Different rebound againt analytic-box and fixed triangle-mesh wall

[pash75-prog]

I am testing DEME mesh-wall contact with pyDEME.

I created two single-sphere cases intended to differ only in the wall representation:

1. Analytic box:
   InstructBoxDomainDimension(...) +
   InstructBoxDomainBoundingBC("all", wall_material)

2. Fixed triangle mesh box:
   AddWavefrontMeshObject(...) using a closed, manifold, inward-facing OBJ
   with the same 50 x 50 x 50 mm internal dimensions.

The mesh OBJ is closed/manifold:

• vertices: 603
• triangles: 1200
• boundary edges: 0
• nonmanifold edges: 0
• shared edges with inconsistent winding: 0
• signed volume: -125000 mm^3
  (faces point inward for a particle inside the box)

Both cases use exactly the same particle, material-pair, solver, and initial
conditions:

• one sphere, radius = 1 mm
• density = 2600 kg/m^3
• initial position = [10, 20, 10] mm
• gravity = [0, -9.81, 0] m/s^2
• particle material:
  E = 1e8 Pa, nu = 0.30, CoR = 0.80, mu = 0.20, Crr = 0
• wall material:
  E = 1e8 Pa, nu = 0.30, CoR = 0.60, mu = 0.50, Crr = 0
• particle-wall pair:
  CoR = 0.50, mu = 0.35
• initial timestep = 1e-6 s
• CD update frequency = 40
• SetExpandSafetyAdder(2.0)
• SetCDNumStepsMaxDriftMultipleOfAvg(1.2)
• SetCDNumStepsMaxDriftAheadOfAvg(6)
• SetSortContactPairs(True)
• output interval = 0.02 s

The predicted first-impact speed is about 0.610557 m/s. Before impact, both
cases match exactly:

| Time | Analytic box | Fixed mesh box |
| 0.06 s | y = 2.34405 mm, speed = 0.588672 m/s | y = 2.34405 mm, speed = 0.588672 m/s |

However, after the first floor collision they differ substantially:

| Time | Analytic box | Fixed mesh box |
| 0.08 s | y = 5.00014 mm, speed = 0.140174 m/s | y = 10.70100 mm, speed = 0.460826 m/s |

The analytic case then settles normally. The fixed triangle-mesh case has a
much stronger rebound. With a larger 5e-6 s timestep, repeated mesh impacts
eventually increase velocity until the solver aborts at the 20 m/s velocity
limit. Reducing to 1e-6 s prevents the immediate instability but the rebound
difference remains.

Questions:

01_analytic_box_1particle_1s.log
01_analytic_box_1particle_1s.py
01_box_contact_comparison_1particle.py
01_mesh_box_1particle_1s.log
01_mesh_box_1particle_1s.py

1. Is a substantially different rebound expected between
   InstructBoxDomainBoundingBC and an equivalent closed fixed triangle mesh
   when material-pair parameters are identical?
2. Does DEME accumulate multiple sphere-triangle contacts for a planar
   tessellated wall in a way that changes the effective normal impulse?
3. Is there a recommended mesh resolution, timestep, or contact-detection
   setting for a 1 mm sphere contacting a fixed closed triangle mesh?
4. Is an inward-facing closed mesh intended/supported as an equivalent
   container boundary in DEME?

Reproduction files attached:

• 01_box_contact_comparison_1particle.py
• 01_analytic_box_1particle_1s.py
• 01_mesh_box_1particle_1s.py
• BOX_BOX_reverse_winding.obj
• analytic and mesh logs

[pash75-prog]

sorry