# TORE Repository Description
## MRI-Validated CFD-DEM and rCFD Dataset for Gas-Solid Fluidized Beds

**Dataset Title**: Exploiting Pseudo-Repetitive Flow Dynamics of Gas-Solid Fluidized Bed for Real-Time Simulation: MRI-Validated rCFD Methodology for Fluidized Beds

**DOI**: https://doi.org/10.15480/882.15949

**Version**: 1.0

**Publication Date**: 2025

---

## FAIR Principles Compliance

### Findable
- **Unique Identifier**: Persistent DOI assigned by TORE platform
- **Rich Metadata**: Comprehensive descriptions in README files at repository and directory levels
- **Keywords**: CFD-DEM, rCFD, fluidized bed, MRI validation, drag models, poppy seeds, recurrence CFD, bubble dynamics
- **Searchable**: Indexed in TORE repository with full-text search capabilities

### Accessible
- **Open Access**: Publicly available via TORE platform (https://www.tore.tuhh.de/)
- **Standard Protocols**: HTTP/HTTPS download, Git clone
- **File Formats**: Open standards (CSV, PNG, SVG, OpenFOAM, VTK, Python)
- **Long-term Storage**: Maintained by Hamburg University of Technology Library

### Interoperable
- **Standard Formats**: CSV (data), PNG/SVG (images), OpenFOAM case structure, VTK files
- **Software Compatibility**: OpenFOAM, LIGGGHTS, ANSYS Fluent, ParaView, Python
- **Documentation**: Detailed README files with usage examples and dependencies
- **Cross-platform**: Linux, Windows, macOS compatible scripts

### Reusable
- **Licensing**: [Specify license - CC BY 4.0 recommended]
- **Attribution**: Citation information provided in all README files
- **Provenance**: Complete workflow documentation from simulation setup to validation results
- **Usage Examples**: Python code snippets, command-line examples, configuration templates

---

## Dataset Overview

This repository contains a **complete research data package** for MRI-validated computational fluid dynamics modeling of gas-solid fluidized beds. The dataset includes CFD-DEM simulations, recurrence-based CFD (rCFD) predictions, experimental MRI measurements, post-processing workflows, and validation results.

**Research Context**: This dataset supports the development and validation of recurrence CFD (rCFD) methodology, which exploits pseudo-repetitive flow patterns in fluidized beds to achieve real-time predictions with orders-of-magnitude computational speedup compared to full CFD-DEM simulations.

**Key Innovation**: First comprehensive MRI validation of rCFD methodology for gas-solid fluidized beds, demonstrating accuracy comparable to CFD-DEM at <1% computational cost.

---

## Repository Structure

The repository is organized into **8 main directories**, each with detailed README documentation:

### 01_CAD_Geometry
**Purpose**: 3D CAD models and technical drawings of experimental fluidized bed apparatus

**Contents**:
- SolidWorks parts and assemblies (`.SLDPRT`, `.SLDASM`)
- STEP files for CAD interoperability
- Technical drawings (PDF)
- Mesh-compatible STL exports

**Key Files**: Fluidized bed column (96mm diameter × 600mm height), distributor plate, pressure taps

---

### 02_Mesh_Generation
**Purpose**: Computational mesh development and grid independence studies

**Contents**:
- OpenFOAM `blockMeshDict` files for 3 geometries (CI, SC, SQ84)
- Grid independence study results at 1.5mm, 2.0mm, 3.0mm resolutions
- Optimal mesh (SC_3mm): 156,800 hexahedral cells
- Python validation scripts for mesh quality metrics

**Key Results**: SC_3mm mesh selected with <2% NRMSE for velocity profiles, 10.31% wall region error

---

### 03_CFD-DEM_Simulations
**Purpose**: Coupled CFD-DEM simulation cases and parametric studies

**Contents**:
- **Drag model comparison**: Beetstra, DiFelice, Gidaspow, Koch-Hill correlations
- **Boundary condition study**: ConsistentVelocityBC vs uniformFixedValue
- **Material properties**: Poppy seed characterization (⚠️ IP of SPE institute - permission required)
- **Simulation templates**: HPC-ready CFDEMcoupling cases with SLURM batch scripts

**Software Stack**: OpenFOAM 6, LIGGGHTS, CFDEMcoupling-PFM, OpenMPI 3.0

**Simulation Scale**: 0.5796 kg particles, 64 CPU cores, ~6 seconds physical time

---

### 04_rCFD_Simulations
**Purpose**: Recurrence-based CFD implementation and validation

**Contents**:
- **Input database**: 799 CFD-DEM snapshots (5 seconds, Koch-Hill drag, 1.5×Umf)
- **rCFD implementation**: ANSYS Fluent setup with custom UDFs (268k cell mesh)
- **Recurrence analysis**: Python scripts for pattern detection and visualization
- **Source code**: ⚠️ Requires access from JKU Linz (Prof. Stefan Pirker)

**Key Achievement**: 100-1000× speedup vs full CFD-DEM with comparable accuracy

**Software**: ANSYS Fluent 2019 R3+, Python with pandas/numpy/scipy/pyvista

---

### 05_MRI_Experimental_Data
**Purpose**: Magnetic Resonance Imaging experimental validation data

**Contents**:
- **Raw MRI video**: Cropped region of interest (MP4)
- **Frame extraction**: Python scripts with OpenCV
- **Image upscaling**: Nearest-neighbor interpolation for edge preservation
- **Particle characterization**: Camsizer data, Sauter mean diameter (d32) calculation

**Acquisition**: Hamburg University of Technology, Institute of Process and Plant Engineering (IPI)

**Processing Workflow**: Video → frame extraction → upscaling (4×) → binary masks → validation

---

### 06_Post_Processing_Scripts
**Purpose**: Automated analysis pipeline from CFD timesteps to MRI comparison

**4-Step Workflow**:
1. **Step 1**: ParaView-based 2D slice extraction at multiple void fraction thresholds
2. **Step 2**: Optimal threshold selection analysis (minimize NRMSE vs MRI)
3. **Step 3**: MRI data processing for threshold-based comparison
4. **Step 4**: Quantitative CFD-DEM/rCFD vs MRI comparison plots

**Key Scripts**:
- `paraview_postprocess_configurable.py`: Automated slice extraction
- `bubble_detection.py`: Bubble identification using OpenCV
- `bubble_analysis_MRI_style.py`: Comprehensive bubble property extraction
- `optimal_threshold_analysis_*.py`: Multi-metric threshold optimization
- `CFD_DEM_vs_MRI_*.py`: Publication-quality comparison plots

**Dependencies**: ParaView 5.11+, Python 3.8+ with OpenCV, pandas, matplotlib, scipy

---

### 07_Validation_Results
**Purpose**: Raw data and figures for Results and Discussion section

**Contents** (96.5 MB total):
- **Drag model comparison** (90 MB): Beetstra, DiFelice, Gidaspow, Koch-Hill vs MRI
- **MRI experimental data** (3.3 MB): Ground truth measurements
- **rCFD validation** (3.2 MB): Time-extrapolated predictions vs MRI
- **Boundary condition study**: ConsistentVelocityBC vs uniformFixedValue

**Data Format**: CSV files with bubble velocity, diameter, count, bed expansion, radial distribution, residence time

**Validation Metrics**: RMSE, NRMSE, MAE, Pearson correlation coefficient

**Key Finding**: Koch-Hill drag model shows best agreement with MRI (NRMSE < 5% for most metrics)

---

### 08_Computational_Performance
**Purpose**: Timing data and computational cost analysis

**Contents**:
- CPU time logs for CFD-DEM simulations
- rCFD speedup quantification
- HPC resource utilization metrics
- Scalability analysis

---

## Case Details and Parameters

### Fluidized Bed Specifications
- **Geometry**: Cylindrical column, 96mm inner diameter, 600mm height
- **Particle Material**: Poppy seeds (Papaver somniferum)
- **Particle Properties**:
  - Sauter mean diameter (d32): ~1000-1500 μm
  - Density: 1040 kg/m³
  - Polydisperse size distribution (500-2500 μm)
- **Operating Condition**: 1.5 × Umf (minimum fluidization velocity)
- **Static Bed Height**: 170 mm
- **Total Particle Mass**: 0.5796 kg

### Computational Setup
- **Mesh Type**: Structured hexahedral (blockMesh)
- **Optimal Mesh**: SC_3mm (squircle cross-section, 3mm cells)
- **Total Cells**: ~156,800
- **CFD Timestep**: 2.5 × 10⁻⁶ s
- **Coupling Interval**: 25 CFD steps (6.25 × 10⁻⁵ s)
- **Parallel Decomposition**: 64 CPU cores (4×4×4)
- **Simulation Duration**: ~6 seconds physical time (CFD-DEM), minutes (rCFD)

### Drag Models Evaluated
1. **Beetstra**: Wide porosity range (0.4-1.0), polydisperse correction
2. **DiFelice**: Semi-empirical, adjustable exponent
3. **Gidaspow**: Hybrid Ergun-Wen Yu, threshold at ε=0.8
4. **Koch-Hill**: Lattice-Boltzmann basis, implicit coupling ✓ **Best performer**

### Validation Metrics Analyzed
- Bubble rise velocity vs bed height
- Bubble equivalent diameter distributions
- Bubble count spatial profiles
- Bed expansion height and ratio
- Radial particle concentration (center-wall ratio)
- Residence time distributions
- Aspect ratio (bubble shape)

---

## Technical Specifications

### Software Requirements

**CFD-DEM Simulations**:
- OpenFOAM 6 or compatible version
- LIGGGHTS (open-source DEM code)
- CFDEMcoupling-PFM framework
- OpenMPI 3.0+ for parallel execution
- ParaView 5.11+ for visualization

**rCFD Simulations**:
- ANSYS Fluent 2019 R3 or later
- rCFD source code (access via JKU Linz)
- Visual Studio compiler (Windows)

**Post-Processing**:
- Python 3.8+ with: numpy, pandas, matplotlib, scipy, opencv-python, Pillow, pyvista
- ParaView with pvbatch/pvpython
- Git for version control

**Operating Systems**: Linux (recommended for CFD-DEM), Windows (for rCFD/Fluent), macOS (for post-processing)

### Hardware Requirements

**Minimum**:
- 16 GB RAM
- 4 CPU cores
- 50 GB free disk space

**Recommended for Full Workflow**:
- 128 GB RAM
- 64+ CPU cores (HPC cluster)
- 500 GB fast storage (SSD)
- GPU for ParaView rendering (optional)

---

## Data Reuse Instructions

### Quick Start for Different User Types

**1. Researchers Validating CFD-DEM Models**:
```bash
# Download repository
git clone https://tore.tuhh.de/[repository-url]

# Use validation data
cd 07_Validation_Results/MRI_postprocess_data
# Compare your simulation with MRI experimental data

# Use post-processing scripts
cd ../../06_Post_Processing_Scripts
# Adapt scripts for your simulation output
```

**2. Students Learning CFD-DEM**:
```bash
# Start with simulation templates
cd 03_CFD-DEM_Simulations/simulation_templates

# Study mesh generation
cd ../../02_Mesh_Generation
# Examine blockMeshDict files and validation scripts

# Explore post-processing
cd ../06_Post_Processing_Scripts/Step1_*
# Learn bubble detection and analysis
```

**3. Engineers Designing Fluidized Beds**:
```bash
# Access CAD geometry
cd 01_CAD_Geometry
# Use STEP files in your CAD software

# Review validation results
cd ../07_Validation_Results/drag_model_comparison
# Select appropriate drag model for your application

# Check particle properties
cd ../05_MRI_Experimental_Data/particle_characterization
# Compare with your particle system
```

**4. Computational Scientists Developing rCFD**:
```bash
# Study rCFD implementation
cd 04_rCFD_Simulations/rCFD_implementation

# Analyze recurrence patterns
cd ../recurrence_analysis
# Adapt for your flow system

# Validate predictions
cd ../../07_Validation_Results/rCFD_validation
# Benchmark against MRI data
```

---

## Citation and Attribution

### Primary Publication
**Please cite**:
```bibtex
@article{Asif2025FluidizedBed,
  title={Exploiting Pseudo-Repetitive Flow Dynamics of Gas-Solid Fluidized Bed
         for Real-Time Simulation: MRI-Validated rCFD Methodology for Fluidized Beds},
  author={Asif, Shaik and Pietsch-Braune, Swantje and Pirker, Stefan and
          Özdemir, Melis and Adriana, Muhammad and Penn, Alexander and Heinrich, Stefan},
  journal={Chemical Engineering Research and Design},
  year={2025},
  doi={10.xxxx/xxxxx}
}
```

### Dataset Citation
```bibtex
@dataset{Asif2025Dataset,
  title={MRI-Validated CFD-DEM and rCFD Dataset for Gas-Solid Fluidized Beds},
  author={Asif, Shaik and Pietsch-Braune, Swantje and Pirker, Stefan and
          Özdemir, Melis and Adriana, Muhammad and Penn, Alexander and Heinrich, Stefan},
  publisher={TORE - TU Hamburg Open Research Data Repository},
  year={2025},
  doi={10.15480/882.15949}
}
```

### Acknowledgments
- **rCFD Source Code**: Prof. Stefan Pirker, Department of Particulate Flow Modelling,
  Johannes Kepler University Linz, Austria (https://github.com/ParticulateFlow/rCFDreloaded)
- **Material Properties**: Institute of Solids Process Engineering and Particle Technology (SPE),
  Hamburg University of Technology (permission required for use)
- **MRI Measurements**: Institute of Process and Plant Engineering (IPI),
  Hamburg University of Technology
- **Funding**: This project is funded by the Deutsche Forschungsgemeinschaft (DFG, German Research
  Foundation) – SFB 1615 – 503850735

---

## Intellectual Property and Licensing

### Open Components
- **Simulation setups**: [Recommended: CC BY 4.0]
- **Post-processing scripts**: [Recommended: MIT or CC BY 4.0]
- **Validation data**: [Recommended: CC BY 4.0]
- **CAD geometry**: [Recommended: CC BY 4.0]

### Restricted Components
⚠️ **Material Properties** (`03_CFD-DEM_Simulations/material_properties/`):
- Intellectual property of SPE, Hamburg University of Technology
- **Permission required** before use
- Contact: Institute of Solids Process Engineering and Particle Technology

⚠️ **rCFD Source Code** (not included):
- Proprietary to JKU Linz
- Access requires permission from Prof. Stefan Pirker (stefan.pirker@jku.at)
- Repository: https://github.com/ParticulateFlow/rCFDreloaded (private)

### Modified Implementation
- `04_rCFD_Simulations/`: Modified version available at https://github.com/shaikasif808/rCFDreloaded_asif
- Case-specific setup files and UDFs (no core algorithm changes)

---

## Contact Information

**Primary Contact**:
- **Name**: Shaik Asif
- **Email**: shaik.asif@tuhh.de
- **Affiliation**: Hamburg University of Technology, Institute of Solids Process Engineering and Particle Technology

**Principal Investigator**:
- **Name**: Dr.-Ing. Swantje Pietsch-Braune
- **Email**: swantje.pietsch-braune@tuhh.de
- **Affiliation**: Hamburg University of Technology, Institute of Solids Process Engineering and Particle Technology

**Repository Maintainer**: TORE - TU Hamburg Open Research Data Repository

**Issue Reporting**: [GitHub Issues link if applicable]

---

## Version History

**v1.0** (2025-10-02):
- Initial release with complete dataset
- All 8 directories with comprehensive README documentation
- CFD-DEM drag model comparison (4 models)
- rCFD validation results
- MRI experimental data
- Post-processing pipeline
- Validation results for publication

**Future Updates**:
- Additional drag models
- Extended operating conditions
- Statistical analysis results (07_Validation_Results/statistical_analysis/)
- Computational performance benchmarks (08_Computational_Performance/)

---

## File Inventory Summary

| Directory | Size | File Count | Key Formats |
|-----------|------|------------|-------------|
| 01_CAD_Geometry | ~50 MB | 15 | SLDPRT, STEP, STL, PDF |
| 02_Mesh_Generation | 41 MB | 25 | blockMeshDict, Python, PNG |
| 03_CFD-DEM_Simulations | 287 MB | 150+ | OpenFOAM, LIGGGHTS, SLURM |
| 04_rCFD_Simulations | 50 MB | 30 | Fluent case, Scheme, UDF, Python |
| 05_MRI_Experimental_Data | 5 MB | 10 | MP4, Python, XLSX |
| 06_Post_Processing_Scripts | 520 KB | 15 | Python, Batch, Markdown |
| 07_Validation_Results | 96.5 MB | 500+ | CSV, PNG, SVG |
| 08_Computational_Performance | TBD | TBD | Log files, CSV |
| **Total** | **~530 MB** | **750+** | Multiple open formats |

---

## Quality Assurance

### Data Validation
✓ All CSV files validated for proper formatting
✓ Mesh quality checked (orthogonality >0.95, aspect ratio <1.2)
✓ Simulation convergence verified
✓ MRI data cross-referenced with experimental logs
✓ Post-processing scripts tested on multiple platforms

### Documentation Quality
✓ README file in each directory (8 total)
✓ Usage examples with actual code snippets
✓ File naming conventions documented
✓ Units clearly specified
✓ Data provenance traced

### Reproducibility
✓ Complete simulation setup files
✓ Exact software versions documented
✓ Configuration files (JSON) for all analyses
✓ Step-by-step execution instructions
✓ Example output provided

---

## Keywords for Searchability

**Primary**: CFD-DEM, rCFD, fluidized bed, MRI validation, recurrence CFD, bubble dynamics

**Secondary**: drag models, OpenFOAM, LIGGGHTS, CFDEMcoupling, ANSYS Fluent, ParaView

**Application**: chemical engineering, particle technology, multiphase flow, process engineering

**Methods**: computational fluid dynamics, discrete element method, magnetic resonance imaging, image processing

**Particles**: poppy seeds, gas-solid flow, granular materials

**Validation**: experimental validation, model comparison, threshold optimization

---

## Related Resources

- **rCFD Methodology**: Lichtenegger & Pirker (2016), *Chemical Engineering Science* 153:394-410
- **Transport-based rCFD**: Pirker & Lichtenegger (2018), *Chemical Engineering Science* 188:65-83
- **CFDEMcoupling**: https://www.cfdem.com
- **OpenFOAM**: https://www.openfoam.org
- **ParaView**: https://www.paraview.org

---

**Repository DOI**: https://doi.org/10.15480/882.15949

**Publication Date**: 2025

**Last Updated**: 2025-10-02

**License**: [Specify - Recommended: CC BY 4.0 for data, MIT for code]

---

*This description follows FAIR (Findable, Accessible, Interoperable, Reusable) principles for research data management.*