Improve CS peak self-field surrogate model

[CoronelBuendia]

Description

Closes #1883

Replaces the PROCESS (original ?) fit to (Boom and Livingstone, 1962) data shown in (Wilson, 1983) with a much better fit using data from BLUEMIRA (previously verified and validated as being quasi-identical to the Boom and Livingstone graph).

Fit is courtesy of @timothy-nunn

This is the resulting fit for $k = B_{CS,peak} / B_{CS,axis}$:

And the corresponding plots for the existing $k$ fit:

The existing fit is not as catastrophic as these are made out to be, because for the most part $k$ ~ 1.0, so the peak self-field is not quite so off as one might think.

I will address a multi-turn factor in a separate issue.

Checklist

I confirm that I have completed the following checks:

• My changes follow the PROCESS style guide
• I have justified any large differences in the regression tests caused by this pull request in the comments.
• I have added new tests where appropriate for the changes I have made.
• If I have had to change any existing unit or integration tests, I have justified this change in the pull request comments.
• If I have made documentation changes, I have checked they render correctly.
• I have added documentation for my change, if appropriate.

[CoronelBuendia]

Regression test failures:

• large_tokamak_eval
• large_tokamak_nof
• low_aspect_ratio_DEMO

These are pulsed devices where the CS is at its limits. Looking at the large tokamak in evaluation mode, all the differences relate to the CS coil and its constraints (as expected):

Variable                                   Ref	       New	  % Change
ineq_con060                             0.0214      0.149     595.72
ineq_con026                             0.0546      0.117     114.33
ineq_con027                               0.13      0.184      41.13
ineq_value_con060                         1.53       1.72      12.51
temp_cs_superconductor_margin              1.53       1.72      12.51
j_cs_critical_flat_top_end             3.78e+07   4.05e+07       7.06
jcableoh_eof                            7.35e+07   7.87e+07       7.06
j_cs_conductor_critical_flat_top_end   3.50e+08   3.75e+08       7.06
ineq_value_con026                        0.567       0.53      -6.60
jcableoh_bop                            7.47e+07   7.96e+07       6.56
j_cs_conductor_critical_pulse_start     3.56e+08   3.79e+08       6.56
j_cs_critical_pulse_start               3.84e+07   4.09e+07       6.56
ineq_value_con027                        0.522       0.49      -6.15
ineq_con072                             -0.553     -0.535       3.24
b_pf_coil_peak[n_cs_pf_coils-1]             14       13.8      -1.91
b_cs_peak_flat_top_end                      14       13.8      -1.91
b_cs_peak_pulse_start                       14       13.7      -1.80
stress_hoop_cs_inner                 7.48e+08   7.35e+08      -1.80
stress_radial_cs_peak                -1.39e+07  -1.36e+07       1.80
stress_mises_cs_peak                 1.02e+09   1.01e+09      -1.24
ineq_value_con072                      1.16e+09   1.15e+09      -1.15
stress_shear_cs_peak                  1.16e+09   1.15e+09      -1.15

Which all stems essentially from the beak field being 0.2-0.3 T lower with the new calculation. For people who don't have the integers in their heads:

• icc = 26 * Central solenoid EOF current density upper limit
• icc = 27 * Central solenoid BOP current density upper limit
• icc = 60 * Central solenoid temperature margin lower limit
• icc = 72 * central solenoid shear stress limit

All fairly straightforward implications of a change to the peak field (higher current density possible, higher temperature margin, lower stress).

In optimisation mode many more outputs see significant changes, because the CS is partially driving the radial build. Now that the CS peak field is lower, the optimal solution vector changes.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 43.75000% with 9 lines in your changes missing coverage. Please review.
✅ Project coverage is 49.95%. Comparing base (a199a00) to head (aec58e5).

Files with missing lines	Patch %	Lines
process/models/pfcoil.py	43.75%	9 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #4655      +/-   ##
==========================================
- Coverage   49.98%   49.95%   -0.03%     
==========================================
  Files         151      151              
  Lines       30450    30450              
==========================================
- Hits        15220    15211       -9     
- Misses      15230    15239       +9     

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.