PRECISION ADDITIVE METAL MANUFACTURING

Abstract

PAM^2, which stands for Precision Additive Metal Manufacturing, is a European MSCA project in which 10 beneficiaries and 2 partners collaborate on improving the precision of metal Additive Manufacturing. Within this project, research is done for each process stage of AM, going from the design stage to modelling, fabricating, measuring and assessment. For each step we aim to progress the state of the art with a view on improving the final AM part precision and quality by implementing good precision engineering practice.
In this article we will list PAM^2’s detailed objectives, the envisioned approach and the results achieved after 1,5 years of research.

Full text

PRECISION ADDITIVE METAL MANUFACTURING
Ann Witvrouw1, Jitka Metelkova1, Rajit Ranjan2, Mohamad Bayat3, David De
Baere3, Mandaná Moshiri3,4, Guido Tosello3, Juliana Solheid5, Amal Charles5,
Steffen Scholz5, Markus Baier6, Simone Carmignato6, Petros Stavroulakis7,
Amrozia Shaheen7, Lore Thijs8
1KU Leuven, Heverlee, Belgium, Member Flanders Make
2Delft University of Technology, Delft, The Netherlands
3Technical University of Denmark, Lyngby, Denmark
4LEGO System A/S, Denmark
5Karlsruhe Institute of Technology, Karlsruhe, Germany
6University of Padua, Vicenza, Italy
7University of Nottingham, UK
83D Systems, Leuven, Belgium
INTRODUCTION
Additive Manufacturing (AM) is an increasingly
used production method with the ability to evoke
a revolution in manufacturing due to its almost
unlimited design freedom and its capability to
produce personalised parts locally and with
efficient material use [1]. Currently still several
technological challenges remain such as a limited
precision due to shrinkage, build-in stresses and
dross formation at overhanging structures (see
Figure 1) and a limited process stability and
robustness. Moreover post-processing is often
needed as as-processed parts have a high
surface roughness and might have, in some
cases, remaining porosity.
FIGURE 1. Deformation due to tensile stresses
and dross formation in overhanging structures [2].
For AM to become fully accepted by endusers
from different sectors, it must be developed into a
true precision manufacturing method. The
production of precision parts relies on three
principles:
(1) The production is robust, i.e. that all
sensitive parameters can be controlled.
(2) The production is predictable, e.g. the
shrinkage or warpage that occurs is
acceptable because it can be predicted
and compensated in the design.
(3) The parts are measurable, as without
metrology accuracy, repeatability and
quality assurance cannot be known.
Next to these principles ensuring robust &
predictable production, we should also work on
improving the overall quality of the produced part
(density, roughness, precision, ..).
PAM2, which stands for Precision Additive Metal
Manufacturing, is a European MSCA project in
which 10 beneficiaries and 2 partners collaborate
on improving the precision of metal Additive
Manufacturing [3]. Within this project, research is
done for each process stage of AM, going from
the design stage to modelling, fabricating,
measuring and assessment. For each step we
aim to progress the state of the art with a view on
improving the final AM part precision and quality
by implementing good precision engineering
practice.
In this article we will list PAM2’s detailed
objectives, the envisioned approach and the
results achieved after 1,5 years of research.
OBJECTIVES
The overall objective of PAM2 is to ensure the
availability of high precision AM processes and
(computational) design procedures. Detailed
objectives to reach this overall goal are:
(1) to develop advanced (computational)
design tools, enabling competitive
designs, better use of AM possibilities

against minimal design costs and
reduced time-to-market
(2) to develop better modelling tools for
first-time-right processing
(3) to optimize selective laser melting
process strategies for improved part
precision and feature accuracy
(4) to understand the link between post-
process metrology and in-process
observations, creating the basis for in-
process quality control and process
stability
(5) To develop innovative in-process and
post-process techniques to reduce or
remove roughness, porosity and internal
stresses and to improve dimensional
accuracy and mechanical properties.
APPROACH
The objectives listed above will be reached by
progressing the scientific/technical knowledge
beyond the state-of-the-art, through imple-
menting good precision engineering practice, by
cross-linking every step along the process chain
and by providing feedback for each stage after
technology and product assessment. The AM
process steps, the interaction between the steps,
the assessment feedback loop and the role of the
end-users are shown in Figure 2.
FIGURE 2. A shematic overview of the research
done within PAM2 for each step of the AM
process chain and the interactions between these
steps.
The specific metal AM processes that are
investigated in this project are powder bed- and
laser-based fusion (e.g. Selective Laser Melting
or SLM) of maraging steel, TiAl6V4 and inconel
718.
RESULTS
Design
To fully exploit the advantages of AM’s
associated design freedom, topology
optimization is often used to generate an optimal
design for an AM part. Current topology
optimization tools however do not take into
account that local overheating during AM
processing might occur. This overheating can
result in microstructural inhomogenities, defect
formation, poor surface finish and undesired
deformation and/or mechanical properties for the
final AM part. Within the design work package of
PAM2 a computationally inexpensive simplified
thermal model, called a ‘hotspot detector’, is
developed to detect zones of local heat
concentration [4]. Finite element implementation
of the hotspot detector can be integrated with the
density based topology optimization in order to
generate robust AM designs that are expected to
be free of local overheating zones (Figure 3). An
important advantage of this physics based
method over already existing geometry based
approaches is that it incorporates the
temperature response of a geometry instead of
imposing explicit prohibition of overhangs. The
later is found to be ocassionally overrestrictive,
leading to sub optimal designs while in some
cases it is insufficient for constraining local
overheating.
FIGURE 3. Normalized temperature fields
superimposed on designs obtained by (a)
Standard Topology Optimization (TO) (b) Hot
spot constrained TO for minimum compliance
against cantilever loading [4].
Modelling
Within PAM2 modelling is done both for the AM
process and the AM post-process. Full numerical
modelling of an SLM process is normally very
time-consuming [5]. Simulating even a single
layer of a tiny part needs a lot of computation
time. As a result, full numerical simulations
cannot be extended to bigger parts. Because of
this issue, a novel strategy has been developed
within PAM2 in order to enable one to do a full

FIGURE 4. Temperature contour and melt pool
shape at four different times [7]
thermo-mechanical simulation for an actual metal
part made by SLM. The method is based on a
combination of a full thermo-fluid dynamic model
(including Marangoni effects) and a lumped
model which reduces the required computation
time, while keeping an acceptable accuracy
(Figure 4). This way complete parts can be
simulated and possible issues can be detected
before processing, again ensuring that final AM
parts have the quality and precision that is
needed. The model can for example be used to
predict porosity formation [6] and the morphology
of the grains of a Ti6Al4V part produced by
selective laser melting [7].
At the same time experimental trials are carried
out to evaluate the effects of different SLM
parameters on the obtainable part quality (see
next paragraph). The results of these
experiments would facilitate the modelling and
subsequent optimization of the SLM process for
producing highly precise components.
FIGURE 5. The initial (columnar) microstructure
on the left (a) and the equiaxed microstructure
after heat treatment on the right (b) [7].
Finally, cellular automata is used to model the
microstructural evolution during post-processing.
Figure 5 shows the result after a uniform heat
treatment at the beta transus temperature. The
model shows good agreement with earlier
experimental results [7].
Processing
In order to evaluate different AM processes and
machines and also to track the progress within
PAM2, a benchmarking design was made (Figure
6 and 7) [8]. This design will enable to evaluate
accuracy and precision of the machine, residual
stresses on the parts, homogeneity (in terms of
density and residual porosity), build speed,
mechanical properties, surface finish and
corrosion resistance. It also includes features that
represent a challenge for AM, which allows us to
track the new developments on improving the
precision of AM within PAM2.
FIGURE 6. PAM2 benchmark part [8].
FIGURE 7. PAM2 benchmark job [8].
One of these new processes that was developed
within PAM2 is the use of a dynamic focusing unit
in order to combine high productivity for the inner
AM part by using a large spot size (defocused
beam) with high outer precision by using the
smallest spot size (focused beam) [9].
The surface quality and dimensional accuracy of
critical down-facing surfaces produced by SLM
was investigated as well [10]. It was found that
the presence of partially melted powder and
dross formation is the major cause of surface
defects within the down-facing surfaces of parts.
Further results indicate that only looking at the
roughness does not provide reliable information
on the presence and formation of dross.
Dimensional accuracy tests are also required and
are the focus in [11].
100 µm
0.3 ms
0.6 ms
1.0 ms
3.5 ms
a)
b)

Post-processing
AM parts might need post-processing as the final
surface roughness is often higher than what is
allowed for most applications. This roughness
can be reduced by laser polishing. A small
amount of material is first ablated and/or molten
through laser irradiation, and subsequently
redistributed to create a surface with a lower
roughness and sometimes also new functionality.
Laser polishing has a high process speed and the
capability for localized surface treatment.
Within PAM2 both in-process (by using a hybrid
AM process) and post-process laser polishing is
investigated. For the post-process laser polishing
different pulse duration (cw, ns, and fs), scan
speed, repetition rate and average laser power
were used and the impact on the material
surface, the melt pool geometry, and
microstructure was analyzed and correlated to
the process parameters. power at high scanning
speeds (see Figure 8). When ultrafast laser
radiation was used, the surface roughness
increased at low scanning speeds, while for
higher scanning speeds the Ra values were
similar to the initial ones [12]. Better results were
obtained by applying a cw laser source as shown
in Figure 8 [13]. Laser polishing with 50 W
average power and high scanning speeds
FIGURE 8. Roughness Ra as function of scan
velocity and pitch distances for a continuous
wave (cw) laser with 50 W average power (100
μm laser beam diameter) [13].
TABLE 1. Roughness Ra for top and side
surfaces of an AM part as-built and after laser
polishing with a cw laser with 50W average power
and scan velocities of 300 and 400mm/s [13].
Speed
pitch
side
top
25 µm
50 µm
25 µm
As-built
6.4 ± 0.8 µm
6.4 ± 0.8 µm
2.7 ± 0.6 µm
300 mm/s
0,6 ± 0,1 µm
0,6 ± 0,1 µm
0,5 ± 0,1 µm
400 mm/s
0,6 ± 0,1 µm
0,8 ± 0,1 µm
0,9 ± 0,4 µm
reduced the surface roughness of the as-built AM
part (2.7 ± 0.6 µm) to Ra values below 1 μm (for
details see Table 1).
Metrology
In PAM2 both in-process and post-process
metrology is used. For in-process metrology both
the use of optical meltpool monitoring and 3D
geometrical measurements through the use of a
compact focus variation system are envisioned.
For post-process metrology currently both micro-
focus X-ray CT (computed tomography) and
fringe projection profilometry (in combination with
other form measurement techniques) are
investigated. For both methods we aim to obtain
high resolution images in order to assess the
more precise AM parts that will be fabricated in
the PAM2 project.
X-ray CT
In X-ray CT, a sufficiently small focal spot size is
necessary to analyse micro-features and
structures. One example where very high
resolution is required is shown in Figure 9. The
top part (Figure 9.a) of this figure shows a region
of the reconstructed surface of a metal AM part
obtained with high resolution CT in comparison
with the same area inspected by confocal
microscopy. The maximum height of texture
surface in this case is in the range of 100 µm, but
very small features well below 100 µm are clearly
visible. The bottom part (Figure 9.b) of the figure
shows a cross sectional profile measured by X-
ray CT and compared to the reference profile
FIGURE 9. (a) Comparison of AM surface areal
topographies measured by X-ray CT at high
resolution and confocal microscopy; (b)
comparison of an AM surface profile measured by
X-ray CT at high resolution with the reference
profile [14,15].

obtained using an imaging probing system after
cutting and polishing the specimen. Some
surface features, including undercuts (e.g. in the
centre of Figure 9.b) are impossible to inspect
with tactile or optical systems but can be
inspected with CT. To be able to evaluate such
features, it is necessary that the resolution is
sufficiently high, otherwise the analysis suffers
from loss of information and high uncertainties.
One of the main limiting factors of CT for high
resolution inspection is the finite size of the focal
spot i.e. where the electron beam impacts the
target and heat and X-rays are generated. Thus
in the last decade, much effort has been put into
the reduction of the spot size. However, existing
standards [16] cannot fully cope with the
achieved reduction in spot size, why it is
necessary to develop new methods and new
standards to close this gap. A redesigned
resolution test chart with feature sizes in the
range of 36 µm to 100 nm (cf. Figure 10) has
been developed to analyse the focal spot size.
FIGURE 10. New designed test chart for micro-
and nano-focus X-ray source evaluation.
This range has been chosen to cover a broad
range of already existing standards, but also to
be able to measure resolutions which can be
achieved with more recently developed X-ray
sources (see e.g. [17]). The obtained values are
additionally compared to already well established
methods as the slanted edge method or JIMA
resolution charts [18].
Optical techniques
Additionally an accurate optical technique is
being developed for form measurements of PAM2
parts [19-21]. In the artificial intelligence (AI)-
enhanced data-fused optical measurement
framework that is being developed [19] for quick
post-process 3D shape measurements, a combi-
nation of three techniques, namely fringe
projection, photogrammetry and deflectometry,
allows the measurement of multiple types of AM
materials (both specularly and diffusely
reflective). The setup employs a high-resolution
camera and an AI network which can quickly
select the appropriate technique for each part of
FIGURE 11. Fused photogrammetry, fringe
projection and deflectometry data to optimally
measure a Ti6Al-4V pyramid (bottom right) a
Nylon 12 pyramid (top right) and a carbon fiber
plate (middle left).
the areal measurement. This type of setup
reduces the cost and complexity of the system by
combining all three measurement systems into
one framework while enabling an expanded
material application range. We have been
currently able to verify the technique by
successfully fusing the measurements for a
Ti6Al-4V, a Nylon 12 and a carbon fiber object
(see Figure 11).
DISCUSSION and CONCLUSION
Ongoing research for each process stage of AM,
going from the design stage to modelling,
processing, post-pocessing and metrology, was
presented. For each step we aim to progress the
state of the art such that the final AM part
precision and quality is improved. Topology
optimization was executed in such a way that hot
spots, and thus resulting AM part deformations,
are avoided. Modelling is done by combining a full
thermo-fluid dynamic model for accuracy and a
lumped model for speed. This way complete parts
can be simulated and possible issues can be
detected before processing, again ensuring that
final AM parts have the quality and precision that
is needed. For evaluating the process, a novel
benchmark part was made. This part also has
advanced features (small pins and channels) that
are beyond current machine capabilities, but
which we hope to be able to fabricate with the
new techniques to be developed in PAM2.

Combining laser additive with laser subtractive
processsing (either in-process or post-process)
should enable us to go beyond the state-of-the-
art in terms of precision. Finally, if we make
precise parts, we also need to be able to measure
them precisely. In PAM2 we use for example
micro-focus X-ray CT to assess parts with
micron-size features.
ACKNOWLEDGEMENTS
This work was done in the H2020-MSCA-ITN-
2016 project PAM2, Precision Additive Metal
Manufacturing, which is funded by The EU
Framework Programme for Research and
Innovation - Grant Agreement No 721383. All
PAM2 beneficiaries and partners are
acknowledged for their contributions to the
project and the results shown in this article.
REFERENCES
[1] Additive Manufacturing: Strategic Research
Agenda, Consultation document prepared by
the AM sub-platform, 2014.
[2] PhD thesis Peter Mercelis, KU Leuven, 2007.
[3] www.pam2.eu
[4] Ranjan R, Yang Y, Ayas C, Langelaar M, Van
Keulen F Controlling Local Overheating in
Topology Optimization for Additive
Manufacturing. Proc. SIG AM Euspen & ASPE,
2017.
[5] Cheng B, Shrestha S and Chou K.,
2016.Additive Manufacturing, 12, pp.240-251.
[6] Bayat M, Moshiri M, Mohanty S, Jesper. Hattel
J H, Multi-scale multiphysics simulation of metal
powder bed fusion AM process and subsequent
mechanical analysis, Proc. Graz, 2018.
[7] De Baere D, Bayat M, Mohanty S, Hattel J H,
Thermo-metallurgical modelling of the selective
laser melting process chain. Proc. LANE 2018.
[8] Moshiri M, Tosello G, Mohanty S A new design
for an extensive benchmarking of Additive
Manufacturing machines, Proc. 18th Euspen
conference, Venice, Italy.
[9] Metelkova J, Kempen K, Kinds Y, de Formanoir
C, Witvrouw A, Van Hooreweder B, On the
influence of laser defocusing in Selective Laser
Melting of 316L, submitted to Additive
Manufacturing.
[10] Charles A P, Elkaseer A, Mueller T, Thijs L,
Hagenmeyer V, Scholz S G, Investigation into
the influence of process parameters on surface
roughness of down-facing surfaces in Selective
Laser Melting, Proc. WCMNM, Portorož,
Slovenia, 2018.
[11] Charles A P, Elkaseer A, Mueller T, Thijs L,
Hagenmeyer V, Scholz S G , Effect of process
parameters on dimensaional accuracy of down-
facing surfaces in selective laser melting of
Ti6Al4V. Proc. ASPE summer topical meeting in
AM, 2018.
[12] Solheid J, Seifert H J, Pfleging W, Laser-
assisted post-processing of additive
manufactured metallic parts, Proc. 18th Euspen
conference, Venice, Italy, 2018.
[13] Solheid J, Seifert H J, Pfleging W, Laser surface
modification and polishing of AM metallic parts,
Proc. LANE 2018.
[14] Zanini F, Sbettega E, Sorgato M, Carmignato S.
(Submitted). New approach for verifying the
accuracy of X-ray computed tomography
measurements of surface topographies in
additively manufactured metal parts. Journal of
Nondestructive Evaluation.
[15] Zanini F, Sbettega E, Sorgato M, Carmignato S.
(2018). Accuracy of surface topography
measurements performed by X-ray computed
tomography on additively manufactured metal
parts. Proc. 18th Euspen conference, Venice,
Italy, p 465-466.
[16] EN 12543 series (1-5):1999, Non-destructive
testing. Characteristics of focal spots in
industrial X-ray systems for use in non-
destructive testing. European Committee for
Standardization
[17] http://www.excillum.com/products-and-
services/nanotube.html (accessed 25.06.2018)
[18] Baier M, Zanini F, Savio E, Carmignato S.
(2018). A new conversion approach between
different characterization methods to measure
the spot size of micro computed tomography
systems. 18th Euspen conference, Venice, Italy,
p 445-446.
[19] Stavroulakis P, Sims-Waterhouse D, Shaheen
A, Catalucci S, Bointon P, Tzimiropoulos G,
Leach R, Doyen E, Kong Y B, Leach R External
Shape Measurement for Industrial Applications
using Artificial Intelligence and Optimised Data
Fusion. In Proceedings of SAI Intelligent
Systems Conference (IntelliSys), London, 2018
(accepted)
[20] Stavroulakis P I and Leach R K Invited review
article: Review of post-process optical form
metrology for industrial-grade metal additive
manufactured parts. Review of Scientific.
Instruments, vol. 87, no. 4, pp. 0411011–
04110115, Apr. 2016.
[21] Stavroulakis P, Sims-Waterhouse D, Piano S,
and Leach R Flexible decoupled camera and
projector fringe projection system using inertial
sensors. Optical Engineering, vol. 56, no. 10, p.
1, Oct. 2017.