Households and developers are adding rooftop solar, batteries and onsite water reuse because the private numbers often close. A usable roof, a clear payback and a lower volumetric bill are enough to move a project forward. The public network those customers still use does not shrink on the same timetable. Feeders, transformers, pipes and treatment plants have to be maintained after dark, during outages, and for people who cannot install anything of their own.
That coupling is the subject of the Substack essay “Private Solar, Public Wires: Partial Exit from Shared Networks” written by Dr Jose Luis Chavez Calva. The essay treats partial exit as a design problem for shared infrastructure. Some customers generate or treat on site. The residual system still has to run, and someone still has to pay for it.
Two research programmes at Arizona State University give the essay its frame. Margaret Garcia, a civil engineer, and Paul Grogan, who works with game theory and design under uncertainty, hold a three-year grant from the US National Science Foundation to model electricity and water networks together. One layer is physical: substations, feeders, laterals, service transformers, pipes and plants. The other is strategic: households, developers, utilities and regulators choosing with incomplete information about neighbours, future rates and how long they will stay in a building. Private kit lands where payback and tenure are favourable. Residual capacity sits where the circuit still has thermal and voltage headroom. Those two maps do not have to match.
Oahu already shows the geography. Hawaiian Electric counted 120,570 grid-connected solar systems by early 2026, with about 82,475 of them on Oahu. Roughly half of the island’s single-family homes have rooftop panels. Hosting capacity is a local remainder, scored on each circuit at daytime minimum load. Saturated laterals slow new connections while neighbouring feeders still have room. Hawaii closed retail net metering to new applicants in October 2015, under Public Utilities Commission Docket 2014-0192, and moved later applicants onto Grid Supply and Self Supply tariffs. The export credit was set below the retail rate because midday injection and evening demand are different products.
The same logic appears in shared irrigation. A cheap bus layout passes one neighbour’s exit downstream. A star layout contains it. A distribution feeder is closer to the bus. Layout is therefore part of the policy problem, not only a drawing in a planning file. When volumetric sales fall and fixed costs remain, the residual bill lands on renters and on owners without a usable roof. US studies at current penetration, including work from Lawrence Berkeley National Laboratory, generally find a modest bill shift. The shift is larger where recovery still rests on kilowatt-hour charges.
Neighbouring cases make the same point in different units. California’s net-load curve has deepened from the familiar duck into a steeper evening ramp. Parts of Australia see voltage rise on feeders with high rooftop shares. In each case the constraint is local and timed, which is why a flat export credit and a flat volumetric rate struggle to send a useful signal.
Tools that sit on the network map: locational incentives, contributions to shared plant that partial exit cannot fully bypass, export credits tied to feeder conditions, and public maps of remaining headroom. Garcia and Grogan’s models are still being built. They are a way to test those tools against observed siting, not a finished tariff.
Source: https://joseluischavezcalva.substack.com/p/private-solar-public-wires-partial
