Instabilities and turbulence in stellarators from the perspective of global codes 2
problem with no possible fully analytic treatment, except in simplified or particular
cases. The physical problem can be treated numerically by using different numerical
implementations, each of them having their own weakness and strengths. Besides, these
kinds of numerical models imply some approximations and simplifications, which makes
it very important to verify the numerical codes by means of comparisons to analytical
models when possible, or against other codes with different numerical implementations.
The axial symmetry in tokamaks makes it possible to use of the local, so-called flux
tube approximation [6], consisting of the simulation of a physical domain surrounding
a magnetic field line and following the line one poloidal turn, which allows reducing
significantly the computational resources required for turbulence simulations with these
codes with respect to the simulation of either the full domain or the full flux surface.
In stellarators, the situation is quite different and different flux tubes lying over the
same flux surface cannot be considered as equivalent. Although first applications of
gyrokinetic codes to stellarators were based on the direct adaptation of the flux tube
paradigm for stellarators [7, 8, 9] and the use of flux tubes with just one poloidal turn
in length, this is not satisfactory in stellarators. The minimum computational domain
required for stellarators was addressed in recent works studying two linear problems
[10, 11] and it was demonstrated that, at fixed radial wavenumber, short flux tubes on the
same flux surface provide different results, in general, which only converge to each other
when the flux tube length is sufficiently increased. In [12] it was shown that the heat flux
computed in full-surface simulations can be significantly larger than that obtained in
full global simulations, suggesting that global simulations are required in stellarators, in
general. Furthermore, global codes are, in principle, required in stellarator turbulence
simulations to properly account for the influence of the long-wavelength electric field
and the global density and temperature profiles.
Several global gyrokinetic codes have been specifically designed for stellarators or
adapted from tokamak codes for thee dimensional geometries, and there is presently a
number of them available for stellarators: XGC-S [13], GTC [14], GT5D [15], EUTERPE
[16, 17], GENE-3D [18] and GKNET [19]. While in tokamaks gyrokinetic simulations
have a reasonable degree of maturity and there is a set of codes cross-benchmarked
and validated, the number of available codes able to target the stellarator geometry
has been limited until recently, and they still lack of verification and validation, in
general. In this work, we present the results of a effort carried out during the last
months for the cross-verification of the codes EUTERPE [16, 17] and GENE-3D [18].
Both are global codes designed specifically for stellarator geometry. They are based on
different numerical models but they also share common features that permit an in-detail
comparison from which both codes can benefit. In addition to the cross verification
of these codes, we address two problems that are considered relevant for turbulent
transport in stellarators and for which the global codes are particularly suited, namely
the localization of instabilities and turbulence over the flux surface and the influence of
the electric field on the instabilities and turbulence.
The structure of the work is as follows. In Section 2, we briefly describe the