DiTEC-WDN: Large-Scale Water Distribution Network Simulation Dataset

Provider
DiTEC project, University of Groningen data science group (rugds)
License
CC-BY-4.0 (source)
Size
8,917 files
Format
Parquet (in sampled files)
Stored at
Hugging Face (https://huggingface.co/datasets/rugds/ditec-wdn)
Access
Free and public. No account with the data provider is needed.
Category
Machine Learning & AI
Source
https://arxiv.org/abs/2503.17167

About DiTEC-WDN: Large-Scale Water Distribution Network Simulation Dataset

DiTEC-WDN, a machine-generated dataset of hydraulic simulations for 36 water distribution networks (WDNs), built for graph machine learning on critical infrastructure. Each network has 1,000 unique scenarios, and each scenario is a time series of directed graphs with a shared topology, so the data form spatiotemporal graphs whose nodes and edges carry multivariate time series. Nodes are reservoirs, junctions or tanks; edges are pipes, pumps or valves, with concrete valve types (PRV, PSV, PBV, FCV, TCV, GPV) and head or power pumps. Inputs include junction base demands, elevations, pipe diameters, lengths and roughness, pump curves and valve settings; outputs include nodal pressure, head and demand and link flow rate, velocity, headloss and friction factor. Networks range from benchmark systems such as Anytown, Hanoi, EPANET Net-3, C-Town, D-Town and L-Town to the KY networks and large real-world systems, in folders named with size and horizon (for example hanoi_8GB_1Y, KL_232GB_1Y). Files are Parquet per parameter and chunk; the repository uses about 7.1 TB (measured 2026-11-14). Uses include state estimation, surrogate hydraulic modelling, leak and demand forecasting and graph neural network benchmarks. Read with pandas, pyarrow or polars. Licensed CC BY 4.0; cite the DiTEC-WDN paper (arXiv:2503.17167).

How to load DiTEC-WDN: Large-Scale Water Distribution Network Simulation Dataset in Python

from datasets import load_dataset
ds = load_dataset("rugds/ditec-wdn")
print(ds)

Data structure

File types in sampled files: Parquet

Files

Example file: 19PipeSystem_4GB_1Y/junction_base_demand-0-dynamic_input.parquet

Analyze this dataset without downloading it

On Scigantic, DiTEC-WDN: Large-Scale Water Distribution Network Simulation Dataset opens in a hosted JupyterLab notebook with the data already mounted, so you can read files where they are stored instead of copying them first. Each notebook includes an AI assistant that can see the mounted files and write and run analysis code. Open the dataset page to launch a notebook, or browse the full catalog.

This page is also available as Markdown and as JSON.

Related Machine Learning & AI datasets