Zhou - Mixture Design Methods For Ultra-High-Performance Concrete - A Review - 2021

Cement and Concrete Composites 124 (2021) 104242
Contents lists available at ScienceDirect
Cement and Concrete Composites

journal homepage: www.elsevier.com/locate/cemconcomp
Research Paper
Mixture design methods for ultra-high-performance concrete - a review

Min Zhou a, b, Zemei Wu a, b, **, Xue Ouyang a, b, Xiang Hu a, b, Caijun Shi a, b, c, *
a
Key Laboratory for Green & Advanced Civil Engineering Materials and Application Technologies of Hunan Province, College of Civil Engineering, Hunan University,
Changsha, 410082, PR China
b
Key Laboratory of Building Safety and Energy Efficiency of Ministry of Education, Changsha, 410082, PR China
c
Department of Civil Engineering, The University of British Columbia, 6250 Applied Science Lane, Vancouver, BC, V6T 1Z4, Canada
A R T I C L E I N F O A B S T R A C T
Keywords: A highly efficient mixture design should have a good balance among workability, strength, durability, economic
Artificial neural network method efficiency, and sustainability. Ultra-high-performance concrete (UHPC) is a relatively novel composite material
Close packing density with ultra-high strength, high toughness, and superior durability, although new solutions are constantly being
UHPC
developed. Previous mixture design methods for normal concrete and high-performance concrete involve a series
Mixture design methods
Rheological properties
of trial tests and empirical analyses, which cannot be directly applied to UHPC without modification. This paper
Statistical mixture design reviews four commonly used mixture design methods, including close packing methods based on dry and wet
packing densities, rheology-based method, statistical analysis method, and artificial neural network (ANN)
model, according to design principles. The design procedures of these methods and the effect of mixture
proportioning on the key performance of UHPC are presented and discussed. The advantages and disadvantages
of these methods are finally compared. Depending on the actual production situation and performance re
quirements, an appropriate method should be chosen to develop UHPC with specified requirements. Integration
of two or more methods can serve as a good approach to deal with the complexities of UHPC mixture design by
exerting each method’s advantages.
1. Introduction 5 MPa [7]. JSCE (Japan) [5] and France [6] agree that the compressive
strength of UHPC is 150 MPa, and the minimum tensile strengths are 5
Ultra-high-performance concrete (UHPC) is a relatively novel com and 8 MPa, respectively. ASTM (United States) [4] and KCI (South
posite material characterized by ultra-high strength, great toughness, Korea) [8] recommend a minimum compressive strength of 120 and
and excellent durability due to low water-to-binder (w/b) ratio of 180 MPa, respectively, for UHPC. However, in practice and laboratory
around 0.2 and the use of superplasticizer and steel fibers [1,2], conditions, the compressive strength of UHPC can reach up to 250 MPa
although new solutions are constantly being developed. France, China, [9], and even 400 MPa [10] due to heat and autoclave curing. UHPC is
the United States (USA), Japan, and South Korea, etc., have formulated widely applied in bridges, explosion-resistant structures, thin-walled
relevant standards and specifications on design, testing, and applica structures, architectural ornaments, marine structures, and rehabili
tions of UHPC [3–7]. Due to the different materials and sample sizes tated and strengthening members [11–13]. However, the use of high
used in each country or region, the definition of UHPC regarding contents of superfine particles and superplasticizer can lead to high
compressive, flexural, and tensile strengths of UHPC are inconsistent viscosity and autogenous shrinkage of UHPC [14–16]. These eventually
with each other. For example, the compressive and flexural strengths of cause difficulty in castability and impair hardened performance of UHPC
UHPC should be greater than 100 and 12 MPa, respectively, according to due to entrapped air bubbles, non-uniform dispersion, and orientation of
GB/T 31387 from China [3]. Asian Concrete Federation (ACF) defines steel fibers [17]. Considering the conflicting requirements of these
UHPC a cementitious composite containing discrete fibers for tensile performances, conventional mixture design methods for normal con
post-cracking ductility with a minimum specified compressive strength crete and high-performance concrete based on empirical parameters
of 120 MPa and a flexural strength of 14 MPa or direct tensile strength of cannot directly deal with these complexities without appropriate
* Corresponding author. Key Laboratory for Green & Advanced Civil Engineering Materials and Application Technologies of Hunan Province, College of Civil
Engineering, Hunan University, Changsha, 410082, PR China.
** Corresponding author. Department of Civil Engineering, The University of British Columbia, 6250 Applied Science Lane, Vancouver, BC, V6T 1Z4, Canada.
E-mail addresses: [email protected] (Z. Wu), [email protected] (C. Shi).
https://doi.org/10.1016/j.cemconcomp.2021.104242
Received 26 May 2021; Received in revised form 25 August 2021; Accepted 27 August 2021
Available online 11 September 2021
0958-9465/© 2021 Elsevier Ltd. All rights reserved.
M. Zhou et al. Cement and Concrete Composites 124 (2021) 104242
modification [18,19]. It requires extensive trial tests due to the lack of fiber can enhance tensile properties and restrain shrinkage of UHPC
theoretical basis [20]. The mixture design of normal concrete is mainly through inhibiting crack propagation [60]. The efficiency of fibers de
based on the water-cement ratio (W/C) law [21], and adjusts the pends on volume, aspect ratio, shape, distribution, type, strength and
workability of concrete by changing the water consumption [19]. elastic modulus, and other characteristics [61]. Yoo et al. [62] investi
Traditional ACI mixture design method gives priority to the workability gated the influence of fiber volume, shape, and aspect ratio on the
of concrete, followed by strength, durability, and economy [22,23]. flexural properties of UHPC. It was found that the optimal volumes of
Although it has been successfully applied in many engineering appli long twisted fiber and medium-length straight fiber to effectively
cations, it often leads to high cement consumption if applied to UHPC improve the flexural strength of UHPC are less than 1% and 1.5%,
[24]. Unlike normal concrete, the key to the UHPC mixture design is to respectively. Wu et al. [63] compared the effects of three shaped steel
produce a tightly packed structure through optimizing particle size fibers with same length and diameter, including straight, corrugated,
distribution [25]. This requires strictly selected materials to fill the pores and hooked-end, on the performance of UHPC made with volumes of 0,
to achieve a possible maximum packing density. The mixture design for 1%, 2%, and 3%. It is showed that the 28-d compressive strengths of
UHPC is more complicated and can be considered a multi-objective UHPC using corrugated and hooked-end fibers were increased by 48%
optimization process [26]. Therefore, it is essential to develop scienti and 59%, respectively, while the flowability values were decreased by
fic and efficient mixture design methods to produce UHPC meeting re 45.1% and 51.2%. Another investigation by Wu et al. [60] that the
quirements of workability, strength, durability, and economy. optimal steel fiber volume for enhancing strength and toughness and
The properties of UHPC are governed by mixture proportion and restraining shrinkage of UHPC was 2%, regardless of fiber shape with a
curing methods, etc. [27–29]. The binder content in UHPC is high of length of 13 mm and a diameter of 0.2 mm. It should be noted that the
800–1000 kg/m3 [1,30,31], And only 30%–40% of cement content increase in steel fiber content can inevitably increase the UHPC cost.
participates in hydration due to the low w/b ratio of UHPC. Thus, a large Steel fiber of 2% volume take up to approximately 50% of the total cost
amount of unhydrated cement is acted as inactive filler [32]. Supple of UHPC.
mentary cementitious materials (SCMs), such as fly ash (FA), ground Hybrid fibers with different sizes [64,65], shapes [63], and types
granulated blast-furnace slag (GGBS) [33], silica fume (SF), metakaolin [66], such as steel, carbon, wollastonite, basalt, PVA, PP-PE, biochip
[34], limestone powder [35], steel slag powder [36], and rice husk ash fiber, are employed to minimize fiber content and exert their combined
[37] are used to reduce cement consumption, carbon emission, and strengthening effects. The use of PVA within 0.5% did not significantly
enhance flowability without sacrificing mechanical properties [38–41]. affect the elastic modulus of UHPC [67]. The 28-d compressive strength
UHPC with a binder content of 500 kg/m3 hardly flows, while mixture of UHPC with 1.2% PVA was 37.5% higher than that without PVA [68].
with a binder content of 700 kg/m3 has a slump of 130 mm [35]. The The tensile strain capacity of UHPC with 33% steel fibers replaced by PE
compressive strength of UHPC is lower than 100 MPa and sharply fiber was increased by 29.7% [69]. UHPC mixed with 1.5% long and
reduced when the amount of cement replacement increased from 50% to 0.5% short fibers exhibited the best static and dynamic mechanical
73%, which is only 280 kg/m3 [42]. UHPC incorporated 20%–30% FA properties [65]. The increase of fiber aspect ratio can promote the
content showed the highest compressive strength and static elastic possibility of non-uniform fiber dispersion and flocculation in UHPC,
modulus [43]. Yalçınkaya et al. [44] produced UHPC with compressive and the fiber volume significantly affects the workability of UHPC [70].
strength over 145 MPa by replacing cement with 30% FA or 50% GGBS. However, uniform fiber dispersion and orientation issues and reduction
Wu et al. [45] pointed out that the addition of 20% GGBS or 3.2% in rheology properties are still main difficulties in the production of
nano-CaCO3 can densify the microstructure of UHPC and increase UHPC [71,72]. Besides, increases in aggregate size and content, espe
fiber-matrix bond strength by 30%–48%. Besides, the addition of SCMs cially the use of coarse aggregate in UHPC, can further intervene with
in UHPC effectively improves the mechanical strength and durability of fiber dispersion and distribution. This is mainly due to the interlocking
UHPC through the filler and pozzolanic effects [44,45]. Superplasticizer between the coarse aggregate and fibers, and the insufficient fiber length
is the core that renders UHPC with a low w/b ratio. Although 0.5% to completely wrap the coarser aggregate, thus eventually affecting the
superplasticizer can significantly increase the flowability of cement flowability and mechanical properties of UHPC [73]. Liu et al. [74]
paste more than 3 times [46], the addition of 0.2% superplasticizer found that when the coarse aggregate content reached 35%, the slump of
delays the initial setting time of concrete for nearly 30 minutes [47]. the fresh UHPC was only 130 mm. Therefore, selecting the type and
Besides, the superplasticizer molecules interfere with the growth of quantity of individual constituents to yield UHPC that meets specifiable
hydration products in pore solution and weaken the positive effects of characteristics for a particular application should be discussed.
SCMs, which is unfavorable to early strength development of UHPC [48, The development of mixture design methods for UHPC has become a
49]. In the research of Li et al. [50], 3% of nano-silica, by mass of research hotspot, especially in recent years. In light of the concept that
cement, is the optimal content to reduce the retardation of UHPC smaller particles fill voids between larger particles, thus reducing void
induced by superplasticizer. volume and increasing packing density of solid constituents, several
In traditional UHPC mixtures, fine aggregate is used to increase close packing models have been proposed [1,75,76]. After Furnas model
homogeneity and eliminate inherent weakness, such as defects of for the binary mixture was put forward in 1931 [75], Toufar model,
interfacial transition zone (ITZ) between matrix and coarse aggregate. Linear packing density model (LPDM), and Compressible packing model
Nowadays, some researchers attempt to introduce coarse aggregates (CPM) were developed. Despite the most significant error occurred at
into UHPC to further reduce cost and achieve better workability [51], the maximum packing density, the linear packing density model (LPDM)
mitigate autogenous shrinkage, and improve projectile impact resis for multi-component mixtures can be applied for concrete, as was
tance [52,53]. The total volume of coarse and fine aggregate accounts pointed out by Stovall et al. [77]. The improved compressible packing
for more than 75% of the total volume of concrete [54,55], and the model (CPM) introduces compaction index and virtual packing concepts
coarse aggregate incorporated is usually 20%–50% [56]. The optimal and successfully develops UHPC with compressive strength over 200
coarse aggregate content for UHPC is affected by its maximum particle MPa [78]. Using a compressible packing model and considering the
size, fiber type, and volume [57,58]. When the maximum aggregate size rheological properties of paste, a suitable binder type was selected to
changed from 3 to 16 mm, the 28-d compressive strength of UHPC design UHPC [55,79]. However, these particle packing models are
decreased by 12 MPa [52]. Although the addition of coarse aggregate established on the packing density of dry particles. The effects of water
can increase workability and reduce the drying shrinkage of UHPC, and superplasticizer on wet packing density are not taken into account,
excessive coarse aggregate content can introduce more ITZs, thus which results in an inaccurate prediction of UHPC properties. Consid
resulting in lower compressive strength, higher permeability, and even ering the effect of liquid in concrete, Larrard et al. [80] used concepts of
impair structural quality [59]. As the essential component of UHPC, wet packing density and maximum paste thickness to design UHPC with
2
satisfactory performance. Hoang et al. [81] proposed an iterative opti effect of liquid phase and its interaction with solid particles of mixture.
mization method to optimize the mixture design of UHPC. Fan et al. [82] Therefore, separating the packing and interaction between particles in
and Yu et al. [83] designed an environmentally friendly UHPC by dry and wet states is indispensable.
achieving the maximum wet packing density using novel approaches, Dry packing density refers to the proportion of solid volume to the
including statistical methods of D-optimal design and response surface total volume of system, as shown in Eq. (1). Fig. 1 illustrates the packing
method (RSM). RSM and D-optimal design methods were found to have density diagram of a granular mixture system. A decrease in binder
the potential to predict the influence of aggregate proportion, admix content and porosity in UHPC can be achieved with a better dry packing
ture, and fibers on UHPC performance [84–86]. Ghafari et al. [87] density of solid materials. Improvement in packing density of solid
proposed an artificial neural network (ANN) model derived from bio skeleton is crucial for the UHPC produce. In mixtures containing
logical systems to predict UHPC performance. They found that it is more aggregate with multiple particle sizes, assuming that at least one
efficient and accurate than traditional statistical methods. Such as aggregate is filled is indispensable, ensuring that the mixture is
backpropagation (BP) algorithm in ANN, can accurately obtain output compacted.
value by varying the weight of input value [88,89].
Vs1 + Vs2 + ... + Vsn Vs
Although different mixture design methods have been developed, dry packing density (φ) = = (1)
Vt Vt
most of the proposed mixture design methods involve a series of trial
tests and empirical reference values. Due to different principles, the where, φ is the packing density (unit-less); Vs is the volume of each solid
design methods exhibit varying characteristics and applicability. To (ml); Vt is the sum of solids and pore volume (ml).
design UHPC with target performance, it is indispensable to compre The value of dry packing density fluctuates due to external
hensively review and compare the principles, procedures, and advan compaction energy exerted and agglomeration among fine particles
tages and disadvantages of these commonly used mixture design under dry state. When cementitious materials contact water, agglom
methods. Table 1 summarizes the design principles and features of these eration easily occurs because of the high surface energy associated with
methods. This review focuses on four categories of mixture design the high specific surface area of fine particles. Besides, the further dried
methods based on design principles, which include close packing water molecules onto particles form hydrogen bonds that bridge with
methods divided into dry or wet packing, method developed from each other, which trigger capillary to shrink and further promote par
rheological properties of paste, statistical design method, and ANN- ticle agglomeration [115]. Unlike dry packing density, wet packing
based method. The basic design principle, procedures, features, and density considers the effects of water and superplasticizer on fine par
advantages and disadvantages of each method are described and ticles and can be calculated according to Eqs. (2) and (3) [116,117].
compared. The influences of key raw materials and interaction among Detailed information regarding the measurement of wet packing of
superplasticizer, aggregate, fibers, and other components on the design particles and its effect on the performance of cement-based material will
efficiency and performance of UHPC are discussed. This paper can be explained in Section 2.1.2.
provide a reasonable scientific basis for the production of UHPC with
meeting requirements of workability, mechanical properties, durability, Vc =
Mmax
and economy. ρw uw + ρSP uSP + ρα Rα + ρβ Rβ + ργ Rγ (2)
2. Mixture design methods of UHPC VC

wet packing density (φ) = (3)
V
2.1. Close packing method
where, φ is the packing density (unit-less); uw and uSP represent the ratio
Studies have proved that UHPC with a densely packed skeleton can of water and superplasticizer volumes to solid volume, respectively
be designed using a particle packing model [76,83,112–114]. Never (unit-less); ρw and ρSP are the densities of water and superplasticizer,
theless, the existing particle packing models without considering the respectively (kg/m3); V is the mold volume (ml); VC is the total solid
volume (ml); ρα, ρβ, ργ are the solid densities of components α, β, and γ,
Table 1 respectively (kg/m3); Rα, Rβ, Rγ represent the volumetric ratio of com
Commonly used mixture design methods for UHPC. ponents α, β, and γ to the total volume of materials in the system,
respectively (unit-less).
Methods Design principles Ref.
UHPC with close particle packing structure is achieved by applying
Close packing method Ensure the tightest packing of the solid [75,77,
particle packing model considering particle shape, content, grading, and
particles of UHPC through varying the 90–99]
proportion of cementing components size. The voids in cement-based materials with a mono size of aggregate
and fine aggregates, and achieve higher are primarily introduced by insufficient filling between aggregates, as
strength and durability. shown in Fig. 2(a). The addition of smaller aggregate sizes reduces
Method based on Based on the relationship between [16, porosity due to the packing of variable sizes particles, as shown in Fig. 2
rheological properties rheology and raw materials, the type 100–102]
(b). Besides, multi-particle aggregate leads to extra paste on the aggre
of paste and proportion of each component are
continuously adjusted to achieve the gate surface at constant binder content that forms a film with a certain
target rheological properties of UHPC to thickness that improves concrete flowability, as shown in Fig. 2(c) [18].
ensure the best performance. The accuracy of close packing model is governed by the interaction
Statistical mixture The proportion of raw materials is [84,85,
between particles, such as wall effect, loosening effect, and wedging
design method regarded as variable, and the 103–108]
performance of UHPC is treated as a effect, as illustrated in Figs. 3 and 4 [94,118,119]. The wall effect sug
response. Determine the optimal gests that large particles prevent the dense packing of small particles.
mixture proportion of UHPC through When the volume of small particles exceeds pore volume between large
employing relevant models to analyze particles, loosening effect occurs. The wedging effect presents two main
experimental results with eliminated
forms. One is that a single fine particle is sandwiched in the gap between
errors.
Mixture design method Train ANN model by using many [87,88, coarse particles, and the other is that the fine particle layer wrapping on
based on ANN model experimental data and constantly adjust 109–111] the surface of coarse particles is disconnected somewhere. The geo
the weights of input and output to metric interactions between particles should be considered during the
reduce the gap between experimental
mixture design of UHPC.
and predicted values.
3
Fig. 1. Packing density diagram of a granular mixture system.
Fig. 2. Packing diagrams of mono-sized and multi-sized aggregate (adapted from Ref. [18]).
Fig. 3. Schematic diagrams of particle loosening and wall effects (adapted from Ref. [120]).
2.1.1. Close packing models based on dry packing density Stovall, De Larrard, Sedran, Glavind, and Dewar are suitable for
Various particle packing models have been proposed to estimate the multi-component mixtures [77,80,121–123]. In 1907, the grading curve
packing density of granular materials at dry state with considering of maximum density of particles was proposed by Fuller and Thomson
structure and interaction effects, such as filling, wedging, loosening, and [96], which laid the foundation of the continuous model.
wall effects [119,120], including Furnas [75], Powers [90], Aim and In terms of classification, the close packing models can be either
Goff [91] for binary mixture, Toufar and modified Toufar model discrete or continuous. The use of a single particle packing model always
extended binary mixture to ternary mixture [93]. The LPDM, Solid brings general mathematical problems in the prediction of particle-
suspension model (SSM), and CPM in discrete models were proposed by graded concrete. When the particle size distribution (PSD) curve of
4
Fig. 4. Illustration of particle wedging effect (adapted from Ref. [119]).
mixture gets closed to the close packing model, the mixture proportion was applied to concrete mixture design, as shown in Eq. (5).
can be considered the target. ( )n
d
CPFT = 100 (4)
D
2.1.1.1. Discrete models. The discrete model assumes that the volume of
each particle can be fully compressed according to specific particle size,
CPFT = Tn (di − d0 )n (5)
and the development of the discrete model is shown in Table 2. In the
Furnas model [75], particles were assumed to not interfere with each where, CPFT is the percentage of cumulative volume (unit-less); n is the
other and PSD is continuous, which was unsuited for multi-component degree of curve equation and value is 0.45 in Eq. (5); T is a coefficient
mixtures, such as UHPC. Subsequently, Powers [90], Aim, and Goff related to the maximum diameter of aggregate (unit-less).
et al. [91] used a simple geometric model to explain the function of Andreasen proposed an ideal packing equation (i.e., AA model) that
porosity for binary mixture. They considered the interference coefficient assumes the smallest particle is infinitesimal. However, this model is still
of wall effect to modify effect on porosity. In the Toufar model, three empirical. Later, Dinger and Funk modified the model considering the
idea assumptions were considered: 1) aggregates are spherical; 2) mono minimum size of PSD, as shown in Eq. (6). Rosin-Rammler-Benuet
aggregate size; 3) sizes of fine and coarse aggregates are different [124], equation described the characteristic diameter of concrete PSD, as
which extends the model from binary to ternary mixtures [93]. Roy et al. shown in Eq. (7).
[92] first modified Toufar model and changed the design parameters. ( )q
The dry packing density suffers from particle shape, distributions, and CPFT =
(d − d0 )
100 (6)
the way of packing if particle surface forces were ignored. LPDM, SSM, (D − d0 )
and CPM were proposed for multi-component mixtures to transfer ( )n
traditional particle packing to virtual packing and modify wall and P = exp
− R
(7)
loosening effect [77,80,121–123]. These discrete models are suitable for R′
multi-component mixtures and have considerable effects in predicting
the packing density of dry particles. However, both the first Furnas where, d and R are the diameters of particle (μm); d0 is the minimum
model and subsequent CPM model exhibit a common defect. Only dry diameter in PSD of particle (μm); D is the maximum diameter of particle
particles are in the mixture system, which is quite different from the (μm); q and n are the distribution coefficients (unit-less); P is the per
actual situation of concrete. centage of residual fraction (unit-less); R′ is the characteristic diameter
(μm).
From the above model, Wang et al. [126] established a relationship
2.1.1.2. Continuous models. With regard to the continuous methods, it
between PSD and dry packing density according to
assumes PSD is continuous and is expressed in densification with a small
Rosin-Rammler-Benuet equation. This equation applies in the following
particle system. It means that all possible particle sizes exist in the
two conditions: 1) cement particles are spherical and particle size does
system. In 1907, Fuller and Thomson [96] proposed a grading curve of
not affect the degree of hydration; 2) packing density of particles with
the maximum density of particles called Fuller’s curve, as shown in Eq.
the same size is independent of particle size. Yu et al. [39] used the
(4). The Fuller model is modified by Shakhmenko and Birsh [97] and
Table 2
Summarization of discrete models for particle packing (adapted from Ref. [125]).
Year Models Packing system Effect on particle packing Main features Ref.
Binary Ternary Multi- Wall Loosening

component effect effect
1929 Furnas model ✓ Assumed that particles are independent [75]

1967 Aim and Goff ✓ ✓ Proposed coefficient of wall effect [91]
model
1969 Powers model ✓ ✓ ✓ Considered wall and loosening effects [90]
1977 Toufar model ✓ ✓ Divided ternary system into multiple binary systems [93]
1986 LPDM ✓ ✓ ✓ Derived packing density when particles are continuously [77,
distributed 121]
1994 SSM ✓ ✓ ✓ Proposed virtual packing density and is widely used [80]
1999 CPM ✓ ✓ ✓
5
improved AA model to design UHPC and adopt the least square method effect depends on superplasticizer type and concentration. Results
(LSM) optimization algorithm to optimize the mixture proportion and showed that the wet packing density of paste mixed with
reduce the error introduced by considering the packing of dry particles. polycarboxylate-based superplasticizer is higher than that of
Hoang et al. [81] proposed a progressively optimized packing of dry naphthalene-based superplasticizer [98]. At low superplasticizer
particles to design the UHPC mixture proportion. In this method, par dosage, the wet packing density of paste is linearly increased with
ticle packing needs to satisfy an assumption that particles in the group superplasticizer dosage [130]. Superplasticizer can improve the wet
(i+1) were always thinner than that in the group (i), and then both of packing density of particle mainly by its dispersion that including
them are combined. If they were compatible, a new combination of electrostatic repulsion and steric hindrance. The dispersion capability of
packing density is obtained. Continuing to combine with group (i+2) superplasticizer is controlled by molecular weight, side-chain density,
and repeating this process to get a combination with the largest packing side-chain length of the polymers, and the superplasticizer amount
density. Fig. 5 illustrates the iterative process of obtaining the maximum adsorbed on the surface of particles [131]. Particles flocculate with each
dry packing density. This iterative packing method takes advantage of other due to the surface charge of the particles at the initial stage of
the diversity of particle size and breaks the limitation of single-size hydration [130,132]. The anions in the superplasticizer that adsorbed
particle packing. This method essentially meets the requirement of onto cement surfaces neutralize the charge, generate electrostatic
continuous particle size, and gradually minimizes the pores in contin repulsion, and weaken the van der Waals force between particles [49,
uous iteration. 129,133]. Steric hindrance arises from the presence of polymers
The mixture design method for UHPC considers dry packing density adsorbed onto particle surfaces, and side-chain length of the polymers
is established under dry conditions without considering the role of has dominant effect on steric resistance [133].
mixed solutions, such as water and superplasticizer. Besides, the dry It is showed that numerous irregular flocculation structures with
packing of particles is greatly governed by the degree of compaction and large size exist in the cement suspension without superplasticizer, as
particle interaction related to particle size, content, and shape. There shown in Fig. 7 (a). However, the flocculation size decreases in the
fore, particle agglomeration might occur and influence the accuracy of presence of superplasticizer (Fig. 7 b) [134]. The water wrapped in the
dry packing density, which affects the performance of UHPC. Therefore, cement particles is released, increasing wet packing density [130,135].
wet close packing method considering the influence of liquid phase is Part of water released inside the particles after mutual repulsion is used
proposed [117,127,128]. to fill the pores of the particles, and other part is used to render thicker
water film onto particles, as shown in Fig. 8, which directly increases the
2.1.2. Close packing models considering wet packing of particles flowability of UHPC [136].
The wet packing density of particles refers to the sum of the mass of The effect of water on particle aggregation is complex, which is
particles and water in voids per unit volume in a wet-mixed state. The classified into two main aspects: the formation of chemical bonds and
maximum wet packing density directly suffers from particle character liquid bridges. The chemical bond is primarily governed by a hydrogen
istics, superplasticizer, the thickness of water film onto solid particles, bond that is formed by the combination of OH− and H+ in water with
degree of particle dispersion, and agglomeration. Due to the filling effect ions on the surface of particles. Besides, the liquid bridge appears in
of fine particles, such as SF, the release of water entrapped in voids of water adsorbed on the particle surface and in the capillary, promoting
particles occurs, increasing the wet packing density of UHPC mixture. the agglomeration of particles [137]. The agglomeration of particles in
However, the pores of the surrounding material will be formed in water leads to a decrease in wet packing density, which is opposite to
agglomerated SF with large size due to the loosening effect, reducing the that of superplasticizer. In addition to filling pores, ultrafine particles
wet packing density of the system, as shown in Fig. 6. also affect the packing density of the mixture through hydration reac
The effect of superplasticizer on the packing of particles can be tion. The gradual diffusion of ultrafine particles in water provides
summarized as dispersion and can increase the wet packing density. The nucleation sites and then promotes hydration, leading to densified
Fig. 5. A new mixture design method based on stepwise optimization of particle dry packing density for UHPC (based on Ref. [81]).
6
Fig. 6. SEM images of size and shape of agglomerated and original SF [129].
Fig. 7. Optical microscope images of cement suspension with and without superplasticizer [134].
Fig. 8. Comparison of dispersion of particles with and without superplasticizer.
C–S–H gels with needle-like morphologies [138–141]. Therefore, the water consumption, and reduced WFT and flowability of cement paste
particles in water exhibit a multi-component, multi-size, and multi-form [142,143]. Therefore, the relationship between WFT, flowability, and
complex rather than the original form. particle wet packing density is indispensable when exploring particles’
Water film thickness (WFT) is a crucial factor used to evaluate wet wet packing.
packing density and defined as the ratio of excess water ratio to the
(8)
′
specific surface area of the total solid, as shown in Eqs. (8-10) [98]. uw = uw − umin
Some studies showed that the increase of particle content, decrease in
7
AS = Aα Rα + Aβ Rβ + Aγ Rγ + ⋯⋯ (9) where, Vpaste is the total volume of cement paste (m3); Vpore is the volume
′
of paste filling voids of aggregate particles (m3); tpaste is the thickness of
WFT =
uw
(10) extra paste (m); MG and MS are amounts of coarse and fine aggregates,
AS respectively (kg); AS,C and AG,C are specific surface area values of coarse
and fine aggregates after correction, respectively (m2/kg); mw/mc is the
where, uw is the ratio of excess water to the volume of solids in paste
′
w/b ratio (unit-less).

(unit-less); As is the specific surface area of all solid components (m2/g); The content range of cementitious materials and the optimal water
Aα , Aβ , and Aγ are the specific surface area values of solid α, β, and γ, consumption in UHPC are determined by this method using the theory of
respectively(m2/g); Rα , Rβ , and Rγ are the volume ratios of solid α, β, and extra paste thickness and the maximum wet packing density (as shown
γ, respectively (unit-less). in Fig. 9). This method considers the role of liquid phase in concrete
Assuming that all the particles are spherical, As can be calculated mixture design and theoretically design mixture proportion.
according to Eqs. (11-12) [144]. Therefore, the role of solid phase and liquid relative packing models
j
∑ is needed to be considered in particle packing. However, no unified
Vi
APSD = 600 (11) standard for the wet packing density of cementitious materials exists
Di
i=1 [146]. Kwan et al. [117,127,147,148] extended wet packing density of
aggregate to cementitious materials and concrete mixtures, and pointed
j
∑
AS = APSD ⋅RS (12) out the wet packing density of fine aggregates or concrete was higher
i=1 than that in the dry state, as shown in Fig. 10. The involved testing
method for wet packing density of cementitious materials can be clas
where, Di is the average particle size between adjacent particle sizes sified into the following four steps: 1) mix particles with water; 2) fill
(μm); Vt is the volume fraction of two adjacent sizes particle (m3); Rs is mixed sample into a designated container; 3) determine whether
the ratio of each solid volume to the total solid volume (unit-less). compaction is required according to experimental requirements; 4) test
The existence of water film onto particles increases flowability. Li wet packing density, as shown in Fig. 11, where SP is the abbreviation of
et al. [145] and Shi et al. [99] calculated wet packing density and the superplasticizer. Some researchers regarded water consumption as an
amount of cementing material by extra paste thickness theory, respec indirect calculation method of wet packing density. However, this
tively, as shown in Eqs. (13-14). method overestimates the packing density for two main reasons [117,
( ) 149].
Vpaste = tpaste g MS AS,C + MG AG,C + Vpore (13)
On the one hand, since the state of water in particles is unstable and
Vpaste ρ indifferent saturated states, causing difficulty in judging whether the
M= (14) liquid is just enough to fill up the voids. On the other hand, the air
(1 + mW /mC )
content in paste is often neglected. Compared with the method
Fig. 9. Mixture proportion based on compact wet packing of particles.
8
Fig. 10. Voids ratio and packing density of concrete mixture with various cement contents under dry and wet conditions [127].
Fig. 11. Testing procedures of the wet packing density of cement-based materials (adapted from Ref. [147]).
mentioned above, Kwan’s method of calculating wet packing density The D-optimal design is a combination method that combines the
using solid concentration shows the following advantages: 1) direct advantages of both the wet close packing model and statistical method.
measurement of solid concentration with negligible experimental error; It transforms concrete mixture design into mathematical and statistical
2) consideration of residual air content in cement paste. models and uses a contour plot to determine the proportion of mixture
9
with maximum wet packing density. Yu et al. [83] used the D-optimal
method and AA model to design a UHPC with a compact structure and
compressive strength close to 135 MPa. This method is more reliable
than previous ones due to considering the effects of liquid phases and
interactions between solid and liquid phases in the mixture.
Theoretically, UHPC with relatively low binder content (about 650
kg/m3) can be produced using the close packing model, and its
compressive strength reaches 140–165 MPa at 28 days [83,112,113].
However, the wet close packing of particles is unstable in the presence of
a large number of fine granular particles, such as SF. The interaction of
SF particles causes the loose structure, and the effect increase as the
particle size decreases. Meanwhile, the loose effect is more evident in a
mixture with cement particles only [150,151]. Therefore, the commonly
used particle packing model shows a significant error in predicting the
packing state of mixture containing nano-particles. The features and
shortcomings of dry and wet close packing models are summarized in
Table 3. Fig. 12. Effects of ingredients on the rheology of concrete [54].
cementitious systems were selected with commonly used content ranges.

2.2. Mixture design method based on rheological properties of UHPC The developed UHPC mixtures were self-consolidating and stable with
28-d compressive strength of 120–125 MPa under standard curing and
2.2.1. Rheology-based mixture design method 178 MPa under heat curing.
The rheology-based mixture design method uses the relationship
between constituents and rheological properties of paste to determine 2.2.2. Effects of particle characteristics on rheological properties of UHPC
appropriate mixture composition and proportion of UHPC with target Before the acceleration period of cement hydration, the effect of
rheology and hardened performance. The rheological properties of particles on the rheological properties of UHPC originates from the
UHPC refer to the deformation and flow properties under the action of surface morphology, volume fraction, gradation, and size of particles
external force. Yield stress and plastic viscosity are two main rheological [54]. Particles with an angular shape and rough surface induce an in
parameters of UHPC. Yield stress is the maximum stress that hinders the crease in contact points between particles, significantly increasing the
plastic deformation of UHPC, and viscosity reflects the flow character yield stress and viscosity due to increased frictional resistance between
istics of UHPC and determines the dispersion and orientation of phases particles [157]. It was found that the yield stress and viscosity increased
in fresh UHPC matrix. UHPC with proper rheological properties can lead with the increase of volume fraction and maximum particle size of
to uniform fiber distribution and satisfactory mechanical strength and coarse particles. The former increases the friction and collision between
toughness [16,152]. Factors affecting rheological properties of UHPC particles [158], while the latter reduces the water consumption of paste
include particle characteristics [153], fibers content [154], the wet due to its lower specific surface area. In contrast, the incorporation of
packing density of skeleton, admixture, water content [155,156], and fine particles reduced the plastic viscosity of UHPC [157]. This can be
external environment humidity and temperature [153]. The influence of explained by the introduced higher solid specific surface area and water
each component on the rheological properties of cement-based materials demand at increasing fine particle content, leading to thicker water film
is shown in Fig. 12. between the particles and reduced contact and friction. In addition, the
Fig. 13 demonstrates the process of mixture design method proposed free water content increased due to the filling effect associated with fine
by Meng et al. [102], which aimed at optimizing rheological properties particles, which enhances the flowability of UHPC [159,160]. However,
and compressive strength of UHPC. After rheological properties the increase in the total surface area of the solid introduced by fine
adjustment and radar chart analysis, binary, ternary, and quaternary
Table 3
Summarization of close packing model-based mixture design method of UHPC.
Model Features Shortcomings Refs.
Based on dry Discrete Furnas model Belongs to binary model; Puts forward the basic Only suitable for mixtures with significant [75]
packing models theory of particle packing. particle size differences.
density Powers model Considered the wall effect and loosening effect; Only considered packing of binary mixtures. [90]
proposed algorithm of minimum porosity.
Aim and Goff model Modified the packing density of binary mixture; Not considered more components. [91]
Mitigated negative effect caused by wall effect.
Modified Toufar model Proposed the Eigen packing; Established contour of Unreasonable hypothesis of model. [92,
packing density of ternary particles. 93]
Linear packing density Belongs to the model of multi-particle packing Suitable for spherical particles with [77]
model (LPDM) density; Derived continuous distribution of grain deviation from the actual situation.
size.
Compressible packing Introduced virtual and actual packing density; The particle size is specific; The packing [80]
model (CPM) Corrected wall effect and porosity effect. density is governed by compaction process.
Continuous Fuller Thomson model Proposed maximum density grading curve. Not suitable for mixture design of concrete. [96]
models Andreassen model The continuous method is used to study PSD; Still based on empirical model;
Proposed an ideal packing equation. Unreasonable assumption for smallest
particle.
Considering wet packing density Kwan model Measured the wet packing density, i.e., solid Rely on experimental data without [117,
concentration of cementitious materials and fine theoretical basis. 148]
aggregate.
Extra paste thickness Considered the influence of liquid phase of concrete. Affected by paste flowability and particle [99]
model packing.
10
Fig. 13. Mixture design of UHPC based on rheological properties and compressive strength of UHPC [102].
particles may result in higher viscosity and low flowability if the paste substitution.
amount to coat the surface of aggregate is not enough. The PSD and size Considering the difference in water consumption for hydration of
of the particles are critical to the rheological properties of UHPC. Par particles with different sizes, Mehdipour et al. [100] used a particle
ticles with better PSD can achieve a denser system to reduce the viscosity packing model to evaluate the effects of superplasticizer and mineral
of paste and improves its flowability [159]. The findings were also admixtures on the performance of cement paste. The results showed that
confirmed by Bentz et al. [160]. water consumption per unit area of cement pastes decreased at a higher
The large amount of SCMs in UHPC can optimize the pore structure replacement ratio of fine particles while flowability increased. There
of UHPC, and make the ITZ denser through its physical filling of sub fore, to ensure the fresh mixing performance of UHPC, particle size and
micron and nanometer particles, as well as the pozzolanic reaction. The content should be reasonably selected and its effects on rheological
reactivity and particle morphology of SCMs are the main factors properties in UHPC mixture design.
affecting the mixture design of UHPC. SCM with a smooth surface
significantly improves the flowability of UHPC by reducing the friction 2.2.3. Effect of fibers on rheological properties of UHPC
between particles, such as FA and SF. However, fresh UHPC containing Fiber is an essential part of UHPC to obtain high tensile strength and
SF has higher consistency than that of UHPC incorporating FA. This is toughness [166]. The strengthening and toughening effects of fibers
caused by the higher specific surface area and the adsorption of a large mainly depend on fibers dispersion and orientation throughout the
amount of water by hydrogen bond on the surface of SF [149,161,162]. system and bonding properties between fibers and matrix [167,168].
Considering the flowability and compressive strength of UHPC, the The fibers commonly used in UHPC can be divided into rigid and flexible
content of SF is usually less than 15% [163]. Limestone, which is a fibers according to Young modulus, the shape of the fibers, and the
substitute for SF, can produce UHPC with significantly improved 28-d consistency of mixture [169]. The addition of fibers can greatly influ
compressive strength and slump of 240 mm by adding 20% limestone ence particle packing and, finally the rheological properties of paste.
[35,164]. When the fibers are altered from disorderly to densely, and the fibers
Vance et al. [165] examined the effects of particle number density, aspect ratio small enough, its packing density is close to that of the
specific surface area, and WFT on the rheological properties of ternary particles, as shown in Fig. 14 [169]. Rigid fiber can partially obstruct the
binder systems containing FA, limestone, or metakaolin. Results showed packing of particles by pushing them away, while flexible fiber can in
that the yield stresses and plastic viscosity of paste both decreased when crease the packing density by filling the space between particles [153].
the median particle size of limestone increased from 0.7 to 15 μm. When Incorporating fibers can render UHPC more viscous, which decreases
the cement replacement ratios were 5% and 20%, the yield stress values fiber dispersion and orientation and eventually reduces flowability and
were reduced by 20% and 50%, respectively. The incorporation of larger impairs hardened performance of UHPC [154]. The underlying action
irregular particles increases the spacing between particles, thereby mechanisms on reducing the flowability involving the following three
reducing the shear resistance of paste and eventually the yield stress and aspects: 1) reduced free water content through absorbing on fibers
viscosity. However, the incorporation of FA reduces friction and pro surface; 2) mutual winding associated with uniform fibers dispersion
motes particle packing due to its spherical shape. Unlike limestone, the and distribution; 3) increased demanded cement paste needed to cover
rheological parameters of paste tend to be stable at a higher level of FA fibers surface [170]. The uniform dispersion and distribution of fibers in
11
Fig. 14. Different packing states of rigid fiber (adapted from Ref. [169]).
paste can be achieved by controlling the flowability of paste [171]. On 1% and 1.6% superplasticizer in UHPC. However, steel fibers are uni
the other hand, the interlocking of steel fibers and coarse aggregate in formly distributed in UHPC with 1% superplasticizer, which exhibits
UHPC significantly reduces the workability of UHPC. The coarse high yield stress, as shown in Fig. 16 [175]. Studies indicated that the
aggregate with large particle size cannot be completely wrapped by steel slump value of UHPC dropped sharply from 4.29 to 1.10 when steel fi
fibers, and the bonding effect between aggregate and fibers would be bers content increased from 0.5% to 2.5% [176]. Moreover, interlaced
weakened. The appropriate length of steel fibers and reasonable fibers introduced more air void, which is unfavorable to the strength of
aggregate gradation should be selected [52]. Studies have shown that UHPC [177]. Teng et al. [178] pointed out that when fiber content
the maximum particle size of coarse aggregate is recommended to be increased from 1% to 3%, the dispersion, and orientation of the fibers
smaller than 25 mm, and the steel fiber length should be between 2 - 5 were linearly related to the plastic viscosity of UHPC.
times of the maximum size of coarse aggregate [53]. Wang et al. [16] selected four w/b ratios and superplasticizers to
The aspect ratio and volume fraction of fiber in UHPC significantly adjust the rheological properties of cement mortar, and controlled fibers
affect the performance of UHPC. The probability of heterogeneous dis distribution in UHPC. Poor dispersion of fibers was observed in the case
tribution and flocculation of fibers in UHPC increases with the increase of low w/b ratio and high fiber content. Besides, the effect of super
of fiber aspect ratio [70]. To avoid excessive fiber content, less fiber is plasticizer on the dispersion of steel fibers was more significant under a
added to UHPC with the use of increased fiber aspect ratio [172]. higher w/b ratio. Fig. 17 indicates that the development trend of the
Research indicated that on the premise of uniform fiber distribution, the distribution coefficients of UHPC specimens with different fiber contents
flocculation of fibers makes flow under UHPC’s own weight impossible are similar, and decrease rheological parameters can aggravate the
when the critical fiber content was surpassed [58]. Generally, the aspect non-uniform dispersion of fibers in a certain range. The better stability
ratio of fiber used in UHPC varied between 50 and 100 [173], and the occurred when the fiber content is 3%, which may be due to the me
volume fraction of fiber is 0.5%–2.5% to make UHPC with good work chanical interlocking of more fibers. The optimal distribution of steel
ability. Ghafari et al. [84] found that 2.5% steel fibers in UHPC are the fibers occurs when the rheological properties of paste are moderate,
maximum content to meet flowability requirements. In the case of the which is also confirmed by other studies [115,178]. The plastic viscosity
same fiber volume, short fiber promotes better compressive strength of of UHPC mortar under the same dispersion and orientation coefficient is
UHPC when compared to long fiber. This may be due to the fact that higher when fiber content increased from 1% to 3%, and the change in
longer fibers are prone to agglomeration, causing UHPC difficulties in fiber distribution is limited after exceeding the optimal value, as shown
vibration and flow [174]. in Fig. 18. It also can be shown that for UHPC with higher fiber content,
Fiber sinks naturally in concrete by gravity and accumulates in the its optimal distribution of the fibers requires a higher viscosity of paste.
middle and lower sections of specimens, as shown in Fig. 15 [16]. Wang The mixture design of UHPC considering fibers dispersion takes the
et al. [175] observed noticeable sedimentation of 2% fibers after mixing rheological property of cement paste as a standard to control the
Fig. 15. Fiber distribution image collected by OLYCIA m3-DSX software [16].
12
Fig. 16. Distribution of steel fibers in UHPC cylinders [175].
Fig. 17. Changes in viscosity and yield stress values of cement mortar with various distribution coefficients at 1% and 3% fiber contents [16].
Fig. 18. Changes in fiber dispersion and orientation coefficients of UHPC with plastic viscosity at various fiber contents [178].
13
distribution of steel fibers. After the continuous adjustment of w/b ratio verify the accuracy of response surface; 3) adjustment of the proportion
and superplasticizer content, satisfactory rheological properties of of each component to get expected response values [84]. The design
UHPC can be obtained to render uniform dispersion and orientation of steps of RSM are shown in Fig. 19. The ultimate purpose of RSM is to
steel fibers. The evaluation of UHPC mixture ratio using rheological determine optimal operating conditions and areas meeting operating
properties alone as standard is often one-sided, and it requires many specifications. In cement-based materials, the use of RSM can satisfy
repeated experiments. In addition, the rheological test data is very nonlinear factors and consider the relationship between variables that
sensitive to variables such as the external environment. affect each other, and ultimately provide an ideal approximation for
optimizing mixture proportion [103]. As shown in Fig. 20, the flow
ability and compressive strength model was proposed through the cen
2.3. Statistical mixture design method tral composite design in RSM. An environment-friendly UHPC was
designed with cement and SF contents of 640 kg/m3 and 56.3 kg/m3,
The statistical analysis model is used to generate decision variables respectively [105]. Results showed that SF had the greatest effect on
and objective data in the mixture proportion of UHPC [179]. It is an flowability reduction, and 2.5% steel fibers were the maximum content
effective experimental design method by varying mixture proportions to to render UHPC with optimal slump flow value [84].
explore how variables affect the performance of mixture [180]. It in Mosaberpanah et al. [104] selected RSM to optimize the perfor
volves designing parameters, such as W/C, proportions of cement, mance of UHPC under normal curing conditions and then analyzed the
mineral and chemical admixtures, and aggregate. An appropriate sta accuracy of the model. The variables are W/C, proportions of SF and
tistical model can be obtained within the reasonable parameter ranges to cement, steel fibers content, and superplasticizer. The response surface
reflect the mathematical relationship between mixture proportioning of
UHPC and fresh and hardened performance.
The purpose of testing or experiment can be achieved by inputting
responses generated by several variables. If the response is well modeled
by a linear function of the independent variables, then the approxi
mating function is the first-order model, as shown in Eq. (15). To satisfy
the experiment data is relatively close to the optimal value, the second-
order model is usually required to predict the response, as shown in Eq.
(16).
y = β0 + β1 x1 + β2 x2 + ⋅⋅⋅ + βk xk + ε (15)
∑
k ∑
k ∑∑
y = β0 + βi xi + βii x2i + βij xi xj + ε (16)
i=1 i=1 i<j
where, Y is the response variable (the unit depends on the response); βi,
βii, and βij are the influence coefficients of factors (unit-less); xi and xj are
variables; ε is the error of response y (unit-less).
An analytical model is established using the relationship between
response and variables and interaction between variables [181]. The
response surface analysis mainly includes three steps: 1) experimental
Fig. 20. Geometric illustration of the central composite design [105].
design and data collection; 2) establishment of a numerical model to
Fig. 19. Design steps of RSM (adapted from Ref. [103]).
14
model employed is a linear or quadratic model with α = 1 plane as center the properties of UHPC. Three variables combined with the second-order
combination design and response. The mixture design procedure is polynomial model of Eq. (16) and an iterative model are used to check
shown in Fig. 21. Although this method established a good relationship the accuracy of predicted results. The accuracy of model was tested by
between UHPC performance and mixture design, the results are greatly analysis of variance after eliminating secondary items. In this study, RDL
affected by the range of parameters and the chosen variables. (%) was used to verify whether the predicted results from the model
For environment-friendly UHPC, RSM can optimize the proportion of were consistent with experimental data. When multiple components
raw materials and adjust the production cost of UHPC. Ferdosian et al. affect each other, the simplex centroid design method can be used for
[105] established a second-order polynomial model with a central mixture design. Shi et al. [108] assumed that the composition of
composite design of RSM by selecting three types of raw materials in cementitious materials includes cement, FA, and GGBS. The model was
UHPC as variables. The environmentally friendly UHPC with cement designed using Eq. (17). Seven groups of different contents of cementi
consumption of less than 650 kg/m3 was successfully designed. The tious materials were selected to draw the contour plots of the ternary
mixture proportion of UHPC designed by this method meets the fresh system. It is worth noting that when the coefficient of a specific
and hardened performance of UHPC and reduces cement content. component is positive, the component will promote the response.
Lightweight sand (LWS) is a common porous material to reduce autog
Y = β1 x1 + β2 x2 + β3 x3 + β12 x1 x2 + β13 x1 x3 + β123 x1 x2 x3 (17)
enous shrinkage of UHPC. Meng et al. [106] used a statistical method to
analyze the effect of lightweight sand replacement ratio to natural sand, where, Y is the responses (the unit depends on the response); βi is the
W/C, and amount of substituted cementitious materials. The equation coefficient (unit-less); x1, x2, and x3 are the three components, respec
and probability value (p) of each material were analyzed by LSM, and tively (unit-less).
the coupling effect between parameters was expressed using significance Within the range of variables, the optimum mixture proportion
degree. The procedures of statistical analysis method considering LWS scheme is based on a large number of well-performing data, which
are illustrated in Fig. 22. Contrary to the assumption that most statistical greatly reduces the number of experiments and labor. However, the
analysis methods are based on independent variables, the components in response and accuracy of this method governed by its variables limit its
UHPC usually affect each other. Therefore, the accuracy of the predic application in UHPC.
tion results of statistical analysis methods is limited.
Statistical analysis can also be used to select binders, such as cement,
SF, and quartz. Ghafari et al. [85] demonstrated the relationship be 2.4. Mixture design method based on ANN model
tween the different contents of each component and the properties of
UHPC using RSM. This method can directly compare and select the Artificial neural networks (ANNs) are mathematical techniques
appropriate type and content of binder, as shown in Fig. 23. It was originating from the biological nervous system [182]. The input values
demonstrated that cement has a dominant role in the strength of UHPC, can be connected to output values through one or more layers of in
and the increase in quartz content enhances the flowability of paste. termediate. The output signal is obtained from the input signal of related
Therefore, a practical reference for the selection of type and content neurons, weight, and activation function. In the UHPC mixture design
binder can be provided by contour plots. The model established in this using the ANN model, the input layer nodes are UHPC mixture decision
method reflects the relationship between variables and compressive variables, and the output layer is the objective [183]. By adjusting the
strength, and predicts the performance of UHPC. Soliman et al. [107] input weights to narrow the gap between the experiment and input
used a statistical model to optimize mixture design based on particle values, the model’s accuracy can be guaranteed.
packing model considering the compatibility of cement and super The backpropagation (BP) algorithm that was widely used in neural
plasticizer and developed environmentally friendly UHPC. network modeling was first proposed by Rumelhart et al. [109]. It
Ghafaric et al. [84] combined regression model and RSM to predict compares the errors between the output and target values. Ghafari et al.
[88] proposed an ANN analysis model for UHPC based on the back
propagation neural network (BPNN) and Levenberg-Marquardt algo
rithm (LMBP). The typical ANN structure consists of input, weight,
function, activation function, and output, as shown in Fig. 24. The
calculation idea of BPNN is divided into the following two points: 1)
error of neural network depends on input and measured value, and
overall error is reduced through spreading error back; 2) weight and
error are readjusted until the output value is closest to the experimental
value. BPNN is simple, but data convergence is slow due to an unknown
mapping relationship. Moreover, the convergence speed of neural net
works changes with adjustment of weight [184]. The weights and their
sums in the BPNN model can be calculated by Eq. (18).
∑n
netj = i=1
wij xi + b (18)
where, Win represents the weight of each parameter (unit-less); netj is the
weighted sums of input components (unit-less).
The predicted compressive strength value of UHPC by the ANN
model reached 240 MPa, but its experimental value was 200 MPa [88].
The maximum wet packing density and slump value of UHPC are 0.83
and 290 mm, respectively, which are 20% higher than UHPC with a poor
proportion [82]. To effectively improve convergence speed and data
fitting accuracy of neural networks, an optimized BPNN was selected by
Tan et al. [110] to predict tensile strength and crack width of UHPC. In
this method, crossover operation is reduced, and the phenomenon of
local overfitting is eliminated through the employment of different
Fig. 21. Statistical mixture design procedure of UHPC (based on Ref. [104]). concentrations and discrete degrees of data.
15
Fig. 22. Statistical analysis method considering LWS (based on Ref. [106]).
Fig. 23. Contour plots of compressive strength and slump flow of UHPC with different types and contents of cementitious materials [85].
Fig. 24. Typical structure diagram of ANN [88].
16
The accuracy of the above model is limited owing to the neglect of with significantly reduced porosity and enhanced strength of
uniformity of steel fibers in UHPC. Qu et al. [111] established two ANN UHPC. Nevertheless, most of the particle packing models only
models to predict UHPC strength. Eighty percent of data was used to consider the packing density of the solid skeleton without
train the model, and the rest was used to examine the model’s accuracy. reflecting the interaction of liquids, such as water and super
Although the results showed that calculated values were close to the real plasticizer, with solid particles. Thus, the measured packing
values, as shown in Fig. 25. The research on the parameters of steel fi density is lower than that under the actual wet state, leading to
bers, including fibers length, diameter, and volume fraction in this inaccurately predicted strength of UHPC.
experiment was not enough since fibers type and shape would signifi (2) Rheology-based mixture design method can produce UHPC with
cantly affect the performance of UHPC. satisfactory flowability, which can ensure relatively uniform
Ghafari et al. [87] selected seven different input values to establish dispersion and orientation of strengthening materials, such as
the ANN model, as shown in Fig. 26(a). The results showed that this ANN fiber and aggregate, to render better performance of UHPC. The
model can efficiently predict concrete performance, as shown in Fig. 26 rheology of UHPC can be adjusted by changing the mixture
(b). The same conclusion was reported by Fang et al. [82]. The ANN proportioning, including superplasticizer dosage, PSD, fibers type
model takes the consistency of fresh UHPC into account and can be used and content, and w/b ratio. The determination of water and
to predict the strength of UHPC. However, artificial nerve requires many superplasticizer demands should consider the particle packing of
experiments to train each neuron. Moreover, over-fitting may occur, UHPC and evaluated the flowability of paste. Repeated experi
which causes neural networks unable to derive new data. ANN model is ments to determine the optimal rheological properties of UHPC
still based on many data and lacks theoretical derivation. mixed with fibers, and the rheological test data is sensitive to
variables such as the external environment.
3. Comparison of different mixture design methods (3) The statistical mixture design method quantitatively analyzes the
relationship between the mixture proportions and performance of
According to the above results, the advantages and disadvantages of UHPC and simplifies the scheme required to optimize mixture
the existing design methods of UHPC are listed in Table 4. The design design. However, many experiments are needed to balance a wide
method based on particle packing is a widely used method at present. range of variables and thus verify the method’s reliability.
However, most of the models do not consider the effect of liquid phase (4) The ANN can reduce the trial tests, enhance the cost-effectiveness
on solid particles, which results in a difference between the prediction of UHPC, as well as provide accurate predictions by continuously
results and the actual situation. Rheology-based mixture design method varying the input values and its weight. The ANN method is
for UHPC needs to consider the effects of various components in UHPC, simpler and more direct than the traditional statistical method
including aggregate, fibers, and superplasticizer on the yield stress and through linear and non-linear programming. However, over
viscosity of UHPC. A lot of trial tests are required to reach the optimal fitting may occur locally, and data cannot be extended due to the
rheological properties through adjusting the mixture proportions. The lack of data to train each neuron.
statistical analysis method can reduce the number of trials, but the ac
curacy is affected by independent variables and the selected model. The 4.2. Future perspectives
ANN model is derived from the biological nervous system, but its
application is limited due to over-fitting. Mixture design is an essential step in concrete production. To pro
mote the manufacture and application of UHPC, the following aspects
4. Conclusions and future perspectives should be considered in future research:
4.1. Conclusions (1) Alternative materials should be used to replace cement, and sci
entific and practical mixture design methods should be selected
Based on the above extensive review and discussion of the UHPC to balance various aspects of performance of UHPC, including
mixture design methods, the following conclusions can be drawn: robustness of various raw materials, flowability, strength, cost,
and sustainability.
(1) The close packing method derived from particle continuously (2) Considering the conflicting requirements of those performances
packing is by far the most widely used mixture design method for of UHPC, integration of two or more mixture design methods to
UHPC. It enables UHPC to achieve a densely compacted structure exert the advantages of each method by meeting multiple
Fig. 25. Experimental and predicted compressive and flexural strengths of UHPC using ANN model [111].
17
Fig. 26. Structure of ANN model and testing and training values through RMSE [87].
References
Table 4
Advantages and disadvantages of various mixture design methods. [1] C. Shi, Z. Wu, J. Xiao, D. Wang, Z. Huang, Z. Fang, A review on ultra high
Methods Advantages Disadvantages performance concrete: Part I. Raw materials and mixture design, Construct. Build.
Mater. 101 (2015) 741–751.
Close Based on Increased packing density Lower dry packing density [2] D. Yoo, S.T. Kang, Y.S.J.C.S. Yoon, Enhancing the flexural performance of ultra-
packing dry of particles; Reduced without considering the high-performance concrete using long steel fibers, Compos. Struct. 147 (2016)
model packing porosity, and improved interaction between solid 220–230.
density strength of UHPC and liquid phases of water [3] China Architecture & Building Press, Reactive Powder Concrete (GB/T 31387-
and superplasticizer; 2015), China Architecture & Building Press, Beijing, China, 2015.
Lower strength of UHPC [4] ASTM C1856/C1586-17, Standard Practice for Fabricating and Testing
Based on More precise and actual Decreased wet packing Specimens of Ultra-high-performance Concrete, ASTM International, West
Conshohocken, PA, 2017.
wet packing density value density due to
[5] JSCE (Japan Society of Civil Engineers), Recommendations for Design and
packing considering the effects of agglomeration of fine
Construction of Ultra High Strength Fiber Reinforced Concrete Structures (Draft),
density water and particles JSCE, Tokyo, Japan, 2004.
superplasticizer on solid [6] French Standard Institute, Ultra-high Performance Fiber-Reinforced Concrete
particles (UHPFRC)-specification, Production and Conformity, NFP, France, 2017,
Mixture design method Adjusted optimal Lacked theoretical model pp. 18–470.
based on rheological rheology to secure for predicting the fresh [7] ACF (Asian Concrete Federation), Guidelines for UHPC Materials, ACF - 120001 -
properties of paste relatively uniform performance of UHPC 2020, 2020. Hunan, China.
dispersion of the solid mixed with fiber [8] KCI (Korea Concrete Institute), Design Recommendations for Ultra-high
skeleton, especially fiber, Performance Concrete KCI-M-12-003, K-UHPC, Korea Concrete Institute, Seoul,
and enhanced hardened 2012.
performance of UHPC [9] J. Dils, V. Boel, G. De Schutter, Influence of cement type and mixing pressure on
Statistical mixture design Established relationship Reduced model accuracy air content, rheology and mechanical properties of UHPC, Construct. Build.
Mater. 41 (2013) 455–463.
method between mixture due to interaction
[10] S. Abbas, A.M. Soliman, M.L. Nehdi, Exploring mechanical and durability
proportioning and between components
properties of ultra-high performance concrete incorporating various steel fiber
performance of UHPC;
lengths and dosages, Construct. Build. Mater. 75 (2015) 429–441.
Reduced number of [11] Y.L. Voo, S.J. Foster, C.C. Voo, Ultrahigh-performance concrete segmental bridge
experiments technology: toward sustainable bridge construction, J. Bridge Eng. 20 (8) (2015)
Mixture design method Predicted consistency Required a large amount B5014001.
based on ANN model and strength of fresh of data to train the model; [12] J. Li, C. Wu, H.J.E.S. Hao, Investigation of ultra-high performance concrete slab
UHPC; Reduced number Occurred data overfit and normal strength concrete slab under contact explosion, Eng. Struct. 102
of experiments (2015) 395–408.
[13] M. Thomas, B. Green, E. O’Neal, V. Perry, S. Hayman, A. Hossack, Marine
performance of UHPC at treat island, in: Proceedings of the 3rd International
Symposium on UHPC and Nanotechnology for High Performance Construction
requirements of UHPC, thus achieving better UHPC mixture
Materials, 2012, pp. 365–370. Kassel, Germany.
proportion. [14] L. Yang, C. Shi, Z. Wu, Mitigation techniques for autogenous shrinkage of ultra-
high-performance concrete – a review, Compos. B Eng. 178 (2019) 107456.
Declaration of competing interest [15] E. Ghafari, S.A. Ghahari, H. Costa, E. Júlio, A. Portugal, L. Durães, Effect of
supplementary cementitious materials on autogenous shrinkage of ultra-high
performance concrete, Construct. Build. Mater. 127 (2016) 43–48.
The authors declare that they have no known competing financial [16] R. Wang, X. Gao, H. Huang, G. Han, Influence of rheological properties of cement
interests or personal relationships that could have appeared to influence mortar on steel fiber distribution in UHPC, Construct. Build. Mater. 144 (2017)
65–73.
the work reported in this paper. [17] H. Huang, X. Gao, L. Li, H. Wang, Improvement effect of steel fiber orientation
control on mechanical performance of UHPC, Construct. Build. Mater. 188 (2018)
Acknowledgment 709–721.
[18] H. Wong, A. Kwan, Packing density: a key concept for mix design of high
performance concrete, in: Proceedings of the Materials Science and Technology in
The authors gratefully acknowledge the financial support from the Engineering Conference, 2005.
Ministry of Science and Technology of the People’s Republic of China [19] P. Kumbhar, P. Murnael, Assessment of suitability of existing mix design methods
of normal concrete for designing high performance concrete mixes, Int. J. Civ.
under Project No. 2018YFC0705400, the National Science Foundation
Struct. Eng. 3 (1) (2012) 158–167.
of China under Project No. 52008164, and the Fundamental Research [20] E. Sadrossadat, H. Basarir, A. Karrech, M. Elchalakani, Multi-objective mixture
Funds for the Central Universities under project No. 531118010484. design and optimisation of steel fiber reinforced UHPC using machine learning
algorithms and metaheuristics, Eng. Comput. (2021).
[21] S.P. Shah, S.H. Ahmad, High Performance Concrete. Properties and Applications,
1994.
18
[22] S.H. Kosmatka, M.L. Wilson, Design and Control of Concrete Mixtures, Portland [51] Y. Industrial and Engineering ChemistryPeng, H. Wu, Q. Fang, J. Liu, Z. Gong,
Cement Association, 2002. Impact resistance of basalt aggregated UHP-SFRC/fabric composite panel against
[23] P.-K. Chang, An approach to optimizing mix design for properties of high- small caliber arm, Int. J. Impact Eng. 88 (2016) 201–213.
performance concrete, Cement Concr. Res. 34 (4) (2004) 623–629. [52] P.P. Li, Q.L. Yu, H.J.H. Brouwers, Effect of coarse basalt aggregates on the
[24] N. Su, B. Miao, A new method for the mix design of medium strength flowing properties of Ultra-high Performance Concrete (UHPC), Construct. Build. Mater.
concrete with low cement content, Cement Concr. Compos. 25 (2) (2003) 170 (2018) 649–659.
215–222. [53] P.P. Li, Y.Y.Y. Cao, M.J.C. Sluijsmans, H.J.H. Brouwers, Q. Yu, Synergistic effect
[25] D. Fan, R. Yu, S. Fu, L. Yue, C. Wu, Z. Shui, K. Liu, Q. Song, M. Sun, C. Jiang, of steel fibres and coarse aggregates on impact properties of ultra-high
Precise design and characteristics prediction of Ultra-High Performance Concrete performance fibre reinforced concrete, Cement Concr. Compos. 115 (2021)
(UHPC) based on artificial intelligence techniques, Cement Concr. Compos. 122 103866.
(2021) 104171. [54] D. Jiao, C. Shi, Q. Yuan, X. An, Y. Liu, H. Li, Effect of constituents on rheological
[26] E. Sharifi, S.J. Sadjadi, M.R.M. Aliha, A. Moniri, Optimization of high-strength properties of fresh concrete-A review, Cement Concr. Compos. 83 (2017)
self-consolidating concrete mix design using an improved Taguchi optimization 146–159.
method, Construct. Build. Mater. 236 (2020) 117547. [55] A. Arora, A. Almujaddidi, F. Kianmofrad, B. Mobasher, N. Neithalath, Material
[27] M. Alkaysi, S. El-Tawil, Z. Liu, W. Hansen, Effects of silica powder and cement design of economical ultra-high performance concrete (UHPC) and evaluation of
type on durability of ultra-high performance concrete (UHPC), Cement Concr. their properties, Cement Concr. Compos. 104 (2019) 103346.
Compos. 66 (2016) 47–56. [56] Y. Chen, J. Feng, H. Li, Z. Meng, Effect of coarse aggregate volume fraction on
[28] W. Li, Z. Huang, G. Hu, W. Hui Duan, S.P. Shah, Early-age shrinkage development mode II fracture toughness of concrete, Eng. Fract. Mech. 242 (2021) 107472.
of ultra-high-performance concrete under heat curing treatment, Construct. [57] P. Rossi, N. Harrouche, Mix design and mechanical behaviour of some steel-fibre-
Build. Mater. 131 (2017) 767–774. reinforced concretes used in reinforced concrete structures, Mater. Struct. 23 (4)
[29] L. Yang, C. Shi, J. Liu, Z. Wu, Factors affecting the effectiveness of internal curing: (1990) 256.
a review, Construct. Build. Mater. (2020) 121017. [58] S. Grünewald, J.C. Walraven, Parameter-study on the influence of steel fibers and
[30] P. Richard, M. Cheyrezy, Composition of reactive powder concretes, Cement coarse aggregate content on the fresh properties of self-compacting concrete,
Concr. Res. 25 (7) (1995) 1501–1511. Cement Concr. Res. 31 (12) (2001) 1793–1798.
[31] N. Randl, T. Steiner, S. Ofner, E. Baumgartner, T. Mészöly, Development of UHPC [59] A. Sadrmomtazi, S. Tajasosi, B. Tahmouresi, Effect of materials proportion on
mixtures from an ecological point of view, Construct. Build. Mater. 67 (2014) rheology and mechanical strength and microstructure of ultra-high performance
373–378. concrete (UHPC), Construct. Build. Mater. 187 (2018) 1103–1112.
[32] A. Korpa, T. Kowald, R. Trettin, Phase development in normal and ultra high [60] Z. Wu, C. Shi, K.H. Khayat, Investigation of mechanical properties and shrinkage
performance cementitious systems by quantitative X-ray analysis and of ultra-high performance concrete: influence of steel fiber content and shape,
thermoanalytical methods, Cement Concr. Res. 39 (2) (2009) 69–76. Compos. B Eng. 174 (2019) 107021.
[33] H. Yazıcı, H. Yiğiter, A.Ş. Karabulut, B. Baradan, Utilization of fly ash and ground [61] A. Bentur, S. Mindess, Fibre Reinforced Cementitious Composites, 2007.
granulated blast furnace slag as an alternative silica source in reactive powder [62] D.-Y. Yoo, S. Kim, G.-J. Park, J.-J. Park, S.-W. Kim, Effects of fiber shape, aspect
concrete, Fuel 87 (12) (2008) 2401–2407. ratio, and volume fraction on flexural behavior of ultra-high-performance fiber-
[34] R. Siddique, J. Klaus, Influence of metakaolin on the properties of mortar and reinforced cement composites, Compos. Struct. 174 (2017) 375–388.
concrete: a review, Appl. Clay Sci. 43 (3) (2009) 392–400. [63] Z. Wu, C. Shi, W. He, L. Wu, Effects of steel fiber content and shape on mechanical
[35] C. Wang, C. Yang, F. Liu, C. Wan, X. Pu, Preparation of ultra-high performance properties of ultra high performance concrete, Construct. Build. Mater. 103
concrete with common technology and materials, Cement Concr. Compos. 34 (4) (2016) 8–14.
(2012) 538–544. [64] Z. Wu, C. Shi, H. Wen, D. Wang, Uniaxial compression behavior of ultra-high
[36] Y. Peng, S. Hu, Q. Ding, Preparation of reactive powder concrete using fly ash and performance concrete with hybrid steel fiber, J. Mater. Civ. Eng. 28 (12) (2016),
steel slag powder, J. Wuhan Univ. Technol.-Materials Sci. Ed. 25 (2) (2010) 06016017.
349–354. [65] Z. Wu, C. Shi, W. He, D. Wang, Static and dynamic compressive properties of
[37] N. Van Tuan, G. Ye, K. van Breugel, O. Copuroglu, Hydration and microstructure ultra-high performance concrete (UHPC) with hybrid steel fiber reinforcements,
of ultra high performance concrete incorporating rice husk ash, Cement Concr. Cement Concr. Compos. 79 (2017) 148–157.
Res. 41 (11) (2011) 1104–1111. [66] K. Hannawi, B. Hui, W. Prince-Agbodjan, B. Raghavan, Effect of different types of
[38] Z. Wu, C. Shi, W. He, Comparative study on flexural properties of ultra-high fibers on the microstructure and the mechanical behavior of Ultra-High
performance concrete with supplementary cementitious materials under different Performance Fiber-Reinforced Concretes, Compos. B Eng. 86 (2016) 214–220.
curing regimes, Construct. Build. Mater. 136 (2017) 307–313. [67] A. Noushini, B. Samali, K. Vessalas, Effect of polyvinyl alcohol (PVA) fibre on
[39] C. Gu, W. Sun, L. Guo, Q. Wang, Effect of curing conditions on the durability of dynamic and material properties of fibre reinforced concrete, Construct. Build.
ultra-high performance concrete under flexural load, J. Wuhan Univ. Technol.- Mater. 49 (2013) 374–383.
Materials Sci. Ed. 31 (2) (2016) 278–285. [68] S.H. Ghasemzadeh Mosavinejad, M.A.M. Langaroudi, J. Barandoust,
[40] Z. Mo, X. Gao, A. Su, Mechanical performances and microstructures of metakaolin A. Ghanizadeh, Electrical and microstructural analysis of UHPC containing short
contained UHPC matrix under steam curing conditions, Construct. Build. Mater. PVA fibers, Construct. Build. Mater. 235 (2020) 117448.
268 (2021) 121112. [69] S.-T. Kang, K.-S. Lee, J.-I. Choi, Y. Lee, B. Felekoğlu, B.Y. Lee, Control of tensile
[41] M.A. Mosaberpanah, O. Eren, Effect of quartz powder, quartz sand and water behavior of ultra-high performance concrete through artificial flaws and fiber
curing regimes on mechanical properties of UHPC using response surface hybridization, Int. J. Concr. Struct. Mater. 10 (3) (2016) 33–41.
modelling, Adv. Concr. Constr. 5 (5) (2017) 481–492. [70] Y. Su, J. Li, C. Wu, P. Wu, Z.-X. Li, Effects of steel fibres on dynamic strength of
[42] M. Ding, R. Yu, Y. Feng, S. Wang, F. Zhou, Z. Shui, X. Gao, Y. He, L. Chen, UHPC, Construct. Build. Mater. 114 (2016) 708–718.
Possibility and advantages of producing an ultra-high performance concrete [71] M. Alkaysi, S. El-Tawil, Effects of variations in the mix constituents of ultra high
(UHPC) with ultra-low cement content, Construct. Build. Mater. 273 (2021) performance concrete (UHPC) on cost and performance, Mater. Struct. 49 (10)
122023. (2016) 4185–4200.
[43] J. Fládr, P. Bílý, T. Trtík, Analysis of the influence of supplementary cementitious [72] K. Srinivas, A.L. Naidu, M. Raju Bahubalendruni, A review on chemical and
materials used in UHPC on modulus of elasticity, IOP Conf. Ser. Mater. Sci. Eng. mechanical properties of natural fiber reinforced polymer composites, Int. J.
522 (2019), 012010. Perform. Eng. 13 (2) (2017).
[44] Ç. Yalçınkaya, H. Yazıcı, Effects of ambient temperature and relative humidity on [73] J. Liu, C. Li, J. Liu, G. Cui, Z. Yang, Study on 3D spatial distribution of steel fibers
early-age shrinkage of UHPC with high-volume mineral admixtures, Construct. in fiber reinforced cementitious composites through micro-CT technique,
Build. Mater. 144 (2017) 252–259. Construct. Build. Mater. 48 (2013) 656–661.
[45] Z. Wu, C. Shi, K.H. Khayat, L. Xie, Effect of SCM and nano-particles on static and [74] J. Liu, F. Han, G. Cui, Q. Zhang, J. Lv, L. Zhang, Z. Yang, Combined effect of
dynamic mechanical properties of UHPC, Construct. Build. Mater. 182 (2018) coarse aggregate and fiber on tensile behavior of ultra-high performance
118–125. concrete, Construct. Build. Mater. 121 (2016) 310–318.
[46] K. Zhang, L. Pan, J. Li, C. Lin, Y. Cao, N. Xu, S. Pang, How does adsorption [75] C. Furnas, Grading aggregates-I.-Mathematical relations for beds of broken solids
behavior of polycarboxylate superplasticizer effect rheology and flowability of of maximum density, Ind. Eng. Chem. 23 (9) (1931) 1052–1058.
cement paste with polypropylene fiber? Cement Concr. Compos. 95 (2019) [76] R. Yu, P. Spiesz, H.J.H. Brouwers, Development of an eco-friendly Ultra-High
228–236. Performance Concrete (UHPC) with efficient cement and mineral admixtures
[47] Z. Yingliang, Q. Jingping, M.A. Zhengyu, G. Zhenbang, L. Hui, Effect of superfine uses, Cement Concr. Compos. 55 (2015) 383–394.
blast furnace slags on the binary cement containing high-volume fly ash, Powder [77] T. Stovall, F. de Larrard, M. Buil, Linear packing density model of grain mixtures,
Technol. 375 (2020) 539–548. Powder Technol. 48 (1) (1986) 1–12.
[48] P.-C. Aïtcin, S.L. Sarkar, M. Regourd, D. Volant, Retardation effect of [78] K. Wille, A.E. Naaman, G.J. Parra-Montesinos, Ultra-high performance concrete
superplasticizer on different cement fractions, Cement Concr. Res. 17 (6) (1987) with compressive strength exceeding 150 MPa (22 ksi): a simpler way, ACI Mater.
995–999. J. 108 (1) (2011), 46-34.
[49] S. Sha, M. Wang, C. Shi, Y. Xiao, Influence of the structures of polycarboxylate [79] A. Arora, M. Aguayo, H. Hansen, C. Castro, E. Federspiel, B. Mobasher,
superplasticizer on its performance in cement-based materials-A review, N. Neithalath, Microstructural packing- and rheology-based binder selection and
Construct. Build. Mater. 233 (2020) 117257. characterization for Ultra-high Performance Concrete (UHPC), Cement Concr.
[50] P.P. Li, Q.L. Yu, H.J.H. Brouwers, Effect of PCE-type superplasticizer on early-age Res. 103 (2018) 179–190.
behaviour of ultra-high performance concrete (UHPC), Construct. Build. Mater. [80] F. De Larrard, T. Sedran, Optimization of ultra-high-performance concrete by the
153 (2017) 740–750. use of a packing model, Cement Concr. Res. 24 (6) (1994) 997–1009.
19
[81] K.H. Hoang, P. Hadl, V. Nguyen, A New Mix Design Method for UHPC Based on [113] J. Van Der Putten, J. Dils, P. Minne, V. Boel, G. De Schutter, Determination of
Stepwise Optimization of Particle Packing Density, First International Interactive packing profiles for the verification of the compressible packing model in case of
Symposium on UHPC, 2016. UHPC pastes, Mater. Struct. 50 (2) (2016) 118.
[82] D. Fan, R. Yu, Z. Shui, C. Wu, J. Wang, Q. Su, A novel approach for developing a [114] D. Hou, D. Wu, X. Wang, S. Gao, R. Yu, M. Li, P. Wang, Y. Wang, Sustainable use
green Ultra-High Performance Concrete (UHPC) with advanced particles packing of red mud in ultra-high performance concrete (UHPC): design and performance
mesostructure, Construct. Build. Mater. 265 (2020) 120339. evaluation, Cement Concr. Compos. 115 (2021) 103862.
[83] X. Wang, R. Yu, Q. Song, Z. Shui, Z. Liu, S. Wu, D. Hou, Optimized design of ultra- [115] W. Yao, G. Guangsheng, W. Fei, W. Jun, Fluidization and agglomerate structure of
high performance concrete (UHPC) with a high wet packing density, Cement SiO2 nanoparticles, Powder Technol. 124 (1–2) (2002) 152–159.
Concr. Res. 126 (2019) 105921. [116] L. Teng, J. Zhu, K.H. Khayat, J. Liu, Effect of welan gum and nanoclay on
[84] E. Ghafari, H. Costa, E. Júlio, RSM-based model to predict the performance of thixotropy of UHPC, Cement Concr. Res. 138 (2020) 106238.
self-compacting UHPC reinforced with hybrid steel micro-fibers, Construct. Build. [117] H.H.C. Wong, A.K.H. Kwan, Packing density of cementitious materials: part
Mater. 66 (2014) 375–383. 1—measurement using a wet packing method, Mater. Struct. 41 (4) (2008)
[85] E. Ghafari, H. Costa, E. Júlio, Statistical mixture design approach for eco-efficient 689–701.
UHPC, Cement Concr. Compos. 55 (2015) 17–25. [118] F. De Larrard, Concrete Mixture Proportioning: a Scientific Approach, CRC Press,
[86] Y. Sun, R. Yu, Z. Shui, X. Wang, D. Qian, B. Rao, J. Huang, Y. He, Understanding 1999.
the porous aggregates carrier effect on reducing autogenous shrinkage of ultra- [119] A.K.H. Kwan, K.W. Chan, V. Wong, A 3-parameter particle packing model
high performance concrete (UHPC) based on response surface method, Construct. incorporating the wedging effect, Powder Technol. 237 (2013) 172–179.
Build. Mater. 222 (2019) 130–141. [120] I. Mehdipour, K.H. Khayat, Understanding the role of particle packing
[87] E. Ghafari, M. Bandarabadi, H. Costa, E. Júlio, Prediction of fresh and hardened characteristics in rheo-physical properties of cementitious suspensions: a
state properties of UHPC: comparative study of statistical mixture design and an literature review, Construct. Build. Mater. 161 (2018) 340–353.
artificial neural network model, J. Mater. Civ. Eng. 27 (11) (2015). [121] E. Pedersen, Packing calculations applied for concrete mix design, in: Proceedings
[88] E. Ghafari, M. Bandarabadi, H. Costa, E. Júlio, Design of UHPC using artificial of the International Conference Held at the University of Dundee, Thomas
neural networks, Brittle Matrix Compos 10 (2012) 61–69. Telford, 1999, Scotland, 1999, p. 121.
[89] O. Abuodeh, J.A. Abdalla, R.A. Hawileh, Prediction of compressive strength of [122] J. Dewar, Computer Modelling of Concrete Mixtures, CRC Press, 1999.
ultra-high performance concrete using SFS and ANN, in: 2019 8th International [123] B. Puri, B. Brown, H. McKee, I. Treasaden, Manual of Ready-Mixed Concrete, CRC
Conference on Modeling Simulation and Applied Optimization, ICMSAO), 2019, Press, 1992.
pp. 1–5. [124] S. Shah, R. Sharma, S. Paswan, A. Bhandari, S.R. Arukala, R.K. Pancharathi,
[90] T.C. Powers, Properties of Fresh Concrete, 1968, p. 301. Prioritizing the aggregate source based on particle packing density using modified
[91] R.B. Aim, P. Le Goff, Effet de paroi dans les empilements désordonnés de sphères Toufar Model and MCDM methods, Adv. Sustain. Constr. Mater. (2020) 171–182.
et application à la porosité de mélanges binaires, Powder Technol. 1 (5) (1968) [125] M. Mangulkar, S. Jamkar, Review of particle packing theories used for concrete
281–290. mix proportioning, Int. J. Sci. Eng. Res. 141 (2013).
[92] D.M. Roy, B.E. Scheetz, M.R. Silsbee, Processing of optimized cements and [126] C.Z.N. Wang, The theoretic analysis of the influence of the particle size
concretes via particle packing, MRS Bull. 18 (3) (1993) 45–49. distribution of cement system on the property of cement, Cement Concr. Res.
[93] P. Goltermann, V. Johansen, L. Palbol, Packing of aggregates: an alternative tool (1999).
to determine the optimal aggregate mix, ACI Mater. J. 94 (5) (1997) 435–443. [127] L.G. Li, A.K.H. Kwan, Packing density of concrete mix under dry and wet
[94] A. Yu, R.P. Zou, N. Standish, Modifying the linear packing model for predicting conditions, Powder Technol. 253 (2014) 514–521.
the porosity of nonspherical particle mixtures, Ind. Eng. Chem. Res. 35 (10) [128] L.G. Li, H.X. Zhuo, J. Zhu, A.K.H. Kwan, Packing density of mortar containing
(1996) 3730–3741. polypropylene, carbon or basalt fibres under dry and wet conditions, Powder
[95] F. de Larrard, T. Sedran, Mixture-proportioning of high-performance concrete, Technol. 342 (2019) 433–440.
Cement Concr. Res. 32 (11) (2002) 1699–1704. [129] X. Wang, J. Huang, B. Ma, S. Dai, Q. Jiang, H. Tan, Effect of mixing sequence of
[96] W.B. Fuller, S.E. Thompson, The Laws of Proportioning Concrete, 1907. calcium ion and polycarboxylate superplasticizer on dispersion of a low grade
[97] G. Shakhmenko, J. Birsh, Concrete mix design and optimization, in: 2nd Int. PhD silica fume in cement-based materials, Construct. Build. Mater. 195 (2019)
Symposium in Civil Engineering, 1998. Budapest. 537–546.
[98] L.G. Li, A.K.H. Kwan, Effects of superplasticizer type on packing density, water [130] J. Liu, K. Wang, Q. Zhang, F. Han, J. Sha, J. Liu, Influence of superplasticizer
film thickness and flowability of cementitious paste, Construct. Build. Mater. 86 dosage on the viscosity of cement paste with low water-binder ratio, Construct.
(2015) 113–119. Build. Mater. 149 (2017) 359–366.
[99] C. Shi, D. Wang, X. An, D. Jiao, H. Li, Method for mixture design of concrete with [131] K. Yamada, T. Takahashi, S. Hanehara, M. Matsuhisa, Effects of the chemical
multiple performance requirements, J. Chin. Ceram. Soc. 46 (2) (2018) 230–238. structure on the properties of polycarboxylate-type superplasticizer, Cement
[100] I. Mehdipour, K.H. Khayat, Effect of particle-size distribution and specific surface Concr. Res. 30 (2) (2000) 197–207.
area of different binder systems on packing density and flow characteristics of [132] S. Chandra, J. Björnström, Influence of cement and superplasticizers type and
cement paste, Cement Concr. Compos. 78 (2017) 120–131. dosage on the fluidity of cement mortars—Part I, Cement Concr. Res. 32 (10)
[101] Q. Wu, X. An, Development of a mix design method for SCC based on the (2002) 1605–1611.
rheological characteristics of paste, Construct. Build. Mater. 53 (2014) 642–651. [133] G. Gelardi, R.J. Flatt, Working mechanisms of water reducers and
[102] W. Meng, M. Valipour, K.H. Khayat, Optimization and performance of cost- superplasticizers, Sci. Technol. Concr. Admixtures. (2016) 257–278.
effective ultra-high performance concrete, Mater. Struct. 50 (1) (2016) 29. [134] Y. Zhang, X. Luo, X. Kong, F. Wang, L. Gao, Rheological properties and
[103] Z. Li, D. Lu, X. Gao, Optimization of mixture proportions by statistical microstructure of fresh cement pastes with varied dispersion media and
experimental design using response surface method - a review, J. Build. Eng. 36 superplasticizers, Powder Technol. 330 (2018) 219–227.
(2021) 102101. [135] A.K.H. Kwan, W.W.S. Fung, Roles of water film thickness and SP dosage in
[104] M.A. Mosaberpanah, O. Eren, Statistical models for mechanical properties of rheology and cohesiveness of mortar, Cement Concr. Compos. 34 (2) (2012)
UHPC using response surface methodology, Comput. Concr. 19 (6) (2017) 121–130.
667–675. [136] M. Daimon, D.M. Roy, Rheological properties of cement mixes: I. Methods,
[105] I. Ferdosian, A. Camões, Eco-efficient ultra-high performance concrete preliminary experiments, and adsorption studies, Cement Concr. Res. 8 (6) (1978)
development by means of response surface methodology, Cement Concr. Compos. 753–764.
84 (2017) 146–156. [137] C. Liu, C. Zhang, C. Zhou, Y. Mi, Z. Wang, Effects of the solidification of capillary
[106] W. Meng, V. Samaranayake, K.H. Khayat, Factorial design and optimization of bridges on the interaction forces between hydrate particles, Energy Fuels 34 (4)
ultra-high-performance concrete with lightweight sand, ACI Mater. J. 115 (1) (2020) 4525–4533.
(2018). [138] A. Gruskovnjak, B. Lothenbach, F. Winnefeld, R. Figi, S.C. Ko, M. Adler, U. Mäder,
[107] N.A. Soliman, A. Tagnit-Hamou, Using particle packing and statistical approach Hydration mechanisms of super sulphated slag cement, Cement Concr. Res. 38 (7)
to optimize eco-efficient ultra-high-performance concrete, ACI Mater. J. 114 (6) (2008) 983–992.
(2017). [139] J.W. Bullard, H.M. Jennings, R.A. Livingston, A. Nonat, G.W. Scherer, J.
[108] C. Shi, D. Wang, L. Wu, Z. Wu, The hydration and microstructure of ultra high- S. Schweitzer, K.L. Scrivener, J.J. Thomas, Mechanisms of cement hydration,
strength concrete with cement–silica fume–slag binder, Cement Concr. Compos. Cement Concr. Res. 41 (12) (2011) 1208–1223.
61 (2015) 44–52. [140] N.K. Lee, K.T. Koh, M.O. Kim, G.S. Ryu, Uncovering the role of micro silica in
[109] D.E. Rumelhart, G.E. Hinton, R.J. Williams, Learning Internal Representations by hydration of ultra-high performance concrete (UHPC), Cement Concr. Res. 104
Error Propagation, California Univ San Diego La Jolla Inst for Cognitive Science, (2018) 68–79.
1985. [141] H.C. Pedrosa, O.M. Reales, V.D. Reis, M.d.D. Paiva, E.M.R. Fairbairn, Hydration
[110] X. Tan, H. Sun, F. Nan, Application of Improved BP Algorithm in Prediction of of Portland cement accelerated by C-S-H seeds at different temperatures, Cement
Axial Tension Performance and Crack Width Control Ability of UHPC, 2020 Concr. Res. 129 (2020) 105978.
International Wireless Communications and Mobile Computing, IWCMC), 2020, [142] J. Qiu, Z. Guo, L. Yang, H. Jiang, Y. Zhao, Effects of packing density and water
pp. 1972–1976. film thickness on the fluidity behaviour of cemented paste backfill, Powder
[111] D. Qu, X. Cai, W. Chang, Evaluating the effects of steel fibers on mechanical Technol. 359 (2020) 27–35.
properties of ultra-high performance concrete using artificial neural networks, [143] H. Ye, X. Gao, R. Wang, H. Wang, Relationship among particle characteristic,
Appl. Sci. 8 (7) (2018). water film thickness and flowability of fresh paste containing different mineral
[112] Y. Shi, G. Long, X. Zen, Y. Xie, T. Shang, Design of binder system of eco-efficient admixtures, Construct. Build. Mater. 153 (2017) 193–201.
UHPC based on physical packing and chemical effect optimization, Construct. [144] I. Navarrete, Y. Kurama, N. Escalona, M. Lopez, Impact of physical and
Build. Mater. 274 (2021) 121382. physicochemical properties of supplementary cementitious materials on
20
structural build-up of cement-based pastes, Cement Concr. Res. 130 (2020) [165] K. Vance, A. Kumar, G. Sant, N. Neithalath, The rheological properties of ternary
105994. binders containing Portland cement, limestone, and metakaolin or fly ash,
[145] L.G. Li, A.K.H. Kwan, Concrete mix design based on water film thickness and Cement Concr. Res. 52 (2013) 196–207.
paste film thickness, Cement Concr. Compos. 39 (2013) 33–42. [166] S. Mindess, J. Skalny, Fiber-reinforced cementitious materials, J. Mater. Civ. Eng.
[146] A.K.H. Kwan, L.G. Li, W.W.S. Fung, Wet packing of blended fine and coarse 5 (3) (1991) 405–483.
aggregate, Mater. Struct. 45 (6) (2012) 817–828. [167] Z. Wu, K.H. Khayat, C. Shi, How do fiber shape and matrix composition affect
[147] W.W.S. Fung, A.K.H. Kwan, H.H.C. Wong, Wet packing of crushed rock fine fiber pullout behavior and flexural properties of UHPC? Cement Concr. Compos.
aggregate, Mater. Struct. 42 (5) (2009) 631–643. 90 (2018) 193–201.
[148] A.K.H. Kwan, H.H.C. Wong, Packing density of cementitious materials: part [168] N. Ozyurt, T.O. Mason, S.P. Shah, Correlation of fiber dispersion, rheology and
2—packing and flow of OPC + PFA + CSF, Mater. Struct. 41 (4) (2007) 773. mechanical performance of FRCs, Cement Concr. Compos. 29 (2) (2007) 70–79.
[149] A.K.H. Kwan, J.J. Chen, Adding fly ash microsphere to improve packing density, [169] L. Martinie, P. Rossi, N. Roussel, Rheology of fiber reinforced cementitious
flowability and strength of cement paste, Powder Technol. 234 (2013) 19–25. materials: classification and prediction, Cement Concr. Res. 40 (2) (2010)
[150] R. Yu, P. Spiesz, H.J.H. Brouwers, Effect of nano-silica on the hydration and 226–234.
microstructure development of ultra-high performance concrete (UHPC) with a [170] S. Jiang, B. Shan, J. Ouyang, W. Zhang, X. Yu, P. Li, B. Han, Rheological
low binder amount, Construct. Build. Mater. 65 (2014) 140–150. properties of cementitious composites with nano/fiber fillers, Construct. Build.
[151] A. Yu, J. Bridgwater, A. Burbidge, On the modelling of the packing of fine Mater. 158 (2018) 786–800.
particles, Powder Technol. 92 (3) (1997) 185–194. [171] L. Ferrara, Y.D. Park, S.P. Shah, Correlation among fresh state behavior, fiber
[152] W. Meng, K.H. Khayat, Improving flexural performance of ultra-high-performance dispersion, and toughness properties of SFRCs, J. Mater. Civ. Eng. 20 (7) (2016)
concrete by rheology control of suspending mortar, Compos. B Eng. 117 (2017) 493–501.
26–34. [172] K. Marar, Ö. Eren, H. Roughani, The influence of amount and aspect ratio of fibers
[153] K.H. Khayat, W. Meng, K. Vallurupalli, L. Teng, Rheological properties of ultra- on shear behaviour of steel fiber reinforced concrete, KSCE. J. Civ. Eng. 21 (4)
high-performance concrete — an overview, Cement Concr. Res. 124 (2019) (2017) 1393–1399.
105828. [173] Ş. Yazıcı, G. İnan, V. Tabak, Effect of aspect ratio and volume fraction of steel
[154] K.G. Kuder, N. Ozyurt, E.B. Mu, S.P. Shah, Rheology of fiber-reinforced fiber on the mechanical properties of SFRC, Construct. Build. Mater. 21 (6) (2007)
cementitious materials, Cement Concr. Res. 37 (2) (2007) 191–199. 1250–1253.
[155] H. Vikan, H. Justnes, F. Winnefeld, R. Figi, Correlating cement characteristics [174] D.V. Soulioti, N.M. Barkoula, A. Paipetis, T.E. Matikas, Effects of fibre geometry
with rheology of paste, Cement Concr. Res. 37 (11) (2007) 1502–1511. and volume fraction on the flexural behaviour of steel-fibre reinforced concrete,
[156] J.J. Chen, A.K.H. Kwan, Superfine cement for improving packing density, Strain 47 (s1) (2011) e535–e541.
rheology and strength of cement paste, Cement Concr. Compos. 34 (1) (2012) [175] R. Wang, X. Gao, J. Zhang, G. Han, Spatial distribution of steel fibers and air
1–10. bubbles in UHPC cylinder determined by X-ray CT method, Constr, Build. Mater.
[157] J. Hu, K. Wang, Effect of coarse aggregate characteristics on concrete rheology, 160 (2018) 39–47.
Construct. Build. Mater. 25 (3) (2011) 1196–1204. [176] R. Yu, P. Spiesz, H.J.H. Brouwers, Mix design and properties assessment of ultra-
[158] O.R. Khaleel, H. Abdul Razak, Mix design method for self compacting metakaolin high performance fibre reinforced concrete (UHPFRC), Cement Concr. Res. 56
concrete with different properties of coarse aggregate, Mater. Des. 53 (2014) (2014) 29–39.
691–700. [177] R. Wang, X. Gao, Relationship between flowability, entrapped air content and
[159] S.H. Lee, H.J. Kim, E. Sakai, M. Daimon, Effect of particle size distribution of fly strength of UHPC mixtures containing different dosage of steel fiber, Appl. Sci. 6
ash–cement system on the fluidity of cement pastes, Cement Concr. Res. 33 (5) (8) (2016) 216.
(2003) 763–768. [178] L. Teng, W. Meng, K.H. Khayat, Rheology control of ultra-high-performance
[160] D.P. Bentz, C.F. Ferraris, M.A. Galler, A.S. Hansen, J.M. Guynn, Influence of concrete made with different fiber contents, Cement Concr. Res. 138 (2020)
particle size distributions on yield stress and viscosity of cement–fly ash pastes, 106222.
Cement Concr. Res. 42 (2) (2012) 404–409. [179] P. Barrett, Review of Statistical Models: Theory and Practice, 2009.
[161] R.P. Khatri, V. Sirivivatnanon, W. Gross, Effect of different supplementary [180] L. Eriksson, E. Johansson, C. Wikström, Mixture design—design generation, PLS
cementitious materials on mechanical properties of high performance concrete, analysis, and model usage, Chemometr. Intell. Lab. Syst. 43 (1–2) (1998) 1–24.
Cement Concr. Res. 25 (1) (1995) 209–220. [181] M.A. Nemeth, Response surface methodology: process and product optimization
[162] C. Artelt, E. Garcia, Impact of superplasticizer concentration and of ultra-fine using designed experiments, J. Qual. Technol. 35 (4) (2003) 428–429.
particles on the rheological behaviour of dense mortar suspensions, Cement [182] S. Shanmuganathan, Artificial Neural Network Modelling, 2016.
Concr. Res. 38 (5) (2008) 633–642. [183] M.A. DeRousseau, J.R. Kasprzyk, W.V. Srubar, Computational design
[163] M. Mazloom, A.A. Ramezanianpour, J.J. Brooks, Effect of silica fume on optimization of concrete mixtures: a review, Cement Concr. Res. 109 (2018)
mechanical properties of high-strength concrete, Cement Concr. Compos. 26 (4) 42–53.
(2004) 347–357. [184] V. Chandwani, V. Agrawal, R. Nagar, Modeling slump of ready mix concrete using
[164] M. Nehdi, S. Mindess, P.C. Aı¨tcin, Rheology of high-performance concrete: effect genetic algorithms assisted training of Artificial Neural Networks, Expert Syst.
of ultrafine particles, Cement Concr. Res. 28 (5) (1998) 687–697. Appl. 42 (2) (2015) 885–893.
21

Zhou - Mixture Design Methods For Ultra-High-Performance Concrete - A Review - 2021

Uploaded by

Copyright:

Available Formats

Zhou - Mixture Design Methods For Ultra-High-Performance Concrete - A Review - 2021

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Zhou - Mixture Design Methods For Ultra-High-Performance Concrete - A Review - 2021

Uploaded by

Copyright:

Available Formats

Cement and Concrete Composites 124 (2021) 104242

Contents lists available at ScienceDirect

Cement and Concrete Composites

Mixture design methods for ultra-high-performance concrete - a review

2. Mixture design methods of UHPC VC

Fig. 1. Packing density diagram of a granular mixture system.

Fig. 4. Illustration of particle wedging effect (adapted from Ref. [119]).

Binary Ternary Multi- Wall Loosening

1929 Furnas model ✓ Assumed that particles are independent [75]

Fig. 8. Comparison of dispersion of particles with and without superplasticizer.

w/b ratio (unit-less).

Fig. 9. Mixture proportion based on compact wet packing of particles.

cementitious systems were selected with commonly used content ranges.

Fig. 16. Distribution of steel fibers in UHPC cylinders [175].

Fig. 19. Design steps of RSM (adapted from Ref. [103]).

Fig. 24. Typical structure diagram of ANN [88].

You might also like