Dependence of energy confinement on edge density

[mkovari]

The updated ITPA global H-mode confinement database: description and analysis
G Verdoolaege et al

The new energy confinement scaling described in this paper has already been implemented (#3164). However, the paper also has a plot of the effect of separatrix density ne,sep (normalised by Greenwald density) on the confinement factor H20, for those entries in the data set where ne,sep is available (only for some points from AUG, AUG-W and JET-ILW):

There is a clear decreasing trend with increasing separatrix density, indicating a residual role of ne,sep that is not reflected by the present scaling.

Note that the volume-average density is already included in the scaling, but this effect is independent.

Verdoolaege et al have not proposed including ne,sep in the scaling, probably because they don't have enough data points. However, we could fit a trendline to this plot and include it as an additional factor in the calculation of the energy confinement time.

(Note that there is also an issue on Correlation of the tokamak H-mode density limit with ballooning stability at the separatrix, #559, but this has not been implemented. I am not sure if it would be sensible to use both of these in the same run.)

The slope of the trend line is modest but not negligible: increasing ne,sep/nGW from 0.05 to 0.2 decreases H20 from about 1.2 to 0.85.

Consequences

There would be two ways to run PROCESS using this new scaling.

1. Set ne,sep or ne,sep/nGW using guidance from SOL experts. This would then increase or decrease the value of H20 depending of the value of ne,sep chosen.
2. Allow ne,sep to vary. The optimiser would tend to push it down towards the lower bound, since ne,sep doesn't have much effect on the other models. It has a small effect on the volume-averaged density, and an effect on radiation loss, probably.

Either way the scaling would improve the realism of the results.

Proposed solution

Fit a linear trendline to this plot and include it as an additional factor in the energy confinement time scaling when ITPA20 is used.

Comments @jmorris-uk @ajpearcey@ @mcoleman @je-cook ?