Neighborhood Microgrid Reactive-Power Optimizer (2D)

Ten houses share one radial feeder run from a substation, and every house draws real power plus lagging reactive power exactly like real inductive household loads do. This 2D simulator solves the linearized DistFlow power-flow equations after every change, so dragging the neighborhood-load slider genuinely sags the far end of the feeder below nominal voltage. A gradient-descent optimizer then tunes each house's capacitor bank — the same reactive-power compensation utilities install on real distribution feeders — to flatten the voltage profile while minimizing a cost function that penalizes both voltage deviation and capacitor usage. A pannable, zoomable feeder diagram is paired with linked voltage-profile, capacitor-bank and cost-descent charts, and feeder resistance, reactance, power factor and the optimizer's own learning rate are all live sliders.