Date: 2026-10-08 Purpose: One REBCO tape, coupled magneto-thermal and periodic, with a contact resistance between the REBCO film and the silver cap and a Jc defect that forces current across it.

What it shows

A single 4 mm tape from the tapestack3d deck, without solder or neighboring tapes, sits in a periodic air cylinder that is 10 mm long. A 150 A sigmoid drives the coupled h-phi/T problem from 77 K. The superconductor uses sp-ap.hdf5 with resistivity type : riva: the power law runs in parallel with the normal-state channel and remains finite when a defect or a temperature near T_crit drives the film normal.

The contact is the rint layer, which is 0.1 µm thick and sits between the silver cap and the REBCO film. It comes from a user material, as in tape_quench_usermat. The plugin sets its resistivity to rho = R_ct / h. For a thin layer, the resistance to current crossing it is rho * h per unit area, so rint imposes the specific contact resistance R_ct on current moving between the REBCO and the stabilizer.

In a uniform periodic tape, current never needs to cross the contact. The deck therefore adds a permanent Jc defect at mid-length. It is 2 mm long, spans the full width, and has smoothed ends and a Jc multiplier of 0.35. This represents a local loss of REBCO thickness, the defect definition of Wozniak et al. 2025 (Sec. III-E). At 77 K, sp-ap gives the sound tape roughly 280 A before the self-field reduction, and the defect section a critical current of roughly 100 A. Because the defect leaves no superconducting path around it, the current above that has to leave the REBCO through rint, run along the stabilizer past the defect, and cross the contact again on the other side.

src/matlib.cpp also carries, commented out, an alternative defect: a 0.4 mm crack that enters from one edge, the typical shape of slitting or handling damage. It needs a mesh resolution of 20 to 50 µm around the crack, which the default tape.geo does not provide.

Where the contact sits and how large it is

interface	contact layer?	why
REBCO / Ag cap	yes, rint	the interfacial resistance that controls current transfer into the stabilizer
REBCO / buffer / Hastelloy	no	the oxide buffer is already an insulator
Ag / Cu, Ag / Hastelloy	no	metallic bonds whose resistance is negligible next to the REBCO/Ag interface
tape / solder / next tape	not in this deck	applies only to stacks and coils (turn-to-turn)

Specific contact resistance of the REBCO/Ag interface, as tabulated in src/fem/maxwell/doc/contact_impedance_theory.md §10:

case	R_ct [Ω·m²]	[nΩ·cm²]	rho at h = 0.1 µm [Ω·m]
as-supplied commercial tape	2.5e-12	25	2.5e-5
typical (deck default)	1e-11	100	1e-4
degraded or aged interface	1e-10	1000	1e-3

Values above about 1e-4 Ω·m make the magnetic system much harder to condition. Turn-to-turn contacts in no-insulation coils are 1e-9 to 1e-7 Ω·m², which this layer model is not designed for.

Knobs

All three knobs live in src/matlib.cpp; rebuild the plugin after changing them.

knob	where	note
rint_Rct	top of matlib.cpp	the specific contact resistance, Ω·m²
rint_h	top of matlib.cpp	must match the rint thickness in input.conf
defect size and depth	my_defect()	the refined box in tape.geo covers z = 3 to 7 mm; widen it if the defect grows

Point ids in tape.geo double as mesh node ids, which the periodic block in input.conf relies on. Keep them gap-free when editing the geometry.

Files

file	role
tape.geo	geometry; mesh it with gmsh
src/matlib.cpp	user plugin usermat.so: the rint material and the Defect
src/user_table.hpp	two-column table reader for the plugin
data/cp_HTS.txt, data/k_HTS.txt	thermal properties of rint
input.conf	deck

Run

Build the plugin first; the deck loads src/build/usermat.so. The CMakeLists requires CMake 4.0 or newer:

cd src && mkdir -p build && cd build
cmake -DBELFEM_DIR=/path/to/belfem .. && make
cd ../..

gmsh -3 tape.geo -o tape.msh

../../build/bin/belfem # or mpirun -np 4 ../../build/bin/belfem

Shared launcher

../scripts/Allrun generates the mesh, chooses ranks and threads, and can submit to Slurm. ../scripts/Allclean removes run output. To use them from this directory, copy the two wrapper scripts Allrun and Allclean from examples/helix/ here and run ./Allrun (./Allrun --dry-run prints the plan first). If you copy this deck out of the tree, copy examples/scripts/ along with it and point the wrappers at its new location. The launcher does not build plugins; build the plugin by hand as described above. Per-deck settings go in a run.conf next to input.conf; see ../scripts/README.md.