MUSE Nuclear Fusion Stellerator Made with Off the Shelf Parts and 3D Printed Shell

The Princeton Physics Lab has built a Stellarator nuclear fusion reactor prototype using permanent magnets. This is a scientific first that enables mostly off the shelf magnet for simple and low cost experiments to test new concepts for future fusion power plants.

It does not generate net energy, but it has some nuclear processes and it is useful for experiments. 100 times cheaper but it is a device that is useful for experimental physics and nuclear fusion tests.

Stellarators typically rely on complicated electromagnets that have complex shapes and create their magnetic fields through the flow of electricity. Those electromagnets must be built precisely with very little room for error, increasing their cost. However, permanent magnets, like the magnets that hold art to refrigerator doors, do not need electric currents to create their fields. They can also be ordered off the shelf from industrial suppliers and then embedded in a 3D-printed shell around the device’s vacuum vessel, which holds the plasma.

“MUSE is largely constructed with commercially available parts,” said Michael Zarnstorff, a senior research physicist at PPPL. “By working with 3D-printing companies and magnet suppliers, we can shop around and buy the precision we need instead of making it ourselves.”

They purchased more than 10 000 magnets at commercially available specifications: 5 % tolerance on magnetisation magnitude, 3∘ tolerance on magnetization orientation and 1 mm tolerance in physical dimensions. Multi-jet fusion 3D-print technology was selected for fabricating the PM holders using nylon plastic.

Using the magnetic surface charge method, the static assembled force between the 4 PM (permanent magnet) holders is calculated to be less than 1000 N. Most of the inter-magnet forces are carried as internal stress in the PM holders. Loading these forces into ANSYS, the peak internal stress was found to be less than 7 MPa, which is safely below the 30 MPa tensile strength of our 3D-print material. Finite permeability effects were also modelled. They found that the small but finite changes to stellarator metrics can be compensated for by slightly adjusting the TF coil current.

Journal of Plasma Physics -Design and construction of the MUSE permanent magnet stellarator

Abstract-
This paper documents the design and construction of MUSE, the world’s first permanent magnet (PM) stellarator and the first quasi-axisymmetric experiment. The purpose of MUSE is to develop and assess a new way of building optimized stellarators that uses simple planar coils PMs. Our PM optimisation algorithm consists of initializing a geometry to pack dipoles densely, running the FAMUS code to minimise surface field error subject to PM constraints and applying discrete jumps to reach a physically realisable solution. FAMUS treats the PM system as a set of ideal point dipoles. From there we construct finite-volume magnet towers to be housed in 3D-printed PM holders. We describe the design of the PM holders, which were validated by laser metrology. We analyze the effects of finite permeability, sensitivity to perturbations and magnetostatic forces. An exact analytic formula for the magnetic field from a finite-volume PM tower is presented to compute PM–PM forces and stress on the PM holder. Stellarator construction is complete and experiments are underway.

With present materials, the poloidal field produced by an individual PM (without the additive effect of Halbach arrays) is less than 1 Tesla.

Coils are better at producing large fields when they are larger, but PMs do not. To make a PM stellarator with larger field, one must increase the PM array’s relative thickness, in addition to photographically scaling the size of the array. One solution to this problem is optimising PM together with mildly shaped coils to partially alleviate coil complexity. This is a staged retreat: start with tilted circular coils, then go to shaped but still planar coils and, finally, lightly shaped coils with non-zero torsion. An issue often raised is the viability of PM in a neutronic environment. The Curie temperature of neodymium magnets is above 500 K. Superconducting coils must be cooled well below 100 K, which is possible only when the superconductor is kept in a cryostat behind a neutron shield. Provided that neutron shielding and cryostat must exist for the TF coils, one can put the PM behind the blanket and pick up extra magnetization from the cryogenic temperature. Whether or not the PM stellarator becomes a fusion reactor, the stellarator concept has many paths to choose from, and the PM approach offers a quick and low-cost path to rapid exploration.

As a next step, they propose the design of new ‘MUSE-like configurations’ defined such that the PM field is produced by single-orientation magnets assembled around a simple circular cross section. They envision a mid-scale experiment CERBERUS ( 𝑅=1 m, 𝐵=1 Tesla) in which three vastly different stellarators (e.g. QA, QH and QI) can be tested in one facility using interchangeable PM sets and a common set of TF coils. Since the axis shape for QH (helical) differs from QA (circular) and QI (polygonal) it may be necessary for CERBERUS to use multiple sets of VVs, but the circular TF coils can still be shared.

Another natural extension of the PM stellarator concept is the use of ‘window-pane coils’ with adjustable current. These form a surface array that does not link the magnetic axis. They can be made from a high-current-density material such as REBCO high-temperature superconductors (HTS). A PM tower produces a magnetic field externally indistinguishable from that of a square solenoid. In MUSE, the magnetisation (one million amps per meter) would be replaced by approximately 30 kA of current around a 0.5×0.5×2.5 cubic centimeter volume. This would not be feasible mechanically, due to the small scale and large stress. For larger systems, however, the required current can be reached by existing HTS (high temperature superconductor) technology. For example, the SPARC Toroidal Field Modular Coil (TFMC) carries 40 kA across a stack of HTS tapes. The key is that the equivalent solenoid for a larger PM requires the same current. Therefore, it could be feasible to make window-pane coils using a larger footprint than that of MUSE to produce of the order of few-tesla fields. This preserves the coil-simplifying properties of the PM stellarator while addressing the two issues of PMs: not being able to be turned off during construction and not being able to be manipulated during experiments. Two independent proposals exploring this concept were presented at the recent APS meeting.