euspen’s 18th International Conference &
Exhibition, Venice, IT, June 2018
A new design for an extensive benchmarking of additive manufacturing machines
Mandaná Moshiri1; Guido Tosello1; Sankhya Mohanty1
1 Department of Mechanical Engineering, Technical University of Denmark
This paper focuses on a new methodology for conducting a comprehensive benchmarking of Additive Manufacturing (AM)
technologies. The quality of the built products using AM strongly depends on the machine capabilities, and it is thus essential to
develop a proper benchmarking design that would allow their comparative evaluation. The benchmarking presented has been
designed with the purpose of conducting a comparison between different AM machines, with a particular focus on metal powder-
bed AM. The main scope is to make an extensive evaluation of the technologies from multiple points of view, covering: accuracy and
precision of the machine, residual stresses on the parts (particularly important in the case of metal AM), homogeneity (in terms of
density and residual porosity), build speed, mechanical properties, surface finish and corrosion resistance. For each evaluation
criteria, a specific analysis method is employed. The aim of this work is to analyse the current technology capabilities and limitations,
in order to assess what different AM machines can deliver in a net-shape process chain scenario. The benchmark is employed for a
statistically designed series of experiments to study in detail the AM machine´s real limitations and their working process windows.
The design also includes features that represent a challenge for the AM machine, and sometimes exceed the machine´s actual
capabilities. Furthermore, the benchmark has been developed to be used as a periodic quality control-job for the operational
performance of the AM machines.
Additive Manufacturing, Selective Laser Melting, Powder Bed Fusion, Benchmarking, Technology evaluation, Accuracy, Repeatability, Homogeneity.
Additive manufacturing (AM) refers to a group of processes
that, starting directly from the CAD model, produces parts by
building the material layer by layer. These manufacturing
techniques are theoretically capable of producing components
of any shape in any material [1,2]. The main aim of using AM in
industrial environment is to produce parts in a net-shape
manner, thus avoiding expensive and time-consuming post-
processes. Currently, net-shape fabrication is not possible yet,
because of the limiting capabilities of current machines.
However, the AM machine that best meets the need can be
identified, and the best way for such identification is a
benchmarking study [3,4]. An extensive review on
benchmarking artefacts has been done by Rebaioli et al. .
In the following paper a new design is proposed, with the focus
on applicability of the metal powder bed fusion AM process
named Selective Laser Melting (SLM), for moulds production for
injection moulding with the aim to understand what currently
can or can not be done.
Considering the near impossibility of defining a single standard
benchmarking artefact, also in terms of dimensions, that can be
used for the evaluation of the newest AM technologies, instead
in the following paper the approach towards building such
artefacts will be presented. The particular focus is on SLM,
specifically to understand the current capabilities and
limitations. For this task, a single benchmarking artefact is not
enough, as the complexity of these processes makes it difficult
to summarise all attributes of interest in the design of a single
part, and instead it is necessary to define a whole benchmarking
job. The following picture is the Design of Experiment triangle to
show the strategy of the benchmarking job presented in this
paper (Fig. 1)
Figure 1. Design of Experiment for the benchmarking.
To limit as much as possible the open variables, the process
(SLM), the design (which follows) and the material (a metal alloy)
have been locked and will be the same for all the experiments.
The only open variable is the technology i.e. the machine under
evaluation. Since only the SLM process is evaluated, none of the
parts will be post-processed, apart from cutting them from the
building platform - and the effects of such support/part removal
will be taken into consideration while evaluating the results. The
parameters for the design of the benchmarking job, and the best
analytical technique for their evaluation, are as follows:
1. Accuracy – dimension: Features as in Fig. 2-3, i.e. holes in
different directions, pins, thin walls cross-shaped, unsupported
pyramids and a writing, are used to assess what the machines
can or can not do. The minimum dimension of the features
overpass the publicly acknowledged limits of the current most
capable machine, as showed in Table 1. The column “min dim”
is the minimum dimension producible from a machine in the
market, as declared by the manufacturer. The preferred analysis
technique for dimensional accuracy is the 3D scanner, wherein
the smallest feature defined on the benchmarking job is still
measurable. Capability to build without support structures (i.e.
overhangs) will be evaluated through 3 hollow pyramids with 3
inclinations of the sides (45°, 35° and 25°). The example of the
artefact is in Fig. 2-3.
2. Accuracy – roughness: The final surface roughness is also
very important, considering post-processes, and will be analysed
with a contact profilometer as well as contactless equipment
depending on the quality of the parts. The measurement will be
done in x, y, 45°, and z direction.
Table 1. Minimum dimension of the feature in the artefact.
Min dim in the artefact
Circular holes (diameter)
Circular pins (diameters)
3. Precision-repeatability: The same parts will be placed in 5
different positions of the building platform (in the corners and
center) and the entire job will be repeated 3 times. The overall
job will look like Fig. 4. For tracking repeatability, all the 3D scans
of the benchmark artefacts, across the different positions, jobs
and machines, will be compared.
4. Homogeneity: The residual porosity in the parts, in terms of
density (with Archimedes test on 15x15x10 mm sample) and
porosity percentage on a polished surface in x-y and z plane
(from a 30x30x20 mm sample), will be used as the indicator.
5. Residual stress: To evaluate the stresses emerging from the
manufacturing, two techniques will be used. The first technique
is to measure the eventual distortion on a long, unsupported,
thin wall (49 mm long, 7 mm high, 0,3 mm width) with the 3D
scanner. The second technique is XRD analyses to quantify the
surface residual stress of the part.
6. Tensile properties: Four tensile test samples (DS/EN ISO
6892-1:2016) printed in z-direction will be used for comparison.
7. Corrosion resistance (ISO 9227:2006): Tests in artificial
atmospheres, using cut-outs from a cylinder (20 mm diameter,
60 mm height), both on the bottom and top of the part, will be
performed considering the anisotropy of the building process.
8. Mould features production: Considering a possible final
application of interest, tool manufacturing for injection
moulding, a feature has been designed to gain insight into the
capacity of the machine in building complex shapes. Three
hollow spirals, with different internal diameters (1,0 mm, 2,5
mm & 5,0 mm) and constant thickness (1 mm), have been
chosen as they resemble conformal cooling channels in moulds.
The analysis method are a fluid flow test, a visual evaluation
using an endoscope camera, and a surface roughness
measurement using computer tomography on the cut parts.
9. Time: Build speeds for the different machines will also be
Simultaneously achieveing excellent results on all chosen
parameters is difficult for current machines, and a trade-off
chart is an expected outcome from such benchmarking activity.
The positions of the parts on the platform (Fig. 4), has been
chosen considering the direction of movement of the recoater.
Figure 3. Top view of the artefact identifying the parameters analysed.
Figure 4. Top view of the benchmarking job containing all the samples.
3. Preliminary results
The proposed benchmarking job has already been successfully
built with an EOSM270 in maraging steel grade 300. From a first
visual analysis of the artefacts, it was possible to identify clearly
the minimum dimension of the features that could be built (Fig.
5) i.e. the smallest thin walls and pins were found missing.
Figure 5. Benchmarking artefact produced in maraging steel.
4. Discussion and conclusion
In this paper, a new approach for performing benchmarking
that can be easily adapted to all AM machines has been
presented. It consists in a benchmarking sequence of jobs,
containing different samples for a global evaluation of the
performance of the technology. The main requirement is to
adapt the dimension of the feature suggested to the technology
under analysis. The results obtained would present the machine
capabilities and limitations directly.
The benchmarking job proposed can also be adapted and used
as a tool for the periodic check of the machines, to control their
The project has received funding from the European Union´s Horizon
2020 Marie Skłodowska-Curie grant agreement No 721383, for the
PAM^2 (Precision Additive Metal Manufacturing) project.
 Gibson I, et al. 2010 Additive manufacturing technologies, 2nd ed.
(New York: Springer)
 Vayre B, et al. 2012 Mech. Ind. 13 89–96
 Moylan S, et al. 2012 A review of test artifacts for additive
 Rivas-Santos V M, et al. 2017 Proc. euspen/ASPE Dimensional
Accuracy & Surface Finish in AM, Leuven, Belgium, Oct 24-27
 Rebaioli L, Fassi I 2017 Int. J. Adv. Manuf. Technol. 1–28
Residual stress-thin wall
Figure 2. Benchmarking artefact