Level demand not fulfilled

[simulutions]

Ch2_user_demands.ipynb

In this model the level demand is not fulfilled, while it has the highest prio.

[Fati-Mon]

I think the user_demand's min_level that is set to zero overwrites in some way the allocation algorithm. Might that be the cause? If so, I was expecting in that case the other way.

[Fati-Mon]

I had a chat with @SouthEndMusic and here is what he explained:

TRC #8 is not allocation-driven, so it simply follows its physical behavior. If TRC #8 withdraws more water from Basin #7 than FlowBoundary #1 can supply, the level of Basin #7 will drop despite the LevelDemand, because allocation has no mechanism to prevent this.
This is actually the reverse of what we had before, and in my view, a much better setup. Allocation now “sees” the flow caused by TRC #8 and recognizes that this prevents LevelDemand from being met.
Another point: now that you’re combining a junction with allocation, this issue becomes relevant: Map links in allocation flow results similar to the physical layer (due to junction) · Issue #2390. In short, the allocation flow output is not entirely correct around junctions. This isn’t an allocation problem per se, but rather a bug in how results are written.
Physical flows determined by connector nodes “just happen,” and allocation must work around them. From that perspective, you could say such flows always have the highest priority. However, this is slightly misleading: physical flows are treated as constraints in the optimization, whereas allocations to demands are part of the objective.

It would be good to document this behavior and include a warning for modelers: if a connector node is used while allocation is enabled, this kind of interaction will occur.

In the meantime, I’ll adjust the TRC setup slightly, starting from a higher level than the LevelDemand itself, and see how that looks.

[Fati-Mon]

Making sure that the TRC starts discharging above the target level defined in LevelDemand gives much better results:

I’m happy to see that the irrigation demand (priority = 4) is now not entirely met, while the higher-priority demands, Industry (2) and City (3), are fully satisfied.

The interaction between the physical and allocation layers becomes much clearer once the modeler understands how to correctly define the physical properties (such as outlets, TRCs, etc.), helping to avoid confusion when interpreting results after enabling the allocation algorithm.