Content uploaded by Vicente Queral
Author content
All content in this area was uploaded by Vicente Queral on May 07, 2018
Content may be subject to copyright.
© 2018 IEEE. Personal use of this material is permitted. Permission
from IEEE must be obtained for all other uses, in any current or future
media, including reprinting/republishing this material for advertising or
promotional purposes, creating new collective works, for resale or
redistribution to servers or lists, or reuse of any copyrighted
component of this work in other works.”
The final version of the preprint shown next is available at DOI :
10.1109/TPS.2018.2790168 and is referenced as
"Prospects for Stellarators Based on Additive Manufacturing: Coil
Frame Accuracy and Vacuum Vessels",
Journal: IEEE Transactions on Plasma Science
Issue Date: MAY 2018
Volume: 46, Issue:5
On Page(s): 1173-1179
Print ISSN: 0093-3813
Online ISSN: 1939-9375
Digital Object Identifier: 10.1109/TPS.2018.2790168
2
Prospects for stellarators based on additive
manufacturing: coil frame accuracy and vacuum vessels
Vicente Queral, Santiago Cabrera, Esther Rincón and Vicente Mirones
Abstract − High geometric complexity at high accuracy
delays the construction of new stellarators and the
related research. Thus, exploring whether additive
manufacturing (AM) might provide advantages to the
fabrication and design philosophy of stellarators appears
valuable. Three coil frame supports were designed,
produced by three AM techniques (SLA, FDM and
PolyJet) and measured by a Coordinate Measuring
Machine. For FDM (similar for SLA), average of ±0.17%
mean deviation was obtained and, 68% of the points (one
sigma) deviate less than ±0.28%. Three assays were
performed for modular stellarator vacuum vessels: i) a
thin copper liner attached to a resin shell cast in an AM
mould, ii) a thin electroformed liner for a test vacuum
vessel, iii) electrodeposited coating on an AM shell. The
assays showed alternative (iii) the simplest and fastest.
The results from such studies and assays are reported
and integrated with previous results. Low stiffness and
strength of AM plastics was previously tackled with
fibre-reinforced resin cast in AM hollow structures
(3Dformwork technique). Thus, additive manufacturing,
particularly combined with other fabrication methods,
proved to be appropriate for the production of certain
unpretentious small and middle size stellarators.
Index terms – additive manufacturing , fusion , stellarators ,
dimensional accuracy , vacuum vessel , electrodeposition
I. INTRODUCTION
The coil supports and vacuum vessels are two essential
mechanical components of stellarators. Casting and milling
are recurrent techniques for the fabrication of coil supports
for modular [1] stellarators, e.g. as employed for NCSX coil
forms [2] and W7-X coil casings [3]. Forming and welding
metallic plates is the conventional technique for the
fabrication of vacuum vessels for modular stellarators. Hot
forming in W7-X [4], cold forming in NCSX [5] and
explosive forming in HSX [6] exemplify the alternatives.
The milling process is difficult and expensive, particularly
for geometrically complex large parts. Accurate casting of
large coil frames is also a specialized work [2].
This work was funded by the Spanish ‘Ministry of Economy,
Industry and Competitiveness’ and the ‘Fondo Europeo de Desarrollo
Regional’ under contract number ENE2015-64981-R (MINECO /
FEDER, EU), project ‘Study of AM for the application to high
performance fusion devices of stellarator type’.
All the authors are with the National Fusion Laboratory in the
‘Centro de Investigaciones Energéticas, Medioambientales y
Tecnológicas’ (CIEMAT), 28040 Madrid, Spain.
This work investigates whether some manufacturing
methods based on additive manufacturing (AM) may be
appropriate for the construction of certain stellarators. If
such methods were feasible, the prospect of faster
production cycle of experimental stellarators and possible
faster advance of fusion plasma science would be
accomplished.
Different AM techniques are available, [7,8]. Among
them, the AM techniques employed in this work are
Selective Laser Sintering (SLS), Stereolithography (SLA),
Fused Deposition Modelling (FDM) and PolyJet. SLS
utilizes a steerable laser to melt and fuse the particles of a
thin layer of plastic or metallic powder. SLA is based on the
localized photo-polymerization of a layer of photopolymer,
typically by a steerable laser. FDM extrudes and deposits a
thermoplastic through a movable nozzle, layer by layer. The
PolyJet technique sprays layers of photopolymer by a nozzle
printing head.
The application of AM to experimental non-nuclear
stellarators is governed by the factors: maximum size of the
AM part, accuracy of the part, properties of the available
AM materials (essentially strength, service temperature,
conductivity and vacuum compatibility) and cost.
The strength of AM plastic parts (maximum ~ 80 MPa,
[9]) is increased by casting short-fibre-reinforced resins
inside hollow AM parts, a technique known as 3Dformwork,
[10,11].
Resistivity comparable to wrought copper is reported for
Electron Beam Melting AM parts [12,13]. The possibilities
of such (still) rather experimental techniques to directly
produce coil windings are not evaluated in the paper.
Maximum size of AM printers, tensile strength of
materials and cost of AM parts, were preliminarily studied in
[10,11,14]. They are not investigated in the present work.
Coil supports and vacuum vessels are the two main
components in a stellarator device. Therefore, in order to
assess the viability of AM for stellarators, the work is
focussed on the study of experimental dimensional accuracy
of coils frames and on the assays performed for
unconventional fabrication methods for fusion vacuum
vessels.
Section II briefly reviews previous studies on accuracy
of AM coil frames, Section III summarizes vacuum vessel
production techniques in previous stellarators, Section IV
stablishes the methods for the dimensional metrology study,
the results from such study are reported in Section V, the
assays on uncommon fabrication methods for vacuum
vessels are described in Section VI, the prospects of
stellarators based on AM are analysed in Section VII and
conclusions in Section VIII.
3
II. PREVIOUS STUDIES ON ACCURACY OF AM
COIL FRAMES
The dimensional accuracy of additively manufactured parts
was previously studied, both in the fusion field and in non-
fusion areas. In the fusion field, in particular for stellarators,
the study [14] by the same authors reported the dimensional
accuracy results for:
i) A twisted coil frame of length ~ 350 mm. The coil
frame was produced by SLS in polyamide in a web-based
AM company. The mean deviation between nominal
dimensions for all the measured points was ±0.23% (relative
to a reference dimension of 250 mm) and 68% of the
measurements (one sigma) deviated lower than ±0.31%.
ii) Subsequently, four identical non-circular planar coils
of major radius ~ 290 mm were produced in different
materials and measured. Web-based AM in SLA resin
proved ±0.1% mean deviations, and 68% of the points (1σ)
deviated lower than ±0.17%. Professional FDM in ABS
(Acrylonitrile butadienestyrene) achieved ±0.07% mean
deviation and ±0.11% deviation for 68% of the
measurements (1σ). Linear and warping deviations [14] are
accounted.
The PolyJet planar coil showed dimensional instability
[14], though brochure specification of accuracy (∼ 0.05%
for large parts) is the highest among all the common AM
techniques.
Dimensional accuracy from 0.15% to 0.56% resulted in
different investigations on non-fusion AM components,
which are summarized in [14].
III. PREVIOUS VACUUM VESSEL PRODUCTION
TECHNIQUES
Concerning the second main stellarator component, the
vacuum vessel, common and experimental production
techniques are summarized next. Such techniques are a
comparison and, in the case of the fibre-reinforced epoxy
vacuum vessel, an inspiration for the initial work on AM of
vacuum vessels for stellarators reported in Section VI.
For modular stellarators, the vacuum vessel shell for
NCSX stellarator was produced [5] by three identical 120º
vessel segments made of Inconel, each one comprising two
welded 60º segments. Each 60º segment is fabricated by
welding ten press-formed panels (press-form, trim borders,
annealing, and second press-form) together over a 60º
welding internal fixture. The vacuum vessel for W7-X
stellarator [4] is composed of five almost identical vessel
modules made of stainless steel plate, each one comprising
two welded half-modules. Each half-module is produced by
welded poloidal rings of 1.8º toroidal width, each one
composed of four hot-formed welded plates. In HSX
stellarator the vacuum vessel modules were produced by
explosive forming, [6].
For subatomic particle experiments, two test vacuum
vessels were produced by fibre-reinforced epoxy resin which
was copper coated by a proprietary galvanic procedure [15].
Ultra-high vacuum was obtained. However, the thin film
(~ 20 µm) detached after harsh performance tests.
IV. METHODS, FOR DIMENSIONAL METROLOGY
The measurements are performed by a Coordinate
Measuring Machine (CMM), brand Mitutoyo (Fig. 1), model
EUROC-574 APEX of maximum permissible error (MPE)
2.2 + (3.0 L / 1000) µm. This measuring method is available
in the CIEMAT research centre and is highly accurate. For a
part dimension of L ~ 150 mm results a MPE of ~ 3 µm. The
bounding box of the studied coil frames is
175 mm × 140 mm × 130 mm.
Dimensions of stellarator coils and their positioning
should achieve deviations lower than 0.1% for satisfactory
plasmas [6,16]. A deviation of 0.1% for a dimension of
100 mm corresponds to 0.1 mm. Thus, appropriate
measurements should have uncertainties lower than
±0.02 mm.
A moderate number of points, of the order of 200, will
be measured for each part.
Only global dimensions of the coil frames are measured,
which includes the external contorted surface and the legs of
the coil frame. Small elements like holes and narrow
grooves or rims achieve an AM tolerance of ±0.2 mm to
±0.3 mm in printers for the part size investigated here [14].
Such deviations are caused by the stepped nature of the
external surfaces of the AM objects, originated by the layer
by layer deposition. Therefore, small elements achieve low
relative accuracy, while keeping the properties of the
deposited material.
The coil frames are measured as received from the AM
supplier, that is, before filling them with fibre-reinforced
resin.
The Meshlab Hausdorff distance [17] between the
measured points and the model is computed. The Meshlab
Hausdorff distance computes the distance from each
measured point to the closest point on the model. The model
is stored in STL file-format and it defines the nominal
dimensions of the part.
Fig. 1. CCM, coordinate system and FDM coil frame.
A. Measuring procedure
The coil frames were measured from one to two months
after fabrication. The temperature of the CMM room was
24±1ºC during the measurements.
The next steps are followed:
1. The coil frame is located on the CMM support table (Fig.
1) with the wider legs on the table.
2. The CAT1000 code (a module for the MCosmos code
from Mitutoyo) for the CMM is utilized to align the
coordinate system of the measurements with the
coordinate system of the model. The coordinate system in
4
CAT1000 code is located (Fig. 2): i) on the CMM
support table (plane Z=0), ii) aligned with the points F0-
F4 (X axis) and, iii) with the Y-axis passing thought Point
C. Point C is a hole located at one leg (Fig. 2) which can
be measured accurately by the CMM.
3. The coordinate system in the STL format model is
located at the same position as in the CAT1000 code.
4. The points to measure are automatically probed by the
code CAT1000 operating on the CMM.
5. Measurements of points are taken on the upper surface of
the top legs, on the lateral sides at the top legs, at the
outboard and inboard of the winding surface, and at the
frontal and back surfaces, see Fig. 3 to Fig. 6.
6. Some points are measured on positioning elements
(Fig. 2): on the three Positioning stops, and on the two
Positioning bulges at the front of the part (points F0…F4).
7. The fitting of the model with the measured points is
performed by locating the model in assembling position
based on the measured points. The points taken on the
positioning bulges are used as main reference.
Fig. 2. Elements for the positioning of the model with respect the
measured points.
V. RESULTS FROM DIMENSIONAL METROLOGY
A. Features of the sample coil frames
Three almost identical coils frames (Fig. 3 to Fig. 6 ) were
produced in different materials and companies, Table I.
Their size is half the size of the coil frames for the UST_2
stellarator [10, 18]. The internal toroidal surface of the SLA
coil frame (Fig. 3) is not printed for the FDM and PolyJet
parts, so to allow the extraction of the AM support material.
Fig. 3. SLA coil frame scaled 1/2 with internal surface (left). FDM
coil frame scaled 1/2 without internal surface whose reinforcement
bars are visible (right).
The SLA coil frame generates a hollow shell to be filled
with fibre-reinforced resin (3Dformwork). The FDM and
PolyJet parts require installation of an internal toroidal
surface to allow resin casting. The structure of the shell and
internal reinforcements are described in [10,18].
Table I compiles for each part: the AM technology, the
bounding box dimensions, material, brochure tensile
strength, and supplier. The price of each coil frame ranges
from 450 € to 730 € depending on the AM technique.
B. Results from the measurements
Table II summarizes the results of the measurements for two
coil frames. The PolyJet coil frame was not measured since
some dimensions distorted more than 10% (see Sec. V.E).
The major radius of the stellarator UST_2 is R = 295 mm
and the diameter of the modular coils at the lower field
region is ~ 200 mm. For coil frames scaled down 1/2, an
average of 125 mm is reasonably considered the reference
dimension to calculate per cent values.
TABLE I. DATA ABOUT THE COIL FRAMES
Tech.
Box (mm)
Mate
rial
TS
(MPa)
Supplier
Ref.
suppl
SLA
175×140×130
Resin
48
Provel Srl./ 3D
systems
[19]
FDM
175×140×130
ABS-
M30
32
Gerg 3D
[20]
PolyJet
175×140×130
F720
50
Non-disclosed
F720 ≡ Fullcure 720.
The SLA and FDM coil frames were measured according to
Sec. IV. The surface roughness was measured at the top
surface of the top legs (Fig. 1) to evaluate its impact on the
uncertainty on the measurements. A standard
deviation σ = 0.006 mm (mean 115.20 mm) for SLA, and
σ = 0.01 (mean 114.48 mm) for FDM was measured from a
set of various observations on a small surface. Thus, a
minimum measuring uncertainty of ±0.02 mm will appear
(two sigma, 95% certainty level) for the measurements on
FDM parts. SLA part is smoother. Additionally, some
surfaces of FDM parts are stepped due to the notable
diameter of the extruded plastic. The measured dimension of
the steps is 0.2 mm for the FDM part. It adds variability to
the measurements on the FDM part. The SLA uncertainty is
acceptable for part deviations ~ 0.1% and dimension
L > 100 mm, Sec. IV.
Table II shows the numerical results for the two
measured coil frames.
TABLE II. RESULTS FROM THE MEASUREMENTS.
n.
Group name
N.
sam
ples
Mean
Haus.
(mm)
Mean
%
σ
(mm)
Max.
(mm)
Max.
%
1
SLA Vertical
40
0.25
0.20
0.03
0.30
0.24
2
SLA Lateral-Legs
117
0.25
0.20
0.13
0.46
0.37
3
SLA Out&Inboard
40
0.12
0.10
0.09
0.27
0.22
4
SLA Front&Back
34
0.27
0.22
0.20
0.54
0.43
5
FDM Vertical
37
0.10
0.08
0.03
0.16
0.13
6
FDM Lateral-Legs
110
0.11
0.09
0.09
0.37
0.62
7
FDM Out&Inboard
65
0.41
0.33
0.23
0.92
0.74
8
FDM Front&Back
37
0.22
0.18
0.23
0.93
0.74
Meaning of fields in Table II:
5
N. samples: Number of measured points in each group of points.
Mean Haus.: Mean MeshLab Hausdorff distance of all the points in
the group.
Mean %: Ratio ‘Mean Haus.’ to reference dimension (125 mm).
σ : Standard deviation of the mean Hausdorff distance.
Max.: Maximum deviation in the group.
Max %: The respective per cent deviation of ‘Max.’ values.
Vertical: Results from the points measured on the upper surface of
the top legs, Fig. 4.
Lateral-Legs: Results from the measurements at the laterals of the
top legs, Fig. 5.
Out&Inboard: Results from the measurements at the convex and
concave winding surface, corresponding to the outboard and
inboard of the torus, Fig. 6.
Front&Back: Results from the measurements at the front and back
of the part.
One point measured for the FDM Lateral-Legs (row 6,
Table II) was discarded since it was located on a residual
extruded plastic.
C. Results for the SLA coil frame.
Fig. 4 to Fig. 6 show the graphical representation of the
results for the SLA coil frame. Reddish points correspond to
the maximum deviation and bluish points to the minimum
deviations as indicated at the top of each figure.
The SLA part resulted ~ 0.2% larger than the model in
the longitudinal Y dimension and in vertical Z dimension,
see Table II and Fig 4 to Fig. 6. The X dimension resulted
fairly accurate.
Average of ±0.18% mean deviation for all the points (n.
1 to n. 4 in Table II) is obtained and, 68% of the points (one
sigma) deviate less than ±0.27%.
Fig. 4. Hausdorff distance from the measured points at the top
surface of the legs to the model for the SLA coil frame.
Fig. 5. Deviations at the laterals of the top legs (Hausdorff
distance) for the SLA coil frame.
D. Results from the FDM coil frame.
Fig. 7 shows the graphical representation of the results for
the FDM coil frame. Colours are expressed as in Fig. 4 to
Fig. 6. Maximum value is indicated in column ‘Max. (mm)’
in Table II.
Fig. 6. Outboard and inboard deviations (Hausdorff distance) for
the SLA coil frame.
Average of ±0.17% mean deviation for all the points (n.
5 to 8 in Table II) is obtained and, 68% of the points (one
sigma) deviate less than ±0.28%.
A fraction of the deviations is originated by the stepping
of some surfaces of the FDM part, particularly the
Front&Back (points n. 8, Table II). The low ‘Mean %’
deviation and σ for the smother FDM Lateral-Legs (n. 6,
Table II) indicates an accurate longitudinal Y dimension.
The large deviation for the FDM Out&Inboard, 0.33%, is
originated mainly by the inboard points, Fig. 7. The supplier
advised on possible problems on circularity on this part, and
this phenomenon may be the cause of increased deviations
in this group of measurements.
Fig. 7. Deviations for the ‘Group names’ indicated in Table II
(rows n. 6 − 8) for the FDM part.
E. Results from the PolyJet coil frame.
The PolyJet coil frame distorted more than 10% for some
dimensions. The phenomenon occurred during a storage
6
period of about one month from fabrication to intended
measurement. The environmental instability of the Fullcure
720 PolyJet material was already noted previously [14] but
this second assay was attempted due to the potential high
accuracy of the PolyJet technique.
F. Discussion of results
The company [19] produced the SLA coil frame and
specified fabrication tolerances of 0.2% for a part of size <
200 mm. Thus, the specification was fulfilled for mean
values (achieved ±0.18%). The same company guarantees
tolerances ±0.15% for parts larger than 200 mm. The current
results provide confidence on the feasibility of such high
accuracy by SLA technique.
The company [20] produced the FDM coil frame and
specified fabrication accuracy of 0.15% for 95% of the
points. This specification was almost fulfilled for mean
values (achieved ±0.17%). For 1σ it cannot be verified from
the present work due to the stepped surface of the small
FDM part, which adds variability to some groups of
measurements, Table II.
Repeatability of the dimensions for the different half-
periods of the stellarator is essential to keep the stellarator
symmetry. It will slightly reduce the minimum required
accuracy of 0.1% for stellarator coil frames. The study of the
repeatability of AM for SLA and FDM techniques remains
for a future work.
The two parts produced by PolyJet AM technique (one
planar coil reported in [14] and the coil frame in this work)
evidenced environmental dimensional instability.
Thus, SLA and FDM parts produced with maximum
quality are at the verge of achieving enough accuracy to
produce coil frames for stellarators.
VI. ASSAYS ON FABRICATION METHODS FOR
VACUUM VESSELS
New alternatives for vacuum vessel manufacturing are
explored to try to find improved manufacturing methods for
certain stellarator vacuum vessels. Traditional vacuum
vessel production techniques are summarized in Section III.
Three alternatives were conceived and assayed:
1. Thin liner, which comprises welded/brazed/soldered
rings, attached to a resin shell. The resin is cast over the
external surface of the liner, inside an AM mould.
2. Thin electroformed liner attached to a resin shell. The
resin is cast as in Alternative 1.
3. Electrodeposited thin film bonded to an additively
manufactured shell.
Indeed, electroforming and electrodeposition are a type of
additive manufacturing of atomic scale.
A. Welded liner attached to a resin shell
This concept is based on the utilization of a thin liner, which
can be formed manually on an additively manufactured
form. The liner is reinforced externally by epoxy resin. The
liner is externally roughed or externally furnished with
welded/brazed/soldered claws in order to obtain reliable
attachment between the liner and the resin. The use of a liner
clamped to a structure, usually of concrete, is typical for
large vacuum vessels, i.e. the hot cell for the IFMIF facility
[21].
This concept avoids the dies for cold/hot forming and the
welding of thick plates. As drawbacks, a casting process is
required and leak testing is hindered.
The concept was assayed on a vacuum vessel sector for
the UST_2 stellarator. The vacuum vessel for UST_2 is
modular and comprises six identical curved vacuum vessel
sectors. The next particular methods and materials were used
for the assay. A plaster form, which is shaped as half
vacuum vessel sector (Fig. 8-left) is generated from AM
plastic mould. Proper thin (0.3 mm) copper strips are cut.
They are manually shaped on the form and soldered among
them. Brazing or e-beam welding are advanced possibilities.
The two half-liners (Fig. 8-right) are soldered longitudinally.
Subsequently, a clamping element (brass ball chain in this
case) is soldered on the external surface of the liner. Leak
testing and leak repair is carried out. The soldered seams
among strips are externally covered by thin foam strips to
allow leak testing after covering the liner with epoxy resin.
Finally, the liner is covered by an AM mould (Fig. 9-left)
and the volume is cast with epoxy resin, Fig. 9-right. The
procedure required ~ 80 hours of skilled work.
Fig. 8 Plaster form and shaping process (left). Half-liner of
vacuum vessel (right).
Fig. 9. AM mould for resin casting (left). Finished vacuum vessel
sector (right).
B. Thin electroformed liner attached to a resin shell
This concept is based on the utilization of a thin liner created
by electroforming [22] on a painted mandrel, and externally
reinforced as in Alternative 1. The liner is devised thin
(unable to withstand the atmospheric pressure) since it is
difficult and costly to electroform thick (several mm
thickness) walls if they are geometrically complex. For
example, a thick (6 mm) electroformed beam dump cone
7
[23] was electroformed, but successive lathing phases were
required to keep the thickness tolerance.
The same drawbacks as in Alternative 1 affect to this
alternative.
The concept was assayed for a UST_2 vacuum vessel
sector scaled down two, with the next particularities (Fig.
10): A wax mandrel (created in an AM mould) was covered
with conductive paint and 0.3-0.5 mm of copper was
electrodeposited on the mandrel by means of the mainstream
copper acid bath [22].
Fig. 10. a) Wax mandrel. b) Mandrel painted with graphite and
silver paint. c) Electrodeposited copper.
C. Electrodeposited coating on an additively manufactured
shell.
A metallic coating is electrodeposited on conductive paint
interior to an AM shell.
Proven techniques exist for plating of plastics [22].
However, they imply complex processes, including
palladium activation of the surface and electroless
deposition (a chemical auto-catalytic process) [22], which
are common for copious batches of parts, i.e. in the
automobile industry, but are rare for single large parts.
Therefore, conductive paint is attempted. As drawbacks,
difficult surface finish for certain conductive paints,
challenging electrodeposition of the interior of the shell and
unbounding of the coating are the main problems of this
alternative.
The concept was preliminarily assayed for a UST_2
vacuum vessel sector, with the next particularities: A sector
shell was produced by an SLA AM plastic liner externally
reinforced by epoxy resin (Fig. 11). A thin layer of epoxy
resin was painted on a portion of the interior of the shell and
conductive paint (made of copper particles coated by silver)
was sprayed on the uncured resin. Finally, copper
electrodeposition on the paint, Fig 11. The bond withstood
vacuum in the vessel.
Fig. 11. Copper coating electroposited on a painted AM vacuum
vessel shell.
D. Discussion of the results from the assays
The simplest and fastest method was deduced from the
assays. Alternative 1 is highly intensive in operator work,
mainly dedicated to welding/soldering the strips and for leak
testing. Alternative 2 required lower operator time than
Alternative 1 for the production of the liner, though the time
needed for wax operations (moulding, wax melting/removal
and cleaning) was notable. Alternative 3 is the most
promising since it involves simple processes that can be
produced quickly in numerous companies.
All three alternatives increase the difficulty of vacuum
leak tests since the resin reinforcement hinders a common
He leak test. Other types of external reinforcement, like
welded nerves, much increases the fabrication time.
VII. PROSPECTS FOR STELLARATORS BASED ON
AM
The previous studies carried out by the authors [10,11,14,18]
together with the results reported in the current work are
integrated in this section to elucidate the prospects for
stellarators based on additive manufacturing.
Mean deviations of 0.17% are measured for a
geometrically complex test coil frame. Considering that: i)
differences among coil frames produced in the same printer
and the same conditions (repeatability) will experience
lower comparative deviations, and ii) larger parts are
specified for tolerance of ±0.15% for SLA, and iii) larger
FDM parts would reduce the variability of the measurements
due to stepped surfaces, there is funded confidence on the
feasibility of achieving enough accuracy to build stellarators
with minimum accuracy (0.1% accuracy).
The strength of AM plastic parts can be increased to
more than 100 MPa by casting short-fibre-reinforced resins
inside hollow AM parts (3Dformwork). Studies in [10]
supports the viability of such method. As an enhancement, if
the feasibility of laminated resins shaped by additive
manufacturing moulds/shells were demonstrated at moderate
cost, strength of the parts would reach more than 1000 MPa.
Certainly, plastics are unsuited for vacuum vessels for
reactors and uncertain for reactor coil supports, but still
many non-nuclear experiments, particularly in the broad
field of stellarator configurations, remain to be produced.
The crucial property of additive manufacturing of
delivering high complexity at fixed cost is paradigmatic for
stellarators. Certainly, all the elements (legs, supports,
positioning elements for the coil frame, fixation for the
conductors during modular coil winding, assembling holes,
etc.) can be inexpensively and quickly produced on a sole
part. Consequently, such approach accelerates the
assembling process and thus, reduces assembling and
winding cost. References [10,18] show evidence of such
potential.
Service temperature of common high temperature epoxy
resin is ∼200 ºC, ultrahigh temperature potting epoxies
withstand 300ºC [10] and the heat deflexion of the best AM
plastics reach 225º C [9]. It appears sufficient for many
applications.
AM parts are more than 4 metres long [10], some
produced in non-commercial experimental 3D-printers.
8
Therefore, size is not a constraint for small or middle size
stellarators.
Numerous coils of extremely curled shape and small
curvatures, and any conceivable contorted magnetic
configuration can be produced additively without extra cost.
Such possibilities reduce the magnetic modular ripple and
open an avenue for new magnetic configurations.
Indeed, uncertainties remain. Still long-term dimension
instability of the fibre-reinforced plastics has not been well
studied for stellarator coil frames. Accuracy of AM parts
remains at the lower required accuracy for coil frames. High
price of AM metallic parts, insufficient accuracy of metallic
AM parts and relatively small commercial 3D-printers for
metals, presently hamper their competitiveness for coil
frames and vacuum vessels. Moreover, AM, perhaps
combined with traditional fabrication methods, still require
further studies, assays and demonstration for vacuum
vessels.
VIII. CONCLUSIONS
The accuracy of AM plastic parts has been proved almost
enough for coils frames for stellarators (0.1% minimum
accuracy). Middle tensile strength of materials is
accomplished by the 3Dformwok method investigated by the
authors, which involves casting short-fibre-reinforced resins.
Potential for high tensile strength, superior to 1000 MPa, is
foreseen for an extension of the 3Dformwork method using
laminated fibre-reinforced resins instead of cast resins. The
size of 3D-printers is large enough for stellarators. The
global cost of the device is reduced since all elements are
produced on a single part. Therefore, logistics, coil winding
and assembling costs are reduced.
Thus, despite the low suitability of plastics for fusion
reactors and the still uncompetitive metal AM for most of
the stellarator components, it is concluded that additive
manufacturing is already globally advantageous for the
construction of certain unpretentious small or middle size
experimental stellarators.
ACKNOWLEDGMENT
The authors would like to thank the ‘Unidad de Fabricacion
y Apoyo a I+D’ for the use of the CMM in CIEMAT.
REFERENCES
[1] M. Kikuchi, K. Lackner, M. Quang (editors), ‘Fusion Physics’,
ISBN 978–92–0–130410–0, IAEA, Vienna, September 2012, p.859.
[2] PPPL Digests, ‘Advanced Technology Essential in NCSX
Component Fabrication’, February 2007.
[3] K. Risse K, Th. Rummel, L. Wegener, R. Holzthüm, N. Jaksic, et
al., Fabrication of the superconducting coils for Wendelstein 7-X,
Fusion Engineering and Design 66-68 965–969, 2003.
[4] J. Reich, W. Gardebrecht, B. Hein, B. Missal, J. Tretter, et al.,
Manufacture of the vacuum vessels and the ports of Wendelstein 7-X,
Fusion Engineering and Design 75–79 (2005) 565–569.
[5] M.E. Viola, T. Brown, P. Heitzenroeder, W. Reiersen, P.
Goranson, et al., NCSX vacuum vessel fabrication, Proceedings of 21st
IEEE/NPS Symposium on Fusion Engineering (SOFE 05), Knoxville,
Tennessee (USA), September 2005.
[6] A. F. Almagri, D. T. Anderson, S. F.B. Anderson, K. M. Likin, P.
G. Matthews, et al., Design and Construction of HSX: a Helically
Symmetric Stellarator, Journal of Plasma Fusion Research SERIES 1
(1998) 422–425.
[7] I. Gibson, D.W. Rosen, B. Stucker, Additive Manufacturing
Technologies, Ed. Springer, 2009, ISBN 978-1-4419-1120-9.
[8] Kaufui V. Wong and Aldo Hernandez, A Review of Additive
Manufacturing, ISRN Mechanical Engineering, 2012.
[9] Materialise Datasheet, March 2016, p. 1-22.
[10] V. Queral, 3D-printed fusion components concepts and validation
for the UST_2 stellarator, Fusion Eng. Des. 96–97 (2015) 343–347.
[11] V. Queral, Concept, production and validation of a 3D-printed
coil frame for the UST2 modular stellarator, Fusion Engineering and
Design 89 (2014) 2145–2149.
[12] P. Frigola, O.A. Harrysson, T.J. Horn, H.A. West, R.L. Aman, et
al. Fabricating Copper Components with Electron Beam Melting,
Advanced Materials & Processes 20-24, July 2014.
[13] Timothy Horn, ‘Material Development for Electron Beam
Melting’, Presentation from the Center for Additive Manufacturing and
Logistics, North Carolina State University, 2013.
[14] V. Queral, E. Rincón, V. Mirones, L. Rios, S. Cabrera,
Dimensional accuracy of additively manufactured structures
formodular coil windings of stellarators, Fusion Engineering and
Desing In Press, DOI:10.1016/j.fusengdes.2016.12.014.
[15] F. Burri, M. Fertl, P. Feusi, R. Henneck, K. Kirch, et al., Copper
coated carbon fiber reinforced plastics for high and ultra high vacuum
applications, arXiv:1308.2563 [physics.ins-det], 12 August 2013.
[16] J.H. Feist, M. Wanner and the W7-X construction team, Status of
Wendelstein 7-X construction, in: 28th EPS Conference on Controlled
Fusion and Plasma Physics, Funchal (Portugal), 18–22 June 2001, ECA
25A (2001) 1937–1940.
[17] ‘MeshLab documentation’ manual for MeshLab code, 2017.
[18] V. Queral, Construction concepts and validation of the 3D printed
UST_2 modular stellarator, Journal of Physics: Conference Series 591
(2015) 012015.
[19] Company ‘Provel Srl.’ A 3D Systems society, Via Roberto Incerti
25, 10064 Pinerolo (TO), Italy.
[20] Company Gerg 3D, Baiern/Piusheim (Germany) www.gerg3d.de.
[21] V. Heinzel, P. Bem, E. Esposito, S. Gordeev, U. Fischer, A.
Moeslang, et al. Overview on the IFMIF Test Cell development,
Journal of Nuclear Materials 329-333 A (2004) 223–227.
[22] M. Schlesinger, M. Paunovic,’Modern Electroplating’, John
Wiley & Sons, ISBN 9781118063149 (2011) 736pp.
[23] F. Arranz, B. Brañas, D. Iglesias, O. Nomen, D. Rapisarda, et al.,
Manufacturing prototypes for LIPAC beam dump, Fusion Engineering
and Design 89 (2014) 2199–2203.
