Two series of conferences (Water Distribution Systems Analysis WSDA) and Computing and Control in the Water Industry CCWI) were combined for the first time and brought together many industry professionals and researchers from around the world, among other things for battle of network modelling skills.
The 1st International Joint Conference in Water Distribution Systems Analysis & Computing and Control in the Water Industry in Kingston, Ontario, Canada, took place on July 23-25, 2018. This joint conference brought together industry professionals and researchers from around the world to solve problems in the field of urban water systems, modelling, and control. The program consisted of themes in water distribution analysis and modelling (as covered by the Water Distribution Systems Analysis (WDSA) series of conferences) as well as themes in water systems modelling and control (as covered by the Computing and Control in the Water Industry (CCWI) series of conferences).
Follow up collaborations
For me and other KWR-participants an important part of the conference was the opportunity for networking. Now that the two top conferences for water infrastructure researchers were combined for the first time into one, there were many people present that we wanted to talk to, or who wanted to talk to us, and also many new people to meet. Even though there were coffee breaks, good long lunches and a conference dinner, it was a bit of challenge to catch up with everyone. To name just a few highlights, there will be follow up collaborations with researchers from Montreal on QMRA modelling, with researchers from Arizona and Toronto on self-cleaning networks, with researchers from South Africa, Waterloo, Cincinnati, Italy, etc. on demand modelling, with researchers from Sheffield on transients and water quality, with researchers from Colombia on drinking water in general, and many more.
Everyone at the conference was aware of Dragan Savic’s move to KWR and as a result participants were much more interested in KWR, how we work and possibilities for collaboration. One good way would be to share data and apply models on real Dutch networks that we have hydraulic models for, and combining this with measurements from the field and a good sense of historic events in the networks. I invited many people to try their ideas on the Dutch networks, instead of the good old EPANET Net 3 or Anytown network models.
Battle of network modelling
Typically the WDSA conference holds a battle of network modelling. This year the battle entailed bringing the drinking water supply up and running after an earth quake that caused pipe breaks and leaks, in such a way that the least number of people were out of water. Water was to be secured especially fast for a hospital and for firefighting. To accomplish this the water company would have three crews available to work around the clock to repair pipes.
The organizers of the battle provided a network model (B city, from Colombia), 6 criteria to judge the solution by, and an answer sheet to be filled out. The competitors (10 teams sent in a solution) had to prepare an abstract, a paper and a 5 min video to explain their methodology. This was a good way to communicate results without repeating 10 times the goals and objectives of the study.
Two Dutch teams
There were two (partly) Dutch teams that entered the competition, one with Edo Abrahams from TU Delft, and the other one with Mario Castro-Gama who now works with KWR. In his blog (link) you can read more about their solution for the problem and see their video.
No winner declared
Most teams used optimization methods, where one team decided that this was not very realistic in a post disaster situation and thus use engineering judgment. All ten teams had to make some choices on what to focus on, it was impossible to fulfill all 6 criteria and some teams did show good results on some criteria, where others scored high on a different set of criteria. The teams also gave some generic recommendations, which were mostly focused on being prepared for disasters and test the systems resilience beforehand and maybe add some extra tanks or valves. In the end, no winner was declared.