Construction and Building Materials 135 (2017) 225232

Contents lists available at ScienceDirect

Construction and Building Materials

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

Characterization of fresh dry-mix shotcrete and correlation to rebound

Julie Armengaud a, Graldine Casaux-Ginestet a,, Martin Cyr a, Bernard Husson a, Marc Jolin b
a
LMDC, Universit de Toulouse, INSAT, UPS, France
b
Centre de recherche sur les infrastructures en bton (CRIB), Universit Laval, 1065 Avenue de la Mdecine, Qubec, QC G1V 0A6, Canada

h i g h l i g h t s

The relation between fresh properties and rebound of shotcrete is studied.

Verification and consolidation of some results are added to the literature.

The correlations between rebound and penetration stresses are verified.

This leads to a discussion on how to study shotcrete, whether as a fluid or a solid.

New insight is brought to understand the relation rebound vs mixture design.

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

Article history: Dry-mix shotcrete has all the components of concrete but its particular placement technique generates
Received 25 March 2016 losses due to rebound. These losses induce a cost increase and a difference between the initial and in-
Received in revised form 22 December 2016 place composition of the concrete. Many parameters influence rebound, but this study focuses on mixture
Accepted 30 December 2016
design via supplementary cementitious materials (SCMs). The aim of the paper is to study the relation
Available online 10 January 2017
between fresh properties and rebound of shotcrete, and evaluate how SCMs act on it. This will eventually
help to understand what mixture properties are needed to limit rebound and help with the design of mix-
Keywords:
tures. In this study, 5 mixtures in which cement was partially replaced by metakaolin, ground granulated
Dry-mix shotcrete
Rebound
blast furnace slag or silica fume were shot at different consistencies, in a full scale laboratory facility.
Consistency Fresh shotcrete was evaluated by penetration tests (static and dynamic). It appeared that the SCMs were
Dynamic contact stress efficient in reducing rebound and their efficiency depended on consistency. Water is one of the main
Static penetration stress parameters in rebound but it is difficult to control because it is adjusted by the operator. Mixtures with
Supplementary cementitious materials metakaolin and silica fume seem to lower this dependence. Static and dynamic penetration stresses seem
(SCMs) to be correlated with rebound and, more precisely, the relation between dynamic measurement and
rebound seems independent of the mixture tested. Understanding the levels of static and dynamic pen-
etration stress could make mixture design easier.
2017 Elsevier Ltd. All rights reserved.

1. Introduction into the machine and conveyed pneumatically through a hose to

This paper studies rebound in dry-mix shotcrete. Shotcrete is a
mortar or concrete, pneumatically projected onto a surface at high
velocity and rebound is defined as the proportion of shotcrete
material that ricochets off the receiving surface [1]. As shotcrete
can be used in wet or dry processes, a clear distinction must be
made between those two methods. Wet-mix shotcrete is batched
with water before being introduced into the delivery hose for
pneumatic placement, while dry-mix shotcrete is a process in
which dry or slightly moist constituents of concrete are introduced
Corresponding author.
E-mail address: geraldine.casaux-ginestet@insa-toulouse.fr
Ginestet).
the nozzle where the water is added. The main difference lies in
the location where water is introduced and therefore in the time
of contact with water before the mixture reaches the receiving sur-
face. In both cases, it is the high velocity imparted to the flow that
induces consolidation of the material on the surface.
The dry-mix process is used in various civil engineering and
construction projects, from tunneling or anchored retaining walls
to repairs or building construction. Unfortunately, the dry-mix
shotcrete process can lead to high losses of concrete due to
rebound [25]. Such losses induce overconsumption of material,
which is detrimental for the cost of the work, the environment,
and to some extent, to the material itself as rebound tends to cre-
(G. Casaux-
ate an in-place mixture that is richer in terms of cement content
and thus undergoes increased shrinkage [2,6]. Rebound in the

http://dx.doi.org/10.1016/j.conbuildmat.2016.12.220
0950-0618/ 2017 Elsevier Ltd. All rights reserved.
226 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232
dry-mix shotcrete process is dependent on many technical param-
eters such as air flow, orientation of the hose and temperature [6
8], and also on mixture properties such as binder content, water,
amount and size of coarse aggregates, and type of admixture
[3,811].
Some mix design solutions exist to reduce rebound, such as par-
tial replacement of cement by silica fume, carbon black or high
reactivity metakaolin [3,12,13] but few explanations have been
given regarding the role of SCMs in shotcrete. Pfeuffer and Kusterle
[3] evaluated rebound and rheological properties, but the material
tested for rheological evaluation was batched with a fixed amount
of water and a liquid admixture, without being shot. This changed
the overall structure of the material and probably altered interpre-
tations. In order to evaluate fresh shotcrete properties and have
direct information on the mixture shot, static and dynamic meth-
ods of measurement have been developed [14,15]. A static method
(based on needle penetration in fresh shotcrete) is often used and
related to rebound but dynamic methods (based on impact mea-
surement) have rarely been exploited. Moreover, the measure-
ments obtained have not been linked to standardized physical or
rheological properties. In order to discuss this connection, static,
dynamic and rebound measurements were made on mixtures con-
taining supplementary cementitious material. The aim of the paper
is to study the relation between fresh properties, measured by sta-
tic and dynamic tests, and rebound of shotcrete made with supple-
mentary cementitious material. The tests performed in this project
are intended to improve the knowledge on this very particular type
of concrete, which differs from traditional concrete in terms of
technology and constituents (finer aggregates and accelerating
agent), but is still a cement-based material with a close behavior
to that of ordinary concrete in certain extreme situations, like in
fire and at high temperature [16].
2. Experimental program
2.1. Method
The test program was carried out at the Laboratory of Materials
and Durability of Constructions (LMDC) in Toulouse (France), with
an original experimental setup designed especially for the study of
dry shotcrete [17]. The shooting took place in an intermodal con-
tainer divided into two parts (Fig. 1). The right part was used for
water adjustment (done by the nozzleman), and the left part for
Fig. 1. Container layout: right part for adjustment, left part for rebound test. The tarp for rebound collection is on the left part of the container.
rebound measurement. Once the setting of water was achieved,
the hose was oriented toward the left part to start the evaluation
of rebound. All mixes were shot with a rotating-barrel type
machine (Meyco Piccola) often used in real construction sites.
The hose used had an internal diameter of 50 mm and the water-
ring was at a fixed distance of 2.5 m from the nozzle. Air flow
was set at 10 m3/min with a pressure of 6.5 bar. In order to study
the influence of only the mixture on rebound, the operational setup
was kept unchanged during the shooting.
2.1.1. Rebound measurement
The rebound measurement was carried out by spraying the
shotcrete on to a 500  500 mm rebound mold, as shown on
Fig. 1, placed near the bottom of a vertical wall of the container
and at a distance of 1.2 m from the nozzle (this distance is opti-
mized experimentally and is dependent of the air flow). The
rebound mold had an opening on the side to allow material to be
evacuated and avoid creation of an aggregate pocket that might
distort the rebound results. The amount of material that did not
stick to the mold was collected in a tarpaulin. At the end of the
shooting, the mold and the tarpaulin with the lost material were
weighed. Rebound was then calculated as:
Mass of material in the tarpkg
Rebound %
Mass of material in the tarp kg Mass in the mold kg
 100
2.1.2. Fresh properties measurement
In addition to the rebound value, dynamic and static consisten-
cies were measured. The static penetration strength (P) was evalu-
ated with a static penetrometer, also called a Proctor needle. This
measurement has been used in many studies with different shapes
of indenters [5,15,18,19]. In our case, the needle was a 6 mm diam-
eter flat indenter as shown in Fig. 2a. The needle was pushed into
the substrate 5 min after spraying, and the strength p was indi-
cated by a spring system. The penetration strength (p) was defined
as being proportional to a yield stress by Johnson [20].
The other fresh parameter evaluated was the stress arising
under dynamic conditions (pd). This value first appeared in the
rebound theory presented by Armelin [21], and was calculated as
the energy of an impacting ball divided by the volume of the
imprint after impact [5,15]. These authors calculated pd as:
 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232 227

Imprint aer impact

(A) (B) (C)

Fig. 2. (a) Proctor needle used for static consistency (P) evaluation (b) 2D method of PD measurement (c) Gun and radar used for dynamic consistency evaluation (PD).

1
mv
pd 2
V mechanically densified. The chemical and physical properties of
where m and v are the mass and velocity of the ball, and V is the vol-
ume of the imprint of the ball on the shotcrete substrate (Fig. 2b). In
the present tests, pd was evaluated with a 16 mm glass ball pro-
jected by a pneumatic launcher at different speeds. Measurements
of pd were done 7 min after spaying. The pneumatic launcher was
a paintball gun (Fig. 2c) and the speed of the ball was measured
by radar. Though evaluating experimentally the dynamic contact
stress is fairly easy, the nature of the relevant phenomena is still
unclear.
In the hardened state, strength was evaluated on
50  50  50 mm cubic samples sawn from the panels after a cure
of 28 days at 20 C and humidity >95%.
2.2. Materials
In this study, concrete was formulated with sand, aggregate,
CEM I 52.5R cement [EN 197-1] and supplementary cementitious
materials (SCMs) with various levels of replacement. The particle
size distribution of 0/4 sand and 4/10 aggregate can be found in
Fig. 3. Sand and aggregate were siliceous alluvial rounded materi-
als with coefficients of water absorption of 1.12% for sand and
1.08% for aggregate.
Three supplementary cementitious materials (SCMs) were
used:
- Ground granulated blast furnace slag (GGBS)
- Flash metakaolin (MK)
- Densified silica fume (SF)
complying with European standard [NF EN 13263-1,2, 2009],
these supplementary cementitious materials (SCMs) are given in
Table 1.
2.3. Mixtures
2.3.1. Dry components
In order to evaluate the influence of supplementary cementi-
tious materials (SCMs) on rebound, five different mixtures were
tested. The proportions of binder and aggregate were maintained
constant. The reference mixture used (named REF) was made with
cement only. The mixture with silica fume (named SF 10%) was
made with 10% replacement of cement. Mixtures with flash meta-
kaolin were tested with 20% (MK 20%) and 30% (MK 30%) replace-
ment of cement, and the mixture with GGBS was made with 50%
replacement of cement (named GGBS 50%). The mixture designs
are presented in Table 2, in kg per m3 of dry matter. The W/B is
not reported as it is not a parameter that is controlled before the
shooting.
2.3.2. Water
As the quality of dry shotcrete is very dependent on the nozzel-
mans water adjustment, each mixture was shot at least 5 times.
Each time, a new water adjustment was made so as to cover the
whole range of consistencies that can be achieved in the field.
Table 1
Chemical and physical properties of cement and SCMs.

Ground granulated blast furnace slag was an industrial by- Cement CEM I GGBS (%) Flash Densified
52.5R (%) metakaolin silica fume (%)
product complying with European standard [NF EN 15167-1,
(%)
2006]. The metakaolin used was obtained by an industrial flash cal-
Specific 4136 cm2/g 3900 cm2/g 12.5 m2/g 22 m2/g (BET)
cination process [22] and consisted in approximately 50% of amor-
surface area (Blaine) (Blaine) (BET)
phous particles of alumina silicate (complying with French
Specific 3.13 g/cm3 2.90 g/cm3 2.54 g/cm3 2.24 g/cm3
standard [NF P18-513]). Silica fume was an industrial by-product
gravity
Shape Angular Angular Flaky Spherical
100 structure

90 Aggregate 4/10 rounded Oxides % % % %

80 Sand 0/4 rounded SiO2 20.3 36.1 67.1 96.5
Percentage passing (%)

70 CaO 65.4 44.0 1.1 0.80

Al2O3 4.5 10.7 26.8 0.21
60 Fe2O3 2.5 0.41 2.6 0.16
50 MgO 1.0 7.3 0.1 0.29
40 K2O 0.14 0.43 0.1 0.52
Na2O 0.13 0.25 0.01 0.17
30 SO3 3.4 0.2 0.0 0.24
20 Loss on 2.1 <0.1 0.8 2.7
10 ignition

0 Mineralogical composition of cement

0.01 0.1 1 10 100 C3S 66.7
Sieve (mm) C2S 13.1
C3A 8.7
Fig. 3. Particle size distribution of sand and aggregate. C4AF 8.2
228 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232

Table 2
Mixture design.

Name of the mix REF MK 30% MK 20% GGBS 50% SF 10%

3
Constituent (kg/m )
Aggregate 343 343 343 343 343
Sand 1555 1555 1555 1555 1555
Cement 400 280 320 200 360
MK 120 80
GGBS 200
SF 40

In order to classify the resulting concrete, three categories of
consistencies were used: wet, normal and dry. There is no official
scale or test to evaluate shotcrete consistencies as there are for
plastic concrete (Abrams cone) or soil (Atterberg limits) but the
combined use of the Proctor needle and the experience of the noz-
zleman enabled a good assessment of the achieved consistency.
However, the adjustment of water during the shooting process
was based purely on the nozzlemans appreciation.
The wettest consistency, or the wettest stable consistency, as
it is called by Parker et al. [8], relates to a limit of liquidity by anal-
ogy with soil behavior. This consistency corresponded to the high-
est water content that may be added before concrete slump
occurred on the wall. The driest consistency was achieved by
reducing the amount of water until the operator judged that the
concrete had become too stiff and the rebound unreasonable. This
corresponded to the driest sticky consistency achievable with the
concrete. Normal consistency lies in between. A visual evaluation
through the degree of shine of the concrete is often used as an indi-
cator to adjust the water in order to obtain a dryer (dull aspect) or
wetter material (glossy aspect).
3. Results
Once the shooting was done, the consistency was evaluated
visually and static and dynamic measurements were performed.
3.1. Influence of consistency on rebound
The first phase of the study was based on the observations of
the operator. After shooting, the consistency of each sample was
classified as dry, normal or wet. When several panels were classi-
fied in the same category, an average value was determined. Fig. 4
shows the rebound results for each mixture, classified by their
consistencies.
The general tendency was that the rebound increased when the
mixture consistencies became drier. This was particularly obvious
with the reference shotcretes. The figure also highlights the role of
45%
40%
35%
30%
Rebound (%)
SCMs in rebound reduction: for each consistency, the REF mixture
had higher rebound values than mixtures with SCMs. The decrease
of rebound could reach 16% in the best cases, mainly for the dry
consistency.
Some mixtures seemed more efficient than others, but their
behavior depended on the consistency: GGBS 50% gave good
results for wet consistency while MK 20% seemed more efficient
for normal and dry consistencies. A higher metakaolin content
did not increase mixture performance, as can be noted by compar-
ing the MK 30% results with those for MK 20%. A hypothesis to
explain this phenomenon is that, beyond a given degree of replace-
ment, agglomerates of metakaolin form. Those agglomerates
would be less dispersed in the stream of material and would
reduce the surface available to capture water.
Another point is that the rebound of mixtures with metakaolin
and silica fume did not vary much among consistencies (difference
of approximately 6% of rebound between dry and wet) compared
to the REF mix or GGBS 50% (difference of 13% between dry and
wet). This means that MK and SF mixtures were less affected by
changes in water proportions. By extension, those formulations
would be less sensitive to the nozzlemans degree of experience
i.e., they can be considered more robust.
In the case of silica fume and metakaolin, SCMs reducing the
rebound may be explained by their action on water consumption.
In the case of cast concrete, previous studies [2225] have pointed
out that those additions tend to increase the water consumption at
constant workability. For shotcrete, this means that, compared to
REF, MK and SF allow more water to be added before reaching
the wettest stable consistency. Pfeuffer and Kusterle [3] explained
that this higher water content formed a plastic suspension with the
binder, favoring adhesion of the impacting flow of material and
thus lowering rebound.
Ground granulated blast furnace slag generally increases work-
ability slightly for a constant W/B ratio and reduces water con-
sumption for the same workability (on slump test) in the case of
cast concrete [26,27]. For shotcrete, when REF and GGBS 50% were
compared, the latter was seen to need less water than REF did to
obtain the wettest stable consistency. However, GGBS 50% showed
the lowest rebound, which tends to contradict the previous obser-
vation on metakaolin and silica fume. The role of water interaction
with SCMs under impact might be dependent on other properties
of SCMs. This makes the choice of SCMs difficult, when the aim is
to reduce losses by rebound, because contradictory properties in
terms of water consumption lead to rebound reduction. It is then

25%
REF necessary to find another characteristic of the mixtures to make
MK 30% this choice easier.
20%
MK 20%
15% GGBS 50%
SF 10%
3.2. Static penetration stress (p)
10%

5% Consistency is a very simple way to classify and compare sam-

0%
wet normal dry
ples but it does not allow very reliable judgements to be made
Consistency
because it is quite a subjective scale. To push the interpretation
further, static penetration stress was used. Fig. 5a presents the
Fig. 4. Rebound according to mixture consistency. penetration stress according to consistency and shows that the dis-
 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232 229

REF MK 30% MK 20% GGBS 50% SF 10% REF MK 30%

Penetration stress p (MPa)

1.2 MK 20% GGBS 50%

1.0 40%

0.8 30%

Rebound
0.6
20%
0.4

0.2 10%

0.0 0%
wet normal dry 0.0 0.5 1.0
Consistency p (MPa)
(A)
(B0)

MK 30% MK 20%
40%
40%

30% 30%
Rebound

Rebound
20% 20%

10% 10%

0% 0%
0.0 0.5 1.0 0.0 0.5 1.0
p (MPa) p (MPa)
(B1) (B2)

GGBS 50% SF 10%

35% 35%
30% 30%
25% 25%
Rebound
Rebound

20% 20%
15% 15%
10% 10%
5% 5%
0% 0%
0.0 0.5 1.0 0.0 0.5 1.0
p (MPa) p (MPa)
(B3) (B4)
Fig. 5. (a) Penetration stress according to consistency. (b) Rebound as a function of penetration stress for different mixtures. (b0) Results for all mixtures, dotted line is the
trend line for the REF mixture. (b1) Results for MK 30%, (b2) Results for MK 20%, (b3) Results for GGBS 50%, (b4) Results for SF 10%.

persion can sometimes be high because the tests were carried out
on a large scale with industrial equipment, so small variations of
the setup might have led to higher variation on the results. How-
ever it was still possible to identify global trends.
On Fig. 5a, it can be seen that, for a given mixture, the highest p
value corresponds to the driest consistency and the lowest p value
corresponds to the wettest stable consistency. Another observable
result is that maximum and minimum p values differ depending on
the mixture studied. When comparing REF and GGBS 50% for
instance, it can be seen that the highest p value of GGBS 50%, cor-
responding to its driest achievable consistency, is lower than the
highest P value of REF. This means that, for the same extreme con-
sistency, p will not be the same.
Fig. 5b presents the rebound as a function of static penetration
stress. The general tendency is an increase in rebound with higher
p values, which correspond to the driest consistencies. At the same
time, lower rebound occurs for lower values of p and, thus, for wet-
ter consistencies. Generally speaking, all mixtures with SCMs show
lower penetration stress. Fig. 5(b0) shows the rebound versus pen-
etration stress (p) for all mixtures. The degree of correlation does
not seem to be the same for all mixtures and, for the sake of clarity,
each mixture is plotted independently on Fig. 5(b1), (b2), (b3) and
(b4), with the dotted trend line of the REF mixture indicated on
each figure.
When looking at Fig. 5(b2), (b3) and (b4) for example, it is inter-
esting to note that the relation between rebound and p depends on
the binder used. This observation corroborates results obtained by
Armelin [21] and Jolin [14]. On Fig. 5(b1) and (b2), it can be seen
that the evolution of rebound as a function of penetration seems
different, even if the SCM is the same. So, the relation is also depen-
dent on the degree of replacement of cement content. To better
understand rebound, it is necessary to find a parameter that is
independent of the mixture used.
3.3. Dynamic contact stress (Pd)
The dynamic contact stress represents the work done by the
ball to penetrate the mortar phase. As shotcreting is a dynamic
event, it seems relevant to evaluate the property of concrete by a
dynamic measure. The results of rebound as a function of dynamic
230 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232

contact stress (Pd) are plotted in Fig. 6, along with those of Jolin [5]. 2.5
It can be seen that rebound and dynamic contact stress are corre- REF

lated in a logarithmic way, and that rebound increases for higher 2.0 MK 30%

values of pd. MK 20%

The solid line on Fig. 6 is the best fit curve for all the mixtures GGBS 50%

Pd (MPa)
1.5
tested in this campaign. The dash-dotted line corresponds to the SF 10%

trend obtained by Jolin [5]. Dynamic contact stress measurements

performed by Jolin [5] were made on two different mixtures shot at
multiple consistencies, with a system that dropped a ball on the
shotcrete substrate from a given height.
Even with slightly different measurement methods, and with
different mixtures, the two curves are close to one another, indicat-
ing that the dynamic contact stress could be considered as a mean-
ingful parameter. This would indicate that there is a relation
between rebound and dynamic contact stress (Pd,), almost inde-
pendent of the mixture tested, and that dynamic contact stress
would be sufficient to characterize a mixture in terms of rebound.
In that case, designing a mixture with low resulting pd would lead
to low rebound.
Dynamic contact stress seems to be a good tool to evaluate the
rebound of a mixture. Understanding how SCMs influence the pd
value would be a starting point to make low rebound mixture
design easier. As the consistency is correlated to the rebound,
and in order to evaluate the influence of SCMs on pd, the relation-
ship between consistency and pd was assessed. Fig. 7 shows the
dynamic contact stress pd for the different mixtures and consisten-
cies. For every mixture, a high value of pd corresponds to a dry con-
sistency, and low dynamic contact stress is obtained for the
wettest consistency.
For a given consistency, mixtures with partial replacement of
cement show lower pd than REF. This difference increases as the
consistencies become drier. Mixtures with MK 20% and SF 10%
have the lowest dynamic contact stress. The capacity of these addi-
tions to mobilize water, as pointed out in Section 3.1, might play a
role in the low dynamic contact stress measured after shooting. As
rebound is a very fast event, capturing water in the very first
moment might be one of the particularities needed for SCMs to
achieve low dynamic contact stress.
In terms of rebound, the best formulations will be those that
allow small variations of dynamic contact stress with the consis-
tency, and low pd, like MK 20% and SF 10%. This implies that the
influence of the nozzelmans degree of experience would be more
limited.
As for the rebound in Fig. 4, it can be noted that increasing the
replacement level of cement does not necessarily diminish
dynamic contact stress, as the case of MK 30% shows. An optimum
degree of substitution for each addition, that will maximize its
action, might exist but further studies need to be run to test this
hypothesis.
1.0
0.5
0.0
wet normal dry
Consistency
Fig. 7. Dynamic contact stress, pd, for different mixtures and consistencies.
3.4. Compressive strength
The results presented above concern only the fresh properties of
mixtures but one of the first expectations for shotcrete is that it
should reach a target performance in strength. Fig. 8 shows the
compressive strengths of mixtures at different consistencies. REF
gave the highest values for all consistencies, while SF 10% was
the most efficient SCM. MK 20% and MK 30% had lower strength
than GGBS 50% and, taking standard deviation into account, there
was not much difference between the two rates of replacement.
For all mixtures, strength stayed relatively high (above 45 MPa).
For cast concrete, adding too much water leads to a decrease in
strength and, in the case of shotcrete, the wettest consistencies
generally exhibit the lowest values. Following the same reasoning,
the driest consistency should lead to the highest strength. How-
ever, for almost every mixture, it was the normal consistency that
gave the best strength. This was due to the particular placement
method for shotcrete, which is compacted by the impacting flow.
A material that is too stiff (because of lack of water) will not be
compacted enough and will present an excess of porosity and thus
a diminution of strength. This evolution of strength is very clear in
the case of MK 20%. To reach the best strength, it is necessary to
target a normal consistency.
4. Comments on the rebound phenomenon and fresh properties
of shotcrete
The results presented show that rebound is linked to the fresh
properties of shotcrete (p, pd, and consistency) but those fresh
properties are empirical and not very well defined. Some hypothe-
ses can be formulated on the model and the physical properties
Compressive strength 28days cube (MPa)

40% 75
y = 0.14ln(x) + 0.27 wet
y = 0,11ln(x) + 0,28 70 normal
35%
R = 0,76
Jolin, 1999
65 dry
REF
30%
Rebound

MK 30% 60
MK 20%
25% 55
GGBS 50%
SF 10% 50
20%
Trend line
45
15%
0 0.5 1 1.5 2 2.5
40
Dynamic contact Stress (MPa) REF MK 30% MK 20% GGBS 50% SF 10%

Fig. 6. Rebound as a function of the dynamic contact stress. Dotted line corresponds Fig. 8. Compressive strength on cube after 28 days of >95% humidity cure for
to Jolins (1999) best fit curve: y = 0.14 ln(x) + 0.27. different mixtures and consistencies.
 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232
needed to reach the best p, pd and consistency and, thus, the lowest
rebound. In particular, an understanding of the mechanism behind
pd is important as this parameter seems to characterize rebound
independently of the mixture tested.
4.1. Rheological approach
The usual way to characterize fresh concrete is to use a rheolog-
ical model like the Bingham model [26]. Armelin and Banthia [28]
highlighted the fact that this model fails to recognize the elastic
component of fresh concrete that causes rebound of particles.
Moreover, the model implies the determination of a viscosity and
a yield value. Pfeuffer and Kusterle [3] made a rheological study
on mixed (non-shot) mortar phase. This solution was used
because measuring viscosity on shotcrete is made impossible by
the fact that it is a relatively stiff material (this kind of concrete
is compacted without prior mixing with water) compared to cast
concrete and thus cannot be evaluated with a classical rotational
rheometer. Nonetheless, Pfeuffer and Kusterle [3] found that
admixtures providing low viscosity were efficient in lowering
rebound. Even though the real viscosity for the shotcrete was not
evaluated, it can be assumed that pd is related to the viscosity of
the shotcrete.
One hypothesis is that SCMs that bring thixotropic behavior to
the concrete are probably the best approach to reducing rebound.
The concrete must have compact, viscous behavior when no shear
is present (no shotcrete stream occurring) and lose viscosity under
shear but must recover its previous properties when stress stops.
Moreover, additions or admixtures that are known to give a thixo-
tropic behavior to concrete must be able to react quickly with
water as the shotcreting event is very fast.
4.2. Soil mechanics approach
As the material under study is stiff, a solution could be to study
it with a soil mechanics approach by determining cohesion and
internal friction angle and applying the Mohr-Coulomb criterion
as suggested by Jolin [5]. A theory of rebound was proposed by
Armelin and Banthia [28] and enable to calculate the probability
that a single particle bounces off a fresh concrete substrate. In this
theory, the fresh concrete substrate is defined as an elastic-
perfectly plastic material with a Tresca yield criterion, and param-
eters such as dynamic and static contact stress are required for the
calculation. According to Armelin and Banthia [28], the plasticity
must be high enough to maintain stability and cohesion under
the concretes own weight but allow particles to enter the medium
and plastically deform it to lower the rebound. Youngs modulus
must be high in order to reach the plastic deformation more
rapidly during the impact (and limit the elastic energy imparted
to the particles).
Both the above approaches can help to explain the physical
properties needed for a mixture to have lower rebound. However
the mechanisms of action of SCMs and the role of water in concrete
under impact are not fully understood, but further studies of pd
might clarify their role. This would then facilitate better choices
when it comes to mixture design.
The rebound phenomenon and the measure of pd have been
studied by making changes in the SCM and water content. Since
the granular skeleton affects the rebound [10], to what extent
the granular particle-size distribution may influence the dynamic
contact stress should be investigated to check (a) whether the
experimental results still agree with the curves proposed in
Fig. 6, and (b) whether the relationship found between the rebound
and the dynamic contact stress has a general validity or applies
only to a given granular skeleton.
231
5. Conclusion
Rebound of dry-mix shotcrete is a complex phenomenon
involving many parameters. To establish the technical parameters
affecting rebound, it is necessary to study the influence of selected
parameters on shotcrete mixtures as shot.

Five different mixtures were shot with various SCMs. Metakao-
lin, silica fume and ground granulated blast furnace slag
decreased rebound. However, their efficiency appears to depend
on the consistency of the mix.

Water is a major parameter in rebound but its control is made
difficult by the fact that it is adjusted by the operator. Mixtures
with metakaolin and silica fume appear to lower this
dependence.

Static penetration stress and dynamic penetration stress, (P)
and (Pd) were used to evaluate fresh shotcrete properties. Both
parameters appear to correlate with rebound.

The relation between dynamic penetration stress and rebound
turns out to be independent of the mixture tested. A better
comprehension of the physical phenomenon behind dynamic
penetration stress would lead to further understanding of con-
crete behavior under impact.
References
[1] ACI Committee 506, Specification for Shotcrete (ACI 506.2-95), 2008.
[2] H.S. Armelin, N. Banthia, D.R. Morgan, C. Steeves, Rebond in dry-mix shotcrete,
Concr. Int. 5460 (1997).
[3] M. Pfeuffer, W. Kusterle, Rheology and rebound behaviour of dry-mix
shotcrete, Cem. Concr. Res. 31 (2001) 16191625.
[4] V. Bindiganavile, N. Banthia, Effect of particle density on its rebound in dry-mix
shotcrete, J. Mater. Civ. Eng. 21 (2009) 5864.
[5] M. Jolin, Mechanisms of Placement and Stability of Dry Process Shotcrete (Ph.
D. thesis), University of British Columbia, Canada, 1999.
[6] S. Grmey, Evaluation des Paramtres Dobtention de la Qualit des Btons
Projets Utiliss Dans des Soutnements Provisoires, des Revtements
Dfinitifs et des Renforcements Douvrages (Ph.D. thesis), INSA Lyon, France,
2003.
[7] U.C. Puri, T. Uomoto, Properties of shotcrete (12) summarized research report
on the numerical modeling of shotcrete, Seisan-Kenkyu 51 (1999) 1720.
[8] H.W. Parker, G. Fernandez-delgado, L.J. Lorig, A practical new approach to
shotcrete rebound losses, in: ACI-SP54-Shotcrete Gr. Support, 1976.
[9] H.S. Armelin, N. Banthia, D.R. Morgan, Particule kinematics in dry-mix
shotcrete-research in progress, in: SCA/ACI Int. Conf., Simon Austin, 1996,
pp. 247255.
[10] M. Jolin, D. Beaupr, Effects of particle-size distribution in dry process
shotcrete, ACI Mater. J. (2004) 131135. 101-M15.
[11] D.R. Morgan, J.T. Wolsiefer, Silica Fume in Shotcrete, in: CANMET/ACI Int.
Work. Use Silica Concr., Washington, DC, 1991.
[12] D.R. Morgan, Dry-mix silica fume shotcrete in western Canada, Concr. Int. 10
(1988) 2432.
[13] V. Bindiganavile, N. Banthia, Rebound in dry-mix shotcrete: influence of type
of mineral admixture, ACI Mater. J. 97 (2000) 16.
[14] M. Jolin, D. Beaupr, S. Mindess, Tests to characterise properties of fresh dry-
mix shotcrete, Cem. Concr. Res. 29 (1999) 753760, http://dx.doi.org/10.1016/
S0008-8846(99)00071-X.
[15] H.S. Armelin, N. Banthia, Mechanics of aggregate rebound in shotcrete (Part
I), Mater. Struct. 31 (1998) 9198.
[16] P. Bamonte, P.G. Gambarova, A. Nafarieh, High-temperature behavior of
structural and non-structural shotcretes, Cem. Concr. Compos. 73 (2016) 42
53, http://dx.doi.org/10.1016/j.cemconcomp.2016.06.009.
[17] J. Armengaud, Optimisation des Proprits des Btons Projets Voie Sche (Ph.
D. thesis), Universit Toulouse III Paul Sabatier, 2016.
[18] A.D. Figueiredo, Rheological behavior of dry-mix shotcrete, in: Second
CANMET/ACI Int. Conf., 1999, pp. 113128.
[19] L.R. Prudncio, H.S. Armelin, P. Helene, Interaction between accelerating
admixtures and Portland cement for shotcrete: the influence of the
admixtures chemical base and the correlation between paste tests and
shotcrete performance, ACI Mater. J. 6 (1996) 619628.
[20] K.L. Johnson, Contact Mechanics, Cambridge University Press, 1985.
[21] H.S. Armelin, Rebound and Toughening Mechanisms in Steel Fiber Reinforced
Dry-mix Shotcrete (Ph.D. thesis), University of British Columbia, Canada, 1997.
[22] R. San Nicolas, Approche Performantielle des Btons Avec Mtakaolins
Obtenus Calcination Flash (Ph.D. thesis), Universit de Toulouse, France,
2011, http://dx.doi.org/10.1017/CBO9781107415324.004.
232 J. Armengaud et al. / Construction and Building Materials 135 (2017) 225232

[23] M. Cyr, Contribution la Caractrisation des Fines Minrales et la [26] G.H. Tattersall, P.F.G. Banfill, The Rheology of Fresh Concrete, Pitman Advanced
Comprhension de Leur Rle Jou Dans le Comportement Rhologique des Pub. Program, Boston, 1983.
Matrices Cimentaires (Ph.D. thesis), INSA Toulouse, France, Universit de [27] H. Zhao, W. Sun, X. Wu, B. Gao, The properties of the self-compacting concrete
Sherbrooke, Canada, 1999. with fly ash and ground granulated blast furnace slag mineral admixtures, J.
[24] J. Ambroise, S. Maximilien, J. Pera, Properties of Metakaolin blended cements, Cleaner Prod. 95 (2015) 6674, http://dx.doi.org/10.1016/j.
Adv. Cem. Based Mater. 1 (1994) 161168, http://dx.doi.org/10.1016/1065- jclepro.2015.02.050.
7355(94)90007-8. [28] H.S. Armelin, N. Banthia, Development of a general model of aggregate
[25] B.B. Sabir, S. Wild, J. Bai, Metakaolin and calcined clays as pozzolans for rebound for dry-mix shotcrete (Part II), Mater. Struct. 31 (1998) 195202.
concrete: a review, Cem. Concr. Compos. 23 (2001) 441454, http://dx.doi.org/
10.1016/S0958-9465(00)00092-5.