search

4flixt / 2019_WNTR_Surrogate_Model

5 stars 2 forks source link

Negative Pressures in new simplified model #7

Closed 4flixt closed 4 years ago

4flixt commented 5 years ago

I have been continuing to work on the clustering and investigated the new results_sim.pklfile which was created with c-town_true_network_simplified.inp (I believe), I know those are only preliminary results but I found that the pressure trajectories look a bit odd ( see below). pressure_new They also don't allow for a similar quality of clustering. Can the pressure really be negative? For comparison, here are the previous pressure trajectories, which yielded the good clustering results: pressure_old

acominola commented 5 years ago

I am working on it, do not trust these results. There was a problem with pump 2 producing negative pressures. That one is solved, but maybe not all pressures are solved. I will fix it

acominola commented 5 years ago

Updated. Now it should be better. If you look at the last figure in the "run_simulation" (sorry, no figure labels, was doing it quickly), pressure levels are now in a similar scale to what you had before (still, some negative values sometimes, seems not many). I am still working on putting some control boundaries that make sense, and later on will try to launch a few randomized (initial tank level, demand patterns) simulations

4flixt commented 5 years ago

ok I'll wait for good results than before I continue the clustering investigation. Thank you!

4flixt commented 5 years ago

I am just starting to realize where this negative pressure results from. I read in the documentation:

In demand-driven simulation, the pressure in the system depends on the node demands. The mass balance and headloss equations described above are solved assuming that node demands are known and satisfied. This assumption is reasonable under normal operating conditions and for use in network design. Both simulators can run hydraulics using demand-driven simulation.

From my understanding, the demand is always satisfied, even when the network can't provide. This results in negative pressures :man_facepalming:

acominola commented 5 years ago

Yes, indeed. Sorry, I thought this was clear. In a Demand Driven scenario (like the one we are adopting), the demand is always satisfied. As said in a previous answer, negative pressures are an artefact (it would mean that the water table is lower than the elevation...). Pressure driven simulations are feasible only with the WNTR simulator...