Construction and Building Materials 86 (2015) 33–43

Contents lists available at ScienceDirect

Construction and Building Materials

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

Material efficiency in the design of ultra-high performance concrete

Kay Wille ⇑, Christopher Boisvert-Cotulio
Department of Civil and Environmental Engineering, University of Connecticut, United States

h i g h l i g h t s

 Material efficiency in ultra-high performance concrete design.

 Using spread value as quick indicator to fine-tune mix design.
 Combine workability, mechanical performance and costs to conclude efficiency.
 Provide UHPC mix designs using materials available in the US.
 Material costs are mainly influenced by fiber reinforcement.

a r t i c l e i n f o a b s t r a c t

Article history: This research investigates the material efficiency in the design of ultra-high performance concrete. The
Received 15 October 2014 material efficiency is influenced by the flowability, mechanical performance, durability and cost.
Received in revised form 16 March 2015 Suitable material constituents have been pre-selected based on their properties and availability in the
Accepted 19 March 2015
United States of America. The efficiency of the constituents is progressively investigated using the follow-
Available online 5 April 2015
ing three step approach emphasizing: (1) ultra-high performance paste, (2) ultra-high performance
matrix, and (3) ultra-high performance fiber composite. Mix design recommendations for performance
Keywords:
and cost effective ultra-high performance concretes are made and conclusions for further enhancements
Ultra-high performance concrete
Ultra-high performance fiber reinforced
are drawn.
concrete Ó 2015 Elsevier Ltd. All rights reserved.
Performance
Efficiency
Cost

1. Introduction the cost of conventional concrete of about $US100/yd3

($US130 per m3), commercially available UHPC is about 20 times
Ultra-high performance concrete (UHPC) has the potential to more expensive. The proprietary nature, increased quality control
address the poor condition of the ageing infrastructure in the and high material costs have hindered an accelerated and wide
United States of America, rated with a D+ by the American spread use of UHPC in the U.S. infrastructure. Addressing these
Society of Civil Engineers (ASCE) in 2013 [1]. ASCE estimated that concerns, this research provides methodical recommendations for
an investment of $US 3.6 trillion is needed by 2020 to remedy material efficiency in the design of UHPC. The efficiency of selected
the infrastructure. The two main reasons for the poor conditions material components will be shown and several UHPC mix propor-
are identified as corrosion of steel reinforcement and concrete tions will be provided based on local available materials.
deterioration through invasive ions. UHPC’s durability and low
permeability against chlorides, sulfates, carbon dioxide and other
2. Definition and material selection
aggressors are the key properties to building structural elements
of longer lifespan and reduced maintenance. UHPC conferences in
Based on information from Rossi [7,8] and international confer-
Kassel, Germany in 2004, 2008, 2012 [2–4] and Marseille, France
ences [2–6] the American Concrete Institute (ACI) committee 239
in 2009, 2013 [5,6] have demonstrated the material’s mechanical
drafted the definition for UHPC as follows: ‘‘Concrete, ultra-high
and durability performance in this regard. In comparison to
performance – concrete with a minimum specified compressive
strength of 150 MPa (22,000 psi) with specified durability, tensile
⇑ Corresponding author at: 261 Glenbrook Road, Unit 3037, Storrs, CT 06269- ductility and toughness requirements; fibers are usually required
3037, United States. Tel.: +1 (860) 486 2074. to achieve specified requirements.’’ A summary of chronological
E-mail address: kwille@engr.uconn.edu (K. Wille). advances in matrix and fiber development, as well as the historical

http://dx.doi.org/10.1016/j.conbuildmat.2015.03.087
0950-0618/Ó 2015 Elsevier Ltd. All rights reserved.
34 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43

development of different high and ultra-high performance con- density has to be increased such that the volume of water-filled
crete mix designs from the 1970s on, is provided by Naaman and voids is reduced. Therefore, less water is physically trapped, leav-
Wille [9]. To the best of the author’s knowledge the first non-pro- ing more of the remaining water available to cover the surface of
prietary concrete mix design achieving ultra-high compressive the particles. This increases the thickness of the water film on
strength in excess of 150 MPa by using materials available in the the surface of the particles and decreases the overall viscosity of
U.S. without the application of heat or pressure treatment has been the paste. Thus, by improving the rheological behavior, the water
reported by Wille et al. [10]. Research efforts by Wille et al. [11–14] to cement ratio (w/c) can be reduced to maintain the same worka-
show that UHPC can be designed achieving compressive strength bility, which is one necessary condition to achieve an ultra-high
in excess of 200 MPa (29 ksi) by using materials available in the strength paste and thus UHPC. Furthermore, increasing the particle
U.S. under ambient curing conditions without the need of special packing density decreases the porosity of the paste, which is the
treatment. The basic principles of UHPC design include high parti- key parameter for improving durability performance [15].
cle packing density (low porosity), high material quality (low Therefore, focusing material design on particle packing optimiza-
impurity), cement hydration chemistry (high density calcium–sili- tion and costs simplifies the research activities.
cate–hydrate [C–S–H]), pozzolanic reactions and filler effect of Based on the relative workability (spread), the relative com-
0
supplemental materials (C–S–H formation and low porosity), high pressive strength ðf c Þ and the relative costs of the paste, the
particle dispersion quality (low porosity and enhanced workabil- dimensionless efficiency parameter E is introduced (Eq. (1)):
ity), optimized particle to high range water reducer (HRWR) inter-
action (enhanced particle dispersion) and excess paste (enhanced f0 spreadN
0:7  f 0 c;N þ 0:3  spreadN;£
workability and robustness). Based on these principles and on E¼ c;N;£
; ð1Þ
costP
the experimental results of prior research [11–14] the material costP;£
constituents for designing UHPC are pre-defined and their approxi-
0
mated median particle size, as well as their range of particle size where f c;N is the 28-day compressive strength normalized at w/
distribution, are listed as recommendations in Table 1. 0
c = 0.25 and at an air content of 3%, f c;N;£ is the average normalized
Additionally prior research results [11–14] suggest the follow- 28-day compressive strength over all pastes of one series, spreadN is
ing mixture proportions for designing UHPC: the spread value normalized at w/c = 0.25, spreadN;£ is the average
normalized spread value over all pastes of one series, costF is the
 Cement (C):silica fume (SF):supplemental material (SM) =
cost of the paste per m3, and costP;£ is the average cost over all
1:0.25:0.25 by weight,
pastes of one series.
 Water (W) to cement (C) = 0.2–0.3 by weight,
The efficiency parameter has been derived to reflect the paste
 Aggregate (A):cement (C) = 1.0–2.0 by weight,
performance over its material costs. Based on the research
 Fiber volume fraction (Vf) = 1.0–2.0 Vol.%.
approach targeting pastes with low porosity and high particle
packing density, thus high compressive strength and workability,
3. Research approach the performance is indicated by the relative strength and relative
workability. The factors 0.7 and 0.3 have been chosen to consider
In order to satisfy time efficiency and material performance in strength with higher priority over workability. While the relative
the UHPC design the following three step progression is proposed performance is defined to be directly proportional to the efficiency
and used in this research: (Level 1) investigation of the cementi- parameter, the relative costs has been defined to be indirectly
tious paste (C + SF + SM + HRWR + W), (Level 2) investigation of proportional.
the cementitious matrix (paste + aggregate), and (Level 3) The effect of each material component on the paste’s relative
investigation of the cementitious composite (matrix + fiber). The strength, spread and cost is evaluated. Then the material compo-
key component in UHPC design is the optimization of the paste nent with the best efficiency out of each series is chosen to form
in its particle packing density providing the basis for mechanical the optimized paste, which is used for the material design of
and durability performance. Based on the strong correlation Level 2 (matrix) and Level 3 (composite).
between rheological and mechanical performance of the cementi- Once the paste has been designed the addition of aggregates fol-
tious paste [11], changes in particle packing density can be lows leading to design Level 2. In accordance to Level 1 the
assessed indirectly through a spread test in accordance with investigation of the matrix is focused on its compressive strength,
ASTM C 230/C 230 M. To increase the flowability of the paste while its workability and its cost efficiency. Investigation parameters
maintaining the amount of water constant (or to maintain the include the type and form of aggregate, the maximum aggregate
same flowability while reducing the water content), the packing size, the particle size distribution, and the aggregate to cement

Table 1
Recommended material constituents for UHPC matrix design based on [11–14].

Type Costa Particle size in lm Comments

$US/ton Median D10% D90%
Water (W) – – – – –
HRWR 13–20b – – – Best in workability and air release
Silica fume (SF) 350–1100 0.2–1 0.1 2 Low carbon content
Supplemental material (SM) 46–879 2–5 1 10 Filler effect, spherical shape and pozzolanic reaction preferred
Cement (C) 92–250 10–20 3 40 Low C3A, high C3S + C2S
Fine agg. 1 8.5–162.5 100 >50 <300 High quality, high strength, low water absorption, optimized particle packing
Fine agg. 2 8.5–162.5 500 >300 <1000
Coarse agg. 8.25–19 >1000 <9000
Fibers 2800–13,300 Tailored bond with matrix, sufficient tensile strength to prevent fiber failure
a
Costs of the material components used in this research.
b
Costs per gallon (3.8 L).
 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43 35

Table 2
Cement properties, costs and performance.

Type Name C2S C3S C2S + C3S C3A C4AF Blaine (m2/kg) Cost w/ca Spreada Compressive strengtha f0 c
(%) (%) (%) (%) (%) $US/ton (mm) MPa (ksi)
White PC I C1 13 74 87 5 1 395 250 0.21 315 171 (24.75)
PC II C2 15 54 69 7 10 377 92 0.26 266 131 (19.01)
PC I/II C3 13 59 72 7 11 396 110 0.21 256 128 (18.50)
PC III C4 14 51 65 8 9 411 92 0.28 265 135 (19.60)
PC III C5 16 53 69 9 8 435 110 0.24 280 146 (21.24)
PC II/V C6 14 58 72 4 11 417 110 0.23 330 184 (26.64)
Oil well cement C7 16 59 75 0.4 18 214 130 0.23 350 183 (26.57)
PC II/V C8 17 59 76 4 15 430 115 0.23 300 196 (28.42)
White PC I C9 17 62 79 9 1 582 250 0.23 300 221 (32.07)
Oil well cement C10 15 59 74 0.3 18 417 130 0.23 300 177 (25.73)
White PC I C11 11 70 81 10 1 518 250 0.24 319 186 (26.97)
PC III C12 13 58 71 8 10 600 N/A 0.27 230 170 (24.71)
a
Values of the reference paste using the type of cement.

Table 3
Silica fume properties, costs and performance.

Type Name SiO2 (%) Carbon (%) Na2O (%) K2O (%) Median particle size (lm) Cost $US/ton w/ Spreada (mm) Compressive strengtha f0 c MPa (ksi)
ca
Gray SF1 >85 0.3 < 0.7 0.1 0.07 0.4 550 0.21 315 171 (24.75)
Gray SF2 >97.5 <0.8 <0.3 <0.6 0.5 1100 0.33 400 175 (25.41)
Gray SF3 >93 2.5b<6.0b <0.4 <0.9 0.6 350 0.31 366 192 (27.87)
White SF4 >96 <0.2 0.15 1000 0.25 329 175 (25.38)
Gray SF5 95 0.6 0.96%c 500 0.23 290 174 (25.29)
a
Values of the reference paste using the type of silica fume.
b
Loss of ignition (LOI).
c
Percent retained on 45 lm diameter sieve.

ratio (A/C). At the composite level (Level 3) fiber reinforcement is C3S) and dicalcium silicate (Ca2SiO4 or C2S) were selected due to
added to the mixture enhancing ductility, sustained tensile their favorable strength development. Furthermore, cements with
strength and limiting crack width. tricalcium aluminate (Ca3Al2O6 or C3A) of <8% are preferred as sug-
gested by Sakai et al. [16]. The hydration process of C3A is very
4. Properties of material constituents rapid and increases the surface area of the particles, leading to
higher water demand, higher viscosity and thus limiting the reduc-
4.1. Paste constituents tion of the w/c ratio. Most Portland cements of type I and III avail-
able in the U.S. have a C3A content higher than 8%. Those cements
In this research materials were preselected based on availabil- with a C3A content of about 8% and lower were included in the
ity, cost, particle size distribution, purity and chemical and physi- investigation. Some Portland cements of type II/V are characterized
cal composition. Twelve cements, five silica fumes, thirteen by a low C3A content and thus selected to be included in the
supplemental materials, eight high range water reducers, eleven research program. However, their availability is limited to the
aggregate variations and four different fiber reinforcements were Eastern regions of the U.S. Oil well cements were included due to
included for experimental investigation. Table 2 summarizes the their very low C3A content. White Portland cements were selected
clinker phase proportions, Blaine fineness and costs of the selected due to their moderate C3A content and the very low amount of
cements. Cements rich in the sum of tricalcium silicate (Ca3SiO5 or C4AF leading to highest values of C2S + C3S, up to 87%. The

Table 4
Supplemental material properties, costs and performance.

Type Name Median particle size (lm) Cost $US/ton w/ca Spreada (mm) Compressive strengtha f0 c MPa (ksi)
Silica powder SM1 1.7 879 0.21 315 171 (24.75)
Fly ash SM2 25 60 0.23 340 192 (27.88)
Fly ash SM3 9.4 46 0.23 364 197 (28.59)
Fly ash SM4 12.7 46 0.23 341 192 (27.87)
Fly ash SM5 10.4 N/A 0.23 313 170 (24.71)
Metakaolin SM6 1.4 500 0.31 333 164 (23.74)
Metakaolin SM7 3 500 0.32 220 160 (23.25)
GGBS SM8 1.5%b 100 0.24 295 202 (29.31)
GGBS SM9 2.0%b 102 0.24 300 181 (26.27)
Fly ash SM10 14%b N/A 0.25 395 187 (27.16)
Limestone powder SM11 3.2 N/A 0.23 300 194 (28.21)
Limestone powder SM12 14%b 122 0.23 260 164 (23.82)
Limestone powder SM13 0.6%b 124 0.23 295 160 (23.22)
a
Values of the reference paste using the type of supplemental material.
b
Percent retained on 45 lm diameter sieve.
36 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43

Table 5 Table 3 summarizes the selected silica fumes with their composi-
HRWR costs and performance. tions and costs. Whereas SF1, SF2, SF4 and SF5 were selected due
Name Cost $US/gal w/ca Spreada (mm) Compressive strengtha to their low carbon content, SF3 was included in the research pro-
f0 c MPa (ksi) gram due to its low cost of $US350 per ton.
HRWR1 20 0.21 315 171 (24.75) Supplemental materials (Table 4) were selected based on med-
HRWR2 13 0.28 100 139 (20.19) ian particle size, preferably about 2–5 lm, availability and costs.
HRWR3 20 0.25 223 191 (27.70) Silica powder (SM1) was included here due to its high material
HRWR4 15 0.26 248 191 (27.72)
HRWR5 14 0.26 223 166 (24.12)
purity and its low median particle size. Four different types of fly
HRWR6 13 0.26 250 171 (24.84) ash were selected for investigation, due to their low cost of about
HRWR7 14 0.27 220 175 (25.42) $US50–$US60 per ton, their spherical particle shape for enhanced
HRWR8 14 0.26 220 167 (24.18) flowability, their acceptable median particle size of about 10 lm,
a
Values of the reference paste using the type of supplemental material. their pozzolanic potential, their abundant availability and their
positive environmental impact. Two types of metakaolin were
included in this research due to their high purity, their low median
Table 6 particle size and their pozzolanic reaction. The high material costs
Fine aggregate nomenclature and costs. of about $US500 per ton potentially reduce the application of this
Type Name Cost $US/ton material in a cost efficient UHPC design. Ground granulated blast
Quartz Q 162.51
furnace slag (GGBS) and limestone powder were selected due to
Basalt FB 13.45 their low cost, their availability, and their suitable particle size.
Limestone FL 8.50 Eight different polycarboxylate ether based high range water
Volcanic rock FVR1 32.00 reducers (HRWR) ranging between $US13 and $US20 per gallon
Volcanic rock FVR2 32.00
were investigated in their efficiency and are listed in Table 5.

Table 7
4.2. Aggregates
Coarse aggregate nomenclature and costs.
Aggregates were primarily selected based on locality, type of
Type Name Cost $US/ton
material, size, and cost. The following four types of aggregates
Basalt CB1 13.45 were included in the research quartz (Q), basalt (B), limestone (L)
Basalt CB2 19.00
and volcanic rock (VR). Basalt aggregates were chosen from the
Limestone CL 8.25
Volcanic rock CVR1 11.00 northeast, limestone from the upper-midwest, and volcanic rock
Volcanic Rock CVR2 11.00 from the northwest area, whereas, Quartz aggregate of high purity
Volcanic rock CVR3 11.00 was selected from a major supplier, which can be ordered in differ-
ent regions of the U.S. At least one type of fine aggregate (smaller
than 1.2 mm) and one type of coarse aggregate (smaller than
drawback of using white cements in a cost efficient UHPC mix 12.5 mm) was selected for each material based on gradation tables.
design are the high material costs of up to $US250 per ton, which Tables 6 and 7 show a list of the fine (F) and coarse (C) aggregates,
is about 2.5 times the price of regular Portland cement. respectively.
Silica fumes were pre-selected based on median particle size, Particle packing density of the aggregates is an important
high silicon dioxide content, low carbon content, and cost. The parameter to control workability, strength, and durability. In this
lower the carbon content and the lower the fineness of silica fume research the selected particle size distribution for all aggregates
the lower the water demand is, and thus, the better the flowability. follows the modified Andreasen and Andersen (A&A) curve [17]

Fig. 1. Particle size distribution (PSD) for fine and coarse aggregates in comparison to modified Andreasen and Andersen (A&A) curve.
 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43 37

Table 8 fiber and surrounding matrix. Several studies have been performed
Fiber nomenclature, dimensions and cost. to investigate the fiber effectiveness regarding single fiber pull-out
Name Type Material Diameter Length Strength Cost embedded in UHPC matrix [19,20] or with regard to direct tensile
(mm) (mm) MPa (ksi) $US/ton tests of fiber reinforced UHPC composites [11–14,21]. Although
F1 Straight Steel 0.2 13 1900 (285) 4760 mechanical bond provided by fiber end-hooks or twists is still ben-
F2 End- Steel 0.5 25 1100 (165) 2800 eficial in crack bridging performance, optimized UHPC matrix with
hooked high packing density allows for enhanced performance of smooth
F3 Straight Glass 0.01 12 1700 (225) 6060
high strength steel fibers, typically with a length-to-diameter ratio
F4 Straight PVA 0.06 12 1600 (240) 13300
of 65 and a fiber diameter of 0.2 mm. Reducing the performance
gap between smooth fibers and fibers with mechanical bond chal-
lenges the need for the latter, if the fiber material efficiency is com-
pensated by the reduced costs of smooth fibers. The selection of
steel fiber reinforcement was restricted to fibers produced in the
U.S. due to federal law (Buy American Act regulated by the
Federal Transit Administration) pertaining to federally funded
infrastructure construction projects. Based on availability, four
suitable fiber types listed in Table 8 have been selected for this
research.

5. Mixing and testing procedure

5.1. Mixing procedure

High particle dispersion quality, thus minimizing particle

(a) no LVDTs for paste and (b) LVDTs for composite agglomeration, is a key factor for achieving high particle packing
matrix testing testing density and enhanced material utilization. Interparticle attraction
forces increase with decrease of particle size and promote
Fig. 2. Setup for testing uniaxial compression strength/behavior. agglomerations. In comparison to conventional concrete the
homogeneous dispersion of small particles is challenging in
UHPC. Whereas conventional concrete includes cement (mean par-
(Fig. 1). It is worth noting that all aggregates were washed to mini- ticle size d50  15 lm) of about 10% per weight, small particles in
mize dust and dirt interfering with the particle size distribution of UHPC mix design, such as silica fume (d50  0.4 lm), silica powder
the mix and to reduce the demand of water. (d50  2 lm) and cement (d50  10 lm), can take up to 46% per
An optimum aggregate to cement ratio of 1.1 was recom- weight [12]. In pursuit of minimizing particle agglomerations,
mended in Park et al. [18]. For highest performance Wille et al. the material constituents were first mixed under dry conditions
[12] used an aggregate to cement ratio of 1.0. In this research an according to the mixing procedure in [11]. Although the paste does
aggregate to cement ratio is targeted to be between 1.0 and 2.0. not include sand or aggregate by definition, a small amount of sand
The higher this ratio can be achieved without significantly sacrifi- was initially mixed with silica fume for 5 min to facilitate breaking
cing workability and strength the more cost-effective the material up agglomerations of silica fume. For the paste, the sand to cement
will be, since the aggregates are significantly less expensive than ratio was kept at 0.68. For UHPC mixes, the full amount of aggre-
the ingredients of the paste. gate or sand was mixed for 5 min with silica fume. Next cement
and supplemental material was added dry and mixed for an addi-
4.3. Fiber reinforcement tional 5 min. One third of HRWR was mixed into the water and
added to the mix. Afterwards, the remaining HRWR was added.
The material efficiency of fiber reinforcement is mainly deter- This step-wise addition based on [22] leads to improved workabil-
mined by the fiber cost and the required fiber volume fraction to ity and more efficient use of the HRWR. An open, 1/2 horsepower
achieve specified mechanical and durability performance. The rotary bench top mixer with a bowl capacity of 11 L was used at
required fiber volume fraction depends on fiber strength, length- low to medium speeds for dry mixing and at high speed once turn-
to-diameter ratio, fiber orientation and bond behavior between over was achieved.

(a) Low amount of air voids (b) Medium amount of air (c) High amount of air voids
voids

Fig. 3. Top view of specimens with visibly enhanced air voids.

38 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43

5.3. Compressive strength

The compressive strength of each mixture (paste, matrix or com-

posite) was investigated by testing three cylinders (76 mm in
diameter and 152 mm in height) in a deformation controlled
hydraulic compression load frame with a capacity of 1740 kN
(400,000 lb force) (Fig. 2). The loading rate was selected to
0.5 mm/min leading to failure within 3–5 min. This equals to about
1 MPa/s (150 psi/s). Prior to testing, the specimens aged for 28 days
in a temperature controlled water bath of 20 °C. No heat, steam or
pressure treatment was applied. In addition to consistency in cur-
ing conditions, high consistency in compressive strength testing
(a) Test setup (b) Crack opening was sought by emphasizing high accuracy of the load faced cylinder
between LVDTs ends in terms of planeness and perpendicularity. By using a com-
bination of cylinder end grinder, rotary grinding/polishing machine,
Fig. 4. Setup for testing uniaxial tensile strength/behavior. and dilatometer, end planeness below 0.1° was achieved, which is
1/10th of the required value reported in [23]. No cylinder capping
material was used due to the insufficient material strength.
Table 9
Mixture design of reference paste.
5.4. Air content
Material Amount (kg/m3) Percent by weight
C1 989.1 41.8 Prior to testing, all specimens were precisely weighted and mea-
SF1 247.3 10.5 sured in size. The air content was back-calculated based on the
SM1 247.3 10.5 known mixture proportions of the material constituents and their
HRWR1 35.6 1.5 specific gravities. In addition, barium sulfate powder was applied
Sand (quartz-Q) 662.5 28
Water 182.8 7.7
to the loading face of the specimens to highlight the air voids for
digital photo-analysis. High resolution pictures (15.1 megapixels)
were taken to assess the air content. Matlab was used to analyze
5.2. Workability the pictures and calculating the air content by counting the number
of pixels representing barium sulfate powder. A good correlation
The workability of each mix (paste, matrix or composite) was between back-calculated air content and optically determined
investigated by testing the spread values according to ASTM value was obtained. Fig. 3 shows examples of specimens with
C230/C230M. For each mix the spread test were performed twice low, medium, and high amount of visible air voids.
and averaged. Special emphasis was placed on keeping the spread
cone and the base plate at similar humidity prior to testing. The 5.5. Tensile strength
targeted spread value of the paste was defined to be in the range
of 280–340 mm, supported by the analysis of previous research Additionally to the compression tests of the fiber reinforced com-
results [11] to achieve a sufficient release of entrapped air and posites their behavior under direct tension was investigated. The
sufficient performance in strength. summary of test setups and specimen geometries provided in [21]

Table 10
Material constituents of pastes investigated in this research.

Series Number of Mixes Cement Silica fume Supplemental material High range water reducer Sand Water
I (Table 2) 12 C1–C12 SF1 SM1 HRWR1 Q W
II (Table 3) 5 C1 SF1–SF5 SM1 HRWR1 Q W
III (Table 4) 13 C1 SF1 SM1–SM13 HRWR1 Q W
IV (Table 5) 8 C1 SF1 SM1 HRWR1–HRWR8 Q W

7 y = -0.0206x + 8.983
35
229
Compressive Strength f'c (ksi)

R² = 0.4189
Compressive Strength (MPa)

6
30 209
Air Content (percent)

5 189
4 25 169
149
3 20
129
2
15 109
1 y (MPa) = -151x + 224 89
R² = 0.25
0 10 69
200 250 300 350 400 0.1 0.2 0.3 0.4 0.5
Spread (mm) w/c . air1/3

(a) Air content of each series (b) Effect of w/c ratio and air content on
compressive strength
Fig. 5. Effect of workability on air content and thus on compressive strength.
 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43 39

shows the wide range of variation in direct tensile testing and the development of multiple cracking prior to forming one major crack
lack of a standardized method. For each series in this research three opening during softening of the material.
prism shaped specimens with geometry of (51 mm by 25 mm by
356 mm) were cast and tested at an age of 28 days (Fig. 4). The speci- 5.6. Freeze–Thaw resistance
mens were centrically placed in a displacement controlled hydraulic
tensile load frame with a capacity of 267 kN (60,000 lb force). The The degradation of UHPC to freezing and thawing cycles was
displacement control allowed capturing the load versus displace- quantified according to ASTM C666. UHPC prisms with the geometry
ment and thus the stress versus strain/crack opening relation of 76- by 102- by 406-mm (3- by 4- by 16-inch) were cast for this
beyond the tensile strength. The specimen’s ends were gripped with test. After 7 days of curing in a temperature controlled water bath
self-aligning, self-tightening mechanical jaws allowing straining the of 20° Celsius, the specimens were subjected to freezing and thaw-
specimen evenly. This mitigated bending stresses and thus prema- ing while submerged in a water bath. The apparatus used was set for
ture cracking. Two LVDTs were attached to the specimen’s side with four freeze–thaw cycles per day. The lower and upper temperature
a measurement length of 140 mm (5.5 inches). This allowed captur- values were set to 23 °C and 3 °C (9 F and 37 F), respectively.
ing potential strain hardening behavior, accompanied by the
6. Results and analysis
1.4
1.3 6.1. Reference paste and material substitution
1.2
Based on the results in [12] a reference paste mixture was
Relative spread

1.1
designed (Table 9) using a w/c ratio of 0.21. The reference mixture
1.0 resulted in a spread value of 315 mm and achieved an average
y = 4.266 x - 0.066
0.9
R² = 0.923 28 day compressive strength of 171 MPa (24.75 ksi).
0.8 Each other mixture was created by voluminously replacing only
0.7 the material in question, e.g. the type of cement for series I
(Table 10). Therefore, the weight ratio of C:SF:SM of 1:0.25:0.25
0.6
0.2 0.22 0.24 0.26 0.28 0.3 0.32 0.34 changes when the density of the replaced material is different. It
w/c ratio is worth noting that the replacement of the type of HRWR was
based on the same solid content to evaluate the performance effec-
Fig. 6. Relationship between relative spread and w/c ratio. tiveness of each HRWR.

400 450
375 400
350
Spread (mm) at w/c=0.25
Spread (mm) at w/c=0.25

350
325
300 300
275 250
250
200
225
200 150
175 100
150
50
125
100 0
SF1

SF2

SF3

SF4

SF5
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
C11
C12

Cement Silica Fume

(a) Paste mixes altered in type of cement (b) Paste mixes altered in type of
silica fume

450 450
400 400
Spread (mm) at w/c=0.25

350 350
Spread (mm) at w/c=0.25

300 300

250 250

200 200
150
150
100
100
50
50
0
0
SM13
SM1
SM2
SM3
SM4

SM10
SM11
SM12
SM5
SM6
SM7
SM8
SM9

(c) Paste mixes altered in type of (d) Paste mixes altered in type of
supplemental material HRWR

Fig. 7. Normalized spread value of altered paste mixtures at w/c = 0.25.

40 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43

210 210
200 200
190 190

Normalized f'c (MPa)

Normalized f'c (MPa)

180 180
170 170
160 160
150 150
140 140
130 130
120 120
110 110
100 100
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
C11
C12
SF1 SF2 SF3 SF4 SF5
Cement Silica Fume
(a) Paste mixes altered in type of cement (b) Paste mixes altered in type of
silica fume

210 210
200 200

Normalized f'c (MPa)

190
Normalized f'c (MPa)

190
180 180
170 170
160 160
150 150
140 140
130 130
120 120
110 110
100
100
SM10
SM11
SM12
SM13
SM1
SM2
SM3
SM4
SM5
SM6
SM7
SM8
SM9

Supplemental Material HRWR

(c) Paste mixes altered in type of (d) Paste mixes altered in type of
supplemental material HRWR

Fig. 8. Normalized compressive strength of altered paste mixtures at w/c = 0.25, air = 3%.

The performance (workability and compressive strength) of the ratio and air content on the compressive strength their combined
altered pastes for series I, II, III and IV are included in Tables 2–5, 1=3
influence (w=c  air ), as discussed in [11], is plotted for all speci-
respectively. It should be noted that the performance values pre- mens in Fig. 5b.
sented in Tables 2–5 are directly related to the w/c ratio used for Based on the results obtained in this research the relationship
the mixtures. To achieve acceptable workability and sufficient air can be described by the following expression for compressive
release the spread was targeted in the range of 280–340 mm. strength ranging between 112 MPa and 223 MPa (16–32 ksi):
Therefore, remixing with higher w/c ratio was necessary in the
0 1=3 0
case the paste did not turn over from the powdered to the fluid f c ¼ 151 ðw=c  air Þ þ 224; f c in MPa
state. This resulted in w/c ratios in the range of 0.21–0.33 and 0 1=3 0
ð2Þ
fc ¼ 21:7 ðw=c  air Þ þ 32:5; f c in ksi
indicated the difference in water demand. The following
chapter addresses the variability of w/c ratio and air content on Eq. (2) was obtained by line-of-best-fit in Fig. 5b, which describes
the performance values. the decrease in strength with increase of the product of
1=3
6.2. Eliminating variability in performance values w=c  air , where air is the percentage of air entrapped. By using
Eq. (2) the compressive strength values will be normalized to a
1=3
Although altering the reference mix provides a quick indication reference value of w=c  air based on the average w/c ratio and
of the suitability of the material constituent for UHPC design, a average air content of all specimens, set to w/c = 0.25 and air = 3%,
direct comparison of spread value and compressive strength is respectively. This allows for isolation of the effect of the type of
not advised because of the variation in w/c ratio and air content. material constituent on the compressive strength of the altered
w/c ratios of 0.21–0.33 were necessary to satisfy the target range paste.
of spread value (280–340 mm). Some mixes, especially those with Similarly, normalizing the spread values to a reference w/c ratio
fly ash as supplemental material, required a higher w/c ratio to allows for isolation of the effect of the type of material constituent
turn over into the fluid state during mixing. Once these mixes on the spread values of the altered paste. A relative spread graph
turned over they exhibited spread values at the end or even (Fig. 6), similar to Fig. 5a in Wille et al. [11], was generated to
exceeding the target spread value (see Table 4, SM2, 3, 4, 10). derive a relationship between relative spread and w/c ratio. Here,
Therefore, adjusting the w/c ratio to achieve the same spread value four pastes were mixed changing the w/c ratio to obtain the influ-
was not feasible for every mix. Since spread reflects yield strength ence of w/c ratio on spread value. Including data from [11], Fig. 6
and viscosity of the mix and air content is related to these rheologi- summarizes the relative spread versus w/c ratio of eight different
cal properties, variation in spread results in variation of air content. mixes in which only the w/c ratio was changed. The equation for
Fig. 5a shows the trend of decreased air content with increase in line-of-best-fit from Fig. 6 is given by Eq. (3), set to result in rela-
spread. In pursuit to minimize the influence of variation in w/c tive spread of 1.0 at w/c = 0.25.
 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43 41

1.20 1.20
1.15 1.15
1.10 1.10

Material Efficiency E
Material Efficiency E
1.05 1.05
1.00 1.00
0.95 0.95
0.90 0.90
0.85 0.85
0.80 0.80
0.75 0.75
0.70 0.70
C1
C2
C3
C4
C5
C6
C7
C8
C9
C10
C11
C12
SF1 SF2 SF3 SF4 SF5
Cement Silica Fume

(a) Paste mixes altered in type of cement (b) Paste mixes altered in type of
silica fume

1.20 1.20
1.15 1.15

Material Efficiency E
1.10 1.10
Material Efficiency E

1.05 1.05
1.00 1.00
0.95 0.95
0.90 0.90
0.85 0.85
0.80 0.80
0.75
0.75
0.70
0.70

Supplemental Material HRWR

(c) Paste mixes altered in type of (d) Paste mixes altered in type of
supplemental material HRWR

Fig. 9. Material efficiency for the constituents used in the altered pastes.

relative spread ¼ 4:266  w=c  0:066 ð3Þ components among one series by minimizing the effect of variable
w/c ratio or air content. The results in Fig. 8a confirm that pastes
with cements of high Blaine values and high in C3S + C2S (C9, C11)
6.3. Normalized spread value
perform best for strength development. Beside SF2 all silica fumes
similarly contributed to the strength of the paste (Fig. 8b).
Fig. 7 summarizes the normalized spread value (spreadN) at w/
Although fly-ash (SM4), GGBFS (SM8) and limestone (SM11)
c = 0.25 of all 4 series using Eq. (3). Larger spread values indicate
achieved the best normalized compressive strengths, all supple-
lower water demand and offer the potential for higher particle
mental materials selected in here are suitable for achieving ultra-
packing density. The results in Fig. 7a confirm that all selected
high strength (Fig. 8c). Despite the moderate performance of
white cements (C1, C9, C11), all oil well cements (C7, C10) and
HRWR3–8 regarding workability, the strength development of all
all Portland cements type II/V (C6, C8) performed suitably for
altered paste mixtures using HRWR1–8 is sufficient for UHPC-de-
UHPC design. The best workability performance showed the paste
sign (Fig. 8d). It is worth noting that despite the normalized
mix with white cement (C1) and the oil well cement (C7), both of
strength in excess of 150 MPa the paste with HRWR2 is not suitable
low Blaine value in comparison to other white (C9, C11) and oil
for UHPC design due to unsatisfactory workability at w/c = 0.25.
well (C10) cements. The results in Fig. 7b confirm that all selected
In addition to the normalized spread and compressive strengths,
silica fumes performed suitably for UHPC design. The best worka-
the cost of each paste is calculated based on the paste proportions
bility performance showed the paste with SF1 of low carbon con-
and the cost of each constituent. By using Eq. (2) the material effi-
tent and moderate fineness. The results in Fig. 7c confirm that
ciency E is calculated and shown in Fig. 9 for each constituent.
beside the pastes with metakaolin (SM6, SM7) all selected supple-
The results in Fig. 9 suggest the following materials as most
mental materials performed suitably. The best results were
suitable for efficient UHPC design: C6–C10, SF1, 3, 5, SM3, 4, 8
achieved with fly ash (SM2–SM5, SM10). Between the HRWR
and HRWR1, 4, 6. Out of these suitable material components C9,
selected for this research HRWR1 outperformed the other
SF1, SM4 and HRWR1 have been selected to form the ultra-high
HRWRs, indicating the importance of this constituent. Based on
performance paste for further investigation. This paste with a w/
the poor workability of the paste with HRWR2, this HRWR is not
c ratio of 0.22 resulted in a spread of 330 mm and reached a com-
suitable for UHPC design.
pressive strength of 189 MPa (27.4 ksi) after 28 days [24].
6.4. Normalized compressive strength
6.5. Cementitious matrix
Fig. 8 summarizes the normalized compressive strength of the
altered pastes at w/c = 0.25 and air content of 3% using Eq. (1). Following the modified Andreasen and Andersen (A&A) curve
This allows comparing directly the performance of material (Fig. 1) aggregates, optimized in their particle size distribution,
42 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43

Table 11
Recommended UHPC with fine aggregates.

Material UHPC-1 (northeast) (kg/m3) UHPC-2 (upper-midwest) (kg/m3) UHPC-3 (northwest) (kg/m3) UHPC-4 (U.S.) (kg/m3)
Cement (C9) 778 753 745 740
Silica fume (SF1) 194 188 186 185
Fly ash (SM4) 189 183 181 180
HRWR (1) 28 27 27 27
Fine aggregate 1166 (FB) 1129 (FL) 1118 (FVR) 1110 (Q)
A/C Ratio 1.5 1.5 1.5 1.5
w/c 0.23 0.24 0.23 0.23
Spread (mm) 290 265 286 315
f 0 c at 28d (MPa) 185 166 162 200
Cost (USD)/m3 494 472 496 652

Table 12
Recommended UHPC with fine and coarse aggregates.

Material UHPC-5 (northeast) (kg/m3) UHPC-6 (upper-midwest) (kg/m3) UHPC-7 (northwest) (kg/m3)
Cement (C9) 778 758 745
Silica fume (SF1) 194 190 186
Fly ash (SM4) 189 184 181
HRWR (1) 28 27 27
Aggregates 1167 (CB1) 1138 (CL) 1118 (CVR3)
A/C ratio 1.5 1.5 1.5
w/c 0.23 0.22 0.23
Spread (mm) 273 265 278
f 0 c at 28 day (MPa) 181 155 159
Cost (USD)/m3 494 475 463

Table 13
Mix design and of fiber reinforced UHPC using selected paste and alternative paste.

Material Name (alternative) Amount Percent by weight Cost ($US/m3) Percent by cost Name (alternative) Cost ($US/m3) Percent by cost
(kg/m3)
Cement C9 790.0 30.2 216 19.4 C6 95 9.7
Silica fume SF1 197.5 7.6 119 10.7 SF5 108 11.0
Supp. material SM4 191.5 7.3 10 0.9 SM3 9.9 1.0
HRWR HRWR1 28.4 1.1 136 12.2 HRWR1 136 13.8
Aggregates FB 1140.5 43.7 17 1.5 FB 17 1.7
Fibers F1 117.8 4.5 618 55.4 F1 618 62.9
Water 146.0 5.6 – – – – –
Total 1116 100 983 100

were added to the paste leading to the development of the ultra- By using Portland cement type II/V (C6 or C8) and less expen-
high performance concrete matrix. The investigation of the aggre- sive silica fume (SF5) the costs for UHPC matrix with acceptable
gate to cement weight ratio on spread and compressive strength performance can be decreased to about U$340 per m3.
suggested a ratio of A/C = 1.5 [FHWA report]. Evaluating the spread
value for each mix proved to be a quick and helpful indicator to
select the A/C ratio.Two types of UHPC matrices were defined as 6.6. Fiber reinforced composite
follows: (1) UHPC with fine aggregates up to a maximum particle
size of 1.2 mm (0.047 inch) and (2) UHPC with fine and coarse UHPC matrix with fiber reinforcement F1 (Table 8) outper-
aggregates up to a maximum particle size of 9.5 mm (0.37 inch). formed the other types of fibers in composite tensile strength
In total 11 matrices were tested using different fine and coarse and workability due to the fiber’s high tensile strength and a fiber
aggregates (Tables 5 and 6). Based on paste development and opti- factor vf ¼ V f  lf =df ¼ 0:98 < 1. This confirms the results in [25],
mized aggregate size distribution all matrices were characterized where the addition of fibers with vf  1 has only minimal effect on
by excellent workability (spread = 265–315 mm) using a w/c ratio workability.
in the range of 0.21–0.24 and compressive strength in excess of UHPC-4 and UHPC-5 (Tables 11 and 12) with and without fibers
150 MPa. Matrices with coarse aggregates demanded less water were further investigated in their freeze–thaw resistance. All four
to achieve comparable spread values in comparison to matrices specimens showed after more than 100 freeze–thaw cycles no
with only fine aggregates. This is attributed to the smaller total indication of damage, visually evaluated and measured by reso-
surface area of the coarse aggregates. Despite the slightly increased nance frequency.
w/c ratio all fine UHPC matrices achieve higher compressive Table 13 shows the cost influence of the material constituents.
strengths than the coarse UHPC matrices with the same type of While supplemental materials and aggregates insignificantly influ-
aggregate. The highest compressive strength values were achieved ence the total costs, the costs of fiber reinforcement represent
by matrices with quartz (Q), followed by basalt (B), volcanic rock about half the cost of the UHPC fiber composite. More research
(VR) and limestone (L) [24]. UHPC mix designs based on local avail- effort is needed to find an alternative cost efficient solution to pro-
able materials, referring to a U.S. region, are recommended in vide sustained tensile strength and enhanced ductility. This can be
Tables 11 and 12. The material costs range between $US463 and achieved by finding an alternative fiber reinforcement of lower
$US652 per m3. cost and by reducing the required amount of fiber reinforcement
 K. Wille, C. Boisvert-Cotulio / Construction and Building Materials 86 (2015) 33–43 43

through improved material utilization. A more effective fiber mate- References

rial utilization could be obtained by tailored matrix fiber bond and
by combining continuous reinforcement with discontinuous fiber [1] American Society of Civil Engineers. Report Card for America’s Infrastructure.
2013. <www.infrastructurereportcard.org>.
reinforcement. [2] Fehling E, Schmidt M, Geisenhanslueke C, editors. International symposium on
ultra high performance concrete. Kassel, Germany; 2004.
[3] Fehling E, Schmidt M, Geisenhanslueke C, editors. Second international
7. Conclusions
symposium on ultra high performance concrete. Kassel, Germany; 2008.
[4] Schmidt M, Fehling E, Glotzbach C, Froehlich S, Piotrowski S, editors.
This research emphasized the material efficiency of different Hipermat-third international symposium on ultra high performance concrete
constituents in the UHPC design. Based on a progressive UHPC and nanotechnology for high performance construction materials. Kassel,
Germany; March 7–9, 2012.
development, ultra high strength paste, matrix and fiber reinforced [5] Toulemonde F, Resplendino, J., editors. International workshop on ultra high
composite were analyzed and optimized in their performance. An performance fibre reinforced concrete – designing and building with UHPFRC:
efficiency factor E was defined to consider the performance in state of the art development; 17–19 November 2009, Marseille, France,
AFGC/fib, 814 p.
workability, compressive strength and costs in order to facilitate [6] Toulemonde F, Resplendino J, editors. Proceedings of the RILEM-fib-AFGC
the selection of most effective materials. international symposium on ultra-high performance fibre-reinforced concrete
– UHPFRC 2013 designing and building with UHPFRC: from innovation to
large-scale realizations. 1–3 October 2013, Marseille, France, 832 p.
 Four UHPC matrices with fine aggregates only and three UHPC [7] Rossi P, Ultra-high performance fibre reinforced concrete (UHPFRC): an
matrices including coarse aggregates were recommended using overview. In: Proceedings of fifth RILEM symposium in fibre-reinforced
local available materials from three different regions, covering concretes (FRC)-BEFIB’ 2000. p. 98–100.
[8] Rossi P, Ultra high performance concretes – a summary of the current
northeast, upper-midwest and northwest. Their material costs
knowledge’’. Concrete international; February 2008. p. 31–4.
without fiber reinforcement ranged between about $US470 [9] Naaman AE, Wille K. The path to ultra-high performance fiber reinforced
and $US650 per m3 and $US465–$US500 per m3 for fine and concrete (UHP-FRC): five decades of progress’’. In: Schmidt M, Fehling E,
editors. Proceedings of hipermat 2012, 3rd international symposium on UHPC
coarse UHPC, respectively. Excellent workability facilitates the
and nanotechnology for high performance construction materials, Kassel,
use of these UHPCs in structural application. The compressive Germany; March 7–9, 2012. p. 3–16.
strength of the recommended UHPC matrix mixes ranges from [10] Wille K, Naaman AE, Parra-Montesinos GJ. Ultra high performance concrete
155 MPa (22.5 ksi) to 200 MPa (29 ksi). with compressive strength exceeding 150 MPa (22 ksi): a simpler way,
‘‘Department of Civil and Environmental Engineering – Report, University of
 Additionally, future research efforts are suggested to tailor the Michigan; 2009. 47 p.
weight ratio of cement:silica fume:supplemental material of [11] Wille K, Naaman AE, Parra-Montesinos G. Ultra high performance concrete
1:0.25:0.25 in terms of performance versus cost ratio. A reduc- with compressive strength exceeding 150 MPa (22 ksi): a simpler way. ACI
Struct Mater J 2011;108(1):46–54.
tion in the amount of the most expensive material and an [12] Wille K, Naaman AE, El-Tawil S, Parra-Montesinos G. Ultra-high performance
increase in the amount of the least expensive material might concrete and fiber reinforced concrete: achieving strength and ductility
lead to a further improvement in performance versus cost. without heat treatment. RILEM Mater Struct 2012;45(3):309–24.
[13] Wille K, Naaman AE, El-Tawil S. Ultra high performance fiber reinforced
This optimization was not the scope of this research project concrete (UHP-FRC) record performance under tensile loading. Concr Int
and has been left for future research efforts. 2011;33(9):35–41.
 Adding fiber reinforcement of 1.5% by volume to the UHPC [14] Wille K, Kim D, Naaman AE. Strain-hardening UHP-FRC with low fiber
contents. Mater Struct 2011;44(3):583–98.
matrix increases the costs per cubic meter by about $US620.
[15] Graybeal BA, Tanesi J. Durability of an ultrahigh-performance concrete. ASCE J
This led to the recommendations of cost-efficient fiber rein- Mater Civil Eng 2007:848–54.
forced UHPC of about $US1120 per m3. More research effort is [16] Sakai E, Akinori N, Daimon M, Aizawa K, Kato H. Influence of superplasticizer
on the fluidity of cements with different amount of aluminate phase. In:
needed to find an alternative cost efficient solution to provide
Fehling E, Schmidt M, Geisenhanslueke C, editors. Second International
sustained tensile strength and enhanced ductility due to the Symposium on Ultra High Performance Concrete. Kassel, Germany: 2008. p.
high costs of fiber reinforcement. This can be achieved by find- 85–92.
ing an alternative fiber reinforcement of lower cost and by [17] Brouwers HJH, Radix HJ, Self-compacting concrete: the role of the particle size
distribution. In: The first international symposium on design, performance and
reducing the required amount of fiber reinforcement through use of self-consolidating concrete (SCC’2005) Changsha, Hunan, China; p. 109–
improved material utilization. A more effective fiber material 18. 2005.
utilization could be obtained by tailored matrix fiber bond [18] Park JJ, Kang ST, Koh KT, Kim SW. Influence of the ingredients on the
compressive strength of UHPC as a fundamental study to optimize the mixing
and by combining continuous reinforcement with discontinu- proportion. In: Fehling E, Schmidt M, Geisenhanslueke C, editors. Second
ous fiber reinforcement. International Symposium on Ultra High Performance Concrete. Kassel,
Germany; 2008. p. 105–12.
[19] Wille K, Naaman AE. Pullout behavior of high-strength steel fibers embedded
in ultra-high-performance concrete. ACI Mater J 2012;109(4):479–88.
Acknowledgements [20] Wille K, Naaman AE. Effect of ultra-high-performance concrete on pullout
behavior of high-strength brass-coated straight steel fibers. ACI Mater J
2013;110(4):451–62.
This research has been supported by the Federal Highway
[21] Wille K, El-Tawil S, Naaman AE. Properties of strain hardening ultra high
Administration (FHWA) and the University of Connecticut. The performance fiber reinforced concrete (UHP-FRC) under direct tensile loading.
writers would also like to acknowledge the support from the fol- Cement Concr Compos 2014;48:53–66.
[22] Tue NV, Ma J, Orgass M. Influence of addition method of superplasticizer on
lowing companies: Lehigh White Cement Company, Holcim U.S.
the properties of fresh UHPC. In: Fehling E, Schmidt M, Geisenhanslueke C,
Inc., CEMEX, Lafarge North America, Elkem Materials, Norchem, editors. Second International Symposium on Ultra High Performance Concrete.
Headwaters, Burgess Pigment Company, Advanced Cement Kassel, Germany: 2008, p. 93–100.
Technologies, Omya Inc., Specialty Minerals Inc., Chryso Inc., [23] Graybeal BA. Material property characterization of ultra-high performance
concrete. Report FHWA-HRT-06-103. August 2006.
Euclid Chemical Company, Grace, Sika, BASF, Tilcon Connecticut [24] Wille K, Boisvert-Cotulio C. Development of non-proprietary ultra-high
Inc., Stoneco, Calportland, Nycon. performance concrete for use in the highway bridge sector.
The publication of this document does not necessarily indicate FHWA Publication No. FHWA-HRT-13-100, NTIS: PB2013-110587;
October 2013. 66 p.
approval or endorsement of the findings, opinions, conclusions, or [25] Naaman AE, Wille K, ‘‘Some correlation between high packing density, ultra-
recommendations either inferred or specifically expressed herein high performance, flow ability, and fiber reinforcement of a concrete matrix’’
by the Federal Highway Administration or the United States for BAC2010 – 2nd Iberian congress on self compacting concrete, University of
Minho – Guimaraes, Portugal, July 1–2, 2010.
Government.