Provided EPANET inp files problems

[Maayowa]

Hello,

I would like to ask if the competition has officially kicked-off. While the timeline specifies December 8, it appears that all the shared networks (.inp files) are currently not fully functional. For example:

1. The networks cause syntax errors when loaded on EPANET or WNTR,
2. The pipe diameters are currently very large (> 25m),
3. There is a large scatter of unconnected reservoirs across the networks.

Could you please confirm if participants are expected to fully design and configure the networks with the existing components as options?

Thanks

[zannads]

Hi @Maayowa
Thanks for your interest in the battle. You raise good questions; let me break down a few concepts because the answer is, in a sense, both yes and no.
PS: since you are not the only one with the question, I renamed the discussion

For us, the competition has officially begun. We released the complete problem description (which should not change but is still under review by some external reviewers) and a good initial batch of data. We understand that there may be some inconsistencies or errors in the data, as it was all generated synthetically. For this, we will release multiple versions, so keep an eye out for that.

[zannads]

Regarding the EPANET inp files, we expect participants to model the system as they deem it most appropriate. We provide all the necessary information to build a model (list of sources, nodes, etc.) in the Excel files; however, such a model doesn't necessarily need to be an EPANET model. We do provide some inp files for support, and we suggest you open them with EPANET.js.

In the data folder, you can find more details about the networks in the epanet-networks/README.md. In short, these are only for visualisation purposes in the EPANET.js web UI.

These networks represent the situation in 2025 across the provinces and the possible sources that can be opened (hence the unconnected reservoirs).
The Netherlands network merges the individual provinces' networks into a unified view and also provides the possible inter-provincial connections that can be created. However, those pipes are not the only ones that can be installed, as the full collection is available in the connections-static_properties.xlsx file. Similarly, these inter-provincial pipes are there for visualisation purposes/debug and have a very large diameter.

The reason for all of this is that EPANET can not model the full complexity and dynamicity of the problem. Hence, inp files would be continuously changing.

[zannads]

PLEASE DISCARD THE PIPES ATTRIBUTES IN THE NETHERLANDS.inp FILE

All pipes should be between 200 mm and 2000 mm (the 10 options described in pipes/pipe_options-static_properties.xlsx).
Only a few exceptions at 4000 mm have been made to make helper pipes frictionless (e.g. connecting reservoirs to pumps). This will change in the next release and we will provide the frictionless effect through other pipe attributes.

Examining the networks, I see that this is correct for all provincial networks. However, only the Netherlands network has the large diameters that you mentioned. As this is just a merge of the individual provincial networks, this is likely a conversion error (most likely something related to units). I will fix it in the next release, but for now, please disregard the EPANET pipes attributes.

[zannads]

Please see #3 for more informaiton about the errors when opening it with EPANET and WNTR.