Presentation in IAEA RUSFD congress, 2014

3D printed modular stellarator UST_2 Vicente Queral L 1

3D printed modular stellarator UST_2

Vicente Queral*

* On leave of absence from NFL, CIEMAT


▪ Background

▪ UST_2 conceptual design

▪ Engineering concepts and design

▪ Experimental validation by construction of UST_2 ▪ Future work

Outline


► The work is developed in a framework of very low funds. Therefore, the relevance of the work resides in the developed methods, not in the size of the device or the plasma performance.

► Up to now only private funds have been obtained for the development and construction of UST_2.

► The work is R&D and innovation in engineering, focused in new construction methods for stellarators.

► The geometrical complexity of stellarators is one of their main drawbacks. Faster and lower cost methods to build stellarators may contribute to increase the potential of stellarators as experimental devices and as future fusion reactors.

Background


► UST_2 is a small modular three period quasi-isodynamic stellarator of major radius 0.26 m and plasma volume 10 litres, being built by 3D printing (additive manufacturing) in Spain. Not only a concept!.

► Objectives: - Develop new better (faster, cheaper, simpler) construction methods for stellarators and other fusion devices.- Validate the construction method by the construction of a small stellarator. - Demonstration effect.- Contribute to my PhD thesis.

Background. UST_2 essential data

Construction status


▪ UST_2 is not the first stellarator I built by innovative construction methods.

▪ UST_1 stellarator was designed, built and operated from 2005 to 2007 in my personal laboratory.

• Cost of materials for the whole facility ~ 3000 € (many 2nd hand pieces).

• The coils were built by a new toroidal milling machine.

Background

UST_1 stellarat

or

UST_1 facility


UST_2 conceptual design


• The aim is to use as much as possible the current physics designs.

• The LCFS of QPS, QIPC2, QIPC3, QIPC6 and NCSX-TU (turbulence improved, [Myn 10]), have been received from researchers and studied.

• The 3 periods Quasi-isodynamic stellarator with poloidal closed contours (QIPC3) has been chosen as starting point.

Physics design

LCFS supplied by J. Nühremberg and team [Mik 04] QIPCC3


Design enhancement

Generate wide ports for fast in-vessel access, diagnostics and divertors

Movable planar non-circular coils for fast and wide in-vessel access and diagnostics. Also space for future large and powerful divertors. Concept similar to [Kul 06], [Ima 11] and [Spo 10]

Wide opening Complex CASTELL code optimization processes using also NESCOIL and DESCUR codes


Specifications

UST_2 properties

Vacuum magnetic surfaces at φ = 0 and Iota profile (from CASTELL code)

Element Specification

Number of periods 3Plasma volume (litres) 10 R, plasma major radius (mm) 260a, ave. plasma minor radius (mm)

~ 37

Aspect ratio ~ 7Bo Magnetic field at axis (T) 0.045 /

0.089 / Higher

ι0 , rotational transform at axis 0.74ιa , rotational transform at edge 0.69Vacuum max. magnetic well 0.2%


Engineering concepts and design


UST_2 specifications

Coil engineering specifications

Element SpecificationType of coils Modular coilsNumber of coils 90Number of non-planar coils 84 (14 x 6) Number of large planar non-circular coils

6 (1 x 6)

Winding pack size (mm) 4 with x 12 depth Conductor type Flexible copper wire

TXL 10 AWG gauge (5mm2)

Turns per pancake 3 layers x 1 turn/lay. = 3


Two concepts developed: Hull and Truss ConceptsHull Concept Two parallel surfaces, similar to the LCFS of the plasma, form a double hull structure. Bars reinforce the structure. The internal volume is filled with a material able to solidify.

Results : Robust, accurate.

3D printed nylon costs about 1-2 € /cm3, still expensive. Consumption of 3D printed material has to be minimised

Test of the Hull Concept

Test of engineering concepts


Two concepts developed: Hull and Truss Concepts

Test of a coil frame sector

Truss ConceptLight structure formed by bars. Coil grooves as 4 parallel 3D curves.The volume is wrapped and filled with a filler.

Results: Cheap but warped, long manufacturing time

3D printed pieces as received

Sketch of the truss structure

Test of engineering concepts


UST_2 engineering design

A mix of the Hull and Truss Concepts was selected

Definitive engineering design of three coils (the internal surface is removed to observe the support bars)

3D printed internal truss structure and external thin surface envelop. Filled with a filler ~ strength.Two parts.

3D printed half-period as produced

Turns compressed in groove to allow fast and easy winding


UST_2 engineering design

One of the six halfperiods

Two halves of the 3D printed halfperiod, before filling with resin

The halfperiod is split in two halves so as to introduce the vacuum vessel in the frame.

3D printed half-period filled with acrylic resin. 700€. A coil is wound


Experimental validation by

construction of UST_2


Validation of 3D printed half period

• Measured dimensional errors (design/real piece) after moulding < 0.3%.

• Dimensional differences among half-sectors might be < 0.1% if the same 3D printer and position of piece is used.

• Result : The 3D printing method is preliminarily considered satisfactory.

Several measurements in x, y, z have been performed

3D printed piece accuracy


Half period prepared for moulding with a

fillerPossible filler materials are studied • Acrylic, epoxy and polyurethane resins, and hard plasters, fibre reinforced and not, have been studied and tested.

• One half period has been filled with acrylic resin (simpler use than epoxy and urethanes. Enough strength).

• Result : Fast moulding (1-2 hours). Complete filling if the piece is internally free of nylon powder.

Validation of 3D printed half period

Some of the materials and samples studied

and tested


Test of coil windings

Testing the crossover

performance

Compression on walls and crossovers• Two coils have been fully wound (the coils require winding after the vacuum vessel is inside the coil frame, not yet).

• Results : - Reasonable pressure of conductor on groove walls.- One coil was wound in about 30 minutes, OK. - The conceived crossover was feasible and satisfactory.

Finished crossover

Test coil


Vacuum vessel construction

Vacuum vessel : Still unclear• 3D printing of metals is still very expensive (about 5-10 times more expensive than nylon per cm3).

• Large 3D printers are uncommon for Ti, Al or Cu and non-magnetic alloys.

• Combination of 3D printing and traditional methods (soldering, welding, laser cutting, etc) are being considered.

Test of soldering segments

on a mandrel

One of many concepts. VV as unfolding surfaces, reinforced


Status of UST_2 construction










Future work


Current status and next future work

Initial tests performed Decision of device to build Conceptual design Detailed design Construction 20%Final validation

Short term : ~ 3 - 4 months▪ Try to rise more funds.▪ Finish the construction and final validation by e-beam field mapping.

Present status


Long term aims

Sequential low-cost rapid manufacturing of larger devices

Notes : - Cost and performance is only a coarse value for rough comparison among devices.- PIGNITRON = Pre-IGNITRON

Ultimate aim: High-field pulsed Allure Ignition Stellarator (AIS). [Que 10] High-field, few ignition pulses. Somewhat similar to the IGNITOR (Alcator), FIRE, CTF concepts, but for a stellarator.

PIGNITRON

IGNITRON


Summary and conclusions

► The potential of 3D printing methods to built small or middle size stellarators is being studied.

► A stellarator, UST_2, is being built to test the methodology, accuracy and possible issues.

► Three concepts combined: 3D printing, truss structures and in situ moulding are combined so as to reduce fabrication costs keeping accuracy.

► If the result is satisfactory, more laboratories or universities might build stellarators for plasma research.


Acknowledgement

I would like to give thanks to all the people and researchers helping in the development, in particular:

Jefrey Harris, Donald Spong and team (ORNL, QPS LCFS and coils)Juergen Nueremberg and team (IPP Max-Planck, QIPCCs LCFS)H. E. Mynick (PPPL, NCSX-TU LCFS)Jesús Romero (NESCOIL teaching, other)Antonio Lopez-Fraguas (DESCUR code update and teaching)Gerardo Veredas (CAD teaching) Juan A. Jiménez (VMEC teaching)Víctor Tribaldos (stellarators)Jose A. Ferreira (vacuum)Cristobal Bellés (I. T. help)Other


3D printed modular stellarator UST_2 Vicente Queral L 32More information in www.fusionvic.org


[Mik 04] “Comparison of the properties of Quasi-isodynamic configurations for Different Number of Periods”, M. J. Mikhailov et al., 31st EPS Conference on Plasma Phys. London, 28 June - 2 July 2004 ECA Vol.28G, P-4.166 (2004).[Myn 10] “Reducing turbulent transport in toroidal configurations via shaping” H. E. Mynick et al., PHYSICS OF PLASMAS 18, 056101 (2011), December 2010 [Kul 06] “Project EPSILON – a way to steady state high b fusion reactor”, V.M. Kulygin, V.V. Arsenin, V.A. Zhil’tsov, et al., IAEA XXI Fusion Energy Conference, 16 -21 October 2006, Chengdu, China.[Ima 11] “Status and plan of gamma 10 tandem mirror program”, T. Imai, et al., TRANSACTIONS OF FUSION SCIENCE AND TECHNOLOGY VOL. 59 Jan. 2011[Que 10] “High-field pulsed Allure Ignition Stellarator”, Stellarator News, n. 125, 2010[Spo 10] “New QP/QI Symmetric Stellarator Configurations”, Donald A. Spong and Jeffrey H. Harris, Plasma and Fusion Research: Regular Articles, Volume 5, S2039 (2010)

References



Extra slides


Grooves mechanised in the plaster frame

Compressing and placing

conductors in the groove

Winding process

Concept

Implementation

Internal crossover and

auxiliary winding coil

(black conductor)


Implementation of the toroidal milling machine concept

12 grooves 7 x 12 mm were mechanised in the plaster frame. Each groove lasted about 2 hours

Method to build the modular coils

Detail


Finished UST_1 stellarator

12 coils finished

Almost finished


Low-cost coil metal casting tests

&D carried out to support the decisionsResults : Inconclusive. Casting not chosen for UST_2

• The coils, the coil frame, the VV, might be casted.• Metal casting tend to be expensive for few units.• For small series (~<10 units) sand casting (non-permanent mould) is the most common and cheaper.• A permanent plaster mould has been tested.

Silver lost wax vacuum casting in plaster mould produced in a specialised company.~ 1000 € in Ag. ~700 € in Cu

~100 mm

Own test of casting in a “permanent” plaster mould. The mould broke. However, some ideas appeared


Process of modification of QIPCC3

The straight section is stretched by CASTELL code, plus re-optimization• Automatic CASTELL code processes: The QIPCC3 straight section is stretched (addition of poloidal cuts and compression of QIPCC3 sections), CASTELL DESCUR-like code application, two NESCOIL runs, confinement, iota and magnetic well profiles calculated by Monte Carlo method. • Only about 500 configurations have been compared. Long lasting computations. • Increasing elongation of the straight section gave decreasing confinement for the best configuration.• The re-optimization is poor (about 3 times less confinement than the original QIPCC3). However, the main objective is engineering.

Stretched and compressed poloidal cuts


Iota ~[0.16 , 0.26] without Ip. A ~ 2.7From CASTELL and VMEC

Several devices have been assessed

QPS• The device is optimised for β~2% but βUST_2 ~0%.♦ Relatively poor confinement obtained for β=0% (to be confirmed by better codes). ♦ Too compact to allocate inboard blankets, if reactor.► Potential improved turbulence transport.►Decision: Not chosen for UST_2

β=2% VMEC Free boundary run

LCFS and winding surface

LCFS, winding surface and coils supplied by J. Harris and D. Spong


QIPCC2


Tip

Using DESCUR code the LCFS was rounded

QIPC6 Large aspect ratio. Not chosen

♦ Considered too large helical excursion.

♦ Configuration with LCFS tips. Unfeasible coils.

► Acceptable confinement.

► Potential improved turbulence transport.

►Decision: Not chosen for UST_2

LCFSs supplied by J. Nühremberg and team



QIPCC3► LCFS and plasma varies little with β.► High confinement obtained for β=0% (to be confirmed by better codes). ► Middle compactness. ► High iota.

►Decision: Chosen for UST_2

Retrospectively:

♦ The distance coil-LCFS must be low, otherwise coil shape is unfeasible or confinement worsen

LCFS supplied by J. Nühremberg and team

Iota [0.67 , 0.71] A~6.8From CASTELL ,[Mik 04],VMEC


NCSX-TU

NCSXMix


β=4%+Ip VMEC-Fix

TipLCFS with tips [Myn

10]

Two overlapped winding surfaces and

coils

Poor surfaces. Require many

coils, very near the LCFS for good

result

♦ The device is optimised for β=4%

+Ip but βUST_2 ~0%.

♦ This particular configuration has

tips.• Complex overlapping

► Potential improved turbulence transport.

►Decision: Not chosen for UST_2

LCFS supplied by H. Mynick

LCFS in vacuum for NCSX and NCSX-TU, Mixed


Neoclassical transport estimation/comparison of possible devices for UST_2. Particle confinement time, from CASTELL code. Er=0. (to be confirmed by well-validated codes)

Several devices assessed

Thinking both in UST_2 size and reactor. Difficult balance of:-Neoclassical confinement.-Potential turbulent confinement.-Alpha particle confinement.-Middle compactness (~inboard blanket).-Simple control (~↓currents,↓shift, …).-Reasonable coil shape and space.-LCFS tips ~ cost ~ performance.-Cost.

Science

Presentation in IAEA RUSFD congress, 2014