How much rooftop PV can a distribution feeder take before a customer sees more than 1.05 per unit? We ran the question in OpenDSS on the IEEE 13-node test feeder: 300 random placements of PV on 1 to 8 customer nodes, ramped up at 30 % of peak load until the first voltage, conductor or transformer violation, with every measure scored on the same placements.
- The median placement takes 1,015 kW (10th to 90th percentile 510 to 1,893 kW). Voltage binds first in 224 of 300.
- Lowering the regulator set-point by 2 V adds the most: +310 kW median, better on 74 % of placements, worse on 10 %, for 0.007 pu less margin at peak load.
- A fixed power factor of 0.95 changed the median by -60 kW and the IEEE 1547-2018 Category B volt-var curve by 0. The regulators' line-drop compensation saw the lower current and held the voltage up. With the taps frozen the same measures add +110 and +60 kW.
The caveats matter more than the headline: one small, heavily loaded feeder; snapshots, not a day; recalled conductor ampacities; the regulator settings of the test case, which a real utility would not share. The point is that a hosting study that leaves out regulator behaviour can rank measures wrongly.
Code, tests (the unmodified feeder reproduces the published IEEE voltages to 0.003 pu) and results: research/infra-hosting.