Skip to content

Revalidate EOS-CG phase assertion in CO2 low-temperature notebook #151

Closed

Description

@EvenSol

Context

A documentation-only catalog update in draft PR #150 triggered the existing CO2 low-temperature notebook validation workflow. The workflow is master-dependent and failed while executing:

notebooks/thermodynamics/CO2_low_temperature_compression_relief_and_depressurization.ipynb

Failed run: https://github.com/EvenSol/NeqSim-Colab/actions/runs/33525804944

This is not caused by the files in PR #150; the notebook and its workflow are unchanged there.

Reproducible environment

  • NeqSim Java source: current equinor/neqsim master at fdf6b227b4240589ebbfb900527b37e667f3efc8
  • NeqSim Python bridge: 3.17.0
  • Python: 3.12
  • Java: 17
  • Workflow: .github/workflows/co2-low-temperature-notebook-validation.yml

The notebook was last recorded as passing against NeqSim commit ef31a3ba78a325b5784a33486bb047a264425412.

Failure

All checks passed except:

EOS-CG detects nominal two-phase state at 25 C: False

The assertion expects two phases for the notebook's nominal state at 70 bara and 25 °C with 1 mol% total impurity split as 50% nitrogen, 30% oxygen, and 20% argon. The retained validated notebook reports two phases and gas beta 0.03087 for that state; the current-master execution no longer satisfies getNumberOfPhases() == 2.

Requested investigation

  1. Reproduce the state against commits ef31a3b and fdf6b22.
  2. Identify whether the phase-count change is an intended EOS-CG correction, a flash/model regression, or an overly brittle notebook acceptance point.
  3. If intended, update the notebook's physical interpretation, validation condition, retained outputs, and ledger with an independently justified phase-boundary comparison.
  4. If unintended, open/link an actionable equinor/neqsim defect and pin the notebook workflow only as a temporary measure.
  5. Restore the existing workflow to green against its declared source policy.

Acceptance criteria

  • The expected phase behavior at the exact nominal state is documented with commit provenance.
  • The check tests a physically justified property rather than simply weakening the assertion.
  • The notebook is cleanly re-executed, rendered, and visually reviewed.
  • The CO2 low-temperature workflow passes.

Activity

  1. EvenSol commented on Sep 9, 2026

    @EvenSol
    OwnerAuthor

    Reproduction update — 2026-09-09

    Draft #156 reproduced this unchanged-notebook failure against a newer NeqSim source state:

    • NeqSim-Colab exact PR head: 308fd1e690eac4d188275c2e862b9d3d6ee6f1fd
    • GitHub merge-ref tested: ad9b725
    • NeqSim Java source built by the workflow: 1d4fcc111b7e80dee49cd0941d8317029b943296
    • Workflow run: 34404033648
    • Attempts: 2; both failed at notebook cell 46
    • Reproduced assertion: EOS-CG detects nominal two-phase state at 25 C: False
    • The other eleven scientific assertions displayed by the cell passed on both attempts.

    The draft changes only comparesimulations2.ipynb and its two maintenance-ledger files; it does not modify the CO₂ notebook, its workflow, or NeqSim thermodynamic sources. This confirms the failure remains an independent source-policy/phase-behavior dependency rather than a branch-attributable defect in #156.

    The acceptance criteria and requested physical investigation in this issue remain appropriate. No assertion was weakened and no thermodynamics change was folded into the engineering-diagram draft.

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Projects

    No projects

      Milestone

      No milestone

      Relationships

      None yet

      Development

      No branches or pull requests

      Issue actions