Groundwater 1D-2D, gridbuilder part

[caspervdw]

Questions:

• [x] What to do with the 1D-2D groundwater exchange of connection nodes? Do we follow the settings of connected channels / pipes? Or is there no exchange unless explicitly set via a manhole?

Discussed with @martijn-siemerink and @nvolp:

For connection nodes we need:

• a total width_1d, which is the sum of the lengths of exchange objects (until the first velocity point)
• coefficients weighted with their length (see width_1d)
• for the manhole exchange itself, we use sqrt(storage_area)

For all objects we need a pipe thickness, which defaults to 10cm. This is stored in ds1d.

[caspervdw]

Darcy flow: $Q = \kappa \frac{A}{\delta} \Delta \zeta$, user gives $(\kappa{in}, \kappa{out}, \delta)$ per object. ( $\kappa$ = hydraulic conductivity, A = crosssectional area $\delta s_{1D} * P(\zeta)$ , $\delta$ = thickness

For the exchange of a single channel or pipe segment, we can write $(\kappa{in}, \kappa{out}, \delta, {ds1d})$ in the gridadmin.

For the exchange of a connection node, however, we have to sum over the manhole and connected pipe/channel exchanges. We have:

$$Q = \Delta \zeta \sum_i{\kappa_i \frac{A_i (\zeta)}{\delta_i}}$$

It is probably best to precompute $\sum_i{\kappa_i \frac{A_i}{\delta_i}}$, but the weights $A_i$ are not completely known (because for pipes and channels they depend on the water level). I think we have to make some approximation here, but I am unsure which one.

@nvolp

[caspervdw]

ref. https://github.com/nens/threedi-schema/pull/3

[caspervdw]

Closing as untestable in API (need "testuser" stuff for that, and this specific feature has probably already been tested by testbank team while making testbank tests)