Energy confinement scaling laws

Overview

Energy confinement scaling laws are mathematical expressions that relate a plasma's energy confinement time (τ_E) to various engineering and physics parameters. The energy confinement time is a critical figure of merit in fusion research, defined as the ratio of the total thermal energy stored in the plasma (W_th) to the net heating power (P_heat) required to sustain that energy: τ_E = W_th / (P_heat - dW_th/dt). A longer τ_E indicates better thermal insulation, meaning less power is required to maintain the high temperatures needed for fusion.

These scaling laws are indispensable for designing and predicting the performance of new fusion devices. Since the performance of a future machine like ITER cannot be tested directly during its design phase, physicists rely on scaling laws to extrapolate from the vast database of results from smaller, existing experiments. The laws condense complex transport physics into a manageable formula, typically a power law, that connects τ_E to global parameters such as plasma current (I_p), toroidal magnetic field (B_T), plasma density (n_e), heating power (P_heat), and machine size (major radius R, minor radius a).

The accuracy of these predictions is paramount. The predicted τ_E for ITER, for instance, directly determines its projected fusion power output and its ability to achieve a burning plasma, where alpha particle heating becomes dominant. The development and validation of robust scaling laws are therefore a central activity in the international fusion research community, underpinning the scientific basis for multi-billion-dollar investments in next-generation facilities.

Physics / Mechanism

The fundamental challenge in magnetic confinement fusion is to contain a plasma at temperatures exceeding 150 million K. While strong magnetic fields confine the charged plasma particles to helical paths, they do not provide perfect insulation. Energy is lost primarily through two mechanisms: collisional (neoclassical) transport and turbulent transport. In the hot, core region of modern tokamaks and stellarators, turbulent transport driven by microinstabilities is the dominant loss channel, often exceeding neoclassical predictions by an order of magnitude or more.

These microinstabilities, such as the Ion Temperature Gradient (ITG) mode, the Trapped Electron Mode (TEM), and the Electron Temperature Gradient (ETG) mode, create small-scale fluctuations in plasma density and temperature. These fluctuations drive a chaotic, cross-field transport of heat and particles that degrades confinement. Scaling laws are an empirical attempt to capture the net effect of this complex, multi-scale, nonlinear physics without resolving the turbulence itself.

Two main approaches are used:

Empirical Scaling: This method involves assembling a large database of experimental results from many different machines. Statistical regression techniques, typically assuming a power-law form, are used to find the best fit for τ_E as a function of global engineering variables (I_p, B_T, n_e, P_heat, R, a, etc.). The widely used IPB98(y,2) scaling is a prime example of this approach. While practical, this method risks conflating different physical effects and can be less reliable for large extrapolations if the underlying physics changes.
Dimensionless Parameter Scaling: This approach is more physically grounded. It recasts the transport problem in terms of fundamental dimensionless physics parameters that govern plasma behavior. The three most important are:
- Normalized gyroradius (ρ)**: The ratio of the ion Larmor radius to the machine size (ρ ∝ √(T)/aB). It quantifies the scale of turbulent eddies relative to the plasma size.
- Normalized collisionality (ν)**: The ratio of the particle collision frequency to the particle bounce frequency (ν ∝ nqR/T²). It measures the importance of collisions in disrupting particle orbits.
- Plasma beta (β): The ratio of plasma pressure to magnetic pressure (β ∝ nT/B²). It indicates how efficiently the magnetic field is being used to confine the plasma.

By performing dedicated experiments that vary each dimensionless parameter while holding the others constant, physicists can determine the plasma's transport properties in a way that is independent of machine size. This provides a more robust basis for extrapolation, as it assumes that if two plasmas in different-sized machines have the same dimensionless parameters, their turbulent transport will behave similarly. This approach, rooted in the Kadomtsev-Connor-Taylor theory of plasma transport, provides a crucial method for validating and improving empirical scaling laws.

Historical development

The development of confinement scaling laws has mirrored the progress of fusion research itself. Early efforts in the 1970s focused on single-machine scalings. One of the first was "Alcator scaling," derived from the Alcator A tokamak at MIT, which found that confinement time scaled linearly with plasma density (τ_E ∝ n_e a²).

As more tokamaks came online in the 1980s, it became clear that a universal scaling was needed. Researchers began compiling multi-machine databases. This led to the first generation of L-mode (low-confinement mode) scalings, such as the Goldston scaling (1984) and the Kaye-Goldston scaling (1985). These scalings revealed a troubling trend: confinement time degraded with increasing heating power (τ_E ∝ P_heat⁻⁰.⁵), implying that reaching higher temperatures would become progressively more difficult.

A major breakthrough occurred in 1982 with the discovery of the H-mode (high-confinement mode) on the ASDEX tokamak in Germany. The H-mode regime, characterized by the spontaneous formation of a transport barrier at the plasma edge, exhibited confinement times roughly double those of L-mode. This discovery revitalized prospects for achieving ignition and necessitated the development of a new set of scaling laws for this improved regime.

Throughout the 1990s, the international fusion community, coordinated through efforts like the International Tokamak Physics Activity (ITPA), refined H-mode databases and scaling laws. This work was critical for establishing the design basis for ITER. An early L-mode scaling for the project was ITER89-P. The key H-mode scaling evolved from ITER93-H to the current standard, IPB98(y,2), published in 1999. This scaling law is derived from a database of H-mode discharges from a dozen different tokamaks and remains the primary reference for predicting ITER's baseline performance. For stellarators, a parallel effort led to the International Stellarator Scaling (ISS) series, with ISS04 being a prominent example.

Current status

As of 2026, the IPB98(y,2) scaling law remains the standard for predicting thermal energy confinement in conventional ELMy H-mode tokamak plasmas. It is the reference scaling used in the ITER design basis to predict a baseline τ_E of 3.7 seconds, which is a key parameter for achieving its goal of Q=10. The formula is given by:

τ_E,IPB98(y,2) = 0.0562 * H_H * I_p⁰.⁹³ * B_T⁰.¹⁵ * P_heat⁻⁰.⁶⁹ * n_e⁰.⁴¹ * M⁰.¹⁹ * R¹.⁹⁷ * ε⁰.⁵⁸ * κ_a⁰.⁷⁸

where H_H is an enhancement factor (typically ~1.0 for standard H-mode), I_p is plasma current (MA), B_T is toroidal field (T), P_heat is heating power (MW), n_e is line-averaged density (10¹⁹ m⁻³), M is the isotopic mass, R is major radius (m), ε is the inverse aspect ratio (a/R), and κ_a is the plasma elongation.

Despite its widespread use, the uncertainty of this empirical law when extrapolating to the large, alpha-heating-dominated regime of ITER is a significant research topic. The root-mean-square error of the fit to the existing database is about 15-20%. Ongoing research focuses on validating this scaling and developing more physics-based models. Experiments on the largest current tokamaks, such as JET and JT-60SA, are particularly important for reducing the extrapolation gap to ITER. These experiments perform dedicated dimensionless parameter scans (especially in ρ*) to test the physical underpinnings of the scaling laws. Furthermore, advanced operating scenarios, such as the I-mode and Super H-mode, offer potentially better confinement than predicted by standard scaling and are being actively investigated.

Notable implementations

Energy confinement scaling is not a technology implemented by a single company but a research framework used across the entire magnetic confinement fusion community.

ITER Organization: The ITER project in France is the most significant consumer of confinement scaling predictions. Its design, operational plan, and performance targets (Q=10) are all fundamentally based on the τ_E predicted by the IPB98(y,2) scaling law. The entire project's success hinges on this extrapolation being largely correct.
ITPA Divertor and Scrape-Off-Layer Topical Group: This international working group, part of the International Tokamak Physics Activity, coordinates multi-machine experiments and database analysis. It is the primary body for maintaining and updating the confinement databases (like the ITPA H-mode database) and for vetting and proposing new or refined scaling laws.
JET (Joint European Torus): As the largest operating tokamak for many years, JET (until its shutdown in 2023) was crucial for confinement studies. Its size and capabilities allowed it to operate closer to ITER's dimensionless parameters (particularly ρ*) than any other machine, providing vital data for validating scaling laws. Its deuterium-tritium campaigns were especially important for testing isotopic effects (the 'M' term in the scaling law).
JT-60SA: This large superconducting tokamak in Japan, a joint project between Europe and Japan, is a key facility for post-JET, pre-ITER research. A major part of its research program is dedicated to confinement scaling studies to reduce the uncertainties in the extrapolation to ITER and future demonstration power plants (DEMOs).
Private Fusion Companies: Companies like /companies/commonwealth-fusion-systems and Tokamak Energy rely on established public-sector scaling laws to design their compact, high-field devices. They often use these laws as a baseline and then incorporate physics-based arguments or proprietary models to justify expected performance enhancements from their specific technological approaches (e.g., high magnetic fields).

Open challenges

Despite decades of progress, several significant challenges remain in the field of confinement scaling.

Extrapolation Uncertainty: The largest open challenge is the uncertainty inherent in extrapolating from current devices to the ITER and reactor scale. ITER's linear dimensions are roughly twice that of JET, and its magnetic field and current are also significantly larger. Small exponents in the power-law scaling can lead to large differences in the predicted τ_E. The role of alpha heating, which will be dominant in ITER but is negligible in most current devices, adds another layer of uncertainty.
Isotope Effect: The physical mechanism behind the observed favorable scaling of confinement with isotopic mass (M) is not fully understood. While experiments show that deuterium and tritium plasmas have better confinement than hydrogen plasmas, first-principles models have struggled to reproduce the effect quantitatively. This is a critical uncertainty for predicting the performance of D-T reactors.
Density Dependence: In some high-density operating regimes, confinement has been observed to saturate or "roll over," deviating from the positive n_e⁰.⁴¹ scaling in IPB98(y,2). Understanding the physics behind this density limit is crucial for optimizing reactor performance, as fusion power scales with density squared.
Integration with Pedestal and Edge Physics: The H-mode confinement is critically dependent on the formation of a transport barrier, or "pedestal," at the plasma edge. The height of this pedestal provides a boundary condition for the core plasma. Standard scaling laws treat the plasma as a single entity, but more advanced predictive models must couple core transport physics with a model for the pedestal height, which is often limited by MHD instabilities like Edge Localized Modes (ELMs).
Applicability to Advanced Scenarios: Scaling laws developed for the standard ELMy H-mode may not apply to alternative, potentially reactor-relevant scenarios like I-mode, quiescent H-mode (QH-mode), or plasmas with strong internal transport barriers (ITBs). Developing robust scaling laws for these advanced regimes is an active area of research.

Outlook

The 5-15 year trajectory for confinement scaling research is focused on reducing the uncertainties for ITER and developing a predictive framework for future power plants. In the near term (5 years), the primary focus will be on analyzing results from the new generation of large tokamaks, especially JT-60SA. These experiments will provide crucial data points at low ρ* and high β, shrinking the extrapolation gap to ITER and rigorously testing the existing scaling laws. The results will either build confidence in the IPB98(y,2) prediction or necessitate its revision.

Concurrently, the development of high-fidelity, physics-based simulation codes (such as gyrokinetic codes like GENE and CGYRO) will continue. While too computationally expensive to replace scaling laws for design studies, they serve as a tool for understanding the underlying physics. A key goal is to use these codes to validate and explain the dependencies seen in empirical scalings, particularly the isotope and power degradation effects.

Over the 10-15 year horizon, the first plasma operations at ITER will provide the ultimate test of the scaling laws. Early ITER experiments in hydrogen and helium, followed by deuterium, will offer the first direct data from a device in the reactor-scale regime. These results will be a watershed moment for the field, either confirming the half-century of research that led to its design or forcing a significant re-evaluation of plasma transport physics. The insights gained will be immediately applied to refine the designs of demonstration power plants (DEMOs), moving beyond empirical laws toward more integrated, physics-based predictive models for fusion performance.

References

ITER Physics Basis Chapter 2: Plasma confinement and transport — Nuclear Fusion (1999)
On the scaling of energy confinement in tokamaks — Nuclear Fusion (1985)
Dimensionless parameter scaling of turbulent transport in tokamaks — Physics of Plasmas (1998)
Regime of improved confinement and high beta in neutral-beam-heated divertor discharges of the ASDEX Tokamak — Physical Review Letters (1982)
Predictions of ITER performance from 0-D and 1.5-D simulations — Nuclear Fusion (2007)
International stellarator confinement scaling — Nuclear Fusion (2004)
Status of the ITPA global H-mode confinement database — Nuclear Fusion (2003)
The physics of the L-H transition — Plasma Physics and Controlled Fusion (1991)