ACE: ACE Bulk Porosity defaults to 0.0 in J2Erosion, silently disabling all ice strength

[lxmota]

Found while reading how porosity enters J2Erosion compared to the new Permafrost cap model. No test or production deck in the tree is affected today, so this is a trap for the next deck rather than a live bug. Reporting rather than fixing, because the choice among the options below is Elyce's.

The three defaults disagree

Consumer	Line	Default
ACEThermalParameters	src/LCM/evaluators/ACEThermalParameters_Def.hpp:506,550	0.60
Permafrost	src/LCM/models/Permafrost_Def.hpp:176	0.0
J2Erosion	src/LCM/models/J2Erosion_Def.hpp:19	0.0

In Permafrost the 0.0 is a deliberate sentinel and is safe. Porosity there does exactly one thing, override the crushable volume W, and the override is guarded:

if (porosity > 0.0) P.W = porosity;   // Permafrost_Def.hpp:596

Omit the key and the Frozen Parameters / Thawed Parameters values of W simply stand. Nothing silently changes.

In J2Erosion there is no such guard. Porosity is not a parameter with a fallback, it is the second coordinate of three empirical surface fits, and 0.0 flows straight into them as a real porosity value:

std::tie(ne, ny, nk) = unit_fit(ice_saturation, porosity);   // J2Erosion_Def.hpp:571
ScalarT E = elastic_modulus_(cell, pt)   * ne;
ScalarT Y = yield_strength_(cell, pt)    * ny;
ScalarT K = hardening_modulus_(cell, pt) * nk;

What porosity 0 does to the fits

All three fits are bilinear in (ice saturation x, porosity y) and dominated by the cross term x*y. Set y = 0 and the cross term vanishes, leaving only negative contributions:

porosity	ice sat	E mult	Y mult	K mult
0.0	0.0	-0.0411	0.0000	-0.0462
0.0	0.5	-0.1806	-0.0344	-0.2277
0.0	1.0	-0.3202	-0.0688	-0.4091
0.6	1.0	0.4719	0.5725	0.4364

E_fit_max is in fact negative for every ice saturation whenever porosity is below about 0.205, and increasing the ice saturation makes it more negative, not less.

The clamps on lines 577-581 then catch it:

E = std::max(E, residual_elastic_modulus_);
Y = std::max(Y, soil_yield_strength_);

so the material does not blow up. It pins to ACE Residual Elastic Modulus and ACE Soil Yield Strength at every point, at every step, at every ice saturation. That is the fully thawed state. A deck that forgets the key gets a J2 Erosion material with no ice strength at all, with no error, no warning, and a run that completes normally and looks plausible.

The failure is quiet in the worst way: ACE_Ice_Saturation is still computed and still written to the output, so the Exodus file shows ice present and the mechanics behaving as if there were none.

Current exposure

Every deck that actually instantiates Model Name: J2 Erosion sets a porosity key, so nothing in the tree is wrong right now:

• tests/LCM/ACE/MiniErosion/materials_mechanical_denudation.yaml:34
• tests/LCM/ACE/MiniErosionClean/materials_mechanical_denudation.yaml:40
• tests/LCM/ACE/MiniErosionClean/materials_mechanical_melt.yaml:72,108
• tests/LCM/ACE/ThermoMechanicalCuboid/materials_mechanical.yaml:33

All five set ACE Bulk Porosity: 0.60, matching the thermal default. That unanimity is itself the argument that 0.60 is the intended value and 0.0 was never meant to be reachable.

Options

1. Match the thermal default. Change J2Erosion_Def.hpp:19 to default 0.60. One line, consistent with every existing deck and with ACEThermalParameters, and a deck that omits the key gets the value it would have gotten on the thermal side anyway. Leaves the two sides of a coupled run agreeing by construction.
2. Require the key. ALBANY_ASSERT that a porosity source is present when the model is J2 Erosion. Loudest and most honest, since there is no physically meaningful default porosity for arbitrary sediment, but it breaks any existing external deck that relies on the current silent behaviour.
3. Treat 0 as a sentinel, as Permafrost does. Skip the fits entirely and use the residual values directly when no porosity source is given, with a one-time warning. Preserves today's numerical behaviour exactly while making it visible.
4. Leave it and document it. Least work, and defensible if the intent was always that decks specify porosity, but the next person to hit this loses a day.

My preference is 1, with 2 as a reasonable stricter alternative if you would rather no one ever runs J2 Erosion without stating a porosity. Either way Permafrost needs no change; its 0.0 is correct as it stands.

Whichever you pick, it is a one-line change plus a rerun of the ACE suite. Note that if you pick 1 or 3 the ACE golds do not move, since every current deck sets the key explicitly.

cc @ebayats

[lxmota]

Related: #122, found in the same read. That one is about the coefficients inside Y_fit_max; this one is about the porosity value fed into all three fits. Independent fixes, but they touch the same block.