ukaea/PROCESS

Optimisation with inequalities doesn't allow for user set bounds

Opened this issue · 0 comments

Some of the constraint equations we have implemented are often in practice used with f-values with upper and lower bounds set in the IN.DAT. If we optimise with inequality constraints and remove the f-values we can not implement these user selected bounds.

For example in DEMO we often select

ixc = 9 * fdene
boundu(9) = 1.2

to allow for a solution with densities above the Greenwald limit but $f_{GW} < 1.2$. The only way to bound the LH fraction ($f_{LH} = P_{sep}/P_{LH}$ ) currently is to bound the f-vlaue flhthresh. For instance

ixc = 103 * flhthresh
boundl(103) = 1.1
boundu(103) = 1.2

Again for the ne0 > neped constraint

ixc = 154 * fne0 
boundu(154) = 0.95

This functionality could be replaced by slightly modifying the constraint equations.