ShapeGen3DCP dataset: filament cross-section geometry for 3D concrete printing

This dataset supports the study “ShapeGen3DCP: A deep learning framework for layer shape prediction in 3D concrete printing” [1]. It contains numerical and experimental cross-sectional geometries of extruded filaments used to train and validate a deep learning model for predicting layer shapes in 3D concrete printing (3DCP). Citation Please cite the related article if you use this dataset: Rizzieri, G., Lanteri, F., Ferrara, L., & Cremonesi, M. (2026). ShapeGen3DCP: A deep learning framework for layer shape prediction in 3D concrete printing. Computers & Structures, 323, 108142. https://doi.org/10.1016/j.compstruc.2026.108142 Dataset content: Numerical cross-sections generated with a Particle Finite Element Method (PFEM) model for extrusion and deposition of cementitious materials. A subset of experimental cross-sections extracted from published studies. Material and process parameters for each case, including density, plastic viscosity, yield stress, nozzle diameter and height, print velocity, and flow velocity. Example visualization scripts in Python (plot_cross_section.py) and MATLAB (plot_cross_section.m). Dataset structure: 1LAYER/ – single printed filament layers 2LAYER/ – two overlapping printed layers Each folder contains a subfolder CROSS_SECTIONS_#L/ with .txt cross-section files and an Excel file (dataset_#L.xlsx) with associated parameters. File format: .txt files: ordered point coordinates of cross-sectional contours (nearest-neighbour ordering) .xlsx files: material and process parameters per case Scripts: Python and MATLAB example code for plotting Notes: File names include a case identifier and suffix num (numerical) or exp (experimental). Each cross-section can be reconstructed from the .txt files using the provided scripts. References: [1] G. Rizzieri, F. Lanteri, L. Ferrara, M. Cremonesi, ShapeGen3DCP: A deep learning framework for layer shape prediction in 3D concrete printing, Computers & Structures 323 (2026) 108142. https://doi.org/10.1016/j.compstruc.2026.108142 [2]–[6] Additional references supporting numerical simulations and experimental data are included in the README.