg, M., Cwirzen, A., (2016), Published in proceeding of the international
RILEM conference, MSSCE2016, Lyngby, Denmark, Aug 22-24, 2016, pp. 31-41.
6 Introduction
The conference paper discusses the possibility of assigning platonic solids as the uniform
shape of the particles instead of the traditional assumption of spherical shape which can
increase the accuracy of size distribution curve based calculation of specific surface area.
Paper IV
“Estimation of Specific surface area of Powder particles Based on Size distribution Curve.”
Ghasemi, Y., Emborg, M., Cwirzen, A., (2017), Submitted to the magazine of concrete
research.
In paper IV, some lab results were presented, the specific surface area of particles were
reported as Blaine value and were compared to the calculation based on the assumption of
different platonic solids as the uniform shape of the particles.
Aggregates in concrete 7
2. AGGREGATES IN CONCRETE
2.1. Purpose and role of aggregates
Understanding the role of aggregates in concrete is fundamental to the production of good

concrete as aggregates have their greatest influence on the performance of fresh concrete
(Alexander and Mindess, 2010).
The aggregates form 60 to 80 per cent of the volume of concrete. Despite the amount of
aggregates used in production of concrete, their influence is sometimes overlooked and the
aggregates are considered only as fillers. However, it is essential to consider and study all of
constitutes in concrete mixtures, including aggregates, in order to obtain a comprehensive
understanding of the behaviour of the final product. As a result of such considerations, it is
clear that the role of aggregates is crucial to ensure satisfactory performance of mixtures since
they govern the volumetric stability of concrete. They also may influence the moisture related
deformations (e.g. shrinkage) as shown in Figure 2.1.
Figure 2.1. Effect of addition of aggregate to a cement paste on reduction of shrinkage (Powers,
1971)(from Addis and Owens, 2001)
8 $JJUHJDWHVLQFRQFUHWH
In addition, aggregates have an important effect on concrete strength by providing rigidity to

the material which governs resistance to applied loads and undesired deformations.
The original w/c ratio law for concrete strength introduced by Abram (1918) was limited to
normal structural concrete with maximum aggregate size of 38 mm and w/c ratio greater than
0.4. The simplified approach proposed by Abram led to common inaccurate assumption that
properties of aggregates have no effect on the compressive strength of concrete causing the
aggregate to be considered solely as inter fillers. While Abram’s law was defined for a special
scenario, at present it is known that other factors than w/c ratio also play a role in estimating
the compressive strength of concrete. This is especially true in case of high strength concrete
(HSC) in which, due to the dense microstructure and strong transition zone, the mechanical
properties of the aggregates become more important (Alexander and Mindess, 2005). In
addition, as an example, a study conducted by Walker and Bloem (1960) on the effect of
maximum size of aggregates on concrete strength independently of w/c ratio suggests that the
strength decreases as aggregate size increases over the full range of w/c, the typical results are
shown in Figure 2.2. However, generally the discussed relationship has limited significance as
aggregates with the same size but from different sources can lead to broader strength
differences comparing to the influence of size.
Figure 2.2. Water/Cement ratio vs. strength for maximum size of aggregates. (Walker and Bloem,
1960)
The mentioned effects should be separated from the characteristics of aggregate that influence
the water requirement and consequently strength of a mix such as shape, surface texture,
grading, and maximum size of aggregate. For example, regarding surface texture, concretes
made with smoother gravels suffer from crack initiation at lower compressive stresses
comparing to concretes made with coarser textured aggregates (Alexander and Mindess,
2005).
2.2. Origin and classification of aggregates
Most of the properties of an aggregate derive from its parental rock chemical and mineral
composition which affects strength, stiffness, density, pore structure and permeability. Rocks
experience geothermal and weathering processes which can produce granular materials in the
form of natural gravels and sands, these types of aggregates can be used in concrete
production without any modification or additional processes. On the other hand, crushed
aggregates, defined as granular materials produced by human related activities such as
blasting, crushing and so on, have rougher surface and more angular shapes comparing to
natural sands. In that sense, the aggregates can be classified into two main categories: natural
and crushed. see Section 2.3.
In terms of origin, rocks are generally classified as igneous, sedimentary, or metamorphic

based on how the rocks were formed over long periods of geological time.
All rocks originate as igneous rocks, which are formed through cooling and solidification of
molten materials that are underlying the earth’s crust. If the solidification occurs slowly, the
rocks will be labelled as intrusive rocks (coarse to medium grained) and in case the molten
material forces its way to the surface and crystallize more rapidly the extrusive rocks will be
formed (fine grained).
Sedimentary rocks are formed by chemical and mechanical breakdown of pre-existing rocks
in a process of deposition and cementation of the materials. The layered structure of particles
settlement can result in undesired shapes and flaky grains in the case of aggregates crushed
from such rocks.
Lastly, metamorphic rocks are made of pre-existing igneous or sedimentary rocks when the
original rock is subjected to high pressure and temperature usually at great depth.
Hydrothermal metamorphism can result in the minerals in these rocks being re-formed and re-
crystallized which makes them less durable and less stable in some cases, for example the
formation of alkali-reactive silicates (Alexander and Mindess, 2005). Metamorphic rocks such
as Quartzite make important contributions to the production of concrete aggregates.
In General, aggregates from different origins and types can be used for making concrete
while some are more suitable for a specific application (e.g. porous material for light weight
or self-healing concrete). Moreover, some undesired minerals such as Mica can be found in
metamorphic or igneous rocks as well as sedimentary rocks.
Figure 2.3. shows rocks formed by each of the explained processes.

Figure 2.3. Typical rocks used as aggregate (a) Granite-igneous (b) Sandstone-sedimentary (c)
Quartzite-metamorphic.
As it was mentioned above, the aggregates can be classified as natural or crushed based on
their source and production technique. Natural aggregates can be obtained from variety of
sources e.g. pits, river banks, beaches or other quarries that require minimum extra effort and
cost in terms of granular material processing. Utilizing natural aggregates in mix design leads
to better performance of the concrete since they have smaller surface area (comparing to
crushed aggregate of the same size) and also their relatively spherical shape facilitate the flow
inside the aggregate structure. However, the negative environmental impact of using natural
aggregate is of concern which includes and not limited to atmospheric pollution, water
pollution, changes in water course, increase in settlement, changes in ecosystem, etc.
(Alexander and Mindess, 2005). Therefore, In Sweden and in some other parts of the world
there are regulations about the amount of natural aggregates that can be taken from the earth’s
crust.
An alternative to usage of natural aggregates is to crush the rocks to desired size fractions.
The crushing technique has a significant effect on quality and properties of the product. This
is especially important in case of crushed sand due to their influence on properties of concrete
in fresh state as a result of tendency of crushed sands to have elongated flaky shapes
(Lagerblad et al, 2013). These issues can be reduced by use of appropriate crushing
techniques, removal of chips and fines, choke and closed-circuit feeding, or other additional
processes that improve the quality of the product by different methods most of which are
based on gravity or centrifugal separation. (Alexander and Mindess, 2005; Neville, 1995;
Gram et. al, 2017)
2.3. Physical properties of aggregates
The section deals with commonly discussed and used physical properties of concrete
aggregates:
Porosity is defined as the internal pore volume as a proportion of the total volume of a solid,
its significance lies on the influence of porosity of aggregates on concrete density.
ܸ௣
‫=݌‬ (2.1)
ܸ௧
where Vp is the volume of internal pores and Vt is the total volume of the solid.
Absorption can be directly related to the porosity as porous materials can absorb water.
Absorption usually expressed as the ratio of changes in mass of an oven dried sample after
saturation to the mass of the saturated surface dried sample. Absorption of the aggregates can
affect concrete strength, workability and durability.
Moisture state defines the condition of aggregates in terms of available moisture in the
particles. Oven dry is a state where the aggregates are conditioned to temperature of 100 to
110o C to evaporated the available moist in the particles, air dry is when the aggregate has
reached equilibrium with the ambient air; saturated surface dry (SSD) is a condition when the
aggregates themselves are saturated but the surface has been dried. The moisture state is
important as it influences the total water content of the mix affecting strength and workability.
Density of a solid is defined as the ratio of its mass to the volume it occupies.
‫ܯ‬ௌ
ߩ= (2.2)
ܸௌ
where Ms and Vs are the mass and the volume of the solid, respectively.
There are several measures of density among which bulk density and relative density are the
most important ones. Bulk density is the density of the material in bulk granular form.
Relative density is the ratio of the density of substance to the density of a given reference
material (usually water). Logically, The aggregates density affects the proportioning of
concrete ingredients.
Void content applies to a collection of particles as the particles do not fit together perfectly
thus leaving voids between them. The volume of the voids is of great importance in concrete
industry as the voids should be filled with paste or matrix. The void content is a function of
particle size, shape, grading and packing properties.
ܸ௏
݁= (2.3)
ܸௌ
where Vv is the volume of the external granular voids and Vs is the total volume of the bulk
sample.
Particle shape is a complex function of aggregates’ formation conditions, the mineralogical

composition, particle size and method of production in case of crushed material. There is
much confusion on definition of various shape parameters and there are no commonly
accepted methods for their measurement (Kwan and Mora, 2002).However, particle shape is
considered as the most important factor influencing water demand and consequently
workability of a concrete mix. Particle shape can be defined by the help of some other
measures such as angularity, flakiness, sphericity and so on.
Flakiness ratio is defined as the thickness-breadth ratio which can be estimated using the
thickness guage. Elongation ratio is defined as the breadth-length ratio obtained from image
analysis or based on ASTM D4791-10. Sphericity is usually defined as the ratio of the surface
area of a sphere having the same volume as the particle to the actual surface area of the
particle. Convexity is a measure of overall roundness, a rounded particle should not contain
too many concave corners and a particle containing many concave corners is not rounded.
Surface texture is difficult to define and is usually being described in quality terms like rough
or smooth. The surface texture depends on pore structure, texture of the parent rock,
production method, amount of wear on the particles, and grain size. Surface texture influences
the surface area and the inter-particle friction of aggregates.
Grading refers to the size distribution of the aggregates, well-shaped well graded aggregates
produce more workable concrete. Grading, particularly grading of the fines, is an essential
factor in production of concrete as it influences the properties of concrete in fresh state.
Grading also affects void content, water demand and available specific surface area of
particles. (Alexander and Mindess, 2005; Neville, 1995).
Mix design approaches 13
3. MIX DESIGN APPROACHES
Mix design can be defined as “the process of choosing the ingredient of concrete and
determining their quantities with the object of producing as economically as possible of
certain concrete of certain minimum properties such as consistence, strength, and durability”
(Neville, 1995).
During past decades several attempts were made on formulating the mix design. One of the
most common approaches to mix design is based on the concept of particle packing density
first introduced by Feret (1892). The aim of particle packing density concept is to minimize
the inter-particle voids between the constituents of concrete in order to reduce the paste
demand. Packing density is the ratio of the volume of solids to the bulk volume of the solid
particles (De Larrard, 1999). Several researchers tried to develop Feret’s model further,
research conducted by Powers (1968) on the interaction of different components in mixture
gave another dimension to the knowledge of particle packing. In 1999, De Larrard (De
Larrard, 1999) introduced a multi component packing model that included interaction
between the particles. With the introduction of computers, packing softwares were developed
to do more detailed and complex calculations for estimation of particle packing, e.g. Europack
(Idorn, 1995), Rene-LCPC (lcpc.fr), and 4C packing (dti.dk).
Another approach to maximizing the packing is by means of developing a method to design

an ideal particle size distribution curve as in the works of Fuller amd Thompson (1907) and
Andreasen and Andersen (1930). Figure 3.1. (adopted form Kumar V and Santhanam, 2003)
shows some of packing models and their relation to each other. The mix design approaches
are listed in three main categories i.e. ideal curves, models based on packing theory and also
layer theories. The ideal curve approach can be also categorized under packing theory as they
eventually change the sieving curve to result in a higher packing. However, since the effect on
the packing density is implemented indirectly, the ideal curve approaches were listed
separated from packing theory branch. Moreover, some of the models such as 4C works based
on combination of ideal curve and packing theory approaches. Under the Multi-phase
approach branch, models that consider concrete as separated liquid and solid phases are listed.
More information on each category can be found in Sections 3.1 to 3.3.
14 Mix design approaches
Figure 3.1. Mix design approaches

3.1. Ideal size distribution curve
As mentioned, the idea of composing an ideal particle size distribution (PSD) curve for
aggregates was first introduced by Fuller and Thompson (1907). The concept behind ideal
size distribution curve lies in reducing void content of the packed aggregates, since at constant
water content, mixes with higher packing will have more available water as lubricant between
particles. The Fuller curve deals with continuous grading curve ranging from 250 ȝm to the
maximum size. As a result, Fuller curve is less suitable for self-compacting concrete (SCC)
and high strength concrete (HSC) ZKHUH ILQHV ZLWK VL]H RI OHVV WKDQ ȝm is commonly
used and also mixes made with gap-graded crushed aggregates, as an example Fuller curve
cannot be used for a material ranging from 4 to16 mm since the diameter of the small class is
not considered by the approach. In that sense, grading by Fuller curve represents a special
case. In a more general case, packing equation were derived by Andreasen and Andersen
(1930), it is suggested that to obtain optimum packing (less voids), the particle size
distribution should be calculated based on the following equation.
‫ܦ‬
ܲ(‫( = )ܦ‬ )௤ (3.1)
‫ܦ‬௠௔௫
where P is the fraction that can pass the sieve with opening D, Dmax is maximum particle size
of the mix and parameter q has value ranging from 0 to 1. According to (Andreasen and
Andersen, 1930), value of 0.37 for q will result in optimum packing while the Fuller curve
can be obtained by using q=0.5.
The A&A curve (Andreasen and Andersen, 1930) was later on modified to account for the
minimum particle size (Dmin) in the mix by Funk and Dinger (1994). The modified A&A
curve can be calculated based on the following equation:
௤
‫ܦ‬௤ െ ‫ܦ‬௠௜௡
ܲ(‫= )ܦ‬ ௤ ௤ (3.2)
‫ܦ‬௠௔௫ െ ‫ܦ‬௠௜௡
It should be mentioned that the value of q is commonly decided based on experience and/or
on qualitative empirically obtained values judged by the type of aggregates - crushed or
natural- (Gram et. al., 2017) or their place of mother rock origin (e.g. a certain q value for
aggregates from Norway).
Figure 3.2. shows ideal curves obtained from Fuller, A&A and modified A&A approaches.
Figure 3.2. Ideal curves according to Fuller, Andreasen & Andersen, and modified A&A (Funk and
Dinger). For maximum and minimum SDUWLFOH VL]H RI PP DQG ȝP respectively. (Fennis and
Walraven, 2012)
While ideal size distribution are practical and easy to use, they don’t take into account the
interaction between the particles and other factors that affect the packing e.g. shape, surface
texture, etc.
3.2. Particle packing theory
In order to produce an optimized concrete product, it is vital to select an acceptably accurate

packing model for estimation of packing density. As a principal, particle packing models aim
to define the size distribution curve in a way that results in high packing thus leaving less
volume of voids to be filled with paste. However, it should be mentioned that maximizing
packing density does not necessarily guarantee a workable mixture (Kwan and Mora, 2002).
This is considered for example in 4C user manual (2009) by suggesting that for making a
workable concrete, the maximum packing density should be avoided. While it is known that
the optimum packing is not equal to maximum packing, the approach to deciding an optimum
packing is not clear.
The packing density of a granular mix is defined as the solid volume in a unit total volume
which is equal to ratio of bulk density to the relative density of the solids.
ߩ௕
ߙ= (3.3)
ߩ௦
wKHUH Į LV SDFNLQJ GHQVLW\ 7KH packing density can also be defined in relation to void
content:
ߙ =1െ݁ (3.4)
It should be mentioned that almost all of the particle packing models are fundamentally rely
on the following geometry based equations (De Larrard, 1999; Furnas, 1929):
ߚଵ
ߙ௧ = Large particles are dominant (1) (3.5)
1 െ ‫ݕ‬ଶ
1
ߙ௧ = ‫ݕ‬ଶ Small particles are dominant (2) (3.6)
‫ݕ‬ଵ + ( )
ߚଶ
wKHUHĮt LVFDOFXODWHGSDFNLQJGHQVLW\RIDPL[WXUHȕ1 is the packing density of the larger size

FODVV ȕ2 is the packing density of the small size class 2,y1 is the volume fraction of size
class 1 and y2 is the volume fraction of size class 2 where for two size classes y1+y2=1.
Eq. (3.5) is valid in case that the amount of large particles is dominating the particle structure
where small particles are filling the voids between the larger particles, on the other hand, Eq.
(3.6) deals with a situation where the amount of small particles is dominating and large
particles are embedded in a matrix of small particles (De Larrard, 1999).
This study deals with the following particle packing models: Toufar (Toufar et al, 1976),
CPM (De Larrard, 1999), and 4C models (Pade et al, 2009; 4C user’s manual, 2009).
3.2.1. Toufar Model

Toufar model is developed for calculating the packing of binary mixes where the diameter
ratio of the class falls in the range of 0.22 < d1/d2 < 1.0 (Toufar et al, 1976). According to the
model, for diameter ratio larger than 0.22, the smaller particles (d1) will be too large to fit in
the voids between larger particles (d2). As a result, factor kd was introduced which relates the
packing density to the diameter ratio of the two particle classes. In addition, Toufar model
assumes that each of the fine particles is placed between exactly four of the coarse particles
which is utilized in factor ks (Fennis, 2011). The packing density, ߙ௧ , is described by
following equations (Goltermann et al. 1997):
1
ߙ௧ =
‫ݕ‬ଵ ‫ݕ‬ଶ 1 (3.7)
ߙଵ + ߙଶ െ ‫ݕ‬ଶ ቀߙଶ െ 1ቁ ݇ௗ ݇௦
݀ଶ െ ݀ଵ
݇ௗ = (3.8)
݀ଵ + ݀ଶ
1 + 4‫ݔ‬
݇௦ = 1 െ (3.9)
(1 + ‫)ݔ‬ସ
‫ݕ‬ଵ ߙଶ
‫=ݔ‬ (3.10)
‫ݕ‬ଶ ߙଵ (1 െ ߙଶ )
wKHUHĮt is the calculated packing, yi LVWKHFODVVLVKDUHRIYROXPH DQG Įi is the packing of

class i.
The Toufar model was later modified by (Golterman et al., 1997) since comparisons with test
results showed that the packing density of a sample of coarse particles does not increase when
a small amount of fine particles is added to the mixture as a result of the unrealistic
assumption that each fine particles is placed in between four coarse particles. Here coarse
particles are considered relative to the fines size and not a certain range e.g. for a binary
mixture the constitute with larger diameter is considered as coarse. The behaviour was
corrected by modifying the ks factor:
0.3881‫ݔ‬
݇௦ = For x < 0.4753 (3.11)
0.4753
1 + 4‫ݔ‬
݇௦ = 1 െ For x ൒ 0.4753 (3.12)
(1 + ‫)ݔ‬ସ
where x is calculated based on Eq. (3.10).
Toufar and modified Toufar models are fairly easy to use. However, estimation of packing
density by these models will lose accuracy in multicomponent systems and tends to
underestimate the packing density.
3.2.2. DTI 4C model

DTI 4C is a computer program developed by Danish Technological Institute and is based on
the Linear Packing Density Model (LPDM), Figure 3.1 and Eq. (3.5) and (3.6). The factors
that are taken into account include particle size distribution of the mixture and packing
densities of each size class. The Linear Packing Density Model (LPDM) is able to predict the
packing density for several particle classes, which makes the model suitable for real concrete
mixtures (Jones et al, 2002). The accuracy of the model depends on interaction formulas
which are relations derived from the packing density of two-component mixtures (Pade et. al,
2009; Glavind and Pedersen, 1999).
4C packing model uses ȝ-value as an interaction factor which indicates possible maximum
ratio (size) between small and large particles without the smaller particle interfering with the
packing of the larger particles. It should be mentioned that ȝis an empirical value and needs
to be calibrated based on the laboratory data obtained from particle packing test. While 4C
model requires values from eigen packing (hard packing), it was found that the calculations
based on loose packing also leads to acceptable results (Ghasemi and Emborg, 2014).
3.2.3. Compressible packing model (CPM)

The CPM is one the most comprehensive multicomponent mix design models which utilize a
refined version of LPDM for grain mixtures. The model calculates the packing density based
on the concept of virtual packing. Virtual packing density is defined as the calculated packing
value in an ideal scenario where all the particles are assumed to be placed in the best possible
position which results in minimum void content. The virtual packing density is higher than the
real achievable packing density. The actual packing density can be derived from virtual
packing by means of compaction index (K) which depends on the compaction energy used in
packing method. K index was suggested to have the value of 4.1, 4.5, 4.75, and 9 for loose
packing, rodding, vibration and vibration + compression respectively (De Larrard, 1999).
Value of 12.5 for the compaction index was suggested by other researchers for Modified CPM
(Jones et. al, 2002).Figure 3.3 shows the effect of compaction index on packing density of a
mixture.
a)
b)
Figure 3.3. (a) Effect of K value on compaction of aggregates where actual packing densities of two
classes assumed to be constant. (b) Variation of K vs. packing density (adopted from Glavind et al,
1993)
LPDM can be considered as a special case of CPM for which the compaction index K tends to
infinity, as shown in Figure 3.2.a by a solid line.
In addition, CPM considers the interaction of components of the mixture based on the
concepts of loosening effect and wall effect. If a smaller grain is inserted in the porosity of a
coarse grain packing, coarse grains being dominant, and if there are no more spaces for the
fine grains to fit inside the voids, there will be a local decrease of volume of the dominant
class. In other word, the finer grains will push the coarse grains apart to make room for fines
to fit (loosening effect). On the other hand, when some isolated coarse grains are immersed in
a sea of fine grains, fine grains being dominant; there is a further amount of voids in the
packing in the interface vicinity (wall effect). Figure 3.4 illustrates the concepts of loosening
and wall effect. The effects can be calculated for multicomponent mixtures by (De Larrard,
1999):
݀௝ ଵ.଴ଶ
ܽ௜௝ = ඨ1 െ (1 െ ) (3.13)
݀௜
݀௜ ଵ.ହ଴
ܾ௜௝ = 1 െ (1 െ ) (3.14)
݀௝
where coefficients aij and bij represent the loosening and wall effect respectively, di is
diameter of dominant particle size class i and dj is diameter of particle class j.
a) b)
Figure 3.4. (a) Loosening effect exerted by a fine grain in a coarse grain packing (b) Wall effect
exerted by a coarse grain on a fine grain packing. (De Larrard, 1999)
For a general case of CPM and for a polydisperse mix, the virtual packing of a mixture, ߚ௧௜
containing n size classes with category i being dominant is expressed as:
ߚ௜
ߚ௧௜ =
1 ܽ௜௝ ߚ௜ (3.15)
1 െ σ௜ିଵ ௡
௝ୀଵ[1 െ ߚ௜ + ܾ௜௝ ߚ௜ (1 െ ߚ )]‫ݕ‬௝ െ σ௝ୀ௜ାଵ ൤1 െ ߚ ൨ ‫ݕ‬௝
௝ ௝
ߚ௜
ߙ௜ = (3.16)
1
1+‫ܭ‬
wKHUHȕti is calculated virtual packing of a mixture when size class i LVGRPLQDQWDQGȕi DQGȕj
are virtual packing densities of size class i and j. For a monosized particle class ߚ௜ can be
determined by Eq. (3.16) from the experimentally determined packing density ߙ௜ . aij and bij
should be calculated based on Eq. (3.13) and (3.14). It should be mentioned that as K value
tends to infinity, the real packing density Įt EHFRPHVFORVHUWRWKHYLUWXDOSDFNLQJGHQVLW\ȕt.
3DFNLQJGHQVLW\Įt is determined indirectly based on:
௡ ௡
‫ݕ‬௜ /ߚ௜
‫ = ܭ‬෍ ‫ܭ‬௜ = ෍ (3.17)
1 1
௜ୀଵ ௜ୀଵ െ
ߙ௧ ߚ௧௜
CPM is more advanced than the previously mentioned models but the complexity of the
model and the number of input data make it more difficult to use.
3.3. Multi-phase approaches
In the current document, multi-phase mix design approaches is defined as the models that
consider concrete as a two phased mixture, i.e. the models consider concrete as a combination
of a solid phase (particles) and a liquid phase (water/paste).
3.3.1. Particle-matrix model

The workability of concrete is governed by the inherent properties of the constituents,
proportioning and the way the constituents are interacting with each other physically and
chemically. Particle-matrix model (PMM) considers concretes in two separate phases, fluid
material and a friction material. Based on this logic the matrix is considered as all of the
particles less than 0.125 mm including cement, fines and possible chemical additives while
particle phase is defined as all of the particles larger than 0.125 mm. The model is particularly
suitable for mixes where matrix phase is dominant e.g. self-compacting and high performance
concrete (Smeplass and Mortsell, 2001; Reknes, 2001).
The main difficulty here is to define the properties of the phases and to model the effect of
these phases on each other. The basic concept of the model is shown in Figure 3.5.
The approach relies on single parameter characterization of each phase (Mortsell et al., 1996)
(Samarakoon et al., 2015):
- The flow resistance ratio of the matrix

- The air voids modulus of the particles
Amount of matrix
Workability
Particle properties Matrix properties
Figure 3.5. Particle matrix model concept for concrete (Bartos et al, 2004)
3.3.1.2. The flow resistance ratio

The flow resistance ratio is calculated based on the results obtained from a modification of
Marsh cone test called FlowCyl. The apparatus consist of a vertical cylindrical steel tube with
a bottom outlet formed as a cone ending in a narrow nozzle and an electronic scale connected
to a data logger where the flow properties of the material are characterized by the
accumulated flow through the nozzle. The flow resistance ratio represents the difference in
accumulation flow between the test material and an ideal fluid flowing through the FlowCyl
and is defined as the ratio between the area under the loss curve (Ft) and the area under the
curve for the ideal fluid without any loss (Fi) (Mortsell et al., 1996). see Figure 3.6. It should
be mentioned that an ideal fluid is defined as an uncompressible nonviscous liquid which does
not actually exist in nature and is commonly used for fluid flow problems (Landau and
Lifshitz, 1987).
Flow rate
Figure 3.6. Typical FlowCyl data for a matrix showing curves for an ideal fluid, measured points for
actual matrix and calculated curve for loss. (Mortsell et al., 1996)
The flow resistance ratio can be calculated based on the following equations:
‫ܨ‬௧
ߣொ = (3.18)
‫ܨ‬௜
݇௖ ܿ ݇௦ ‫݇ ݏ‬௙ ݂ ௡
ߣொ = ݇ ( + + ) (3.19)
‫ݓ‬ ܿ ܿ
where k, kc, ks, kf and n are constants found by regression analysis of test data, c/w is cement
water ratio by weight, s/c is silica fume to cement ratio by weight and f/c is filler to cement
ratio by weight.
3.3.1.3. The air voids modulus

The air void modulus is defined as the air void space ratio of the fines (ranging from 0.125
mm to 4mm fractions) and course (>4 mm) portion of the particle system. It should be noted
that the effect of properties of fine fractions portion is greater than the effect of coarse
aggregates properties. Hence, the fineness modulus (Fm) of the fine/coarse aggregates is
utilized as correction factor when defining the air voids modulus. A preliminary definition of
the air voids modulus is (Mortsell et al., 1996):
‫ݒ = ݉ܪ‬௦௔ (‫ݏܪ‬/(‫݉ܨ‬௦ )଴.ହ + ܶ‫ )ݏ‬+ ‫ݒ‬௖௢ (‫݌ܪ‬/(‫݉ܨ‬௣ )଴.ହ + ܶ‫)݌‬ (3.20)
where vsa and vco are volume fraction of sand (0.125 – 4 mm) and volume fraction of coarse
aggregates (>4 mm) respectively. Hs is air voids ratio in the sand (vol %), Hp is air voids ratio
in the coarse aggregate, Ts is an aggregate parameter for the sand, Tp is an aggregate
parameter for the coarse aggregate, Fms and Fmp are the fineness modulus for sand and coarse
aggregates respectively. The aggregate parameters can be determined in different ways based
on different conditions (Mortsell et al., 1996).
The physical interpretation of air voids modulus (Hm) according to the authors (Mortsell et
al., 1999) is that it is equal to the paste volume (vol %) when the mix is changing from no
slump to a small but measurable slump.
3.3.1.4. Workability function for concrete

The workability of the concrete can be measured using slump or flow table test and it can also
be calculated based on the PMM as a function of the flow resistance of the matrix, the air
voids modulus of the particle phase and the volume fraction of the matrix. The correlation
between workability and volume fraction of the fluid phase is assumed to be S-shaped utilized
by using the Tanh-function. Workability function Kp can be defined as:
(݊ െ ݉)(݄ܶܽ݊(2ߙ௞ (‫ ݌ܨ‬െ ‫)݉ܪ‬/100 െ 1) + 1

‫= ݌ܭ‬ +݉ (3.21)
2
where m and n are lower and uppeU DV\PSWRWH YDOXHV UHVSHFWLYHO\ Įk is expressed as the
steepness of Kp function, Fp is volume fraction of fluid phase and Hm is the air void
PRGXOXV7KHIDFWRUĮk LVDIXQFWLRQRIWKHIORZUHVLVWDQFHUDWLRȜQ :
ߙ௞ = ‫ ܣ‬. ݁ ି஻ఒೂ (3.22)
where A and B are constants found by regression analysis.

The main application of the model is to adjust concrete mix design, from a workability point
of view, when properties of proportions of the constituent materials are changed (Mortsell et
al., 1996).
3.3.2. Layer theories

A different approach to mix design roots in excess paste layer theory first proposed by
Kennedy (1940). The paste and water layer theories were originally developed for self-
compacting concrete (Reschke et al., 1998; Fraaij and Rooij, 2008; Grünewald, 2004).
According to these theories, the cement paste should not only fill the void inside the
aggregates structure but also surround all the particles with a thin layer of paste to fulfil
workability requirements. In turn, the required amount of paste or water depends directly on
the specific surface area of the particles, as increase in the surface area of the particles will
reduce the thickness of surrounding layer for constant amount of water.
In excess paste layer theory approach, the packing density of the aggregates is measured in
order to determine the amount of cement paste necessary to fill up the voids, the remaining
paste added in the mixture for allowing workability is counted as excess paste. The volume of
excess paste, Vep , can be calculated using Eq. 3.7.
1 െ ߙ௔௚௚
ܸୣ୮ = ܸୡ୮ െ ܸ௔௚௚ ( ) (3.23)
ߙ௔௚௚
where Vcp and Vagg are volumes of cement paste and aggregate respectively and Įagg is the
packing density of the aggregate which can be determined experimentally. Once the volume
of excess paste is known, the thickness of paste layer tep can be calculated by dividing the
volume of excess paste to the specific surface area of the particles. It should be mentioned that
there are no universally accepted and accurate method for measuring the specific surface area
of particle and available methods involve high cost and complex devices. However, it is
possible to mathematically calculate the thickness of the surrounding layer based on the
following assumptions (Midorikawa, et al., 2009):
1. Particles in each size group are assumed to be spherical

2. The thickness of the excess paste layer is constant for different sizes of
particles.
Based on the above conditions, the volume of excess paste for spherical particles can be
calculated by utilizing the size distribution curve of the material.
௡
1 ܸ௜
ܸୣ୮ = ෍[ ߨ {(݀௜ + 2‫ݐ‬௘௣ )ଷ െ ݀௜ଷ } ] (3.24)
6 1 ଷ
௜ୀଵ
6 ߨ݀ ௜
where tep is the paste layer thickness, di is the diameter of particles in fraction i and Vi is the
volume share of fraction i.
In case that the packing density of the total particle structure (including cement) was used as
value forȘagg in Eq. (3.23) instead of the packing density of the aggregate structure, the
flowability arises from the excess amount of water present in the particle mixture as in the
excess water layer theory (Fennis, 2011) see Figure 3.7.
Figure 3.7. Volume of water, divided into excess water and void filling water, within a concrete
mixture in a unit volume (Midorikawa et al, 2009).
In a sense, excess water layer theory and excess paste layer theory follow a similar principal.
Based on excess water layer theory, the water is partly used to fill the voids in the particle
structure and the rest of the water form a layer with thickness of tew. The thickness of the
water film can be calculated by: (Krell et al, 1985).
1 െ ߙ௧
ܸ௘௪ ܸ௪ െ ܸ௣ ( )
ߙ௧ (3.25)
‫ݐ‬ୣ୵ = =
‫ܣ‬௣ ݉௣ ‫ܣ‬௣ ݉௣
where:
Vew is the volume of excess water, Ap is surface area of particles, mp is the mass of the
particles in a mixture, Vw is the volume of the water in the mixture, Vs is the volume of all
particles in a mixture and ߙ௧ is the packing density of a mixture. Research conducted by
Teichmann (2008) found a good relation between viscosity measurements and excess water
layer thickness.
The issue with water layer and paste layer theories roots in lack of a comprehensive method to
measure specific surface area of the particles. Current common methods of measuring the
specific surface area, including the Blaine test and BET, involve high instrumental complexity
and are costly. It is also possible to estimate the specific surface area based on the particle size
distribution curve information and the assumption of spherical shape for the particles.
However, particle shapes are far from being spherical due to 3D randomness in their shape,
related to the origin of the aggregates, and production method. This is particularly true for
crushed aggregate.
It should be noted that the total surface area of a set of aggregates is chiefly governed by the
fine aggregate fraction according to square-cube law (Section 3.3.2.1). Assuming that all
particles were spherical in shape, the surface area can be calculated based on grading curves
according to the following equation (McCabe et. al., 1993):
௡
݉௜
ܽ௦௣௛ = 6 ෍ (3.26)
݀ҧ௜ . ߩ௦
௜ୀଵ
where ݉௜ is the mass of a grain fraction i, being the mass percentage of the fraction between
di and di+1. ݀ҧ௜ is the mean diameter of fraction i and i+1. ߩ௦ is the relative density of the
particles.
Appended papers III and IV, focus on the possibility of substituting ideal polyhedron shape
for spheres which can be related to the actual shape of aggregate with better accuracy. The
details and the method are presented in the following headings:
3.3.2.1. Square-cube law

The square-cube law was first described by Galilei in 17th century (Galilei and Drake, 1946)
and it defines a mathematical principle which describes the relationship between surface area
and volume related to changes in size. When an object undergoes a proportional increase in
size, its volume grows faster than its surface area. The effect of square-cube law becomes
especially significant for calculation of specific surface area of finer particles namely powders
and cement i.e. for a given mass of aggregate, the surface area increases with reducing
particle size. Figure 3.8 shows the effect of size changes of platonic solids on the ratio
between surface area and volume.
400
350
300
Surface Area
250
200
150
100
50
0
0 100 200 300 400 500
Volume
Figure 3.8. Surface area against volume of the platonic solids and a sphere (see Table 3.1).
Eq. 3.26 can be written in its general form where the ratio of surface area to the volume
defines the shape:
௡
ܵܵ‫ܣ‬௜ . ݉௜
ܽ௣௢௟௬ = ෍ (3.27)
ܸ௜ . ߩ௦
௜ୀଵ
where SSAi/Vi is the surface area to volume ratio of fraction i and is related to the shape as
shown in Table 3.1.
Table 3.1: Platonic solids used in the calculation of specific surface area.
Shape Surface Area Volume SSA/V
Tetrahedron ξ3ܽ ଶ
ξ2ܽ ଷ 14.697
12 ܽ
Cube 6ܽଶ ܽଷ 6
ܽ
Octahedron 2ξ3ܽଶ 1 7.348
ξ2ܽଷ
3 ܽ
Dodecahedron య 1 2.694
ට25 + 10ξ5ܽଶ (15 + 7ξ5)ܽଷ
4 ܽ
Icosahedron 5ξ3ܽଶ 5 3.970
(3 + ξ5)ܽଷ
12 ܽ
Sphere 4ߨܽଶ 4ߨܽଷ 3
3 ܽ
Substituting spheres with the platonic solids will not only change the calculated volume and
surface area but also affects the pace of growth in SSA/Volume ratio according to square-
cube law.
3.3.2.2. Equivalent polyhedron shape

In order to calculate the specific surface area of spherical particles, the mean diameter of the
particle sizes di and di+1 of fraction i, as the characteristic particle size, is required. The mean
diameter di can be calculated using either arithmetic mean or geometric mean using Eq. (3.28)
and (3.29) respectively.
݀௜ +݀௜ାଵ
݀௜,௔௥௜௧௛ = (3.28)
2
݀௜,௚௘௢ = ට݀௜ ଶ + ݀௜ାଵ ଶ (3.29)
When replacing the spheres with other polyhedron shape, the length of the sides of
polyhedrons should be calculated based on the mean diameter of the spheres. The length of
the sides of polyhedrons can be computed based on the assumptions of how the shapes are
defined in respect to geometric properties using the concepts of circumshpere and midsphere,
see Figure 3.9. In geometry, a circumscribed sphere or circumsphere of a polyhedron is
a sphere that contains the polyhedron and touches each of the polyhedron's vertices.
Midsphere is defined as a sphere that touches all of the polyhedron edges. The midsphere
does not necessarily pass through the midpoints of the edges, but is rather only tangent to the
edges at same point along their lengths (Cundy and Rollett, 1961).
It is also possible to calculate the side length of the polyhedrons with the assumption that the
polyhedrons have the same volume as the sphere it replaced (volumetric equivalency).
Circumsphere -Cube Midsphere - Cube
Equivalent volume
Figure 3.9. Circumsphere, Midsphere and volume equivalency of a cube, edge length of a.
The geometric equivalent edge lengths of the polyhedrons, a (ac for circumsphered and am for
midsphere edge length), can be calculated by equations listed in Table 3.2. Median radius of
equivalent spheres, r, can be calculated by either Eq. (3.28) or (3.29).
Table 3.2: Edge lengths (Circumsohered ac and Midsphered am) of polyhedron.

Shape Circumsphered edge Midsphered edge length
length (ac) (am)
4‫ݎ‬ 4‫ݎ‬
Tetrahedron ξ6 ξ2
2‫ݎ‬ 2‫ݎ‬
Cube
ξ3 ξ2
2‫ݎ‬
Octahedron 2‫ݎ‬
ξ2
4‫ݎ‬ 4‫ݎ‬
Dodecahedron
ξ3(1 + ξ5) (3 + ξ5)
4‫ݎ‬ 4‫ݎ‬
Icosahedron
ඥ10 + 2 × ξ5 (1 + ξ5)
3.4. Concluding remarks
Different approaches in mix design were briefly discussed in the current chapter. It should be
mentioned that using the models will not necessarily lead to production of acceptable concrete
in term of required specification of the final product and the recipes usually need fine tuning
and corrections which is often based on trial batches of concrete. Moreover, some of the
models may be useable for a specific region or certain type of aggregates, especially the
models that are focusing on ideal size distribution curves.
As for the more advanced models like CPM, the number of required input data and the tests
that are needed to be conducted renders the model as a complex and time consuming
approach.
In addition, the models that are based on specific surface area suffer from inaccuracy and
complexity of conducting specific surface area measurements as there are no commonly
accepted methods for the tests.
A simple fairly accurate mix design model could potentially reduce the number of tests that
are needed for using the model and also the number of trail batches can be lowered by a more
accurate estimation of the properties of the final product.
30 Methods
4. METHODS
The current chapter deals with the test methods utilized in the laboratory for obtaining the
input data to the models in addition to the methods that were used in calculations of the
models.
4.1. Test methods
4.1.1. Sampling of aggregates

It is essential to conduct tests on the aggregates to obtain their physical and mechanical
properties. Since the aggregates are used in vast quantities in production of concrete; it is of
great importance to take smaller samples of the aggregate in an appropriate and represented
way. The method of sampling can cause variations in samples comparing to the pile it was
taken from such as segregation in large buckets and changes in moisture content in different
depths of soil. ASTM D75 includes instructions on appropriate sampling depending on how
and where the samples were taken from e.g. conveyor belt, stockpile etc. In the current study,
the sampling was done in the laboratory taken from large bags. The samples were taken from
different depths of the bags to avoid the problems that can occur because of segregation of the
materials on surface and the base edge. The samples were mixed to produce a composite
sample before further reducing of the size of the sample.
4.1.2. Sieve analysis

Sieve analysis is defined as the operation of dividing a sample of aggregates into certain
fractions, each containing of particles of similar size (Neville, 1995). Sieving can be
conducted in different ways; ASTM C136-14 describes the method for dry sieving of
aggregates which was used as the method for obtaining the particle size distribution curve
(PSD) of the materials used in Paper II, however, the test was not designed for accurate
determination of PSD fRU PDWHULDOV ILQHU WKDQ ȝP )RU ILQHU PDWHULDOV VLHYLQJ VKRXOG EH
done either by washing (ASTM C117) or by means of low angle laser light scattering
technique (LALLS). The LALLS technique was employed for obtaining the sieving curve of
the powders used in Paper IV.
4.1.3. Specific gravity

Specific gravity of soil solids is the ratio of the mass of a unit volume of a soil solids to the
mass of the same volume of gas free distilled water at 20°C and should be conducted
according to the instructions of ASTM D854-14 for soil solids that pass the 4.75 mm sieve by
Methods 31
means of a water Pycnometer. As for the retained soil on the 4.75 mm sieve ASTM C127
should be utilized. Specific gravity can be measured as absolute or apparent. The absolute
specific gravity refers to the volume of the solid material excluding all pores while apparent
specific gravity is calculated to include the impermeable pores but not the capillary ones. The
latter is commonly used in concrete industry and as an input to particle packing models.
4.1.4. Packing methods

The methods of packing can be classified based on the compaction energy used in the process
of packing. According to European standard EN-1097:3, loose packing can be achieved by
pouring the aggregates in a standard cylinder from the distance of maximum 50 mm.
American standard ASTM C29 deals with the process of measuring hard packing in which the
container is filled to one third of its volume, then the layer of material is compacted using
rodding with 25 strokes and the process continues with filling the next one third of the volume
and rodding as the same manner as the first layer and then the procedure continues for the
third layer.
Another procedure for packing the particles is suggested by De Larrard (1999), in vibration +
compaction method, the aggregates will be poured in a cylinder that is closed with a 20 kg
steel piston and the whole setting will be put on a vibration table and shaken by following
vibration sequences: 2 min,40 s and 1 min at amplitudes 0.4 mm, 0.2 mm and 0.08 mm
respectively.Obviously, the choice of packing method affects the packing density measured
on the same aggregates (Johansson and Emborg, 2014).
The selection of the packing method has an obvious effect on the packing density. However,
among the discussed particle packing models, CPM is the only models that considers the
effect of packing method by means of compaction index K (see Section 3.2.3 and Figure 3.2.)
which is an empirical value that can be calibrated based on the method of packing. For 4C
model, the manual suggests using hard packing while Toufar model does not mention a
specific packing method. In the study conducted in Paper II, all of the packing density values
were obtained based on loose packing since the method is less energy and time consuming.
Moreover, loose packing data was chosen as input in all of the models so they can be treated
in the same way which makes the comparison of the models more sensible.
4.1.5. Surface area measurements

Measurement of surface area is of great importance in calculations based on layer theories and
has a significant effect on water requirement of the concrete mixtures. However, no
commonly accepted methods of measurement exist. The surface area is usually measured
either by Blaine test or BET test.
Blaine test derives surface area from the resistance to flow of air through a porous bed of a
powder and should be conducted based on instructions in ASTM C204-16 standard. The
method was originally introduced for measurement of surface area of cement; however, the
test has been used for determination of the measures of fineness of various materials. It should
32 Methods
be understood that the obtained values are relative to fineness of cement and are not absolute
measures of surface area.
BET test (Brunauer et. al., 1938) works with adsorption of gas molecules (usually Nitrogen)
on a solid surface which can be used for measurement of SSA. The results from BET depend
on the adsorbate molecule utilized and its adsorption cross section. BET can lead to results
that are several time larger than results from Blaine test for a given material since the values
from BET test include surface area of inner pores of the material as well as external surfaces.
Both Blaine and BET tests are fairly expensive and complex to perform and will eventually
lead to not so accurate results. Surface area can also be mathematically calculated based on
some basic assumptions and by using the particle size distribution. Calculation of specific
surface area is one of the main scopes of this thesis and is discussed more in detail in Papers
III and IV in Appendix 1 and also in Section 3.3.2.
4.2. Theoretical methods
4.2.1. Studied particle packing models

Among the several existing particle packing models, three were chosen for comparison of
calculated packing density values to conducted measurements in laboratory and the results
were reported in Paper II in Appendix 1. The chosen models included Modified Toufar, CPM
and 4C. Modified Toufar was chosen for its simplicity, CPM for its supposedly accuracy and
4C for its approach to utilizing both particle packing theory and ideal curve fitting.
The models are different in terms of calculation, required input data and application. Modified
Toufar was originally designed for binary mixtures and mentions no requirements about the
packing methods. CPM is capable of handling multi component mixtures including
aggregates and cement and requires packing density of each fraction of materials. Moreover,
the method of packing is considered in the model by utilizing index K. DTI 4C model can
calculate the packing density of three component mixture and requires hard packing density of
constitutes.
It should be mentioned that in order to treat the models in a similar way, loose packing data
was used as an input to all of the discussed models. Furthermore, the packing densities were
measured on two wide range fractions defined as fines (0-8 mm) and coarse (8-16 mm) and
not every sieve fraction as it is suggested by CPM.
4.2.2. Layer theories

Water and paste layer theories can be used as a basis for compiling a mix design. The main
difference between the water and paste layer theories roots in the way that the theories view
concrete mixtures, water layer theory considers concrete as a mixture of particles and water
while paste layer theory divides concrete in two phases of paste and aggregates. Water/paste
layer theories are less used in literature and practice comparing to particle packing theory
since the measurement of surface area is complex, expensive and in cases inaccurate. The
water/paste layer theories were not directly used in this document but rather explained to
Methods 33
emphasize the research gap concerning measurement of specific surface area (SSA).
Calculation of SSA based on particle size distribution curve and the conditions for replacing
assumed spherical shape of particles with other polyhedrons is described in Section 3.3.2 and
Papers III & IV.
34 Experimental results
5. EXPERIMENTAL RESULTS
The following chapter includes the results that were used in compiling the appended papers.
The data were obtained either in laboratory, by calculation or extracted from papers published
by other researchers.
Papers I & II focus on the subject of particle packing and the accuracy of three of the
developed models namely modified Toufar, CPM and 4C, see Sections 3.2 and 3.3. Papers III
& IV deal with estimation of specific surface area and its application in water/paste layer
theories as described in Section 3.3.2.
5.1. Packing models study
Figure 5.1. shows a typical result of packing density study (more results can be found in
Papers I & II) for binary mixes obtained from the laboratory experiment versus modified
Toufar model, 4C and CPM. The volume share of fines in the mixture was increased by 10%
in each step.
The interactions of particles were utilized in 4C software by means of ȝ-values (see Section
3.2.2). For sensitivity analysis purposes, three different ȝ-values of 0.07, 0.05, and 0.03 were
assigned to the model. Results from CPM were calculated by assigning compression index K
equal to 4.1 which represents loose packing.
The models’ estimation had a point to point deviation of 0.5 % to 5.8 % in packing density
comparing to the laboratory data.
Experimental results 35
0,8
Riksten Crushed 0-2 + Riksten Crushed 8-16
0,75
0,7
Packing Density
4C μ=0.07
4C μ=0.05
0,65
Lab
0,6 Toufar
CPM
0,55 4C μ=0.03
0,5
0 20 40 60 80 100
% Fine / total volume
Figure 5.1. Loose packing density of a binary mix from Riksten quarry in Sweden, CPM, modified
Toufar and 4C models ZLWKWKUHHGLIIHUHQWȝYDOXHVvs. lab data.
As it can be seen in Figure 5.1, using different models will lead to different proportions of
aggregates for any given mixture. The difference also exists in estimation of maximum
packing density depending on which model was used where CPM slightly overestimates
packing values comparing to modified Toufar. Moreover, while the models tend to agree on
the packing density as the finer material becomes dominant in the mixture (above 60% of fine
content), the difference between estimated packing density becomes larger on the coarse side
(left side) of the diagram.
It was concluded that generally in the tests performed, the accuracy of CPM and modified
Toufar increase as the ratio between fines mean diameter to the coarse mean diameter
decreases. On the contrary 4C shows better agreement with the test results for higher mean
size ratios. For more detailed results see Paper II.
Figure 5.2. shows the total comparison of differences between measured and calculated
packing densities. Considering all the data obtained in the laboratory, Modified Toufar
showed 1.72 % mean difference while the mean difference for CPM and 4C were 1.79 % and
1.84 % respectively. It should be mentioned that the comparison was done on 4C with ȝYDOXH
of 0.07 as it is suggested by 4C manual for aggregates originated in Scandinavia.
36 Experimental results
Figure 5.2. Comparison of difference between calculated and measured values based on loose packing
for Modified Toufar, CPM and 4C (see also Paper II).
5.2. Computation of specific surface area
Calculated specific surface areas and their corresponding Blaine values are presented in Table
5.1. and Figure 5.3. where the computations were based on the concept of midsphere and
arithmetic mean approach. The bold numbers in the following table marks the shape that gives
closest estimation to the Blaine value. More detailed data on the subject can be found in
Papers III & IV.
Table 5.1: Calculated specific surface area of the powders

Calculation based on Midsphere and Arithmetic mean
Material Specific surface area (cm2/g)
Blaine
CEMIII/B42.5 4500 8011 6543 5664 5433 5215 4625

Marble 4580 5540 4525 3917 3761 3644 3199
Limestone 4040 8936 7300 6319 6064 5812 5160
Quartz 2600 4685 3828 3313 3005 2995 2705
Experimental results 37
9000
Limestone
8000
CEM III
7000
Tetrahedron
Calculated SSA
Cube
6000
Octahedron
Marble
Dodecahedron
5000
Icosahedron
Quartz
Sphere
4000
Equality line
Linear (Equality line)
3000
2000
2000 2500 3000 3500 4000 4500 5000
Blaine value
Figure 5.3. Blaine values vs. calculated SSA for different polyhedrons based on Midsphere-arithmetic
mean assumption. (see Paper IV)
It should also be mentioned that the Blaine test is designed for measuring SSA of cement and
not necessarily any non-spherical powder, in other words Blaine value is a relative value and
not absolute for non-cement materials.
In the case of most of the studied powders, the calculated spherical values of SSA were close
to the Blaine value with some deviation for the particles with less spherical shape (e.g. Marble
powder). It should be emphasized again that the Blaine value is calculated based on the
assumption of spherical shape for the particle and so shows better agreement with the
calculations based on the same assumption, in other words the Blaine value is not
representative of the actual surface area, Hence to perform a more scientific comparison, the
specific surface area should be measured by means of a test that produces more realistic
measure of specific surface area (e.g. Microtomography, see Section 6.3).
38 Discussion and conclusions
6. DISCUSSION AND CONCLUSIONS

6.1. Accuracy of particle packing models
Particle packing theory is commonly used as a tool for concrete mix design and thus it is
necessary to study the accuracy of packing models. Suitability of the models was decided
based on the mean difference from the laboratory data and it was concluded that all of the
models can predict the packing density of the mixture with acceptable accuracy while some
models are able to estimate packing density of specific mixtures better comparing to the
others. The judgement of appropriateness in accuracy of the models comes from the fact that
experimental packing density of a given sample after correction for container wall effect
includes a deviation of 1 to 2% as a result of randomness in arrangement of the particles.
Hence, errors less than 2% for the models are considered acceptable. Moreover, it was found
that the deviation of the calculated data from laboratory measurements can be related to the
mean diameter ratio between the two classes (mean diameter of fine and coarse fractions).
Results of the studies imply that as the mean size ratio of fines over coarse material decreases
the accuracy of CPM and Modified Toufar increase. Contrariwise, 4C is more suitable with
higher mean size ratios.
While the descriptions on how the packing density is calculated are rather strictly
demonstrated, the concept of optimum packing and the approach to define it remains unclear.
6.2. Estimation of specific surface area
Water and paste layer theories can be used as a tool in mix design where it is necessary to
have an accurate measure of specific surface area (SSA) as an input in the models. However,
there are no commonly accepted approaches to measuring the specific surface area of
aggregates and in general the tests that can measure SSA are expensive and include highly
device related complexities.
An easier method for estimating the specific surface area is to calculate SSA derived from
grading curve and based on the assumption of spherical shape for the particles. However, the
mentioned method does not include the effect of square-cube law. In order to make the
calculation of SSA more accurate, a group of scenarios were studied where the assumption of
spherical shape was replaced by the assumption of platonic solid shape for the aggregates to
count for the angularity and also to implement the square-cube law in the equations. It was
found that the SSA can be much higher for certain aggregates if particles were assumed to be
Discussion and conclusions 39
cubic in shape instead of spherical. As an example, SSA of Quartz powder (see Table 5.1)
becomes 50% larger if the assumption of spherical shape is replaced by cubical shape. This
shows the significance of shape of aggregates and the corresponding surface area/volume
ratio growth (based on square-cube law) on computation of SSA and consequently the
calculations related to water or paste layer theories. Thus, for more angular elongated flaky
particles, the assumption of an angular platonic solid will lead to better estimation of SSA this
is especially true for the materials with high fine content.
6.3. Future work
In general, it is essential to estimate the water demand of a concrete mixture as it directly

influences some of the most important properties of the product e.g. strength and workability.
Water demand of a mixture is in turn related to both specific surface area of the particles since
the water should cover all of the surfaces of the particle and also packing density as less voids
means less water to fill the voids between the aggregates.
The major issue here is that while maximum packing leads to least voids to be filled with
water, the concrete produced on this basis usually turns out to be “harsh”. Hence, the concept
of optimum packing should be defined and a model for its estimation needs to be compiled.
Thus, continuation of the current research aims to explain the concepts of optimum packing
and zero slump concrete as foundations for a mix design model. The optimum packing in a
mixture can be defined as a high packing density that at the same time has low water
requirement. In turn, water requirement for a mixture is considered as all the water that is
needed to fill the voids in between the aggregates in addition to the amount of water needed
for covering the particles with a certain thickness which governs workability and puts the
mixture at the onset of flow (based on water layer theory). For this purpose it is necessary to
have a fairly accurate estimation of 1) SSA of particles and 2) the water requirement of
mixtures.
6.3.1. Estimation of SSA

As it is discussed in this thesis, proper estimation of surface area of particles is very important
for calculating the amount of water needed to cover all the surfaces of solid constituents.
Relating calculated SSA based on assumption of certain shape for the particle to actual
measured SSA can provide a basis for categorizing aggregates based on their shape.
An advantage of categorizing the shape of aggregates and relating them to a polyhedron is

that the SSA of that specific aggregate can be recalculated once the sieving curve changes.
Moreover, the data can be used for judging the suitability of the techniques that are used for
production of crushed aggregates. This is usually done by studying the packing density and
size distribution curve, while the importance of SSA is overlooked.
Among the methods for measuring SSA, X-ray microtomography can be considered as a
method with high accuracy. Microtomography is a radiographic imaging technique that can
produced 3D images of a material’s internal structure at a spatial resolution in micrometer
scale (Landis & Keane, 2010).
40 Discussion and conclusions
Accurate measurement of SSA, in combination with the data from particle packing can be
used to define optimum packing. By knowing SSA and packing densities, the optimum
packing – defined as a point in packing diagram that has high packing and low water
requirement - can be estimated by drawing the packing densities against water requirement of
a mix at any given point. Water requirement can be mathematically calculated based on the
void ratio in aggregates structure in addition to the water that surrounds the particles based on
water layer theory. see Figure 6.1. showing schematic diagrams of packing and water
requirement.
Figure 6.1. Schematic diagram of particle packing vs. water requirement in a binary system.
While microtomography produces results with high accuracy, it still suffers from the same
problems as Blaine and BET tests, namely inaccessibility, device related complexity and high
cost. Hence, a fairly accurate but simpler method for estimation of SSA is required.
6.3.2. Water requirement of mixtures

As mentioned before, it is possible to estimate the amount of water requirement of a mixture
based on the concept of water layer theory which is calculated by utilizing SSA and
assumption of certain thickness of the water layer. However, an empirical test should be
conducted in order to find the actual water requirement that puts a mixture on the onset of
flow for the sake of comparing calculation to actual laboratory data. This can be achieved by
using the Marquardt test or other commonly used tests developed for the purpose of
measuring water demand such as spread-flow test, Puntke or Vicat needle test.
Marquardt test characterizes the water demand as the percentage of water, adhesively bound
on particle surface (Hunger & Brouwers, 2009). The test can be performed on powders,
mortar or the entire concrete mixture. According to the test, the amount of water can be
determined by monitoring the power consumption of a mixer while water is being added to
the mixture in small amounts. A dry powder has a low shear resistance and therefore requires
less power for stirring the mix, as the water is being added to the mixture, agglomeration of
individual particles takes place which results in more consumption of power. Finally, the
Discussion and conclusions 41
power consumption reaches its maximum which reflects a point where all particles are wetted
and the particle themselves are connected via water films. Figure 6.2. shows a scheme of
development of power consumption in Marquardt test.
Figure 6.2. Schematic development of power consumption in Marquardt test. (Hunger & Brouwers,
2009).
Assuming the water layer thickness is constant for all of the particles, and in case calculated
layer thickness is acceptably close to the measured one, Marquardt test can be used as an
approximate test for measuring SSA.
Once the amount of water that is required to put a mix on the onset of flow is known, the data
can be used further for starting point of a mix design approach.
42 References
References
Abrams, D. A. (1919). Design of concrete mixtures (Vol. 1). Structural Materials Research
Laboratory, Lewis Institute.
Addis, B.J. and Owens, G. (2001) Fulton’s concrete technology, 8th edition, Midrand: Cement
and Concrete Institute.
Alexander, M., & Mindess, S. (2005). Aggregates in concrete. Taylor & Francis group.
Andreasen, A. M., & Andersen, J. (1930). Relation between grain size and interstitial space in
products of unconsolidated granules. Kolloid-Zeitschrift,50(3), 217-228.
Bartos, P. J., Cleland, D. J., & Marrs, D. L. (Eds.). (2004). Production methods and
workability of concrete (Vol. 32). CRC Press.
Brouwers, H. J. H., & Radix, H. J. (2005). Self-compacting concrete: theoretical and

experimental study. Cement and Concrete Research, 35(11), 2116-2136.
Brunauer, S., Emmett, P. H., & Teller, E. (1938). Adsorption of gases in multimolecular
layers. Journal of the American chemical society, 60(2), 309-319.
Cundy, H. M., & Rollett, A. P. (1961) Mathematical Models. Oxford university press.
De Larrard, F. (1999). Concrete mixture proportioning: a scientific approach. CRC Press.
Fennis, S. A. A. M. (2011). Design of ecological concrete by particle packing optimization.

TU Delft, Delft University of Technology.
Fennis, S. A., & Walraven, J. C. (2012). Using particle packing technology for sustainable
concrete mixture design. Heron, 57 (2012) 2.
Feret, R. (1892). Sur la compacité des mortiers hydrauliques. As mentioned in Fennis (2011).
Fraaij, A.L.A. & Rooij, M.R. de (2008) The workability of concrete: is there an easy way to
produce self-compacting concrete? In: Dhir, R.K., Hewlett, P.C., Csetenyi, L.J. and
Newlands, M.D. (eds). Role for concrete in global development. Dundee, Scotland, UK,387-
396.
Fuller, W. B., & Thompson, S. E. (1907). The laws of proportioning concrete.

References 43
Funk, J. E., & Dinger, D. (2013). Predictive process control of crowded particulate
suspensions: applied to ceramic manufacturing. Springer Science & Business Media.
Furnas, C. C. (1929). Flow of gases through beds of broken solids (Vol. 300). US
Government Printing Office.
Galilei, G., & Drake, S. (1946). Two new sciences. Madison, WI: University of Wisconsin
Press.
Gram, H-E., Vogt, C., Ericsson, J. (2017). Mix design – General, method, and for special
applications. Betonghandbok – Material, Svensk Byggtjänst. 467-531. (In Swedish)
Glavind, M., & Pedersen, E. J. (1999, September). Packing calcuations applied for concrete
mix design. In Utilizing Ready Mixed Concrete and Mortar: Proceedings of the International
Conference Held at the University of Dundee, Scotland, UK.
Goltermann, P., Johansen, V. and Palbøl, L. (1997) Packing of Aggregates: An Alternative

Tool to Determine the Optimal Aggregate Mix. ACI Materials Journal, Vol. 94 (5), pp. 435-
443.
Grünewald, S. (2004) Performance-based design of self-compacting fiber reinforced concrete.

PhD Thesis. Delft: Delft University of Technology.
Hunger, M. (2010). An integral design concept for ecological self-compacting concrete.

Eindhoven University.
Hunger, M., & Brouwers, H. J. H. (2009). Flow analysis of water–powder mixtures:

Application to specific surface area and shape factor. Cement and Concrete
Composites, 31(1), 39-59.
Idorn, G.M. (1995) Europack V1.1 User Manual Europack (G.M. Idorn Consult A/S,
Denmark).
Johansson, N., Emborg, M., (2014). Methods to optimize aggregate distribution: Evaluation
by concrete and mortar experiments. Nordic Concrete Research, Vol. 51, pp 145-157.
Jones, M. R., Zheng, L., & Newlands, M. D. (2002). Comparison of particle packing models
for proportioning concrete constitutents for minimum voids ratio. Materials and
structures, 35(5), 301-309.
Kennedy, C. T. (1940, February). The design of concrete mixes. In American Concrete

Institute Journal Proceedings .Vol. 36, No. 2, 373-400.
Krell, J. Die Konsistenz von Zementleim, Mörtel, und Beton und ihre zeitliche Veränderung.
Faculty of Civilengineering. Aachen, Germany: Rheinisch-Westfälische Technische
Hochschule Aachen (1985). As mentioned in: Fennis, S. A. A. M. Design of ecological
concrete by particle packing optimization PhD thesis, Delft University of Technology. (2011)
Kronlöf, A. (1994). Effect of very fine aggregate on concrete strength.Materials and

Structures, 27(1), 15-25.
44 References
Kumar, S. V., & Santhanam, M. (2003). Particle packing theories and their application in
concrete mixture proportioning: A review. Indian Concrete Journal 77(9), 1324-1331.
Kwan, A. K. H., & Mora, C. F. (2002). Effects of various, shape parameters on packing of
aggregate particles. Magazine of concrete Research 53(2),91-100.
Lagerblad, B., Gram, H. E., & Westerholm, M.(2013) Quality of fine materials from crushed
rocks in sustainable concrete production. In Proceedings SCMT3, Kyoto, Japan.
Landau, L. D., & Lifshitz, E. M. (1987). Fluid mechanics. 1987. Course of Theoretical
Physics.
Landis, E. N., & Keane, D. T. (2010). X-ray microtomography. Materials

characterization, 61(12), 1305-1316.
Maeyama, A., Maruyama, K., Midorikawa, T., & Sakata, N. (1998, August). Characterization
of powder for self-compacting concrete. In International Workshop on Self Compacting
Concrete.
Mahasenan, N., Smith, S., Humphreys, K., & Kaya, Y. (2003, August). The cement industry
and global climate change: current and potential future cement industry CO2 emissions.
In Greenhouse Gas Control Technologies-6th International Conference (Vol. 2. 995-1000).
Elsevier.
Marquardt, I., Vala, J., & Diederichs, U. (2002). CONCRETE TECHNOLOGY-

Determination of the composition of self-compacting concretes on the basis of the water
requirements of the constituent materials. Betonwerk und Fertigteiltechnik, 68(11), 22-31.
McCabe, W. L., Smith, J. C., & Harriott, P. (1993). Unit operations of chemical
engineering (Vol. 5, p. 154). New York: McGraw-Hill.
Midorikawa, T., Pelova, G. I., & Walraven, J. C. (2009). Application of" The Water Layer
Model" to self-compacting mortar with different size distributions of fine aggregate. Heron,
54 (2/3).
Mørtsell, E., Maage, M., & Smeplass, S. (1996). A particle-matrix model for prediction of
workability of concrete. In RILEM PROCEEDINGS 1996, 429-438.
Neville, A. M. (1995). Properties of concrete. Cartermills Publishing.
Pade, C., Thrane, L.N., Kaasgaard, M. (2009) 4C-Packing user’s manual.
Powers, T. C. (1969). The properties of fresh concrete. Treval Clifford.
Powers, T.C. (1971) ‘ Foundamental aspects of concrete shrinkage’ , Rev. Materiaux et

constructions, 545: 79-85.
Quiroga, P. N., & Fowler, D. W. (2004). Guidelines for proportioning optimized concrete
mixtures with high microfines. ICAR Technical Reports.
References 45
Reknes, K. (2001, August). Particle-matrix model based design of self-compacting concrete

with lignosulfonate water reducer. In Proceedings of the Second International Symposium on
Self-Compacting Concrete, 247-256.
Reschke, T., Siebel, E., & Thielen, G. (1998). Influence of the granulometry and reactivity of
cement and additions on the development of the strength and microstructure of mortar and
concrete. Concrete Technology Reports,200, 25-38.
Samarakoon, S., Vie, Ø. S., & Fjelldal, R. K. (2015). Self-Compacting White Concrete Mix
Design Using the Particle Matrix Model. International journal of civil, environmental,
structural, concstructions and arch. Eng. Vol 9(7), 802-806.
Smeplass, S., & Mørtsell, E. (2001, October). The particle matrix model applied on SCC.
In The Second International Symposium on Self-Compacting Concrete , 23-25.
Teichmann, T. (2008). Influence of the granulometrie and the water content on the strength
and density of cement stone. PhD Thesis. Kassel University
Toufar, W., Born, M., & Klose, E. (1976). Contribution of optimisation of components of
different density in polydispersed particles systems.Freiberger Booklet A, 558, 29-44.
Walker, S., & Bloem, D. L. (1960, September). Effects of aggregate size on properties of
concrete. In Journal Proceedings ,Vol. 57(9), 283-298.
Standards
BS EN 1097-3: 1998. Tests for mechanical and physical properties of aggregates,
determination of loose bulk density and voids.
ASTM C29 / 29M – 16. Standard test method for bulk density and voids in aggregate.
ASTM C127-15. Standard test method for Relative Density (Specific Gravity) and absorption
of coarse aggregate.
ASTM C136/136M – 14. Standard test method for sieve analysis of fine and coarse
aggregates.
ASTM C204-16. Standard test method for fineness of hydraulic cement by air permeability
apparatus.
ASTM D75 / D75M -14. Standard practice for sampling aggregates.
ASTM D854-14. Standard test method for specific gravity of soil solids by water Pycnometer.
ASTM D4791-10. Standard test method for flat particles, elongated particles, or flat and
elongated particles in coarse aggregate.
46 References
Websites
www.lcpc.fr. RENÉ-LCPC, http://www.lcpc.fr/fr/produits/rene/index.dml. visited: 2016/11.
www.dti.dk. 4C-Packing, http://www.dti.dk/2783. visited: 2016/11.
http://www.heidelbergcement.com/no/no/norcem/sustainability/Karbonfangst/Nyhetsbrev.htm
,Carbon capture newsletter, visited: 2015/10
http://www.hcne-sustainability.nu/en/node/2183, Meeting the challenge through zero vision,

visited: 2015/10
Doctoral and licentiate theses 47
Doctoral and Licentiate theses

Structural and Fire Engineering
Luleå University of Technology
Doctoral theses
1980 Ulf Arne Girhammar: Dynamic fail-safe behaviour of steel structures. Doctoral Thesis
1980:060D. 309 pp.
1983 Kent Gylltoft: Fracture mechanics models for fatigue in concrete structures. Doctoral
Thesis 1983:25D. 210 pp.
1985 Thomas Olofsson: Mathematical modelling of jointed rock masses. Doctoral Thesis
1985:42D. 143 pp. (In collaboration with the Division of Rock Mechanics).
1988 Lennart Fransson: Thermal ice pressure on structures in ice covers. Doctoral Thesis
1988:67D. 161 pp.
1989 Mats Emborg: Thermal stresses in concrete structures at early ages. Doctoral Thesis
1989:73D. 285 pp.
1993 Lars Stehn: Tensile fracture of ice. Test methods and fracture mechanics analysis.
Doctoral Thesis 1993:129D, September 1993. 136 pp.
1994 Björn Täljsten: Plate bonding. Strengthening of existing concrete structures with epoxy
bonded plates of steel or fibre reinforced plastics. Doctoral Thesis 1994:152D, August 1994.
283 pp.
1994 Jan-Erik Jonasson: Modelling of temperature, moisture and stresses in young concrete.
Doctoral Thesis 1994:153D, August 1994. 227 pp.
1995 Ulf Ohlsson: Fracture mechanics analysis of concrete structures. Doctoral Thesis
1995:179D, December 1995. 98 pp.
1998 Keivan Noghabai: Effect of tension softening on the performance of concrete structures.
Doctoral Thesis 1998:21, August 1998. 150 pp.
48
1999 Gustaf Westman: Concrete creep and thermal stresses. New creep models and their
effects on stress development. Doctoral Thesis 1999:10, May 1999. 301 pp.
1999 Henrik Gabrielsson: Ductility in high performance concrete structures. An experimental

investigation and a theoretical study of prestressed hollow core slabs and prestressed
cylindrical pole elements. Doctoral Thesis 1999:15, May 1999. 283 pp.
48 Doctoral and licentiate theses
2000 Patrik Groth: Fibre reinforced concrete – Fracture mechanics methods applied on self-
compacting concrete and energetically modified binders. Doctoral Thesis 2000:04, January
2000. 214 pp. ISBN 978-91-85685-00-4.
2000 Hans Hedlund: Hardening concrete. Measurements and evaluation of non-elastic

deformation and associated restraint stresses. Doctoral Thesis 2000:25, December 2000. 394
pp. ISBN 91-89580-00-1.
2003 Anders Carolin: Carbon fibre reinforced polymers for strengthening of structural
members. Doctoral Thesis 2003:18, June 2003. 190 pp. ISBN 91-89580-04-4.
2003 Martin Nilsson: Restraint factors and partial coefficients for crack risk analyses of early
age concrete structures. Doctoral Thesis 2003:19, June 2003. 170 pp. ISBN: 91-89580-05-2.
2003 Mårten Larson: Thermal crack estimation in early age concrete – Models and methods
for practical application. Doctoral Thesis 2003:20, June 2003. 190 pp. ISBN 91-86580-06-0.
2005 Erik Nordström: Durability of sprayed concrete. Steel fibre corrosion in cracks.
Doctoral Thesis 2005:02, January 2005. 151 pp. ISBN 978-91-85685-01-1.
2006 Rogier Jongeling: A process model for work-flow management in construction.

Combined use of location-based scheduling and 4D CAD. Doctoral Thesis 2006:47, October
2006. 191 pp. ISBN 978-91-85685-02-8.
2006 Jonas Carlswärd: Shrinkage cracking of steel fibre reinforced self compacting concrete
overlays – Test methods and theoretical modelling. Doctoral Thesis 2006:55, December 2006.
250 pp. ISBN 978-91-85685-04-2.
2006 Håkan Thun: Assessment of fatigue resistance and strength in existing concrete
structures. Doctoral Thesis 2006:65, December 2006. 169 pp. ISBN 978-91-85685- 03-5.
2007 Lundqvist Joakim: Numerical analysis of concrete elements strengthened with carbon
fiber reinforced polymers. Doctoral Thesis 2007:07, March 2007. 50 pp. ISBN 978-91-85685-
06-6.
49
2007 Arvid Hejll: Civil structural health monitoring – Strategies, methods and applications.
Doctoral Thesis 2007:10, March 2007. 189 pp. ISBN 978-91-85685-08-0.
2007 Stefan Woksepp: Virtual reality in construction: Tools, methods and processes.
Doctoral Thesis 2007:49, November 2007. 191 pp. ISBN 978-91-85685-09-7.
2007 Romuald Rwamamara: Planning the healthy construction workplace through risk
assessment and design methods. Doctoral Thesis 2007:74, November 2007. 179 pp. ISBN
978-91-85685-11-0.
2008 Björnar Sand: Nonlinear finite element simulations of ice forces on offshore structures.
Doctoral Thesis 2008:39, September 2008. 241 pp.
2008 Bengt Toolanen: Lean contracting: relational contracting influenced by lean thinking.
Doctoral Thesis 2008:41, October 2008. 190 pp.
2008 Sofia Utsi: Performance based concrete mix-design: Aggregate and micro mortar
optimization applied on self-compacting concrete containing fly ash. Doctoral Thesis
2008:49, November 2008. 190 pp.
2009 Markus Bergström: Assessment of existing concrete bridges: Bending stiffness as a

performance indicator. Doctoral Thesis, March 2009. 241 pp. ISBN 978-91-86233-11-2.
2009 Tobias Larsson: Fatigue assessment of riveted bridges. Doctoral Thesis, March 2009.
165 pp. ISBN 978-91-86233-13-6.
2009 Thomas Blanksvärd: Strengthening of concrete structures by the use of mineral based
composites: System and design models for flexure and shear. Doctoral Thesis, April 2009.
156 pp. ISBN 978-91-86233-23-5.
2011 Anders Bennitz: Externally unbonded post-tensioned CFRP tendons – A system

solution. Doctoral Thesis, February 2011. 68 pp. ISBN 978-91-7439-206-7.
2011 Gabriel Sas: FRP shear strengthening of reinforced concrete beams. Doctoral Thesis,
April 2011. 97 pp. ISBN 978-91-7439-239-5.
2011 Peter Simonsson: Buildability of concrete structures: processes, methods and material.
Doctoral Thesis, April 2011. 64 pp. ISBN 978-91-7439-243-2.
2011 Stig Bernander: Progressive landslides in long natural slopes. Formation, potential
extension and configuration of finished slides in strain-softening soils. Doctoral Thesis, May
2011, rev. August 2011 and April 2012. 250 pp. ISBN 978-91-7439-238-8. (In collaboration
with the Division of Soil Mechanics and Foundation Engineering).
2012 Arto Puurula: Load carrying capacity of a strengthened reinforced concrete bridge:
non-linear finite element modeling of a test to failure. Assessment of train load capacity of a
two span railway trough bridge in Örnsköldsvik strengthened with bars of carbon fibre
reinforced polymers (CFRP). Doctoral Thesis, May 2012. 100 pp.ISBN 978-91-7439-433-7.
2015 Mohammed Salih Mohammed Mahal: Fatigue behaviour of RC beams strengthened

with CFRP, Analytical and experimental investigations. Doctoral Thesis, March 2015. 138
pp. ISBN 978-91-7583-234-0.
2015 Jonny Nilimaa: Concrete bridges: Improved load capacity. Doctoral Thesis, June 2015.
180 pp. ISBN 978-91-7583-344-6.
2015 Tarek Edrees Saaed: Structural control and identification of civil engineering structures.
Doctoral Thesis, June 2015. 314 pp. ISBN 978-91-7583-241-8.
2015 Majid Al-Gburi: Restraint effect in early age concrete structures. Doctoral Thesis,
September 2015. 190 pp. ISBN 978-91-7583-374-3.
2017 Cosmin Popescu: CFRP strengthening of cut-out openings in concrete walls – Analysis
and laboratory tests. Doctoral Thesis, February 2017. 159 pp. ISSN 1402-1544.
Licentiate theses
1984 Lennart Fransson: Bärförmåga hos ett flytande istäcke. Beräkningsmodeller och
experimentella studier av naturlig is och av is förstärkt med armering. Licentiate Thesis
1984:012L. 137 pp. (In Swedish).
1985 Mats Emborg: Temperature stresses in massive concrete structures. Viscoelastic models
and laboratory tests. Licentiate Thesis 1985:011L, May 1985. rev. November 1985. 163 pp.
1987 Christer Hjalmarsson: Effektbehov i bostadshus. Experimentell bestämning av

effektbehov i små- och flerbostadshus. Licentiate Thesis 1987:009L, October 1987.72 pp. (In
Swedish).
1990 Björn Täljsten: Förstärkning av betongkonstruktioner genom pålimning av stålplåtar.

Licentiate Thesis 1990:06L, May 1990. 205 pp. (In Swedish).
1990 Ulf Ohlsson: Fracture mechanics studies of concrete structures. Licentiate Thesis
1990:07L, May 1990. 66 pp.
1990 Lars Stehn: Fracture toughness of sea ice. Development of a test system based on
chevron notched specimens. Licentiate Thesis 1990:11L, September 1990. 88 pp.
1992 Per Anders Daerga: Some experimental fracture mechanics studies in mode I of concrete
and wood. Licentiate Thesis 1992:12L, April 1992, rev. June 1992. 81 pp.
1993 Henrik Gabrielsson: Shear capacity of beams of reinforced high performance concrete.
Licentiate Thesis 1993:21L, May 1993. 109 pp.
1995 Keivan Noghabai: Splitting of concrete in the anchoring zone of deformed bars. A
fracture mechanics approach to bond. Licentiate Thesis 1995:26L, May 1995. 123 pp.
1995 Gustaf Westman: Thermal cracking in high performance concrete. Viscoelastic models
and laboratory tests. Licentiate Thesis 1995:27L, May 1995. 125 pp.
1995 Katarina Ekerfors: Mognadsutveckling i ung betong. Temperaturkänslighet, hållfasthet

och värmeutveckling. Licentiate Thesis 1995:34L, October 1995. 137 pp.(In Swedish).
1996 Patrik Groth: Cracking in concrete. Crack prevention with air-cooling and crack
distribution with steel fibre reinforcement. Licentiate Thesis 1996:37L, October 1996. 128 pp.
1996 Hans Hedlund: Stresses in high performance concrete due to temperature and moisture
variations at early ages. Licentiate Thesis 1996:38L, October 1996. 240 pp.
2000 Mårten Larson: Estimation of crack risk in early age concrete. Simplified methods for
practical use. Licentiate Thesis 2000:10, April 2000. 170 pp.
2000 Stig Bernander: Progressive landslides in long natural slopes. Formation, potential
extension and configuration of finished slides in strain-softening soils.. Licentiate Thesis
2000:16, May 2000. 137 pp. (In collaboration with the Division of Soil Mechanics and
Foundation Engineering).
2000 Martin Nilsson: Thermal cracking of young concrete. Partial coefficients, restraint
effects and influences of casting joints. Licentiate Thesis 2000:27, October 2000. 267 pp.
2000 Erik Nordström: Steel fibre corrosion in cracks. Durability of sprayed concrete.
Licentiate Thesis 2000:49, December 2000. 103 pp.
2001 Anders Carolin: Strengthening of concrete structures with CFRP – Shear strengthening
and full-scale applications. Licentiate thesis 2001:01, June 2001. 120 pp. ISBN 91-89580-01-
X.
2001 Håkan Thun: Evaluation of concrete structures. Strength development and fatigue
capacity. Licentiate Thesis 2001:25, June 2001. 164 pp. ISBN 91-89580-08-2.
2002 Patrice Godonue: Preliminary design and analysis of pedestrian FRP bridge deck.
Licentiate Thesis 2002:18. 203 pp.
2002 Jonas Carlswärd: Steel fibre reinforced concrete toppings exposed to shrinkage and
temperature deformations. Licentiate Thesis 2002:33, August 2002. 112 pp.
2003 Sofia Utsi: Self-compacting concrete – Properties of fresh and hardening concrete for
civil engineering applications. Licentiate Thesis 2003:19, June 2003. 185 pp.
2003 Anders Rönneblad: Product models for concrete structures – Standards, applications
and implementations. Licentiate Thesis 2003:22, June 2003. 104 pp.
2003 Håkan Nordin: Strengthening of concrete structures with pre-stressed CFRP. Licentiate
Thesis 2003:25, June 2003. 125 pp.
2004 Arto Puurula: Assessment of prestressed concrete bridges loaded in combined shear,
torsion and bending. Licentiate Thesis 2004:43, November 2004. 212 pp.
2004 Arvid Hejll: Structural health monitoring of bridges. Monitor, assess and retrofit.
Licentiate Thesis 2004:46, November 2004. 128 pp.
2005 Ola Enochsson: CFRP strengthening of concrete slabs, with and without openings.
Experiment, analysis, design and field application. Licentiate Thesis 2005:87, November
2005. 154 pp.
2006 Markus Bergström: Life cycle behaviour of concrete structures – Laboratory test and
probabilistic evaluation. Licentiate Thesis 2006:59, December 2006. 173 pp. ISBN 978-91-
85685-05-9.
2007 Thomas Blanksvärd: Strengthening of concrete structures by mineral based composites.

Licentiate Thesis 2007:15, March 2007. 300 pp. ISBN 978-91-85685-07-3.
2008 Peter Simonsson: Industrial bridge construction with cast in place concrete: New
production methods and lean construction philosophies. Licentiate Thesis 2008:17, May
2008. 164 pp. ISBN 978-91-85685-12-7.
2008 Anders Stenlund: Load carrying capacity of bridges: Three case studies of bridges in
northern Sweden where probabilistic methods have been used to study effects of monitoring
and strengthening. Licentiate Thesis 2008:18, May 2008. 306 pp. ISBN 978-91-85685-13-4.
2008 Anders Bennitz: Mechanical anchorage of prestressed CFRP tendons – Theory and
tests. Licentiate Thesis 2008:32, November 2008. 319 pp.
2008 Gabriel Sas: FRP shear strengthening of RC beams and walls. Licentiate Thesis
2008:39, December 2008. 107 pp.
2010 Tomas Sandström: Durability of concrete hydropower structures when repaired with
concrete overlays. Licentiate Thesis, February 2010. 179 pp. ISBN 978-91-7439-074-2.
2013 Johan Larsson: Mapping the concept of industrialized bridge construction: Potentials
and obstacles. Licentiate Thesis, January 2013. 66 pp. ISBN 978-91-7439-543-3.
2013 Jonny Nilimaa: Upgrading concrete bridges: Post-tensioning for higher loads.
Licentiate Thesis, January 2013. 300 pp. ISBN 978-91-7439-546-4.
2013 Katalin Orosz: Tensile behaviour of mineral-based composites. Licentiate Thesis, May
2013. 92 pp. ISBN 978-91-7439-663-8.
2013 Peter Fjellström: Measurement and modelling of young concrete properties. Licentiate
Thesis, May 2013. 121 pp. ISBN 978-91-7439-644-7.
2014 Majid Al-Gburi: Restraint in structures with young concrete: Tools and estimations for
practical use. Licentiate Thesis, September 2014. 106 pp. ISBN 978-91-7439-977-6.
2014 Niklas Bagge: Assessment of Concrete Bridges Models and Tests for Refined
Capacity Estimates. Licentiate Thesis, November 2014. 132 pp. ISBN 978-91-7583- 163-3.
2014 Tarek Edrees Saaed: Structural identification of civil engineering structures. Licentiate
Thesis, November 2014. 135 pp. ISBN 978-91-7583-053-7.
2015 Cosmin Popescu: FRP strengthening of concrete walls with openings. Licentiate Thesis,
December 2015. 134 pp. ISBN 978-91-7583-453-5.
2016 Faez Sayahi: Plastic shrinkage cracking in concrete. Licentiate Thesis, October 2016.
146 pp. ISBN 978-91-7583-678-2.
2016 Jens Häggström: Evaluation of the load carrying capacity of a steel truss railway
bridge: testing, theory and evaluation. Licentiate Thesis, December 2016. 139pp. ISBN: 978-
91-7583-739-0.
Paper I:
Particle Packing of Aggregates for Concrete Mix Design:
Models and Methods.
Ghasemi, Y., Johansson, N. (2014), Published in proceeding of XXII Nordic Concrete
Research symposium, Reykjavik, Iceland, August 13-15, 2014, pp. 109-112.
109
Particle packing of aggregates for concrete mix design: Models and methods.
Yahya Ghasemi
PhD-student
Div. Structural and Construction Engineering, Luleå Universtiy of
Technology
S-971 87 Luleå
E-mail: yahya.ghasemi@ltu.se
Tekn Lic, Niklas Johansson

PhD-student
Cementa AB
P.O. Box 104, S-624 22 Slite
Div. Structural and Construction Engineering, Luleå University of
Technology
S-971 87 Luleå
E-mail: niklas.johansson@cementa.se
niklas.johansson@ltu.se
ABSTRACT
An optimized aggregate particle packing density- used as a base for concrete mix design-
provides economic, environmental and technical advantages. The particle packing density can
be determined by many methods and predicted by different models. This paper reviews
common packing and procedures and compares predictions of three packing models. It was
found that the models tend to show different packing densities and percentage of ingredients at
maximum packing for the same mixture. A test setup is proposed to determine the accuracy of
each model’s prediction.
Keywords: Aggregate, Testing, Mix design, Dry packing
1. INTRODUCTION
The prediction of the packing density for aggregate mixture, together with fine particle (i. e.
matrix) modelling, is essential in concrete mix design. It is known that besides particle size
distribution, the packing density of the aggregates is influenced by the packing process and also
the shape of the grains /Proske & Ramge 2004/. Over the past few decades, several
mathematical models have been introduced for the purpose of predicting packing of granular
mixes. This study aims to compare three common particle packing models. The models were
used on 4 mixtures of aggregate fragments with known aggregate distribution and packing
density for each fragment. In addition to that, the paper will briefly review four common dry
packing methods and aims to illustrate how the variation in packing process affects the outcome
of packing density.
2. PACKING MODELS
2.1. Modified Toufar Model
110
In this model, the packing density and characteristic diameter of each material are used to
calculate the packing densities of particle combinations /Goltermann et al. 1997/. Modified
Toufar model can be used for estimating packing of a multicomponent system. However,
according to /Fennis 2011/.calculations of multi-component mixtures based on this procedure
tends to underestimate the packing density.
2.2. 4C Model
4C is a computer program developed by Danish Technological Institute and is based on the
Linear Packing Density Model (LPDM). The factors that are taken into account include particle
size distribution of the mixture and packing densities of each size class. The accuracy of the
model depends on interaction formulas which are relations derived from the packing density of
two-component mixtures /dti.dk; Glavind, et al. 1999/.
2.3. Compressible Packing Model

De Larrard /1999/ introduced the concept of virtual compactness that is defined as the packing
that can be obtained by placing the aggregates one by one in a mixture in such a way that the
minimum amount of space is left. The actual packing density can be derived from virtual
packing density by use of compaction index K. The value of K depends on the compaction
energy applied in the process of packing.
3. PACKING METHODS
Several methods exist for packing process; one method is suggested by de Larrard /1999/ using
vibration+compaction method. According to vibration+compaction approach, the aggregates
should be poured in a cylinder that is closed with a 20 kg steel piston and the whole set should
be put on a vibration table and submitted to the following vibration sequences: 2 min,40 s and 1
min at amplitudes 0.4 mm, 0.2 mm and 0.08 mm respectively.
American standard ASTM C29 enforces conducting rodding or jigging method. Rodding
method is defined as filling one third of a cylinder and rodding the layer of aggregate with 25
strokes and continuing with filling the two-thirds and again rodding as above and finally, filling
the cylinder to overflowing and rodding again. Jigging procedure includes filling the container
in three approximately equal layers and compacting each layer by placing the measure on a firm
base and raising the opposite sides alternately about 50 mm and allowing the measure to drop on
the surface blow. Each layer should be compacted by dropping the measure 50 times, 25 times
on each side.
Moreover, European standard EN-1097:3 suggests using loose packing by means of pouring the
aggregates in a standard cylinder from the distance of maximum 50 mm.
Obviously the selection of the method affects the packing density. Among the packing models
compiled for particle packing prediction, compressible packing model (CPM) takes the effect of
packing method into account by means of a compaction factor index- K -which was suggested to
have the value of 4.1, 4.5, 4.75 and 9 for loose packing, rodding, vibration and
vibration+compression respectively. Figure 1 shows the effect of compaction index on packing
density of a mixture based on CPM.
111
Figure1: Effect of K value on compaction. Actual packing densities of two classes were measured on
loose packing.
4. EXPERIMENTS AT LTU
In order to examine the reliability of the models, a series of tests have been designed. The
experiment will include mixing different types and fragments of aggregates to construct binary
mixes. The fines will be mixed with coarser aggregates in steps by replacing 10 % of coarse
materials by fines in each step. The trend for mixture packing density will be drawn and can be
compared to the results obtained from mentioned models. Eventually, models will be calibrated
based on the lab results. As a reliable model was found or compiled, the mixtures with higher
packing density will be used for making concrete samples with the aim of measuring the
workability of mixtures. The experiments will be done on loose packing in a standard cylinder.
As a part of the research, effect of packing method on packing density will be studied.
5. THEORETICAL COMPARISON OF THE MODELS

Examples of results for theoretical comparison of models’ predictions made so far are presented
in Figure 2. Each model suggested a different optimum contribution of the aggregates to the
mixture and moreover, the models did not agree on the maximum packing density that can be
obtained in the mixtures. 4C and Toufar model tend to almost merge as the finer aggregates
became dominant in the mixture. The predicted packing from CPM was significantly different
from 4C and Toufar. In the mixtures with crushed aggregate the difference between 4C and
CPM was about 3 % while for the mixture of crushed and natural the difference was increased
up to 7 %. CPM tends to predict higher packing density than the other two model whereas 4C
showed the least packing density.
6. FINAL COMMENTS
As it can be seen in Figure 2, the models led to different proportions of aggregates for the same
mixture, also the maximum predicted packing densities were unalike. It is noticeable that the
models tend to agree on the packing density as the finer material became dominant in the
mixture (above 60% of fine material). However, the maximum packing density of a mixture is
usually achieved when fine materials had 40 to 60% share of aggregate mixture, where the
models showed the highest deviations from each other. That emphasizes the need of further
studies on experimental data in order to calibrate the existing models or compiling a new one.
112
Figure 2: Comparison of packing models for four available aggregate combinations. Actual packing
density of fragments were measured in the lab.
REFERENCES
ASTM C 29
“Standard test method for bulk density and voids in aggregates”. ASTM International.
Fennis, S. A. A. M., 2011
“Design of ecological concrete by particle packing optimization” Dissertation, Delft
University of Technology.
Glavind, M., Olsen, G.S., Munch Petersen, C., 1999
”Packing calculations and concrete mix design”. Danish Technological Institute
Goltermann, P., Johansen, V., Palbol, L., 1997
“Packing of aggregates: an alternative tool to determine the optimal aggregate mix.”
ACI Materials Journal, Vol 94, pp. 435-443
Larrard, F. de., 1999
“Concrete mixture proportioning: a scientific approach.” London: E & FN Spon.
Proske, T.,Ramge, P., 2004
“Influence of the particle shape on the packing density of aggregates.” Darmstadt
Concrete – Annual Journal on Concrete and Concrete Structures, 19.
SS-EN 1097-3:1998
“Tests for mechanical and physical properties of aggregates-Part3: Determination of
loose bulk density and voids”. SIS..
Paper II:
Particle Packing for Concrete Mix Design: Models vs.
Reality.
Ghasemi, Y., Emborg, M. (2014), Published in Nordic Concrete Research, Vol. 51, No.3,
December 2014, pp. 85-94.
85
Particle Packing for Concrete Mix Design: Models vs. Reality
Yahya Ghasemi
PhD-student
Div. Structural and Construction Engineering, Luleå University of
Technology
S-971 87 Luleå
yahya.ghasemi@ltu.se
Mats Emborg
Professor LTU/Head of R&D Betongindustri AB
971 87 Luleå, Sweden./Betongindustri AB, 100 74 Stockholm,
Sweden
mats.emborg@betongindustri .se
mats.emborg@ltu.se
ABSTRACT
The packing density of aggregates is of great importance in

concrete mix design as obtaining a higher packing density leads to
less usage of cement paste which has technical, environmental and
economic benefits. It is thus of interest to model particle packing
correctly. Hence, in this study, packing densities of seven mixes of
aggregate were attained in the laboratory using the loose packing
method and were compared to values suggested by three models:
4C, Compressible Packing Model and Modified Toufar Model.
Modified Toufar showed 1.7% mean difference from the
laboratory values while CPM and 4C had mean differences of
1.8% and 1.9% respectively. In addition, it was found that some of
the models are preferable in certain mixtures.
Keywords: Aggregates, Loose Packing, Mix Design, CPM, 4C,

Modified Toufar
86
1. INTRODUCTION
Aggregate is a major component of the concrete, occupying 60% to 80% of its total volume.
Increasing the amount of aggregates corresponds to less usage of cement in the concrete which
has several beneficial effects, e.g. reduction in the cost of producing concrete, decrease in most
of the durability problems of hardened concrete, reducing shrinkage and cracking, etc.
In addition, reduction in usage of cement leads to a decrease in pollution caused by its

production. The cement industry produces about 5% of global man-made CO2 emissions; the
amount of CO2 emitted by the cement industry can be as high as 900 kg of CO2 for every 1000
kg of cement produced [1]. It should be noted that the cement industry worldwide and
especially in Scandinavia and Europe takes its responsibility and strong efforts are taken to
reduce the CO2 emissions at production. Some companies have formulated a zero-vision [2] and
others are engaged in carbon capturing of emitted gas see [3] describing a Heidelberg Cement
supported project. Also, concrete producers are now striving to reduce the amount of clinker and
thus CO2 by replacements such as fly ash, blast furnace slag, lime stone filler etc.
The packing density concept can be used as a part of concrete mix design with the aim of
minimizing the inter-particle voids between the constituents of concrete in order to reduce the
paste demand. Packing density is the ratio of the volume of solids to the bulk volume of the
solid particles [4, 5]. The date for one of the first articles on particle packing goes as far as 1892
[6] further researches were conducted mainly concentrating on designing an ideal aggregate size
distribution curve [7, 8]. In 1929 the first analytical packing model was designed to predict the
void ratio of a mixture of two particle groups [9]. Since then, plenty of researches were
conducted on the subject resulting in development of several analytical models and computer-
aided mix design software.
Particle packing models can be used as a tool to determine the optimum combination of
aggregate mix constitutes that will provide a maximum packing density and minimize the
remaining voids. Although it has been recognized nowadays that the binder phase can also be
graded just as the aggregate phase for the purpose of achieving tight particle packing or
minimum void, research results have shown that improvements achieved in the reduction of
void ratio are far greater with the aggregate phase than with the binder phase [10].
The aim of this paper is to examine the reliability and accuracy of analytical particle packing
models by comparing the suggested values by the models to actual aggregate packing values
obtained in the laboratory. For this purpose, three of more common packing models –Modified
Toufar, 4C and CPM- were studied. The results from this type of study can assist the
development of future mix design philosophies.
The study only dealt with the precision of packing models considering solely dry aggregates.
2. PARTICLE PACKING MODELS
As mentioned above, it is of extreme importance to minimize voids for optimising concrete

mixes. In order to fulfil this requirement it is vital to select an acceptably accurate packing
model to estimate the packing density. A number of particle packing models were developed
over the past 80 years. However, some of them were proved to be unsuitable for concrete mix
87
constituent proportioning [11, 12]. Among the remaining models, three of more common ones
were chosen and are described briefly in the following sections.
2.1. Modified Toufar
According to Toufar et al. [13] the packing density depends on the diameter ratio of the two
particle class that are to be mixed. It is assumed that each of fine particles is placed between
exactly four of the course particles. The Toufar model was later modified by Golterman et al.
[14] since it was shown that the original model predicts that the packing density of a sample of
coarse particles does not increase when a small amount of fine particles is added to the coarse
particles, which is in contrast with reality.
The required input data for Modified Toufar model includes packing density ĳand characteristic
diameter dchar of each material that is used in the combination. Modified Toufar model can be
used for estimating packing density of a multicomponent system. However, calculation of multi-
component mixtures based on this procedure tends to underestimate the packing density [15].
On the other hand Modified Toufar model is fairly easy to use and can be implemented in a
spreadsheet with a little effort.
2.2. 4C Model
4C is a computer program developed by Danish Technological Institute and is based on the

Linear Packing Density Model (LPDM). The Linear Packing Density Model can be used to
optimize the grading curve of a concrete mixture. The key elements of the LPDM used to
determine packing density are:
i. Calibrate the eigenpacking density ai of each constituent material

ii. Calculate ai of combination for each clustered size class di
iii. Calculate yi (volume) of combination for each clustered size class di
iv. Claculate the total packing density
The Linear Packing Density Model can predict the packing density for several particle classes,
which makes the model suitable for real concrete mixtures [10]. The accuracy of the model
depends on interaction formulas which are relations derived from the packing density of two-
component mixtures [16, 17]. 4C packing model uses ȝ-value as an interaction factor which
indicates possible maximum ratio (size) between small and large particles without the smaller
particle interfering with the packing of the larger particles. For the purpose of this study three
values of ȝ were compared to actual data obtained from laboratory tests. It should be mentioned
that ȝis an empirical value and needs to be calibrated based on the laboratory data.
2.3. Compressible Packing Model (CPM)
The CPM is a refined version of a previous model (LPDM) for grain mixtures [18]. The
Compressible Packing Model calculates packing density via the virtual compactness ȕi instead
of ai (eigenpacking) and a compaction index K. Virtual packing density is defined as the
maximum potential packing density of a mixture if the particles would have been placed one by
one in such a way that they use the minimum amount of space.
Obviously, the virtual packing density is higher than the real packing density. The difference
depends on the applied compaction energy. To drive the real packing density from virtual
88
density, compaction index K is used; the value of K depends on the compaction energy applied
to the mixture. K index was suggested to have the value of 4.1, 4.5, 4.75 and 9 for loose
packing, rodding, vibration and vibration+compression respectively. Figure 1 shows the effect
of compaction index on packing density of a mixture based on CPM [18].
a) b)
Figure1: (a) Effect of K value on compaction where actual packing densities of two classes
assumed to be constant. (b) Variation of K vs. packing density [18].
LPDM can be considered as a special case of CPM for which the compaction index K tends to
infinity.
Another difference between 4C and CPM is in the way that models handle the interaction of
packing densities of components in the mixture. While this is implanted in 4C by means of ȝ-
value, CPM suggest calculating the effect by using mathematical formulas.
In order to use the CPM model, packing density of each fraction, mean diameter and K-value
needs to be introduced to the model. CPM can be used to predict the packing density/void ratio
for combination of any given number of fractions. However, comparing to the other two models
CPM is more complex and requires more input data.
3. MATERIALS AND METHODS
3.1. Materials
Seven binary aggregate mixtures were made out of eight fractions of three quarries. The mixes
were made in several steps by adding 10% to 20% of volume of fines to coarse material in each
step for every binary mixture. Table 1 shows the consumables used in the experiments, Figure 2
illustrates the sieving curves for the materials and Table 2 shows the ingredients for the
mixtures.
89
Table 1 – Aggregates used for the experiments

Quarry Type Fraction
Riksten Natural 0 - 4 mm
8 - 16mm
Riksten Crushed 0 - 2 mm
0 - 4 mm
4 - 8 mm
8 - 16 mm
Ledinge Cubic crushed 0 - 4 mm
8 - 16 mm
Figure 2 – aggregate sieving curves
Table 2 – Mixtures ingredients

Combo Fine Coarse d50(fine) / d50(coarse)*
Mix1 Riksten Crushed 0-4 mm Riksten Crushed 8-16 mm 0.12
Mix4 Riksten Natural 0-4 mm Riksten Natural 8-16 mm 0.08
Mix5 Ledinge Cubic Crushed 0-4 mm** Ledinge Cubic Crushed 8-16 mm 0.07
Mix6 Ledinge Cubic Crushed 0-4 mm Riksten Natural 8-16 mm 0.09
Mix7 Riksten Natural 0-4 mm Ledinge Cubic Crushed 8-16 mm 0.07
* The ratio of mean diameter of fine aggregate class to mean diameter of coarse aggregate class.
**crushed aggregates formed in a crusher to a cubic shape so they become less angular.
3.2. Packing Method
The packing densities for each of listed materials and their mixtures were determined according
to European standard EN-1097:3 [19] by means of pouring the aggregate in a standard cylinder
90
form the distance of maximum 50mm. The dry particle densities and the bulk densities were
determined and their packing densities were calculated as the ratio of bulk density of the
aggregate to the solid density of the dry aggregate particles. Table 3 shows packing density of
aggregates.
Table 3 – Packing density of aggregates in un-compacted condition

Quarry Fraction Bulk density (kg/m3) Particle density Packing density*
(kg/m3)
Riksten Natural 0-4 mm 1682 2645 0.636
8-16 mm 1642 2645 0.621
Riksten Crushed 0-2 mm 1572 2674 0.588
0-4 mm 1695 2674 0.634
4-8 mm 1342 2674 0.502
8-16 mm 1562 2674 0.584
Ledinge Cubic Crushed 0-4 mm 1863 3064 0.608
8-16 mm 1657 3064 0.541
* Particle packing densities were corrected for cylinder wall effect.
4. RESULTS
Figure 3 illustrates the results of packing densities obtained from the laboratory experiment
versus Modified Toufar model, 4C and CPM. For the purpose of sensitivity analysis three
different ȝ-values were introduced in 4C software as 0.07, 0.05 and 0.03. The models’
estimation had a point to point deviation of 0.5 % to 5.8 % in packing density comparing to the
laboratory data. The least error occurred in mixtures with higher mean diameter ratio between
fine and coarse aggregate fractions. The maximum error took place in prediction of packing
densities for mixtures combined of natural aggregate as coarse and cubic crushed material as
fine.
Figure 4 shows the total comparison of differences between measured and calculated packing
densities. Considering all the data obtained in the laboratory, Modified Toufar showed 1.7 %
mean difference while the mean difference for CPM and 4C were 1.8 % and 1.9 % respectively.
However, comparing each mixture leads to different values as some of the models are more
suitable with specific mixtures; see Table 4 in Section 5.
For the purpose of unified comparison only results with assumption of ȝ-value = 0.07 was
considered for 4C model as this number is suggested in 4C manual and furthermore, it is not
possible to choose the best value of ȝ prior to conducting actual laboratory tests.
91
Figure 3 – Packing densities of binary mixes, models vs. lab data.

92
a) Modified Toufar b) CPM

Mean difference = 1.7% Mean difference = 1.8%
Line of Equality Line of Equality
c) 4C ȝ=0.07
Mean difference = 1.9%
Line of Equality
Figure 4 – Comparison of difference between calculated and measured values
5. DISCUSSION AND CONCLUSION
There were some subtle differences between the models: Modified Toufar and CPM suggested
more or less the same values of fine percentage corresponding to the packing density. However,
CPM slightly overestimated packing values comparing to Modified Toufar model. This is in
agreement with what was concluded earlier by Jones et al. [10]. Both CPM and Modified
Toufar overestimated the packing density where volume of fines engaged in the mixture was
less than 40% of the total volume. 4C tends to behave more inconsistently with variation of
mixtures.
According to Figure 4, Modified Toufar’s predictions show the least deviation from the
measured packing densities (1.7%). However, it had been showed that the model’s accuracy
decreases as the number of ingredients in a mixture increases [15].
CPM was able to predict the packing density with a mean difference of 1.8%. CPM is more
suitable for using with multicomponent mixtures as it was formulated in a way that theoretically
it can predict packing density of any given number of fractions in one equation, the same cannot
be said about Toufar as for Modified Toufar model, packing density of classes of aggregates
need to be calculated in a binary system and then the result should be added to a third class and
93
so on. 4C is capable of calculating the packing density for a mixture with up to three constitutes.
However, for a binary mixture, Toufar is preferable comparing to CPM and 4C as it has higher
accuracy and is simpler to use.
It should be mentioned that, all of the models trends merge and became more in agreement with
each other and the laboratory data on the finer side of the packing diagrams.
In order to compare the data in more detail, Table 4 is presented which shows the mean
difference between the calculated data by each of three models and measured values for each
mixture. It should be mentioned that for the calculation of mean difference, only the data
corresponding to 40 % to 60 % of fine/ total volume of aggregate was taken into account. The
reason for choosing the mentioned range is firstly, the maximum packing density usually occurs
in this domain and secondly, it is more practical to have concrete recipes with 40 % to 60 % of
fine / total volume of aggregate since using more than 60 % of fines will lead to a very viscous
concrete with large demand for superplasticizer and less than 40% of fines results in “too stony”
concrete.
Table 4 – Comparison of models suitability on each mixture

d50(fine) / Mean difference Suitability
Combo
d50(coarse) M.Toufar CPM 4C High Medium Low N/A
Mix1 0.12 0.78% 1.86% 1.37% 4C, Toufar CPM -- --
Mix2 0.09 1.12% 0.98% 1.64% CPM, Toufar 4C -- --
Mix3 0.52 1.83% 3.40% 1.40% 4C Toufar -- CPM
Mix4 0.08 1.24% 1.07% 1.67% CPM, Toufar 4C -- --
Mix5 0.07 1.62% 1.99% 1.62% 4C, Toufar 4C, Toufar, CPM -- --
Mix6 0.09 1.80% 1.34% 3.94% CPM Toufar -- 4C
Mix7 0.07 2.65% 1.75% 1.52% 4C CPM Toufar --
Note that in Table 4, suitability of the models was decided based on the mean difference from
the laboratory data where high suitability was assigned to predictions with less than 1.5 % mean
difference, medium was used for predictions with mean difference between 1.5 % to 2 % and
low suitability for predictions with error higher than 2 % and finally, predictions with error
higher than 3 % were considered as unsuitable. The criterion can be justified by the fact that
trials in the lab consisted of up to 2% variation in the packing density due to changes of size
distribution and randomness of aggregate shapes in the quarries.
Results from Table 4 imply that as the mean size ratio of fines over coarse material decreases
the accuracy of CPM and Modified Toufar increase. Contrariwise, 4C is more suitable with
higher mean size ratios. Note that Mix6 was consisting of cubic crushed materials as fine and
natural aggregate as coarse; this led to incapability of 4C model to predict the packing density
with good accuracy. Nevertheless, further work is necessary to examine the effect of changes in
packing density of a mixture on workability of concrete so that a basis for acceptable error range
of packing density/void ratio prediction of the models can be established.
AKNOWLEDGEMENT
The authors would like to acknowledge the financial support provided by the Swedish Research
Council Formas and also Mr Ismael Lensol whom carried out most of laboratory experiments as
a part of his master thesis.
94
REFERENCES
1. Mahasenan, N., Smith, S., Humphreys, K., & Kaya, Y. “The cement industry and global
climate change: current and potential future cement industry CO2
emissions.” Greenhouse Gas Control Technologies-6th International Conference,
August 2003, Oxford: Pergamon ,pp. 995-1000
2. “Meeting the challenge through a zero vision” retrieved 10th Oct 2014 from:
http://www.hcne-sustainability.nu/en/node/2183
3. “Carbon capture newsletter”, 2014-2, retrieved 10th Oct 2014 from:
http://www.heidelbergcement.com/no/no/norcem/sustainability/Karbonfangst/Nyhetsbre
v.htm
4. Wong, H. H., & Kwan, A. K. “Packing density: a key concept for mix design of high
performance concrete”. Proceedings of the materials science and technology in
engineering conference, HKIE materials division, Hong Kong, May 2005, pp. 1-15
5. Quiroga, P. N., & Fowler, D. W. “Guidelines for proportioning optimized concrete
mixtures with high microfines”. International Center for Aggregates Research Report:
104-2, 2004.
6. Feret, R. “Sur la compacité des mortiers hydrauliques”. Annales des Ponts et Chaussees.
Vol. 4 (2e semestre), 1892,(pp. 5-16).(In French)
7. Fuller, W. B., & Thompson, S. E. “The laws of proportioning concrete”. Transactions of
the American Society of Civil Engineers, 57(2), 1907.
8. Andreasen, A. M., & Andersen, J. “Relation between grain size and interstitial space in
products of unconsolidated granules.” Kolloid Z journal, 50, 1930, pp. 217-218
9. Furnas, C. C. “Flow of gases through beds of broken solids” (Vol. 300). US Govt. print.
off. 1929.
10. Jones, M. R., Zheng, L., & Newlands, M. D. “Comparison of particle packing models for
proportioning concrete constituents for minimum voids ratio” Journal of Material and
Science, Vol 35, June 2002, pp. 301-309
11. Dewar, J. ”Computer modelling of concrete mixtures.” CRC Press, 2002.
12. Johansen, V., & Andersen, P. J. “ Particle packing and concrete properties”, Journal of
Materials and Science of Concrete 2. 1996, pp. 111-147
13. Toufar, W., Born, M., & Klose, E. “Contribution of optimisation of components of
different density in polydispersed particles systems”. Freiberger Booklet A, 1976, pp.
29-44
14. Goltermann, P., & Johansen, V. (1997). “Packing of aggregates: an alternative tool to
determine the optimal aggregate mix” ACI Materials Journal, 94(5). 1997, pp. 435-443
15. Fennis, S. A. A. M.“Design of ecological concrete by particle packing optimization”
Dissertation, Delft University of Technology. 2011.
16. Stovall, T., Larrard, F. de and Buil, M. “ Lineair Packing Density Model of Grain
Mixtures” Powder Technology, Vol. 48, 1986, pp. 1-12
17. Larrard, F. de. “Concrete mixture proportioning: a scientific approach”. London: E & FN
Spon. 1999.
18. Glavind, M., Olsen, G. S., & Munch-Petersen, C. ”Packing calculations and concrete
mix design”. Nordic Concrete Research, 13(2).1993.
19. Swedish Standards Institute, “Tests for mechanical and physical properties of aggregate
– Part 3: Determination of loose bulk density and voids”. SS-EN-1097:3. October 1998.
Paper III:
Quantification of the Shape of Particles for Calculating
Specific Surface Area of Powders.
Ghasemi, Y., Emborg, M., Cwirzen, A. (2016), Published in proceeding of the Materials,
Systems and Structures in Civil Engineering Conference, MSSCE 2016, Lyngby, Denmark,
August 22-24, 2016, pp. 31-41.
,QWHUQDWLRQDO5,/(0&RQIHUHQFHRQ0DWHULDOV6\VWHPVDQG6WUXFWXUHVLQ&LYLO(QJLQHHULQJ
&RQIHUHQFHVHJPHQWRQ6LPXODWLRQWRROVXVHGLQWKHH[HFXWLRQSKDVHRI&RQFUHWH6WUXFWXUHV
$XJXVW7HFKQLFDO8QLYHUVLW\RI'HQPDUN/\QJE\'HQPDUN
48$17,),&$7,21 2) 7+( 6+$3( 2) 3$57,&/(6 )25

&$/&8/$7,1*63(&,),&685)$&($5($2)32:'(56

<DK\D*KDVHPL0DWV(PERUJ$QGU]HM&ZLU]HQ

/XOHnWHFKQLFDOXQLYHUVLW\/XOHn6ZHGHQ
%HWRQJLQGXVWUL$%6WRFNKROP6ZHGHQ

$EVWUDFW
7KHFRQFHSWVRISDUWLFOHSDFNLQJDQGZDWHUSDVWHOD\HUWKHRULHVDUHFRPPRQO\XVHGIRUEDVLV
RI FRQFUHWH PL[ GHVLJQ PRGHOV :KLOH SDUWLFOH SDFNLQJ LQVLVWV RQ DFKLHYLQJ IHZHU YRLGV LQ
DJJUHJDWH PDWUL[ E\ HJ DGGLQJ ILQH DJJUHJDWHV ZDWHUSDVWH OD\HU WKHRULHV VWDWH WKDW
LQFUHDVLQJWKHDPRXQWRIILQHVZLOOOHDGWRKLJKHUZDWHUGHPDQGVLQFHWKHVSHFLILFVXUIDFHDUHD
RI SDUWLFOHV ZLOO LQFUHDVH ,Q RUGHU WR FDOFXODWH WKH WKLFNQHVV RI H[FHVV SDVWH LQ ZDWHUSDVWH
OD\HU WKHRULHV LW LV HVVHQWLDO WR KDYH DQ HVWLPDWLRQ RI VSHFLILF VXUIDFH DUHD RI WKH SDUWLFOHV
+RZHYHUWKHFRPSOH[LW\RIWKHLQVWUXPHQWVUHTXLUHGWRPHDVXUHVSHFLILFVXUIDFHDUHDPDNHV
WKH PHWKRGV FRVWO\ DQG DOVR FDXVHV DQ LVVXH UHJDUGLQJ WKH DYDLODELOLW\ RI WKH LQVWUXPHQWV
OHDGLQJ WR OHVV SRSXODULW\ RI ZDWHUSDVWH OD\HU WKHRULHV FRPSDUHG WR PL[ GHVLJQ DSSURDFKHV
EDVHG RQ WKH SDUWLFOH SDFNLQJ WKHRU\ ,QWURGXFLQJ D PRGHO WKDW FDQ URXJKO\ HVWLPDWH WKH
VSHFLILFVXUIDFHDUHDRISDUWLFOHVPD\HOLPLQDWHWKHQHHGIRUPHDVXULQJVSHFLILFVXUIDFHDUHD
ZKLFK LQ WXUQ FRQWULEXWHV WR FRVW UHGXFWLRQ DQG DOVR OD\LQJ D IRXQGDWLRQ IRU FDOFXODWLRQ RI
ZDWHUOD\HUWKLFNQHVV
7KH SDSHU DLPHG WR VWXG\ FDOFXODWHG VSHFLILF VXUIDFH DUHD E\ UHSODFLQJ UHJXODU FRQYH[
SRO\KHGURQVZLWKDFWXDOSDUWLFOHVKDSHVEDVHGRQWKHVL]HGLVWULEXWLRQFXUYHVRIWKHVDPSOHV
DQGWRFRPSDUHWKHUHVXOWVZLWKGDWDREWDLQHGE\%ODLQHWHVW
7KH FRQGXFWHG VWXG\ VXJJHVWV WKDW LQ PRVW FDVHV DVVXPSWLRQ RI VSKHULFDO VKDSH IRU WKH
SDUWLFOHV OHDGV WR DQ DFFHSWDEOH HVWLPDWLRQ RI 66$ +RZHYHU 66$ RI SRZGHUV ZLWK PRUH
DQJXODUSDUWLFOHVFDQEHFDOFXODWHGPRUHDFFXUDWHO\ZLWKWKHDVVXPSWLRQRI3RO\KHGURQVKDSH
UDWKHUWKDQVSKHUHV

,QWURGXFWLRQ

%DFNJURXQG
'XULQJSDVWGHFDGHVVHYHUDODWWHPSWVZHUHPDGHRQIRUPXODWLQJWKHPL[GHVLJQRIFRQFUHWH
0RVWRIWKHVHPRGHOVDUHEDVHGRQWKHDVVXPSWLRQWKDWWKHSURSHUWLHVRIFRQFUHWHLQIUHVKVWDWH
QDPHO\IORZSURSHUWLHVDQGZRUNDELOLW\DUHFKLHIO\JRYHUQHGE\WKHSDUWLFOHVL]HGLVWULEXWLRQ
36' DQG WKH SDUWLFOH SDFNLQJ >@ $FFRUGLQJ WR WKH DERYHPHQWLRQHG PRGHOV SDUWLFOH
SDFNLQJ FDQ EH LQFUHDVHG E\ PRGLI\LQJ 36' ZKLFK LQ WXUQ XVXDOO\ OHDGV WR LQFUHDVLQJ WKH
VKDUHRIILQHV3DFNLQJWKHRU\DVVXPHVWKDWDGGLQJILQHSDUWLFOHVWRDSDUWLFOHVWUXFWXUHKHOSV
ILOOXSWKHYRLGVLQWKHSDUWLFOHVWUXFWXUHOHDYLQJRQO\PLQLPXPVSDFHIRUZDWHU,QWKLVZD\
DGGLQJ ILQH SDUWLFOHV ZLOO UHGXFH WKH ZDWHU UHTXLUHPHQW >@ +RZHYHU WKH SDFNLQJ RI
DJJUHJDWH LV GHSHQGHQW DOVR RQ WKH VKDSH RI WKH DJJUHJDWH SDUWLFOHV 7KH HIIHFW RI SDUWLFOH
VKDSH LV PRUH GLIILFXOW WR FRPSUHKHQG DQG LW LV LQGLUHFWO\ DFFRXQWHG IRU E\ PHDVXULQJ WKH
SDFNLQJRIPRQRVL]HGIUDFWLRQV
$QRWKHU DSSURDFK WR FRPSLOLQJ D PL[ GHVLJQ PRGHO LV EDVHG RQ H[FHVV SDVWHZDWHU OD\HU
WKHRULHVILUVWLQWURGXFHGE\>@$PDMRUK\SRWKHVLVE\>@VWDWHVWKDWWKHUHODWLYHVOXPSRID
ZDWHUSRZGHUPL[WXUHEHFRPHVDIXQFWLRQRIWKHVSHFLILFVXUIDFHDUHD66$ZKHQVXIILFLHQW
ZDWHULV SUHVHQW WR IORZ %DVHGRQWKHK\SRWKHVLVDWKLQOD\HURIDGVRUEHGZDWHUPROHFXOHV
DURXQGWKHSDUWLFOHVLVQHFHVVDU\WRDVVXUHWKHIORZFKDUDFWHULVWLFVRIWKHK\GUDWLQJV\VWHP,W
LVUHSRUWHGWKDWWKHWKLFNQHVVRIWKLVZDWHUOD\HULVUHODWHGWRVHQVLWLYLW\RIWKHPL[RQFKDQJHV
LQ WKH ZDWHU FRQWHQW DQG DOVR WKH VSHFLILF VXUIDFH DUHD RI WKH PDWHULDO XVHG >@ DV ODWHU
FRQILUPHG E\ >@ 0RUHRYHU WKH ZDWHU OD\HU WKHRU\ DVVXPHV WKDW WKH ZDWHU GHPDQG RI D
PL[WXUH GHSHQGV RQ WKH VXUIDFH DUHD RI WKH SDUWLFOHV LQ WKDW PL[WXUH ,QFUHDVLQJ WKH VXUIDFH
DUHDE\DGGLQJVPDOOSDUWLFOHVZLOOLQFUHDVHWKHZDWHUUHTXLUHPHQW>@
%RWK DSSURDFKHV 3DUWLFOH SDFNLQJ DQG :DWHUSDVWH OD\HU WKHRULHV VWURQJO\ GHSHQG RQ WKH
VKDSHRIWKHDJJUHJDWHLQRQHZD\RUDQRWKHUWKDWLVHVSHFLDOO\PRUHHVVHQWLDOZKHQLWFRPHV
WRZDWHUSDVWHOD\HUWKHRULHVZKLFKUHTXLUHVWKHVSHFLILFVXUIDFHDUHDDVDQLQSXWIRUWKHPRGHO
:KLOH LW LV SRVVLEOH WR GLUHFWO\ PHDVXUH 66$ WKH FRPSOH[LW\ RI LQVWUXPHQW UHTXLUHG IRU WKH
PHDVXUHPHQWLPSRVHVLVVXHVQDPHO\WKHDYDLODELOLW\RIWKHWHVWLQJLQVWUXPHQWVDQGWKHFRVW,W
LV DOVR SRVVLEOH WR HVWLPDWH WKH 66$ EDVHG RQ WKH DVVXPSWLRQ WKDW SDUWLFOHV KDYH LGHDO
VSKHULFDOVKDSHVXVLQJWKH36'GDWD>@
7KHSDSHUDLPVDWGHILQLQJDQDOWHUQDWLYHZD\WRLPSURYHWKHDFFXUDF\RIFDOFXODWHG66$E\
DVVXPLQJ LGHDO SRO\KHGURQ VKDSHV IRU WKH SDUWLFOHV LQ DGGLWLRQ WR WKH SUHYLRXVO\ PHQWLRQHG
VSKHULFDO VKDSHV DQG E\ FRPSDULQJ WKH FDOFXODWHG 66$ WR WKH PHDVXUHPHQWV REWDLQHG E\
%ODLQHWHVW $FFXUDWHHVWLPDWLRQRI66$FDQSURYLGHDWRROIRUFDOFXODWLQJZDWHUGHPDQGRI
WKHSRZGHUVDWWKHRQVHWRIIORZDQGDOVRFDQEHXVHGLQZDWHUSDVWHOD\HUWKHRULHV

6SHFLILFVXUIDFHDUHD66$
7KHVSHFLILFVXUIDFHDUHDLVWKHTXRWLHQWRIWKHDEVROXWHDYDLODEOHVXUIDFHH[FOXGLQJDOORSHQ
LQQHU VXUIDFHV SRUH ZDOOV GLYLGHG E\ WKH PDVV >PJ@ )RU FRQFUHWH PL[ GHVLJQ RQO\ WKH
RXWHUVXUIDFHEHLQJLQFRQWDFWZLWKZDWHULVRILQWHUHVW:LWKWKHFRQVLGHUDWLRQRIWKHVSHFLILF
GHQVLW\WKHVSHFLILFVXUIDFHDUHDFRXOGDOVREHH[SUHVVHGDVDUHDSHUYROXPH>PP@%HVLGHV
PDVVDQGVROLGYROXPHWKH66$FDQDOVREHUHODWHGWRDEXONYROXPHWDNLQJDFFRXQWIRUWKH
YRLGIUDFWLRQRUDFURVVVHFWLRQDODUHD>@
7KHUHDUHVHYHUDOZD\V RIGHWHUPLQLQJ66$EDVHGRQGLUHFWDQGLQGLUHFWPHDVXUHPHQWV HJ
%ODLQH WHVW /HD DQG 1XUVH PHWKRG %RWK WHVWV JLYH FORVHO\ VLPLODU UHVXOWV EXW DUH QRW

DSSOLFDEOH WR ILQH DQG XOWUDILQH SRZGHUV $QRWKHU PHWKRG WKDW KDV EHHQ XVHG WR GHWHUPLQH
66$ LV WKH YROXPHWULF VWDWLF PXOWLSRLQW PHWKRG EHWWHU NQRZQ DV WKH %(7 PHWKRG >@
+RZHYHU GHWHUPLQLQJ WKH 66$ YDOXH XVLQJ WKH PHQWLRQHG WHVWV LV DVVRFLDWHG ZLWK D KLJKO\
GHYLFH UHODWHG FRPSOH[LW\ $Q LQGLUHFW SRVVLELOLW\ IRU GHWHUPLQLQJ WKH 66$ FDQ EH LWV
GHYLDWLRQIURPWKHSDUWLFOHVL]HGLVWULEXWLRQ>@$VVXPLQJWKDWDOOWKHSDUWLFOHVDUHVSKHUHV
WRWDO66$IRUVSKHULFDOSDUWLFOHVFDQEHFDOFXODWHGE\(T

௡
߱௜
ܽ௦௣௛ ൌ ͸ ෍
݀ҧ௜ǡ௔௥௜௧௛ Ǥ ߩ௦
௜ୀଵ
ZKHUH
߱௜ LVWKHPDVVRIDJUDLQIUDFWLRQiEHLQJWKHPDVVSHUFHQWDJHRIWKHIUDFWLRQEHWZHHQdi DQG
di+1
݀ҧ௜ǡ௔௥௜௧௛ LVWKHDULWKPHWLFPHDQGLDPHWHURIIUDFWLRQi DQGi+1
ߩ௦ LVWKHVSHFLILFGHQVLW\RIWKHSDUWLFOHV

:DWHU3DVWHOD\HUWKHRULHV
7KHSDVWHOD\HUWKHRU\RULJLQDWHVLQWKHGHVLJQRIVHOIFRPSDFWLQJFRQFUHWH,WLVNQRZQWKDW
FHPHQW SDVWH VKRXOG QRW RQO\ ILOO WKH YRLGV DPRQJ WKH DJJUHJDWHV EXW VXUURXQG DOO SDUWLFOHV
ZLWK D WKLQ OD\HU RI SDVWH WR IXOILOO ZRUNDELOLW\ UHTXLUHPHQWV >@ 7KH UHTXLUHG DPRXQW RI
SDVWHIRUWKLVOD\HUGHSHQGVRQWKHVSHFLILFVXUIDFHDUHDRIWKHSDUWLFOHV7KHODUJHUWKHVXUIDFH
DUHDRIWKHSDUWLFOHVWKHVPDOOHUWKHWKLFNQHVVRIWKHVXUURXQGLQJSDVWHOD\HUDQGWKHODUJHUWKH
DYDLODEOH VXUIDFH IRU FHPHQW PDWUL[–DJJUHJDWH ERQG 7KH WKHRU\ KDV SURYHQ XVHIXO LQ VHOI
FRPSDFWLQJFRQFUHWH>@
,QSDVWHOD\HUWKHRU\DSSURDFKWKHSDFNLQJGHQVLW\RIWKHDJJUHJDWHLVPHDVXUHGLQRUGHUWR
GHWHUPLQHWKHDPRXQWRIFHPHQWSDVWHUHTXLUHGWRILOOXSWKHYRLGVWKHUDPLQLQJSDVWHDGGHG
LQWKHPL[WXUHLVFRQVLGHUHGDVH[FHVVSDVWH:DWHUOD\HUWKHRU\IROORZVDVLPLODUSULQLFSOHDV
SDVWHOD\HUWKHRU\%DVHGRQWKHH[FHVVZDWHUOD\HUWKHRU\WKHWKLFNQHVVRIZDWHUOD\HUFDQEH
FDOFXODWHG E\ GLYLGLQJ WKH DPRXQW RI ZDWHU E\ DYDLODEOH VSHFLILF VXUIDFH DUHD RI DOO WKH
SDUWLFOHVLQFOXGLQJFHPHQWDQGRWKHUPLFURILQHVVHH)LJXUH

)LJXUH9ROXPHRIZDWHUGLYLGHGLQWRH[FHVVZDWHUDQGYRLGILOOLQJZDWHUZLWKLQDFRQFUHWHPL[WXUH
LQDXQLWYROXPH>@


(VWLPDWLRQRIVSHFLILFVXUIDFHDUHDRISDUWLFOHVSOD\VDPDMRUUROHLQFDOFXODWLRQRIWKLFNQHVVRI
ERWKSDVWHDQGWKHZDWHUWKDWVXUURQGVWKHSDUWLFOHV

0DWHULDOV

,QRUGHUWRH[DPLQHWKHDFFXUDF\RI66$FDOFXODWLRQWKHGDWDQHHGVWREHFRPSDUHGWRDFWXDOO
PHDVXUPHQWVRI66$IRUWKLVSXUSRVHIRXUSRZGHUVZHUHFKRVHQDQGWKHLUSURSHUWLHVQDPHO\
VSHFLILF GHQVLW\ EXON GHQVLW\ %ODLQH YDOXH DQG WKHLU SDUWLFOH VL]H GLVWULEXWLRQ FXUYHV ZHUH
H[WUDFWHGIURP>@DQG>@DQGDUHUHSUHVHQWHGLQ7DEOH

7DEOH'HQVLWLHVDQG66$RIWKHSRZGHUV>@>@
6SHFLILFGHQVLW\ /RRVHSDFNLQJ %XONGHQVLW\ 66$EDVHGRQ
0DWHULDO
JFP GHQVLW\ JFP %ODLQHFPJ
&(0,,,%1
0DUEOH3RZGHU
/LPHVWRQH
4XDUW]3RZGHU

3DUWLFOHVL]HGLVWULEXWLRQRIWKHPDWHULDOVLVDQLPSRUWDQWIDFWRULQFDOFXODWLRQRIWKHVSHFLILF
VXUIDFHDUHDWKHUHDUHVHYHUDOPHWKRGVIRUGHWHUPLQLQJWKH36'7KHVL]HGLVWULEXWLRQFXUYH
ZDVREWDLQHGE\GHSOR\LQJORZDQJOHODVHUOLJKWVFDWWHULQJWHFKQLTXH/$//6FRQGXFWHGE\
>@6HHWKHSDUWLFOHVL]HGLVWULEXWLRQFXUYHVRIWKHVHOHFWHGSRZGHUVLQ)LJXUH

ϭϬϬ
ϵϬ D///ϰϮ͕ϱE
ϴϬ DĂƌďůĞWŽǁĚĞƌ
Cumulative finer (Vol. %)
ϳϬ >ŝŵĞƐƚŽŶĞ
YƵĂƌƚǌ
ϲϬ
ϱϬ
ϰϬ
ϯϬ
ϮϬ
ϭϬ
Ϭ
Ϭ͕ϭ ϭ ϭϬ ϭϬϬ ϭϬϬϬ
Particle size (micron)

)LJXUH3DUWLFOHVL]HGLVWULEXWLRQRIWKHSRZGHUV>@>@


$GGLWLRQDOO\ LQ RUGHU WR IXUWKHU YDOLGDWH WKH UHODWLRQ EHWZHHQ SRO\KHGURQV DQG WKH SDUWLFOH
VKDSH DQG WR EH DEOH WR GLVWLQJXLVK WKH GLIIHUHQFH LQ SDUWLFOH JHRPHWU\ 6(0 PLFURJUDSKV
KDYHEHHQSUHSDUHGE\>@DQGVKRZQLQ)LJXUH
D&(0,,,%1 E'RORPLWLFPDUEOHSRZGHU

F/LPHVWRQHSRZGHU

)LJXUH6(0PLFURJUDSKVRIWKHLQYHVWLJDWHGSRZGHUV;PDJQLILHG>@

&RPSXWDWLRQRI66$EDVHGRQSDUWLFOHVL]HGLVWULEXWLRQ

$VPHQWLRQHGEHIRUHXVLQJ(TLWLVSRVVLEOHWRFDOFXODWH66$EDVHGRQWKHSDUWLFOHVL]H
GLVWULEXWLRQ IRU VSKHUHV ,Q DGGLWLRQ WR VSKHUHEDVHG FDOFXODWLRQ RI 66$ RWKHU SRO\KHGURQV
ZHUH H[DPLQHG WR UHFDOFXODWH WKH 66$ 7DEOH VKRZV WKH VHOHFWHG SRO\KHGURQV DQG WKHLU
JHRPHWULFDOSURSHUWLHV


7DEOH3RO\KHGURQVXVHGLQWKHFDOFXODWLRQRI66$
6KDSH 6XUIDFH$UHD 9ROXPH 66$9
ξʹܽଷ ͳͶǤ͸ͻ͹
7HWUDKHGURQ ξ͵ܽଶ
ͳʹ ܽ

͸
&XEH ͸ܽଶ ܽଷ

ܽ
ͳ
ʹξ͵ܽଶ ξʹܽଷ ͹Ǥ͵Ͷͺ
2FWDKHGURQ ͵
ܽ

య ͳ ʹǤ͸ͻͶ
'RGHFDKHGURQ ටʹͷ ൅ ͳͲξͷܽଶ ሺͳͷ ൅ ͹ξͷሻܽଷ
Ͷ ܽ
ͷξ͵ܽଶ ͷ ͵Ǥͻ͹Ͳ
,FRVDKHGURQ ሺ͵ ൅ ξͷሻܽଷ
ͳʹ ܽ
Ͷߨܽଷ ͵
6SKHUH Ͷߨܽଶ
͵ ܽ

5HSODFLQJ VSKHUHV ZLWK RWKHU SRO\KHGURQV ZLOO QRW RQO\ FKDQJH WKH FDOFXODWHG YROXPH DQG
VSHFLILF VXUIDFH DUHD EXW DOVR DIIHFWV WKH SDFH RI JURZWK LQ 66$9ROXPH UDWLR DFFRUGLQJ WR
VTXDUHFXEHODZ

6TXDUH&XEHODZ
7KH VTXDUHFXEH ODZ ZDV ILUVW LQWURGXFHG E\ *DOLOHR>@ DQG LW GHILQHV D PDWKHPDWLFDO
SULQFLSOHZKLFKGHVFULEHVWKHUHODWLRQVKLSEHWZHHQWKHYROXPHDQGWKHDUHDLQDFFRUGDQFHWR
FKDQJHVLQVL]H$FFRUGLQJWRWKHSULQFLSOHDVDVKDSHJURZVLQVL]HLWVYROXPHJURZVIDVWHU
WKDQLWVVXUIDFHDUHD&RQVHTXHQWO\DVWKHVL]HGHFUHDVHVLWVVXUIDFHDUHDJURZVIDVWHUWKDQLWV
YROXPH 7KH HIIHFW RI VTXDUHFXEH ODZ EHFRPHV HVSHFLDOO\ VLJQLILFDQW IRU FDOFXODWLRQ RI
VSHFLILF VXUIDFH DUHD RI ILQHU SDUWLFOHV QDPHO\ SRZGHUV DQG FHPHQW )LJXUH LOOXVWUDWHV WKH
GLIIHUHQFHLQJURZWKRI66$9ROUDWLRRISRO\KHGURQVDFFRUGLQJWRVTXDUHFXEHODZ


ϰϬϬ
ϯϱϬ
ϯϬϬ
ϮϱϬ
Surface Area
ϮϬϬ
ϭϱϬ
ϭϬϬ
ϱϬ
Ϭ
Ϭ ϱϬ ϭϬϬ ϭϱϬ ϮϬϬ ϮϱϬ ϯϬϬ ϯϱϬ ϰϬϬ ϰϱϬ ϱϬϬ
Volume

)LJXUH6XUIDFHDUHDDJDLQVWYROXPHRIWKHSODWRQLFVROLGVDQGDVSKHUH

%\ZULWLQJ(TLQLWVJHQHUDOIRUPWKHVTXDUHFXEHODZHIIHFWFDQEHLPSOHPHQWHGLQWKH
HTXDWLRQ

௡
ܵܵ‫ܣ‬௜ Ǥ ߱௜
ܽ௣௢௟௬ ൌ ෍
ܸ௜ Ǥ ߩ௦
௜ୀଵ

ZKHUH
66$L9LLVWKHVSHFLILFVXUIDFHDUHDWRYROXPHUDWLRRIIUDFWLRQi DQGLVUHODWHGWRWKHVKDSHDV
VKRZQLQWDEOH
7RGHILQHWKHHTXLYDOHQWSRO\KHGURQWRWKHVSKHUHVLWZDVDVVXPHGWKDWWKHSRO\KHGURQVKDYH
WKH VDPH YROXPH DV WKH UHWDLQHG VSKHUHV RQ HDFK IUDFWLRQ 7KH OHQJWK RI WKH VLGHV RI
SRO\KHGURQV LV FDOFXODWHG EDVHG RQ WKH PHQWLRQHG DVVXPSWLRQ 7KH DYHUDJH GLDPHWHU RI WKH
VSKHUHVLQHDFKIUDFWLRQLVGHWHUPLQHGEDVHGRQWKHDULWKPHWLFPHDQDSSURDFK

5HVXOWVDQGGLVFXVVLRQ

&DOFXODWHG VSHFLILF VXUIDFH DUHD DFFRUGLQJ WR (T DQG WKHLU FRUUHVSRQGLQJ %ODLQH YDOXHV
IURPODERUDWRU\WHVWV>@>@DUHSUHVHQWHGLQ7DEOH

7DEOH6XUIDFHDUHDRIWKHSRZGHUVDQGWKHGHYLDWLRQIURPWKHFDOFXODWHG66$
6SHFLILFVXUIDFHDUHDFPJ
0DWHULDO %ODLQH

&(0,,,%
0DUEOH
/LPHVWRQH
4XDUW]

&DOFXODWHGYDOXHVRI66$LQ FDVHRI&(0,,,% /LPHVWRQHSRZGHU DQG4XDUW]SRZGHU
ZHUHRYHUHVWLPDWHGE\DQGUHVSHFWLYHO\LQFRPSDULVRQWRWKHLUFRUUHVSRQGLQJ
%ODLQHYDOXHV7KHKLJKPDUJLQRIHUURUIRU/LPHVWRQHSRZGHUFDQEHUHODWHGWRWKHSULQFLSOHV
RIWKH%ODLQHPHWKRGDVLWLVDUHODWLYHWHVWZKLFKGHSHQGVRQWKHVKDSHRIWKHSDUWLFOHVDQGWKH
FRPSDFWLRQOHYHORUSRURVLW\RIWKHLUFRPSDFWHGEHG>@>@
,WVKRXOGEHQRWHGWKDWWKHPRGHORYHUHVWLPDWHG66$LQPRVWRIWKHVWXGLHGPDWHULDOV)RUWKH
FDVH RI 0DUEOH SRZGHU FDOFXODWLRQV EDVHG RQ 7HWUDKHGURQ VKDSH IRU SDUWLFOHV JLYH D FORVHU
HVWLPDWLRQRI66$FRPSDUHGWRWKHRWKHUSODWRQLFVROLGV

&RQFOXVLRQ

$QDO\VLVRIWKHPRGHO
:KLOH LW LV VKRZQ WKDW XVLQJ WKH ZDWHU OD\HU WKHRU\ FDQ EH D KHOSIXO WRRO IRU SUHGLFWLRQ RI
IORZDELOLW\RIFRQFUHWH>@WKHFRPSOH[LW\RILQVWUXPHQWVUHTXLUHGIRUPHDVXULQJWKH66$DV
DQHVVHQWLDOLQSXWLQWKHZDWHUOD\HUWKHRU\PRGHOPDNHVWKLVDSSURDFKOHVVIDYRXUHGE\WKH
LQGXVWU\FRPSDULQJWRWKHPL[GHVLJQPRGHOVEDVHGRQSDUWLFOHSDFNLQJ7KXVIRUPXODWLQJDQ
HTXDWLRQWKDWFDQSUHGLFW66$XVLQJVLPSOHLQSXWGDWDWKHZDWHUOD\HUWKHRU\FDQEHXVHGZLWK
PRUHVLPSOLFLW\DQGHIILFLHQF\DVDWRROIRUDQWLFLSDWLQJWKHZRUNDELOLW\RIWKHFRQFUHWH
7KH FRQGXFWHG VWXG\ VXJJHVWV WKDW LQ PRVW FDVHV DVVXPSWLRQ RI VSKHULFDO VKDSH IRU WKH
SDUWLFOHV OHDGV WR DQ DFFHSWDEOH HVWLPDWLRQ RI 66$ +RZHYHU 66$ RI SRZGHUV ZLWK PRUH
DQJXODUSDUWLFOHVFDQEHFDOFXODWHGPRUHDFFXUDWHO\ZLWKWKHDVVXPSWLRQRI3RO\KHGURQVKDSH
UDWKHUWKDQVSKHUHVDVLWZDVWKHFDVHIRUWKH0DUEOHSRZGHU6HH)LJXUHDQG7DEOH

)XWXUHZRUN
0RUH GDWD LV UHTXLUHG LQ RUGHU WR FDWHJRUL]H WKH VKDSH RI SDUWLFOHV EDVHG RQ WKH VXJJHVWHG
SRO\KHGURQV,QDGGLWLRQVLQFHWKH%ODLQHWHVWZLOOUHVXOWLQUHODWLYHYDOXHVIRUSRZGHUVRWKHU
WKDQFHPHQWDQDFFXUDWHPHWKRGRIPHDVXULQJ66$RISDUWLFOHVLVQHFHVVDU\WRFDOLEUDWHWKH

PRGHOHJPLFURWRPRJUDSK\$VDQH[W VWHSLQ WKHVWXG\WKHDFWXDO66$FDQEHPHDVXUHG

XVLQJ'PRGHOVRIWKHSDUWLFOHVWUXFWXUHJHQHUDWHGE\PLFURWRPRJUDSK\VFDQQLQJDQGFDQEH
FRPSDUHGZLWKWKHUHVXOWVREWDLQHGIURPWKHPRGHO0RUHRYHUWKHHIIHFWRIFKDQJHVLQ36'
FDQEHVWXGLHGE\PRGLI\LQJWKHGLVWULEXWLRQFXUYHVRIWKHSDUWLFOHWRDWDUJHWFXUYH

5HIHUHQFHV

>@'H/DUUDUG)&RQFUHWHPL[WXUHSURSRUWLRQLQJDVFLHQWLILFDSSURDFK&5&3UHVV
>@ *ODYLQG 0 2OVHQ *6 0XQFK 3HWHUVHQ & 3DFNLQJ FDOFXODWLRQV DQG FRQFUHWH PL[
design”. Danish Technological Institue.
>@ 7RXIDU : %RUQ 0 .ORVH ( &RQWULEXWLRQ RI RSWLPLVDWLRQ RI FRPSRQHQWV RI
GLIIHUHQWGHQVLW\LQSRO\GLVSHUVHGSDUWLFOHVV\VWHPV)UHLEHUJHU%RRNOHW$
>@ Kronlof, A., “Effect of Very Fine Aggregate on Concrete Strength,” 0DWHULDOV DQG
6WUXFWXUHV9RO
>@)HQQLV6$$0'HVLJQRIHFRORJLFDOFRQFUHWHE\SDUWLFOHSDFNLQJRSWLPL]DWLRQ3K'
WKHVLV'HOIW8QLYHUVLW\RI7HFKQRORJ\
>@.HQQHG\&KDUOHV77KHGHVLJQRIFRQFUHWHPL[HV-RXUQDO3URFHHGLQJV9RO1R

>@ %URXZHUV + - + DQG + - 5DGL[ 6HOIFRPSDFWLQJ FRQFUHWH WKHRUHWLFDO DQG
H[SHULPHQWDOVWXG\&HPHQWDQG&RQFUHWH5HVHDUFK
>@ +XQJHU 0 $Q LQWHJUDO GHVLJQ FRQFHSW RI HFRORJLFDO VHOIFRPSDFWLQJ FRQFUHWH 3K'
WKHVLV(LQGKRYHQ8QLYHUVLW\
>@ )UDDLM $/$ DQG 5RRLM 05 7KH ZRUNDELOLW\ RI FRQFUHWH LV WKHUH DQ HDV\ ZD\ WR
SURGXFH VHOIFRPSDFWLQJ FRQFUHWH" ,Q 'KLU 5. +HZOHWW 3& &VHWHQ\L /- DQG
1HZODQGV0'5ROHIRUFRQFUHWHLQJOREDOGHYHORSPHQW'XQGHH6FRWODQG8.

>@ +XQJHU 0 DQG + - + %URXZHUV )ORZ DQDO\VLV RI ZDWHU–SRZGHU PL[WXUHV
$SSOLFDWLRQ WR VSHFLILF VXUIDFH DUHD DQG VKDSH IDFWRU&HPHQW DQG &RQFUHWH
&RPSRVLWHV
>@ 0DH\DPD $ 0DUX\DPD . 0LGRUDNDZD 7 DQG 6DNDWD 1 &KDUDFWHUL]DWLRQ RI
3RZGHU IRU 6HOI&RPSDFWLQJ &RQFUHWH 6HFRQG ,QW :RUNVKRS RQ 6&& &RQFUHWH
(QJLQHHULQJ6HULHV-DSDQ6RFLHW\RI&LYLO(QJLQHHUV
>@0LGRULNDZD73HORYD*,DQG:DOUDYHQ-&$SSOLFDWLRQRIWKHZDWHUOD\HUPRGHO
WRVHOIFRPSDFWLQJPRUWDUZLWKGLIIHUHQWVL]HGLVWULEXWLRQRIILQHDJJUHJDWH,Q2]DZD.
DQG 2XFKL 0 HGV 7KH 6HFRQG ,QWHUQDWLRQDO 6\PSRVLXP RQ 6HOI&RPSDFWLQJ
&RQFUHWH7RN\R-DSDQ
>@7HLFKPDQQ7,QIOXHQFHRIWKHJUDQXORPHWULHDQGWKHZDWHUFRQWHQWRQWKHVWUHQJWKDQG
GHQVLW\RIFHPHQWVWRQH3K'7KHVLV.DVVHO8QLYHUVLW\
>@ 0F&DEH :/ 6PLWK -& DQG +DUULRWW 38QLW RSHUDWLRQV RI FKHPLFDO HQJLQHHULQJ
9RO1HZ<RUN0F*UDZ+LOO
>@ %UXQDXHU 6 (PPHWW 3 + DQG 7HOOHU ( $GVRUSWLRQ RI *DVHV LQ 0XOWLPROHFXODU
/D\HUV-RXUQDORIWKH$PHULFDQ&KHPLFDO6RFLHW\–
>@5HVFKNH7(6LHEHODQG*7KLHOHQ,QIOXHQFHRIWKHJUDQXORPHWU\DQGUHDFWLYLW\RI
FHPHQW DQG DGGLWLRQV RQ WKH GHYHORSPHQW RI WKH VWUHQJWK DQG PLFURVWUXFWXUH RI PRUWDU DQG
FRQFUHWH&RQFUHWH7HFKQRORJ\5HSRUWV
>@*UQHZDOG63HUIRUPDQFHEDVHGGHVLJQRIVHOIFRPSDFWLQJILEUHUHLQIRUFHG

FRQFUHWH3K'7KHVLV'HOIW'HOIW8QLYHUVLW\RI7HFKQRORJ\
>@-HQQLQJV+.URSS-6FULYHQHU.7KHPRGHOOLQJRIPLFURVWUXFWXUHDQGLWVSRWHQWLDO
IRUVWXG\LQJWUDQVSRUWSURSHUWLHV DQGGXUDELOLW\ 9RO6SULQJHU6FLHQFH %XVLQHVV
0HGLD
>@*DOLOHL*'UDNH67ZRQHZVFLHQFHV0DGLVRQ:,8QLYHUVLW\RI:LVFRQVLQ3UHVV

>@)HUUDULV&)*XWKULH:$YLOpV$,+DXSW50F'RQDOG%&HUWLILFDWLRQRI650
T3DUW,1,6763–
>@ )HUUDULV &) *XWKULH : $YLOpV $ , 3HOW] 0 +DXSW 5 0DF'RQDOG %6
&HUWLILFDWLRQRI650T3DUW,,3DUWLFOH6L]H'LVWULEXWLRQ1,6763–

Paper IV:
Estimation of Specific Surface Area of Particles Based on
Size Distribution Curve.
Ghasemi, Y., Emborg, M., Cwirzen, A. (2017), Submitted to the Magazine of Concrete
Research.
ESTIMATION OF SPECIFIC SURFACE AREA OF PARTICLES BASED
ON SIZE DISTRIBUTION CURVE
Yahya Ghasemi (1), Mats Emborg (1)(2), Andrzej Cwirzen (1)
(1) Luleå technical university, Luleå, Sweden

(2) Betongindustri AB, Stockholm, Sweden
Abstract
Workability at fresh state is one of the most important factors in design and production of
concrete and can be related to the water demand of the mixture, which in addition to other
factors, is a function of particle shape of aggregates and binders and their specific surface
area. While it is known that the shape of fine particles have a significant effect on the water
demand, there are uncertainties regarding how the various shape parameters would affect the
specific surface area, mainly because up to now many of the shape parameters are not yet
clearly defined and there are no commonly accepted methods for their measurement and/or
estimation.
In this research the actual particle shapes was replaced with regular convex polyhedrons to
calculate the total specific surface area using size distribution curves of the samples.
The obtained results indicate that while in some cases, assumption of spherical shape of
particles leads to an acceptable estimation of the specific surface area when compared with
Blaine tests results, the specific surface area of powders with more angular particles could be
calculated more accurately with the assumption of Polyhedron shape rather than spheres.
1. Introduction
Concrete in the plastic state can be characterized by several parameters among which
workability is probably the most important one and is influenced by the water requirement,
which in turn is a function of aggregates’ shape, size, and fine content. Thus, understanding
the role of aggregates is fundamental for the production of high performance concrete
(Alexander and Mindess, 2010).
Aggregates have a large variability in mineral composition, shape, surface roughness and
surface texture and their specific surface area. One major parameter influencing the water
demand is a comprehensive measurement of size, shape and roughness (Wang and Lai, 1998).
The shape of particles is a complex function of their formation conditions, the mineralogical
composition, and particle size and not only refers to basic shape of aggregates, but also to
other measures such as angularity, flakiness, etc. There is a considerable confusion on how
various shape parameters are defined. There are also no commonly accepted methods for their
measurement (Kwan and Mora, 2001). Particle shape can be classified by measuring the
length, width and thickness of particles. Estimation is easier for larger particles. The specific
surface area can be used as an indicator of size, shape and surface roughness of particles.
In asphalt mixtures, the specific surface area of the aggregate can be directly related to the
asphalt concrete binder thickness and therefore related to the rutting and fatigue performance
of asphalt concrete, (Alexander and Mindess, 2010). Furthermore, Hunger (2010) concluded
that in the case of a self-compacting concrete, a certain thickness of water layer surrounding
the particles in water-powder dispersion will put the mixture at the on-set of flow. In other
words, the relative slump of a water-powder mixture becomes a function of the specific
surface area when sufficient water is present to enable the flow (Brouwers and Radix, 2005).
It is also possible to estimate the specific surface area using particle size distribution data
based on the assumption that particles have spherical shape. However, particle shapes are far
from being spherical due to 3D randomness in their dimensions, related to the origin of the
aggregates, and their production method. This is particularly true for example in the case of
crushed aggregate.
The specific surface area is the quotient of the absolute available surface inclusive all open
inner surfaces (pore walls) divided by the mass [m2/g]. For concrete mix design, only the
outer surface being in contact with water is of interest. With the consideration of the specific
density, the specific surface area could also be expressed as area per volume [m2/m3]. The
total surface area of a set of aggregates is governed by the fine aggregate fraction according
to the square-cube law. Assuming that all particles were spherical in shape, the Specific
Surface Area (SSA), Įsph , would be easy to calculate based on the particle size distribution
and grading curves,(McCabe et. al., 1993) :
௡
߱௜
ܽ௦௣௛ = 6 ෍ (1)
݀ҧ௜ . ߩ௦
௜ୀଵ
where
߱௜ is the mass of a grain fraction i, being the mass percentage of the fraction between di and
di+1.
݀ҧ௜ is the mean diameter of fraction i and i+1.
ߩ௦ is the specific density of the particles.
Since the solid constituents of concrete mixtures seldom have spherical particle shape, some
error should be expected in the results from Eq (1). It has been found that the specific surface
area of the aggregate can be much larger than that of spheres of equivalent size (Wang and
Frost, 2003).
There are several ways of determination of SSA based on direct and indirect measurements,
e.g. Blaine test (ASTM C204, 2016), Lea and Nurse Method (Lea and Nurse, 1939). Both
tests give similar results but are not applicable to fine and ultra-fine powders. The Blaine test
method was developed exclusively for measurement of the specific surface area of cement
and is based on the assumption of spherical particle shape which leads to relative measures
for materials other than cement.
Another method that has been used to determine SSA is the volumetric static multi-point
method, better known as the BET method (Brunauer et. al., 1938). Results from BET test
include the measure of surface area of internal pores, which is not of interest for calculation of
water demand in concrete mixtures.
Determination of the SSA value using these three test methods includes complex measuring
devices. As a result developing a cheaper and easier to use method for estimation of SSA is
necessary. The main aim of this research was to verify the effect of the assumption of the
ideal polyhedron shapes of the particles instead of spheres on calculation of the SSA. For this
purpose, the specific surface areas of the particles were mathematically calculated based on
the size distribution curve and the assumption that particles have a uniform shape. The
particle shapes were substituted with the shape of standard platonic solids. The calculated
values were compared to the specific surface area of the samples measured using Blaine
method.
2. Materials
Four types of powders were used in this study and all required input data were extracted from
earier test results. Characteristics of the first three materials shown in Table 1 were extracted
from Hunger & Brouwers, (2009) and included specific density, bulk density, grading curve,
Blaine values, and Scanning Electron Microscope (SEM) images. The information on the
Quartz powder was obtained from Jennings et. al., (2013).
Table 1: Densities and SSA of the powders, from.(Hunger and Brouwers,( 2009) and Jennings
et. al.,(2013)
Material Specific density Loose packing Bulk density SSA based
3 3
(g/cm ) density (g/cm ) on Blaine
(cm2/g)
CEM III/B 42.5N 2.96 0.72 2.13 4500
Marble Powder 2.80 0.64 1.79 4580
Limestone 2.21 0.69 1.87 4040
Quartz Powder 2.60 0.64 1.66 2600
Particle size distribution curve of the materials was obtained by deploying low angle laser
light scattering technique (LALLS) conducted by (Hunger and Brouwers, 2009), Figure 2.
100
90 CEM III 42,5N
80 Marble Powder
Cumulative finer (Vol. %)
70 Limestone
Quartz
60
50
40
30
20
10
0
0,1 1 10 100 1000
Particle size (micron)
Figure 2. Particle size distribution curves. (Hunger and Brouwers, 2009)(Jennings et. al., 2013).
Additionally, in order to further validate the relation between polyhedrons and the particle
shape and to be able to distinguish the difference in particle geometry, SEM images from
Hunger and Brouwers (2009) Figure 3.
(a) CEM III/B 42.5 N (b) Dolomitic marble powder

(c) Limestone powder
Figure 3. SEM-SE images of studied powders, 1000X magnification. (Hunger and Brouwers, 2009).
3. Computation of SSA
3.1. Square-Cube law
The square-cube law defines a mathematical principle describing the relationship between the volume
and the area related to changes in size and was first introduced by Galilei and Drake, (1946).
According to the principle, as a shape grows in size, its volume grows faster than its surface area.
Consequently, as the size decreases its surface area grows faster than its volume. The effect of the
square-cube law becomes especially significant for calculation of specific surface area of finer
particles namely powders and cement i.e. for a given mass of aggregate, the surface area increases
with reducing particle size. The specific surface area can be calculated mathematically by assumption
of spherical shapes for the particle. In case when spheres were replaced by another shape, the
difference in calculations is caused the fact that different shapes have different volumes and also the
ratio between specific surface area and volume changes based on the chosen shape according to
square-cube law. Figure 4 shows the difference in pace of growth of SSA/Volume ratio of so called
Platonic solids - a set of five 3D regular convex polyhedrons - obeying the square-cube law.
400
350
300
250
Surface Area
200
150
100
50
0
0 50 100 150 200 250 300 350 400 450 500
Volume
Figure 4. Surface area versus volume of the platonic solids and a sphere. (Ghasemi et. al,2016)
The formula presented in Eq. (1) deals with a special case of calculating the SSA for spherical
particles, the equation can be written in its general form where the ratio of SSA/V implements
the square-cube law in the formula:
௡
ܵܵ‫ܣ‬௜ . ߱௜
ܽ௣௢௟௬ = ෍ (2)
ܸ௜ . ߩ௦
௜ୀଵ
where:
SSAi/Vi is the specific surface area to volume ratio of fraction i and is related to the shape as
shown in Table 2.
As mentioned before, it is possible to calculate SSA based on particle size distribution curves
and with the assumption of mono-shaped particles. The Platonic solids that were examined to
re-calculate the SSA are shown in Table 2. Substituting spheres with the platonic solids will
not only change the calculated volume and specific surface area but also affect the rate of
growth in SSA/Volume ratio according to the square-cube law.
Table 2: Platonic solids used in the calculation of SSA and their volumes.
Shape Surface Area Volume SSA/V
Tetrahedron ξ3ܽ ଶ
ξ2ܽଷ 14.697
12 ܽ
Cube 6ܽଶ ܽଷ 6
ܽ
Octahedron 2ξ3ܽଶ 1 7.348
ξ2ܽଷ
3 ܽ
Dodecahedron య 1 2.694
ට25 + 10ξ5ܽଶ (15 + 7ξ5)ܽଷ
4 ܽ
Icosahedron 5ξ3ܽଶ 5 3.970
(3 + ξ5)ܽଷ
12 ܽ
Sphere 4ߨܽଶ 4ߨܽଷ 3
3 ܽ
3.2. Equivalent polyhedron shape
The spherical surface area of each fraction can be calculated using Eq (1). To do so, the mean
diameter of the particle sizes di and di+1 of a fraction i as the characteristic particle size, is
required. The mean diameter di can be calculated using either arithmetic mean or geometric
mean, see . Eq (3) and Eq (4) respectively:
݀௜ +݀௜ାଵ
݀௜,௔௥௜௧௛ = (3)
2
݀௜,௚௘௢ = ට݀௜ ଶ + ݀௜ାଵ ଶ (4)

For the corresponding calculations for the polyhedrons, the length of the sides is needed since
the Platonic solids should be defined in relation to the spheres. This relation can be
conditioned based on geometric properties of the spheres using the concepts of circumsphere
and midsphere or by equivalent volume (mass) to the spheres. In geometry, a circumscribed
sphere or circumsphere of a polyhedron is a sphere that contains the polyhedron and touches
each of the polyhedron's vertices. Midsphere is defined as a sphere that touches all of the
polyhedron edges.
The midsphere does not necessarily pass through the midpoints of the edges, but is rather only
tangent to the edges at same point along their lengths (Cundy and Rollett, 1989). The length
of edges of platonic solids are smaller for the circumsphere approach comparing to
midsphere.
For volumetric equivalency the sides of the polyhedrons can be back-calculated by replacing
the volume of the polyhedrons by the volume of spheres assumed for each fraction, see Figure
4.
Circumsphere -Cube Midsphere - Cube
Equivalent volume
Figure 4. Circumsphere, Midsphere and volume equivalency of a cube.

It is also possible to define the equivalency based on the concept of Insphere. Insphere is
a sphere that is contained within the polyhedron and is tangent to each of the polyhedron's
faces. The issue with calculation based on Insphere is for some polyhedrons e.g. Tetrahedron,
only a relatively small sphere can be contained in comparison with other shapes. This would
affect the calculation of SSA and therefore calculation based on Insphere was ignored.
In this study different approaches have been examined to define the equivalent polyhedrons to
the spheres by utilizing the concepts of circumsphere, midsphere, and volume equivalency.
The computation was done based on both arithmetic and geometric means.
The lengths of sides of the polyhedrons were calculated for different assumptions:
x The polyhedrons are contained in spheres (circumsphere) with diameter calculation
based on arithmetic mean.
x The polyhedrons are contained in spheres (circumsphere) with diameter calculation
based on geometric mean.
x The sphere touches all of the polyhedron edges (midsphere) with diameter calculation
based on arithmetic mean.
x The sphere touches all of the polyhedron edges (midsphere) with diameter calculation
based on geometric mean.
x The polyhedrons have the same volume as the spheres (volumetric) with diameter
calculation based on arithmetic mean.
The polyhedrons have the same volume as the spheres (volumetric) with diameter
calculation based on geometric mean.
The edge lengths of the polyhedrons, a, were calculated by equations listed in Table 3.
Median radius of equivalent spheres, r, can be calculated by either Eq (3) or (4).
Table 3: Edge lengths of polyhedron.

Shape Circumsphered edge Midsphered edge length
length (ac) (am)
4‫ݎ‬ 4‫ݎ‬
Tetrahedron ξ6 ξ2
2‫ݎ‬ 2‫ݎ‬
Cube
ξ3 ξ2
2‫ݎ‬
Octahedron 2‫ݎ‬
ξ2
4‫ݎ‬ 4‫ݎ‬
Dodecahedron
ξ3(1 + ξ5) (3 + ξ5)
4‫ݎ‬ 4‫ݎ‬
Icosahedron
ඥ10 + 2 × ξ5 (1 + ξ5)
4. Results and discussion
Calculated specific surface areas according to Eq (2) and their corresponding Blaine values
obtained from laboratory tests of the referred previous studies are presented in Table 4 where
specific surface was calculated based on the size distribution curve assuming different
platonic solids.
In the case of most of the studied powders, the calculated spherical values of SSA were close
to the Blaine value. The marble powder showed the largest deviation and thus its calculation
should be conducted with the assumption of a different shape rather than spheres. This result
corresponds well to the observed elongated, flaky particle shape of that material, Figure 3.
Moreover, in case of CEMIII/B42.5, Limestone powder, and Quartz powder, the calculation
based on the assumption of spherical shape for the particles led to an overestimation of the
SSA. It should be noted that the less spherical is a particle, the greater is its specific surface
area. Since the particle shapes of powders are normally anything but spherical, therefore, the
SSA of the actual particles should be higher than the calculated one based on the assumption
of spherical shapes. Slight overestimation of spherical SSA can be related to the approach that
is taken for determining the mean diameter of a particle.
In the case of the Marble powder, the calculated spherical SSA has a more noticeable
difference to the Blaine value and is also the least spherical in terms of particle shape, Figure
3. It should also be mentioned that the Blaine test is a relative test designed for measuring
SSA of cement and not necessarily any non-spherical powder, in other words Blaine value is a
relative value and not absolute.
To sum up, among the studied scenarios used for defining the equivalent shape and mean
diameter, The assumption of midsphere equivalency and arithmetic mean results in less error
comparing to other approaches. See a compilation made in Figure 5.
As it can be seen in Figure 5, the assumption of spherical shape agrees with the Blaine values
for CEM III and Quartz. While cement particles usually have round shape, the same cannot be
said about Quartz. The reason that calculated SSA for quartz agrees with the spherical shape
could be related to the fact that the source of information for Quartz comes from a different
research (Jennings et. al., 2013).
In the case of Marble powder, assumption of cubical shape leads to a better estimation of
specific surface area, which this can be directly related to angularity of Marble grains.
Moreover, it should be noted that for the finer particles, there is a larger difference in the
calculated specific surface area for different shapes which can be related to the principle of
square-cube law.
It should also be mentioned that each column of data in Figure 5 shows the difference in
calculated specific surface area based on different shape for a given size distribution curve.
The difference in SSA for different shapes becomes more significant as the fine content of
studied materials increases as a result of the square-cube law, e.g. see the difference in SSA
for Limestone comparing to Quartz.
Table4: Calculated specific surface area of the powders

Calculation based on Circumsphere and Arithmetic mean
Blaine
CEMIII/B42.5 4500 13874 8014 8014 5822 5833 4625

Marble 4580 9596 5542 5543 4027 4036 3199
Limestone 4040 15478 8940 8940 6495 6512 5160
Quartz 2600 8116 4689 4688 3397 3415 2705
Calculation based on Circumsphere and Geometric mean
Blaine
CEMIII/B42.5 4500 14642 8458 8454 6140 6143 4881

Marble 4580 10112 5838 5838 4240 4245 3371
Limestone 4040 16325 9426 9426 6846 6851 5441
Quartz 2600 8562 4944 4944 3581 3594 2854
Calculation based on Midsphere and Arithmetic mean

Blaine
CEMIII/B42.5 4500 8011 6543 5664 5433 5215 4625

Marble 4580 5540 4525 3917 3761 3644 3199
Limestone 4040 8936 7300 6319 6064 5812 5160
Quartz 2600 4685 3828 3313 3005 2995 2705
Calculation based on Midsphere and Geometric mean
Blaine
CEMIII/B42.5 4500 7625 6904 5978 5743 5224 4881

Marble 4580 5164 4767 4128 3969 3608 3371
Limestone 4040 8534 7696 6665 6404 5824 5442
Quartz 2600 4656 4037 3495 3354 3054 2584
Calculation based on equivalent volume and Arithmetic mean

Blaine
CEMIII/B42.5 4500 6892 5737 5469 5078 4927 4625

Marble 4580 4766 3968 3784 3515 3406 3199
Limestone 4040 7689 6401 6103 5668 5495 5160
Quartz 2600 4031 3355 3200 2964 2881 2705
Calculation based on equivalent volume and Geometric mean
Blaine
CEMIII/B42.5 4500 7273 6056 5772 5358 5195 4881

Marble 4580 5022 4182 3986 3703 3588 3371
Limestone 4040 8109 6751 6434 5977 5791 5442
Quartz 2600 4253 3541 3375 3128 3037 2584
9000
Limestone
8000
CEM III
7000
Tetrahedron
Calculated SSA
Cube
6000
Octahedron
Marble
Dodecahedron
5000
Icosahedron
Quartz
Sphere
4000
Equality line
Linear (Equality line)
3000
2000
2000 2500 3000 3500 4000 4500 5000
Blaine value
Figure 5. Blaine values vs. calculated SSA based on Midsphere-arithmetic mean assumption.
5. Conclusion
The specific surface area of aggregates, fillers and binders affects the fresh and hardened
concrete properties. The water layer theory is a potentially useful tool for prediction of fresh
concrete flow ability. However, the complexity of the analytical instruments required to
measure the specific surface area limits a wider usage of that approach for e.g. concrete mix
design. An alternative is to formulate an equation enabling theoretical prediction of the
specific surface area using only simple input data. The results of the present study showed
that while assumption of spherical shape for particles leads to an acceptable estimation of the
specific surface area for round particles, in case of more angular flaky particles, substituting
polyhedrons with sphere improves the accuracy of specific surface area estimation.
Acknowledgement
The authors would like to thank Dr. Martin Hunger for his guidance and The Swedish
Research council (Formas) for the financial support.
References
Alexander, M., and Mindess, S., Aggregates in concrete. CRC Press, (2010).
Wang, L.B. and Lai, J.S., Quantify specific surface area of aggregates using an imaging
technique. TRB. (1998).
Hunger, M., An integral design concept of ecological self-compacting concrete. PhD thesis,
Eindhoven University. (2010)
Brouwers, H. J. H., and H. J. Radix. Self-compacting concrete: theoretical and experimental

study. Cement and Concrete Research 35.11 (2005): 2116-2136.
Wang, L,. and Frost, D.J., "Quantification of the specific aggregate surface area using X-ray
tomography." Recent Advances in Materials Characterization and Modelling of Pavement
Systems. ASCE, 2003.
Kwan, A.K.H. and Mora, C.F. Effects of various shape parameters on packing of aggregate
particles. Magazine of Concrete Research, Vol. 53 (2), (2001) 91-100.
McCabe, W.L., Smith J.C., and Harriott P., Unit operations of chemical engineering. Vol. 5.
New York: McGraw-Hill, (1993)
Standard Test Methods for Fineness of Hydraulic Cement by Air-Permeability Apparatus.

ASTM C204, 2016
Lea, F. M., and R. W. Nurse. The specific surface of fine powders. J. Soc. Chem. Ind. 58.9
(1939): 277-283.
Brunauer, S., Emmett, P. H. and Teller, E. Adsorption of Gases in Multimolecular Layers,

Journal of the American Chemical Society 60(2) (1938): 309 – 319.
Galilei, G., Drake, S., Two new sciences. Madison, WI: University of Wisconsin Press,
(1946).
Hunger, M, and H. J. H. Brouwers. Flow analysis of water–powder mixtures: Application to

specific surface area and shape factor. Cement and Concrete Composites 31.1 (2009): 39-59.
Jennings, H., Kropp, J., Scrivener K., The modelling of microstructure and its potential for
studying transport properties and durability. Vol. 304. Springer Science & Business Media,
(2013).
Cundy, H. and Rollett, A. Mathematical Models, 3rd ed. Stradbroke, England: Tarquin Pub.,
(1989). P. 117.
Fraaij, A.L.A. and Rooij, M.R. The workability of concrete: is there an easy way to produce
self-compacting concrete? In: Dhir, R.K., Hewlett, P.C., Csetenyi, L.J. and Newlands, M.D.
Role for concrete in global development. Dundee, Scotland, UK, (2008) 387-396.
Fennis, S. A. A. M. Measuring water demand or packing density of micro

powders: comparison of methods. (2008).

Full Text 01

Transféré par

Informations du document

Copyright

Formats disponibles

Partager ce document

Partager ou intégrer le document

Options de partage

Avez-vous trouvé ce document utile ?

Ce contenu est-il inapproprié ?

Droits d'auteur :

Formats disponibles

Full Text 01

Transféré par

Droits d'auteur :

Formats disponibles

L ICE N T IAT E T H E S I S

Department of Civil, Environmental and Natural Resources Engineering

Aggregates in Concrete Mix Design

Yahya Ghasemi Aggregates in Concrete Mix Design

Luleå University of Technology 

Printed by Luleå University of Technology, Graphic Production 2017

Room F1031, Luleå University of Technology

Discussion leader: Dr. Hans-Erik Gram, Cementa AB

Denna licentiatuppsats behandlar de vanligaste principerna för betongproportionering

Det har observerats att partikelpackningsmodeller kan förutsäga packningsdensiteten med

Nyckelord: partikelpackning, teorier för överskott av vatten/cementpasta, specifik yta,

Įi Actual packing density of class i [-]

Įagg Packing density of aggregates [-]

Įk Steepness of Kp function [-]

Įt Calculated packing density [-]

ȕi Virtual packing density of class size i [-]

ȕj Virtual packing density of class size i [-]

ȕti Calculated virtual packing of a mixture size i being dominant [-]

ȜQ Flow resistance ratio [-]

ȡb Bulk density [kg/m3]

ȡs Relative density [kg/m3]

a Polyhedron edge length [mm]

ac Circumsphered edge length [mm]

aij Coefficient presenting the loosening effect [-]

am Midsphered edge length [mm]

asph Spherical specific surface area [cm2/cm3][cm2/Kg]

Ap Specific surface area of particles [cm2/cm3][cm2/Kg]

bij Coefficient presenting the wall effect [-]

di Diameter of dominant particle size class i [mm]

di,arith Arithmetic mean of fractions i and i+1 [mm]

di,geo Geometric mean of fractions i and i+1 [mm]

dj Diameter of dominant particle size class j [mm]

D Diameter of particles [mm]

Dmax Maximum particle size of the aggregate in the mix [mm]

Dmin Minimum particle size of the aggregate in the mix [mm]

e Void content [-]

Fi Accumulated flow of an ideal fluid through FlowCyl [m3/s]

Fmp Fineness modulus for the coarse aggregate [-]

Fms Fineness modulus for the sand [-]

Fp Volume fraction of fluid phase of concrete [-]

Ft Accumulated flow of test material through FlowCyl [m3/s]

Hm The air void modulus [-]

Hp Air void ratio in the coarse aggregate [-]

Hs Air void ratio in the sand [-]

K Compaction index [-]

mi Mass percentage of the fraction between di and di+1 [Kg/Kg]

mp Mass of all particles in a mixture [Kg]

Ms Mass of the solid [Kg]

P(D) Fraction that pass the sieve with opening D [-]

r Median radius of equivalent sphere [mm]

SSAi Specific surface area of particles in fraction i [cm2/cm3][cm2/Kg]

Tp Aggregate parameter of the coarse aggregate [-]

Ts Aggregate parameter of the sand [-]

Vagg Volume of aggregates [m3]

Vco Volume fraction of coarse aggregate [-]

Vcp Volume of cement paste [m3]

Vep Volume of excess paste [m3]

Vew Volume of excess water [m3]

Vi Volume of fraction i [m3]

Vp Volume of internal pores [m3]

Vs Volume of the solid [m3]

Vsa Volume fraction of sand in the aggregate [-]

Vt Total volume of the solid [m3]

Vv Volume of the external granular voids [m3]

wKHUHĮt LVFDOFXODWHGSDFNLQJGHQVLW\RIDPL[WXUHȕ1 is the packing density of the larger size