Virtual and Physical Prototyping

ISSN: (Print) (Online) Journal homepage: https://www.tandfonline.com/loi/nvpp20

Design of a 3D printed concrete bridge by testing

Theo A. M. Salet, Zeeshan Y. Ahmed, Freek P. Bos & Hans L. M. Laagland

To cite this article: Theo A. M. Salet, Zeeshan Y. Ahmed, Freek P. Bos & Hans L. M. Laagland
(2018) Design of a 3D printed concrete bridge by testing, Virtual and Physical Prototyping, 13:3,
222-236, DOI: 10.1080/17452759.2018.1476064

To link to this article: https://doi.org/10.1080/17452759.2018.1476064

© 2018 The Author(s). Published by Informa

UK Limited, trading as Taylor & Francis
Group

Published online: 31 May 2018.

Submit your article to this journal

Article views: 13226

View related articles

View Crossmark data

Citing articles: 110 View citing articles

Full Terms & Conditions of access and use can be found at

https://www.tandfonline.com/action/journalInformation?journalCode=nvpp20
VIRTUAL AND PHYSICAL PROTOTYPING
2018, VOL. 13, NO. 3, 222–236
https://doi.org/10.1080/17452759.2018.1476064

Design of a 3D printed concrete bridge by testing*

Theo A. M. Saleta, Zeeshan Y. Ahmeda, Freek P. Bos a
and Hans L. M. Laaglandb
a
Department of the Built Environment, Eindhoven University of Technology, Eindhoven, Netherlands; bWitteveen+Bos, Deventer, Netherlands

ABSTRACT ARTICLE HISTORY

The current state of research and development into the additive manufacturing of concrete is Received 22 April 2018
poised to become a disruptive technology in the construction industry. Although many Accepted 11 May 2018
academic and industrial institutions have successfully realised full-scale structures, the limitations
KEYWORDS
in the current codes of practice to evaluate their structural integrity have resulted in most of 3D printing; concrete;
these structures still not being certified as safe for public utilisation and thus deemed as test prestress; reinforcement;
prototypes for display purpose only. To realise a 3D concrete printed (3DCP) structure which bridge
could be certified safe for public use, a bridge was realised using the print facility of the
Eindhoven University of Technology (TU/e) based on the concept of ‘Design by Testing’. This
paper holistically discusses the complications encountered while realising a reinforced 3DCP
bridge in a public traffic network and decisions taken to find solutions for overcoming them.

Introduction
A promising development from this perspective is a
The built environment worldwide is facing major chal- global shift from in situ construction towards Design
lenges for the decades ahead. A global migration for Manufacturing and Assembling. The controlled
towards cities, energy transition, smart mobility, a environment of a precast factory simplifies automation
growing lack of natural resources, climate change, an such as the use of robots. A next step is not to pour con-
ageing population and rapidly changing urban econom- crete but instead apply additive manufacturing tech-
ies from industry and services towards knowledge and niques and print concrete to be safe on natural
creativity are just examples of issue matters that will resources and decrease weight. In addition, 3D Concrete
have an impact on our buildings and infrastructure. Printing (3DCP) offers the benefit of mass customisation.
This includes both new cities as well as urban transform- The technology of 3DCP was introduced about two
ations. The pace of the developments puts a pressure on decades ago (Khoshnevis and Dutton 1998, Khoshnevis,
the building industry to innovate and ensure that the 2004, Khoshnevis et al. 2001, 2006). Ever since, ground-
construction productivity meets the new demand, costs breaking projects and case studies are being presented
are in control and all are in a sustainable manner. on a regular basis to showcase the potential of digital
The digital disruption that already has affected most fabrication with concrete. On a trial-and-error basis, the
industries could easily become a game changer to frontiers of the new technology are rapidly being
meet future demands in the building and civil industry explored, including buildings (3ders 2015, 2016, 2017a,
as well. Design skills and workmanship become rapidly 2017b, 2017c, Cnet 2016, Mediaoffice 2016, cementon-
supported by data-driven algorithms, and the design line 2017, De Ingenieur 2017). This paper is an elabor-
information is getting stored systematically in Building ation of Salet et al. (2018) and extensively describes the
Information Models. The digital storage obviously design, manufacturing and assembling of a 3D Printed
enhances the collaboration among the many design dis- Concrete bridge.
ciplines and reduces costs of failure. However, the use of Despite the capability to actually print concrete, the
the same design data for the actual construction, without design and manufacturing of a bridge was a challenge
human intervention, is of more recent date and is still in a on its own for more than one reason. In the first place,
process of becoming matured. codes to test the structural properties of printed concrete

CONTACT Theo A. M. Salet [email protected] Department of the Built Environment, Eindhoven University of Technology, Groene Loper 6, Eindhoven
5612 AZ, Netherlands
*Chua, C.K., Yeong, W.Y., Tan, M.J., Liu, E.J., and Tor, S.B., eds. 3rd International Conference on Progress in Additive Manufacturing (Pro-AM 2018), 14–17 May 2018,
Singapore.
© 2018 The Author(s). Published by Informa UK Limited, trading as Taylor & Francis Group
This is an Open Access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivatives License (http://creativecommons.org/licenses/by-nc-nd/
4.0/), which permits non-commercial re-use, distribution, and reproduction in any medium, provided the original work is properly cited, and is not altered, transformed, or built upon in
any way.
 VIRTUAL AND PHYSICAL PROTOTYPING 223

are not available yet. The NEN-EN-206-1 tests vibrated facility has been described by Bos et al. (2016). Despite
concrete in cylinders and are not suitable for printed the fact that the full bridge coincidentally fits in the
material that cannot be vibrated. This issue has been current print box, it was decided not to print it in one
overcome by performing material tests on printed con- piece from the bottom to the top of the deck, but
crete. Structural calculations have been made of the instead in multiple elements that need to be assembled.
bridge, based on these tested materials properties, The reason is to benefit from the printer to optimise the
using existing codes (NEN-EN-1992-1-1) on prestressing. cross-section of the bridge. The bridge consists of
However, the application of these codes on printed con- printed elements that are rotated 90° after printing,
crete structures has not been validated yet. To overcome and then pressed together by post-tensioned prestres-
this issue, a scale model of the bridge was tested structu- sing tendons. This allows for an optimised section
rally in the laboratory to prove on safety and compliance design composed out of splines. Figure 1(a,b) illustrates
with the codes. Finally, the actual bridge was tested on- the design concept.
site again prior to opening for public usage last year on Once ready, an element is released from the print bed
the load-bearing capacity. This final test was done to and flipped into its final orientation. This means that the
exclude any failure during the production since the height of the printed element is related to the span of
print process is not certified yet. In addition to these the bridge and the printer. The printer is capable to
structural tests, various other tests have been performed print 2.8-m-high elements, meaning at least three
to find the best print strategy and the best possible way elements need to be printed and assembled to achieve
to assemble the printed elements. All of these tests the 6.5 m span.
meant that the design of this bridge can be best The need to print and stack several single elements
described as learning on the job and is referred to as with a hollow core by itself makes it self-evident to pres-
design by testing. tress the components to overcome the lack of bending
moment resistant of unreinforced concrete, a technology
commonly used in the construction industry. So, pres-
Specifications and conceptual design
tress tendons are placed in the openings of the printed
The bridge is part of the renewal of an existing bicycle elements, stressed, anchored at the beginning and
track named Lieve Vrouwensteeg in the village Gemert the end, and finally released. The printed concrete
in the Netherlands. It crosses a small local canal called elements are stressed to a level that only compression
the Peelse Loop. The span of the bridge is 6.5 m and remains in the section, and no additional passive
the width 3.5 m. The uniformly distributed design load reinforcement in that direction is required. The main
(qEd) is 5.0 kN/m2. The realisation of the printed bridge advantage of this approach is that the lack of tensile
was treated as a living lab within the overall project, strength of concrete is overcome, without a need of
involving the triple helix of government, industry and a reinforcement. Furthermore, the uncertainty with
knowledge institute (Eindhoven University of Technol- regard to the tensile capacity of the printed concrete
ogy, TU/e) from the start. with all its interfaces, which has been shown by several
The bridge has been designed, starting with the capa- researchers to depend on the interface interval time
bilities of the 3D concrete printer at TU/e in mind. The and may be less than the bulk material tensile strength

Figure 1. (a, b) 3DCP bicycle bridge conceptual design (illustration by BAM).

224 T. A. M. SALET ET AL.

Figure 2. Principle of prestress applied to a structure loaded in bending. (a): no prestress, only bending; (b): prestress and bending: the
tensile stress in the bottom of the structure is eliminated.

(Le et al., 2012, Zareiyan & Khoshnevis 2017, Panda et al. designations (C30/37, etc.). Extensive published and
2018), becomes irrelevant. The full cross-section is unpublished research were therefore carried out by the
stressed in compression only at the design load. The TU/e and the material supplier into compressive, flexural
principle is illustrated in Figure 2. and uniaxial tensile strength, modulus of elasticity,
It is noted that, even with concrete as a material, it is density and directional dependency (Doomen, 2016,
possible to print some cantilevering and realise a unique Slager, 2017).
shape and some surface texture. However, the cantilever- In view of the prestress force, the compressive
ing is limited due to the nature of the material that will strength was particularly important for this project,
harden only after about 2 hours after deposition. The since the bending moment resistance (MRd) is ruled by
limits can be stretched by applying a fast hardening type the prestress force. To resist the external shear force,
of concrete. However, this has been disregarded in this the shear resistance of the concrete cross-section (VRd,c)
project. The focus was to realise a structurally safe needs to be known. The design features embedded
bridge. The architectural design can certainly be improved reinforcement cables (see further discussion below)
in the future. For the same reason, it was decided not to that might act as stirrups, but their presence is not con-
integrate a parapet in the printed bridge part yet. sidered in the shear resistance yet, as their performance
The cross-section of the bridge elements consists of a in terms of shear resistance has not been established.
series of connected bottle shapes, alternatively posi- The shear strength of the concrete is not determined
tioned upside down, in combination with a continuous in a direct shear test, but instead derived from the
connecting straight line at the bottom. This shape pro- tensile and compression strength, based on a Mohr–
vides a sufficient amount of deposited material to resist Coulomb failure criterion.
both the bending moment and the shear force. At the It is noticed that the strength of printed concrete,
front and at the end of the printed bridge deck elements, different from poured concrete, might differ based on
two solid concrete bulkheads are added to introduce the the test direction compared to the print direction. In
prestress forces. These have been traditionally cast and view of this, tests have been performed in three indepen-
reinforced to overcome the spitting tensile forces that dently perpendicular directions.
are typical for the local anchoring of the prestress Apart from the strength, also the stiffness of the con-
tendons. The bridge rests, with the two bulkheads, on crete must be known. The structure shortens under pres-
two traditional abutments with a pile foundation due to tress, which results in prestress loss. The amount of
the soft soil conditions. The steel parapet spans the full shortening depends on the stiffness of the section, and
length of the bridge and is supported, independently thus on the section area and modulus of elasticity. The
from the printed bridge, on the foundation. lower the stiffness, the more initial prestress is needed
to actually achieve the required final prestress level in
the concrete. However, the short-term stiffness is not
Materials testing
enough. Concrete, like many other materials, creeps
To be able to develop the conceptual design into a final under a permanent load and also shrinks due to
design, the (structural) material properties of the applied drying. This is expected to be more than normal concrete
print mortar Weber 3D 115-1, globally described by Bos because of a lack of coarse aggregates due to limitations
et al. (2016) (and since been given its trade name), had of the rotor stator pump that is used. Shrinkage and
to be established. Obviously, this custom designed creep gradually reduce the prestress load over time
material does not comply with standard concrete and must therefore be known and compensated for.
 VIRTUAL AND PHYSICAL PROTOTYPING 225

Table 1. Structural properties of Weber 3D 115-1 print mortar, as used in the structural design of the bridge.
Property Dir. Age Symbol Value
Density 28 days Ρ 2,000 kg/m3
Modulus of elasticity 28 days E 19,000 MPa
Average compressive strength u 28 days fck,u 23.2 MPa
v 28 days fck,v 21.5 MPa
w 28 days fck,w 21.0 MPa
Average tensile strength (also used for flexural tension) u 28 days fck,u 1.9 MPa
v 28 days fck,v 1.6 MPa
w 28 days fck,w 1.3 MPa
Creep factora 7 days w7 1.0
14 days w14 2.5
56 days w56 3.0
Shrinkage 7 days ε7 0.6
14 days ε14 1.2
56 days ε56 1.5
a
After 28 days.
Notes: For the directional dependency, a relative orientation of axis u, v, w is used (Bos et al., 2016), indicating the direction parallel to the print in the horizontal
plane, perpendicular to the print direction in the horizontal plane, and vertically perpendicular to the print direction (or parallel to the robot arm), respectively.

The material properties applied to the structural design to apply the load to the samples which were stored in
of the bridge are summarised in Table 1. a climate-controlled chamber. The load was applied at
The compressive strength and modulus of elasticity a concrete age of 14 days to correspond with the
have been determined by subjecting 40 × 40 × 40 mm expected age of application of the prestress on the
printed cubes consisting of four layers of 40 × 10 mm bridge elements. The deformation was regularly
to a displacement-controlled compression test based measured along four sides of the test samples and aver-
on the EN 196 series, in three different directions aged to determine the creep and shrinkage deformation.
(Figure 3(a)). The tensile strength was determined in a
hinged set-up shown in Figure 3(b). The test was per-
Final design
formed on 40 × 40 × 40 mm printed cubes in three direc-
tions, similar to those used for the compression tests. The final cross-section of the bridge is 3440 × 920 mm.
However, a 6-mm radius notch was applied on two Structural calculations showed a need for sufficient
opposing sides to ensure failure in the middle of the shear capacity due to the relatively low (tensile) strength
specimen, which for the w-direction corresponds with of the concrete that was used. This resulted in a typical
the layer interface. Creep and shrinkage, finally, were design of the cross-section as shown in Figure 4(a,b).
determined by subjecting printed samples to a constant The design in Figure 4(a) was used for the 1:2 scale test
load of approximately 30–35% of the compressive on the bridge (discussed below). Figure 4(b) shows the
strength for 90 days (Figure 3(c)) and comparing them final design, which was an optimised version of the
to the deformation in unloaded samples. The former design used for the test as it (relatively) reduced the
records the deformation caused by creep and shrinkage print path length (and thus the printing time and
combined, while the latter only records deformation material use) by about 4%.
from shrinkage. The creep factor can be determined The cross-section exists of a (curved) top and (straight)
from the difference between both. A spring was used bottom flange that combined with the prestress force

Figure 3. Test set-up of a compression test (a), a direct tensile test (b) and a creep/shrinkage test (c) used to determine structural
properties of the print mortar.
226 T. A. M. SALET ET AL.

Figure 4. Print paths of the 1:2 scale model for testing (a) and the actual bridge section (b). The optimised pattern of the latter with
regard to the former saves 4% of print path length.

provides the bending moment resistance. In between requirements. The elements could not exceed the
the flanges is a bottle-shaped type of pattern to take lifting capacity of the crane in the laboratory (5 tons) to
care of the shear force. The concrete filament width in remove the elements from the print bed. In total, the
this design is 60 mm. The standard TU/e 3DCP system bridge features 535 printed layers, with a length of 25.1
operates with a downward printing 10 × 40 mm2 m each, and a total print path length of 13.4 km.
nozzle opening, but for this project, a specific nozzle The prestress is applied by post-tensioning 16
was custom designed (see further discussion below). In Dywidag-system tendons to an initial load P0 of 150 kN.
the scaled element, yet another size was used, namely The prestress load has been determined so that the
of 30 mm filament. These differences could be achieved entire section will always remain in compression under
by changing the nozzle, print speed and pump pressure the design load, even in the ultimate limit state. The
in the print system. It requires some trial-and-error centre of gravity of the tendons is slightly below the
testing to obtain the correct settings for each case. centre of gravity of the concrete section to compensate
The height of each element is approximately 1.08 m, for effects of the self-weight. The tendons can be ten-
or 89–90 layers of 12 mm. This results in six elements sioned again in the future, if creep and shrinkage
to compose the bridge. The element height was exceed the considered values. Thus, the required level
limited due to handling and transportation of prestress can always be maintained.

Figure 5. Finite element analysis of the 3DCP bicycle bridge: stress distribution under variable load (a) and under prestress load (b)
(illustrations provided by Witteveen+Bos).
 VIRTUAL AND PHYSICAL PROTOTYPING 227

2017b). Apart from acting as a reinforcement, the wire

also serves a confinement, as demonstrated in Figure 8.
The figure shows a small cylinder that is printed in con-
crete with the embedded wire. The cylinder has been
stressed in compression in the (vertical) axis, perpendicu-
lar to the filaments, like the situation at the prestressed
bridge. As shown, the concrete from the cover eventually
spalls off, but the concrete at the inside the reinforce-
ment still remains one solid piece due to the continuous
wire. The wire prestresses the element in compression
transverse to the load due to the restrained lateral
deformation.

Figure 6. Positioning of prestress tendons and reinforcement Printing

cable in the bridge.
Print path

The final design was structurally calculated in more In the original design, on which the 1:2 scale test was
detail by means of finite element analysis (Figure 5(a, based, the overall length of the print path for each
b)). From the analysis, it was noticed that the bridge layer was 26.0 m. For the final production design, the
has a relatively large width to span ratio of about 1:2. path length was optimised to 25.1 m (Figure 2). The
This ratio makes the bridge sensitive for torsion that advantage of the design is that the print path could be
can occur if the two abutments are not fully aligned programmed in such a manner that the start and the
with the bridge itself, e.g. due to settlements. To avoid end locations of each layer are at the same position.
brittle failure in that case, an automatically entrained The printing of the bridge is shown in Figure 9.
reinforcement cable was introduced in the filament. To
limit the required cable length, only specific sections of
Nozzle shape and concrete flow direction
the elements were reinforced (Figure 6). This way,
these sections act similar to shear reinforcement stirrups The type of nozzle for the project was not self-evident
used in conventional structural concrete. and studied upfront. A down-flow nozzle (Figure 10(a,
The reinforcement consists of a thin, high-strength d)) is commonly used for printing concrete. But for this
steel cable that becomes embedded in the concrete fila- project, the nozzle had to be adjusted to embed the
ments during printing with the Reinforcement Embed- reinforcement cable. Initially, a cable was introduced in
ding Device (Figure 7(a,b)) developed at TU/e. This a down-flow printing nozzle. However, in this set-up,
concept has been described by Bos et al. (2017a, the cable cuts the concrete filament in two during

Figure 7. (a, b) Reinforcement Entraining Device, developed by the TU/e.

228 T. A. M. SALET ET AL.

printing, resulting in a poor bond between the concrete

and cable.
Therefore, a back-flow nozzle was developed (Figure 10
(b,e)). The cable wire is again from the front and joins the
concrete that is coming from the top. The filament of con-
crete with the cable leaves the nozzle from the back. The
back-flow nozzle allows for the entrainment of the cable
into the printed layer and does not split the layer into
two. The back-flow concrete deposition helps with the
cable placement as the angle of concrete flow and the
movement of the cable direction are the same.
However, it was discovered that an improper design of
a back-flow nozzle can have a disastrous effect on the
interface bond between the filament layers, due to the
concrete being placed without any fluidic pressure.
Figure 11 illustrates that the bond between the layers
placed by the back-flow nozzle is so poor that the layers
can be lifted from each other effortlessly. In addition,
the side walls of the back-flow nozzle take material
away from the adjacent printed layer and hence the
bond between the adjacent in the x–y plane is also poor.
This anomaly was corrected by making a hybrid of the
Figure 8. Confinement of failed printed concrete column with down-flow and back-flow nozzle, where the bottom
cable reinforcement. surface of the back-flow nozzle was removed making it

Figure 9. (a, b, c) Printing of an element for the actual bridge.

 VIRTUAL AND PHYSICAL PROTOTYPING 229

Figure 10. Nozzle design and concrete flow. When a down-flow printing nozzle (a, d) is equipped with a reinforcement cable, the
filament is cut and the cable–concrete bond is poor. This may be solved with a back-flow nozzle (b, e). However, an improper
design of the geometry at the nozzle mouth can have a disastrous effect on the interface bond between concrete layers. An improved
hybrid down/back-flow nozzle design (c, f) solves this issue

a down-flow nozzle, which improved the bond between To achieve a print filament of proper consistency, the fila-
two consecutively placed layers in the z-axis (Figure 10(c, ment section size, print speed, and pump pressure (and
f)). However, the front wall surface of the back-flow related material flow) need to be attuned carefully, as
nozzle was kept intact for placing the reinforcement they are directly related. It was decided to maintain the
tube through which the reinforcement cable was pump pressure that is normally used as changing it
entrained into the printed layer. Also, to improve the would influence several parameters of which the influ-
bond between two layers in the x–y plane, the edge of ence is difficult to oversee, such as the material rheology
the front surface of the nozzle was removed so that and internal friction of the system. The pump pressure is
the material from the already printed layer is not controlled by the pump motor frequency, which was set
removed while printing the adjacent layer. at 150 rpm. The default print speed of the TU/e set-up is
100 mm/s. In order to obtain a comparable filament con-
sistency, the speed was reduced by 20% in this project to
Filament size, pump pressure and print speed
80 mm/s. Even though this meant a significant increase
For this project, a print nozzle with a mouth opening of printing time, this was preferred to altering the
larger than that of the standard TU/e set-up was used. pump pressure.
230 T. A. M. SALET ET AL.

a switch was made to the other system to continue print-

ing within minutes, so that dry joints could be avoided.
The duration of 2 hours is based on the open time of
about 3 hours of the concrete. After that time period,
the materials inside the hose and parts of the pump
start hardening. This results in increasing the tempera-
ture and print pressure which affect the quality of the
printed filaments and in extreme cases can lead to
pump failure. Changing and cleaning the wet system
avoided pieces of hardening concrete damaging the
system or decreasing the print material quality.

Destructive 1:2 scale test

One could argue if only the testing of the printed con-
Figure 11. Bad layer adhesion resulting from the use of the print crete material satisfies the needs to be in compliance
head from Figure 10(b).
with structural codes, since the prestress part is not
different from the poured concrete. However, since this
Further consideration is needed to be given to the is the first time a concrete bridge was constructed this
radii of the curves (Ahmed et al., 2016). Depending on way, it was decided to realise a 1:2 scale model and
the sharpness of the curve, the print speed decreases test the overall structural performance to prove the
temporarily owing to the robot kinematics. But if the structural integrity. The assembly of the scale model is
pump speed remains constant, the material gets shown in Figure 12(a–c).
unevenly distributed: more material is deposited at the The dimensions of the printed elements were scaled
angles, resulting in inconsistent layer dimensions. Fur- down to half the design size: i.e. 1720 mm in width,
thermore, in any corner, more concrete is deposited in 500 mm in length and 460 mm in height. To achieve
the inside of the corner than on the outside due to the that the dimensions of the printed filaments were also
difference in path length. This causes a difference in scaled down in width, the nozzle dimension was scaled
the vertical height (z-direction) of the filaments and down to 30 mm wide. However, the height of the
eventually may cause a failure of the object during print- nozzle was kept the same, and the number of layers
ing. This limitation has been investigated experimentally for each element was adjusted accordingly.
in the past and has been used as design rules for the The scaled bridge was loaded with three tendons
bridge. Also, smoothing the curvature of the angles era- Dywidag tendons (Ø15.7 mm, Y1860 steel grade) top
dicated the issue of varying filament dimensions. and six tendons of the same type at the bottom. The
Finally, a single layer took around 5 minutes and 20 tendons were stressed to a load of an initial load P0 =
seconds to print; it took 48 hours to complete printing 120 kN. One tendon at both the top and the bottom
of all the bridge elements. Two full wet systems, includ- were attached to a load cell, to allow the monitoring of
ing two mixing devices and pumps, were used in the the prestress forces. After wedge setting and elastic defor-
project. Each system was used for about 2 hours before mation (immediate losses), they recorded a prestress load

Figure 12. (a, b, c) Assembly of 1:2 scale model of the bridge for destructive 4-point bending.
 VIRTUAL AND PHYSICAL PROTOTYPING 231

Table 2. Prestress loss due to creep and shrinkage, and axial deformation due to prestress load.
Time (days) Prestress load, top (kN) Prestress load, bottom (kN) Elongation, top (mm) Elongation, bottom (mm) Elongation, average (mm)
0 99.9 (100%) 87.4 (100%) 0.905 1.190 1.047 (100%)
2 99.1 (99%) 86.8 (99%) 1.130 1.202 1.166
7 97.9 (98%) 85.7 (98%) 1.362 1.361 1.361
10 97.1 (97%) 85.0 (97%) 1.493 1.451 1.472
16 95.7 (96%) 83.7 (96%) 1.683 1.609 1.646

P = 99.9 kN at the top and P = 87.4 kN at the bottom reloaded to 150 kN, released, reloaded to 180 kN, and
(see Table 2). so forth, until 300 kN, after which the bridge was
Besides, the elongation of the bridge was measured loaded a final time to 350 kN and the test was discontin-
during and prestressing (Table 2). The expected instan- ued. The load-displacement curve is shown in Figure 14.
taneous deformation expected due to the prestress Prior to the test, the pass/fail criteria were set. Consid-
force was 0.5 mm, while 0.9 mm was measured. The ering design bending moments and shear loads, it was
difference is attributed to setting in the joints between determined that the scale model should be able to
the bulkheads and the elements. The prestress force carry at least 176 kN (88 kN per jack). The crack
reduces over time, but the pace of the force reduction moment, flexural shear failure or diagonal tensional
is slowing down rapidly due to the fast shrinkage that shear failure were expected not to occur before 213,
is explained by the nature of the printed concrete and 210 and 454 kN, respectively – although considerable
the small dimensions of the filaments. The prestress statistical variations could lead to deviatory results in
force is reduced by 4% after 2 weeks of prestressing. this specific model.
Thereafter, the bridge was tested on the load-bearing The loading–unloading sequence allows to evaluate
capacity. whether damage has occurred in the model. In structural
The bridge was tested in a load-controlled four-point terms, damage, as caused by cracks, for instance, is
bending test, as shown in Figure 13. The load span was shown by a reduced stiffness and a permanent defor-
one-third of the support span. The load was applied by mation. In a loading–unloading sequence, this translates
two hydraulic jacks (one for each load point), in steps into an unloading curve that does not return to zero dis-
of 30 kN/min (15 kN/min per jack). An alternating placement and a reloading curve that is less steep than
loading–unloading sequence was applied from 120 kN the previous load curve. The test showed no damage
onwards, for each 30 kN of additional load, i.e. the load occurred until 240 kN load. A permanent deformation
was released at 120 kN, then the scale model was and loss of stiffness were only found at the unloading

Figure 13. Scale model in test 4-point bending test set-up.

232 T. A. M. SALET ET AL.

Figure 14. Load-displacement curve of the 4-point bending test on 1:2 scale model, measured through linear variable differential trans-
formers (LVDTs).

and reloading at 270 and 300 kN, respectively. The first Assembly and in situ test
visually noticeable crack appeared between the load
Assembly
points, at approximately 300 kN loading (Figure 15).
Initially, it runs vertically along one of the interfaces The assembly process of the printed elements was
between the filaments. Later, it diverges horizontally studied simultaneously along with the printing research
through the printed mass. The location of the crack indi- and structural research. This was done in two phases. The
cates that it is induced by bending. This is a desired initial phase involved the research on the interface
failure mechanism, as it is more ductile than shear failure. material, defining tolerance between interface material
When comparing the results to the set requirements, and printed elements, and studying the concept of pre-
it should be noted that the first signs of damage occur stressing for printed elements. The second phase
well above the minimum required capacity (>240 kN involved working on a mock-up drill for the assembly
versus 170 kN), and between 10% and 20% above the process on the 1:2 scale bridge, which involved defining
calculated crack or failure loads. Visible cracking protocols for logistics of lifting of printed elements,
occurred at a significantly higher load, and final failure checking alignment tolerances while assembling all the
was not observed. As the scale model carried the six elements and defining requirements for the bulkhead
required load with a large margin and the failure behav- for the prestressing cables.
iour was shown to be bending-induced, the test was For the initial assembly phase, a conceptual cross-sec-
stopped. tional geometry of the bridge was printed in a 1:1 scale,

Figure 15. (a, b) Bending crack, becoming visible at 300 kN total load.
 VIRTUAL AND PHYSICAL PROTOTYPING 233

Figure 16. (a–f) Assembly study, showing the sequential process from lifting of an element to final prestressing.

Figure 17. (a, b) On-site assembly and prestressing.

with the 60 mm filament geometry. A sequential mock-

up of the entire process from lifting the elements
from print bed to assembly was done, as shown in
Figure 16(a–f).
Figure 15(a) shows the lifting of the printed element
from the print bed using a gantry crane in the structural
laboratory at TU/e and flipping the element 90° and
placing it in a temporary wooden scaffold and aligning
the two elements with minimum deviation, as shown
in Figure 15(b,c). Before completely merging the
elements, a synthetic epoxy-based interface material
was applied on the cross-section as a binding material
(Figure 15(d)). The reason for applying the interface
material is to smoothen the surface of the cross-
section. Otherwise when prestressed, the printed
surface of the cross-section being not completely
smooth may have local peek stresses (Figure 15(e)). Figure 18. Hoisting the bridge into position.
234 T. A. M. SALET ET AL.

Figure 19. In situ testing of the bridge with water-filled containers.

Finally, Figure 15(c,f) shows the makeshift bulkhead for The final assembly of the printed elements was based
applying prestressing with steel cables. The study con- on the research conclusions of the first two phases and is
cluded with recommendations on improving the shown in Figure 17(a,b); in Figure 18, the bridge is
process for lifting elements from the print bed, proving hoisted into place.
the strategy of placing of elements in a secondary
wooden scaffold, understanding the thickness of the
In situ test
interface material after prestressing, which should be
taken into account for fixing the length of each A final full-scale test was performed in situ to guarantee
element and finally of the design requirements. the bridge to behave as expected and be structurally

Figure 20. Completed bridge at the opening (photo: Kuppens fotografie).

 VIRTUAL AND PHYSICAL PROTOTYPING 235

safe. Tests like this are obviously non-destructive and are Disclosure statement
used on a regular basis to verify the load-bearing
No potential conflict of interest was reported by the authors.
capacity of older and existing infrastructure.
As shown in Figure 19, the bridge was loaded with 10
containers filled with 500 litres of water. Including their Notes on contributors
self-weight, this resulted in a total load of 57 kN on the
Theo is full professor of Concrete Structures at the Eindhoven
bridge. The containers were positioned in a pattern
University of Technology. Furthermore, he is senior partner
such that 100% of serviceability limit state bending with Witteveen+Bos consulting engineers. The main focus of
moment was reached. The resulting deflections were his research is 3D printing of sustainable concrete structures.
too small to measure. As also no other response was Among others, he develops new types of printable concretes,
observed, and in consideration of the previous material from structural point of view, by using advanced finite
element tools at a heterogeneous level (Numerical Concrete).
and scale testing, the bridge was considered to comply
with the Dutch building regulations. Zeeshan obtained his MSc degree from UCL The Bartlett. He is a
PhD candidate in the chair of Concrete Structures at the Eind-
A concrete surface finishing was applied at the top of
hoven University of Technology. His project evolves around
the bridge, and the parapet was mounted at the abut- the development of a generic approach to 3D Concrete
ments to finish the bridge. On 17 October 2017, it was Printing.
opened to the public (Figure 20). Freek P. Bos as assistant professor of Concrete Structures at the
Eindhoven University of Technology, Freek focuses his research
on the development of reinforcement methods for 3D printed
Conclusion concrete. He obtained his PhD from the Delft University of
Technology.
A 3DCP bicycle bridge was presented. It was printed at
the TU/e 3DCP facility and was the first of its kind (pre- Hans is a senior structural engineer at Witteveen+Bos consult-
ing engineers, where he heads the Structural Engineering
stressed, fused depositioning modelling printed) world-
group in the Buildings department. He has over 25 years of
wide to be put into service. It also includes the experience on a wide variety of projects.
innovative use of cable reinforcement in the print
filament.
The process of design, printing and assembly and ORCID
their interaction was discussed. A thorough testing pro-
Freek P. Bos http://orcid.org/0000-0002-6666-2395
cedure was followed both for the structural and the con-
struction aspects. Assembly trials were performed to
identify potential problems and hazards. Material References
testing, destructive scale model testing and in situ
3ders, 2015. Available from: http://www.3ders.org/articles/
testing was performed to show that the bridge complies
20150909-lewis-grand-hotel-andrey-rudenko-todevelop-
with Dutch building regulations. The bridge was opened worlds-first-3d-printed-hotel.html [Accessed 26 September
in October 2017 and now forms part of the cycling infra- 2017].
structure around Gemert, Netherlands. 3ders, 2016. Available from: http://www.3ders.org/articles/
20161214-spain-unveils-worlds-first-3d-printedpedestrian-
bridge-made-of-concrete.html [Accessed 26 September 2017].
Acknowledgements 3ders, 2017a. Available from: http://www.3ders.org/articles/
20170602-cybe-construction-completes-3dprinting-of-168-
The following partners were involved in the 3DCP bicycle sq-m-rdrone-laboratory-in-dubai.htm [Accessed 26
bridge in Gemert: September 2017].
3ders, 2017b. Available from: http://www.3ders.org/articles/
- Client: Province Noord-Brabant, Netherlands; 20170213-3d-printing-construction-company-apiscor-prints-
- General contractor, initiator: BAM Infra, Netherlands; 37-m2-house-near-moscow-plans-global-expansion.html
[Accessed 26 September 2017].
- Structural design and engineering: Witteveen+Bos 3ders, 2017c. Available from: http://www.3ders.org/articles/
consulting engineers, Netherlands; 20170907-3d-printed-concrete-office-hotel-comingto-
- Research, print design and manufacturing of printed copenhagen.html [Accessed 24 September 2017].
elements: Eindhoven University of Technology, Ahmed, Z.Y., et al., 2016. Design considerations due to scale
Netherlands; effects in 3D concrete printing. Proceedings of 8th
International Conference of the Arab Society for Computer
- Material supplier print mortar: Saint-Gobain Weber
Aided Architectural Design ASCAAD 2016. London, United
Beamix, Netherlands; Kingdom.
- Prestress system and application: Dywidag, Netherlands; Bos, F., et al., 2016. Additive manufacturing of concrete in con-
- Reinforcement cable supplier: Bekaert N, Belgium. struction: potentials and challenges of 3D concrete printing.
236 T. A. M. SALET ET AL.

Virtual and Physical Prototyping, 11, 209–225. doi:10.1080/ Khoshnevis, B., 2004. Automated construction by contour
17452759.2016.1209867. crafting-related robotics and information technologies.
Bos, F.P., et al., 2017a. 3D printing concrete with reinforcement. Automation in Construction, 13, 5–19.
In: M. Luković and D.A. Hordijk, eds. High tech concrete: Khoshnevis, B., et al., 2001. Crafting large prototypes. IEEE
where technology and engineering meet. Cham: Springer Robotics & Automation Magazine, 8 (3), 33–42.
International Publishing, 2484–2493. doi:10.1007/978-3- Khoshnevis, B., et al., 2006. Mega-scale fabrication by contour
319-59471-2_283. crafting. International Journal of Industrial and Systems
Bos, F.P., et al., 2017b. Experimental exploration of metal cable Engineering, 1 (3), 301–320.
as reinforcement in 3D printed concrete. Materials, 10, 1314. Le, T., et al., 2012. Hardened properties of high-performance
doi:10.3390/ma10111314. printing concrete. Cement and Concrete Research, 42, 558–566.
Cementonline, 2017. Available from: https://www.cementonline. Mediaoffice, 2016. Available from: http://mediaoffice.ae/en/
nl/proefstuk-eerste-3d-geprinte-autobrugsterker-dan-verwacht media-center/news/23/5/2016/3d-printed-officebuilding.
[Accessed 24 September 2017]. aspx [Accessed January 2016].
Cnet, 2016. Available from: http://www.cnet.com/news/dubai- Panda, B., et al., 2018. Measurement of tensile bond strength
unveils-worlds-first-3d-printed-office-building [Accessed of 3D printed geopolymer mortar. Measurement, 113,
January 2016]. 108–116.
De Ingenieur, 2017. Available from: https://www.deingenieur. Salet, T.A.M., et al., 2018. 3D printed concrete bridge. 3rd
nl/artikel/betonnen-fietsbrug-uit-de-printer [Accessed 26 International Conference on Progress in Additive
September 2017]. Manufacturing Pro-AM 2018, 14–17 May 2018. Singapore.
Doomen, C.C.M., 2016. The effect of layered manufacturing on Slager, G.J., 2017. Influence of the Interface between Layers on
the strength properties of printable concrete. Graduation the Tensile Properties of 3D-Printed Concrete. Graduation
Thesis. Eindhoven: TU/e. Thesis A-2017.194. Eindhoven: TU/e.
Khoshnevis, B. and Dutton R., 1998. Innovative rapid prototyp- Zareiyan, B., and Khoshnevis, B., 2017. Interlayer adhesion and
ing process makes large sized, smooth surfaced complex strength of structures in contour crafting – effects of aggre-
shapes in a wide variety of materials. Materials Technology, gate size, extrusion rate, and layer thickness. Automation in
13, 53–56. Construction, 81, 112–121.