Question About Fracture Property Scaling in Lower-Dimensional Elements

[SirGardener]

First, thank you for providing such detailed documentation and examples for mixed-dimensional modeling. While implementing both DFN and CDFM models, we noticed some interesting behavior regarding property scaling that we'd like to discuss.

We're working with the example Flow and solute transport along 2D elliptical fractures embedded in a 3D porous matrix as a reference and implementing similar models. The documentation suggests scaling both permeability and porosity with aperture:

[Materials]
  [poro1]
    porosity = 6e-4   # = a * ϕ
  []
  [permeability1]
    permeability = '1.8e-11'   # = a * k
  []

Observations
We've tested different scaling approaches and observed the following:

Configuration A (Double scaling as suggested in documentation):

k = (a²/12) * a # Permeability scaled with aperture
ϕ = 1 * a # Porosity scaled with aperture

Results:

DFN model:

• Extremely rapid pressure propagation
• Convergence issues even with very low injection rates (e.g., 2e-7 kg/s)
• Nearly instantaneous equilibrium regardless of injection rate

CDFM model:

• More stable behavior
• Different pressure propagation compared to DFN

Configuration B (Single scaling):

k = (a²/12) * a # Permeability scaled with aperture
ϕ = 1 # Unscaled porosity (MOOSE handles volume scaling)

Results:

Both DFN and CDFM models:

• Stable convergence
• Physically reasonable pressure propagation
• Good agreement between models (2-5% deviation)
• Expected sensitivity to injection rates

Questions

Given that MOOSE automatically scales the volume term (and thus porosity) with aperture for lower-dimensional elements, is the additional manual scaling of porosity necessary or potentially problematic?
The observed behavior suggests that Configuration B (scaling only permeability) produces more physically reasonable results. Is this the intended approach?
Could the documentation be clarified regarding:

• The automatic scaling performed by MOOSE
• Which properties need manual scaling and why
• The physical interpretation of single vs. double scaling

Additional Context
Our testing involved:

• Model domain: 100m × 100m × 100m
• Aperture: 6e-4 m
• Matrix permeability: 1e-22 m²
• Various injection rates from 2e-7 to 20 kg/s
• Both pure DFN and CDFM implementations
• PorousFlowSquarePulsePointSource
• We're happy to provide additional details or test cases if helpful.