# ConcreteShellFEA: A surrogate modelling dataset for the buckling and stress behaviour of concrete thin-shells

## Overview

ConcreteShellFEA is a dataset designed for the training of deep learning models to predict buckling loads and stress fields in concrete thin-shell structures.
It contains 3 smaller datasets, which can be used for different use cases:

1. PerfectShell_LinearFEA: A dataset of 20,000 thin-shells (with various span, height, thickness, and Young's modulus), for which buckling factors and stress fields under design loads were determined using linear Finite Element analysis. The data is presented in three formats (tabular, image, graph) to enable different types of deep learning models (Multilayer Perceptrons, Convolutional Neural Networks, and Graph Neural Networks) to be trained.

2. ImperfectShell_LinearFEA: A dataset of 20,000 imperfect thin-shells (with various span, height, thickness, Young's modulus, and geometric imperfections), for which buckling factors and stress fields under design loads were determined using linear Finite Element analysis. The data is presented in two formats (tabular, image) to enable different types of deep learning models (Multilayer Perceptrons, Convolutional Neural Networks) to be trained.

3. PerfectShell_NonlinearFEA: A dataset of 25,000 thin-shells (with various span, height, thickness, and Young's modulus, and geometric imperfections), for which buckling factors under design loads were determined using Finite Element analysis. The buckling factors were determined using linear Finite Element analysis for 20,000 thin-shells, and using nonlinear Finite Element analysis for 5,000 thin-shells, to enable mixed-fidelity applications.  The data is presented in a single format (tabular).

Further details about the methods used to generate this dataset can be found in:
Pollet, M. (2025). Rapid structural analysis of prefabricated thin concrete shells using deep learning (Thesis). University of Bath.

## Repository structure

```
ConcreteShellFEA/
│── datasets/
│ ├── PerfectShell_LinearFEA/
│ │ ├── tabular/                 # Tabular data
│ │ │ ├── input/                 # Input features
│ │ │ ├── output/                # Output features
│ │ ├── images/                  # Image-based data
│ │ │ ├── input/                 # Input images
│ │ ├── graphs/                  # Graph data
│ │ │ ├── input_and_output/      # Input and output graphs
│ ├── ImperfectShell_LinearFEA/
│ │ ├── tabular/                 # Tabular data
│ │ │ ├── input/                 # Input features
│ │ │ ├── output/                # Output features
│ │ ├── images/                  # Image-based data
│ │ │ ├── input/                 # Input images
│ ├── PerfectShell_NonlinearFEA/
│ │ ├── tabular/                 # Tabular data
│ │ │ ├── linear/                # Linear FE data
│ │ │ │ ├── input/               # Input features
│ │ │ │ ├── output/              # Output features
│ │ │ ├── nonlinear/             # Nonlinear FE data
│ │ │ │ ├── input/               # Input features
│ │ │ │ ├── output/              # Output features
│── scripts/                     # Scripts for loading data
│ ├── load_tabular_datasets.py
│ ├── load_image_datasets.py
│ ├── load_graph_datasets.py
```

## License

The contents of this dataset are licensed under Creative Commons: Attribution 4.0 (CC BY), which allows for the derivation of new datasets and other resources from the data, and for the redistribution of the data and their derivations, both openly and commercially, as long as the creators of this dataset are credited for the original creation.


## Crediting/citing the creators of this dataset

The creators of this repository can be acknowledged by citing the related research article:

Pollet, M. (2025). Rapid structural analysis of prefabricated thin concrete shells using deep learning (Thesis). University of Bath.

or using BibTeX:

```
@phdthesis{Pollet2025,
   author = {Maxime Pollet},
   school = {University of Bath},
   title = {Rapid structural analysis of prefabricated thin concrete shells using deep learning},
   year = {2025}
}
```

## Related publications

This dataset has been used in the following publications:

- Pollet, M. (2025). Rapid structural analysis of prefabricated thin concrete shells using deep learning (Thesis). University of Bath.

- Pollet, M., Shepherd, P., Hawkins, W., & Costa, E. (2026). Fast structural analysis of concrete thin-shells using deep learning. Computers & Structures, 320, 108042. https://doi.org/10.1016/J.COMPSTRUC.2025.108042