doc: Add Ansys Fluent QoI-based workflow example (#399)

<!--
Please use a PR title that conforms to *conventional commits*:
"<commit_type>: Describe your change"; for example: "fix: prevent race
condition". Some other commit types are: fix, feat, ci, doc, refactor...
For a full list of commit types visit
https://www.conventionalcommits.org/en/v1.0.0/
-->

#### Relevant issue or PR
<!-- If the changes resolve an issue or follow some other PR, link to
them here. Only link something if it is directly relevant. -->

#### Description of changes
<!-- Add a high-level description of changes, focusing on the *what* and
*why*. -->

#### Testing done
<!-- Describe how the changes were tested; e.g., "CI passes", "Tested
manually in stagingrepo#123", screenshots of a terminal session that
verify the changes, or any other evidence of testing the changes. -->

---------

Co-authored-by: andrinr <[email protected]>
Co-authored-by: Matteo Salvador <[email protected]>
Co-authored-by: Dion Häfner <[email protected]>
Co-authored-by: Alessandro Angioi <[email protected]>

Author: 5 people

demo/_showcase/ansys-qoi/README.md

@@ -0,0 +1,152 @@
+# QoI-Based Surrogate Modeling Workflow
+
+This showcase demonstrates an end-to-end workflow for building QoI-based surrogate models from Ansys simulation data using Tesseract.
+
+## Overview
+
+This workflow implements a **QoI-based surrogate modeling pipeline** that:
+1. Processes Ansys simulation runs (CAD files + simulation results)
+2. Extracts point clouds, CAD parameters, boundary conditions, and Quantities of Interest (QoI)
+3. Trains hybrid ML models (PointNet + Random Forest) to predict QoI
+4. Performs inference on new CAD designs
+
+### Workflow Components
+
+The workflow consists of three Tesseract components:
+
+```
++------------------+
+|   qoi_dataset    |  <- Processes Ansys simulation runs
+|   (Data Gen)     |  -> Generates NPZ datasets
++--------+---------+
+         |
+         v
++------------------+
+|    qoi_train     |  <- Trains ML models on processed data
+|   (Training)     |  -> Outputs trained models + scalers
++--------+---------+
+         |
+         v
++------------------+
+|  qoi_inference   |  <- Import new geometry design
+|   (Inference)    |  -> Generates QoI predictions
++------------------+
+```
+## Input Data
+
+The workflow expects **Ansys simulation runs** in the following structure:
+
+```
+inputs/Ansys_Runs/
+├── Run_0001/
+│   ├── geometry.stl              # CAD geometry file
+│   ├── metadata.json.series      # BC and CAD parameteres
+│   ├── all_pressure.txt          # QoI outputs
+│   └── ...
+├── Run_0002/
+│   └── ...
+└── ...
+```
+
+Each simulation run directory should contain:
+- **CAD file** (`.stl` format): 3D geometry for point cloud sampling
+- **Boundary condition data**: Parameters varied across simulations
+- **CAD parameters**: Parameters used during the CAD design process
+- **Simulation results**: Target QoI values to predict
+
+## Configuration
+
+Each component uses a `config.yaml` file in `inputs/config.yaml` that specifies:
+- Point cloud sampling strategy
+- Parameter extraction rules
+- QoI definitions
+- Model architecture and training hyperparameters
+
+## Running the Workflow
+
+### Option 1: Run Complete Workflow
+
+Execute all three components sequentially:
+
+```bash
+cd demo/_showcase/ansys-qoi
+tesseract build ./qoi_dataset/
+tesseract build ./qoi_inference/
+tesseract build ./qoi_train/
+python workflow.py
+```
+
+This will:
+1. Process all Ansys runs into NPZ datasets
+2. Train QoI-based surrogate models on the dataset
+3. Run inference using the latest trained model
+
+**Output locations:**
+- Dataset: `outputs/dataset/*.npz`
+- Models: `outputs/models/experiment_hybrid_YYYYMMDD_HHMMSS/`
+- Predictions: `outputs/predictions_YYYYMMDD_HHMMSS.csv`
+
+### Option 2: Run Components Individually
+
+#### 1. Dataset Generation (`qoi_dataset`)
+
+Process Ansys simulation runs into point cloud datasets:
+
+```bash
+cd qoi_dataset
+tesseract build .
+python main.py
+```
+
+**What it does:**
+- Samples point clouds from CAD files (`.stl`)
+- Extracts boundary condition parameters
+- Extracts CAD sketch design parameters
+- Parses QoI from simulation outputs
+- Generates compressed NPZ files with all data
+
+**Outputs:** `outputs/dataset/*.npz` files
+
+#### 2. Model Training (`qoi_train`)
+
+Train hybrid ML models on the processed dataset:
+
+```bash
+cd qoi_train
+tesseract build .
+python main.py
+```
+
+**What it does:**
+- Loads NPZ dataset files
+- Creates train/val/test splits
+- Fits data scaler
+- Trains QoI-based surrogate model
+- Evaluates model performance
+- Saves trained models and scalers
+
+**Outputs:**
+- `outputs/models/experiment_hybrid_YYYYMMDD_HHMMSS/`
+  - `models/hybrid_pointnet_small.pkl` - Trained model weights
+  - `scaler.pkl` - Data normalization scaler
+  - `config.yaml` - Training configuration
+  - `model_metrics.json` - Performance metrics
+
+#### 3. Inference (`qoi_inference`)
+
+Run predictions on new geometries using trained models:
+
+```bash
+cd qoi_inference
+tesseract build .
+python main.py
+```
+
+**What it does:**
+- Loads trained model and scaler
+- Processes input geometries
+- Generates QoI predictions
+- Saves predictions to CSV
+
+**Outputs:**
+- `outputs/predictions_YYYYMMDD_HHMMSS.csv`