DOI 10.1617/s11527-012-9828-z
ORIGINAL ARTICLE
A. G. F. Gibb • T. Thorpe
Received: 4 February 2011 / Accepted: 4 January 2012 / Published online: 19 January 2012
Ó RILEM 2012
Abstract This paper presents the experimental Keywords Additive manufacturing Build
results concerning the mix design and fresh properties Concrete Extrusion Open time Printing
of a high-performance fibre-reinforced fine-aggregate
concrete for printing concrete. This concrete has been
designed to be extruded through a nozzle to build 1 Introduction
layer-by-layer structural components. The printing
process is a novel digitally controlled additive Concrete is normally placed into formwork and then
manufacturing method which can build architectural vibrated to fabricate building components. Two alter-
and structural components without formwork, unlike native construction strategies—self-compacting and
conventional concrete construction methods. The sprayed concretes—have been developed to eliminate
most critical fresh properties are shown to be extrud- the compaction process. The basic principle of
ability and buildability, which have mutual relation- self-compacting concrete is that aggregate particles
ships with workability and open time. These properties form a smooth grading with a minimum void content
are significantly influenced by the mix proportions and and a considerable volume of cementitious paste with
the presence of superplasticiser, retarder, accelerator a superplasticiser, which fill the gap between aggre-
and polypropylene fibres. An optimum mix is identi- gate particles to create flowability [1, 2]. However,
fied and validated by the full-scale manufacture of a concrete construction using self-compacting concrete
bench component. still has to use formwork which needs considerable time
and labour for setting up and constrains geometries to
minimise the materials cost by re-using the forms.
Sprayed concrete [3, 4] is another approach to
making concrete components usually using a backing
T. T. Le (&) S. A. Austin S. Lim R. A. Buswell material (natural, like rock, or man-made) to eliminate
A. G. F. Gibb T. Thorpe temporary formwork. The mix proportions are
Department of Civil and Building Engineering,
designed for a minimum voids by a grading with the
Loughborough University, Loughborough,
Leicestershire LE11 3TU, UK largest size particles filled with particles of the next
e-mail: lttnvn@gmail.com size fraction down, and so on. This is particularly
important with the wet process to facilitate plug flow
Present Address:
in the delivery hose [4] and is achieved by using a
T. T. Le
National University of Civil Engineering (NUCE), Hanoi, combined grading (aggregate, cement and powdered
Vietnam additions). A relatively high cement content is needed
1222 Materials and Structures (2012) 45:1221–1232
to facilitate adhesion and build-up thickness and form without any assistance of vibration) and sprayed
a lubricating layer around the inside of conveying concrete (i.e. fresh concrete is expelled from a nozzle
pipes. This is also to make sure that the mix can be to fabricate complex forms) to meet the critical
pumpable and sprayable. Formwork is reduced when requirements of a freeform construction process. The
using this method but the shape and form of building self-compacting performance of a printing concrete is
components is not precise. dependent on the ability to extrude consistent fila-
Concrete printing is an innovative construction ments while the use of wet-process sprayed concrete
process for fabricating concrete components employ- principles can help ensure that the fresh concrete is
ing an additive, layer-based, manufacturing technique, conveyed effectively in the pipe and pump system
also called freeform construction [5, 6]. This method without blocking. An additional requirement is the
can be used to build complex geometrical shapes ability to build layers without significant deformation
without formwork, and thus has a unique advantage due to self-weight before setting.
over conventional construction methods. Briefly, A systematic research programme and experimen-
components are designed as volumetric objects using tal results are presented to identify optimum mix
3D modelling software. They are next sliced and proportions together with the ways developed to
represented as a series of two dimensional layers. The control and measure the fresh properties, namely
data are exported to a printing machine layer-by-layer extrudability, workability, open time and buildability.
in order to print structural components by the
controlled extrusion of a cementitious material. The
potential advantages of this process include: (a) inte- 2 Concrete printing process
gration of mechanical and electrical services within
voids formed in the structure could optimise materials The prototype concrete printing machine operates
usage and site work; (b) better control of the deposi- within a 5.4 m (length) 9 4.4 m (width) 9 5.4 m
tion of build material can produce novel internal and (height) steel frame (see Fig. 1). The system comprises
external finishes; (c) creating integrated units will a printing head which is digitally controlled by a CNC
reduce interface detailing and hence the likelihood of machine to move in X, Y and Z directions via three
costly remedial works; and (d) the coupling of a chain-driven tubular steel beams. A material container
layered construction process with solid modelling is mounted on top of the printing head and connected to
techniques will give greater design freedom. a pump to convey the material to the printing nozzle
The development of a printing concrete is reported to which is activated by the CNC machine.
exploit the potential of this advanced method of The printing process is in three stages: data
construction. The concrete needs to have an acceptable preparation, concrete preparation, and component
degree of extrudability to be extruded through a printing printing. In the data preparation stage a component
head containing nozzles to form small concrete is designed as a 3D CAD model, then converted to an
filaments. The filaments must bond together to form STL file format and sliced with a desired layer depth.
each layer, as the fresh concrete is continuously The printing path for each layer is then generated to
extruded to form consecutive filaments layered on the create a G-Code file for printing. Concrete preparation
previous ones to build complete 3D components. involves mixing and placing it into the container. Once
Furthermore, the material must have sufficient build- the fresh concrete has been placed into the container, it
ability characteristics to enable it to lay down correctly, can be conveyed smoothly through the pump–pipe–
remain in position, be stiff enough to support further nozzle system to print out self-compacting concrete
layers without collapsing and yet still be suitable to filaments, which can build layer-by-layer structural
provide a good bond between layers. A high strength components. The schematic in Fig. 2 shows the
(of the order of 100 MPa in compression) was targeted delivery system of the concrete printing process.
in this research because the layered structure of compo- Experiments showed that the most critical proper-
nents fabricated by this method is likely to be inherently ties in the fresh state of this printing process are
weaker than conventional in situ and precast concrete. extrudability and buildability. Extrudability can be
In principle, printing concrete has the advantages of defined as the capacity of concrete to pass through the
both self-compacting concrete (i.e. self-compacting small pipes and nozzles at the printing head. It is
Materials and Structures (2012) 45:1221–1232 1223
(c) printing a large-scale freeform component to buildability. The methods used to measure the fresh
evaluate the feasibility of the printing concrete properties are described as follows. The strategy was
working with the concrete printing machine. to use very practical indications of successful printing,
in terms of the ability to extrude filaments into a layer
3.1 Materials and mix design (extrudability) and to then build-up multiple layers
(buildability). This allows an appropriate mix design
The mix design of concrete needed to meet the to be established. We then adopted shear strength, and
performance requirements of the fresh and hardened its change with time (to define workability and open
concrete. The performance of the former comprises time) to characterise the material scientifically and
extrudability and buildability which are significantly allow a mix to be fine-tuned.
influenced by the workability and open time. The key
properties of the hardened concrete are the compres- 3.2.1 Extrudability
sive and flexural strengths of both cast and printed
specimens. Initially, targets of compressive strength of Extrudability here relates to the ability to transport the
100 MPa and flexural strength of 12 MPa at 28 days fresh concrete through a hopper and pumping system
were set for mould-cast specimens. to a nozzle where it must be extruded as a continuous
A 2 mm maximum size sand was selected because filament. Earlier research concerning this ability
of the small nozzle diameter (i.e. 9 mm) to give a high referred to the pumping and spraying of concrete
printing resolution. Cement CEM type I 52.5, fly ash [3, 4, 7], but the fresh concrete exits the system as a
and undensified silica fume formed the binder com- particle stream, not an extrusion. Extruding concrete
ponent. The gradings, measured by a Mastersizer 2000 to build components has been mentioned in previous
machine, of sand, cement, fly ash and silica fume were research [8–11] but a suitable test method to evaluate
combined in various proportions to form smooth this ability of concrete has not been described.
grading curves of test mixtures. The dry components In this research, the extrudability was evaluated
were added to the mixing water together with a with 9 mm wide filaments (printed from a 9 mm
polycarboxylate based superplasticiser to lower the nozzle) that comprised five groups from one to five
water–binder ratio and hence increase its workability filaments as shown in Fig. 3. Each filament was
as well as strength. A retarder, formed by amino-tris 300 mm long and the total continuous length of
(methylenephosphonic acid), citric acid and formal- filaments extruded out for a test was 4,500 mm. The
dehyde, was added to maintain a sufficient open time, test shape was designed to represent the typical way of
facilitating a constant flow during printing stages. building freeform construction components. The test
An accelerator, formed by sulphuric, aluminium salt result was evaluated as YES or NO, the former being
and diethanolamine, was also investigated to control when the complete length of 4,500 mm was deposited
setting. The concrete contained 12/0.18 mm length/ successfully without a blockage or fracture.
diameter polypropylene micro fibres to reduce shrink-
age and deformation in the plastic state. The optimum 3.2.2 Workability
mix was considered to be the one with the lowest
content of binder that could be printed and built with Conventional methods evaluating workability include
the recommended dosage of fibres from the supplier slump, compacting factor and flow tests, for which
(i.e. 1.2 kg/m3) and gain the target strengths. various national standards are available. However,
these do not measure fundamental physical properties.
3.2 Experimental procedures An alternative approach is to conduct a more robust
rheological investigation, such as a two-point test
First, preliminary mixtures were tested for extrudabil- (e.g. Tattersall [12]) which can characterise Bingham
ity to find the optimum particle grading which include fluid behaviour in terms of shear strength and torque
sand, cement, fly ash and silica fume. Then, the mix viscosity. Such methods have their own difficulties
with optimum particle composition was tested with (especially in calibration) and cannot be conducted in
fibres and the admixture dosages varied to approach situ. Austin et al. [3, 4, 7] adopted a shear vane
the optimum extrudability, workability, open time and apparatus (originally for measuring the shear strength
Materials and Structures (2012) 45:1221–1232 1225
The open time in this research, was determined as compressive strength concrete based upon a previous
the time period in which the workability of fresh study of ultra high-performance concrete [18].
concrete was at a level that maintained extrudability. A pumpable sprayed mortar [3] and a commercial
The ending of open time was found to be once the shear pre-packaged mortar [19] are also shown for compar-
strength had increased by 0.3 kPa from the initial shear ison. The percentage passing 0.6 mm was over 95%
strength of the concrete as this increase in shear for all mixes, whilst the reference sprayable mixes
strength, i.e. decrease in workability, was found to were in the range of 83–90%. The printing mixes were
coincide with increasing difficulty of printing a good made finer to be more suited to deposition through the
quality filament (with an optimum mix). The work- 9 mm diameter nozzle.
ability of fresh concrete was therefore measured every The mix proportions of five preliminary mixes were
15 min using the shear vane to determine the open calculated assuming a density of 2,300 kg/m3
time. Preliminary tests showed that fresh concrete (Table 1). The dosages of superplasticiser, retarder,
samples usually had higher shear strength if they were accelerator and polypropylene fibres were varied to
not agitated before testing. As a result, shear strength identify optimum mix proportions for printing.
was tested in two ways: the first measurement was done
with non-agitated samples and the samples were then 4.2 Extrudability
agitated by shaking the container 10 times and the
second measurement taken (i.e. agitated sample). The extrudability was affected significantly by the
combination of particle sizes of the constituent mate-
3.2.4 Buildability rials. Mixes 1 and 2 were difficult to pass through the
pipe–pump–nozzle system due to the high sand content
Most concrete is placed into formwork as a fluid and that caused sand segregation. More water and super-
there is no need for it to be self-supporting, i.e. the plasticiser were added to make them more flowable but
buildability is not an issue. Sprayed concrete is an the concrete still segregated in the pipes which led to
exception, and research quantifying the buildability of blocking. The binder content of 25 and 30% in these
a mix has been done by Austin et al. [7]. The work mixes appear to be insufficient to create a suitable
focused on estimating the adhesion and cohesion of extrudability to cope with this printing process.
sprayed concrete in a horizontal build test. The One to two percent superplasticiser was also added
horizontal build is not appropriate for the current in Mixes 3 and 4 to improve the extrudability. They
process. In absence of other methods, the buildability could be extruded from the nozzle and printed.
of fresh concrete was quantifying as the number of However, Mix 4 was more suitable when incorporat-
filament layers which could be built up, based on the ing micro polypropylene fibres as it could be printed
shape used to evaluate the extrudability, without with up to 1.6 kg/m3 while Mix 3 could only be printed
noticeable deformation of lower layers. with a fibre dosage of up to 0.4 kg/m3. Above these
dosages the 9 mm diameter nozzle became blocked.
As Mix 4 passed the extrudability criteria with a fibre
4 Results and discussions dosage over that of 1.2 kg/m3 recommended from the
fibre supplier, it was identified as the optimum mix in
4.1 Mix design terms of extrudability and binder content. Mix 5,
which had even higher binder content, was therefore
Five preliminary mixes with different sand/binder not evaluated further.
proportions were designed by combining their particle Concerning the effect of workability on the extrud-
analysis results, see Fig. 5. The sand content was ability, a shear strength in the range of 0.3–0.9 kPa
reduced in 5% increments from 75 to 55% by weight (controlled by superplasticiser dosage) was optimum
of dry mixture in Mix 1 to Mix 5 whilst the binder for printing Mix 4. Lower than 0.3 kPa, the fresh
content was increased from 25 to 45%, respectively. concrete was too wet and segregation occurred in the
The binder in all mixes comprises 70% cement, 20% pipe–pump–nozzle system. The concrete filaments
fly ash and 10% silica fume. A water–binder ratio of extruded out were deformed with unstable cross
0.28 was selected to achieve over 100 MPa sections. Above 0.9 kPa, the fresh concrete was too
Materials and Structures (2012) 45:1221–1232 1227
60.00 Sand
Mix 1 (75% sand : 25% binder)
Mix 2 (70% sand : 30% binder)
40.00
Mix 3 (65% sand : 35% binder)
Mix 4 (60% sand : 40% binder)
stiff for printing and could not finish the extrudability optimise the other fresh properties, i.e. workability,
shape of 4,500 mm. The concrete filaments extruded open time and buildability.
out of the nozzle fractured and the mix could not be
printed continuously. 4.3 Workability
The cross section of extruded filaments had an
elliptical shape. Mix 4, with an initial shear strength of The workability was influenced significantly by the
0.55 kPa (1% superplasticiser and the water–binder dosage of superplasticiser, retarder and accelerator. In
ratio of 0.26 as the water content was reduced from the series without retarder and accelerator, increasing
232 to 216 kg/m3 to obtain the best extrudability), the superplasticiser dosage from 0 to 2% by weight of
gave consistent filaments with a cross section of binder increased the workability, i.e. reduced the shear
approximately 9 mm transverse diameter and 6 mm strength, see Fig. 6. The workability of Mix 4 with 0%
conjugate diameter. Mixes using 1.5–2% superplasti- superplasticiser could not be determined as the
ciser appeared too wet (initial shear strength of concrete was too stiff. To achieve a measurable
0.2–0.4 kPa) and led to deformed concrete filaments. workability (shear strength of 2.24 kPa) for this 0%
However, using 1% superplasticiser resulted in a mix superplasticiser mix, the water–binder ratio was
which could not be printed 30 min after mixing. increased up to 0.36 and then to 0.44 to gain a shear
Therefore a retarder was added to extend the time for strength of 0.57 kPa for printing. The superplasticiser
extrudability, i.e. open time. thus proved crucial for printing concrete to attain a
As Mix 4 was the optimum in terms of extrudabil- reasonable workability and also to attain high strength
ity, it was subsequently used to investigate and with a low water–binder ratio.
1228 Materials and Structures (2012) 45:1221–1232
ability of the printing concrete, increasing the retarder Retarder dosage (% by weight of binder)
or accelerator dosage reduced the workability, i.e. Fig. 7 Effect of retarder dosage on workability
increased the shear strength. The trend of changing
workability caused by the retarder appeared linear, 3.50
while that due to accelerator appeared nonlinear
(quadratic form), see Figs. 7 and 8. Accelerator was 3.00
1.50
4.4 Open time 1st measurement
1.00
2nd measurement
The open time was investigated first with varying 0.50 3rd measurement
2.10
1.80
1.50
1.20
0.90
0.60
Shear strength zone to pass printability
0.30
0.00
0 30 60 90 120 150 180 210 240 270 300
Time (minute)
40
4.5 Buildability
20
The buildability of the optimum mix in terms of
0 extrudability, workability and open time (0.5%
0 0.5 1 1.5 2 2.5
Superplasticier dosage (% by weight of binder)
retarder with a 100 min open time) was examined to
find the optimum by varying the dosage of superplas-
Fig. 10 Superplasticiser dosage versus open time ticiser which resulted in different shear strength of the
fresh concrete. The results confirmed that outside of
For the non-agitated samples, the results showed 0.3–0.9 kPa shear strength the concrete could not
that adding retarder did not influence the open time build a test sample due to being either too wet or too
significantly. After less than 10 min the shear strength stiff (Fig. 13). A mix with 0.3 kPa shear strength
of all samples increased over 0.3 kPa, and the same could only build four layers for a one filament group
phenomenon was observed in the mix without and seven layers for a five filament group and the
retarder, see Figs. 9 and 10 (dotted curves). filaments were deformed considerably. A mix with
For the agitated samples, the open time was 0.9 kPa shear strength could not build two layers
extended significantly to 100 min when 0.5% retarder correctly as some broken points occurred in the
was added. However, increasing the dosage beyond filaments. The optimum mix in terms of buildability
0.5% appeared to decrease it. One percent retarder was again one with a 0.55 kPa shear strength as it
increased the shear strength by 0.3 kPa after 14 min could build up to 15 layers for a one-filament group
but after that it increased slowly for 90 min by another and up to 34 layers for a five-filament group, see
0.3 kPa. The retarder appeared not to be effective in Fig. 13. Indeed 57, 60, 61 and 61 layers could be
extending the open time with the mixes having low attained for 2, 3, 4 and 5 filament groups, respectively,
initial workability (i.e. high shear stress). Another with only one filament in each group collapsing
disadvantage of 1.0–2.0% retarder was that it reduced (Fig. 14).
1230 Materials and Structures (2012) 45:1221–1232
1.80
1.50
1.20
0.90
0.00
0 30 60 90 120 150 180 210 240 270 300
Time (minute)
140
Non-agitated samples
120
Agitated samples
Open time (minute)
100
80
60
40
20
0
0 0.5 1 1.5 2 2.5
Retarder dosage (% by weight of binder)
40
35
Number of layers built
30 y = -1.43x 2 + 12.97x + 4 Fig. 14 The buildability of a 0.55 kPa shear strength concrete
R² = 0.97
25 mix
Shear strength = 0.55 kPa
20
Shear strength = 0.30 kPa
15
was best suited to printing freeform components,
which had a 60:40 sand:binder ratio, comprising 70%
10 y = -0.14x2 + 1.66x + 2.4
R² = 0.98 cement, 20% fly ash and 10% silica fume, plus 1.2 kg/
5
m3 micro polypropylene fibres. This mix also needed
0
0 1 2 3 4 5 6 1% superplasticiser and 0.5% retarder to attain an
Number of filaments optimum workability of 0.55 kPa shear strength, an
optimum open time of up to 100 min and the ability to
Fig. 13 Effect of workability on buildability
build a large number of layers with various filament
4.6 Optimum mix proportions groups. The compressive strength of this mix, deter-
mined by casting 100 mm cube specimens, was
Five trial mixes were designed to identify optimum measured at 1, 7, 28 and 56 days in accordance with
mix proportions for printing concrete. The extrudabil- BS EN 12390-2:2009 [20], and found to be 20, 80, 110
ity tests showed that Mix 4 (water–binder ratio of 0.26) and 125 MPa, respectively.
Materials and Structures (2012) 45:1221–1232 1231
References
3. Austin SA, Robins P, Goodier CI (1999) The rheological 12. Tattersall GH, Banfill PFG (1983) The rheology of fresh
performance of wet-process sprayed mortars. Mag Concr concrete. Pitman Books Ltd., London
Res 51:341–352 13. British Standards Institution (1990) BS 1377-9:1990
4. Austin SA, Robins P, Goodier CI (2002) Construction and methods of test for soils for civil engineering purposes.
repair with wet-process sprayed concrete and mortar. Milton Keynes
Technical Report 56. The Concrete Society UK 14. Previte RW (1977) Concrete slump loss. ACI Mater J
5. Buswell R, Soar RC, Gibb A, Thorpe T (2007) Freeform 74:361–367
construction: mega-scale rapid manufacturing for con- 15. Ravina D et al (1994) Slump loss and compressive strength
struction. Autom Constr 16:224–231 of concrete made with WRR and HRWR admixtures and
6. Lim S, Le T, Webster J, Buswell R, Austin S, Gibb A, subjected to prolonged mixing. Cem Concr Res
Thorpe T (2009) Fabricating construction components 24:1455–1462
using layer manufacturing technology. Proc Int Conf Glob 16. Kim BG, Jiang SP, Aitcin PC (2000) Slump improvement
Innov Constr, Loughborough, pp 512–520 mechanism of alkalies in PNS superplasticized cement
7. Austin S, Robins P, Goodier CI (2005) Low-volume wet- pastes. Mater Struct 33:363–369
process sprayed concrete: pumping and spraying. Mater 17. Alhozaimy MA (2009) Effect of absorption of limestone
Struct 38:229–237 aggregates on strength and slump loss of concrete. Cem
8. Jacobsen S (2009) Flow conditions of fresh mortar and Concr Compos 31:470–473
concrete in different pipes. Cem Concr Res 39:997–1006 18. Le T, Soutsos MN, Millard SG, Barnett SJ (2007)
9. Mu B et al (1999) Cementitious composite manufactured by UHPFRC—optimisation of mix proportions. Proc Concr
extrusion technique. Cem Concr Res 29:237–240 Platf Int Conf, Belfast, pp 339–348
10. Shao Y et al (2011) Microstructure of extruded cement- 19. Tarmac Limited (2007) CEMROK 40A—product
bonded fibreboard. Cem Concr Res 31:1153–1161 information
11. Lombois-Burger H et al (2006) Kneading and extrusion of 20. British Standards Institution (2009) BS EN 12390-3:2009
dense polymer–cement pastes. Cem Concr Res 36:2086– testing hardened concrete. Milton Keynes
2097