You also want an ePaper? Increase the reach of your titles
YUMPU automatically turns print PDFs into web optimized ePapers that Google loves.
Plasma Phys. Control. Fusion 53 (2011) 093001<br />
Topical Review<br />
resistance (in SI units)<br />
l<br />
( µ0<br />
) 3/2 I<br />
R Lovberg =<br />
4(N i m i ) 1/2 π a . (7.2)<br />
This determined <strong>the</strong> rate <strong>of</strong> transfer into ‘turbulent kinetic energy’ <strong>of</strong> <strong>the</strong> plasma. To get<br />
agreement with HDZP-II deuterium fibre Z-<strong>pinch</strong> experiment this heating was added only to<br />
<strong>the</strong> ion energy equation, and was described as a ‘boiling’ <strong>of</strong> <strong>the</strong> <strong>pinch</strong>. The inward radial<br />
velocity <strong>of</strong> <strong>the</strong> bubbles is equated to <strong>the</strong> Alfvén speed. Lovberg states that only 10% <strong>of</strong> <strong>the</strong><br />
<strong>pinch</strong> length is affected in this way, and agreement on <strong>the</strong> low neutron yield is obtained because<br />
<strong>of</strong> <strong>the</strong> drop in density resulting from a faster expansion <strong>of</strong> <strong>the</strong> corona.<br />
In Rudakov et al [235, 331] <strong>the</strong>re is a discussion <strong>of</strong> <strong>the</strong> inward velocity <strong>of</strong> bubbles based<br />
on buoyancy leading to a velocity v bubble given by<br />
(<br />
v bubble = 2α π r )<br />
b 1/2<br />
vA , (7.3)<br />
a<br />
where r b is <strong>the</strong> radius <strong>of</strong> <strong>the</strong> bubble, v A <strong>the</strong> mean Alfvén speed and α an undetermined constant.<br />
For ‘definiteness’ Rudakov et al take r b /a = 0.2 and α = 0.3 to give <strong>the</strong> effective additional<br />
nonlinear resistance<br />
l<br />
( µ0<br />
) 3/2 I<br />
R Rudakov =<br />
2 · (N i m i ) 1/2 π a . (7.4)<br />
This mechanism is fur<strong>the</strong>r explained and employed by Velikovich [329] to compare with<br />
experiments. It is essentially identical with <strong>the</strong> earlier work by Lovberg, differing only by a<br />
factor <strong>of</strong> 2. Both models have some arbitrariness in <strong>the</strong> numerical coefficient. The effective<br />
resistance in Haines’ viscous heating model, equation (5.13) has no arbitrary constant and<br />
employs well established MHD instability growth rates, saturation levels and stress tensors.<br />
Taking <strong>the</strong> growth rates in Coppins [149] R visc becomes<br />
R visc = Ɣ(1+Ɣ/2)1/2 l<br />
( µ0<br />
) 3/2 I<br />
8(N i m i ) 1/2 π a . (7.5)<br />
For <strong>the</strong> ratio <strong>of</strong> specific heats Ɣ <strong>of</strong> 5/3 <strong>the</strong> coefficient becomes 0.282 which is just 10% larger<br />
than in R Lovberg . Considering <strong>the</strong> large physical differences in <strong>the</strong> two models this is remarkable.<br />
One reason for this coincidence is that in choosing a viscous Lunqvist number <strong>of</strong> 2 <strong>the</strong> product<br />
<strong>of</strong> wave number and viscosity is eliminated to give <strong>the</strong> left-hand side <strong>of</strong> equation (5.10) for <strong>the</strong><br />
local ion-viscous heating rate. The positive aspect is that <strong>the</strong> resistance formula enables 0D<br />
and 1D models to agree with experiments.<br />
In Haines’ model <strong>the</strong> wavelength is short and corresponds to <strong>the</strong> fastest growing mode.<br />
This saturates at a low amplitude, but <strong>the</strong> viscous heating for high k is substantial. Depending<br />
on <strong>the</strong> ratio <strong>of</strong> ion to electron viscosity, <strong>the</strong> heating will be distributed between ions and<br />
electrons. There is no need to postulate turbulence and associated cascading. If Menik<strong>of</strong>f’s<br />
<strong>the</strong>ory [500] is adapted to <strong>the</strong> MHD instability <strong>the</strong> wavelength <strong>of</strong> <strong>the</strong> fastest growing mode<br />
is increased. In contrast a ‘flux limit’ to <strong>the</strong> momentum transport caused by <strong>the</strong> mean-free<br />
path being only slightly smaller than <strong>the</strong> wavelength is equivalent to an increase in k. A<br />
full nonlinear Fokker–Planck stability analysis is needed in which electron viscosity is also<br />
included, not only in <strong>the</strong> equation <strong>of</strong> motion but also in Ohm’s law.<br />
In <strong>the</strong> Lovberg/Rudakov/Velikovich (LRV) model, <strong>the</strong> implied wavelength is long and in<br />
<strong>the</strong> linear phase will exponentiate only once in one Alfvén transit time. Thus <strong>the</strong>re is little time<br />
at stagnation for it to grow unless it is already well established as a MRT mode. There is even<br />
less chance for <strong>the</strong> cascading to shorter wavelengths implicit in establishing MHD turbulence.<br />
In contrast Haines chooses <strong>the</strong> fastest growing mode and has no need for cascading. Indeed<br />
δB/B ∼ (ka) −1 here, whilst LRV requires δB/B ∼ 1. In LRV <strong>the</strong>re is a problem with current<br />
111