Integrated Pedestal-Stability Workflow

• Integrated pedestal–stability workflow is a coupled system that computes pedestal height, width, and core-edge parameters self-consistently to maintain reactor performance.
• It links models like ASTRA, TGLF, and EPED-NN to integrate core transport, equilibrium, impurity radiation, and divertor detachment in a closed iterative loop.
• Systematic sensitivity scans reveal separatrix density as a key lever, with Ar seeding yielding robust H-mode access while impurity effects primarily influence transport and power balance.

Searching arXiv for the cited workflow papers and closely related pedestal-stability works to ground the article in current literature. Integrated pedestal–stability workflow denotes a coupled modeling procedure in which pedestal structure, core transport, impurity behavior, equilibrium, and edge or divertor constraints are solved in a closed loop rather than as isolated subproblems. In reactor-relevant studies, the workflow typically links a pedestal stability model to transport solvers, equilibrium reconstruction, impurity and radiation calculations, and reduced or full divertor constraints, so that pedestal pressure, pedestal width, separatrix conditions, and global performance remain mutually consistent. A representative implementation is the ARC core-edge study of impurity-seeded H-modes, which couples ASTRA, SPIDER, FreeGS, TGLF, FACIT, STRAHL, an EPED-based neural-network pedestal model, and an X-Lengyel divertor model into an iterative stationary workflow (Muraca et al., 8 Jun 2026). Closely related workflows couple ELITE or GATO ideal-MHD boundaries with gyrokinetic flux calculations (McClenaghan et al., 24 Jun 2026), or embed EPED and transport solvers within integrated frameworks such as STEP and MAESTRO (Lyons et al., 2023, Rodriguez-Fernandez et al., 7 May 2026).

1. Definition and scope

An integrated pedestal–stability workflow is a modeling chain in which pedestal height and width are not treated as fixed external inputs, but are computed in concert with core profiles, equilibrium, transport, and edge constraints. In the ARC study, the workflow is explicitly described as coupling pedestal physics, core transport, impurity radiation, and a reduced divertor or neutral model in a closed loop, followed by systematic sensitivity scans (Muraca et al., 8 Jun 2026). This places pedestal stability within a broader core–edge systems context rather than within a standalone peeling–ballooning calculation.

The workflow is “integrated” in two senses. First, it couples distinct physical domains: core turbulence, neoclassical transport, pedestal structure, impurity charge-state evolution, radiation, and detachment. Second, it uses feedback between these domains. In ARC, ASTRA and TGLF determine core profiles and losses; X-Lengyel uses those losses and the assumed separatrix density to determine the impurity seeding needed for detachment; enrichment and edge transport then set pedestal impurity content and $Z_{eff,ped}$; EPED-NN returns updated pedestal top pressure $p_{top}$ and pedestal width; these modified pedestal conditions alter core gradients and therefore fusion power and power to the scrape-off layer (Muraca et al., 8 Jun 2026).

This integrated interpretation is consistent with broader recent work. One workflow maps pedestal operating space by combining equilibria, ELITE and GATO stability boundaries, and CGYRO or QLGYRO flux predictions over pedestal height and width scans (McClenaghan et al., 24 Jun 2026). Another class of frameworks, including STEP and MAESTRO, organizes equilibrium, transport, pedestal, and auxiliary-source modules around centralized data exchange or surrogate-based optimization, permitting feedback and optimization loops instead of a single forward pass (Lyons et al., 2023, Rodriguez-Fernandez et al., 7 May 2026).

2. Core architecture and code coupling

The ARC implementation centers on ASTRA as the 1D transport solver, evolved to stationary flat-top solutions rather than full time-dependent discharges (Muraca et al., 8 Jun 2026). Magnetic equilibrium is handled through SPIDER for fixed-boundary equilibria and FreeGS for free-boundary equilibrium and LCFS geometry, both solving the Grad–Shafranov equation

$\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$

These equilibria provide the geometry used by TGLF and EPED-NN (Muraca et al., 8 Jun 2026).

Core turbulence is modeled with TGLF using the SAT2 rule, electromagnetic physics, Miller geometry, and 5 species. TGLF is called twice per time step to separate impurity diffusion and convection. ASTRA advances density and energy through the standard transport equations

$$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$

$$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$

with $q_s$ supplied by TGLF and source terms such as collisional power exchange, ICRH, and fusion handled in ASTRA (Muraca et al., 8 Jun 2026).

Impurity transport is split into turbulent and neoclassical parts. FACIT supplies fast analytical neoclassical impurity coefficients, including rotation and poloidal asymmetries, through

$$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$

which are combined with TGLF impurity transport coefficients (Muraca et al., 8 Jun 2026). STRAHL is then called at each ASTRA step to evolve charge-state-resolved impurity profiles and line radiation. For each charge state $Z_i$,

$$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$

and the radiation power density is

$$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$

This radiation enters ASTRA power balance and the definition of $p_{top}$0 (Muraca et al., 8 Jun 2026).

A further global stability constraint is imposed through TRANSP with the Porcelli model, including fast-ion stabilization, to predict the pre-crash $p_{top}$1-profile. The resulting profile with $p_{top}$2 is imposed in ASTRA, fixing one important core stability condition associated with sawteeth (Muraca et al., 8 Jun 2026).

This modular composition resembles other integrated frameworks, but with different emphases. STEP organizes EFIT or CHEASE, TGYRO, EPED or EPED-NN, DCON or GATO, and source models through a centralized ITER-IMAS data structure, enabling open-loop and feedback workflows (Lyons et al., 2023). MAESTRO similarly couples PORTALS to external equilibrium, pedestal, divertor, and heating modules, using surrogate-based optimization to reduce the number of expensive transport evaluations in stiff regimes (Rodriguez-Fernandez et al., 7 May 2026).

3. Pedestal model and stability constraints

In the ARC workflow, pedestal height and width are predicted by an EPED-based neural network trained on EPED scans for ARC, referred to as “EPED-NN” (Muraca et al., 8 Jun 2026). Inputs include prescribed pedestal density $p_{top}$3, global stored energy or pressure, geometry, and $p_{top}$4, while outputs include pedestal top pressure $p_{top}$5 and pedestal width $p_{top}$6. EPED-NN implicitly encodes peeling–ballooning and KBM constraints through predicted $p_{top}$7 as a function of $p_{top}$8, $p_{top}$9, and $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$0 (Muraca et al., 8 Jun 2026).

The underlying EPED structure is summarized through a peeling–ballooning limit

$\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$1

and a KBM-related pressure-gradient limit often expressed through a pedestal-width scaling like $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$2 (Muraca et al., 8 Jun 2026). In the ARC scans, these constraints manifest in the transition between peeling-limited and ballooning-limited pedestals.

The key control parameters are pedestal density $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$3 and separatrix density ratio $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$4, both treated as scan variables rather than predicted quantities. At fixed $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$5, the study scans $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$6, roughly $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$7 (Muraca et al., 8 Jun 2026). Fusion power rises from about 820 MW to a maximum near 900 MW around $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$8, then plateaus as the pedestal transitions to ballooning-limited. By contrast, varying $\Delta^* \psi = -\mu_0 R j_\phi(\psi, R).$9 from 0.3 to 0.5 at fixed $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$0 leaves the pedestal peeling-limited and $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$1 for 0.3–0.4, but at 0.5 produces a ballooning-limited state with $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$2 dropping from about 370 kPa to about 330 kPa and fusion power dropping to about 800 MW (Muraca et al., 8 Jun 2026).

This behavior supports a specific interpretation of pedestal–stability integration: separatrix density is not only an exhaust parameter, but a pedestal-stability control parameter through its effect on peeling–ballooning stability (Muraca et al., 8 Jun 2026). A closely related recent study reaches a similar structural conclusion from a different direction, mapping pedestal operating space with ELITE, GATO, and gyrokinetic flux predictions and identifying an intermediate regime bounded by local KBM second stability and global finite-$$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$3 ballooning (McClenaghan et al., 24 Jun 2026). That result suggests the pedestal operating window may be narrower and more structured than the simplest EPED picture when global effects are retained.

The data also indicate that moderate changes in $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$4 have weaker effect on $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$5 than changes in $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$6 within the ARC parameter range (Muraca et al., 8 Jun 2026). This does not imply that impurities are unimportant; rather, it locates their strongest integrated role primarily in transport, dilution, and power balance, with only modest direct impact on pedestal pressure in the scanned $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$7 range.

4. Edge, divertor, and impurity feedback loops

The edge and divertor part of the ARC workflow is handled by the extended Lengyel 0D divertor model, X-Lengyel (Muraca et al., 8 Jun 2026). Given separatrix density $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$8, power to the SOL $$\frac{\partial n_s}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V'\left( -D_s \frac{\partial n_s}{\partial \rho} + v_s n_s \right)\right] + S_s,$$9, and geometry inputs consistent with the ARC divertor, X-Lengyel returns the required SOL impurity fraction $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$0, separatrix temperature $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$1, and divertor neutral pressure $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$2, while enforcing a divertor target temperature $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$3 as the detachment criterion (Muraca et al., 8 Jun 2026).

The impurity coupling is organized through the enrichment factor

$$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$4

with $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$5 meaning the core concentration is smaller than the SOL concentration (Muraca et al., 8 Jun 2026). A nominal scaling from Kallenbach’s regression is used:

$$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$6

This is used to estimate edge enrichment and thus $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$7 at the pedestal top, with $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$8 supplied by X-Lengyel (Muraca et al., 8 Jun 2026).

Edge impurity transport coefficients are then tuned so that $$\frac{3}{2}\frac{\partial (n_s T_s)}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' q_s \right] + P_s,$$9 is consistent with the enrichment, $q_s$0, and $q_s$1. The resulting $q_s$2 is passed back into EPED-NN (Muraca et al., 8 Jun 2026). This produces a closed loop in which detachment requirements determine edge seeding, edge seeding determines pedestal impurity content, pedestal impurity content modifies $q_s$3, and $q_s$4 modifies pedestal predictions.

The principal feedback loops can be summarized as follows.

Loop	Direction	Returned quantity
Core to SOL	ASTRA + TGLF + STRAHL to X-Lengyel	$q_s$5
SOL to seeding	X-Lengyel to enrichment model	$q_s$6
Impurity to pedestal	STRAHL + ASTRA to EPED-NN	$q_s$7
Pedestal to core	EPED-NN to ASTRA	$q_s$8

This looped structure is the defining feature of the workflow. The paper does not spell out a numerical convergence algorithm, but the description implies iteration between ASTRA, EPED-NN, and X-Lengyel until a consistent stationary solution satisfies core power and particle balance, EPED-NN pedestal pressure, divertor detachment with $q_s$9, and consistency of $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$0, $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$1, and the pedestal state (Muraca et al., 8 Jun 2026).

A plausible implication is that such workflows are especially sensitive to boundary-condition closures: $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$2 and $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$3 remain imposed scan variables in ARC rather than fully predicted quantities. The paper explicitly identifies this as a limitation typical of present-generation integrated modeling (Muraca et al., 8 Jun 2026).

5. Sensitivity studies and integrated behavior

The ARC study uses systematic scans to stress-test the integrated workflow (Muraca et al., 8 Jun 2026). Three are central: pedestal density, separatrix density ratio, and enrichment factor.

For pedestal density, at fixed $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$4 and Ar seeding, $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$5 spans about 820–900 MW across $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$6, peaking near $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$7. Across this scan, SOL parameters such as $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$8, $$\Gamma_z^{NC} = -D_z^{NC}\,\nabla n_z + n_z v_z^{NC},$$9, and $Z_i$0 vary little, while impurity profiles remain moderately peaked but near-flat enough radially to justify simplified constant-concentration assumptions used in earlier ARC modeling (Muraca et al., 8 Jun 2026).

For separatrix density, with $Z_i$1, varying $Z_i$2 leaves $Z_i$3 near 900 MW for $Z_i$4, but lowers it to about 800 MW at 0.5. The key point is that performance degradation is dominated by the pedestal stability curve changing with $Z_i$5, not by impurity effects, since changes in $Z_i$6 are modest (Muraca et al., 8 Jun 2026).

For enrichment, Ar-seeded cases with $Z_i$7 ranging from 50–200% of the Kallenbach scaling plus a case with $Z_i$8 show $Z_i$9 varying only modestly, about 820–920 MW. Higher $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$0 reduces core impurity concentration and core radiation, raising $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$1 and $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$2, but pedestal pressure changes only mildly through $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$3 (Muraca et al., 8 Jun 2026). This is one basis for the conclusion that Ar performance is relatively robust to enrichment uncertainty provided detachment is maintained.

The most extensive sensitivity analysis compares Ar and Ne seeding across roughly 30 cases per species (Muraca et al., 8 Jun 2026). For Ar, fusion power spans about 750–1000 MW, often near 900–950 MW. For Ne, it spans about 600–850 MW. The lower Ne performance is attributed to stronger DT dilution and higher core radiation because lower enrichment allows more core accumulation. Averaged values from table 3 are $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$4 and $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$5 for Ar, versus $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$6 and $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$7 for Ne (Muraca et al., 8 Jun 2026).

H-mode access is assessed using the Schmidtmayr L–H threshold scaling through

$$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$8

Ar cases generally satisfy $$\frac{\partial n_{z,i}}{\partial t} = -\frac{1}{V'}\frac{\partial}{\partial \rho}\left[ V' \Gamma_{z,i} \right] + S^{ion}_{z,i-1} - S^{ion}_{z,i} + S^{rec}_{z,i+1} - S^{rec}_{z,i},$$9, indicating robust H-mode access, whereas Ne cases are often only marginal, with $$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$0. The same qualitative outcome holds with Delabie and Martin threshold scalings (Muraca et al., 8 Jun 2026).

These scans collectively establish the integrated response of the workflow rather than only the response of a single submodel. A plausible implication is that systematic scans are not merely uncertainty analysis but an essential diagnostic of workflow closure: they reveal which couplings dominate reactor performance. In ARC, separatrix density is the dominant pedestal-stability lever, while enrichment uncertainty is secondary for Ar-seeded operation (Muraca et al., 8 Jun 2026).

6. Comparative frameworks, limitations, and broader significance

The ARC workflow sits within a broader family of integrated pedestal–stability approaches. One direct comparison is to classic core-plus-EPED-plus-SOLPS workflows such as JINTRAC, CORSICA, or TGYRO–TGLF–EPED chains. The ARC study differs by replacing SOLPS with the reduced X-Lengyel model and direct EPED evaluations with EPED-NN, while retaining self-consistent charge-state-resolved impurity transport through STRAHL, TGLF, and FACIT (Muraca et al., 8 Jun 2026). This combination makes large scans feasible while keeping impurity physics substantially more explicit than in many reduced integrated studies.

Another comparison is to the equilibrium-scan workflow combining ELITE, GATO, and CGYRO or QLGYRO. That approach explicitly maps regions of KBM first stability, KBM instability, KBM second stability, MTM-dominated transport, and global finite-$$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$1 ideal-MHD boundaries in the $$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$2 plane (McClenaghan et al., 24 Jun 2026). Its main contribution is to show that accessible pedestal operation may lie in a narrow region bounded by local KBM second stability on one side and global finite-$$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$3 ballooning on the other, with low-$$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$4 peeling especially important in spherical tokamaks (McClenaghan et al., 24 Jun 2026). This suggests that integrated workflows can be organized either as a stationary coupled loop, as in ARC, or as a precomputed operating-space map, with both strategies aiming at the same underlying closure problem.

STEP and STEP-0D illustrate a more general software architecture in which pedestal physics, equilibrium, transport, and stability are modularized through a central data structure. STEP couples EFIT or CHEASE, TGYRO, EPED or EPED-NN, and ideal-MHD tools such as DCON or GATO; pedestal calculations have been validated across more than 500 discharges of the $$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$5 database, with a mean error in confinement time from experiment less than 19% (Lyons et al., 2023). STEP-0D extends this concept by generating initial profiles from 0D reactor parameters and iterating equilibrium, transport, and pedestal models to a self-consistent stationary solution, again with a mean relative error in confinement time of less than 19% on a large H-mode subset (Slendebroek et al., 2023). MAESTRO adds a different numerical philosophy, using PORTALS and Gaussian-process surrogates to handle stiff and even discontinuous quasilinear flux responses during steady-state optimization (Rodriguez-Fernandez et al., 7 May 2026).

The main limitations emphasized in the ARC study are that $$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$6 and $$P_{\rm rad}(\rho) = \sum_{z,i} n_e n_{z,i} L_{z,i}(T_e).$$7 are not predicted self-consistently, ELMs and their pedestal effects are neglected, no explicit 2D SOLPS neutrals or kinetic neutrals are included, and H-mode access is assessed by static threshold criteria rather than dynamic L–H transition simulations (Muraca et al., 8 Jun 2026). These limitations recur in other workflows with different emphasis. The ELITE–CGYRO workflow is static and does not model ELM dynamics (McClenaghan et al., 24 Jun 2026). STEP presently relies on ideal-MHD and EPED assumptions that do not cover type-III ELMs or fully ELM-free regimes (Lyons et al., 2023). MAESTRO is steady-state only and does not resolve fine pedestal structure or transient stability events (Rodriguez-Fernandez et al., 7 May 2026).

An important recent development is that impurity-driven turbulence has been shown experimentally to alter pedestal transport and peeling–ballooning stability itself, producing a decoupling of peeling and ballooning boundaries and opening a stability channel toward long ELM-free phases (Banerjee et al., 1 Jun 2026). This suggests that future integrated pedestal–stability workflows may need to elevate impurity-driven turbulence from a perturbation to an active control variable.

Taken together, these studies imply that “integrated pedestal–stability workflow” is best understood not as a single code stack but as a methodological class: a closed, iterative, or mapped system in which pedestal limits, transport, equilibrium, impurity physics, and edge constraints are jointly solved. In the ARC case, that joint solution supports the viability of high-performing H-mode operation with detached divertors and fusion power approaching a gigawatt, particularly with Ar seeding (Muraca et al., 8 Jun 2026).