Article published in
Rapid Prototyping Journal, Vol. 29 No. 3 (2023), pp. 639-654.
A sustainable formwork system based on ice pattern and
sand mould for fabricating customised concrete components
Wei Li, Xiaoshan Lin and Yi Min Xie
Centre for Innovative Structures and Materials, School of Engineering, RMIT
University, Melbourne 3001, Australia
Purpose – Optimised concrete components are often of complex geometries, which are
difficult and costly to cast using traditional formworks. This paper aims to propose an
innovative formwork system for optimised concrete casting, which is eco-friendly,
recyclable and economical.
Design/methodology/approach – In the proposed formwork system, ice is used as
mould pattern to create desired geometry for concrete member, then sand mould is
fabricated based on the ice pattern. A mix design and a mixing procedure for the
proposed sand mould are developed, and compression tests are also performed to ensure
sufficient strength of the sand mould. Furthermore, surface preparation of the sand
mould is investigated for easy demoulding and for achieving good concrete surface
quality. Additionally, recyclability of the proposed sand mould is tested.
Findings – The proposed mix design and mixing procedure can provide sufficient
strength for sand mould in concrete casting. The finished components exhibit smooth
surfaces and match designed geometries, and the proposed sand mould can be fully
recycled with satisfactory strength.
Originality/value – This is the first study that combines ice pattern and sand mould to
create recyclable formwork system for concrete casting. The new techniques developed
in this research has great potential to be applied in the fabrication of large-scale concrete
structures with complex geometries.
Key words: Concrete formwork, Ice pattern, Sand mould, Surface roughness,
Recyclability, Optimisation, Advanced manufacturing technologies, Manufacturing
technology, Fabrication, Structures.
Paper type: Research paper
Concrete is one of the main construction materials. It is widely used in civil engineering
structures, and the application of concrete has a long history. On one hand, concrete
possesses excellent features, such as high structural performance, outstanding
durability and good accessibility. On the other hand, concrete exhibits excellent
formability to achieve any shapes regardless of geometric complexity. Formwork is
essential in the construction of concrete structures. It is required to shape fresh concrete
into desired geometries and support concrete structures to gradually develop strength.
As traditional formworks usually account for 35 to 60 percent of the total cost in the
construction of concrete structures (Krawczyńska-Piechna 2017), the repetitive use of
formworks is highly desirable. However, for modern structures with complex geometry
designs, traditional formworks cannot be reused in other structures, leading to a huge
waste of material and cost. Thus, the economy of formwork system should be
thoroughly considered in the formwork design. Another major concern for researchers
and engineers is the environmental impact of traditional formworks due to the waste
generation, energy and resource consumptions and lack of recyclability. Therefore, a
sustainable formwork system for achieving optimised efficient concrete design is of
Rigid materials like wood and steel have been traditionally used in formwork systems
(Li et al. 2022). Wooden formwork has been commonly used for the construction of
cast-in-situ concrete structures. Nevertheless, wooden formwork requires highly
qualified carpenters working on site, and it usually takes a significant amount of time.
Besides, for larger or more time-consuming jobs, wood is a less ideal choice due to its
tendency to warp, shrink and swell. Steel and aluminium formworks have been used as
alternatives to wooden formwork to enhance stiffness and durability. However, steel
and aluminium modular formwork systems are most suitable for high rise buildings
with regular and orthogonal shapes (Gaddam & Achuthan 2020). Very few structures
with complex curved geometries are found to be constructed using steel or aluminium
formwork as material waste and undesirable cost can be significant without repetitive
In recent years, with the advancement and application of optimisation techniques in
civil engineering constructions, structures with complex shapes are designed to achieve
better structural efficiency and aesthetic value. It has become difficult for traditional
formworks to cope with the increasing need of concrete structures with customised
geometries due to the restrictions in geometry, labour, material waste and cost. Under
this circumstance, flexible formwork systems have been developed for casting freeform
concrete components. A comprehensive study on fabric formwork was carried out by
West and his co-workers (West 2016). A series of columns, walls, beams and trusses
with irregular geometries were successfully constructed using various textiles as
formworks. One of the major advantages of fabric formwork is that it allows the
fabrication of concrete structures with complex geometries and keeps the cost at a low
level. However, the pollution, wrinkling of fabric during concrete casting and material
waste are the problems. Moreover, to maintain the desired shape of fabric formwork,
exterior load and frame are required. In addition, digitally fabricated formwork system
has shown a significant development over the last decades. Both subtractive and
additive technologies have been used to produce concrete moulds in various shapes
(Menna et al. 2020). In the study conducted by Dombernowsky and Søndergaard
(Dombernowsky & Søndergaard 2012), an optimised non-uniform, doubly curved
concrete element was cast using expanded polystyrene (EPS) formwork fabricated by
computer numerical control (CNC) milling technology. However, to achieve industrial-
grade surface smoothness, a lot of machining time is required for the EPS formwork.
Also, the raw material was wasted by being removed from the original block in this
process, and the reuse of formwork was limited to the same shape. 3D printed plastic
formwork was investigated by Katzer and Szatkiewicz (Katzer & Szatkiewicz 2019)
for constructing concrete components. Although plastic formwork with freeform
geometry could be achieved using 3D printing technology without much labour
involved, the printing process is time-consuming, and the material is costly. Moreover,
as the used 3D printing material could not be recycled, it may pose additional burden
to the environment. Therefore, a cost-effective formwork system which could be used
to cast concrete elements with arbitrary geometries with minimum material waste is in
Sand casting has been widely used in metal forming industry for fabricating complex
metal parts. The alkaline-phenolic resin system has been one of the most adopted
approaches for making sand mould, which is the combination of sand, binder and
hardening chemical (Brown 2000). Once the three ingredients are mixed together,
binder and hardener would react and allow the sand mixture to gain strength gradually.
Sand mould made with this system has nice surface finish as well as very low fume and
odour during sand casting. The alkaline-phenolic resin system consists of water soluble
alkaline phenolic resin and liquid ester hardener. When preparing sand mixture, 2% to
3% of alkaline resole by weight of sand is firstly blended into sand, followed by 20%
to 30% of hardener based on the weight of resin. The hardening speed of sand mould is
mainly dependent on the type of organic ester, the mix design as well as the mixing
temperature (Huang et al. 2016) Recently, 3D sand mould printing technology has been
extensively explored in metal forming industry (Upadhyay et al. 2017). Nevertheless,
there are limitations for it to be applied in civil engineering. 3D printing of sand mould
has a relatively high initial investment and operational cost. Compared to metal parts,
concrete elements usually have larger dimensions, which will be restricted by the size
of 3D printed sand mould. Moreover, the printing process is much slower than
traditional pattern moulding process for large scale production. Thus, pattern moulding
process is preferred in the fabrication of sand mould for concrete construction. When
sand mould is used for concrete casting, loose particles on the surface of sand mould
may be absorbed into fresh concrete, causing poor surface finish and difficulties in
demoulding process. Therefore, additional investigation on the improvement of
interface quality is required when sand mould is to be used for concrete casting.
Additionally, concrete casting on ice formwork was investigated by Sitnikov (Sitnikov
2019). Based on the idea of waste free production, free-form prefabricated concrete
elements were cast using the CNC-milled ice formwork (Sitnikov 2018). Compared
with conventional formwork materials, a decreased embodied energy and carbon
footprint was achieved. However, due to the low temperature requirement, the
application is limited to a specially designed frost-resistant high-performance concrete
In this study, an innovative formwork system is proposed to realise freeform
construction with economical and eco-friendly formwork materials and fabrication
process. In the new formwork system, ice is used as mould pattern to create desired
geometry for concrete member, and then sand mould is fabricated based on the ice
pattern. Once sufficient strength is developed in the sand mould, ice pattern is allowed
to naturally melt away. The sand mould is then served as formwork for casting concrete.
Compared to other formwork systems, the advantages of the proposed formwork
system should be highlighted. First of all, the use of ice and sand as formwork materials
could resolve the problem of material waste. They could be reused and formed into
different shapes after demoulding, which efficiently closes the material loop of
production. It is usually costly to carve wood into complex shapes, and wood mould
cannot be reused if a different pattern is required. Polymers are also costly compared to
ice and sand, and the cast concrete is hard to be demoulded from the polymer mould.
Besides, polymer mould cannot be recycled, which would cause material waste and lay
burden on the environment (Czarnecki & Sadowski 2022). The proposed formwork
system with ice and sand is amenable and recyclable, which is more sustainable and
suitable for fabricating customised concrete components obtained from optimised
design. Secondly, water and sand are easily accessible in any inhabited area, and the
material cost is low. Digital fabrication methods have been increasingly used for mass
production of metal components, while sand casting is suitable for one-time customised
product. In concrete construction, the proposed formwork is intended to be used in the
fabrication of customised concrete components. Once the component is obtained, the
formwork material can be recycled and reused for other projects. Thirdly, considering
the use of ice pattern, the proposed formwork system can be easily applied in the
frigid/subfrigid zones. In other areas, the fabrication of sand mould could be
accomplished in winter to save energy. Additionally, the resin and hardener adopted for
the proposed sand mould are environmentally friendly, as they produce very low fume
and odour during sand casting. Finally, as ice naturally melts away, the sand mould can
be retrieved from ice pattern without labour involved. The environmental impact of
concrete and its formwork can be endlessly debated. The use of sustainable and
environmentally friendly material as concrete formwork could largely alleviate the
This paper is organised as follows. The parametric studies for the proposed sand mould
are introduced in Section 2. The compression tests of sand mould specimens are
presented in this section to ensure the proposed sand mould has sufficient load bearing
capacity for concrete casting. Different release agents are tested for concrete casting on
the sand mould. The selected release agent shall not only meet the requirement for easy
demoulding, but also make sure that the final concrete components have satisfactory
surface roughness. Cases studies on the fabrication of optimised concrete structures are
presented in Section 3 to demonstrate the effectiveness of the proposed formwork
system. The recyclability tests of sand mould and the concrete construction with
recycled sand mould are also presented. A detailed discussion of the proposed
formwork system, including its eco-friendliness, geometric flexibility, surface finish
and fabrication accuracy, is provided in Section 4. Finally, the conclusions are drawn
based on the present study.
2. Materials and methods
2.1. Mix design for sand mould
In this study, silica sand, ZTS-968 alkaline phenolic resin and ZTG-01 ester hardener
are used for making sand mould. The silica sand used for sand mould casting is the
waste sand collected from lab with chemical composition listed in Table 1. It can be
reused for many times, which closes the loop of material waste. The resin is a highly
viscous liquid, while the hardener has a low viscosity. As ice is served as mould patten
for sand casting, the ZTG-01 ester hardener is selected, which is suitable for casting
sand in low temperature environment. The characteristics of ZTS-968 alkaline phenolic
resin and ZTG-01 ester hardener are listed in Table 2. Sand mixing is conducted by a
continuous sand mixer, and the sand specimens are fabricated using standard moulds
with dimensions of 50 mm × 50 mm × 50 mm. The strength tests are measured by a
universal strength testing machine.
Table 1 Chemical composition of silica sand.
Chemical composition (wt%)
SiO2 Al2O3 Fe2O3 Na2O + K2O
≥98 <1 <1 0.2 to 0.4
Table 2 Characteristics of ZTS-968 alkaline phenolic resin and ZTG-01 ester hardener
(25 °C, mPa s)
(25 °C g/cm3)
ZTS-968 resin 70 to 90 1.2 to 1.3 - ≤0.2 ≤0.2
ZTG-01 hardener low 1.10 to 1.18 ≥90 - -
2.2. Compression tests on sand mould
Two environmental temperatures (-5°C representing temperature of ice pattern and
25°C representing room temperature) and three mixing times (150 s, 300 s and 450 s)
are tested to study the effects of temperature and mixing time on the final quality of the
hardened sand specimen. Three sand specimens are cast under the same condition for
each group, e.g., 1-1,1-2,1-3. In alkaline-phenolic resin casting system, the mass of
resin is suggested to be 2% to 3% of the mass of sand, while the mass of hardener
should be 20% to 30% of the mass of resin. Higher ratios of resin and hardener would
increase the compressive strength and reduce the curing time of sand sample (Brown
2000). In this study, 300 g sand, 9 g resin (3% of sand mass) and 2.7 g hardener (30%
of resin mass) are adopted to obtain each sand specimen with high compressive strength.
Figure 1 shows the flowchart for the process of casting sand specimen. Sand and
hardener are firstly mixed with a specific mix proportion at the selected temperature for
one minute. The resin is then added, and the mixing time is counted until finish. After
that, the mixed sand paste is cast into the mould. The sand specimens are cured in the
air at the selected temperature for 10 h. Finally, compression test is conducted by
universal testing machine to determine the strength of the sand specimens. The results
for the average compressive strengths of sand samples under different casting
conditions are shown in Figure 2. The compressive strength of each sand specimen is
also provided in Table A1 in the Appendix.
Figure 1 Casting process.
Figure 2 Average compressive strengths of sand samples at different casting conditions.
It can be seen from Figure 2 that with the same mixing time of 150 s, the sand specimens
cast at -5°C temperature have less compressive strength compared to the specimens cast
at 25°C. However, when the mixing time exceeds 150 s, the sand specimens cast at
25°C almost have no strength, and the compressive strength of sand specimens cast at
-5°C also decreases with the increase in the mixing time. When the mixing time is
longer, the resin and hardener might have started reacting and setting in the mixer. Thus,
the hardened sand may be torn apart by the mixer, resulting in lower strength of the
sand specimens. Furthermore, as higher temperature would accelerate the chemical
reaction and setting process, even weaker sand is obtained in this case. In addition, it is
also optimal to cast sand on ice pattern at sub-freezing temperature, as higher
temperature may result in the melting of ice pattern during casting process. Therefore,
it is suggested that the sand mould is mixed at -5°C temperature for less than 300 s. As
a minimum compressive strength of 0.049 MPa is usually required for silica sand-based
mould for casting products with density of up to 7800 kg/m3 (Sadarang & Nayak 2021),
the suggested mixing temperature and mixing time could ensure a sufficient sand mould
strength (above 0.6 MPa) to carry the weight of concrete components in most cases
(with density of 2500 kg/m3).
2.3. Surface preparation for sand mould
The quality of concrete surface is essential for the durability and aesthetic impact of the
final construction. Common issues such as surface voids, deviation due to sticking
concrete and concrete stains shall be avoided in the construction process. Therefore, it
is important that the care is taken with the surface preparation of the mould to achieve
good surface quality and assist easy demoulding. When the sand mould is used to cast
concrete, the sand particles on the mould surface may be absorbed by the fresh concrete
or attached to the surface of concrete element, which may result in poor surface finish
and difficulty in demoulding. The quality of the formed concrete surface finish mainly
depends on the type of release agent and the type of form face material (Savukaitis et
al. 2021). Thus, the selection of a proper release agent is particularly important for
casting concrete by using the proposed sand mould. In this section, the standards and
methods for quantifying concrete surface quality are introduced, and release agents are
tested to achieve an ideal concrete surface finish.
2.3.1. Surface roughness
A concrete member with a high surface quality can not only ensure the performance of
a structure, but also boost its aesthetical value. In general, concrete surface quality can
be measured by surface roughness. The average roughness (𝑅) is usually defined as
the average deviation of the profile in relation to its mean line, which is indicated by
where 𝑛 is the number of discrete measurements, and 𝑍 is the amplitude of each
measurement which indicates the distance between the specific profile point and the
mean line. Based on the roughness parameter, the surface roughness can be classified
into four categories, i.e. very smooth (𝑅 is not measurable), smooth (𝑅 1.5 mm),
rough (1.5 mm 𝑅 3 mm) and very rough (𝑅 3 mm) (Santos & Júlio 2013).
The surface roughness of concrete components can be detected by contact and non-
contact methods (Santos & Júlio 2010). For contact method, mechanical stylus is the
most commonly used device for assessing surface texture. It is composed of a stylus, a
conditioner/amplifier, a mechanical unit for advancement and a computer for data
acquisition. The surface texture is measured by dragging the probe over the surface of
an object. Laser scanning method is usually used for non-contact method, which can be
used to characterise the geometric properties of an object (Santos & Júlio 2008). In this
method, photographic images are taken from different positions and angles, and then a
3D model, comprising a dense cloud of points, of the object can be generated using
mathematical algorithms (Tonietto et al. 2019). The shape and surface roughness of the
finished concrete components could be accurately measured with this method.
In addition, the roughness of concrete surface can be qualified through visual inspection,
which is the most well-known method and is suggested by the International Concrete
Repair Institute (Millman & Giancaspro 2013). The observed concrete surface is
compared with nine standard Concrete Surface Profiles (CSP) as shown in Figure 3
(Santos & Júlio 2013). The nine concrete surfaces from CSP 1 to CSP 9 represent
varying degrees of concrete roughness, with CSP 1 generally regarded as the least rough,
CSP 4 the moderate rough, and CSP 9 the roughest. Visual inspection is fast and easy
to perform, but the results are subjected to the expertise of the inspector who performs
CSP 1 CSP 2 CSP 3
CSP 4 CSP 5 CSP 6
CSP 7 CSP 8 CSP 9
Figure 3 Nine standard concrete surface profiles (ICPI)(Santos & Júlio 2013).
2.3.2. Release agents for sand mould
When fresh concrete is cast into sand mould, loose particles on the surface of sand
mould could be absorbed into concrete or attached on the concrete surface, which not
only damages the surface quality of final concrete product, but also causes difficulties
in demoulding. Regarding this issue, both solid and liquid release agents are tested in
this study to improve the surface quality of finished concrete. To cast concrete with
curved surface using the proposed sand mould, Stretchy fabric is firstly tested. Stretchy
fabric is usually fabricated using textile sheets made of synthetic fibres, e.g. nylon,
polyesters and polypropylene (West 2016). The selection of the fibre type is dependent
on load-bearing requirements. For example, stiff fabrics like carbon fibre textile could
be used when a high resistance to deformation is required, while other fabrics such as
spandex may be stretched easily to fit a complex shape (Hawkins et al. 2016). In this
study, stretchy fabric made of spandex is used. Stretchy fabric is extensible under
pressure, which can be stretched to form a smooth film on curved surface without
wrinkling. Figure 4 shows a sand mould with spherical geometry and a finished
concrete member cast in the sand mould with stretchy fabric as release agent. With the
help of stretchy fabric, the separation of the finished concrete from the sand mould is
easy. The shape of the finished concrete part corresponds well with the shape of sand
mould, and the surface of the finished concrete aligns with CSP 1 in Figure 3. An
average surface roughness of 4.91 μm is detected using the mechanical stylus device,
which indicates that the obtained surface can be classified as smooth.
(a) sand mould (b) finished concrete member
Figure 4 Concrete surface finished with stretchy fabric.
An oil-based liquid concrete release agent is also tested in the study. This liquid release
agent could form a natural barrier on the surface of the sand mould, which would
separate fresh concrete from the sand mould. A curved concrete part is cast with this
release agent applied on the sand mould surface, and the final concrete product is shown
in Figure 5. As can be seen, the shape of the finished concrete part is well moulded by
the sand mould. As concrete is in direct contact with the sand surface, the surface
smoothness of the finished concrete part is not as good as that with solid release agent,
i.e., stretchy fabric. By performing the visual inspection and comparing against the
concrete surface profiles in Figure 3, the surface of the concrete member is found to
align with CSP 1. An average surface roughness of 12.36 μm is detected using the
mechanical stylus device, which indicates that the obtained surface can be classified as
smooth. Thus, the surface quality of the concrete cast with liquid release agent is also
(a) concrete cast on sand mould (b) finished concrete member
Figure 5 Concrete surface finished with liquid release agent.
3. Case studies and results
Topology optimisation techniques have been increasingly employed in the design of
modern concrete structures to achieve high structural efficiency and aesthetic value.
The proposed formwork system based on ice pattern and sand mould is promising for
the fabrication of these optimised concrete components. On one hand, the proposed
formwork system is eco-friendly with low fabrication cost. On the other hand, it is
flexible which could assist the fabrication of customised concrete members. The
process for the fabrication of concrete element using the proposed formwork system is
shown in Figure 6. The feasibility of the proposed formwork system is demonstrated
by the fabrication of concrete components with customised shapes, i.e., optimised wall,
tree, bridge and cantilever beam.
Figure 6 Concrete fabrication process using the proposed formwork system.
The concrete mix design used in this study is shown in Table 3. The fresh concrete has
high flowability, which can easily fill in the sand mould without vibration. The
compressive strength of the concrete is 90 MPa.
Table 3 Concrete mix design.
Material Cement Slag Silica
Sand Water Superplas
457 315 262 1049 215 21 80
3.1. Fabrication of 2D concrete components
In this section, the fabrications of an optimised concrete wall and a concrete tree are
presented to demonstrate the effectiveness of the proposed formwork system. Solid
release agent and liquid release agent are applied to concrete wall and concrete tree
respectively to test their effects on the demoulding process and the quality of resulting
3.1.1. Fabrication of a topologically optimised concrete wall with solid release
In this section, a rectangular concrete wall with 600 mm width, 400 mm height and 5
mm thickness is optimised using the Bi-directional Evolutionary Structural
Optimisation (BESO) topology optimisation method (Huang & Xie 2010). The concrete
wall is subjected to a uniformly distributed load on the top and is fixed along the bottom
as shown in Figure 7 (a). The optimisation goal is to reduce 60% of its volume and
obtain the maximum stiffness with the remaining material. After optimisation process,
the edges are smoothed, and the resulted shape (designed shape) is shown in Figure 7
(a) Initial concrete wall design setup
(b) Optimised concrete wall
Figure 7 Concrete wall design.
A sand mould with the same shape is needed
in order to cast concrete wall with the
designed shape. Therefore, an ice pattern with the designed shape is firstly fabricated
for the proposed formwork system. In this study, as the dimension of the designed shape
is within one metre, the ice pattern is manually carved with electrical drill. When
dealing with larger ice objects, digital fabrication techniques such as CNC milling and
3D printing could be employed so that the production of free-form geometries with high
accuracy is possible. The carved ice pattern with the designed shape is shown in Figure
Figure 8 Ice pattern with designed shape.
Once the ice pattern is obtained, sand is mixed with the proposed mix design by
following the recommended mixing procedure discussed in Section 2. After that, the
fresh sand is cast on the ice pattern immediately. The sand mould is then stored in an
environmental chamber at -5°C temperature for 10 h until it is hardened. Afterwards,
the sand mould is taken out from the environmental chamber and placed under the sun
for ice to melt naturally. The final sand mould is shown in Figure 9, and it is in ideal
As sand mould has already hardened before taking out from the
environmental chamber, the melting water has no impact on the quality of the sand
mould, and the mould surface is very smooth after the water is evaporated. The whole
process goes well without much operation.
(a) Before ice melting (b) After ice melting
Figure 9 Finished sand mould for optimised wall.
When the sand mould is ready for concrete casting, stretchy fabric is placed on the
surface of the sand mould. Fresh concrete is then cast into the sand mould. Figure 10
shows the concrete cast on the sand mould and the finished concrete member. The
demoulding process is very easy, and no crack or air void is observed on the concrete
surface. Several overlap marks caused by the solid release agent can be observed on the
surface of the concrete wall, but they do not influence the overall smoothness.
(a) Concrete cast on sand mould
(b) Top surface of finished concrete (c) Bottom surface of finished concrete
Figure 10 Finished concrete component.
The 3D geometry of the finished concrete component is detected by a portable laser
scanning equipment to determine the fabrication accuracy of the proposed formwork
system as shown in Figure 11. This device can be held by hand to scan the surface of
the targets no matter of their dimensions. The scanned point cloud has very high
accuracy with error up to 0.01 mm. Once the point cloud data are obtained, the finished
concrete shape can be analysed and compared with the designed shape. The shape
comparison is shown in Figure 12. It can be seen that the maximum deviation is 7.94
mm along the right boundary, and the root mean square error (RMSE) which is used to
measure the average shape error is 2.51 mm only. The results indicate that the geometry
of the finished concrete wall highly coincides with the designed shape, and the error is
mainly introduced by the hand-carved ice pattern.
In addition, in order to understand the effects of solid release agent and sand mould on
the concrete surface finish, the surface roughness of the concrete wall is detected by a
mechanical stylus device as shown in Figure 13. For each surface, ten random areas are
selected for the test, and the average roughness is calculated. The obtained average 𝑅
of the top surface is 3.66 μm, and the average 𝑅 of the bottom surface (in contact with
the sand mould) is 5.84 μm. The surface roughness results indicate that the finished
concrete surface can be regarded as smooth.
Figure 11 Laser scanning of concrete component.
Figure 12 Comparison of finished concrete with designed model.
Figure 13 Surface roughness tested by mechanical stylus device.
3.1.2. Fabrication of a tree-shape concrete component with liquid release agent
A tree-shape target is designed in this section to further demonstrate the efficiency and
accuracy of the proposed formwork system as well as the effect of liquid release agent.
The height of the designed target is 600 mm, and the width is 400 mm. The thickness
of the tree-shape component is set to be 5 mm. The ice pattern of the designed shape is
fabricated using electronic cutter and shown in Figure 14. The sand mould is fabricated
by following the same process, and the final tree-shape sand mould is shown in Figure
15. It can be seen that the sand mould is in ideal condition and the surface is smooth.
Figure 14 Designed target shape and its ice pattern.
(a) Before ice melting (b) After ice melting
Figure 15 Finished sand mould for tree-shape concrete component.
The liquid release agent is applied on the internal surface of the sand mould. Fresh
concrete is then cast in the cavity of the sand mould. The demoulding process is also
very easy, and no crack or air void is observed on the concrete surface. The finished
concrete tree is shown in Figure 16. The shape of the finished concrete is also detected
by laser scanning equipment. Surface roughness of the finished concrete tree is detected
the mechanical stylus device at ten random locations, and the average roughness
is obtained accordingly. Figure 17 shows the comparison between the finished concrete
shape and its designed model. In general, the shape of the finished concrete tree
coincides well with the designed shape. The maximum deviation is 12.60 mm, and the
RMSE for the entire concrete component is 3.81 mm. The errors mostly come from the
areas with narrow branches. This is because the ice pattern is cut a little wider than the
designed shape in these narrow areas to avoid the breaking apart of the ice pattern. The
of the top surface is 3.24 μm, and the average 𝑅
of the bottom surface (in
contact with the sand mould) is 10.56 μm. Although the surface roughness value of the
concrete cast using liquid release agent is a little higher than that with solid release
agent, the surface of the finished concrete tree is still qualified as smooth.
Figure 16 Finished concrete tree: top surface (left) and bottom surface (right).
Figure 17 Comparison of finished concrete with designed model.
3.2. Fabrication of optimised 3D concrete structures with recycled sand
In this section, compression tests on cubic sand specimens are firstly conducted. In the
mix design, the only variable is the ratio of the used sand (i.e. 0% to 100%). The other
ingredients remain the same. All specimens are mixed for 150 s at 25°C. The average
compressive strengths obtained from specimens with different used sand ratios are
shown in Figure 18. The compressive strength of each sand specimen is presented in
Table A2 in Appendix.
Figure 18 Average compressive strengths of sand specimens with different used sand ratios.
It can be seen that with the increase in the amount of used sand, the compressive
strength of sand specimen is slightly decreasing. When the mix ratio of used sand
exceeds 70%, no significant change in the compressive strength can be observed. In
addition, compared to sand specimens made fully with fresh sand, the average
compressive strength of specimens made with 100% used sand only decreases by
around 15%. This indicates that the sand mould made with 100% used sand also has
sufficient strength to be used for concrete casting. In the following sections, two
examples (optimised bridge and cantilever beam) are fabricated with fully recycled
sand to further validate the feasibility and recyclability of the proposed method.
3.2.1. Fabrication of an optimised bridge
In this example, an initial model with design domain of 600 mm × 400 mm × 200 mm
is proposed. A uniformly distributed load is applied on the top surface of this model,
while both edges of the bottom surface are fixed. The optimisation goal is to reduce 90%
of its original volume and keep the remained structure at its maximum stiffness. The
initial model setup and the optimised structure are shown in Figure 19.
Figure 19 Initial model setup (top) and optimised concrete bridge design (bottom).
During the fabrication process, the ice pattern of the same shape is made firstly. Sand
mould is then cast on the ice pattern and stored in the environmental chamber for curing.
Once the sand mould is fully cured, it is taken out of the environmental chamber for ice
to melt. Finally, fresh concrete is cast into the sand mould when it is dry. The fabrication
process of the optimised concrete bridge is illustrated in Figure 20.
Figure 20 Fabrication process of optimised concrete bridge.
When concrete is cured for two days, the sand mould is smashed into particles and the
finished concrete structure is obtained. The concrete bride can be further decorated with
white pigment paste to boost its esthetical value as shown in Figure 21.
(a) Before decorated with white pigment
(b) After decorated with white pigment
Figure 21 Finished optimised concrete bridge
3.2.2. Fabrication of an optimised concrete beam
To further demonstrate the superiority of the proposed recyclable sand mould system,
an optimised concrete beam is presented in this section. The design domain is a
cantilever beam. Its length is 600 mm, and the edge length of the squared cross section
is 200 mm. The beam is fixed at one end, and a twisting force (torque) is applied at the
other end. The optimisation goal is to reduce 80% of its original volume. The initial
model setup and the optimised concrete beam are shown in Figure 22. The fabrication
of the optimised concrete beam follows the same process as that for the optimised
concrete bridge and is illustrated in Figure 23. Figure 24 shows the finished concrete
beam and the concrete beam decorated with white pigment paste.
In the two 3D fabrication examples, the geometries of the finished 3D concrete
structures well match their designs with excellent surface quality and high esthetical
values, which further validates the effectiveness of the proposed recyclable sand mould
Figure 22 Initial model setup (top) and optimised concrete beam design (bottom).
Figure 23 Fabrication process of optimised concrete bridge.
(a) Before decorated with white pigment
(b) After decorated with white pigment
Figure 24 Finished optimised concrete beam.
With the increasing worldwide concerns on carbon footprint and its environmental
impact, the material reduction in concrete construction has become an urgent task. The
use of optimisation techniques in concrete design is becoming more and more popular
in recent years, which is because of its ability to generate efficient and aesthetical
geometries of concrete members. However, the optimised geometry is usually complex
and offers many curvatures. In contemporary constructions, the applications of these
complex designs are mainly restricted to iconic buildings due to the cost and flexibility
of formwork systems. Therefore, a recyclable and cost-effective formwork system that
can be used in the construction of customised concrete structures has become highly
desirable. The advantages of the proposed formwork system based on sand mould and
ice pattern should be highlighted. First of all, the use of ice and sand as formwork
materials could resolve the problem of material waste. They could be reused and formed
into different complex shapes after demoulding, which efficiently closes the material
loop of production. Secondly, as water and sand are easily accessible in any inhabited
area, the material cost is low. Thirdly, the proposed formwork system is flexible and the
fabrication of ice pattern with desired shapes can be achieved with the help of digital
techniques, making the production of free-form and precise geometries feasible. Digital
fabrication of ice structure has been reported in literature (Kamble et al. 2021), which
could be incorporated into the proposed formwork system for future applications. In
addition to CNC milling, 3D printing of ice may be an alternative for the fabrication of
ice pattern. In the study conducted by Barnett (Barnett 2013), a robot-assisted rapid
prototyping system was proposed for producing ice object by incremental deposition
and crystallization of water. Also, 3D printing of ice composites was proposed by Pronk
et al. (Pronk et al. 2019; Pronk et al. 2017). In their study, cellulose-reinforced ice
composite was 3D printed by using a mixture of water with the additions of Guar Gum,
Xanthan Gum and cellulose fibres. Finally, as ice naturally melts away, the effort
involved in separation process is eliminated.
The ecological value of the ice pattern should also be noticed. Compared to traditional
wood pattern used in metal forming industry, the fabrication of ice pattern could result
in less carbon footprint. In wood products sector, manufacturing 1 m3 of plywood would
generate more than 600 kg CO2eq emissions (Kutnar & Hill 2014). As for ice pattern,
the carbon footprint could come from the refrigeration process where electricity is
consumed. According to the study reported by Sitnikov (Sitnikov 2019), producing 1
m3 of ice (-10°C) at ambient air of 38°C (worst scenario) would require 30.5 kWh
energy. Life cycle greenhouse gas emissions generated from non-renewable power
(fossil fuels) could range from 480 g to 1000 g CO2eq/kWh, while the emissions could
drop significantly with renewable power (Levasseur et al. 2021). This indicates that the
possible carbon footprint from the fabrication of 1 m3 ice pattern is less than 30 kg
CO2eq, which is much less compared to the fabrication of 1 m3 plywood pattern. In fact,
the energy required in this study for making ice pattern would be even less as ice pattern
is fabricated at -5°C. Moreover, the carbon footprint could be further eliminated as ice
pattern could be manufactured in winter or cold areas where refrigeration process could
be completed in nature. In addition, potential methods could be explored to increase the
melting point of ice pattern. For example, heavy water (D2O) has a higher freezing point
compared to normal water. Freezing point can be increased by using a certain
proportion of heavy water in making ice pattern, but this is not economical at present.
Further investigation shall be conducted towards reducing the environmental constraint
to the ice pattern.
When resin bonded sand is used as mould, one of the critical issues is that the loose
sand particles from sand mould could be absorbed into fresh concrete. This not only
causes difficulties in demoulding process, but also results in poor concrete surface
finish. To address this issue, both solid and liquid release agents have been tested in this
study. It is found that stretchy fabric is suitable for casting on curved surfaces. For more
complex designed shapes, liquid release agent can be applied. The release agents could
be used alone or concurrently, depending on different surface conditions. In the present
study, organic binder is used for making sand mould as high compressive strength and
good surface smoothness can be easily achieved, and the recycled sand mould material
can be used in new projects. To further improve the eco-friendliness of the proposed
formwork system, the inorganic binders such as clay can be investigated in the future
In addition, during concrete casting, although vibration is not performed, no crack or
air void is observed on the surface of the finished concrete components, which might
be attributed to the porous nature of the sand mould so that the entrapped air in concrete
mix could move through the porous sand particles to the ambient air. This could
significantly improve the surface quality of concrete components and minimize the
production costs on the repair of surface imperfections. In the two examples shown in
Section 3.1, the overall fabrication process is simple without many operations. The
shapes of the finished concrete components match very well with the designed shapes
with minor errors. In the study conducted by (Lim et al. 2020), wooden moulds were
fabricated into dome and saddle shapes, and the average errors were 3.4 mm and
5.9 mm respectively. In this study, the average deviations for concrete wall and tree
components are 2.51 mm and 3.81 mm respectively. The errors in the present study are
mainly from the manual fabrication of the ice patterns, ranging from 0 to 10 mm at
different locations, which could be easily eliminated by using digital fabrication method.
Based on the test results for the surface roughness of both concrete components, it is
safe to conclude that the smoothness of concrete components cast from sand mould is
highly desirable. In Section 3.2, the use of recycled sand mould is validated. Two
optimised 3D structures are fabricated with satisfactory surface conditions and
geometries. Therefore, the proposed formwork system has been demonstrated to be
feasible and effective. With the characteristics of low cost, eco-friendliness and full
recyclability, the proposed formwork system is promising for large-scale applications
in the future.
In this paper, an innovative formwork system based on ice and sand for concrete casting
is proposed. The new formwork system is eco-friendly with minimum material waste
and construction cost, which could be used for the fabrication of customised concrete
components with complex geometries. The required sand mould strength and surface
quality of the final concrete products are tested and discussed in detail. The application
of the proposed formwork system and its fabrication process are demonstrated by a
series of concrete casting examples. The conclusions drawn from the present study are
A mix design and a mixing procedure of the proposed sand mould are developed
and tested. The compressive strength of the sand mould mixed at -5°C for less than
300 s is 0.651 MPa which is sufficient to bear the load from concrete casting.
To achieve easy demoulding and obtain high quality concrete surface, stretchy
fabric is used as a solid release agent, which is found to be suitable for the proposed
sand mould. Liquid release agent can be applied to the sand mould with more
complex geometries to achieve satisfactory results.
The feasibility and effectiveness of the proposed formwork system are
demonstrated by two 2D fabrication cases, i.e., optimised concrete wall and
concrete tree. The geometries of the finished concrete components match well with
their designed shapes, with average deviations of 2.51 mm and 3.81 mm
respectively. The surface roughness values of the top and bottom surfaces of the
optimised concrete wall are 3.66 μm and 5.84 μm respectively, and those for the
concrete tree are 3.24 μm and 10.56 μm respectively, which indicate the finished
concrete components are in excellent surface conditions.
The compressive strength of the fully recycled sand only decreases by about 15%
compared to the new sand, which indicates that the proposed sand mould can be
fully recycled. The recyclability of the proposed sand mould is further validated by
the fabrication of two complex optimised 3D concrete structures.
Although the proposed formwork system is currently tested at lab-scale, it has great
potential to be applied in larger structures with complex designs, such as double
curved surfaces. In addition, digital fabrication methods can be adopted to further
improve the accuracy and automation of making ice patterns.
This work was supported by the Australian Research Council (FL190100014) and
the China Scholarship Council (201706440017).
Brown, J 2000,
Foseco ferrous foundryman's handbook
Czarnecki, S & Sadowski, Ł 2022, 'Morphological properties of the cement skin: Understanding the
effect of contact with formwork',
Case Studies in Construction Materials
, vol. 16, p. e01007.
Dombernowsky, P & Søndergaard, A 2012, 'Design, analysis and realisation of topology optimized
Journal of the International Association for Shell and Spatial Structures
53, no. 4, pp. 209-216.
Gaddam, MS & Achuthan, A 2020, 'A comparative study on newly emerging type of formwork
systems with conventional type of form work systems',
Materials Today: Proceedings
, vol. 33, pp.
Hawkins, WJ, Herrmann, M, Ibell, TJ, Kromoser, B, Michaelski, A, Orr, JJ, Pedreschi, R, Pronk, A,
Schipper, HR & Shepherd, P 2016, 'Flexible formwork technologies–a state of the art review',
, vol. 17, no. 6, pp. 911-935.
Huang, R, Zhang, B & Tang, Y 2016, 'Application conditions for ester cured alkaline phenolic resin
, vol. 13, no. 4, pp. 231-237.
Huang, X & Xie, YM 2010,
Evolutionary topology optimization of continuum structures: methods
, John Wiley & Sons.
Ka tzer, J & Sz atkiewicz , T 2019, 'Pro perties of co ncrete ele ments with 3-D printed formworks which
substitute steel reinforcement',
Construction and Building Materials
, vol. 210, pp. 157-161.
Krawczyńska-Piechna, A 2017, 'Comprehensive Approach to Efficient Planning of Formwork
Utilization on the Construction Site',
, vol. 182, pp. 366-372.
Kutnar, A & Hill, C 2014, 'Assessment of carbon footprinting in the wood industry', in
of Carbon Footprint in Different Industrial Sectors, Volume 2
, Springer, pp. 135-172.
Levasseur, A, Mercier-Blais, S, Prairie, Y, Tremblay, A & Turpin, C 2021, 'Improving the accuracy of
electricity carbon footprint: Estimation of hydroelectric reservoir greenhouse gas emissions',
Renewable and Sustainable Energy Reviews
, vol. 136, p. 110433.
Li, W, Lin, X, Bao, DW & Xie, YM 2022, 'A review of formwork systems for modern concrete
, vol. 38, pp. 52-63.
Lim, JH, Weng, Y & Pham, Q-C 2020, '3D printing of curved concrete surfaces using Adaptable
Construction and Building Materials
, vol. 232, p. 117075.
Menna, C, Mata-Falcón, J, Bos, FP, Vantyghem, G, Ferrara, L, Asprone, D, Salet, T & Kaufmann, W
2020, 'Opportunities and challenges for structural engineering of digitally fabricated concrete',
Cement and Concrete Research
, vol. 133, p. 106079.
Millman, LR & Giancaspro, JW 2013, 'Three-dimensional optical profilometry analysis of the
international concrete repair institute concrete surface profiles (CSPs)',
ACI Materials Journal
110, no. 5, p. 519.
Sadarang, J & Nayak, RK 2021, 'Utilization of fly ash as an alternative to silica sand for green sand
mould casting process',
Journal of Manufacturing Processes
, vol. 68, pp. 1553-1561.
Santos, P & Júlio, E 2008, 'Development of a laser roughness analyser to predict in situ the bond
strength of concrete-to-concrete interfaces',
Magazine of Concrete Research
, vol. 60, no. 5, pp.
Santos, P & Júlio, E 2010, 'Comparison of Methods for Texture Assessment of Concrete Surfaces',
ACI Materials Journal
, vol. 107, pp. 433-440.
Santos, PM & Júlio, EN 2013, 'A state-of-the-art review on roughness quantification methods for
Construction and Building Materials
, vol. 38, pp. 912-923.
Savukaitis, G, Daukšys, M, Juociūnas, S, Grinys, A & Kriptavicius, D 2021, 'The influence of new and
used formwork coated with different release agents on the appearance of the formed concrete
Journal of Building Engineering
, vol., p. 102807.
Sitnikov 2018, 'Ice formwork for ultra-high performance concrete: simulation of ice melting
Humanizing Digital Reality
, pp. 395-406.
Sitnikov 2019, 'Ice formwork for high-performance concrete: a model of lean production for
prefabricated concrete industry',
, vol. 18, pp. 109-116.
Tonietto, L, Gonzaga, L, Veronez, MR, de Souza Kazmierczak, C, Arnold, DCM & da Costa, CA 2019,
'New method for evaluating surface roughness parameters acquired by laser scanning',
, vol. 9, no. 1, pp. 1-16.
Upadhyay, M, Sivarupan, T & El Mansori, M 2017, '3D printing for rapid sand casting—A review',
Journal of Manufacturing Processes
, vol. 29, pp. 211-220.
West, M 2016,
The fabric formwork book: Methods for building new architectural and structural
forms in concrete
To study the effects of temperature and mixing time on the final quality of the hardened
sand specimen, two environmental temperatures (-5°C and 25°C) and three mixing
times (150 s, 300 s and 450 s) are tested. For each group, three sand specimens are cast
under the same condition, e.g., 1-1,1-2 and 1-3 for specimens cast at -5°C for 150 s.
The test results of all specimens are summarized in Table A1.
Table A1 Compressive strengths of sand mould specimens.
1-1 -5 150 1762 0.705
1-2 -5 150 1815 0.727
1-3 -5 150 1892 0.757
2-1 -5 300 1621 0.650
2-2 -5 300 1683 0.674
2-3 -5 300 1574 0.630
3-1 -5 450 378 0.151
3-2 -5 450 417 0.167
3-3 -5 450 392 0.157
4-1 25 150 3959 1.584
4-2 25 150 3844 1.538
4-3 25 150 3980 1.592
5-1 25 300 98 0.040
5-2 25 300 106 0.043
5-3 25 300 91 0.037
6-1 25 450 N/A N/A
6-2 25 450 N/A N/A
6-3 25 450 N/A N/A
To investigate the influence of the used sand on the sand mould strength, sand
specimens with six different used sand ratios (i.e., 0% to 100%) are tested.
For each group, three sand specimens are cast. For example, 7-1,7-2 and 7-3 are for the
specimens with 0% used sand. All specimens are mixed for 150 s at 25°C, and the other
ingredients remain the same. The test results of all specimens are summarized in Table
Table A2 Compressive strengths of specimens with different used sand ratios.
Specimen No. Used sand ratio Maximum
7-1 0% used sand 3959 1.584
7-2 0% used sand 3844 1.538
7-3 0% used sand 3980 1.592
8-1 60% used sand 3714 1.486
8-2 60% used sand 3813 1.526
8-3 60% used sand 3777 1.511
9-1 70% used sand 3305 1.322
9-2 70% used sand 3514 1.406
9-3 70% used sand 3469 1.388
10-1 80% used sand 3256 1.302
10-2 80% used sand 3328 1.331
10-3 80% used sand 3362 1.345
11-1 90% used sand 3204 1.282
11-2 90% used sand 3270 1.308
11-3 90% used sand 3325 1.330
12-1 100% used sand 3375 1.350
12-2 100% used sand 3309 1.324
12-3 100% used sand 3300 1.320
About the authors
Wei Li is a PhD candidate in the Centre for Innovative Structures and Materials (CISM)
at RMIT University, Australia. He obtained his BEng Degree in mechanical engineering
from South China University of Technology, China. His research focuses on recyclable
formwork design, topology optimisation, concrete construction, and finite element
Xiaoshan Lin is a Senior Lecturer in Civil and Infrastructure Engineering discipline in
the School of Engineering at RMIT University, Australia. She received her PhD degree
in Civil Engineering from the University of New South Wales. Dr Lin’s areas of
expertise include finite element development for accurate and efficient numerical
simulation, high performance reinforced concrete and composite materials, and
structural analysis under extreme conditions. Xiaoshan Lin is the corresponding author
and can be contacted at: firstname.lastname@example.org.
Yi Min Xie is an Australian Laureate Fellow, Distinguished Professor and Director of
the Centre for Innovative Structures and Materials (CISM) at RMIT University,
Australia. He is a Fellow of the Australian Academy of Technology and Engineering.
He obtained his B.Eng. degree from Shanghai Jiao Tong University, China and his PhD
from Swansea University, UK. His research focuses on developing new techniques for
designing and manufacturing innovative structures and materials.