arxiv: 2604.22728 · v1 · submitted 2026-04-24 · ⚛️ physics.plasm-ph

Recognition: unknown

Revisiting confinement scalings and fusion performance with a perspective optimized for extrapolation

Egemen Kolemen, Geert Verdoolaege, Jalal Butt, Stanley M. Kaye

Pith reviewed 2026-05-08 09:06 UTC · model grok-4.3

classification ⚛️ physics.plasm-ph

keywords tokamak confinementH-mode scalingplasma currentfusion extrapolationreactor performanceITPA databasefusion powermetallic wall penalty

0 comments

The pith

Fusion triple product scales approximately as plasma current squared in tokamaks

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper reexamines the updated ITPA global H-mode confinement database from an explicitly extrapolation-focused viewpoint rather than pure fit optimization. It identifies low-order empirical models with three to four terms as striking the best balance between capturing observed variance and avoiding overfit for future use, with plasma current, machine size, heating power, and elongation as dominant factors plus a metallic-wall penalty. Recast as direct performance measures, both the fusion triple product and fusion power then show primary dependence on plasma current with near-quadratic scaling across the highest-performing discharges from several machines. This perspective matters because reactor performance projections remain anchored to such scalings, and the results indicate that compact high-field designs may need higher currents than standard models suggest to reach target output. The analysis therefore supplies revised guidance on the engineering levers required for reactor-grade fusion.

Core claim

By systematically searching for minimally complex confinement scalings that optimize the tradeoff between variance capture and extrapolative robustness, the authors determine that models centered near N=3 to N=4 perform best, with plasma current, machine size, heating power, and elongation emerging as the main engineering levers together with an empirically inferred metallic-wall penalty. When translated into reactor-performance terms, the fusion triple product scales approximately as I_p squared while the empirical fusion power scaling exhibits a similarly near-quadratic dependence on plasma current over the survey of top discharges. Projecting to reactors, these results suggest that high-f

What carries the argument

Extrapolation-optimized low-order empirical confinement scaling that identifies plasma current as the dominant term for fusion performance metrics

If this is right

The fusion triple product scales approximately as I_p squared.
Empirical fusion power shows a near-quadratic dependence on plasma current across high-performing discharges.
High-field devices with metal walls may require higher plasma current than standard IPB98(y,2) projections imply.
Gigawatt-class tokamak performance likely demands operation at I_p greater than or equal to 20 MA.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Design studies for compact reactors could shift emphasis toward maximizing achievable plasma current alongside field strength.
The metallic-wall penalty may compound with impurity or divertor physics in ways that require separate experimental checks at higher parameters.
Cross-validation against other confinement databases could test whether the quadratic dependence persists when additional variables are included.

Load-bearing premise

Low-order empirical models optimized on the current ITPA database will continue to describe confinement physics at the much higher magnetic fields, densities, and wall conditions of future reactors without new dominant loss channels or regime changes.

What would settle it

A high-performance discharge in a high-field metallic-wall tokamak whose measured triple product deviates substantially below the I_p squared trend at parameters outside the present database range would falsify the extrapolated claim.

Figures

Figures reproduced from arXiv: 2604.22728 by Egemen Kolemen, Geert Verdoolaege, Jalal Butt, Stanley M. Kaye.

**Figure 1.** Figure 1: FIG. 1: Peak plasma current achieved in major machine since the T-3 view at source ↗

**Figure 2.** Figure 2: FIG. 2: Confinement-time scalings with the highest view at source ↗

**Figure 3.** Figure 3: FIG. 3: Fusion power scaling with and without the machines currently under view at source ↗

**Figure 4.** Figure 4: FIG. 4 view at source ↗

**Figure 5.** Figure 5: FIG. 5: Machine size for safety factor view at source ↗

**Figure 7.** Figure 7: FIG. 7 view at source ↗

**Figure 6.** Figure 6: FIG. 6: Model view at source ↗

**Figure 8.** Figure 8: FIG. 8: RMSE and median absolute error of ln view at source ↗

**Figure 9.** Figure 9: FIG. 9: Schematic of the procedure used for the percentile-shell view at source ↗

**Figure 10.** Figure 10: FIG. 10: Source-machine to held-out machine models’ RMSE error and view at source ↗

**Figure 12.** Figure 12: FIG. 12 view at source ↗

**Figure 11.** Figure 11: FIG. 11: Wall-discriminating coefficient’s variation with model complexity. view at source ↗

**Figure 13.** Figure 13: FIG. 13: Single-variable power-law regressions with and view at source ↗

read the original abstract

Recent advances in high-temperature-superconductor technology have made substantially higher toroidal magnetic fields technologically accessible, reopening the design space for compact, high-field tokamak reactors. Because reactor performance projections remain anchored to empirical confinement scalings, the recent update to the ITPA global H-mode confinement database raises an important question: what does the present experimental record and its uncertainty imply for the path to reactor-grade fusion performance? In this work, we revisit confinement extrapolation from an explicitly extrapolation-oriented perspective and, to complement its implications in terms of a direct reactor performance measure, present a cross-machine empirical scaling for fusion power. We systematically search for a minimally complex confinement scaling that optimizes the tradeoff between variance capture and extrapolative robustness. We find that low-order models centered near $N=3$ to $N=4$ optimize this tradeoff, with plasma current, machine size, heating power, and elongation emerging as the dominant engineering levers, together with an empirically inferred confinement penalty associated with metallic walls. Recast in reactor-performance terms, the results indicate that both the fusion triple product and fusion power are governed primarily by plasma current: the triple product scales approximately as $I_p^2$, and the empirical fusion power scaling exhibits a similarly near-quadratic dependence over a survey of the highest performing discharges across several machines. Projecting to reactors, these results suggest that high-field devices with metal walls may require higher plasma current than standard IPB98$(y,2)$-based expectations imply, and that gigawatt-class tokamak performance likely demands operation at $I_p \gtrsim 20\mathrm{MA}$.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

The paper reanalyzes the ITPA database to get an Ip^2 scaling for triple product and fusion power, but the claim rests on how other variables are fixed during recasting.

read the letter

The main takeaway is that low-order confinement models optimized for extrapolative robustness on the updated ITPA data point to plasma current as the dominant lever, with the triple product scaling roughly as Ip squared and fusion power showing a similar near-quadratic dependence across high-performance shots. This leads to the suggestion that high-field metal-wall reactors may need Ip of 20 MA or more for strong performance, above what standard IPB98(y,2) projections give. They also flag a confinement penalty for metallic walls. The work is useful because it shifts the focus from the usual tau_E fits to direct reactor performance measures and keeps the models simple by design. The systematic search for minimal complexity that still captures the data variance is a clear step, and pulling in the wall material effect matches current device trends. The empirical fusion power scaling from multiple machines adds a practical check. The soft spots sit in the recasting to reactor terms. Deriving a clean Ip^2 dependence requires holding or substituting the other parameters at specific values, and the abstract gives no explicit list of those choices or sensitivity runs to show whether the exponent stays near 2 when the fixed values shift within plausible reactor ranges. Without those tests the Ip dominance could be tied to the particular assumptions rather than the data alone. The paper also skips cross-validation on held-out discharges and does not propagate uncertainties into the extrapolated regime or compare against physics models, so the robustness for future devices stays partly untested. The fits are internal to the database, though the robustness criterion they apply reduces the circularity somewhat. This is for reactor designers and scaling-law users who need updated empirical anchors for compact high-field concepts. A reader building performance projections or magnet requirements would find the Ip emphasis and wall penalty worth weighing. It shows straightforward engagement with the database and a focused question, so it deserves a serious referee even with the gaps in validation. I would send it for peer review.

Referee Report

2 major / 2 minor

Summary. The manuscript analyzes the updated ITPA global H-mode confinement database from an extrapolation-oriented perspective, systematically identifying low-order empirical models (N=3 to N=4) that optimize the tradeoff between variance capture and extrapolative robustness. Plasma current, machine size, heating power, and elongation are identified as dominant levers, along with an empirically inferred metallic-wall confinement penalty. Recast in reactor-performance terms, the fusion triple product is reported to scale approximately as I_p^2, with a complementary cross-machine empirical scaling for fusion power also showing near-quadratic I_p dependence. The work concludes that high-field metal-wall tokamaks may require I_p ≳ 20 MA for gigawatt-class performance, exceeding standard IPB98(y,2)-based expectations.

Significance. If the I_p^2 scaling and associated extrapolations prove robust, the results would carry substantial implications for compact high-field tokamak reactor design by indicating that higher plasma currents are needed than conventional scalings suggest, particularly under metallic-wall conditions. The paper earns credit for its explicit optimization criterion for extrapolative robustness, the cross-machine survey of highest-performing discharges for the fusion-power scaling, and the parameter-free aspects of the low-order model selection. These elements strengthen the data-driven perspective, though the empirical foundation without external validation limits the strength of the conclusions.

major comments (2)

[Abstract] Abstract: the claim that the triple product scales approximately as I_p^2 after recasting the N=3–4 model does not specify the fixed values chosen for the remaining variables (R, P, kappa) or the substitution rules for I_p-dependent operational limits. Without these details or sensitivity tests, it is unclear whether the quadratic exponent is robust or an artifact of the recasting procedure, directly affecting the load-bearing conclusion that I_p ≳ 20 MA is required.
[Results (scaling optimization and reactor projections)] The section presenting the scaling optimization and reactor projections: the I_p^2 dependence and fusion-power scaling are derived from fits to the same selected high-performance discharges used to evaluate them, with no reported cross-validation on held-out data, uncertainty propagation into the extrapolated regime, or comparison against independent physics-based models. This circularity in the fit procedure undermines the claimed extrapolative robustness of the central I_p^2 result.

minor comments (2)

[Abstract] Abstract: the number of discharges and machines included in the highest-performing survey should be stated explicitly to allow readers to gauge the statistical basis of the near-quadratic fusion-power scaling.
[Introduction or Methods] Notation: the exponent N in the low-order models should be defined on first use with a clear statement of how it relates to the standard power-law form of confinement scalings.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thoughtful and constructive report. We address the two major comments point by point below. We agree that the abstract requires explicit details on the recasting procedure and will revise to include them. We also acknowledge the need for additional validation steps in the results and will incorporate cross-validation and uncertainty propagation in the revised manuscript.

read point-by-point responses

Referee: [Abstract] Abstract: the claim that the triple product scales approximately as I_p^2 after recasting the N=3–4 model does not specify the fixed values chosen for the remaining variables (R, P, kappa) or the substitution rules for I_p-dependent operational limits. Without these details or sensitivity tests, it is unclear whether the quadratic exponent is robust or an artifact of the recasting procedure, directly affecting the load-bearing conclusion that I_p ≳ 20 MA is required.

Authors: We agree that the abstract and main text lack sufficient specification of the recasting assumptions. In the revised manuscript we will state the representative fixed values (R = 6.2 m, P = 50 MW, κ = 1.8) chosen to reflect a high-field metal-wall reactor design, and we will detail the substitution rules that incorporate I_p-dependent operational limits (Greenwald density, Troyon beta, and q95 constraints). We will also add a short sensitivity study varying these fixed parameters over plausible ranges to demonstrate that the near-quadratic I_p dependence remains stable. These clarifications will be inserted in both the abstract and the reactor-projection section. revision: yes
Referee: [Results (scaling optimization and reactor projections)] The section presenting the scaling optimization and reactor projections: the I_p^2 dependence and fusion-power scaling are derived from fits to the same selected high-performance discharges used to evaluate them, with no reported cross-validation on held-out data, uncertainty propagation into the extrapolated regime, or comparison against independent physics-based models. This circularity in the fit procedure undermines the claimed extrapolative robustness of the central I_p^2 result.

Authors: The referee correctly notes the absence of explicit validation steps. The low-order model selection (N = 3–4) was performed on the full ITPA database, while the fusion-power scaling is a separate survey of the highest-performing discharges; however, we recognize that this distinction is not sufficient without further checks. In revision we will add k-fold cross-validation on the confinement database to assess stability of the I_p exponent, Monte-Carlo propagation of coefficient uncertainties into the extrapolated triple-product and power values, and a concise comparison of our empirical exponents with those from physics-based models (e.g., the ITER physics basis and selected gyrokinetic results). These additions will be placed in a new subsection on robustness and will strengthen the extrapolative claims without altering the core empirical approach. revision: partial

Circularity Check

0 steps flagged

Empirical model selection and recasting contain no definitional or self-referential reduction

full rationale

The paper selects low-order empirical scalings by optimizing a tradeoff between variance explained and an internal robustness metric on the ITPA H-mode database, then algebraically recasts the resulting multi-variable expression into reactor-relevant quantities (triple product and fusion power) while holding auxiliary parameters at representative values. This procedure is self-contained: the I_p^2 dependence emerges from the fitted exponents and the chosen fixed-point substitution rather than from any prior assumption that the output must equal the input. No self-citation supplies a uniqueness theorem, no fitted coefficient is relabeled as an independent prediction, and the abstract's claims remain falsifiable against future data outside the fitting set. The derivation therefore does not collapse to its own inputs by construction.

Axiom & Free-Parameter Ledger

2 free parameters · 2 axioms · 0 invented entities

The central claims rest on the assumption that the ITPA database spans the relevant physics for extrapolation, that a low-order power-law form is sufficient, and that the metallic-wall effect can be captured by a single multiplicative factor. No new physical entities are postulated.

free parameters (2)

exponents in the N=3 to N=4 scaling
Fitted coefficients for Ip, size, power, elongation, and wall material that define the reported scalings.
metallic-wall penalty factor
Empirically inferred multiplicative correction applied to confinement time for metal-walled devices.

axioms (2)

domain assumption The ITPA global H-mode database is representative of the physics that will govern confinement in high-field reactors.
Invoked when using the database to optimize and validate the scaling for extrapolation.
domain assumption A low-order power-law model is adequate to capture the dominant dependencies without missing higher-order interactions or regime transitions.
Basis for searching N=3 to N=4 models as the optimal tradeoff.

pith-pipeline@v0.9.0 · 5599 in / 1667 out tokens · 57542 ms · 2026-05-08T09:06:19.326051+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

6 extracted references · 1 canonical work pages

[1]

out-of-distribution

Systematic determination ofN−parameter model noninferiority to highest complexity model on a low-τE to high-τE basis Before carrying out the simulated extrapolation error infla- tion exercise, we first split the standard set of the ITPA H- mode confinement database inτ E by distribution percentile. We built N-parameter models for incrementing subsets of t...

2044
[2]

incl" columns include all points from table I, while

Machine-organized extrapolation performance: chronology,τ E, and composite reactorwardness The preceding appendix subsection asked whether a model fit only on the lower-τ E portion of the database can extrap- olate to progressively higher-τ E data. That construction was intended to mimic a low-performance→high-performance extrapolation task, and it showed...
[3]

With the same assumptions,Q∝ H2 I2 pA2/f aux

Invoking the optimal low-order confinement scaling from this work (N=3parameters) Approximating our optimal low-order scaling for the ther- mal energy confinement time asτ E,th ∝I pRgeo/P1/2 L,th; neglect- ing radiative losses; and assuming all of the thermal loss power comes from the auxiliary heating, fusion power takes the form, Pf us ∝(nT) 2 V∝ (Pl,th...
[4]

Resulting scaling from toroidal beta This invokation finds that fusion power density scales quar- tically with the toroidal magnetic field and square of toroidal beta,β T = ⟨p⟩ B2 T /2µ0 , Pf us ∝n 2T 2 ∝p 2 ∝β 2 T B4 T While it’s true thatB T andβ T are technically independent quantities, in practice they are difficult to independently move at high perfo...
[5]

Pf us ∝n 2T 2 ∝β 2 p I4 p

Resulting scaling from poloidal beta This invokation finds that fusion power density scales quar- tically withI p and quadratically with poloidalβ,β p = ⟨p⟩ I2p/2µ0 . Pf us ∝n 2T 2 ∝β 2 p I4 p. As is the case with toroidalβ,β p’s dependence onI2 p makes Pf us’s parametric dependence difficult to interpret. Let us at- tempt to disambiguate this by substitu...

2020
[6]

arXiv:2505.03834 [physics]. 26P.N. Yushmanov, T. Takizuka, K.S. Riedel, O.J.W.F. Kardaun, J.G. Cordey, S.M. Kaye, and D.E. Post. Scalings for tokamak energy confinement.Nu- clear Fusion, 30(10):1999–2006, October 1990. 27Hartmut Zohm. On the size of tokamak fusion power plants.Philosophical Transactions of the Royal Society A: Mathematical, Physical and E...

work page arXiv 1999