Content uploaded by Daniel Türk
Author content
All content in this area was uploaded by Daniel Türk on Jun 29, 2017
Content may be subject to copyright.
Available via license: CC BY-NC-ND 4.0
Content may be subject to copyright.
Available online at www.sciencedirect.com
2212-8271 © 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 26th CIRP Design Conference
doi: 10.1016/j.procir.2016.04.202
Procedia CIRP 50 ( 2016 ) 402 – 407
ScienceDirect
26th CIRP Design Conference
Combining Additive Manufacturing with Advanced Composites for Highly
Integrated Robotic Structures
Daniel-Alexander Türka*, Lukas Triebea& Mirko Meboldta
aProduct Development Group Zurich (pd|z), ETH Zürich, Leonhardstr. 21, 8092 Zürich, Switzerland
* Corresponding author. Tel.: +41 44 63 33045; E-mail address: dtuerk@ethz.ch
Abstract
The combination of additive manufacturing with advanced composites offers potentials in the development of highly integrated
lightweight structures. This paper investigates a novel manufacturing process route where binder jetting is used to produce a
water soluble sand core for hand layup of autoclave prepreg composites. The novel approach is applied to a hollow high
performance robotic part demonstrating the following two design potentials: First, binder jetting of water soluble sand material
is a suitable technology for the production of very complex composite parts at low tooling costs. Second, tailored load
introduction elements made of selective laser melting enable lightweight designs. Weight savings of 54.3 % compared to a state-
of-the-art aluminum robotic part indicate that the approach is competitive for complex low volume parts.
© 2016 The Authors. Published by Elsevier B.V.
Selection and peer-review under responsibility of Professor Lihui Wang.
Keywords: Additive Manufacturing, Composites, Lightweight Design, Binder Jetting, Selective Laser Melting
1. Introduction
Advanced composites (AC) are materials with high
strength and stiffness at low weight and therefore are of great
interest for the development of high performance lightweight
structures. AC have two fundamental characteristics that
make design and fabrication challenging. For one, all AC
parts are made by processes that require hard tooling for
support during layup and cure. Traditional toolmaking cycles
are cost-intensive and time consuming as investment in
tooling is high, especially for low volume production [1]. The
growth of advanced composite applications with complex
contours has heightened the need for tooling that affords
design flexibility. Another important characteristic is that the
fibrous nature of the material makes joining a critical step in
manufacturing. The weight advantages of composites are
often compromised at the joint as quasi-isotropic layup and
additional reinforcements are necessary [2].
Additive manufacturing (AM) produces geometrically
complex three-dimensional objects and offers new
possibilities in composite tooling. In fact, the trade journal
Composites world states that 3D printing of aerospace tooling
is a growing trend [3]. The authors mention applications as
master models, layup tools used to produce composite parts
for autoclave processes and washout mandrels for trapped
tooling. With AM it is easily possible to add or remove
material and thereby to tailor the mechanical performance
according to local loads. This makes AM suitable for tailored
load introduction elements.
In previous work we demonstrated how the combination of
AM with AC can lead to innovative integral lightweight
designs by using the example of the lower leg structure of a
hydraulically actuated robot (HyQ). In this example carbon
fiber reinforced polymers are laid on a core made with Fused
Deposition Modeling and a load introduction made by SLM
[4]. However, there are some limitations: The part was
manufactured with vacuum pressure only and the printed core
remained in the structure adding additional weight to the part.
In this paper we extend the approach combining AM with
AC and present a novel manufacturing process route. Binder
jetting is used to produce a water soluble sand core for hand
layup of prepreg composites that are then cured in a standard
autoclave process. Finally, the core is dissolved in water
© 2016 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/4.0/).
Peer-review under responsibility of the organizing committee of the 26th CIRP Design Conference
403
Daniel-Alexander Türk et al. / Procedia CIRP 50 ( 2016 ) 402 – 407
allowing the production of complex shaped composite
structures at no additional weight.
This paper proceeds as follows: Section 2 provides
background information about the three relevant technologies
in this research: binder jetting, SLM and advanced
composites. Section 3 presents the novel manufacturing
process route combining AM with AC. In Section 4 the
extended approach is applied to the lower leg structure of the
HyQ robot. Design and analysis of the selected part are
presented. Section 5 compares the weight of the novel leg to
the state-of-the-art aluminum reference. Section 6 discusses
the manufacturing process by using the example of the lower
leg. Recommendations are given. Section 7 concludes.
2. Background
2.1. Additive manufacturing (AM)
Additive Manufacturing is a process of joining materials to
make objects from 3D model data, usually layer upon layer,
as opposed to subtractive manufacturing technologies [5].
2.1.1. Binder jetting
Binder jetting is an AM process in which a liquid binding
agent is selectively deposited to join powder particles. Layers
of material are then bonded to form an object. The build
platform is lowered and the next layer of powder will be laid
out on top. The process is repeated until the final part is built.
Binder Jetting does not require support structures as the part
lies in the job box of not bonded powder [6]. Binder jetting is
a rather fast and cost-effective technology, working with a
wide range of materials including metals, sands and ceramics
[7]. Some materials like sand require no additional post
processing and large building envelopes are available.
2.1.2. Selective laser melting
Selective Laser Melting (SLM) is a powder based fusion
process where a laser beam is directed onto the powder bed in
such way that it thermally melts the powder material to form
the slice of the parts’ cross section. [8] This leads to the
consolidation of the powder particles in the scanned area,
resulting in a nearly fully dense layer of the part being built.
The building platform is lowered and the process is repeated
until the final physical part is produced. Due to the high
melting temperature of metals, thermal gradients in the build
chamber can lead to thermal stresses and warping. [9] To
overcome this effect, manufacturers attach the part to the base
plate with anchors and add support structures for some
inclined geometries. Parts made with SLM machines are
increasingly used for final products.
2.2. Advanced composites (AC)
Carbon Fiber Reinforced Polymers (CFRP) consist of
aligned continuous carbon fiber reinforcements that are
embedded in a polymeric resin (e.g. epoxy). For many high
performance applications PREimPREGnated fibers are used.
Prepregs are resin-impregnated semifinished products in fiber,
fabric or mat form, in which the resin is kept in a non-
polymerized state through cooling. Semifinished products
with fibers laid at 0° orientation are called unidirectional
(UD). The autoclave process is the state-of-the-art
manufacturing techniquefor high performance applications.
In this process prepregs are cut and laid down in the desired
fiber orientation. The layup is vacuum bagged and put in an
autoclave where defined temperatures ranging up to 180°C
and pressures up to 10 bar are applied for curing and
consolidation of the part. The prepreg process is labor
intensive but shows great flexibility in design and therefore is
suited for low volume production [10]. In co-curing no
additional adhesive film but only the resin of the prepreg is
used to generate bonding, thus reducing the number of joints
and process steps.
3. Novel Manufacturing Process Route
The manufacturing technique applied in this research is a
sequential combination of additive manufacturing and co-
curing autoclave prepregs. The novelty in the process is the
use of the binder jetting technology to produce a dense sand
core for the autoclave processing. The detailed steps are:
xBinder jetting &sealing of the core:Water soluble binder
is deposited on sand powder to obtain a complex shaped
core that is produced without tooling. A sealing liquid is
repeatedly sprayed on the coreto prevent resin inflow
during autoclave curing. The sealed core is packed under
vacuum for storage and transport.
xSelective laser melting of complex elements: Load bearing
elements are made by SLM. The partsare sand blasted,
shot peened,heat treated and machined.
xLayup: The printed parts are assembled and cleaned. The
prepreg is cut in the desired orientation to net-shape. The
plies are draped on the tooling and the SLM part according
to the ply book.
xBagging: A vacuum bagging is conducted. This includes
the application of a release film, bleeder and the covering
with a vacuum bag. The assembly is placed in the
autoclave.
xCuring & consolidation: The composite is co-cured and
consolidated in one shot according to the cure cycle data of
the resin system in a standard autoclave process.
xCore Removal& Post Processing:The part is demolded
and the core is washed out with tap water. The part is
trimmed where necessary. Holes are drilled and surface
finishing methods are applied.
4. Experimental
4.1. Reference
The Hydraulically Quadruped (HyQ) robot was developed
at the Italian Institute of Technology to serve as a platform to
study highly dynamic motions such as running and jumping
over rough terrain. HyQ is about 1 meter tall and weighs 91
kg. Each of its four legs has 3 joints that are actuated by
hydraulic cylinders and motors [11].The lower leg consists of
404 Daniel-Alexander Türk et al. / Procedia CIRP 50 ( 2016 ) 402 – 407
an aluminum tube that is pinned to two milled plates. It exists
in its second lightweight version and is structurally optimized.
Fig. 1. (a) HyQ Robot; (b) CAD model of the leg structure [11]
4.2. Lower leg concept with AM & AC
A novel lower leg for a robotic structure is developed using
additive manufacturing and autoclave prepregs. The combined
approach is used in a way to benefit from the specific
advantages of each material and production technology,
resulting in a highly integrated lightweight structure as shown
in Figure 2. A titan part made by SLM is used for the load
introduction to withstand stress peaks and multi-axial loads.
The bearing fittings, necessary for attaching the lower leg to
the upper robot structure, are integrated into the titan part. An
aluminum stopper limits the flexion of the lower leg. The
inner tooling is made of a binder jetted water soluble sand
core. An aluminum insert featuring a thread for mounting the
Misumi foot is pinned to the sand core. CFRP prepregs cover
the titan load introduction, the core and the aluminum insert.
The load is transferred from the titan part into the CFRP shell
where biaxial bending is the dominant loading condition.
Fig. 2. Concept of a lower leg made with AM and CFRP
4.3. Design analysis
4.3.1. Loads
The critical load case is the flying trot, where the robot
runs at a speed of 3 to 4 m/s. Diagonal leg pairs move
together and the robot lifts off completely between the steps.
The robot lands on two diagonal legs, inducing a highly
dynamic and cyclic load into the lower leg structure.
The global ground reaction forces are measured for the
load case of the flying trot and transformed into a local
coordinate system of the lower leg displayed in Figure 3. The
local ground reaction forces consist of normal and lateral
forces around the local z1 and y1 axis. (blue and black).
Fig. 3. Local ground reaction forces acting on the lower leg
The two local lateral forces are combined for every
measurement point to a resulting lateral force that is fully
described by its absolute force value and a corresponding
angle. The resulting forces are used to calculate the principal
bending moments for every data point. Figure 4(a) shows the
polar plot of the principal bending moments over the
corresponding angle. Two rough box section profiles are
designed to withstand the biggest bending moments in the
outmost angles. Splines join the profile. The smoothed profile
is shown in Figure 4(b).
Fig. 4. (a) Principal bending moments with corresponding angles and overlaid
rough profile; (b) Cross section of smoothed profile.
The three data points with the highest bending moments are
selected as the limiting design loads, shown in Table 1.
Table 1. Design loads in local coordinate system
Load vectors Fx1 (N) Fy1 (N) Fz1 (N)
Load triplet 1 - 565 465 801
Load triplet 2 -632 302 842
Load triplet 3 -499 402 788
4.3.2 Material allowables
The maximum material allowable for the CL 41 TI ELI
Titanium alloy is a fatigue limit of 600 MPa. The allowables
for the CFRP can be found in Table 2.
Table 2. Material allowable for the CFRP
Material Strain fiber
parallel (%)
Strain fiber
transverse (%)
Interlaminar
Shear stress (MPa)
CFRP UD 0.4 0.4 75
CFRP Fabric 0.4 0.4 70
405
Daniel-Alexander Türk et al. / Procedia CIRP 50 ( 2016 ) 402 – 407
4.3.2. FEM results
The result of the FEM analysis of the SLM part is shown in
Figure 5. The von Mises stress failure criteria is applied. High
stresses occur in the box sections. The maximum stress to be
found is 330 MPa and is located at an inner edge. This value
is admissible and results in a safety factor of 1.51.
Fig. 5. Von Mises failure criteria of the SLM part
Figure 6 shows the strain in the critical UD ply. High
strains occur in the area of the load transmission between the
titan part and the hollow composite shell. The absolute
maximum amounts to -0.19 % which is admissible and
corresponds to a safety factor greater than 2.
Fig. 6. Maximum strain failure criteria
4.4. Manufacturing
4.4.1. Additive manufacturing of the titan part
Fig. 7. (a) As-built SLM part; (b) SLM part with stopper and bolt after heat
treatment and machining.
The SLM part is manufactured by 3D Precisions from a
titanium alloy Ti6Al4V powder on a Concept Laser M2
machine with a layer thickness of 30 µm. The part has a
bounding box of 93 x 60 x 71 mm and the building time
amounts to 40 h. Figure 7(a) shows the raw SLM part. The
support structure is removed mechanically. The part is
sandblasted and shot peened to improve its surface quality and
its dynamic properties. The part is heat treated to release
residual stresses at a rate of 5°C/min, held at a 840°C for 2h
and cooled to room temperature at a rate of 1°C/min. The
aluminum stopper is mounted to the SLM part and the bearing
fittings are machined. The final SLM part is depicted in
Figure 7(b).
4.4.2. Binder jetting of the sand core
Figure 8 shows the core made with an ExOne S-Max
binder jetting machine featuring a layer thickness of 0.28 -
0.38 mm. An inorganic, water soluble binder system was
applied on Silica sand with a grain size of 0.14 to 0.19 mm. A
water soluble sealing layer is sprayed on the printed sand core
and dried in the oven. The sealing process is repeated five
times amounting to a total sealing layer thickness of
approximately 0.3 mm. The core features four bore holes with
a diameter of 4 mm each to attach the foot insert. The SLM
part is positioned via the step and fixed with a bolt.
Fig. 8. Sealed sand core made with binder jetting
4.4.3. Autoclave process
The sand core, the stopper, the SLM part and the foot
insert are assembled and fixed with pins. A glass fiber fabric
is draped around the foot insert to prevent contact corrosion
between the carbon plies and the aluminum. The net-shape cut
prepreg plies are draped on the parts according to the ply-
book. A four-part silicone mold is casted and used to enclose
the part.
Fig. 9. Final part
The composite assembly is vacuum bagged and cured in the
autoclave at 2 bar and 80°C. The part is heated at a rate of
1°C/min, held at 80°C for 10h and cooled down to room
temperature at a rate of 1°C/min. Next, the upper silicone
mold is opened and surface finish is applied. The final part is
depicted in Figure 9.
406 Daniel-Alexander Türk et al. / Procedia CIRP 50 ( 2016 ) 402 – 407
5. Results
5.1. Weight
Figure 10 compares the weight of the aluminum reference
part with the demonstrator leg manufactured in this research.
The part is divided into components that are within and
without the design scope. The latter remain unchanged and
amount to 159 g of fixed weight: these are the bolt, the foot,
the stopper, the bearing and the downholders. Within the
scope are the load carrying aluminum structure and the
bearing fittings that weigh 714 g. The total weight of the
lower aluminum leg amounts to 873 g. This stands against
399 g when made with AM and CFRP. The CFRP weighs 60
g, the SLM part weighs 161 g and the foot insert amounts to
19 g, resulting in 240 g. This corresponds to weight savings of
66,4 % or 474 g within the design scope. The results show
total weight savings of 54,3 % or 474 g.
Fig. 10. Weight comparison
5.2. SLM part deviation
Figure 11 shows the deviation of the circular surfaces of
the SLM titan part before and after stress-relief annealing
compared to the nominal distance. It can be seen that the
deviation increases further away from the base plate. The
graph shows a maximum deviation greater than 1.6 mm at the
top of the part. The deviation of 2.1 mm at the tip of the part
when heat treated indicates thermally induced inner stresses.
Fig. 11. Deviation of functional surfaces before and after heat treatment
6. Discussion
The approach combining AM with CFRP is analyzed along
the manufacturing process by using the example of the lower
leg. Recommendations are given based on the results.
6.1.1. Binder Jetting of the sand core
Binder jetting of sand materials shows limited possibilities
for the integration of functional elements compared to other
AM technologies such as FDM. This is due to the brittleness
of the sand material that can not withstand concentrated loads.
Furthermore the resolution of the process is limited by the
size of the sand grains that is in the area of one to two tenth of
a millimeter compared to a particle size of approximately 60
µm for polyamide powders used in selective laser sintering.
[12] We therefore recommend to avoid interfaces with
concentrated loads and use interfaces with greater surfaces
such as tapered cones. It is recommended to use a different
technology for areas where tight positional tolerances are
necessary.
6.1.2. Warping of the SLM part
Parts made with SLM are prone to warping as the high
internally induced stresses distort the material. This can result
from material accumulations that depend on the cross
sectional area of the powder layer. Design rules for steel
recommend to avoid cross sectional areas greater thanAlimit =
400 mm2[13]. Figure 12 shows the cross sectional area of the
SLM part at the locations z = -21.8 mm, z = 0 and z = + 21.8
mm. The three cross sections are compared with the slope of
the deviation curve from Figure 11. It can be seen that the
deviation increases sharply near the midpoint z = 0, which is
the location with the greatest cross sectional area A2= 447.7
mm2. This seems to confirm one cause of the part deviation
being accumulated material. However, deviations are
observed for cross sections below the suggested value,
indicating that the thermally induced residual stresses are
more distinct for titan. To minimize part deformation, we
recommend the following design strategies:
xReduce material accumulations: Substitute material with
structures of lower density (eg. Lattices) where admissible
by design.
xReduce length of heat flux: Use support structures to lead
the heat flux directly in the base plate. This reduces part
deformation during build up. Consider breakage points to
simplify its removal.
Fig. 12. Cross sectional areas of the SLM titanium part
407
Daniel-Alexander Türk et al. / Procedia CIRP 50 ( 2016 ) 402 – 407
6.1.3. Layup
The prepreg is cut to net-shape eliminating further
machining steps and composite waste. The complex geometry
makes hand layup a suitable technique to form a
reinforcement without wrinkles or tears but the following
considerations should be taken into account:
xThe hand layup for complex shapes necessities a
skilled operator.
xDraping on complex shaped tools can lead to deviated
fiber angles and fiber volume contents resulting in
locally different mechanical properties.
xThe design is limited by the draping characteristics of
the reinforcement.
xPly complexity can be opted for more but simpler
geometries.
xConcave geometries require tooling to ensure
sufficient laminate compaction.
6.1.4. Autoclave curing
The curing time depends on the resin system and often
shows a logarithmic relation to the curing temperature.
Industrial autoclave processes typically use a curing
temperature of 120°C which would reduce the cycle time for
the SGL epoxy prepreg resin E022 used in this research to
only 50 min. This enhances the economic efficiency of the
process. Increasing the autoclave pressure would reduce the
void content and therefore improve the mechanical properties.
However, the upper limits in terms of temperature and
pressure are not yet available for printed and sealed sand
materials. Furthermore, differences in the coefficient of
thermal expansion between the materials can lead to shape
distortions and stresses when processed at high temperatures.
6.1.5. Removing the sand core
The sand core has a volume of 287 cm3and is dissolved in
tap water within one hour. Fig. 13(a) shows sand
solidifications on the inside of the CFRP shell as a
consequence of incomplete sealing. Resin entered the sand
core and solidified during curing. Such accumulations are
removed mechanically, making accessibility a key issue. To
avoid this effect, an increase in the number of sealing layers
is recommended.
6.1.6. Final machining
High demands on surface qualities and part tolerances
require machining of parts made by SLM. Therefore
machining should be considered during the part design. Figure
13(b) shows the clamped SLM part. Perpendicular alignment
of the planes simplifies clamping. The plane of the building
platform can preferably be used as a reference plane since
parts are typically removed by wire eroding. Demands for
flatness and parallelism tolerances require sufficient
allowances for machining. In the depicted part 1 mm
allowance per side was considered to encounter process
induced part deviations. It is recommended to consider the
location of functional surfaces in areas with low deviations to
minimize necessary allowances.
Fig. 13. (a) Sand solidifications due to incomplete sealing of the sand core;
(b) SLM part with clamping jig
7. Conclusion
In this paper a novel manufacturing route combining
additive manufacturing and advanced composites is
investigated and applied to a hollow robotic leg structure.
Binder jetting is used to produce a water soluble sand core for
autoclave processing. SLM is used for the tailored load
introduction. The following results are drawn: First, weight
savings of 54,3 % are achieved compared to a state-of-the art
aluminum leg. Second, the approach opens doors to novel
lightweight designs. The technical feasibility of complex load
orientated structures at low tooling cost is demonstrated.
However, there are some limitations: Sealing technology has
to be improved to eliminate resin inflow during curing.
Furthermore, the binder jetting technology with water soluble
sand materials has to be characterized for higher autoclave
pressures and temperatures to increase the efficiency of the
process. Finally, new design challenges and restrictions
emerge from the combined approach that call for proficient
knowledge in the design of AM and AC.
References
[1] Black S. Tooling roundup: new materials, new methods. Composites
World 2004.
[2] Schürmann H. Konstruieren mit Faser-Kunststoff-Verbunden. 2nd ed.
Darmstadt: Springer; 2007.
[3] Black S. A growing trend: 3D printing of aerospace tooling. Composites
World 2015.
[4] Türk D-A, Klahn C, Meboldt M. Combining additive manufacturing with
CFRP Composites: design potentials. Proceedings of the 20th International
Conference on Engineering Design ICED; 2015.
[5] ASTM Standard F2792-12a, Standard Terminology for Additive
Manufacturing Technologies ASTM International; 2012.
[6] Gebhardt A. Understanding Additive Manufacturing. Munich: Carl
Hanser Verlag; 2011.
[7] Utela B, Storti D, Anderson R, Ganter M. A review of process
development steps fro new material systems in three dimensional printing
(3DP). Journal of Manufacturing Processes 2008; 10:96-104.
[8] Gibson I, Rosen DW, Stucker B. Additive manufacturing technologies.
New York: Springer Science+Business Media; 2010.
[9] Mercelis P, Kruth J-P. Residual stresses in selective laser sintering and
selective laser melting. Rapid Prototyping Journal 2006; (12)5:254-265.
[10] Gutowski T. Advanced composites manufacturing. Cambridge: John
Wiley & Sons; 1997.
[11] Semini C. HyQ – Design and development of a hydraulically actuated
quadruped robot. Istituto Italiano di Tecnologia (IIT). Genoa; 2010.
[12] Berretta S, Ghita O, Evans KE, Anderson A, Newmann C. Size, shape
and flow of powders for use in selective laser sintering. White paper.
[13] Adam G, Zimmer D. Design for additive manufacturing – element
transitions and aggregated structures. CIRP Journal of Manufacturing
Science and Technology 2014; 7:20-28.