abaqus-evidence-simulation

name	abaqus-evidence-simulation
description	Abaqus evidence-first modeling and simulation workflow for Codex. Use when the user asks to build, run, submit, validate, postprocess, or report Abaqus/CAE or Abaqus input-deck simulations with real logs, job status, ODB evidence, screenshots, CSV exports, and AI CAE Automation Lab run records.

Abaqus Evidence Simulation

Use this skill for Abaqus work that must be executable and auditable. Prefer real Abaqus CLI or Abaqus/CAE Python execution; do not present generated scripts as solved simulations.

Required Start

1. Check environment:

ai-cae-toolbox env-check --json

2. Create a run:

ai-cae-toolbox create-run --solver abaqus --case <case-name> --objective "<objective>"

3. Write all Abaqus files inside the run folder:

• scripts/model.py for Abaqus/CAE model creation
• inputs/job.inp for input-deck workflows
• logs/ for command output, .sta, .msg, .dat
• outputs/ for .odb and other native solver outputs
• exports/ for CSV, PNG, JSON, and report artifacts

Execution Patterns

If the MCP server is available, prefer:

• abaqus_run_no_gui_script
• abaqus_submit_input_deck

Otherwise use one of these CLI patterns:

Use one of these patterns depending on the task:

abaqus cae noGUI=<run-dir>\scripts\model.py

abaqus job=<job-name> input=<run-dir>\inputs\job.inp interactive

If ABAQUS_COMMAND is set, use it explicitly:

& $env:ABAQUS_COMMAND cae noGUI=<run-dir>\scripts\model.py

Modeling Checklist

• Create geometry, material, section, assembly, step, loads, boundary conditions, mesh, job, and output requests.
• Use meaningful model, part, material, set, and surface names.
• For generated scripts, add comments only where a future engineer needs orientation.
• Save the Abaqus model or input deck before submitting the job.
• Keep private CAD files and large solver outputs out of the public repository.

Evidence Checklist

After running, collect:

• command transcript or terminal log
• .sta, .msg, .dat, and .inp
• .odb presence and size, if a real solve completed
• extracted CSV/JSON result summaries
• viewport image or plotted result image, if available
• report.md generated by the toolbox

Then run:

ai-cae-toolbox scan-evidence <run-dir>
ai-cae-toolbox generate-report <run-dir>

Failure Handling

If Abaqus is missing, license checkout fails, the job aborts, or no ODB is created, report the failure stage and exact log evidence. Do not replace failed solver evidence with screenshots or assumptions.