Finite Element Analysis (FEA) Assisted Agent

Overview

The FEA Assisted Agent automates Abaqus finite element simulations. It leverages OpenAI models through LlamaIndex and presents a Streamlit interface for executing jobs, extracting stresses and running parametric studies. Phoenix provides tracing and evaluation capabilities so each run can be analysed in detail.

---

Features

1. Abaqus Input File Generator – Creates parameterised .inp files and stores them under src/abaqus_files.
2. Abaqus Job Executor – Runs the simulation and gathers all generated output files.
3. Von‑Mises Stress Extractor – Parses the ODB file and records the peak stress value in max_vm_stress.txt.
4. Parametric Studies – Uses a ReAct-based agent to automate multiple displacement trials.
5. Real-Time Stress Evaluation – Logs whether the stress exceeds the STRESS_THRESHOLD during each step.

---

Workflow

1. User Query Input: Submit a request describing the desired analysis.
2. Task Automation: The agent selects tools and performs the simulation steps.
3. Real-time Updates: Intermediate reasoning and outputs are shown while the job runs.
4. Final Results: The agent summarises the outcome and recorded evaluations.

---

Technologies Used

• Streamlit for the user interface.
• LlamaIndex for agent and tool integration.
• Phoenix for tracing and grading tool calls and answers.
• OpenAI models for both the agent and evaluation judges.
• Custom Modules: tools, prompt_temp, and eval_utils.

Evaluation Features

• Tool-Calling and Unit Evaluation – Grades whether the chosen tool and units are correct.
• Final Result Evaluation – Detects hallucination and checks if stress exceeds the configured threshold.
• Stress Threshold Logging – Adds a metric to the latest agent step whenever a new stress value is produced.

---

Running the Application

1. Install the requirements:

   pip install -r requirements.txt

2. Start the Streamlit app:

   streamlit run app.py

   Set the STRESS_THRESHOLD environment variable if a different limit is desired.