Planar Coil Stellarator

• Planar coil stellarators are magnetic confinement devices that employ flat coils to generate helical, toroidal, and poloidal magnetic fields for stable plasma control.
• Advanced optimization methods, including sparse regression and autodifferentiation, minimize field errors and enforce quasisymmetry, leading to improved device performance.
• The simplified, manufacturable coil geometries enable scalable, cost-effective designs with robust operational flexibility in reactor-scale applications.

A planar coil stellarator is a class of magnetic confinement device employing exclusively or predominantly planar (flat) coils—rather than fully three-dimensional modular windings—to generate the helical, toroidal, and poloidal magnetic field components necessary for confining high-temperature plasmas under conditions suitable for nuclear fusion. This architecture seeks to combine the optimized neoclassical and MHD stability properties of advanced stellarator configurations with significant reductions in coil manufacturing complexity, cost, and maintainability, leveraging the geometric and logistical simplicity of planar coil geometries. Recent research demonstrates that with sophisticated optimization (notably sparse regression and autodifferentiation-based force minimization), planar coil arrays—when properly sited and individually powered—can achieve quasisymmetric or quasi-isodynamic equilibria, plasma performance metrics competitive with modular-coil-based stellarators, and scalability to reactor-relevant designs (Wu et al., 11 Feb 2025, Kaptanoglu et al., 2024, Swanson et al., 8 Dec 2025).

1. Planar Coil Architectures: Conceptual Advances and Classification

Planar coil stellarators have evolved from early figure-8 designs to highly engineered configurations explicitly optimized for neoclassical confinement, stability, flexibility, and engineering integration. The fundamental distinction is the restriction of current-carrying elements to lie within planar surfaces—either simple geometric circles, rounded rectangles, or other convex planar shapes—sometimes augmented by minimal nonplanar (but still geometrically simple) elements for enhanced field tailoring.

Major architectures include:

• Encircling/TF planar arrays: Large, toroidally extended planar coils, analogous to traditional tokamak TF coils, set the bulk toroidal field and principal linkage (Swanson et al., 8 Dec 2025, Nash et al., 19 Mar 2025).
• Shaping arrays: Arrays of hundreds of planar dipole ("windowpane") or circular coils, positioned on a conformal surface outside the plasma, generate the non-axisymmetric (helical and poloidal) field components responsible for confining the plasma and enforcing quasi-symmetry (Wu et al., 11 Feb 2025, Kaptanoglu et al., 2024).
• Minimal/compact planar sets: Designs such as the Zhejiang Compact Stellarator (ZCS) employ only four planar coils (two interlocked ellipses and two circles) to produce nested flux surfaces and excellent neoclassical performance (Yu et al., 2021, Salguero-Martínez et al., 4 May 2026).

A particularly notable realization is the two-field-period, quasi-axisymmetric "Helios" power plant, which utilizes 12 large encircling planar coils and 324 shaping coils to yield a 1.1 GW_th / 390 MW_e reactor system with reactor-scale dimensions and operational lifetimes (Swanson et al., 8 Dec 2025).

2. Magnetic Field Theory and Optimization Metrics

Magnetic field generation is governed by the Biot–Savart law, with the total field at point $\mathbf{x}$ due to $N$ filamentary coils given by:

$$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$

Design optimization requires matching $\mathbf{B}(\mathbf{x})$ and its harmonics (in Boozer or cylindrical coordinate expansions) to a desired target on a prescribed plasma boundary or reference surface. The principal quality metric is the surface-averaged normal field error:

$$B_n(\mathbf{x}) = \mathbf{B}(\mathbf{x}) \cdot \mathbf{n}(\mathbf{x}), \qquad B_{n,avg} = \frac{1}{\|s\|_1} \|(A i - b) \circ s\|_1,$$

where $A$ is the "induction matrix" mapping currents to field-normal error, $b$ encodes boundary conditions, and $s$ are area weights (Wu et al., 11 Feb 2025).

Optimization objectives additionally include coercing quasi-symmetry, as quantified in Boozer coordinates by the minimization of non-axisymmetric field strength harmonics $B_{m,n}$ with $m \neq 0$, or directly via local $N$0 or $N$1 volume integrals (Salguero-Martínez et al., 4 May 2026, Kaptanoglu et al., 2024). The effective ripple $N$2 is minimized to suppress $N$3 neoclassical transport (Yu et al., 2021, Salguero-Martínez et al., 4 May 2026).

3. Planar Coil Optimization Methodologies

Recent work employs advanced optimization frameworks:

• Sparse regression/regularization: The current distribution across an extensive planar coil array is optimized for field quality and coil count using LASSO ($N$4), relax-and-split ($N$5), mixed-integer quadratic programming (MIQP), and heuristic "optimize-then-delete" (OtD) approaches. This yields a Pareto front between coil sparsity and $N$6 error, with MIQP achieving up to 20% lower mean $N$7 at equal sparsity (Wu et al., 11 Feb 2025).
• Automatic differentiation and force/torque minimization: Reactor-scale optimization jointly minimizes field error, net and pointwise coil-coil forces, and torques using autodifferentiation (JAX implementation), ensuring mechanical loads are compatible with superconducting technology and avoiding unbalanced force distributions typical of naïvely placed coil sets (Kaptanoglu et al., 2024).
• Multi-configuration optimization for flexibility: Inclusion of planar control coils during modular coil optimization enables experimental flexibility across multiple values of rotational transform $N$8 ("flexible stellarator"), while preserving flux surface volume and reasonable quasisymmetry (Lee et al., 2022).
• Free-boundary MHD equilibrium: Sophisticated equilibrium solvers such as DESC allow single-stage optimization of coil shapes, plasma boundary, and target profiles (including finite pressure), directly coupling field quality (via $N$9 or $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$0) to coil parameters (Salguero-Martínez et al., 4 May 2026).

Manufacturability, robustness to misalignment, and integration with feedback control are routinely incorporated, especially in power plant designs (Swanson et al., 8 Dec 2025, Nash et al., 19 Mar 2025).

4. Demonstrated Devices and Experimental Validation

Prototyping campaigns, such as the "Canis" 3×3 HTS planar coil array (Nash et al., 19 Mar 2025), have experimentally validated the ability of planar shaping coils to produce stellarator-like magnetic field topologies with $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$1 RMS field error relative to target, and closed control loops utilizing in-vessel Hall sensor diagnostics. This proof-of-concept demonstrates not only field accuracy but also manufacturability—via SMI (soldered-metal-insulation) coil constructions, mass-producibility, and sandwich assembly—and operational resilience (self-protection, rapid control convergence) at 20 K and 3.7 T peak self-field per shaping coil.

At reactor scale, Helios leverages individually powered planar HTS shaping coils (324 units at 20 T max field) to restore quasi-symmetry and provide real-time error correction, with design targets of $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$2 and a 40-year neutron fluence limit for coil lifetime (Swanson et al., 8 Dec 2025). Engineering analysis confirms the feasibility of such arrays for plasma boundary control, breed-and-shield function allocation, and sector-based maintenance.

Compact devices such as the ZCS and the four-coil quasi-axisymmetric systems illustrate that with precise optimization—even with as few as four planar coils—excellent neoclassical transport, low ripple ($$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$3), and stable nested flux surfaces are achievable (Yu et al., 2021, Salguero-Martínez et al., 4 May 2026).

5. Physics Performance: Quasisymmetry, Transport, and Stability

A central achievement of planar coil stellarators is the demonstration that simple coil sets, when properly optimized:

• Enforce quasi-axisymmetry or quasi-isodynamicity, as shown in the first planar-coil quasi-isodynamic device (Plunk et al., 2024), and QA equilibria with reactor-relevant aspect ratios (Swanson et al., 8 Dec 2025).
• Achieve nested flux surface volumes that exceed those of classical modular-coil benchmarks (e.g., NCSX), with low quasisymmetry error ($$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$4 on axis) and suppressed $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$5 neoclassical diffusion by one to two orders of magnitude relative to unoptimized circular-coil configurations (Lee et al., 2022, Yu et al., 2021).
• Control the rotational transform $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$6 via tuning of planar shaping coil currents, including multi-state flexibility (up to three targeted $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$7 values on a common coil set) and transform inversion for enhanced profile control (Lee et al., 2022, Salguero-Martínez et al., 4 May 2026).
• Realize robust MHD equilibrium and stability, as evidenced by β-stable configurations at up to $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$8 and absence of ballooning instabilities below $$\mathbf{B}(\mathbf{x}) = \sum_{i=1}^N I_i \frac{\mu_0}{4\pi} \int_{C_i} \frac{d\boldsymbol\ell' \times (\mathbf{x} - \mathbf{x}')}{|\mathbf{x} - \mathbf{x}'|^3}$$9 (Swanson et al., 8 Dec 2025).

Finite-β effects and pressure-driven expansion require coupled optimization, as observed in both compact and reactor-scale devices. There is a clear trade-off between field quality, transform, volume, and the number/placement of planar coils.

6. Engineering, Manufacturability, and Operational Integration

Planar coil architectures confer significant practical benefits:

• Simplicity and cost: Coil winding, inspection, and structural support for planar (especially convex) coils are straightforward, supporting modular field shaping units (FSUs) and sector-based maintenance (Swanson et al., 8 Dec 2025, Nash et al., 19 Mar 2025).
• Cryogenics and high-field operation: Use of REBCO-based HTS in simple flat geometries enables high current densities ($\mathbf{B}(\mathbf{x})$0) at $\mathbf{B}(\mathbf{x})$1 K and $\mathbf{B}(\mathbf{x})$2 T, permitting compact, high-performance systems (Swanson et al., 8 Dec 2025, Nash et al., 19 Mar 2025).
• Robustness: Sparse shaping arrays permit reduced part count, fewer current leads, and higher tolerance to manufacturing imperfections or misalignments (mean $\mathbf{B}(\mathbf{x})$3 error change decreases as sparsity increases) (Wu et al., 11 Feb 2025).
• Active control: Individual shaping-coil supplies, coordinated via real-time field feedback, facilitate magnetic equilibrium correction and plasma scenario flexibility, particularly when paired with diagnostic integration (Nash et al., 19 Mar 2025, Swanson et al., 8 Dec 2025).
• Mechanical integrity: Minimization of net and pointwise Lorentz loads via optimization ensures coil and support structure stresses remain below HTS and steel limits; comprehensive FEA modeling is employed at reactor scale (Kaptanoglu et al., 2024, Swanson et al., 8 Dec 2025).

Challenges remain in wiring/cooling integration for arrays of hundreds of coils, enforcing minimum coil–plasma/shield distances, and maintaining high field accuracy at large scale. The planar arrays cannot, by themselves, generate net toroidal flux; separate encircling or TF planar coils are required for complete functionality.

7. Outlook: Design Trade-offs and Open Challenges

The adoption of planar coil stellarator architectures is motivated by manufacturability, cost, and maintainability, but subject to physical and engineering trade-offs.

• Planarity vs field quality: Restricting coils strictly to circles or other simple planar shapes fixes degrees of freedom but can be partially compensated by placement, orientation, and current optimization; allowing shape Fourier modes enhances field quality at some cost to manufacturability (Kaptanoglu et al., 2024).
• Coil count vs robustness: Sparse optimization (MIQP, LASSO, relax-and-split) enables flexibility in coil count and current allocation, balancing complexity with error resilience (Wu et al., 11 Feb 2025).
• TF-coil complexity vs dipole arrays: Planar shaping coil arrays relieve the engineering burden on TF coils by allowing shorter, smoother TF coil design and enabling ripple correction, but require more superconducting material overall; reactor-scale studies demonstrate TF-coil length reductions by $\mathbf{B}(\mathbf{x})$420–60 m and field accuracy at $\mathbf{B}(\mathbf{x})$5 (Kaptanoglu et al., 2024).
• Flexibility and multi-configuration capability: Simultaneous optimization for flexibility in $\mathbf{B}(\mathbf{x})$6 and plasma shape is feasible with relatively few (e.g., 2 per field period) planar control coils, but nested flux surface volume and absolute quasisymmetry degrade as $\mathbf{B}(\mathbf{x})$7 is pushed higher (Lee et al., 2022).

A plausible implication is that planar coil stellarators are poised to deliver competitive physical performance alongside practical engineering advantages, but full realization at reactor scale demands further advances in integrated electromagnetic, mechanical, and plasma boundary optimization, enhanced by rapid diagnostic feedback and automated coil manufacturing (Swanson et al., 8 Dec 2025, Kaptanoglu et al., 2024).

---

Key references:

• Planar coil sparse optimization and Eos design: (Wu et al., 11 Feb 2025)
• HTS planar array prototyping: (Nash et al., 19 Mar 2025)
• Helios power plant with planar coil arrays: (Swanson et al., 8 Dec 2025)
• Force-optimized planar arrays at reactor scale: (Kaptanoglu et al., 2024)
• Minimal planar coil QA optimization: (Yu et al., 2021, Salguero-Martínez et al., 4 May 2026)
• Planar control coils for configuration flexibility: (Lee et al., 2022)
• First planar coil quasi-isodynamic design: (Plunk et al., 2024)