Flux-Tube versus Global Gyrokinetics
Gyrokinetic simulations trade domain size for cost: flux-tube runs are local and cheap, global runs capture large-scale structure at far higher expense.
The flux-tube approximation
A flux-tube simulation follows a thin domain elongated along a single field line, assuming turbulence is statistically uniform across the small perpendicular extent and that gradients are locally constant. This local approximation is inexpensive and excellent for computing a transport coefficient at one radius, which is why it is the workhorse for building transport models.
When local breaks down
The flux-tube picture fails when the turbulence has structures comparable to the machine size, when profile gradients vary sharply, or in compact devices where the ratio of turbulence scale to minor radius is not small. Then a global simulation, covering a full radial extent or the whole poloidal cross-section, is required.
Global effects
- Turbulence spreading from unstable into stable regions
- Zonal flows and avalanche-like transport events spanning many correlation lengths
- Profile shearing that regulates the turbulence self-consistently
- Finite machine-size corrections to transport scaling
Choosing the right tool
Analysts start local for speed and move global only when the physics demands it. A common practice is to validate a flux-tube result against a global run at one point, then use the cheaper local model for scans. The choice is documented so that the validity range of the result is clear.
Relevance to compact devices
Spherical tokamaks such as the Hyperion breeder have relatively large ratios of gyroradius to minor radius and strong flow shear, conditions where global and electromagnetic effects can matter. Turbulence estimates for such devices are treated with appropriate caution and cross-checked across configurations.
The general lesson holds across codes: match domain and physics to the question, and never extrapolate a local result into a regime where non-local physics governs.