Stable and Adjustable Mechanisms for Optical Instruments and Implants

Abstract

Metal printing, in particular powder bed fusion with laser beam (LPBF) of metals, combined with appropriate design techniques permits the creation of yet unknow monolithic mechanisms that make use of engineered compliance to allow specific motions. This subject has been recently explored in the SAMOII project, in which a team of researchers at TU Delft and Fontys worked on creating mechanisms for alignment of optical components as well as spine implants with tailored flexibility. This article covers two topics that can be of broad interest: the developed design techniques to generate complex flexures and mechanisms, and the challenges and solutions in printing metal flexures. Following that, an application case study is presented as well as a discussion on remaining challenges.

Full text

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
1 | 15
Dit is de nieuwsbrief van IAMM waarmee binnen het consortium kennis en nieuws gedeeld
wordt. In deze editie komt een samenwerking tussen Fontys en de TU Delft aan bod.
Veel leesplezier.
Stable and Adjustable Mechanisms for Optical Instruments and Implants
Koppen, S.1, Yasir, A.1, Langelaar, M.1, Herder, J. 1, Mast, R. van der2
1TU Delft, Faculty of Mechanical Engineering, Department of Precision and Microsystems
Engineering, Mekelweg 2, 2628 CD Delft, The Netherlands
2Fontys, Department of Mechanical Engineering, BIC 1, 5657 BX Eindhoven, The Netherlands
Abstract
Metal printing, in particular powder bed fusion with laser beam (LPBF) of metals, combined
with appropriate design techniques permits the creation of yet unknow monolithic
mechanisms that make use of engineered compliance to allow specific motions. This subject
has been recently explored in the SAMOII project, in which a team of researchers at TU Delft
and Fontys worked on creating mechanisms for alignment of optical components as well as
spine implants with tailored flexibility. This article covers two topics that can be of broad
interest: the developed design techniques to generate complex flexures and mechanisms, and
the challenges and solutions in printing metal flexures. Following that, an application case
study is presented as well as a discussion on remaining challenges.
Key words: Topology optimization, compliant mechanisms, flexures, additive manufacturing,
laser powder bed fusion
Wetenschappelijk
Introductie:

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
2 | 15
Figure 1: Various compliant mechanism prototypes designed and additively manufactured
under the SAMOII project.
Motivation and Aim
Throughout the high-tech and medical engineering fields, instrumentation and high-
performance equipment critically rely on adjustable mounts for fine alignment of optical
components. State-of-the-art adjustable mounts and mechanisms typically consist of many
parts, are voluminous and heavy, and lack control of stiffness of individual degrees of freedom.
This results in cumbersome and time-consuming manual adjustment procedures of optical
setups and instrumentation.
Recent advances in additive manufacturing offer new possibilities for the development of
highly complex monolithic structures and compliant mechanisms. However, additive
manufacturing of compliant structures comes with its own challenges.
The research project `Stable and Adjustable Mechanisms for Optical Instruments and
Implants’ (SAMOII), consisting of two academic groups and seven industrial partners (TU Delft,
Fontys, TNO, ASML, VDL, Airbus, Nexperia and BAAT Medical), developed new design
methodologies combining kinematic synthesis and computational topology optimization
methods to generate monolithic compliant mechanisms with multiple fully decoupled
kinematic adjustment modes. Some examples are shown in Fig. 1 above. To enable additive

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
3 | 15
manufacturing of the designs, the project also focused on printing of thin metal flexures. These
form critical elements in most compliant mechanisms and pose a challenge in terms of support
placement and removal, heat input, surface quality and overall reliability, as their dimensions
approach the critical dimensions of the printing process. Both research directions, i.e. the
design and printing of complex 3D compliant mechanisms, are elaborated below.
Computational mechanism design
Topology optimization of compliant mechanisms
Topology optimization allows for automated computational design of complex structural
layouts simultaneously considering various conflicting user-defined performance measures
(Bendsoe, 2013). A key advantage is that the designer does not need to provide an initial
design concept but only an outline, as the design is generated by the algorithm itself. This
property makes that very innovative designs can be generated and evaluated. In addition, the
shape of the design is virtually unrestricted, up to the resolution of the finite element analysis
that is applied in the process. This provides an enormous design freedom that allows for new
design solutions (Bendsoe, 2013).
A schematic illustration of the topology optimization process is provided in Fig. 2. The process
typically starts with the definition of a so-called design domain, that is the region of interest
to the user, and its initial design (typically homogeneous material with intermediate density).
This initial design is now iteratively improved by subsequent (i) filtering technique(s), (ii) finite
element analysis, (iii) sensitivity analysis, and (iv) design update by solving an approximate
optimization problem. Special filtering techniques are employed in attempt to regularize the
spatial density description and/or imitate a manufacturing process. The finite element analysis
is required to determine the behaviour of the structure, and thereto its performance. Next,
sensitivities of this performance with respect to the local densities are calculated and form
the crucial information to judge design improvement. Since the constrained optimization
problem is typically too hard to solve directly, standard practice is to solve successive
approximate (e.g. linearized) optimization problems. The 0resulting design variables are
updated, representing a new, typically improved, design. This process of automated analysis
and design is continued until a user-defined termination criterium is met.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
4 | 15
Figure 2: Schematic illustration of the topology optimization process including most crucial
steps; initialization of the design domain, iterative execution of design filtering, finite element
and sensitivity analysis and design optimization until convergence.
Despite various contributions since the early work of Sigmund (2001) focusing on topology
optimization methods for compliant mechanism design, state-of-the-art methods mostly
consider only the planar case, may easily induce design artifacts and do not consider undesired
interaction between multiple inputs and outputs. What is more, despite their crucial
importance, additive manufacturing limitations such as maximum overhang angle, minimum

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
5 | 15
printing resolution, precision and accuracy as well as thermo-mechanical (re)design
considerations have only been considered to a limited extent (Koppen, 2021).
In collaboration with industrial partners, the TU Delft team developed a simple, versatile, and
efficient topology optimization framework that allows for systematic design of decoupled
parallel compliant mechanisms, flexures, springs, and shape-morphing structures (Koppen,
2022a & Koppen 2022b). Using high-performance computing, the framework can generate
high performance designs with a resolution exceeding additive manufacturing capability, see
Fig. 3. Simultaneous modelling and measuring the performance of slightly eroded and dilated
geometries allows for control of the minimum printable feature size, and robustness of the
performance measures with respect to fabrication variations (Wang, 2011 & Koppen, 2021).
Figure 3: Example of an optimized flexure designed using the topology optimization
framework (Koppen, 2021 & artofscience, 2020). The figure displays the flexure in its
undeformed (left) and deformed (right) configurations. This flexure is optimized to exhibit a
very high ratio between the stiffness in axial, shear and specifically torsion mode, as compared
to the stiffness’ in bending.
Additive manufacturing of precise flexures
As discussed above, flexures typically constitute the most functionally important but also
thinnest parts of a compliant mechanism. Printing them correctly by laser-based powder bed

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
6 | 15
fusion of metals (PBF-LB/M) and without artifacts requires special attention. In the SAMOII
project, this has been explored using a set of representative flexure geometries.
The objective was to evaluate the additive manufacturing system and improve it where
necessary, so that a successful production of the topology-optimized designs could be
achieved. That is why some crucial printer parameters such as layer thickness, laser power,
exposure time and point distance, as well as support material layout had to be optimized to
ensure a reliable print quality. The additive manufacturing system employed was the
Renishaw AM400 at Fontys University of Applied Sciences (see Fig. 5), equipped with a 400-
watt modulated continuous-wave fibre laser (pulse-like output) with a 70-micron focal
diameter. The applied metal powder was non-virgin stainless steel ‘PowderRange 316L F’ by
`Carpenter Additive’.
Figure 4: the `Objexlab Metal’ team of Fontys, led by Rein van der Mast (2nd from right).
Table 1: Parametric specimen design and the parameters for a flexure printing test case.
L
T
O
Build
direction
1 mm
R

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
7 | 15
For the testing phase, a cross-axis flexure design was chosen, as it allowed simultaneous
evaluation of the minimum reliable flexure thickness and overhang angle limits in a single
parametric design. Several sample sets were prepared, each featuring different flexure
thicknesses and overhang angles, which were subsequently printed for assessment. In Table
1, the parametric model of a cross-axis flexure is depicted, and a set of parameters is
presented for the initial test case.
Overcoming challenges in flexure printing
The initial set of prints (see Fig. 6) revealed localized heat concentrations, leading to
distortions and even cracks. To address this, the second set of prints (see Fig. 7) incorporated
several design modifications. Dummy volumes to act as contactless support were strategically
placed under the thin flexures, positioned close enough to act as heat sinks without interfering
with the flexures themselves. Cavities were introduced in the longer sections to avoid bulk
material accumulation, which could cause excessive heat concentrations and deformations.
Furthermore, to address distortions between the interconnected blocks via flexures, locking
support bars were added to the sides of the part. These bars were to be removed after the
printing process. Lastly, a finer support structure has been employed to securely connect the
samples to the build-plate. Also, the orientation of the scan path in the flexures was adjusted
in a way that the amount of inner hatch lines not parallel to the primary orientation of the
flexures was minimized.
Later this became the basis of the current Ph.D. research project by Van der Mast at KU
Leuven, which focuses on miniaturization in parts additively manufactured by laser-powder
bed fusion, including thin walls that function as leaf springs. This includes his `Scanpath Centric
Design’ method, which allows him to define the exact location of the scan path while the part
is being digitally constructed, rather than having a slicer/hatcher interpret a solid model. By
its nature, this is only applicable for small elements, for example leaf springs that may only
consist of scan lines along the longitudinal direction, instead of having a contour that encloses
many very short lines in different directions (an array of parallel lines: hatch), of different
lengths. Due to these variations, the resulting solidification is inconsistent and therefore often
lacks homogeneity and density [8-10]. Scanpath Centric Design aims to eliminate such
variations to achieve consistent material properties. From the basis laid in the SAMOII project,
Van der Mast is currently working to further develop and evaluate this method for printing
high quality precision parts.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
8 | 15
Post-processing involved the removal of the printed pieces from the build plate using a band
saw. Subsequently, an additional milling process was required to eliminate the remaining
support material from the parts. Upon completion of these steps, the locking support bars
were carefully cut, freeing the flexures to move as intended.
The efforts yielded promising results. Flexures with a 30° (and higher) overhang angle were
successfully printed without any defects for a flexure thickness of 400 microns. Moreover,
flexures with a thickness of 350 microns were printed with a 40° overhang angle.
Figure 5: The first set of test flexures with coarse support material.
In conclusion, the tuning of printer parameters and the introduction of design modifications
enabled successful manufacturing of samples that provided valuable insights into the
limitation on minimum feature size and overhang limits. These findings contributed to the
reliable printing of more complex designs, such as those originating from topology
optimization.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
9 | 15
Figure 6: The second set of flexures with design modifications: finer support material, the
support bars on sides, dummy volumes (contactless support) and the cavities to reduce heat
concentration.
Application showcase: Customized spinal implant
Lower back pain is the second most common symptom for a visit to a physician in the US, with
associated costs exceeding $100 billion per year (Dagenais, 2008). One of the leading causes
is spinal instability due to degenerative disc diseases. Degenerated intervertebral discs have a
different stiffness, causing overloading of adjacent components, tearing of fibers and
decreased disc height causing chronic pain. Traditional spinal fusion surgery permanently fixes
two or more bones in the spine, providing stability, though restricting range of motion and
increasing the stress on adjacent discs (so-called stress shielding).
As an alternative to traditional fusion surgery, we showcase the conceptual design of a fully
monolithic non-fusion spinal implant. Such designs do not wear and tear and retain the
patients range of motion. What is more, computational design optimization allows for patient-
Dummy volumes as heat sinks

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
10 | 15
specific design by straightforward modifications to the allowed design volume and values of
design requirements, such as the desired stiffness.
The goal of the computational design optimization approach is to find the design that
maximizes the stability of two adjacent vertebrae while allowing for range of motion in
(lateral) flexion, extension and rotation of the spine. Fig. 8 shows a part of the spine with one
disc replaced by a conceptual design, located and oriented with respect to the axes displayed.
We may rephrase the goal in a more formal optimization language: that is, find the value of
the design variables that maximizes the translational stiffnesses TX and TY (shear), while
limiting the translational stiffness TZ (extension) and rotational stiffnesses RX (flexion), RY
(lateral flexion), and RZ (rotation). We furthermore restrict the allowable use of material for
weight and cost reductions.
Additive manufacturing limitations are considered by constraining the maximum radius of
curvature and minimum feature thickness, thereby effectively eliminating small cavities and
thin flexures (Wang, 2011). Note that these constraints negatively impact the achieved
performance (here shear stiffness) but allows for manufacturability of the part and limits the
sensitivity to unavoidable manufacturing errors (such as uniform geometric errors). Overhang
limitations can have a tremendous negative effect on the performance (Koppen, 2021).
Therefore, we opt to allow for overhanging features and deal with those by smart design of
the support structure and printing parameters as previously discussed.
The design for one specific set of input parameters (geometry and desired stiffnesses) is
shown in Fig. 9. The framework can readily address patient specific data, such as outer
geometry and desired stiffness’. Therefore, the displayed result is one of the many possible
personalised designs that may arise from this framework. The design exhibits a complicated
three-dimensional interconnection of beams under various angles. The working principle of
such design is hard to perceive, let alone engineering such a piece, which illustrates the power
of computational design to find innovative solutions. The obtained complexity furthermore
emphasizes the utilization of the extreme freedom given to the optimization scheme, fully
exploiting the capabilities of additive manufacturing.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
11 | 15
Figure 7: Overview of one of the non-fusion spinal implants, including an artist impression of
a placed non-fusion spinal implant in situ and definition of the relevant degrees of freedom
(top left), additively manufactured part (bottom left), and various views (right) and cross-
sections (middle) of the CAD model.
To address overhang constraints without introducing a significant amount of post-processing
tasks, which is common in general support concepts, alternative supports were developed:
'needles' were added manually to a digital representation of the design, to be printed with the
part, and then manually removed in a way that is not labour intensive, leaving only some small
scars on the downskin surfaces, thereby minimizing differences in roughness and porosity as
compared to the other sides of the printed object.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
12 | 15
Figure 8: Custom-made support structures, created at Fontys, for a satisfactory result that is
easy to remove.
The new approach to the requirement of support structures, shown in Fig. 10, worked
satisfactorily in 316L, in a Renishaw AM400, with a layer thickness of 50 µm. However,
manually inserting the needles digitally (in Rhino 5), especially those that were not straight,
was time-consuming. Removing them once materialized was very easy. With the smallest
diameter at the top, they all broke loose there. In this case, the diameter of the needles is 0,4
mm, with only its contour scanned as well as an offset to the inside of 0,1 mm. So, the needles
are hollow. At the top the diameter is narrower than below, leaving only the narrower
contour, so without the offset. It is the weakest spot, where the part and the support are to
break loose. (Very short) hatch lines are not included because their added value would be
limited, and the data processing time and printing time would increase.
This case study demonstrates the enriching possibilities of the proposed framework; designing
customized high-performant fully monolithic compliant structures that are readily additive
manufacturable. The intricate designs arising from this framework are not only highly detailed
and unique but can also provide academics with novel insights about the working principles
and synthesis of novel conceptual designs. In combination with smart use of sacrificial support
structures and heat sinks, and in-depth knowledge of the process, it was possible to

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
13 | 15
successfully print complex compliant mechanisms, which illustrates the potential of combining
computational compliant mechanism design and metal additive manufacturing.
Conclusions and outlook
The SAMOII project has set new standards in the creation of complex compliant mechanisms.
Firstly, regarding their design: with a newly developed topology optimization approach,
computational design of effective compliant mechanisms and flexures was enabled. Designs
exhibiting exactly the desired functionality in terms of stiffnesses in different rotations and
translations can be generated automatically. Secondly, new approaches for the laser-powder
bed fusion additive manufacturing of such geometrically complex compliant mechanisms
were developed. These allow for fully functional flexures and other detailed features with the
required dimensions and properties, through dedicated use of sacrificial support and heat
evacuation structures. In this article a spine implant case study was presented, where the
methods proved capable of generating and manufacturing a design that met all specifications
in a very confined space. The new implant concept has resulted in a patent application and
will hopefully benefit many patients.
Next to flexures and implants, the SAMOII project also has considered compliant mechanisms
with multiple inputs and outputs, such as fully compliant monolithic optical mounts with
alignment adjustments. New designs with minimal crosstalk between different alignment
degrees of freedom were produced and are expected to find their way into various high-tech
and space applications. However, this is not the end of the story regarding printed compliant
mechanisms, but only its beginning. A fundamental assumption in all cases considered in this
project was that deflections and rotations remain sufficiently small, that geometric
nonlinearity could be neglected. For the fine adjustment of optical components this certainly
holds, but in other applications where large-stroke compliant mechanisms are required,
nonlinear effects must be considered at the computational design stage. This adds new
computational challenges both regarding computation time and stability, and research into
this direction is presently ongoing. Quite certainly larger ranges of motion will also place new
demands on material quality and printing accuracy, that are to be addressed by the metal
printing community. Another open question concerns the fatigue life of printed flexures,
particularly those with dimensions near the minimum feature size. The SAMOII specimens
thus far have performed reliably, however a systematic assessment of fatigue performance
also remains a topic for future research.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
14 | 15
Acknowledgements
This work was supported by the Dutch Research Council (NWO), Applied and Engineering
Sciences (AES) project 16191 titled Stable and Adjustable Mechanisms for Optical Instruments
and Implants (SAMOII), and carried out by academic partners TU Delft and Fontys. The support
by project partners TNO, ASML, VDL, Airbus, Nexperia and BAAT Medical is also gratefully
acknowledged.
References
1. Sigmund, O. (2001). Design of multiphysics actuators using topology optimization–Part
I: One-material structures. Computer methods in applied mechanics and engineering,
190(49-50), 6577-6604.
2. Koppen, S., Langelaar, M., & van Keulen, F. (2022a). A simple and versatile topology
optimization formulation for flexure synthesis. Mechanism and Machine Theory, 172,
[104743]. https://doi.org/10.1016/j.mechmachtheory.2022.104743
3. Koppen, S. (2022b). Topology optimization of compliant mechanisms with multiple
degrees of freedom. Delft University of Technology.
https://doi.org/10.4233/uuid:21994a92-e365-4679-b6ac-11a2b70572b7
4. Wang, F., Lazarov, B. S., & Sigmund, O. (2011). On projection methods, convergence
and robust formulations in topology optimization. Structural and multidisciplinary
optimization, 43, 767-784.

17 apr. 24
NIEUWSBRIEF IAMM 03
Auteur(s): TU Delft & Fontys
15 | 15
5. Koppen, S., Hoes, E., Langelaar, M., & Frecker, M. I. (2021). Local redesign for additive
manufacturability of compliant mechanisms using topology optimization. In ASME
2021 International Design Engineering Technical Conferences and Computers and
Information in Engineering Conference : 45th Mechanisms and Robotics Conference
(MR) (Vol. 8A). [V08AT08A002] ASME. https://doi.org/10.1115/DETC2021-67642
6. artofscience. (2020, August 13). Topology optimization of 3D compliant mechanism
flexures [Video]. YouTube. https://www.youtube.com/watch?v=jHqkL6dLo2M
7. Dagenais, S., Caro, J., & Haldeman, S. (2008). A systematic review of low back pain cost
of illness studies in the United States and internationally. The spine journal, 8(1), 8-20.
8. Wu, Z., Prabha Narra, S. and Rollett, A. (2020). Exploring the fabrication limits of thin-
wall structures in a laser powder bed fusion process. International Journal of Advanced
Manufacturing Technology 110: 191–207. https://doi.org/10.1007/s00170-020-
05827-4 .
9. Taban, L., Al Mangour. B., Grzesiak, D., Walunj, G. and Tushar, B. (2021). Effect of Laser
Spot Size, Scanning Strategy, Scanning Speed, and Laser Power on Microstructure and
Mechanical Behavior of 316L Stainless Steel Fabricated via Selective Laser Melting.
Journal of Materials Engineering and Performance 31: 2205–2224.
https://doi.org/10.1007/s11665-021-06387-8.
10. Ehsan, M. and El Mounayri, H. (2021) Scanning Strategies in the PBF Process: A Critical
Review. Proceedings of the ASME 2020 International Mechanical Engineering Congress
and Exposition. Virtual, Online: ASME. https://doi.org/10.1115/IMECE2020-24589.
11. Bendsoe, M., and Sigmund, O. (2013) Topology optimization: theory, methods, and
applications. Springer Science & Business Media, 2013.