Choetal 2019
Choetal 2019
net/publication/336021780
CITATIONS READS
10 7,932
4 authors:
Some of the authors of this publication are also working on these related projects:
Material characterization procedures for extrusion-based 3D concrete printing (3DCP) View project
All content following this page was uploaded by Gideon van Zijl on 25 September 2019.
ABSTRACT A microcontroller unit (MCU), the core of the 3D printer, controls the
Growing interest in additive manufacturing in the construction industry input/output (I/O) signals and is powered by a 12 V power supply unit
has promoted research in 3D printing of concrete. A gantry-type 3D (PSU) to accommodate other electronic features like fans and lighting.
printer for concrete was designed and manufactured at the Department The software generated G-code according to the 3D model is interpreted
of Civil Engineering, Stellenbosch University. Characteristics of 3D by the MCU and communicates with the stepper motors to produce the
printable concrete are investigated and reported. In the fresh state, required nozzle movement.
thixotropic behaviour is required in the form of high static shear yield The gantry printer allows three translational degrees of freedom
stress, but relatively low dynamic yield shear stress when the material (DOF), with the option for future expansion to rotational DOF to enable
is agitated by pumping and extrusion in the 3D printing process. Once complex geometrical detail with alternative nozzle shape and size.
agitation stops, i.e. when a layer has been 3D printed, fast re-building of Currently, a 25 mm diameter nozzle is utilised to print a uniform cross-
the static yield stress is required to retain the printed shape despite self- section around bends in the absence of rotational DOF. The movements
weight and that of upper layers. In the hardened state, the interfaces in of the gantry system in the horizontal plane (x and y-direction) and
the layered structure may be regions of weakness. Results of rheometer vertical direction are allowed by the mounted linear guide and rail
tests to characterise thixotropy, mechanical strength and stiffness tests, systems. In the horizontal plane, a belt and pulley system driven by
as well as interfacial bond strength tests in the hardened state, are selected stepper motors are employed for each axis, while ball screws
presented. An evolution of 3D concrete printed elements is presented. are the chosen mechanism for vertical movement of the gantry in which
back drive can be prevented if there is a loss of power. The 3D printing
1. INTRODUCTION platform shown in Fig. 2 is a stabilized trolley with 12 polyurethane
A research program on 3D printing of concrete (3DPC) is executed by the wheels with an ultimate capacity of 700 kg each. The trolley facilitates
Centre for Development of Sustainable Infrastructure (CDSI) at Stellenbosch the removal and transport of the printed object from the gantry frame.
University. An industrial-grade gantry type 3D printer of roughly 1 m cube The gantry frame is paired with a 3-phase 380V 3kW locally
build volume, was designed and manufactured in 2017-2018. The point of manufactured concrete pump (Rockcrete TSL), originally purchased for
departure was to develop a robust, versatile laboratory printer that enables research on the sprayed application of fibre concrete, also known as
research on a range of concrete material classes, printing speeds and shotcrete, with a maximum aggregate size of 4 mm. A variable
geometrical complexity. Knowledge and experience in concrete pumping, frequency drive (VFD) was supplemented for pump motor speed control,
spraying and extrusion [1-3] informed the selection of an appropriate which is also manipulated by the MCU. This enables control of the
concrete pumping pressure range for the standard to high performance concrete pumping speed, required for instance when bends of small
(HPC) grades of concrete and fibre reinforced concrete (FRC). radius are printed and excessive deposition may result in inconsistent
filament thickness, or bulging.
The Marlin firmware based MCU is paired with two G-code based
2. DEVELOPMENT OF A 3D CONCRETE PRINTER
software packages in operator PC, namely ‘Slic3r’ and ‘Simplify3D’.
Three types of 3D concrete printers are typically used, gantry, robotic,
Both packages provide various customizing options for different types
and crane systems (see Fig.1). Each type has merits and drawbacks,
of printer and material. Particularly, vase mode printing, printing spirally
however, the gantry type was selected since it is relatively simple and
upwards instead of printing each level of layers, is functional in both
inexpensive to build, versatile and the most common type in the
software. The option saves time and improves the printing quality for
research field. Conceptual, mechanical and electrical design commenced
certain objects such as circular columns.
in 2017. The industrial-grade, laboratory scale, gantry-type 3D printer
The 3D concrete printer enables the research team to produce
with outer dimensions 1.3 x 1.3 x 1.67 m with a build volume of roughly
laboratory scale structural elements (walls, beams, columns etc.) for
1 m3 was constructed within 7 months.
Figure 1 - 3D concrete printers: (a) Gantry (b) Robotic (c) Crane type [1]
1
Department of Civil Engineering, Stellenbosch University, South Africa
CONCRETE BETON 11
technical paper
and shape of a printed object also influence the print quality, but
the distinct rheological characteristics of the material will notably
influence the buildability and pumpability.
Rheology is the branch of physics that studies the deformation
and flow phenomena of matter. Material flow initiates when the
applied shear stress exceeds the static yield stress of the material.
Once the material starts to flow, the measured shear stress reaches
a quasi-equilibrium plateau, defined as the dynamic shear stress. A
distinct difference between the static and dynamic yield stress of
the material is required for 3DPC and is defined as thixotropic
behaviour. Thixotropic fluids as dispersions build an intermolecular
system of forces at rest, which results in flocculation of particles
that increases the viscosity of the fluid, enabling it to support its
own weight [7]. Flow initiates when the intermolecular structures
are broken by external energy via agitation, resulting in shear-
thinning of the material. This phase is called de-flocculation, which
Figure 2 - 3D concrete printer component layout and final assembly
reduces the viscosity of the material. Once the external energy is
removed, the thixotropic material starts to re-flocculate and
regains pseudo-solid behaviour as the original microstructure is
rebuilt and the original viscosity is restored. This 3-phase thixotropic
testing, with the goal to contribute towards fundamental research for
behaviour is depicted in Fig. 3 (c). Each phase illustrated in the
the 3DPC industry. The printer complies with structural design criteria of
figure can be closely associated with the following timescales:
strength and stiffness, in order to accommodate high volume concrete
• Flocculation: concrete is placed into the hopper of the pump.
printing and nozzle head speed without excessive vibration.
• De-flocculation: the material is agitated by rotational paddle or
auger, and pumped through the hose.
3. MATERIAL CHARACTERISTICS FOR 3D PRINTABLE • Re-flocculation: the material is extruded and deposited onto
CONCRETE the printing bed or a previous layer.
3.1 Rheology
3D printing is widely used in various industries, such as the The thixotropic behaviour is determined by several factors including
automobile, aviation, biomedical, dental, food, fashion and the water content, temperature, chemical admixtures, cement
manufacturing industries. Despite the various and sophisticated extenders, aggregate grading, and rate of hydration. The
additive manufacturing techniques developed, the material restoration rate between the static and dynamic yield stress can be
deposition and formation are mainly thermal-processed. Hence, considered as the degree of thixotropy. A higher degree of
heating and cooling techniques are vital features of the printing thixotropy is desired for 3D printing application in concrete. The
process. However, the conventional 3D printing deposition and degree of thixotropy can be characterised by measurement of the
thermal-processed formation, such as metals or plastics, are not shear stress of the fresh concrete with a rheometer. The measured
essential for concrete, although heating to 60 – 70 ºC can be used initial static and dynamic yield stresses are 1860.4 Pa and 1474.4
for faster setting and strength development. Note that no thermal Pa according to Fig. 3 (b). The test result further shows rapid
activation was used in the 3DPC research results reported here. restoration and build-up of the static yield stress within a short
Instead, concrete requires modified rheology through optimised period of less than 1 minute. Note that return of the 3D printer
mix design, admixtures and other additions including nano- nozzle to deposit an upper layer varies according to nozzle speed
materials [4,5] and fibres [6] to be suitable for 3D printing in terms of and the printed structural element’s geometry and size. Typical
transport (pumpability) and buildability, or extrudability and shape return periods can be several minutes. For a laboratory test 3D
retention under self-weight and subsequently deposited upper print of a 400 mm diameter circular hollow column, the return
layers. period is 21 seconds with a 60 mm/s printing head speed. The re-
As additive manufacturing is formwork free, the freshly extruded building rate of the material static shear yield stress, is used to
material must have appropriate shape retention, and develop design a 3D concrete printing process, in order to prevent instability
sufficient strength and stiffness rapidly to support subsequent and collapse of the unsupported printed object, considering the
layers. Hence, traditional concrete designed for casting in formwork geometry and print settings [8]. Fig. 4 presents a bilinear time-
cannot be directly used. Concrete can be designed to be stiff, or evolution of static yield shear stress, collecting the peak shear
highly viscous, in the form of so-called zero slump / pseudo-solid stress values from Fig. 3 (b), and more at longer waiting periods up
concrete, by modification of the constituents and their proportions to initial set of the concrete. The graph in Fig. 4 serves as a 3DPC
to retain the extruded shape without significant deformation. design tool for the time-dependent resistance, which must exceed
Shape retention is one of the prime factors influencing the quality the demand, typically the weight of upper layers as 3D printing
of the extruded layer. In contrast, the material is required to have progresses. The initial gradient of static yield shear strength
low viscosity at transport during pumping, minimizing the evolution is depicted by RThix, referring to the rapid rebuilding by
probability of ingredient segregation under high pumping pressure re-flocculation. The second gradient is depicted AThix, which refers
in the range of 1 to 4 MPa [1]. Segregation may lead to blockage or to a lower rate of strengthening, brought about a structuration
poor dispersion and associated inferior mechanical and durability process following re-flocculation (Fig. 3 c).
properties of the extruded product. The printing speed, path, size
Shear Stress
Static Yield Stress
Time
(a)
1000
0
0 2 4 6 8 10 12
(b) Time (s)
(d)
Silica Fume 83
The 3D printable mix was designed mainly based on the rheological
Sand 1167
performance of the material. From the literature [9,10], the optimal
mix design was developed with locally available constituents, Water 261
including CEM II/A 52.5N cement, Class S fly ash, silica fume Superplasticizer 1.48% by mass of binder
(Microfume supplied by Silicon smelters), and natural Malmesbury
Viscosity Modifying Agent 0.13% by mass of binder
sand with fineness modulus of 2.12. The final mix design, with
CONCRETE BETON 13
technical paper
% passing
loaded at third points (x = 50 mm). All specimens failed at an interface
within the uniform bending moment zone (between load points). From
40
the ultimate load, the maximum flexural strength was calculated,
considering linear elastic behaviour, i.e. the modulus of rupture was
20 calculated. These flexural strengths are denoted as the interfacial
(flexural) bond strength (IBS) in Table 2, and determined at 28 day age
0 only on three specimens. Note that the 3D printed column was kept in
Pan 0,075 0,15 0,3 0,6 1,18 2,36 4,75 laboratory conditions at 23 ±2ºC and 65 ±5% relative humidity until
Sieve Size (mm) sawing the specimens and subsequent testing.
The 19% lower interfacial flexural strength (IBS) of 6.8 MPa,
Figure 5 - Particle size distribution of the binder and sand combined, compared with the cast specimen flexural strength of 8.4 MPa could be
compared to Fuller-Thompson’s ideal curve ascribed to both the curing regime and the interfacial bond
characteristics. Note that there was a short time lapse of less than one
The Fuller Thompson theory [11], which aims to achieve maximum
packing density, is used as optimal aggregate grading target. The Table 2 – Mechanical characteristics of 3D printable mortar
shear rate of cement paste is inversely proportional to granular (coefficient of variation in brackets)
aggregate packing fraction [12] in which the shear rate is determined
Age ffl (MPa) fcu (MPa) E (GPa) IBS (MPa)
by the state of flocculation. Maximum packing and minimum voids
are expected to yield lower shear rates and higher state of 1 day 1.7 (0.081) 7.9 (0.026)
flocculation. A recent study by Weng et al. [13] confirms that the 7 days 7.3 (0.050) 55.6 (0.006) 26.6 (0.023)
Fuller Thompson theory can serve as a reasonable rheological
28 days 8.4 (0.057) 70.6 (0.103) 30.8 (0.020) 6.8 (0.021)
design approach for 3DPC. A natural sand available in the Western
Cape is selected for use in the 3DPC mix reported here. It is 56 days 8.7 (0.02) 80.0 (0.151)
possible to achieve a Fuller-Thompson grading curve by sieving and
re-mixing, which is generally an expensive process. Given the
reasonable agreement with the ideal curve shown in Fig. 5, and
associated cost of re-grading, the natural sand is accepted. Note
that Fig. 5 shows the combined grading of the binder and sand
particles.
Nozzle diameter 25 mm
Filament diameter 25 mm
Extrusion width 25 - 30 mm
(a)
Choose
Print head speed
(b)
Calibrate
Extrusion Rate
(Using G code)
If unacceptable
Extrude layer
width check
(c)
If acceptable
Ready to print
(Save the setting for future)
(d)
Recall from Section 3 that freshly mixed concrete has a certain stiffness
and flowability based on the constituents. Since the material rheology Figure 8 - Typical poor condition or failure during the printing
process
cannot be modified simply without changing mechanical properties,
the stiffness and flowability of the material must be characterised or
investigated in advance to be acceptable for printing, while also directly affects the interlayer bond strength, which influences the
attaining appropriate mechanical properties in the hardened state. integrity of the structural component and its mechanical properties.
Thus, the print head speed is only considered as a variable parameter to The layer thickness was obtained in the range of 10 – 15 mm depending
determine the optimum extrusion rate. The printing speed was tested on the condition of the material.
within a range of 50 – 100 mm/s based on the previous study by Nerella
et al.[9]. Since the standardised test for printability of the material has During the calibration process, a few typical observations (Fig. 8) of
not yet been developed, the determination of the parameter was trials performed show imperfection and irregularity in quality or global
judged by the authors. For each printing speed, the extrusion rate must failure:
be calibrated until the extruded layer width does not exceed the nozzle (a) A loose translational pulley not engaged properly with stepper
diameter by 5 mm, thus a layer width in the range of 25 – 30 mm. motors resulting in a slip. The slip causes an inaccurate print path
The structural component in the construction environment is, first or shifting of the printing plane entirely.
of all, desired to have global stability. In a first approach the buckling (b) Residual water inside of the hose altering the flowability of the
effect as a cause of failure is disregarded to investigate the buildability material.
of the concrete material. Hence, cylindrical hollow column is chosen for (c) At the vertical lifting position for each layer, extra material is
a buildability test model. The printing time, the compressibility of each deposited due to gravity feed. Thus, an actuator at the nozzle is
printed layer under the self-weight and printed object height were required to control material flow more accurately.
compared in each printing outcome. The optimum layer thickness was (d) Printing a bathtub. The action described in (c) causes global
investigated by printing the buildability model. The layer thickness deflection of the wall.
CONCRETE BETON 15
technical paper
CONCRETE BETON 17
technical paper