The combination of redispatch and a nodal market design using the complex_clearing method produced highly unintuitive results.
Even though a nodal market design inherently accounts for all technical constraints of the underlying grid, we still observe redispatch in the subsequent simulation step.
Why is redispatch happening at all?

Problem Description

From an energy economics perspective, redispatch after a nodal market clearing should not occur, since nodal pricing internalizes grid constraints directly in the optimization.
One possible explanation could be that the grid representation in complex_clearing is too simplified, resulting in cases that are considered feasible in the clearing but become infeasible once evaluated with the more detailed grid model in PyPSA, thereby triggering redispatch.

However, this discrepancy should not occur if both models describe the same physical system consistently.

Open Questions

Is our market → redispatch workflow reliable, or is there a conceptual issue? Am I missing something obvious?

Did someone see such behavior with zonal markets and redispatch so that redispatch is happening between zones still?

Or is this caused by another bug (clearing logic, grid representation, data transfer, etc.)?

Reproducabilty

You can reproduce that by running example_01d on this branch: redispacth_and_nodal_pricing_bug