Discrepancy between Abaqus and SimScale for simple uniaxial 3D hex-mesh bar: rigid body modes and stress distribution

Using SimScale Solid Mechanics / FEA
1

I am comparing a very simple uniaxial bar problem between Abaqus and SimScale and observing a discrepancy that I would like to understand better.

Problem setup

  • Simple rectangular 3D bar

  • Linear elastic steel

  • Pure hexahedral mesh in SimScale:

    • 160 hexahedra

    • 0 triangles

  • Uniform pressure/load applied in X direction on right face

  • Left face constrained

The mesh in SimScale reports:

  • Mesh order: 2nd

  • Number of hexahedra: 160

  • Number of triangles: 0

So this appears to be a genuine structured hex mesh.

Expectation

For ideal uniaxial loading, I expect approximately uniform axial stress:

σxx = F/A

especially away from the constrained edge.

Observation in Abaqus

Using a structured hex mesh:

  • the stress field becomes almost perfectly uniform (~25 MPa)

  • even when only UX = 0 is prescribed on the left face, Abaqus solves without singularity/convergence issues

Observation in SimScale

Even with the pure hex mesh:

  • I still observe stress disturbances/gradients near the constrained edge

  • if I constrain only UX = 0, the solver reports singularity/convergence issues

  • the simulation only runs robustly when additional constraints (UY/UZ) are added

This surprised me because the same conceptual setup runs smoothly in Abaqus.

My questions

  1. Is this primarily due to differences in:

    • rigid body mode handling,

    • solver stabilization,

    • stress recovery/averaging,

    • or element formulation?

  2. Does SimScale require stricter removal of rigid body modes for full 3D solid mechanics problems?

  3. Is Abaqus internally stabilizing or regularizing this near-singular system automatically?

  4. Are the stress gradients near the support physically expected due to Saint-Venant / Poisson constraint effects, while Abaqus visually smooths them more aggressively?

I am asking partly because I plan to use this example pedagogically to teach:

  • structured vs unstructured meshes,

  • convergence,

  • and boundary-condition effects.

Any clarification would be greatly appreciated.