Constraint 17 temporary fix

[clmould]

@timothy-nunn and I found that constraint 17 was no longer behaving as expected after #4299. In #4406 a temporary fix has been added to revert constraint 17 back to it's working state:

Reverts constraint 17 to pre-#4299.

The issue is that constraint 17 used to be formatted as

PROCESS/process/core/solver/constraints.py

Lines 594 to 605 in a68bb7f

	pradmaxpv = (
	data.current_drive.p_hcd_injected_total_mw / data.physics.vol_plasma
	+ data.physics.pden_alpha_total_mw * data.physics.f_p_alpha_plasma_deposited
	+ data.physics.pden_non_alpha_charged_mw
	+ data.physics.pden_plasma_ohmic_mw
	)
	return leq(
	data.physics.pden_plasma_rad_mw / pradmaxpv,
	data.constraints.fradpwr,
	constraint_registration,
	)

Importantly, it used pden_plasma_rad_mw. Now, the constraint uses f_p_plasma_separatrix_rad which is calculated as

PROCESS/process/models/physics/physics.py

Lines 1037 to 1042 in 024a2f2

	self.data.physics.f_p_plasma_separatrix_rad = (
	self.exhaust.calculate_radiation_fraction(
	p_plasma_rad_mw=self.data.physics.p_plasma_rad_mw,
	p_plasma_heating_mw=self.data.physics.p_plasma_heating_total_mw,
	)
	)

However this uses p_plasma_rad_mw which also includes the SOL radiation

PROCESS/process/models/stellarator/stellarator.py

Line 2170 in a68bb7f

self.data.physics.p_plasma_rad_mw += self.data.physics.psolradmw

which is never accounted for in pden_plasma_rad_mw. This changed the constraint, making the problem harder to solve, and led to constant failures when running the stellarator-helias test.

The diffs in the stellarator-helias test should revert it back to this file.

Thanks @clmould for identifying constraint 17 was the problem!

This needs to be investigated by people with stellarator knowledge - was constraint 17 wrong all along? Does something need to change in the stellarator? Or with the new format of constraint 17?

@grmtrkngtn

[timothy-nunn]

To add to this, I believe psolradmw should probably form part of the f_p_plasma_separatrix_rad calculation for stellarators and constraint 17 was correct after #4299. However, adding in this term to f_p_plasma_separatrix_rad made the constraint harder to satisfy and caused problems in the solver, with it flip-flopping between different solutions and sometimes not solving at all.

There is also the question of whether pden_plasma_rad_mw should be consistent with p_plasma_rad_mw which it currently would not be in the Stellarator.