figures are almost certainly higher than what San Francisco would pay, but they provide a reasonable, if conservative, estimate. LADWP's cost per customer is $383; multiplied by the number of customers in San Francisco, that cost is $131 million a year. Then there's the question of generating or buying the electricity. Here San Francisco has a huge advantage over other public power agencies: The city owns a large hydro<\h>electric dam that can generate enough to cover some of the local power needs - and it's already paid for. Power from the Hetch Hetchy dam is cheap: the cost of operating the system is only about 2¢ a kilowatt-<\h>hour. Unfortunately, the city also has to pay PG&E to ship the power over its lines to the city borders, since the city has no complete transmission line to carry the power here; San Francisco pays PG&E $9.6 million a year in what's known as "wheeling fees." San Francisco currently sells most of the available Hetch Hetchy power to the Turlock and Modesto Irrigation Districts. Our analysis assumes that those contracts will be broken and that much of the power - 425 million kilowatt-<\h>hours' worth - will be available to the MUD or WPA. The city also has a very expensive contract with Calpine to provide backup energy when water is low at the dam. The wheeling fees and Calpine deal boost the actual cost of Hetch Hetchy power to about 4¢ a kilowatt-hour. But the Calpine deal ends in five years, at which point Hetch Hetchy power will be far less expensive - and the MUD's costs will go down. Green power Our analysis is based on the assumption that San Francisco will move as rapidly as possible to reduce its reliance on fossil fuels (see "Green City," 9/26/01). Not all of the alternative-<\h>energy sources that should ultimately be part of the city's mix are likely to be online when the MUD starts operating, so we've again been conservative, assuming in our worst-case scenario only a modest amount of solar power to supplement Hetch Hetchy power. In our best-case scenario we assume that the city will be able to develop 200 megawatts of solar and wind power - five times as much as projected in the solar bond measure, Proposition B, and enough to power 200,000 homes. The cost of solar and wind is easy to determine: it's the cost of the interest on the bonds needed to buy and install the windmills and panels. Once they're up and running, they cost very little to operate - and the fuel, of course, is free. Based on the San Francisco Public Utilities Commission staff's analysis of Prop. B), 40 megawatts of solar, wind, and efficiency programs - the equivalent of 98 million annual kilowatt-<\h>hours - will cost about $7.5 million a year. Our ambitious plan - for five times that much solar and wind power- would cost $38 million a year. (Again, the actual costs will probably be lower; once a big agency orders a large amount of solar- or wind-<\h>generating facilities, the price goes down substantially.) The rest of the power the city needs will have to be bought on the open market. Because the market is so volatile, it's hard to say exactly what that cost would be. But futures contracts for power are listed on the New York Mercantile Exchange Web site, and they're currently running at less than 4¢ a kilowatt-hour. That price is expected to decline in the future. Again, we've stuck to conservative numbers, assuming the MUD or WPA would have to pay 6.9¢ a kilowatt-<\h>hour for power generated locally, by Mirant Corp.'s Potrero Hill power plant (one energy expert told us that Mirant is unlikely to accept less than the 6.9¢ the state is now paying for power), and 5.5¢ a kilowatt-<\h>hour for power bought from out-of-town sources. We assumed that the Potrero plant would operate at its capacity.