Clarification on reference='local' in stress_field(): does index 0 represent the fiber direction?

[whhong-research]

I'm using ANBA4 via SONATA to perform cross-sectional stress analysis on an
IEA-15MW blade and have encountered an inconsistency in interpreting the
coordinate system when using reference='local'.

Issue Description

For a CarbonUD layer with fiber_orientation = 0.0 rad (fibers aligned
along the spanwise/axial direction), I applied a unit spanwise force and
compared stress components:

Component	Value (Pa)	Expected if Voigt[0] is fiber direction
σ (Voigt index 0)	~0.0207	should be ~1.0 for axial load
σ (Voigt index 1)	~0.00535	—
σ (Voigt index 2)	~0.977	← largest response
Other components	~0	—

The dominant response is at index 2 (σ_zz), not index 0 as I would expect
if index 0 represented the fiber direction.

Analysis

The ratio σ[0] / σ[2] ≈ 0.0207 / 0.977 ≈ 0.02, which is consistent with a
Poisson effect (lateral stress ≈ ν × axial stress), not a fiber direction
stress.

This suggests that:

• Voigt index 2 (σ_zz / σ_33) carries the axial load (fiber direction)
• Voigt index 0 (σ_xx / σ_11) is the transverse/chordwise stress
  (perpendicular to fiber)

My current understanding is:

 - reference='global' → stress in global beam cross-section frame (x=chordwise, y=flapwise, z=spanwise)
 - reference='local' → stress in material frame via RotatedStressElasticModulus = C_local * TM^T
 - Voigt index 2 = σ_zz (spanwise) = fiber direction stress for spanwise fibers

Questions

1. Is this interpretation correct? Does reference='local' output stress in
   a material coordinate system where index 0 is not the fiber direction,
   but rather a transverse direction?

2. If so, for extracting true fiber-direction stress from an axially-loaded
   composite, should we use index 2 (σ_33) rather than index 0 (σ_11)?

3. Is there documentation that explicitly defines what each Voigt index
   represents in the material coordinate system as output by ANBA4?

[PorterJustin]

Instead of manually indexing those arrays, I would suggest using c.stress.sigma11 etc. for global and c.stressM.sigma11 etc. for local/material coordinates. Those should have been verified through a number of models that I have run. For indexing the array you mention, I think some options change the ordering.

In material coordinates, 11 should be aligned with the fibers (may only be verified at fiber_orientation=0), 22 should be in plane wrapping around the section/perpendicular to fibers, 33 should be through thickness.

This file has a number of tests for the stress and strain values. Look at those tests to see how to get the stress/strain. I am not sure if one of the tests specifically covers checking the axial stress in local coordinates.

If this does not resolve your problem, can you please share a minimal example script that highlights the issue. Please also specific which branch you are currently using.