search

ukaea / PROCESS

PROCESS is a systems code at UKAEA that calculates in a self-consistent manner the parameters of a fusion power plant with a specified performance, ensuring that its operating limits are not violated, and with the option to optimise to a given function of these parameters.
https://ukaea.github.io/PROCESS/
MIT License
36 stars 11 forks source link

Better conducted power to TF coil cold mass #1656

Closed jonmaddock closed 1 year ago

jonmaddock commented 2 years ago

In GitLab by @pc6783 on May 16, 2022, 15:16

Currently, the heat conducted to the TF coils is assumed to be 2 W/m^2. This can be seen in this line in plant_power.f90:

if ( i_tf_sup == 1 ) qss = qss + 2.0D0*tfsai

where qss is the steady-state heat load to the cold mass in Watts, and tfsai is the area of the inboard TF coil legs in m^2.

This level of fidelity is a fine first approximation. The next level of fidelity could increase the accuracy by a lot. Here at PPPL our engineers commonly use a combined conduction and many-layer radiation model to estimate this power. We assume a specific ordering of materials (thermal shield, polished gold radiation gap, MLI + spacer, coil), but I'm open to input on how to provide sufficient generality.

I'd be interested to hear from @mkovari and @stuartmuldrew on how best to implement this.

Thanks

jonmaddock commented 2 years ago

In GitLab by @mkovari on May 16, 2022, 16:02

Please see issue #1483, which describes the calculation of the heat conducted through the gravity support.

The problem with estimating the radiative load using a detailed model of the sort you describe is that as far as I know the relevant data is not really known. If data collected under stringent laboratory conditions is used, the result is likely to be an underestimate, since the radiative load will be dominated by holes, areas that have the wrong emissivity, and so on.

Using data collected using a real-world cryostat might be better, but it is very difficult to separate conducted from radiated loads.

I do agree, however, that a more careful model is called for - certainly it should use the temperature of the vacuum vessel as an input. As mentioned in #1483, the existing model is wrong - it seems to take account solely of the inward-facing (plasma-facing) area of the TF coil legs, neglecting the outward facing surface.

Do you know if ITER and JT60-SA have MLI?

mkovari commented 1 year ago

I have incorporated this into Discussion #2925, and closed this issue.