The best and most complete theory of transport in magnetically confined systems is the Neoclassical theory. However, it is found that transport often exceeds Neoclassical expectations by an order of magnitude or more (also see Non-diffusive transport). [1] The difference between actual transport and the Neoclassical expectation is called "anomalous" transport. It is generally assumed that the anomalous component of transport is generated by turbulence driven by micro-instabilities. [2]

How important is turbulence?

In spite of lengthy studies into the subject, it is still controversial how important turbulent transport really is. In part, this may be because turbulent transport gives a variable contribution to transport (depending on local and global parameters), whereas Neoclassical transport is always present. And in part, because no complete theory for anomalous transport is available. [3]

Arguments in favour

An important argument suggesting that anomalous transport is important to the degree that it often dominates the total transport is the scaling of transport with heating power and machine size. [4] The phenomenon of power degradation, universally observed in all devices, is an indication that standard transport theories are inadequate to explain all transport, since these would not predict power degradation. Following Freidberg, [2] the cited scaling laws can be rewritten in terms of the temperature dependence (eliminating the heating power dependence). Then, classical and neoclassical estimates would predict that the confinement increases with T (namely: τET0.5, associated with collisionality). However, the experimental scalings give a decrease with T (namely: τETα with α < -1). This unexpected behaviour is explained from increased turbulence levels (and enhanced transport) at higher values of (the gradients of) T.

Profile consistency indicates that self-organisation plays an important role in transport, and this can only be the case when instabilities or turbulence are able to regulate the profiles, i.e., when they carry an important fraction of transport.

The suppression of turbulence is possible, either actively (by imposing an external radial electric field), or spontaneously (H-modes, Internal Transport Barriers). As a consequence, transport is reduced significantly (to Neoclassical levels). This is a clear indication that turbulence is responsible for the main fraction of anomalous transport.

Arguments against

It has been argued that turbulence cannot be responsible for a significant fraction of the anomalous component of transport, since that would lead to high resistivity (due to collisions), which contradicts experimental observation. [5] However, this argument fails to note that transport events may be collective (e.g., via streamers), which do not require an enhanced collisionality.

Can anomalous transport be modelled?

There are several answers to this question. Since all equations describing the motion of charged particles in fields are known, including the effects of collisions, detailed numerical (gyrokinetic) simulations are possible. [6] However, due to the enormous disparity between the minimum and maximum scales involved (collision times vs. transport times, and the gyroradius vs. the machine size), this is a major challenge.

An alternative approach is to model the net effect of turbulence without simulating the fine detail. In doing so, it is not sufficient to introduce a simple additional "turbulent diffusivity", as this cannot possibly reproduce the observed global transport scaling behaviour. It is probably necessary to use a non-diffusive description. [7]

Can anomalous transport be controlled?

Yes. The impression is that anomalous transport is more difficult to control in tokamaks than in stellarators. However, limited control in tokamaks is possible by making use of edge transport barriers (cf. H-mode) and Internal Transport Barriers (ITBs). This reduces transport to Neoclassical levels, at least transiently.

Particularly in optimised stellarators (W7-AS), transport can be close to Neoclassical levels. [8]

References