Idea/Feature Request: "Zero impedance" power flow

[bergkvist]

How it would work:

1. Ignore any impedance in lines/transformers
2. Turn cycles in the network graph into single buses by merging all buses in each cycle.
3. Calculate the flow of active and reactive power by simply summing up p_mw and q_mvar in all the nodes as your traverse the graph from the loads up to the slack bus/ext_grid.
4. It might be possible to get an estimate for current by dividing the power by vn_kv (reference voltage).
5. We don't calculate bus voltages.

Advantages:

• Will always converge, even if the regular powerflow doesn't. Because of this, it allows us to identify power transfer bottlenecks which might cause a regular powerflow to diverge.
• Gives a fairly good estimate for power and current flow, typically within 10% of what a regular powerflow would give.
• Allows for finding direction of power flow/current everywhere except cycles.

Shortcomings:

• We won't know the direction of current/power flow within a "cycle", only what is going into a cycle, and coming out of it - so this will not work well for meshed networks.
• Doesn't take line power losses into account
• Is not able to provide an estimate for voltage

[SteffenMeinecke]

Hi @bergkvist, thanks for another great idea. Just some comments for discussion:

What is the reason for diverging power flows?

Is it the complexity caused by reactive power and voltages? -> then a dc power flow might be already a sufficient simplification Is it the complexity of a large and meshed grid? Then this proposal is not the solution. Is it a by far radial net but with ill-conditioned r/x values? Then your proposal could be helpful.